J. Phys. Chem. B 2005, 109, 2149-2154
2149
Catalysis in High-Temperature Fuel Cells† K. Fo1 ger* and K. Ahmed Ceramic Fuel Cells Ltd., 170 Browns Road, Noble Park, Victoria 3174, Australia ReceiVed: March 2, 2004; In Final Form: May 24, 2004
Catalysis plays a critical role in solid oxide fuel cell systems. The electrochemical reactions within the cells oxygen dissociation on the cathode and electrochemical fuel combustion on the anodesare catalytic reactions. The fuels used in high-temperature fuel cells, for example, natural gas, propane, or liquid hydrocarbons, need to be preprocessed to a form suitable for conversion on the anodessulfur removal and pre-reforming. The unconverted fuel (economic fuel utilization around 85%) is commonly combusted using a catalytic burner. Ceramic Fuel Cells Ltd. has developed anodes that in addition to having electrochemical activity also are reactive for internal steam reforming of methane. This can simplify fuel preprocessing, but its main advantage is thermal management of the fuel cell stack by endothermic heat removal. Using this approach, the objective of fuel preprocessing is to produce a methane-rich fuel stream but with all higher hydrocarbons removed. Sulfur removal can be achieved by absorption or hydro-desulfurization (HDS). Depending on the system configuration, hydrogen is also required for start-up and shutdown. Reactor operating parameters are strongly tied to fuel cell operational regimes, thus often limiting optimization of the catalytic reactors. In this paper we discuss operation of an authothermal reforming reactor for hydrogen generation for HDS and start-up/ shutdown, and development of a pre-reformer for converting propane to a methane-rich fuel stream.
1. Introduction A fuel cell is a device that converts the chemical energy in a fuel to electrical energy via electrochemical reactions. These electrochemical reactions are the oxidation of the reductant (ideally hydrogen) and reduction of the oxidant (ideally oxygen). The electrochemical reactions, reduction and oxidation, are catalyzed by the cathode and the anode, respectively. However, catalysis in fuel cell systems is not restricted only to these electrochemical reactions. In real systems there are chemical reactions and associated catalysis relating to processing the fuel to a form suitable for the fuel cellsremoval of contaminants which damage the electrocatalytic activity of the electrode and reforming/converting the fuel to hydrogen or some other form that can be oxidized electrochemically at the anode. Furthermore, there is catalysis associated with oxidation of unutilized fuel. The catalytic processes in a solid oxide fuel cell (SOFC) system are highlighted in the system process flow diagram shown in Figure 1. Fuel processing is a two-step process: (1) In the first step, referred to as “cleanup”, contaminants are removed from the fuel. The most common contaminant is sulfur, which poisons the hydrocarbon reforming catalyst and the fuel cell anode. Both of these catalysts require the sulfur content in the fuel to be reduced to sub-parts-per-million levels. If the fuel contains other poisons, then the fuel-processing step must also include processes to remove these contaminants. (2) The second step, known as “pre-reforming”, converts the fuel to a mixture of methane, hydrogen, carbon monoxide, and carbon dioxide for internally reforming fuel cells. For fuel cells operating on hydrogen, the product gas from the second step consists of hydrogen, carbon monoxide, and carbon dioxide. The exact composition of the product gas in both cases depends on the reforming route and process parameters, i.e., steam reforming, †
Part of the special issue “Michel Boudart Festschrift”. * Towhomcorrespondenceshouldbeaddressed.E-mail:
[email protected].
partial oxidation, and autothermal reforming with their temperature, steam-to-carbon ratio, and air-to-fuel ratio. Since hightemperature fuel cells can operate on carbon monoxide, a CO cleanup step is not required for these fuel cells. The least component sensitive “sulfur cleanup” process is hydro-desulfurization (HDS) coupled with an absorber bed for hydrogen sulfide. In the first step organic sulfides such as thiophenes and mercaptans are hydrogenated over a Co-Mo or Ni-Mo catalyst. In the second step hydrogen sulfide and carbonyl sulfide usually present in the fuel and generated from the first step are adsorbed in a fixed bed of high-surface-area zinc oxide. Ceramic Fuel Cells Ltd. (CFCL) has developed SOFC anodes that are active for steam reforming of methanesinternally reforming cellssand thus are able to operate on methane-rich gases without degradation from carbon formation. Indeed, the endothermic reforming reaction assists in heat removal from the fuel cell stack. However, the higher hydrocarbons in the fuel, e.g., natural gas, need to be converted to methane, hydrogen, and carbon monoxide as they lead to carbon deposition in the fuel passages and preheat heat exchangers. This is carried out by pre-reforming via steam reforming, partial oxidation, or autothermal reforming. In steam reforming (SR) the hydrocarbon fuel is contacted with steam over a nickel catalyst. The overall steam reforming process consumes heat and can therefore be used for thermal management of the fuel cell stack.1 The product spectrum varies depending on the process conditions. When the process is carried out at relatively low temperatures, the product gas contains hydrogen, carbon dioxide, carbon monoxide, and methane. Such a gas mixture is suitable for internally reforming fuel cells. When the process is carried out at relatively high temperatures, the product gas contains little or no methane and is suitable for noninternally reforming fuel cells, i.e., fuel cells that operate on hydrogen only.
10.1021/jp0490507 CCC: $30.25 © 2005 American Chemical Society Published on Web 06/29/2004
2150 J. Phys. Chem. B, Vol. 109, No. 6, 2005
Fo¨ger and Ahmed
Figure 1. Process flow diagram of a solid oxide fuel cell system.
In catalytic partial oxidation (POX) the hydrocarbon reacts with a substoichiometric amount of oxygen over a catalyst to produce a mixture of hydrogen and carbon monoxide. In practice air is used as the source of oxygen so that the product gas is diluted with residual nitrogen, resulting in reduced cell voltage and efficiency. Some carbon dioxide is formed as some of the hydrocarbon goes to complete combustion. The reaction is highly exothermic and attractive for fast heat-up applications. Autothermal reforming (ATR) combines endothermic steam reforming and exothermic partial oxidation reactions. The heat produced by the partial oxidation reaction is used in the steam reforming reaction to generate hydrogen and carbon monoxide. It is desirable to balance the reactions in such a way that they are essentially thermoneutral, requiring no net input of heat. The catalytic reactions on an internally reforming SOFC cell are the steam reforming reaction and water gas shift reactions on the anode surface and the electrochemical hydrogen and CO combustion reaction at the anode-electrolyte interface. Practical fuel conversions in the fuel cell stack are restricted to 10 to minimize axial dispersion and DT/dp > 10 to eliminate wall effects. Checks were also made for internal diffusion by carrying out experiments with two different catalyst particle diameters. These data were required to design a bench-scale reactor as part of a fuel processor for a kilowatt-sized internally reforming SOFC. The Mears4 criterion for avoiding a significant dispersion effect is
L/dp > (20n/Pe) ln(1/[1 - XA]) The criterion for the absence of a significant concentration difference between the bulk fluid and the outside surface of the catalyst is
(-r)rp/(Cbkc) < 0.15/n Furthermore, Cb - Cs was calculated to check that the deviation was not more than 5%. Similarly, Tb - Ts was calculated to check that temperature differences were less than 5 °C. 3. Results and Discussion 3.1. Autothermal Reforming. Autothermal reforming of methane, the principal component of natural gas, is a combination of steam reforming of methane
CH4 + H2O a CO + 3H2, ∆H298 ) 206.2 kJ/mol (1) and partial oxidation of methane
CH4 + 0.5O2 a CO + 2H2, ∆H298 ) -35.7 kJ/mol
(2)
ATR 1 was performed under various operating conditions to investigate the effect of the steam-to-carbon ratio (S/C), oxygento-carbon ratio (O2/C), and space velocity on catalyst bed temperatures and product compositions. Table 1 summarizes the experimental results from ATR 1. The results indicate that changing the S/C ratio, O2/C ratio, and space velocity strongly influences the catalyst bed temperature and product compositions. However, just increasing the space velocity from 12596 to 23840 h-1 while holding the two other parameters constant showed no significant changes in product composition. Further doubling the space velocity to 47675 h-1 increased the catalyst bed temperature while reducing hydrogen production and methane conversion. The absence of changes in product composition between space velocities of 12596 and 23840 h-1 may be due to an oversized catalytic reactor. However, when the space velocity was increased to 47675 h-1, the significant reduction in residence time led to reduced hydrogen and methane conversion and also increased the catalyst bed temperature. Reducing the O2/C ratio resulted in less hydrogen production and decreased methane conversion due to less oxygen available for combustion. Reducing the S/C ratio had a similar effect with a reduction in both hydrogen production and methane conversion due to less available steam for steam reforming. However, a reduction in the S/C ratio also showed an increase in CO due to a lack of steam for the water gas shift. At a low S/C ratio there is a higher risk of carbon formation primarily due to the decomposition of methane and catalyst deactivation. After the ATR reactor was run at a low ratio of S/C ) 0.5 (a previous run at conditions GHSV ) 12596 h-1, S/C ) 1, and O2/C ) 0.6 was repeated to check the reproducibility) an increase in catalyst temperature and significant reduction in hydrogen production and methane conversion were observed, suggesting catalyst deactivation. However, SEM analysis performed on the ATR catalyst showed no evidence of carbon formation on the catalyst surface. Another possibility may be deactivation due to sulfur present in the natural gas, which is a known poison for autothermally reforming catalysts.5 Palm et al.6 reported no decrease in conversion during autothermal reforming of various hydrocarbon feeds in a small-scale tubular reactor (inner diameter 11 mm) with a precious metal catalyst at O2/C ratios between 0.34 and 0.47, S/C ratios between 1.5 and 2.2, and space velocities between 13000 and 18000 h-1 in the absence of any sulfur. Addition of 1-benzothiophene (10-30 wt ppm) in the feed caused deactivation of the catalyst, resulting in a decrease in conversion. They commented that the hydrocarbon mixtures could be converted into hydrogen-rich products by autothermal reforming with a wide range of operating variables. Without the heat loss of the small-scale tubular reactor, higher reaction temperatures and a less pronounced temperature profile in a larger reactor will enable operation at higher GHSV and also decrease the tendency of sulfur to adsorb on the catalyst.
2152 J. Phys. Chem. B, Vol. 109, No. 6, 2005
Fo¨ger and Ahmed
TABLE 2: Product Compositions for ATR 1, ATR 2, and ATR 3 conditions expt
GHSV (h-1)
S/C
ATR 1 ATR 2 ATR 3
12596 12596 25000
1 1 1
product composition (dry basis)
O2/C
Tcat (°C)
H2
O2
N2
CH4
CO
CO2
C2H6
C3H8
total
X(CH4)
0.6 0.6 0.6
777 793 912
35.85 36.41 35.05
0.69 0.75 0.99
41.61 41.62 42.57
5.89 5.11 4.91
5.85 6.27 7.74
9.95 9.75 8.61
0.16 0.08 0.12
0.00 0.00 0.00
100 100 100
73.2 3 76.02 77.18
Figure 3. Temperature profiles for ATR 3 (GHSV ) 25000 h-1, S/C ) 1, O2/C ) 0.6) at two locations in the bed.
Figure 2. Composition profile for ATR 3, GHSV ) 25000 h-1, S/C ) 1, O2/C ) 0.6.
Indeed, in our larger reactor much higher space velocities were used at higher temperatures. In addition to investigating the effects of the S/C ratio, O2/C ratio, and GHSV on catalyst temperature and product composition, these initial sets of experiments also determined a catalyst light-off temperature of about 300 °C. No reaction was observed for an ATR feed inlet temperature of less than 300 °C. Once the light-off temperature was reached, spontaneous combustion occurred, as observed by the rapid increase in catalyst bed temperature. It was also observed that the mixer section below the reactor absorbed significant radiation from the combustion reaction. However, the mixer temperature did not exceed the autoignition of the reactants, and hence, it was clear that combustion only occurred in the catalytic reactor. The results reported in Table 1 indicate the required catalyst light-off temperature and clearly demonstrate that the ATR reactor was able to produce a substantial amount of hydrogen under welldefined conditions and could run for weeks before deactivating. A further set of experiments, ATR 2, determined the ability of the ATR reactor to replicate the results from run 1 of ATR 1 at GHSV ) 12596 h-1, S/C ) 1, and O2/C ) 0.6. In ATR 1, it was found that doubling this space velocity had little effect on product composition. Experiment ATR 3 was then run at GHSV ) 25000 h-1, S/C ) 1, and O2/C ) 0.6 to confirm this observation. Table 2 summarizes the findings. Table 2 presents the average results over a period of operation time. As indicated above, results from ATR 1 and ATR 2 at the same operating conditions show similar catalyst temperatures and product compositions. Doubling the space velocity to 25000 h-1 shows little change in composition, suggesting the ATR reactor can be sized to a space velocity of 25000 h-1, therefore reducing the catalyst volume and weight without significantly influencing its performance. Figure 2 shows that the ATR reactor is capable of providing a stable and continuous supply of hydrogen as shown by the constant composition profile with only minor fluctuations. Figure 3 shows the catalyst temperature profile as a function of operational time for the same experiment. Similarly to the
composition profiles, there is an initial increase in temperature followed by a reduction, which steadies off, giving a relatively constant profile for a period of time. These observations are consistent with the exothermic heat of reaction of the partial oxidation reaction, which “kicks off” the reforming process and supplies heat for the endothermic steam reforming reaction. The “leveling off” of the temperature occurs when all the heat generated by the exothermic partial oxidation reaction is consumed by the endothermic steam reforming reaction. However, the steady temperature profile starts to increase after 250 h of continuous operation, suggesting the catalyst may start to deactivate; however, Figure 2 shows that the composition profiles are not greatly affected. Figure 3 also provides some information on the temperature profile within the length of the catalyst bed. All experiments have shown that, at light-off temperature, the immediate addition of air results in a spontaneous combustion reaction, observed by a rapid increase in temperature. Figure 3 suggests that the majority of combustion occurs at the bottom of the catalyst bed followed by the endothermic steam reforming, resulting in a reduction in temperature along the length of the catalyst bed. 3.1.1. Thermal Cycle. ATR 3 was also subjected to thermal cycles after running continuously at GHSV ) 25000 h-1, S/C ) 1, and O2/C ) 0.6 for about 3 weeks. The thermal cycling consisted of shutting down the reactor and starting it up each day to the same operating conditions to determine whether the catalyst could handle daily changes in load and temperature and still maintain its activityssimulating transient fuel cell operation. Table 3 (ATR 3 and thermal cycle results at operating conditions GHSV ) 25000 h-1, S/C ) 1, and O2/C ) 0.6) shows that apart from slight differences in the catalyst bed temperatures, the product compositions after each thermal cycle are similar with possibly a slight reduction in hydrogen and methane conversion. Further thermal cycle experiments are needed to determine with more confidence whether this trend was due to catalyst deactivation or only minor fluctuations in the system or gas chromatograph. 3.1.2. Comparison between Experimental Results and Thermodynamic Predictions. Equilibrium compositions can be calculated from equilibrium data as a function of temperature, pressure, and S/C and O2/C ratios. Predictions of the system running at O2/C ) 0.5 and S/C ) 1.0 are shown in Figure 4.
Catalysis in High-Temperature Fuel Cells
J. Phys. Chem. B, Vol. 109, No. 6, 2005 2153
TABLE 3: Thermal Cycles: GHSV ) 25000 h-1, S/C ) 1, O2/C ) 0.6
ATR 3 thermal cycle 1 thermal cycle 2
product composition (dry basis)
Tcat(bot) (°C)
Tcat(mid) (°C)
H2
O2
N2
CH4
CO
CO2
C2H6
C3H8
total
X(CH4)
913 952 969
831 871 849
35.05 34.92 34.50
0.99 0.73 0.71
42.57 42.73 42.95
4.91 5.29 5.66
7.74 7.21 7.36
8.61 8.98 8.65
0.12 0.13 0.16
0.00 0.00 0.00
100 100 100
77.18 75.67 74.26
Figure 4. Thermodynamic data at S/C ) 1 and O2/C ) 0.6 as a function of temperature.
TABLE 4: Experimental and Thermodynamic Compositions at S/C ) 1 and O2/C ) 0.5 observed thermodynamic prediction based on measured Tcat ) 750 °C thermodynamic prediction based on measured CH4 concentration, Tcat ) 589 °C thermodynamic prediction based on measured H2 concentration, Tcat ) 641 °C thermodynamic prediction based on measured CO/CO2 ratio, Tcat ) 538 °C thermodynamic prediction based on measured Toutlet ) 535 °C thermodynamic prediction based on measured average ATR reactor T ) 642 °C
It is clearly seen that H2 production occurs from ∼200 °C. The reaction is complete by ∼800 °C with no significant change in the concentrations of H2, H2O, CO, and CO2 from 800 to 1000 °C. As expected, at equilibrium, the ratio of H2 to CO is about 2.0. As mentioned above, a range of experimental conditions were investigated in the first set of experiments, ATR 1. These included the O2/C ratio, S/C ratio, and SV of the reactants. One such run was performed at conditions SV ) 11274 h-1, S/C ) 1.0, O2/C ) 0.5, Tcat ) 750 °C, and Toutlet ) 535 °C. Thermodynamic predictions based on the average ATR reactor temperature agree closely with the observed results for the major component H2. However, less CH4 and CO2 are predicted, and more CO is predicted. This indicates the combustion/conversion of methane is kinetically limited and the extent of partial oxidation is less than predicted (on the basis of the measured catalyst bed temperature, Tcat), indicating, when reacted, more of the methane goes to complete combustion to CO2 than predicted by thermodynamics. However, the hydrogen concentration prediction based on the average ATR reactor temperature is equal to the measured value. Interestingly, there is some slippage of O2 at that temperature in the presence of substantial amounts of unreacted H2, indicating the need to lower the space velocity or the presence of channeling effects. Thermodynamic analysis7 applied to natural gas autothermal
H2
O2
N2
CH4
CO
CO2
C2H6
C3H8
38.2 43.3 32.4 37.9 26.2 26.0 37.8
0.6 0.0 0.0 0.0 0.0 0.0 0.0
37.2 35.3 42.1 38.7 46.0 46.3 38.6
8.3 0.7 8.6 4.7 13.0 13.5 4.7
5.6 18.0 9.4 13.5 5.3 5.0 13.5
10.0 2.6 7.4 5.2 9.5 10 5.3
0.24 0.0 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0.0 0.0 0.0
reforming showed that the optimal O2/C ratio and S/C ratio for maximum hydrogen yield under the constraints of carbon-free and minimized carbon monoxide and residual methane content in the reformed gases are 0.67 and 2.3-3.6, respectively. With this condition, the product’s temperature under the assumption of adiabatic reaction is 547-598 °C. In our tests carbon-free operation was achieved at a much lower S/C ratio of 1.0 with the O2/C ratio and product temperature similar to those reported by Chan and Wang.7 3.2. Propane Pre-reforming. For internally reforming SOFC stacks, a methane-rich fuel stream needs to be supplied to the stack. The internal steam reforming reaction then assists in thermally managing the stack. This work had the objective to arrive at design data for a pre-reformer (using steam reforming) to convert propane to a methane-rich fuel for a multiple-kilowatt SOFC system. For carrying out an integral analysis of the reaction data, the heat- and mass-transfer parameters were calculated. These are shown with the reaction data in Table 5. It shows that if we assume n e 1 (first-order or lower in propane), the following criteria apply: (i) The axial dispersion criterion of L/dp > (20n/ Pe) ln(1/(1 - XA)) is satisfied. (ii) The criterion for the absence of a significant concentration difference between the bulk fluid and solid surface, (-r)rp/(Cbkc) < 0.15/n, is also satisfied. (iii) Deviations between the bulk fluid and solid surface partial
2154 J. Phys. Chem. B, Vol. 109, No. 6, 2005
Fo¨ger and Ahmed
TABLE 5: Calculations of Mass- and Heat-Transfer Effects reactant space velocity, h-1 fractional conversion rate of reaction, mol/(cm3 h) fluid mass velocity, kg/(m2 h) NRe jD Cb - Cs, mol/cm3 Pb - Ps, atm Pb, atm %(Pb - Ps)/Pb mass-transfer coefficient, m/h (-r)rp/(Cbkc) (20n/Pe) ln(1/(1 - XA)) Ts - Tb, °C
57840 .027 0.07 1596 9.0 0.47 4.74E-09 2.54E-04 0.747 0.0 1883 0.0006 0.3 0.3
pressures are less than 1%. (iv) Temperature differences between the bulk fluid and solid surface are less than 5 °C. Therefore, it is established that reaction data generated with 0.25 g of catalyst particles in the size range of 0.425-0.5 mm are free from heat- and mass-transfer effects at conversions below 40%. Having established these criteria, an integral analysis of the data was carried out:
W ) FAO
dX
∫k C A n c AO
The right-hand side of the equation was integrated for n ) 1 and n ) 0.5. By plotting the integral of dXA/((1 - XA)n) against WCAOn/FAO, the fit of the equation was checked and an estimate of the rate constant at 380 °C was made, with the result that the reaction is half-order in propane with a rate constant of 91.4 (cm3 mol)0.5/(g of catalyst h) at 380 °C. This rate constant formed the basis of the reactor scaling for designing an LPG pre-reformer. This empirical power law rate expression is in sharp contrast to the mechanistic rate expression reported by Ma,8 where the reaction was found to be 0.93-order in propane, -0.53-order in steam, and 0.86-order in hydrogen. 4. Conclusions 4.1. Autothermal Reforming. The performance of the ATR reactor has proven its capability of providing a continuous and stable supply of hydrogen under defined operating conditions, and similar results are obtained after thermal cycles and load changes-as required in fuel cell operation. However, a newly loaded catalyst does require a short stabilizing period of a few days before maintaining constant catalyst activity. The lightoff temperature for the ATR reactor using a platinum catalyst was found to be around 300 °C. Below this temperature little reaction occurred in the ATR reactor. Addition of air at the light-off temperature resulted in spontaneous combustion and a rapid increase in temperature. The reduction in temperature along the length of the catalyst bed suggested the majority of combustion occurs at the inlet, after which steam reforming takes over. Changing the S/C and O2/C ratios affects both catalyst temperature and product composition. A reduction in the S/C ratio resulted in less hydrogen production and methane conversion due to less available steam for steam reforming and a higher risk of catalyst deactivation due to carbon formation. After catalyst deactivation from ATR 1, carbon could not be detected by SEM. Thus, the cause for deactivation remains unclear, with deactivation due to sulfur present in the natural gas being another possible cause. It was also found that, below a critical space velocity, few changes in product composition were observed. Once the critical space velocity is exceeded, there is a reduction in hydrogen production and methane conversion primarily due
18000 0.175 0.116 497 2.8 0.754 1.91E-08 8.42E-04 0.392 0.3 942 0.0047 1.9 1.2
12000 0.302 0.162 331 1.9 0.89 2.79E-08 1.5E-03 0.309 0.5 741 0.0087 3.6 1.7
9000 0.388 0.156 248 1.4 1.0 3.19E-08 1.7E-03 0.238 0.7 625 0.0128 4.9 2.0
to the significant reduction in residence time. This finding provides important information for sizing the ATR reactor. 4.2 Propane Pre-reforming. The propane reforming rate determined by this work provided the basis for the design of a bench-scale pre-reformer reactor that has been successfully used in demonstrating operation of an SOFC stack on pre-reformed LPG.9 Acknowledgment. We acknowledge the contribution of all employees of Ceramic Fuel Cells Ltd. who took part in these investigations. Nomenclature CAO Cb Cs dp DT FAO GHSV jD kc L n NRe Pb Pe Ps r rp Tb Ts W XA
concentration of the reactant at the reactor inlet bulk-gas-phase concentration of the reactant solid-phase concentration of the reactant diameter of the catalyst particle diameter of the reactor tube molar flow rate of the reactant at the reactor inlet gas hourly space velocity j-factor for mass transfer reaction rate constant length of the reactor reaction order Reynold’s number bulk-gas-phase partial pressure of the reactant Peclet number solid-phase partial pressure of the reactant rate of reaction radius of the catalyst pellet bulk-gas-phase temperature solid-phase temperature weight of the catalyst conversion of the reactant
References and Notes (1) Fo¨ger, K.; Barrett, S.; Pham, T.; Ahmed, K. Fuel Cell System. International Patent WO03098728, 2003. (2) Christensen, T. S.; Primdahl, I. I. Hydrocarbon Process. 1994, 73, 39. (3) Krumpelt, M.; Krause, T.; Kopasz, J.; Wilkenhoener, R.; Ahmed, S. DOE Annual Laboratory ReView; Richland, WA, June 7-8, 2000. (4) Mears, D. E. Chem. Eng. Sci. 1971, 26, 1361. (5) Mawdsley, J.; Ferrandon, M.; Rossignol, C.; Ralph, J.; Miller, L.; Kopasz, J.; Krause, T. Hydrogen, Fuel Cells, and Infrastructure Technologies, 2003 Merit ReView; Berkeley, CA, May 19-22, 2003. (6) Palm, C.; Cremer, P.; Peters, R.; Stolten, D. J. Power Sources 2002, 106, 231. (7) Chan, S. H.; Wang, H. M. Fuel Process. Technol. 2000, 64, 221. (8) Ma, L. Ph.D. Thesis, University of New South Wales, Sydney, 1995. (9) Ahmed, K.; Gamman, J.; Fo¨ger, K. Solid State Ionics 2002, 152153, 485.
