New Mechanism for the Enhancement of the Oxygen Reduction Reaction on Stepped Platinum Surfaces

1. Introduction

There has been an intensive search for ways to increase the activity of catalysts for the oxygen reduction reaction (ORR) in order to decrease the catalyst cost in H₂-O₂ fuel cells, which are an integral component of the much-anticipated hydrogen energy economy. Much research has been devoted to fine-tuning the adsorption of reactants and intermediates in order to reach the summit of the famed volcano plot, in which reaction rate is plotted versus the adsorption strength of oxygenated species, OH for example. The adsorption strength of most oxygenated species, including O₂, O, OH, HO₂ all appear to be proportional to each other, the so-called scaling relationship [1]. Thus, if it is assumed that OH adsorption should be weakened, O₂ adsorption would also be weakened. This is a dilemma that has been largely ignored thus far. Nevertheless, it has been found experimentally that the use of Pt alloys such as Pt-Co with decreased OH adsorption strength has resulted in increased ORR rates. Interestingly, no Pt alloys have been reported for which O₂ adsorption is too weak, even though theoretical results show that adsorption on Pt-covered Pt-Fe should be very weak indeed, on the order of -0.15 to -0.20 eV, which should in principle have a detrimental impact upon O2 capture [2,3,4]. At this point, O₂ must compete with water for adsorption sites, as shown later. Clearly, this approach has its limits, since eventually _O2 adsorption would be impaired.

One key to this puzzle has been the recognition that platinum surfaces based on Pt(111) that include steps, either (110) or (100), can have increased ORR activity based on area, usually with a maximum activity being observed at intermediate terrace widths [5,6,7,8]. Early theoretical work predicted that stepped surfaces would be less active than Pt(111), based on the strong adsorption of OH at steps [9]. Bandarenka et al. proposed that the maximum in ORR activity at intermediate step density is due to a trade-off between OH removal and OH formation rates, based on H-bonded water network stabilization of OH [10]. Subsequently, Jinnouchi et al. have similarly explained the presence of a maximum in the ORR activity at intermediate step widths based on the trade-off between destabilization of OH and stabilization of HO₂ adsorbed on (111) terraces by water molecules [11]. In contrast, Hoshi and coworkers have argued that, since the step structures differ between (110) and (100) steps, with the activities being similar, the steps themselves cannot be the active sites, which are proposed to be the (111) terraces immediately adjacent to the steps [8]. Subsequently, Kodama et al. reported that a variety of Pt stepped surfaces in which the steps are capped with gold are still active for the ORR [12]. This result seems to strongly suggest that the steps themselves are not the active sites, based on the assumption that gold is inactive for the ORR, at least in acid electrolytes, although high activity can be observed in alkaline electrolytes [13,14].

The present work aims to take into account the clearly observed experimental results that the ORR activity in acid electrolyte mainly increases with step density, both (110) and (100) steps on Pt(111) surfaces except for very narrow (111) terraces. The simplest assumption is that the steps have an increased adsorption strength for O₂ so that the ORR current is proportional to the O₂ coverage at the step. Using DFT calculations, we have found that indeed the O₂ adsorption strength is significantly higher at steps compared with the terraces. In addition, we have found that water molecules adsorbed on either side of the bridging O2 can assist in its dissociation, creating 4 OH, and the activation energies are much smaller at both types of steps than on the (111) terrace. Thus, water is being used as a proton donor, which has been shown to be important for alkaline electrolytes but not thus far for acid [13].

In principle, these results can help to escape the scaling relation dilemma by the use of Pt alloys on which OH adsorption on (111) terraces is weak, but the _O2 adsorption at steps is still strong enough so that the overall reaction rate is enhanced. Rurigaki et al. have studied PtNi alloy stepped surfaces experimentally and have found some deviations from the trends observed for pure Pt surfaces, however [15]. Nevertheless, we believe that the present model may be useful in the search for more active ORR catalysts.

2. Results

O₂ has been found to be relatively stable on Pt(111) and has been observed directly with STM [16] [17]. Its adsorption energy is compared on these surfaces in Table 1. The adsorption energies for O2 adsorption on the (100) edge of Pt(533) and the (110) edge are quite similar, -1.96 and -1.86 eV, respectively, far larger than that on Pt(111), -0.89 eV. The presence of water stabilizes O2 on both steps slightly but destabilizes that on Pt(111). The adsorption energies of O2 on the (111) terraces of Pt(533) and Pt(553) are also similar to those on Pt(111) itself. Thus, O₂ is expected to adsorb at the steps rather than on the terraces, as long as the latter are not covered with OH, which adsorbs strongly. However, it must be kept in mind that the adsorption energy for OH is referred to gas-phase OH and thus is misleadingly large.

To get a more realistic idea of the first steps in the ORR, the energies of the O₂-2H₂O system have been plotted as the O-O bond is stretched in ca. 0.2 Å increments on Pt(533), Pt(553) and Pt(111) (Fig. 1), in much the same way as was done for O2 dissociation on Pt(111), Pt(211) and Pt(221) in a UHV study by Gambardella et al. [17]. As with the latter study, the behavior on the two stepped surfaces is somewhat similar, but the activation energy E act is somewhat smaller on Pt(533), 0.21 vs. 0.28 eV (Table 2). Both energies are far smaller than that on Pt(111), 0.55 eV. In contrast, Gambardella calculated values of 0.9 eV for O₂ dissociation energies on three surfaces, Pt(111), Pt(211) and Pt(221) using DFT. However, Gee and Hayden found O₂ to be dissociated more readily on Pt(533) than Pt(111) in UHV [18]. The corresponding atomic models for the overall process are shown in Figure 2, Figure 3 and Figure 4 for Pt(111), Pt(533) and Pt(553), respectively.

The reasons for the great differences in E _act are proposed to involve both the stronger adsorption of O₂ but also that of H₂O, which is ca. 0.75 eV stronger at the steps than on the (111) terrace. Thus, the reactants are held in place firmly during the overall reaction. Even more importantly, O₂ adsorbs more strongly at the steps than does water, in contrast to the situation on Pt(111), on which water is calculated to adsorb more strongly than O₂, which would lead to an impairment of the ORR activity. This is a result that has not been reported thus far to our knowledge. It should also be noted that water adsorption does not follow the usual scaling relations, so that it is possible for water to adsorb more strongly than O₂.

The overall reaction can also be plotted to show the differences in the initial O₂ adsorption energies (Fig. 2). This representation emphasizes the great advantage of the stepped surfaces in terms of capturing O₂ from solution. These reaction profiles follow the well-known BEP relation, which predicts that, for dissociation reactions, the activation energy should decrease as the energy of the products is decreased, as long as the activated state is similar to the product state [25].

Figure 5. Reaction profiles for the first steps in the ORR involving O₂ adsorbed in the bridging configuration at (100) steps on Pt(533) (green circles), at (110) steps on Pt(553) (red triangles), and on Pt(111) (blue squares), as in Fig. 1 but with the energies referred to the adsorption energy of O₂ on the surface with two water molecules present.

Figure 5. Reaction profiles for the first steps in the ORR involving O₂ adsorbed in the bridging configuration at (100) steps on Pt(533) (green circles), at (110) steps on Pt(553) (red triangles), and on Pt(111) (blue squares), as in Fig. 1 but with the energies referred to the adsorption energy of O₂ on the surface with two water molecules present.

In contrast to most recent DFT studies, we have elected to plot the results in terms of the directly calculated DFT energies of the systems as the O-O bond is stretched, with the numbers of atoms necessarily remaining constant. The approach is essentially that which would be used in a purely chemical reaction in a closed system, so that, as the reaction proceeds, a reverse driving force builds up. The way this occurs in the present approach is that the effective electrochemical potential U becomes more positive during the reaction, which would cause the rate of a reduction reaction such as the ORR to slow down.

U can be thought of as a linear function of the Fermi energy E_F, which can be plotted versus the reaction coordinate (Figure 6). In order to produce a reaction profile in which U is constant, it will be necessary to perform calculations on a series of similar systems in which the initial E_F is varied and then curve-fit the energy surface, i.e., the energy vs. E_F vs. reaction coordinate.

3. Discussion

The present work attempts to shed light on the reasons for the increased ORR rate in acid electrolytes on stepped Pt surfaces with increasing step density, based on the simplest possible assumption, which is that the O₂ adsorbs more strongly at the steps than on the terraces. This point is crucial, because, on Pt(111), water adsorption is similar in strength to that of O₂, and, on certain alloys, O₂ adsorption can be significantly weaker than that of water. Therefore, steps are necessary in order to ensure that O₂ can be captured effectively from the electrolyte. It has been proposed by various groups that the effect of the step is to modulate the water structure close to the catalyst surface and thus stabilize or destabilize intermediates in the ORR. However, this model does not take into account the competition between O₂ and water for adsorption sites on the (111) terrace.

4. Materials and Methods

DFT calculations were carried out with the use of the DMol³ program (BIOVIA, Dassault Systemes, v. 2023) with periodic boundary conditions for Pt(533), Pt(553) and Pt(111). The first steps in the ORR have been modeled starting with one O₂ molecule and two H₂O molecules.