An adhesive elastomeric supramolecular polyurethane healable at body temperature

We report a non-cytotoxic supramolecular polyurethane network whose mechanical properties can be recovered efficiently (>99%) at body temperature.


Introduction
Materials capable of self-repair offer attractive advantages in many applications, especially in terms of performance and longevity. [1][2][3][4][5][6] In recent years, polymers have been reported that can heal when external stimuli (such as heat, 7-13 pressure, 14,15 light [16][17][18] ) are applied to damaged sites. 19,20 Potential applications for healable materials within modern society include paints, [21][22][23] aerospace composites, 24 regenerative medicine, 25 (in particular articial skin [26][27][28] ) and plastic surgery. [29][30][31] Successful approaches to healable polymer networks that have been reported to date include encapsulated-monomer systems, 24,[32][33][34] reversible covalent bond formation, 15,35,36 utilisation of irreversible covalent bond processes 37 and more recently supramolecular self-assembly. [38][39][40] In the latter approach, network formation is facilitated 40 by a combination of non-covalent bond association (hydrogen bonds, 39,41-44 electrostatic interactions, 45 aromatic p-p stacking interactions [7][8][9][10]46 or dynamic metal-ligand bonds 36,47 ) and phase separation between the polar end-groups and apolar polymer chains, which serves to strengthen the end-groups aggregation, resulting in enhancement of the supramolecular interactions. 10,48 The weak nature of non-covalent interactions permits the materials to possess thermo-responsive and thermo-reversible properties, thereby delivering dramatic viscosity changes over well-dened and tuneable temperature ranges. These addressable and tuneable characteristics 41,49 are highly desirable in both bulk commodity and value-added applications, such as adhesives, 50 shape-memory materials, 51,52 healable coatings 10 and impactresistant structures (e.g. protection for mobile electronics). An important class of supramolecular polymers, which have been developed in the last decade, are polyurethanes. Supramolecular polyurethanes (SPUs) 53,54 are synthesised via reaction of diols or polyols with polyisocyanates and alcohols or amines. 55 The physical properties of SPUs have been shown to directly correlate 41 to the nature of the hydrogen bond receptors that are generated by the reaction of isocyanate end groups and alcohols or amines. 48,[56][57][58] The generation of synthetic materials able to mimic human skin or that are suited for rapid wound isolation is a notable challenge in the biomedical industry. 59 Supramolecular polymers have already been employed in biomedicine as biocompatible thermoplastic elastomers 38,60 and acrylic copolymers utilised in wound dressings in the form of commercially available spray plasters (e.g. Elastoplast®, OpSITE™ and TCP®). [61][62][63][64] Whilst elastomeric, these acrylic-based materials have not been described as healable in nature and thus new skin coatings whose properties offer the ability to repair physical damage in situ by taking advantage of the thermal energy provided by the host but without the requirement of a stimulus such as electrical current 65,66 represent a signicant practical advancement. In this paper, we report the synthesis and healing ability of a SPU whose mechanical properties can be recovered at the temperature of the human body (37 C). In order to demonstrate the potential use of this SPU system in a biomedical setting, we also reveal that lms of this material adhere to pig skin and can be healed in situ post damage. 67

Results and discussion
The supramolecular polyurethane 1 was synthesized 41,42,48 using an established two-step process. Firstly, a hydrophobic and elastomeric diol, Krasol™ HLBH-P2000, was reacted with methylene diphenyl diisocyanate (MDI) at 80 C for three hours to afford a prepolymer featuring isocyanate end-groups. 4-(2-Aminoethyl)morpholine was then added to the prepolymer to install the receptor end-groups via urea bond formation and afford the desired polyurethane 1 in a yield of 93% (see Fig. 1 plus ESI Fig. S1-S3 † for spectroscopic and thermal data). Chain extension in polyurethane 1 was minimal, 1 H NMR spectroscopic analysis revealed a ratio of 1 : 1 for the integrals of the proton resonances of the urethane and urea groups of the endcapping units, consistent with the feed ratios used in the prepolymer and end-capping steps. GPC analysis (THF, room temperature) revealed a material with M n and M w values of 4097 and 4287, respectively.
Dynamic rheological testing was employed to characterise the viscoelastic properties of the polyurethane 1. Temperature and frequency sweeps were performed using an Anton-Paar Physica MCR301 Rheometer, in oscillatory shear. In the low temperatures regime (from 0 to 35 C, Fig. 2a), the elastic rubbery character clearly dominates the properties; there is a gradual decrease of storage modulus and concomitant constant loss modulus with increasing temperature. However, the drop of the storage modulus accelerates above 37 C, owing to the dissociation of the supramolecular polyurethane network formed, 10 and viscous behaviour governs the properties of polyurethane 1 at the temperature of 50 C and above. The master curve shown in Fig. 2b was constructed by manual shiing of isothermal frequency sweep data (the raw data are available in the ESI, Fig. S4 †). 68 The overlapping of data obtained at different temperatures indicates that time/temperature superposition analysis was applied successfully to the rheological data for polyurethane 1. The resultant master curve shows the highly rate dependent behaviour of polyurethane 1 with a typical terminal zone, transition zone to ow, plateau zone (rubbery), and transition zone to glassy behaviour, as indicated in Fig. 2b, over the full investigated frequency regime. Fig. 2c shows the shi factors, a T , used to produce the master curve, as a function of temperature. It is observed that a T changes by over 9 orders of magnitude between À30 and +80 C. This large change in a T over a small, readily accessible temperature range is consistent with dissociation of the supramolecular network, leading to a large drop in viscosity that thereby facilitates healing of damage sites. This dramatic change is attributed to additional relaxation processes, which do not occur in amorphous covalently bonded polymers, and to the disengagement of the supramolecular (p-p stacking or hydrogen-bonding) interactions. For amorphous, covalently bonded polymers, there is a linear relationship between a T and normalised temperature, but recent research shows that this trend is not observed in the case of supramolecular polymers. 69,70 The behaviour of polyurethane 1, shown in Fig. 2d, is consistent with other supramolecular polymer systems: 8,10,68-70 two linear zones are evident, with a transition associated with the dissociation of the network, but with a lower transition temperature than observed in previously reported materials. Importantly, although the mechanical response of polyurethane 1 is comparable to other supramolecular polymers and blends, the temperature at which the intermolecular interactions disengage is close to body temperature.
In order to understand the morphology of polyurethane 1, variable temperature (20 to 100 C) wide angle X-ray scattering (WAXS) and small angle X-ray scattering (SAXS) analyses were conducted. The WAXS scattering pattern shows a lattice spacing of 5.43Å corresponding to the stacking of the urea moieties ( Fig. 3a). 71 It is interesting to note that with increasing temperature the lattice spacing becomes consistently less sharp, suggesting that the hydrogen bonding interactions between the urea moieties of polyurethane 1 are being disrupted. In particular, the change of the morphology is evident at a temperature of 60 C in accordance with the rheology data.
Two Bragg peaks (61 and 145Å) are evident in the SAXS prole ( Fig. 3b), suggesting a microphase-separated morphology arising from the immiscibility of the hard hydrogen bonding end-groups with the so polymer backbone. In addition, in the SAXS prole a drastic change in the morphology is evident at a temperature of 60 C.
Optical microscopy was rst used to probe the healing ability of this supramolecular material. In the light of the rheological data polyurethane 1 was exposed to a temperature of 37 C and the dynamics of the healing process at this temperature was monitored. A sample was cut along the centre, transverse to its long axis with a razor, and then positioned with the cut edges in close contact. Fig. 4 shows two dimensional microscopy images, which reveal that the material surrounding the cut ows into the damage site upon heat treatment and aer 120 minutes, the material becomes essentially homogenous with the position of the cut barely visible. Furthermore, the three dimensional surface prolometry images (Fig. 4) revealed the same behaviour and the surface rough prole shows that the roughness around the cut area is both qualitatively (i.e. visibly) and  quantitatively comparable to the other areas in the surface of the sample, indicating a fully topological recovery of the cut interface.
The microscopy data reveal that the physical integrity of polyurethane 1 can be recovered within two hours at a temperature of 37 C. To examine the recovery of the mechanical properties aer healing, tensile tests were performed on specimens ca. 0.5 mm thick, 40 mm long and 5 mm wide. For the healed samples, the two cut edges were positioned in contact, but not overlapped, as described in previous studies. [7][8][9][10] Experiments were performed on pristine materials and specimens healed for different time periods at 37 C. Pristine material heated to 37 C for 120 minutes was also tested. Mean stress-strain curves with error bars are shown in Fig. 5, and the corresponding mechanical properties calculated from the individual stress-strain curves (see Fig. S5 †) are shown in Table  1. It should be noted that in these experiments, the strain was calculated using digital image correlation on images of the specimen surface. The tensile modulus was calculated from the slope of the stress-strain curve between 0 and 4% strain. It was observed that thermal annealing improved the mechanical properties of polyurethane 1, as revealed by the apparent  aer 120 minutes, respectively). Therefore, it can be concluded that polyurethane 1 can completely recover all the mechanical performance aer healing at a temperature of 37 C for 60 minutes.
The investigation of the healing properties of polyurethane 1 demonstrates that it can recover its mechanical properties fully at body temperature aer 60 minutes, which suggests that it has the potential to be used as a biomedical material, such as articial skin and as an adhesive within temporary wound dressings. 72 Therefore, the mechanical performance of polyurethane 1 under different physiological conditions was investigated -lms of polyurethane 1 were soaked in distilled water and a phosphate buffered saline (PBS) solution at both room temperature (19 AE 0.5 C for 48 hours) and body temperature (37 C for 12 hours). Mean stress-strain curves are shown in Fig. 6, and the corresponding mechanical properties calculated from the individual stress-strain curves (see Fig. S6 †) are shown in Table 2. It was observed that both the modulus and strength decreased up to 30% aer soaking in distilled water or PBS solution, but the elongation to break increased. This suggests that water diffuses into the polymer network and acts as a plasticiser, thus decreasing the stiffness and strength but improving failure resistance. Despite the water absorption, the chemical integrity of the bulk material is stable under exposure to these physiological conditions, as indicated by the mechanical performance of polyurethane 1 (compared to pristine polyurethane 1) at different physiological conditions (see Table 2).
Polyurethanes are routinely employed as adhesives in a diverse range of applications including biomedical devices. 26,29,67 Within this context, the adhesive properties of polyurethane 1 were investigated using pig skin as a model substrate. A simple manual peel off test was carried out to  a This modulus was calculated using forces measured by the mechanical testing machine and local strains measured using an optical technique (digital image correlation) between strains of 0 and 3.5%. The value is therefore a linear approximation to the true, non-linear, polymer behaviour. investigate the adhesive properties of polyurethane 1 rst. The skin was washed with acetone to remove residual fats and preservatives; a lm of polyurethane 1 was then placed between two pieces of washed skin. The obtained sandwich structure with polyurethane 1 lm in the middle was placed in an oven at a temperature of 37 C for a period of 4 hours. Fig. 7 shows a This modulus was calculated using forces measured by the mechanical testing machine and local strains measured using an optical technique (digital image correlation) between strains of 0 and 3.5%. The value is therefore a linear approximation to the true, non-linear, polymer behaviour. images of polyurethane 1 being removed from the skin manually. It can be observed that during the peel off, there is a large deformation of polyurethane 1 lm, and that cohesive failure occurs occasionally (as shown in the pictures of the failure from the pig skin surface and the large deformation of polyurethane 1 before failure), which indicates that good bonding properties can be achieved between the pig skin and the lm of polyurethane 1. To quantify the adhesive strength peel tests were performed on samples of width 1.25 mm and length 80 mm, using a commercial tensile test frame. A rig was designed to hold the sample and apply the loading. The specic arrangement of the test sample and setting of the rig is shown in Fig. S7 in the ESI. † As a result of the difficulties in cutting sufficiently at skin samples, it proved impossible to maintain uniform contact (thus uniform pressure) across the whole sample during preparation, which leads to signicant variation of the peel force during the test, 73 as observed in Fig. 7 (force vs. displacement curve). However, preliminary results strongly suggest that stable peel strength can be generated and a peel force of 2 N can be achieved (the high force region corresponding to the dendritic failure surface due to the large deformation polyurethane 1 experienced during the peel test). In all, both the qualitative evidence (large plastic deformation of polyurethane 1 lm and cohesive failure during the peel off) and quantitative data (peel strength) suggest good adhesive properties of polyurethane 1 to bind skin substrates. The healing capability of the adhered polyurethane 1 on the pig skin surface was also investigated. A sample was cut in the centre gently, transverse to its long axis, with a razor and positioned with the cut edges in close contact on the surface of pig skin. The sample was then placed in the oven at 37 C for two hours. The images before healing and aer healing for two hours were captured by high-resolution digital camera and optical surface prolometry, and the corresponding results are shown in Fig. 8. It was observed that a clear cut existed before healing, which disappeared completely aer two hours healing at the temperature of 37 C, as indicated by the surface roughness prole which shows that the roughness around the cut area is both qualitatively (i.e. visibly) and quantitatively comparable to the other areas in the surface of the sample, indicating a fully topological recovery of the cut interface. Therefore, the excellent healing capability of polyurethane 1 was maintained even when attached on the surface of pig skin.
Creep recovery experiments were performed in the same rheometer as described above to further characterise the viscoelastic response of the materials (see Fig. 9). It is observed that the creep behaviour of polyurethane 1 shows linear dependence on the stress level, but is very sensitive to temperature. For example, at 10 C, 0.17% deformation is observed at a load of 200 Pa aer about 1 hour, and 38% of this deformation can be recovered aer 1 hour. This indicates good elasticity recovery of polyurethane 1, and is expected to be due to the strong non-covalent interaction between polymer chains from the hydrogen bonds. However, at a temperature of 20 C, the deformation increases signicantly to 1.43%, and only 5% of this deformation is recovered aer 1 hour, which is consistent with the disruption of the hydrogen bonds at elevated temperatures; although this is not observed in the 5 Hz rheometer data (Fig. 2) until higher temperatures, it does have a signicant effect on the creep and recovery behaviour on these longer timescales. The behaviour of polyurethane 1 is similar to another polyurethane we reported recently, 74 but with a lower disruption temperature for the secondary interaction, which is consistent with the lower healing temperature (37 C) of this material compared with 45 C for that reported before, and also consistent with the observed recovery data for the two materials. Further creep recovery experiments were performed at 37 C, the recovery at larger stresses was minimal, although at 10 Pa it was about 20%. It is anticipated that for biological applications the creep and recovery behaviour would be improved through the production of composite materials with suitable llers.
Whenever a new material is suggested for therapeutic purposes, toxicity assessment is important to ensure that it is safe for use. Cytotoxicity studies were carried out on the human skin broblasts, 161BR cells by MTT assay. Polyurethane 1 was found to be non-toxic (cell viability aer exposure to liquid extracts from the polymer >94% at all concentrations, and non-signicantly different from the negative control, see Fig. 10).

Conclusions
A well-dened supramolecular polyurethane capable of selfassembling via hydrogen bonding interactions has been synthesised. The material presents rheological behaviour characteristic of a supramolecular polymer, but with a low dissociation temperature for the network, which permits healing at 37 C. Results show that aer 60 minutes at body temperature, the material can fully recover its mechanical performance. In addition, the investigation of the mechanical performance under physiological conditions shows that the material can maintain its structural integrity. In addition, when adhered to pig skin, the healable properties of polyurethane 1 were fully conserved suggesting this material could be used for biomedical applications such as articial skin or adhesives for plastic surgery. Fig. 10 Cytotoxicity profile of liquid extracts from polyurethane at different concentrations (from 100%: PU100 to 25%: PU25). Polyethylene (PE) and polyurethane (PU) containing 0.1% (w/w) zinc diethyldithiocarbamate (ZDEC) were used as negative and positive controls, respectively. Data indicate average AE SEM, n ¼ 3. Statistical significance with respect to untreated sample (medium) was determined by ANOVA followed by Bonferroni post hoc test and is indicated in the figure (* ¼ P < 0.05; ns ¼ non-significant).