3.1 Molecular design and structure characterization
We synthesized a range of PUDS with varying mechanical properties using the synthetic route outlined in Fig. 1a, in which PCL-diol served as the soft chain and IPDI and 2-AFD acted as the hard chain. Notably, the presence of disulfide bonds within the PUDS molecular chain enabled exchange reactions with neighboring molecular chains, while hydrogen bonding in both the soft and hard segments facilitated the reorganization of the molecular network, promoting self-healing capabilities (Fig. 1b and 1c). Elastomers were synthesized with varying molar ratios of hard to soft segments, and their properties were optimized, with hard segment and disulfide content presented in Table S1. The Mn, Mw, and PDI were determined by GPC and listed in Table S2. For the PUDS range of polymers, the PDI ranged from 1.67 to 1.99, demonstrating a narrow molecular weight distribution. NMR spectroscopy (Fig. S2 and S3) confirmed the structure of the adhesive and demonstrated a gradual increase in the hydrogen of the imino group on the IPDI and the benzene ring on the 2-AFD as the hard segment content increased. ATR-FTIR spectroscopy (Fig. S4) characterized the final product, with the peaks at 3366 and 1644 cm− 1 assigned to the -N-H and -C = O stretching vibrations of the urea bond, respectively. The disappearance of the characteristic peak of the -NCO stretching band at 2269 cm− 1 provided compelling evidence that the -NCO group reacted with the amino group of the chain extender to form the urea group at the end of the reaction.
To further investigate the intermolecular hydrogen bonding behavior, we employed a Gaussian function to fit the stretching vibrational region of the C = O group, revealing three absorption peaks (Fig. 2a): the stretching vibrational of the free hydrogen bonded C = O group at 1734 cm− 1, the disordered hydrogen bonded C = O group at 1720 cm− 1, and the ordered H-bonded C = O group at 1693 cm− 1. The percentage of hydrogen bonding content in the PUDS samples was quantified and plotted in Fig. 2b. As the hard segment content increases, the degree of ordered hydrogen bonding in the system gradually increases, promoting an ordered structure. The enhanced hydrogen bonding interactions offer a strong driving force for microphase separation. The microstructure of the PUDS sample was further investigated using XRD. The XRD pattern displayed only a broad amorphous peak centered at approximately 20°, indicating that the polymer chains assume a loosely packed conformation, precluding crystallization in the system (Fig. 2c)[24, 54]. SAXS analysis revealed with the period length gradually increased as the hard segment content increased indicating fuller micro-phase separation within the material (Fig. 2d). The AFM phase diagrams provide a visualization of the microphase separation in the PUDS samples, with dark and bright regions indicating soft and hard domains, respectively (Fig. 2e and S5). The diagrams reveal a significant microphase separation structure and a loose hard domain arrangement that becomes more pronounced with increasing hard segment content. This achievement enables the microscopic modulation of the polymer hard domains. Due to the relatively high content of ether oxygens in the PUDS-1 system, a large number of hydrogen bonding sites are provided for the hard domains to form weak hydrogen bonds. This interaction between the two phases can lead to incomplete microphase separation. Quantum chemical calculations revealed the binding energies (ΔE) relative to the monomer, providing evidence of hydrogen bond aggregation and multiphase hydrogen bonding within the system. These findings, as illustrated in Fig. S7-S9 and Table S4, demonstrate a graded strength of the hydrogen bonds. ATR-FTIR analysis was utilized to investigate the hydrogen bonding interactions in PUDS within the temperature range of 25–95°C, as depicted in Fig. S10. As the temperature increased, the stretching vibrations of N-H at 3366 cm− 1 and the N-H bending vibrational band at 1537 cm− 1 shifted towards lower wave numbers (red-shift), indicating an increase in bond length and a decrease in stretching vibrational frequency. In contrast, the stretching vibrational band of C = O (urethane) at 1725 cm− 1, C = O (urea) at 1644 cm− 1, and -O- (PCL) at 1160 cm− 1 displayed a blue shift, indicating that reversible hydrogen bond dissociation was taking place within the PUDS network.
To adapt the adhesive to the environment in which the ECMs are to be used and stored, a qualitative assessment of the thermal stability of the adhesive is usually performed. As shown in Fig. 2f, the DSC curve of PUDS from − 80 to 100°C indicated that the glass transition temperature (Tg) of the adhesive is around − 42°C, which is attributed to the excellent low-temperature properties of PCL. Furthermore, the thermal weight loss curves of self-healing polyurethanes with varying disulfide levels were studied by TGA (Fig. S11). The PUDS series adhesives exhibited good thermal stability, with an initial decomposition temperature greater than 220 ℃. At approximately 450°C, all weight was lost, and higher disulfide content resulted in greater weight loss.
3.2 Self-healing performance and mechanical properties
Integrating self-healing properties into adhesives is an effective approach to enhancing material performance and prolonging service life. We evaluated the self-healing performance of the PUDS series adhesives through scratch recovery experiments, as shown in Fig. 3a and S12. The width of the approximately 80um scratch on the surface of the PUDS-3 sample gradually decreased and became almost flat after 10 minutes at 60°C without external forces. This self-healing ability is related to the mobility of free radicals on the fracture surface. Molecular chains can effectively promote the exchange reaction of disulfide bonds under heating conditions, contributing to the rapid healing of scratches. To provide a more visual demonstration of self-healing, a rectangular film with a cut area of 5mm*1mm was reassembled and, after 10 minutes of recovery at 60°C, the two films bonded well enough to withstand manual pulling, demonstrating excellent self-healing ability, as shown in Fig. 3b.
Further mechanical properties and self-healing efficiency of the PUDS samples were evaluated through tensile tests using a universal testing machine, as shown in Fig. 3c. The initial ultimate tensile strength (UTS), strain at break, and toughness are listed in Table S5, with UTS increasing from 2.79 MPa to 9.01 MPa with increasing hard segment content. We cut the PUDS film completely in half and then reattached it at 60°C at different times to obtain stress-strain curves, as shown in Fig. 3d and S13. The self-healing efficiency, defined as the ratio of the toughness of the healed material to that of the original material, was assessed to evaluate the self-healing properties of the PUDS films (Fig. 3e). The self-healing efficiency of PUDS-1 reached 97%, which is much higher than that of the other samples. As the healing time continued, the mechanical properties of all samples gradually increased, and the final healing efficiency remained above 90%. Increasing the disulfide content facilitated the formation of more compact and robust physical cross-linking, enhancing the mechanical strength of the adhesive. However, it also resulted in the formation of regular hydrogen bonds between the molecules, which could reduce the mobility of the molecular chains and impede the self-healing effect. The structures of PUDS-1 and PUDS-5 were determined using all-atom molecular dynamics (MD) simulations, as depicted in Figs. 4a and 4b. The hard segments of both polymers aggregate and form a structure consisting of separate soft and hard phases due to hydrogen bonding cross-linking. While the H-bonding sites on both polymer chains are identical, PUDS-5 forms a denser array of hydrogen bonds with 3.5 H-bonds formed on each hard segment, compared to only 2.8 H-bonds formed on each hard segment in PUDS-1.
The recovery properties of the adhesive are of great importance for its use under various loads. In Fig. 4c, the cyclic stretching curve of PUDS-3 is presented at a maximum strain of 100%. The stress-strain curve exhibits significant hysteresis in the first cycle, indicating that the breakage of disulfide and hydrogen bonds during the stretching process consumes a considerable amount of energy, and these bonds cannot be reconnected quickly. Therefore, the tensile stress decreases as the number of cycles increases. However, after five cycles and 60 minutes of recovery, the sample was stretched again, and the cyclic stretching curve largely overlapped with the original curve, indicating excellent mechanical recovery. The rapid rearrangement of hydrogen and disulfide bonds is a prerequisite for the recovery of deformation. Moreover, the PUDS-3 spline can withstand approximately 5,500 times its weight without tearing (Fig. 4d), and after removing the weight, it largely restored its original state within 10 minutes, demonstrating its high toughness and rapid recovery from deformation. The results of the time-aging mechanical test show that after 180 days of exposure to air, the tensile properties of the PUDS-3 film remained consistent (Fig. 4e), and the self-healing properties of the polymer chains exhibit little change. A high concentration of dynamic bonding aids in the recombination of broken regions but also results in more hydrogen bonding interactions between molecular chains. Therefore, adjusting the proportion of hard segments is a feasible approach to achieving a reasonable configuration of mechanical and self-healing properties. Compared to the relevant references reported so far (Fig. 4f), the PUDS system designed in this work not only exhibits excellent toughness but also demonstrates better self-healing capacity at moderate temperatures.
3.3 Rheological Properties
A rheological perspective can provide further insights into the dynamic behavior of the adhesive and the crack self-healing mechanism, as illustrated in Fig. 5a. PUDS-1 exhibits a low-stress relaxation in the same period, with the relaxation modulus decreasing rapidly with time and gradually approaching zero stress. As the hard segment content increases, the physical connection network becomes more complex, resulting in a gradual increase in relaxation time [64]. Additionally, the sample's relaxation modulus exhibits clear temperature dependence, as illustrated in Fig. 5b. Specifically, relaxation of the PUDS system occurs more quickly at 60°C than at room temperature, due to the weakening of intermolecular hydrogen bonds and the promotion of exchange reactions among dynamic bonds, which accelerates the relaxation rate.
We also investigated the temperature-dependent loss factor (tan δ) of the PUDS system in the range of 25–100°C (Fig. 5c). The PUDS-3/4/5 sample exhibited a tan δ value lower than unity (tan δ = 1) across the entire temperature range, indicating a solid-like behavior. This behavior is attributable to an increase in the hard segment content, which can partially immobilize the dynamic bonds and leave the material in an elastic state, thereby adversely affecting the self-healing performance of PUDS. Moreover, a frequency sweep of PUDS within the linear viscoelastic region (Fig. 5d and Fig. S14) revealed increased values of both G' and G" with increasing hard segment content. This effect was attributed to the formation of minute crystalline domains within the benzene ring structure of the molecular chain, which enhances the molecular cohesion of PUDS, imparting a more solid-like character to the material and resulting in an increase in the G' value. On the other hand, intermolecular hydrogen bonding enhances the entanglement of PUDS chains and the energy dissipation due to deformation, leading to a gradual increase of the G'' value. The lower the test temperature of the PUDS series samples, the more frequency at which G' and G'' intersect (ωcross) tends to occur at lower frequencies, as at that time, the molecular chains have stronger interactions with each other as well as lower kinematic activity. Conversely, higher temperatures facilitate the dissociation of physical networks and reversible disulfide bonds in the material. The cross point of G' and G'' appears when the scanning temperature of the PUDS-3 sample exceeds 60°C, indicating that a significant disulfide exchange reaction has occurred in the system. The reciprocal of the corresponding ωcross indicates the characteristic relaxation time (\(\tau =1/\omega\))[52, 65], which represents the timescale for the stresses in a material caused by cracking to return to their initial state[66]. PUDS-1 exhibits a lower τ than other adhesives over the entire frequency range at the same temperature. We obtained the apparent activation energy (Ea) of the relaxation process by linearly fitting Lnτ and 1/T at different temperatures according to the Arrhenius equation, and the Ea values for PUDS-1/2/3 were 99.6 kJ/mol, 100.2 kJ/mol, and 124.4 kJ/mol, respectively (Fig. S15). Ea increases with the increase of hard segment content, indicating that the mechanical strength of the sample is strengthened at room temperature while ensuring sufficient self-healing ability after increasing a certain temperature. At 60°C, PUDS-5 shows high yield stress and low Cole-Cole slope values, further indicating that it is a relatively hard system (Fig. 5e and S16)[53, 67].
Overall, the results of our study demonstrate the importance of the hard segment content in determining the mechanical and self-healing properties of PUDS adhesives. The increase in hard segment content leads to a higher degree of physical crosslinking, which enhances the mechanical strength and rigidity of the material but also limits its deformation ability and self-healing performance. On the other hand, a lower hard segment content results in a more flexible and deformable material with faster disulfide exchange kinetics, but at the expense of reduced mechanical strength and stability. PUDS-3, with an intermediate hard segment content, exhibits a good balance between these properties, making it a promising candidate for further development as a self-healing ECMs.
3.4 Adhesive properties and fracture healing ability of ECMs
When the interfacial bond between the energetic material and the adhesives is weak, air trapped at the interfaces can increase the thermal resistance of the ECMs system, resulting in a higher risk of explosions. However, a strong adhesive bond can effectively dissipate energy and prevent debonding, reducing the vulnerability of ECMs during production, transportation, and storage [68]. To investigate this effect, we performed tensile tests on PUDS adhesive films sandwiched between two TATB pellets (Fig. 6a) to measure the interfacial bond strength between the film and the energy-containing compound. We hypothesize that the presence of -NH2 and -NO2 groups on the TATB molecule creates strong interfacial forces between TATB and PUDS, as shown in Fig. 6b where possible interfacial hydrogen bonding interactions are illustrated. The bond strength increased with increasing disulfide content, with PUDS-5 exhibiting the highest adhesion strength of 186.2 kPa (Fig. 6c and S17). The interfacial bonding strength between particles and the polymer matrix in the interfacial region can be directly assessed by the interaction energy (Einter) [69]. In this study, interfacial models of TATB and PUDS chains were constructed using molecular dynamics (MD) simulations. The optimized interfacial adhesion structure yielded a trend of Einter that aligns with the results obtained from tensile tests, suggesting that hydrogen bonding interactions effectively enhance interfacial adhesion (Fig. 6d and S18).
In consideration of the comprehensive properties required for ECMs, including self-healing ability, mechanical strength, and adhesion, PUDS-3 emerges as a highly promising polymer. To explore its potential application in ECMs systems, we prepared ECMs containing 20 wt% PUDS-3 and 80 wt% TATB using the slurry method. Compatibility is a crucial factor in ensuring the safety and reliability of explosives. According to the compatibility assessment criteria outlined in Table S6, we performed a pressure transducer method (VST) test, and the result showed a net increment of 0.041, indicating that PUDS is compatible with TATB. The resulting ECMs powder was then compacted using the press-fitting technique to obtain a sample strip measuring 30 mm × 5 mm × 2 mm, the original σ and ε values of the specimen were 3.84 MPa and 13.03%, respectively. To investigate the effects of initial damage and healing on the mechanical behavior of ECMs, samples with scratches (width: 0.25 mm, depth: 0.5 mm) were allowed to self-heal at 60°C for varying periods, as shown in Fig. 6e. It was found that the maximum tensile strength of freshly scratched samples was only 36% of the original value. However, after 6 hours, the samples recovered to 64% of their maximum tensile strength, and after 24 hours, more than 96% recovery was achieved, demonstrating the excellent self-healing efficiency of PUDS-3. The self-healing of cracks in ECMs is primarily dependent on the self-healing ability of the adhesive and the thermal movement of its chains[7]. However, the less plastic crystal structure of TATB explosives restricts the movement of polymer chains in the ECMs system, which impedes the rearrangement of the polymer network. Consequently, it takes time for the molecular chains to come closer spatially, enabling the disulfide bond exchange reaction and H-bonding at the fracture interface, which ultimately facilitates the self-healing effect. Compared to currently reported ECMs with crack healing ability [6, 43, 46, 70–72], this ECMs not only exhibits high self-healing efficiency but also possesses exceptional mechanical properties. Finally, we evaluated the storage stability of ECMs placed at room temperature for 180 days and found that it retained a tensile stress of 3.83 MPa and 13.19% elongation at break, indicating good stability (Fig. S19).
In addition, we used confocal laser scanning microscopy to observe the recovery of scratches in the longitudinal plane. Three-dimensional surface and depth profiles of ECMs with microcracks quantitatively demonstrate the self-healing process of the notches (Fig. 6f and Fig. 6g), which gradually decrease in depth with time after being heated at 60°C for 12 h until they become flat again after 24 h. This process is achieved through the migration of adhesive molecular chains and internal tension, effectively healing the damaged interface and maintaining good structural integrity. We then evaluated the self-healing efficiency of ECMs at different temperatures, as shown in Fig. 6h. The gap size was 0.5 mm deep and the rate of self-healing was defined as follows:
Self-healing speed = notch depth/self-healing period
The self-healing rate increased with increasing temperature and reached a maximum of 41.67 mm/h at 80°C, which was higher than that at 60°C. This is due to the increased energy stimulating molecular relaxation and conformational transitions, and breaking more dynamic disulfide bonds. The broken short chains migrated and diffused more readily than the long chains, thus increasing polymer chain activity and promoting crack healing [46]. Figure 6i shows the dissolution recovery characteristics of the ECMs sample, where TATB in the sample can be fully recovered after heating in DMAc at 60°C for 12 h. Dynamic disulfide bonding not only provides ECMs with healing ability but also degrades in solvents at high temperatures, thus bringing significant economic benefits and reducing the recovery cost of ECMs.