Molecular electrocatalysts transform CO into C2+ products effectively in a ow cell

Shaoxuan Ren University of British Columbia Arthur Fink University of British Columbia Eric Lees University of British Columbia https://orcid.org/0000-0002-0524-3816 Zishuai Zhang University of British Columbia Wen Yu Wu University of British Columbia https://orcid.org/0000-0002-1528-7257 David Dvorak University of British Columbia Curtis Berlinguette (  cberling@chem.ubc.ca ) University of British Columbia https://orcid.org/0000-0001-6875-849X


Introduction 1
Using renewable electricity to drive the CO2 reduction reaction (CO2RR) is an appealing way to 2 synthesize carbon-neutral chemicals and fuels. 1 Electrolytic CO2RR at a high rate of product formation 3 requires the continuous feed of reactants to the respective electrodes in a flow cell (or "electrolyzer"). 1,2 4 A microfluidic flow cell is a type of flow cell proven to be particularly effective for mediating CO2RR 5 electrolysis at high rates of product formation. 3,4 The highest performing microfluidic CO2RR flow cells 6 pass an aqueous 7 M KOH solution between an anion exchange membrane (AEM) and cathodic gas 7 diffusion electrode (GDE). 5 This high pH serves to promote the CO2RR by suppressing the competing 8 hydrogen evolution reaction (HER). However, these strongly alkaline conditions convert up to 70% of 9 reacted CO2 into unreactive (bi)carbonates (Eq. 1-2), which also inducing solid salt formation that The use of CO as a reagent (instead of CO2) can produce the same carbon-containing products as 16 CO2RR electrolysis while bypassing the formation of (bi)carbonates that reduce the efficiency and 17 durability of a flow cell. 8 An additional advantage of the CO reduction reaction (CORR) electrolysis is 18 that it inherently involves saturating the electrocatalyst surface with adsorbed CO (*CO), 9,10 thereby   19 increasing the probability of carbon-carbon coupling to form multi-carbon (C2+) products. 11,12 This 20 Faradaic efficiency is appealing because C2+ products are generally more economically valuable than the 21 C1 products (e.g., CO, HCOOH, CH4) that are more readily produced by CO2RR electrolysis. 13

23
Solid-state copper is the only electrocatalyst reported to mediate high rates of CORR (i.e., J >100 24 mA/cm 2 ). [14][15][16][17][18][19] This electrocatalyst produces a range of C2+ products (e.g., acetate, ethylene, and alcohols), 25 but not selectively. A major challenge for addressing CORR electrolysis is discovering an electrocatalyst 1 that can selectively produce a single C2+ product. The highest reported CORR selectivity is merely 48% 2 (for the formation of acetate). 16 While there are a number of studies exploring ways to modify solid-state 3 electrocatalysts, our program is intrigued by the use of molecular electrocatalysts to control reaction 4 selectivity in flow cells because of the acute synthetic control over the electronic environment of the active 5 site and the primary coordination and secondary coordination spheres. 3,20-24 While we have shown that 6 molecular electrocatalysts can operate at high current densities in a CO2RR flow cell for the selective 7 formation of CO (i.e., FECO of 95% at 150 mA/cm 2 ), 25 the use of molecular electrocatalysts that reduce 8 CO at high current densities under flow conditions has not yet been documented. 26,27 9 We report here that a copper phthalocyanine (CuPc; Fig. 1) molecular electrocatalyst can reduce 10 CO into C2+ products at a Faradaic efficiency (FEC2+) >70% at current densities up to 200 mA/cm 2 . These 11 performance parameters were realized in a microfluidic flow cell by continuously delivering gaseous CO 12 and 2 M KOH electrolyte to the surface of a GDE. The C2+ formation rates we report here [i.e., current 13 density for C2+ products (JC2+) of ~150 mA/cm 2 ] constitute a marked improvement relative to previous 14 reports of molecular electrocatalysts (i.e., JC2+ <10 mA/cm 2 in prior art; Fig. 1). Moreover, we show that 15 CuPc maintains stable C2+ formation at 100 mA/cm 2 for >20 h without significant declines in 16 electrocatalyst activity (i.e., <10% decrease in FEC2+). These results are not only commensurate with state-17 of-the-art solid-state electrocatalysts, 15,16 but the performance metrics also establish molecular  Faradaic efficiencies towards C2+ (FEC2+) production as a function of current density (J) for high-3 performing molecular electrocatalysts 28-33 (blue; details in Supplementary Table 1) and the state-of-the-art 4 solid-state Cu electrocatalysts. 5,14-18 (gray) are presented. The data for CuPc featured in this study is 5 indicated in orange. 6 7 8 All electrochemical experiments in this study were performed in a commercial three-compartment 9 flow cell reactor (Micro Flow Cell ® ) operating in a "flow-through" configuration ( Fig. 2). This "flow- 10 through" configuration forces gaseous reactants to penetrate through the electrocatalyst on the electrode, 11 differing from the traditional "flow-by" configuration whereby gas reactants reach the electrocatalyst only 12 by diffusion (see Supplementary Fig. 1 for details). 34 The electrochemical reactor consisted of two PTFE 13 housing pieces, a Pt/Ni anode current collector, a Ti cathode current collector, three 14 polytetrafluoroethylene (PTFE) flow frames containing expanded PTFE meshes for turbulence, and 15 rubber gaskets. The first flow frame delivered CO(g) at 10 standard cubic centimeters per minute (sccm) 16 from the back side and through the CuPc-coated gas diffusion electrode (GDE) that is fixed on the cathode 17 current collector with a Ti frame. The unreacted CO and product mixture is delivered into the second flow frame carrying 2 M KOH catholyte solution at a flow rate of 6 mL/min between the anion exchange 1 membrane (AEM, Sustainion ® ) and the electrocatalyst side of the GDE. A leak-free Ag/AgCl reference 2 electrode was inserted into the second flow frame to measure the cathodic potential. On the opposing side 3 of the AEM, 2 M KOH anolyte was circulated at 6 mL/min through a third flow frame between the AEM 4 and the nickel foam fixed on the anode current collector. The outlet catholyte was recycled to a reservoir 5 that was continuously purged through the headspace with N2(g) at 44 sccm to carry CO and gaseous product 6 to the in line gas chromatograph (GC) for analysis. The flow rate of the gas mixture was measured by a 7 flow meter before the gas entered into the GC. Liquid products collected from both the catholyte and 8 anolyte were quantified by 1 H NMR spectroscopy using methylsulfonylmethane (MSM) and potassium 9 hydrogen phthalate (KHP) as internal standards with corresponding calibration curves. CuPc was immobilized on carbon black by sonication in ethanol for 1 h. Nafion ® ionomer was 21 then added to the solution and sonicated for another hour. Oxide-derived copper electrocatalysts (OD-Cu) were synthesized for control experiments according to a previously reported procedure. 17 These two 1 electrocatalyst ink solutions were spray-coated separately onto the microporous layer (MPL) of 2 hydrophobic gas diffusion layers (GDLs; 2 × 2 cm 2 ; Sigracet 39BB) to form GDEs. Each batch of GDE 3 furnished an average CuPc loading of ~0.6 mg/cm 2 . The loading of OD-Cu was measured to be ~0.6 4 mg/cm 2 . The OD-Cu coated GDEs were subjected to a preconditioning step where a constant cathodic 5 current of 15 mA/cm 2 was applied for 800 s.   The predominant gaseous product was C2H4 with a FEC2H4 of ~20% at all applied current densities that 19 were tested. The highest partial current density for C2+ formation (JC2+) was measured to be 147 ± 6 20 mA/cm 2 , and was achieved at a cathode potential of -1.61 V (vs. RHE, without iR correction). This with CuPc at current density of 100 mA/cm 2 . See Supplementary Fig. 3 for FE of each electrolysis product. 12

Results and Discussions
The value of FEC2+ is obtained by the summation of FEC2H4, FEAcO-, FEC2H6O, FEC3H8O, and FEC3H6O. 13 14 As a control experiment, we then tested the state-of-the-art electrocatalyst OD-Cu under the same 15 operating conditions as those used for the CuPc electrocatalyst (i.e., electrolyte composition, current 16 density, etc.). These electrolysis experiments with OD-Cu yielded a similar product distribution of 17 ethylene, oxygenates, and hydrogen that matched literature reports (Fig. 3c). 17 The relative amounts of 18 C2+ produced matched that observed with CuPc, yet CuPc was 25% more selective towards AcO -1 production than OD-Cu, while OD-Cu was 12 % more selective for alcohols (e.g., ethanol and propanol). 2 3 We tested the stability of CuPc in a flow cell by measuring the FE of the different products over 4 20 h of electrolysis at 100 mA/cm 2 (Fig. 3d). During this stability test, the KOH catholyte was periodically 5 replenished to ensure the concentrations of liquid products remained within the calibration ranges for 1 H 6 NMR analyses. A high FEC2+ of ≥64% was maintained throughout the 20 h period of electrolysis. While 7 a 10% decrease in FEC2+ was observed over this period, this value compares remarkably favorably to 8 solid-state Cu electrocatalysts that lasted merely 3 h at the same current densities. 16,17 One reason our 9 system may be more stable than previous reports is because we operated the microfluidic flow cell in a 10 flow-through configuration. This flow-through configuration pushes gaseous CO through the GDE and 11 helps to mitigate "flooding" which is a common flow cell failure mode due to excessive electrolyte 12 percolation into the GDE. 38 13 14 We analyzed the morphologies of the electrocatalyst layers on the carbon GDEs before and after 15 2 h of electrolysis at 150 mA/cm 2 using scanning electron microscopy (SEM) and energy dispersive X-16 ray (EDX) spectroscopy. We observed similar porous morphologies and evenly distributed elemental 17 signals in the SEM and EDX images, respectively, but there was evidence of some electrocatalyst 18 aggregation ( Supplementary Figs. 4-5). X-ray diffraction (XRD) spectroscopy measurements showed an 19 identical crystallinity pattern before and after the electrolysis (Supplementary Fig. 6). X-ray photoelectron 20 spectroscopy (XPS) measurements of CuPc before and after 20 h of electrolysis at 100 mA/cm 2 showed 21 no changes in Cu 2+ signals (~935.3 eV and ~955.2 eV; Supplementary Fig. 7). Collectively, these XRD 22 and XPS measurements confirm CuPc remains largely intact after electrolysis (i.e., no solid-state copper 23 was detected). 1 We show in this study how a molecular electrocatalyst, CuPc, coated on a carbon-based GDE can 2 mediate the electrolytic conversion of CO into C2+ products in a microfluidic flow cell at high rates of C2+ 3 formation (i.e., JC2+ of 147 ± 6 mA/cm 2 ). This conversion rate exceeds all previous reports of molecular 4 electrocatalysts. These findings present the opportunity to explore the rich library of molecular catalysts 5 available in the literature, as well as new derivatives, for effective CO electrolysis at industrially-relevant 6 rates of product formation. Moreover, the well-defined structure of molecular electrocatalysts can be used 7 to elucidate reaction mechanisms pertinent to the rate-limiting carbon-carbon formation step that is 8 relevant to both CO2RR and CORR. We contend that further investigations into molecular electrocatalysts 9 in flow reactors will broaden the scope of energy-dense products that can be produced efficiently from    Kα2 with a nickel filter to remove Kβ. The detector is LynxEye silicon strip and the slits are 1 mm divergent, 3 8 mm anti-scatter, and 2.5˚ soller. A FEI Helios NanoLab 650 scanning electron microscope (SEM) was 4 utilized to obtain micrographs of the GDE surface morphology. Energy dispersive X-ray (EDX) maps of 5 size 256 × 200 pixels were collected at 25 kX magnification using an accelerating voltage of 5 kV and 6 beam current of 1.6 nA. X-ray photoelectron spectra (XPS) were collected with Kratos Analytical Axis 7 Ultra DLD XPS system. The XPS spectra was calibrated with the C1s binding energy of 284.8 eV. 8 9 Electrolysis and product analysis. The electrolysis experiments were performed in a flow cell purchased 10 from Electrocell ® . A pre-calibrated leak-free Ag/AgCl reference electrode was used to record the cathodic 11 potential. The CO flow rate to the cathode was set to 10 standard cubic centimeters per minute (sccm) by   Where qj is the number of electrons exchanged, F is Faraday's constant (F = 96,485 C mol -1 ), nj is 20 the moles of product j, I is the applied current, t is the electrolysis time. Turnover frequency (TOF) was Where q is the number of electrons exchanged, F is Faraday's constant (F = 96,485 C mol -1 ), ncat 2 is the moles of catalyst deposited on the gas diffusion electrode, I is the applied current, FE is Faraday 3 efficiency. The consumption of CO (in mL/min) was obtained by equation 6. Then, the single pass 4 conversion was calculated using equation 7 (exemple for AcO -).  The total single pass conversion was calculated by dividing the sum of CO consumed (in mL min -10 1 ) for each product by the inlet CO flow rate of 10 mL min -1 . CuPc molecular electrocatalyst competes with state-of-the-art solid-state copper electrocatalysts.

Conclusion
Faradaic e ciencies towards C2+ (FEC2+) production as a function of current density (J) for highperforming molecular electrocatalysts28-33 (blue; details in Supplementary Table 1) and the state-of-theart solid-state Cu electrocatalysts.5,14-18 (gray) are presented. The data for CuPc featured in this study is indicated in orange.

Figure 2
Illustration of the micro uidic ow cell architecture. a, Exploded view of the micro uidic cell used in this study. b, Scheme of the ow of species at the cathode electrocatalyst. Reactant CO gas ows through the gas diffusion electrode (GDE) for its reduction to various products. KOH catholyte furnishes an alkaline environment at the cathode and transports gaseous and liquid products to the reservoir for analysis by GC or 1H NMR, respectively. The ow of CO and KOH catholyte are indicated in black dash line. The molecular structure of CO, C2H4 and AcO-are shown. (red, oxygen; grey, carbon; white, hydrogen)

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SIpdf.pdf