Design, synthesis, and structural characterization. High density of mechanically interlocked units is the structural basis for the integration and amplification of microscopic mechanical motions in MINs. In our work, we employed the [2]rotaxane as a monomer to construct MINs rather than as a pure cross-linker of conventional polymers as reported33, which could guarantee a high density of rotaxane moiety in the network. In specific, the [2]rotaxane monomer 1 was formed based on host−guest interaction between benzo-21-crown-7 (B21C7) wheel and secondary ammonium salt axle, of which one alkene unit was decorated each on the wheel and axle (Fig. 1c). Thanks to the designed [2]rotaxane, mechanically interlocked networks were constructed via photo-induced, radical-mediated thiol−ene click chemistry to transmit force on the mechanical bonds and ensure the integration and amplification effect. The variation in the feed molar ratios of DODT and PETMP enables facile modulation of the cross-linking densities of the target MINs. For this purpose, seven MINs named as MINs-1 − 7 were prepared with the DODT/PETMP molar ratios of 0/1, 0.3/1, 1/1, 3/1, 8/1, 15/1, and 1/0, respectively. It is noteworthy that in all samples, the contents of the rotaxane units always maintain at a high level, because the amounts of rotaxane monomer in the reactions are weakly affected by the ratios of DODT/PETMP.
The 1H NMR spectra were firstly obtained to prove the preparation of the [2]rotaxane 1. On the one hand, the aromatic protons H1 and H3 of B21C7 wheel moved downfield after the formation of [2]rotaxane 1 (Fig. 2a,b). Meanwhile, the ethyleneoxy protons H5−9 on the wheel showed complexed signals because of the host−guest interactions. On the other hand, the aromatic protons Ha and Hb on the secondary ammonium salt motif showed a significant upfield shift while the methylene peaks Hc and He shifted downfield, which could be ascribed to the combined macrocyclic shielding effect and hydrogen bonds of [N−H⋅⋅⋅O] and [C−H⋅⋅⋅O] (Fig. 2b,c). In addition, the appearance of signals Hg and Hh validated the success of the stoppering process. These observations, together with the characterizations of 13C NMR and electrospray ionization mass spectrometry (ESI-MS) (Supplementary Figs. 8−10), clearly demonstrated the formation of the [2]rotaxane 1 and similar results were also observed on the [2]rotaxane 2 (Supplementary Figs. 11−13).
With the [2]rotaxanes in hand, we then performed the thiol−ene click polymerization to prepare the MINs and the control. The Fourier transform infrared spectroscopy (FTIR) measurements indicated that the polymerization reaction was complete (Fig. 2d and Supplementary Fig. 16). The characteristic peaks corresponding to C − H stretching vibration at 3076 cm− 1 and C = C stretching vibration at 1641 cm− 1 for the alkene unit of [2]rotaxane 1 and S − H stretching vibration at 2556 cm− 1 of the DODT and PETMP disappeared in MIN-4. These results implied the formation of the covalently cross-linked MINs, and the same phenomena were also observed in control (Supplementary Fig. 17). We then conducted swelling experiments of MIN-4 in seven solvents with different polarities to confirm the formed network structure. The MIN-4 did not swell in nonpolar n-hexane, and even in good solvents of the other six, it only showed limited swelling ratios (< 480%), indicative of the cross-linked nature (Fig. 2e and Supplementary Fig. 23).
Fundamental performance characterizations of MINs. Network structure is closely related to the integration and amplification effect of MINs, and thus its influence on the properties of MINs was firstly investigated. Differential scanning calorimetry (DSC) was carried out to probe the glass transition temperatures (Tg) of MINs with different cross-linking densities. The results showed that the Tg values increased from 1.47 to 32.4°C with the increased content of PETMP cross-linker (Fig. 3a). Such a tendency was further confirmed by dynamic mechanical analysis (DMA) measurements (Supplementary Fig. 20). It is reasonable that denser networks are more likely to restrict the motion of the polymer segments, hence leading to higher Tg. Considering the fact that the motion of the mechanical bonds in the rubbery state of the material is more favorable than those in the glassy state, and thus the MINs with Tg lower than room temperature are more suitable for the study of structure−property relationships.
We then proceeded to study mechanical properties of the MINs with different cross-linking densities by tensile tests. The stress−strain curves of MINs-1 − 6 were shown in Fig. 3b and the corresponding mechanical property profiles were summarized in Supplementary Table 2. Upon increasing cross-linking density, the Young’s modulus grew significantly from 37.2 (MIN-6, DODT/PETMP = 15/1) to 281 MPa (MIN-1, DODT/PETMP = 0/1) (Fig. 3c). However, high cross-linking density undoubtedly sacrificed the tensile strain and toughness of the sample, as manifested by MIN-1 whose maximum strain and toughness are 31.6% and 7.0 MJ/m3, respectively. With a medium cross-linking density, taking MIN-4 for an example (DODT/PETMP = 3/1), its tensile strain and toughness reached maximum values of 542% and 47.2 MJ/m3, respectively, compared with the other MINs (Fig. 3b,c). Notably, because of the linear nature of MIN-7 (DODT/PETMP = 1/0), it cannot form free-standing film for tensile test. These observations indicated that cross-liking density of the MINs had a significant effect on their mechanical properties. Given a more balanced mechanical performance and also a proper Tg, MIN-4 was chosen as a representative in the following study.
The network structure of MIN-4 was clearly revealed by the master curves based on the time-temperature superposition (TTS) principle at a reference temperature of 30°C. As illustrated in Fig. 3d, three regions on the master curves were shown in a wide frequency range of 104 to 10−8 rad/s. The high frequency (f > 10) and the intermediate frequency (0.03 < f < 10) regions were assigned as the glassy and transition regions, respectively37. After the frequency lower than 0.03, a pronounced plateau with G′ > G″ referred to as a rubbery plateau emerged, where the material exhibited a predominately elastic property, suggestive of a decent network structure.
The comparison between our MIN-4 and the reported few rotaxane cross-linked networks in terms of swelling ratio and mechanical performance could reflect more favorable features of MINs with high-density rotaxane units. As illustrated in Fig. 3e, the swelling ratio of MIN-4 in chloroform was only 336%. By contrast, the swelling ratios of the reported rotaxane cross-linked networks were at least three-fold higher than that of the MIN-4 in the same good solvent33,34,38. Together with aforementioned results, we are able to speculate that the low swellability of MIN-4 can be ascribed to the densely rotaxanated cross-linked network. Generally, polymer networks exhibit a trade-off between swelling resistance and toughness; for example, increasing cross-linking density simultaneously leads to reduced swellability and toughness. However, our MIN-4 showed lower swellability but better toughness than those of reported counterparts (Fig. 3e). These results suggest that the large property enhancement mainly originates from the densely rotaxanated backbone and proper cross-linking density, wherein the former enables the increase of toughness via collective sliding motion of [2]rotaxanes and the latter as well as the host−guest binding suppress the expansion of the network in good solvents.
Macroscopic mechanical properties of high-density [2]rotaxane MIN. To better exhibit the property advantages of our MIN and understand its origin, a control sample was also designed, which consisted of [2]rotaxane 2 and the same thiol feeding ratio with MIN-4. The control could be regarded as a traditional covalent polymer network because the [2]rotaxane with two alkene groups on the ends of its axle, so that the B21C7 wheel cannot slide along the axle under applied force (Supplementary Fig. 25). As shown by the tensile curves in Fig. 4a, MIN-4 exhibited much better mechanical performances than those of the control in terms of stretchability (542 vs 268%), stiffness (Young’s modulus: 101 vs 1.68 MPa), toughness (47.2 vs 1.70 MJ/m3), and strength (maximum stress: 14.3 vs 1.26 MPa), indicating the important role of slidable mechanical bonds in improving mechanical properties.
Subsequently, the tensile tests with different deformation rates showed that the mechanical properties of MIN-4 are stretching rate dependent (Fig. 4b). In detail, Young’s modulus of MIN-4 increased with the increase of deformation rate, and the same tendency was also observed for the maximum stress (Fig. 4c). Although the fracture strain kept reducing, the changes were relatively small compared to the other two parameters. It has been proved that the stretching rate-dependent mechanical behaviors are often observed for non-covalently cross-linked networks39−42, and the noncovalent bond in our MIN could be attributed to the host−guest recognition.
Cyclic tensile tests for MIN-4 and control were also performed (Fig. 4d). Under the same strain of 100%, the hysteresis area between the loading and unloading curves of MIN-4 was about 12 times larger than that of control, as reflected by their energy dissipation values of 4.55 and 0.38 MJ/m3, respectively (Fig. 4e). Moreover, the ratio of energy dissipation to the incoming energy (damping capacity)43 was calculated to be 95.4% for MIN-4 and 50.7% for control (Fig. 4e). These results suggested that the dissociation of host−guest recognition and the consequent motion of mechanical bonds provided an effective pathway to dissipate energy.
Puncture resistance reflects the ability of a material to avoid breakage by a sharp object44. Corresponding tests for MIN-4 and control were conducted by using a toothpick at a compression speed of 50 mm/min. For MIN-4, the maximal puncture force and displacement reached 7.85 N and 5.75 mm, respectively (Fig. 4f). The puncture energy of MIN-4 was up to 23.5 mJ which was about 22 times higher than that of the control (1.09 mJ). The superior puncture resistance of MIN-4 could also benefit from the intramolecular motion of the [2]rotaxane unit: when the B21C7 wheel slides along its axle, the chain length of the network strand between cross-links increases. As a result, the tension of the polymer chains can be equalized, which diminished the localization of the stress and thus showed good puncture resistance (Supplementary Fig. 26).
Microscopic mechanical motion of [2]rotaxane moiety in MIN. Now, the macroscopic mechanical features of MINs have been elucidated, but to correlate them with the integration and amplification effect, the mechanical motion of [2]rotaxane unit also needs to be revealed. For this purpose, a continuous multi-step stress relaxation (SR) approach on a strain-controlled rotational rheometer was adopted (Fig. 5a). The first step of stress relaxation (i.e., SR-1) was under a linear strain amplitude of 0.5% to obtain the initial relaxation behavior of MIN-4. Subsequently, the second step of stress relaxation (i.e., SR-2) was carried out at a defined constant strain amplitude (γ0) ranging from 1 to 300% to trigger the mechanical motion of the [2]rotaxane moiety. We extracted the average plateau moduli (G0 ave) from SR-2 curves in 60−100 s to evaluate the state of the sample under different strains. The normalized average plateau modulus (G0 N) was calculated by the equation G0 N = G0 ave/G0, where G0 and G0 ave represent the average plateau moduli of the SR-1 and SR-2 stages, respectively. Finally, three consecutive stress relaxation tests (i.e., SR-3, 4, and 5) were performed under the strain amplitude of 0.5% to examine the response of the sample. It should be noted that the five consecutive stress relaxation curves represent a set of tests under each a defined strain.
As mentioned above, G0 N extracted from the stage of SR-2 could be employed to analyze the network behavior under different strains. There are prominent differences between the results of MIN-4 and control as shown in Fig. 5b. Significant decrease of G0 N values for MIN-4 started at smaller strains (< 20%) than that for control (> 30%), and the slope of its curve was also larger than that for control. In addition, even at small strains (< 10%), the values of G0 N for MIN-4 also declined slowly, which were well exhibited in Supplementary Fig. 27. It is well-known that polymeric networks under large strains would exhibit a decrease in modulus due to the network deformation. As for our MIN-4, the motion of mechanical bonds could also lead to the reduction of modulus, because sliding of B21C7 wheel on the axle is able to loosen the network17. Based on the results of MIN-4, we speculated that at small strains (< 10%), the intramolecular motion of the [2]rotaxane unit in the network had already begun, but after the strain of 20%, the mechanical motion and the deformation of the network structure took place simultaneously.
The speculation was verified by analyzing the whole process of SR-1−5. When the applied strain was set to be 7%, the stress relaxation results of the SR-1−5 were shown in Fig. 5c. Compared with SR-1, the modulus of SR-2 had a distinct decrease, which was consistent with the loosening of the network due to the intramolecular motion of [2]rotaxane moiety. During the processes of SR-3−5, the moduli declined continuously but slowly. Because SR-3 started immediately after the SR-2 without an interval, the decreased modulus indicated that after a relatively large-scale motion of mechanical bonds at SR-2, tiny mechanical motion continued to occur even under a small strain of 0.5% during the processes of SR-3−5. The same phenomenon was also observed for the experiment with the defined strain of 13% (Supplementary Fig. 28a). Such conclusion was further supported by the opposite results of the control in which a gradual increase of modulus was observed from SR-3 to SR-5 due to the recovery of the deformed network as analyzed below (Supplementary Fig. 29a).
Furthermore, we increased the defined strain to 20% to probe the dynamic behaviors of [2]rotaxanes in MIN (Fig. 5d). Similarly, the modulus at SR-2 was also much lower than that of SR-1. However, different from the results in Fig. 5c, the stress relaxation curves of SR-3−5 displayed more notable relaxation behaviors, and the moduli gradually increased from SR-3 to SR-5 at the same time. These were typical relaxation behaviors of deformation network. If a strain at the SR-2 stage gives rise to an elastic deformation of the network rather than just the motion of mechanical bonds, the deformed network would tend to recover to the original state at the SR-3 stage. The recovery process involves the retraction of the polymer chains, which would effectively relax the stress. And the elevated moduli from SR-3 to SR-5 reflected the reduced deformation of network strands in successive stress relaxations. This interpretation was verified by corresponding results for control which always presented gradual increased moduli in the procedures of SR-3 to SR-5 with a series of deformations (Supplementary Fig. 29). Furthermore, a more obvious phenomenon was exhibited by the test results with the defined strain of 30% (Supplementary Fig. 28b). Therefore, we can conclude that at small strains the motion of mechanical bonds took place, and larger strains (> 20%) led to the mechanical motion accompanied by the elastic deformation of MIN.
To further clarify the dynamic behavior of the MIN upon different strains, the stress relaxation curves at the stages of SR-3−5 were fitted by the equation of G(t) = G0exp(-t/τ), where G0 is the elastic plateau modulus and τ is the macroscopic relaxation time of the network. As shown in Fig. 5e, there is a cross point at the strain of around 20%. When the strains were lower than 20%, the τ values were in the following order: SR-5 < SR-4 < SR-3, indicating that SR-5 was able to relax stress faster than SR-3 and SR-4. Such a result was consistent with our above interpretation that under small strains, the motion of mechanical bond was dominant, and even under the strain of 0.5% during SR-3−5, the motion still slowly and continuously occurred. For the strains larger than 20%, a reverse trend was observed with the relaxation time in the following order: SR-5 > SR-4 > SR-3. As described above, the stage of SR-2 induced a large elastic deformation of the network whose recovery at stages of SR-3−5 caused notable stress relaxation, and most of deformation recovered at SR-3, thus leading to faster relaxation in this stage. Therefore, before the strain of 20%, the motion of mechanical bond was dominant, and after 20% strain, considerable elastic deformation of the network took place.
More information about the microscopic motion of mechanical bonds in MINs was provided by constrained geometries simulate external force (CoGEF) calculations45 (Fig. 5f and Supplementary Fig. 30). When a [2]rotaxane moiety was under tension, the elongation profile could be divided into four regions46. The initial stage (0−5 Å) was relatively flat without significant change of the energy, and this stage was mainly related to the initial stretch of the chains in the molecular structure. Subsequently, an obvious energy increase (up to 238.5 kJ/mol) was observed in the second stage (6−10 Å), which could be ascribed to gradual dissociation of the host−guest interaction. The third stage (10−18 Å) exhibited a relatively stable energy profile, where the B21C7 wheel moved away from its ammonium station and slid along its axle, namely, a motion stage. Finally, an enthalpic regime (> 18 Å) with a sharp increase in energy was observed. In this stage, the B21C7 wheel reached the site of the stopper, and then bond deformation occurred with the increase of strain. When the [2]rotaxane moiety sits in the fourth stage, further applied force would induce the elastic deformation of the network, as we analyzed above.
After proving the microscopic motion of mechanical bonds in MINs, the mechanical features of the MIN-4 with densely rotaxanated backbone could be well understood based on the integration and amplification mechanism. From the perspective of structural design, two factors are mainly responsible for the mechanical behaviors of the MINs: one is the cumulative motion of dense mechanical bonds, and the other one is the cross-linked structures. Meanwhile, the role of the cross-linked structure on the mechanical properties could be evaluated by the control (Fig. 4), and thus the enhanced properties of the MIN-4 compared with the control mainly stemmed from the integrated motion of dense [2]rotaxane units. For a single [2]rotaxane unit, the axial part between secondary ammonium salt site and stopper is initially presented as a dangling chain of the network, which becomes a part of the network after the sliding motion of B12C7 wheel along the axle upon stress. Due to the high-density [2]rotaxane units in the MINs, a large amount of such microscopic motion could be integrated and amplified into a distinct extension of the whole network, endowing the MIN-4 with good performance in ductility and puncture resistance (Fig. 4a,f). Moreover, the dissociation of host−guest recognition combining with subsequent sliding of B21C7 wheel provide an effective pathway to dissipate energy. Based on the contribution of numerous [2]rotaxane units, our MIN-4 is capable of dissipating the applied energy efficiently, showing markedly improved toughness. In a word, through integrating and amplifying microscopic motion of [2]rotaxane units, the MINs achieve the scale change of the network structure and macroscopic response of mechanical force simultaneously, which is reminiscent of the working mechanism of muscles. Moreover, such a mechanism is obviously different from the slide-ring materials in which the mechanical bonds are used to modulate the distribution of applied stress on conventional polymeric chains32, and thus MINs with densely mechanically interlocked backbones represent a new class of elastomeric materials.
Recovery properties of high-density [2]rotaxane MIN. The force induced motion of [2]rotaxane moiety has been elucidated, then we attempt to explore the recovery property of the mechanical bond after unloading. To this end, we developed a combined rheological method to evaluate the recovery behaviors of MINs under large strains47 (Fig. 6a). Firstly, a frequency sweep in linear viscoelastic region was carried out to characterize the initial state of our MINs. Then, a flow ramp was performed from 0 to a defined shear rate in each set of testing, where the range of the defined shear rate ( γ) increased from 0.02 to 0.64 s−1. Accordingly, the range of shear strain (γ0) grew from 1 to 1024% by the conversion of shear rate. This step was equivalent to apply different strains or stresses on MINs in the shear flow field. After that, a stress relaxation was conducted under maximum reverse strain amplitude (−γ0) to eliminate the applied strain or stress in each flow ramp step. Finally, a frequency sweep like the initial step was conducted to verify the degree of recovery for MINs. As shown in Fig. 6b, even the shear strain γ0 reached up to 1024%, the frequency sweep curves at the first and last stages overlapped well with each other. And the measurements under other shear strains also exhibited similar phenomena (Supplementary Fig. 32). Further, cyclic temperature ramp tests of MIN-4 were also conducted (Fig. 6c). Similar to the aforesaid results, the temperature ramp curves almost overlapped completely. These results indicated that the B21C7 wheel pulled away from the secondary ammonium salt site could move back quickly in the rheological tests.
To further study the recovery performance of MIN-4, the creep and recovery experiments were carried out at room temperature (Fig. 6d). Upon loading, creep strain gradually increased, and the strain reached ∼75% after 45 min. When the applied stress was released, the strain recovered largely with the residual strain lower than 5% within 55 min. In addition, the recovery property at room temperature was further investigated by cyclic tensile tests (Fig. 6e). The cyclic tensile curve of MIN-4 at a strain of 200% showed an obvious hysteresis loop, indicating efficient energy dissipation caused by the sliding motion of [2]rotaxanes during stretching. Then the second cycle without relaxing showed a small hysteresis loop, which was possibly because partial temporarily dissociated host–guest interactions did not fully recover in the time scale of an individual cycle. After about 30 min, the curve almost completely overlapped with the first cycle. The results of creep and cyclic tensile tests implied that even worked at room temperature, the force-induced sliding motions are reversible.