Supporting Information. Active and Stable Core-Shell Catalysts for. Electrochemical Oxygen Reduction

Transcription

1 Supporting Information Active and Stable Core-Shell Catalysts for Electrochemical Oxygen Reduction Alaina L. Strickler, Ariel Jackson, and Thomas F. Jaramillo* Department of Chemical Engineering, Stanford University, Stanford, California 94305, United States SUNCAT Center for Interface Science and Catalysis, SLAC National Accelerator Laboratory, Menlo Park, California 94025, United States * 1

2 1. Experimental Details 1.1 Nanoparticle Synthesis In a typical synthesis, Ir cores were first formed by combining 0.05 mmol sodium hexachloroiridate hexahydrate (99.9+%, Acros Organics) with 40 ml ethylene glycol (99+%, Acros) and 56 mg polyvinylpyrrolidone (55,000 g/mol MW, Sigma-Aldrich) in a two-neck 100 ml round bottom flask. Monodisperse Ir nanoparticles were formed by quickly heating the solution to reflux (~196 o C) and reacting for 3 hours under vigorous stirring. After cooling to room temperature, a portion of the Ir nanoparticles was removed for material and electrochemical characterization. An appropriate amount of 100 mm chloroplatinic acid hydrate (99.9+%, Sigma- Aldrich) in ethylene glycol was then added, and the solution was slowly heated (1-2 o C/min) to reflux and reacted for 1.5 hours to form the Pt shells. Pt-only nanoparticles were formed using an identical method as described for Ir-only particles with chloroplatinic acid hydrate as the Pt precursor. 1.2 Catalyst Ink Preparation Catalyst inks were prepared by combining a certain amount of synthesis solution (typically 1.4 ml for a 2:1 molar Pt:Ir) with carbon black (Vulcan XC-72) to achieve 20 wt% metal. The supported particles were then washed 3 times by adding acetone, centrifuging at 10,000 rpm for 30 min, redispersing in a water and ethanol mixture, and sonicating for 20 min. Supported particles were dried in a box furnace at 100 o C. Catalyst inks were made by adding 3 ml water, 2 ml isopropanol, and 20 µl Nafion 117 solution (~5% in a mixture of lower aliphatic alcohols and water, Aldrich) to the dried supported particles and sonicating for 1 hour. 1.3 Material Characterization Samples were prepared for transmission electron microscopy (TEM) imaging by dropcasting the nanoparticle solution onto ultrathin carbon on holey carbon support 400-mesh copper TEM grids (Ted Pella). A FEI Tecnai operated at 200 kv was used to obtain TEM micrographs. EDS spectra were obtained in STEM mode using an EDAX SUTW (super ultra thin window) and analyzer with 0.3 srad EDS solid angle. Extended EDS point scans were performed by alternating between signal collection and drift correction until sufficient signal was collected to resolve the Ir and Pt peaks. 2

3 Particle size distributions and EDS spectra were analyzed using ImageJ and Tecnai Imaging Analysis (TIA) software, respectively. EDS scans were smoothed using the adjacent averaging function with 20 points in Origin Pro 9.1 software. XPS characterization was carried out on a PHI versaprobe with a monochromatized Al(kα) source. Peaks were fitted using CasaXPS software. 1.4 Electrochemical Characterization Catalyst samples were prepared by a method described previously. 1 Briefly, the catalyst ink (5-20 µl) was dropcast onto a polished glassy carbon disk (0.196 cm 2 ) attached to an inverted rotating disk electrode (RDE, Pine Research Instrumentation) and spun at 700 rpm until dry. Electrochemical characterization was performed using a RDE, 3-electrode set up with a Pt wire counter electrode, a custom built reversible hydrogen reference electrode (RHE), and 0.1 M perchloric acid electrolyte. The series resistance of the cell was measured at 100 khz and IR loss was compensated at 85%. Typical resistances were around 22 Ω. To remove capping ligands from the surface, the synthesized catalysts were electrochemically cleaned prior to ORR testing by performing 300 potential cycles between 0.05 and 1.1 V vs. RHE at 500 mv/s in nitrogen saturated electrolyte. After electrochemical cleaning, the electrolyte was replaced. Electrochemical cleaning was not performed on the commercial Pt catalyst as this was found to decrease its electrochemically active surface area (ECSA) and ORR activity. Cyclic voltammograms (CVs) were performed at 20 mv/s and 1600 rpm with baseline and ORR CVs collected in nitrogen and oxygen saturated electrolyte, respectively. Anodic ORR sweeps are corrected for the background current by subtracting the anodic sweep baseline. 1.5 Mass and specific activity Mass and specific activities of the catalysts were calculated by first extracting the kinetic current by removing mass transfer effects using the experimentally measured mass transfer limited current. The mass loading for Ir@Pt catalysts was calculated from the concentration of Pt in the synthesis solution. It thus represents the lower bound of mass activity as it ignores loss mechanisms during synthesis (yield) and ink preparation. The specific activity was determined from the ECSA. This was found by integrating the double layer corrected charge in the HUPD region between V vs. RHE. The average of the cathodic and anodic charge was converted to surface area using 210 µc/cm 2 Pt. 2 All reported activities are measured at 0.9 V vs. RHE and are compared to commercial 3

4 Pt/C (46.6 wt% Pt on high surface area carbon from TKK). The shift in the OH*/O* desorption was found by determining the minimum of the desorption peak of the baseline voltammograms using a parabolic fit. 1.6 Electrochemical stability tests Prior to stability testing, Ir@Pt particles were electrochemical conditioned by cycling between V at 500 mv/s for 25 cycles and electrochemically cleaned for 75 cycles under conditions described above. 3 Stability cycles were performed under oxygen saturation without rotation and the potential was cycled between V vs. RHE at 125 mv/s. Before measuring ORR activity, the catalyst was electrochemically cleaned and the electrolyte was replaced. 2. Further Analysis 2.1 STEM-EDS STEM energy dispersive x-ray spectroscopy (STEM-EDS) was used to determine spatial compositions of the particles and verify the core-shell structure. Due to the significant overlap of the Pt and Ir peaks, the small particle size, and the substantial drift that occurs while imaging, line scans across the particle and composition maps were not able to both maintain reliable spatial resolution and collect adequate signal to resolve the Pt and Ir peaks. To overcome this issue, a complete line scan across the particle was replaced by two extended point scans, one performed at the center of the particle and one at the particle edge. These scans were performed by alternating between 5 seconds of signal collection and drift correction for a total of seven minutes. To aid in visual interpretation of the spectra, the raw EDS data was smoothed using adjacent averaging and is overlaid on the raw data. The Ir(L), Pt(L), and Cu(K) (Cu from the TEM grid) peaks were fitted using TIA software. From this analysis, it was found that the center of the particle had a molar ratio of 1:1 Pt:Ir, while the edge of the particle was Pt rich with a molar ratio of 8:1 Pt:Ir. This is consistent with an Ir@Pt core-shell structure. Pt enrichment at the particle edge is visually apparent in the EDS spectra when comparing the highest intensity L peaks, Lα1, of Ir and Pt at 9.2 and 9.4 kev, respectively. At the center of the particle, this peak is clearly visible for both Ir and Pt, whereas at the edge of the particle only the Pt Lα1 peak can be seen. Point scans of multiple particles found that the majority of the particles contain both Pt and Ir; however, some smaller 4

5 particles were found to contain only Ir. This is consistent with the bimodal particle size distribution of observed in Figure 1 and confirms the presence of uncoated Ir particles. Of the numerous particles surveyed, no Pt-only particles were found. 2.2 Comparison to Larger Pt-only Nanoparticles The activities of Ir@Pt and Pt/C (TKK) were compared to Pt-only nanoparticles synthesized by the polyol method discussed above. TEM and electrochemistry of the Pt-only nanoparticles are shown in Figures S1 and S2, respectively. Commercial TKK had an ECSA of 67 m 2 /gpt, much larger than Pt-only particles (30 m 2 /gpt) due to their smaller average particle size (2.5 nm vs. 5.8 nm, respectively). The specific activities of Pt, Ir@Pt, and Pt/C (TKK) at 0.9 V vs. RHE are 0.74, 0.90, and 0.47 ma/cm 2 Pt, respectively. The greater specific activity of the Pt-only nanoparticles compared to Pt/C (TKK) can be rationalized by their larger particle size. This results in a lower proportion of undercoordinated sites that are well known to provide reduced ORR kinetics. 4 The specific activity of Ir@Pt exceeds that of both the synthesized Pt-only and commercial Pt/C. Since the Ir@Pt specific activity exceeds that of the larger Pt-only nanoparticles, the larger particle size of Ir@Pt over Pt/C (TKK) cannot be the only factor leading to enhanced activity. At 0.9 V vs. RHE, the mass activities of Pt, Ir@Pt, and Pt/C (TKK) are 0.215, 0.35, and 0.31 A/mgPt, respectively. Despite a greater specific activity, Pt-only nanoparticles have a significantly lower mass activity than Pt/C (TKK) due to their larger particle size that results in a 2.2-fold smaller ECSA. The Pt-mass activity of Ir@Pt far exceeds that of the Pt-only nanoparticles as well as Pt/C (TKK). 2.3 Ir Loss During Cleaning To further investigate the effect of Ir loss during electrochemical cycling, base and ORR CVs were compared before and after electrochemical cleaning for synthesized Ir, Pt, and Ir@Pt particles. After electrochemical cleaning, the Ir nanoparticles exhibited a complete loss in HUPD current and became almost fully inactive for ORR (Figure S3). In contrast, the HUPD current and ORR activity of the synthesized Pt-only nanoparticles increased after electrochemical cleaning (Figure S4). This is the expected result for the removal of capping ligands from the Pt surface. Upon cleaning, the Ir@Pt HUPD current deceased while its ORR activity increased. The decrease in HUPD current is consistent with the loss of uncoated Ir-only particles that were observed in TEM, while the increase 5

6 in ORR activity is in accordance with the removal of capping ligands from the surface of particles. The loss of Ir during potential cycling was confirmed with XPS. The Pt and Ir 4f peaks are shown in Figure S9 for as-synthesized particles (top) and after electrochemical cleaning and ORR testing (bottom). Before and after electrochemical testing, Pt is mostly metallic with 4f7/2 and 4f5/2 peaks at 71.6 ev and 74.9 ev, respectively, with small native oxide features at 73.0 ev and 76.4 ev. Similarly, Ir is mostly in the metallic state with 4f7/2 and 4f5/2 peaks at 61.3 ev and 64.3 ev, respectively, with oxide peaks at 62.9 ev and 65.9 ev. While signal attenuation due to the different depths of metal in a core-shell structure does not allow for accurate global composition analysis, it does provide a semi-quantitative guide to compositional changes during electrochemistry. Using typical XPS peak fitting, it was found that the as-synthesized particles have a molar composition of 37% Ir and 63% Pt, which is similar to the 2:1 molar ratio of Pt:Ir used in synthesis. After electrochemical testing, however, there is a significant loss in Ir with the composition becoming 24% Ir and 76% Pt. Ir loss with electrochemical cycling is not indefinite, however. XPS was performed between a series of electrochemical cleaning and ORR tests on the same sample. As cleaning cycles are performed, the Pt:Ir ratio increases indicating Ir loss (Figure S7). As the cleaning continues, this loss slows and plateaus to a constant value after approximately 900 cleaning cycles. Decreases in HUPD current mirror XPS results with a significant drop in current and stabilization after 900 cleaning cycles reflecting the loss of surface Ir atoms (Figure S8). Correspondingly, specific activity increases and stabilizes as the HUPD determined ECSA plateaus. These results are consistent with Ir loss from the removal of uncoated Ir cores as opposed to a continual leaching out of Ir from Ir@Pt particles. 6

7 Figure S1: TEM images of (a) 1:2, (b) 1:1, and (c) 2:1 molar ratios of Pt:Ir core-shell nanoparticles and (d) Pt-only nanoparticles synthesized via the polyol method. Size distributions of the nanoparticles are shown below each respective image. Particle sizes were found to be 2.1 ± 0.9 nm, 2.5 ± 1.3 nm, and 2.9 ± 1.1 nm for the 1:2, 1:1, and 2:1 Pt:Ir, respectively, and 5.8 ± 1.5 nm for Pt-only nanoparticles. Particle agglomeration on the carbon support shown in (b) and (c) likely leads to non-optimal mass activities of the core-shell catalysts. 7

8 Figure S2: Electrochemical characterization of commercial Pt/C (TKK) (black) and synthesized Pt (pink), Ir (purple), 2:1 (green), 1:1 (orange), and 1:2 (blue) nanoparticles. All experiments were performed in 0.1 M HClO4 electrolyte solution at 20 mv/s and 1600 rpm using a reversible hydrogen reference electrode. All samples (except the Ir only) were electrochemically cleaned prior to testing. (a) Cyclic voltammograms in N2 saturated electrolyte. (b) Background corrected anodic sweep cyclic voltammograms in O2 saturated electrolyte. Mass activity (c) and specific activity (d) of the catalysts. The small error (<10%) in limiting current of 1:2 Pt:Ir and Pt only nanoparticles is likely due to significant particle agglomeration on the carbon support leading to an incomplete coating on the electrode surface. 8

9 Figure S3: Effect of electrochemical cleaning on Ir only nanoparticles. (a) Cyclic voltammograms in N2 saturated electrolyte. (b) Background corrected anodic sweep cyclic voltammograms in O2 saturated electrolyte. The current in the base CV, especially in the HUPD region, and the ORR activity decrease after electrochemical cleaning indicating the loss of Ir nanoparticles. Figure S4: Effect of electrochemical cleaning on Pt nanoparticles. (a) Cyclic voltammograms in N2 saturated electrolyte. (b) Background corrected anodic sweep cyclic voltammograms in O2 saturated electrolyte. Mass activity (c) and specific activity (d) of the catalyst. The increase in the 9

10 HUPD and oxygen adsorption/desorption features in the base CV as well as the increase in ORR activity are consistent with the removal of organic ligands from the catalyst surface. Figure S5: Effect of electrochemical cleaning on nanoparticles. (a) Cyclic voltammograms in N2 saturated electrolyte. (b) Background corrected anodic sweep cyclic voltammograms in O2 saturated electrolyte. Mass activity (c) and specific activity (d) of the catalyst. The decrease in the HUPD and oxygen adsorption/desorption features in the base CV may be due to the removal of uncoated Ir cores. The enhancement in ORR activity is consistent with the removal of organic ligands from the surface. 10

11 Figure S6: XPS spectra of the 1:1 core-shell catalyst before and after electrochemical cleaning and testing. A significant decrease in the Ir signal is observed after electrochemistry. Fluorine is from the Nafion binder. 11

12 Figure S7: Series of high resolution XPS spectra for (a) Pt 4f and (b) Ir 4f taken on the same sample of 1:1 between the electrochemical experiments indicated on the left. ORR indicates that 5 potential sweeps were performed under typical oxygen reduction conditions. Clean indicates that 300 electrochemical cleaning cycles were performed. Spectra are temporally organized from bottom to top of the figure. After electrochemical cleaning a significant loss in Ir is observed as illustrated by the calculated molar ratio of Pt/Ir and the molar percent Ir calculated from the XPS spectra. After 900 cleaning cycles, the ratio of Pt to Ir stabilizes to a constant value. This is consistent with the loss of uncoated Ir cores which were observed with TEM and STEM-EDS. 12

13 Figure S8: Electrochemical series corresponding to the XPS spectra in Figure S7. Legend labels indicate the total number of cleaning cycles performed previously and colors correspond with Figure S7. (a) Cyclic voltammograms in N2 saturated electrolyte. (b) Background corrected anodic sweep cyclic voltammograms in O2 saturated electrolyte. (c) Specific activity of the catalyst. A large change in the base CV is observed after the first 300 cleaning cycles which corresponds with a large decrease in the amount of Ir seen in Figure S7. The change in shape is also consistent with that observed in the base CV of the Ir only nanoparticles before and after cleaning in Figure S3. Throughout cleaning, the catalyst activity continues to increase. After 900 cleaning cycles, the specific activity reaches a constant value which corresponds to when the HUPD current and Pt/Ir ratio stabilize (Figure S7). These observations are consistent with the loss of uncoated Ir cores during electrochemical cleaning where the Pt shells serve as protection to the Ir cores preventing their corrosion during electrochemical testing. The Ir loss during testing can be alleviated by achieving a more uniform coating of Pt on all particles. 13

14 Figure S9: XPS spectra of 2:1 Pt:Ir nanoparticles before (top) and after (bottom) electrochemical testing indicates initial Ir loss with testing. The Ir content stabilizes after 900 cleaning cycles (see Figure S7) possibly indicating uncoated Ir cores are removed. Figure S10: Cyclic voltammograms in the H adsorption region in N2 saturated 0.1 M HClO4 for Pt/C (TKK) (black) and 2:1 Pt:Ir (green). The potentials to absorb an estimated 1/8 and 1/4 of a monolayer (ML) of H* on the respective catalysts are indicated by dashed lines. The H* adsorption potentials are negatively shifted for the Ir@Pt with respect to Pt/C. This is consistent with the average destabilization of the Pt-H bond for the Ir@Pt catalyst compared to pure Pt, a trend that has been shown to correlate well with increased ORR activity. 5 14

15 Figure S11: Average performance retention of (2:1 Pt:Ir) (green) and Pt/C (TKK) (purple) after accelerated stability cycles from V vs. RHE at 125 mv/s in O2 saturated 0.1 M HClO4. Data represents the average of three independently measured samples with the error bars as standard deviations. The average mass activity, specific activity, and ECSA retentions after AST for Ir@Pt are 100 ± 10 %, 120 ± 20 %, and 85 ± 6 %. The average mass activity, specific activity, and ECSA retentions after AST for Pt/C (TKK) are 78 ± 2 %, 97 ± 9 %, and 81 ± 8 %. Ir@Pt outperforms Pt/C (TKK) for all three stability metrics. 15

16 Figure S12: TEM of (2:1 Pt:Ir) before AST (a and b) and after cycles of AST (c and d). TEM-EDS before (e) and after (f) cycles of AST. Before testing, appears as individual particles whereas after testing, clusters of individual particles have agglomerated. Isolated particles appear to have maintained their original size. Agglomeration of clustered particles likely results in a reduced ECSA as was observed during the AST and can be ameliorated by increasing the original particle dispersion on the carbon support. Both Pt and Ir appear to be present before and after testing; however, the observed increase in molar ratio of Pt:Ir after testing is indicative of some Ir loss during testing, consistent with the removal of uncoated Ir cores. 16

17 (1) Jackson, A.; Viswanathan, V.; Forman, A. J.; Larsen, A. H.; Nørskov, J. K.; Jaramillo, T. F. Climbing the Activity Volcano: Core-Shell Electrocatalysts for Oxygen Reduction. ChemElectroChem 2014, 1, (2) Green, C. L.; Kucernak, A. Determination of the Platinum and Ruthenium Surface Areas in Platinum-Ruthenium Alloy Electrocatalysts by Underpotential Deposition of Copper. I. Unsupported Catalysts. J. Phys. Chem. B 2002, 106, (3) Jackson, A.; Viswanathan, V.; Forman, A. J.; Norskov, J.; Jaramillo, T. F. Effects of a New Electrochemical Cleaning Protocol on Ru@Pt Core-Shell ORR Catalysts. ECS Trans. 2013, 58, (4) Viswanathan, V.; Hansen, H. A.; Rossmeisl, J.; Nørskov, J. K. Universality in Oxygen Reduction Electrocatalysis on Metal Surfaces. ACS Catal. 2012, 2, (5) Escudero-Escribano, M.; Malacrida, P.; Hansen, M. H.; Vej-Hansen, U. G.; Velazquez- Palenzuela, A.; Tripkovic, V.; Schiotz, J.; Rossmeisl, J.; Stephens, I. E. L.; Chorkendorff, I. Tuning the Activity of Pt Alloy Electrocatalysts by Means of the Lanthanide Contraction. Science 2016, 352,