Introduction

Structural isomerism in organic molecules is a common occurrence due to the bonding diversity of carbon. However, for nanoscale or even larger scale materials, experimental observation of structural isomerism has been largely impeded by the challenge of unravelling the intrinsic structure at the atomic level1. Nevertheless, theoretical and experimental efforts2,3,4,5,6,7,8 in searching for structural isomerism in such materials continue, because such a finding would provide precise and insightful structure–property correlations and meaningful guidance for designing and synthesizing unique functional materials. The recently developed ultrasmall, thiolated metal nanoparticles (also called nanoclusters) provide opportunities for investigating structural isomerism, as they can now be controlled with atomic precision9,10,11,12,13,14,15,16,17,18,19,20 and their structures can be resolved by single-crystal X-ray crystallography (SCXC) as well. To date, the structures of a series of thiolated metal nanoparticles with various sizes have been elucidated experimentally and theoretically21,22,23,24,25,26,27,28; however, to the best of our knowledge, no structural isomerism in thiolated metal nanoparticles has been reported, albeit Au24(SCH2Ph-tBu)20 and Au24(SePh)20 were revealed to have different Au24 core structures29,30. In a strict sense, Au24(SCH2Ph-tBu)20 and Au24(SePh)20 are not structural isomers, since their ligands are different. Thus, structural isomerism in thiolated nanoparticles remains a mystery.

In the current work, using a modified synthesis method for Au25, we synthesize a nanocluster, whose composition is determined to be the same as that of the previously reported Au38(PET)24 (refs 31, 32, 33) (PET, phenylethanethiolate), as evidenced by electrospray ionization mass spectrometry (ESI–MS) in combination with X-ray photoelectron spectroscopy (XPS) and thermogravimetric analysis (TGA). SCXC reveals that the structure of this nanocluster is different from that of the previously reported structure24. To differentiate the two structures, the previous Au38 is denoted as Au38Q and our nanocluster is denoted as Au38T (where Q and T are the surname initial of the first author of the previous and current work, respectively). Au38T and Au38Q are therefore structural isomers and they represent the first pair of structural isomers in nanoparticles as revealed by SCXC, to the best of our knowledge. The two isomers exhibit distinctly different optical, stability and catalytic properties, and the less stable Au38T can be irreversibly transformed to the more stable Au38Q at 50 °C.

Results

Characterization

Au38T was synthesized using a modified one-pot method34 and isolated using preparative thin-layer chromatography (PTLC)35,36. ESI–MS was employed to determine the exact molecular mass and formula of the novel nanoparticle (note: caesium acetate was added to form positively charged adducts). Three distinct peaks centred at m/z 10910.186, 7317.312 and 5522.217 were observed in the mass spectrum (Fig. 1a). The peaks at m/z 10910.186 and 5522.217 (almost half of 10910.186) can be readily assigned to [Au38(PET)24Cs]+ (theoretical m/z value: 10910.658; deviation: 0.472) and [Au38(PET)24Cs2]2+ (theoretical m/z value: 5522.772; deviation: 0.555), respectively. The peak at m/z 7317.312 can be assigned to Au26(PET)16 (theoretical m/z value: 7316.818; deviation: 0.494), which could be a fragment of Au38(PET)24, because the nanoparticles are monodisperse, as demonstrated by TLC, and it is also observed in the ESI spectrum of Au38Q (see below). Based on the ESI–MS results, it is concluded that the as-prepared nanoparticle is neutral, and that its composition is Au38(PET)24, which is also corroborated by the TGA and XPS. TGA shows a weight loss of 30.39 wt% (Fig. 1b), corresponding to the theoretical loss of 30.55 wt% according to the formula. No other elements (including Cl, Br, N and Na) was detected by XPS (Fig. 1c), which excludes the possibility of existence of potential counterions such as Cl, Br, [N(C8H17)4]+ and Na+; thus, the as-prepared nanoparticle is neutral. Quantitative measurement reveals that the Au/S atomic ratio is 38.0:24.3 (Supplementary Figs 3 and 4), in good agreement with the expected ratio (38.0:24.0) for the composition of Au38(PET)24. Thus, the formula is identical to that of the nanoparticle previously reported in ref. 31; however, the absorption spectrum of our nanoparticle distinctly differs from that of the previous nanoparticle. The ultraviolet–visible–near-infrared spectrum of the novel Au38(PET)24 (abbreviated as Au38T) shows six absorption peaks at 505 nm (ɛ: 3.86 gcm l−1), 540 nm (ɛ: 3.22 gcm l−1), 610 nm (ɛ: 1.46 gcm l−1), 700 nm (ɛ: 0.69 gcm l−1), 880 nm and 1,090 nm (Fig. 1d and Supplementary Fig. 2). The previous Au38(PET)24 (abbreviated as Au38Q) shows six absorption peaks centred at 480 nm (ɛ: 4.62 gcm l−1), 520 nm (ɛ: 3.72 gcm l−1), 570 nm (ɛ: 2.86 gcm l−1), 627 nm (ɛ: 2.59 gcm l−1), 740 nm (ɛ: 0.58 gcm l−1) and 1,035 nm (Fig. 1d and Supplementary Fig.1). TLC also indicates that they are not the same nanoparticle (Fig. 1d, inset). Indeed, they are a pair of structural isomers (vide infra).

Figure 1: Characteriztion of Au38T.
figure 1

(a) ESI mass spectrum of the Au38T. (b) TGA of Au38T. (c) XPS spectrum of Au38T. (d) Ultraviolet–visible–near-infrared absorption spectra of Au38T (blue) and Au38Q (black) in toluene (measurement temperature: 0 °C). Insets are the photo of thin-layer chromatography, and enlarged absorption spectra in the range from 800 to 1,200 nm of Au38T and Au38Q.

Atomic structure

The structure of the previous Au38Q was determined by SCXC and it has a core-shell structure consisting of a face-fused bi-icosahedral Au23 core, which is capped by a second shell composed of the remaining 15 gold atoms (Fig. 2f). To confirm that our nanoparticle (Au38T) is an isomer of Au38Q, we grew high-quality single crystals and successfully elucidated the structure via SCXC. Briefly, the new structure of Au38T is composed of one Au23 core and one mixed capping layer of thiolate ligands and gold–thiolate complex units. The Au23 core consists of one icosahedral Au13 and one Au10 unit, and the mixed surface layer contains two Au3(SR)4 staple units, three Au2(SR)3 staple units, three Au1(SR)2 staple units and one bridging thiolate SR ligand. The anatomy of the Au38T structure starts with the central Au23 core (Fig. 2b), which can be viewed as one Au12 cap and one Au13 icosahedron (Fig. 2a) fused together via sharing two gold atoms (Fig. 2b, dark green gold atoms), which is in distinct contrast with the case of Au38Q; for the latter, the two Au13 icosahedra are fused together via sharing a face (three gold atoms) to form a bi-icosahedral Au23 core. The Au12 cap is composed of three tetrahedra and the Au–Au bond lengths in each tetrahedron range from 2.71 to 2.88 Å. In the Au13 icosahedron, the Au–Au bond lengths between the central atom and the shell Au atoms (except for the two shared gold atoms) vary from 2.71 to 2.82 Å. The bond lengths between the two shared gold atoms and the central atom of Au13 icosahedron are 2.77 and 2.78 Å, respectively. The different Au23 core in Au38T (in contrast to the biicosahedral Au23 core of Au38Q) leads to various surface-binding structures. The Au23 core in our case was capped by two Au3(SR)4 units and two Au(SR)2 units, and the average Au–S bond lengths/Au–S–Au bond angles were 2.33 Å/96.47° and 2.32 Å/94.43° in the Au3(SR)4 and Au(SR)2 staple units, respectively (Fig. 2c). Interestingly, in addition to the two Au3(SR)4 units and two Au(SR)2 units, one bridging thiolate (SR) is also found to link the Au13 icosahedron and the Au12 cap (Fig. 2c, the sulfur atom is marked in red), the two Au–S bond lengths are 2.33 and 2.30 Å, respectively, and the Au–S–Au bond angle is 92.66°. It is noteworthy that in Au38Q, the Au23 core was protected by six Au2(SR)3 and three Au(SR)2 staple units; no comparable Au3(SR)4 staple units and bridging thiolate (SR) was observed. The Au13 icosahedron in Au38T is exclusively capped by two Au2(SR)3 staple units (the average Au–S bond lengths in the Au2(SR)3 staple units are 2.35 and 2.34 Å, respectively, and the average Au–S–Au bond angles in the Au2(SR)3 staple units are 92.20° and 90.33°, respectively) and the Au12 cap is capped by one Au2(SR)3 staple unit (the average Au–S bond length is 2.33 Å and the average Au–S–Au bond angle is 97.03°). In addition, the Au12 cap is also capped by one Au(SR)2 staple unit, and the average Au–S bond length/Au–S–Au bond angle are 2.33 Å and 99.91° (Fig. 2d). The structure resolved by X-ray diffraction was further analysed by computations: the simulated ultraviolet–visible–near-infrared spectrum is close to the experimental one (Fig. 3).

Figure 2: Structures of Au38T and Au38Q.
figure 2

(a) Anatomy of the Au23 core, which consists of a Au12 cap unit and Au13 icosahedral unit. (b) Au23 core, which is constructed by one Au12 unit and one Au13 unit sharing two gold atoms. (c) Two Au3(SR)4, two Au(SR)2 and one SR linking the Au12 cap and Au13 icosahedron. (d) Three Au2(SR)3 and one Au(SR)2 protecting the Au23 core. (e) Back view of Au38T. (f) The Au38Q structure.

Figure 3: Comparison of ultraviolet–visible–near-infrared absorption spectra of Au38T.
figure 3

Blue: experimental; black: calculated by time-dependent density function (TDDFT) method. Inset is the enlarged spectra in the range from 800 to 1,200 nm.

As discussed above, the structure of Au38T is remarkably different from that of Au38Q and the main differences between the two structures lie in the type of Au23 core and the surface capping mode of the Au23 core. Au38T and Au38Q have an identical composition but completely different structures; thus, they are literally a pair of structural isomers. Notably, the structure of Au38T reported in this work is novel and also differs from those theoretical structures predicted by Hakkinen et al.37, Tsukuda and colleagues38, Jiang et al.39 and Zeng and colleagues40, among others.

Transformation

Au38T exhibits relatively high stability at low temperatures, as no obvious spectral changes was detected when a solution of Au38T was stored at −10 °C for as long as 1 month in toluene (Fig. 4a). However, the absorption spectrum of Au38T gradually changed to that of Au38Q at 50 °C in toluene (Fig. 4b), which indicates that Au38T can transform to Au38Q at elevated temperatures. TLC and ESI–MS further support this transformation (Fig. 4b (inset) and Fig. 4c). However, the reverse transformation (that is, from Au38Q to Au38T) was not successful under various investigated conditions. These results indicate that Au38T is less stable than Au38Q, and that Au38T can only be irreversibly transformed to Au38Q. The reason for why the relatively unstable Au38T is formed rather than the stable Au38Q during the synthesis is probably because the former is kinetically favourable in our reaction conditions, similar to some previous reports41,42.

Figure 4: Difference in stability and catalysis between Au38T and Au38Q.
figure 4

(a) Time-dependent ultraviolet–visible–near-infrared absorption spectra of Au38T at −10 °C in toluene. (b) Ultraviolet–visible–near-infrared absorption spectral transformation at 50 °C in toluene (the isosbestic points are at 360 and 700 nm). Inset: thin-layer chromatography of Au38T before and after the transformation. (c) ESI mass spectrum of the transformed product. (d) Catalytic activities of Au25, Au38T and Au38Q.

Catalysis

Au38T exhibits remarkably higher catalytic activity than Au38Q at low temperature (for example, 0 °C) in reduction reactions. For example, 4-nitrophenol can be reduced to 4-aminophenol in 44% yield with 0.1 mol% Au38T catalyst in half an hour, whereas no reduction occurred when Au25(PET)18TOA+ (Au25 for short, TOA+: tetra-n-octylammonium) or Au38Q was used as the catalyst (Fig. 4d and Supplementary Fig. 5) under the same reaction conditions. It is noteworthy that in other cases, Au25 was reported to exhibit good catalytic reduction activity43,44. The high catalytic activity of Au38T may be due to its surface being not as densely protected as the surfaces of Au25 and Au38Q; further investigation is underway. A previous work44 implied that the catalytic properties of gold nanoclusters are not only size dependent but also structure sensitive. However, the structure dependence of catalytic properties was unclear at that time, because the ligands were different in Au44(PET)32 and Au44(TBBT)28 (TBBT: 4-tert-butylbenzenethiolate), and the ligand effect should be considered. Herein, it is unambiguously demonstrated that the structure effect indeed exists, because Au38T and Au38Q exhibit remarkably different catalytic performance. Au38T is relatively robust and can retain its ultraviolet–visible–near-infrared spectrum even after 18 catalytic cycles (Supplementary Fig. 6). It is noteworthy that the gradual decrease in the yield is primarily due to the unavoidable mass loss of the catalyst during the isolation by column chromatography. However, after 21 cycles, the catalyst transformed to more stable Au38Q demonstrated by the ultraviolet–visible–near-infrared spectra and accordingly the loss of catalytic activity (see Supplementary Table 1). The high catalytic activity at low temperatures indicates the potential application of Au38T in some catalytic processes.

Discussion

In summary, we have discovered a pair of structural isomers Au38T and Au38Q, which were identified using ESI–MS, TGA, XPS and SCXC. Although both species have the same composition (that is, Au38(PET)24), they have distinctly different structures, which results in differences in their optical and catalytic properties, as well as structural stability. The less stable Au38T can be irreversibly transformed to the more stable Au38Q at high temperatures. The structure of Au38T is very interesting: it is composed of a Au23 core (fused by one Au13 icosahedron and one Au12 cap by sharing two atoms) and a mixed layer of thiolate ligands and gold–thiolate complex units for surface protection. This structure is unique (that is, not found in other reported gold nanoclusters). In particular, the diversity of staple units and the bridging thiolate found in Au38T provide a new direction for structural studies of metal nanoclusters. The significance and novelty of this work are as follows. (i) A novel synthesis method is developed, with which a novel gold nanoparticle is readily synthesized, and the composition of the as-prepared nanoparticle is precisely determined using ESI–MS in conjunction with XPS and TGA. (ii) The structure of Au38T is resolved using SCXC and the unique structural features provide important implications for nanocluster structural studies. (iii) Significantly, structural isomerism is observed in nanoparticles for the first time. (iv) The distinctly different properties (in particular the catalytic properties) of the two structural isomers indicate a structure–property correlation and this will have important implications for future catalytic studies. It is expected that our work may motivate more studies on structural isomerism and structure–property correlations in nanoscale or even larger scale materials.

Methods

Reagents

All chemicals and reagents are commercially available and were used as received. Tetraoctylammonium bromide (TOAB, 98.0%) and 4-nitrophenol (99.0%) were obtained from Aladdin; 2-phenylethanethiol (PhC2H4SH, 99.0%) was purchased from Sigma-Aldrich; Au38Q and Au25(PET)18TOA+ were synthesized following reported methods31,34.

Synthesis

Au38T was synthesized using a modified one-pot method and separated using PTLC. Briefly, HAuCl4·4H2O (0.20 g, 0.48 mmol) was mixed with 1.03 equivalents of TOAB (0.27 g, 0.49 mmol) in CH2Cl2 (40 ml). Then, 9 equivalents of phenylethanethiol (0.61 ml, 4.50 mmol) were added to this solution and the solution was stirred for 2 h until it became colourless. To this solution, 5.3 equivalents of NaBH4 (0.11 g, 2.80 mmol) in cold water (5 ml) was added in one shot under vigorous stirring and the reaction was allowed to proceed under constant stirring for 6 h. CH2Cl2 was removed via rotary evaporation at 20 °C to isolate the crude product. For purification, the crude product was extracted with a small amount of tetrahydrofuran (THF) and washed with ice water three times and with CH3OH two times; during this procedure, traces of inorganic salt, excess TOAB and phenylethanethiol were thoroughly removed. Next, the as-obtained crude products were separated using PTLC (dichloromethane: petroleum ether=3:4) and finally the target product was isolated from the reddish brown band of PTLC after extraction with CH2Cl2, with a yield of 5%.

Characterization

The ultraviolet–visible–near-infrared absorption spectrum was measured on a UV-3600 spectrophotometer (Shimadzu, Japan) at room temperature. TGA analysis was conducted under a N2 atmosphere (3 mg sample used, flow rate 50 ml min−1) on a TG/DTA 6300 analyzer (Seiko Instruments, Inc.) and the heating rate was 10 °C min−1. XPS measurements were performed on an ESCALAB 250Xi XPS spectrometer (Thermo Scientific, USA), using a monochromated Al Kα source and equipped with an Ar+ ion sputtering gun. All binding energies were calibrated using the C (1 s) carbon peak (284.8 eV). ESI–MS data were acquired on a Waters Q-TOF mass spectrometer equipped with a Z-spray source. The sample was dissolved in toluene (1 mg ml−1) and diluted 1:1 in dry ethanol (5 mM CsOAc). The sample was directly infused at 5 μl min−1. The source temperature was fixed at 70 °C. The spray voltage was set at 2.20 kV and the cone voltage was set at 60 V.

Single-crystal growth and analysis

Black crystals were formed from a CH2Cl2/hexane solution of the nanoclusters at 4 °C after 5 days. The diffraction data for Au38(PET)24 were collected at 173 K on a Bruker APEX DUO X-ray diffractometer using Cu Kα radiation (λ=1.54184 Å).

Theoretical methods

All calculations were performed using density functional theory with the pure functional Perdew-Burke-Ernzerhof45,46 and the all electron basis set 6–31 g (d, p) for H, S, pseudopotential basis set LANL2DZ for Au, as implemented in the Gaussian 09 program package47. Time-dependent density functional calculations48 were performed to reproduce the experimental ultraviolet–visible spectrum and –R group was replaced by –H to minimize computational work31. The Gaussian half-width at half-height of 0.15 eV in the Multiwfn software49 was used to simulate the ultraviolet–visible spectrum.

General procedure for the catalyses

4-Nitrophenol (69.50 mg, 0.500 mmol), Au38Q, Au25 or Au38T (0.100 mol%, not adsorbed on a support or calcined) and THF (5 ml) were mixed in a reaction tube at 0 °C. The mixture was stirred at this temperature for 5 min. NaBH4 (189.00 mg, 5.000 mmol) dissolved in 1.0 ml of H2O was added slowly to the mixture. After stirring at 0 °C for 30 min, a large amount of water was added to quench the reaction. The mixture was extracted with dichloromethane twice (2 × 10 ml) and then the organic layers were collected and concentrated. The reduction product (4-aminophenol) was purified by column chromatography on silica gel, with ethyl acetate and petroleum ether (ethyl acetate/petroleum ether=1/1) as the eluant.

General procedure for the recovery of Au38T

When the reduction was completed, the reaction mixture was quenched with water. Au38T and other organic compounds were extracted with dichloromethane. The extract was collected and concentrated. After the other organic compounds were isolated by column chromatography with ethyl acetate and petroleum ether, Au38T was recovered using dichloromethane as the eluant. The dichloromethane was evaporated under reduced pressure and then Au38T was re-used in the next cycle without further treatment.

Additional information

Accession codes: The X-ray crystallographic coordinates for structures reported in this study (see Supplementary Table 2 and Supplementary Data 1) have been deposited at the Cambridge Crystallographic Data Centre (CCDC), under deposition number CCDC 1423153. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

How to cite this article: Tian, S. et al. Structural isomerism in gold nanoparticles revealed by X-ray crystallography. Nat. Commun. 6:8667 doi: 10.1038/ncomms9667 (2015).