Rejuvenation of annealed and grown LCEsFigure 2a illustrates the compositions of LCEs and catalysts for rejuvenation. LCEs were synthesized from base-catalyzed thiol-Michael addition reaction among the liquid crystal monomer 1,4-bis-[4-(3-acryloyloxypropyloxy)benzoyloxy]−2-methylbenzene (RM257), chain extender 2,2’-(ethylenedioxy) diethanethiol (EDDET), and crosslinker pentaerythritol tetra(3-mercaptopropionate) (PETMP), with stoichiometric acrylates and thiols. RM257 and PETMP provided ester bonds for further transesterification. Fourier transform infrared (FTIR) spectroscopy showed that acrylates and thiols were totally converted, where typical IR signals assigned to the acrylate group of RM257 (C = C stretching at 1637 cm−1) and sulfhydryl peak (S-H stretching at 2568 cm−1) disappeared (Supplementary Fig. 1). Swelling experiment showed the gel content was 97.7% (Supplementary Note 1 and Supplementary Fig. 2). For non-fresh LCEs, annealed samples (A-LCE) were prepared through full annealing of freshly prepared LCEs (F-LCE), while grown samples (G-LCE) were obtained after the self-growth of F-LCE (Methods).Fig. 2: Synthesis, rejuvenation and regrowth of self-growing LCEs.a Composition of the self-growing LCE, and the catalyst for rejuvenation. b FTIR spectra of nTBD, A-LCE, and annealed R-LCE. c Rejuvenation of A-LCE and G-LCE. Scale bars: 5 mm.Rejuvenated LCE (R-LCE) was obtained by swelling the non-fresh LCEs with a transesterification catalyst (neutralized 1,5,7-triazabicyclo- [4.4.0]-dec-5-ene17, nTBD). Unless otherwise noted, this was achieved by immersing non-fresh A-LCE and G-LCE in a dichloromethane (DCM) solution containing 3 mg mL−1 of nTBD for 48 h at ambient temperatures. The synthesis and introduction of nTBD was detailed in Method and Supplementary Fig. 3. Swelling experiments showed that the catalyst nTBD content in the LCE was approximately 0.84 wt% after rejuvenation (Supplementary Note 1). FTIR spectroscopy of annealed R-LCE showed the appearance of N-H stretching peak at 1646 cm−1, indicating that nTBD was swelled into the sample (Fig. 2b). After swelling for an extended period, the shape of A-LCE remained the similar, while the G-LCE reverted back to its original orientated state as the growth history was erased (Fig. 2c).After swelling with nTBD, the obtained R-LCEs were effectively reverted back to the initial state as they exhibited similar properties to F-LCEs. As differential scanning calorimetry (DSC) curves showed, the F-LCE and R-LCE exhibited similar glass transition temperatures (Tg) and isotropic transition temperatures (Ti) after annealing, with values of Tg ~ 2.9 °C, Ti ~ 74 °C for the former and Tg ~ 6.6 °C, Ti ~ 69 °C for the latter (Supplementary Fig. 4). The thermogravimetric analysis (TGA) curves indicated that both annealed F-LCE and R-LCE showed 5% thermal decomposition temperatures well above the operating temperature, with values of 342.5 °C and 325.9 °C respectively (Supplementary Fig. 5). The stress-strain curves of F-LCE and R-LCE before and after annealing were compared (Supplementary Fig. 6, Supplementary Fig. 7). Results showed that the mechanical properties of A-LCE and annealed R-LCE were similar. Specifically, A-LCE and annealed R-LCE exhibited fracture stresses of 2.56 MPa and 2.25 MPa, and fracture elongations of 409% and 390%, respectively (Supplementary Fig. 7). Birefringence behavior of two samples was found to be exactly the same under polarizing optical microscope (POM). During the heat-cooling cycle, both R-LCE and F-LCE initially showed strong birefringence at room temperature while their birefringence disappeared after annealing (Supplementary Fig. 8). The d-spacing of R-LCE and F-LCE were also very similar, indicating similar microstructures (Supplementary Fig. 9). X-ray diffraction (XRD) revealed that a peak at 2θ ~ 20.0° was observed for both R-LCE and F-LCE, indicating their d-spacing of 4.44 Å. In addition, peaks of F-LCE and R-LCE after annealing were observed at 2θ ~ 19.5°, and their d-spacing was calculated to be 4.55 Å.In addition, creep resistance of annealed R-LCE was analyzed under a normal force of 10 N and a constant shear stress of 3 kPa under different temperatures (80 °C, 90 °C, 100 °C, 120 °C, 140 °C), and the induced deformation was recorded over the measurement time (Supplementary Fig. 10). At the actuation temperature of 80 °C, there was almost no deformation. Even at 90 °C, which was higher than the actual usage temperature (80 °C), no substantial deformation appeared (only 0.36% in 600 min), indicating that the actuators have good creep resistance under this temperature. As the temperature was increased to 120 and 140 °C, the onset of creep was obviously detected.Regrowth into soft actuators of R-LCER-LCE, whether rejuvenated from A-LCE or G-LCE, could spontaneously regrow at room temperature just like F-LCE (Fig. 3a). After being pre-stretched to a 30% strain and fixed on both ends, the R-LCE spontaneously elongated beyond the length before growth and arched itself over time in ambient environments, without the need for external stimuli or energy input (Fig. 3b).Fig. 3: Regrowth and characterization of R-LCE.a Regrowth of R-LCE. Scale bar: 5 mm. b Images of self-growing process of R-LCE. Scale bar: 5 mm. c Length, POM and X-ray diffraction images of the rejuvenation and regrowing process of G-LCE. Scale bars: 500 μm.The R-LCE also naturally became soft actuators at room temperature after self-growing, similar to F-LCE (Fig. 4a). POM images were taken to track the rejuvenation-regrowth process (Fig. 3c). Birefringence of both G-LCE and regrown R-LCE showed significant 45o contrast inversion to the analyzer under the crossed polarizers and analyzers, indicating uniaxial alignment. XRD patterns displayed a pair of arcs along the alignment direction for both the G-LCE and regrown R-LCE and the order parameters (S) were calculated to be 0.61 and 0.59 respectively, indicating well orientation (Supplementary Note 2 and Supplementary Fig. 11). In contrast, after annealing, the R-LCE obtained from G-LCE exhibited minimal birefringence under POM. In addition, XRD images of R-LCE in Fig. 3c showed a uniform ring of scattering with minimal azimuthal bias of intensity, indicating that the growth history was effectively erased.Fig. 4: Potential influencing factors for rejuvenation, regrowth and soft actuators.a Preparation of a soft actuator formed from R-LCE. Scale bar: 5 mm. b Actuation strains of five rejuvenation cycles. Actuation strain = (Lc – Li) / Li, Li: the length of sample at isotropic phase; Lc: the length of sample at anisotropic phase. c Elongation strain and actuation strain under different swelling times. Elongation strain = (Lg – L0) / L0; L0: the original length; Lg: the length after self-growth. d Elongation strain and actuation strain for different concentrations of catalyst. e Elongation strain and actuation strain for different catalysts (nTBD and DPA).The rejuvenation-regrowth process can be repeated more than 5 cycles, while maintaining a relatively consistent actuation strain. As Fig. 4b showed, the rejuvenation process was repeated on a single G-LCE sample for five times. Each time, the sample transformed into a soft actuator after self-growth. The initial actuation strain of the original G-LCE was recorded at 51.0%. Subsequent rejuvenation cycles resulted in actuation strains of 49.3%, 47.1%, 48.7%, and 50.2%, respectively. The LCE soft actuators exhibited high actuation stability at 80 °C over 1000 cycles of deformation (Supplementary Note 3 and Supplementary Fig. 12), despite fixing orientation at room temperature through exchange reactions. It should be noted that nTBD was stable within the annealed R-LCEs, not only during annealing at 50 °C but also in retaining its functionality for more than six months (Supplementary Note 4, Supplementary Figs. 13 and 14). Therefore, in the next rejuvenation cycles, a solution containing less catalyst or even just solvent itself also worked for rejuvenation and regrowth (Supplementary Fig. 15).Potential influencing factors for rejuvenation and regrowthThe swelling time of the rejuvenated process was investigated. As shown in Fig. 4c, the elongation strain of R-LCEs and the actuation strain of resulting actuators both increased with longer swelling times (1 h, 6 h, 12 h, 24 h, and 48 h). The average elongation strain for samples swelling for 1 h, 6 h, 12 h, 24 h and 48 h is 38.8%, 45.7%, 48.2%, 51.5% and 52.1%, respectively. The average actuation strain of the obtained actuators when swelling for 1 h, 6 h, 12 h, 24 h and 48 h is 38.7%, 42.7%, 47.2%, 49.9% and 51.5%, respectively. We noticed that there’s only a slight increase in elongation strain when swelling for 24 h and 48 h (51.5%-52.1%), indicating the transesterification reaction reached equilibrium, and a maximum elongation has been achieved.The concentration of catalyst during the swelling process also influenced the growth results. We subjected A-LCEs to swelling with different concentrations of nTBD (0.3 mg mL−1, 1.5 mg mL−1, 3 mg mL−1) for 48 h. The elongation strain and actuation strain of the resulting actuators increased with the higher catalyst concentration (Fig. 4d). The average elongation strain for concentrations of 0.3 mg mL−1, 1.5 mg mL−1, and 3 mg mL−1 is 47.0%, 51.5%, and 52.1%, respectively, while the corresponding average actuation strain is 43.3%, 45.4%, and 51.5%.Catalysts were also compared for their effect on rejuvenation. Two transesterification catalysts, DPA and nTBD, were used as examples. We chose a catalyst concentration of 3 mg mL−1 for comparison. As shown in Fig. 4e, when nTBD was used as catalyst, the average elongation strain of R-CLEs reached 51.5% when swelling for 24 h, and 52.1% when swelling for 48 h. For DPA, the average elongation strain was 41.6% when swelling for 24 h, and 46.2% for 48 h. The results suggested that nTBD was a more effective catalyst for rejuvenation. Moreover, DPA and nTBD also resulted in different actuation strains in the resulting actuators. An average actuation strain of 51.5% was obtained by swelling with nTBD for 48 h. In contrast, the actuation strain was 32.6% by swelling with DPA for 48 h. It should be noted that the average elongation strain reached 46.2% under this condition, indicating a relatively poor capability of DPA in fixing alignment. The reason could lie in the volatility of DPA22. The DPA gradually evaporated before the alignment was completely fixed, resulting in a smaller actuation strain compared to samples processed with nTBD. Therefore, we chose nTBD in the swelling process in the following study.Investigation on the mechanism of rejuvenationRejuvenating non-fresh A-LCE and G-LCE required resetting their network structure back to the initial state of F-LCEs. We tried to investigate the necessary conditions for rejuvenation. Pure solvent was first examined. During swelling, solvents physically altered the conformation of polymer chains by disrupting the π–π interaction of benzene rings in molecular pairs23 in liquid crystals. As a result, the A-LCE and G-LCE transformed from the liquid crystal phase to the isotropic phase temporarily. However, non-fresh LCEs that were swelled with pure solvent failed to rejuvenate and regrow. Once the solvent evaporated, they reverted to the annealed or grown state before swelling (Supplementary Fig. 16). This indicated that the solvent alone only temporarily changed the conformation of the polymer chain but cannot reorganize the structure. Consequently, rejuvenation cannot be achieved solely by the physical effect of solvents.Rearranging the topological structure of the network through dynamic covalent bonds was essential for rejuvenation. Only when swollen with transesterification catalyst, both A-LCEs and G-LCEs were successfully rejuvenated (Fig. 5a). Shear stress relaxation tests were conducted to measure the impact of nTBD on transesterification within the network (Fig. 5b). Results showed that the relaxation time τ* (time needed to relax to below 1/e of the initial stress relaxation modulus) follows an Arrhenius law with elevated temperatures. The activation energy (Ea) was calculated to be 104.3 kJ mol−1, and the topology freezing transition temperature (Tv) was calculated to be 57.1 °C (Fig. 5c, Supplementary Note 5, Supplementary Note 6 and Supplementary Fig. 17). However, the transesterification in this thiol-acrylate LCE system can intrinsically occur at ambient temperatures, as reported in our previous work24. We also replotted the stress relaxation data in to log-log scale without normalization, as the interpretation of stress relaxation data was influenced by how the data were plotted (Supplementary Note 5, Supplementary Fig. 18). In addition, the stress relaxation data were fitted with more advanced stretched exponential model, which was also called as Kohlrausch-Williams-Watts (KWW) model (Supplementary Note 7, Supplementary Fig. 19)25,26,27. The results suggested that stress could fully relax at infinite time due to the exchange reaction.Fig. 5: Network rearrangement for rejuvenation.a Comparison of swelling with and without nTBD. b Normalized shear stress relaxation of sample after swelling with nTBD at varying temperatures. c Arrhenius plot of the measured relaxation time.Ester bonds involved in the transesterification was also investigated (Supplementary Note 8). Zhao and co-workers reported that heterolytic reaction could occur between the aromatic ester group in mesogen core and aliphatic esters. Although the work said that heterolytic reaction involving aromatic esters is rarely activated at 100 °C16, well above the rejuvenation temperature, we investigated that whether the heterolytic reaction occurred during rejuvenation. DSC results showed that Tis of A-LCE (without nTBD) and annealed R-LCE (containing nTBD) were both around 74 °C during the heating/cooling cycle between −50 to 100 °C (Supplementary Fig. 20). 1H nuclear magnetic resonance (NMR) spectra of RM275 monomer, mixture of RM257 and nTBD at 0 h, and mixture of RM257 and nTBD after stirred at room temperature for 48 h were also compared (Supplementary Figs. 21–23). The signals from 6.45 ppm to 5.83 ppm corresponding to the protons of CH2 = CH- in acrylate groups of RM257 maintained at the same position and no additional peak appeared. In addition, the integral of these protons did not change in all the three samples. These results indicated that the aromatic esters did not involve in the transesterification under our rejuvenation situation.Based on these results, we propose that the synergistic effects of solvents and dynamic covalent bonds played crucial roles in the rejuvenation process. During swelling, the solvent exerted a physical influence. It disrupted the π–π interaction of liquid crystals, causing the R-LCE and G-LCE samples to temporarily transition from the liquid crystal phase to the isotropic phase. The orientation of G-LCE was also temporarily disrupted during swelling. Simultaneously, the introduction of the alkaline catalyst nTBD activated the dynamic nature of the network, enabling the topological rearrangement through transesterification, thus exerting a chemical influence. As a result, the network structure of R-LCE and G-LCE samples permanently reverted to the initial state of the freshly prepared samples. In addition, the orientation of G-LCEs were permanently disrupted, effectively erasing the growth history.Rejuvenation for reprogramming G-LCEReprogramming actuation modes through rejuvenating G-LCEs has never been proposed previously, suggesting an alternative choice for designing soft actuators that combines reusability, recyclability and durability. As Fig. 6a shows, a soft actuator exhibited elongation/contraction deformation was obtained in the first self-growing procedure of a F-LCE. The G-LCE was swelled in solution containing nTBD of 3 mg mL−1 for 48 h to erase the original actuation mode. The obtained R-LCE was stretched and rotated for 540° with both ends fixed. The R-LCE regrew and arched at 30 °C to form the soft actuator with a twisting deformation mode (Fig. 6b).Fig. 6: Demonstration of reprogramming actuation modes.a Reprogramming process of G-LCE. b Actuation mode of G-LCE before and after reprogramming. Scale bars: 5 mm.Local regrowth via selective rejuvenationLocal control of self-growth was realized by selective rejuvenation. As shown in Fig. 7a, one half of the A-LCE was swelled in the solvent. After swelling, the R-LCE was stretched and fixed, only the half that swelled grew, while the other half remained at its original length. By dropping solution containing catalyst (3 mg mL−1) onto the A-LCE, the growth mode was locally controlled. The dropping parts grew, while the other parts remained unchanged after we stretched and fixed the sample (Fig. 7b).Fig. 7: Demonstration of local rejuvenation and controllable growth modes.a Half-rejuvenated A-LCE and its growth. b Control rejuvenation area through locally dropping solvents with nTBD. c Rejuvenation and growth of a “T” region in a square A-LCE. d Rejuvenation and growth of a “+” region in a square A-LCE. Scale bars: 5 mm.More sophisticated three-dimensional actuators were fabricated by rationally designing the rejuvenation region. As shown in Fig. 7c and d, there was a “T”-shaped region and a “+”-shaped region in the square frame, respectively. We rejuvenated the “T” and “+” in the middle while leaving the square frame unchanged by dropping solution containing catalyst (3 mg mL−1). Then the swelling part was stretched by 30% and fixed. The rejuvenated parts grew and arched, and eventually formed an actuator, enabling a reversible transition between a flat 2D structure and a stereoscopic 3D structure.
