Supporting Information Trace Levels of Copper in Carbon Materials Show Significant Electrochemical CO 2 Reduction Activity Yanwei Lum,,,, Youngkook Kwon,,, Peter Lobaccaro,,,# Le Chen,, Ezra Lee Clark,,,# Alexis T. Bell, *,,Δ,# and Joel W. Ager *,,, Joint Center for Artificial Photosynthesis, Materials Science Division, and Δ Chemical Sciences Division, Lawrence Berkeley National Laboratory, California 9472, United States Department of Materials Science and Engineering and # Department of Chemical and Biomolecular Engineering, University of California, Berkeley, California 9472, United States These authors contributed equally to this work. * A.T.B (email: alexbell@berkeley.edu) and J.W.A. (email: jwager@lbl.gov) S1
X-ray photoelectron spectroscopy a 15 1 5 x 1 4 C 1s O 1s GO 3 6 9 12 485 48 475 47 465 46 92 94 96 25 24 23 22 21 Cu Mn 635 64 645 65 655 66 23 22 21 2 12 11 1 Ni 85 86 87 88 89 Ag 36 365 37 375 38 24 23 22 21 7 71 72 73 74 14 12 1 8 Fe Au 78 81 84 87 9 93 b 4 3 2 1 x 1 5 C 1s GC O 1s 2 4 6 8 1 12 28 Cu 26 24 22 92 94 96 34 32 3 Mn 28 64 65 66 28 26 24 4 38 36 34 Ni 85 86 87 88 89 Ag 32 Fe 3 28 26 7 71 72 73 74 8 36 37 38 8 84 88 92 2 16 12 Au S2
Intesntiy / CPS c 5 4 3 2 1 x 1 5 C 1s CNT O 1s Cu 26 25 24 92 94 96 27 26 Mn 27 26 25 31 3 29 Ni 85 86 87 88 89 Ag 28 Fe 27 26 25 7 71 72 73 74 16 14 12 Au 2 4 6 8 1 12 25 64 65 66 36 37 38 1 8 84 88 92 d 5 4 3 2 1 x 1 5 C 1s PG O 1s 2 4 6 8 1 12 28 Cu 26 24 92 94 96 29 28 27 Mn 26 64 65 66 27 26 25 24 36 34 32 Ni 85 86 87 88 89 29 Fe Ag Au 28 27 26 25 7 71 72 73 74 1 36 365 37 375 38 8 84 88 92 16 14 12 Figure S1. XPS survey scans for (a) graphene oxide (GO), (b) glassy carbon (GC), (c) carbon nanotubes (CNT), and (d) pure graphite (PG). The corresponding narrow scans for Cu, Ni, Fe, Mn, Ag, and Au are shown on the right. S3
Electroactive Surface Area (EASA) of carbon materials The carbon materials (8 µg of carbon for each sample) were first drop-cast onto glassy carbon electrodes and allowed to dry. The electrochemically active surface area (EASA) was then determined using the ferri-/ferrocyanide redox couple ([Fe(CN) 6 ] 3 /4 and typical CVs using this procedure are shown below. The EASA of the carbon materials (GC, PG, GO, and CNT) was determined using the Randles-Sevcik equation. Cyclic voltammetry was carried out in a nitrogen-purged 5 mm K 3 Fe(CN) 6 in.1 M KCl solution at a scan rate of 5 mv s -1. From this, the EASA per unit mass of carbon could be calculated by subtracting the contribution of the electroactive surface area from the glassy carbon support and normalizing by the mass of carbon deposited. The results of this procedure are summarized in Table S1. 4 Current (ma) 2-2 -4 GC PG GO CNT -.2..2.4.6 E (V) vs. Ag/AgCl Figure S2. Voltammograms for electrochemically active surface area (EASA) of the carbon materials (GC, PG, GO, and CNT) determined using the Randles-Sevcik equation. Cyclic voltammetry was carried out in a nitrogen-purged 5 mm K 3 Fe(CN) 6 in.1 M KCl solution at a scan rate of 5 mv s -1. From the EASA, the EASA per unit mass of carbon material was calculated to be 948, 118 and 155 cm 2 g -1 for PG, GO and CNT respectively. S4
Table S1 The electrochemically active surface area (EASA) of the various electrodes prepared by spincoating were also determined using the ferri-/ferrocyanide redox couple ([Fe(CN) 6 ] 3 /4 ). The EASA per unit mass was then used to obtain an estimate of the actual loading of carbon material on the spin-coated electrodes. Electrode EASA of electrode/ cm 2 EASA per unit mass Estimate of carbon / cm 2 g -1 loading / µg GC 4.43 - - PG/GC (spin-coated) GO/GC (spin-coated) 4.92 9,478 53 4.89 11,751 4 CNT/GC (spin-coated) 5.2 15,521 38 S5
Efficacy of the high purity nitric acid washing procedure Table S2. Calculated impurity concentrations in ppm/w of metallic impurities in the as-received carbon materials based on performing the extraction procedure twice with ultrapure nitric acid and measuring the extracted metals with ICP-MS. Standard deviation is abbreviated as stdv and is based on 3 repeats. N.D: Not Detected. Sample Cr Mn Fe Ni PG.191.58 N.D - 3.71.716 N.D - PG rewash N.D - N.D - N.D - N.D - GO 5.78.423 36 135 5.8 8.21 3.46.851 GO rewash N.D - 182 32.8 2.32 1.5 1..12 CNT 3.42 1.55 1.7 2.23 84.9 11.8 115 196 CNT rewash N.D - N.D - 3.94.72 6.2 19.6268 Sample Co Cu Ag Pb PG.111.1758.1.183 N.D - N.D PG rewash N.D - N.D - N.D - N.D - GO N.D - 119 4.58.122.121.59.827 GO rewash N.D - 1.9.255 N.D - N.D - CNT 6.69 2.21249 3..66.5.161 N.D CNT rewash N.D - N.D - N.D - N.D - S6
Table S3. CO2RR product distribution on as-received carbon materials. Standard deviation is abbreviated as stdv and is based on 3 repeats. N.D: Not Detected. Sample Current density (ma/cm 2 ) Faradaic efficiency (%) H 2 CO Methane Formic Acid GC.598 92.5 3.41 1.596.127 N.D -.43.151 PG/GC 1.33 88.7 2.87 1.41.187 N.D -.329.274 GO/GC 1.3 78.3 2.35 9.21 1.56 4.3.288 1.73.112 CNT/GC 3.42 89.9 2.45.297.221 N.D -.62.587 Table S4. CO2RR product distribution on the carbon materials after treatment with nitric acid. Standard deviation is abbreviated as stdv and is based on 3 repeats. N.D: Not Detected. Sample Current density (ma/cm 2 ) Faradaic efficiency (%) H 2 CO Formic Acid PG/GC.981 84. 2.97 5.93.382.277.121 GO/GC.498 79.4 1.61 13.4.891 1.77.152 CNT/GC 1.7 9.4 3.11 1.63.512 1.24.164 S7
Comparison of impurity concentrations in graphene oxide from 4 sources Table S5. Calculated impurity concentrations in ppm/w of metallic impurities in the as-received graphene oxide from 4 different sources. This was done based on performing the extraction procedure with ultrapure nitric acid and measuring the extracted metals with ICP-MS. The graphene oxide as described in the main text wa purchased from Sigma-Aldrich and is also labeled here as GO. Graphene oxide from ACS Material LLC is labeled as ACS GO, reduced graphene oxide from Sigma Aldrich is labeled as SA RGO and graphene oxide from Graphene Supermarket is labeled as GS GO. Standard deviation is abbreviated as stdv and is based on 3 repeats. N.D: Not Detected. Sample Cr Mn Fe Ni Co Cu ACS GO 3.98 1. 151 399 45.6 11.9 3.93 1.86.519.141 2.84 1.21 SA RGO.91.641 377 98.5 16.3 2.5.599.191.676.418 8.47.177 GS GO 3.61 2.28 271 73.2 422 9.84 4.62 1.37.643.298 31.9 1.1 GO 5.78.423 36 135 5.8 8.21 3.46.69 N.D - 119 4.58 Zn Ga Ag Pb Bi Sn Sample (ppb) ACS GO 14.1 3.38.129.268.22.978.296.917.927.165 2.13.864 SA RGO 4.2.295.899.137.119.549.158.792.177.38 1..216 GS GO 55.4 17.1.47.292.93 1.2 4.17 1.59.354.23 2.25.335 GO.486.114 N.D -.122.121.59.827 N.D - N.D - S8
1 1 ACS GO SA RGO GS GO GO Concentration / ppm (by mass) 1 1 1.1.1 Cr Mn Fe Ni Co Cu Zn Ga Ag Pb Bi Sn Figure S3. Plot showing the concentrations of different impurities present in the as-received graphene oxide materials. The graphene oxide as described in the main text was purchased from Sigma-Aldrich and is also labeled here as GO. Graphene oxide from ACS Material LLC is labeled as ACS GO, reduced graphene oxide from Sigma-Aldrich is labeled as SA RGO and graphene oxide from Graphene Supermarket is labeled as GS GO. S9
Faradaic efficiency comparison for graphene oxide from 4 different sources before and after washing with high purity nitric acid a 1 H 2 CO CH 4 HCOOH 6 8 FE / % 6 4 4 2 j (ma cm -2 ) 2 ACS GO SA RGO GS GO GO Before nitric acid treatment b 1 H 2 CO HCOOH 2. 8 1.5 FE (%) 6 4 2 1..5 j (macm -2 ) ACS GO SA RGO GS GO GO After nitric acid treatment. Figure S4. Faradaic efficiencies (FE, shown as bars, left hand axis) and current densities (open white circles with error bars, right hand axis) of graphene oxide from 4 different sources dispersed on glassy carbon electrodes. (a) as-received and (b) after nitric acid treatment. Electrolysis was carried out at -1.3V vs RHE for 2 hours in.1m NaHCO 3 solution. See Tables S6 and S7 for detailed product distributions. S1
Table S6. CO2RR product distribution on as-received graphene oxide from 4 different sources. Standard deviation is abbreviated as stdv and is based on 3 repeats. N.D: Not Detected. Sample Current density (ma/cm 2 ) Faradaic efficiency (%) H 2 CO Methane Formic Acid Acetic Acid ACS GO.983 66.6.723 9.81 1.38 N.D - 2.5.12.437.29 SA RGO GS GO GO 4.54 4.36 1.3 96.4 2.3.1.797.627.314.133.153 N.D - 95.7 2.8.167.115.483.14.191.361 N.D - 78.3 2.35 9.21 1.56 4.3.288 1.73.112 N.D - Table S7. CO2RR product distribution on graphene oxide from 4 different sources after treatment with nitric acid. Standard deviation is abbreviated as stdv and is based on 3 repeats. N.D: Not Detected. Sample Current density (ma/cm 2 ) Faradaic efficiency (%) H 2 CO Formic Acid ACS GO.386 86.6 1.65 8.61.747 2.17.188 SA RGO.482 91.5 1.18 5.4.565.743.168 GS GO.324 89.4 1.83 7.35.386 1.7.267 GO.498 79.4 1.61 13.4.891 1.77.152 S11
Table S8. Table showing partial current density to methane and weight normalized partial current density to methane (based on copper concentration). Sample Cu impurity concentration Partial current density to methane(ma/cm 2 ) Normalized partial current density to methane (ma cm -2 ppm -1 ) SA RGO GS GO GO 8.47.285.336 31.9.211.66 119.443.372 Normalized partial current density to methane (ma cm -2 ppm -1 ).3.2.1. SA RGO GS GO GO Figure S5. Comparison of weight normalized partial current density to methane (based on copper concentration) of various graphene oxide materials. S12
Washing carbon materials with reagent grade nitric acid Carbon materials were sonicated with reagent grade nitric acid for 3 hours. The carbon materials were then recovered via filtration and the leachate was analyzed via ICP-MS to identifiy the trace metals present. The CO2RR activity of the materials was then tested after acid treatment. It was found that metallic impurities persist in the carbon materials after this treatment and in some cases, additional impurities were introduced as well, resulting in an increased CO2RR activity and substantial Faradaic efficiency for methane formation. a H2 CO CH4 C2H4 HCOOH 1 5 FE / % 8 6 4 4 3 2 j / ma cm -2 2 1 PG GO CNT Figure S6: Faradaic efficiencies and current densities for CO2RR when a reagent grade nitric acid was used to clean the carbon materials. S13
Table S9: Calculated impurity concentrations in ppm of metallic impurities in the carbon materials when a reagent grade nitric acid was used for washing. Wash 2 refers to a second wash of the same sample. Sample Reagent grade nitric acid Reagent grade nitric acid wash 2 Calculated impurity concentrations Mn Fe Ni Cu PG. 1.89.75.24 GO 35.58 27.9 1.4 2.13 CNT 4.97 81.55 1273.15.64 PG.96 3.59 1.14 1.43 GO 7.1 34.12 2.25 29.64 CNT.51 6.94 266.81.87 S14
Supplemental Cu loading and FE data Table S1. Actual amounts of Copper deposited in-situ as a result of electrodeposition from copper containing electrolytes at -1.3 V vs. RHE. The concentration of Cu in the electrolyte in ppm is indicated in parentheses. These values were obtained by determining the concentration of copper remaining in the electrolyte using ICP-MS. Standard deviation is abbreviated as stdv and is based on 3 repeats. Carbon Initial mass of Cu in electrolyte (µg) Mass of Cu remaining in electrolyte postelectrolysis (µg) Stdv. of mass of Cu remaining in electrolyte postelectrolysis (µg) Actual mass of Copper deposited (µg) Percentage of Copper deposited (%) Cu(.1)GO/GC.5.37.113.193 38.7 Cu(.1)GO/GC.5.771.795.423 84.6 Cu(1)GO/GC.5.137.367.363 72.5 Cu(2)GO/GC 1.349.676.651 65.1 Cu(2)PG/GC 1.27.654.73 73. Cu(2)GC 1.36.589.639 64. S15
Table S11. CO2RR product distribution on cleaned GO/GC with different concentrations of Cu ions introduced into the electrolyte. Standard deviation is abbreviated as stdv and is based on 3 repeats. N.D: Not Detected. Cu ion / ppm Current density (ma/cm 2 ) Faradaic efficiency (%) H 2 CO Methane Ethylene Formic Acid Acetic Acid Ethylene glycol Ethanol.498 79.4 1.61 13.4.891 N.D - N.D - 1.77.152 N.D - N.D - N.D -.1 1.4 86.1 2.17 5.32.775.51.421 N.D - 2.1.146 N.D - N.D - N.D -.1 1.61 76.1 2.38 1.8.235 12.5.136 N.D - 1.36.19 N.D - N.D - N.D - 1 3.84 56.7 1.89.641.271 37.8 2.16 1.11.14.718.546 N.D -.11.178 1.72.181 2 4.67 45.8 1.58.822.656 42.9 3.57.654.546.693.273.187.154.171.667 3.27.197 Table S12. CO2RR product distribution on PG/GC with different concentrations of Cu ion introduced into the electrolyte. Standard deviation is abbreviated as stdv and is based on 3 repeats. N.D: Not Detected. Cu ion / ppm Current density (ma/cm 2 ) H 2 CO Methane Ethylene Faradaic efficiency (%) Formic Acid Acetic Acid Ethylene glycol Ethanol 1.33 88.7 3.73 1.4.21 N.D - N.D -.331.658 N.D - N.D - N.D -.1 2.94 81.1 2.89.8.352 6.7.421.11.125.569.879.298.452.64.238.15.337 1 3.37 59.5 2.12.6.455 29.7 1.47.418.337.427.747.24.316.119.897.59.521 2 4.49 46.4 1.38.5.316 48.8 2.11 1.63.294.42.556.296.289.61.39.74.694 S16
Table S13. CO2RR product distribution of Cu NPs loaded on GC. Standard Standard deviation is abbreviated as stdv and is based on 3 repeats. N.D: Not Detected. Cu NPs Current mass density loading (ma/cm2) / µg H 2 CO Faradaic efficiency (%) Methane Ethylene Formic Acid Acetic Acid Ethylene glycol Ethanol.346 1.6 54.3 2.47 9.1.566 27.2 2.4.851.612 4.7.25 N.D - N.D - N.D -.692 1.75 4.1 1.98 5.46.331 4.1 3.62 2.28.378 1.98.184.679.581 N.D - 1.55.621 9.69 2.67 47.4 2.43 1.79.247 3.3 2.81 1.7 1.18 3.37.261.688.956 N.D - 4.9.87 Table S14. CO2RR product distribution on GC with different concentrations of Cu ion introduced into the electrolyte. Standard deviation is abbreviated as stdv and is based on 3 repeats. N.D: Not Detected. Cu ion / ppm Current density (ma/cm 2 ) H 2 CO Methane Ethylene Faradaic efficiency (%) Formic Acid Acetic Acid Ethylene glycol Ethanol.598 92.5 2.97 1.58.227 N.D - N.D - N.D - N.D - N.D - N.D -.1 1.5 87.1 2.56.33.451 2.1.112.471.512.547.235 N.D - N.D - N.D - 1 3.36 88.8 1.71.11.133 3.58.217.373.147.92.487 N.D - N.D - N.D - 2 3.71 77.2 1.81.498.335 12.9 1.8.54.384.656.266 N.D - N.D - N.D - S17
Scanning electron microscopy SEM images were obtained of the glassy carbon electrode after electrodeposition of copper as well as the various carbon materials before and after electrodeposition of copper. Figure 3a and c of the main text shows GO and PG prior to and after Cu deposition. In Figure S7c and e, additional images of GO and PG are provided. As seen from the SEM images, a high coverage of the glassy carbon surface is achieved for all the carbon materials. In-situ electrodeposition of copper results in the formation of copper nanoparticles, with the size dependent on the carbon material on which it forms. Cu is observable as particles at low and high resolution when deposited on the the flat GC electrode, Fig. S7a and b. Highly-selective deposition was observed on GO, with copper nanoparticles deposited preferentially on the GO rather than on the exposed glassy carbon surfaces, Fig. S7d. S18
a Cu(2)GC b Cu(2)GC 3 µm 3 nm c GO/GC d b Cu(2)GO/GC 1 µm 3 5 nm e PG/GC f Cu(2)PG/GC 3 µm 3 nm Figure S7. SEM images of (a, b) electrodeposited Cu on the surface of GC (Cu average size 31.2 nm). (c) GO before, (d) GO after electrodeposition (Cu average size 8.6 nm), (e) PG before, and (f) PG after electrodeposition (Cu average size 27. nm) with 2 ppm Cu containing electrolyte. (d) and (f) are reproduced from Fig. 3 of the main text to provide a comparison to electrodeposited Cu on GC, (b). S19
Transmission electron microscopy (TEM) TEM was carried out on the GO.1 sample (after CO 2 electroreduction) to determine the morphology of the copper nanoparticles deposited onto the carbon. TEM was also used to characterize the copper nanoparticles prepared using colloidal synthesis. a b 1 nm 1 nm Figure S8. (a) TEM image of graphene oxide.1 ppm copper (GO.1). (b) HRTEM image of a copper nanoparticle found on the graphene surface. Inset shows the FFT of the copper nanoparticle demonstrating its polycrystalline nature. S2
5 nm Figure S9. TEM image of copper nanoparticles (Cu NPs). The average particle size is approximately 12 nm. S21
Long duration CO2RR using PG and cleaned SA RGO dispersed on GC a 2 b 1 CO H2 Current density / ma cm-2 1-1 -2 Faradaic Efficiency / % 8 6 4 2-3 2 4 6 8 1 12 14 16 18 2 Time / s 1 min 1hr 2hr 3hr 4hr 5hr Time Figure S1. Long duration CO2RR with PG/GC. Electrolysis was carried out at -1.3V vs RHE in.1m NaHCO 3 solution for 5 hours. (a) Current density profile vs. time (b) Product distribution vs. time. Formic acid formation was not analyzed as on-line detection of liquid products was not available and evaporation of electrolyte after long durations becomes significant. Current density and product distribution remains uniform throughout. See Table S13 for detailed product distributions. a 2 b 1 CO H2 1 Current density / ma cm-2-1 -2 Faradaic Efficiency / % 8 6 4 2-3 2 4 6 8 1 12 14 16 18 2 Time / s 1 min 1hr 2hr 3hr 4hr 5hr Time Figure S11. Long duration CO2RR with SA RGO/GC (nitric acid treated). Electrolysis was carried out at - 1.3V vs RHE in.1m NaHCO 3 solution for 5 hours. (a) Current density profile vs. time (b) Product distribution vs. time. Formic acid formation was not analyzed as on-line detection of liquid products was not available and evaporation of electrolyte after long durations becomes significant. Current density and product distribution remains uniform throughout. See Table S13 for detailed product distributions. S22
Table S15. Long duration CO2RR product distributions as a function of time for PG/GC and SA RGO/GC (nitric acid treated). Faradaic efficiency (%) Time PG/GC SA RGO/GC H 2 CO H 2 CO 1 minutes 96.2 1.21 92.3 4.89 1 hour 97.3 1.9 91.7 5.63 2 hours 95.4.98 93.3 4.63 3 hours 94.9 1.11 91.2 4.86 4 hours 96.9 1.32 93.8 5.1 5 hours 97.1 1.27 94.7 4.77 S23