快速时时彩: O2-reduction at high temperature MCFC.pdf
O 2 -reduction at high temperature: MCFC K. Hemmes 1 andJ.R.Selman 2 1 Delft University of Technology, Delft, The Netherlands 2 Illinois Institute of Technology, Chicago IL, USA 1 INTRODUCTION Oxygen reduction in the cathode of the molten carbonate fuel cell (MCFC), stoichiometrically described by the reaction 0.5O 2 + CO 2 + 2e ? ===? CO 3 2? (1) has been intensively studied as part of MCFC development programs since 1975. The cathode exhibits by far the largest polarization of the two electrodes in an MCFC. This is the main reason for the lower power density of the state-of-the- art MCFC relative to that achieved by other types of fuel cells, in particular the proton exchange membrane fuel cell (PEMFC) and solid oxide fuel cell (SOFC) (except in area cells smaller than the state-of-the-art MCFC). Comparative studies have been carried out on the catalytic activity of several candidate cathode materials. Extensive effort has been applied to elucidate the reaction mechanism of oxygen reduction and the specific reac- tion/diffusion processes occurring in the porous cathode, to identify the cause(s) of the relatively high polarization losses and mitigate them if possible. The number of possible electrode materials is very limited: few metals or alloys are stable enough to persist in the very corrosive molten car- bonate environment, and those few that do (generally noble metals) are too expensive to be of practical importance. Oxide electrocatalysts that are sufficiently stable, such as the state-of-the-art material lithiated NiO, form naturally porous structures which cannot be characterized simply in terms of reaction kinetics, but require modeling of the entire electrode process. The stability of the state-of-the-art cathode material NiO in relation to the melt chemistry is of great importance for the MCFC and has been extensively investigated. Since the melt chemistry also impacts on the electrode kinetics, it is discussed below in detail in the context of both stability and kinetics of NiO. A theoretical description of the equilibria in the melt is worked out using concepts of linear alge- bra. From this theory it becomes clear that many reaction mechanisms are possible and will occur in parallel. Fur- thermore, a shift from one dominant mechanism to another is expected, depending on the gas composition. This has been confirmed experimentally using gold flag electrodes in bulk melt. Determination of the reaction mechanism in a porous electrode such as NiO is complicated by the complex interaction between kinetics and transport in the potential distribution, and therefore overall polarization, of the electrode. ThepeculiarroleofCO 2 in the oxygen reduction process is discussed. There is general agreement that, in addition to one or more electrochemical steps breaking oxygen or reduced oxygen down to oxide ions, O 2? , the recombination of CO 2 with oxide ions: O 2 ? + CO 2 ===? CO 3 2? (2) constitutes a rate-limiting step in the mechanism, especially within a porous electrode. This has been attributed to slow kinetics of the homogeneous recombination reaction. Recent evidence, however, suggests that the dissolution Edited by Wolf Vielstich, Hubert A. Gasteiger, Arnold Lamm and Harumi Yokokawa. ? 2010 John Wiley however, in the case of oxygen reduction, it can make a difference in modeling predictions (unless a vastly superior electrocatalyst than the above transition metal oxides is identified!). 4.6 The role of CO 2 in the rate limitation of oxygen reduction Early investigators of the obvious rate limitation associ- ated with oxygen reduction in molten carbonate, even at 650 ? C and higher, tried to identify a slow step among those elements of the oxygen reduction process that were experi- mentally evident, as well as other, hypothetical elements of the reaction mechanism. Thus elements such as gas disso- lution, dissolved-gas transport, electron transfer kinetics in one or several steps, ionic transport, and possible chemical reaction steps were considered. Dissolved-gas transport and electron transfer kinetics are the classical “usual suspects” in porous-electrode rate limitation. However, in the case of the MCFC cathode, special attention has been focused for a long time on the (supposedly) last step, the chemical reaction recombining oxide ion and CO 2 , equation (2) or (A4), and more recently on the first physical step, gas dis- solution, in particular CO 2 dissolution. Both gas dissolution and oxide–CO 2 recombination received considerable attention from early investigators of the MCFC electrode processes. A study by Borucka  reporting work carried out at the Institute of Gas Tech- nology (IGT) in the early 1970s led later researchers to believe that dissolution of O 2 and CO 2 in molten carbonate is fast. Her conclusion was based on the fact that switch- ing the O 2 /CO 2 gas supply on and off caused relatively fast changes in the potential of the gold flag electrode. This was observed when the potential was measured with respect to a counter electrode, and also when a reference elec- trode was used. However, as discussed below, scrutiny of Borucka’s data, combined with evidence from more recent experiments, suggest that her conclusion was premature. In the same paper by Borucka, potential relaxation mea- surements after addition of oxide ions (in the form of Li 2 O) were reported.  These experiments, in contrast to the gas dissolution rate measurements, generated considerable interest and helped give the hypothesis of a slow recombina- tion reaction an aura of strong probability. It is noteworthy that Borucka’s interpretation agreed with the strongly held belief of Broers, the MCFC pioneer, in this slow chemical step.  It also appeared to corroborate systematic observa- tions by early workers in molten carbonate chemistry, such as Janz’s group, that oxide generated cathodically was not rapidly recombined with CO 2 so that carbonate electrolysis was characterized by marked hysteresis.  Areviewwas given by Selman and Maru.  During the 1980s, several efforts were made to repeat and extend the oxide addition/potential relaxation measure- ments of Borucka, and interpret them quantitatively. The oxide concentration in the melt was monitored by means of a gold probe whose potential was continuously mea- sured with respect to an oxygen reference electrode. It was found that the oxide concentration decays with a time constant of the order of tens of minutes, suggesting a reac- tion rate constant for the recombination reaction (2) of the order of 5 × 10 ?3 s ?1 in Li/K eutectic at 650 ? C, with an unexpectedly low activation energy of 19 kJ mol ?1 .  In these same experiments, and further extensive measure- ments by Ramaswami,  the potential relaxation upon addition of peroxide and superoxide was also investigated. Peroxide and superoxide appeared to be decomposed to oxide rapidly. However, the potential responses of these reduced oxides exhibit, depending on the CO 2 /O 2 ratio, sharp initial potential shifts (both positive and negative), preceding a slow final relaxation as upon oxide addi- tion. This suggests a very complicated interaction among the six or seven interacting oxygen-containing species, and does not clearly point to one potential-determining reaction. Analyzing quantitatively the wealth of data gath- ered, Ramaswami  presented an improved kinetic model including peroxide decomposition kinetics, but was not able to give a unifying interpretation, accounting for the CO 2 /O 2 effect. Hence there is no clear convergence on the recombination kinetics as a single cause of rate limitations. Moreover, porous cathode modeling that included a recom- bination reaction rate constant of the order found by Lu  12 The oxygen reduction/evolution reaction did not predict correctly the weak dependence of overall cathode polarization on pCO 2 (in the range 0.10–0.15).  Faced with these contradictory results, other researchers have investigated the effect of humidification as a pos- sible way to reconcile the apparent kinetic slowness of the oxide–CO 2 recombination reaction with the observed trends of polarization. Reactions (16a) and (16b) above pro- vide an alternative to the direct recombination of O 2? and CO 2 provided that H 2 O is present and the forward reac- tions of (16a) and (16b) are sufficiently fast. Nishina et al. have confirmed that when pO 2 and pCO 2 are kept con- stant, the addition of water vapor has a positive effect on the overall reaction speed, provided that pCO 2 is not too high (in which case oxide activity is very low).  How- ever, they found the effect to be small. This suggests that the recombination reaction may not be intrinsically slow. Peelen et al., who repeated Borucka’s potential relaxation experiments in their entirety, arrived at the same conclusion by a different route, i.e., by a critical review of various results which suggest that gas dissolution rate, far from being intrinsically fast, may under some circumstances be slow enough to explain the observed potential relaxation results.  4.7 Dissolution rate of O 2 and CO 2 into the melt A critical review of Borucka’s results  leads one to the question of whether or not gas dissolution may be excluded as a rate-limiting step. When, in Borucka’s experiments, the gas was switched off, the potential drifted in the negative direction. According to the text of Borucka’s paper, the decay rate was 10 mV s ?1 , but the figure shown suggests a much slower rate, of 1 mV s ?1 . The potential recovered rapidly (within a time period of the order of 1 s) after switching on the gas flow again. However, the potential did not fully recover to the original value (?35 mV) but remained at ?60 mV, which was not explained. In the same experiment, but this time performed with a reference electrode (of much smaller dimensions), the potential recovered with an overshoot. This, likewise, was not clarified. Hence there are a number of unresolved issues in relation to the results which are in themselves interesting. Therefore, the conclusion that O 2 and CO 2 dissolution is fast seems premature. Although Peelen et al.,  following the same approach as Borucka  and Lu,  consistently found apparent relax- ation times of the order of minutes, they do not believe that the oxide–CO 2 recombination reaction (2) would suffice as an explanation. Instead, they state that the assumption of slow CO 2 dissolution yields the same theoretical relaxation curves and could equally well explain the results. In later oxide addition experiments using an improved oxide concentration probe, Peelen et al. confirmed that CO 2 dissolution is rate-limiting.  In these oxide addition experiments Ni 2+ ions were used as a probe in combination with the very sensitive electrochemical technique of square- wave voltammetry (SWV). Peelen et al. observed a slow relaxation of the Ni 2+ and CO 2 peaks in their SWV traces back to equilibrium values, at the same rate. Therefore the CO 2 dissolution, not its reaction with oxide,mustbe causing the potential relaxation. CO 2 dissolution therefore is rate-limiting. They determined a dissolution rate of 25(±5) × 10 ?4 cm s ?1 and a Henry’s constant for CO 2 of 20(±5) μmol cm ?3 atm ?1 . The latter value is very close to that determined by Broers and co-workers in ternary eutectic at 700–830 ? C, namely 17 μ mol cm ?3 atm ?1 . [42, 49] In the same paper, Peelen et al. also presented a cal- culation which strongly suggests that the recombination reaction cannot be slow.  In an operating MCFC, cathode oxide ions are produced in large concentrations, irrespec- tive of the precise reaction mechanism. If the recombination reaction were as slow as calculated from the results of addi- tion experiments when slow CO 2 dissolution is ruled out (k f = 8.1 × 10 ?2 atm ?1 s ?1 ), then not only would the melt within the pores of the cathode become very basic, but also the oxide concentration would actually exceed the solubil- ity limit of Li 2 O. Its solidification in the cathode would immediately lead to a steep decrease in performance by blockage of the electrode interface. This, however, is not observed, so the recombination reaction must be faster than was assumed, and slow dissolution of CO 2 must be taken into account as a rate-limiting factor, perhaps together with a somewhat slow recombination reaction. Assuming exclusively slow gas dissolution as the cause of rate limitation, Peelen et al. estimated the gas dissolution rate constant to be k CO 2 = 2.5 × 10 ?3 cm s ?1 From this, they calculated limiting current densities in an MCFC cathode as a function of pCO 2 and active surface area as follows i lim = nFk CO 2 K H pCO 2 A eff (19a) or, with the dimensions indicated: i lim [mA cm ?2 ] = 10pCO 2 [atm] × A eff [cm 2 cm ?2 ] (19b) This is graphically shown in Figure 5.  At a typical specific active surface area A eff of 250 cm 2 cm ?2 ,the limiting current density would be about 150 mA cm ?2 when pCO 2 is approximately 0.06 atm. This appears to agree with reality in practical MCFC systems, since pCO 2 is always kept at or slightly above 0.05 atm to avoid steeply increased cathode polarization encountered at partial pressures below this value. 13 O 2 -reduction at high temperature: MCFC 0 250 500 750 1,000 A eff (cm 2 cm ?2 ) 0.0 0.3 0.6 0.9 CO 2 rate limited i max (A cm ? 2 ) BET surface area pCO 2 = 0.3 bar pCO 2 = 0.15 bar pCO 2 = 0.05 bar Figure 5. Predicted CO 2 dissolution rate-limiting current densi- ties of the MCFC cathode as a function of specific gas/electrolyte surface area in the cathode for several pCO 2 in the cathode gas. If pCO 2 of the cathode gas were 0.05 atm at the standard current density of 150 mA cm ?2 , the melt would still be acidic, from the viewpoint of NiO dissolution, in spite of the large oxide ion production. This was confirmed in specially designed experiments to assess NiO dissolution under load in a pressurized cell.  Upon increasing the total pressure (and thus the pCO 2 at the same gas composition), NiO dissolution did increase as predicted by the acidic dissolution mechanism. If the recombination reaction were slow, the melt would become very basic and one would expect a basic dissolution mechanism (as discussed in Section 3) and a decrease in NiO dissolution with increased pressure (see Figure 1). Again, this is not observed, thus undercutting the long popular hypothesis of slow recombination kinetics. The fact that CO 2 dissolution is slow need not completely contradict the experiment and conclusions by Borucka.  It is not unlikely that CO 2 dissolution is slow, but O 2 dissolution much faster. This is even plausible, in the light of the accepted interpretation of their respective gas solubilities, i.e., that CO 2 dissolves only physically in molten carbonate, whereas O 2 dissolution is predominantly chemical, e.g., O 2 + CO 3 2?