Normal & reversed spin mobility in a diradical by electron-vibration coupling

π−conjugated radicals have great promise for use in organic spintronics, however, the mechanisms of spin relaxation and mobility related to radical structural flexibility remain unexplored. Here, we describe a dumbbell shape azobenzene diradical and correlate its solid-state flexibility with spin relaxation and mobility. We employ a combination of X-ray diffraction and Raman spectroscopy to determine the molecular changes with temperature. Heating leads to: i) a modulation of the spin distribution; and ii) a “normal” quinoidal → aromatic transformation at low temperatures driven by the intramolecular rotational vibrations of the azobenzene core and a “reversed” aromatic → quinoidal change at high temperatures activated by an azobenzene bicycle pedal motion amplified by anisotropic intermolecular interactions. Thermal excitation of these vibrational states modulates the diradical electronic and spin structures featuring vibronic coupling mechanisms that might be relevant for future design of high spin organic molecules with tunable magnetic properties for solid state spintronics.

"reversed" aromatic to quinoidal change at high temperatures can be activated by a bicycle pedal motion. A unique and spin-vibration mechanism has been revealed and clearly discussed. This work is of general interests for chemists majoring in diradical chemistry and of high value in organic spintronics. Besides, the paper is well organized with a high-level scholar presentation. Before its consideration for formal publication, I would suggest some revisions to be made.
1. Although the rich transformations between aromatic and quinoidal canonical forms can be clearly observed in the single crystal, the magnetic susceptibility measurements of CAR should also be performed in the crystal state. It is known that there may exist several phases in multicrystalline or power states that may affect the discussed transformations. So please indicate this information in the paper.
2. Following the first suggest, the other measurements should also be performed under the crystal state. If it is not the case or it is difficult to perform such measurements, I would suggest the authors the check XRD of the solid and confirm the molecular stacking is consistent in crystals and powers.
3. Since the triplet state is thermally activable, we should keep in mind that the triplet state and single state coexist in the crystal and the geometry information actually is a statistical data. In other words, with the increasing temperature, the triplet state will populate, which may affect the molecular geometry and maybe the packing. The authors are suggested to carefully consider this possible issue.

A clear explanation on the varied intramolecular torsional motion under different temperatures
should be provided in the abstract, introduction, and conclusion parts, which may induce more consideration of this design and get a larger influence. 5. About the importance of such a unique phenomenon, the authors are suggested to put more efforts. I think it is not just a new phenomenon, and my feel is that it should be quite meaningful for organic spintronics considering its tunable spin distribution.
6. There are a few typos in the main text and in the information. Please check and correct them. For example, in "… a solid state building scaffold able to bringing intramolecular torsional mobility…", "bringing" should be corrected; In "… bicycle pedal motion in in azobenzene (B)…," please remove one "in".
Reviewer #3: Remarks to the Author: Shen et al reported that they prepared new azobenzene-based organic diradical. The diradical has been proven as an open-shell singlet ground state by EPR, SQUID and DFT calculation. They claimed that the geometry diradical is flexible with temperature, accompanied with spin density changes. However, based on the std (3-4, and higher at high T in cifs) of bond lengths of X-ray structures, it is hard to see there is any change of geometries at different temperatures. The line of XT is very smooth, consistent with the property of a singlet diradical, indicating there is no changes of the electron coupling. Since the std is so large, that DFT (spin density and y values) based on X-ray structures result is totally unreliable. In words, the finding is just a normal diradical with an open-shell singlet ground state, and basically no alternation of geometries with temperature based on crystal structures. The current version of the manuscript is unpublishable and is supposed to be rewritten. Several points need consideration: 1. ST gap difference between determined values by EPR and SQUID (1.31 vs 3.41 Kcal/mol) is large. One of them must be incorrect.
2. Thermal driven geometry change of organic diradicals have been reported before, especially by Juan and Wang etal. Those literature work should be clearly described in the text.
3. STD of geometry parameters shod be added in the text and tables.

REVIEWER #1:
Reviewer point 1. My major concern is that the only experimental data that supports this change is provided by a single crystal structure. The XRD data is generally good, with some exceptions as noted below, but this is still only a single piece of experimental evidence for their conclusions. The magnetic data does not show any changes which would be consistent with the proposal. What other data can the authors acquire that would support this? Can they get variable temperature spectroscopic or vibrational data to support this change? I think that the changes are subtle enough that some corroborating data would strengthen their arguments.
Authors´s response 1. Solid state Raman spectra in the polycrystalline sample of CAR have been obtained as a function of the temperature (for completeness two excitation Raman lines have been analyzed). The new data, their description and discussion have been incorporated in the new version of the article in pages 9-11 including a new Figure 4. Moreover, quantum chemical calculations of the Raman spectra on the experimental molecular structures at different temperatures have been obtained which nicely reproduced the experimental Raman spectra which therefore support the new discussion. The Raman spectra nicely confirm the previous arguments in the paper and represent a new piece of evidence (such as the reviewer required) in line with the whole discussion. We very much appreciate the comment of the reviewer.
The new theoretical Raman spectra have been included in the new version of the electronic supporting information (new Supplementary Figures 12-14). The text with the Raman discussion has been incorporated in the revised version of the paper in pages 9-11. This text added in the main body of the article follows below: Raman spectra with the 633 and 785 nm laser excitations have been measured as a function of the temperature and the spectra at 130, 290 and 340 K are shown in Figure 4.   [(CC)phO] bands (A1/A2 in Figure 4) are similar in line with similar quinoidal contributions in the two types of benzene rings. In the 290 K Raman spectrum, the (CC)phO band clearly decreases regarding the vicinal (CC)ph one, in line with a gaining of aromatic character in the azobenzene rings. In the 340 K spectrum, however, the (CC)phO band regains intensity relative to its parent one in accordance with a partial recovery of quinoidal character in the azobenzene rings. The same is found in another experiment with the 633 nm laser excitation also in Figure 4. In the medium wavenumber region of the 785 nm Raman spectra, a similar discussion can be done with the relative intensities of the two (CN)azo bands (i.e., denoted as (CN)azo1 and (CN)azo2); hence, heating from 130 to 290 K produces a redistribution of the intensity consisting in a decrease of the (CN)azo2 band an effect that is reversed by further heating to 340 K in accordance with a sequence of quinoidalaromatic transformation on 130240 K, followed by an aromaticpseudo-quinoidal post-conversion at 340K. For the experiment with the 633 nm laser excitation in the medium wavenumber region, the thermal evolution of the Raman spectra, though with less clarity, also shows the same behavior. Authors´s response 2 and 3. The ESD of the bonds and angles were marked in the text and tables in the revised manuscript as follows: In page 5 and in line 2, "38.76°" was replaced with "38.8(2)°", "5.78°" was replaced with "5.8(5)°".
Reviewer point 4. Some additional information on the refinement of the crystal data would be valuable. The ellipsoids look a little strange, what restraints were applied? If large restraints were applied to make the ellipsoids more isotropic they should be relaxed, or the authors should at least compare the changes in bond lengths with or without the restraints. This is particularly crucial given the importance of the structural data in the author's arguments.
Authors´s response 4. For the single crystal data at 130 K and 200 K, no additional geometric and ADP restraints were applied in the structure refinement.
For the single crystal data at 250 K and 290 K, overall ADP restraints were used during the refinement. Removing them did not affect the structure. For the single crystal at 340 K, the N atom ellipsoid seems strange compared to surrounding atoms, due to some disorder and EDAP constraint was used to make the ADP to look more reasonable. Removing them makes N=N bonds a little shorter. Very nicely the appearance of disorder in the 340K structure is compatible with the activation of the bicycle pedal motion which simultaneously produces the aromaticpseudoquinoidal electronic change. To expand this discussion, please see response to point number 5 below for this referee.
This new discussion has been added in the revised version of the article in page 5. In addition, in page 5 of revised electronic supplementary information, the corresponding contents and discussion of the structural disorder were added.
Reviewer point 5. The authors invoke a pedal motion to explain the different structures (as cited in reference 13). However, as stated in ref. 13, "disorder in crystal structures is an important indicator of pedal motion in the crystal". Why do the authors not see any evidence of this disorder? If a pedal motion were present, wouldn't the atom positions be disordered? It is possible that only one conformer would be favored, as discussed in the cited reference, but the author's structural data seems to suggest that there are two conformers close in energy. I might expect that some evidence of disorder would appear at 340 K. At the very least, this needs to be explained more clearly. These results indicate that a conformational equilibrium takes place in the crystal of CAR at every temperature. Therefore, this temperature dependence of populations of the conformers is a further proof of the existence of the pedal motion in CAR. Though the number of temperature data and relative populations is small, these have been fitted to a van´t Hoff plot producing an energy gap between the states in equilibrium of 1.26 kcal/mol which corresponds to roughly 440 cm -1 . Though this value must be taken with caution given the small number of data in the plot, it already suggests the existence of a low energy vibrational mode around 440 cm -1 that would correspond to the torsional plus double pedal motion that couples to the electronic structure. However, we did not go in more details due to the weakness of the plot but leaves here this description for the benefit of the reading of the reviewer. In page 6 of the revised version of electronic supplementary information, the explanation about the change of the two conformers from 130 to 340 K was added.
Reviewer point 6. What is the feature at 55 K in the magnetic data?
Authors´s response 6. The SQUID experiment has been measured again and the new XT and T curve is shown in Figure R2, which indicates that the feature at 55 K for the old spectrum was accidental. In the new version of Figure 3 the new SQUID data have replaced the old one.

Figure R2
Temperature-dependent plots of χT versus T and fitted χT-T curve for CAR measured at 1.0 T from 2 to 400 K.

Reviewer point 7.
It seems that some more recent citations to aggregation or thermally modulated diradical character are absent. Examples include: Chem.

REVIEWER #2:
Reviewer point 1. Although the rich transformations between aromatic and quinoidal canonical forms can be clearly observed in the single crystal, the magnetic susceptibility measurements of CAR should also be performed in the crystal state. It is known that there may exist several phases in multicrystalline or power states that may affect the discussed transformations. So please indicate this information in the paper. Following the first suggest, the other measurements should also be performed under the crystal state. If it is not the case or it is difficult to perform such measurements, I would suggest the authors the check XRD of the solid and confirm the molecular stacking is consistent in crystals and powers.
Authors´s response 1. Owing to the difficulty of growing single crystals, it is impossible to make magnetic susceptibility measurements, and related measurements, in the single crystal state. The criticism of the reviewer referring to the use of a polycrystalline sample instead of a single crystal, however, can be solved by comparing the powder XRD experiment in the polycrystalline sample in Figure R3 below (i.e., the same sample we used for the magnetic measurements) with that obtained by simulation of a XRD pattern considering the structure of the single crystal of 290 K which is also shown in Figure R3. The comparison is good by which we are confident that the magnetic measurements carried out in the polycrystalline sample are closely correlated with those we would obtain by measurement of the SQUID in the single crystal. This result can indicate that the molecular stacking is consistent in crystals and powders.
In the new version of the article, we have included this Figure R3 in page 12 of electronic supplementary information as Supplementary Fig.7. Reviewer point 2. Since the triplet state is thermally activable, we should keep in mind that the triplet state and single state coexist in the crystal and the geometry information actually is a statistical data. In other words, with the increasing temperature, the triplet state will populate, which may affect the molecular geometry and maybe the packing. The authors are suggested to carefully consider this possible issue.
Authors´s response 2. The reviewer would agree with us in the fact that the singlet triplet energy gap over which the transitions take place are certainly small or very small. This means that molecular species (the singlet and the triplet) with very similar formation energies would disclose very similar molecular structures.
So, in our case, we could not expect significant (even detectable) differences between the structures of the singlet and of the triplet. What we observe by heating is a transformation of the ground electronic singlet diradical state between forms with different portion of aromatic and quinoidal contribution and, for each of these, there are associated triplets that would be similar to their singlets. So, in principle, this issue should not be affecting considerably in our case and, therefore, our argumentation holds in the same terms.

Reviewer point 5.
There are a few typos in the main text and in the information.
Please check and correct them. For example, in "… a solid state building scaffold able to bringing intramolecular torsional mobility…", "bringing" should be corrected; In "… bicycle pedal motion in in azobenzene (B)…," please remove one "in".

Authors´s response 5.
We have checked and corrected grammatical mistakes in the revised manuscripts and the amendments are highlighted in yellow.

REVIEWER #3:
Before addressing the points posed by this reviewer, we would like to draw her/his attention on the new experimental data we provide in the revised version. The data further support our main hypothesis which have allowed to add some more new discussions of the underlying mechanism that overall make a nicer story for the general reader. We hope the referee would agree with us and would be certainly happy if she/he reconsiders his opinion.

Reviewer point 1. ST gap difference between determined values by EPR and
SQUID (1.31 vs 3.41 Kcal/mol) is large. One of them must be incorrect.

Authors´s response 1. In order to check this, new samples and new SQUID
magnetometric measurements have be carried out which result in the data represented in Figure R2 of the response to reviewer 1. By fitting the curves with a modified Bleaney-Bowers equation, the single-triplet energy gap (ΔEST) now amount to 3.68 kcal mol -1 which is closer with the result from ESR (-3.41 kcal mol -1 ). The new experimental value is also closer to the average value, -2.95 kcal/mol, of the theoretical singlet-triplet gaps data from theory; namely, -4.94 kcal/mol on the crystalline structure at 130K, -1.58 kcal/mol on the crystalline structure at 290K and -2.34 kcal/mol on the crystalline structure at 340K. The new SQUID data have been incorporated to the new version of Figure 3 in the main text together with the discussion of their comparisons. Figure R2. Temperature-dependent plots of χT versus T and fitted χT-T curve (solid line) for CAR measured at 1.0 T from 2 to 400 K. Therefore, in pages 12 and 13 of the electronic supplementary information, the new related SQUID data were revised.
In addition, in page 8 line 3 of manuscript, "0.35 emu K mol -1 " was replaced with "0.32 emu K mol -1 ". In Page 8 line 6 of manuscript, "-1.31 emu K mol -1 " was replaced by "-3.68 emu K mol -1 ". And in page 8 line 9 of manuscript, "2-300 K" was replaced with "2-400 K". In reference 15, the unique reversed effect of the aromaticpseudoquinoidal conversion by heating was absent, which is now described here for CAR. The article cited by the referee is below which has been added to the reference list as reference 30. Rivero, S. M. et al. Isomerism, Diradical Signature, and Raman Spectroscopy: Underlying Connections in Diamino Oligophenyl Dications. ChemPhysChem. 19, 1465-1470(2018.

Reviewer point 3. STD of geometry parameters shod be added in the text and tables.
Authors´s response 3. The STD of the bonds and angles were marked in the text and tables in revised manuscript as follows: In page 5 and in line 2, "38.76°" was replaced with "38.8(2)°", "5.78°" was replaced with "5.8(5)°".

Reviewers' Comments:
Reviewer #1: Remarks to the Author: I think this revised manuscript is significantly improved and addresses many of my previous concerns. I think that the additional spectroscopic characterization is very helpful. I have only a few minor questions and concerns prior to publication: 1. The Raman data is a very nice addition to the paper, and strengthens the authors' claims substantially. However, it would be nice to show the analysis from Figure 4 top right for all of the vibrational regions. The key trend on going to 340 K is only one data point in both the SXRD data and the Raman data. Is it reproduced in the other Raman features?
2. I just noticed that the TGA data shows significant mass loss upon warming, what is the explanation of this? This is particularly important as the physical phenomena are occurring at 340 K which appears to be in a region where significant mass loss has occurred.
Reviewer #2: Remarks to the Author: I believe the paper has been well revised and can be considered for publication in Nat. Commun.
Reviewer #3: Remarks to the Author: Shen et al have redone the SQUID and added some previously reported work. Authors also mentioned the disorder of CAR molecule requested by one reviewer. Authors have not completely clarified my questions but just pick up some questions to answer. Based on the provided new data andinformation, I have to say the conclusion of this manuscript is incorrect.
1. The claimed spin mobility is only based on the crystal structures. Authors now mentioned that CAR is disorded with two conformers especially at high temperatures. The whole molecule is disordered including N-N bond! How can the bond length changes be reliable for a disordered crystal structure? 2. Thus other evidences are badly needed. Authors calculated the spin density directly using the disordered crystal structures and claimed that electron spin density has large changes from low temperatures to high temperatures. Such spin density change needs experimental evidences. Solid state EPR and SQUID should demonstrate this! But they did not! As they states in the text "Unfortunately, these three Δ E ST gaps 270 could not be resolved experimentally by SQUID magnetometry which overall detects one single process", which actually is a strong proof that the bond length changes based on the disordered crystal structures are unreliable! There are no apparent spin density changes with temperature! 3. Authors have not excluded the monoradical inpurity that readily exists during synthesis. If we define the parent molecule as CAR-2H, the monoradical is CAR-H and "diradical" is CAR. CAR-H can be buried in the same crystal! The parent molecule CAR-2H can also readily co-crystalize with CAR. In fact, the EPR spectra appear from monoradical CAR-H. Solution NMR and EPR measurements are much better to exclude CAR-2H and CAR-1H.
In conclusion, the purity of claimed CAR molecule is still in doubt. Despite this, CAR is only a normal diradical with a singlet ground state. There is no clear evidence for bond length and spin density changes. This manuscript should not be published in any journal.

REVIEWER 1 Reviewer point 1:
The Raman data is a very nice addition to the paper, and strengthens the authors' claims substantially. However, it would be nice to show the analysis from Figure 4 top right for all of the vibrational regions. The key trend on going to 340 K is only one data point in both the SXRD data and the Raman data. Is it reproduced in the other Raman features? Author´s response 1: We have added the representation of the behavior of the areas of of the Raman bands in the 1200-1100 cm -1 as a function of the temperature such as it was done for the 1620-1540 cm -1 region in Figure 4 (top-right). The new data are included in the same graph (Figure 4) in red colors. The behavior is fully consistent with the previous one. The new Figure 4 is below: (2) ν(CC) phO ; (3) ν(CN) azo1 and (4) ν(CN) azo2 . Top, right: representation as a function of temperature of the ratio of intensities of 1 (A1) and 2 (A2) as black circles and 3 (A3) and 4 (A4) bands as red squares as spectroscopic markers of the degree of quinoidal/aromatic character which delineates three regions: small ratio means more quinoidal, large ratio more aromatic and pseudoquinoidal (PQ) is in between. Bottom, right: topologies of the vibrational modes from the theoretical Raman spectra in Supplementary Figs. 12, 13 and 14.
Reviewer point 2: I just noticed that the TGA data shows significant mass loss upon warming, what is the explanation of this? This is particularly important as the physical phenomena are occurring at 340 K which appears to be in a region where significant mass loss has occurred. Author´s response 2: Sorry for both TGA (in ºC) and Raman/XRD (in K) data being in different temperature scales. According to the TGA in Figure S3 up to approx. 130 ºC ( approx. 400 K) no significant loss of mass is advised. So, the XRD and Raman data taken at the highest 340 K temperature are in conditions free of decomposition.

REVIEWER 3
Reviewer point 1: Shen et al have redone the SQUID and added some previously reported work. Authors also mentioned the disorder of CAR molecule requested by one reviewer. Authors have not completely clarified my questions but just pick up some questions to answer. Based on the provided new data and information, I have to say the conclusion of this manuscript is incorrect. Author´s response 1: We regret the reviewer does not find our article of interest. We, however, disagree with her/his statement about the incorrect interpretation of the data. We will describe our reasons below. Now, we wish to clarify that we have tried to give responses to all her/his previous questions. So, we did not exercise in selecting some points to positively respond them (ignoring the rest). We would like to remark that we addressed all questions from all reviewers: in some cases, these imply changes in the manuscript and in others no. But always we provide the best responses we can or we are aware of.

Reviewer point 2:
The claimed spin mobility is only based on the crystal structures. Authors now mentioned that CAR is disorded with two conformers especially at high temperatures. The whole molecule is disordered including N-N bond! How can the bond length changes be reliable for a disordered crystal structure?
Author´s response 2: The bicycle pedal motion is a well-established feature of the crystal of azobenzene, and of its derivatives, and it is well reported in the literature that the way (one of them) to proof this is by seeing some disorder in the azo moiety. This must be remarked since we are dealing with a local effect and it is not a disorder feature of the whole molecule as a function of the temperature. This is clearly stated in the literature [Acta Crystallogr., Sect. B:Struct. Sci., 1997, 53, 662-672] for unresolved disordered structure of azobenzene. The paper concludes: "the influence caused by the displacement of the positions of the other atoms is so small that it can be negligible which can be proved from their single crystal at 82 K and 296 K", what is actually our results. The small displacement of the atoms for phenoxyl and azobenzene rings is marginal and thus the bond length changes in the rings are reliable and not caused by disorder. Nonetheless, we have to admit that the observed NN bond length might be slightly shorter than the true length owing to the disorder at 250, 290 and 340 K, but this deviation is assumed to be small because the changes of population of the minor conformer is from 10% to 17% on heating while the NN bond lengths show large changes from 1.285(3) to 1.253(4) Å. So, the presence of disorder in the azo group (insist, not in the whole molecule) is the corroboration of the activation of the bicycle pedal motion which detunes the reversal of the aromaticquinoidal transition.
The following paragraph has been added to page 6 of the ESI file of the second revision of the article: In addition, in spite of the presence of the disorder, the population of the minor conformer is very small all along the whole temperature range analyzed even if its ratio is rising with temperature increasing. The very little influence of the disorder on bonds length is mainly existing in the NN bond, but for the other bonds, this is negligible according to the literature report by: Harada, J.; Ogawa, K.; Tomoda, S. Molecular motion and conformational interconversion of azobenzene in crystals as studied by X-ray diffraction. Acta Cryst. 1997, B53, 662-672.
Finally, we would like to mention that upon requests of reviewers the addition of the variable temperature Raman data considering up to 12 different temperatures in this range corroborate all the behavior deduced from X-ray analysis.
Reviewer point 3: Thus other evidences are badly needed. Authors calculated the spin density directly using the disordered crystal structures and claimed that electron spin density has large changes from low temperatures to high temperatures. Such spin density change needs experimental evidences. Solid state EPR and SQUID should demonstrate this! But they did not! As they states in the text "Unfortunately, these three ∆ E ST gaps could not be resolved experimentally by SQUID magnetometry which overall detects one single process", which actually is a strong proof that the bond length changes based on the disordered crystal structures are unreliable! There are no apparent spin density changes with temperature! Author´s response 3: We agree that the solid state ESR and SQUID did not show very clear transitions. In fact, If one looks closely on the SQUID data, there is a small transition at 290 K. We have tried to make Bleany-Bowers fits of χΤ versus T in the temperature ranges lower and higher than 290 K from which we did obtain two ∆E s-t values. However, the error of one of them is relatively large. Therefore, we did not discuss it. On the other hand, she/he mentioned that other evidences are needed to confirm our hypothesis and these have been provided by variable temperature Raman spectroscopy which further corroborates our arguments. Briefly, Raman spectra have been carried out up to in 12 different temperatures between 130 and 340 K and the spectral changes were analyzed in two spectra regions. In all these temperatures and in the two spectral regions the behavior is fully pointing to the presented hypothesis (see Figure 4 and discussion). So, it is not totally true that we sustain our claims only based on XRD data and only in some selected temperatures. In addition, we reproduce all these features by quantum chemical calculations which are further corroborations of our study. It is a well accepted procedure to obtain the spin distribution from the x-ray data by quantum chemical calculations and this has been done for innumerable cases of magnetic organic molecules.

Reviewer point 4:
Authors have not excluded the monoradical inpurity that readily exists during synthesis. If we define the parent molecule as CAR-2H, the monoradical is CAR-H and "diradical" is CAR. CAR-H can be buried in the same crystal! The parent molecule CAR-2H can also readily co-crystalize with CAR. In fact, the EPR spectra appear from monoradical CAR-H. Solution NMR and EPR measurements are much better to exclude CAR-2H and CAR-1H. Author´s response 4: Thanks for your comment. The oxidation of CARH to CAR takes place in very high yield (>95%). Also, the polarities of CARH and CAR are very different on silica gel, which can be easily separated by column chromatography. The CAR is very stable and was purified from silica gel column (following by thick layer silica plate purification for 2 times). According to our experience on radical purification, monoradical based on phenoxyl unit (CAR-1H) is unstable in silica gel. Therefore, all monoradical impurity can be easily removed. Meanwhile, we store our sample in the glovebox. Furthermore, radical compounds are normally NMR silent, so it is hard to identify their purity based on NMR measurements. We also obtained the crystal structure of CAR-2H (CARH, Figure  1), the packing mode is very different from that of CAR. It will be unlikely that both could co-crystalize. Therefore, we think our sample is very pure enough for all the measurements.

Reviewer point 5:
In conclusion, the purity of claimed CAR molecule is still in doubt. Despite this, CAR is only a normal diradical with a singlet ground state. There is no clear evidence for bond length and spin density changes. This manuscript should not be published in any journal.

Author´s response 5:
Most of the experimental data presented in the article are scrutinized by the corresponding simulations (XRD, EPR, Raman, singlet-triplet gaps, etc.) and all simulations reproduce very well the experiments. So, assuming there would a fraction of impurity due to precursor radical or others, these seem not to interfere in the analysis and thus in the conclusions. The referee knows well (she/he is a diradical expert) that small fraction of radical impurities in the samples of ALL diradicals are unavoidable. The important thing is that these impurities do not interfere the analysis and this is warranted in our article. We obviously respect the referee conclusion but are not in agreement. This is not another diradical of the many in the list. CAR diradical represents the unique and first (to the best of our knowledge) example in which molecular vibrations in the ground electronic state modulate the structure and subsequently the spin distribution. So far, most of the existing diradicals are described at their "vibrationally frozen" structure neglecting the role of vibrations. This is logical to some point since the immense majority of diradicals can be divided in two main groups: i) those constructed in rigid or very rigid polycyclic aromatic platforms who vibration-structural relationship is insensitive to rather small-moderate changes of temperature. And ii) those diradicals based on more flexible structures that, unfortunately, are tough to crystalize and become much more reactive. On these flexible diradicals, temperature studies are impractical in the sense that temperature degrades the samples. In this scenario, CAR is in the intermediate situation between the (i) and (ii) cases above in which the effect of temperature in the solid state dynamics can be addressed and this is very significant. In this context, we disagree with CAR being just one more diradical example.