# Fileset

[MI_Pt-CCA_re-revised main_final version.docx](https://mdr.nims.go.jp/filesets/240a1cba-04f8-4137-b4a1-78239e857abe/download)

## Creator

Yoshihiro Chida, [Sae Dieb](https://orcid.org/0000-0002-8111-2009), Hiraku Masui, Arata Umehara, [Keitaro Sodeyama](https://orcid.org/0000-0002-9228-0729), Toshimasa Wadayama

## Rights

This document is the Accepted Manuscript version of a Published Work that appeared in final form in ACS Applied Materials & Interfaces, copyright © 2025 American Chemical Society after peer review and technical editing by the publisher. To access the final edited and published work see https://doi.org/10.1021/acsami.4c22052.[In Copyright](http://rightsstatements.org/vocab/InC/1.0/)

## Other metadata

[Surface Nanostructures of Pt-Compositionally Complex Alloy Single-Crystal Model Catalyst Surfaces for Improved Oxygen Reduction Reaction: Machine-Learning-Assisted Exploration](https://mdr.nims.go.jp/datasets/99e22a27-11f2-44fc-be93-1e29fc38ba78)

## Fulltext

Template for Electronic Submission to ACS JournalsSurface Nanostructures of Pt-Compositionally Complex Alloys Single-Crystal Model Catalyst Surfaces for Improved Oxygen Reduction Reaction: Machine-Learning-Assisted ExplorationYoshihiro Chida,*a§† Sae Dieb,b§ Hiraku Masui,a Arata Umehara,a Keitaro Sodeyama,b and Toshimasa Wadayama aa Graduate School of Environmental Studies, Tohoku University, Sendai 980-8579, Japanb National Institute for Materials Science, Tsukuba 305-0044, JapanKEYWORDS: oxygen reduction reaction; platinum; compositionally complex alloy; high entropy alloy; materials informatics; Bayesian optimization; atomically resolved cross-sectional STEM. TOC GraphicABSTRACT: We investigated the oxygen reduction reaction (ORR) properties of Pt-containing compositionally complex alloy (Pt-CCA) single-crystal model catalyst surfaces to optimize the dry-process synthesis conditions, that is, CCA compositions of less-noble alloying elements and their synthesis (annealing) temperatures. Using a machine-learning approach, we effectively navigated the large space of possible synthesis conditions to minimize the experimental workload. The ORR activity and durability of the Pt/CCA/Pt(111) model catalyst surfaces—synthesized through vacuum deposition on a Pt(111) substrate of non-equiatomic Cr-Mn-Fe-Co-Ni or Mn-Fe-Co-Ni alloy (111) lattice stacking layers, followed by a surface Pt(111) layer—depends upon the alloy composition and synthesis temperature: the model catalyst surfaces synthesized with specific combinations of these two parameters outperformed benchmark surfaces such as Pt/equi-atomic Cr-Mn-Fe-Co-Ni/Pt(111) in terms of the ORR durability during potential-cycle loading. The outstanding ORR properties are attributed to the use of machine learning to predict synthesis conditions that are closely linked to the atomic-level surface microstructures that favor enhanced ORR properties. These microstructures enable the formation of a so-called "pseudo-core-shell like structure", i.e., surface Pt(111) underlaid with CCA(111) lattice stacking layers with atomically distributed active elements (Co and/or Ni) close to the surface that are beneficial for ORR property enhancements. This study demonstrates that, not only the "high-entropy" effect of charged less-noble CCA elements, but also the precise control of elemental distributions in the near-surface vicinity in the pristine state, resulting from optimized CCA compositions and synthesis temperatures, are the key factors to improve Pt-CCA catalyst material systems. INTRODUCTIONHigh-performance electrocatalysts are urgently required for the development of various renewable energy systems, particularly for energy conversion and storage devices: material science including informatics plays an essential role for the synthesis of novel catalyst materials, semiconductors, and compounds. For example, binary or ternary alloys have been the main targets of the research and development of high-performance novel catalytic materials, where the main subject is responsible for the target reaction and the promoter has the task(s) of boosting the reaction.1–3 More recently, the idea of multi-elemental alloying, mainly high-entropy alloys (HEAs), has gained attention because of the unique properties of these materials.4–8 In general, HEAs represent the combined chemical properties of their individual constituent elements. This enables these multi-element alloys to exhibit more diverse and complex characteristics owing to electronic interactions among the constituent elements, compared with binary or ternary alloys. Indeed, numerous studies have attested to their potential as excellent catalytic materials for various catalytic reactions involving hydrogen evolution9–11 and methanol oxidation.9,12 Additionally, the increased mixing entropy enhances the thermodynamic stability of alloy systems, leading to unique surface atomic-level microstructures compared to binary alloys. This can improve the surface structural stability in catalytic environments, i.e., the unique microstructures on the surface can enhance the durability of the catalyst.11,13–16 In particular, catalytic HEA materials for polymer electrolyte membrane fuel cells (PEMFCs) have been widely studied as cathode materials. In PEMFCs, the oxygen reduction reaction (ORR) takes place under strongly acidic electrochemical conditions with the potential loading of this reaction fluctuating severely during cycling.6,11,15–20 Actually, we demonstrated that Pt and equiatomic Cr-Mn-Fe-Co-Ni HEA (the so-called Cantor alloy21) single-crystal model catalyst surfaces (Pt/Cr-Mn-Fe-Co-Ni(hkl) (hkl = 111, 110, 100)) exhibit outstanding ORR properties. That is, relative to the corresponding Pt/Co(hkl) binary alloy systems, they demonstrate high specific activity and structural stability against potential cycle (PC) loading, essentially the result of the clear separation of the Pt-enriched surface and the underlying Cr-Mn-Fe-Co-Ni-enriched coherent stacking lattices.22 Furthermore, despite the presence of fewer constituent elements (i.e., lower mixing entropies), the atomic-level surface microstructures and hence the resulting ORR properties are sensitive to the combinations of HEA constituent elements.23 In other words, optimizing the combination of elemental types and controlling the charged compositions of HEAs are essential for developing HEA ORR catalysts by considering the chemical characteristics of the individual HEA elements. We conducted and reported ORR property evaluations of various Pt-“equiatomic multi-element alloys”(111) derived from the Pt-Cantor alloy/Pt(hkl) with fewer constituent elements. The Pt-“Cantor alloy”(111) model catalyst surface delivered the best ORR performance (in terms of its initial activity and durability under the PCs loading) among the Pt-HEA and Pt-medium entropy alloy systems that were tested.Further improvement of the ORR properties and clarifications of the ORR mechanisms of Pt-“multi-element alloys” surfaces would necessitate the experimental investigation and optimization of a huge number of potential combinations of synthesis conditions, such as the constituent elements, multi-component alloy compositions (not only with equiatomic ratios but also extended to compositionally complex alloys (CCAs) with non-equiatomic ratios),24–27 and the thermal annealing (sintering) conditions. However, in general, refining the procedures for synthesizing CCA catalysts to determine the optimal elemental combinations and thermal treatments etc., is much more complex, and would involve a hugely time-consuming exercise to develop a practical “recipe” for CCA catalysts. The efficiency of materials exploration based on conventional experimental approaches has been improved considerably by incorporating machine-learning approaches. Machine learning is useful to learn complex, nonlinear relationships between different variables and, therefore, has been applied to various complex material design problems, such as improving the mechanical and chemical (catalytic) properties of HEAs.28–35 In these complex materials systems, the “recipe” of constituent elements for the extremely complicated chemical and electronic interactions between multiple constituent elements make it exceedingly difficult to predict the properties of a material based on classical bottom-up experimental and theoretical approaches. In this regard, materials informatics, particularly machine-learning methods, offers an efficient way to explore target material spaces by identifying correlations between well-known, quantifiable experimental descriptors, such as composition and synthesis conditions, and evaluating their chemical properties, for example the catalytic activity.36 Therefore, the characteristics of machine learning enable the efficient exploration and discovery of novel materials with unique properties with fewer experimental trials, even when the essential descriptors are unclear. Using these approaches, we attempted to expand our study of the ORR model catalyst Pt-HEA(hkl) by determining  the optimal synthesis conditions, such as the constituent elemental types, their combinations, and their thermal annealing temperatures, to improve the ORR properties and durability of the Pt-CCAs(hkl) surfaces beyond the activity of the initial pristine surface.Our present target is Pt/CCA/Pt(111), which consists of a non-equiatomic Cr–Mn–Fe–Co–Ni or Mn-Fe-Co-Ni CCA underlying layer and a Pt layer covering the surface thereof. In particular, we focused on the composition ratios of Fe, Co, and Ni, which are well-known ORR activity-enhancing elements through alloying with Pt,3,37 as well as the promoter elements of Cr and Mn, both of which likely contribute to enhanced surface microstructural stability (durability), compared to the initial, pristine activity,22,23 and additionally considered the thermal annealing temperatures of the dry-process synthesis. Utilizing an electrochemical experimental platform for Pt-HEA(hkl) model catalyst surfaces that enables both the alloy composition ratios and heat treatment conditions to be controlled with high precision,22,23,38  we first designed and prepared single-crystal model catalyst surfaces and evaluated their ORR properties. To clarify and mathematically optimize the relationship between the aforementioned two synthesis parameters and the ORR properties, the experimental results obtained for Pt-“equiatomic multi-element alloys”(111) surfaces were used as training data for the machine-learning method. To incorporate these experimental variables into the descriptors for the machine learning method, we propose the ORR performance index (OPI) to evaluate the ORR activity and durability of the prepared Pt/CCA/Pt(111) model catalyst surfaces. This design process allows us to iteratively refine the ORR properties of the Pt/CCA/Pt(111) catalyst until they are comparable or superior to those of the Pt-Cantor alloy(111) for specific synthesis conditions (specific ratios of charged less-noble elements and synthesis temperatures).METHODS1. Machine-learning-assisted optimization of the synthesis conditionsThe ability of machine learning to identify complex patterns in large datasets makes it a powerful tool for material design and discovery.39 Several machine-learning methods have been applied in catalysis research to develop new catalysts.40 Active learning is a machine learning approach used to explore large number of potential candidates iteratively by selecting the most informative data point at each iteration. While several methods are established in this context (such as evolutionally algorithms), numerous studies41,42 have shown that Bayesian optimization (BO)42,43 has proved to be a highly efficient machine-learning approach capable of exploring complex and high-dimensional parameter spaces iteratively with a limited number of actual experiments or simulations. BO methods have demonstrated competitive performance in several material design problems, including catalyst design.44-47 BO functions by selecting an optimal solution from the search space of candidates that optimize an objective function. BO approximates the objective function that maps the synthesis conditions to this objective function using a probabilistic surrogate model. By balancing exploration and exploitation, BO can efficiently explore the candidate space to select the next optimal point for the experiment. .Figure 1. Schematic illustration of the machine-learning-assisted exploration of the synthesis conditions for Pt/CCA/Pt(111) and evaluation of the ORR properties. Figure 1 illustrates the procedure for optimizing the dry-process synthesis conditions for improving ORR properties of Pt/CCA/Pt(111) surfaces using  BO. To achieve the research objective of exploring the synthesis conditions for the highly durable Pt-CCAs surface system, we defined a quantitative objective function, named the “ORR performance index (OPI),” to evaluate the ORR properties of model catalyst surfaces. OPI was defined with the following two factors: (1) maintaining high specific activity under PCs, and (2) activity retention rate during the first 1,000 PCs, which is an indicator of surface structural stability.22,23,48,49 We tested various formulas and weighting schemes with our training data shown in Table S1 and S2, which we previously obtained for the Pt-“equiatomic multi-element alloys”(111) (Pt-HEA(111)) in our laboratory and were used as the training data,22,23 iteratively refining the ranking of ORR performances of the fabricated various Pt-CCA(111) to ensure it matched the objectives. Through this trial-and-error process, we arrived at the OPI definition as the following equation:     (eq.1)Here, Ax is the ORR activity of the sample surface during x cycles of PCs used to simulate the PEMFC operating conditions.50 The synthesis conditions to be optimized were as follows: the thermal annealing temperature during the dry-process synthesis, for example, the synthesis temperature of the model catalyst surfaces after the deposition of a four-monolayer(MLs)-thick Pt layer (TC) on the surface, and the composition ratios of the underlaid CCA layers derived from the Cantor alloy (Cr: Mn: Fe: Co: Ni = a: b: c: d: e; where a to e, respectively, are integers of which the sum is equal to 10). The constraints on the synthesis conditions were as follows: TC was set to increase from 473 up to 873 K in increments of 50 K (a total of nine temperatures). To explore the CCA composition ratios, the sum of a (Cr), b (Mn), c (Fe), d (Co), and e (Ni) was 10 (a total of 210 compositions). BO was implemented using PHYSBO (optimization tools for “PHYSics based on BO”), an open source freely available Python library for fast and scalable BO.51 By applying BO to correlate the OPI with the practical dry-process synthesis conditions of Pt/CCA/Pt(111), we iteratively evaluated the ORR properties of three surfaces prepared at various synthesis temperatures and CCA composition ratios at each iteration. In the first batch, the synthesis conditions were designed using only the training data, and in subsequent batches, the evaluated ORR properties of the prepared Pt/CCA/Pt(111) surfaces were fed back to the BO model of the synthesis conditions. Eighteen proposed sets of synthesis conditions for up to six batches are summarized in Table S3. 2. Pt/CCA/Pt(111) model catalyst surface fabricationThe model catalyst surfaces of Pt/CCA/Pt(111) were fabricated in an ultrahigh vacuum (UHV; <1 × 10⁻⁷ Pa) chamber equipped with a manipulation stage with pyrolytic graphite heaters, an ion source gun (PSP; ISIS3000) for surface cleaning (Ar⁺ sputtering), and multiple arc-plasma deposition (APD) sources (ADVANCED RIKO; APS-1). The Pt(111) single-crystal substrate (MaTeck; φ10 mm, t = 1 mm, <0.1º miscut) was cleaned by repeated Ar⁺ sputtering and annealing at 1273 K under UHV conditions. Subsequently, 3 nm-thick (10 monolayer (ML)-thick equivalent) CCA layers were deposited on the surface-cleaned Pt(111) at 773 K of substrate temperature using the APD method, followed by thermal treatment at 773 K for 30 min. By considering the atomic radii, the 1 ML-equivalent thickness (0.3 nm) for the alloys and Pt was determined using a quartz oscillator (Sigma Instruments: SQM-160) installed in the UHV-APD chamber. A bias arc voltage of 70 V and a pulse frequency of 2 Hz were used for the APD processes of both the alloy layers and the subsequent Pt layers.To achieve the charged CCA compositions predicted by machine learning, custom-made alloy and/or single-element targets (Cr-Mn-Fe-Co-Ni, Cr-Mn-Co-Ni, Mn-Fe-Co-Ni, Cr-Mn-Fe, Cr-Mn-Co, Cr-Co-Ni, Mn-Co-Ni, Fe-Co-Ni, Cr-Co, Mn-Co, Fe-Co, Co-Ni, Co and Ni) were prepared by mixing and sintering different constituent elements (Cr, Mn, Fe, Co, and Ni) in various combinations in equal composition ratios (>99.9% overall purity; Toshima Manufacturing Co., Ltd.). By alternately depositing material from these targets, it was possible to form CCA layers with the desired composition ratios.22,23 For example, to form the CCA layer with the composition ratio Cr1Mn2Fe1Co3Ni3, we used both equiatomic Mn-Co-Ni and Cr-Fe-Co-Ni targets. We alternately deposited layers with approximate thicknesses of 0.2 and 0.1 nm from the Mn-Co-Ni and Cr-Fe-Co-Ni targets, respectively, by repeating this deposition cycle for a total of ten cycles to fabricate a CCA layer with a total thickness of 3.0 nm and then performing thermal treatment to promote interdiffusion. The composition elements and combinations of targets used in the CCA-layer-APD for each Pt/CCA/Pt(111) surface, as well as the amount of deposition irradiated from each target, are shown in Table S4. Finally, 1.2 nm-thick (4 ML-thick equivalent) Pt layer (>99.95% purity; Nilaco Corporation) was deposited on the pre-deposited CCA layer at 298 K, followed by annealing at various substrate temperatures (synthesis temperature: TC). The APD-synthesized Pt/CCA/Pt(111) samples are designated as “CraMnbFecCodNie_TCK@X batch”, where a, b, c, d, and e denote the composition ratio (Cr, Mn, Fe, Co, and Ni, respectively), and X indicates the batch of data acquired. In some figures, simplified designation, abcde_TcK, is used to fit them into the limited space.3. Electrochemical measurements Electrochemical measurements were performed using a conventional three-electrode cell equipped with a potentiogalvanostat (HZ-5000, Hokuto Denko) and a rotating disk electrode (RDE) system (HR-301, Hokuto Denko). A reversible hydrogen electrode (RHE) and a Pt wire served as reference and counter electrodes, respectively. Using a dedicated transfer vessel, the APD-synthesized Pt/CCA/Pt(111) samples were transferred from the UHV chamber to a glove box purged with 1 atm of N₂ to prevent oxidation and contamination.52 The geometric surface area of the working electrodes (sample surface) was controlled using a Karletz O-ring (Dupont-P4).22,23,38,52 All potentials are presented with respect to the reversible hydrogen electrode (RHE). CV scans were performed at 0.05 V s⁻¹ of sweep speed in the potential range of 0.05–1.0 V without disc rotation until the CV shape stabilized. Subsequently, CO-stripping voltammetry (CO-SV) was conducted by bubbling CO into the solution at 0.08 V vs. RHE until the electrode surface was saturated, then re-purging with N₂ to remove dissolved CO, and finally using the same CV protocol.22,23,38 Linear sweep voltammetry (LSV) was then performed in O₂-saturated 0.1 M HClO₄ (298 K) at rotation rates of 400–2,500 rpm, using a positive sweep of 10 mV s⁻¹ between 0.05 and 1.05 V. The ORR activity was evaluated from the jk values of LSV estimated at 0.9 V vs. RHE using the Koutecky-Levich equation.53 Finally, to investigate structural stability (i.e., ORR durability), square-wave PCs were applied between 0.6 V and 1.0 V every 3 s in O₂-saturated 0.1 M HClO₄ at 25 °C up to 5,000 PCs.50 The ORR activity trends were monitored at 1,000-cycle intervals, and CO-SV was also performed before and after these PCs to estimate changes in the effective electrochemical surface area (ECSA) of Pt.4. Surface microstructure observationAtomically resolved surface cross-sectional images and EDS intensity line profiles of each constituent element in the vicinity of Pt/CCA/Pt(111)  surface were collected using HAADF-STEM-EDS (JEOL; JEM-ARM200F). For atomic-level observations, the Ga-focused ion beam process on the surface of the model catalysts was preprocessed after coating the model catalyst surfaces with carbon.RESULTS AND DISCUSSION1. Electrochemical properties evaluationFigure 2. LSV curves for the (a) pristine-state and (b) 5,000 potential-cycle loadings (PCs) of the Pt/CCA/Pt(111) (Cr1Mn1Fe3Co3Ni2_573K@4th batch (green) and Mn1Fe2Co5Ni2_673K@6th batch (blue)) surfaces recorded under O2-saturated conditions. The current densities of the working electrodes (jk) were calculated by using the respective geometrical surface area (cmgeo-2). Those of the Pt/Cantor alloy/Pt(111) (red), Pt/Co/Pt(111) (yellow) and clean Pt(111) (black dashed) collected in our previous study are shown as references. (c) ORR activity trends for these sample surfaces during 5,000 PCs in the range of 0.6 V (3s) – 1.0 V (3s) vs. RHE. CV curves for the samples with surfaces in the (d) pristine state and (e) after applying 5,000 PCs collected after purging with N2 in 0.1 M HClO4. (f) ECSAs of the Cr1Mn1Fe3Co3Ni2_573K@4th batch and Mn1Fe2Co5Ni2_673K@6th batch surfaces estimated after 5,000 PCs by CO-SV. That of the Pt/Cantor alloy/Pt(111) and Pt/Co/Pt(111) are shown as references (also as horizontal dashed lines). Table S5 summarizes the ORR activity trends every 1,000 cycles for the Pt/CCA/Pt(111) surfaces (A0 to A5k) prepared under the synthesis conditions listed in Table S3, along with the corresponding OPI. Electrochemical measurements result of respective samples are presented in Figure S1-S20. Regarding the initial activity (A0) and the activity estimated after 5,000 PCs (A5k), several Pt/CCA/Pt(111) surfaces, Cr2Mn2Fe2Co1Ni3_573K@1st batch, Cr2Mn3Fe1Co2Ni2_623K@2nd batch, Mn2Fe2Co4Ni2_623K@3rd batch and Mn1Fe3Co4Ni2_573K@5th batch etc., maintained higher activity than a Pt-Co alloy system (Pt/Co/Pt(111)), which is used as a practical catalyst material. Most notably, as indicated in Figure 2(a)-(c), Cr1Mn1Fe3Co3Ni2_573K@4th batch and Mn1Fe2Co5Ni2_673K@6th batch exhibited activities comparable to and higher than those of the Pt/Cantor alloy/Pt(111) in our previous study,22 suggesting superior ORR durability. The CV curves of these pristine surfaces shown in Figure 2(d) exhibit typical oxidation/reduction features of Pt-based alloy (111) surfaces, such as decrease in the adsorption/desorption charges of H-related species (below 0.4 V) and higher shift in the onset potential for O/OH adsorption (above 0.6 V) than that of clean Pt(111) (dashed line).2,22,23,38,54-58 These behaviors indicate changes in Pt electronic state due to the alloying, and especially, the fact that comparable shift of the O/OH adsorption peak suggests that all the Pt/CCA(111) surfaces possess a similar level of initial ORR activity. In contrast, Figure 2(e) shows that after 5,000 PCs, the H-related adsorption/desorption charges on the Pt/Co/Pt(111) surface have changed more markedly than on the other surfaces. As demonstrated by the electrochemical surface area (ECSA) measurements via CO-SV in Figure 2(f), this result indicates that model Pt-alloy(111) surface roughening, i.e., structural degradation induced by the PC-loading,22,23,38,59-61 was most advanced on the Pt/Co/Pt(111). Meanwhile, the two Pt/CCA/Pt(111) surfaces exhibit almost the same level suppression of structural degradation as the Pt/Cantor alloy/Pt(111).22,23,38 In model Pt-alloy(111) catalysts, surface roughening caused by PC-loadings relieves the compressive strain inherent in the alloyed Pt(111) and exposes undercoordinated crystal facets, thereby leading to deactivation. Consequently, the extent of this structural degradation is a direct and important indicator of ORR durability. Therefore, by controlling the alloy compositions and synthesis temperatures, one can synthesize Pt/CCA/Pt(111) surfaces that maintain high activity under the PC loading condition, thereby achieving ORR performance equal or even superior to that of Pt/Cantor alloy/Pt(111). In addition, we should emphasise that the Pt-CCA(111) surface synthesis conditions that exhibits high ORR performance exceeding the benchmark Pt/Cantor alloy/Pt(111) were successfully found through a search of two percent of the total exploration space, including training data.2. Surface microstructure observationFigure 3. Atomically resolved cross-sectional high-angle annular dark field (HAADF)-scanning transmission electron microscopy (STEM) images collected for the pristine-state surfaces of (a-1,a-2) Cr1Mn1Fe3Co3Ni2_573K@4th batch and Mn1Fe2Co5Ni2_673K@6th batch. Corresponding line profiles of the energy-dispersive X-ray spectroscopic (EDS) signal intensities of each element in the direction normal to the surface (along the yellow arrows in the STEM images) are presented in (c-1,c-2) and (d), respectively.Figure 3 presents the atomically resolved cross-sectional HAADF-STEM images and the corresponding EDS intensity line profiles of each constituent element in the direction normal to the surface for the pristine Cr1Mn1Fe3Co3Ni2_573K@4th batch (Figure 3(a-1, a-2, c-1,c-2), respectively) and Mn1Fe2Co5Ni2_673K@6th batch (Figure 3(b,d), respectively). As shown in Figure S21(a), Cr1Mn1Fe3Co3Ni2_573K@4th batch displays a unique atomic-level microstructure with irregularly distributed local regions of different contrast (Z-contrast) of the image in the horizontal direction of the surface. In Figure 3(a-1), there is a darker region about 2-4 nm deep from the top-most surface (marked by a red dotted line), and a large amount of CCA constituent elements are distributed within this region, accompanied by a lower Pt signal intensity (Figure 3(c-1)). Hence, the Cr1Mn1Fe3Co3Ni2_573K@4th batch surface comprises Pt-enriched surface layer and underlaid alloying-element-rich  lattice stacking layer of (111), forming what we have previously referred to as “pseudo-core-shell structure”.22,23 Yet compared with other Pt-“multi-element alloys”(111) surfaces that exhibit clearer phase separation, such as Pt/Cantor alloy/Pt(111), the Z-contrast on this sample surface is relatively unclear.22,23 Indeed, in another field of view (Figure 3(a-2) and (c-2)), the Z-contrast is barely obscure, suggesting that localized interdiffusion between the surface Pt and the underlying alloying elements is activated. In contrast, the Mn1Fe2Co5Ni2_673K@6th batch surface exhibits coherent lattice stacking from the top-most surface down to the substrate interface without any authenticate Z-contrast (Figure 3(b)), similar to the Pt/Co/Pt(111).22 Moreover, across the entire APD region (3 nm for the alloy layer and 1.2 nm for the surface Pt layer) both Pt and the CCA constituents display a wider distribution to Cr1Mn1Fe3Co3Ni2_573K@4th batch (Figure 3(d)).One possible origin of the “pseudo-core-shell structure” is the high thermodynamic stability, i.e., low Gibbs free energy, of the multi-elemental alloy phase which depends on both the chemical interactions (mixing enthalpy) among the constituent elements and the mixing entropy, itself governed by the number of elements and their composition ratios.23 Since Cr1Mn1Fe3Co3Ni2 contains five elements in the CCA phase and has large mixing entropy just like Cantor alloy, thermal diffusion of the constituent atoms is suppressed and phase separation tendency is fostered due also to sluggish diffusion,22,23,62-65 which is one of the so-called high-entropy effects. However, deviations from an equi-atomic ratio lower the mixing entropy from its theoretical maximum, potentially triggering the local collapse of any phase-separated structure.23,66-68 Moreover, in Mn1Fe2Co5Ni2, which lacks Cr and combines four other elements in significantly off-equi-atomic ratio, more active thermal diffusion of the constituent elements can be expected. Consequently, the observed near-surface microstructures in these samples can be rationalized in terms of the CCA composition and ratios, appear highly reasonable.3. Key factors to optimize the ORR performance of the Pt-CCA materials systemFigure 4. Relationship between the three ACP factors (CCA composition), TC (synthesis temperature), and OPI (ORR performance index). The color coding between data points is based on extrapolation.Here, we provide an overview of the extent to which the ORR properties (initial, pristine activity, and durability of ORR) depend on the dry-process synthesis conditions of the Pt/CCA/Pt(111) surfaces, based on the present machine-learning descriptors of the composition ratios of a(Cr), b(Mn), c(Fe), d(Co), and  e(Ni) and the synthesis temperature (TC). Numerous studies on alloy-based ORR catalysts have identified Co and Ni as elements capable of efficiently boosting the ORR activity by alloying with Pt.2,3,37,69 Fe is also classified as an ORR-enhancing alloying element for Pt;3,69 however, Fe is known to catalyze the formation of  highly reactive hydroxyl radicals from hydrogen peroxide in PEMFCs; therefore, the use of Fe should be avoided from the viewpoint of degradation of the PEM.70,71 Based on the aforementioned elemental restrictions on the alloying elements of Pt, we categorized the CCA constituent elements into three groups based on the enhancement of the activity of the Pt surface by alloying with (i) Cr and Mn, (ii) Fe, and (iii) Co and Ni. Thus, weighting factors were applied to the composition ratios of a(Cr), b(Mn), c(Fe), d(Co), and e(Ni) by, for example, multiplying category (i) by 5, (ii) by 10, and (iii) by 20. By summing the above-mentioned weighted composition ratios of a to e, we defined the alloy composition parameter (ACP) for the composition ratios of the CCA constituent elements, as expressed by the equation below:0                    (eq.2)For example, the value of the ACP of the Pt/Cantor alloy/Pt(111) (TC = 623 K), which has an equal composition ratio of a(Cr), b(Mn), c(Fe), d(Co), and e(Ni), was 120. In this way, we weighted the composition ratios of Co and Ni relative to the other elements and used the ACP instead of the “raw” (charged) composition ratios for a to e. Figure 4 shows a scatter plot of experimental data for the OPI of the prepared Pt/CCA/Pt(111) model catalyst surfaces, where the  x- (horizontal)  and y- (vertical) axes represent the ACP value and TC. The color coding in the diagram corresponds to the OPI values evaluated for each data point, and the coloring between data points is based on extrapolation. As shown in Figure 4, for ACP values ranging from 120 to 140, the regions with relatively higher OPI values (indicated in red and orange) tended to expand with decreasing TC. This tendency reveals that when the composition ratios of d(Co) and e(Ni) (well-known ORR boosting elements when alloyed with Pt) are high, a TC of between 623 and 573 K is desirable for achieving Pt-CCA catalysts with high ORR performance. For ACP values exceeding 140, the aforementioned tendency becomes unclear, that is, discrete values of OPI against TC can be seen in the scatter plot in Figure 4. In other words, the OPI values are sensitive to the number (N) and type of CCA constituent elements: even for the CCA surface without Cr (Mn1Fe2Co5Ni2_673K@6th batch; N < 5), the ORR performance was overwhelmingly, which was equivalent to or slightly exceeded that of the Cr-included CCA with the Cr1Mn1Fe3Co3Ni2_573K@4th batch surface (N = 5).Figure 5. Schematical illustration of the pristine Cr1Mn1Fe3Co3Ni2_573K@4th batch (left side) and Mn1Fe2Co5Ni2_673K@6th batch (right side) surfaces (before ORR evaluation).Considering the results shown in Figure 4, we discuss the surface nanostructures of the high-performance Pt-CCA catalyst of Cr1Mn1Fe3Co3Ni2_573K@4th batch and Mn1Fe2Co5Ni2_673K@6th batch surfaces. Figure 5 schematically depicts the surface structures of these samples in the pristine state.Cr1Mn1Fe3Co3Ni2_573K@4th batch surface tends to comprise “pseudo-core-shell structure,” namely Pt-enriched surface layer with an underlying CCA-rich phase. Our previous studies on Pt/Cantor alloy/Pt(111) have shown that forming such “pseudo-core-shell structure” suppresses the dissolution of alloying elements and suppress surface structural degradation, thereby improving durability.22,23 This suggests that the characteristic surface structure of Cr1Mn1Fe3Co3Ni2_573K@4th batch is a rational explanation for its high ORR performance. However, ACP value is 120 for an equi-atomic ratio, so increasing ACP value from 120 to 140 (which is the value of Cr1Mn1Fe3Co3Ni2) means the overall composition to become more off-equi-atomic. As noted in STEM-EDS observation (Figure 3(a) and (c)), this ACP value shift weakens the high-entropy effect that ordinarily suppresses thermal diffusion of the constituent atoms. Hence, lowering the synthesis temperature (Tc) to be inactivate for the thermal diffusion of Pt-Cr-Mn-Fe-Co-Ni should yield a more durable Pt-CCA(111) surface structure. In other words, by appropriately balancing the CCA composition, particularly the relative amounts of Cr and Mn versus Co and Ni and increase in synthesis temperatures that activate relative diffusions, one can form an initial surface distribution corresponding to a “pseudo-core-shell structure.” We propose that this approach is one of the crucial factors for achieving high ORR performance in Pt-CCA surface systems. In fact, to gain a deeper understanding of the correlation between the synthesis temperature and ORR performance for Pt/Cr1Mn1Fe3Co3Ni2/Pt(111), we conducted additional experiments on a sample synthesized at a temperature 100 K higher than that of Cr1Mn1Fe3Co3Ni2_573K@4th batch (namely, Cr1Mn1Fe3Co3Ni2_673K). As shown in Table S6 and Figure S22, this surface exhibited a characteristic activity transition wherein significant deactivation occurs early in the PC-loading, revealing that alloying elements distributed near the surface had dissolved. We added these data points to the correlation mapping in Figure 4, and presented the result in Figure S23, confirming that the overall correlation trend remained unchanged. This insight further underscores the above discussion about the relationship between the surface microstructure of Pt/Cr-Mn-Fe-Co-Ni CCA/Pt(111) and corresponding ORR properties. Unlike Cr1Mn1Fe3Co3Ni2_573K@4th batch, Mn1Fe2Co5Ni2_673K@6th batch surface does not form “pseudo-core-shell structure,” as shown in Figure 3(b). In other words, Cr-free Pt/Mn-Fe-Co-Ni CCA/Pt(111) improves ORR performance through different material factors from those in Pt/Cr–Mn–Fe–Co–Ni CCA (or HEA)/Pt(111). This finding is particularly interesting as an example of how machine-learning-assisted exploration of constituent elements and synthesis conditions within minimal bias and few trials can yield “unexpectedly” promising outcomes. Although the main objective of this study is not to pinpoint the decisive reason behind the improved ORR performance of Mn1Fe2Co5Ni2_673K@6th batch, previous attempts with Pt-quaternary alloy(111) surfaces (Mn2Fe2Co4Ni2_623K@3rd batch, Mn2Fe2Co4Ni2_573K@5th batch, Mn1Fe3Co4Ni2_573K@5th batch) suggest that high ORR performance arises in case that the Mn ratio (b) is minimized and the synthesis temperature Tc is relatively high. The EDS results (Figure 3(d)) show that, under these conditions, all CCA constituents were facilitated to be widely distributed throughout the deposited layer due to relatively high synthesis temperatures and easily diffusible CCA composition. In particular, Co, which strongly contributes to improving ORR activity and has a large composition ratio (d(Co) = 5), is also enriched in the top-most surface. Meanwhile, Mn, which contributes less to activity enhancement,69 remains depleted in the near-surface region. This may be related to its inherently low vapor pressure,72 which could cause Mn to be removed from the surface during vacuum annealing. Naturally, such the elemental distributions likely to be beneficial for the activity enhancement during the early stages of PC-loading. Furthermore, considering (i) that Pt-quaternary alloy(111) exhibits marked difference in the ECSA after 5,000 PCs depending on the synthesis conditions (Figure S2), and (ii) that Mn has a low mixing enthalpy with Pt, Co, and Ni, indicating strong inter-element interactions,66 it appears that enhancing the overall thermodynamic stability of homogeneous Pt–Mn–Fe–Co–Ni solid solution formed by a specific synthesis condition in the sub-surface region, rather than relying on the thermodynamic stability of CCA phase, plays an essential role for significant influence on maintaining surface structural stability (ORR durability).In any case, it has become evident that the nanostructure and materials-science characteristics required for Pt-“multi-element alloys” surface system to apply practical ORR catalysts, specific combinations of constituent elements are desirable. This finding is one of the major achievements of this study, which was made possible by the materials exploration assisted by machine-learning.CONCLUSIONSBy applying a machine-learning (BO) method and newly defined ORR performance descriptors for Pt/CCA/Pt(111) model catalyst surface such as OPI, TC, and ACP, we explored the optimal synthesis conditions required to boost the ORR performance of the surfaces. Specifically, two synthesis conditions of Pt/CCA/Pt(111), Cr1Mn1Fe3Co3Ni2 with 573K annealing and Mn1Fe2Co5Ni2 with 673K annealing, exhibiting ORR properties either equivalent to or exceeding those of the benchmark Pt/Cantor alloy(111) were successfully determined by conducting a small number of experimental trials during which approximately two percent of the total candidate space was explored due to the iterative design process of BO, which enabled the stepwise and efficient exploration of the optimal conditions for high ORR properties. It is undoubtedly a crucial insight that, for Pt-“Cr-containing CCA (Cr-Mn-Fe-Co-Ni)”(111), we have deepened the understanding of surface design guidelines based on the cumulative knowledge of high-ORR-performance strategies for Pt-“multi-element alloys”(111) systems. Equally noteworthy, however, is our finding that even Pt-“Cr-free CCA (Mn-Fe-Co-Ni)”(111) can achieve high ORR performance through optimal synthesis conditions, which intriguingly broadens the potential of Pt-CCA catalysts. The results demonstrate that our experimental study platform for a single-crystal Pt-HEA (Pt-CCA) model catalyst becomes more powerful with the incorporation of the machine-learning method. Notably, the results are expected to provide valuable insights into atomic-level surface designs and the development of high-performance electrocatalysts; furthermore, they could be applied to a wide variety of catalysis studies, particularly in multi-component alloys and/or systems consisting of compound materials.ASSOCIATED CONTENTSupporting Information. The data supporting this study are included in the ESI (Electronic Supplementary Information) and the material is available free of charge via the Internet. Training dataset for the Bayesian optimization prediction, information of machine-learning predicted synthesis condition, APD target information of CCA-layer synthesis, supplemental electrochemical measurement results for Pt/CCA/Pt(111), and cross-sectional STEM images of the samples (a different magnification of Figure 3(a) and (c)).・MI_Pt-CCA_ESI (docx)AUTHOR INFORMATIONCorresponding Author* Yoshihiro Chida - Graduate School of Environmental Studies, Tohoku University, Sendai 980-8579, JapanEmail: yoshihiro.chida@aist.go.jpPresent Addresses† Yoshihiro Chida - National Institute of Advanced Industrial Science and Technology, Ikeda 563-8577, JapanAuthor ContributionsThe manuscript was written with contributions from all the authors. All the authors approved the final version of the manuscript. Y.C. conducted experimental investigation, data curation, formal analysis, visualization, writing of the original draft, review, and editing. S.D. implemented the optimization of the synthesis conditions through the coding and operation of Bayesian optimization, writing of the original draft, and reviewing and editing. M.H. and A.U. were involved in experimental investigation, data curation, review, and editing. K.S. contributed to review and editing. T.W. contributed to conceptualization, funding acquisition, project administration, supervision, review, and editing. §These authors contributed equally.)Funding SourcesThe New Energy and Industrial Technology Development Organization (NEDO) of Japan (Grant Number: 20001184-0 for T.W.) and JSPS KAKENHI (Grant Number: JP21H01645 for T.W. and JP23KJ0111 for Y.C.).ACKNOWLEDGMENTWe gratefully acknowledge support by the New Energy and Indus-trial Technology Development Organization (NEDO) of Japan and JSPS KAKENHI. Additionally, Y.C. acknowledges support in the form of a Grant-in-Aid for JSPS Fellows.REFERENCES(1) Oezaslan, M.; Hasché, F.; Strasser, P. PtCu 3 , PtCu and Pt 3 Cu Alloy Nanoparticle Electrocatalysts for Oxygen Reduction Reaction in Alkaline and Acidic Media. J. Electrochem. Soc. 2012, 159, B444–B454. https://doi.org/10.1149/2.106204jes.(2) Stamenkovic, V. R.; Fowler, B.; Mun, B. S.; Wang, G.; Ross, P. N.; Lucas, C. A.; Marković, N. M. Improved Oxygen Reduction Activity on Pt3Ni(111) via Increased Surface Site Availability. Science. 2007, 315, 493–497. https://doi.org/10.1126/science.1135941.(3) Li, C.; Zhao, W.; Lu, X.; Chen, Z.; Han, B.; Zhang, X.; Chen, J.; Shao, Y.; Huo, J.; Zhu, Y.; Deng, Y.; Sun, S.; Liao, S. PtCoNi Ternary Intermetallic Compounds Anchored on Co, Ni and N Co-doped Mesoporous Carbon: Synergetic Effect between PtCoNi Nanoparticles and Doped Mesoporous Carbon Promotes the Catalytic Activity. J. Energy Chem. 2023, 78, 340–349. https://doi.org/10.1016/j.jechem.2022.12.005.(4) Rößner, L.; Patiño Soriano, D. T.; Tiryaki, O.; Burkhardt, U.; Armbrüster, M. Synthesis of Isostructural Intermetallic Sn-Pb-Bi-Pt Platform Materials for Catalytic Investigations. Inorg. Chem. 2023, 62, 4688–4695. https://doi.org/10.1021/acs.inorgchem.3c00163.(5) Löffler, T.; Meyer, H.; Savan, A.; Wilde, P.; Garzón Manjón, A.; Chen, Y. T.; Ventosa, E.; Scheu, C.; Ludwig, A.; Schuhmann, W. Discovery of a Multinary Noble Metal–Free Oxygen Reduction Catalyst. Adv. Energy Mater. 2018, 8, 1802269–1802274. https://doi.org/10.1002/aenm.201802269.(6) Yao, Y.; Li, Z.; Dou, Y.; Jiang, T.; Zou, J.; Lim, S. Y.; Norby, P.; Stamate, E.; Jensen, J. O.; Zhang, W. High Entropy Alloy Nanoparticles Encapsulated in Graphitised Hollow Carbon Tubes for Oxygen Reduction Electrocatalysis. Dalton Trans. 2023, 52, 4142–4151. https://doi.org/10.1039/d2dt03637a.(7) He, Y.; Zhu, X.; Zhang, C.; Liu, Z.; Cao, B.; Sheng, L.; Yang, Y.; Feng, Q.; Wang, N.; Ou, J. Z.; Xu, Y. Self-Circulating Adsorption–Desorption Structure of Non-Noble High-Entropy Alloy Electrocatalyst Facilitates Efficient Water Splitting. ACS Sustainable Chem. Eng. 2023, 11, 5055–5064. https://doi.org/10.1021/acssuschemeng.2c06888.(8) Dey, G. R.; McCormick, C. R.; Soliman, S. S.; Darling, A. J.; Schaak, R. E. Chemical Insights into the Formation of Colloidal High Entropy Alloy Nanoparticles. ACS Nano. 2023, 17, 5943–5955. https://doi.org/10.1021/acsnano.3c00176.(9) Wang, Y.; Luo, W.; Gong, S.; Luo, L.; Li, Y.; Zhao, Y.; Li, Z. Synthesis of High-Entropy-Alloy Nanoparticles by a Step-Alloying Strategy as a Superior Multifunctional Electrocatalyst. Adv. Mater. 2023, 35, (2302499), 1–11. https://doi.org/10.1002/adma.202302499.(10) Sivanantham, A.; Lee, H.; Hwang, S. W.; Lee, H. U.; Cho, S. B.; Ahn, B.; Cho, I. S. Complementary Functions of Vanadium in Boosting Electrocatalytic Activity of CuCoNiFeMn High-Entropy Alloy for Water Splitting. Adv. Funct. Materials 2023, 33, (2301153). https://doi.org/10.1002/adfm.202301153.(11) Hu, Z.; Chen, K.; Zhu, Y.; Liu, B.; Shen, J. Synergistic Effects of PtRhNiFeCu High Entropy Alloy Nanocatalyst for Hydrogen Evolution and Oxygen Reduction Reactions. Small. 2024, 20, (2309819), 1–8. https://doi.org/10.1002/smll.202309819.(12) Subhash, B.; Unocic, R. R.; Lie, W. H.; Gallington, L. C.; Wright, J.; Cheong, S.; Tilley, R. D.; Bedford, N. M. Resolving Atomic-Scale Structure and Chemical Coordination in High-Entropy Alloy Electrocatalysts for Structure-Function Relationship Elucidation. ACS Nano. 2023, 17, 22299–22312. https://doi.org/10.1021/acsnano.3c03884.(13) Kormányos, A.; Dong, Q.; Xiao, B.; Li, T.; Savan, A.; Jenewein, K.; Priamushko, T.; Körner, A.; Böhm, T.; Hutzler, A.; Hu, L.; Ludwig, A.; Cherevko, S. Stability of High-Entropy Alloys under Electrocatalytic Conditions. iScience 2023. iScience. 2023, 26, (107775), 1–17. https://doi.org/10.1016/j.isci.2023.107775.(14) Zhao, X.; Cheng, H.; Chen, X.; Zhang, Q.; Li, C.; Xie, J.; Marinkovic, N.; Ma, L.; Zheng, J. C.; Sasaki, K. Multiple Metal-Nitrogen Bonds Synergistically Boosting the Activity and Durability of High-Entropy Alloy Electrocatalysts. J. Am. Chem. Soc. 2024, 146, 3010–3022. https://doi.org/10.1021/jacs.3c08177.(15) Chen, T.; Ning, F.; Qi, J.; Feng, G.; Wang, Y.; Song, J.; Yang, T.; Liu, X.; Chen, L.; Xia, D. PtFeCoNiCu High-Entropy Solid Solution Alloy as Highly Efficient Electrocatalyst for the Oxygen Reduction Reaction. iScience 2023. iScience. 2023, 26, (105890), 1–13. https://doi.org/10.1016/j.isci.2022.105890.(16) Yu, Y.; Xia, F.; Wang, C.; Wu, J.; Fu, X.; Ma, D.; Lin, B.; Wang, J.; Yue, Q.; Kang, Y. High-Entropy Alloy Nanoparticles as a Promising Electrocatalyst to Enhance Activity and Durability for Oxygen Reduction. Nano Res. 2022, 15, 7868–7876. https://doi.org/10.1007/s12274-022-4432-1.(17) Ding, Q.; Zhang, Y.; Chen, X.; Fu, X.; Chen, D.; Chen, S.; Gu, L.; Wei, F.; Bei, H.; Gao, Y.; Wen, M.; Li, J.; Zhang, Z.; Zhu, T.; Ritchie, R. O.; Yu, Q. Tuning Element Distribution, Structure and Properties by Composition in High-Entropy Alloys. Nature. 2019, 574, 223–227. https://doi.org/10.1038/s41586-019-1617-1.(18) Qiu, H. J.; Fang, G.; Wen, Y.; Liu, P.; Xie, G. Nanoporous High-Entropy Alloys for Highly Stable and Efficient Catalysts. J. Mater. Chem. A, 2019, 7, 6499-6506. https://doi.org/10.1039/C9TA00505F(19) Li, S.; Tang, X.; Jia, H.; Li, H.; Xie, G.; Liu, X.; Lin, X.; Qiu, H. J. Nanoporous High-Entropy Alloys with Low Pt Loadings for High-Performance Electrochemical Oxygen Reduction. J. Catal. 2020, 383, 164–171. https://doi.org/10.1016/j.jcat.2020.01.024.(20) Yang, C. L.; Wang, L. N.; Yin, P.; Liu, J.; Chen, M. X.; Yan, Q. Q.; Wang, Z. S.; Xu, S. L.; Chu, S. Q.; Cui, C.; Ju, H.; Zhu, J.; Lin, Y.; Shui, J.; Liang, H. W. Sulfur-Anchoring Synthesis of Platinum Intermetallic Nanoparticle Catalysts for Fuel Cells. Science. 2021, 374, 459–464. https://doi.org/10.1126/science.abj9980.(21) Cantor, B.; Chang, I. T. H.; Knight, P.; Vincent, A. J. B. Microstructural Development in Equiatomic Multicomponent Alloys. Mater. Sci. Eng. A. 2004, 375–377 (1-2 SPEC. ISS.), 213–218. https://doi.org/10.1016/j.msea.2003.10.257.(22) Chida, Y.; Tomimori, T.; Ebata, T.; Taguchi, N.; Ioroi, T.; Hayashi, K.; Todoroki, N.; Wadayama, T. Experimental Study Platform for Electrocatalysis of Atomic-Level Controlled High-Entropy Alloy Surfaces. Nat. Commun. 2023, 14, 4492. https://doi.org/10.1038/s41467-023-40246-5.(23) Chida, Y.; Tomimori, T.; Todoroki, N.; Wadayama, T. Pt-Surface Stabilization by High-Entropy Alloys for Enhancing Oxygen Reduction Reaction Property: Single-Crystal Model Catalyst Study. Electrochem. Commun. 2024, 159, (107657). https://doi.org/10.1016/j.elecom.2023.107657.(24) Inui, H.; Tsuji, N. Preface to Special Issue on Recent Advances in the Understanding of Peculiar Properties of High-Entropy Alloys. Mater. Jpn. 2018, 57, 311–311. https://doi.org/10.2320/materia.57.311.(25) Manzoni, A. M.; Glatzel, U. New Multiphase Compositionally Complex Alloys Driven by the High Entropy Alloy Approach. Elsevier Inc. January 1, 2019. Materials Characterization. 2019, 147, 512–532. https://doi.org/10.1016/j.matchar.2018.06.036.(26) Ikeda, Y.; Grabowski, B.; Körmann, F.; Phase, A. I. Ab Initio Phase Stabilities and Mechanical Properties of Multicomponent Alloys: A Comprehensive Review for High Entropy Alloys and Compositionally Complex Alloys. Elsevier Inc. January 1, 2019. Materials Characterization. 2019, 147, 464–511. https://doi.org/10.1016/j.matchar.2018.06.019.(27) Fantin, A.; Lepore, G. O.; Manzoni, A. M.; Kasatikov, S.; Scherb, T.; Huthwelker, T.; d’Acapito, F.; Schumacher, G. Short-Range Chemical Order and Local Lattice Distortion in a Compositionally Complex Alloy. Acta Mater. 2020, 193, 329–337. https://doi.org/10.1016/j.actamat.2020.04.034.(28) Lee, J.; Jinnouchi, R. Machine Learning-Based Screening of Highly Stable and Active Ternary Pt Alloys for Oxygen Reduction Reaction. J. Phys. Chem. C. 2021, 125, 16963–16974. https://doi.org/10.1021/acs.jpcc.1c02890.(29) Ghosh, S.; Sotskov, V.; Shapeev, A. V.; Neugebauer, J.; Körmann, F. Short-Range Order and Phase Stability of CrCoNi Explored with Machine Learning Potentials. Phys. Rev. Materials 2022, 6, (113804). https://doi.org/10.1103/PhysRevMaterials.6.113804.(30) Meyer, B.; Sawatlon, B.; Heinen, S.; Von Lilienfeld, O. A.; Corminboeuf, C. Machine Learning Meets Volcano Plots: Computational Discovery of Cross-Coupling Catalysts. Chem. Sci. 2018, 9, 7069–7077. https://doi.org/10.1039/c8sc01949e.(31) Qiao, L.; Liu, Y.; Zhu, J. A Focused Review on Machine Learning Aided High-Throughput Methods in High Entropy Alloy. J. Alloys Compd. 2021, 877, (160295). https://doi.org/10.1016/j.jallcom.2021.160295.(32) Jinnouchi, R.; Asahi, R. Predicting Catalytic Activity of Nanoparticles by a DFT-Aided Machine-Learning Algorithm. J. Phys. Chem. Lett. 2017, 8, 4279–4283. https://doi.org/10.1021/acs.jpclett.7b02010.(33) Zahrt, A. F.; Henle, J. J.; Rose, B. T.; Wang, Y.; Darrow, W. T.; Denmark, S. E. Prediction of Higher-Selectivity Catalysts by Computer-Driven Workflow and Machine Learning. Science. 2019, 363. https://doi.org/10.1126/science.aau5631.(34) Braconi, E.; Godineau, E. Bayesian Optimization as a Sustainable Strategy for Early-Stage Process Development? A Case Study of Cu-Catalyzed C-N Coupling of Sterically Hindered Pyrazines. ACS Sustainable Chem. Eng. 2023, 11, 10545–10554. https://doi.org/10.1021/acssuschemeng.3c02455.(35) Nugraha, A. S.; Lambard, G.; Na, J.; Hossain, M. S. A.; Asahi, T.; Chaikittisilp, W.; Yamauchi, Y. Mesoporous Trimetallic PtPdAu Alloy Films Toward Enhanced Electrocatalytic Activity in Methanol Oxidation: Unexpected Chemical Compositions Discovered by Bayesian Optimization. J. Mater. Chem. A. 2020, 8, 13532–13540. https://doi.org/10.1039/D0TA04096G.(36) Matsui, K.; Kanamori, K.; Toyoura, K.; Takeuchi, I. Introduction to Bayesian Optimization and Its Application to Material Engineering. Mater. Jpn. 2019, 58, 12–16. https://doi.org/10.2320/materia.58.12.(37) Wang, T.; Liang, J.; Zhao, Z.; Li, S.; Lu, G.; Xia, Z.; Wang, C.; Luo, J.; Han, J.; Ma, C.; Huang, Y.; Li, Q. Sub-6nm Fully Ordered L10. Adv. Energy Mater. 2019, 9, 1–9. https://doi.org/10.1002/aenm.201803771.(38) Kobayshi, T.; Chida, Y.; Todoroki, N.; Wadayama, T. Enhanced Oxygen Reduction Reaction Activity on the Melamine-Modified Pt-High-Entropy Alloy Single-Crystal Lattice Stacking Surface. ACS Catal. 2024, 14, 11512–11521. https://doi.org/10.1021/acscatal.4c02191.(39) Dieb, T. M.; Tsuda, K. Machine Learning-Based Experimental Design in Materials Science. In Nanoinformatics, Tanaka, I., Ed., Springer: Singapore, 2018. https://doi.org/10.1007/978-981-10-7617-6_4.(40) Toyao, T.; Maeno, Z.; Takakusagi, S.; Kamachi, T.; Takigawa, I.; Shimizu, K. Machine Learning for Catalysis Informatics: Recent Applications and Prospects. ACS Catal. 2020, 10, 2260–2297. https://doi.org/10.1021/acscatal.9b04186.(41) Lookman, T., Balachandran, P.V., Xue, D.; Yuan, R. Active learning in materials science with emphasis on adaptive sampling using uncertainties for targeted design. npj Comput. Mater. 2019, 5, 21, 1–16. https://doi.org/10.1038/s41524-019-0153-8.(42) Jones, D. R.; Schonlau, M.; Welch, W. J. Efficient Global Optimization of Expensive Black-Box Functions. J. Glob. Optim. 1998, 13, 455–492. https://doi.org/10.1023/A:1008306431147.(43) Snoek, J.; Larochelle, H.; Adams, R. Practical Bayesian Optimization of Machine Learning Algorithms. Adv. Neural Inf. Process. Syst. 2012, 25, 2951–2959. https://doi.org/10.48550/arXiv.1206.2944.(44) Ju, S.; Shiga, T.; Feng, L.; Hou, Z.; Tsuda, K.; Shiomi, J. Designing Nanostructures for Phonon Transport via Bayesian Optimization. Phys. Rev. X. 2017, 7, 021024. https://doi.org/10.1103/PhysRevX.7.021024.(45) Karadaghi, L. R.; Williamson, E. M.; To, A. T.; Forsberg, A. P.; Crans, K. D.; Perkins, C. L.; Hayden, S. C.; LiBretto, N. J.; Baddour, F. G.; Ruddy, D. A.; Malmstadt, N.; Habas, S. E.; Brutchey, R. L. Multivariate Bayesian Optimization of CoO Nanoparticles for CO2 Hydrogenation Catalysis. J. Am. Chem. Soc. 2024, 146, 14246–14259. https://doi.org/10.1021/jacs.4c03789.(46) Pedersen, J. K.; Clausen, C. M.; Krysiak, O. A.; Xiao, B.; Batchelor, T. A. A.; Löffler, T.; Mints, V. A.; Banko, L.; Arenz, M.; Savan, A.; Schuhmann, W.; Ludwig, A.; Rossmeisl, J. Bayesian Optimization of High-Entropy Alloy Compositions forElectrocatalytic Oxygen Reduction. Angew. Chem. Int. Ed. 2021, 60, 24144 – 24152.  https://doi.org/10.1002/anie.202108116(47) Mints V. A.; Pedersen, J. K.; Bagger, A.; Quinson, J.; Anker, A. S.; Jensen, K. M. Ø.; Rossmeisl, J.; Arenz, M. Exploring the Composition Space of High-Entropy Alloy Nanoparticles for the Electrocatalytic H2/CO Oxidation with Bayesian Optimization. ACS Catal. 2022, 12, 18, 11263–11271. https://doi.org/10.1021/acscatal.2c02563(48) Kusunoki, K.; Kudo, D.; Hayashi, K.; Chida, Y.; Todoroki, N.; Wadayama, T. Oxygen Reduction Reaction of Third Element-Modified Pt/Pd(111): Effect of Atomically Controlled Ir Locations on the Activity and Durability. ACS Catal. 2021, 11, 1554–1562. https://doi.org/10.1021/acscatal.0c04054.(49) Kudo, D.; Kaneko, S.; Myochi, R.; Chida, Y.; Todoroki, N.; Tanabe, T.; Wadayama, T. Ligand-Effect-Induced Oxygen Reduction Reaction Activity Enhancement for Pt/Zr/Pt(111) Surfaces with Tensile Strain Relieved by Stacking Faults. ACS Appl. Energy Mater. 2019, 2, 4597–4601. https://doi.org/10.1021/acsaem.9b00873.(50) Shinohara, K.; Ohma, A.; Iiyama, A.; Yoshida, T.; Daimaru, A. Membrane and catalyst performance targets for automotive fuel cells by FCCJ membrane, catalyst, MEA WG. ECS Meet. Abstr. 2011, MA2011-02, 1103–1103. https://doi.org/10.1149/MA2011-02/16/1103.(51) Motoyama, Y.; Tamura, R.; Yoshimi, K.; Terayama, K.; Ueno, T.; Tsuda, K. Bayesian Optimization Package: PHYSBO. Comput. Phys. Commun. 2022, 278, 108405. https://doi.org/10.1016/j.cpc.2022.108405.(52) Wadayama, T.; Todoroki, N.; Yamada, Y.; Sugawara, T.; Miyamoto, K.; Iijama, Y. Oxygen Reduction Reaction Activities of Ni/Pt(111) Model Catalysts Fabricated by Molecular Beam Epitaxy. Electrochem. Commun. 2010, 12, 1112–1115. https://doi.org/10.1016/j.elecom.2010.05.042.(53) Marković, N. M.; Adžić, R. R.; Cahan, B. D.; Yeager, E. B. Structural Effects in Electrocatalysis: Oxygen Reduction on Platinum Low Index Single-Crystal Surfaces in Perchloric Acid Solutions. J. Electroanal. Chem. 1994, 377, 249–259. https://doi.org/10.1016/0022-0728(94)03467-2.(54) Yamada, Y.; Miyamoto, K.; Hayashi, T.; Iijima, Y.; Todoroki, N.; Wadayama, T. Oxygen Reduction Reaction Activities for Pt-Enriched Co/Pt(111), Co/Pt(100), and Co/Pt(110) Model Catalyst Surfaces Prepared by Molecular Beam Epitaxy. Surf Sci 2013, 607, 54–60. https://doi.org/10.1016/j.susc.2012.08.016.(55) Asano, M.; Kawamura, R.; Sasakawa, R.; Todoroki, N.; Wadayama, T. Oxygen Reduction Reaction Activity for Strain-Controlled Pt-Based Model Alloy Catalysts: Surface Strains and Direct Electronic Effects Induced by Alloying Elements. ACS Catal. 2016, 6, 5285–5289. https://doi.org/10.1021/acscatal.6b01466.(56) Torihata, M.; Nakamura, M.; Todoroki, N.; Wadayama, T.; Hoshi, N. Activity for the ORR on Pt-Pd-Co Ternary Alloy Electrodes Is Markedly Affected by Surface Structure and Composition. Electrochem. Commun. 2021, 125, 107007. https://doi.org/10.1016/j.elecom.2021.107007.(57) Wakisaka, M.; Morishima, S.; Hyuga, Y.; Uchida, H.; Watanabe, M. Electrochemical Behavior of Pt-Co(111), (100) and (110) Alloy Single-Crystal Electrodes in 0.1 M HClO 4 and 0.05 M H2SO4 Solution as a Function of Co Content. Electrochem. Commun. 2012, 18, 55–57. https://doi.org/10.1016/j.elecom.2012.02.008.(58) Suzuki, A.; Nakamura, M.; Hoshi, N. Structural Effects of the Oxygen Reduction Reaction on the High Index Planes of Pt3Fe. Electrochem. Commun. 2022, 136. https://doi.org/10.1016/j.elecom.2022.107235.(59) Conway, B. E. Electrochemical Oxide Film Formation at Noble Metals as a Surface-Chemical Process. Prog. Surf. Sci. 1995, 49, 331–452. https://doi.org/10.1016/0079-6816(95)00040-6.(60) Khalakhan, I.; Vorokhta, M.; Václavů, M.; Šmíd, B.; Lavková, J.; Matolínová, I.; Fiala, R.; Tsud, N.; Skála, T.; Matolín, V. In-Situ Electrochemical Atomic Force Microscopy Study of Aging of Magnetron Sputtered Pt-Co Nanoalloy Thin Films during Accelerated Degradation Test. Electrochim. Acta. 2016, 211, 52–58. https://doi.org/10.1016/j.electacta.2016.06.035.(61) Zhang, J.; Yuan, Y.; Gao, L.; Zeng, G.; Li, M.; Huang, H. Stabilizing Pt-Based Electrocatalysts for Oxygen Reduction Reaction: Fundamental Understanding and Design Strategies. Adv. Mater. 2021, 33, e2006494. https://doi.org/10.1002/adma.202006494.(62) Yeh, J. W. Alloy Design Strategies and Future Trends in High-Entropy Alloys. JOM. 2013, 65, 1759–1771. https://doi.org/10.1007/s11837-013-0761-6.(63) Tsai, M. H.; Yeh, J. W.; Alloys, H. E. High-Entropy Alloys: A Critical Review. Mater. Res. Lett. 2014, 2, 107–123. https://doi.org/10.1080/21663831.2014.912690.(64) Amiri, A.; Shahbazian-Yassar, R. Recent Progress of High-Entropy Materials for Energy Storage and Conversion. J. Mater. Chem. A. 2021, 9, 782–823. https://doi.org/10.1039/D0TA09578H.(65) Miracle, D. B.; Senkov, O. N. A Critical Review of High Entropy Alloys and Related Concepts. Elsevier Ltd. 2017, 122, 448–511. https://doi.org/10.1016/j.actamat.2016.08.081.(66) de Boer, F. R.; Boom, R.; Mattens, W. C. M.; Miedema, A. R.; Niessen, A. K. Cohesion in Metals. Transition Metal Alloys; North-Holland, 1988.(67) Ye, Y. F.; Wang, Q.; Lu, J.; Liu, C. T.; Yang, Y. High-Entropy Alloy: Challenges and Prospects. Mater. Today. 2016, 19, 349–362. https://doi.org/10.1016/j.mattod.2015.11.026.(68) Ye, Y. F.; Wang, Q.; Lu, J.; Liu, C. T.; Yang, Y. The Generalized Thermodynamic Rule for Phase Selection in Multicomponent Alloys. Intermetallics. 2015, 59, 75–80. https://doi.org/10.1016/j.intermet.2014.12.011.(69) Čolić, V.; Bandarenka, A. S. Pt Alloy Electrocatalysts for the Oxygen Reduction Reaction: From Model Surfaces to Nanostructured Systems. ACS Catal. 2016, 6, 5378–5385. https://doi.org/10.1021/acscatal.6b00997.(70) Kusoglu, A.; Weber, A. Z. New Insights into Perfluorinated Sulfonic-Acid Ionomers. Chem. Rev. 2017, 117, 987–1104. https://doi.org/10.1021/acs.chemrev.6b00159.(71) Inaba, M.; Kinumoto, T.; Kiriake, M.; Umebayashi, R.; Tasaka, A.; Ogumi, Z. Gas Crossover and Membrane Degradation in Polymer Electrolyte Fuel Cells. Electrochim. Acta. 2006, 51, 5746–5753. https://doi.org/10.1016/j.electacta.2006.03.008.(72) Ledieu, J.; Feuerbacher, M.; Thomas, C.; de Weerd, M. C.; Šturm, S.; Podlogar, M.; Ghanbaja, J.; Migot, S.; Sicot, M.; Fournée, V. The (110) and (320) Surfaces of a Cantor Alloy. Acta Mater. 2021, 209, 116790. https://doi.org/10.1016/j.actamat.2021.116790.11image3.tiffimage4.tiffimage5.tiffimage6.tiffimage1.jpgimage2.tiff