Thiolate-protected Au₃₈(SR)₂₄ nanocluster: size-focusing synthesis, structure determination, intrinsic chirality, and beyond

Introduction

Ultrasmall Au nanoparticles (ca. 1-2 nm core size), often called Au nanoclusters, have become one of the important types of nanomaterials targeted in current nanoscience research [1–25]. Unlike their larger counterpart – crystalline Au nanoparticles with size above 2 nm, whose optical properties are dominated by surface plasmon resonances due to excitation of electrons in the continuous conduction band, Au nanoclusters instead exhibit discrete electronic structure and molecular-like properties [26–36],such as HOMO-LUMO electronic transition [1, 37, 38], intrinsic magnetism [29, 39], chiroptical properties [40–42], and enhanced catalytic properties [43–45]. Considering the extreme sensitivity of the properties of Au nanoclusters to the number of Au atoms in the particle, these ultrasmall nanoclusters are better represented as Au_nL_m, where n is the number of Au atoms and m the number of ligands (such as thiolate, –SR).

Early work on thiolate-protected Au nanoparticles was pursued in the 1990s [46–49]. Brust et al. developed a two-phase protocol for preparing thiolate-protected Au nanoparticles. However, the nanoparticles obtained at that time were quite polydisperse (typically ranging from ~2 to 5 nm) [46]. Mirkin et al. developed thiolated-DNA for functionalization of Au nanoparticles, which are extremely robust and have acquired wide applications in sensing, biological labeling, and biomedicine [47]. In 1996, Whetten et al. synthesized a series of narrow size distributed nanoparticles in the range of 1.0–3.5 nm by solvent fractionation, and several distinct critical sizes were identified by laser desorption-ionization mass spectrometry (LDI-MS) [48]. The atomically monodisperse Au nanoclusters were not obtained at that time.

Au₃₈(SR)₂₄ nanoclusters as the first Au nanoclusters that were observed to show the molecular-like properties [26, 27] have been extensively studied. The history of Au₃₈(SR)₂₄ can be traced back to 1997 [26]. A series of ultrasmall Au nanoparticles were isolated using solvent fractionation. The smallest Au nanoparticles have a Au core mass of ~8 kDa determined by LDI-MS (Au core refers to Au_xS_y). However, due to the limitation of characterization techniques, the formula of ~8 kDa species was not determined. Whetten et al. estimated the ~8 kDa species consists of ~38 Au atoms [26]. After that, a sample of ~8 kDa Au:SC₆H₁₃ species was analyzed by X-ray photoemission spectroscopy (XPS), indicating it had an elemental composition of 62 % Au and 38 % S. If the Au core has 38 Au atoms, the number of thiol ligands should be 23–24 [50]. The first theoretical model of Au₃₈(SR)₂₄ was proposed based on the assignment of this formula [50].

In 1998, Chen et al. observed the molecular-like charging properties for Au₃₈(SR)₂₄, where the HOMO-LUMO energy gap is nearly 0.9 eV [27]. It is noted that some early work (prior to 2007) from Murray and co-workers erroneously assigned another ultrasmall nanocluster – Au₂₅(SR)₁₈ as Au₃₈(SR)₂₄ due to their similar molecular like properties and Au/S ratio [51–55]. Advances in electrospray ionization-mass spectrometry (ESI-MS) and matrix-assisted laser desorption ionization-mass spectrometry (MALDI-MS) facilitated the successful determination of intact cluster ions without fragmentation. These techniques provided the exact composition assignment with atomic precision [56, 57]. Now it is easy to distinguish these two nanoclusters (Au₂₅ and Au₃₈).

After successful crystal structure determination of Au₂₅(SR)₁₈ by the Murray [58] and Jin [37] groups in 2008, some researchers doubted whether the Au₃₈(SR)₂₄ really existed. Shortly after that, the formula of Au₃₈(SR)₂₄ was unambiguously identified by the Tsukuda group with ESI-MS. A series of Au₃₈(SC_nH_2n+1)₂₄ (n = 6, 10, 12, and 18) clusters were isolated using recycling size-exclusion chromatography (SEC) [59] or solvent fractionation [60]. The hexanethiolate-protected Au₃₈ clusters were also prepared by etching polydispersed Au clusters with excess thiol; the formula was assigned as Au₃₈(SC₆H₁₃)₂₂ due to the loss of ligands in MALDI-MS characterization [61]. However, the above synthesis suffered from a complicated purification process and low yield (<1 %). In order to study the fundamental properties and explore the potential applications of Au₃₈(SR)₂₄ nanoclusters, it was therefore highly desirable to achieve a more efficient large-scale synthesis.

Size-focusing synthesis of Au₃₈(SR)₂₄ nanoclusters

On the basis of the early work from various groups, a new size-focusing method was developed to synthesize Au₃₈(SR)₂₄ nanoclusters with high yield (~25 %, Au atom basis) [62–64]. This method was based on the thermal thiol etching process, as shown in Fig. 1a. In the first step, glutathione (GSH) protected Au_n(38≤n≤~100) nanoclusters were prepared by reducing Au(I)-SG in acetone solution; in the second step, the size-mixed Au_n(SG)_m nanoclusters reacted with excess phenylethylthiol (PhC₂H₄SH) for ~40 h at 80°C and monodisperse Au₃₈(SC₂H₄Ph)₂₄ clusters were finally obtained [63].

Fig. 1

(a) Schematic diagram of a two-step procedure for synthesizing monodisperse Au₃₈(SC₂H₄Ph)₂₄ nanoclusters in high yield. (b) MALDI-mass spectrum of Au_n(SR)_m before and after thiol etching. Adopted with permission from ref. [63].

Figure 1b shows the size-focusing mechanism by MS. The size-focusing process occurred over the etching-induced growth process. The formula and high purity of Au₃₈(SC₂H₄)₂₄ nanoclusters were further confirmed by ESI-MS, MALDI-MS, and SEC. The major advance of this method was the control of size distribution of Au_n(SR)_m mixture before the thermal thiol etching step. In previous procedures [60, 65], a mixture of Au_n(SR)_m resulted with the modified Brust method, and the size distribution of Au_n(SR)_m was quite broad. Even after thermal etching, it still contained several stable species in the solution, including Au₃₈(SR)₂₄, Au₁₄₄(SR)₆₀, and others. The separation process was nontrivial and time-consuming. To solve this problem, the initial size distribution of Au_n(SR)_m (i.e., “precursors” for the size-focusing step) was adjusted by controlling the type of thiol ligands and reaction solvent. The combination of glutathione as thiol ligands and acetone as solvent was finally chosen after screening various thiols and solvents [63]. The Au_n(SG)_m mixture cannot be dissolved in acetone. Therefore, after the nucleation of Au_n(SG)_m, the nanoclusters immediately precipitated out of solution and the further growth was inhibited. The size distribution of the formed Au_n(SG)_m ranged from Au₃₈ to Au_~100. In the subsequent thermal thiol etching process, those relatively large Au nanoclusters were found to be decomposed and converted to Au₃₈(SR)₂₄ nanoclusters. This process was largely driven by thermodynamic stability of Au₃₈(SR)₂₄, which was the most robust species in this particular size range. The final product only contained monodisperse Au₃₈(SR)₂₄ and Au(I)-SR polymer; no larger or smaller nanoclusters were detected. The Au(I)-SR polymer was poorly soluble in almost all solvents, hence easy to separate from the product. This method was highly reproducible [41, 66–68] and versatile for various alkanethiols as ligands for Au₃₈ nanoclusters, i.e., Au₃₈(C_nH_2n+1)₂₄ (n = 4, 6, 8, 12, 16) [69, 70]. The size-focusing method developed in the preparation of Au₃₈(SR)₂₄ cluster has been successfully applied in the synthesis of other magic-sized Au_n(SR)_m nanoparticles with molecular purity, including Au₂₈(SR)₂₀ [71], Au₅₅(SR)₃₁ [72], Au₁₄₄(SR)₆₀ [73, 74], and Au₃₃₃(SR)₇₉ [18], etc.

The truly monodisperse Au₃₈(SC₂H₄Ph)₂₄ nanocluster displays distinct molecular-like optical and electrochemical properties [63]. The UV-vis-NIR absorption spectrum exhibits a series of peaks at 1050 nm (1.18 eV), 750 nm (1.66 eV), 620 nm (2.00 eV), 560 nm (2.21 eV), 520 nm (2.39 eV), and 490 nm (2.53 eV), as shown in Fig. 2a. By extrapolating the lowest energy absorption peak to zero absorbance, the optical energy gap (HOMO-LUMO gap) is determined to be 0.92 eV (Fig. 2a inset). The highly structured optical spectrum serves to give convenient “fingerprints” for one to quickly identify Au₃₈(SR)₂₄ nanoclusters (e.g., in routine synthesis). Figure 2b shows differential pulse voltammogram (DPV) of Au₃₈(SC₂H₄Ph)₂₄ nanoclusters. The electrochemical energy gap is ~1.2 V, i.e., the gap between the first oxidation wave (ox1, +0.48 V) and the first reduction wave (re1, -0.71 V). The charging energy can be estimated from the difference between the first oxidation wave (ox1, +0.48 V) and the second oxidation wave (ox2, +0.68 V), i.e., 0.68 –0.48 = 0.2 V. The HOMO-LUMO gap is thus 1.0 V after subtracting the charging energy from the electrochemical energy gap, which is very close to the optical energy gap (0.92 eV).

Fig. 2

(a) UV-vis-NIR absorption spectrum of Au₃₈(SC₂H₄Ph)₂₄ nanoclusters; the inset shows the spectrum on the energy scale (eV). (b) Differential pulse voltammogram (DPV) at room temperature. Adapted with permission from ref. [63].

Structure determination of Au₃₈(SR)₂₄ nanoclusters

These atomically precise nanoclusters allow for the growth of X-ray quality single crystals and subsequent determination of their total structure (core plus surface) by X-ray crystallography [2, 37, 58, 75, 76]. Understanding the total structure, in particular the surface atom arrangement, is of paramount importance in order to study the catalytic properties of metal nanoclusters for future design of highly efficient catalysts [43, 77–80]. In order to determine the total structure by single crystal X-ray diffraction, crystallization of nanoclusters is the critical step, but unfortunately the growth of single crystals, in general, has long been a huge challenge. Significantly, the Au₃₈(SR)₂₄ nanoclusters with molecular purity achieved by the size-focusing methodology have led to successful crystallization [81]. The unit cell of Au₃₈(SCH₂CH₂Ph)₂₄ crystals contains a pair of enantiomeric nanoparticles (left- and right-handed, Fig. 3).

Fig. 3

Crystal structure of Au₃₈(SC₂H₄Ph)₂₄ nanoclusters. Adapted with permission from ref. [81]. A single crystal (~0.2 mm) was shown in the photograph (upper right).

The left-handed isomer is chosen for a detailed analysis of the Au₃₈ structure. As shown in Fig. 4, the nanocluster shows a prolate shape (diameter ~0.8 nm, length ~1.1 nm, Au center-to-center distances). The core of Au₃₈(SR)₂₄ consists of a face-fused biicosahedral Au₂₃ motif (13 + 13 – 3 = 23), Fig. 4a. The fusion of the two icosahedra occurs along a common C₃ axis. The rod-like Au₂₃ kernel is structurally strengthened by three monomeric staples (–SR-Au-RS–) (Fig. 4b). Then, the top icosahedron is further capped by three dimeric staples (–SR-Au-SR-Au-RS–) (Fig. 4c), which are arranged in a rotating fashion, like a tri-blade “propeller”. A similar arrangement of the other three staples is found on the bottom icosahedron, but the bottom “propeller” rotates by ~60° relative to the top one, forming a staggered dual-propeller configuration (Fig. 4d). The entire nanoparticle becomes chiral due to the rotative arrangement of the dimeric staples, that is, one enantiomer is clockwise and another is anti-clockwise. The determination of the crystal structure of the Au₃₈(SR)₂₄ nanoparticle paves the way for the deep understanding of its electronic and catalytic properties.

Fig. 4

Structure analysis of Au₃₈(SR)₂₄ nanoclusters. (a) Au₂₃ core, (b) Au₂₃ core with three monomeric Au(SR)₂ staples, (c) Au₂₃ core with three dimeric Au₂(SR)₃ staples, (d) overall Au₃₈S₂₄ framework. Color labels: yellow, S; all the other colors are for Au atoms in different positions. The carbon tails (–CH₂CH₂Ph) are not shown for clarity. Adapted with permission from ref. [81].

It is worth mentioning that several candidate structures had been proposed by theoretical chemists. In 1999, Häkkinen et al. proposed a structure of Au₃₈(SR)₂₄ by placing 24 isolated thiolates on the surface of a truncated octahedron containing 38 Au atoms [50]. After that, Garzon [40, 82, 83] Jiang [84, 85], Tsukuda [60], Zeng [86], and Aikens [87] proposed several structures with symmetric or disordered Au core. The energy difference between Häkkinen’s original model in 1999 to Zeng’s model in 2008 is ~9 eV, a substantial reduction in energy (i.e., Zeng’s model being significantly more stable) [88]. Of them, the models predicted by Zeng and Aikens showed good agreement with the crystal structure. Recently, a theoretical explanation for the unusual stability of Au₃₈(SR)₂₄ cluster was reported [89]. Based on the super valence bond model, the 23c-14e double-icosahedral Au₂₃⁽⁺⁹⁾ core of Au₃₈(SR)₂₄ is proved to be a superatomic molecule, which is analogous to the F₂ molecule in both molecular orbital (MO) diagrams and bonding patterns [89].

Zhang and co-workers reported solution-phase X-ray absorption spectroscopy (XAS) studies of Au₃₈(SR)₂₄ nanoclusters. Distinct solvation effects on the structure and bonding of Au-thiolate clusters were observed [90]. With respect to bimetal nanoclusters, two-Pd-atom doping of Au₃₈(SC₂H₄Ph)₂₄ has been reported [91, 92], and highly pure Pd₂Au₃₆(SC₂H₄Ph)₂₄ clusters were successfully isolated, in which two Pd atoms were doped at the two center positions of Au₃₈(SC₂H₄Ph)₂₄. The stability investigation revealed that Pd₂Au₃₆(SC₂H₄Ph)₂₄ was more robust than the homo-Au cluster Au₃₈(SC₂H₄Ph)₂₄ against degradation in solution and core etching by thiols [92]. In addition, the studies of selenolate-protected Au₃₈ clusters found that a highly pure form of Au₃₈(SeC₁₂H₂₅)₂₄ could be synthesized by replacing all the SC₂H₄Ph ligands of Au₃₈(SC₂H₄Ph)₂₄ with SeC₁₂H₂₅, and selenolate-protected Au₃₈ clusters showed the same degree of stability as thiolate-protected Au₃₈ clusters [93].

Chirality of Au₃₈(SR)₂₄ nanoclusters

The Au₃₈(SCH₂CH₂Ph)₂₄ nanocluster shows intrinsic chirality. The chirality of Au₃₈(SR)₂₄ originates from the dual-propeller-like distribution of the six dimeric Au₂(SR)₃ staples, with one enantiomer exhibiting clockwise (L-) and the other anti-clockwise (R-) arrangements [81]. To probe the internal chiral structure of Au nanoclusters, circular dichroism (CD) is, generally speaking, a good method, but the racemic pair of Au nanoclusters results in a net zero signal. That was why the chiral structure of Au₃₈ nanoclusters was not revealed until the determination of their crystal structure. The determination of crystal structures of nanoclusters is still highly challenging since it is very difficult to culture high-quality single crystals of Au nanoclusters. It is thus highly desirable to develop an analytical method to detect the intrinsic chirality, especially in racemic mixtures. Recent research showed that NMR spectroscopy was capable of probing the nature of chiral nanoclusters using the diastereotopicity induced in the α-CH₂ protons of the –SCH₂CH₂Ph chains (Fig. 5) [94].

Fig. 5

(a) 1D NMR of Au₃₈(SCH₂CH₂Ph)₂₄ nanoclusters (¹H-¹³C HSQC). The α-CH₂ and β-CH₂ are as labeled (see the top of panel A); 1–4 denote the four sets of ligands; a and b represent each diastereotopic proton for each carbon (α- or β-CH₂) in the four sets of ligands. (b) Diastereotopicity in the CH₂ protons of the ligands on nanoclusters (chiral metal-core-induced diastereotopicity). Adapted with permission from ref. [94].

The intrinsically chiral Au₃₈(SCH₂CH₂Ph)₂₄ and nonchiral Au₂₅(SCH₂CH₂Ph)₁₈^-TOA⁺ nanoclusters were used as two model systems to investigate how the NMR spectroscopy correlates with the nature of chirality in nanoclusters.

From 1D to 2D NMR of Au₃₈(SCH₂CH₂Ph)₂₄ nanoclusters, four sets of PhCH₂CH₂S– ligands were detected, corresponding to four different chemical environment of ligands. Of them, three sets of –SR ligands come from the dimeric –SR-Au-SR-Au-SR– staples and one set of –SR ligands comes from the monomeric –SR-Au-SR– [94]. The chiral Au₃₈(SR)₂₄ clusters showed different ¹H signals for the two germinal protons in the α-CH₂ group of the ligand, and the same for the β-CH₂, so-called diastereotopicity. For α-CH₂, a chemical shift difference of up to ~0.8 ppm for each germinal proton was observed (Fig. 5a). As for the nonchiral Au₂₅(SCH₂CH₂Ph)₁₈^-TOA⁺ clusters, we found that two chemically inequivalent sets of –SCH₂CH₂Ph ligands, which were corresponding to the interior and exterior sites of the ligands in the dimeric staples. However, no diastereotopicity was detected. Two protons of each CH₂ group are chemically equivalent, indicating nonchiral Au₂₅ core [94]. This diastereotopicity principle will allow for future identification of intrinsically chiral nanoclusters prior to crystal structure determination.

The separation of enantiomers of Au₃₈(SCH₂CH₂Ph)₂₄ by high-performance liquid chromatography (HPLC) column was reported by Bürgi and co-workers [41]. The racemic Au₃₈(SCH₂CH₂Ph)₂₄ nanoclusters were separated by using a chiral cellulose-based analytical HPLC column and hexane/isopropanol (80:20) as eluent at room temperature. As shown in Fig. 6a, two well-separated peaks are observed at 8.45 and 17.45 min, corresponding to enantiomers 1 and 2. The UV-vis spectra of racemic clusters, enantiomers 1 and 2 are identical (Fig. 6b). The CD spectra of enantiomers 1 and 2 show perfect mirror images, and 11 clear signals are observed between 230 and 800 nm (Fig. 6c) [41]. These CD peaks match well with transitions in the absorption spectrum at room temperature [63] and at low temperature [95]. The strong anisotropy factor (up to 4×10^–3) indicates that the intrinsic chirality due to ligand and surface Au atom arrangements can contribute significantly to the net optical activity (Fig. 6d) [41].

Fig. 6

(a) HPLC-chromatogram and (b) UV-vis spectra of HPLC-separated rac-Au₃₈(SCH₂CH₂Ph)₂₄. (c) CD spectra of enantiomers 1 (black) and 2 (red) and the racemic Au₃₈(SCH₂CH₂Ph)₂₄ (blue) before separation; (d) corresponding anisotropy factors of enantiomers 1 and 2 and of the racemate. Adapted with permission from ref. [41].

Bürgi et al. also reported the racemization of intrinsically chiral Au₃₈(SCH₂CH₂Ph)₂₄ nanoclusters at modest temperature (40–80°C) [96]. The drastic surface rearrangement took place without significant decomposition. The determined activation energy for the inversion reaction was only 28 kcal/mol, which indicated that the process occurred without complete break of the Au–S bond [96]. The ligand exchange between chiral R-1,1′-binaphthyl-2,2′-dithiol (R-BINAS) and racemic Au₃₈(SCH₂CH₂Ph)₂₄ was investigated. It was found that this process was diastereoselective with a preference of R-BINAS for the left-handed enantiomer [66]. The partial exchange of dithiol (R-BINAS) made the Au-thiolate interface more rigid and effectively stabilized the cluster against inversion [97].

Size-adjacent nanoclusters – Au₄₀(SR)₂₄ and Au₃₆(SR)₂₄

Au₃₈(SR)₂₄ is not the only cluster in the core mass ~8 kDa species. Indeed, Au₄₀(SR)₂₄ is another member in the ~8 kDa species. In the size-focusing synthesis of Au₃₈(SCH₂CH₂Ph)₂₄, by applying a shorter thermal thiol etching time (~18 h), instead of ~40 h, Au₄₀(SCH₂CH₂Ph)₂₄ was found to co-exist with Au₃₈(SCH₂CH₂Ph)₂₄ and can be isolated by SEC [98]. On the other hand, the complete ligand exchange from –SCH₂CH₂Ph to –SPh-t-Bu induced the significant size and structure conversion from Au₃₈(SCH₂CH₂Ph)₂₄ to Au₃₆(SPh-t-Bu)₂₄ [99, 100].

The synthesis and isolation of Au₄₀(SCH₂CH₂Ph)₂₄ clusters required three main steps [98]. Au_n(SG)_m mixture (38≤n≤~100) was first prepared by reducing Au(I)-SG polymer with NaBH₄ in an acetone solution. Then, the obtained Au_n(SG)_m mixture was etched with excess thiol (PhCH₂CH₂SH) at 80°C for ~18 h. The starting polydispersed Au_n(SR)_m clusters were gradually converted to Au₃₈(SCH₂CH₂Ph)₂₄ and Au₄₀(SCH₂CH₂Ph)₂₄ clusters due to higher stability of these two clusters than other sizes in the distribution. Without further etching, both Au₃₈(SCH₂CH₂Ph)₂₄ and Au₄₀(SCH₂CH₂Ph)₂₄ were present in the product and successful separation was achieved by SEC [98]. Figure 7a shows a typical SEC chromatograph of ~18 h etched product, in which two peaks (centered at 14.0 and 14.4 min) were observed. The MALDI-MS and on-line recorded UV-vis spectra indicated that 12.5 to 14.0 min eluate was pure Au₄₀(SCH₂CH₂Ph)₂₄, the 14.4 to 15.6 min eluate was pure Au₃₈(SCH₂CH₂Ph)₂₄, and the in-between eluate is the mixture of these two species [98]. Similarly, Bürgi also reported the SEC for semipreparative scale separation of Au₃₈(SR)₂₄ and Au₄₀(SR)₂₄ [101].

Fig. 7

(a) SEC chromatograms of nanoclusters (b) MALDI mass spectra of the crude product(a) and isolated Au₄₀(SR)₂₄ (b) and Au₃₈(SR)₂₄ (c) nanoclusters. Adapted with permission from ref. [98].

The optical properties of isolated Au₄₀(SC₂H₄Ph)₂₄ clusters were also investigated. Au₄₀(SC₂H₄Ph)₂₄ does not show prominent peaks in the absorption spectrum, in which only three steps at ~2.2, 3.2, and 4.4 eV were observed. The HOMO-LUMO gap is estimated to be ~1.0 eV from optical absorption [98]. The enantioseparation of Au₄₀(SCH₂CH₂Ph)₂₄ was successfully achieved by using chiral HPLC [102]. The chiroptical responses of two separated enantiomers with achiral thiolate ligands (PhCH₂CH₂SH) demonstrated that Au₄₀(SR)₂₄ also has an intrinsically chiral structure [102]. Based on the experimental characterization of Au₄₀(SR)₂₄ via ligand exchange, chiroptical spectroscopy, and transmission electron microscopy, Häkkinen et al. proposed two models for the structure of Au₄₀(SR)₂₄ cluster [103]. Both of them contained a Au₂₆ core, six monomeric –RS-Au-SR– staples, and four dimeric –RS-Au-SR-Au-SR– staples. It is worth mentioning that Jiang also studied the structures of Au₄₀(SR)₂₄ based on the staple-fitness concept and found a most promising model which contains a Au₂₅ core with three monomeric staples and six dimeric staples. The alternative model with Au₂₅ core is more stable than the previously proposed model with Au₂₆ core [9].

Of note, we found that the isolated Au₄₀(SCH₂CH₂Ph)₂₄ can be converted to Au₃₈(SCH₂CH₂Ph)₂₄, which explains why the ~40 h long size-focusing product did not contain Au₄₀(SCH₂CH₂Ph)₂₄. Nevertheless, Au₄₀(SCH₂CH₂Ph)₂₄ is quite stable (though somewhat less than Au₃₈(SCH₂CH₂Ph)₂₄), and future work may reveal its structure. A comparison with the structure of Au₄₀(SCH₂CH₂Ph)₂₄ would be interesting.

The availability of atomically precise Au₃₈(SCH₂CH₂Ph)₂₄ nanoclusters provided new starting materials for the synthesis of different-sized nanoclusters via ligand exchange process [99]. Recently, the molecularly pure Au₃₆(SPh-t-Bu)₂₄ was synthesized by performing thiol-to-thiol ligand exchange with 4-tert-butylbenzenethiolate (denoted as SPh-t-Bu) using Au₃₈(SCH₂CH₂Ph)₂₄ as starting clusters. The crystal structure of Au₃₆(SPh-t-Bu)₂₄ was successfully determined by single-crystal X-ray crystallography [100]. Surprisingly, the structure of Au₃₆(SPh-t-Bu)₂₄ is significantly different from that of Au₃₈(SCH₂CH₂Ph)₂₄. The Au₃₆(SR)₂₄ cluster contains a truncated FCC tetrahedral Au₂₈ kernel. The exposed (111) and (100) facets in Au₂₈ kernel are capped by a new type of thiolate binding mode, that is, 12 of the 24 ligands bind to the underlying Au atoms on the (100) facets in a simple bridging mode, and the remaining 12 thiolates forming the known dimeric staple motifs on the (111) facets [100]. First-principle DFT calculations reveal a large HOMO-LUMO energy gap (approximately 1.7 eV, c.f. 0.9 eV for its parent cluster Au₃₈(SCH₂CH₂Ph)₂₄), which is in good agreement with the experimental result from the optical absorption spectroscopy. The extremely high stability of the nanocluster comes from the geometric structure and organization of the electronic states into superatom shells [100].

A disproportionation mechanism was further identified in the transformation of rod-like biicosahedral Au₃₈(SCH₂CH₂Ph)₂₄ to tetrahedral Au₃₆(SPh-t-Bu)₂₄ nanoclusters [99]. Figure 8 shows the detailed size-conversion pathway (roughly four stages) from the analysis of time-dependent mass spectrometry and optical spectroscopy. In stage I, the ligand exchange of Au₃₈(SCH₂CH₂Ph)₂₄ with bulkier HSPh-t-Bu occurred. A maximum of ~12 HSPh-t-Bu ligands were exchanged on the surface of Au₃₈ cluster. In stage II, the ligand exchange reaction continued, but meanwhile it started to induce structural distortion to the original Au₃₈(SR)₂₄ cluster. In the optical absorption spectroscopy, Au₃₈(SPh-t-Bu)_m(SCH₂CH₂Ph)_24–m (~12<m<~21) clusters showed some new absorption bands. In stage III, a disproportionation reaction from Au₃₈(SR)₂₄ to Au₃₆(SR)₂₄ and Au₄₀(SR)₂₆ was observed. In this process, one Au₃₈(SR)₂₄ cluster released two Au atoms and eventually generated Au₃₆(SR)₂₄. Meanwhile, another Au₃₈(SR)₂₄ cluster captured the two released Au atoms together with two free HSPh-t-Bu ligands to form Au₄₀(SR)₂₆. In the last stage, a size-focusing process accompanied by the ligand exchange rendered the pure Au₃₆(SPh-t-Bu)₂₄ clusters [99]. The conversion from rod-like biicosahedral Au₃₈(SCH₂CH₂Ph)₂₄ to tetrahedral Au₃₆(SPh-t-Bu)₂₄ nanoclusters provides a good example that the bulkiness of ligand would significantly affect the size and structure of thiolate-protected Au nanoclusters.

Fig. 8

Schematic diagram of reaction pathway for conversion of Au₃₈(SC₂H₄Ph)₂₄ to Au₃₆(SPh-t-Bu)₂₄. Stage I, ligand exchange; II, structure distortion; III, disproportionation; IV, size focusing. Adapted with permission from ref. [99].

Conclusions

This review summarized the research on one of the representative Au nanoclusters – Au₃₈(SR)₂₄, including its identification, size-focusing synthesis, structure determination, intrinsic chirality, and two size-adjacent nanoclusters. There is still tremendous science in Au₃₈(SR)₂₄ nanoclusters to be discovered in future research. As the properties of Au₃₈(SR)₂₄ continue to be fleshed out, the results summarized in this review will be useful for further analyses and applications.

Future research on the Au₃₈(SR)₂₄ cluster will focus on the application in catalysis, optics, photovoltaics, and biomedicine, etc. In catalysis, Au₃₈(SR)₂₄ clusters, including the doped ones or surface-functionalized clusters, hold great promise in asymmetric catalysis. Well-defined Au₃₈(SR)₂₄ nanocluster catalysts will lead to fundamental understanding of the origin of nanogold catalysis, especially how the intrinsic chiral surface and quantum confinement in nanoclusters affects their catalytic properties. The semiconducting electronic properties and wide absorption spectrum spanning the UV-vis-NIR range also render them great potentials in optics and photovoltaics applications. The systematic and comprehensive studies on nanocluster materials will stimulate broad scientific and technological interests.