Surface environment complication makes Ag29 nanoclusters more robust and leads to their unique packing in the supracrystal lattice

Silver nanoclusters have received unprecedented attention in cluster science owing to their promising functionalities and intriguing physical/chemical properties. However, essential instability significantly impedes their extensive applications. We herein propose a strategy termed “surface environment complication” to endow Ag29 nanoclusters with high robustness. The Ag29(S-Adm)18(PPh3)4 nanocluster with monodentate PPh3 ligands was extremely unstable and uncrystallizable. By substituting PPh3 with bidentate PPh2py with dual coordination sites (i.e., P and N), the Ag29 cluster framework was twisted because of the generation of N–Ag interactions, and three NO3 ligands were further anchored onto the nanocluster surface, yielding a new Ag29(S-Adm)15(NO3)3(PPh2py)4 nanocluster with high stability. The metal-control or ligand-control effects on stabilizing the Ag29 nanocluster were further evaluated. Besides, Ag29(S-Adm)15(NO3)3(PPh2py)4 followed a unique packing mode in the supracrystal lattice with several intercluster channels, which has yet been observed in other M29 cluster crystals. Overall, this work presents a new approach (i.e., surface environment complication) for tailoring the surface environment and improving the stability of metal nanoclusters.

Herein, a "surface environment complication" strategy has been exploited to endow the Ag 29 nanocluster with high robustness. By substituting the monodentate PPh 3 (with only the P coordination site) in previously reported Ag 29 -PPh 3 with bidentate PPh 2 py (with P and N dual coordination sites), the nanocluster surface structure underwent a twist due to the generation of N-Ag interactions. Besides, three NO 3 ligands were further anchored onto the nanocluster surface, making the metallic kernel entirely wrapped. The obtained Ag 29 (S-Adm) 15 (NO 3 ) 3 (PPh 2 py) 4 (Ag 29 -PPh 2 py for short) nanocluster was much more robust relative to Ag 29 -PPh 3 , and its structure was successfully determined by single-crystal X-ray diffraction. Furthermore, based on this nanocluster template, the metalcontrol and ligand-control effects on stabilizing the Ag 29 framework were evaluated. Moreover, at the supramolecular level, Ag 29 -PPh 2 py followed a unique packing mode in the crystal lattice with several intercluster channels, while such an aggregation pattern has yet been discovered in other M 29 cluster crystals.

Synthesis of Ag 29 (S-Adm) 18 (PPh 3 ) 4 (Ag 29 -PPh 3 )
The preparation of Ag 29 -PPh 3 was based on a reported method. 46 Synthesis of Pt 1 Ag 28 (S-Adm) 18 (PPh 3 ) 4 (Pt 1 Ag 28 -PPh 3 ) The preparation of Pt 1 Ag 28 -PPh 3 was based on a reported method. 46 Preparation of Ag 29 (S-Adm) 15 (NO 3 ) 3 (PPh 2 py) 4 (Ag 29 -PPh 2 py) In a 50 mL round-bottom ask, 94 mg of AgNO 3 was dissolved in 5 mL of MeOH and 10 mL of EtOH, and 50 mg of Adm-SH was added under vigorous stirring. Aer 20 min, 100 mg of PPh 2 py was added. Shortly aer this, 10 mg of NaBH 4 (dissolved in 1 mL of EtOH) was poured in, and the reaction was continued for 12 hours. The obtained solution was centrifuged at 10 000 rpm for 5 minutes, and then the supernatant was collected and evaporated to get the dry product, which was then washed several times with n-hexane to get the nal product, i.e., Ag 29 -PPh 2 py. The yield was about 30% based on the Ag element (calculated from AgNO 3 ).

Characterization
The optical absorption spectra of nanoclusters were recorded using an Agilent 8453 diode array spectrometer. Electrospray ionization mass spectrometry (ESI-MS) measurements were performed by using a Waters XEVO G2-XS QTof mass spectrometer. The sample was directly infused into the chamber at 5 mL min À1 . For preparing the ESI samples, nanoclusters were dissolved in CH 2 Cl 2 (1 mg mL À1 ) and diluted Infrared (IR) measurements were recorded on a Bruker Vertex 80sv Fourier transform IR spectrometer.

X-ray crystallography
The data collection for single-crystal X-ray diffraction (SC-XRD) of Ag 29 -PPh 2 py was carried out on a Bruker Smart APEX II CCD diffractometer under a nitrogen ow, using graphitemonochromatized Mo Ka radiation (l ¼ 0.71073Å). The data collection for single-crystal X-ray diffraction (SC-XRD) of Pt 1 Ag 28 -PPh 2 py was carried out on a Stoe Stadivari diffractometer under a nitrogen ow, using graphite-monochromatized Cu Ka radiation (l ¼ 1.54186Å). Data reductions and absorption corrections were performed using the SAINT and SADABS programs, respectively. The structure was solved by direct methods and rened with full-matrix least squares on F 2 using the SHELXTL soware package. All non-hydrogen atoms were rened anisotropically, and all the hydrogen atoms were set in geometrically calculated positions and rened isotropically using a riding model. All crystal structures were treated with PLATON SQUEEZE. The diffuse electron densities from these residual solvent molecules were removed. The CCDC number of the Ag 29 -PPh 2 py nanocluster is 2115749. The CCDC number of the Pt 1 Ag 28 -PPh 2 py nanocluster is 2117814.

Results and discussion
Ag 29 -PPh 3 was prepared by a literature method. 46 Although the Ag 29 -PPh 3 nanocluster was uncrystallizable because of its weak stability, several of its alloyed derivatives have been structurally determined, including Pt 1 Ag 28 -PPh 3 , Au 1 Ag 28 (S-Adm) 18 (PPh 3 ) 4 , and Pt 1 Ag 12 Cu 16 (S-Adm) 18 (PPh 3 ) 4 . [44][45][46] In this context, alloying has been used as an efficient approach to improve the stability of the M 29 framework. 46 Fig. 1 depicts the proposed structure of Ag 29 -PPh 3 . Of note, the Ag 13 kernel in Ag 29 -PPh 3 might follow a FCC (face-centered cubic) conguration for two reasons: (i) the consistent FCC conguration of the M 13 kernel in PPh 3 and S-Adm co-stabilized M 29 nanoclusters, [44][45][46] and (ii) the different absorption proles of Ag 29 -PPh 3 and Ag 29 -PPh 2 py (discussed below). However, such a verication calls for more experimental efforts.
At the same time, we unremittingly made efforts to stabilize the homo-silver Ag 29 and determine its atomically precise structure. Considering that (i) the unchanging S-Adm ligand could retain the basic framework of the Ag 29 nanocluster 47,48 and (ii) the introduction of N-coordination sites in original ligands would generate new N-metal interactions that might enhance the structural robustness, [49][50][51][52] we were motivated to substitute the PPh 3 ligand with PPh 2 py while retaining the S-Adm ligand in the nanocluster synthesis. A new Ag 29 nanocluster, formulated as Ag 29 (S-Adm) 15 (NO 3 ) 3 (PPh 2 py) 4 (Ag 29 -PPh 2 py), was synthesized and further structurally determined owing to its high stability ( Fig. 1 and S1 †).
Compared with Ag 29 -PPh 3 , Ag 29 -PPh 2 py contained three fewer S-Adm ligands and three more NO 3 ligands, and the number of the phosphine ligands retained was four (Fig. 1). Because of the interactions between N (in PPh 2 py) and Ag (in the cluster), the surface structure of Ag 29 -PPh 2 py displayed more obvious distortion relative to Ag 29 -PPh 3 ( Fig. 1 and S2 †). Besides, three NO 3 ligands were observed on the nanocluster surface via Ag-O interactions. For the three O atoms in each NO 3 , the two inward O linked to two Ag atoms or one Ag atom, while the outward O was naked ( Fig. 1 and S2 †). The presence of NO 3 in the cluster system has been veried by IR measurement (Fig. S3 †). ESI-MS measurement was performed to validate the molecular composition and determine the valence state of the nanocluster. As shown in Fig. S4 15 (NO 3 ) 2 (PPh 2 py) 4 ] 3+ , respectively. In this context, the NO 3 ligand on the nanocluster surface was more prone to be dissociated relative to S-Adm and PPh 2 py ligands. Besides, the "+3" valence state of Ag 29 -PPh 2 py was tallied with the presence of 3SbF 6 À counterions with an Ag 29 cluster molecule in the crystal lattice (Fig. S1 †). According to the valence state of the Ag 29 -PPh 2 py nanocluster, its nominal electron count was determined to be 8, 53 i.e., 29(Ag) À 15(SR) À 3(NO 3 ) À 3(charge) ¼ 8e, the same as that of Ag 29 -PPh 3 . Structurally, the Ag 29 -PPh 2 py nanocluster contains an icosahedral Ag 13 kernel ( Fig. 2A). Of note, for other structurally determined M 29 (S-Adm) 18 (PR 3 ) 4 nanoclusters, their Ag 13 kernels follow a FCC conguration. 46 The difference between these two kernel congurations originates from their distinguishable surface environments via a "surface-kernel structure transfer effect". The Ag 13 kernel of Ag 29 -PPh 2 py is rst wrapped by three same Ag 4 (S-Adm) 2 (PPh 2 py) 1 motif structures that are further xed by three S-Adm bridges ( Fig. 2B and C), giving rise to an Ag 25 (S-Adm) 9 (PPh 2 py) 3 structure (Fig. 2D). Such three Ag 4 (S-Adm) 2 (PPh 2 py) 1 motifs or three S-Adm bridges are in C 3 axial symmetry. Besides, an Ag 4 (S-Adm) 6 (PPh 2 py) 1 surface unit caps the Ag 25 (S-Adm) 9 (PPh 2 py) 3 structure to present an Ag 29 (S-Adm) 15 (PPh 3 py) 4 structure ( Fig. 2E and F). In this context, the four PPh 2 py ligands follow different bonding modes in the nanocluster framework: three PPh 2 py are dually bonded onto the nanocluster via both Ag-P and Ag-N interactions, while the remaining one is singly bonded onto the nanocluster vertex via the Ag-P interaction (Fig. S2 †). Of note, the Ag 29 (S-Adm) 15 (-PPh 3 py) 4 structure is still bare to a certain extent, and three NO 3 ligands, which originated from the AgNO 3 reactant, are further anchored onto the nanocluster surface (Fig. 2G), making the Ag 29 kernel fully protected and yielding the overall structure of Ag 29 -PPh 2 py (Fig. 2H). The complete structure of Ag 29 -PPh 2 py follows a C 3 axial symmetry, and the axis of the symmetry passes through the vertex P and the innermost Ag atoms (Fig. S5 †).
In the crystal lattice of Ag 29 -PPh 2 py, two nanocluster enantiomers were observed, labeled as the R-nanocluster enantiomer and S-nanocluster enantiomer in Fig. 2I and J. Each type of enantiomer displayed a bilayer rotation: (i) for the S-nanocluster enantiomer, the inner-layer (i.e., the Ag 4 (S-Adm) 6 (PPh 2 py) 1 ) was counterclockwise while the outer-layer (i.e., assembly of three surface Ag 1 (S-Adm) 1 (PPh 2 py) 1 ) was clockwise (Fig. 2I); (ii) for the R-nanocluster enantiomer, the rotations of the inner-layer and outer-layer were opposite to those of the S-nanocluster enantiomer (Fig. 2J). Since the quantities of Rand S-nanocluster enantiomers are the same in the crystal lattice, the nanocluster samples were racemic.
The Ag 29 -PPh 3 and Ag 29 -PPh 2 py nanoclusters with distinguishable kernel structures and surface environments exhibited different optical absorptions. The CH 2 Cl 2 solution of Ag 29 -PPh 3 showed an intense absorption at 413 nm and a shoulder band at 506 nm (Fig. S6, † black line). By comparison, the CH 2 Cl 2 solution of Ag 29 -PPh 3 showed several apparent UV-vis signals at 401, 438, and 530 nm (Fig. S6, † red line). The difference in optical absorptions of these two Ag 29 nanoclusters suggested their distinct electronic structures. 54,55 The photoluminescence properties of Ag 29 -PPh 3 and Ag 29 -PPh 2 py nanoclusters were further compared. As shown in Fig. S7, † the CH 2 Cl 2 solution of Ag 29 -PPh 3 was red emissive with an intense signal at 622 nm. By comparison, the Ag 29 -PPh 2 py was non-emissive in the solution state. The different photophysical properties originated from their distinct electronic structures. 54,55 The thermal stability of these two Ag 29 nanoclusters was then compared in air. As shown in Fig. 3A, the characteristic optical peaks of Ag 29 -PPh 3 continuously decreased in the rst three hours and completely disappeared within six hours, demonstrating the decomposition of the nanoclusters. In this context, the Ag 29 -PPh 3 nanocluster was unstable. In vivid contrast, the optical absorptions of Ag 29 -PPh 2 py remained unchanged for 24 hours (Fig. 3B), which suggested the high robustness of this nanocluster. Besides, the difference in stability was primarily responsible for the crystallographic discrepancy of these two Ag 29 nanoclusters: the Ag 29 -PPh 3 nanocluster was uncrystallizable, whereas the crystal structure of Ag 29 -PPh 2 py was successfully determined.
Collectively, as depicted in Fig. 4A, two approaches have been presented to endow the unstable Ag 29 -PPh 3 nanocluster with enhanced stability: (i) the metal control approach (e.g., from unstable Ag 29 -PPh 3 to stable Pt 1 Ag 28 -PPh 3 ), 46 and (ii) the ligand control approach (i.e., from unstable Ag 29 -PPh 3 to stable Ag 29 -PPh 2 py). These two disparately stabilizing approaches raised an interesting question: which type of the Pt 1 Ag 28 nanocluster would be generated when the metal control and the ligand control were performed simultaneously in the synthesis (Fig. 4B)?
As inspired by the aforementioned results, two types of Pt 1 Ag 28 nanoclusters with different surface environments might be generated (Fig. 4B): Pt 1 Ag 28 (S-Adm) 18 (PPh 2 py) 4 with a maintained framework or Pt 1 Ag 28 (S-Adm) 15 (NO 3 ) 3 (PPh 2 py) 4 with a twisted framework. Aer the crystallographic analysis, we determined its structure as the framework-retained Pt 1 Ag 28 (S- Adm) 18 (PPh 2 py) 4 (Pt 1 Ag 28 -PPh 2 py for short). The structure of Pt 1 Ag 28 -PPh 2 py was almost the same as that of Pt 1 Ag 28 -PPh 3 (Fig. S8 †). 44,46 Although the four PPh 2 py ligands in Pt 1 Ag 28 -PPh 2 py exposed N coordination sites, these N sites remained uncoordinated in the nanocluster formation (Fig. S8 †). Consequently, in the competition between metal control and ligand control in this nanocluster system, the metal control seized a dominant position (Fig. 4B). In other words, when the Pt heteroatom was introduced into the innermost region of the nanocluster, the M 29 structure was robust enough to hinder the formation of surface Ag-N interactions, which resulted in a retained cluster framework without any distortion. Besides, in the previously reported intercluster transformation from Pt 1 Ag 28 -PPh 3 into Pt 1 Ag 28 (BDT) 12 (PPh 3 ) 4 (BDT ¼ 1,3-benzenedithiolate), the presence of BDT afforded the kernel transformation from FCC into icosahedron. 56 In this context, for the Pt 1 Ag 28 cluster template, the bidentate thiolate ligand (i.e., BDT) showed enhanced ability for directing the nanocluster conguration relative to the bidentate phosphine ligand (i.e., PPh 2 py).
The Ag 29 -PPh 2 py nanocluster molecules followed a crystallographic pattern of "lamellar eutectic" between R-nanocluster and S-nanocluster enantiomers, viewed from both x and y axes ( Fig. S9A-C †). The interlayer distance along the z axis was determined to be 34.064Å (from cluster kernel to cluster kernel, as shown in Fig. S9B †). Signicantly, the supracrystal lattice of Ag 29 -PPh 2 py showed several intercluster channels with the same diameter of 18.875Å from the (001) crystalline plane ( Fig. 5A and S9D †), which was reminiscent of the behavior of MOFs (metalorganic frameworks). 57,58 However, the channel diameter should be remarkably less than 18.875Å due to the presence of carbon tails from peripheral ligands of nanoclusters (Fig. S10 †). The intercluster channel was constructed by symmetrically assembling six cluster molecules into a hexagon, where three molecules were R-nanocluster enantiomers (marked in orange in Fig. 5B), while the other three were S-nanocluster enantiomers (marked in blue in Fig. 5B). Specically, the intercluster hexagon was composed of two cluster-based triangles in parallel planes in opposite directions, and each triangle contained three cluster molecules in the same enantiomeric conguration ( Fig. 5B and C). The intermolecular distance of the cluster-based triangle was 22.224Å, and the interlayer distance between two adjacent triangles was 18.816Å ( Fig. 5B and C). Furthermore, the arrangement of SbF 6 À counterions in the supracrystal lattice was analyzed. As shown in Fig. S11, † 2/3 of SbF 6 À counterions were uniformly organized in the intercluster channels while the others were packed along the C 3 axis of symmetry of Ag 29 -PPh 2 py nanoclusters. Of note, such a hexagon-like crystallographic packing of Ag 29 -PPh 2 py cluster molecules in the supracrystal lattice was unique, which has yet been detected in other M 29 nanocluster crystals. [44][45][46]48,59,60 For example, for the crystal lattice of Pt 1 Ag 28 -PPh 2 py, the nanocluster molecules were packed in   a layered assembly mode from the x axis, y axis, or z axis, and no intercluster channel was detected (Fig. S12 †). In this context, such unique intercluster channels may render the Pt 1 Ag 28 -PPh 2 py crystals potential nanomaterials for gas adsorptionrelated applications. 61-65

Conclusions
In summary, a strategy termed "surface environment complication" has been exploited to render unstable Ag 29 highly robust. The surface structure of unstable Ag 29 (S-Adm) 18 (PPh 3 ) 4 underwent directional distortion due to the generation of Ag-N interactions by substituting the monodentate PPh 3 ligand with bidentate PPh 2 py. Besides, three NO 3 ligands were anchored onto the nanocluster surface to entirely protect the Ag 29 kernel, yielding a new Ag 29 (S-Adm) 15 (NO 3 ) 3 (PPh 2 py) 4 nanocluster with high robustness. Owing to its enhanced stability, the Ag 29 (S-Adm) 15 (NO 3 ) 3 (PPh 2 py) 4 nanocluster was crystallizable, and its atomically precise structure was successfully determined. On the supramolecular level, the Ag 29 (S-Adm) 15 (NO 3 ) 3 (PPh 2 py) 4 nanocluster molecules followed a unique crystallographic packing mode and displayed several intercluster channels. This study thus presented a novel strategy for tailoring the surface environment of metal nanoclusters, and also provided fundamental insights into the controllable synthesis of highly robust silver nanoclusters. Future work will focus on promoting this strategy to other ligand-protected metal nanoclusters.

Data availability
All the data supporting this article have been included in the main text and the ESI. †

Author contributions
C. X. and Q. Y. carried out the experiments and analyzed the data. X. W., H. L. and H. S. assisted in the analysis. X. K. and M.
Z. designed the project, analyzed the data, and wrote the manuscript.

Conflicts of interest
There are no conicts to declare.