Potential-Dependent CO2 Electroreduction Pathways on Cu(111) Based on an Improved Electrode/Aqueous Interface Model: Determination of the Origin of the Overpotentials.

Potential-dependent CO2 electroreduction pathways on Cu(111) are systematically studied with the aim of applying an improved electrode/aqueous interface model in this paper. The results indicate that our present defined CH2O and CHOH pathways may be able to parallelly take place at low overpotentials. Notably, the applied potentials will not alter the optimal CO2 reduction mechanisms. However, the presence of high overpotentials makes CO2 electroreduction more favorable, thus explaining why high overpotentials at experiments are required during CO2 electroreduction on Cu. Based on the potential-dependent energetics, the results suggest that COOH and CHO intermediates may be unstable at low overpotentials, in which COOH can easily change back to CO2 and CHO can easily change back to CO, thus preventing CO2 electroreduction. However, the high overpotentials will facilitate the formation and further electroreduction of CO and CHO. Thus, we can speculate that CO formation and then further electroreduction into CHO are the possible potential-limiting steps during CO2 electroreduction, which are regarded as the origin of experimentally observed high overpotentials. The present comprehensive understanding on CO2 electroreduction pathways can provide theoretical guidelines for efficiently designing Cu-based alloy electrocatalysts operated under the conditions of low overpotentials.

Introduction

Cu is the only monometallic electrocatalyst that can electrochemically reduce CO2 into valuable fuels and chemicals with reasonable current density, such as hydrocarbons CH4 and C2H4.[1−3] The selectivity of hydrocarbon products had been revealed on Cu electrodes by early experiments, which is crystal face-dependent.[4−8] Especially, CH4 is more facile to form on the close-packed Cu(111) surface because of its greater selectivity to C1 species than the more open (100) crystal face of Cu. However, the practical application of CO2 electroreduction on the Cu electrodes is hindered because of required high overpotentials (ca. 1 V vs RHE).[5] Furthermore, complexities of the electrochemical environment make electroreduction mechanisms of CO2 remain elusive.[6−13] Therefore, comprehensive understanding on CO2 electroreduction pathways and rate-limiting steps will help rationally design Cu-based alloy electrocatalysts.

Although some spectroscopy techniques had provided some important insights into the nature of adsorbed species during CH4 production, such as CO2, CO, and HCOO– and speculated that intermediate CH2 is a precursor at the electrode/aqueous interface,[14−16] experimental understanding of the CO2 electroreduction mechanism has been challenged and impeded for lack of in situ spectral characterization technologies. On the basis of previous electrokinetic experiments, thus, only very limited mechanistic insights into CO2 electroreduction can be concluded. At present, to obtain various crucial electrocatalytic reaction mechanisms, theoretical calculations have become a unique tool.[17,18] The favored intermediates that are not accessible at experiments for product formations can be identified by the calculated results and thus mechanistic insight can be offered. Unfortunately, complete understanding on the major electroreduction pathways of CO2 on Cu(111) that result in product CH4 is still not achieved and disagreements still exist because of complexity of electrode/aqueous interfaces despite numerous theoretical efforts based on the various theoretical methodologies, especially the determination of key intermediates, such as CHO, COH, CH2O, CHOH and so forth.[19−32] Thus, the simulation of an electrochemical interface is still a subject of ongoing discussion, in which the early experimental results need be employed to examine and confirm whether the proposed computational methods for describing the electrochemical interface are accurate enough and identify the lacking intermediate in the indeterminacy of the electrocatalytic reaction pathways.

In this paper, an improved electrode/aqueous interface model based on a model of CO coverage with two explicit relaxed H2O bilayers is used to study CO2 electroreduction pathways, in which the formation of CH4 production of Cu(111) surface is taken as an example as the (111) facet of Cu exhibits a high selectivity for CH4. In fact, the previous study conducted by Nørskov et al. had showed that using of a model of two relaxed H2O bilayers can rationally model the interactions between solvent and adsorbates and determine the kinetic barriers by calculating the capacitance across the double electric layer.[33] Furthermore, our most recent study has elaborated the interactions between the present used solvation model and adsorbents.[34,35] As a part of our effort, the onset potentials and potential-dependent energetics for each elementary step are calculated at electrode/aqueous interfaces based on the previous theoretical models. The excellent consistencies between the present calculated results and available experimental data and partial theoretical calculations confirm the validity of the present used model and method.

Results and Discussion

CO Coverage-Dependent Equilibrium Potentials

The relationship between CO coverage and equilibrium potentials has been given in our most recent study.[36] As can be seen in Figure a, the reasonable linear relationship is exhibited on Cu(111) between the differential adsorption energy of CO, ΔE(θ), and CO coverage (θCO), suggesting that CO adsorption may approximately follow the Termkin adsorption isotherms on Cu(111). We can calculate ΔE(θ) at any θCO through linearly fitting the ΔE(θ) ≈ θCO data to a straight line. Thus, the linear relationships between the calculated equilibrium potentials (U) and CO coverage (θCO) on Cu(111) can be obtained based on eq , as shown in Figure b. It can be observed that more CO coverage will result in more negative equilibrium potential for CO2 electroreduction on the Cu surface, implying that the overpotential may derive from the adsorbed CO. The thermodynamic equilibrium potential is calculated as ca. 0.27 V (vs RHE) when θCO is close to zero, which is comparable with the previous proposed thermodynamic equilibrium potential of 0.17 V for CO2 electroreduction, partially confirming the reasonability of the present used computational model.[25] In our present investigation, CO2 electroreduction pathways on Cu(111) will be first focused at the low CO coverage of 1/9 ML for comparison, which corresponds to the calculated equilibrium potential of ca. 0.14 V and the conditions of low overpotentials. Considering the expensive computational cost, further analysis will be carried out at more negative applied potentials based on the approach from Nørskov et al. and Asthagiri et al. to determine the effect of applied potentials on kinetic barriers and reaction free energies (see Computational Details).[24−26,29]

Figure 1

(a) Relationships between the differential adsorption energy of CO, ΔE(θ) and CO coverage (θCO); (b) relationships between the calculated equilibrium potentials (U) and CO coverage (θCO).

CO2 Electroreduction Pathways at Low Overpotential

Although CO2 reduction mechanisms at low overpotential have been given in our most recent work,[36] they are still elucidated in this paper for comparison. Thereinto, CO2 electrochemical reduction pathways are simulated via direct transfer of surface adsorbed H atoms to adsorbed species and the previous proposed H-shuttling mode forming acid hydronium ion is not considered as electrocatalytic CO2 reduction generally occurs in nonacidic conditions.[1−5,10,37] As shown in Figure S1, formation of the intermediate trans-COOH is more favorable in the first elementary reaction step through direct hydrogenation of physical adsorbed CO2 as relatively lower activation barrier of ca. 0.87 eV is required than cis-COOH formation (ca. 1.20 eV). The subsequent hydrogenative dissociation of trans-COOH can produce the key intermediate CO along with H2O formation with a low and surmountable barrier of ca. 0.20 eV at room temperature. The minimum energy pathways (MEPs) analysis is also performed for CO2 hydrogenative dissociation into CO in order to further validate the possibility of the trans-COOH formation. For this pathway, an almost identical barrier of ca. 0.87 eV is obtained with trans-COOH formation, implying that trans-COOH is a key intermediate during initial CO2 electroreduction into CO. The COOH species was also speculated to be a possible intermediate in the previous theoretical studies involving various solvation models.[29,30,32,38] Furthermore, our calculated barriers using the present electrode/aqueous interface model exhibit excellent consistencies compared with the earlier report form Asthagiri et al. for the first elementary step during CO2 electroreduction.[32]

For production of hydrocarbons CH4, further electroreduction of CO is one of the crucial steps. The direct hydrogenation of CO can lead to the formation of intermediates CHO and COH on Cu(111) and the barriers are calculated as 0.85 and 1.39 eV, respectively, as can be seen in Figure S2, showing that CHO formation is more facile to occur at low overpotential. The present conclusions are coincident with earlier studies from Nørskov et al. based on density functional theory (DFT) calculation where CHO can be more easily formed than COH on Cu(111).[25−27] However, the disagreeable result was also obtained by others previous theoretical reports,[29,30,32] namely, COH formation is more favorable via the proton–electron transfer reaction at an applied potential of ca. −0.60 V (vs RHE) because of the shorter path between the hydrated proton and the O atom of CO, finally resulting in CH4 formation through sequential electroreduction. By applying a similar potential and an identical solvation model on Cu(111) with our present paper, in fact, our most recent study also showed that the barrier of CHO formation via proton–electron transfer is notably lower than COH formation (ca. 0.98 vs 1.83 eV).[39] The preferred CO adsorption configuration that adsorbed via C atoms may be able to explain why CHO pathways are more favorable on Cu(111), in which the direct electronic donation from the Cu surface to adsorbed CO can lead to more localized electrons on the C atom. Thus, the surface-adsorbed H atom or solvated proton tends to reach the C atom of CO to form CHO. Our present result for CO electroreduction into CHO exhibits excellent inconsistencies compared with the most recent experimental report from Koper et al. for CO2 electroreduction into CH4 on Cu(111).[8,13] The difference of the solvation model may lead to the disagreements between earlier theoretical and our present studies. Only a single H2O molecule and fixed H2O bilayer model were employed in earlier studies, which are inadequate to capture and describe the solvent effect.

Further hydrogenation of CHO may be able to produce intermediates CH2O and CHOH. The almost identical barriers of ca. 0.45 eV suggest that further electroreduction of CHO into CH2O and CHOH both are favorable at low overpotential and facile to take place because of surmountable barriers at room temperature, as shown in Figure S3. Further electroreduction of CH2O may be able to form CH2OH and CH3O and the calculated barriers are 0.32 and 0.23 eV, respectively. This extremely low value and subtle difference make the formations of these both intermediates favorable (see Figure S4). CHOH hydrogenation is also able to form CH2OH and a low barrier of ca. 0.38 eV is required (see Figure S5). Simultaneously, a possible pathway is proposed, namely, CHOH hydrogenative dissociation into intermediate CH along with H2O formation, which is almost a nonactivated process with the barrier of only ca. 0.19 eV, suggesting that CH species is also favorable on Cu(111) and it opens a pathway via CHO species that can lead to CH4 production. However, the previous DFT studies conducted by Asthagiri et al. showed that CHO only can lead to CH3OH production because of a notably high barrier for CHOH reduction into CH, whereas COH leads to CH4 production.[12,29,32] The abovementioned difference of the solvation model can be also used to explain the inconsistencies between earlier and present theoretical studies. Our calculated barriers are comparable with some previous reported theoretical values at electrochemical interfaces,[29,37,40] validating the rationality of our present employed electrode/aqueous interface model to some degree.

There may be three possibilities for further CH2OH electroreduction, including direct hydrogenation into CH3OH, hydrogenative dissociation along with H2O formation, and direct dissociation into CH2 through cleavage of the C–O bond. The results indicate that CH2OH reduction inclines to produce CH2 along with H2O formation because of an extremely low barrier of ca. 0.07 eV (see Figure S6). CH3OH formation kinetically may be prohibited because of relatively higher barriers for further CH2OH and CH3O electroreduction (see Figures S6 and S7). The present barriers for CH2OH electroreduction into CH3OH and CH2 along with H2O formation by surface hydrogenation are comparable with those of corresponding processes through proton–electron transfer reaction from Skúlason et al.,[30] in which surface-adsorbed H atoms are first formed by the electron-coupled proton transfer via a Volmer reaction and then hydrogenation of the surface-adsorbed intermediates results in formation of the corresponding species. The agreement of barriers may be able to confirm the validity of our present employed direct transfer model of surface-adsorbed H atoms. Further CH hydrogenation is also able to form CH2 species with a surmountable barrier of ca. 0.33 eV at room temperature. Subsequently, CH4 can be produced by CH2 sequential hydrogenation. The considerably higher barrier of ca. 1.31 eV for CH4 formation may be able to be attributed to the observed physisorption state of CH4 on Cu(111) (see Figure S8). The calculated barriers are also comparable with the earlier report from Asthagiri et al.[29] On the basis of the energies in Table S1, two possible mechanisms, namely, CH2O and CHOH pathways, can be defined for CO2 electroreduction. As can be seen in Figure , the identical barrier of ca. 0.87 eV is observed for these two possible mechanisms by comprehensively considering on the overall energy profiles, suggesting that CH2O and CHOH pathways through common intermediates CHO and CH2 may be able to parallelly take place on Cu(111) at a low overpotential. CH3OH formation is also thermodynamically less favorable as CO2 electroreduction into CH4 is a stronger exothermic process. Thus, the present study explains why only CH4 is experimentally observed on the Cu electrodes at electrochemical interfaces, rather than CH3OH.[5,41] Although CH3 final electroreduction into CH4 has a significantly higher barrier than others’ steps, the strongly exothermic process during CH2 sequential electroreduction makes production of CH4 easily occur.

Figure 2

Overall energy profiles of CO2 electroreduction into products CH4 and CH3OH on Cu(111) at low overpotentials: (a) CH2O pathway; (b) CHOH pathway.

By scrutinizing the overall energy profiles, we find that the barrier height of overall energy profiles can be determined by initial CO2 electroreduction into the COOH pathway. Moreover, intermediate COOH further electroreduction into CO and the inverse process of COOH formation and may be competitive due to almost equal and surmountable barriers of ca. 0.20 eV at room temperature, which will inhibit key intermediate CO formation. Simultaneously, it is also noted that the inverse process of CO electroreduction into CHO only needs an extremely low barrier of ca. 0.20 eV, suggesting that CHO may be not stable at low overpotentials, which can be facile to change back to CO. Thus, we can speculate that CO formation and further reduction may be the potential-limiting steps during CO2 electroreduction, which are regarded as the origin of an experimentally observed high overpotential on the Cu surfaces. In following sections, on the basis of the energetics obtained at low overpotentials by the currently used electrochemical model, the onset potential and potential-dependent kinetic barriers and reaction free energies for each elementary step will be considered to further determine the origin of the overpotential, the potential-limiting step, and the effect of the applied potentials on CO2 electroreduction pathways on Cu(111) using the approaches from Asthagiri et al. and Nørskov et al.

Determination of the Onset Potential

The onset potential (Uonset) is relative with our DFT-calculated kinetic barriers, namely, the lower barrier and faster reaction rate means more positive onset potential for CO2 electroreduction on Cu(111). Thus, to further ascertain the optimal CO2 electroreduction pathways and the potential-limiting step at the present used electrochemical interface, the onset potentials are determined for each elementary step during CO2 electroreduction based on eq based on the energetics obtained at low overpotentials by the present improved CO coverage-dependent electrochemical model, in which the value of equilibrium potential (U0) can be calculated by eq , as can be seen in Table . A surmountable barrier of 0.40 eV is assumed to be the threshold, which will make a specific pathway open up at room temperature and give significant reduction rates. For CO2 electroreduction into key intermediate CO on Cu(111), adsorbed COOH is found to be an intermediate, which then is electroreduced into adsorbed CO and a H2O molecule. A negative onset potential of −0.69 V (vs RHE) is required for CO2 electroreduction into COOH, whereas a positive onset potential of 0.59 V (vs RHE) is required to overcome the assumed kinetic barrier for CO formation through the COOH intermediate, suggesting that CO2 electroreduction into COOH is the potential-limiting step during CO2 electroreduction into CO on Cu(111), as observed above for a higher kinetic barrier (ca. 0.87 eV). The present calculated onset potential for CO2 electroreduction into COOH is also well agreeable with that of the corresponding process on Cu(111) from Asthagiri et al. (ca. −0.64 eV vs RHE),[32] further indicating that the validity of our present used an electrochemical model. There are two ways for further electroreduction of CO, including CHO or COH formation by CO hydrogenation. A significantly more negative onset potential of −1.87 V (vs RHE) is required for COH formation than that of CHO formation (−0.79 V vs RHE). The results show CHO can be more facile to be formed on Cu(111), which is consistent with the above MEP analysis on Cu(111), namely, CO electroreduction into CHO is kinetically more favorable than electroreduction into COH.

Table 1

Calculated Equilibrium Potential (U0 vs RHE) and Onset Potential (Uonset vs RHE) for the Possible Elementary Steps for CO2 Electroreduction on Cu(111) at the Present Used Electrode/Aqueous Interface

reactions	U0, V	Uonset, V
CO2* + (H+ + e–) → COOH*	0.25	–0.69
COOH* + (H+ + e–) → CO* + H2O	0.19	0.59
CO* + (H+ + e–) → CHO*	0.11	–0.79
CO* + (H+ + e–) → COH*	0.11	–1.87
CHO* + (H+ + e–) → CH2O*	0.13	0.03
CHO* + (H+ + e–) → CHOH*	0.13	0.07
CH2O* + (H+ + e–) → CH2OH*	0.14	0.30
CH2O* + (H+ + e–) → CH3O*	0.14	0.48
CHOH* + (H+ + e–) → CH2OH*	0.21	0.25
CHOH* + (H+ + e–) → CH* + H2O	0.21	0.63
CH2OH* + (H+ + e–) → CH2* + H2O	0.14	0.80
CH2OH* + (H+ + e–) → CH3OH*	0.14	0.22
CH3O* + (H+ + e–) → CH3OH*	0.06	–0.52
CH* + (H+ + e–) → CH2*	0.01	0.15
CH2* + (H+ + e–) → CH3*	0.04	0.20
CH3* + (H+ + e–) → CH4*	0.10	–1.72

Further electroreduction of CHO into CH2O and CHOH has a slightly positive and almost identical onset potential, ca. 0.05 V (vs RHE), suggesting that CH2O and CHOH formations may be able to simultaneously occur on Cu(111). A similar conclusion is also obtained by the above MEP analysis. Thus, further electroreduction of CH2O and CHOH is considered. It is found that CH2OH formation requires positive and almost identical onset potential through CH2O and CHOH electroreduction, ca. 0.30 V (vs RHE), suggesting that CH2OH may be able to be formed by CH2O and CHOH electroreduction. Simultaneously, the more positive onset potentials of 0.48 and 0.63 V (vs RHE) are required for CH2O further electroreduction into CH3O and CHOH further electroreduction CH and H2O, respectively. Thus, these results show that CH2OH, CH3O, and CH may be able to be easily formed at this stage, which can be confirmed by the above obtained extremely low barriers for these electroreduction steps. CH3OH may be able to be formed by further electroreduction of CH2OH and CH3O, in which the required onset potentials are 0.22 and −0.52 eV (vs RHE), respectively. Thus, CH3OH seems to be able to be produced by CH2OH electroreduction because of considerably negative onset potential for CH3O electroreduction during CO2 electroreduction into hydrocarbons. A similar conclusion is also given by the above calculated barriers. However, a significantly more positive onset potential of 0.80 V (vs RHE) is required for further electroreduction of CH2OH to CH2 and H2O, thereby suggesting that CH2OH electroreduction into CH2 is favored over CH3OH on Cu(111) at the electrochemical interface. An identical conclusion is also obtained by the above MEP analysis. Simultaneously, CH2 is also able to be formed by further electroreduction of CH with a positive onset potential of 0.15 V (vs RHE) and a surmountable barrier at room temperature (ca. 0.33 V). Finally, CH4 can be produced by serial electroreduction of CH2 with a strongly exothermic process.

The investigations on onset potentials again explain why experimentally only CH4 can be observed on the Cu electrodes at electrochemical interface. Combining MEP analysis with the calculated onset potentials, it can be speculated that CO formation and then further electroreduction into CHO may be the potential-limiting step during CO2 electroreduction into hydrocarbons because of significantly more negative onset potentials than others, which again confirms that this step is the origin of experimentally observed high overpotential on the Cu electrodes. Accordingly, the optimal CO2 electroreduction mechanism on Cu(111) can be presented at the present used electrode/aqueous interface by comprehensive consideration of the MEP and onset potentials, which is in complete agreement with the optimal electroreduction pathways of CO2 obtained at low overpotentials in our most recent study,[36] as shown in Figure . In spite of being CH2O or CHOH pathways, the present results all show that CO2 electroreduction into CH4 on Cu(111) mainly proceeds by the common intermediates CHO and CH2 species, which are found to be well consistent with the available experimental data. For instance, the most recent experimental study conducted by Koper et al. indicated that CHO is the possibly crucial intermediate toward CH2 formation on Cu(111) using online electrochemical mass spectrometry;[8,13] the study from Dewulf et al. also indicated that adsorbed CHO and CH2 species are key intermediates for CO2 electroreduction into CH4 production on the Cu electrodes using the in situ X-ray photoelectron spectroscopy and auger electron spectroscopy.[16] Thus, the rationality of the present employed electrochemical model can be further validated.

Figure 3

Optimal CO2 electroreduction pathways on Cu(111) obtained by comprehensive consideration of the MEP and onset potentials: (a) CH2O pathway; (b) CHOH pathway.[36]

Potential-Dependent Activation Barriers and Reaction Free Energies

To further determine the effect of applied potentials on CO2 electroreduction pathways on Cu(111), the potential-dependent activation barriers and reaction free energies are calculated by eqs and 7, respectively, in which Eact0(U0) is the activation barrier of each elementary step at equilibrium potential (U0) and can be directly determined by DFT calculations. Three representative conditions are considered, namely, the present calculated thermodynamic equilibrium potential of ca. 0.27 V (vs RHE) when θCO is close to zero, low overpotential (0.14 and 0 V vs RHE), and high overpotential (−0.8 and −0.9 V vs RHE), in which the applied potential of 0.14 V (vs RHE) corresponds to our above-simulated electrochemical interface. The comparison of energetics at 0.14 V (vs RHE) between the present theoretical model and previous methodologies from Asthagiri et al. and Nørskov et al. may be able to confirm the validity of our used model. The calculated potential-dependent activation barriers and reaction free energies are listed in Tables and 3, respectively. It is found that the activation barriers and reaction free energies linearly change as the applied potentials. The observed activation barriers and reaction free energies of various elementary steps at equilibrium potential of ca. 0.27 V (vs RHE) are more positive than those at more negative applied potentials. Especially at high overpotentials, the kinetic barriers are significantly decreased and the reactions are all almost strongly exothermic. Furthermore, the barriers are surmountable at room temperature except the CH4 formation step when the applied potential reaches −0.8 V (vs RHE) based on the kinetic calculations. Although a high barrier of ca. 0.80 eV is required for CH4 formation even if the potential is −0.9 V (vs RHE), the strongly exothermic processes for CH2 serial hydrogenation (ca. 4.00 eV) at high overpotential still make its formation occur easily. Simultaneously, we also note that the kinetic barriers of some elementary steps are close to 0 eV at relatively lower overpotentials during CO2 electroreduction into CH4 on Cu(111). For example, the barrier of 0 eV is reached for COOH hydrogenative dissociation into CO at −0.21 V, CHO hydrogenation into CH2O and CHOH at ca. −0.75 V (vs RHE), CH2O hydrogenation into CH2OH and its hydrogenative dissociation into CH2 at 0 V (vs RHE), CH2O hydrogenation into CH3O at −0.32 V (vs RHE), CHOH hydrogenation into CH2OH, and its hydrogenation into CH3OH at ca. −0.56 V (vs RHE), CHOH hydrogenation into CH at −0.17 V, CH hydrogenation into CH2 at −0.65 V (vs RHE), and its hydrogenation into CH3 at 0 V (vs RHE), suggesting that these steps may be nonactivated processes at the applied potentials of higher than −0.8 V (vs RHE). Further electroreduction of CO into CHO and COH is scrutinized at different applied potentials. Even at high applied potential of −0.9 V (vs RHE), COH formation still has a notably higher barrier of ca. 0.90 eV than CHO formation and positive reaction free energy of ca. 0.10 eV, showing that CO electroreduction into CHO is also kinetically more favorable at high overpotentials.

Table 2

Potential-Dependent Activation Barriers (Eact(U), eV) for the Possible Elementary Reaction Steps for CO2 Electroreduction on Cu(111) in Aqueous Phasea

		Eact(U), eVb
reactions	Eact0(U0), eV	0.27 V	0.14 V	0 V	–0.8 V	–0.9 V
CO2* + (H+ + e–) → COOH*	0.87	0.88	0.82	0.75	0.35	0.30
COOH* + (H+ + e–) → CO* + H2O	0.20	0.24	0.18	0.11	–0.30	–0.35
CO* + (H+ + e–) → CHO*	0.85	0.93	0.87	0.80	0.40	0.35
CO* + (H+ + e–) → COH*	1.39	1.47	1.41	1.34	0.94	0.89
CHO* + (H+ + e–) → CH2O*	0.45	0.52	0.46	0.39	–0.02	–0.07
CHO* + (H+ + e–) → CHOH*	0.43	0.50	0.44	0.37	–0.04	–0.09
CH2O* + (H+ + e–) → CH2OH*	0.32	0.39	0.32	0.25	–0.15	–0.20
CH2O* + (H+ + e–) → CH3O*	0.23	0.30	0.23	0.16	–0.24	–0.29
CHOH* + (H+ + e–) → CH2OH*	0.38	0.41	0.35	0.28	–0.13	–0.18
CHOH* + (H+ + e–) → CH* + H2O	0.19	0.22	0.16	0.09	–0.32	–0.37
CH2OH* + (H+ + e–) → CH2* + H2O	0.07	0.14	0.07	0.00	–0.40	–0.45
CH2OH* + (H+ + e–) → CH3OH*	0.36	0.43	0.36	0.29	–0.11	–0.16
CH3O* + (H+ + e–) → CH3OH*	0.69	0.80	0.73	0.66	0.26	0.21
CH* + (H+ + e–) → CH2*	0.33	0.46	0.40	0.33	–0.08	–0.13
CH2* + (H+ + e–) → CH3*	0.32	0.44	0.37	0.30	–0.10	–0.15
CH3* + (H+ + e–) → CH4*	1.31	1.40	1.33	1.26	0.86	0.81

The asterisk (*) indicates that the species is adsorbed on Cu(111).

In kinetic barrier calculations, the entropies obtained from the literature of Nørskov and co-workers are considered for gaseous molecules,[25,26] whereas the entropies of the adsorbed species are ignored. The DFT calculated zero-point energies for all species are included in the activation barrier calculations.

Table 3

Potential-Dependent Reaction Free Energies (ΔGreac(U), eV) for the Possible Elementary Reaction Steps for CO2 Electroreduction on Cu(111) in the Aqueous Phasea

		ΔGreac(U), eVb
reactions	ΔGreac(U0), eV	0.27 V	0.14 V	0 V	–0.8 V	–0.9 V
CO2* + (H+ + e–) → COOH*	0.63	0.65	0.52	0.38	–0.42	–0.52
COOH* + (H+ + e–) → CO* + H2O	–1.00	–0.75	–0.88	–1.02	–1.82	–1.92
CO* + (H+ + e–) → CHO*	0.65	0.81	0.68	0.54	–0.26	–0.36
CO* + (H+ + e–) → COH*	1.09	1.26	1.13	0.99	0.19	0.09
CHO* + (H+ + e–) → CH2O*	–0.27	–0.14	–0.27	–0.41	–1.21	–1.31
CHO* + (H+ + e–) → CHOH*	0.07	0.20	0.07	–0.07	–0.87	–0.97
CH2O* + (H+ + e–) → CH2OH*	–0.22	–0.09	–0.22	–0.36	–1.16	–1.26
CH2O* + (H+ + e–) → CH3O*	–0.98	–0.85	–0.98	–1.12	–1.92	–2.02
CHOH* + (H+ + e–) → CH2OH*	–0.55	–0.49	–0.62	–0.76	–1.56	–1.66
CHOH* + (H+ + e–) → CH* + H2O	0.09	0.16	0.03	–0.11	–0.91	–1.01
CH2OH* + (H+ + e–) → CH2* + H2O	–0.34	–0.21	–0.34	–0.48	–1.28	–1.38
CH2OH* + (H+ + e–) → CH3OH*	–0.94	–0.81	–0.94	–1.08	–1.88	–1.98
CH3O* + (H+ + e–) → CH3OH*	–0.33	–0.12	–0.25	–0.39	–1.19	–1.29
CH* + (H+ + e–) → CH2*	–0.80	–0.53	–0.66	–0.80	–1.60	–1.70
CH2* + (H+ + e–) → CH3*	–1.04	–0.80	–0.93	–1.07	–1.87	–1.97
CH3* + (H+ + e–) → CH4*	–0.88	–0.71	–0.84	–0.98	–1.78	–1.88

The asterisk (*) indicates that the species is adsorbed on Cu(111).

In reaction free energy calculations, the entropies obtained from the literature of Nørskov and co-workers are considered for gaseous molecules,[25,26] whereas the entropies of the adsorbed species are ignored. The DFT calculated zero-point energies for all species are included in the reaction free energy calculations.

As noted above, CO formation and then further electroreduction into CHO may be the potential-limiting step during CO2 electroreduction into CH4 on Cu(111) at a low overpotential (ca. 0.14 V vs RHE) with high kinetic barrier of ca. 0.86 eV. On the basis of the potential-dependent barrier calculations, the formation of CO and its further electroreduction into CHO may be able to occur more and more facilely as increasing overpotential, as shown in Figure . The kinetic barriers of CO2 electroreduction into COOH and CO further electroreduction will be reduced to ca. 0.40 eV when the applied potential reaches −0.8 V (vs RHE), which can be overcome at room temperature. The corresponding barriers will be further lowered to 0.30 and 0.35 eV at the applied potential of −0.9 V (vs RHE), respectively. Interestingly, we also note that the barrier linearly increases with increasing overpotentials for the reverse processes of these both potential-limiting steps (see Figure ), which can be obtained by Eact(U) – ΔGreac(U). The equal barrier for forward and reverse processes of CO2 electroreduction into COOH and CO electroreduction into CHO may be able to be reached at the applied potential of ca. −0.38 and −0.54 V (vs RHE), respectively. As can be seen in Figure , at thermodynamic equilibrium potential and low overpotentials, the considerably high kinetic barriers for CO2 electroreduction into COOH and CO electroreduction into CHO is required, whereas their reverse process is almost nonactivated because of extremely low barriers (ca. 0.20 eV). The results suggest that COOH and CHO intermediates may be unstable at the present calculated thermodynamic equilibrium potential and low overpotentials, in which COOH can easily change back to CO2 through O–H bond cleavage and is competitive with further electroreduction of COOH, and CHO can easily change back to CO through C–H bond breaking and is competitive with further electroreduction of CHO, thus leading to not easy formations of some key intermediates and preventing CO2 electroreduction on the Cu electrodes. However, the barriers are lowered to surmountable values at room temperature at high overpotentials for CO2 electroreduction into COOH and CO electroreduction into CHO, whereas they are notably increased for the corresponding reverse processes, which can facilitate CO and CHO formations and further electroreduction. Thus, the present analyses of potential-limiting steps may be able to explain partially why the high overpotential (ca. 1 V vs RHE) is required for CO2 electroreduction into CH4 in at experiments on the Cu electrodes.[4,5]

Figure 4

Relationships between the applied electrode potentials and activation barriers of forward and reverse processes for (a) CO2 electroreduction into COOH and (b) CO electroreduction into CHO on Cu(111).

Simultaneously, we observe that the calculated potential-dependent barriers and reaction free energies of each elementary step at low overpotentials based on the methodologies from Asthagiri et al. and Nørskov et al. are almost identical with the corresponding values (Eact0(U0) and ΔGreac(U0)) obtained by our present simulated conditions of low overpotential, as can be seen in Tables and 3, further confirming the rationality of our used electrochemical interface. For comparison, the overall energy profiles of CO2 electroreduction into CH4 and CH3OH on Cu(111) at high overpotential are also given, as shown in Figure . We find that the optimal CO2 electroreduction mechanisms are not altered at high overpotentials compared with that of low overpotentials, namely, CH2O and CHOH pathways through common intermediates CHO and CH2 are still able to occur simultaneously, suggesting that CO2 electroreduction pathways on the Cu electrodes may be not affected by the applied potentials. It is also observed that CO2 electroreduction into CH4 is a more strongly exothermic process than into CH3OH at high overpotentials (−8.80 vs −6.22 eV), again confirming that CH4 formation is thermodynamically more favorable than CH3OH and explaining why only CH4 is observed experimentally on the Cu surface at the electrochemical interface. It is noteworthy that the kinetic barrier of ca. 0.35 eV for the overall energy pathway is observed at high overpotentials (see Figure ), which is significantly lower than that of low overpotentials (ca. 0.87 eV). Moreover, CO2 electroreduction into CH4 is also significantly stronger exothermically at high overpotentials than that of low overpotentials (−8.80 vs −2.47 eV). Thus, CO2 electroreduction into CH4 is thermodynamically and kinetically more favorable at high overpotentials, on the whole explaining why the high overpotentials are required during CO2 electroreduction at experiments on the Cu electrodes.

Figure 5

Overall energy profiles of CO2 electroreduction into CH4 and CH3OH on Cu(111) at the applied potential of −0.8 V (vs RHE): (a) CH2O pathway; (b) CHOH pathway (the calculated obtained negative barriers are considered as 0 eV in the overall energy profiles).

Conclusions

In this paper, potential-dependent CO2 electroreduction pathways on Cu(111) are systematically studied with the aim of applying an improved electrode/aqueous interface model. The results indicate that the present defined CH2O and CHOH pathways may be able to parallelly take place at low overpotentials, which proceed via common intermediate CHO and CH2 species. Notably, the applied potentials will not alter the optimal CO2 electroreduction mechanisms. However, the presence of high overpotentials makes CO2 electroreduction more favorable because of the significantly reduced kinetic barriers of overall energy pathway and various elementary steps, thus explaining why the high overpotential at experiments is required during CO2 electroreduction on the Cu electrodes. On the basis of the analyses of potential-dependent energetics, the results suggest that COOH and CHO intermediates may be unstable at low overpotentials, in which COOH can easily change back to CO2 and CHO can easily change back to CO, thus preventing CO2 electroreduction. However, the high overpotentials can significantly reduce the barriers for CO2 electroreduction into COOH and CO electroreduction into CHO and notably increase the barriers for the corresponding inverse processes, facilitating CO and CHO formations and further electroreduction. Thus, we can speculate that CO formation and then further electroreduction into CHO may be the potential-limiting steps of CO2 electroreduction, which are regarded as the origin of experimentally observed high overpotentials. The present comprehensive understanding on CO2 electroreduction pathways can provide theoretical guidelines for efficiently designing Cu-based alloy electrocatalysts operated under the conditions of low overpotentials.

Computational Details

Determination of CO Coverage-Dependent Equilibrium Potentials

Surface and solvation model and computational parameters are demonstrated in the Supporting Information in detail. The method for determining CO coverage-dependent equilibrium potentials had been reported in our most recent study,[36] which was first employed to model electroreduction pathways of CO2 at low overpotentials. For clarity, the corresponding method is again elaborated in this paper. The CO intermediate is formed during CO2 electroreduction by the following reaction

Under different CO coverage conditions based on the approach proposed by Nørskov et al. and Chen et al. for hydrogen evolution and oxygen reduction reactions, the Gibbs free energy, ΔG(θ), can be calculated according to eq ,[24,42−45] where ΔE(θ), ΔS(θ), ΔZPE, and kBT ln(θ/1 – θ) represent the contributions of CO differential adsorption energy, entropy, zero-point energy, and configurational entropy to ΔG(θ), respectively. Here, coverage θ = n/N, in which n is the number of adsorbed CO molecules on the surface and N is the total number of the surface Cu atoms. Thus, the CO coverage-dependent equilibrium potential U (vs RHE) can be determined when the Gibbs free energy, ΔG(θ) is zero.

The differential adsorption energy of CO, ΔE(θ), can be expressed by eq , in which is the total energy of the surface with different CO coverages.

Considering that the entropies of the adsorbed molecules are small when compared with the entropies of gaseous and that the zero-point energy of the adsorbed CO molecules on Cu(111) is based little on our present calculations, the contributions from the changes of entropy and ZPE together to ΔG(θ) has been estimated to be −0.42 eV for standard temperature (298 K) according to the available data from Nørskov et al.[25] Thus, eq can be converted to the following form of eq .

The values of , EH, ECO, and EH are directly available through DFT calculations. A series of values of can be obtained by changing the coverage of adsorbed CO on Cu(111). The values of ΔE(θ) at various coverages can be obtained according to eq by differentiating the plots of against θ.

Potential-Dependent Activation Barriers

To consider the effect of applied potential (U) on activation barriers, Eact(U), potential-dependent barriers are obtained by using the previously reported approach from Asthagiri et al.,[29,32] as shown in eq , where Eact0(U0) is the activation barrier of each elementary step at equilibrium potential (U0) and can be directly determined by DFT calculations when the adsorbed H species and proton–electron pair has identical chemical potential, β is the reaction symmetry factors. In this paper, the average value of β (ca. 0.5) is used on the Cu surface based on the available data from Asthagiri et al. and deemed to be a reasonable assumption by Head-Gordon et al.[46]

The value of equilibrium potential (U0) is calculated by the CHE model proposed by Nørskov et al., in which U0 of any elementary reaction step 6 can be determined by eq when reaction free energy is zero. The potential-dependent reaction free energies are also able to be calculated by eq .

Simultaneously, the onset potential of each elementary step is calculated using the approach proposed by Asthagiri et al.,[32] in which a specific pathway may be able to easily occur at room temperature. In the present paper, the threshold barrier that can be overcome at room temperature with significant reaction rate is assumed to be 0.4 eV based on early studies.[47,48] Thus, the onset potential can be calculated by solving eq for each elementary step including proton-electron transfer process.