Mechanistic Study of Electrooxidation of Ethanol on PtSn Nanoparticles in Alkaline and Acid Media

In this study, we report on the structure-to-property relationship for PtSn ethanol electrooxidation catalysts in alkaline and acidenvironment. Relationship between surface chemistry, as determined by XPS and electrochemical performance in both acid andalkaline media, is facilitated by multivariate analysis. Upon transitioning from acidic to an alkaline environment, changes in thematerial structure and electrochemical reaction mechanisms are observed. Electrocatalysts containing larger particles with largerrelative amounts of metallic Pt and Sn perform better in acid indicating inner-sphere electron transfer reaction on active PtSn alloyphase. PtSn electrocatalysts containing larger amounts of oxidized Pt and Sn perform better in alkaline indicating that hydroxylspecies that are natively present on oxidized Pt and Sn are promoting an outer-sphere electron transfer.© The Author(s) 2015. Published by ECS. This is an open access article distributed under the terms of the Creative CommonsAttribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in anymedium, provided the original work is properly cited. [DOI: 10.1149/2.0401506jes] All rights reserved.Manuscript submitted January 7, 2015; revised manuscript received February 4, 2015. Published February 27, 2015.

Direct hydrocarbon and direct alcohol fuel cells could find practical application due to the high energy densities of the fuels and the easier storage and delivery of liquid fuels. Direct ethanol fuel cells (DEFCs) are particularly attractive because ethanol can be generated from sustainable biomass derived feed stocks. There are several primary challenges that need to be overcome in order for DEFCs to see practical use. One of them is a relatively slow oxidation kinetics, which limits the power levels that can be achieved with the fuel cells. This is partly due to incomplete oxidation of ethanol to CO 2 , which effectively negates the high energy density of the fuel. Recent developments in OH − anion exchange membranes will allow for DEFCs to operate not only in acid media, where reaction kinetics is less favorable, but also under alkaline conditions where the oxidation kinetics is much faster.
Platinum is found to be active for the ethanol electrooxidation in both acid and alkaline media, however Pt favors incomplete oxidation forming acetic acid and acetaldehyde in acid media and acetate and aldehyde in alkaline solutions. 1 Complete oxidation of ethanol to CO 2 through the direct fission of the C-C bond at low overpotentials is the ultimate goal of direct ethanol fuel cells. 2 Much of the current research into direct ethanol fuel cells is focused on the development of platinum group metal (PGM) electrocatalysts that contain metallic promoters to assist in the complete oxidation of ethanol to carbon dioxide and resist poisoning. Bimetallic Pt-based electrocatalysts such as PtSn, PtRu, PtPd, PtW, [3][4][5] and ternary electrocatalysts such as PtRuW, PtRuMo, PtRuSn, 6 PtSnNi 7 and PtRhSnO 2 8 have been tested for promotion of ethanol oxidation in acid media. Among several promoters, Sn or SnO x were identified as the best element to enhance the ethanol electrooxidation on Pt in acid 9,10 and Pt and Pd in alkaline media. 11 DFT calculations of effect of alloyed metal on the catalytic activity of Pt for ethanol oxidation showed than among Ru, Sn, Re, Rd and Pd, only Sn enhances the rate constant of the dehydrogenation on the Pt site of Pt3Sn as compared with that on Pt alone. 12 Several factors are established to play a role in the performance of Pt-Sn systems, such as surface and bulk composition, structure, morphology, particle size, degree of alloying and the presence of surface oxides. It was found that in acid media, the electrocatalyst performance depends substantially on the atomic ratios of elements present, and the optimal composition for PtSn varies depending on the synthesis procedure. Lamy et al. 3 tested Pt:Sn compositions ranging from 9:1 to 4:1 and found that the addition of Sn always promoted the ethanol electrooxidation if compared to Pt, whereas the presence of 10-20 at% Sn resulted in the best resistance to catalyst poisoning or deactivation. Zhou et al. 9 observed the PtSn composition of 33-40 at% Sn to provide the best ethanol electrooxidation performance. The justification for the Pt:Sn ratio relates to the existing mechanism of the ethanol electrooxidation. The addition of Sn improves catalyst performance due to the bifunctional mechanism and/or the electronic effect. Well-ordered PtSn surface alloys supported on Pt (111) were shown to be particularly active toward ethanol electrooxidation in acidic media for structures where one Pt atom is in immediate contact to one or two surrounding Sn atoms. 13 It is well-established that in the bifunctional mechanism, the dissociative adsorption of water produces oxygen containing species (OH − ads ) on the promoter that can oxidize adsorbed carbon monoxide and other intermediates on Pt at lower potentials than on pure Pt. 5 The bifunctional mechanism can also facilitate the oxidation of acetaldehyde to acetic acid. 3 In the electronic effect, the added Sn decreases the bond strength between Pt and CO ads and/or other intermediates thus favoring ethanol adsorption. 14,15 The electronic effect may also cause water activation (OH ads ) on Pt at lower potentials. 16 An increase in the amount of surface Sn or Sn oxides may have a negative effect on PtSn performance since the adsorption and decomposition of ethanol occurs on Pt and contiguous Pt sites are needed to split the C-C bond. 17 Promoters may cover Pt sites and reduce the number of available Pt sites. 9 Although the electronic effect is beneficial in reducing the Pt-CO ads bond strength, excess electronic effect decreases the Pt-C bond strength and favors acetaldehyde formation. 18 Furthermore, the difference in surface energy between Sn (0.673 J/m 2 ) and Pt (1.95 J/m 2 ) may result in surface enrichment with Sn and hence a large difference between overall and surface composition. 19 The performance of high at. % Sn electrocatalysts can also be reduced due to the low electronic conductivity of SnO x .
Furthermore, degree of PtSn alloying and presence of surface oxides appear to have an influence on PtSn catalytic performance in acid electrolytes. Zhu et al. 15 showed that a large PtSn alloying degree promotes the ethanol oxidation reaction through the electronic effect while low alloying degree promotes ethanol oxidation through the bifunctional mechanism. Godoi et al. 20 found that catalytic activity of PtSn/C for ethanol oxidation is strongly influenced by changes in the amounts of Sn in alloyed and oxidized forms and that the increase in the amount of alloy at the expense of the oxides improves the catalytic activity.
Despite the large number of publication on the ethanol electrooxidation on PtSn catalysts, the evaluation of the role of various parameters, e.g., surface and bulk composition, role of alloyed phases and oxides, particle size effect on the catalytic enhancement, is not a straightforward task and remains to be elucidated. Therefore, in this study, we report on the structure-to-property relationship for PtSn nano-particles made by the polyol method that have been previously studied in alkaline and acid media. 21 Under electrochemical conditions in acid and alkaline solutions, solvation effects and surface adsorbed species play a critical role in rate-determining step of the ethanol electrooxidation. The chemical structure of the catalysts, i.e., distribution of metallic and oxidized (oxyphilic) within electrocatalysts structure will, thus, play a critical role in the mechanism of the electrooxidation. As was reported for ORR, among the many elementary reaction steps involved in ORR, there could be a surface-independent outer-sphere electron transfer component in the overall electrocatalytic inner-sphere electron transfer reaction. Upon transition from acidic to an alkaline environment, double layer structure and electrochemical reaction mechanisms may change. Specifically adsorbed hydroxyl species may promote an outersphere electron transfer mechanism in alkaline media. The amount of the adsorbed hydroxyls will depend significantly on surface composition of the catalyst particles.
In the present work X-ray photoelectron spectroscopy was used to study the surface chemistry of carbon-supported Pt 7 Sn 3 catalysts reported earlier. 21 The ability to discriminate between different chemical environments, not just elemental compositions, is one of the primary advantages of XPS in the characterization of catalysts. Correlation of XPS structural information with catalytic performance, particle size, and structural characteristics is accomplished by application of multivariate statistical methods of data analysis (MVA). MVA becomes of critical importance in structure-to-property relationship understanding. 22 Principal Component Analysis (PCA) is used herein as an analysis tools to facilitate visualization of the types of chemistries responsible for the better performance in acid versus alkaline media and to highlight differences between two categories of structures of catalysts obtained.

Experimental
Synthesis of the carbon-supported PtSn nanoparticles.-Synthesis of the carbon-supported PtSn nanocatalysts with the atomic ratio of Pt to Sn of 70:30 at% is described in details elsewhere. 21 In short: tin (II) chloride anhydrous (ACROS Organics, 98% Anhydrous) and Platinum (IV) chloride (Alfa Aesar, 99.9% metal basis, Pt (57.75%)) were used as precursor salts. Synthesis of PtSn nanoparticles was carried out in ethylene glycol (EG) solution (anhydrous 99.8% Sigma-Aldrich). First, precursor salts were dissolved in EG containing various concentration of NaOH (EM Science, ACS grade): 0.08, 0.1, 0.12, 0.15, 0.2 and 0.3 M. The solution was stirred for 1 hour at room temperature and then refluxed at 190 • C for 2 h. Appropriate aliquots of the resulting colloidal solutions were mixed with carbon black (Vulcan XC-72R, Cabot, corporation) in a large beaker for up to 24 h, resulting in the deposition of the colloids on the carbon substrates (20 wt% metal loading). The carbon supported PtSn catalysts were extensively washed with DI water (18 M cm) and then dried in the air at 80 • C for 3 h. No further pre-treatment was performed to the catalysts.
High resolution TEM.-HRTEM was performed on a JEOL 2010F FASTEM field emission gun scanning transmission electron microscope equipped with EDS. Probe size for high resolution work was 0.2 nm, 1.0 nm for analytical work, and accelerating voltage was 200 kV. Gatan Microscopy Suite Software with DigitalMicrograph was used for calculation of lattice spacing from HR-TEM images.
XRD.-The X-ray diffraction powder patterns were collected using a Bruker AXS D8 Advance system powder diffractometer, equipped with a Cu tube and a Vantec position-sensitive detector with radial Soller slits to reduce the background at low angles. The diffractograms were collected between 30 • and 100 • 2θ with a step of 0.0142 • 2h. The Pt7Sn3/C 20 wt% catalyst samples were deposited in a flat-surfaced single-crystal silicon wafer holder. The average crystallite sizes for the Pt7Sn3/C catalysts were evaluated using the full width at half maximum (FWHM) and full width at three-quarter maximum (3/4). The width of the Pt (220) peak was used to calculate the average crystallite size according to the Debye's formula as described in Ref. 21. The (220) peak was used for the analysis, because Vulcan XC-72 has reflections near 25 and 45 • 2θ, corresponding respectively to graphite reflections 002 and 101, which overlap with the main fcc (111) and (200)  Catalysts characterization.-Average crystallite size and structure of the carbon-supported PtSn catalysts were obtained from Xray diffraction patterns as described in details in Ref. 21. Carbonsupported Pt 7 Sn 3 catalysts were analyzed by Kratos Axis Ultra DLD X-ray Photoelectron Spectrometer. The XPS analysis was conducted at 140 W using monochromatic Al kα source and pass energy of 20 eV. The peak positions were corrected for sample charging by setting the maximum of C 1s peak to the binding energy of 284.7 eV. Data analysis and quantification were performed using CasaXPS software. A linear background subtraction was used for quantification of C 1s and O 1s and Shirley for quantification of Pt 4f and Sn 3d spectra. Sensitivity factors provided by the manufacturer were utilized. 70% Gaussian/30% Lorentzian line shape was used in the curve-fit. For species identification, Pt 4f spectra have been deconvoluted based on the constraints of equal spin-orbit splitting of 3.4 eV. FWHM was used based on that for Pt 4f spectrum from Pt foil.
Data analysis.-The curve-fit results from Pt 4f and Sn 3d spectra were combined with electrochemical surface area (ECSA) and normalized current density at 0.6 V vs. RHE (i) for both acid and alkaline media and average particle size found from XRD (d) into one table for PCA analyses. PCA (in PLS_Toolbox 5.0 23 in MATLAB) using autoscaling as data preprocessing method were applied. Autoscaling mean centers and scales data to unit variance making sure that all of the variables are equally weighted so that species present in very small amounts have the same importance as species present in large quantities. PCA extracts the key mathematical principal components (PC) from a large data matrix, by converting it into two smaller matrices that are easier to examine and interpret. The first PC accounts for the largest part of the variance in the data; the second PC accounts for the second-largest part of the variance, and so forth. The results of PCA are usually displayed as score plots (reflecting the significance of each sample in a principal component), loading plots (reflecting the significance of each variable in a principal component) and biplots (showing both samples and variables for two principal components). Biplots of PCA components will be used herein for analysis of correlation between variables as they provide a more instructive visualization of the clustering of samples for identification of variables that are the most or least important for a specific sample grouping. Correlated variables and samples will be located in the same regions on a biplot.

Results and Discussion
HRTEM and XRD.- Figure 1 shows HRTEM images representative of two types of PtSn catalysts -alloy and bi-phase. The lattice spacing between 0.1988 and 0.228 is assigned to Pt(1 1 1) or (2 0 0) planes. 24 For bi-phase sample #6, lattice spacing is within this range confirming presence of both phases. For alloy sample, however, elongation of spacing in lattice due to Pt alloying with Sn is observed. Figure 2 shows XRD patterns of PtSn/C catalysts # 3 and #5 that are representative of alloy and bi-phase catalysts. As can be seen from the figure, the catalysts show face-centered cubic (fcc) structure  Table I.  220) and (311), respectively. The (220) reflection was used to calculate the average crystallite size according to Debye formula for scattering by randomly oriented molecules as shown in Refs. 21 and 25, because this peak does not overlap with any diffraction peaks from carbon Vulcan XC-72. The 2θ position at maximum intensity (2θ max ) for reflection (220) was derived from a third order polynomial fit close to the top of the peak. 25 For PtSn nanoparticles #1-3, the 2θ max of (220) reflection is shifted to lower 2θ values (∼66.5 • 2θ), compared to the 220 position for bulk fcc Pt, (67.5 • 2θ) and PtSn nanoparticles #4-6 (67.2-67.4 • 2θ). As was shown in Ref. 21, nanoparticles with a disordered alloy structure of the nominal Pt:Sn ratio of 7:3 were formed for catalysts #1, 2 and 3. These samples correspond to the three smallest NaOH concentrations used for synthesis (Table I). For catalysts #4, 5 and 6, prepared using higher NaOH concentrations of respectively 0.15, 0.2 and 0.3 M, only partial alloying occurred and nanoparticles with a structure and composition close to that of pure Pt were formed. Table I summarizes the average particle sizes of the carbon-supported Pt 7 Sn 3 catalysts estimated by XRD.
Electrochemistry.-Ethanol electrooxidation on the six carbon supported Pt 7 Sn 3 catalysts were reported earlier in acid 21 and alkaline 26 solutions. Representative cyclic voltammograms and chronoamperograms in ethanol containing electrolyte is shown in Figure 3, while CO stripping in alkaline and acid solutions is shown in the Supporting Information (Figs 1S-2S). Several characteristics, such as electrochemical surface area (ECSA) found from CO stripping CVs and electrooxidation current densities (i) obtained from CV experiments in 1.0 M ethanol containing solutions are summarized in Table I. The table also shows average diameter of all six samples and the bulk structure found from XRD measurements 21 as well as XPS data. 27 XPS. -Table II shows atomic % and relative distribution of Pt and Sn types as determined by XPS. The stoichiometry of the sample depends on the concentration of the NaOH solution during synthesis. At higher % NaOH, the composition of bi-phases samples, as determined by XPS is much closer to stoichiometric Pt70Sn30 composition. Alloy samples, on the other hand, have Sn enrichment in comparison with the theoretical composition. As we will discuss below, the absolute composition, is not as important as chemical environment of Pt and Sn. Figure 4 shows Pt 4f and Sn 3d spectra representative for all samples. Pt 4f 7/2 part of the spectrum was curve fitted with four symmetrical peaks due to metallic Pt (71.6 eV), PtO (72.3 eV), Pt(OH) 2 (73.1 eV) and PtO 2 (74.2 eV). The peak at lower BE may also have contribution  from PtSn alloy. The Sn 3d 5/2 core level peaks were deconvoluted using two components. The much smaller peak at 485.9 eV was assigned to Sn metal and major peak at 487.4 eV to tin oxide SnO x in good correspondence with reference reported values. 28-30 The position of Sn(0) that is alloyed to a metal with different electronegativity may shifts toward higher values. The highest shift reported in the literature is 1.8 eV which is close to the position of 487.4 eV that we have identified as SnO x . 30 There may be small contribution of the Sn alloy into this peak for alloyed samples, but large contribution of this peak for bi-phase samples and prominent peak at the position of metallic  Sn (0) present in both alloyed and bi-phase samples make us believe that the major contribution is due to oxidized tin. Binding energies for SnO 2 and SnO are quite similar, so it is difficult to distinguish between these two species. We have combined all but metallic types of Pt as oxidized Pt (PtO + Pt(OH) 2 + PtO 2 ), Pt x O y , for correlations as shown in Table I, which shows the relative % of metallic Pt and Sn and relative % of oxidized Pt and Sn.
Bi-phase samples show a larger range of relative % of metallic Pt. Sample #6 with the largest average size shows largest relative amounts of both Pt and Sn metallic. We have reported before that relative distribution of types of Pt and Sn is more important than the absolute amounts. 31 Largest ECSA in both acid and alkaline media is present for samples #3 and 4 that have largest absolute amounts of Sn and hence largest absolute amounts of both metallic Sn and SnO. Largest density current in alkaline is observed for bi-phase samples that also have the largest amount of total Pt and hence largest amounts of both metallic and oxidized Pt. Therefore, correlating performance to the absolute amount of metallic and oxidized part of metal present will point to the same conclusions as total amounts of metal detected. Relative % of metallic and oxidized Pt and Sn were, therefore, combined for statistical structure-to-property correlation analysis.  Table I and Table II. Loading shows the relationship between variables and are coefficients of linear combinations of the original variable that generate the principal component. Scores describe the relationship between samples and contain coordinates of the original data in the new coordinate system of principal components. Biplots displaying both loadings for each variable and scores for each sample for the PC1 and PC2 produce the way to study clustering of the samples with respect to variables, such as amounts of certain chemical species and electrochemical parameters. Correlated variables and samples are located within the same half or quadrant of the biplot. Figure 5 shows PCA biplot calculated from data in Table I and  Table II combined which shows both samples (diamonds) represented as scores and variables (circles and stars) represented as loadings. The major difference between samples properties such as surface chemistry and electrochemical behavior is captured by the principal component #1, which represents ∼57% of the variance in the data. Positively contributing variables into PC1 are total amounts of Pt and Sn detected, ECSA in both media, density current in alkaline and relative % of Pt oxide and Sn oxide detected. Negatively contributing variables are density current in acid, the relative amount of metallic Pt and Sn and size of particles as determined by XRD. There is no sharp separation between alloy and biphase samples in terms of better performance in acid vs. alkaline media. The overall trend, however, is  Tables I and II showing loadings (variables) and scores (samples). Catalysts synthesized at different NaOH concentrations marked as diamonds.
) unless CC License in place (see abstract  that biphase composition shows better performance in alkaline than alloy samples, 26 while alloy samples show better performance in acid than in alkaline.
Direct correlations between the overall amount of both metals present and relative distribution of types of metals are shown in Figures 6 and 7. Considering heterogeneous nature of these materials, these relationships between chemistry, as determined by XPS, and performance in acid and alkaline, as manifested by ECSA measures, are evaluated for presence of any positive or negative trend pointing toward higher/lower amounts of certain species present contributing toward particular type of electrochemical behavior.
ECSA values obtained by CO stripping, show that in both acid and alkaline solutions the same types of chemical species contribute to the removal of CO. The higher the total amounts of Sn in the sample, as detected by XPS, the larger ECSA (Figure 6a and 6b) in both media. This dependence is a little stronger in alkaline media than in acidic. In alkaline media, there is also slightly more pronounced dependence of ECSA on the total amount of Pt present. There is no significant dependence between the relative amount of Pt in oxidized environment and ECSA (Figure 6c). However, stronger dependence of ECSA is observed as a function of the relative amount of SnO ( Figure 6). It indicates that the presence of Sn that is oxyphilic and, therefore, present as mainly Sn oxide at the surface of the particle enhances the number of active sites. It is known that CO adsorbs on Pt and does not adsorb on Sn atoms, therefore the increase of ECSA with the higher amount of SnO x may be explained by the formation of additional active sites, PtSnO x neighboring sites that are active toward the reaction.
Striking differences are observed when one looks into dependence of the intrinsic performance of these catalysts in both environments as a function of surface structure as determined by XPS. Figure 7 plots the most notable dependencies observed. Relative amounts of both Pt and Sn metallic have high positive contribution into current densities in acid while these dependencies are reversed in alkaline, i.e. the larger relative amount of metallic Pt and Sn, which is also correlated with larger particle size, the poorer the performances observed. The opposite is observed when one looks into the performance dependence on relative amounts of oxidized part, i.e. Pt x O y and SnO. Higher current densities are detected in alkaline media for larger amounts of oxides of both metals that are present for smaller particles. The dependence of performance on relative speciation of Pt is quite stronger than on that of Sn.
To summarize, in acid media the major contributions into ECSA are from oxidized part of the material while intrinsic performance mainly depends on a metallic part of the material. In alkaline media, the major dependencies for ECSA and current densities are the same, i.e. the larger total amount of metals, the better performance, and larger ECSA. The contributions into electrochemical parameters are the same, in contrast with acid media, i.e., larger current densities and surface areas come from relative amounts of oxidized part of PtSn catalysts, not the metallic one. Such difference in contribution of metallic versus oxidized part of both metals in intrinsic activity may be explained by different mechanism of electron transfer in acid and alkaline. It is suggested that the mechanism of the ethanol electrooxidation in acid is via inner-sphere electron transfer while the mechanism of the ethanol electrooxidation in alkaline media within outer-sphere plane. We suggest that in acid, metallic Pt and Sn are the active phases for molecular adsorption of reactant species necessary for electron transfer to occur. The trend observed for the acidic environment, shows that higher activity is indeed observed for materials with higher relative amount of both Pt and Sn. The rate determining step on PtSn in acid is believed to be the first electron transfer step to the adsorbed reagents with or without rapid proton transfer. The trend observed in an alkaline environment, shows the opposite behavior -higher activities are observed for samples with smaller amounts of metallic and higher degree of oxidized metals. We suggest that a surface-independent outer-sphere electron transfer is facilitated by larger amounts of hydroxyl groups that are native to oxidized Pt and Sn.