1. Introduction

Non-metallic gold nanoclusters (Au NCs, 1–2 nm) have attracted considerable attention as new promising nanomaterials in the nanoscience and nanotechnology [1]. These well-defined Au NCs are ligated by organic ligands, e.g., thiolate, phosphine, alkyne, and a few to name. The Au NCs are widely exploited for a serial of catalytic reactions in organic transformations [2-5], which is highly valuable and desirable due to their monodispersity fully determined crystal frameworks. And it enables efficient and accurate simulations for better understanding of homogeneous/ heterogeneous catalysis over Au NCs [6-10].

In recent years, remarkable developments have been achieved in wet synthesis of Au NCs by "size-focusing" and "ligandexchange" methodologies [11]. Specific-size ligand-capped Au NCs (i.e., Au_n(L)_m, "L" represents organic ligands) is composed of a precise number of gold atoms, n, and organic ligands, m, (n can vary from ten to a few hundred atoms, equivalent Au core size range of 0.7–2 nm). For example, the size of the Au₂₅(SR)₁₈ cluster is 1.3 nm and these for the Au₁₀₂(SR)₄₄ and Au₁₄₄(SR)₆₀ are 1.7 nm and 1.9 nm, respectively. Au NCs with atomic precision can be ligated by different functional ligands (e.g., thiolate, phosphine, and alkyne) to tune their electronic properties and structures and further tailor their catalytic performance. Of note, the electronic property of the Au NCs also can be slightly turned by the type of the capping ligands (e.g., aliphatic vs. aromatic ligands) [6, 12].

Au NCs are of particular interests in catalysis, as the ultrasmall size of the Au NCs originates strong electron-energy quantization effects [13], in contrast with the continuous conduction band of the metallic Au NPs or bulk gold. In addition, some foreign atoms (e.g., palladium, platinum, silver, copper, etc.) can dope into the core of the Au_nL_m clusters to form bimetallic alloy clusters, i.e., Au_n-xM_xL_m, which can largely turn the cluster's electronic property and give rise to new or superior catalytic activity [14-17].

In this paper, we mainly summarize Au NCs as a novel promising nanogold catalyst for oxygen activation (normal O₂ to singlet O₂ via a photocatalytic process) and aerobic oxidation, including CO transfers to CO₂, sulfides to sulfoxides, alcohol to aldehyde, styrene to styrene epoxide as a major product (benzaldehyde and acetophenone as the minor products), amines to imines, and glucose to gluconic acid, as shown in Scheme 1. Of note, the oxygen activation and oxidation using oxygen as oxidant is an important application in the industry. The correlation of catalytic behaviors with intrinsic frameworks of Au NCs is discussed in details. And then the size-specificity, ligand engineering, and the metal doping effects of the clusters also are included.Finally, the proposal catalytic active sites on the gold clusters and the tentative catalytic mechanisms simulated via DFT (based on the Au clusters' frameworks) are well presented.

2. Optical property, synthesis, and framework of gold nanoclusters 2.1. Optical property

The optical property of gold nanoclusters have greatly attracted scientist interests due to the strong quantum confinement effect, and it is related to their photo-catalysis, in which the Au cluster can absorb specific wavelength [18-22]. For example, the absorption spectrum of Au₂₅(SR)₁₈ shows three bands at 400 (3.10 eV), 450 (2.76 eV), and 670 nm (1.85 eV), which are all due to single-electron transitions between quantized electronic energy levels [23]. The HOMO-LOMO gap of Au₂₅(SR)₁₈ clusters is ca. 1.3 eV (Fig. 1). And the behavior of the gold cluster as molecular species is different from the optical absorption of larger gold nanocrystals, which exhibit a distinct SPR band at 520 nm due to collective excitation of conduction electrons [24]. It is worthy to note that the UV–vis property of the atomically precise gold nanocluster is unique and characteristic "fingerprint". Thus, gold cluster of different gold numbers exhibit diverse UV and optical gaps. Thus, the gold clusters can absorb different light due to their native optical property, and they show distinct activity in the photocatalysis.

2.2. Controlled synthesis of gold nanoclusters

In this review, we only take the controlled preparation of the [Au₂₅(PPh₃)₁₀(SR₁)₅X₂]²⁺ nanorods (H-SR₁: alkyl thiol, H-SC₂H₄Ph and H-S(n-C₆H₁₃)) and Au₂₅(SR₂)₁₈ nanospheres (H-SR₂: aromatic thiol, H-SPh and H-SNap) as an example to introduce the ligand effects on the cluster synthesis. These Au₂₅ nanorods and nanospheres are adequately investigated in the catalytic reactions (see Section 3). The two Au₂₅ clusters were obtained through an one-phase thiol etching reaction of the polydisperse 1.3 nm Au_n(PPh₃)_m particles (Fig. 2). It is interesting that the alkyl thiol gave rise to Au₂₅ nanorods and the aromatic thiol leaded to the Au₂₅ nanospheres during the conversion process. And it is also observed that the Au₂₅ nanorods cannot convert to Au₂₅ nanospheres in the presence of excess thiol (both the alkyl and aromatic thiol) even under thermal conditions [25].

2.3. Framework

Au clusters exhibit distinct crystal structures resolved by X-ray crystallography. The structures of [Au₂₅(PPh₃)₁₀(SR)₅Cl₂]²⁺Cl₂ nanorod and Au₂₅(PET)₁₈ nanosphere are briefly discussed, which are used as model to explore catalytic mechanisms (vide infra). The Au₂₅ nanorod are composed of two icosahedral Au₁₃ units by sharing one common vertex (Fig. 2) [14, 15]. The two Au₅ pentagonal rings at the end of the rod are ligated by ten phosphine ligands. Two Cl atoms bind to two apical Au atoms, and the thiolate ligands bridge the two Au₁₃ icosahedrons (Fig. 2). Whilst, the Au₂₅ nanosphere comprises a 13-atom Au₁₃ icosahedral core and six Au₂(SR)₃ staples (Fig. 2) [16].

3. Catalytic oxidation

During the catalytic oxidation reaction, the oxygen interaction over Au cluster is a key issue. And the gold clusters show strong redox property and electron transfer capacity. For example, the Au₃₈S₂(SAdm)₂₀ cluster [26] exhibited good charge transfer capacity to the redox probe molecules (e.g., K₃Fe(CN)₆) in the cyclic voltammetry analysis. However, it is interesting that the redox property of the cluster was disappeared when β-cyclodextrins (β-CDs) existed in the system. After DFT calculation, it is found that the β-CDs acted as an umbrella to cover the fragile metal cores of the Au₃₈S₂(SAdm)₂₀ nanoclusters, thereby blocking direct interaction with destabilizing agents and hence quenching the charge transfer process [27].

3.1. Photocatalysis for oxygen activation

In catalysis, O₂ activation is of major importance since O₂ is widely applied in many oxidation reactions [28]. Light-induced electrons with large reduction potentials can easily activate normal O₂ to produce singlet O₂ (¹O₂) via a photocatalytic process [29-31].

We have demonstrated that ¹O₂ can be efficiently generated through direct sensitization under visible/near-IR (532, 650, and 808 nm) irradiation by organic-soluble Au₂₅(PET)₁₈^- (PET = phenylethanethiolate) and water-soluble Au₂₅(Capt)₁₈^- (Capt = captopril) in the absence of conventiona organic photosensitizers. ¹O₂ was successfully detected by direct observation of the characteristic ¹O₂ emission around 1276 nm as well as three different ¹O_2^- selective probes. Water-soluble Au₂₅(Capt)₁₈^- clusters were explored for cytocompatibility and photo-dynamic activity toward cancer cells. The quantity of ¹O₂ generated was dependent on the size of clusters; Au₃₈(PET)₂₄⁰ clusters were found to be less effective for the generation of ¹O₂ in comparison to Au₂₅(PET)₁₈^-. In addition, the long lifetime of the electronic excited states of Au₂₅(SR)₁₈^- and the well-defined O₂ adsorption sites are the key factors that promote energy transfer from Au₂₅(SR)₁₈^- to molecular oxygen, thus facilitating the formation of ¹O₂. Finally, neutral Au₂₅(SR)₁₈0 can also produce ¹O₂ as efficiently as does the anionic Au₂₅(SR)₁₈^-.

Very recently, Li et al. reported a new protocol for the synthesis of Au₃₈(SAdm)₂₀ nanoclusters (-SAdm = 1-adamantanethiolate) with moderate efficiency (ca. 10% yield) in comparison to previous reported methods. It is shown that the Au₃₈(SAdm)₂₀ nanoclusters can be applied as efficient photosensitizers for the generation of ¹O₂ under visible/near-IR (e.g., 532 and 650 nm) irradiation (Fig. 3). The formation of ¹O₂ was detected by 1, 3-diphenylisobenzofuran as the chemical trapping probe as well as direct observation of the characteristic ¹O₂ emission (ca. 1276 nm). The efficiency of the ¹O₂ formation using the Au₃₈(SAdm)₂₀ nanoclusters is found to be notably higher than that of Au₂₅(SR)₁₈ nanoclusters [23]. This work demonstrates the promise of Au₃₈(SAdm)₂₀ nanoclusters in the generation of activated singlet oxygen for selective catalytic reactions. Further, they also found that the [Au₁₃(dppe)₅Cl₂]³⁺ cluster with a HOMO-LUMO gap of 1.9 eV showed a very high quantum yield of Φ = 0.71, which is compared with the that of organic dye [32].

3.2. Catalytic oxidation of CO to CO₂

The oxidation reaction of CO to CO₂ has been extensively investigated in gold catalysis [33, 34]. The TiO₂ or CeO₂-supported gold catalyst is generally identified to be the most effective catalyst. Lin and coworkers investigated the electronic properties and the steric effects of the protecting ligands of the Au₂₅ nanoclusters in the CO oxidation [25, 34, 35]. The Au₂₅ nanoclusters include Au₂₅(PPh₃)₁₀(PET)₅X₂ nanorod, and Au₂₅(SR)₁₈ nanosphere (-SR = -PET and -SNap, and -SNap = naphthalenethiol). The three CeO₂-supported catalysts exhibit different performances in CO oxidation. The Au₂₅(PET)₁₈ catalyst starts to show activity with 3.7% at 80 ℃ and reaches up to 80.7% (based on CO conversion) at 100 ℃, while the Au₂₅(SNap)₁₈ and Au₂₅(PPh₃)₁₀(PET)₅X₂ show no catalytic activity at 80 ℃ and only catalytic activity at 100 ℃ with 3.3% and 10.2% CO conversion. Meanwhile, Au₂₅(PET)₁₈, Au₂₅(SNap)₁₈ and Au₂₅(PPh₃)₁₀(PET)₅X₂ catalyst show catalytic activity at 150 ℃ with 98.5%, 98% and 94.6% CO conversion, respectively. The prolonged reaction temperature of Au₂₅(PET)₁₈ (100 ℃) and longer induction period (>20 h) of Au₂₅(SNap)₁₈ and Au₂₅(PPh₃)₁₀(PET)₅X₂ are at 150 ℃, which is caused by the richelectron system of Au₂₅(SNap)₁₈ and Au₂₅(PPh₃)₁₀(PET)₅X₂ and the bulky protecting ligands.

Heteroatom-doped gold clusters have been of great interest to modify the electronic and physical properties and catalytic performance of the clusters in the CO oxidation. Of note, the dopants (e.g., Ag, Cu, Pd, Pt, etc.) are preferentially located at cluster kernel instead of the staple motif [36-40]. Li et al. explore the catalytic activity of bimetallic gold-based M_xAu_25-x(SR)₁₈ (M = Cu and Ag, and SR = PET) clusters and compare with that of homogold Au₂₅(SR)₁₈ in the CO oxidation reaction. The results showed catalytic activity of the CeO₂-supported clusters in the order of Cu_xAu_25-x(SR)₁₈ > Au₂₅(SR)₁₈ > Ag_xAu_25-x(SR)₁₈. The density functional theory calculations (DFT) and Fourier-transform infrared (FT-IR) analyses show that the thiolate ligands are partially removed under reaction conditions (T > 120 ℃). The exposed metal atoms (Au, Ag, and Cu) are deemed as the catalytic active sites (Fig. 4) [41].

Further, DFT calculations showed that the CO adsorption on the doped systems would be more preferable to occur in order of Au₁₃@Cu₂Au₁₀-(SCH₃)₁₅ > Au₁₃@Au₁₂(SCH₃)₁₅ > Au₁₃@Ag₂Au₁₀- (SCH₃)₁₅. Three thiolate ligands on one "Au₂(SCH₃)₃" staple was deemed to be detached at the high reaction temperature (100 ℃). And the CO reactant was adsorbed on the naked Au atoms of the Au₂ unit (Fig. 5). The DFT is in good agreement with the experimental results of catalytic performance of the clusters. Of note, the CO adsorption and activation should be the key step of the Au cluster catalyzed CO oxidation. And it is worthwhile to mention that the trend of the CO adsorption (ΔE_ad) results are in well consistent with those previously reported on the pure metals (ΔE_{ad, Cu} > ΔE_{ad, Au} > ΔE_{ad, Ag}) [42]. Another possible reason includes smaller size of Cu and Au in comparison with that of Ag, thereby facilitating a better interaction with CO.

3.3. Aerobic oxidation of alcohol

Zhang et al. explored the activity of [Au_25-x(PET)_18-xM]NH₃ (M = Cu, Co, Ni, and Zn) nanoclusters in aerobic alcohol oxidation [43]. The results show that the CeO₂-supported nanoclusters showed good catalytic activity and fellow the order, [Au_25-x(PET)_18-xNi] > [Au_25-x(PET)_18-xCu] > [Au_25-x(PET)_18-xZn] > [Au_25-x(PET)_18-xCo]. The [Au_25-x(PET)_18-xM] nanoclusters also showed different product selectivity. The [Au_25-x(PET)_18-xZn] and [Au_25-x(PET)_18-xCo] catalysts preferably yield benzaldehyde, [Au_25-x(PET)_18-xNi] yields benzyl acid, and [Au_25-x(PET)_18-xCu] yields benzaldehyde and benzyl acid. The exposed metal atoms are considered as the catalytic active sites. The catalytic activity and selectivity of the [Au_25-x(PET)_18-xM] catalysts are greatly turned and mediated by the transition metal atom dopants (Scheme 2).

3.4. Selective oxidation of styrene

Three main products, benzaldehyde, styrene epoxide and acetophenone, can be found in the gold cluster-catalyzed oxidation of styrene [44]. Zhu et al. investigated the styrene oxidation catalyzed by Au cluster catalysts only using O₂ as the oxidant [45]. The reaction selectively gave rise to benzaldehyde (up to ~70%) and ~25% styrene epoxide and < 5% acetophenone. Further, they compared the reactions under three oxidant systems: (Ⅰ) only using tert-butyl hydroperoxide (TBHP); (Ⅱ) using TBHP as the initiator and O₂ as the oxidant; (Ⅲ) only using O₂. Of note, TBHP is more reactive than O₂. Au₂₅(SR)₁₈ gave high conversion of styrene (86%) with 100% selectivity for benzaldehyde in the system Ⅰ. Meanwhile in the system Ⅱ, the gold nanoclusters showed much lower activity (25% conversion with 100% selectivity for benzaldehyde), and system Ⅲ (solely O₂) was even lower (18% conversion with 80% selectivity for benzaldehyde). However, Dreier et al. found that the high activity of Au₂₅(SR)₁₈ in the system Ⅰ was catalyzed by active Au(Ⅰ) species instead of Au clusters; Au(Ⅰ) species was yielded by the oxidative decomposition of gold nanoclusters [46].

To compare with catalytic activity of the Au₂₅(SR)₁₈, Qian et al. also investigated the monoplatinum-doped Pt₁Au₂₄(SR)₁₈ clusters in the selective epoxidation of styrene using PhI(OAc)₂ as oxidant; of note, PhI(OAc)₂ is a much softer oxidant than TBHP and it cannot decompose gold nanoclusters under mild conditions (e.g., < 80 ℃) [46]. The TiO₂-supported Pt₁Au₂₄(SR)₁₈ catalyst performed higher catalytic activity (i.e., styrene conversion and benzaldehyde selectivity) than that for Au₂₅(SR)₁₈. The Pt₁Au₂₄(SR)₁₈ prefer to yield benzaldehyde as major product (~90% selectivity) [47]. Further, the bimetallic M_xAu_25-x(PET)₁₈ (M = Cu and Ag) nanoclusters was investigated in the selective oxidation of styrene [11]. The results indicate that the selectivity for benzaldehyde and styrene epoxide are affected by the inner shell of the icosahedral M_xAu_25-x(SR)₁₈ clusters. The Cu_xAu_25-x(SR)₁₈ clusters primarily increase the selectivity to benzaldehyde without obviously changing the activity, while the Ag_xAu_25-x(SR)₁₈ clusters exhibit higher activity and benzaldehyde selectivity than the homogold cluster. The results provide the atomic level insights into the factors influencing the catalytic activity and selectivity in styrene oxidation [48].

3.5. Oxidation of sulfide to sulfoxide

Sulfoxides (R-S(=O)-R′) are versatile intermediates for the preparation of important medicinally and biologically products [49, 50]. Sulfoxides can be prepared by oxidation conversion from sulfides (R-S-R′) to sulfoxides catalyzed by gold clusters due to the interaction between the sulfur atom and the surface gold atoms. While, the product sulfoxides only weakly bind to the gold surface, which can prevent the over-oxidation of the sulfone (R-S (=O)₂-R′) products. So, the selective oxidation sulfides to sulfoxides catalyzed by gold clusters should be easily achieved [51, 52].

TiO₂ oxide has been proved to be a good support for the Au nanoclusters in the selective sulfoxidation. Therefore, in 2012, Au₂₅(PET)₁₈ was investigated in the selective oxidation of organic sulfide to sulfoxide using PhIO as the oxidant [53]. The results show that the TiO₂-supported Au₂₅(PET)₁₈ nanocluster catalysts give rise to high catalytic activity (e.g., 97% conversion of Ph-S-CH₃ and 92% selectivity for Ph-S(=O)-CH₃ sulfoxide) and show excellent recyclability and durability in the sulfoxidation process.

In 2015, we report the one-pot synthesis of gold nanoclusters-Au₁₄₄(SCH₂Ph)₆₀ with moderate efficiency (ca. 20% yield based on HAuCl₄)byetching thepolydispersed Au_n(SG)_m nanoclusters inthe presence of the excess H-SCH₂Ph via a combination of "ligandexchange" and "size-focusing" processes [54]. Moreover, the Au₁₄₄(SCH₂Ph)₆₀/TiO₂ catalyst exhibits good catalytic performance (92% conversion of methyl phenyl sulfide with 99% selectivity for sulfoxide) in the selective sulfoxidation (Table 1), and size dependence of the atomically precise gold nanocluster catalyst is observed in the catalytic reactions: Au₁₄₄(SCH₂Ph)₆₀ > Au₉₉(SPh)₄₂ >Au₃₈(SCH₂CH₂Ph)₂₄ >Au₂₅(SCH₂CH₂Ph)₁₈. Further, the steric effect also was investigated; the catalytic performance of the Au₁₄₄ nanoclusters capped with different thiolate ligands (including -SCH₂Ph, -SC₂H₄Ph, and -SC₆H₁₃) was tested and compared. The Au₁₄₄(SCH₂Ph)₆₀ nanoclusters showed the best catalytic activity among the three Au₁₄₄ nanoclusters (Table 1, entries 4–6) [54].

In 2016, solvent-solvable Au₁₀₂(SPh)₄₄ nanoclusters, prepared via a kinetically controlled synthetic protocol, can convert to Au₉₉(SPh)₄₂ with equivalent thiophenol ligands by the thermodynamic "size-focusing" method. And, the TiO₂-supported Au₁₀₂(SPh)₄₄ nanocluster catalyst was further investigated in the selective oxidation of sulfides to sulfoxides using PhIO as the oxidant under mild conditions (at 40 ℃). It gave rise to high catalytic activity (e.g., 80%–99% conversion of R-S-R′ sulfides and 96%–99% selectivity for R-S(=O)-R′) [55].

Very recently, Au₃₈S₂(SAdm)₂₀ nanoclusters were explored in the sulfoxidation in open quartz vessels with O₂ bubbling under irradiation at 532nm [55]. The Au₃₈(SAdm)₂₀ nanoclusters showed no activity in the dark (i.e., absence of light) or without oxygen. Its activity increased to 57% when the solution is exposed to exciting light at 532nm. Of note, the cluster gave a decent 100% selectivity to produce methyl phenyl sulfoxide. Finally the Au₃₈ nanoclusters presented a better catalytic activity than Au₂₅, which is mainly due to the higher efficiency of ¹O₂ photogeneration over Au₃₈(SAdm)₂₀.

3.6. Efficient aerobic oxidation of D-glucose

Recently, the activated carbon (AC) supported gold clusters were found to exhibit excellent catalytic performance in the selective aerobic oxidation of D-glucose into gluconic acid (Scheme 3). We explore the catalytic performance of the activated carbon (AC)-supported gold cluster-Au₃₈(PET)₂₄ in the aerobic oxidation of glucose to gluconic acid [56]. The Au₃₈(PET)₂₄/AC-120 catalysts (annealed at 120 ℃ in air) exhibited high catalytic activity and significantly better performance than the corresponding catalysts Au₃₈/AC-150 and Au₃₈/AC-300 (treated at 150 ℃ and 300 ℃ to remove the protecting thiolate ligands). The high activity of the robust Au cluster was a result of the partial ligand removal, providing catalytically active sites. Au₃₈(PET)₂₄/AC-120 also showed excellent recyclability (up to seven cycles). This new ultra-small gold nanomaterial is expected to find great potential for further applications.

It is well known that the catalytic properties of nanocatalysts are closely associated with the surface area of nanoparticle as particle surfaces are usually deemed as the catalytic active sites. Thus, the size-dependence of gold nanoparticles becomes a very crucial factor in the nanocatalysis. So, we explored the catalytic performance of the atomically precise Au_n(PET)_m/AC nanocluster catalysts of different gold atoms and sizes, including Au₂₅(PET)₁₈ (~1.2nm), Au₃₈(PET)₂₄ (~1.5nm), and Au₁₄₄(PET)₆₀ (~1.9nm) nanoclusters [57]. The results showed good catalytic activity in the aerobic oxidation of D-glucose into gluconic acid (or gluconates) with ~98% selectivity. And size dependence of the atomically precise gold nanocluster catalyst is observed in the catalytic reactions: Au₁₄₄(PET)₆₀/AC >Au₃₈(PET)₂₄/AC > Au₂₅(PET)₁₈/AC. On the whole, the core size of the gold nanoclusters played an important role in the catalytic process and these well-defined gold nanoclusters can be expanded to another mild catalysis in the fundamental studies.

3.7. Oxidation of amines to imines

Selective oxidation of amines to imines is of significant importance due to the versatile applications of the imines product in fine pharmaceuticals and chemicals. We have synthesized atomically precise Au₂₅(PPh₃)₁₀[SC₃H₆Si(OC₂H₅)₃]₅Cl₂/TiO₂ nanoclusters exhibiting superior selective visible light photocatalytic oxidation of amines to imines with molecular oxygen [58]. The turnover frequency (TOF) of 4-methylbenzylamine oxidation is found to be 1522h^-1, which is considerably higher than that conventional gold catalysts (Fig. 6). The gold nanoclusters present good recyclability and stability for the oxidation of a wide range of amines. The high catalytic activity is associated with its unique electronic structure and framework. This work might lay a firm foundation of visible light photocatalytic selective organic transformation over gold nanoclusters.

Further, DFT simulations showed that the benzylamine reactant cannot be adsorbed on the surface of gold clusters. Therefore, we initially speculated that there should be partial ligand removal under the reaction conditions, providing active sites for the oxidation (Fig. 7). We here choose [Au₂₅(PH₃)₁₀(SH)₅Cl₂]²⁺ as a model for the simulations to rationalize the possible ligand removal process in the presence of amine. Three organic ligands (i.e., PPh₃, MPTES, and Cl) ligated on the gold nanoclusters can be lost during the catalytic process. The calculated results indicated the detachment of a Cl^- or PH₃ ligands required 0.9 eV/mol and 1.3 eV/mol, respectively, which are considerably lower than that for removing a thiolate ligand (-SH, 2.7 eV/mol, Fig. 7). Thereby, Cl^- and phosphine ligands provide gold open metal sites (maybe associated with the cataclytic active sites) due to easily leaving the nanoclusters.

Moreover, we investigated the photocatalysis properties of Au₃₈(SAdm)₂₀ nanoclusters via a one-step chemical transformation of benzylamine to N-(phenylmethylene) benzenemethanamine (Table 2) [21]. No products were found in the blank experiment, in which either the gold nanoclusters were absent or the experiment was run in the dark. When the O₂ atmosphere was replaced by N₂ gas, the catalytic activity was suppressed, which imply that selective aerobic oxidation occurs due to the photogeneration of ¹O₂ at Au₃₈(SAdm)₂₀ nanoclusters. The work further illustrates the promise of Au₃₈(SAdm)₂₀ nanoclusters in the generation of activated singlet oxygen for selective catalytic reactions.

4. Summary

Compared with conventional nanoparticle catalysts, the atomically precise Au nanoclusters exhibit a well-known surface structure, providing an opportunity for in-depth understanding of the nature of catalysts and the correlation of the structure and catalytic properties. To summarize, a series of Au_n(SR)_m clusters (including Au₂₅(SR)₁₈, Au₃₈(SR)₂₄, Au₁₄₄(SR)₆₀, Au₁₀₂(SR)₄₄ and Au₉₉(SR)₄₂) have been explored the utilization of such clusters as active sites in selective oxidation and photocatalysis. With regard to catalyst applied research, it has been demonstrated that precisely synthesized Au_n(SR)_m clusters can also contribute to the creation of highly active water-splitting photocatalytic materials. It has been demonstrated that the catalytic performances (including activity and selectivity) are affected by the distinct size-dependence, electronic structure, framework and steric effect of the gold nanoclusters.

Our next goal is further improvement of photocatalytic activity of gold nanoclusters catalysts. Future research is expected to achieve steady advances in revealing the nature of the cluster surface structure, ligand binding modes, and many other fundamental issues. We believe that the gold nanocluster catalyst should be a promising new class of catalysts and is expected to find more applications in catalysis in future research.

Acknowledgment

We acknowledge financial support by the Program for the Outstanding Innovative Teams of Higher Learning Institutions of Shanxi (OIT), Shanxi Province Hundred Talent Project.