Carbon dioxide electroreduction to C2 products over copper-cuprous oxide derived from electrosynthesized copper complex

Efficient electroreduction of carbon dioxide to multicarbon products in aqueous solution is of great importance and challenging. Unfortunately, the low efficiency of the production of C2 products limits implementation at scale. Here, we report reduction of carbon dioxide to C2 products (acetic acid and ethanol) over a 3D dendritic copper-cuprous oxide composite fabricated by in situ reduction of an electrodeposited copper complex. In potassium chloride aqueous electrolyte, the applied potential was as low as −0.4 V vs reversible hydrogen electrode, the overpotential is only 0.53 V (for acetic acid) and 0.48 V (for ethanol) with high C2 Faradaic efficiency of 80% and a current density of 11.5 mA cm−2. The outstanding performance of the electrode for producing the C2 products results mainly from near zero contacting resistance between the electrocatalysts and copper substrate, abundant exposed active sites in the 3D dendritic structure and suitable copper(I)/copper(0) ratio of the electrocatalysts.


1.36
The Cu and C contents in the complexes were determined by ICP method. 14

Determination of the Cu-Complex loading on Cu substrate
The loading of the complexes on the Cu substrate was determined by gravimetric method, which was obtained with an electrodeposition time of 1 h. The results are given in Supplementary Table 3.  Figure 11 shows the lnI(h) vs. ln(h) plots of the complexes obtained from the SAXS data. Surface fractal (D s ) existed in the complexes, indicating that the surface of the complexes was coarse. Figure 11. lnI(h) vs. ln(h) plots of the complexes obtained from the SAXS data.  Analysis of the single crystals of the resulting complex powers generated by the electrosynthesis method yield a framework which is depicted in Supplementary Fig. 13 Fig. 7). These ligands grow in pairs in the cavity and perpendicular to each other. They not only act as the connection units to construct two-dimensional layer structure, but also serve as arms protruding both sides of the sheet. The multipoint hydrogen bonding links, existed among lattice water molecules, coordinated aqua ligands and carboxyl groups, further extent the two-dimensional layer into a 3D supramolecular network. Obviously, the π-π stacking interactions between the aromatic rings also play an important role in stabilizing the whole crystal structure 2-7 . The most intriguing feature of the complexes is that the ligands can be assembled around metal centers in diverse arrangement. Such a unique arrangement manner generates a driving force that prompt the edge of a pore to deviate from one plane, thus leading to the formation of the helical shaped channel.

Supplementary note 2. In situ formation and characterization of Cu-Cu 2 O catalysts via electroreduction of the Cu-Complexes on the Cu substrate
We found the Cu-Complexes can undergo in situ reduction during the CO 2 reduction reaction. Thus, the original Cu-complex species change to diffraction peaks of Cu 2 O and Cu, respectively. The ligand is versatile for the fabrication of Cu complexes, which have different morphologies, lattice parameters, and spatial structures. Therefore, the growing interconnected Cu and Cu 2 O grains from the constrained environment of the precursor, the obtained dendritic Cu 2 O/Cu catalytic electrode surfaces were different apparently in morphologies, electrochemical active areas and the ratio of Cu 2 O and Cu on the surface of the catalysts. All of them resulted in the dendritic structure with quantity of the active sites. The structural changes are known from XRD and SEM characterizations (Supplementary Figs. 19 and 20). Concomitantly, The SEM images show dendritic structures with a layer thickness of 20~50 µm which mainly consist of cuprite and copper.
In order to get more information about the morphology of the catalysts and to characterize the phases, high-resolution TEM measurements were performed on Cu-Cu 2 O-1 catalyst after electroreduction of Complex-1 for 5h. The HR-TEM image in Fig.  2g shows a highly crystalline core with a disordered layer, the lattice spacings of the nanoparticle corresponding to the (111) spacing of Cu 2 O (2.47 Å) and the (111) spacing of Cu (2.09 Å). The oxygen atoms were seen to be aligned between the layers in the interstitial tetrahedral sites. The lattice disorder on the surface suggests the existence of oxygen vacancies in the crystal structure. Moreover, the two spacings are seen to be directly related to each other, which indicate high density grain boundary between Cu and Cu 2 O. Therefore, ionic transport properties could be improved by creating a high density of grain boundaries that act as channels to allow ions to enter the particles [9][10][11]     XAFS technique was used to investigate the change in the chemical states of copper species in the catalysts prepared using Complex-1 as the precursor during electrolysis. In this experiment, XAFS was applied to monitor the dynamic evolution of Cu species in the induction period of CO 2 reduction. Supplementary Fig. 22 shows the normalized O Ledge and Cu K-edge XAFS spectra and its first derivative profiles of the selected complex and reference catalysts. In situ Cu K-edge XAFS spectroscopy results show that the energy of adsorption edge (E 0 ) of Cu species increases with increasing oxidation degree. The E 0 at 8979.1, 8980.9 and 8984.3 eV are characteristic of Cu foil, Cu 2 O and CuO reference electrodes respectively. A drastic shrink occurred to the Cu I at 8981.0 eV and Cu 0 peak at 8979.1eV in the first derivative curve after the reduction. Supplementary  Fig. 23 shows that the near-edge oscillation of the complex quickly changed to the pattern similar to that of Cu 2 O after electrolysis.

Comparison study of the other Cu based electrodes
In this work, metallic Cu foam, Cu 2 O and CuO were also used for CO 2 electrolysis. Therefore, purified metallic Cu was used as model Cu 0 , Cu 2 O was used as model Cu I and Fig. 3 shows the catalytic results for CO 2 reduction over various reference electrodes at the reaction time of 5h. It is interesting that all the reference electrodes does not have significant effect on the catalytic activity for C 2 products, the applied potential is also far negative than the in-situ synthesized Cu-Cu 2 O electrode. This strongly supports that synergetic effect occurs between Cu I and Cu 0 species during the CO 2 reduction.

Supplementary
Overview of Cu and oxide derived Cu based electrocatalyst to C 1 products     Supplementary Fig. 30 illustrates the effect of the current density in KCl and KHCO 3 aqueous electrolytes. It is clear that the total current density is higher in KCl electrolyte, and this trend is observed at all applied potentials.
Supplementary Table 11 summarizes the selectivity of product for various Cu-Cu 2 O electrodes in KHCO 3 electrolyte. Interestingly, the electrolysis experiments demonstrate that the product distribution in KHCO 3 aqueous solvent lead to production of 47.6% of ethanol as the major C 2 product and 32.4% of formic acid as the major C 1 product on Cu-Cu 2 O-1 electrode. The most striking observation is that, for all of the catalysts, higher selectivity and partial current densities toward C 2 products can be attained using a KCl electrolyte compared to a KHCO 3 electrolyte at all applied potentials. In Supplementary  Fig. 31, the total current density and FE of ethanol increase initially and then decline as the potential is negatively shifted, reaching the maximum FE at -0.4V vs RHE in 0.5M KHCO 3 aqueous solvent. The products distributions are also similar on other electrodes. These metrics are extremely noteworthy, as the results demonstrate that the changes of the catalytic activity of Cu-Cu 2 O-1 electrode in different electrolytes have also been attributed to the different nature of the solvatized ions in solution. These electrolytes can control the conversion of CO 2 to different product with a level of efficiency and selectivity that is typically only accessible with precious metal catalysts that are both rare and expensive. Similarly, the thickness of the complex film was tuned to optimize the performance of the Cu-Complex by varying the deposition time from 5 min to 2h ( Supplementary Fig. 32). The performance of the Cu-Cu 2 O-1 catalyst in KHCO 3 aqueous solvent also increased with increasing active-site loading until reaching a maximum at deposition time of 1h. Therefore, we can conclude that adding of KCl as electrolyte affects the C 2 product selectivity.    From the 1 H NMR spectra, we can only observe 13 C signal for acetate and ethanol, which indicate that the C 2 products were derived from CO 2 rather than other C-based chemicals in our reaction system. The spectrum of the virgin electrolyte is also given for comparison. No C signal attributed to acetate or ethanol was detected when we used 12 CO 2 as the feedstock. But strong C signal attributed to acetate and ethanol can be found when 13 CO 2 was used. The results also indicate that all the carbon atoms in the product were from CO 2 .

EIS study of the two methods
In this work, we performed the electrochemical impedance spectroscopy (EIS) 86,87 . To measure the solution resistance (R s ), charge transfer resistance (R ct ) between catalyst surface and the reactant, as well as film resistance (R f ) between the Cu-Cu 2 O and substrate (Cu or carbon paper) for Cu-Cu 2 O-1 and Cu-Cu 2 O-3 electrodes prepared by two methods (Cu-Complex that prepared by in situ electrosynthesis and solvothermal methods after electrolysis for 5h). The impedance spectra was recorded in CO 2 saturated KCl aqueous solution at an open circuit potential (OCP) with an amplitude of 5 mV of 10 -2 to 10 -5 Hz (Supplementary Fig. 41). Two electrical equivalent circuits were used in simulation of impedance behavior of the electrode films from the experimentally obtained impedance data. The simulated results are given in Supplementary Table 13. It indicates that R s was nearly independent of the method for preparation of the electrodes. However, the two components of the impedance R ct and R f with in situ synthesized Cu-Cu 2 O were much lower than that prepared by solvothermal method. This indicates that surface of electrode prepared via electrosynthesized method has less effective barriers to charge transfer. It was properly due to better coverage and better adherent of the catalyst to the substrate electrode compared to the solvothermal method. The conductivities of substrates also provide a promising support to enhance the charge transport. Even though coatings of solvothermal method covered a large fraction of the surface of the panels, the poor charge transfer behavior appeared to be dominated by defects in the interfacial conductivity. The EIS result confirms that charge transfer can easily occur on the Cu-Cu 2 O-1 or Cu-Cu 2 O-3 electrodes prepared by electrosynthesis method. Randles' equivalent circuit used for fitting the experimental impedance data of the electrode prepared by solvothermal method: solution resistance (R s ), constant phase element (CPE) of the electrical double layer, electron transfer resistance (R ct ), Warburgtype impedance (Z w ), film capacitance (C f ) and film resistance (R f ). (D) Randles' equivalent circuit used for fitting the experimental impedance data of the electrode prepared by in situ electrodeposition method: solution resistance (R s ), double layer capacitance (CPE dl ), electron transfer resistance (R ct ), film capacitance (C f ), film resistance (R f ) and Warburg-type impedance (Z w ). In situ electrosynthesis method:

Supplementary Table 13. EIS characterization of various electrodes.
Values of the main parameters of Randles equivalent circuit elements obtained by fitting the EIS spectra with Randles' equivalent circuit and R(Q(RW))(CR) and R(C(R(Q(RW)))) at OCP.

A:
Entry Electrode R s (Ω·cm 2 ) R ct (Ω·cm 2 ) R f (Ω·cm 2 )  The result indicates that the active surface Cu I or Cu 0 sites alone do not improve the efficiencies of CO 2 RR and indeed deteriorate the efficiency. It is synergy between surface Cu I and surface Cu 0 that improves significantly the kinetics and thermodynamics of both CO 2 activation and CO dimerization, while making C 1 unfavorable, thereby boosting the efficiency and selectivity of CO 2 RR to C 2 products.  Supplementary note 6. Proposed reaction mechanism for the C-C bond formation

Test with possible intermediates.
With the aim to understand the mechanistic pathway towards the formation of acetic acid and ethanol, some tests of CO 2 reduction were made in the presence of the possible reaction intermediates, such as CO, formic acid, formaldehyde, acetaldehyde and acetic acid. Before testing the system with adding of possible intermediates, the air in cathode compartment was replaced by N 2 via bubbling the electrolyte using pure N 2 . Then possible intermediate was added and electrolysis was started. The results are given in Supplementary Table 14. When CO or formaldehyde was used, acetic acid and ethanol were formed with high rates on Cu-Cu 2 O-1/Cu electrode, respectively, indicating that CO and formaldehyde are a possible intermediate towards the formation of C 2 products. In the presence of acetaldehyde large amount of ethanol was formed, but acetic acid was not detected, indicating that acetaldehyde was a possible intermediate towards ethanol, but not for acetic acid. However, adding of formic acid only produced trace amount of ethanol and acetic acid, suggesting that formic acid was not the main intermediate of C 2 products. Moreover, adding of acetate did not yield detectable reduction product, this result suggests that acetate cannot be reduced further under our experimental condition, which is in agreement with the conclusion of the previous study 88 . All the results demonstrate that CO and formaldehyde are possible intermediates for both ethanol and acetic acid, and acetaldehyde is the possible intermediate for ethanol. The conclusion is the same when Cu-Cu 2 O-3/Cu electrode was used, as shown in Supplementary Table 14. 2. Proposed mechanism for the formation of C 2 products. The experimental results discussed above support the tentative mechanistic pathway. Supplementary Scheme 2 shows the scheme of the possible mechanistic pathway for the electrocatalytic production of acetic acid and ethanol. The three key steps we focus on are (1) CO 2 activation, which we previously showed to be the rate determining step for CO 2˙-free radical; (ii) C 1 product formation, which was found to compete with C 2 products; (iii) CO dimerization, which can be important *COCHO or *COCO intermediate for C 2 products formation. After a first step of reduction with the initial formation of the radical anion CO 2˙-, the radical anion CO 2˙-may strongly interact with the electrode surface depending on the nature of the electrocatalyst. This is a rate-determing step which can be detected from the Tafel slop of the reaction. The catalytic sites may stabilize CO 2˙-, which can further reduce to more hydrogenated species. The experimental evidence suggests that after one proton and one electron transfer, CO 2 could be reduced to either *CHO or *COO adsorbed on the catalytic surface. Early mechanistic studies found that formic acid cannot be reduced to other products, suggesting that the mechanistic pathway toward formic acid is thus separate from the hydrocarbon pathway, which must go through CO. Therefore, the reaction is divided into two paths subsequently. In the formation route of acetic acid, *CO could be further reduced at low overpotential and undergoes a CO dimerization step mediated by electron transfer rendering a *COCHO intermediate. The intermediate *COCHO can further reduce at the electrode surface until a -CH 3 COOspecies is formed. It is noteworthy that these half-reactions occur in a strong reducing environment. That is the electrons coming from the anode side through an external circuit and the protons reaching the cathode from the Nafion membrane in direct contact with the electrocatalyst.
At this point, the CH 3 COOspecies may desorb to form acetic acid. In the formation route of ethanol, the*COCHO or *COCO both can be considered as the precursor for the formation of ethanol. The C-C bond formation is subsequently formed and undergoes a CH 3 CHO intermediate and reduced to ethanol. Therefore, the pathway towards ethanol should be isolated by two routes until reaching the final CH 3 CHO intermediate. This tentative mechanistic pathway is able to explain the experimental observations that we obtained in our electrocatalytic tests.