Density functional theory (DFT) studies of CO oxidation reaction on M₁₃ and Au₁₈M clusters (M = Au, Ag, Cu, Pt and Pd): the role of co-adsorbed CO molecule

Abstract

Using the icosahedra M₁₃ (M = Au, Ag, Cu, Pt, Pd) and hetero-atom doped Au₁₈M (M = Ag, Cu, Pt, Pd) clusters as model systems, we have systematically investigated the role of the co-adsorbed CO molecule played in the CO oxidation reaction on the basis of density functional theory (DFT) calculations. The results indicate that the co-adsorbed CO molecule at a triangular active site can induce the dissociation of the OCOO* intermediate via a tri-molecular reaction route. This mechanism is also validated on other larger single doped gold alloy clusters such as Au_nAg and Au_nCu (n = 32–34, 54). The underlying reason for promoting the oxidation effect of a co-adsorbed CO molecule is unraveled. It is found that the relatively weaker d–π* back bonding of CO on group 11 elements like Au, Ag and Cu may increase its electrophilic activity, which can facilitate the dissociation of nearby OCOO* intermediates. For the CO molecule that is bounded to the Pd and Pt atoms, it can also induce the dissociation of OCOO* intermediate, but shows weaker electrophilic activity. By explicitly considering the elementary reaction steps in a Kinetic Monte Carlo (KMC) simulation, we have shown that the tri-molecular reaction route is an alternative reaction channel of CO oxidation, which is competitive to the conventional bi-molecular route on a doped Au₁₈M cluster.

---

1. Introduction

CO oxidation is an important model reaction in industrial and environmental applications.^1,2 Since Haruta et al. reported that the ultra-small gold nanoparticles supported on the oxides of 3d transition metals of group VIII exhibit an exceptional catalytic effect toward CO oxidation at temperatures as low as −70 °C, which have been considered one of most promising low-temperature catalysts towards CO oxidation compared to the conventional Hopcalite catalysts (mixed oxides mainly composed of Mn and Cu).³

Numerous experimental investigations have shown that the activity of nanosized gold clusters is related to the size and shape,^4–20 oxide supports^21–23 and doping heteroatom, etc.^24–30 For instance, nanosized gold clusters with diameters within the scope of 0.5–3 nm show more reactivity than those of larger size.^31,32 Moreover, nano-gold catalysts on reducible oxide supports (e.g., TiO₂, CeO₂, MgO, Fe₂O₃) exhibit a higher activity for catalyzing CO reactions than on non-reducible oxide supports (e.g., SiO₂, Al₂O₃).^33–39 On the other hand, hetero atom doping can also improve the catalytic activities of gold nanoparticles.⁴⁰ For example, alloyed nanoparticles, including Au–Pd nanoalloys,⁴¹ Na-doped Au clusters,⁴² Cu-doped and Ag-doped Au nanoparticles,^43,44 all show relatively higher catalytic activities than the pure gold nanoparticles.

Thus far, the understanding of gold catalyzed CO oxidation is based on two well-known mechanisms, namely, the Eley–Rideal (ER) and Langmuir–Hinshelwood (LH) mechanisms.^45–53 For both mechanisms, the elementary reaction processes are composed of an initial reaction of one CO molecule and one O₂ molecule. Then, the scission of the O–O bond of the reaction intermediate such as OCOO* (in the LH mechanism) leads to the formation of one CO₂ molecule. We call these mechanisms conventional bi-molecular reaction mechanisms. In fact, the bi-molecular LH or ER mechanism has also been used to understand the CO oxidation on other noble metal catalyst such as Ag, Cu, Pd and Pt et al.

Recently, a novel tri-molecular LH mechanism, also named as a CO self-promoting oxidation effect, has been proposed. Liu et al.⁵⁴ found that the co-adsorbed CO molecule at a triangular Au₃ active site can act as an electrophilic agent to induce the dissociation of an OCOO* intermediate, which results in the spontaneous formation of two CO₂ molecules in one reaction step with fairly low energy barriers. The discovery of a promoting oxidation effect of co-adsorbed CO on Au clusters raised several interesting questions. For example, in CO oxidation over other noble metal nanoparticles, like Ag, Cu, Pt and Pd, will the co-adsorbed CO molecule at the active site show a similar promoting oxidation effect? Which factors control the activity of co-adsorbed CO molecule for promoting OCOO* intermediate dissociation?

In order to understand the role of the co-adsorbed CO molecule in CO oxidation reactions over different types of noble metal clusters, we have systematically investigated CO oxidation of two kinds of model catalysts: the icosahedra M₁₃ (M = Au, Ag, Cu, Pt, Pd) and the hetero-atom doped Au₁₈M (M = Ag, Cu, Pt, Pd) alloy clusters. According to the recently proposed tri-molecular reaction mechanism, the triangular Au₃ site can provide an active site for the adsorption of multi reactant molecules such as CO and O₂. During the proceeding of the reaction, the co-adsorbed CO molecule could electrophilically attack the formed OCOO* intermediate so as to induce the breaking of the O–O bond and hence lead to the formation of two CO₂ molecules in one reaction step. Herein, the icosahedra M₁₃ cluster contains twenty triangular M₃ faces, which provides an ideal model system to examine the role of co-adsorbed CO during oxidation reactions on different kinds of metallic nanoparticles. On the other hand, the Au₃ site on a pyramid-like Au₁₉ cluster was demonstrated to be a highly active site for CO oxidation.⁵⁴ By investigating the CO oxidation over the doped Au₂M site of the Au₁₈M cluster, we can obtain a deeper insight into the origin of promoting the oxidation effect of the CO molecule at different metal site.

2. Computational method and details

In this theoretical study, we employ the density-functional theory (DFT) calculations implemented in the DMol³ 6.1 package within a generalized-gradient approximation (GGA) in the Perdew–Burke–Ernzerhof (PBE) form.^55–57 The double numerical (DND) basis set and semicore pseudopotential are used to treat atomic orbitals and core electrons, respectively. The real-space global cutoff radius is set to be 4.0 angstrom. The convergence tolerances of the geometrical optimization are set to be 2.0 × 10⁻⁵ hartree for the energy change, 2.0 × 10⁻³ hartree Å⁻¹ for the gradient, and 3.0 × 10⁻³ Å for the maximum displacement, respectively. The reaction transition states are located by using a combination of linear synchronous transit (LST)/quadratic synchronous transit (QST) algorithms⁵⁸ with conjugated gradient optimization, which are all confirmed by frequency calculations. For the Au₁₃ and Pt₁₃ clusters, their atomic positions are fixed at the icosahedron structure during the study of the CO oxidation route. The Hirshfeld population analysis is used to determine the atomic charges.

3. Results and discussion

3.1 CO oxidation on M₁₃ (M = Au, Ag, Cu, Pt, Pd) clusters: effect of co-adsorbed CO molecule

In order to illustrate the role of co-adsorbed CO during the CO oxidation on different noble metallic clusters, two lines of reaction pathways, based bi-molecular LH and tri-molecular LH mechanisms, were investigated over the triangular M₃ site of different icosahedra M₁₃ clusters (cf. Fig. 1). The computed energy curves of CO oxidation over various M₁₃ clusters are displayed in Fig. 1.

Fig. 1 (a)–(f) Energy profiles of CO oxidation on M₁₃ (M = Au, Ag, Cu, Pt, Pd) clusters under bi- and tri-molecular LH mechanisms. The symbol ‘*’ refers to the atom or molecule or the intermediate being adsorbed on the M₁₃ cluster. The bond length is in units of Å. (f) Energy profiles of the removal of an O* atom by a CO molecule on various M₁₃ clusters.

The CO oxidation reaction in the framework of a conventional bimolecular LH mechanism can be described by the four elementary steps:

CO(g) + * → CO* (1-B)

O₂(g) + * → O₂* (2-B)

CO* + O₂* → OCOO* + * (3-B)

OCOO* → CO₂ + O* (4-B)

In the present study, the energy curves of the four elementary reactions, steps (1-B)–(4-B), are examined. According to reactions (1-B) and (2-B), we can calculate the adsorption energies of CO and O₂ on M₁₃ clusters, which are defined as:

E_ad(CO) = E(M₁₃–CO) − E(M₁₃) − E(CO) (M = Au, Ag, Cu, Pt, Pd)

E_ad(O₂) = E(M₁₃–O₂) − E(M₁₃) − E(O₂) (M = Au, Ag, Cu, Pt, Pd)

From the elementary reaction steps (3-B) and (4-B), we can calculate the corresponding energy barriers of OCOO* formation and dissociation (i.e. E_a(TS1) and E_a(TS2)). The computed adsorption energies and activation energies on the different M₁₃ models are shown in Table S1 in the ESI.†

The elementary reaction steps of the trimolecular reaction mechanism are described by reaction steps (1-T)–(5-T):

CO₍₁₎(g) + * → CO₍₁₎* (1-T)

CO₍₂₎(g) + * → CO₍₂₎* (2-T)

O₂(g) + * → O₂* (3-T)

CO₍₁₎* + O₂* → OCOO* + * (4-T)

OCOO* + CO₍₂₎* → 2CO₂ + 2* (5-T)

Based on the trimolecular reaction mechanism, the first four elementary reaction steps (1-T)–(4-T) are the same as those of the bimolecular reaction mechanism discussed above. In the following step (reaction (5-T)), the co-adsorbed CO can electrophilically attack the OCOO* intermediate and hence induce the dissociation of the OCOO* intermediate into two CO₂ products in one reaction step.

Based on both the bi-molecular LH and tri-molecular LH mechanisms, the computed adsorption energies and activation energies on the different M₁₃ models are shown in Table S1.† (For Au₁₈M clusters, energy data are given in Table S2 in the ESI†).

For the CO oxidation over different noble metallic clusters, we find that the reaction barrier of both the OCOO* formation step (TS1) and the O–O bond breaking step (TS2) under the bi-molecular LH mechanism are closely related to the adsorption energy of CO. From Fig. 1, it can also be found that the initial reaction step of the electrophilic attack of the CO molecule on the nearby O₂ molecule on Pt₁₃ and Pd₁₃ exhibits a much higher energy barrier than that on Cu₁₃, Ag₁₃ and Au₁₃ clusters. The larger energy barrier of TS1 can be ascribed to the stronger binding interactions between CO and Pt₁₃ and Pd₁₃, which will be discussed in the follow section. However, we note that the O–O bond breaking barrier of the OCOO* intermediate on both Pt₁₃ and Pd₁₃ clusters is fairly small (0.002 eV and 0.06 eV, respectively), which is much smaller than those on Cu₁₃, Ag₁₃ and Au₁₃ clusters.

For the CO oxidation proceeding via the tri-molecular reaction route, the reaction barrier of the initial OCOO* formation step (TS1) is slightly affected by the co-adsorbed CO molecule. After the formation of the OCOO* intermediate, a tri-molecular transition state involving the nearby co-adsorbed CO molecule attacks the OCOO* intermediate is obtained on all M₁₃ clusters. From Fig. 1, the distances (r_C–O) between the C atom (in the co-adsorbed CO, denoted as C_CO) and the nearby O atom (in the OCOO* intermediate, denoted as O_OCOO*) range from 1.559 to 1.943 Å on five types icosahedra M₁₃. However, we find that even though the co-adsorbed CO molecule can induce O–O bond dissociation on various types of M₁₃ clusters, this process does not show obvious advantages of lowering the barrier height of OCOO* dissociation except on the Ag₁₃. From Fig. 1, the OCOO* dissociation barriers in the tri-molecular routes are 0.442, 0.590, 0.648 and 1.102 eV on Au₁₃, Cu₁₃, Pd₁₃ and Pt₁₃, respectively, which are larger than those of bi-molecular routes (0.311, 0.216, 0.002 and 0.06 eV, respectively). Even so, we note that the CO oxidation via a tri-molecular route can shorten the reaction steps. From Fig. 1, the tri-molecular route only involves two elementary reaction steps in the whole catalytic cycle, while the bi-molecular route needs an extra step of removal of the residue O* atom, which has energy barriers that range from 0.617 eV to 1.390 eV (Fig. 1f).

3.2 CO oxidation on doped Au₁₈M (M = Ag, Cu, Pt, Pd) clusters

The above discussions of bi- and tri-molecular reaction routes of CO oxidation on five icosahedra M₁₃ clusters indicate that the co-adsorbed CO does not show obvious advantages of lowering the barrier height of OCOO* dissociation except on the Ag₁₃ cluster. In fact, Liu et al. found that the protuberance degree of the Au₃ site can significantly affect the rate of the tri-molecular reaction route.⁵⁴ For example, the co-adsorbed CO on planar and flat-cage gold clusters like Au_7–10 and Au_12–14 does not show obvious advantages on lowering the barrier height of OCOO* dissociation. This is consistent with the current tendency on the M₁₃ cluster.

Recent theoretical studies show that the co-adsorbed CO molecule on the protuberant triangular Au₃ site of the Au₁₉ cluster can greatly accelerate the breaking of the OCOO* intermediate due to the large protuberance degree of the Au₃ site. The O–O scission caused by the attack of a nearby co-adsorbed CO has an energy barrier less than 0.1 eV.⁵⁴ At present, a hetero metal atom, M = Cu, Ag, Pd or Pt, is introduced at the triangular Au₃ site of the Au₁₉ cluster. By studying the CO oxidation over the Au₂M site, we try to understand the effect of co-adsorbed CO molecules in CO oxidation on the gold alloy cluster.

Six different reaction pathways of CO oxidation on the Au₂M triangular site of doped Au₁₈M (M = Cu, Ag, Pd or Pt) clusters were investigated, as shown in Scheme 1. For the bi-molecular reaction mechanism, we considered two possible initial adsorption modes of CO and O₂ on the triangular active site. In adsorption mode 1 the CO adsorbs on the Au site and the O₂ bridge links to Au and a hetero-atom site. Starting from this kind of adsorption configuration, two diverse reaction pathways (path a and path b) are induced. In adsorption mode 2, the CO adsorbs on a hetero-atom site and the O₂ bridge links to two Au atoms. In this kind of adsorption configuration, only one reaction pathway results (path c). When the co-adsorbed CO is taken into consideration, we also consider two possible initial adsorption configurations. Adsorption mode 3 has two CO molecules both linked to an Au atom and adsorption mode 4 includes two CO molecules that occupy Au and hetero-atom sites, respectively. Adsorption mode 4 it leads to two possible tri-molecular transition states which consist of an attacking CO molecule from either the hetero-atom site (path b′) or the Au site (path c′), respectively.

Scheme 1 Catalytic cycle of CO reactions on a triangular Au₂M site of Au₁₈M (M = Ag, Cu, Pt, Pd) cluster based on two oxidation mechanisms: (a) conventional bi-molecular LH mechanism and (b) tri-molecular LH mechanism involving attack of the nearby co-adsorbed CO.

Fig. S1 in the ESI† summarizes the calculated energy curves of CO oxidation starting from different reactant adsorption modes. In order to find out the most favorable reaction route under either the bi-molecular mechanism or the tri-molecular mechanism, a simplified kinetic treatment termed the “Sabatier analysis” is used to estimate the reaction rate. The Sabatier analysis provides a measure of the intrinsic ability of a metal surface to catalyze a given reaction, which was recently used to estimate reaction rates and shown to quantitatively reproduce the temperature dependent reactivity of Au nanoparticles and nanoporous Au.^59,60 The details of the formula and energy parameters for the computation of the reaction rates are given in the ESI.†

Fig. 2 displays the energy curves of the reaction routes with the highest reaction rates among either the bi-molecular or tri-molecular reaction mechanism. From Fig. 2, a general trend is found that these reaction routes adopt the same initial reactant adsorption configuration (mode 1 in the bi-molecular route and mode 3 in the tri-molecular route as shown in Scheme 1). That is, the CO molecule adsorbed on the Au site and the O₂ link to the hetero-atom leads to the highest reaction rate of CO oxidation on a doped Au₁₈M cluster.

Fig. 2 (a)–(d) Energy profiles of CO oxidation on Au₁₈M (M = Ag, Cu, Pt, Pd) clusters with the highest reaction rates. The bi-molecular and the tri-molecular reaction routes are marked by green and black lines, respectively. The symbol ‘*’ refers to the atom or molecule or the intermediate being adsorbed on the cluster. The energies (in eV) are computed at the PBE/DND level. The bond length is in units of Å.

For CO oxidation on the Au₁₈Ag nanoalloy, we find that the CO prefers to adsorb on the Au site and O₂ prefers to adsorb on the Ag site. The energy barriers of formation and dissociation of the OCOO* intermediate in the bi-molecular route are 0.247 eV and 0.251 eV, respectively. If the co-adsorbed CO molecule is considered, we find that the energy barrier in the step of the OCOO* formation reduces to 0.04 eV. Moreover, no transition state is located in the OCOO* dissociation step. The neighboring co-adsorbed CO molecule can induce the OCOO* dissociate into two CO₂ molecules spontaneously.

For the CO oxidation on the Au₁₈Cu cluster, the bi-molecular reaction route with the highest reaction rate has energy barriers of TS1 and TS2 of 0.211 and 0.148 eV, respectively. If a CO molecule is present at the nearby Au-corner, the formation of the OCOO* intermediate becomes barrierless. In addition, the energy barrier of the OCOO* dissociation decreases to 0.085 eV upon the attack of the co-adsorbed CO molecule.

On Au₁₈Pt and Au₁₈Pd clusters, the presence of the co-adsorbed CO at the Au-corner also lowers the energy barriers of both TS1 and TS2, similar to those observed on Au₁₈Cu and Au₁₈Ag alloy clusters. From Fig. 2, the energy barriers of the OCOO* formation step are 0.023 and 0.127 eV on the Au₁₈Pt and Au₁₈Pd clusters with the presence of co-adsorbed CO, which are much lower than those under the bi-molecular process (0.513 eV and 0.352 eV, respectively). On the other hand, we find that the energy barriers of direct OCOO* dissociation are 0.009 and 0.213 eV, respectively, on Au₁₈Pt and Au₁₈Pd clusters. Upon the CO attack, the dissociation barriers are 0.016 and 0.159 eV, respectively.

From the discussion above, it is seen that at the protuberant triangular Au₂M site, the presence of co-adsorbed CO can promote the whole catalytic oxidation process by lowering the energy barriers of both TS1 and TS2. Moreover, an interesting tendency is also found that the attack of a co-adsorbed CO at an Au site leads to a low dissociation energy barrier of OCOO* on all the doped Au₁₈M clusters (see Fig. S1 in the ESI†). For comparison, the attack of a co-adsorbed CO that is bound to a hetero-atom like Ag, Cu, Pt or Pd results in larger barrier of the OCOO* dissociation step. These results indicate that the reaction activity of co-adsorbed CO is significantly dependent on its binding site and the geometric structure of the triangular active site.

3.3 Correlation of CO oxidation rate and the electrostatic interactions

To understand the distinct reaction activity of co-adsorbed CO on different metallic nanoparticles, we also undertake a comparison of reaction rates between bi- and tri-molecular reaction pathways for CO oxidation on both icosahedra M₁₃ (M = Au, Ag, Cu, Pt, Pd) and heteroatom doped Au₁₈M (M = Ag, Cu, Pt, Pd) clusters (see Tables 1 and 2). From Table 1, Ag₁₃ has the maximum bi-molecular reaction rate of 2.273 × 10² s⁻¹ among all the M₁₃ (M = Au, Ag, Cu, Pt and Pd) clusters. Pt₁₃ and Pd₁₃ show an extremely inert catalytic activity due to the low reaction rates (8.548 × 10⁻²⁸ s⁻¹ and 5.067 × 10⁻¹⁴ s⁻¹, respectively). Moreover, in comparison to the bi-molecular reaction routes, we find that the participation of the co-adsorbed CO molecule in oxidation reactions does not show an obvious improvement in the reaction rate. From Table 1, the bi-molecular reaction routes on Au₁₃, Ag₁₃ and Cu₁₃ clusters have much higher reaction rates than those of the tri-molecular routes.

Table 1 Summary of the reaction rates of CO oxidation on different icosahedra M₁₃ (M = Cu, Ag, Au, Pd, and Pt) clusters under bi-molecular and tri-molecular reaction routes (s⁻¹). The details of the micro-kinetics analysis of the reaction rates are given in the ESI

Clusters	Reaction rate (s^−1)
	Bi-molecular route	Tri-molecular route
Au13	6.499 × 10^−8	3.467 × 10^−9
Ag13	2.273 × 10^2	5.646 × 10^−2
Cu13	2.387 × 10^−1	9.611 × 10^−5
Pt13	8.548 × 10^−28	1.193 × 10^−28
Pd13	5.067 × 10^−14	2.499 × 10^−14

---

Table 2 Summary of the reaction rates for CO oxidation on Au₁₈M (M = Cu, Ag, Au, Pd, and Pt) clusters under bi-molecular and tri-molecular reaction routes (s⁻¹). The reaction paths a–c and a′–c′ are described in Scheme 1, and correspond to different initial reaction adsorption configurations. The details of the micro-kinetics analysis of reaction rates are given in the ESI

Clusters	Reaction rate (s^−1)
	Bi-molecular route	Tri-molecular route
Au19	9.337 × 10^2	2.798 × 10^8
Au18Ag	6.477 × 10^5 (path a)	1.490 × 10^11 (path a′)
	2.922 × 10^6 (path b)	7.080 × 10^−3 (path b′)
	3.377 (path c)	1.007 × 10^7 (path c′)
Au18Cu	3.869 × 10^2 (path a)	1.255 × 10^9 (path a′)
	1.049 × 10^−3 (path b)	1.045 × 10^1 (path b′)
	2.396 × 10^−2 (path c)	1.642 × 10^3 (path c′)
Au18Pt	3.141 × 10^3 (path a)	1.185 × 10^8 (path a′)
	7.349 × 10^−5 (path b)	4.810 × 10^−50 (path b′)
	6.028 × 10^−48 (path c)	5.037 × 10^−42 (path c′)
Au18Pd	3.348 × 10^2 (path a)	1.827 × 10^2 (path a′)
	3.692 × 10^−3 (path b)	4.373 × 10^−26 (path b′)
	4.784 × 10^−29 (path c)	9.457 × 10^−24 (path c′)

---

In contrast to the icosahedra M₁₃ cluster, we find that the tri-molecular reaction route is favored over the bi-molecular route on doped Au₁₈M clusters. From Table 2, one may find that the maximum reaction rates are 1.490 × 10¹¹ s⁻¹, 1.255 × 10⁹ s⁻¹ and 1.185 × 10⁸ s⁻¹ on Au₁₈Ag, Au₁₈Cu and Au₁₈Pt, respectively, which are all higher than the maximum reaction rate from the bi-molecular reaction route (2.922 × 10⁶ s⁻¹, 3.869 × 10² s⁻¹ and 3.141 × 10³ s⁻¹). Here, we suggest that the protruding triangular site on Au₁₈M is a key to enhancing the tri-molecular reaction route, which results in shorter C_CO⋯O_OCOO* distances (generally less than 3.0 Å) that can facilitate the electrophilic attack of the C_CO atom to the nearby OCOO* unit.

In order to understand why the co-adsorbed CO molecule possesses a very different reaction activity towards inducing OCOO* intermediate dissociation on M₁₃ and Au₁₈M clusters, we have summarized the barrier heights of TS2 under tri-molecular reaction routes and their corresponded inter-molecular interaction styles in Table 3. Moreover, the local atomic charge of C_CO and O_OCOO* atoms in the co-adsorbed CO molecule and OCOO* intermediate are analyzed using the Hirshfeld population analysis. As discussed in previous studies,⁵⁴ the O–O bond scission (TS2 step) in the tri-molecular LH mechanism is induced by the electrophilic attack of the C_CO atom on the O_OCOO* atom. The energy barrier of this step is therefore strongly influenced by the electrostatic interactions between positively charged C_CO and negatively charged O_OCOO* units.

Table 3 Hirshfeld atomic charges of C_CO and O_OCOO* of OC*⋯OCOO* intermediate on the M₁₃ (M = Au, Ag, Cu, Pt, Pd), Au₁₉ and Au₁₈M (M = Ag, Cu, Pt, Pd) clusters. The symbol ‘*’ denotes that no stable OC*⋯OCOO* intermediate is located. The Hirshfeld charge analysis is performed with the PBE/DND level of theoretical calculation

	Clusters	Reaction pathways	CCO (e)	OOCOO* (e)	dC–O (Å)	Ea(TS2) (eV)
CO oxidation over M13 clusters
	Au13		0.112	−0.205	4.093	0.442
	Ag13		0.077	−0.275	4.468	0.160
	Cu13		0.063	−0.238	4.002	0.590
	Pt13		0.06	−0.157	3.899	1.102
	Pd13		0.001	−0.225	4.582	0.648
CO oxidation over Au18M clusters
	Au19		0.113	−0.211	2.401	0.050
	Au18Ag	Path a′	*
		Path b′	0.123	−0.220	2.683	0.231
		Path c′	0.109	−0.213	2.509	0.058
	Au18Cu	Path a′	0.118	−0.226	2.628	0.085
		Path b′	0.098	0.211	2.646	0.261
		Path c′	0.111	−0.206	2.456	0.059
	Au18Pt	Path a′	0.106	−0.147	4.942	0.016
		Path b′	0.076	−0.213	4.501	0.367
		Path c′	0.118	−0.187	2.926	0.312
	Au18Pd	Path a′	0.115	−0.221	2.923	0.159
		Path b′	0.047	−0.224	3.363	0.427
		Path c′	0.113	−0.219	2.886	0.252

---

At present, for the M₁₃ system, we find that the more positively charged C_CO atom and the more negatively charged O_OCOO* atom generally lead to lower barrier heights of OCOO* dissociation. For example, the CO*⋯OCOO* intermediates on Ag₁₃ and Pd₁₃ show a similar distance between C_CO and O_OCOO* atoms (4.468 and 4.582 Å, respectively). However, the O–O scission energy barrier caused by the electrophilic attack of C_CO on a OCOO* intermediate on an Ag₁₃ cluster is decreased by 0.40 eV in comparison to that of the bi-molecular route, while the energy barrier of O–O scission caused by the attack of a neighboring CO molecule on the Pd₁₃ cluster is raised by 0.588 eV. The reason for the very different effect of the attack of the co-adsorbed CO molecule can be attributed to the more positively charged C_CO atom and more negatively charged O_OCOO* atom on Ag₁₃ in comparison to those on Pd₁₃. Likewise, a similar tendency is also found on Pt₁₃. The attack of the co-adsorbed CO on a Pt₁₃ cluster leads to a significant increase of the OCOO* dissociation barrier (∼1.10 eV) due to the less positive charge on the C_CO and a negative charge on O_OCOO* atoms. It was well known that the Pd and Pt catalyst is readily poisoned by the CO molecule due to strong metal-CO bonding. Herein, the weaker reaction activity of CO molecules adsorbed on Pd₁₃ and Pt₁₃ correlates with their larger bonding energies.

In the case of CO oxidation over the hetero-atom doped Au₁₈M cluster, an interesting tendency is found that the attack of a co-adsorbed CO on an Au site leads to a lower dissociation energy barrier of OCOO* among the three kinds of reaction pathways (paths a′–c′). By examining the local atomic charge of C_CO and O_OCOO* atoms, one may again find that the barrier height of the TS2 step is strongly correlated with the electrostatic interactions between C_CO and O_OCOO* atoms. Table 3 illustrates three different CO attacking modes on each Au₁₈M cluster, which have distinct barrier heights. For Au₁₈Cu, Au₁₈Pd and Au₁₈Pt clusters, the attack of the CO from a hetero-atom site leads to the highest barrier height of OCOO* dissociation. Correspondingly, we find that the C_CO atom adsorbed on the hetero-atom site indeed carries less positive charge. Taken the above analyses together, we conclude that the observed very different promoting oxidation effect of co-adsorbing CO molecules can be partially attributed to the electrostatic interactions.

The chemical adsorption of CO onto a noble metal cluster is often described in terms of the Blyholder model.⁶¹ According to this model, the on-top site binding of the CO molecule to the noble metal atom consists of a σ forward donation from the 5σ lone pair of CO into an orbital and the occupied metal d orbitals donate electrons back to the anti-bonding CO–2π*. In the present studies, the Au, Ag and Cu have a relatively larger s–d valence level splitting than Pd and Pt, which renders the bonding of CO to Au, Ag and Cu less favorable due to the weaker d–π* back bonding. This qualitatively bonding analysis is in good agreement with our DFT results that the adsorption energies of CO on Pd and Pt are much larger than on Au, Ag and Cu (cf. Fig. 1 and 2). Because of the decreased d–π* back bonding of CO on Au, Ag and Cu, the relatively stronger σ forward donation may lead to more net electron transfer from CO binding to the metal atom. As a result, one may find that the C_CO atom adsorbed on Au, Ag and Cu sites possesses a more positive charge, which eventually results in a higher electrophilic activity of the CO molecule upon attacking the OCOO* intermediate.

The partial density of states (PDOS) was further analyzed for the Au₁₈M and Au₁₈M–CO systems (M = Au, Ag, Cu, Pt, Pd). From the PDOS in Fig. S2,† it can be found that the electron back-donation from the 5d-orbitals of Pt and Pd to the adsorbed CO molecule is much stronger than that from the Au, Ag and Cu atoms. This is in agreement with the Blyholder model and explains the intrinsic difference of electrophilic activity of the absorbed CO molecules.

We also calculate the reaction pathways of CO oxidation on the doped Au_nAg and Au_nCu (n = 32–34, 54) clusters to investigate the heteroatom effect of Ag or Cu. With Fig. 3, we address two possible initial adsorption modes, which correspond to mode 3 and mode 4 (cf. Scheme 1). In accordance with expectations, the CO molecule binding on the Au site leads to a stronger electrostatic interaction between C_CO on the Au site and the OCOO* intermediate on the Ag or Cu site, which facilitates the CO molecule attacking the OCOO* intermediate (i.e. mode 3).

Fig. 3 (a)–(h) Energy diagrams of CO oxidation on Au_nAg and Au_nCu (n = 32–34, 54) doped clusters under the tri-molecular LH mechanism. The symbol ‘*’ refers to the atom or molecule or intermediate being adsorbed on the cluster models. The energies (in units of eV) are computed at the PBE/DND level. The bond length is in units of Å.

3.4 Competitive effect of bi-molecular and tri-molecular routes from the Kinetic Monte Carlo (KMC) simulations

Taking the above discussions together, we find that the role of co-adsorbed the CO molecule could not be ignored during investigating the reaction mechanism of CO oxidation over the noble metallic clusters, in particular for the hetero-atom doped Au clusters on which the tri-molecular reaction channels possess much higher reaction rates than bi-molecular ones.

However, in a real catalytic system, the adsorption of reactants such as CO and O₂ are competitive on different sites, which may strongly affect the reaction selectivity. It is therefore of significance to assemble all the pieces of information from DFT into a solid kinetic model to yield a more complete microscopic description on the origin of the activity and selectivity. To give a more realistic description of the bi- and tri-molecular reaction channels, we employed a KMC simulation based on the elementary steps obtained from DFT calculations. Here, the Au₁₉ and doped Au₁₈M (M = Ag and Cu) were chosen as the model catalytic systems.

Table 4 displays all elementary steps under the bi- and tri-molecular reaction routes considered in the KMC simulations. With the whole reaction network being determined, we are in the position to assess which reaction route dominates the CO₂ formation and what is the theoretical selectivity. We have explicitly considered 6 elementary reaction steps over Au₁₉ and 16 steps on Au₁₈Ag or Au₁₈Cu. The DFT calculated reaction barriers utilized have been tabulated. The detailed procedures of the KMC simulations are given in the ESI.†

Table 4 Forward (E_f) and reverse (E_r) activation energies (in units of eV) of 6 elementary steps in the KMC simulation of CO₂ formation from CO/O₂ over the Au₁₈M (M = Au, Ag, Cu) clusters. All the adsorption energies of CO and O₂ molecules are taken from the bi-molecular route. —* The reversion reaction of CO₂ formation step is not considered

Elementary step on Au19	Ef	Er
CO(g) + *Au ↔ CO*Au	0	0.64
O2(g) + *Au ↔ O2*Au	0	0.40
CO*Au + O2*Au ↔ OCOO*Au–Au	0.09	0.15
OCOO*Au–Au ↔ CO2(g) + O*Au–Au	0.26	2.33
CO*Au + O*Au–Au → CO2(g) + 2*Au	0	—
CO*Au + OCOO*Au–Au → 2CO2(g) + 3*Au	0.05	—

---

Elementary step on Au18M (M = Ag/Cu)	Ef (M = Ag/Cu)	Er (M = Ag/Cu)
CO(g) + *Au ↔ CO*Au	0/0	0.843/0.837
CO(g) + *M ↔ CO*M	0/0	0.627/1.158
O2(g) +*Au ↔ O2*Au	0/0	0.725/0.651
O2(g) +*M ↔ O2*M	0/0	0.755/0.795
CO*Au + O2*M ↔ OCOO*Au–M	0.040/0	0.424/0.349
OCOO*Au–M ↔ CO2(g) + O*Au–M	0.251/0.149	2.245/2.373
CO*Au + O*Au–M → CO2(g) + 3*	0.043/0.018	—*/—*
CO*Au + OCOO*Au–M → 2CO2(g) + 3*	0/0.085	—*/—*
CO*Au + O2*Au ↔ OCOO*Au–Au	0.025/0.03	—*/—*
OCOO*Au–Au ↔ CO2(g) + O*Au–Au	0.251/0.149	2.245/2.373
CO*M + O*Au–Au → CO2(g) + 3*	0.062/0.049	—*/—*
CO*M + OCOO*Au–Au → 2CO2(g) + 3*	0.231/0.261	—*/—*
CO*M + O2*Au ↔ OCOO*M–Au	0.116/0.137	0.176/0.129
OCOO*M–Au ↔ CO2(g) + O*Au–M	0.102/0.147	2.560/2.637
CO*Au + O*Au–M → CO2(g) + 3*	0.043/0.0175	—*/—*
CO*Au + OCOO*M–Au → 2CO2(g) + 3*	0.058/0.059	—*/—*

---

Our statistical results were obtained by averaging the output of 20 KMC runs, and each simulation involves 5 × 10⁴ simulation steps. It is found that the CO₂ formation produced via the bi-molecular and tri-molecular pathways has a ratio of about 1.0 on the Au₁₉ cluster, which means the tri-molecular reaction is very competitive with the bi-molecular process. For comparison, we find that when the hetero Cu and Ag atom is introduced, the tri-molecular events decrease to different extents. For CO oxidation on Au₁₈Cu and Au₁₈Ag, the bi- and tri-molecular reaction ratios increased to 1.4 and 5.6, respectively. We attribute the greatly increased ratio between the bi-molecular and tri-molecular events on Au₁₈Cu to the larger formation barrier of the OCOO* bridge link on Cu and Au sites [i.e. CO*_M + O₂*_Au ↔ OCOO*_M–Au in Table 4], as well as the larger dissociation barrier of OCOO* upon CO attack from the Cu site [i.e. CO*_M + OCOO*_Au–Au → 2CO₂(g) + 3* in Table 4]. The analysis of KMC runs indicates that the CO prefers to occupy the Cu site and hardly desorbs. As a result, most tri-molecular adsorption configurations observed in the KMC runs adopt an adsorption mode 4 (see Scheme 1). Owing to the larger formation barrier of OCOO* bridging the Au and Cu sites (0.137 eV), the formation of an OCOO* bridge linking two Au sites (pathway b′ in Scheme 1 with an energy barrier of 0.030 eV) prevails over the formation of an OCOO* bridge linking Au and Cu sites. At this moment, the OCOO* tends to dissociate directly due to the smaller energy barrier (0.149 eV). For comparison, the attack of CO from the Cu site needs to overcome a higher energy barrier (0.261 eV from Table 4). From these discussions, we find that the co-adsorbed CO on a hetero-atom site like Ag, Cu, Pt and Pd has a weaker activity to induce OCOO* dissociation than to adsorb on the Au site. The presence of the hetero-atom at the triangular active site will suppress the tri-molecular reactions of CO oxidation on a gold cluster.

4. Conclusion

In this study, we have classified the different reaction activities of the co-adsorbed CO molecule in a CO oxidation reaction over both noble metallic clusters and gold alloy clusters by comprehensively studying the bi-molecular and tri-molecular reaction routes of CO oxidation over the icosahedra M₁₃ (M = Au, Ag, Cu, Pt, Pd) and hetero-atom doped Au₁₈M (M = Ag, Cu, Pt, Pd) clusters. The results indicate that the reaction activity of the co-adsorbed CO is significantly affected by its binding site, due to the different bonding properties of CO to Au, Ag, Cu, Pd and Pt. For CO binding on main group 11 elements like Au, Ag and Cu, the relative weaker d–π* back bonding entails more positive charges on the C_CO atom, which leads to a higher electrophilic activity of the CO unit. In the case of the CO binding on Pd and Pt, the stronger d–π* back bonding induces more charge back donation from the metal site to the CO, which may weaken the reaction activity of CO upon attacking the OCOO* intermediate. Furthermore, our KMC simulations indicate that, although the participation of co-adsorbed CO in the oxidation reaction may significantly increase the reaction rate on doped Au clusters, the tri-molecular reactions are also affected by the geometric structure of the active site and the adsorption configuration of the reactants. On an ideal triangular active site of a gold cluster, the presence of hetero-atoms like Ag, Cu, Pd and Pt will suppress the tri-molecular reactions.

5. Conflict of interest

The authors declare no competing financial interests.

Acknowledgements

This work is supported by NSFC (21373176, 21422305), Scientific Research Fund of Hunan Provincial Education Department (13A100).

References

1. T. Engel and G. Ertl, J. Chem. Phys., 1978, 69, 1267–1281.
2. H. J. Freund, G. Meijer, M. Scheffler, R. Schlögl and M. Wolf, Angew. Chem., Int. Ed., 2011, 50, 10064–10094.
3. (a) M. Haruta, T. Kobayashi, H. Samo and N. Yamada, Chem. Lett., 1987, 2, 405–408; (b) M. Haruta, N. Yamada, T. Kobayashi and S. Iijima, J. Catal., 1989, 115, 301–309.
4. P. Pyykkö, Angew. Chem., Int. Ed., 2004, 43, 4412–4456.
5. M. Shekhar, J. Wang, W. S. Lee, W. D. Williams, S. M. Kim, E. A. Stach, J. T. Miller, W. N. Delgass and F. H. Ribeiro, J. Am. Chem. Soc., 2012, 134, 4700–4708.
6. X. Tong, L. Benz, P. Kemper, H. Metiu, M. T. Bowers and S. K. Buratto, J. Am. Chem. Soc., 2005, 127, 13516–13518.
7. S. L. Gonzalez-Cortes, T. C. Xiao, P. M. F. J. Costa, B. Fontal and M. L. H. Green, Appl. Catal., A, 2004, 270, 209–222.
8. M. C. Daniel and D. Astruc, Chem. Rev., 2004, 104, 293–346.
9. J. Oliver-Meseguer, J. R. Cabrero-Antonino, I. Dominguez, A. Leyva-Perez and A. Corma, Science, 2012, 338, 1452–1455.
10. A. Fukuoka and P. L. Dhepe, Chem. Rec., 2009, 9, 224–228.
11. M. Turner, V. B. Golvoko, O. P. H. Vaughan, P. Abdulkin, A. Berenguer-Murcia, M. S. Tikhov, B. F. G. Johnson and R. M. Lambert, Nature, 2008, 454, 981–983.
12. M. Haruta, Catal. Today, 1997, 36, 153–166.
13. M. Chen and D. W. Goodman, Acc. Chem. Res., 2006, 39, 739–746.
14. G. J. Hutchings, M. Burst and H. Schmidbaur, Chem. Soc. Rev., 2008, 37, 1759–1765.
15. C. D. Pina, E. Falletta, L. Prati and M. Rossi, Chem. Soc. Rev., 2008, 37, 2077–2095.
16. J. K. Nørskov, T. Bligaard, B. Hvolbaek, F. Abild-Pedersen, I. Chorkendorff and C. H. Christensen, Chem. Soc. Rev., 2008, 37, 2163–2171.
17. I. X. Green, W. J. Tang, M. Neurock and J. T. Yates, Science, 2011, 333, 736–739.
18. U. Landman, B. Yoon, C. Zhang, U. Heiz and M. Arenz, Top. Catal., 2007, 44, 145–158.
19. L. Li, Y. Gao, H. Li, Y. Zhao, Y. Pei, Z. Chen and X. C. Zeng, J. Am. Chem. Soc., 2013, 135, 19336–19346.
20. M. D. Kane, F. Sloan Roberts and S. L. Anderson, Int. J. Mass Spectrom., 2015, 377, 263–277.
21. F. Boccuzzi, A. Chiorino, M. Manzoli, D. Andreeva and T. Tabakova, J. Catal., 1999, 188, 176–185.
22. M. Comotti, W. C. Li, B. Spliethoff and F. Schuth, J. Am. Chem. Soc., 2006, 128, 917–924.
23. H. Fujikia, N. Yamauchia, T. Chijiiwaa, S. Araia, S. Tsubotab and M. Haruta, Catal. Today, 1997, 36, 115–123.
24. M. Daté and M. Haruta, J. Catal., 2001, 201, 221–224.
25. M. Daté, M. Okumura, S. Tsubota and M. Haruta, Angew. Chem., Int. Ed., 2004, 43, 2129–2132.
26. J. Saavedra, H. A. Doan, C. J. Pursell, L. C. Grabow and B. D. Chandler, Science, 2014, 345, 1599–1602.
27. Z. Duan and G. Henkelman, ACS Catal., 2015, 5, 1589–1595.
28. L. Li and X. C. Zeng, J. Am. Chem. Soc., 2014, 136, 15857–15860.
29. G. C. Bond and D. T. Thompson, Gold Bull., 2000, 33, 41–50.
30. N. Austin, J. Karl Johnson and G. Mpourmpakis, J. Phys. Chem. C, 2015, 119, 18196–18202.
31. A. Herzing, C. J. Kiely, A. F. Carley, P. Landon and G. Hutchings, Science, 2008, 321, 1331–1335.
32. M. Valden, X. Lai and D. W. Goodman, Science, 1998, 281, 1647–1650.
33. M. M. Schubert, S. Hackenberg, A. C. Veen, M. Muhler, V. Plzak and R. J. Behm, J. Catal., 2001, 197, 113–122.
34. S. H. Overbury, L. Ortiz-Soto, H. G. Zhu, B. Lee, M. D. Amiridis and S. Dai, Catal. Lett., 2004, 95, 99–106.
35. L. Delannoy, N. El Hassan, A. Musi, N. N. Le To, J.-M. Krafft and C. Louis, J. Phys. Chem. B, 2006, 110, 22471–22478.
36. M. S. Chen and D. W. Goodman, Science, 2004, 306, 252–255.
37. Q. Fu, H. Saltsburg and M. Flytzani-Stephanopoulos, Science, 2003, 301, 935–938.
38. B. Yoon, H. Häkkinen, U. Landman, A. S. Wörz, J. M. Antonietti, S. Abbet and U. Heiz, Science, 2005, 307, 403–407.
39. C. Y. Peng and H. B. Schlegel, Isr. J. Chem., 1993, 33, 449–454.
40. H. Zhang, T. Watanabe, M. Okumura, M. Haruta and N. Toshima, Nat. Mater., 2012, 11, 49–52.
41. J. S. Jirkovský, I. Panas, E. Ahlberg, M. Halasa, S. Romani and D. J. Schiffrin, J. Am. Chem. Soc., 2011, 133, 19432–19441.
42. L. M. Molina and B. Hammer, J. Catal., 2005, 233, 399–404.
43. S. W. T. Price, J. D. Speed, P. Kannan and A. E. Russell, J. Am. Chem. Soc., 2011, 133, 19448–19458.
44. W. Q. Ma and F. Y. Chen, Catal. Lett., 2013, 143, 84–92.
45. W. An, Y. Pei and X. C. Zeng, Nano Lett., 2008, 8, 195–202.
46. G. Chen, S. J. Li, Y. Su, V. Wang, H. Mizuseki and Y. Kawazoe, J. Phys. Chem. C, 2011, 115, 20168–20174.
47. S. Nigam and C. Majumder, ACS Nano, 2008, 2, 1422–1428.
48. Y. Gao, N. Shao, S. Bulusu and X. C. Zeng, J. Phys. Chem. C, 2008, 112, 8234–8238.
49. C. Zhang, B. Yoon and U. Landman, J. Am. Chem. Soc., 2007, 129, 2228–2229.
50. C. Harding, V. Habibpour, S. Kunz, A. N. S. Farnbacher, U. Heiz, B. Yoon and U. Landman, J. Am. Chem. Soc., 2009, 131, 538–548.
51. W. Liu, Y. F. Zhu and Q. Jiang, J. Phys. Chem. C, 2010, 114, 21094–21099.
52. D. Y. Tang and C. W. Hu, J. Phys. Chem. Lett., 2011, 2, 2972–2977.
53. D. H. Kim, K. Shin and M. L. Lee, J. Phys. Chem. C, 2011, 115, 24771–24777.
54. C. Liu, Y. Tan, S. Lin, H. Li, X. Wu, L. Li, Y. Pei and X. C. Zeng, J. Am. Chem. Soc., 2013, 135, 2583–2595.
55. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868.
56. B. Delley, J. Chem. Phys., 1990, 92, 508–517.
57. B. Delley, J. Chem. Phys., 2000, 113, 7756–7764.
58. T. A. Halgren and W. N. Lipscomb, Chem. Phys. Lett., 1977, 49, 225–232.
59. T. Jiang, D. J. Mowbray, S. Dobrin, H. Falsig, B. Hvolbæk, T. Bligaard and J. K. Nørskov, J. Phys. Chem. C, 2009, 113, 10548–10553.
60. P. Wang, X. Tang, J. Tang and Y. Pei, J. Phys. Chem. C, 2015, 119, 10345–10354.
61. G. Blyholder, J. Phys. Chem., 1964, 68, 2772–2777.

---

Footnote

† Electronic supplementary information (ESI) available: The energy diagrams of the CO oxidation over doped Au₁₈M clusters with different reaction pathways and the PDOS plots of heteroatom M in Au₁₈M clusters are supplied. The details of the micro-kinetics analysis and the KMC simulations are also given. See DOI: 10.1039/c6ra07566e

---