Shape-memory effects in molecular crystals

Molecular crystals can be bent elastically by expansion or plastically by delamination into slabs that glide along slip planes. Here we report that upon bending, terephthalic acid crystals can undergo a mechanically induced phase transition without delamination and their overall crystal integrity is retained. Such plastically bent crystals act as bimorphs and their phase uniformity can be recovered thermally by taking the crystal over the phase transition temperature. This recovers the original straight shape and the crystal can be bent by a reverse thermal treatment, resulting in shape memory effects akin of those observed with some metal alloys and polymers. We anticipate that similar memory and restorative effects are common for other molecular crystals having metastable polymorphs. The results demonstrate the advantage of using intermolecular interactions to accomplish mechanically adaptive properties with organic solids that bridge the gap between mesophasic and inorganic materials in the materials property space.

The second key aspect that stands out in this work is that they were able to provide strong evidence for the proposed mechanism by obtaining crystal structures on the concave and convex sides of the bent crystal to demonstrate the partial phase transformation, and they argue convincingly how these changes in crystal packing relate to the rod deformation.
The combination of the unique crystalline behaviors and the strong structural evidence makes this paper worthy of publication in Nature Comm. I have only a couple very minor comments: -Is phase II of TA generally more stable under isotropic/hydrostatic pressure, or does the phase transition only result from the anisotropic nature of the local forces being applied here? -The paper notes large differences (1.80 vs 1.44 Ang) in the hydrogen bond lengths for the two polymorphs, but the crystal structure details for the COOH groups in the CIF files are suspect. In both polymorphs, the intermolecular O..O distances are very similar at 2.62 and 2.63 Angstroms. However, in one case, the O-H covalent bond is suprisingly short (0.82 Ang), and in the other it seems too long at 1.19 Ang. This then translates to the large differences seen in the H..O hydrogen bond lengths. I am not a crystallographer, and I recognize that H atom positions can be difficult to infer, but this still seems wrong (or at least something to discuss). None of this will likely change any of the important arguments, but they should look into the issue.
Reviewer #4 (Remarks to the Author): The authors report synchrotron microfocus X-ray diffraction analysis of the bent crystal of terephthalic acid (TA) that has two coexisting polymorphs and suggest a new mechanism for bent crystals as a mechanically induced phase transition. The introduced mechanism provides some insight on shape memory and restorative effects seen in other molecular crystals. However, many of the results discussed in the paper including the shape memory and self-healing effects of this crystal have already been discussed in their previous work (J. Am. Chem. Soc. 2016, 138, 13298). Specifically, most of the results presented in Figures 1~3 of this work have been discussed before in detail. There are numerous additional results that should be added overall and some of the arguments are not well supported. Thus this work requires major revision by addressing all questions and points suggested below before being considered for publication in Nature Communications.
• Abstract and conclusion generalize the mechanism seen in TA crystal to soft crystals and organic solids, but this study only focuses on the TA crystal (which most of the results have been discussed in JACS 2016). There is not enough evidence to support that what happens in the TA crystals is generalizable to other molecular crystals. • Phase identification in the bent region requires much more detail. This is the critical part of the paper. The data presented does not fully support the conclusion that the bent region is bimorphic. The microdiffraction results presented in Figure S5 is not of very high quality (showing doublets). Also the result is only presented in the compressed region, a single diffraction experiment, but not over the entire bent region as a mapping result. Mapping is needed to show how the polymorphs vary: is it continuous or an abrupt change from form I to II? The diffraction image presented needs to be indexed and the unit cell refinement results need to be presented with R values (Table  S1 does not contain this info).
Section "Phase transition and shape restorative effects" • Timescale on transition: In the first paragraph, when they indicate that the crystals "do not transition sharply", "rapid reshaping upon mechanical stimulation", etc. what does that mean? Is it 1st order or 2nd order transition? Need to show DSC results which show 1st order transition o On a related note, Supplementary Videos do not have time stamps or scale bars • "Twinned crystals" can indicate that the two forms are twins, indicating that they are of same polymorph. In this work, they are form I and II, and are different polymorphs • From Supplementary Movie 1, it is hard to believe that the crystal springs off due to phase transition rather than force of the sharp object pushing down on the edge of the crystal. In the video the partial transition from I to II already takes place in the beginning before jumping off. Is there proof that the crystal completely transitioned to form II? Also, this phenomenon was already discussed in JACS 2016, it is unclear what the purpose of mentioning it again here is. • Figure 1B: Is there structural proof that the region that the sharp object touches is Form II? Was micro X-ray also done in that region? If so, it should be discussed, if not, there should be supporting evidence that the transitioned region is indeed the stable Form II, using other methods to cross validate such as spectroscopy.
• Figure 1D: This is identical to Figure 4A of JACS 2016, where they bend the crystal and heat. It is unclear what new information this figure adds to this work. Also, what happens to the crystal after cooling? Would the crystal transform back to Form II? Would you able to start from Form I and induce mechanical bending as well or does bending only occur on Form II? Need more details. • Figure 1E: Supplementary Video shows clear restoration, but from the figure it is hard to see. Also, need to show evidence that after heating, the crystal is fully transformed to Form I. Was Xray repeated on the crystal after heating? Since it did not completely restore is it possible that it may not be completely Form I? Heating alone can't guarantee that the crystal has completely transformed to Form I. • Figure 2: Again, how is it proven that it is entirely Form I at 358K? The bottom figures and the supplementary movie seems to show that maybe the crystal has partially transformed from Form II to I, as the phase propagation does not seem to go through all the way. Does it eventually transform completely? Is phase transformation affected by defect density or location of defects caused by the metal plate?
Section "Microfocus X-ray diffraction analysis of a bent crystal" • As indicated in manuscript, Supplementary Figure 5 shows long range order, but also shows several diffraction peaks close together, which can indicate twinning or gliding of the different layers. Explanation of what they are is needed. • The manuscript describes Supplementary Table 1 as the result of refinement on the bent convex and concave side crystals, but the title of the table is: Crystallographic data and refinement details of terephthalic acid structures refined from the same crystal at different temperatures. It seems like the title is labeled incorrectly. • It would be interesting to see if there is a critical strain for forming I on the compressed concave side of the crystal. With less than critical strain, then both sides of the crystal even after bending may still be Form II. • It would be interesting to see if the crystal integrity can still be maintained if the crystal is bent the opposite way. • A comparison of the bent crystal Form II to the pristine Form II at same temperatures would be interesting. Since the convex part still experiences expansion as Form II, does it have expanded/contracted unit cell values when compared to the pristine case?
Section "Mechanism of plastic bending by partial phase transition" • Authors described in last paragraph before Figure 5 "This causes sliding and rotation of nearly 30 of the tapes on the concave side, ..." which angle is this referring to? The author never described in the manuscript.
• Also authors mentioned in the same paragraph that: "The slight mismatch of the two unit cells, which-similar to their coexistence as alternating layers in a straight partially converted crystalremain attached at the phase boundary, stabilizes the bimorph and the bent shape of the crystal is retained ( Figure 4D)." As authors show the lattice structure in Figure 5A the mismatch at the phase boundary looks significant (for instance around 26.3 % decrease in a-axis length); which do not mean slight mismatch of the two unit cells. And in Figure 5B on the right schematic exaggerated the lattice dimension of Phase I to seem like it matches to that of Phase II. • Incorrect statement right before Figure 5: "When the bent crystal is heated over the phase transition, the form I on the concave side is transformed back to form II" (it should be form II to form I) We thank all reviewers for the valuable comments which have contributed significantly to improve the quality of the manuscript. We considered all comments, and we tried to address them to the best of our ability. Unless stated otherwise, all numbers of the figures and supplementary materials refer to the revised version of the manuscript. Together with this submission we have provided marked copies of the main text and the Supplementary Materials where the changes to the original version have been marked. For convenience, in what follows, the original comments from the reviewers are highlighted in blue color, our response is provided in black color, and the text that was modified or added to the manuscript is marked with red color.

Response to the comments from Reviewer #1
Comment: Key results: The authors have investigated some of the mechanical properties of crystals of terephthalic acid. They find that upon application of strain within the crystal they can induce a phase change which leads to deformation of the crystal. The authors claim it is reversible and a new mechanism of bending.
The claims of the authors are simply not supported. There are many definitional issues throughout the manuscript and statements that are factually incorrect. These statements are not supported by references or experiments. The biggest flaw and most significant is a requirement that the authors demonstrate that these are single crystals -indeed their microfocus diffraction experiments show that they are not single and probably not crystals. The propose a mechanism but it is mere conjecture it has not been established by experiment. Insuffficient experimental details have been given to allow the reproducibility of the results.
Response to the comment: We appreciate the reviewer's assessment of our results, and in the response below we address their concerns.
We are quite perplexed with the reviewer's conclusion that the "microfocus X-ray diffraction experiments show that their crystals are not single crystals", particularly in view of the fact that three out of the four authors on this manuscript are trained crystallographers. The corresponding author of this manuscript has over 20 years of experience in small-molecule crystallography and has determined and published hundreds of crystal structures. One of the other crystallographers on this manuscript works in a crystallography beamline at a synchrotron, and the third is specialist for crystallography with in-house diffractometers. same data have been deposited within the Cambridge Structure Database, and are accessible for inspection if needed. In the original version of the manuscript, we have also provided images of diffractions that confirm that the samples are single crystals. There is no better way to confirm the single crystal nature of these samples than the inspection of the primary diffraction images and fully refined crystal structures that clearly evidence that the material is in form of single crystals. In view of the collective experimental data, we do not agree that that the mechanism which we propose is unfounded, because we have provided firm experimental data that clearly support that mechanism.
We would also like to note that the definitions of the terms "single crystal" and "crystallinity" are out of scope of this work, because these terms are already well defined in the crystallography, they are commonly accepted, and can not be a subject of personal interpretation. We would like to draw the editor's and the reviewer's attention to a recent peer-reviewed publication where we have discussed this subject in greater detail, and based on a large amount of results on bent crystals from the literature: P. Commins, D. P. Karothu, P. Naumov, "Is a bent crystal still a single crystal?", Angew. Chem., Int. Ed. 2019, in press (https://doi.org/10.1002/anie.201814387). The conclusion in this peerreviewed publication is very clear: bent crystals are single crystals as long as they are not completely delaminated in separate crystals. We would like to reiterate here that the definition of a single crystal given by the International Union of Crystallography is very clear and can not be subject to different interpretations. The above article explains how it applies on bent molecular crystals, and we strongly recommend that the reviewer becomes acquainted with the definition and its interpretation. We do not find it necessary to redefine these established definitions in the current article, because they are available from the above article and the general crystallographic literature.
In response to the reviewer's comment, we provide a copy of the conclusions paragraph of the above paper: "Going to the question posed in the title of this article, among the articles that describe bending crystals, the amount of reports that provide thorough X-ray diffraction analysis of bending crystals that were single crystals before being bent is still surprisingly small and thus a general consensus on the single crystallinity of these compounds has not been established. The available data, however, clearly show-within the IUCr definition of a single crystal-crystals that bend elastically and crystals that bend plastically without physical separation are clearly single crystals after they have been deformed, the latter having expectedly higher concentration of defects. Single crystals where visible separation occurs during bending would normally be considered multiple crystals after the bending. Even when the layers do not separate after the first bending, plastic crystals are expected to partially delaminate during repeated bending and unbending; they may maintains single crystallinity after one cycle and gradually become more polycrystalline after successive bending cycles. We hope that this collection of results will clarify some of the confusion, both semantic and factual, with how mechanically complaint crystals should be regarded in future." The crystal nature of the terephthalic acid has been well established in the earlier literature which is cited in our manuscript. For example, see the article by Roger Davey: Davey, R. J. et al. Morphology and polymorphism in molecular crystals: terephthalic acid. J. Chem. Soc., Faraday Trans. 90, 1003-1009 (1994). Our samples of terephthalic acid conform to the earlier literature, and they are clearly single crystals, as can be also inspected from the diffraction images recorded from both forms provide below: The details of the refined crystal structures can also be inspected from the crystallographic details in Supplementary

Response to the comment:
The authors of this work are very well aware of the work of Miyamoto and Takamizawa. However, both the material and the phenomenon described in the above article are very different from those studied in this work (we would also like to refrain from commenting on one article as being "superior" to another one). Namely, the compound studied by Takamizawa and Miyamoto is terephthalamide, and is a so-called "superelastic" organic crystal. By applying shear stress on this crystal, the crystal transforms to another form. Once the stress is released, the crystal reverts to its original form. This behavior is completely different from the one that we describe in the current work.
The terephthalic acid (TA) crystals that we describe here show thermosalient and mechanosalient behaviors which are not observed with the above compound. Moreover, the form II crystals of TA show shape-memory and self-healing like properties. Most importantly, the crystals of TA can be bent plastically and they do not revert to their initial shape spontaneously. These bent crystals can be further reshaped into straight and bent shapes by applying force for several cycles without any deterioration. What we report in this manuscript is the first ever example where upon bending of form II TA crystal, the stable form II in the bent region is transformed to the metastable form I. This has not been reported before. Due to the large differences in the Young's modulus and hardness of the two forms and the relatively soft nature of form I, the two forms can coexist as a bimorph in the bent region. We have characterized this bimorph in the bent region by using X-ray diffraction and scanning electron microscopy, in addition to other methods.
The figure below describes visually the difference between the two systems. Response to the comment: We thank the reviewer for the generally positive assessment of the contents of our manuscript, and for the constructive suggestions. All suggestions were considered and taken into account in the revised version.
We have revised the abstract, the introduction, and to a greater extent, the conclusions section in order to place the results within a more general context. We attempted to add to the generality of the results in the abstract of the original submission, where we state that "Heating over the phase transition temperature partially recovers their original straight shape and appears as a shape memory effect akin of that observed with polymers or metals." in hope that the general audience will recognize the more general implications of the results presented in this work. Although we are very limited with word count, in the revised version we added text to highlight the possible applications of these materials. Below, we have copied the new version of the abstract: "Molecular crystals can be bent elastically by expansion, or plastically by delamination into slabs that glide along slip-planes. Here we report that organic crystals can also be bent by a distinctly different mechanism, a mechanically induced phase transition that occurs without delamination and preserves the overall crystal integrity. While generally the shape of elastically bent crystals is restored immediately after removal of the external force and plastically bent crystals remain deformed indefinitely, crystals where two phases coexist effectively act as bimorphs and their phase uniformity can be recovered thermally-heating over the phase transition temperature partially recovers their original straight shape and upon cooling they revert to their bent habit-resulting in shape memory effects akin of those observed with some metal alloys and polymers. The phase transition also accounts for partial recovery of the integrity of mechanically damaged crystals. We anticipate that similar memory and restorative effects are common for other molecular crystals which have polymorphs that can coexist in the same crystal with low interfacial energy. The results demonstrate the advantage of using intermolecular interactions to accomplish mechanically adaptive properties with organic solids that bridge the gap in the materials space between mesophasic and inorganic materials."

Comment: How general is this result? Is it transformative and if yes, give some concrete examples.
Response to the comment: We share reviewer's sentiment that at first reading, the text might appear rather technical and narrow in scope. We would like to note that the maijn point of originality in this work is that this is the first example of a molecular crystal which shows phase transformation during bending. At ambient temperature, the resulting bent crystal remains a stable bimorph and the bent region having form I and form II in the concave and convex regions, respectively. So far, we have not observed this phase transformation in the bent region in any other organic molecule, and we are not aware of another reported example. However, as we have also delineated in the revised text, we do expect that other crystals could undergo similar phase transition during bending: Abstract: "We anticipate that similar memory and restorative effects are common for other molecular crystals which have polymorphs that can coexist in the same crystal with low interfacial energy. The results demonstrate the advantage of using intermolecular interactions to accomplish mechanically adaptive properties with organic solids that bridge the gap in the materials space between mesophasic and inorganic materials." Conclusions: "The TA crystal provides the first example of pressure-induced phase transition in a bending crystal where the two phases coexist at ambient conditions and establishes a hitherto unreported mechanism of crystal bending. It is likely that this mechanism can account for shape-memory and restorative effects of other organic crystalline materials that are increasingly observed, but are almost never explained." We are rather limited with the word count in the abstract and the introduction to provide more information of the general significance and broader impact of the results. Instead, we elaborated more on this in the conclusions section in the revised version. Comment: I also would like to see discussion on the change in mechanical properties between the two polymorphs and how it may contribute to the apparent partial shaperestoration effect. Below please find several additional specific questions and suggestions: Response to the comment: We thank the reviewer for this constructive suggestion.
Although it was challenging exercise for one of the polymorphs, in the meantime we succeeded to measure the Young's modulus and hardness for both forms II and I by using nanoindentation. In the revised version of the manuscript we have included the discussion on the mechanical properties of both forms and their contribution towards the partial shape-restoration effect.  Figure 22). The minimal force required to induce the phase transition of form II to form I by bending of the crystal obtained from the force-displacement profile along with the critical strain and averaged over eight crystals was determined to be 92 ± 5 mN (an exemplary force-displacement and stress-strain curves are shown in Supplementary Figure 7)." Comment: 1. Figure 1D - Response to the comment: We thank the reviewer for this important remark. We have replace some of the panels in Figure 1 (panels B, C and D, and added panel F) with new images that show the same phenomena and we hope they illustrate better the point that we discuss in the text.
As we have explained in the revised text, form II crystals of terephthalic acid are only obtained as elongated parallelepipeds by hydrothermal crystallization. These straight crystals can be bent on their (010) face. As we also highlighted in the revised text, the degree of shape restoration does depend on the way the initial mechanical bending has been performed as well on the thickness of the crystal. Partial shape restoration to straight shape is observed during phase transition (above 350 K) to form I. However, in several cases we have observed complete shape restoration to a straight crystal. In the revised version, we have now included the data on four crystals that were bent to a different degree, and then heated for shape restoration during the phase transition to form I. The shape restoration of these bent form II crystals can only observed by heating and occurs above 350 K.
The new examples were included in the revised version as Supplementary Figure 8. The following text was added in the main text of the manuscript: "We noticed that among other factors-such as defects that are beyond the experimenter's control-the ability for recovery of the straight shape is determined by the degree of bending and crystal thickness. In some cases, we were astonished to observe Response to the comment: Following the reviewer's suggestion, we have performed additional experiments to verify this. A crystal of form II was bent by applying mechanical force on its (010) face. This bent crystal is then transformed to form I by heating above the phase transition temperature (354 K). During this phase transition the bent shape transformed to straight shape. Upon cooling to room temperature, form I crystal transformed to form II, and the shape of the crystal as again converted to the initial bent shape. This transformation between bent shape and straight shape can be repeated several times. We have now included a figure   Response to the comment: We thank the reviewer for this suggestion. We performed nanoindentation measurements on both form II and form I and the corresponding graphs are now included as Supplementary Figure 21 (shown below). The Young's modulus of form II is 6.2 ± 0.7 GPa and the hardness of the crystals is 0.33 ± 0.05 GPa. However, form I is comparatively softer, and its Young's modulus is 0.6 ± 0.2 GPa and hardness of the crystal is 0.05 ± 0.03 GPa. The difference in mechanical properties between the two forms could the reason for form I being able to accommodate the local pressure in the bent region. The following text was added to the main text: "As established by nanoindentation, the crystals of both polymorphs are generally soft ( Supplementary Figures 7 and 21) Figure 1B, Supplementary Figure 1 Response to the comment: We thank the reviewer for noticing the long H···O hydrogen bond lengths, and we share the reviewer's opinion on the unreliable positioning of the hydrogen atoms with X-rays. We have now refined the hydrogen positions, using the common constraints, to acceptable distances. All data that pertain to the crystal structures have been corrected, and the new set of data are deposited with the revised version. The respective files in the CCDC were also replaced.

Response to the comments from Reviewer #4
Comment: The authors report synchrotron microfocus X-ray diffraction analysis of the bent crystal of terephthalic acid (TA) that has two coexisting polymorphs and suggest a new mechanism for bent crystals as a mechanically induced phase transition. The introduced mechanism provides some insight on shape memory and restorative effects seen in other molecular crystals. However, many of the results discussed in the paper including the shape memory and self-healing effects of this crystal have already been discussed in their previous work (J. Am. Chem. Soc. 2016, 138, 13298). Specifically, most of the results presented in Figures 1~3 of this  We do share the reviewer's sentiment about our preliminary results which were reported together with other results in our previous publication which focused on the mechanosalient effect. Indeed, we noticed some of the salient features described here in that stage of the research work, although at that time we were not able to fully explain our observations, which were also made on a very limited number of crystals. Some of these observations were reported at conference meetings, but remained in the form of mere observations. With this report, we intended to explore the full details of these effects, in an attempt to provide thorough experimental results and to explain the mechanism, as well as to reach more general conclusions. For instance, the partial recovery of the shape of the crystal or the shape-memory effect were noted in the earlier report, but the reasons for these observation remained unexplained because we did not have experimental facilities to pursue in that direction.
Having said that, the authors of this work strongly believe that experimental observations should be reported, regardless of whether they can be explained or not, because (as it has been the case in so many instances in science, and with an example most closely to our researchthe thermosalient effect) they could become the seed or motivation for further in-depth studies, and in some cases, even lead to major discoveries. Perhaps the most important aspect of the new results presented in this work is that we have succeeded to obtain evidence for the first ever example of crystal deformation (bending) which is accompanied by a phase transition. This mechanism has not been reported before. We think it could be important because in principle the mechanism should be applicable to other crystals which have stable and metastable phases that can coexist at a certain temperature and pressure. We also provide evidence of coexistence of two different phases in the bent region, and we note again that this result has not been reported before. We hope that considering the depth and the breadth of the new results (in view of the different techniques used to characterize the effects) presented in this manuscript, the work can hopefully stand on its own, due to both the different focus of the main research theme, the methods used (SR XRD, SEM, optical microscopy, thermal data), the new mechanism, and also the other related properties that might be of a more general relevance and could potentially have a broader impact. We do, however, realize that additional results would contribute to strengthen further the originality and novelty aspect of this work.
In the revised version of the manuscript, and also because of the recommendation from another reviewer of this manuscript (see above), we have revised the abstract, to a smaller extent the introduction, and to a larger extent the conclusions sections, to reflect the possible general impact of the results presented in the manuscript, and in line with the above applicability of the mechanism to other crystals. We believe that once the details of the mechanism become available to the research community, they could be useful to others to try and apply the mechanism in order to accomplish reversible plastic bending of organic crystals without delamination. The enhanced mechanical compliance in such materials would be the true benefit and the general relevance of the results presented here.
Comment: • Phase identification in the bent region requires much more detail. This is the critical part of the paper. The data presented does not fully support the conclusion that the bent region is bimorphic. The microdiffraction results presented in Figure S5 is Response to the comment. We thank the reviewer for this important comment, and we agree with them that the phase identification of the bent section is one of the central results in this manuscript.
During the work on this problem, we have of course, attempted mapping of the bent region, similar to the beautiful work on elastically bendable crystal published by McMurtrie Clegg and the collaborators recently (Nat. Chem. 10, 65−69 (2018). After many trials, however, we concluded that similar analysis by mapping was not possible in our case, both because we have two distinct phases, as well as because of technical limitations (the limited beamtime we could obtain for these experiments). As the reviewer might be aware, these experiments are extremely challenging, and this is the reason none of these crystals have been characterized yet by X-ray diffraction, even when using bright X-rays and small X-ray beam, although the literature now contains a number of reports which show images of diffraction from bent crystals (see our recent minireview: P. Commins, D. P. Karothu, P. Naumov, "Is a bent crystal still a single crystal?", Angew. Chem. Int. Ed. 2019, in press (https://doi.org/10.1002/anie.201814387)). Namely, with our experimental geometry at the synchrotron it was extremely difficult to design an appropriate data collection strategy in order to collect sufficient coverage for one of the forms in presence of the other form. Unlike the case where the structure changes continuously across the kink (i.e. the work off McMurtrie and Clegg), in case of a bimorph, the presence of two phases places limitations on the rotational freedom of the crystal because at diferent pint of time the beam passes through one, second or both phases in different ratios, bot of them having spatially non-uniform orientation of their crystal lattice. The difficulties which are encountered with one phase are doubled when working with two phases. These technical issues are further complicated with the thickness of the crystal in respect to the limited diameter of the microfocus beam required to obtain sufficient count without damaging the crystal, restrictions with the phi rotation and of course, the strain on both lattices close to the habit plane between the two phases. Facing these difficulties, we had to resort to structure determination of the two domains representing different phases, although this exercise also proved extremely challenging and required careful planning of the data collection strategy and technical experience with the data collection and processing. Inspection of the reciprocal lattice showed absence of twinning in the bent region. The initial inspection of the diffraction images and the rocking analysis farther from the habit plane that they were of sufficient quality, comparable to that of the straight section of the same crystal, and we collected full data on positions which were promising in view of the data quality. In the revised version, we have provided exemplary images of the diffraction collected at three sections of the crystal, the concave side, the convex side, and the straight section of the same crystal, with indices for several of the more characteristic reflections. They are now provided as Supplementary Figures 17, 18 and 19. Any deformations of any of the reflections in these images probably comes from the strain imposed on the two lattices by the habit plane between them. In the revised version of the supplementary material, we also provide complete data for the refined structures (Supplementary Table 4; the table was mislabeled in the original version, and we thank the reviewer for bringing that to our attention). As it can be inspected from the table, the crystal structures of both phases have been refined to good acceptable statistics. The same data were deposited and are available from the CCDC.
In addition to the X-ray diffraction data, we sought for other means to understand better the nature of the interface between the two forms. Optical micrographs, and especially, the scanning electron micrographs, shown in Figure 3A and Supplementary Figure 16 were particularly helpful to identify the striations on the surface of the crystal and to determine their orientation for correlation with the crystal axes, and these were applied to a number of crystals to confirm the formation of two phases in the bent region. More importantly they helped to directly visualize the habit plane and the slight difference in striations on both phases that are joined at the plane. For bent crystals, we provided additional evidence by using in-house diffractometer (the data quality was only sufficient to confirm the unit cells) and Raman spectroscopy. We rely on these additional experimental results, in addition the SR X-ray diffraction data as the central evidence for the mechanism of bending that we advance in this work.

Comment: Section "Phase transition and shape restorative effects" • Timescale on transition: In the first paragraph, when they indicate that the crystals "do not transition sharply", "rapid reshaping upon mechanical stimulation", etc. what does that mean? Is it 1st order or 2nd order transition? Need to show DSC results which show 1st order transition. On a related note, Supplementary Videos do not have time stamps or scale bars
Response to the comment: We considered that the general information on the nature of the phase transition is available from the literature, and this is why we did not provide more information in the original version of the manuscript. In the revised version, we included a new figure in the supplementary material (Supplementary Figure 2) which contains detail thermal analysis, by DSC, of the transition between the two forms (for convenience, the figure is also shown below) after one thermal cycle and after four thermal cycles, without grinding and with light grinding of the sample (this was necessary, because from our prior experience we were aware that grinding can have a substantial effect on the thermal profile of thermosalient transitions).
Ass can be inspected from Supplementary Figure

Comment: • "Twinned crystals" can indicate that the two forms are twins, indicating that they are of same polymorph. In this work, they are form I and II, and are different polymorphs
Response to the comment: We agree with the reviewer on the importance of the accurate use of the term "twinning". The term "twinning" is used to refer to two crystal components that have grown together. In the more exact formulation of twinning, which is given by a twin law for a particular twinned crystal, a twinned crystal is an aggregate in which different domains are joined together and are related to each other with a specific symmetry operation. The diffraction patterns derived from different domains are rotated, reflected or inverted with respect to each other, depending on the symmetry relationship between the different domains. The diffraction pattern measured during complete data collection (which includes also phi rotation) is a superposition of all of these. In the case of the form II crystal of TA, the diffraction data were collected at mainly two locations along the crystallographic b axis by restraining the phi and 2θ angles. This excludes the overlapping of diffraction spots with the other form in the convex or in the concave region. We would like to point out that within the convex or concave regions the twinning may indeed occur, however we have evaluated the diffraction data of the convex and concave regions separately for twinning, and concluded that the region is devoid of twinning. In line with this, we have removed the term "twinning" in the text where it specifically referred to a bent crystal of TA.

Comment: • From Supplementary Movie 1, it is hard to believe that the crystal springs off due to phase transition rather than force of the sharp object pushing down on the edge of the crystal. In the video the partial transition from I to II already takes place in the beginning before jumping off. Is there proof that the crystal completely transitioned to form II? Also, this phenomenon was already discussed in JACS 2016, it is unclear what the purpose of mentioning it again here is.
Response to the comment: The thermodynamically stable polymorph, form II, is converted to form I by heating. Upon cooling of the thus obtained form I, most of the crystals are transformed back to the (more stable) form II. However, some crystals remain untransformed and are "trapped" as form I, which is metastable at room temperature. We have repeatedly observed that the transformation of these crystals can be induced by mechanical stimulation, i.e., by applying a local pressure, for example, by lightly touching the crystals with a hard object. Upon mechanical stimulation, they are rapidly transformed to form II. The rapid progression of the phase front results in simultaneous reshaping of the crystal and hopping off the stage, a phenomenon which was termed "mechanosalient effect". In the revised version, we have included a few additional videos as supplementary material (Supplementary Movies 1-5) which clearly illustrate this effect with multiple crystals.
We would like to clarify the content of the original Supplementary Video 1. The original version of the video showed a crystal of form I, obtained as described in the preceding paragraph, which was lightly pressed at the edge, occasionally slightly sliding along the surface. In the first portion of the video (time 00:00 to about time 01:30 s), the crystal of form I was stepwise and partially converted to form II (the converted domain is visible and can be inspected from the shifting of the "kink" on the edge of the crystal, which represents the end point of the phase boundary between the two phases). If the crystal has defects, the transformation starts and is terminated at a defect, and therefore only one domain is converted at a time, and the entire crystal is converted over a longer timespan. As it can been from the original video, the first domain is transformed around 00:08-00:12 s. The second domain is transformed around 00:28 s (note that these are high speed recordings, so what appears as slow transformation actually occurs very fast). Further application of pressure on the crystal and against the base (the needle visible in the video is towards the viewer pointing in the viewer's direction) by applying very light pressure on this transformed region) results in (indirect) pressure on the remaining part of the crystal, which then at 01:34 transforms completely. This transformation is fast and results in jumping of the crystal.
Because in this experiment the crystal was touched always on one side (from the top and close to the edge), in the revised version of the manuscript we have added four new videos (the new Supplementary Videos 25) which show this phenomenon more clearly. In these videos, the crystal is not pressed from the top or at the edge, but instead it is lightly tapped from the side, and therefore it does not experience any pressure applied from the top/edge that would appear as a reason for the crystal motion simply as a reaction to the action. Moreover, the high-speed recordings clearly show that the crystal springs off the support at a rate which is much faster than the rate by which it is approached from the side, and confirms the mechanosalient effect.
We agree with the reviewer that this phenomenon has been discussed in our previous work, however we felt that a combined narrative on the polymorphism, bending, and the mechanosalient effect, and a more detailed study of the latter that we provide here may shed further light on the mechanism, especially in view of the fact that the phenomenon is rooted in a phase transition which also is the reason for the new mechanism of crystal bending that we describe in this work. Presenting all these exotic phenomena together appeared to us to add to the completeness of this report. However, considering the preceding study, the mechanosalient effect was intentionally given much less attention and space in the manuscript relative to the bending, shape-memory and shaperestorative effects. We hope that by presenting another aspect of the same phase transition, this portion of the results adds to the completeness of the narrative in the overall manuscript. In  Table 1). We also recorded Raman spectra that confirm the phase identity of the two phases before and after mechanical stimulation (Supplementary Figure 5). These results confirmed that after the transformation induced by touch, the crystal was in form II (see the response to the following comment). Response to the comment: As we discussed in our response to some of the previous comments, only the crystal of form I, which is metastable at room temperature, can be transformed to form II by applying local pressure (mechanosalient effect). The original figure 1B may have not been very descriptive because the portion of the crystal that was attached to the base was in form II (converted upon affixation of the crystal), while the remaining, larger portion of the crystal was in form I. The section that was in form I was actually contacted with a metal object.

Supplementary Figure 3 │ Mechanosalient effect of crystals of form I terephthalic acid
In order to clarify this aspect of the transformation further, in the revised version we have replaced the series of snapshots in Figure 1B with another series of a images recorded by using high-speed camera ( Figure 1F). We also performed analysis of the mechanosalient effect in a crystal before, during and after the phase transition by determining the unit cell in all cases. This proved to be challenging because once the transition is initiated, it proceeds in very short intervals, even with lightest mechanical stimulation; it is thus very difficult to "trap" the crystal in a stage where only portion has been converted.
To that end, a single crystal mounted on the diffractometer head was first heated over the transition from form II to form I several times, and then slowly cooled to room temperature. At room temperature, form I is metastable. Once this metastable form I is contacted with a metal needle, it starts transforming to form II. This partially transformed crystal was covered in Paratone to slow down the conversion to form II, and the unit cell was determined with an in-house diffractometer at different locations to confirm the phase identity. After 15 min, the crystal was completely transformed to form II and single crystal diffraction measurements were performed again to confirm that the crystal is completely converted to form II. The locations where the unit cell was determined are marked in a new figure, which is now included in the revised version as Supplementary Figure 4. The crystallographic data are provided as new Supplementary Table 1. The identity of the phase in the two regions was additionally confirmed by using Raman spectroscopy. The Raman spectra are provided as Supplementary Figure 5.  Figure 1D: This is identical to Figure 4A  Response to the comment: We share the reviewer's sentiment regarding the possible conceptual overlap with Figure 4A in our previous publication, at least in view of the visual presentation of the phenomenon in Figure 1D, although the two figures show different samples. As we have elaborated earlier in this response, in our previous publication the phenomenon was noted, but its mechanism remained speculative and it was not substantiated by experimental evidence. Specifically, we were not able to explain the mechanism of what appeared to be a shape recovery similar to that encountered with mesophasic materials, such as polymers; nevertheless we decided to report our observations. The authors of this work strongly believe in the importance of reporting observation, regardless whether they can be explained at the time when they were made. We also believe that the relevance of a follow-up, in-depth research should not be alleviated by avoiding to report such observations.

Supplementary Figure 4 │ A single crystal used for in situ X-ray diffraction analysis of a mechanically stimulated phase transition. A crystal of form II terephthalic acid was covered in Paratone to slow down the transition and mounted on the diffractometer head. The crystal was taken over the transition between form II and form I several times by
The scope of the present work was to provide an in-depth insight into the phenomenon and to investigate the details of this unusual property by using different analytical techniques. For completeness of the current study, we needed to show one panel with this phenomenon, so as to avoid the necessity for reference to a figure in an earlier publication. Specifically, in the work presented here we provide exhaustive experimental evidence with results that explain and provide evidence for our earlier observations. Moreover, by approaching the preliminary results with a further in-depth study we not only provide direct evidence for the structure of a crystal that can be bent without delamination, but we also establish the mechanism by which the crystal of TA bends as a new mechanism for bending of organic crystals, which could be of more general importance than the single example studied here. Namely, as we have also noted in the manuscript, this mechanism for crystal bending could be applicable (and indeed, practically more relevant) to other crystals which have metastable phases at ambient conditions. These conclusions, we believe, should warrant publication of the new results presented here.
In order to provide these new details on the phenomenon, Figure 1D was now replaced with another series of images which show a new property that we believe has an added value to the property that we observed beforethe reversibility of the shape change that can be induced mechanically once and reverted thermally multiple times. We also show, both in the new Figure 1D and, due to space limitations to a greater extent in Supplementary Figures 8, 9 and 10, the behavior of the crystals of TA after single or repeated alternative treatment with hat and/or mechanical force. We hope that this novel aspect is acceptable to be presented together with the other results on this material.  Figures 11 and 13) and tables ( Supplementary Tables 2 and 3) that show images and contain data related to the restoration of crystal during heating (the tables contain the unit cell parameters). When the crystal of form II is heated to 359 K, it is completely transformed to form I. In order to confirm that, we have performed single crystal X-ray diffraction analysis on different locations of such crystal. Other crystals were damaged and we recorded X-ray powder diffraction above 359 K. the powder diffraction pattern in the high temperature phase corresponds to that of form I calculated from its crystal structure. These results confirmed that in all cases the samples were fully converted to form I after heating. The corresponding powder diffraction and single crystal X-ray diffraction results are now included in the Supplementary Material (new Supplementary Figure 15). We would like to retain Figure 1E in the main text because we consider that it depicts well the partial shape-restoration on a heavily damaged crystal. For convenience, the Supplementary  Figures 11-15 showing images of damaged crystals and locations where the diffraction data were collected, as well as the new Supplementary Tables 2 and 3, are provided below.  Response to the comment: The crystal shown in Figure 2 was mechanically damaged by compressing it uniformly with a metal plate. During this procedure, a kink developed along the long axis of the crystal. Careful examination of the supplementary movie 17 shows that upon heating to the phase transition temperature, the shape restoration process propagates starting from both ends of the crystal and is eventually completed at the kink. In the revised version, we have included Supplementary Figures (see above) and Supplementary Movies of several crystals that show complete phase propagation during heating. We have also performed single crystal X-ray diffraction on the cracked crystals of TA before and after the phase transition. These results confirm that the transformation is complete (the corresponding data are now included in the Supplementary Material, and are also provided below for convenience). The phase transformation of form II to form I is partially affected by the defects caused by the metal plate. During heating, the propagation is terminated at defects and the transformation of a larger portion of the crystal is delayed for a few milliseconds. However, the transformation is eventually completed once the temperature is raised above 358 K.

Comment:
Section "Microfocus X-ray diffraction analysis of a bent crystal" • As indicated in manuscript, Supplementary Figure 5 shows long range order, but also shows several diffraction peaks close together, which can indicate twinning or gliding of the different layers. Explanation of what they are is needed.
Response to the comment: We have carefully evaluated the synchrotron X-ray diffraction data collected from the bent region (both on the convex and the concave side of the kink) for twinning, and the data was also checked against alternative unit cell parameters. The inspection for possible twinning and the reciprocal lattice view of form I and form II in the bent region did not show any twinning. However, a few peaks appear close together and we believe that this is due to the compression of the layers during the bending. We have included more diffraction images in the revised supplementary material  Response to the comment: In order to verify this, we bent several crystals using threepoint bending method and the critical strain was found to be 2.5 ± 0.2%. We did not observe any phase transition below the critical strain. The following sentence was added to the main text: "The critical strain for bending, based on several crystal samples, was found to be 2.5 ± 0.2%; below this strain no phase transition occurs (Supplementary Figure 22). The minimal force required to induce the phase transition of form II to form I by bending of the crystal obtained from the force-displacement profile along with the critical strain and averaged over eight crystals was determined to be 92 ± 5 mN (an exemplary forcedisplacement and stress-strain curves are shown in Supplementary Figure 7)."

Supplementary Figure 22 |
Stress-strain curves for TA crystals of form II bent by the three-point bending method. The critical strain was found to be 2.5 ± 0.2%.

Comment: • It would be interesting to see if the crystal integrity can still be maintained if the crystal is bent the opposite way.
Response to the comment: Form II crystals of TA can be bent either be exterting pressure either on their (010) face or on their (01 ̅ 0) face. Our experiments showed that the crystal integrity is well preserved in both cases. We have included in the revised version of the supplementary information a new figure (Supplementary Figure 10) that shows the mechanical bending by applying force on both faces. Consistent with the other results, partial shape recovery was observed when this crystal was heated above the phase transition temperature. Upon cooling the crystal regains its original bent shape on both faces.  Comment: Section "Mechanism of plastic bending by partial phase transition" • Authors described in last paragraph before Figure 5 "This causes sliding and rotation of nearly 30 of the tapes on the concave side, ..." which angle is this referring to? The author never described in the manuscript.

Supplementary
Response to the comment: In the revised version we modified the sentence to clarify its contents. In form II, due to - stacking, the molecular tapes form infinite stacks along the crystallographic a axis. Furthermore, the molecular tapes are connected via C-H···O contacts along the crystallographic c axis. The molecular tapes in form II are stacked at a tilt angle ~55 with respect to the crystallographic a axis. During the phase transition this tilt angle changes to ~83, and the molecular tapes slide and are realigned to new positions.
The following sentence was modified: "This causes sliding and rotation of the molecular tapes in form II to nearly 30 with respect to the crystallographic a axis on the concave side, where they reform the sheets along the crystallographic c axis, and the concave part of the crystal is transformed to form I." Comment: • Also authors mentioned in the same paragraph that: "The slight mismatch of the two unit cells, which-similar to their coexistence as alternating layers in a straight partially converted crystal-remain attached at the phase boundary, stabilizes the bimorph and the bent shape of the crystal is retained ( Figure 4D)." As authors show the lattice structure in Figure 5A the mismatch at the phase boundary looks significant (for instance around 26.3 % decrease in a-axis length); which do not mean slight mismatch of the two unit cells. And in Figure 5B on the right schematic exaggerated the lattice dimension of Phase I to seem like it matches to that of Phase II.

Response to the comment:
We have now changed the sentence accordingly in the revised version and also removed the exaggerated lattice dimension of phase I in Figure  5B: "The mismatch of the two unit cells, which-similar to their coexistence as alternating layers in a straight partially converted crystal-remain attached at the phase boundary, stabilizes the bimorph and the bent shape of the crystal is retained ( Figure 4D)." Comment: • Incorrect statement right before Figure 5: "When the bent crystal is heated over the phase transition, the form I on the concave side is transformed back to form II" (it should be form II to form I) Response to the comment: We have rechecked the mentioned sentence and it is indeed correct. In the bent crystal, the convex side is in form II and the concave side is form I. When the bent crystal is heated over the phase transition, the form I on the concave side is transformed to form II.
We thank all reviewers for the valuable comments which have contributed significantly to improve the quality of our manuscript. We considered all comments, and we tried to address them to the best of our ability. Unless stated otherwise, all numbers of the figures and supplementary materials refer to the revised version of the manuscript. Together with this submission we have provided marked copies of the main text and the supplementary materials where the changes to the original version have been marked. For convenience, in what follows, the original comments from the reviewers are highlighted in blue color, our response is provided in black color, and the text that was modified or added to the manuscript is marked with red color.  Response to the comments: We thank the reviewer for their additional comments, and for the very careful reading of the revised manuscript. As authors of this manuscript, we would like to reassure the reviewer that we take each of their comments (as well as those from other reviewers) with utmost respect and very seriously, and as in the first revision, in this second revision we have also tried our best to address their newly expressed concerns to the best of our ability. As authors (and reviewers or many other authors' manuscripts) we truly appreciate and value the time that each reviewer has taken to reread the revised manuscript. As always, we very carefully read, discuss, try to understand, and we work to address every single comment from every reviewer, and we hope that in the revised version of this manuscript we have made the necessary changes to the reviewer's satisfaction. We are very happy to accept sensible and substantiated criticisms based on actual results and the respective conclusions, because that discussion always contributes to improvement of the manuscript's quality.
The article to which the reviewer refers in their comment discusses an issue that was indeed raised in one of the reviewer's comments (that is, whether the samples we have used are single crystals, I quote: "The biggest flaw and most significant is a requirement that the authors demonstrate that these are single crystals -indeed their microfocus diffraction experiments show that they are not single and probably not crystals"), and we thought that instead of elaborating the source of confusion around the terminology and its proper use in great details, a reference to this recently published minireview will help clarify some of the concerns regarding the crystallinity of the samples. In our response to the reviewer's comments we have provided both a written explanation and experimental evidence (diffraction images) regarding the crystallinity of our sample, and we have further supported our explanation with this recent reference which provides additional information based not only on our results but also results from other authors, which we think saves space and time with our response. We do strongly believe that this recent peer-review greatly clarifies the terminology and the proper and adequate use of both terms "single crystal" and "crystalline". However, if the reviewer thinks otherwise, we are happy to provide more detailed explanation in our response to the reviewer's comments and in support of our claims.

Comment:
Having noted and read the extensive, but too long and unconvincing responses to the referees I have reread the paper fresh to consider it in the absence of my previous comments -ie as if it were a fresh submission.
Response to the comment: We again thank the reviewer for their time and attention with the careful reading of the manuscript and the comments. Our extensive comments are aimed at better explaining the changes we have made to respond to the reviewer's comments, and we hope the reviewer appreciates the effort we have taken to address their new comments, which were not part of their first set of comments.

Comment: 1. Line 52 -this is statement is incorrect. Plastic bending results in the loss of discrete diffraction -see Thomas et al.
Response to the comment: We thank the reviewer for expressing their concern regarding the given statement. We are well aware of the work by Thomas, Spackman et al., and we have now modified the statement, and we have also included the relevant references. The following text was added in the main text of the revised manuscript together with additional references.
"Upon elastic or plastic bending of crystals their diffraction ability is generally retained, as concluded from their discrete diffraction patterns. 8,[42][43][44][45][46] However in some cases loss of the discrete diffraction has been observed upon plastic bending. 47 Comment: 2. Line 64-68 the authors assert a new mechanism of bending. This is not supported -it appears that the mechanism of bending results from slippage of molecules (which is accompanied by a phase change), not a new mechanism. This is also a chicken and egg question.
Response to the comment: We are not sure from the comment whether the reviewer in their comment is referring to the mechanism of bending or the mechanism of phase transition. The slippage of molecular hydrogen-bonded tapes in this phase transition is the mechanism of the phase transition which causes bending, so the two processes are related to each other. We provide further clarification and more elaborate answer below.
We would like to reiterate here that as the main concept in this work we have proposed and provided experimental evidence (obtained by X-ray diffraction as well as by using other methods) of a distinctly new mechanism for plastic bending of molecular crystals that occurs by mechanically induced phase transition.
Therefore, the mechanism of the bending is mechanically induced phase transition. An illustration of this mechanism is provided in Figure 4D. This is entirely new mechanism for plastic bending of molecular crystals, where in the bent crystal two phases co-exist across a phase boundary, and the bending occurs irrespective of delamination and sliding of molecular slabs, a mechanism which is well established and common for plastic bending of some other plastically bendable crystals. As a support of this distinct mechanism, we have also demonstrated that the bent crystals of TA undergo a shapememory effect, which is based on the reverse phase transition. This effect would not have been possible if the bending occurred simply by delamination and sliding of slabs.
Specifically, application of localized pressure in form II of TA induces a phase transition on the concave side of the crystal, and the two phases coexist in the bent region, as we have confirmed by X-ray diffraction analysis. We would like to note that the material that we report here provides the first ever example of plastic bending by a mechanically induced phase transition, and the form II of TA is the first compound which shows phase transition in the bent region upon application of force. We believe that similar bending is possible with other biphasic materials. The relatively soft nature of form I and the small structural difference between the two phases stabilizes the resulting bimorph in the bent region at room temperature. The bimorph can be converted to a single phase by heating or cooling, and this results in apparent partial recovery of the straight shape of the crystal, which visually appears as a shape-memory effect.
Having that said, the mechanism of the phase transition, or the change between the two phases, which can be induced by either heating/cooling, occurs via slippage of molecular chains in a single-crystal-to-single-crystal manner. This means that the main difference between the two phases is offset of the molecular tapes from each other, which results in slight difference in the hydrogen bonds, as is clearly explained in the text. During the phase transition the molecular tapes of hydrogen-bonded TA molecules slide in respect to each other and take up new positions relative to each other, which results in change in the centroid-to-centroid distances. In the current work, the mechanism for plastic bending of the crystal by a mechanically induced phase transition and formation of a bimorph in the bent region is unique, although the phase transition does proceed through slippage of molecular tapes of hydrogen-bonded molecules. We also confirm that TA form II is the first example which shows phase transition in the bent region upon mechanically induced bending, and exist with form II and form I in the convex and concave regions respectively in the bent part of the bent crystal.

Comment: 3. There are not statistics or errors or objective measures the restorative effects. It is likely that the origins of these effects are defects and/or phase boundaries introduced upon deformations
Response to the comment: We note that it is generally very difficult to quantify the restorative effects, however, in the meantime, to respond to the reviewer's comments, we have performed additional experiments that further confirm our claims. The new results are provided in the revised version of the supplementary material. We also note that the defects and phase boundaries play very different role in this mechanism.
We would like to reiterate here that according to the mechanism that we have established, the shape restoration of form II TA crystals occurs after partial phase transition to form I. Upon mechanical bending, the side that is impacted of form II TA crystal transforms to form I and remains in that phase in the bent region. Therefore, both form II and form I coexist in the convex and the concave regions of the kink of the bent crystal, and can be converted to each other either by heating or cooling. This results in apparent partial recovery of the straight shape of the crystal, and is observed as the shape-memory effect.
The main reason for the bent shape is the formation of a bimorph, which necessarily includes a phase boundary between the two phases. Namely, the conversion of form II to form I in the bent region during mechanical bending leads the generation of phase boundary between the two phases. When the crystal is heated, one of the phases converts to the other phase, so that the phase boundary disappears and the crystal straightens. Therefore, the main driving force of the shape-memory effect is the formation of a phase boundary. If there was no phase boundary, and the bending was due only to formation of defects, the crystal would not be able to recover its shape. Having said that, defects are always created when a soft crystal is impacted mechanically. However, these mechanically created defects can not be restored thermally, that is, if the crystal was bent by delamination and sliding of slabs, its original shape will never be recovered by heating.
In order to respond to the reviewer's comments, we have now added the following figure in the supplementary material. This figure shows quantification, expressed as the angle of the bent crystal by heating and cooling of the three different crystals. These results clearly show the reversibility of the shape memory effect.