category: literaturenote citekey: wursterbentsorensenhydrogenfuel2007 title: "Bent Sorensen, Hydrogen and fuel cells, emerging technologies and applications, Elsevier, Oxford, UK (2005). ISBN: 10:0-12-655281-2, ISBN: 13:978-0-12-655281-2 (450pp, $67.95, Hard cover)." authors: "WURSTER, R" year: 2007 date: 2007-00-00 2007 doi: 10.1016/j.ijhydene.2007.09.008 publication: International Journal of Hydrogen Energy url: "https://doi.org/10.1016/j.ijhydene.2007.09.008" zotero_key: N8XWRN65 zotero_storage: DV3C6W5C collections: transpordi_planeerimine folder: 001_artiklid firstAuthor: "WURSTER, R"

status: converted

CHAPTER 1 INTRODUCTION

1.1 Possible role of fuel cells and hydrogen

The pollution from automobiles, particularly in city areas, is becoming increasingly unacceptable to people living in, visiting, or working in the cities of the world. Demands for zero-emission vehicles have been voiced, and the automobile industry has been facing louder and louder criticism for not addressing the problem. The simplest solution to reducing emissions is to make the vehicle more efficient. This is the route taken by several European car manufacturers, using a combination of low-weight car structure, low air resistance, high performance but low fuel consumption engines such as the common-rail diesel engine, brake energy recuperation, computer-optimised gearshift operation, engine close down as an alternative to idling, and so on. Presently, this leads to fuel consumption for a four-person standard car of about 3 litres of diesel fuel per 100 km or 4-5 litres of gasoline if the less efficient Otto engines are used. Three litres of diesel fuel per 100 km corresponds to 0.1 GJ or 1 MJ km-1. The fuel-to-wheel efficiency is about 60% higher than that of current global stocks of passenger cars (27% rather than 17%), while the overall fuel-to-transportation work efficiency for delivering the transport service to the end-user, as, for example, measured by the number of passengers times the distance driven (see section 6.2) is as much as about 2.3 times better than that of present average cars (12-13 km per litre of gasoline, the energy content of which is about 10% lower than that of diesel fuel). The average efficiency of current passenger car fleets is lowest in China and the United States and highest in Europe, a fairly direct consequence of the prevailing fuel prices seen by the customers (i.e., including subsidies and taxes).

Other options include electric and fuel cell-based vehicles, as well as hybrids between these or other concepts. If the original fuel is of fossil origin, environmental pollution depends of the total well-to-transportation service energy use. For electric vehicles, assuming a motor efficiency of 80%, a transmission efficiency of 98%, and a 40% fuel-to-electricity conversion efficiency at current state-of-the-art power plants, the well-to-wheel efficiency calculated after consideration of battery-cycle losses is 26%. For fuel cell cars, the hydrogen fuel-to-wheel efficiency is about 36% (see section 6.2), implying a fuel-to-wheel efficiency of around 25% for the chain starting from hydrogen production from natural gas, over proton exchange membrane (PEM) fuel cells and electric motors to wheels, all for a standard mixed driving cycle.

In the electric car case, pollution is moved from the street level to the locations of power plants, which if these are operated on fossil fuels allows for better exhaust gas cleaning and generally, the dispersal at higher altitudes is much less likely to cause pollutant concentrations as high as from currently common vehicle exhaust. If renewable energy sources such as wind or photovoltaics are used to produce the electricity, there are no operational phase pollutants. It follows from these estimates that the higher cost of electric or fuel cell vehicle solutions will in many cases be difficult to defend, relative to the simple efficiency improvements of current thermodynamic engines using conventional fuels, and the real case for electric or fuel cell vehicles is therefore strongly dependent on the need for a transition to nonfossil fuels. Of course, this does not exclude the use of fuel cell vehicles based on hydrogen produced from fossil fuel in demonstration projects during an interim period, where the infrastructure for hydrogen production based on sustainable energy sources is not yet available.

Considering hydrogen produced from renewable energy such as wind power and taking the efficiency of state-of-the-art electrolysis plants as being around 85%, this also leads to an overall primary energy-to-wheel efficiency of about 25% for a fuel cell vehicle, but this time without emission of pollutants. For vehicles operated purely on batteries, the corresponding efficiency is around 50%, but the weight penalty of battery operation, even using lithium-ion stacks, is usually higher than that of fuel cell operation; this is also the case when this includes some battery storage. If both batteries and fuel cell technologies were economically viable, the optimisation would consist of balancing the weight and required engine rating to obtain the lowest cost. However, none of the two technologies is economical today, and a central question for the role of fuel cells is whether cost reductions will be easier to achieve than with batteries, despite a certain level of technology similarity. The cost implications of selecting a range of possible hybrid concepts is

discussed in Chapter 6.

While direct pollution from motor vehicles, whether involving particles, SO2, or NOx emissions that are only partially reduced by filters and catalytic devices, is the largest visible social (health) and environmental impact, emissions of greenhouse gases have increasingly been moved up on the list of impacts not to be tolerated in the future. The reason is the implied additional warming of the atmosphere and the negative consequences it may have, both regarding the stability of overall climate and also in particularly vulnerable regional contexts (Sørensen, 2011a). Presently, climatic impacts seem to top the list of reasons to move away from fossil fuels. Not long ago, supply security and resource depletion were quoted as the key reasons to develop alternatives based upon sustainable energy sources. The temporary decline in such worries was due to the halt of the exponential growth in energy use, that was achieved after the oil supply crises in 1973 and 1979 by concerted efforts to improve efficiencies of energy use. While the most advanced industrialised nations have not noticeably increased their total energy use during the past three decades, despite continued economic growth, new high-growth players have appeared on the scene. Further, the transportation sector has in most parts of the world continued to increase its energy use, only compensated for by less usage of fuels for space heating and other sectors seemingly more amenable to efficiency improvements than cars, ships, and aircraft. As the preceding automobile example shows, there are no technical reasons for not improving the efficiency of vehicles for transportation, and indeed, an effort to catch up finally seems in progress.

It is clear that the resource issue is not gone permanently. Production of fossil fuels, and particularly oil, is expected to reach its maximum sometime in the next decade or two and then to decline despite enhanced recovery techniques, as the rate of discovering new wells continues to diminish. This will necessarily lead to price increases, although likely in an irregular way, due to the fact that the actual production costs are still very low in some regions, notably in the Middle East, and prices are therefore dependent on the dayto-day market situation and on political issues such as cartel ceilings on production and warfare in the regions of production. These conditions make the development of alternatives to fossil fuels increasingly more attractive, both for economic reasons and also for reasons of security of supply. Because renewable energy sources are more evenly available geographically (although the best mix of renewable sources may vary from one region to another), they are seen as attractive in scenarios placing emphasis on local control, often referred to as "decentralisation", although they certainly do not eliminate the need for transmission and trade of electricity and (renewablebased) fuels or for developing suitable forms of energy storage (see Sørensen, 2010a).

The reasoning just presented will be qualified by the more detailed discussion offered in the following chapters.

The possibility of using hydrogen as a general energy carrier has long been recognised (see, e.g., Sørensen, 1975, 1983, 1999; Sørensen et al., 2004). The issues raised for attention include hydrogen production, storage and transmission, and the use of hydrogen, notably as fuel for fuel cells. The hope is that fuel cells will experience a significant price reduction along with the development of new fields of application and that infrastructure problems will eventually be solved. This could happen through a series of steps, with hydrogen first being used in those niche areas where the required change in infrastructure is modest, such as fuel cell buses travelling by fixed routes from a single filling station. While the present price of producing hydrogen fuel (whether from fossil or renewable energy) is higher than that of the fuels already in use, it might be expected to decline if the market expands and production technology is refined. The cost of central hydrogen storage in underground facilities such as those already in use for natural gas will have a rather insignificant impact on the overall cost of hydrogen usage, while that of transmission is expected to be similar to or slightly higher than the cost of natural gas transmission. The cost of local hydrogen storage costs, typically using pressure containers, is not negligible but still has a fairly modest influence on overall cost, while the critical cost item remains the fuel cell converter used in all cases where the end-use energy form is electricity, including traction through electric motors. Thus the critical development items for allowing the penetration of hydrogen as a general energy carrier is the fuel cell. Hydrogen could be used at a more limited market in conventional thermal power plants, but this would only be interesting with nonfossil energy sources, and most of these (wind power, photovoltaics) do not need hydrogen to produce electricity but may need it to store energy for handling intermittency. One therefore has to consider the transportation sector as crucial for a serious introduction of hydrogen as a general energy carrier.

The disposition of the book is as follows: Chapter 2 discusses hydrogen production by different methods, storage, and transmission. Chapter 3 covers fuel cell basics. Chapter 4 presents fuel cell systems, followed by implementation issues (including safety and norms) and scenarios for future use in Chapter 5. Chapter 6 discusses economic issues such as direct and life-cycle costs and Chapter 7 summarises lessons learned. These distinctions are not watertight, since it is often useful to discuss systems options in connection with individual technologies or to mention implementation issues along with the technologies, but cross-references are provided.

CHAPTER 2

HYDROGEN

2.1 Production of hydrogen

Hydrogen is the most abundant element in the universe. The main isotope consists of one proton and one electron occupying the lowest angular momentum zero atomic state (i.e., the electron ground state, denoted 1s) at an energy of —2.18 × 10-18 J, relative to the electron being at infinity. Hydrogen is found as interstellar gas and as the chief constituent of main-sequence stars. On planets such as Earth, hydrogen is found as part of the molecules of water, methane, and organic material, whether fresh or fossilised. The average isotope ratio ² H to ¹ H in the natural abundance on Earth is 1.5 × 10⁻⁵ . The normal molecular form is H2. Further properties are listed in Table 2.1.

Hydrogen production involves extracting and isolating hydrogen in the form of independent molecules, at the level of purity required for a given application. The processes naturally depend on the starting point, and the currently dominant scheme of production from methane only makes sense if the energy is initially contained in methane or can easily be transferred to methane. Thus in the case of fossil fuels, the transformation of natural gas into hydrogen is relatively easy and that of oil is a little bit more elaborate, while transformation of coal requires an initial step of high-temperature gasification. For energy already transformed to electricity, electrolysis is presently the most common process for hydrogen production. Among the renewable energy sources, biomass sources require special attention, depending on their form. Photo-induced or direct thermal decomposition of water at high temperatures may be considered, while more complex, multistep schemes are required at lower temperatures, such as those offered by steam from nuclear reactors.

Atomic number, H	1
Electron binding (ionisation) energy in 1s ground state	2.18	aJ
Molar mass, H2	2.016	10-3 kg mol-1
Average distance between nucleons, H2	0.074	nm
Dissociation energy, H2 to 2H at infinite separation	0.71	aJ
Ionic conductance of diluted H+ ions in water at 298 K	0.035	m2 mol-1 Ω-1
Density, H2, at 101.33 kPa and 298 K	0.084	kg m-3
Melting point at 101.33 kPa	13.8	K
Boiling point at 101.33 kPa	20.3	K
Heat capacity at constant pressure and 298 K	14.3	kJ K-1 kg-1
Solubility in water at 101.33 kPa and 298 K	0.019	m3 m-3

Table 2.1. Properties of hydrogen^∗ .

2.1.1 Steam reforming

Current industrial production of hydrogen starts from methane, CH4, which is the main constituent of natural gas. A mixture of methane and water vapour at an elevated temperature is undergoing the strongly endothermic reaction,

$$CH_4 + H_2O \rightarrow CO + 3H_2 - \Delta H^0,$$ (2.1)

where the enthalpy change ∆H⁰ equals 252.3 kJ mol-1 at ambient pressure (0.1 MPa) and temperature (298 K), and 206.2 kJ mol-1 if the input water is already in gas form. The carbon monoxide and hydrogen mixture on the right-hand side of (2.1) is called "synthesis gas". This step requires a catalyst (nickel or more complex nickel on aluminium oxide, cobalt, alkali, and rareearth mixtures) and typically is made to proceed at a temperature of about 850°C and a pressure of around 2.5 × 10⁶ Pa.

The process is controlled by design of the reactor used for the reforming process, by input mixture (typical steam/methane ratios are 2 to 3, i.e., above the stoichiometric requirement), and, as mentioned, by reaction temperature and catalysts. Other reactions may take place in the reformer, such as the inverse of (2.1). In order to obtain a high conversion efficiency, some heat inputs are taken from cooling the reactants, and from the heat outputs derived from the subsequent water-gas "shift reaction" (WGS reaction), usually taking place in a separate reactor,

$$CO + H_2O \rightarrow CO_2 + H_2 - \Delta H^0,$$ (2.2)

∗ For units, prefixes, and conversion factors, see pp. xii-xiv.

with ∆H⁰ equal to —41.1 kJ/mol when all reactants are in gas form at ambient pressure and temperature, and —5.0 kJ/mol if the water is liquid. Heat is recovered and recycled back to the first reaction (2.1). This involves two heat exchangers and is the main reason for the high cost of producing hydrogen by steam reforming. The use of a steam/methane ratio above unity is necessary in order to avoid carbon (char) and excess CO formation (Oh et al., 2003).

Industrial steam reformers typically use direct combustion of a fraction of the primary methane (although other heat sources could, of course, be used) to provide the heat required for the process (2.1),

$$CH_4 + 2O_2 \rightarrow CO_2 + 2H_2O - \Delta H^0,$$ (2.3)

where ∆H⁰ = —802.4(g) or —894.7(l) kJ/mol, (g) for gaseous end products and (l) for liquid, condensed water (the CO2 is supposed to stay gaseous at the relevant process temperatures and pressures). The heat evolution with products in gas form, 802.4 kJ/mol, is called the "lower heat value" of methane, while the one including heat of condensation to liquid form is called the "upper heat value". Only a modest fraction of the heat required for (2.1) is obtained from recycling. The reactions (2.1) and (2.2) combine to

$$CH_4 + 2H_2O + \Delta H^0 \rightarrow CO_2 + 4H_2$$ , (2.4)

with ∆H0 = 165 kJ/mol for gaseous water input and ∆H0 = 257.3 kJ/mol for liquid water input. As a chemical energy conversion process, (2.4) has an ideal efficiency of 100%: the sum of the burning value for CH4, 894.7 kJ/mol ((2.3) with gaseous CO2 but all steam condensed) and the required heat input to the process (2.4), 257.3 kJ/mol, exactly equals the 1152 kJ/mol burning value of 4H2 (again steam condensed). The enthalpies depend on temperature and pressure, but considering the process as starting and ending at ambient temperature and pressure, then the theoretical efficiency does not change by additional heating and pressurising, since the energy spent in these processes could be regained at the end. In practice, only some 50% of the energy from combustion heat input contributes to the chemical process (2.1), and the remaining heat gets transferred to the product stream and is eventually recaptured and recycled, adding further to the capacity requirement for heat exchangers (Joensen and Rostrup-Nielsen, 2002). Kinetic effects influencing the catalyst action of the shift reaction (2.2) have been studied by Ovesen et al. (1996). The actual efficiency of industrial steam reforming of methane is rarely above 80%. As mentioned, the need for heat recycling is a dominating cost item reflected in the hydrogen product cost. If there is sulphur contamination, for example, by H2S, of the input methane used for (2.1) above the limit set by the effect on the integrity of the catalysts, then a sulphur removal step must precede the synthesis gas step.

The role of the catalyst − say, a simple nickel metal surface with periodic lat-

tice structure − is to adsorb methane molecules and step-by-step strip them of hydrogen atoms, ready for the following conversion steps. The basic mechanism of the surface dissociation was already suggested by Lennard-Jones (1932), who postulated the presence of a double-minimum potential energy as a function of the distance between the surface and the impinging molecule, as illustrated in Fig. 2.1. The outer minimum is due to molecular adsorption by the long-range Coulomb forces (sometimes referred to as van der Waals forces), while the inner minimum is due to dissociative chemisorption. The barrier between the two minima may have a height exceeding the potential energy of the molecule at large distances, in which case the molecule must possess enough kinetic energy to be able to penetrate the barrier (Ceyer, 1990). The thermal motion associated with elevating the process temperature achieves this. Clearly, the height of the potential barrier depends on both surface material and impinging molecule, and thus allows engineering by selection of catalyst. New catalysts with increased levels of hydrogen production have been brought into use, from the traditional Ni catalysts on Al2O3 (Yokota et al., 2000) over Ni/ZrO2 (Choudhary et al., 2002) to a recent Ni/Ce-ZrO2 catalyst, which yields 15% more hydrogen than implied by the reaction (2.1), because the excess steam used in the first reaction step contributes hydrogen production by the shift reaction (2.2), even in the first process reactor (Roh et al., 2002).

Figure 2.1. Dissociation of a molecule at a catalyst surface may require enough kinetic energy to overcome a potential energy barrier. Without this extra energy the molecule just gets adsorbed to the surface (Lennard-Jones, 1932).

The elevated temperatures used in the reactor may damage the catalysts. Of particular importance is the possible carbon formation by methane cracking:

$$CH_4 + 74.9 \text{ kJ/mol} \rightarrow C + 2H_2.$$ (2.5)

This carbon may appear as filaments on the catalyst surface, encapsulating catalyst atoms and largely blocking the steam reforming reaction (Clarke et al., 1997). As previously mentioned, excess steam is effective in preventing cracking or at least diminishing the negative effects by cleaning the catalyst surfaces.

The shift reaction (2.2) employs additional sets of catalysts, traditionally Fe or Cr oxides, but in newer plants with staged-temperature processing these would be used only at the first step at around 400°C, while new catalysts such as Cu/ZnO/Al2O3 are used for a second low-temperature step, following heat removal for the recycling just mentioned. Carbon contamination at the catalysts can proceed via the Boudouard reaction (Basile et al., 2001),

$$2CO \rightarrow C + CO_2 + 172.4 \text{ kJ/mol.}$$ (2.6)

New catalysts with higher stability against such effects are being sought. The final CO2 removal is, on the other hand, a fairly simple process that can be achieved by water sprinkling. Current natural gas steam reforming in large industrial plants is not aimed at fuel cell use of the hydrogen produced, and the CO contamination remaining after the shift reaction stays in the range of 0.3 to 3% (Ghenciu, 2002; Ladebeck and Wagner, 2003).

For use in proton exchange membrane (PEM) fuel cells (see section 3.6), the CO contamination in the hydrogen produced must be below 50 ppm (parts per million). This is due to the poisoning limit of typical platinum catalysts used in PEM cells. The implication is the need for a final CO cleaning treatment, unless the main reaction steps (2.1) and (2.2) can be controlled so accurately that all reactants are accounted for. This CO cleaning stage may involve one of the following three techniques: preferential oxidation, methanation (the reverse of (2.1)), or membrane separation (Ghenciu, 2002). The process currently in most widespread use is partial oxidation,

$$CO + \frac{1}{2}O_2 \rightarrow CO_2 + 283.0 \text{ kJ/mol},$$ (2.7)

where "preferential" means that catalysts are applied, which favour this reaction over the corresponding hydrogen oxidation,

$$H_2 + \frac{1}{2}O_2 \rightarrow H_2O - \Delta H^0$$ , $\Delta H^0 = -242(g)$ or $-288(l)$ kJ/mol, (2.8)

(enthalpies at ambient temperature and pressure; in (2.7) CO2 is assumed to be gaseous). Catalysts currently used are based on Pt or platinum alloys. They require high surface areas and low temperature (80-200°C), with smaller being better (Shore and Farrauto, 2003).

A way to avoid some of the problems mentioned while maintaining a high efficiency is to continually remove the hydrogen produced from the second reactor. This can be done by using a membrane allowing hydrogen molecules to pass, but none of the other reagents. A membrane reactor can replace both reactors of the conventional steam reforming process in achieving the overall reaction (2.4). It may consist of a tubular reactor with packed catalysts (Pd catalysts have been employed in several pilot studies) adjacent to the membrane and inlet flow channel. The removal of hydrogen produced will shift the thermodynamical equilibrium of (2.1) and (2.2) towards the product side, allowing a lowering of reaction temperature and the use of less excess steam (Kikuchi et al., 2000; Gallucci et al., 2004). A cleaning step has, in some cases, been needed, but CO contamination can both theoretically and in practice be kept as low as 0.001%, as recently demonstrated (Yasuda et al., 2004; at a total efficiency of 70%). The reactor may be small in size relative to a comparable, conventional steam reforming reactor, and the efficiency of conversion higher, at least in the laboratory experiments performed.

Reforming hydrocarbons other than methane, notably oil products, can be achieved with reactions similar to (2.1):

$$C_n H_m + n H_2 O \to n CO + (n + \frac{1}{2}m) H_2 - \Delta H^0.$$ (2.9)

The successive breaking of C-C terminals of higher hydrocarbons with suitable catalysts is more involved than for methane, due to differences in reaction rate and increased problems with thermal cracking (in analogy to (2.5), also called pyrolysis). To avoid this problem, the carbon stripping is usually done in a separate pre-reformer (Joensen and Rostrup-Nielsen, 2002), making this way of producing hydrogen more complex and more expensive than natural gas reforming. It is also possible to employ new types of catalysts (such as Pd/ceria) in order to reform specific higher hydrocarbons with acceptable catalyst integrity (Wang and Gorte, 2002).

2.1.2 Partial oxidation, autothermal, and dry reforming

The moderately exothermic catalytic partial oxidation process for methane,

$$CH_4 + \frac{1}{2}O_2 \rightarrow CO + 2H_2 + 35.7 \text{ kJ/mol},$$ (2.10)

or more generally,

$$C_n H_m + \frac{1}{2} n O_2 \rightarrow n CO + \frac{1}{2} m H_2 - \Delta H^0,$$ (2.11)

is considerably more rapid than steam reforming. When oxygen (usually in air) and methane are passed over a suitable catalyst (e.g., Ni/SiO2), the reaction (2.10) occurs, but to a degree also full oxidation (2.3) and product oxidation ((2.7) and (2.8)), opening up for further processes such as methanation (the reverse of (2.1)), the shift reaction (2.2), and dry reforming (see (2.14) following). When oxygen is supplied through air, nitrogen has to be removed from the hydrogen product. This usually takes place in a separate stage following the oxidation reactor. Partial oxidation is suitable for smallscale conversion, such as in a motor vehicle with fuel cells. The process can be stopped and started, as required for on-board reformation, and when in progress, it provides elevated temperatures that may start steam reforming along with the oxidation processes. This is called "autothermal" reforming and involves all the reactions mentioned so far, plus stoichiometric variations on (2.10) in the possible presence of water,

$$CH_4 + \frac{1}{2} x O_2 + (1 - x) H_2O \rightarrow CO + (3 - x) H_2 - \Delta H^0,$$ (2.12)

written in a way suitable for x < 1, which need not be fulfilled, as the following special case shows:

$$CH_4 + \frac{3}{2}\,O_2 \rightarrow CO + 2H_2O - \Delta H^0, \ \ \Delta H^0 = 611.7(l) \ or \ 519.3(g) \ kJ/mol.(2.13)$$

The use of air is convenient for automotive applications, but if the autothermal scheme were to be used for large industrial plants, dedicated oxygen production for the feed would be preferable−for example, due to concerns over the volume of N2 to be handled and the associated size of heat exchangers (Rostrup-Nielsen, 2000).

Hydrogen production by partial oxidation from methane increases with process temperature, but reaches a plateau value at around 1000 K (Fukada et al., 2004). The theoretical efficiency is similar to that of conventional steam reforming, but less water is required (Lutz et al., 2004).

Gasoline and other higher hydrocarbons may be converted to hydrogen on board cars by the autothermal processes, using suitable catalysts (Ghenciu, 2002; Ayabe et al., 2003; Semelsberger et al., 2004). Partial oxidation may also be combined with the palladium-catalyst membrane reactors mentioned in section 2.1.1 (Basile et al., 2001).

As an alternative to conventional steam reforming, methane could be reformed in a stream of carbon dioxide rather than of water steam,

$$CH_4 + CO_2 \rightarrow 2CO + 2H_2 + 247.3 \text{ kJ/mol},$$ (2.14)

again followed by the shift reaction (2.2). This reaction may be used for disposal of CO2, for example, in situ during coal extraction (Andrés et al., 2011), and also for operating at fairly low temperatures, when this is desirable for conventional steam reforming (Abashar, 2004).

2.1.3 Water electrolysis: reverse fuel cell operation

The conversion of electric energy into hydrogen (and oxygen) by water electrolysis has been known for a long time (demonstrated by Faraday in 1820 and widely used since about 1890), but if the electricity is produced by use of fossil fuels, then the cost of hydrogen obtained in this way is higher than

the one associated with steam reforming of natural gas. On the other hand, high hydrogen purity needed in some applications is easier to achieve, and for this reason electrolysis hydrogen currently has a market share of about 5%. As the cost of electricity dominates the cost of hydrogen production from electrolysis, the situation is quite different, if the electricity used is surplus electricity from variable resources such as wind or solar radiation. In cases where such electricity is produced at times of no local demand and no evident option for export to other regions, the value of this electricity may be seen as zero. This makes it very favourable to store the energy for later use, and storage in the form of hydrogen is one option that may be favourable, provided that uses of hydrogen or regeneration of electricity are economically attractive in a specific situation.

Conventional electrolysis uses an aqueous alkaline electrolyte, for example KOH or NaOH at a weight percentage of around 30, with the positive and negative electrode areas separated by a microporous diaphragm (replacing earlier asbestos diaphragms). The overall reaction at the positive electrode^∗ (e.g., Ni or Fe) is

$$H_2O \rightarrow \frac{1}{2}O_2 + 2H^+ + 2e^-,$$ (2.15)

where electrons are leaving the cell by way of the external circuit (Fig. 2.2) and where the three products may be formed by a two-step process,

$$2H_2O \rightarrow 2HO^{-} + 2H^{+} \rightarrow H_2O + \frac{1}{2}O_2 + 2e^{-} + 2H^{+}.$$ (2.16)

The reaction at the negative electrode is

$$2H^+ + 2e^- \to H_2,$$ (2.17)

grabbing electrons from the external circuit. The hydrogen ions are to be transported through the electrolyte by the electric potential difference V. The role of the alkaline component is to improve on the poor ion conductivity of water, favouring the use of KOH. However, this limits process temperature to values below 100°C, in order to avoid strong increases in alkali corrosion of electrodes. The overall reaction is the inverse of (2.8),

$$H_2O - \Delta H^0 \rightarrow H_2 + \frac{1}{2}O_2$$ , $\Delta H^0 = -242 (g)/288 (l) kJ/mol$ (2.18)

$$\Delta H = \Delta G + T \Delta S. \tag{2.19}$$

At ambient pressure and temperature (298 K), the change in enthalpy and free energy for liquid water is ∆H = —288 kJ mol-1 and ∆G = 236 kJ mol-1. The electrolysis process thus requires a minimum amount of electric (highquality) energy of 236 kJ mol-1, while the difference between enthalpy and

^∗ See section 3.1 for a discussion of the names anode and cathode sometimes used in connection with the two electrodes of an electrochemical system.

free energy changes, ∆H − ∆G, in theory could be heat taken from the surroundings. Since the apparent conversion efficiency would then be ∆H/∆G, since −∆H is also the (upper) heating value of H2, it could theoretically exceed 100% by as much as 22%. However, with the use of ambient heat at 25°C, the process would be very slow. Temperatures used in classical electrolysers are around 80°C, and in some cases active cooling has to be applied. The much lower conversion efficiency (50-70%) obtained with simple electrolysers in practical cases is largely a consequence of electrode "overvoltage", mainly stemming from polarisation effects. The cell potential V for water electrolysis may be expressed by

Figure 2.2. Schematic layout of a water electrolysis cell.

$$V = V_r + V_a + V_c + R j, (2.20)$$

where Vr is the reversible cell potential,

$$V_r = -\Delta G/(z\mathcal{F}) = 1.22 \text{ V}, \tag{2.21}$$

using the free energy change ∆G, the Faraday constant F = 96 493 C mol-1, and the number z of electrons involved in the reaction, here (2.15). The three next terms in (2.20), which constitute the overvoltage, have been divided into negative ("anode") and positive ("cathode") electrode parts Va and Vc, and the resistance contribution. The current is j, and R is the internal resistance of the cell. The three last terms in (2.20) represent electrical losses, and the voltage efficiency ηV of an electrolyser operating at a current j is defined by

$$\eta_V = V_r / V. \tag{2.22}$$

Efforts are being made to increase the efficiency to 80% or higher. One method is to increase the operating temperature, typically to above 1500°C, and to optimise electrode design and catalyst choice.

Figure 2.3. Sodium chloride solution near a metal surface (at bottom). Single time-step results of molecular dynamics simulation for a positive surface charge, From E. Spohr (1999). Molecular simulation of the electrochemical double layer. Electrochimica Acta 44, 1697-1705, reproduced with permission from Elsevier.

The alkaline electrolysers described here are one type of fuel cells (described as such in section 3.4). Although a detailed discussion of the quantum chemical processes involved in such devices is the subject of Chapter 3, some explanation will be given here of the mechanisms responsible for gas dissociation and the electrode losses appearing in (2.20). How do the water molecules and ions of the electrolyte behave near the electrode surfaces, or on the surface of any catalysts placed on (or near) the electrodes in order to speed up the transfers? Experimental methods such as tunnelling electron micro-scopy (TEM) and a range of spectral measurements allow the establishment of visual impressions of molecular structure that were unthinkable to early constructors of electrochemical devices. Furthermore, recent advances in semi-classical and quantum chemistry allow insights into the mechanisms of electron transfer on a level of detail that was unheard of until the 1990s.

Figure 2.4. Charge density as a function of distance from the electrode, for the electrolyte modelled in Fig. 2.3. The three cases represent positive, zero, and negative electrode charge. Solid lines are total charge densities, longer dashes are ionic charge densities, and shorter dashes are aqueous charge densities. The curves have been smoothed. From E. Spohr (1999). Molecular simulation of the electrochemical double layer. Electrochimica Acta 44, 1697-1705, reproduced with permission from Elsevier.

Consider a metal surface (electrode or catalyst) with a regular lattice atomic structure. As an electrode in an electrolyser, the external voltage applied to the system causes a build-up of charges on the metal surface. In the classical description of electrochemical devices, charges of the opposite sign, derived from ions in the electrolyte, are drawn towards the electrodes to form a double layer of opposite charges. At the positively charged electrode, a layer of negative ions builds up in the electrolyte, along the electrode surface. This is then the environment in which the water splitting takes place. To model the structure in this double-layer region, molecular mechanics has been used.

"Molecular mechanics" is a classical mechanics approximation to a description of the motion of atomic or molecular centres, neglecting electrons entirely except for keeping track of charges, but retaining the possibility of a non-symmetric shape of each molecule, that may be characterised by its orientation in space. The molecules are assumed to interact via a heuristic force, given by parametrised potential functions such as the one depicted in Fig. 2.1. Figure 2.3 shows an example (frozen in time) of the size of problems that can be treated in this way: there are 400 water molecules and 32 ions (Na⁺ or Cl- ) in a rectangular space, with one charged side representing the electrode. The figure shows that the calculation provides patterns that look realistic. Averaging over the total simulation time of a few nanoseconds, the charge density variations as a function of distance from the electrode can be established, as shown in Fig. 2.4. The three cases shown are the results of calculations for a positively, zero, and negatively charged electrode, respectively. It is seen that, as expected, the Cl⁻ ions accumulate near a positively charged electrode and the Na⁺ ions accumulate near a negative charged electrode, but, in addition, the water molecules have become polarised so that the total charge as a function of distance from the electrode is oscillating as a result of interference between the ionic and water-related charges. The model is of course limited by its use of a phenomenological short-range interaction, which is the same for water and ions, by the molecular dynamics calculation and by not treating the specific quantum reactions that may take place at the electrode surface (Spohr, 1999).

In an electrolyser, the charged double layers (or more complex oscillatory charge densities) exist at both electrodes, and the ions and water molecules in the electrolyte have a thermal distribution of kinetic energies, depending on the operating temperature. The kinetic energy causes impact of the reactants on the electrodes, which again may change the rate of reactions. A quantum mechanical description thus entails first the determination of electron density distributions and potential energies along possible paths of reaction, and then a time-dependent solution of the Schrödinger equation or a semi-classical approximation to it, for reactant molecules in each of the states possible within their thermal distribution. These can be determined by finding the possible vibrational excited states of each molecule involved.

Figure 2.5 shows the potential energy surface of the simpler reaction (2.17) at the negative electrode, as a function of the distance z of the hydrogen molecule from the surface and the distance d between the two hydrogen atoms. Quantum calculations have been performed as a function of all the six hydrogen co-ordinates (three for each atom), and it has been found that a hydrogen-pair position parallel to the surface has the lowest energy. Near the surface, centres of mass for the hydrogen molecule may approach top surface metal atoms or the "holes" between them. In the case of a platinum negative electrode, the first possibility seems to provide the easiest reaction path at larger distances (Pallassana et al., 1999; Horch et al., 1999), but once separated, the hydrogen atoms appear with their centre of mass at the "hole" site, that is, the face centre in case of the cubic (fcc) Pt lattice pattern with Miller indices (111)¤ . In Fig. 2.5, the approach of the hydrogen pair is along a vertical line from the fcc position for a Ni(111) lattice.

Figure 2.5. Potential energy surface in eV along two co-ordinates for hydrogen dissociation on a Ni catalyst surface. The abscissa is the distance d between the two hydrogen atoms, and the ordinate is the distance z of the hydrogen pair from the second layer of the Ni surface. The pair is assumed parallel to the surface. The quantum chemical calculation of the potential energies employs the density functional theory B3LYP (see section 3.2) and a set of 508 basis functions (SV) for 2 × 337 electrons in 26 atoms, including two layers of Ni atoms and the two hydrogen atoms, positioned as shown in the top-view insert (Sørensen, 2004b).

Figure 2.5 shows that the hydrogen molecule may approach the surface at a fixed separation of d = 0.074 nm (equal to the experimentally determined separation in an isolated hydrogen molecule) down to a distance of about z = 0.5 nm from the lower surface. There, it needs to pass through a saddle point in the potential energy (which is possible due to the vibrational energy

The three Miller indices are the reciprocal of the displacements along the three major axes defining the unit cell of a lattice structure (i.e., the distance to the position of the corresponding atom of the next unit cell in each direction). See solid state physics textbooks such as Kittel (1971).

associated with a finite temperature, kinetic energy, or quantum tunnelling) in order to reach the potential energy valley associated with the dissociated state at z = 0.317 nm. A steeply rising potential energy prevents the hydrogen atoms from getting closer to the surface, although there are variations in the potential as a function of the position of the Ni atoms. The calculated saddle point (transition structure) barrier height is about 0.90 eV, as seen from the hydrogen molecular state, and d is only slightly increased. The final energy minimum once the hydrogen atoms have been separated is about 0.33 eV below the initial one. The separation between the hydrogen atoms at this minimum is around d = 0.1 nm, but the energy stays flat for increasing d.

Figure 2.6. Potential energy surface in eV along two co-ordinates for oxygen dissociation on a Ni catalyst surface. The abscissa is the distance d between the two oxygen atoms, and the ordinate is the distance z of the oxygen pair from the second layer of the Ni surface. The pair is assumed parallel to the surface. The quantum chemical calculation of the potential energies employ Hartree-Fock for the Ni atoms and adds correlations from density functional theory (B3LYP, see section 3.2) for the oxygen atoms, a set of 508 basis functions (SV) for 2 × 344 electrons in 26 atoms, including two layers of Ni atoms and the two oxygen atoms, positioned as shown in the topview insert. Like in Fig. 2.5, the absolute value of the energy scale is taken arbitrarily, so that only potential energy differences matter (Sørensen, 2004b).

At the positive electrode, the pair of oxygen atoms seem to prefer separating

over a face centre of a Pt-lattice, but the published quantum calculations are not able to reproduce the experimental energies well (Eichler and Hafner, 1997; Sljivancanin and Hammer, 2002). The precise nature of the surface is important, and the occurrence of steps in the crystal structure termination at the surface can greatly enhance the dissociation of, for example, oxygen molecules on a platinum surface (Stipe et al., 1997; Gambardella et al., 2001). Similar effects have also been found for hydrogen adsorption (Kratzer et al., 1998). Figure 2.6 gives the potential energy surface for two O-atoms over a Ni surface, along co-ordinates similar to those of Fig. 2.5. Dissociation at the surface is here clearly demonstrated. Further discussion of the quantum calculations for oxygen formation or dissociation taking place during water electrolysis or fuel cell operation will be given in sections 3.1 and 3.2.

Finite temperature effects may be calculated on top of the stationary quantum calculations of electronic structure, either by classical molecular dynamics calculations (with the argument that quantum effects are less important for the atomic nuclei than for the electrons) or by Monte Carlo calculations based on sampling path integrals (Weht et al., 1998) or plane wave density functional methods (Reuter et al., 2004). The path integral method is a general method of treating many-body systems in a quantum statistical way. It involves calculating a two-point density matrix, which under certain conditions may be expressed in terms of effective potentials from calculations such as those used in Figs. 2.5-2.6. The Monte Carlo technique is then used to determine the average thermal behaviour of the system.

The overvoltage introduced in connection with (2.20) can now be interpreted in terms of two identifiable deviations from the potential determined by overall reaction energies. One is caused by the non-static situation of an electrochemical device, in which a current flows through the electrolyte, causing ions to interact with the molecules they pass. The other, occurring at each electrode, is connected with the polarisation build-up, as illustrated in Fig. 2.4 and involving both water and ions in the electrolyte. The dynamic nature of both these effects imply that the overpotential could be defined as the difference between the real value and that for a completely static situation without any currents across the device (Hamann et al., 1998). With interactions between the primary reactants and the surrounding molecules, the simple potentials depicted in Figs. 2.1, 2.5, and 2.6 could, strictly speaking, no longer be drawn as a function of a single variable.

The effect of the solvent can be modelled at various levels of sophistication. The simplest is to consider the molecule studied as surrounded by an infinite, homogeneous dielectric medium (the solvent) and interacting only via its total dipole moment. To avoid infinities in the numerical calculations, this also involves invoking a void (cavity) around the molecule studied, so the distance between a point in the solvent and a point belonging to the molecule is never zero. This dipole model was first proposed by Onsager (1938). Refinements include replacing Onsager's fixed spherical cavity by the unification of spheres around each atom of the molecule and introducing various parametrisations of the dipole interaction. At the other extreme, each molecule of the solvent might be included in the quantum chemical calculations, which, of course, would then exclude use of the most detailed models.

Considering current commercial alkaline water electrolysers with 25-30% KOH and Ni electrodes, typical overpotentials are Va + Vc = 0.32 V and Rj = 0.22 V at j = 0.2 A m-6, to be added to the reversible cell potential of 1.19 V (Andreassen, 1998).

Figure 2.7. Schematic layout of proton exchange membrane electrolysis cell.

Improvement of electrolyser efficiency involves the design of catalysts able to promote high rates of water dissociation at the positive electrode and high rates of hydrogen recombination at the negative electrode. The liquid electrolyte may be replaced by solid polymer membrane electrolytes, offering the possibility of higher stability and longer life than the liquid electrolytes. Overpotentials as low as 0.016 V have been obtained for currents up to 10⁴ A m⁻² (Rasten et al., 2003). It is also possible to provide the water feed at only one place (the negative electrode side in contrast to the example illustrated in Fig. 2.2), choosing a membrane that allows water penetration to the positive electrode side. The water penetration in one direction and the H+ migration in the opposite direction can be balanced so that the oxygen is produced free of water (Hamilton Sundstrand, 2003). The hydrogen gas still contains water that has to be separated through a cleaning stage.

As mentioned, higher process temperatures may also increase efficiency. This is not possible with simple liquid electrolytes such as the KOH solution, but it can be achieved with suitable solid oxide membranes similar to those used in solid oxide fuel cells and accepting water input in the form of steam (Dutta et al., 1997; recent progress reported by Ni et al., 2008).

An interesting idea investigated by Cheng et al. (2002) is to subject the entire electrolysis device to a constant, fast rotation. The centrifugal acceleration lowers the overvoltage and improves efficiency more than it is lowered by having to spend energy on maintaining the rotation.

Since hydrogen is often to be distributed at elevated pressures (e.g., in containers for vehicle use), a compression step may be avoided by carrying out already the electrolysis at high pressure. This requires suitable protection against explosion risks (Janssen et al., 2004).

All types of fuel cells may be operated in reverse mode to split water. Proton exchange membrane fuel cells are being developed both for electrolysis and for two-way operation, at efficiencies for electrolysis from 50 to 95% for one- or two-way systems (Shimizu et al., 2004; Agranat and Tchouvelev, 2004; see section 3.5.5). Optimisation for hydrogen production involves the use of larger membrane areas, a low number of cells in a stack, and efficient hydrogen removal ducts (Yamaguchi et al., 2001). Figure 2.7 gives an overview of the layout of a proton exchange membrane electrolyser. Recently, new catalysts have been developed, allowing high efficiency for operation in either direction (Ioroi et al., 2002). Fuel cells for direct or reversed operation are discussed in more detail in Chapter 3.

2.1.4 Gasification and woody biomass conversion

An emerging technology for producing hydrogen from natural gas or heavy fuel oil, albeit with substantial electricity inputs, is high-temperature plasma-arc gasification, based on which a pilot plant operates on natural gas at 1600°C at Kvaerner Engineering in Norway (Zittel and Würster, 1996). The resulting products are in energy terms: 48% hydrogen, 40% carbon, and 10% water vapour. Since all the three main products are useful energy carriers, the conversion efficiency may be said to be 98% minus the modest amounts of energy needed for the process. However, conversion of natural gas to carbon is not normally desirable, and the steam can be used only locally, so quoting a 48% efficiency is more meaningful. Concentrating solar collector energy input to decomposition of natural gas is investigated as an alternative, but expensive, route (Hirsch and Steinfeld, 2004; Dahl et al., 2004).

Gasification is seen as a key pathway towards hydrogen when starting from coal or lignin-containing biomass (wood, wood scrap, or other solidstructure plant material). The gasification takes place by heating with steam:

$$C + H2O \rightarrow CO + H2 - \Delta H0, \qquad (2.23)$$

where ∆H⁰ is 138.7 kJ/mol for water already in the gas phase. In air, competition may derive from the oxygen gasification (combustion) process

$$C + O_2 \rightarrow CO_2 - \Delta H^0, \tag{2.24}$$

with ∆H⁰= −393.5 kJ/mol for CO2 in the gas phase. Additional processes include the Boudouard process (2.6) and the shift reaction (2.2). For biomass, the carbon is initially contained in a range of sugar-like compounds, such as the cellulose materials listed in Table 2.2. Without catalysts, the gasification takes place at temperatures above 900°C, but use of suitable catalysts can bring the process temperature down to around 700°C. If additional hydrogen is to be produced by the shift reaction, this has to take place in a separate reactor operating at a temperature of about 425°C (Hirsch et al., 1982).

Figure 2.8. Schematic diagram of catalytic gasification process (SNG: synthetic natural gas). From Hirsch et al. (1982). Science 215, 121-127, reprinted with permission. Copyright 1982 American Association for the Advancement of Science.

Coal may be gasified in situ, that is, before extraction. If the coal has already been mined, the traditional methods include the Lurgi fixed-bed gasifier (providing gas under pressure from non-caking coal at a conversion efficiency as low as 55%) and the Koppers−Totzek gasifier (oxygen input, gas at ambient pressure, also of low efficiency). By using suitable catalysts, it is possible to shift the temperature regimes and combine all the processes within a single reactor, an example of which is shown in Fig. 2.8. In some cases, such as waste feedstock containing plastic, pyrolysis is an alternative to gasification (see, e.g., Ahmed and Gupta, 2009).

Peat and wood can be gasified in much the same way as coal. Wood gasification has a long history. The processes may be viewed as "combustionlike" conversion, but with less oxygen available than needed for burning. The ratio of oxygen available and the amount of oxygen that would allow complete burning is called the "equivalence ratio". For equivalence ratios below 0.1, the process is called "pyrolysis", and only a modest fraction of the biomass energy is found in the gaseous product, with the rest being in char and oily residues. If the equivalence ratio is between 0.2 and 0.4, the process is called a proper "gasification". This is the region of maximum energy transfer to the gas (Desrosiers, 1981). Table 2.2 lists a number of reactions involving polysaccharidic material, including pyrolysis and gasification. In addition to the chemical reaction formulae, the table gives enthalpy changes for idealised reactions (i.e., neglecting the heat required to bring the reactants to the appropriate reaction temperature). The specific heat of the material is 3 kJ g-1 wood at the peak of energy in the gas, increasing to 21 kJ g-1 wood for combustion at an equivalence ratio equal to unity. Much of this sensible heat can be recovered from the gas, so that process heat inputs for gasification can be kept low. The energy content of the gas produced peaks at an equivalence ratio of about 0.25 and falls fairly rapidly both below and above this ratio (Reed, 1981; Sørensen, 2010a).

Chemical reaction	Energy consumed (kJ	g−1 ) a Products / process
→ 6C + 5H2 + 2.5 O2 C6H10O5	5.94b	elements, dissociation
→ 6C + 5H2O(g) C6H10O5	−2.86	charcoal, charring
→ 0.8 C6H8O + 1.8 H2O(g) + 1.2 CO2 C6H10O5	−2.07c	oily residues, pyrolysis
→ 2C2H4 + 2CO2 + H2O(g) C6H10O5	0.16	ethylene, fast pyrolysis
C6H10O5 + ½O2 → 6CO + 5H2	1.85	synthesis gas, gasification
C6H10O5 + 6H2 → 6"CH2" + 5 H2O(g)	−4.86d	hydrocarbons, −generation
C6H10O5 + 6O2 → 6CO2 + 5 H2O(g)	−17.48	heat, combustion

a Specific reaction heat.

b The negative of the conventional heat of formation calculated for cellulose from the heat of combustion of starch.

Calculated from the data for the idealised pyrolysis oil C6H8O (∆Hc = —745.9 kcal mol⁻¹ , ∆Hf = 149.6 kcal g⁻¹ , where Hc = heat of combustion and Hf = heat of fusion). d

Calculated for an idealised hydrocarbon with ∆Hc as above. H2 is consumed.

Figure 2.9 gives the equilibrium composition calculated as a function of the equivalence ratio. Equilibrium composition means that the composition of reaction products occurring after the reaction rates and reaction temperature have stabilised adiabatically. The actual processes are not necessarily adiabatic; in particular the low-temperature pyrolysis reactions are not. For the cellulosic wood considered in Table 2.2, the average ratios of carbon, hydrogen, and oxygen are 1:1.4:0.6. Figure 2.10 shows three examples of wood gasifiers: the updraft, the downdraft, and the fluidised bed types.

Figure 2.9. Calculated gas composition resulting from equilibrium processes between air and biomass, as a function of equivalence ratio. (From T. Reed (1981). Biomass Gasification, reprinted with permission. Copyright 1981, Noyes Data Corporation.)

The drawback of the updraft type is a high rate of oil, tar, and corrosive chemical formation in the pyrolysis zone. This problem is solved by the downdraft version, where such oils and other matter pass through a hot charcoal bed in the lower zone of the reactor and become cracked to simpler gases or char. The fluidised bed reactor may prove superior for large-scale operations because passage time is smaller. The drawback of this is that ash and tars are carried along with the gas and have to be removed later in cyclones and scrubbers. Several variations on these gasifier types have been suggested (Drift, 2002; Gøbel et al., 2002). Ni-based catalysts are used for both steam and dry reforming (Courson et al., 2000). Practical ways of continuous hydrogen production by fluidised bed gasification are under investigation−for example, within the Japanese energy programme (Matsumura and Minowa, 2004). For complex types of biomass, such as mixed urban waste or agricultural residues, gasification often leads to a gas of low hydrogen content (25-50%), requiring additional reaction or purification steps (Mérida et al., 2004; Cortright et al., 2002).

Figure 2.10. Gasifier types: (a) updraft; (b) downdraft; and (c) fluidised bed (from B. Sørensen, Renewable Energy, 2004a, used with permission from Elsevier).

The gas produced by gasification of biomass is a "medium-quality gas" with a burning value in the range of 10-18 MJ m⁻³ . It may be used directly in Otto or diesel engines or to drive heat pump compressors. However, in order to serve as a source of useful hydrogen, it has to be upgraded to pipelinequality gas (about 30 MJ m-3) and a purity, depending on the application envisaged. If methanol is the desired fuel, as it might be for some fuel cell operations, the following reaction is carried out at elevated pressure:

$$2H_2 + CO \rightarrow CH_3OH. \tag{2.25}$$

The mixture on the left-hand side of (2.25) is called "synthesis gas". More details on the methanol production from biomass may be found in Sørensen (2010a).

There are environmental impacts to consider in connection with biomass production, collection (e.g., by forestry), and transport to the gasification site, from the gasification and related processes and finally from the use made of the gas. The gasification residues−ash, char, liquid wastewater, and tar−have to be disposed of. Char may be recycled to the gasifier, while ash and tars could conceivably be utilised in the road or building construction industry.

Figure 2.11 gives an overview of the conversions of various types of biomass, including the gasification and pyrolysis processes described in this section. Biological production by fermentation and other bacterial processes will be described in section 2.1.5, while direct splitting by light or by heat will be covered in sections 2.1.6 and 2.1.7. The fluid "end" products such as ethanol or methanol may be further converted into hydrogen, although it is presumably more appropriate to use them directly, for example as fuels in the transportation sector. Of course, methane can also be used in hightemperature fuel cells such as solid oxide or molten carbonate fuel cells, without further conversion into hydrogen.

Figure 2.11. Pathways for non-food uses of biomass (based on Sørensen, 2010a).

2.1.5 Biological hydrogen production

Production of hydrogen from biomass may be achieved by biological fermentation or by other bacterial or algae decomposition of water or another suitable substrate. The conversion processes may proceed in the dark or with the assistance of light. Growing the biological substance in the first place requires energy input, normally from sunlight, and there are thus several conversion efficiencies involved: from primary energy source to biological material, from energy in biological material to energy in hydrogen produced, or the overall efficiency from solar radiation to hydrogen. Because production of molecular hydrogen is rarely the purpose of natural biological systems, they have to be modified in order to serve this purpose, e.g., by genetic engineering. Plants have very low efficiencies of converting solar energy, which poses stringent cost requirements on the equipment to be used.

Some primary sources ("substrates") for biological hydrogen production include those listed in the top area of Fig. 2.11. To these should be added water in case of the direct photolysis process (discussed in section 2.1.6) and as the primary substrate for photosystem II discussed following. In these cases the biological system may directly produce hydrogen (which then has to be prevented from subsequent use by the biological organism itself). Some organisms instead use an indirect route to produce hydrogen, starting not from water, but from an organic compound such as sugar, that requires less energy to complete the conversion. In organic waste fermentation, a succession of decomposition processes precedes the hydrogen forming process. Many of the reactions depend on enzymes in order to work. There are enzymes in biological systems tailored to catalyse nitrogen to ammonia conversion (nitrogenases), with hydrogen as a possible by-product and hydrogen oxidation enzymes (uptake hydrogenases) removing any hydrogen produced if not prevented. There are also bi-directional enzymes performing either function and, of course, enzymes catalysing other processes.

Photosynthesis

The overall process of splitting water,

$$2H_2O \rightarrow O_2 + 4H^+ + 4e^- - \Delta H^0,$$ (2.26)

requires an energy input of ∆H⁰ = 590 kJ/mol or 9.86 × 10-19 J = 6.16 eV for the molecular process (2.26) splitting two water molecules. The solar spectrum, shown in Fig. 2.12, and the plant absorption spectrum, shown in Fig. 2.13, are such that in most cases four light quanta are required for delivering enough energy. Because of this, green plants have had to develop an intricate system for collecting light, for transferring it to the site of water splitting, and for further collecting and storing the energy in a form useful to subsequent biological processes. All of these processes take place in a photosynthesis complex, consisting of two photosystems and a number of auxiliary components, organised in and around a membrane structure similar to what would be needed in non-biological water splitting systems for separating the oxygen and hydrogen produced. The risk of recombination is usually prevented by not producing free molecular hydrogen at all, but rather some other energy-containing substance. Alternatively, oxygen and hydrogen should be on different sides of the membrane.

Figure 2.14 shows the components of the photosystems residing along the thylakoid membranes. Insights into the structure of the membrane and component architecture have been obtained by atomic force microscopy (Bahatyrova et al., 2004). The space inside the membranes is called the lumen, while on the outside there is a fluid, the stroma, that is rich in dissolved proteins. The whole assembly is often enclosed within another membrane, defining the chloroplasts, entities floating within the cytoplasma of living cells, while while in some bacteria there is no outer barrier: these are called "cell-free".

Figure 2.12. Spectrum of solar radiation, on top of the atmosphere, at the Earth's surface on a clear day, and examples of the scattered part on typical overcast and clear days (based on NASA, 1971; Gates, 1966).

Figure 2.13. Spectrum of absorptance for green plants and for purple bacteria (see page 36; based on Clayton, 1965). Figs. 2.12 and 2.13 are reproduced from B. Sørensen, Renewable Energy, 2004a, by permission from Elsevier.

Although some of the basic processes responsible for photosynthesis have been inferred long ago (see Trebst, 1974), the molecular structure of the main components has only recently been identified. The reason is that the huge molecular systems have to be crystallised in order to perform spectroscopy such as X-ray diffraction that allows identification of atomic positions. Crystallisation is achieved by some gelling agent, typically a lipid. Early attempts either damaged the structure of the system of interest, or the agent used to crystallise showed up in the X-ray pictures in a way difficult to separate from the interesting atoms. As a result, very slow progress has finally resulted in finding suitable agents and then improving the resolution from initially being over 5 nm down to first some 0.8 nm by 1998 and recently to about 0.2 nm. This allows most of the structure to be identified, although not the precise position−for example of hydrogen atoms. In the future, it is expected that the actual splitting of water may be followed as a result of timeslice experiments under pulsed illumination.

Figure 2.14. Schematic picture of the components of the photosynthetic system in green plants and cyanobacteria. See the text for an explanation (Sørensen, 2004c; with use of Protein Data Bank IDs 1IZL (Kamiya and Shen, 2003), 1UM3 (Kurisu et al., 2003), and 1JBO (Jordan et al., 2001)).

Figure 2.14 places the main photosystems along the thylakoid membrane, shown schematically but with the three systems exhibiting extrusions on the lumen or stroma side according to experimental findings. The splitting of water by solar radiation takes place in photosystem II, with the hydrogen emerging in the form of *pq*H2, where pq is plastoquinone. The cytochrome b6f (resembling the cytochrome bc1 at work in the mitochondria of higher animals) transfers the energy in *pq*H2 to plastocyanin (pc) and recycles pq to photosystem II. The plastocyanin migrates on the lumen side of the membrane to photosystem I, where more trapping of sunlight is needed to transfer the energy to the stroma side of the membrane, where it is picked up by ferredoxin (fd). Ferredoxin is the basis for the transformation in the stroma, of NADP to NADPH2, which is able to assimilate CO2 by the Benson-Bassham-Calvin cycle (see Sørensen, 2010a) and thereby form sugarcontaining material such as glucose and starch. However, a few organisms are capable of alternatively producing molecular hydrogen rather than NADPH2. Possibly, the selection of the hydrogen-producing process can be stimulated by genetic engineering. More details of the molecular processes involved in photosynthesis are provided in Sørensen (2010a).

An X-ray study at 0.37 nm resolution has revealed 72 chorophylls in the photosystem II dimer (Ferreira et al., 2004). Chlorophylls are rod-like molecules capable of absorbing light. Some of these are organised in rings with a unique manganese cluster believed to act as a catalyst. These chlorophylls are part of an antenna system capable of absorbing solar radiation at the lower wavelength (higher energy) of the two peaks shown in Fig. 2.14 for green plants. Peripheral chlorophyll molecules may capture light and transfer the energy to the central region of the photosystem II. The process for an absorbed light quantum of wavelength λ = 680 nm will deliver an amount of energy given by

$$E = hc/\lambda = 2.9 \times 10^{-19} \text{ J} = 1.8 \text{ eV},$$ (2.27)

where c is the vacuum speed of light (3 × 10⁸ m s-1) and h is Planck's constant (6.6 × 10-34 Js). The excited chlorophyll state is a collective state involving electrons in all parts of the central ring structure. The excited state may decay and re-emit the energy as a photon (fluorescence), but the probability of transferring the energy to a neighbouring chlorophyll molecule is higher. It is in this way the energy is handed from one chlorophyll molecule to another until it reaches one of the central chlorophyll-a molecules.

Figure 2.14 indicates schematically how the capture energy is transferred through a cytochrome cluster to another photosystem (number "I"), from where the energy may be stored for life processes in the plant by a process that precisely avoids the formation of free hydrogen, which naturally is seen as a problematic molecule by a living organism,

$$ADP^{3-} + H^{+} + HPO{4}^{2-} \rightarrow ATP^{4-} + H{2}O,$$ (2.28)

where ADP and ATP are adenosine di- and triphosphates. ATP is formed on the stroma side of the membrane, implying a transport of protons across the membrane.

Brought to the stroma side by the succession of photosystems, one now finds a number of electrons carried by the proteins ferredoxin and flavodoxin. Furthermore, protons (positively charged hydrogen ions) have been separated and moved across the thylakoid membrane (thereby creating an H⁺ gradient from lumen to stroma). This makes room for two types of recombination processes on the stroma side (Fig. 2.14). One is the "normal" reaction in plants and most bacteria:

$$NADP + 2e^{-} + 2H^{+} \rightarrow NADPH_{2}, \qquad (2.29)$$

where NADPH2 combines with ATP4- from (2.28) and water to assimilate CO2 from the atmosphere for the synthesis of sugars and other molecules needed for the organism in question (see Sørensen, 2010a). The second possibility is the subject of interest here−namely−to produce molecular hydrogen by some process equivalent to

$$2e^{-} + 2H^{+} \rightarrow H_{2}$$ (2.30)

where, as in the inorganic case (2.17) and in Fig. 2.5, an "agent" (in the inorganic case a catalyst) is needed for promoting the reaction. In the organic case, such an agent may be hydrogenase.

Bio-hydrogen production pathways

Except for some rare bacteria, few organisms in nature produce hydrogen through (2.30). Clearly, the path in (2.29) is more profitable for the organism itself. To achieve bio-hydrogen production, it may therefore be necessary to trick suitable organisms into producing hydrogen by removing the subsequent steps of the "natural" pathway−for example by genetic engineering. The problem is that this would also remove the life-support system of the organism and prevent it from growing and reproducing. It is therefore most likely that a compromise will have to be struck, where only a part of the solar energy collected by the organism will be used for hydrogen production, while the rest is left for the purposes of the organism itself. This has implications for the overall efficiency of bio-hydrogen production. While the photosynthetic processes are efficient as judged by the requirements of the organism, they are low when the organism is seen as a solar collector, intercepting on average (over day and night) an energy flow of some 100-200 W m⁻² . The energy efficiency of photosynthetic production of biomass relative to incident solar radiation is about 0.2% as a global average (Sørensen, 2010a), rising to at most 2% in special circumstances such as on coral reefs, and higher only for special energy-subsidised agri- or aquaculture. This implies maximum theoretical efficiencies for bio-hydrogen production of about 1% and the ensuing economic consequences discussed in Chapter 6.

Cyanobacteria (also called blue-green algae) convert nitrogen to ammonia, with the use of a catalyst called nitrogenase. The structure of a nitrogenase is shown in Fig. 2.15. Important components are an iron protein and a molybdenum-iron protein, the latter exhibiting an interesting mid-point atom (Fig. 2.16), assumed to be nitrogen and only recently discovered (Einsle et al., 2002). Nitrogen fixing bacteria live in symbiosis with a range of green plants, such as legume crops, and account for about half of the nitrogen needed by human society agriculture. The other half is provided by chemical fertilisers based upon the Haber-Bosch ammonia synthesis route. The interesting feature of the nitrogenase Mo-Fe centre is its similarity to the tools used in the industrial ammonia synthesis process. Two amino acids keep the Mo-Fe cluster in place: a histidine attached to the Mo end and a cysteine attached to the single Fe atom at the opposite end. The middle atom stabilises the two Fe-triangles, the detailed working of the catalyst still debated (Smith, 2002).

Figure 2.15. Nitrogenase Mo-Fe protein from Azotobacter vinelandii. Hydrogen and amino acid atoms are not shown. Based on Protein Data Bank ID: 1M1N (Einsle et al., 2002). There are four sets of Fe8S7 and MoFe7S9N clusters, one enlarged in Fig. 2.16.

Figure 2.16. The highly specific Mo-Fe cluster of nitrogenase with centre ligand suggested to be N (blow-up of rectangle in Fig. 2.15). Based on the Protein Data Bank ID: 1M1N (Einsle et al., 2002). Resolution: 0.116 nm.

The overall reaction for ammonia synthesis, with ATP as the energy source and an inorganic phosphate ion as a by-product, can be written

$$N_2 + 8 H^+ + 8 e^- + 16ATP^{4-} + 16H_2O \rightarrow 2NH_3 + H_2 + 16 ADP^{3-} + 16 H_2PO_4^-.$$ (2.31)

The presence of oxygen, from the exterior or the interior of the organism, has a deleterious effect on nitrogenase, and cyanobacteria separates nitrogen fixation and oxygen evolution, either spatially or temporally (Berman-Frank et al., 2001).

Because hydrogen evolution is a side effect of nitrogen conversion, the energetic efficiency is modest. Under normal circumstances, cyanobacteria use the hydrogen by-product of nitrogen fixation to fuel energy-requiring processes, by a conversion process catalysed by the enzyme uptake hydrogenase (Tamagnini et al., 2002). The active site of uptake hydrogenase contains a nickel-iron complex, the structure of which was first identified in Desulfovibrio gigas (Volbeda et al., 1995) and later in D. vulgaris Miyazaki F (Ogata et al., 2002). The extraction of hydrogen for human uses thus requires genetically modifying the cyanobacteria to suppress the action of uptake hydrogenase (Happe et al., 1999).

Under anaerobic (oxygen-free) conditions, green algae can either use hydrogen as an electron donor in the CO2 assimilation process or can, in the absence of oxygen, produce molecular hydrogen by combining protons with electrons from ferredoxin. These processes involve an enzyme (catalyst) called reversible hydrogenase, probably similar to the bi-directional hydrogenase seen in certain cyanobacteria (Tamagnini et al., 2002; Pinto et al., 2002). Use of genetic engineering to transfer some of the properties of specific hydrogenases such as the "Fe-only" type from the bacterium Clostridium pasteurianum (Peters et al., 1998; shown in Fig. 2.17), which can only produce hydrogen in dark, oxygen-free environments, to photosynthetic cyanobacteria such as Synechococcus elongates, has been carried out successfully (Asada et al., 2000), with H2 production capability as a result.

The persistent role of Fe-S molecules in electron transfer is quite remarkable. It is involved in photosystem I, where three Fe4S4 clusters sits prominently on the stroma side, in ferredoxin, where each branch features a Fe2S2 cluster, and now in the nitrogenase and hydrogenase enzymes where similar but not identical clusters of typically 2-5 iron-sulphur molecules are performing the energy transfer processes of a variety of organisms.

The rate of photosynthesis varies between organisms, for example reflecting the amounts of solar radiation available at typical growing sites. In highradiation environments, many plants have protective mechanisms ranging from passive coatings on leaves or reduced numbers of chlorophyll staves to active regulation of the amounts of radiation absorbed. These features influence the overall solar energy conversion efficiency. For natural cyanobacterial species, one of the highest rates of hydrogen production measured for the Anabaena variabilis is just under 1%¤ (1.6% relative to the photochemical active part of the solar spectrum; see Figs. 2.12 and 2.13), and only over short periods of time (about a half hour), low irradiation (50 W/m2 ), and in a pure argon atmosphere (Masukawa et al., 2001, 2002). Average efficiencies over 24 h with outdoor solar radiation exposure are only about 0.05% (Tsygankov et al., 2002a). The biomass addition to the bacterial culture itself is typically between 0.5 and 1% of incident radiation accumulated over a number of growth days. When solar radiation is intense, bacteria and algae harvest more photons through their photosystems than the metabolic and hydrogen-producing processes can handle, and the conversion efficiencies go down. The ensuing fluorescence and heating further reduces hydrogen production, and to improve efficiencies, genetic manipulation truncating the chlorophyll antenna size has been used, for example, for the unicellular algae Chlamydomonas reinhardtii (Polle et al., 2002). Obviously, this does not in itself remedy the poor utilisation of available sunlight, but experiments have been made with two-layer collectors, where the first layer is genetically modified and the second is not (Kondo et al., 2002; using bacteria of the genus Rhodobacter sphaeriodes and obtaining about 2% efficiency at 500 W/m2 irradiation for 24 h). The same effect can be achieved by sulphur deprivation, but still obtaining only low levels of hydrogen production (Tsygankov et al., 2002b).

In understanding the efficiencies of hydrogen production by algae and bacteria, it should be kept in mind that there are fundamental limitations to the fraction of solar radiation that can be retrieved in the form of hydrogen. The solar spectrum on clear and overcast days is not the same (see Fig. 2.12), implying that a single collection system with a fixed spectral sensitivity cannot

Only efficiencies relative to the radiation energy received are mentioned here. In biochemical literature, one may find yields of hydrogen production quoted as mol per kg of chlorophyll-a per hour, or per kg of culture, and the light input quoted as einsteins (E) per square metre and second. These data are useful only if the geometry of the reactor (area exposed to light and depth) is known and the light source used emits monochromatic light. The non-SI unit einstein is defined as Avodagro's number times the energy of a single light quantum, *hν,* and is thus not strictly a unit but linearly proportional to the frequency of the light and therefore useless for a distribution of light frequencies such as the one found in solar radiation or in laboratory white light sources such as tungsten lamps. The nearest SI unit is lumen, which assumes the specific frequency distribution of black-body radiation at the temperature 2040 K. This distribution peaks at a wavelength of 555 nm (where 1E = 0.214 MJ) and resembles white light of certain lamps, but is not useful for the solar spectrum, as shown in Fig. 2.12.

work well at all times. In addition to the losses also found in photovoltaic devices (due to the effects of band-gap and internal resistance; see Sørensen, 2010a), the photosystem in plants and bacteria have much more narrow frequency bands of light acceptance, as seen from Fig. 2.13, implying that substantial parts of the solar spectrum cannot be used. Furthermore, the radiation harvested has to be used for the growth and life support of the organism (respiration) before some of it can be diverted to producing hydrogen not used by the organism.

Figure 2.17. Iron-only hydrogenase from Clostridium pasteurianum, showing a string of five Fe-clusters: one Fe2S2 + , three Fe4S4 2+, and a peculiar C5H4O7S2Fe2 molecule. Based on Protein Data Bank ID: 1FEH (Peters et al., 1998). Resolution: 0.18 nm. Hydrogen and amino acid atoms are not shown.

The behaviour of hydrogen production efficiencies are typically uneven, with an initial delay while the organism switches from light absorption and respiration to hydrogen production, and later diminishing due to saturation or lack of feedstock. It is therefore important to distinguish between the peak efficiencies that may be experienced for short intervals and the average efficiencies over long periods of time. For comparison with other energy systems, an averaging time of a year should be used, due to the seasonal effects of solar radiation. As pointed out by Bolton (1996), the efficiency may be factorised to reflect incoming solar radiation and the fraction of solar radiation captured and useful to the hydrogen-producing processes, thus involving the quantum yield after losses of light quanta below the threshold energy and excess energy exciting various quantum states and eventually emerging as heat, as well as the hydrogen transport losses within the organic system, similar to an internal resistance of a semiconductor solar cell device.

For cyanobacteria, the distribution of radiation energy use between biomass and hydrogen production has been found to be about 10:1 (Tsygankov et al., 2002a). The smallness of the hydrogen production efficiency for direct bioproduction is thus an inescapable condition, both for direct hydrogen production by the organism and for alternate production of biomass during the day and hydrogen in the dark. The implications of this are important for cost, as there has to be either a reactor device with cover, if hydrogen is to be retrieved from the same system as the one receiving solar radiation, or two separate systems. For alternate biomass/hydrogen production one may have an open large-area system for growth under sunlight and then transfer the biomass to a less expensive reactor with smaller surface area for the hydrogen-producing process. This points to the third possibility, where the biomass is just a feedstock (possibly in the nature of a residue, of discarded refuse, or from some other inexpensive biomass source), which is collected and fed into some sort of reactor converting it into hydrogen. This route may employ several conversion techniques, such as the gasification schemes (section 2.1.4) or fermentation, a biological process that does not involve light input for its bacterial processes (see fermentation section following).

Hydrogen production by purple bacteria

Purple bacteria do not have the two photosystems depicted in Fig. 2.14, but a single one incapable of water splitting but capable of delivering the energy required for CO2 assimilation, with acetic acid or hydrogen disulfide as the electron donor, in the case of non-sulphur purple bacteria. Purple sulphur bacteria use sulphur or sulphur compounds as a substrate. The working of the single photosystem is similar to that of photosystem I (Minkevich et al., 2004). Purple bacteria also contain nitrogenase and hydrogenases and may in the absence of nitrogen develop molecular hydrogen, without the problems encountered in organisms with both photosystems, related to the damage to nitrogenase caused by oxygen. For direct light-dependent hydrogen production, efficiencies are low (the absorption spectrum is shown in Fig, 2.13). The non-sulphur purple bacteria are considered for use by fermentation of organic waste, with hydrogen as one possible output. Some purple non-sulphur bacteria such as Rubrivivax gelatinosus can produce hydrogen by the water-gas shift reaction (2.2), while at the same time using organic substrates by means of an oxygen-tolerant hydrogenase (Levin et al., 2004).

Fermentation and other processes in the dark

The production of an energy-rich gas from an organic substrate under oxygen-free and dark conditions is called fermentation. It is becoming a conventional energy conversion tool in biogas reactors, where the gas emerging is chiefly methane and carbon dioxide (Sørensen, 2010a). However, direct production of hydrogen is possible by selection of suitable bacteria for the degradation process. In biogas production, the bacterial cultures are often not controlled, and there may be hundreds of different bacteria working in different temperature regimes (from the mesophilic 25-40°C to the thermophilic 40-65°C region or even higher temperatures). For obtaining a high hydrogen content of the gas produced, it is necessary to use specific bacterial cultures, and even so, the CO2 contamination is unavoidable and some amounts of CH4, CO, and H2S are usually present and require the hydrogen to be extracted by a subsequent purification process.

The feedstock may be carbohydrates such as glucose, starch, or cellulose or more complex waste streams, including industrial or habitat liquid and solid wastes of plant or animal origin, including food waste and animal dung. The cost of feed may be lower for the less specific types, but hydrogen yields are higher for transformation of glucose to acetic or butanoic acid (butyrate):

$$C6H{12}O_6 + 2H_2O \rightarrow 2CH_3COOH + 2CO_2 + 4H_2 + 184.2 \text{ kJ/mol},$$ (2.32)

$$C6H{12}O_6 \rightarrow CH_3CH_2CH_2COOH + 2CO_2 + 2H_2 + 257.1 \text{ kJ/mol.}$$ (2.33)

In praxis, yields of 2 to 4 mol H2 per mol of glucose have been reported, corresponding to realising the stoichiometric ratios for the processes just listed (Ueno et al., 1996; Hawkes et al., 2002; Lin and Lay, 2004; Han and Shin, 2004; Hallenbeck, 2009). The theoretical energy efficiency of hydrogen production from cellulose would then be 17-34%. Hydrogen build-up inhibits production, so continuous removal of produced hydrogen is essential (Lay, 2000). Among the bacterial cultures used in current laboratory experiments are several strains of Clostridium (such as pasteurianum, butyricum, and beijerinkii). The naturally occurring bacteria cultures found in cow dung and generally in sewage sludge also seem dominated by Clostridia. These bacteria are suitable for pure sugar fermentation and in proper mixtures also for the more complex types of feedstock mentioned. Like in biogas plants, the ability to handle varying compositions of feedstock as a function of time requires the presence of several types of bacteria, each of which should be able to grow to an important culture when the right feed becomes present.

A number of commonly available waste streams are suitable for fermentation. Specific factors influencing the possibility of obtaining hydrogen as a major end product have been discussed by Wang and Wan (2009). They include the selection of an appropriate reactor type and avoidance of inhibiting substances such as metal compounds in the waste.

The fermentation processes depend on temperature, usually producing more hydrogen in the thermophilic region (about 55°C) than in the mesoplilic region (for starch; Zhang et al., 2003). They also depend on pH, with optimum production in acidic environments (pH = 5.5 for cow dung; Fan et al., 2004). The production in some cases varies considerably with time, while in other cases a rather stable production is obtained with continuous feedstock input. Stable production seems to require modest rates of new input (e.g., diluted sludge, used for olive mill waste using Rhodobacter sphaeroides; Eroglu et al., 2004). High performance is found for fermentation of granular sludge high in sucrose, producing a gas of 63% hydrogen and a volume yield of 280 litres of hydrogen per kg of sucrose, over 90 days (which is about 100% of the ratio in (2.33)), at a rate of 0.54 litres of H2 per hour and per litre of sludge (using culture with 69% Clostridium and 14% Sporolactobacillus racemicus; Fang et al., 2002). Also, the unicellular cyanobacterium Gloeocapsa alsicola has given high hydrogen yields in laboratory glucose fermentation experiments (Troshina et al., 2002).

Figure 2.18 shows a generally accepted model for the main process steps in the enzymatic conversion of glucose, with energy and hydrogen transfers mediated by the ATP-ADT process as in plants and the NADH2-NAD process (NAD = nicotinamide-adenine dinucleotide) replacing the NADPH2- NADP process (2.29) of green plants and algae. The thick arrows indicate carbon flows, while fainter arrows are used to indicate electron and proton flows. Hydrogenase, as well as several other enzymes, is seen to play a role. Variants of this diagram have been demonstrated for other fermentation processes.

Industrial-scale production of bio-hydrogen

The two hydrogen production routes involving either photosynthetic algae/ cyanobacteria or fermentation bacteria converting earlier produced organic residues pose different challenges to a profitable industrial scale-up of the current laboratory-scale experimental work. Only the fermentation route is currently in industrial use, albeit not with hydrogen as the end product.

Several of the organisms invoked in experiments with hydrogen production are modifications of naturally occurring and often very abundant species. Some 40 years ago, the fairly large, filamentous photosynthetic cyanobacterium Trichodesmium was demonstrated to fix nitrogen by means of nitrogenase, and it was believed to be the main plankton species responsible for ammonia production in the open oceans. Subsequently, a number of unicellular cyanobacteria of a size under 10-5 m have been shown to be responsible for 20-60% of chlorophyll biomass and carbon fixation in the oceans (Zehr et al., 2001; Palenik et al., 2003; Sullivan et al., 2003). Dominant species are Synechococcus and Prochlorococcus, and abundance may exceed 10⁹ per litre of seawater. In addition, large quantities of bacterial rhodopsin have been found, for example, in Escherichia coli (Béja et al., 2000). This rhodopsin is capable of converting solar radiation and forming an electric proton gradient across the cell membrane and was earlier believed to exist only in archaea.

Figure 2.18. Model of fermentation processes in Clostridium pasteurianum. Enzymatic components are labelled: 1, glyceraldehydes-3-phosphate dehydrogenase; 2, lactate dehydrogenase; 3, pyruvate-ferredoxin oxidoreductase; 4, hydrogenase; 5, NADHferredoxin oxidoreductase; 6, acetaldehyde dehydrogenase; 7, ethanol dehydrogenase; 8, β-hydroxybutyryl-CoA dehydrogenase; 9, butyryl-CoA dehydrogenase; 10, butyraldehyde dehydrogenase; 11, butanol dehydrogenase. P, phosphate. From Lin and Lay, 2004 (used by permission from Int. Assoc. Hydrogen Energy), which is a modified version based on Dabrock et al. (1992), American Society for Microbiology.

Finally, the abundance of viruses in ocean water has been estimated to be as

high as 1010 per litre, being important pathogens both for fish and for marine phytoplankton, assuming responsibility for the toxic algae blooms that have created severe problems both for holiday swimmers and for commercial aquaculture and fisheries (Culley et al., 2003; Azam and Worden, 2004). Similarly, freshwater cyanobacteria are known to host microcystins responsible for outbursts of extreme toxicity blooms in lakes and other freshwater reservoirs all over the world (see, e.g., Shen et al., 2003). The cyanobacteria most common in freshwater systems include Microsystis, Anabaena, Nostoc, and Oscillatoria. These facts must be taken into account in designing new uses of the hydrogen-producing variants of the organisms in question.

Figure 2.19. Schematic layout of photobiological hydrogen production plant with auxiliary plants for production of modified bacterial strains and for hydrogen gas purification (Sørensen, 2004c).

As mentioned, natural hydrogen-producing bacteria are unlikely to produce industrially interesting amounts of hydrogen, and direct photoproduction of hydrogen is thus dependent on genetically modified strains such as the ones described in the previous sections. Because a solar collector is required with a large area due to the modest efficiency of conversion, a possibility would be to place this collector on the ocean surface, presumably offering lower area prices compared to on-shore land prices, even in marginal areas. The need for hydrogen recuperation to a pipeline demands a closed reactor system containing the cyanobacteria, a glass, or otherwise transparent cover and inlet/outlet devices for hydrogen transport and for replenishing or replacing the cultures as needed. The need to avoid accumulation of additional cyanobacteria in the surrounding water, as well as the require-ment of not allowing genetically altered species to mix with the wild type in oceans, calls for a closed system for bacterial supply, photoreactor, and hydrogen handling. The layout could in schematic terms be like the one shown in Fig. 2.19. The off-shore reactor could, instead of the glass-covered flat-plate collector indicated (and used in the experimental ocean facility developed by RITE in Kyoto; Miyake et al., 1999), have a reactor system consisting of tubular pipes, allowing a higher portion of solar radiation to hit the cyanobacterial substrate in case of a non-perpendicular impact angle and easier hydrogen transport to the pipe ends (Akkerman et al., 2002). This system is in use for thermal solar collectors, where it gives a modest increase in energy production (some 10%) at a cost, which, however, is more than 10% higher than that of flat-plate systems. Such ideas have recently been reviewed (Eroglu and Melis, 2011).

Figure 2.20. Attenuation of solar radiation as a function of penetration into a biohydrogen reactor containing partly modified cultures of Rhodobacter spheroides. Reprinted from J. Miyake, M. Miyake, Y. Asada (1999). Biotechnological hydrogen production: research for efficient light energy conversion. Journal of Biotechnology 70, 89-101, with permission from Elsevier.

The algae or cyanobacterial cultures must occupy a shallow box because the solar radiation diminishes rapidly with depth of penetration (through water plus culture), as shown in Fig. 2.20. The penetration of radiation is highly wavelength-dependent, with low wavelengths of light being able to penetrate deeper (see Sørensen, 2010a; section 2.2.2). For an actual tank containing a photosynthetic purple bacterium Rhodobacter spheroides RV culture, Miyake et al. (1999) found an even more rapidly declining intensity, having dropped to 10% at 1 cm depth and to 1% at 2 cm depth. At the specific wavelengths of absorption by the two types of chlorophyll antennas present in the single photosystem, practically all light is absorbed within the first 0.5 cm. In accordance with the remarks on saturation in the previous sections, the efficiency of hydrogen production is found to be low in the first 0.5 cm layer and high in the deeper layers, but with negligible absolute production due to the near absence of light.

Most hydrogen produced from biomass, whether by thermo- or biochemical routes, will require a process of purification before being suitable for its intended use. The most efficient ways to accomplish high purification is by employment of membrane technology. For most bio-hydrogen, this will entail an additional process step, not directly integrated into the production step as it might be for reverse fuel cell and some reformation-type pathways to hydrogen (Lu et al., 2007).

Figure 2.21. Schematic layout of substrate fermentation plant for hydrogen production, with infrastructure for residue collection/recycling and hydrogen gas purification. The blanket hydrogen reactor is similar to those used in biogas plants (Sørensen, 2010a).

Figure 2.21 shows a possible layout for industrial bio-hydrogen production based upon bacterial fermentation of already available biomass, such as harvested residues or wastes from other biomass uses (food, wood, etc.). The blanket-type reactor has been used in several experimental hydrogen fermentation studies (Chang and Lin, 2004; Han and Shin, 2004). The feedstock in this case has to be collected and transported to the site of the hydrogen production plant, which will generally have the appearance and components of biogas plants, either low-technology traditional ones or the large, industrial biogas plants currently used in most of the world in connection with waste water treatment plants and in a few cases in connection with biogas production from mixed household and (non-toxic) industry wastes (see Sørensen, 2010a). After hydrogen extraction, the remaining residues may be returned to agriculture and, due to their high nutrient value, may replace or reduce the use of fertiliser. Most fermentation plants use the thermophilic temperature regime, but going to slightly higher temperatures (70-80°C) may be possible without too severe an economic penalty from having to heat sections of the plant, because extensive recovery and recycling of heat is possible (Groenestijn et al., 2002).

Recently, certain microbial systems have demonstrated the ability to generate electricity directly (with little or no hydrogen production) in fuel cell-like devices. These are described in section 3.7. Production of hydrogen by photoelectrochemical devices may employ microbial sensitisers and enzymes or engineered substances derived from bacterial constituents. These are discussed in section 2.1.6 after the inorganic devices.

2.1.6 Photodissociation

In analogy to photovoltaic and photoelectrochemical devices aimed at electricity production (see Sørensen, 2010a), efforts have been made to modify the devices in question to deliver hydrogen directly rather than electricity. Of course, there is the alternative to convert the electricity to hydrogen in a second step−for example, by conventional electrolysis. This is conveniently used as a backstop technology for economic comparison with new proposals.

The basic problem that must be overcome is that hydrogen production does not start until a sufficiently large cell voltage is reached, and further along the road, there is the familiar problem of separating oxygen- and hydrogenevolving streams, for reasons including safety.

Figure 2.22a,b. Schematic picture of a solar cell with electron excitation and external circuit (a: left) and current I vs. voltage V diagram under illumination (b: above). Isc is shortcircuit current, and Voc is open-circuit voltage. The solar cell may consist of one or more p-n semiconductor junctions or of p-i-n amorphous junctions.

Figure 2.23a shows a hydrogen-producing cell, and Fig. 2.22a shows the corresponding electricity-producing cell. Figures 2.22b and 2.23b show currentvoltage (IV) diagrams for the two systems, to be explained following.

Figure 2.23a,b. Schematic picture of a solar hydrogenproducing cell aiming to excite a sufficient number of electrons for creating the voltage difference needed for water splitting (a: above) and current-voltage curves for such devices with insufficient (A) or sufficient (B) voltage for unassisted water splitting (b: left).

The devices shown in Figs. 2.22a and 2.23a employ a solar cell with at least one semiconductor p-n junction. A p-layer contains lower atoms (i.e., atoms with a lower number of protons, Z) than the bulk of the material, manufactured by doping with foreign atoms, and an n-layer contains higher atoms (see, e.g., Sørensen, 2010a, section 4.4.1). Upon incident solar radiation, an electron is excited from the valence band of the semiconductor into its conduction band, and the potential behaviour across the junction will cause the excited electron to move towards the n-side terminal and the hole left behind in the valence band to move towards the p-side terminal. A current will flow through the external circuit connecting the terminals (in Fig. 2.22a), depending on the external load (resistance in the external circuit) as shown in Fig. 2.22b. Voc is the open circuit voltage corresponding to infinite external resistance (circuit cut-off), and Isc is the maximum photo-induced current as obtained with short-circuited external circuit (zero resistance). The maximum power that can be delivered to the external load is the maximum product IV. It deviates from the product of Isc and Voc by the fill factor FF = IV/ Isc Voc.

In Fig. 2.23a, one of the terminal electrodes is separated from the solar cell with the other electrode, to make room for an electrolyte containing water and some ion-conducting medium (a salt or redox couple). This design is known from photoelectrochemical solar power cells, where the photoelectrode is usually made of nanocrystalline TiO2 covered by a metal- or pureorganic colour dye acting as a photo-sensitiser. In the case of the modified device aimed at producing hydrogen (and oxygen), the outer electron circuit is short-circuited in order to obtain a maximum current through the electrolyte. The voltage across the electrolyte is not a simple quantity, but depends on the chemical reactions taking place within the electrolyte, notably the water splitting and hydrogen (oxygen)-evolving processes. The IV-curve is in this case a hysteresis curve−that is, it has a different shape depending on whether the voltage is increasing or decreasing. The simple curve shown in Fig. 2.23b is just a limited part of the increasing voltage curve for large V, out of a total voltammogram, a technique further discussed following. It shows a photocurrent for device A that indicates activity in producing hydrogen, while for device B, the current only appears after an external bias voltage, that is, an additional electromotive force, has been added to the external circuit. The indication is that a considerable "internal resistance" has to be overcome before hydrogen production can take place. This is equivalent to stating that the voltage difference between the electrodes must exceed the amount needed for the hydrogen splitting reaction, given in (2.21), based on the reactions (2.16) and (2.17) and also indicated in Fig. 2.23a, plus any internal electrode losses and resistance terms, as formally given by (2.18).

As an example, Khaselev and Turner (1998) used a high-efficiency tandem solar cell consisting of first a p-n junction (for electricity production as in Fig. 2.22) or a Schottky-type junction (for hydrogen production as in Fig. 2.23) made of GaInP2 with a bandgap of 1.83 eV (absorbing the central part of the visible solar spectrum) and then a tunnel diode interconnect and a second semiconductor layer consisting of an n-p GaAs junction with a bandgap of 1.42 eV (suitable for the near-infrared portion of the solar spectrum, passing unhindered through the first layer). Solar irradiation may excite electrons in either layer, and if one electron is excited in GaInP2 while another is already in the conduction band of the GaAs layer, then one of these electrons (most likely the one from GaAs, as it has the lowest excitation energy) may combine with the hole left in the valence band of the other layer, so that the net effect is to raise one electron to a state with energy more than 3.2 eV over the initial state. Two such excited electrons provide energy enough for the water-splitting process (2.18). Like in the photosynthetic water-splitting systems described in section 2.1.5 (see also Chapter 3, Fig. 3.3), it takes four light quanta of typical solar radiation (only the most energetic tail of the distribution could do with less) to supply the 6.16 eV needed (equal to 590 kJ/mol or 2 times 4.93 × 10-19 J for the two water molecules) to split water into hydrogen and oxygen.

The excited electrons will emerge on the electrolyte side of the semiconductor junctions (see Fig. 2.23a), due to the potential gradient created, with holes in the valence band being filled by electrons from the counterelectrode, moving through the external circuit. The full potential gradient now makes its presence felt across the electrolyte, completing the positive and negative electrode reactions (2.16) and (2.17), provided that the electrolyte can provide transport for the protons H+ , either directly (similar to the flow processes shown in Chapter 3, Fig. 3.7) or through charge transfers involving redox couples present in the electrolyte. Hydrogen will then evolve near this negative electrode, while oxygen emerges at the positive counterelectrode, which in the experimental device is made of metallic platinum. The efficiency of the Khaselev-Turner device, illuminated by a tungsten lamp with about 10 times the energy flow typical of solar light, is quoted as 12.4%, with an efficiency they define as

$$\etaC = 1.23 \ I/P{in},$$ (2.34)

in terms of the photocurrent I (A/m2 ) and the incident power of radiation Pin (W/m2 ). This corresponds to considering the electrolysis to be 100% efficient, as 1.23 V is the lower heating value of hydrogen at 25°C. A better method would obviously be to measure the volume of hydrogen produced per unit of time and calculate the output power from that. This device provides enough voltage to produce hydrogen without any energy subsidy, that is, according to the scheme (A) in Fig. 2.23b.

In the general case, where an external bias voltage Vbias may provide additional energy for the production of hydrogen, the solar-to-hydrogen efficiency may be defined by (Bolton, 1996),

$$\eta = (R \Delta G - I V{bias})/P{in}. \tag{2.35}$$

Here, R is the rate of hydrogen production, ∆G is the corresponding free energy, I is the photocurrent, and Pin is the incident solar radiation. Both numerator and denominator have to be inserted in the same units−for example Wm-2. In case it is necessary to apply a bias voltage, the hydrogen production can be said to be made by a solar-assisted hybrid generator.

Various systems have been investigated, such as use of a Cu-doped TiO2

electrode similar to the ones used in organic dye solar cells (Yoong et al., 2009). The system discussed previously uses a high-efficiency tandem solar cell to obtain energy, but there are many further possibilities, such as, for example, those discussed by Jing et al., (2010). The system shown in Fig. 2.24 uses an amorphous solar cell with three p-i-n layers as the negative electrode (Yamada et al., 2003). The p-i-n structure of an amorphous solar cell functions similarly to the p-n junction of a crystalline solar cell (Sørensen, 2010a), with an intrinsic *i-*layer having interstituent hydrogen atoms (a-Si:H) in order to enhance light absorption. Instead of expensive platinum or Ru-based electrodes, simpler materials are used: a Co-Mo compound at the hydrogenevolving electrode and a Ni-Fe-O compound at the oxygen-evolving electrode. The electrolyte is a strongly alkaline (pH = 13) Na2SO4 plus KOH solution. A solar-to-hydrogen conversion efficiency of 2.5% is claimed.

Figure 2.24. Single-chip photoelectrochemical electrolyser, with electrode assembly simply immersed into the electrolyte, according to the laboratory-scale design of Yamada et al. (2003).

An obvious device modification is to separate the two electrodes, so that the gases evolving may be collected easier through separate channels. Typically, a separator in the form of a perfluorised membrane would be used in analogy to proton exchange membrane fuel cells (see section 3.7).

The electrode reactions may be weakened by back-reactions such as conduction band electrons from the positive electrode recombining with protons formed in the electrolyte or electrons from the electrolyte filling vacancies in the electrode valence band. These reactions depend on the pH of the electrolyte, and it may therefore seem desirable to be able to control the alkalinity of the electrolyte near the positive electrode independently from that at the negative counter-electrode. Milczarek et al. (2003) use a two-compartment photoelectrochemical electrolyser separated by a perfluorised membrane to perform this separate optimisation. The device, which is illustrated in Fig. 2.25, uses 2.5 molar Na2S at the positive electrode and 1 molar H2SO4 at the negative electrode. The negative electrode is made of a Pt slab covered with a film of a perfluorinated Nafion (trademark of Dupont de Nemours) and a further layer of electrodeposited Pt, while the positive electrode consists of Ti covered by a CdS film. The two electrolytes are separated by the membrane (tradename Aldrich Nafion-417), allowing proton transport. The efficiency, based on the photocurrent as in (2.34), was claimed to be 7% on a sunny day, rising to 12% on an overcast day with a radiation level of 200 Wm-2. However, the use of two electrolytes implies a difference in chemical potential, which acts like an external bias voltage, so the real efficiency is lower and will degrade with time.

Figure 2.25. Two-compartment photoelectrochemical electrolyser (Milczarek et al., 2003), suggesting optimum performance for an H⁺ transporting membrane area 2-3 times larger than that of the light-exposed electrode. Gas outtakes are only indicated.

Many photoelectrochemical hydrogen-producing devices described in the scientific literature since the first one by Fujishima and Honda (1972) are unable to deliver enough potential energy difference for water splitting and therefore apply an additional external bias voltage in order to enable the system to draw the missing energy (e.g., Kocha et al., 1991; Mishra et al., 2003; Radecka, 2004). There are also suggestions to use concentrators to increase the amount of solar radiation reaching the hydrogen-producing device (Aroutiounian et al., 2005), but again this is unlikely to be compatible with the low cost needed because of the modest efficiency achievable.

Whereas the device with incorporation of what is close to a full photovoltaic cell necessarily leads to a high hydrogen production cost, devices such as that illustrated in Fig. 2.24 are likely to be limited by low efficiency.

The behaviour of the electrolyte can be understood in terms of the description of redox processes, which are processes by which the oxidation level of a substance can increase and decrease (reduction, hence "red-ox"), whereby ionic charges can be transported through the electrolyte. The relation between the potential difference across the electrolyte and the Gibb's free energy of the redox reactions

$$X \Leftrightarrow X^+ + e^-, e^- + Y \Leftrightarrow Y^-;$$ or $X + Y \Leftrightarrow X^+ + Y^-$ (2.36)

is given by Nernst's equation (see, e.g., Bockris et al., 2000; Hamann et al., 1998), relating the reversible potential (2.21) to the concentrations ci of the substances entering into (2.36) at a given temperature T (in degrees Kelvin)

$$-z \mathcal{F} V{r} = \Delta G = -z \mathcal{F} V^{0} + \mathcal{R} T \log (c{X+} c{Y-} / c{X} c_{Y}), \qquad (2.37)$$

where z is the number of electrons (one in (2.36)), F is Faraday's constant given in connection with (2.21), R is the gas constant (8.315 J K-1 mol-1), and the constant V⁰ is called the standard potential. It is taken at some reference temperature, normally 298 K, and may be looked up in tables such as those found in CRC (1973).

A way to gain insight into the response of an electrolyte to application of a voltage across a pair of electrodes enclosing the electrolyte is to perform a two-way voltage sweep measurement, the result of which can be displayed in a voltammogram (Bard and Faulkner, 1998). The voltage is first increased from zero to a maximum value and then reduced back to zero. The outcome may look as shown in Fig. 2.26. It shows different regions of behaviour determined for small applied voltage by electrode processes, then by the transformation of the reactants such as those described by the hydrogen-forming reactions indicated in Fig. 2.23a, and finally, a region where the process is limited by diffusion of reactants into the electrode region. This assumed a fairly quick voltage sweep. If the voltage is increased so slowly that there is always time for diffusion to the active region, this behaviour disappears. On the way back, one observes a lower current, due to the possibility that the oxidised substance may experience other reactions (Wolfbauer, 1999).

A micro photoelectrochemical system for portable equipment using polyaniline as a catalyst has been studied by Han and Furukawa (2006). Some of the biological systems for hydrogen production considered in section 2.1.5 mimic photoelectrochemical systems. The possibility of making hydrogenase produce hydrogen molecules rather than hydrogen bound in organic substances has spurred efforts to insert components such as the hydrogenase into inorganic systems, in order to improve control over the hydrogen-evolving processes. Also, the photosynthesis could be "industrialised" by inserting light-collecting biological antennas such as chlorophylls or porphyrins (Sørensen, 2010a) into electrochemical devices in the place of the (expensive) solar cells used in the devices described above.

Figure 2.26. Schematic sketch of voltammogram for the electrolyte of a hydrogenproducing photoelectrochemical cell.

Hydrogenase purified from the bacterium Pyrococcus furiosus has been combined with the enzymes of the pentose phosphate cycle to produce hydrogen from glucose-6-phosphate and NADP⁺ (Woodward et al., 2000). Similar experiments have been made using hydrogenase from Thiocapsa roseopersicina (Wenk et al., 2002; Qian et al., 2003) immobilised in viologen polymer matrices. These experiments used an applied bias voltage. More close analogies to the photoelectrochemical cells have been studied by Saiki and Amao (2003) and Tomonou and Amao (2004). Here a light absorbing system based on tetraphenyl-porphyrin tetrasulphate or Mg-chlorophyll-a (from spirulina) was coupled partly to an NADH or NADPH system able to extract energy from glucose (or with a further enzyme from polysaccharides) and partly a methylviologen system with a platinum catalyst capable of producing hydrogen from the protons transferred to it. Absence of oxygen is essential to all of these systems.

2.1.7 Direct thermal or catalytic splitting of water

Another route contemplated for hydrogen production from water is thermal decomposition of water. As the direct thermal decomposition of the water molecule requires temperatures exceeding 3000 K, which is not possible with presently available materials, attempts have been made to achieve decomposition below 800°C by an indirect route using cyclic chemical processes and catalysts. These thermochemical or water-splitting cycles were originally designed to reduce the required temperature to the low values attained in nuclear reactors, but could of course be used with other technologies generating heat, say at around 400°C. An example of the processes considered in early studies is the three-stage reaction (Marchetti, 1973).

$$\begin{split} &6\text{FeCl}_2 + 8\text{H}_2\text{O} \rightarrow 2\text{Fe}_3\text{O}_4 + 12\text{HCl} + 2\text{H}_2 \ (850^\circ\text{C}), \ &2\text{Fe}_3\text{O}_4 + 3\text{Cl}_2 + 12\text{HCl} \rightarrow 6\text{FeCl}_3 + 6\text{H}_2\text{O} + \text{O}_2 \ (200^\circ\text{C}), \ &6\text{FeCl}_3 \rightarrow 6\text{FeCl}_2 + 3\text{Cl}_2 \ (420^\circ\text{C}). \end{split} \tag{2.38}$$

The first of these reactions still requires a high temperature, implying a need for external energy to be supplied in addition to problems related to the corrosive substances involved. A similar scheme using CaBr2 involves three reactions at 730, 550, and 220°C (Doctor et al., 2002). This research is still a long way from having created a practical technology.

At temperatures of 700-900°C, a number of mixed metal oxides of perovskite structure (ABO3) have been studied as proton conductors, for example, for use in solid oxide fuel cells. Water vapour introduced at one electrode of such a proton conductor sustaining a sufficiently strong electric current (with the use of externally provided electric energy) is able to split water and produce some amounts of hydrogen (Matsumoto et al., 2002; Schober, 2001). Introducing a membrane to separate hydrogen as it is produced, thermal water splitting can be augmented, but the production rates are still very minute at the temperatures in question (Balachandran et al., 2004).

Other reaction schemes are considered in section 5.4.2. A general review of hydrogen production methods has recently been provided by Holladay et al. (2009).

2.2 Issues related to scale of production

2.2.1 Centralised hydrogen production

Different technologies are suited for application on a scale determined both by the type of application and by the characteristics of the technology itself. If the cost of a given technology exhibits an economy of scale, the technology may be used in a centralised fashion, with large individual units. If the technology is cheapest in smaller units (for fixed overall production), the situation is of course opposite, but this is rare, and a common situation is that the cost is insensitive to the scale of production. This allows the kind of applications aimed for to determine the scale selected. However, there may also be specific scale requirements set by the type of usage. For example, technology for passenger cars must have a size and weight suitable for typical motor vehicles. Here one finds little flexibility, due to existing infrastructure such as roads, size of garages and parking spaces, and so on. Generally, if a new technology requires changes in infrastructure, both the cost and the inconvenience associated with changing to it must be taken into consideration.

Traditional hydrogen production technologies, such as steam reforming or electrolysis, do exhibit some economy of scale−for example for heat exchangers or associated with environmental control options. For production of hydrogen by photosynthesis, the economy of scale resembles that of agriculture, while for dark fermentation, known economy of scale advantages for biogas plants should be transferable. Photodissociation by devices resembling solar cells or ambient temperature fuel cells are likely not to show any marked economy of scale, while high-temperature direct splitting of water may derive advantages from a large scale of operation.

2.2.2 Distributed hydrogen production

The identification of hydrogen production by reverse operation of lowtemperature fuel cells as a technology not suffering by small scale of operation explains the interest in developing this technology for dispersed (decentralised) employment. Visions of hydrogen production in individual buildings (see section 4.5) may be based on reversible proton exchange membrane fuel cells taking over the function of existing natural gas burners, while at the same time offering to produce hydrogen for vehicles parked in the building and to regenerate electricity from stored hydrogen in situations of insufficient electricity supply from the outside (e.g., in connection with the desire of electric utility suppliers to perform load-levelling or manage intermittent renewable energy resources). If solar collector-based hydrogen generators become developed to a state of acceptable efficiency and cost, they can also be used in a decentralised fashion. Present experience from electricity-producing photovoltaic installations is that the economy of scale obtained for centralised plants (e.g., reducing installation cost) is often offset by the possibility of saving conventional roof and facade elements when such solar panels are installed.

The question of suitability for decentralisation becomes more critical in the case of vehicle-integrated hydrogen production systems, which must be economic at a small scale. Along with problems of reliability, this has caused the interest in on-board fuel reformation (see section 2.2.3) to fade.

2.2.3 Vehicle on-board fuel reforming

The basis for on-board production of hydrogen in principle could be fuels

such as biofuels or fossil fuels, notably gasoline, as well as methanol, ethanol, and similar intermediate stages between fuels produced from natural organic or from more artificial industrial primary materials. Of these, only methanol can be obtained by a reformation process similar to that of natural gas, (2.1), at moderate temperatures of 200-300°C. Reformation of other hydrocarbons usually requires temperatures above 800°C. Methanol is also interesting because of its similarity to gasoline in terms of fuelling infrastructure. The methanol energy content of 21 MJ kg⁻¹ or 17 GJ m⁻³ is lower than that of gasoline, but because fuel cell cars are more efficient than gasoline cars, the fuel tank size will be similar.

Figure 2.27. Layout of methanol-to-hydrogen vehicle power system with fuel cell and electric motor. The power controller allows a shift from direct drive to battery charging (from B. Sørensen, Renewable Energy, 2004a, used with permission from Elsevier).

Compressed and liquefied hydrogen are limited for use in automobiles by low energy-density and safety precautions for containers and by the requirement of creating a new infrastructure for fuelling, so there is a strong interest in converting conventional fuels to hydrogen on-board. It is primarily to avoid having to make large changes to the current gasoline and diesel fuel filling stations that schemes based on methanol as the fuel distributed to the vehicle fuel tank have been explored. The energy density of methanol just given corresponds to 4.4 kWh litre⁻¹ , which is roughly half that of gasoline. Hydrogen is then formed on-board by a methanol reformer before being fed to the fuel cell to produce the electric power for an electric motor. The set-up is illustrated in Fig. 2.27. Prototype vehicles with this set-up have been tested in recent years (see Takahashi, 1998; Brown, 1998).

Methanol, CH3OH, may eventually be used directly in a fuel cell without the extra step of reforming to H2. Such fuel cells are similar to proton exchange membrane fuel cells and are called direct methanol fuel cells (see section 3.7). Because methanol can be produced from biomass, hydrogen may in this way be eliminated from the energy system. On the other hand, handling of surplus production of power systems with intermittent production may still conveniently involve hydrogen as an intermediate energy carrier, as it may have direct uses and thus may improve the system efficiency by avoiding the losses in methanol production. The electric power to hydrogen conversion efficiency is currently around 65% in high-pressure alkaline electrolysis plants (Wagner et al., 1998; with prospects for higher efficiencies in the future, see section 2.1.3), and the efficiency of the further hydrogen (plus CO or CO2 and a catalyst) to methanol conversion is up to around 70%. Also, the efficiency of producing methanol from biomass is about 45%, whereas higher efficiencies are obtained if the feedstock is methane (natural gas) (Jensen and Sørensen, 1984; Nielsen and Sørensen, 1998).

Production of methanol

Because methanol may serve as a substitute for hydrogen in fuel cells and an intermediate fuel that can be used to produce hydrogen, its own production is of relevance and will be discussed in brief. Methanol may be produced from fossil sources such as natural gas or from biological material.

Conventional steam reforming of natural gas may produce methanol through a reaction of the form

$$2CH_4 + O_2 \rightarrow 2CH_3OH,$$ (2.39)

as one among a number of possible reactions, of which several were mentioned in section 2.1.1 (Wise, 1981). The initial production of a synthesis gas by (2.1) is a likely pathway for the methanol synthesis (2.39). The process typically yields 0.78 mol of methanol per mol of methane at a thermal efficiency of 64% including considerations of heat inputs (Borgwardt, 1998). A direct synthesis may be obtained by partial oxidation (see section 2.1.2), at temperatures in the range of 425-465°C (Zhang et al., 2002) and with similar efficiencies as for steam reforming.

There are various ways of producing methanol from biomass sources, as indicated in Fig. 2.11. Starting from wood or isolated lignin, the most direct routes are by liquefaction or by gasification. The pyrolysis alternative gives only a fraction of the energy in the form of a producer gas (Güllü and Demirbas, 2001). By high-pressure hydrogenation, biomass may be transformed into a mixture of liquid hydrocarbons suitable for further refining or synthesis of methanol (Chartier and Meriaux, 1980).

Current biomass-to-methanol production schemes use a synthesis gas (a mixture of H2 and CO) derived from wood gasification by a process similar to that of coal gasification. The low-quality "producer gas" resulting directly from the wood gasification (extensively used in cars throughout Europe during World War II) is a mixture of carbon monoxide, hydrogen gas, carbon dioxide, and nitrogen gas, varying with reaction temperature. If air is used for gasification, the energy conversion efficiency is about 50%. If pure oxygen is used instead, some 60% efficiency is possible, and the gas produced would have less nitrogen content (Robinson, 1980).

Gasification could conceivably be performed with heat from (concentrating) solar collectors−for example, in a fluidised bed gasifier maintained at 500°C. A probably more economic alternative is to use some of the biomass to produce the required heat, although this would entail environmental emissions that would have to be controlled.

Once the producer gas has been cleaned, removing CO2 and N2 (e.g., by cryogenic separation) as well as impurities (the nitrogen), methanol is generated at elevated pressure by the reaction

$$2H_2 + CO \rightarrow CH_3OH. \tag{2.40}$$

Changing the H2/CO stoichiometric ratio to suit (2.40) may be achieved by the "shift reaction" (2.2) discussed in section 2.1.1. Steam is to be added or removed in the presence of a catalyst (e.g., iron oxide or chromium oxide).

For woody biomass consisting of 46% C, 7% H, and 46% O plus minor constituents, the process may yield 16 mol of methanol per kg of biomass, with a thermal efficiency of 51% when accounting for the necessary input of electric power for wood drying, and so forth (Borgwardt, 1998).

Non-woody biomass may be converted into methane contained in biogas (see Sørensen, 2004a), followed by the methane-to-methanol conversion process just mentioned, e.g., using steam reformation.

The conversion efficiency of the synthesis gas-to-methanol step is about 85%, and it is usually assumed that improved catalytic gasification techniques are required to raise the overall conversion efficiency above the 51% just mentioned (Faaij and Hamelinck, 2002). The total efficiency will become lower if all life-cycle contributions involving energy inputs−for example, for collecting and transporting biomass−are included (EC, 1994).

The octane number of methanol is similar to that of ethanol, but the heat of combustion is less, as just mentioned. Methanol can be mixed with gasoline in standard engines or used in specially designed Otto or diesel engines. An example would be a spark ignition engine run on vaporised methanol, with the vaporisation energy being recovered from the coolant flow (Perrin, 1981). Methanol uses generally are similar to those of ethanol, but several differences exist in the assessment of environmental impacts, from production to use (e.g., toxicity of fumes at filling stations).

Historical costs of methanol have varied between US\$6 and U\$16 per GJ (Lange, 1997), and estimated future costs with production stimulated by demand from the transportation and power utility sectors could be in the range of US\$5.5 to US\$8 per GJ (Lange, 1997; Faaij and Hamelinck, 2002). In regard to environmental considerations, the gasification would be made in closed environments, where all emissions are collected, as well as ash and slurry. Cleaning processes in the methanol formation steps will recover most catalysts in reusable form, but other impurities would have to be disposed of along with the gasification products. Precise schemes for waste disposal have not been formulated, but probably only a subset of nutrients could be recycled to agri- or silviculture (like in the case of ethanol fermentation; SMAB, 1978).

Production of ammonia by a process similar to the one yielding methanol is an alternative use of the synthesis gas. Production of methanol from eucalyptus rather than from woody biomass has been studied in Brazil (Damen et al., 2002). Extraction of methanol from tobacco cells (Nicotiana tabacum) has been achieved by genetic engineering, incorporating a fungus (Aspergillus niger) to decompose cell walls and infect the plant with a mosaic virus promotor (Hasunuma et al., 2003). More fundamental studies aiming to better understand the way in which conventional methanol production relies on degradation of lignin are ongoing (Minami et al., 2002).

Methanol-to-hydrogen conversion

The steam reforming of methanol to hydrogen involves the reaction analogous to (2.1),

$$CH_3OH + H_2O \rightarrow 3H_2 + CO_2 - \Delta H^0,$$ (2.41)

where ∆H⁰ = 131 kJ mol⁻¹ (liquid reactants) or 49 kJ mol⁻¹ (gaseous reactants) has to be added as heat. The process works at a temperature of 200- 350°C. The water-gas shift reaction or its reverse may be operating as well,

$$CO_2 + H_2 \leftrightarrow CO + H_2O + \Delta H^0, \qquad (2.42)$$

with ∆H⁰ = −41 kJ mol⁻¹ . This could lead to CO contamination of the hydrogen stream, which is unacceptable for fuel cell types such as the proton exchange membrane or alkaline fuel cells (above ppm-level) and only little acceptable (up to 2%) for phosphorous acid fuel cells. Fortunately, the CO production is low at the modest temperatures needed for steam reforming, and adjusting the amount of surplus steam (H2O) may be used to force the reaction (2.42) to go towards the left at a rate achieving the desired reduction of CO (Horny et al., 2004). At the high end of the temperature regime, this control of CO becomes more difficult. However, use of suitable membrane reactors with separate catalysts for steam reforming and water-gas shift allows this problem to be overcome (Lin and Rei, 2000; Itoh et al., 2002; Wieland et al., 2002). Figure 2.28 shows a typical layout of a modest-scale tubular reactor for on-board automotive applications. A typical thermal efficiency of hydrogen formation by this method is 74%, with near 100% conversion of the methanol feed.

It is possible to make the process autothermal−in other words, to avoid having to heat the reactants−by adding the possibility of the exothermic partial oxidation process (see section 2.1.2),

$$CH_3OH + \frac{1}{2}O_2 \rightarrow CO_2 + 2H_2 - \Delta H^0.$$ (2.43)

Here, ∆H⁰ = −155 kJ mol⁻¹ . By a suitable combination of (2.41) and (2.43) the overall enthalpy difference may become approximately zero. There are still problems in controlling the temperature across the reactor, because the oxidation reaction (2.43) is considerably faster than the steam reforming (2.41). Proposed solutions include the use of a catalyst filament wire design leading to near-laminar flow through the reactor (Horny et al., 2004).

Figure 2.28. Schematic cross section of membrane steam reformer for automotive uses (Lin and Rei, 2000; Itoh et al., 2002).

Catalysts traditionally used for the steam reforming process include a Cu-Zn catalyst containing mole fractions of 0.38 CuO, 0.41 ZnO, and 0.21 Al2O3 (Itoh et al., 2002; Matter et al., 2004) and a metallic Cu-Zn catalyst in case of the wire concept (Horny et al., 2004). For the oxidation reaction, a Pd catalyst is suitable−for example, forming the membrane illustrated in Fig. 2.28 (using Pd91Ru7In2; Itoh et al., 2002). Also Pt has been used, as indicated in Fig. 2.27.

For higher hydrocarbons, such as gasoline, steam reforming has to be performed at high temperatures. Using conventional Ni catalysts, the temperature must exceed 900°C, but the addition of Co, Mo, and Re or the use of zeolites allows a reduction of the temperature by some 10% (Wang et al., 2004; Pacheco et al., 2003).

Figure 2.29. Minimum energy path of a hydrocarbon (methylcyclohexane) moving through an Al2O3 catalyst cavity channel with additional Pt atoms. In the upper part, energy in kcal/mol is shown as a function of distance along the path in Å. The lower part gives a top view of the hydrocarbon trail through the catalyst. A molecular dynamics model was used. Reproduced from J. Szczygiel and B. Szyja (2004). Diffusion of hydrocarbons in the reforming catalyst: molecular modelling. Journal of Molecular Graphics and Modelling 22, 231-239. With permission from Elsevier.

Small-scale applications of fuel cells, aimed at increasing operation time before recharge over that of current battery technology, have stimulated the development of miniature reformers for both methanol and other hydrocarbons (Palo et al., 2002; Presting et al., 2004; Holladay et al., 2004). Although the thermal efficiency of these devices aiming at use in conjunction with fuel cells rated in the range of 10 mW to 100 W is lower than for larger units, system efficiencies compare favourably with those of existing devices in the same power range. Applications will be discussed in section 4.6.

A theoretical understanding of the diffusion of hydrocarbons through the porous catalyst layer (see Fig. 2.28) may be obtained by simulations using semi-classical molecular dynamics (as in Fig. 2.3). Such calculations have been performed for the penetration of various hydrocarbons through Al2O3 catalysts with and without Pt insertions (Szczygiel and Szyja, 2004). As indicated in Fig. 2.29, it is found that fuel transport depends on both cavity structure and the adsorption on internal catalyst walls.

For high-temperature fuel cells (such as solid oxide fuel cells), reforming of methanol and certain other hydrocarbons to hydrogen occurs naturally within the fuel cell itself, in the presence of a catalyst such as Ru, but with an efficiency varying somewhat according to feed (Hibino et al., 2003).

2.3 Hydrogen conversion overview

This section gives an overview of current and contemplated uses of hydrogen: for combustion, further conversion into liquid or gaseous fuels and fuel cell production of electricity, possibly with associated heat production. Details of a number of the technologies employed, specifically those involving fuel cells, are the subject of subsequent chapters. The non-energy uses and energy uses not involving fuel cells are only described in the present section.

2.3.1 Uses as an energy carrier

The use of hydrogen as an energy carrier may be contemplated without necessarily assuming it also to be an energy source. Energy forms difficult to transport or store for time-displaced uses could be converted into hydrogen and the hydrogen transported in containers or pipelines, possibly combined with hydrogen stores, and later converted to another energy form suited for the final energy uses. The low energy-density by volume (see Table 2.1) sets certain limits to such uses of hydrogen, but there are many applications where density is not a big issue.

Advantages of hydrogen as an energy carrier include low environmental impacts and versatility in usage. A possible disadvantage might be the high requirements for tightness of containers and pipelines needed to avoid leakage.

Particular areas where an intermediate energy carrier is desirable include intermittent primary energy sources such as solar and wind energy, but also other energy sources linked to not easily storable electricity−for example, nuclear energy. If the use of fossil energy is extended, the quest for avoiding CO2 emissions could lead to extended conversion of, for example coal to hydrogen, in order to be able to control greenhouse gas emissions at the primary stage (e.g., by forming carbonates and sending them to a repository such as old natural gas wells). A closer inspection of scenarios for fossilhydrogen, nuclear-hydrogen, and renewables-hydrogen energy supply systems is made in sections 5.3-5.5.

2.3.2 Uses as an energy storage medium

Seen as a storage medium, gaseous hydrogen is convenient in underground stores such as aquifers or flushed-out salt dome extrusions, requiring only a better lining compared to present use of the same geological formations for natural gas storage. The low volume density makes hydrogen storage in manufactured containers somewhat expensive, but compressed hydrogen storage is still considered a convenient solution for many applications in industry and for at least the first generations of fuel cell hydrogen vehicles and household-size generators. These and other storage options will be described further in section 2.4, including liquefied and molecularly trapped hydrogen stores.

It is possible for hydrogen to play a role as storage medium in connection with several types of energy systems, independent of whether or not hydrogen is also used as a general energy carrier. Renewable energy systems require energy storage in order to become self-contained solutions, and hydrogen satisfies a range of the storage requirements of such systems, particularly if an affordable fuel cell becomes available for recovery of energy in the versatile form of electricity. As just mentioned, just about any future energy system would derive benefits from access to hydrogen energy storage.

2.3.3 Combustion uses

Hydrogen can be used as a fuel in conventional spark-ignition engines such as the Otto and Diesel engines used in motorcars. The engine efficiency is as high as for gasoline or diesel fuels, and the hydrogen flame expands rapidly from the kernel of ignition (see Table 2.3). However, due to the lower energy density at the pressures suitable for piston cylinders, the volume of displacement must be two to three times as large as for gasoline engines, causing problems of space within engine compartments of passenger vehicles. One motorcar manufacturer active in developing hydrogen-fuelled passenger cars has used huge 8- or 12-cylinder engines with displacement volumes of over 4 litres in order to approach an acceptable performance (BMW, 2004). Efficient conventional gasoline or diesel cars have total displacement volumes of around 1.2 litres, distributed on three to four cylinders (VW, 2003).

Property	Hydrogen	Methanol	Methane	Propane	Gasoline	Unit
Minimum energy for ignition	0.02	−	0.29	0.25	0.24	10-3 J
Flame temperature	2045		1875		2200	°C
Auto-ignition temperature in air	585	385	540	510	230−500	°C
Maximum flame velocity	3.46	−	0.43	0.47		m s-1
Range of flammability in air	4−75	7−36	5−15	2.5−9.3	1.0−7.6	vol.%
Range of explosivity in air	13−65		6.3−13.5		1.1−3.3	vol.%
Diffusion coefficient in air	0.61	0.16	0.20	0.10	0.05	10-4 m2 s-1

Table 2.3. Safety-related properties of hydrogen and other fuels (with use of Dell and Bridger, 1975; Zittel and Wurster, 1996).

Figure 2.30 shows the results of a computer simulation of the hydrogen combustion process in air (in a combustion chamber that could represent an engine cylinder or perhaps rather a gas turbine), substantiating the remark on the rapid consumption of the hydrogen injected. H2 enters from the left side. The oxygen distribution shows "unused" oxygen along the chamber, into which air is drawn through numerous holes in the outer surface. At the bottom of Fig. 2.30 the distribution of nitrogen oxides formed is shown. It is similar to the distribution of high temperatures just shown, as the formation of nitrogen oxides increases dramatically for temperatures above 1700 K. High levels of NOx formation are a problem for combustion uses of hydrogen, which in this case is no longer pollution-free.

The behaviour of hydrogen in thermodynamical engines has been established both by simulations such as the one shown in Fig. 2.30 and also by measurements−for example to establish the influence of compression ratios (Verhelst and Sierens, 2003; Karim and Wierzba, 2004). The wide range of flammability and easy ignition has two sides. One is the smooth operation at low loads. However, at high loads there are a number of problems with preignition, backfiring or knocking that have to be addressed (Shoiji et al., 2001). There are significant safety aspects, due to the wide flammability range of hydrogen (see Table 2.2). These have to be controlled, such as by doublevalve systems on each cylinder (NFC, 2000). Work has also been done on engines fuelled by mixtures of hydrogen with gasoline or with natural gas (Fontana et al., 2004; Akansu et al., 2004).

Due to the large amounts of hydrogen by volume needed for use in internal combustion engines, a decent operational range for a passenger car of conventional dimensions requires much higher H2 storage densities than possible on gas form. Therefore, liquid hydrogen is used, implying cooling to a

Figure 2.30. Simulation of a hydrogen combustion chamber. Top to bottom: H2, O2, temperature, and NO distributions (From Weydahl et al. (2003). Proc. 14th World Hydrogen Conf. Used with permission from the Canadian Hydrogen Association).

temperature of 20 K (with an energy use that further reduces the efficiency) and using very special filling stations. These technologies have been developed and used for the prototype hydrogen combustion cars (Fischer et al., 2003). Special tanks for containing liquid hydrogen have been constructed, using many layers of metal cylinders with high-grade insulation between them (Michel et al., 1998). Even so, there is an escape of heat and a hydrogen-leaking problem to handle.

Hydrogen combustion is somewhat easier to incorporate into buses, because more generous sizing of the engine compartment is a smaller fraction of the overall size and−due to slower operating speeds (in urban use)−buses can accommodate stored hydrogen on the roof, as it is done in the prototype hydrogen buses manufactured by MAN (Knorr et al., 1998). Also, ships could accommodate hydrogen fuel instead of diesel in conventional engines.

The suggestion of using hydrogen for direct combustion has received a good deal of attention in the case of airplanes. Stores of liquid hydrogen may be incorporated into wings or fuselage, and near-conventional gas turbine engines can be used. Early tests of liquid hydrogen aircraft include the 1957 Boeing B57 bomber and in 1988 the Tupolev TU-154, and recently, a similar project named Cryoplane based on the Airbus A310 aircraft has been started (Pohl and Malychev, 1997; Klug and Faass, 2001). The more bulky fuel stores have lead to suggestions that these planes should operate at lower heights, which would have positive impacts on NOx emissions (Svensson et al., 2004). This also makes the impact of water emissions less important, because the absorption of solar radiation by water (leading to greenhouse warming) is more pronounced above 10 km. Design for an airplane capable of operating from zero to Mach 5 (5 times the speed of sound, which is 1.22 × 10⁶ m s −1 at sea level) has been forwarded, with the vehicle using liquid hydrogen in the conventional gas turbine turbofan mode up to Mach 3 and a ramjet rocket mode above that speed (Qing and Chengzhong, 2001).

Liquid H2 has been used in space exploration vehicles for a number of years. Here the total mass to lift-off is an essential limiting parameter. Thus the high energy content by weight makes hydrogen preferable to any other fuelbased system, and the additional complication of cryogenic storage is deemed worthwhile. The engine may be a gas turbine or a rocket engine, depending on the mode of operating the space vehicle (cruising within Earth's atmosphere or escaping the gravitation of Earth). Both hydrogen and oxygen have to be carried in liquid form, as the space environment outside Earth's atmosphere does not contain air or oxygen. A basic feature of a rocket engine is the nozzle through which propellant gases are exhausted in order to provide the forward thrust. New developments in highperformance multiple-nozzle design are still ongoing (Yu et al., 2001), as are experiments with multiple fuels (typically liquid hydrogen and solid hydrocarbons) to suit the requirements at different stages of a near-Earth space journey with reusable shuttle-type space vehicles (Chibing et al., 2001). The optimum relative location of liquid hydrogen and liquid oxygen injection sites has been studied by experiment and simulation (Kendrich et al., 1999).

2.3.4 Stationary fuel cell uses

The uses of hydrogen in fuel cells for stationary markets have received considerable attention. Two areas of usage have been investigated. One is the electric power production sector, where the most efficient fuel cell types (solid oxide or molten carbonate) may have a future when they are sufficiently developed with regard to both technology and cost. The other area, which seems closer to the market introduction, aims at small, singlebuilding units based on the proton exchange membrane fuel cell technology, replacing natural gas-fired heat burners and thereby making the house owner a provider of both heat and electricity. The details of these developments will be dealt with in section 4.5 and in the scenarios in Chapter 5.

2.3.5 Fuel cell uses for transportation

Hydrogen and fuel cells could make the most important contribution in the transportation sector, where the introduction of alternative energy sources such as renewables has been most elusive. While hydrogen produced from renewable energy sources is the only sustainable solution for the long term, the use of hydrogen based upon natural gas may constitute a convenient transition solution. The most significant cost reduction needed for these scenarios to come true is for the fuel cell itself (assumed to be a proton exchange membrane type), but also the infrastructure change associated with distribution of hydrogen may constitute a significant cost. These issues will be dealt with in detail in sections 4.1-4.3 and in Chapter 5 on implementation.

2.3.6 Direct uses

It has been suggested that one of the first uses of hydrogen was for brick making (Bao, 2001). Fired bricks replaced sun-dried ones for uses in luxury buildings more than 5000 years ago in Mesopotamia, and lead-glazed ones appeared from about 3000 years ago (Hodges, 1970). Only much later did kiln bricks become common in the Mediterranean region, due to the abundance of other building materials such as stone and marble (although the kiln technology was itself in widespread use for pottery). In China, pottery making seems imported from the West (of Asia). Little evidence for fired bricks exists before the construction of the Great Wall some 1000 years ago. The bricks used here are grey (rather than the red shades characterising many modern clay bricks), which is interpreted as them being heated with water and carbon monoxide in high-temperature kilns, where they are subjected to the water-gas shift reaction followed by reduction of clay,

$$CO + H_2O \rightarrow CO_2 + H_2$$

$$\downarrow \qquad (2.44)$$

Fe2O3 + H2 → 2FeO + H2O. A 17th-century Chinese source describes this process as "fire and water

stimulate each other to make the best-quality bricks" (Yinxing, 1637). The element hydrogen was unknown until the 19th century. Processes like (2.44) involving H2 are in use today for the production of a range of materials.

The current main use of hydrogen is still in industry, where it is a standard commodity usually distributed in pressure containers in the form of gas flasks. In industrial areas such as Ruhr in Germany or London in England, hydrogen distribution through pipeline networks has been common.

Some 60% of the hydrogen use in industry is for ammonia production. The process is

$$N_2 + 3H_2 \rightarrow 2NH_3 - \Delta H^0$$ (2.45)

with ∆H⁰ = −107 kJ mol⁻¹ at 425°C and 21 MPa. An iron catalyst is required. Since the process is exothermal, the temperature is as low as consideration of reaction velocity allows. At this temperature, the conversion efficiency is 15 to 20%, depending on the catalyst used (Superfoss, 1981; Zhu et al., 2001).

Ammonium salts can be used for hydrogen storage. Hydrogen is further used in hydro-cracking and -refining, methanol synthesis (section 2.2.3) and aldehyde production. Future coal liquefaction also relies on hydrogen input.

2.4 Hydrogen storage options

The most suitable form of hydrogen storage depends on the application. Applications in the transportation sector require storage at a volume that can be accommodated within the vehicle and at a weight that does not limit the performance of the vehicle (for space applications one that still allows lift-off and disengagement from Earth's atmosphere). Also, for buildingintegrated applications, storage volume must usually be restricted, while dedicated storage at power plants or remote locations may allow for more latitude. Examples of storage density by mass or by volume are given in Table 2.4. Hydrogen storage is discussed following in terms of compressed gas, liquefaction, cryo-adsorbed gas storage in activated carbon, metal hydride storage, carbon nanotube storage, and reversible chemical reactions.

| Storage form | | Energy density | | | |------------------------------------|----------------|-----------------|--------|--| | | kJ kg−1 | MJ m−3 | kg m−3 | | | Hydrogen, gas (ambient 0.1 MPa) | 120 000 | 10 | 0.090 | | | Hydrogen, gas at 20 MPa | 120 000 | 1 900 | 15.9 | | | Hydrogen, gas at 30 MPa | 120 000 | 2 700 | 22.5 | | | Hydrogen, liquid | 120 000 | 8 700 | 71.9 | | | Hydrogen in metal hydrides | 2 000−9
000 | 5 000−15
000 | | | | Hydrogen in metal hydride, typical | 2 100 | 11 450 | 5 480 | | | Methane (natural gas) at 0.1 MPa | 56 000 | 37.4 | 0.668 | | | Methanol | 21 000 | 17 000 | 0.79 | | | Ethanol | 28 000 | 22 000 | 0.79 | |

Table 2.4. Energy density by weight and volume for various hydrogen storage forms, and mass density, including for comparison natural gas and biofuels (with use of Sørensen, 2004a; Wurster, 1997b).

2.4.1 Compressed gas storage

Storage of hydrogen in compressed gas form is the most common storage form today. Standard cylindrical flasks use pressures of 10-20 MPa, and fuel cell motorcar stores are currently in the range of 25-35 MPa. Tests are ongoing with pressure increased to 70 MPa in order to be able to store enough energy to obtain acceptable ranges for passenger cars, in particular the less efficient ones used in North America. While flasks for stationary uses are usually made of steel or aluminium-lined steel, weight considerations make composite fibre tanks more suitable for vehicle application. A typical design involves a carbon-fibre shell with inside polymer liner and outside reinforcement (in the United States, gunfire-proof). The first approved 70-MPa system holds 3 kg of hydrogen with an overall system weight of 100 kg (Herrmann and Meusinger, 2003).

Compression may take place at a filling station, receiving hydrogen from a pipeline. The energy requirement depends on the compression method. The work required for isothermal compression at temperature T from pressure P1 to P2 is of the form

$$W = A T \log \left( P_2 / P_1 \right), \tag{2.46}$$

derived from the modified gas law

$$PV = AT, (2.47)$$

where A is the gas constant R = 4 124 J K⁻¹ kg⁻¹ times an empirical, pressuredependent correction valid particularly for hydrogen and taken to decrease from 1 at low pressures to around 0.8 at 70 MPa (Zittel and Wurster, 1996; Herrmann and Meusinger, 2003). Adiabatic compression requires more energy input, and the preferred solution is a multistage intercooled compressor, which typically reduces the energy input to about half of the one for a single stage (Magazu et al., 2003). For compression to about 30 MPa, the compression energy for the multistage concept is about 10 MJ/kg or some 10% of the energy in the hydrogen stored (Table 2.4). The further energy required for transferring the compressed gas to the tank in a vehicle is minute in comparison, and the transfer time for the flask sizes mentioned is under 3 minutes. Although this is still some 60 times slower than the energy flow from a gasoline filling station to the car tank (Sørensen, 1984), it is deemed acceptable.

Figure 2.31. Types of underground compressed air storage: (a) storage in salt cavity, (b) rock storage with compensating surface reservoir, and (c) aquifer storage (from B. Sørensen, Renewable Energy, 2004a, used with permission from Elsevier).

Safety of compressed hydrogen tanks in proximity to vehicle passengers has been the subject of intense investigations, including crash tests and drops from considerable heights, as well as behaviour during fires−particularly those confined in the long tunnels that are found in many highway systems, leading traffic through mountains or under straits and other water bodies (Carcassi et al., 2004; FZK, 1999). Incidences spanning from minor leaks to tank failure and explosion can usually not be handled by the safety distances used for stationary stores. Questions of safety standards will be discussed in section 5.2.

Power requirements of portable hydrogen equipment such as cameras, mobile and smart phones, or laptop computers currently covered by batteries, typically of lithium ion type, could with a small fuel cell and some 10-20 g of directly or indirectly stored hydrogen prolong operational time by a factor of 5-10. A discussion of such options using direct methanol fuel cells is made in section 4.6.

For stationary hydrogen storage on a large scale, underground caverns or cavities are an appealing option, often offering storage solutions at a very low cost. Three possibilities of interest are salt dome intrusions, cavities in solid rock formations, and aquifer bends.

Cavities in salt deposits (Fig. 2.31a) may be formed by flushing water through the salt. The process has already been used successfully in a number of cases in connection with storage of compressed natural gas. Salt domes are salt deposits extruding upwards toward the surface, therefore allowing cavities to be formed at modest depths. The former industrial hydrogen salt dome store at Teeside, UK, is currently being restored as a hydrogen technology demonstration site (Taylor et al., 1986; Roddy, 2004).

Rock cavities (Fig. 2.31b) may be either natural or excavated, with walls properly sealed to ensure air-tightness. If excavated, they are considerably more expensive to make than salt caverns, but the latter only exists in a limited number of locations around the world.

Aquifers are layers of high permeability, permitting underground water flows along the layer. Gases may be stored in aquifers by displacing water (Fig. 2.31c), provided that the geometry includes upward bends allowing a gas pocket to be held in place by water on the sides. Above and below the aquifer must be layers of little or no permeability for the gas in question. This is often the case for clay layers. Aquifers that may be suitable for hydrogen storage are found in many parts of the world, except for areas dominated by rock all the way to the surface. These will have to use the more expensive rock storage shown in Fig. 2.31b.

The selection and preparation of sites for any of these gas stores is a fairly delicate process, because tightness can rarely be guaranteed on the basis of geological test drillings and modelling. The detailed properties of the cavity will not become fully disclosed until the installation is complete. The ability of the salt cavern to keep an elevated pressure may turn out not to live up to expectations. The stability of a natural rock cave, or of a fractured zone created by explosion or hydraulic methods, is also uncertain until actual fullscale pressure tests have been conducted. For the aquifers, the decisive measurements of permeability can only be made at a finite number of places, so surprises are possible due to rapid permeability change over small distances of displacement (Sørensen, 2010a).

The stability of a given cavern is further influenced by two factors: the temperature variations and the pressure variations. It is possible to keep the cavern wall temperature nearly constant, either by cooling the (possibly compressed) hydrogen before letting it down into the cavern or by performing the compression so slowly that the temperature only rises to the level prevailing on the cavern walls. The latter possibility (isothermal compression) is impractical for most applications because excess power must be converted at the rate at which it comes. Most systems therefore include one or more cooling steps. With respect to the problem of pressure variations, when the store holds different amounts of energy, the solution may be to store the hydrogen at constant pressure but variable volume (a possibility offered naturally by some aquifer stores). For the underground rock cavern, a similar behaviour may be achieved by connecting the underground reservoir to an open surface reservoir (indicated in Fig. 2.31b), so that a variable water column may take care of the variable amounts of hydrogen stored at the constant equilibrium pressure prevailing at the depth of the cavern. This kind of compressed energy storage system may alternatively be viewed as a pumped hydro storage system, with extraction taking place through gasdriven turbines rather than through water-driven turbines.

The aquifer storage system shown in Fig. 2.31c would have an approximately constant working pressure, corresponding to the average hydraulic pressure at the depth of the hydrogen-filled part of the aquifer. The stored energy E may, in general, be expressed as

$$E = -\int_{V_0}^{V} P \, dV, \tag{2.48}$$

considering the compressed gas cavern store at pressure P as a cylinder with a piston at positions describing the volumes V0 and V.

In the aquifer case, E simply equals the pressure P times the volume of hydrogen gas displacing water in the aquifer. This volume equals the physical volume V times the effective porosity *p*−that is, the fractional void volume accessible to intruding hydrogen gas (there may be additional voids that the incoming gas cannot reach) −so the energy stored may be written

$$E = pVP. (2.49)$$

Typical values are p = 0.2 and P around 6 × 10⁶ N m-2 at depths of some 600 m, with useful volumes of 10⁹ to 1010 m3 for each site. Such sites are already in use for storage of natural gas (see section 5.1.1).

An important feature of an energy storage aquifer is the time required for charging and emptying. This time is determined by the permeability of the aquifer. The permeability is basically the proportionality factor between the flow velocity of a fluid or gas through the sediment and the pressure gradient causing the flow. The linear relationship assumed may be written

$$v = -K (\eta \rho)^{-1} \partial P / \partial s, \tag{2.50}$$

where v is the flow velocity, η is the viscosity of the fluid or gas, ρ is its density, P is the pressure, and s is the path length in the downward direction. K is the permeability. In SI units it has the dimension of m2 . The unit of viscosity is m2 s-1. Another commonly used unit of permeability is the darcy. One darcy equals 1.013 × 1012 m2 . If filling and emptying of the aquifer storage are to take place in a matter of hours rather than days, the permeability has to exceed 1011 m2 . Sediments such as sandstone are found with permeabilities ranging from 1010 to 3 × 1012 m2 , often with considerable variation over short distances.

In practice, there may be additional losses. The cap-rock bordering the aquifer region may not have negligible permeability, implying a possible leakage loss. Friction in the pipes leading to and from the aquifer may cause a loss of pressure, as may losses in the compressor and turbine. Typically, losses of about 15% are expected in addition to those of the power machinery. Large salt dome and aquifer stores in operation for natural gas are described in the implementation chapter (section 5.1.1), as they may have future uses for hydrogen, along with new installations in similar geological formations.

2.4.2 Liquid hydrogen storage

Liquid hydrogen storage requires refrigeration to a temperature of 20 K, and the liquefaction process requires an industrial facility expending a minimum of 15.1 MJ/kg. The actual energy spending is nearly three times higher for the refrigeration techniques available at the present. The liquefaction process requires very clean hydrogen, as well as several cycles of compression, liquid nitrogen or helium cooling, and expansion. The subsequent transfer to a filling station and from there to a store within a vehicle (assuming automotive application) uses little energy by comparison, and like the situation for compressed hydrogen filling, can be accomplished in a few minutes. Pressures used are only slightly above the atmospheric, 0.6 MPa being a typical value. The liquid hydrogen storage technology was initially developed for space vehicles.

A vessel for storing liquid hydrogen consists of several layers (of metal) separated by highly insulating material, often involving low "vacuum-like" pressures of about 0.01 Pa and a resulting heat conductivity of 0.05 W K⁻¹ (Chahine, 2003). Including the entire storage device, the storage capacity by volume becomes roughly half the figures given in Table 2.4, for store sizes of the order of 130 litres. This means that liquefaction energy amounts to at least 30% of the stored energy.

A significant problem is the boil-off of hydrogen from the store caused by the need to control tank pressures by venting valves (associated with the heat release of ortho- to para-H2 conversion). Insulation properties influences the boil-off, which will start after a dormancy period of a few days and then proceeds at a level of 3-5% per day (Magazu et al., 2003), with the lower values obtained by installing heat exchangers between outgoing cold hydrogen and air intake to regulate pressure. The boil-off has security implications, say for cars parked in a garage. Boil-off limits the usefulness of liquid vehicle storage except where a fairly continuous driving pattern is involved.

2.4.3 Hydride storage

Molecular hydrogen is dissociated into hydrogen atoms in the neighbourhood of metals and certain compounds as illustrated in Fig. 2.5. If the lattice structure of the metal or alloy is suitable, there are interstitial positions capable of accommodating the relatively small hydrogen atoms. The energy associated with these changes can be quite modest, as demonstrated by the thermodynamical plot of pressure versus hydrogen concentration shown in Fig. 2.32. Heat is released when hydrogen enters the lattice, and heat must be supplied to drive the hydrogen out of the lattice again. The mechanisms involved in Fig. 2.32 can be understood in terms of chemical thermodynamics (see, e.g., Morse, 1964).

Chemical thermodynamics

Consider a chemical reaction, which may be described in the form

$$\sum_{i=1}^{N} v_i M_i = 0, (2.51)$$

where Mi is the *ith substance (reactant or product) and ^νⁱ* is its stoichiometric coefficient. As the reaction proceeds, the amounts of each substance changes, and eventually an equilibrium is formed. Introducing the concentration, χⁱ , defined as the mole fraction of the *i*th substance, the stoichiometry (2.51) requires that changes of the substances involved have the relationship

$$d\chi_i = v_i dx$$ with the same $dx$ for all $i$ . (2.52)

For near-ideal multi-component gases, it is possible to use macroscopic average variables in a thermodynamical description, and write the change in the Gibb's free energy G as

Figure 2.32. The left side of the figure shows the equilibrium pressure (1 bar ≈ 0.1 MPa) as a function of the hydrogen concentration (hydrogen to metal ratios, the metal being here LaNi5) at three different temperatures. For temperatures below a critical value, the non-equilibrium α-phase (metal lattice intact) and β-phase (lattice expanded/modified by the presence of hydrogen atoms) are separated by a pressure plateau region where the reactants remain in thermodynamical equilibrium during the absorption of more hydrogen. In the right-hand part of the figure, the logarithm of the plateau pressure is plotted against inverse temperature. From L. Schlapbach and A. Züttel (2001). Hydrogen-storage materials for mobile applications, Nature 414, 353-358, reprinted by permission.

$$dG = \sum_{i=1}^{N} \mu_i \nu_i \, dx, \tag{2.53}$$

where µⁱ is the chemical potential of the *i*th substance. The equilibrium situation is characterised by a minimum of G. The quantity

$$KP = \prod{i=1}^{N} P_i^{\nu_i} \tag{2.54}$$

is called the equilibrium constant. It is expressed in terms of the partial pressures Pi of the *i*th substance raised to a positive power for the reaction products and to a negative power for the reactants. For ideal gases, the chemical potentials may be written as

$$\mu_i = \mu_i^o + RT \log(P_i), \tag{2.55}$$

where R is the gas constant and the superfix "0" denotes a reference situation. Inserting (2.55) into (2.53), one gets for the equilibrium situation

$$\Delta G^{0} = \sum{i=1}^{N} v{i} \mu{i}^{0} = -RT \sum{i=1}^{N} \log(P{i}) = -RT \log K{p}.$$ (2.56)

Introducing the enthalpy from the definition H = G + TS, the temperature dependence of the quantities in (2.56) can be expressed by van't Hoff's equation

$$\frac{\mathrm{d}\log K_P}{\mathrm{d}T} = \frac{\Delta H^0}{RT^2}.\tag{2.57}$$

In integrated form this reads (using ∆H⁰ = ∆*G⁰ + T*∆S⁰ to fix the integration constant)

$$\log K_P = -\frac{\Delta H^0}{RT} + \frac{\Delta S^0}{R} \,. \tag{2.58}$$

Finally, with use of Pi = χi P and the equilibrium conditions, one can rewrite (2.58) in the form (given in a slightly different way by Morse, 1964; Schlapbach and Züttel, 2001)

$$\log P - \log P^0 = -\frac{\Delta H}{RT} + \frac{\Delta S}{R} \tag{2.59}$$

valid for the equilibrium (plateau) values of P and P0 and at a given temperature T. This equation describes the linear curve in the right-hand side of Fig. 2.49, and it allows determination of the plateau pressures as a function of temperature illustrated in the left-hand side of Fig. 2.32. The dominant contribution to ∆S is the dissociation energy of hydrogen molecules, −130 J K⁻¹ mol⁻¹ , and for the T = 300 K (25°C) curve to have its plateau pressure equal to a normal atmospheric sea-surface value of 0.1 MPa (≈ 1 bar), ∆H must be −39 kJ mol⁻¹ (of hydrogen; Züttel, 2004). Actual enthalpy values for different substances are discussed following.

Metal hydrides

Some metal alloys can store hydrogen at volume densities more than twice that of liquid hydrogen. However, if the mass storage density is also important, as is normally the case in automotive applications, then storage densities are still only 10% or less, compared to those of conventional fuels (see Table 2.4 and the summary in Fig. 2.37). This makes the concept doubtful for mobile applications, but interesting for decentralised hydrogen storage (as required, e.g., in one of the scenarios described in section 5.4), due to the positive aspects of near loss-free storage at near-ambient pressures (0.06−6 MPa) and high safety in operation with hydrogen transfer accomplished by adding or withdrawing modest amounts of heat, according to

$$Me + \frac{1}{2}xH2 \leftrightarrow MeHx,$$ (2.60)

where Me stands for a metal or metal alloy, such as a binary AmBn or higher order alloy. Among single-metal hydrides, MgH2 and PdH0.6 are the most well-studied ones, with hydrogen mass percentages of 7.6 and 0.6 and decomposition temperatures of 330 and 25°C (Grochala and Edwards, 2004). Binary and higher metal alloys generally store similar amounts of hydrogen by volume, but less hydrogen by mass (see Fig. 2.37).

Figure 2.33. Structure of MgH2 proposed on the basis of X-ray spectroscopy and quantum chemical calculations (see text). The larger atoms are Mg, and the smaller ones are hydrogen.

The hydride of the composition MgH2 seems the most interesting singlemetal hydride, with its fairly high hydrogen fraction of 7.6% by mass. Although the desorption temperature is higher than ambient (about 330°C), it could be made available in motorcars, although this would lower the overall efficiency. The enthalpy of the exothermal transfer of hydrogen into the metal lattice is —74.5 kJ mol⁻¹ (Sandrock and Thomas, 2001), and Fig. 2.33 shows the tetragonal crystal structure of MgH2. Quantum chemical calculations are used to follow the swelling of the lattice with incorporation of hydrogen in sheets or bulk materials (Shang et al., 2004; Liang, 2003a). The kinetic restraints at near-ambient pressures make the absorption and desorption processes very slow, and desorption times in the range of hours are unacceptable, at least for mobile applications (see also the modelling subsection at the end of this section).

Other binary hydrides can be formed with light elements, without improvement in performance. The structure of NiH is shown in Fig. 2.34. For heavy metals, only palladium has received some attention, due to rapid absorption and desorption at ambient temperature and pressure. However, only 0.6% hydrogen by mass is stored, which, together with high metal cost, makes this option unattractive.

Figure 2.34. Structure of an NiH lattice. The large atoms are Ni, and the small ones are H. Note that the Ni lattice structure with Miller indices (111) is the same as that used in the catalyst reaction modelling shown in Figs. 2.5 and 2.6.

Hydrides with two metal constituents studied include a range of Mg alloys, including interesting ones with Ni and Fe, as well as alloys containing Al, Fe, Ni, Ti, and La, as well as other metals. Of the total number of atoms−say, 6 × 1028 per m3 −the hydrogen fraction is usually below the value of 2 for MgH2. The mass fraction achieved for these hydrides is often below 2%, with the LaNi5H7 illustrated in Figs. 2.35 and 2.36 as a typical example. Its volumetric storage density is about 115 kg H2 per m3 , and the enthalpy evolution during absorption is —30.8 kJ mol⁻¹ (Sandrock and Thomas, 2001). While the maximum number of hydrogen atoms that can be accommodated within the LaNi5 hexagonal structure is 7 (Fig. 2.36), sometimes only 6 are incorporated, possibly due to the swelling of those unit cells that already have accepted 7 H atoms. Attempts have been made to perform quantum chemical calculations to explore these effects and to determine the locations of absorbed hydrogen atoms, which is not at midpoints of the interstitial sites but must be strongly influenced by the Coulomb binding forces exerted by the lattice atoms, see Fig. 2.35 (Tatsumi et al., 2001; Morinaga and Yukawa, 2002).

Figure 2.35. Structure of LaNi5H7 proposed on the basis of X-ray spectroscopy and quantum chemical calculations (see text). The atoms are illustrated with declining size from La over Ni to H (not all of the H's are visible; compare Fig. 2.53).

Figure 2.36. LaNi5H7 as in Fig. 2.35, but from an angle better showing the hydrogen atom positions. The atoms are illustrated with declining size from La over Ni to H.

Figure 2.37. Volume change of a metal hydride for increasing H content. From S. Orimo and H. Fujii (2001). Materials science of Mg-Ni-based new hydrides. Applied Physics A72, 167-186. Used by permission from Springer-Verlag.

Figure 2.38. Overview of hydrogen storage properties. A number of hydrides are characterised by mass and volume H2 densities (axes) as well as by simple density (scale on top). The circles represent chemisorbed hydrogen on carbon, and for comparison the curves (in lower part of the figure) illustrate the properties of liquid and compressed hydrogen storage in steel and composite containers. Abbreviations: dec = decomposition, m.p. and b.p. = melting and boiling point. From A. Züttel (2004). Hydrogen storage methods. Naturwissenschaften 91, 157-172. Used by permission from Springer-Verlag. See Züttel et al. (2004) and Schlapbach and Züttel (2001).

The swelling as a function of hydrogen absorbed is shown in Fig. 2.33, for another well-studied hydride, Mg2NiH4, the storage capacity of which is seen in Fig. 2.38. The lattice absorption cycle also performs a cleaning of the gas, because impurities in the hydrogen gas are too large to enter the lattice.

The metal hydride with the highest mass percentage of hydrogen found so far is Mg2FeH6 (Fig. 2.38) but decomposing at a temperature near 400°C for ambient pressure (Fig. 2.39). Several higher alloys have been studied, such as LaNi4.7Al0.3 based on the alloy shown in Fig. 2.35 (Asakuma et al., 2004). Increased stability is achieved by such substitution, but without definite improvements in storage properties (Züttel, 2004). The van't Hoff plots (as in the right-hand side of Fig. 2.32) for this and other alloys are shown in Fig. 2.39, again showing the difficulty in bringing the properties needed for automotive applications together.

Figure 2.39. Plateau pressure (see Fig. 2.32) for various metal hydride combinations as a function of inverse temperature. From A. Züttel (2004). Hydrogen storage methods. Naturwissenschaften 91, 157-172. Used by permission from Springer-Verlag.

The time needed for absorption depends not only on the alloy used, but also on its physical properties such as grain size. Figure 2.40 shows how dramatic such differences can be, using as an example the LaNi5 alloy discussed previously, with a catalyst helping the absorption process. The range of absorption and desorption times encountered in the alloys studied is illustrated in Fig. 2.41. The storage charge and discharge times are for most alloys in the range of 30-60 min, but there are ranges down in the area of 1 min, suitable for automotive applications. These require additives based on vanadium and carbon, as well as elevated temperatures and for absorption pressures of 10-15 times the ambient pressure. Nuclear Magnetic Resonance (NMR) studies have been performed in the hope of understanding how small additions of transitional metals can improve the performance (Kasperovich et al., 2010).

In addition to studying alloys with additives (e.g., Zhao and Ma, 2009; Lin et al., 2011), the possible advantage of using alloys under high pressure has been explored (Mori and Hirose, 2009). In all cases, the advantages observed have been marginal and often offset by other disadvantages (such as having to work with high pressure in consumer vehicles aimed for urban driving).

Figure 2.40. Hydrogen absorption in different forms of LaNi5:

a: polycrystalline;
b: nanocrystalline;
c: nanocrystalline with addition of catalyst.

From A. Zaluska, L. Zaluski, J. Ström-Olsen (2001). Structure, catalysis, and atomic reactions on the nano-scale: a systematic approach to metal hydrides for hydrogen storage. Appl. Phys. A72, 157-165. Used by permission from Springer-Verlag.

Models for following the hydrogen diffusion process may be based on Monte Carlo simulation or network simulation incorporating both a regular lattice structure and irregular grain regions (Herrmann et al., 2001).

A maximum hydrogen storage capacity in chemical structures may be estimated from close packing of hydrogen atoms with a distance of about 0.2 nm, as closer distances are unlikely if lattice atoms are to be accommodated (cf. Problem 2.6.2 in section 2.6). Some of the metal hydrides in Fig. 2.38 are not far below the implied density, so emergence of novel miracle materials is not to be expected.

Although failure to meet all the criteria required for use of metal hydride stores in automobiles so far has barred their introduction, there are in fact manufacturers offering the technology for special stationary purposes, where, for example, safety considerations disfavour use of compressed or liquid hydrogen tanks. The products offered have dealt with the problem of expansion in different ways. One company offers rolled up mats of hydride material (and heating coils) with generous room for expansion (the "Hy-Stor" trademark of Ergenics, Ringwood, NJ, owned by Hera in Quebec), while other products use stacks of boxes filled with granular hydride material (the simplest design using cylindrical boxes with a centre hole for heat application equipment).

Complex hydrides

The term complex hydrides has been used for metal hydrides thought to be involved in complex bindings, but the distinction as a separate class is perhaps not entirely warranted, because the nature of the quantum chemical binding exhibits a sliding transition between the different classical binding types. The hydrides called complex typically involve light atoms, which in general terms would seem suitable for high hydrogen-to-metal ratios. Hydrogen is often not located in the most obvious interstitial sites, but rather near certain lattice atoms as a result of favoured binding to the atoms in these locations. This is also the case in the metal hydride shown in Fig. 2.35, exposing the gradual scale of assigning hydrides as metal or complex. An equally important parameter is the macroscopic structure of the material, from solid lattices through grains of different sizes to nanoporous structures with a high ratio of surface to bulk atoms.

Figure 2.41a,b. Kinetics of hydrogen absorption (a: top) and desorption (b: below) in Mg alloys with V and C additives, at the temperature 350°C and pressure 1.4 MPa (absorption) and 10 kPa (desorption). The fastest reactions are for MgH2. Based on Liang, 2003a.

Figure 2.38 shows mass percentages up to 18 for LiBH4 at about 280°C, a substance first synthesised by Schlesinger and Brown (1940). A similar compound, NaAlH4, shows reversibility with absorption and desorption at the

more convenient temperatures of about 195°C, but with a considerably smaller mass percentage (Bogdanovic and Schwickardi, 1997; the 7.5% indicated in Fig. 2.38 includes H atoms not involved in the reversible storage reactions, as do the other complex hydride entries in the figure). The reactions involved for these substances, called alanates, are more complicated than simple hydrogen molecule dissociation and diffusion into

the lattice:

$$6$$ NaAlH₄ $\neq$ 2Na₃AlH₆ + 4Al + 6H₂ $\neq$ 6NaH + 6Al +9H₂ (2.61)

$$2Na_3AlH_6 \neq 6NaH + 2Al + 3H_2,$$ (2.62)

usually complemented by reactions for an accompanying Li-substituted alanate,

$$2Na_2LiAlH_6 \neq 4NaH + 2LiH + 2Al + 3H_2.$$ (2.63)

Figure 2.42. Molecular structure of NaAlH4 (a) and Na3AlH6 in its α-form (b) and βform (c). The small grey atoms are H, the large dark ones are Na, and the large lighter ones are Al.

The preceding four reactions in isolation each contribute 3.7, 5.5, 3.0, and 3.5% to the hydrogen mass percentage, which for the combined reaction system is about 5.6%. Figures 2.42a-c show the molecular structure of the sodium alanates. In particular, Na3AlH6 has two forms, of which the simple face centred cubic structure (Fig. 2.42c) only dominates at temperatures above 252°C (Arroyo and Ceder, 2004). A prototype of this type of system, in the form of a cylindrical canister with separate hydrogen access and cooling pipes, has been tested (Mosher et al., 2007) and modelled by finite element heat and mass transfer calculations (Hardy and Anton, 2009). Figure 2.43 gives the concentration building up of Na3AlH6 and NaAlH4, after injection of hydrogen into the cannister, according to (2.61) and (2.62). Only a 3 dimensional slice of the cylinder is modelled, due to the rotational and translational symmetry of the device.

Figure 2.43. Build-up of alanate-concentrations in model calculation of Hardy and Anton (2009). Upper slice-indent is hydrogen inlet. Lower indent is for cooling fluid. Note that each segment has a different scale. Used with permission from Elsevier.

The complex sequences of reactions may make the filling and emptying of the storage system too slow for some applications, although speed can be increased by using catalysts (typically Ti) in the reactions (2.61)-(2.63). Hydrogen evolution in early experiments took 10-30 h (Bogdanovic et al., 2000), and questions of stability have been raised (Nakamori and Orimo, 2004). The new modelling in Fig. 2.43 suggests possible future automotive uses.

In the lithium borohydride case, the "desorption" process of regaining hydrogen is (Züttel et al., 2003),

$$2\text{LiBH}_4 \rightarrow 2\text{LiH} + 2\text{B} + 3\text{H}_2, \tag{2.64}$$

a process proceeding at 300°C under the influence of an SiO2 catalyst, with a hydrogen mass yield of 13.8%. The choice of catalyst is not obvious. A Ti catalyst works for the reactions (2.61)-(2.63) and for Li3AlH6, but not for Li-AlH4. An explanation has been sought in terms of the energy changes following absorption of Ti atoms to a position well into the surface structure (Løvvik, 2004). Charging the system with hydrogen by the process reverse to (2.64) is possible, but complete reversibility has so far only been achieved at elevated pressure (10 MPa) and temperature (550°C) (Sudan et al., 2004).

If the questions of efficient synthesis, catalyst use, and stability are resolved, there are still issues of cost and safety to be addressed. The complex hydrides and related structures are relatively new research areas, where new ideas keep coming up. For example, hydrogen storage may occur in zeolites at a temperature of 77 K (Züttel et al., 2004), and in metal-organic frameworks (Rosi et al., 2003), with hydrogen mass adsorption on surface sites changing from 1 to 4.5%, as the temperature is decreased from ambient to 78 K. Another avenue of investigation is clathrate hydrates at 145 K (Mao et al., 2002). Clathrates are water-ice structures that form cages capable of accommodating extra hydrogen atoms (Sluiter et al., 2003; Profio et al., 2009). A different approach is to look at metal ammine complexes such as Mg(NH3)6Cl² for efficient H2 storage at 620 K (Christensen et al., 2005).

Modelling metal hydrides

In order to understand (or predict) the energetic advantage for hydrogen of penetrating a metal lattice and forming a hydride, quantum calculations of the chemical structures involved may be performed.

Consider first the behaviour of hydrogen approaching a nickel surface. This was modelled in Fig. 2.5 in connection with the description of the ability of a metal catalyst such as Ni to split the hydrogen molecule into two hydrogen atoms. This process was shown to take place about 0.1 nm outside the Ni surface. It is easy to extend the calculation to see what happens if the hydrogen atoms are allowed to penetrate into the Ni lattice, simply by using the same method to calculate potential energies for H positions below the top Ni layer. The result is shown in Fig. 2.44.

In order to penetrate into the Ni lattice, the two hydrogen atoms must pass a barrier that is lowest when the two H atoms form a molecule (d = 0.076 nm). The barrier is considerably higher than the dissociation barrier outside the top Ni surface. If the hydrogen atoms get inside the lattice, they can reach a fairly deep minimum in potential energy, located midway between the two Ni-layers and at a hydrogen atom distance of d ≈ 0.12 nm. The position of the Ni atoms is further away. One Ni atom is responsible for the rise in potential energy at the right-hand side of Fig. 2.44 (white area in top Ni layer). The next Ni atom is at the left-hand side of Fig. 2.44, just to the left of the boundary but not in the same plane as the one spanned by the H atoms and the Ni atom in the white area. So the most favourable place for the H atoms inside the Ni structure is as expected midway between the layers and approximately midway between the Ni atom positions. However, the depth of the potential energy minimum is nearly 2 eV less than that of separating the H atoms outside the Ni surface. As a result, Ni is not a suitable material for absorption of hydrogen and forming hydrides. Ni has this property in common with the other metal catalysts in the same group, Pd and Pt.

Figure 2.44. Potential energy surface in eV along two co-ordinates for two hydrogen atoms near a nickel surface: distance between hydrogen atoms d and height z of the centre of mass of the two H atoms over the (1,1,1) surface of the Ni-layer below the top layer. The quantum chemical calculation uses the density functional theory B3LYP and a set of basis functions called SV (see section 3.2). The energy scale has an arbitrarily chosen origin (Sørensen, 2004f).

The metals of group II in the periodic system comprise magnesium, the metal identified previously in the metal hydride subsection as forming a metal hydride without needing a binary metal structure. It is thus interesting to see if a quantum chemistry calculation will show magnesium as better suited for absorbing hydrogen than nickel.

This calculation presented following is carried out using density functional theory with periodic boundary conditions and fast multipole techniques (Kudin and Scuseria, 1998, 2000), rather than explicitly treating the periodic lattice as a single large molecule in the way done for the preceding Ni hydride.

Figure 2.45. Two configurations of Mg-2H periodic structure used in the potential energy calculations of Figs. 2.46-2.47. The situation on top corresponds to the MgH2 hydride equilibrium configuration (same as the one shown in Fig. 2.33), while the H atoms in the lower figure have been moved 0.5 nm out of the Mg lattice, to the left along the negative x-axis perpendicular to one of the lattice surface planes (Sørensen, 2004f).

Figure 2.45 (top) shows the optimised structure of MgH2, with H atoms occupying positions within the lattice leading to a minimum in potential energy. In order to compare this energy to that of the same system but with hydrogen atoms outside the Mg lattice, a sequence of calculations were made, pulling the hydrogen atoms out of the lattice but preserving their relative positions among them. This allows the periodic boundary condition method to be used. Figure 2.45 (bottom) shows a situation with the hydrogen set pulled out 0.5 nm from the equilibrium position. Figure 2.46 shows the map of the potential energy surface, as function of x, the pull direction perpendicular to the Mg lattice surface, and y, the co-ordinate describing displacements parallel to the Mg surface. Figure 2.47 shows the potential energy curve along the y = 0 line of displacement (the potential minimum for fixed x as a function of y, as seen in Fig. 2.46). For completeness, the potential energy of fully separated hydrogen atoms (calculated as the energy of the Mg lattice alone, plus the energy of the hydrogen atoms alone) is also indicated. All potential energies are given per unit cell of 2 Mg atoms and 4 H atoms, with arbitrary origin but the same for all displacement calculations.

It is seen, first of all, that the MgH2 metal hydride indeed has a potential energy 3.5 eV (per unit cell) lower than that of the Mg and H atoms separated, thereby substantiating the ability of the Mg lattice to absorb hydrogen

Figure 2.46. Potential energy plot of Mg-2H system as a function of displacement of the H atoms in two directions relative to the equilibrium position within the hydride. See text for further explanation (Sørensen, 2004f).

Figure 2.47. Potential energy plot for Mg-2H system as a function of *x*displacement (as in Fig. 2.46 but with y fixed at zero). See text for further explanation (Sørensen, 2004f).

without energy added externally. On their way in, the hydrogen atoms make the potential energy oscillate in concord with the positions of Mg atomic centres, with an excursion range of 3 eV but at all times more than 2 eV higher than the final hydride energy (Fig. 2.47). The y = 0 path into the Mg lattice allows the most free passage, as seen by the effect of varying y (Fig. 2.46), which is to rapidly increase the potential energy. The calculation gives the enthalpy of formation at standard temperature and pressure as −71.93 kJ mol⁻¹ . The experimental value at 435°C is —75.2 kJ/mol (Bogdanovic et al., 2000). Details of the calculation are the use of the SV basis (Schaefer et al., 1992) with automatic optimisation (Gaussian, 2003), and the PBEPBE parametrisation of the density functional exchange and correlation parts (Perdew et al., 1996; see the discussion in section 3.2). The periodicity was allowed to prevail for about 1.6 nm in every direction, which was verified to be enough to ensure stability of calculated energies.

Figure 2.48. Iso electron-density surfaces (at the value of 0.05) for the Mg (darkest) and H (lighter shaded) atoms of a unit cell in the equilibrium structure of the metal hydride MgH2 (Sørensen, 2004f).

Figure 2.48 shows the electron wavefunctions in one unit cell of the system, illustrated by the density surface where the squared density has fallen to 0.05. It is seen that the "size" of the wavefunction parts surrounding Mg and H atoms is very similar, denouncing the common explanation of hydrogen absorption in metal lattices based on hydrogen being "such a small atom".

Quantum chemistry calculations using theoretical approaches similar to the preceding one have recently been made for other absorption storage media, based, for example, on magnesium and titanium lattices (Tao et al., 2011).

2.4.4 Cryo-adsorbed gas storage in carbon materials

Hydrogen molecules may adsorb not just to metal surfaces, but also to surfaces of various solid materials including carbon, qualitatively following an adsorption and/or chemisorption curve of the type shown in Fig. 2.1. Adsorption of hydrogen takes place some 0.1 nm above the surface, in a single (mono) layer at a suitable low temperature (but above the hydrogen liquid phase transition temperature of 20 K). Too high a temperature leads to losses of adsorbed molecules by thermal disturbance, and for the only substance studied extensively so far, carbon, suitable temperatures are around the boiling point of nitrogen (77 K) and at pressures of about 10 MPa. Because only a monolayer is formed, the surface area must be as large as possible in order to get interesting storage parameters such as hydrogen mass or volume fraction. For a fully occupied monolayer, the hydrogen concentration is 1.2 × 10⁻⁵ mol m⁻² . The maximum amount of hydrogen adsorbed is thus directly proportional to the carbon surface area, and for thin sheets of graphene, activated carbon, or graphite, the proportionality factor is about 2% by mass (H/(C+H)) per m2 kg⁻¹ carbon surface area, with 1.5% established in practice (Nijkamp et al., 2001; Züttel, 2004). The possibility of raising pressures to near 30 MPa has been considered (Ahluwalia et al., 2010).

Substances like activated carbon contain micropores that allow adsorption through surface areas 2-3 orders larger than the geometrical surface. Compared with storage in liquid hydrogen, the difference between the 20 and 77 K temperature should reduce filling costs and reduce problems with escaping hydrogen. The mass percentage seems too small for being attractive in mobile applications (although the company Zevco in the UK has proposed the concept in connection with its alkaline fuel cell taxi prototype). Theoretical modelling of the physisorption (i.e., not chemisorption, see Fig. 2.1) process has been done using Monte Carlo simulation (Williams and Eklund, 2000), and charging behaviour of a concrete activated carbon medium has been studies by Jurewicz (2011).

The use of nanotubes as the carbon material attracted some attention a few years back. However, it has become clear that nanotubes cannot store hydrogen gas in its voids, but only as adsorbed to the sides, implying a storage capability no greater than that of any other carbon surface (in fact smaller due to the curvature of the surface) (Zhou et al., 2004).

Surface absorption of hydrogen on metal surfaces is a well-known phenomenon in the electrodes of batteries. It may also be at work for electrodes made of nano-structured carbon and accomplishes nearly the same mass percentage as cryogenic storage for suitable electrolytes, but at ambient temperature. Charging or discharging takes several hours or even days, so the concept is unsuitable for use in cars (Jurewicz et al., 2004).

2.4.5 Other chemical storage options

Once again the distinction between different types of chemical hydrogen storage is a bit artificial, as reactions beyond dissociation were already involved in several of the forms just described and notably in the use of hydrides called complex. But generally, any reversible reaction scheme with different amounts of molecular hydrogen on either side of the equation may be contemplated as a hydrogen storage device, and even a non-reversible device may serve the process if it can be manufactured with a hydrogen content that may be recovered under controlled circumstances, such as in a vehicle. Thus, the methanol reformer in a car storing methanol but converting it to hydrogen before using it in a fuel cell may be considered such a device, with hydrogen stored in the form of methanol.

A number of other chemical reactions have been contemplated for storage applications. Here only a single example shall be mentioned, involving the hydrocarbons decalin and naphthalene,

$$C{10}H{18} \neq C_{10}H_8 + 5H_2, \tag{2.65}$$

which proceeds to the right at ambient pressure and a temperature of T ≈ 200°C, with a platinum-based catalyst and heating corresponding to an enthalpy supply of ∆H⁰ = 8.7 kWh kg⁻¹ of H2 (Hodoshima et al., 2001). The reaction proceeds to the left at ambient or slightly lower temperatures (5°C). The weight percentage of hydrogen in decalin is 7.3, which is favourable compared to most of the hydrides discussed above (although also here, only a part of the hydrogen is available as a gas on the right-hand side of (2.65)).

Comments on the practical options for use of various types of storage in hydrogen energy carrier systems are given below in section 2.4.6 and, for particular implementations, in Chapter 4 and section 5.1.

2.4.6 Comparing storage options

The choice of a suitable hydrogen store depends on the application aimed

for and should compare options at a system level.

Figure 2.49 summarises the energy density by volume and mass for systems considered for use in fuel cell passenger cars, based on the data from the preceding subsections, supplemented with estimates of system package overheads (containment, safety features, controls) according to Herrmann and Meusinger (2003). No presently available system is satisfactory on a mass basis, and only liquid hydrogen stores approach viability on a volume basis. At the same time, the latter systems have undesirable traits related to boil-off losses.

Figure 2.49. Energy density for storage systems of a total capacity suitable for automotive applications. Densities are estimated for a total package solution in each case.

The US Department of Energy has set some performance goals for the US domestic industry engaged in development of hydrogen storage concepts for automotive use (Satyapal et al., 2007). For year 2015, a lifetime exceeding 1500 cycles is envisaged, and a 5 kg of hydrogen filling time of under 2.5 minutes. Maximum costs of storage is by 2015 set at US\$2 per kWh and the 2015 target-cost of hydrogen itself is US\$67 per kg. As pointed out by Hua et al. (2011), the compressed hydrogen tanks currently incorporated in hydrogen prototype vehicles are not likely to be able to meet all the requirements, neither at 35 nor at 70 MPa.

Assessments similar to the above may be made for other applications, notably stationary uses, where the requirements in most cases are less stringent. An alternative tool in evaluation of system performance is a plot of energy density versus power density. Some of the systems under consideration (e.g., many of the metal hydride stores) have filling and delivery times leading to power densities too low for many applications.

2.5 Hydrogen transmission

2.5.1 Container transport

Hydrogen can be transported in containers of compressed or liquid hydrogen, as well as emerging storage forms in hydrides or other chemical substances, as described in section 2.4.

Intercontinental hydrogen transport, which may become required if conditions for hydrogen production are geographically distributed in a way different from that of demand, may make use of containers transported on ships, similar to those transporting liquefied natural gas today. Because the density of liquefied hydrogen is so much smaller than that of natural gas, the transportation costs are going to be higher. In addition, there are problems with the leakage from the containers (section 2.4.2), and safety in case of containment accidents from on-board causes, including those related to hydrogen loading and unloading or caused by ship collisions. Conceptual designs include spherical or cylindrical containers (Abe et al., 1998). Intercontinental transport of hydrogen by ship has been estimated to cost around US\$25/GJ or US\$3/kg (Padró and Putsche, 1999).

| Closed-loop system | Enthalpya | Temperature | | |-------------------------|--------------------|--------------|--| | | ∆H0
(kJ mol -1) | range
(K) | | | CH4 + H2O ↔ CO+ 3H2 | 206 (250)b | 700−1200 | | | ↔ 2CO+ 2H2
CH4 + CO2 | 247 | 700−1200 | | | CH4 + 2H2O ↔ CO2 + 4H2 | 165 | 500−700 | | | ↔ C6H6 + 3H2
C6H12 | 207 | 500−750 | | | ↔ C7H8 + 3H2
C7H14 | 213 | 450−700 | | | ↔ C10H8 + 5H2
C10H18 | 314 | 450−700 | |

Table 2.5. High-temperature, closed-loop chemical C-H-O reactions (Hanneman et al., 1974; Harth et al., 1981). a

Alternative materials for hydrogen storage during long-range transport are, as mentioned in section 2.4.5, methanol and higher hydrocarbons. The hightemperature reactions listed in Table 2.5 allow methane and other hydrocarbons to be transformed into producer gases with high hydrogen content,

Standard enthalpy for complete reaction.

Including heat of evaporation of water.

typically with more modest temperature requirements for higher hydrocarbons (such as the decalin mentioned in Eq. (2.65)).

The transport of hydrogen in the form of chemical compounds would seem to reduce losses as well as costs, relative to transport of liquid hydrogen or the in any case likely to be too bulky compressed gaseous hydrogen.

For shorter distances of transport−for example, from central stores to filling stations−all the forms of transport can in principle be contemplated. The conversion or liquefaction/deliquefaction processes may carry too high energy losses and costs, so that despite the bulkiness of compressed hydrogen it may be a more acceptable solution.

2.5.2 Pipeline transport

The alternative for hydrogen transport over short or medium distances is pipelines, with the viable distances being determined by ease and cost of establishing pipelines, usually over land but conceivably also in some cases off-shore. Again, the technology is basically already developed for natural gas transport, and only modest alterations are believed to be required in order to contain the smaller hydrogen molecules with acceptable leakage rates.

Current polymer pipes used for transportation of natural gas at pressures under 0.4 MPa have sufficient strength to carry hydrogen, and diffusion losses arising mainly from pipe connectors have been estimated at three times higher for hydrogen than for natural gas (Sørensen et al., 2001). Transmission of natural gas at pressures up to 8 MPa uses steel pipes with welded connections. Most metals can develop brittleness after absorption of hydrogen, especially at locations contaminated with foreign substances (dirt or H2S) (Zhang et al., 2003). Adding a little oxygen (about 10⁻⁵ by volume) inhibits these effects. Inspection programmes similar to those carried out today should suffice to make the risk comparable to that for natural gas transmission. Auxiliary components such as pressure regulators and gauges are not expected to cause problems, although lubricants used should be checked for hydrogen tolerance. Like for natural gas pipelines, spark ignition is a potential hazard. If hydrogen is distributed to individual buildings, gas detectors must be installed, and outdoor emergency venting devices should be considered. Hydrogen has previously been extensively used in town gas systems. Recent test installations in the United States and Germany have been investigated (Mohitpour et al., 2000). Hydrogen pipeline costs are estimated at about US\$625 000/km (Ogden, 1999). Leighty (2008) similarly estimates a cost of (2005-)US\$700 000/km for a 300 km pipeline and a cost estimate of US\$2-6 for piping 1 kg hydrogen by 320 to 1600 km, with the purpose of smoothing intermittent renewable energy production.

2.6 Problems and discussion topics

2.6.1

Try to estimate the amount of hydrogen storage required if all electricity demand in your region should be covered either from wind power alone or from photovoltaic power alone (produced at your latitude). Compare with the scenario results in section 5.4. How many days of full demand should the store be able to hold?

2.6.2

An estimate of the maximum amount of hydrogen that can be stored in a molecular lattice may be obtained by close packing of hydrogen atoms with a distance between H atoms determined from empirical values in, for example, metal hydrides. Assuming such a distance to be not less than 0.2 nm, show that the total amount of hydrogen than can be stored is about 250 kg m⁻³ .

For comparison, the distance between the hydrogen atoms in a hydrogen molecule is 0.074 nm.

2.6.3

Give an estimate of how much land (or water surface) area should be set aside for allowing worldwide needs for hydrogen as a transportation fuel to be derived by microbial photosynthesis. How much arable land would be used if the same amount of hydrogen were to be produced by fermentation. Discuss the possibility for combining the land (or ocean/waterway) use for hydrogen production with other uses.

2.6.4

Try to make a plot of energy density (by mass) versus estimated power density as suggested in section 2.4.6 for the energy storage systems discussed.

2.6.5

How much hydrogen should be stored in hydrogen fuel cell passenger vehicles in order to give them a typical driving range of 800 km between fillings (make assumptions on the size of the fuel cell and its efficiency; compare with the examples given later in Chapter 4, including those for hybrid vehicles).

FUEL CELLS

3.1 Basic concepts

3.1.1 Electrochemistry and thermodynamics of fuel cells

The electrochemical conversion of energy is a conversion of chemical energy to electrical energy or vice versa. An electrochemical cell is a device either converting chemical energy deposited within the device, or chemical energy supplied through channels from the outside to the cell, into electricity, or it is operating in reverse mode, converting electricity to either stored chemical energy or an outward-going mass flow of chemical energy. At all stages, associated heat may be generated or drawn from the surroundings. The general layout of such a device is shown in Fig. 3.1.

According to thermodynamics (consult textbooks, e.g., Callen, 1960), the maximum amount of chemical energy of the system that in a given situation can be converted into a high-quality energy form such as electricity is given by the free energy G, also called Gibbs energy,

$$G = U - T{ref} S + P{ref} V, \tag{3.1}$$

where U is the internal energy of the system (such as chemical energy in the cases considered here), S is its entropy, and V is its volume, while (Tref, Pref), the absolute temperature and pressure of the surroundings, defines what is meant by "in a given situation". The classical law of energy conservation (which is also the first law of thermodynamics) states that the increase in internal energy is given by the net energy added to the system from the outside:

$$\Delta U = \int dQ + \int dW + \int dM. \tag{3.2}$$

Here M is the net energy of material flowing into the device, W is the net amount of mechanical or electric work performed on the system by the surroundings, and Q is the net amount of heat received from the environment.

Figure 3.1. Schematic picture of an electrochemical device, showing mass (fuel), heat, and power exchange to or from the device (from B. Sørensen, Renewable Energy, 2004a, used by permission from Elsevier).

The amount of work performed by the device on its surroundings is −W. Integrated over a period of time, it may be expressed as

$$-\Delta W = -\Delta W_{elec} + \int P \, dV, \tag{3.3}$$

where the last term is zero if the system is held at a fixed volume. The electric work -∆Welec, which the electrochemical system can deliver to a device (such as an electric motor) attached to an external circuit, may be expressed in terms of the electric potential difference between the positive and negative electrodes (see following) and the number of electrons travelling,

$$-\Delta W_{elec} = n_e NA e \Delta \phi{ext} = ne \mathcal{F} \Delta \phi{ext}$$ (3.4)

where the electron charge is e = 1.6 × 10-19 C (and *e*∆φext thus the energy difference for a single electron), ne is the number of moles of electrons, NA = 6 × 1023 Avogadro's constant (the number of particles−here electrons−per mole) and F = NA e = 96 400 C mol-1 is Faraday's constant. The energy (3.4) can also be expressed in terms of internal properties of the electrochemical device, namely the chemical potential difference ∆^µ across the device, eventually made up by more components ∆µⁱ ,

$$-\Delta W{elec} = \sum{i} n{e,i} \, \Delta \mu{i}. \tag{3.5}$$

The chemical potential for an electrolyte is an expression of the additional energy of its ions relative to that of the pure solvent. With the electrolyte components labelled as above by i, their contributions may be expressed as

$$\mu_i = \mu_i^0 + \mathcal{R} T \log (f_i x_i), \tag{3.6}$$

where R = 8.3 J K-1 mol-1 is the gas constant, T is the temperature (K), and xi is the mole fraction of the particular component. fi is called the activity coefficient and may be regarded as an empirical constant. µⁱ ⁰ is the chemical potential of the *i*th component if it were the only one present, at the given pressure and temperature (see, e.g., Maron and Prutton, 1959).

In case the electrochemical device has a constant volume and the exchange of heat with the environment can be neglected, then the electricity produced must correspond to the loss (conversion) of free energy from the cell,

$$-\Delta G = -\Delta W_{elec}. (3.7)$$

The free energy G, as given by (3.1), is the maximum work that can be drawn from the system under conditions where the exchange of (any kind of) work is the only interaction between the system and its surroundings. A system of this kind is said to be in thermodynamic equilibrium if its free energy is zero. Technically speaking, the expression of the free energy given in (3.1) implies dividing the total system into two subsystems: a small one (the electrochemical device) with extensive variables (i.e., variables proportional to the size of the system) U, S, V, and so on, and a large one (the surroundings) with intensive variables Tref, Pref, and so on. The reason for introducing the large system is to be able to consider its intensive variables (but not its extensive variables Uref, Sref, etc.) as constants, regardless of the processes by which the entire system approaches equilibrium.

This implies that the intensive variables of the small system, which may not even be defined during the process, approach those of the large system when the combined system approaches thermodynamical equilibrium (G = 0). The maximum work is found by considering a reversible process between the initial state and the equilibrium state. It equals the difference between the initial internal energy, Uinit = U + Uref, and the final internal energy, Ueq, and hence takes the form given in (3.1).

The approach of the entire system towards equilibrium, likely involving internal, irreversible processes, gives rise to a dissipation of energy given by

$$D = -dG/dt = T_{ref} dS(t)/dt,$$ (3.8)

assuming that entropy is the only time-dependent variable. It may not be

possible to realise the approach to equilibrium in a finite span of time, and constraints such as confining the (small) system by walls to a finite volume may prevent reaching the true equilibrium state of zero free energy. If the extensive variables in such a constrained equilibrium state are denoted U⁰ *, S*0 , V⁰ , and so on, then the available free energy becomes modified to an expression of the form

$$\Delta G = (U - U^{0}) - T{ref}(S - S^{0}) + P{ref}(V - V^{0}), \tag{3.9}$$

where chemical reaction energy is assumed included in the internal energy.

When the small system is constrained by walls, then the free energy reduces to the Helmholtz potential U − TS, and if the small system is constrained so that it is incapable of exchanging heat, the free energy reduces to the enthalpy, H = U + PV. The corresponding forms of (3.9) may be used to find the maximum work that can be obtained from a thermodynamic system with the given constraints. The general difference between enthalpy and free energy is made explicit through the relations (see, e.g., Sørensen, 2010a)

$$dU = T dS - P dV,$$

$$dH = T dS + V dP.$$ (3.10)

It should be emphasised that thermodynamics is a theory for the behaviour of systems near their equilibrium state, obtained from statistical treatment of the Newtonian laws of mechanics. In other words, thermodynamics tries to establish simple laws for the time development of averages of certain quantities (like temperature defined through averages of squared particle velocities). As weather forecasters know, it is not generally possible to find any simple behaviour for averaged quantities, or differently stated, no simple theory has yet been found for averages of thermodynamic quantities far away from equilibrium. Probably such basic laws simply do not exist. Since the systems of actual interest in electrochemistry are always away from equilibrium and are involved in complex, irreversible reactions or developments, then it should be clear that things like the maximum thermodynamical efficiency of a device can only be used as a theoretical reference point for comparing measured efficiencies, and in many cases there are no device designs that could bring the actual efficiency close to the thermodynamical maximum.

A description of the dynamical behaviour of the system must then go beyond the consideration of thermodynamics and try to establish relations between the rates of flows in the system:

$$J_Q = dQ/dt$$ (heat flow rate),
$J_m = dm/dt$ (mass flow rate), (3.11)
$J_q = dq/dt = I$ (charge flow rate or electrical current),

and the factors that may influence them, that is, interactions (generalised forces) between the components of the system. From Fig. 3.1, the simple energy efficiency of an electrochemical device converting fuel and heat into electricity (thus in this case the electricity input is zero) is then at a given instant in time given by

$$\eta = \frac{J{Q,in} - J{Q,out} + Jm(w{in} - w{out})}{J{Q,in} + Jm w{in}},$$ (3.12)

where the specific energies win and wout of the ingoing and outgoing substances (fuels) serve to convert the mass flow rates into energy flow rates. The efficiency of providing work (also called the second law efficiency or exergy efficiency) is defined as

$$\eta^{2.law} = \frac{W}{\max(W)},\tag{3.13}$$

where W is the actual power output and max(W) the maximum work that might be produced, according to the theoretical understanding of the actual irreversible, not-close-to-equilibrium processes. The expression carefully avoids specifying which "theoretical understanding" one should use or if there exists a known, valid, and applicable theory at all. This warning is meant to emphasise that although we believe there are valid theories on the basic level (classical and quantum mechanics, electromagnetic theory, etc.), it is often not possible to pre-calculate the behaviour of an actual, macroscopic energy conversion system without having to introduce uncertainties associated with the approximate descriptions used to engineer systems too large, or having too many elementary constituents, to allow direct application of the fundamental theories available.

Electrochemical device definitions

Electrochemical devices are named according to the following conventions. A device that converts chemical energy supplied as input fuels to the device into electric energy is called a fuel cell. If the free energy-containing substance is stored within the device rather than flowing into the device, the name "primary battery" is used. A device that accomplishes the inverse conversion (e.g., electrolysis of water into hydrogen and oxygen) may be called a driven cell. The energy input for a driven cell need not be electricity, but could be solar radiation, for example, in which case the process would be photochemical rather than electrochemical. If the same device can be used for conversion in both directions (or if the free energy-containing substance is regenerated outside the cell with addition of energy and recycled through the cell), it may be called a regenerative or reversible fuel cell, and finally, if the free energy-containing substance is stored inside the device, it may be called a regenerative or secondary battery.

The basic ingredients of all these electrochemical devices are two electrodes (positively and negatively charged) and an intermediate electrolyte layer capable of transferring (most often) positive ions in either direction, while a corresponding flow of electrons in an external circuit provides the desired power or uses power to produce fuels. I use the name negative electrode for the electrode where negative charges build up as a result of the presence of excess electrons and the name positive electrode for the electrode where positive charges build up as a result of a deficit of electrons. In contrast to the historical names anode and cathode, these definitions ensure that the electrodes retain their names, no matter whether electrons are transferred from electrode to electrolyte or the other way (i.e., the fuel or power production modes of a fuel cell) ¤ . Use has been made of solid electrodes and fluid electrolytes, as well as fluid electrodes and solid electrolytes (e.g., conducting polymers). Before entering into the description of particular fuel cell types, a few general characteristics will be described with use of the theoretical concepts just introduced.

Fuel cells

Figure 3.2 shows the general layout of a fuel cell, based on the free energy change ∆G = —7.9 × 10-19 J for the reaction (left-to-right for fuel cell production of electricity, right-to-left for reverse fuel cell performing electrolysis):

$$2H_2 + O_2 \leftrightarrow 2H_2O - \Delta G,$$ (3.14)

(see Chapter 2, (2.16)). For electricity production, a hydrogen gas is led to the negative electrode, to which it may lose electrons and thereby form H+ ions capable of diffusing through the electrolyte (and membrane if present), while the electrons flow through the external circuit. The involved splitting of hydrogen molecules, H2, into protons and electrons is typically speeded up by the presence of a catalyst (typically a metal, which could be the electrode itself or, for example, a platinum film placed on the electrode surface). A reaction of the form

$$2H_2 \leftrightarrow 4H^+ + 4e^-$$ (3.15)

thus takes place at the negative electrode, where suitable catalyst choices may make it proceed at an enhanced rate (see, e.g., Bockris and Reddy, 1998; Bockris et al., 2000; Hamann et al., 1998).

Gaseous oxygen (or oxygen-containing air) is similarly led to the positive electrode, where a more complex reaction takes place, with the net result

Some authors used to define the cathode as the electrode, to which electrons in the outer circuit flow and become transferred to the electrolyte. This gives rise to confusion, as the electrodes will change names each time the current is reversed (battery charge/discharge or electricity/fuel production in the case of fuel cells).

Figure 3.2. Schematic picture of a hydrogen−oxygen fuel cell. The fuel inlets are drawn so that they can illustrate inputs on either side of the electrodes.

This reaction is likely built up by simpler reactions, which could involve oxygen first picking up electrons or first associating with a hydrogen ion. Like in biological material, the fuel cell may utilise membranes allowing H⁺ to diffuse through, but not H2. It is instructive to compare the way the reactions in (3.16) proceed in biological systems such as photosystem II (considered in section 2.1.5) and in fuel cells and batteries. The proposed mechanism for water splitting in photosystem II is one of sequential stripping of four protons, one at a time, as illustrated in Fig. 3.3a. For manmade electrochemical devices, the process is by conventional chemical (ionic) theory considered to consist of two steps, as illustrated in Fig. 3.3b:

$$4H_2O \leftrightarrow 4OH^- + 4H^+$$ and $4OH^- \leftrightarrow 2H_2O + O_2$ . (3.17)

The underlying thought is that the second proton is not, in the case of an electrochemical device, able to directly leave the OH⁻ radical, but will form new water molecules plus molecular oxygen in conjunction with a second, similar process (Fig. 3.3b, top). The combination of the two oxygen atoms to an oxygen molecule (or the opposite process in normal fuel cell operation) requires the presence of a catalyst, such as Ni or Pt. The first step (bottom to middle line in Fig. 3.3) is the same for organic and inorganic water splitting. The reason that the organic molecule in Fig. 3.3a (middle to upper step) can get away with direct O2 formation from just two OH⁻ radicals is, if this is the right explanation, that the molecules are held in place by the amino acid chains (capable of storing and releasing any supplementary energetic requirements), in conjunction with the action of a special cluster of four Mn atoms (see Sørensen, 2010a; Fig. 3.79b) acting as a Pt (or Pt compound) catalyst does in the fuel cell. Early suggestions that Mn existed in photosystem II as oxides (Hoganson and Babcock, 1997) have not been confirmed by the structural studies (Kamiya and Shen, 2003). Although not all details are revealed at the resolution achieved (0.37 nm), it is unlikely (but not impossible) that 8 oxygen atoms could have remained undetected near the 4 Mn atoms detected.

Figure 3.3. Reaction mechanisms for water splitting (upward arrows) or the reverse power-producing reaction (downward arrows), proposed for organic photosystem II (a) and for inorganic electrolysers and fuel cell systems (b). Charges are omitted in the figure, but are discussed in the text (Sørensen, 2010a).

Quantum mechanical calculations are made in the following section in order to shed light on the reaction pathways. They can help determine the distribution of charges over the entire system of a metal surface and the O and H atoms. In Figure 3.3, no charges are indicated. In classical electrochemistry, these are as given in (3.17), and the specific need for a catalyst is not very clear. Looking at the fuel cell case (Fig. 3.3b), the hydrogen atoms need to be positively charged in order to move from the negative to the positive electrode. At the negative electrode, they may transfer 4 electrons to the metal catalyst. This would explain why it has to be there. For electrolysis (left-toright reaction in (3.17)), energy has to be added by applying a voltage across the external circuit between the electrodes. This means having a surplus of electrons migrate through the external circuit to the electrode surface on the hydrogen-producing side of the membrane and away from the electrode on the oxygen side. The difference from the classical explanation is that the OH molecules may not be in the form of simple ions. The charge distribution may be more complex, which turns out to be an important observation for carrying out the quantum mechanical calculation considered in section 3.1.2.

This type of explanation means that in the left-side case, the intermediate situation in Fig. 3.3a is one where there may still be 2 charges on the metal electrode, while in both cases, the OH entities may have no overall charge. In the upper situation in both Figs. 3.3a and Fig. 3.3b, there must be 4 positive charges on the hydrogen atoms, in other words, they are 4 protons without electrons attached, ready to transfer to the negative electrode in the electrolysis case.

Returning to the thermodynamical level of description, the drop in free energy (3.7) is usually considered to be associated with the negative electrode reaction (3.15), and ∆G may be expressed in terms of the chemical potential (3.5) of the H+ ions dissolved in the electrolyte. Writing the single chemical potential µ involved as Faraday's constant times a suitable potential φ, the free energy for n moles of hydrogen ions is (see (3.4) and (3.5))

$$G(H^{+}) = n\mu = n\mathcal{F}\phi = nN_{A}e\phi. \tag{3.18}$$

When the hydrogen ions "disappear" as a result of the reaction at the positive electrode ((3.17) in the direction towards the left), this chemical free energy is converted into the electrical energy by (3.7), and since the numbers of electrons and hydrogen ions in (3.16) are equal, n = ne, the chemical potential µis given by

$$\mu = \mathscr{F}\phi = \mathscr{F}\Delta\phi_{\text{ext}}.\tag{3.19}$$

The quantity φ is usually referred to as the electromotive force (e.m.f.) of the cell, or the "standard reversible potential" of the cell, if taken at standard atmospheric pressure and temperature. From the value of ∆G just quoted in connection with (3.14) for two molecules of water, the corresponding value of −2.37 × 10⁵ J is obtained for producing (or splitting) a mole of H2O. The cell e.m.f. φ then becomes

$$\phi = -\Delta G / n \mathcal{F} = 1.23 \text{ V}, \tag{3.20}$$

with n = 2, since there are two H⁺ ions for each molecule of H2O formed. The chemical potential (3.19) may be written in the form (3.6), and the cell e.m.f. thus expressed in terms of the properties of the reactants and the electrolyte (including the empirical activity coefficients appearing in (3.6) as a result of generalising the expression obtained from the definition of the free energy, (3.1), assuming P, V, and T to be related by the ideal gas law, PV = R T, valid for one mole of an ideal gas (see, e.g., Angrist, 1976)).

The efficiency of a fuel cell is the ratio between the electrical power output (3.4) and the total energy lost from the fuel. However, it is possible to exchange heat with the surroundings, and the energy lost from the fuel may thus be different from ∆G. For an ideal (reversible) process, the heat added to the system is

$$\Delta Q = T \Delta S = \Delta H - \Delta G, \tag{3.21}$$

and the efficiency of the ideal process thus becomes

$$\eta^{ideal} = -\Delta G / (-\Delta G - \Delta Q) = \Delta G / \Delta H. \tag{3.22}$$

For the hydrogen−oxygen fuel cell just considered, the enthalpy change during the two processes (3.15) and (3.16) (in the direction left to right) is ∆H = −9.5 × 10-19 J or —2.86 × 10⁵ J per mole of H2O formed, and the ideal efficiency of power production in this case is thus

$$\eta^{ideal} = 0.83.$$

There are reactions with positive entropy change, such as 2C+O2 → 2CO, which may be used to cool the surroundings and at the same time create electric power with an ideal efficiency above one (1.24 for CO formation).

In actual fuel cells, a number of factors tend to diminish the power output. They have traditionally been expressed in terms of "expenditure" of cell potential fractions on processes not contributing to the external potential,

$$\Delta \phi_{ext} = \phi - \phi_1 - \phi_2 - \phi_3 - \dots, \tag{3.23}$$

where each of the terms −φ ⁱ corresponds to a specific loss mechanism. Examples of loss mechanisms are blocking of pores in the porous electrodes, for example, by piling up of the water formed at the positive electrode in connection with the processes (3.17), internal electric resistance of the cell (heat loss), and the building up of potential barriers at or near the electrolyte-electrode interfaces, such as due to impurities. Most of these mechanisms limit the reaction rates and thus tend to place a limit on the current of ions that may flow through the cell. There will be a limiting current, IL, beyond which it will not be possible to draw any more ions through the electrolyte, because of either the finite diffusion constant in the electrolyte (in case the ion transport is dominated by diffusion) or the finite effective surface of the electrodes at which the ions are formed. Figure 3.4 illustrates the change in ∆φext as a function of current,

$$I = \Delta \phi{ext} R{ext} = I{-} + I{+} = (\phi{-} - \phi{+}) R_{ext}, \tag{3.24}$$

expressed as the difference between potential functions at each of the electrodes, ∆φext = φ⁻ − φ+ = φ*c− φ^a* (the latter notation used in Fig. 3.4). This representation gives some indication of whether the loss mechanisms are connected with processes at one or the other electrode, and it is seen that the largest fraction of the losses is connected with the more complex positive electrode reactions in this example. For other types of fuel cells it may be negative ions that travel through the electrolyte, with corresponding changes in characteristics (see Jensen and Sørensen, 1984).

Equation (3.24) also expresses the total current as a sum of contributions from each of the electrode areas. They can be related to the potential loss terms in (3.23), which describe the loss terms at each electrode, as well as Ohmic loss terms for each of the components of the cell. At each electrode these involve obstacles to transferring electrons from electrode to electrolyte (especially at low current) and at higher currents the obstacles to diffusing charges into the electrolyte from their creation near the electrodes, or to maintain sufficient amounts of reactants in the electrode area. The exponential dependence of electrode currents on these loss potentials is described by the Butler-Volmer equations (Hamann et al., 1998; Bockris et al., 2000),

$$\begin{split} I{-} &= I{-}^{0} \left( \frac{C{-}}{C{-}^{0}} \right)^{\gamma{-}} \left[ \exp \left( \frac{\alpha{-} \mathcal{F}}{\mathcal{R} T} \phi{+} \right) - \exp \left( \frac{\alpha{+} \mathcal{F}}{\mathcal{R} T} \phi{-} \right) \right], \ I{+} &= I{+}^{0} \left( \frac{C{+}}{C{+}^{0}} \right)^{\gamma{-}} \left[ \exp \left( \frac{\alpha{+} \mathcal{F}}{\mathcal{R} T} \phi{-} \right) - \exp \left( \frac{\alpha{-} \mathcal{F}}{\mathcal{R} T} \phi{+} \right) \right]. \end{split} \tag{3.25}$$

The parameters ^α⁻ and ^α+ are charge transfer coefficients for each of the electrodes, C⁻ and C+ are the concentrations of the reactants (such as hydrogen and oxygen) near the electrode, while C0 and C0 ⁺ are the corresponding concentrations in the bulk electrolyte. The exponents γ⁻ and ^γ+ are empirical quantities taken as 0.5 for oxygen and 0.25 for hydrogen (Nguyen et al., 2004). Finally, I 0 and I⁰ + are reference exchange currents for each electrode. They are the currents corresponding to neglecting the loss contributions to the potentials, which is the Nernst potential depending on the activities of the reversible electrode reactions without losses (such as (3.20) for the combined oxygen and hydrogen reactions). Further investigations of catalyst layer behaviour departing from (3.25) may be found in Kulinovsky (2010).

The Butler-Volmer equations simplify in the case where one of the exponen-

tial terms dominate, which will happen if the loss potential in question is large. In this case the electrode potential is linearly related to the logarithm of the corresponding current, and if the losses occur equally at both electrodes, then the total potential becomes linearly related to log(I). This is called the Tafel relation, and the slope of the line is called the "Tafel slope". The following sections will give many examples of potential-current relationships, and except for the interval of smallest currents, the Tafel approximation is often a valid one.

Figure 3.4. Fuel cell negative electrode potential φa and positive electrode potential φc, as a function of current. The main cause of the diminishing potential difference ∆φext for increasing current is at first incomplete electrocatalysis at the electrodes, for larger currents also ohmic losses in the electrolyte solution, and finally a lack of ion transport (see Bockris and Shrinivasan, 1969). From B. Sørensen, Renewable Energy, 2004a, used by permission from Elsevier.

It follows from diagrams of the type shown in Fig. 3.4 and from the equations (3.25), that there will be an optimum current, usually lower than IL, for which the power output will be maximum,

$$\max(E) = I^{opt} \, \Delta \phi_{ext}^{opt}. \tag{3.26}$$

Dividing by the rate at which fuel energy ∆H is added to the system in a steady state situation maintaining the current I opt, one obtains the actual efficiency for the conversion maximum,

$$\max(\eta) = I^{opt} \Delta \phi_{ext}^{opt} / (dH/dt). \tag{3.27}$$

The potential losses within the cell are fundamentally similar to the thermodynamical energy dissipation described in the beginning of this section, implying that energy cannot be extracted in a finite time without losses.

3.1.2 Modelling aspects

Quantum mechanical modelling can be used to describe processes involving individual molecules. Employing such models is also possible for describing surface processes, such as the adsorption and chemisorption at electrode surfaces and the action of catalysts on individual molecules, as well as describing reactions taking place between molecules moving within a solvent such as an electrolyte (e.g., Redox couples). The theoretical framework used in quantum chemical modelling is described in section 3.1.3.

Gas flows in channels leading to the electrodes, including the gas diffusion layers, are modelled, for example, by finite element models of flows such as those described in section 3.1.5. Such models may also, with suitable modifications, be applied to the ion flow through membranes.

Based upon the results of detailed models of these two types, overall models for the performance of fuel cells may be formulated in terms of simple equivalent electric circuit models that parametrise the loss terms and allow calculations of overall efficiencies as a function of such parameters.

The electrical circuit analogues have been used in connection with impedance spectroscopic measurements. These are measurements of the response of a total cell upon application of an alternating current across its outer electrodes and thus are aimed at the type of characterisation possible without dismantling the device for experiments on individual constituents (which, however, can also be done, e.g., for a half cell, as in Fig. 3.5). The response to the application of an external time-dependent potential of the form V=V0 cos(ωt) is a measurable current that may be out of phase with the applied voltage variation. In circuit theory, the phase differences are modelled by an equivalent description in terms of complex functions, which allows the use of imaginary numbers for quantities 90° out of phase with the potential. In general, the cosine is replaced by exp(iωt) = cos(ωt) + i sin(ωt), where i = (−1)-1/2, and the measurements will then determine a complex impedance Z, which differs from the Ohmic resistance by an imaginary part describing a phase delay between voltage and current. The prototype causes for this behaviour are capacitors and coils, but other components, for example, with a dependence on the frequency ω, may be invoked. Writing Z = Re(Z) + i Im(Z), the outcome of an experiment may be expressed in a diagram such as the one for a proton exchange membrane (PEM) fuel cell shown in Fig. 3.5, with Re(Z) as the abscissa and Im(Z) as the ordinate (Ciureanu et al., 2003). The straight line at smaller frequencies may be interpreted in terms of diffusion of H⁺ through the membrane, while the curve at high frequencies can be modelled by bulk capacitance and Ohmic resistance of the cell.

In Fig. 3.6, impedance spectroscopy has been performed on the negative electrode-side half cell of a direct methanol fuel cell (Müller et al., 1999), under conditions of (a) limited methanol supply (about two times the supply rate required by stoichiometry) and (b) generous methanol supply. The four parameters of the equivalent circuit proposed for case (b) may be simply fitted, or alternatively they may by modelling be associated with physical properties of the cell (Harrington and Conway, 1987). A similar behaviour is

found for proton exchange membrane half cells (Ciureanu et al., 2003). The behaviour in Fig. 3.6b can be explained in terms of the rates of electrons received from the negative electrode, the net rate of CO adsorption on the electrode, and the fractional coverage of the electrode by CO. The slow relaxation of adsorbed CO makes modelling in terms of an inductance L do well in reproducing the phase delay. Figure 3.6a exhibits three arcs. The one for low frequencies varies with methanol flow rate and thus may have to do with the probability of the methanol in reaching the active sites; the middle arc is due to methanol oxidation kinetics; and the arc at high frequencies, which is independent of electrode potential, represents an Ohmic loss that is insignificant at lower frequencies (Mueller and Urban, 1998).

Figure 3.5. Overall impedance response of a proton exchange membrane (PEM) fuel cell for different cell temperatures, depicted as corresponding values of the real and imaginary parts of the complex impedance (sometimes denoted a Nyquist plot). Each sequence of points represents frequencies ranging from 10-1 to 107 Hz; the highest values corresponding to the leftmost points. From M. Ciureanu, S. Mikhailenko, S. Kaliaguine (2003). PEM fuel cells as membrane reactors: kinetic analysis by impedance spectroscopy. Catalysis Today 82, 195-206, with permission from Elsevier.

The negative electrode reaction (3.15) involves the processes depicted in Chapter 2, Fig. 2.5: adsorption of a hydrogen molecule at d = 0.74 nm followed by splitting into two chemisorbed H atoms. The mechanism for these to part from the surface and move across to the opposite electrode involves for each hydrogen atom the delivery of an electron to the metal surface. The actual travelling need not be performed by H⁺ ions or by any other ions (say, of an electrolyte) in a bulk fashion, but may involve short-distance hopping from molecule to molecule, as suggested by Hamann et al. (1998) and illustrated in Fig. 3.7. The water molecule lends itself perfectly to this process. Due to its asymmetry (and hence dipole moment), it may easily accept the approach of a third hydrogen atom and form H3O+ as shown in the Fig. 3.7. A proton on the other side of the molecule may then leave and move the charge to a neighbouring molecule, which is now the H3O+ , and so on.

Figure 3.6. Impedance spectra for negative electrode half of direct methanol fuel cell. (a) Methanol mass flow limited situation (see text for explanation of the three arcs). (b) Situation for generous mass flow, with impedance behaviour shown for different current densities. (c) Equivalent circuit model for the situation in (b). Based on J. Mueller and P. Urban (1998). Characterization of direct methanol fuel cells by ac impedance spectroscopy, J. Power Sources 75, 139-143; J. Müller, P. Urban, W. Hölderich (1999). Impedance studies on direct methanol fuel cell anodes. J. Power Sources 84, 157-160. Used by permission from Elsevier.

Figure 3.7. Transfer of a proton through water: the proton attaches to the "unoccupied" side of a water molecule, the electron density rearranges, and another proton may leave the other side of the molecule and move to another water molecule. In this way, any bulk motion of single H⁺ atoms through the water is avoided. A potential slope is assumed to have been created by a voltage applied to the electrodes, or the displacement of the protons will create such a potential, depending on whether the electrochemical device is operated as an electrolyser or as a fuel cell.

There have been considerable amounts of theorising about the behaviour of ions in a solution such as an electrolyte. The layer of oppositely charged ions near a charged electrode is well established, but the possible extension of this layer into an ionic cloud diffusing far out into the electrolyte, as suggested by Gouy and Chapman (see Bockris et al., 2000), is less clear. The model calculation presented in Chapter 2, Fig. 2.4 only exhibits very minor variations in charge away from the electrode, and this is probably what one has to expect due to the falling strength of the Coulomb force with distance, in a model where the electrode charge is supposed to be responsible for the rearrangement of ions in the electrolyte. Within the electrolyte, only minor variations in charge are expected, due to charge diffusion by "hopping" mechanisms such as the one depicted in Fig. 3.7.

3.1.3 Quantum chemistry approaches

The many-body problem of describing molecules in terms of their nuclear and electronic constituents requires some level of simplification of the general quantum mechanical approach in order to be numerically tractable. The first approximation made is usually the Born-Oppenheimer assumption, that only the dynamic behaviour of electrons has to be modelled. The atomic nuclei are much heavier and thus move less rapidly, implying that one may look at the electron motion for each configuration of the atomic centres (called the molecular structure). One can then repeat the calculation for different structures, in case one is interested in finding one or more equilibrium configurations of a molecular structure, and hope to identify minima in the total energy calculated, as a function of structure parameters. Often there will be more than one minimum, e.g., indicating isomerism in ring structures.

The Schrödinger equation for describing a molecular system has the form

$$H \Psi = i\hbar D_t \Psi, \tag{3.28}$$

where ' is Planck's constant over 2π (' = 1.05 × 10-34 J), Dt is the partial derivative, here with respect to time, Ψ is the wave function of the system, and H is the Hamiltonian, an energy operator which in the molecular case may be written

$$\begin{split} H &= T + V &= -\frac{1}{2}\hbar^2 \sum_{i \in {e,N}} m_i^{-1} (D_x^2 + D_y^2 + D_z^2) \ &+ e^2 (4\pi\varepsilon0)^{-1} \Biggl( \sum{i < j \in {e}} \left| r_i - rj \right|^{-1} + \sum{i < j \in {N}} Z_i Z_j \left| R_i - Rj \right|^{-1} - \sum{i \in {e}, j \in {N}} Z_j \left| r_i - R_j \right|^{-1} \Biggr) \end{split}$$ (3.29)

in terms of kinetic energy T (first line), potential energy V, with mi the mass of an electron or of a particular atomic nucleus, depending on whether i belongs to the set of electrons in the system, {e}, or to the set of atomic nuclei, {N} (each consisting of a number of protons and neutrons not modelled here), and in terms of the Coulomb interaction between all pairs of the charged particles, whether negatively charged electrons or positively charged atomic nuclei. The radius vector is denoted ri for the *i*th electron and Rj for the *j*th nucleus. The elementary charge is denoted e (= 1.60 × 10⁻¹⁹ C) and the dielectric constant in vacuum ^ε⁰ (= 8.85 × 10⁻12 C2 m-1 J⁻¹ ). Thus *e /(4*πε0 ) = 2.30 × 10⁻28 J m. Although no special notation is used, all radius vectors ri or Ri are indeed vectors, with three co-ordinates such as x, y, and z.

The wavefunction Ψ = Ψ (ri ^∈{e}, Rj ^∈{N},t) is assumed normalised for any fixed time t by the condition

$$\int_{all\,space} \Psi^* \Psi \, dr1 ... dr{n_e} dR1 ... dR{n_N} = 1 \,, \tag{3.30}$$

and the density (a function of all space variables as well as time) is given by

$$\rho(r_i \in {e}, R_j \in {N}, t) = \Psi^* \Psi = |\Psi|^2. \tag{3.31}$$

^Ψ * is the complex conjugate of ^Ψ (that is, ^Ψ * = a − ib if Ψ = a + ib). Note that by this definition, ρ is normalised to unity and not, for example, to the total number of particles, an equally valid alternative, discussed following. The interpretation of this density is that it describes the probability of finding the system with its particles at the specified space vectors {ri ,Rj } at a given time t.

That is the very beauty of quantum mechanics. It is a perfectly deterministic theory for calculating the development of a physical system, once you know it at a given time. But because only the absolute square of the wavefunction is accessible to experiment, it is not possible for us to know the state precisely at such an initial time, and hence all the further development of the system is to the human observer clouded with uncertainty relations and similar indeterminacy, despite the fact that the system itself knows perfectly well what it is doing and follows a uniquely given path of development in time and space.

The solutions to (3.28) have the following general properties: there are a countable number of bound, stationary state solutions, ψq, q = 1,2,3, …, with energies Eq. This is the meaning of calling the spectrum of solutions quantised and the theory quantum mechanics. Additionally, there may exist unbound state solutions that may form a continuum in energy space. The set of quantised solutions are labelled by q. The collection of values that q may have is called quantum numbers. Although time dependence is sometimes important (e.g., for chemical reactions), many problems can be reduced to finding the stationary states, and finding these is also a good starting point, if time dependence has to be subsequently studied. The form of the wavefunction for a particular stationary case q is

$$\Psi{q}(r{i} \in {e}, R{j} \in {N}, t) = \psi{q}(r{i} \in {e}, R{j} \in {N}) \exp(-iE_{q}t/\hbar). \tag{3.32}$$

Since the H in (3.28) does not have an explicit time dependence, this equation reduces to what is called the stationary Schrödinger equation,

$$H\psi_q = E_q \psi_q. \tag{3.33}$$

The Born-Oppenheimer approximation can now be stated as follows: treat all the variables of the atomic nuclei, Rj , as constants and forget about the nuclear part of the wavefunction. More precisely, one may assume that the wavefunction can be written in the form

$$\psi_q(r_i \in {e}, R_j \in {N}) = \varphi_q(r_i \in {e}) \Phi_q(R_j \in {N}),$$ (3.34)

which allows the Schrödinger equation to be written, with obvious notation for the electronic and nuclear parts of the kinetic and potential energy operators in (3.29),

$$(T_e + T_N + V_e + VN + V{eN}) \varphi_q \Phi_q = E_q \varphi_q \Phi_q.$$ (3.35)

Although the electron wavefunction ϕ*q(ri ; i∈{e})* is primarily a function of the electron co-ordinates, it indirectly depends also on the nuclear co-ordinates because they appear in the potential terms VN + VeN. The Born-Oppenheimer approximation now neglects the influence of nuclear motion, that is, the kinetic term with TN, on the solutions ϕq for the electron part of the system, by solving for the latter by the following equation:

$$(T_e + V_e + VN + V{eN}) \varphiq = E{q,eff} \varphi_q. \tag{3.36}$$

All that is done is to leave out the action of the operator TN and, as a consequence, get a modified, effective energy Eq,eff. The rest of the energy, Eq,rest = Eq ⁻ Eq,eff, may subsequently be found by solving the remaining part of (3.35), i.e., what was left out in (3.36). If the indirect dependence of ϕq on the Rj s is disregarded, what is left is only

$$(TN + E{q,eff}) \Phi_q = E_q \Phi_q \text{ or } T_N \Phiq = E{q,rest} \Phi_q.$$ (3.37)

The formation of molecules in a model with only Coulomb forces does of course rely strongly on the presence of both positive and negative particles, but the approximations made here have shifted all the non-trivial calculations to the electron treatment in (3.36). Because all electrons are fermion identical particles, quantum theory makes additional formal requirements of antisymmetry on their wavefunctions, to be discussed in the following.

Hartree-Fock approximation

The splitting of a wavefunction into product form, as in (3.34), is a widely used method for seeking computationally simple solutions to the Schrödinger equation. The Hartree (1928) approximation consists in writing the overall wavefunction ϕq as a product of wavefunctions depending only on the co-ordinate of one electron,

$$\varphiq = \prod{i \in e} \psi_{q,i}(r_i). \tag{3.38}$$

This wavefunction does not satisfy the antisymmetry property required for electrons, and it does not include the spin (s = ½ for electrons, with projection ms = −½ or +½ on the quantisation axis). To remedy these shortcomings, the wavefunction should rather be of the form

$$\varphiq = (n!)^{-1/2} \det \left( \prod{i \in e} \psi_{q,i}(ri) \chi{q,i} \right),$$ (3.39)

where the product over i goes over all electrons e of the system, and the determinant ensures that the interchange of any two electrons will give an overall minus sign on the wavefunction (this is a standard property of determinants). The spin functions are denoted χq,i. They must equal one of the two "universal" spin eigenfunctions previous mentioned as possible for electrons. They are universal in the sense that there is only one spin-up (ms = ½) and only one spin-down (ms = −½) spin function and no variables known to describe them further (except in superstring theory and the like). The "guess" (3.39) is now ready to be introduced into (3.36) for the solution.

A well-known theorem in quantum mechanics states that the expectation value of the Hamiltonian, taken in the state defined by any wavefunction, is minimum when the wavefunction equals the exact ground state wavefunction. Starting from a trial wavefunction (that most likely is different from the exact ground state wavefunction) involving some parameters, one can then use a variational approach to find the best wavefunction within the space spanned by the trial wavefunctions considered. The energy average to be varied is (from here on dropping the index q labelling the stationary energies)

$$\langle E_{eff} \rangle = \frac{\int \varphi^* (T_e + V_e + VN + V{eN}) \varphi \ d\tau_1 ... d\tau_e}{\int \varphi^* \varphi \ d\tau_1 ... d\tau_e}, \tag{3.40}$$

where ^τⁱ is a shorthand for the space and spin variables of the *i*th particle, and where in contrast to (3.36), the division by the normalising integral has to be there because variations of ϕ may not preserve normalisation. This energy has a minimum,

$$\delta \langle E_{eff} \rangle / \delta \varphi = 0, \tag{3.41}$$

where it is closest to (but not smaller than) the exact ground state energy, and the corresponding function ϕ of space and spin variables is the corresponding ground state wavefunction, in the form implied by the approximation used. Performing the variation (3.41) with a wavefunction of the form (3.39) leads to e equations of the form (see, e.g., Scharff, 1969):

$$(H_k + U^{(k)}(r_k)) \varphi_k = E_k \varphi_k; \quad k = 1, ..., e,$$ (3.42)

where Hk is the part of the Hamiltonian in (3.36) that depends only on the *k*th electron (using following a suffix ",k" to indicate that only the *k*th term in the sum over electrons in (3.29) should be included),

$$Hk = T{e,k} + VN - V{eN,k} (3.43)$$

and

$$U^{(k)}(r_k) = \frac{\int \varphi^{(k)} * V_e \; \varphi^{(k)} \; d\tau1 ... d\tau{k-1} d\tau_{k+1} ... d\tau_e}{\int \varphi^{(k)} * \varphi^{(k)} \; d\tau1 ... d\tau{k-1} d\tau_{k+1} ... d\tau_e},$$ (3.44)

where

$$\varphi^{(k)} = (n!)^{-1/2} \det \left( \prod_{i \in e, i \neq k} \psi_i(r_i) \chi_i \right)$$ (3.45)

as distinguished from ^ϕ k = ψk(rk) χk. These equations constitute the Hartree-Fock approximation (Fock, 1930; in Heisenberg form: Foresman et al., 1992).

The physical significance of this approximation is clear from the form of (3.42): each electron is supposed to move in a potential, which in addition to the atomic nuclei consists of the average over the Coulomb interactions with all electrons other than the one focussed upon. The construction ensures that the Pauli principle demanding antisymmetry for the total electron wavefunction is satisfied in regard to other electrons with the same spin direction ms as the one focussed upon. However, it is not satisfied with respect to other electrons with opposite spin direction. This can be a serious restriction in the appropriateness of the Hartree-Fock approximation. Solving for the best description of a single electron moving in the average potential of all the other particles of the system (nuclei and e − 1 electrons) will clearly not address the issue of antisymmetry, and only if the initial antisymmetric states are not mixed appreciably will the solution retain a reasonable amount of this property. Furthermore, constructing wavefunctions comprising several configurations of many-electron states characterised by ms equal to +½ or −½ will not ensure that the many spin-½ components couple correctly to one definite total spin (such as either S = 0 or S = 1 for two electrons), as they should in quantum mechanics.

Basis sets and molecular orbitals

Most numerical algorithms solve the Hartree-Fock equations by iterative methods, implying that a good initial wavefunction guess is useful, but also that calculations can be lengthy. To cater to these problems, it is customary to expand the solutions (seeked) in terms of known basis wavefunctions, implying that only the expansion coefficients need to be modified at each iteration, while all integrals in principle can be kept on file. Use of a basis set is an exact method, provided that the basis set is complete (i.e., sufficiently large to span the whole space of solutions). However, and again for reasons of computational rationality (read: computing time required), practical calculations do not use complete, infinite sets of basis functions, but truncate them and often quite dramatically−say, to just 2 to 6 basis functions per electron in the system modelled. It is then important to have "good" basis functions, with "good" meaning functions that approximate well in a truncation at low order, and functions applicable for a wide range of different molecules (rather than basis sets doing well only for selected problems). It is important to be able to go into uncharted territory and still have reasonable confidence in using the methods that worked well in problems studied previously. Such basis sets are usually taken from the electron wavefunctions of single atoms, which in many cases are known with high precision or even are exactly known in analytical form, such as the Laguerre polynomials and spherical harmonic functions entering the solution for the hydrogen atom.

The most widely used bases for molecular calculations are constructed with use of Gaussian functions, due to the ease with which such functions can be differentiated or integrated. A Gaussian is a real function of the form

$$g_{l,m,n,\alpha}(x,y,z) = A x^n y^m z^l e^{-\alpha(x^2+y^2+z^2)},$$ (3.46)

where A ensures that the integral of g² over all space is unity (normalisation) and x,y,z are co-ordinates relative to a particular atomic centre. The parameter α determines the width of the density distribution. The actual basis functions are constructed as linear combinations of the primitive Gaussian functions (3.46), bj = ∑*ⁱ ai,j gli,mi,ni,αⁱ* , and are fixed quantities for solving a given problem: one chooses which basis set to use and then starts calculating, seeking to solve the equations (3.42) and find solutions of the form (3.39) with

$$\psi{q,k} = \sum{j} c{q,kj} b{j,} \tag{3.47}$$

where the set of *ckj*s for the lowest energy state defines both the Hartree-Fock approximation to the actual ground state q = 0 of the system through (3.39) and the wavefunctions ψ0,k called molecular orbitals. Because the solutions ckj also appear in the integrand of (3.44), the solution must be found through iteration. The lowest e levels (some possibly degenerate, i.e., with same energy) will be filled by electrons in the ground state, but the Hartree-Fock solution through (3.42) gives a full set of orbital solutions comprising not only occupied orbitals, that is, those filled when electrons are added from the lowest one−but also the higher energy solutions corresponding to virtual orbitals that may be excited to form states other than the ground state.

Figure 3.8. Molecular orbitals for the oxygen atom, with indication of their quantum numbers (main, orbital angular momentum, and projection along axis of quantisation). Shown is the oxygen nucleus and the electron density ψ***ψ (where it has fallen to 0.0004; it is identical for each pair of two spin projections), but with two different shades used for positive and negative parts of the wavefunction. The calculation uses density functional theory (B3LYP) and a Gaussian basis of 9 functions formed out of 19 primitive Gaussian functions (see text for further discussion). The first four orbitals (on the left) are filled in the ground state, while the remaining ones are unoccupied.

Figures 3.8 and 3.9 show¤ the molecular orbitals for an oxygen atom and an oxygen molecule at a proper atomic distance (R = 0.12 nm). The atomic orbitals are good indications of the spatial shape of the Gaussian functions and may be compared with the Laguerre polynomial/spherical harmonics solutions often depicted in elementary chemistry textbooks. The molecular orbitals of Fig. 3.9, on the other hand, give some insight into the types of solutions involved in calculations for more complicated systems. The accuracy of the Hartree-Fock approximations for the O2 energy will be discussed in the next section. The SV basis used in the figures (Schaefer et al., 1992; 1994) contains 9 basis functions for O, formed out of 19 primitive Gaussian functions, and twice as many for O2. This is only about 2 basis functions per electron in the system, but using instead the EPR-III basis (Barone, 1996) with about 8 basis functions per electron does not change either the shapes shown or the energy of the ground state significantly.

¤ The calculations presented in these and the following figures are made using the software Gaussian (2003) and displayed by GaussView or WebLab ViewerLight.

Figure 3.9. Molecular orbitals (iso-density at 0.0004) for the diatomic oxygen molecule, calculated as in Fig. 3.8 using a Gaussian basis of 18 functions formed from 38 primitive Gaussian functions. The basis states numbered 1 to 8 are occupied in the O2 ground state.

Higher interactions and excited states: Møller-Plesset perturbation theory or density function phenomenological approach?

Perturbation theory is a standard method in quantum physics (Griffith, 1995). It consists in writing the Hamiltonian as H0 + λH1, where H0 may be the Hartree-Fock Hamiltonian and H1 the part containing the remaining correlations, and then expanding the overall wavefunction solution as a power series of λ. The introduction of λ is only for identifying the order of approximation: in the end, λ is put equal to unity, meaning that it is actually H1 that is assumed "small". The application of perturbation theory for calculating corrections to Hartree-Fock wavefunctions and energies arising from the interaction between electrons was first proposed by Møller and Plesset (1934).

As one progresses through second, third, fourth, or higher order corrections the calculations become increasingly lengthy, despite refinements (Schültz et al., 2004), and efforts have been made to find other methods. As it is seen from (3.44) and (3.45), the integrals in the Hartree-Fock approximations can be cast into a form depending only on matrix elements of Ve and of the unity matrix between two electron wavefunctions ϕk1*^ϕk2. Around 1930, it was recognised that the indistinguishability of electrons meant that instead of keeping track of the co-ordinates of each particle, one might introduce an overall density function, ρ(r), normalised to the actual numbers of particles and a complete set of basis states *bj (τ) = bj (r,σ)*, where τ as in (3.40) is a shorthand for r, the position vector of a point in space, and σ, a general spin-related coordinate, assumed discrete so that integrals over dr must be accompanied by sums over σ. In the Heisenberg matrix formalism, all properties of the system can therefore be related to the density matrix, as already noted by Dirac (1930) and used in Hartree-Fock related theories such as that of Møller and Plesset in 1934 for both stationary and time-dependent situations. The density matrix may be constructed in the space of the real wavefunctions of different states of the system, or it may be formed for an arbitrary complete basis such as the bj s considered in the previous subsection.

This suggests a method for introducing electron correlations based on the density matrix, which was taken up by Hohenberg and Kohn (1964) and Kohn and Sham (1965), but which had its roots in much earlier work (Thomas, 1927; Fermi, 1928; Slater, 1928, 1951, and several papers in between). The method is now called density functional theory (DFT).

The first observation of density functional theory is that there is a one-to-one correspondence between the ground state total electron density and the ground state energy, under the assumption that there is no degeneracy of states. In chemistry, in the past the interest was often confined to the ground state, and using a real density rather than the complex wavefunction ap-

pealed to many people working in chemistry¤ . The ground state energy for a given density is of the form (Kohn and Sham, 1965)

$$E_0 = \int (VN + V{eN}(\mathbf{r}))\rho(\mathbf{r})d\mathbf{r} + \int V_e(\mathbf{r}_1, \mathbf{r}_2)\rho(\mathbf{r}_1)\rho(\mathbf{r}_2)d\mathbf{r}_1d\mathbf{r}_2 + Te[\rho] + E{xc}\rho$$

$$\rho(\mathbf{r}) = \sum_{k} \varphi^* \varphik = \sum{k} \chi_k^* \chi_k \psi_k(\mathbf{r})^* \psi_k(\mathbf{r}) \qquad k = 1, ..., e.$$ (3.49)

Here, the square brackets in (3.48) denote a possibly complicated and indirect dependence on the density ρ. The second observation is that a minimum found by varying this energy with respect to the density will describe the ground state. Unfortunately, the terms describing antisymmetry and deviations from independent electron motion beyond those described by the previous term cannot be written in simple terms as a function of the density and therefore must be approximated and/or parametrised. Zero and first order gradients of the density are used, under the names of local spin density and generalised gradient approximation (Perdew et al., 1996, 1999). Effectively, these methods may be viewed as replacing the antisymmetrisation by a short-range repulsive potential, which in the local spin approximation is just a function of a radial co-ordinate.

Iterative solution of the Kohn-Sham equations is similar to that of the Hartree-Fock equations, and a number of popular methods use a parametrisation with both Hartree-Fock and density function bits, each with a phenomenological weight factor. The density functional bit may consist of the local spin density part plus a phenomenological parametrised part representing exchange and other correlations, such as the popular threeparameter B3LYP method (Becke, 1993; with use of Vosko et al., 1980; Lee et al., 1988). The ad hoc parametrisations are made such that on average they fit selected properties for a range of simple molecules, but without necessarily doing well for a particular molecule, even in the sample used for parameter choice. There is continued effort to find new approximations capable of addressing the known problems and preferably reducing the use of non-exact treatment of the basic Schrödinger equation (see, e.g., Adamo and Barone, 2002; Kudin et al., 2002; Kümmel and Perdew, 2003; Staroverov et al., 2004).

Treatment of excited states in Hartree-Fock or in a density functional-type approach is not possible using simple variational methods. Considering the time-dependent Schrödinger equation (3.28), one may instead describe those excited states that can be reached by a one-step transition from the ground

Of course, use of complex numbers is essential to quantum mechanics and necessary for explaining the limitations to observability. Indeed, the matrix elements in any "density functional theory" aimed at explaining more than just one state of the system must have this essential feature.

state, by adding a time-dependent potential and looking at the resulting density distributions of the states that can be formed. Giving the transition matrix a simple form, for example, by using a Taylor expansion−it can be inverted to yield a correspondence between the excited state densities and the potentials responsible for them (Runge and Gross, 1984). Switching on these potentials adiabatically allows the calculation of the energy dependence of the density function by solving a matrix diagonalisation eigenvalue problem of twice the physical dimension (Bauernschmitt and Ahlrich, 1996). This is called time-dependent Hartree-Fock or time-dependent density functional theory depending on the starting point.

Method	E1(O2) (eV)	E2(OO) (eV)	E3(2O) (eV)	E1−E2 (eV)	E1−E3 (eV)
HF	−4 064	—4 044	—4 059	—19.5	—4.4
MP2	—4 071	—4 401	—4 063	+319	—11.7
MP3	—4 071	+4 083	—4 064	—8 154	—6.8
MP4	—4 071	—163 400	—4 064	+159 279	—7.3
B3LYP	—4 084	—4 074	—4 076	—10.1	—8.1
PBE	—4 080	—4 073	—4 071	—7.8	—9.4
PBE1	—4 080	—4 069	—4 071	—11.1	—8.4

Table 3.1. Calculations of ground state energies for the oxygen molecule, E1(O2), for the oxygen atoms moved 2.0 nm apart, E2(O–O), and twice the energy of a single oxygen atom, E3(2O), plus O2 binding energies estimated as the difference between E1 and each of the two calculations for separated atoms. The value determined experimentally is —5.2 eV. All calculations use the SV basis set. HF = Hartree-Fock, MP = Møller-Plesset. For the remaining rows, see discussion in text.

As an example of the accuracy of the higher-order corrections to Hartree-Fock solutions obtained by perturbation methods and by parametrised density functional methods, Table 3.1 looks at the binding energy (or minus the atomisation energy) for an oxygen molecule, calculated in a number of different approximations. All calculations use the same basis set SV as Figs. 3.7 and 3.8 (Schaefer et al., 1992), and adding polarisation to the basis functions as in Schaefer et al. (1994) only increases the binding energy by some 10%. The binding energy is already too large in all calculations except HF, as the measured value is —5.2 eV (CRC, 1973). It is seen that the Møller-Plesset perturbation expansion is well behaved for the oxygen atom or molecule, but when the atoms of the molecule are pulled apart, the expansion breaks down. The pure Hartree-Fock also does poorly with the two atoms far apart, while the HF difference *E1*−E3 = −4.4 eV is closer to the measured value than any of the higher approximations, of which the MP3 comes next, but overshooting. The parametrised calculations are the B3LYP (Becke, 1993) mentioned above and two variants using PBE, based on Perdew et al. (1996), with PBE1 and PBE having different weightings between the exchange and correlation parts of the functional Exc in (3.48) (0.25:0.75 and 0.5:0.5).

The conclusion drawn from this small exercise is that the perturbational but otherwise exact methods can be dangerous, especially for large molecules with parts bound loosely together, where the expansion may diverge as it does in the O--O case studied in Table 3.1. The density functional methods seem computationally much more stable, but are not necessarily correct in predicting absolute energies. Yet relative energies may be all right, as the measured O2 bond distance of 0.12 nm is very well reproduced by the potential surface charting by most of the methods used in Table 3.1. This can be seen from Chapter 2, Fig. 2.6, using HF for the Ni surface and B3LYP for the oxygen molecule. The same is true for the bond distance of H2, shown in Chapter 2, Fig. 2.5, and generally, the structure determination obtained even with the simpler quantum chemical methods seems acceptable, at least in cases without multiple, close-lying minima in the potential surface. However, the problem in reproducing binding energies encountered with oxygen suggests that one should exercise much care in attributing too much confidence to the results of energy calculations, although the general features of excitation spectra, such as succession of low-lying states, would in many cases come out in fair agreement with data, for cases where such are available. In other words, the quantum chemical tools available at the moment give a reasonable qualitative picture, without being necessarily reliable at a quantitative level.

3.1.4 Application to water splitting or fuel cell performance at a metal surface

In Fig. 3.3, mechanisms were presented for water splitting in an electrolyte or the reverse reaction of electric power production in a fuel cell. Here I shall attempt to use quantum chemical calculations to assess if such proposed schemes are possible and in which way they need the assistance of a metal surface ("catalysis"). Is the metal a template or catalyst satisfying the older textbook definition of "not participating in the reaction", or does it play a much more essential role in the processes, as the calculations shown earlier in Chapter 2, Figs. 2.5 and 2.6 have already suggested?

The dissociation of a hydrogen molecule at a distance of just over 0.12 nm from the metal surface shown in Figs. 2.5 and 2.44 lead to two hydrogen atoms separated by around 0.12 nm, but rather insensitive to the precise distance. Similarly, an oxygen molecule would dissociate at a distance from the metal surface only slightly larger than for hydrogen (about 0.16 nm, reflecting the larger "radius" characterising the oxygen electron density distribution), and the distance between the two oxygen atoms is around 0.15 nm, again slightly larger than for hydrogen, as expected (Fig. 2.6).

The dissociation of molecular hydrogen is the main story of what happens at the negative electrode of a fuel cell. The "liberated" hydrogen atoms can move through the (liquid or solid membrane) electrolyte towards the positive electrode, by the mechanism described in Fig. 3.7, and the electrons will enter the electrode metal and take part in an external current, if the cell is connected to a load, or will just build up a potential across the cell, creating the conditions for the transfer of positive hydrogen ions. The expected outcome of quantum chemical calculations at the electrodes is sketched in Fig. 3.10, where the energy gain at the negative electrode (A) must be large enough to furnish the kinetic energy necessary for transport through the electrolyte membrane (B), and where the remaining energy gained at the positive electrode (C) makes the total capable of expelling the water generated from the cell and of enabling external work to be performed by the electric potential difference established between the electrodes.

Figure 3.10. Suggested potential energy behaviour during transfer of protons (H⁺ ) between fuel cell electrodes (Sørensen, 2007e).

At the positive electrode of a fuel cell, a more complex set of reactions must be at work in order for the hydrogen ions to join with oxygen and form water (as illustrated by the sequences in Fig. 3.3, read from top to bottom). Figure 3.3 does not indicate ionic charges, because part of the investigation is to find out at which stage molecules are ionised or electrons are captured to form neutral molecules. The electrons are taken from or donated to the electrode metal, which then becomes positively charged in the fuel cell case (or already is due to the external potential in the electrolyser case). It follows that the electrons likely react with ions quite near to the electrode surface. Figure 3.11 shows the distribution of electron density in a small two-layer chunk of nickel, in the highest occupied molecular orbital (HOMO) and for the lowest unoccupied orbital (LUMO). The latter shows large balloons of electron density above (negative density) and below (positive density) the Ni-surface, ready to interact with any charged particle entering the area.

In order to determine more precisely what is happening, the quantum calculations must be made in the presence of either water molecules or hydrogen atoms and OH-molecules capable of making use of the effective positive charge caused by the deficiency of electrons on the positive metal electrode surface. The types of calculations described in the previous subsection in principle may predict the distribution of charges over the entire system, but in case of divided systems such as the metal surface and the molecules in the electrolyte, this may not emerge correctly (due to the deficiencies of the approximate quantum mechanical treatment, which as illustrated by the ionisation energy calculations behind Table 3.1 are not doing too well for separated subsystems). Indeed, straight calculations using the density functional theory fail to describe the water formation and subsequent movement away from the metal surface in a fuel cell, and inspection of the resulting wave functions show that electrons are invariably moved from the metal surface to the protons, even when they are far from the surface. This failure to distribute charges correctly is due to poor initial guesses and to optimisation routines being unsuccessful due to multiple minima in the potential energy surface.

Figure 3.11. Representing the catalyst by the simple 2-layer Ni lattice shown on top, the electron density is calculated using B3LYP density functional theory with an SV basis set. Left: highest occupied molecular orbital. Right: lowest unoccupied orbital.

The solution used here to avoid such problems is to divide the system into parts, with the metal surface being one part and the hydrogen and oxygen atoms included in the calculation forming one or two other parts (Fig. 3.12). This type of calculation allows each part of the system to have a prescribed charge. It was initially proposed to allow lower approximations to be used for parts of the system allowing this, and it uses a prescription called ONIOM to sum the energy contributions from different levels of theory (Dapprich et al., 1999). The upper part of Fig. 3.12 shows the division of the quantum chemistry calculation into one for describing the transfer of four unit charges from the negative electrode to four protons, and a second calculation for the transfer of these four charges to the positive electrode, associated with the formation of charge-neutral water. In the lower part of Fig. 3.12, an external circuit is connected, allowing parts of or all the electrode charges to be carried (by electrons) through an external device. Figure 3.13 shows the IV-characteristics, as the external current reduces the charge on the positive electrode from +4 elementary charges (open circuit) to 0 charges (short-circuit external loop, i.e., a load with no resistance).

Figure 3.12 (left). Selection of fuel cell system parts to include in separate calculation steps (Sørensen, 2007e).

Figure 3.13 (below). Current-voltage (IV-) characteristics of fuel cell, with indication of the charges (C) on the positive electrode subsystem, suitable for use in the ONIOM calculations described in the text.

In simple cases, the quantum mechanical calculation using density functional theory can be used to optimise the molecular configuration, that is, to find the position of each atom in the molecular structure−that will lead to the minimum overall energy. The method is to use an iterative procedure, where each point calculation is followed by a change of atomic co-ordinates in the direction of steepest potential energy descent. This is illustrated in Fig. 3.14, for the case of two oxygen atoms over a Ni-surface (the potential surface of which is calculated directly in Fig. 2.6), by giving the total energy at each step and the root-mean-square gradient of the energy surface at each step of the configuration, characterised by distance between the oxygen atoms and their distance from the Ni surface. For more complex situations with several minima of various depths, such simple optimisation if not possible, and a reasonable approach is to calculate the potential energy surface at a carefully selected number of configurations for the molecular structure, capturing expected types of structural arrangements pertaining to the significant atomic constituents, as the total potential energy multi-dimensional surface is rarely within the reach of computational capabilities, because it

has three dimensions of variable position co-ordinates for each atom in the molecule. I now illustrate this for the positive fuel cell electrode, where initially, the catalyst surface and oxygen molecules are met by four protons from the electrolyte.

Figure 3.14. Course of optimisation calculation for two oxygen atoms over a Nisurface. The inserts show the position of the two O-atoms at the start of the optimisation and after 137 iterations. The upper part gives total energy, the lower part the root-mean-square energy gradient at a given step.

Consider the process (a) in Fig. 3.3, involving two water molecules or six atoms for which to vary the position co-ordinates over a metal surface. To limit the variations to a manageable set¤ , I first look at a case where the wa-

To draw a potential energy surface, you need, say, 10 values for each of the 6 × 3 position co-ordinates. In total that is 1018 point-energy calculations. Even with a fast computer at your disposal, assuming each point to require an hour of CPU time, you see that this is not practical. Thus, some restricted variation strategy must be found.

ter molecules are both in a plane at a distance of z above the metal surface, with the centre of mass above a low-density location in the surface layer of the metal lattice (Fig. 3.15). For reasons of display, it is convenient to illustrate variations of two parameters at a time. Assuming first, that the hydrogen atoms are at the water equilibrium distance d = 0.096 nm from the relevant oxygen atom, while the distance 2e between the oxygen atoms is varied, the potential energy surface as a function of z and e takes the form shown in Fig. 3.15. It is seen that with the four hydrogen atoms close, the two oxygen atoms cannot form a molecule at 2e ≈ 0.12 nm as in Chapter 2, Fig. 2.6, but they willingly separate further when they are at a distance of z ≈ 0.4 nm from the second Ni surface (like in the figures of Chapter 2, I use the lower of the two Nickel layers to measure distances from, purely out of convenience based on the way I set up the calculations). Nickel is a commonly used catalyst, in the same transition group as Pd and Pt.

Figure 3.15. Potential energy surface of two water molecules over a Ni-surface, as a function of height over the surface (z measured from the second layer of Ni atoms) and 2e, the distance between the oxygen atoms. The quantum chemical calculation of potential energies employs Hartree-Fock for the two layers of 24 Ni atoms and density functional theory (B3LYP) for the rest, with SV basis functions (see section 3.2). The position of the water molecules for one set of parameter values is shown in the insert. The zero of the energy scale is chosen arbitrarily (Sørensen, 2005a).

Figure 3.16 shows that once the oxygen atoms have moved away from each other (0.24 nm), the two water molecules can drift away from the Ni surface, but soon just be inertia, without gain or loss in energy. In Fig. 3.17, the dependence on the distance g from the hydrogen atoms to the oxygen atom of the water molecule to which they belong is explored. All four *g*s are still the same, and the H-O-H angles are kept at 104.4°. It is seen that if the hydrogen atoms start far away from the oxygen atoms, they will move towards them until they get below the distance (0.096 nm) characterising their equilibrium distance in a water molecule and get repelled if they try to move closer.

Figure 3.16. Detail of Fig. 3.15 showing the potential energy as a function of z, the height over the Ni surface, for a fixed distance between the Oatoms of 0.24 nm. The overall charge in Figs. 3.15-3.19 is taken as zero (Sørensen, 2005a).

Figure 3.17. Potential energy for the Ni plus two water molecule system as in Fig. 3.15, with the distance of the H atoms from the relevant O-atom varied from the value g = 0.096 nm valid for an isolated water molecule. The O-atoms are 2e = 0.24 nm apart, and the distance over the Ni surface is kept at the value z = 0.4 nm, leading to the minimum potential energy for all values of g (Sørensen, 2005a).

The next step is to explore the energy behaviour when each oxygen atom has the two H atoms at different distances from the oxygen atom. This is illustrated by repeating the calculation with one H atom in each water molecule staying at g = 0.096 nm, but letting the other ones increase their distance, which I shall denote h. Figure 3.18 shows the potential energy for h larger than 0.096 nm. If the height over the Ni surface is at the same time kept equal to the value z = 0.4 nm that led to a minimum potential energy in Figs. 3.15 and 3.16, then one hydrogen atom from each molecule will slide away for h larger than 0.15 nm, but approach the oxygen atoms if h is below that value. If the water molecules are closer to the Ni surface (z ≈ 0.32 nm), then one hydrogen atom from each water molecule could theoretically move away, but as the overall energies are higher, it will not do that. Instead, it will fall into the higher z minimum (for h > 0.14 nm, the energy is lower for z = 0.32 than for z = 0.4, and here the H atom will run away in any case).

Figure 3.18. From a second set of calculations for two water molecules over a nickel surface, now allowing one hydrogen atom in each water molecule to move, while the other is fixed at g = 0.096 nm. The dependence of the total potential energy on the distance h of the two H atoms allowed to move is shown for two heights over the Ni surface and for the lower height z for two distances between the oxygen atoms (the distance being 2e or 0.2 and 0.44 nm) (Sørensen, 2005a).

The implication of Fig. 3.18 is that capturing the second H atom to form a water molecule (by the process going from the middle to the lower line in the left-hand side of Fig. 3.3a) cannot proceed spontaneously. What about attaching the first H atom to an O atom (top-to-middle line in Fig. 3.3a)?

Figure 3.19 shows the results of the quantum chemical calculation for this situation (overall charge still 0). The potential energy is calculated as a function of the distance between the oxygen atoms (2e) and the H-O distance (g) for the single hydrogen atoms (the other remaining far away). It is seen that the two already dissociated oxygen atoms will willingly move further away from each other, while for all distances above 2e = 0.12 nm, single hydrogen atoms from a larger distance will move towards the O atom until they reach the equilibrium distance g = 0.096 nm characterising H-O distances in water molecules. In other words, getting the first H atom to join with an oxygen atom is energetically favoured (top-to-middle line in Fig. 3.3a).

Figure 3.19. Potential energy of two HO molecules over an Ni surface, as a function of the O–O distance 2e and the O–H distance g, calculated with the same methods as those used for Figs. 3.15-3.18 (Sørensen, 2005a).

To proceed to water formation something more is needed, such as another

two water molecules (Fig. 3.3b) or a catalyst, which could be the metal surface or a nearby organic molecule such as the tyr molecule in photosynthetic water splitting (see Sørensen, 2010a). The preceding calculations have been done for neutral molecules. It is now time to look in more detail at the distribution of charges, starting with the four positive charges brought along by the four H atoms from the negative electrode processes, after passage through the electrolyte.

Figure 3.20. Mulliken distribution of atomic charges, for DFT calculation similar to those of Figs. 3.15-3.19, for two water molecules and a piece of a Ni-surface.

In Fig. 3.20, the distribution of charges following a DFT point calculation of the same type as the ones just used are given for three cases with overall charges 4, 2, and 0 atomic units. Charge assignment to atoms is not a quantum mechanical observable for molecular systems and the electron charge is distributed over all space, in the same way as the electron density. The positive charge distribution inside an atomic nucleus is anyway assigned to the centre point in a molecular calculation. The Mulliken method (Mulliken, 1955) is just one out of several approximate ways of assigning charges to atoms−that is, to neighbouring particle centres.

Figure 3.21 shows the results of a series of calculations with an overall charge of +4 elementary charge units on the positive electrode system, using the experience of the preceding explorations (Figs. 3.15-3.19) and trying to identify the sequence of steps in the process of water formation.

The first one has the 4 charges on the H+ (protons) entering above the electrode surface area from the electrolyte membrane. One proton is then transferred to each O atom to form OH, while two charges remain at the distant H atoms. In a following step, these are drawn towards the Ni-surface, but even then, water is not formed. Instead, the hydrogen atoms retain their positive charge and the oxygen atoms their negative charge, even sitting close to each other at the Ni-surface. Reducing the total charge from 4 to 2 and then to zero further reduces the energy (right-hand side in Fig. 3.21), but the msurface-attached H- and O-atoms still retain their charges and no neutral water is formed. What is wrong is that the DFT calculation is unable to render a description of the dynamic processes responsible for moving the charges introduced by the four incoming protons through the correct sequence of steps. Actually, since only the electron structure is modelled, the process that should have been emerging would be one of taking electrons from the Ni-lattice and using it to bind the oxygen and hydrogen atoms together at a height over the surface, ending with formation of neutral water. Figure 3.21 indicates the model prediction of (Mulliken) charges for each step, and they are seen to be systematically incorrect: Starting with moving electrons high up above the Ni-surface to falsely give the 4 hydrogen atoms a total charge of 1 instead of 4, and subsequently, when the Ni-surface has lost 4 electrons as it should, then to incorrectly maintain a positive net charge sum on the surface atoms that should have formed neutral water molecules.

Figure 3.21. Early positive electrode calculation using the DFT PBEPBE-method, treating the entire system as one. It is seen that the charge distribution is highly unrealistic, particularly for the columns to the left, and consequently the energy differences cannot be trusted. This problem could conceivably be reduced by enlarging the Ni-lattice size (Sørensen, 2005a).

The problem with distribution of charges is further illustrated in Fig. 3.22, for three additional calculations with total charge +4 units and various positions of the four protons. In Fig. 3.22a, the predicted charge distribution even puts an excess negative charge on two of the hydrogen atoms, while in Fig. 3.22c, the charge of the surface atoms is only slightly off zero, as in Fig. 3.21. The ONIOM method allows each subsystem to have a subtotal charge fixed at a prescribed value, as illustrated in Fig. 3.22d, for a case with total charge equal to +2 atomic units.

Figure 3.22. a-c: Examples of DFT calculations without constraints on subsystem charges. d: example of calculation with subsystem charges prescribed.

The ONIOM method is used in the DFT calculation shown in Fig. 3.23, first with all four charges on the hydrogen atoms high above the metal surface, in the third step by two charges moved to OH+ molecules, in the forth step these charges are neutralised by electrons from the Ni-surface, and the two remaining protons are attached to the now neutral OH. Now, another pair of electrons may be transferred from the Ni surface to the weakly bound OH-H⁺ group, forming water at little energy change. The charge of the Ni surface now increases towards +4. Like in Fig. 3.21, electrons from the outer circuit have to be drawn in to allow the formation of a water molecule that is capable of leaving the surface (reducing the charge on the catalyst from 4 to 2 units). This is a realistic outcome, explaining that unless there is an external circuit with a load, the water will not leave the catalyst surface but just pile up on the surface. Further reduction of the charge on the Ni surface to near zero (in the condition of a high external current) does not necessarily increase the release of water molecules. Calculated total energy gains exceed the observed (<5 eV) value, particularly for situations with hydrogen atoms far from the rest (in concord with the discussion of Table 3.1), and Fig. 3.23 is still an incomplete description of the processes taking place at the positive electrode of a fuel cell, but qualitatively acceptable.

Some of the short-comings of the quantum chemistry DFT model may be explained by the simplicity of the small chunk of catalyst surface modelled. An alternative method taking less computer time would be to use the simplification of a periodic layout (using periodic boundary conditions), as furnished by an infinitely extending nickel surface (see the calculation behind Fig. 2.46). However, this method requires periodicity also in the atoms above the catalyst surface, implying an assumption of synchronised transfer of a number of protons along the entire catalyst surface. Okazaki et al. (2004) use this method to calculate the positive electrode water formation process and finds an energy gain smaller than but closer to the experimental value than the one predicted by the ONIOM method (Fig. 3.23), where step-3 overestimates the energy gain, or the combined-structure calculation (Fig. 3.21), which very grossly overestimates the potential energy gain.

Figure 3.23. A series of potential energy calculations for a sequence of structural arrangements of a Ni surface, two O atoms, and four H atoms. The calculations use an ONIOM method with the Ni surface as the low layer in a HF calculation, the two O or O-H pairs as the intermediate layer, and the last four or two H atoms as the high layer, using the B3LYP density functional method. The overall charge is +4, except for the two rightmost columns, where it is +2. At first (left side), the charges are placed on the 4 H atoms, then first 2 and then all 4 charges are moved to the Ni layer. The two columns to the right have electrons added from an external circuit. Their energy has been shifted by the average binding energy. Creating an external current allows the water formed to escape from the Ni surface. The too large energy difference between columns 2 and 3 is due to the inability of DFT calculations to correctly describe atoms far from the main molecule (Sørensen, 2005a).

One would think that the behaviour at the negative electrode is easier to describe on a quantum chemical level. The splitting of the hydrogen molecules has already been explained through calculations such as those behind Fig. 2.5. Figure 3.24 shows that just as at the positive electrode, an external current has to be at work for the hydrogen ions (protons) to be able to move away from the Ni catalyst-surface, and that also in this case, the overall energy gain is considerably overestimated.

Figure 3.24. Hydrogen dissociation and ionisation at the negative electrode of a fuel cell, followed by the protons leaving the surface and entering the electolyte layer. The DFT calculation uses the ONIOM method, a SV basis sert, Hartree-Fock for the catalyst layer and B3LYP for the surface atoms. The total charge is +4 atomic units for the 3 left columns, and +2 for the two to the right. The predicted energy gain exceeds the measured one (Sørensen, 2005a).

In summary, the presence of a metal such as Ni as a catalyst-electrode is found to be essential both for the charge transfer constituting the electric current of the fuel cell and also for making it energetically favourable to form water molecules and move them away from the reaction site. The simple calculations described here provide a reasonable qualitative explanation of the mechanisms in operation for an operating fuel cell on the quantum mechanical level, but fails to provide correct energy steps, due to the general problems of methods based density functional theory to describe collections of molecules, which are not close to each other, as discussed in Chapter 2.

3.1.5 Flow and diffusion modelling

While an understanding of the molecular processes at the fuel cell electrodes requires a quantum mechanical description, the flows through the inlet channels, the gas diffusion layer and across the electrolyte can be described by classical physical theories such as fluid mechanics and diffusion theory. The equivalent of Newton's equations for continuous media is an Eulerian transport equation of the form

$$\frac{\partial}{\partial t}(\rho A) + \operatorname{div}(\rho \, v A) + \operatorname{div}(s_A) = E_A, \tag{3.50}$$

where t is time, ρ is the density, v is the velocity at a given point (x,y,z) in

space, s^A is a vector describing microscopic transport due to molecular processes, and EA describers external sources of the quantity A looked at. For fluid flow, A would be a component of the velocity v, while for A = 1, (3.50) is an equation of continuity,

$$\frac{\partial}{\partial t}(\rho) + \operatorname{div}(\rho v) = 0. \tag{3.51}$$

When A = v, the molecular transfer of velocity is related to the stress tensor ^τij by (see, e.g., Sørensen, 2004a)

$$-(s_{v_i})i = \tau{ij} = \left(-P + \left(\eta - \frac{2}{3}\eta\right)\rho \operatorname{div}(v)\right)\delta_{ij} + \eta\rho\left(\frac{\partial v_i}{\partial x_j} + \frac{\partial v_j}{\partial x_i}\right),\tag{3.52}$$

where P is the pressure, η is the kinematic viscosity, and η' is the volume viscosity. If the latter is neglected and the fluid is treated as incompressible, the simpler version of equation (3.50) for A = v with (3.52) inserted is called the Navier-Stokes equation.

Numerical solution of equations such as (3.50)-(3.52) can be done in various ways, each suitable for certain classes of problems. Flow modelling is used in many problem areas, including climate modelling, weather forecasting, aircraft and vehicle air resistance and flow over wings and similarly wind turbine blades, flow in pipes and combustion boilers and turbines, and so on. Flow in the gas channels of fuel cells is just one example. The numerical methods used comprise two types. In one, the integration space is discretised, that is, the functions considered are given only on a mesh of points in space and differential equations become difference equations. Each point is assigned average values of the quantities studied, and the solutions may be sought in terms of an initial value problem (levels and velocities, corresponding to variables and their derivatives in the continuous case, known at each point at one time t0) or of a boundary problem (levels and velocities specified at the boundaries that define the volume for which calculations are performed).

The other method of calculation defines a discrete set of (generally) threedimensional regions spanning the volume of interest in space and expands the solutions on a polynomial basis set, typically just piecewise linear functions, for each of the elemental regions (called finite elements). Each of the basis polynomials is zero, except for a few elemental regions, and the solution to, for example, the flow field v is a linear combination of the basis functions. The coefficients are determined either from applying the differential equations at a finite number of points spanning the elements or by minimising a suitable aggregate function such as the total energy, in case the solution sought corresponds to a stationary point in energy space. The finite elements, or elemental regions, selected would typically be large in regions where little variation in v is expected, but smaller in regions where interesting variations are expected. For some problems, symmetry in design may allow the transport equations to be solved in less than three dimensions.

Examples of employing the fluid dynamics or finite element method will be shown in the following sections. The methods can obviously be used also for variables other than the velocity, by a suitable selection of A in (3.50), such as, for example, temperature, once the basic velocity field has been determined (as it enters in the equations for all other quantities). In the case of temperature T, the source term ET in (3.50) includes external sources of heat and condensation, and the ideal gas law may be used to connect temperature and pressure.

Neither the finite element method nor a discretised integration will "catch" eddy or other motion below a certain scale determined by the choice of mesh. Small-scale motion in many cases may be better described as random, in which case the transport of the quantity A is called diffusion. Diffusion can be described by Fick's law, assuming that the flux density *f*−that is, the number of "particles" (here a small parcel of a gas), passing a unit square in a given direction−is proportional to the negative gradient of particle concentration n (Bockris and Despic, 2004),

$$f = -D \operatorname{grad} n. \tag{3.53}$$

The proportionality factor D characterising the medium is called the diffusion constant. It is assumed that each particle−that is, each parcel of the fluid−is moving under the influence of the average field from the rest of the fluid. Fick's law is then combined with the continuity equation similar to (3.51)

$$\operatorname{div} f + \frac{\partial n}{\partial t} = 0, \tag{3.54}$$

expressing that changes in concentration are equal to minus the sum of fluxes out of the parcel volume in all directions. Combining (3.53) and (3.54), one gets the diffusion equation

$$D \Delta n = \frac{\partial n}{\partial t},\tag{3.55}$$

where the notation ∆ = ^∇ ² = (div grad) is used. Equation (3.55) describes diffusion of a single substance. In practical cases such as the fuel cell types considered in the following sections, there will often be several chemical components characterised by their mole fractions xi (see (3.6)). The equations corresponding to Fick's law are now called the Stefan-Maxwell equation, and may, with use of the ideal gas law in the form valid for partial pressures pi and molar fractions of the *i*th substance, ni , be written (Bird et al., 2001)

$$\operatorname{grad} xi = \frac{\mathcal{R}T}{P} \sum{j \neq i} \frac{x_i f_j - x_j fi}{D{ij}},$$ (3.56)

$$p_i = n_i \,\mathcal{R} \, \mathrm{T/V}, \tag{3.57}$$

$$P = \sum p{i}, \quad n = \sum n{i}, \quad n{i} = x{i} n, \tag{3.58}$$

where R is the gas constant (see (3.6)), T is the temperature, and P is the total pressure. The flux density of the *i*th component is denoted fi and the total volume is denoted V. The diffusion constant Dij describes the inter-diffusion of components i and j and is proportional to the mean free path for collisions between the substances i and j and to their relative speeds ((vi + vj 2 )/2)1/2. The flux density of the *i*th substance is related to the velocity of this substance at a given point by

$$f_i = \varepsilon c_i v_i \tag{3.59}$$

where ε is the porosity and ci is the molar concentration of substance i ( ∑ ci = ctotal). For the small pore sizes found in many gas diffusion layers, collisions with pore walls become as important as collisions between the particles in the substances, and the right-hand side of (3.56) receives an additional contribution (Knudsen, 1934),

$$-\frac{\mathscr{R}Tf_i}{PD_i},\tag{3.60}$$

where D'i is a pressure-independent diffusion coefficient depending on pore structure, and where the particle collision-type diffusion parameters Dij in (3.56) may have to be somewhat adjusted by the presence of pore walls (Bruggeman, 1935).

Many numerical models make additional assumptions, valid if only some specific questions are being asked. For example, if one is not interested in the start-up phase or in changing the operation of a fuel cell, one may apply the steady state condition that only time-independent solutions are requested. In certain problems, one may disregard temperature variations, and in the free gas ducts, laminar flow may be imposed. The diffusion in porous media is often approximated by an assumption of isotropy for the gas diffusion or membrane layer, and the coupling to chemical reactions is often simplified or omitted. Water evaporation and condensation, on the other hand, are often a key determinant for the behaviour of a fuel cell and thus have to be modelled at some level.

The management of water is an integral part of operating many fuel cell types. For those operating at high temperatures, water vapour can be modelled as just one additional component of gas flow. However, for cells operating at lower temperatures, a two-phase flow model is required, and in general, there will be both liquid and gas forms of water present in both gas diffusion layer and gas channels (plus in the liquid electrolyte or membrane). It is thus necessary to consider both transport of both phases and also evaporation and condensation processes taking place in various parts of the cell. The water generated by the cell from hydrogen and oxygen may at its origin be considered as gaseous and generated in proportion to the current density. Subsequently, the water vapour condenses to liquid form, likely to appear predominantly at the positive electrode. The motion of the liquid water is then proceeding through the gas diffusion pores, with capillary pressure gradients driving the flow being the most probable mechanism of transport (Wang et al., 2001; Nguyen et al., 2004). The relation between the fluid velocity and the pressure gradient may be expressed by Darcy's law

$$v = -\frac{K(s)}{\varepsilon \eta} \operatorname{grad} P, \tag{3.61}$$

where K is the permeability that may be taken to depend on the level of water saturation in the pore, *s*−that is, the liquid volume fraction. The pressure P is for this unsaturated flow proportional to a capillary pressure function that also depends on s (Stockie, 2003; Stockie et al., 2003). The transport of water in pores may be enhanced by impregnating the gas diffusion layer with, for example, teflon to reduce adhesion to pore walls.

3.1.6 The temperature factor

The following sections discuss a number of fuel cell concepts. The reason there are so many in part has to do with the historical development of the technology, but also to a large extent the temperature regime for which they are designed. In many cases, high operation temperature will entail a higher efficiency of electricity production, but of course maintenance of a high temperature will also influence efficiency and in some cases, such as coldstart of a vehicle, be inconvenient.

Generally, large stationary fuel cell installations aim at the highest possible efficiency, which points in the direction of molten carbonate fuel cells (operating around 670°C) or solid oxide fuel cells (currently operating at around 800°C or higher). For decentralised stationary uses, lower temperatures may be contemplated, although presently used boiler units for heating operate at temperatures similar to the high-temperature fuel cell concepts. For automotive applications, the absence of combustion implies that heat would have to be supplied initially and sustained during driving with insulation and topup heat supply−for example, from batteries. This is generally considered inconvenient, because fuel cells are often seen as an alternative to electrically powered vehicles, and even in hybrid concepts, the energy use for heating would substantially decrease the overall efficiency of operation. This statement can be made more precise by considering the loss of efficiency implied by using low-temperature fuel cells and comparing it with the energy cost of having to heat the fuel cell to a higher operating temperature.

The bottom line is that at the present conception of the various technologies, proton exchange membrane (PEM) fuel cells operating near ambient temperature are considered the best choice for (road) vehicles, while the hightemperature fuel cells are considered as suitable for centralised stationary applications. The market development may interfere with this line of thought, because the automotive market is seen as a key entry point for fuel cell technology, and if this large market succeeds in bringing the cost of PEM fuel cell technology down, such cells may out-compete the other options. Signs of this are already evident in the current development phase, because several manufacturers take advantage of the strong effort made to ready PEM cells for the automotive market to develop small buildingintegrated PEM fuel cell units for replacement of natural gas boilers. Because low-efficiency and high-pollution problems are much more severe in the transportation sector than in the fossil fuel-based power production sector, the initial market for high-temperature fuel cells for general power production may be very small and thereby prevent such fuel cells from taking advantage of a production volume-related cost decline.

3.2 Molten carbonate cells

A fuel cell cycle employing carbonate ions penetrating a solid matrix electrolyte at high temperatures is schematically illustrated in Fig. 3.25. It is aimed at stationary applications and promises high efficiency. The electrode reactions for this electricity-producing molten carbonate fuel cell (MCFC) are

$$H_2 + CO_3^{2-} \rightarrow CO_2 + H_2O + 2e^-,$$ (3.62)

$$\frac{1}{2}O_2 + CO_2 + 2e^- \rightarrow CO_3^{2-}$$ (3.63)

The carbon dioxide from (3.63) is in most designs recycled as input to (3.62) together with the hydrogen fuel. It has been suggested that CO2 emissions from fossil power plants could be used as input as a way of reducing greenhouse gas emissions from the current system. However, then the output CO2 from (3.62) would have to be collected and disposed of, for example, by transforming it once more to the form of carbonates for deposition as waste (Lusardi et al., 2004). As indicated in Fig. 3.25, all the gases may have to be present in the appropriate channels in order to provide the required flow and to recycle fuel gases not used in a particular pass. This signals a need for fuel processing loops separated from the fuel cell stack. The cell electrolyte consists of Li-K (for systems operating near atmospheric pressure) or Li-Na compounds (for elevated pressure systems) able to carry the molten alkaline carbonate mixture, while for stability and strength the assembly is being supported within a porous aluminate (e.g., LiAlO2) matrix.

Figure 3.25. Schematic picture of a molten carbonate fuel cell. The gases in parentheses are not necessarily taking part in the electrochemical processes, but they are required either for ensuring flow or for carrying excess gases away. The arrangement of CO2 recycling from negative to positive electrode can be made in different ways.

Electrode materials are for the negative electrode Ni with Cr or Al additives for providing material strength and for the positive electrode NiO with additives of Mg or Fe, mainly serving the purpose of avoiding short-circuit phenomena (conductivity across the electrolyte from positive to negative electrode). The carbonate ions formed at the positive electrode travel through the matrix and combine with hydrogen at the negative electrode at an operating temperature of 660°C or higher. The process was originally aimed at hydrogen supplied from coal gasification or natural gas conversion. The first research installations (being built in the 1980s, after a few European precursors) suffered severe degradation problems in the electrolyte, including cracking of the matrix (Mugikura, 2003). Also the prototype units put into operation around the year 2000 are considered vulnerable to possible structural change effects after longer periods of operation, notably affecting seal coatings and the integrity of the alumina matrices (Jun et al., 2002; Frangini and Masci, 2004; Mendoza et al., 2004). The first prototype tests of MCFC power plants (250 kW or larger) were conducted during the period 1996-2000 in the United States, Japan, Italy, and Germany (Farooque and Ghezel-Ayagh, 2003). Conversion efficiencies in the range of 55-60% were reached, but, theoretically, the efficiency could become higher (near 70%). Despite the relatively short periods of operating the demonstration plants, degradation of key components affecting plant lifetime were observed, and efforts to improve the situation are ongoing. It should be kept in mind that degradation of electrolyte and electrodes is inherent in all electrochemical systems and is the main reason for the limited life of conventional or advanced batteries. The challenge is whether it is possible to extend the lifetime for fuel cells beyond the under-5-year level characterising most battery types, which by some are seen as necessary for reaching economic viability.

The electrolyte constitutes a separator for the gas flows shown in Fig. 3.25, in addition to providing ionic transport. Physically, it consists of a compressed powder behaving as a soft paste at the operating temperatures, thus substantiating the name "molten carbonate" for the liquid Li-Na or Li-K carbonates inhabiting the electrolyte substrate. The LiAlO2 material used for the support structure can exist in three different forms (phase α, β, and γ), and transitions have been observed (e.g., γ to α) after extended fuel cell operation times, leading to enhanced degradation of the electrolyte integrity. Once the degradation processes have reached a certain level, the paste pore structure opens up and shorting by precipitation of Ni between the electrodes ends the electrolyte life. The Li-K electrolyte suffers a higher rate of acidic dissolution than Li-Na, but the latter shows a more pronounced sensitivity to high temperatures (Hoffmann et al., 2003). Using α-type LiAlO2 rather than the conventional γ -type may reduce degradation (Batra et al., 2002).

The stability of the negative electrode is fairly good, but the gas flowing past the positive electrode is highly corrosive and responsible for a significant dissolution of the NiO material. The Ni erosion increases with pressure and is very high at the temperatures of 660-700°C prevailing in MCFCs. It declines when the temperature is raised above 1000 K. NiO dissociation rates in the range of 0.01-0.02 g m⁻² h⁻¹ have been measured at ambient pressure, rising by an order of magnitude for higher pressures (0.7 MPa; Freni et al., 1998). The high-pressure value implies a service life of only a few months, and even the lower value for ambient pressure systems leads to a lifetime (under 3 years) lower than what is considered necessary. Impedance measurements of the catalytic activity of Li-Ni-O (used to enhance the conductivity of the positive NiO electrode) under the influence of an eutectic Li-K melt and a corrosive CO2-O2 atmosphere have shown a transfer of lithium and related morphological change (expansion) of the NiO electrode material (Escudero et al., 2003). Also, the reduction of electrode area has a negative influence on cell performance (Freni et al., 1998). It has been proposed that separating the CO2 and O2 inlet flows (Fig. 3.25) into two separate streams, with the CO2 flow entering the electrolyte matrix directly, on the opposite side of the positive electrode relative to the O2 flow, may reduce the corrosion problem significantly, at a modest loss of efficiency (Au et al., 2003). Specific corrosion problems arising for MCFCs using biogas as a fuel, due to sulphur and other impurities, have been looked into by Zaza et al. (2011).

In actual designs of MCFCs, flows are controlled by separator plates with corrugated patterns, and wet seals are used to separate corrosive gas flows from the sensitive electrolyte matrix and still allow the molten carbonate to pass through. Adding carbonates of Ca, Sr, and Ba may diminish the NiO dissociation (Tanimoto et al., 2004). The cell voltage at zero current density is about 1 V and diminishes by around 30% for a current density of 2 kA m⁻² (Freni et al., 1997). It is hoped that a lifetime exceeding 5 years can be achieved by implementing improvements as just suggested.

Simple models for performance simulation have been developed for stationary flow situations, with the options of verifying the voltage-current relations and calculating the overall efficiency under consideration of total plant auxiliary power and heat inputs. The efficiency calculated with such models exhibits a few percent efficiency improvement from raising the pressure above ambient by the factors considered above and, as expected, a decline with increased gas flow rates (Simon et al., 2003). Consideration of thermodynamic constraints has been used to define optimum working areas in terms of fuel inputs (Zhang et al., 2011).

3.3 Solid oxide cells

The high-temperature fuel cell to which most attention is paid at present is the solid electrolyte cell. Solid oxide fuel cells (SOFC) use zirconia compounds as the electrolyte layer to conduct oxygen ions formed at the positive electrode. The electrode reactions involve oxygen ion transport (in contrast to the hydrogen ion transport in the basic scheme given by (3.15) and (3.16)), whence the appearance of "oxygen" in the cell name,

$$H_2 + O^{2-} \rightarrow H_2O + 2e^-,$$ (3.64)

$$\frac{1}{2}O_2 + 2e^- \to O^{2-}$$ . (3.65)

These reactions take place at the surface of a solid state electrolyte at a temperature of 600-1000°C. The lower temperatures are desirable, due to a wider choice of materials capable of maintaining integrity. A number of materials have been contemplated for use as electrodes and electrolyte.

The electrolyte should be able to conduct oxygen ions, but must be impermeable to the hydrogen and oxygen gases flowing along its sides. For this purpose, a solid membrane structure is used as electrolyte. It may consist of a thin layer of yttrium-stabilised zirconia, ZrO2, doped with 3-8 mol% Y2O3 and possibly adding other dopants such as Sc2O3. A molecular structure is shown in Fig. 3.26. This electrolyte material may be sprayed onto the negative electrode as a ceramic powder with a thickness of around 10 µm (Kahn, 1996) or may be self-supporting at a thickness above 100 µm (Weber and Ivers-Tiffée, 2004). Generally, electrolyte losses increase with electrolyte thickness, so when possible the thinnest sheets not giving rise to short circuiting or gas penetration are preferable. For the lowest temperature cells, alternative electrolytes are needed in order to achieve sufficient conductivity. Among these are (Y-)doped ceria (CeO2), used for direct hydrocarbon SOFCs or in two-layer electrolytes. The second layer may be a lanthanum gallate electrolyte (La1-xSrxGa1-yMgyO3), which exhibits high oxide ion conductivity but may be discouraged by stability problems and gallium cost.

The electrodes should be able to catalyse the relevant reactions (i.e., (3.64), (3.65) and necessary fuel transformation reactions, in case the initial fuel is not hydrogen). They need to have a large effective surface, but still possess long-term stability at the operating temperatures and with considerations of the temperature cycling taking place in practical applications. For designs such as the one shown in Fig. 3.29, where the gas flow is on the outer side of the electrodes, the electrodes must also be permeable for the gases involved in the cell reactions. The electrodes suitable for the temperature regime relevant to SOFCs are typically made of oxides of metals in or near the rare earth sequence in the periodic system. For the positive electrodes, the materials used include compounds such as La1-xSrxMn1-yCoyO3, possibly with Fe instead of Mn and Ca instead of Sr. The molecular structure of LaCoO3 is shown in Fig. 3.27. Typically, some 20% of the La atoms are replaced by Sr (x = 0.2), and the amount of Co (y) controls the electric conductivity and the thermal expansion rate of the electrode, both increasing with y. In particular, the positive electrode requires the dopants to be stable in a hightemperature oxygen atmosphere.

The negative electrode may consist of an Ni-based compound, using, for example, a mixture of NiO with a gadolinium-doped cermet, Ce1-xGdxO1.95 (x around 0.1), and RuO2 as a catalyst for use at around 600°C (Hibino et al., 2003). Temperatures around 600°C are required for the direct use of hydrocarbon fuels (e.g., methane, ethane, or propane) instead of hydrogen, which is then formed in a second step by internal reformation. Also in use are negative electrodes combining the same yttrium-stabilized zirconia as used for the electrolyte with Ni and Ce oxide (Weber and Ivers-Tiffée, 2004).

Techniques used in metal−organic photoelectrochemistry (see Sørensen, 2004a; Chapter 4, section 4.2.3) may be used to create a large surface area of contact between the electrolyte and the electrode material.

Figure 3.26. Structure of eight unit cells of an yttrium-stabilised zirconium oxide, Zr0.75Y0.25O2. The smaller (dark) atoms are oxygen, and the larger ones are Y (lightest grey shade) or Zr, sitting at cube corners or face centred. The oxygen atoms have been placed in the most symmetric configuration. For independent ZrO2 molecules, the angle between the O atoms is only 21° and the O-Zr distance is 0.2 nm, a little larger than in the lattice.

Figure 3.27. Structure of eight unit cells of LaCoO3, with body-centred La atoms (large, light shaded) and Co (medium, darker) at the unit cell corners and O atoms (smallest, darkest) halfway between them, in the same arrangement as for perovskite (e.g., CaTiO3 or MgSiO3). Perovskites can exist in another structure, but only at the very high pressure (125 GPa) and temperature (2700 K) found at the Earth's core−mantle interface (Murakami et al., 2004).

Figure 3.28 summarises data on conductivity of some materials considered for SOFC electrolytes, as a function of temperature (values along top, inverse temperatures along bottom abscissa). The uniform decline in conductivity at lower temperatures illustrates the compromise on efficiency associ-

Figure 3.28. Electrolyte conductivity σ for La0.8Sr0.2Ga0.8Mg0.115Co0.085O3 (open circles) and other materials (the unit AV⁻¹ is sometimes referred to as a siemens), as a function of temperature. (From Ishihara et al. (2004). Novel fast oxide ion conductor and application for the electrolyte of solid oxide fuel cell. J. European Ceramic Soc. 24, 1329-1335. Used by permission from Elsevier).

Figure 3.29 shows a cylindrical layout often used for high-temperature SOFCs. Alternatives are a stack of planar cells or a disk concept with feed tubes in the centre. A consideration of efficient heat exchange is the common design strategy for the high-temperature fuel cell geometry.

Modelling the behaviour of solid oxide fuel cells employs a combination of an electrochemical model and a heat and materials flow model. The electrochemical model starts from calculating the internal cell potential by combination of Eqs. (3.5)−(3.7) with (3.20) and (3.23),

$$\phi = \sum{i} n{e,i} \, \Delta \mu{i} - \phi{losses} = \sum{i} n{e,i} \mu{i}^{0} + \mathcal{R} T \sum{i} n{e,i} \log(f{i} x{i}) - \phi{losses}.$$ (3.66)

For the SOFCs, the sums in (3.66) have three terms, two positive ones for H2 and O2 and one negative one for H2O. The constant part of the first terms on the right-hand side, φ⁰ , deviates from the one given in (3.20) for an ideal gas at ambient pressure and temperature, as it must take into account the high temperature in the SOFC. It is estimated as (Campanari and Iora, 2004),

$$\phi^{0} = 1.2723 - 2.7645 \times 10^{-4} T_{cell}. \tag{3.67}$$

Figure 3.29. Layout of a cylindrical SOFC pipe. Air (or oxygen) enters the centre tube and is deflected back through the next cylinder ring after being heated to operating temperature. The hydrogen (or other fuel) flow is external to the cylinder, and cells may be assembled in packages with electric interconnects as indicated.

The second sum on the right-hand side of (3.66) is called the Nernst potential, and finally, the loss term is as in (3.23) made up of contributions from bulk resistive losses and losses occurring at each of the two electrodes. These terms can be modelled in terms of measured Ohmic resistance and parametrised activation losses at electrodes (Costamagna et al., 2004; Campanari and Iora, 2004).

For describing heat flow and mass transport, models based on the finite element method or some volume-averaging scheme for handling general Eulerian transport are used, as described in section 3.1.5. Several programs are available for calculating the time-dependent behaviour of such systems, in terms of fields of velocity, matter concentration, and temperature, and considering couplings to the electrochemistry and other chemical reactions that may take place. Examples of the results of such calculations for SOFCs are given following. Also simulations of entire SOFC stacks are available (Roos et al., 2003).

Figure 3.30 shows an example of the air velocity field in an SOFC of the shape given in Fig. 3.29. The segment shown goes from the centre of the air inlet to the wall of the second tube used for return air (vertical axis) and horizontally covers the innermost part of the tube, where the air (with its oxygen) is forced to deflect back into the outer of the two tubes. This is evidently the most difficult part to model, and most fluid dynamics models allow the number of computation elements to be increased in such regions.

Figure 3.30. Velocity field calculated by a fluid dynamics model for the turning area of the air/oxygen flow in a solid oxide fuel cell at operating temperature. (Reprinted from S. Campanari and P. Iora (2004). Definition and sensitivity analysis of a finite volume SOFC model for a tubular cell geometry. J. Power Sources 132, 113-126. Used by permission from Elsevier.)

The temperature field from a similar simulation is shown in Fig. 3.31, comprising both the two air/O2 carrying tubes and their internal wall and also the cell area (electrodes and electrolyte; see Fig. 3.29), which in the model cell is at radii between 0.5 and 0.72 cm, and finally, a piece of the channel for fuel flow. The temperature distribution over the length of the cell is not quite uniform, but shows 100-150°C lower values near the ends of the tube cylinder for the cell and fuel compartments. It is seen that the air/oxygen entering the inlet tube at lower temperature does not acquire a uniform temperature over the radial dimension of the tube, but only reaches the cell operating temperature in the part of the cylinder closest to the electrode assembly (and only in the middle of the cylinder). The assembly of positive electrode, solid electrolyte, and negative electrode, as well as the fuel flow region adjacent to the negative electrode, on the other hand exhibits a temperature dependent only on the position x along the tube length. Figure 3.32 shows similar temperature profiles as a function of x only.

The two calculations underlying Figs. 3.31 and 3.32 agree on the behaviour of the cell assembly and fuel temperatures, but the calculation shown in Fig. 3.32 assumes pre-heating of the air/O2 flow to a higher inlet temperature than the one appearing from Fig. 3.31. The return air/O2 flow in Fig. 3.32 is therefore losing temperature when it is slowed down at the internal bend, under influence of the lower temperature in the cell assembly. It indicates that, for this particular case, less heating of the inlet air might be better and that a higher inlet fuel temperature could be desirable, although possibly problematic for safety reasons. In any case, the figures illustrate the usefulness of being able to model the dynamic behaviour of the cells.

Figure 3.31. Temperature field along the length x of a tubular SOFC cylinder (air/O2

inlet to the right, tube end with flow bending at left, radially from r = 0 to 0.5 cm), as simulated by a dynamic flow model. The middle horizontal band is the electrodeelectrolyte assembly, and the top part (above r = 0.72 cm) is the fuel channel. (Reprinted from P. Li and M. Chyu (2003). Simulation of the chemical/electrochemical reactions and heat/mass transfer for a tubular SOFC in a stack. J. Power Sources 124, 487-498. Used by permission from Elsevier.)

The fuel input streams for the models illustrated in Figs. 3.31 and 3.32 are considered as the mixture of gases typically arising from previous steam reforming of natural gas. The calculation shown in Fig. 3.32 and the following assumes an inlet fuel composition of 26% H2 (molar fraction), 11% CH4, 23% CO2, 6% CO, 6% N2, and 28% water (Campanari and Iora, 2004). The calculation shown in Fig. 3.31 assumes more methane and no nitrogen. The option of internal reforming of fossil fuels within the SOFC will be discussed following. The inclusion of the possible reactions between fuel and gas flow components, in addition to those at the electrodes, must be considered in models for combined reforming and fuel cell reactions.

A recent model using computational fluid dynamics (Ni, 2010) finds that the maximum temperature has a dependence on the inlet gas velocity, whereas heat transfer is less depending on electrode geometry.

Figure 3.32. Temperature profiles from an SOFC flow model, as a function of distance from the H2/fuel inlet (see Fig. 3.29). The four temperature curves are Ta: inlet O2/air centre tube; Tc: return O2/air tube along positive electrode; Ts: electrode-electrolyte assembly; and Tf: H2/fuel flow. (Reprinted from S. Campanari and P. Iora (2004). Definition and sensitivity analysis of a finite volume SOFC model for a tubular cell geometry. J. Power Sources 132, 113-126. Used by permission from Elsevier.)

The temperature in an SOFC bundle is high enough for reforming of many hydrocarbons to take place within the reactor, so that the hydrogen required will be created internally, and the fuel to handle externally could be natural gas (methane), gasified coal, or liquid hydrocarbons. For methane, the reactions (2.1), (2.2), (2.10), and (2.14) in Chapter 2 may take place near the hot fuel cell electrode, and for higher hydrocarbons the reactions (2.9) and (2.11) take place. Figure 3.33 shows a process diagram for a proposed natural gasbased SOFC, showing some of the additional problems intro-duced with the use of fuels other than pure hydrogen. The natural gas has to be desulphurised and eventually pre-reformed in a more efficient reformer than the fuel cell. The fuel cell faces additional corrosion problems due to carbon depositing on electrodes, formed by partial cracking of the types

$$C_n H_m \to nC + (m/2) H_2.$$ (3.68)

For high-n hydrocarbons, cracking is abundant at the temperatures prevailing in an SOFC, while for methane, it is only a problem in the presence of certain catalysts, particularly Ni-based ones. A further problem is associated with the gases leaving the fuel cell area. Figure 3.34 shows the composition of gases along the flow path in the SOFC of Fig. 3.29 and used in Fig. 3.32, the initial gas composition being described following. It is seen that the CH4

reformation happens during the early part of the flow path. The carbon is seen to be mainly continuing in the form of CO. The hydrogen formed plus the amount already present in the input fuel serves as fuel for the SOFC during the middle flow path, and at the end, some 30% of the hydrogen leaves the fuel cell unprocessed, along with the quantities of water initially present or formed by the fuel cell reactions. In Fig. 3.33, it is proposed to add a catalytic afterburner to burn the remaining hydrogen and possibly remove more pollutants. The scheme does not sequester the CO2 in the output stream and considers a fuel with 88% methane, with the rest made up by higher hydrocarbons. As it stands, SOFCs operating on fuels other than hydrogen do not live up to the fuel cell ideal of emitting "just water".

Figure 3.33. System layout for SOFC plant based on natural gas fuel (NG), with gas cleaning except for CO2 and with the possibility of combined power and heat output. "Anode" and "cathode" are obsolete names for the negative and positive electrodes. (From Fontell et al. (2004). Conceptual study of a 250 kW planar SOFC system for CHP application. J. Power Sources 131, 49-56. Used by permission from Elsevier.)

The cell losses discussed in connection with (3.66) include the three components depicted in Fig. 3.35 as a function of the SOFC flow path for the mixed gas fuel cell underlying Figs. 3.30 and 3.32. They are the bulk Ohmic losses and losses associated with overcoming the potential barrier for each of the electrodes. Further bulk loss terms associated with diffusion polarisation are much smaller (Campanari and Iora, 2004; see also the dissociation barrier discussion in Chapter 2, sections 2.1.1 and 2.1.3). Figure 3.35 shows that the voltage losses mainly take place in the part of the electrolyte layer residing near the fuel inlet side of the cell cylinder, due to the lower temperatures prevailing in this area. At the maximum temperature locations along the electrodes (see Fig. 3.32), the loss terms are substantially smaller.

In principle, reverse operation of SOFCs is possible, although little data on efficiency obtained in praxis are available (European Platform for Hydrogen and Fuel Cell Technologies, 2004).

Figure 3.34. Molar composition profiles along the flow path in the fuel channel, calculated using a flow model of the tubular SOFC, with the abscissa being the distance from the H2/fuel inlet (see Fig. 3.29). The six curves reflect the six components in the fuel used in the cell modelled (see text). (Reprinted from S. Campanari and P. Iora (2004). Definition and sensitivity analysis of a finite volume SOFC model for a tubular cell geometry. J. Power Sources 132, 113-126. Used by permission from Elsevier.)

Overall efficiency of SOFCs depends on fuels, materials, and operating temperature and on the total device capability of supplying and recycling heat required for maintaining the operating temperature, beyond what can be generated as waste heat of the fuel cell reactions themselves. Figure 3.36 shows the cell potential and current density for the modelling exercise pursued through Figs. 3.32, 3.34, and 3.35, as a function of the path through the length of the cell cylinder. The Nernst part of the cell potential φ (3.66) declines after the initial stretch, at least as fast as the decline in hydrogen content in the fuel stream (Fig. 3.34). The current density i exhibits a maximum in the early middle of the path, when the fuel cell reactions are most active and the fuel is not yet much depleted. Figure 3.37 gives a number of measured values of φ , now as a function of current density i for a planar single cell SOFC device, at different operating temperatures. The right-hand axis gives the corresponding power density, e = φ i. The voltage drops to zero at higher *i*s, and the power density thus declines to zero after having passed the maximum (approximately where the curves in the figure stop).

Figure 3.35. Modelled main cell-potential loss terms. Ohmic: bulk losses in electrolyte; Act. an: activation losses at negative electrode; Act. cat: activation losses at positive electrode. (Reprinted from S. Campanari and P. Iora (2004). Definition and sensitivity analysis of a finite volume SOFC model for a tubular cell geometry. J. Power Sources 132, 113-126. Used by permission from Elsevier.)

Figure 3.36. The Nernst part of the cell potential (3.66) and the current density, from an SOFC model calculation and like the previous figures given as a function of distance from the fuel inlet. (Reprinted from S. Campanari and P. Iora (2004). Definition and sensitivity analysis of a finite volume SOFC model for a tubular cell geometry. J. Power Sources 132, 113-126. Used by permission from Elsevier.)

Efforts to optimise design are not only concerned with the selection of materials for electrodes and electrolyte, but also with the microstructure of the surfaces between them. Activity areas must be large, causing as mentioned previously interest in the surface-increasing deposition techniques developed for metal-organic solar cells. The surface control reactions comprise bulk and surface diffusion, adsorption, dissociation, charge transfer, and chem.ical reactions, with consideration of the dynamics of motion ("kinetics") for each constituent, whether a molecule, an atom, or a charged entity (ion or electron) (Kawada and Mizusaki, 2003). Figure 3.38 shows a cut through the electrode−electrolyte interface of a laboratory-scale SOFC.

Figure 3.37. Voltage and power density for a single, planar SOFC, as a function of current density in the range from zero to approximately the value giving maximum power. The different curves are for different operating temperatures, as indicated in the box. (From D. Ghosh (2003). Development of stationary solid oxide fuel cells at Global Thermoelectric Inc. Presented at "14th World Hydrogen Energy Conf., Montréal 2002". Used by permission from Canadian Hydrogen Society.)

The bipolar plates (Fig. 3.29) providing electric contact between cells and forming impermeable channels for gas transport are in most SOFC designs made of ferritic stainless steel (low thermal expansion coefficient), although at the highest operating temperatures, more advanced (and expensive) metal structures must be used.

Target design lifetime is of the order of 40 000 h for stationary installations (Tu and Stimming, 2004; a 5000 h target for mobile systems is less relevant for SOFC). The planar designs seem limited to the lower temperature regimes, and high-temperature systems all use the tubular design shown in Fig. 3.29 or disk array designs of a similar complexity. Measured degradation rates for current planar cell prototypes are about 1.7% per 1000 h of operation (Borglum, 2003), which is an order of magnitude too high.

A number of SOFC prototype plants (in the 100-kW size range) are in operation. The current direct conversion efficiency (i.e., neglecting system auxiliaries) is about 55%, but could reach 70−80% in the future, at least for the high temperature range. System power efficiencies are currently around 45% and around 75% for SOFCs producing both power and heat.

The efficiencies quoted are based on the thermodynamical efficiency (3.22), corrected for voltage losses (see (3.23) and the discussion of (3.66)) and for fuel passing to exhaust without being reacted (Fig. 3.34).

SOFCs share with other fuel cell types a very low tolerance to fuel impurities of H2S (< 1 ppm), NH3 (< ½%), HCl, and other halogens (< 1 ppm), but in contrast to the low-temperature cells accept fuel impurities of CO in addition to the methane and hydrocarbons reformed internally (Dayton et al., 2001).

(%)	Natural gas (pipeline)	Coal (O2 blown)	Biomass (steam/air blown)
H2O		27—62	0—40
H2		38—42	15—21
N2	0.06	0.3—0.8	0—40
CO2	1.3	23—31	13—22
CO		15—37	11—43
CH4	88.1	0.1—9.0	11—16
CnHm	10.4	¿0.8	0.1—5.0
H2S	10 ppm	0.2—1.3	0.01—0.1
NHx	0.3	0.3—0.8	0.1—0.4
Tars		¿0.24	0.3—0.4
HCl		200 ppm

Table 3.2. Composition of potential SOFC fuels derived from North Sea natural gas (vol%) and from gasification of coal and biomass (mol%). Except for water, the mol% given is on a dry basis. Steam gasified biomass is at low pressure, and the air blown is at high pressure. Based on Fontell et al. (2004) and data collected by Dayton et al. (2001).

Table 3.2 shows typical compositions of potential fuels other than pure hydrogen. It is seen that gasified coal and biomass both fail to meet the criteria regarding poisoning by the mentioned poorly tolerated substances. The same is the case for MCFCs and phosphoric acid cells, while proton exchange membrane fuel cells do not tolerate CO in the fuel stream. The implication is that if solid oxide fuel cells are not used with hydrogen, there has to be additional cleaning operations such as the sulphur removal (see Fig. 3.33) in the case of natural gas or higher hydrocarbons, and in the case of coal or biomass based fuels removal by some cleaning process of both sulphur and halogens, plus possibly ammonia. Simplification by mixing CH4 fuel and O2 in a single gas channel has been investigated (Hibino et al., 2003).

Figure 3.38. Scanning electron microscope picture of the electrode-electrolyte structure along a perpendicular cut. Top: screen-printed La0.6Sr0.4Co0.2Fe0.8O3 positive electrode. Middle: spray-deposited electrolyte, YSZ = 8 mol% Y2O3 stabilised ZrO3. Bottom: negative electrode, NiO and YSZ in ratio 7:3, in cermet CeO2. (From D. Perednis and L. Gauckler (2004). Solid oxide fuel cells with electrolytes prepared via spray pyrolysis. Solid State Ionics 166, 229-239. Reprinted by permission from Elsevier.)

Modelling of gas flow and cell performance with inclusion of chemical reactions and heat transfer between components has been performed for more industrially relevant assemblies of SOFC tubes, such as the one shown in Fig. 3.39 (Colclasure et al., 2011). A similar design has been studied by a three-dimensional CFD (computational fluid dynamics) model coupled to a one-dimentional tube model by Kattke et al. (2011), obtaining complicated temperature and oxygen distributions such as those illustrated in Fig. 3.40. For planar cell structures, the start-up transient behaviour has been modelled (Colpan et al., 2010), and density functional theory has been applied to study of the relative merits of different catalysts on the positive electrode, similarly to the simple calculations presented in section 3.1.4, but for bimetallic catalysts made from Ni-metal alloys suitable for operating at the high temperatures relevant for SOFC cells (An et al., 2011).

Figure 3.39 (left). Design of a multi-tubular SOFC system, indicating gas flows (Colclasure et al., 2011; reproduced by permission from Elsevier).

Figure 3.40 (below). Model results for temperature and oxygen distributions in upper (left panel) and lower (right panel) part of tubes. In each panel, the left part depicts °C temperature (from periphery to centre), the right part similarly mole fraction of oxygen (Kattke et al., 2011; reproduced by permission from Elsevier).

One key feature of the SOFC is its ability to auto-reform fuels. This means that impure hydrogen, hydrocarbon fuels and methanol can often be accepted directly. Auto-reforming is facilitated by high temperatures and this property may be lost in efforts to lower the operating temperature of SOFCs. Eveloy (2010) discusses a problem of carbon deposits on the electrodes, if the otherwise desirable Ni-based catalysts are employed in SOFCs aimed at using methanol as a fuel.

For many fuel cell applications, it would be desirable to employ SOFCs operating at a lower temperature than that of current prototypes. One way of achieving this is the use of ceria-carbonate composites as electrolytes (Wang et al., 2011). As all other fuel cells, reverse operation is possible (sometimes denoted SOEC in case of SOFCs, "E" standing for electrolyser, but there is no compelling reason for introducing a new acronym). Again, the high operating temperature implies that such electrolysers may be quite efficient (Jin and Xue, 2010). However, as with other fuel cell types, the crucial question is if the same device can be used with high efficiency in both direction, or if optimisation for hydrogen and electricity production has to be different.

Several studies have been made on the lifetime and degradation of SOFC systems (Nagel et al., 2009; Zhang and Xia, 2010). Particularly sulphur has a negative effect on electrodes, which makes the use of hydrocarbon fuels problematic. The current phase is still one of trying out many new concepts for SOFC design without having a clear "winner". For instance, a particular effort has been made to design systems that will accept fuel input at low pressures (Shi and Xue, 2010).

3.4 Acid and alkaline cells

Phosphoric acid cells have been developed for stationary uses. They employ porous carbon electrodes with a platinum catalyst and phosphoric acid as electrolyte and feed hydrogen to the negative electrode and oxygen (or air) to the positive electrode, with the basic electrode reactions given by (3.15) and (3.16). The operating temperature is in the range of 175-200°C, and water is continuously removed. Several 200-kW installations have been operating for a number of years for emergency power at hospitals and military facilities.

Fluid acid electrolytes are generally good conductors with reasonable levels of stability (as witnessed by the 40 000 h operation between major overhauls achieved by some operating plants according to King and McDonald, 2003). Corrosion is a problem at the electrodes and is the reason for having to use catalysts based on noble metals such as Pt on electrodes made of porous graphite paper. However, even platinum catalysts experience degradation, suggested to arise from migration along the carbon surface upon which the Pt molecules are adhered (Aindow et al., 2011). The amount of graphite is reduced to 20% of the original after 40 000 h of operation (Kordesch and Simader, 1996). The use of H3PO4 as the electrolyte, rather than other conventional fluid acids, is due to considerations of low evaporation and stability at the temperatures of 150-200°C suitable for operation of the phosphoric acid fuel cell (PAFC). If the fuel is based on reformed natural gas, it would contain CO2 at a level of typically 20%, which is acceptable for the PAFC reactions to proceed at an overall efficiency of around 40%. The cell voltage versus current density relationship is similar to the ones shown in Fig. 3.37 for SOFCs, but maximum voltages are slightly lower and drop as fast as the

SOFC curve for 650°C (Kordesh and Simader, 1996). Modelling of the electrochemical performance has been undertaken, for example, by Sprague and Dutta (2011). Like the high-temperature fuel cells, PAFCs need hours to start up and are thus not suited for automotive applications (Spakovsky and Olsommer, 2002). Despite sales of several hundred units, the price has remained over US\$3000/kW, and it is generally held that radical breakthroughs are required for this technology to compete with the other fuel cell types.

In general, because PAFCs are proton conductors like the proton exchange membrane (PEM) and the subcategory direct methanol fuel cells (sections 3.5 and 3.6), there is a continuous conceptual transition between them. The polymers used in PEM cells usually contain weakly acidic components such as HSO3, but may be reinforced with a stronger acid in order to increase conductivity or allow operation at higher temperatures.

Figure 3.41. Proton conduction in acid polymer PBI membrane based upon benzimidazole. The smallest molecules are hydrogen, and the polymer backbone chains at the top and bottom are mainly carbon in three benzene rings, but with two attached pentagonal rings with two nitrogen atoms each. In the middle are four phosphoric acid molecules (the 4:2 ratio of doping acid to PBI gives maximum protonation, the largest atoms being phosphor and the middle ones being oxygen). The proton transport is illustrated by transfer from a PBI nitrogen atom (A) and an acid-to-acid molecule transfer (B). If water is present, further proton transport is facilitated by reactions such as those illustrated in Fig. 3.7. The next polymer chain segments are attached at points C and D (eventually twisted by 180°).

Several suggestions have been made for adding acids to a polymer material,

while maintaining the solid structure as much as possible. One proposal adds phosphor-tungsten acid (H3PW12O40) to a mixed organic−inorganic polymer network based on an organic hydrocarbon plus inorganic zirconium structure (Kim and Honma, 2004). This material has a conductivity of around 10⁻³ A V⁻1 cm⁻¹ (slightly higher when humidity-saturated) at temperatures in the interval 100-160°C (compare, e.g., with Fig. 3.28) and good temperature stability below 200°C. A development along similar lines is the replacement of the perfluorinated membranes common in PEM cells by polybenzimidazole (PBI), for example, as a polymer chain with two benzimidazole molecules followed by an extra benzene ring (a unit segment seen at the top and bottom of Fig. 3.41). This polymer is stable between 100 and 200°C and may be doped with H3PO4 to get a strongly increased proton conductivity (Li et al., 2004). The mechanism of proton transfer is illustrated in Fig. 3.41. Laboratory fuel cells made with this technology offer high CO tolerance and need no water management, but may be operated using dry gas inputs (Jensen et al., 2004). Loss of acid molecules from the membrane may be a problem (Wang, 2003).

Another possibility is to introduce a solid acid directly as electrolyte in a cell similar to the PAFC (i.e., using a Pt catalyst and carbon materials for electrodes). Materials tried include CsHSO4 (Haile et al., 2001), CsH2PO4 (Boysen et al., 2004), and Tl3H(SO4)2 (Matsuo et al., 2004). These solid electrolytes exhibit rapidly rising conductivities, from some 10⁻⁸ A V⁻1 cm⁻¹ at ambient temperature to over 10⁻² A V⁻1 cm⁻¹ at temperatures in the range of 150- 200°C. This leads to cell power levels of about 40 mW cm⁻² or at least 5 times lower than those of conventional direct methanol fuel cells (see section 3.6), but 25 times lower than some of the SOFC power densities in Fig. 3.37, with methanol cells being one of the competing technologies over which the solid acid cells have the advantage of not requiring careful water management. Stability problems have been reduced by initial water pressure cycling.

Alkaline fuel cells (AFCs) use an aqueous potassium hydroxide (KOH) solution (around 30%) as electrolyte and have electrode reactions of the form

$$H_2 + 2OH^- \rightarrow 2H_2O + 2e^-,$$ (3.69)

$$\frac{1}{2}O_2 + H_2O + 2e^- \rightarrow 2OH^-.$$ (3.70)

These cells operate in the temperature range of 70-100°C, but the specific choice of catalysts (e.g., Pt or Ni) requires maintenance of a fairly narrow temperature regime. Also, the hydrogen fuel must have a high purity and notably not contain CO2, due to the alkaline pH prevailing. Carbon dioxide reacts with OH⁻ to form carbonate (K2CO3), which reduces ion generation and transport within the electrolyte. There is some dispute over the importance of this effect, and a recent paper has suggested that alkaline fuel cells degrade equally rapidly with and without CO2 impurities in the fuel (Gülzow and Schulze, 2004). The likely explanation is that CO2 poisoning is very important with the small amounts of KOH present in fixed matrix configurations, while the currently preferred design (Fig. 3.42) with KOH being circulated through the cell (in order to be able extract the extra water generated by the reactions (3.69) and (3.70) and evaporate it) diminishes the problem (Gouérec et al., 2004). However, using large quantities of circulating electrolyte also has implications for the physical size of the system, which is important, for example, for automotive uses. The alternative of cleaning the electrolyte for carbonates has not been put in practical use and would add to bulkiness and cost.

Figure 3.42. Schematic layout of an alkaline fuel cell.

Alkaline fuel cells have been used extensively on early spacecraft until they were superseded by more reliable solar cells. The high cost of the space cells and the use of corrosive compounds requiring special care in handling have been held against AFCs. Current AFC development employs multi-component electrodes using Ni for structural stability and as catalyst, carbon black as electron conductor and polytetrafluoroethylene (PTFE) pore-forming powder for gas diffusion and water repulsion (hydrophobic quality, Fig. 3.42).

General objections to liquid electrolytes include corrosion and difficulty in minimising physical volume requirements for applications such as vehicle systems. The lifetime of AFCs are of the order of 5000 h with a large spread in experience (McLean et al., 2002). The electrolyte penetrates into the electrode pores and degrades functionality by increasing diffusion paths (Cifrain and Kordesch, 2004). Extending the lifetime requires that KOH be drained out of the system when not in use.

The individual components used in an AFC are not necessarily expensive compared to those of other fuel cell types under development. Use of Pt catalysts can conceivably be avoided, while the bipolar plates collecting the electron flows typically have to be made of fairly expensive black carbon to avoid corrosion. The peripherals needed for water management and electrolyte draining add to the cost, but do not necessarily lead to drawbacks such as long start-up times if proper controls are part of the system design. Process heat is used to evaporate water in the electrolyte circulation cycle and would have to be recovered by condensation for energetic efficiency.

The energy efficiency of the AFC itself is similar to or slightly better than that of other low-temperature cells, lying in the range of 45-60%, with an open circuit cell voltage around 0.9 V and a current density of 0.2-1.0 A cm⁻² , highest for space cells (Jo and Yi, 1999; Spakovsky and Olsommer, 2002; McLean et al., 2002).

Mass production cost of AFCs is quoted as US\$400-500/kW (Gülzov, 1996) and in another study as US\$155-643/kW (McLean et al., 2002), which compares favourably with the estimated cost of proton exchange membrane (PEM) cells (US\$60-1220/kW), except for a lower expectation of options for cost reduction by mass production. The difference in lower bounds may be decreased if peripheral costs are added, for reasons of, for example, water management costs for the PEM cells. However, as just mentioned, similar or even more elaborate water and electrolyte management costs might have to be added to AFC modules for durability and efficiency in practical applications. The lifetimes of prototype AFCs are fairly low for stationary uses, but are considered acceptable to mobile applications, where the bulkiness of the system is a more likely limitation. Cost comparisons are difficult to make because all systems have too high costs at present, and the decisive parameters have to do with the possible advantages obtained by mass production and reductions in material use, which by many observers is judged to be less probable for AFC technology than for PEM fuel cell technology. Variations on the alkaline reaction scheme (3.70) at the positive electrode are the possible creation of hydroperoxide,

$$O_2 + H_2O + 2e^- \rightarrow OH^- + HO_2^-.$$ (3.71)

This is an example of the use of alkaline fuel cells for production of industrially interesting compounds rather than for electricity production (Alcaide et al., 2004).

3.5 Proton exchange membrane cells

The type of fuel cell recently undergoing the most rapid development is the proton exchange membrane (PEM) fuel cell¤ . It has been developed over a relatively short period of time, and it is hoped to be able to serve economic application in the transportation sector. It contains a solid polymer membrane sandwiched between two gas diffusion layers and electrodes. The membrane material is typically polyperfluorosulphonic acid. A platinum or Pt−Ru alloy catalyst is used to break hydrogen molecules into atoms at the negative electrode, and the hydrogen atoms are then capable of penetrating the membrane and reaching the positive electrode, where they combine with oxygen to form water, again with the help of a platinum catalyst. The allsolid state design makes the cells compact and suitable for stacking (Fig. 3.43). Figure 3.44 gives a schematic view of the assembly.

Figure 3.43. Layout of a PEM fuel cell layer, several of which may be stacked. Electrode areas include gas diffusion and electrode components in a grid-type structure (see Fig. 3.33) (from B. Sørensen, Renewable Energy, 2004, used by permission from Elsevier).

^¤ In recent years, some public research programmes have used the term "polymer electrolyte membrane", which happens also to be abbreviated as "PEM".

Figure 3.44. Schematic picture of a proton exchange membrane fuel cell. Modelling of reactions at the gas diffusion layer/catalyst/membrane interfaces A and B is discussed in section 3.5.2. Details of design are discussed in the following subsections.

Figure 3.45. Expected part-load efficiencies for a 50-kW PEM fuel cell system, projected from measurements involving 10-20 cell test stacks (Patil, 1998).

The electrode reactions are (3.15) and (3.16), and operating temperatures are

in the range of 50-100°C. Protons (H+ ) travel through the membrane material. Figure 3.44 shows a typical layout for an individual cell. Due to a low operating temperature and flexibility in design, proton exchange membrane (PEM) fuel cells can be used for a range of applications from portable over vehicular to general power supply. PEM cell stacks are dominating the current wealth of demonstration projects in road transportation and dispersed building-integrated areas of use. The efficiency of conversion for these systems is between 40 and 50%, and lifetimes are under 5000 h with intense efforts to improve stability. As indicated in Fig. 3.45, an advantage of particular importance for automotive applications is the high efficiency at part loads, which alone gives a factor of two improvement over current internal combustion engines.

In the following subsections, each component of the PEM fuel cell will be described, and modelling and empirical experience of the overall system will be presented, as well as stability and durability evaluations.

Figure 3.46. Two design options for the bipolar electrode¤ plates of a PEM fuel cell.

3.5.1 Current-collectors and gas delivery system

The mechanical stability of a PEM fuel cell is usually provided by a set of bipolar plates, which also act as collecting terminals for the electric current generated and which are further forming the walls of the access flow channels for hydrogen and oxygen (air), as well as the outlet channels for water and excess gases. Additionally, in connection with the plate configuration, coolant flow (e.g., water) has to be considered, as indicated in Figs. 3.44 and 3.46. The plates may be made of suitable metals or graphite. In case metal is selected, a corrugated plate design (Fig. 3.46A) is a natural choice, while for graphite, a machined structure as shown in Fig. 3.46B may be considered (Wilkinson and Vanderleeden, 2003). Metals are subject to corrosion in the

^¤ I use the term "electrode" for the electron conductors. A different terminology is also in use, where the name "electrode" is reserved for the catalyst and gas diffusion layer assembly.

fuel cell chemical environment, and many metals must be coated to protect them from decay. Titanium has been considered, but cost speaks against it. While most metals have both adequate conductivity and high mechanical strength, graphite has only fair conductivity and poor mechanical strength. It is therefore used in composites, for example, with a polymer resin as support, which also allows carbon powder to be used instead of solid graphite.

The gas diffusion layer (GDL) must be capable of transporting gases (hydrogen or oxygen) from the gas input channels to the active area at the catalyst-membrane interface. At the same time, it must be able to transport electrons to or from the active area and deliver them to or take them from the bipolar plates connected to an external circuit. In other words, the pore structure should be such that there are contiguous gas channels through the material, as well as contiguous electron conducting wall elements.

Figure 3.47. Different design options for the gas flow channels: (A) straight, (B) interdigitated, (C) serpentine, and (D) spiral. For the interdigitated design, the incoming flow must proceed through the gas diffusion layer to reach the outlet channel.

The design of gas inlet and outlet channels offers some potential to control cell behaviour. Figure 3.47 illustrates some of the possibilities. The serpentine (C) and spiral (D) designs used with a higher pressure than the conventional straight channel design (A) are capable of reducing the amount of liquid water accumulating in the gas channels, and the interdigitated design (B) goes one step further by forcing the flow to pass through the gas diffusion layer to the adjacent channel, thereby helping to clean the gas diffusion layer for accumulating water from the reaction (3.16) or from the water initially soaking the membrane (as a condition for its proton conducting property). An analysis of the designs has been made by Sierra et al. (2011).

The co-ordinates indicated in Figs. 3.46 and 3.47 and used in Figs. 3.48-3.51 and Figs. 3.48-3.49 are defined such that the y-axis is in the perpendicular direction (to the paper plane in Fig. 3.47) through the sequence hydrogen inlet channel → gas diffusion layer → negative electrode catalyst → membrane → positive electrode catalyst → gas diffusion layer → oxygen (air) channel. The x-axis is along the main gas channel flow directions, and the *z*axis is across the design as shown in both figures.

Figure 3.48. Computed velocity fields (m/s) for flows in adjacent interdigitated oxygen channels (with gas diffusion layer on the left side) of a PEM fuel cell (A: inlet, C: outlet, in x-y plane). In the middle (B), the flow from one gas channel through the gas diffusion layer to the adjacent gas channel is shown in the z-y plane, for the midpoint of the cell extension in the z-direction. The flows in A and C are for the midpoint value of z. (From S. Um and C. Wang (2004). Three-dimensional analysis of transport and electrochemical reactions in polymer electrolyte fuel cells. J. Power Sources 125, 40-51. Used with permission from Elsevier.)

Figure 3.48 shows the flow field in two adjacent oxygen channels and a bridging piece of the gas diffusion layer, from simulation of an interdigitated design with use of a three-dimensional fluid dynamics model (as discussed in section 3.1.5) including modelling of the electrochemical reactions. The left part (A) of Fig. 3.48 shows that the gas flow (as expected) is slowed to zero at the dead end of the channel. The middle part (B) of Fig. 3.48 shows the behaviour in the y-z plane, highlighting a strong and asymmetric flow around the bridge caused by forced convention (in contrast to the straight channel design in Fig. 3.47A, for which penetration into the gas diffusion layer is by diffusion and where the flow field therefore is symmetric around the mid-point current collector). The right-hand side (C) of Fig. 3.48 shows the flow in the uptake channel, which is building up along the x-axis and is largest at the end of the channel.

Figure 3.49. Oxygen concentrations in two adjacent outlet flow channels of an interdigitated PEM fuel cell structure, with gas diffusion layer on top. The three planes are at x-values corresponding to the flow channel exit (a: large x), the middle (b), and the entrance (c: x = 0). The cell current is at its maximum (about 0.8 A cm⁻² ). (From M. Hu et al. (2004). Three-dimensional, two-phase flow mathematical model for PEM fuel cell: Part II. Analysis and discussion of the internal transport mechanism. Energy Conversion & Management. 45, 1883-1916. Used with permission from Elsevier.)

In addition to improving water management, the interdigitated design causes higher currents at lower voltages, because the limiting factor is mass transport in the positive electrode gas diffusion layer (Um and Wang, 2004).

Figure 3.49 shows the variations in oxygen concentration from a similar fluid dynamics calculation, also for two adjacent oxygen channels of an interdigitated design and the bridging gas diffusion layer, where a substantial reduction in oxygen concentration is seen to take place. At the exit of the outgoing channel, 34% of the incoming oxygen has been removed to the electrochemical reaction at the catalyst. Adding more channel pairs in the flow path design of Fig. 3.47 allows larger fractions of the oxygen to become processed.

3.5.2 Gas diffusion layers

In order for the gas diffusion layers (GDLs) to handle the hydrogen and oxygen flows, respectively, water from the recombination (3.16) must be led to the exit flow channels. This water is created at the oxygen side membrane to the gas diffusion layer interface, but it may find it easier to move through the membrane to the hydrogen side than to penetrate into the oxygen gas diffusion layer. This means that the membrane will get soaked even if it was dry at first (although in that situation, the H+ penetration would be reduced) and hence that the functionality of the membrane sheet will build up. Whether the reaction water will leave through the oxygen or hydrogen side channels is therefore difficult to predict intuitively. Figure 3.50 presents the results of the model calculation also underlying Fig. 3.49, showing the calculated water concentrations (for three x-values) in the two adjacent hydrogen channels and the corresponding gas diffusion layer (top of each pair) and for the two oxygen channels on the other side of the cell, with their diffusion layer. In the hydrogen channels, water concentrations are high, indicating that much water has crossed the membrane after its creation at the oxygen catalyst by (3.16), back to the gas diffusion layer on the hydrogen side and reaching the hydrogen outlet channel.

For the straight channel geometry (Fig. 3.47A), water concentration declines uniformly when progressing along the channel length x, both in the hydrogen and in the oxygen channels, with liquid water forming in the gas diffusion layer on the oxygen side and eventually (for larger x) showing up in one of the gas channels (Wang et al., 2001; Yu and Liu, 2002; Hu et al., 2004). For the interdigitated design, Hu et al. (2004) find condensed water only on the hydrogen side when the current density is low (≈0.17 A cm−2), but on the oxygen side when the current density is high (≈0.8 A cm−2), as shown in Fig. 3.51. In the first case, water is transported across the membrane to the hydrogen side, while, in the second case, the high reaction rate creates a need to replenish water in the membrane, thus removing water from the hydrogen side, while at the oxygen side, the high level of water generation by (3.16) exceeds what can be absorbed by the membrane, so that a surplus appears in the oxygen channel.

Figure 3.50. Water concentration in an interdigitated PEM fuel cell structure, for three planes at x-values corresponding to the flow channel exit (a: top left), the middle (b: top right), and the entrance (c: bottom left). In each pair of pictures, the y-z plots depict the hydrogen side at the top (GDL is lower bar) and the oxygen side at the bottom (GDL is upper bar). The cell current is at its maximum (about 0.8 A cm⁻² ). (From M. Hu et al., (2004). Threedimensional, two-phase flow mathematical model for PEM fuel cell: Part II. Analysis and discussion of the internal transport mechanism. Energy Conversion & Management. 45, 1883-1916. Used with permission from Elsevier.)

Figure 3.51. Saturated liquid water distributions (continued).

Figure 3.51 (continued). Saturated liquid water distributions for cell with interdigitated channel design, at low current density (≈0.17 A cm⁻² , left column: a, c) and at high current density (≈0.8 A cm⁻² ; right column: b, d). Two planes are depicted, at *x*values corresponding to the flow channel exits (a: top left, b: top right) and to the middle (c: bottom left, d: buttom right). Each of the four pairs of pictures depicts y-z plots for the hydrogen side at the top (GDL is lower bar) and the oxygen side at the bottom (GDL is upper bar). (From M. Hu et al., (2004). Three-dimensional, two-phase flow mathematical model for PEM fuel cell: Part II. Analysis and discussion of the internal transport mechanism. Energy Conversion & Management. 45, 1883-1916. Used with permission from Elsevier.)

The water profiles show a dependence on pressure (Futerko and Hsing, 2000), and most PEM cells have been found capable of recovering from water flooding (Nguyen and Knobbe, 2003). Recently, direct views of the water distribution have been obtained by neutron imaging techniques, taking advantage of the different neutron cross-sections of different atoms (Satija et al., 2004). Like the differences between straight and interdigitated cell designs seen here (see Fig. 3.48), the water handling patterns are also different for serpentine designs (Nguyen et al., 2004).

Materials commonly used for the gas diffusion layers are carbon paper or woven carbon mats (examples of which are shown in Fig. 3.52). They combine the connectivity allowing electron transport with a pore structure suitable for hydrogen or oxygen gas access to the catalyst layer. In cell manufacture, the catalysts may be deposited either on the gas diffusion layer or on the membrane.

Figure 3.52. Structure of carbon paper (left) and carbon cloth (right) used for gas diffusion layers in PEM fuel cells. A coating of 20% (by weight) fluorinated ethylene propylene has been applied. (From C. Lim and C-Y. Wang (2004). Effects of hydrophobic polymer content in GDL on power performance of a PEM fuel cell. Electrochimica Acta 49, 4149-4156; G. Lu and C-Y. Wang (2004). Electrochemical and flow characterization of a direct methanol fuel cell. J. Power Sources 134, 33-40. Used with permission from Elsevier.)

Because of the importance of the gas diffusion layer for the performance of a PEM fuel cell, notably considering the negative effects of flooding the layer with water, simulation and theoretical investigations have been performed to a considerable extent, in order to identify the best operation conditions, in cases of different design of gas channels and different choice of the material (such as the one shown in Fig. 3.52) to use. The macroscopic modelling of Figs. 3.50 and 3.51 has been performed in three dimensions (Cordiner et al., 2010) and has been taking into account the micro-porosity of the carbon paper or cloth material (Nishiyama and Murahashi, 2011; Kopanidis et al., 2011). From the latter source, Fig. 3.53 shows for a gas diffusion layer based on carbon cloth (see Fig. 3.52, right) the temperature and humidity distributions along the gas channel, for a specific case of fairly high gas velocity, selected among several cases studied.

Figure 3.53a, b. a (right): Temperature distribution, increasing along gas diffusion layer in the carbon cloth. The foreground bipolar plate walls are not shown, except for end contours. b (below): Maximum local relative humidity at gas velocity 143 ms-1, temperature 353 K and an inlet relative humidity of 97%.

Reproduced from Kopanidis et al. (2011), by permission from Elsevier.

Following the water transport through the membrane area, as well as in the gas channel and electrode regions, provides new insights into the options for design optimisation (Berning et al., 2011). Advanced experimental techniques such as microstructure optical inspection of semi-transparent model cells have added to the understanding of flow in a PEM fuel cell (Bazylev et al., 2011). Electric current investigations have progressed along similar lines (Carcadea et al., 2007). An overview of some recent studies on the micropore level may be found in Shah et al., (2011).

In Figs. 3.54a to c, cathode channel water content and current density are shown for an interdigitated cell design (see Fig. 3.47B), based on CFD calculations. It is seen that an uneven distribution is emerging, presumably as a result of pressure changes at channel bends, and due to flow impediment when encountering assemblies of water (Le and Zhou, 2009, with more examples given).

Figure 3.54a, b, c. For the PEM fuel cell serpentine geometry in a (indicating gas velocity vectors), the simulated water content and current density distributions are shown in b and c, for a position (y-value) at the negative electrode gas diffusion layer-catalyst interface. All three plots correspond to a definite point in time during the simulation (Le and Zhou, 2009). Used with permission from Elsevier.

3.5.3 Membrane layer

The membrane layer consists of a polymer structure capable of transporting hydrogen ions with high conductivity, inspiring the name "proton exchange membrane". The membrane is thus a solid electrolyte. For the current emphasis on PEM fuel cells operating below 100°C, the Nafion® (trademark of Dupont de Nemours) or analogous perfluorosulphonic acid membranes have been the dominating choice. The structure of the repeat structure of the polymer fluorocarbon backbone and a side chain with sulphonic acid ends upon which Nafion is based is shown in Fig. 3.55 (the commercial product is sold with various thicknesses and dimensions denoted by a number code such as "Nafion-117", related to non-SI units). The membranes should have high protonic conductivity, low gas permeability, and, of course, a suitable mechanical strength and low temperature sensitivity.

Figure 3.56. Ambient humidity Nafion-115 structure from small-angle X-ray spectroscopy. Lighter areas are cluster structures in the material. (Reprinted with permission from J. Elliott, S. Hanna, A. Elliott, G. Cooley (2000). Interpretation of the small-angle X-ray scattering from swollen and oriented perfluorinated ionomer membranes, Macromolecules 33, 4161-4171. Copyright American Chemical Society.)

Figure 3.55. Segment of a perfluorinated sulphonate ion exchange membrane of Nafion® type. The top polymer chain is made from C atoms as backbone and pairs of F atoms on the sides (end hydrogens are replaced when the chain is enlarged). The acid HSO3 molecule promoting H transport is attached at the end of an arm of 4 CF2 molecules interspersed with two O atoms, and with one F replaced by a CF3 group.

Spectroscopic studies of the structure of Nafion membranes suggest the presence of nodules of diameter 1-10 nm (the lighter parts in Fig. 3.56). The nodules are fairly regularly distributed and explained as spherical or oblong extrusions (with radius around 2 nm) of a pair of (CF2)n sheets, that is, a double layer. Some models have the acidic side chains on the outside of the membrane bi-layer (Gierke and Hsu, 1982), while other models place the SO3 ions on the inside of the spheres (Vankelecom, 2002). It has not so far been possible to carry out full quantum chemical calculations confirming the tendency of such structures to "fold up" into a nodular structure. However, molecular dynamics calculations (see Chapter 2, Fig. 2.3) have been made (Jang et al., 2004) for systems containing four Nafion chains of 70 CF2s and 10 side chains, together with 560 water molecules and 40 hydronium molecules (H3O+ ). Figure 3.57 shows results from this study, indicating a structure with clusters of curled-up Nafion backbone structures with the side chain S-atoms close to both water and hydronium molecules. The figure assumes a Nafion structure similar to Fig. 3.55 with side chains every 7 units. If the side chains are lumped at fewer places, the backbone nodules become larger than in Fig. 3.57. The average distance between pairs of S atoms in the case shown in Fig. 3.57 is 0.68 nm, with a rather narrow distribution (spread about ±0.2 nm).

Figure 3.57. Molecular dynamics model calculation of Nafion-117 (for small system, left, and large system, right). The Nafion backbone structure is omitted and therefore shows up as white areas. Large, light grey atoms are S, and large dark ones are the O's of hydronium molecules. (Reprinted with permission from S. Jang, V. Molinero, T. Cagin, W. Goddard III (2004). Nanophase-segregation and transport in Nafion 117 from molecular dynamics simulations: Effect of monomeric sequence. J. Phys. Chem. B108, 3149-3157. Copyright American Chemical Society.)

The molecular dynamics simulation behind Fig. 3.57 does not consider double polymer layers of the kind suggested for the nodule structure, and it does not include the hopping process of proton transfer between the water molecules (as in Fig. 3.7), because the water and hydronium molecules are treated as fixed entities. The stability and regularity of the polymer membranes is also poorly explained by the "loose" structure of Fig. 3.57. An alternative modelling approach is to depart from a double Nafion membrane structure such as the one shown in Fig. 3.58. The distance between the two membrane layers is initially taken as 1.8 nm, corresponding to the thin channels identified experimentally (Barbi et al., 2003). The two sheets would have to be pulled away from each other to roughly the double distance in order to form the observed nodules of diameter around 4 nm. The side chains have in Fig. 3.58 been assumed to face the inter-membrane volume, but could alternatively face outward. The sulphuric acid terminals are already in the double-membrane structure facing "empty" areas of the structure, where water molecules might be placed to help the H+ transport, but pulling the side chains apart would furnish a more continuous route for transfer. Proton transport has been studied in isolation, using a model including one acid side chain and a number of water molecules (Paddison, 2001).

Figure 3.58. A 2 × 3 unit cell chunk of a possible structure for the Nafion doublemembrane channels in the absence of nodule enlargement. The structure has been loosely optimised by 50 iterations of a Hartree-Fock calculation (Sørensen, 2005a).

The reason for attempting to explore the structure by theoretical means is that the small-angle X-ray and nuclear magnetic resonance scattering experiments are not very selective techniques for structure determination, and interpretations of results such as those shown in Fig. 3.54 involve a choice between a multitude of possibilities. Alternative guesses were made earlier by M. Ise (quoted in Kreuer, 2001), suggesting like Jang et al. (2004) that the membrane bundles observed are disorganised chunks of (CH2)n and that all acid and water molecules are placed outside the clusters. I find the acidinside and acid-outside structures to have identical energies. The inconsistency of some literature findings is pointed out by Barbi et al. (2003). They use their own X-ray experiments to confirm the magnitude of the nodule radius found earlier (they find 1.9 nm) and to determine an average distance between domains (3.6 nm), but refrain from further guessing.

Figure 3.59. Conductivity of Nafion-115 as a function of humidity (partial pressure of water relative to saturation pressure, the latter corresponding to 18% water by weight or 11 water molecules per sulphonate molecule). (From C. Yang, S. Srinivasan, A. Bocarsly, S. Tulyani, J. Benziger (2004). A comparison of physical properties and fuel cell performance of Nafion and zirconium phosphate/Nafion composite membranes. J. Membrane Science 237, 145-161. Used with permission from Elsevier.)

Figure 3.59 shows that dry Nafion is a poor conductor and that operating the fuel cell membrane close to full water saturation is essential. Proton conductivity increases modestly with temperature in the range of 20-80°C, that is, below the values shown in Fig. 3.59 (Gil et al., 2004).

The membrane water balance and conductivity were also studied by the model behind Figs. 3.49-3.51. The water model included electro-osmotic drag due to the H3O+ ion transport (Springer et al., 1991) and diffusive processes. Figure 3.60 shows the results for the two cell designs A and B in Fig. 3.47, at high current densities. For the straight gas channel design (left), the water content is high on the oxygen side, but low on the hydrogen side, due to the fact that the calculation was started with a dry membrane and water creation is at the oxygen side, combined with the water transport to the oxygen side caused by the drag from the positive H⁺ ion movements. At low currents, the water content would be highest on the hydrogen side. For the interdigitated design (Fig. 3.60, right), the more complex water distribution varies across the membrane plates (z-direction) and the oxygen side content is high on the outflow side of the gas channel, whereas the content on the hydrogen side is higher at the inflow side of the gas channel. The cell processes will start even with low initial water content and will saturate the membrane at high power output, causing membrane resistance to increase in the straight design but less so in the interdigitated design.

Figure 3.60. Distributions of water content across the membrane of a PEM cell with either straight (left) or interdigitated (right) channel design, at maximum current density (≈0.8 A cm⁻² ). The membrane thickness is 0.16 mm. The figures correspond to an x-value at the middle of the gas flow channels. The y-z co-ordinates at the floor of each picture have the y-axes in opposite directions in order to better display the different behaviour in the two cases: the left-hand picture has the oxygen side (y = 0) at the back, while the right-hand picture has the oxygen side in front. Also, the z coordinate across the cell is opposite in the two pictures, corresponding to them representing views from either the front or the back side of the cell. (From M. Hu et al. (2004). Three-dimensional, two-phase flow mathematical model for PEM fuel cell: Part II. Analysis and discussion of the internal transport mechanism. Energy Conversion & Management. 45, 1883-1916. Used with permission from Elsevier.)

Figure 3.61. Distributions of local current density along the membrane of a PEM cell with either straight (left) or interdigitated (right) channel design, at maximum average current density (≈0.8 A cm⁻² ). The membrane thickness is 0.16 mm. The figure coordinate system is x-z (see Fig. 3.36) rather than y-z as used in several previous figures. (From M. Hu et al. (2004). Three-dimensional, two-hase flow mathematical model for PEM fuel cell: Part II. Analysis and discussion of the internal transport mechanism. Energy Conversion & Management. 45, 1883-1916. Used with permission from Elsevier.)

The current density along the direction of gas flow direction (x) and the z coordinate across the membrane surface is shown in Fig. 3.61, for the cell designs A and B in Fig. 3.47 and for a high average current density. In both cases the current density decreases somewhat along the x-axis, as fuel becomes depleted, and as a function of z, the distribution is symmetric for the straight design, but favouring the inlet side of the gas channel (z = 0) for the interdigitated design, and as a whole showing less variation with the x coordinate. Similar studies have been made by Sivertsen and Djilali (2005).

Alternatives to Nafion membrane materials are investigated, partly in the hope to reduce cost and partly in consideration of the advantages of increasing operating temperature by modest amounts (50-100 K) in order to obtain more efficient operation. Key objectives are to produce a thin and flat membrane without compromising on gas crossover. One such membrane is Goreselect (trademark of W. L. Gore & Assoc. Inc.), which is only 20-30 µm thick (as compared to 160 µm for the Nafion membrane studied in Fig. 3.60) and yet has high tensile strength and a potentially very small H2 crossover (Nakao and Yoshitake, 2003). It is based on a porous, helical-structure polytetrafluoroethane (PTFE) backbone and a perfluorinated ionomer side chain.

Polyether ketones (PEEKs) are studied, but generally do not give performance advantages over Nafion, although cost may be lower. Polyphenylene sulphonic acid membranes may improve performance at low humidity, and fluorinated polyarylene ethers may be produced at lower cost, but so far have low conductivity. Also, membranes based on bacterial cellulose with added precious metals have been investigated. Generally, hydrocarbonbased materials have lesser mechanical strength than the C-F based materials (Evans et al., 2003; Gil et al., 2004; Lee et al., 2004).

For use at higher temperatures, the polybenzimidazole (PBI)-type membranes (see Fig. 3.41 in section 3.4) are considered. They have to be "doped" with acid at the order of around 10 mol per kg of solvent, and although the proton conductivity increases with temperature up to about 450 K, it is still slightly lower than that of Nafion membranes. Tensile strength diminishes with doping level. Among the advantages of PBI are high CO tolerance and simpler water management, because all water is in gas form at operating temperatures of 150-200°C (Schuster et al., 2004; Li et al., 2004).

Membranes are subject to degradation from peroxide (H2O2) that may be formed within the H2SO4-H2O-H3O+ environment, from temperatures above the normal operating range and from a number of trace metal ions such as Fe2 + , Fe3 + , Cr3 + or Ni2 + , possibly coming from metallic end-plates used in the cell construction. With time the membrane loses sulphonic acid, particularly at the negative electrode side, and small amounts of fluoride ions and CO2 may be found in the product water. Several of the mechanisms are poorly understood. Some studies suggest that degradation is largest during low power production and small at full and steady power output. A mixture of enhanced ageing experiments and model estimates has been used to establish a basis for further work (Fowler et al., 2002; LaConti et al., 2003; Okada, 2003; Kulinovsky et al., 2004).

3.5.4 Catalyst action

The central part of a PEM fuel cell is the catalysts that help the basic reactions (3.15) and (3.16) to proceed at sufficiently high rates. Due to the design of the PEM cells (see Figs. 3.44 and 3.46), the catalysts are the real electrochemical electrodes¤ , receiving gases from the gas diffusion pores and delivering split molecules for ionic transport through the membrane. The conducting parts of the gas diffusion layer just carry the electrons made available to the outer circuit and its loads.

For this reason the name "electrode" is sometimes used to denote only the catalyst, while in other sources, it denotes the gas diffusion layer plus catalysts, and in still other treatments it denotes the bipolar conductors to which the gas diffusion layer delivers its electrons.

Figure 3.62. Slice of a PEM cell showing gas diffusion layer (A), catalyst layer (B), and membrane layer (C), at a magnification factor of 200 (a). Tunnelling electron microscope pictures of catalyst layer at a magnification factor of 500 (b), 18 400 (c), and in (d) 485 500. (From N. Siegel, M. Ellis, D. Nelson, M.v.Spakovsky (2003). Single domain PEMFC model based on agglomerate catalyst geometry. J. Power Sources 115, 81-89. Used with permission from Elsevier.)

Figure 3.62 shows scanning and tunnelling electron microscope (SEM, TEM) pictures of the catalyst layer sitting between the gas diffusion layer and the membrane and at increasing magnification. One notes a pore structure (1-10 µm, see Fig. 3.62b) and the presence of irregular lumps (≈3 nm, see Fig. 3.62c, d) of the Pt catalyst, when it is in a mixed environment including carbon structures offering support for the Pt as well as electron conduction, and ionomer intrusions (presumably Nafion). The white "hole" in Fig. 3.62b is from a carbon support that has been withdrawn during manufacturing. No macroscopic pores are seen on the magnified portions of Figs. 3.62c and d, indicating that all pores are at a scale above 1 µm. It is therefore inferred that hydrogen and oxygen gases initially flow through the macroscopic pores, but then have to become dissolved and to diffuse to the reaction sites at the Pt catalyst particles (Siegel et al., 2003). Since the ionomer channels indicated in Fig. 3.62c seem to fully enclose the Pt lumps, being capable of transporting H+ but not H2 or O2, this transport mechanism is difficult to understand. The membrane-electrode assembly (MEA) used in the experiment is manufactured by deposition of Pt on carbon paper followed by hot pressing onto a Nafion membrane. Possibly the observed polymer blocking of small pores in the catalyst layer is due to the sample preparation technique, or to application of an epoxy resin to fixate the sample before slicing. More remarks on the nearness of access and exit channels from the catalyst follow.

If the catalyst particles during manufacturing are deposited on the membrane, they should remain chiefly on its surface (although some mixing is caused by heat application), but if they are alternatively deposited on the carbon material of the gas diffusion layer, an amount of penetration into the gas channels is acceptable, as long as the dissociated gases can reach the membrane. All three layers must be in close sequential contact to enable transport (in opposite directions) of both electrons and protons. Early designs of the catalyst layer used polytetrafluoroethene to bind the platinum and subsequently applied a Nafion impregnation by spray techniques (some 2 mg cm⁻² ). This method required rather high percentages of Pt (some 20% by weight or over 400 µg cm⁻² ). Performance increased up to a certain Nafion loading and then decreased, presumably due to pore blocking as discussed above (Lee et al., 1998; Qi and Kaufman, 2003; review by Lister and McLean, 2004). Current manufacturing uses Nafion directly to bind platinum in a thin-film process. This ensures easy proton transport from the reaction site, but as mentioned requires care in retaining gas access.

The choice of catalyst is influenced by considerations of tolerance of contaminants in the gas flows. A pure Pt catalyst works well for extremely pure hydrogen feed, but even small amounts of CO in the H2 fuel will deteriorate performance. This problem can be reduced by using a Pt-Ru alloy as catalyst (Liu and Nørskov, 2001).

Molecular level modelling of the catalyst reactions are made in the manner described in section 3.1.4. Such quantum chemical modelling can also be performed for CO adsorbing to the Pt or Pt-Ru catalyst surface at the negative electrode, where a number of reactions may take place in addition to (3.15), describing the competition between splitting of hydrogen molecules and the conversion of CO and any water present to CO2 and protons,

H2 + catalyst surface → 2H at catalyst surface (see Fig. 2.5),

2H at catalyst surface $$\rightarrow$$ 2H⁺ + 2e^-. (3.72)

H2O at catalyst surface → OH at catalyst surface + H+ + e⁻ ,

OH and CO both at catalyst surface $$\rightarrow$$ CO₂ + H⁺ + e^-, (3.73)

with the two latter equations adding to

$$H_2O$$

CO both at catalyst surface $\rightarrow CO_2 + 2H^+ + 2e^-$ . (3.74)

The reaction (3.74) would remove CO, but quantum calculations find it energetically unfavourable (see, e.g., the calculation with just two Pt catalyst atoms performed by Narayanasamy and Anderson, 2003). However, Liu and Nørskov (2001) were able to prove that although unfavourable at either Pt or Ru surfaces, the combined presence of Pt and Ru allows OH to attach to Ru faster than to Pt, while the presence of Ru reduces the binding of CO to Pt sites and thus furthers the second reaction in (3.73), even though CO is not willing to adsorb to Ru itself.

Other alloys such as Pt with Cr or Ni have been investigated, but without any definite advantages as catalysts emerging. Karmazyn et al. (2003) looked at the behaviour of CO contamination at Pt and Ni catalyst surfaces, taking into account the quantum modelling experience of catalytic reactions often being favoured on steps in the catalyst surface structure. Steps are layer discontinuities deviating from simple surface structures with Miller indices such as (1,1,1) (see the discussion of Ni catalytic reactions with H2 in Chapter 2, section 2.1.3, with the crystal structures of Ni and Pt being basically identical), as first described by Hammer and Nørskov (1997).

Macroscopic modelling of the catalyst electrode−electrolyte interface behaviour is made by use of the Butler-Volmer equations (3.25), and this model is then incorporated into the gas flow and ion diffusion models described in sections 3.5.2 and 3.5.3.

The loading of catalyst particles influences cost, and values have been decreased over time, with the lowest ones about 14 µg/cm2 achieved by using a sputtering process (O'Hayre et al., 2002). Occasional flooding by water should be tolerated by the catalysts as well as by the other cell layers.

Figure 3.63. Oxygen dissociation at catalyst surface. a: Pt(111) surface with steps (S) and various clusters. b: O2 at face-centred site (F) and at bridge site (B), and an O-O pair under dissociation (f, h). c: two O2 molecules at face-centred sites (F). d: dissociated O atoms (f, h) with neighbouring O2 molecule. Triangles indicate Pt atoms (distance 0.277 nm). (From B. Stipe, M. Rezaei, W. Ho, S. Gao, M. Persson, B. Lundqvist (1997). Single-molecule dissociation by tunnelling electrons. Phys. Rev. Lett. 78, 4410- 4413. Used with permission from the American Physical Society.)

A new angle on the proximity needed between catalyst and the adjacent layer is offered by an alternative thin-film catalyst process requiring neither carbon fibre backup nor channels of Nafion or other ionomer proton conductors (Debe, 2003). Pt-Ru catalysts are deposited by sputtering on oriented crystalline organic whiskers (length about 1 µm), and this assembly is then transferred to a Nafion membrane surface. The voltage-current behaviour is decent and fairly independent on whisker orientation, which is taken to suggest that ionomer intrusions into the catalyst layer are not necessary for proton transport and that carbon conductors within the catalyst layer are not required for efficient electron transport. That proton transport through the void structure if possible is plausible, while the unaided electron transfer over distances up to 1 µm is more difficult to understand. Most likely, the catalyst particles sit so densely on the whiskers that they are in mutual contact, so that electrons can move within the catalyst to the surface facing the electric conductors of the gas diffusion layer.

Quantum chemical description of the catalyst action includes the dissociation processes depicted in Chapter 2, Figs. 2.5 and 2.6, and the water recombination processes discussed in section 3.1.4. Among the first realistic calculations of the H2 dissociation at the negative electrode is the work of Hammer and Nørskov (1995), followed by calculations of increasing sophistication (e.g., Penev et al., 1999; Horch et al., 1999) The behaviour at the positive electrode involves the well-studied dissociation of O2 over a catalyst surface, including the exploration of the importance of steps on the catalyst surface, just mentioned (e.g., Gambardella et al., 2001). Figure 3.63 shows a scanning tunnelling microscope picture with identification of steps as well as a specific O2 splitting process. The more complex reactions involved in the splitting (or formation) of water are under study by density functional methods, as discussed in section 3.1.4. Investigations using the simpler molecular dynamics methods have also been undertaken (Wang and Balbuena, 2004; Malek et al., 2007), as well as direct simulation of component built-up (Siddique and Liu, 2010), both confirming the importance of pore structure and sizes of catalyst-carbon agglomerates and ionomer domains.

3.5.5 Overall performance

The overall performance of PEM cells may be evaluated on the basis of the current-voltage diagram, in the same way as for other electrochemical devices (see, e.g., Fig. 3.37). Figure 3.64 gives an example of the temperature dependence of IV-curves for a PEM cell with very low Pt loading, a possibility mentioned in section 3.5.4. The power density superimposed on the figure shows that densities as high as 0.7 W cm⁻² can be achieved in this way. Similar curves published over the last decade show that it has been possible to improve the maximum power density performance gradually over the last decade, from under 0.5 W cm⁻² (Starz et al., 1999) to the present level.

Relative to the ideal thermodynamical efficiency (3.22), the practical efficiency is diminished by the electrochemical losses through the system (3.23), discussed previously for each step in the process. Furthermore, as also mentioned in the preceding subsections, not all the hydrogen fuel is utilised and there is a hydrogen content in the outflow from the fuel channel. When cells are combined to form a fuel cell stack, a fuller utilisation may be achieved by passing unused fuel from one cell to the next, but on the other hand, if the amount of fuel reaching the last part of the stack is small, less power is produced than could be accomplished with adequate hydrogen supply. Thus, the overall efficiency may by written in the form

$$\eta = \eta{ideal} \, \eta{voltage} \, \eta{fuel} \, \eta{stack'} \tag{3.75}$$

with ηideal = ∆G/∆H, ηvoltage expressing the correction for "voltage loss" by cell processes not contributing to the electric current, ηfuel the fraction of fuel used, and ηstack the stack correction to efficiency caused by non-optimal flow to each cell or other losses that are not counted for the individual cell.

Figure 3.64. Single PEM fuel cell performance. Current-voltage relationship (open symbols) and implied power density (filled symbols), are shown for different operational temperatures in the range of 45-90°C, for a cell with a catalyst layer with incorporated PTFE (polytetrafluoroethene) to reduce water flooding, a low loading of a carbon-supported Pt catalyst layer (120 µg Pt cm⁻² ), and finally Nafion intrusions. (From Z. Qi and A. Kaufman (2003). Low Pt loading high performance cathodes for PEM fuel cells. J. Power Sources 113, 37-43. Used with permission from Elsevier.)

A review of most of the processes occurring in various parts of a PEM fuel cell has been undertaken by Jiao and Li (2011), and modelling of the types described above has been applied to a concrete, early commercial PEM stack (Lee and Yang, 2011).

3.5.6 High-temperature and reverse operation

In Fig. 3.65, the efficiency of a reversible PEM fuel cell is depicted. Any fuel cell can in principle be operated in either direction, to produce electricity from hydrogen or to produce hydrogen from electricity. However, in most cases, the cell design optimised for one intended use is not very efficient for the reverse mode of operation. Electricity-producing PEM cells usually have a stack of many, fairly small cell assemblies, converting hydrogen with an efficiency typically below 60%, whereas a hydrogen-producing PEM electrolyser will typically have a few, large-area cells in order to produce hydrogen at an efficiency as high as 95% (Yamaguchi et al., 2001). Reversible fuel cells would have an important role in dispersed, building-integrated applications, but typical electricity-producing PEM cells will produce hydrogen with an efficiency of only around 50% (Proton Energy Systems, 2003). Efforts are in progress for finding a better compromise between the efficiencies of the two modes of operation, or preferably to improve the performance of both. One recent suggestion has been the use of new catalyst where Pt is mixed with IrO2 (Ioroi et al., 2002; 2004). Figure 3.65 shows the power and hydrogen efficiencies together with the implied round trip efficiency that would be relevant if the cell is used to produce hydrogen from electricity and to store it for later regeneration of electricity (e.g., in connection with intermittent primary energy sources such as solar or wind energy).

Figure 3.65. Efficiency of a reversible PEM fuel cell as a function of the amount (at. % or mol %) of Ir in the form of IrO2 relative to Pt in the positive electrode catalyst, for fuel cell electricity production (FC) or for water electrolysis (WE). Also the product of the two efficiencies relevant for storage cycles is shown. The catalyst is otherwise similar to that of Fig. 3.64, with PTFE and Nafion channels. (From T. Ioroi, K. Yasuda, Z. Siroma, N. Fujiwara, Y. Miyazaki (2002). Thin film electrocatalyst layer for unitized regenerative polymer electrolyte fuel cell. J. Power Sources 112, 583-587. Used by permission from Elsevier. See also Ioroi et al. (2004)).

The efficiency of electrolysis takes a strong dip when the IrO2 content drops and the catalyst approaches pure Pt, in accordance with earlier experience. However, already at 10% Ir content, the electrolysis efficiency is near 95%, rising only a little with further addition of Ir. At 10% Ir, the drop in the fuel cell efficiency for electricity production has been reduced only from 55 to 53%, so the technology for acceptable operation in both modes seems finally to be established. This will prove important for some of the hydrogen introduction scenarios described in Chapter 5.

Commercial PEM fuel cell stacks currently have electricity production efficiencies in the range of 30-60% at temperatures from 50 to 100°C (highest efficiency at the highest temperatures, as shown for a single cell in Fig. 3.64). An example of stack efficiency was given in Fig. 3.58. Upstart times are of the order of seconds, making the technology suitable for vehicle and other applications requiring quick upstart on demand.

Figure 3.66a, b. Comparison between high-temperature PEM-cell data and behaviour as calculated by a threedimensional model, describing the effect of CO from the fuel stream on electric behaviour (IV-diagrams) at different operating temperatures (a: 125°C, b: 150°C) (Jiao et al., 2011). Reproduced by permission from Elsevier.

There is an interest in constructing PEM fuel cells capable of operating at temperatures above the boiling point of water. This will solve some of the water management problems, but materials other than conventional Nafiontypes will have to be developed for the membrane. Modelling of a potential fuel cell of this kind, operating at temperatures around 160°C, has been performed, for example, by Shamardina et al. (2010), deducing current-voltage curves and comparing them with experiments. Poisoning by CO in hightemperature PEM cells has been modelled by Jiao et al. (2011) and found to cause less problems than below 100°C, as seen in Fig. 3.66.

Micro fuel cells intended for use, for example, with portable electronics, will be mentioned below in section 3.6, as they are often based on direct methanol fuel. Direct methanol fuel cells are also PEM fuel cells, as they are based on the transport of hydrogen ions through a solid polymer electrolyte.

The ability to operate in reverse mode exists for all types of fuel cells, the traditional electrolyser for example being an alkaline fuel cell. Reverse operation of SOFC cells, sometimes denoted SOEC ("E" for electrolysis), has been surveyed, for example, by Ebbesen et al. (2009).

3.5.7 Degradation and lifetime

As mentioned in the preceding individual subsections, degradation of PEM cells may occur as a result of structural change in electrodes (as for batteries), precipitated by the presence of foreign substances such as CO or water. Removal of CO requires a second reactor, such as a reformer or a catalytic reduction compartment in combination with a fuel cell using for example, liquid polyoxometalates in a positive electrode circuit (Kim et al., 2004). The overall lifetime of the current "first generation" of industrially produced PEM cells is expected to be of the order of 4000 h of continuous operation in contrast to the 1000 h lifetime of most prototype cells. For automotive applications, target lifetimes of 5000 h or more are desirable, under conditions of typical driving cycles, while for stationary uses, 5 years (43 800 h) is the minimum acceptable lifetime. It is hoped that such targets can be met within the next 5 years. Other concerns include, e.g., survivability at extreme temperatures. Current cells have problems if subjected to temperatures under about −25°C, e.g., due to failure of seals preventing gas transport through the diffusion layers by mechanisms other than diffusion (Schulze et al., 2004).

A variety of degradation processes have been investigated, from fuel impurities (Serinkan et al., 2010) over membrane degradation by OH-radicals (Panchenko, 2006), crossover (Nam et al., 2010) or (Jung and Williams, 2011, employing three-dimensional Monte-Carlo simulation) to gas channel and electrode damage, e.g., by water (Seidenberger et al., 2011) or CO as discussed above. Nafion membranes may be damaged if used at temperatures over 100°C, and for thin membranes, the crossover of hydrogen or oxygen remains a problem that must be handled, just as strict management of water in the membrane-electrode assembly (MEA) is required. Models for estimating the magnitude of MEA resistance as it increases over time have been formulated (Fowler et al., 2002). The combined measurement and simulation modelling study by Jung and Willliams (2011) of damage induced by prolonged operation has further suggested the variable build-up of H2O2 during start, operation and stop as a possible problem (see Fig. 3.67). Detailed measurements of catalyst layer degradation have been performed by Lee et al. (2009).

Figure 3.67. Hydrogen peroxide concentration across PEM fuel cell during open circuit operation (top), switch-on (large current; middle) and switch-off (small current; bottom) (Jung and Williams, 2011). Used with permission from Elsevier.

3.6 Direct methanol and other non-hydrogen fuel cells

The similarity of methanol (and ethanol) to conventional fuels used for transportation has spurred an interest in using such fuels directly in a proton exchange membrane fuel cell, thereby retaining the advantage of the simpler methanol storage and infrastructure, relative to hydrogen, while avoiding the additional component of a methanol-to-hydrogen reformer for accommodation in passenger vehicles where size and weight is at a premium.

The electrochemical reactions at the electrodes are in the methanol case

$$2CH_3OH + 2H_2O \rightarrow 2CO_2 + 12H^+ + 12e^-,$$ (3.76)

$$3O_2 + 12H^+ + 12e^- \rightarrow 6H_2O,$$ (3.77)

and the thermodynamical ideal cell voltage is φ = 1.20 V, similar to the one for hydrogen (3.20). The membrane must let some of the produced water pass through to the methanol side, and the byproduct besides water is carbon dioxide, which is a problem in a world needing to reduce emissions of greenhouse gases, except if the methanol is originally derived from (woody) biomass, so that both CO2 from methanol production and from methanol fuel cell conversion can be considered as balanced by CO2 assimilated earlier from the atmosphere (Sørensen, 2010a).

Figure 3.68. IV-curve and power density for direct methanol fuel cell with electrodes of carbon black coated on a carbon paper substrate, with Pt-Ru (ratio 1:1) on the negative electrode side and Pt alone on the positive electrode side, both with Nafion intrusions and hot-pressed on a Nafion-112 membrane. The 2-mol methanol solution was fed at a rate of 21 ml min-1 and at the other side non-humidified air at a rate of 700 ml/min. The temperature was 85°C. (From G. Lu and C. Wang (2004). Electrochemical and flow characterization of a direct methanol fuel cell. J. Power Sources 134, 33-40. Used with permission from Elsevier.)

It is possible to use methanol in all acid fuel cells, but the key research area is in PEM cells. A major difference between hydrogen PEM and direct methanol PEM cells (DMFCs) is that the latter produces exhaust gases on both sides of the membrane-electrode assembly. Both CO2 and N2, if air is used at the positive electrode, can block pores of the diffusion layers and catalysts. The reaction (3.76) is relatively slow, and power densities obtained so far are at least ten times lower than for hydrogen-fuelled PEM fuel cells, as illustrated in Fig. 3.68. Better catalysts than Pt are under investigation, taking into account evidence that the Pt attracts a CO subunit of the methanol molecule in a way depending on the presence of steps in the catalyst surface (see Fig. 3.63a). Catalyst candidates include Ru-Pt alloys and Pt-Sn alloys on amorphous Ni-Nb substrates (Sistiaga and Pierna, 2003). A problem is that of stability, both of more complex catalysts (such as Mo-Ru-S alloys) and also of the membrane, where thin membranes give higher performance but also promote methanol crossover and degradation (Hamnett, 2003).

The maximum power density of 0.093 W cm⁻² shown in Fig. 3.68 is one of the highest currently demonstrated, but still nearly ten times lower than that of hydrogen PEM cells (see Fig. 3.64), which also achieves more than twice the current density. Typical single cell overall efficiencies are of the order of 40% (Müller et al., 2003).

The crossover of methanol has caused problems in finding a suitable membrane material. On the positive electrode side, methanol combines with oxygen to form CO2. Among the alternatives to pure Nafion are Nafion filled with zirconium phosphate or grafted with styrene to inhibit methanol transport (Bauer and Willert-Porada, 2003; Sauk et al., 2004), as well as non-Nafion membrane materials such as sulfonated polyimide (Woo et al., 2003). None have achieved performance as good as the one shown in Fig. 3.68, which, however, has a substantial methanol crossover rate.

Cell modelling techniques for flow and electrochemistry are basically the same for DMFC as for other PEM fuel cells−that is, the techniques described in sections 3.1.5 and 3.5.1-3.5.3 have been employed for DMFCs by Fuhrman and Gärtner (2003).

Fuels other than methanol have been contemplated, such as formic acid, which exhibits lower crossover fluxes than methanol and thus may be an alternative for use for example, in small systems contemplated for portable applications (Ha et al., 2004; Zhu et al., 2004).

Due to the inferior performance, DMFCs are not the ideal choice for automotive applications, although the possibility is being investigated (e.g., by DaimlerChrysler-Ballard). However, for small portable devices, the convenience in carrying fuel and minimising the number of components may weigh more than efficiency in determining acceptability. The proper alternative to compare with for the small portable applications is lithium ion batteries, which currently have a power density of around 130 Wh/kg but with further development may reach 200 Wh kg-1 (Sørensen, 2010a).

The theoretical power density of a DMFC at 0.5 V is about 1600 Wh per kg of methanol fuel, but in practice, small DMFCs for portable applications have achieved much less. If small DMFCs are designed like conventional PEM cells, including a membrane-electrode assembly (MEA), two gas diffusion layers, fuel and air channels with forced flows and current collectors, they may achieve power densities of about 0.015-0.050 W cm⁻² at temperatures in the range of 23-60°C (Lu et al., 2004), consistent with the value found at 85°C in Fig. 3.68.

An emerging new approach to design is to simplify the number of components and to dispense from forced flows, as these require mechanical parts not always desirable for portable applications. A resulting design may look as shown in Fig. 3.69. There are no gas diffusion layers, no forced fuel flow (the area between two MEAs constitutes the fuel container) and no air channels, but there is free exposure to air on the outside of the membrane assemblies, supported by a meshed conductor (Kubo, 2004). Modelling the performance of a passive DMFC has been done by Yang et al. (2011), assuming operation with diluted methanol (3-5 M) to reduce crossover problems.

Figure 3.69. Direct methanol fuel cell designed for passive operation (no forced flows), with two membrane-electrode assemblies (MEA) and a central fuel container.

In order to compensate for the decline in power density implied by not using forced flow and by operating at ambient temperatures, Kubo used an improved Pt catalyst structure with high surface area and finer Pt particles (about 2 nm diameter) than conventional "lumpy" structures (such as those shown in Fig. 3.62c,d). The structure is one of "nanohorns" produced from graphene, with a sheet structure (similar to single-wall nanotubes) shown in Fig. 3.70. They assemble in the "sea urchin" landscapes shown in the lower right part of the figure, and Fig. 3.71 shows the tunnelling electron microscope pictures after application of Pt particles (a) and for comparison in (b) the same for a conventional Pt catalyst of the type used in PEM cells such as those of Fig. 3.62. With the improved catalyst structure, the measured power density was about 0.045 W cm⁻² at ambient temperature, implying that the catalyst improvement has brought the performance up to the same level other DMFCs reach at 60°C with forced air and methanol flows and without the simplifications in design such as the absence of gas diffusion layers and air channel.

Figure 3.70. Carbon nanohorn substrates used for catalyst backing in NEC's passive DMFC design. (From Y. Kubo (2004). Micro fuel cells for portable electronics. In Proc. 15th World Hydrogen Energy Conference, Yokohama. Used with permission.)

Micro fuel cells for portable applications could also be conventional hydrogen PEM cells, or a micro-reformer could be included in the concept, again due to the convenience of carrying hydrocarbon or methanol fuels as compared with carrying hydrogen around (e.g., Holladay et al., 2004). Small planar PEM cells (size range 0.01-1.0 cm² ) have been developed for hydrogen fuel in the same manner as the passive DMFC discussed previously (Hahn et al., 2004). Again, there is no gas diffusion layer and no forced flow, but in contrast to the device in Fig. 3.69, there are flow channels for both air and hydrogen, delimited by a patterned current collector with a gold layer deposited at a thickness of around 0.01 mm. This was required in order to get reliable long-term performance, resulting in a power density peaking at 0.09 W cm⁻² for a current density of 0.20-0.25 A cm⁻² . Alternative use of methanol fuel may match the performance of the structure shown in Fig. 3.69.

Figure 3.71. Platinum catalyst (dark spots) deposited on carbon nanohorn substrates in NEC's passive DMFC design (a), compared to Pt deposited on a conventional carbon substate (b). (From Y. Kubo (2004). Micro fuel cells for portable electronics. In Proc. 15th World Hydrogen Energy Conference, Yokohama. Used with permission.)

Recent developments in the passive designs have concentrated on trying to prolong the operating time between refills, without losing the advantages of diluted methanol fuel. This is done by doubling the number of reservoirs, adding one with highly concentrated methanol while keeping the diluted methanol reservoir, hoping that the concentrated methanol could be reserved for short high-performance intervals and the limited crossover rate still be maintained. Basic design of such cells is discussed by Shaffer and Wang (2010), and transient behaviour is modelled by Cai et al. (2011).

Compared to the (modest as seen from the user's point of view) goal of 5000 h operational life set for several of the other types of fuel cells, DMFC lifetimes have been disappointingly short, typically below 300 h of operation (Bae et al., 2010). A number of new ideas have been proposed for electrode design and catalyst choice, including RuSe material supported by carbonnanotubes like those shown in Fig. 3.70, but so far without much success (Jeng et al., 2011). The short lifetimes can perhaps be accepted if passive DMFCs are seen as replacements for advanced batteries as power source in portable equipment, an application further discussed in Chapter 4. Modelling and control strategies for such devices are discussed by Zenith and Krewer (2010). Figure 3.72 shows the result of a recent model-calculation of DMFC efficiencies, as function of current density, temperature, and methanol concentration.

Figure 3.72. Conversion efficiency of DMFC estimated by semi-empirical model of Chiu et al. (2011). Low current density, low methanol loading, and elevated temperature yield the highest efficiency (Used by permission from Elsevier).

3.7 Biofuel cells

The concept of biofuel cells was briefly mentioned at the end of section 2.1.6, with microbial sensitisers employed in photoelectrochemical devices. The production of hydrogen by photosynthesis or dark fermentation (section 2.1.5) in principle allows hydrogen to be used as a fuel independent of the source. However, a number of integrated systems have been contemplated, where the biological reactor is adjacent to or integrated into a fuel cell in order to save transport. The problem with this approach is that the hydrogen production from biological systems is variable, particularly of course for photosynthetic systems. This means that hydrogen cannot always be fed into the fuel cell at an optimum rate, with lower efficiency as a likely result.

In a recent experiment, using Chlamydomonas reinhardtii to produce hydrogen, the CO-free gas was fed into a PEM fuel cell, using the accompanying gases (CO2, N2) to generate a flow (Dante, 2005). With 100 m3 of algae culture, a power production of 475 W appeared over a period of 25 h, with rather sharp drops before and after. Of course, production intermittency could be smoothed by adding a hydrogen store. Earlier experiments used Anabena variabilis, with even more modest results (Yagishita et al., 1996). In all cases, the difficulty in handling the low and intermittent photosynthetic fuel production makes integration with a fuel cell unattractive. Fermentation could conceivably produce a more steady flow of hydrogen. However, if these avenues of hydrogen production should ever become viable, the hydrogen would probably be better utilised by storage and use in the same fashion as hydrogen from other sources.

Other schemes use biological molecules as catalysts in order to obtain direct enzymatic oxidation of a fuel such as glucose (Katz et al., 2003; Chaudhuri and Lovley, 2003). The advantage is that a natural biological material could be used in a special fuel cell without having to first convert it to hydrogen, but there are still many problems to overcome, related to the poor ability of the biological substances to conduct−that is, to transfer electrons to the external electrodes. Attempts are being made to integrate the biological material in nano-structured carbon materials or to find biological organisms capable of transferring electrons without special mediators.

Figure 3.73. Suggested docking of Rhodoferax ferrireducens surface to a carbon electrode for efficient transfer of electrons derived from oxidation of glucose. (From G. Tayhas and R. Palmore (2004). Bioelectric power generation. Trends Biotechnology 22, 99-100. Used with permission from Elsevier).

One such organism is Rhodoferax ferrireducens, which is claimed to be able to convert over 80% of electrons from glucose to cell current (Chaudhuri and Lovley, 2003). Whether this concept can be translated into a practical biofuel cell remains to be seen. It does gain all the 24 electrons from glucose and seems to attach to a carbonic electrode in a way that allows electron transfer without the help of enzymes or catalysts. The methods for this are not understood but seem to involve the R. ferrireducens outer boundary ("biofilm") having an affinity for attaching to the carbon surface (Fig. 3.73). The energy transfer from glucose involves reduction of Fe(III),

$$C6H{12}O_6 + 6H_2O + 24 \text{ Fe(III)} \rightarrow 6CO_2 + 24 \text{ Fe(II)} + 24H^+,$$ (3.78)

as suggested by Chaudhuri and Lovley (2003) and Finneran et al. (2003). Placing the bacterium in the negative electrode compartment of a biofuel cell, separated from the positive electrode chamber by a membrane, a current of up to 0.6 mA, or 0.3 mA m⁻² , was observed.

Biofuels have several uses in the energy sector (Sørensen, 2010a), and it is generally not attractive to transform them to hydrogen first, although this can be done by conventional methods (gasification, steam reforming). Suggestions to use fuel cells to transform biofuels to power are abundant, but if the conversion is through hydrogen, the overall conversion efficiency is usually very low, and if the biofuels are used directly, whether in SOFCs (Mermelstein et al., 2011), MCFCs (Hernández et al., 2011), or "microbial" PEMs (Lee and Nirmalakhandan, 2011), the negative impacts of air pollution usually remains, in addition to a cost increase that is hardly paid for by the modest increase in efficiency relative to non-fuel-cell uses.

Figure 3.74. Voltage and power curves for a microbial fuel cell, as function of current (Katuri and Scott, 2011). Used by permission from Elsevier.

A typical performance of a microbial biofuel cell is shown in Fig. 3.74. The conversion efficiency stays below about 100 mW/m2 , which is 10 000 times lower than for a methanol fuel cell (Fig. 3.68) and more than 100 000 times lower than the efficiency of a SOFC (Fig. 3.37). The only way to regain these large factors is by similar reductions in specific fuel cell cost, which is plainly unrealistic.

Direct photovoltaic hydrogen production (section 2.1.6) has been proposed to be based on sensitisers made from organic dyes or Ru-complexes (Kalyanasundaram and Graetzel, 2010), a technology that failed to make it for solar cells (Sørensen, 2010a).

3.8 Problems and discussion topics

3.8.1

Write expressions in analogy to (3.12) for the efficiency of fuel cells in reverse operation, used to produce either hydrogen or both hydrogen and heat from electric power.

3.8.2

Using fossil fuels in solid oxide fuel cells, it is possible in cars to reform the fuels to hydrogen within the device. CO2 will be a side product, unlikely to be collectable as in stationary installations. What are the implications for greenhouse gas emissions? For natural gas as fuels, what is the global gas requirement and CO2 emissions? Same question if coal is used, although here the direct carrying of coal in cars is improbable, and it would rather be a "producer gas" in compressed form.

3.8.3

Think of a fuel cell and its fuels as an energy store and compare it with other storage options. For example, estimate the energy density on a weight basis for a conventional lead battery, an advanced lithium ion battery and for several of the low-temperature fuel cells (such as PEM and DMFC). In contrast to the storage point of view in Chapter 2, section 2.4 (or in Chapter 5 of Sørensen, 2010a), you will need not only the weight of the fuel, but also of all the equipment making up the fuel cell and fuel handling system, including the fuel container. Some data on this may be found in Chapter 6. Energy and power densities have a particular relevance for portable devices, so make sure that you include some portable concepts such as the passive DMFC.

Try to make a similar comparison, not on a mass, but on a volume basis. For fuels, this may be taken from tables such as Table 2.4, but for equipment, you may have to look at actual equipment (e.g., on the manufacturer's website) to get an idea of just how tiny or bulky the devices are.

3.8.4

Try to design a fuel cell with all-solid components (except possibly for water), having the benefits of a PEM fuel cell but without its drawbacks. Would it look like a 200°C acid cell?

3.8.5

Could a direct carbon fuel cell work without employing, as the MCFC, melting temperatures?

SYSTEMS

This chapter examines devices using hydrogen and fuel cells to provide traction power for vehicles or heat and electricity for stationary purposes. Such devices are often denoted "systems", but the term "system" is used fairly generally, because the distinction between systems and components is not always sharp, and in consequence, systems will be mentioned in many places throughout this book. This chapter is concentrating on a systematic overview of several types of systems employing hydrogen and fuel cells, constituting comprehensive aggregates of components serving particular demands, such as providing personal or freight transportation or providing heat and power to a building. These individual systems will in Chapter 5 be combined to form nation-wide or global networks of interconnected energy supply systems, this being another conventional use of the term "system". The reuse of the term may be defended as being less ambiguous than alternatives such as "economies" or "societies", often found in constellations like "hydrogen economy" or "hydrogen society".

4.1 Passenger cars

4.1.1 Overall system options for passenger cars

The simplest system for using hydrogen in a passenger car, not considering straight combustion (see section 2.3.3), has a fuel storage tank, a fuel cell, and an electric motor. The electric motor is rated at the maximum power required by the vehicle, and as there is no traction battery included, the fuel cell must be able to deliver the same power input to the motor, while the hydrogen store must be large enough to give the vehicle the desired range.

In case fuels other than hydrogen are used, the fuel cell must be capable of

accepting them (direct methanol fuel cell, etc.), or they must be converted to hydrogen, for example by use of a reformer (natural gas, gasoline, methanol, etc.; see Chapter 2, Fig. 2.27). The storage tank is now accommodating the fuel of choice.

A control system administers the flows of fuel and timing of the components' functions. There should in most cases also be a water management system, capable of keeping the fuel cell (if it is of PEM type) at the proper humidity. In many cases, it is not convenient to use hydrogen for cold starts, and a start-battery is used to furnish start-up power. This could be a conventional car battery (of lead-acid type) with modest storage capacity, but generally, the functionality will be improved by using a larger, high-voltage battery.

Figure 4.1. Various hybrid car concepts. (From G. Suppes, S. Lopes, C. Chiu (2004). Plug-in fuel cell hybrids as transition technology to hydrogen infrastructure. Int. J. Hydrogen Energy 29, 369-374. Used by permission from Elsevier.)

When a battery aimed at traction is added, the system is called a hybrid system. Now the power required at the wheels is delivered either by the fuel cell or by the energy already stored in the battery. In hybrid systems, the option of delivering power directly from the fuel cell to the electric engine (termed parallel operation) is not mandatory, as the fuel cell may deliver all its power output through the battery (serial operation). Whether this is the case or not, the hybrid concept allows the fuel cell rating to be smaller than that of the engine. One option is in this case to operate the fuel cell at a constant power level and have the battery charge up when the power is not needed for traction. An alternative is to allow battery charging off-road, for example, when the car is parked or at a filling station. Such fuel cell vehicles are called plugin hybrids (Bitsche and Gutmann, 2004; Suppes et al., 2004). Figure 4.1 shows some possible hybrid layouts.

For plain combustion hydrogen vehicles, the components include the engine and usually a liquid hydrogen store (to get sufficient range). The control equipment must therefore also comprise an exhaust system capable of safely handling the hydrogen boil-off from such a store (Ochmann et al., 2004). The specific problems and optimising issues arising when using hydrogen in an internal combustion engine are discussed by Verhelst and Wallner (2009).

Figure 4.2. Schematic layout of power system for a PEMFC vehicle. (From R. Ahluwalia, X. Wang, A. Rousseau, R. Kumar (2004). Fuel economy of hydrogen fuel cell vehicles. J. Power Sources 130, 192-201. Used by permission from Elsevier.)

In assessments of energy efficiency, it is important to include all components of the system. Each energy-converting device is characterised by a conversion efficiency (energy output over energy input), as well as an exergy efficiency (free energy out over free energy in), with the latter reflecting the quality of energy (Sørensen, 2010a). The fuel-to-electricity efficiency of various fuel cell types is 30-70% (Chapter 3), but to this comes the upstream efficiency of producing the fuel and the downstream efficiency of using it. Hydrogen production from fossil fuels or biofuels has an efficiency of 45-80%, while production from electricity is at 60-90% efficiency (Chapter 2). Further, the efficiency characterising the production of electricity in the first place may add another factor of 30-45% to the efficiency, if the basis for electricity is fossil resources. For renewable energy sources, such as wind power or photovoltaic panels, the conversion efficiency is usually not included in this context because the primary source is "free". Downstream efficiencies ranges for automobile traction are typically 35-45% (Chapter 6), while for electric lights and appliances, nearly the entire scale of efficiencies is met in various actual devices. It follows that the overall efficiency from primary energy to an energy service at the end-user, such as mobility, may be as low as 5%. The positive message contained in this fact is that there is plenty of room for improvement by proper design and combination of components.

Figure 4.3. Placement of fuel cells, hydrogen tank, and auxiliary equipment in the early DaimlerChrysler prototype car Necar 4. (Based on G. Friedlmeier, J. Friedrich, F. Panik (2001). Test experiences with the DaimlerChrysler fuel cell electric vehicle NE-CAR 4. Fuel Cells 1, 92-96. Used by permission from Wiley.)

4.1.2 PEM fuel cell and fuel cell hybrid cars

As most current automotive fuel cell efforts use proton exchange membrane

(PEM) fuel cells, these will be described in a little more detail in this section and will be used as templates for the performance calculations presented in section 4.1.3. A typical passenger car pure (i.e., not hybrid) PEM fuel cell system is outlined in Fig. 4.2. Included are heaters for bringing the equipment from ambient temperatures to the operating temperature of around 80°C and humidifiers for ensuring the level of water in the membrane and electrode areas required for operation (see Chapter 3, Figs. 3.50, 3.51, and 3.60). The water management equipment includes a condenser, which is integrated into a conventional radiator, but operating at much lower temperatures than that of internal combustion engine cars.

Figure 4.4. The General Motors skateboard concept, having all power equipment and physical controls placed in a flat frame construction placed below the passenger cabin, and all driving control instructions transmitted electronically from the cabin to the board equipment. (From M. Herrmann and J. Meusinger (2003). Hydrogen storage systems for mobile applications. Presented at 1st European Hydrogen Energy Conf., Grenoble. Used by permission from GM.)

The fuel cell equipment adds considerably to the weight of the vehicle, implying lower efficiency but also the possibility of increased stability, achieved by placing the heavy equipment low in the vehicle construct. Figure 4.3 shows the under-floor placement of fuel cells, storage tanks, and auxiliary equipment in the Necar 4 hydrogen-fuel prototype fuel cell car of Daimler-Chrysler, making it much more stable than its commercial combustion engine cousin, the commercial Mercedes-Benz A-class car. While the Necar 4 has a 75-kW fuel cell stack, the newer DaimlerChrysler 0-series f-cell car (see Chapter 6, section 6.2.4) uses a 85 kW Mark-902 fuel cell stack from Ballard Power Systems. Use of these huge power systems (relative to the size of the class-A cars, normally working with a 40-50 kW diesel engine) is considered necessary because of the extra weight of the fuel cell-related equipment, and in order not to attract any criticism for lowering the "sporty" performance. A larger, B-class version of the f-cell has been built, featuring smaller volume fuel tanks by use of 70 MPa compressed hydrogen rather than the earlier 35 MPa (Orecchini and Santiangeli, 2010).

An even more advanced step is the concept first proposed by General Motors (Fig. 4.4), where not only all fuel cell equipment is placed below the passenger cabin, but this "skateboard" is totally isolated from the cabin and receiving all instructions electronically (for steering, for braking, and for accelerating). Achieving optimum control is being assisted by having not just one but four electric motors, one at each wheel. The concept is used in prototype hybrid cars such as Sequel (with lithium-ion batteries) and HydroGen4 (with nickel-metal hydride batteries; Eberle and Helmolt, 2010). Other car manufacturers, such as Toyota, in their prototype fuel cell vehicles have placed the fuel cell stacks more conventionally in what traditionally constitutes the engine space in front of the passenger cabin (Takimoto, 2004).

The methanol-reformer systems (e.g., DaimlerChrysler's Necar 5) were for some time developed in parallel with the direct hydrogen fuel cell vehicles, considering that the advantages of only needing minor changes in fuelling infrastructure would outweigh the somewhat lower overall efficiency (Boettner and Moran, 2004). However, technical problems with the reformer performance have currently brought this line of development to a halt. Direct methanol fuel cell vehicles are still under development, for example, by DaimlerChrysler (Lamm and Müller, 2003).

Basic performance modelling of PEM fuel cells was undertaken in Chapter 3. Further such studies, directed specifically at thermal performance in automotive applications, may be found in Nolan and Kolodziej (2010).

A considerable number of investigations have been carried out in an attempt to determine the best control strategies for vehicles with PEM fuel cells with or without a traction battery or other complementary power-providing device such as a flywheel or a capacitor (fuel cell control, Al-Durra et al., 2010; hybrid system without plug-in option, Bernard et al., 2010, Ryu et al., 2010, Fadel and Zhou, 2011; hybrid with plug-in capability, Bubna et al., 2010a). Some 5% fuel saving may typically be obtained by refining the simpler control strategies. As mentioned, the hybrid system may be either serial (all fuel cell power is delivered through the energy store) or parallel (the electric motor may be powered directly by the fuel cell or by the storage). The hybrid systems studied in the references just quoted are parallel systems. An example of a serial system is given in section 4.1.3.

Hybrid systems combine hydrogen fuel cells with electrochemical energy stored, typically in a battery, and thus constitute a natural continuation of the technology choice embedded in current fossil fuel-battery hybrid vehicles such as the Toyota Prius and the similar Lexus hybrids (gasoline and nickelmetal-hydride battery; Orecchini and Santiangeli, 2010). Plug-in hybrids using advanced lithium-ion batteries include the General Motors Volt and Ampera. The Li-ion batteries are also the choice for the range of commercial purely electric vehicles marketed by Renault and by Citroën/Peugeot, because of their excellent capability for following complex and demanding driving cycles (Corbo et al., 2010). As with Li-ion batteries for small electronic products, there is a safety issue usually solved by a self-releasing emergency vent (Arora et al., 2010).

Because the reaction time of a fuel cell is slower than that of most batteries, there is usually a modest-size battery even in "pure" fuel cell vehicles, aimed to provide better performance during changing operating condition (such as acceleration or deceleration). This function could conceivably be taken over by other technologies, such as capacitors. Flywheels have been considered but for safety reasons, they are more appropriate for stationary applications, where they can be placed underground. Capacitors, on the other hand, can display extremely high power densities (compared with usually adequate and sufficient batteries) and thus provide the thrills of motorcar racing and reckless driving. Control systems for capacitors and supercapacitors integrated into fuel cell drive trains with or without batteries have been considered by Ayad et al. (2010) and by Lin and Zheng (2011). Incorporating all three expensive components, fuel cells, batteries and supercapacitors, may well be overkill (Yu et al., 2011). In the opposite vein one finds the suggestion that pure battery hybrids may perform well enough to be used for road vehicles, at the extreme without use of a voltage converter (a device that allows battery terminal and motor bus voltages to be different) to maximise efficiency (Bernard et al., 2011).

While early prototype passenger pure fuel cell vehicles were rated at about 100 kW, the advantage of hybrids is that the expensive fuel cells do not have to provide the peak power. The smallest vehicles may need only a few kW of rated fuel cell power (Tang et al., 2011) and a four-person car 10-20 kW (see section 4.1.3). At the same time, the presence of the fuel cell makes it possible to do with less battery capacity, affecting another expensive component. One purpose of the simulation studies below is to explore the optimum rating of the fuel cell and battery in a hybrid vehicle.

4.1.3 Performance simulation

In parallel with actual tests on the road, simulations are used by both scientists and auto manufacturers as a means of getting a first orientation at low cost, both before a new car is actually constructed and also during the sequences of testing and revision phases. Here, a brief simulation study will be made in order to illustrate the capability of simplified, but fairly realistic, model assumptions.

Models are available, capable of simulating the behaviour of various vehicle types including those with conventional propulsion systems using Otto or Diesel engines, pure electric vehicles or pure fuel cell vehicles as well as hybrids among any of these, either by a detailed physical modelling or by a semi-empirical method, where different processes are simply parametrised and the parameters are adapted to measured data. The latter method is presented here, based on the software ADVISOR originally developed at the US Renewable Energy National Laboratory (Markel et al., 2002). The users can write their own subroutines, or use the existing collection of parametrised models for fuel cell stacks, electric motor, battery energy storage, power-train control in fuel cell cars with battery, exhaust control, power behaviour of the wheel/axle system under prescribed driving conditions (slope, road surface, and resistance, etc.), and also auxiliary electric energy use in the vehicle. The ADVISOR programme core uses the required driving speed at a given time to calculate torque, rotational speed, and power in each drive-train element, a procedure called a backward-facing simulation approach. However, this is combined with a forward-facing approach based on the control logic, and the simulation proceeds forward in time, but with backward consistency checks at each step. Program users can both change and add subroutines for specific processes, or change the flow between the modules.

The fuel cell modelling is either a simple one where power output and efficiency are linked by an empirical curve such as the one shown in Fig. 4.5, or a combination of relationships for each component in the fuel cell system (Fig. 4.2). More detailed models−for example, treating water management−are available (Markel et al., 2002; Maxoulis et al., 2004). For the battery in hybrid modelling, additional calculations were made with the battery assumed to be rechargeable by the fuel cell. Both conventional lead-acid or nickel-metal-hydride batteries and advanced lithium ion batteries have been modelled, using internal resistance battery models. Also, for the battery segment of the simulation, much more detailed subroutines have been developed for use by technical vehicle designers.

The fuel cell vehicle considered in the simulation model is an artefact loosely modelled over the Volkswagen Lupo TDI-3L (see Chapter 6, Table 6.6), so for simplicity, I shall call it Little Red Riding Hood. The 45 kW diesel engine is replaced by a PEM fuel cell engine in a hybrid configuration with battery power, ranging from pure fuel cell operation to a pure electric vehicle. Suitable component rating is initially determined on the basis of efficiency for a given driving cycle rather than (poorly known) cost. This may lead to several systems of comparable performance.

Figure 4.5. Power curve for a 50 kW PEM fuel cell used in simulations.

The hydrogen fuel consumption for this vehicle is assumed to take the form shown in Fig. 4.6. The fuel efficiency is closer to the target value for current research than to the (lower) efficiency of the early prototypes that have been test-driven over the last decade.

Figure 4.6. Hydrogen consumption of the Little Red Riding Hood vehicle assumed in simulation model, as function of power level. Note that the partload performance implied by Figs. 4.5 and 4.6 is nearly independent of power level down to around 1 kW, in clear contrast to the situation for internal combustion engines.

The total mass of the each Little Red Riding Hood configuration has the distribution shown in Table 4.1 (compare to Chapter 6, Table 6.6 for the Diesel Lupo 3L).

Component mass (kg)	pure FC	hybrid	pure EV
Basic vehicle (incl. Li-ion start battery)	570	570	570
Fuel cell equipment (40, 20 and 0 kW)	150	100	0
Exhaust management	8	5	0
Li-ion batteries (0, 15, and 250 MJ)	0	70	1134
Electric motor (50 kW)	60	60	60
Transmission (manual 1-speed equivalent)	50	50	50
Passengers and cargo (average)	136	136	136
Total	974	991	1950

Table 4.1. Mass distribution for selected Little Red Riding Hood fuel cell-electric hybrid vehicles. For lead-acid or Ni-MeH batteries, the battery mass is 2-3 times higher.

Figure 4.7. The driving cycle used in simulations (Sørensen, 2005c, d).

The simulations are employing a mixed driving cycle (Fig. 4.7) put together from pieces of the driving cycles used in the United States and in the European Union for regulatory and taxation purposes and containing both highway driving, suburban stretches, and city driving with frequent stops at red lights. The overall speed frequency distribution of the 89 km route is shown in Fig. 4.8.

Figure 4.8. Frequency distribution of driving speeds for the driving cycle shown in Fig. 4.7.

Figure 4.9. Simulated power output from 40 kW pure fuel cell vehicle.

For a pure fuel cell vehicle rated at 40 kW, the efficiency assumptions of Figs. 4.5 and 4.6 make it possible to traverse the driving cycle of Fig. 4.7 without noticeable deviations from the prescribed speeds along the trip (maximum deviation some 2%). Figure 4.9 shows the power output from this fuel cell, as function of driving time at the specified speeds. For simplicity, no grades have been assumed present. The average energy consumption of the 40 kW fuel cell during the driving-cycle trip is 1.138 MJ/km (equivalent to about 3.5 litre of gasoline per 100 km) (Sørensen, 2010b). Figure 4.9 suggests that even with a 30 kW fuel cell rating, the particular driving cycle could be completed at the required speeds, in which case the fuel consumption would be reduced to 0.855 MJ/km (or 2.7 litre of gasoline-equivalent per 100 km) (Sørensen, 2007b, c). However, in a real car, the higher rating of at least 40 kW would be used, because the driving cycle selected for the study may not represent the actual maximum requirement met on the road (for instance, it does not contain slopes). The range of the vehicle on a 4 kg hydrogen fuel tank (equivalent to 178 litre of hydrogen at a pressure of 30 MPa) would be over 650 km and thus similar to the range of most present passenger cars.

However, Fig. 4.9 also shows that the upper-level fuel cell power rating is only required during quite short parts of the trip. This suggests that an advantageous configuration would by a fuel cell-battery hybrid, where the fuel cell is rated at a lower level, such as 20 kW or even down in the range of 5-10 kW, and where the peaks are taken care of by a traction battery feeding power into the electric motor (which at little extra cost can be rated at the higher power level envisaged, here chosen as 50 kW). In the following, the performance of a range of such hybrid configurations will be simulated, using a serial coupling, i.e., that the fuel cell delivers power to the battery and the motor is driven entirely from the battery (avoiding the direct use of electric power from the fuel cell, in a parallel configuration with or without voltage conversion).

As a prelude to the model simulation of various hybrid layouts, a pure electric vehicle will be treated, as it forms the end-point opposite to that of the pure fuel cell vehicle considered. Several detailed models of battery performance exist (for instance, Albertus et al., 2008), but only a simplified equivalent Kirchhoff circuit model will be employed here (Johnson, 2001; Liaw and Dubarry, 2010), judged to be sufficient for identifying the general behaviour of hybrid vehicles. For a single battery module of which about a thousand makes up the 250 MJ lithium-ion battery electric vehicle specified in Table 4.1, it predicts the voltage and resistance behaviour illustrated in Figs. 4.10 and 4.11, as function of charge state and for different temperatures. As with most batteries, low temperature (around 0°C) degrades performance, while high ambient temperatures (about 40°C) do not cause significant problems except when the battery is close to being completely discharged. The implications for power delivery is illustrated in Fig. 4.12, suggesting that the best

performance is ensured if the battery is never discharged much under 40%. Finally, Fig. 4.13 shows the round-trip efficiency of the battery at 25 °C, as function of current and state of charge. Except for the poor performance near zero charge, the efficiency is quite high at low currents and declines with increasing currents drawn, but still stays above 70% for currents up to 100 A.

Figure 4.10. Voltage diagram (volt for a given charge state) for a module of a "Saft" Liion battery (Johnson, 2001), for three ambient temperatures.

Figure 4.11a, b. Resistance diagram (ohm for a given charge state) for a module of a "Saft" Li-ion battery (Johnson, 2001), for three ambient temperatures, during charging (a) and during discharging (b).

Figure 4.12. Instantaneous power (W for given state of charge) delivered by a module of a "Saft" Li-ion battery (Johnson, 2001), for three ambient temperatures.

Figure 4.13. Round-trip efficiency as function of current and charge state for a module of a "Saft" Li-ion battery (Johnson, 2001) at ambient temperature 25°C.

Figure 4.14. Simulation of pure electric vehicle based on Little Red Riding Hood with Liion batteries: Net torque delivered by electric motor during the driving cycle.

Figure 4.15. Simulation results for Little Red Riding Hood in a pure electric vehicle version with Li-ion batteries: Power delivered by battery assembly during driving cycle. Negative values represent brake energy recovery.

Figure 4.16. Simulation results for Little Red Riding Hood in a pure electric vehicle version with 250 MJ Li-ion batteries: Battery charge state as function of cycle time. The average energy use−that is, drop in battery stored energy during the driving cycle−is 0.617 MJ/km.

The results of carrying out simulation of the performance of the pure electric Little Red Riding Hood vehicle is shown in Figs. 4.14 to 4.16, giving first the torque and power provided by the battery/motor power train and in Fig. 4.16 the charge state of the battery during the trip. From the total energy drawn from the battery during the trip, an average energy use of 0.617 MJ km-1 can be calculated, which corresponds to an equivalent 2.6 l of gasoline per 100 km. This is better than the energy performance the VW Lupo on which the vehicle is modelled (that has a fuel use of 3 l diesel fuel per 100 km or about 3.3 l gasoline per 100 km, cf. further discussion in Chapter 6). The stated electric vehicle energy use does not include the upstream losses in charging the batteries from some external source of electric power, or the losses that may have occurred in generating that electricity from some primary source.

Turning now to fuel cell/battery hybrid vehicles, the results of a simulation for a specific combination of a vehicle equipped with a 20 kW fuel cell and a modest 15 MJ Li-ion battery (as specified in Table 4.1) will be presented. The battery can be recharged by the fuel cell, in the serial configuration assumed. Figures 4.17 and 4.18 show the net power output from the battery and from the fuel cell during the mixed driving cycle of Fig. 4.7. The size of the battery has been chosen so that the charge state at the end of the assumed driving cycle is approximately the same as at the beginning, as illustrated in Fig. 4.19.

For other trips made with the vehicle, this may of course not be the case.

Figure 4.17. Simulation results for Little Red Riding Hood vehicle with 20 kW fuel cells and 15 MJ Li-ion batteries: Power delivered by battery assembly during driving cycle. Negative values represent brake energy recuperation.

Figure 4.18. Simulation results for Little Red Riding Hood vehicle with 20 kW fuel cells and 15 MJ Li-ion batteries: Power delivered by fuel cells during driving cycle.

Figure 4.19. Simulation results for Little Red Riding Hood vehicle with 20 kW fuel cells and 15 MJ Li-ion batteries: Battery charge state as function of cycle time. The average energy drawn from the fuel cell during the driving cycle is 0.796 MJ/km.

It is seen from Fig. 4.19, that particularly during the high-speed highway part of the driving cycle (at 5000-6000 sec), the fuel cell input to the battery is smaller than the power required, and the battery charge state thus declines rapidly. At other times, the fuel cells deliver more power charging the batteries than drawn by the motor. The equivalent gasoline fuel use would be 2.5 l per 100 km, after subtracting a small battery charge state gain at the end of the cycle (the end-point of the curve in Fig. 4.19 is slightly above the startlevel).

Lower than 20 kW fuel cell rating leads to insufficient battery charging, implying that the vehicle cannot be electrically autonomous, but has to obtain battery recharging from an external source. The latter situation describes a plug-in hybrid vehicle. The plug-in configurations allow a less costly size of fuel cells and permit recharging of batteries whenever the car is parked at a suitable power outlet (at home, at a working place, or at a special publicly located dispenser charging for a fee). The cost of recharging facilities is typically smaller than that of hydrogen filling stations, but for a plug-in hybrid, both are required. The technical performance of plug-in and self-recharging fuel cell hybrids will be considered following, and guidelines for optimising the economy of the solutions will be dealt with in Chapter 6.

The electric motor used in the simulations has an efficiency of 0.92 and a manual 1-speed. In conventional vehicles the transmission often constitutes a significant loss factor, and losses in using 5-speed or automatic transmissions (Cuddy, 1998) are usually higher than for the 1-speed gearbox. However, the Lupo 3L on which the current vehicle models are based has a computeroperated automatic transmission that actually has a higher efficiency than the corresponding manual transmission, and the 1-speed model was therefore selected in order to avoid distorting the simulation results by use of outdated gearbox technology. Figure 4.20 shows the motor characteristics during motoring and regeneration, for the 50 kW motor used in the simulations shown in Figs. 4.17 to 4.19. The limiting curves with crosses (motoring) or circles (generating) describe torque as function of rotational speed, while the free-standing crosses indicate operating points during the mixed driving cycle assumed in the simulation. Motor efficiencies are indicated inside the operating area with lines and numbers.

Figure 4.20. Torque-speed characteristics of 50 kW electric motor during motoring (× with line) or battery charging (ο with line), motor efficiencies (numbers at lines), and points of operation (crosses) during the Fig. 4.7 driving cycle for a 20 kW fuel cell and 15 MJ Li-ion battery configuration of the Little Red Riding Hood vehicle. More operational points are in the (upper) region of motoring than in the (lower) region of generating.

The time-sequence of the efficiency of the motor-controller system is shown in Fig. 4.21 (equivalent to collecting the efficiencies of the operating points in

Fig. 4.20 and ordering them along the driving cycle of the trip).

Figure 4.21. Efficiency of motor-controller assembly for the 20 kW/15 MJ Little Red Riding Hood vehicle during the mixed driving cycle of Fig. 4.7.

The efficiency of the battery charge and discharge operations stays close to 98% except for the rapid discharge during highway driving (5000-6000 s), where it goes down to about 92%. Also, the fuel cell hydrogen conversion efficiency stays between the maximum of 59% (as assumed, see Fig. 4.5) and a low of 48% during start-up, as shown in Fig. 4.22.

Figure 4.22. Little Red Riding Hood 20 kW fuel cell efficiency over Fig. 4.7 driving cycle.

One interesting feature of the power distribution between fuel cell and bat-

tery (Figs. 4.17 and 4.18) is that the battery contribution is largest during the high-speed motorway driving and occasionally even higher than the fuel cell contribution, and that during the urban sections, fuel cells are providing at least as much power as the batteries. This is a feature of the component rating, where neither power source is sufficient on its own, but it is contrary to the conventional view of (fossil fuel-electric) hybrids, where it is intended that batteries should provide most of the urban driving and the polluting fuel most of the highway driving. For the hydrogen fuel cell-battery design, the division is unimportant, since both components are pollution-free during driving.

The simulation takes into account losses from rolling resistance of tyres against the road surface and from aerodynamic formation of turbulence, as well as the losses occurring along the entire drive train from fuel input or stored energy to the final transfer of power to the wheel axle. The size of individual losses during motoring in the mixed driving cycle is shown in Fig. 4.23, and Fig. 4.24 shows the losses during the segments in the driving cycle, where regeneration of charge to the batteries takes place. The motoring losses are dominated by the hydrogen conversion losses, although they are smaller than those of a vehicle with an internal combustion engine. For battery charging, the recovery of brake energy is incomplete, and the highest losses are brake energy not retrieved because it is escaping in the form of heat from braking discs or other equipment.

The final step in passenger car modelling is to explore the dependence on the relative component sizes of PEM fuel cells and Li-ion batteries, going from the pure battery over plug-in hybrids to self-charging hybrids (at 20 kW fuel cell rating or higher), ending at the pure fuel cell car.

Figure 4.25 shows how the charge state along the mixed driving cycle changes when passing from plug-in to grid-independent vehicles. Figures 4.16 and 4.19 already showed two situations, and in the top panels of Fig. 4.25, it is seen that a 15 kW fuel cell stabilises the (25 MJ) batteries, except during the highway part of the trip, while a 5 kW fuel cell does not change the (125 MJ) battery discharge curve much, relative to the 250 MJ battery discharge curve of Fig. 4.16. In the lower panels, it is seen that with increasing fuel cell rating and decreasing battery rating, the battery charge state excursions become more pronounced. Because a vehicle should also perform well in cases going beyond the particular driving cycle used for fair comparison between different vehicles, one would hardly reduce the battery size to values significantly below 10 MJ (for the four-passenger size of car considered). The range of hybrids considered and their calculated performance are summarised in Table 4.2 and Fig. 4.26.

Figure 4.23. Distribution of total losses during the motoring segments of the mixed driving cycle for the Little Red Riding Hood simulation (in 104 kJ).

Figure 4.24. Distribution of total losses during the battery charging segments of the mixed driving cycle for the Little Red Riding Hood simulation (in kJ).

Figure 4.25. Charge state as function of driving time in the mixed driving cycle used for simulation of selected Li-ion batteries in Little Red Riding Hood hybrid vehicles with increasing fuel cell rating and decreasing battery rating. The two hybrids in the upper panel are of plug-in type, while the two in the lower panel are gridindependent. Note that the vertical scale is different in each panel.

The hybrid simulation summary in Table 4.2 shows the total energy consumption (during driving) through the simulated trip is smallest for a pure battery vehicle, increases slightly for the plug-in hybrids and the gridindependent vehicle with the smallest fuel cell, but then increased with increased rating of the fuel cell, emphasising that electric vehicles are more efficient than even the best fuel cell vehicles. However, the table also shows the mass penalty of large batteries, making the hybrids with small fuel cells and large batteries very heavy and in consequence having shorter rage between recharging of batteries. This range more than doubles from the pure battery car to the 15 kW fuel cell plug-in hybrid car. Once into the self-recharging range, there is little advantage in increasing the fuel cell rating relative to the smallest one making the vehicle grid-independent. The battery size is modest in any case.

Table 4.2. Summary of results for hybrid fuel cell-battery vehicle simulations.

Plug-in hybrids
Fuel cell rating (kW)	0	5	10	15
Fuel cell system mass (kg)	0	43	55	68
Fuel cell energy use (MJ/km)	0	0.435	0.666	0.751
Battery capacity (MJ)	250	125	57.5	25
Battery system mass (kg)	1136	567	261	113
Battery fuel use (MJ/km)	0.617	0.263	0.1	0.028
Battery recharging range (km)	405	468	574	890
Self-recharging hybrids
Fuel cell rating (kW)	20	25	30	35	40
Fuel cell system mass (kg)	80	93	105	118	130
Fuel cell energy use (MJ/km)	0.796	0.809	0.818	0.842	1.138
Battery capacity (MJ)	15	10	10	7.5	0
Battery system mass (kg)	68	45	45	34	-

Figure 4.26. Comparison of simulated energy usage for mixed driving cycle traversed by diesel Lupo 3L, plug-in, and self-recharging hybrids (Sørensen, 2010b).

In Fig. 4.26, the energy use values found in the hybrid vehicle calculations are compared to each other and to the case of the same vehicle powered by a common rail diesel engine (the commercial VW Lupo 3L discussed in Chapter 6). The efficient diesel engine performs slightly better than the pure fuel cell configuration, but all the hybrids have better energy efficiencies, and as expected, the purely electric vehicle has the best efficiency and the plug-in hybrids with large battery components perform better than the gridindependent fuel cell-battery hybrids. The cost implications of these results are discussed further in Chapters 6 and 7.

4.2 Other road vehicles

At an early stage, interest in incorporating fuel cells in larger vehicles arose, due to the easier fitting of the (then) bulky equipment as compared with installation in passenger cars. While freight companies have shown little interest, bus companies (often under public control) were among the first to volunteer to test fuel cell technology. Presently, there are about some 100 fuel cell buses driving in regular route patterns in various cities of the world. One positive consideration is that the fixed route driving and use of dedicated filling stations have made it easy to accommodate the limited range of current fuel cell buses and to establish dedicated hydrogen filling stations at suitable locations in the test cities. Typical fuel cell bus layouts have the stacks in the back and compressed hydrogen tanks on top, as illustrated in Fig. 4.27.

Figure 4.27. Hybrid fuel cell bus, showing placement of power equipment (68 kW PEM cell, 2×75 kW motor and NiMH batteries) (MAN, 2004; used by permission.)

Practical experience is accumulating through several ongoing and completed fuel cell bus programmes. Data have been obtained−for example, for a 13-t (including passengers), 52-passenger SCANIA hybrid fuel cell bus with a fuel cell rating of 50 kW and 44 standard 12-V lead-acid batteries capable of delivering power to two 50 kW wheel hub motors (Folkesson et al., 2003). Figure 4.28 shows the energy flows in this system, based on the Braunschweig driving cycle (a multiple stop-go bus cycle with maximum speeds of 35-60 km/h). The bus is equipped with a regenerative breaking system that reduces the fuel consumption with a diesel equivalent of 9.9 l/100 km to 25.8 l/100 km. The special characteristics of the driving pattern cause the modest size power supply system to be sufficient, even with an air conditioning system installed.

Figure 4.28. Average energy flows during a bus driving cycle for a SCANIA hybrid fuel cell bus (based on lower heating value of hydrogen). (From A. Folkesson, C. Andersson, P. Alvfors, M. Alaküla, L. Overgaard (2003). Real life testing of a hybrid PEM fuel cell bus. J. Power Sources 118, 349-357. Used by permission from Elsevier.)

A large fleet of fuel cell buses were put in operation with support from an EU program: about 30 DaimlerChrysler Evobus Citaro F buses (200 kW PEM fuel cells, 1629 l of H2 compressed to 35 MPa), in route application in various European cities, including a range of hydrogen production schemes delivering to the associated filling station (Mercedes-Benz, 2004). Similar demonstration projects have been implemented in other parts of the world, including the United States, Japan, Australia, and China. Operating experience from such programmes is accumulating (e.g., Bubna et al., 2010b), including indications of fuel cell system lifetimes under different control strategies. Figure 4.29 shows the power supply sharing between an 80 kW PEM fuel cell and a Ni-MeH battery for two systems manufactured in China, during operation since 2008 in a Beijing route path. Two patterns of controlling the fuel cell use characterise the two systems: The Foton-II bus has a load-forecasting strategy, leading to periods of constant fuel cell power production with only fluctuations induced by demand, while the Foton-III bus controls fuel cell operation in an attempted instant load-following mode, causing much more strongly varying power output. It is found that the degradation of model III is larger than that of model II, and the lifetime similarly shorter.

Figure 4.29. Power distributions for hybrid city-buses operated in Beijing. Top: Foton-II bus total power (upper curve) and power provided by fuel cell (middle with plateaus), plus power to or from batteries (lower curve). Bottom: Same for Foton-III bus, except that there are no fuel cell power plateaus, due to use of a different control strategy (from Li et al., 2010; used with permission).

Use of fuel cells in two-wheeled road vehicles, such as bicycles and scooters, has been demonstrated (Huang et al., 2009; SiGNa, 2011).

Several initiatives have been taken to introduce fuel cells in specialised vehicles, for use not on public roads but privately on the premises of various enterprises. Typical tasks performed include hauling of merchandise within a warehouse (Wilhelm et al., 2011) or transporting disabled persons and luggage in airports (e.g., Munich project). For warehouse applications, fuel cell powered forklifts have been manufactured in small series. Figure 4.30 gives the test performance of such a device comprising a 5 kW PEM fuel cell, an ultracapacitor, and a lead-acid battery. The fuel cell is operated at a fairly constant power level and the capacitor takes care of most of the load variations, except for the period of high load, where the battery contributes.

Figure 4.30. Performance simulation of forklift system with fuel cell, ultracapacitor, and battery: Current contributions during operational cycle at top, voltage at bottom (from Keränen et al., 2011; used with permission).

4.3 Ships, trains, and airplanes

The use of fuel cells on ships, as well as direct hydrogen combustion, has been suggested, but so far only tried for smaller pleasure boats and for partial provision of auxiliary power for small equipment happening to be on a boat (Tse et al., 2011). A European project looked at environmental impacts from using hydrogen on ferries (see Chapter 6), and various proposals for use of PEM or SOFC fuel cells for propulsion have been forwarded. The first one realised (in 2008) was a European Commission-funded 48 kW fuel cell riverboat for use on the Alster at Hamburg (Anon., 2008).

An early proposal considers submarines, where air-independent propulsion systems may be of particular importance (Sattler, 2000). Such a submarine would need to carry both hydrogen and oxygen in containers, presumably in liquid form in both cases. Figure 4.31 shows a possible layout of such an arrangement, suitable for retrofitting of existing submarines. Suggestions for dealing with the problems of such designs have recently been made by Nikiforov and Chigarev (2011).

Figure 4.31. Proposed submarine propulsion system based on PEM fuel cells and liquid hydrogen and oxygen. (From G. Sattler (2000). Fuel cells going on-board. J. Power Sources 86, 61-67. Used by permission from Elsevier.)

A project to explore the possibilities for using hydrogen for trains has been undertaken by Tokyo Gas Co. (2004), including assessments of arrangements for fuelling. Specific use of SOFC fuel cells in a hybrid configuration with supercapacitors and batteries has been studied for a switcher locomotive (Guo et al., 2011), taking advantage of its particular driving cycle with little displacement from a fixed location (where filling stations and battery recharging can conveniently be located). A more general railway application simulates the traversal of a 40 km stretch in the UK by a train with a PEM fuel cell (470-670 kW) and a nickel-cadmium battery, and compares various hybrid and pure diesel-power configuration (Meehagawatte et al., 2010). The driving cycle is shown in Fig. 4.32, including occasionally quite substantial grades. These grades seem to present the largest modelling problem, as the parameter settings working well for the main part of the route have difficulties in reproducing performance on the large grades near the end of the trip.

Figure 4.32a, b. Driving cycle for train route between Stratford upon Avon and Birmingham in the UK (a, top: with model simulation results), and grades along the track route (b, bottom). From Meegahawatte et al. (2010); used with permission.

Another possible tracked fuel cell application is for urban tramways, metros, or monorails. Figure 4.33 shows an example of a tramway application, aimed for Seville in Spain (Fernandez et al., 2011). The system consist of PEM fuel cells rated at 254 kW, a motor rated at 540 kW, and a 34 Ah Ni-MeH battery, and Fig. 4.33 shows the urban driving cycle (speed versus time) and the total motor power drawn during the trip. The fuel cell contribution consists of a series of shifting plateaus of up to 200 kW power generation.

Figure 4.33. Driving cycle for Sevilla tramway, and power requirement (from Fernandez et al., 2011; used with permission).

The use of hydrogen in aircraft gas turbines has been proposed several times over the past decades (Jensen and Sørensen, 1984). Within the European Cryoplane project, it has been suggested that lower altitude cruising will make propulsion by hydrogen advantageous with respect to environmental impacts (Svensson et al., 2004). However, it is doubtful if this suggestion is practical in a world where air corridors are becoming overcrowded and new approaches to making better use of airspace are being sought.

A different approach is to reconsider the airship as a means of air travel. A first approach to this is considering an airship for high-altitude cruising (or as a stratospheric platform) powered by photovoltaic panels and using a reversible fuel cell system to store surplus solar power and use it when the sun is not visible. In this way, carrying possibly heavy batteries may be avoided. The envisaged relative shares of direct use of solar power, of electrolyser operation and of fuel cell power production are shown in Fig. 4.34. So far, testing of the equipment sketched in Fig. 4.34 has been performed on a 1 kW scale in the laboratory and in simulated airship conditions.

Testing fuel cell technology in a small, unmanned lightweight aircraft has been done by Bradley et al. (2007). The glider-looking plane has a 0.5 kW PEM fuel cell, capable of providing more power than the similar planes aimed for human bicycle propulsion. An example of the power disposition during maximum propulsion is given in Fig. 4.35. For cruising, the power levels are reduced by at least a third, and during descent, idling is possible.

Figure 4.34. Power modes and their timeshares for reversible fuel cell (RFC) plus solar cell system proposed as a power source for a stratospheric airship. (From K. Eguchi, T. Fujihara, N. Shinozaki, S. Okaya (2004). Current work on solar RFC technology for SPF airship. From Proc. 15th World Hydrogen Energy Conf., Yokohama. 30A-07, CD Rom by Hydrogen Energy Soc. Japan.)

Figure 4.35. Power disposition for fuel cell-propelled, light unmanned aircraft at full throttle (takeoff; from Bradley et al., 2007; used with permission).

A step up in aircraft size is the manned plane considered by Boeing (Lapeña-Rey et al., 2008). It has a PEM fuel cell rated slight over 20 kW and a similarly rated Li-ion battery. Both are required during take-off, while the fuel cell can provide enough power for cruising, as indicated in Fig. 4.36.

Figure 4.36. Performance of hybrid PEM fuel cell-Li-ion battery in light manned aircraft, shown as breakdown of contributions to power, according to pre-flight workshop tests (Lapeña-Rey et al., 2008; used with permission).

The prospect of saving fuel-storage mass in aircraft by replacing petroleum fuels by hydrogen has been discussed by Dollmayer et al. (2006), Barbir et al. (2005), and Verstraete et al. (2010). The influence of vibrations caused by turbulence on fuel cell stability and performance has been analysed by Rouss et al. (2008), concluding that such considerations should enter into the design of fuel cell-powered aircrafts.

4.4 Power plants and stand-alone systems

Stationary power generation on a larger scale may use either low- or hightemperature fuel cell systems, and several systems rated at up to a few hundred kW have been operated (Barbir, 2003; Bischoff et al., 2003; Veyo et al., 2003). The systems comprise the basic units of PEMFC, MCFC, or SOFC as described in Chapter 3, combined with fuel preparation and exhaust cleaning equipment where necessary. Placement at conventional power plant locations allows convenient accumulation of all the processes required, or access to a hydrogen pipeline where this is preferred, without having to cope with the limited space constraints of vehicles. Scenarios for the expansion of this type of system are dealt with in Chapter 5, including the special requirements in cases where the hydrogen plants serve as stores for primary energy sources such as wind or solar power. Also discussed in Chapter 5 is the incorporation of large hydrogen stores in a centralised set-up. Similar discussions of the requirements for large penetration of stationary fuel cell technologies have been taken up for Japan by Fukushima et al. (2004). In the transitional phase between the present and a fuel cell-based system, it has been suggested to lower the cost of hydrogen for vehicles by generating it from off-peak electricity from the present system (Oi and Wada, 2004). This idea is similar to the use of surplus wind power for the production of hydrogen, as described in Chapter 5, section 5.5. In conjunction with intermittent renewable energy sources, the fairly long up-start times for high-temperature fuel cell systems may be a problem, although the relevant time-interval for forecasting of, for example, wind power production can be handled rather accurately (Meibom et al., 1999).

In order to give stationary fuel cell systems an optimum performance, shortterm storage in the form of ultra-capacitors may be incorporated (Key et al., 2003). The fast response of these storage devices allows a very precise load matching. On the other hand, it is generally assumed that a mature fuel cell technology will enable power systems to respond fast enough for most normal operations. Small-scale capacitor storage is perhaps more likely to become incorporated into electric appliances, in order to make them more resilient. Because many new appliances with growing penetration in the market (such as laptop computers and smart phones) include on-board battery storage and stabilising circuits, the demand for high power-quality (suppression of frequency and voltage excursions) has generally been declining in recent years, which is good news for many of the emerging power systems, including both primary converters for wind and solar energy and also intermediate converters such as fuel cells. One already emerging niche application of fuel cells is for emergency power system (or UPS = "uninterruptible power systems").

A different kind of stationary system application is for remote power supply. Such systems are most likely based on primary energy from renewable energy converters, because long-distance transportation of hydrogen by road will increase the current problem of transportation cost, making remote power much more expensive than the same type of power in densely inhabited areas, due to the cost of transporting fossil fuels in trucks over land. Locations that can be served by waterways may not have this problem. The fuel cell system will not be particularly different from what is used in other locations, except that low maintenance requirements will have a higher priority. Use of modest lifetime MCFCs for peak-shaving has been tested (MPS, 2004).

Power plant use of hydrogen is likely for a while to be more attractive using diesel or gas engines than fuel cells, due to the much lower cost combined with the absence of the negative effects of volume requirements by internal combustion engines using hydrogen found in automotive applications (e.g., the first BMW demonstration of hydrogen cars used 12-cylinder engines or similarly bulky ignition chambers). A study has been carried out for the North-Atlantic island of Utsira, with a couple of dozens inhabitants, based on power generated by wind, used directly or for producing hydrogen by an alkaline electrolyser, and using the hydrogen either in a PEM fuel cell or in an engine similar to those used for diesel oil (Fig. 4.37; Ulleberg et al., 2010).

Figure 4.37. Power production system on the Norwegian island Utsira in the North Sea. The components are rated as follows: Wind turbine 600 kW, electrolyser 50 kW, hydrogen store 2400 Nm3 , fuel cell 10 kW, hydrogen engine 55 kW, battery 50 kWh, and flywheel 5 kWh. (From Ulleberg et al. (2010); used with permission).

Utsira is a very windy place, and the intermittency of wind power production causes need for back-up power only in relatively short periods. Figure 4.38 shows the system behaviour during such a five-hour period: When the wind power production minus electricity load drops, the electrolyser is switched off and the flywheel is employed to ensure stability of the now quite strong fluctuations in wind power, securing smooth supply by taking off surplus production and applying it during deficit, along with the engine generating power from stored hydrogen (but reacting slower than the flywheel). When wind power production again starts to rise, the engine is first shut down, and a little later, the electrolyser is restarted.

Figure 4.38. Operational data from the Utsira set-up shown in Fig. 4.37, during a 12 hour period in March (see text). From Ulleberg et al. (2010); used with permission.

4.5 Building-integrated systems

Smaller, building-integrated fuel cell systems, notably of the PEM type, have attracted much attention over recent years. This is partly related to the motion to more decentralised energy systems, where traditional energy supply has maintained a distinction between heat, which often has been decentralised (individual building oil or gas burners), and electric power supplied from central sources. The third important energy type, that of fuels for vehicles, has been delivered through supply chains ending at communal filling stations, and small-scale, portable power has been entirely supplied through the purchase of small batteries (where only the rechargeable ones have offered personal control). Fuel cells offer the possibility for individual buildings and thus their owners to become their own electricity providers and possibly also to provide individual filling stations for vehicles parked in building-attached garages (Sørensen, 2000). At the same time, waste heat from on-site power and hydrogen production may cover or contribute to providing heat/hot water to the building (combined power and heat, CPH). Finally, fuel cell technology may also replace small batteries for portable applications, allowing individuals to decentralised control all their energy supplies for heat, vehicle fuels and stationary or portable electricity uses.

Figure 4.39. Vision of building-integrated fuel cell system supplying heat, power, and hydrogen as a vehicle fuel. From Honda (2004), used with permission.

These possibilities are incorporated in some of the scenarios presented in Chapter 5. Figure 4.39 gives a vision for building-integrated fuel cell systems deriving their primary energy from renewable energy supply through electric power networks. An alternative would be hydrogen supply through pipelines in analogy to natural gas distribution or district heating networks.

A first approach to the building-integrated use of hydrogen may be replacing the natural gas boiler unit used in many countries for home heating and hot water needs by a PEM fuel cell unit with reformer and thus capable of using the existing supply of natural gas. Such fuel cell plus reformer units have already been developed and may reach a competitive cost within the next one or two decades (Vaillant, 2004; Osaka Gas Co., 2004). Test results have been forthcoming for a number of prototype installations, such as a 4.0 plus 6.8 kW H-Power combined heat and power producing unit operated in Italy (Gigliucci et al., 2004). The natural gas reformer plus PEM fuel cell installations are often referred to as micro-CPH plants, and several discussions of design and modelling have recently emerged−for example, by Beausoleil-Morrison (2010). Arsalis et al. (2011) proposes the use of high-temperature (above 100°C) PEM fuel cells. However, the possible efficiency improvement is small and heat distribution temperatures in buildings (and district heating systems) have been declining over the last several decades, making conventional PEM cells delivering heat at 50-60°C quite adequate and probably

more convenient. An interesting approach is taken by Li and Ogden (2011), adding a hydrogen compressor and making the reformer so large that it can provide enough hydrogen to cover both building electricity use (plus some heat for hot water, space heating being poorly correlated to power demand at the location considered) by the PEM fuel cell and also refill a fuel cell vehicle parked in/at the building during nights. Figure 4.40 shows the performance of the system on a typical day, under conditions prevailing in northern California, assuming a fixed-schedule, slow vehicle filling at night.

Figure 4.40. Typical performance of a residential system with a 2 kW PEM fuel cell, an 8 kW natural gas reformer (including shift reactor and gas purifier) producing 0.9 kg vehicle H2 over a 10 hour interval, plus electricity and heat. The building power demand is met and the associated heat produced mostly exceeds demand. Based on Li and Ogden (2011); used with permission.

Development of 1-2 kW combined cycle SOFC modules is ongoing, based, for example, on disc-shaped sub-stacks and integrated into a small unit with a fuel cleaner (for removing sulphur from natural gas), an afterburner (to reduce fuel in exhaust) and a battery storage (EnBW, 2004; CFCL, 2004; Kazempoor et al., 2010). The advantage of using SOFC technology is to be able to avoid the reformer. A recent study comparing micro-CPH systems based on PEM and SOFC technologies concludes that the PEM systems are more efficient in both energy and exergy respect (Barelli et al., 2011). This confirms the generally held view that SOFC technologies are interesting only for largescale plants, where the necessary high temperatures are easier to establish.

In a next phase, assuming that resources of natural gas are so limited that alternatives should be sought and/or that greenhouse gas emission issues must be taken seriously, micro-CPH units based on an established hydrogen supply infrastructure would be introduced (Erdmann, 2003; Kato and Suzuoki, 2004). Buildings would serve to both supply fuel-cell hybrid vehicles with plug-in power and hydrogen (Syed et al., 2010). As regards micro-CPH plants, one could envisage both systems with piped hydrogen supplies (say, taking over or replacing the natural gas network) and systems where the hydrogen is produced in a decentralised fashion. Such production would be particularly attractive if the capability of fuel cells to perform in reversible modes can be exploited at high efficiency (Sørensen, 2000; 2003a; and Chapter 5).

Actually, most building-integrated systems would benefit greatly from being able to use reversible fuel cells (see Chapter 3, section 3.5.5), rather than having to install two expensive components: fuel cell and electrolyser. The problem has been that reverse operation of a fuel cell optimised for power production used to have a low efficiency (around 50%), making it less efficient than conventional alkaline electrolysis (also technically a one-way type of fuel cell). However, a breakthrough in technology, so far on laboratory scale (Ioroi et al., 2004), discussed in connection with Fig. 3.52 in Chapter 3, suggests that reversible PEM fuel cell technologies may in the future live up to the expectations made in visions of widespread in-building use, such as the decentralised scenarios presented in Chapter 5, section 5.5.

An important issue for building-integrated fuel cells is the infrastructure serving the building. It might consist of an electricity grid connection and attachment to a hydrogen pipeline network. If the building is receiving hydrogen, it can use it to produce electricity and associated heat. If more electricity is produced than can be used in the building, it may be exported to the electricity grid. In this way the fuel cell capacity is used to a fuller degree, and the need for additional power stations not part of the building-based system becomes smaller. However, another mode of operation (see Chapter 5, Fig. 5.3) is to receive electricity from the grid and use it to produce hydrogen to be dispensed to vehicles parked in or at the building or to be stored for later regeneration of power and associated heat. This option is most relevant for primarily renewable energy systems with intermittent power input, because then the hydrogen can be used to produce power and heat also when no primary production is available.

This points to a further possibility in case a reversible fuel cell system is used−namely that the building has a connection to an electricity grid, but not to any hydrogen network. It would then generate hydrogen from excess electricity not used immediately in the building, store it, and use some of it for regenerating power in case of low supply and the rest of it as fuel for vehicles. In such a vision, storage of hydrogen should be possible in close connection with the building. Only when there is a hydrogen network, does a centrally located hydrogen store make sense. Building-integrated hydrogen storage may be in the form of storage tanks in vehicles that happen to be occupying the garage (if present) or parked near the house. However, this option does not guarantee availability of sufficient storage space in all cases where required. It is therefore necessary to think of other hydrogen storage options dedicated to the building. These could be compressed gas containers or, as some would prefer for safety reasons, metal hydride stores. As the scenario in Chapter 5 will show, only modest volumes of such storage (about a third m3 for a family home) are required for handling the fluctuating energy production of a wind-solar primary energy production system, including consideration of losses in all components of the system. Aki et al. (2004) have also pointed out that there are system stability advantages obtained by using a local hydrogen pipeline grid to interchange hydrogen between individual buildings.

4.6 Portable and other small-scale systems

Consuming patterns have over recent years seen a dramatic increase in the use of portable equipment for entertainment and work (such as music and video players, laptop computers, and mobile phones with multi-functionality). This has increased the demand for batteries, but at the same time revealed limitations of the battery technology that seem difficult to avoid despite the steady but necessarily modest increase in conversion efficiency. Fuel cells with small-scale stores would be an obvious solution to these problems, because the technical performance in one important area already surpasses that of batteries, namely that fuel cell operation of a state-of-the-art laptop computer can offer autonomy for days rather than a few hours. The difference between these otherwise similar technologies is the external storage of chemicals for a fuel cell versus the internal storage in batteries. This is also the bottleneck, because direct storage of hydrogen would be inconvenient (high compression needed for pocket-size containers of suitable capacity), and use of portable reformers constitutes an added component presenting its own concerns. Two avenues are pursued at the moment: avoiding reformation by use of a direct methanol fuel cell, or development of a micro-reformer working on a substance with high energy density, both as regards mass and volume.

Battery replacement technologies are also sought for appliances other than laptops, both with smaller energy use (e.g., smart phones) and with larger energy requirements (from garden equipment to military portable weaponry and devices for intelligence and communication). A cellular phone in use will need less than 200 mW of electric power, a video camera will need less than 6 W, and a laptop computer or a portable CD player will typically need less than 20 W. The more specialised portable equipment includes field environmental monitors, medical mobile life-support systems, and soldier communication and signalling devices used in military operations (Palo et al., 2002). Power requirements for such equipment would typically be in the range of 10-500 W. A standard size Li-ion battery has a storage capacity of 750 mAh, and the largest one convenient for a laptop computer has a capacity of 3600 mAh. For a portable video camera, the standard Li-ion battery may deliver 5 W at 7 V for an hour or two, while for the laptop at 12 V, current batteries provide about 4 h at an average power consumption of 10 W.

The current high price paid for small-scale Li-ion batteries today (some 15-25 euro or US\$ per kWh for typical devices for cameras and laptops, with a 4 year lifetime, as opposed to the lower but still significant prices for the larger batteries used for craftsman tools or electric vehicles) makes portable applications an inviting market for alternative technologies, and particularly those that may offer prolonged autonomy.

The options that may be considered for portable fuel cell applications include PEM fuel cells with compressed hydrogen canisters, metal hydrides, or fuels such as methanol requiring use of a reformer, and as mentioned direct methanol fuel cells. Compressed hydrogen at 30 MPa and the best metal hydrides have an energy density on a volume basis (Chapter 2, Table 2.4) of 2.7 and 15 GJ m⁻³ , while that of methanol is 17 GJ m⁻³ . These should be compared to the energy density of Li-ion batteries, which is 1.4 GJ m⁻³ (Sørensen, 2010a). The implication is that by adding a 50% efficient PEM fuel cell, the performance of a device using stored compressed hydrogen at up to 30 MPa plus a fuel cell is not better than that of a Li -battery, while the other possibilities listed at least theoretically may outperform the best batteries.

The consideration of size (volume) is relevant for portability, but even more is the energy density by mass, as humans have to physically carry the devices around (a tautology for "portable" equipment). In fact, lower mass would be a general advantage for many laptop computers that in their present form are more "towable" than "portable" (one reason that the market is currently splitting up into desktop replacements, laptops, notebooks, tablets, and so on, basically just signalling weight and distinguishing themselves from the smart-phone and GPS devices without a keyboard of anatomically acceptable dimensions). Without the mass of the container, the energy density of hydrogen in any form is 120 MJ kg⁻¹ , but for storage in metal hydrides, the overall density goes down to under 9 MJ kg⁻¹ (Chapter 2, Table 2.3). For methanol the value is 21 MJ kg⁻¹ , and for Li-ion batteries it is 0.7 MJ kg⁻¹ (Sørensen, 2010a). This comparison on a mass basis makes most fuel cell solutions preferable to Li-ion batteries (and much more to lead-acid batteries 5 times lower in energy density), as long as the volume can be tolerated.

For the best metal hydride stores, the capacity (hours of autonomous operation) could be increased some 5-fold over Li-ion batteries of the same volume, and this option has attracted some attention. Güther and Otto (1999) describe a Siemens-Nixdorf experimental laptop computer with such a store, obtaining a 3-fold increase over a Li-ion battery. Recently, a 2 W smart-phone re-charger consisting of a small PEM fuel cell and a replaceable 15 Wh MeH cartridge has been marketed (Geek with laptop, 2011). Non-metal hydrides should be able to reach even better values.

The use of methanol and a PEM fuel cell in combination with a miniature reformer has been contemplated by several Japanese electronics industries and in a series of US army projects (Palo et al., 2002; Patil et al., 2004). In the early project, an energy density of 2.6 MJ kg⁻¹ was found for a 15 W, 1 kg portable power device with a small auxiliary battery used to start the reformer system up. From the next project, Fig. 4.41 shows a 40 W micro methanol-reformer. In 2006, still with US Army funding, the company UltraCell developed another methanol cartridge-based micro-fuel cell capable of running a laptop for two or three days. Today, versions from some hours to a month of rugged, autonomous operation are offered, still primarily for military use (Fig. 4.42; UltraCell, 2011). Various theoretical models have been constructed to deal with optimisation issues (Besser, 2011), and ways of avoiding environmental emissions were already suggested at an early stage (Muradov, 2003).

Figure 4.41. Prototype of 40 W methanol-tohydrogen reformer, produced for the US Army by Pacific Northwest National Laboratory. The coin appearing in the picture has a diameter of 24 mm (Patil et al., 2004). Used with permission.

Figure 4.42. Left: Ultracell XX25 hydrogen fuel cell with methanol cartridge and reformer, for rugged military usage (2008-2011). Middle and right: NEC direct methanol prototype fuel cells with methanol cartridge (2003 and 2004 design). From UltraCell (2011); Y. Kubo, NEC Corp. (2004); NEC (2011). Used with permission.

The good energy density of methanol can be used without the extra reforming component, by using a direct methanol fuel cell, described in Chapter 3 section 3.6 (see Fig. 3.69). One drawback is the lower conversion efficiency of methanol PEM cells relative to their hydrogen cousins. This route is being pursued (Meyers and Maynard, 2002; Zenith et al., 2010), but methanol/DMFC devices are not only less efficient but also further from commercial readiness than the reformer/PEM fuel cell combination. Key factors in making the DMFC work are ambient humidity, condenser temperature, and excess air, according to the model calculations by Zenith et al. (2010).

Early prototypes of a notebook personal computer powered by a micro-DMFC are shown in Fig. 4.42 (middle and right). In the 2003 model, a replaceable methanol cartridge is placed behind the computer, and the 280 cm² DMFC is placed under the laptop keyboard. The fuel cell produces 14 W at 12 V, and if the methanol volume is some 30 cm3 (judged from photo), then the storage capacity is 142 Wh. This would allow the computer to be operated for 14 h at an average consumption of 10 W. A "docking station" design was exhibited in 2004. Both Japanese and Korean computer manufacturers presented several fuel cell notebook prototypes between 2003 and 2006, and sometimes with promises of "next year" entrance into the commercial market, but none of them have by 2011 been appearing in shops. However, the German company Efoy/SFC Energy is marketing a range of DMFC units (e.g., model 2200 XT: 90 W, weight 9 kg, capable of delivering 2160 Wh of electricity over 24 hours) aimed primarily at recreational uses, such as in pleasure boats (Efoy, 2011). All the portable fuel cell units discussed use a small battery for starting the system, a process often taking 10-20 minutes.

In order to take advantage of standard PEM hydrogen fuel cells without reforming, a chemical reaction concept based on sodium silicide has been employed to generate hydrogen for a small PEM fuel cell, first for bicycle use (500 g furnishing some 200 W over 50 km) and recently for off-grid recharging of small portable devices (Fig. 4.43; SiGNa, 2011; MyFC, 2011). The reaction scheme is (NSF, 2011; Lefenfeld et al., 2006)

2 NaSi + 5 H2O → Na2Si2O5 + 5 H2 + 350 kJ/mol

Figure 4.43. PowerTrekk PEM fuel cell for recharging small appliances, using stored NaSi to produce hydrogen, when water is added, and to recharge the Li-ion battery that serves to allow instant start-up (from MyFC, 2011; used by permission).

The production of NaSi from sand and kitchen salt should be straightforward (Modic, 2011), and the exogenic process producing hydrogen is quite fast but easily controllable, in contrast to some other sodium reactions.

Other types of fuel cells, such as microbiological ones, have also been contemplated for small-scale, portable use (Dunn-Rankin et al., 2005), but prospects are poor due to the extremely low overall efficiency.

4.7 Problems and discussion topics

4.7.1

How much hydrogen must an airplane be able to carry for a flight from London to Tokyo?

4.7.2

Discuss the optimum rating of fuel cell and battery for a hybrid car, as a function of the relative price of batteries and fuel cell equipment. Note that the energy rating of a battery is a poorly defined parameter. The reason is that the energy released from a battery depends on the discharge rate and hence on the driving cycle (Jensen and Sørensen, 1984). Manufacturers rarely mention this when they quote energy stored E in kWh or charge, C = E/V, stored in Ah. Here V is the potential across all units connected in series (some systems use a combination of parallel and series connection of individual battery module units).

4.7.3

Write a list of those power-consuming activities in your private or work life that you believe cannot ever be expected to become powered by batteries or portable fuel cells.

Based on the fraction these activities constitute of the total power usage, estimate which percentage of total electricity use it would be possible to base on renewable energy sources such as wind or photovoltaic power, used solely to charge either batteries or fuel cells (two separate estimates).

CHAPTER 5 IMPLEMENTATION SCENARIOS

5.1 Infrastructure requirements

In this section, the different components required for using hydrogen in various systemic environments are viewed as a network of infrastructure that has to be convenient and feasible as judged by technical and economic analysis. In particular, the additional components required to connect the system components are identified and assessed.

5.1.1 Storage infrastructure

Traditional energy sources may be divided into those that employ storage and those that do not. Fuels like wood, oil, natural gas, or coal are conventionally stored at various stages from production (extraction) to use: central stores, intermediate stores at retail traders and decentralised ones at the enduser. Because of its gas form (implying higher storage volumes), natural gas has in its early years of use been distributed (by pipeline) without storage, whereas presently, containers (for intermediate storage on land or for liquefied natural gas during ship transport) or underground storage is usually employed to manage any mismatch between the most economical rate of production and the patterns of use. On the other hand, electricity is largely produced, distributed, and used without any storage components (except where hydro stores are available), implying that power plants have to be able to cope with the variations in demand. Increasingly, small-scale users demand storable power, causing a widespread use of batteries (simple leadacid or advanced high-power density versions such as lithium ion batteries), originally only for portable applications, but increasingly also for electronic equipment used in quasi-stationary environments (from transistor radios to laptop computers used at distances from grids of a few tens of metres).

The situation changes dramatically if energy sources either with variable

availability or which need production methods preferring constant production are put into use. With renewable energy sources such as wind power or photovoltaic power, an autonomous system requires integration of storage, and, if the system is not local, also transmission. Such storage is the same as that considered with conventional electricity systems, in that it is not kinetic energy or radiation which is stored, but produced electricity that is converted into an energy form storable, in order to be ready for later regeneration of electric power. Hydrogen is one of several possible storage options catering to electricity storage, by storage as a fuel. It is in competition with mechanical and electrochemical storage options, and with direct storage of electricity in superconducting rings (see Sørensen, 2010a).

Alternatively, one can view hydrogen as the basic energy carrier of a given system and then ask how intermittent sources can be used to produce the hydrogen, and how hydrogen can be used to cover society's variable demands of different forms of energy. The layout of such systems determines if it may be convenient to consider centralised storage facilities or if storage is better done locally near the end-user. For automotive uses, there clearly has to be onboard storage of hydrogen if this is the energy carrier used by the car system.

The implications for infrastructure are that hydrogen produced centrally will typically be divided into two parts: one going directly into pipelines to points of use and another being stored at suitable (large) storage facilities. These are likely to be compressed gas stores, because liquefaction entails considerable energy loss, and storage types such as batteries, hydrides, or other chemical storage are unsuited for large-scale storage. When a hydrogen transmission pipeline system exists, there is no particular reason to store hydrogen near the point of use, and availability of cheap storage options is the overriding consideration. Here the solution is likely to be the same as used for natural gas, that is, underground storage in suitable geological formations. In many regions, it is possible to find such formations that allow inexpensive formation of the cavity for holding hydrogen, implying that the volume of the store becomes less important and therefore the pressure can be kept low in order to minimise leakage and avoid expensive lining to tighten the cavity.

Figures 2.31a and c in Chapter 2 showed two such options, both currently in use for natural gas storage. One takes advantage of the salt intrusions found in many areas characterised by ice-age clay deposits over areas once covered by oceans. The Danish installation of the type shown in Chapter 2, Fig. 2.31a was formed by flushing the salt out of the intrusion over a period of a year or more, at very low energy expenditure. The 7.2 × 10⁸ m3 underground cavity formed allows gas storage cycles corresponding to a working volume of 4.2 × 10⁸ m3 to be employed at a pressure interval of 16-23 MPa and a temperature of about 45°C (DONG, 2003; Energinet.dk, 2007). If used for hydrogen, a pressure of 5-10 MPa would be employed. Several dozen additional underground salt domes of this kind exist in Denmark as well as in other countries with glacier deposit geology, for possible future use.

Figure 5.1. Hydrogen storage in an aquifer, with electrolysis and hydrogen recovery plant above, receiving electric power and delivering hydrogen to filling stations−for example, through pipelines.

The underground formation shown in Chapter 2, Fig. 2.31c and (in a system setting) in Fig. 5.1 is an aquifer, which is a water-carrying (sand-based) layer trapped between two impenetrable (clay) layers. Such aquifers exist in most of the world and usually bend up and down in a way that in some cases allows them to store gases in an upward bend, again with very modest capital investments. Factors determining the suitability include the motion of water along the aquifers. A Danish installation of the type shown in Fig. 2.31c (total volume 12 × 10⁸ m3 ) can store 3.6 × 10⁸ m3 of natural gas at a pressure of 17 MPa (this being the pressure at extraction, with higher pressures at the bottom of the cavity, as in the salt dome case; DONG, 2003; Energinet.dk, 2007). Again for hydrogen, it would be prudent to employ a lower storage pressure, because of the higher permeability of the surrounding geological formations for hydrogen than for natural gas. The two Danish underground stores could hold the equivalent of some 14 + 8 = 22 PJ of stored hydrogen.

Decentralised storage may invoke other of the storage solutions described in Chapter 2, section 2.4. For building-integrated storage of hydrogen, compressed gas containers are already in use industrially. However, for home and high-rise city buildings, safety considerations may make metal hydride or similar stores preferable, should they become technically and economically viable. For a family dwelling, a hydride store may be buried under the house or under the garden, much in the same way as oil containers are today (Sørensen et al., 2001, 2004).

For automotive applications, all the storage options mentioned in Chapter 2, section 2.4 are in consideration, although most of the presently emerging fuel cell vehicles use compressed gas stores. Some prototypes have used liquefied hydrogen (combustion engine hydrogen cars typically requiring larger amounts of fuel) or cryogenic activated carbon (acid fuel cell cars). Acceptable sizes of 30 MPa compressed hydrogen stores in small passenger cars are considered to be at most 5 kg of hydrogen, currently giving driving ranges of about 450 km, which is a little lower than the average of the present stock of gasoline-driven cars. This has spurred an effort to construct containers for around 70 MPa pressure with the same level of safety as the 30 MPa units in widespread use today for industrial hydrogen, with an excellent safety record. For buses and lorries, it is usually possible to integrate enough hydrogen storage to match the driving ranges of the current fleets of such vehicles. Metal hydrides are unlikely to be used in mobile applications due to high weight, but some of the chemical hydrides under study could be considered if they are found viable in the technical and economic sense.

Storage at filling stations poses problems of a nature similar to the other stationary stores just mentioned. Current hydrogen filling stations (see Fig. 5.1 to the right) are parts of demonstration programmes, e.g., for fuel cell city buses, and use compressed gas storage.

5.1.2 Transmission infrastructure

Hydrogen transmission techniques were discussed in Chapter 2, section 2.5. Leaving alone intercontinental trade of hydrogen (the ship transport option), long-range transport from the area of production (e.g., a park of wind turbines) to the area of large users or hubs of distribution to end-users (such as filling stations) is most likely done by pipeline. This could be new, dedicated hydrogen pipelines, or it could be natural gas pipelines upgraded to use for hydrogen at a time when natural gas is no longer available or in demand (Sørensen et al., 2001). The current Danish transmission grids for electricity and natural gas are indicated in Fig. 5.2, along with the gas stores and the location of existing or planned offshore wind parks, the locations of which are important for the grid infrastructure, also in a hydrogen-based energy system, due to the necessary production of hydrogen by excess wind power production. The Danish scenario discussed in section 5.5 takes advantage of the existing natural gas pipeline transmission system consisting of the major grid shown in Fig. 5.3, and, not shown, minor pipelines to the filling stations that are also indicated in Fig. 5.3. The two storage facilities just mentioned remain operative, now for hydrogen. Clearly, if hydrogen and natural gas systems would be in use at the same time, two parallel transmission systems would be required. Small amounts of hydrogen (around 10%) can easily be added to the natural gas transmitted in present lines, but this is not of interest for use of hydrogen in the transportation sector.

Figure 5.2. Structure of the present Danish natural gas and electricity transmission systems. Gas stores are indicated, as are areas contemplated for offshore wind. Nearly a thousand wind turbines have already been or are being installed in the two areas furthest to the west and to the south.

Discussion of transmission infrastructure in connection with surplus wind power or other off-peak power generation has also been made for Ireland (González et al., 2003), for Japan (Oi and Wada, 2004), and in connection with coal-based hydrogen production for China (Feng et al., 2004).

5.1.3 Local distribution

As in the case of general transmission lines, the local natural gas distribution lines reaching individual buildings in urban areas can with little change be used for hydrogen. This allows a decentralised hydrogen scenario such as the one described in section 5.5. Key cost items will be the in-building installations, which include interface with hydrogen stores, with fuel cell equipment, and possibly with the storage tanks of vehicles parked in a garage at the building. There is little doubt that the cost of placing such installations decentralised in buildings will be higher than for more centralised facilities covering the same needs, but traditionally, consumers have been willing to pay a premium price for what is perceived as individual control over the technology (here energy supply for home and vehicle usage).

Figure 5.3. Infrastructure envisaged in a 2050 scenario for hydrogen use in Denmark, with emphasis on the transportation sector, including two underground hydrogen stores, pipeline transmission and distribution of hydrogen to motorcar filing stations, seaports and airports, as well as to centralised fuel cell power plants. The scenario, which is further discussed in section 5.5, uses electricity for railroad operation and electric cars in the city of Copenhagen (Sørensen, 2003a).

5.1.4 Filling stations

Hydrogen filling stations for buses, passenger cars, and special vehicles cur-

rently exist worldwide in connection with demonstration projects (some 20- 50 units), but of course have to be present in approximately the same numbers as gasoline and diesel fuel filling stations today, for a future with extensive use of hydrogen in the transportation sector. It is sometimes argued that the current number of filling stations is larger than required, and the number is on its way down in several countries. Whether this means that the more expensive hydrogen filling stations can be reduced in number is unclear, as the hydrogen vehicle range may be lower, indicating a denser network of filling stations. The issue is linked to questions of storage pressure and mass of hydrogen storage in future commercial fuel cell vehicles. Figure 5.3 shows an example of filling station density, based on the assumption that the fuel cell vehicle driving range can be kept at the same level as for the current average (but not aiming for the larger range of over 1000 km characterising the fuel tank size of the most fuel-efficient current passenger cars, e.g., Volkswagen Lupo 3L). Most likely, as suggested in Chapter 4, hydrogen vehicles will be hybrids, and the amount of hydrogen to carry onboard thus smaller.

Figure 5.4. Layout of a decentralised, building-integrated hydrogen and fuel cell system based on intermittent primary power sources (such as wind or solar energy), reversible fuel cells and local stores, including stationary and maybe vehicle-based stores, and possibly capable of interchanging hydrogen with users in other buildings through pipelines (Sørensen, 2002a).

5.1.5 Building-integrated concepts

As mentioned in section 5.1.1, building-integrated hydrogen systems may possess storage in one or both of the following forms: (1) stores in vehicles parked in or at the building and (2) stationary storage in the basement or under the building or in its surroundings, such as metal hydride storage or pressurised containers. Decentralising storage means that equipment for filling and extracting hydrogen from the stores must be available in each building possessing such stores. Although use of the tanks in the vehicles parked at the building may seem attractive, the additional cost of providing a filling station for pressure upgrade of pipeline hydrogen (presumably at considerably lower pressure than the automobile storage containers) would presumably be too high. The problem would be much smaller for metal hydride stores, that need only a modest pressure above ambient for the charging and discharging operations (in the latter case also a modest temperature increase).

The building integration concept used in the decentralised scenario of section 5.5 is shown in Fig. 5.4. The concept is based on the availability of reversible fuel cell technology, which as mentioned in Chapter 3, section 3.5.5 seems a fairly realistic assumption, at least for a future scenario. At an increased cost, the reversible fuel cell could of course be replaced by a pair of a powerproducing PEM fuel cell and an electrolyser cell. Both storage options (underground or in vehicle) are left open, and the heat from the fuel cell operation is envisaged as contributing to or covering the heat loads of the building. Full coverage of space and hot water demands would be possible even with highly efficiency fuel cells (less excess heat) if the building is insulated and otherwise energy-optimised to the best current standards (Sørensen, 2010a).

Finally, portable fuel cell applications may entail infrastructure requirements in terms of networks of locations (e.g., shops) selling hydrogen storage units for exchange like batteries or offering to recharge hydrogen or methanol canisters. As for vehicle hydrogen containers, high-pressure filling is unlikely to be decentralised to home level (in contrast to battery recharging).

5.2. Safety and norm issues

5.2.1 Safety concerns

The history of technology reveals that safety and risk issues have often been handled with a mixture of attitudes and approaches, including consideration of the following (Sørensen, 1982).

Direct risk, defined as the probability of an accident (or other risk-related event) times the magnitude of the consequences.
Social risk, defined as the damage inflicted upon society by an accident or risk-related event.

• Perceived risk, defined as the average public perception of the severity of a given risk.

The difference between direct and social risks is particularly clear in cases where the magnitude of the consequences is such that a given society has difficulties in handling them (for instance, a disruptive accident with many casualties affecting the functioning of the administrative headquarters of the society). Perceived risks are of course as real as the technical ones when it comes to influencing political decisions regarding technology choices. This is independent of whether the perceived risks are based on real or false information.

A well-known case is the detrimental influence on the development and use of hydrogen in the transportation industry caused by the Hindenburg accident in 1936. The Zeppelin company chose to blame hydrogen for the accident, although they seemed to have known that the cause was the use of a highly flammable compound in waterproofing the blimp cloth, which was further electrically isolated from the airship frame, so that frictional spark ignition during weather patterns with thunderstorms could start an explosive fire. It has recently been concluded that the accident would have evolved identically had the airship been using helium rather than hydrogen (Bain and van Vorst, 1999).

Public perception has also in recent years slowed the expansion of nuclear power to a halt. Also in this case, the defensive attitude of the industry involved probably played as decisive a role as the public criticism. Turning points here were the 1986 Chernobyl accident in the former Soviet Union and the 2011 Fukushima accident in Japan. The former nullified all previous industry statements of "the worst statistically significant accident being only of local consequences", by causing a global fallout of nuclear radiation-active substances and bans on particular isotope-accumulating food many thousands of kilometres from the accident site (Sørensen, 1979b, 1987). In this case, the public perception turned out closer to reality than the theoretical accident models used to calculate direct risk for similar installations (Rasmussen, 1975). The recent Fukushima accident exposed the hubris of a reactor industry building nuclear power plants on top of seismic faults in tsunamiprone locations, claiming that everything was under control.

It is clear that public attitude to "small" and frequently occurring accidents is different from the one towards catastrophic accidents. Killing thousands of people in a large number of individual traffic accidents every year has not lead to a ban on motorcars or to a halt on technology development, as did the 35 people killed in the Hindenburg accident. The logic may be difficult to appreciate, when one looks at airplane accident rates which over the past 50 years have diminished from near 10⁻⁴ to 10⁻⁶ chance of perishing in an airplane accident per trip taken. Airplane accidents are also in many cases catastrophic, claiming several hundreds of victims at a time. Yet there has been no reaction in the direction of closing down air travel in general. It is sometimes claimed that this has been due to the already established acceptance of automobile transportation (and that this might not have been forthcoming, had the true accident tolls been known around 1900, when the automobile made its entry), but this does not explain the negative effect of the Hindenburg accident on hydrogen development. By 1936 the death toll of automobile driving was well established.

Figure 5.5a. Photo from accident report No. 2 (1984) by Svenska Miljöverket; Stockholm County Police has confirmed that the photo is in the public domain. Hydrogen canisters are encircled in black.

Figure 5.5b. Analysis of the hydrogen accident in a Stockholm street 1983 depicted above: flow pattern and H2 concentration (shaded areas) after 10 s in a vertical plane halfway between the canisters on the truck and the building wall. (From A. Venetsanos, T. Huld, P. Adams, J. Bartzis (2003). J. Hazardous Mat. A105, 1-25. Used by permission from Elsevier.)

Although some of these questions have to be left unanswered, they clearly have significant implications for a second attempt to introduce hydrogen in the energy system. Are there risk-related events that could halt the introduction of hydrogen and fuel cells in the transportation and the domestic sectors, even if the problems related to economy and technical performance have found a solution? This has considerable bearing on the issues, discussed briefly below, of common norms and standards to be applied world-wide, but it also has non-technical dimensions related to openness on the part of the industries and controlling authorities involved and on improving the understanding of the factors influencing not only direct and social but also perceived risk.

Industrial use of hydrogen has experienced a few accidents, which subsequently have been useful in defining norms and procedures. One occurred in a narrow Stockholm street in 1983, where 13.5 kg of H2 escaped from a set of 20 MPa pressure tanks with defective connections and exploded (Fig. 5.5a): 16 people were injured, and 10 cars and the adjacent building were heavily damaged. This accident has recently been modelled by computational fluid dynamics methods, giving the distribution of H2 velocities and concentrations as shown in Fig. 5.5b (lower flammability limit is at a volume concentration ratio of 0.04 in air, delimiting the mid-grey area in the figure).

A hydrogen road accident occurred in Germany in 2001, with a truck crashing into a trailer loaded with hydrogen canisters. Escaping hydrogen ignited and caused a fire in the truck, killing the driver. Firefighters arrived quickly and were able to prevent further hydrogen release by spraying water onto the remaining canisters for several hours (Wurster, 2004).

5.2.2 Safety requirements

The approach to handling safety issues for hydrogen applications in the energy sector, affecting society broadly in a number of different ways, requires consistent standards in several areas from production to use and also requires a way of handling the unanticipated safety-related events bound to occur in a technology that is significantly different from that currently used.

Starting with production of hydrogen, this is a conventional industrial activity already in place, including safety prescriptions related to production by steam reforming or by alkaline electrolysis. The implications of raising the pressure of electrolyser to a proposed 12-20 MPa have spurred new safety investigations, focussed on hydrogen-oxygen explosive reactivity (Janssen et al., 2004). For biological production of hydrogen, little safety-related work has been performed, as the would-be industrial processes are largely unknown.

The majority of hydrogen safety studies focus on the properties of hydrogen as such, in relation to burning, to explosions and to diffusion into unwanted locations. The basic properties of hydrogen were compared to those of other fuels in Table 2.3 in Chapter 2, section 2.3.3. Some of these numbers are still being refined by new measurements−for example, the flammability safety limits (Chan et al., 2004).

Confined and semi-confined (vented) hydrogen gas explosions are being studied in experimental settings (Carcassi et al., 2004), and by simulation. Of particular concern is garage spaces used for hydrogen vehicles and perhaps housing a decentralised filling outlet, and hydrogen transport trough tunnels, because of the large number of vehicles and people potentially involved (Breitung et al., 2000).

Figure 5.6. Closed space H2-release in 6 × 3 × 2.8 m3 garage at a flow rate of 5.6 litre min-1. Based on data from Hayashi and Watanabe (2004). Hydrogen safety for fuel cell vehicles. In 15th World Hydrogen Energy Conf., Yokohama.

Figure 5.7. With the purpose of determining the best ventilation arrangements for a domestic FC-car garage, dispersal following a leak was calculated by several fluid dynamics software codes (the acronyms in the figure distinguishes the models employed). Shown is the predicted volume of air-hydrogen mixtures at flammable hydrogen concentrations (4−75%), and the mass of helium in an experiment performed with helium substituting for hydrogen. The differences indicate the error range of currently available software tools (Papanikolaou et al., 2010; used with permission).

Several studies have looked at leakage from a hydrogen tank in a vehicle parked in a closed garage, as well as on flame characterisation by punctuation of a hydrogen container (Hayashi and Watanabe, 2004; Fig. 5.6) or hydrogen installations in buildings (Tchouvelev et al., 2004), using computational fluid dynamics calculations. Alternatively, scale experiments may be performed in wind tunnels, a method also used to assess the hazards of spills at hydrogen fuelling stations (Chitose et al., 2004). A recent European Union project has looked at garage releases from vehicle or store by a number of computational (Papanikolaou et al., 2010; Fig. 5.7) and experimental methods (Friedrich et al., 2011; Royle and Willoughby, 2011), such as the samples of rapid-sequence optical photos shown in Fig. 5.8. Hydrogen release in a garage-like space has also been studied by Merilo et al. (2011).

Figure 5.8. Test of spark ignition of a hydrogen release, simulating a pipework-failure induced accident in a high-compression hydrogen store. Out of a sequence of 1000 pictures per second at optical wavelengths, a shows the situation 296 ms after ignition, b 390 ms after ignition (from Royle and Willoughby, 2011; used by permission).

Figure 5.9. Simulated hydrogen diffusion following tunnel accident releasing 7 kg of H2 in 15 min. The H2-distribution is shown at 900 s, just before ignition. The H2 volume-concentrations drops from 9 to 4% from the first to the third car relative to the release point. From Breitung et al. (2000), used by permission from the IAHE.

Figure 5.9 shows simulation results for the tunnel release. As expected, the hydrogen accumulates at the ceiling of the tunnel or room. As a consequence of several tunnel accidents in Europe (not related to hydrogen), the release of hydrogen and possible ignition in the confined space of road tunnels are seen as a key safety concern. Queues of cars may be trapped in tunnels unable to escape from the walls of fire travelling through the narrow confinements of typical tunnel structures. The simulations of hydrogen dispersion in a simulated hydrogen car accident within a tunnel shown in Fig. 5.9 takes departure in the liquefied-hydrogen containers used, for example, by BMW in their combustion engine hydrogen prototype cars (FZK, 1999). Hydrogen concentrations initially around 15 vol% are found to disperse by several car lengths in a few hundred seconds, before igniting.

All components of hydrogen systems for vehicles (fuel container, handling equipment, fuel cell) must be assessed for safety under the conditions of both normal operation and accidents involving collisions between the car and other stationary or moving objects. It is well known that road traffic gives rise to a very large number of accidents, including many that cost lives. Table 5.1 gives an idea of the fatality rates in present traffic systems.

The numbers in Table 5.1 are from Germany and imply a person fatality rate of 10⁻⁴ y⁻¹ or 4.8 × 10⁻⁵ y⁻¹ per car registered in Germany (EEA, 2002). A similar distribution is found in the United States, except that the per capita fatality rate is 1.6 times higher than in Germany. Since most accidents are caused by driver errors, auto-pilot technology could presumably reduce the rate of accidents, but it is clear that introduction of new automobile technology such as hydrogen and fuel cells should as a minimum not aggravate the outcome of a given accident, relative to the present car technology. It seems reasonable to aim for strongly reducing the hydrogen-specific accident types involving leakage, fires, or explosions to negligible levels, rather than to just make them rare compared to the "conventional" accidents, as these are already unacceptable to many observers.

German traffic fatalities 1999:	7842
Solo accidents:	40%
Collisions with pedestrians:	13%
Collisions with other vehicles:	47%

Table 5.1. German fatalities in car accidents 1999 (out of a total number of road accidents with person injury amounting to 397 689) (with use of data seen in Daimler-Chrysler, 2001).

Specific safety analyses are made for hydrogen vehicles used in public transportation, such as city buses, as it is generally accepted that per passenger accident and fatality rates should be lower in public transport than for individual driving (Perrette et al., 2003).

Also, the non-hydrogen components of the hydrogen fuel cycle should be assessed for risks, such as equipment for fuel processing. This would be relevant for technology using methanol−for example, for reforming or DMFC operation. The health impacts of methanol are debated, in addition to questions being raised on CO2 emissions, primarily related to the possibility of inhaling fumes of methanol during filling of fuel into a vehicle or a container. Similar risks exist with gasoline or diesel fuels, which have led to the introduction of cuffs around the fuelling nozzles. The toxicology of methanol has recently been reassessed (Center for the Evaluation of Risks to Human Reproduction, 2004), offering an improved basis for assessment of the risks from using methanol in connection with fuel cells.

5.2.3 National and international standards

The introduction of a new technology is very much eased if the same norms and standards are applicable in all the markets of interest. In the past, differences in norms and standards have cost the consumers large sums of money (for example, in connection with video recorders, battery physical design, software and computer operating systems, information storage media), and particularly in the automotive market, there are still no world-covering rules allowing an unmodified model to be sold everywhere. Marauder companies use their reluctance to comply with global norms as a branding tool, showing plain contempt for the consumers.

In order to minimise the already substantial problems associated with introducing hydrogen vehicles, international norms and standards should be established in the early phase of finalising the technical designs, with eventual refinement as the technology matures. The same is of course true for other uses of hydrogen and fuel cells, such as the home units that may replace natural gas boilers.

The international scene for establishing common codes and standards is comprised of a number of United Nation work programmes (with regional sub-organisation, such as the Economic Commission for Europe) and the International Standards Organisation technical committees (Dey, 2004). Some individual countries have long traditions for such work, and many national standards working groups exist (e.g., as the basis for the Deutsche Industrie Norm). Often, smaller countries in a region lean towards these lead countries when establishing norms and standards. Finally, many industry sector organisations and professional organisations conduct work on norms and standards. In Europe, this type of work is performed for hydrogen by the European Integrated Hydrogen Project, globally by the US proposed International Partnership for Hydrogen Economy (Ohi and Rossmeissl, 2004), and by the Society of Automotive Engineers. In Japan, the Japan Automobile Research Institute performs norm work based on a number of specific legislative initiatives in the areas of collisions, container and fire safety, underground parking lots and road tunnels, and so on (Hayashi and Watanabe, 2004; Kikuzawa et al., 2004).

As in other industrial areas, the establishment of a safety-conscious culture is a very important goal. Efforts to make designs with high degrees of fail-safe operation are important, as are frequent reviews and formal hazard modelling and evaluations, in areas such as hydrogen leak detection, spontaneous ignition, and flame and shock wave progression, for a variety of different types of confinement and accident definitions.

5.3 Scenarios based on fossil energy

5.3.1 Scenario techniques and demand modelling

Scenarios are snapshot pictures of a future situation, with sufficient detail to allow them to be checked for consistency. Scenarios are not forecasts, but rather an exploration of various different visions considered attractive or for some other reason being discussed by a group of people in society. Wellthought-out scenarios are important tools for decision-makers by offering the possibility of orderly change from one situation to a new one, an option that conventional economic theory is unable to offer, as it is based on trend forecasts departing from analysis of the past and application of economic rules also derived from past behaviour.

Visions of hydrogen as a fundamental energy carrier have been held for a long time (Bockris, 1972; Sørensen, 1975). Hydrogen implementation roadmaps specifying plans for research, development, demonstration, and commercialisation are available in abundance (e.g., USDoE, 2002a, 2004; Industry Canada, 2003; UKDTI, 2003; European Commission, 2008), but where these explore the early steps towards establishing a hydrogen energy system, scenarios aim at a fairly comprehensive description of the future system once implemented, in order to be able to test its viability and identify problem areas (that may have repercussions on the introductory phase actions). Specific focus on the US transportation sector is found in Andress et al. (2011), notable because of its focus on improving vehicle efficiencies before introducing hydrogen, but leaving the sources to be used for producing hydrogen quite open (fossil, nuclear, or renewable).

Suggestions for using fossil fuels with carbon capture to produce hydrogen, as in the scenario discussed following, remains an option intensely discussed (Cormos, 2011). A similar Canadian study using fossil and biomass resources including waste, but without carbon capture, is discussed by Gnanapragasam et al. (2010).

Scenarios for global energy demand and supply by the year 2050 are discussed following. For ease of comparison, they are based on the same assumptions regarding energy demand: concern for human welfare and environmental sustainability are assumed to lead to highly efficient energy consumption patterns based on increased concern for materials conservation and emphasis on non-material ("information society") types of activity.

The scenario year 2050 is selected sufficiently far into the future to allow the necessary changes in infrastructure and technology to be introduced in a smooth manner, without premature discarding of existing equipment before the end of its economic life. The use of a geographical information system (GIS) makes it possible to express all quantities on an "energy flow per unit of land area" basis, a methodology which offers new perspectives on energy supply and use, as compared with the conventional country-based statistics. This method is used for demand and for renewable energy supply, where the decentralised production makes an area-based assessment interesting. For fossil and nuclear energy with centralised production facilities, the areabased method would lead to point-like production maps, and here I use the conventional country averages.

Scenarios for a particular section of society, such as the energy sector focussed upon here, require a general vision of the main characteristics of the global 2050 society, but one that need only be a broad-brush description of some general conditions. A detailed specification on the general shape of all aspects of society is clearly a formidable task, which is here approximated by the two-resolution approach. By necessity, the energy system is based on assumed developments for presently known technologies, but only as a means of proving the feasibility of the scenario: if better options emerge with time, as they almost certainly will, then they will replace some of the scenario technologies. However, the choice made at least constitutes a default and as such proves that there exists a possible system of the kind envisaged.

The actual development may comprise a combination of some reference scenarios selected for analysis, with each reference scenario being a clear and perhaps extreme example of pursuing a concrete line of political and technological preference. It is important that the scenarios selected for political consideration be based on values and preferences judged as important in the society in question. The value basis should be made explicit in the scenario construction. Although all analysis of long-term policy alternatives is effectively scenario analysis, particular studies differ in the comprehensiveness of the treatment of future society. For example, most of the current analysis makes normative scenario assumptions only for the aspects of society directly having a direct influence on the energy sector.

The economic viability of technology choices made for the scenarios can be assured only within considerable uncertainties, because the final cost of emerging technologies is based on necessarily uncertain and sometimes too optimistic estimates. However, for the end-use energy demand scenario, the conservative technology assumption made below is that the average 2050 technology is identical to the best technology available today, in terms of energy efficiency. It may in connection with assuming that such technologies are also economically viable be necessary to refer to inclusion of externality costs (i.e., indirect costs not currently included in market prices, as discussed in connection with life-cycle analysis in Chapter 6).

A global population scenario is important for determining energy demand. I use a detailed United Nations population study estimating a year 2050 world population of 9.4 × 10⁹ in its middle variant. Details of the population model are discussed in Sørensen (2010a), and the assumed population distribution is shown in Table 5.2, also offering migration patterns (mostly movements into cities) and a country list used as a basis for the figures given in the next subsections.

COUNTRY	CONTINENT	REGION	POPULATION ('000) 1996	POPULATION ('000) 2050	Urban % 1992	Urban % 2050
Afghanistan	Asia	III	20883	61373	18.92	55.97
Albania	Europe	II	3401	4747	36.34	72.65
Algeria	Africa	VI	28784	58991	53.34	89.65
Andorra	Europe	II
Angola	Africa	VI	11185	38897	29.86	75.80
Anguilla	North America	IV	8	13	0.00	0.00
Antarctica	Antarctica	-
Antigua & Barbuda	North America	IV	66	99	35.56	60.94
Argentina	South America	IV	35219	54522	87.14	90.00
Armenia	Europe	II	3638	4376	67.98	89.11
Aruba (Netherlands)	North America	IV	71	109	0.00	0.00
Australia	Australia	II	18057	25286	84.94	90.00
Austria	Europe	II	8106	7430	55.44	77.52
Azerbaijan	Europe	II	7594	10881	54.96	83.15
Azores (Portuguese)	Europe	II
Bahamas	North America	I	284	435	84.76	85.18
Bahrain	Asia	III	570	940	88.62	90.00
Bangladesh	Asia	III	120073	218188	16.74	57.62
Barbados	North America	IV	261	306	45.84	53.15
Belarus	Europe	III	10348	8726	68.56	90.00
Belgium	Europe	II	10159	9763	96.70	90.00
Belize	North America	IV	219	480	47.28	69.64
Benin	Africa	VI	5563	18095	29.92	68.73
Bermuda	North America	IV
Bhutan	Asia	III	1812	5184	6.00	28.85
Bolivia	South America	IV	7593	16966	57.80	90.00
Bosnia & Herzegovina	Europe	II	3628	3789	49.00	84.15
Botswana	Africa	VI	1484	3320	25.10	77.65
Brazil	South America	IV	161087	243259	72.04	90.00
British Virgin Islands	North America	I	19	37	0.00	0.00
Brunei Darussalam	Asia	III	300	512	57.74	79.29
Bulgaria	Europe	II	8468	6690	68.90	90.00

Burkina Faso	Africa	VI	10780	35419	21.62	90.00
Burundi	Africa	VI	6221	16937	6.78	31.77
Cambodia	Asia	V	10273	21394	18.84	63.06
Cameroon	Africa	VI	13560	41951	42.14	85.83
Canada	North America	I	29680	36352	76.64	89.58
Cape Verde	Africa	VI	396	864	48.24	68.91
Cayman Islands	North America	IV	32	67	100.00	100.00
Central African Republic	Africa	VI	3344	8215	39.00	74.15
Chad	Africa	VI	6515	18004	20.86	52.74
Chile	South America	IV	14421	22215	83.54	90.00
China	Asia	V	1232083	1516664	27.84	75.58
Colombia	South America	IV	36444	62284	71.08	90.00
Comoros	Africa	VI	632	1876	28.96	48.36
Congo	Africa	VI	2668	8729	55.62	90.00
Cook Islands	Oceania	IV	19	29	0.00	0.00
Costa Rica	North America	IV	3500	6902	48.14	84.80
Croatia	Europe	III	4501	3991	61.64	90.00
Cuba	North America	IV	11018	11284	74.56	90.00
Cyprus	Europe	III	756	1029	52.48	65.70
Czech Republic	Europe	III	10251	8572	65.10	84.26
Denmark	Europe	II	5237	5234	84.96	90.00
Djibouti	Africa	VI	617	1506	81.54	90.00
Dominica	North America	IV	71	97	0.00	0.00
Dominican Republic	North America	IV	7961	13141	62.08	90.00
Ecuador	South America	IV	11699	21190	56.24	90.00
Egypt	Africa	VI	63271	115480	44.26	75.44
El Salvador	North America	IV	5796	11364	44.38	75.35
Equatorial Guinea	Africa	VI	410	1144	38.30	90.00
Eritrea	Africa	VI	3280	8808	17.00	50.39
Estonia	Europe	III	1471	1084	72.32	90.00
Ethiopia	Africa	VI	58243	212732	12.74	43.08
Falkland Islands	South America	IV
Fiji	Oceania	IV	797	1393	39.86	75.26
Finland	Europe	II	5126	5172	62.12	86.52
Macedonia (former Yugosl.)	Europe	III	2174	2646	58.64	85.64
France	Europe	II	58333	58370	72.74	89.02
French Guiana	South America	IV	153	353	75.36	83.52

French Polynesia	Oceania	IV	223	403	56.40	80.30
Gabon	Africa	VI	1106	2952	47.42	87.11
Gambia	Africa	VI	1141	2604	23.76	68.12
Georgia	Asia	III	5442	6028	57.00	86.88
Germany	Europe	II	81922	69542	85.78	90.00
Ghana	Africa	VI	17832	51205	34.92	75.48
Gibraltar	Europe	II	28	28	100.00	100.00
Greece	Europe	II	10490	9013	63.64	90.00
Greenland	Europe	II	58	72	78.90	86.11
Grenada	North America	IV	92	134	0.00	0.00
Guadeloupe	North America	IV	431	634	98.86	90.00
Guam	Oceania	IV	153	250	38.08	59.03
Guatemala	North America	IV	10928	29353	40.24	78.48
Guinea	Africa	VI	7518	22914	27.32	72.45
Guinea Bissau	Africa	VI	1091	2674	20.82	63.32
Guyana	South America	IV	838	1239	34.64	77.45
Haiti	North America	IV	7259	17524	29.80	72.33

Honduras	North America	IV	5816	13920	41.98	80.68
Hungary	Europe	III	10049	7715	63.14	90.00
Iceland	Europe	II	271	363	91.00	90.00
India	Asia	V	944580	1532674	26.02	59.38
Indonesia	Asia	IV	200453	318264	32.52	82.58
Iran	Asia	V	69975	170269	57.38	88.35
Iraq	Asia	V	20607	56129	72.92	90.00
Ireland	Europe	II	3554	3809	57.14	81.50
Israel	Asia	III	5664	9144	90.42	90.00
Italy	Europe	II	57226	42092	66.66	83.08
Ivory Coast	Africa	VI	14015	31706	41.68	80.91
Jamaica	North America	IV	2491	3886	52.38	83.35
Japan	Asia	II	125351	109546	77.36	90.00
Jordan	Asia	III	5581	16671	69.40	90.00
Kazakhstan	Asia	III	16820	22260	58.44	87.55
Kenya	Africa	VI	27799	66054	25.24	70.52

Kiribati	Oceania	IV	80	165	35.04	61.33
Korea Dem. People's Rep.	Asia	V	22466	32873	60.40	86.06
Korea	Asia	V	45314	52146	76.80	90.00
Kuwait	Asia	III	1687	3406	96.34	90.00
Kyrgyzstan	Asia	III	4469	7182	38.48	71.03
Laos	Asia	V	5035	13889	22.00	62.42
Latvia	Europe	III	2504	1891	71.84	90.00
Lebanon	Asia	III	3084	5189	85.16	90.00
Lesotho	Africa	VI	2078	5643	20.88	66.79
Liberia	Africa	VI	2245	9955	43.26	81.47
Libya Arab Jamahiriy	Africa	VI	5593	19109	83.84	90.00
Liechtenstein	Europe	II
Lithuania	Europe	III	3728	3297	70.12	90.00
Luxembourg	Europe	II	412	461	87.42	90.00
Madagascar	Africa	VI	15353	50807	25.12	68.85
Malawi	Africa	VI	9845	29825	12.48	46.79
Malaysia	Asia	IV	20581	38089	51.36	89.39
Maldives	Asia	V
Mali	Africa	VI	11134	36817	25.08	68.88
Malta	Europe	II	369	442	88.28	90.00
Marshall Islands	Oceania	IV
Martinique	North America	IV	384	518	91.62	90.00
Mauritania	Africa	VI	2333	6077	49.60	90.00
Mauritius	Africa	VI	1129	1654	40.54	71.23
Mexico	North America	IV	92718	154120	73.68	90.00
Micronesia	Oceania	IV	126	342	27.04	49.82
Moldova	Europe	III	4444	5138	49.36	87.39
Monaco	Europe	II
Mongolia	Asia	V	2515	4986	59.16	88.76
Morocco	Africa	VI	27021	47276	47.02	80.38
Mozambique	Africa	VI	17796	51774	29.76	84.67
Myanmar	Asia	V	45922	80896	25.36	63.39
Namibia	Africa	VI	1575	4167	34.10	86.65
Nauru	Oceania	IV	11	25	0.00	0.00
Nepal	Asia	V	22021	53621	12.02	50.65
Netherlands	Europe	II	15575	14956	88.82	90.00
New Caledonia	Oceania	IV	184	295	60.78	76.98
New Zealand	Australia	II	3602	5271	85.32	90.00

Nicaragua	North America	IV	4238	9922	61.04	90.00
Niger	Africa	VI	9465	34576	15.92	51.21
Nigeria	Africa	VI	115020	338510	36.84	81.06
Niue	Oceania	IV	2	2	0.00	0.00
Northern Mariana Islands	Oceania	IV	49	92	0.00	0.00
Norway	Europe	II	4348	4694	72.66	89.08
Oman	Asia	III	2302	10930	11.88	49.00
Pakistan	Asia	V	139973	357353	33.08	75.12
Palau Islands	Oceania	IV	17	35	0.00	0.00
Panama	North America	IV	2677	4365	52.34	83.38
Papua New Guinea	Asia	V	4400	9637	15.40	44.58
Paraguay	South America	IV	4957	12565	50.42	88.35
Peru	South America	IV	23944	42292	70.76	90.00
Philippines	Asia	IV	69282	130511	50.96	90.00
Poland	Europe	III	38601	39725	63.38	89.08
Portugal	Europe	II	9808	8701	34.34	70.65
Puerto Rico	North America	IV	3736	5119	72.14	77.17
Qatar	Asia	III	558	861	90.50	90.00
Reunion	Africa	VI	664	1033	65.46	82.23
Romania	Europe	III	22655	19009	54.14	83.77
Russian Federation	Asia	III	148126	114318	74.80	90.00
Rwanda	Africa	VI	5397	16937	5.80	21.97
Saint Lucia	North America	IV	144	235	46.84	52.39
San Marino	Europe	II
Sao Tome & Principe	Africa	VI	135	294	0.00	0.00
Saudi Arabia	Asia	III	18836	59812	78.46	90.00
Senegal	Africa	VI	8532	23442	40.80	78.06
Seychelles	Africa	VI	74	106	51.68	75.09
Sierra Leone	Africa	VI	4297	11368	33.80	78.09
Singapore	Asia	IV	3384	4190	100.00	100.00
Slovakia	Europe	II	5347	5260	57.42	86.56
Slovenia	Europe	III	1924	1471	60.80	90.00
Solomon Islands	Oceania	IV	391	1192	15.60	40.91
Somalia	Africa	VI	9822	36408	24.80	62.06
South Africa	Africa	VI	42393	91466	49.84	83.52
Spain	Europe	II	39674	31755	75.84	90.00
Sri Lanka	Asia	V	18100	26995	21.80	59.06
St. Kitts & Nevis	North America	IV	41	56	40.72	57.03
St. Vincent & Grenadine	North America	IV	113	174	0.00	0.00
Sudan	Africa	VI	27291	59947	23.34	63.17
Suriname	South America	IV	432	711	48.66	86.17

Swaziland	Africa	VI	881	2228	28.32	78.73
Sweden	Europe	II	8819	9574	83.10	90.00
Switzerland	Europe	II	7224	6935	60.02	84.59
Syrian Arab Rep.	Asia	III	14574	34463	51.08	84.33
Taiwan	Asia	IV	2087	2583.706	0.00	0.00
Tajikistan	Asia	III	5935	12366	32.20	63.48
Tanzania	Africa	VI	30799	88963	22.24	67.52
Thailand	Asia	IV	58703	72969	19.22	53.98
Togo	Africa	VI	4201	12655	29.42	69.11
Tonga	Oceania	IV	98	128	37.50	59.47
Trinidad & Tobago	South America	IV	1297	1899	70.18	90.00
Tunisia	Africa	VI	9156	15907	55.82	87.77
Turkey	Europe	III	61797	97911	64.06	90.00

Turkmenistan	Asia	III	4155	7916	44.90	73.20
Turks And Caicos Islands	North America	IV	15	32	0.00	0.00
Tuvalu	Oceania	IV
US Virgin Islands	North America	I	106	158	0.00	0.00
Uganda	Africa	VI	20256	66305	11.72	42.09
Ukraine	Europe	III	51608	40802	68.62	90.00
United Arab Emirates	Asia	III	2260	3668	82.20	90.00
United Kingdom	Europe	II	58144	58733	89.26	90.00
United States	North America	I	269444	347543	75.60	90.00
Uruguay	South America	IV	3204	4027	89.46	90.00
Uzbekistan	Asia	III	23209	45094	40.88	72.73
Vanuatu	Oceania	IV	174	456	18.82	38.47
Vatican City (Holy See)	Europe	II
Venezuela	South America	IV	22311	42152	91.36	90.00
Vietnam	Asia	V	75181	129763	20.26	53.20
Western Sahara	Africa	VI	256	558	41.00	56.82
Western Samoa	Oceania	IV	166	319	57.86	79.20
Yemen	Asia	III	15678	61129	30.78	78.62
Serbia & Montenegro	Europe	III	10294	10979	54.46	88.80
Zaire	Africa	VI	46812	164635	28.50	66.29
Zambia	Africa	VI	8275	21965	42.04	73.61
Zimbabwe	Africa	VI	11439	24904	32.00	72.42

Table 5.2. Regional assignments of countries used in scenarios, with population distributions. The selection of countries/territories and their names is based on the definitions and borders used in the GIS software employed (MAPINFO, 1997); population in thousands are from UN (1996) and urbanisation percentages from UN (1997). More recent UN population projections are modified only after year 2050.

With respect to economic activity and energy demand, the scenario expects the development until the mid-21st century to be dominated on one hand by efforts to "catch up" by many currently poor countries, depending on several factors including education policy, issues of global equity in trade conditions, and also issues of local conflicts and government corruption. On the other hand, a determinant will be the nature of the "new activities" being added to or replacing the current inventory of energy demanding undertakings. New, information-related activities often have smaller energy requirements than the activities they displace, leading to a de-coupling of economic growth and energy demand: industrialised country energy demands have in recent years risen much less than measures of economic activity such as gross national products. The expectation is that this trend will continue, and due to technical requirements, the energy intensity of, for example, computer-related activities will continue to decrease, while the number of installations will increase, as will the activity level in general.

As an example, the overall activity growth in Western European countries was a factor of 5.6 during the 60-year period 1930-1990, from 2200 to 12 370 euro/cap. average GNP. The growth was uneven (depression, World War II, reconstruction period, unprecedented growth 1956-1971, stabilising from 1973), but over the entire 60 years perhaps symptomatic of the technology progress achieved during this quite exceptional period in world history. The likely European growth over the next 60 years will be lower, and high growth rates will be seen predominantly in certain Asian regions. The IPCC Second Assessment (IPCC, 1996) estimated in its high-growth scenario that the growth in Western Europe will reach a GNP of 45 300 1990-US\$/cap. by 2050, as contrasted to 69 500 1990-US\$/cap., if the growth factor equals that seen in the period 1930-1990. It would perhaps be a more realistic approach to estimate growth during the next 60 years as being at most the same in absolute terms as between 1930 and 1990.

A specific European study assumed the emergence of an information society with two thirds of the growth de-coupled from energy and use of materials (Nielsen and Sørensen, 1998), implying in simplistic terms that the growth factor for demand of energy services should be one-third of the GNP growth. The relationship between energy and GNP is complex and depends both on attitudes and on technology developments. The ratio between energy and GNP growth 1930-1990 first declined from 1.5 to 1.0, then rose to 2.0 during the exceptional period and at times became negative after 1973 (Sørensen, 2011b). This is partly an effect of energy (and particularly oil) prices, but also technology requirements have played a role by improving energy efficiency after 1975 in ways often exceeding a purely cost-driven transition. The change in delivered energy service may have been even lower (e.g., the improvement in service between bicycle transport and automobile transport is not always as big as the increase in energy use would suggest).

Based on factors such as the ones just outlined, the energy demand by year 2050 has been estimated for different regions in the world, using the following methodology (Sørensen, 1996b, 2010a; Kuemmel et al., 1997): current energy supplied to the end-user (taken from statistics, such as OECD, 1996) is analysed with regard to the efficiency of the final conversion step, translating energy into a useful product or service. The net energy demand is taken as that used by the most efficient equipment available at present (when the scenarios were first published in 1996, they included some technology available only at the prototype stage, but subsequently, most of that technology has reached the commercial markets). The assumption that, as briefly mentioned in the beginning of this section applies to all scenarios discussed, is thus that by 2050 the average of any equipment used will have an energy efficiency equal to the best found in the current marketplace. This would seem a fairly cautious assumption, as it neglects the possibility of new methods and equipment, based on different principles and capable of improving efficiency by much more than marginal factors.

The overall consequence of these assumptions will in the demand scenario employed be a global increase to year 2050 in end-use energy amounting to an average factor of 4.8, with much larger increases in the currently poor countries. The implied average 1994-2050 per capita growth factor for enduse energy is 2.7, and the consequences in terms of primary energy input will be different for the different supply scenarios, due to different conversion efficiencies of the intermediate conversions. The de-coupling implies that the GNP per capita growth to year 2050 will be substantially larger than the factor 2.7.

The basic methodology for estimating energy demand at the end-user is taken from Sørensen (2010a, Ch. 6). Energy demand futures are sometimes discussed in terms of changes relative to historic and current patterns (such as those shown in Fig. 5.10, together with the distribution on sources of energy). This is of course a suitable basis for assessing marginal changes, while for changes over a time horizon of 50 years, it is not likely to capture all important issues. The alternative approach used is to look at human needs, desires, and goals and from this build up first the material demands required for satisfying these and then the energy required under certain technology assumptions. This is a bottom-up approach based on the view that certain human needs are basic needs−that is, non-negotiable−while others are secondary needs that depend on cultural factors and stages of development and knowledge and could turn out differently for different societies, subgroups or individuals within a society. The basic needs include those of adequate food, shelter, security and human relations, and there is a continuous transition to more negotiable needs that incorporate material possessions, art, culture, and human interactions and leisure. Energy demand is associated with satisfying several of these needs, with manufacturing/constructing the equipment and products entering into the fulfilment of the needs and with procuring the materials needed along the chain of activities and products.

In normative models aimed at environmental sustainability, the natural approach to energy demand is to translate needs and goal satisfaction to energy requirements consistent already at demand level with environmental sustainability. For market-driven scenarios, basic needs and human goals play an equally important role, but secondary goals are more likely to be influenced by commercial interest rather than by personal motives. It is interesting that the basic needs approach is always taken in discussions of the development of societies with low economic activity, but rarely in discussions of highly industrialised countries.

The methodology thus first identifies needs and demands, commonly denoted human goals, and then discusses the energy required to satisfy them in a chain of steps backwards from the goal-satisfying activity or product to any required manufacture and then further back to materials and primary energy (this method is sometimes denoted "backcasting"). The evaluation is done on a per capita basis (averaging over differences within a population), but taking into account different geographical and social settings.

Figure 5.10. Historic trends in the world's gross energy use and its distribution on sources, 1860-2000. During the early years, food energy includes an estimate of the draft animal power utilised. Wood includes straw and biofuels ethanol and biogas (with use of Jensen and Sørensen, 1984, USDoE, 2003, and Sørensen, 2010a).

The assumption that it is meaningful to specify the energy expenditure at the end-use level without caring about the system responsible for delivering the energy is only approximately valid. In reality there may be couplings between the supply system and the final energy use, and the end-use energy demand therefore in some cases becomes dependent on the overall system choice. For example, a society rich in resources may take it upon itself to produce large quantities of resource-intensive products for export, while a society with less resources may instead focus on knowledge-based production, with both doing this in the interest of balancing an economy to provide satisfaction of the goals of their populations, but possibly with quite different implications for energy demand. The end-use energy demands will be distributed on the following energy qualities:

1. Cooling and refrigeration 0-50°C below ambient temperature.
2. Space heating and hot water 0-50°C above ambient temperature.
3. Process heat below 100°C.
4. Process heat in the range 100-500°C.
5. Process heat above 500°C.
6. Stationary mechanical energy.
7. Electrical energy (no simple substitution possible).
8. Energy for transportation (mobile mechanical energy).
9. Food energy.

The goal categories used to describe the basic and derived needs are:

A: Biologically acceptable surroundings
B: Food and water
C: Security
D: Health
E: Relations and leisure
F: Activities
- f1: Agriculture
- f2: Construction
- f3: Manufacturing industry
- f4: Raw materials, and energy industry
- f5: Trade, service and distribution
- f6: Education
- f7: Commuting

Here, categories A-E refer to direct goal satisfaction, f1-f4 refer to primary derived requirements for fulfilling the needs, and finally, f5-f7 refer to indirect requirements for carrying out the various manipulations stipulated. The individual energy requirement estimates are discussed in detail in Sørensen (2010a). A summary of the scenario for demand at the end-user is given in Fig. 5.11.

5.3.2 Global clean fossil scenario

The interest in a future scenario based on fossil energy resources is to be able to maintain as much as possible of the current energy infrastructure, which is almost entirely based on the fossil option (Fig. 5.10). The reason for having to depart from the present practice is partly air pollution, which is becoming unacceptable, particularly for motorcar traffic in cities, and increasingly greenhouse gas emissions, the adverse effect of which on climate is becoming increasingly well established. Air pollution from power plants has been diminished by a number of technical devices, but for the small engines of motorcars with breathing height emissions, no effective means have been found to reduce the health impacts sufficiently (the reduction achieved by catalytic converters placed in the exhaust system is not nearly enough to avoid negative impacts). The greenhouse issue is seen as even more severe, placing demands of at least 60% reduction in carbon dioxide emissions for just stabilising the atmospheric content (IPCC, 1996; Sørensen, 2011a).

Figure 5.11. Total required energy delivery to final consumers (including environmental heat and food energy), for the 2050 scenarios and averaged over each country (unit: Wm-2; note that scale is logarithmic). The global total delivered energy flow is 6.93 TW or 0.74 W/cap. From Sørensen, 1999; the same data are shown without country averaging in Fig. 5.37.

There is also concern over the finite magnitude of fossil resources, which some decades ago, when energy demand increased exponentially, was predicted to cause supply problems within the next 50 years (to be further discussed in Chapter 7). The problem is aggravated by the move away from the most abundant fossil resource, coal, due to its higher CO2 emissions per unit of energy, and because it seemed to have more air pollution impacts than, for example, natural gas (although this depends very much on the technology used). However, the current scenarios for the mid-21st century assume that energy efficiency measures are pursued, notably on the demand side, which will stretch the availability of fossil resources considerably. How much will be discussed following. One may also remark that, particularly for natural gas, the current resource estimates may be on the low side. All-together, it will be shown that with the scenario demand assumptions, the fossil resources could cover supply for at least a hundred years, provided that techniques are introduced that allow coal to be used outside the power sector, and that CO2 is actually captured and safely disposed of (all with acceptance of the induced energy cost increases).

A number of technical options for allowing fossil fuels to be used without emission of CO2 to the atmosphere are currently being explored. Research topics include primary conversion of fossil fuels before conversion or use, notably to hydrogen, for which combustion or fuel cell uses do not emit carbon dioxide, although some of the hydrogen conversion steps may. Another idea is to recover carbon dioxide from the flue gases of fossil fuel combustion and to sequester already emitted CO2, for example, by biological processes. The fossil scenario explores these technologies and makes a choice of the mix of options to use. It is clear that both carbon dioxide recovery and hydrogen use constitute a significant change in the way human societies will use fossil fuels. The sheer mass of carbon dioxide to be handled is orders of magnitude above that of the pollutants SO2 and NOx currently being processed. This also means that the carbon dioxide recovered before or after use of the fossil fuels will not be easy to dispose of. Options for safely depositing CO2 are therefore an integral part of the technology discussion. All of the technologies considered will add to the cost of using fossil fuels. However, the extra cost should be compared with the estimated externality cost of the negative impacts of the present emission of pollutants and CO2, and preliminary estimates suggest that the expenditure can in this way be defended.

This clean fossil energy system coming out of these considerations is not so similar to the present system as one may have hoped. Road-based vehicles will have to use zero-emission technologies, pointing to hydrogen and/or electric vehicles. Only for air and ship transportation may oil products possibly retain a role. Also for the power sector, conventional power plants may not emerge as the best solution, due to the limitations on removing CO2 from flue gases (decreased power plant efficiency, incomplete removal), and therefore fuel cell conversion technologies are seen as providing a better solution also for stationary purposes. Some of the hydrogen produced by conversion of natural gas and coal can be used directly for process heat in industry. This all points to hydrogen as playing a major role in the clean fossil scenario. It also plays a significant role in the nuclear and the renewable energy scenarios constructed in the following sections, but not in quite the dominating fashion found to optimise the fossil scenario. One could therefore with good reason use the term hydrogen scenario as an alternative name for the clean fossil scenario.

Clean fossil technologies

Removal of CO2 after conventional combustion may be achieved by absorbing CO2 from the flue gas stream (using reversible absorption into, e.g., ethanol amines), by membrane techniques, or by cryogenic processes leading to the formation of solid CO2. These techniques have the disadvantage of requiring substantial energy inputs, and the most accessible techniques (absorption) further only lead to a partial capture of CO2 (Meisen and Shuai, 1997). However, with heat recycling there is hope to achieve about 90% recovery, which would be quite acceptable for greenhouse gas mitigation, and the energy requirements may be reduced to around 10% (of the power generated) for natural gas-fired units and 17% for coal-fired ones (Mimura et al., 1997). In the scenario, it is further assumed than an average power plant conversion efficiency of 40% can be achieved for a modern combined power and heat-producing plant with removal of CO2 from the flue gas.

An alternative "after combustion" type of CO2 removal is to convert atmospheric CO2 to methanol by a catalytic process at elevated temperature and pressure. Catalyst based on Cu and ZnO are used and laboratory demonstrations performed at a temperature of 150°C and a pressure of 5 MPa (Saito et al., 1997). Additional reaction products are CO and water. Other options include carbon sequestering by enhanced biomass growth, where increasing forest areas can provide a long time interval between carbon assimilation and subsequent decay and release (Schlamadinger and Marland, 1996). Such options have not been incorporated into the present scenario.

The most promising option for avoiding CO2 is to transform the fossil fuels to hydrogen and then use this fuel for subsequent conversions. Currently, hydrogen is produced from natural gas by steam reforming with water vapour, with a conversion efficiency around 70%, as discussed in Chapter 2, section 2.1. If the initial fossil fuel is coal, a gasification process with partial oxidation followed by the shift reaction would be used. Nitrogen from the air is used to blow oxygen through the gasifier, and impurities in the crude gas would be removed. With impurities removed, the hydrogen fuel is now of pipeline quality, ready to be transported to the points of use. Overall conversion efficiencies of about 60% are expected.

Because the quantities of CO2 to be disposed of after extraction are huge, proposed storage in aquifers or abandoned wells may well be insufficient (available capacity is less than 100 Gt of coal equivalent; Haugen and Eide, 1996). This leaves ocean disposal of CO2 as the only serious option. Storage here would be by dissolving liquefied CO2 in sea water at depths of 1000- 4000 m through special pipelines from land or from ships or by converting the CO2 to dry ice form and simply dropping it from a ship into the ocean (Koide et al., 1997; Fujioka et al., 1997). The CO2 is supposed to be subsequently dissolved into seawater, and if sites are suitably selected, it may stay in cavities or at the ocean floor indefinitely, due to its higher density.

The cost includes that of liquefaction or dry ice formation, plus operational costs and pipelines if used. Fujioka et al. (1997) estimate these costs to be about US\$0.03 per kWh of fuel (US\$0.08/kWh of electricity if that is what is produced) for the liquefied pipeline and ocean tanker disposal scheme, and US\$0.05 per kWh of fuel for the dry ice scheme.

The CO2-rich waters will stimulate biological growth and may seriously alter marine habitats (Herzog et al., 1996; Takeuchi et al., 1997). Stability of the deposits, and the subsequent fate of any escaped CO2 will also have to be established−for example, by empirical methods.

The clean fossil scenario selected includes both hydrogen produced from natural gas and from coal, with the efficiencies stated previously. For use in fuel cells, the hydrogen to electricity conversion efficiency in 2050 is taken as 65%. Losses in hydrogen storage and transmission are taken as 10%, as compared with 5% for electricity transmission.

Fossil resource considerations

Fossil resources are biomass that has undergone transformations over periods of millions of years. Their use as fuels is already limited to a fairly short interval in history and should be seen as a unique opportunity to smooth the road to a more sustainable energy system.

The discussion of fossil resources and their geographical distribution will be made on the basis of a simple version of the standard distinction between reserves and other resources. Three categories will be used:

Proven reserves are deposits identified and considered economic to exploit with current price levels.
Additional reserves are deposits that exist and may become economic, with a probability over 50%.
New and unconventional resources are all other types of deposits, typically inferred from geological modelling or identified but not presently being considered technically or economically feasible to exploit.

The sum of all known and inferred (with reasonable probability) resources without consideration of economy of extraction is the resource base. The level of investigation is uneven among regions, and therefore additional amounts may be discovered, particularly in areas not well studied today. Also, extraction methods vary with time, and new techniques (e.g., enhanced oil recovery) may alter the amount of reserves assigned to a given physical resource.

*Figure 5.12a*−c. Proven reserves of oil (and natural gas liquids) in Wym-2 (top), offshore part (middle), and additional reserves (bottom) (World Energy Council, 1995).

Figure 5.13. Total resources of oil and natural gas liquids in place (unit Wym-2; in this and adjacent figures representing the average number of years for which an energy flow of 1 W per m2 of land surface could be derived at 100% energy extraction efficiency, for each country). The resources have been distributed over the country land areas only, although many of the reserves and resources actually occur offshore; see Fig. 5.12 (based upon data from World Energy Council, 1995; with area-based layout from Sørensen, 1999).

Figure 5.12a gives the proven reserves of oil (including natural gas liquids), Fig. 5.12b gives the part of them located offshore in the territorial waters of the countries indicated, and Fig. 5.12c estimates additional reserves. Figure 5.13 shows the total amounts in place (resource base). Figure 5.14a gives the proven resources of oil shale and natural bitumen, Fig. 5.14b gives the estimated additional reserves, and Fig. 5.14c gives the total known amounts (resources) in place, for these possibly exploitable resources. The rate of new oil finds are currently lower than depletion by production, which has led to predictions of peak production occurring within the next few decades. It is by no means certain that usage will follow the underlying Hubbard bellshaped curve, and production levels are influenced by many factors other than those related to resource depletion. Political choices could lead to high production levels continuing until a final rapid decline (see Chapter 7). The scenario is based on the consideration that new hydrogen schemes should not be based on the politically unstable outlooks for oil.

Figures 5.15a and 5.16a show the country distribution of proven reserves of bituminous coal ("hard coal") and other coal (sub-bituminous coal or lignite), and Figs. 5.15b and 5.16b give the additional reserves for the same two categories. Finally, Fig. 5.16c gives the total amount of coal estimated in place (i.e., the resource base).

Figure 5.17a shows the proven reserves of natural gas, Fig. 5.17b shows the additional reserves, and Fig. 5.17c shows the total amounts in place.

*Figure 5.14a*−c. Proven (top) and additional reserves (middle), and total (bottom) resources of oil shale and natural bitumen (unit Wym-2; World Energy Council, 1995).

*Figure 5.15a*−b. Proven (upper) and additional (lower) reserves of bituminous coal (unit Wym-2; World Energy Council, 1995).

It is clear that estimates of reserves and resources are uncertain, especially where little exploration has been made. Assessments found in the literature therefore differ among themselves, and the reporting of national agencies to international compilations may be of varying quality, for reasons of incomplete knowledge or for commercial reasons. Figures 5.12-5.17 are based on fairly old estimates collected by the World Energy Council (WEC, 1995). They agree fairly well with other estimates for coal and oil, including newer ones, but are lower for natural gas than found in the assessment made for IPCC by Nakicenovic et al. (1996). For example, Middle East oil-producing countries quote very small associated gas reserves, which may just indicate that they do not find it economic to bring these resources to suitable markets. From geological considerations, the view is sometimes expressed that there is more natural gas to be found, and that the 1995 assessment of the WEC is almost certainly an underestimate (of some 50%).

*Figure 5.16a*−c. Proven (upper), additional (lower) reserves of sub-bituminous coal and lignite, and total coal resources in place (bottom; Wym-2; WEC, 1995).

*Figure 5.17a*−c. Proven (top) and additional (middle) reserves of natural gas, and total gas resources in place (bottom) (Wym-2; World Energy Council, 1995).

The average rate of using coal during the 1990s was 2886 GW, for oil it was 4059 GW, and for natural gas 2188 GW. This means that the proven reserves shown in Figs. 5.12a, 5.15a, and 5.17a at a constant rate of use equal to the one just referred to would last 167 years for coal, 46 years for oil, and 70 years for natural gas. Relative to the total estimated resource in place, the numbers would be 3554 years for coal, 210 years for oil plus another 184 years from oil shale and bitumen, and 96 years for natural gas. The total natural gas resources as mentioned may well be underestimated. In Chapter 7, an update is made and a closer look taken at the depletion of oil.

The fossil scenario

In constructing a clean fossil scenario for satisfying the demand scenario described in section 5.3.1, the basic consideration has been to use of the technologies just described in such a way that the resource utilisation becomes acceptable. Because oil resources are considered the most limiting constraint, and also because oil is in widespread use outside the energy sector as lubricants and feedstock for a number of industrial products including first of all plastics, it is decided not to use oil in the energy sector, except possibly for a few sectors where substitution with other fossil technologies may be difficult, notably as an aviation fuel. Non-energy uses of oil could alternatively be substituted with, for example, biomass-based raw materials, but the way the current set of scenarios is constructed, this would rather belong in the renewable energy scenario. The past trends shown in Fig. 5.10 also suggests that oil usage is levelling out and thus lend support to those who estimate declining oil production starting about a decade from now, even without considerations of political constraints, and thereafter declining production.

In regard to the balance between natural gas and coal, it is for resource reasons considered appropriate to increase the relative use of coal. Both fossil resources can be transformed into hydrogen (as could oil), for providing a carbon-free energy carrier, but the efficiency of transforming coal to hydrogen is a somewhat lower than that of transforming natural gas, and the cost is considerably higher. However, the price of natural gas is currently above that of coal and expected to develop in the direction of an increased difference, in the short term due to use of gas substitution to reduce CO2 emissions, in the longer range for reasons of resource constraints. The increase in the cost of natural gas and certainly that of oil will almost certainly happen before the selected scenario year, 2050.

The scenario choice is then in the transportation sector to cover half the demand by electric vehicles (urban vehicles, trains) and the other half by hydrogen (used in fuel cells directly or with methanol from natural gas or coal as a precursor). The fuel cell vehicles are of course also electric vehicles, but with the power-production taking place on board. Because de-carbonisation favours a large fraction of the fossil fuels being transformed to hydrogen, hydrogen is in the clean fossil scenario that is also used directly in industry for medium- and high-temperature process heat, thereby avoiding a second conversion step to electricity and the associated conversion losses.

By 2050, the scenario assumes a substantial demand for electricity proper− that is, electricity such as for electric apparel that cannot be powered by other sources−plus stationary mechanical energy, which for environmental reasons and efficiency optimisation is better served by electricity as input. The production of electricity in the scenario is partly performed at conventional power plants, but also by fuel cell power plants (presumably SOFCs) that can use hydrogen. Again, this is because it is easier to transform the fossil fuels to hydrogen before conversion than to recover carbon dioxide from stack emissions. In the scenario, about two-thirds of the power generation is from fuel cells. The higher efficiency is partly offset by losses in hydrogen production and storage, but the versatility is higher, because the fuel cells may become decentralised and integrated into individual buildings. This is made possible by a hydrogen distribution pipeline system, which in many countries could serve as a useful continuation of using the existing natural gas distribution network. Transmission (regionally) and distribution (locally) of hydrogen and natural gas will both be needed in this scenario.

For low-temperature heat, the conversion of fuels in both power plants and fuel cells provides considerable amounts of associated heat production, which can be employed after distribution by district heating lines, and in the case of building-integrated fuel cells may serve to provide heat to the same building and its activities. However, because of the considerable direct use of hydrogen and the fairly high electric power conversion efficiencies achieved, in the scenario there is not quite enough "waste" heat to cover all lowtemperature requirements, and the remaining space and process heat (under 20% of the total, varying with the season) are assumed to be provided by heat pumps. The scenario lumps space cooling by electricity-driven coolers and heat from heat pumps together, assuming the same coefficient of performance (ratio of heat output, added or removed, to electricity input), COP = 3.33. This is not the highest technically possible value, but because much of the space heating is in climates where the source of environmental heat to the heat pumps is of low temperature, it is considered reasonable to use a cautious estimate.

Although the scenario aims at only using fossil resources, the existing hydro electricity production is retained. This is a renewable energy resource, but it works well together with the ingredients of the fossil system, and only existing plants and plants currently under construction are assumed to be present in the fossil scenario. The distribution of power production from hydro is shown in Figs. 5.18a,b. The total potential contribution is an average production of 440 GW. A debate has taken place over large hydro installations:

Figure 5.18a,b. Potential delivery of hydropower to final consumers in the 2050 scenario (upper panel) and current annual average production of hydropower at existing hydropower plants plus estimated production from plants under construction (lower panel). (Based upon data from World Energy Council (1995), with area-based layout from Sørensen (1999)).

Hydropower has in the past been used in a way not compatible with environmental considerations (flooding areas of high landscape value, dislocating people, etc.). It is here considered as a renewable energy source worthy of inclusion in any sustainable scenario, provided that it is used with proper considerations. These may include modular construction of large plants (cascading the water), allowing a highly reduced disturbed area, and planned location of reservoirs in locations where there is little conflict with preservation concerns, even if the water transport layout may become more complex and expensive.

The 2050 scenarios have chosen to include all existing hydropower plants and plants under construction. It would hardly be feasible to attempt to close those existing plants that have caused human and environmental suffering. At least some studies suggest that reverting to pre-hydro landscapes will entail negative environmental effects quite similar to the original change in the opposite direction, due to the already completed adaptation of flora and fauna to the changed conditions (Tasmanian Hydro Commission, private communication, 1998).

Also included in the scenarios are small-scale hydro installations planned in several countries, considering the impacts to be acceptable for most schemes. As regards total potentials identified and actual schemes proposed but not started, a modest inclusion is made of only those for which actual proposals including environmental statements exist. As a source for such appraisals of the potential hydro generation a survey made by the World Energy Council (1995) has been used. Including all categories in that study would roughly double the current hydropower generation. Figure 5.18a shows the distribution of this potential power generation on countries. A large fraction of such installations is or will be reservoir based, so no additional storage cycle is considered necessary (although pumped hydro would be an obvious option). Figure 5.18b shows the average generation from existing hydro plants plus those under construction, again averaged over country areas. These are the ones included in the fossil as well as in the other scenarios discussed in the following sections.

It is not attempted to show the actual reservoir areas on the maps of Figs. 5.18a or 5.18b. The figures simply spread the power production evenly over each country. In any case, most hydro is a centralised resource in the sense that it is not (and in most cases could not be) used by the local population. South America is the most endowed region, implying that the utilisation choices made in this region will be decisive for the global contribution of hydropower.

The clean fossil scenario will now be presented in terms of annual average energy flows between supply and demand. The handling of diurnal and seasonal variations is much the same as it is today. The only critical area is in the provision of space heating and cooling, where the co-produced heat from power plants and fuel cells on average is very close to the average demand. For space cooling, the high season just requires more electricity input, which is provided by a sufficient installed capacity of the power plants and fuel cells. For winter heating, the co-producing power and heat plants can already with technology currently used in many places avoid a locked ratio between electricity and heat production, and by regulation of the heat to power ratio within the available range cater to any heat demand situation, without need for adding storage to the system, except possibly for diurnal storage, which is already today economic in some cases.

Figure 5.19a,b. Hydrogen derived from fossil energy, to be used for covering half the needs in the transportation sector (using PEM fuel cells) and all requirements for medium- and high-temperature process heat (using hydrogen furnaces), after transmission and storage losses (top). Electricity derived from fossil energy (by conventional power plants), to be used for covering half the needs in the transportation sector and all requirements for stationary mechanical energy, dedicated electricity and electricity for cooling, refrigeration and heat pumps, after transmission losses (bottom). Average flows in W per m2 of land area for each country (Sørensen, 1999).

Figure 5.19a shows country distributions of the amount of hydrogen needed to provide energy for the demand scenario medium- and high-temperature industrial process heat plus 50% of the transportation needs, making allowance for 10% losses in storage and piping of hydrogen to the end-user. Figure 5.19b gives the electricity from fossil fuels required for stationary mechanical energy, electric appliances, cooling, refrigeration, and the heat pumps covering 20% of space heating needs, plus the other half of the transportation energy needs.

*Figure 5.20a*−c. Coal for conventional combined power and heat plants (top), and coal (middle) and natural gas (bottom) required for hydrogen production. Average flows in W per m2 of land area for each country (Sørensen, 1999).

*Figure 5.21a*−c. Hydrogen required as input to stationary fuel cells (top), space heating/cooling provided by heat pumps and electric coolers (middle), and lowtemperature district heating (waste heat from fuel cell and other power plants; bottom). Average flows in W per m2 of land area for each country (Sørensen, 1999).

The power provided by hydro is given first priority, explaining that there is on average no fossil power requirement in hydro-abundant countries such as Canada, Sweden, or Norway. In practice, the fast regulation of hydro plants allows them to be used in load-levelling efforts. In Fig. 5.21a, the amount of hydrogen that is needed as fuel for electricity-producing stationary fuel cells is shown, assuming a conversion efficiency of 65% (for electricity, the rest being available as low-temperature heat). Figure 5.20a similarly gives the coal input to power plants, at an electric efficiency of 40%, again with the remaining 60% being available for district heating. One-third of the electricity in the scenario is assumed to come from coal-fired power plants and two-thirds are from fuel cells (probably SOFCs) with hydrogen as input. In Figs. 5.20b and 5.20c, the coal and natural gas inputs to produce the required amounts of hydrogen are shown. The scenario assumes 40% to be based on natural gas, with a conversion efficiency of 70%, and 60% to be from coal with a conversion efficiency of 60%. Figure 5.21b shows the energy for space heating (20%) and cooling (100%) covered by heat pumps and coolers, while Fig. 5.21c gives the rest of the low-temperature requirements for space heating and process heat covered by district heating based on power plant and fuel cell waste heat. Figure 5.22 gives an overview of the scenario energy system, with annual energy flows indicated. The coverage of heat with assistance from co-produced heat from power plants and fuel cells is indicated.

With the production of natural gas and coal now having been determined, the year 2050 scenario assumes that the annual requirement is produced by each country in proportion to its total estimated resources of natural gas and coal, respectively. This creates a need for trade of these commodities, much in the same way as today. The fixed percentages of natural gas and coal inputs would in practice be modified for countries with resources primarily of one type. For example, Norway and Saudi Arabia would increase their use of natural gas in order not to have to import coal. For countries able to cover their own electricity requirements with existing hydro−for example, Canada and Norway−the scenario construction does not initially take into account export options, and the hydro input in Fig. 5.22 is therefore less than the potential production (by 177 GW). This additional capacity will in reality be used for export of electric power to neighbouring countries and thus will slightly diminish the use of fossil fuels, on a global level.

Figures 5.23a and b give the surplus and deficit of coal production on a national basis, Figs. 5.23c and 5.24a similarly for natural gas, and Figs. 5.24b and c give the totals. As expected, trade follows the same patterns as at present. Although the fossil reserves are more evenly distributed over the globe than thought 40 years ago, the matching with demand is most negative for countries in South America, Africa, and Southeast Asia, due to the low level of identified resources and in Southeast Asia the high population density, plus the assumed high standard of living in the scenario year 2050.

Figure 5.22. Overview of clean fossil scenario (flows in GW or GWy/y). For comparison, the 1990 totals are indicated at the bot- tom, with consumption total taken from the analysis in section 5.3.1. A large fraction of the stationary fuel cells might be placed in a strongly decentralised fashion, down to individual buildings. Food production has been detached from energy production; see discussion in Sørensen (2010a). (From Sørensen, 1999.)

*Figure 5.23a*−c. 2050-scenario coal surplus (top), deficit (middle), and natural gas surplus (bottom). Surplus/deficit is country production minus required supply, if positive/negative. Average flows in W per m2 of national land area (Sørensen, 1999).

*Figure 5.24a*−c. 2050-scenario natural gas deficit (top), total fossil surplus (middle), and deficit (bottom). Surplus/deficit is country production minus required supply, if positive/negative. Average flows in W per m2 of national land area (Sørensen, 1999).

Evaluation of the clean fossil scenario

Any fossil scenario has to be regarded as a parenthesis in the grand-scale history of mankind (Sørensen, 1979, 2010a). The 2050 scenario just described allows the expected future demand to be covered by a rate of fossil fuel input no greater than the current one (see Fig. 5.22). Even then, the proven reserves of natural gas and coal will sustain the scenario primary energy use only for 73 and 86 years, respectively (the similarity of these numbers was aimed at in selecting the use of gas and coal in the scenario). Taken in relation to the total resource base (whether exploitable or not), the scenario energy use can be sustained for 101 and 1830 years, for gas and coal, respectively, with the natural gas value possibly being underestimated, as just discussed. It has been a fundamental consideration to avoid using oil in the scenario, partly because reserves are believed to be smaller than those of gas and coal, and also because of the geopolitical implications of pawning the energy supply to nations with unstable or fundamentalist regimes.

The carbon removal and deposition technologies described in section 5.3.1 have in most cases not been demonstrated in full scale. There may be technical problems and likely new environmental problems not foreseen, leading at least to higher prices if they can at all be fixed. Particular concern is in place regarding the ocean disposal of carbon dioxide, which could produce environmental problems as large as those one is trying to avoid, if the assumed stability turns out to not be secured. The cost estimates quoted in section 5.3.1 suggest that the ocean disposal will increase the cost of energy derived from fossil fuels by a factor of 2−3. To this one must add the cost of the new technologies for producing hydrogen, whereas the current high cost of fuel cells, whether stationary or mobile, has to be assumed to decline to near the general level of conventional power plants, per unit of energy produced. Altogether, the cost of energy produced in the clean fossil scenario will be larger than the present one by amounts having to be justified by:

The avoided greenhouse warming damage; this may be defended by referring to the higher ones among the externality estimates found in the literature, but not by the lower ones (Kuemmel et al., 1997; IPCC, 1996).
The avoided uncertainty of oil availability, for reasons of political conflicts or of declining extraction due to resource depletion.

Apart from the question mark on the environmental implications of ocean CO2 disposal, the fossil scenario seems to be technically feasible. However, the original justification for continuing use of fossil fuels was to be able to use existing infrastructure. This is not really the outcome of the scenario analysis: the transportation sector will be very different from the current one (key technologies hydrogen, various types of fuel cells, batteries and electric motors), although a development which may well be needed in any case due to urban pollution considerations. District heating, which is presently common only in a few countries, will have to be expanded, and an increase must take place in the use of heat pumps for heating. The industry sector will not feel much change in going from, for example, natural gas to hydrogen, and electricity use will generally expand.

Considering the likelihood of the assumed technical advances of technologies for fossil gasification to hydrogen fuel and subsequent use in fuel cells, despite optimism one must realise that substantial further development is needed in order to achieve technical and economic viability. Particularly for the fuel cells, current costs are beyond what can be swept under the heading of "including externalities" in cost comparisons. For the disposal technologies, and particularly the ocean disposal required for a large-scale handling of carbon dioxide, there is little doubt that the CO2 can be dumped into the ocean (as ice or using suitable pipes) at an affordable cost. However, it needs to be proven by experiment that this simple procedure does in fact constitute a safe form of disposal, leaving the carbon-containing substance away from the biosphere for sufficiently long periods of time. One may take the position that the known reserves of fossil fuels are too small to warrant the difficulties of a major transition from today's fossil technologies to the new "clean" ones, and only the belief that a high proportion of the general resources known to exist can indeed in the future be converted to "reserves" and exploited in practical, economic, and environmentally acceptable operations will make the fossil scenario interesting.

The scenario technologies are selected from a much larger catalogue of options as being the most promising ones. Nobody envisages this cost to be as low as the present cost of energy, and that would also not be a proper starting point for comparison, as the current supply system has to be changed over the period considered, for reasons of the environment if not for reasons of resource depletion. Without subsidy, such as accepting damage or paying for environmental cleanup, the present energy cost would be much higher. The only exception from the expectation of rising energy costs is the impact that energy efficiency measures could have. These are assumed taken into account in the demand scenario and they all make sense even at current energy cost. For the supply and conversion technologies, the best that can be hoped for is that the cost can be kept within reasonable limits in a frame of reference defined by adding externality costs to all energy-related activities.

One might alternatively suggest to turn the problem around and ask that a "standard price" or "target price" be determined for each new technology−that is, the price that would allow this technology to penetrate the market to the extent assumed in the particular scenario (Nielsen and Sørensen, 1998). However, this assumes knowledge of the future society that cannot be taken for granted. Changes in values and paradigms will alter the valuation of externalities and likely add other externalities not identified today. Therefore, it may well turn out to be impossible to calculate such standard prices to the accuracy that would render them useful for the discussion. In other words, what can be done is only to point to the plausibility of the selected technologies as having the potential for becoming economic in the eyes of a future society and to remind us of the fact that all decisions on changes so profound as the ones considered here for the energy system have in the past been made on the basis of normative convictions−in other words, by visions of determined individuals along with their persistence in pursuing specific goals. No radical progress was ever based upon economic evaluations, and most historic innovations had to fight claims of defying economic rationality. The rationality of the past is a concept of little use in shaping the future.

Implementation of the situation to be reached by 2050 according to the scenario involves sketching a path of moving from the current system primarily based on fossil fuels to the very different system and identifying the conditions that have to be fulfilled to make the transition happen. These would involve economic milestones for the new technologies involved, as well as political decisions needing to be taken. It is here assumed that the social environment, in which the transitions happen, is governed by a mixture of free market competition and regulation by society, as it is the case in most regions of the world today. The regulatory part would impose requirements, such as building codes and minimum technical standards for reasons of safety and consumer protection, and would set maximum energy use for certain appliances and activities. A possible public handle is to use environmental taxes that incorporate indirect costs which otherwise would distort the competition between different solutions in the marketplace. A consistent method of estimating the environmental taxes that will make the market "fair" is to perform life-cycle analysis of the entire energy supply chains and the technologies involved. The methodology for doing this, with examples for many of the renewable energy systems considered here, is described in Sørensen (2010a; 2011a) and a condensed overview is given in Chapter 6. However, also life-cycle analysis has many uncertainties, pointing back to the normative element in any energy policy decision.

5.4 Scenarios based on nuclear energy

5.4.1 History and present concerns

Electricity generated at nuclear power stations presently accounts for some 8.4 EJ y-1 or 2% of global energy use (USDoE, 2003). The technology used is primarily light water reactors, a commercial spin-off from the submarine nuclear-powered propulsion systems introduced in the 1950s. The situation after World War II was characterised by two factors of some importance for the development of nuclear energy and the specific reactor choice:

There was the existence of a highly skilled group of nuclear physicists with a desire to be able to employ their knowledge for peaceful purposes, as opposed to the wartime bomb making. In other words they wanted to demonstrate that nuclear techniques could be used for beneficial purposes as well as for destruction. No group of researchers and developers held in similar esteem were available for the development of competing energy technologies such as those based on renewable energy sources.
There was general shortage of economic means, relative to the magnitude of tasks needing to be performed in rebuilding the assets destroyed during the war. This led decision-makers to prefer the cheapest solution, without regard to supply security and environmental impacts, that later became part of energy policy decisions. It meant that although alternative nuclear reactor designs were indeed suggested and developed (e.g., CANDU and high-temperature types), the economic advantage of copying an already existing military technology was very tempting, even if some critics found the design less than suited for civilian deployment.

The subsequent fate of nuclear power in the energy sector may be seen as a consequence of the special circumstances suggested by these observations. A major public dispute erupted in a number of countries with tradition for public participation, basically centred on three issues:

♦ The nuclear reactor technology selected poses risks of large accidents with major releases of radioactivity and unpredictable consequences.
♦ Operation of nuclear power plants creates volumes of radioactive waste, to be kept separate from the biosphere during intervals of time that cannot be guaranteed as they are much longer than the lifetime of economic entities or even national states.
♦ The nuclear fuel chain generates plutonium and other material useful for nuclear weapons production and thus increases the risk of such material being diverted to belligerent states or to terrorist organisations for use in actions of nuclear blackmail or war.

The early proponents of nuclear power claimed that any risk associated with the technology was zero or negligible and that the engineering and operational safety was far superior to that of any other technology. Lengthy reports were procured, which seemed to prove these statements as being virtually mathematical tautologies.

This provoked a number of independent scientists to look closer at the claims, and they concluded that there were basic flaws in both the methodologies used and the actual calculations (see overview by Sørensen, 1979b). It is therefore not surprising that when major accidents started to happen the nuclear proponents lost credibility in a way that could perhaps have been avoided with a more honest approach to the information issue.

Important accidents involving commercial plants have been the Three Mile Island reactor partial meltdown accident, which did not breach the outer containment but totally ruined the plant, the Chernobyl meltdown accident, which caused major releases of radioactivity to the atmosphere and global fallout, and the Fukushima accident with four ruined reactors plus fallout.

Gradually, national energy policies in most democratic countries were changed to exclude additional construction of nuclear power plants, either directly or indirectly by enforcing strong licensing conditions that could only be met with increased time intervals between the decision to build and the final commissioning, and consequently with considerable increases in costs. Concurrent efforts to make the electricity industry more competitive caused power plant operators to forego the "difficult" nuclear option, thereby avoiding to deal with its uncertainties.

A clear indication of this development is found in the market projections of future world uranium requirements, as, for example, stated by the US Department of Energy (deMouy, 1998). The global 2015 requirements are estimated as only 85% of the current ones, as shown in Fig. 5.25, despite some growth in countries in Southeast Asia and South America.

Figure 5.25. World uranium requirements according to US Department of Energy reference case projection (based on deMouy, 1998).

The purpose of the "safe nuclear scenario" is to investigate if nuclear technologies different from those employed today may solve the problems or strongly reduce the concerns stated above, while forming a solution that is viable in terms of the transition time needed and resource depletion and has reasonable economy. As it turns out, it is hardly possible to address all three of the voiced concerns, and most improvements of nuclear technology are aimed at solving one or at most two of the key problems, while often at the same time improving other performance aspects.

5.4.2 Safe nuclear technologies

The technologies selected for the safe nuclear scenario should address the main objections to current nuclear power technologies: proliferation issues, large nuclear accidents and long-term storage of waste. A wealth of ideas and new technologies for avoiding or reducing these problems has been discussed for some years, but all are still fairly speculative. A few have been tested at laboratory scale, but their implementation will take further technical development and would presumably make nuclear power more expensive than today. These additional costs have to be justified by including social costs (including those external to current economic valuation) of the problems associated with current nuclear technologies.

Among the reactor types studied over the past 4 or 5 decades, but not reaching the stage of market introduction on a commercial basis, are hightemperature gas-cooled reactors and sodium-cooled fast breeders. Current proposals are aware of the issues raised (p. 295), but still far from deal with all of them. The reactor industry has recently concluded that a new generation of safer reactors will require substantial breakthroughs (particularly in materials science) that may push commercialisation at least 25 years into the future (USDoE, 2002b). The proposed concepts are summarised in Table 5.3.

Technology	Coolant	Pressure	Temp. (K)	Issues
Conventional breeders	sodium	low	820	safety, cost, reprocessing
Supercritical water	water	very high	800	safety, materials, corrosion
Very high temperature	helium	high	1300	safety, materials, accidents
Gas-cooled breeders	helium	high	1130	materials, fuels, recycling
Lead-cooled breeders	lead-bismuth	low		800−1100 materials, fuels, recycling
Molten salt	fluoride salts	low	1000	materials, salts, reprocessing

Table 5.3. Industry proposals for new generations of nuclear reactors (USDoE, 2002b; Butler, 2004).

The four proposed reactor types operating at temperatures over 1000 K can be used to produce hydrogen directly. All the concepts operate at temperatures higher than the ≈600 K of existing light water reactors and therefore would produce electric power at a higher efficiency. The likely cheapest of the systems is the extreme pressure water-cooled reactor, but it does not solve problems of large accidents and large amounts of nuclear waste. The sodium-cooled breeder has already been carried to the large-scale demonstration stage, but has had troubled operational experiences and as a concept does not solve problems of safety and cost. Like the three other types in Table 5.3 requiring reprocessing of spent fuel, it has severe weapons proliferation dangers. Helium-cooled reactors have also been researched for many years, with the most recent prototype just going critical in Japan. The very high-temperature proposals do pose materials problems expected to require many years of research and development efforts. While the Japanese prototype uses fuel pellets in a honeycomb graphite structure, future versions are expected to be of the pebble-bed type, consisting of millions of 5-10 cm diameter spheres of fuel coated with graphite (acting as moderator) and a very hard ceramics layer, to capture and encapsulate fission products. In this way, it is hoped that an accident will not lead to release of large amounts of radioactivity to the environment. However, this depends on temperature control in the event of an accident, a problem still needing to be resolved. No proposal for safeguarding plutonium created during reprocessing has yet been found credible.

Among the chemical cycles available for hydrogen production at temperatures above 1000 K, as an alternative to electricity production by a thermodynamical Brayton cycle (Sørensen, 2010a), the following set currently appears most attractive (Summers et al., 2004; Elder and Allen, 2009),

at 830°C: $$H_2SO_4 + heat \rightarrow \frac{1}{2}O_2 + SO_2 + H_2O$$ ,
at 120°C: $I_2 + SO_2 + 2H_2O \rightarrow H_2SO_4 + 2HI$ , (5.1)
at 320°C: $2HI \rightarrow I_2 + H_2$ .

The net reaction is water splitting and all the chemicals may in principle be recycled between the reaction steps, while recuperated heat from the first reaction is used for the following lower temperature steps. Reactors for hydrogen production should have module sizes of at least 500 MW, for convenience in sizing Brayton turbines, but is far from small enough to constitute an inherently safe design as defined following.

An alternative method for hydrogen production by a nuclear reactor operating above 100 K is a high-temperature fuel cell unit in reverse mode (Fujiwara et al., 2008).

Inherently safe designs

This concept means that the risk of core meltdown in case the heat from fission processes cannot be led away must be absent. Two examples of proposed inherently safe reactor designs are:

♦ Either to reduce the size so much that core melt accidents almost certainly can be contained by the vessel used (this involves maximum unit sizes of 50−100 MW in a traditional design, while the pebble-bed reactor may circumvent this limitation, if the integrity of the pebbles can be guaranteed).
♦ Or to use a design in which the core of a conventional pressurised water reactor (PWR) is enclosed within a vessel of boronated water that will flood the core if pressure is lost: there is no barrier between the core and the pool of water, which in case pressure in the primary system is lost will shut the reactor down and continue to remove heat from the core by natural circulation. It is calculated that in an accident situation, replenishing of cooling fluid can be done at weekly intervals (in contrast to hours or less required for current light water reactor designs) (Hannerz, 1983; Klueh, 1986).

To avoid proliferation, fissile material such as plutonium should never accumulate in large amounts or should be difficult to separate from the stream of spent fuel. This can be addressed by using accelerators to produce fissile material at the same pace as it is used to produce energy. Accelerators are also among the options for "incineration" of current nuclear and military waste, in order to reduce waste storage time and again avoid storage of waste from which weapons material could be extracted.

There are two proposed technologies that use accelerators. One is called accelerator-breeding, as it aims to convert fertile material (e.g., Th-232) into fissile fuel to be used elsewhere in a (conventional) reactor (Lecocq and Furukawa, 1994). This does not in itself reduce accident probabilities, and to achieve this, the reactors should be of the mentioned inherently safe type.

The other accelerator concept proposed during the last decade (Rubbia, 1994; Rubbia and Rubio, 1996) integrates the accelerator and the reactor-like device into what is termed an energy amplifier. The central point in this design is that the energy amplifier does not have to be critical (i.e., the nuclear processes do not have to sustain themselves), as there is a continuous supply of additional neutrons from spallation processes caused by the accelerated protons. This is supposed to greatly reduce the risk of criticality accidents. Exactly how much needs to be further studied. The main components of the fuel chain are illustrated in Fig. 5.26. Although this concept is not supported by the existing reactor industry, it will be used here as a template for the scenario presented below, because of its unique promise to reduce all three objections to the current variety of nuclear reactors. For this reason follows a brief outline of the technology proposed.

Technical details of energy amplifier

The proposed energy amplifier technology involves a proton accelerator (linear or cyclotron) for the purpose of producing fast neutrons. This is achieved by letting high-energy protons impinge on some heavy target (which could be lead, uranium, or thorium). The process is called spallation and typically produces around 50 neutrons of energy 20 MeV per 1 GeV proton. The device proposed by Rubbia resembles a reactor or the blanket of a reactor. Fertile material such as thorium-232 (23290Th) needs bombardment with energetic neutrons to transform into nuclear isotopes capable of fissioning−that is, splitting into two approximately equally heavy products under release of the energy difference (see, e.g., Sørensen, 2010a). Examples of fertile materials are ²³² 90Th and ²³⁸ 92U, whereas examples of isotopes that may undergo fission already by absorption of slow (little energetic) neutrons are 233 92U, 235 92U, and 239 94Pu. In current light water moderated reactors, the fissile isotope is 23592U, and the additional neutrons released by fissioning are capable of sustaining a chain reaction (or in absence of careful control a run-away reaction, which is the criticality problem leading to accidents such as the one that occurred in 1986 in the Chernobyl reactor; see Kurchatov Institute, 1997). In the dense core of fast breeder reactors, the number of ²³⁸ 92U isotopes hit by neutrons and transformed into 239 94Pu is so large that the potential energy derived from the fissile material may be 60 times that of the 23592U input.

Figure 5.26. The energy amplifier concept. The fraction of the generated electricity used for powering the accelerator is indicated. The reprocessing step is essential for the resource viability of the system (From C. Rubbia and J. Rubio (1996). A tentative programme towards a full-scale energy amplifier. CERN. Used with permission).

Figure 5.27. Layout of the central heat-generating unit of an "amplifier", a 30 m high silo to be placed underground. Note the emergency beam dump volume indicated as "EBDV", (From Rubbia et al. (1995). Conceptual design of a fast neutron operated high power energy amplifier. European Organisation for Nuclear Research, CERN. Used with permission.)

The accelerator-breeding efficiency depends strongly on the neutron multiplication factor k, which in the proposed design should be about 0.95 (as opposed to very slightly above 1 in a conventional reactor). This is equivalent to an energy gain (breeding factor) of about 50, and it is believed that the value of k = 0.95 is sufficient to avoid k ever exceeding the value 1, in situations of irregularity. If the accelerator is halted, the nuclear process will stop. The heat-generating unit will be placed underground, with the proposed design illustrated in Fig. 5.27.

Sub-criticality is the first essential feature of the concept; the second is the use of a thorium cycle, based on the accelerator-induced reactions (T1/2 is the half-life, i.e., the time after which activity has been reduced to half),

$$\begin{array}{l} ^{232}{90} Th + n \ \to \ ^{233}{90} Th \ (T{1/2} = 23m) \ \to \ ^{233}{91} Pa \ (T_{1/2} = 27d) + e^- \$$

Figure 5.28. Activity of radioactive waste from nuclear power plants after 40 years of continuous operation, indicating the advantage of accelerator-breeder concepts. The calculations are preliminary and do not correspond to expected final designs (based on Rubbia et al., 1995; Lung, 1997).

The fissionable end product 23392U is not the only outcome, as some of the 23390Th will undergo an (n,2n) reaction to ²³¹ 90Th that will further decay to ²³¹ 91Pa and 23292U, leading eventually to 20881Tl, the strong gamma-activity of which makes reprocessing difficult, although not impossible. Because of the breeding property, reprocessing of fuel elements is expected to take place only with intervals of five years or more. The thorium cycle as just described has significant advantages in terms of reducing the amounts of nuclear waste to be kept separated from the biosphere for long periods of time. Figure 5.28 gives the calculated radio-toxicity of waste from the thorium energy amplifier (one-pass through or recycling reprocessed waste) and compares it with current reactor waste toxicity (Lung, 1997; Magill et al., 1995). The accelerator-breeder will accept current reactor and military waste as input, thus offering a way of disposing obsolete nuclear weapons and high-level waste from light water reactors.

Concern has been expressed that the interface between the accelerator and the breeder could constitute a weakness in the Rubbia design, allowing escape of radioactivity together with molten lead, in case of sudden accelerator failure (J. Maillard as quoted by Mirenowicz, 1997).

The combination of accelerator-breeding and use of the thorium cycle with possible inclusion of uranium resources is needed in the global "safe nuclear scenario" for resource reasons, as further explained in the section below: the available U-235 reserves used in conventional reactor types would only last for a short time at the rate of usage assumed in a global nuclear scenario, and although additional resources may become available, the time horizon is not going to expand significantly towards a sustainable system. Therefore, breeding is necessary in any long-term nuclear scenario, and inclusion of thorium resources would extend the resources to last some 1000−2000 years in scenarios of efficient energy usage such as the present (IPCC, 1996). Current thorium extraction is discussed by Hedrick (1998).

The ideas of accelerator-breeding and thorium-based nuclear power have been around from the start of the effort to transfer nuclear techniques from the military to the civil sector, first for resource reasons (Lawrence as quoted by RIT, 1997), later for reasons of reducing the waste disposal problem (Steinberg et al., 1977; Grand, 1979; OECD, 1994) and getting rid of military nuclear waste and obsolete warheads (Toevs et al., 1994; National Academy of Sciences, 1994), and finally for avoiding proliferation and reducing accident risks (Rubbia et al., 1995).

The ideas involving separation of the accelerator-breeding stage and the energy production stage (Furukawa, as quoted by Lung, 1997) were used in an early version of the present scenario (Sørensen, 1996b). This is the preferred concept of the Los Alamos Group as well as the French and Japanese groups, and it would use a mixture of lithium, beryllium, and thorium fluorides as input to the accelerator-breeder, allowing this molten-salt fuel to be transported to and used in graphite-moderated nuclear reactors (see Table 5.3), despite negative experiences with such a reactor operated for some years at Fort St-Vrain in Colorado (Bowman et al., 1994; Lung, 1997). Early schemes for waste transmutation proposed the use of breeder reactors (Pigford, 1991), which do not fulfil the demand for inherent safety. The following scenario is based on the Rubbia idea, but it could include parallel operation of inherently safe reactors, from which spent fuel would be shipped to the reprocessing plant of the accelerator-amplifier installation. It should be stated clearly that this scenario is based on largely unproven technology and would require huge research and development investments over significant periods of time. Of course, the same can be said for the current light water technology and certainly for all the advanced concepts contemplated over the past 40 years, including those of Table 5.3.

Nuclear resources assessment

Current estimates of the available reserves and further resources of uranium and thorium, and their global distribution, are shown in Figs. 5.29−5.31. The uranium proven reserves indicated in Fig. 5.29a can be extracted at costs below US\$130/t, as can the probable additional reserves indicated in Fig. 5.29b. Figure 5.29c shows new and unconventional resources that may later become reserves. They are inferred on the basis of geological modelling or other indirect information (OECD and IAEA, 1993; World Energy Council, 1995). The thorium resource estimates are from the US Geological Survey (Hedrick, 1998) and are similarly divided into reserves (Fig. 5.30a), additional reserves (Fig. 5.30b), and more speculative resources (Fig. 5.30c). The thorium situation is less well explored than that of uranium: the reserves cannot be said to be "economical", as they are presently mined for other purposes (rare earth metals), and thorium is only a by-product with currently very limited areas of use. The "speculative" Th-resources may well have a similar status as some of the additional U-reserves.

The magnitude of nuclear sources identified is similar to that of oil or natural gas, if the nuclear fuels are used in conventional non-breeder reactors (capable of lasting about 100 years). Thus, if nuclear fission energy is to constitute a reasonably sustainable resource, some type of breeding is required. The breeding ratio for liquid metal fast breeders (which in the scenarios are excluded because the technology of liquid sodium cooling close to a very dense nuclear fuel core makes the occurrence of accidents with conventional explosions followed by fuel meltdown and possible criticality leading to nuclear explosions very high) may theoretically reach about 60, whereas for the accelerator-breeder concept just outlined, a breeding factor of 10 is assumed to be technically feasible. This brings the thorium resources up to about 1000 years (the similar magnitude of uranium resources may not be of immediate interest if thorium is selected as the principal fuel). On the other hand, once economic interest in thorium is established, the exploration of these resources will take a more serious shape, and it is possible that additional reserves will be identified. The overall resource base is quite large.

The different levels of reserves and resources are summed up in Figs. 5.31a and 5.31b, giving a probable magnitude of total exploitable resources. The one for thorium will be used as a basis for the following nuclear scenario.

*Figure 5.29a*−c. Proven (top; OECD and IAEA, 1993), additional reserves (middle), and new/unconventional (bottom; World Energy Council, 1995) uranium resources, given as kg of uranium oxide per m2 (country averages). GIS layout Sørensen (1999).

*Figure 5.30a*−c. Proven (top) and additional thorium reserves (middle), and new/unconventional thorium resources (bottom), given as kg of thorium oxide per m² (country averages). (Hedrick, 1998; GIS layout Sørensen, 1999).

Figure 5.31a,b. All estimated uranium resources (top; sum of Figs. 5.29*a*−c) and all estimated thorium resources (bottom; sum of Figs. 5.30*a*−c), given as kt of uranium/thorium oxide per m2 (country averages; from Sørensen, 1999).

The highest grade of thorium resources is that found in veins of thorite (ThSiO4), containing 20 to 60% of equivalent ThO2. However, the most common mineral considered for exploitation is monazite (MPO4, where M is Ce, La, Y, or Th, often in combination), containing about 10% pure ThO2 equivalent. A number of other minerals have been identified, which contain Th at a lower percentage (Chung, 1997).

Safe nuclear scenario construction

The safe nuclear scenario employs the common energy demand scenario described in section 5.4.1 and the following conversion technologies: the main source of energy is the energy amplifier described above, assumed primarily to use thorium as fuel (although some U-233 may be used during the first start-up, i.e., for the initial reactor cores). The scenario construction task is then to determine the thorium fuel annual input necessary for covering the demand. The conversion routes included are partly production of hydrogen for use in the transportation sector and possibly for industrial process heat and partly electricity for direct use or for powering heat pumps and cooling devices to provide low-temperature heating and cooling, with additional inputs of environmental heat. Whereas this is the preferred way of producing heat in renewable energy scenarios (because no large sources of waste heat exist), for scenarios such as the fossil and nuclear, the associated heat production from power plants should have first priority, and the local heat production by heat pumps should only be used when distances between power plants and load points make it inconvenient or too expensive to transport district heat to the end-user. In the nuclear case, "waste heat" is abundantly present, so only the transmission conditions limit their use.

The demand for food energy is covered by agriculture and the distribution of production as described in detail in Sørensen (2010a). Another renewable energy source that has been incorporated in the nuclear scenario is hydropower, where existing plants and plants under construction are retained. This is due to their long productive life and ability to operate in harmony with the nuclear power plants, both being components of a centralised system with similar requirements for transmission. The fast regulation of incorporating hydro plants in a given system also adds to the technical viability of nuclear power plants. The scenario presented below assumes that the nuclear plants can be regulated to a sufficient extent to allow following the load with minor backup from hydro and stored energy. However, should it turn out to be technically or economically inconvenient to perform this regulation, the solution is to increase the hydrogen production and use hydrogen as a storage medium, for subsequent use either directly for process heat or by regenerating electricity, presumably using fuel cell technology. This in indicated in Fig. 5.36, where lines of storage cycle supply are shown, but at the moment with zero flow.

The demand for transportation energy is assumed covered by equal amounts delivered to electric vehicles and to fuel cell-based vehicles. For the former, a 50% storage cycle loss associated with the entire battery cycle operation is assumed, and for the fuel cell-based vehicles, operation is assumed based upon either hydrogen or a more storable derivative (e.g., methanol). Fuel cells are considered to have 50% conversion efficiency. In both cases, the minimal losses that occur in the final electric motor conversion to traction power are considered as included in the 50% overall loss. The nuclear electricity going through a hydrogen conversion process is assumed subject to a 20% conversion loss, which would be typical of state-of-the-art electrolysers. For conventional electrolysis processes in use today, the loss is more like 35% (see Chapter 2, section 2.1.3). Figures 5.32 and 5.33 show the detailed assumptions on covering demands in the 2050 safe nuclear scenario.

*Figure 5.32a*−c. 2050-scenario nuclear hydrogen production (top), delivered electricity from nuclear or hydro (middle), and total electricity production (including that used for hydrogen) from nuclear (W per m2 country average; Sørensen, 1999).

Figure 5.33a, b. 2050-scenario nuclear waste heat piped as district heat (top) and environmental heat captured by heat pumps (bottom). (W per m2 country average; Sørensen, 1999).

Figure 5.32a shows the amounts of hydrogen, which in this scenario will be produced from nuclear power. Out of this, 50% of global transportation energy needs will be covered, through use of fuel cell equipped cars or various hybrid vehicles. The remaining energy delivered as electricity comprises all energy used for electric apparel, stationary mechanical energy and input to cooling devices, refrigerators, and heat pumps, which are assumed to provide 50% of the low-temperature heat (space heating, hot water, and other process heat under 100°C). The other half of the low-temperature heat demand is covered by district heating based on nuclear waste heat. Due to the modest efficiency of nuclear reactors, including accelerator-based ones, the available associated nuclear heat is much larger (see Fig. 5.36), but the amount included is considered the maximum amount that can be delivered to load areas at reasonable transmission distances.

*Figure 5.34a*−c. Thorium requirements for 2050 nuclear scenario: Country and time averaged kt of thorium oxide per year and per m2 (top) or per country (middle, note different scale), and potential country Th-production (bottom) (Sørensen, 1999).

Figure 5.35a, b. 2050 safe nuclear scenario: Thorium surplus (top) or deficit (bottom) (positive or negative country values of production minus required supply; kt of thorium oxide per year for each country) (Sørensen, 1999).

The required electricity delivery for all these purposes (except that used to generate hydrogen, i.e., power delivered to the final consumer, covering dedicated electricity use including electric input to heat pumps and for refrigeration, stationary mechanical energy, medium- and high-temperature process heat) is shown in Fig. 5.32b and it will in the 2050 scenario be delivered by hydro or nuclear power, in that order of priority. Figure 5.32c shows the total electric energy input required from nuclear energy−that is, that for hydrogen production (Fig. 5.32a) plus the part of Fig. 5.32b that cannot be supplied by hydropower. The amounts shown in Fig. 5.32c have been augmented by transmission losses and thus give the power that must leave the nuclear power plants, after in-plant uses for the accelerator-based energy amplifiers. Note that countries with large hydropower production (such as Canada) need no or little nuclear energy, as hydro is given first priority.

Figure 5.36. Overview of a safe nuclear scenario (flows in GW or GWy/y, the thorium input is indicated at the left). 1990 con- sumption is indicated at the bottom (based on Sørensen. 1999).

Figure 5.33a gives the amounts of nuclear waste heat to be delivered through district heating lines (the heat leaving the power plants being some 25% higher due to transmission losses), and Fig. 5.33b shows the amounts of environmental heat drawn by the heat pumps used in locations unsuited for district heating (this may come from air, soil, or waterways, and the assumed heat pump coefficient of performance (COP)−that is, the ratio between heat output and electricity input, is taken as 3.33). The two components shown in Figs. 5.33a and 5.33b roughly cover 50% of the low-temperature energy demand each.

Having now determined the total amount of nuclear electricity required, the thorium fuel input to the energy amplifiers can be calculated from the design data of Rubbia and Rubio (1996). The thermal output from the prototype design reactor is 1500 MW, with a fuel amount of 27.6 t in the reactor (Fig. 5.27). The fuel will sit in the reactor heat-generating unit for 5 years, after which the "spent" fuel will be reprocessed to allow for manufacture of a new fuel load with only 2.9 t of fresh thorium oxide supply. This means that 2.6/5 t y⁻¹ of thorium fuel is required for delivery of 5 × 1500 MWy of thermal power over 5 years, or 675 MWy of electric power, of which the 75 MWy is used for powering the accelerator and other in-plant loads. The bottom line is that 1 kg of thorium fuel produces very close to 1 MWy of electric power, and 1 kt thorium produces close to 1 TWhe.

Figure 5.34a gives the scenario requirements of thorium input per year and per m2 , averaged over each country, and Fig. 5.34b gives the total amounts per year for each country. It is seen that countries with abundant hydropower (such as Canada) do not need any nuclear energy in the scenario, even if they do today. The reason is the substantial increase in energy efficiency assumed, a fairly stable population in Canada, and maintaining all hydro plants currently operating or under construction. The hydro contribution is the same in all the scenarios constructed here (fossil, nuclear, and renewable). The thorium requirement per country may finally be compared with the assumed production in each country, shown in Fig. 5.34c. It is determined by assuming that the world production of thorium matches the world requirements, and that each country produces in proportion to its total thorium resources, as estimated in Fig. 5.31b.

It is now possible to determine the required trade of nuclear fuels by comparing the production and use of thorium fuels. This is done in Figs. 5.35a and 5.35b, showing the surpluses and deficits in thorium balances for each country. It is seen that the world is fairly sharply divided into energyproducing countries and energy-importing countries, the latter being in South America (except Brazil), Africa, Europe (minus Sweden, Norway, and Russia), and Asia (minus the former Soviet Union countries). In some of these areas (e.g., in Africa), poor knowledge of Th resources may play a role.

A summary of the scenario is shown in Fig. 5.36. The more than four times increase in energy services at the end-use level, relative to the present level, is here achieved by primary energy inputs 50% above the current ones. Although an improvement over the current systemic losses, this is more "overhead" than in the renewable energy scenarios discussed below in section 5.5, due mainly to the associated thermal energy from the energy amplifiers, that cannot all be made useful.

Evaluation of the safe nuclear scenario

The nuclear scenario presented above has zero emissions of greenhouse gases. It largely avoids the accumulation of nuclear material with risk of diversion for military or terrorist purposes, with the only sensitive material being the spent fuel. It is suggested to join the power production and reprocessing units at the same location (Fig. 5.26), where the reprocessing is fairly delicate due to the presence of strongly radioactive Tl-208 (see the technical description of the energy accelerator on p. 299 and following). This makes it necessary to perform the reprocessing by robots in a heavily gammashielded environment, a technology just starting to be technically feasible and currently excessively expensive (Lung, 1997; Chung, 1997). On the other hand, this fact makes theft of the spent fuel more unlikely.

As regards nuclear accidents, the sub-criticality suggests dramatic reduction in the risk (Buono and Rubbia, 1996), but it is clear that more work is required to fully assure this proposition.

With respect to radioactive waste, Fig. 5.28 suggests large advantages over current nuclear reactor technology in reducing the period in which wastes must be kept separate from the biosphere from some 10 000 years to 500 years. However, this is still a very long time to require integrity of depository sites, in fact of the same order of magnitude as predicted for nuclear fusion wastes (speculatively, as the decisive factor is the choice of materials for the fusion reactor, the design of which—should controlled commercial fusion ever become feasible—is unknown). In the first 100-year period of time, the energy amplifier waste produces a surge of radioactivity coming from the U-233 formation, which requires special precautions in storage and disposal. Figure 5.28 indicates that this "early" surge of radioactivity is still a factor of 3 to 10 lower than the level for light-water reactor waste. It is in any case difficult to see a decisive advantage of the energy accelerator over current reactors with waste containing plutonium-uranium mixtures. A more accurate statement will have to wait for a detailed comparison of the handling techniques for each isotope in the two different waste compositions.

When Rubbia and Rubio (1996) proposed the energy accelerator they noted that a unique opportunity to rapidly evaluate the energy amplifier concept on a realistic scale was offered by the planned year-2000 retirement of the LEP superconducting cavity electron accelerator previously in use for research at the European Organisation for Nuclear Research (CERN). This accelerator could have been modified to accelerate proton beams at a current of 20 mA to energies above 1 GeV, which would make it suitable for an energy amplifier of up to 1500 MW thermal power. The proposal for such a facility, submitted to the European Commission, was evaluated by a working group dominated by current nuclear industry representatives, assigned by the European Commission's DGXII. They recommended (Pooley, 1997), and the Scientific and Technical Committee of the European Commission adopted the view, that the project should not be realised, because it would put nuclear power "back to square one", discarding all existing technology and building a novel one (which was indeed the intention), with associated expenditures as huge as those leading to the existing nuclear technology and with little hope that the public would understand the subtle difference between the old and the new nuclear power and better accept the latter.

Unfortunately, this reasoning could equally well be directed at all the other new nuclear concepts in Table 5.3, as well as against nuclear fusion, and seems to convey the fatalistic attitude that the nuclear industry has already lost so much credibility in the general populations that it is not worthwhile to spend money trying to remedy the situation. Not all funding agencies share this negative attitude, but it is probably true that the nuclear industry, most of which is privatised or in the process of being privatised, finds it difficult to imagine spending as much money on the next generation of nuclear reactors as the military institutions and governments of nuclear nations have put into the development of the light water reactor.

What the construction of a global scenario based on the energy amplifier concept, as presented here, has shown is that if the energy amplifier project could indeed be successfully developed, it would enable the use of nuclear energy on a much larger scale, globally, than ever envisaged for the present nuclear technology. The estimated resources of thorium makes it possible to sustain the thorium-cycle nuclear scenario at the level contemplated for some 1000 years, with a additional, similar interval being possible for uraniumbased concepts with a similar breeding ratio. The direct cost will certainly be higher than that of present nuclear reactor technologies, with the reprocessing and the waste management costs considerably higher and still very uncertain. However, waste handling and plant decommissioning costs have systematically been underestimated for the current generations of reactors (e.g., by use of depreciation rates that do not fairly represent long-term social equity), and the final cost of accidents (such as Three Mile Island, Chernobyl, Fukushima) will be very high, once all clean-up and health damage have been paid for (which, of course, it will, eventually, also if not by the responsible actors and their institutions; see Sørensen, 2011a; 2011b).

Clearly, from a resource point of view (Figs. 5.29-5.31), the present lightwater nuclear technology is uninteresting as a long-term solution. Furthermore, it is not useful as a short-term remedy for the current problems of unreliable oil supply, as it would only be replacing coal in power plants. The future of nuclear energy technology thus depends on developing breeder technologies that successfully address the three objections raised in the beginning of section 5.4.1.

The safe nuclear scenario is weaker than the clean fossil and renewable energy scenarios discussed in this chapter, as it depends on a technology development that has hardly even started and is still far away from the commercial markets. The expertise for handling both accelerator and reactor technologies is probably still present today, although nuclear science and engineering is being chosen by fewer and fewer students. Still, a deeper reason for questioning this technology development is a very real concern over whether a practical version of new concepts such as the energy amplifier will indeed get rid of the problems characterising current nuclear technologies. Would there not still be a risk that other new developments will bring the proliferation risk back, just as the centrifuge technology changed enrichment from being in the hands of a very exclusive club of huge and expensive laboratories to becoming accessible to poor countries and potentially to determined terrorist organisations? May there not turn out be novel accident routes that will become revealed only when the technology has reached a more concrete form? Because of the time required to transform a scientist's idea into an industrial product, the safe nuclear scenario (and even more any use of nuclear fusion) is likely to have problems reaching the envisaged high share in energy supply already by the middle or end of the 21st century. Lessons from renewable energy tell us that the transition time required for such a substantial technology shift may well be 25 years or more, from the day where the new technology is technically ready for commercialisation.

5.5 Scenarios based on renewable energy

For the fossil and nuclear scenarios based on hydrogen as an energy carrier just described, the relevant data were available on a country basis and the geographical distributions were thus simple country maps. In the case of renewable energy, the sources are closely connected to the geography of climates, and the use of an area-based description is much more rewarding in revealing the detailed correlation between potential energy production and final energy use, with the latter being already for many forms of energy clearly area-based, as shown in Fig. 5.37, without the country averaging of Fig. 5.11 (both from Sørensen, 1999). Following, the renewable energy counterpart to the fossil and nuclear scenarios is first presented on a global scale.

This work is already discussed in detail in Sørensen (2010a) and will only be summarised here. However, subsequent work has expanded the insights considerably, while at the same time working on a regional or national scale, the latter allowing a spatial resolution of 500 × 500 m2 and a time resolution of 1 h (in contrast to the seasonal time scale of the global scenarios). This makes the model capable of catching both realistic demand variations and also variations in fluxes of solar and wind energy at a particular geographical spot. With these levels of resolution, a realistic appraisal of energy storage requirements in connection with the fluctuations in renewable energy inputs can be made, which is important for the hydrogen aspects of the scenario, because hydrogen is used both as an energy carrier (for mobile and stationary applications) and also as an energy storage medium. The details of such a scenario will be described in section 5.5.2.

Figure 5.37. Total required energy delivery to final consumers (including environmental heat and food energy), for the 2050 scenarios (Wm-2 shown on 56-km grid).

Several investigations have focussed on finding the best ways of transforming renewable energy into hydrogen, either by electrolysis of surplus renewable electricity (Levene et al., 2007; Rodríguez et al., 2010) or thermochemically for biomass waste products (Levin and Chahine, 2010). Hydrogen production from solar thermal sources is discussed by Pregger et al. (2009).

5.5.1 Global renewable energy scenarios

Two global scenarios based on renewable energy have been constructed, in analogy to the clean fossil and safe nuclear ones. The scenarios differ by one (termed the decentralised one) allowing no centralised energy production, such as wind parks or solar megawatt arrays placed in marginal land areas (deserts, etc.), and the other including such options (termed the centralised renewable scenario). The latter turn out to be much more resilient to misjudgements of the future energy demands or errors in defining acceptable resource utilisation.

The outcome of this "centralised renewable energy scenario" is illustrated in Fig. 5.38, showing the surplus and deficit of demand coverage by the mix of renewable resources available to each region (and inviting trade). A scenario summary is shown in Fig. 5.39. Evidently, city conglomerations use more energy than can be supplied by rooftop solar energy and wind energy placed within the urban setting, whereas many rural areas turn out to be able to supply far more energy than used locally, allowing an export to urban areas profitable to both.

Figure 5.38a, b. 2050 renewable energy scenario: Energy surplus (top) or deficit (bottom) (positive or negative values of production minus required supply; food energy not included). The totals of surpluses and of deficits are equal. (Sørensen, 1999).

Implementation of either of the 2050 scenarios, or a combination of them, involves sketching a path of moving from the current system primarily based on fossil fuels to a very different system and identifying the conditions that have to be fulfilled to make the transition happen.

Figure 5.39. Overview of renewable energy scenario with centralised components (from Sørensen, 1999).

These could be economic milestones for the new technologies involved or they could be political decisions needing to be taken. It is here assumed that the social environment, in which the transitions happen, is governed by a mixture of free market competition and regulation by society, as is the case in most regions of the world today. The regulatory part would impose requirements, such as building codes and minimum technical standards for reasons of safety and consumer protection, and maximum energy use for appliances. Another public handle is environmental taxes that incorporate indirect costs which otherwise would distort the competition between different solutions in the marketplace. A consistent method of estimating the environmental taxes that will make the market a fair one is life-cycle analysis of the entire energy supply chains and the technologies involved. The methodology for doing this, with examples for many of the renewable energy systems considered here, may be found in Sørensen (2011a). A summary is outlined in Chapter 6.

In a fair market, the price that new technologies have to measure up against is the price of currently used coal, oil, natural gas, hydro, and nuclear technologies, all augmented with the externalities not included in the present market prices and reflecting, as well as our knowledge allows, the negative impacts that the new scenarios propose to remedy. Renewable technologies such as wind power, the cost of which today is only slightly above current fossil fuel-based systems, will clearly be economic if externalities are included (as these are very small for wind power and for fossil fuels are at least of the same order of magnitude as the current price). Other renewable technologies such as production of vehicle fuels from biomass, which today involves a cost about twice that of fossil fuels, would in the fair market be able to enter by standard competitive forces. For photovoltaic power and the new conversion and storage technologies such as those based upon fuel cells, present costs are still a bit higher than what can be remedied by introducing externalities for comparison with fossil fuels. Therefore, these have to be assumed to pass through a technical development over the transition period to the scenario year, which would bring the price down below the threshold value. Subsidies may be contemplated in order to speed up this process in the initial phase, but already the political readiness to include externalities in prices would constitute a strong motivation for the development of alternative solutions.

For the alternatives of coal gasification and advanced nuclear energy technologies, a fair comparison should similarly include externality costs. The extra cost of coal gasification may in this way become acceptable, whereas the ocean disposal of carbonates requires further environmental assessment. For nuclear energy such as high-temperature generation of hydrogen or the accelerator technologies, there exist preliminary cost estimates exhibiting the same kind of over-optimism that characterised the early development of nuclear power ("too cheap to meter") (Fernandez et al., 1996). The real cost of the safe nuclear fuel cycle cannot be realistically estimated today, but is likely to be at least two to three times the current energy costs.

The assumptions that the future transition will be driven by fair market rules are somewhat at variance with the present realities. On one hand, there are hidden subsidies in many regions (e.g., to fossil and nuclear energy, where society pays for environmental and health impacts and assumes the responsibility for risk-related events), and on the other hand, monopolies and generally differences in size and power of the energy industries involved in different technologies make the actual price setting likely not to follow those prescribed by the life-cycle analysis in a fair market philosophy.

An obvious solution is to regulate the market not by general taxation (as currently done in Europe) but by regulation and legislation, requiring, for example, power providers to use specific technologies. This alleviates the need to accumulate tax money at the state level (avoiding this is by some nations seen as a positive feature), but makes the system rather stiff, as each technical change has to be followed up by possibly altering the legislative regime. The taxation method where tax levels are set to reflect life-cycle impacts is more flexible, and once the level of environmental tax is decided by governments (who thereby cut through scientific uncertainty), the market functions exactly as before, but should give the manufacturers of the new technologies with smaller environmental impacts a good chance to compete, even if they are initially smaller than the established market players. It is important that externalities are set politically, because there will always be a continued debate over the uncertainties of scientific assessment, which itself is not static.

A problem is the possible differences in tax levels that different nations may see as fair. International synchronisation is highly desirable, as in all policy aimed at reducing global threats. Depending on the degree of planning tradition in different societies, the energy transition might also benefit from "setting goals" and continually monitoring if they are fulfilled. If, for example, the market does not respond well enough to the price signals set by the environmental taxes, it is then possible to adjust the size of the imposed externality (within limits which would not violate the scientific basis for it) or introduce specific legislation to remove the obstacles to a free and fair market.

Remaining technical problems include the determination of the optimum share of different energy forms in the total mix and the requirement for energy storage and infrastructure implied. The next section gives an example of how these issues may be approached on a national level. Scenarios focussing on a single form of energy, such as fossil, nuclear, or renewable, may be illustrative in disclosing the nature of problems posed by the different solutions, but in practice, there may of course be an advantage in combining the energy sources in the actual energy policy of a given nation, unless the "purity" of the solution is a demand of the public in a given democracy.

5.5.2 Detailed national renewable energy scenario

Two detailed scenarios for 2050 will be discussed in this section. They pertain to Denmark, a country already having a share of renewable energy in its energy system, and a vague policy plan to become totally based on renewable energy, first formulated more clearly with a 50% goal share of renewables by the year 2025 (Danish DoE, 1998). The scenarios use hydrogen either in a centralised way or in a decentralised, building-integrated mode. The descriptions will start with a demand projection, followed by a resource assessment and finally the supply-demand matching that constitutes the two scenarios.

Danish energy demand in 2050

The energy demand depends on population size and level of activities. The Danish population is expected to decline slightly, but may likely be compensated by immigration. As future integration policy is difficult to predict, the assumption of unchanged total population has been made. This is the same outlook as the one upon which United Nations (1996) population predictions for Denmark is based. The current population data are thus used, with a small correction based on the tendency to decentralise economic activities by moving businesses and production factories away from cities. This feature of highly developed societies is in contrast to that of the developing nations, where migration is rather from rural to urban areas. As a basis for creating area-based projections, the Danish building database is used (Danish National Agency for Enterprise and Construction, 1999), describing current buildings on a geographical basis (a 500 m × 500 m grid was used), distributed on types and use of each building (categories are detached family homes, dense-low dwellings, farm houses, public and privately owned highrise apartments, public and privately owned office and service buildings, institutions, public and privately owned production enterprises, and finally vacation and recreational buildings). This database, which includes information on present energy installations, is used for determining the distribution of energy use, but also for estimating population, for which statistics only exist on a county level. The totals for each county are distributed on the areabased grid by assuming a population distribution given by building occupancy such that the number of people is proportional to the floor area of all residential category buildings, but with recreational buildings counting only 0.33. This is the estimated use of vacation houses for permanent living. In this way, variations in dwelling area per person in different parts of the country are reflected in the model. The outcome is shown in Fig. 5.40, with the underlying dwelling floor area shown in Fig. 5.41 (total summed floor area for multi-storey buildings). The migration between urban and rural areas has been modelled by assuming a 30% growth in population for grid cells that in the year 1996 had less than 50 inhabitants and a corresponding decline in those with more people in 1996, maintaining a total 2050 population of 5.2 million.

Figure 5.40. Danish population distribution in 2050 used in scenarios (Sørensen et al., 2001).

A number of energy demand types are assumed to be directly proportional to the population, so the energy use in a grid unit is proportional to the population density. Table 5.4 gives a summary of 2050 end-use assumptions made for the scenario construction. The space heating requirement is calculated from the floor area by use of a model taking into account hourly time series of outdoor temperatures, a wind-dependent chill-factor, and solar irradiation on various building surfaces (according to their angle to the direction of direct solar radiation and with a model for the absorption of both direct and scattered solar radiation through windows). The gradual improvements of the building shell (higher insulation, control of ventilation) that have taken place over the last 30 years are assumed to continue to lower the average heating consumption per square metre of building floor area at the same rate as during the past 30 years, where the space heating requirement

was reduced by 50%. The same measures make space cooling unnecessary under Danish climatic conditions. Details of this model are described in Chapter 6 of Sørensen (2010a).

Figure 5.41. Geographical distribution of floor space in the Danish residential sector of 2050 (note that this figure is on a logarithmic scale, in contrast to Fig. 5.66) (Sørensen et al., 2001).

The calculation of space heating requirements takes into account occupancy of building space, so that office and production space is not heated outside working hours and dwelling space has zonal heating with time variations in temperature depending on occupancy (the body heat from which is also included), as controlled by computerised space management units already having a certain penetration in the building sector today. Figures 5.42 and 5.43 give the floor space geographical distribution similar to the one shown in Fig. 5.41 for living areas, but now for production (including agriculture) and commercial building areas. The geographical distribution of all these building types has been taken as unchanged from the present, although relocation of businesses and new preferences regarding residential preferences and types of services offered could lead to a number of (difficult to predict) changes in geographical distribution of floor areas.

Figure 5.42. Geographical distribution of floor space area in the Danish production sector 1998, including farm buildings other than the residential ones included in Fig. 5.41 (Sørensen et al., 2001).

Figure 5.43. Geographical distribution of floor space area in the Danish service sector 1998 (Sørensen et al., 2001).

Figure 5.44 gives the total annual requirement of end-use low-temperature heat. Figures 5.45*a*−d give the geographical variations on a seasonal basis, using a coarse grid of 56 × 56 km2 , because this is the resolution of the satellite temperature data used because temperature variations across Denmark are rather small. The actual heat use map of the model has the full 1-hour and 0.5 × 0.5 km2 resolution, and only the calculations of heat loss use the same outdoor temperatures for the larger grid together with the building descriptions

of the smaller grid data. Note that even in July, there is a space-heating requirement in western coastal location. Finally, Fig. 5.46 gives the hourly time dependence of heat requirements for the whole country, distinctly showing the fairly constant hot water use and the variable space heating.

Energy type/quality	Annual average use (W/cap.)
Space heating (climate dependent, cf. Fig. 5.68)	389
Hot water and other low-temperature heat	150
Medium-temperature process heat (100−500°C)	50
High-temperature process heat (over 500°C)	40
Space cooling (climate dependent)	0
Other cooling and refrigeration	35
Stationary mechanical energy	150
Electric appliances and other electric equipment	150
Transportation energy	150
TOTAL	1214

Table 5.4. End-use energy assumed in the Danish 2050 scenario (Sørensen et al., 2001).

Figure 5.44. Geographical distribution of all Danish 2050 low-temperature heat requirements based on Table 5.4 and the heated floor areas shown in Figs. 5.41−43 (Sørensen et al., 2001).

*Figure 5.45a*−d. Geographical distribution of average Danish space heating requirements in January (top left), April (top right), July (bottom left), and October (bottom right) of the year 2050, based on current satellite measurements of temperature (on the gross scale of a 56 km grid) in combination with the scenario's assumed building standards. The variations thus reflect both climate differences across Denmark and differences in heated space per unit cell. (Sørensen et al., 2001).

The end-use energy given in Table 5.4 is the energy flow measured after any final conversion by the end-user. For example, that means the light (radiation) energy, not the electric power supplied to a light-producing device (fluorescent tube, etc.). For electric appliances the 2050 value is twice that for the year 2000, and for transportation it is three times the 2000 value, while as mentioned the space heating energy is lower than the present one. In Fig. 5.76, the time variation of the necessary electricity requirement for Denmark is shown. It is assumed to be the same as the measured one for the year 2000, except that the absolute magnitude will be increased.

Available renewable resources

Denmark is well endowed with renewable energy sources, including an identified wind potential far exceeding power demands, forestry, and particularly agriculture producing primary food for more than five times the Danish population, plus plenty of residues that could be used for energy purposes, and finally solar energy which on average is not small, but due to the high latitude (around 56°N) has a pronounced seasonal anti-correlation to loads (in addition to the diurnal mismatch).

Figure 5.46. Hourly time variation of low-temperature heat (space heating and hot water) in the 2050 scenario with use of current weather data (Sørensen et al., 2001).

Figure 5.47. Normalised hourly variation of electricity use in Denmark in year 2000, used for the 2050 scenario to model time variation of necessary electricity use (i.e., electricity not used to produce process or low-temperature heat, the use of which may have a different time variation) (Sørensen et al., 2001).

Figure 5.48 shows the wind potential for Denmark, on land and inland seas. The off-shore potential has been estimated for a number of reserved locations, shown in Fig. 5.49. These have been selected as suitable for wind turbine parks without disturbing fishery activities, boat routes for passengers and freight, military exercise areas, and so on. The total exploitable wind power is given in Fig. 5.51 for each reserved area. It by far exceeds the envisaged 2050 electricity use. In several of the areas, wind power production has already been initiated, albeit at a total level substantially lower than that of the 2050 scenario. The year 2003 installed capacity is about 3.3 GW.

Figure 5.48. The annual average power in the wind at 70 m over Danish land area, estimated on the basis of geostrophic winds (winds aloft) and the surface roughness (caused by vegetation, buildings, etc.) traced in 24 different directions at each grid point. In the legend, the number of grid cells with power in a given interval is given. A wind turbine is typically able to convert 30-40% of the power in the wind to electricity. For off-shore wind potential, see Fig. 5. 49 (based on Energi- og Miljødata, 1999; Sørensen et al., 2001).

Time variations in wind power are determined by passage of weather front systems. Typical passage of fronts over Denmark has a pattern exhibiting similar conditions over periods of a few days. This has implications for the amount of energy storage that would remove the mismatch between variable wind power production and demand. Lulls are followed by periods of high winds with delays of typically 3-6 days and very rarely over 12 days. Time variability of wind is discussed in more detail in Sørensen (2010a). An impression of the situation can be obtained from looking at Fig. 5.55, which gives the hourly variation for the total Danish production for the wind turbine capacity needed for the 2050 scenario described in the following subsection.

Figure 5.49. Off-shore areas set aside during 1990s for wind power production by Danish planning legislation. For each area, the legend gives an estimate of the total annual production possible from wind parks in that area. Considerations of minimum spacing between turbines and optimal configuration enter these estimates (based on Danish Power Utilities, 1997; Sørensen et al. 2001).

For bio-energy resources, residues from households and food industry are available year round, and fresh biomass, whether harvested or collected, can usually be stored, so that conversion may take place at a time convenient for the conversion capacity or desirable from the point of view of users. The scenario use of biomass is primarily for production of biogas, liquid biofuels such as methanol or eventually hydrogen.

For solar energy, the availability in Denmark, including spatial and time variations, was used in Sørensen (2010a, Ch. 6) to illustrate system simulation techniques. In the present scenario, only modest contributions from solar energy (thermal panels and photovoltaics) are counted on, so the details will not be repeated here.

Construction of 2050 scenarios for Denmark

←A

Figure 5.50. Structure of hydrogen and renewable energy scenarios for 2050. The main source of hydrogen is surplus renewable energy (wind and solar power available at times where production exceeds demand). In the centralised scenario (A, on top), hydrogen production and storage are at central locations and pipeline transmission carries hydrogen to load areas such as filling stations. For the decentralised scenario (B, below; see Fig. 5.4), hydrogen production from excess power takes place in buildings, as does storage and vehicle filling. Reversible fuel cells may regenerate electricity at times of insufficient production (Sørensen et al., 2004).

←B

The idea behind the two hydrogen and renewable energy scenarios considered for Denmark in the year 2050 (Fig. 5.50) is generally applicable for any part of the world with proper substitution of the mix of renewable energy sources most relevant for the particular region considered. The centralised scenario (A) assumes hydrogen to be produced and stored in a few locations and distributed by pipeline. The decentralised scenario (B) assumes buildingintegrated production and storage.

Figure 5.51. Total annual wind power production in the centralised 2050 scenario for Denmark. The caption indicates how much off-shore production is derived from each numbered area (Sørensen et al., 2001)

Centralised scenario

The centralised scenario corresponding to the outline in Fig. 5.50A is based on primary energy production from a mix of renewable energy sources available in Denmark. As seen from the overview scenario energy flows in Fig. 5.54, the most significant Danish energy source is definitely wind power, followed by bio-energy production from agricultural and forestry residues (i.e., not interfering with food or wood production by introducing dedicated energy crops).

Figure 5.52. All electric power production from renewable sources in the Danish 2050 centralised scenario, including the wind production onshore and offshore (as indicated in Fig. 5.51), plus solar power from rooftop photovoltaic installations placed on south-facing building surfaces and roofs (of which about 25% of all suitable surfaces are assumed to be in use). Grid cells with less than 10 TJ/y production represent solar energy, the ones with higher cell production wind.

The renewable energy mix available in Denmark is dominated by energy suited for conversion into electricity, making available electric power far exceed demand of this energy form. This means that electricity can be used for all medium- and high-temperature process heat and can supplement solar heat (the total production of which is given in Fig. 5.54) in case of insufficient heat production (heat stores empty). This conversion is assumed to use heat pumps for maximum efficiency. The time distribution of power production is such that large surpluses are produced during windy periods, as indicated in Fig. 5.55 (a fuller discussion of time variations in wind production may be found in Sørensen, 2010a). The surplus power production is assumed to be used for production of hydrogen in centralised facilities, placed approximately where there are current electric power plants (of which Denmark currently has about 30 000, including larger plants of 200-1000 MW as well as smaller combined heat and power plants, typically rated at 10-200 MW). The distribution of such hydrogen production sites is shown in Fig. 5.53.

Figure 5.53. Hydrogen production in the Danish 2050 scenario (Sørensen et al., 2001).

In Fig. 5.52, the geographical distribution of the total electricity production in the 2050 scenario is shown, adding photovoltaic production (21 PJ y-1) to that of wind (280 PJ y-1). In climatic conditions other than the Danish one, the optimum share of photovoltaic relative to wind power may be much higher. The hourly time distributions of wind and photovoltaic power are given in Figs. 5.55 and 5.56 and are combined in Figs. 5.57 (year) and 5.58 (week), with the demand for electric power indicated (dedicated power plus other demands covered by electricity in the scenario; see the overview in Fig. 5.54).

The 2050 scenario assumes that electricity from intermittent sources is preferably used directly, in order to avoid losses associated with going through storage cycles. Dedicated electricity uses have top priority, followed by mechanical energy needs (motors) and high-temperature industrial process heat. Then follows heat pumps and lower temperature heat−for example, refilling heat stores. The surplus still remaining is used to produce hydrogen. This still amounts to 242 PJ y-1, implying that most of the electricity produced goes through the hydrogen cycle, either to be used as hydrogen in industry or in transportation or to be reconverted to electricity at times when production is lower than demand.

Figure 5.54. Overview of centralised hydrogen and renewable energy scenario for Denmark 2050. Each box represents a conversion step. The number in the lower left of the box is the input energy, and that in the lower right the output energy. The middle number is the difference between input and output−that is, the energy loss (all numbers in TJ/y). The number in the upper part of the box is the assumed conversion efficiency. In case of combined power and heat production there are two efficiencies. The different signatures of lines between boxes indicate energy forms. In some cases, there is recycling, for example, of waste heat. In the upper right is a breakdown of transportation end-use energies (based on Sørensen et al., 2001).

Figure 5.55. Hourly wind power production in the 2050 scenario for Denmark.

Figure 5.56. Hourly photovoltaic power production in the 2050 scenario.

Figure 5.57. Hourly total power production and use in the Danish 2050 scenario (Sørensen et al., 2001).

Figure 5.58. As Fig. 5.57, but for a single spring week.

Figure 5.59. The amount of electricity demand that can be covered directly from wind and photovoltaic power as it is produced (Sørensen et al., 2001).

Figure 5.59 shows the direct electricity delivery, showing the effect of the low wind periods. They are seen to be time-wise shallow and fairly evenly distributed over the year. One reason for this is the positive seasonal correlation between demand and power production, and the other is the structure of weather system passages. As a result, the storage requirements for hydrogen are favourable, requiring no more than days and occasionally a few weeks of energy usage to be held in the hydrogen stores. The scenario assumption regarding inclusion of hydrogen stores is discussed following. The location of stores is shown in Figs. 5.2 and 5.3.

The amount of solar heat produced by thermal collectors in central configurations or mounted on rooftops or building surfaces is shown in Fig. 5.60. Its variation over the year is similar to the power production by photovoltaic collectors mounted on similar surfaces, except for a spring/autumn asymmetry caused by the higher thermal storage and hence collector inlet temperature causing lower efficiency in autumn. Some of the installations may be combined photovoltaic-thermal collectors, where the waste heat from a photovoltaic panel is made useful (see Sørensen, 2010a). As power production by the photovoltaic surfaces is only around 15% of incident solar radiation, the thermal collection may still be near 50% of the energy in the solar radiation, as in the case of pure thermal collectors.

Figures 5.61 and 5.62 show the hourly collected solar heat along with the low-temperature heat demand. Building heat losses and performance of the solar heat system are modelled as described in Chapter 6 of Sørensen (2010a). It is seen as expected that demand cannot be met in winter, whereas summer heat collection with suitable heat storage (day-night or a few overcast days) is consistent with the demand, which during summer is primarily for hot water.

Figure 5.60. The hourly amount of heat for space heating and hot water produced by solar thermal collectors in the 2050 scenario.

Figure 5.63 shows the annual time variation in use of electricity for hydrogen production, by use of reversely operated fuel cells or conventional electrolysers, along with the much smaller use for heat pumps. A detailed view for a single spring week of the same is shown in Fig. 5.93. Figures 5.94 and 5.95 show how waste heat from hydrogen fuel cells may provide additional lowtemperature heat, although smaller amounts than those drawn from central heat stores, except for the winter period.

In Figs. 5.67 and 5.68, the time variations in the disposition of hydrogen in the energy system are depicted. The automotive sector is assumed to use hydrogen through filling stations distributed geographically as shown in Fig. 5.3. No seasonal variation has been assumed, but only a variation over the hours of the day. Regeneration of electric power by use of fuel cells appears in Fig. 5.67 (and for a single week in Fig. 5.68) as fairly narrow features. Those in the lower part of the figures are based on direct hydrogen supply during hours of excess power production. The larger amounts appearing in the upper part of the figures are based on hydrogen drawn from one of the two central hydrogen storage facilities shown in Fig. 5.2 or Fig. 5.3.

Figure 5.61. Comparison of low-temperature heat collection by thermal solar panels with demand, on an hourly basis, for the 2050 scenario.

Figure 5.62. Same as Fig. 5.61, but for a single spring week.

Figure 5.63. Hourly amounts of electricity used for hydrogen production by fuel cell electrolysers and for heat pumps in the Danish 2050 scenario.

Figure 5.64. Same as Fig. 5.63, for a single spring week.

Figure 5.65. Sources of low-temperature heat in the 2050 Danish scenario. Time variations are shown for heat drawn from central stores (filled using heat pumps at times of excess electricity production) and waste heat from central fuel cell operation to generate power. This heat is distributed through district heating pipelines.

Figure 5.66. Same as Fig. 5.65, but for a single spring week.

Figure 5.67. Hourly conversion of hydrogen in the 2050 scenario. The curve varying regularly between 70 and 80 PJ/y is the (schematic) consumption by vehicles, and the strongly varying curves are the fuel cell electricity production in cases of insufficient direct power production. The lower part is direct use of hydrogen when production is ongoing, and the top part is based on hydrogen drawn from stores.

Figure 5.68. Same as Fig. 5.67, but for a single spring week.

According to the overview in Fig. 5.54, 97 PJ y-1 of hydrogen is used to regenerate 50 PJ y-1 of electricity, while the hydrogen used in the transportation sector is 100 PJ y-1. The scenario assumes hydrogen used not only in passenger cars, but also in airplanes and ships. The demand scenario is based on a continued expansion of air traffic, making the energy requirement for planes and cars about equal in the year 2050.

The large Danish agricultural sector, along with managed silviculture, produces biomass residues allowing a fairly large energy production, which in the scenario was assumed in the form of methanol (18.4 PJ y-1). This liquid fuel was taken to be used in particular transportation sub-sectors, such as lorries and possibly buses. Additionally (in view of the abundance of electricity characterising a renewable energy-based future Danish situation), electric vehicles are assumed to be present, including all trains and a special fleet of small city cars available in the only large city of Denmark, Copenhagen. These details are shown in Fig. 5.3. It is also seen how the two central hydrogen stores in the scenario are conveniently located in different regions of the country. This is not surprising, since they are assumed to be modifications of the already existing natural gas stores serving the current energy system as strategic security stores created as a safeguard against rupture of the main natural gas pipeline from the North Sea.

Figure 5.69. Hourly level of hydrogen stored in the two central underground facilities (top) and hydrogen production (lower part of figure) in the centralised 2050 renewable energy scenario for Denmark (Sørensen et al., 2004).

In order to quantify the hydrogen storage requirement, a time simulation of the entire system has been made, using a specific and typical year of hourly data shown preceding for wind, solar radiation, and various components of energy demand (scaled to 2050 usage assumptions). The outcome is a sequence of hourly data quantifying the degree of filling of the two hydrogen stores (one aquifer and one salt cavern; see Fig. 5.3 in section 5.1), shown in Fig. 5.69, along with variations in total hydrogen production (the upper ceiling being the rated capacity of the production facilities). The conclusion is that 60 000 PJh y-1 of storage capacity allows the system to run smoothly at all times.

The amount of 60 000 PJh y-1 or around 7 PJ is actually only a fraction of the capacity of the existing natural gas stores (see section 2.4.1 in Chapter 2 and Figs. 5.1-5.3), implying that hydrogen may be stored at low pressure (say 5 MPa), which means low cost and little need to seal the storage caverns that are imbedded in clay layers of low hydrogen diffusivity. Figure 5.3 indicates that the centralised renewable energy hydrogen system has a fairly high infrastructure similarity to the present system.

Decentralised scenario

One may notice already today a tendency for decentralisation in society, replacing the centralised systems characterising early industrialisation by increasingly decentralised facilities. Much business activity (offices, sales outlets) has moved away from city centres. It has become attractive to offer employees work in rural environments for reasons of avoiding urban traffic and obtaining a relaxed environment. This has been made possible by current electronic facilities allowing communication from any location, facilitating not only industrial activities, but also residential recreation. Transport by road has offered strong competition to earlier long-distance freight transport by railroads, increasing numbers of cars have substantially augmented mobility, and air transport for business and leisure has become commonplace in industrialised countries, where working schedules allow individual vacation trips to any destination in the world, often several times each year. Citizens of many countries have acquired personal command over energy for residential heating as well as automobile transportation, and it is likely that the prospect of becoming one's own electricity producer by building-integrated fuel cells is appealing to many people. This would imply an acceptance of a possibly higher price for such installations, relative to the centralised solutions. Of course, this is true only within certain limits, but precisely fuel cell technology, at least of the PEM type, appears to offer a modular concept with rather modest mass production advantage. Furthermore, if the residential installation can not only provide electric power and associated heat, but also provide hydrogen for the hybrid-vehicle(s) parked in a garage, the concept does look very appealing to the paradigm of being in personal control.

Figure 5.70. Wind power production in the 2050 decentralised Danish scenario.

The outline shown in Fig. 5.50B indicates how this may be achieved. The diagram assumes that reversible fuel cells are available in 2050, with high efficiency both for direct and reversed operation. The higher efficiency of fuel cells relative to current power plants implies that if waste heat should be able to cover all space and hot water heat requirements, more efficient means than the presently used ones should be employed.

The decentralised 2050 Danish scenario is constructed in a somewhat different way from the centralised one, with significantly lower energy demand, particularly in the transportation sector. The centralised scenario assumes a standard activity development typical of current thinking by economists, where the transportation end-use demand by 2050 is three times higher than today. The decentralised scenario, on the other hand, considers this development as unattractive, in view of the already annoying congestion on roads and in the air, and opts for a more modest growth, as seen in the scenario overview shown in Fig. 5.71. The transportation activity level and energy use for electric appliances remains at the present level. The latter assumption does of course not mean that the activity level is stationary, but expresses a balance between new and increasing activities and efficiency improvements.

Figure 5.71. Overview of decentralised hydrogen and renewable energy scenario for Denmark 2050. Each box represents a conversion step. The numbers in the lower left of the box are the input energy, and that in the lower right is the output energy. The middle number is the difference between input and output, i.e., the energy loss (all numbers in TJ/y). The number in the upper part of the box is the assumed conversion efficiency. In case of combined power and heat production, there are two efficiencies. The different signatures of lines between boxes indicate energy forms. In some cases, there is recycling, for example, of waste heat. In the upper right is a breakdown of transportation end-use energies (based on Sørensen et al., 2001).

Figure 5.72. Geographical distribution of hydrogen production in the 2050 decentralised scenario for Denmark. Most producers are in buildings.

The fairly extreme emphasis on energy efficiency underlying the decentralised scenario assumptions is not based on a conservationist "enough is enough" philosophy, but rather expresses the well-documented fact that there are numerous energy efficiency improvements that are not undertaken today, despite the fact that they are by far cheaper (per unit of energy use avoided compared with an energy unit to be purchased) than any of the new energy system components considered in the scenarios (renewable energy and hydrogen converters). At present, many of the economically viable energy and technically feasible efficiency measures are left undone, because they have a lower status among decision-makers and consumers than new energy provision technologies (Sørensen, 1991).

As a result of these assumptions, the required energy production is less in the decentralised scenario as compared with the centralised one, allowing wind power production to be reduced to 106 PJ y-1 and therefore occupying fewer of the designated off-shore sites around Denmark, as shown in Fig. 5.70. The hydrogen production sites are now in buildings, as illustrated in Fig. 5.72, in contrast to the limited number of sites used in the centralised scenario (Fig. 5.53).

Figure 5.73a, b. Hourly level of hydrogen stored in all building-integrated hydrogen storage facilities in the decentralised 2050 scenario (top), and same with off-shore wind production increased from 99 to 180 PJ/y (bottom) (Sørensen et al., 2001).

Figure 5.73a shows the hourly combined filling of all the hydrogen stores assumed placed decentrally in buildings. The combined storage capacity is taken as the same as for the two central stores in the centralised scenario. It is seen that the system is more prone to having low storage filling, but still fulfils its purpose except for a few hours of the year. This is due to the higher demand on storage caused by not having hydrogen production based on biomass (omitted due to perceived concerns about removing carbon from the soil, carbon which is released to the atmosphere as CO2 in compensation for carbon assimilated by the plants at an earlier time). Furthermore, the use of hydrogen in the building system is different from that of the centralised facilities. Increasing the stores does not alter this picture, and the emptying of the decentralised stores can only be avoided by increasing the production, for example, by adding more electricity production from wind (the surplus being then converted into hydrogen). Figure 5.73b shows that completely emptying stores can in this way be avoided, although variations in degree of filling are still substantial. The extra cost of increasing production just for avoiding a few hours of hydrogen supply-demand mismatch is not considered justifiable, as the tiny amounts of extra hydrogen can be produced from imported electricity or from small reserve fossil plants being operated for only a few hours a year (e.g., gas turbines).

Assessment of renewable energy scenarios

The renewable energy scenarios are more complex to model than scenarios based on fossil and nuclear energy. This is primarily because the efficiency of solar heat panels depends on the temperature of the water (or other medium) circulated through the solar collectors, causing a complicated coupling between the time development of heat loads, heat stores, and solar collectors, where the performance of the system depends on its previous history of behaviour. The other complication is the substantial amounts of energy storage required to cope with the intermittency of renewable sources such as wind and solar energy. In fuel-based systems the fuel itself acts like an energy store. This behaviour is mimicked in the renewable energy scenarios by using intermediate fuels such as liquid biofuels and hydrogen, which can fairly easily be stored by use of the different kinds of devices discussed in Chapter 2. The scenario method has allowed a demonstration of the ability of the renewable energy systems to function in a dependable and resilient manner, under conditions of varying demand and natural variations in renewable energy inputs.

An important condition for this behaviour is the maintenance of a strong electricity grid of both regional transmission lines and local distribution lines. This grid transports renewable power to loads or to fuel cells (or other electrolysers) producing hydrogen and again transports power generated in fuel cells from hydrogen during periods of insufficient renewable energy inputs to the system (see Sørensen, 1081; Meibom et al., 1999). The additional grid systems available today are natural gas pipelines and district heating lines (both of which are abundant in the Danish energy system although not available to every building). These systems are supposed to be retained: the district heating lines for carrying heat to seasonal stores (which are not building-integrated) and for adding heat supply security, and the gas lines (converted to transport hydrogen) for additional flexibility in providing hydrogen for either transportation or stationary uses.

Heat demands are covered by a prioritised succession of options: first solar collectors, then heat stores and finally electric heating by heat pumps (which in areas with district heating lines may be placed centrally, for maximum efficiency). High-quality energy demands are first covered by direct renewable electricity and then by hydrogen (or methanol) or electricity regenerated from hydrogen. Surplus electricity after direct usage and heat pump demands is converted to hydrogen, and the power production capability (total rating) of wind and photovoltaic converters is chosen such that all stationary demands can be met by the preceding succession of options and all transportation energy needs can be met by available hydrogen, methanol or electric power. This determines the rating of the primary renewable energy collectors required, with values as given in Figs. 5.54 and 5.71 for the centralised and decentralised renewable energy scenarios. For wind power, the current 4000 turbines placed on land are supposed to be replaced by new ones before the year 2050, with a unit turbine rating of at least 2 MW. For off-shore wind parks, the decentralised scenario requires less than 2000 turbines, and the centralised scenario about 4000, if the average unit rating is 4 MW. The offshore installations have a rated power of 6.8 GW in the decentralised scenario and 14.7 GW in the centralised one. The density of turbines on the water has been assumed to be 8 MW km⁻² , in order to avoid significant "shadowing" effects (reduced power of narrowly spaced turbines). Methanol production from biomass uses technology described in Sørensen (2010a).

In describing a plan for implementation over the period from now until the year 2050, a gradual lowering of price as a function of accumulated manufacture of critical cost items such as fuel cells and solar cells must be assumed, as further discussed in Chapter 6. An important technical issue in determining cost is the lifetime of fuel cell components. Other technical issues relate to infrastructure changes implied in the transition from the present system to the scenario pictures of hydrogen filling stations, hydrogen stores and transmission. There are further areas presently very far from viability, such as hydrogen-fuelled airplanes. It would be natural during the transition period to reserve the declining fossil fuel usage to such applications. Filling station replacement suggests an implementation strategy beginning with dedicated fleets of vehicles to operate on hydrogen, and regarding hydrogen pipeline systems it may be necessary to establish new parallel ones during a transition period when natural gas pipelines are still used for natural gas. The alternative of driving hydrogen containers around on the road system is not attractive in a system aiming at using resources efficiently. While central hydrogen stores can be established fairly easily (also without using existing natural gas facilities, as many suitable geological formations are available in most parts of the world), the safer alternatives to compressed hydrogen storage in containers, such as metal hydrides, will need a few decades to reach viability, even if the technical problem solution goes smoothly. Fuel cell electricity production may be by low-temperature PEM cells (or further developments using intermediate temperatures) and SOFCs for stationary uses. Electrolysis may be by reversible fuel cells, optimised for two-way operation. Conventional electrolysis (one-way reverse fuel cells) is the fallback option.

Countries accomplishing an early penetration of fluctuating renewable energy sources will be more interested in an early entry of hydrogen technologies. The main driving forces of increasing fossil fuel prices and delivery problems in connection with the bulk of such resources located in politically unstable regions make it less likely that early introduction of hydrogen based, during a transitional period, on fossil fuels will be successful, except in particular areas where a surplus of fuels, particularly natural gas, may be available locally.

5.5.3 New regional scenarios

In recent years, scenario work has focussed on the additional advantages that may be derived from regional energy collaboration. This is particularly relevant for neighbouring countries with different renewable assets, such as one country having forestry or agriculture that could generate large quantities of biomass waste, another having coastal locations with high wind potential, and a third having many hours of sunshine that would perhaps not be suited for large energy production when seen isolated (say, due to mismatch between seasonal variations in solar radiation and electricity loads), but where coordinating the energy production and coverage in the combined region, assuming presence or establishment of the required transmission and piping facilities, could greatly stabilise the energy supply situation and offer a resilient system perhaps even at a lower cost than if the countries only acted on their own. Such synergisms have been identified for the North European region (Sørensen, 2008a), the North American region (Sørensen, 2007d), and the Mediterranean region (Sørensen, 2010a, 2011c). Also the discussion of future energy demand has been expanded, to comprise several demand scenarios reflecting different development in goals and preferences in human communities (Sørensen, 2008b; 2010a).

This chapter will close by mention of a few highlights from the regional study made for five North European countries: Norway, Sweden, Finland, Denmark, and Germany, because it forms a nice extension of the Danish scenario work of section 5.5.2. Table 5.5 gives a survey of the renewable energy resources identified for the five countries. A cautious level of exploitation has been adopted: Potential biomass production as shown in Fig. 5.74 comprise only an easily collectable part of residues from forestry and agriculture, plus an aquaculture harvest deriving from near-shore locations (i.e., not expanding present inland freshwater fish ponds). Similarly, the wind potential detailed on a per land/ocean surface area or per capita basis in Fig. 5.75 assumes use of only 0.01% of the total land area for inland wind turbines (i.e., the vertical area swept by the turbines equals 0.01% of the horizontal land area), and 0.1% of off-shore areas with water depth below 40 metres and not in use for other purposes (such as fishing, boat traffic, military, etc.). The higher production on land identified in Denmark is consistent with using the locations where a wind turbine is already installed (although in many cases an older model with smaller production than that of a current turbine), and the potential off-shore energy production is similar to that of a study identifying site-specific locations suitable for off-shore wind farms, made by the Danish utilities in the late 1990s (Danish Power Utilities, 1997).

Country:	DK	N	S	SF	D
Wind on-shore	64	167	201	147	157
Wind off-shore	358	974	579	391	177
Biofuels from agriculture	241	51	111	49	1993
Biofuels from forestry	58	523	1670	1180	892
Biofuels from aquaculture	153	223	320	205	108
Hydro	-	510	263	49	27
Solar PVT electricity	-	-	-	-	129
Solar PVT heat	-	-	-	-	275

Table 5.5. Summary of potential renewable energy supply available for use in the North-European countries considered in a 2060 scenario (unit PJ/y). PVT is combined photovoltaic and thermal collectors (Sørensen, 2008a).

It is seen that both wind and bio-energy resources are abundant in the Nordic countries. For wind, comparing Figs. 5.75a and 5.75b shows that relative to land area, Denmark is most endowed, but per capita Norway leads. Currently, wind power is neglected in Norway, where off-shore fossil fuels are exploited in a way that recover much of the revenues at state level, planned to be used to benefit future Norwegian generations after the oil and natural gas resources have started to vane. Presently, Norway contributes significantly to CO2 and methane emissions worldwide, from its extraction, treatment and use of oil and gas, and many Norwegian citizens have had difficulty in adjusting gracefully to the sudden wealth, just like in the Middle Eastern oil-producing countries. For Germany, resources of wind and other renewable resources are seen to be small when the high population is taken into consideration. Solar power or heat may contribute more significantly than in the Nordic countries, but self-sufficiency based on renewable resources will require a much harder exploitation than what have been assumed in this 2060 scenario. However, the scenario offers a different solution, based on regional trade and collaboration in meeting energy demands.

Figure 5.74. Potential energy production from biomass residues in the five North European countries considered for the 2060 scenario (W/cap.). This and following pictures in Chapter 5 are from Sørensen et al. (2008).

Figures 5.76a and 5.76b give an example of the annual variations in wind energy production, for all offshore and all onshore locations Norway, at 6-hour intervals. The data are from a model using blended satellite and ground based data in a grid of about 25 × 25 × cos(ϕ) km2 , where ϕ is the latitude. The satellite scatterometer data are most accurate over sea, where measurements are rare, and all wind data have acquired improved spatial consistency by being fed into a global circulation model. The scatterometer radar data represent wind speeds at a few metre above sea level, and a standard method based on the roughness parameter characterising the ocean surface has been used to extrapolate to the height of turbine hubs where rotor blades would typically be mounted (some 80 m for current wind turbines); see Sørensen (2010a).

The Norwegian part of the 2060 scenario assumes the disposition of the power produced by wind, for covering loads in Norway, shown in Figs. 5.77 and 5.78. This leaves a substantial surplus of produced electricity, which cannot be made useful in Norway. The scenario further predicts a surplus of electricity from hydro and a surplus of liquid biofuels produced from biomass residues. These are shown in Fig. 5.79, distributed on 6-h intervals over the scenario year, according to the assumed share of wind- and hydropower.

Figure 5.75a, b. Estimated potential wind power production with a turbine-swept area that is 0.01% of the land area (except 0.02% for Denmark, as this area is already currently in use for wind) and 0.1% of the area at off-shore locations identified as being suited for foundation work and not interfering with other establish waterway uses. The potential in annual average kW or W is evaluated on a land-area basis (top) or on a per capita basis (bottom).

Figure 5.76a, b. Potential wind power production in Norway, at 6-hour intervals during a typical year. The top panel gives offshore production at 0.1% of all suitable locations near the coast, and the lower panel the production on 0.01% of the land area. The calculations are made using the blended wind data from a general circulation model for the land areas and satellite scatterometer data for the ocean parts, as described in Sørensen (2008b; 2008c): The circulation model used for land areas serves to smooth the observed wind data used as input. Scaling to a typical wind turbine hub height of about 80m has been done here and in Table 5.5/Fig. 5.75.

Figure 5.77. Partial disposition of wind power produced by Norway in 2060 scenario, to direct Norwegian electricity loads and devices for providing high-temperature heat (primarily to industry) and low-temperature heat (primarily for heat-pumps used to produce space and utility heating in buildings).

Figure 5.78. Further disposition of wind power produced by Norway in 2060 scenario, through hydrogen production to furnish coverage of 50% of the end-use energy for transportation. The remaining 50% is assumed covered by biofuels, but the primary biofuel energy exceeds 50%, due to the lower conversion efficiency relative to that assumed for the hydrogenusing fuel cells.

The 2060 scenario message is that the surplus of electricity and biofuels in Norway, together with similar surpluses in the other Nordic countries, are more than sufficient to cover the deficits in a (recently decided) non-nuclear Germany. There is even a choice between exporting electricity to Germany, where it would primarily be transformed into hydrogen for use in the transportation sector, or alternatively to export biofuels that Germany may use to cover most of its deficits, which occurs mainly in the transportation sector. In other words, the Nordic countries and Germany would be self-sufficient in energy coverage by from indigenous renewable sources, and even have additional surpluses to export to other countries further south in Europe.

Figure 5.79. Surpluses for export from various renewable energy sources in Norway. The wind power is shown by 6-hour intervals over the 2060 scenario year (using wind data from a recent year), the hydropower is shown monthly (based on reservoir filling), and the biofuels just shown as an average (as the biomass or biofuels can easily be stored before or after production).

5.6 Problems and discussion topics

5.6.1

Try to predict the succession of events that would be needed for a smooth transition from the present infrastructure to the one required for hydrogen use in stationary and mobile sectors.

5.6.2

Construct a scenario for your country or region in analogy to the ones described in this chapter. First, enumerate the energy sources that would be most appropriate in your setting, and then use available data (or the scenarios in this chapter) to determine the amount of each primary energy form needed for a consistent system matching supply and demand. You may want to make different assumptions regarding future demand (activity expansion and efficiency improvement), and thereby see what restrictions the demand assumptions make on the other options.

5.6.3

Discuss how long it would be possible to maintain a hydrogen-based energy system if all hydrogen had to be produced from natural gas as the sole primary resource (e.g., use the natural gas reserve estimates given in section 5.3).

CHAPTER 6

SOCIAL IMPLICATIONS

6.1 Cost expectations

6.1.1 Hydrogen production costs

The cost¤ of hydrogen production from natural gas by steam reforming methods is well established once the fuel input cost is known (Amos, 1998; Longanbach et al., 2002; Simbeck and Chang, 2002). At low (pipeline) pressure, it is about US\$1.0 kg-1 based on inexpensive natural gas available at the time of estimation (assumed at US\$1.5 GJ-1 or under half of the 2004 price). Filled into pressurised flasks the price is 30% higher, and if liquefied more than twice higher. On the other hand, the smaller volume makes transportation costs lower, and dispensed from a central hydrogen production plant to the final user, the hydrogen from pipeline natural gas has the highest cost (US\$5 kg-1) and that based on liquefied gas has the lowest cost (US\$3.7 kg-1), depending of course on the assumptions regarding transportation distances and technology. Simbeck and Chang (2002) estimate that the delivered H2 cost from biomass waste is only about US\$2.5 kg-1 and from electricity (at 2001 US electricity costs) by electrolysis about US\$5 kg-1. Conventional smallscale electrolyser H2 costs have been estimated by Fingersh (2003) and by Padró and Putsche (1999) as US\$8-12 kg-1, but larger units using surplus wind power may attain a cost down to US\$2 kg-1. Based on coal gasification, the H2 cost is estimated as over US\$12 kg-1. No externalities are included in these cost estimates.

^¤ One must generally be careful in using cost estimates from different countries with different taxation and subsidy structures. In particular, one should be concerned over use of current energy costs for estimating break-even costs of new technologies.

If conversion of surplus electricity to hydrogen is made by a fuel cell in reverse operation, but paid for on the basis of its electricity production, and if the cost of a reversible fuel cell can be assumed similar to that of a one-way cell, then the hydrogen cost based upon fuel cell electrolysis may be as low as the cost of the power used (depending on how costs are allocated to the two process directions). The power used for hydrogen production could thus be off-peak power or surplus power from renewable energy systems such as wind and photovoltaics, which cannot be made useful for covering demand at the time of generation. In recent years, the cost of such power at auction pools is often lower than 2 (US- or euro-)cents per kWh, corresponding to about US\$6 GJ-1 of hydrogen at some 90% conversion efficiency.

The cost of hydrogen from (most likely lignocellulosic) biomass gasification and shift reaction (see Chapter 2) has been estimated as US\$9-14 GJ-1 by Faaij and Hamelinck (2002), with a long-term potential for reduction to US\$4-9 GJ⁻¹ . The higher value reflects inclusion of the production cost under European conditions (most likely from forestry), and if discarded residues of no economic value can be used, lower price outcomes are possible. Production of biomass-based hydrogen from photosynthetic algae involves even higher costs. Kondo et al. (2002) estimate hydrogen costs of 235-574 yen m⁻³ or US\$188-460 GJ-1, for a 5000 m3 day-1 prototype system based on a Rhodobacter sphaeroides mutant.

In case methanol is used as an intermediate fuel or in direct methanol fuel cells, the cost of methanol production is of interest. Produced from fossil fuels, notably natural gas, at a price of US\$3 GJ-1, reforming or series reactor schemes lead to a methanol production cost estimated around US\$5.5 GJ-1 (Lange, 1997). Advanced micro-structured string-reactors for this concept are under development (Horny et al., 2004).

6.1.2 Fuel cell costs

Concepts such as molten carbonate or solid oxide fuel cells may not reach a commercial stage during the next decade. Preliminary cost estimates for a possible SOFC commercialisation are at US\$3200 kW-1 by 2010, 1300 by 2050 (Fukushima et al., 2004), or as low as US\$350 kW-1 (EG&G, 2004). Problem areas include the availability of La, currently used mainly in high-tempera-ture ceramics. Recycling should in this case be pursued to a high rate.

Due to their early success, alkaline fuel cells (AFCs) appear less expensive than proton exchange membrane (PEM) fuel cells, although both are still above the requirements for commercial use. However, the prospects for rapid cost decline envisaged (and necessary) for PEM cells do not seem to be paralleled for AFCs. McLean et al. (2002) explain this as due to lack of interest in further development of the AFC technology. They quote the cost of a small batch of AFCs as US\$1750 kW-1 with a potential for reduction to US\$155 kW-1 for mass production at a large but unknown scale. They further assume the cost of a small batch of PEM cells as above US\$2000 kW-1, but with potential for reduction to US\$20 kW-1 in mass-produced automotive 50 kW stack. This is lower than the target of 30 euro kW-1 by 2025 for a level of production amounting to 250 000 fuel cell units per year (Danish Hydrogen Committee, 1998). Peripheral components (gas circulation systems, piping, and electronics plus electrolyte recirculation for AFC, humidifier for PEM) are estimated to cost three times more for PEMs than for AFCs, and although these are small compared with current fuel cell stack costs, they would become more important if the stack costs reach the lower bound.

More recent estimates of the gain by massive introduction of PEM cells into vehicles arrive at a future cost as a function of both the accumulated volume of production and the following two parameters: power density improvement (increasing from current 2 to 5 kW m⁻² ) and speed of maturing (taken as slope of an assumed logarithmic learning curve), that ranges from 15 to 392 US\$ kW-1 (Tsuchiya and Kobayashi, 2004). The lower cost estimate assumes an accumulated 5 million fuel cell vehicles by 2020, with average rated fuel cell power of 110 kW, while the upper estimate assumes 50 000 vehicles and a fuel cell power density having reached 3 kW m⁻² , with the cost being suggested to be a possible minimum achievement for the year 2010. The production volume assumptions may be overestimates, as the most likely role of fuel cells in the automotive sector would be in hybrids (see Chapter 4), with unit rating no larger than 20 kW.

Wang et al. (2011) find 2007 PEM fuel cell costs dominated by the cost of the Pt-catalyst, but predict that thinner catalyst layers will soon reduce this component considerable, making air delivery a major cost. This is in contrast to Mahadevan et al. (2010), who finds the bipolar plates to be the dominating cost, followed by catalyst and gas diffusion layer costs, for a 5 kW fuel cell aimed for stationary uses. It is clear that many factors of importance are still unclear, as regards cost reduction potentials. If the cost goals for specific future years proclaimed by the US and European Departments of Energy are to be realised, the balance of system costs will end up dominating the total.

Vehicle PEM fuel cells are currently aiming for a lifetime of around 5000 h (but having by 2010 reached only about 2000 h), as compared with the 40 000 h set as the minimum for continuous stationary uses. Lipman et al. (2004) expect future oil and gas prices that will make stationary fuel cells break even at US\$1200 kW-1 for a 5 kW home systems and US\$700 kW-1 for larger, 250 kW systems. For vehicle fuel cells, they suggest that vehicles when parked in a garage could contribute to electricity production for the building system. This would seem at variance with the shorter lifetime of the fuel cells designed for vehicle use. The US Department of Energy estimates that fuel cell costs using current technology could be as low as US\$50/kW, if only the production volume would be half a million units per year (US DoE, 2010a). The 80 kW PEM fuel cells considered as a basis for many US cost studies (e.g., Carlson et al., 2005) may as mentioned be oversized for realistic vehicle propulsion systems, where a smaller fuel cell could become accepted at an earlier time, even at higher per kW cost (Baptista et al., 2010).

Figure 6.1. Observed learning curves for wind turbine and photovoltaic module costs used to suggest possible learning behaviour for PEM fuel cell stacks. The total accumulated capacity at a given time is along the abscissa, and the average cost of power production is along the ordinate. It is taken as the cost per kW of installed power divided by cP, which is the ratio between average annual production and the device rated power level (i.e., the equivalent fraction of time at maximum power production). The extrapolation lines for automotive PEM fuel cells correspond to learning at typical rates characterising the wind and photovoltaic industry over prolonged periods. Data vary considerably, depending on the siting of devices such as wind turbines (Lauritsen et al., 1996; Neij et al., 2003: Morthorst et al., 2008; Madsen, 2002; GWEC, 2010; WWEA, 2011) or photovoltaic modules (IEA-PVPS, 2010, with limited geographical coverage; Schaeffer et al., 2004; Borenstein, 2008; Lushetsky, 2010). The early penetration PEM fuel cell cost is based on Danish Hydrogen Programme and European Commission Citaro F bus tenders 1998-2002, the more recent cost estimates on Mahadevan et al., 2010, and the lower trend curve on US DoE, 2010a.

It may be instructive to compare fuel cell prospects with the empirical learning curves for other technologies in the energy sector, such as wind and photovoltaic power, or with high-density battery developments. Figure 6.1 gives the results of an analysis of the wind and photovoltaic learning curves. Economists often describe such findings in terms of straight-line logarithmic behaviour, writing the cost Y as a function of accumulated volume of production X,

$$\log Y(X) = -r \log X + \text{constant.} \tag{6.1}$$

The slope —r is sometimes quoted in terms of a "progress ratio" PR = 2−^r or the "learning rate" LR = 1 − PR.

Figure 6.1 shows that quite complicated features are found in empirical learning curves (in 2004 euros by use of deflators from UK Treasury, 2004).

For wind turbine costs, the flat or rising cost curve during the 1980s reflects a period where tax evasion schemes made it possible to sell wind parks in the United States (particularly California) without worrying too much about cost. By the end of the decade, the Californian market collapsed and prices rapidly went down in an effort to penetrate new markets in other parts of the world. This was not successful, and several manufacturers went bankrupt (vertical part of the curve). After a restructuring of the industry, the diversification strategy worked better, and after a few years with a flat cost curve, the markets really took off and learning curve behaviour set in. The decade 1990-2000 exhibits a learning rate of about 10% in a steadily expanding market (Germany, India, Spain, Denmark, plus several others). The curve in Fig. 6.1 pertains to the performance at the best locations, such as offshore or coastal class-1 wind condition site, where the coefficient of performance reached cP = 0.294 by year 2000, compared with 0.20 by 1981. During the years 1997-2002, much if the expansion has been in offshore wind farms, where the coefficient of performance by 2002 was as high as 0.46 (Sørensen, 2010a). Neither this nor the later decline in cP is due to any radical technical modification, but rather to a change in design philosophy away from aiming at maximum annual energy production to one where more annual operational hours have a higher priority (in practice obtained by keeping the blade design and pitch angle roughly as it were for the on-shore turbines).

The subsequent behaviour in Fig. 6.1 reflects a large expansion in wind projects worldwide, causing prices to stop declining and actually increase, but with new price falls appearing after 2007 due to the worldwide economic crisis following years where banks could not find enough meaningful projects to finance use of the deposits made in a period of strong growth in economic liquidity, notably caused by bubble-economy in the real estate sector. It should be added, that prices even of mature technologies such as wind have become more difficult to obtain, due to the disappearance of manufacturer's standard price lists, and the often very site-specific civil and transmission work included in tenders for large wind projects.

For photovoltaic panels, the straight-line cost approximation works poorly, as inspection of the early curve in Fig. 6.1 reveals jumps from one plateau to a lower one every 3−5 years. This may reflect the introduction of better technology, each time production equipment is replaced. However, a competing influence, particularly visible after 1999, is from the many public subsidy programs (particularly in Germany and Japan). These make it less attractive for the manufacturers to decrease prices, unless tender condition set by the large public programmes forces them to do so. The overall learning rate is initially above 20%, considerably higher than for the more mature wind technology. The recent drop in prices may be connected with closure of major subsidy programmes. The coefficient of performance, cP, is around 0.16, meaning that on average over a year the panels produce energy corresponding to 16% of the one they would have produced if operated in full solar radiation at their maximum (rated) power all seasons and day and night.

Turning now to PEM fuel cells, the emerging technology working only in demonstration programmes does not allow prices to be determined with much accuracy. In 1995, fuel cells were sold to research laboratories and prototype vehicles at prices lower than the starting price shown in Fig. 6.1. However, these cells were without any warranty and often worked for less than a year. When the technology moved from prototype level to small series of 30-80 units after year 2000, with PEM fuel cell buses and passenger cars coming into operation around 2005, most fuel cell manufacturers entered contractual agreements with automobile manufacturers at prices not generally known. Efforts to purchase PEM fuel cells even by national programmes faced difficulties, and the few manufacturers willing to quote prices demanded much more than in 1995. The market still lacks clarity and the cells do not yet have the goal of 5 years of operational life, but rather about 2 years (US DoE, 2006). The prices used for Fig. 6.1 are taken from tenders such as from the European bus project material, assuming that the cell price constitutes about half of the total system price (the estimated cost of a Citaro F bus relative to a similar bus with a conventional diesel traction system). The power coefficient cP is taken as 0.06, reflecting an assumed 1000 h of annual operation and another multiplicative factor of 0.5 due to the vehicles not operating all their driving time at full fuel cell rated power (this factor may vary according to the amount of battery charging from excess power, in the case of hybrid FC-battery vehicles). Current numbers of PEM fuel cells in different applications are discussed by Wang et al. (2011).

The future reduction in PEM fuel cell prices is indicated in Fig. 6.1 in terms of two curves corresponding to learning rates of 10 and 20%, corresponding to the limiting cases of photovoltaic modules and wind turbines over the recent decade. Even with the lower curve, break-even with current vehicle engines would not be reached until the cumulated production reaches more than 500 GW. However, including fuel prices in the assessment, supply problems for oil products (peaking production, instability of major supplier countries) may make hydrogen PEM fuel cell vehicles competitive at higher prices than the currently seen break-even price. Forecasts of market penetration and cost behaviour are abundant but with clear bias of optimism on behalf of hydrogen and fuel cell technologies (USDoE, 2010; Park et al., 2011).

This discussion has shown that many factors influence price development, in addition to industrial "learning". One should also be careful with methods used in the literature, which are often very different from one another, for example, using running instead of as here inflation-corrected prices or comparing prices and cumulative installations only for a single country (Ibenholt, 2002; Junginger et al., 2004; Rosenberg et al., 2010). The learning of manufacturers in a single country may be a reasonable indicator, as technological progress spreads rapidly across regions when not prevented by patent issues, but it is not always the case (Chen et al., 2011). Prices, on the other hand, may vary geographically due say to a new kind of industry not possessing worldwide sales and maintenance infrastructure. Another reminder is that use of double-logarithmic plots can make just about anything look linear, especially if several decades are included for both abscissa and ordinate. Care has been taken in Fig. 6.1 not to encourage this feature, which is striking in a large part of the literature on the subject of learning curves.

Care should be exercised when comparing prices of equipment with different lifetimes. Wind turbines have an established lifetime of around 25 years. The lifetime of photovoltaic panels is believed to be similar or even longer, at least for those based upon crystalline and multicrystalline cells. In contrast, the 5-year lifetime currently aimed at for both automotive and stationary fuel cells should be considered in comparing technologies. One may even argue that the fuel cell curves in Fig. 6.1 should be moved upwards by a factor of five, if the purpose is to compare different technologies fairly. The breakeven of fuel cell vehicles is not affected, because it is already based on the assumed equipment life. Only in case the equipment duration goal cannot be reached (or is surpassed) will the assessment have to be modified.

The time required to meet the target costs for both stationary and mobile PEM fuel cell systems has been explored by the Delphi method (interviewing a number of experts) (Kosugi et al., 2004). The result was a distribution centred around 17 years for both technologies. It matches very well the assumptions of Tsuchiya et al. (2004) just mentioned and is consistent with the goals of US\$40-200/kWrated expressed as needed to break-even in regional fuel cell programmes in Europe, Japan, and the United States (European Commission, 1998; Iwai, 2004; USDoE, 2002a; Chalk et al., 2004; USDoE/T, 2006). The fact that the United States estimate of the break-even price is at the lower end of the range is due to the low price (ignoring externalities) of current subsidised vehicle fuels in that country. Tsuchiya and Kobayashi (2004) arrive at a lower set of curves as compared with Fig. 6.1, because they use a lower present cost of PEM fuel cells, US\$1833/kWrated or about 15 000 euro /kWav, as their starting point for learning. In any case it is evident that Delphi-type oracle-forecasts cannot be taken as proxy for scientific assessment.

6.1.3 Hydrogen storage costs

Hydrogen storage costs comprise capital costs for the equipment used and operational costs, such as power for compression or liquefaction. Shayegan et al. (2004) quote the additional cost of hydrogen recovered from a liquefied store as about US\$5 kg-1 for small units, decreasing to around US\$1 kg-1 for high compressor/liquefier rating. The additional H2 cost going through a compressed hydrogen store in containers is around US\$0.4 kg-1 for shortterm storage, but rises with length of storage (Amos, 1998; Padró and Putsche, 1999). The large-scale underground hydrogen storage in caverns, abandoned natural gas wells, aquifers, or salt domes mentioned in Chapter 2, section 2.4 and Chapter 5, section 5.1 has much lower costs (capital cost of establishing the store at US\$3-20 kg-1, leading to a storage cost of one order of magnitude below that of liquefied H2 storage and two orders of magnitude below that of compressed storage) and constitutes a natural choice for centralised storage of hydrogen.

For decentralised stationary storage, metal hydride stores, or one of the similar concepts described in Chapter 2, section 2.4, may be more attractive than compressed gas flasks, but the cost is difficult to estimate because the final designs are not available (e.g., geometrical layout allowing sufficiently rapid extraction). Capital costs for metal hydride stores have been estimated as in the range of 2 000-80 000 2004-US\$ per kg of H2 capacity (Amos, 1998). Neither choice of hydride or estimating cost of storage has noticeably advanced since that estimate was made. The storage cycle costs are estimated at US\$0.4-25 kg-1 (Padró and Putsche, 1999). For automotive applications, the weight will be a problem except for some non-metal chemical or carbon storage types. When the technology is mature, the cost is expected to be a few times higher than that of compressed storage containers. General surveys of storage options may be found at Electricity Storage Association (2009, mainly batteries), Perez et al. (2010) or Sørensen (2010a, Chapter 5). Several options (from batteries to hydro) may be contemplated for storing renewable energy from wind or photovoltaics before or instead of conversion to hydrogen, and flywheels and supercapacitors have been suggested for short-term storage in vehicles (Doucette and McCulloch, 2011).

6.1.4 Infrastructure costs

The cost of hydrogen transmission by pipeline depends on the pipe diameter

and the hydrogen flow rate. By increasing the pressure difference through the pipeline, cost can be reduced more than the additional cost of compressors. Amos (1998) arrives at a cost of about US\$5 GJ-1 for an optimised flow rate of 10⁶ kg day-1 through a 160 km pipeline. Marin et al. (2010) list transmission cost-dependence on distance and flow-rate, unfortunately without specifying the kind of \$ used (Canada, US, …) or deflators (the references given for costs range from 1985 to 2009). The cost of liquid hydrogen transport by road is estimated as lower, but only when using current transportation fuel (diesel) costs, and in any case the additional cost of liquefying the hydrogen makes the total price unattractive except for the very longest transportation distances (such as intercontinental hydrogen transport by ship, in analogy to liquefied natural gas ocean carriers).

Converting filling stations for road vehicles to dispense compressed hydrogen may add some US\$0.1 kg-1 to the hydrogen price, but alternatively, the hydrogen production may take place at the filling station site using any of the methods available. The necessary filling station conversion to hydrogen cost is less than one year's maintenance costs for the current system (see review by Padró and Putsche, 1999; Campbell, 2004b, and a more recent European estimate by Köhler et al., 2010).

In-building generation of hydrogen (by electrolysis) and dispensing to garage-parked vehicles ("one-car filling station") are likely to double the price of hydrogen production and filling, respectively (Padró and Putsche, 1999). The reversible fuel cell units sized as traditional natural gas burners have not reached commercialisation, despite early efforts, and the saving by combining hydrogen and electricity production in one device thus remains a theoretical possibility.

In case hydrogen is produced from natural gas, there is a substantial environmental impact (Ozbilen et al., 2011) and it may be considered to recover CO2 (which is already separated in most current steam reforming plants) and store it away from the atmosphere−for example, in abandoned wells or as carbonate on ocean floors. The additional cost has been estimated at US\$0.05- 0.1 per kg of hydrogen (Padró and Putsche, 1999).

6.1.5 System costs

The cost of fuel cell systems is partly the cost of vehicles or building-based systems and in a wider context the total cost of a hydrogen economy with production, various types of usage and infrastructure such as storage and transmission, distribution, and filling outlets.

One reason that system costs may reveal things not possible to derive from the component costs is that each system component has an efficiency characteristic that often differs from that of the equivalent component (if there is one) in the current energy system.

Looking first at the cost of PEM fuel cell vehicles, the studies aimed at probing into future fuel cell costs have also addressed the total system cost of fuel cell, hydrogen storage, handling, batteries in case of hybrid vehicles, and power control, in order to arrive at a total price development of car manufacture. Figure 6.2 shows the result of an early Japanese study, with the assumption of a fuel cell cost declining to US\$40 kW-1 by 2020 (one of several scenarios) and a corresponding increase in the stock of fuel cell vehicles (passenger cars, lorries, buses, etc.) to 5 million by 2020 and 15 million by 2030. One might have expected the demand for fuel cell vehicles to rise as the price comes down and not only after it has come down, but the scenario is just an indication for three selected years and does not pretend to be dynamic. By 2020, the fuel cell vehicle cost (US\$15 788 in constant \$s) is nearly as low as that of current gasoline cars (US\$13 136 assumed in the study).

Figure 6.2. Scenario results for market introduction of fuel cell vehicles in Japan, giving the stock of fuel cell vehicles as a function of time, along with the decline in cost of passenger fuel cell cars (the stock comprises other types of fuel cell vehicles as well) (based on Tsuchiya et al., 2004).

Figure 6.3 shows the amounts of hydrogen required in the Japanese scenario, along with the associated cost, and Figure 6.4 shows the number of hydrogen filling stations required and the annual cost of constructing them.

The scenario assumptions made by Tsuchiya et al. (2004) comprise a slightly declining population in Japan, a modestly rising gross national product (GNP) and an unchanged demand for energy (achieved by the introduction of more efficient energy using equipment, not only in the transportation sector). The production of fuel cell vehicles rises from 50 000 in 2010 to 1.3 million in 2020 to 3.1 million per year in 2030, by which year the annual sales revenue is 59 × 10⁹ US\$ (production cost plus a 15% mark-up). The hydrogen activities constitute 1% of total Japanese GNP by the year 2030.

Figure 6.3. Japanese hydrogen demand and price delivered to the customer in the scenario of Fig. 6.2 (based on Tsuchiya et al., 2004).

The cost estimates of key hydrogen handling equipment could be used to expand cost scenarios such as the Japanese one for road transportation to other sectors; for example, the building-integrated use of fuel cells envisaged in the decentralised Danish scenario described in Chapter 5, section 5.5. In any case, such cost projections are highly uncertain, due to the wide range of future costs that may materialise for some of the most important components in a hydrogen-based energy system (see Fig. 6.1). The failure so far of automotive fuel cell solutions to reach the marketplace makes does not suggest any gain to be forthcoming from trying to update the early price and penetration scenarios with new optimistic guesses (Hajimiragha et al., 2011).

One criticism sometimes launched against a hydrogen infrastructure system is that the many conversion steps involved will make the overall efficiency of the system low (Bossel, 2004). From wind electricity through electrolysers to hydrogen, with transportation and storage losses, to finally regenerate electricity in buildings or fuel cell vehicles entails an overall efficiency of around 25%, which is seen as low compared with the current conversion efficiencies, typically between 13 and 30%, including refinery losses (10%) and either vehicle combustion engines (average efficiency about 15%) or power plants (average efficiency at or below 35%). I believe the criticism is largely ill placed, because the whole point is that the present fossil fuel system is drawing towards its end and has to be replaced by another energy system. Since renewable energy offers the only sustainable future energy system, the handling of variable primary energy streams is a must. Handling such fluctuations requires storage of energy, and hydrogen seems to be the cheapest way of accomplishing this. Clearly, converting back and forth involves additional losses, but that cannot be helped, and a fair comparison of hydrogen systems would be with other systems capable of handling the large fluctuations of solar and wind energy, not with the obsolete system on its way out. Such a fair comparison will most likely still designate hydrogen as the most viable energy storage medium, with underground hydrogen storage as very attractive compared with batteries, pumped hydro and other storage possibilities.

Figure 6.4. Japanese requirement for hydrogen filling stations and their cost (annual expenditure) for the scenario of Fig. 6.2 (based on Tsuchiya et al., 2004).

6.2 Life-cycle analysis of environmental and social impacts

The life-cycle assessment of hydrogen production is, in the cases where conventional fuels are converted into hydrogen, similar to many existing studies, in that major impacts are found to derive from air pollution and global warming issues (see Sørensen, 2011a). I shall first, in section 6.2.1, give the definition of life-cycle analysis and assessment used here and subsequently a number of examples of the application of life-cycle analysis to hydrogen systems. This also brings about a discussion of environmental impacts from each component of the hydrogen systems, as well as other impacts that have been identified.

Among the examples of life-cycle analysis given will be one of direct conversion of biomass to hydrogen, aimed to clarify the claim that such hydrogen production pathways will reduce impacts. Although many of the technologies remain emerging, making a complete description of the industrial processing is impossible at the moment, there will be important implications from the generic assessment that can be made at present. For fuel cell conversion of hydrogen, the industrial progress is such that meaningful lifecycle studies can start to be made on a fairly detailed basis. A concrete example of this will be given.

As an example of a life-cycle analysis of an entire system, the case of the passenger car important. This involves both the analysis of the car manufacture, including the specific additions to traditional cars necessary for fuel cell operation, and also the infrastructure impacts and contributions from the fuel provision and from the final disposal of the product.

The short summary of the life-cycle method preceding the actual analyses is necessary, because life-cycle analysis is not a standard calculation. In the literature, one finds various definitions ranging from restricted net energy analyses over environmental impact studies to the full consideration of both environmental and social impacts. It is the latter methodology, more fully described in Sørensen (2011a), which will be employed here.

6.2.1 Purpose and methodology of life-cycle analysis

Technology assessment is a natural tool to use for evaluating both individual products and also more comprehensive systems aimed at handling general sectors such as that of energy supply. The purpose of the assessment may be licensing or establishing health and environmental requirements for a product or a system. To the individual customers, technology assessment is a tool for choosing between alternative ways of satisfying a given need, and for decision-makers, it may serve as a tool for evaluating industrial policy and allocating public investments.

Side effects of a particular technology may show up during its use, but might as well be associated with its production or procurement or with the disposal or recycling of the technology after use. The technology may also, during any of its life phases, require inputs of materials and services from other sectors of society, and they may again have negative effects associated with either of the life-cycle phases. It is therefore necessary to perform what is termed an entire "life-cycle analysis" (LCA), including such imbedded impacts, in order to allow a fair comparison between different technologies, which may have their problem areas anywhere during the course of the life-cycle. It would certainly be wrong to compare energy technologies only on the basis of how they are produced and equally wrong to compare them only on their behaviour during use. For example, fossil fuels have their most severe impacts from emissions in case they are burned, whereas solar of wind energy have little impacts during the use phase. For instance, nuclear energy has impacts (associated with radioactive waste) long after its usage phase, so, quite generally, fair comparison must be based upon a complete life-cycle assessment, following the gathering of necessary information (inventory building), which constitutes the first step of a life-cycle analysis (Kuemmel et al., 1997; Sørensen, 2011a).

The assessment problem is further aggravated by the fact that there may be damage that neither manufacturer, society, or the customer is paying for. In such cases the damage will be left to deteriorate the collective value of society. Such impacts that escape the economic handling of technology may still be important for any comparative evaluation of alternatives. Costs not reflected in the marked prices associated with a technology are called "externalities". A more detailed discussion of the purpose and scope of life-cycle analysis, including ways to use an LCA, how to handle geographic distance and time displacement between occurrences of benefits and damage, and methodology for aggregation of data and for seeking common measures (such as monetary units) for presenting damage, may be found in Sørensen (2010a; 2011a). The purpose of the few remarks made here is to facilitate understanding the case studies presented following.

The actual work required for making a life-cycle analysis and assessment of a technology such as fuel cells may be summarised in the following way:

Make an inventory of substances or processes with potential harm, from production through the use phase to final decommissioning.
Perform an impact analysis including environmental effects, health effects and other less tangible social effects.
Conduct a damage assessment in terms of actual physical damage.
Make an assessment in comparative terms, weighing the different damage contributions on a common scale as far as possible.
Suggest alternative procedures for production, materials choice, conditions of use, and decommissioning that will reduce those identified impacts deemed critical, and possibly formulate norms and regulation, if it is not preferred to impose taxation or otherwise create economic incentives or disincentives in order to guide the use of technology of a particular society in the way indicated by the life-cycle assessment made.

A fairly comprehensive list of items to include in a life-cycle analysis mention the following, without claiming the list to be complete:

• Economic impacts, as seen from the point of view of the owner (private economy) or from society (public economy, including considerations of employment and balance of foreign payments).

Environmental impacts, such as land use, noise, visual impact, pollution of air, water, soil and biota, on a local, regional, or global scale, including climate change induced by emissions of greenhouse gases or ozone depleting substances.
Social impacts, including health impacts, accident risks, and effects on work environment and on satisfaction of human needs.
Security impacts, including terror actions, misuse, as well as supply security issues.
*Resilience*−that is, sensitivity to system failures, to planning uncertainty, and to future changes in value and impact assessments.
*Impact on development*−that is, furthering or countering the development goals of society (assuming that such exist, as they usually do at least for currently less developed nations).
Political impacts, including requirements for control, regulation and issues of centralisation of decision-making.

In order to analyse some (or all) of these impacts, including upstream and downstream components relative to the system or device of primary interest, it is as mentioned a good idea to establish an inventory of substances and processes of relevance and to assess their individual impacts, first in physical terms (emissions, etc.) and then in damage terms (injuries, disease, death, etc.). This list can in some cases be reused for assessing other technological items, provided that the inventory is not site- or time-specific. The outcome will be in different units, or in some cases non-quantifiable, and must be submitted to an assessment by decision-makers or by public debate. Only in some cases may it be meaningful to convert different impacts into common units (such as ∈ = euro or \$ = dollar or "environmental points"). This involves many difficult issues, as damage often occurs at other places or times than the benefits of using the technology. For instance, one would have to estimate the cost to society of accidentally losing one member (the "statistical value of a life"), which has spurred discussion on whether to use the insurance "cost" of a life in Western Europe to assess a life lost in a Colombian coal mine, rather than a Colombian value, perhaps corrected by use of purchase parity conversion. If the impacts are kept in different units, the assessment is said to be "multivariate".

6.2.2 Life-cycle analysis of hydrogen production

Conventional production by steam reforming

Production of hydrogen by steam reforming of natural gas gives rise to a number of impacts mediated by emissions to the atmosphere (Table 6.1), as well as impacts from the equipment production and disposal stream and from materials used (such as Ni catalysts) without complete recycling. The impacts caused by the emissions include global warming (from emission of CO2, CH4, N2O, etc.), waterway acidification (from SOx), and eutrophication (from N and P) and human respiratory diseases (from SOx, NOx, benzene and particles, as well as from soot and winter smog involving C formed by the side reaction given by Chapter 2, Eq. (2.5); Koroneos et al., 2004).

Impact category	Physical amount g/kWh of H2	Monetised value euro-c/kWh of H2	Uncertainty (range)
Environment:	Emissions:
Plant operation: CO2	320	12.1	(8−30)
SOx	0.29	0.17	high
NOx	0.38	0.23	high
CH4	4.4	2.0	(1−4)
C6H6	0.042	NQ
CO	0.18	NQ
N2O	0.0012	−
non-C6H6 hydrocarbons	0.79	NQ
particulates	0.06	0.04	high
Ni catalyst material	NA
Plant construction/decommissioning	NA
Occupational:	Number:
Industrial disease and accident	0.5 major injury/TWh	0.0004	low
Economic:
Direct economy (production costs)		3−6
Resource usage	serious in long run	NQ
Labour needs for manufacture	5 person-years/MW	NQ
Import fraction	NA
Benefits (value of product)		6−12
Other:
Supply security	low to fair	NQ
Robustness	medium	NQ
Geopolitical	competition	NQ

Table 6.1. Life-cycle impacts from H2 production by steam reforming of natural gas. Mortality from stack emissions of (SO2, NOx) aerosols and particulate matter is taken as 2 × 10-9 per g, and morbidity is taken as 935 × 10-6 workdays lost per g, corresponding to European population densities. Global warming cost is 0.38 euro/kg CO2-eq. NA/NQ: not analysed/not quantified (the table is constructed with use of data from Spath and Mann, 2001; Sørensen, 2004a).

It is seen from Table 6.1 that the externality cost of hydrogen produced on the basis of natural gas is very high, due to the global warming caused by emission of greenhouse gases. This is also true if the lower estimates of global warming impacts are used. The issues related to these figures are discussed in Sørensen (2010a; 2011a). The main question is related to the fact that negative consequences of warming mainly occur in the less developed countries near the Equator, and one has to decide whether to value lives lost in these regions at the same level as lives lost in Europe or the United States, as it is done in the central estimate, based on insurance payments for premature deaths in Europe.

As stated in section 6.1.4, the cost of removing CO2 from the steam reforming plant appears to be much lower than that of the global warming externality associated with not handling the CO2 issue.

Production by electrolysis

If hydrogen production is based on surplus renewable energy such as wind power (as in the scenario in Chapter 5, section 5.5), the impacts are dramatically reduced. Only the occupational impacts are larger, at least at present, due to their roughly scaling with the cost of the conversion equipment, whether conventional electrolysers or reversely operated fuel cells.

Direct bio-production of hydrogen from cyanobacteria or algae

The potential advantage of bio-hydrogen is to be able to use a production device or system, where major components are already part of the natural solar radiation disposition system operating on our planet to produce biomass. The same philosophy is behind agriculture, where a low conversion efficiency is accepted in return for obtaining a free "production system" requiring as its only input some seeds and some work (such as weeding and harvesting), and based on these inputs the system produces food and related biomass products.

However, both for agriculture and for bio-hydrogen production, it is basically wrong to consider plant or bacterial material "taken" from nature as free. There may be substantial life-cycle impacts and associated costs involved. The area used for plant growth or ocean farming is affected by the formation of large monocultures, and the ecological implications of residues left over, of water usage, and so forth, may be significant. It is therefore necessary to go through all the steps involved in bio-hydrogen production, including the farming itself, collection of crops, treatment made at any stage of the cultivation, residue handling and waste disposition, in order to make an inventory of impacts and the subsequent assessment in monetary or other terms. One should distinguish between systems where biomass is harvested and later fed into a hydrogen-producing device, and those where hydrogen is directly produced under the influence of sunlight.

For the latter type of system, a bio-reactor construction must cover the exposed area (presumably of water, since land use would entail the higher land costs) from which solar radiation is to be collected, in order to form a closed construction capable of keeping the hydrogen produced from escaping before it can be piped to the shore. The cost can therefore be expected to be similar to that of thermal solar collectors. It follows from the discussion in Chapter 2 that the efficiency (defined as the energy value of the hydrogen produced divided by the incoming solar radiation on the reactor surface) will be considerably below 1%. Table 6.2 assumes for illustration an efficiency of 0.1% and estimates cost on the basis of the similarity to solar collectors.

Impact category	Physical impacts	Monetised value euro-c/kWh of H2	Uncertainty, assumptions
Environment:
Plant construction/decommissioning	NA
Land or ocean use	large	NQ
Use of genetic engineering	problematic	NQ
Hydrogen cleaning	NA
Occupational:
Industrial disease and accident	NA
Economic:
Direct economy (production costs)		> 40	efficiency 0.1%
Resource usage	area covered	NQ
Labour needs for manufacture	5 person-years/MW	NQ
Import fraction	NA
Benefits (value of product)		6−12
Other:
Supply security	good	NQ
Robustness	medium	NQ
Geopolitical	positive	NQ

Table 6.2. Life-cycle impacts from photo-induced H2 production using cyanobacteria, as yet poorly quantified. See caption to Table 6.1 (see Sørensen, 2004c).

The implication of these estimates is that the productivity of diverting hydrogen from (possibly genetically modified) cyanobacteria or algae, which is only a fraction of the total energy turnover of the organism, must be substantially lower than the biomass productivity in agriculture, which on average is about 0.2% (a maximum for special plants being 4%; see Sørensen, 2010a). Even if the techniques suggested in Chapter 2 could be realised, it would therefore seem that the prospects for viable hydrogen production by direct photosynthesis are meagre.

Impacts from use of genetically engineered organisms

Although the form of genetic engineering has changed, the concept itself has been employed from the earliest periods of systematic crop farming, some 10⁴ years ago. Over the course of time, cereal species were improved to give higher yields, by sorting and crossbreeding of species or sub-species. Only recently, a few hundred years ago, did a theoretical understanding of the mechanisms underlying the processes of genetic modification start to emerge, and the process remained a very slow way of improving crops until some 50 years ago, when more systematic genetic experiments began to be performed with mutations induced, for example, by nuclear radiation. Today, the sequencing of a number of genes with identification of groups of code responsible for particular properties of the plant or organism has widely increased the options for modifying biological systems, such as changing a CO2 or nitrogen fixing organism to become a hydrogen-producing organism, by suppression of the unwanted pathways of energy transfer. Still, the present knowledge of genetic functionality is very limited, and genetic engineering is not based on precise knowledge of what the smallest individual coding sequences do exactly, but rather on identifying the gross responsibilities of more substantial chunks of code. For this reason, genetic engineering is still to a considerable extent a game of trial and error, requiring monitoring and tests of the behaviour of the manipulated organisms, often during extended periods of time and with possible surprises in the form of unexpected side effects occurring only after delays and in particular circumstances.

The impacts of genetic manipulations could have been large, even during the Stone Age. We have no precise records to support this view, but it is likely that agricultural species obtained by cross-breeding would sometimes turn out not only to improve yields or quality of the parts of the plant used for food, but also to make the plant more susceptible to infections or predator organisms, in which case the result could have been famine and an associated decimation of populations. Exactly the same problems could arise with present genetically modified species, which are often employed on a large scale after testing that must necessarily be incomplete and restricted to periods of time and diversity of environmental factors insufficient to rule out a number of adverse long-time effects. The increase in the industrial use of genetic manipulation techniques has rather had the effect that screening, especially for long-term effects, has become more superficial and less systematic. As it is required by current legislation, testing is often left to the industries themselves, as public control institutions with often limited budgets can only overcome a limited amount of independent verification (sometimes only by spot tests). This is like the situation for use of chemicals in industrial products, where the rate of new chemicals introduced is much too high to provide public overseers with the time needed to perform long-time screening.

A profound moral (as well as practical) issue is the juxtaposition of benefits and risks associated with the introduction of genetically engineered crops and products. We have recently seen crops genetically engineered to accept the use of pesticides not compatible with previous semi-wild species, marketed by the same company that sells the pesticide in question. Would we accept crops engineered to accept the pesticide of one company, but made incompatible with pesticides of competing companies? Should we accept at all genetic engineering aimed at improving tolerance to pesticides that due to their possible wider environmental impacts would be better banned in favour of ecological methods of cultivation? Clearly, the questions raised make it very difficult to calculate any precise quantitative value for the externality associated with genetically engineered crop modification, whether for food or for hydrogen production.

Impact category	Physical impacts	Monetised value euro-c/kWh of H2	Uncertainty, assumptions
Environment:
Plant construction/decommissioning	NA
Agricultural production	substantial	3	high
Land or ocean use	large	NQ
Transportation of feedstock	emissions, accidents	40	high
Bio-reactor operation	NA
Hydrogen cleaning	NA
Occupational:
Industrial disease and accident	NA
Economic:
Direct economy (production costs)		> 10	efficiency 30%
Resource usage	agricultural area	NQ
Labour for manufacture and operation	10 person-years/MW	NQ
Import fraction	low
Benefits (value of product)		6−12
Other:
Supply security	good	NQ
Geopolitical	positive	NQ

Table 6.3. Life-cycle impacts from H2 production by fermentation of agricultural waste, as yet poorly quantified. Transportation impacts include greenhouse gas warming, air pollution and traffic accidents, assuming current vehicles such as diesel trucks. See caption to Table 6.1 (with use of Sørensen, 2004c).

Hydrogen from fermentation of biomass

The alternative route from photosynthesis to hydrogen mentioned in Chapter 2 separates the biomass production from the hydrogen extraction, proposing fermentation for the latter, in analogy to biogas (methane plus CO2) production.

This technology has the potential for producing hydrogen at social costs that are acceptable. However, Table 6.3 indicates that a likely condition will be the use of less polluting vehicles for the substantial requirement for transport of biomass feedstock, which is here considered to be by road using the current variety of trucks. The transportation externality estimates in Table 6.3 are derived from impacts from driving passenger cars, and the agricultural externalities are from a similar study of biogas production (Sørensen, 2004a).

The sparsely populated Tables 6.2 and 6.3 should be regarded as indications of work to be done. A number of studies have identified environmental impacts for a range of hydrogen production pathways, ready for incorporation in life-cycle analyses (GM, 2001; Wang, 2002; Wurster, 2003; EC, 2004; El-Sharkh et al., 2010).

6.2.3 Life-cycle analysis of fuel cells

The further away from commercialisation a technology is, the more difficult it is to perform an accurate life-cycle analysis (LCA). On the other hand, precisely at an early stage, the LCA may be most useful. It may identify traits of a technology discouraging further development, and it may point to the most proper path for the development of a given technology, allowing important environmentally or socially relevant choices to be made at an early time during the development process. In many cases a broad-brush LCA may identify the most important externality cost items in the design and may allow them to be optimised or replaced by alternatives at a time when it is still possible to make such changes. Had LCA assessments been applied to the early-stage electric and fossil fuel-based vehicles around 1900, the technology choice might not have been the internal combustion engine. Thus, no excuse will be made for attempting to carry out life-cycle analysis now, but of course the outcome must be viewed with appropriate caution.

SOFCs and MCFCs

For the solid oxide fuel cells (SOFCs), a number of environmentally critical items have been identified (Zapp, 1996). The carrier sheet electrolyte may be produced from yttrium-stabilised zirconium oxide with added electrodes made of, for example, LaSrMn-perovskite and NiO-cermet. Nitrates of these substances are used in manufacturing, and metal contamination of wastewater is a concern. The high temperature of operation makes the assembly very difficult to disassemble for decommissioning, and no process for recovering yttrium from the YSZ electrolyte material is currently known.

For molten carbonate fuel cells (MCFCs), a full life-cycle analysis has been attempted (Lunghi and Bove, 2003). Both electrodes and the electrolyte matrix are manufactured by mixing powdered constituents with binders and solvents to form a sheet after casting and drying. Some constituents of the slurry preparation have been omitted from the LCA due to industrial secrecy. The results of the analysis comprise resource usage and emissions to air, wastewater, and soil. A critical resource may be Ni, for which the largest supplier is Cuba. The main life-cycle impacts for the areas mentioned are summarised in physical terms in Table 6.4.

It is seen from Table 6.4 that the largest life-cycle impacts come from the negative electrode manufacture. The attempt made in Table 6.5 to monetise these impacts shows a global warming impact that is dominated by the emission of methane, and health impacts that are dominated by the very large emissions of sulphur oxides indicated in Table 6.4. Presumably these values, derived from an Italian pilot production, would be substantially reduced in case of a realistic commercial production of MCFCs.

Life-cycle impact	Negative electrode	Positive electrode	Electrolyte matrix	Bipolar plate	Total	Unit
Electric energy input	153.2	82.6	73.03	5.47	314	kWh
CO2	508	214	127	8.03	857	kg
CH4	423	131	36.2	0.502	591	g
N2O	12.2	3.8	3.5	0.0014	19.4	g
SO2	10.9	6.67	1.5	0.26	19.4	kg
SOx (as SO2)	2.01	0.61	0.08	–	2.7	kg
CO	121	45.4	25.2	37.6	229	g
NO2	366	224	–	14.3	604	g
NOx (as NO2)	697	214	27.8	–	939	g
Non methane VOC	420	129	16.8	0.03	566	g
VOC	–	–	–	15.5	15.5	mg
Benzene	0.895	0.31	0.031	0.01	1.3	g

Table 6.4. Contributions to life-cycle impacts from different components of a 1 m2 unit single molten carbonate fuel cell (based on data from Lunghi and Bove, 2003).

Impact category	Physical amount	Monetised value euro-c/m2 of cell	Uncertainty (range)
Environment:	Emissions:
Manufacture (global warming)	Table 6.4	301	(200−400)
Manufacture (health effects)	Table 6.4	13 467	high
Operation	NA
Occupational:
Health effects (included above)	NA
Economic:
Direct economy (production costs)		NQ
Resource usage	Ni availability	NQ
Labour needs for manufacture	5 person-years/MW	NQ
Benefits (value of product)		6−12

Table 6.5. Monetised life-cycle impacts corresponding to the physical impacts given in Table 6.4, using the same translation as in Table 6.1 The impacts are given for a 1 m² cell, as the lifetime power production of the cell is not available.

PEM fuel cells

Figure 6.5 shows the main steps in the manufacture of a proton exchange membrane fuel cell (PEMFC) stack for automotive purposes. Several components in the fuel cell assembly are not significantly different from components found in other industrial products (metals, carbon as fibre or graphite, plastics), and their life-cycle impacts may be taken from generic studies. However, there are exceptions. First, the use of a polymer membrane in each cell unit, based on perfluorinated ionomers (Barbi et al., 2003) of hydrocarbons (Kreuer, 2003), or of organic material (Evans et al., 2003), may cause special concerns regarding decommissioning or recycling (Handley et al., 2002). Recycling can be difficult, and incineration is recommended, although in some cases care should be taken to separate difficult materials such as Pd used with organic membranes before burning. Second, while the steel and carbon contents of the bipolar plates can easily be reused or in the case of carbon incinerated, there are again small amounts of special materials requiring care. The most important among these is Pt or Pt-compounds used as a catalyst at the electrodes. Pt causes strongly negative impacts due to the emissions during extraction and purifying, particularly in Third World plants with modest environmental experience, such as seen in the case study of South African platinum performed by Pehnt (2001).

Figure 6.5. Life-cycle flow path for the industrial manufacture of PEM fuel cell stacks (Sørensen, 2004d).

The effect of these impacts can be greatly reduced if Pt is recovered and reused, as all heavy metals should be. A strategy for supply of Pt to the fuel cell industry is discussed by Jaffray and Hards (2003). In terms of weight, the stack breakdown on components is shown in Fig. 6.6 for two conventional material choices for the bipolar plates: graphite or aluminium. Recently, bipolar plates made of conducting polymers have been developed (Middleman et al., 2003), with thickness and weight reduction as a consequence.

If the PEM fuel cell is requiring a reformer, there will be additional impacts, depending on the fuel and equipment used. The same is true for direct methanol fuel cells, with a dependence on the way methanol is produced, based on natural gas or bioenergy.

The short lifetime of present PEM fuel cells will in general terms increase life-cycle impacts relative to technologies with a longer operational life. Current cells degrade significantly after less than 2000 h of operation (Ahn et al., 2002; Myers et al., 2009), and even with the target 5-year life, replacement will be much more frequent than for most other energy technologies.

Figure 6.6. Distribution of material use for stacks with aluminium or graphite bipolar plates (Mepsted and Moore, 2003; Sørensen, 2004d).

6.2.4 Life-cycle comparison of conventional passenger car and passenger car with fuel cells

The passenger cars selected for the LCA study presented here are characterised by the features listed in Table 6.6. The DaimlerChrysler f-cell was the first fuel cell passenger car to enter the stage of limited series production (estimated at 60-80 units) demonstrated first in Japan (Tokyo Gas Co., 2003) and subsequently in Europe and North America. It follows the Citaro F fuel cell bus entering a similar phase in 2003 (demonstration in Europe of a small series of about 30 units). The f-cell car is based on a slightly longer version of the commercial A2 series of Mercedes-Benz gasoline and diesel fuel cars, and Table 6.6 reflects the limited data available at the time when the calculation was made. The two non-fuel cell cars studied for comparison are a Toyota Camry gasoline/Otto engine car used as a typical US year-2000 vehicle in a previous life-cycle study (Weiss et al., 2000, 2003), and the Lupo 3L TDI Diesel car topping the European list for mixed driving efficiency (VW, 2002, 2003). The car is no longer in production, but still holds the record for fuel efficiency of commercial passenger vehicles. After some year without a fuelefficient car on their programme, the VW in 2011 again offers a vehicle almost as efficient as the Lupo (Polo Blue Motion). Table 6.6 gives a gross material usage survey, as well as the weight and fuel consumption details to be used in the life-cycle analysis. These are also summarised in Figs. 6.7 and 6.8.

Figure 6.7. Comparison of mass distribution for the three passenger cars included in the LCA analysis (Sørensen, 2004d).

Figure 6.7 shows that while the Lupo has diminished weight as compared to an average car through use of lightweight materials where possible (but still being in the top safety category according to crash tests), the f-cell car, although small of appearance, has a higher mass than even the conventional car, due to the heavy equipment associated with hydrogen management and conversion. Figure 6.8 compares the efficiencies of the three cars studied. In terms of energy content, the fuel use of the f-cell car is slightly below that of the Lupo, both being considerably below the current average car. The fuel-towheel efficiency is seen to improve considerably for the fuel cell vehicle, over the efficient diesel car and of course over the conventional gasoline car. This, however, is due to using theoretical simulation to calculate the fuel usage. The actual f-cell cars did not achieve this goal. The simulations make use of the new European driving cycle used for official rating of cars in Europe, the shape of which is shown in Fig. 6.9.

Passenger car (1-5 persons plus luggage) average USA, 2000 best Europe, 2000 Description	Otto engine Toyota Camry	VW Lupo 3L	35 MPa H2 fuel common-rail Diesel PEMFC/elec. motor DaimlerChrysler f-cell Unit		Reference
Bare mass (body, chassis)	930	570	800	kg	3,4,7,est.
Propulsion system mass	340	220	600	kg	3,est.
Battery mass	12	10	40	kg	3,4, est.
Fuel and container/handling mass	<40	<35	3+100	kg	3,4, est.
Proper mass (unloaded)	1300	825	1589	kg	4,5,6
Mass of steel		410		kg	4
Mass of plastics, rubber		130		kg	4
Mass of light metals		130		kg	4
Load mass	<350	<340	<340	kg	3,4
Total mass (occupancy: 2, 0.67 full tank)	1440	980	1725	kg	3,4
Coefficient of rolling resistance	0.009	0.0068	0.0068		3,est.
Drag coefficient	0.33	0.25	0.25		3,4
Auxiliary power	0.7	0.6	1	kW	3,est.
Engine/fuel cell rating	109	45	69	kW	3,5,6
Electric motor rating			65	kW	5,6
Battery rating		4/732	20/1400	kW/Wh 5,6
Reformer efficiency (not applicable)
Engine/fuel cell efficiency*	0.38	0.52	0.68		3,7,calc.
Gear and transmission efficiency*	0.75	0.87	0.93		3,est.
Electric motor efficiency			0.8		3
Fuel use*	2.73	1.08	0.8-1.44¤		MJ/km 3,7,calc.
Fuel use*	12	33		km/l	3,4,5
Fuel to wheel efficiency*	0.15	0.27	0.36		3,calc.

Table 6.6. Basic vehicle data used (Sørensen, 2004d).

Environmental impact analysis

Table 6.7 gives the environmental LCA data available for the selected cars, in terms of energy used and emissions occurring during the phases of the vehicle life-cycle, based on the studies mentioned with addition of my own calculations and estimates. The impacts are given in physical units as required for the LCA inventory and will subsequently be translated into concrete impacts on health and environment, including global warming effects. Of particular interest are the impacts from the manufacture and use of the fuel cell component in the f-cell car, discussed in section 6.2.3.

^* For standard mixed driving cycle. Fuel-to-wheel efficiency is the work performed by the car to overcome air and road friction, plus the net work performed against gravity and for acceleration/deceleration, all divided by the fuel input (note that this efficiency concept varies linearly with the combined drag and rolling resistance).

^{¤ 1.44} MJ/km for the first f-cell cars manufactured, with hope of lowering (DC, 2004). Key to references: 3: Weiss et al., 2000; 4: VW, 2002; 5: VW, 2003; 6: Tokyo Gas Co., 2003; 7: Weiss et al., 2003.

Figure 6.8. Comparison of contributions to and total efficiency of the three cars studied in the LCA (Sørensen, 2004d).

Figure 6.9. Driving cycle used for emission certification of passenger cars in Europe, with an urban stop-go section and at the end a continuous sequence at speeds up to 120 km/h (EC, 2001).

Passenger car (1-5 persons & luggage) LCA environmental impacts Life-cycle emissions	Otto engine Toyota Camry	Average USA, 2000 Best Europe, 2000 common-rail Diesel VW Lupo 3L	H2 from natural gas PEMFC/elec. motor	H2 from wind surplus PEMFC/elec. motor DaimlerChrysler f-cell DaimlerChrysler f-cell Unit Ref.
Car manufacture Energy use Greenhouse gas emissions SO2 emissions CO emissions NOx emissions Non-methane volatile organic compounds Particulate matter emissions Benzene Benz(a)pyrine	87 1.7	Production+materials 37+51 0.5+0.5 1.6+10.0 1.8+4.6 2.0+1.3 0.3+4.0	Total/FC stack 93/?, 178/80 § 1.7/?, 2.8/1.4 § 36/14.5 § ?/1.7 ?/14.5 ?/1.7 ?/2.6 ?/2.3 ?/0.034	93/?, 178/80 1.7/?, 2.8/1.4 § 36/14.5 § ?/1.7 ?/14.5 ?/1.7 ?/2.6 ?/2.3 ?/0.034	GJ kg kg kg kg kg g g	7,4,9 tCeq 7,4,9 4,8 8 4,8 4,8 4,8 8 8
Fuel production (for 300 000 km) Energy use Greenhouse gas emissions SO2 NOx Non-methane volatile organic compounds Particulate matter Pd (if reformer is used)	156 3.6	67 0.4 9 40 60 1	185 8.6	185 0 0 0 0 0	GJ tCeq 4,7 kg kg kg kg	4,7 4 4 4 4
Lifetime operation (15 y, 300 000 km) ¤ Energy use Greenhouse gas emissions SO2 CO NOx Non-methane volatile organic compounds Particulate matter PAH N2O: effect on stratospheric ozone	819 16.1 13	Incl. decomm. est.: 324 6.5 1.6 30 75 2.7 6 1.5 1	240 0 0 0 0 0 0 0 ~0	240 0 0 0 0 0 0 0 0	GJ tCeq 3,4 kg kg kg kg kg kg kg	3,4 4 4 4 4 4 4 4,calc.
Decommissioning (not estimated separately) Totals Energy use Greenhouse gas emissions SO2 CO NOx Non-methane volatile organic compounds Particulate matter	1062 21.4 61 70 12	479 7.9 22.2 30 121 66 11.3	603 11.4 36	603 2.8 36	GJ kg kg kg kg kg	3,4,7,9 tCeq 3,4,7,9 4,8 4,8 4,8 4,8 4,8

Table 6.7. Environmental life-cycle impacts (Sørensen, 2004d).

§ Pt manufacturing (assumed in South Africa) accounts for 30% of energy, 40% of greenhouse gases and 67% of acidification, with no recycling assumed (Ref. 8).
¤ Maintenance impacts not estimated.

Key to references: 3: Weiss et al., 2000; 4: VW, 2002; 5: VW, 2003; 7: Weiss et al., 2003; 8: Pehnt, 2001; 9: Pehnt, 2003.

The fuel cell vehicle considered in this study uses hydrogen directly. If reformation of methanol or gasoline were used, additional impacts would be derived from the reformer, including often quite large impacts from a catalyst such as Pd, which again should be recycled as fully as possible.

No separate data have been found for decommissioning impacts, although VW (2002) claims to have included them under "lifetime operation". In Denmark, cars delivered to a recycling station pay a fee of about 500 euro, assumed to cover the decommissioning costs minus income from selling extracted parts for reuse. European regulation is discussed, where decommissioning is part of the initial purchase price and the manufacturer is obliged to optimise assembly for decommissioning and to take the vehicle back at the end of service for maximum recycling.

The Volkswagen report (VW, 2002) is a detailed and site-specific LCA for the manufacturing plant at Wolfsburg, including materials and water delivered to or coming out of the plant. It is centred on the Golf cars, but the scaling for application to Lupo made here has in gross terms already been made in the VW environmental report (2002). The environmental impacts of Table 6.7 are summarised in Fig. 6.10.

Passenger car (1-5 persons & luggage)# LCA social impacts and other environmental impacts	Otto engine Toyota Camry	Average USA, 2000 Best Europe, 2000 VW Lupo 3L	H2 from natural gas common-rail Diesel PEMFC/elec. motor	H2 from wind surplus PEMFC/elec. motor DaimlerChrysler f-cell DaimlerChrysler f-cell	Unit Ref.
Car manufacture/decommissioning Job creation Occupational risk: death Occupational risk: severe injury Occupational risk: minor injury	0.3 0.0001 0.003 0.015	0.3 0.0001 0.002 0.013	1.8 0.0005 0.015 0.08	1.8 person-year 1 0.0005 0.015 0.08	1,12 1,12 1,12
Maintenance Job creation Occupat. risks (death/major/minor injury)	0.3	0.3 0.0001/0.003/0.015 0.0001/0.002/0.013			1 1,12
Driving Accidents (death/severe injury)£ Stress/inconveniences Mobility	0.005/0.050 some advanced	0.005/0.050 some advanced	0.005/0.050 some advanced	0.005/0.050 some advanced	1 1
Time use as social factor (different perception by individuals) Noise (economic quantification in Table 6.10) some Visual impacts (of cars in environment; different perception by individuals)		some	less	less	1

Impacts from road infrastructure (road construction, maintenance, visual impacts: estimated in monetary terms in Table 6.10) Impacts from car infrastructure (service, repair, traffic police & courts, insurance: mostly included in cost given in Table 6.9)

Table 6.8. Social life-cycle impacts (Sørensen, 2004d).

All figures for service life of 15 years, 300 000 km.

£ Statistics for Denmark has been used.

Key to references: 1: Kuemmel et al., 1997; 12: European Commission (1995).

Social and economic impact analysis

Table 6.8 gives occupational risks during the life-cycles of the vehicles, based on standard industrial data (i.e., the impacts are proportional to cost). The job content is based on statistics from the energy sector in Denmark (Kuemmel et al., 1997). The rate of accidents on the road is taken from several Danish studies and is considerably higher in some parts of the world. Evaluation of the health and injury impacts are again based on several Danish studies (references in Kuemmel et al., 1997), as are the less tangible visual and noise impacts (estimated by hedonic pricing) and the inconveniences such as children having to be supervised when near public roads or pedestrians in general having to use round-about paths to get to street crossings with traffic lights, where also the waiting time is valued.

Figure 6.10. Comparison of environmental impacts from the three passenger vehicles considered in the life-cycle analysis, with hydrogen for the fuel cell car being derived from either natural gas or excess wind power (Sørensen, 2004d).

Cars need roads for driving, and the road infrastructure is thus an "externality" to vehicle LCA, which has to be evaluated along with the car operation infrastructure. This is done in monetary terms based on Kuemmel et al. (1997) and is included in Tables 6.9 and 6.10. Table 6.9 gives the direct costs involved (and for comparison the cost of public transportation), without including any of the substantial taxes and/or subsidies characterising the actual consumer costs in many countries. The f-cell car never became a commercial vehicle, so its market price is not available. Instead, the price of the corresponding Mercedes-Benz (smallest car) is used, after augmenting it by an assumed fuel cell stack price taken as 100 euro/kW, and the other hydrogen handling and storage costs are then assumed to be similar to that of the stack. Finally, a factor of two is applied in order to account for the small series of production that would prevail for some period of time. This distribution of this assumed price on components is similar to the one estimated for the Citaro F fuel cell bus (based on published Evobus EC project material).

Passenger car (1-5 persons & luggage) LCA economic impacts Life expectancy: 15 years, 300 000 km	Otto engine Toyota Camry	Average USA, 2000 Best Europe, 2000 H2 from natural gas VW Lupo 3L	common-rail Diesel PEMFC/elec. motor	H2 from wind surplus PEMFC/elec. motor DaimlerChrysler f-cell DaimlerChrysler f-cell Unit Ref.
Direct economy
Car (est. cost without taxes/subsidies)	15 000	13 000	80 000¤	80 000¤		euro est.
Roads (monetary evaluation in Table 6.10)
Fuel cost (at filling station*, no tax)	15 000	5 455	15 600	15 600		euro est.
Service and maintenance	15 000	13 000	80 000	80 000		euro est.
Decommissioning (becoming included in purchase price)
Time use (of time that could have been paid, depends on individual case)
Reference cost of satisfying mobility needs 35 000		35 000	35 000	35 000	euro §
Resource use See materials in Table 6.6, recycling will modify these)
Balance of labour and trade Job intensity (near 50% local, even if no local car or fuel production)

Table 6.9. Economic life-cycle impacts (Sørensen, 2004d).

* Oil price staying at current level, hydrogen price dropping linearly from 100 to 30 euro/GJ (projected for 50 000 vehicle penetration; Jeong and Oh, 2002 ) during the 15-year period, initial cost of hydrogen filling stations not included.
¤ Small series cost is reflected; the current 85 kW PEMFC stack cost is about 10000 euro (with ~2500 euro projected for 2025) (Sørensen, 1998; Tsuchiya and Kobayashi, 2004).
§ Public transportation estimated cost.

Import and export fractions (varying between countries)

Passenger car (1-5 persons plus luggage) Life-cycle assessment externality monetising exercise	Otto engine Toyota Camry	VW Lupo 3L	Average USA, 2000 Best Europe, 2000 H2 from natural gas common-rail Diesel PEMFC/elec. motor	H2 from wind surplus PEMFC/elec. motor DaimlerChrysler f-cell DaimlerChrysler f-cell Unit Ref.
Vehicle-related environmental emissions (based on Table 6.7)
Human health impacts	38 100	14 000-40 000#	22 500§	22 500§	euro 1
Global climate impacts*	32 100	12 000	14 700	4 200	euro 1
Quantified social impacts (based on Tables 6.8, 6.9)
Occupational health risks	648	632	3 241	3 241	euro 1
Traffic accidents, incl. rescue & hospital costs 31 200¤		31 200¤	31 200¤	31 200¤	euro 1
Traffic noise	9 000	9 000	5 000	5 000		euro 1,est.
Road infrastructure (envir. & visual impacts) 28 000		28 000	28 000	28 000	euro 1
Inconvenience (to children, pedestrians etc.) 30 000		30 000	30 000	30 000	euro 1

Table 6.10. Overall externality assessment (Sørensen, 2004d).

* Mainly caused by tropical diseases and with accidental deaths valued by European standards (3 Meuro/death), cf. discussion in Kuemmel et al. (1997).
# The upper estimate is due to possible increased impacts associated with NOx compared to earlier valuations (may be reduced by NOx exhaust cleaning).
§ Can be reduced by recycling of Pt (Pehnt, 2001).
¤ About half of this number is from the 3 Meuro valuation of accidental death.

Key to reference: 1: Kuemmel et al., 1997.

Maintenance costs are taken as a fixed fraction of capital cost and thus are large for the f-cell car (hardly unrealistic for a new construction). The hydrogen cost is that of production from natural gas, ramped down as a function of time. It does not include the initial high cost of establishing hydrogen filling stations. No separate estimate is made for the cost of producing hydrogen from wind, discussed in Sørensen et al. (2004). The fuel price for gasoline and diesel fuel has been taken at the current level, disregarding possible increases during the period of operating the vehicles. The social life-cycle impacts are summarised in Fig. 6.11.

Figure 6.11. Comparison of social impacts from the three passenger vehicles considered in the life-cycle analysis (Sørensen, 2004d).

Overall assessment

The total externality costs (i.e., those not reflected in direct consumer costs) are summarised in Table 6.10. This involves translating the impacts from physical units to common monetary units, with the problems inherent in such an approach, notably valuing the loss of a human life to society. The caveats are associated with the fact that impacts such as accidental deaths are not always occurring in the same society that harvests the benefits of car driving. These issues have been discussed, for example, in Sørensen (2010a; 2011a). All monetised impacts are summarised in Fig. 6.12 for the three vehicles considered. A very large fraction of the impacts is derived from road infrastructure, traffic accidents and annoyance. These are identical for all vehicles, except for noise that is smaller for hydrogen vehicles. The other large contribution is from emissions of pollutants to the air. They are in part from manufacture and maintenance and, in the case of the gasoline and diesel cars, from emissions in breathing height, despite attempts of exhaust cleaning (much less efficient than for central power plants). This component is larger for the average car than for the Lupo 3L, as is the fuel cost. Regarding greenhouse gas emissions, the f-cell car using hydrogen from natural gas is no better than the Lupo car, but with hydrogen from renewable energy sources the advantage is substantial.

Figure 6.12. Summary of monetised life-cycle impacts from Tables 6.9 and 6.10 (Sørensen, 2004d).

Concern over particulate air emissions involving small-diameter particles has made some countries prefer gasoline cars over diesel cars, except for trucks and buses where the higher efficiency has been the overriding factor. The mechanisms involved in the dispersion of such particles have been the subject of intense study (see, e.g., Kryukov et al., 2004). The Lupo diesel car considered above has reduced the particulate emissions (Table 6.7) to levels comparable to those of gasoline cars, but all newer European diesel cars, including both efficient passenger cars, buses and trucks, have by 2010 regulation in the European Union been required to use electrostatic filters reducing the particle emissions by over 90%, which is better than the SO2 removal by the small catalyst devices used in gasoline cars (but in both cases not as good as the exhaust cleaning at large, stationary power plants).

For fuel cell cars carrying methanol and using an on-board reformer, there is direct emission of greenhouse gases, as well as additional impacts from the manufacture of fuel and reformer, leading to an overall CO2-equivalent contribution some 10% higher than for a corresponding car with a pure hydrogen fuel stream (Patyk and Höpfner, 1999; Pehnt, 2002; MacLean and Lave, 2003; Ogden et al., 2004).

The plug-in and autonomous hybrid vehicles considered to offer the best opportunities for market acceptance (see Chapter 4 and 7) have life-cycle impacts deriving both from the fuel cell components and from the batteries, in addition to general system impacts common for all road vehicles. Life-cycle assessments for batteries indicate that impacts from Li-ion and NaNiCl batteries are smaller than those of lead-acid and Ni-metal batteries (Bossche et al., 2006). For a specific manufacturing process of new types of Li-ion batteries, Zackrisson et al. (2010) find greenhouse gas emissions from production to be at least as large as excess emissions during battery use in a plug-in gasoline/battery hybrid vehicle (not calculating impacts from the battery recycling assumed in effect). Only the extra emissions caused by having to carry the weight of the batteries, plus leakage, are included, and a few impacts other than global warming are only given in relative terms (a half-hearted LCA method strongly criticized, e.g., in Sørensen, 2011a).

Key factors are the sources of energy inputs and using, for example, the current electricity mix of a given country can give very misleading results when applied to a future situation where, for example, coal fired power has been replaced by wind power. Figure 6.13 illustrates this, giving the direct costs plus a low, medium, and high estimation of warming and pollution social costs, for conventional and pure fuel cell cars operating in Korea (and with optimistic cost figures assigned to a year-2015 scenario). Inventories for climate and pollution studies for current hydrogen and hybrid vehicles have been collected by Colella et al., (2005), and comparison of various battery and fuel cell cars including hydrogen fuel cell/battery hybrids are made by Offer et al. (2010), finding positive life-cycle impacts for the hybrids, by US DoE (2010b), and by Thomas (2009), republished by Veziroglu and Macario (2011).

The assessment of Thomas (2009), summarised in Fig. 6.14, is interesting, because it identifies geopolitical impacts of oil import as major contributors to life-cycle impacts of oil-based vehicles. A large externality cost is found to be associated with military expenditures aimed at securing access to oil supply, and an even larger impact from the indirect influence on the economy (exemplified by that of the US), by having to adapt to the trade conditions and policy wishes of oil-producing countries. Such a connection between military policy and oil security is frequently suggested (see Sørensen, 2011a; Chapter 5), but debate is still ongoing on the strength of the coupling, and on the extent to which the recent Iraq war (that started in 2003) can be used as a valid example. Some of these ideas are taken up in another life-cycle comparison between a US gasoline and pure fuel cell vehicle (Sun et al., 2010), but without considering the more likely market entrants such as hybrid configurations. The conclusions are similar to those of Thomas (2009), that externalities are larger for gasoline vehicles than for fuel cell vehicles, over a range of different assumptions.

Figure 6.13. "Well-to-wheel" direct costs in 2015 Korean scenarios (WTW, left column), and same plus social costs taken as the externality costs of global warming and air pollution (three right-hand columns, for three monetising assumptions: minimum, average, and maximum encountered in a literature survey). The life-cycle assessment considers current gasoline (G) and diesel (D) cars using fuels at (increased) 2015 prices, and compare them with pure fuel cell cars, assumed economically competitive by 2015, possible assisted by a Korean government subsidy, and using hydrogen derived from a number of sources (Hx; x = A, …, F). Acronyms are SR = steam reforming, NG = natural gas, LPG = liquid petroleum gas, WE = water electrolysis, C = centralised production, S = decentralised production at filling stations. Note the increasing difference between wind-produced hydrogen and hydrogen produced by the (current) Korean electricity mix (Lee et al., 2009; used by permission).

6.2.5 Life-cycle assessment of other vehicles for transportation

Life-cycle assessments have been made for fuel cell buses and hybrids, as this has been a major demonstration area for the new technologies, along with passenger cars. Ally and Pryor (2007) finds that greenhouse gas emissions of fuel cell buses under Western Australian conditions are far greater than those of conventional diesel buses, unless hydrogen for the fuel cells are produced by wind energy and not grid power or steam reforming. Ou et al. (2010) compare life-cycle impacts of alternative bus technologies under currently prevailing Chinese and US conditions and finds substantial differences due to differences in the energy systems in the two countries, and conclude that under Chinese conditions, while some alternatives such as electric buses are advantageous in an assessment including life-cycle externalities, fuel cell buses are not and require major R&D breakthroughs. Similar investigations have been made earlier for buses (Tzeng et al., 2004).

Figure 6.14. Social costs derived from life-cycle analysis for a range of passenger car options. Included are global warming, air pollution, and the economic and military costs estimated to be associated with imported oil, increasing at a rate displayed by the top curve for internal combustion engine vehicles. The following curves are for oil-battery hybrids and the bottom curves for electric and fuel cell vehicles. The military cost of protecting oil supply is taken as US\$1.18 × 1011 y-1, the socio-economic cost of transfer of wealth, production capacity and disruption losses as US\$2.65 × 1011 y-1, adding up to about US\$10 per GJ of oil. The left scale translates these figures into indicative US \$/y for a single car. Based on Thomas (2009), used by permission.

Although fuel cell operation of trains are under consideration (e.g., in Japan), the advantage is not obvious, considering that most train services are electrified or planned to become electrified and thus can use renewable power say from wind directly. An interesting life-cycle analysis of a proposed highspeed, underground maglev-train in Switzerland finds that life-cycle impacts under wide ranges of assumptions are higher than for present trains and that the valuation of time saved cannot outweigh the increased impacts caused by the augmented use of the faster trains (Spielmann et al., 2008).

For ships, life-cycle studies have largely been confined to special cases, such as auxiliary electric equipment powered by SOFC (Strazza et al., 2010), while for land transportation, studies covering a range of transport modes have been carried out, at least for the impacts of greenhouse gas emissions (Uherek et al., 2010). Other environmental assessments have been for possible use of fuel cells for propulsion of ships (Altmann et al., 2004). This European Commission study considers both a 400 kW PEM fuel cell propulsion system based on compressed hydrogen gas for a small ferry and also a 2 MW auxiliary power system for a larger ferry on the route Oslo-Kiel. Here, a natural gas-based SOFC or MCFC system is compared to a liquid hydrogen-based PEMFC power unit. In the small ferry case, the base alternative used today is a diesel engine using light fuel oil, where the larger ship would be using heavy fuel oil. The basic database used in this life-cycle study is said to be proprietary, but the numbers are very similar to those one would derive from publicly available emission and impact databases.

Ship's energy system (engine or PEM fuel cell)	CO2-eq. from fuel production (g/kWhth)	CO2-eq. from propulsion (t/y)	SO2. from fuel production and use (t/y)
Diesel engine	38	302	0.125
H2 from wind	25	18	0.014
H2 from natural gas	445	230	0.10
H2 from gasification of forest residues	14	7	small
H2 from agricultural waste	54	28	small

Table 6.12. Emissions of greenhouse gases and SO2 from fuel production and use in propulsion for a small ferry, using different technologies (annual fuel consumption 991 MWh of diesel or 539 MWh of hydrogen). Based on Altmann et al., 2004.

Table 6.12 gives rounded values found for the greenhouse emission associated with selected routes of hydrogen fuel production and usage for the small ship, for different energy systems and fuels, along with SO2 emissions (omitting a discussion of a possible bonus for producing useful byproducts). The same investigation for the auxiliary power system of the large ship of course gives similar results for the hydrogen-based systems, but not for the diesel engine, where two conditions are different: the larger diesel engine for the large ship is considerably more efficient, giving lower per kWh emissions, but the use of heavy fuel oil often entails higher emissions of SO2, with details depending on the origin (and sulphur content) of the fuel.

6.2.6 Life-cycle assessment of hydrogen storage and infrastructure

Most of the life-cycle assessments of hydrogen applications contain an analysis of the storage system used in that particular application. For several types of energy storage systems, the energy requirement for construction is a major concern. Table 6.11 gives the energy consumption for four current storage options for mobile applications, based on Neelis et al. (2004) and references quoted therein.

Materials used in construction	Weight (kg)	Primary energy (MJ)
Carbon reinforced epoxy resin container for compressed hydrogen storage
Low-density polyethylene	0.5	16
Epoxy resin	16.6	2 460
Poly-acrylonitrile	38.9	4 232
Stainless steel	2.8	125
Total	58.7	6 833
Container for liquid hydrogen
Aluminium	30.7	7 140
Stainless steel	5.0	223
Total	35.7	7 363
Low-temperature Fe-Ti hydride system
Iron	271.3	12 107
Titanium	232.7	99 945
Aluminium	88.0	20 467
Total	592	132 519
High-temperature Mg hydride system
Magnesium	119	35 486
Aluminium	84	19 536
Total	203	55 022

Table 6.11. Material and energy use for hydrogen storage containers (based on data from Neelis et al., 2004).

For small containers suitable for portable applications, Paladini et al. (2003) give weights around 100 g and volumes around 20 cm3 for metal hydride stores using LaNi5 or Mg2Ni.

For fuel cell vehicles carrying reformers and using methanol as a fuel, additional life-cycle impacts from the methanol life-cycle have to be considered. A few of these have been analysed, for example, by Pehnt (2003). In a sequel to the work mentioned in section 6.2.4 (Fig. 6.13), Lee et al. (2010) looks further at a wind powered filling station with electrolyser, hydrogen storage, and dispenser. General remarks on long-term energy storage, relevant to hydrogen infrastructure, are provided in Bielmann et al. (2011), using Switzerland as an example.

6.2.7 Life-cycle assessment of hydrogen systems

Individual hydrogen application systems include those for passenger cars, for which life-cycle analysis is discussed in section 6.2.4, and for other means of passenger or freight transportation (section 6.2.5), including storage and dispensing infrastructure (section 6.2.6). Similar life-cycle analysis can be performed for a building-integrated fuel cell system, for a larger building as well as for a detached residence (Fleischer and Oertel, 2003). Combined heat and power production may be considered for various types of fuel cells (see Fig. 5.50, including the option of using reversible fuel cells) and may be compared to similar systems using natural gas (Ren and Gao, 2010). The possibility of distributing the negative impacts on two types of energetic outputs, heat and power, makes the specific impact for each of them smaller per kWh of useful energy. A specific investigation of a residential combined power and heat system based on PEM fuel cells has been made for Malaysia (Mahlia and Chan, 2011). The system is autonomous, as far as electricity is concerned, aimed for residences far from power grids (and yet with access to piped natural gas for a fuel reformer in front of the fuel cell!). The system includes a battery store and according to a lifetime direct cost comparison (called LCA although no climate, pollution, or other social costs are included) could become marginally feasible under certain cost conditions.

Turning from individual hydrogen systems to a complete hydrogen-based energy supply system, such as one of the scenarios considered in Chapter 5, one may ask if additional positive or negative impacts can be imagined. One suggestion put forward by Tromp et al. (2003) is that small amounts of unconverted hydrogen could escape from the transmission or conversion equipment and be released to the atmosphere as H2 molecules and subsequently possibly diffuse to the stratosphere, where they might assist in forming additional water vapour. Such an increase could have substantial impacts on cloud cover, ozone chemistry and derived climatic changes. In a more detailed modelling work, Schultz et al. (2003) show that even with a larger than likely escape of hydrogen, the tropospheric burden of OH will decrease in a hydrogen-based energy system, due to reductions in NOx emissions. As a result, surface ozone concentrations become significantly increased, and the amounts reaching the stratosphere will more than compensate for the under 2% depletion that could be caused by added hydrogen molecule pollution.

6.3 Uncertainties

As a set of technologies still under development, hydrogen system components are difficult to evaluate, and the preceding attempts described, both as regards cost and life-cycle impacts, are necessarily incomplete and uncertain.

Highest accuracy is expected for those hydrogen technologies that are similar to existing technologies, such as underground gas storage and container gas and liquid stores, and such as hydrogen production technologies based on steam reforming and related technologies, or acid electrolysis.

Figure 6.15. Humidifier (Lossnay™) concept for positive electrode area. (From K. Mitsuda, H. Maeda, T. Mitani, M. Matsumura, H. Urushibata, H. Yoshiyasu (2004). Technical issues of polymer electrolyte fuel cells. In 15th World Hydrogen Energy Conference, Yokohama 2004, Paper 01PL-02. Used by permission from Mitsubishi Electric Corp.

The most uncertain technologies are the core fuel cell technologies and the new storage technologies such as hydrides and carbon materials. Although laboratory experiments have in many cases been followed by prototype demonstration projects, there are still major uncertainties in future cost and operational stability and durability of the equipment. The learning curves shown in Fig. 6.1 indicate the difference it would make, say, by shifting the curves upwards, if the durability of fuel cells for both mobile applications turn out to be less than the anticipated 5000 h (like present advanced batter-

ies), or if that of stationary applications should turn out less than the 40 000 h hoped (as compared to the 200 000 h characterising, e.g., wind turbines and being considered a lower limit for nearly all other current energy technology).

A large technical problem remaining for PEM fuel cells seems to be water management. The cost of the membrane-electrode assembly is being decreased by use of graphite powder-polymer composites as separators, and the possibility of using new low-humidification concepts, where heat and water are simultaneously exchanged between oxygen/air flow channels, is under investigation, as indicated in Fig. 6.15.

Of course, many other factors will contribute to determining the fate of new energy technologies. Will the overall efficiency goals from primary energy through hydrogen and fuel cells to electric power be met? Will the oil production peak over the next decade as predicted, and exactly what impact will that have on cost? How will the other fossil fuels fare? Will the resources of natural gas be able to hold up to the demand, and will coal gasification and CO2 removal make coal an acceptable fuel for future expansion? If fossil fuels are becoming increasingly problematic, is this the chance for hydrogen in the growing transportation sectors of China and India? Will these countries make 2 billion people take over the inefficient energy usage habits of countries such as the United States, or will they gain a factor of two by looking to Europe or even better develop still higher energy efficiency? Current trends indicate that the emerging mega energy-users will be wasteful, at least initially, perhaps influenced by the multinational energy and automobile companies already establishing themselves in Asia and often focussing on customers not caring much about energy efficiency. The questions are many, and the uncertainty regarding outcomes is large.

6.4 Problems and discussion topics

6.4.1

Try to extend the Japanese cost scenario for the transportation sector, described in section 6.1.5, to decentralised use of hydrogen and fuel cells in buildings, along the lines indicated in Chapter 5, Fig. 5.2.

6.4.2

Look at the production of hydrogen by fermentation and on the basis of biomass from waste or from dedicated crops. Estimate current and possibly reduced future costs of such a hydrogen production system, taking into account the cost of bio-feedstock, its transportation as well as the hydrogenproducing conversion equipment (see Chapter 2, section 2.1.5).

6.4.3

How would you suggest the introduction of hydrogen in less developed countries to proceed? Would the preferred systems be different from those envisaged in currently industrialised countries?

CONCLUSION: A CONDITIONAL OUTCOME

The preceding chapters have considered the possible ingredients in a hydrogen system, which is an energy system where hydrogen plays a central role as energy carrier and possibly also as storage medium. The latter function is required for intermittent renewable energy primary sources, but is also useful for systems based on exhaustible resources such as coal or thorium. The story is still an unfinished one, not only because of challenging scientific and technical problems, but also due to the difficulties of economic systems such as current national entities to cope with radically new systems with a different cost structure from the ones that have prevailed for the last century. On the other hand, hydrogen technologies offer ways of handling some critical problems facing current societies, ways that could be preferable to other ways of coping with the problems. In this chapter, some of the main issues are recapitulated.

7.1 Opportunities

Looking broadly at the fuel cell technologies for converting fuels such as hydrogen into electric power, one striking feature is the fact that just about any primary source of energy can be used. Fuels can be fed into the hightemperature fuel cells or be converted into hydrogen for the low-temperature cells. Electricity can be transformed into hydrogen, either by fuel cells in reverse mode or by other schemes. This allows the flexibility of going back and forth between non-storable electricity and storable hydrogen. Of course there will be losses in these processes, but as ways of handling the fluctuating production from solar or wind power devices, there is no technology more efficient for matching supply and demand. Of course, only electricity that cannot be used when produced will have to be sent through the loss-prone conversions. This feedstock flexibility may prove invaluable in creating a smooth path for transition from the present to a hydrogen-based energy system.

The structure of the energy market may be influenced in a positive way by the hydrogen technologies. This is at least true for the PEM fuel cells, because they are modular and can be installed at nearly the same price per kW at any size, from family installations to central power plants. This means a large freedom for independent power producers who may avoid the restraints of conventional energy companies in regard to inertia in technologies used and services offered. Everyone can install a PEM fuel cell in the basement, or purchase a car with one, to produce power and associated heat for the building or for portable equipment on the basis of piped or locally stored hydrogen. Primary, intermittent power may come from rooftop photovoltaic panels or from a wind farm in which the user may possess shares. The implication is a decentralisation and restructuring of the energy industry that deserves the name much better than the current shifts in large-scale ownership.

In terms of environmental impacts, the hydrogen conversion and infrastructure technologies are generally benign. Some of the high-temperature fuel cells may produce nasty wastes that have to be dealt with delicately, but generally, from hydrogen through storage and converters to end-use energy, the negative impacts are small. The inefficient end-of-pipe remedies such as car exhaust catalysts used with current automotive fuels can totally be done away with for vehicles using hydrogen in fuel cells. This is not necessarily true for the production of hydrogen from primary energy sources or from on-board reformers in vehicles. The primary hydrogen production can be done with small as well as large negative impacts. Potentially large impacts arise if nuclear or fossil energy sources are used. In the nuclear case, remedies would require decades of research efforts in developing new, safe reactor types, just to partially solve the problems (see Chapter 5). In the fossil case, a possible remedy is offered by CO2 sequestering and development of acceptable deposition and long-term CO2-capture and storage methods. Also production of hydrogen from biomass can have large negative impacts. A number of remedies appear possible, although few have been tried on a realistic scale as yet. Only hydrogen production from wind or solar power that is not used immediately seems to offer genuinely sustainable and ecologically acceptable avenues.

The decentralised potential of renewable and hydrogen technologies offers ways of delivering power to remote areas not yet deemed economic to reach by electricity transmission lines. This may turn out to be a very important feature for many developing regions of the world. Due to high cost, it is an advantage that fuel cells can be introduced gradually through hybrids.

7.2 Obstacles

The obstacles to hydrogen and fuel cell development as a backbone energy carrier are both technical and social. The social obstacles are in part political and institutional, as they relate to the decision and economic structure prevailing in present societies. In the past, almost all new technology introduction has been made by foresighted individuals (sometimes in government, sometimes in industry) and in plain disrespect for economic views at the time. Today, it seems that vested interests have become stronger, making transitions away from the way things happen to be at the moment more difficult. Furthermore, traditional economic thinking has managed to increase its influence on decisions made in private or public spheres. The introduction of the personal automobile 100 years ago may not even have happened if economists had had an influence on the decision.

Evaluating the cost of road infrastructure and the negative impacts of collisions between these at the time of introduction poorly manoeuvrable vehicles, as well as the astronomical cost of producing them, plus considerations of the cost of extracting enough oil products to operate them, would certainly have discouraged any decision-maker listening to his economic advisors. As another example, electrification of the rural areas of currently industrialised regions by extension of grid lines to these areas would certainly never have happened if city electricity-users had not subsidised rural users through the uniform (or at least more uniform than suggested by direct costs) pricing policy of public power utilities. Had the electricity industry been privatised 100 years ago rather than now, no one would probably have dared to propose extending grid lines to thinly populated areas, as profits clearly were visible only in locations with high concentrations of potential power users. Yet the comprehensive electrification is precisely what distinguishes the presently rich, industrialised countries from the less fortunate countries where rural life has hardly changed over the past several thousand years (except for a few transistor radios operated on expensive batteries).

The technical obstacles are more tangible. Many of the methods for producing hydrogen should be upgraded for supplying the amounts needed for energy supply rather than the current limited quantities for industrial feedstocks. Hopefully, the price of traded hydrogen can be brought down along with the larger volumes. The alternative to such mass supply is decentralised production, which diminishes the need for transmission or transport. Although decentralised production is normally more expensive than mass production, it also offers advantages to the customers, in terms of control and resilience.

The fuel cell technology is currently at its beginning, with a probably long

learning process ahead. Even for the PEM technology currently receiving most attention, the cost is high, the performance has problems (such as water management and cross over), and the working life is too short. Hightemperature fuel cells have some promising attributes for stationary uses, but are probably further away from the marketplace. Fuel cell conversion, including that of reversible cells, is probably another very important area for development and cost reduction efforts.

Hydrogen storage in containers (as compressed gas or liquefied) is an existing technology needing only modest refinement, unless radical changes such as ultra-high pressure is considered necessary. Also, underground storage is likely to be feasible as a bulk storage option at low cost for many geographical locations. Only the increased safety of metal or chemical hydride stores, or the possible advanced carbon storage forms, requires substantial research and development, both in terms of technology and cost, in order to become viable alternatives for storage. The acceptance of extra cost by an end-user liking to be in control of his energy supply is probably not stretching as far as accepting the current cost of metal hydride stores, in case cheap underground stores are centrally available, and provided that transmission costs are not excessive.

It is possible to consider a hydrogen "economy" based on hydrogen as the storage medium for intermittent energy sources, also in case direct use of fuel cells as power sources does not turn out to be technically or economically viable. The inexpensive cost of, for example, the underground storage options will make such an energy storage system feasible, even when one has to use gas turbines (presumably with efficiencies lower than that of fuel cells) for regenerating electricity. However, the present efficiency of gas turbines is not much lower than that of fuel cells, and the factor of three penalty in volume relative to natural gas should make the cost of the store at most a factor of three higher, which is quite acceptable in a stationary application. The turbine cost is in any case low.

From this line of argument it follows that a large penetration of intermittent renewable energy primary sources often thought to require either storage or backup from fossil fuels (which may not be possible in a not so distant future), could as well derive the storage needed from a hydrogen store that may be both cheaper and more convenient than other forms of storage, even if the roundtrip efficiency is slightly lower. However, many of the advantages of hydrogen as a storage medium are enhanced, if hydrogen can also be used in fuel cell systems, notably for the transportation sector. Therefore, the development of viable fuel cell systems remains a top priority.

Furthermore, the infrastructure required for hydrogen to be useful as a general energy carrier must become implemented. Although both pipeline transportation, container transportation and dispensing fuel stations are all fairly well-established technologies needing only modest further development, the present uncertainty of which fuels should be used in the future could cause some confusion, say, if filling stations have to offer both methanol, compressed hydrogen, and gasoline. This is not impossible, but a likely situation will be one with different options offered in different locations. Incompatibility of filling station infrastructure in different countries (e.g., in Europe) would cause trouble for people driving their fuel cell vehicle from one country to another (just as present drivers of bio-diesel cars find that some countries do not offer this fuel but only truck diesels). The same is true if efforts to internationally standardise safety norms and equipment standards for hydrogen (and other electric) vehicles are not successful.

7.3 The competition

In both shorter and long-term appraisals, battery-electric vehicles have been seen as the main competitor to fuel cell-electric vehicles. The simulations carried out in section 4.1.3 indicate the likely winner is a hybrid between battery and fuel cell power systems, for general vehicle usage, but that battery vehicles may be the best solution for specific urban transportation tasks facilitated by having fixed routes, limited range requirements and the possibility of establishing a flexible array of recharging points. Elements of an economic comparison of these options are discussed in section 7.4.

Another competitor is biofuels from renewable use of biomass residues or sustainable crops not having negative impacts on the prospects of feeding the world population. In terms of environmental acceptability, biofuels is one step down the ladder from hydrogen, due to emissions from combustion (cf. Chapter 6). They must be used in vehicles with both particulate matter filters (like current diesel vehicles) and with filters for conventional air pollutants such as NOx'es or SOx'es. Despite the cost of these arrangements, biofuels in internal combustion engines may still, at least for a while, be less expensive than fuel cell or battery traction. Biofuels may be traded and used as oil products. Sobrino et al. (2010) note that some countries would need to import either biomass residues or biofuels in order to satisfy their demand.

The internal combustion of biofuels such as ethanol in an Otto engine or biodiesel in a diesel engine is less efficient over a driving cycle such as the one shown in Fig. 4.7 than the electric motor of a battery or fuel cell vehicle, because the efficiency at part load is less than at optimal load, whereas for a fuel cell, partial load is less important (see Fig. 4.5). That part load is a condition of automobile driving is clear from the power duration curve shown in Fig. 7.1. It is based on a typical passenger car, used about 1140 hours in a year of 8760 hours, and with an average load during operation, which is less than 25% of the rated power. The duration curve in Fig. 7.1 is constructed from Fig. 4.9 (the mixed driving cycle), assuming that this driving cycle is representative for the entire year of driving. Taking the power output from a pure fuel cell car rated at 40 kW gives a capacity factor not lower than for other types of passenger cars, for example, one driven on biodiesel fuel.

Figure 7.1. Power output duration curve for annual use of a passenger car based on the driving pattern of Fig. 4.7 and with a 40 kW fuel cell system and minimal battery (1.67 MJ capacity to ensure stable input to motor). The 6 kW and 0 kW segments of the duration curve during operation represent idling and stopping−for example, at red traffic lights.

The efficiency of the diesel engine of a passenger car similar to the Little Red Riding Hood varieties considered in Table 4.1, but running on biodiesel fuel, is shown in Fig. 7.2, as function of engine (rotational) speed and torque delivered. As function of time during the mixed driving cycle, the efficiency varies as shown in Fig. 7.3. The operating points in a speed-torque diagram are influenced by gear exchange ratios. An example of this relationship in shown in Fig. 7.4, based on measurements for a slightly larger vehicle (NREL, 2001).

Figure 7.3 clearly exhibits the reduced efficiency of the internal combustion engine in various situations of part load. The corresponding efficiency for the fuel cell version of the vehicle was shown in Fig. 4.22. It stays much closer to the maximum efficiency, under most driving conditions. Figure 7.4 shows the position of operating points for different gearbox exchange ratios, when driving along the Fig. 4.7 cycle. While the operating points are fairly localised in the torque and speed diagram while driving in the lowest gear, this is not the case for higher gear level settings (level 2 to 5).

Figure 7.2. Calculated engine efficiency of biodiesel version of the Little Red Riding Hood passenger car considered in Chapter 4, as function of torque delivered and rotational shaft speed. The curve with crosses indicate the maximum torque that can be realised on the road, for a given rotational shaft speed.

Figure 7.3. Simulated engine conversion efficiency as function of driving time along Fig. 4.7 mixed driving cycle, for biodiesel version of the Little Red Riding Hood vehicle.

Figure 7.4. Calculated operating points along the mixed driving cycle, for the biodiesel Little Red Riding Hood, as function of torque delivered and shaft rotational speed, with indication if the gearbox exchange setting at each point.

The efficiency behaviour as function of torque delivered and rotational shaft speed (Fig. 7.2) is also influenced by the temperature of the engine. Figure 7.5 shows the fuel consumption pattern during cold start and for a well-heated engine having been in operation for a while. The difference is substantial, with typically about 50% higher use of diesel fuel in cold conditions. The temperature has similarly large influence on the amount of pollutants emitted. Figures 7.6 to 7.9 shows the simulated emissions of four important pollutants, NOx, unburned hydrocarbons, CO, and particulate matter, for operation in cold conditions and in mid-trip hot engine conditions. As mentioned, the underlying measured environmental data are for a slightly larger vehicle, but scaled to the Lupo-specifications of Tables 4.1 and 6.6. Figure 7.10 shows the total emissions of the same four substances, as function of time during the mixed driving cycle of Fig. 4.7.

One notes that for pollutants such as NOx and particles, the cold-start penalty is similar to that of fuel consumption in Fig. 7.5, while for HC and CO, the emissions at cold-start are several times higher than later at stable cruising. Figure 7.10 summarises the time-wise emission results for the entire trip.

Other aspects of the biodiesel and fossil diesel-car (the VW Lupo) simulation project may be found in Sørensen (2006b; 2010b), and costing of the impacts in a life-cycle context is discussed in Sørensen (2011a, Chapter 7).

Figure 7.5. Calculated fuel consumption for biodiesel version of the Little Red Riding Hood passenger car, as function of torque delivered and rotational shaft speed. The upper panel pertains to cold start conditions, the lower panel to operation in warm conditions (where curves are similar to the efficiency curves in Fig. 7.2).

Figure 7.6. Calculated NOx emissions from biodiesel version of the Little Red Riding Hood passenger car, as function of torque delivered and rotational shaft speed. The upper panel pertains to cold start conditions, the lower panel to operation in warm conditions. Consistent with fuel consumption results, the NOx emissions are about 50% higher during cold-start.

Figure 7.7. Calculated unburned hydrocarbon emissions from biodiesel version of the Little Red Riding Hood passenger car, as function of torque delivered and rotational shaft speed. The upper panel pertains to cold start conditions, the lower panel to operation in warm conditions. The cold-start emissions are here 5-7 times higher than those found after reaching cruising temperature levels.

Figure 7.8. Calculated carbon monoxide emissions from biodiesel version of the Little Red Riding Hood passenger car, as function of torque delivered and rotational shaft speed. The upper panel pertains to cold start conditions, the lower panel to operation in warm conditions. The cold-start emissions are 4 to 7 times higher at cold-starts, more similar to the situation for HC than for NOx.

Figure 7.9. Calculated particulate matter emissions from biodiesel version of the Little Red Riding Hood passenger car, as function of torque delivered and rotational shaft speed. The upper panel pertains to cold start conditions, the lower panel to operation in warm conditions. The cold-start emissions are about twice as high at cold start than later, in similarity to the NOx emissions.

Figure 7.10. Time-wise distributions of the emission-types from Figs. 7.6-7.9 for a biodiesel Little Red Riding Hood vehicle driving the mixed driving cycle of Fig. 4.7, here shown on a logarithmic scale.

The analysis of emissions underlines the possible role of biofuels as being temporary and eventually one of facing replacement by zero-emission technologies. However, this does not exclude that biofuels may offer the most benign solution in a transition phase between the fossil fuels, also characterised by emissions of pollutants and without the CO2 benefits offered by biofuels, and some truly sustainable solution. In the long range, biomass could be one of several renewable sources for producing clean fuels such as hydrogen. The biodiesel investigations are quite similar to those of diesel fuels derived from fossil oil, and the remarks on diesel as one possible transition fuel are underlined by the fact, that common-rail diesel engines today are the most energy-efficient internal combustion engines, and that a reduction in oil demand by a factor of 2 to 3 can be achieved by replacing current average automobile engines by the best common-rail diesel engines, a policy that will furnish badly needed time to implement the most urgent sustainable global warming actions without at first worsening the problem. Compared to any new energy supply option, just using energy more efficiently is a remedy already available, and at zero or minimal cost, as long as the efficiency gain aimed for is just a factor of 3-5 (Weizsäcker et al., 2009; Sørensen, 2010a).

Outside the transportation sector, the competitors to using hydrogen (as a combustion fuel or for electricity production through a fuel cell power plant) are linked to the primary sources of energy. Renewable sources employed in wind turbines or photovoltaic panels can produce electricity directly and going through a separate energy carrier like hydrogen is neither required nor desirable, except as one particular way of dealing with the intermittency of these renewable sources. Here hydrogen storage may prove more viable than other storage solutions. This will be looked at in section 7.4.1, with details of options for power restoration from hydrogen dealt with in sections 7.4.3 and 7.4.5, depending on whether the preferred structure of the energy system is centralised or decentralised.

7.4 The way forward

The assessments made in this book suggests five key areas of hydrogen introduction, which could happen individually or in combination. They are:

Hydrogen as energy store for variable renewable energy systems.
Applications of hydrogen and fuel cells in the transportation sector.
Stationary applications of fuel cells in building environments.
Portable applications of fuel cells.
Stationary application of fuel cells in large power plants.

Following, each option is briefly assessed on the basis of the treatment made in previous chapters of this book. Such explorations constitute a set of suggestions or proposals for development efforts (in the US called "roadmaps", in Europe "action plans") in the hydrogen and fuel cell areas.

7.4.1 Hydrogen storage in renewable energy systems

In order for fluctuating renewable energy sources such as wind and solar radiation to attain a dominant share in any energy system, storage must be part of the system. Only for modest levels of penetration can demand-supply mismatch be handled by trading (e.g., in international power pools), but in energy systems with favourable transmission infrastructure, the penetration can still be substantial. Sudden needs for large imports are costly and sudden spot-market exports must accept low prices, so such events must be fairly infrequent (Meibom et al., 1999). In any case, this set-up only works if neighbouring systems are based on systems that can adjust their production levels as required (i.e., fuel-based or hydro systems). If the transition to variable primary sources is taking place generally, there may be no alternative to energy storage. An assessment of possible energy storage technologies (Sørensen, 2010a) does not identify any other storage option more suitable than hydrogen underground storage. The options of using aquifer or salt intrusion storage for hydrogen are natural extensions of their proven usefulness for natural gas storage and promise a very low storage cost. In areas without these options, rock cavern storage is technically feasible but somewhat more expensive.

If stored hydrogen is not used to regenerate electric power, it may alternatively be forwarded to dedicated hydrogen users, by pipeline, truck, ship, or other means of transport. The cost of such infrastructure is often high, and an optimisation of hydrogen transport requirements by considering locations of components such as filling stations for vehicles or industry using hydrogen should be made. The additional cost of small-scale hydrogen storage nearer to the customer (at filling stations or completely decentralised to individual buildings) should be evaluated in relation to the total cost and convenience to the final consumer.

In any case, this discussion indicates that using hydrogen as storage medium in connection with variable renewable energy systems based on wind turbines or photovoltaic collectors is an option that can and very likely should be used, independent of the outcome of the development of fuel cell technologies towards a wider role for hydrogen in the energy system. Without fuel cells, conventional methods for regeneration of electricity from hydrogen (e.g., gas turbines, Sterling engines) have to be used, implying a lower efficiency than that of stationary fuel cells, but not in a dramatic way (40-50% rather than 50-65%).

In summary, the consideration of hydrogen stores should be part of any energy policy aiming to replace environmentally adverse and politically if not resource-wise uncertain reliance on fossil fuels by renewable energy technologies.

7.4.2 Fuel cell vehicles

The transportation sector is clearly the sector where a transition to sustainable energy sources is most difficult. Battery-based electric vehicles have taken many decades to reach the present stage, in terms of technical performance, and the economic goals are still not fulfilled. In any case, a purely battery-operated vehicle is expected to remain limited in terms of range.

The analysis in Chapter 4, such as Fig. 4.26, indicates that fuel cell-battery hybrids may well turn out to be a better solution, due to the complementary advantages of the two systems. The current emphasis on PEM fuel cells requires substantial further technological development of PEM fuel cells in regard to reliability and long service life, and particularly in regard to cost, implying that a fuel cell totally responsible for powering the vehicle will be a very distant possibility. The hybrid fuel cell-battery option may offer full performance with a lower sum of the ratings of each of the two systems, as compared with either a pure fuel cell vehicle or the a pure battery-operated vehicle. Considering vehicle weight, the batteries much be of lithium-ion type or other concepts with as favourable mass and energy density. The cost optimisation is illustrated in Fig. 7.11, for different guesses regarding the relative price of fuel cell and battery systems. Comparing with Fig. 4.26, it is seen that the optimum price is obtained for upper-end plug-in hybrids, or perhaps for the lowest autonomous hybrids, but in no case for vehicles with larger than 50% dependence on fuel cells.

Figure 7.11. Cost of the PEM fuel cell Li-ion battery hybrid systems considered in section 4.1, based on an estimate of the current price of a purely battery-based or purely fuel cell-based propulsion system (upper curve end-points). The effect of halving either fuel cell or battery price is indicated by the dashed curves. In all cases, the price optimum is for a hybrid configuration with a fuel cell rating of 10 to at most 20 kW and a battery rating of 15 to 60 MJ, which means plug-in hybrid configurations (Sørensen, 2010b).

If the PEM cell development does not meet the set goals, a possible alternative would be acid polymer cells operating at temperatures around 200°C. However, the development stage of this concept is currently much less advanced, and a shift to this technology will likely have the effect of further delaying the deployment of viable vehicle fuel cells in the general automobile manufacturing lines, already behind schedule due to cell lifetime issues.

Assuming that the PEM fuel cells eventually do reach the technical goals, their rate of introduction will depend on the size of cost reductions obtained as market penetration and hence accumulated production advances. This phase may be speeded up by creating infrastructure facilities in dedicated areas at an early stage, and by market introduction initiatives such as offering a reward for the absence of pollution during operation of the vehicles (and also during hydrogen production if this is based on renewable energy sources). The transition will be assisted by a gradual introduction of hybrid vehicles, starting with the current generation of gasoline-battery or dieselbattery cars and shifting to fuel cell hybrids after the market has become accustomed to the hybrid concept. The fossil fuel-battery hybrids may help lowering the cost of advanced batteries and thus act as a catalyst for subsequent introduction of fuel-cell-battery hybrids.

Initial introduction of fuel cell-battery hybrid vehicles could be in niche markets where the range requirements are modest, such as fixed route buses, delivery vehicles and trucks, but this approach is more relevant for pure battery vehicles. The markets for fuel cell ships seem more interesting that current activities indicate. The auxiliary system components such as filling stations and transport/transmission systems for of delivering hydrogen to these vehicles (or producing it on-site) should not be neglected, as their cost can substantially influence the overall attractiveness of the fuel cell system.

7.4.3 Building-integrated fuel cells

The PEM fuel cell is the leading alternative for building-integrated applications, where multi-functionality would allow current natural gas burners to be replaced by combined heat-and-power systems, possibly with the additional option of supplying hydrogen to a one-vehicle filling station. One may think that if the automobile industry is successful in developing a viable PEM fuel cell for vehicles, then it will be directly applicable for stationary purposes. However, this is only partially true, as the service-life requirements are much higher for stationary uses. Current natural gas users could be the target for building-size fuel cells, and in the transition period where natural gas is available, it could be used to generate hydrogen if a reformer is integrated in the system. The cost of such a system appears high, considering that there are many ways a building could obtain inexpensive waste heat and cheaper electricity from wind (or perhaps with time solar power). Avoiding a separate gas reformer had also led to proposing small SOFC for domestic uses, but again cost is presently too high. The customers that currently have access to piped natural gas anyway constitute only a segment of the total market, with a share that varies between countries.

Looking a bit further ahead and assuming a successful reduction of fuel cell costs, an interesting system is that of a reversible PEM cell capable of converting excess electricity supplies (from renewables) to hydrogen, to be stored in the buildings for a moderate period of time. Experiments having led to laboratory increases in the efficiency of the PEM electrolysis operation of specially designed PEM cells (section 3.5.5) makes this an interesting proposition for the future (with further application in the area of filling station on-site conversion). One remaining problem is suitable hydrogen stores for safe operation in building environments (beyond what can be directly stored in vehicles parked at the building, cf. Chapter 5, but not without concern of safety during fires). Finalising development of hydride storage systems may be required to realise this degree of decentralisation.

Overall, the cost and technical performance of PEM fuel cells exhibit some of the same problems for decentralised stationary as for mobile applications, with the mentioned qualification that durability requirements are substantially higher. Both applications are in areas, where consumers in most parts of the world are accustomed to paying a fairly high price for energy, compared with that characterising central power plant operation.

7.4.4 Fuel cells in portable equipment

The current generation of portable consumer products are in several areas close to the technical limits of battery technology. This is true for portable computers, where high performance requirements have led to considerable efforts to make energy use as efficient as possible. Flat screens of very low power consumption have been developed, and both central processors and peripheral equipment are rapidly approaching a very good energy performance. Driving forces in this direction include also the stationary computers, because damage caused by excess heat is a decisive determinant of lifetime and performance. For these reasons, the current autonomous operating time for laptops of some 4 hours between battery recharging is a major limitation both for present users and for further performance development.

This suggests that portable applications may offer a very attractive upstart niche market for fuel cells and small-scale hydrogen or methanol stores, just as they did some years ago for advanced battery types (first NiMH then Liion batteries). The discussion in section 4.6 suggests that the most appropriate technology for this type of application could be a direct methanol fuel cell, due to the volume requirements of the fuel store. Increasing the computer's operational time between recharging or reloading to some 10-20 hours, this option would offer advantages to users that they would likely be willing to pay extra for, as they already do for advanced Li-ion batteries typically consuming up to US\$200 of the cost of a portable computer.

Also for the fuel cell development in general, the existence of such niche markets can have a positive effect, reminding of the success of Japanese manufacturers obtained by incorporating solar cells in consumer products such as watches and calculators and thereby earning a profit capable of covering the total Japanese development costs of solar cells, at least for an initial ten year period.

7.4.5 Fuel cells in centralised power production

Because the current bulk power production sector is dominated by coal-fired power plants and coal reserves seem adequate for another couple of hundred years (Chapter 5), the fuel price issue may be seen as less critical than for oil in the transportation sector. However, considerations of greenhouse gas emissions put coal lowest on the list of acceptable fuels, and ways of decarbonising coal are intensely discussed. Here, primary hydrogen conversion (if not coal to hydrogen replacement) is high on the agenda, because recovering carbon from power plant stack exhaust gases is a fairly low-efficiency and energetically unfavourable option. In case the renewable energy transition is successful in the power industry, hydrogen will as mentioned in section 7.3.1 also have an important role already for central energy storage, even if the re-generation of electricity is not done by fuel cells.

If the somewhat higher efficiency of fuel cells warrants their higher cost, the most likely fuel cell type for the bulk power sector is the SOFC. However, there are still many technical issues to resolve, especially if fuels other than pure hydrogen are to be used (the poisoning problem from sulphur or nitrogen compounds, as well as chlorides and other halogens mentioned in section 3.3, and the long cold-upstart times). Because the power sector is not pushing this development as strongly as the automobile sector is pushing the PEMFC development, it may take longer before a cost-effective SOFC is available, or the PEM cells make take over the utility market due to a lower cost achieved through its use in the transportation sector, despite lower conversion efficiency (still above that of conventional steam turbines).

The feeling of a lower urgency for introduction of fuel cells in the power sector may not be warranted, considering the possibility of an increased effort to combat the global warming problem by introduction of renewable energy, which has its primary market aim precisely in the area of power generation. Both coal and nuclear power-production plants are exhibiting unattractive features related to emissions and safety. The scenarios in Chapter 5 point to centralised renewable energy solutions that may be considerably lower in cost than the building-integrated solutions. Fuel cell installations serving to make use of stored hydrogen may achieve a cost advantage due more to balance of system costs than to mass production costs, as the stacks themselves are modular and hence should not exhibit much little scale effect. Further cost advantages of centralised systems may be associated with a lower-cost infrastructure for bringing hydrogen from stores to power plant sites, and with using existing electricity transmission networks, all in comparison with the duplication of many infrastructure components in a multitude of buildings.

7.4.6 Efficiency considerations

In section 6.1, a criticism sometimes forwarded against fuel cells and hydrogen going through energy stores was briefly discussed. The claim is that the efficiency of hydrogen energy systems is unacceptably low when all stages of conversion is included in the considerations. My reply was that use of variable renewable energy sources in any case must involve some part of the energy going from an initial form of electricity through stores based on other energy forms and subsequently back to electricity, with the associated unavoidable losses. In order to put this argument in perspective, Figs. 7.12 and 7.13 show the accumulated efficiencies through all conversion stages from renewable energy leading to the two specific end-uses of personal transportation and electric appliances. For comparison, the corresponding efficiency chains are shown for oil and coal as used in the current energy system. I shall now describe the individual steps involved in these calculations.

Figure 7.12. Stepwise accumulated efficiency of chains of energy conversion from primary sources to end-use, taken here as personal transportation. Current vehicles based on oil products are compared to fuel cell vehicles using methanol from biomass or hydrogen from wind (Sørensen, 2004d; see text for details).

Consider first the routes to personal transportation. For fossil and biomass routes, Fig. 7.12 starts with the production of biomass from solar radiation, at an average efficiency of 0.2% (Sørensen, 2010a). For the petroleum products, the efficiency of the (million year) fossilisation processes of kerogen formation (2%), oil formation in rocks (67%, the rest being natural gas, of which some escapes), and flow into exploitable reservoirs (2.8%) are taken from Dukes (2003). Average extraction efficiencies are estimated as 24.5% and refinery losses as 15%. The combustion efficiency in vehicle Otto or Diesel engines is about 40%, power train efficiency 75%, and the end-use benefits of overcoming air resistance and friction or elevation losses 50% (see section 6.2.4). The overall accumulated fossil fuel car efficiency is thus as low as 2.3 × 10⁻⁸ .

For the methanol route the solar-to-biomass efficiency of 0.2% is followed by a methanol-production efficiency (50%), a DMFC or reformer-PEMFC efficiency of 40%, an electric motor efficiency of 93%, and the same power train and driving efficiencies as for the fossil fuel car.

Finally, for the hydrogen fuel cell car, the solar radiation to wind conversionefficiency is taken as 100% (following arguments of Sørensen, 1996c), the wind turbine efficiency as 35%, the electrolysis efficiency as 80%, the fuel cell conversion efficiency as 55%, and the rest as for methanol. The overall accumulated efficiencies are 1.4 × 10⁻⁴ for the methanol route and 0.054 for the wind-hydrogen route.

Turning now to the routes leading to energy for electronic devices such as computers, consider first the coal route. In Fig. 7.13, following the solar energy to biomass efficiency (0.2%), the successive transformation into peat, highly volatile bituminous coal, and finally hard coal (anthracite) carries efficiencies of 15%, 92.5%, and 63% (Dukes, 2003). A typical efficiency of coal extraction (average of surface and deep mining) is 69%, and that of refining and transport 90%. The steam power plant efficiency is taken as 42%, and the electricity transmission and distribution efficiency as 94%. Finally, an average to high-end end-use efficiency of using electricity (say in a microelectronic device) is taken as 20%.

The first steps of the renewable routes are as for Fig. 7.12 (35% wind turbine), but under half of the power produced is assumed to be used directly and the remaining goes through a hydrogen store (as discussed in the Chapter 5 scenarios), with an assumed average storage cycle efficiency of 75%, covering electrolysis and hydrogen storage/transmission losses, but not electricity regeneration (some 60% for the centralised SOFC route, and 50% electric plus 40% heat efficiency for the decentralised PEM route).

In the centralised case, a transmission and distribution loss of 6% is subtracted, while for the decentralised case, the in-house losses are confined to 1%. The electricity to microelectronic device performance efficiency is still taken as 20%, but a 40% in-house use of waste heat is added in the decentralised case.

Figure 7.13. Stepwise accumulated efficiency of chains of energy conversion from primary sources to end-use, taken here as the operation of electronic equipment by electricity. Current electricity production based on coal is compared to renewable electricity from wind, of which some go through hydrogen stores and are converted back to electricity with and without utilisation of associated heat (Sørensen, 2004d; see text for details).

In the long-term perspective adopted by these two examples, the renewable energy, hydrogen store and fuel cell routes are much more efficient than the current systems.

Another remark on efficiency relates to the way that performance of vehicles is expressed. In daily situations, a car is described as driving a certain number of kilometres per litre of gasoline used, or by how many litres of fuel is required on average for driving 100 km. The problem with such definitions is that in comparing different fuels, one must regard their different energy content per litre. A litre of diesel oil contains about 10% more energy than a litre of gasoline. It is therefore preferable to quote the performance in terms of kilometres driven per unit of energy (say, in MJ) rather than fuel. This was done for the Chapter 4 simulations.

Figure 7.14a, b. Passenger car performance for models available on the Danish market in 2004. Top (a): Conventional rating according to km/MJ on regulatory EU driving cycle. Bottom (b): Rating according to transport work efficiency (km driven times kg payload over MJ of energy used). The vehicles are plotted against payload, taken as the maximum regulatory-permitted addition to proper weight (Sørensen, 2007b).

A further obstacle is met when trying to compare vehicles of different size, and vehicles serving different purposes. Instead of inventing new criteria of efficiency for each category of vehicle, one can get close to a fair comparison across sizes and applications by focussing not on the kilometres driven, but on the transport work performed. Transport work is defined as the product of the mass to be moved and the distance moved. For passenger transport the weight of the people is used, for freight transport the weight of the cargo. In this way, the efficiency of passenger road vehicles can be meaningfully compared to lorry transport of goods, or passenger transport by airplane and good's transport by ship (Fig. 7.15).

Figure 7.14 displays either the kilometres or the transport work that various passenger motorcars can provide per unit of energy, traversing the European regulatory standard driving cycle (the first 1200 seconds of the driving cycle in Fig. 4.7). Using the more appropriate transport work (km × kg), it is seen that cars with little space for passenger of cargo (e.g., Smart) loose in efficiency rating, while cars with room for more people and cargo and still driving fairly economically (e.g., Skoda Octavia diesel version) gain in rating.

Figure 7.15. Typical transport work performance levels (km × kg/MJ) for different types of passenger and freight vehicles for use on roads, tracks, air, or sea, primarily based on US fleets of vintage 2000-2002 (Sørensen, 2007c).

Figure 7.16 looks specifically at freight transport options (expressed as MJ energy used over the product of km driven and tons of payload, the inverse of the ordinate in Figs. 7.14 and 7.15), exhibiting on the left substantial differences in transport work efficiencies between trucking in different European countries, presumably related to long-haul vehicle size and to variations in habits of accepting partial loads. Freight train hauling (data from the US) and international shipping is displayed and seen to be an order of magnitude better in fuel economy than road vehicles. To the right in Fig. 7.16, the highestefficiency ships, trains, and trucks are shown, indicating that for ships and trains, the average is close to the best current efficiencies, but not for trucks. Also to the right, indicative values are added for commercial aircraft and light trucks. These are consistent with the values in Fig. 7.15, and for light trucks poorer than the country averages for (heavy) truck hauls, presumably due to the often very low part loads of delivery vans and trucks.

Figure 7.16. Average specific energy use for delivering transport work with various modes of freight transportation, for different countries, and to the right selected "best technology" values (Sørensen, 2011b, Chapter 12).

7.5 How much time do we have?

The timeframe for the necessary changes in the energy system and thus the urgency of the development efforts related to hydrogen is illustrated by the prospects for continued reliance on oil and expectations of oil price developments. Figure 7.17 shows the decline in finding new exploitable oil fields. The energy contents of new finds topped before 1970, and in contrast to the erratic exploration efforts prior to 1950, the subsequent global efforts have been much more systematic and hence the probability of unexpected large finds in the future is considered low. Enhanced recovery techniques beyond the gas injection method already in widespread use could lead to an upgrading of existing reserve estimates. However, the proposed methods (such as in situ combustion, using bacteria to release oil from rock pores, high-pressure chemistry) do not have universal applicability and can at best release some 20% of the 60% of oil resources presently left trapped in the geological formations (see Giles, 2004). High investment costs are associated with increasing production above the present level, whether using new enhanced recovery methods or exploiting new reservoirs, such as Canadian tar sands or Venezuelan shale oil.

Figure 7.17. Year-by-year global oil discoveries and the results of a model for future finds. Based on data from Longwell (2002) and Campbell (2004a) (Sørensen, 2004e).

Figure 7.18 gives some model results for the future development of oil production and oil prices. On the basis of the data shown in Fig. 7.17 and the remarks on unconventional oil recovery techniques, it is assumed that half of the exploitable reserves have been used by 2010. The uncertainty of this estimate is low (of the order of ±10 years, unless quite unexpected new oil finds materialise) and is supported by many international investigations (Campbell, 2004a; PFC Energy, 2004). However, the fact that we are at the midpoint of reserves does not determine the rate of oil use during the next decades. In principle, the rate can be freely chosen. High use will be followed by a more abrupt decline, and price excursions are likely to increase. Figure 7.18 shows a range of possible developments.

Oil production and consumption could be growing, stay constant, or decline. A decline symmetrical with its historical growth is the outcome of a simple model (Hubbert, 1962), which is highly unlikely because it assumes readily available substitution of oil by other energy forms at similar prices. Constant production will require availability of a partial substitution option, such as liquefaction of coal or conversion of biomass into oil-substituting fuels. These currently have price tags not much over US\$100/bbl of equivalent oil, which would seem acceptable relative to current oil prices (note that Fig. 7.18 uses 2001-US\$). In this case, reserves will be exhausted around 2043 (consistent with the appraisal given in section 5.3.2). As Fig. 7.18 indicates, the future oil price can be expected to fluctuate unpredictably due to a nervous market, economic recession, or growth, as well as geopolitical issues including wars and political unrest in oil-producing countries.

Figure 7.18. The historical global consumption of oil (top) and the historical price levels of crude oil (bottom), supplemented with a spread of possible future behaviours. The figure uses historical consumption data from Fig. 5.10 and historical prices from Enquete Kommission des Deutschen Bundestages (1995), Energy Information Agency of the US DoE (2011) and Sørensen (2004e; 2011b).

The upper level of oil consumption, despite dwindling resources, would be the result of not seeing any substitution options as viable, together with a rising demand, say, in growth countries such as China. If China increases its oil use by 25% until 2030 and there is a 30% increase in world car ownership over the next 25 years (hardly extravagant expectations), then such a scenario could become realised. The reserves would is this case last only to around year 2038. A recent IEA report (IEA, 2004) has an even higher 60% increase in fossil energy use to 2030 in its reference scenario. The implication of such expectations is also a 60% increase in CO2 emissions. The increased production will take place entirely in the OPEC countries, because non-OPEC oil production is predicted to decline during the period, according to PFC Energy (2004).

The geopolitical and supply-security implications of the latter development are huge. Both IEA and the PFC study recommend energy policies reducing demand and speeding up a transition away from oil dependence. They differ in that IEA believes that increasing the oil production sufficiently is possible at least to 2030, while PFC predicts a demand-supply gap already during the period 2014-2020, for a corresponding global demand growth varying from 2.4 to 1.1% per year. A recent study by the oil industry supports the PFC view (ExxonMobile, 2004), while the analysis in Fig. 7.18 would also be consistent with IEA, when year 2038 is set as the limit for covering demand by increased oil production (at any price).

In all scenarios, the average price of oil is predicted to go up substantially. In the high production scenarios, the sudden onset of decline in supply is likely preceded by crisis conditions that could have been avoided by a planned winding-down of production. Short-term behaviour of the oil price may exhibit even larger fluctuations in this nervous and highly politicised market than have been seen so far. However, the solidity of the reasoning behind expecting a phase-out of conventional cheaply produced oil is high.

The preceding analysis provides a clear indication of the importance of making viable energy alternatives, including if possible hydrogen technologies, available as soon as possible, and emphasises the urgency of stepping up R&D in order to achieve the required technical maturity and economic acceptability levels of alternatives to oil.

At the same time it is clear that hydrogen technologies cannot reach a penetration in the marketplace large enough to solve the near-term problems. Other solutions are required for the short term, and as underlined by investigations such as the one made for passenger cars in section 6.2.4, efficient use of energy is the only technology that is ready and can be implemented in time (as also recommended by Sørensen, 2010a; IEA, 2004; PFC Energy, 2004). Hydrogen and fuel cell technologies may be the solution for the longterm energy transition, but short-term oil supply problems can only be handled by the improving energy efficiency.

7.5 The end, and a beginning

It is time to bring this book to an end, but not without a few concluding remarks signalling the need to create the conditions for a new beginning, by developing the partnership between research and industry and between voters and parliamentarians in a direction that will make the conditions for a hydrogen society realistic. The motivation is the fact that few other alternatives are available, and thus access to the hydrogen route may help preserve the progression in wealth for both the rich and the presently poor.

The success of the hydrogen project is conditional on a number of things falling in place, notably solving the technical and economic problems of fuel cells. There are plenty of drivers to spur us in this direction: peaking and then declining fossil fuel production, political instability of several significant supplier countries and the global quest for a transition to a sustainable energy system that most voters like to see based upon renewable energy sources. The hydrogen technologies have already brought a large number of new players into the field. But numbers do not guarantee success, and there are many "conventional" perceptions that may function as barriers. Do hydrogen and fuel cell systems have to match the current low prices of conventional systems, without consideration of externality costs? I do not think that is possible, not for hydrogen and not for several among the primary sources of energy that may replace fossil fuels. This has further implications, for example, for the transportation sector.

Should we continue to encourage (e.g., by accepting advertising) driving in oversized cars or one person in a four-wheel-drive inflated van (going by the name SUV = special utility vehicle; the one day a year we may need to drive off-road we could rent a four-wheel-driven car), or should we rather go for a super-efficient small car (see the assessment made in section 6.2.4)? The first option would more than double the energy demand in existing industrialised societies, plus add the additional demand from developing economies. The second option could reduce existing demand by a factor of three, leaving more room for newcomers to the industrialised world. Some national fuel cell programmes appear to be construed in order to push the automobile problem 25 years into the future by passively waiting for a massive penetration of fuel cell vehicles in the marketplace, rather than use the already available low or no extra cost energy efficient car option capable of reducing the oil supply problem here and now by a factor of three or more.

The energy efficient automobile is also a perfect starting point for subsequent introduction of fuel cell drives, where one would no longer need to put a 65- 100 kW fuel cell in to get a decent performance but could do with a smaller hybrid vehicle and still get acceptable range (see section 4.1.3). In any case, future cars must have an electronic control system not only for optimum performance, but also for avoiding collisions and preventing speeding. Energy is not the only problem of the current personal transportation system, so solutions that solve more than one problem should be considered, and the possibly higher cost would then have a better chance to become accepted by customers.

One thing is clear to me: The ideal energy future is a world without combustion. Biofuels or hydrogen use in steam power plants may be an intermediate solution that we will have to live with for a small while, but in the long run, energy conversion should proceed without flames and emissions.

All this requires a new attitude to shaping our way of life. We want more self-control, and technologies such as the reversible fuel cells offer us just that. But being more in charge ourselves also implies a greater responsibility for how it all affects other people, which again should make us think of changing the priorities of what we want to spend money on. What do we do with people willing to pay more for a car capable of going three times the maximum speed limit, but unwilling to pay just a little bit more for an energy-efficient car? There must be a balance between increased freedom of choice and setting common rules for society to work decently. "Freedom of choice" often just means that the best advertiser wins, and the muchdevalued concept of regulation may be unequivocally needed to make society an acceptable place to be for every member of it. Individualism is often to the detriment of the common goods or−environmentally speaking−of the commons.

The law locks up the man or woman who steals the goose from off the common but lets the greater villain loose who steals the common from the goose

(English rhyme quoted from McMichael, 2001).

Hydrogen and Fuel Cells Emerging technologies and applications

Other books by the author:

A History of Energy. Northern Europe from the Stone Age to the Present Day, 2011

Life-Cycle Analysis of Energy Systems: From Methodology to Applications, 2011.

Renewable Energy Reference Book Set (ed., 4 volumes of reprints), 2010

Renewable Energy −physics, engineering, environmental impacts, economics and planning, 4th edition, 2010 (Previous editions 1979, 2000 and 2004)

Renewable Energy Focus Handbook (with Breeze, Storvick, Yang, Rosa, Gupta, Doble, Maegaard, Pistoia and Kalogirou), 2009

Renewable Energy Conversion, Transmission and Storage, 2007

Hydrogen and Fuel Cells, 1st ed., 2005

Life-cycle analysis of energy systems (with Kuemmel and Nielsen), 1997

Blegdamsvej 17, 1989

Superstrenge, 1987

Fred og frihed, 1985

Fundamentals of Energy Storage (with Jensen), 1984

Energi for fremtiden (with Hvelplund, Illum, Jensen, Meyer and Nørgård), 1983

Energikriser og Udviklingsperspektiver (with Danielsen), 1983

Skitse til alternativ energiplan for Danmark (with Blegaa, Hvelplund, Jensen, Josephsen, Linderoth, Meyer and Balling), 1976

*Music across times and fences*−a book project in progress

More information on the author's work is available at http://energy.ruc.dk

Hydrogen and Fuel Cells

Emerging technologies and applications

Second Edition

Bent Sørensen Roskilde University Institute of Environmental, Social and Spatial Change

Academic Press is an imprint of Elsevier The Boulevard, Langford Lane, Kidlington, Oxford, OX5 1GB, UK 30 Corporate Drive, Suite 400, Burlington, MA 01803, USA

No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher's permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions.

This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein).

Notices

Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary.

Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility.

To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein.

Library of Congress Cataloging-in-Publication Data

Sørensen, Bent, 1941-

Hydrogen and fuel cells: emerging technologies and applications hydrogen and fuel cells emerging technologies and applications / Bent Sørensen. -- 2nd ed.

p. cm.

ISBN: 978-0-12-387709-3

Hydrogen as fuel. 2. Fuel cells. I. Title.

TP359.H8S68 2011 621.31'2429--dc23

2011038769

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library.

For information on all Academic Press publications visit our website at www.elsevierdirect.com.

Typeset by author.

Printed in the United States of America

11 12 13 14 9 8 7 6 5 4 3 2 1

Preface

When the first edition of this book was written, there was an enthusiastic belief in the fuel cell and automobile industry that hydrogen vehicles would start to penetrate the market already in the first decade of the 21st century. This did not happen−for several reasons. The target specifications of 5-year fuel cell life and a cost under US\$10 000 for an array with 1 kW average production were not too ambitious but still proved unreachable in the short term. A 5-year fuel cell lifetime is really on the low side, as cars currently approach operational lifetimes of 20 years, spurred by concerns for environmental impact during manufacture and resource stewardship. The 5-year fuel cell would have to be replaced three times during the lifetime of the vehicle, greatly augmenting the real cost, considering both equipment and replacement work. Another factor in the present market evaluation is the strength of the competition, offered by electric cars, hybrids combining batteries with oil-based fuels, and new generations of biofuels for the transportation sector. A careful appraisal of the merits of these alternatives relative to a hydrogen-fuelled transportation sector is undertaken in this new edition, along with a similar assessment of other areas of fuel cell uses.

For the researchers and manufacturers of fuel cells and related equipment, the temporal setback is not uniquely negative, as it provides an opportunity for rethinking the basic design concepts and developing several new ideas that perhaps were left unprocessed during the feverish rush to reach the early goals. Fortunately, the relative absence of hydrogen-vehicle stories from the news headlines has not slowed down the pace of science and engineering efforts, and this edition has many new, exciting advances to present and discuss. This goes both for the basic understanding of electrochemical processes inside the cell structures, and for the development of vehicle and system concepts, with a promise to overcome the many challenges posed by the introduction of new technology that in several ways is departing from the traditional operational behaviour and existing infrastructure layout.

This edition has expanded the coverage of hybrid systems, considering the advantages of combining battery and fuel cell technologies, rather than seeing them as mutually exclusive competitors. It is demonstrated that suitable combinations of either plug-in hybrids between fuel cell and advanced battery use, or stand-alone concepts, can achieve performance and economic viability not obtainable for pure fuel cell or pure battery vehicles. The fuel cell R&D community has attracted some of the most skilled scientists and engineers, and it is those people I hope to inspire further by this volume.

Preface to first edition

These years, many scientists and engineers move into the field of hydrogen and fuel cells because it is exciting and well funded. The aim is to transform the way energy is delivered and used over the coming years, with major changes in technologies for production, distribution and conversion, as a response to political instability of many oil-producing countries, uncertainty about resources and increasing concerns over environmental effects.

This book is written to the many newcomers to the field, to students at the increasing number of courses given in the subject and to well-established scientists and developers who already have in-depth knowledge in certain sub-fields but like to keep informed about the entire field from technology to policy considerations, economical and environmental assessment. I aim to provide an introduction to people with general science background, but no special hydrogen and fuel cell experience, and also to give an up-to-date account of the frontiers of research and development to readers who need to be able to connect the emerging terminology to the concepts of conventional fields of science.

At the end of each chapter are problems and discussion topics, several of which can be used for problem-oriented mini-projects.

In fact, the pace of knowledge development in the field of hydrogen and fuel cells is so rapid that about half the content of this book is based on material less than a year old (as of the time of writing) and often not even published or found in the "in press" basket of the scientific journal publishers. It has been a pleasure to work with this extremely fresh material from the desks of my colleagues. New techniques have enabled specific investigations to be performed along with the book writing that would have been impossible 5-10 years ago.

To achieve the aim stated above I try to avoid specialist jargon or, if it is important, to define and explain the special terminology that the reader will meet in the latest scientific journals, providing the connection back to the concepts familiar to people with a physics, chemistry or biology background. The policy planner and assessor will similarly find the newest ideas and methods in these fields linked back to conventional economy and to environmental and planning sciences.

First of all, I want to convey the richness of the hydrogen and fuel cell fields and to present the challenges calling for a dedicated effort by the cream of human ingenuity, which is you, dear reader.

Bent Sørensen Gilleleje, October 2004 boson@ruc.dk

Units and conversion factors

Powers of 10¤

Prefix	Symbol	Value	Prefix	Symbol	Value
atto	a	10-18	kilo	k	103
femto	f	10-15	mega	M	106
pico	p	10-12	giga	G	109
nano	n	10-9	tera	T	1012
micro	µ	10-6	peta	P	1015
milli	m	10-3	exa	E	1018

SI units

Basic unit	Name	Symbol
length	metre	m
mass	kilogram	kg
time	second	s
electric current	ampere	A
temperature	Kelvin	K
luminous intensity	candela	cd
plane angle	radian	rad
solid angle	steradian	sr
amount#	mole	mol

Derived unit	Name	Symbol	Definition
energy	joule	J	kg m2 s-2
power	watt	W	J s-1
force	newton	N	J m-1
electric charge	coulomb	C	A s
potential difference	volt	V	J A-1 s-1
pressure	pascal	Pa	N m-2
electric resistance	ohm	Ω	V A-1
electric capacitance	farad	F	A s V-1
magnetic flux	weber	Wb	V s
inductance	henry	H	V s A-1
magnetic flux density	tesla	T	V s m-2
luminous flux	lumen	lm	cd sr
illumination	lux	lx	cd sr m-2
frequency	hertz	Hz	cycle s-1

^¤ G, T, P, E are called milliard, billion, billiard, trillion in Europe, but billion, trillion, quadrillion, quintillion in the USA. M as million is universal.

The amount containing as many particles as there are atoms in 0.012 kg 12C.

Conversion factors

Type	Name	Symbol	Approximate value
energy	electon volt	eV	1.6021 × 10-19 J
energy	erg	erg	10-7 J (exact)
energy	calorie (thermochemical)	cal	4.184 J
energy	British thermal unit	Btu	1055.06 J
energy	Q	Q	1018 Btu (exact)
energy	quad	q	1015 Btu (exact)
energy	tons oil equivalent	toe	4.19 × 1010 J
energy	barrels oil equivalent	bbl	5.74 × 109 J
energy	tons coal equivalent	tce	2.93 × 1010 J
energy	m3 of natural gas		3.4 × 107 J
energy	kg of methane		6.13 × 107 J
energy	m3 of biogas		2.3 × 107 J
energy	litre of gasoline		3.29 × 107 J
energy	kg of gasoline		4.38 × 107 J
energy	litre of diesel oil		3.59 × 107 J
energy	kg of diesel oil/gasoil		4.27 × 107 J
energy	m3 of hydrogen at 1 atm		1.0 × 107 J
energy	kg of hydrogen		1.2 × 108 J
energy	kilowatthour	kWh	3.6 × 106 J
power	horsepower	hp	745.7 W
power	kWh per year	kWh/y	0.114 W
radioactivity	curie	Ci	3.7 × 108 s-1
radioactivity	becqerel	Bq	1 s-1
radiation dose	rad	rad	10-2 J kg-1
radiation dose	gray	Gy	J kg-1
dose equivalent	rem	rem	10-2 J kg-1
dose equivalent	sievert	Sv	J kg-1
temperature	degree Celsius	°C	K — 273.15
temperature	degree Fahrenheit	°F	9/5 C+ 32
time	minute	min	60 s (exact)
time	hour	h	3600 s (exact)
time	year	y	8760 h

Type	Name	Symbol	Approximate value
pressure	atmosphere	atm	1.013 × 105 Pa
pressure	bar	bar	105 Pa
pressure	pounds per square inch	psi	6890 Pa
mass	ton (metric)	t	103 kg
mass	pound	lb	0.4536 kg
mass	ounce	oz	0.02835 kg
length	Ångström	Å	10-10 m
length	inch	in	0.0254 m
length	foot	ft	0.3048 m
length	mile (statute)	mi	1609 m
volume	litre	l	10-3 m3
volume	gallon (US)		3.785 × 10-3 m3

REFERENCES

Abashar, M. (2004). Coupling of steam and dry reforming of methane in catalytic fluidized bed membrane reactors. Int. J. Hydrogen Energy 29, 799-808.
Abe, A., Nakamura, M., Sato, I., Uetani, H., Fujitani, T. (1998). Studies of the largescale sea transportation of liquid hydrogen. Int. J. Hydrogen Energy 23, 115-121.
Adamo, C., Barone, V. (2002). Physically motivated density functionals with improved performances: the modified Perdew-Burke-Ernzerhof model. J. Cham. Phys., 116, 5933-5940.
Adamson, K-A., Pearson, P. (2000). Hydrogen and methanol: a comparison of safety, economics, efficiencies and emissions. J. Power Sources 86, 548-555.
Agranat, V., Tchouvelev, A. (2004). CFD modelling of gas-liquid flows in water electrolysis units. In Proc. 15th World Hydrogen Energy Conf., Yokohama. 28C-05, CD Rom, Hydrogen Energy Soc. Japan.
Ahluwalia, R., Hua, T., Peng, J-K., Lasher, S., McKenney, K., Sinha, J., Gardiner, M. (2010). Technical assessment of cryo-compressed hydrogen storage tank systems for automotive applications. Int. J. Hydrogen Energy 35, 4171-4184.
Ahluwalia, R., Wang, X., Rousseau, A., Kumar, R. (2004). Fuel economy of hydrogen fuel cell vehicles. J. Power Sources 130, 192-201.
Ahmed, I., Gupta, A. (2009). Hydrogen production from polystyrene pyrolysis and gasification: Characteristics and kinetics. Int. J. Hydrogen Energy 34, 6253-6264.
Ahn, S-Y., Shin, S-J., Ha, H., Hong, S-A., Lee, Y-C, Lim, T., Oh, I-H. (2002). Performance and lifetime analysis of the kW-class PEMFC stack. J. Power Sources 106, 295-303.
Aindow, T., Haug, A., Jayne, D. (2011). Platinum catalyst degradation in phosphoric acid fuel cells for stationary applications. J. Power Sources 196, 4506-4514.
Akansu, S., Dulger, Z., Kahraman, N., Veziroglu, T. (2004). Internal combustion engines fueled by natural has-hydrogen mixtures. Int. J. Hydrogen Energy 29, 1527- 1539.
Aki, H., Yamamoto, S., Kondoh, J., Maeda, T., Yamaguchi, H., Murata, A., Ishii, I., Sugimoto, I. (2004). Fuel cells and hydrogen energy networks in urban residential buildings. In Proc. 15th World Hydrogen Energy Conf., Yokohama. 28I-05, CD Rom, Hydrogen Energy Soc. Japan.
Akkerman, I., Janssen, M., Rocha, J., Wijffels, R. (2002). Photobiological hydrogen production: photochemical efficiency and bioreactor design. Int. J. Hydrogen Energy, 27, 1195-1208. The article is difficult to read because of editing errors.
Albertus, P., Couts, J., Srinivasan, V., Newman, J. (2008). II. A combined model for determining capacity usage and battery size for hybrid and plug-in hybrid electric vehicles. J. Power Sources 183, 771-782.
Alcaide, F., Brillas, E., Cabot, P-L. (2004). Limiting behaviour during the hydroperoxide ion generation in a flow alkaline fuel cell. J. Electroanalytical Chem. 566, 235- 240.
Al-Durra, A., Yurkovich, S., Guezennec, Y. (2010). Study of nonlinear control schemes for an automotive traction PEM fuel cell system. Int. J. Hydrogen Energy 35, 11291-11307.
Ally, J., Pryor, T. (2007). Life-cycle assessment of diesel, natural gas and hydrogen
fuel cell bus transportation systems. J. Power Sources 170, 401-411.
Altmann, M., Weindorf, W., Wurster, R., Mostad, H., Weinberger, M., Filip, G. (2004). FCSHIP: environmental impacts and costs of hydrogen, natural gas and conventional fuels for fuel cell ships. In Proc. 15th World Hydrogen Energy Conf., Yokohama. 30A-05, CD Rom, Hydrogen Energy Soc. Japan.
Amos, W. (1998). Cost of storing and transporting hydrogen. Internal Report, Nat. Renewable Energy Lab., Golden, CO.
An, W., Gatewood, D., Dunpal, B., Turner, C. (2011). Catalytic activity of bimetallic nickel alloys for solid-oxide fuel cell anode reactions from density-functional theory. J. Power Sources 196, 4724-4728.
Andreassen, K. (1998). Hydrogen production by electrolysis,. In "Hydrogen Power: Theoretical And Engineering Solutions" (Sætre, T., ed.), p. 91. Kluwer, Dordrecht.
Andrés, M-B., Boyd, T., Grace, J., Lim, C., Gulamhusein, A., Wan, B., Kurokawa, H., Shirasaki, Y. (2011). In-situ CO² capture in a pilot-scale fluidized-bed membrane reformer for ultra-pure hydrogen production. Int. J. Hydrogen Energy 36, 4038- 4055.
Andress, D., Nguyen, T., Das, S. (2011). Reducing GHG emissions in the United States' transportation sector. Energy for Sustainable Development, doi:10.1016/ j.esd.2011.03.002
Angrist, S. (1976). "Direct Energy Conversion", 3rd ed. Allyn and Bacon, Boston.
Anonymous (2008). First fuel cell passenger ship unveiled in Hamburg. Fuel Cell Bulletin, October, 4-5.
Antonkine, M., Jordan, P., Fromme, P., Krauss, N., Golbeck, J., Stehlik, D. (2003). Assembly of protein subunits within the stromal ridge of photosystem I: structural changes between unbound and sequentially PS I-bound popypeptides and correlated changes of the magnetic properties of the terminal iron sulphur clusters. J. Mol. Biol. 327, 671-697.
Arora, A., Medora, N., Livernois, T., Swart, J. (2010). Safety of Lithium-ion batteries for hybrid electric vehicles. Ch. 18 in Electric and Hybrid Vehicles (G. Pistoia, ed.), 463-491. Elsevier, Amsterdam.
Aroutiounian, V., Arakelyan, V., Shahnazaryan, G. (2005). Metal oxide photoelectrodes for hydrogen generation using solar radiation-driven water splitting. Solar Energy 78, 581-592.
Arroyo, M. y de Dompablo, Ceder, G. (2004). First principles investigations of complex hydrides AMH⁴ and A3MH⁶ (A = Li, Na, K, M = B, Al, Ga) as hydrogen storage systems. J. Alloys Compounds 364, 6-12.
Arsalis, A., Nielsen, M., Kær, S. (2011). Modeling and parametric study of a 1 kWe HT-PEMFC-based residential micro-CHP system. Int. J. Hydrogen Energy 36, 5010-5020.
Asada, Y., Koike, Y., Schnackenberg, J., Miyake, M., Uemura, I., Miyake, J. (2000). Heterologous expression of clostidial hydrogenase in the cyanobacterium Synechococcus PCC7942. Biochim. Biophys. Acta 1490, 269-278.
Asakuma, Y., Miyauchi, S., Yamamoto, T., Aoki, H., Miura, T. (2004). Homogenization method for effective thermal conductivity of metal hydride bed. Int. J. Hydrogen Energy 29, 209-216.
Au, S., Hemmes, K., Woudstra, N. (2003). Flowsheet calculation of a combined heat and power fuel cell plant with a conceptual molten carbonate fuel cell with
separate CO² supply. J. Power Sources 122, 19-27.
Ayabe, S., Omoto, H., Utaka, T., Kikuchi, R., Sasaki, K., Teraoka, Y., Eguchi, K. (2003). Catalytic autothermal reforming of methane and propane over supported metal catalysts. Appl. Catalysis A: General 241, 261-269.
Ayad, M., Becherif, M., Henni, A. (2010). Vehicle hybridization with fuel cell, supercapacitors and batteries by sliding mode control. Renewable Energy, doi:10.1016/ j.renene.2010.06.012
Azam, F., Worden, A. (2004). Microbes, molecules and marine ecosystems. Science 303, 1622-1623.
Badsberg, U., Jørgensen, A., Gasmar, H., Led, J., Hammerstad, J., Jespersen, L., Ulstrup, J. (1996). Solution structure of reduced plastocyanin from the bluegreen alga Anabena variabilis. Biochemistry 35, 7021-7031.
Bae, S., Kim, S-J, Park, J., Lee, J-H., Cho, H., Park, J-Y. (2010). Lifetime prediction through accelerated degradation testing of membrane electrode assemblies in direct methanol fuel cells. Int. J. Hydrogen Energy 35, 9166-9176.
Bahatyrova, S., Frese, R., Siebert, C., Olsen, J., Werf, K. van der, Grondelle, R. van, Niederman, R., Bullough, P., Otto, C, Hunter, C. (2004). The native architecture of a photosynthetic membrane. Nature 430, 1058-1062.
Bain, A., Vorst, W. van (1999). The Hindenburg tragedy revisited: the fatal flaw found. Int. J. Hydrogen Energy 24, 399-403.
Balachandran, U., Lee, T., Wang, S., Dorris, S. (2004). Use of mixed conducting membranes to produce hydrogen by water dissociation. Int. J. Hydrogen Energy 29, 291-296.
Bao, D. (2001). A panoramic review of hydrogen energy activity in China. In "Hydrogen Energy Progress XIII, Proc. 13th World Energy Conf., Beijing 2000" (Mao, Z., Veziroglu, T., eds.), pp. 181-185. Int. Assoc. Hydrogen Energy, Beijing.
Baptista, P., Tomás, M., Silva, C. (2010). Plug-in hybrid fuel cell vehicles market penetration scenarios. Int. J. Hydrogen Energy 35, 10024-10030.
Barber, J. (2002). Photosystem II: a multisubunit membrane protein that oxidises water. Current Opinion Structural Biology 12, 523-530.
Barbi, V., Funari, S., Gehrke, R., Scharnagl, N., Stribeck, N. (2003). Nanostructure of Nafion membrane material as a function of mechanical load studied by SAXS. Polymer 44, 4853-4861.
Barbir, F. (2003). System design for stationary power generation. In "Handbook of Fuel Cells, Vol. 4" (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 51. Wiley, Chichester.
Barbir, F., Molter, T., Dalton, L. (2005). Efficiency and weight trade-off analysis of regenerative fuel cells as energy storage for aerospace applications. Int. J. Hydrogen Energy 30, 351-357.
Bard, A., Faulkner, L. (1998). "Electrochemical methods. 2nd ed." Wiley, New York.
Barelli, L., Bidini, G., Gallorini, F., Ottaviano, A. (2011). An energeticeexergetic comparison between PEMFC and SOFC-based micro-CHP systems. Int. J. Hydrogen Energy 36, 3206-3214.
Barone, V. (1996). Chapter in "Recent Advances In Density Functional Methods, Part I" (Chong, D., ed.). World Scientific Publ., Singapore.
Basile, A., Paturzo, L., Laganà, F. (2001). The partial oxidation of methane to syngas in a palladium membrane reactor: simulation and experimental studies. Catalysis Today 67, 65-75.
Batra, V., Maudgal, S., Bali, S., Tewari, P. (2002). Development of aplha lithium aluminate matrix for molten carbonate fuel cell. J. Power Sources 112, 322-325.
Bauer, F., Willert-Porada, M. (2004). Microstructural charaxterization of Zrphosphate-Nafion membranes for direct methanol fuel cell (DMFC) application. J. Membrane Science 233, 141-149.
Bauernschmitt, R., Ahlrichs, R. (1996). Treatment of electronic excitations within the adiabatic approximation of time dependent density functional theory. Chem. Phys. Lett. 256, 454-464.
Bazylev, N., Fomin, N., Galiano, H., Meleeva, O., Martemianov, S., Penyazkov, O. (2011). PEMFCs flow microstructure analysis by advanced speckle technologies. Int. J. Heat & Mass Transfer 54, 2341-2348.
Beausoleil-Morrison, I. (2010). The empirical validation of a model for simulating the thermal and electrical performance of fuel cell micro-cogeneration devices. J. Power Sources 195, 1416-1426.
Becke, A. (1993). Density-functional thermochemistry. III: the role of exact exchange. J. Chem. Phys. 98, 5648.
Béja, O., Aravind, L., Koonin, E., Suzuki, M., Hadd, A., Nguyen, L., Jovanovich, S., Gates, C., Feldman, R., Spudich, J., Spudich, E., DeLong, E. (2000). Bacterial Phodopsin: evidence for a new type of phototrophy in the sea. Science 289, 1902- 1906.
Bell, B. (1998). Looking beyond the internal combustion engine: the promises of methanol fuel cell vehicles. Paper presented at "Fuel Cell Technology Conference, London, September". IQPC Ltd., London.
Berman-Frank, I., Lundgren, P., Chen, Y., Küpper, H., Kolber, Z., Bergman, B., Falkowski, P. (2001). Segregation of nitrogen fixation and oxygenic photosynthesis in the marine cyanobacterium Trichodesmium. Science 294, 1534-1537.
Bernard, J., Delprat, S., Guerra, T., Büchi, F. (2010). Fuel efficient power management strategy for fuel cell hybrid power trains. Control Eng. Practice 18, 408-417.
Bernard, J., Hofer, M., Hannesen, U., Toth, A., Tsukada, A., Büchi, F., Dietrich, P. (2011). Fuel cell/battery passive hybrid power source for electric powertrains. J. Power Sources 196, 5867-5872.
Berning, T., Odgaard, M., Kær, S. (2011). Water Balance Simulations of a PEM Fuel Cell Using a Two-Fluid Model. J. Power Sources, doi:10.1016/j.jpowsour.2011.03. 068.
Berthold, O., Bünger, U., Niebauer, P., Schindler, P., Schurig, V., Weindorf, W. (1999). Analyse von Einsatzmöglichkeiten und Rahmenbedingungen von Brennstoffzellen in Haushalten and im Kleinverbrauch in Deutschland und Berlin. Ludwig-Bölkow Systemtechnik GmbH, Ottobrun.
Besser, R. (2011). Thermal integration of a cylindrically symmetric methanol fuel processor for portable fuel cell power. Int. J. Hydrogen Energy 36, 276-283.
Bielmann, M., Vogt, U., Zimmermann, M., Züttel, A. (2011). Seasonal energy storage system based on hydrogen for self sufficient living. J. Power Sources 196, 4054- 4060.
Bird, R., Stewart, W., Lightfoot, E. (2001). "Transport phenomena", 2nd ed., John Wiley & Sons, New York.
Bischoff, M., Farooque, M., Satou, S., Torazza, A. (2003). MCFC fuel cell systems. In "Handbook of Fuel Cells, vol. 4" (Vielstich, W., Lamm, A., Gasteiger, H., eds.). Ch. 92. Wiley, Chichester.
Bitsche, O., Gutmann, G. (2004). Systems for hybrid cars. J. Power Sources 127, 8-15. BMW (2004). Model 745h, 750hL. Website: http://www.bmwworld.com/models
Bockris, J. (1975). "Energy: The Solar−hydrogen Alternative". Australia and New Zealand Book Co., Brookvale, Australia.
Bockris, J. (1972). A hydrogen economy. Science 176, 1323.
Bockris, J., Despic, A. (2004). Principles of physical science: fields. "Encyclopædia Brittannica Library", CDROM Deluxe Ed., London.
Bockris, J., Reddy, A. (1998). "Modern Electrochemistry", 2nd ed., Vol. 1 (and Vol. 2B, 2000). Plenum Press, New York.
Bockris, J., Reddy, A., Gamboa-Aldeco, M. (2000). "Modern Electrochemistry", 2nd ed., Vol. 2A. Plenum Press, New York.
Bockris, J., Shrinivasan, S. (1969). "Fuel Cells: Their Electrochemistry". McGraw−Hill, New York.
Boettner, D., Moran, M. (2004). Proton exchange membrane (PEM) fuel cell-powered vehicle performance using direct-hydrogen fueling and on-board methanol reforming. Energy 29, 2317-2330.
Bogdanovic, B., Brand, R., Marjanovic, A., Schwickardi, M., Tölle, J. (2000). Metaldoped sodium aluminium hydrides as potential new hydrogen storage materials. J. Alloys Compounds 302, 36-58.
Bogdanovic, B., Schwickardi, M. (1997). Ti-doped alkali metal aluminium hydrides as potential novel reversible hydrogen storage materials. J. Alloys Compounds 253-204, 1-9.
Bolton, J. (1996). Solar photoproduction of hydrogen: a review. Solar Energy 57, 37-50. Borenstein, S. (2008). The Market Value and Cost of Solar Photovoltaic Electricity Production. Center for the Study of Energy Markets, Report WP 176, Berkeley CA. http://www.ucei.org
Borglum, B. (2003). From cells to systems: Global Thermoelectric's critical path approach to planar SOFC development. Presentation at "8th Grove Fuel Cell Symposium 2003", http://www.globalte.com.
Borgwardt, R. (1998). Methanol production from biomass and natural gas as transportation fuel. Industrial Eng. Chem. Res. 37, 3760-3767.
Bossche, P. v. d., Vergels, F., Mierlo, J., Matheys, J., Autenboer, W. (2006). SUBAT: An assessment of sustainable battery technology. J. Power Sources 162, 913-919.
Bossel, U. (2004). Hydrogen: why its future in a sustainable energy economy will be bleak, not bright. Renewable Energy World 7, No. 2, 155-159.
Bowman, C., Arthur, E., Heighway, E., Lisowski, P., Venneri, F., Wender, S. (1994). Accelerator-driven transmutation technology, Vol. 1, pp. 1-11. Los Alamos Laboratory, report LALP-94-59.
Boysen, D., Uda, T., Chisholm, C., Haile, S. (2004). High-performance solid acid fuel cells through humidity stabilization. Science 303, 68-70.
Bradley, T., Moffitt, B., Mavris, D., Parekh, D. (2007). Development and experimental characterization of a fuel cell powered aircraft. J. Power Sources 171, 793-801.
Brehm, N., Mayinger, F. (1989). A contribution to the phenomenon of the transition from deflagration to detonation. VDI−Forschungsheft No. 653/1989, pp. 1- 36. (website: http://www.thermo−a.mw.tu−muenchen.de/lehrstuhl/foschung /eder_gerlach.html).
Breitung, W., Bielen, U., Necker, G., Veser, A., Wetzel, F-J., Pehr, K. (2000). Numerical simulation and safety evaluation of tunnel accidents with a hydrogen pow-
ered vehicle. In "Proc. 13th World Energy Conf., Beijing 2000" (Mao, Z. and Veziroglu, T., eds.), pp. 1175-1181. Int. Assoc. Hydrogen Energy, Beijing.
Brettel, K., Leibl, W. (2001). Electron transfer in photosystem I. Biochim. Biophys. Acta, 1507, 100-114.
Brown, S. (1998). The automakers' big-time bet on fuel cells. Fortune Mag., 30 March, 12 pages (http://www.pathfinder.com/fortune/1998/980330).
Bruggeman, D. (1935). Ann. Phys. 24, 636.
Bubna, P., Brunner, D., Advani, S., Prasad, A. (2010a). Prediction-based optimal power management in a fuel cell/battery plug-in hybrid vehicle. J. Power Sources 195, 6699-6708.
Bubna, P., Brunner, D., Gangloff, J., Advani, S., Prasad, A. (2010b). Analysis, operation and maintenance of a fuel cell/battery series-hybrid bus for urban transit applications. J. Power Sources 195, 3939-3949.
Buchmann, I. (1998). Understanding your batteries in a portable world. Cadex Inc., Canada, http://www.cadex.com/cfm.
Buono, S., Rubbia, C. (1996). Simulation of total loss of power accident in the energy amplifier. CERN/ET internal note 96-015, 10 pp.
Butler, D. (2004). Nuclear power's new dawn. Nature 429, 238-240.
Cai, W., Li, S., Feng, L., Zhang, J., Song, D., Xing, W., Liu, C. (2011). Transient behavior analysis of a new designed passive direct methanol fuel cell fed with highly concentrated methanol. J. Power Sources 196, 3781-3789.
Callen, H. (1960). "Thermodynamics", John Wiley & Sons, New York.
Camara-Artigas, A., Williams, J., Allen, J. (2001). Structure of cytochrome c² from Rhodospirillum centenum. Acta Cryst., Biol. Cryst. D57, 1498-1505.
Campanari, S., Iora, P. (2004). Definition and sensitivity analysis of a finite volume SOFC model for a tubular cell geometry. J. Power Sources 132, 113-126.
Campbell, C. (2004a). Oil and gas liquids 2004 scenario. http://www.peakoil.net/ uhdsg/Default.htm (update of "ASPO Statistical Review of Oil and Gas, 2002" (Aleklett, K., Bentlay, R., Campbell, C., eds.).
Campbell, D. (2004b). Fuel cells international: PEM perspective. Oral presentation at Proc. 15th World Hydrogen Energy Conf., Yokohama.
Carcadea, E., Ene, H., Ingham, D., Lazar, R., Ma, L., Pourkashanian, M., Stefanescu, I. (2007). A computational fluid dynamic analysis of a PEM fuel cell system for power generation. Int. J. Numerical Methods for Heat & Fluid Flow 17 (3), 302-312.
Carcassi, M., Cerchiara, G., Marangon, A. (2004). Experimental studies of gas vented explosion in real type environment. In "Hydrogen Power — Theoretical and Engineering Solutions, Proc. Hypothesis V, Porto Conte 2003" (Marini, M., Spazzafumo, G., eds.), pp. 589-597. Servizi Grafici Editoriali, Padova.
Carlson, E., Kopf, P., Sinha, J., Sriramulu, S., Yang, Y. (2005). Cost Analysis of PEM Fuel Cell Systems for Transportation. US National Renewable Energy Laboratory Report NREL/SR-560-39104, Golden CO.
Casida, M., Casida, K., Jamorski, C., Salahub, D. (1998). J. Chem. Phys. 108, 4439.
Ceder, G., Chiang, Y−M., Sadoway, D., Aydinol, M., Jang, Y−I., Huang, B. (1998). Identification of cathode materials for lithium batteries guided by first−principles calculations. Nature 392, 694-696.
Center for the Evaluation of Risks to Human Reproduction (2004). NTP-CERHR expert panel report on the reproductive and developmental toxicity of methanol. Reproductive Toxicology 18, 303-390.
Ceyer, S. (1990). New mechanisms for chemistry at surfaces. Science 249, 133-139.
CFCL (2004). Ceramic Fuel Cells Ltd. Distributed Generation Product Concept, web: http://www.cfcl.com.au.
Chahine, R. (2003). Review of progress in H² storage technologies. In "Proc. 1st European Hydrogen Energy Conf., Grenoble 2003", CDROM produced by Association Francaise de l'Hydrogène, Paris.
Chalk, S., Devlin, P., Gronich, S., Milliken, J., Sverdrup, G. (2004). The United States' FreedomCAR and hydrogen fuel initiative. In Proc. 15 th World Hydrogen Energy Conf., Yokohama. 28A-09, CD Rom, Hydrogen Energy Soc. Japan.
Chan, S.H., Abou-Ellail, M., Yan, T. (2004). Prediction and measurement of fuel cell flammability limits. Int. J. Green Energy 1, 101-114.
Chang, F., Lin, C. (2004). Biohydrogen production using an up-flow anaerobic sludge blanket reactor. Int. J. Hydrogen Energy, 29 33-39.
Chartier, P., Meriaux, S. (1980). Recherche 11, 766—776.
Chaudhuri, S., Lovley, D. (2003). Electricity generation by direct oxidation of glucose in mediatorless microbial fuel cells. Nature Biotechnology 21, 1229-1232.
Chen, Y-H., Chen, C-Y., Lee, S-C. (2011). Technology forecasting and patent strategy of hydrogen energy and fuel cell technologies. Int. J. Hydrogen Energy, doi:10.1016/j.ijhydene.2011.03.063.
Cheng, H., Scott, K., Ramshaw, C. (2002). Intensification of water electrolysis in a centrifugal field. J. Electrochem. Soc. 149, D172-D177.
Chibing, S., Qinwu, L., Chunlin, J., Jin, Z., Zhenguo, W. (2001). Combustion performance of H2/O2/hydrocarbon tripropellant engine operating in dual mode. In "Hydrogen Energy Progress XIII, Proc. 13th World Energy Conf., Beijing 2000" (Mao, Z., Veziroglu, T., eds.), pp. 677-683. Int. Assoc. Hydrogen Energy, Beijing.
Chitnis, P. (2001). Photosystem I: function and physiology. Ann. Rev. Plant Physiol. Mol. Biol. 52, 593-626.
Chitose, K., Takeno, K., Kouchi, A., Yamada, Y., Okabayashi, K. (2004). Activities on hydrogen safety for hydrogen refueling stations — experiment and simulation of gaseous hydrogen dispersion. In Proc. 15th World Hydrogen Energy Conf., Yokohama. CD Rom, Hydrogen Energy Soc. Japan.
Chiu, Y-J., Yu, T., Chung, Y-C. (2011). A semi-empirical model for efficiency evaluation of a direct methanol fuel cell. J. Power Sources 196, 5053-5063.
Choudhary, V., Banerjee, S., Rajput, A. (2002). Hydrogen from step-wise steam reforming of methane over Ni/ZrO² : factors affecting catalytic methane decomposition and gasification by steam of carbon formed on the catalyst. Appl. Catalysis A: General 234, 259-270.
Christensen, C., Sørensen, R., Johannessen, T., Quaade, U., Honkala, K., Elmøe, T., Køhler, R., Nørskov, J. (2005). Metal ammine complexes for hydrogen storage. J. Materials Chemistry 15, 4106-4108.
Chung, T. (1997). The role of thorium in nuclear energy. US Dept. Energy Info. Agency, http://www.eia.doe.gov/cneaf/nuclear/uia/thorium/thorium.html.
Cifrain, M., Kordesch, K. (2004). Advances, aging mechanism and lifetime in AFCs with circulating electrolytes. J. Power Sources 127, 234-242.
Ciureanu, M., Miklailenko, S., Kaliaguine, S. (2003). PEM fuel cell as membrane reactors: kinetic analysis by impedance spectroscopy. Catalysis Today 82, 195-206.
Clarke, S., Dicks, A., Pointon, K., Smith, T., Swann, A. (1997). Catalytic aspects of the
steam reforming of hydrocarbons in internal reformiing fuel cells. Catalysis Today 38, 411-423.
Clayton, R. (1965). "Molecular Physics in Photosynthesis". Blaisdell, New York.
Colclasure, A., Sanandaji, B., Vincernt, T., Kee, R. (2011). Modeling and control of tubular solid-oxide fuel cell systems. I: Physical models and linear model reduction. J. Power Sources 196, 196-207.
Colella, W., Jacobsen, M., Golden, D. (2005). Switching to a U.S. hydrogen fuel cell vehicle fleet: The resultant change in emissions, energy use, and greenhouse gases. J. Power Sources 150, 150-181.
Colpan, C., Hamdullahpur, F., Dincer, I. (2010). Transient heat transfer modeling of a solid oxide fuel cell operating with humidified hydrogen. Int. J. Hydrogen Energy, doi:10.1016/j.ijhydene.2010.11.127.
Consoli, F. et al. (eds.) (1993). "Guidelines for Life−cycle Assessment: A Code of Practice." Society of Environmental Toxicology and Chemistry (SETAC).
Corbo, P., Migliardini, F., Veneri, O. (2010). Lithium polymer batteries and proton exchange membrane fuel cells as energy sources in hydrogen electric vehicles. J. Power Sources 195, 7849-7854.
Cordiner, S., Lanzani, S., Mulone, V. (2010). 3D effects of water-saturation distribution on polymeric electrolyte fuel cell (PEFC) performance. Int. J. Hydrogen Energy, doi:10.1016/j.ijhydene.2010.09.063.
Cormos, C-C. (2011). Hydrogen production from fossil fuels with carbon capture and storage based on chemical looping systems. Int. J. Hydrogen Energy, doi:10.1016/ j.ijhydene.2011.01.170
Cortright, R., Davda, R., Dumesic, J. (2002). Hydrogen from catalytic reforming of biomass-derived hydrocarbons in liquid water. Nature 418, 964-967.
Costamagna, P., Selimovic, A., Borghi, M., Agnew, G. (2004). Electrochemical model of the integrated planar solid oxide fuel cell (IP-SOFC). Chem. Eng. J. 102, 61-69.
Courson, C., Makaga, E., Petit, C., Kiennemann, A. (2000). Development of Ni catalysts for gas production from biomass gasification. Reactivity in steam- and dryreforming. Catalysis Today 63, 427-437.
CRC (1973). "Handbook of Chemistry and Physics" (Weast, R., ed.). The Chemical Rubber Co., Cleveland, OH.
Cuddy, M. (1998). Volkswagen gearbox description for ADVISOR software. File notes, National Renewable Energy Lab., Golden, CO.
Culley, A., Lang, A., Suttle, C. (2003). High diversity of unknown picorna-like virusses in the sea. Nature, 424, 1054-1057.
Dabrock, B., Bahl, H., Gottschal, G. (1992). Parameters affecting solvent production by Clostridium pasteurianum. Appl. Environm. Microbiol. 58, 1233-1239.
Dahl, J., Buechler, K., Weimer, A., Lewandowski, A., Bingham, C. (2004). Solarthermal dissociation of methane in a fluid-wall aerosol flow reactor. Int. J. Hydrogen Energy 29, 725-736.
DaimlerChrysler (2001). Accident-free driving a vision. Strategies for safety, Hightech Report, 18-23.
DaimlerChrysler-Ballard (last accessed 1998). Fuel-cell development programme. http://www.daimler−benz.com/research/specials/necar/necar_e.htm.
Damen, K., Faaij, A., Walter, A., Souza, M. (2002). Future prospects for biofuel production in Brazil. In "12th European Biomass Conf.", Vol. 2, pp. 1166−1169. ETA Firenze & WIP Munich.
Danish DoE (1998). Energy-2100. Plan scenario and Action plan with Update (1999). Danish Department of Energy and Environment, Copenhagen.
Danish Hydrogen Committee (1998). Brint et dansk energi perspektiv (Sørensen, B., ed.). Danish Energy Agency, Copenhagen.
Danish National Agency for Enterprise and Construction (1999). Building Registry (extract). Description: http://www.ebst.dk/Publikationer/0/10 (2003).
Danish Power Utilities (1997). Action plan for off-shore wind parks (in Danish). SEAS Wind Dept.
Danish Transport Council (1993). Externaliteter i transportsektoren. Report 93−01, Copenhagen.
Danish Windpower Industry Association (2003). Web: http://www.windpower.dk.
Dante, R. (2005). Hypotheses for direct PEM fuel cells applications of photobioproduced hydrogen by Chlamydomonas reinhardtii. Int. J. Hydrogen Energy 30, 421- 424.
Dapprich, S., Komáromi, I., Byun, K., Morokuma, K., Frisch, M. (1999). A new ONIOM implementation in Gaussian98. Part I. The calculation of energies, gradients, vibrational frequencies and electric field derivatives. J. Molec. Structure (Theochem) 461-462, 1-21.
Dayton, D., Ratcliff, M., Bain, R. (2001). Fuel cell integration a study of the impacts of gas quality and impurities. Report NREL/MP-510-30298, National Renewable Energy Lab., Golden CO.
DC (2004). F-Cell brochure (Japanese/English). http://www.daimlerchrysler.co.jp
Debe, M. (2003). Novel catalysts, catalysts support and catalysts coated membrane methods. In "Handbook of Fuel Cells - Fundamentals, Technology and Applications (Vielstich, W., Gasteiger, H., Lamm, A., eds.), Vol. 3, Ch. 45, pp. 576-589. John Wiley & Sons, New York.
Dell, R., Bridger, N. (1975). Hydrogen the ultimate fuel. Appl. Energy 1, 279-292.
deMouy, L. (1998). Projected world uranium requirements, reference case. USDoE Energy Information Administration: Int. Energy Information Report, website: http://www.eia.doe.gov/cneaf/nuclear/n_pwr_fc/apenf1.html.
Desrosiers, R. (1981). In "Biomass Gasification" (Reed, T., ed.), pp. 119-153. Noyes Data Corp., Park Ridge, NJ
Dey, R. (2004). Facilitating commercialization of hydrogen technologies through the activities of ISO/TC 197. In Proc. 15th World Hydrogen Energy Conf., Yokohama. CD Rom, Hydrogen Energy Soc. Japan.
Dirac. P. (1930). Proc. Cambridge Phil. Soc. 27, 240.
Doctor, R., Wade, D., Mendelsohn, M. (2002). STAR-H2: a calcium-bromine hydrogen cycle using nuclear heat. Paper for "American Inst. Chem. Eng. Spring Meeting", New Orleans, http://www.eere.energy.gov/hydrogenandfuelcells.
Dollmayer, J., Bundschuh, N., Carl, U. (2006). Fuel mass penalty due to generators and fuel cells as energy source of the all-electric aircraft. Aerospace Sci. & Technology 10, 686-694.
DONG (2003). Gas stores. Danish Oil and Gas Co. http://www.dong.dk/dk/publikationer/lagerbrochure/(last assessed 2003).
Doucette, R., McCulloch, M. (2011). A comparison of high-speed flywheels, batteries, and ultracapacitors on the bases of cost and fuel economy as the energy storage system in a fuel cell based hybrid electric vehicle. J. Power Sources, 196, 1163- 1170.
Drift, A. (2002). An overview of innovative biomass gasification concepts. In "Proc. PV in Europe Conf.". WIP, Munich & ETA, Florence.
DTI (2000). 1996 Danish Reference Year, obtained from Danish Technological Institute, Tåstrup.
Duffie, J., Beckman, W. (1991). "Solar Energy Thermal Processes", 2nd ed., Wiley, New York.
Dukes, J. (2003). Burning buried sunshine: Human consumption of ancient solar energy. Climate Change 61, 31-44.
Dunn-Rankin, D., Leal, E., Walther, D. (2005). Personal power systems. Prog. Energy & Combustion Sci. 31, 422-465.
Dutta, S., Morehouse, J., Khan, J. (1997). Numerical analysis of laminar flow and heat transfer in a high temperature electrolyzer. Int. J. Hydrogen Energy 22, 883-895.
Ebbesen, S., Graves, C., Mogensen, M. (2009). Production of synthetic fuels by coelectrolysis of steam and carbon dioxide. Int. J. Green Energy 6, 646-660.
Eberle, U., Helmolt, R. von (2010). Fuel cell electric vehicles, battery electric vehicles, and their impact on energy storage technologies: an overview. Ch. 9 in Electric and Hybrid Vehicles (G. Pistoia, ed.), 247-273. Elsevier, Amsterdam.
EC (1994). "Biofuels" (M. Ruiz−Altisent, ed.), DG XII Report EUR 15647 EN, European Commission, Brussels.
EC (2001). The ECE-EUDC driving cycle. European Commission Report 90/C81/01, Brussels.
EC (2004). Well-to-wheels analysis of future automotive fuels and powertrains in the European context. Joint study of the European Council for Automotive R&D, European Oil Companies' Association for environment, health and safety in refining and distribution (CONCAWA), the Institute for Environment and Sustainability of the European Commission's Joint Research Centre, L-B Systemtechnik and Institut Francais de Pétrole. WTW Report 220104. CORDIS.
EC-ATLAS (2003). European Commission DG Energy: ATLAS programme, http://europa.eu.int/energy_transport/atlas/htmlu/lbpot2.html.
EEA (2002). Size of vehicle fleet. Indicator fact sheet TERM 32AC, European Environmental Agency, Copenhagen.
Efoy (2011). EFOY Pro 2200 XT: 100% guaranteed off-grid power! At http://www. efoy.com/en/traffic-new-efoy-pro-2200-xt.html
EG&G (2004). Fuel Cell Handbook (7th ed,). Technical Services work for US DoE, contract DE-AM26-99FT40575.
Eguchi, K., Fujihara, T., Shinozaki, N., Okaya, S. (2004). Current work on solar RFC technology for SPF airship. In Proc. 15th World Hydrogen Energy Conf., Yokohama. 30A-07, CD Rom, Hydrogen Energy Soc. Japan.
Eichler, A., Hafner, J. (1997). Molecular precursors in the dissociative adsorption of O2 on Pt(111). Phys. Rev. Lett. 79, 4481-4484.
Einsle, O., Tezcan, F., Andrade, S., Schmid, B., Yoshida, M., Howard, J., Rees, D. (2002). Nitrogenase MoFe-protein at 1.16Å resolution: a central ligand in the FeMo-cofactor. Science 297, 1696-1700.
Elder, R., Allen, R. (2009). Nuclear heat for hydrogen production: Coupling a very high/high temperature reactor to a hydrogen production plant. Progress Nucl. Energy 51, 500-525.
Electricity Storage Association (2009). Technology Comparison. Washington DC, http://www.electricitystorage.org/ESA/technologies
Elliott, J., Hanna, S., Elliott, A., Cooley, G. (2000). Interpretation of the small-angle Xray scattering from swollen and oriented perfluorinated ionomer membranes. Macromolecules 33, 4161-4171.
El-Sharkh, M., Tanrioven, M., Rahman, A., Alam, M. (2010). Economics of hydrogen production and utilization strategies for the optimal operation of a grid-parallel PEM fuel cell power plant. Int. J. Hydrogen Energy 35, 8804-8814.
Eltra/Elkraft (2001). Time series of Danish wind power production 2000, available at websites http://www.eltra.dk and http://www.elkraft-system.dk.
Eltra (2003). Søkabel, Danish power utility webpage http://www.eltra.dk.
EnBW (2004). Sulzer-Hexis SOFC-field test. Energie Baden-Würtemburg AG. Website: http://www.enbw.com.
Energinet.dk (2007). Press releases concerning purchase of gas store from DONG. At http://www.energinet.dk/da/ (accessed 2007).
Energi- og Miljødata (1999). Windresource Mapper. CD-Rom from EMD, Aalborg.
Energi- og Miljødata (2002). Danish wind turbine price lists. Previously published by Danish Energy Agency and then Renewable Energy Information Secretariat, beginning 1981. http://www.emd.dk.
Energy Information Agency of the US DoE (2011). World crude oil prices. Spreadsheet covering the period 1946 to 11. February 2011, when accessed. Available at http://www.eia.gov/nev/pet/pet_pri_wco_k_w.xls
Enquete Kommission des Deutschen Bundestages (1995). Mehr Zukunft für die Erde. (Lippold et al., eds). Economica Verlag, Bonn.
Erdmann, G. (2003). Future economies of the fuel cell housing market. Int. J. Hydrogen Energy 28, 685-694.
Eroglu, E., Gündüz, U., Yücel, M., Türker, L., Eroglu, I. (2004). Photobiological hydrogen production by using olive mill wastewater as a sole substrate source. Int. J. Hydrogen Energy 29, 163-171.
Eroglu, E., Melis, A. (2011). Photobiological hydrogen production: Recent advances and state of the art. Bioresource Technology, doi:10.1016/j.biortech.2011.03.026.
Escudero, M., Rodrigo, T., Soler, J., Daza, L. (2003). Electrochemical behaviour of lithium-nickel oxides in molten carbonate. J. Power Sources 118, 23-34.
European Commission (1995). ExternE: externalities of Energy, Vols. 1-6. Reports EUR 16520-16525 EN, DGXII, Luxembourg.
European Commission (1997). Energy in Europe, 1997 Annual Energy Review, DGXII: Science, Research & Development, Special Issue, 179 pp.
European Commission (1998). A fuel cell RDD strategy for Europe to 2005. DGXIIF, Brussels.
European Commission (2008). Hyways, the European hydrogen roadmap. Report from an industry consortium, EUR 23123, Brussels. ftp://ftp.cordis.europa.eu/ publ/fp7/energy/docs/hyways-roadmap_en.pdf
European Platform for Hydrogen and Fuel Cell Technologies (2004). Strategic Research Agenda and Deployment Strategy. Draft report available on website: http://www.hfpeurope.org/
Evans, B., O'Neill, H., Malyvanh, V., Lee, I., Woodward, J. (2003). Palladiumbacterial cellulose membranes for fuel cells. Biosensors Bioelectronics 18, 917-923.
Eveloy, V. (2010). Numerical analysis of an internal methane reforming solid oxide fuel cell with fuel recycling. Applied Energy, doi:10.1016/j.apenergy.2010.10.045.
ExxonMobil (2004). A report on energy trends, greenhouse gas emissions and alter-
native energy, Houston, TX.
Faaij, A., Hamelinck, C. (2002). Long term perspectives for production of fuels from biomass; integrated assessment and R&D priorities. In "12th European Biomass Conf." Vol. 2, pp. 1110-1113. ETA Firenze & WIP Munich.
Fadel, A., Zhou, B. (2011). An experimental and analytical comparison study of power management methodologies of fuel cell—battery hybrid vehicles. J. Power Sources 196, 2171-3279.
Fan, Y., Li, C., Lay, J., Hou, H., Zhang, G. (2004). Optimization of initial substrate and pH levels for germination of sporing hydrogen-producing anaerobes in cow dung compost. Bioresource Technology, 91 189-193.
Fang, H., Liu, H., Zhang, T. (2002). Characterization of a hydrogen-producing granular sludge. Biotechnology Bioeng. 78, 44-52.
Farooque, M., Ghezel-Ayagh, H. (2003). System design. In "Handbook of Fuel Cells, Vol. 4" (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 68. Wiley, Chichester.
Feng, W., Wang, S., Ni, W., Chen, C. (2004). The future of hydrogen infrastructure for fuel cell vehicles in China and a case of application in Beijing. Int. J. Hydrogen Energy 29, 355-367.
Fermi, E. (1928). Zeitschrift f. Physik 48, 73.
Fernandez, L., Garcia, P., Garcia, C., Jurado, F. (2011). Hybrid electric system based on fuel cell and battery and integrating a single dc/dc converter for a tramway. Energy Conversion & Management 52, 2183-2192.
Fernandez, R., Mandrillon, P., Rubbia, C., Rubio, J. (1996). A preliminary estimate of the economic impacts of the energy amplifier. Report CERN/LHC/96-01(EET), 75 pp.
Ferreira, K., Iverson, T. Maghlaoui, K., Barber, J., Iwata, S. (2004). Architecture of the photosynthetic oxygen-evolving center. Science 303, 1831-1835.
Fingersh, L. (2003). Optimized hydrogen and electricity generation from wind. National Renewable Energy Lab., Report NREL/TP-500-34364, Golden, CO.
Finneran, K., Johnsen, C., Lovley, D. (2003). Rhodoferax ferrireducens sp. nov., a psychrotolerant, facultatively anaerobic bacterium that oxidizes acetate with the reduction of Fe(III). Int. J. Systematic Evolutionary Microbiology 53, 669-673.
Fischer, G., Schnagl, J., Sarre, C., Lechner, W. (2003). Function of the liquid hydrogen fuel system for the new BMW 7 series. In "Proc. 1st European Hydrogen Energy Conf.", Grenoble 2003. CDROM published by Association Francaise de l'Hydrogène, Paris, 6 pp.
Fleischer, T., Oertel, D. (2003). Fuel cells impact and consequences of fuel cell technology on sustainable development. Report EUR 20681 EN, European Commission JRC, Sevilla.
Fock, V. (1930). Zeitschrift f. Physik 61, 126.
Folkesson, A., Andersson, C., Alvfors, P., Alaküla, M., Overgaard, L. (2003). Real life testing of a hybrid PEM fuel cell bus. J. Power Sources 118, 349-357.
Fontana, G., Galloni, E., Jannelli, E., Minutillo, M. (2004). Different technologies for hydrogen engine fuelling. In "Hydrogen Power: Theoretical and Engineering Solutions, Proc. Hypothesis V Conf., Porto Conte 2003" (Marini, M., Spazzafumo, G., eds.), pp. 917-927. Servizi Grafici Editoriali, Padova.
Fontell, E., Kivisaari, T., Christiansen, N., Hansen, J-B., Pålsson, J. (2004). Conceptual study of a 250 kW planar SOFC system for CHP application. J. Power Sources 131, 49-56.
Foresman, J., Head-Gordon, M., Pople, J., Frisch, M. (1992). Towards a systematic molecular orbital theory for excited states. J. Phys. Chem., 96, 135-149.
Fowler, M., Mann, R., Amphlett, J., Peppley, B., Roberge, P. (2002). Incorporation of voltage degradation into a generalised steady state electrochemical model for a PEM fuel cell. J. Power Sources 106, 274-283.
Frangini, S., Masci, A. (2004). Intermetallic FeAl based coatings deposited by the electrospark technique: corrosion behavior in molten (Li+K) carbonate. Surface Coatings Tech. 184, 31-39.
Freni, S., Barone, F., Puglisi, M. (1998). The dissolution process of the NiO cathodes for molten carbonate fuel cells: state-of-the-art. Int. J. Energy Res. 22, 17-31.
Freni, S., Passalacqua, E., Barone, F. (1997). The influence of low operating temperature on molten carbonate fuel cells decay processes. Int. J. Energy Res. 21, 1061- 1070.
Friedlmeier, G., Friedrich, J., Panik, F. (2001). Test experiences with the Daimler-Chrysler fuel cell electric vehicle NECAR 4. Fuel Cells 1, 92-96.
Friedrich, A., Veser, A., Stern, G., Kotchourko, N. (2011). Hyper experiments on catastrophic hydrogen releases inside a fuel cell enclosure. Int. J. Hydrogen Energy 36, 2678-2687.
Frisch, M. J., G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, J. A. Montgomery, Jr., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J. B. Cross, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C. Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, C. Gonzalez, and J. A. Pople, (2003). Gaussian 03 software, Revision B.02. Gaussian, Inc., Pittsburgh PA (use of this reference format is part of user licence).
Fromme, P., Jordan, P., Krauss, N. (2001). Structure of photosystem I. Biochim. Biophys. Acta 1507, 5-31.
Fromme, P., Melkozernov, A., Jordan, P., Krauss, N. (2003). Structure and function of photosystem I: interaction with its soluble electron carriers and external antenna systems. FEBS Lett. 555, 40-44.
Fuhrmann, J., Gärtner, K. (2003). A detailed numerical model for DMFC: discretization and solution methods. Paper for "Computaional fuel cell dynamics II", Banff Int. Res. Center, USA.
Fujiwara, S., et al. (2008). Hydrogen production by high temperature electrolysis with nuclear reactor. Progress Nucl. Energy 50, 422-426.
Fujioka, Y., Ozaki, M., Takeuchi, K., Shindo, Y., Herzog, H. (1997). Cost comparison of various CO² ocean dispersal options. Energy Conversion & Management 38, S273-S277.
Fujishima, A., Honda, K. (1972). Electrochemical photolysis of water at a semiconductor electrode. Nature 283, 37.
Fukada, S., Nakamura, N., Monden, J. (2004). Effects of temperature, oxygen-tomethane molar ratio and superficial gas velocity on partial oxydation of methane for hydrogen production. Int. J. Hydrogen Energy 29, 619-625.
Fukushima, Y., Shimada, M., Kraines, S., Hirao, M., Koyama, M. (2004). J. Power Sources 131, 327-339.
Futerko, P., Hsing, I-M. (2000). Two-dimensional finite-element method study of the resistance of membranes in polymer electrolyte fuel cells. Electrochimica Acta 45, 1741-1751.
FZK (1999). LH2 release in tunnel. Hydrogen research at Forschungszentrum Karlsruhe. In "Proc. Hydrogen Workshop at European Commission DG XII", website http://www.eihp.org/eihp1/workshop
Gallucci, F., Paturzo, L., Basile, A. (2004). A simulation study of the steam reforming of methane in a dense tubular membrane reactor. Int. J. Hydrogen Energy 29, 611- 617.
Gambardella, P., Sljivancanin, Z., Hammer, B., Blanc, M., Kuhnke, K., Kern, K. (2001). Oxygen dissociation at Pt steps. Phys. Rev. Lett. 87, 056103.1-4.
Gates, D. (1966). Science 151, 523—529.
Gaussian (2003). Software package, see Frisch et al. (2003).
Geek with laptop (2011). Horizon Fuel cell Technologies: MiniPak hydrogen fuel cell charger for portable devices. http://www.geekwithlaptop.com and http:// www.horizonfuelcell.com/electronics.htm
Ghenciu, A. (2002). Review of fuel processing catalysts for hydrogen production in PEM fuel cell systems. Current Opinion Solid State Material Sci. 6, 389-399.
Ghosh, D. (2003). Development of stationary solid oxide fuel-cells at Global Thermoelectric Inc. In "14th World Hydrogen Energy Conference", Montreal 2002, File B001g, 5 pp. CD published by CogniScience Publ. for l'Association Canadienne de l'Hydrogène, revised CD issued 2003.
Gierke, T., Hsu, W. (1982). The cluster-network model of ion clustering in perfluorosulfonated membranes. In "Perfluorinated Ionomer Membranes", American Chemical Society Symp. Series 180, Washington, DC.
Gigliucci, G., Petruzzi, L., Cerelli, E., Garzisi, A., LaMendola, A. (2004). Demonstration of a residential CHP system based on PEM fuel cells. J. Power Sources 131, 62-68.
Gil, M., Ji, X., Li, X., Na, H., Hampsay, J., Lu, Y. (2004). Direct synthesis of sulfonated aromatic poly(ether ether ketone) proton exchange membranes for fuel cell applications. J. Membrane Sci. 234, 75-81.
Giles, J. (2004). Every last drop. Nature 429, 694-695.
GM (2001). Well-to-wheel energy use and greenhouse gas emissions of advanced fuel/vehicle systems — North American analysis. Report from General Motors Corp., Argonne Nat. Lab., BP, ExxonMobil and Shell. (For adjacent European study see Wurster, 2003).
Gnanapragasam, N., Reddy, B., Rosen, M. (2010), Feasibility of an energy conversion system in Canada involving large-scale integrated hydrogen production using solid fuels. Int. J. Hydrogen Energy 35, 4788-4807.
Gøbel, B., Bentzen, J., Hindsgaul, C., Henriksen, U., Ahrenfeldt, J., Houbak, N., Qvale, B. (2002). High performance gasification with the two-stage gasifier, In "12th European Biomass Conf.", pp. 289-395. ETA Firenze & WIP Munich.
González, A., McKeogh, E., Gallachóir, B. (2003). The role of hydrogen in high wind
energy penetration electricity systems: the Irish case. Renewable Energy 29, 471- 489.
Gouérec, P., Poletto, L., Denizot, J., Sanchez-Cortezon, E., Miners, J. (2004). The evolution of the performance of alkaline fuel cells with circulating electrolytes. J. Power Sources 129, 193-204.
Grand, P. (1979). Nature 278, 693-696.
Griffith, D. (1995). "Introduction to quantum mechanics". Prentice Hall, New Jersey. Grochala, W., Edwards, P. (2004). Thermal decomposition of the non-interstitial hydrides for the storage and production of hydrogen. Chem. Rev. 104, 1283-1315.
Groenestijn, J. v., Hazewinkel, J., Nienroord, M., Bussmann, P. (2002). Energy aspects of biological hydrogen production in high rate bioreactors operated in the thermophilic temperature range. Int. J. Hydrogen Energy 27, 1141-117.
Güllü, D., Demirbas, A. (2001). Biomass to methanol via pyrolysis process. Energy Conversion Management 42, 1349-1356.
Guo, L., Yedavalli, K., Zinger, D. (2011). Design and modeling of power system for a fuel cell hybrid switcher locomotive. Energy Conversion & Management 52, 1406- 1413.
Gülzow, E. (1996). Alkaline fuel cells: a critical view. J Power Sources 61, 99—104.
Gülzow, E., Schulze, M. (2004). Long-term operation of AFC electrodes with CO² containing gases. J. Power Sources 127, 243-251.
GWEC (2010). Global wind report. Annual market update. Global Wind Energy Council, Brussels. http://www.gwec.net
Güther, V., Otto, A. (1999). Recent developments in hydrogen storage applications based on metal hydrides. J. Alloys Compounds 293-295, 889-892.
Ha, S., Adams, B., Masel, R. (2004). A miniature air breathing direct formic acid fuel cell. J. Power Sources 128, 119-124.
Hahn, R.,Wagner, S., Schmitz, A., Reichl, H. (2004). Development of a planar micro fuel cell with thin film and micro patterning technologies. J. Power Sources 131, 73-78.
Haile, S., Boysen, D., Chisholm, C., Merle, R. (2001). Solid acids as fuel cell electrolytes. Nature 410, 910-913.
Hajimiragha, A., Cañizares, C., Fowler, M., Moazeni, S., Elkamel, A., Wang, S. (2011). Sustainable convergence of electricity and transport sectors in the context of a hydrogen economy. Int. J. Hydrogen Energy, doi:10.1016/j.ijhydene. 2011.02.070.
Hallenbeck, P. (2009). Fermentative hydrogen production: Principles, progress, and prognosis. Int J. Hydrogen Energy 34, 7379-7389.
Hamann, C., Hammett, A., Vielstich, W. (1998). "Electrochemistry". Wiley-VCH, Weinheim.
Hamilton Sundstrand Inc. (2003). Water electrolysis. Website http:// xnwp021.utc.com/ssi/ssi/Applications/Echem/Background/waterelec.html.
Hammer, B., Nørskov, J. (1995). Why gold is the noblest of all the metals. Nature 376, 238-240.
Hammer, B., Nørskov, J. (1997). Adsorbate reorganization at steps: NO on Pd(211). Phys. Rev. Lett. 79, 4441-4444.
Hamnett, A. (2003). Direct methanol fuel cells (DMFC). In "Handbook of Fuel Cells Vol. 1" (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 18. Wiley, Chichester.
Han, Y., Furukawa, Y. (2006). Conducting polyaniline and biofuel cell. Int. J. Green
Energy 3, 17-23.
Han, S., Shin, H. (2004). Biohydrogen production by anaerobic fermentation of food waste. Int. J. Hydrogen Energy 29, 569-577.
Handley, C., Brandon, N., van der Vorst, R. (2002). Impact of the European vehicle waste directive on end-of-life options for polymer electrolyte fuel cells. J. Power Sources 106, 344-352.
Hanneman, R., Vakil, H., Wentorf Jr., R. (1974). Closed loop chemical systems for energy transmission. In "Proc. 9th Intersociety Energy Conversion Engineering Conf.". American Society of Mechanical Engineers, New York.
Hannerz, K. (1983). Nuclear Engineering International Dec., p. 41
Happe, T., Schütz, K., Böhme, H. (1999). Transcriptional and mutational analysis of the uptake hydrogenase of the filamentous cyanobacterium Anabaena variabilis ATCC 29413. J. Bacteriology 182, 1624-1631.
Hardy, B., Anton, D. (2009). Hierarchical methodology for modeling hydrogen storage systems. Part II: Detailed models. Int. J. Hydrogen Energy 34, 2992-3004.
Harrington, D., Conway, B. (1987). Electrochimica Acta 32, 1703.
Harth, R., Range, J., Boltendahl, U. (1981). EVA-ADAM system, a method of energy transportation by reversible chemical reactions. In "Energy storage and transportation" (Beghi, G., ed.), pp. 358-374. Reidel, Dordrecht.
Hartree, D. (1928). Proc. Cambridge Phil. Soc. 24, 89.
Hasunuma, T., Fukusaki, E., Kobayashi, A. (2003). Methanol production is enhanced by expression of an Aspergillus niger pectin methylesterase in tobacco cells. J. Biotechnology 106, 45-52.
Haugen, H., Eide, L. (1996). CO² capture and disposal: the realism of large scale scenarios. Energy Conversion Management 37, 1061-1066.
Hawkes, F., Dinsdale, R., Hawkes, D., Hussy, I. (2002). Sustainable fermentative hydrogen production: challenges for process optimisation. Int. J. Hydrogen Energy, 27, 1339-1347.
Hayashi, T., Watanabe, S. (2004). Hydrogen safety for fuel cell vehicles. In Proc. 15th World Hydrogen Energy Conf., Yokohama. CD Rom, Hydrogen Energy Soc. Japan.
Hedrick, J. (1998). Thorium. US Geological Survey, Mineral commodity summary & Yearbook. http://minerals.er.usgs.gov/minerals/pubs/commodity/thorium.
Hernández, S., Scarpa, F., Fino, D., Conti, R. (2011). Biogas purification for MCFC application. Int. J. Hydrogen Energy, doi:10.1016/j.ijhydene.2011.01.055.
Herrmann, M., Meusinger, J. (2003). Hydrogen storage systems for mobile applications. In "Proc. 1st European Hydrogen Energy Conf., Grenoble 2003", CDROM produced by Association Francaise de l'Hydrogène, Paris.
Herrmann, A., Schimmele, L., Mössinger, J., Hirscher, M., Kronmüller, H. (2001). Diffusion of hydrogen in heterogeneous systems. Appl. Phys. A72, 197-208.
Herzog, H., Adams, E., Auerbach, D., Caulfield, J. (1996). Environmental impacts of ocean disposal of CO² , Energy Conversion Management 37, 999-1005.
- Hibino, T., Hashimoto, A., Yano, M., Suzuki, M., Yoshia, S., Sano, M. (2002). J. Electrochem. Soc. 149, A133.
Hibino, T., Hashimoto, A., Yano, M., Suzuki, M., Sano, M. (2003). Ru-catalyzed anode materials for direct hydrocarbon SOFCs. Electrochimica Acta 48, 2531-2537.
Hirsch, R., Gallagher, J., Lessard, R., Wesselhoft, R. (1982). Science 183, 909-915.
Hirsch, D., Steinfeld, A. (2004). Solar hydrogen production by thermal decomposi-
tion of natural gas using a vortex-flow reactor. Int. J. Hydrogen Energy 29, 47-55.
Hodges, H. (1970). "Technology in the ancient world". Barnes & Nobles, New York.
Hodoshima, S., Arai, H., Saito, Y. (2001). Liquid-film type catalytic decalin dehydrogeno-aromatization for mobile storage of hydrogen. In "Hydrogen Energy Progress XIII, Proc. 13th World Energy Conf., Beijing 2000" (Mao, Z., Veziroglu, T., eds.), pp. 504-509. Int. Assoc. Hydrogen Energy, Beijing.
Hoffmann, P. (1998). ZEVCO unveils fuel cell taxi. Hydrogen and Fuel Cell Letter, feature article, August (http://www.mhv.net/~hfcletter/letter).
Hoffmann, J., Yuh, C-Y., Jopek, A. (2003). Electrolyte and material challenges. In "Handbook of Fuel Cells, Vol. 4" (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 67. Wiley, Chichester.
Hoganson, C., Babcock, G. (1997). A metalloradical mechanism for the generation of oxygen from water in photosynthesis. Science 277, 1953-1956.
Hohenberg, P., Kohn, W. (1964). Inhomogeneous electron gas. Phys. Rev. 136, B864- B871.
Holladay, J., Hu, J., King, D., Wang, Y. (2009). An overview of hydrogen production technologies. Catalysis Today 139, 244-260.
Holladay, J., Wainright, J., Jones, E., Gano, S. (2004). Power generation using a mesoscale fuel cell integrated with a microscale fuel processor. J. Power Souces 130, 111-118.
Holmqvist, M., Lindblad, P., Sørensen, B. (2005). Life-cycle analysis pf bio-hydrogen production. In "Proc 2nd European Hydrogen Energy Conf., Zaragoza", pp. 294-296, also on CDROM, Idea Madrid.
Honda (2004). Honda's vision of future home life. http://www.honda.com.
Horch, S., Lorensen, H., Helweg, S., Lægsgaard, E., Stensgaard, I., Jacobsen, K., Nørskov, J., Besenbacher, F. (1999). Enhancement of surface self-diffusion of platinum atoms by adsorbed hydrogen. Nature 398, 134-136.
Horny, C., Kiwi-Minsker, L., Renken, A. (2004). Micro-structured string-reactor for autothermal production of hydrogen. Chem. Eng. J. 101, 3-9.
Hu, M., Zhu, X., Wang, M., Gu, A., Yu, L. (2004). Three dimensional, two phase flow mathematical model for PEM fuel cell: Parts I and II. Analysis and discussion of the internal transport mechanism. Energy Conversion & Management 45, 1861- 1882 and 1883-1916.
Hua, T., Ahluwalia, R., Peng, J-K., Kromer, M., Lasher, S., McKenney, K., Law, K., Sinha, J. (2011). Technical assessment of compressed hydrogen storage tank systems for automotive applications. Int. J. Hydrogen Energy 36, 3037-3049.
Huang, T-W., et al. (2009). Application of fuel cell hybrid electric scooter and development of hydrogen supply technology. Renewable Energy, doi:10.1016/j.renene. 2008.12.037
Hubbert, M. (1962). Energy resources, a report to the Committee on Natural Resources. Nat. Acad. Sci., Publ. 1000D.
Ibenholt, K. (2002). Explaining learning curves for wind power. Energy Policy 30, 1181-1189.
IEA (2002). Key world energy statistics, International Energy Agency, Paris, Available at http://www.iea.org
IEA (2004). World Energy Outlook 2004. Executive Summary. OECD/IEA, Paris.
IEA-PVPS (2010). Trends in photovoltaic applications. Report IEA-PVPS T1-19:2010. International Energy Agency, Paris.
Industry Canada (2003). Canadian fuel cell commercialization roadmap. Industry Canada, Government Canada, Vancouver, BC.
Ioroi, T., Yasuda, K., Siroma, Z., Fujiwara, N., Miyazaki, Y. (2002). Thin film electrocatalyst layer for unitized regenerative polymer electrolyte fuel cells. J. Power Sources 112, 583-587.
Ioroi, T., Yasuda, K., Miyazaki, Y. (2004). Polymer electrolyte-type unitized regenerative fuel cells. In "15th World Hydrogen Energy Conference, Yokohama 2004". Paper P09-09. Hydrogen Energy Systems Soc. of Japan (CDROM).
IPCC (1996). "Climate Change 1995: Impacts, Adaptation and Mitigation of Climate Change: Scientific-Technical Analysis. Contribution of WGII" (Watson et al., eds.), 572 pp. Cambridge University Press, Cambridge.
Ishihara, T., Shibayama, T., Ishikawa, S., Hosoi, K., Nishiguchi, H., Takita, Y. (2004). Novel fast oxide ion conductor and application for the electrolyte of solid oxide fuel cell. J. European Ceramic Soc. 24, 1329-1335.
Itoh, N., Kaneko, Y., Igarashi, A. (2002). Efficient hydrogen production via methenol steam reforming by preventing back-permeation of hydrogen in a palladium membrane reactor. Industrial Eng. Chem. Res. 41, 4702-4706.
Iwai, Y. (2004). Japan's approach to commercialization of fuel cell/hydrogen technology. In Proc. 15th World Hydrogen Energy Conf., Yokohama. 28PL-02, CD Rom, Hydrogen Energy Soc. Japan.
Iwata, S., Lee, J., Okada, K., Lee, J., Iwata, M., Rasmussen, B., Link, T., Ramaswamy, S., Jap, B. (1998). Complete structure of the 11-subunit bovine mitochondrial cytochrome bc¹ complex. Science 281, 64-71.
Jaffray, C., Hards, G. (2003). Precious metal supply requirements. In "Handbook of Fuel Cells - Fundamentals, Technology and Applications", Vol 3, Ch. 41 (Vielstich, W., Gasteiger, H., Lamm, A., eds.), pp. 509-513. John Wiley & Sons, New York.
Jang, S., Molinero, V., Cagin, T., Goddard, W. III (2004). Nanophase-segregation and transport in Nafion 117 from molecular dynamics simulations: effect of monomeric sequence. J. Phys. Chem. B108, 3149-3157.
Janssen, H., Bringmann, J., Emonts, B., Schroeder, V. (2004). Safety-related studies on hydrogen production in high-pressure electrolysers. Int. J. Hydrogen Energy 29, 759-770.
Jeng, K-T., Hsu, N-Y., Chien, C-C. (2011). Synthesis and evaluation of carbon nanotube-supported RuSe catalyst for direct methanol fuel cell cathode. Int. J. Hydrogen Energy 36, 3997-4006.
Jensen, J., Li, Q., He, R., Xiao, G., Gao, J-A., Bjerrum, N. (2004). High temperature polymer fuel cells and their interplay with fuel processing systems. In "Hydrogen Power — Theoretical and Engineering Solutions, Proc. Hypothesis V, Porto Conte 2003" (Marini, M., Spazzafumo, G., eds.), pp. 675-683. Servizi Grafici Editoriali, Padova.
Jensen, J., Sørensen, B. (1984). "Fundamentals of Energy Storage". Wiley, New York, 345 pp.
Jeong, K., Oh, B. (2002). Fuel economy and life-cycle cost analysis of a fuel cell hybrid vehicle. J. Power Sources 105, 58-65.
Jiao, K., Alaefour, I., Li, X. (2011). Three-dimensional non-isothermal modeling of carbon monoxide poisoning in high temperature proton exchange membrane fuel cells with phosphoric acid doped polybenzimidazole membranes. Fuel 90,
568-582.
Jiao, K., Li, X. (2011). Water transport in polymer electrolyte membrane fuel cells. Progress in Energy & Combustion Science 37, 221-291.
Jin, X., Xue, X. (2010). Mathematical modeling analysis of regenerative solid oxide fuel cells in switching mode conditions. J. Power Sources 195, 6652-6658.
Jing, D., et al. (2010). Efficient solar hydrogen production by photocatalytic water splitting: From fundamental study to pilot demonstration. Int. J. Hydrogen Energy 35, 7087-7097.
Jo, J-H., Yi, S-C. (1999). A computational simulation of an alkaline fuel cell. J. Power Sources 84, 87-106.
Joensen, F., Rostrup-Nielsen, J. (2002). Conversion of hydrocarbons and alcohols for fuel cells. J. Power Sources 105, 195-201.
Johnson, V. (2001). Module ESS_L17_temp documentation file for Advisor. National Renewable Energy Laboratory, Golden, CO.
Jordan, P., Fromme, P., Witt, H., Klukas, O., Saenger, W., Krauss, N. (2001). Threedimensional structure of cyanobacterial photosystem I at 2.5 Å resolution, Nature 411, 909-917.
Jun, J., Jun, J., Kim, K. (2002). Degradation behaviour of Al-Fe coatings in wet-seal area of molten carbonate fuel cells. J. Power Sources 112, 153-161.
Jung, M., Williams, K. (2011). Effect of dynamic operation on chemical degradation of a polymer electrolyte membrane fuel cell. J. Power Sources 196, 2127-2724.
Junginger, M., Faaij, A., Turkenburg, W. (2004). Global experience curves for wind farms. Energy Policy 33, 133-150.
Jurewicz, K. (2011). Influence of charging parameters on the effectiveness of electrochemical hydrogen storage in activated carbon. Int. J. Hydrogen Energy 34, 9431- 9435.
Jurewicz, K., Frackowiak, E., Béguin, F. (2004). Towards the mechanism of electrochemical hydrogen storage in nanustructured carbon materials. Appl. Phys. A78, 981-987.
Kahn, J. (1996). Fuel cell breakthrough doubles performance, reduces cost. Berkeley Lab. Research News, 29. May (http://www.lbl.gov/science−articles/archive/ fuel−cells.html).
Kalkstein, L. (1993). Health and climate change: direct impacts in cities. The Lancet 342, 1397-1399.
Kalkstein, L., Smoyer, K. (1993). The impact of climate change on human health: some international implications. Experientia 49, 969−979.
Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin, L., Iredell, M., Saha, S., White, G., Woollen, J., Zhu, Y., Leetmaa, A., Reynolds, R., Chelliah, M., Ebisuzaki, W., Higgins, W., Janowiak, J., Mo, K., Ropelewski, C., Wang, J., Jenne, R., Joseph, D. (1996). The NCEP/NCAR 40-year reanalysis project. Bull. Am. Met. Soc. (March).
Kaltschmitt, M., Reinhardt, G., Stelzer, T. (1996). LCA of biofuels under different environmental aspects, In "Biomass for Energy and the Environment" (Chartier, P., Ferrero, G., Henius, U., Hultberg, S., Sachau, J., Wiinblad, M., eds.), Vol. 1, pp. 369-386. Pergamon/Elsevier, Oxford.
Kalyanasundaram, K., Graetzel, M. (2010). Artificial photosynthesis: biomimetic approaches to solar energy conversion and storage. Current Opinion in Biotechnology 21, 298-310.
Kamiya, N., Shen, J-R. (2003). Crystal structure of oxigen-evolving photosystem II from Thermosynechococcus vulcanus at 3.7-Å resolution. Proc. Nat. Acad. Sci. (US) 100, 98-103.
Karim, G., Wierzba, A. (2004). The lean flammability and operational mixture limits of gaseous fuel mixtures containing hydrogen in air. In "Hydrogen Power: Theoretical and Engineering Solutions, Proc. Hypothesis V Conf., Porto Conte 2003" (Marini, M., Spazzafumo, G., eds.), pp. 839-845. Servizi Grafici Editoriali, Padova.
Karmazyn, A., Fiorin, V., Jenkins, S., King, D. (2003). First-principles theory and microcalorimetry of CO adsorption on the {211} surfaces of Pt and Ni. Surface Sci. 538, 171-183.
Kasperovich, V., et al. (2010). NMR study of metal-hydrogen systems for hydrogen storage. J. Alloys & Compounds, doi:10.1016/j.jallcom.2010.10.195.
Kato, T., Suzuoki, Y. (2004). Energy saving potential of home co-generation system using PEFC in both individual household and overall energy system. In Proc. 15th World Hydrogen Energy Conf., Yokohama. P12-02, CD Rom, Hydrogen Energy Soc. Japan.
Kattke, K., Braun, R., Colclasure, Goldin, G. (2011). High-fidelity stack and system modeling for tubular solid oxide fuel cell system design and thermal management. J. Power Sources 196, 3790-3802.
Katuri, K., Scott, K. (2011). On the dynamic response of the anode in microbial fuel cells. Enzyme and Microbial Technology 40, 351-358.
Katz, E., Shipway, A., Willner, I. (2003). Biochemical fuel cells. In "Handbook of Fuel Cells, Vol. 1" (Vielstich, W., Lamm, A., Gasteiger, H., eds.). Wiley, Chichester.
Kawada, T., Mizusaki, J. (2003). Current electrolytes and catalysts. Ch. 70 in "Handbook of fuel cells vol. 4" (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 21. Wiley, Chichester.
Kazempoor, P., Dorer, V., Weber, A. (2010). Modelling and evaluation of building integrated SOFC systems. Int. J. Hydrogen Energy, doi:10.1016/j.ijhydene.2010. 11.003
Kendrich, D., Herding, G., Scouflaire, P., Rolon, C., Candel, S. (1999). Effects of a recess on cryogenic flame stabilization. Combustion Flame 118, 327-339.
Keränen, T., Karimäki, H., Viitakangas, J., Vallet, J., Ihonen, J., Hyötylä, P., Uusalo, H., Tingelöf, T. (2011). Development of integrated fuel cell hybrid power source for electric forklift. J. Power Sources, doi:10.1016/j.jpowsour.2011.01.025
Key, T., Sitzlar, H., Geist, T. (2003). Fast response, load-matching hybrid fuel cell. Report NREL/SR-560-32743, Nat. Renewable Energy Lab., Golden, CO.
Khaselev, O., Turner, J. (1998). A monolithic photovoltaic-photoelectrochemical device for hydrogen production via water splitting. Science 280, 425-427.
Kikuchi, E., Menoto, Y., Kajiwara, M., Uemiya, S., Kojima, T. (2000). Steam reforming of methane in membrane reactors: comparison of electroless-plating and CVD membranes and catalyst packing methods. Catalysis Today 56, 75-81.
Kikuzawa, H., Ohmura, T., Yamaguchi, R., Ohtuka, M., Sawada, Y., Tomihara, I. (2004). Japanese national project for establishment of codes and standards for stationary PEM fuel cell system. In Proc. 15th World Hydrogen Energy Conf., Yokohama. CD Rom, Hydrogen Energy Soc. Japan.
Kim, J-D., Honma, I. (2004). Synthesis and proton conducting properties of zirconia bridged hydrocarbon/phosphotungstic acid hybrid materials. Electrochimica Ac-
ta 49, 3179-3183.
Kim, W., Voiti, T., Rodriguez-Rivera, Dumesic, J. (2004). Powering fuel cells with CO via aqueous polyoxometalates and gold catalysts. Science 305, 1280-1283.
King, J., McDonald, B. (2003). Experience with 200 kW PC25 fuel cell power plant. In "Handbook of Fuel Cells, Vol. 4" (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 61. Wiley, Chichester.
Kittel, C. (1971). "Introduction to Solid State Physics". Wiley, New York.
Klueh, P. (1986). New Scientist 3. April, 41-45.
Klug, H., Faass, R. (2001). Cryoplane: hydrogen fuelled aircraft status and challenges. Air & Space Europe 3, 252-254.
Knorr, H., Held, W., Prümm, W., Rüdiger, H. (1998). The MAN hydrogen propulsion system for city bus. Int. J. Hydrogen Energy 23, 201-208.
Knudsen, M. (1934). "The kinetic theory of gases". Methuen, London.
Kocha, S., Turner, J., Nozik, A. (1991). J. Electroanalyt. Chem. 367, 27.
Köhler, J., Wietschel, M., Whitmarsh, L., Keles, D., Schade, W. (2010). Infrastructure investment for a transition to hydrogen automobiles. Technological Forecasting & Social Change 77, 1237—1248.
Kohn, W., Sham, L. (1965). Self-consistent equations including exchange and correlation effects. Phys. Rev. 140 (1965) A1133.
Koide, H., Shindo, Y., Tazaki, Y., Iijima, M., Ito, K., Kimura, N., Omata, K. (1997). Deep sub-seabed disposal of CO² — the most protective storage. Energy Conversion Management 38, S253-S258.
Kok, B., Forbush, B., McGloin, M. (1970). Photochem. Photobiol. 11, 457-475.
Kondo, T., Arakawa, M., Wakayama, T., Miyake, J. (2002). Hydrogen production by combining two types of photosynthetic bacteria with different characteristics. Int. J. Hydrogen Energy 27, 1303-1308.
Kopanidis, A., Theodorakakos, A., Gavaises, M., Bouris, D. (2011). Pore scale 3D modelling of heat and mass transfer in the gas diffusion layer and cathode channel of a PEM fuel cell. Int. J. Thermal Sciences 50, 456-467.
Kordesch, K., Simader, G. (1996). "Fuel cells and their applications". VCH Verlag, Weinheim.
Koroneos, C., Dompros, A., Roumbas, G., Moussiopoulos, N. (2004). Life cycle assessment of hydrogen fuel production processes. Int. J. Hydrogen Energy 29, 1443-1450.
Kosugi, T., Hayashi, A., Tokimatsu, K. (2004). Forecasting development of elemental technologies and efficiency of R&D investments for polymer electrolyte fuel cells in Japan. Int. J. Hydrogen Energy 29, 337-346.
Kratzer, P., Pehlke, E., Scheffler, M., Raschke, M., Höfer, U. (1998). Highly sitespecific H² adsorption on vicinal Si(001) surfaces. Phys. Rev. Lett. 81, 5596-5599.
Kreuer, K. (2001). On the development of proton conducting polymer membranes for hydrogen and methanol fuel cells. J. Membrane Sci. 185, 29-39.
Kreuer, K. (2003). Hydrocarbon membranes. In "Handbook of Fuel Cells Fundamentals, Technology and Applications", Vol. 3 (Vielstich, W., Gasteiger, H., Lamm, A., eds.), ch. 33. John Wiley & Sons, Chichester.
Kryukov, A., Levashov, V., Sazhin, S. (2004). Evaporation of diesel fuel droplets: kinetic versus hydrodynamic models. Int. J. Heat Mass Transfer 47, 2541-2549.
Kubo, Y. (2004). Micro fuel cells for portable electronics. In Proc. 15th World Hydrogen Energy Conf., Yokohama. CD Rom, Hydrogen Energy Soc. Japan.
Kudin, K., Scuseria, G. (1998). A fast multipole algorithm for the efficient treatment of the Coulomb problem in electronic structure calculations of periodic systems with Gaussian orbitals. Chem. Phys. Lett. 289, 611-616.
Kudin, K., Scuseria, G. (2000). Linear-scaling density-functional theory with Gaussian orbitals and periodic boundary conditions: efficient evaluation of energy and forces via the fast multipole method. Phys. Rev. B61, 16443.
Kudin, K., Scuseria, G., Martin, R. (2002). Hybrid density-functional theory and the insulating gap of UO² . Phys. Rev. Lett. 89, 266402.
Kuemmel, B., Nielsen, S., Sørensen, B. (1997). "Life−cycle Analysis of Energy Systems". Roskilde University Press, Copenhagen, 216 pp.
Kulinovsky, A. (2010). The regimes of catalyst layer operation in a fuel cell. Electrochimica Acta 55, 6391-6401.
Kulinovsky, A., Scharmenn, H., Wippermann, K. (2004). Dynamics of fuel cell performance degradation. Electrochem. Comm. 6, 75-82.
Kümmel, S., Perdew, J. (2003). Simple iterative construction of the optimized effective potential for orbital functionals, including exact exchange. Phys. Rev. Lett. 90, 043004.
Kurchatov Institute (1997). Hypertext data base: Chernobyl and its consequences, Website http://polyn.net.kiae.su/polyn/manifest.html (last accessed 1999).
Kurisu, G., Zhang, H., Smith, J., Cramer, W. (2003). Structure of the cytochrome b⁶ f complex of oxygenic photosynthesis: tuning the cavity. Science 302, 1009-1014.
Kussmaul, K., Deimel, P. (1995). Materialverhalten in H² -Hochdrucksystemen. VDI Berichte 1201, pp. 87-101.
LaConti, A., Hamdan, M., McDonald, R. (2003). Mechanisms of membrane degradation. In "Handbook of Fuel Cells", Vol. 3 (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 49. Wiley, Chichester.
Ladebeck, J., Wagner, J. (2003). Catalyst development for water-gas shift. In "Handbook of Fuel Cells" (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 16. Wiley, Chichester.
Lamm, A., Müller, J. (2003). System design for transport applications. In "Handbook of Fuel Cells", Vol. 4 (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 64. Wiley, Chichester.
Lange, J-P. (1997). Perspectives for manufacturing methanol at fuel value. Industrial Eng. Chem. Res. 36, 4282-4290.
Lapeña-Rey, N., Mosquera, J., Bataller, E., Orti, F., Dudfield, C., Orsillo, A. (2008). Environmentally friendly power sources for aerospace applications. J. Power Sources 181, 353-362.
Lauritsen, A., Svendsen, T., Sørensen, B. (1996). A study of the integration of wind energy into the national energy systems of Denmark, Wales and Germany as illustrations of success stories for renewable energy. Wind Power in Denmark, EC project report RENA.CT94-0012 (106 pp). IMFUFA, Roskilde University.
Lay, J. (2000). Biohydrogen generation by mesophilic anaerobic fermentation of microcrystalline cellulose. Biotechnology Bioeng. 74, 280-287.
Lecocq, A., Furukawa, K. (1994). Accelerator molten salt breeder. In "Procedings 8th Journées Saturne, Saclay", pp. 191-192. Website (last accessed 1999): http:// db.nea.fr/html/trw/docs/saturne8/.
Le, A., Zhou, B. (2009). A generalized numerical model for liquid water in a proton exchange membrane fuel cell with interdigitated design. J. Power Sources 193,
665-683.
Lee, S., Bessarabov, D., Vohra, R. (2009). Degradation of a cathode catalyst layer in PEM MEAs subjected to automotive-specific test conditions. Int. J. Green Energy 6, 594-606.
Lee, C., Yang, W., Parr, R. (1988). Development of the Colle-Salvetti correlationenergy formula into a functional of the electron density. Phys. Rev., B37, 785.
Lee, C-H., Yang, J-T. (2011). Modeling of the Ballard-Mark-V proton exchange membrane fuel cell with power converters for applications in autonomous underwater vehicles. J. Power Sources 196, 3810-3823.
Lee, H., Hong, H., Kim, Y-M., Choi, S., Hong, M., Lee, H., Kim, K. (2004). Preparation and evaluation of sulphonated-fluorinated poly(arylene ether)s membranes for a proton exchange membrane fule cell (PEMFC). Electrochimica Acta 49, 2315- 2323.
Lee, J-Y., Yoo, M., Cha, K., Lim, T., Hur, T. (2009). Life cycle cost analysis to examine the economical feasibility of hydrogen as an alternative fuel. Int. J. Hydrogen Energy 34, 4243-4255.
Lee, J-Y., An, S., Cha, K., Hur, T. (2010). Life cycle environmental and economic analyses of a hydrogen station with wind energy. Int. J. Hydrogen Energy 35, 2213- 2225.
Lee, S., Mukerjee, S., McBreen, J., Rho, Y., Kho, Y., Lee, T. (1998). Effects of Nafion impregnation on performances of PEMFC electrodes. Electrochimica Acta 43, 3693-3701.
Lee, Y., Nirmalakhandan, N. (2011). Electricity production in membrane-less microbial fuel cell fed with livestock organic solid waste. Bioresource Technology, doi: 10.1016/j.biortech.2011.02.090.
Lefenfeld, M., Dye, J., Barton, S. (2006). Sodium silicide amd alkali metal-silica gel for convenient hydrogen production. Materials Engineering News, Premier Issue, June (year missing), 16-17 (journal discontinued).
Leighty, W. (2008). Running the world on renewables: Hydrogen transmission pipelines and firming geologic storage. Int. J. Energy Res. 32, 408-426.
Lennard-Jones, J. (1932). Trans. Faraday Soc. 28, 333.
Levene, J., Mann, M., Margolis, R., Milbrandt, A. (2007). An analysis of hydrogen production from renewable electricity sources. Solar Energy 81, 773-780.
Levin, D., Pitt, L., Love, M. (2004). Biohydrogen production: prospects and limitations to practical application. Int. J. Hydrogen Energy 29, 173-185.
Levin, D., Chahine, R. (2010). Challenges for renewable hydrogen production from biomass. Int. J. Hydrogen Energy 35, 4962-4969.
Li, P-W. and M. Chyu, M. (2003). Simulation of the chemical/electrochemical reactions and heat/mass transfer for a tubular SOFC in a stack. J. Power Sources 124, 487-498.
Li, Q., Jensen, J., He, R., Bjerrum, N. (2004). New polymer electrolyte membranes based on acid doped PBI for fuel cells operating above 100°C. In "Hydrogen Power — Theoretical and Engineering Solutions, Proc. Hypothesis V, Porto Conte 2003" (Marini, M., Spazzafumo, G., eds.), pp. 685-696. Servizi Grafici Editoriali, Padova.
Li, X., Li, J., Xu, L., Yang, F., Hua, J., Ouyang, M. (2010). Performance analysis of proton-exchange membrane fuel cell stacks used in Beijing urban-route buses trial project. Int. J. Hydrogen Energy 35, 3841-3847.
Li, X., Ogden, J. (2011). Understanding the design and economics of distributed trigeneration systems for home and neighborhood refueling–Part I: Single family residence case studies. J. Power Sources 196, 2098-2108.
Liang, G. (2003a). Magnesium-based alloys for hydrogen storage. In "Hydrogen and Fuel Cells Conference. Towards a greener world", Vancouver June, CDROM published by Canadian Hydrogen Association and Fuel Cells Canada, Vancouver.
Liang, J. (2003b). Theoretical insight on tailoring energetics of Mg hydrogen absorption/desorption through nano-engineeering. Appl. Phys. A DOI:10.1007/s00339 -003-2383-3.
Liaw, B., Dubarry, M. (2010). A roadmap to understand battery performance in electric and hybrid vehicle operation. Ch. 15 in Electric and Hybrid Vehicles (G. Pistoia, ed.), 375-403. Elsevier, Amsterdam.
Lim, C., Wang, C-Y. (2004). Effects of hydrophobic polymer content in GDL on power performance of a PEM fuel cell. Electrochimica Acta 49, 4149-4156.
Lin, C., Lay, C. (2004). Carbon/nitrogen-ratio effect on fermentative hydrogen production by mixed microflora. Int. J. Hydrogen Energy 29, 41-45.
Lin, H., Ouyang, L., Wang, H., Liu, J., Zhu, M. (2011). Phase transition and hydrogen storage properties of melt-spun Mg3LaNi0.1 alloy. Int. J. Hydrogen Energy, doi:10. 1016/j.ijhydene.2011.02.071
Lin, Y.-M., Rei, M.-H. (2000). Process development for generating high purity hydrogen by using supported palladium membrane reactor as steam reformer. Int. J. Hydrogen Energy 25, 211-219.
Lin, W-S., Zheng, C-H. (2011). Energy management of a fuel cell/ultracapacitor hybrid power system using an adaptive optimal-control method. J. Power Sources 196, 3280-3289.
Linden, S. van der (2003). The commercial world of energy storage: a review of operating facilities. Presentation for "1st Ann. Conf. Energy Storage Council", Houston, Texas.
Lipman, T., Edwards, J., Kammen, D. (2004). Fuel cell system economics: comparing the costs of generating power with stationary and motor vehicle PEM fuel cell systems. Energy Policy 32, 101-125.
Lister, S., McLean, G. (2004). PEM fuel cell electrodes. J. Power Sources 130, 61-76.
Liu, P., Nørskov, J. (2001). Kinetics of the anode processes in PEM fuel cells the promoting effect of Ru in PtRu anodes. Fuel Cells 1, 192-201.
Liu, S., Takahashi, K., Ayabe, M. (2003). Hydrogen production by oxidative methanol reforming on Pd/ZnO catalyst: effects of Pd loading. Catalysis Today 87, 247- 253.
Longenbach, J., Rutkowski, M., Klett, M., White, J., Schoff, R., Buchanan, T. (2002). "Hydrogen Production Facilities, Plant Performance and Cost Comparisons". US DoE, Nat. Energy Technology Lab., Reading, PA.
Longwell, H. (2002). The future of the oil and gas industry: Past approaches, new challenges, World Energy (Houston) 5, 100-104.
Losciale, M. (2002). Technical experiences and conclusions from introduction of biogas as a vehicle fuel in Sweden. In "12th European Biomass Conf.", vol. 2, pp. 1124-1127. ETA Firenze & WIP Munich.
Lostao, A., Daoudi, F., Irún, M., Ramón, Á., Fernández-Cabrera, C., Romero, A., Sancho, J. (2003). How FMN binds to Anabaena apoflavodoxin. J. Biol. Chem. 278,
24053-24061.
Løvvik, O. (2004). Adsorption of Ti on LiAlH⁴ surfaces studied by band structure calculations. J. Alloys Compounds 373, 28-32.
Lu, G., Costa, J., Duke, M., Giessler, S., Socolow, R., Williams, R., Kreutz, T. (2007). Inorganic membranes for hydrogen production and purification: A critical review and perspective. J. Colloid & Interface Science 314, 589-603.
Lu, G., Wang, C. (2004). Electrochemical and flow characterization of a direct methanol fuel cell. J. Power Sources 134, 33-40.
Lu, G., Wang, C. (2005). Development of micro direct methanol fuel cells for high power applications. J. Power Sources 144, 141-145.
Lu, G., Wang, C., Yen, T., Zhang, X. (2004). Development and characterization of a silicon-based micro direct methanol fuel cell. Electrochimica Acta 49, 821-828.
Lung, M. (1997). Reactors coupled with accelerators. Joint Research Center (ISPRA) seminar paper (5 pp., revision 12.3.1997), Website (last accessed 1999): http://itumagill.fzk.de/ADS/mlungACC.htm.
Lunghi, P., Bove, R. (2003). Life cycle assessment of a molten carbonate fuel cell stack. Fuel Cells 3, 224-230.
Lusardi, M., Bosio, B., Arato, E. (2004). An example of innovative application in fuel cell system development: CO² segregation using molten carbonate fuel cell. J. Power Sources 131, 351-360.
Lushetsky, J. (2010). The prospect for \$1/watt electricity from solar. Solar Energy Technology Program of the Office of Energy Efficiency and Renewable Energy, US Department of Energy. \$1/W Workshop, http://www.solar.energy.gov
Lutz, A., Bradshaw, R., Bromberg, L., Rabinovich, A. (2004). Thermodynamic analysis of hydrogen production by partial oxidation reforming. Int. J. Hydrogen Energy 29, 809-816.
MacLean, H., Lave, L. (2003). Evaluating automobile fuel/propulsion system technologies. Progress Energy Combustion Sci. 29, 1-69.
Madsen, B. (2002). International wind energy development. Annual reports, BTM consult. http://www.btm.dk.
Magazu, V., Freni, A., Cacciola, G. (2003). Hydrogen storage: strategic fields and comparison of different technologies. In "Hydrogen Power — Theoretical and Engineering Solutions, Proc. Hypothesis V, Porto Conte 2003" (Marini, M., Spazzafumo, G., eds.), pp. 371-386. Servizi Grafici Editoriali, Padova.
Magill, J., O'Carroll, C., Gerontopoulos, P., Richter, K., van Geel, J. (1995). Advantages and limitations of thorium fuelled energy amplifiers. In "Proc. Unconventional Options for Plutonium Dispositions, Obninsk", Int. Atomic Energy Agency TECDOC-840, pp. 81-86.
Mahadevan, K., Contini, V., Goshe, M., Price, J., Eubanks, F., Griesemer, F. (2010). Economic analysis of stationary PEM fuel cell systems. Presentation at US DoE meeting in Washington, DC. http://www.hydrogen.energy.gov/pdf/…/fc050 _mahadevan_2010_p_web.pdf
Mahlia, T., Chan, P. (2011). Life cycle cost analysis of fuel cell based cogeneration system for residential application in Malaysia. Renewable and Sustainable Energy Reviews 15, 416-426.
Malek, K., Eikerling, M., Wang, Q., Navessin, T., Liu, Z. (2007). Self-Organization in Catalyst Layers of Polymer Electrolyte Fuel Cells. J. Chemical Physics C 111, 13627-13634.
MAN (2004). Nutzfahrzeuge. http://www.brennstoffzellenbus.de/bus/bus.html.
Mao, W., Mao, H., Goncharov, A., Struzhkin, V., Guo, Q., Hu, J., Shu, J., Hemley, R., Somayazulu, M., Zhao, Y. (2002). Hydrogen clusters in clathrate hydrate. Science 297, 2247-2249.
MAPINFO (1997). Professional GIS Software v 4.5, country boundaries. Troy, NY.
Marchetti, C. (1973). Chem. Econ. & Eng. Rev. 5, 7.
Marin, G., Naterer, G., Gabriel, K. (2010). Rail transportation by hydrogen vs. electrification - Case study for Ontario, Canada, II: Energy supply and distribution. Int. J. Hydrogen Energy, 35, 6097-6107.
Markel, T., Brooker, A., Hendricks, T., Johnson, V., Kelly, K., Kramer, B., O'Keefe, M., Sprik, S., Wipke, K. (2002). ADVISOR: a systems analysis tool for advanced vehicle modeling. J. Power Sources 110, 255-266. A commercial version of the software is under preparation by AVL. Graz, Austria; http://www.avl.com
Maron, S., Prutton, C. (1959). "Principles of Physical Chemistry". Macmillan, New York.
Masukawa, H., Nakamura, K., Mochimaru, M., Sakurai, H. (2001). Photohydrogen production and nitrogenase activity in some heterocystous cyanobacteria. In "BioHydrogen II" (Miyake, J., Matsunaga, T., Pietro, A., eds.), pp. 63-66.
Masukawa, H., Mochimaru, M., Sakurai, H. (2002). Hydrogenases and photobiological hydrogen production utilizing nitrogenase system in cyanobacteria. Int. J. Hydrogen Energy 27, 1471-1474.
Matsumoto, H., Okubo, M., Hamajina, S., Katahira, K., Iwahara, H. (2002). Extraction and production of hydrogen using high-temperature proton conductor. Solid State Ionics 152-3, 715-720.
Matsumura, Y., Minowa, T. (2004). Fundamental design of a continuous biomass gasification process using a supercritical water fluidized bed. Int. J. Hydrogen Energy 29, 701-707.
Matsuo, Y., Saito, K., Kawashima, H., Ikehata, S. (2004). Novel solid acid fuel cell based on a superprotonic conductor Tl3H(SO⁴ )2 . Solid State Comm. 130, 411-414.
Matter, P., Braden, D., Ozkan, U. (2004). Steam reforming of methanol to H² over nonreducing Zr-containing CuO/ZnO catalysts. J. Catalysis 223, 340-351.
Maxoulis, C., Tsinoglou, D., Koltsakis, G. (2004). Modeling of automotive fuel cell operation in driving cycles. Energy Conversion & Management 45, 559-573.
McLean, G., Niet, T., Prince-Richard, S., Djilali, N. (2002). An assessment of alkaline fuel cell technology. Int. J. Hydrogen Energy 27, 507-526.
McMichael, T. (2001). "Human frontiers, environments and disease". Cambridge University Press, Cambridge.
Meegahawatte, D., Hillmansen, S., Roberts, C., Falco, M., McGordon, A., Jennings, P. (2010). Analysis of a fuel cell hybrid commuter railway vehicle. J. Power Sources 195, 7829-7837.
Meibom, P., Svendsen, T., Sørensen, B. (1999). Trading wind in a hydro-dominated power pool system. Int. J. Sustainable Development 2, 458-483.
Meisen, A., Shuai, X. (1997). Research and development issues in CO² capture. Energy Conversion Management 38, S37-S42.
Mendoza, L., Baddour-Hadjean, R., Cassir, M., Pereira-Ramos, J. (2004). Raman evidence of the formation of LT-LiCoO² thin layers on NiO in molten carbonate at 650°C. Appl. Surface Sci. 225, 356-361.
Mepsted, G., Moore, J. (2003). Performance and durability of bipolar plate. In
"Handbook of Fuel Cells Fundamentals, Technology and Applications", Vol. 3 (Vielstich, W., Gasteiger, H., Lamm, A., eds.), Ch. 23. John Wiley & Sons, Chichester.
Mercedes-Benz (2004). Fuel cells in field trials. http://www.mercedes-benz.com under "citarofcell".
Mérida, W., Maness, P., Brown, R., Levin, D. (2004). Enhanced hydrogen production from indirectly heated, gasified biomass, and removal of carbon gas emissions using a novel biological gas reformer. Int. J. Hydrogen Energy 29, 283-290.
Merilo, E., Groethe, M., Colton, J., Chiba, S. (2011). Experimental study of hydrogen release accidents in a vehicle garage. Int. J. Hydrogen Energy 36, 2436-2444.
Mermelstein, J., Millan, M., Brandon, N. (2011). The interaction of biomass gasification syngas components with tar in a solid oxide fuel cell and operational conditions to mitigate carbon deposition on nickel-gadolinium doped ceria anodes. J. Power Sources 196, 5027-5034.
Meyers, J., Maynard, H. (2002). Design considerations for miniaturized PEM fuel cells. J. Power Sources 109, 76-88.
Michel, F., Fieseler, H., Meyer, G., Theissen, F. (1998). On-board equipment for liquid hydrogen vehicles. Int. J. Hydrogen Energy 23, 191-199.
Middleman, E., Kout, W., Vogelaar, B., Lenssen, J., Waal, E. de (2003). Bipolar plates for PEM fuel cells. J. Power Sources 118, 44-46.
Milczarek, G., Kasuya, A., Mamykin, S., Arai, T., Shinoda, K., Tohji, K. (2003). Optimization of a two-compartment photoelectrochemical cell for solar hydrogen production. Int. J. Hydrogen Energy 28, 919-926.
Mimura, T., Simayoshi, H., Suda, T., Iijima, M., Mituoka, S. (1997). Development of energy saving technology for flue gas carbon dioxide recovery in power plant by chemical absorption method and steam system. Energy Conversion Management 38, S57-S62.
Minami, E., Kawamoto, H., Saka, S. (2002). Reactivity of lignin in supercritical methanol studied with some lignin model compounds. In "12th European Biomass Conf.", pp. 785–788. ETA Firenze & WIP Munich.
Minkevich, I., Laurinavichene, T., Tsygankov, A. (2004). Theoretical and experimental quantum efficiencies of the growth of anoxygenic phototrophic bacteria. Process Biochem. 39, 939-949.
Mirenowicz, J. (1997). Le CERN reste silencieux face au nucléaire "propre" proposé par Carlo Rubbia, Journal de Géneve, 21. June, p. 13.
Mishra, P., Shukla, P., Singh, A., Srivastava, O. (2003). Investigation and optimization of nanostructures TiO² photoelectrode in regard to hydrogen production through photoelectrochemical process. Int. J. Hydrogen Energy 28, 1089-1094.
Mitsuda, K., Maeda, H. , Mitani, T., Matsumura, M., Urushibata, H., Yoshiyasu, H. (2004). In Proc. 15th World Hydrogen Energy Conf., Yokohama. 01PL-02, CD Rom, Hydrogen Energy Soc. Japan.
Miyake, J., Miyake, M., Asada, Y. (1999). Biotechnological hydrogen production: research for efficient light energy conversion. J. Biotechnology 70, 89-101.
Modic, E. (2011). Sodium Silicide. Today's Energy Solutions, March. At http:// www.onlineTES.com/tes-0311-hydrogen-fuel-cells-sodium-silicide.htm
Mohitpour, M., Golshan, H., Murray, A. (2000). "Pipeline Design & Construction". ASME Press, New York.
Molina-Heredia, F., Wastl, J., Navarro, J., Bendall, D., Hervás, M., Howe, C., Rosa, M.
(2003). A new function for an old cytochrome? Nature 424, 33-34.
Møller, C., Plesset, M. (1934). Note on an approximation treatment for many-body systems. Phys. Rev. 46, 618.
Morales, R., Chron, M., Hudry-Clegeon, G., Pélillot, Y., Nørager, S., Medina, M., Frey, M. (1999). Refined X-ray structures of the oxidized, at 1.3 Å, and reduced, at 1.17 Å, [2Fe-2S] ferredoxin from the cyanobacterium Anabaene PCC7119 show redox-linked conformation changes. Biochemistry, 38, 16764-15773.
Mori, D., Hirose, K. (2009). Recent challenges of hydrogen storage technologies for fuel cell vehicles. Int. J. Hydrogen Energy 34, 4569-4574.
Morinaga, M., Yukawa, H. (2002). Nature of chemical bond and phase stability of hydrogen storage compounds. Materials Science Eng. A329, 268-275.
Morse, P. (1964). "Thermal Physics". W. A. Benjamin, New York.
Morthorst, P., Auer, H., Gerard, A., Blanco, I. (2008). The economics of wind power. Part III of Wind energy — the facts. European Wind Energy Association, at http:// www.wind-energy-the-facts.org
Mosher, D., Tang, X., Arsenault, S., Laube, B., Cao, M., Brown, R., Saitta, S. (2007). High Density Hydrogen Storage System Demonstration Using NaAlH⁴ Complex Compound Hydrides. In "Proc. DOE hydrogen program 2007 annual merit review, Arlington. Available from: http://www.hydrogen.energy.gov/pdfs/ review07 /stp_33_mosher.pdf (accessed April 2011).
MPS (2004). Hot module MCFCs. Modern Power Systems Inc. Website: http://www. connectingpower.com.
MSI (2000). WebLab ViewerLite software, v3.7. Molecular Simulations Inc.
Mueller, J., Urban, P. (1998). Characterization of direct methanol fuel cells by ac impedance spectroscopy. J. Power Sources 75, 139-143.
Mugikura, Y. (2003). Stack material and stack design. In "Handbook of Fuel Cells", Vol. 4 (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 66. Wiley, Chichester.
Müller, J., Frank, G., Colbow, K., Wilkinson, D. (2003). Transport/kinetic limitations and efficiency losses. Ch. 62 in "Handbook of fuel cells vol. 4" (Vielstich, W., Lamm, A., Gasteiger, H., eds.). Wiley, Chichester.
Müller, J., Urban, P., Hölderich, W. (1999). Impedance studies on direct methanol fuel cell anodes. J. Power Sources 84, 157-160.
Mulliken, R. (1955). Electronic population analysis on LCAO-MO molecular wave functions, I. J. Chem. Physics.23, 1833-1837.
Muradov, N. (2003). Emission-free fuel reformers for mobile and portable fuel cell applications. J. Power Sources 118, 320-324.
Murakami, M., Hirose, K., Kawamura, K., Sata, N., Ohishi, Y. (2004). Post-perovskite phase transition in MgSiO³ . Science 304, 855-858.
Myers, D., et al. (2009). Polymer Electrolyte Fuel Cell Lifetime Limitations: The Role of Electrocatalyst Degradation. Project kick-off Meeting, Washington DC. At http://www.hydrogen. energy.gov
MyFC (2011). PowerTrekk fuel cell charger. At http://www.powertrekk.com
Nagel, F., Schildhauer, T., Sfeir, J., Schuler, A., Biollaz, S. (2009). The impact of sulfur on the performance of a solid oxide fuel cell (SOFC) system operated with hydrocarboneous fuel gas. J. Power Sources 189, 1127-1131.
Nakamori, Y., Orimo, S. (2004). Destabilization of Li-based complex hydrides. J. Alloys Compounds 370, 271-275.
Nakao, M., Yoshitake, M. (2003). Composite perfluorinate membranes. In
"Handbook of Fuel Cells", Vol. 3 (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 32. Wiley, Chichester.
Nakicenovic, N., Grübler, A., Ishitani, H., Johansson, T., Marland, G., Moreira, J., Rogner, H-H. (1996). Energy Primer, pp. 75-92 in IPCC (1996).
Nam, J., Chippar, P., Kim, W., Ju, H. (2010). Numerical analysis of gas crossover effects in polymer electrolyte fuel cells (PEFCs). Applied Energy 87, 3699-3709.
Narayanasamy, J., Anderson, A. (2003). Mechanism for the electrooxidation of carbon monoxide on platinum by H2O. Density functional theory calculation. J. Electroanalytical Chem. 554-555, 35-40.
NASA (1971). Report No. R-351 and SP-8005, May.
National Academy of Sciences (US) (1994). "Management and Disposition of Excess Weapons Plutonium". National Academy Press, Washington, DC.
Nazeeruddin, M., et al. (2001). Engineering of efficient panchromatic sensitizers for nanocrystalline TiO² -based solar cells. J. Am. Chem. Soc. 123, 1613-1624.
NCAR (1997). The NCAR Community Climate Model CCM3 with NCAR/CSM Sea Ice Model. University Corporation for Atmospheric Research, National Center for Atmospheric Research, and Climate and Global Dynamics Division, http://www.cgd.ucar.edu:80/ccr/bettge/ice.
NCEP−NCAR (1998). The NOAA NCEP−NCAR Climate Data Assimilation System I, described in Kalnay et al. (1996), data available from Univ. Columbia at http://ingrid.ldgo.columbia.edu.
NEC (2011). October 19., 2004, press release: Development of notebook PC & fuel cell unit set. http://www.nec.co.jp/press/en/0410/1901.html (accessed May 2011).
Neelis, M., Kooi, H. van der, Geerlings, J. (2004). Exergetic life cycle analysis of hydrogen production and storage systems for automotive applications. Int. J. Hydrogen Energy 29, 537-545.
Neij, L., Andersen, P., Durstewitz, M., Helby, P., Hoppe-Kilpper, M., Morthorst, P. (2003). Experience curves: a tool for energy policy assessment. EC Extool project report ENG1-CT2000.00116. Lund University.
NFC (2000). Brintbil med forbrændingsmotor et pilot projekt. Report 1763/99-003 to Danish Energy Agency, Nordvestjysk Folkecenter for Vedvarende Energi, Hurup.
Nguyen, P., Berning, T., Djilali, N. (2004). Computational model of a PEM fuel cell with serpentine gas flow channels. J. Power Sources 130, 149-157.
Nguyen, T., Knobbe, M. (2003). A liquid water management strategy for PEM fuel cell stacks. J. Power Sources 114, 70-79.
Ni, M. (2010). Modeling of a planar solid oxide fuel cell based on proton-conducting electrolyte. Int. J. Energy Res. 34, 1027-1041.
Ni, M., Leung, M., Leung, D. (2008). Technological development of hydrogen production by solid ocide electrolyzer cell (SOEC). Int. J. Hydrogen Energy 33, 2337- 2354.
Nielsen, S., Sørensen, B. (1996). Long-term planning for energy efficiency and renewable energy. Paper presented at "Renewable Energy Conference, Cairo April 1996"; revised as: Interregional power transmission: a component in planning for renewable energy technologies, Int. J. Global Energy Issues 13, No. 1-3 (2000) 170-180.
Nielsen, S., Sørensen, B. (1998). A fair market scenario for the European energy system. In "Long-Term Integration of Renewable Energy Sources into the Euro-
pean Energy System" (LTI-research group, ed.), pp. 127-186. Physica-Verlag, Heidelberg.
Nijkamp, M., Raaymakers, J., Dillen, A. van, Jong, K. de (2001). Hydrogen storage using physisorption: materials demands. Appl. Phys. A72, 619-623.
Nikiforov, B., Chigarev, A. (2011). Problems of designing fuel cell power plants for submarines. Int. J. Hydrogen Energy 36, 1226-1229.
Nishiyama, E., Murahashi, T. (2011). Water transport characteristics in the gas diffusion media of proton exchange membrane fuel cell — Role of the microporous layer. J. Power Sources 196, 1847-1854.
Nolan, J., Kolodziej, J. (2010). Modeling of an automotive fuel cell thermal system. J. Power Sources 195, 4743-4752.
NREL (2001). Documentation for software routine package ADVISOR 3.2, described by Markel et al. (2004). National Renewable Energy Laboratory, Golden, CO.
NSF (2011). ChemPrime. Teaching tool from US National Science Foundation, at: http://wiki.chemprime.chemeddl.org/index.php/Sodium_Silicide_Fueled_Bic ycles
Ochmann, F., Fürst, S., Müller, C. (2004). Industrialization of automotive hydrogen technology. In Proc. 15th World Hydrogen Energy Conf., Yokohama. 01PL-26, CD Rom, Hydrogen Energy Soc. Japan.
OECD (1994). Overview of Physics Aspects of Different Transmutation Concepts, 118 pp. Nuclear Energy Agency, Paris, Report New/Nsc/Doc(94)11.
OECD (1996). Energy Balances and Statistics of OECD and Non-OECD Countries. Annual Publications, Paris.
OECD and IAEA (1993). Uranium: Resources, production and demand. Nuclear Energy Agency, Paris.
OECD/IEA (2002a). Energy Balances of OECD Countries, 1999-2000. International Energy Agency, Paris.
OECD/IEA (2002b). Energy Balances of Non-OECD Countries, 1999-2000. International Energy Agency, Paris.
Offer, G., Howey, D., Contestabile, M., Clague, R., Brandon, N. (2010). Comparative analysis of battery electric, hydrogen fuel cell and hybrid vehicles in a future sustainable road transport system. Energy Policy 38, 24-29.
Ogata, H., Mizoguchi, Y., Mizuno, N., Miki, K., Adachi, S., Yasuoka, N., Yagi, T., Yamauchi, O., Hirota, S., Higuchi, Y. (2002). Structural studies of the carbon monoxide complex of [NiFe] hydrogenase from Desulfovibrio vulgaris Miyazaki F: suggestions for the initial activation site for dihydrogen. J. Am. Chem. Soc. 124, 11628-11635.
Ogden, J. (1999). Developing an infrastructure for hydrogen vehicles: a Southern California case study. Int. J. Hydrogen Energy 24, 709-730.
Ogden, J., Williams, R., Larson, E. (2004). Societal lifecycle costs of cars with alternative fuels/engines. Energy Policy 32, 7-27.
O'Hayre, R., Lee, S., Cha, S., Prinz, F. (2002). A sharp peak in the performance of sputtered platinum fuel cells at ultra-low platinum loading. J. Power Sources 109, 483-493.
Oh, Y., Roh, H., Jun, K., Baek, Y. (2003). A highly active catalyst, Ni/Ce-ZrO2/θ-Al2O³ , for on-site H² generation by steam methane reforming: pretreatment effect. Int. J. Hydrogen Energy 28, 1387-1392.
Ohi, J., Rossmeissl, N. (2004). Hydrogen codes and standards: an overview of US
DoE Activities. In Proc. 15th World Hydrogen Energy Conf., Yokohama. CD Rom, Hydrogen Energy Soc. Japan.
Oi, T., Wada, K. (2004). Feasibility study on hydrogen refueling infrastructure for fuel cell vehicles using off-peak power in Japan. Int. J. Hydrogen Energy 29, 347- 354.
Okada, T. (2003). Effect of ionic contaminants. In "Handbook of Fuel Cells", Vol. 3 (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 48. Wiley, Chichester.
Okazaki, K., Kokubu, R., Fushinobu, K., Uchimoto, Y. (2004). Reaction mechanisms on the Pt-based cathode catalyst of PEFCs — QMD and XAFS-analyses for atomic and electronic structures. In "Proc. 15th World Hydrogen Energy Conf., Yokohama". 29K-01, CD Rom, Hydrogen Energy Soc. Japan.
Onsager, L. (1938). J. Am. Chem. Soc., 58, 1486.
Orecchini, F., Santiangeli, A. (2010). Automaker's powertrain options for hybrid and electric vehicles. Ch. 22 in Electric and hybrid vehicles (G. Pistoia, ed.), 579-636. Elsevier, Amsterdam.
Orimo, S., Fujii, H. (2001). Materials science of Mg-Ni-based new hydrides. Appl. Phys. A72, 167-186.
Osaka Gas Co. (2004). Super compact on-site hydrogen production unit: Hyserve-30. http://www.osakagas.co.jp.
Ou, X., Zhang, X., Chang, S. (2010). Alternative fuel buses currently in use in China: Life-cycle fossil energy use, GHG emissions and policy recommendations. Energy Policy 38, 406-418.
Ovesen, C., Clausen, B., Hammershøi, B., Steffensen, G., Askgaard, T., Chorkendorff, I., Nørskov, J., Rasmussen, P., Stoltze, P., Taylor, P. (1996). A microkinetic analysis of the water-gas shift reaction under industrial conditions, J. Catalysis 158, 170-180.
Ozbilen, A., Dincer, I., Rosen, M. (2011). A comparative life cycle analysis of hydrogen production via thermochemical water splitting using a Cu-Cl cycle. Int. J. Hydrogen Energy, doi:10.1016/j.ijhydene.2010.12.035.
Pacheco, M., Sira, J., Kopasz, J. (2003). Reaction kinetics and reactor modelling for fuel processing of liquid hydrocarbons to produce hydrogen: isooctane reforming. Appl. Catalysis A: General 250, 161-175.
Paddison, S. (2001)., J. New Materials Electrochem. Sys. 4, 197.
Padró, C., Putche, V. (1999). Survey of the economics of hydrogen technologies. US National Renewable Energy Lab. Report NREL/TP-570-27079. Golden, CO.
Paladini, V., Miotti, P., Manzoni, G., Ozebec, J. (2003). Conception of modular hydrogen storage systems for portable applications. In "Hydrogen and Fuel Cells Conference. Towards a Greener World", Vancouver June, CDROM, 12 pp. Published by Canadian Hydrogen Association and Fuel Cells Canada, Vancouver.
Palenik, B., Brahamsha, B., Larimer, F., Land, M., Hauser, L., Chain, P., Lamerdin, J., Regala, W., Allen, E., McCarren, J., Paulsen, I., Dufresne, A., Partensky, F., Webb, E., Waterbury, J. (2003). The genome of a motile marine Synechococcus. Nature 424, 1037-1042.
Pallassana, V., Neurock, M., Hansen, L., Hammer, B., Nørskov, J. (1999). Theoretical analysis of hydrogen chemisorption on Pd(111), Re(0001) and PdML/Re(0001), ReML/Pd(111) pseudomorphic overlayers. Phys. Rev. B60, 6146-6134.
Palo, D., Holladay, J., Rozmiarek, R., Guzman-Leong, C., Wang, Y., Hu, J., Chin, Y.- H., Dagle, R., Baker, E. (2002). Development of a soldier-portable fuel cell power
system. Part I: a bread-board methanol fuel processor. J. Power Sources 108, 28- 34.
Panchenko, A. (2006). DFT investigation of the polymer electrolyte membrane degradation caused by OH radicals in fuel cells. J. Membrane Science 278, 269-278.
Papanikolaou, E., Venetsanos, A., Heitsch, M., Baraldi, D., Huser, A., Pujol, J., Garcia, J., Markatos, N. (2010). HySafe SBEP-V20: Numerical studies of release experiments inside a naturally ventilated residential garage. Int. J. Hydrogen Energy 35, 4747-4757.
Park, S., Kim, J., Lee, D. (2011). Development of a market penetration forecasting model for Hydrogen Fuel Cell Vehicles considering infrastructure and cost reduction effects. Energy Policy, doi:10.1016/j.enpol.2011.03.021.
Patil, A. , Dubois, T., Sifer, N., Bostic, E., Gardner, K., Quah, M., Bolton, C. (2004). Portable fuel cell systems for America's army: technology transition to the field. J. Power Sources 136, 220-225.
Patil, P. (1998). The US DoE fuel cell program. Investing in clean transportation. Paper presented at "Fuel Cell Technology Conference, London, September", IQPC Ltd, London.
Patyk, A., Höpfner, U. (1999). Ökologischer Vergleich von Kraftfarhzeugen mit verschiedenen Antriebsenergien unter besonderer Berücksichtigung der Brennstoffzelle. IFEU, Heidelberg.
Pehnt, M. (2001). Life-cycle assessment of fuel cell stacks. Int. J. Hydrogen Energy 26, 91-101.
Pehnt, M. (2002). Ganzheitliche Bilanzierung von Brennstoffzellen in der Energieund Verkehrstechnik. Dissertation, Fortschrittsberichte 6, No. 476. VDI-Verlag Dusseldorf.
Pehnt, M. (2003). Life-cycle analysis of fuel cell system components. In "Handbook of Fuel Cells - Fundamentals, Technology and Applications, Vol. 4 (Vielstich, W., Gasteiger, H., Lamm, A., eds.), ch. 94. John Wiley & Sons, Chichester.
Penev, E., Kratzer, P., Scheffler, M. (1999). Effect of the cluster size in modelling the H² desorption and dissociative adsorption on Si(001). J. Chem. Phys. 110, 3986- 3994.
Perdew, J., Burke, K., Ernzerhof, M. (1996). Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 3865-3868. Erratum: 78, 1396-1397.
Perdew, J., Kurth, S., Zupan, A., Blaha, P. (1999). Accurate density functional with correct formal properties: a step beyond the generalised gradient approximation. Phys. Rev. Lett. 82, 2544-2547. Erratum: 5179.
Perednis, D., Gauckler, L. (2004). Solid oxide fuel cells with electrolytes prepared via spray pyrolysis. Solid State Ionics 166, 229-239.
Perez, R., Hoff, T., Perez.M. (2010). Quantifying the cost of high photovoltaic penetration. Paper for American Solar Energy Society Annual Conference.
Perrette, L., Chelhaoui, S., Corgier, D. (2003). Safety evaluation of a PEMFC bus. In "Hydrogen Power — Theoretical and Engineering Solutions, Proc. Hypothesis V, Porto Conte 2003" (Marini, M., Spazzafumo, G., eds.), pp. 599-610. Servizi Grafici Editoriali, Padova.
Perrin, G. (1981). Verkehr und Technik, issue no. 9.
Peters, J., Lanzilotta, W., Lemon, B., Seefeldt, L. (1998). X-ray crystal structure of the Fe-only hydrogenase (Cpl) from Clostridium pasteurianum to 1.8 Ångström resolution. Science 282, 1853-1858.
PFC Energy (2004). Global crude oil and natural gas liquids supply forecast. Presentation at Center for Strategic & Int. Studies (CSIS), Washington DC, http:// www.csis.org/energy/040908_presentation.pdf
Pigford, T. (1991). In Transmutation as a waste management tool, pp. 97-99. Unpublished Conf. Proc.
Pinto, F., Troshina, O., Lindblad, P. (2002). A brief look at three decades of research on cyanobacterial hydrogen evolution. Int. J. Hydrogen Energy 27, 1209-1215.
Pohl, H., Malychev, V. (1997). Hydrogen in future civil aviation. Int. J. Hydrogen Energy 22, 1061-1069.
Polle, J., Kanakagiri, S., Jin, E., Masuda, T., Melis, A. (2002). Truncated chlorophyll antenna size of the photosystems — a practical method to improve microalgal productivity and hydrogen production in mass culture. Int. J. Hydrogen Energy 27, 1257-1264.
Pooley, D. (chairman) (1997). Opinion of the scientific and technical committee on a nuclear energy amplifier. European Commission, Nuclear Science and Technology Report EUR 17616 EN 1996; UKAEA Government Division, Harwell, UK, assessed 1997 at http://itumagill.fzk.de/ADS/pooley.html.
Pöpperling, R., Schwenk, W., Venkateswarlu, J. (1982). Abschätzung der Korrosionsgefärdung von Behältern und Rohrleitungen aus Stahl für Speicherung von Wasserstoff und wasserstofhältigen Gasen unter hohen Drücken. VDI Zeitschriften Reihe 5, No. 62.
Pregger, T., Graf, D., Krewitt, W., Sattler, C., Roeb, M., Möller, S. (2009). Prospects of solar thermal hydrogen production processes. Int. J. Hydrogen Energy 34, 4256- 4267.
Presting, H., Konle, J., Starkov, V., Vyatkin, A., König, U. (2004). Porous silicon for micro-sized fuel cell reformer units. Materials Sci. Eng. B108, 162-165.
Profio, P. di, Arca, S., Rossi, F., Filipponi, M. (2009). Comparison of hydrogen hydrates with existing hydrogen storage technologies: Energetic and economic evaluations. Int. J. Hydrogen Energy 34, 9173-9180.
Proton Energy Systems (2003). Unigen. Website http://www.protonenergy.com.
Qi, Z., Kaufman, A. (2003). Low Pt loading high performance cathodes for PEM fuel cells. J. Power Sources 113, 37-43.
Qian, D., Nakamura, C., Wenk, S., Wakayama, T., Zorin, N., Miyake, J. (2003). Electrochemical hydrogen evolution by use of a glass carbon electrode sandwiched with clay, poly(butylviologen) and hydrogenase. Materials Lett. 57, 1130-1134.
Qing, H., Chengzhong, Y. (2001). Application of liquid hydrogen in hypersonic aeroengine. In "Hydrogen Energy Progress XIII, Proc. 13th World Energy Conf., Beijing 2000" (Mao, Z., Veziroglu, T., eds.), pp. 670-676. Int. Assoc. Hydrogen Energy, Beijing.
Radecka, M. (2004). TiO² for photoelectrolytic decomposition of water. Thin Solid Films 451/2, 98-104.
Rasmussen, N. (1975). Project leader, Reactor Safety Study. Report WASH-1400 NUREG 75/014. US Nuclear Regulatory Commission, Washington, DC.
Rasten, E., Hagen, G., Tunold, R. (2003). Electrocatalysis in water electrolysis with solid polymer electrolyte. Electronica Acta 48, 3945-3952.
Reed, T. (ed.) (1981). "Biomass gasification". Noyes Data Corp., Park Ridge, NJ.
Ren, H., Gao, W. (2010). Economic and environmental evaluation of micro CHP systems with different operating modes for residential buildings in Japan. Energy
and Buildings 42, 853-861.
Reuter, K., Frenkel, D., Scheffler, M. (2004). The steady state of heterogeneous catalysis, studied by first-principle statistical mechanics. Phys. Rev. Lett. 93, 116105.
RIT (1997). Accelerator driven systems. 3 pp., Royal Institute of Technology, Stockholm, Website http://www.neutron.kth.se/introduction/.
Robinson, J. (ed.) (1980). "Fuels from Biomass". Noyes Data Corp., Park Ridge, NJ.
Roddy, D. (2004). Making a viable fuel cell industry happen in the Tees Valley. Fuel Cells Bulletin Jan. 10-12.
Rodríguez, C., Riso, M., Yob, J., Ottogalli, R., Cruz, R., Aisa, S., Jeandrevin, G., Leiva, E. (2010). Analysis of the potential for hydrogen production in the province of Córdoba, Argentina, from wind resources. Int. J. Hydrogen Energy 35, 5952-5956.
Roh, H., Jun, K., Dong, W., Chang, J., Park, S., Joe, Y. (2002). Highly active and stable Ni/Ce-ZrO² catalyst for H² production from methane. J. Molec. Catalysis A: Chemical 181, 137-142.
Roos, M., Batawi, E., Harnisch, U., Hocker, T. (2003). Efficient simulation of fuel cell stacks with the volume averaging method. J. Power Sources 118, 86-95.
Rosa, M. de la, Navarro, J., Díaz-Quintana, A., Cerda, B. de la, Molina-Heredia, F., Balme, A., Murdoch, P., Díaz-Moreno, I., Durán, R., Hervás, M. (2002). An evolutionary analysis of the reaction mechanisms of photosystem I reduction by cytochrome c⁶ and plastocyanin. Bioelectrochemistry 55, 41-45.
Rosenberg, E., Fidje, A., Espegren, K., Stiller, C., Svensson, A., Møller-Holst, S. (2010). Market penetration analysis of hydrogen vehicles in Norwegian passenger transport towards 2050. Int. J. Hydrogen Energy 35, 7267-7279.
Rosi, N., Eckert, J., Eddaoudi, M., Vodak, D., Kim, J., O'Keeffe, M., Yaghi, O. (2003). Hydrogen storage in microporous metal-organic frameworks. Science 300, 1127- 1129.
Rostrup-Nielsen, J. (2000). New aspects of syngas production and use. Catalysis Today 63, 159-164.
Rostrup-Nielsen, J., Aasberg-Petersen, K. (2003). Steam reforming, ATR, partial oxidation: catalysts and reaction engineering. In "Handbook of Fuel Cells" (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 14. Wiley, Chichester.
Rouss, V., Candusso, D., Charon, W. (2008). Mechanical behaviour of a fuel cell stack under vibrating conditions linked to aircraft applications part II: Threedimensional modelling. Int. J. Hydrogen Energy 33, 6281-6288.
Royle, M., Willoughby, D. (2011). Consequences of catastrophic releases of ignited and unignited hydrogen jet releases. Int. J. Hydrogen Energy 36, 2688-2692.
Rubbia, C. (1994). The energy amplifier, In "Proc. 8 th Journées Saturne, Saclay", pp. 115-123. http://db.nea.fr/html/trw/docs/saturne8/ (last accessed 1999).
Rubbia, C., Rubio, J., Buono, S., Carminati, F., Fiétier, N., Galvez, J., Gelès, J., Kadi, Y., Klapisch, R., Mandrillon, P., Revol, J., Roche, C. (1995). Conceptual design of a fast neutron operated high power energy amplifier. European Organization for Nuclear Research, preprint collection CERN/AT-95-44.
Rubbia, C., Rubio, J. (1996). A tentative programme towards a full scale energy amplifier. European Organization for Nuclear Research, preprint CERN/LHC/96- 11(ET). 36 pp. Website http://sundarssrv2.cern.ch/search.html
Runge, E., Gross, E. (1984). Density-functional theory for time-dependent systems. Phys. Rev. Lett. 52, 997-1000.
Ryu, J., Park, Y., Sunwoo, M. (2010). Electric powertrain modeling of a fuel cell hy-
brid electric vehicle and development of a power distribution algorithm based on driving mode recognition. J. Power Sources 195, 5735-5748.
Saiki, Y., Amao, Y. (2003). Bio-mimetic hydrogen production from polysaccharide using the visible light sensitation of zinc porphyrin. Biotechnology Bioeng., 82, 710-714.
Saito, M., Takeuchi, M., Watanabe, T., Toyir, J., Luo, S., Wu, J. (1997). Methanol synthesis from CO² and H² over a Cu/ZnO-based multicomponent catalyst, Energy Conversion Management 38, S403-S408.
Sandrock, G., Thomas, G. (2001). Database administrators for an online hydride database of the International Energy Agency and the US Department of Energy at http://hydpark.ca.sandia.gov.
Satija, R., Jacobsen, D., Arif, M., Werner, S. (2004). In situ neutron imaging techniques for evaluation of water management systems in operating PEM fuel cells. J. Power Sources 129, 238-245.
Sattler, G. (2000). Fuel cells going on-board. J. Power Sources 86, 61-67.
Satyapal, S., Petrovic, J., Read, C., Thomas, G., Ordaz, G. (2007). The U.S. Department of Energy's National Hydrogen Storage Project: Progress towards meeting hydrogen-powered vehicle requirements. Catalysis Today 120, 246-256.
Sauk, J., Byun, J., Kim, H. (2004). Grafting of styrene to Nafion membranes using supercritical CO² impregnation for direct methanol fule cells. J. Power Sources 132, 59-63.
Schaefer, A., Horn, H., Ahlrichs, R. (1992). Fully optimized contracted Gaussian basis sets for atoms Li to Kr. J. Chem. Phys. 97, 339.
Schaefer, A., Huber, C., Ahlrichs, R. (1994). J. Chem. Phys. 100, 5829.
Schaeffer, G., Alsema, E., Seebregts, A., Buerskens, L., Moor, H. de, Durstewitz, M., Perrin, M., Boulanger, P., Laukamp, H., Zuccarro, C. (2004). Synthesis report Photex-project ECN Report, Petten.
Scharff, M. (1969). "Elementary quantum mechanics". John Wiley & Sons, London.
Schlamadinger, B., Marland, G. (1996). Full fuel cycle carbon balances of bioenergy and forestry options. Energy Conversion Management 37, 813-818.
Schlapbach, L., Züttel, A. (2001). Hydrogen-storage materials for mobile applications. Nature 414 353-358.
Schlesinger, H., Brown, H. (1940). Metallo borohydrides, III: Lithium borohydride. J. Am. Chem. Soc. 62, 3429-3435.
Schober, T. (2001). Tubular high-termperature proton conductors: transport numbers and hydrogen injection. Solid State Ionics 139, 95-104.
Schultz, M., Diehl, T., Brasseur, G., Zittel, W. (2003). Air pollution and climateforcing impacts of a global hydrogen economy. Science 302, 624-527.
Schültz, M., Werner, H-J., Lindh, R., Manby, F. (2004). Analytical energy gradients for local second-order Møller-Plesset perturbation theory using density fitting approximations. J. Chem. Phys. 121, 737-750.
Schulze, M., Knöri, T., Schneider, A., Gülzow, E. (2004). Degradation of sealings for PEFC test cells during fuel cell operation. J. Power Sources 127, 222-229.
Schuster, M., Meyer, W., Schuster, M., Kreuer, K. (2004). Towards a new type of anhydrous organic proton conductor based on immobilized imidazole. Chem. Meterials 16, 329-337.
Scrosati, B. (1995). Challenge of portable power. Nature 373, 557-558.
Seidenberger, K., Wilhelm, F., Schmitt, T., Lehnert, W., Scholta, J. (2011). Estimation
of water distribution and degradation mechanisms in polymer electrolyte membrane fuel cell gas diffusion layers using a 3D Monte Carlo model. J. Power Sources 196, 5317-5324.
Semelsberger, T., Brown, L., Borup, R., Inbody, M. (2004). Equilibrium products from autothermal processes for generating hydrogen-rich fuel-cell feeds. Int. J. Hydrogen Energy 29, 1047-1064.
Serincan, M., Pasaogullari, U., Molter, T. (2010). Modeling the cation transport in an operating polymer electrolyte fuel cell (PEFC). Int. J. Hydrogen Energy 35, 5539- 5551.
Shaffer, C., Wang, C-Y. (2010). High concentration methanol fuel cells: Design and theory. J. Power Sources 195, 4185-4195.
Shah, A., Luo, K., Ralph, T., Walsh, F. (2011). Recent trends and developments in polymer electrolyte membrane fuel cell modelling. Electrochimica Acta 56, 3731- 3757.
Shamardina, O., Chertovich, A., Kulikovsky, A., Khokhlov, A. (2010). A simple model of a high temperature PEM fuel cell. Int. J. Hydrogen Energy 35, 9954- 9962.
Shang, C., Bououdina, M., Song, Y., Guo, Z. (2004). Mechanical alloying and electronic simulations of (MgH² + M) systems (M = Al, Ti, Fe, Ni, Cu and Nb) for hydrogen storage. Int. J. Hydrogen Energy 29, 73-80.
Shayegan, S., Hart, D., Pearson, P., Bauen, A., Joffe, D. (2004). Hydrogen infrastructure costs: what are the important variables? In "Hydrogen Power — Theoretical and Engineering Solutions, Proc. Hypothesis V, Porto Conte 2003" (Marini, M., Spazzafumo, G., eds.), pp. 499-508. Servizi Grafici Editoriali, Padova.
Shen, P., Shi, Q., Hua, Z., Kong, F., Wang, Z., Zhuang, S., Chen, D. (2003). Analysis of microsystins in cyanobacteria blooms and surface water samples from Meilang Bay, Taihu Lake, China. Environment Int. 29, 641-647.
Shi, J., Xue, X. (2010). CFD analysis of a novel symmetrical planar SOFC design with micro-flow channels. Chemical Engineering J. 163, 119-125.
Shimizu, K., Fukagawa, M., Sakanishi, A. (2004). Development of PEM water electrolysis type hydrogen production system. In "15th World Hydrogen Energy Conference, Yokohama 2004". Hydrogen Energy Systems Soc. of Japan (CDROM).
Shoiji, M., Houki, Y., Ishiyama, T. (2001). Feasibility of the high-speed hydrogen engine. In "Hydrogen Energy Progress XIII, Proc. 13th World Energy Conf., Beijing 2000" (Mao, Z., Veziroglu, T., eds.), pp. 641-647. Int. Assoc. Hydrogen Energy, Beijing.
Shore, L., Farrauto, R. (2003). PROX catalysts. In "Handbook of Fuel Cells" (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 18. Wiley, Chichester.
Siddique, N., Liu, F. (2010). Process based reconstruction and simulation of a threedimensional fuel cell catalyst layer. Electrochimica Acta 55, 5357-5366.
Siegel, N., Ellis, M., Nelson, D., Spakovsky, M. von (2003). Single domain PEMFC model based on agglomerate catalyst geometry. J. Power Sources 115, 81-89.
Sierra, J., Moreira, J., Sebastian, P. (2011). Numerical analysis of the effect of different gas feeding modes in a proton exchange membrane fuel cell with serpentine flow-field. J. Power Sources 196, 5070-5076.
SiGNa (2011). Fuel cells using hydrogen produced from sodium silicide; bicycle use; myFC PowerTrekk charger. http://www.signachem.com
Simbeck, D., Chang, E. (2002). Hydrogen supply: cost estimate for hydrogen pathways — scoping analysis. National Renewable Energy Lab., Report NREL/SR-540-32525, Golden, CO.
Simon, G. de, Parodi, F., Fermeglia, M., Taccani, R. (2003). Simulation of process for electrical energy production based on molten carbonate fuel cells. J. Power Sources 115, 210-218.
Sistiaga, M., Pierna, A. (2003). Application of amorphous materials for fuel cells. J. Non-Crystalline Solids 329, 184-187.
Sivertsen, B., Djilali, N. (2005). CFD-based modelling of proton exchange membrane fuel cells. J. Power Sources 141, 65-78.
Slater, J. (1928). The self consistent field and the structure of atoms, Phys. Rev. 32, 339-348.
Slater, J. (1951). Phys. Rev. 81, 385.
Sljivancanin, Z., Hammer, B. (2002). Oxygen dissociation at close-packed Pt terraces, Pt steps, and Ag-covered Pt steps studied with density functional theory. Surface Sci. 515, 235-244.
Sluiter, M., Belosludov, R., Jain, A., Belosludov, R., Adachi, H., Kawazoe, Y., Higuchi, K., Otani, T. (2003). Ab initio study of hydrogen hydrate clathrates for hydrogen storage within the ITBL environment. In "Proc. ISHPC Conference 2003" (Veidenbaum, A. et al., eds.) , pp. 330-341. Springer-Verlag, Berlin.
SMAB (1978). "Metanol sam drivmedel," Annual Report, Svensk Metanol-utveckling AB, Stockholm.
Smith, B. (2002). Nitrogenase reveals its inner secrets. Science 297, 1654-1655.
Sobrino, F., Monroy, C., Pérez, J. (2010). Critical analysis on hydrogen as an alternative to fossil fuels and biofuels for vehicles in Europe. Renewable and Sustainable Energy Revs. 14, 772-780.
Sørensen, B. (1975). Energy and resources. Science 189, 255-260, and in "Energy: Use, Conservation and Supply" (Abelson, P., and Hammond, A., eds.), Vol. II, pp. 23-28. Am. Ass. Advancement of Science, Washington, DC (1978).
Sørensen, B. (1979a). "Renewable Energy". Academic Press, London.
Sørensen, B. (1979b). Nuclear power: the answer that became a question. An assessment of accident risks. Ambio 8, 10-17.
Sørensen, B. (1981). A combined wind and hydro power system. Energy Policy, March, pp. 51−55.
Sørensen, B. (1982). Comparative risk assessment of total energy systems. In "Health Impacts of Different Sources of Energy", pp. 455-471. Report IAEA-SM-254/105, Int. Atomic Energy Agerncy, Vienna.
Sørensen, B. (1983). Stationary applications of fuel cells. In "Solid State Protonic Conductors II — for Fuel Cells and Sensors", pp. 97-108 (Goodenough, J., Jensen, J., Kleitz, M., eds.), Odense University Press, Odense.
Sørensen, B. (1984). Energy storage. Ann. Rev. Energy 9, 1-29.
Sørensen, B. (1987). Chernobyl accident: assessing the data. Nuclear Safety 28, 443- 447.
Sørensen, B. (1991). Energy conservation and efficiency measures in other countries. Greenhouse Studies No. 8. Commonwealth of Australia, Dept. Arts, Sport, Environment, Tourism and Territories, Canberra.
Sørensen, B. (1993). What is life-cycle analysis? In "Life-cycle Analysis of Energy Systems", pp. 21-53. Workshop Proceedings, OECD Publications, Paris.
Sørensen, B. (1995). History of, and recent progress in, wind−energy utilization. Ann. Rev. Energy & Environment 20, 387−424.
Sørensen, B. (1996a). Life-cycle approach to assessing environmental and social externality costs. In "Comparing Energy Technologies", Ch. 5, pp. 297-331. International Energy Agency, IEA/OECD, Paris.
Sørensen, B. (1996b). Scenarios for greenhouse warming mitigation. Energy Conversion & Management 37, 693-698.
Sørensen, B. (1996c). Does wind energy utilization have regional or global climate impacts? "Proc. 1996 European Union Wind Energy Conference", pp. 191-194. H. Stephens & Ass., Bedford UK.
Sørensen, B. (1997a). Impacts of energy use. In "Human Ecology, Human Economy" (Diesendorf and Hamilton, eds.), pp. 243-266. Allen and Unwin, New South Wales.
Sørensen, B. (1997b). Externality estimation of greenhouse warming impacts, Energy Conversion Management 38, S643−S648.
Sørensen, B. (1998). Brint (Strategy note from Danish Hydrogen Committee). Danish Energy Agency, Copenhagen.
Sørensen, B. (1999). Long-term scenarios for global energy demand and supply: four global greenhouse mitigation scenarios. Final Report from a project performed for the Danish Energy Agency, IMFUFA Texts 359, Roskilde University, pp. 1- 166.
Sørensen, B. (2000). Role of hydrogen and fuel cells in renewable energy systems. In "Renewable Energy: the Energy for the 21st Century", Proc. World Renewable Energy Conference VI, Reading, Vol. 3, pp. 1469-1474. Pergamon, Amsterdam.
Sørensen, B. (2002a). Handling fluctuating renewable energy production by hydrogen scenarios. In "14th World Hydrogen Energy Conference", Montreal, 2002. File B101c, 9 pp. CD published by CogniScience Publ. for l'Association Canadienne de l'Hydrogène, revised CD issued 2003.
Sørensen, B. (2002b). Understanding photoelectrochemical solar cells. In "PV in Europe, from PV Technology to Energy Solutions", Roma Int. Conf. (Bal, J., et al., eds.), pp. 3-8. WIP Munich and ETA Florence.
Sørensen, B. (2003a). Scenarios for future use of hydrogen and fuel cells. In "Hydrogen and Fuel Cells Conference. Towards a Greener World", Vancouver June, CDROM, 12 pp. Published by Canadian Hydrogen Association and Fuel Cells Canada, Vancouver.
Sørensen, B. (2003b). Time-simulations of renewable energy plus hydrogen systems. In "Hydrogen Power — Theoretical and Engineering Solutions. Proc. Hypothesis V, Porto Conte 2003" (Marini, M., Spazzafumo, G., eds.), pp. 35-42. Servizi Grafici Editoriali, Padova.
Sørensen, B. (2003c). Hydrogen scenarios using fossil, nuclear or renewable energy. In "Proc. 1st European Hydrogen Energy Conference, Grenoble 2003", paper CO5/78, CDROM, 14 pp. Published by Association Francaise de l'Hydrogène, Paris.
Sørensen, B. (2004a). "Renewable Energy". 3rd ed. Elsevier Academic Press, Burlington. MA. Previous editions 1979; 2000, new edition 2010.
Sørensen, B. (2004b). Total life-cycle analysis of PEM fuel cell car. In "Proc. 15th World Hydrogen Energy Conf., Yokohama". 29G-09, CD Rom, Hydrogen Energy Soc. Japan.
Sørensen, B. (2004c). Quantum chemical exploration of PEM fuel cell processes. In "Proc. 15th World Hydrogen Energy Conf., Yokohama". 28K-02, CD Rom, Hydrogen Energy Soc. Japan.
Sørensen, B. (2004d). Readiness of hydrogen technologies. Hydrogen and fuel cell futures conference, Perth, CDROM by Western Australia Govt. Dept. Planning & Infrastructure.
Sørensen, B. (2004e). The last oil? A strategy for development of hydrogen technologies in Denmark, a report commissioned by the Danish Energy Agency. Both available at http://energy.ruc.dk
Sørensen, B. (2005a). Understanding fuel cells on a quantum physics level. In "Proc. World Hydrogen Technology Conference, Singapore", paper A16-229 on CDROM, IESE, Nanyang University.
Sørensen, B. (2005b). Quantum chemical exploration of hydrogen energy storage in metal hydrides. In "Proc 2nd European Hydrogen Energy Conf., Zaragoza", pp. 112-113, also on CDROM, Idea Madrid.
Sørensen, B. (2005c). On the road performance simulation and appraisal of hydrogen vehicles (1). In "Proc. World Hydrogen Technology Conference, Singapore", paper A16-230 on CDROM, IESE, Nanyang University.
Sørensen, B. (2005d). On the road performance simulation and appraisal of hydrogen vehicles (2). In "Proc. 2nd European Hydrogen Energy Conference, Zaragoza", on CDROM, European Hydrogen Association.
Sørensen, B. (2006a). Appraisal of bio-hydrogen production schemes. In "Proc 16th World Hydrogen Energy Conf., Lyon", Paper S07-109, IHEA CDROM #109, Sevanova, France.
Sørensen, B. (2006b). Comparison between hydrogen fuel cell vehicles and bio-diesel vehicles. In "Proc 16th World Hydrogen Energy Conf., Lyon", Paper S24-111, IHEA CDROM #111, Sevanova, France.
Sørensen, B. (2006c). Description of hydrogen storage in hydrides and related compounds by quantum chemical calculations. In "Proc 16th World Hydrogen Energy Conf., Lyon", Paper S14-110, IHEA CDROM #110, Sevanova, France.
Sørensen, B. (2007a). Biological hydrogen production. Final Report to Nordic Energy Research (Oslo) from project "Bioenergy, 2004-2007". ENCPAC, Roskilde University.
Sørensen, B. (2007b). On the road performance simulation of hydrogen and hybrid cars. Int. J. of Hydrogen Energy 32, 683-686.
Sørensen, B. (2007c). Assessing current vehicle performance and simulating the performance of hydrogen and hybrid cars. Int. J. of Hydrogen Energy 32, 1597-1604.
Sørensen, B. (2007d). Geological hydrogen storage. In "Proc. World Hydrogen Technology Conf., Montecatini", on CDROM, It-Forum.
Sørensen, B. (2007e). Quantum mechanical description of catalytical processes at the electrodes of low-temperature hydrogen-oxygen fuel cells. In "Proc. Hypothesis VII International Conf., Merida, Mexico", CDROM, ISBN 9686114211.
Sørensen, B. (2008a). A renewable energy and hydrogen scenario for northern Europe. Int. J. of Energy Research, 32, 471-500 (published online 2007).
Sørensen, B. (2008b). A sustainable energy future: Construction of demand and renewable energy supply scenarios. Int. J. Energy Research 32, 436-470 (published online 2007).
Sørensen, B. (2008c). A new method for estimating off-shore wind potentials. Int. J.
Green Energy 5, 139-147.
Sørensen, B. (2009). Scenarios for the roles of hydrogen in a future energy system based on renewable energy. Int. J. Nuclear Hydrogen Production and Application, 1 (4), 287-294.
Sørensen, B. (2010a). Renewable Energy. Physics, engineering, environmental impacts, economics & planning. 4th Edition, Academic Press-Elsevier, Burlington.
Sørensen, B. (2010b). On the road performance simulation of battery, hydrogen and hybrid cars. Ch. 10 in Electric and Hybrid Vehicles (G. Pistoia, ed.), 247-273. Elsevier, Amsterdam.
Sørensen, B. (2011a). Life-cycle Analysis of Energy Systems. From Methodology to Applications. RSC Publishing, Cambridge.
Sørensen, B. (2011b). A history of energy. Northern Europe from Stone Age to the Present Day. Earthscan-Taylor & Francis, London and Cambridge.
Sørensen, B. (2011c). Mapping potential renewable energy resources in the Mediterranean region. Ch. 3 in Recent developments in energy and environmental research (E. Maleviti, ed.), 23-36. ATINER SA, Athens, ISBN 978-960-85411-2-2.
Sørensen, B., Meibom, P., Nielsen, L., Karlsson, K., Pedersen, A., Lindboe, H., Bregnebæk, L. (2008). Comparative assessment of hydrogen storage and international electricity trade for a Danish energy system with wind power and hydrogen/fuel cell technologies. Final Report for Danish Energy Authority Project EFP05 033001/033001-0021. EECG Research Paper No. 1/08, available at Roskilde University website http://rudar.ruc.dk/handle/1800/2431
Sørensen, B., Petersen, A., Juhl, C., Ravn, H., Søndergren, C., Simonsen, P., Jørgensen, K., Nielsen, L., Larsen, H., Morthorst, P., Schleisner, L., Sørensen, F., Petersen, T. (2001). Project report to Danish Energy Agency (in Danish): Scenarier for samlet udnyttelse af brint som energibærer i Danmarks fremtidige energisystem, IMFUFA Texts No. 390, 226 pp., Roskilde University; report download at http://rudar.ruc.dk/handle/1800/3500, file IMFUFA_390.pdf
Sørensen, B., Petersen, A., Juhl, C., Ravn, H., Søndergren, C., Simonsen, P., Jørgensen, K., Nielsen, L., Larsen, H., Morthorst, P., Schleisner, L., Sørensen, F., Petersen, T. (2004). Hydrogen as an energy carrier: scenarios for future use of hydrogen in the Danish energy system. Int. J. Hydrogen Energy 29, 23-32 (summary of Sørensen et al., 2001).
Sørensen, B., Sørensen, F. (2000). A hydrogen future for Denmark. In "Hydrogen Energy Progress XIII, Proc. 13th World Energy Conf., Beijing 2000" (Mao, Z., Veziroglu, T., eds.), pp. 35-40. Int. Assoc. Hydrogen Energy, Beijing.
Spakovsky, M., Olsommer, B. (2002). Fuel cell systems and system modeling and analysis perspectives for fuel cell development. Energy Conversion Management 43, 1249-1257.
Spath, P., Mann, M. (2001). Life cycle assesment of hydrogen production via natural gas steam reforming. Revised USDoE contract report NREL/TP-570-27637, National Renewable Energy Lab., Golden, CO.
Spielmann, M., Haan, P. d., Scholz, R. (2008). Environmental rebound effects of highspeed transport technologies: a case study of climate change rebound effects of a future underground maglev train system. J. Cleaner Production 16, 1388-1398.
Spohr, E. (1999). Molecular simulation of the electrochemical double layer. Electrochimica Acta 44, 1697-1705.
Sprague, I., Dutta, P. (2011). Role of the diffuse layer in acidic and alkaline fuel cells.
Electrochimica Acta, doi:10.1016/j.electacta.2011.02.060.
Springer, T., Zawodzinski, T., Gottesfeld, S. (1991). Polymer electrolyte fuel cell model. J. Electrochem. Soc. 138, 2334-2342.
Staroverov, V., Scuseria, G., Tao, J., Perdew, J. (2004). Tests of a ladder of density functionals for bulk solids and surfaces. Phys. Rev. B 69, 075102.
Starz, K., Auer, E., Lehmann, T., Zuber, R. (1999). Characteristics of platinum-based
electrocatalysts for mobile PEMFC applications. J. Power Sources 84, 167-172. Steele, B., Heinzel, A. (2002). Materials for fuel-cell technologies. Nature 414, 345-352.
Steinberg, M., Takahashi, H., Ludewig, H., Powell, J. (1977). Linear accelerator fission product transmuter, Paper for American Nuclear Society meeting, New York, 17 pp.
Stipe, B., Rezaei, M., Ho, W., Gao, S., Persson, M., Lundqvist, B. (1997). Singlemolecule dissociation by tunneling electrons. Phys. Rev. Lett. 78, 4410-4413.
Stockie, J. (2003). Modeling hydrophobicity in a porous fuel cell electrode. Paper for "Computational Fuel Cell Dynamics II", Banff Int. Res. Center, USA.
Stockie, J., Promislow, K., Wetton, B. (2003). A finite volume method for multicomponent gas transport in a porous fuel cell electrode. Int. J. Numerical Methods Fluids 41, 577-599.
Strazza, C., Borghi, A. d., Costamagna, P., Traverso, A., Santin, M. (2010). Comparative LCA of methanol-fuelled SOFCs as auxiliary power systems on-board ships. Applied Energy 87, 1670-1678.
Stroebel, D., Choquet, Y., Popot, J., Picot, D. (2003). An atypical haem in the cytochrome b⁶ f complex. Nature 426, 413-418.
Sudan, P., Wenger, P., Mauron, P., Gremaud, R., Züttel, A. (2004). Reversible properties of LiBH⁴ . In "Hydrogen Power — Theoretical and Engineering Solutions, Proc. Hypothesis V, Porto Conte 2003" (Marini, M., Spazzafumo, G., eds.), pp. 433-440. Servizi Grafici Editoriali, Padova.
Sullivan, M., Waterbury, J., Chisholm, S. (2003). Cyanophages infecting the oceanic cyanobacterium Prochlorococcus. Nature, 242, 1047-1052.
Summers, W., Gorensek, M., Danko, E., Schultz, K., Richards, M., Brown, L. (2004) Analysis of Economic and Infrastructure Issues Associated with Hydrogen Production from Nuclear Energy. In Proc. 15th World Hydrogen Energy Conf., Yokohama. CD Rom, Hydrogen Energy Soc. Japan.
Sun, Y., Ogden, J., Delucchi, M. (2010). Societal lifetime cost of hydrogen fuel cell vehicles. Int. J. Hydrogen Energy 35, 11932-11946.
Superfoss (1981). En dansk industri. Danmarks Radio Skole-TV, Glostrup.
Suppes, G., Lopes, S., Chiu, C. (2004). Plug-in fuel cell hybrids as transition technology to hydrogen infrastructure. Int. J. Hydrogen Energy 29, 369-374.
Svensson, F., Hasselrot, A., Moldanova, J. (2004). Reduced environmental impact by lowered cruise altitude for liquid hydrogen-fuelled aircraft. Aerospace Sci. Tech. 8, 307-320.
Sveshnikov, D., Sveshnikova, N., Rao, K., Hall, D. (1997). Hydrogen metabolism of mutant forms of Anabaena variabilis in continuous cultures and under nutritional stress. FEBS Lett. 147, 297-301.
Syed, F., Fowler, M., Wan, D., Maniyali, Y. (2010). An energy demand model for a fleet of plug-in fuel cell vehicles and commercial building interfaced with a clean energy hub. Int. J. Hydrogen Energy 35, 5154-5163.
Szczygiel, J., Szyja, B. (2004). Diffusion of hydrocarbons in the reforming catalyst:
molecular modelling. J. Molecular Graphics Modelling 22, 231-239.
Takahashi, K. (1998). Development of fuel cell electric vehicles. Paper presented at "Fuel Cell Technology Conference, London, September", IQPC Ltd., London.
Takeuchi, K., Fujioka, Y., Kawasaki, Y., Shirayama, Y. (1997). Impacts of high concentrations of CO² on marine organisms: a modification of CO² ocean sequestration, Energy Conversion Management 38, S337-S341.
Takimoto, M. (2004). Development of fuel cell hybrid vehicles in Toyota. In Proc. 15th World Hydrogen Energy Conf., Yokohama. 01PL-08, CD Rom, Hydrogen Energy Soc. Japan.
Tamagnini, P., Axelsson, R., Lindberg, P., Oxelfelt, F., Wünchiers, R., Lindblad, P. (2002). Hydrogenases and hydrogen metabolism of cyanobacteria. Microbiol. Mol. Biol. Rev. 66, 1-20.
Tang, Y., Yuan, W., Pan, M., Wan, Z. (2011). Experimental investigation on the dynamic performance of a hybrid PEM fuel cell/battery system for lightweight electric vehicle application. Applied Energy 88, 68-76.
Tanimoto, K., Kojima, T., Yanagida, M., Nomura, K., Miyazaki, Y. (2004). Optimization of the electrolyte composition in a (Li0.52Na0.48)2-2xAExCO³ (AE = Ca and Ba) molten carbonate fuel cell. J. Power Souirces 131, 256-260.
Tao, S., Notten, P., Santen, A. van, Jansen, A. (2011). DFT studies of hydrogen storage properties of Mg0.75Ti0.25. J. Alloys & Compounds 509, 210-216.
Tatsumi, K., Tanaka, I., Inui, H., Tanaka, K., Yamaguchi, M., Adachi, H. (2001). Atomic structures and energetics of LaNi⁵ -H solid solution and hydrides. Phys. Rev. B64, #184105 (10 pp).
Tayhas, G., Palmore, R. (2004). Bioelectric power generation. Trends Biotechnology 22, 99-100.
Taylor, J., Alderson, J., Kalyanam, K., Lyle, A., Phillips, L. (1986). Technical and economic assessment of methods for the storage of large quantities of hydrogen. Int. J. Hydrogen Energy 11, 5-22.
Tchouvelev, A., Howard, G., Agranat, V. (2004). Comparison of standards requirements with CFD simulations for determining sizes of hazardous locations in hydrogen energy station. In "Proc. 15th World Hydrogen Energy Conf., Yokohama". CD Rom, Hydrogen Energy Soc. Japan.
Thomas, C. (2009). Transportation options in a carbon-constrained world: Hybrids, plug-in hybrids, biofuels, fuel cell electric vehicles, and battery electric vehicles. Int. J. Hydrogen Energy 34, 9279-9296.
Thomas, L. (1927). Proc. Cambridge Phil. Soc. 23, 542.
Toevs, J., Bowman, C., Arthur, E., Heighway, E. (1994). Progress in accelerator driven trans-mutation technologies, In "Proc. 8th Journées Saturne, Saclay", pp. 22-28. Website: http://db.nea.fr/html/trw/docs/saturne8/ (accessed 1999).
Tokyo Gas Co. (2003). Press Release 16/10/03, Corporate Communications Dept.
Tokyo Gas Co. (2004). Completion of the plans by Tokyo Gas and the Railway Technical Res. Inst. for railway hydrogen station. http://www-tokyo-gas.co.jp
Tomonou, Y., Amao, Y. (2004). Effect of micellar species on photoinduced hydrogen production with Mg chlorophyll-a from spirulina and colloidal platimum. Int. J. Hydrogen Energy 29, 159-162.
Toyota (1996). High-performance hydrogen−absorbing alloy. Website http:// www.toyota.co.jp/e/november_96/electric_island/press.html.
Trebst, A. (1974). Ann. Rev. Plant Physiol. 25, 423-447.
Tromp, T., Shia, R-L., Allen, M., Eiler, J., Young, Y. (2003). Potential environmental impact of a hydrogen economy on the stratosphere. Science 300, 1740-1742.
Troshina, O., Serebryakova, L., Sheremetieva, M., Lindblad, P. (2002). Production of H2 by the unicellular cyanobacterium Gloeocapsa alpicola CALU 743 during fermentation. Int. J. Hydrogen Energy 27, 1283-1289.
Tse, L., Wilkins, S., McGlashan, N., Urban, B., Martinez-Botas, R. (2011). Solid oxide fuel cell/gas turbine trigeneration system for marine applications. J. Power Sources 196, 3149-3162.
Tsubomura, H., Matsumura, M., Nomura, Y., Amamiya, T. (1976). Dye sensitised zinc oxide: aqueous electrolyte: platinum photocell. Nature 261, 402-403
Tsuchiya, H., Inui, M., Fukuda, K. (2004). Penetration of fuel cell vehicles and hydrogen infrastructure. In Proc. 15th World Hydrogen Energy Conf., Yokohama. 30A-01, CD Rom, Hydrogen Energy Soc. Japan.
Tsuchiya, H., Kobayashi, O. (2004). Mass production cost of PEM fuel cell by learning curve. Int. J. Hydrogen Energy 29, 985-990.
Tsygankov, A., Fedorov, A., Kosourov, S., Rao, K. (2002a). Hydrogen production by cyanobacteria in an automated outdoor photobioreactor under aerobic conditions. Biotechnology Bioeng. 80, 777-783.
Tsygankov, A., Kosourov, S., Seibert, M., Ghirardi, M. (2002b). Hydrogen photoproduction under continuous illumination by sulphus-deprived, synchronous Chlamydomonas reinhardtii cultures. Int. J. Hydrogen Energy 27, 1239-1244.
Tu, H., Stimming, U. (2004). Advances, aging mechanisms and lifetime in solid-oxide fuel cells. J. Power Sources 127, 284-293.
Tzeng, G-H., Lin, C-W., Opricovic, S. (2004). Multi-criteria analysis of alternative-fuel buses for public transportation. Energy Policy (in print).
Ueno, Y., Otauka, S., Morimoto, M. (1996). Hydrogen production from industrial wastewater by anaerobic microflora in chemostat culture. J. Ferment. Bioeng. 82, 194-197.
Uherek, E., et al. (2010). Transport impacts on atmosphere and climate: Land transport. Atmospheric Envir. 44, 3772-4816.
UK Treasury (2004). GDP deflators at market prices.In "Economic data & tools", http://www.hm-treasury.gov.uk/economic_data_and_tools/gdp_deflators.
UKDTI (2003). A fuel cell vision for the UK the first steps. Taking the White Paper forward. UK Department of Trade and Industry, The Carbon Trust and EPSRC.
Ulleberg, Ø., Nakken, T., Eté, A. (2010). The wind/hydrogen demonstration system at Utsira in Norway: Evaluation of system performance using operational data and updated hydrogen energy system modeling tools. Int. J. Hydrogen Energy 35, 1841-1852.
UltraCell (2011). Mobile power: brochures for XX25 and XX55 fuel cell systems. http://www.ultracellpower.com
Um, S., Wang, C. (2004). Three-dimensional analysis of transport and electrochemical reactions in polymer electrolyte fuel cells. J. Power Sources 125, 40-51.
UN (1996). "Populations 1996, 2015, 2050". United Nations Population Division and UNDP: available at the website http://www.undp.org/popin/wdtrends/pop /fpop.htm.
UN (1997). "UN urban and rural population estimates and projections as revised in 1994". United Nations Population Division and UNDP, Washington. Website: http://www.undp.org/popin/wdtrends/urban.html.
USDoE (2002a). National hydrogen energy roadmap. Towards a more secure and cleaner energy future for America. United States Department of Energy, Washington, DC.
USDoE (2002b). A technology roadmap for Generation IV Nuclear Energy Systems. US Department of Energy NERAC/GIF report, Washington, DC, weblocation: http://gif.inel.gov/roadmap/pdfs/gen_iv_roadmap.pdf.
USDoE (2003). International Energy Annual 2001. Energy Information Administration report DOE/EIA-0219(2001), US Department of Energy, Washington, DC.
USDoE (2004). Hydrogen posture plan. An integrated research, development and demonstration plan. United States Department of Energy, Washington, DC.
USDoE (2006). Hydrogen Program Record # 5036. At http://www.hydrogen. energy.gov
USDoE (2010a). Hydrogen Program Record # 10004. At http://www.hydrogen. energy.gov
USDoE (2010b). Program Record, Office of Vehicle Technologies & Fuel Cell Technologies # 10001. At http://www.hydrogen. energy.gov
USDoE/T (2006). Hydrogen Posture Plan. US Departments of Energy and Transportation, Washington DC.
Vaillant (2004). Zukunft Brennstoffzellen. http://www.valiant.de.
Vankelecom, I. (2002). Polymer membranes in catalytic reactors. Chem. Rev. 102, 3779-3810.
Venetsanos, A., Huld, T., Adams, P., Bartzis, J. (2003). Source, dispersion and combustion modelling of an accidental release of hydrogen in an urban environment. J. Hazardous Mat. A105, 1-25.
Verhelst, S., Sierens, R. (2003). Simulation of hydrogen combustion in spark-ignition engines. In "La planète hydrogène", Proc. 14th World Hydrogen Conf., Montréal 2002, CDROM published by CogniScience Publ., Montréal.
Verhelst, S., Wallner, T. (2009). Hydrogen-fueled internal combustion engines. Prog. Energy & Combustion Sci. 35, 490-527.
Verstraete, D., Hendrick, P., Pilidis, P., Ramsden, K. (2010). Hydrogen fuel tanks for subsonic transport aircraft. Int. J. Hydrogen Energy 35, 11085-11098.
Veyo, S., Fukuda, S., Shockling, L., Lundberg, W. (2003). SOFC fuel cell systems. In "Handbook of Fuel Cells", Vol. 4 (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 93. Wiley, Chichester.
Veziroglu, A., Macario, R. (2011). Fuel cell vehicles: State of the art with economic and environmental concerns. Int. J. Hydrogen Energy 36, 25-43.
Volbeda, A., Charon, M., Piras, C., Hastchikian, E., Frey, M., Fonticilla-Camps, J. (1995). Crystal structure of the nickel-iron hydrogenase from Desulfovibrio gigas. Nature 373, 580-585.
Vosko, S., Wilk, L., Nusair, M. (1980). Accurate spin-dependent electron liquid correlation energies for local spin density calculations: a critical analysis. Canadian J. Phys. 58, 1200.
VW (2002). Environmental Report 2001/2002: Mobility and sustainability; Schweimer, G., Levin, M. (2001). Life cycle inventory for the Golf A4 (internal report), Volkswagen AG.
VW (2003). Lupo 3 litre TDI, Technical Data, Volkswagen AG, Wolfsburg.
Wagner, U., Geiger, B., Schaefer, H. (1998). Energy life cycle analysis of hydrogen systems. Int. J. Hydrogen Energy 23, 1-6.
Wang, C-Y. (2003). Two-phase flow and transport. In "Handbook of Fuel Cells", Vol. 3 (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 29. Wiley, Chichester.
Wang, J., Wan, W. (2009). Factors influencing fermentative hydrogen production: A review. Int. J. Hydrogen Energy 34, 799-811.
Wang, L., Murta, K., Inaba, M. (2004). Development of novel highly active and sulphur-tolerant catalysts for steam reforming of liquid hydrocarbons to produce hydrogen. Appl. Catalysis A: General 257, 443-47.
Wang, M. (2002). Fuel choices for fuel-cell vehicles: well-to-wheels energy and emission impacts. J. Power Sources 112, 307-321.
Wang, X., Gorte, R. (2002). A study of steam reforming of hydrocarbon fuels on Pd/ceria. Appl. Catalysis A: General 224, 209-218.
Wang, X., Ma, Y., Kashyout, A-H., Zhu, B., Muhammed, M. (2011). Ceria-based nanocomposite with simultaneous proton and oxygen ion conductivity for lowtemperature solid oxide fuel cells. J. Power Sources 196, 2754-2758.
Wang, Y., Balbuena, P. (2004). Roles of proton and electric field in the electroreduction of O² on Pt(111) surfaces: results of an ab-initio molecular dynamics study. J. Chem. Phys. B108, 4376-4384.
Wang. Y., Chen, K., Mishler, J., Cho, S., Adroher, X. (2011). A review of polymer electrolyte membrane fuel cells: Technology, applications, and needs on fundamental research. Applied Energy 88, 981-1007.
Wang, Z., Wang, C., Chen, K. (2001). Two-phase flow and transport in the air cathode of proton exchange membrane fuel cells. J. Power Sources 94, 40-50.
Weber, A., Ivers-Tiffée, E. (2004). Materials and concepts for solid oxide fuel cells (SOFCs) in stationary and mobile applications. J. Power Sources 127, 273-283.
Weht, R., Kohanoff, J., Estrin, D., Chakravarty, C. (1998). An ab initio path integral Monte Carlo simulation method for molecules and clusters: application to Li⁴ and Li⁵ . J. Chem. Phys. 108, 8848-8858.
Weiss, M., Heywood, J., Drake, E., Schafer, A., AuYeung, F. (2000). On the road 2020. Report MIT EL 00-003, Laboratory for Energy and Environment, Massachussetts Institute of Technology, Cambridge, MA.
Weiss, M., Heywood, J., Schafer, A., Natarajan, V. (2003). Comparative assessment of fuel cell cars. Report LFEE 2003-001 RP, Massachussetts Institute of Technology, Cambridge, MA.
Weizsäcker, E., Hargroves, K., Smith, M., Desha, C., Stasinopoulos, P. (2009). Factor Five. Earthscan Publ., London.
Wenk, S., Qian, D., Wakayame, T., Nakamura, C., Zorin, N., Rögner, M., Miyake, J. (2002). Biomolecular device for photoinduced hydrogen production. Int. J. Hydrogen Energy 27, 1489-1493.
Weydahl, T., Gruber, A., Gran, I., Ertesvåg, I. (2003). Mathematical modelling and numerical simulations of different diffusion effects in hydrogen-rich turbulent combustion. In "La planète hydrogène", Proc. 14th World Hydrogen Conf., Montréal 2002, CDROM published by CogniScience Publ., Montréal, 9 pp.
Wieland, S., Melin, T., Lamm, A. (2002). Membrane reactors for hydrogen production. Chem. Eng. Sci. 57, 1571-1576.
Wilhelm, J., Janßen, H., Mergel, J. Stolten, D. (2011). Energy storage characterization for a direct methanol fuel cell hybrid system. J. Power Sources 196, 5299-5308.
Wilkinson, D., Vanderleeden, O. (2003). Serpentine flow design. In "Handbook of Fuel Cells", Vol. 3 (Vielstich, W., Lamm, A., Gasteiger, H., eds.), Ch. 27. Wiley,
Chichester.
Williams, K., Eklund, P. (2000). Monte Carlo simulations of H² psysisorption in finite-diameter carbon nanotube ropes. Chem. Phys. Lett. 320, 352-358.
Wise, D. (1981). Solar Energy 27, 159-178.
Wolfbauer, G. (1999). The electrochemistry of dye sensitized solar cells, their sensitizers and their redox shuttles. Ph. D. Thesis, Monash University, Melbourne.
Woo, Y., Oh, S., Kang, Y., Jung, B. (2003). Synthesis and characterization of sulphonated polyimide membranes for direct methanol fuel cell. J. Membrane Science 220, 31.45.
Woodward, J., Orr, M., Cordray, K., Greenbaum, E. (2000). Enzymatic production of biohydrogen. Nature 405, 1014-1015.
World Energy Council (1995). "Survey of Energy Resources" 17th ed. World Energy Conference, London.
Wurster, R. (1997a). PEM fuel cells in stationary and mobile applications. Paper for Biel Conference (http://www.hyweb.de/knowledge).
Wurster, R. (1997b). Wasserstoff-Forschungs- und Demonstrations-Projekte. Kryotechnik 26. Feb. 1997, VDI-Tagung (http://www.hyweb.de/knowledge).
Wurster, R. (1998). Paper presented at Deutsche Kälte-Klima-Tagung, Würzburg.
Wurster, R. (2003). GM well-to-wheel-Studie Ergebnisse and Schlüsse. LB Systemtechnik website http://www.HyWeb.de.
Wurster, R. (2004). Daily use of hydrogen in road vehicles and their refueling infrastructure: safety, codes and regulation. In "Proc. HYFORUM: Clean Energies for the 21st Century", Beijing. Available at http://www.lbst.de.
WWEA (2011). World Wind Energy Report 2010. World Wind Energy Association, Bonn, http://www.wwindea.org
Yagishita, T., Sawayama, S., Tsukahara, K., Ogi, T. (1996). Photosynthetic bio-fuel cells using cyanobacteria, In "Renewable Energy, Energy Efficiency and the Environment: World Renewable Energy Congress", vol. II, pp. 958-961. Pergamon, Elmsford, NJ.
Yamada, Y., Matsuki, N., Ohmori, T., Mametsuka, H., Kondo, M., Matsuda, A., Suzuki, E. (2003). One chip photovoltaic water electrolysis device. Int. J. Hydrogen Energy 28, 1167-1169.
Yamaguchi, M., Horiguchi, M., Nakanori, T., Shinohara, T., Nagayama, K., Yasuda, J. (2001). Development of large-scale water electrolyzer using solid polymer electrolyte in WE-NET, In "Hydrogen Energy Progress XIII", Vol. 1 ("Proc. 13th World Hydrogen Energy Conf., Beijing 2000"; Mao and Veziroglu, eds.), pp. 274-281. Int. Assoc. Hydrogen Energy & China Int. Conf. Center for Science and Technology, Beijing.
Yang, C., Srinivasan, S., Bocarsly, A., Tulyani, S., Benziger, J. (2004). A comparison of physical properties and fuel cell performance of Nafion and zirconium phosphate/Nafion composite membranes. J. Membrane Sci. 237, 145-161.
Yang, W., Zhao, T., Wu, Q. (2011). Modeling of a passive DMFC operating with neat methanol. Int. J. Hydrogen Energy, doi:10.1016/j.ijhydene.2011.02.117.
Yasuda, I., Shirasaki, Y., Tsuneki, T., Asakura, T., Kataoka, A., Shinkai, H., Yamaguchi, R. (2004). Development of membrane reformer for high-efficient hydrogen production from natural gas. In "15th World Hydrogen Energy Conference, Yokohama 2004". Hydrogen Energy Systems Soc. of Japan (CDROM).
Yinxing, S. (1637). High skills in materials production (Tian gong kai wu). China.
Yokata, O., Oku, Y., Sano, T., Hasegawa, N., Matsunami, J., Tsuji, M., Tamura, Y. (2000). Stoichiometric consideration of steam reforming of methane on Ni/Al2O3 catalyst at 650°C by using a solar furnace simulator. Int. J. Hydrogen Energy 25, 81-86.
Yoong, L., Chong, F., Dutta, B. (2009). Development of copper-doped TiO2photocatalyst for hydrogen production under visible light. Energy 34, 1652-1661.
Yu, L., Liu, H. (2002). A two-phase flow and transport model for the cathode of PEM fuel cells. Int. J. Heat Mass Transfer 45, 2277-2287.
Yu, L., Wuye, D., Xianchen, C., Bin, M. (2001). Aerospike engine and single-stage-toorbit return transportation. In "Hydrogen Energy Progress XIII, Proc. 13th World Energy Conf., Beijing 2000" (Mao, Z., Veziroglu, T., eds.), pp. 654-663. Int. Assoc. Hydrogen Energy, Beijing.
Yu, W., Zinger, D., Bose, A. (2011). An innovative optimal power allocation strategy for fuel cell, battery and supercapacitor hybrid electric vehicle. J. Power Sources 196, 2351-2359.
Zackrisson, M., Avellán, L., Orlenius, J. (2010). Life cycle assessment of lithium-ion batteries for plug-in hybrid electric vehicles - Critical issues. J. Cleaner Production 18, 1519-1529.
Zaluska, A., Zaluski, L., Ström-Olsen, J. (2001). Structure, catalysis and atomic reactions on the nano-scale: a systematic approach to metal hydrodes for hydrogen storage. Appl. Phys. A72, 157-165.
Zapp, P. (1996). Environmental analysis of solid oxide fuel cells. J. Power Sources 61, 259-262.
Zaza, F., Paoletti, C., LoPresti, R., Simonetti, E., Pasquali, M. (2011). Multiple regression analysis of hydrogen sulphide poisoning in molten carbonate fuel cells used for waste-to-energy conversions. Int. J.Hydrogen Energy, doi:10.1016/ j.ijhydene.2011.01.174.
Zehr, J., Waterbury, J., Turner, P., Montoya, P., Omoregle, E., Steward, G., Hansen, A., Karl, D. (2001). Unicellular cyanobacteria fix N² in the subtropical North Pacific Ocean. Nature 412, 635-638.
Zenith, F., Krewer, U. (2010). Modelling, dynamics and control of a portable DMFC system. J. Process Control 20, 630-642.
Zenith, F., Weinzierl, C., Krewer, U. (2010). Model-based analysis of the feasibility envelope for autonomous operation of a portable direct methanol fuel-cell system. Chem. Engineering Sci. 65, 4411-4419.
ZFK (1999). Hydrogen research at Forschungszentrum Karlsruhe. In "Proc. EIHP Workshop on Dissemination of Goals, Preliminary Results and Validation of Methodology", European Commission DG XII, acsessed 2003 at the website http://www.eihp.org/eihp1/workshop/intro.html.
Zhang, H., Lin. G., Chen, J. (2011). Performance analysis and multi-objective optimisation of a new molten carbonate fuel cell system. Int. J. Hydrogen Energy 36, 4015-4021.
Zhang, Q., He, D., Li, J., Xu, B., Liang, Y., Zhu, Q. (2002). Comparatively high yield methanol production from gas phase partial oxidation of methane. Appl. Catalysis A: General 224, 201-207.
Zhang, T., Chu, W., Gao, K., Qiao, L. (2003). Study of correlation between hydrogeninduced stress and hydrogen embrittlement. Materials Sci. Eng. A147, 291-299.
Zhang, T., Liu, H., Fang, H. (2003). Biohydrogen production from starch in wastewa-
ter under thermophilic condision. J. Environm. Managem. 69, 149-156.
Zhang, Y., Suenaga, K., Colliex, C., Iijima, S. (1998). Coaxial nanocables: silicon carbide and silicon oxide sheathed with boron nitride and carbon. Science 281, 973−975.
Zhang, Y., Xia, C. (2010). A durability model for solid oxide fuel cell electrodes in thermal cycle processes. J. Power Sources 195, 6611-6618.
Zhao, X., Ma, L. (2009). Recent progress in hydrogen storage alloys for nickel/metal hydride secondary batteries. Int. J. Hydrogen Energy 34, 4788-4796.
Zhou, L., Zhou, Y., Sun, Y. (2004). A comparative study of hydrogen adsorption on superactivated carbon versus carbon nanotubes. Int. J. Hydrogen Energy 29, 475- 479.
Zhu, Q., Li, J., Wei, J. (2001). Production and utilization of hydrogen in China. In "Hydrogen Energy Progress XIII, Proc. 13th World Energy Conf., Beijing 2000" (Mao, Z., Veziroglu, T., eds.), pp. 105-109. Int. Assoc. Hydrogen Energy, Beijing.
Zhu, Y., Ha, S., Masel, R. (2004). High power density direct formic acid fuel cells. J. Power Sources 130, 8-14.
Zittel, W., Wurster, R. (1996). "Hydrogen in the Energy Sector". Ludwig− Bölkow−ST Report: http://www.hyweb.de/knowledge/w−i−energiew−eng.
Zouni, A., Witt, H., Kern, J., Fromme, P., Krauss, N., Saenger, W., Orth, P. (2001). Crystal structure of photosystem II from Synechoccoccus elongatus at 3.8Å resolution. Nature 409, 739-743.
Züttel, A. (2004). Hydrogen storage methods. Naturwissenschaften 91, 157-172.
Züttel, A., Rentsch, S., Fischer, P., Wenger, P., Sudan, P., Mauron, P., Emmenegger, C. (2003). J. Alloys Compounds 356, 515.
Züttel, A., Wenger, P., Sudan, P., Mauron, P., Orimo, S. (2004). Hydrogen density in nanostructured carbon, matals and complex materials. Materials Sci. Eng. B108, 9-18.

INDEX

pathways, 26

A	Bio-production of hydrogen
	life-cycle impacts, 377
Absorptance spectrum	Bipolar plates
green plants, bacteria, 28	corrosion issues, 162
Acid polymer fuel cell, 159	for PEMFC, 165
Activated carbon storage, 89	Blue-green algae
Activity coefficient, 97	also called cyanobacteria, 31
ADP, 30	Boil-off of liquid hydrogen, 71
AFC, 160	Born-Oppenheimer approximation,
cost, 162	111
efficiency, 162	Boudouard reaction, 9
lifetime, 162	Brick-making, 64
Airplanes	Building-integrated fuel cell systems,
accident rates, 253	236, 237
liquid hydrogen engines, 63	Building-integrated hydrogen con
Algae, 34	cept, 251
Alkaline electrolysis plants	Buses
efficiency, 54	hydrogen combustion engine, 63
Alkaline fuel cells, 160	and lorries with fuel cells, 225
Ammonia production, 65	Butler-Volmer equations, 105
Ammonia synthesis, 31
Anode, 100

Antisymmetry, 113	C
Aquifer storage of hydrogen, 69, 247	Car accident fatalities, 258
Atomic force microscopy, 27	Carotenes, beta, 30
ATP, 30	Catalysis, 6, 7, 122
Autothermal hydrogen production,	Catalyst
57	for fuel cells, 100
Autothermal oxidation, 10	for gasification, 22
Available free energy, 98	for methanol steam reforming, 56
Avogadro's constant, 96	for PEMFC, 181
	Catalytic water splitting, 50
B	Cathode, 100
	Cavern hydrogen storage, 68
B3LYP, 17, 18	Cellulose
B3LYP method, 120	fermentation, 37
Basis sets of wavefunctions, 116	thermal conversion, 23
Battery modelling, 208	Centralised hydrogen production, 51
Biofuel cells, 197	Charge flow, 98
Biogas, 55	Chemical potential, 72, 97, 103
Bio-hydrogen, 31	Chemical thermodynamics, 71
Biological hydrogen production, 26	Chemisorption, 8
Biomass methanol production from, 55	Clean fossil scenario, 270 appraisal of feasibility, 292

Diffusion, 137

Climatic impacts from hydrogen	particle-collision type, 138
emissions, 399	Diffusion constant, 137
CO contamination, 56	Diffusion equation, 137
CO2 emissions, 271	Diffusion modelling, 135
CO2 removal, 55, 273	Diffusivity of fuels in air, 60
Coal reserves and additional re	Dipole model, 20
sources	Direct combustion hydrogen vehicles,
global distribution, 276	203
Complex hydride storage, 80	Direct methanol fuel cell, 191
Compressed gas hydrogen storage, 65	impedance measurements, 107
Compressed gas storage, 69	Dissipation of energy, 97
Compression of hydrogen, 66	Dissociation
Conductivity of SOFC electrolyte ma	hydrogen, 17
terials, 145	oxygen, 18
Control logic simulation, 208	diatomic molecules, 122
Corrosion of electrodes, 158	Distributed hydrogen production, 52
Cost expectations, 361	DMFC
Coulomb interaction, 111	passive operation, 194
Cyanobacteria, 29, 31, 36	power density, 193
Cytochrome b6f, 29	Double-layer, 15
	Drive-train modelling, 208
D	Driven cell, 99
	Driving cycles for passenger cars, 210
Damage assessment in life-cycle stud	Dynamical behaviour
ies, 374	thermodynamic systems, 98
Darcy's law, 139	Driving cycle
Darcy unit, 70	new European, 385
Decentralised hydrogen storage, 247	Dry reforming, 10
Decentralised hydrogen system, 252	fuel cell, 104
Decentralised use of fuel cells, 236	ideal, 104
Degradation of AFC, 162	photoproduction of hydrogen, 34
Delphi method	solar hydrogen, 47
for estimating technology learning
progress, 367	E
Demand modelling, 260
Denmark	Einstein unit, 34
hydrogen infrastructure scenario,	Electric circuit models, 107
250	Electric potential difference, 96
Density functional theory, 17, 18, 119	Electric vehicle simulation, 218
Density of hydrogen and other fuels,	Electrochemical devices, 15, 96
66	defined, 99
Density matrix, 119	Electrochemistry, 95
Density of electron wavefunction, 112	Electrode
De-sulphurisation of natural gas for	negative, 12
SOFC, 150	positive, 12
Diesel cars, particulate emissions, 393	potentials, 106
Diesel engine, 25	Electrolysis
Difference equation approach, 136	alkaline, 12

by reverse fuel cell operation, 11,	Faraday's constant, 96
100	Fermentation, 36
Electrolyte	of biomass
alkaline, 12	life-cycle impacts, 380
chemical characterisation, 97	industrial plant, 42
fluid and solid, 100	Fermions, 113
model, 15	Fe-S centre, 33
Electromotive force, 103	Fick's law, 137
Electron charge, 96	Filling stations
Emergency power systems, 234	individually building-integrated,
End-use energy, 262, 269	236
Energy conservation, 95	Finite elements method, 137
Energy delivered to final consumer,	First law of thermodynamics, 95
271	Flame temperature, fuel burning, 60
Energy demand	Flammability of fuels, 60
categories, 268	Flavodoxin, 30
for Denmark 2050, 324	Flow modelling, 136
scenario, 262	PEMFC, 168
scenario for Denmark, 328	Fluid dynamics, 137
Energy density of hydrogen and other	modelling
fuels, 66	applied to hydrogen accident,
Energy efficiency	254
fuel cell systems, 203	Fluidised bed gasification, 24
Energy quality, 268	Fluid mechanics, 135
Enthalpy, 6, 98	Fossil energy resources, 270
Entropy, 95, 97	Fossil/hydrogen scenario for global
Environmental impact, 372	supply by 2050, 281
gasification, 25	Free energy, 95
Environmental impact analysis	available, 98
fuel cell and internal combustion	Frequency response, 107
engine cars, 386	Fuel cells
Environmental sustainability, 261	basic layout, 100
Equation of continuity, 136	in cars
Equilibrium constant, 72	life-cycle analysis, 385
Equilibrium pressure in metal	costs, 362
hydrogen stores, 72	defined, 100
Equivalence ratio, 23	efficiency, 104
Eulerian transport equation, 135	electrochemistry and thermody
Excited states, 120	namics, 95
Exergy, 99	fuel and power production modes,
Explosivity of fuels, 61	100
Externality assessment, passenger	power production, 100
cars, 391	quantum chemical calculations,
External versus internal storage of	130
chemical energy, 240	reverse operation, 11, 100
	reversible used in scenarios, 334
F	in ships
	environmental impacts, 399
Fair market considerations, 322

Heat of combustion, 23

vehicle modelling, 206 Fuel cell systems combined heat, power and fuel delivery, 237 control strategies, 203 cost, 369 stationary power production, 233 G Gas constant, 97, 138 Gas delivery system, PEMFC, 165 Gas diffusion layer modelling, 135 Gas diffusion layer, PEMFC, 166, 169 Gas flow channel design, PEMFC, 166 Gasification for hydrogen production biomass, wood, 21 coal, peat, 22 natural gas and oil, 21 Gasifier types updraft, downdraft, fluidised bed, 25 Gas law, modified, 66 Gaussian wavefunctions, 116 GDL, 166 Generalised forces, 99 Generalised gradient approximation, 120 Geographical information system (GIS), 261 Danish scenario model, 325 Geological hydrogen storage, 65, 246 Gibb's free energy, 71, 95 GIS; see Geographical information system Glucose fermentation, 38 Goal satisfaction, 268 Greenhouse warming damage, 292 Green plants, 29 H Hamiltonian, 111 Hartree-Fock approximation, 113 Hazard modelling, 260 Health impacts of methanol, 259 Heat flow, 98 Heat of fusion, 23 Heat pump use in scenarios, 312 use in renewable energy scenarios, 348 Helmholtz potential, 98 High temperature fuel cells, 139 Hindenburg accident, 253 Home fuel cell units with reformer, 237 Human welfare, 261 Hybrid fuel cell bus simulation, 226 Hybrid fuel cell cars, 202 Hydride storage, 71 Hydrocarbons use in SOFC, 150 Hydrogen accident, 254 bioproduction yield, 34 centralised production, 51 from coal scenario, 273 combustion, 60 conversion, 59 dissociation, 17 distributed/decentralised production, 52 as energy carrier, 59, 246 as energy storage medium, 60 leak detection, 260 liquid in vehicles, 63 photobiological production, 40 production by fermentation, 42 properties, 6 safety related properties, 61 storage in underground caverns, 246 Hydrogen and renewable energy scenarios centralised and decentralised structures, 332 Hydrogenase uptake, reversible, bi-directional, 33 Hydrogen combustion simulation, 62 Hydrogen diffusion during accident, 257 Hydrogen filling stations, 250 Hydrogen gas explosions, 256 Hydrogen molecule splitting, 100

Hydrogen pipeline cost, 93 Hydrogen production, 5 costs, 361 in Danish scenarios, 348 life-cycle analysis, 375 by steam reforming of natural gas life-cycle impacts, 376 Hydrogen storage	part-load behaviour relative to PEMFC, 165 Internal energy, 95 Inventory analysis in life-cycle stud ies, 374 Ions in electrolyte, 97 Irreversible processes, 99 ISO, 259 IV-curve, DMFC, 192
at buildings, 240 on carbon surfaces, 89
chemical reactions, 92	K
compressed gas, 66	Kinematic viscosity, 136
costs, 368	Kinetic energy, 111
cryoadsorbed gases, 89
density and energy density, 66	L
energy density overview, 91
hydride properties in comparison,	Laminar flow, 138
78	LaNi5H7, 76 Laptop computer with fuel cell, 240
infrastructure, 245	Learning curves
liquefied, 70	influence on fuel cell cost, 363
materials and energy use, 398	Lennard-Jones potential, 8
maximum capacity of hydrides, 80	Life-cycle analysis, 372
metal hydrides, 73	of fuel cells, 381
salt domes, aquifers, rocks, 68	Life-cycle assessment, 374
Hydrogen transmission, 92	hydrogen storage and infrastruc
infrastructure, 248	ture, 398
Hydrogen transportation in containers, 92	Lifetime of fuel cells
transportation cost	influence on cost, 363
in ship containers, 92	Li ion battery, 240—241
pipelines, 93	Liquid hydrogen storage, 70
Hydrogen use in fossil scenario, 284	Local spin density, 120
Hydrophobic quality, 161	Lower heat value, 7
	Low temperature fuel cells, 140
I	Lumen, 27
	Lumen unit, 34
Identical particles, 113	Lurgi gasifier, 22
Ignition energy, fuels, 60
Impact analysis in life-cycle studies,	M
374	Mass distribution in passenger cars
Impedance spectroscopy, 107	fuel cell-battery hybrids, 210
Industrial use of hydrogen, 65

Mass flow, 98 Mass production

MCFC, 140

influence on PEMFC cost, 363 Material and energy use

hydrogen storage devices, 398

Industrial use of hydrogen, 65 Infrastructure requirements for hy-

Intermittent power handling, 239 Internal combustion engines

drogen system, 245 Infrastructure costs, 368

life-cycle impacts, 381 MEA, 183	model of hydrocarbon in catalyst, 58
Medium-temperature membrane fuel	Molecular dynamics model,
cells, 155	membrane, 176
Membrane-electrode assembly, 183	Molecular mechanics, 15
Membrane reactor, 9—10	Molecular orbitals, 115
Membrane steam reformer, 57	Molecular structure, 111
Metal hydrides batteries, 241	Mole fraction, 97
hydrogen absorption	Møller-Plesset perturbation theory, 119
density functional theory, 85	Molten carbonate fuel cells, 140
periodic boundary condition	Monetised life-cycle impacts for
calculation, 87	passenger cars, 393
quantum chemical modelling,
80—81	N
storage, 74
absorption/desorption times, 80	Na3AlH6, 82
volume change by hydrogen ab	NADP, 30
sorption, 77	Nafion
Methane, 5	membrane, 175
cracking, 8	Nanotube hydrogen storage, 89
steam reforming, 6	National renewable energy scenario,
Methanol	323
cost, 56	Natural gas reserves and additional
crossover in DMFC, 193	resources
as fuel for fuel cells, 53	global distribution, 277
health impacts, 259	Navier-Stokes equation, 136
hydrogen production by steam re	Negative electrode, 100
forming, 56	Nernst potential, 105, 147
production, 54	Niche markets for fuel cells, 241
production from biomass, 54	NiH, 75
reforming in SOFC cells, 59	Nitrogenase, 31
vehicle hydrogen production by	Nitrogen fixation, 32
reforming, 52—53	Normative models of the future, 268
Methodology of life-cycle analysis,	Norms and standards, 259
373	Nuclear accelerator-breeding, 299
MgH2, 74	Nuclear accidents, 295
potential energy surface, 87	discussion, 253
quantum chemical model of	Nuclear energy amplifier, 299
absorption, 85	Nuclear energy usage, 295
Miller indices, 17	Nuclear hydrogen, 297
Mismatch between power production	chemical cycles, 298
and demand Danish scenarios, 342	in global scenario, 309
Modified gas law, 66	Nuclear power history, 294—295
Mo-Fe centre, 31	Nuclear reactors
Molar fraction, 137	inherently safe design, 298—299
Molecular dynamics, 15	novel high-temperature types, 297
	Nuclear resources assessment, 304
	Nuclear terrorism, 295

Nuclear waste, 295	stack cost, 363
	stack materials use, 384
O	temperature modelling, 186
	vehicle power system, 203
Ocean disposal of CO2, 273	water management, 170
Oil reserves and additional resources	water modelling, 179
finding rate and usage, 429	Perfluorosulphonic acid membranes,
global distribution, 275	175
On-board fuel reforming, 52—53	Performance simulation
ONIOM calculations, 124	passenger car, 207—208
Onsager model, 20	Permeability, 69, 139
Otto engine, 25	Perovskite
using hydrogen, 60	as proton conductor, 51
Overvoltage, 13, 19	Perovskite-type structure, 145
Oxidation	Perturbation theory, 119
autothermal, 10	Phosphoric acid cells, 158
partial, 10, 11	Photobiological hydrogen produc
Oxygen dissociation, 18	tion, 40
at catalyst surface, 185	Photodissociation, 43
	Photoelectrochemical electrolyser
P	with amorphous solar cell, 47
	membrane type, 47
PAFC, 158	Photoelectrochemistry, 145
Partial oxidation, 10, 11	Photosynthesis, 27
Partial pressure, 137	Photosynthetic system, 29
Part-load efficiency of PEMFC, 164	Photosystem I, 29, 33
Passenger cars, 201	Photosystem II, 29, 30
with fuel cells or internal combus	Photovoltaic power and storage, 246
tion engines	Planck's constant, 30, 111
mass distribution comparison,	Plastocyanin, 29
385	Plastoquinone, 29
life-cycle analysis, 385	Plateau pressures of metal hydrides,
PEMFC use, 203	79
simulations, 212	Platinum, environmental impact, 383
PBI membrane, 159, 181	Platinum catalyst, 196
Peak-shaving fuel cell power, 234	Polybenzimidazole membranes, 160
PEEKs, 181	Polytetrafluoroethylene pore layers,
PEMFC; see PEM fuel cell	161
PEM fuel cell	Population distributions, 266
cars, 204—205	Pore structure, influence on diffusion,
catalysts, 181	138
current density calculation, 180	Portable fuel cell uses, 241
degradation and lifetime, 190	Positive electrode, 100
efficiency, 186	reactions, 123
gas diffusion layer, 169	Potential energy, 111
general layout, 163	calculations
life-cycle impacts, 382—383	positive fuel cell electrode area,
molecular modelling, 184 PEM cell, 162	134

surface, 16 surface calculations, 124 Potential loss, 105 Power density SOFC, 154 Power density improvements influence on PEMFC cost, 363 Power plants using fuel cells, 233 Power-quality, 234 Producer gas, 54 Proton exchange membrane impedance measurements, 107 Proton exchange membrane fuel cell, 163 electrolysis, 20 Proton transfer in water, 110 PTFE gas diffusion layer, 161 Purple bacteria, 36 Pyrolysis, 10, 23

Q

Quantum chemical calculation, 17 Quantum chemistry, 111 Quantum mechanical calculation difference from classical, 102 Quantum mechanical modelling, 106 Quantum mechanics, 17, 112

R

Recombination, 27 Redox couple, 49 Reference scenario, 261 Reformers, 6, 53 miniature, 58 Regenerative battery, 99 Regenerative fuel cell, 99 Relaxation, 108 Remote power supply, 234 Renewable energy use in Danish scenarios, 334 Renewable energy scenarios, 332 appraisal of Danish scenarios, 364 for Denmark, 324 hydrogen produced decentrally, 361 use of reversible fuel cells, 332 Renewable sources

estimate for Danish scenario, 332 Reserves of fossil fuels, 274, 429 Resources of fossil fuels, 274 Response function, 107 Reverse fuel cell operation, 11 Reversible fuel cell, 99 Reversible PEM fuel cell performance, efficiency, 187 Reversible PEMFC in buildings, 239 Reversible reactions, storage in, 90 Risk definitions, 252 Roadmaps for hydrogen implementation, 260 Rock storage of hydrogen, 67

S

Safe nuclear scenario, 307 Safe nuclear technologies, 297 Safety culture, 260 Safety issues, 252 Safety of compressed hydrogen, 67 Safety-related properties of hydrogen, 61 Salt intrusion hydrogen storage, 246 Salt storage of hydrogen, 68 Scenario techniques, 260 Schrodinger equation, 111 Second law efficiency, 99 Secondary battery, 99 Semiconductor p-n junction for photodissociation, 43 Sequestration of carbon dioxide, 272 Shift reaction, 6 Ships, trains and airplanes using hydrogen, 228 Skateboard vehicle, 206 Social impacts, 372 Social impacts from passenger cars, 392 SOFC, 143 autoreforming, 150 cell-potential loss, 153 efficiency, 152 lifetime, 154 modelling, 146 system layout, 151 voltage and power density, 154

Solar cell, 43

Solar hydrogen-producing cell, 44	U
Solar radiation spectrum, 28
Solid oxide fuel cells, 143	Uncertainties
Solvent, 19	fuel cell technology and cost, 400
Space vehicles	Uncertainty avoided in hydrogen
use of liquid hydrogen, 63	scenario, 292
Spacecraft fuel cells, 161	Underground compressed air storage, 67
Speed of sound, 63	Underground storage of hydrogen,
Spin functions, 114	246
Standard reversible potential, 103	Upper heat value, 7
Standards, 255 Stationary states, 112	Uranium reserves and resources, 304
Steady state condition, 138
Steam reformer	V
membrane type, 57
Steam reforming, 6	Van't Hoff's equation, 73
gasoline, 57	Variational theory, 114
methanol to hydrogen, 56	Vehicle reformers, 52
natural gas to methanol, 54	Vehicle simulations, 211
Stefan-Maxwell equation, 137	Velocity fields
Stratospheric airship, 232	in PEMFC, 167
Stress tensor, 136	in SOFC, 148
Stroma, 27	Voltammogram, 49
Substrates for biohydrogen produc

	W
tion, 26
Sulphur bacteria, 36	Waste
SV basis, 117	as basis for hydrogen production,
Synthesis gas, 6	42
	electrolysis, 11
T	management, 138
Tafel slope, 106	management in PEMFC, 168, 172,
Temperature field in SOFC, 148	401
Thermal water splitting, 51	splitting, 27
Thermodynamical description of elec	transport in fuel cells, 139
trolysis, 103	Wavefunction, 112
Thermodynamics of fuel cells, 95	of electrons
Thorium reserves and resources, 306	metal hydride absorption, 89
Thorium-cycle nuclear reactors, 304	Wind power
Thylakoid membranes, 27	Danish resources on-and off-shore,
Time-dependent density functional	333
theory, 121	Wind power and storage, 246
Time-dependent Hartree-Fock theory,	Wind power production and use
121	Danish 2050 scenario, 335
Tolerance for poisoning substances,	Woody biomass, 21
155	Work, 96
Transport in PEMFC, 167 Tunnel accidents, 258	Work, maximum, 97 World energy use, 269

X Y

X-ray diffraction, 28 Yttrium-stabilised zirconium oxide, 145

wursterbentsorensenhydrogenfuel2007.md 1.1 MB История Исходник