Constructing of an interfacial microenvironment through CO₂-reactive IL

Linear sweep voltammetry (LSV) was performed with neat [EMIM][2-CNpyr] to investigate the electrochemical behavior of the IL under N₂ and CO₂ on the Cu surface, as shown in Figure 2A. Under CO₂ environment, a reduction potential at −1.8 V vs. Ag/Ag⁺ is observed and attributed to CO₂RR since it was absent under N₂. In addition, the open circuit potential (E_oc) shifts from −1.0 to −1.11 V. This is associated with the changes in the pH and composition of the electrolyte due to CO₂ absorption. The surface species are [EMIM]⁺, 2-CNpyrH, carbamate, and carboxylate, as shown in Figure 1, formed upon CO₂ chemisorption. Specifically, [EMIM]⁺ adsorption is evident as more clearly seen under the N₂ environment where the current gradually increases in the potential range of −1.0 V to −2.1 V. Beyond −2.1 V, current increases sharply due to the reduction of [EMIM]⁺ to an adsorbed radical or carbene.⁴⁰ Other general observations are the lower current under CO₂ compared to N₂ and the shallower slope in current beyond −1.8 V. This is due to the complexation with CO₂ that results in increased viscosity and decreased conductivity; consequently, species diffuse slower to the electrode surface, slowing charge transport rate.

Differential capacitance (C_diff) measurements were performed to probe the electrode-electrolyte interface, as shown in Figure 2B. The capacitance curve starts with a plateau at −1.1 V vs. Ag/Ag⁺ followed by a decay in capacitance with negative polarization up to −2.0 V and another plateau at potentials more negative than −2.0 V where a minimum of 0.25 μF.cm⁻² is reached for the Cu|[EMIM][2-CNpyr] interface under both N₂ and CO₂. Initial capacitance of the interface is lower under CO₂, suggesting a difference in the interface, consistent with the shift in E_oc. As increasing negative polarization causes more and more cations to accumulate on the electrode surface, [EMIM]⁺ cations crowd the interface, as interpreted from the Goodwin-Kornyshev model which treats the interface through the mean field theory.⁴¹ Accordingly, the charge density at the interface increases with increased applied potential until the ions feel their excluded volumes. Since the neat IL has no solvent molecules to displace, it can only admit voids upon increased polarization. However, when the cations admit all the voids, further increase in potential results in the increased double layer thickness, known as crowding, which leads to decrease in C_diff as seen in Figure 2B.

To further investigate the nature of C_diff and the interfacial species, in situ SERS was performed with N₂ and CO₂-saturated electrolytes (Figures 2C and 2D). Table S1 summarizes the observed peaks and vibrational assignments. The prominent peaks are consistent with both N₂ (Figure 2C) and CO₂ (Figure 2D), and the spectra point to the co-existence of both the anion and the cation on the electrode surface. However, as discussed below, there are notable changes related to the [EMIM]⁺ orientation with increased polarization and compositional changes with CO₂ absorption.

Spectral peaks related to [EMIM]⁺ ring located at 1347, 1380, and 1423 cm⁻¹ under N₂ (Figure 2C) intensify with applied polarization, indicating a stronger interaction with the electrode surface and further blueshift to 1348, 1383, and 1432 cm⁻¹, respectively, under CO₂ (Figure 2D), confirming the change in its environment due to CO₂ absorption. In parallel, [2-CNpyr]⁻ ring stretching at 1398 cm⁻¹ (Figure 2C) that is weakly seen under N₂ intensifies and shifts to 1405 cm⁻¹ (Figure 2D). This is due to the closer approach of the neutral 2-CNpyrH specie that forms upon CO₂ saturation to the surface. The presence of 2-CNpyrH upon CO₂ exposure is also evidenced by the shift in the −C≡N stretching vibration at 2189 cm⁻¹ under N₂ to 2223 cm⁻¹ under CO₂. The peak at 1423 cm⁻¹ under N₂ and 1432 cm⁻¹ under CO₂ are attributed to the symmetric stretching of the CH₃ group. Since it presents very low enhancement with negative polarization, CH₃ groups are interpreted to be away from the surface, pointing towards the bulk electrolyte. On the contrary, the peaks at 1116, 1342, and 1380 cm⁻¹ associated with C₄ and C₅ of [EMIM]⁺ and N−CH₂ moiety of the ethyl group increase in intensity with negative polarization, as seen in Figure 3A, indicating a change in orientation of the cation. This behavior is persistent with N₂ and CO₂, although it is better seen under N₂ without interference from complexation with CO₂. Since the surface is already crowded by [EMIM]⁺, the intensification of these specific ring modes is interpreted to be due to the changes in the orientation of [EMIM]⁺ from tilted (C₂ position facing the electrode surface) to parallel configuration with respect to the electrode surface.

The potential at which [EMIM]⁺ transitions from tilted to parallel can be easily seen from the recorded intensities of the associated peaks plotted in Figure 3A. Specifically, the intensities of the 1342 and 1380 cm⁻¹ peaks associated with N-CH₂ stretching increase sharply beyond −1.5 V vs. Ag/Ag⁺, which also coincides with the potential where the decrease in capacitance observed in Figure 2B during the negative potential sweep in EIS. When compared between N₂ and CO₂, the relative enhancement in the peak intensities used to identify the orientation change is lower under CO₂ as the [EMIM]⁺−CO₂⁻ adduct forms. The change in the orientation of Im-species is further supported by IC-QM/MM calculations of the optimized geometries as shown in Figure 3B. At −1.0 V, the lowest energy structure shows a tilted orientation, while at −1.7 V the lowest energy structure changes to a parallel orientation. IC-QM/MM calculations at −1.5 V still show an energetic preference for the tilted orientation, however there is a clear transition to the parallel preference at more negative potentials, marking −1.5 V vs. Ag/Ag⁺ as the potential of transition. This trend continues into −2.0 V where the preferred orientation is parallel as shown in the relative energies given in Table S2.

In the light of SERS analysis, the interfacial liquid structure can be considered a cation-rich layer infiltrated with anions under N₂. Under CO₂, the surface is dominated by [EMIM]⁺, [EMIM]⁺−CO₂⁻, 2-CNpyrH, and fewer [2-CNpyr]⁻. The crowding and thickening of the interface with [EMIM]⁺ upon negative polarization accompanied by reorientation are believed to impair the ability of the anion to escape from the interface, especially when the neutral form of the anion (2-CNpyrH) gets reduced, the effect of which is discussed later. Therefore, we observe features of the anion in SERS at 2189 cm⁻¹ under N₂ and 2194 cm⁻¹ under CO₂. The existence of the pyrrole species contributes to the formation of a unique microenvironment for CO₂RR since the anion also complexes with CO₂ ([2-CNpyr]−CO₂⁻) and can serve as a CO₂ carrier to the electrode surface. In addition, the protonated anion, 2-CNpyrH, can act as HBD.

CO₂RR catalytic behavior modulated by the microenvironment

Neat [EMIM][2-CNpyr] has a viscosity of 68 cP at 25 °C, which increases to 247 cP upon saturation with CO₂;²⁵ these high viscosities is a limiting factor for CO₂RR. Therefore, acetonitrile was used as a diluent to minimize the mass transfer limitation and increase the ionic conductivity in CO₂RR. The concentration-dependent ionic conductivity and the viscosity of diluted [EMIM][2-CNpyr] (0.0 to 3.0 M) are provided in Figure S3. A maximum ionic conductivity at 1.5 M and a consistent increase in viscosity with increasing IL concentration was observed. Because the measured ionic conductivity of [EMIM][2-CNpyr]/acetonitrile decreased after CO₂ saturation, 0.1 M tetraethylammonium perchlorate (TEAP) was also added as the supporting electrolyte to maintain a stable ionic conductivity. Therefore, further analysis of the interface and CO₂RR with IL/acetonitrile electrolytes was performed with 0.1, 0.5, and 1.0 M concentrations of IL and 0.1 M TEAP supporting salt.

Interestingly, the general behavior of the C_diff curves is nearly identical between the IL electrolytes in acetonitrile (Figure 4A) and with the neat IL (Figure 2B), which suggests that the ionic microhabitat and the structure of the double-layer formed on the electrode surface are similar. With 0.1 and 0.5 M IL electrolytes, the decay in capacitance with a negative potential sweep gradually starts at about −1.5 V vs. Ag/Ag⁺. The decline in capacitance occurs at a notably faster rate with 1.0 M IL. This decay further confirms the higher [EMIM]⁺ concentration covering the electrode surface, thickening the double-layer faster than the relatively diluted electrolytes (0.1 and 0.5 M IL), and impeding the diffusion of new IL−CO₂ adducts towards the electrode surface, i.e., increased mass transfer overpotential. In addition to the similarities in C_diff, the comparison of the SERS spectra obtained from the neat (Figure 2C and 2D) and diluted IL electrolytes (Figure S4) further confirm the similar microenvironment for CO₂RR where Im-species (i.e., [EMIM]⁺ and [EMIM]⁺−CO₂⁻) dominate the interface with entrapped [2-CNpyr]⁻ and HBDs (2-CNpyrH). This microenvironment is unique compared to prior literature involving Im-based ILs since more active species exist in the interface along with native HBD (in situ generated due to CO₂ absorption).

The LSV measurements under CO₂ suggest the onset potential for CO₂RR is at about −1.9 V vs. Ag/Ag⁺ for all the IL electrolytes studied (Figure 4B). Consistent onset potential implies the activation energy for the charge transfer does not depend on IL concentration. However, the IL concentration significantly impacts CO₂RR current densities, similar to our previous report with the Ag electrode.²⁴ Increasing the IL concentration from 0.1 to 0.5 M results in a more than two-fold increase in CO₂RR current due to the increased availability of CO₂ at the electrode surface. However, the catalytic current decreases with further increase in IL concentration to 1.0 M. [EMIM]⁺ was reported by Urushihara et al.⁴² to cover the entire surface even at low concentrations. Therefore, the decrease in CO₂RR activity is attributed to the surface layer formation by the cation and thickening of the double layer which prevents easy access of reactants to the electrode surface, thus increased mass transfer overpotential and lowered catalytic activity.

With respect to the role of the IL in lowering the overpotential needed for CO₂RR, one widely acknowledged explanation involves the formation of the IL−CO₂ adducts, i.e., carboxylate formation at the interface with CO₂ binding to the C₂ position of the Im ring. The facilitation of CO₂RR by the carboxylate results in a decreased overpotential compared to its absence;⁴³ however, after a particular concentration, the reaction becomes impaired, and the rate slows down, as seen with 1.0 M IL in Figure 4C, due to the double layer dynamics discussed above. Another explanation of the role of the IL focuses on the hydrogen bonding that also modulates the electrical field and interaction with the intermediates at the interface, especially with CO₂⋅⁻. To probe specifically the impact of the hydrogen bonding network on CO₂RR at the interface, we performed CV measurements with additional 2-CNpyrH (0.1 and 0.5 M) to vary HBD concentration independently in the IL electrolyte (Figure 4C).

We hypothesized that incorporating HBDs could enhance the CO₂ reduction activity by increasing the -COOH intermediate formation. As seen in Figure 4C, additional 2-CNpyrH at 0.1 M does not influence the CO₂RR onset potential whereas it increases the catalytic current, compared to 0.1 M IL electrolyte without the HBD addition. When the 2-CNpyrH concentration was increased to 0.5 M, the catalytic activity also increased along with a positive shift in the onset potential by 120 mV. These results suggest a substantial modification to the energetics of the first proton-coupled electron transfer step in CO₂RR (i.e., formation of -COOH) as a function of HBD concentration.

The enhanced catalysis by HBD is also evident from the SERS spectra shown in Figure 4D, where the surface adsorbed CO peak (*CO_ads in 2046–2088 cm⁻¹ range) gains intensity at potentials more negative than −1.5 V vs. Ag/Ag⁺ compared to that without HBD (full spectra is given in Figure S5). The observed broad peaks indicate the presence of intramolecular C≡O stretching due to the binding of CO to the Cu electrode surface, consistent with multiple binding configurations.⁴⁴ Furthermore, the results suggest that an optimal ratio of [EMIM]⁺ (hence carboxylate) and HBD may exist, highlighting the significance of the ionic microenvironment on the electrode surface for CO₂RR and demonstrating a new approach to tune the microenvironment in non-aqueous electrolytes.

In complement, computational analysis of the simplified interface shown in Figures 4E and 4F using IC-QM/MM optimized geometries of [EMIM]⁺−CO₂⁻ with 2-CNpyrH and [EMIM]⁺−CO₂⁻ with 2-CNpyr, respectively, at −1.7 V demonstrate the hydrogen bonding in the presence of the native HBD. The HBD (CNpyrH) prefers to orient itself towards the carboxylate moiety interacting through hydrogen bonding and stays close to the surface while [EMIM]⁺−CO₂⁻ prefers a tilted orientation, consistent with the lower increase in intensities under CO₂ (Figure 3A). Upon deprotonation and the formation of the anion, the [CNpyr]⁻ prefers to move away from the negative Cu surface, however, gets trapped in the microenvironment due to the double layer thickening as seen from 2194 cm⁻¹ feature in Figure 2D. From the computed anion-cation configurations, it is also seen that there is still hydrogen bonding between [CNpyr]⁻ and [EMIM]⁺−CO₂⁻ through the C4, C5 protons of Im. This causes the cation to move more towards a parallel orientation as in Figure 4F.

To distinguish between the microenvironments formed by different electrolyte concentrations, constant potential electrolysis experiments were conducted at −2.1 V vs. Ag/Ag⁺ for 3 hours in a two-compartment cell (Figure S6) coupled with in-line GC analysis. The gaseous products were identified and quantified by in-line GC (example in Figure S7) while the liquidus products were analyzed after electrolysis by NMR (example in Figures S8). Table S3 summarizes the quantified products from electrolysis. The catholyte was the IL electrolyte of interest and the anolyte was aq. 0.1 M H₂SO₄ serving as the proton source. The calculated Faradaic efficiencies (FE) for H₂ and C₁ products, such as CO, formate, and CH₄, are given in Figure 5A, and that of the C₂₊ products, such as succinate, C₂H₄, and C₂H_6, are shown in Figure 5B. Selectivity for CO (C₁, 2e⁻) increases from 23 % to 43 % as the concentration of IL increases from 0.1 M to 0.5 M, as seen in Figure 5A. Further increase in the concentration slightly decreases the selectivity for CO to 37 %. High local CO₂ concentrations cause premature CO desorption from the Cu surface and lead to high CO selectivity. Increasing IL concentration increases CO₂ concentration on the electrode surface, leading to CO desorption.⁴⁵ Furthermore, [EMIM]⁺ adsorption may also replace the CO_ads on the electrode surface, causing high CO selectivity.

The second major product is formate (C₁, 2e⁻), with 45 % FE for 0.1 M IL, 26 % FE for 0.5 M IL, and 36 % FE for 1.0 M IL. The spectral evidence to -COOH in complement to significant formate presence in products strengthen the argument for the role of the proton dynamics between the Im species and HBD in controlling the mechanism leading up to CO formation and beyond in these non-aqueous electrolytes. As seen in Figure 5A, the selectivity of formate decreases from 45 % to 26 % when the electrolyte concentration is increased from 0.1 M to 0.5 M IL. This decrease is accompanied by an increase in the selectivity towards succinate (C₄, 14e⁻), as seen in Figure 5B. Succinate has a relatively low selectivity with 6 %, 11 %, and 4 % FE for 0.1 M, 0.5 M, and 1.0 M concentrations, respectively. The initial increase in selectivity towards succinate may be due to the −COOH coupling that proceeds it; however, further increase in the concentration of −COOH does not result in increased FE for succinate, likely due to the over saturation of the electrode with [EMIM]⁺ as discussed earlier, thus preventing the C₂−C₂ coupling. FE for H₂ is small, less than 10 %, since HER is suppressed in IL containing electrolytes, as generally reported in the literature⁴⁶ and most of the observed HER is from water cross-over from the anolyte. However, HER enhances when HBD is introduced at 0.5 M (39 % FE) suggesting a high surface coverage by H⁺, which further explains the positive potential shift observed in Figure 4C. This finding underlines the role of HBDs in CO₂RR selectivity and further suggests that selectivity can be tuned by IL : HBD composition and by replacing the aqueous anolyte with a non-aqueous system as the proton source. Increased H⁺ availability due to HBD also promotes the formation of CH₄, C₂H_4, and C₂H₆, as seen in Figure 5B. Faradaic efficiencies in Figure 5 add up to 90 % of the total efficiency. Apart from the reported products in Figure 5, butane and formaldehyde were also observed in GC (Figure S9) and NMR (Figure S10), respectively, in negligible quantities. Table 1 compares the obtained Faradaic efficiencies with previous reports utilizing IL electrolytes in CO₂RR on Cu-based electrodes. Uniquely, C₄ products are obtained in this study owing to the constructed microenvironment presented.