Evolution from unimolecular to colloidal-quantum-dot-like character in chlorine or zinc incorporated InP magic size clusters

Magic-sized clusters (MSCs) can be isolated as intermediates in quantum dot (QD) synthesis, and they provide pivotal clues in understanding QD growth mechanisms. We report syntheses for two families of heterogeneous-atom-incorporated InP MSCs that have chlorine or zinc atoms. All the MSCs could be directly synthesized from conventional molecular precursors. Alternatively, each series of MSCs could be prepared by sequential conversions. 386-InP MSCs could be converted to F360-InP:Cl MSCs, then to F399-InP:Cl MSCs. Similarly, F360-InP:Zn MSCs could be converted to F408-InP:Zn MSCs, then to F393-InP:Zn MSCs. As the conversion proceeded, evolution from uni-molecule-like to QD-like characters was observed. Early stage MSCs showed active inter-state conversions in the excited states, which is characteristics of small molecules. Later stage MSCs exhibited narrow photoinduced absorptions at lower-energy region like QDs. The crystal structure also gradually evolved from polytwistane to more zinc-blende.

Cl-incorporated InP MSC-399? 2. What is the composition-structure-property relationship for Zn-incorporated InP MSC-360, Znincorporated InP MSC-408, and Zn-incorporated InP MSC-393? 3. Is it possible to provide the very experimental evidence on the transformation of Cl-incorporated InP MSC-399 to QDs, and Zn-incorporated InP MSC-408 to Zn-incorporated InP MSC-393 to QDs? 4. On Page 17, the authors wrote for Figure S18 that the "a continuous blue-shift indicates the existence of numerous intermediate species between Zn-incorporated InP MSC-408 and Zn-incorporated InP MSC-393". What is the nature of the evolution? The authors claimed that the two clusters have different compositions. 5. On Page 21, the authors should provide in depth explanations on how to control the Zn atom to be on the surface, and the definition of "periphery" (does not affect the absorption peak position) vs "quantum-confined core" (does affect the absorption peak position). What is the composition difference between ODPA-capped InP MSC-397 and Zn-incorporated InP MSC-393? For Zn-incorporated InP MSC-360, 408, and 393, do they have different cores? 6. What is the composition-structure-property relationship for MA-capped InP MSC-386 (exhibiting one absorption singlet) and ODPA-capped InP MSC-397 (displaying one absorption doublet)? 7. For the InP MSC-386 to Cl-incorporated InP MSC-360 evolution shown in Figure 1c at 80 C, the authors attributed the absorption peaking at 416 nm as one intermediate species . The half width at half-maximum (HWHM) of "416-IS" is 20 nm, the same as that of Cl-incorporated InP MSC-360. Can this intermediate be . For the evolution from InP MSC-386 to Cl-incorporated InP MSC-360 shown in Figure S3b at 50 C and Figure S3c at 25 C, the authors claimed (Lines 141 to 142 on Page 6) "intermediates which have not been properly detected". Regarding the evolution from InP MSC-386 to Cl-incorporated InP MSC-360 at these different temperatures studied as shown in Figures 1c, S3b, and S3c, any explanation why the experimental observation on "intermediates" is influenced by the temperatures? 9. For the Page 9 equation, any numbers can be assigned to x? 10. For the Page 9 equation, if y = 0, the reaction becomes an anionic exchange reaction between P and Cl. Any explanation? 11. All the figures should be improved, including the color usage, the legend font size, and dpi. For example, the Figure 1 legend font size is far too small. 12. The format of the refs should be edited according to the magazine required format.
Comment 3: EDX appears to be the primary composition method used in this manuscript. Was the EDX acquired on the TEM samples? It seems surprising to me that the clusters would survive under the harsh beam conditions needed to acquire EDX analysis. It is more typical to assess compositions by a method like ICP-MS (or OES/AES). Do those data compare favorably with the observed EDX numbers? Overall the reported values seem reasonable and consistent with what others have seen, but I am surprised the data are reliable for such fragile objects.
Our response: We thank the reviewer for the insightful comment. Elemental analysis by ICP-MS or OES/AES methods typically requires acid digestions, and meanwhile phosphorous-containing compounds (especially InP) often evolve volatile phosphorous compounds (eg. PH 3 ). Technically, proper pre-oxidation and subsequent digestion should yield accurate phosphorous contents of InP MSCs by ICP-AES or by ICP-OES. However, in our cases, the values obtained multiple times by ICP-AES fluctuated significantly. On the other hand, our MSCs were quite stable under e-beam (200 kV) at least for a few minutes, and the EDX data were reproducible.
Comment 4: What was the rationale for adding ODPA in the synthesis of the zinc containing clusters? This seems like a non-intuitive choice and as a reader I would appreciate understanding what the authors were thinking there. What happens when no ODPA is used? Given the precedent for ODPA-capped InP clusters reported in reference 6, do the authors have controls that help us understand how much of the optical properties come from ligand effects versus the incorporation of zinc?
Our response: We thank the reviewer for the insightful comment. When no ODPA was used, zinccontaining clusters grew very rapidly to NPs before any clusters could be properly isolated. Figure R1 shows the control reaction which omitted ODPA in direct F393-InP:Zn MSC synthesis. As heating up the reaction mixture, broad absorption was observed at 200°C which quickly turned into NPs ( Figure   R1a). The final products were InP NPs of a few nm in size ( Figure R1b).
Indeed, adding ODPA in the synthesis of zinc-containing clusters is non-intuitive. ODPA seems be a strongly binding ligand to InP MSCs, and can stabilize certain InP MSCs as inhibiting the further growth to NPs. We used ODPA for InP:Zn MSCs simply because we were not able to synthesize Znincorporated InP MSCs without using ODPA.
Regarding the issue of how much of the optical properties come from ligand effects versus the incorporation of zinc, we believe that ODPA is involved in formation of the inorganic skeleton of MSCs which strongly governs the optical properties. In other words, we speculate that ODPA (as bound on MSC surfaces) made both F397-InP MSC (in reference 6) and our F393-InP:Zn MSC to share an inorganic skeleton which in turn shared similar optical properties such as the doublet features.
386-InP MSC can be converted into F397-InP MSC upon heating with ODPA addition. Figure R2 shows the absorption change during the conversion from 386-InP MSC to F397-InP MSC. The conversion is made via intermediates which shows the absorption peak at around 360 nm. It is noted that F397-InP MSC was formed after apparent dissolution of the intermediates at 360 nm and that the doublet optical feature appeared only after the dissolution of the 360 nm intermediates ( Figure R2). This suggests that the inorganic skeleton of F397-InP MSC can be very different from that of 386-InP MSC. XRD pattern of F397-InP MSC is also very different from that of 386-InP MSC ( Figure R3).
As for another control, we have performed experiments of adding small amounts of ODPA to 386-InP MSC. The reaction was monitored by absorption and 31 P NMR ( Figure R4). Up to 10 equivalent ODPA addition, the absorption peak did not noticeably change. However, it slightly broadened in the UV region presumably due to the ODPA binding and partial etching as monitored by NMR. These results show that binding of ODPA on InP MSC does not alter the optical properties as significantly as the inorganic skeleton structural change.
Based on the additional experimental data, the revised manuscript has been also properly expanded as shown below.
(Revised manuscript page 16, omission of what precedes) In contrast, T R and the amount of ODPA critically affected the fate of product MSCs. When no ODPA was used, the growth was uncontrollable and yielded InP NPs ( Figure S16). Reducing the amount of ODPA by quarter and elevating T R to 300°C for 3.5 h yielded F408-InP:Zn MSCs, which had LEET at 408 nm ( Figure S17)     Response to reviewer 2 Comment: In this manuscript, the authors reported the synthetic pathways of doped InP clusters. By introducing Cl atom or Zn atom either in situ or in post synthesis step, two families of InP clusters InP:Cl and InP:Zn with different size have been prepared. Furthermore, in an individual family, the authors observed the conversion of clusters between different sizes. Although, the authors did present a large amount of results in this manuscript, the novelty is still missing. This manuscript looks more like a following work of the one published on JACS in 2016 (JACS, 2016(JACS, , 138, 1510. Besides, the characterization is not complete. For example, what is the actual crystal structures of the new InP clusters? How does the Cl or Zn bind with clusters, on the surface or in the cluster? Is the method present here a general method and does it works for other cluster system besides InP? Therefore, with these concerns in my mind, I don't recommend to publish this manuscript on Nature Communication. Our response: Admittedly, we do not have actual crystal structures of the new InP magic size clusters (MSCs). However, we do not believe that our works are like a following work of the one published on JACS in 2016 (JACS, 2016(JACS, , 138, 1510. The 2016 JACS paper reported a crystal structure of 386-InP MSC using single crystal crystallography. 386-InP MSC is a single species. To the contrary, all our samples are a family of plural isomers. We have clearly stated it in the beginning of our manuscript, and it is also clearly shown in the nomenclatures that all the samples start with 'F' as representing 'family'. All our samples have the mass weight of tens of kDa. With the co-existence of isomers, it is practically impossible to grow single crystals suitable for crystallography. We believe that characterizations of each and every MSCs (and of each isomers to be precise) using single crystal crystallography is beyond the scope of our manuscript.
Our manuscript reports conversions among MSCs. We report two series of MSC series and the evolutions from uni-molecule-like to colloidal-QD-like characters observed as the conversions proceed. Early stage MSCs possess the low-symmetry polytwistane structures, and they show unimolecular-like characters such as active inter-state transition in the excited state dynamics. Later stage MSCs exhibit more QD-like characters: crystal structure evolution to more zinc-blende toward the bulk thermodynamic equilibrium that possesses higher symmetry, LO and TO Raman bands, Auger and trap-associated exciton dynamics, and narrow photoinduced absorptions at the lower-energy region.
Regarding the issue "how the Cl or Zn binds with clusters, on the surface or in the cluster?", we believe that Cl and Zn reside in the surface and periphery of the MSCs. Upon incorporation of Cl ions, XPS spectra showed the In 3d 5/2 peaks increased by 0.56 eV for F360-InP:Cl MSCs and by 0.77 eV for F399-InP:Cl MSCs when compared to that of 386-InP MSC. This large increase in the binding energies is indicative of the direct bindings of Cl to In atoms. Considering the valency of Cl ion, we believe it is reasonable to believe that Cl atoms exist on the surface of MSC binding directly to adjacent In atoms. The case of InP:Zn MSC is less clear. F393-InP:Zn MSCs show the intensified and broader PL emission over 445 nm. In addition, the band-edge emission band of F393-InP:Zn MSCs is 1.3 times broader than that of F397-ODPA InP MSCs ( Figure S26 and Figure S27). These emissive features suggest that the incorporated Zn cations facilitate the carrier trapping process and intensifies the trap emission in F393-InP:Zn MSC. In the TA data, the broader photoinduced absorption signal in the lower-energy region of band-edge bleaching of F393-InP:Zn MSC with the longer-lived TA decay profile well depict the denser trap states by the Zn incorporation ( Figure 8e). Based on the optical properties, we speculate that the incorporated zinc cations may be mostly located at the periphery of the InP MSC.
Regarding the issue if our method present here is a general method and works for other cluster systems besides InP, unfortunately our method is not directly expandable to other cluster systems.
However, other dopants could be incorporated into InP clusters using our method. For an example, InP:Sn MSC and InP:Mn MSC could be successfully obtained (See Figure R5 for absorption spectra of InP:Sn MSC and InP:Mn MSC). We have tried to extend our method to other cluster systems including GaP MSCs and InAs MSCs. In our attempts to synthesize GaP MSCs, the growth was uncontrollable and NPs were typically obtained ( Figure R6). In our attempts to synthesize InAs MSCs using our method, uncontrolled growth or amorphous clusters of size 1-2 nm were obtained. Similar results were also reported by others (Chem. Mater. 2016, 28, 8119-8122). It might be possible to modify our method and extend it for other MSCs however we believe it to be beyond the scope of this manuscript.  Our response: We thank the reviewer for the insightful comment. Due to the discrete nature of nanoclusters, size or composition of clusters cannot be continuously tuned and synthesized. As the result, it is practically impossible to continuously change the composition and elucidate the structure and properties to clearly map the composition-structure-property relationship as conventionally performed for bulk materials. In addition, the small size of nanoclusters makes the structural analysis extremely difficult because of the lack of many repeating units. By the size, our nanoclusters should have only a few repeating unit cells at best within any single cluster. Technically, single crystal (made of periodic superlattice of nanoclusters) crystallography is the ultimate way to elucidate the nanocluster structure. However, all our samples are a family of plural isomers as all our samples start with 'F' as representing 'family'. They have the mass weight of tens of kDa. With the co-existence of isomers, it is very difficult to grow single crystals of each MSCs (and of each isomers) suitable for crystallography. We had to resort to spectroscopic tools for the structural analysis of our nanoclusters, which only afforded limited structural information. As the reviewer pointed out, our structural analysis is rather limited and as the result the final conclusion sentence is admittedly an overstatement. We have altered the sentence to tone down the claim in the revised manuscript as shown below. Raman spectroscopy suggests that F360-InP:Cl MSC is structurally close to polytwistane, whereas F399-InP:Cl MSC is more matched with zinc-blende. F360-InP:Cl MSC possesses the polytwistane structure with pseudo-C 2 symmetry. The polytwistane structure has large fluctuations in the length and angle of In-P bonds, and In atoms in polytwistane structure are irregularly configured with the pseudo-octahedral and tetrahedral geometries. We conjecture that relatively small F360-InP:Cl MSC takes polytwistane structure to gain favorable bond energies between In and P atoms. The larger F399-InP:Cl MSCs should evolve to more zinc-blende structure, which is close to bulk thermodynamic equilibrium that possesses higher symmetry. The composition-structure relationship strongly determines the optical properties. F360-InP:Cl MSCs is difficult to have quantum confinement along with their heterogeneous structures and they show the molecule-like character. differentiate P atoms in the inorganic InP structure and P atoms in the phosphonate ligands, we tried to strip the phosphonate ligands without altering the inorganic structure however it was never successful. According to the XRD and Raman data, F360-InP:Zn MSC has the structure close to polytwistane whereas F408-InP:Zn MSCs and F393-InP:Zn MSCs share an inorganic structure which can be assigned to a slightly distorted zinc-blende structure. Similar to the case of Cl incorporated InP MSCs, structural evolution from low symmetry polytwistane to high symmetry zinc-blende was observed.
Because of the limited composition information and also because of our intention to compare Zn MSC. These PL features indicate that the trap-associated emission is more enhanced in F408-InP:Zn MSC, which manifests smaller Zn content of F393-InP:Zn MSC than F408-InP:Zn MSC. This is also well in consistency with the fact the conversion from F408-InP:Zn MSCs to F393-InP:Zn MSCs involves loss of Zn complexes (i.e., detachment of zinc stearate complexes), and as the result F393-InP:Zn MSC has smaller Zn to In ratio than F408-InP:Zn MSC. Therefore, the smaller Zn content of F393-InP:Zn MSC must have formed less trap sites than F408-InP:Zn MSC and showed the abovementioned properties as nicely showing the composition-structure-property relationship.
Based on the additional experimental data, the revised manuscript has been also properly expanded as shown below.
(Revised manuscript page 20, omission of what precedes) From the peak intensities of the two characteristic peaks, the number ratio of carboxylate to phosphonate bound onto F408-InP:Zn MSCs was deduced to be 1:5.9, which was eventually changed to solely phosphonate for F393-InP:Zn MSCs.
Emission of F360-InP:Zn MSC was very broad and weak, whereas those of F408-InP:Zn MSC and When the temperature reached 100°C, conversion to F360-InP:Cl MSC was only observed, which was followed by complete conversion to F399-InP:Cl MSC at 150°C. Upon further heating to 200°C, broad and featureless absorption was observed at 180°C and higher. The final product was collected and characterized as InP QDs. XRD returned the crystal structure as zinc-blende InP ( Figure R8b) and their average size was measured as 7.50 nm by TEM ( Figure R8c).
Contrary to the case of Cl-incorporated InP MSCs, F393-InP:Zn MSCs could not be converted into QDs by heating. F393-InP:Zn MSC has the notable thermal stability, and it does not lose any integrity upon continuous heating at 300°C for weeks. Further heating of F393-InP:Zn MSC to a higher temperature than 300°C eventually resulted in the complete dissolution of clusters. However, when F393-InP:Zn MSCs were heated with addition of extra cationic or anionic molecular precursors such as In(MA) 3 or (TMS) 3 P, F393-InP:Zn MSCs could be further grown into InP QDs.
When In(MA) 3 were added into F393-InP:Zn MSCs at room temperature, the absorption peak redshifted to 404 nm and the second transition disappeared ( Figure R9a). Spectroscopic analysis suggests the bridging bondings between two oxygen atoms of an ODPA and surface In atoms on F393-InP:Zn MSCs was opened and inserted by an incoming In(MA) 3 molecule ( Figure R10). FT-IR bridging mode of carboxylate indicates two oxygen atoms of an ODPA are coordinated to two different In atoms ( Figure R11). A few In(MA) 3 were additionally attached to the surface of F393-InP:Zn MSCs, which dramatically reduced the thermal stability. When F393-InP:Zn MSCs were added by 70 equivalents of In(MA) 3 and heated up, the red-shift of absorption peak and disappearance of the second transition were followed by further gradual red-shift. At 300°C, the absorption profile became broad and featureless showing the tail down to 660 nm. InP QDs of atypical shapes were obtained ( Figure R9b). The destabilized cationic-precursor-adsorbed F393-InP:Zn MSCs were thought to have dissembled into monomers and fragments, which subsequently re-assembled and grew into the atypical InP QDs.
When the anionic precursor (TMS) 3 P was added to F393-InP:Zn MSCs, the absorption did not change up to 60°C ( Figure R9c). Further heating over 200°C yielded red-shift of the absorption, and the absorption became featureless at 270°C. The absorption was also significantly broadened with the tail reaching over 500 nm, which is indicative of InP QD growth. The final product was aggregated InP QDs ( Figure R9d). The initial red-shift upon the addition of (TMS) 3 P was due to the silylester reaction between (TMS) 3 P and phosphonate and carboxylate ligands on F393-InP:Zn MSCs. The    Octadecylphosphonate, SA: Stearate). Both absorption spectra and FT-IR spectra confirm that adding the larger amount of ODPA significantly accelerate the conversion to F393-InP:Zn MSCs ( Figure   R12). When 0.925 ODPA/SA ratio was used, carboxylate ligands were completely removed in 24 hrs ( Figure R12a and Figure R12b). To the contrary, 0.875 ODPA/SA ratio needed 72 hrs for such conversion. During the exchange reaction, narrowing of the P-O(H) vibration peaks between 1000 and 1200 cm -1 was observed, which attributes to the change to deprotonated ODPA species from mixed ODPA species of (partially) protonated and deprotonated ODPAs. All the deprotonated ODPA should bind to the surface In atoms more strongly than partially protonated ODPAs, and this makes the final F393-InP:Zn MSCs to be very stable. ODPA is known to be a stronger binding ligand than SA or zinc stearate (J. Phys. Chem. Lett. 2011, 2, 145-152). The nature of the evolution can be considered as the enthalpy gain by exchanging the ligand to a stronger ODPA ligand. Admittedly, direct use of the terms such as core and periphery without proper characterizations is an overstatement and can be misleading. We have corrected the text in the revised manuscript as shown below to avoid such confusions.
We were not able to find a way to isolate InP MSC-397 from its organic phosphorous-containing byproducts. As the result, we were not able to obtain proper elemental analysis. InP MSC-397 was first reported by Cossairt group without any elemental analysis till now, we assume they suffered from the same problem. The Zn content difference is the main composition difference between ODPA-capped InP MSC-397 and Zn-incorporated InP MSC-393.
The purification problem of InP MSC-397 also had inhibited us from obtaining the XRD structural data on InP MSC-397. To respond to the reviewer's comment, we have performed additional experiments and successfully obtained a sample partially purified but enough to acquire XRD signals. Our response: We thank the reviewer for the comment. As stated above in response to comment 5, we were not able to obtain elemental analysis data on InP MSC-397. We have obtained XRD patterns for InP MSC-386 and InP MSC-397 ( Figure R14). 386-InP MSCs showed the polytwistane structure whereas InP MSC-397 was zinc-blende. We believe the structure has changed from low symmetry mins from the 25°C reaction showed the red-shifted absorption because of the higher content of 416-IS than the one previously obtained under conventional 80°C condition ( Figure R15a). Aliquots were taken at 10 mins, 6 hrs, and 14 hrs during the conversion reaction from InP MSC-386 to InP MSC-360 at 25°C. Absorption, XRD, and elemental analysis were characterized for those three aliquots along with InP MSC-386 and InP MSC-360 ( Figure R15b, 15c, and 15d). The 416-IS contains significant amount of Cl atoms and is considered as another Cl-incorporated InP MSC.
Based on the additional experimental data, the revised manuscript has been also properly expanded as shown below. to Cl-incorporated InP MSC-360 at these different temperatures studied as shown in Figures 1c, S4b, and S4c, any explanation why the experimental observation on "intermediates" is influenced by the temperatures?
Our response: We thank the reviewer for the helpful comment. At 80°C evolution from InP MSC-386 to Cl-incorporated InP MSC-360, an isosbestic point was observed. However, the isosbestic point was missing at the same reaction at 25 or 50°C. Based on the results, we assumed that the concentration of intermediates or the kinds of intermediates should be different by the different reaction temperatures. Different reaction temperature may affect differently for the kinetics of fragmentations and assemblies. Different reaction temperature may also yield different fragment species, partial assemblies of which result in the same final product. Our response: We thank the reviewer for the comment. We have corrected the format as suggested.

REVIEWERS' COMMENTS:
Reviewer #1 (Remarks to the Author): The manuscript appears to have been revised with great care and attention to the reviewer comments. I believe the work now merits publication.
Reviewer #3 (Remarks to the Author): The authors have tried hard to "answer" the concerns and suggestions raised. Importantly, they need to "improve" their manuscript, based on the concerns and suggestions raised. Although some explanations or interpretations have been provided, the reviewer believes that the authors are able to push harder to do a better job, so that the study will have a bigger impact. In Discussion, the authors should better reflect what is actually known and unknown about these clusters and the implications. Also, the authors need to improve their English.
Before Reviewer 3 Comment 1, "The structural studies do not seem to be convincing." The authors may address the reason in their revised manuscript why the cluster structures are difficult to identify at present (as partially indicated by The future of colloidal semiconductor magic-size clusters. ACS Nano 2020, 14, 1227-1235).
It is unclear to the reviewer "Our findings highlight the mechanistic insight that MSCs which appear before the growth to QDs transform by multi-step sequential conversions of MSCs or evolutions of MSCs." Do they want to say "Our studies introduce the possible pathways, regarding the evolution of MSCs which may include multi-step sequential conversions among MSCs"? Comment 1: the significant figure should be taken care of for the compositions. The authors should address "the composition-structure-property relationship between Cl-incorporated InP MSC-360 and Clincorporated InP MSC-399" in their revised version. In Discussion?
Comment 2: the significant figure should be taken care of for the compositions. On Page 12 of the pointby-point response file, the authors seem to write down a lot. The reviewer recommends that the authors should concisely summarize "the composition-structure-property relationship for Znincorporated InP MSC-360, Zn-incorporated InP MSC-408, and Zn-incorporated InP MSC-393" in their revised version. In Discussion?
Comment 3: On Page 14 of the response letter, for the response of reviewer 3, comment 3, the authors claimed that "When In(MA)3 were added into F393-InP : Zn MSCs at room temperature, the absorption peak red-shifted to 404 nm and the second transition disappeared ( Figure R9a)." Here is one example that the authors should carefully improve their English.
Any explanation for the disappearance of the second transition? Is it due to disappearance of the MSCs and the growth of QDs? On Page 15, Figure S8 was presented, showing a sharp absorption peak obtained for a 150 C sample. Why Figures R8 and R9 are presented, while the edited words on Page 15 mentioned Figures S6 and S7? Again, for the present Figure R8 on Page 15, the two traces for the 110 and 180 C samples look similar.
Comment 4: The authors should make a clear statement in their revised version "What is the nature of the evolution? The authors claimed that the two clusters have different compositions".
Comment 5: The authors should straightforwardly comprehend in their revised version "do they have different cores?" Comment 6: The authors should provide an answer in their revised version "the composition-structureproperty relationship for MA-capped InP MSC-386 (exhibiting one absorption singlet) and ODPA-capped InP MSC-397 (displaying one absorption doublet)?" For example, if the authors do not have an answer, that is fine. The authors may point out such a concern but without an answer, saying "we do not know why MA-capped ….." This is a kind reminder for to readers to think as well.
Comment 7: The authors edited in their revised manuscript "…..We speculate that these partially-etched 386-MA InP MSCs are 416-IS." Do the author mean "due to more Cl incorporated"? The reviewer hopes that the authors is able to make their point more directly.
Comment 8: The authors should concisely explain in their revised version "why the experimental observation on "intermediates" is influenced by the temperatures?" Comment 9: The authors should provide an answer in their revised version "any numbers can be assigned to x" Comment 10: The authors should improve their revised version answering " if y = 0, …?" Comment 11: The authors should improve their figure presentation. For example, 50 C can be the title of Part a, and does not need to be present for each trace. The four subfigures can be arranged in two panels (two top and two bottom), so do Figure 2. There is no need to use "F", while keep "MSCs". The author can use, for example, MSC-360 Cl : InP, for their main text and for their figures.

REVIEWERS' COMMENTS:
(Response to reviewer 3) General Comment 1: The authors have tried hard to "answer" the concerns and suggestions raised. Importantly, they need to "improve" their manuscript, based on the concerns and suggestions raised.
Although some explanations or interpretations have been provided, the reviewer believes that the authors are able to push harder to do a better job, so that the study will have a bigger impact. In Discussion, the authors should better reflect what is actually known and unknown about these clusters and the implications. Also, the authors need to improve their English.
Before Reviewer 3 Comment 1, "The structural studies do not seem to be convincing." The authors may address the reason in their revised manuscript why the cluster structures are difficult to identify at present (as partially indicated by The future of colloidal semiconductor magic-size clusters. ACS Nano 2020, 14, 1227-1235).
Our response: We thank the reviewer for the insightful comment. Thanks to the precious comments, we were able to significantly improve the manuscript by the revision. The Discussion session has been properly expanded as shown below. General Comment 2: It is unclear to the reviewer "Our findings highlight the mechanistic insight that MSCs which appear before the growth to QDs transform by multi-step sequential conversions of MSCs or evolutions of MSCs." Do they want to say "Our studies introduce the possible pathways, regarding the evolution of MSCs which may include multi-step sequential conversions among MSCs"?
Our response: We thank the reviewer for the insightful comment. As the reviewer nicely pointed out, our studies introduce the possible pathways regarding the evolution of MSCs which may include multi-step sequential conversions among MSCs. Evolutions of MSCs are typically observed during early stages of nucleation and growth of QDs. It may not be a coincidence that MSCs become more and more QD-like as getting close to QDs. In this vein, we had stated that "Our findings highlight the mechanistic insight that MSCs which appear before the growth to QDs transform by multi-step sequential conversions of MSCs or evolutions of MSCs.". However, admittedly it was an overstatement. In the revised manuscript, we have revised the discussion as shown below. Our response: We thank the reviewer for the comment. In the comment "On Page 15, Figure S8 was presented, showing a sharp absorption peak obtained for a 150C sample.", we believe the S8 is a typo for Figure R8. In Figure R8  Our response: We thank the reviewer for the comment. We believe that F408-InP:Zn MSCs and F393-InP:Zn MSCs share a similar inorganic skeleton close to zinc-blende core but differ in Zn trap site density and that F393-InP:Zn MSCs and F397-ODPA InP MSCs may share a more similar core (both zinc-blende and electronically doublet absorption feature) but differ in Zn trap site existence.
We believe the initial F360-InP:Zn MSCs to have the inorganic skeleton disparate from the rest of the series. We have properly added the discussion in the revised manuscript as shown below.
(Revised manuscript page 22, omission of what precedes) Here, compared to F397-ODPA InP MSCs, the blue-shifted absorption spectrum of F393-InP:Zn MSCs may delineate the decreased effective size of 'quantum-confined core', which suggests that the incorporated Zn cations may be mostly located at the 'periphery' of the MSC. Such Zn incorporations can also reduce the symmetry of F393-InP:Zn MSCs, which yielded the smaller absorption cross section value ( Figure 9f) and broad absorption and emission spectral features. We believe that F408-InP:Zn MSCs and F393-InP:Zn MSCs share a similar inorganic skeleton close to zinc-blende core but differ in Zn trap site density and that F393-InP:Zn MSCs and F397-ODPA InP MSCs may share a more similar core (both zinc-blende and electronically doublet absorption feature) but differ in Zn trap site existence. We believe the initial have an answer, that is fine. The authors may point out such a concern but without an answer, saying "we do not know why MA-capped ….." This is a kind reminder for to readers to think as well.
Our response: We thank the reviewer for the comment. (omission of what follows) Comment 9: The authors edited in their revised manuscript "…..We speculate that these partiallyetched 386-MA InP MSCs are 416-IS." Do the author mean "due to more Cl incorporated"? The reviewer hopes that the authors is able to make their point more directly.
Our response: We thank the reviewer for the helpful suggestion. We believe that these partially Comment 11: The authors should provide an answer in their revised version "any numbers can be assigned to x" Our response: We thank the reviewer for the comment. For an example, in conversion from 386-InP MSCs to F360-InP:Cl MSCs, the x value can be assigned approximately as six by using the elemental analysis data of the reactant and product. As suggested, the revised manuscript has been also properly expanded as shown below.
(Omission of what follows) Comment 12: The authors should improve their revised version answering "if y = 0, …?" Our response: We thank the reviewer for the comment. As suggested, the revised manuscript has been properly expanded with the case y=0 as shown below. Our response: We thank the reviewer for the comment. Figure 1 and Figure 2 were improved in the revised manuscript and we have also corrected the x axis and style in Figure 8.   ).

Comment 14:
There is no need to use "F", while keep "MSCs". The author can use, for example, MSC-360 Cl:InP, for their main text and for their figures.
Our response: We thank the reviewer for the comment. Indeed, there is no need to use "F" in our nomenclature. However, we believe the definition of MSC does not yet have a consensus in the scientific community. We intended to use "F" to clearly define our samples as families of plural isomers. For an example, we intentionally did not use "F" for 386-InP MSC because it is a single species. This may appear odd but we wish to express the terms with such rigorousness.