Mechanically robust, readily repairable polymers via tailored noncovalent cross-linking

See allHide authors and affiliations

Science  05 Jan 2018:
Vol. 359, Issue 6371, pp. 72-76
DOI: 10.1126/science.aam7588

A healing squeeze

The very long molecules found in synthetic polymers, and their tendency to entangle and partially crystallize, impart many of the polymers' useful properties. However, these same characteristics also mean that chain dynamics are slow, which impedes potential self-healing. Yanagisawa et al. developed a family of ether-thiourea linear polymers that form hydrogen-bonded networks and still manage to stay amorphous. The polymers are stiff, showing the strength of the hydrogen bonding; however, because these bonds can easily reform, the polymer is also able to self-heal when compressed.

Science, this issue p. 72


Expanding the range of healable materials is an important challenge for sustainable societies. Noncrystalline, high-molecular-weight polymers generally form mechanically robust materials, which, however, are difficult to repair once they are fractured. This is because their polymer chains are heavily entangled and diffuse too sluggishly to unite fractured surfaces within reasonable time scales. Here we report that low-molecular-weight polymers, when cross-linked by dense hydrogen bonds, yield mechanically robust yet readily repairable materials, despite their extremely slow diffusion dynamics. A key was to use thiourea, which anomalously forms a zigzag hydrogen-bonded array that does not induce unfavorable crystallization. Another key was to incorporate a structural element for activating the exchange of hydrogen-bonded pairs, which enables the fractured portions to rejoin readily upon compression.

Healable polymers with enhanced longevity and reliability are particularly attractive as next-generation materials for the realization of a sustainable society (13). On the basis of their healing mechanisms, such materials are classified as exhibiting extrinsic (1, 2) or intrinsic (1, 3) healing behavior. Extrinsic healing depends on the presence of finely dispersed small capsules or vascular networks that separately entrap monomers and initiators that are mixed when the containers are broken, resulting in the formation of new polymer chains that connect the fractured parts (4, 5). This approach is reliable, but it allows the material to heal only a limited number of times. In contrast, intrinsic mechanisms enable polymeric materials to heal many times, in some cases at ambient temperatures, through the reorganization of dynamic covalent bonds (610) or through noncovalent interactions, mostly involving hydrogen bonds (H bonds) (1115). These healable materials are usually soft and deformable. Some healable materials with high mechanical robustness have also been developed by cross-linking with dynamic covalent bonds. However, in most cases, heating to high temperatures (on the order of 120°C or more) to reorganize their cross-linked networks is necessary for the fractured portions to repair (68).

In general, noncrystalline, high-molecular-weight polymers form mechanically robust materials owing to the entanglement of long polymer chains. However, once the materials are fractured, they are difficult to repair unless they are heated to melt, because the entangled polymer chains diffuse too sluggishly to unite fractured portions within reasonable time scales (2). On the other hand, low-molecular-weight polymers, when they are noncovalently cross-linked by H bonds, may be elaborated into mechanically robust yet healable polymeric materials, because their dynamic properties are modulable by changing the cross-linking density. So far, rubber-like soft materials (11) and thermoplastic elastomers (14) with H-bonding motifs have been designed to heal upon gentle compression. However, the presence of a large number of H bonds often leads to crystallization or clustering of polymeric materials, thereby making them brittle (1618). In other words, high mechanical robustness and healing ability tend to be mutually exclusive.

Here we report that poly(ether-thioureas) such as TUEG2 and TUEG3 (Fig. 1A) anomalously form amorphous materials, despite carrying dense H-bonding thiourea units. These materials are highly mechanically robust yet can be readily repaired by compression at fractured surfaces. Why are they amorphous despite carrying a large number of H-bonded thiourea units? We found that their H-bonded thiourea arrays are geometrically nonlinear (less ordered), so that they do not induce crystallization. Originally, we prepared TUEG3 as a synthetic intermediate for guanidinium ion–appended polymers that strongly bind to oxyanionic groups in biomacromolecules (1921). During the isolation of TUEG3, we noticed that cut surfaces of this polymer are self-adhesive, despite their rigid and nontacky nature. Cut pieces of a 2-mm-thick rectangular sheet of TUEG3 (10 mm by 20 mm) formed a merged sheet that withstood a 300-g load when their fractured surfaces were manually compressed for 30 s at an ambient temperature of 21°C (fig. S1).

Fig. 1 Molecular structures and characterization of poly(ether-thiourea) TUEG3 and its reference polymers.

(A) Schematic structures of poly(ether-thioureas) with diethylene glycol (TUEG2), triethylene glycol (TUEG3), and tetraethylene glycol (TUEG4) as spacers; schematic structures of poly(alkylene-thioureas) with octamethylene (TUC8) and dodecamethylene (TUC12) chains as spacers; and schematic structure of a poly(ether-urea) with triethylene glycol (UEG3) as a spacer. (B) Differential scanning calorimetry thermograms for TUEG2, TUEG3, TUEG4, TUC8, TUC12, and UEG3 on the second heating from –20°C to 200°C at a rate of 10°C/min. Tg, Tc, and Tm denote the glass-transition temperature, crystallization temperature, and melting temperature, respectively. (C) Characterization of polymers. Mn is the number-average molecular weight, DP is the degree of polymerization as estimated by 1H NMR end-group analysis, Mw is the weight-average molecular weight obtained by multiangle light scattering, PDI is the polydispersity index (Mw/Mn), and E is the elastic modulus evaluated from the stress-strain curve in tensile testing (D). (D) Stress-strain curves of TUEG2, TUEG3, TUC8, TUC12, and UEG3 at 21°C.

TUEG3 (Fig. 1A) can be prepared by a one-pot polycondensation of commercially available monomers (supplementary materials, p. S4). It was characterized by means of 1H and 13C nuclear magnetic resonance (NMR) spectroscopy and elemental analysis. 1H NMR end-group analysis and multiangle light-scattering studies showed that its number- and weight-average molecular weights (Mn and Mw) were 9500 and 22,300, respectively (Fig. 1C). TUEG3 is amorphous, as shown by its differential scanning calorimetry profile on the second heating (Fig. 1B). No sharp peaks owing to crystallization appeared in the temperature range from –20° to 200°C at any scan rate from 1° to 10°C/min, whereas a broad peak corresponding to a glass transition (Tg) appeared at 27°C (fig. S8B). Furthermore, in its x-ray diffraction profile recorded at 22°C, TUEG3 exhibited a broad diffraction peak centered at 2θ = 23°, typical of an amorphous material (fig. S10B). The thiourea group is special as an H-bonding motif (22), because UEG3, the urea analog of TUEG3, is semicrystalline (Fig. 1, A and B, and fig. S10E).

According to a thorough crystallographic study of the geometries of H-bonded thiourea and urea derivatives (23), the former mostly form nonlinear zigzag arrays of H-bonded thiourea units adopting cis/trans and strained trans/trans conformations (I in Fig. 2A), whereas the latter form linear arrays of H-bonded urea units adopting only a trans/trans conformation (II in Fig. 2A). We consider that the amorphous nature of TUEG3 originates from such less ordered, nonlinear geometries of its H-bonded thiourea units. Fourier transform infrared (FTIR) spectroscopy of TUEG3 at 22°C showed broad vibrational bands at around 3290 and 3060 cm–1 (Fig. 2B), the former of which can be assigned to the NH stretching vibration, whereas the latter is characteristic of an NH deformation vibration of nonlinearly H-bonded thiourea units (24). Because the estimated energy gap between the cis/trans and trans/trans conformations is 2.9 to 4.2 kJ/mol (25), these two conformers likely coexist in the polymer matrix of TUEG3. Semicrystalline UEG3, the urea analog of TUEG3, did not display an IR absorption at around 3060 cm–1 (Fig. 2B). This is in accord with a previous report that H-bonded urea units adopt only a trans/trans conformation (Fig. 2, A and B) (23). The contrasting FTIR spectral features observed for amorphous TUEG3 and semicrystalline UEG3 highlight an important principle—namely, that densely located thiourea units, like urea units, tightly cross-link the polymer main chains through H-bonding interactions but do not induce crystallization because the resultant H-bonded arrays are nonlinear and less ordered.

Fig. 2 Hydrogen-bonding interactions of thiourea and urea moieties in polymer matrices.

(A) Schematic representations of the H-bonding modes of thiourea and urea. (B) FTIR spectra (bulk) at 22°C of poly(ether-thioureas) TUEG2, TUEG3, and TUEG4; poly(alkylene-thioureas) TUC8 and TUC12; and poly(ether-urea) UEG3.

To follow up on our preliminary healing test of TUEG3 (fig. S1), we fabricated tensile bars by a method analogous to that used for the preparation of the polymer sheet, then subjected the bars to tensile tests (test rate, 10 mm/min) at an ambient temperature of 21°C. The stress-strain curve (red solid curve, Fig. 1D) displayed an abrupt increase in stress of as much as 45 ± 8 MPa, and the sample yielded at an applied strain of 6 ± 2%. After this yielding stage, the specimen elongated by up to 393 ± 5% before fracture (fig. S11). The elastic modulus of TUEG3, as estimated from the initial slope of the stress-strain curve (Fig. 1D), was 1.4 GPa. Because dynamic mechanical properties are important for understanding the healing events at fractured portions, we performed temperature-dispersion tests for TUEG3 at 200° to 0°C (5°C/min) at a constant frequency (ω) of 10 rad/s. The storage modulus (G′) and loss modulus (G′′) curves of TUEG3 allowed for confirmation of the glass transition at around 30°C (fig. S12B). TUEG3 exhibited a rubber plateau region, typical of long polymer chains upon being entangled, at around 50° to 70°C, despite the fact that its molecular mass is less than 10 kDa.

We then tested the healing of TUEG3 by a method analogous to that reported by Corté and co-workers (26). A 0.1-mm-thick sheet of Teflon (10 mm by 10 mm) with a 1-mm-diameter hole at its center (Fig. 3A) was sandwiched between two discs of TUEG3 (diameter, 8 mm; thickness, 1.0 mm). This assembled specimen was set on a rheometer stage, gently pressed with a jig, then annealed for 20 min at 90°C (hot melt), so that the touching parts of the upper and lower discs merged inside the hole area of the Teflon sheet. The resultant specimen was cooled to 24°C at a rate of 5°C/min, then the stage and jig in the rheometer were rapidly separated from one another at a rate of 100 mm/min to cause brittle fracture in the merged part of the specimen. After 3 min, the fractured surfaces were mechanically brought back into contact and compressed at the designated temperature under a constant applied stress. The compressed specimen was then subjected to tensile testing (tensile rate, 2 mm/min) to determine the stress at fracture.

Fig. 3 Healing behaviors of polymeric thiourea derivatives upon compression.

Healing tests using a rheometer (26) are illustrated in (A) to (D). (A) Schematic representation of the assembled specimen. (B and C) Elastic moduli (E, red circles) of poly(ether-thioureas) (B) TUEG3 and (C) TUEG2 at various temperatures, as estimated by compression testing (10 mm/min), and stresses at break for the assembled specimens (black circles) prepared by compression for 1 hour at various temperatures under a constant applied stress of 1.0 MPa. (D) Percentage recoveries of the stress at break for assembled specimens of TUEG2, TUEG3, and TUC8pristine = 26.5 ± 0.3, 31.7 ± 0.5, and 39.1 ± 1.6 MPa, respectively) prepared by compression for various periods of time at Tcomp = 58°, 24°, and 44°C, respectively, under a constant applied stress of 1.0 MPa. (E) Master curves for TUEG3 at a reference temperature of 24°C, prepared by the time-temperature superposition treatment of storage modulus (G′) and loss modulus (G′′) values obtained at 24°, 40°, 60°, and 80°C in frequency-dispersion tests [frequency (ω) = 628 to (6.28 × 10−3) rad/s] at an applied strain of 0.05 or 0.1%. (F) Relaxation times (τ) of TUEG3 for the flow transition, estimated from the crossover points of G′ and G′′ at various temperatures (20° to 36°C) (fig. S21).

Although the contact area between the upper and lower discs was only 0.79 mm2, the tensile tests were highly reproducible (fig. S13). Even when the contact area was enlarged from 0.79 to 1.76, 3.14, or 19.6 mm2 (figs. S14 and S15), the results obtained were consistent (fig. S16). When the applied compression stress was increased from 0.12 to 1.0 MPa, the recovery leveled off at 0.4 MPa after the initial abrupt increase (fig. S17). On the basis of these results, we carried out systematic healing tests by setting the contact area and applying compression stress at 0.79 mm2 and 1.0 MPa, respectively. Figure 3B shows the stresses at break (black circles) observed for the assembled specimens of TUEG3 prepared by compression for 1 hour at various temperatures (Tcomp = 10° to 36°C). The elastic moduli (red circles) of the compressed specimens at ≤32°C were high (≥0.9 GPa). Figure 3D shows the percentage recoveries of the stresses at break (σ/σpristine × 100) for TUEG3; the specimen compressed at 24°C completely recovered its original mechanical strength (σpristine) in 6 hours. Accordingly, the interfacial adhesion energy was estimated to be ~0.9 J/m2 (supplementary materials, p. S4). When Tcomp was raised from 24°C to the higher range of 28° to 36°C, the specimen completely recovered within 1 hour (Fig. 3B), and the healing could be repeated multiple times without noticeable deterioration (fig. S18). Even at Tcomp = 12°C, the specimen recovered partially.

On the basis of the frequency-dispersion properties of the G" and G" values at 20° to 100°C with ω = 628 to (6.28 × 10−3) rad/s (figs. S19 and S20), we estimated the relaxation times (τ) for the flow transition of TUEG3 by using their master curves (27) at 24° to 32°C (Fig. 3E and fig. S21). The τ values were in the range of 107 s (on the order of months) to 105 s (on the order of days) (Fig. 3F), far longer than the compression times required for TUEG3 to heal completely at Tcomp = 24° to 32°C (≤6 hours). This contrast suggests that the rapid healing behavior of TUEG3 is not a consequence of the flow of polymer chains (11, 14) but is dominated by a much more rapid process.

Other polymeric thiourea derivatives including TUEG2, TUEG4, TUC8, and TUC12 (Fig. 1A)—in which diethylene glycol, tetraethylene glycol, octamethylene, and dodecamethylene spacer units, respectively, connect the thiourea units—were synthesized. Like TUEG3, TUEG2 and TUEG4 were both amorphous and exhibited only single, broad glass-transition peaks at 58° and 5°C, respectively (Fig. 1B and figs. S8 and S10). TUC8 was also amorphous (Tg = 39°C), whereas TUC12 was semicrystalline (Fig. 1B and figs. S8 and S10). As in the case of TUEG3, the FTIR spectra (Fig. 2B) of these reference polymers exhibited NH stretching and deformation vibrations between 3250 and 3300 cm–1 and at around 3,060 cm–1, respectively, the latter of which is typical of nonlinear H-bonded arrays of thiourea units (23).

We investigated the tensile behaviors of these polymers and of UEG3 at an ambient temperature of 21°C with a tensile rate of 10 mm/min (Fig. 1D). In contrast to TUEG3, all of these polymers were highly brittle, except TUEG4, which was too fluid-like to be tested. However, just below their glass-transition temperatures, TUEG2 did not fracture in a brittle manner but yielded like TUEG3, whereas TUC8 again showed brittle fracture (fig. S23). We then conducted healing tests on amorphous TUEG2 and TUC8 at Tcomp = 58° and 44°C, respectively. These temperatures were used to compare the healing behaviors of these polymers with those of TUEG3 at 24°C (Fig. 3D), because their viscosities, as determined by shear creep tests, were comparable to one another (2.5 × 109 Pa·s for TUEG2 at 58°C, 2.2 × 109 Pa·s for TUC8 at 44°C, and 3.3 × 109 Pa·s for TUEG3 at 24°C) (fig. S25). Figure 3D shows that TUEG2 at Tcomp = 58°C exhibited an 85% recovery of its mechanical strength upon compression for 6 hours. Even when Tcomp was lowered to 48°C (Fig. 3C, black circles), compressed TUEG2 showed a partial recovery (22%) in 1 hour, although the elastic modulus of TUEG2 at temperatures up to 70°C remained high (≥1 GPa) (Fig. 3C, red circles). In contrast, TUC8 barely healed, even upon compression at 44°C over a long period of 24 hours.

Analogously to the case of TUEG3, we evaluated the relaxation times of TUEG2 and TUC8 for the flow transition (fig. S22) from their temperature-dispersion (fig. S12) and frequency-dispersion properties (figs. S19 and S20). As shown in fig. S22, TUEG2 and TUC8 exhibited a rubber plateau region similar to that of TUEG3. Furthermore, both of these polymers showed extremely slow relaxation properties at Tcomp [τ = 107 s (on the order of months) to 106 s (on the order of weeks)].

Why can TUEG2 as well as TUEG3 readily heal despite such slow diffusion dynamics? We consider that their healing properties are dominated by a segmental motion such as the exchange of H-bonded thiourea pairs, leading to the interpenetration of polymer chains at the fractured portions upon compression (28). The H-bond exchange of thiourea pairs should occur far more rapidly than the flow relaxation of these polymers (Fig. 3E and figs. S21 and S22). Consequently, virtually frozen TUEG2 and TUEG3 can heal only in a few hours. Then why does TUC8 hardly heal? This question prompted us to consider a special role of the spacer units that connect the thiourea units. Arrhenius plots of the intersection frequencies for the G′ and G′′ curves (fig. S20) allowed us to estimate the apparent activation energies (Ea) for the slippage of polymer chains (Fig. 4A). As shown in Fig. 4B, the Ea values for TUEG2, TUEG3, and TUC8 were 171, 149, and 203 kJ/mol, respectively. In H-bonded polymeric materials, polymer chains presumably slip and interpenetrate by exchanging their H-bonded pairs. According to the Flory-Huggins interaction parameters (χ) of the model structures in Fig. 4C, thiourea is much more miscible with ethers than with hydrocarbons. In ethereal media, the ether oxygen atoms might facilitate the exchange of H-bonded thiourea pairs by serving as temporal H-bond acceptors (Fig. 4D); in other words, the ether-containing TUEG2 and TUEG3 use this mechanism to lower the energy barrier for the slippage of their polymer chains. This is not the case in TUC8, because the thiourea group is incompatible with hydrocarbons (Fig. 4C). It is therefore reasonable that the Ea value for TUC8 is larger than those for ether-containing TUEG2 and TUEG3, and the same might hold true even in a lower temperature range.

Fig. 4 Slip motions of polymer chains in poly(ether-thiourea) and poly(alkylene-thiourea) through exchange of H-bonded pairs.

(A) Arrhenius plots of the intersection frequencies (f = ω/2π) for the G′ and G′′ curves of poly(ether-thioureas) TUEG2 and TUEG3 and poly(alkylene-thiourea) TUC8, estimated from frequency sweep tests [ω = 100 to 0.05 rad/s] at a constant applied strain of 0.1% in the temperature range for the flow transition (fig. S20). (B) Apparent activation energies (Ea) of TUEG2, TUEG3, and TUC8 for the slippage of polymer chains, estimated from the Arrhenius plots. (C) Flory-Huggins interaction parameters χ (DMTU-spacer) for N,N"-dimethylthiourea (DMTU) in combination with dimethyl ether (a model mixture for TUEG2), ethylene glycol dimethyl ether (a model mixture for TUEG3), or hexane (a model mixture for TUC8). These values were estimated from the corresponding Hildebrand solubility parameters, determined by the Fedors method (29, 30). (D) Proposed mechanism of how the exchange of H-bonded thiourea pairs in TUEG3 is enhanced.

This work provides the essential structural elements for the design of mechanically robust yet healable polymeric materials: (i) relatively short polymer chains that permit greater segmental motions, (ii) tight cross-links by a large number of H bonds for better mechanical properties, (iii) nonlinear (less ordered) H-bond arrays that do not induce crystallization, and (iv) implemented mechanisms to facilitate the exchange of H-bonded pairs. The triethylene glycol spacer that connects the thiourea units in TUEG3 optimally modulates the activation energy for the exchange of H-bonded thiourea pairs; consequently, the polymeric material, although highly cross-linked noncovalently, can heal by compression without heating.

Supplementary Materials

Materials and Methods

Figs. S1 to S25

Reference (31)

Movie S1

References and Notes

Acknowledgments: We appreciate valuable discussions with Z. Guan (University of California, Irvine), K. Mayumi and K. Ito (The University of Tokyo), H. Sasaki (Toagosei), and E. Silver and K. Morishita (The University of Tokyo). We also thank H. Ejima and N. Yoshie (The University of Tokyo) and Y. Kamei, H. Ochiai, and A. Masumoto (Shimadzu Corporation) for performing the mechanical tests, and we thank M. Ishikawa (Tokyo Institute of Technology) and M. Nakamura, K. Kurono, and T. Tokai (Shoko Science) for the multiangle light-scattering analysis. This work was financially supported by a Grant-in-Aid for Specially Promoted Research (25000005, “Physically perturbed assembly for tailoring high-performance soft materials with controlled macroscopic structural anisotropy.”) We also acknowledge assistance from the ImPACT Program of the Council for Science, Technology and Innovation (Cabinet Office, Government of Japan). Y.Y. acknowledges support from the Japan Society for the Promotion of Science’s Research Fellowship for Young Scientists and the Program for Leading Graduate Schools (MERIT).

Stay Connected to Science

Navigate This Article