The electrochemistry of polycrystalline platinumWhile most fundamental studies focus on the low-index facets of platinum, we performed our analysis on Pt(pc). While the Pt(pc) surface has the disadvantage of not being uniquely defined on the surface atom level, the low index planes are unstable when cycled to potentials above 900 mV (Pt(100), Pt(110)) and 1050 mV (Pt(111))3. In contrast, the Pt(pc) exhibits exceptionally stable features and kinetics in a potential window ranging from the HER to the OER region and can be viewed, to a crude approximation, as a superposition of various Pt facets30. Compared to a single crystalline surface, the polycrystalline surface can be repeatedly cycled to high potentials, maintaining cleanliness and reproducibility over an extended series of measurements, see Fig. S1b and the associated discussion. The cyclic voltammogram (CV) of Pt(pc) can be separated into five regions (Fig. 2a). Starting from the double layer region (DL), the water and electrolyte dipole interactions provide a capacitance of around 80–100 µF/cm2. Going negative, supposed underpotential deposited hydrogen (H-upd) forms reversibly, and molecular hydrogen evolves (HER) when approaching the RHE potential at 0 mV; noting that all potentials in this contribution refer to the RHE scale at pH = 1. When reversing the scan direction, any adsorbed molecular hydrogen is oxidized to water (HOR), transitioning into the oxidation of the H-upd until the DL region is reached, after which oxygen species form on the surface. The nature of these species is not fully understood. Likely, reversible hydroxyl species (*OH) form on the surface, while most of the subsequent charge stems from other, electrochemically irreversible, oxygen species31. From the symmetry of CVs on Pt(111) and Pt(100), we know that the charge associated to *OH is strongly reversible32, similarly to the H-upd charge, and is also affected by the presence of specifically adsorbing anions such as bisulfate. Using scanning potential controlled electrochemical impedance spectroscopy (SPEIS), substituting the equivalent AC Voltammetry technique, used previously to determine oxygen species on platinum33,34 (see the SI for details, Fig. 2b, c), we can separate the *OH charge (40 ± 5 µC/cm2) from the irreversible charge. We find that species causing the irreversible charge dominates the oxygen region, denoting them as *O from hereon. The *OH and the elusive *O mimic the pseudocapacitive nature of the H-upd, as indicated from CVs recorded at various scan rates (Fig. S1c). This supports the distinction between different oxygen species made previously35. We, therefore, use the term O-upd for the entire surface oxygen region, a term initially introduced in 197336. Additionally, if the Pt surface is subjected to high potentials for a sufficiently long time, more persistent oxide species form, that are only being reduced in the H-upd region. These oxide species are not the subject of this study and are avoided through our experimental protocol. At potentials above 1500 mV, molecular oxygen evolves, marking the start of the OER region. Reversing the scan direction leads to the reduction of the O-upd in one characteristic peak followed by the initial DL region.Fig. 2: Cyclic voltammetry and impedance spectroscopy on Pt(pc).a Different CVs of Pt(pc) and their net surface charge (corrected by the DL capacitance CDL = 100 µF/cm2) MLH and MLO stands for the charge of a hypothetical monolayer of one hydrogen and one oxygen atom per surface platinum atom, respectively. The label “initial” stands for a CV after activating the surface at a low upper vertex potential and before rearrangement to the typical CV of Pt(pc). The different regions of the CV are marked in blue. b, c CV and AC voltammogram from the imaginary/capacitive part of the SPEIS spectra of Pt(pc) in 0.1 M HClO4 and 0.1 M H2SO4. The blue/gray in (b) area represents the possible *OH charge blocked by bisulfate in (c).The intrinsic ORR kineticsPresently, the most common kinetic description (Eq. 1a) of the ORR on platinum is derived from the BV equation37 that simplifies to a Tafel equation at sufficiently high overpotentials \(\left|E-{E}_{{\rm{eq}}}\right|/b \, \gg \, 1\). The contribution of the Tafel equation can be written as the kinetic current, \({i}_{\text{kin}}\). \(b\) is the Tafel slope, \(E\) is the electrode potential, \({E}_{{\rm{eq}}}\) is the equilibrium potential of the ORR/OER on the RHE scale (Eq. 1b) with \({E}^{0}\) as its standard potential, and \({a}_{{O}_{2}}\) as the oxygen activity. Blocking and electronic effects of adsorbents are included by additional terms, \({T}_{\text{Blocking}}\) and \({T}_{\text{Electronic}}\), respectively. The blocking effect here refers to physically blocking a reaction site by an observer species, while the electronic effect refers to a change in the binding energy of the reaction site by a nearby adsorbed observer species.$${i}_{\text{ORR}}\left(E\right)=\underbrace{{\left(1-\gamma \frac{Q\left(E\right)}{{Q}_{0}}\right)}^{n}}_{{T}_{\text{Blocking}}}\cdot \underbrace{\exp \left(\frac{\omega (E)}{{RT}}\cdot \frac{Q\left(E\right)}{{Q}_{0}}\right)}_{{T}_{\text{Electronic}}}\cdot \underbrace{{i}_{0}\cdot {10}^{\frac{{E}_{{\rm{eq}}}-E}{b}}}_{{i}_{\text{kin}}\left(E\right)}$$
(1a)
$${E}_{{\rm{eq}}}={E}^{0}+\underbrace{\frac{{RT}}{4F}\mathrm{ln}\left({a}_{{O}_{2}}\right)}_{\approx 0}$$
(1b)
\({i}_{{ORR}}\) is the apparent, but background and mass transport corrected ORR activity. \(Q\) is the surface charge of specifically adsorbing species, \({Q}_{0}\) is its maximum value. \(\gamma\) and \(n\) are empirical constants loosely connected to the number of sites blocked and the number of sites involved in the rds, respectively. \(\omega\) is an empirical energy interaction parameter of the adsorbent and \({i}_{0}\) is the Tafel exchange current density. The value of the Tafel slope \(b\) is understood here as the value that would be observed in the absence of blocking and electronic effects from adsorbents. Commonly, only the anodic scan is used to determine the ORR activity at the chosen benchmark potential without explicitly accounting for either the blocking or electronic effect. These effects contribute to the apparent Tafel slope, which then would lead to the prevalent curved Tafel plot with a varying slope between 60 and 120 mV/decade exemplified in Fig. 1. By increasing the upper vertex potential, \({E}_{{uVtx}}\), more O-upd charge is passed. When the scan direction is reversed, this charge is not immediately recovered (Fig. 3a). Hence, the O-upd charge and ORR current observed at a particular potential in the cathodic scan depend on the chosen \({E}_{{uVtx}}\), see Fig. 3b, c. Subsequently, Eq. 1a can be linearized (\(y\left(x\right)=a+{mx}\)) yielding a test for the blocking effect (Eq. 2a) and for the electronic effect (Eq. 2b) by omitting the alternative effect, respectively. \({C}_{{DL}}\) is the double-layer capacitance measured at 450 mV and \(v\) is the scan rate:$${\left.\underbrace{{i}_{\text{ORR}}^{\tfrac{1}{n}}(E,\, {E}_{{\rm{uVtx}}})}_{y(x)}=\underbrace{{i}_{\text{kin}}^{\tfrac{1}{n}}(E)}_{a}-\underbrace{\gamma \frac{{i}_{\text{kin}}^{\tfrac{1}{n}}(E)}{{Q}_{0}}}_{-m}\cdot \underbrace{Q\left(E,\, {E}_{{\rm{uVtx}}}\right)}_{x}\right|}_{E={\rm{const}}.}$$
(2a)
$${\left.\underbrace{{\mathrm{ln}}\left({i}_{\rm{ORR}}(E,\,{E}_{{\rm{uVtx}}})\right)}_{y(x)}=\underbrace{\mathrm{ln}\left({i}_{\text{kin}}(E)\right)}_{a}+\underbrace{\frac{\omega (E)}{{RT}{Q}_{0}}}_{m}\cdot \underbrace{Q\left(E,\,{E}_{{\rm{uVtx}}}\right)}_{x}\right|}_{E={\rm{const}}.}$$
(2b)
$$Q\left(E,\,{E}_{{\rm{uVtx}}}\right)=\underbrace{\frac{1}{v}{\int }_{450{\rm{mV}}}^{{E}_{{\rm{uVtx}}}}{i}_{\text{Ar},\text{anodic}}{dE}}_{{Q}_{\text{anodic}}}- \underbrace{\frac{1}{v}{\int }_{{E}_{{\rm{uVtx}}}}^{E}{i}_{\text{Ar},\text{anodic}}{dE}}_{{Q}_{\text{cathodic}}}\pm v\cdot {C}_{{\rm{DL}}}\cdot \left(E-450 \,{\rm{mV}}\right)$$
(2c)
Fig. 3: Effect of O-upd on the cyclic voltammograms and the ORR current.a CV in Argon with increasing upper vertex potentials. b O-upd charge of calculated from the CVs in (a). c Linear sweep voltammograms in oxygen (background corrected) with different upper vertex potentials and scan directions. d Linearization of Eq. 2b for the blocking effect with \(({{n}}=1)\). e Linearization of Eq. (2c) for the electronic effect. f The same conditions as (e), but the potential is held at 1200 mV for a set time. The colors represent the measurement potential, while the gray inset potentials display the respective upper vertex potential (e) or hold time (f). The lines in (b) and (c) highlight the effect of different upper vertex potentials on the O-upd charge or ORR current, respectively, at a particular potential.This linearization requires that the O-upd charge is related to the surface coverage \(\theta\) via \(\theta=Q/{Q}_{0}\). \(Q\) is determined through the integration of pseudocapacitive current, \({i}_{{Ar}}\), in the anodic and cathodic direction (Eq. 2c). In both, the blocking and the electronic effect, \({Q}_{0}\) is part of the slope of the linearized form and is incorporated into the empirical interaction parameter \(m\) (Eq. 2a, b). Wang et al.19 found a value of \(n=1\) or \(n=2\) for the blocking effect of *OH on Pt(111), providing reasonable guesses for \(n\). In the case of Pt(pc), Fig. 3d demonstrates that for \(n=1\) (Case for \(n=2\), Fig. S1e) the blocking effect of the O-upd cannot be linearized according to Eq. (2a). In contrast, the linearization of Eq. (2b) as displayed in Fig. 3e is successful with (R2 > 0.99), hence, on Pt(pc), the O-upd acts on the ORR through an electronic effect rather than via a site-blocking mechanism. Extrapolating \({i}_{{ORR}}\) to \(Q=\,\)0 µC/cm2 in Fig. 3e yields a Tafel plot of the ORR kinetics (Fig. 4a) that is free of the effects of the O-upd over a potential range of about 250 mV. We also tested this technique by varying the amount of the O-upd charge by holding the potential at the upper vertex potential for a set of time durations and found the same relationship (Figs. 3f, S1e). This suggests that the O-upd charge formed potentiodynamically or potentiostatically (at a sufficiently small timescale) has the same origin. Additionally, the slope of the linearization contains information on the interaction of the species responsible for the O-upd effect on the ORR. Within this analysis, a site-blocking effect is only possible when setting \(n\) to an unreasonably high value (>5) or as a minor contribution dwarfed by the electronic effect. These results are not in contradiction with the existence of *OH on the Pt(pc) surface: First, *OH adsorption on Pt(pc) appears to be a minor contribution as suggested by the AC voltammograms in Fig. 2. Secondly, in the potential range of near maximum *OH coverage, its blocking effect manifests itself as a constant factor. Due to its reversibility, its contribution cannot be quantified directly on Pt(pc) and is therefore considered part of the value of the intrinsic exchange current density itself.Fig. 4: The electronic effect of the O-upd on Pt.a Tafel plots of polycrystalline platinum in HClO4 at 50 mV/s. The upper vertex potential of the cathodic scan is 1200 mV. b Tafel parameters for the O-upd free ORR activity at different scan rates in 0.1 M HClO4 and 0.1 M H2SO4. c The interaction parameter \(m\) at different scan rates in 0.1 M HClO4 and 0.1 M H2SO4 with uncertainties stemming from the linear regressions.Both from the cathodic and the anodic branches, we can perform a Tafel analysis using the definition in Eq. (3).$${i}_{{\rm{kin}},{\rm{cathodic}}}\left(E\right)={i}_{{\rm{O}}}\cdot {\left(1-\gamma {\theta }_{\text{ads}}\right)}^{n}\cdot \exp \left({m}_{\text{ads}}{\theta }_{\text{ads}}\right)\cdot {10}^{\frac{{E}_{{\rm{eq}}}-E}{b}}={i}_{0}^{*}\cdot {10}^{\frac{{E}_{{\rm{eq}}}-E}{b}}$$
(3)
The apparent exchange current density \({i}_{0}^{*}\) includes the effects of specific adsorbates other than from the O-upd. While the cathodic branch is obtained by extrapolation, the anodic branch is constructed according to Eq. (4).$${i}_{{\rm{kin}},{\rm{cathodic}}}\left(E\right)={i}_{{\rm{O}}}(E)\cdot \exp \left(m(E)\cdot {Q}_{{\rm{anodic}}}(E)\right)$$
(4)
Notably, the Tafel plot in Fig. 4a exhibits a constant slope and is virtually identical in the anodic and cathodic branches. Repeating this analysis at scan rates up to 500 mV/s yields comparable Tafel parameters (Fig. 4b). At scan rates >500 mV/s, the background correction from the pseudocapacitive current contribution on the apparent ORR activity becomes too large, introducing great uncertainty in the analysis. Nearly identical results are obtained in 0.1 M sulfuric acid (Fig. 4b, c, S1f), except for a sharp decrease in the apparent exchange current density, suggesting a high but incomplete surface coverage of bisulfate. This decrease may be a combination of a blocking effect, effects on the electrochemical double layer, and a shift in the point of zero charge. When the electrolyte contained 0.1 M chloride, no ORR current was observed until chloride leaves in the H-upd region (see S1i). The complete and incomplete coverage of the halides and bisulfate, respectively, has been demonstrated on Pt(111) by STM studies38 and the latter also by radiotracer methods39.The obtained Tafel slope of around 120 ± 10 mV/decade is consistent with a one-electron one-step process being the rds. Using the thermodynamic reversible potential of the ORR of \({E}^{0}\) = 1229 mV vs RHE in Eq. 1b, the exchange current density in the absence of specific adsorbates calculates to around 13 ± 4 µA/cm2 and 1 ± 0.7 µA/cm2 in the presence of bisulfate. The uncertainty of the value originates from the extrapolation in Eq. (2b) and averaging over multiple scan rates. The values of the interaction parameter \(m\) agree for different scan rates, displaying that the effect of the O-upd on the ORR is, as expected, independent of the scan rate (Fig. 4c). Interestingly, the impact of the O-upd on the ORR increases at higher overpotentials. Details of the analysis can be found in the supplementary information. The value for the exchange current density agrees well with previous estimates obtained for platinum surfaces (7–25 µA/cm2)19,29. In contrast, Luo and Koper5 confirmed a Tafel slope of 60 mV/dec on Pt(111), attributed to the effect of *OH surface groups11,19. Therefore, the transfer of the kinetics of Pt(111) to that of Pt(pc) and its derivates, such as the ubiquitous Pt (alloy) nanocatalysts, is not straightforward. To the best of our knowledge, no previous analysis of the ORR on Pt surfaces has yielded such consistent kinetic data: The agreements of the Tafel kinetics over such a wide potential range, range of scan rates, and of both the anodic and cathodic branch is unprecedented. We conclude that the ORR indeed follows the rate equation (Eq. 1a) on Pt(pc) when the O-upd charge is accounted for (Eq. 5). To demonstrate the universal applicability of this analysis, we conduct the same analysis on Pt(pc) surfaces with different levels of rearrangement, on the alloys Pt3Co(pc) and Pt3Ni(pc), and on supported Pt nanoparticles.$${i}_{{\rm{ORR}}}\left(E\right)={\left(1-\gamma {\theta }_{{\rm{ads}}}\right)}^{n}\cdot \exp \left(m(E)\cdot {Q}_{{\rm{O}}-{\rm{upd}}}\right)\cdot {i}_{0}\cdot {10}^{\frac{{E}_{{\rm{eq}}}-E}{b}}$$
(5)
Surface rearrangementCommonly, a freshly polished surface of platinum is “activated” by potential cycling until a stable voltammogram is obtained. In each cycle O-upd species are formed and then reduced which leads to dissolution of a small amount of platinum40. The high reduction potential of Pt ions leads to their immediate redeposition on the surface, resulting in surface rearrangement towards the lowest surface energy while at the same time, possible contaminants are oxidized or expelled from the surface. Both processes are functions of the upper vertex potential (Fig. 2a) and to what extend the surface rearrangement originates from a redistribution of facets/reconstruction or the removal of contaminates is not answered in this study. We find that the apparent ORR current decreases slightly with increased surface rearrangement while the Tafel slope, the exchange current density and the O-upd interaction parameter remain unchanged, see Fig. 5c. This suggests that the tendency to form O-upd species is altered in the process. Through our XPS measurements on the Pt crystal surfaces before and after cycling (Fig. S3a, S3b) it appears that our surface is clean to an extend that can be probed by XPS. We find that when the surface is cycled to lower upper vertex potentials, higher apparent ORR activities are observed meaning that the effect of any contaminants on the ORR is neglectable. Hence, we cautiously suggest that the surface rearrangement from a polished surface to a cycled (activated) surface is a form of reconstruction, ultimately altering the binding energy of the O-upd species binding stronger and earlier on an activated surface.Fig. 5: The effect of the O-upd in different ORR environments.a–c Plots of the kinetic parameters and the apparent ORR activity (900 mV, anodic scan, 50 mV/s) averaged (leading to the displayed uncertainties) over multiple samples (as indicated by the number in the graph): a 0.1 M HClO4 vs 0.1 M H2SO4 and Pt(pc) vs Pt/C. b Effect of alloying. c Effect of rearrangement: Notice that the strain is calculated from the bulk lattice constant obtained by x-ray powder diffraction. d XPS results (see Fig. S3a and Fig. S4a) of the investigated surfaces after polishing, a typical measurement, and cycling to 1400 mV for 50 cycles. To not disturb the platinum overlayer, the measurement protocols did not exceed 1000 mV for the measurements in (b). The specific values in the plots (b) can also be found in Table 1.Platinum alloy surfacesPlatinum surfaces exhibit increased ORR activity when alloyed with an appropriate metal, such as nickel or cobalt, to induce negative lateral surface strain. The resulting downshift in the d-band center lowers the binding energy of oxygen species, such as the intermediates in the ORR10,41. We find that on Pt3Ni(pc) and P3Co(pc) surfaces, the H-upd features are slightly depressed compared to Pt(pc), while those of the surface oxygen regions behave virtually identically regardless of surface preparation method (polishing or Ar-sputtering) (Fig. S1g). Moreover, the formation of non-adventitious surface oxygen (M-O, associated with PtxOy), see Fig. 5d, does not change on the platinum alloys compared to pure platinum before and after the measurement. These observations and the high platinum-to-metal ratio (Fig. 5d) of the non-sputtered samples agree with the presence of a protective platinum overlayer on Pt alloys42. Consequently, the ORR appears to follow the same mechanism on all surfaces, as the kinetic analysis finds that the Tafel slope is unchanged. Here, \(\kappa\) is the rate at which the ORR activity increases with the contraction of the Pt-layer. Its value is comparable for the apparent ORR activity and the exchange current density at around \(\kappa\) = 0.15%/decade. The interaction parameter \(m\) becomes more negative at more compressive surface strain (Fig. 5b), implying that the strength of the electronic effect of the O-upd increases with a downshift in the d-band center. Overall, the ORR activity increases observed for Pt3Co and Pt3Ni mainly arise from an increased intrinsic exchange current density to 25 ± 5 and 30 ± 5 µA/cm2, respectively, consistent with literature data (Table 1) and theory9. While the O-upd is unchanged, it seems to interact slightly stronger with the ORR kinetics, reducing some of the ORR activity gains compared to Pt(pc). The stronger O-upd effect could be attributed to the greater proximity of the ORR intermediates and transition states to the sphere of influence of the elusive O-upd species. While the surface strain relationship is conserved for Pt(pc), Pt3Ni(pc), and P3Co(pc), the O-upd effect may account for discrepancies observed in Volcano relationships between experiment and theory, e.g. those discussed by Stephens et al.9.Table. 1 Collection of the kinetic benchmark currents in literature and this work on different Pt surfaces together with kinetic Tafel parameters and interaction parameters of our O-upd analysisSupported platinum nanoparticlesWhile extended Pt(pc) surfaces are the focus of this work, supported Pt nanoparticles (Pt/C) dominate catalyst applications due to their orders of magnitude higher Pt-utilization2. The surface of Pt/C closely resembles that of Pt(pc), as is evident from the CV in Fig. 6. Despite the similarities, the ORR activities on Pt/C are significantly lower than for Pt(pc), both in this work and in the literature (Table 1), known as the particle size effect1,43. Since the O-upd onset and reduction occurs at lower potentials on Pt/C compared to Pt(pc), this phenomenon is attributed to greater coverage of oxygen species on Pt nanoparticles2 in perchloric and sulfuric acid, which is likely linked to stronger binding on Pt nanoparticles than extended surfaces. Our analysis now provides a method to quantify this effect by comparing the O-upd corrected kinetic parameters of the ORR on Pt/C to Pt(pc) (Fig. S1f). The interaction parameter \(m\), the intrinsic Tafel slope, and the exchange current density (only for HClO4) on Pt/C match their respective values on Pt(pc). The apparent exchange current density of Pt/C (for H2SO4) is only half of the value found on Pt(pc). The reason becomes apparent when one notices that bisulfate adsorbs differently on Pt/C than on Pt(pc) (Fig. 6), with a characteristic peak at 600 mV instead of 300 mV. This peak is likely a contribution of the (111) facet of Pt nanoparticles44 since parts of the characteristic adsorption feature of bisulfate on Pt(111) have been shown to persist even after its surface has been subjected to reconstruction45. It is also known that bisulfate adsorbs stronger and reduces the ORR activity on Pt(111) more than on the other low-index planes46. Together, this explains how bisulfate has a stronger effect on Pt/C than on Pt(pc), while when no bisulfate is present, their exchange current densities are nearly equal. Overall, we conjecture that these results translate to Pt alloy nanoparticles in that activity enhancements, as predicted from their respective extended-surface counterparts (Table 1), are not guaranteed and are subject to changes in the properties of the O-upd.Fig. 6: CVs on the extended Pt surface and on supported Pt nanoparticles.Comparison of the CVs of Pt(pc) and Pt/C in perchloric and sulfuric acid under the same test conditions. Note that the electrochemical active surface area for Pt/C is about 8x larger than for Pt(pc).As we have demonstrated, the electronic effect of the O-upd on the ORR is consistent, quantifiable, and universal among polycrystalline platinum surfaces. Our results on Pt/C provide a valuable link to applications of such platinum catalysts in hydrogen fuel cells (PEMFC and AEMFC): The testing and operating modus of these systems are commonly chrono-potentiometric at 600–700 mV, about 200 mV more positive than where the O-upd is fully reduced and may already have appeared in in-situ spectroscopic studies during fuel cell operation47. Under operating conditions at elevated temperatures, the present O-upd species reduce the ORR activity significantly and, therefore, also reduce the overall performance of the fuel cell as has been found previously48,49. Notably, we find a Tafel slope of around 120 mV/dec on Pt/C rather than the commonly observed 60 mV/dec on Pt/C in low temperature PEMFCs2. A contributing factor to this discrepancy could be sulfonic acid groups interacting with the platinum surface50,51, as the adsorption of sulfate and sulfonic acid groups from the Nafion ionomer coincides with the examined potential range of PEMFCs. In contrast to applied fuel cells, the common half-cell RDE protocol aims to minimize the formation of the O-upd by choosing comparably fast anodic scans to determine the ORR activity of a given catalyst. Consequently, the standard protocol does not provide the intrinsic kinetic values of the ORR on platinum, nor does it predict the performance of a given catalyst in a fuel cell, as illustrated in Fig. 1: For fundamental analysis, one needs to obtain the curve of the intrinsic kinetics, while for fuel cell applications one needs to use a steady-state-like curve. However, the common protocol can only deliver ORR activities between these boundary cases. Alternative ways to test ORR catalysts for fuel cells have been developed by multiple workgroups e.g., by utilizing gas diffusion half-cells52,53. However, experimental challenges limit the wider adoption of these techniques, and the common half-cell RDE protocol remains the measurement tool of choice in developing new catalysts and fundamental studies. We suggest expanding the common protocol to incorporate the effects of the O-upd so that more accurate, comparable, and robust kinetic values are obtained for fundamental studies or when developing new catalysts54,55. Additionally, instead of tweaking the ORR intermediate binding energies to increase the intrinsic ORR activity, reducing the susceptibility towards the formation of surface oxygen may be a more fruitful pathway to better ORR catalysts.Weakening of the ORR/OER intermediate binding energies by O-updTo investigate the electronic effect of surface oxygen, we use DFT to calculate the binding energies of reaction intermediates *OO, *OOH, *O, and *OH in various configurations and oxygen coverages on Pt(111) and Pt(100). We calculate the change in binding energy \(\Delta {\mathbb{E}}\) of these intermediates in the presence of the possible *Oupd structures at 25% coverage intervals and find that the presence of *Oupd weakens the binding energy of all ORR intermediates (Fig. 8a).The *Oupd structures considered, mainly consists of surface *O, because subsurface *O is kinetically unstable at low oxygen coverages. This agrees with previous DFT studies, by e.g. Gu and Balbuerna56 which found that on Pt(111) >0.75 ML *O is required before it becomes favorable to add oxygen to the subsurface rather than the surface. Simulations by Hu and co-workers57 have shown that at about 0.5 ML average O* on Pt(111), high local O coverage can pull Pt out of the Pt surface plane forming Pt2O6 units. Motifs with PtOx pulled out from the plane, have been widely considered as part of the place exchange mechanism, which allows *O to be incorporated into the subsurface below PtOx units58,59,60. We have therefore included *Oupd structures with Pt2O6 units and subsurface *O on Pt(111), and find that *Oupd weakens the binding of ORR intermediate regardless of the exact *Oupd structure.Since the weakening of ORR intermediates appears to be a universal phenomenon on Pt(111) and Pt(100), it likely applies to the Pt(pc) surface as well. Notably, the weakening of the *OO and *OOH binding energies is correlated, showing the robustness of their scaling relationship beyond surface strain. Our results are in good agreement with previous works that found the O2 binding energies to vary approximately linearly with *Oupd coverage, with variations linked to changes in the metal d-band center induced by the *Oupd adsorbates61,62. The Bell/Brønsted-Evans-Polanyi (BEP) principle correlates the changes in the change in reaction energy for elementary steps,11,63, which our calculations show depends approximately linearly on *Oupd coverage. For a multistep reaction where the influence of blocking species has been removed, the activation energy will take the form in eq. 6a with \(\alpha\) as the transfer coefficient and \(\beta\) as scaling constant. The ORR current density will therefore depend exponentially on the shifts in intermediate binding energy, the *Oupd coverage, or its surface charge, \({Q}_{O-{upd}}\). The effect of the shift in intermediate binding energy and the interaction parameter \(m\) are then linked through Eq. 6b. If one ML (\({\theta }_{O-{upd}}=1\)) of *Oupd corresponds to a transferred charge of \({Q}_{0}=\,\)420 µC/mol we find that \(m\) linearly correlates with the *Oupd induced shift in binding energies of the reactants and products of the rds, corroborating our experimental observations.$${G}_{{\rm{P}}\to {\rm{R}}}=\alpha \cdot \left({\Delta G}_{{\rm{P}}}-{\Delta G}_{{\rm{R}}}\right)+\beta=\alpha \cdot \left({{{\Delta }}{\mathbb{E}}}_{{\rm{P}}}-{{{\Delta }}{\mathbb{E}}}_{{\rm{R}}}\right)+{\beta }^{{\prime} }$$
(6a)
$${i}_{0}\exp \left(-m\cdot {Q}_{\text{O}-\text{upd}}\right)={i}_{0}\exp \left(-\alpha \frac{\left({{{\Delta }}{\mathbb{E}}}_{{\rm{rds}},{\rm{P}}}-{{{\Delta }}{\mathbb{E}}}_{{\rm{rds}},{\rm{R}}}\right)}{{RT}}\right)$$
(6b)
$$\frac{d{{{\Delta }}{\mathbb{E}}}_{{\rm{rds}},{\rm{P}}\to {\rm{R}}}}{d{\theta }_{\text{O}-\text{upd}}}=-\,\frac{1}{\alpha }m{Q}_{0}{RT}\, \approx \, 0.21\,\frac{{\rm{eV}}}{{\rm{ML}}}$$
(6c)
Now, we can estimate what change in binding energy of *OO or *OOH would reflect the experimentally determined energy interaction parameter (\(m \, \approx \, 10\) cm2/mC). Assuming α = 0.5, results in 0.21 eV/ML which is somewhat lower than from our DFT calculations (0.8 – 1 eV/ML), which implies the transition state for O2 activation depends weakly on O-upd coverage. However, other effects may play important roles in the electronic effect on the ORR47,59,60,64,65, such as changes in intermediate coverages affecting available free sites, interactions between the *Oupd and the liquid water layer or hydronium ions; requiring further investigation. The experimentally determined interaction parameter can consequently assist in the identification of such effects. Interestingly, the O-upd charge is stabilized by the presence of O-upd species, apparent from the delayed onset (up to 50 mV) of the reduction peak in Fig. 7a, while the binding energy of the supposed *Oupd itself is weakened. Overall, the observed behavior on Pt, likely applies to other catalyst systems, as linear adsorbate-adsorbate interaction models are prevalent in micro-kinetic modeling and is considered in, e.g., the Temkin and Fowler-Guggenheim adsorption isotherms.Fig. 7: The effect of the O-upd on the OER.a CVs with different upper vertex potentials (1400 mV to 1850 mV, 50 mV steps, color-coded). The O-upd current is calculated from the cathodic charge and used in the background correction of the OER current in (b) and (c). The inset is a zoom-in on the onset of the reduction of the O-upd. b OER currents vs O-upd charge in a similar plot as in Fig. 3e. The green lines represent the slope of different values for the interaction parameter \(m\). c Measurement and a prediction of the OER using ORR kinetic parameters and a reasonably chosen O-upd interaction parameter. The inset displays a Tafel plot of the measured cathodic OER branch.The oxygen evolution reactionThe principles of the presented kinetic analysis of the ORR should also apply to the OER case. First, one must determine the O-upd charge during the OER. At potentials before the onset of the OER, the oxidation/reduction (anodic/cathodic branch) O-upd charges are balanced until 1600 mV (Fig. 7a), suggesting that the entire pseudocapacitive contribution is captured by the O-upd and the double layer charge. During the OER, the oxidation charge includes the electron transfer to oxidize water to molecular oxygen. However, due to the rotation of the electrode at 1600 rpm in Ar-saturation, the molecular oxygen is rapidly removed from the Pt surface. Due to the absence of oxygen during the cathodic scan, the reduction charge must equal the O-upd charge formed in the anodic scan. In connection, the high scan rate of 500 mV/s limits the oxygen production and formation of stronger oxides during the OER, so that the metallic platinum surface remains the same as during our investigation of the ORR. Since the O-upd charge is only reduced at potentials <1100 mV during the cathodic OER scan, the present O-upd charge is only dependent on its upper vertex potential and otherwise constant. This allows collecting the background corrected OER currents at the same potential but different O-upd charges, displayed in Fig. 7b. The resulting plot is of the same kind as the one used to extract the Tafel kinetics for the ORR in Fig. 3e. However, while the graphs are close to being linear with respect to the logarithmic current, they exhibit a slight curvature, meaning that the interaction parameter \(m\) is not constant with O-upd coverage. While in the case of the ORR, the relevant O-upd charge is less than one monolayer equivalent of oxygen (<420 µC/cm2), in the OER case, the O-upd charge is greater than two equivalent monolayers of oxygen. It is reasonable that the effect of the O-upd charge on the OER kinetics is not any longer linear at such high oxygen coverages. The interaction parameter for each potential branch lies between −5 and −10 cm2/mC, close to the one found for the ORR, see Fig. 5a–c. The negative curvature suggests that at increasing O-upd coverages, the marginal effects on the OER kinetics become less negative. Even with a constant interaction parameter, obtaining a Tafel plot for the ORR is not feasible since it would require an extrapolation over 9 times the range of the known values to \({Q}_{O-{upd}}=0\) µC/cm2. However, since the O-upd charge, in contrast to the ORR, is constant in the cathodic scan direction, one can determine the Tafel slope to be around 100 mV/decade (Inset in Fig. 7c). This value is somewhat lower than for the ORR, but sufficiently close to be considered the same within a mechanistic discussion. Setting the interaction parameter to \(m\) = −8 cm2/mC and using the exchange current density obtained for the ORR (\({i}_{0}=\) 13 µA/cm2), we observe a very close match to the measured OER currents, including the characteristic hysteresis between the anodic and cathodic scan (Fig. 7c): During the anodic scan the O-upd coverage increases, causing a reduction in the observed OER compared to its expected value from its intrinsic activity. On the cathodic scan the O-upd remains constant during the OER, meaning that for any given potential below the vertex potential, the O-upd coverage is higher causing the reduction in the apparent OER activity in the cathodic scan.A great concern is the relatively high scan rates chosen in determining the OER currents: Depending on the rate of the studied electrochemical reaction, it may not be at a kinetic steady state as its rate constant may be too small to follow the potential sweep, and one would observe an increase in currents for a given potential when the scan rate is decreased. However, like in the analysis of the ORR, the OER currents decrease when the scan rate is decreased (see Fig. S1h). The decrease stems from the increased formation of the O-upd species, analogue to the ORR case. When this effect is accounted for, the OER currents recorded at 200 mV/s and 500 mV/s match, demonstrating that the OER, even at these relatively high scan rates, is still kinetically in steady state. All things considered, first, the nature of the potentiodynamically formed O-upd charge appears to be the same in the potential range from 600 mV to 2000 mV indicating that it is not a trivial intermediate of the OER. Secondly, due to the absence of noticeable electrochemical features past 900 mV (associated with <0.5 ML *O), there is no distinct upper bound of *O on the Pt surface, meaning that increasing positive potentials overcome associated energetic penalties. At lower scan rates or during potentiostatic operation, this is likely not the case anymore (Fig. S1e) and other, stronger oxide species form. Before the onset of the OER, bisulfate was desorbed39, and the results are virtually identical in sulfuric and perchloric acid electrolytes (Fig. S1i). Consequently, for Pt(pc) surfaces, the ORR and OER kinetics can phenomenologically be described by a single rate equation (Eq. 7). Such a description suggests that the electronic effect of the O-upd on the transition state of the rds of both the ORR and OER is the same, hence both reactions likely share the same transition state and are opposite directions of the same reaction path.$${i}_{{\rm{ORR}}/{\rm{OER}}}\left(E\right)={i}_{0}^{*}\cdot \exp \left(m\cdot {Q}_{{\rm{O}}-{\rm{upd}}}\right)\cdot \left({10}^{\frac{E-{E}_{{\rm{eq}}}}{b}}-{10}^{\frac{{E}_{{\rm{eq}}}-E}{b}}\right)$$
(7)
When no specifically adsorbing species are present on the surface, \({i}_{0}^{*}\) equals \({i}_{0}\). An overview of the unified ORR/OER kinetics is displayed in Fig. 8a. The H-upd coverage was extended into the HER region using potential step experiments, a method adopted from66 to show the similarities between the HER/HOR and H-upd region with the ORR/OER and O-upd region. During the absence of O-upd current in the cathodic scan direction and corrected by the double layer capacitance, the scans cross the potential axis at around 1229 mV. This supports that the thermodynamic reversible potential of the ORR/OER is the relevant value in their kinetic description rather than the measured open circuit potentials (900–1000 mV). While our results are conclusive on various kinds of platinum surfaces, it is unclear if they can be transferred to other relevant metallic or oxide surfaces, such as that of iridium oxide67. Figure 8c depicts an scheme summarizing the relevant species on a Pt(111) slab: The *O of the O-upd would occupy the high coordinated hollow sites (bridge sites on Pt(100)) or subsurface vacancies and can be seen as a spectator. The intermediate *O of the reaction pathway and the specific adsorbates68,69 responsible for a blocking effect would exist on the more exposed, low coordinated, top sites surrounded by a water adsorption layer. Therefore, species such as *O would then fulfill different roles, depending on which binding site they are found. However, despite tremendous efforts by the scientific community, it is not certain how *OH and *O manifest themselves and lead to their ascribed electrochemical features. While the efforts to complete the picture on the atomic level continue, e.g., by ins-situ Raman spectroscopy34,70,71 and molecular dynamic simulations64, the study of the electrochemical effects of the elusive O-upd species can be carried out independently, as demonstrated in this contribution.Fig. 8: A unified picture of the O-upd with the ORR and OER.a Change in the binding energies \(\triangle {\mathbb{E}}\) of ORR intermediates at different *O configurations. Displayed for each *O configuration is the binding energy for the most favorable intermediate position, including structures with the presence of subsurface oxygen. b The unified kinetics of the ORR/OER on Pt(pc) together with the HOR/HER, the H-upd, and the O-upd currents. The HER/HOR and ORR/OER branches have been measured individually to avoid residual O-upd from high upper vertex potentials. All currents are corrected by the double-layer capacitance. c Scheme of the relevant species of the ORR/OER on a Pt(111) slap.In this work, we presented an experimental method to extract the intrinsic ORR kinetics on polycrystalline platinum surfaces in acidic media, irrespective of scan rate and scan direction. For this, we distinguished between stronger oxides that are not the subject of this study, and weaker oxygen species, called the O-upd. We could successfully separate the effects of the O-upd from the kinetics and found that the ORR follows the Tafel kinetics handsomely. We measured a Tafel slope of about 120 mV/decade, suggesting that one electron is transferred during the rate-determining step in agreement with mechanistic models with an exchange current density determined in the Tafel region of 13 ± 4 µA/cm2, a substantial step towards a mechanistic description of the ORR/OER. We show that our method delivers consistent results on both less well and highly rearranged polycrystalline platinum, on platinum alloys, in the presence of bisulfate, and on supported Pt nanoparticles. The negative effect of the O-upd seems to stem from weakening the binding energies of the ORR/OER intermediates, altering potential kinetic and thermodynamic barriers for O2 activation. Our experimental findings are in good agreement with density functional theory calculations showing that surface and subsurface oxygen atoms weaken binding of the ORR/OER intermediates. The ORR activity on Pt(pc) at 900 mV is ~7 mA/cm2 when accounting for the O-upd effect, over 3-fold higher than the commonly reported 2 mA/cm2. This effect may explain some of the discrepancy between expectations and experiments regarding the Volcano relationship of ORR catalysts. Due to the difference between the Tafel slope on Pt(111) and on Pt(pc), our analysis substantiates limits in transferring the ORR kinetics on model surfaces to applied platinum-based catalysts. With these results, we argue that the RDE half-cell protocol must be expanded to determine the intrinsic ORR kinetics for effective screening and development of Pt-based catalysts and for providing more accurate kinetic data to inform computational methods that are commonly performed on metallic rather than oxidized surfaces. Using an analogous analysis for the OER reveals that the ORR/OER can be described by the same rate equation with nearly identical parameters and the same negative effect of the O-upd, suggesting a deeper mechanistic connection. Finally, we propose that *O has different roles depending on the binding site where it occurs. We conjecture that high-coordinated, strong binding sites are responsible for the O-upd, low-coordinated, weak binding sites are responsible for the kinetic activity. Since the reduction of the O-upd species during a cathodic scan is the main driving force behind catalyst degradation, a reduction in the O-upd formation and its interaction with reaction intermediates holds the key to both improved ORR kinetics and catalyst stability. Arguably, an ORR catalyst with a sufficiently reduced O-upd formation could be more stable and more catalytically active despite an intrinsically lower exchange current density.
Unifying the ORR and OER with surface oxygen and extracting their intrinsic activities on platinum
