Copper has many applications, particularly in electro-catalysis, where the oxidation state of the copper electrode plays a significant role in the selectivity towards products. Although copper-based materials have clear potential as catalysts in the reduction of CO2 and conversion to products, fundamental understanding of CO2 adsorption and activation on different copper oxide surfaces is still limited. We have used DFT+U methodology to study the surface reconstruction of the three most exposed (111), (110), and (001) surfaces of Cu2O with different possible terminations. Considering several adsorbate geometries, we have investigated CO2 adsorption on five different possible terminations and proposed eight different configurations in which CO2 binds with the surface. Similar to earlier findings, CO2 binds weakly with the most stable Cu2O(111):O surface showing no molecular activation, whereas a number of other surfaces, which can appear in the Cu2O particles morphology, show stronger binding as well as activation of the CO2 molecule. Different CO2 coverages were studied and a detailed structural and electronic charge analysis is presented. The activation of the CO2 molecule is characterized by structural transformations and charge transfer between the surface and the CO2 molecule, which is further confirmed by considerable red shifts in the vibrational frequencies.

Copper is a unique metal owing to its ability to selectively produce hydrocarbons through the electro-reduction of CO2,1,2 where the oxidation state of the Cu electrode plays an important role in the product selectivity. The direct reduction of CO2 to methanol (CH3OH) is known to occur on oxidized Cu electrodes, which show an increase in the methanol formation by an order of magnitude compared to metallic copper.3 The surface structure of oxidized copper resembles the Cu2O (111)4,5 surface and it reduces CO2 to CH3OH at rates remarkably higher than either air-oxidized or anodized Cu.5 In addition, our recent density functional theory (DFT) based calculations of CO2 hydrogenation on the most stable (111) surface of Cu2O show that it is a suitable catalyst for CO2 conversion to formate and formic acid under mild conditions.6 Surface analysis of these oxides, before and after the reaction, shows mixed oxidation states (Cu2O, Cu4O3, and CuO) depending on the method of preparation.7 Recently, it has been demonstrated that CuO–Cu2O nanorod arrays prepared on Cu substrates can drive the efficient solar photo-conversion of CO2 to methanol.8 

The catalytic process is affected considerably by the catalyst structure, with different shapes and surface arrangements having a large impact on the catalyst’s activity and stability. Furthermore, surface structures and crystallographic facets of metal oxides have been found to control the gas sensing properties of metal oxide-based sensors.9–11 By controlling the size and morphology, one can fine tune the strength of surface adsorption and reactivity to meet the stringent selectivity and activity requirements in a catalytic process. For example, our recent investigations of CO2 activation on a number of Cu(ii) oxide surfaces revealed that surface structures have significant effects on CO2 activation and binding energies.12 

The most exposed surfaces of Cu2O are the (111), (110), and (001),13 with the (111) surface the most stable and most studied among these surfaces.14–19 However, shape-controlled synthesis of Cu2O crystals has been investigated widely and a variety of morphologies has been synthesised successfully.20–25 Recently, a study by Sun et al. on the crystal facet-dependent effect of polyhedral Cu2O micro-crystals, that exposed different Miller index facets, showed that the catalytic performance can be enhanced by high-index facets,26 Furthermore, copper(i) oxide nano clusters have been studied recently to understand the methanol formation through DFT based calculations.27 

The adsorption of molecules on a catalyst surface is the first step in their activation and conversion in any catalytic process. CO2 adsorption on the Cu2O(111) surface has been investigated by Wu et al.,18,19 and Bendavid and Carter,28 using DFT calculations. Wu et al.,19 investigated CO2 adsorption on the Cu2O(111) surface, using the standard generalized gradient approximation (GGA) and identified that CO2 binds as a linear molecule in a tilted configuration to the surface, with its oxygen atom coordinated to a coordinatively unsaturated surface copper atom, releasing an adsorption energy of 26.8 kJ/mol. However, it is now well known that pure GGA can lead to considerable errors when calculating reactions where 3d-metal oxides are oxidized by means other than by oxygen. Reaction energies for these processes become more accurate when the so-called DFT+U method is applied.29 Bendavid and Carter28 recently investigated CO2 adsorption on the Cu2O(111) using the DFT+U method and showed via comparison to adsorption energies derived by standard DFT that the U parameter is necessary. Their choice of U = 6 eV was based on their earlier work,30 where they determined and compared different values of U to earlier DFT+U studies on Cu2O and CuO.31–33 The selection of their U value was based solely on its accuracy to predict the equilibrium lattice constant for Cu2O. However, experimentally it is found that copper oxide surfaces consist of mixed Cu2O and CuO surface species, whereas molecule interactions can also alter the oxidation state of the copper oxide surface, i.e., through −OH groups.30 Therefore, we recently determined a single U parameter to describe adequately both Cu2O and CuO in terms of experimental properties.12 In the present work, we have employed DFT with this Hubbard U correction to explore CO2 adsorption on different non-polar stoichiometric terminations of the (111), (110), and (001) surfaces of Cu2O. We first describe the reconstruction of the different surfaces and their electronic properties, followed by a detailed discussion of the CO2 adsorption behaviour.

All the calculations were performed using the Vienna Ab initio Simulation Package (VASP) with plane-wave basis set.34–37 We have employed plane-wave DFT+U38 with the PBE39,40 exchange-correlation functional and the formalism of Dudarev et al.38 The different Cu2O surfaces were obtained by the METADISE code,41 providing different non-polar surface terminations.42 At the base of the surface simulation cell, two layers of atoms were fixed at their optimised bulk positions to simulate the bulk phase of the Cu2O. Above these two layers, the surface is represented by three layers of atoms, whose positions are allowed to change freely during optimization. In each case, the vacuum region above the surface was 12 Å, i.e., large enough to avoid interactions between the periodic slabs. We sampled (1 × 1) and (2 × 2) supercells with 5 × 5 × 1 and 3 × 3 × 1 Monkhorst and Pack43 k-point mesh, respectively. Such dense grids and a truncation kinetic energy of 450 eV for the plane waves ensured an accurate description of properties that are influenced by sharp features in the density of states. A total convergence better than 10−5 was reached and the interatomic forces were minimized to 0.01 eV/Å for structural relaxation calculations.

The surface energies of the relaxed slabs were obtained using a combination of calculations for the relaxed and unrelaxed surfaces. After surface relaxation, the top and bottom surfaces are not equivalent and therefore we also need to consider the unrelaxed surface energy (γu) in order to calculate the final surface energy of the relaxed surface. The unrelaxed surface energy is the surface energy before any surface optimisation and is calculated as

γ u = E slab,u n E bulk 2 A ,
(1)

where Eslab,u is the energy of the unrelaxed slab, nEbulk is the energy of an equal number of bulk atoms, and A is the surface area of one side of the slab. Using this value, it is then possible to calculate the relaxed surface energy γ r from the total energy of the relaxed slab.

The relaxed surface energy, γr, is given by

γ r = E slab,r n E bulk A γ u ,
(2)

where Eslab,r is the energy of the relaxed slab.

The equilibrium morphology of a Cu2O particle (ignoring higher Miller indices) was constructed using Wulff’s method,44 which requires that the distance to a given surface from the center of the particle is proportional to the surface energy.

While modelling the CO2 molecule, we have also used the implementation of the DFT-D2 approach described by Grimme45 to account for long-range dispersion forces. The isolated molecule was modelled in the centre of a big cell with broken symmetry and lattice constants of 20 Å, sampling only the gamma-point of the Brillouin zone with the same accuracy parameters described for the surfaces.

The adsorption energy per molecule was calculated from the relation

E ads = E surf + mol ( E surf + E mol ) ,
(3)

where Esurf+mol is the total energy of the adsorbate-substrate system, Esurf is the energy of the naked surface slab, and Emol is the energy of the isolated CO2 molecule. Within this definition, a negative adsorption energy indicates an exothermic process.

In a recent work, we found that a value of Ueff = 7 eV results in the accurate reproduction of the structural parameters of Cu2O and a proper description of the Cu(ii) oxide.12 At this Ueff value, we found the lattice parameter of Cu2O to be 4.270 Å, which is very close to the experimental value of 4.2696 Å.46 Other structural parameters were also found to be in close agreement with the experimental values.46 We have therefore modelled the different Cu2O surfaces using the same Ueff value and employing the same bulk structural parameters.12 

In this section (Sec. III A), we have described in detail the reconstruction of the different terminations of three low-index Cu2O surfaces: (111), (110), and (001). We have calculated the surface energies of the different surface terminations from Equation (2) and determined the Wulff morphology44 of the Cu2O crystal, as shown in Fig. 1. The calculated surface energies γ r , the work functions, and the electronic band gaps of the different surfaces are listed in Table I.

FIG. 1.

The Wulff morphology of Cu2O particle determined from calculated surface energies.

FIG. 1.

The Wulff morphology of Cu2O particle determined from calculated surface energies.

Close modal
TABLE I.

The calculated relaxed surface energies γ r , work functions (ϕ), and the bandgaps (Eg) of different Cu2O surfaces.

Surface γr (J/m2) ϕ (eV) Eg (eV)
(111):O  1.08  4.98  0.78 
(111):Cu  1.92  5.10  … 
(110):Cu  1.24  5.41  0.30 
(110):Cu–O  1.54  4.39  0.15 
(100):Cu  1.62  4.54  … 
Surface γr (J/m2) ϕ (eV) Eg (eV)
(111):O  1.08  4.98  0.78 
(111):Cu  1.92  5.10  … 
(110):Cu  1.24  5.41  0.30 
(110):Cu–O  1.54  4.39  0.15 
(100):Cu  1.62  4.54  … 

1. Cu2O(111) surface

a. (111):O.

In agreement with Soon et al.,47 we found that the most stable surface is the stoichiometric non-polar oxygen-terminated (111) surface, (111):O, with a surface energy of 1.08 J/m2. The work function calculated with DFT+U is 4.98 eV, which is close to the experimental range of 4.62-4.84 eV.48 This surface consists of four distinct types of atoms: unsaturated (singly coordinated) surface copper atoms CuCUS, outermost surface oxygens OSUF, saturated copper atoms with linear O–Cu–O bond symmetry CuCSA, and sub-surface oxygens that are 4-fold coordinated OSUB (Fig. 2). The unsaturated copper atoms (CuCUS) act as Lewis acid sites, where most of the surface reactions are believed to take place.49 

FIG. 2.

The Cu2O(111):O terminated relaxed surface side view (a) and top view (b). We have shown a (2 × 2) cell in side view with periodic images of atoms for clearer visualization of bonding in all surface figures. Blue and red balls indicate Cu and O atoms, respectively, in all figures. The bond length values are in Å.

FIG. 2.

The Cu2O(111):O terminated relaxed surface side view (a) and top view (b). We have shown a (2 × 2) cell in side view with periodic images of atoms for clearer visualization of bonding in all surface figures. Blue and red balls indicate Cu and O atoms, respectively, in all figures. The bond length values are in Å.

Close modal

After relaxation, the distance of the CuCSA atoms to OSUF atoms decreases from 1.85 to 1.82 Å, but increases to the OSUB atom to 1.86 Å. As a result, these CuCSA atoms become more exposed. The top CuCUS atoms also move outwards so that the vertical bond length between CuCUS and the topmost O atoms found in the second trilayer increases from 1.85 to 1.91 Å, while the vertical bond length from the sub-surface oxygen atoms to the copper atoms in the second layer also increases to 1.89 Å. We investigated the electronic density of states (DOS) (Fig. 3(a)) of this surface and found that the bandgap slightly decreases by 0.78 eV from the calculated value of 0.89 eV for the bulk Cu2O material. The calculated values of the bandgap are expected to be under-estimated as DFT+U fails in the accurate prediction of bandgaps for Cu2O.12,32 The calculated projected DOS shows that both valence band maxima (VBM) and conduction band minima (CBM) mainly consist of O (2p) and Cu (3d) orbitals, respectively, while contributions from other orbitals are much less.

FIG. 3.

Electronic DOS of Cu2O (a) (111):O and (b) (111):Cu terminated surfaces with Fermi-level set to zero.

FIG. 3.

Electronic DOS of Cu2O (a) (111):O and (b) (111):Cu terminated surfaces with Fermi-level set to zero.

Close modal
b. (111):Cu.

We reconstructed another non-polar stoichiometric (111) surface with a Cu termination ((111):Cu), which, however, is found to be less stable by 0.84 J/m2 than the (111):O surface. The work function is found to increase slightly to 5.10 eV. The presence of two Cu atoms at both top and bottom of the slab makes the (111):Cu surface non-polar, while maintaining the bulk Cu2O ratio of Cu and O atoms (an unrelaxed (2 × 2) supercell is shown in Fig. S1 of the supplementary material). After relaxation, we noted significant changes in the positions of the top copper atoms, which moved down below the level of the O atoms. As a result, the O atoms in the relaxed surface are more exposed than the Cu atoms (Fig. 4). The Cu–O bond distance to these two Cu atoms increases slightly by 0.01 Å, while the vertical bond distance to the top O atoms from Cu atoms in the second layer decreases slightly by 0.01 Å. Other Cu–O bond distances in the second and third layers remain unchanged. We observed a finite number of states near the Fermi level in the electronic DOS of this surface and hence propose that this surface is conducting (Fig. 3(b)).

FIG. 4.

The Cu2O(111):Cu terminated relaxed surface side (a) and top view (b). The bond length values are in Å.

FIG. 4.

The Cu2O(111):Cu terminated relaxed surface side (a) and top view (b). The bond length values are in Å.

Close modal

2. Cu2O(110) surface

a. (110):Cu.

This surface consists of Cu atoms at the top of the first layer (Fig. S2 of the supplementary material) and hence we labelled this termination as (110):Cu. This is the second most stable surface with a surface energy of 1.24 J/m2, while the work function is further increased to 5.41 eV. The top Cu atoms are connected to the 4-coordinated oxygen atoms (marked OA), which are connected tetrahedrally to three more Cu atoms. The other type of O atoms (marked OB) are 3-coordinated to copper atoms. After relaxation, these top copper atoms bend along the x-axis thereby increasing their distance to OA atoms from 1.85 to 1.90 Å (Fig. 5). During the surface relaxation, the OA atoms moved up, increasing the distance from the Cu atoms of the second layer from 1.85 to 1.91 Å. The OB oxygens also move so that their distance to the lower Cu atoms changes to 1.84 from 1.85 Å. The bond length changes in the second layer are about 0.02 Å, while in the third layer they are less than 0.01 Å.

FIG. 5.

The Cu2O(110):Cu terminated relaxed surface side (a) and top view (b). The bond length values are in Å.

FIG. 5.

The Cu2O(110):Cu terminated relaxed surface side (a) and top view (b). The bond length values are in Å.

Close modal
b. (110):Cu–O.

In this termination, the surface consists of both Cu and O atoms at the top (labelled (110):Cu–O) and the calculated surface energy is 1.54 J/m2. The work function is found to be the lowest of the surfaces considered at 4.39 eV. During the reconstruction to remove the surface dipole, while keeping the ratio of Cu and O atoms the same as in the bulk, the oxygen atoms are rearranged at the top and bottom of the surface (Fig. S3 of the supplementary material). There are two distinct types of copper atoms below the top Cu–O layer, marked CuA and CuB. The CuA atoms are doubly coordinated to oxygens in the top and second layers, while the CuB atoms are only singly coordinated to an oxygen atom in the second layer. After relaxation, the top Cu and O atoms are closer and create weak Cu–O bonds of 2.10 and 2.18 Å in length (Fig. 6). The CuB type atoms are also rearranged and, after relaxation, these atoms connect with top O atoms (dCuB–O = 1.87 Å). The Cu–O bond distances in the second tri-layer increase up to 1.89 Å, while there are no structural changes in the third tri-layer.

FIG. 6.

The Cu2O(110):Cu–O terminated relaxed surface side (a) and top view (b). The bond length values are in Å.

FIG. 6.

The Cu2O(110):Cu–O terminated relaxed surface side (a) and top view (b). The bond length values are in Å.

Close modal

We calculated the electronic DOS for both terminations and found that the bandgaps for the (110) surfaces are quite low, at 0.30 and 0.15 eV for the (110):Cu and (110):Cu–O terminations, respectively (Fig. 7).

FIG. 7.

Electronic DOS of Cu2O (a) (110):Cu, (b) (110):Cu–O, and (c) (001):Cu terminated surfaces with Fermi-level set to zero.

FIG. 7.

Electronic DOS of Cu2O (a) (110):Cu, (b) (110):Cu–O, and (c) (001):Cu terminated surfaces with Fermi-level set to zero.

Close modal

3. Cu2O(001) surface

a. (001):Cu.

The (001):Cu is the only non-polar stoichiometric termination of the (001) surface. Its surface energy is calculated at 1.62 J/m2, which is 0.46 J/m2 larger than the surface energy of the most stable Cu2O(111):O surface, while the work function is 4.54 eV. This surface consists of Cu atoms in the top layer connected to oxygen atoms below, which in turn are connected to two copper atoms in the layer below (Fig. S4 of the supplementary material). We noted that after relaxation, the top Cu atoms moved down and became less exposed and the Cu–O bond distance increased from 1.85 to 1.88 Å (Fig. 8). Cu atoms in the second layer move up to shorten the bond length to oxygen atoms in the top layer from 1.85 to 1.83 Å. We also noted that the Cu–O bond distance in all other relaxed surfaces increases from 1.85 Å and varies from 1.86 to 1.88 Å. With finite states near the Fermi level, this surface is also found to be conducting (Fig. 7).

FIG. 8.

The Cu2O(001):Cu terminated relaxed surface side (a) and top view (b). The bond length values are in Å.

FIG. 8.

The Cu2O(001):Cu terminated relaxed surface side (a) and top view (b). The bond length values are in Å.

Close modal

1. Cu2O(111) surface

a. (111):O surface.

A (1 × 1) slab (a = b = 6.04 Å) consists of 20 copper and 10 oxygen atoms. We first considered the (1 × 1) cell of the (111) surface for CO2 adsorption and investigated a number of initial configurations with different orientations of the CO2 molecule. We found that the CO2 molecule moved away from the (111):O surface for all configurations, except where we placed it near the coordinatively unsaturated surface copper, CuCUS. In this configuration one of the oxygen atoms, O1, of the CO2 molecule binds weakly with this CuCUS copper atom, as shown in Fig. 9. The CO2 molecule remains almost linear with an angle of 176.9°. The distance between the oxygen atom O1 of the CO2 molecule and CuCUS is found to be 2.05 Å, and the C–O bond between C and this O1 atom is slightly stretched at 1.19 Å, while the C–O2 bond length is found to be around 1.17 Å. Cu–O bond lengths in the slab also change slightly as a result of CO2 adsorption, where the vertical bond distance between CuCUS (coordinated to the O1 atom of the CO2 molecule) and the topmost O atom found in the second trilayer shortens from 1.91 to 1.88 Å. The adsorption energy in this configuration is −51.0 kJ/mol.

FIG. 9.

The CO2 molecule adsorbed on the Cu2O(111):O terminated surface. Black balls indicate C atom of CO2 molecule, green balls indicate O atoms of the molecule, while blue and red balls denote the surface Cu and O atoms in all the figures.

FIG. 9.

The CO2 molecule adsorbed on the Cu2O(111):O terminated surface. Black balls indicate C atom of CO2 molecule, green balls indicate O atoms of the molecule, while blue and red balls denote the surface Cu and O atoms in all the figures.

Close modal

In order to assess the effect of CO2 coverage, we repeated our calculation by placing one CO2 molecule in a (2 × 2) supercell; we found that the adsorption energy increases to −56.1 kJ/mol, but with negligible changes in the CO2 geometry. Adsorption geometries of the CO2 molecule on both (1 × 1) and (2 × 2) supercell are given in Table II. Our calculated geometrical parameters of the adsorbed CO2 molecule and the binding energies are in reasonable agreement with the recent work of Bendavid et al., where they used similar DFT(D)+U (6 eV) methodology and found ∠CO2 to be 177.1°and an adsorption energy of −36.4 kJ/mol.28 This small change in adsorption energy value is expected as we have not included entropy and enthalpy energy corrections in our calculated adsorption energies.

TABLE II.

The adsorption energies and the characteristic parameter values of the CO2 adsorbed geometry in the (1 × 1) and the (2 × 2) supercell of the Cu2O(111):O surface.

Supercell Eads (kJ/mol) ∠CO2 (deg) dC–O1 (Å) dC–O2 (Å) dO1–CuCUS (Å)
(1 × 1)  −51.0  176.9  1.19  1.17  2.05 
(2 × 2)  −56.1  178.3  1.18  1.18  2.05 
Supercell Eads (kJ/mol) ∠CO2 (deg) dC–O1 (Å) dC–O2 (Å) dO1–CuCUS (Å)
(1 × 1)  −51.0  176.9  1.19  1.17  2.05 
(2 × 2)  −56.1  178.3  1.18  1.18  2.05 

A Bader charge analysis of the (1 × 1) cell (Table III) shows that the oxygen atom O1 of the CO2 molecule (bonded to the coordinatively unsaturated surface copper CuCUS) gains 0.03e, resulting from a small charge transfer from the surface copper atom CuCus, which becomes more oxidized after CO2 adsorption. This very small charge transfer between the surface and the CO2 molecule, as well as small changes in vibrational frequencies (Table III) indicates weak activation of the CO2 molecule.

TABLE III.

Vibrational frequencies (cm−1) and Bader charges (e) comparison of the atoms in the adsorbed CO2 molecule and the Cu2O(111):O surface atoms bonded with the molecule to that of the atoms in the isolated CO2 molecule and the bare surface in the (1 × 1) cell.

Atoms and vibrational modes C O1 O2 CuCUS υas υs υb
Adsorbed CO2 molecule  2.08  −1.07  −1.02  0.50  2332  1292  567 
Isolated CO2 molecule  2.08  −1.04  −1.04  …  2355  1316  632 
Bare surface  …  …  …  0.44  …  …  … 
Atoms and vibrational modes C O1 O2 CuCUS υas υs υb
Adsorbed CO2 molecule  2.08  −1.07  −1.02  0.50  2332  1292  567 
Isolated CO2 molecule  2.08  −1.04  −1.04  …  2355  1316  632 
Bare surface  …  …  …  0.44  …  …  … 
b. (111):Cu surface.

We calculated the CO2 adsorption of numerous input configurations, placing the CO2 molecule at different sites on the surface in different orientations, and we found that the CO2 molecule binds in two configurations. In the first configuration (config. 1) after optimisation, the top Cu atoms CuA and CuB have moved upwards to interact with one of the CO2 oxygen atoms O1, as shown in Fig. 10(a). The O1–Cu distances are 1.89 and 1.97 Å for CuA and CuB, respectively. The other oxygen atom, O2, of the CO2 molecule remained unbound in this configuration. The CO2 molecule bends with ∠CO2 = 125.5°, as the carbon atom moved down to interact with a surface oxygen atom, OSUF, in the second layer (dC–OSUF = 1.41 Å). The C–O1 bond length becomes slightly elongated, dC–O1 = 1.34 Å, while the C–O2 bond is 1.22 Å long, i.e., longer than in the gas phase, which, together with the bending of the CO2, is related to the activation of the molecule.50 Upon CO2 adsorption, the bond distance between the top Cu and O atoms changes from 1.86 Å to 1.83. We noted that the surface oxygen atoms, which were connected in a vertical linear manner to Cu and O atoms in the second and third layer, respectively, bend towards the CO2 molecule with loss of linearity. The adsorption energy calculated in this configuration is −117.1 kJ/mol.

FIG. 10.

The CO2 molecule adsorbed on the Cu2O(111):Cu terminated surface in the (a) (1 × 1) cell, (b) (1 × 2) supercell in config. 1 and in the (c) (1 × 1) cell, (d) (1 × 2) supercell in config. 2.

FIG. 10.

The CO2 molecule adsorbed on the Cu2O(111):Cu terminated surface in the (a) (1 × 1) cell, (b) (1 × 2) supercell in config. 1 and in the (c) (1 × 1) cell, (d) (1 × 2) supercell in config. 2.

Close modal

We noted that due to the orientation of the CO2 molecule, the lateral distance in the x-direction between the CO2 molecule and its periodic image is 6.04 Å, while in the y-direction, it is only 3.80 Å. Hence, to minimize the effect of the periodic images on the CO2 adsorption, we carried out calculations on a (1 × 2) supercell. At this lower coverage, CO2 adsorbs in a slightly different manner, as the top surface Cu atoms (CuA and CuB) interact with both CO2 oxygen atoms at distances of 2.03 Å and 2.01 Å, respectively (Fig. 10(b)). As a result, the CuA and CuB bond lengths with oxygen atoms in the surface change to 1.91 and 1.90 Å, respectively. Because of the lower coverage of CO2 molecules on the surface, other surface Cu atoms (further away from the CO2 molecule) bend inwards to bind to O atoms in the second layer, as shown in Fig. 10(b). As expected, the adsorption energy increases to about −161.5 kJ/mol. Similar to the (1 × 1) cell configuration, the C atom of the CO2 molecule bends towards a surface oxygen atom OSUF in the second layer (dC–OSUF = 1.42 Å). The angle of the adsorbed CO2 molecule is 129.0° and both C–O bond lengths are 1.27 Å. We have given parameters of the CO2 adsorption geometries in the (1 × 1) and (1 × 2) simulation cells in Table IV.

TABLE IV.

The adsorption energies and the characteristic parameter values of the CO2 adsorbed geometry in the (1 × 1) and the (1 × 2) supercell of the Cu2O(111):Cu surface in config. 1 and config. 2.

Supercell Eads (kJ/mol) ∠CO2 (deg) dC–O1 (Å) dC–O2 (Å) dO1–CuA (Å) dO1–CuB (Å) dC–OSUF (Å)
Config. 1               
(1 × 1)  −117.1  125.5  1.34  1.22  1.89  1.97  1.41 
(1 × 2)  −161.5  129.0  1.27  1.27  1.91  1.90  1.42 
Config. 2               
(1 × 1)  −97.1  133.2  1.26  1.26  2.14  2.14  1.44 
(1 × 2)  −232.6  119.2  1.30  1.30  1.85  1.85  1.32 
Supercell Eads (kJ/mol) ∠CO2 (deg) dC–O1 (Å) dC–O2 (Å) dO1–CuA (Å) dO1–CuB (Å) dC–OSUF (Å)
Config. 1               
(1 × 1)  −117.1  125.5  1.34  1.22  1.89  1.97  1.41 
(1 × 2)  −161.5  129.0  1.27  1.27  1.91  1.90  1.42 
Config. 2               
(1 × 1)  −97.1  133.2  1.26  1.26  2.14  2.14  1.44 
(1 × 2)  −232.6  119.2  1.30  1.30  1.85  1.85  1.32 

Bader charge analysis of the (1 × 2) supercell shows charge transfer between the CO2 molecule and the surface, as both molecular oxygens O1 and O2 gain 0.08e and 0.07e charge densities, respectively. This charge transfer originates mainly from the interacting surface copper atoms CuA and CuB, which become more positively charged after adsorption. The OSUF atom bound to the molecule also gains 0.11e charge density (Table V). We also note some charge redistribution on the Cu2O surface as a result of CO2 adsorption. Bader analysis indicates the CO2 molecule as a chemisorbed anion on the surface, in agreement with the molecular orbital occupation and bending of the molecule. This activation of the CO2 molecule is also reflected in terms of changes in the vibrational frequencies of the molecule, as asymmetric (υas) and symmetric (υs) stretching modes change to 1560 and 1200 cm−1 from their values of 2355 and 1316 cm−1, respectively, in the isolated gas phase molecule (Table V).

TABLE V.

Vibrational frequencies (cm−1) and Bader charges (e) comparison of the atoms in the adsorbed CO2 molecule and the Cu2O(111):Cu surface atoms bonded with the molecule in the (1 × 2) supercell to that of the atoms in the isolated CO2 molecule and the bare surface in config. 1 and config. 2.

Atoms and vibrational modes C O1 O2 CuA CuB OSUF υas υs υb
After CO2 adsorption (config. 1)  2.02  −1.12  −1.11  0.54  0.55  −1.03  1560  1200  748 
After CO2 adsorption (config. 2)  2.09  −1.06  −1.07  0.56  0.58  −1.15  1395  1257  858 
Isolated CO2 molecule  2.08  −1.04  −1.04  …  …  …  2355  1316  632 
Bare surface  …  …  …  0.41  0.40  −0.92  …  …  … 
Atoms and vibrational modes C O1 O2 CuA CuB OSUF υas υs υb
After CO2 adsorption (config. 1)  2.02  −1.12  −1.11  0.54  0.55  −1.03  1560  1200  748 
After CO2 adsorption (config. 2)  2.09  −1.06  −1.07  0.56  0.58  −1.15  1395  1257  858 
Isolated CO2 molecule  2.08  −1.04  −1.04  …  …  …  2355  1316  632 
Bare surface  …  …  …  0.41  0.40  −0.92  …  …  … 

In a different configuration (config. 2), CO2 binds to the (111):Cu terminated surface through its C atom to a surface oxygen atom (dC–OSUF = 1.44 Å), while both oxygen atoms of the molecule bind to CuA and CuB (dO–Cu = 2.14 Å), as shown in Fig. 10(c). The CO2 molecule again bends to ∠CO2 = 133.2°, while the Cu–O–Cu angle in the surface is about 145.4°. We found the surface Cu–O bonds to be slightly more stretched with bond distances of 1.98 Å. The adsorption energy at this coverage is −97.1 kJ/mol, which is slightly less than the same coverage in config. 1. Similar to config. 1, we also investigated a lower coverage of CO2 at the surface in a (1 × 2) supercell (Fig. 10(d)). At this coverage, after CO2 adsorption, surface rearrangement takes place where copper atoms CuA and CuB break their bonds with the OSUF atom to form new bonds to surface oxygen atoms nearby, as well as bind to both CO2 oxygen atoms (dO–Cu = 1.85 Å). The carbon atom binds more strongly to surface atom OSUF (dC–OSUF = 1.32 Å) as the CO2 angle changes to ∠CO2 = 119.0°, and we noted that ∠O1–C–OSUF and ∠O2–C–OSUF are ∼120.0°. The adsorption energy increases to −232.6 kJ/mol (Table IV). Despite this large adsorption energy, Bader charge comparison (Table V) of the free CO2 molecule with that in the adsorbed geometry shows that there is very little charge transfer, although large charge redistribution takes place among the surface atoms bonded to the molecule. OSUF atom gains 0.23e charge density, while CuA and CuB both lose 0.16e and 0.17e in charge densities, respectively. This charge redistribution together with the change in the surface results in a CO3 like-species on the (111):Cu surface (Fig. 10(d)). Unstable surfaces are often highly reactive, which is exemplified by this behaviour of the (111):Cu surface. This strong activation of the CO2 molecule is further confirmed by considerable changes in the vibrational modes of the adsorbed CO2 molecule, where asymmetric stretch (υas), symmetric stretch (υs), and bending (υb) frequencies change to 1395, 1257, and 858 cm−1, respectively, from their original values of 2355, 1316, and 632 cm−1 in the isolated gas phase molecule.

2. Cu2O(110) surface

a. (110):Cu.

For this surface, we first considered a (1 × 1) unit cell and tried different initial configurations with several orientations of the CO2 molecule, but we found only one configuration in which CO2 binds to the surface. Here, CO2 binds strongly (Eads = − 100 kJ/mol) in a configuration where the molecule bends to bind with an oxygen atom in the second layer (dC–OSUF = 1.45 Å), while its oxygen atoms O1 and O2 bind to surface atoms, CuA and CuB, at 1.97 Å (Fig. 11(a)). We noted that the CO2 molecule is activated with an angle of ∠CO2 = 128.0°. From Fig. 11(a), we observe that the distance between the CO2 molecule and its image in the x-direction is 4.3 Å, while in the y-direction it is only 3.78 Å. We therefore repeated the calculations of all the different configurations in (2 × 1) and (1 × 2) supercells.

FIG. 11.

The CO2 molecule adsorbed on the Cu2O(110):Cu terminated surface in the (a) (1 × 1) cell and in the (b) (1 × 2) supercell.

FIG. 11.

The CO2 molecule adsorbed on the Cu2O(110):Cu terminated surface in the (a) (1 × 1) cell and in the (b) (1 × 2) supercell.

Close modal

Keeping the same input orientations, we first assessed the effect of a lower CO2 coverage by placing one molecule in a (2 × 1) supercell and found that Eads increased to −105.0 kJ/mol, while in the (1 × 2) supercell, Eads increased to −116.7 kJ/mol. This increase in Eads was expected because of the small distance between the CO2 molecule and its periodic image in the y-direction in the (1 × 1) cell. Because of the significant difference in Eads in the (1 × 2) supercell compared to the (2 × 1) supercell, we have limited our discussion only to the more favourable (1 × 2) supercell system. In the (1 × 2) supercell, the carbon atom of the molecule binds strongly to the surface oxygen atom (dC–OSUF = 1.42 Å), while CuA–O1 and CuB–O2 bond lengths reduce to 1.89 Å (Fig. 11(b)). We have given geometrical parameters of the adsorbed geometry of the (1 × 1) and (1 × 2) supercells in Table VI. Bader analysis of the (1 × 2) supercell (Table VII) shows charge transfer between the oxygen atoms of CO2 and surface copper atoms. Oxygen atoms O1 and O2 gain 0.05 and 0.06e, respectively, while both surface copper atoms CuA and CuB lose 0.12e. There is a very small charge transfer to the carbon atom of the CO2 molecule of ∼0.01e. This amount of charge transfer is consistent with the charge transfer in the (111):Cu surface, where the molecule’s oxygen gains ∼0.08e and surface copper atoms lose charge of ∼0.15e. Here also, frequencies for asymmetric stretch (υas), symmetric stretch (υs), and bending (υb) vibrations change to 1639, 1247, and 808 cm−1, indicating activation of the CO2 molecule on the (110):Cu surface. In all other configurations considered, the CO2 molecule does not bind to the copper oxide surface.

TABLE VI.

The adsorption energies and the characteristic parameter values of the CO2 adsorbed geometry in the (1 × 1) and the (1 × 2) supercell of the Cu2O(110):Cu surface.

Supercell Eads (kJ/mol) ∠CO2 (deg) dC–O1 (Å) dC–O2 (Å) dO1–CuA (Å) dO1–CuB (Å) dC–OSUF (Å)
(1 × 1)  −100.4  128.0  1.26  1.26  1.97  1.97  1.45 
(1 × 2)  −116.7  126.2  1.26  1.26  1.89  1.89  1.42 
Supercell Eads (kJ/mol) ∠CO2 (deg) dC–O1 (Å) dC–O2 (Å) dO1–CuA (Å) dO1–CuB (Å) dC–OSUF (Å)
(1 × 1)  −100.4  128.0  1.26  1.26  1.97  1.97  1.45 
(1 × 2)  −116.7  126.2  1.26  1.26  1.89  1.89  1.42 
TABLE VII.

Vibrational frequencies (cm−1) and Bader charges (e) comparison of the atoms in the adsorbed CO2 molecule and the Cu2O(110):Cu surface atoms bonded with the molecule to that of in the isolated CO2 molecule and the bare surface in the (1 × 2) supercell.

Atoms and vibrational modes C O1 O2 CuA CuB OSUF υas υs υb
After CO2 adsorption  2.07  −1.09  −1.10  0.56  0.56  −1.04  1639  1247  808 
Isolated CO2 molecule  2.08  −1.04  −1.04  …  …  …  2355  1316  632 
Bare surface  …  …  …  0.44  0.44  −0.95  …  …  … 
Atoms and vibrational modes C O1 O2 CuA CuB OSUF υas υs υb
After CO2 adsorption  2.07  −1.09  −1.10  0.56  0.56  −1.04  1639  1247  808 
Isolated CO2 molecule  2.08  −1.04  −1.04  …  …  …  2355  1316  632 
Bare surface  …  …  …  0.44  0.44  −0.95  …  …  … 
b. (110):Cu–O.

Here again, we carried out calculations on a (2 × 1) supercell, exploring different configurations for CO2 to bind with the surface. In the first configuration (config. 1), after placing the CO2 molecule parallel to the Cu–O–Cu linear bond in the top layer, we found that this bond breaks when Cu atoms move up to bind to oxygen atoms of the CO2 molecule, while the carbon atom bends down to bind to the oxygen atom of the top surface layer (dC–OSUF = 1.36 Å), as shown in Fig. 12(a). One of the CO2 oxygen atoms (O1) binds to one of the nearest Cu atoms (CuA) in the top layer with a bond distance dO1–CuA = 1.84 Å, while the second oxygen (O2) binds to another surface copper atom with a bond distance dO2–CuB = 1.86 Å, causing the Cu–O distances of CuA and CuB to their neighbouring surface oxygen atoms to change from 2.10 and 2.18 Å to 1.83 and 1.84 Å, respectively. The CO2 molecule bends to an angle of ∠CO2 = 123.6° and adsorbs strongly with an adsorption energy Eads = − 138.1 kJ/mol. In this configuration, the C–O bond length is 1.27 Å, which is slightly longer than the normal bond length in a CO2 molecule. The adsorption energy and geometrical parameters of the system are given in Table VIII. Bader charge analysis (Table IX) reveals that the O1 and O2 oxygen atoms of the CO2 molecule gain 0.13 and 0.10e in charge density, respectively. Meanwhile, the surface copper atoms, CuA and CuB, which are bound to the O1 and O2 lose a charge density of 0.49 and 0.39e, respectively, while the surface oxygen atom (OSUF) bound to the carbon atom of the CO2 molecule gains 0.14e. Hence, upon CO2 adsorption, a large charge redistribution occurs at the surface. This strong adsorption and activation of the CO2 molecule are reflected in changes in the vibrational frequencies, as listed in Table IX.

FIG. 12.

The CO2 molecule adsorbed on the Cu2O(110):Cu–O terminated surface in the (2 × 1) cell in (a) config. 1 and (b) config. 2.

FIG. 12.

The CO2 molecule adsorbed on the Cu2O(110):Cu–O terminated surface in the (2 × 1) cell in (a) config. 1 and (b) config. 2.

Close modal
TABLE VIII.

The adsorption energies and the characteristic parameter values of the CO2 adsorbed geometry in the (2 × 1) supercell of the Cu2O(110):Cu–O surface in config. 1 and config. 2.

Supercell Eads (kJ/mol) ∠CO2 (deg) dC–O1 (Å) dC–O2 (Å) dO1–CuA (Å) dO2–CuB (Å) dC–OSUF (Å)
Config. 1               
(2 × 1)  −138.1  123.6  1.27  1.27  1.84  1.86  1.36 
Config. 2               
(2 × 1)  −169.9  122.7  1.27  1.30  1.85  1.86  1.35 
Supercell Eads (kJ/mol) ∠CO2 (deg) dC–O1 (Å) dC–O2 (Å) dO1–CuA (Å) dO2–CuB (Å) dC–OSUF (Å)
Config. 1               
(2 × 1)  −138.1  123.6  1.27  1.27  1.84  1.86  1.36 
Config. 2               
(2 × 1)  −169.9  122.7  1.27  1.30  1.85  1.86  1.35 

In the second configuration (config. 2) (Fig. 12(b)), the carbon atom binds to the OSUF atom (dC–OSUF = 1.35 Å) and O1 binds to CuA (dO1–CuA = 1.85 Å) in the top layer. The CO2 molecule bends so that O2 binds with CuB of the second layer (dO2–CuB = 1.86 Å), as shown in Fig. 12(b). In this configuration, we found that CO2 binds more strongly with an adsorption energy of about −170.0 kJ/mol. CO2 bends to an angle of ∠CO2 = 122.7° which indicates stronger activation (Table VIII). Oxygen–carbon bond lengths, i.e., dC–O1 and dC–O2, are found to be 1.27 Å and 1.30 Å, respectively. After CO2 adsorption, the Cu–O bonds in the top of the surface change from 2.10 Å and 2.18 Å to 1.86 Å and 1.82 Å for CuA and CuB, respectively. Bader charge analysis (Table IX) shows that the oxygen atoms O1 and O2 of the CO2 molecule gain −0.13 and −0.08e, respectively, while the change in the carbon atom charge is negligible. The surface copper atom CuA bonded to one of the oxygen atoms of the CO2 molecule becomes more positively charged with a loss of 0.26e, while the surface oxygen atom OSUF gains 0.23e. Charge transfer to CuB is negligible. This charge transfer, the adsorption energy, geometry changes together with large changes in vibrational frequencies (Table IX) indicates activation of the CO2 molecule.

TABLE IX.

Vibrational frequencies (cm−1) and Bader charges (e) comparison of the atoms in the adsorbed CO2 molecule and the Cu2O(110):Cu–O surface atoms bonded with the molecule to that of in the isolated CO2 molecule and the bare surface in config. 1 and config. 2 of (2 × 1) supercell.

Atoms and vibrational modes C O1 O2 CuA CuB OSUF υas υs υb
After CO2 adsorption (config. 1)  2.07  −1.14  −1.17  0.91  0.81  −1.03  1580  1291  869 
After CO2 adsorption (config. 2)  2.07  −1.17  −1.12  0.53  0.69  −1.12  1508  1250  851 
Isolated CO2 molecule  2.08  −1.04  −1.04  …  …  …  2355  1316  632 
Bare surface  …  …  …  0.42  0.42  −0.89  …  …  … 
Atoms and vibrational modes C O1 O2 CuA CuB OSUF υas υs υb
After CO2 adsorption (config. 1)  2.07  −1.14  −1.17  0.91  0.81  −1.03  1580  1291  869 
After CO2 adsorption (config. 2)  2.07  −1.17  −1.12  0.53  0.69  −1.12  1508  1250  851 
Isolated CO2 molecule  2.08  −1.04  −1.04  …  …  …  2355  1316  632 
Bare surface  …  …  …  0.42  0.42  −0.89  …  …  … 

3. Cu2O(001) surface

a. (001):Cu.

The (1 × 1) cell (a = 4.27 Å, b = 4.27 Å) is too small to study adsorption of the isolated molecule, as we discovered earlier that the interaction between neighbouring CO2 molecules affects the geometry and adsorption energies. Hence, we carried out calculations on the (2 × 2) supercell to minimize the effect of interactions between periodic images. Similar to the other surfaces, we studied different sites at the surface for possible CO2 adsorption. In the first adsorption configuration (config. 1), oxygens O1 and O2 of the molecule bind to the nearest topmost copper atoms CuA and CuB with bond distances of 1.85 and 1.87 Å, respectively (Fig. 13(a)). The CO2 molecule bends to an angle of 121.5° as the carbon atom moves down to bind with the nearest available oxygen atom (OSUF) of the top layer with a bond length dC–OSUF of 1.35 Å. Upon CO2 adsorption, the bond between CuB and OSUF is broken, as shown in Fig. 8(a). The adsorption energy, Eads, calculated for this configuration is −138 kJ/mol. Carbon–oxygen bond lengths in the CO2 molecule are found to be 1.29 and 1.28 Å for O1 and O2, respectively. Copper–oxygen bond lengths in the surface for CuA and CuB change from 1.88 to 1.84 Å and 1.86 Å, respectively (Table X). Bader charge analysis (Table XI) shows significant charge transfer between the CO2 molecule and the surface atoms. After CO2 adsorption, both O1 and O2 oxygens of the CO2 molecule lose charge density of 0.24e and 0.22e, respectively, while surface copper atoms CuA and CuB which bind to these two oxygen atoms lose charge density of 0.12 and 0.10e, respectively. However, the carbon atom gains 0.56e after binding to surface oxygen atom OSUF, which loses 0.12e. Hence, charge transfer has occurred to the CO2 molecule from nearby surface atoms. As shown in Table XI, we note considerable changes in the frequencies of the different vibrational modes (υas = 1509, υs = 1281, and υb = 869 cm−1) of the activated CO2 molecule on this surface.

FIG. 13.

The CO2 molecule adsorbed on the Cu2O(001):Cu terminated surface in the (2 × 2) cell in (a) config. 1 and (b) config. 2.

FIG. 13.

The CO2 molecule adsorbed on the Cu2O(001):Cu terminated surface in the (2 × 2) cell in (a) config. 1 and (b) config. 2.

Close modal
TABLE X.

The adsorption energies and the characteristic parameter values of the CO2 adsorbed geometry in the (2 × 2) supercell of Cu2O(001):Cu surface in config. 1 and config. 2.

Supercell Eads (kJ/mol) ∠CO2 (deg) dC–O1 (Å) dC–O2 (Å) dO1–CuA (Å) dO1–CuB (Å) dO2–CuB (Å) dO2–CuC (Å) dC–OSUF (Å)
Config. 1                 
(2 × 2)  −138.1  121.5  1.29  1.28  1.85  …  1.87  …  1.35 
Config. 2                 
(2 × 2)  −98.7  121.7  1.32  1.27  1.95  1.91  …  1.91  1.32 
Supercell Eads (kJ/mol) ∠CO2 (deg) dC–O1 (Å) dC–O2 (Å) dO1–CuA (Å) dO1–CuB (Å) dO2–CuB (Å) dO2–CuC (Å) dC–OSUF (Å)
Config. 1                 
(2 × 2)  −138.1  121.5  1.29  1.28  1.85  …  1.87  …  1.35 
Config. 2                 
(2 × 2)  −98.7  121.7  1.32  1.27  1.95  1.91  …  1.91  1.32 
TABLE XI.

Vibrational frequencies (cm−1) and Bader charges (e) comparison of the atoms in the adsorbed CO2 molecule and Cu2O(001):Cu surface atoms bonded with the molecule to that of in the isolated CO2 molecule and the bare surface in config. 1 and config. 2 of (2 × 2) supercell.

Atoms and vibrational modes C O1 O2 CuA CuB CuC OSUF υas υs υb
After CO2 adsorption (config. 1)  1.52  −0.80  −0.82  0.46  0.44  …  −0.77  1509  1281  869 
After CO2 adsorption (config. 2)  1.86  −0.96  −1.11  0.59  0.61  0.65  −0.93  1461  1242  875 
Isolated CO2 molecule  2.08  −1.04  −1.04  …  …  …  …  2355  1316  632 
Bare surface  …  …  …  0.34  0.34  0.51  −0.89  …  …  … 
Atoms and vibrational modes C O1 O2 CuA CuB CuC OSUF υas υs υb
After CO2 adsorption (config. 1)  1.52  −0.80  −0.82  0.46  0.44  …  −0.77  1509  1281  869 
After CO2 adsorption (config. 2)  1.86  −0.96  −1.11  0.59  0.61  0.65  −0.93  1461  1242  875 
Isolated CO2 molecule  2.08  −1.04  −1.04  …  …  …  …  2355  1316  632 
Bare surface  …  …  …  0.34  0.34  0.51  −0.89  …  …  … 

In another configuration (config. 2) the CO2 molecule remains almost parallel to the surface, as shown in Fig. 13(b). The carbon atom binds to a top oxygen atom (OSUF) (dC–OSUF = 1.32 Å), while one oxygen (O1) of the CO2 molecule binds to surface copper atoms CuA and CuB, with bond lengths 1.95 and 1.91 Å, respectively. The second oxygen atom (O2) of the molecule binds to another surface copper atom, CuC with a bond length of 1.91 Å. In the adsorbed CO2 molecule we find O–C bond lengths of 1.32 and 1.27 Å for oxygen atoms O1 and O2, respectively, and the CO2 angle is ∠CO2 = 121.7°. The adsorption energy in this configuration is −98.7 kJ/mol, which is almost 40 kJ/mol less than in the first configuration (Table X). The Bader charge analysis shows charge redistribution, with oxygen O1 losing 0.08e charge density, while the CuA and CuB atoms also lose 0.25e and 0.27e, respectively. Oxygen atom O2 gains 0.07e charge density after binding to surface copper atom CuC, which loses 0.14e. Charge transfer also occurs between surface oxygen atom, OSUF and the C atom which gains 0.22e, while OSUF gains 0.04e (Table XI). Here again, we note considerable changes in the vibrational frequencies of the adsorbed CO2 molecule, as shown in Table XI.

We noted that the activation of the CO2 molecule is related to the adsorption energy, as shown in Fig. 14(a), where we considered the molecule’s angle as a measure of its activation.50 Furthermore, the adsorption energy, Eads, depends almost linearly on the surface stability (Fig. 14(b)): The most stable (111):O surface shows weak adsorption (approximately −56 kJ/mol) and the CO2 molecule remains linear, while the least stable (111):Cu surface shows the strongest binding (Eads = − 233 kJ/mol) with the CO2 molecule bent to 119.2°. The second most stable surface, (110):Cu, shows an adsorption energy of approximately −117 kJ/mol, while the (001):Cu surface, binds the CO2 molecule with the release of ∼138 kJ/mol, with the CO2 molecule bending to 126.2 and 121.5°, respectively. This trend is similar to that found in CO2 adsorption on CuO surfaces, where the most stable surface, (111), shows weak binding of the CO2 molecule compared with other low index surfaces, which cause significant CO2 activation.11 

FIG. 14.

The variation of CO2 activation (∠OCO) (a) and adsorption energy (Eads) (b) with surface energies of different surfaces. For the sake of simplicity we have plotted only lowest coverage values of only those configurations that show strongest binding with CO2.

FIG. 14.

The variation of CO2 activation (∠OCO) (a) and adsorption energy (Eads) (b) with surface energies of different surfaces. For the sake of simplicity we have plotted only lowest coverage values of only those configurations that show strongest binding with CO2.

Close modal

Using DFT+U methodology, we have studied the reconstructions of the (111), (110), and (001) surfaces and proposed different non-polar terminations. We further analysed the structural geometries, energetics, and electronic properties for the process of carbon dioxide adsorption to different stoichiometric Cu2O surfaces, at different coverages. While the CO2 adsorption to stoichiometric Cu2O(111):O is weak, causing no significant changes to the geometry or electronic structure of the adsorbate, CO2 adsorption to all other surfaces is energetically favourable. The (110):Cu surface, which is only less stable by ∼0.16 J/m2 compared to the most stable (111):O surface, shows adsorption energies up to approximately −117 kJ/mol, while the third most stable (110):Cu–O surface exhibits strong chemisorption of the CO2 molecule, releasing ∼170 kJ/mol. We found that CO2 coverage affects the adsorption energy as Eads increases for all surfaces at lower coverage, where CO2 is found to be chemisorbed as the CO2 anion. The Cu2O(111) surface with Cu termination is found to be the least stable surface and a detailed structural and Bader charge analysis shows that the CO2 molecule affects the surface geometry, rearranging itself to resemble a [CO3] species on the surface.

Activation of carbon dioxide is the most important step in its conversion into valuable chemicals and large structural transformations and significant charge transfer between different surfaces and the CO2 molecule demonstrate that Cu2O is capable of activating CO2. For all the bent CO2 configurations, we note a significant red-shift on the C–O symmetric (υs) and asymmetric (υas) stretching modes relative to the linear gas phase molecule, indicating that the CO2 molecule is considerably activated. It is worth noting, however, that our calculations are valid only at 0 K and only stoichiometric surfaces were considered in the present study. Nevertheless, the results presented in this paper provide fundamental mechanistic insights into CO2 activation on stoichiometric (111), (110), and (001) surfaces, which will still be relevant to our general understanding of CO2 adsorption by Cu2O, as the different surfaces studied here include a wide variety of the kind of surface sites, that can be expected to occur on experimental surfaces.

The reconstructed unrelaxed surface structures of (111):Cu, (110):Cu, (110):CuO–O, and (001):Cu are given as Figs. S1–S4 in the supplementary material, respectively.

This work was carried out as part of the Engineering and Physical Sciences Research Council (EPSRC) “4CU” programme grant, aimed at sustainable conversion of carbon dioxide into fuels, led by The University of Sheffield and carried out in collaboration with University College London, University of Manchester, and Queens University Belfast. The authors acknowledge the EPSRC for supporting this work financially (Grant Nos. EP/K001329/1 and EP/K035355/1). Via our membership of the UK’s HEC Materials Chemistry Consortium, which is funded by EPSRC (Grant No. EP/L000202), this work used the ARCHER UK National Supercomputing Service (http://www.archer.ac.uk). The authors also acknowledge the use of the IRIDIS High Performance Computing Facility, and associated support services at the University of Southampton, in the completion of this work. N.H.d.L. thanks the Royal Society for an Industry Fellowship.

1.
Y.
Hori
, in
Modern Aspects of Electrochemistry
, edited by
C.
Vayenas
,
R.
White
, and
M.
Gamboa-Aldeco
(
Springer
,
New York
,
2008
), Vol.
42
, Chap. 3, pp.
89
189
.
2.
M.
Gattrell
,
N.
Gupta
, and
A.
Co
,
J. Electroanal. Chem.
594
,
1
(
2006
).
3.
J.
Szanyi
and
D. W.
Goodman
,
Catal. Lett.
10
,
383
(
1991
).
4.
T.
Matsumoto
,
R. A.
Bennett
,
P.
Stone
,
T.
Yamada
,
K.
Domen
, and
M.
Bowker
,
Surf. Sci.
471
,
225
(
2001
).
5.
S. Y.
Lee
,
N.
Mettlach
,
N.
Nguyen
,
Y. M.
Sun
, and
J. M.
White
,
Appl. Surf. Sci.
206
,
102
(
2003
).
6.
A. K.
Mishra
and
N. H.
de Leeuw
, “
Mechanistic insights into the Cu(I) oxide-catalyzed conversion of CO2 to fuels and chemicals: A DFT approach
,”
J. CO2 Util.
(in press).
7.
M.
Le
,
M.
Ren
,
Z.
Zhang
,
P. T.
Sprunger
,
R. L.
Kurtz
, and
J. C.
Flake
,
J. Electrochem. Soc.
158
,
E45
(
2011
).
8.
G.
Ghadimkhani
,
N. R.
de Tacconi
,
W.
Chanmanee
,
C.
Janaky
, and
K.
Rajeshwar
,
Chem. Commun.
49
,
1297
(
2013
).
9.
C.
Wang
,
L.
Yin
,
L.
Zhang
,
D.
Xiang
, and
R.
Gao
,
Sensors (Basel)
10
,
2088
(
2010
).
10.
O.
Lupan
,
V.
Cretu
,
V.
Postica
,
N.
Ababii
,
O.
Polonskyi
,
V.
Kaidas
,
F.
Schütt
,
Y. K.
Mishra
,
E.
Monaico
,
I.
Tiginyanu
,
V.
Sontea
,
T.
Strunskus
,
F.
Faupel
, and
R.
Adelung
,
Sens. Actuators, B
224
,
434
(
2016
).
11.
V.
Cretu
,
V.
Postica
,
A. K.
Mishra
,
M.
Hoppe
,
I.
Tiginyanu
,
Y. K.
Mishra
,
L.
Chow
,
N. H.
de Leeuw
,
R.
Adelung
, and
O.
Lupan
,
J. Mater. Chem.
4
,
6527
(
2016
).
12.
A. K.
Mishra
,
A.
Roldan
, and
N. H.
de Leeuw
,
J. Phys. Chem. C
120
,
2198
(
2016
).
13.
Z. K.
Zheng
,
B. B.
Huang
,
Z. Y.
Wang
,
M.
Guo
,
X. Y.
Qin
,
X. Y.
Zhang
,
P.
Wang
, and
Y.
Dai
,
J. Phys. Chem. C
113
,
14448
(
2009
).
14.
B. Z.
Sun
,
W. K.
Chen
, and
Y. J.
Xu
,
J. Chem. Phys.
133
,
154502
(
2010
).
15.
C.
Li
,
F.
Wang
,
S. F.
Li
,
Q.
Sun
, and
Y.
Jia
,
Phys. Lett. A
374
,
2994
(
2010
).
16.
B. Z.
Sun
,
W. K.
Chen
,
X.
Wang
,
Y.
Li
, and
C. H.
Lu
,
Chin. J. Inorg. Chem.
24
,
340
(
2008
).
17.
W.-K.
Chen
,
B.-Z.
Sun
,
X.
Wang
, and
C.-H.
Lu
,
J. Theor. Comput. Chem.
7
,
263
(
2008
).
18.
H.
Wu
,
N.
Zhang
,
Z.
Cao
,
H.
Wang
, and
S.
Hong
,
Int. J. Quantum Chem.
112
,
2532
(
2012
).
19.
H.
Wu
,
N.
Zhang
,
H.
Wang
, and
S.
Hong
,
Chem. Phys. Lett.
568-569
,
84
(
2013
).
20.
Y.
Tan
,
X.
Xue
,
Q.
Peng
,
H.
Zhao
,
T.
Wang
, and
Y.
Li
,
Nano Lett.
7
,
3723
(
2007
).
21.
M.
Pang
and
H. C.
Zeng
,
Langmuir
26
,
5963
(
2010
).
22.
Y.
Sui
,
W.
Fu
,
H.
Yang
,
Y.
Zeng
,
Y.
Zhang
,
Q.
Zhao
,
Y.
Li
,
X.
Zhou
,
Y.
Leng
,
M.
Li
, and
G.
Zou
,
Cryst. Growth Des.
10
,
99
(
2009
).
23.
X.
Zhao
,
Z.
Bao
,
C.
Sun
, and
D.
Xue
,
J. Cryst. Growth
311
,
711
(
2009
).
24.
S.
Sun
,
F.
Zhou
,
L.
Wang
,
X.
Song
, and
Z.
Yang
,
Cryst. Growth Des.
10
,
541
(
2009
).
25.
S.
Sun
,
C.
Kong
,
S.
Yang
,
L.
Wang
,
X.
Song
,
B.
Ding
, and
Z.
Yang
,
CrystEngComm
13
,
2217
(
2011
).
26.
S.
Sun
,
X.
Song
,
Y.
Sun
,
D.
Deng
, and
Z.
Yang
,
Catal. Sci. Technol.
2
,
925
(
2012
).
27.
E. L.
Uzunova
,
N.
Seriani
, and
H.
Mikosch
,
Phys. Chem. Phys.
17
,
11088
(
2015
).
28.
L. I.
Bendavid
and
E. A.
Carter
,
J. Phys. Chem. C
117
,
26048
(
2013
).
29.
L.
Wang
,
T.
Maxisch
, and
G.
Ceder
,
Phys. Rev. B
73
,
174112
(
2006
).
30.
L. Y.
Isseroff
and
E. A.
Carter
,
Phys. Rev. B
85
,
235142
(
2012
).
31.
M.
Nolan
and
S. D.
Elliott
,
Phys. Chem. Chem. Phys.
8
,
5350
(
2006
).
32.
D. O.
Scanlon
,
B. J.
Morgan
, and
G. W.
Watson
,
J. Chem. Phys.
131
,
124703
(
2009
).
33.
C. E.
Ekuma
,
V. I.
Anisimov
,
J.
Moreno
, and
M.
Jarrell
,
Eur. Phys. J. B
87
,
23
(
2014
).
34.
G.
Kresse
and
J.
Hafner
,
Phys. Rev. B
47
,
558
(
1993
).
35.
G.
Kresse
and
J.
Hafner
,
Phys. Rev. B
49
,
14251
(
1994
).
36.
G.
Kresse
and
J.
Furthmüller
,
Comput. Mater. Sci.
6
,
15
(
1996
).
37.
G.
Kresse
and
J.
Furthmüller
,
Phys. Rev. B
54
,
11169
(
1996
).
38.
S. L.
Dudarev
,
G. A.
Botton
,
S. Y.
Savrasov
,
C. J.
Humphreys
, and
A. P.
Sutton
,
Phys. Rev. B
57
,
1505
(
1998
).
39.
J. P.
Perdew
,
K.
Burke
, and
M.
Ernzerhof
,
Phys. Rev. Lett.
77
,
3865
(
1996
).
40.
J. P.
Perdew
,
K.
Burke
, and
M.
Ernzerhof
,
Phys. Rev. Lett.
78
,
1396
(
1997
).
41.
G. W.
Watson
,
E. T.
Kelsey
,
N. H.
de Leeuw
,
J. D.
Harris
, and
S. C.
Parker
,
J. Chem. Soc., Faraday Trans.
92
,
433
(
1996
).
42.
P. W.
Tasker
,
J. Phys. C: Solid State Phys.
12
,
4977
(
1979
).
43.
H. J.
Monkhorst
and
J. D.
Pack
,
Phys. Rev. B
13
,
5188
(
1976
).
44.
45.
S.
Grimme
,
J. Comput. Chem.
27
,
1787
(
2006
).
46.
A.
Werner
and
H. D.
Hochheimer
,
Phys. Rev. B
25
,
5929
(
1982
).
47.
A.
Soon
,
M.
Todorova
,
B.
Delley
, and
C.
Stampfl
,
Phys. Rev. B
75
,
125420
(
2007
).
48.
J. A.
Assimos
and
D.
Trivich
,
Phys. Status Solidi A
26
,
477
(
1974
).
49.
B. Z.
Sun
,
W. K.
Chen
,
X.
Wang
, and
C. H.
Lu
,
Appl. Surf. Sci.
253
,
7501
(
2007
).
50.
H. J.
Freund
and
M. W.
Roberts
,
Surf. Sci. Rep.
25
,
225
(
1996
).

Supplementary Material