Electrochemical determination of the degree of atomic surface roughness in Pt–Ni alloy nanocatalysts for oxygen reduction reaction

Abstract

Pt–Ni alloy nanocrystals with Pt‐enriched shells were prepared by selective etching of surface Ni using sulfuric acid and hydroquinone. The changes in the electronic and geometric structure of the alloy nanoparticles at the surface were elucidated from the electrochemical surface area, the potential of zero total charge (PZTC), and relative surface roughness, which were determined from CO‐ and CO₂‐displacement experiments before and after 3000 potential cycles under oxygen reduction reaction conditions. While the highest activity and durability were achieved in hydroquinone‐treated Pt–Ni, sulfuric acid‐treated one showed the lower activity and durability despite its higher surface Pt concentration and alloying level. Both PZTC and $Q_{{\mathrm{CO}}_2}$/Q_CO ratio (desorption charge of reductively adsorbed CO₂ normalized by CO_ad‐stripping charge) depend on surface roughness. In particular, $Q_{{\mathrm{CO}}_2}$/Q_CO ratio change better reflects the roughness on an atomic scale, and PZTC is also affected by the electronic modification of Pt atoms in surface layers. In this study, a comparative study is presented to find a relationship between surface structure and electrochemical properties, which reveals that surface roughness plays a critical role to improve the electrochemical performance of Pt–Ni alloy catalysts with Pt‐rich surfaces.

INTRODUCTION

In recent years, it has been demonstrated that alloying Pt with a transition metal M (e.g., Fe, Co, Ni, and Cu) decreases the strength of bonding between Pt and hydroxyl group, ^1,2 by downshifting the electron energy of the d-band center of surface Pt. ^2-4 As a result, the activity for the oxygen reduction reaction (ORR) increases. In addition, these Pt–M alloys with Pt-enriched shell have shown further enhanced stability and activity, resulted from the stabilization of their electronic structures by removing surface M atoms. ^5,6 The formation of surface Pt-enrichment can be simply achieved by chemical ^7,8 and electrochemical methods. ^9,10

The bimetallic Pt skin surface has unique properties which favor the enhancements in specific activity and durability during ORR, such as no charge transfer from surface Pt to surface M, ¹¹ low reactivity toward oxygen, ^12,13 and weak bond strength of oxygen, ^14,15 in comparison to Pt–M alloy surface. Furthermore, Pt-enriched surface increases the resistance against Pt dissolution by shifting the dissolution potential toward higher potentials, leading to an increase of thermodynamic and kinetic stabilities of metal nanoparticles (NPs), especially under severe corrosive conditions for ORR. ¹⁶ Finally, unalloyed surface M, which is mainly produced from oxide formation due to its high oxophilicity, ¹⁷ can bring a significant gap between practical and theoretical electron energy levels of bimetallic NPs. The surface enrichment of Pt atoms is expected to prevent the formation of new compounds.

Another critical factor to affect the stability of surface Pt atoms is surface defect density. Specifically, electrocatalytic activity and durability can vary significantly depending on surface defects. ^18,19 To date, however, the effects of defects of Pt-enriched surface on electrochemical properties including ORR activity and durability have been rarely studied except on oxidation behaviors of CO_ad on electrode films ^20-22 or extended X-ray absorption fine structure (EXAFS) analysis of average coordination number (CN) of surface Pt atoms, ²³ because quantitative analysis of surface defects in small nanocrystals (∼3 nm) is particularly challenging. Therefore, there is a strong need for electrochemical characterizations that can determine at least a relative change in the atomic surface roughness.

Here we report a study on the relationship between electrochemical properties and several factors related to surface structures in Pt₃Ni/Pt_3+x Ni_1−x core/shell nanocrystals, which were prepared using two different chemical solutions (sulfuric acid aqueous solution ²⁴ or hydroquinone dissolved in ethanol ²⁵ ) to leach out Ni at the surface. From the changes in electrochemical and structural properties of Pt-enriched Pt₃Ni, and commercial Pt occurring during 3000 potential cycles in oxygen-saturated acidic aqueous solution, we first suggested that charge ratio for the oxidation of CO ²⁶ and reduced CO₂ (“CO₂”) adlayers ²⁷ can be one of the key indicators to describe surface structures as well as ORR activity at pristine and degraded states. This will be expected to be employed with other catalysts as well, and it will also provide new insight into the methods to design robust and active ORR catalysts.

EXPERIMENTAL Synthetic procedures

Pt₃Ni/C (40 wt% of metal loading) was prepared from PtCl₄ and NiCl₂ · 6H₂O by reduction with sodium borohydride (NaBH₄) in a solution containing anhydrous ethanol (180 mL), sodium acetate (C₂H₃O₂Na, 99.995%), and carbon black (Vulcan XC 72R). 0.15 g of carbon black was dispersed in anhydrous ethanol containing sodium acetate (17.56 mmol). Pt and Ni precursors were added into the mixture and stirred for 3 h. After sonication for 5 min, anhydrous ethanol of 20 mL containing NaBH₄ (2.9 mmol) was quickly added with vigorous stirring. The resultant solution was further stirred for 2 h, followed by filtration, washing with ethanal, and drying in a vacuum oven at 60°C for 12 h. The obtained powder was redispersed in ethanol and sonicated for 5 min and then filtered. The mixture solution was heated at 75°C for 1 h, and then filtered, washed, and dried in a vacuum oven at 40°C for 12 h. This additional washing process was repeated three times and the product was collected. To leach Ni out at the surface, the pristine Pt₃Ni/C (0.1 g; denoted as PR) was dispersed in two different solutions, 0.5 M sulfuric acid (95%–98%) aqueous solution (denoted as SA) and 0.17 M 1,4-benzohydroquinone (C₆H₆O₂, 99%) dissolved in anhydrous ethanol (denoted as HQ). The mixture solutions were heated at 80°C for acid treatment, and 75°C for hydroquinone treatment, for 2 h in a flow of N₂ gas under stirring. Then, they were cooled down to room temperature and washed with deionized water and ethanol, respectively. Catalysts in two solutions were filtrated and dried at 60°C for 12 h in a vacuum oven. The additional washing process was repeated after the Ni leaching process.

Electrochemistry

Catalyst inks were prepared by mixing 0.01 g of Pt–Ni/C with 20 µL of distilled water, 57.2 µL of 5 wt% Nafion solution (Aldrich Chemical Co) as a binding material, and 800 µL of 2-propanol. The ink was ultrasonicated, and then, the suspension was pipetted onto a glassy carbon disk of a rotating disk electrode (0.196 cm² geometric surface area; Pine Instrument Co), resulting in a metal loading of 85 µg_Pt–Ni/cm². The dried electrode was then transferred to the electrochemical cell containing argon-saturated 0.1 M HClO₄ (Aldrich). For the ORR, the oxygen was purged into 0.1 M HClO₄ solution, and the cell temperature was kept at 293 K. Polarization curves for the ORR of commercial Pt/C and Pt–Ni samples were measured at a scan rate of 5 mV/s before and after 3000 potential cycles. The stability test was conducted by subjecting the catalysts to the continuous potential cycling between 0.632 and 1.132 V (reversible hydrogen electrode; RHE) at 50 mV/s in an O₂-saturated 0.1 M HClO₄ solution. All current densities were normalized to the geometric area of the glassy carbon disk.

The potential of zero total charge (PZTC) was calculated with CO-displacement charge at a constant potential, at which CO molecules cover the surface by replacing preadsorbed species such as protons and anions. Briefly, CO was introduced into the Ar-purged electrolyte at 0.108 V versus RHE. The displacement current was recorded as functions of time and integrated. The resulting charge was subtracted from the H_upd charge that was integrated from the CO-dosing potential. CO₂ was reductively adsorbed by introducing CO₂ gas into the Ar-purged electrolyte at 0.088 V for 20 min, followed by admitting Ar into the inlet of the line for 20 min. The subsequent anodic potential sweep generated an oxidation peak at about 0.67 V.

All electrochemical measurements were conducted in standard three-compartment electrochemical cells, consisting of glassy carbon electrodes, Pt wires, and saturated calomel electrodes as working, counter, and reference electrodes, respectively. All potentials in this study are referred with respect to RHE.

Characterization

The metal loadings of PT, PR, HQ, and SA were obtained by measuring C, H, N, and S with an elemental analyzer (Elementar). Energy dispersive-X-ray fluorescence (ED-XRF; SEA1200VX; Seiko) was conducted to measure the atomic ratios of the Pt–Ni alloy catalysts. The particle size and shape were confirmed by transmission electron microscopy (TEM; Technai F20 at 200 kV). The crystal structure of the prepared bimetallic NPs was analyzed by powder X-ray diffraction (XRD) patterns obtained at the 3D XRS beamline of the Pohang Accelerator Laboratory (PAL). By combining the (220) peak position and the atomic ratio of Pt to Ni, which was confirmed by XRD and ED-XRF, respectively, the alloying level of the Pt–Ni alloy NPs was calculated by applying Vegard's law. X-ray photoelectron spectroscopy (XPS; Al Kα; Sigma Probe, VG Scientifics) was also conducted to identify atomic fractions of various electronic states of both Pt and Ni.

Pt L₃-edge EXAFS spectra were collected in transmission mode at 10C beamline of PAL. Background subtraction and normalization with respect to the main edge jump were performed using the ATHENA software package. ²⁸ After being k ³-weighted, the EXAFS function was Fourier transformed from k-space to r-space to generate the χ(R) versus R (or FT-EXAFS) spectra in terms of the real distance from the central absorbing atom. To simulate the FT-EXAFS data of Pt foil, theoretical Pt structure (space group-Fm3m, a = b = c = 3.9200 Å, α = β = γ = 90°) was generated using the ARTEMIS (ATOM and FEFF codes) software package. ²⁹ The data range taken for the transformation was 3–10 Å⁻¹ in the k-space. Structural parameters were obtained, without phase corrections, by fitting the data in the R-space within the interval of 1–5 Å. According to the above analysis and results, for the Pt L₃-edge, a single shell (Pt–Pt) model was chosen for the Pt foil, two shells (Pt–O and Pt–Pt) model was chosen for pure Pt NPs and a three shell model (Pt–O, Pt–Ni, and Pt–Pt) was chosen for the Pt₃Ni samples treated with sulfuric acid or hydroquinone.

RESULTS AND DISCUSSION

Figure 1 shows TEM images and size distribution histograms of commercial Pt/C (PT), pristine Pt₃Ni without chemical treatment (PR), sulfuric acid-treated (SA), and hydroquinone-treated Pt₃Ni (HQ). All the samples have similar metal loadings (40 wt% of the expected nominal value), as proved by elemental analysis in Table S1. The mean particle sizes of PT, PR, HQ, and SA, which were measured from 100 selected particles with a spherical shape, are 3.3 ± 1.1, 3.6 ± 0.9, 3.5 ± 0.8, and 3.4 ± 0.9 nm, respectively. All the samples, including PT, have little difference in particle size within the deviation range. Powder XRD and ED-XRF data were collected and compared to determine the structural and chemical modifications (Figure S1 and Table S2). The most noticeable difference between PT and the Pt–Ni catalysts is the decrease in the lattice parameters of the Pt–Ni catalysts by alloying of Pt and Ni. The atomic ratios of Pt and Ni from ED-XRF and the (220) peak positions from XRD patterns are listed in Table S2. In addition, degrees of alloying (doa, %) were also presented in Table S2, which were calculated with XRD and ED-XRF results. In SA, most of the remaining Ni are alloyed with Pt and contribute to the diffraction patterns as indicated by its high doa value (92.4% in Table S2), while the significantly lower doa values for PR (70.6%) and HQ (75.4%) imply that PR and HQ have higher ratios of the surface to core Ni atoms.

Figure 1. Transmission electron microscopy images and particle size distribution histograms of (A and E) PT, (B and F) PR, (C and G) HQ, and (D and H) SA, respectively

To probe the change in the oxidation level, atomic fractions of metallic phases (Pt(0) and Ni(0)) were calculated by fitting Ni 2p and Pt 4f core-level X-ray photoelectron spectra (Figure 2). The atomic ratio of Pt increased in the order of PR < HQ < SA, which is fully consistent with the ratios determined by ED-XRF in Table S2. It should be noted here that the SA catalyst showed the highest values in both Ni(0) and Pt(0), as shown in Figures 2C and 2D, respectively. The Pt(0) fractions of both SA and HQ were significantly higher than that of PT (see Figure S2 for PT). The trend of both Pt(0) and Ni(0) in PR, HQ, and SA enabled us to conclude that Ni dissolution by the chemical treatments (HQ and SA) causes the increase in the metallic states of surface metals.

Figure 2. (A) Ni 2p3/2 and Pt 4f core-level X-ray photoelectron spectroscopy spectra. (B) Atomic ratios (%) of Pt to Ni, (C) Pt(0) and Pt oxides, and (D) Ni(0) and Ni oxides

Pt L₃- and Ni K-edge absorption spectra were also obtained (Figure S3). The absorption intensities around Pt L₃ and Ni K edges were calculated with the obtained spectra. Because the absorption area of Pt L₃ edge increases with the number of holes in the Pt 5d band, the relative change in the oxidation state can be seen from the comparison of the white-line intensities. The white-line area decreases in the order of PR > HQ > SA > PT > Pt foil, which is well consistent with Pt(0) phase ratio obtained from XPS analysis in Figure 3C except for the fact that the ratio of metallic Pt for PT is higher than PR but lower than both HQ and SA, indicating that amount of the absorption decreases with increasing the atomic ratio of Pt. In the case of Ni K edge, the absorption features at Ni K edge arise from electron transition from 1s to valence shells, higher energy bound states, or eventually the continuum. The lowest energy transition of 1s to 3d at the pre-edge region around 8333 eV, which is a forbidden transition of 1s to 3d bands, is always prominent for direct Ni–Ni bonds in metallic Ni. At the pre-edge energy region, SA showed the highest absorption compared with HQ and PR. Therefore, the intensity of the white-line, which is plotted in Figure S3C, shows exactly the same trend with that at Pt L₃ edge, meaning both Pt and Ni in SA have the higher metallicities of both Pt and Ni, compared with those of HQ as well as PR.

Figure 3. Hydrodynamic voltammograms of PT, PR, HQ, and SA before and after 3000 potential cycles using rotating disk (glassy carbon) electrode in O2-saturated 0.1 M HClO4. The rotation rate was 1600 rpm for all catalysts

Polarization curves of PT, PR, HQ, and SA for the ORR, with support of carbon black, were measured, before and after potential cycling (Figure 3). As expected, both SA (122 A/g_Pt for mass_Pt-specific activity at 0.9 V) and HQ (224 A/g_Pt) exhibited much higher ORR activities than those of PT (82 A/g_Pt) and PR (38 A/g_Pt), which confirms that Pt-enriched surface is beneficial for the ORR. After 3000 potential cycles, all the catalysts showed the decreased mass_Pt-specific activity in the following order: HQ (127 A/g_Pt) > SA (39 A/g_Pt) > PT (23 A/g_Pt) > PR (16 A/g_Pt). In particular, HQ showed the highest ORR activities, both before and after 3000 potential cycles, and more specifically, 1.8 and 2.0 times higher mass_Pt-specific activities than those of SA, respectively. In our previous study, ²⁵ it was found that the high activity and durability of HQ are attributed to the increase of the average size of Pt skin domain in the alloy surface layer. Although the origin of the difference in the catalytic performance of HQ and SA were identified in terms of the size and the roughness of the Pt skin domain, the increased surface roughness for SA has not been verified. In this study, therefore, we focused on the new electrochemical measurement of relative change in surface roughness as an important factor in affecting the electrocatalytic activity and the durability of Pt–Ni alloy surfaces. The new electrochemical analysis quantifies surface roughness at an atomic scale and enables us to understand the reason why SA has poor electrochemical activity and durability.

The current for CO adlayer (CO_ad) oxidation was detected in CO-displacement experiments, and it was analyzed to estimate the PZTC ^30,31 and electrochemical surface area (ECSA) of PT, PR, SA, and HQ before and after the potential cycling. CO_ad was formed on the catalyst surface by holding the potential at 0.108 V for 10 min, then the 0.1 M HClO₄ (aq.) electrolyte was purged with Ar gas for 10 min. The oxidation currents were recorded, then normalized to Pt loading (Figure 4A,B). Since CO_ad oxidation on Pt-based alloy surface follows the kinetics of an electrochemical Langmuir–Hinshelwood reaction (i.e., water activation and subsequent CO₂ formation as a result of diffusion of CO_ad to OH_ad), ^31,32 the CO_ad oxidation peak potential strongly depends on water activation and molecular mobility of CO_ad. Therefore, with the voltammograms only, it is exceedingly limited to determine which surface has a lower water activation or a higher CO-metal bond strength. ^33-35

Figure 4. The fifth forward scans of cyclic voltammograms in Ar-saturated electrolyte in Hupd potential region and CO adlayer oxidation currents, (A) after and (B) before the potential cycling. (C) Electrochemical surface area (ECSA) of PT, PR, HQ, and SA before and after 3000 potential cycles. ECSA was calculated from the CO adlayer oxidation current. (D) Potential of zero total charge (PZTC) of PT, PR, HQ, and SA before and after 3000 potential cycles. PZTC was calculated from the Hupd and COad oxidation charges

ECSA normalized with Pt loading was calculated from CO_ad oxidation charge (Figure 4C). Before cycling, SA (48.6 m²/g) and HQ (36.6 m²/g) had the highest and lowest ECSA values, respectively. After 3000 potential cycles, SA still exhibited the highest ECSA (32.1 m²/g), but the ECSA value for HQ became the second-highest (27.9 m²/g), even though it is slightly higher than those of PT (27.4 m²/g) and PR (27.3 m²/g). In particular, HQ had the lowest loss (31%, relative to the initial value) in ECSA during 3000 cycles, suggesting that HQ has the most durable surface owing to the formation of a bulk alloy of surface Pt and underlying Ni atoms. ¹⁶ PZTC is an important value to understand electronic and geometric modifications from interfacial charge transfer in specific adsorption for compensating potential-induced excess, or deficit surface charge that is proportional to the gap between an applied potential and surface electron energy level. ^7,36 Therefore, the shift of PZTC can indicate the relative change of electrode work function, and also provide information to understand the surface structural change.

Figure 4D presents PZTC values before and after potential cycling. HQ showed the highest PZTC (0.291 V), even compared with those of PR (0.281 V) and PT (0.276 V). After the potential cycling, PZTC of HQ decreased to 0.274 V (94% of its initial value), but still higher than that of the degraded PT (0.270 V). PZTC of SA increased from 0.238 to 0.267 V after potential cycling. It is worth noting that PZTC of PR is the highest among the four samples, although PR exhibited the worst ORR rates. One plausible explanation is that, during the durability test, surface Pt in PR was less corrugated owing to the sacrificial dissolution of Ni. Similarly, SA has a high surface defect density by acid treatment initially, but a large part of the surface defects might be removed after the potential cycling through repetitive surface oxidation and reduction of Pt.

To estimate relative change in surface roughness of the bimetallic surface, the oxidation charge of reductively adsorbed CO₂ (“CO₂”) was obtained and combined with CO_ad oxidation charge (Figure 5A,B, insets). The subsequent anodic potential sweep generated an oxidation peak at about 0.67 V. The measured currents basically reflect surface Pt area because the reductive adsorption of CO₂ reacts with hydrogen adsorbed on surface Pt only. ^27,37 Comparison of the oxidation charges normalized by Pt loading (Figure 4C) reveals that HQ has the lowest charge, indicating the lowest surface concentration of Pt. The trend of CO₂ oxidation charge is very similar to that of ECSA, although the absolute charge is only about half of the corresponding CO_ad oxidation charge. This low CO₂ coverage is ascribed to the surface coverage of H_ad/Pt and steric interaction between reductively adsorbed radicals. ³⁷

Figure 5. Oxidation curves of reductively adsorbed CO2 on PT, PR, HQ, and SA (A) before and (B) after 3000 potential cycles in 0.1 M HClO4. Insets in A and B show current-time transients for reductive adsorption of CO2 at 0.088 V before and after the potential cycling, respectively. Comparisons of (C) “CO2” adlayer oxidation charges and (D) the charges normalized with respect to the corresponding stripping charges of COad

The charge ratio of “CO₂” to CO_ad ($Q_{{\mathrm{CO}}_2}$/Q _CO) can be affected mainly by average domain size and defect density of Pt domains in the surface layer. The saturation coverage of CO_ad is achieved up to about 0.65 V, which supports that CO_ad oxidation current can represent ECSA. ³⁸ Therefore, the $Q_{{\mathrm{CO}}_2}$/Q _CO ratio eliminates the contribution of surface area to the decrease in “CO₂” coverage caused by the severe corrosive conditions of the potential cycling. Importantly, the $Q_{{\mathrm{CO}}_2}$/Q _CO ratio can be useful to evaluate the relative order of defect density and domain size of Pt skin, with the assumption that binding energies of “CO₂” and CO_ad are independent of surface roughness, particle size, electronic modification of Pt by alloying with Ni, or oxidation levels of surface Pt and Ni, because surface defect density can affect “CO₂” coverage by steric interaction between reductively adsorbed radicals such as –COOH, –COH, –HCO, and –CO. ³⁷ On the basis of the previous findings, we propose the $Q_{{\mathrm{CO}}_2}$/Q _CO ratio as an indirect measure of determining surface roughness change under the assumption that the amount of reductively adsorbed radicals decreases with increasing surface roughness due to the increase of low-coordinating Pt sites.

HQ had much higher $Q_{{\mathrm{CO}}_2}$/Q _CO ratio than those of PT and SA, both before and after the potential cycling (Figure 5D), indicating that HQ had relatively low defect density and/or high Pt skin domain size. In particular, it is reasonably accepted that the highest value of PR before the potential cycling comes from its lowest defect density. Also, considering the order in the ratios of the four samples, we can conclude that the charge ratio depends mainly on surface defects, rather than Pt domain size. $Q_{{\mathrm{CO}}_2}$/Q _CO ratio of PR and HQ decreased during potential cycling, suggesting that a large amount of Ni was dissolved and the surface became rougher at the atomic scale. On the other hand, the increase in $Q_{{\mathrm{CO}}_2}$/Q _CO ratio of SA and PT after potential cycling can be understood as a result of the reduction in the defect density, owing to the repeating adsorption and desorption of oxygen on the surface Pt of SA and PT during the cycling under the corrosive conditions. It is reasonable to postulate that SA and commercial Pt consist of high surface concentration of low coordinated and energetically unstable Pt atoms, which makes surface Pt atoms easily lose the ability to form the reductive CO₂ adsorption by irreversible surface oxide formation. ³⁹

To identify the initial local structures as the origin of the different activities and durabilities of Pt-enriched Pt–Ni NPs, EXFAS analysis was performed. Even though EXAFS measurement around Ni K and Pt L₃ edge has a high bulk-sensitivity, the extracted parameters can reflect the structural change in the surface layer because acid- and hydroquinone-treatments leach out Ni atoms from the surface layer of the alloy NPs. The CN and bond distance (R) extracted from the simulation are listed in Table 1 (for details see Table S4). It is interesting that the extracted CN values are different significantly between SA and HQ. The fact that SA has the lower CN_Pt–O, as compared with HQ is well consistent with the lowest d band vacancies and oxidation states of both Pt and Ni in the XANES (Figure S3) and the XPS results (Figure 2). The lower CN_Pt–Pt and the higher CN_Pt–Ni can be caused by the preferential dissolution of unalloyed Ni cluster (Ni oxides) and consequently, the higher CN of Ni despite the highest surface concentration of Pt, strongly supporting that SA has rough surface layers consisting of electronically unstable Pt and a lower surface concentration of Ni than in HQ. In contrast, HQ exhibited the opposite trend to SA, that is, higher CN_Pt–O and CN_Pt–Pt, lower CN_Pt–Ni, values, highlighting that hydroquinone treatment leaves the largest domain size of Pt-skin in the surface layer by selectively removing alloyed Ni atoms close to Pt. ²⁵

Table 1 Pt L₃-edge EXAFS structure parameters of Pt₃Ni alloy nanoparticles

Catalyst	Pt–O		Pt−Ni		Pt−Pt
	CN a	R b (Å)	CN	R (Å)	CN	R (Å)
Pt foil	N/A c	N/A	N/A	N/A	11.90	2.78
PT	0.90	1.95	N/A	N/A	9.40	2.75
PR	1.90	1.98	3.13	2.63	6.07	2.70
HQ	1.78	1.98	1.73	2.63	7.26	2.70
SA	1.61	1.98	2.07	2.65	6.56	2.72

Abbreviation: EXAFS, extended X-ray absorption fine structure.

^aCN is the coordination number.

^b R is interatomic distance.

^cNot applicable (N/A) to Pt foil and PT because N and R values of Pt–O (Pt foil) and Pt–Ni (Pt foil and PT) are not variables included in the fit.

The initial surface states before and after leaching surface Ni out using sulfuric acid and hydroquinone were illustrated in Figure 6. Both HQ and SA had higher surface Pt concentration than that of PR. At the same time, the schematics of SA and HQ incorporate the present observations of differences in the surface compositional geometries. Surface Pt sites in HQ have the largest number of neighboring Ni(0) and the longest bond distances to O and Ni(0). This can explain the highest electrochemical performance of HQ, although it has the lowest ECSA.

Figure 6. Schematic of (A) HQ, (B) PR, and (C) SA. Metallic Ni is assumed to be not present at the surface layer

CONCLUSION

In summary, we have evaluated the relative surface roughness of Pt–Ni alloy NPs with Pt-rich surfaces, by comparing electrochemical properties such as PZTC and $Q_{{\mathrm{CO}}_2}$/Q _CO ratios from CO- and “CO₂”-displacement experiments. Hydroquinone-treated Pt–Ni alloy (HQ) catalyst showed the highest ORR activity and durability compared with pristine (PR) and sulfuric acid-treated catalysts (SA). Contribution of the surface roughness to the enhancement of the performance of HQ was investigated by comparing the electrochemical properties before and after 3000 potential cycles. PR showed the highest initial value and the largest decrease in $Q_{{\mathrm{CO}}_2}$/Q _CO ratio, strongly supporting that $Q_{{\mathrm{CO}}_2}$/Q _CO ratio is sensitive to surface roughening by potential cycling. The Pt-enriched surfaces, that is, HQ and SA, showed the opposite trend in both $Q_{{\mathrm{CO}}_2}$/Q _CO ratio and PZTC. The initial and final values of HQ are higher than those of SA and slightly decreased after the potential cycles. On the contrary, SA showed the initially lowest values except for PT and increased rather than decreased after potential cycling. Considering that, in comparison to SA, HQ has the lower ECSA, the higher PZTC, and the higher $Q_{{\mathrm{CO}}_2}$/Q _CO, this opposite tendency results mainly from atomic-scale surface roughness. In other words, hydroquinone treatment causes a relatively small decrease in surface roughness and, at the same time, generates a more durable Pt-skin domain in the surface layer. This result described here implies surface Pt roughness is another important factor for ORR activity and durability. This finding led us to conclude that both surface smoothness and Pt-skin are critical for more active and durable Pt-based alloy electrocatalysts. In addition, all our results indicate that the $Q_{{\mathrm{CO}}_2}$/Q _CO ratio obtained from a combination of CO- and “CO₂”-displacement experiments can be employed as an efficient probe for estimating the surface roughness change of Pt–Ni bimetallic NPs.

ACKNOWLEDGMENT

This study was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. 2019R1F1A1062193).

CONFLICT OF INTEREST

The authors declare no conflict of interest. [Correction added on 15 June 2021, after first online publication: Conflict of Interest section has been added.]