Key-and-lock commodity self-healing copolymers

See allHide authors and affiliations

Science  12 Oct 2018:
Vol. 362, Issue 6411, pp. 220-225
DOI: 10.1126/science.aat2975

Simple routes to self-healing

Biology provides many routes for self-healing or repair, but this trait is hard to endow into engineering materials. Although self-repair has been demonstrated for some polymers, it usually required specialized monomers. Urban et al. demonstrate that for a very narrow range of compositions, simple vinyl polymers based on methyl methacrylate and n-butyl acrylate show repeatable self-healing properties (see the Perspective by Sumerlin). A key characteristic of this system is that it relies on van der Waals interactions rather than the reformation of hydrogen or covalent bonds for repair.

Science, this issue p. 220; see also p. 150


Self-healing materials are notable for their ability to recover from physical or chemical damage. We report that commodity copolymers, such as poly(methyl methacrylate)/n-butyl acrylate [p(MMA/nBA)] and their derivatives, can self-heal upon mechanical damage. This behavior occurs in a narrow compositional range for copolymer topologies that are preferentially alternating with a random component (alternating/random) and is attributed to favorable interchain van der Waals forces forming key-and-lock interchain junctions. The use of van der Waals forces instead of supramolecular or covalent rebonding or encapsulated reactants eliminates chemical and physical alterations and enables multiple recovery upon mechanical damage without external intervention. Unlike other self-healing approaches, perturbation of ubiquitous van der Waals forces upon mechanical damage is energetically unfavorable for interdigitated alternating/random copolymer motifs that facilitate self-healing under ambient conditions.

Advances in the last two decades in materials capable of self-healing focused primarily on incorporating physical and chemical mechanisms into polymer networks. These mechanisms can be conveniently classified into the following categories: embedding reactive encapsulated fluids that burst open upon damage to fill and repair damaged areas (1); incorporating covalent (26) or supramolecular (712) dynamic bonds that, upon cleavage, reform polymer networks; physically dispersing nanomaterials that enable repair in response to magnetic or electromagnetic fields (13, 14); introducing phase-separated morphologies that facilitate damage closure (11, 15); and incorporating living organisms capable of remending damaged structures (16). Polymers—and in particular copolymers, if designed properly—can encode molecular features by placement of repeating units that interact with each other (17). However, for self-repair to occur, synchronized chemical and physical events (18, 19), potentially driven by van der Waals (vdW) interactions, must take place.

We synthesized a series of copolymers using atom transfer radical polymerization (ATRP), statistical free radical polymerization, and colloidal polymerization. The methyl methacrylate/n-butyl acrylate (MMA/nBA) molar ratios were varied from 30/70 to 70/30, while maintaining similar molecular weights for all compositions for each synthesis method (ATPR: ~25 kD; statistical: ~60 kD; colloidal: ~700 kD). Copolymer synthesis and properties are summarized in table S1. Figure 1A illustrates selected optical images of p(MMA/nBA) copolymer films produced by ATRP in the 40/60 to 55/45 compositional range that were damaged (0 hours) and allowed to self-heal (~14 hours). Self-repair occurs without external intervention only within narrow 45/55 to 50/50 MMA/nBA compositional ranges (movie S1) Outside this range, self-repair does not take place even days after damage, even though the glass transition temperature (Tg) for nBA-richer 40/60 copolymers is below ambient conditions (25°C, relative humidity = 50%). For undamaged copolymer films, when MMA/nBA molar ratios increase, Young’s moduli (E) also increase (Fig. 1B). However, ~14 hours after damage, only 45/55 to 50/50 p(MMA/nBA) copolymer compositions recover 90 to 100% (±5%) of their original tensile strains, respectively (Fig. 1B, B5 and C5). The 45/55 self-healing copolymer exhibits moderate toughness with tensile strain of ~550% and stress values of ~8.6 MPa after self-healing (~600% and 10 MPa before damage). By contrast, the copolymer films outside this range exhibit ~55 and 10% recovery (Fig. 1B, A5 and D5), respectively. Similar behavior, although with longer self-healing times (~86 hours), are observed for copolymers produced by colloidal radical polymerization (fig. S1 and table S1-B).

Fig. 1 Self-healing of copolymer films and their mechanical analysis.

(A) Optical images of damaged p(MMA/nBA) copolymers with the following MMA/nBA molar ratios: 40/60 (A1 to A4), 45/55 (B1 to B4), 50/50 (C1 to C4), and 55/45 (D1 to D4). The copolymers were allowed to repair under ambient conditions. A video of the self-healing process is shown in movie S1. (B) A5 to D5: The corresponding stress-strain curves before damage and 14 hours after repair for each copolymer composition in (A).

It is reasonable to hypothesize that for copolymers with 45/55 to 50/50 MMA/nBA molar ratios, the neighboring MMA and nBA copolymer units and their distribution may play some role in self-healing as these compositions are expected to form random and/or alternating chain topologies. To test this hypothesis, MMA and nBA monomers were copolymerized to obtain number average molecular weight Mn = ~20- to 30-kDa pMMA-b-pnBA block copolymers with controlled bock sizes and the number of blocks ranging from two to six (tables S2 and S3). These block copolymers do not exhibit self-healing under the same conditions.

To experimentally assess molecular events associated with self-healing or lack thereof, we used internal reflection infrared imaging (IRIRI), proton nuclear magnetic resonance (1H NMR), and electron spin resonance (ESR), along with stress-strain and dynamic mechanical analysis (DMA). The results of these experiments show that reversible spectroscopic changes are only observed for self-healable copolymer compositions. In IR analysis (figs. S2 and S3), they are manifested by the intensity changes of the C=O (1728 cm−1) and C-O-C (1158 cm−1) normal vibrations due to conformational changes of MMA and nBA repeating units (20). In 1H NMR, the key features are the changes in the methyl group shielding-deshielding during the damage-repair cycle for self-healing copolymer compositions (figs. S4 to S6 and tables S4 and S5) (21, 22). Upon mechanical damage, the resonances at 0.98 parts per million (ppm) (a) and 0.96 ppm (b) increase (deshielded) at the expense of diminishing 0.93-ppm (c) and 0.90-ppm (d) peaks (shielded), suggesting a closer chain packing. If strong vdW forces contribute to interchain cohesiveness, mechanical damage will alter the distribution of shielded and deshielded methyl groups along the polymer backbone. Because mechanical damage may also lead to the formation of free radicals, ESR analysis of damaged copolymers showed that, regardless of the copolymer composition, the concentrations of free radicals are in the 4.5 to 8 × 10−7 mol/liter range (fig. S7) and appear to have no relation to self-healing. Junction densities (νj) due to chain entanglements or adjacent chain interactions were obtained from the measurements of viscoelastic length transitions (VLTs) in dynamic mechanical analysis (DMA) as a function of copolymer composition (fig. S8 and table S6) (23). For MMA and nBA homopolymers, the νj values are 93.1 and 60 mol/m3, respectively, but an increase up to 123.6 mol/m3 is observed for self-healable compositions.

Molecular dynamics (MD) simulations were employed under isothermal (NVT) and isoenergetic equilibration (NVE) conditions as a function of copolymer composition to determine copolymer conformations, end-to-end distances (r), and cohesive energy densities (CEDs). These results are plotted in Fig. 2A (table S7A), and further experimental details along with the results of MD simulations are provided in the supplementary materials. Figure 2A shows that the equilibrium cohesive energy densities (CEDeq) (curve a), as well as the end-to-end chain distances (req) (curve a’) both reach maxima for self-healable 45/55 to 50/50 MMA/nBA compositions (range II). Copolymer interchain packing (Fig. 2B) is greater within self-healing compositional range II, whereas non–self-healable ranges I and III exhibit less interwinding chains. Further, representative copolymer chains extracted from each range shown in Fig. 2C indicate that, within self-healing range II, the chains exhibit extended helix-like conformations with average req values of ~34 Å, whereas within ranges I and III, globular shapes with req values in the ~25 to 29 Å range are observed. The results of MD simulations are summarized in Table 1 and show that the CED values upon reaching equilibrium (CEDeq) increase for self-healing compositions (range II). Within range II, the vdWeq density values reach 1.96 × 105 kJ/m3, thus indicating that the extended-chain helix-like conformations are energetically preferable. It is also useful to examine chain conformations equilibrated in the absence of interchain vdW interactions for all compositions. The results of MD simulations for single isolated 30/70, 45/55, and 70/30 p(MMA/nBA) chains shown in Fig. 2D illustrate that, regardless of the copolymer composition, globular conformations, similar to non–self-healable compositions (ranges I and III) are preferable, and the single-chain end-to-end distances (req) are within the 21.7 to 27.8 Å range. Thus, without interchain vdW interactions, globular chain conformations prevail regardless of the copolymer topology.

Fig. 2 The results of MD simulations as a function of copolymer composition.

(A) Cohesive energy densities at equilibrium (CEDeq) (curve a), end-to-end equilibrium distances (req) (curve a’), cohesive energy densities (CEDhl) of forced helical conformations (curve b), and end-to-end chain distances for forced helical conformations (rhl) (curve b’) as a function of molar % of MMA in p(MMA/nBA) copolymers. (B) Representative examples of copolymer morphologies in range I (MMA/nBA molar ratio: 30/70), range II (MMA/nBA molar ratio: 45/55), and range III (MMA/nBA molar ratio: 70/30); circles denote examples of non-interwinding chains. (C) Average end-to-end distances for macromolecular chains extracted from MD simulations in (B). (D) Average end-to-end distances for single isolated chains (reqs) in range I (MMA/nBA molar ratio: 30/70), range II (MMA/nBA molar ratio: 45/55), and range III (MMA/nBA molar ratio: 70/30). The req and reqs values were measured from 3D chain images and may appear not to scale.

Table 1 Cohesive energy density of equilibrated (CEDeq) and forced helix-like (CEDhl) p(MMA/nBA) copolymer conformations, van der Waals (vdW) density, end-to-end distance (req), flexibility parameter (feq), and enthalpy changes (ΔHeq) as a function of MMA/nBA molar ratios [italics indicate self-healing (range II of Fig. 2) copolymer compositions].

View this table:

In all MD simulations, an experimental average copolymer density of 1.125 g/cm3 was used. In separate simulations conducted under the same conditions, copolymer chains were allowed to have excess free volume by assuming an initial density of 0.5 g/cm3, thus enabling chain motion in and out of the physical cell boundaries upon reaching an equilibrium. The premise behind these simulations was to examine the role, if any, of vdW interactions as a function of copolymer composition in their ability to assume higher- or lower-density states. With an initial density of 0.50 g/cm3,  respective copolymer chains were isothermally equilibrated. Only for self-healing compositions (range II) did the density increase to the 0.529- to 0.562-g/cm3 range, whereas for non–self-healing compositions, the density decreased (fig. S9), supporting the hypothesis that enhanced vdW forces facilitate favorable interchain interactions and return to denser packing upon physical separation. The question then arises from these experimental and modeling exercises: What are the molecular entities within this narrow compositional range that lead to stronger interchain interactions and subsequent self-healing?

To determine the role of the monomer sequences and the vdW contributions to self-healing, we examined vdW forces and cohesive energies (CEp) for model pentads containing selected sequences of M and B monomer units (where M and B represent MMA and nBA monomers, respectively). Under NVT MD conditions, selected pentads were placed into one cell and equilibrated. Figure 3A illustrates BMBMB/BMBMB, BMBMB/BMBBM, BMBMB/BMMBB, and BMBMB/MMBBB pentad pairs and the CEp values due to their interactions. The highest CEp value (313.6 kJ/mol) exhibits an alternating BMBMB/BMBMB pair (1:1). By contrast, more ‘blocky-type’ MMBBB/MMBBB pentads (Fig. 3B, pair 4-4) have the lowest CEp value (258.2 kJ/mol). Similarly, other “blocky-type” combinations (Fig. 3B) also exhibit lower CEp values, thus indicating that the alternating BMBMB-type monomer sequences of the neighboring chains favor overall higher CEp values. Notably, for alternating BMBMB-type segments composed of MMA (M) and nBA (B) units, there is an average ~120 Å3 space (~7.1 Å by 4.2 Å by 4.0 Å) between two neighboring nBA monomers separated by one MMA unit along one chain, thus being spatially capable and energetically favorable for hosting an nBA unit of an adjacent chain and thereby enabling the key-and-lock interactions stabilized by vdW forces.

Helix-like chain conformations may also contribute to the high CEDeq values within self-healable compositions (range II; Fig. 2). To examine this hypothesis, we analyzed cohesive energy densities for fixed helix-like conformations (CEDhl) as a function of copolymer composition. All copolymers across the compositional range were forced to retain a 34.0 ± 0.2 Å end-to-end distance (Fig. 2C”) and extended helix-like chain conformations of the 45/55 self-healable copolymer. The results are summarized in Table 1, and CEDhl and end-to-end distance values are plotted in Fig. 2A (curves b and b’, respectively). As shown, regardless of p(MMA/nBA) composition, the CEDhl values are higher compared to their corresponding CEDeq counterparts, suggesting that the helix-like conformations resulting from alternating monomer sequences are the main contributing factors to higher CEDeq and strong vdW interchain forces. The monomer sequence contributions to self-healing (range II, Fig. 2) are also supported by the most negative ΔHeq values (Table 1). Because pMMA-b-nBA block copolymers do not self-heal and exhibit lower CEDb values (table S7B), these results further substantiate that the presence of alternating/random BMBMB-like sequences favors strong interchain vdW interactions reflected in higher vdW densities that facilitate self-healing (Table 1, italicized rows). Average MMA and nBA reactivity ratios (r1 = 1.75 to 3.15 and r2 = 0.2 to 0.39) indicate that it is unlikely that copolymers in range II will form purely alternating copolymers. However, 1H NMR analysis shows the presence of minute homopolymer blocks manifested by the presence of CH3 protons due to MMA triads (fig. S6 and table S5) for self-healing compositions (range II), but their content is small compared to non–self-healing compositions. MD simulations conducted for average r1 and r2 values (2.61 and 0.36, respectively) showed that the maximum CED values are still reached for self-healing compositions (table S7), the probability of finding alternating topologies are also greater, and chain conformations follow the same trend.

A lack of interfacial fluidity attributed to the elevated Tg at damage on the MMA-rich compositional end (range III), and limited quantities of vdW interactions on the MMA-poor end (range I), inhibit self-healing outside the 50/50 to 45/55 region (range II). Because the increase in the CEDeq values parallels the increasing number of neighboring MMA/nBA units (table S7C), the formation of key-and-lock configurations between adjacent chains will be favorable for alternating/random copolymer topologies within region II, as reflected by higher junction densities. Assuming that chain entanglements (E) and side-by-side (S) chains are the primary contributors to enhanced junction densities (νj) experimentally obtained in DMA measurements (table S6), we extracted both types of interactions from MD simulations and examined the distribution of the induced dipoles due to vdW interactions that contribute to the enhanced νj values. Figure 4A-1 illustrates that within the self-healing range II, δ+- δ–induced dipole interactions dominate the entanglement (MD-E’) and side-by-side (MD-S’) chain interactions. By contrast, Fig. 4A-2 shows extracted copolymer chains just outside the self-healable range (range III) in which randomized orientation of induced dipoles for entangled (MD-E”) and side-by-side (MD-S”) chains dominate. νj values significantly increased for self-healing compositions, clearly supporting MD predictions. Enhanced segmental chain mobility within interfacial regions generated during damage may also aid the self-healing process, owing to lower Tg values near surfaces (24, 25) which can be boosted by collective structural rearrangements at the interfacial regions (26).

Fig. 4 Visual representation of MD simulations and proposed self-healing mechanism.

(A1) Extracted interchain interactions from MD simulations for self-healable compositions (range II) of entangled (MD-E’) and side-by side (MD-S’) chains. (A2) Extracted interchain interactions from MD simulations for non–self-healable compositions (range III; 55/45 MMA/nBA ratio) of entangled (MD-E″) and side-by side (MD-S″) chains. To visually differentiate copolymers, the neighboring chains were colored in orange and blue. The color scale represents relative distributions of induced dipoles (red, high; blue, low). (B) Proposed self-healing mechanism responsible for the restoration of vdW interactions; the presence of key-and-lock associations (red) facilitates chain recovery upon mechanical damage. (C1) Pictorial representation of the distribution of induced dipole moments in self-healable entangled (E’) and side-by side (S’) chains. (C2) Pictorial representation of distribution of induced dipole moments for none self-healable entangled (E’’) and side-by side (S’’) chains.

Further evidence for interchain interactions can be found in determining the flexibility parameter (feq) (27), defined as the fraction of bonds capable of bending out of the collinear direction of previous segments expressed as feq = rmax/ [req2(l (2 − f))], where: rmax is fully extended chain length, req is the end-to-end distance obtained from MD simulations, and l is length of the repeat unit. The feq values as a function of copolymer composition are summarized in Table 1. When chains are in the equilibrium state (feq), the chain flexibility is the smallest for self-healing compositions, indicating that if chains are deformed as a result of external forces, they will store energy and act like mechanical springs capable of returning to the original state. As was shown for pentad model MD simulations (Fig. 3), these interactions are stabilized by BMBMB/BMBMB key-and-lock junctions between neighboring chains, resulting in recovery upon displacement. Similar behavior is observed for methylmetacrylate/n-pentyl acrylate (PA)– and methylmetacrylate/n-hexyl acrylate) (HA)–based pentads (table S8), in which also alternating copolymer compositions favor enhanced CEp. The optical images (fig. S10) of selected copolymer compositions show similar self-healing behavior, and stress-strain curves recorded before and after damage are strong indicators of mechanical property recovery after ~14 hours (fig. S10). Comparison of mechanical properties before damage and after self-healing for selected p(MMA/nBA), p(MMA/nPA), and p(MMA/nHA) copolymers is summarized in tables S10 and S11. To illustrate that vdW interactions can be highly effective in self-healing of thermoplastic materials, we severed and physically reattached ~200-μm-thick 46/54 p(MMA/nBA) film. After reattachment, self-healing occurred within a few minutes, but to regain ~70 to 85% mechanical properties took ~80 hours under ambient conditions. The tensile strength of these materials before damage and after self-healing is in the range of 6 to 9 MPa (fig. S11). Repetitive damage and self-healing by making parallel cuts over the same area does not affect self-healing efficiency (fig. S12).

Fig. 3 Cohesive energies (CEp) for selected pentad pair combinations.

(A) CEp values for (1–1), (1–2), (1–3), (1–4). (B) (4–2), (4–5), (4–2), (4–4) pentad pair interactions (1, BMBMB; 2, BMBBM; 3, BMMBB; 4, MMBBB 5, MBMBM).

The presence of strong vdW interchain forces for predominantly alternating/random copolymer compositions forming helix-like conformations creates a viscoelastic response that energetically favors self-recovery upon chain separation because of key-and-lock associations of neighboring chains (Fig. 4B). In the presence of these interactions, vdW forces stabilize key-and-lock neighboring junctions reflected in the enhanced CEDeq values. When chains are separated as a result of mechanical damage and an external force is removed, copolymer chains return to their initial conformations by restoring helix-like chain conformations in a spring-like manner and reforming key-and-lock junctions manifested by increased CEDeq and req distances for self-healing compositions (range II). Outside self-healing compositions (ranges I and III), irreversible chain dislocations and insufficient interchain vdW forces inhibit complete chain recovery. Thus, the presence of directional vdW forces due to induced dipole interactions enhances CEDeq of entangled or side-by-side chains (Fig. 4C).

For comparison with vdW forces, when supramolecular interactions, such as H-bonding, were employed in self-healing of rubber, the tensile strength ~3.5 MPa at similar elongation levels was reached (7). Although the underlying mechanisms responsible for self-healing using supramolecular and vdW forces are substantially different, they may result in somewhat similar responses. Considering directionality and polarity as commonly accepted differences between H-bonding and vdW interactions, the former facilitates localized bonding directionality because of the orientation of interacting molecular orbitals and high polarity (hydrophilicity). The main feature of vdW interactions is high polarizability (hydrophobicity) with a tendency to form ubiquitous nondirectional contacts between neighboring macromolecular segments. However, in layered systems with large individual atomic planes, individual weak vdW attractive forces in two-dimensional materials (e.g., graphene, others) are directional and become collectively strong. In amorphous polymers, at first approximation, vdW interactions are nondirectional, but the magnitude of vdW forces will strongly depend on the proximity of the neighboring units (28). As extended semihelix macromolecules are in closer proximity to their alternating/random copolymer neighbors, vdW forces will increase because of the preferable bearings of the side groups, resulting in interdigitated key-and-lock interchain morphologies that facilitate self-healing.

Supplementary Materials

Materials and Methods

Figs. S1 to S12

Tables S1 to S11

References (2940)

Movies S1 and S2

References and Notes

Acknowledgments: We thank K. Ivey for technical assistance in GPC, DSC, and DMA measurements. Funding: This work was supported by the National Science Foundation under Award DMR 1744306 and partially by the J.E. Sirrine Foundation Endowment at Clemson University. Author contributions: The experiment was designed by M.W.U., D.D., Y.Y., and L.C. (EPR). Experimental work was conducted by D.D., T.D., and Y.Z. Data analysis was performed by M.W.U., Y.Y, D.D., and L.C. M.W.U. wrote the manuscript. Competing interests: None declared. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the supplementary materials. Patent application no. 62/702,410 was filed 24 July 2018; contact: C. Gesswein, Clemson University Research Foundation (CURF); email: agesswe{at}

Stay Connected to Science

Navigate This Article