The adsorption and activation of carbon dioxide over copper cluster (Cu4) and copper doped on the alumina support (Cu4/Al2O3) catalytic systems have been investigated by using density functional theory and climbing image nudged elastic band. The adsorption energies, geometrical configurations, and electronic properties are analysed. The results show the strong chemical interaction between the copper cluster and the alumina support. Both the Cu4 cluster and Cu4/Al2O3 systems have a high adsorption ability for CO2, and the adsorption process is of chemical nature. The role of the alumina support in the adsorption and activation of CO2 has been addressed. The calculated results show that the “synergistic effect” between Al2O3 and Cu4 is the key factor in the activation of CO2.

1. Introduction

The greenhouse effect causing global climate change is now a special concern for all scientists. Capture and utilization of carbon dioxide (CO2) through its conversion into useful products such as methane, methanol, and hydrocarbon not only benefit economically but also reduce CO2 emissions into the environment [13]. Because CO2 is thermodynamically very stable, the C–O bond dissociation of CO2 is not easy and it always requires novel catalytic processes. In CO2 conversion, the adsorption and activation of CO2 over catalytic surfaces are crucial steps.

Density functional theory (DFT) today has become an effective tool for studying adsorption and activation of molecular substances on metal oxide surfaces. Therefore, adsorption and activation of CO2 have been extensively theoretically studied using the DFT method on many simple catalytic models such as flat transition metal surfaces [49] and oxide single crystals [1016]. For example, in the work of Liu et al. [6], adsorption and decomposition of CO2 on the face (100) of four metals (Fe, Co, Ni, and Cu) in the 3d group element were investigated to understand the intrinsic chemical properties of the 3d elements. Their calculations indicated that all four studied metals are capable of adsorbing and activating CO2. On the metal’s surfaces, CO2 chemisorption occurs spontaneously and is thermodynamically favourable. Ding et al. [8] studied the CO2 adsorption on the Ni (110) surface, and they indicated that CO2 is weakly adsorbed on Ni (110). Studies on comparing catalytic activities on different faces of a metal surface are also of interest. For example, O’Shea et al. [9] studied the adsorption of CO2 on Co (100), Co (110), and Co (111) fcc surfaces. Their results showed that the (110) surface exposes the strongest interaction with CO2, and the activation process involves charge transfer from the metal surface to the CO2. Adsorption of CO2 on the surfaces of several cesium oxides (Cs2O, Cs2O2, and CsO2) have been studied by Tai et al. [17]. By analysing the Mulliken charges, the authors reported there is a decrease on the basicity when going from Cs2O > Cs2O2 > CsO2.

Transitional metals supported on oxides or porous materials have been shown to be very effective in CO2 conversion [1823]. Aluminium oxides (Al2O3) are a class of advantaged materials with widespread applications as adsorbents and catalysts [24, 25]. Particularly, γ-Al2O3 is one of the most employed supports for heterogeneous industrial catalysts. The assets of alumina include low cost, relatively high surface area, large porosity, and good chemical and thermal stability [2629]. Moreover, Al2O3 also possesses a special surface property, and the simultaneous presence of acid and base centers will certainly affect the ability of the catalytic system to interact with CO2 [27, 28]. Although many works have been devoted to the study of interaction between CO2 with different oxides, theoretical studies of using γ-Al2O3 as a substrate were not common due to the complexity in its structure [3033]. Adsorption and protonation of CO2 on the (100) and (110) partially hydroxylated surfaces of γ-Al2O3 have been investigated by Pan et al. [10] using the DFT slab calculations. Their calculated results indicated that, on the partially hydroxylated γ-Al2O3 (100) surface, the formation of the bicarbonate species through the reaction between CO2 and the neighbouring surface hydroxyl is both thermodynamically and kinetically more favourable than that on the partially hydroxylated γ-Al2O3 (110) surface. Casarin et al. [30] studied CO2 adsorption on α-Al2O3 using density functional molecular cluster calculations. They found that CO2 interacts with the α-Al2O3 (0001) surface to form a bidentate-chelating carbonate species.

Recently, noble metal clusters have attracted much attention in scientific and technological fields because of their thermodynamic, electronic, optical, and catalytic properties in nanomaterials [3438]. Among them, size-selected subnanometer Cu clusters have received considerable attention and copper-based catalysts are widely investigated for hydrocarbon productions from CO2 due to their ability to catalyse CO2 related reactions [3942]. Especially, highly dispersed Cu cluster on γ-Al2O3 was reported to exhibit high catalytic activity and selectivity for methanol synthesis from CO2 [4348]. Therefore, many efforts have been devoted to discover the reaction mechanism and the nature of the active sites in Cu catalysts, including cluster size, oxidation state, and support effects [4042, 4850]. It is known that Cu nanoparticles show a better performance for CO2 electroreduction than bulk Cu. However, the roles of the size and symmetry of the Cu clusters as well as the temperature in the CO2‐reduction process remain elusive, which hinders the development of advanced catalysts. In the work of Zhao et al. [41], the density functional theory (DFT) method is applied to investigate these factors. They found that the decrease in icosahedron Cu cluster size but the increase in truncated octahedron Cu clusters’ size contributes to the selectivity of CO2 reduction. However, it was found that, for oxide-supported metal catalysts, the metal/support interface plays an important role in the catalytic activity of the metal, and reaction mechanisms on a supported metal cluster can be quite different from the mechanisms on pure metal catalysts. For example, Padama et al. [42] studied the interaction of CO, O, and CO2 on the Cu (111) surface and Cu3 cluster/Cu (111) surface. Their obtained calculations revealed that the Cu atoms in the cluster are more reactive towards adsorption than the Cu atoms of the flat surface. Therefore, the performance of transition metals depends not only on the chemical properties of the metals themselves but also on other factors such as the size of the cluster and the role of the support, and it is necessary to study the characterization of oxide-supported transition metal clusters [47, 5153]. However, the size effects in catalysis have been well addressed for Cu nanoparticle studies on the effects of the support and of the cluster’s size for supported ultrasmall Cu clusters with a few atoms are still rare [42, 54, 55].

Liu et al. [48] who studied on a subnanometer Cu4 catalyst have revealed that ultrasmall copper clusters can exhibit extraordinary catalytic activity for methanol synthesis at near atmospheric pressure compared with other, larger size catalysts. Yang et al. [55] studied the catalytic performances of Cun (n = 3, 4, 20) supported on Al2O3 for CO2 hydrogenation into CH3OH both theoretically and experimentally. The authors found that the activity for the methanol product strongly depends on the size of cluster and on the charge transfer interaction between clusters and the support. They observed a strong size dependence in the activity of Cun clusters with similar oxidation states at elevated temperatures, and their results indicated that the Cu4/Al2O3 possesses the highest activity for CH3OH formation. However, their work does not pay attention to the effect of the cluster structure when interacting with the support.

In this paper, we report a study of CO2 adsorption and activation on the Cu4 cluster and Cu4 cluster placed on Al2O3 using DFT-based calculations. The reason for choosing the Cu4 cluster is that, based on the experimental results reported by Yang et al. [55], DFT simulation will provide insights into the geometrical structure of the Cu cluster and the effects of the support of the Cu cluster for CO2 adsorption and activation. In order to better understand the role of Al2O3 in CO2 conversion and find out the key factors affecting the activity and selectivity of catalysts, we perform the similar calculations of CO2 interaction with Al2O3. The adsorption energies, geometrical configurations, and analysis of the electronic properties are presented.

2. Computational Details

According to Yang et al. [55], among the Cu3, Cu4, and Cu20 supported on Al2O3 catalysts, the Cu4/Al2O3 system shows the highest turnover rate for methanol production. Therefore, in this study, cluster of Cu4 was selected to investigate the adsorption behaviors. Two configurations of Cu4—tetrahedral and rhombus—are investigated to find out the most stable structure. Then, the selected stable cluster will be placed on the alumina carrier (Al2O3). The Al2O3 (104)-(3 × 3 × 1) was chosen based on the XRD experimental results by Yang et al. [56].

All the geometry and energy calculations were performed using the density functional theory (DFT) approach. The generalized gradient approximation (GGA) with the Perdew, Burke, and Ernzerhof (PBE) nonlocal gradient-corrected functional was employed to estimate the exchange correlation energy [57]. The double zeta basis plus polarization (DZP) orbitals was used for valence electrons, while the core electrons were treated using the norm-conserving pseudopotentials (NCPs) in its fully nonlocal (Kleinman–Bylander) form [58]. The Coulomb potential was expanded in a plane-wave basis with an energy cut-off of 150 Ryd. Spin-polarized calculations have been performed for all systems including metals. For all calculations with alumina supports, the periodic boundary conditions were applied. The sizes of the simulation box were 20.9891 × 14.277 × 16.7703 Å. All equilibrium structures were obtained using the Quasi Newton algorithm with the convergence criteria is that the forces acting on the dynamic atoms are smaller than 0.05 eV/Å. The DFT calculations were performed using the SIESTA (Spanish Initiative for Electronic Simulations with Thousands of Atoms) code [59] due to its advantages in robust and accurate aspect.

The adsorption energy (Eads) was calculated using the following equation:

The adsorption energy is considered as thermodynamic criterion to predict the possibility of the process. If the changes in the geometrical parameters are significant, the analysis will be done to get more inside the nature of the interaction between adsorbate and substrate. Besides, the atomic partial charges, estimated by means of the Voronoi deformation density (VDD) method, were reported. A remarkable charge transfer between atoms may indicate the chemisorption occurs. The VDD method avoids the problems inherent to basis set-based schemes and provides meaningful charges that conform to chemical experience [60]. The Mayer bond orders [61] were used in bonding analysis to describe the chemical interaction (if any) between molecules. The Mayer quantities are close to the corresponding classical values and are less dependent on the basis set choice. The Mayer bond order between two atoms A and B is defined as follows:where Pα and Pβ are the density matrices for spin α and β [61].

The transition states were determined using a climbing image nudged elastic band (CI-NEB) method [62]. The total number of configurations that appeared in the reaction coordinates is seven. All forces acting on the dynamic atoms were <0.1 eV/Å. If the adsorption involves a transition state, the activation energy is considered as the difference between the energy of the transition state and the energy of the initial structure. The activation energy (if any) will be considered to estimate kinetics of the process.

3. Results and Discussion

3.1. Electronic Properties of Cu4 and Cu4/Al2O3 Systems

Copper cluster is considered as an active center, and thus, the stability and the electronic properties of cluster are important. To determine the most stable structure of Cu4 clusters, the binding energies (Eb = [4E (Cu) − E (Cu4)]/4) for the tetrahedral and rhombus structures (Figure 1) are calculated and presented in Table 1. The total spin polarization for the optimal structure (QupQdown, where Qup and Qdown are the number of electrons with α-spin and β-spin, respectively) is defined using collinear spin-polarized option in the SIESTA code.

Since the binding energy of the rhombus structure is higher than that of the tetrahedral one, the former structure is more stable than the latter. The total spin polarization for the tetrahedral structure is two, whereas for the rhombus form, it is zero. This indicates the presence of two unpaired electrons in the tetrahedral cluster, while the rhombus structure has no unpaired electron. It is noted that the isolate copper atom has one unpaired electron. The reduction in the number of unpaired electrons in the clusters compared to that in the copper atoms alone suggested that the formation of copper clusters involves the electron pairings. In other words, the chemical bonds between copper atoms in the cluster have metallic covalent characterizations as well. This suggestion is confirmed by evaluating the geometrical parameters and the population analysis. The minimal distances between copper atoms in the tetrahedral and the rhombus clusters are determined to be 2.341 and 2.355 Å, respectively, which are significantly smaller than twice of the atomic radii of Cu (1.35 Å) [63]. The calculated total bond order in the Mayer scale between atoms for the rhombus structure is determined to be 1.819 which is higher than that for the tetrahedral structure (1.288). This result suggests that the covalent bonds (contributed to the electron pairing) in the rhombus are stronger than that in the tetrahedral one. This suggestion is in a good agreement with the calculated binding energies and spin polarization for Cu4 clusters. Therefore, the rhombus structure is more stable than the tetrahedral one.

In the next step, to evaluate the interaction between the cluster and the support, the rhombus Cu4 cluster is placed on the Al2O3 surface and the Cu4/Al2O3 structure is optimized. The obtained results are quite interesting: in the most stable geometrical configuration, the initial rhombus Cu4 cluster changed into the tetrahedral structure, as illustrated in Figure 2.

The reason for changing the geometrical structure may lay on the lack of unpaired electrons in the rhombus structure; therefore, in order to be able to interact with the surface oxygen atoms of the alumina support, rhombus has tended to convert into a tetrahedral structure. The total spin polarization (QupQdown) calculated for the Cu4/Al2O3 structure equals to zero. This result confirms the above suggestion. The total bond order between copper atoms in Cu4/Al2O3 is 1.272, which closes to that in the parent tetrahedral cluster (1.288). A total bond order of 0.659 indicates a strong interaction between copper atoms and oxygen atoms of the support. And this interaction is attributed to the partly electron transfer from the metal clusters to the alumina support. Due to the charge transfer, the VDD charge on the Cu4 cluster increases from 0 to 1.045 in Cu4/Al2O3. Significant changes in the atomic charges of the oxygen atoms in the alumina which directly bound with the copper atoms are also observed and presented in Table 2.

Analysis of the composition of the molecular orbitals of the Cu4/Al2O3 structure reveals the positive overlap between 3d orbitals (AOs) of copper atoms and 2p orbitals of oxygen atoms, as demonstrated in Figure 3.

The expression of the HOMO-35 molecular orbital of the Cu4/Al2O3 structure is as follows:where the last terms correspond to the AOs of the other atoms.

The positive overlap mainly between p AOs of oxygen in Al2O3 and 3d AOs of copper atoms leads to the chemical adsorption of Cu4 on the alumina support. The calculated projected density of states (PDOSs) for 3d orbitals of copper atoms in Cu4 clusters as well as in the Cu4/Al2O3 structure is presented in Figure 4.

Figure 4 shows that there is an increase in the density of states which is higher than the Fermi level for 3d orbitals of Cu4 in the Cu4/Al2O3 system in comparison to that in the Cu4 clusters. This result indicates the electron transfer from the d orbitals of copper atoms to the oxygen atoms of the alumina support. This finding is in a good agreement with the total bond order (0.659) between copper atoms and oxygen atoms discussed above. Due to the formation of chemical bonds, the adsorption and activation ability of Cu4/Al2O3 for CO2 are expected to differ from the pristine Cu4 clusters. To clarify this hypothesis, in the next steps, we investigate the adsorption of CO2 on the Al2O3 support, on the Cu4 tetrahedral cluster and on the Cu4/Al2O3 systems.

3.2. Adsorption of CO2 on Al2O3 Support

Alumina support is considered as a fairly good adsorbent for CO2 [64]. Our calculation results have shown that the adsorption of CO2 on Al2O3 releases an amount of 79.31 kJ·mol−1. The adsorption configuration of CO2 on Al2O3 support is presented in Figure 5. When adsorbing on the alumina support, the angle OCO is slightly decreased from 180° to 177.89°. The C–O bond lengths are determined to be 1.168 and 1.192 Å, which are not significantly differ from the C–O bond in the gas phase (1.175 Å). The minimal distance from CO2 molecule to the Al2O3 surface is found to be 2.245 Å, which is much longer than the sum of the covalent radii of O (0.60 Å) and Al (1.25 Å) [59]. Therefore, the adsorption of CO2 over the alumina surface can be considered as physisorption.

3.3. Adsorption and Activation of CO2 on Cu4 Cluster and Cu4/Al2O3 Systems

When carbon dioxide is adsorbed on the Cu4 cluster or Cu4/Al2O3, there are two possibilities: (i) CO2 molecule is bound to one copper atom via O atom (d1-perpendicular orientation); (ii) CO2 molecule is simultaneously bound to two copper atoms via C and O atoms (d2-parallel orientation). There will be four configurations, therefore, for the CO2 adsorption on the Cu4 tetrahedral cluster and on the Cu4/Al2O3 (Figure 6). They are named as follows: for Cu4 cluster: d1a (perpendicular orientation) and d2b (parallel orientation); for Cu4/Al2O3, d1c (perpendicular orientation) and d2d (parallel orientation). The adsorption energies for d1 and d2 configurations are presented in Table 3.

From the calculated results, one can obtain the following: (i) the d2-configurations are thermodynamically more favourable compared to the d1-configuration due to the more negative adsorption energy; (ii) the CO2 molecule is more strongly adsorbed by the tetrahedral cluster than by the cluster supported on alumina. The lowest adsorption energy (−150.49 kJ·mol−1) is obtained for d2b configuration. However, to compare the adsorption ability for CO2 between Cu4 and Cu4/Al2O3 structures, it is necessary to include the kinetic factor. Thus, in the next step, the CI-NEB calculations have been performed to investigate the kinetics of the CO2 adsorption process over the Cu4 cluster and Cu4/Al2O3. The obtained results show that there is a continuous decrease in energy from the initial configuration over five intermediates to the final product on the reaction pathways. It means that the CO2 adsorption process does not involve a transition state on both the Cu4 cluster and the Cu4/Al2O3 as well. Figure 7 demonstrates the relative energies of the initial, final, and five configurations during the process of CO2 adsorption on the tetrahedral Cu4 cluster.

Because the adsorption process of CO2 is not governed by the kinetic factors, the adsorption energy (Eads) can be considered as a criterion to estimate the adsorption ability. Due to the lower Eads, the Cu4 tetrahedral cluster is predicted to adsorb CO2 more effectively than Cu4/Al2O3. To figure out the nature of the interaction during the adsorption process, the changes in geometrical parameters and partial VDD atomic charges on atoms for the d2b and d2d configurations are determined. The results are presented in Table 4.

For CO2 adsorbed configurations, the OCO angles are bent from 180 in the initial CO2 gas molecule to 134.82 and 157.33° in the tetrahedral Cu4 cluster and Cu4/Al2O3, respectively. The C–O bond lengths are significantly elongated. The values of OCO angles and the total charges on the carbon dioxide molecule indicate that CO2 is more strongly bound to the Cu4 cluster than to Cu4/Al2O3. In the d2b configuration, an amount of 0.349 e is transferred from the Cu4 cluster to CO2, while in the d2d configuration, CO2 receives only 0.026 e. The total atomic charge on Cu4 in d2d is determined to be 1.092, whereas the value on the Cu4 cluster in Cu4/Al2O3 is 1.045. This means the charge transfer from the Cu4 cluster to CO2 is more significant than from Cu4 doped on the alumina support. We further investigate the molecular orbital compositions of the CO2 gas molecule and CO2–Cu4 cluster (d2b) configurations. Figure 8 shows that electrons are transferred from the copper atoms to the LUMO-antibonding -MO of CO2. Consequently, an elongation of C–O bond lengths is observed (from 1.175 Å to 1.273 Å) due to the decrease in the bond orders of CO2. Obviously, copper plays an active site to weaken the C–O bonds.

When the Cu4 cluster is placed on Al2O3, electrons from the Cu4 cluster are strongly transferred to the support (the total charge of Cu4 on Al2O3 is +1.045 e), and thus, in the CO2–Cu4/Al2O3 system (in d2d configuration), CO2 receives less electrons from Cu4 compared to the pristine Cu4 cluster.

3.4. Type of Adsorption

The adsorption of CO2 on the Cu4 cluster as well as on the Cu4/Al2O3 system can be considered as chemisorption due to the relative negative adsorption energies and remarked charge transfers from the Cu4 cluster to the CO2 molecule. The CO2 molecule is activated via C and O atoms. The bond orders between the CO2 molecule and the catalytic system are determined for d2 configurations and presented in Table 5. We also calculated the vertical binding energy C–O (Ebind), which is characterized for the following step:where the subscript (ads) denoted the adsorbed species on the catalytic surface. To calculate Ebind, all atoms are fixed except oxygen atom cleaved.

Clearly, from Table 5, it can be seen that CO2 molecule is “rigid” over the Cu4 cluster due to the bond formation with the total bond order of 1.275. Meanwhile, for CO2–Cu4/Al2O3 system (d2d configuration), the CO2 molecule is hold by copper atoms more weakly. The total bond order between CO2 and copper atoms is determined to be 0.34. The difference in the CO2 adsorption ability of the Cu4 cluster and Cu4/Al2O3 systems will result in the difference of CO2 activation efficiency. Obviously, two C–O bonds in the adsorbed CO2 molecule are not equally activated on the catalytic systems. For the CO2 activation on Cu4 (d2b), the bond orders of two C–O bonds are determined to be 1.197 and 1.571, while for the CO2 activation on Cu4/Al2O3 (d2d), these values are 1.750 and 1.910. The sharp decrease in the C–O bond order of adsorbed CO2 in d2b configuration compared to that of adsorbed CO2 in d2d and isolated CO2 gas molecule is expected to result in the significant decrease in the binding energies. In other words, the vertical C–O binding energies for the CO2–Cu4 system (d2b) are predicted to be the lowest. However, it occurs in the opposite manner: Ebind of two oxygen atoms of CO2 in d2b configuration are calculated to be 669.56 and 696.31 kJ·mol−1 which are although lower than that for the isolated CO2 gas molecule, but they are higher than that of adsorbed CO2 in Cu4/Al2O3 (d2d) (648.11 and 632.58 kJ·mol−1). Thus, it could be said that, on the Cu4/Al2O3 system, the CO2 molecule is more strongly activated and C–O bonds are more easily broken. The obtained calculations for Ebind (C–O) do not conflict with the calculated results for bond orders discussed above. It is noted that the bond orders characterize for the covalent properties of a bond, while the calculated Ebind are the sum of all interactions between atoms including electrostatic forces. Because the distances between adsorbed CO2 molecule and adsorbents are close to each other, we suggest that the key factor in the electrostatic interaction will be the charge on atoms. The copper atoms in Cu4/Al2O3 have a higher positive charge than that of Cu atoms in the Cu4 cluster; therefore, the higher electrostatic forces between oxygen atoms of CO2 and copper atoms in Cu4/Al2O3 system may govern the effectiveness of the CO2 activation process and result in the decrease of Ebind.

The calculated results suggest that when CO2 is activated over Cu4/Al2O3 catalyst, Ebind of two C–O bonds are close (648.11; 632.58); thus, the products will contain long carbon chain, whereas, over the pristine Cu4 cluster, Ebind of two C–O are significantly different; therefore, if one C–O bond of CO2 is activated, the products will contain one carbon atom such as CH3OH and HCHO. If both C–O bonds are broken, the products with more carbon atoms will be dominant due to the insertion chain growth mechanism.

4. Conclusions

By using density functional theory (DFT) and CI-NEB methods, we have studied the adsorption and activation of carbon dioxide over copper cluster (Cu4) and copper cluster doped on the alumina support (Cu4/Al2O3) catalytic systems. Our results have showed that, for the Cu4 cluster, the rhombus structure is more stable than the tetrahedral one. When being placed on the alumina support, the rhombus structure changed to the tetrahedral due to the strong chemical interaction with the support. The calculated results have also indicated that both the Cu4 cluster and Cu4/Al2O3 systems possess a high CO2 adsorption ability. When adsorbing on the Cu4 cluster as well as on Cu4/Al2O3, the C–O bond orders significantly decreased due to the electron transfer from the copper atoms to the antibonding MO of the CO2 molecule. During the adsorption and activation of CO2, the alumina oxide plays not only role of a support but also as an electron acceptor. The later role results in the increasing electrostatic interaction between copper active centers and the negative charged oxygen atoms of CO2, and thus, the C–O bonds are activated. This “synergistic effect” between Al2O3 support and Cu4 cluster plays an essential role in the activation of CO2.

Data Availability

The coordinate files (word files) of all optimized structures used to support the findings of this study are included within the supplementary information files.

Conflicts of Interest

The authors declare that there are no conflicts of interest regarding the publication of this paper.


This research was funded by the NAFOSTED (grant number 104.06-2014.84).

Supplementary Materials

Coordinates of all optimized structures. (Supplementary Materials)