An alternative decomposition reaction pathway for CO oxidation at Au_5-x-yAg_xCu_y (2 ≤ x + y ≤ 4) nanoclusters

Abstract

Different reaction pathways of CO oxidation at the surface of trimetallic nanoclusters, Au_(5-x-y)Ag_xCu_y of varying compositions with 2 ≤ x + y ≤ 4 (x, y > 0) are investigated. The starting point of all pathways is the surface reaction between adsorbed CO and O₂ via the Langmuir-Hinshelwood mechanism. The conventional reaction pathway leads to the formation of intermediate Au_(5-x-y)Ag_xCu_y.O* followed by a reaction with CO to release CO₂. However, on these trimetallic clusters, the formation of the nanocluster-carbonate adduct is greatly facilitated, which is known to poison the catalytic surface because of its greater stability. Here, we propose a possible Eley-Rideal (ER) pathway leading to carbonate decomposition and catalytic surface regeneration. This ER-pathway involves interaction between nanocluster-carbonate adduct and CO in the gaseous state, forming Au_(5-x-y)Ag_xCu_y.O₂COCO*, which decomposes to release CO₂. The study suggests the best trimetallic clusters for CO oxidation are the ones where both reaction sites are Ag-sites.

Graphical abstract

Different reaction pathways of CO oxidation at the surface of trimetallic nanoclusters, Au_(5-x-y)Ag_xCu_y, are investigated. These include the conventional LH-LH pathway and a newer alternative LH-ER pathway. The proposed LH-ER pathway involves an interaction between nanocluster-carbonate adduct and CO(g), forming Au_(5-x-y)Ag_xCu_y.O₂COCO*, which decomposes to release CO₂.

1 Introduction

With the continuous need to upgrade environment remediation technologies, significant focus has been on developing gold-nanocluster-based catalysts over the past several decades.¹ CO is a toxic pollutant generated due to the incomplete oxidation of fossil fuels. The interest in gold catalysts for CO oxidation was generated following the report by Haruta et al., of exceptional reactivity of small nanosized gold catalysts (≈5 nm) dispersed on different transition metal oxides like α- Fe₂O₃ at temperatures as low as 200 K.² Due to promising catalytic activity of gold nanoparticles at low temperatures, the use of gold-based catalysts in catalytic convertors installed in fossil fuel-powered vehicles can also provide a solution to cold-start emission problem. Cold-start emission is a known problem with conventional Pt and Pd-based catalytic converters.³ The gold nanoparticles are catalytically active only when particle size lies in the range between 1-5 nm; the bigger-sized particles or bulk gold are nonreactive.^4,5,6 There are a large number of factors, such as charge effect of support, low coordination sites, strain, particle size, composition, and presence of quantum effect at the nano level that influence the catalytic activity of nanoparticles.^{7,8,9,10,11,12,13} As a result, a mechanistic investigation into catalytic activity of such systems requires a study from different perspectives.

A great contrast is observed in the catalytic properties of hydrogen chemisorbed silver clusters (Ag_nH) and pure silver clusters (Ag_n) for CO oxidation reaction, where it is found while the former is catalytically active, the latter is not.¹⁴ Addition of Na to Au₂₀ nanocluster (NaAu₂₀) results in two types of structures – tetrahedral and caged. Both of these structures are reported to have enhanced binding energies and substantially reduced activation energy barriers for CO oxidation.¹⁵ The studies on hydrogen-doped gold nanoclusters (Au_(n-1)H) found the adsorption properties of the clusters to be the same as that of undoped gold clusters (Au_n).^16,17 These studies indicate the profound effect of the composition of nanoparticles on structural and thereby on their catalytic properties. Darby et al., report the structural difference between pure copper and pure gold nanoclusters. Further, replacing just one Au atom with Cu in a gold nanocluster significantly changes its structure.¹⁸

Nanoalloys as catalysts are interesting materials to explore as their catalytic capabilities can be fine-tuned with changes in the composition and position of dopants. In the case of Ag doping in Au₁₃, an increase in co-adsorption energy and a decrease in the activation energy is reported due to enhanced electron density at shell atoms over the core.¹⁹ Zeng et al., in a systematic study of CO oxidation on Au₁₈M (M = Ag, Cu, Pt, Pd) nanoclusters investigated in addition to conventional bi-molecular CO oxidation pathway, a tri-molecular reaction pathway in which co-adsorbed CO facilitates the dissociation of OCOO* intermediate.²⁰ A similar but distinct alternative route for CO oxidation is also presented in the current investigation. Zeng et al., found doped clusters, especially with M = Ag, Cu, to be catalytically more active than parent gold nanocluster via a tri-molecular reaction mechanism. The nanocluster Au₁₈M with M = Ag, Cu is found to be more reactive via the tri-molecular route due to the weaker d-π* back bonding in agreement with the Blyholder model.²⁰ There are a number of studies that analyze the different aspects of catalytic activity of bi-/tri-metallic nanoclusters on the basis of local densities of states, electron localized function, and charge distribution to identify the catalytically most active composition of these nanoalloys.^{21,22,23,24,25,26} For example, dispersed nanoalloys of Au-Ag-Pd created by laser irradiation are investigated for catalytic activity, and these are found to be more effective catalysts for Heck coupling reactions as compared to monometallic or bimetallic nanoclusters.²³ A uniquely designed Cu-Ag-Au trimetallic nano bowls having a cavity of few nanometers are found to exhibit maximum efficiency of 98.6% for reducing 4-Nitrophenol to 4-Aminophenol, which are reported to be 14, 23, and 43-fold more effective than the monometallic Au, Cu, and Ag systems, respectively.^24. A comparative study of activation energy of rate-determining step of oxygen reduction reaction (ORR) reports trimetallic Pt-Co-Ni cluster as having higher catalytic performance over Pt containing bimetallic or pure Pt nanoclusters.²⁵ The slightly weaker adsorption of reactant/intermediates on the surface of the catalyst is correlated with the better catalytic activity of these trimetallic nanoclusters. Kang et al., demonstrated the free valence electron centralization strategy to be an effective route for preparing ultrastable silver nanoclusters.²⁶ The strategy involves stepwise preparation of nanoalloy of Ag with Cu and Au, which not only imparts higher thermal stability to these silver-based nanoclusters but is also shown to have improved catalytic performance. They reported a considerable reduction in reaction time from 5 h to 3 min upon employing trimetallic nanocluster Au₁Ag₁₆Cu₁₂ as the catalyst for a multicomponent coupling reaction at high temperatures.²⁶

In the current study, we investigate the effect of the composition of trimetallic Au-Ag-Cu nanocluster on activation energy barriers of CO oxidation reaction and also propose an alternative route for CO oxidation via decomposition of adsorbed carbonate species initiated by CO at the surface of these nanoclusters.

2 Computational details

All calculations are performed through Gaussian09 using density functional theory with PW91 exchange-correlation (xc) functional.²⁷ For gold, silver, and copper atoms, relativistic effective core potential (ECP) LANL2TZ(f) is used, while for small atoms C and O 6-311+G(d) basis set is used. The LANL2TZ(f) is a LANL2 relativistic ECP of Hay and Wadt with a triple-zeta basis set supplemented with f polarization functions for treating the atom’s valence shell.²⁸ The choice of ECP, basis set, and xc functional is justified and discussed by Saini et al.²⁹ The bare nanoclusters are optimized for both doublet and quartet spin multiplicity, and nanoclusters with doublet multiplicity are found to be more stable. The geometries for various intermediates are also optimized for doublet spin multiplicity. No symmetry constraints are imposed in all calculations, and these calculations are spin unrestricted (UDFT). The structures are optimized using the keyword integral=ultrafine. The frequency analysis is carried out to verify the optimized geometries as minima or transition states. The intrinsic reaction coordinate (IRC) calculations are carried out to validate the corresponding transition state for a given set of reactants and products. All energies reported here include zero-point energy (ZPE) corrections.

3 Results and Discussion

3.1 Trimetallic clusters

The monometallic, bimetallic, and trimetallic nanoclusters have been optimized for different spin multiplicities. The clusters with doublet spin multiplicity are found to be the most stable. The binding energy per atom (B.E./atom) for these clusters is reported in Table 1 and compared with binding energy for monometallic clusters Au₅, Ag₅, Cu_5, and bimetallic clusters Au₃Ag₂ and Au₃Cu₂. The binding energy is defined as the energy required to break the cluster into its constituent atoms. B.E. per atom is calculated using equation (A) given below. Here, E(Au_5-x-yAg_xCu_y) is the energy of a cluster while E(M), M=Au, Ag, Cu is the energy of a single atom of Au/Ag/Cu in an unbound state.

$${\text{B.E.}}/{\text{atom}}= \frac{B.E}{5} = \frac{{(5 - x - y)*E(Au) + x*E(Ag) + y*E(Cu) - E({\text{Au}}_{{{\text{5 - x}} - y}} {\text{Ag}}_{x} {\text{Cu}}_{y} )}}{5}\quad \quad \quad ({\text{A}})$$

Table 1 Binding energy per atom in eV for Au_(5-x-y) Ag_x Cu_y clusters and the comparison with monometallic clusters and bimetallic clusters Au₃Ag₂ and Au₃Cu₂ at the same level of theory.³⁰

As the silver atom is replaced by a copper atom in trimetallic cluster AuCuAg₃, it is noted that B.E. per atom increases. A similar trend is observed as a gold atom in the trimetallic cluster AuCuAg₃ replaces a silver atom. It can be interpreted from the table that, in general, replacement by Ag atom decreases B.E./atom while replacement by Cu increases B.E./atom. But in the case of trimetallic cluster Au₃AgCu, the replacement of the gold atom by copper atom does not bring much change in B.E./atom for the cluster. Contrarily, B.E. per atom for Au₂AgCu₂ is more than that for AuAgCu_3, albeit the increment is minor. This can be understood in terms of the strength of the bond between two atoms. The study of Table 2 shows that the strength of the Au-Cu bond is highest and least for the Ag-Ag bond. Through careful examination of the geometries of most stable trimetallic clusters (Figure 1), it is found that in Au₂AgCu₂, there are four Au-Cu bonds and one Ag-Cu bond, while in cluster AuAgCu₃, the number of Au-Cu bonds reduces to two and there are two Ag-Cu bonds. This count explains why despite the replacement of Au by Cu in Au₂AgCu₂, the B.E. per atom for Au₂AgCu₂ (1.82 eV) is slightly greater than that for AuAgCu₃ (1.81 eV). The count of different kinds of bonds in six trimetallic clusters is provided in the supplementary information (SI) Table S.1. The formation energy for these trimetallic clusters has also been calculated, and the results are detailed in SI (Table S.2). The formation energies for all six clusters are negative indicating the stability of these clusters.

Table 2 Binding energy per atom for diatomic clusters in eV.

Figure 1

Trimetallic nanoclusters of varying composition with important bond lengths (Å) and binding energy per atom (eV).

3.2 Reaction profile of CO oxidation on Au_(5-x-y)Ag_xCu_y clusters

Here, we discuss the reaction profile of CO oxidation at the surface of various trimetallic clusters with different compositions i.e., Au_(5-x-y)Ag_xCu_y where x, y varies between 1 to 3 with the restriction that 2 ≤ x + y ≤ 4. There are six such trimetallic clusters. CO oxidation catalysed by these clusters is a bimolecular surface reaction that can proceed either via the Langmuir-Hinshelwood mechanism (LH) or the Eley-Rideal mechanism (ER). In the present study, the initial point of all the reaction profiles is taken to be a cluster with CO and O₂ adsorbed (LH mechanism) on adjacent three coordinated sites in top adsorption geometry. This very initial geometry is selected as we aimed to study the effect of the nature of the adsorption site (Au/Ag/Cu) on a given reaction profile, and secondly, the molecules adsorbed at these sites are oriented preferably for the reaction to proceed. The main steps involved in the CO oxidation on different compositions of trimetallic nanocluster considered in the current work are as follows.

$$\left. \begin{gathered} {\text{Au}}_{{{\text{5 - x}} - y}} {\text{Ag}}_{x} {\text{Cu}}_{y} {\text{ + CO(g) + O}}_{2} {\text{(g)}} \to {\text{Au}}_{{{\text{5 - x}} - y}} {\text{Ag}}_{x} {\text{Cu}}_{y} {\text{.CO}}^{*} {\text{.O}}_{{2}}^{*} \mathop{\longrightarrow}\limits^{{{\text{LH}}}}{\text{Au}}_{{{\text{5 - x}} - y}} {\text{Ag}}_{x} {\text{Cu}}_{y} {\text{.O}}^{*} {\text{ + CO}}_{{2}} {\text{(g)}}\begin{array}{*{20}c} {} & {} \\ \end{array} - - - (i) \hfill \\ {\text{Au}}_{{{\text{5 - x}} - y}} {\text{Ag}}_{x} {\text{Cu}}_{y} {\text{.O}}^{*} {\text{ + CO}}_{{2}} {\text{(g)}} \to {\text{Au}}_{{{\text{5 - x}} - y}} {\text{Ag}}_{x} {\text{Cu}}_{y} {\text{.OOCO}}^{*} {\text{(Carbonate formation)}}\begin{array}{*{20}c} {} & {} \\ \end{array} - - - (ii) \hfill \\ {\text{Au}}_{{{\text{5 - x}} - y}} {\text{Ag}}_{x} {\text{Cu}}_{y} {\text{.O}}^{*} {\text{ + CO(g)}}\mathop{\longrightarrow}\limits^{{}}{\text{Au}}_{{{\text{5 - x}} - y}} {\text{Ag}}_{x} {\text{Cu}}_{y} {\text{.O}}^{*} {\text{.CO*}}\mathop{\longrightarrow}\limits^{{{\text{LH}}}}{\text{Au}}_{{{\text{5 - x}} - y}} {\text{Ag}}_{x} {\text{Cu}}_{y} {\text{ + CO}}_{{2}} {\text{(g)}}\begin{array}{*{20}c} {} & {} \\ \end{array} - - - (iii) \hfill \\ {\text{Au}}_{{{\text{5 - x}} - y}} {\text{Ag}}_{x} {\text{Cu}}_{y} {\text{.OOCO}}^{*} {\text{ + CO(g)}}\mathop{\longrightarrow}\limits^{{{\text{ER}}}}{\text{Au}}_{{{\text{5 - x}} - y}} {\text{Ag}}_{x} {\text{Cu}}_{y} {\text{ + 2CO}}_{{2}} {\text{(g)(Decomposition)}}\begin{array}{*{20}c} {} & {} \\ \end{array} - - - (iv) \hfill \\ \end{gathered} \right\}\quad \quad \quad \left( {\text{B}} \right)$$

The first step involves a surface reaction between adsorbed CO and O_2, which is an activated process (Figure 2). In figures depicting the reaction profile, the chemical formula in the text implies that the species, e.g., CO, O₂ and/or CO₂, are infinitely separated from each other and from the cluster. Also, for all the transition states reported here, the coordinates, distance matrix, and a list of angles and dihedral angles are provided at the end of SI. Table 3 indicates the activation of O-O bond is maximum when O₂ is adsorbed on a 3-coordinated Cu-site. Due to the least electronegativity of Cu among the three metals, the charge transfer is more from Cu to O₂, leading to observed longest/weaker O-O bond among all six trimetallic clusters, which can account for the smallest barrier obtained for the surface reaction between CO* and O₂* at the surface of Cu₃AgAu. On the other five clusters, O-O activation is nearly of the same order and is supported by O-O bond length and charge on O₂. The CO and O₂ molecule adsorption studies on these six trimetallic clusters (Table S.3) indicate that C-atom forms the weakest bond with Ag, followed by Au and Cu. This can be explained based on s-d splitting. In the case of Ag, d levels lie deeper compared to Au and Cu. As a result, the strength of the Ag-CO bond is weaker due to the lesser availability of Ag d-orbitals for π back donation.^31,32 Note that wherever the strength of interaction between the metal atom and C/O is discussed, it is supported by initial adsorption studies of CO and O₂ carried out on these trimetallic clusters.

Figure 2

Reaction pathway for oxidation of first CO molecule by LH mechanism on the surface of Au₃AgCu nanocluster. The probability of forming carbonate species by recapturing liberated CO₂ is high due to the small energy barrier. The energies are reported in eV. For reaction profiles on remaining nanoclusters, refer to supplementary information.

Table 3 Activation energy barrier involved for the surface reaction between adsorbed CO and O₂ on different trimetallic clusters and other supporting parameters.

Except for Au₃AgCu, the surface reaction is a multistep reaction on the surface of trimetallic clusters. The multistep reaction facilitates CO oxidation. Table 3 tabulates the activation energy barrier involved in this surface reaction on different clusters. In this step, the interaction between adsorbed CO and O₂ generates transiently CO₂* and O* surface species. O* is stabilized in µ₂ configuration, which facilities the desorption of CO₂ from the nanocluster. After the desorption of CO₂, two possibilities exist: (a) another CO molecule adsorbs on the cluster and interacts with O* surface species leading to the liberation of the second molecule of CO_2, and the surface of nanocluster becomes available for the next catalytic cycle (Figure 3, LH-LH Pathway) or (b) the desorbing CO₂ may be reabsorbed resulting in the formation of carbonate surface species that poisons the surface of nanocluster for the further oxidation reaction. The carbonate nanocluster adduct (IV in Figure 2) is highly stable relative to starting reactants (I in Figure 2 +CO+ O₂). This is a common observation for the reaction profile on all six trimetallic nanoclusters. In our current study, we have further explored the possibility of a decomposition reaction involving Structure IV and CO which has been discussed in section 3.2.2. This pathway is depicted as an LH-ER pathway as after the initial LH step (Figure 2), the reaction between structure IV and CO takes place via ER mechanism. To clearly state, the LH-LH pathway consists of steps (i) and (iii) of equation (B), while the LH-ER pathway consists of steps (i), (ii), and (iv) of equation (B). Section 3.2.1 below discusses the possibility (a), Figure 3 with reaction proceeding via conventional LH-LH pathway.

Figure 3

Reaction pathway for oxidation of the second molecule of CO by LH mechanism on the surface of Au₃AgCu nanocluster. A high barrier of 0.99 eV is involved. The reported energies are in eV.

3.2.1 Reaction of Au_(5-x-y)Ag_xCu_y.O* with CO

To explore the surface reaction between Au_(5-x-y)Ag_xCu_y.O* and CO via LH mechanism, the adsorption study of CO on both 3-coordinated sites of Au_(5-x-y)Ag_xCu_y.O* has been carried out, and the most stable adduct Au_(5-x-y)Ag_xCu_y.O*.CO* is selected for further reaction mechanism study. The reaction pathway for Au₃AgCu is presented in Figure 3. The reaction profiles are presented in the supplementary information for the rest of the nanoclusters. Table 4 lists the activation energy barrier involved in this surface reaction on different trimetallic clusters. The activation energy for the reaction is least when both the 3-coordinated sites involved in the reaction are Ag sites which is justified as interaction energy for CO and O on the Ag site is lowest among all three metal sites – Au, Ag, and Cu. The orientation of CO* with respect to O* also significantly affects the activation energy barrier. Table 4 shows that the larger is the C-M1-O bond angle greater is the energy barrier. It is the most for Au₃AgCu, where the bond angle is 174.2°, and the barrier is 0.99 eV, followed by Cu₂Au₂Ag with a bond angle of 170.3°. In both clusters, one of the 3-coordinated sites is the Au site. For other clusters, the orientation is better suited for the surface reaction, which is also reflected in lower values of activation energy.

Table 4 Activation energy barrier involved for surface reaction via LH mechanism between the second molecule of CO adsorbed on different trimetallic Au_(5-x-y)Ag_xCu_y.O* adducts and other supporting parameters.

For cluster Cu₃AgAu both reaction sites are Cu sites, and the angle C-M1-O is 153.8°. This angle is lower than the corresponding angle for Cu₂Au₂Ag, but the activation energy barrier for surface reaction on Cu₃AgAu surface is only 0.01 eV less than on Cu₂Au₂Ag. This is explained by the stronger bonding interaction between Cu and O, as evident from the shortest bond length. The stronger the M1-O and M2-O interactions, the more energy would be required for a surface reaction between adsorbed CO and O, where O makes a new bond with the carbon of CO and breaks away completely/partially from metal sites. Once this surface reaction completes, the next step is the desorption of the product, that is, CO₂, from the surface of the metal cluster, making it once again available for the next catalytic cycle. A significant activation energy barrier is involved in the desorption of CO₂ from the surface of Cu₃AgAu, which is expected as the interactions of C and O with Cu are stronger. An additional barrier of 0.35 eV is reported in this case (Figure S.16). Another point to be noted here is that Cu₃AgAu.CO₂^* is considerably more stable than Cu₃AgAu, similar to surface energetics of trimetallic clusters, Au₃AgCu (Figure 3) and Cu₂Ag₂Au (Figure S.24). For the remaining three trimetallic clusters, an alternative pathway exists in which CO₂ is desorbed immediately after completion of the surface reaction.

3.2.2 Decomposition of cluster-carbonate adduct

It is widely reported in the literature that the formation of carbonate on the surface of a nanocluster affects the catalytic activity of the CO oxidation reaction. In the present study, we explore the new aspect of the decomposition of the cluster-carbonate adduct in the presence of CO. The reaction between carbonate-nanocluster adduct and CO is an activated process. For the clusters, the activation energy required is less if the reaction site is (Ag, Ag) or (Au, Ag) [Refer Table 5 and 6].

Table 5 Energy barriers (eV) for various steps involved in the decomposition reaction pathway via Route1 involving dihedral angle rotation at the surface of different compositions of Au_(5-x-y)Ag_xCu_y nanoclusters.

Table 6 Energy barriers (eV) for various steps involved in the decomposition reaction pathway via Route2 at the surface of Au_(5-x-y)Ag_xCu_y nanoclusters of different compositions.

However, if reaction site is Cu-site, the activation energy required for the reaction between carbonate-nanocluster adduct and CO is considerably higher which is because of strong interaction between Cu and O. Further, once Au_(5-x-y)Ag_xCu_y.O₂COCO* is formed, the decomposition may proceed by two alternate routes – the one which involves dihedral angle rotation about C-O-C-O by nearly 180 degrees and second without any dihedral angle rotation. The activation energy required for decomposition is higher for the route without dihedral angle rotation (Refer to discussion in SI). Tables 5 and 6 present the activation energy barriers involved for Route1 (with dihedral angle rotation) and Route2 (without dihedral angle rotation) for all six trimetallic nanoclusters, respectively. The reaction profile via Route1 at the surface of Au₃AgCu is shown in Figure 4, while the reaction profile via Route2 on the same cluster is presented in Figure 5. The reaction profiles for the rest of the trimetallic nanoclusters are provided in supplementary information.

Figure 4

An alternative decomposition reaction pathway by ER mechanism involving dihedral angle rotation (Route1).

Figure 5

An alternative decomposition reaction pathway by ER mechanism without dihedral angle rotation (Route2).

Again, in the clusters where at least one of the reaction sites is Cu, the activation energy is greater for Route2 (without dihedral angle rotation) and highest when both the reaction sites are Cu sites. For Route1 where after the formation of Au_(5-x-y)Ag_xCu_y.O₂COCO*, the next step is dihedral angle rotation which is as well an activated process, but the barrier for rotation is nearly the same for all trimetallic clusters and is of the order of 0.24 eV. However, as with Route2, even for Route1, the sum of all activation barriers for the decomposition of carbonate on the surface of the trimetallic cluster with Cu-site as one of the reaction sites is higher. Also, when both the reaction sites are Cu-sites, the CO₂ molecule does not desorb from the surface. As per Sabatier principle, the best catalyst for a reaction is one that has optimal interaction with reactants and products that is neither too strong nor too weak. The clusters with Cu as reaction sites are not good catalysts for CO oxidation reaction as Cu interacts strongly with the reactants O₂ and CO as well with product CO₂.

Another important observation is the presence of two different kinds of intermediates (Figure 6) during the decomposition of Au_(5-x-y)Ag_xCu_y.O₂COCO* adduct on nanoclusters where one of the reaction sites is Cu-site. The decomposition of Au_(5-x-y)Ag_xCu_y.O₂COCO* adduct leads to the release of two molecules of CO₂. The release of the second CO₂ molecule upon decomposition of Au_(5-x-y)Ag_xCu_y.O₂COCO* adduct may either be directly from intermediate A, B or it may be that firstly intermediate A transforms to intermediate B followed by the release of CO₂. The release of CO₂ from intermediate B may or may not occur depending on the strength of interaction between the metal site and C. The study shows that if the intermediate involved is of type A (Cu₂Ag₂Au, Figure S.25 and Figure S.26), the release of CO₂ is energetically more favourable compared to if the intermediate is of type B (Cu₂Au₂Ag, Figure S.21, and S.22) is involved. For clusters Cu₃AgAu (Figure S.17 and Figure S.18) with reaction site as (Cu, Cu), because of stronger interactions between Cu and C, the transformation of intermediate A to B is facilitated, which increases the energetic cost for the release of CO₂ molecule and also for the regeneration of catalytically active sites. On the other hand, for trimetallic clusters with reaction sites (Au, Ag); (Ag, Ag), and (Cu, Ag) as one the reaction site is Ag site which has weaker interactions with C, the release of CO₂ takes place either directly from intermediate A or from a structure with bonding characteristics similar to intermediate A.

Figure 6

Two types of possible intermediates involved in the release of second molecule of CO₂ via Route1 or Route2.

The carbonate decomposition via ER mechanism through Route1 (with dihedral angle rotation) is most favourable on the surface of Au₃AgCu, Au₂Ag₂Cu, and Ag₃AuCu where at least one reaction site is Ag-site.

4 Conclusions

CO oxidation at the surface of different trimetallic clusters can occur through several pathways, albeit the starting point is the surface reaction between adsorbed CO and O₂ via the Langmuir-Hinshelwood mechanism. This surface reaction leads to the formation of Au_(5-x-y)Ag_xCu_y.O* in µ₂-configuration, and one molecule of CO₂ is released. From this point onwards, the study presents several possible ways for the release of subsequent CO₂ molecules. Broadly, it can be summarized that two main pathways exist for CO oxidation at the surface of trimetallic clusters: the LH-LH pathway and the LH-ER pathway. Both CO₂ molecules are released following surface reaction via the LH mechanism in the LH-LH pathway. In the LH-ER pathway, the initial step occurs via LH mechanism, followed by carbonate formation and further decomposition via ER pathway. ER-pathway involves an interaction between nanocluster-carbonate adduct and CO in the gaseous state leading to the formation of Au_(5-x-y)Ag_xCu_y.O₂COCO*, which decomposes to release two molecules of CO₂.

The ER mechanism may happen either via Route1 or Route2. Route1 is where an additional step, that is, dihedral angle rotation, is involved that considerably lowers the activation energy required for decomposition reaction in comparison to Route2. Table 7 details the total activation energy barrier involved for these different pathways. The study of Table 7 suggests the best trimetallic cluster for CO oxidation is the one where both the reaction sites are Ag-sites. The trimetallic clusters Au₂Ag₂Cu and Ag₃AuCu are better suited for CO oxidation. Initially, both CO₂ molecules may be generated at the surface of these trimetallic clusters via the LH-LH pathway. However, when the concentration of CO₂ is considerable, the chances of carbonate formation will be higher, and hence the possibility of the reaction proceeding via the LH-ER pathway will be more. It is to be noted that for clusters where even if, one of the reaction sites is Cu-site, the total energy barrier is higher for both the LH-LH pathway and LH-ER pathway making Cu₃AuAg, Cu₂Au₂Ag, and Cu₂AuAg₂ trimetallic clusters less suited for the reaction. This is mainly because of stronger interactions of the Cu-site with the reactants (CO, O₂) and product (CO₂).

Table 7 Activation Energy Barrier in eV for different steps in LH-LH and LH-ER pathways on the surface of different trimetallic clusters.

Also, our study on CO oxidation at the surface of the Au₅ cluster shows that the chances of the formation of carbonate are negligible here as the activation energy required for its formation is quite high, i.e., 0.97 eV.²⁹ Therefore, at the surface of Au₅ the reaction proceeds via the LH-LH pathway involving a total energy barrier of 0.64 eV (0.50 eV + 0.14 eV). To conclude, it is suggested that due to the ease of formation of carbonate species at the surface of trimetallic clusters, the oxidation of CO has equal chances of proceeding via the LH-LH and LH-ER pathway, as discussed here. A similar study on the surface of Au_xAg_y and Au_xCu_y bimetallic clusters has also been carried out.³⁰