Probing the Site-Specific Reactivity and Catalytic Activity of Ag n (n = 15-20) Silver Clusters.

Density functional theory calculations within the framework of generalized gradient approximation (GGA), meta-GGA, and local functionals were carried out to investigate the reactivity and catalytic activity of Ag n (n = 15-20) clusters. Our results reveal that all the Ag n clusters in this size range, except Ag20, adsorb O2 preferably in the bridged mode with enhanced binding energy as compared to the atop mode. The O2 binding energies range from 0.77 to 0.29 in the bridged mode and from 0.36 to 0.15 eV in the atop mode of O2 adsorption. The strong binding in the case of the bridged mode of O2 adsorption is also reflected in the increase in O-O bond distance. Natural bond orbital charge analysis and vibrational frequency calculations reveal that enhanced charge transfer occurs to the O2 molecule and there is significant red shift in the stretching frequency of O-O bond in the case of the bridged mode of O2 adsorption on the clusters, thereby confirming the above results. Moreover, the simulated CO oxidation reaction pathways show that the oxidation of the CO molecule is highly facile on Ag16 and Ag18 clusters involving small kinetic barriers and higher heats toward CO2 formation.

Introduction

Silver in the nano and sub-nano size regime has attracted a great deal of attention due to its wide range of catalytic and optical properties. This has led to many interesting experimental studies involving the synthesis and application of silver nanoclusters for catalyzing a wide range of reactions which include NO reduction,[1] reactions involving hydrocarbons,[2−21] and CO oxidation.[22−31] Despite the progress during the past decade to synthesize silver clusters supported on variable supports and their subsequent applications, an exact understanding of the reaction mechanism and underlying factors which govern the silver cluster reactivity is still lacking due to the complexity of heterogeneous catalysis.

Compared to the supported metal clusters, simple gas-phase clusters are considered ideal to understand the reaction mechanism and factors governing the catalytic properties of metal clusters. The stability of gas-phase clusters and the lack of direct experimental probes to understand their structure pose a great challenge in this direction. However, recent advances in experimental techniques aided by first-principles calculations have resulted in a number of studies to understand the structure and reactivity of silver nanoclusters with small molecules such as O2. Most of these studies were carried out on small anionic Ag (n = 1–10) clusters as their geometric and electronic structures are easily accessible through photoelectron spectroscopy. For example, a joint experimental and theoretical study[32] revealed a pronounced size and structure dependence of O2 binding on (n = 1–5) clusters. The O2 binding exhibited significant odd–even oscillations with the open-shell clusters showing significantly enhanced O2 adsorption in comparison to closed-shell counterparts. Moreover, the authors found that a pre-adsorbed oxygen molecule on the closed-shell and significantly enhances the binding and activation of the second O2 molecule. Socaciu et al.[33] carried out an identical type of investigation on the interaction of clusters (n = 1–11) with oxygen and carbon monoxide molecules. The cluster anions were found to readily react with O2 molecules and showed a parallel type of odd–even O2 binding pattern. Castleman and co-workers[34] reached similar conclusions while investigating the interaction of O2 with clusters (n = 1–17) using synergistic experimental and theoretical calculations. It was found that clusters with an even number of electrons exhibit markedly less reactivity than those with an odd number of electrons toward O2 molecules. Further, was identified as a highly stable cluster due to its large highest occupied molecular orbital–lowest unoccupied molecular orbital gap and large spin excitation energy. Ma et al.[35] using kinetic measurements on reactions of (n = 6–69) with O2 revealed that reactivity of anionic silver clusters with O2 in the nano size domain is still dominated by global electronic properties such as the spin state and electron binding strength of the cluster.

Compared to anionic silver clusters, the interaction of O2 molecules with neutral silver clusters is less explored due to the lack of direct experimental probes. In this context, machine learning and density functional theory (DFT) act as suitable tools to probe the structure and properties of such materials.[36−48] This has mostly resulted in theoretical calculations based on DFT to evaluate the O2 reactivity of neutral silver clusters. For example, DFT calculations by Klacar et al.[49] on the binding strength of molecular and dissociative adsorption of oxygen showed marked odd–even oscillations with the dissociative adsorption being more favorable than molecular adsorption on silver clusters with more than five atoms. Wang et al.[50] using the particle swarm and minimum hopping approach in combination with DFT calculations studied the dissociative chemisorption of O2 on Ag and AgIr (n = 3–26). The results pointed that O2 dissociation entails very high barriers for clusters with n = 11–26 for both pristine and doped counterparts. DFT calculations by Liao et al.[51] also revealed odd–even oscillation in the O2 adsorption energies on neutral silver clusters.

Although impressive progress has already been done in decoding the reactivity/catalytic behavior of small silver clusters, more effort needs to be put in to understand the catalytic activity of medium and large silver nanoclusters. Thus, in this work, we have carried out comprehensive DFT calculations on structurally well-characterized Ag (n = 15–20) clusters to investigate their site-dependent reactivity toward O2 molecules. Further, followed by O2 adsorption, we also simulated the CO oxidation reaction to understand the catalytic properties of these clusters.

Computational Details

All the calculations were accomplished by using DFT with Perdew–Burke–Eernzerhof (PBE), Tao–Perdew–Staroverov–Scuseria (TPSS), and M06-L functionals as implemented in the Gaussian-16[52] software. The geometry optimizations were carried out by using the Berny algorithm with the default convergence criterion. Vibrational frequency calculations were carried out to confirm the true ground-state nature of the optimized structures. To simulate the reactivity of Ag clusters, O2 adsorption was studied at different possible sites in the atop mode with one Ag–O bond and the bridged mode with two to three Ag–O bonds. Furthermore, O2 adsorption was investigated for both the singlet and triplet multiplicities for even electron clusters and doublet and quartet multiplicities for odd electron clusters. O2 preferentially adsorbs in doublet multiplicity on the odd electron clusters and triplet multiplicity on even electron clusters expect for Ag16 where singlet multiplicity is the favored spin state. The LANL2DZ basis set and the corresponding Los Alamos relativistic effective core potential were used for the silver atoms, and the TZVP basis set was used for carbon and oxygen atoms. TZVP types of basis sets have been widely used for interactions of small molecules with transition metals and clusters.[53−56]

The binding energy (Eb) of the O2 molecule was computed using the below formulawhere is the energy of the O2 molecule in the triplet state, is the energy of the isolated cluster, and is the energy of the O2-adsorbed Ag complex. The transition states for simulating CO oxidation pathways on Ag clusters were found by using the linear synchronous transit method and were characterized by the presence of one imaginary frequency.

Results and Discussion

We proceed with a discussion on the adsorption of the O2 molecule on the different sites of Ag (n = 15–20) clusters. The lowest-energy structures of Ag clusters are in the size range of n = 15–20 atoms as presented in Figure and are already well documented by previous studies.[34] To characterize O2 reactivity of these clusters, we first adsorbed O2 in atop and bridged modes at the different possible sites in these clusters as shown in Figures S1 and S2. The optimized geometries of the lowest-energy AgO2 complexes in both the bridged and atop modes with their relative energies in eV are displayed in the Figure . As can be seen, O2 preferentially undergoes the bridged type of bonding with all the silver clusters except Ag20, which shows the atop mode of bonding with O2. The energy difference between the bridged and atop modes is particularly pronounced for Ag16O2 and Ag17O2 clusters. In the case of Ag20O2, the atop mode of adsorption is stronger with no bridged adsorption involving both oxygen atoms bonded to Ag atoms. It is noteworthy to mention here that earlier reports by Liao et al.[51] contrastingly pointed the atop mode of O2 binding as the most favorable adsorption on Ag15–Ag18 clusters. To further confirm our results, we optimized the structures using TPSS and M06-L functionals. The results also revealed the bridged O2 binding on all Ag clusters except Ag20 rather than atop mode of binding.

Figure 1

Ground-state structures of Ag (n = 15–20) clusters obtained using the PBE functional.

Figure 2

Lowest-energy bridged and atop configurations of O2 adsorption on Ag clusters (n = 15–20), with their relative energies and relevant geometrical parameters computed at the PBE level of theory.

The interaction of molecular oxygen can be quantitatively understood in terms of the O2 binding energies on the Ag clusters. Tables and 2 enlist the computed O2 binding energies in bridged and atop modes, respectively, obtained by PBE, TPSS, and M06-L functionals with excellent agreement for all the clusters under consideration. As can be seen, all the silver clusters except Ag20 show higher binding energies in the bridged mode as compared to the atop mode. Further, the O2 binding energies decrease as we go from Ag15 to Ag19 in the bridged mode with Ag18 showing the lowest binding energy of 0.29 eV and Ag15 and Ag16 showing the highest binding energies of 0.77 and 0.66 eV, respectively, at the PBE level of theory. Moreover, the odd–even trends in the bridged mode of O2 binding energy are not quite visible in the studied cases, which is in sharp contrast to small silver clusters having less than 10 atoms. The atop mode of adsorption reveals significantly lower binding energies with strong odd–even oscillation. The enhanced binding of O2 adsorption is reflected in the significant increase in the O–O bond distances in the bridged mode as compared to the atop mode of O2 adsorption. The bridged Ag–O2 complexes reveal an average O–O bond distance of ∼1.30 Å, whereas atop Ag–O2 complexes reveal an average O–O bond distance of ∼1.26 Å. The above results are corroborated by natural bond orbital (NBO) charges on O2. As can be seen in Tables and 2, the bridged mode of adsorption leads to a significantly higher charge transfer of ∼0.40 e to O2 as compared to ∼0.25 e in the case of atop adsorption. Further, a significant amount of red shift corresponding to the O–O frequency in the bridged mode compared to the atop mode of the O2 adsorption also reflects the strong activation of the O–O bond.

Table 1

Binding Energies of O2 (Eb) and Optimized Geometrical Parameters Such as Ag–O Bond Length (rAg–O), O–O Bond Length (rO–O), and O–O Stretching Frequency (vO–O) and of the AgO2 in the Bridged Mode Computed Using the PBE Level of Theory

system	Eb(PBE) (eV)	Eb(TPSS) (eV)	Eb(M06-L) (eV)	rAg–O (Å)	rO–O (Å)	vO–O (cm–1)
Ag15O2	0.77	0.82	0.86	2.38/2.38	1.30	1134	–0.42
Ag16O2	0.66	0.89	0.87	2.37/2.38	1.30	1118	–0.43
Ag17O2	0.61	0.54	0.65	2.34/2.40	1.31	1106	–0.46
Ag18O2	0.29	0.25	0.26	2.38/2.43	1.30	1113	–0.41
Ag19O2	0.56	0.59	0.56	2.33/2.36	1.32	1110	–0.54
Ag20O2

Table 2

Binding Energies of O2 (Eb) and Optimized Geometrical Parameters Such as Ag–O Bond Length(rAg), O–O Bond Length (rO–O), and O–O Stretching Frequency (vO–O) and of the AgO2 in the Atop Mode Computed Using the PBE Level of Theory

system	Eb(PBE) (eV)	Eb(TPSS) (eV)	Eb(M06-L) (eV)	rAg–O (Å)	rO–O (Å)	vO–O (cm–1)
Ag15O2	0.36	0.44	0.31	2.35	1.27	1244	–0.28
Ag16O2	0.29	0.28	0.29	2.39	1.26	1281	–0.21
Ag17O2	0.30	0.24	0.27	2.39	1.26	1265	–0.22
Ag18O2	0.25	0.24	0.36	2.40	1.26	1285	–0.17
Ag19O2	0.48	0.42	0.39	2.34	1.28	1228	–0.29
Ag20O2	0.15	0.14	0.13	2.50	1.24	1358	–0.13

We next looked at the CO oxidation ability of Ag clusters using the Langmuir–Hinshelwood mechanism. In the Langmuir–Hinshelwood type of reaction mechanism, the O2 and CO molecules first co-adsorb on the adjacent sites of a cluster, leading to the formation of the O–O–C–O intermediate in the first step and release of the CO2 molecule in the second step. Figure displays the structures and relative energies of the involved reactants, intermediates, transition states, and final products on the Ag clusters. The lowest-energy configurations of O2 and CO co-adsorbed Ag clusters were obtained by adsorbing CO molecules on different possible sites next to the adsorbed O2 on the clusters. From Figure , we note that the first step leading to the formation of the O–O–C–O intermediate on Ag20 involves a very high activation barrier of 0.55 eV. The second step involving the cleavage of the O–O bond of the O–O–C–O intermediate and leading to the release of CO2 entails a barrier of 0.22 eV. However, with the activation of molecular oxygen and O–O–C–O intermediate formation being the rate-determining step for CO oxidation on metal clusters,[57−62] CO oxidation reaction on Ag20 is less feasible. Contrary to this, for Ag15 to Ag19, the first step leading to the formation of the O–O–C–O intermediate involves very low activation barriers of 0.11, 0.18, 0.15, 0.01, and 0.14 eV, respectively. The reaction barriers for the second step are 0.40, 0.31, 0.33, 0.32, and 0.51 eV, respectively. Interestingly, we note that the CO oxidation reactions on the Ag16 and Ag18 clusters involve the intermediates and transition states significantly lower in energy than the rest of the clusters. This makes the reaction more facile and thereby increases the probability of the CO oxidation on the Ag16 and Ag18 clusters. NBO charge analysis along the reaction pathway of Ag16, Ag18, and Ag20 was carried out to understand the stabilization of transition states along the CO oxidation reaction pathways on Ag16, Ag18, and Ag20. The net NBO charges on the O2 molecule are presented in Table . The calculated NBO charges analysis reveals that significant charge transfer occurs from the cluster to the O2 molecule in TSI and TS2 in the case of Ag16 and Ag18 clusters, leading to their stabilization and subsequent reduction in barrier heights. A similar trend is seen in the frontier molecular orbital analysis of TSI and TS2 (Figure S3), wherein we note significant overlap in the case of Ag16 and Ag18 compared to Ag20.

Figure 3

Simulated reaction pathways (schematic) for CO oxidation on the Ag clusters (n = 15–20) computed at the PBE level of theory.

Table 3

Calculated NBO Charges on O2 along the CO Oxidation Reaction Pathway on Ag16, Ag18, and Ag20

system	QO2 (Int-1)	QO2 (TS-1)	QO2 (Int-2)	QO2 (TS-2)
Ag16	–0.40	–0.62	–0.76	–0.93
Ag18	–0.25	–0.64	–0.47	–0.90
Ag20	–0.18	–0.49	–0.64	–0.77

Furthermore, the calculated heat of formation for the final products in the case of Ag16 and Ag18 are −3.61 and −3.22 eV, respectively. Thus, among the above-studied silver clusters, Ag16 and Ag18 are the most promising model catalysts for CO oxidation with low activation barriers.

Conclusions

In summary, we have investigated the reactivity and catalytic properties of Ag (n = 15–20) clusters using first-principles simulations. The findings point that the extent of O2 adsorption depends largely on the mode of adsorption. It is found that O2 adsorbs strongly with enhanced binding energy in the bridged mode on all the clusters except Ag20 as compared to the atop mode. A substantial increase in the O–O bond lengths accompanied with a pronounced red shift in the O–O frequency in the bridged adsorption also corroborates the obtained results. Simulated reaction pathways reveal that Ag16 and Ag18 clusters entail very low barriers and formation energies for CO oxidation to CO2. Furthermore, the stabilization of transition states is reflected in the enhanced NBO charge transfer to O2 and significant spatial overlap between frontier molecular orbitals, eventually leading to their stabilization. Thus, Ag16 and Ag18 clusters act as efficient catalysts for the environmentally important CO oxidation reaction.