With DCR, our HPWPUE can be processed into transparent thin films at a large scale (1.5 m × 1.5 m × 1 mm) using commercially available feedstocks (Fig. 1b). It achieves an impressive maximum stretch ratio of 31.9, a tensile strength of 81.8 MPa, and a toughness of 0.959 GJm⁻³, surpassing the performance of typical commercial TPUE (Fig. 1c; see Supplementary Table 1 for details).
The synthetic route of our HPWPUE is shown in Fig. 1d. We use hexamethylene diisocyanate (HDI) and methylene-bis(4-cyclohexylisocyanate) (HMDI) as isocyanates, poly (adipic acid-hexylene glycol-neopentyl glycol) (PAA-PHG-PNPG) (Number average molecular weight (M) = 2000 g/mol) as polyol, and 2,2-bis(hydroxymethyl) propionic acid (DMPA), ethylenediamine (EDA), and 2-(2-aminoethylamino) ethanol (AEEA) as chain extenders. Nuclear magnetic resonance data can be found in Supplementary Fig. 2. Due to its hydrophilic nature, DMPA enables the stability of HPWPUE in aqueous solutions (Supplementary Fig. 3). Using various measurements and analyses, we confirm the successful synthesis of the product (Supplementary Fig. 4) and establish that HPWPUE is thermally stable up to ~250 °C (Supplementary Fig. 5). The glass transition temperature (T) of HPWPUE is below room temperature (Supplementary Fig. 1b), indicating its elastomer behavior. Furthermore, it exhibits a non-crystalline, microphase-separated structure (Supplementary Fig. 1b and Supplementary Fig. 6). Stress-strain cycle measurements at various delay times (Supplementary Fig. 7) following an initial cycle confirm the presence of reversible inter-segment bonds, which we attribute to inter-segment hydrogen bonds. These bonds function as sacrificial bonds (as elaborated below), preventing molecular bond breakage during tensile stretching and enhancing toughness of the elastomer.
To evaluate the mechanical strength of HPWPUE, we first conducted uniaxial tensile tests at an extension rate of 50 mm/min. Notably, HPWPUE demonstrates impressive tensile strength (σ = 81.8 MPa), elongation at break (ε = 30.9), and toughness (τ = 0.959 GJ/m³) (Fig. 1e). We also observed that HPWPUE starts off completely transparent (leftmost photograph in Fig. 1a) but becomes opaque as the stretch ratio increases to 30 (rightmost photograph in Fig. 1a), a change attributed to strain-induced crystallization. Remarkably, the toughness of HPWPUE is 2.7 times greater than that of Darwin's bark spider silk, the toughest known natural material. During tensile tests, the sample's cross-sectional area significantly shrinks, making true tensile strength a better measure of HPWPUE's actual strength, which is determined to be 2.65 GPa. We determine that a HPWPUE specimen weighing only 81.7 mg can lift a load 61,200 times its weight (Fig. 1f). Additionally, we conducted puncture resistance tests, revealing that HPWPUE is difficult to pierce (right side of Fig. 1g). Force-distance curves from these tests show that HPWPUE outperforms commercial TPUE in puncture resistance (left side of Fig. 1g). Furthermore, HPWPUE exhibits a tear energy of 238.7 kJ/m and a fracture energy of 147.3 kJ/m, which are significantly higher than those of human tendons and ligaments (20-30 kJ/m) (Supplementary Fig. 8).
We believe that not all WPUE are suitable for the DCR as a reinforcement strategy. To investigate the requirements for this strategy, we synthesized additional WPU polymers using different chain extenders in place of the original EDA. The new chain extenders are diethylenetriamine (DETA), N, N'-di-tert-butylethylenediamine (DBEDA), and isophorone diamine (IPDA(NH)). The chemical structures of these extenders are shown in Fig. 2a. The rationale for selecting these chain extenders is as follows: DETA has an extra -NH- group compared to EDA, which enhances cross-linking density. DBEDA contains tertiary nitrogen that may hinder hydrogen bond formation and the absence of urea groups in WPU-DBEDA may also reduce interchain bonding. Lastly, IPDA has an asymmetric structure, which allows us to explore how molecular symmetry affects the DCR process. We also synthesized new WPUE by substituting the original isocyanates, HDI and HMDI, in HPWPUE with one of the following diisocyanates: hexamethylene diisocyanate (HDI), methylene-bis(4-cyclohexyl isocyanate) (HMDI), isophorone diisocyanate (IPDI(OCN)), and diphenylmethane diisocyanate (MDI). Among these, HDI is an aliphatic diisocyanate, HMDI is a cycloaliphatic diisocyanate, MDI is an aromatic diisocyanate, and IPDI(OCN) is an asymmetrically structured diisocyanate (see Fig. 2a). These four diisocyanates allow us to examine the influence of diisocyanate type on the DCR process. Gel permeation chromatograph (GPC) shows that all the new WPUE have approximately the same M of ~20 kg/mol and a polydispersity index of ~2 (Supplementary Table 2), similar to HPWPUE, ensuring that variations in molecular weight will have minimal impact performance. The resulting polymers are designated as WPU-X, with X representing the specific monomer used.
Tensile testing of these WPUE shows that HPWPUE exhibits significantly higher toughness and strength compared to the new WPUEs (Fig. 2b, c). In WPU-DETA, high cross-linking causes the material to fracture at small elongations, because the cross-linked network restricts the movement of the polymer chain segments. In WPU-DBEDA, insufficient hydrogen bonding leads to premature fracture of the material. In WPU-IPDA(NH), reinforcement does not occur at the stretch ratios where HPWPUE shows crystallization (Fig. 2b), suggesting that crystallization may be hindered in WPU-IPDA(NH). Among the WUP-X with varying diisocyanates, WPU-HMDI exhibits mechanical enhancements similar to those from SIC. This behavior can be attributed to its higher chain-segment rigidity, which enhances microphase separation and thereby impedes fragmentation of hard domains into separate segments required for DCR. WPU-HDI, which incorporates a flexible aliphatic diisocyanate, forms tightly bonded hard domains that do not effectively dissipate stress, resulting in low tensile strength and an absence of delayed crystallization. WPU-MDI exhibits an elevated elastic modulus as well as superior tensile strength and elongation at break, which can be ascribed to π-π stacking of the aromatic segments. Notably, both WPU-MDI and WPU-HDI exhibit the same qualitative trend during stretching and do not undergo self-enhancement before fracture. This observation indicates a physical network structure with strong hydrogen bonding, likely corresponding to tightly stacked hard domains within the polyurethane matrix before stretching. The strength of WPU-IPDI(OCN) is notably lower than that of WPU-HMDI due to its asymmetric structure, which hinders orderly stacking of hard domains. Additionally, the steric hindrance of the methyl side groups inhibits the formation of interchain hydrogen bonding, resulting in a lower elastic modulus compared to WPU-HMDI (Fig. 2b).
The structures of the seven WPUEs were analyzed using atomic force microscopy (AFM), Fourier Transform Infrared (FT-IR) spectroscopy, small-angle X-ray scattering (SAXS), and other characterization techniques (details provided in the discussion beneath Supplementary Figs. 9-11). Based on these analyses, we propose the following prerequisites for effective mechanical reinforcement through DCR: First, interchain hydrogen bonding must provide sufficient energy dissipation to prevent premature chain breakage during stretching. Second, the elastomer should not have a highly cross-linking structure, as this can hinder molecular rearrangements and lead to stress concentration. Third, the hard-domain structures within the polyurethane network cannot pack tightly, as this would impede their dissociation and reorganization during stretching. Finally, the elastomer chain segments should exhibit a strong tendency to crystallize upon alignment.
Additional results from stress-strain tests on HPWPUE at various temperatures reveal that temperature influences the effective implementation of the DCR process (Supplementary Fig. 12). As temperature increases to 40 and 50 °C, the stretch ratio at which strain hardening begins to occur decreases. We believe that the additional energy provided by the temperature increase disrupts some of the hard domains and disentangles the soft segments, allowing for an earlier crystallization process. Furthermore, the concave upward trend in the stress-strain curves gradually disappears at temperatures above 60 °C, preventing the realization of the delayed self-enhancement process. This phenomenon occurs because elevated temperature weakens the interactions between polymer chain segments and break interchain hydrogen bonds, leading to reductions in both tensile strength and elastic modulus. Elevated temperature also promotes chain mobility and thereby chain disentanglement, explaining decreases in both stress and strain at break (Supplementary Fig. 12b).
To understand the mechanical self-reinforcement of HPWPUE during the DCR process, we analyze the microphase behavior of HPWPUE at various stretch ratios (λ) or engineering strains (ε) where λ = 1 + ε. The polarizing optical microscopic (POM) images in Fig. 2d show the development of a notable alignment signature, indicating crystallization, when λ reaches 19. This observation is supported by AFM imaging at different stretching ratios (Supplementary Fig. 13). For 1 < λ < 12, blurred phase boundaries, smaller hard domain sizes, and weakened phase contrast are observed, indicating disruption of hard domains and mixing with soft segments. As deformation continues, the two phases start to blend. Beyond a stretch ratio of 20, a crystalline phase with distinct boundaries emerges. Figure 2e displays the differential modulus (dσ/dε) against ε. As ε increases from zero, dσ/dε initially decreases rapidly, indicating softening behavior. Between ε = 1.25 and 18, dσ /dε stabilizes. Beyond this range, it increases with ε, demonstrating strain-hardening behavior, similar to how a hard solid transitions to a soft rubber after yielding. We also calculated the true stress (σ) to analyze changes in the microphase structure. Treating HPWPUE as a cross-linked Gaussian-chain network, σ can be expressed as σ = G (λ -1/λ), where G represents the rubber elastic modulus, calculated as ρRT /<M>. Here, ρ is the density (1.12 g cm for HPWPUE), R is the gas constant, T is the absolute temperature, and <M> is the average molecular weight between cross-links. To relate the mechanical properties of HPWPUE to its microstructure, we plot σ against (λ-1/λ) in Fig. 2f. The inset reveals that HPWPUE quickly yields into a softer elastomer when λ -1/λ ≈ 5 or ε ≈ 1, consistent with Fig. 2e. Beyond this yield point, the data follows the relationship, σ = Y + G (λ-1/λ), which is also exhibited by TPUEs. Here, Y represents the yield stress, and G is the rubber elastic modulus during strain-hardening. Using this formula along with the G value obtained from the inset of Fig. 2f, we calculate <M> to be 2216 g/mol. This value is close to the molecular weight of the soft segments (PAA-PHG-PNPG) in our HPWPUE, suggesting that the hard domains act as network junctions.
The distinctly higher modulus for small λ-1/λ values below 5 is attributable to the soft segments in the soft domains wrapping around the hard domains, effectively shortening the length of the soft segments between junctions. However, this wrapping can loosen under tension. When λ -1/λ exceeds 300 (or λ > ~19), the slope of the σ versus (λ -1/λ) curve increases significantly beyond G, indicating the onset of crystallization, which aligns with the emergence of macroscopic alignment observed by POM at the same λ in Fig. 2d.
The presence of C = O, ether, alkoxy (acceptor), and -NH- (donor) in WPU leads to abundant hydrogen bonding in HPWPUE. These bonds form reversible cross-links during stretching, which we believe significantly influences the DCR process. To investigate this, we collect stretching-dependent FT-IR spectra of HPWPUE from λ = 1 to 31, as shown in Fig. 3a for the C = O and Fig. 3b -NH- spectral bands. The relative spectral intensities in these bands change noticeably with stretching, indicating dynamic alterations in hydrogen bonding between the C = O and -NH- groups. We conduct further analysis by calculating synchronous and asynchronous two-dimensional correlation spectra (2DCOS), with the results presented in Fig. 3c-f.
By applying Noda's judging rules to the 2DCOS of the carbonyl (C = O) band and analyzing the signs of cross-peaks in the synchronous and asynchronous spectra (Fig. 3c, d), as discussed further below Supplementary Fig. 14, we determine the response order of four sub-peaks in the C = O band with stretching: 1730/1740 cm → 1693 cm → 1710 cm → 1674 cm. This sequence represents the progression from free or disordered C = O in urethane groups to free C = O in urea groups, followed by ordered C = O in urethane groups, and finally ordered C = O in urea groups (see Supplementary Fig. 14e for the specific chemical environments of these C = O groups).
From this sequence, we infer that upon stretching, the hydrogen bonds in the disordered state with the lowest bond energy respond first, followed by those of C = O in the hard domains. We also deduce the order of response for the -NH- sub-peaks as follows: 3450 cm → 3260 cm → 3340 cm → 3320 cm (Fig. 3e, f). This sequence indicates the progression from free (non-hydrogen-bonded) -NH- groups (3450 cm) to -NH- groups hydrogen-bonded to ether oxygen (3260 cm), then to disordered (hydrogen-bonded) -NH- groups (3340 cm), and finally to ordered (hydrogen-bonded) -NH- groups (3320 cm) (refer to Supplementary Fig. 14e for details on the chemical environments of these -NH- groups). Using similar reasoning, this sequence indicates that changes in hydrogen bonding between the soft and hard domains occur first, followed by changes within the hard domain.
A large number of hydrogen bonds increase the energy dissipation, allowing HPWPUE to undergo interchain breakage and morphological change during the initial stretching stage, providing the basis for delayed crystallization. As the structure of the stretched HPWPUE becomes more compact and uniform, it is evident upon the initiation of crystallization (at λ > ∼21), the hydrogen bonding transitions to a stable structure. To further elucidate the role of hydrogen bonds in the DCR process, we employed density functional theory (DFT) to calculate the hydrogen bond binding energies between various donor and acceptor groups in the elastomer. Please refer to the supplementary information for details.
The calculation results show that the binding energies of the hydrogen bonds formed by different donor/acceptor groups vary (Supplementary Table 3). In the initial state, the bonding energy of hydrogen bonds in HPWPUE are calculated as follows: -9.000 kcal/mol between urea-urea groups, -7.949 kcal/mol between urethane-urea groups, and -8.096 kcal/mol between urethane-urethane groups. After stretching, these values increase to -14.879, -13.942 and -13.490 kcal/mol, respectively, indicating transformation of the hydrogen bonds from monodentate to bidentate (Fig. 3g and Supplementary Table 3) as the polymer chains are arranged along the stress direction. It is generally agreed that the primary driving force for hard domain aggregation is the strong intermolecular interaction between the urethane units, which facilitate the formation of inter-urethane hydrogen bonds and microphase separation. During the reorganization process of hydrogen bonds, the formation of additional bidentate hydrogen bonds enhances the structural ordering of the hard domains. This increase in binding energy from the bidentate hydrogen bonds can help delay early fracture in HPWPUE. As we will discuss further, this characteristic is essential for the high strength of HPWPUE.
To further elucidate how DCR enhances mechanical properties, we examine the strain-induced microstructure evolution of HPWPUE using small-angle and wide-angle X-ray scattering (SAXS and WAXS). Figure 4a displays the 2D-SAXS patterns of HPWPUE at various λ, along with their 1D intensity profiles, with the equatorial direction corresponding to the stretching direction. These patterns reveal changes in the microphase-separated structure, consistent with prior findings (see Supplementary Fig. 15). The mean inter-domain spacing, d, is calculated using Bragg's law d = 2π/q, representing the distance between hard domains before stretching. Initially, the SAXS pattern exhibited a ring pattern with maximum intensity at q = 0.58 nm (d = 10.83 nm, λ = 1), indicating a microphase-separated structure of hard domains imbedded within a soft-segment matrix. The circular and hence isotropic nature of the 2D SAXS pattern suggests random dispersion of the hard domains.
As strain increases, the 2D SAXS pattern changes from circular to oval and then to rhombus-shaped. Figure 4b presents the azimuthal 1-D profiles of the 2D pattern at varying λ, averaged over q = 0.10-1.11 nm. These profiles transition from no peaks at λ = 1 to two distinct peaks at 0° and 180° at larger λ signifying a shift from isotropic to anisotropic behavior. As λ increases from 0 to 13, peaks in the intensity profiles -- along both the equatorial and meridional directions (Fig. 4c, d, respectively) -- gradually disappear. This is attributed to the dissociation of hydrogen bonds and the breakdown of hard domains (Supplementary Fig. 1f). At the same time, the peak in Fig. 4c shifts to smaller q, indicating that the spacing between hard domains parallel to the stretching direction increases.
As λ increases from 14 to 31, the rhombus-shaped 2D-SAXS pattern enlarges and elongates, sharpening at the vertices perpendicular to the stretching direction. The final pattern resembles highly oriented lamellae. Along the stretching direction (Fig. 4c), a new peak appears around q = 0.44 nm (d = 14.27 nm) and shifts to larger q values with increasing λ, indicating the emergence of new domains with a decreasing interdomain distance. We attribute these domains to newly formed crystals in HPWPUE. The elongation of the 2D-SAXS pattern in the meridional direction with λ shows that these crystalline domains aligned increasingly with the stretching direction. Additionally, the scattering intensity in the meridional direction increases, with a faint peak appearing at around q = 0.43 nm (Fig. 4d). These results suggest that the hard chains rearranged due to the breakage of hard domains and concomitant formation of lamellar microdomains perpendicular to the chain axis.
Figure 4e shows the 2D-WAXS patterns of HPWPUE during stretching, with the stretching direction still in the equatorial direction as in the 2D-SAXS patterns. Initially, a broad halo is observed around q = 11.8 nm (Fig. 4f), corresponding to the average separation between soft-soft and soft-hard segments. As stretching progresses, the intensity of this halo decreases. In the region where λ > 20, a bright (blue) spot emerges in the upper right quadrant of the pattern, along with a new peak at q = 12.2 nm in the 1D meridional profiles (Fig. 4f). This implies that delayed co-crystallization occurs after the soft and hard segments have been rearranged during the stretching process. Figure 4g shows the 1D profiles along φ = 70°. The scattering peaks do not disappear during stretching, but rather gradually shift towards smaller q, reaching 11.37 nm at the end of the stretch. Therefore, changes in the intensity in the φ = 70° direction is attributed to the correlations between the PAA-PHG-PNPG domains, which diminishes as the co-crystallization domains between the hard and soft segments grow with stretching. This further reinforces the interpretation that co-crystallization occurs within this range of λ, with crystal orientation aligning parallel to the stretching direction.
The SAXS and WAXS results demonstrate that the microstructural evolution of HPWPUE corresponds with the DCR process illustrated in Fig. 1a during stretching. For comparison, we present a scatter plot of tensile strength (σ) versus elongation at break (ε) for various strategies aimed at enhancing polyurethane elastomers (Fig. 4h). This plot demonstrates that the DCR process yields both higher σ and ε for our HPWPUE (Fig. 4h and Supplementary Table 4), resulting in the highest toughness among these strategies.
Notably, our HPWPUE exhibits significantly greater toughness and tensile strength compared to other WPUEs reported to have high toughness (Fig. 4i and Supplementary Table 5). Furthermore, when comparing the mechanical properties of HPWPUE with those of recently developed tough gels and rubbers, we found that HPWPUE exhibited superior strength and toughness (Supplementary Fig. 16 and Supplementary Tables 6 and 7). Finally, we validated the scalability and application potential of HPWPUE. The large-scale HPWPUE retains its excellent mechanical performance, and the mechanical protective gloves coatings produced from it significantly outperform commercial products (Supplementary Figs. 17 and 18).