Underlying Substrate Effect on Electrochemical Activity for Hydrogen Evolution Reaction with Low‐Platinum‐Loaded Catalysts

Platinum is known as the best catalyst for the hydrogen evolution reaction (HER) but the scarcity and high cost of Pt limit its widespread applicability. Herein, the role of the underlying substrate on the HER activity of dispersed Pt atoms is uncovered. A direct current magnetron sputtering technique is utilized to deposit transition metal (TM) thin films of W, Ti, and Ta as underlying substrates for extremely low loading of Pt (<1.5 at%). The electrocatalytic performance of as‐synthesized samples for the HER is examined in both alkali and acidic media. The results show that despite the low loading of Pt, the Pt/TM catalysts produce hydrogen at a rate comparable to that of pristine bulk Pt. Pt/TM catalysts also display good stability with less than 5% decay in performance after 10 h of continuous HER operation. Based on the computational study, the excellent performance is attributed to the modified electronic properties of the Pt atoms, offering ideal binding energy for HER due to interaction with the underlying substrates. This work provides a robust and industry‐friendly route toward designing efficient catalytic systems for important electrochemical reactions such as HER and others.


Introduction
Hydrogen (H 2 ) production varies widely around the world, but most of the H 2 is produced through steam methane reforming, where natural gas is reacted with steam to produce H 2 .However, this process is energy intensive; it relies on natural gas as a feedstock and generates carbon dioxide (CO 2 ) as a byproduct, contributing to greenhouse gas emissions and climate change.Despite these drawbacks, steam methane reforming remains the most economically viable method for producing large quantities of H 2 due to its low cost and high efficiency.Ongoing research and development of alternative methods of H 2 production include approaches based on photo-and electrochemical hydrogen evolution reaction (HER).The electrochemical HER is currently considered a promising method for sustainable Platinum is known as the best catalyst for the hydrogen evolution reaction (HER) but the scarcity and high cost of Pt limit its widespread applicability.Herein, the role of the underlying substrate on the HER activity of dispersed Pt atoms is uncovered.A direct current magnetron sputtering technique is utilized to deposit transition metal (TM) thin films of W, Ti, and Ta as underlying substrates for extremely low loading of Pt (<1.5 at%).The electrocatalytic performance of assynthesized samples for the HER is examined in both alkali and acidic media.The results show that despite the low loading of Pt, the Pt/TM catalysts produce hydrogen at a rate comparable to that of pristine bulk Pt.Pt/TM catalysts also display good stability with less than 5% decay in performance after 10 h of continuous HER operation.Based on the computational study, the excellent performance is attributed to the modified electronic properties of the Pt atoms, offering ideal binding energy for HER due to interaction with the underlying substrates.This work provides a robust and industry-friendly route toward designing efficient catalytic systems for important electrochemical reactions such as HER and others.
hydrogen production, especially as renewable energy sources become more widely adopted.Moreover, hydrogen is promising for grid energy storage at large scale and for powering energy-dense transport applications in comparison to advanced batteries. [1,2]Effective catalysis will play a vital role in bringing hydrogen via the HER process to the mainstream of the world economy.The choice of catalyst for HER depends on various factors, such as cost, efficiency, stability, and environmental impact.
Among metal catalysts, platinum (Pt) is considered the most effective catalyst for HER due to its high intrinsic catalytic activity, but it is also among the most expensive. [3][16] Therefore, Pt nanoclusters with a certain size (of %30 atoms) have shown higher HER activity in comparison to smaller or larger nanoclusters. [17][20][21][22] Moreover, it has been observed that Pt-based multi-metallic nanocatalysts exhibit superior HER activities due to modified electronic (e.g., d-band center, oxidation state) properties and morphological (e.g., porosity, interfacial structure, and dimensionality) effects and decoration of surface steps. [12,13,23]To achieve higher HER atomic efficiency via efficient utilization of the catalytic sites, Pt atoms or nanoclusters can be dispersed on a substrate.In our recent study, we have established that Pt deposited on Mo film exhibits HER performance comparable to bulk Pt in both acidic and basic media. [24]ased on the X-Ray photoelectron spectroscopy (XPS) analysis, it has been suggested that Pt atoms interact with underlying Mo atoms and work synergistically to promote HER.In another study, Huang et al. found that Pt NPs with sizes of 1-3 nm on the MoS 2 nanosheets exhibit HER activity comparable to that of the bulk Pt-C catalysts despite %70% lower Pt loading; this effect was attributed to cocatalytic activities that stem from the Pt interaction with the substrate. [25]Here it should be noted that electronic properties (e.g., electronic configuration and electronegativity) of the substrate atoms can significantly alter the catalytic activity of the Pt atoms either via modifying electronic state or due to induction effect. [22,26]However, it is not experimentally clear how the variation in the electronic properties of underlying substrate atoms will affect the atomic efficiency and thus overall catalytic activity of the designed system.
In this study, we reveal the role of underlying substrate electronic properties on the catalytic behavior of dispersed Pt nanostructure by performing controlled experiments and computational analysis of the properties of Pt dispersed on different substrate materials.We have selected titanium (Ti), tantalum (Ta), and tungsten (W) as substrates due to their diverse periodic properties, such as electron negativity, electron distribution, atomic size, and other characteristics (Table S1, Supporting Information).The catalysts were prepared using a chemical-free, scalable, and reproducible magnetron sputtering technique in inert media (Figure 1A,B).Our results provide evidence that Pt dispersed on W (Pt d -W) catalyst has superior performance in comparison to Pt dispersed on Ti (Pt d -Ti), Ta (Pt d -Ta), and bulk Pt regardless of the nature of the electrolyte (either acidic or basic).Based on XPS and theoretical analyses, this behavior has been attributed to the alteration in electronic properties of the Pt and synergistic effect.

Morphological and Elemental Characterization of Electrocatalysts
Scanning electron microscopy (SEM) images (Figure 1C-H) suggest the formation of closely packed random nano-/ microparticles on degenerately doped Si substrate.However, the film roughness remains between 10 and 20 nm, as measured by the height variations in profilometer experiments (Figure S1, Supporting Information).The thickness of the deposited films varied from 85 to 130 nm (Figure S1, Supporting Information).The SEM images clearly show variations in the size of nano-/microparticles.Here it should be noted that during the sputtering process, highly energized argon ions (Ar þ ) strike the target; then ablated ionized target atoms adhere to the substrate due to electrostatic forces.Therefore, ionization energy plays an important role in controlling the quality (e.g., roughness) of the deposited film under similar experimental conditions, such as applied power and chamber pressure.The ionization energy of metallic polycrystalline Ti, Ta, W, and Pt varies in the order Ti (658.81 kJ mol À1 ) < Ta (728.42 kJ mol À1 ) < W (758.76 kJ mol À1 ) < Pt (864.39 kJ mol À1 ).Lower ionization energies result in a more efficient formation of ions and may thus lead to larger sputtered nano-/microparticle formation.This is consistent with larger particles observed in the Ti films in comparison to the other sputtered materials (Figure 1C-E).However, other physical parameters such as density of crystallization sites, and crystal formation rate, would also play a critical role in defining the size of observed particles.After thin-film deposition (30 min), Pt atoms were sputtered only for 10 s creating submonolayer (ML) or nanoislands.Due to the low density of deposited Pt nanoislands and relatively high roughness of the Ti, Ta, and W substrates, Pt nanostructures were not readily resolvable in the SEM images (Figure 1F-H) and corresponding energy dispersive spectroscopy (EDS) analysis (Figure S2-S4, Table S2-S4, Supporting Information).Notably, the morphology of the films remains unchanged (Figure 1F-H).
To investigate the morphological and electronic nature of the dispersed Pt atoms, we also performed transmission electron microscopy (TEM) characterization (see Figure 2A).The TEM samples were prepared by sputtering Pt atoms directly on the TEM grids to mimic the Pt/TM film deposition conditions.Interestingly, the TEM imaging suggests the formation of homogeneously distributed Pt nanoislands with an average diameter of %3 nm.However, it was not feasible to visualize these Pt nanoislands on W and other thick films due to insufficient intensity of electron scattering.Additionally, the small size and uniform deposition of Pt nanoislands limit the characterization of their sizes and shapes in SEM experiments and, therefore, the dependence of their structural characteristics on the supporting  surface.That being said, earlier experimental studies of bimetallic systems suggest that Pt deposited at subML coverages wets the W(110) surface in the absence of oxygen and forms ML-thick islands. [27,28]These observations are consistent with the results of our ab initio simulations that sub-ML Pt forms a continuation of the body-centered-cubic (BCC) lattice (i.e., a pseudomorphic layer) on the (110) surfaces of W, Mo, and Ta, and that this layer is stable relative to the bulk Pt (Figure S13, Supporting Information).Specifically, we compared the stabilities of face-centered cubic (FCC) bulk-like two-and three-layer Pt clusters supported on W(110) and showed that they transform into such pseudomorphic layers with the energy gain of 1.3-1.7 eV per each Pt atom displaced to the interface with W(110) (Figure S13a-c, Supporting Information).The higher calculated stability of the ML clusters is consistent with findings of a seminal study of Au clusters on MgO(001) and thin MgO films supported on Mo(100) surface, [20] which shows that the propensity of the Au clusters to stabilize in the form quasi-2D increases with increasing proximity to the Fermi pool of the subsurface metal and is associated with the electron transfer between the metal substrate and the cluster.
We also performed XPS characterization to understand the effect of substrates on the electronic states of the Pt nanoislands (Figure 2B-D, S5-S7 and Table S5-S7, Supporting Information).Figure 2 shows the high-resolution deconvoluted Pt 4f spectra for Pt d -Ti, Pt d -Ta, and Pt d -W catalysts.All systems exhibit peaks attributed mostly to Pt 0 with some Pt 2þ , suggesting the presence of both metallic Pt and some oxidized Pt atoms. [29,30]We observed slight variations in the Pt 0 peak positions for different samples.Specifically, Pt 4f 5/2 and Pt 4f 7/2 peaks were found at 74.0 and 70.7 eV for Pt d -Ti, at 74.1 and 70.8 eV for Pt d -Ta, and at 74.6 and 71.2 eV for Pt d -W, respectively, indicating substrate effect on the electronic structure of the supported Pt clusters. [31]However, the separation in the 4f peaks remains constant, that is, 3.33 eV except slightly increased for Pt d -W (3.34 eV).The Pt 2þ 4f 7/2 and Pt 2þ 4f 5/2 characteristic peaks appeared at 71.9 and 74.9 eV for Pt d -Ti, at 72.5 and 75.5 eV for Pt d -Ta and at 72.1 and 75.6 eV for Pt d -W, respectively.
[32][33] According to our calculations (see Figure 2E), oxygen atoms bind strongly to both W(110) and Ta(110) surfaces even though their binding energies (E b ) decrease by %0.5 eV with increasing O coverage.In the presence of ML-thick Pt islands (see Figure S14, Supporting Information), the average binding energies of the O species at the exposed W(110) and Ta(110) surfaces decrease slowly with increasing Pt coverage.In contrast, the adsorption of O species on the ML of Pt on both W(110) and Ta(110) is much weaker (Figure 2E), which suggests that in the regime of sub-ML Pt coverage, residual O species are predominantly captured by the exposed substrate surfaces.Furthermore, since the Ta-O interaction is stronger than the W-O interaction, one can expect that a larger fraction of the O species is bound to Ta than to W atoms and; accordingly, a smaller fraction of O atoms interacts with Pt/Ta than with Pt/W.This conclusion is consistent with a smaller Pt 2þ peak intensity observed for Pt 4f in the case of Pt d -Ta than in Pt d -W (Figure 2C,D).As reported previously, the appearance of Pt 2þ species correlates with the higher HER activity, [34] which suggests that the sputtered Pt d -TM systems can be promising HER catalysts.We ought to note that the role of partial oxidation of metal catalysts in chemical conversions is still to be understood as discussed in a recent perspective that highlights key challenges of working with these materials. [35]o examine the role of substrates on the HER activity, we performed linear sweep voltammetry (LSV) experiments (ohmic resistance corrected) in both acidic (0.5 M H 2 SO 4 ) and basic (1 M NaOH) aqueous electrolytes (Figure 3A,B, S8, Supporting Information).Hydrogen as a single product was confirmed by online differential electrochemical mass spectroscopy (DEMS) (Figure S9 and S10, Supporting Information).[38][39] Thus, we compare overpotentials at 10 mA cm À2 for all catalysts (Figure 3).The TM films alone, that is, before Pt deposition, exhibit relatively high overpotentials in acidic media with 10 mA cm À2 current density obtained at 835 mV (Ti), 975 mV (Ta), and 384 mV (W).However, the presence of Pt significantly reduces the overpotential (at 10 mA cm À2 ) for all substrates (Figure 3A) to 283, 541, and 48 mV for Pt d -Ti, Pt d -Ta, and Pt d -W, respectively.We note that the overpotential for Pt d -W (48 mV) is even lower than that of bulk Pt (commercially available nanowire) (94 mV) in acidic media examined under similar experimental conditions.The results show that the overpotential of Pt d -W is approximately 2, 6, 11, 17, and 20 times lower than those measured for the bulk Pt, Pt d -Ti, Pt d -Ta, Ti, and Ta catalysts, respectively.41] Therefore, higher overpotentials are expected for all catalysts.In our case, 10 mA cm À2 current density was recorded at 0.94, 1.13, 0.83, and 0.37 V for Ti, Ta, W, and Pt catalysts, respectively.Notably, the same order of catalytic performance was observed for the metals in acidic conditions.Also, for the Pt/TM catalysts, the overpotentials at 10 mA cm À2 decrease to 0.16, 0.  S8, Supporting Information).The Tafel slopes for the bulk Pt (42.5 mV dec À1 ) and Pt d -W (43.1 mV dec À1 ) catalysts were approximately identical, suggesting the similar Volmer-Tafel mechanism, where recombination is the rate-limiting step. [42]In general, the Pt nanocatalysts (e.g., Pt NPs mixed with carbon) studied using the rotating disc electrode (RDE) method have relatively high surface area and fast kinetics.In our case, we have used bulk Pt wire as a working electrode without RDE.Therefore, we have observed a slightly higher Tafel slope (%42 mV dec À1 ) in comparison to the other Pt/C catalysts (%30-40 mV dec À1 ) studied using RDE.In the case of Pt nanocluster-decorated TiO 2 , nanotubes, and commercial 20 wt% Pt/C examined under similar experimental conditions without RDE, have also shown a Tafel slop of 37 and 42 mV dec À1 respectively. [43]The calculated exchange current density for bulk Pt (0.93 mA cm À2 ) and Pt d -W (0.91 mA cm À2 ) is identical and approximately two orders of magnitude higher in comparison to the other examined catalysts in similar conditions, thus further evidencing their superior intrinsic catalytic activity for HER (Table S8, Supporting Information).
In basic media, the Volmer step ( ) is considered as the rate-limiting step. [44]Ti, Ta, W, and bulk Pt exhibit a Tafel slope of 171.64, 179.87, 318.9, and 111.54 mV dec À1 , respectively (Table S9, Supporting Information).The high magnitude of the Tafel slopes indicates sluggish Volmer step kinetics for HER.However, in the case of Pt d -W, we observed a very low Tafel slope (65.00 mV dec À1 ), suggesting significantly enhanced HER activity.This value is comparable to those observed for some of the most advanced catalysts: NiO/Pt (50 mV dec À1 ) and 20% Pt/C (50.6 mV dec À1 ). [44]The Tafel slope for Pt d -Ti (79.03 mV dec À1 ) is also relatively low, but the exchange current density of Pt d -W is approximately one order of magnitude higher in comparison to the Pt d -Ti catalyst (Table S9, Supporting Information).The performance of the catalysts was also compared by plotting the Tafel slope versus overpotential (at 10 mA cm À2 ).Pt d -W catalyst exhibited the lowest Tafel slope and lowest overpotential in both acidic and basic environments (Figure 3E and S11, Supporting Information).
Considering the high catalytic activity of the Pt d -W catalyst, we also performed a long-term stability test for it in acidic media at À0.08 V versus reversible hydrogen electrode (RHE) (Figure 4A-C).In standard experimental conditions, the performance of the Pt d -W catalyst sharply decreases within the first hour and a 50% lower current (%5 mA cm À2 ) has been observed after 10 h of operation (Figure 4A).Similar behavior has been observed for atomically dispersed Pt and Pt nanocluster-based catalysts. [11,12]he decay in performance can be attributed to the deactivation of active sites due to the adsorption of ionic impurities, however, it requires further study to identify the origin of the deactivation of the catalysts.This effect can be reversed via an anodization process, that is, applying positive potential for short periods (10 s in our case), which results in restoring the catalyst properties via eliminating the adsorption-induced poisoning effect.Accordingly, we applied anodization potential (1.1 V vs. RHE) for 60 s after every 10 min and continued the stability test after that (Figure S12, Supporting Information).Interestingly, the Pt d -W catalyst shows remarkable stability with an almost unchanged current density even after 10 h of operation period, as manifested by the recorded LSV before and after 10 h.After this test, the Pt d -W catalyst was stored in ambient conditions for 90 days and re-examined for HER activity in an acidic environment.These tests yield similar LSV characteristics, which suggest the high durability of the Pt d -W catalyst (Figure 4C).Finally, the catalytic activity of Pt d -W was compared with those of nine Pt-based catalysts reported in the literature by comparing Tafel slope and overpotential at 10 mA cm À2 current density (Figure 4D).Despite ultralow loading and easy catalyst preparation technique, the Pt d -W catalyst exhibits performance comparable to or better than some of the current state-of-the-art catalysts (e.g., Pt-2H MoS 2 , Pt-1T MoS 2 , and ep-WS 2 -Pt).

Discussion
To obtain the atomic-scale insight into the high and low efficiency of the H 2 production in these systems, we turn to ab initio simulations and examine how energies of H and OH adsorption on surfaces of these crystals partially capped with Pt MLs depend on the local atomic structure.To isolate the effects of the lattice strain and charge redistribution, we perform this analysis for Pt supported on the isostructural W and Ta surfaces only.[47] We purposefully omitted the Pt d -Ti system from this analysis because Ti metal has a different structure (hexagonal closed packed (HCP)).
Given the relatively small amount of Pt deposited on the TM surfaces, we considered four structural models corresponding to Pt coverage of 0.02 (i.e., one atom per lateral supercell), 1/3, 2/3, and one full ML for each substrate.In the cases of 1/3 and 2/3 ML Pt coverages, the Pt atoms were arranged to create an island and a pit in a Pt ML, respectively (Figure 5, S17, Supporting Information).The shapes of these structures were selected so that they exhibit characteristic structural elements, including terraces, edges, corners, and inverse corners; this choice allows us to quantify the interactions of H and OH with characteristic low-coordinated sites and downselect the most catalytically relevant cluster features.While their structures were not a result of a global minima search, we found it instructive to assess their stability relative to other possible structures.This assessment is provided in Figure S13e (Supporting Information), where we plot the adsorption energies (per Pt atom) of the Pt 14 island configuration and several randomly selected configurations of 14 Pt atoms as a function of the average number of the nearest Pt-Pt neighbors.The overall trend suggests that the Pt structures become more stable as the number of nearest neighbors increases, that is, the Pt structures are expected to exhibit steps oriented along the (111) directions of the supporting BCC W lattice.Our calculations suggest that up to 1 ML, Pt atoms form a continuation of the BCC lattice in all cases; however, the interaction of Pt with W and Mo surfaces is different from that with Ta.Specifically, as shown in Figure S13d (Supporting Information), the binding of Pt to the W(110) and Mo(110) surfaces becomes stronger with increasing Pt coverages, which suggests that Pt atoms deposited on these surfaces tend to aggregate into larger clusters.In contrast, Pt binds to Ta(110) most strongly in the limit of dispersed atoms (0.02 ML in our case).At higher Pt coverages, the average adsorption energy decreases considerably, which suggests that Pt tends to remain atomically dispersed or forms small clusters on the Ta(110) surface.
We note that since the interaction of residual oxygen with Pt MLs supported on W(110) and Ta(110) is much weaker than that with TM surfaces (Figure 2E), that is, in the regime of sub-ML Pt coverages considered here, residual oxygen is predominantly trapped at the exposed TM surfaces.Furthermore, since Pt catalysts tend to be more reduced under HER conditions than they appear in XPS characterization after synthesis, [48] we expect that the Pt clusters themselves and their vicinity are oxygen free during the HER measurements.Under these conditions, Pt clusters transform into a pseudomorphic configuration forming quasi-2D single-layer thick islands, which is the model we adopted below.
At sub-ML coverages, Pt exhibits a range of strain states depending on the proximity to the cluster edges (Figure S15, Supporting Information).The differences between the local atomic structure of the outermost Pt plane in the Pt(111) surface and Pt/TM(110) (TM = W, Mo, Ta) systems can be illustrated by the comparison of the Pt-Pt distances between the nearest-and second-nearest neighbors within the Pt layer as well as their ratio t (Figure S15, Supporting Information).These structural variations suggest that locations of the H adsorption sites and the corresponding binding energies on Pt/TM(110) differ from those on Pt(111) and can vary depending on the details of the local atomic structure.We quantified the extent of deformations in the supported Pt clusters in terms of the radial distributions of the Pt-Pt distances (Figure S15, Supporting Information).These distributions were calculated for the models of a Pt island (Pt 14 % 1/3 ML Pt coverage), a pit in the Pt ML (Pt 34 % 2/3 ML Pt coverage), and a full Pt ML on both W(110) and Ta(110) substrates.The distribution peak broadening obtained for the 1/3 ML and 2/3 ML models suggests significant variations in the Pt local atomic structure at sub-ML Pt coverage (Figure S15, Supporting Information).
To examine the effect of the local atomic structure on the Pt chemical reactivity, we calculated the H atom binding energies for %20 sites, including patches of the W(110) and Pt terraces, step edges, corners, and inverse corners (see Figure S17, Supporting Information).Our calculations show that E b varies from %0 to 0.8 eV depending on the H atom location.H binds most strongly at the exposed regions of W(110) (E b % 0.7-0.8eV).In the other limit, H atoms located on terraces of Pt islands (Figure 5) are weakly bound, which suggests that they are likely to migrate toward the island edge and stabilize there, on either Pt or W planes.The calculated E b for H adsorbed near the step edges fall in the range of 0.25 AE 0.05 eV.We note that, in addition to the hydrogen adsorption energy (defined as -E b ), Gibbs free energy (ΔG) corresponding to the H þ H ! H 2 reaction includes contributions due to differences in zero-point energy (ΔE ZPE ) and vibrational entropy (TΔS) between the adsorbed and gasphase hydrogen. [49,50]We adopt the value of þ0.24 eV for these contributions. [50]With this correction, the estimated values of ΔG for the step edges in Pt/W(110) are close to zero, which correspond to optimal conditions for the HER.The effect of the W(110) and Mo(110) substrates on H adsorption energies is quantitatively similar.This is consistent with an earlier study of the HER catalytic activity of Pt d -Mo that reported overpotential values similar to those found here for Pt d -W. [24]In contrast, H interaction with Pt/Ta(110) is skewed by stronger binding to the Ta(110) substrate and weaker binding to the Pt terraces supported by Ta(110), which is consistent with the lower activity of the Pt d -Ta catalysis.
To rationalize this observation, we analyze the charge distribution at the Pt/TM(110) interface (Figure 5D).Pt islands draw the electron charge from the substrate and become negatively charged.Since H adsorbed on Pt also tends to be slightly negatively charged, the reduced Pt layer weakens H binding.In contrast, H adsorbed on the exposed TM(110) patches draws up to 0.6e and binds to both W and Ta with energies of 0.8 and 1 eV, respectively.This correlation between the H binding energy and the amount of trapped charge is captured in Figure 5D.
To assess the behavior of the Pt d -W and Pt d -Ta catalysts in the alkaline environments, we compared the energy change in the H 2 O !H þ OH reaction for the reaction products located: 1) on the W(110) and Ta(110) surfaces far from Pt, 2) on the surface in the vicinity of the Pt supported clusters, and 3) on top of these clusters, both near the Pt cluster edge and at their terraces.These calculations were conducted on the same footing as for the H binding energies and compared with the energies of H 2 O decomposition on the Pt(111) surface.Selected configurations of the adsorbed H and OH are shown in Figure S17 (Supporting Information), and the corresponding binding energies are shown in Figure 5.
To discuss the results of these calculations, we turn to a 3D volcano plot reported recent study. [51]This study suggests that if OH binding to the catalyst surface was %0.95 eV stronger than on Pt(111) and, simultaneously, H binding to the same catalyst was 0.25 eV weaker, the maximum of HER rates would have been achieved.Our calculations (see Figure 5B,C) suggest that, indeed, adsorption of ½ H 2 near step edges of Pt/W(110) is %0.2 eV weaker than that on Pt(111), while adsorption of H þ OH at the same sites is %0.8 eV stronger than on Pt(111).From this, we deduce that the adsorption of OH alone at these sites is %1 eV stronger than that on Pt(111).Hence, assuming that entropic contributions to the free energies are comparable, the binding energies of H and OH on Pt/W(110) are close to those predicted to maximize HER rates. [51]We find the situation is very different for Pt/Ta(110): the binding energies of both H and OH are weaker than on Pt/W(110) resulting in lower HER activity.
Finally, we note that while the parameter t is a convenient measure of the Pt cluster distortions, its direct effect on the H adsorption energy may not be straightforward to establish.This is because deformations are coupled with the charge redistribution between Pt and its substrates.We deconvoluted the effects of local structural variations, electron charge redistribution between the substrate and the supported Pt clusters, and substrate-induced deformation; we used the following approach (see Figure 6).
To isolate the effect of the Pt structural variations, we compared the H binding energies calculated for the fully relaxed configurations and configurations where Pt atoms were fixed at the substrate (110) surface sites.These structural changes are accompanied by changes in the Pt atomic charges by as much as 10% (i.e., by %0.05 e).As shown in Figure 6, their average effect on H adsorption energies is %0.1 eV for W(110) and appears to be negligible for Ta(110).In turn, to isolate the effect of the electron transfer from the W and Ta substrates to the Pt clusters, we substituted all W and Ta atoms with Pt atoms and recalculated the H binding energies.In these calculations, the lateral positions of all metal atoms were fixed at the corresponding BCC sites.Since the systems are monometallic, the substrateto-Pt electron transfer is suppressed, while the strain associated with the substrate symmetry and lattice parameters remains unaffected.As a result, the H binding energies have increased significantly (Figure 6) and more so for the case of Ta than W, which indicates a stronger effect of substrate-to-Pt electron transfer and is consistent with the average Pt charges: -0.46 |e| and -0.64 |e| for Pt on W(110) and Ta(110), respectively.This comparison shows that the more negative the surface Pt, the lower is H binding energy, which, in turn, allows one to define the bounds for electronegativities of the substrates.Finally, a comparison with the binding energy of H adsorbed at FCC hollow sites of the Pt(111) surface suggests that BCC-like deformation of the Pt(111) surface increases H binding energy, that is, has the effect opposite to the substrate-Pt charge transfer.

Conclusion
In summary, we have designed Pt dispersed catalysts (Pt d -W) consisting of Pt nanoclusters dispersed on W film via magnetron sputtering.The designed catalyst had an overpotential of 47 mV at a current density of 10 mA cm À2 , a Tafel slope of 43 mV dec À1 , and stability for greater than 10 h in acidic media, exhibiting competitive performance to the state-of-the-art Pt-based catalyst in both acidic and basic media despite a low loading of Pt.This significantly improved performance can be attributed to the origin of a Pt active site with a 2þ oxidation state due to interaction with underlying substrate atoms, resulting in a synergistic effect by creating an edge site with a more favorable binding energy.In this work, we present a simple and robust approach to guiding the Pt-based catalyst design with excellent electrochemical properties.This approach can be applied to design compositions of the TM substrate that provide optimal H binding energies across a wide range of structural motifs, thus paving the way toward the development of new-generation catalysts with industrial-friendly techniques.

Experimental Section
Chemical Reagents: Sodium hydroxide (NaOH) pellets were purchased from Millipore Corp. Sulfuric acid (H 2 SO 4 , 0.5 M) acetone, ethanol, and buffer oxide etchant (BOE) were purchased from VWR chemicals.Pt, W, Ti, and Ta metal targets, with purity of 99.95% and a diameter of 2 inches, were obtained from Kurt J. Leskar company.
Thin-Film Deposition: Pt dispersed on metal (W, Ti, Ta, and Cu) thin-film samples was prepared using the DC magnetron sputtering technique.Initially, degenerately doped p-type Si substrates were cleaned by sonicating sequentially in acetone, ethanol, and deionized (DI) water for 5 min and then immersed in BOE for 2 min.Then, we washed the substrate with DI water and dried it with nitrogen gas.Metal thin film (W, Ti, and Ta) was deposited on cleaned p-type Si substrate using commercially available metal targets, (Kurt J. Leskar) with purity of 99.95% and a diameter of 2 inch as a sputtering target.The deposition chamber was evacuated to achieve a base pressure of about 10 À8 Torr.The deposition was carried out for an hour at room temperature with an applied DC power supply of 100 W under 10 mTorr ultrahigh-pure argon gas (99.999%) atmosphere.During deposition, rotating degenerately doped p-type Si (0.001-0.005Ohm-cm) substrate was placed 25 cm away from the target to achieve uniform deposition.After depositing metal (W, Ti, and Ta) thin film, under the same pressure and temperature conditions, Pt (Kurt J. Leskar, 99.99% purity) was sputtered on top of the TM (W, Ti, and Ta) thin film for only 10 s with an applied DC supply of 30 W to acquire Pt-dispersed TM (W, Ti, and Ta) electrocatalyst thin film on a silicon substrate.
Material Characterization: The morphological structure and elemental composition of the samples were characterized using XPS and SEM techniques.XPS data was acquired using SPECS XPS with 300 W Mg-anode (hν = 1253.6eV) X-Ray source.Initially, the survey spectrum was scanned in the range of 0-1100 eV binding energy with a 0.5 eV step size to identify all the elements present on the electrocatalyst thin film.Subsequently, the high-resolution regional spectra of C 1s, O 1s, Pt 4f, W 4f, Ti 2p, and Ta 4f with 0.1 eV step size were recorded.Deconvolution of peaks from XPS spectra was performed using Casa XPS software.The binding energies were calibrated with the adventitious carbon C 1s peak position at 285 eV.Surface morphologies of electrocatalyst thin-film samples were imaged using JEOL JSM-6060LV and JEOL JSM-6010PLUS/LA SEM.Chemical analysis was carried out using a Thermo Scientific Ultradry EDS detector attached to the SEM.
Electrochemical Measurements for HER Activity: A typical twocompartment three-electrode electrochemical H-cell was utilized to perform the electrochemical characterizations using a BioLogic potentiostat SP-300.In this study, as-prepared electrocatalyst thin-film devices were used as working electrodes, Pt wire was used as counter electrode, and Ag/AgCl saturated in KCl was used as reference electrodes.An anion-exchange membrane was used to separate the cathode and anode compartments.Hydrogen evolution kinetics were studied by performing LSV measurements by sweeping the potential essential to achieve the 10 mA cm À2 current density with the scan rate of 20 mV s À1 .[54][55] All potentials were converted to the RHE using the Nernst equation.
Potential versus RHE ¼ applied potential versus Ag=AgCl The in situ differential electrochemical mass spectrometry (DEMS) was utilized to detect hydrogen.During the LSV measurement, the evolved hydrogen at the cathode was transferred into the mass spectrometry device (Hyden Analytical HPR-40).An electron energy of 70 eV was used for ionization of all species, with an emission current of 500 μA.The massto-charge ratio (m z À1 ) of 2 was selected for detection of the hydrogen generation rate.
Computational Modeling: The Pt/TM(110) system (where TM = W, Mo, Ta) was represented using the periodic slab model.The lateral supercell of the slab terminated with the (110) surface of the BCC lattice is formed by the 4 Â 6 extension of the a 1 = (1,À1,0) and a 2 = (0,0,1) vectors.The out-of-plane vector was a 3 = (1,1,0).The numerical values of the lateral supercell parameters were obtained from the corresponding bulk lattice parameters calculated for the crystallographic cells using k-mesh of 14 Â 14 Â 14 points.Accordingly, the sizes of the supercells used in this work were 18.02 Â 19.10 Å 2 for W(110) and 18.78 Â 19.92 Å 2 for Ta (110).For comparison, we performed selected calculations for the Mo(110) surface as well; the corresponding lattice parameters were 17.87 Â 18.97 Å 2 .The Pt/W slab contained three atomic planes of W with an additional partial or full ML of Pt.The out-of-plane supercell parameter was set to 23 Å, which left a vacuum gap of over 15 Å.Binding energies of oxygen and hydrogen atoms on the Pt/W(110) surfaces were calculated relative to the half of the corresponding gas-phase (O 2 and H 2 ), molecules, respectively: positive if dissociative adsorption is favorable.A similar approach was used to calculate the binding energies of H þ OH resulting from H 2 O decomposition.Atoms in the W plane furthest from the Pt/W interface were fixed at the sites corresponding to the ideal bulk lattice.The total energy of each system was minimized with respect to the internal coordinates of all other atoms.[58] Projector-augmented wave potentials were used to approximate the effect of the core electrons. [59]To determine the k-mesh, we compared total energies of Pt/ W(110) systems for 1/3, 2/3, and 1 ML Pt coverages as well as average Pt binding energies for k-meshes 1 Â 1 Â 1, 2 Â 2 Â 1, and 3 Â 3 Â 1 (Figure S18, Supporting Information).We also compared binding energies of H atoms on the W(110) surface partially covered with Pt atoms and found that the values calculated using Γ-point only and 2 Â 2 Â 1 k-mesh were nearly identical (Figure S18, Supporting Information).Therefore, all subsequent calculations were conducted using Γ-point only.The planewave basis-set cutoff was set to 500 eV.The total energy convergence criterion was set to 10 À5 eV.The effects of spin polarization were found to be negligible (see Supplementary Information section).Therefore, all calculations were performed in the spin-unpolarized mode.Methfessel-Paxton smearing (ISMEAR = 1; SIGMA = 0.2) was used in all calculations.

Figure 1 .
Figure 1.Catalyst preparation and morphological analysis: A) Schematic of HER process on the bulk Pt and on Pt/TM catalysts, and B) schematic of magnetron sputtering technique employed for catalyst deposition.The SEM images of C) Ti, D) Ta, E) W, F) Pt d -Ti, G) Pt d -Ta, and H) Pt d -W.The scale bar is 200 nm in each case.

Figure 2 .
Figure 2. Characterization of the Pt/TM catalysts: A) TEM of Pt sputtered on Cu TEM grid.The average size of Pt nanoclusters was %3 nm (black dots).High-resolution XPS spectra of B) Pt d -Ti, C) Pt d -Ta, and D) Pt d -W showing Pt 0 and Pt 2þ oxidation states of the Pt atoms.E) Calculated average binding energies of oxygen species as a function of Pt coverage.
79, and 0.15 V for Pt d -Ti, Pt d -Ta, and Pt d -W catalysts, respectively, that is, under basic environment, Pt d -W exhibits 1.1, 5.4, and 2.6 times lower overpotential in comparison to the Pt d -Ti, Pt d -Ta, and Pt catalysts.Depending on the reaction environment (acidic or basic) and intrinsic properties of the catalysts, the HER consists of either Tafel/Volmer or Heyrovsky/Volmer mechanistic steps.The Tafel equation, η = b*log( j) þ a, where η is the overpotential, b is the Tafel slope, j is the current density, and a is a constant, was used to analyze the HER mechanism.In acidic media, the calculated Tafel slopes further evidence that deposition of Pt significantly improves the HER activity as decreased Tafel slopes are observed for Pt-based catalysts [Pt d -Ti (89.3 mV dec À1 ), Pt d -Ta (149.5 mV dec À1 ), and Pt d -W (43.1 mV dec À1 )] in comparison to catalysts without Pt [Ti (173.9 mV dec À1 ), Ta (182.2 mV dec À1 ), and W (101.1 mV dec À1 )] (Table

Figure 3 .
Figure 3.The HER electrocatalytic performance of the Pt, TM, and Pt/TM catalysts: LSV curves at 20 mV s À1 scan rate (with iR correction) in A) acidic (0.5 M H 2 SO 4 ) and B) basic media (1 M NaOH).C) Tafel slopes in acidic media.D) Tafel slopes in basic media.Comparison of examined catalyst in E) acidic media.

Figure 4 .
Figure 4. Stability test of Pt d -W and comparison with other catalysts: A) CA test in acidic (0.5 M H 2 SO 4 ) media in standard conditions and with anodization (10 s) process.B) LSV before and after 10 h stability test in both acidic and alkaline environments.C) LSVs recorded on first day and after 90 days of exposure to the environment.D) Comparison with Pt-based state-of-the-art catalyst in acidic media.

Figure 5 .
Figure 5. A) Selected configurations of H atoms (magenta) adsorbed in the vicinity of Pt clusters (blue) supported on TM(110) (TM = W, Ta).An extended set of configurations for both adsorbed H and OH is provided in Figure S17 (Supporting Information).B,C) Binding energies of H atoms (B) and dissociated water molecules, as H þ OH (C), on the Pt/W(110) and Pt/Ta(110) systems.D,E) Correlations of the H and H þ OH binding energies and charge trapped by these species.These correlations suggest that "starving" H and OH of the electron charge using Pt buffer layer can be exploited to tune their binding energies.Insets show the schematics of charge distribution in respective systems.

Figure 6 .
Figure 6.Binding energies of H atoms adsorbed on Pt/W(110) and Pt/Ta(110) calculated for the fully relaxed systems (black), Pt fixed at the BCC substrates in-plane lattice sites (red), and the substrate W and Ta atoms replaced with Pt atoms (green).Dashed line corresponds to H binding energy on the hollow FCC site on the Pt(111) surface.Insets show the qualitative charge distribution for the Pt/TM (weaker binding) and model Pt/BCC-Pt (stronger binding) systems.See text for details.