Small symmetry-breaking triggering large chiroptical responses of Ag70 nanoclusters

The origins of the chiroptical activities of inorganic nanostructures have perplexed scientists, and deracemization of high-nuclearity metal nanoclusters (NCs) remains challenging. Here, we report a single-crystal structure of Rac-Ag70 that contains enantiomeric pairs of 70-nuclearity silver clusters with 20 free valence electrons (Ag70), and each of these clusters is a doubly truncated tetrahedron with pseudo-T symmetry. A deracemization method using a chiral metal precursor not only stabilizes Ag70 in solution but also enables monitoring of the gradual enlargement of the electronic circular dichroism (CD) responses and anisotropy factor gabs. The chiral crystals of R/S-Ag70 in space group P21 containing a pseudo-T-symmetric enantiomeric NC show significant kernel-based and shell-based CD responses. The small symmetry breaking of Td symmetry arising from local distortion of Ag−S motifs and rotation of the apical Ag3 trigons results in large chiroptical responses. This work opens an avenue to construct chiral medium/large-sized NCs and nanoparticles, which are promising for asymmetric catalysis, nonlinear optics, chiral sensing, and biomedicine.

Reviewer #2 (Remarks to the Author): Crystallography Report This took me a while because I am entering unfamiliar territory here. For this study, the question is not: Are these fully refined crystal structures? They clearly aren't, and don't claim to be. The question is rather: Do the data support the conclusions the authors draw from them? My tentative answer to this is: As far as I can tell they do. I have my doubts about the charges of the clusters, simply due to the use of SQUEEZE, but from what I can tell the main conclusions about the chiral shape of the clusters are not affected by the number of valence electrons within it (and I do not understand the spectroscopic evidence, so whether that supports the charges of the clusters other referees have to answer). I have a slight reservation about the P21 structure: I agree with the authors that the most likely space group is indeed P21; I have tried to solve the structure in P21/c and got nowhere, and ADDSYMM from PLATON did not suggest any higher symmetry. However, as the authors state correctly the FLACK parameter is very high, and this can sometimes resolve itself upon full refinement of the structure. What I have also sometimes observed is that distortions in molecules occur together with high FLACK parameters in early refinement stages which both disappear when the structure is fully resolved. I cannot rule out from the data presented here that this has not occurred here, i.e. that the small differences between the AG70 cluster in P-3c1 and in P21 are due to incomplete refinement. Again, I do not understand the spectroscopic data about the shape of the clusters, so if that indeed shows that the additional distortion is real then my comment above is void.
Reviewer #3 (Remarks to the Author): The article can be published after the following questions are addressed.
In the structural description, only the metal framework was discussed. The coordination of thiolates, trifluoroacetates, and DMF ligands should be added. Are there any changes of Ag…Ag, Ag-S, and Ag-O bonds between T-and Td-symmetric Ag70? The bond lengths or angles should be reported, which are the benchmark to identify whether the structure is distorted.
The crystals of Rac-Ag70 and Ag70‧Ag12 were obtained from the same synthetic procedure (in experimental section, page 14). How did the authors said that "By using a small Ag(I) cluster as a stabilizer,…"? Based on the description of the experimental section, those two crystals can be obtained in the same condition. How to separate the two in the same reaction since the colors of the crystals are similar?
The metal framework possesses Td symmetry in Ag70‧Ag12. Will it be the same symmetry if it contains the surficial ligands? In page 12, the author mentioned that "the small Ag(I)12 cluster not only stabilizes Ag70 but also changes the crystal packing, which probably weakens the intercluster strain in Rac-Ag70 and result in Td-symmetric Ag70 in Ag70‧Ag12." Are there any interactions such as H…F contacts between clusters in Rac-Ag70? If so, please consider them for discussion.
It is clever to add chiral ligands (TFL-) to get optically pure R-/S-Ag70. R-Ag70 crystallized in a noncentrosymmetric space group P21, but the flack parameter (0.42) shows large uncertainty of the determination of the absolute structure. It could be due to racemic or twinned. However, the CD spectrum shows optical activity nicely. Thus this reviewer believes the R-/S-Ag70 were synthesized. The Figure 4d is quite unclear while R-/S-TFLAg and Rac-Ag70+R/S-TFLAg are stacked together. Please revise.
Line 352, Page 15, "no-H" should be " non-hydogen". Line 180, Page 9, "m/z" should be "m/z" (in italic style) Reviewer #4 (Remarks to the Author): Small Symmetry-Breaking Triggering Large Chiroptical Responses of Ag70Nanoclusters -Review This paper presents the structure and optical properties of chiral Ag-70 nanoclusters. A crystal structure of racemic Ag70 nanoclusters is first presented. This material was then de-racemized by incorporating a chiral silver precursor into the synthesis, and the crystals of resulting R/S enantiomeric pairs were also solved. De-racemization was also monitored using CD and was found to be reversible. In particular, deracemization of such clusters for R/S structure solutions is of particular significance, as it remains a challenge.
The charge state of rac-Ag70 is determined to be -2, based on the jellium model, and supported by ESI-MS data. The UV-Vis-IR spectra also conform to that expected of a molecular cluster, and are supported by theoretical simulations.
The structural analysis provided suggests the surface Ag-S moieties cause a minor Td symmetry breaking distortion, resulting in two enantiomers with large chiroptical properties. The CD data supports this. However, the author's note that the ligand layer structure was unable to be solved with single crystal XRD, which is not entirely unexpected given the fragility of similar metal nanoclusters when exposed to X-rays.
A co-crystal of Ag70-Ag12 was also solved, and found to impart greater stability than the Rac-Ag70 in solution. A stability mechanism of Ag70 clusters in solution is proposed. The paper also presents a novel Ag16S4 FCC inner core structure within the Ag70 cluster, which is particularly of interest to the community.
On the whole, the data in the paper is convincing and well presented, and should be published for the community. I do not have major remarks other than the following: A few parts of the main text are unclear, which could undermine the importance of this work. In particular, the introductory sentences are somewhat clunky/confusing (lines 40 -45) in relation to the material that follows. Additionally, expanding on the novelty of the Ag16S4 inner core structure (lines 141-144) would be welcomed.
Also, the author's do not mention if the as-synthesized clusters were "washed" or re-crystallized after isolation to remove excess impurities prior to subsequent analyses. This can be helpful for obtaining higher quality single crystals for XRD.
1). The formulation of the title cluster is {Ag70S4(S-iPr)24(TFA)20(DMF)3} 2− (Rac-Ag70). Here CF3COO − is abbreviated as TFA. What is the formulation of companion cluster Ag12 in the cocrystal Ag70·Ag12? It looks like this small Ag12 cluster is a highly distorted cuboctahedron (Fig. S43). Yet the large Ag70 cluster, under the influence of this small companion Ag12 cluster, has "reverted" back to achiral of Td symmetry! With TFA ligands, Ag70 crystalizes as a racemate Rac-Ag70 with mirror-images of R and S chiral clusters: how did this small, distorted Ag12 cluster, help make the large Ag70 cluster lose its chirality? We think that the lower T-symmetry of chiral clusters is induced by the slight lattice distortion, according to single-crystal structural analysis . When the small Ag12 cluster occupies the interstitial space, it probably weakens the strain in individual chiral cluster in Rac-Ag70 crystals, resulting in a rise of symmetry to higher Td-symmetry.
2). Please provide structural details of the Ag12 cluster, complete with the ligand positions and interatomic distances, in the cocrystal Ag70·Ag12. Also provide the UV-Vis and CD of the Ag12 cluster in pristine form if possible. If this Ag12 cluster cannot be made, I would like to suggest DFT calculations to identify its contribution to the UV-Vis and CD spectra of the cocrystal, if any.

Response:
As suggested by Reviewer 1, the structural details of the Ag12 cluster together with the ligand positions and interatomic distances in the cocrystal Ag70·Ag12 have been added in the revised manuscript.
"The smaller Ag(I)12 cluster possesses a slightly distorted cuboctahedral metal framework (Ag···Ag distances: 3.06(1)-3.09(2) Å, Supplementary Figs. 46-48). The six μ4-S i Pr − ligands are anchored on the quadrilateral surface of Ag(I)12 with Ag−S bond lengths in the range of 2.44(4)−2.50(4) Å, and the CF3COO − ligands on the edges in μ2η 1 ,η 1 coordination mode (Ag−O bond lengths: 2.29(2)−2.44(4) Å). " As suggested by Reviewer 1, we identify the contribution to the UV-Vis spectrum of Ag12 in the cocrystal Ag70·Ag12 through DFT calculations (at the PBE0/def2SVP level, Fig. R1), because we indeed cannot isolate the identical Ag12 after a lot of trials. The theoretical results show that the Ag12 cluster has not contribution to the absorbance in the visible-light region, implying that the CD signals in the ligand-exchange process come from the superatomic Ag70. Figure R1. Experimental (Ag70·Ag12 and Rac-Ag70) and calculated (Ag70 and Ag12) absorption spectra.
3). As the TFA is replaced by the chiral TFL ligands, the CD of the Ag70 is progressively enhanced, indicating deracemization. Is this experimental observation the result of (1) the instant conversion of, say, S to R enantiomer, or (2) a progressive distortion of the metal framework as more and more TFAs are replaced by the chiral TFLs? If the answer is (1), how many TFAs must be replaced by chiral TFLs to cause this transformation?
Response: This is a very good comment. We think that the two processes (1) and (2) could cooccur as the TFA is replaced by the chiral TFL ligands.
First, for process (1), we could find implications in the back CD titration of deracemization solution by using achiral TFA. When n(CF3COOAg):n(R-TFLAg) reaches 1000:50, the CD signal nearly approaches zero and there is basically no change afterwards. ESI-MS results show that the chiral TFL nearly disappears . In addition, the successful crystallization of chiral crystals with mirror-image CD response using chiral TFL also suggests that the conversion of chirality really occurs in this system (Fig. 5c). Nevertheless, because of the complex dynamic process in the solution system, where each cluster could have a different occurrence for the number of chiral ligands, the conversion point could not be completely quantified for the time being. In future work, we will work on this interesting topic.
Second, for process (2), the progressive distortion of the Ag70 framework also can be deduced, because the CD intensity rises progressively and reaches a peak, after which CD profiles keep constant, implying that the distortion has a limit. The more serious distortions of Ag70 skeleton are actually found in the monochiral single-crystals ( Fig. 5a-b). In addition, after several iterative crystallization using TFL ligands, we fortunately obtain a higher-quality single crystal for diffraction, whose cif (S-Ag70-3.cif) is uploaded for Reviewers and whose CD spectra (Fig. R2) agree well with the original data, supporting the distortion become more and more with the increasing of chiral TFLs. Response: As suggested by Reviewer 1, the exchange reaction is in sync tracked by CD ( Fig.4 and Supplementary Figs. 31−32) and 19 F-NMR spectra ( Fig. R3 and Table  R1).
As commented by Reviewer 1, the ratio on the interface of the actual cluster does not match the experimental ratio used, which is observed in the ESI-MS data .
In the range of n(R-/S-TFLAg):n(Ag70) = 0-15 (the experimental ratio used), the CD intensity is almost linearly enhanced ( Fig. 4c and Supplementary Fig. 32c), which should include two basic inducements, e.g. chiral conversion of a part of clusters and chiral enlargement of another part of clusters. Therefore, in combination ESI-MS data, we proposed that m value should be an average in some range.
For in sync 19 F-NMR spectra of exchange reaction using a chiral metal precursor (Fig. R3), we found that with different amounts of precursors (R-TFLAg) added, there are always only two peak positions, which are different from those in precursor of TFAAg and TFLAg ( Supplementary Fig. 63), indicating the different shielding effects of the Ag70 skeleton on F nuclei of TFA and TFL. 19 F-NMR spectra show that during exchange reaction, the ratios of TFA to TFL are close to the feed ratio (Table R1). Figure R3. 19 F-NMR spectra of Rac-Ag70 using a chiral metal precursor (R-TFLAg) in DMF and CDCl3 at room temperature.   Supplementary Fig. 29), Rac-Ag70 + CF3COOAg ( Supplementary Fig. 29), Rac-Ag70 + C2F5COOAg ( Supplementary Fig. 30), and Ag70‧Ag12 ( Supplementary Fig. 54). We think that there could be potential sites on the surface of Ag70 clusters, which have more affinity with the small molecules including (TFL)Ag, CF3COOAg and C2F5COOAg.
In fact, we also pay attention to this interesting point and are working on it to assemble such large superatomic clusters by virtue of these potential sites. 6). This paper provides no structural details on the surface ligands for the STAR cluster: the chiral R-Ag70. What is the TFL:TFA ratio on the surface? Are they badly disordered? Perhaps FTIR or NMR can provide an estimate (see Item 4 above) Response: As commented by Reviewer 1, the surface ligands of cluster surface are badly disordered, resulting in the loss of structural details.
As suggested by Reviewer 1, we measured 19 F-NMR of R-Ag70 samples in CDCl3 and found two resonances at −74.5 and −75.8 ppm in ca. 12:8 ratio ( Supplementary Fig.  63), corresponding to the -CF3 groups in TFA and TFL ligands. So, the TFA:TFL ratio on the surface is ca. 12:8 by 19 F-NMR.
"The 19 F-NMR spectrum of Rac-Ag70 shows one resonance (−74.2 ppm), which corresponds to the −CF3 group in the CF3COO − ligand, while the resonance shift to −73.3 ppm for CF3COOAg in DMF and CDCl3. The 19 F-NMR spectra of R-Ag70 and S-Ag70 show two resonances (−74.5 and −75.8 ppm) in ca. 12:8 ratio, corresponding to the −CF3 groups in the CF3COO − and TFL − ligands, respectively. For comparison, the resonance corresponding to the −CF3 groups in the TFL − ligand shifts to −76.2 ppm for TFLAg in CDCl3." 7). It was stated in the paper that "Only one case of enantiomerically pure single crystals of a medium-sized silver nanocluster, Ag78, protected by a chiral diphosphine ligand, has been characterized.21" This is utterly not true.

Response:
We define a medium-sized silver nanocluster with the number of Ag atoms over 50 (J. Am. Chem. Soc. 2019, 141, 19754-19764). Therefore, we claim that Ag78 reported by Zheng group is the most outstanding one. In addition, these significant small-sized Ag clusters (< 50 Ag) recommended by Reviewer 1 are also referenced (Ref 5,6,8,10).

Reviewer 2:
Crystallography Report This took me a while because I am entering unfamiliar territory here. For this study, the question is not: Are these fully refined crystal structures? They clearly aren't, and don't claim to be. The question is rather: Do the data support the conclusions the authors draw from them? My tentative answer to this is: As far as I can tell they do. I have my doubts about the charges of the clusters, simply due to the use of SQUEEZE, but from what I can tell the main conclusions about the chiral shape of the clusters are not affected by the number of valence electrons within it (and I do not understand the spectroscopic evidence, so whether that supports the charges of the clusters other referees have to answer).
I have a slight reservation about the P21 structure: I agree with the authors that the most likely space group is indeed P21; I have tried to solve the structure in P21/c and got nowhere, and ADDSYMM from PLATON did not suggest any higher symmetry. However, as the authors state correctly the FLACK parameter is very high, and this can sometimes resolve itself upon full refinement of the structure. What I have also sometimes observed is that distortions in molecules occur together with high FLACK parameters in early refinement stages which both disappear when the structure is fully resolved. I cannot rule out from the data presented here that this has not occurred here, i.e. that the small differences between the AG70 cluster in P-3c1 and in P21 are due to incomplete refinement. Again, I do not understand the spectroscopic data about the shape of the clusters, so if that indeed shows that the additional distortion is real then my comment above is void.
Response: Thank this reviewer for the modest comments very much.
The severe disorder and some other ambiguous facts in the crystal data indeed plagued us a lot. Although as the reviewer stating "I cannot rule out from the data presented here that this has not occurred here", there is also possibility that the structure should be in centrosymmetric space group, which we admit, is actually a more common selection, particularly for small molecule crystallography. Luckily, after hundreds of crystallization experiments, we obtained a new crystalline phase for the same cluster that allowed us to unambiguously determine the space group as P212121(S-Ag70-3.cif). The single-crystal X-ray diffraction experiments showed that the structure is in an orthorhombic system and the unit cell parameters (23.89, 29.99, 46.30, 90, 90, 90) are completely different from the one in the previous manuscript (23.83, 33.56, 37.56, 90, 91.23, 90) and more closer to the parent structure of Rac- Ag70 (29.64,29.64,46.99,90,90,120). Although only the Ag−S skeleton (the same Ag70S28) can be well solved in the new structure, the statistic diffraction intensity gives a lower │E*E-1│ value of 0.915, and only chiral space groups are suggested according to the systematic absence. Finally, after solving the Ag70S28 core, a much lower Hooft y parameter (0.158(12)) Ref 1 was computed from the least-squares refinement on F 2 , implying the overwhelming probability of the R-chirality of the cluster in the crystal. Note that Hooft y parameter has been widely used in present to determine the absolute configuration in modern crystallography. It is regarded as giving even greater enantiomer distinguishing power than Flack parameter.
In addition, we carefully compared the Ag70 skeleton (Fig. R4) in the new structure with that in the structure of Rac-Ag70. Clear distortion of the new cluster can be observed, which also supports its non-centrosymmetric geometry induced by the peripheral ligands. The new crystal data (S-Ag70-3.cif) can better support our conclusions in the manuscript, which is uploaded for the Reviewer's evaluation. Figure R4. Compared the Ag70 skeleton in the new structure (blue, from S-Ag70-3.cif) with that in the structure of Rac-Ag70 (green).

Reviewer 3:
The article can be published after the following questions are addressed. 1). In the structural description, only the metal framework was discussed. The coordination of thiolates, trifluoroacetates, and DMF ligands should be added. Are there any changes of Ag…Ag, Ag−S, and Ag−O bonds between T-and Td-symmetric Ag70?
The bond lengths or angles should be reported, which are the benchmark to identify whether the structure is distorted.

Response:
As suggested by Reviewer 3, the coordination of thiolates, trifluoroacetates, DMF ligands, and relevant Ag···Ag, Ag−S, and Ag−O bonds had been discussed in the revised manuscript and revised Supplementary Information.
In addition, the relevant differences of Ag···Ag, Ag−S bonds between T-and Tdsymmetric Ag70 is included in Supplementary Table 1. 2). The crystals of Rac-Ag70 and Ag70‧Ag12 were obtained from the same synthetic procedure (in experimental section, page 14). How did the authors said that "By using a small Ag(I) cluster as a stabilizer,…"? Based on the description of the experimental section, those two crystals can be obtained in the same condition. How to separate the two in the same reaction since the colors of the crystals are similar.
Response: It is likely that we did not clearly describe the difference of synthesis conditions between Rac-Ag70 and Ag70‧Ag12, leading to Reviewer 3's misunderstanding.
The synthetic procedure of the crystals of Rac-Ag70 and Ag70‧Ag12 are similar, but the solvent is different. For Rac-Ag70: 3.5 mL i PrOH and 1.5 mL DMF; For Ag70‧Ag12: 3.5 mL EtOH and 1.5 mL DMF. We use different alcohols to tune the reducibility of DMF in reaction systems. Therefore, pure Rac-Ag70 and Ag70‧Ag12 crystals are separately prepared in respective systems.
In order to avoid the readers' misunderstanding, we revise the related description in Methods-Synthesis, which is highlighted in yellow in the revised manuscript.
3). The metal framework possesses Td symmetry in Ag70‧Ag12. Will it be the same symmetry if it contains the surficial ligands? In page 12, the author mentioned that "the small Ag(I)12 cluster not only stabilizes Ag70 but also changes the crystal packing, which probably weakens the intercluster strain in Rac-Ag70 and result in Td-symmetric Ag70 in Ag70‧Ag12." Are there any interactions such as H…F contacts between clusters in Rac-Ag70? If so, please consider them for discussion.

Response:
The 70-nuclearity Ag NC including metal framework and its surficial ligands in Ag70·Ag12 exhibits nearly perfect Td symmetry.
At the same time, the C-H···F and C-H···O interactions between clusters in Rac-Ag70 have been discussed in detail in the revised manuscript. It is found that there are multiple C-H···F hydrogen bonds between CF3COOand -CH3 groups of S i Prand DMF ligands ( Supplementary Fig. 16), suggesting the strong interactions between the clusters.
Supplementary Figure 16. The hydrogen bonds of Rac-Ag70. (a) The intra-cluster and inter-cluster C-H···F, and intra-cluster C-H···O hydrogen bonds. (b) The inter-cluster C-H···F hydrogen bonds. (c) The intra-cluster C-H···F, and C-H···O hydrogen bonds. 4). It is clever to add chiral ligands (TFL -) to get optically pure R-/S-Ag70. R-Ag70 crystallized in a non-centrosymmetric space group P21, but the flack parameter (0.42) shows large uncertainty of the determination of the absolute structure. It could be due to racemic or twinned. However, the CD spectrum shows optical activity nicely. Thus this reviewer believes the R-/S-Ag70 were synthesized. The Figure 4d is quite unclear while R-/S-TFLAg and Rac-Ag70+R/S-TFLAg are stacked together. Please revise.

Response:
We appreciate Reviewer 3's positive comments on our conclusions.
As Reviewer 3 suggested, Fig. 4d has been revised in the revised manuscript. And the control CD of R/S-TFLAg and deracemization of Rac-Ag70 solution have been shown in Supplementary Fig. 34. 5). Line 352, Page 15, "no-H" should be " non-hydogen".
Line 180, Page 9, "m/z" should be "m/z" (in italic style) Response: As Reviewer 3 suggested, the errors mentioned have been corrected in the revised manuscript.

Reviewer 4:
This paper presents the structure and optical properties of chiral Ag70 nanoclusters. A crystal structure of racemic Ag70 nanoclusters is first presented. This material was then de-racemized by incorporating a chiral silver precursor into the synthesis, and the crystals of resulting R/S enantiomeric pairs were also solved. De-racemization was also monitored using CD and was found to be reversible. In particular, deracemization of such clusters for R/S structure solutions is of particular significance, as it remains a challenge. The charge state of rac-Ag70 is determined to be -2, based on the jellium model, and supported by ESI-MS data. The UV-Vis-IR spectra also conform to that expected of a molecular cluster, and are supported by theoretical simulations. The structural analysis provided suggests the surface Ag-S moieties cause a minor Td symmetry breaking distortion, resulting in two enantiomers with large chiroptical properties. The CD data supports this. However, the author's note that the ligand layer structure was unable to be solved with single crystal XRD, which is not entirely unexpected given the fragility of similar metal nanoclusters when exposed to X-rays. A co-crystal of Ag70-Ag12 was also solved, and found to impart greater stability than the Rac-Ag70 in solution. A stability mechanism of Ag70 clusters in solution is proposed. The paper also presents a novel Ag16S4 FCC inner core structure within the Ag70 cluster, which is particularly of interest to the community. On the whole, the data in the paper is convincing and well presented, and should be published for the community. I do not have major remarks other than the following: 1). A few parts of the main text are unclear, which could undermine the importance of this work. In particular, the introductory sentences are somewhat clunky/confusing (lines 40 -45) in relation to the material that follows. Additionally, expanding on the novelty of the Ag16S4 inner core structure (lines 141-144) would be welcomed.

Response:
We really appreciate the constructive comment.
As Reviewer 4 suggested, the introduction section (lines 40 -45) has been rewritten which is highlighted in yellow in the revised manuscript.
The novelty of Ag16S4 inner core structure has been highlighted and expanded in the revised manuscript (Page 7).
2). Also, the author's do not mention if the as-synthesized clusters were "washed" or re-crystallized after isolation to remove excess impurities prior to subsequent analyses. This can be helpful for obtaining higher quality single crystals for XRD.