Crosslinking-determined mechanical-response for desynchronized LCN actuators. The LCN actuator, capable of executing two-stage shape morphing towards opposite directions on only stimulation on or off, is fabricated based on the concept of “desynchronized actuation”. It means that the two sides of a LCN strip start to deform at different temperatures and impose different strains. To validate this strategy, the material used in this work is a main-chain liquid crystalline polymer containing biphenyl groups as mesogens and the cinnamyl groups as photocrosslinkable moieties (Fig. 1c, synthesis procedures, previously reported in ref 21, are detailed in Supplementary Information). The liquid crystalline polymer film (typically 0.35 mm in thick if not otherwise stated) was given a planar, uniaxial alignment of mesogens by mechanical stretching to 400% at 58 oC (LC phase). After being irradiated with UV light (320 nm, 180 mW/cm2), the polymer was photocrosslinked through the generated cinnamyl dimers, thereby locking the oriented mesogens for the reversible deformation. Interestingly, the crosslinking density can affect the LCN’s order-disorder phase transition and further influence the actuation behavior. Fig. 1d shows the DSC curves of the uniaxially aligned LCN of which both sides were crosslinked for 20 min, 60 min and 120 min, respectively. As the crosslinking time increases, the LC-isotropic phase transition temperature range widens while the peak position shows no significant change. The broader phase transition of the more crosslinked LCN also corresponds to a wider actuation temperature range, which can be reflected by the thermally induced reversible strain. With the three strips subjected to the same cooling and heating run (3 oC/min), their length changes were recorded using a dynamic mechanical analyzer in the iso-stress mode (Fig. 1e). In all cases, the LCN strip extends on cooling into the LC phase and then contracts on heating into the isotropic state. However, the changing crosslinking gives rise to notably different actuation behavior. The LCN crosslinked for 20 min on each side shows lower mid-point actuation temperatures on cooling (50 oC) and on heating (72 oC), determined as the temperature at which 50% of the reversible strain occurs, while the LCN crosslinked for 120 min on each side shows higher mid-point actuation temperatures (55 oC on cooling and 78 oC on heating). In addition to the actuation temperature range difference, less crosslinked LCN shows larger reversible deformation degree41,42, as can be seen from the plateau strain in Fig. 1e, being 64%, 55% and 48% for LCNs crosslinked for 20 min, 60 min and 120 min on both sides, respectively (the measurement is detailed in Supplementary Methods). As shown below, such differences in actuation temperature and deformation degree enable the desynchronized actuation leading to the deformation reversal behavior of LCN actuators.
Deformation reversal behavior. Firstly, the deformation-reversal actuators with bending-based motions were fabricated by stretching the LCNs to 400% (draw ratio λ) followed by crosslinking both sides of the film for different times, thus crosslinking degrees. The prepared actuators are coded as LCN1(X,Y) (X and Y represent the crosslinking time of side A and side B, respectively, and the subscript 1 denotes the first series of samples used for varying the crosslinking dissymmetry of the two sides while keeping a fixed λ at 400%). Fig. 2a and Supplementary Movie 1 show a typical deformation reversal process of LCN1(20,120) during cooling from 75 to 40 oC, which can be explained by the crosslinking-dependent actuation behaviors shown in Fig. 1e. In isotropic phase, the film exhibits a bent shape toward side A because the lightly crosslinked side A contracts more than the highly crosslinked side B. On cooling from 75 to 57 oC, the highly crosslinked side B elongates while the lightly crosslinked side A shows no obvious deformation, which causes the actuator to curl to side A and generates a coiled structure (deformation I). When the temperature drops below 57 oC, the bending direction of the sample reverses to side B as the extension of the side A proceeds with a higher strain (deformation II). Hence, in the cooling process, the LCN1(20,120) undergoes both deformation I and II stages having opposite bending directions and realizes the transformation among three different shapes.
To investigate the factors that dictate the deformation reversal behavior, we prepared different samples with fixed crosslinking time of side B (120 min) and variable crosslinking time of side A (0, 20, 40 and 60 min). The change in bending angle of the LCNs as a function of temperature is shown in Fig. 2b. As expected, since the uncrosslinked side shows no reversible deformation, LCN1(0,120) can only bend unidirectionally during cooling. In contrast, the three samples crosslinked on both sides all exhibit deformation reversal behavior, i.e., bidirectional bending. As the crosslinking-time difference increases, the deformation reversal behavior becomes more pronounced with larger changes in bending angle during deformation I (|dαI|) and deformation II (|dαII|) (Fig. 2d). Furthermore, accompanying the bending-reversed bending process on cooling, all samples crosslinked on both sides extended about 50% of their length in isotropic phase (Fig. 2c). Further expanding the combinations of crosslinking times for side A and side B of the LCN (see Supplementary Fig. 1-3), a graph showing the influence of the crosslinking time on the actuation behavior was obtained, on which the largest bending angle changes in the two directions, dαI and dαII, are listed for each actuator (Fig. 2e). As can be seen, to obtain larger |dαI| and |dαII| and more significant deformation reversal behavior (see the colored region in Fig. 2e), the crosslinking times of the two sides should be adjusted to meet two criteria: 1) large difference in crosslinking time, and 2) effective actuation via sufficient crosslinking on both sides of the film. Therefore, the optimal crosslinking condition shown in the map is 20 min/120 min on the film’s two sides. The area near the diagonal dashed line indicates insufficient crosslinking-time difference; the area close to while at the far end of either coordinate axis (meaning one side highly crosslinked while the other barely crosslinked) cannot meet the second criterion; the area near the origin of the coordinate could fail to satisfy both criteria. What’s more, other factors, such as film thickness, stretching temperature and draw ratio (λ, lager than 400%), did not show a decisive effect on the deformation reversal behavior (Supplementary Fig. 4). Moreover, as revealed by the X-ray diffraction analysis (Supplementary Fig. 5), the LCN used possesses a smectic phase. For comparison, we synthesized a new LCN with only a nematic phase (Supplementary Fig. 6, synthesis procedures are detailed in Supplementary Methods). The nematic LCN actuator also exhibits deformation reversal behavior (see Supplementary Fig. 7), which means that this interesting phenomenon is not only for smectic LCNs.
Enhancing deformation reversal capability. As demonstrated above, by tuning the crosslinking times for both sides of a uniaxially stretched LCN, we can control the direction and extent of the deformation reversal behavior, corresponding to the sign and absolute value of dα, respectively. However, the obtained shape change conforms to either α1>0, dαI>0, dαII<0 (if X<Y), or α1<0, dαI<0, dαII>0 (if X>Y). This exposes the limitation of adjusting the deformation reversal behavior only by changing the crosslinking times. This is because crosslinking not only determines the signs of dαI and dαII (the shape change direction), but also affects the signs of α1 (the shape in isotropic phase). In this regard, we further enriched the attainable deformation reversal behaviors of our LCN films by not only generating crosslinking difference, but also creating draw-ratio difference on both sides of the LCNs. Typically, a strip was stretched to a draw ratio and one side was exposed to UV light for photocrosslinking; then the strip was stretched again to a larger draw ratio and the other side was photocrosslinked. In other words, the draw ratio on one side indicates the strain at which that side is photocrosslinked, though the final strain on both side is the same for the LCN strip. As shown in Fig. 3a, the crosslinking times of side A and B for all samples were fixed at 20 min and 120 min, but for the draw ratio, the λ of side B was fixed as 400% for all samples, while that on side A varied from 0% to 1000%. The obtained actuators are named as LCN2(M,N) (M and N represent the λ of side A and B respectively, and the subscript 2 denotes this second series of actuators). As seen from the photographs in Fig. 3a, we achieved different types of deformation reversal behaviors. For example, LCN2(700%,400%) can complete a -360o to 360o to -360o deformation mode during one cooling run; LCN2(1000%,400%) undergoes ring to flat to ring two-stage deformation. These LCN actuators all show deformation where α1<0, dαI>0, dαII<0. It is easy to notice that the α1 of LCNs with λ of side A greater than 400% are shifted to negative values, indicating that, as λ increases, the shape in isotropic phase can be changed from rolling toward side A to curling toward side B (Fig. 3b). This is understandable because a higher λ results in a longer LCN actuator. Concurrently, high bending angle changes (large |dαI| and |dαII|) were achieved. Among them, the |dαI| and |dαII| of LCN2(540%,400%) reach up to 770o and 735o respectively (Supplementary Fig. 8a). In comparison, the samples with low λ of side A show no or only slight deformation reversal behavior due to the lack of good mesogen alignment which leads to ineffective/insufficient actuation on side A and an unmet second criterion (see Fig. 3b and Supplementary Fig. 8). Therefore, the above results demonstrate effective enhancement and tuning of the deformation reversal behavior through adjusting the λ of the two sides prior to crosslinking of a monolithic LCN film. A reasonable assumption is that the deformation reversal behavior can be regulated on-demand by fine-tuning these two processing factors (crosslinking time and λ), generating a large variety of shape change modes. More interestingly, when encoding the asymmetrical crosslinking and λ on the two halves of the LCN along its width direction instead of its two sides along the thickness, controllable in-plane bending and bending reversal can also be readily realized (see Supplementary Fig. 9).
Not limited to simple bending-based deformation reversal behavior, we also achieved twisting-based deformation reversal using the LCN strip with tilted LC orientation relative to its long axis. As shown in Fig. 3c, the LCN1(60,120) having its mesogens aligned -45o with respect to the principal axis of the strip shows a slightly twisted form in isotropic phase, which then winds up and unwinds in the following cooling process. Furthermore, Fig. 3d presents a contrary deformation reversal behavior. The film was prepared by firstly crosslinking a LCN strip (λ=400%, +45o mesogen alignment) for 20 min and 120 min on side A and B respectively, then stretching the film to double its length and further crosslinking the side A for 20 min. The resulting LCN displays a helical shape in isotropic phase, with the highly crosslinked side as the inner of the helix. Upon cooling, the helical LCN first unwinds and then winds up, showing opposite deformation trend compared with that in Fig. 3c.
Photomobility of deformation-reversal actuators. The central interest of the deformation reversal resides in the fact that the number of bidirectional shape switch can be doubled under one stimulation on/off cycle. Since repeated shape switch under repeated on/off stimulation is the basis of moving LCN soft robots, this feature means that the number of driving stimulation on/off cycles may be reduced, which may simplify the control in many ways. Therefore, on the basis of the diverse forms of the attainable deformation reversal behaviors, we further went on to demonstrate that these thermal responsive shape changes can also be controlled by light. To this end, thin layers of polydopamine (PDA) were coated on the crosslinked LCNs to serve as photothermal agents. The PDA-coated LCN film showed unchanged thermal-induced actuation behavior (see Supplementary Fig. 10), while possessing similar but more rapid light-driven deformation reversal behaviors based on the photothermal effect, regardless of which side is exposed to the light source (see Supplementary Fig. 11-12). In contrast to the conventional light-driven actuation, our LCN actuators can bend and unbend (or unbend and bend) twice in one light on/off cycle (see Supplementary Fig. 13 and Supplementary Movie 5). As light allows remote and convenient control of the fast and reversible deformation, light-guided locomotion based on deformation reversal behavior has been explored. Shown in Fig. 4a is a miniature walker of 3.5 mm in length, which is made of PDA-coated LCN1(20,120). The sample is initially flat with the highly crosslinked side facing upward. After exposure to light (30.5 mW/mm2) for ~2 s, the film arches up with the right side as the stationary point; further exposure for ~1 s, the film flattens down with the left side as the stationary point. As such, the film already takes a step forward with the light spot on it. Ceasing the light, the film cools down and re-arches up with the right side as the stationary point, followed by flattening down as it cools further. In this way, the film can implement two steps forward in one light on/off cycle. Due to the doubled number of steps, the microwalker can move at a speed of ~ 2 body length/min even slipping occurs during locomotion. Combining photo-patterning with asymmetrical crosslinking for deformation reversal behavior, we further achieved a wave-shaped LCN walker that can move in-plane with one edge on the substrate without “body” lifting in air. (Fig. 4b). The LCN, prepared to have three sections with alternating highly crosslinked sides (120 min, 400%) and lightly crosslinked sides (20 min, 400%), can first fold into a compact accordion shape that then expands into a loose wave under continuous light irradiation; upon retrieving the light, the loose wave tights up into an accordion shape and further expands into a large wave shape, thus completing one shape change cycle (see Supplementary movie 7 for the shape change process). When this actuator is obliquely exposed to light on/off cycles from one side as shown in Fig. 4b (light of 16.8 mW/mm2, shining from ~ 25o relative to the substrate surface), it can walk with the two end sections performing deformation reversal to adjust the friction during motion, while its middle part, being self-shadowed after curling up, displays only unidirectional deformation (speed: 8 mm/min, see Supplementary Fig. 14 and Supplementary movie 7). It should be emphasized that, unlike laser scanning process where only part of the actuator is activated at any point in time, the light is evenly applied to the actuator in the present case. Therefore, there is no need to precisely position the light spot on the micro-scale device, which is beneficial to light-fueled locomotion43.
Multimodal light-driven locomotion of a single deformation-reversal actuator. The deformation reversal behavior also imparts light-fueled LCN soft robot with unprecedented versatility in motion. This is because the entire shape change process with doubled number of shape switches can be divided into many stages by simply adjusting the time period of light-on and light-off, corresponding to different actuation temperature oscillation profiles. As a result, different locomotion modes can readily be achieved using the same LCN actuator. As a proof of concept, we were able to obtain five types of motion patterns using a single actuator prepared as illustrated in Fig. 5a: one side of the strip was crosslinked for 120 min at λ=400%, it was then further stretched to λ=540% and had the two halves of the other side crosslinked for 20 min and 40 min respectively. When exposed to light, this actuator can evolve from a coil shape to an S shape and then to a C shape, while when light is off, it transforms from the C shape to the S shape and then to the coil shape (Supplementary Fig. 15a). The reversible shape change can be made to occur between different stages of the complex shape morphing process to realize different locomotion modes. For the sake of clarity, only three modes are shown in Fig. 5 and Supplementary Movie 8 (2 others in Supplementary Information). In Mode I (Fig. 5b), the actuator is subjected to repeated entire shape change process (coil-S-C) under sufficiently long light-on and light-off times (light of 16.8 mW/mm2), which results in its motion through body rotation (visible from the changing state at the end of each light on/off cycle, at a speed of 9 mm/min). In mode II (Fig. 5c), the light-on and light-off times are shortened to enable the reversible shape switch only between coil and the S shape (light of 16.8 mW/mm2). Under this condition, the actuator moves through the unidirectional uncurling (light on) and curling (light off) movements, at a faster speed of 31 mm/min. In Mode III (Fig. 5d), the light irradiation condition is set to generate the back-and-forth switch between the shape C (deformation II) and an intermediate shape during deformation I (after obtaining shape C, light-off time is controlled to prevent the complete shape restoration during cooling). In this case, the imbalance in shape evolution could be enhanced by the higher light intensity 23.5 mW/mm2 used and the incomplete shape change, which causes the actuator to advance by flipping along its long axis, with an intermediate locomotion speed (21 mm/min). Moreover, when the actuator lying on its edge flips under light irradiation and has its planar side parallel to the surface, it can also rotate (Mode IV shown in Supplementary Fig. 16) or walk (Mode V shown in Supplementary Fig. 17) when the entire or partial deformation process is activated, respectively (see also Supplementary Movie 9).