Promoting CO2 Electroreduction to Acetate by an Amine-Terminal, Dendrimer-Functionalized Cu Catalyst

Acetate derived from electrocatalytic CO2 reduction represents a potential low-carbon synthesis approach. However, the CO2-to-acetate activity and selectivity are largely inhibited by the low surface coverage of in situ generated *CO, as well as the inefficient ethenone intermediate formation due to the side reaction between CO2 and alkaline electrolytes. Tuning catalyst microenvironments by chemical modification of the catalyst surface is a potential strategy to enhance CO2 capture and increase local *CO concentrations, while it also increases the selectivity of side reduction products, such as methane or ethylene. To solve this challenge, herein, we developed a hydrophilic amine-tailed, dendrimer network with enhanced *CO intermediate coverage on Cu catalytic sites while at the same time retaining the in situ generated OH– as a high local pH environment that favors the ethenone intermediate toward acetate. The optimized amine-network coordinated Cu catalyst (G3-NH2/Cu) exhibits one of the highest CO2-to-acetate Faradaic efficiencies of 47.0% with a partial current density of 202 mA cm–2 at −0.97 V versus the reversible hydrogen electrode.


■ INTRODUCTION
Acetate is an important chemical widely used in manufacturing, medicine, and food.The commercial acetate production is mainly based on the thermal carbonylation of methanol and carbon monoxide (CO), which results in a substantial carbon footprint. 1,2The recent development of CO electroreduction has featured a potential alternative means of producing acetate, with a reported acetate Faradaic efficiency (FE) of >70% and partial current density (|j acetate |) over 425 mA cm −2 . 3On the other hand, although the direct carbon dioxide reduction reaction (CO 2 RR) using renewable electricity has the potential of both reducing greenhouse emission and generating valueadded chemicals, the selective CO 2 -to-acetate (or acetic acid) conversion has received only limited progress.
Previous studies in CO electroreduction have suggested that the critical factors of producing acetate include a high coverage of *CO adsorbates on the catalyst surface as well as the formation of ethenone intermediate (*H 2 CCO) under high alkalinity. 3Nonetheless, in CO 2 RR, the *CO intermediate is in situ generated from a CO 2 source, and thus its coverage is generally lower than the direct use of CO reactant. 4In addition, the side reaction between CO 2 and the high alkaline electrolyte also results in the fast depletion of CO 2 molecules and generation of (bi)carbonate that can gradually deactivate the catalyst sites, 5 further inhibiting the retention of high surface *CO coverage.Even when optimizing the Cu-based electrocatalysts to achieve highest multicarbon (C 2+ ) product selectivities of >80%, such as confinement Cu structures 6,7 or metal-doped Cu oxides 8,9 to enhance the binding with *CO intermediates, the main C 2+ products are ethylene 7,8 and ethanol. 6,9To date, the highest partial current density of producing acetate from CO 2 RR is less than 50 mA cm −2 . 10hemical modification of the catalyst surface has recently been investigated to enhance CO 2 capture and increase surface *CO coverage. 11−13 Nonetheless, as *CO is a shared critical intermediate for most of the CO 2 RR products, the scaling relation between different reaction pathways poses a critical limit on the selectivity tuning.For instance, coating of hydrophilic molecules such as polyamides or amino acids on the Cu surface for CO 2 capture also increases the *H coverage on the catalytic sites, which not only promotes the competitive hydrogen evolution reaction 14 but also can facilitate the formation pathway of the *CHO intermediate toward CH 4 . 15−18 Thus, it places a challenging dilemma of both increasing *CO coverage and enhancing the acetate pathway.
To address this dilemma, we propose that a hydrophilic network can be beneficial for the high coverage of *CO on the catalytic sites while retaining the in situ generated OH − to sustain a high local pH environment that favors the *H 2 CCO intermediate toward acetate.Herein, we developed a Cu nanoparticle catalyst functionalized with a highly dendritic polymer with amine (-NH 2 ) tails, designated as G 3 -NH 2 /Cu, as an efficient CO 2 -to-acetate electrocatalyst.Compared to pristine (bare) Cu (Figure 1a), the abundant NH 2 -containing tail chain of the dendritic polymer exhibited a highly intertwined network to capture CO 2 toward higher *CO coverage on Cu, and in the meantime allowed strong coordination with the Cu surface to reserve the in situ generated OH − , thus maintaining a stable high local pH (Figure 1b).The dendritic polymer-coated Cu electrocatalyst enabled an outstanding CO 2 -to-acetate performance, including one of the highest acetate partial current densities (|j acetate |) of 202 ± 14 mA cm −2 with a corresponding Faradaic efficiency (FE acetate ) of 47.0 ± 3.1% at −0.97 V versus the reversible hydrogen electrode (vs RHE), suggesting an attractive strategy of surface molecular engineering to tune the scaling relation of different CO 2 RR products.
The X-ray absorption spectroscopy of the Cu K-edge was then performed to investigate the chemical state and coordination structure.The X-ray absorption near-edge fine structure (XANES) spectroscopy of Cu K-edge in G 3 -NH 2 /Cu and G 3 -OCH 3 /Cu showed different shapes of the rising edge and the post edge (Figure 2c), suggesting their different chemical states and local coordination environments. 24The first edge absorption (the first inflection point, black dash in Figure 2c) of G 3 -NH 2 /Cu (∼8978.0eV) was lower than the Cu foil (∼8979.0eV), 25,26 suggesting the electron transfer from the coordination atoms to Cu in G 3 -NH 2 /Cu.The N 1s XPS spectra were recorded to investigate the chemical states of N-containing groups (Figure 2d).The G 3 -OCH 3 /Cu sample retained a C−N bond (∼399.5 eV), while the N 1s XPS spectrum of G 3 -NH 2 /Cu was deconvoluted into two sub-peaks centered at 399.5 and 398.5 eV, ascribed to C−N and Cu−N interactions, respectively, 27,28 suggesting the electron transfer from N to Cu in G 3 -NH 2 /Cu.
Electrochemical CO 2 RR Measurements.The electrocatalytic CO 2 RR performances of G 3 -NH 2 /Cu, G 3 -OCH 3 /Cu, and pristine Cu were evaluated in flow cells (Methods in the Supporting Information).The G 3 -NH 2 /Cu catalyst not only presented the largest total current density among those three samples (Figure S8) but also showed favorable selectivity toward acetate (Figure S9 and Table S4).The peak acetate partial current density (|j acetate |) reached 202 ± 14 mA cm −2 at −0.97 V vs RHE, with corresponding FE acetate of 47.0 ± 3.1% and a half-cell energy efficiency (EE) of 23.9 ± 1.6% (Figure 3a, b).The FEs of ethylene and CH 4 were measured as 16.3% and 8.0%, respectively.In comparison, the G 3 -OCH 3 /Cu and Cu catalysts showed limited selectivity toward acetate (Figure 3b).For G 3 -OCH 3 /Cu, CH 4 was the main product with a FE CH4 of 73.2 ± 4.7% at −0.97 V vs RHE (Figure 3c and Figure S10), while the FE values of acetate and ethylene were ∼3%.For the Cu catalyst, ethylene was the main product with FE C2H4 of 45.8 ± 2.7% at −0.97 V vs RHE (Figure 3c and Figure S11), and the FE values for acetate and CH 4 were 5.0 ± 0.3% and 20.3 ± 1.1%, respectively (Figure 3c and Table S5).−38 In addition, by tuning the number of amine tail chains by dendrimer generation (G i -NH 2 , i = 1, 2, 3, 4) for optimization of amine network (Table S7), the highest acetate selectivity in C 2 products (i.e., FE acetate /(FE acetate + FE ethylene + FE ethanol )) was obtained from G 3 -NH 2 /Cu (Figure S12 and Table S8).It was observed that while the CO 2 was captured and converted to the *CO intermediate on G 3 -NH 2 , no significant CO 2 RR performance was observed on G 3 -OCH 3 (Figure S13).Furthermore, after >100 h of a continuous electrochemical CO 2 RR test at a constant current density of −400 mA cm −2 , the potential of G 3 -NH 2 /Cu catalyst was stable between −0.9 to −1.1 V vs RHE, and the FE acetate was retained at ∼38.5%, corresponding to ∼82% retention of its initial value (Figure 3e).The morphology (Figure S14) of G 3 -NH 2 /Cu was preserved as before electrolysis, suggesting its electrocatalytic stability.
Mechanistic Investigation.As the electrochemical active surface areas (ECSAs) of the three electrodes evaluated by double layer capacitance (C dl ) 39 were similar (Figure S15), we hypothesized that the distinctive acetate selectivity of the G 3 -NH 2 /Cu catalyst was attributed to the high *CO coverage and high local pH.To verify this hypothesis, density functional theory (DFT) calculations were conducted to investigate the CO 2 -to-acetate pathways in those catalysts.As Cu(111) is the main exposed facet 40,41 as well as the dominant facet in the G 3 -NH 2 /Cu, G 3 -OCH 3 /Cu, and bare Cu, we chose the Cu(111) planes for calculations (Figure S3, Figure S16, Methods in Supporting Information).The adsorption energy of *CO (ΔE *CO ) was first compared on these models, as it affects the *CO coverage on Cu(111) surface. 7Compared to pure Cu(111) (−84.9 kJ mol −1 ), the ΔE *CO values on G 3 -NH 2 / Cu(111) and G 3 -OCH 3 /Cu(111) were calculated as −108.3 and −109.6 kJ mol −1 , respectively (Figure 4a), indicating their enhanced *CO binding capabilities with surface modifications.Then, the *CO-COH coupling pathway and the competitive *CHO pathway were studied. 4,42The *CHO pathway was more efficient than the *CO-COH coupling in G 3 -OCH 3 / Cu(111) (Figure S17a), which led to CH 4 formation.In contrast, both Cu(111) (Figure S17b) and G 3 -NH 2 /Cu(111) (Figure 4b) showed more efficient *CO-COH coupling than the*CHO route, indicating preferable C 2 selectivity.The mechanism was then employed on Cu(111), where acetate is formed through the *H 2 CCO intermediate 4,43 and competes with other C 2 products. 40,44n addition, the Gibbs free energy change (ΔG) value of the *H 2 CCO formation step was 0.73 eV lower than *CCOH formation step on G 3 -NH 2 /Cu(111) (Figure 4b), indicating the acetate route is more favorable than the ethylene or ethanol route.As the adsorption energies of the most favorable configuration of key *CCO and *H 2 CCO intermediates on G 3 -NH 2 /Cu(111) were lower than those on Cu(111) (Figure S18 and Table S9), the *H 2 CCO species stabilized on G 3 -NH 2 /Cu(111) can promote the CO 2 -to-acetate conversion, while the main product of bare Cu(111) was C 2 H 4 , consistent with our experimental results.Thus, the functionalization of G 3 -NH 2 on Cu(111) not only led to a strong *CO binding capability for increasing the surface *CO coverage but also enhanced the adsorption of key *H 2 CCO intermediates toward acetate production.The in situ electrochemical surface-enhanced Raman spectroscopy (SERS) was further performed at potentials ranging between −0.37 and −1.07 V vs RHE to confirm the contribution of *CO coverage and OH − confinement on G 3 -NH 2 /Cu during CO 2 RR (Methods in Supporting Information).For G 3 -NH 2 /Cu (Figure 5a,b), the Raman peaks of C− N stretching (∼1200 cm −1 ), C−C stretching (∼1320 cm −1 ), CO 2 − stretching (∼1330 and 1545 cm −1 ), N−H stretching (∼1595 cm −1 ), and N−H deformation (∼1609 cm −1 ) showed gradually increasing signals than G 3 -OCH 3 /C at more negative potentials, 11 confirming the continued CO 2 capture by the G 3 -NH 2 network during CO 2 electrolysis. 45,46This result was also verified with pure G 3 -NH 2 and G 3 -OCH 3 as electrodes for CO 2 RR at different applied potentials ranging from −0.37 to −1.07 V vs RHE (Figure S13).
In addition, the G 3 -NH 2 /Cu exhibited clearer Raman peaks located at 270−350 and 1800−2100 cm −1 (Figure 5c), attributed to atop-adsorbed *CO (*CO atop , ∼334 and 2070 cm −1 ) and bridge-adsorbed *CO (*CO bridge , ∼275 and 1850 cm −1 ). 47As shown in Figure 5c, the*CO atop peaks located at 2000−2100 cm −1 showed a red-shift from −0.57 to −0.77 V vs RHE, indicating the *CO atop vibration was affected by the vibrational Stark effect. 48In addition, the gradual blue shift of *CO atop peaks and the appearance of *CO bridge peaks at a more negative potential of −0.77 to −1.07 V vs RHE, suggesting the increased higher *CO coverage. 4,49Furthermore, a weak peak at ∼1970 cm −1 was also detected on G 3 -NH 2 /Cu, which was identified as the C�C�O stretching of an *CCO intermediate toward acetate. 47In comparison, for G 3 -OCH 3 / Cu (Figure 5d) and Cu (Figure S19), the *CO bridge peaks were not observed and the *CO atop peaks had much weaker intensities, suggesting their low surface *CO coverage.
Moreover, the Raman peaks observed on both G 3 -NH 2 /Cu and Cu showed O−H bending peaks (∼525 cm −1 ) with the increase of applied negative voltages, assigned to the in situ generated *OH species. 50Those peaks were assigned to the O−H bending modes of the surface *OH species, which are hydrogen-bonded with surrounding water molecules, as suggested by previous Raman studies. 51These peaks were red-shifted from 533 to 517 cm −1 with applied more negative potentials from −0.67 to −1.07 V vs RHE on G 3 -NH 2 /Cu (Figure 5e), implying that they were affected by the vibrational Stark effect with a Stark tuning rate of 41 ± 1.5 cm −1 /V (Figure S20). 47,51,52hile on the Cu surface, another C−O stretching (∼1067 cm −1 ) was observed for applied negative potentials as small as −0.47 V vs RHE (Figure 5f), corresponding to the CO 3 2− species. 16,52,53However, this peak only became observable on G 3 -NH 2 /Cu for potentials more negative than −0.97 V vs RHE along with another peak at ∼1035 cm −1 corresponding to *COOH (Figure 5e). 50The carbonate accumulation on the Cu surface lowers the local pH 5 and induces a higher chemical potential (Table S10), which hinders the H 2 CCO-to-acetate pathway. 54Taken together, the G 3 -NH 2 /Cu catalyst was allowed to confine the in situ generated OH − on the Cu surface due to the repulsion between OH − and G 3 -NH 2 /G 3 -NHCOO − during the CO 2 electroreduction.Subsequently, the confined OH − reacted with H 2 CCO to form acetate, instead of reacting with CO 2 to form CO 3 2− . As a result, the G 3 -NH 2 /Cu enabled a high surface *CO coverage and high local pH with in situ generated OH − to be achieved, which facilitates the production of a *H 2 CCO intermediate toward acetate, thus boosting the CO 2 -to-acetate electrosynthesis with a peak FE acetate of 47.0%, a 9.4 times improvement than that of Cu.

■ CONCLUSION
In summary, we have demonstrated a -NH 2 -tailed, dendrimerfunctionalized Cu surface aiming to enhance the CO 2 capture and increase the local *CO concentration.The -NH 2 -rich network allowed an increase of the *CO intermediate coverage on the Cu catalytic sites, while at the same time retained the in situ generated OH − and a high local pH environment, favoring the formation of a *H 2 CCO intermediate toward acetate.The catalyst exhibited a high CO 2 -to-acetate performance, with an FE acetate of 47.0% and corresponding partial current density of 202 mA cm −2 .Our work suggests an attractive strategy of surface molecular engineering to tune the selectivity of acetate from CO 2 RR.

Figure 1 .
Figure 1.Schematic of the CO 2 RR process on (a) Cu and (b) -NH 2 -terminal, dendrimer-functionalized Cu.The abundant -NH 2 -terminals of the dendritic polymer provide a highly intertwined network to capture CO 2 for high *CO coverage on Cu and allow strong coordination with the Cu surface to retain the in situ generated OH − , thus favoring the CO 2 -to-acetate selectivity.

Figure 3 .
Figure 3. (a) Partial current densities (the upper panel) and corresponding FE values (the lower panel) of CO 2 RR products using the G 3 -NH 2 /Cu catalyst at various applied constant potentials (without ohmic correction).(b) EE values for acetate at various applied constant potentials with the G 3 -NH 2 /Cu, G 3 -OCH 3 , and Cu catalysts.(c) FE values at −0.97 V vs RHE on each catalyst.(d) Summary of |j acetate | vs FE acetate of this work with other Cu-based CO 2 RR catalysts.(e) The stability obtained at a constant negative current density of −400 mA cm −2 (without ohmic correction).Error bars in (a−c) correspond to a mean ± standard deviation of >3 measurements.