Suppressing H2 Evolution and Promoting Selective CO2 Electroreduction to CO at Low Overpotentials by Alloying Au with Pd

S1 Alloying Au with Pd suppresses H2 evolution and promotes selective CO2 electroreduction to CO at low overpotentials Supporting Information Marco Valenti*1, Nitin P. Prasad1, Recep Kas1, Divya Bohra1, Ming Ma1, Vignesh Balasubramanian1, Liangyong Chu1, Sixto Gimenez 2, Juan Bisquert 2, Bernard Dam1 & Wilson A. Smith*1 1Materials for Energy Conversion and Storage (MECS), Department of Chemical Engineering, Faculty of Applied Sciences, Delft University of Technology, Delft 2628-BL, The Netherlands 2Photovoltaic and Optoelectronics Devices Group, Departament de Fisica, Universitat Jaume I Av Sos Baynat s/n, 12071, Castello, Spain. SI-1. Surface Roughness The surface roughness of the Au-Pd thin films deposited on quartz substrates were characterized by atomic force microscopy (AFM). The images are shown in Figure 1S. The average roughness and the RMS are tabulated in Table 1S. Figure 1S. AFM images for Pd-Au thin film alloys of composition (a) Pd (b) Au25Pd75 (c) Au50Pd50 (d) Au75Pd25 (e) Au Composition RMS roughness (nm) Average roughness (nm) Au 1.127 0.872 Au75Pd25 1.362 1.060 Au50Pd50 1.443 1.105 Au25Pd75 1.486 1.157 Pd 0.878 0.656 Table 1S. Average roughness and RMS roughness of the surface as calculated using the atomic force microscopy (AFM) S2 SI-2. Valence band spectra from XPS before and after etching The valence band spectra of all the electrodes were taken before and after 29 cycles of etching using an ion beam of 500 eV with the Ta2O5 estimated sputter rate of 0.02 nm/s. Each etch cycle lasted for 1 minute. The results are shown in Figure 2S and 3S Figure 2S. XPS spectra of valence band of the sputtered surface (black solid) and the surface after 29 cycles of ion beam etching (red dotted) for Au-Pd thin films alloys with composition (a) Au75Pd25 (b) Au50Pd50 (c) Au25Pd75 prior to CO reduction. Figure 3S. XPS spectra of valence band of the sputtered surface (black solid) and the surface after 29 cycles of ion beam etching (red dotted) for Au-Pd thin films alloys with composition (a) Au75Pd25 (b) Au50Pd50 (c) Au25Pd75 after CO2 reduction. The surface after etching for 30 minutes can be considered as bulk from the fact that the carbon peak which was prominent before the start of the etching process almost disappeared after etching (Figure 4S). It must be also noted that the etching process does involving some mixing of components also. Figure 4S. XPS spectra of C species after each cycle of etching using ion beam for Au25Pd75. The carbon peak is found to reduce with etching confirming that the etching process takes places. The peak almost disappears by 29 cycles of etching. Increasing cycles of etching S3 SI-3. Surface composition using X-Ray Photoelectron spectroscopy (XPS) The surface composition of the thin film alloys were calculated using both Au 4d3/2 – Pd 3d3/2 and Au 4p3/2 – Pd 3d3/2. The calculations were done for the electrodes before CO2 reduction and after CO2 reduction at -0.5 V. The results are shown in Table 2S and Table 3S. Expected Composition Using Au 4d 3/2 and Pd 3d 3/2 Using Au 4p 3/2 and Pd 3d 3/2 75% 77.66 % 75.57 % 50% 56.19 % 51.58 % 25% 19.14 % 16.37 % Table 2S Composition of the Au-Pd thin film alloys before CO2 reduction as calculated from XPS survey spectra. All values show % of Au content. Expected Composition Using Au 4d 3/2 and Pd 3d 3/2 Using Au 4p 3/2 and Pd 3d 3/2 75% 78.90 % 75.01 % 50% 56.15 % 47.59 % 25% 21.01 % 16.85 % Table 3S Composition of the Au-Pd thin film alloys after CO2 reduction at -0.5 V as calculated from XPS survey spectra. All values show % of Au content. SI-4. Electrochemical Reduction of CO2 All electrochemical experiments were carried out in a custom-made electrochemical cell with two compartments – working electrode compartment and the counter electrode compartment. The two compartments were separated by a Nafion 115 proton exchange membrane. Pt plate was used as the counter electrode and saturated Ag/AgCl was used as reference electrode. 0.1 M KHCO3 (99.95%, Sigma Aldrich) solution was filled as electrolyte in both compartments. Prior to electrolysis, CO2 was bubbled through the electrolyte in both compartments for 20 minutes at a flow rate of 15 ml min-1. All electrolysis experiments were conducted at room temperature (25˚C) and 1 atm with pH of the electrolyte at 6.8. A potentiostat (Parstat 4000, Princeton Applied Research) was used to apply the potential versus the reference electrode which was later converted to RHE. The potentials applied were manually corrected for the ohmic drop. CO2 was continuously purged throughout the experiment in both the electrolyte compartments. A magnetic stirrer rotating at 960 rpm was used in the working electrode compartment in order to effectively remove the gaseous products from the liquid phase. The gases from the cell were led into the sampling loop of a gas chromatograph (CompactGC 4.0, Global Analyser Solutions) with a run time of 3min 16s. Each electrolysis experiment was performed for 20 minutes and 5 GC injections were taken. The average of all 5 readings of the GC are used to calculate the FE which is reported. Due to similar roughness, the ECSA of the Pd electrode was used to calculate the current density of each of the electrodes (see supporting information SI-7) Prior to performing CO2 reduction experiments, chronopotentiometry was performed on Au in 0.1 M KHCO3 without bubbling CO2 . Since no CO2 was bubbled through the electrolyte, only hydrogen is expected to be produced. The gases produced during electrolysis are periodically analyzed using gas chromatography (CompactGC 4.0, Global Analyser Solutions). The analysis confirms that only hydrogen is produced, and the faradaic efficiency calculations add up ~ 100% as shown in Figure 5S. The mean and standard deviation of the faradaic efficiencies are also estimated. The GC takes one reading for stable faradaic efficiency and concentration and hence atleast 3 readings are performed at each potential. S4 Figure 5S. Faradaic efficiencies and concentrations of hydrogen detected by the GC during the chronopotentiometry of pure Au thin film in 0.1 M KHCO3 without bubbling CO2 at -3.5 mA. The mean faradaic efficiency and its standard deviation are indicated in the figure. The flow rate of 15 ml min-1 and stirring rate of 960 rpm are used for all the CO2 reduction experiments. All electrochemical measurements were performed in a custom made electrochemical cell with two compartments (working and counter electrode compartments). The compartments were separated by a proton exchange membrane, Nafion 115. A three electrode configuration was used for all experiments with a Pt plate as counter electrode and saturated Ag/AgCl electrode as reference electrode. Both compartments were filled with 0.1 M KHCO3 (99.95%, Sigma Aldrich). Prior to CO2 reduction experiments, both compartments were purged with CO2 at constant flow rate of 15 ml/min. The electrolyte was bubbled for 20 min to make it saturated with CO2 (pH = 6.8). Chronoamperometry was started immediately after 20 min and the flow rate of CO2 was kept constant throughout. Five GC injections were taken for each potential and the electrolysis was performed for ~20 minutes. The gaseous products were analysed using gas chromatography (CompactGC 4.0, Global Analyser Solutions). All mesurements were done at room temperature and atmospheric pressures. All potentials were applied versus saturated Ag/AgCl and converted to reversible hydrogen electrode (RHE) using the following equation. 𝐸𝑅𝐻𝐸 =𝐸𝐴𝑔/𝐴𝑔𝐶𝑙 +0.197+0.0591×𝑝𝐻 SI-5. Hydrogen adsorption/absorption calculations The charge per unit area required for the adsorption of a monolayer of hydrogen on Pd can be calculated using the amount of palladium per unit area. Each Pd atom is expected to bind to one H atom and the process involves a one electron (n). Charge per unit area for H adsorption (Q) =𝑁𝑞𝑛= 1.526×1015 ×1.602×10−19 ×1=244 µ𝐶 𝑐𝑚−2 Although the amount of charges calculated correspond to the adsorption of H on pure Pd, it might be different for alloys. However, it is reasonable to assume that charges of the same order of magnitude or lesser are indicative of adsorption while charges much greater than this is an indication of absorption. Since all the alloys were prepared by magnetron sputtering and the RMS roughness calculated from AFM show similar values, the area of all the electrodes is assumed to be the same. Taking a suitable baseline, the area of the peaks corresponding to the hydrogen desorption are calculated from the respective voltammograms as shown in Figure 6S. S5 Figure 6S. Calculated areas for Au-Pd films in CO2 saturated 0.1 M KHCO3 corresponding to the hydrogen desorption peak in the cyclic voltammograms with cycling upto -0.4 V. Using the calculated area and the scan rate, the total charges corresponding to the hydrogen desorption can be estimated. The total charges obtained for the different alloys and pure Pd are shown in Table 4S. Composition Total charge (µC cm-2) Au75Pd25 130 Au50Pd50 1091 Au25Pd75 3222 Pd 1885 Table 4S. Total charges corresponding to the hydrogen absorption/adsorption peak of Au-Pd alloys The charges corresponding to the hydrogen desorption peak for Pd is larger than the value calculated for adsorption of hydrogen on Pd. The larger charges could be attributed to the absorption of hydrogen by Pd. It is known that Pd absorbs hydrogen which finds a direct application in diffusion membranes to purify H2. The total charges for Au50Pd50 and Au25Pd75 are also larger than 244 µC cm-2 indicating that in these alloys also, hydrogen absorption occurs. The charge for hydrogen desorption in the Au75Pd25 alloy is much smaller than that calculated for hydrogen adsorption. This is a clear indication of the absence of hydrogen absorption. SI-6. Cycling voltammetry of adsorbed intermediates The relationship between the hydrogen absorption and the binding of CO2RR intermediates was studied using cyclic voltammetry of 25% Au electrode in CO2 saturated 0.1 M KHCO3. Initially, two scans were performed from 0.4 RHE to -0.9 RHE with the cathodic sweep first. The first scan showed a cathodic peak for the adsorbed intermediates as shown in Figure 7S. S6 Figure 7S. Cyclic voltammetry of 25% Au electrode in 0.1 M KHCO3 from 0.4 V to -0.9 V at scan rate of 0.05 V s-1 The hydrogen absorption/adsorption region (0 to 0.4 V) was cycled to observe if hydrogen was adsorbed/absorbed or desorbed. The current in the nA range shown in Figure 8S confirm that no process takes place in this region. Figure 8S. Cyclic voltammetry of 25% Au electrode in 0.1 M KHCO3 from 0 V to 0.4 V at scan rate of 0.05 V s-1 On extending the potential range for the scan to 1.6 V, a sharp peak was observed ~1.2 V which corresponds to CO desorption. Another cycle in the same range leads to the appearance of a peak at 0.3 V as shown Figure 9S which corresponds to that of hydrogen desorption. This might lead to the conclusion that hydrogen desorption does not occur as the electrode surface is covered with CO2. However, it must be noted that in Figure 9S, a broad peak at ~0.9 V overlaps with the sharp peak at ~1.2 V. The broad peak at 0.9 V in scan 1 probably corresponds to hydrogen desorption. The late appearance of this peak in this scan suggests that hydrogen desorption still occurs but is made more difficult by the adsorbed intermediates. Once these are removed, hydrogen desorption peak shifts back to its origin position. Thus it may be concluded that the hydrogen absorbed finds it difficult to escape out from the electrode when it is covered by the CO2RR intermediates. S7 Figure 9S. Cyclic voltammetry of 25% Au electrode in 0.1 M KHCO3 from 0 V to 1.6 V at scan rate of 0.05 V s-1 SI-7. Electrochemical Surface Area (ECSA) calculations The electrochemical surface area (ECSA) for the electrodes were calculated using the cyclic voltammogram of pure Pd. Pd forms a face centred cubic (FCC) crystal structure. The arrangement of atoms in a (111) crystal plane of a typical FCC crystal is shown in Figure 10S. Figure 10S. Arrangement of atoms in a typical FCC crystal. Number of atoms in the (111) plane =1 2×3+1 6×3=2 atoms Lattice parameter for Pd (a) = 389 pm = 3.89 × 10-8 cm Number of atoms of Pd per unit area (N) =2 √3 2𝑎2=2 √3 2(3.89 × 10−8 )2= 1.526×1015 atoms/cm2. Since the reduction peak corresponds to PdO is being reduced to Pd, two electrons (n=2) are involved in the process. The charge of one electron (q) is 1.602×10-19 C. Total charges per unit area for reduction =𝑁𝑞𝑛=1.526×1015 ×1.602×10−19×2=488 µ𝐶 𝑐𝑚−2 The CV of pure Pd in the absence of CO2 is taken to calculate the total charge corresponding to the reduction of PdO to Pd as shown in Figure 11S. A scan rate of 0.05 V s-1 was used for the cyclic voltammetry. S8 Figure 11S. Area of the reduction peak of Pd in 0.1 M KHCO3 without CO2 bubbling Total charges from CV =𝐴𝑟𝑒𝑎 𝑜𝑓 𝑡ℎ𝑒 𝑝𝑒𝑎𝑘 𝑆𝑐𝑎𝑛 𝑟𝑎𝑡𝑒 = 0.000100782 0.05 =2015 µ𝐶 ECSA of the thin films = 𝑇𝑜𝑡𝑎𝑙 𝑐ℎ𝑎𝑟𝑔𝑒𝑠 𝑓𝑟𝑜𝑚 𝐶𝑉 𝑇𝑜𝑡𝑎𝑙 𝑐ℎ𝑎𝑟𝑔𝑒𝑠 𝑝𝑒𝑟 𝑢𝑛𝑖𝑡 𝑎𝑟𝑒𝑎 𝑓𝑜𝑟 𝑟𝑒𝑑𝑢𝑐𝑡𝑖𝑜𝑛 = 2015 488 =4.12 𝑐𝑚−2 SI-8. X-Ray Diffraction (XRD) X-ray diffraction was used for the characterization of all the Au-Pd alloys. The lattice parameter for all the compositions were calculated using the following equation. 𝑎= √3 𝜆𝑐𝑜𝑏𝑎𝑙𝑡 2sin𝜃 where θ is the angle between the incident X-ray and the scattering planes and λcobalt is 0.178897 nm. XRD patterns of all the alloys are shown in Figure 12S and the lattice parameters are tabulated in Table 5S. The lattice parameters of Au and Pd closely match with the reported values. S9 Figure 12S. X-ray diffraction (XRD) patterns of pure Au, pure Pd and Au-Pd thin film alloys (FCC, 50 nm thickness) of different compositions using Co-Kα (λ=0.178897 nm). The peak is found to shift gradually on increasingly adding Pd to Au. Composition Lattice parameter (A0) Au 4.071464 Au75Pd25 4.043865 Au50Pd50 4.002828 Au25Pd75 3.948924 Pd 3.884979 Table 5S. Lattice parameter as calculated using XRD for all compositions of Au-Pd alloys Figure 13S Lattice parameter as a function of the composition. The lattice expands almost linearly with composition thus verifying Vegard’s law.