Hollow nanoclusters are intriguing for bridging the gap between small metal-organic cages and hollow nanoparticles and provide enormous opportunities for functional materials. Rational construction of hollow nanoclusters remains a challenge owing to the poor understanding of their formation pathways and difficulties in obtaining the intermediate structures. Here we report a catalyzed assembly strategy to construct hollow silver–sulfide clusters using a self-releasable carbonate template. The hollow Ag56 cluster and the carbonate-templated Ag33 intermediate are characterized by single-crystal X-ray analysis. Combining this data with a time-dependent UV-Vis spectroscopic investigation, we reveal the catalyst-like behavior of the self-releasable carbonate template and propose the corresponding mechanism. The assembly of hollow clusters using a self-releasable template may provide insight into understanding the formation mechanisms of other hollow nanoclusters and facilitate the design and construction of new hollow nanoclusters.
Noble metal nanoclusters1,2,3,4 have experienced great advances in the past decade as an important bridge between smaller metal-organic coordination complexes5,6,7 and larger ligand capped nanoparticles8,9, providing enormous opportunities on magnetism1, catalysis4,10,11, luminescent materials12,13,14,15, biomedicine16,17, and molecular sensing15,18. As a prime example, thiolate-protected silver18,19,20,21,22,23, gold3,12,24, and the bimetallic nanoclusters25,26,27 have been studied extensively for understanding their size-dependent evolutions of structures and properties. Because atomically precise nanoclusters have the same monodisperse nature as small coordination molecules1,24, they can be analyzed by single-crystal X-ray crystallography, revealing various sophisticated structures with many kinds of symmetries28,29,30,31. Among these structures, hollow nanoclusters32,33,34,35,36 are particularly intriguing for bridging the gap between small supramolecular cages5,6,7 and hollow nanoparticles37,38, which presents increased structural complexity and may bring more sophisticated functions. However, constructing hollow nanoclusters and precisely determining their structures remain great challenges39,40,41,42, because of the lack of understanding of the formation pathway of nanoclusters28,43,44,45 and the inherent difficulties in getting high-quality single crystals of hollow structures46. To understand the formation mechanism of hollow nanoclusters, it is crucial to track growth kinetics and obtain single-crystal structures of the intermediate clusters. Single crystals of intermediate clusters can be hard to obtain, and other characterization methods, such as advanced mass spectrometry, also could be used to monitor the growth mechanism of nanoclusters and to measure precise intermediate molecular formulas16,17.
Here, we report a catalyzed assembly strategy to construct a hollow silver-sulfide cluster, Ag56S12(StBu)20(Sal)12 (1, HStBu = tert-Butylthiol, H(Sal) = salicylic acid), using a self-releasable anion template. This hollow Ag56 cluster is synthesized from small precursors, whereas a carbonate-templated Ag33 nest-like intermediate is found during the growth of Ag56 cluster. Structures of Ag56 cluster and Ag33 intermediate are characterized by single-crystal X-ray analysis, and the growth kinetics are investigated by time-dependent ultraviolet (UV)–visible (Vis) spectroscopy, revealing the catalyst-like behavior of the self-releasable carbonate template.
Structure of 1
Cluster 1 was synthesized by adding silver salicylate hydrate (Ag(Sal)·H2O, 1.2 mmol) and AgStBu (4.5 mmol) into a mixture of acetonitrile, dichloromethane, and methanol (v/v = 1:1:1). The reaction proceeded at 80 °C for 5 days and was directly reflected by the color change from colorless to yellow and finally to orange. After reaction, orange crystals were obtained and analyzed by single-crystal XRD (Supplementary Table 1), revealing a neutral spherical Ag56S12(StBu)20(Sal)12 structure of 1 (Fig. 1a and Supplementary Fig. 1). Cluster 1 have a core–shell structure containing a [Ag14S12] core (Fig. 1b) and a [Ag42(StBu)20(Sal)12] shell (Fig. 1c).
In the [Ag14S12] core, 14 Ag+ ions form a hollow rhombic dodecahedron, whereas 12 S2– ions are located outside each face of Ag+ rhombic dodecahedron to form an icosahedron. The rhombic dodecahedron@icosahedron geometry of the [Ag14S12] core resembles the [Ag14(C≡CtBu)12Cl] and [Ag14S13] cores in the reported clusters47,48,49. However, all these previously reported clusters encapsulated a guest anion (Cl– or S2–) at the center of [Ag14] rhombic dodecahedron.
In the [Ag42(StBu)20(Sal)12] shell, Ag+ ions can be classified into two types: 18 Ag+ ions that are only protected by thiolates, and the other 24 Ag+ ions that are protected by both thiolate and salicylate ligands. The 20 StBu ligands form a dodecahedron and coordinate with Ag+ in μ3 and μ4 bridge binding modes, with the Ag–S bond lengths ranging from 2.419 to 2.944 Å. The 12 Sal ligands form an icosahedron and coordinate with Ag+ in μ2-O,O’ binding mode, with Ag–O bond lengths ranging from 2.257 to 2.357 Å. Therefore, cluster 1 can be resolved as five layers of polyhedral from inner to outer surface: the Ag14 rhombic dodecahedron, the S12 icosahedron, the spherical Ag42 polyhedron, the (StBu)20 dodecahedron and the Sal12 icosahedron. All these polyhedra share a common S6 symmetry axis, making the cluster achiral. In crystal, spherical clusters are packed in the R-3c space group with the S6 axis of the cluster in parallel to the c axis of the unit cell. The diameter of the spherical 1 approximately equals the lattice parameter a (27.1 Å), and the volume of one cluster is 10,436 Å3.
Structure of 2
During the synthesis of 1, we observed an obvious color change from colorless to yellow and finally to orange (Supplementary Fig. 2). Therefore, we speculated that the synthetic pathway of 1 could go through an intermediate stage with the formation of a smaller yellow cluster. By carefully crystallizing the reaction solution at intermediate stage, we obtained a small amount of yellow hexagonal crystals of CO3Ag33S3(StBu)16(Sal)9 (2) concomitantly with orange rhombic crystals of 1. Cluster 2 has a nest-like structure having a C3v-symmetry (Fig. 2 and Supplementary Fig. 3), crystallized in a P-3m1 unit cell. The nest-like structure of 2 is similar to the reported NO3−-templated Ag33 cluster13. However due to the charge difference between CO32− and NO3−, cluster 2 is neutral without any counter-anion outside the cluster. The 33 Ag+ ions in 2 can be identified as two layers: the inner layer contains 9 Ag+ ions (green in Fig. 2), holding the center CO32− anion; and the outer layer contains 24 Ag+ ions (magenta in Fig. 2), which are connected to the inner layer by three S2− ions (orange in Fig. 2) and six StBu ligands (yellow in Fig. 2). The Ag–O bond-length between the center CO32− anion and the Ag+ inner layer was measured as 2.7 Å, which is longer than the 2.3 Å Ag–O bond in Ag2CO350, suggesting the relatively weaker interaction between the central CO32− anion and the Ag+ inner layer. It was surprising that this intermediate 2 contains a CO32− anion template, as no CO32− species was intentionally added into the reaction. We speculated that the CO32− was introduced from the atmospheric carbon dioxide fixation51,52. To confirm the reaction pathway and understand the reaction kinetics, we added a small amount of Ag2CO3 (4% molar ratio to AgStBu) into the previously described reaction for synthesizing 1. Two parallel reactions with and without Ag2CO3 were investigated by time-dependent UV–Vis spectroscopy.
The UV–Vis spectra of the clusters (Fig. 3a) corroborate the observed color change from yellow 2 to orange 1. Compared to the solution of reactants, the smaller 2 shows a peak of absorbance increase at 370 nm, whereas larger 1 shows a wide increase in UV–Vis absorption from 300 to 550 nm (Fig. 3b). Two parallel reactions with and without Ag2CO3 were monitored by UV–Vis spectroscopy at 1 h time intervals. As shown in Fig. 3c, the UV–Vis spectra of the reaction solution without Ag2CO3 increased obviously upon reaction time. By comparing with the initial spectrum, we derived differential spectra for different reaction time (Fig. 3d), showing an increasing peak at 370 nm in the first 4 h, which corresponds to the formation of 2. After 5 h, the differential spectra get broader from 300 to 550 nm with a new increasing peak at 330 nm, corresponding to the formation of 2. For the reaction assisted by Ag2CO3, a similar absorption increase was observed by spectroscopy (Fig. 3e and Supplementary Fig. 4), yet with a higher increasing rate upon reaction time. As shown in Fig. 3f, it took approximately 5 h for UV–Vis spectra to increase for 0.08 at 330 nm without introducing Ag2CO3, however, accelerated by 4% amount of Ag2CO3, it took only 2 h to achieve the similar increase. In fact, the acceleration caused by Ag2CO3 could even be reflected by the considerable difference in the rate of color change observed by bare eyes, as shown in Supplementary Fig. 2.
Combining structural analyses and kinetics investigation, we can conclude that the CO32− anion acts as a self-releasable template and accelerates the formation process of 1 in a way similar to a catalyst, which is essentially different from the conventional templates that cannot be removed or can only be removed by extra post-synthesis methods53,54,55,56. Since the formation processes of supramolecular clusters are usually referred to molecular assembly22,28,31, which mainly involve reversible coordination bonding57, we prefer to classify the as-described process in this work as a catalyzed assembly (catassembly) process58. The turn over number (TON) is calculated by measuring the amount of 1 over Ag2CO3 as 4.8. Catassembly pathway for forming 1 contains three steps as shown in Fig. 4a, and the proposed mechanism of each step is shown in Fig. 4b.
Compared to tert-butyl thiolate and salicylate, CO32− has a smaller size and higher negative charge, thus enriching a higher local concentration of Ag+ ion for the nucleation of 2. The chemical equation to form 2 can be derived as Eq. (1).
As shown in Fig. 4b, proton is generated along with the formation of 2. During the growth from 2 to 1, the local concentration of proton is further accumulated. As indicated by the single-crystal X-ray analysis, the interaction between the CO32− template and the nest structure of 2 is relatively weak. Thus, the CO32− template is removed from the nest and becomes free H2CO3 when the local proton concentration gets higher during the cluster growth. The free H2CO3 further defuses and reacts with AgStBu to generate Ag2CO3 again. The chemical equation from 2 to 1 can be derived as Eq. (2).
The side products in Eqs. (1) and (2), i.e., HStBu and MeOtBu, were confirmed by the capillary gas chromatography and mass spectrometry analyses of the reaction solution and also the gas phase in the reaction vial (Supplementary Fig. 5). Moreover, if extra Ag2CO3 (e.g., 18% molar ratio to AgStBu) was introduced into the initial reactants, it would impede the release of CO32− from 2 and stop the further growth into 1. As a result, with adding 18% amount of Ag2CO3, the main product of the reaction would become 2, corroborating the proposed mechanism in Fig. 4b.
We have constructed a hollow silver-sulfide cluster with a catassembly strategy. Through carefully tracking and analyzing the growth process, we found a carbonate-templated nest-like intermediate, which accelerated the cluster growth in a similar way as catalyst accelerates a chemical reaction. The method to assemble hollow cluster through self-releasable template provides insights into understanding the formation mechanism of other hollow nanoclusters and may facilitate the design and construction of new hollow nanoclusters with various properties.
Chemicals and materials
Silver oxide (Ag2O), ammonia (NH3·H2O), sodium hydroxide (NaOH), trimethylamine (Et3N), methanol (MeOH), acetonitrile (MeCN), dichloromethane (DCM), chloroform (HCCl3), and diethyl ether (Et2O) were purchased from Sinopharm Chemical Reagent Co. Ltd. (Shanghai, China). 2-methylpropane-2-thiol (HStBu) was purchased from J&K Chemicals. Silver carbonate (Ag2CO3) was purchased from Adamas-Beta, Silver salicylate hydrate (AgC7H5O3·H2O) was purchased from Alfa-Aesar. AgStBu was prepared from the reaction of Ag2O, HStBu in NH3·H2O in an ice bath, and further washed in turn with H2O, MeOH, MeCN, and Et2O. Other reagents and solvents used without further purification.
Synthesis of Ag56S12(SC4H9)20(O3H5C7)12 (1)
Mixed acetonitrile, dichloromethane and methanol (1:1:1, 75 mL) was added 3.7 mg Ag2CO3, 306.1 mg AgC7H5O3·H2O, and 892.2 mg AgStBu. Under ultrasound condition, a white suspension was formed and further refluxed under nitrogen at 80 °C with a balloon for 5 days. The reaction solution changed from colorless to yellow and finally to orange, after cooled to room temperature, orange-red solution and orange solids were obtained. The solution was further concentrated by evaporation to generate more orange solids. Orange solids were collected, and then washed with diethyl ester (50 mL, twice) and dissolved in 30 mL dichloromethane. To this dichloromethane solution, mixed acetonitrile and methanol (1:1, 15 mL) was added to give the orange powder of 1 (633.2 mg, yield 89.7% based on AgC7H5O3·H2O). Anal. Calcd. for 1: Ag56S12(SC4H9)20(O3H5C7)12: C, 19.989; H, 2.455. Found: C, 20.279; H, 2.367. IR spectrum of 1 is shown in Supplementary Fig. 6a.
Synthesis of (CO3)Ag33S3(SC4H9)16(O3H5C7)9 (2)
Mixed acetonitrile and methanol (1:1, 9 mL) was added 25.6 mg Ag2CO3, 102.8 mg AgC7H5O3·H2O, and 137.9 mg AgStBu. Under ultrasound condition, a white suspension was formed and further refluxed under nitrogen at 80 °C for 7 h. To the cooled reaction solution was added 10 mL diethyl ether, after filtered, the yellow filtrate was dried and gave yellow powder of 2 (194.2 mg, yield 91.6% based on AgStBu). Anal. Calcd. for 2: (CO3)Ag33S3(SC4H9)16(O3H5C7)9(CH3CN)3(H2O)9: C, 24.158; H, 3.268; N, 0.631. Found: C, 23.929; H, 3.064; N, 0.631. IR spectrum of 2 is shown in Supplementary Fig. 6b.
Kinetic measurements for the reaction accelerated by Ag2CO3
In a 20 mL headspace bottle, mixed acetonitrile and methanol (1:1, 5 mL) was added into 2.9 mg Ag2CO3, 40.3 mg AgC7H5O3·H2O and 109.2 mg AgStBu. Under ultrasound condition, a white suspension was formed. Then 10 mL chloroform was added, the suspension dissolved and gave a colorless transparent solution. The reaction vial was tightly sealed and stirred in an oil bath at 80 °C. At 1 h time intervals, the reaction solution (50 µL) were drawn out and injected into 5.00 mL chloroform. The diluted solutions were immediately refrigerated in a refrigerator at 4 °C for the latter UV–Vis characterizations. The spectra of refrigerated solutions did not change after being kept in liquid nitrogen for more than 24 h.
Kinetic measurements for the reaction without adding Ag2CO3
In a 20 mL headspace bottle, mixed acetonitrile and methanol (1:1, 5 mL) was added into 40.3 mg AgC7H5O3·H2O and 110.4 mg AgStBu. Following treatment was exactly the same as the above reaction accelerated by Ag2CO3.
UV–Vis absorption spectra were collected on an Agilent Cary 5000. Fourier-transform infrared spectroscopy (FT-IR) spectra were collected from KBr pellets in the range of 4000–400 cm-1 with a Nicolet AVATAR FT-IR 380 spectrometer. Nuclear magnetic resonance (NMR) data were collected on a Bruker Avance II spectrometer (600 MHz). GC-MS analyses were performed on Shimadzu GC-2010 MS with a capillary column (Rtx-5MS). X-ray crystallography is detailed in Supplementary Note 1 and Supplementary Data 1.
The data that support the findings of this study are available from the corresponding author upon reasonable request. The X-ray crystallographic coordinates for structures reported in this Article have been deposited at the Cambridge Crystallographic Data Center (CCDC), under deposition number CCDC 1825401 and 1825402. These data can be obtained free of charge from The Cambridge Crystallographic Data Center via http://www.ccdc.cam.ac.uk/data_request/cif. All other data associated with this study are available from the authors upon reasonable request.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank Mr. Kuang-Yu Wang at Peking University for assistance in experiments. This work was supported by National Natural Science Foundation of China (21473139, 21631007, 91227111, and 91427304).