Thermodynamically stable whilst kinetically labile coordination bonds lead to strong and tough self-healing polymers

There is often a trade-off between mechanical properties (modulus and toughness) and dynamic self-healing. Here we report the design and synthesis of a polymer containing thermodynamically stable whilst kinetically labile coordination complex to address this conundrum. The Zn-Hbimcp (Hbimcp = 2,6-bis((imino)methyl)-4-chlorophenol) coordination bond used in this work has a relatively large association constant (2.2 × 1011) but also undergoes fast and reversible intra- and inter-molecular ligand exchange processes. The as-prepared Zn(Hbimcp)2-PDMS polymer is highly stretchable (up to 2400% strain) with a high toughness of 29.3 MJ m−3, and can autonomously self-heal at room temperature. Control experiments showed that the optimal combination of its bond strength and bond dynamics is responsible for the material’s mechanical toughness and self-healing property. This molecular design concept points out a promising direction for the preparation of self-healing polymers with excellent mechanical properties. We further show this type of polymer can be potentially used as energy absorbing material.

The present manuscript by Z. Bao et al. describes the synthesis of self-healing polymers based on PDMS and ligands -namely bimcp, which is capable of complexing zinc ions. The synthesis of these novel materials is described in detail. Additionally, the complexation of the metal ions was investigated revealing the structure of the complexes. The mechanical properties of the polymers as well as the ability for self-healing was studied extensively. Noteworthy, the resulting materials feature very good mechanical properties, in particular a very high the toughness. Scratch healing as well as healing of specimen was investigated. Futhermore, these materials feature very good energy absorbing properties, which was finally demonstrated by the coating of a sponge. Minor issues: -The authors depict the complexes as the charged species with protons. Alternatively, the complexes could also be neutral (zinc (II) and the phenolates. What is the evidence for these charged complexes? If it is positively charged, what are the counter ions? -Considering the proton at the imine (see above) imine metathesis might also contribute to the healing process in the solid state. Imines, in particular in acidic media, are prone to hydrolysis or can self-exchange. Have the authors observed an exchange? Is the molar mass distribution of the polymer changing? (decomplexation by HEDTA and SEC measurement of the resulting linear polymer) -Force spectroscopy measurements have been performed. Have the authors compared these results with the contour length of the polymers? For PDMS with higher molar masses around 12 nm are given in literature (DOI: 10.1038/NCHEM.2492. With the smaller molar mass in the present study much smaller values could be expected. However, this effect seems not to be present in the measured curves, which feature much larger distances.
-The rupture force of the zinc complexes could be compared with literature values of other complexes. Are these values high or low? -The healing efficiencies for the other healing times could be given in the Supporting Information.
-The higher zinc content in Zn2(bimcp)2-PDMS makes the polymer films stiffer compared to the Zn(bimcp)2-PDMS. How would the Zn1.5(bimcp)2-PDMS behave? A smaller amount of Zn2(bimcp)2-complexes could potentially reinforce the polymer. Supporting Information -Is the dispersity of the PDMS known? - Figure 1: Where do the protons come from? At low temperature there are two protons, the imines are depicted as protonated form. These moieties would be charged positively. These protons could / should also appear in the NMR spectrum.
-13C NMR data could be added - Figure 10: A scale bar for the microscopic images is missing - Figure 11: The strain could be added - Figure 16: The legend (top left) is missing -General comment for the NMR spectra: Sometimes the integral (green line) seems to be smaller than the corresponding peaks (consequently lowering the numbers) -for instance Figure 25 peak at around 1 ppm.
Reviewer #3 (Remarks to the Author): The authors report a design principle to synthesize a strong, stretchable, and selfhealable elastomer using Zn-bimcp coordination bonds. This kind of coordination complex is thermodynamically stable whilst kinetically labile. They use PDMS elastomer as a representative material and show combined excellent mechanical properties and selfhealing capabilities. The PDMS elastomer exhibits Young's modulus~43.68 MPa, stretchability~24, toughness~29.3 MJ/m3, and self-healing effiency~98.9%. This type of material is potentially used as energy absorbing material in many fields.
The reviewer suggests the paper be published if the authors can address the following issues.
• In the introduction part, the authors are recommended to give additional explanation on why the Zn-bimcp coordinate system is unique, such as high association energy and high healing efficiency, etc., compared to other metal-coordination systems, disulfide bonds and many types of hydrogen bonds that are described in the reference 13, 18, 23, and 24.
• In Fig. 1e, two peak forces are observed in stretching bimcp-PDMS polymer chain, what is the event corresponding to each peak?
• In Fig. S7, why is the force in second stretch much lower than the first stretch? Are those continuous tests in a same polymer chain?
• In describing Fig. 2C, please add explanation on how the Zn-bimcp plays role in the ratedependent stress-strain curves?
• In Fig.2C, the Zn-bimcp-PDMS demonstrates excellent stretchability, even with 2400% strain at a loading rate of 10 mm/min, does it restore its initial configuration given enough time? Does the hysteresis also depend on the loading rate? How does the hysteresis change with loading cycles? The reviewer suggests conducting stress-strain curves with multiple cycles for each strain rate.
• In the self-healing experiment, do the authors press the damaged parts together to make them contact intimately? Are there gaps between the healing interface that hinder healing?
• In the coating of sponge, what is the typical thickness of the coating? How does the quality fraction of coated Zn-bimcp-PDMS influence the coating thickness?
• On line 195, "module" should be "modulus", please carefully check typos in the manuscript. 1

Point-by-point response to reviewers' comments
Reviewer #1 (Remarks to the Author): The manuscript "Thermodynamically stable…," by Lai et al reports the method to overcome the trade-off between the mechanical properties and dynamic nature of self-healing by the design and synthesis of a polymer containing thermodynamically stable whilst kinetically labile coordination complex of Zn-based, three-dimensional, alterdentate ligands. The authors focus on the tunable nature of unique noncovalent interactions between a metal ion and its surrounding organic ligands. Several material designs based on metal-ligand coordination using different metal ions and ratios are selected and they are introduced as crosslinking units into PDMS polymer to synthesize self-healing polymers. For those material designs and synthesized polymers, various spectroscopic analyses such as NMR, ESI-MS, FT-IR, UV-vis and force spectroscopy are well conducted. Also, the mechanical and rheological properties are well-studied. Its scientific novelty is well-recognized. Furthermore, the authors suggest their potential applications as energy absorbing materials together with elastic polyurethane sponge. Although the work contains no apparent serious flaw, several critical issue and minor issues as listed below should be addressed before the publication of this manuscript in Nature Communications.
Response: We appreciate your positive comments. Zn(Pr-Hbimcp)Cl 2 is the only clean and characterizable product both in solution and at solid state. The purity can be evidenced by comparison of the X-ray powder diffraction peaks of the experimental and simulated data ( Figure R3).  However, we believe the lack of time-and temperature-dependent 1 H NMR of Zn 1 (Hbimcp) 2 and Zn 2 (bimcp) 2 will not affect our conclusion in our manuscript. The mono-nuclear complex Zn(Pr-Hbimcp)Cl 2 showed rapid ligand exchange reaction above room temperature, indicating that the Zn(II)-Hbimcp complexes are highly dynamic. As for the ligand exchange reaction rate, we can get a quantitative evaluation by comparing the relaxation time from stress-relaxation test and timetemperature superposition data of the polymers. As shown in Figure R4a, the stress of Zn 2 (bimcp) 2 -PDMS polymer decrease much slower than that of Zn(Hbimcp) 2 -PDMS at different initial strain. The residual stress is still significant after 3600 s ( Figure   R4b), which indicated there was no obvious ligand exchange phenomenon in the polymer of Zn 2 (bimcp) 2 -PDMS within 1 h. Temperature dependent characteristic relaxation time can also manifest the different ligand exchange speed in these polymers. As derived from time-temperature superposition (TTS) of rheological data, the characteristic relaxation time of Zn 2 (bimcp) 2 -PDMS at 25 °C is as long as 4 × 10 4 s ( Figure R5), which is much longer than Zn(Hbimcp) 2 -PDMS (8 s).  Figure R7). The intra-molecular ligand exchange is fast and can be observed at room temperature. Moreover, the intra-molecular ligand exchange can accelerate the inter-molecular ligand exchange due to the generation of meta-stable three-coordinated intermediates ( Figure R6). In contrast, the inter-molecular ligand exchange is slow and has to be activated by heating or solvation effect. That is why the Zn 2 (bimcp) 2 -PDMS polymer show much longer relaxation time. We have added these data and discussions in the revised manuscript. Figure R6. The intra-molecular and inter-molecular ligand exchange process in Zn-Hbimcp complexes.
2. The authors are highly encouraged to conduct more experiments using Zn-bimcp-PDMS coated sponge as energy absorbing materials to emphasize its high novelty and impact. For example, drop a heavy stuff onto the both Zn-bimcp-PDMS coated sponge and reference sponge, and compare their shock absorbing capabilities.
The data in current form is not enough to make the potential readers impressed.
-More about energy absorbing experiment, the reviewer wants to check the effect of compression speed (or the duration time at full compression). Does the polymer coated sponge return to its original shape directly when the speed is faster (or duration time is shorter)? Or more lately return to original shape when the speed is lower (or duration time is longer)?
-Please describe the experimental procedures in detail in the Supporting Information when the compression speed was higher or the duration time was shorter ( Figure   R9a and Figure R10a), but there is no obvious effect on the blank sponge ( Figure   R9b and Figure R10b). This observation is also reasonable because less time is allowed for the ligand exchange processes and re-formation of the complexes at 10 higher compression speed or shorter duration time, thus the viscoelastic Zn(Hbimcp) 2 -PDMS behaves like an elastic polymer. Therefore, there is no significant residual strain and the configuration of the Zn(Hbimcp) 2 -PDMS coated sponge can be recovered quickly.  Response: According to your suggestion, we have investigated the self-healing and stress-relaxation properties of Zn 2 (bimcp) 2 -PDMS. The self-healing properties of Zn 2 (bimcp) 2 -PDMS were indeed inferior. We have partially damaged the Zn 2 (bimcp) 2 -PDMS polymer (remain 10% uncut to keep the incision better contact) and measured the self-healing properties at room temperature (25 °C). As shown in Figure R11, the healing efficiency for 48 h was still very low (< 2%). The stress-relaxation results showed that the stress of Zn 2 (bimcp) 2 -PDMS polymer decreased much slower than that of Zn(Hbimcp) 2 -PDMS at different initial strains.
The residual stress was still significant after 3600 s ( Figure R4b), which indicated there was no obvious ligand exchange phenomenon in the polymer of Zn 2 (bimcp) 2 -PDMS within 1 h. As derived from time-temperature superposition (TTS) of rheological data, the characteristic relaxation time of Zn 2 (bimcp) 2 -PDMS at 25 °C is as long as 4 × 10 4 s ( Figure R5), which is much longer than Zn(Hbimcp) 2 -PDMS (8 s). These data indicate that although the inter-molecular ligand exchange process may be possible in the polymer film, the speed is too slow.
That is why the Zn 2 (bimcp) 2 -PDMS polymer showed inferior self-healing properties. Figure R11. The self-healing properties of Zn 2 (bimcp) 2 -PDMS at 25 °C. b is the partial enlargement of a.
4. The reviewer wants to know the reason why the authors designed a polymer structure containing (5-chloro-2-oxy-1,3-phenylene)dimethanimine, especially, Response: From our previous study, we found that the mechanical properties of a polymer are determined by the thermodynamic stability of the crosslinking sites. The more stable (i.e. higher association constant) of the crosslinking sites, the stronger and tougher but less dynamic of the polymer. In contrast, the self-healing rate of a polymer is determined by the kinetic lability of the crosslinking sites. Therefore, to achieve both high toughness/high modulus while having a rapid self-healing rate, a molecular design concept for the crosslinking site that is both thermodynamically stable whilst kinetically labile is needed. Alterdentate ligand can provide two equivalent donor centers. However, due to the steric hindrance, the two equivalent donor centers can not coordinate with the same metal ion at the same time and therefore the two coordination atoms are alternative and interchangeable. Take N-O-N chelation for example, when one imine-N is coordinated with Zn, the other uncoordinated imine-N has the ability to replace the previous coordinated imine-N and coordinate with Zn.
We envisage that such a unique coordination system would be an ideal crosslinking site that is both thermodynamically stable and kinetically labile, and can lead to strong and tough self-healing polymers. That's why we designed the polymer structure containing (5-chloro-2-oxy-1,3-phenylene)dimethanimine ligands. We have revised the introduction part of the manuscript to make this design concept more clear to the readers.
5. The reviewer found a similar reference (Nat. Commun. 2018Commun. , 9, 2725 by Prf. Jing-Lin Zuo. It is recommended to make mention of this ref. in the main text; about differences of association constant, mechanical, self-healing properties. The reviewer 13 is concerned why Zn-bimcp polymer (this study) exhibited lower tensile strength, though it has a higher association constant than PDMS-COO-Zn (ref.).

Response:
The mechanical properties of a polymer are determined not only on the strength but also on the density of the crosslinking sites. In this study, although the crosslinking site has a higher association constant as compared to PDMS-COO-Zn (Nat. Commun. 9,2725), the density of the crosslinking site is much less. Therefore, the degree of freedom of the polymer segment between the two crosslinking sites is higher, leading to higher tensile strain and higher toughness. Prof. Von Zelewsky. According to your suggestion, we have added the description of "alterdentate ligand" in the Supporting Information as "According to the reference 35, the definition of "alterdentate ligand" is that a species which offers to a metal ion more than one equivalent coordination site. In an "alterdentate ligand" there is, principally, always a re-arrangement possible in which the metal is transferred from one site to another one. This can be either an inter-or intra-molecular process. The rearrangement reaction is kinetically controlled by the activation energy and entropy experienced by the metal on the reaction path. The free energy difference is zero by definition, if the coordination sites are equivalent."

2018,
1. In abstract, it is recommended not to use abbreviations such as "Zn-bimcp".
Response: Thank you for pointing out this mistake. We have added the full name of the abbreviation in the abstract.
Response: Thank you for pointing out this mistake. We have added the label for chlorine on the structure.
Response: Thank you for pointing out this mistake. We have replaced "roughness" with "toughness" in the revised manuscript. Response: Thank you for pointing out this mistake. The mistake has been corrected.
6. Supporting Figure 18. Check strain unit (% -> mm/mm) Response: Thank you for pointing out this mistake. The strain unit has been modified to mm/mm.  Table R1, the comparison between this manuscript and the prior published results revealed that the polymer reported in this work has the highest Young's modulus, toughness, healing efficiency and competitive healing temperature, healing time, maximum strain. The excellent mechanical and self-healing properties was achieved through the design of a thermodynamically stable whilst kinetically labile coordination system, which has never been reported before.    Figure R12) and [Zn(Bz-Hbimcp)]Cl 2 ( Figure R13) again. This time we use DMSO as solvent to observe the signals of active hydrogen, which are hard to observe in protic solvents. As show in Figure R12 and Figure   For [Zn 2 (bimcp) 2 ]Cl 2 type complexes, although we were not able to get the 1 H NMR spectra to prove the absence of signals of active hydrogen, the crystal structure clearly shows that the phenol group was deprotonated. This is due to that, when both the O and N atoms are coordinated to Zn(II) metal ions, the O-H or N-H bonds are significantly weakened and the H atom is easy to leave although no base was added.
Based on these observations, we can conclude that for Zn-Hbimcp complexes with unbounded N atoms, the phenol group was not deprotonated. The H atom was located between O and N atoms (more close to N atoms as revealed by single crystal X-ray crystallography). For Zn-Hbimcp complexes without unbounded N atoms, the phenol group was not protonated. In order to differentiate these structures, we denote 2) We tracked the changes of the molar mass distribution of the polymer Hbimcp-PDMS by GPC under different condition. One is the as prepared sample, 22 one is the sample stored at room temperature for 7 days, and the other one is the sample complexation by ZnCl 2 and then decomplexation by terpyridine in the dichloromethane solution (HEDTA was not adopted because it is insoluble in dichloromethane). The results showed that there were no obvious molar mass distribution changes under the different conditions ( Figure R16), indicating that the self-exchange process in the polymer is unconspicuous. Based on the above observations, we believe that imine metathesis do not contribute significantly to the healing process for Zn(Hbimcp) 2 -PDMS polymer.
Presumably due to that the PDMS matrix is not favorable for imine exchange since the imine exchange rate is sensitive to the environment (Macromolecules 2016, 49, 6277−6284). Such conclusion can be evidenced by other observations. First, as we stated in our original manuscript, the polymer Zn 2 (bimcp) 2 -PDMS and Ni(Hbimcp)-PDMS do not show self-healing behavior although they contain the same Hbimcp ligand with imine groups. Moreover, we crosslinked the bis(3-aminopropyl) terminated linear oligomer H 2 N-Hbimcp-PDMS-NH 2 with tri-functional homopolymer of hexamethylene diisocyanate (THDI) to obtain the crosslinked polymer THDI-Hbimcp-PDMS and measured the self-healing properties at room 23 temperature ( Figure R17). We also did not observe obvious self-healing phenomenon even after two days ( Figure R18). ( Figure R20). Therefore, the contour length increments for Zn(Hbimcp) 2 -PDMS will not show significant difference with those of Fe-Hpdca-PDMS. Response: The rupture force of the zinc complexes is about 108.5 ± 40.9 pN, which is similar to the Fe(III)-pdca complexes in our previous study (103 ± 12 pN, Nat. Chem. 2016, 8, 618-624), but lower than the ferric-thiolate bonds (160 ± 60 pN at pH = 6 and 211±86 pN at pH = 7.4, Nat. Commun. 2015, 6:7569) and gold-thiolate bonds (165 ±55 pN, J. Am. Chem. Soc. 2015, 137, 15358−15361). However, it should be noted that the rupture force from AFM study reveals the kinetics of the transition from the bound state to the unbound state over a potential energy surface. The rupture forces will vary significantly with different loading rates and probe stiffnesses (Methods 2013, 60, 142-150). Therefore, the force measured in different situations and for different systems can not be directly compared. What we can know for sure from these data is that the single-chain of Fe-Hpdca-PDMS and Zn(Hbimcp) 2 -PDMS molecule can be unfolded and quickly refolded due to the dynamic rupture and reconstruction of coordination bonds.
-The healing efficiencies for the other healing times could be given in the Supporting

Information.
Response: Thank you for your suggestions. The healing efficiencies for the other healing times have been given in the Supporting Information and list in Table R2. Response: Yes, a smaller amount of Zn 2 (bimcp) 2 -complexes could reinforce the polymer films and make them stiffer. As shown in Supplementary Figure 21, with increasing the content of Zn(II), the obtained polymers exhibited the higher modulus and maximal strength, but the breaking strains were also declined. Response: The protons were from the phenolic hydroxyl group after coordination. The signals of protons appeared in the 1 H NMR spectra with the solvent of DMSO, which showed in Figure R12 and R13.
-13 C NMR data could be added Response: The 13 C NMR data has been added in the Supplementary information. -General comment for the NMR spectra: Sometimes the integral (green line) seems to be smaller than the corresponding peaks (consequently lowering the numbers) -for instance Figure 25 peak at around 1 ppm.
Response: Thank you for your kind remind. The integral have been adjust to fit each peaks.
Reviewer #3 (Remarks to the Author): The authors report a design principle to synthesize a strong, stretchable, and complex has a relatively large association constant. Such a unique coordination system would be an ideal crosslinking site that is both thermodynamically stable and kinetically labile, and can lead to strong and tough self-healing polymers which can not be easily achieved through designing of covalent bonds (such as disulfide bonds) or non-covalents interactions (such as hydrogen bonds)." • In Fig. 1e, two peak forces are observed in stretching bimcp-PDMS polymer chain, what is the event corresponding to each peak?
Response: The first peak can be assigned to the force when the AFM cantilever tip was in contact with the substrate. The second peak can be assigned to the force when the polymer chain was detached from the glass substrate.
• In Fig. S7, why is the force in second stretch much lower than the first stretch? Are those continuous tests in a same polymer chain?
Response: In Fig. S7 (Supplementary Figure 10 in the revised manuscript), we didn't put the two curves in the same vertical axis but use the same scale bar in order to make the two curves distinguishable to readers. The force in the second stretch is 30 actually similar to the first stretch. The continuous tests were performed with the same polymer chain.
• In describing Fig. 2C, please add explanation on how the Zn-bimcp plays role in the rate-dependent stress-strain curves? Response: Thank you for your suggestion. When the strain speed increases, as manifested by the curve at 100 mm min -1 , less time is allowed for the ligand exchange processes and re-formation of the complexes, which reduce the fracture tolerance and increase the tensile stress. In contrast, when the strain speed decreases, as manifested by the curve at 10 mm min -1 , more time is allowed for the ligand exchange processes and re-formation of the complexes, which increase the fracture tolerance and reduce the tensile stress. The similar explanation has been reported in our previous articles (Nat. Chem. 2016, 8, 618-624). The sentence of "When the strain speed increases, less time is allowed for the ligand exchange processes and re-formation of the complexes, which reduce the fracture tolerance and increase the tensile stress" has been added in the manuscript.
• In Fig.2C, the Zn-bimcp-PDMS demonstrates excellent stretchability, even with 2400% strain at a loading rate of 10 mm/min, does it restore its initial configuration given enough time? Does the hysteresis also depend on the loading rate? How does the hysteresis change with loading cycles? The reviewer suggests conducting stress-strain curves with multiple cycles for each strain rate.
Response: Thank you for your suggestions.
1) The stretched Zn(Hbimcp) 2 -PDMS polymer will not restore its configuration even given enough time. The configuration of stretched polymer has been tracked by camera with the interval of 1 day. As shown in Figure R21, there are no obvious changes in length and shape after 1 day. The reason why the stretched polymer could keep its configuration is that there is rapid ligand exchange processes which fixes the shape after stretching; 31 Figure R21. The configuration keeping property of the stretched Zn(Hbimcp) 2 -PDMS polymer.
2) Yes. The hysteresis depends on the loading rate and loading cycles. When the loading rate was decreased, a higher residual strain and more obvious hysteresis were observed ( Figure R22). When increasing the loading cycles, the residual strain of the stretched polymer will increase, and the area of the hysteresis loop will decrease ( Figure R23). We have added these data in the revised manuscript; • In the self-healing experiment, do the authors press the damaged parts together to make them contact intimately? Are there gaps between the healing interface that hinder healing?
Response: Yes, before the self-healing experiment, we need to press the damaged parts to make them contact, but no additional pressure was required during the self-healing process. The gaps between the healing interfaces will hinder healing, because the polymer has high mechanical properties and the fluidity of the polymer is not very good.

33
• In the coating of sponge, what is the typical thickness of the coating? How does the quality fraction of coated Zn-bimcp-PDMS influence the coating thickness? Response: The typical thickness of the coating (d) was calculated using the following equation: Where m p is the mass of coated polymer, ρ p is the density of coated polymer, m b is the mass of blank sponge, S b is the apparent surface area of blank sponge which is estimated using BET methods, wt% is the weight ratio of the coated polymer. The density of the Zn(Hbimcp) 2 -PDMS polymer ρ p was 1.07 g cm -3 and apparent surface area of blank sponge S b in this work was estimated as 1.22 m 2 g -1 .
The quality fraction of coated Zn(Hbimcp) 2 -PDMS influence significantly on the coating thickness. According to this equation, the thickness of the coatings with different quality fraction of Zn(Hbimcp) 2 -PDMS were determined in Table R3 as followed. Response: Thank you for pointing this out. The mistake has been corrected.