Potential-Dependent Competitive Electroreduction of CO2 into CO and Formate on Cu(111) from an Improved H Coverage-Dependent Electrochemical Model with Explicit Solvent Effect.

An improved density functional theory-based H coverage-dependent electrochemical model with explicit solvent effect is proposed for Cu(111), which is used to identify potential-dependent initial competitive CO2 electroreduction pathways considering HER. We find that a chemisorbed CO2 molecule at the present electrode/aqueous interface can be spontaneously formed and the overpotentials can affect its coordination pattern. The Eley-Rideal mechanism may be more favorable during the initial CO2 electroreduction into CO, whereas chemisorbed CO2 reacting with adsorbed H into HCOO- via the Langmuir-Hinshelwood mechanism is more facile to occur. The analyses of energetics suggest that the low overpotentials have a negligible influence on CO and HCOO- formation, and HCOO- species with monodentate and bidentate configurations may also parallelly form with the surmountable barriers at room temperature. However, the high potentials have an interruptive effect on initial CO2 electroreduction because of the significantly increased barriers, indicating that the chemisorbed CO2 can be stabilized by imposing more negative potentials and thus going against initial CO2 electroreduction. By analyzing the competing HER with initial CO2 electroreduction into CO, we find that HER is competitive with initial CO formation because of the required lower overpotentials. Simultaneously, the present study shows that the blocked Cu surface by adsorbed H and CO can explain why the initial CO formation pathway is unfavorable at the high overpotentials. Our present conclusions can also confirm the previous experimental report on initial formation of CO and HCOO-.

Introduction

Electrochemical reduction of CO2 into useful and stable chemical fuels represents a promising approach and attracts significant attention to mitigate CO2 emissions in the atmosphere by supplying renewable electricity that generated from hydropower, wind, and solar.[1−3] Among a series of metals that can electrochemically reduce CO2,[4−15] Cu electrodes have unique ability to electro-reduce CO2 into hydrocarbons, such as CH4 and C2H4.[4−6] Moreover, early experimental and theoretical investigations had reached consensus on the existence of a crucial intermediate CO adsorbed on the Cu surface during CO2 electroreduction.[5,16−19] However, a high overpotential of ca. 1.0 V for a significant current density and low selectivity toward a particular species, such as CO, limited its application.[4−9] Because of unique electrocatalytic ability of Cu electrodes, it is usually used to determine and validate CO2 electroreduction mechanisms to seek new Cu-based alloy electrocatalysts in a more efficient way that can lower overpotentials and improve the selectivity of CO against its competitors, formate (HCOO–) species and H2.[20−23] Thus, CO2 electroreduction mechanisms on the Cu surface have been extensively investigated by using experimental and theoretical methodologies.[24−37]

In situ IR adsorption spectroscopy technique had been previously used to characterize chemisorbates relevant to the production of CO and HCOO– species on the polycrystalline Cu electrodes with varying applied potentials.[24] However, only a few intermediates were identified because of the complexity of spectra, which is insufficient for rigorously determining CO2 electroreduction mechanisms. The product distribution of CO2 electroreduction is electrode potential-dependent.[5,25,26] The earlier studies from Hori et al. showed that CO starts to form at −0.50 V (vs RHE) and predominates at potentials less negative than −0.8 V (vs RHE), where it competes with the formation of HCOO– species.[19,27] The adsorbed CO intermediate on the Cu electrode during CO2 electroreduction will also interfere with the cathodic hydrogen evolution reaction (HER). At applied potentials more negative than −0.9 V (vs RHE), the adsorbed CO is electrochemically reduced to hydrocarbons, such as CH4 and C2H4, with a decrease in CO Faradaic efficiency. Recent developments of nanostructured Cu electrocatalysts have enabled CO2 electroreduction at low overpotentials. For example, recent reports from Wang et al. have elucidated selective CO2 electroreduction at potentials more positive than −0.50 V (vs RHE).[28,29] CO and HCOO– species are typically found to be the two major products at such low overpotentials with the Faradaic efficiencies generally increasing as the applied potentials become more negative, which is considered to be the key intermediate toward the formation of more valuable hydrocarbons at higher overpotentials.[5,25,28−31] Indeed, CO reduction results in very similar product distributions and onset potentials to those of CO2 based on experimentally measured reaction kinetics.[7,19,27] Thus, it is important to understand the initial electroreduction mechanisms of CO2 into CO and HCOO– species at applied electrode potentials. However, the mechanism of CO2 electroreduction into CO and HCOO– species on the Cu electrodes remains elusive at the different applied potentials because of lack of in situ characterization techniques. The efficient theoretical methodologies can be used to advance the mechanistic understanding of initial CO2 electroreduction. However, theoretically studying the electrocatalytic mechanisms occurring at the electrode/aqueous interfaces is a challenging task because the considerations of solvation and electrode potential are required. Previously, the vacuum/solid interfacial model or the fixed H2O molecules were usually employed for the investigations of CO2 reduction mechanism on the Cu electrodes in many theoretical calculations to reduce computational cost, leading to be not very convincing results.[32,33] To more realistically simulate the electrode/aqueous interfaces, some theoretical methods have been proposed.[34−41] For example, ab initio molecular dynamics simulations (AIMD) may be able to describe the dynamic nature of H2O molecules and provide atomic insights into the mechanisms of CO2 electroreduction into CO on Cu(100).[34−37] Using the AIMD simulations model and an aqueous interfacial model with explicit H2O molecules, electrical field effects from solvated proton–electron transfer have been showed to play an important role in C–O bond breaking and CO formation mechanism.[35,42] The electroreduction pathways to CO and HCOO– species on Cu(100) were also examined by Goddard et al. using AIMD simulation with an explicit description of H2O molecules.[43] They concluded that chemisorbed CO2δ− can be an intermediate in forming CO with the conversion from CO2δ− to physisorbate CO2 as the rate-determining step, while HCOO– species formation proceeded via a different route with surface-adsorbed H reacting directly with the physisorbate CO2. However, the most recent AIMD simulations showed that the physisorbate CO2 can rapidly transform into chemisorbed CO2δ− during CO formation at high overpotentials,[35] suggesting that the stabilities of reactive species may be relative with the applied potentials. In contrast to the vacuum/solid interfacial model or the fixed H2O molecules, AIMD simulations at the aqueous interfaces require sampling of the fluctuating H2O molecules, resulting in extremely time-consuming calculations. Focusing on the formation of CO and HCOO– reduction products, Liu et al. studied the effect of applied potentials on initial CO2 reduction and determined the origin of the overpotentials on Cu(211) using density functional theory (DFT) with an implicit solvent model, showing that the large overpotentials were required for CO and HCOO– formations.[44] However, the implicit solvent model may not be able to determine accurately the energetics of various reactions occurred in aqueous phase and describe the interactions between adsorbates and solvents. In spite of numerous experimental and theoretical studies, the mechanistic inconsistencies still exist for initial CO2 electroreduction because of the difference of electrode/aqueous interface models. Further theoretical investigations need to be performed to explore the effect of applied electrode potentials on the initial competitive CO2 electroreduction pathways.

In this paper, by employing the validated explicit solvent model with two relaxed H2O bilayer structures on Cu(111) which allows us to better determine the kinetic barriers for various electrochemical reduction pathways and simulate the interactions between adsorbates and solvents,[45−48] an improved DFT-based H coverage-dependent electrochemical model is proposed. Based on the present model, we can identify potential-dependent competitive electroreduction mechanisms of CO2 into CO and HCOO– species with the consideration of competing HER on Cu(111), which is essential for optimizing the design strategy of Cu-based alloy electrocatalysts and product selectivity.

Results and Discussion

H Coverage-Dependent Equilibrium Potentials

Various possible surface adsorption sites of H atoms and coverage dependence are considered in our present study. It is observed that H atoms prefer to adsorb at threefold face-centered cubic (fcc) hollow sites on Cu(111) so that they can stay away from each other to minimize the repulsive reactions in our present research scope on θH. The coverages of H atoms above 1 monolayer (ML) are not further analyzed because of the observed formations of H2 molecules by adjacent adsorbed H atoms. As can be seen in Figure , a reasonable polynomial relationship on Cu(111) is exhibited between the differential adsorption energy of H atoms, ΔE(θ) and θH, suggesting that the Langmuir adsorption isotherms may be nearly followed for the adsorption of H atoms. We are able to calculate ΔE(θ) at any H coverage (θH) via polynomial fitting of the present ΔE(θ)−θH data. Therefore, the equilibrium potentials (U) can be calculated according to the above eq , and then, the polynomial relationships between θH and U can be obtained on Cu(111). We find that the more negative equilibrium potentials can be obtained with the more H coverage, in which the potentials only change slightly when θH is below 1/3 ML, whereas they are significantly changed into more negative values when θH is above 1/3 ML, implying that the adsorbed H atoms produced by HER may be able to occupy surface-active sites of the Cu electrodes at high overpotentials and impede the further occurrence of CO2 electroreduction. The equilibrium potential is calculated as ca. 0.16 V (vs RHE) when θH is equal to zero based on the abovementioned polynomial relationships, which is well agreeable with the thermodynamically required value of ca. 0.17 V (vs RHE) for CO2 electroreduction, validating the reasonability of our present proposed electrochemical model to a certain extent. Our present study will focus on potential-dependent competitive electroreduction pathways of CO2 into CO and HCOO– on Cu(111) with the aim of applying the present proposed H coverage-dependent electrochemical model.

Figure 1

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

Initial Electroreduction of CO2 into CO and HCOO–

Early experimental investigations showed that CO and HCOO– species are initially formed during CO2 electroreduction on the Cu electrodes prior to the production of hydrocarbons.[5,8,9,18,19] Further CO electroreduction can result in a similar product distribution as observed in CO2 electroreduction, suggesting that CO is a key intermediate on the Cu electrodes. Experimental tests beginning with HCOO– do not lead to any measurable products. Thus, initial electroreduction of CO2 into CO and HCOO– may be competitive at Cu. Based on the relationships between H coverage (θH) and the equilibrium potentials (U), it is found that the calculated equilibrium potential is ca. 0.16 V (vs RHE) when θH is zero, which may correspond to thermodynamically required equilibrium potential for CO2 electroreduction into hydrocarbons. The potentials become more and more negative with increasing θH. The calculated equilibrium potentials locate the range from ca. −0.01 to ca. −0.06 V (vs RHE) when θH is below ca. 1/3 ML, corresponding to the conditions of low overpotentials compared with the equilibrium potentials when θH is zero. The calculated potentials are ca. −0.50 V (vs RHE) when θH locates the range from ca. 1/3 to ca. 2/3 ML, corresponding to the conditions of medium overpotentials. When θH is above ca. 2/3 ML, the calculated potentials locate above ca. −0.90 V (vs RHE), which corresponds to the conditions of high overpotentials. Thus, the potential-dependent initial electroreduction pathways of CO2 into CO and HCOO– on Cu(111) can be obtained by using the present proposed H coverage-dependent electrochemical model.

CO2 adsorption on the Cu electrodes is of key importance during initial CO2 electroreduction in the determination of product selectivity. Early experimental and theoretical studies showed that CO2 may physically or chemically adsorb with a linear or bent structure on the Cu surfaces.[49−52] Therefore, several configurations for the chemisorbed bent CO2 molecule have been postulated, involving coordination of a single C or O atom and mixed C–O coordination to the surfaces.[50] Thus, the adsorption state and the coordinate pattern of the CO2 molecule on the Cu surfaces are not unambiguously characterized and remain to be unveiled. Using the H coverage-dependent electrochemical model, we find that a physisorbed CO2 molecule placed at the electrode/aqueous interface would spontaneously adsorb chemically on Cu(111) with a bent and mixed C–O coordination configuration at the low and medium overpotentials. However, the coordination pattern of a single C atom is observed at the high overpotentials, as can be seen in Figure . The results suggest that the overpotentials will have an effect on the adsorption geometry of the surface CO2 molecule, which can be characterized by the changes of bond length of C–O and bond angle of O–C–O. At the low and medium overpotentials, the bond lengths of C–O are ca. 1.256 and 1.296 Å, and the bond angle of O–C–O is ca. 123° in the chemisorbed CO2 molecule, which are comparable with those obtained in calculated equilibrium potential (θH = 0). However, the bond length and bond angle are significantly changed at the high overpotentials, in which the bond lengths of C–O are shortened to ca. 1.249 and 1.270 Å, whereas the bond angle of O–C–O is extended to ca. 126°. Simultaneously, the analyses of geometry configuration for the chemisorbed CO2 molecule also suggest that the further electroreduction ability of CO2 will be reduced on the Cu electrodes at the high overpotentials.

Figure 2

Adsorption configuration of the chemisorbed bent CO2 molecule at the different H coverages (θH): (a) 1/9 ML, (b) 2/9 ML, (c) 1/3 ML, (d) 4/9 ML, (e) 5/9 ML, and (f) 2/3 ML.

Previous studies have established that the adsorbed COOH species is a crucial intermediate during CO2 electroreduction into CO on various Cu facets;[7,32,33,35] therefore, the initial electroreduction pathways of CO2 into CO through intermediate COOH are studied. Starting from chemisorbed CO2 on the Cu(111) surface, intermediate COOH may be able to be formed through two possible reaction mechanisms, namely, a proton–electron transfer to the C atom of chemisorbed CO2 via the Eley–Rideal reaction and a surface-adsorbed H atom reacting with chemisorbed CO2 via the Langmuir–Hinshelwood mechanism. It is found that the formation of COOH species requires a barrier of ca. 1.0 eV via Langmuir–Hinshelwood mechanism on Cu(111) at the calculated equilibrium potential of ca. 0.16 V (vs RHE) and its subsequent hydrogenative dissociation into CO and H2O needs to overcome the barrier of ca. 0.70 eV. As can be seen in Figure S1, the barrier for the overall energy profile is ca. 1.37 eV for CO2 electroreduction into CO. However, the barriers for COOH and CO formations are decreased to ca. 0.30 eV when proton–electron transfers to the chemisorbed CO2 through the Eley–Rideal reaction at the calculated equilibrium potential (θH = 0). Furthermore, the barrier for the overall energy profile is also reduced to only ca. 0.30 eV (see Figure S1). Thus, we can conclude that the Eley–Rideal reaction mechanism may be more favorable during the initial CO2 electroreduction into CO on Cu surface. As can be seen in Figure , the activation barrier for COOH formation via the proton–electron-transfer process is ca. 0.35 eV at the low overpotentials, and it slightly rises to ca. 0.50 eV at the medium overpotentials (CO coverage of 5/9 ML). Surprisingly, the barrier is notably increased to ca. 0.96 eV at the high overpotential of ca. −1.0 V (CO coverage of 2/3 ML), indicating that significantly more negative potentials have an interruptive effect on the initial reduction into COOH species of chemisorbed CO2 on Cu(111). The initial electroreduction product CO can be formed via the proton–electron transfer of intermediate COOH along with H2O formation. At the thermodynamically required equilibrium potential of ca. 0.16 V (vs RHE), the calculated barrier is ca. 0.28 eV for further COOH electroreduction into CO on Cu(111), whereas the almost identical and negligible barriers of ca. 0.10 eV are obtained with a strong exothermic process at the low, medium, and high overpotentials (see Figure ), suggesting that the overpotentials may have negligible influence on CO formation through COOH species at Cu. By scrutinizing the overall energy profile shown in Figure , the barrier height is determined by COOH formation pathways in the range of the different overpotentials. Thus, COOH formation pathways may be rate-limiting steps during the initial CO2 electroreduction into CO. Notably, it is observed that the reverse reaction barrier of COOH via O–H bond cleavage into chemisorbed CO2 is extremely low (ca. 0.20 eV) and comparable with its further electroreduction into CO and H2O molecule at the high overpotential of ca. 1.0 V, which indicates that the chemisorbed CO2 may be able to be stabilized by imposing more negative potentials on the Cu electrodes, in contrast to the initial CO2 electroreduction.

Figure 3

Overall energy diagram of CO2 electroreduction into CO through intermediate COOH at different H coverages via the Eley–Rideal mechanism.

HCOO– formation had been thought experimentally to be competitive with CO formation on the Cu electrodes.[5,25] Even though the hydrogenative mechanism of physisorbed CO2 to produce HCOO– species had been theoretically proposed for Cu surfaces,[44,47,53] the initial electroreduction mechanisms of chemisorbed CO2 are focused at the present study because of spontaneous formation of chemisorbed state of CO2 molecule using the present model. Two possible adsorption configurations of HCOO– species are considered, namely, the monodentate adsorbate Mo–HCOO- with one O atom coordinated to the Cu surface and the bidentate adsorbate Bi–HCOO– with the coordination of two O atoms. One possible mechanism of HCOO– formation is by proton–electron transfer to the C atom of chemisorbed CO2 via the Eley–Rideal reaction. However, we find that the formations of Bi–HCOO– need to overcome a considerably higher barrier of ca. 1.0 eV via the proton–electron transfer than that via the adsorbed H atom on Cu(111) (ca. 0.18 eV) at the calculated equilibrium potential, as shown in Figure S2. Thus, the Eley–Rideal reaction pathway may be not responsible. An alternative reaction pathway for the formation of HCOO– species can be proposed, namely, chemisorbed CO2 reacts with a surface-adsorbed H atom through the Langmuir–Hinshelwood mechanism. As can be seen in Figure , the barriers of only ca. 0.15 eV are required for the formation of Mo–HCOO– species at the low and medium overpotentials, and the almost identical barriers are obtained for Bi–HCOO– formation at the low overpotentials, which are comparable with the barrier value of ca. 0.18 eV at the calculated equilibrium potential, suggesting that the low overpotentials will have a negligible influence on HCOO– formation on Cu(111), and two adsorption configurations of HCOO– species are possibly formed at the low overpotentials. However, the barriers are significantly increased into ca. 0.60 and 0.80 eV for formations of Mo–HCOO– and Bi–HCOO– species at a high overpotential of ca. 1.0 V, respectively, indicating that high overpotentials have notable effect on HCOO– formation. Furthermore, Bi–HCOO– is more difficult to be formed than Mo–HCOO– species at the high overpotential. Simultaneously, we also note that the barrier is increased into ca. 0.40 eV at the medium overpotentials for Bi–HCOO– formation, which is significantly larger than that of Mo–HCOO– formation (see Figure ), suggesting that Bi–HCOO– formation may be difficult on Cu(111) at the medium and high overpotentials.

Figure 4

Energy profile of CO2 electroreduction into (a) monodentate HCOO– and (b) bidentate HCOO– species at different H coverages via the Langmuir–Hinshelwood mechanism.

To further ascertain the formation mechanisms of HCOO– species, the minimum energy path (MEP) analysis is carried out for the conversion of Mo–HCOO– into Bi–HCOO– on Cu(111) within the range of explored overpotentials, as shown in Figure . The results show that the barriers of only ca. 0.25 eV are required for conversion process in the range of low overpotentials, which are surmountable at room temperature, further indicating that HCOO– species with monodentate and bidentate configurations may be able to be parallelly formed at the low overpotentials. However, the barriers locate the range from ca. 0.50 to 0.90 eV at the medium and high overpotentials. Moreover, the reverse processes for the conversion from Mo–HCOO– to Bi–HCOO- are almost nonactivated, as can be seen in Figure , indicating that the formation of Bi–HCOO– species is unfavorable and unstable on Cu(111) in the range of medium and high overpotentials. By comprehensively considering the initial electroreduction mechanisms of CO2 into COOH, CO, and HCOO– species on Cu(111), we can conclude that CO through COOH intermediate and HCOO– species with monodentate and bidentate configuration is facile to be formed within the range of low overpotentials because of the surmountable barriers at room temperature, confirming the previous experimental report on initial formation of CO and HCOO– from Hori et al.[17] In fact, the most recent experimental study conducted by Shao et al. also implied the coexistence of COOH and HCOO– species after initial CO2 electroreduction on Cu surfaces at the low overpotentials using attenuated total reflection surface-enhanced infrared absorption spectroscopy techniques.[54] The formations of these two species are unfavorable because of significant increased barriers at the high overpotentials, which can be confirmed by the above analyses of geometry configuration for chemisorbed CO2 molecule and explain why experimentally observed productions of hydrocarbons dominate over CO and HCOO– species on Cu surface within the range of high overpotentials.

Figure 5

Energy profile of conversion of monodentate HCOO– into bidentate HCOO– species at different H coverages.

Competing HER with Initial CO2 Electroreduction

Occurrence of the rate-determining step after CO formation has experimentally established that adsorbed CO is a key reaction intermediate in the formation pathways of hydrocarbons and can block the Cu electrode surfaces.[7,8,18,19] Thus, a crucial issue during CO2 electroreduction into hydrocarbons is competing HER with initial CO2 electroreduction into CO, in which a hydrated proton does not attach to the adsorbed molecules but rather adsorbs on the Cu electrodes via Volmer reaction. Based on our most recent proposed CO coverage-dependent electrochemical model, the linear relationships between CO coverage (θCO) and the overpotentials can be obtained on Cu(111) compared with the calculated equilibrium potentials when θCO = 0.[47] The polynomial relationships between θH and the overpotentials can be obtained on Cu(111) based on the present proposed electrode/aqueous interface model, as can be seen in Figure .

Figure 6

Obtained relationships between overpotentials and θH and θCO based on H and CO coverage-dependent electrochemical models.

It is observed that at the overpotentials of below ca. 0.15 V, θH and θCO is low and almost identical and locates the range of 0 to 1/9 ML, implying that both HER and CO formation are not facile to occur at such low overpotentials. The required overpotentials gradually increase as the θH and θCO increase on Cu(111). The overpotentials are ca. 0.20 V when θH locates the range of 1/9 ML to 1/3 ML, whereas it is in the range of ca. 0.40–0.60 V for the corresponding θCO. The overpotentials locate the range of ca. 0.40–0.60 V when θH is in the range of 1/3 ML to 2/3 ML, whereas the significantly higher overpotentials from ca. 0.80–1.0 V are obtained for the corresponding θCO. When θH is 2/3 ML, the overpotential of ca. 1.0 V is required, which is also lower than that for the corresponding θCO. The results show that HER requires lower overpotentials than CO formation in the range of 1/9 to 2/3 ML of θH and θCO, suggesting that HER may be able to be more facile to occur than CO formation at the overpotentials of below ca. 1.0 V on the Cu surfaces and further confirming the experimentally observed results that the current is predominant for HER during initial CO2 electroreduction.[5,17] Above 2/3 ML of θH and θCO, the overpotentials tend to achieve very high and almost equal values (see Figure ), showing that the Cu electrode surface is blocked by adsorbed H and CO and thus suggesting that the initial CO formation pathway is unfavorable at such high overpotentials. As can be seen in Figure , θH is 4/9 ML, whereas θCO is 2/9 ML at the low overpotential of ca. 0.40 V. The corresponding value for θH and θCO is increased into 5/9 and 1/3 ML at a medium overpotential of ca. 0.60 V, suggesting that HER is competitive with initial CO2 electroreduction into CO, and the adsorbed H can inhibit further CO reduction in this voltage range, thus explaining the previous experimental and theoretical observations.[5,55] However, because of relatively lower θCO at the low overpotentials, we can speculate that CO is facile to be formed during CO2 electroreduction, as above-elucidated low barriers. At the high overpotentials of above ca. 1.0 V, the high θH and θCO of above 2/3 ML can occupy the surface-active site of Cu electrodes and lead to difficult occurrences of initial CO2 electroreduction into CO, as above-observed high barriers in MEP analysis. These results may be able to explain why CO formation competes with HER.

Conclusions

In this paper, we propose an improved DFT-based H coverage-dependent electrochemical model with explicit solvent effect on Cu(111). We identify potential-dependent competitive electroreduction pathways of CO2 into CO and HCOO– species with the consideration of competing HER using the present model. We find that a physisorbed CO2 molecule placed at the electrode/aqueous interface would spontaneously adsorb chemically on Cu(111) with a bent and mixed C–O coordination configuration at the low and medium overpotentials. However, the coordination pattern of a single C atom is observed at the high overpotentials. Thus, it can be speculated that the overpotentials will have an effect on the adsorption geometry of the surface CO2 molecule. By comprehensively considering the initial CO2 electroreduction mechanisms, we can conclude that the Eley–Rideal reaction via proton–electron transfer may be more favorable during initial CO2 electroreduction into CO through intermediate COOH, whereas chemisorbed CO2 reacting with a surface-adsorbed H into HCOO– via Langmuir–Hinshelwood mechanism is more facile to occur. The analyses of energetics show that initial CO2 electroreduction into CO and HCOO– species with monodentate and bidentate configurations can proceed with the surmountable barriers at room temperature in the range of low overpotentials, suggesting that the low overpotentials will have a negligible influence on CO and HCOO– formation and two adsorption configurations of HCOO– may be able to be parallelly formed. However, the high potentials have an interruptive effect on initial electroreduction of chemisorbed CO2 because of significantly increased barriers, indicating that the chemisorbed CO2 may be able to be stabilized by imposing more negative potentials and the initial CO2 electroreduction is unfavorable, which can be confirmed by analyzing geometry configuration for the chemisorbed CO2 molecule. Furthermore, HCOO– formation with bidentate configuration is unfavorable at the medium and high overpotentials. By analyzing the competing HER with initial CO2 electroreduction into CO, we find that HER may be able to be more facile to occur than CO formation at the overpotentials of below ca. 1.0 V, suggesting that HER is competitive with initial CO formation. Simultaneously, the present study shows that the blocked Cu surface by adsorbed H and CO at the high overpotentials can explain why initial CO formation pathway is unfavorable. Our present conclusions can also confirm the previous experimental report on initial formation of CO and HCOO–.

Model and Computational Details

Surface and Solvation Model

The closed-packed Cu(111) crystal planes are generally chosen as the representative surfaces for both experimental and theoretical studies because of their high selectivity to CO2 electroreduction. Considering the complexity of real CO2 electroreduction systems, the aqueous-phase environment is included in the present study, in which 12 explicit H2O molecules with two relaxed bilayer structures chosen to fill up the vacuum region were used to model the solvation effect in order to better simulate the interactions between the solvent and adsorbates and decrease the size of the simulated systems as much as possible. Simultaneously, a hydrated proton is included in solvation model to better simulate real electrode/aqueous interfaces because of HER and the proton–electron transfer process during CO2 electroreduction. In fact, the formation of an ordered H2O bilayer structure in a hexagonal arrangement with 2/3 ML saturation coverage with respect to the surface normal had been demonstrated by X-ray absorption spectroscopy, thermal desorption spectroscopy, low-energy electron diffraction, X-ray photoelectron spectroscopy, and scanning electron microscopy along with DFT calculations in previous experimental and theoretical studies on the meal surface.[48,56,57] Our solvation model is on the basis of the previous studies on structure and orientation of H2O. However, many different H2O solvation structures may also exist, which are all approximate in energy.[58] Because all energies of interest in this study are energy differences, which are not sensitive to the accurate model of H2O as long as the same model is consistently used and a reasonable model in a local minimum structure is chosen when calculating the energy differences. Considering the coverage is 2/3 of H2O ML, thus, a (3 × 3) Cu(111) slab model with nine metal atoms per layer and a theoretical equilibrium lattice constant of 3.66 Å by using four metal layers was created.

Computational Parameters

Using the generalized gradient approximation of the Perdew–Burke–Ernzerhof exchange correlation functional, calculations were performed in the framework of DFT.[59] Ultrasoft pseudopotentials were employed to describe the nuclei and core electrons and the Kohn–Sam equations were self-consistently solved using a plane-wave basis set.[60] A kinetic energy cutoff of 30 Ry and a charge density cutoff of 300 Ry were used to make the basis set finite. The Fermi surface has been treated by the smearing technique of Methfessel–Paxton with a smearing parameter of 0.02 Ry.[61] The PWSCF codes in Quantum ESPRESSO distribution were employed to perform all calculations.[62] Brillouin zone integrations were implemented using a (3 × 3 × 1) uniformly shifted k-mesh for (3 × 3) supercell with the special point technique, which was tested to converge to a subset of the relative energies reported herein. A vacuum layer of 16 Å was placed above the top layer of the slab, which is sufficiently large to ensure that the interactions are negligible between repeated slabs in a direct normal to the surface. The Cu atoms in the bottom two layers are fixed at the theoretical bulk positions, whereas the top two layers and all adsorbates including solvent are allowed to relax to minimize the total energy of the system. Structural optimization was performed until the Cartesian force components acting on each atom were brought below 10–3 Ry/bohr and the total energy was converged to within 10–5 Ry. Using the climbing image nudged elastic band method, the saddle points and MEPs were located.[63,64] Zero point energy (ZPE) corrections were applied into the calculations of the activation and reaction energies from MEP analysis, in which the density functional perturbation theory within the linear response was used to study the vibrational properties.[65] The ZPEs were calculated using the PHONONS code that contained in the Quantum ESPRESSO distribution.[62] The entropies of the adsorbed surface species are ignored because of their little and almost negligible contributions to activation and reaction energies.

H Coverage-Dependent Electrochemical Model

It is known that HER and proton-coupled electron transfer during CO2 electroreduction on the Cu electrodes may involve the adsorbed H atoms as the intermediate.[7] Furthermore, the electrostatic potential difference in double electric layer can be tuned by changing the number of H atoms adsorbed on the surface.[48,66,67] Thus, H coverage-dependent CO2 electroreduction kinetics can be speculated and the corresponding equilibrium potentials can be determined. H intermediate is formed during CO2 electroreduction by the following reaction

According to eq , the Gibbs free energy of eq , ΔG(θ) can be calculated under different H coverage conditions on the basis of the approach proposed by Nørskov et al., Chen et al., and Strasser et al. for hydrogen evolution, oxygen reduction, and CO2 electroreduction reactions,[48,68−72] where ΔE(θ), ΔS(θ), ΔZPE, and kBT ln(θ/1 – θ) represent the differential adsorption energy of H, entropy change, zero-point energy change, and the contributions of configuration entropy to ΔG(θ), respectively. Here, coverage θ = n/N, in which n is the number of adsorbed H atoms on the surface and N is the total number of the surface Cu atoms. Thus, the H coverage-dependent equilibrium potential U (vs RHE) can be determined when the Gibbs free energy, ΔG(θ), is equal with zero.

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

Considering that the zero point energy of the adsorbed H atom is small based on our present calculation and the entropy of the adsorbed molecule is little when compared with the entropy of gaseous molecule, the contribution from ZPE and entropy changes together to ΔG(θ) has been estimated to be ca. 0.24 eV at standard temperature (298 K) according to the available data from previous literature.[32] Thus, eq can be transformed to the following form of eq .

The values of and EH can be obtained directly through DFT calculations. A series of values of is available by changing the coverage of adsorbed H atoms on Cu(111). At various coverages, the values of ΔE(θ) can be calculated by differentiating the plots of against θ based on eq .