# 快速时时彩: Reaction mechanisms of the O2 reduction_evolution.pdf

Reaction mechanisms of the O 2 reduction/evolution reaction M. Gattrell and B. MacDougall Institute for Chemical Process and Environmental Technology, Ottawa, Canada 1 INTRODUCTION The electrochemistry of oxygen (its reduction to water and evolution from water) has been extensively and intensively studied because of its fundamental complexity and impor- tance for many practical systems. Oxygen reduction is the key to achieving the overall fuel oxidation reaction in fuel cells. It is also of interest for metal–air batteries, and as a cathodic reaction to reduce the cell voltage in electrochem- ical processes. A deep understanding of oxygen reduction is also important because it is often the counter electrode reaction for many corrosion systems. Oxygen evolution occurs in electrolyzers, as a counter electrode reaction for electrowinning, and as a competing process for many oxida- tive synthesis reactions (e.g., chlorine, ozone). It is also important for recharging of metal–air batteries and regener- ative fuel cells. One interesting application combining both reactions is electrochemical oxygen pumps. These reduce oxygen at one electrode from a low pressure and/or low purity feed stream, then evolve oxygen from the other elec- trode at high purity and possibly also high pressure. The reaction is also of fundamental interest because it is a good example of an electrocatalytic reaction (thus involving adsorption) and so shows a great sensitivity to the electrode surface and the presence of other adsorbed species. Studies have been carried out investigating the effects of different electrode materials, different crystal faces, surface modification by adatoms, and the influence of different electrolytes or contaminants. The progress made in the understanding of oxygen elec- trochemistry over the years and by different groups has been the subject of numerous reviews. These include those by J. P. Hoare (1968), [1] A. Damjanovic (1969), [2] J. P. Hoare (1974), [3] M. R. Tarasevich, A. Sadkowski, and E. Yeager (1983), [4] A. J. Appleby (1992), [5] A. Damjanovic (1992), [6] K. Kinoshita (1992), [7] J. O’M. Bockris (1993), [8] R. Adzic (1998), [9] among others. The importance of better understanding, and so hopefully better catalyzing, the oxygen reduction reaction can be seen in the high overvoltage in an operating fuel cell related to oxygen reduction (see Figure 1). Being from a real fuel cell, these curves also reflect some transport limits within the catalyst and backing layers in addition to the kinetic losses. However, for the 5 atm results, where the transport losses have been estimated to be small, the electrode potential is still about 400 mV below the theoretical potential. The potential drop at the cathode is even more important for direct methanol fuel cells, where methanol that crosses the cell membrane further decreases the cathode potential. The overall reaction for the 4-electron reduction of oxy- gen to water is shown below (In acid) O 2 + 4e ? + 4H + ???→ 2H 2 O E 0 = 1.229 V vs. SHE (1) From thermodynamic data, the potential has been calculated to be 1.229 V vs. standard hydrogen electrode (SHE). [4] The reaction, however, involves multiple intermediates, Edited by Wolf Vielstich, Hubert A. Gasteiger, Arnold Lamm and Harumi Yokokawa. ? 2010 John Wiley ? disk current, and x ring current. (0.1 N H 2 SO 4 , ω = 120 s ?1 , ring at 1.4 V vs. RHE). (Reproduced from Damjanovic et al. (1967) [23] by permission of The Electrochemical Society, Inc.) 6 Reaction mechanisms of the O 2 reduction/evolution reaction This example illustrates how the RRDE allows one to distinguish between 2- and 4-electron reductions. It does not directly allow one to distinguish between a parallel 4- electron reduction and a 4-electron reduction via peroxide. Therefore a significant amount of work has focused on examining the dependence of the RRDE response with rotation rate and disk potential to try to obtain more information. Various reaction pathway models have been developed and their expected responses evaluated, then compared to experimental data. This approach has to be used with care, however, because any number of mechanisms may be able to fit a given set of data. An initial approach (Scheme 1) was described by Dam- janovic et al., that distinguished between the series mech- anism (through peroxide (k 2 + k 3 )) or the parallel mech- anism (with earlier breaking of the oxygen bond then reduction of the two fragments to water (k 1 )). [24] O 2 bulk 2H 2 O HOOH HOOH bulk diff. K 2 (2e ? )K 1 (4e ? ) K 3 (2e ? ) 2H 2 O Scheme 1 At steady state, the rate equations can be solved to yield I D I B = I D (I R /N) = parenleftbigg 4k 1 2k 2 + 1 parenrightbigg + parenleftbigg 4k 1 2k 2 + 2 parenrightbigg k prime ω ?1/2 (15) where I D is the disk current, I R the ring current, N the ring collection efficiency, I B the ring current corrected for the collection efficiency, ω the RRDE rotation rate, and k prime = 1.61D ?2/3 ν 1/6 k 3 (where D is the diffusion coefficient of peroxide and ν is the solution kinematic viscosity). The rotational dependence of the ring-disk data was then used to distinguish between the different pathways using a plot of I D /I B versus ω ?1/2 . It should be noted, however, that the relationships derived are based on the assumption that the rate of release of peroxide from the disk is governed by mass transport, which results in its dependence on rotation rate. If this is not the case, due for example to desorption-controlled release of the peroxide, the intercept at infinite rotation rate will be more complicated and the results of the I D /I B versus ω ?1/2 plots more difficult to interpret. Thus this model is not sufficiently general, and by not including such reaction steps as peroxide decomposition, and the adsorption and desorption of reactants and products, the results of an analysis with this model (and other simplified models) will reflect the starting assumptions. Unfortunately, it has been shown that for more general models, there is insufficient data available to determine all the rate constants. [25] Accordingly, different simplified models have been used for specific electrode materials, and data on the electrochemistry of peroxide solutions has also been used to try to determine some of the constants. Listings and comparisons of the models can be found elsewhere. [4,7,9,26] In addition, work has been done to develop diagnostic criterion in the form of plots, which can be used to compare the relative rates of different reaction pathways. A commonly used plotting method was developed by Wroblowa et al., for the proposed reaction pathways, shown in Scheme 2. [27] O 2 bulk O 2 surface 2H 2 O H 2 O HOOH ads k ?2 (?2e ? ) k 2 (+2e ? ) k 5 k ?5 HOOH surface HOOH bulk diff. k 3 (+2e ? ) 2H 2 O k 1 (+4e ? ) k 4 Scheme 2 This scheme does not explicitly include an O 2ads species (essentially lumping together the O 2 surface and O 2ads ,which is mathematically equivalent to assuming a rapid equilib- rium between O 2 surface and a low coverage of O 2ads ). For this scheme, the resulting equations are I D I B = I D (I R /N) = 1 + 2 k 1 k 2 + A + k 5 γ A ω 1/2 A(16) where A = parenleftbigg 2k 1 k 2 k ?5 parenrightbigg (k ?2 + k 3 + k 4 ) + (2k 3 + k 4 ) k ?5 (17) and γ A =?0.62D 2/3 ν ?1/6 (18) It was noted that for an I D /I B versus ω ?1/2 plot of these equations, the intercept (J) at infinite rotation rate would be given by Intercept = J = 1 + 2 k 1 k 2 + A(19) 7 The oxygen reduction/evolution reaction and the slope (S)by Slope = S = k 5 γ A A(20) Then for a series of curves measured at different potentials, the intercepts and slopes can be related by J = 1 + 2 k 1 k 2 + γ A k 5 S(21) The resulting J –S plots provide an additional method for viewing data. If k 1 and k 2 have the same potential dependence and if k 5 is potential independent, a linear plot with an intercept of (1 + 2k 1 /k 2 ) will result. This intercept can be used to find the ratio of k 1 (the parallel mechanism) to k 2 (the series mechanism). If k 1 and k 2 have different potential dependencies, the plot will appear curved. An example of such a plot is shown in Figure 8 for oxygen reduction in 85% phosphoric acid. [28] The straight line relationships in the I D /I R plot confirm that the reaction is first order in oxygen. The straight line J –S plot indicates that k 1 and k 2 have the same potential dependency with the intercept (1 + 2k 1 /k 2 ) of 5.2 giving an estimate of k 1 /k 2 of 2.1. On gold, J –S plots show an intercept of close to one, indicating that k 1 is essentially zero, [27] and for platinum in 0.55 M H 2 SO 4 a curved plot is reported, indicating 1 0 10 20 30 2 3 4 100 0 8 12 16 50 150 j I D / I R S × 10 2 I / √ω (rpm) ?1/2 × 10 2 0.30 V 0.32 V 0.35 V 0.37 V 0.40 V 0.45 V 0.50 V Figure 8. I D /I R versus ω ?1/2 ,andJ –S plots for oxygen reduc- tion on platinum. (85% H 3 PO 4 ,25 ? C, ring potential 1.2 V, volt- ages versus RHE, N = 0.40). (Reproduced from Adzic (1998) [9] by permission of Wiley-VCH.) (according to the model) different potential dependencies for k 1 and k 2 . [26] In the latter paper, a comparison of some simplified models was made and it was concluded that, for their case, the model of Damjanovic was adequate. Then, using a combination of I D /I B versus ω ?1/2 and I Dlim /(I Dlim ? I D ) versus ω ?1/2 plots, values of k 1 , k 2 and k 3 were obtained at various potentials. These results are shown in Figure 9. The parallel route dominates (k 1 greatermuch k 2 ) from about 0.8–0.6 V RHE. However, below about 0.6 V, k 1 becomes potential independent (which corresponded to an observed flattening of the mass transfer corrected Tafel slope) leading to speculation about an oxygen adsorption or chemical rate- determining step before the first electron transfer. At very low potentials (around 0.4 V), the series pathway begins to dominate, though k 3 remains larger than k 2 and so most of any peroxide produced would be expected to be further oxidized to water. A very general reaction pathway has been developed by Anastasijevic et al. [29] (see Scheme 3: modified for acid conditions for ease of comparison with the other schemes). This scheme is written using only single electron transfer steps, and explicitly includes oxygen irreversible dissociative adsorption at sites for strong adsorption, versus weak adsorption sites for the series route. It also allows for transfer of series pathway intermediates at weakly adsorbing sites to strongly adsorbing sites of the parallel pathway, effectively blurring the difference between a series and parallel route. They found when using the more general model that interpretation of the various plots was more complex, but a number of conclusions could 10 ?3 10 ?2 10 ?1 10 ?4 0.8 0.7 0.6 0.5 0.4 P otential (V vs . RHE) k 1 k 2 k 3 Electrolyte: H 2 SO 4 (0.55 M) Model 1 Rate constant (cm s ?1 ) Figure 9. Rate constants for the intermediate steps of O 2 reduc- tion on Pt as a function of potential. Constants calculated based on Scheme 1. (0.55 M H 2 SO 4 ,25 ? C). (Reproduced from Hsueh et al. (1983) [26] by permission of Elsevier Science.) 8 Reaction mechanisms of the O 2 reduction/evolution reaction O S ads k 22 O 2 surface O 2 bulk diff. k 23 k 24 HO 2 S ads O 2 ads HO 2 ads HOOH ads k 1 (e?) k 2 (e?) k ?2 (e?) HOOH surface HOOH bulk diff. k 25 k ?25 k 4 (e?) k 5 (e?) k ?5 (e?) OH S ads HOOH S ads k 6 (e?) H 2 O H 2 O k 12 k 13 k 10 (2) H 2 O k 11 (2) k 21 k 21 k ?20 k 3 (e?) Scheme 3 still be drawn. [29] A linear I D /I B versus ω ?1/2 plot is characteristic of a pathway with first order (or pseudo first order) elementary reaction steps that produces only one compound that reacts at the ring. Non-linear J –S plots are characteristic of coupled series and parallel pathways (i.e., k 22 and/or k 23 0). A linear J –S plot with an intercept greater than 2/n R (where n R is the average number of electrons exchanged per oxygen) implies that a parallel pathway is operating and the ratio of parallel to series path can be estimated. And the J –S plot consists of a single point when only the 2-electron series pathway is operating. Finally, while fitting such a complex model does not seem possible using electrochemical data alone, simplified models must be used with caution to avoid misleading conclusions that can be made using a model that overlooks possible reaction intermediates or steps. Nevertheless, some useful qualitative and quantitative information can be obtained. The concept of different sites for the parallel 4-electron and series 2- and/or 4-electron reductions of oxygen has developed from numerous experiments over the years. These experiments have shown the non-equivalent influence of adsorbed and sub-monolayer deposits of species (under potential deposition (UPD) metals or partially reduced oxides) on the two reactions. [4, 9] In general on platinum, while such species tend to inhibit the overall reduction of oxygen, they more strongly inhibit the parallel reaction pathway. An example of such work is shown in Figure 10. This work shows the influence of different anions on the RRDE response at a palladium electrode (which behaves generally similarly to platinum). It can be seen that in the presence of anions such as H 2 PO 4 ? ,Cl ? ,andBr ? ,the overall process of oxygen reduction at the disk decreases, but the amount of peroxide oxidized at the ring increases. [30] This data was also examined in terms of a reaction pathway model (similar to Scheme 2) and it was found that these results were related to a decrease in k 1 /k 2 (rather than simply a decrease in k 3 ). One explanation for this is that the parallel mechanism requires two adjacent sites for dissociative adsorption, while oxygen reduction to peroxide may only require one site. This would have the general effect of making the 4-electron reduction second order in available sites, and so be more sensitive to changes in the number of available sites. [4] With the ability to prepare good quality single crystal RRDEs, these approaches are now being extended to investigating the effects of surface structure on the oxygen reduction reaction pathways. [9, 31] This work has confirmed the important role of structure-sensitive adsorption of anions. Adsorbed hydrogen (in the H UPD region) decreases the disk current and quantitatively increases the ring current, which is felt to be due to a blocking of sites for breaking of the O–O bond. [31] On the other hand HSO 4 ? was found to inhibit O 2 reduction, but not to affect the 9 The oxygen reduction/evolution reaction 1.0 0.8 0.6 0.4 0.2 0.0 0.25 0 0.75 0.5 I 0 (mA) 40 20 I R ( μ A) ? (V) 12 34 5 6 7 8 9 10 11 12 1′ 2′ 3′ 4′ 5′ 6′ 7′ 8′ 9′ 10′ 11′ 12′ Figure 10. RRDE results for oxygen reduction at a Pd electrode showing the effect of adsorbed anions. 1: 0.1 N H 2 SO 4 ,2:0.1N H 3 PO 4 , 3–6: 0.1N H 2 SO 4 + HCl, 7–12: 0.1 N H 2 SO 4 + HBr; at concentrations of: 7: 10 ?8 N, 8: 10 ?7