Selective ligand exchange synthesis of Au₁₆(2-PET)₁₄ from Au₁₅(SG)₁₃

Thiolate protected gold (Au) nanoclusters have received widespread attention in the recent years, owing to their size-selective and tunable properties.^1,2 They consist of a small gold core stabilized by thiolate ligands (SR⁻), which surround the inner metal atoms by forming -SR-(Au-SR-)_x staple units.³ A series of magic-number clusters was predicted by Häkkinen and co-workers, exhibiting extraordinary stability due to their closed electron shells.⁴ Several candidates of this series have been synthesized and characterized, including Au₂₅(SR)₁₈⁻,⁵ Au₁₀₂(SR)₄₄,⁶ and Au₁₄₄(SR)₆₀.⁷ 

However, ultrasmall thiolate protected Au nanoclusters with 20 or less atoms have only scarcely been studied so far. This is related to their limited stability in the size-focusing processes taking place during synthesis,⁸ following modified Brust procedures:⁹ Typically, a gold salt precursor is dissolved and partially reduced by addition of thiols. This step is then followed by a second reduction to yield Au nanoclusters.² The clusters are initially polydisperse but can be focused to predominantly one size by reacting for a defined time interval under reducing conditions in which unstable sizes will eventually convert to more stable ones.^10,11

Several ultrasmall nanoclusters were obtained by Negishi et al., employing the water soluble tripeptide, L-glutathione (GSH), as a ligand.¹² However, the presented synthesis method yielded a size dispersion of structures consisting of between 10 and 39 Au atoms. Monodisperse samples were only accessible by post-synthetic polyacrylamide gel electrophoresis (PAGE) separation. Significant progress in selective production of small Au nanoclusters protected by water soluble thiolates was later made by Xie and co-workers, establishing selective protocols toward Au₁₅(SG)₁₃ and Au₁₈(SG)₁₄.^13,14 Nevertheless, even though these ultrasmall nanoclusters can be obtained with significant yield and purity, their reactivity has remained widely unexplored.

Over the last decade, fueled by the isolation and identification of a variety of different gold nanoclusters, understanding the structural transformations in Au nanocluster formation and reactions has become a major research focus in metal nanocluster chemistry. An emerging field is related to cluster growth schemes from simple kernel building blocks to superatom structures.^15–17 So-called ligand-exchange-induced size/structure transformations (LEISTs) are currently also intensively studied, since they present synthetic approaches toward novel cluster compositions and structures while avoiding post-synthetic separation steps.^16,18–22 During exchange reactions of thiolate-protected Au nanoclusters with thiolate ligands, Au-SR is transformed to Au-SR′, i.e., the interaction between the protecting ligand sphere and the Au kernel should remain mostly unaltered.¹⁸ However, structural transformations of the Au nanoclusters are known to occur because of differences in the properties or structures of the original and the exchange ligand.^18–21 These processes usually require the addition of a large excess of exchange ligands as well as elevated temperatures,¹⁹ though exceptions are known.²³ Several clusters obtained by the LEIST strategy have been reported so far.^18,21

The degree of changes imposed to the cluster structure is known to depend on the specific exchange conditions and may vary significantly. For example, reacting the glutathione-protecting Au₁₈(SG)₁₄ with cyclohexanethiol led to a simple structural rearrangement of the cluster core while maintaining its size.²⁴ This seems to be related to the interplay of the cluster, ligand, and solvent, clearly showing that the whole ligand exchange system has to be taken into account.²⁵ However, the number of Au atoms may also be altered upon thiol-to-thiol ligand exchange, e.g., in the growth of Au₂₅(S-But)₁₈ to Au₂₈(PPT)₂₁ (S-But: butanethiol, PPT: 2-phenylpropanethiol)²³ or from Au₁₅(SG)₁₃ to Au₁₆(S-Adm)₁₂ (S-Adm: adamantanethiol).²⁶ Reductions in cluster size are also common, for example, in the reaction of Au₂₅(2-PET)₁₈ with TBBT, yielding Au₂₂(2-PET)₄(TBBT)₁₄ (2-PET: 2-phenylethanethiol, TBBT: 4-tert-butylbenzenethiol).²⁷ Au₂₀(TBBT)₁₆ has also already been obtained from Au₂₅(2-PET)₁₈⁻,²⁸ as well as Au₃₆(TBBT)₂₄ from Au₃₈(2-PET)₂₄. Overall, a range of cluster species can be accessed through selective ligand exchange transformations, some of which are challenging to obtain by direct synthesis.¹⁹ This confirms the use of LEIST post-synthetic modifications as a synthesis tool toward novel cluster structures.

In this work, the ligand exchange of Au₁₅(SG)₁₃ precursor clusters with 2-phenylethanethiol ligands was explored, selectively yielding Au₁₆(2-PET)₁₄ clusters. The reaction was found to proceed only when a large excess (as compared to conventional ligand exchange processes) of the organic thiol was added. Moreover, a dependence on the purity of the water-soluble precursor clusters could be noted. The unexpected formation of Au₁₆(2-PET)₁₄ in this exchange is explained by its high stability against thiolate etching. The selective ligand exchange synthesis of Au₁₆(2-PET)₁₄ is yet another application of the LEIST approach, enriching the library of nanocluster transformation reactions.

To obtain the Au₁₆(2-PET)₁₄ clusters, a simple two-phase ligand exchange procedure is used, which is described in the supplementary material in more detail. First, a solution of Au₁₅(SG)₁₃ in H₂O is prepared. After addition of a solution of 2-PET (>500 eq. compared to mol precursor cluster) in dichloromethane (DCM), the mixture is stirred vigorously at room temperature. Over a few hours, the yellow color of the aqueous phase gradually vanishes, whereas the shade of orange in the previously colorless organic phase darkens (see photographs in Fig. S5). This is accompanied by a change in the UV–visible (UV–vis) absorption profile of the organic phase, which starts to show defined bands at ∼365, 420, and 485 nm [Fig. 2(b)]. Size exclusion chromatography of the ligand exchange product in the organic phase then yields mainly Au₁₆(2-PET)₁₄, with only very small amounts of smaller polydisperse clusters eluting at the end.

The nature of the exchange product was evaluated by electrospray ionization-time of flight mass spectrometry (ESI-TOFMS). As can be seen from Fig. 1, a sharp peak is observed at 4933 Da, which could be identified as [Au₁₆(2-PET)₁₃]⁺, as a comparison of the experimental isotope pattern with a simulated one shows. The spectrum further shows cluster fragments obtained by the loss of several (Au-2-PET) units from [Au₁₆(2-PET)₁₃]⁺, as well as further in-source fragmentation and gas phase chemistry products. It is worth noting that this might be the result of the relatively high concentration of the sample used for this specific measurement (as compared to the other ESI-MS measurements in this Communication) to ensure a signal-to-noise ratio sufficient for the comparison with isotope pattern fits, as only by that a definitive assignment of the nature of the exchange product was possible. The top graphic of Fig. 1 shows an enlarged view of the patterns associated with [Au₁₆(2-PET)₁₄-Me]⁺ ions (Me: Na, K, Cu). A comparison with their isotope pattern fits can be found in Fig. S7. A low intensity [Au₁₆(2-PET)₁₄]⁺ peak could be identified as well in Fig. 1. This indicates formation of a neutral Au₁₆(2-PET)₁₄ cluster species upon ligand exchange.

It should be noted that the formation of Au₂₀(2-PET)₁₆ was assumed in the beginning, since the UV–vis spectrum obtained after the exchange showed close similarities with a reported spectrum.²⁹ However, no significant signals associated with [Au₂₀(2-PET)₁₅]⁺ (5995 Da) or [Au₂₀(2-PET)₁₆-Me]⁺ adducts (6150–6200 Da) could be identified. Moreover, while Au₁₆(S-Adm)₁₂ has been reported and has even been crystallized before,^26,30 reports on a Au₁₆ cluster with 14 protecting units are scarce. Formation of Au₁₆(SG)₁₄ in their syntheses was, however, observed by Hamouda et al. with ESI-MS in negative mode, and it was determined to be of neutral charge.³¹ 

This selective transformation of Au₁₅(SG)₁₃ toward Au₁₆(2-PET)₁₄ was also unexpected because the relatively small 2-phenylethanethiol does not usually induce a change in the cluster size, even though size transformations regularly happen upon exchange with bulky exchange ligands.²⁰ Moreover, Au₁₅(SG)₁₃ is known to react with adamantanethiol to give Au₁₆(S-Adm)₁₂.²⁶ This cluster possesses the same number of Au atoms but has only 12 instead of 14 protecting ligands. The reaction proceeds via a two-phase ligand exchange process comparable to the one described in this manuscript, also employing DCM as the organic solvent. However, the attempt of Zhu and co-workers of conducting the exchange with 2-PET instead of S-Adm resulted in a featureless UV–vis absorption spectrum, indicating decomposition of the sample.²⁶ 

Therefore, to investigate the origin of this unexpected size transformation and to optimize the procedure, several parameters of the protocol were altered. Thereby, the amount of excess 2-PET added to the reaction mixture was found to have the biggest influence. Whereas the reaction proceeded within a few hours if 500 or more equivalents of 2-PET with respect to the moles of Au₁₅(SG)₁₃ were added to the reaction mixture, no reaction was observed if less than 300 equivalents were used. This was indicated by the organic phase still being completely colorless after a few days of stirring, whereas a significant amount of white precipitate had formed at the interface.

It should be noted that even though ligand exchange processes require an excess of exchange thiol to be added to the reaction, the 500–1000 eq. used for this reaction can be considered a large excess for a conventional ligand exchange maintaining the kernel size. However, this has been found to promote size transformation reactions also by other authors.^19,28,32 Au₁₆(2-PET)₁₄ was indeed found to be very stable in the presence of free thiol ligands. As shown in Fig. S1, even after stirring in an excess thiol solution for 3 days, the UV–vis profile of Au₁₆(2-PET)₁₄ shows only minor weakening of the absorption features. It is therefore assumed that the selective formation of Au₁₆(2-PET)₁₄ originates from its exceptional stability in the presence of free thiols. This also explains why no Au₁₆(2-PET)₁₄ was found by Zhu and co-workers, since significantly less thiol was used in their experiments.²⁶ Moreover, adamantanethiol, the exchange ligand used in that case, is significantly more bulky than 2-phenylethanethiol, which can likely result in different cluster structures and might explain the different Au/ligand ratios observed of the product Au₁₆ clusters.

To obtain a crude understanding of the reaction kinetics, the ligand exchange was followed by UV–vis spectroscopy [Figs. 2(a) and 2(b)] and ESI-TOFMS [Figs. S8(a) and S8(b)]. As can be seen, the characteristic absorption features of Au₁₅(SG)₁₃ at 380 and 420 nm¹³ disappeared quickly and mostly featureless spectra were obtained from then on. This was accompanied by formation of a white dispersion in the H₂O phase shortly after the start of the reaction, which disappeared again after some hours. The disappearance of the Au₁₅(SG)₁₃ clusters within the first 6 h of the reaction was also evident in the ESI-MS spectra [Fig. S8(a)]. The DCM phase was originally colorless and did not show significant UV–vis absorption. However, after 6 h, the bands associated with Au₁₆(2-PET)₁₄ were already observed, even though the solution was still almost colorless. The ESI-MS spectrum of the DCM reaction phase also already showed a prominent [Au₁₆(2-PET)₁₃]⁺ peak [Fig. S8(b)]. The features then intensified up to 15 h, after which no major change could be observed anymore. The H₂O phase was completely colorless at this point of the reaction. In addition, formation of a yellow precipitate at the interface was noted, which did not disappear anymore in the subsequent hours of stirring.

It is therefore assumed that the product cluster Au₁₆(2-PET)₁₄ had already been formed upon transition of the cluster from the aqueous to the DCM phase, since no indication of a polydisperse reaction mixture could be seen from the UV–vis spectra of the organic phase. The size transformation and focusing process thus seems to occur at the reaction interface. This is in agreement with the general understanding of LEIST processes of Au nanoclusters, which suggests a reconstruction mechanism rather than decomposition and subsequent reformation of cluster species.^19,32 However, further studies would be required to elucidate the exact nature of the structural modifications, which are beyond the scope of this Communication.

To elucidate the role of Au₁₅(SG)₁₃ in this reaction, the ligand exchange was also attempted with a polydisperse mixture of water-soluble Au_x(SG)_y nanoclusters. However, even though a color transfer to the DCM phase was noticed, the UV–vis spectrum showed no sign of the formation of Au₁₆(2-PET)₁₄ (Fig. S3). This indicates that the structural transformation might originate from the building blocks already present in the Au₁₅(SG)₁₃ structure.

Moreover, also a “one-pot ligand exchange synthesis” approach was tested, for which the Au₁₅(SG)₁₃ was not isolated first and then only subsequently subjected to the ligand exchange reaction. Therefore, the Au₁₅(SG)₁₃ was set up in H₂O and toluene as described by Yao et al.,¹³ and after the reduction process was completed, 2-PET was directly added to the toluene phase and the stirring continued until complete color transfer to the organic phase was observed [Fig. S2(a)]. However, even though some Au₁₆(2-PET)₁₄ could be isolated after size exclusion chromatography, also polydisperse clusters were found [Fig. S2(b)]. Compared to the ligand exchange starting from isolated Au₁₅(SG)₁₃, the yield was significantly lower and the process less selective.

Further attempts were made by adjusting the solvent mixture by addition of methanol (MeOH) to tune the polarity. The influence of the pH of the aqueous phase was also investigated, since it plays a crucial role in the synthesis of Au₁₅(SG)₁₃.¹³ Therefore, the mixture was acidified to pH = 2.0 by the addition of 1 N HNO₃ (pH was ∼4.2 before). However, neither of these seemed to have a significant influence on the reaction dynamics.

Complementary structural information of the clusters was obtained by attenuated total reflection-Fourier transform infrared spectroscopy (ATR-FTIR) by probing the characteristic vibrations of the ligands at mid (MIR) and the Au–S vibrations at far-infrared (FIR). The spectra of the clusters before and after the ligand exchange are shown in Fig. 3. The initial glutathione ligand of Au₁₅(SG)₁₃ features intense bands in the amide I and II regions and a broad band centered at 3300 cm⁻¹ related to OH and NH₃⁺ stretching vibrations [Fig. 3(a), blue line].^33,34 The broadness of the bands can be explained by the presence of residual H₂O due to the hygroscopic nature of the clusters. In contrast, after the ligand exchange (Au₁₆(2-PET)₁₄), only vibrations associated with the 2-PET ligand were detected, i.e., vibrations of the aromatic ring (1600–1450 cm⁻¹).³⁵ This confirms that all original GSH ligands were replaced by 2-PET. The absence of a band associated with the S–H stretch (2500 cm⁻¹)³⁵ confirmed that no free ligands were present in the sample.

The two distinctly different ligands before and after ligand exchange also exhibit different vibrations in the far-infrared energy range [Fig. 3(b)]. In this region, Au–S vibrations related to different staple configurations can be observed, based on previous experimental and theoretical studies.^36–38 Depending on the nature of the Au–S vibration (tangential or radial), Au_core–S_staple or Au_staple–S_staple vibrations can be distinguished,^36,37 in some cases even the length of the staple motif.³⁹ The range below 110 cm⁻¹ is associated with weak Au–S–Au bending modes and bands above 300 cm⁻¹ may be assigned to vibrations of Au and S both located inside the staple, whereas bands between 200 and 300 cm⁻¹ may originate from Au_core–S_staple.³⁶ 

The two clusters investigated in this work, Au₁₅(SG)₁₃ and Au₁₆(2-PET)₁₄ [Fig. 3(b)], showed quite different spectra, denoting the structural evolution. The precursor nanocluster Au₁₅(SG)₁₃ with GSH ligands showed an intense band at 348 cm⁻¹, which could be due to Au_staple–S_staple vibrations. Below 300 cm⁻¹, a broad band centered at 167 cm⁻¹ prevents further identification of potential Au–S modes. Overall, the spectrum matches the one of Au₂₅(SG)₁₈ clusters reported by Valušová et al.³³ After the ligand exchange, characteristic FIR vibrations of the 2-PET ligand are observed at 564 and 492 cm⁻¹.³⁷ Moreover, a band appeared at 319 cm⁻¹, associated with Au–S vibrations of both atoms inside a staple. A weak band at 283 cm⁻¹ can also be identified, which can be assigned to tangential or radial Au–S vibrations.³⁷ 

Furthermore, extended x-ray absorption fine-structure (EXAFS) studies were performed to inspect the structural differences from the starting Au₁₅(GSH)₁₃ cluster to the one obtained after ligand exchanged (Au₁₆(2-PET)₁₄). The x-ray absorption near edge structure (XANES) spectrum of Au₁₆(2-PET)₁₄ shows a small increase in white line intensity (most prominent at 11 923 eV) compared to Au₁₅(GSH)₁₃ [Fig. 4(a)], which is usually related to a difference in the valence electron occupancy.⁴⁰ This would indicate an increase in the valence state from Au₁₅ to Au₁₆, as the higher Au/S ratio leads to more unoccupied d orbitals to which electrons can be excited, resulting in higher white line intensity. However, Yamazoe and co-workers reported that the change in white line intensity can also be related to differences in the nature of the Au–S bonds as well as due to different types of ligands.⁴⁰ This is also denoted in the R space spectrum [Fig. 4(b)], which shows a small increase in the intensity of the scattering peak at 2 Å (Au–S bond). The following peak at around 2.5 Å originating from the Au–Au bonds was also increasing, which could denote the increase of the Au cluster kernel.⁴¹ Au₁₅(SG)₁₃ has been predicted to possess a Au₄ kernel,^42,43 whereas the resolved crystal structure of Au₁₆(S-Adm)₁₂ showed a Au₇ kernel.²⁶ However, it should be noted that it is unclear if the Au₁₆(2-PET)₁₄ obtained in this study would possess the same kernel structure, considering that it has two additional protecting thiolate ligands. Similar cluster kernels before and after exchange would explain the similarity of their EXAFS spectra.

Both clusters, Au₁₅(SG)₁₃ and Au₁₆(2-PET)₁₄, also show photoluminescence (PL) in the near-infrared region, as can be seen in Fig. 5. The discrete energy levels, which are characteristic of the nanoclusters, result in molecular-like optical properties and enable luminescence, though usually with bands in the red wavelength region.⁴⁴ To further characterize the properties of the obtained Au₁₆(2-PET)₁₄ nanoclusters and to compare them to those of the initial Au₁₅(SG)₁₃, a photoluminescence analysis was performed. The PL spectra of both Au₁₅(SG)₁₃ and Au₁₆(2-PET)₁₄ shown in Fig. 5 exhibit a broad band centered at around 670 nm and another maximum above 800 nm, in agreement with published spectra.^13,30,45 The dual fluorescence profile of both clusters indicates electron donation from the ligands when excited, as determined by previous studies.⁴⁶ Slight differences were noticed when measuring the emission decay (probed at 650 and 700 nm; see Fig. S9) after excitation at 375 nm: The lifetimes found for the Au₁₅(SG)₁₃ nanoclusters were a bit longer at both emission wavelengths (12.2 vs 9.6 ns at 650 nm and 14.5 vs 11.3 ns at 700 nm), which seems to be related to the electronegative substituents of the GSH ligand.⁴⁶ This further confirms the influence of the protecting ligand on the cluster properties and the importance of selective ligand engineering, for example, by ligand exchange.

In summary, this work reports on the synthesis of Au₁₆(2-PET)₁₄ by ligand exchange from Au₁₅(SG)₁₃. The selective size transformation occurred only upon addition of a sufficiently large excess of 2-phenylethanethiol. The obtained Au₁₆(2-PET)₁₄ is exceptionally stable in the presence of free thiols in solution, preserving its optical absorption features even after days of exposure. Complete replacement of the protecting ligand sphere was confirmed by ESI-TOFMS and ATR-FTIR, whereas EXFAS measurements showed that both clusters seem to bear structural and electronic similarities. Moreover, they both show photoemission in the near-infrared region. Overall, this application of the LEIST methodology presents a feasible synthesis approach toward Au₁₆(2-PET)₁₄, one of the smallest thiolate-protected nanoclusters isolated so far.

The supplementary material contains all experimental details, additional UV–vis of the ligand exchanges, and photographs showing the phase transfer of the clusters through the reaction, as well as MS and additional PL spectra.

The staff of CLAESS beamline at ALBA synchrotron, especially Dr. Wojciech Olszewski and Dr. Carlo Marini, are thanked for help in acquiring the EXFAS spectra reported in this manuscript (Proposal No. 2017092492). EQ VIBT GmbH and the BOKU Core Facility Mass Spectrometry are acknowledged for providing mass spectrometry instrumentation. We acknowledge support from the Austrian Science Fund (FWF) via grants Single Atom Catalysis (Grant No. I 4434-N) and Elise Richter (Grant No. V831-N). The authors acknowledge TU Wien Bibliothek for financial support through its Open Access Funding Programme.

The authors have no conflicts to disclose.

The synthesis of the Au nanoclusters and the exchange reactions as well as characterization were performed by V.T. and S.P. ESI-MS was measured by H.D. Photoluminescence measurements were conducted by S.P.N. and D.E. Data evaluation and manuscript preparation were performed by V.T. and N.B., with contributions from all authors. Funding was acquired by N.B. and G.R.

The data that support the findings of this study are available from the corresponding author upon reasonable request.