Supporting Information

Transcription

1 Supporting Information Selective CO 2 Reduction to CO in Water using Earth-Abundant Metal and Nitrogen-Doped Carbon Electrocatalysts Xin-Ming Hu, a,b Halvor Høen Hval, a,b Emil Tveden Bjerglund, a,b Kirstine Junker Dalgaard, b Monica Rohde Madsen, a,b Marga-Martina Pohl, c Edmund Welter, d Paolo Lamagni, a,b Kristian Birk Buhl, a,b Martin Bremholm, b Matthias Beller, c Steen U. Pedersen, a,b Troels Skrydstrup, a,b and Kim Daasbjerg*,a,b a Carbon Dioxide Activation Center (CADIAC), Interdisciplinary Nanoscience Center (inano), Aarhus University, Gustav Wieds Vej 14, 8000 Aarhus, Denmark b Department of Chemistry, Aarhus University, Langelandsgade 140, 8000 Aarhus, Denmark c Leibniz-Institut für Katalyse an der Universität Rostock, Albert-Einstein-Straße 29a, Rostock, Germany d Deutsches Elektronen-Synchrotron - A Research Centre of the Helmholtz Association, Notkestraße 85, D Hamburg, Germany * Corresponding author: kdaa@chem.au.dk S1

2 Figure S1 shows that the distribution of N species is similar for all M- electrocatalysts. The red, blue, pink, and green colors represent pyridinic, pyrrolic, graphitic, and oxidized N species, respectively. All atomic percentages of the four kinds of N species of the material are slightly lower than those of M Atomic Percentage (%) Pyridinic Pyrrolic Graphitic Oxidized Figure S1. Percentages of the four kinds of N species present in the and M- (M = Fe, Co, or Ni) materials. S2

3 Figure S2 depicts the experimental and fitted EXAFS spectra in R-space. The position of the first, main peak in χ(r) at Å suggest that the first shell has light backscatters like C, N, and O. More likely, the peak should represent metal-nitrogen or metal-carbon paths. However, EXAFS is poorly suited to distinguish scattering from nitrogen and carbon. The fits using the model with the metal coordinated to four nitrogen in a planar configuration is approximately following the second and third peaks, suggesting that metal-metal coordination is not necessary for a description of R in the range of Å. In other words, the local structure of metal species could all be atomically dispersed ions. All fitting parameters are collected in Tables S4 S Experimental Fit Experimental Fit (R) (Å -3 ) (R) (Å -3 ) R (Å) R (Å) Experimental Fit (R) (Å -3 ) R (Å) Figure S2. Comparison of the experimental FT-EXAFS curves of the M- (M = Fe, Co, or Ni) materials with fitted curves and the structure model used for the fitting. S3

4 Figure S3 shows the EDX spectra of the and M- electrocatalysts. In agreement with the XPS analysis all M- materials reveal mostly carbon, a small amount of nitrogen, and trace amount of the respective metal. In addition, small peaks for oxygen, sodium, silicon, sulfur, and chlorine are detected which originate from either residual silica or glassware contaminants. The pronounced copper signal is from the grid used for these measurements. Intensity (a.u.) C Intensity (a.u.) N O Na Si S Cl Fe Co Cu Ni Energy (kev) Energy (kev) Figure S3. EDX spectra of the and M- (M = Fe, Co, or Ni) materials (left: full range spectra; right: zoom in). S4

5 Figure S4 depicts the HAADF-STEM images spectra of and electrocatalysts. Figure S4. HAADF-STEM images of (A,B) and (C,D). The pictures B and D are taken ~1 min after A and C, respectively, to show the immovability of the positions of single atoms (marked with arrows). S5

6 Figure S5 shows the TEM images of and the graphitic carbon domains in the,, and electrocatalysts. In comparison, Figure 2c in the main text depicts the TEM images of the porous amorphous carbon structure in,,, and. Figure S5. TEM images of (a), (b), (c), and (d). S6

7 Figure S6 shows the diffraction patterns acquired for the,, and electrocatalysts on amorphous and graphitic carbon domains, i.e. both domains exist. a c e /nm /nm /nm b d f /nm /nm /nm Figure S6. Diffraction patterns of (a), (c), and (e) on amorphous carbon domains and of (b), (d), and (f) on graphitic carbon domains. S7

8 Figure S7 depicts the PXRD patterns of the and M- electrocatalysts. The Fe, Co, and Ni doped carbon materials all have two broad peaks with 2θ = 24 and 43, characteristic of graphitic domains in amophous carbon. and show higher percentage of graphitic carbon domains than and. Noteworthy, a sharp peak at 2θ = 26.3 assigned to multi-layer graphene sheets is observed for and, but only weakly for, which indicates that the graphitic domains in are mostly few-layer graphene-like sheets. No peaks from metal based particles could be detected in the three PXRD patterns for M- showing the effectiveness of the acid washing. The two small peaks (2θ = 45.6 and 46.9 ) observed for are assigned to the presence of unknown contaminants as they cannot originate from any of the elements (Ni, C, N, O, and Si) detected by XPS (Table S2) according to the Crystallography Open Database ( Cristobalite Intensity (a.u.) (degree) Figure S7. PXRD patterns of the and M- (M = Fe, Co, or Ni) materials. S8

9 Figure S8 depicting the Raman spectra of the and M- electrocatalysts shows two intense broad peaks centered around 1350 and 1590 cm 1, which are assigned to the characteristic D and G peaks of graphitic carbon, respectively. The D peak originates from the disorder induced defects, while the G peaks is related to the vibration of sp 2 bonded carbon atoms in a two-dimensional hexagonal lattice. Two small broad peaks are visible in the range of cm 1, corresponding to the 2D and D+Dˈ peaks, respectively. In some cases these two peaks overlap. The Raman spectra clearly show the presence of graphitic structures in the M- electrocatalysts. D G 2D D+D' Intensity (a.u.) Raman Shift (cm -1 ) Figure S8. Raman spectra of the and M- (M = Fe, Co, or Ni) materials. S9

10 Figure S9 shows the nitrogen adsorption-desorption isotherms of the and M- electrocatalysts. The surface areas are calculated from the adsorption isotherms using the Brunauer Emmett Teller (BET) method. The surface areas for the silica-templated,,, and are 807, 742, 559, and 680 m 2 g 1, respectively. In contrast, the -NS material has a much lower surface area (= 107 m 2 g 1 ), highlighting the importance of using the silica templated synthesis. V ads, STP (cm 3 g -1 ) NS Relative Pressure (P/P 0 ) Figure S9. Nitrogen adsorption desorption isotherms of the, M- (M = Fe, Co, or Ni), and -NS materials at 77 K. S10

11 Figure S10 depicts the cyclic voltammograms recorded at the,, and electrocatalysts in the absence and presence of CO j (ma cm -2 ) Ar CO 2 j (ma cm -2 ) Ar Potential (V vs RHE) -14 CO Potential (V vs RHE) 2 0 j (ma cm -2 ) Ar -12 CO Potential (V vs RHE) Figure S10. Cyclic voltammograms recorded at the,, and materials using a sweep rate of 10 mv s 1 in 0.5 M KHCO 3. S11

12 Figure S11 shows cyclic voltammograms of the and M- electrocatalysts recorded between V vs RHE at three different sweep rates in CO 2 saturated aqueous 0.5 M KHCO 3. The double layer capacitive current density was obtained by averaging the cathodic and anodic current density from voltammograms (a c) at 0.25 V vs RHE. From a plot of these current densities against the sweep rate (Figure S11d), the layer capacities are calculated to be 50, 47, 41, and 49 mf cm 2 for,,, and, respectively. The relative electrochemically active surface area (ECSA) is, by and large, the same for all materials, indicating that the activity differences observed for the CO 2 -to-co conversion must originate from the nature of the different active sites. Note that the ECSA ś cannot be compared directly with the relative surface areas obtained by the BET method, considering that the hydrated K + in the capacitive measurements with an effective diameter of 0.6 nm could easily be precluded from populating some of the micropores a 0.8 b j (ma cm -2 ) j (ma cm -2 ) Potential (V vs RHE) Potential (V vs RHE) c d 0.8 j (ma cm -2 ) DL-j (ma cm -2 ) Potential (V vs RHE) (mv s -1 ) Figure S11. Cyclic voltammograms of the and M- (M = Fe, Co, or Ni) materials recorded between V vs RHE at the sweep rate,, of (a) 5, (b) 10, and (c) 20 mv s 1 in CO 2 saturated aqueous 0.5 M KHCO 3. (d) Plot of double layer current densities, DL-j, obtained at 0.25 V vs RHE from voltammograms (a c) vs. S12

13 Figure S12 presents the FE and partial current density for H 2 evolution, j H2, of the 15 min electroylsis performed on the and M- electrocatalysts at various potentials (supplementary to Figure 4c e). Figure 4d shows that the total current density, j, follows the order > > >. However, the situation changes for the partial current density of CO production, j CO, which is almost the same for and (Figure 4e). This indicates that the cobalt doping adds nothing to the CO 2 RR activity. Rather, it leads to a considerably enhanced HER (Figure S12). Figures S12a and S12b also reveal that the activity of and for the HER is small at potentials > 0.6 V, but becomes dominant at more negative potentials. In contrast, the HER activity on remains small, even at extreme potentials, which explains its good selectivity for CO 2 -to-co conversion. FE H2 (%) a j H2 (ma cm -2 ) b Potential (V vs RHE) Potential (V vs RHE) Figure S12. (a) FE and (b) partial current density of H 2 production using and M- (M = Fe, Co, or Ni) electrocatalysts in 15 min electrolyses at various potentials in 0.5 M KHCO 3. S13

14 Figure S13 shows the CO 2 RR activity of Fe, Co, and Ni porphyrins immobilised on multi-walled carbon nanotubes (MTPP-MWCNT) to make a comparison with the corresponding M- electrocatalysts possible. CoTPP-MWCNT exhibits both higher FE CO and j than FeTPP-MWCNT and, in particular, NiTPP-MWCNT that shows no activity for CO production H 2 CO j FE (%) j (ma cm -2 ) 0 FeTPP-MWCNT CoTPP-MWCNT NiTPP-MWCNT 0 Figure S13. FE and j of 15 min electrolyses of CO 2 at 0.57 V vs RHE using three different carbon supported metalloporphyrins in 0.5 M KHCO 3. S14

15 Table S1. Details on the Preparation of Carbon Electrocatalysts. Material T ( C) Metal salt Template Acid washing 2nd pyrolysis Yield (%) a 1000 No Yes Yes Yes FeCl 3 Yes Yes Yes CoCl 2 Yes Yes Yes NiCl 2 Yes Yes Yes 4 -NS 1000 FeCl 3 No Yes Yes 43 -NA 1000 FeCl 3 Yes No Yes 41 -N2P 1000 FeCl 3 Yes Yes No 31 a Yield is calculated relative to the mass of the organic precursor, i.e. o-phenylenediamine. S15

16 Table S2. Elemental Composition of Carbon Electrocatalysts Determined by XPS. Material C N M Si O (at%) (at%) (at%) (at%) (at%) NS NA N2P XPS is a semi-quantitative technique, where the determination of low metal contents in the carbon material is associated with a large uncertainty. Hence, we employed a more accurate technique, ICP-OES, to determine the metal content in the and M- electrocatalysts (Table S3). As seen the results obtained by the two techniques are fully consistent. Table S3. Comparison of Metal Content in Carbon Electrocatalysts Determined by XPS and ICP-OES. Metal content XPS (wt%) a ICP-MS (wt%) a Weight percentage of the metal is calculated from the corresponding atomic percentage listed in Table S2. S16

17 Tables S4 S6 show the fit parameters from the EXFAS analysis. Although the fits match well with the experimental data in Figure S2, some issues emerge in the fits of the model. Firstly, the fitted S 2 0 value for Fe is larger than 1.0. Secondly, the refined energy shift, E 0, for the Ni edge is larger than 10 ev. Finally, the radial shifts, ΔR, with respect to the model distances for the carbon atoms in the Fe and Co sample do not all have the same sign, resulting in contraction except for M-C4. These issues may to some extent be resolved by e.g. inclusion of more multiple scattering paths, but the presence of multiple of varying local coordination is likely intrinsic to these samples. Some variation in the metal coordination is expected, i.e. the coordination number and the ratio of carbon and nitrogen atoms vary within the same system. The atomic structure model is considered as an effective description of the three systems. The analysis shows the metal atoms are atomically dispersed, but the quantitative results should be interpreted with caution. Table S4. EXAFS Fitting Parameter for. a Path Coordination numbers R (Å) σ 2 (Å 2 ) Fe-N (1) 0.012(2) Fe-C (2) 0.016(3) Fe-C (2) - Fe-C (9) - Fe-C (6) - a Transmission data were used; R-factor: 0.020; Reduced χ 2 : 1203; Energy shift, E 0 : ev; S 0 2 : Table S5. EXAFS Fitting Parameter for. a Path Coordination numbers R (Å) σ 2 (Å 2 ) Co-N (2) 0.014(3) Co-C (1) 0.009(2) Co-C (1) - Co-C (6) - Co-C (6) - a Fluorescence data were used; R-factor: 0.041; Reduced χ 2 : 1050; Energy shift, E 0 : ev; S 0 2 : S17

18 Table S6. EXAFS Fitting Parameter for. a Path Coordination numbers R (Å) σ 2 (Å 2 ) Ni-N (1) 0.011(2) Ni-C (3) 0.014(4) Ni-C (3) - Ni-C (9) - Ni-C (8) - a Transmission data were used; R-factor: 0.037; Reduced χ 2 : 2473; Energy shift, E 0 : ev; S 0 2 : S18