Revealing the Active Phase of Copper during the Electroreduction of CO2 in Aqueous Electrolyte by Correlating In Situ X-ray Spectroscopy and In Situ Electron Microscopy.

The variation in the morphology and electronic structure of copper during the electroreduction of CO2 into valuable hydrocarbons and alcohols was revealed by combining in situ surface- and bulk-sensitive X-ray spectroscopies with electrochemical scanning electron microscopy. These experiments proved that the electrified interface surface and near-surface are dominated by reduced copper. The selectivity to the formation of the key C-C bond is enhanced at higher cathodic potentials as a consequence of increased copper metallicity. In addition, the reduction of the copper oxide electrode and oxygen loss in the lattice reconstructs the electrode to yield a rougher surface with more uncoordinated sites, which controls the dissociation barrier of water and CO2. Thus, according to these results, copper oxide species can only be stabilized kinetically under CO2 reduction reaction conditions.

Closing the carbon loop toward a clean energy economy requires the reduction of CO2 emissions using substitute fuels for powering engines. Other solutions entail the sequestration of CO2 by pulling it out of the atmosphere and burying it in geologic reservoirs. Combining the two previous concepts yields another approach, wherein CO2 is pulled out of the air and transformed into valuable chemical feedstocks via electrochemistry,[1−5] yielding a negative carbon cycle if this process is powered with renewable sources of energy. Among the different chemical elements, copper can uniquely perform the electrocatalytic reduction of CO2 into valuable hydrocarbons and alcohols.[6−9] Unfortunately, the cathodic CO2 reduction reaction (CO2RR) suffers from low current density, high overpotential,[10,11] low selectivity, and electrode deactivation over time,[12] hindering its industrial application.[13] Consequently, the understanding of Cu activation in the electrocatalytic reduction of CO2 is a crucial factor to develop highly active, selective, and stable electrocatalysts[14] because of its unique property to produce the C–C bond. Surface-sensitive spectroscopies during electrochemical reaction can shed light on this issue, but it is technically challenging to realize such measurements.[15] Thus, revealing the electronic structure of copper during the electrochemical reduction of CO2 requires the use of element-specific techniques and the combination of surface- and bulk-sensitive detection schemes in order to identify the catalyst’s active centers.[16−18] Furthermore, these changes in the electronic structure of the catalyst are accompanied by modifications in the electrode morphology requiring microscopy characterization for the complete process understanding.

Element-specific X-ray spectroscopy is used for the investigation of catalysts during reaction conditions, though it is technically challenging in the soft X-ray regime. A bulk-sensitive version has been realized by thin Si3N4 membranes that separate the liquid phase from vacuum allowing the collection of photons in a total fluorescence yield mode (TFY).[19,20] Another approach is to record the intensity decay of the incoming X-ray in transmission mode, yielding information on the whole catalyst.[21] Surface-sensitive contributions under operando/in situ conditions are, however, scarce. The reason lies in the short inelastic mean free path (IMFP) of the photoelectrons in dense media.[22] This technical limitation forces the use of ex situ characterizations and bulk-sensitive methods, thereby yielding inaccurate pictures of the active/stable copper oxidation state during the CO2RR. In order to provide an unambiguous description of the oxidation state of copper during the electrocatalytic reduction of CO2 to valuable hydrocarbons, we enabled the use of surface-sensitive X-ray absorption spectroscopy (XAS) using an atomically thin-film membrane (single-layer graphene) that is semitransparent to the photoelectrons. Furthermore, these results were compared with bulk-sensitive XAS-TFY mode in order to link these changes on the surface with the variations that the catalyst undergoes in the bulk. In addition, we performed in situ scanning electron microscopy (SEM) under reaction conditions in the presence of aqueous electrolyte using a continuous flow cell. Previous SEM experiments were limited in the detection of species through a Si3N4 window because of a high-contrast effect.[23−27] Furthermore, we overcame the problem of the stability of free-standing graphene reported in the literature during the characterization of liquids[28] for its use in electron spectroscopy and microscopy. Thus, these experiments provide spatial resolution of the modifications that the catalyst undergoes as a consequence of its contact with the liquid electrolyte and applied potential using the same specimen for both microscopy and spectroscopy investigations under operando conditions.

First, the change in the electronic structure and morphology of an electrodeposited electrode was investigated during the CO2RR by means of bulk-sensitive XAS-TFY. In order to do that, we take advantage of an electrochemical cell (EC-cell) based on a thin-film Si3N4 membrane (100 nm thick) semitransparent to the incoming X-ray and out-going photons, which enables photon-in/photon-out (PIPO) spectroscopy while still being leak-tight[29,30] (a schematic drawing of this cell operation is shown in Figure A1, including the detection scheme). A 20 nm thick Au and 3 nm of Cr (adherence layer) electrode was deposited onto this membrane by physical vapor deposition (PVD) and used as the working electrode for the electrodeposition of copper[31,32] from 50 mM CuSO4 (around ∼300 nm thick as shown in Figure S1A) as described in the Supporting Information. The spectra were collected at the ISISS beamline of the BESSY II synchrotron facility (Berlin, Germany)[33] using a photodiode to collect the TFY signal. In addition, the electrode holder is exchangeable between the in situ flow EC-cells compatible with the scanning electron microscopy and TFY spectroscopy setups, allowing the investigation of the same specimen correlating with both techniques. This compatible cell was fabricated and implemented in an environmental scanning electron microscope (ESEM), which we will refer to as electrochemical SEM (ECSEM). It allows for the investigation of electrified electrodes in the presence of liquid, yielding images with nanometric resolution (more details of this approach can be found in the Supporting Information). Using this setup, copper was electrodeposited on the PVD Au film from a 50 mM CuSO4 solution (see the Supporting Information for more details). Figure A2 shows in situ ECSEM images of the electrodeposited copper on the Au electrode at OCV in the presence of 100 mM KHCO3 saturated in CO2. During the CO2RR (at −1.8 V vs Ag/AgCl) the electrode shrinks, indicating a reduction of the electrode due to oxygen loss from the copper oxide lattice. We measured the relative shrinking on different particles and estimated its distribution (see Figure S2C). On average, the copper phase shrinks by 17% of its original length during reaction, which would correspond to 59% of the original volume for a spherical particle. This result agrees well with the value expected for the reduction of either Cu2O or CuO into metallic Cu. Furthermore, the reduction of the copper electrode prompts the transformation of a smooth surface into a rough one that contains crevices and channels with more uncoordinated sites which may contribute to the CO2RR activity/selectivity, as shown in Figure S2. This reduction of the electrode is confirmed by the in situ energy-dispersive X-ray spectroscopy (EDX) measurements shown in Figure S3. The EDX spectra indicate a clear decrease in the oxygen signal in line with copper reduction and oxygen loss observed in the in situ ECSEM measurements. Unfortunately, EDX is not sensitive enough to the details of the electronic structure, and further investigation with other spectroscopic techniques is required to reveal the atomistic aspects of the electrode under reaction conditions. Using the exchangeable electrode in the in situ EC-XAS cell, we investigated the electronic structure of copper during reaction conditions. Figure A3 shows the Cu L3-edges spectra collected in TFY mode depending on the applied potential. Under OCV conditions the electrode is composed of a mixed-oxide of 40% Cu0 and 60% Cu+, as determined by fitting the spectrum with a linear combination of reference spectra (Figure S4A). On the other hand, under CO2RR conditions (−1.8 V vs Ag/AgCl) the electrode is reduced to metallic copper. It is expected that copper oxides are not stable under most of the cathodic potentials at different pH concentrations, as seen in the Pourbaix diagram for copper (see Figure S5),[34] indicating that copper oxide could then only be stabilized because of kinetics during CO2RR. One should remark at this point that the operation in the in situ microreactor cell and in a standard electrochemical setup can be different because of the “current gap” that is related to the difference between the current achievable in both systems being dominated by the cell geometry and charge transport limitations.

Figure 1

(A1) Schematic drawing of the approach used to investigate electrified electrodes (surface and bulk), which are composed of a photon semitransparent Si3N4 membrane (100 nm thick) and electrodeposited Cu electrode onto a 20 nm thick PVD gold electrode. (A2) In situ ECSEM measurements in 100 mM KHCO3 saturated in CO2 under OCV and CO2RR. (A3) In situ XAS measurements in TFY in 100 mM KHCO3 saturated in CO2 under OCV and −1.8 V vs Ag/AgCl. (B1) Schematic drawing of the approach used to investigate the electrified interface (surface), which is composed of a photoelectron semitransparent single layer of graphene on a CEM. (B2) In situ ECSEM measurements in 100 mM KHCO3 saturated in CO2 under OCV and CO2RR. (B3) XAS measurements in TEY and AEY in 100 mM KHCO3 under OCV and −1.8 V vs Ag/AgCl.

TFY-XAS is a bulk-sensitive technique, making it mostly insensitive to the electronic structure modification at the electrified interface. In order to complete the understanding of the CO2RR process at the electrified interface, we investigated the variation in the electronic structure that an electrodeposited copper electrocatalyst undergoes in the course of the electroreduction of CO2 to hydrocarbons using Auger electron yield (AEY) and total electron yield (TEY) XAS, which are surface-sensitive techniques because of the short IMFP of the measured electrons (especially AEY-XAS). Thus, we take advantage of an EC-cell based on an ion-conductive membrane[35,36] upgraded to be used with a semitransparent ultrathin working electrode (single-layer graphene), which allows better wetting of the sample (because this membrane acts as a barrier for water evaporation), while retaining an appropriate escape depth of the photoelectrons.[37,38] Distinct from previous work, we replaced the acidic proton exchange membrane (PEM, Nafion) by a cation exchange membrane (CEM, FKD-type sourced from FuMA-Tech GmbH, Bierigheim-Bissingen, Germany) that also allows the transport of OH–. Figure B1 shows the detection scheme and the different parts of this approach, including the electrode, electrolyte, and CEM. Such a sample assembly allows (among other techniques) for X-ray photoelectron spectroscopy at near-ambient pressure (NAPXPS) and ECSEM, even with back scattered electrons. Technical details of the single-layer graphene transfer, EC-cell description and electrodeposited copper electrode preparation can be found in the Supporting Information. After the copper electrode is electrochemically deposited (around ∼300 nm thick as shown in Figure S1B) on the single-layer graphene membrane, 100 mM HKCO3 (saturated in CO2) electrolyte was flowed into the cell at the same time that the partial pressure of CO2 in the main chamber was set to 0.5 mbar to ensure it reaches the electrode surface. Using these parameters, we investigated our electrode with in situ ECSEM and XAS in TEY and AEY modes. In this manner, Figure B2 shows the ECSEM image of the electrodeposited copper in the presence of 100 mM KHCO3 under OCV conditions; after the electrode is polarized to −1.8 V vs Ag/AgCl, there is a modification of the electrode structure, yielding the formation of lumps on the copper surface. Close inspection of the electrode reveals that there is a reconstruction of the electrode as a consequence of the reduction and oxygen loss in the copper lattice. According to these measurements the surface is initially flatter and during the reaction is reconstructed to a rougher surface with more uncoordinated sites, one in good agreement with the previous measurements with the Si3N4-based EC-cell. The reconstruction occurs over the whole electrode surface as indicated by the image recorded in a fresh spot (see Figure S6). Furthermore, Figure B3 shows the AEY and TEY XAS Cu L3 edge spectra revealing that the surface is reduced from a mixed copper oxide with a composition of 25% Cu0, 45% Cu+, and 30% Cu2+ (see Figure S4B) at OCV to metallic copper during CO2RR. These measurements confirm that the electrified interface is dominated by the formation of reduced copper during the CO2RR, in good agreement with the TFY-XAS measurements. The modification in the electronic structure of the electrode, surface and bulk, to a more metallic copper, is accompanied by the electrode morphology reconstruction to a rougher one, possibly with remaining dissolved oxygen atoms (known as subsurface oxygen[39,40]) due to slow diffusion/reduction kinetics, during the CO2RR. The reversibility of the process depending on the applied potential, for both setups, is discussed in the Supporting Information. Therefore, the surface and bulk are in the most stable phase according to the Pourbaix diagram,[34] indicating that it is exactly the same on the surface and in the near-surface of the catalyst where copper oxide can be stabilized only under CO2RR conditions by the time necessary for the oxygen to diffuse from the bulk to the surface. The experiments reported in this Letter are fully consistent with previous results found recently by Scott et al.[41] using grazing incidence X-ray diffraction (GIXRD) where no copper oxide phase was found during CO2RR. In addition to the information related to the phase of the copper yielded by the in situ GIXRD measurements, the combination of ECSEM/XAS provides an accurate description, with submonolayer sensitivity, of the electronic structure of the catalyst (i.e., oxidation state). X-ray spectroscopy characterization is an element-specific and pure surface-sensitive technique in the case of the AEY in contrast to GIXRD, which is near-surface-sensitive in the best case. In addition, the in situ ECSEM characterization provides important morphology information with nanometric lateral resolution which is not easily accessible with diffraction techniques, indicating that these approaches are complementary. Consequently the results reported in this investigation go a step further in the process description, confirming that the surface of the electrode is reduced to metallic copper with more uncoordinated sites due to surface reconstruction under CO2RR.

From our experiments it is obvious that the applied potential drives the electrode reduction and oxidation and thus rules the reaction performance. In addition to the previous results, we further investigated the variation in the electronic structure and the products evolved from the reaction depending on the applied cathodic potential. Complete sets of spectra with their respective linear sweep voltammetries (LSV) and reference sample comparison are shown in Figure S7 and S8. These measurements were accomplished in TFY and TEY, the latter being the surface-sensitive measurements more relevant for this investigation. Figure A shows how the copper metallicity increases as the electrode is further reduced at higher cathodic potentials, as indicated by the orange arrow in the figure. In addition, Figure B shows the gas-chromatography measurements (GC) of the evolved products depending on the applied potential and the estimated metallicity (in orange). This characterization was done in a setup different than the in situ EC-cell (see the Supporting Information for more details). These measurements confirm the unique ability of copper to produce the C–C bond during the electrocatalytic reduction of CO2, where formic acid is a spectator. The formation of C2 products is enhanced at more cathodic potentials, indicating that the degree of metallicity, rougher surface, and accumulated charge within the electrical double layer (EDL) controls this process. This result is of the utmost relevance for the CO2RR. It is obvious from these experiments that the CO2 reduction is happening on a reduced Cu electrode depending on the applied potential according to the in situ spectroscopy. Furthermore, the calculated barrier associated with the dissociative CO2 adsorption on different copper catalysts[32] indicates that the copper oxide has a higher dissociation barrier due to the formation of extra oxygen on the surface with a CuO-like structure and high dissociation energy, which is the case for OCV conditions, as shown in Figure C. Otherwise, when copper is reduced (as during CO2RR) the dissociation barrier for CO2 is diminished, facilitating the dissociative adsorption of CO2 as shown in Figure C. Furthermore, based on the ECSEM data, the selective CO2 reduction toward hydrocarbons probably takes place on a rougher surface controlling the selectivity of the catalysts due to the formation of more uncoordinated sites. Consequently, these experiments indicate that the nature of the reduced copper active sites dominates the activity and selectivity of copper during the reaction. To conclude, one should note that the beam induces water radiolysis, which potentially has several effects on the electrode changes[42] and local redox potential, pH, and temperature.[43] Strategies that were undertaken to mitigate the electron beam interaction with the electrode and electrolyte are discussed in the Supporting Information.

Figure 2

(A) Cu L3-edge in TEY and (B) Faradaic efficiency for different reaction products depending on the applied potential. (C) Calculations and drawings of the copper structures[32] indicating the difference between Cu2O under OCV and Cu0 during the CO2RR.

In summary, we first accomplished the investigation of the CO2RR combining surface and bulk-sensitive in situ X-ray absorption spectroscopy with in situ ECSEM and GC product analysis using the same specimen. These experiments point out the importance of combining in situ/operando spectroscopy and microscopy in order to describe accurately the electrode modification during the reactions at certain applied potential. We found direct evidence of the nature of the active sites, which rules the electroreduction of CO2 to industrially valuable hydrocarbons and alcohols. Under CO2RR conditions, the surface and the bulk of the electrode are dominated by the existence of reduced copper, which correlates with enhanced C–C bond formation at higher cathodic polarization due to the increased metallicity of the copper electrode. In addition, the formation of reduced copper is accompanied by the electrode reconstruction prompting the formation of a rougher surface that contains crevices and channels with more uncoordinated sites, which in the last term controls the dissociation barrier and thus the activity, selectivity, and stability of the catalysts.