SC I ENCE ADVANCES | R E S EARCH ART I C L E

APPL I ED SC I ENCES AND ENG INEER ING

1Department of Mechanical and Aerospace Engineering, University of California,San Diego, La Jolla, CA 92093, USA. 2Materials Science and Engineering Program,University of California, San Diego, La Jolla, CA 92093, USA.*Corresponding author. Email: shqcai@ucsd.edu

He et al., Sci. Adv. 2019;5 : eaax5746 11 October 2019

Copyright © 2019

The Authors, some

rights reserved;

exclusive licensee

American Association

for the Advancement

of Science. No claim to

originalU.S. Government

Works. Distributed

under a Creative

Commons Attribution

NonCommercial

License 4.0 (CC BY-NC).

Do

Electrically controlled liquid crystal elastomer–basedsoft tubular actuator with multimodal actuationQiguang He1, Zhijian Wang1, Yang Wang2, Adriane Minori1,Michael T. Tolley1,2, Shengqiang Cai1,2*

Soft tubular actuators can be widely found both in nature and in engineering applications. The benefits oftubular actuators include (i) multiple actuation modes such as contraction, bending, and expansion; (ii) facilefabrication from a planar sheet; and (iii) a large interior space for accommodating additional components or fortransporting fluids.Most recently developed soft tubular actuators are drivenbypneumatics, hydraulics, or tendons.Each of these actuation modes has limitations including complex fabrication, integration, and nonuniform strain.Here, we design and construct soft tubular actuators using an emerging artificial muscle material that can be easilypatterned with programmable strain: liquid crystal elastomer. Controlled by an externally applied electricalpotential, the tubular actuator can exhibit multidirectional bending as well as large (~40%) homogenous contrac-tion. Using multiple tubular actuators, we build a multifunctional soft gripper and an untethered soft robot.

wn

on February 19, 2021

http://advances.sciencemag.org/

loaded from

INTRODUCTIONIn nature as well as engineering applications, we can often find ac-tuators of cylindrical shape, for instance, the trunk of the elephant,the arms of the octopus, and the tube feet of the starfish. Concen-tric tubular surgical robots (1, 2) and endoscopes (3–5) are repre-sentative examples from biomedical engineering, which can exhibitmultimodal actuation. Those tubular actuators are usually made ofstiff materials, in which a gear-and-shaft system is used to achievetheir multimodal actuation.

Recent work has intensively explored soft continuum robots andhas demonstrated theirmany unique and attractive characteristics, suchas a large number of degrees of freedom and high biocompatibility(6–10). Notably, researchers have designed and fabricated varioussoft actuators with cylindrical shape to achieve various bioinspiredmovements, such as octopus-inspired robotic tentacles (11, 12),trunk-inspired manipulators (13), and worm-inspired architectures(14). However, most previously constructed soft actuators were eitherpneumatically or hydrodynamically driven, which often require abulky external control system, complex internal channels for diverseactuationmodes, and the prevention of fluid leakage (15–17). There isa clear need for stimuli-responsive material to construct soft actuators(18, 19), which has the potential to simplify fabrication and assemblyprocesses and reduce the complexity of the controlling.

Liquid crystal elastomer (LCE), a thermally driven actuating ma-terial that combines polymer network and liquid crystal mesogens, asshown in fig. S1 (A and B), has attracted much attention recently be-cause of its unique properties, including large (~40%) and reversibleactuation, high processability, and programmability (20–25). As thetemperature increases, liquid crystal mesogens transition from the ne-matic phase to the isotropic phase, leading to a notable and macro-scopic deformation in the material. A variety of LCE-based actuatingstructures have been designed and successfully fabricated, which areoften actuated by direct environmental heating (26, 27) or by lightthrough photothermal and photochemical effects (28–34). However,for most practical applications, electronically powered actuators pro-

vide notable convenience for system control and integration. A fewrecent studies have successfully integrated stretchable resistive heatersinto LCE (35–37), whose actuation can then be easily controlled byelectrical potential. Here, we describe the design and fabrication ofan LCE-based tubular actuator with multiple actuation modes andcontrolled by the externally applied electrical potential of several volts(1.0 to 3.0 V). To demonstrate the advantages of the soft tubular ac-tuator in potential applications, we further fabricate an untethered softrobot with on-board power source andmicrocontroller, mainly drivenby the LCE-based tubular actuators. These features are often highlydesired in constructing compact and multifunctional devices.

RESULTSFabrication of LCE-based tubular actuatorFigure 1A and fig. S2 show themain steps of fabricating an LCE-basedtubular actuator. We sandwiched three separate thin stretchable ser-pentine heating wires (as shown in fig. S3) between two layers of looselycross-linked LCE films (1 mm, each), as shown in steps (i) and (ii). Instep (iii), we made the tubular shape by rolling the sandwich-like struc-ture and stretching it by lp. The LCE tubular actuator could be finallyobtained once the entire structure was exposed to ultraviolet (UV) ir-radiation to fix the alignment of the liquid crystal mesogens inside.

In nature, animals such as mammals can contract muscles in dif-ferent parts in their bodies to achievemultimodal actuation. Similarly,the fabricated LCE-based tubular actuator contains three separateheating wires that can be controlled independently, as shown inFig. 1B. The electric current going through the metallic wires gener-ates Joule heating and increases the temperature of the adjacent LCE.Consequently, the heated LCE can contract in the longitudinal direc-tion. By applying an electrical potential onto different embeddedheating wires, we achieve various actuating modes. For example, ifonly one of the heating wires generates Joule heating, the tube canbend toward one direction, but if all three heating wires generateJoule heating, the tube can contract homogenously.

Electrical potential–driven actuation of an LCE thin filmTo better control the actuation of the tubular actuator, we first fab-ricated and characterized the actuation behavior of an LCE artificialmuscle film. The fabrication of LCE artificial muscle film is similar to

1 of 7

SC I ENCE ADVANCES | R E S EARCH ART I C L E

Dow

nloaded from

that of the LCE tubular actuator. A freestanding LCE artificial musclefilm contracted by 41% of its initial length as 3.0 V was applied, asshown in Fig. 2A and movie S1. When we turned off the electricalpotential, the temperature of LCE artificialmuscle gradually decreasedto room temperature, and the LCE artificial muscle recovered to itsoriginal shape within 4 min. To quantitatively characterize the actua-tion of the LCE-based artificial muscle, we define the actuation strainas e = (L − l)/L × 100%, where L and l are the lengths of LCE artificialmuscle in the initial and actuated states, respectively.

Thermal images (taken by FLIR E75-42 Advanced Thermal Imag-ing Camera) in fig. S4A showed the surface temperature of LCE arti-ficial muscle film during the heating process when we applied 3.0 V toit. The highest temperature on the surface increased from20° to 120°Cwithin 30 s, and the actuation strain of the LCE artificial musclereached 41%. We also noticed that nearly homogeneous temperaturedistribution could be generated on the surface of the artificial musclevia Joule heating and thermal diffusion after the 30 s of heating. If weapplied the electrical potential onto the heatingwires for a longer time,the surface temperature of the artificial muscle could be furtherincreased to 220°C, while the actuation strainmaintained its 41% con-traction. We had also shown, in fig. S1C, that the embedded heatingwires of serpentine shape have a negligible influence on the actuationbehavior of LCE films.

He et al., Sci. Adv. 2019;5 : eaax5746 11 October 2019

Next, we quantitatively characterized the actuation behaviors ofLCE artificial muscle when different electrical potentials were applied.In Fig. 2B, we showed the actuation strain of LCE artificial muscle ver-sus time with four different levels of electrical potential (1.0, 1.5, 2.0,and 3.0 V) for 360 s. For a given electrical potential, the actuationstrain of LCE artificial muscle increased at the beginning and thenreached a plateau value either because the temperature field in the ar-tificial muscle reached a steady state or because the actuationwas at itsmaximal value (41%). As we turned off the electrical potential, the

on February 19, 2021

http://advances.sciencemag.org/

Fig. 1. Design, fabrication, and operation principle of an LCE tubular actuator.(A) Fabrication steps of an LCE-based tubular actuator: (i) Three serpentineheating wires were sandwiched between two layers of loosely cross-linked LCEfilms. (ii) The sandwiched structure was compressed slightly to promote adhesion.(iii) A tube was made by rolling the thin film. (iv) An LCE tubular actuator wasobtained by stretching and then exposing the actuator under UV irradiation.(B) Principle of operation of LCE tubular actuators. Bending motion can be real-ized by applying an electrical potential to one of the heating wires (colored red);homogeneous contraction can be obtained by applying an electrical potential toall heating wires (colored red).

Fig. 2. Thermomechanical characterizations of LCE artificial muscle film.(A) Optical images of reversible actuation of the artificial muscle: The filmcontracted by 40% of its initial length when an electrical potential of 3.0 V wasapplied. (B) Plot of actuation strain of LCE artificial muscle versus time with dif-ferent applied electrical potential ranging from 1.0 to 3.0 V. For a given electricalpotential, actuation strain increased and then reached a plateau value (steadystate). When the electrical potential increased from 1.0 to 2.0 V, the actuationstrain at steady state increased from 15 to 41%. As the electrical potential furtherincreased from 2.0 to 3.0 V, the actuation strain of LCE artificial muscle at steadystate remained the same, while the response time (time for reaching the steadystate) decreased from 100 to 30 s. (C and D) Actuation stress of the artificialmuscle film (with fixed length) versus time with different electrical potentials rang-ing from 1.0 to 3.0 V for 30 s. The actuation stress increased from 0.05 to 0.35 MPaas the electrical potential was increased from 1.0 to 3.0 V. (E) An LCE artificialmuscle film could lift a load of 3.92 N (the stress was 0.312 MPa) by 38% of its initiallength. (F) Actuation strain of the LCE artificial muscle film versus time under threedifferent levels of applied stresses. As the load was increased, the time needed forthe LCE film to contract was almost unchanged, while the time needed for it torecover to its original length decreased. Inset: Maximal work density of LCE artificialmuscle under different applied stresses. Scale bars, 1 cm (A, C, and E). Photo credit:Qiguang He, University of California, San Diego.

2 of 7

SC I ENCE ADVANCES | R E S EARCH ART I C L E

on February 19, 2021

http://advances.sciencemag.org/

Dow

nloaded from

actuation strain decreased to zerowithin 240 s.Whenwe increased theelectrical potential from0.5 to 2.0 V, themaximal actuation strain rosefrom 15 to 41%. As we incremented the electrical potential furtherfrom 2.0 to 3.0 V, the maximal actuation strain maintained the samevalue (41%), but the response time decreased. It took about 30 s for theLCE artificial muscle to reach the maximum contraction (41%) when3.0 V was applied compared with 150 s while 2.0 V was applied. Asshown in fig. S4B, as we increased the electrical potential from 1.0 to2.0 V, the surface temperature of the LCE films in the steady state wasincreased from 55° to 120°C. Correspondingly, the actuation strain ofthe LCE artificial muscle was increased from 16 to 41%, as shown inFig. 2B. If we further increased the electrical potential from2.0 to 3.0Vand the maximal surface temperature increased from 120° to 220°C(fig. S4B), the actuation strain of LCE film stayed unchanged (Fig. 2B).We also showed that the performance of the LCE artificial muscle filmalmost remained unchanged after more than 50 and 100 cycles of actu-ation with an applied electrical potential of 3.0 and 2.0 V, respectively(fig. S5).Moreover, nodelamination betweenLCE layer andheatingwirewas observed in cyclic heating and cooling process in the experiments.

By fixing the length of the LCE artificial muscle film (Fig. 2C), westudied its actuation stress of fixed lengthwith different applied electricalpotentials ranging from 1.0 to 3.0 V for 30 s. For a given electricalpotential, the actuation stress increased fromzero to themaximumvaluewithin 30 s and dropped to zero when we turned off the electricalpotential. As the electrical potential increased from1.0 to 3.0V, themax-imal actuation stress increased from 0.05 to 0.35MPa, which is shown inFig. 2D. We noted that the typical actuating stress generated by skeletalmuscles of most mammals is also around 0.1 to 0.35 MPa (18, 38).

We also measured the work density of LCE artificial muscle film byapplying a constant load onto the LCE artificialmuscle film and an elec-trical potential of 3.0V (Fig. 2, E andF). In the experiment, wemeasuredthe contraction of the artificialmuscle filmwith five different loads rang-ing from 0.49 to 3.92 N. We calculated the mechanical work done bythe artificialmuscle filmbymultiplying themagnitude of the load by themaximal displacement of contraction, as shown in Fig. 2E. Themaximalwork density can be further obtained by dividing the mechanical workby the volume of the LCE artificial muscle film. As shown in the inset ofFig. 2F, themaximalwork density of the LCE artificialmuscle film couldreach as high as 150 kJ/m3when the applied stress was 0.31MPa, whichwas comparable to that of dielectric elastomermembrane andmamma-lian skeletal muscle but higher than that of ionic polymer metal com-posites and piezoelectric actuators (18).

Last, we would like to point out that, like other thermally respon-sive materials, the actuation bandwidth of LCE actuators dependsprimarily on the characteristic time of heat transfer in the material.The thermal conductivity of LCE is low, so their response time is rel-atively slow, as shown in Fig. 2 (B, D, and F). The most effective wayto reduce the response time is to decrease the size of the material or,more precisely, the characteristic size of the heating area. However,this modification could adversely affect other actuation performanceof the LCE actuators, such as the actuation force and strain. There-fore, an optimized design will certainly be needed to achieve a desir-able combination of actuation speed, strain, and stress for specificapplications.

Electrical potential–controlled multimodal actuation of atubular actuatorWe also characterized the multimodal actuation of an LCE tubularactuator. In Fig. 3 (A and B), we demonstrated that applying an

He et al., Sci. Adv. 2019;5 : eaax5746 11 October 2019

electrical potential onto one or two of the heating wires (same elec-trical potential for both heating wires) could generate bending ac-tuation. Depending on which heating wires we distributedly activated,the tube could bend toward six different directions (red color in Fig. 3Bindicates the activated heating wires). Similarly, by simultaneouslyapplying the electrical potential to all the heating wires, the entireLCE tube was heated up, resulting in a homogenous contraction.The LCE tubular actuator recovered to its original shape and geom-etry after the electrical potential was turned off.

On the basis of the measurement of the actuation strain on a singleLCE artificialmuscle thin film as described previously, we could quan-titatively predict the bending angle of the LCE tubular actuator, asshown in Fig. 3C. In the experiment, we applied different electricalpotentials to an LCE artificialmuscle thin film for 30 s and then turnedoff the electrical potentials. We measured the actuation strain of thethin film as a function of time for different electrical potentials. Theresults are shown in Fig. 3D, together with the contraction of the LCEtubular actuator whenwe applied the electrical potential onto all threeembedded heating wires. It can be seen that those two results are verysimilar to each other.

To predict the bending angle of the tubular actuator when we ap-plied the electrical potential only onto one of the heating wires, weadopted themodel developed for calculating the deflection of a beamcaused by inhomogeneous thermal expansion (39). As shown in fig. S6,when one of the three embedded heating wires was subjected to anelectrical potential, we assumed that one-third of the tubular actuator

Fig. 3. Multimodal actuation of an LCE-based tubular actuator. (A) Real andthermal images of an LCE tubular actuator during actuation: Activating one ortwo heating wires resulted in bending motion; activating all three heating wiresresulted in homogeneous contraction. (B) Different actuation modes of the tubu-lar actuator. Six directional bending were realized by activating one or twoheating wires. Contraction was achieved by simultaneously activating all threeheating wires. (C) Plot of bending angle of LCE tubular actuator versus time byapplying electrical potentials from 1.0 to 3.0 V. The heating time was 30 s, and thecooling time was 270 s. Solid lines represent theoretical prediction, and dots repre-sent experimental results. (D) Plot of contraction of LCE tubular actuator versustime through the activation of all three heating wires. The electrical potentialwas set to 3.0 V, and the time of electrical potential on was set to 30 s; the timeof electrical potential off was set to 270 s. Solid lines represent theoretical predic-tion, and dots represent experimental results. Scale bars, 1 cm (A and B). Photocredit: Qiguang He, University of California, San Diego.

3 of 7

SC I ENCE ADVANCES | R E S EARCH ART I C L E

on February 19, 2021

http://advances.sciencemag.org/

Dow

nloaded from

behaves just like the LCE artificial muscle thin film and that the othertwo-thirds of the tubular actuator acts like passive elastic material.Therefore, the bending angle of the tube can be computed as q ¼MTI l0, where l0 is the length of the tube; I ¼ p

64 E½ðdoÞ4 � ðdiÞ4�, knownas the bending stiffness of the tubewithE as themodulus of the LCE anddo and di as the outer and inner diameter of the tube, respectively;MT =∫ eTZdA as the thermal bending moment, with eT as the actuationstrain of the thin film andZ andA as denoted in fig. S6.With the input ofthe actuation strain eTmeasured for LCE thin film, as shown in Fig. 3D,the above equation allows us to predict the bending angle of the tubularactuator as a function of time. The prediction agrees very well with theexperimental measurements, shown in Fig. 3C. We observed that boththe length of the tube and the inner and outer diameters of the tube canbe measured. There is no fitting parameter for the prediction.

The position of the center of the unconstrained cross-sectionalarea of the tubular actuator is important to understand the actua-tion performance. We plotted the trajectory in fig. S7 (A and B), inwhich one or two heating wires were subjected to electrical potentials.The trajectory of the center point on the cross-sectional area of thetubular actuator when we applied in one wire an electrical potentialof 3.0 V for 30 s is shown in fig. S7A and movie S2. When the powersupply delivered the same electrical potential (3.0 V) for 20 s to twoheating wires, the trajectories of the actuator’s cross section have thedisplacements shown in fig. S7A and movie S3. When all threeheating wires had the same electrical potential (3.0 V) for 30 s, weobserved a homogeneous contraction, as shown in movie S4.

Because of the thermal diffusion, when we applied high electricalpotential (e.g., 3.0 V) to one or two heating wires, larger areas of thetube could be heated up if the heating time was too long (e.g., 360 s),which might induce unexpected geometrical distortion of the tube, asshown in fig. S8. To address this problem, we adopted the followingstrategy: We first applied a high electrical potential (V1 = 3.0 V) to theheating wires for 30 s to induce fast bending, and then the electricalpotential was reduced to a lower value (V2 = 1.5 V) for the rest of timeto maintain the deformation, as shown in fig. S8. We show the cyclicbending of the LCE tubular actuator in fig. S9, whenwe turned on andoff, repeatedly, the electrical potential into one or two heating wires.

Because of the internal heating mechanism of the tubular actua-tor, we further show that the LCE tubular actuator could also workeffectively in a water environment (fig. S10) with slightly increasedelectrical potential (6.0 V) due to the higher thermal conductivityand heat capacity of water as compared to air.

Soft gripper based on LCE-based tubular actuatorWe next built a soft gripper (Fig. 4) using the tubular actuator dem-onstrated previously as the main building block. As shown in Fig. 4A,we could construct an electronically controlled soft gripper by usingthree tubular actuators. Three tubular actuators were first attached to acircular plate, which was further connected to an LCE artificial musclefilm. We used a microcontroller to control the electrical potential ap-plied to the actuator and, thus, the deformation of each tubular actu-ator. By selectively activating the heating wires in each tubularactuator of the gripper, the gripper can grasp and lift a vial (Fig. 4Cand movie S5) and also twist its cap (Fig. 4B and movie S6) withoutadditional external control.

An untether soft robotWe next designed and fabricated the untethered robot using theLCE tubular actuators. We integrated a microcontroller (Arduino

He et al., Sci. Adv. 2019;5 : eaax5746 11 October 2019

Mini, 3.0 V), electrical components [metal oxide semiconductorfield-effect transistors (MOSFETs) and wires)], battery (lithium/polymer, 3.7 V), and LCE tubular actuators containing two separateheating wires (for the detailed characterizations, see fig. S7, C and D)with an acrylic plate to build an untethered robot shown in Fig. 5A.

In Fig. 5B, with automatically controlled LCE tubular actuators,the untethered robot could walk on a flat surface. The robot startedto walk by simultaneously activating a pair of its diagonal tubularactuators for 17 s to bend forward. After that, the other pair of tu-bular actuators was activated simultaneously for 18 s. In the last step,the electrical potential was turned off for all the tubular actuators for205 s, and all the actuators turned back to the original resting state.The displacement in one period was around 8.5 mm (see Fig. 5D).After 30 min of walking, the robot can move forward for a distanceof about one body length (8 cm), as shown in movie S7.

Figure 5C demonstrates three cycles of the untethered robot trans-porting an item on the top. In the experiment, we placed a cardboardon top of the four tubular actuators and a weight of 10 g on the plate.During the operation, the two front actuators first bent forward grad-ually using around 45 s and then recovered back to the straight stateaccompanied by contraction. During the recovery and contraction ofthe front actuators, the other two rear actuators bent forward also usingabout 45 s, which was followed by the state that the electrical potentialwas turned off for all the actuators. These repetitive actions enabled therobot to move the rigid plate and the weight on top of the plate forwardabout 5 cm within 15 min, as shown in Fig. 5D and movie S8.

The costs of transport (COTs) are estimated as 5.0 × 104 and 1.0 ×105 for the robot walking and transporting experiments, respectively.Compared with other untethered soft robots (40), the magnitude ofCOTs of the LCE-based untethered robot is very large because of thelow energy efficiency of thermally driven actuating materials.

DISCUSSIONHere, we have discussed the design and fabrication of an LCE-basedtubular actuator, which can exhibit multidirectional bending andhomogenous contraction as a low electrical potential is selectively

Fig. 4. Amultifunctional soft gripper composed of three LCE tubular actuators.(A) Schematic of the assembly of the soft gripper. (B) Soft gripper twisting the cap ofa vial. (C) Soft gripper grasping and lifting the vial (50 g). Scale bars, 1 cm (B and C).Photo credit: Qiguang He, University of California, San Diego.

4 of 7

SC I ENCE ADVANCES | R E S EARCH ART I C L E

on February 19, 2021

http://advances.sciencemag.org/

Dow

nloaded from

applied to heating wires embedded in the LCE. In addition, on thebasis of the tubular actuators, we have further constructed soft grip-pers that can grasp a vial and twist its cap. By integrating a battery, amicrocontroller, and the tubular actuators, we have built an untetheredrobot that can crawl on the ground and transport an item.

The distributed actuation of a tubular LCE caused the multimodalactuation of the soft tubular actuator developed here. Because of theembedded serpentine heating wires, we can fully activate the tubularactuator by applying an electrical potential. Although LCE has beenintensively explored in recent years todesign and fabricate various pre-viously unknown structures and devices (41–43), the multifunctionalrobot based on the tubular actuators, to our knowledge, is the first un-tethered robotic system only using LCE as the actuating material.

Previous work in electrically controlled soft structures such asmostdielectric elastomer actuators (DEAs) has shown much faster speedand higher energy efficiency compared to the LCE-based tubular ac-tuators developed here. However, the LCE-based tubular actuatordoes not require any stiff frames to maintain large prestretch forachieving large actuation (44). Furthermore, the tubular actuator alsouses a low-voltage power source (around three orders of magnitudelower than the electrical potential required to actuate DEAs). This fea-ture makes our tubular actuator compatible with most low-cost, com-mercially available electronic devices and batteries.

MATERIALS AND METHODSMaterials1,4-Bis-[4-(3-acryloyloxypropyloxy)benzoyloxy]-2-methylbenzene(RM257) (Wilshire Technologies; 95%), (2-hydroxyethoxy)-2-

He et al., Sci. Adv. 2019;5 : eaax5746 11 October 2019

methylpropiophenone (HHMP; Sigma-Aldrich; 98%), 2,2′-(ethylenedioxy)diethanethiol (EDDET; Sigma-Aldrich; 95%), pentaerythritol tetrakis(3-mercaptopropionate) (PETMP; Sigma-Aldrich; 95%), dipropyla-mine (DPA; Sigma-Aldrich; 98%), 3% hydrogen peroxide solution(Sigma-Aldrich), sodium iodide (Sigma-Aldrich; 99%), and sodiumthiosulfate pentahydrate (Fisher Scientific) were used as receivedwith-out further purification. The mechanical tests were conducted usingthe Instron Universal Testing Machine (5965 Dual Column TestingSystems; Instron) with 1000-N loading cell. The surface temperatureof LCE artificial muscle was measured by thermal imaging technique(FLIR E75-42 Advanced Thermal Imaging Camera).

Fabrication of loosely cross-linked LCE filmWe prepared the loosely cross-linked LCE following the previouslyreported work with little modification (45). Liquid crystal mesogenRM257 (10.957 g, 18.6mmol)was added into toluene, and themixturewas heated at 85°C. The photoinitiator HHMP (0.077 g, 0.3 mmol)was added into the mixture. Then, we added 3.076 g (16.9 mmol) ofspacer EDDET, 0.244 g (0.5mmol) of cross-linker PETMP, and 0.038 g(0.4mmol) of catalyst DPA into the solution. After stirring and degas-sing, we poured the mixture into the rectangular mold (1 mm thick-ness) and put it at room temperature for 24 hours. Then, we placed theloosely cross-linked LCE film into an 85°C oven for solvent evapora-tion. After that, we obtained the loosely cross-linked LCE film.

Fabrication of heating meshThe fabrication process of the heatingmesh is shown in fig. S3. First, a2-mm-thick layer of polyimide (PI) precursor was spin-coated onto the20-mm-thick layer of copper. After that, the one-side coated copper

Fig. 5. Electrically activated, untethered soft robot. (A) Schematic of the robot, mainly composed of a microcontroller, battery, and four LCE tubular actuators.(B) Frames from a video of the robot walking. Walking began from rest (all the actuators were in the deactivated state); then, a pair of diagonal tubular actuators wassimultaneously activated for 17 s to bend forward. After that, the other pair of tubular actuators was activated simultaneously for 18 s. In the last step, the electricalpotential to all actuators was turned off for 205 s, returning the actuators to their original states. (C) Frames from the video of the robot transporting an object on a plate ofcardboard. During the operation, the two front actuators first bent forward for 45 s and then straightened without touching the top plate. During the straightening of thetwo front actuators, the two rear actuators bent forward for approximately 45 s; then, the electrical potential to all the actuators was turned off. Repeating this sequenceenabled the robot to move the rigid plate and the weight on top forward about 5 cm within 15 min. (D) Plot of displacement of the untether robot (black curve) anddisplacement of the item on the top plate (red curve) versus time. Scale bars, 2 cm (B and C). Photo credit: Qiguang He, University of California, San Diego.

5 of 7

SC I ENCE ADVANCES | R E S EARCH ART I C L E

on February 19, 2021

http://advances.sciencemag.org/

Dow

nloaded from

was cured under 300°C oven under nitrogen protection. Then, weplaced the PI-coated copper with the copper layer up onto a substrate,as shown in step (i). The photoresist AZ1512was spin-coated onto thecopper layer. Photolithography and wet etching define patterns of thecopper wires (step ii). Later, we removed the residual photoresist withacetone; then, another layer of PI was spin-coated on the patternedcopper layer and baked in 300°C oven under nitrogen protection. Af-ter that, photolithography defined amask for dry etching process withCF4 and O2 through the PI layer. Last, we obtained the heating meshby removing the photoresist with acetone.

Fabrication of LCE tubular actuatorWe showed the fabrication process of the LCE tubular actuator infig. S2. First, we sandwiched three heating wires between two layersof loosely cross-linked LCE film (39 mm × 30 mm × 1 mm). Then,the whole structure was compressed, forming a sandwich-like struc-ture. In step (iii), we rolled the sandwich-like structure to a tube andstretched it along the longitudinal direction by lp = 1.8 to align theliquid crystal mesogens. Next, we obtained the LCE tubular actuatorby exposing under UV irradiation for 30 min, as shown in step (iv).Last, we soldered the electrical wires to the heating wires.

Preparation of soft gripperWe glued three fabricated LCE tubular actuators (embedded withthree heatingwires) to the acrylic plate. ArduinoMini (3.0V)was usedto control the on/off of the heating mesh, MOSFET (IRLB3034PBF)was used tomagnify the signal, and an electrical power source (Dr.MeterPS-305DM) was used to provide the electrical energy.

Preparation of untethered robotWe fixed four fabricated LCE tubular actuators (embedded with twoheating wires) to the acrylic plate and controlled the actuator usingArduinoMini (3.0 V). AMOSFET (IRLB3034PBF) was used tomag-nify the signal, anda lithium/polymerbattery (uxcellDC3.7V, 1300mAh)was used as the power source for the system.

Characterization of LCE artificial muscle/LCEtubular actuatorWe characterized the actuation strain of LCE artificial muscle by ana-lyzing the videos taken by the digital camera (Canon 60D) usingImageJ. The surface temperature of LCE artificial muscle duringactuation was measured by thermal imaging technique (FLIRE75-42 Advanced Thermal Imaging Camera). The actuation stressof LCE artificial muscle was measured using the Instron UniversalTesting System (5965 Dual Column Testing Systems; Instron) with1-kN loading cell. During the mechanical measurement, we fixedthe strain of LCE artificial muscle between two loading cells.

We characterized the actuation (bending and contraction) of theLCE tubular actuator by analyzing the videos taken by the digital cam-era (Canon 60D) using ImageJ. The surface temperature of the LCEtubular actuator during actuationwas characterized using thermal im-aging technique (FLIR E75-42 Advanced Thermal Imaging Camera).

SUPPLEMENTARY MATERIALSSupplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/5/10/eaax5746/DC1Fig. S1. Chemical component, fabrication process, and actuation performance of an LCE thin film.Fig. S2. Process of the fabrication of an LCE tubular actuator.Fig. S3. Design, fabrication, and morphology of heating wire.

He et al., Sci. Adv. 2019;5 : eaax5746 11 October 2019

Fig. S4. Temperature change of the LCE artificial muscle film during contraction and recovery.Fig. S5. Cyclic actuation test of the LCE artificial muscle film with applied stress of 0.078 MPa.Fig. S6. Theoretical prediction of bending angle of LCE tubular actuator by using the actuationof LCE artificial muscle thin film.Fig. S7. Characterizations of LCE tubular actuators.Fig. S8. The actuation behavior of LCE tubular actuator with different voltage controls.Fig. S9. Cyclic test of LCE tubular actuator with different heating wires.Fig. S10. The reversible actuation (bending motion and contraction) of LCE tubular actuator inthe water environment.Movie S1. Reversible actuation of LCE thin film actuator.Movie S2. Reversible bending actuation of LCE tubular actuator (activate one heating wire).Movie S3. Reversible bending actuation of LCE tubular actuator (activate two heating wires).Movie S4. Homogeneous contraction of LCE tubular actuator (activate three heating wires).Movie S5. Soft gripper grasps the vial.Movie S6. Soft gripper twists the cap.Movie S7. Untether soft robot walks on the ground.Movie S8. Untether soft robot manipulates the weight.

REFERENCES AND NOTES1. H. B. Gilbert, D. C. Rucker, R. J. Webster III, Concentric tube robots: The state of the art and

future directions, in Robotics Research (Springer, 2016), pp. 253–269.2. P. E. Dupont, J. Lock, B. Itkowitz, E. Butler, Design and control of concentric-tube robots.

IEEE Trans. Robot. 26, 209–225 (2010).3. A. J. Madhani, G. Niemeyer, J. K. Salisbury Jr., The black falcon: A teleoperated surgical

instrument for minimally invasive surgery, in Proceedings of the IEEE/RSJ InternationalConference on Intelligent Robots and Systems (IROS-98). Innovations in Theory, Practice andApplications (Cat. No. 98CH36190) (IEEE, 1998).

4. K. Xu, N. Simaan, An investigation of the intrinsic force sensing capabilities of continuumrobots. IEEE Trans. Robot. 24, 576–587 (2008).

5. A. Degani, H. Choset, B. Zubiate, T. Ota, M. Zenati, Highly articulated robotic probe forminimally invasive surgery. Conf. Proc. IEEE Eng. Med. Biol. Soc. , 4167–4172 (2006).

6. D. Rus, M. T. Tolley, Design, fabrication and control of soft robots. Nature 521, 467–475(2015).

7. M. T. Tolley, R. F. Shepherd, B. Mosadegh, K. C. Galloway, M. Wehner, M. Karpelson,R. J. Wood, G. M. Whitesides, A resilient, untethered soft robot. Soft Robot. 1, 213–223(2014).

8. F. Ilievski, A. D. Mazzeo, R. F. Shepherd, X. Chen, G. M. Whitesides, Soft robotics forchemists. Angew. Chem. 123, 1930–1935 (2011).

9. C. Christianson, N. N. Goldberg, D. D. Deheyn, S. Cai, M. T. Tolley, Translucent softrobots driven by frameless fluid electrode dielectric elastomer actuators. Sci. Robot. 3,eaat1893 (2018).

10. G. M. Whitesides, Soft robotics. Angew. Chem. Int. Ed. 57, 4258–4273 (2018).11. R. V. Martinez, J. L. Branch, C. R. Fish, L. Jin, R. F. Shepherd, R. M. D. Nunes, Z. Suo,

G. M. Whitesides, Robotic tentacles with three-dimensional mobility based on flexibleelastomers. Adv. Mater. 25, 205–212 (2013).

12. C. Laschi, M. Cianchetti, B. Mazzolai, L. Margheri, M. Follador, P. Dario, Soft robot arminspired by the octopus. Adv. Robot. 26, 709–727 (2012).

13. M. W. Hannan, I. D. Walker, Kinematics and the implementation of an elephant's trunkmanipulator and other continuum style robots. J. Robot. Syst. 20, 45–63 (2003).

14. S. Seok, C. D. Onal, K.-J. Cho, R. J. Wood, D. Rus, S. Kim, Meshworm: A peristaltic soft robotwith antagonistic nickel titanium coil actuators. IEEE/ASME Trans. Mechatronics 18,1485–1497 (2013).

15. A. D. Marchese, K. Komorowski, C. D. Onal, D. Rus, Design and control of a soft andcontinuously deformable 2d robotic manipulation system, in 2014 IEEE InternationalConference on Robotics And Automation (ICRA) (2014).

16. C. D. Onal, D. Rus, Autonomous undulatory serpentine locomotion utilizing bodydynamics of a fluidic soft robot. Bioinspir. Biomim. 8, 026003 (2013).

17. D. Drotman, M. Ishida, S. Jadhav, M. T. Tolley, Application-driven design of soft, 3-Dprinted, pneumatic actuators with bellows. IEEE/ASME Trans. Mechatronics 24, 78–87(2019).

18. S. M. Mirvakili, I. W. Hunter, Artificial muscles: Mechanisms, applications, and challenges.Adv. Mater. 30, 1704407 (2018).

19. Q. Li, Intelligent Stimuli-Responsive Materials: From Well-Defined Nanostructures toApplications (John Wiley & Sons, 2013).

20. Z. Wang, H. Tian, Q. He, S. Cai, Reprogrammable, reprocessible, and self-healable liquidcrystal elastomer with exchangeable disulfide bonds. ACS Appl. Mater. Interfaces 9,33119–33128 (2017).

21. C. Ohm, M. Brehmer, R. Zentel, Liquid crystalline elastomers as actuators and sensors.Adv. Mater. 22, 3366–3387 (2010).

22. T. J. White, D. J. Broer, Programmable and adaptive mechanics with liquid crystal polymernetworks and elastomers. Nat. Mater. 14, 1087–1098 (2015).

6 of 7

SC I ENCE ADVANCES | R E S EARCH ART I C L E

http://advances.sciencemag.or

Dow

nloaded from

23. Z. Pei, Y. Yang, Q. Chen, E. M. Terentjev, Y. Wei, Y. Ji, Mouldable liquid-crystallineelastomer actuators with exchangeable covalent bonds. Nat. Mater. 13, 36–41 (2014).

24. H. K. Bisoyi, Q. Li, Light-driven liquid crystalline materials: From photo-induced phasetransitions and property modulations to applications. Chem. Rev. 116, 15089–15166(2016).

25. S. W. Ula, N. A. Traugutt, R. H. Volpe, R. R. Patel, K. Yu, C. M. Yakacki, Liquid crystalelastomers: An introduction and review of emerging technologies. Liq. Cryst. Rev. 6,78–107 (2018).

26. X. Qian, Q. Chen, Y. Yang, Y. Xu, Z. Li, Z. Wang, Y. Wu, Y. Wei, Y. Ji, Untethered recyclabletubular actuators with versatile locomotion for soft continuum robots. Adv. Mater. 30,1801103 (2018).

27. C. Ahn, X. Liang, S. Cai, Inhomogeneous stretch induced patterning of molecularorientation in liquid crystal elastomers. Extreme Mech. Lett. 5, 30–36 (2015).

28. Z. Wang, K. Li, Q. He, S. Cai, A light-powered ultralight tensegrity robot with highdeformability and load capacity. Adv. Mater. 31, 1806849 (2019).

29. S. Palagi, A. G. Mark, S. Y. Reigh, K. Melde, T. Qiu, H. Zeng, C. Parmeggiani, D. Martella,A. Sanchez-Castillo, N. Kapernaum, F. Giesselmann, D. S. Wiersma, E. Lauga, P. Fischer,Structured light enables biomimetic swimming and versatile locomotion ofphotoresponsive soft microrobots. Nat. Mater. 15, 647–653 (2016).

30. A. H. Gelebart, D. Jan Mulder, M. Varga, A. Konya, G. Vantomme, E. W. Meijer,R. L. B. Selinger, D. J. Broer, Making waves in a photoactive polymer film. Nature 546,632–636 (2017).

31. H. Tian, Z. Wang, Y. Chen, J. Shao, T. Gao, S. Cai, Polydopamine-coated main-chain liquidcrystal elastomer as optically driven artificial muscle. ACS Appl. Mater. Interfaces 10,8307–8316 (2018).

32. L. Liu, M.-H. Liu, L.-L. Deng, B.-P. Lin, H. Yang, Near-infrared chromophore functionalizedsoft actuator with ultrafast photoresponsive speed and superior mechanical property.J. Am. Chem. Soc. 139, 11333–11336 (2017).

33. Z. Pei, Y. Yang, Q. Chen, Y. Wei, Y. Ji, Regional shape control of strategically assembledmultishape memory vitrimers. Adv. Mater. 28, 156–160 (2016).

34. H. K. Bisoyi, A. M. Urbas, Q. Li, Soft materials driven by photothermal effect and theirapplications. Adv. Opt. Mater. 6, 1800458 (2018).

35. S. Schuhladen, F. Preller, R. Rix, S. Petsch, R. Zentel, H. Zappe, Iris-like tunable apertureemploying liquid-crystal elastomers. Adv. Mater. 26, 7247–7251 (2014).

36. S. Petsch, R. Rix, P. Reith, B. Khatri, S. Schuhladen, D. Ruh, R. Zentel, H. Zappe,A thermotropic liquid crystal elastomer micro-actuator with integrated deformablemicro-heater, in IEEE 27th International Conference on Micro Electro Mechanical Systems(MEMS), 2014, 6765788 (2014).

37. C. Wang, K. Sim, J. Chen, H. Kim, Z. Rao, Y. Li, W. Chen, J. Song, R. Verduzco, C. Yu, Softultrathin electronics innervated adaptive fully soft robots. Adv. Mater. 30, 1706695 (2018).

He et al., Sci. Adv. 2019;5 : eaax5746 11 October 2019

38. Q. He, Z. Wang, Z. Song, S. Cai, Bioinspired design of vascular artificial muscle.Adv. Mater. Technol. 4, 1800244 (2019).

39. R. Clark, The bending and stretching of plates (E. H. Mansfield). SIAM Rev. 6, 468 (1964).40. W. C. Messner, J. Paik, R. Shepherd, S. Kim, B. A. Trimmer, Energy for biomimetic robots:

Challenges and solutions. Soft Robot. 1, 106–109 (2014).41. Z. Wang, W. Fan, Q. He, Y. Wang, X. Liang, S. Cai, A simple and robust way

towards reversible mechanochromism: Using liquid crystal elastomer as a mask.Extreme Mech. Lett. 11, 42–48 (2017).

42. H. Zeng, O. M. Wani, P. Wasylczyk, R. Kaczmarek, A. Priimagi, Self-regulating iris based onlight-actuated liquid crystal elastomer. Adv. Mater. 29, 1701814 (2017).

43. C. Yuan, D. J. Roach, C. K. Dunn, Q. Mu, X. Kuang, C. M. Yakacki, T. J. Wang, K. Yu, H. J. Qi,3D printed reversible shape changing soft actuators assisted by liquid crystal elastomers.Soft Matter 13, 5558–5568 (2017).

44. G. Gu, J. Zou, R. Zhao, X. Zhao, X. Zhu, Soft wall-climbing robots. Sci. Robot. 3, eaat2874(2018).

45. C. Yakacki, M. Saed, D. P. Nair, T. Gong, S. M. Reed, C. N. Bowman, Tailorable andprogrammable liquid-crystalline elastomers using a two-stage thiol–acrylate reaction.RSC Adv. 5, 18997–19001 (2015).

Acknowledgments: We thank Q. Wang for helping with the thermal measurement andacquiring IR images. Funding: The authors acknowledge the support from ONR through grantno. N00014-17-1-2062 and the NSF through grant no. CMMI-1554212. This work wasperformed, in part, at the San Diego Nanotechnology Infrastructure (SDNI) of UCSD, a memberof the National Nanotechnology Coordinated Infrastructure (NNCI), which is supported by theNSF (grant no. ECCS-1542148). Author contributions: Q.H., Z.W., and S.C. conceived theresearch. Q.H., Z.W., and Y.W. performed the experiments and analyzed the experimentalresults. Q.H., A.M., M.T.T., and S.C. composed the manuscript. All authors reviewed themanuscript. Competing interests: The authors declare that they have no competing interests.Data and materials availability: All data needed to evaluate the conclusions in the paperare present in the paper and/or the Supplementary Materials. Additional data related to thispaper may be requested from the authors.

Submitted 16 April 2019Accepted 14 September 2019Published 11 October 201910.1126/sciadv.aax5746

Citation: Q. He, Z. Wang, Y. Wang, A. Minori, M. T. Tolley, S. Cai, Electrically controlled liquidcrystal elastomer–based soft tubular actuator with multimodal actuation. Sci. Adv. 5, eaax5746(2019).

og/

7 of 7

n February 19, 2021

actuationbased soft tubular actuator with multimodal−Electrically controlled liquid crystal elastomer

Qiguang He, Zhijian Wang, Yang Wang, Adriane Minori, Michael T. Tolley and Shengqiang Cai

DOI: 10.1126/sciadv.aax5746 (10), eaax5746.5Sci Adv 

ARTICLE TOOLS http://advances.sciencemag.org/content/5/10/eaax5746

MATERIALSSUPPLEMENTARY http://advances.sciencemag.org/content/suppl/2019/10/07/5.10.eaax5746.DC1

REFERENCES

http://advances.sciencemag.org/content/5/10/eaax5746#BIBLThis article cites 39 articles, 0 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.Science AdvancesYork Avenue NW, Washington, DC 20005. The title (ISSN 2375-2548) is published by the American Association for the Advancement of Science, 1200 NewScience Advances

License 4.0 (CC BY-NC).Science. No claim to original U.S. Government Works. Distributed under a Creative Commons Attribution NonCommercial Copyright © 2019 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of

on February 19, 2021

http://advances.sciencemag.org/

Dow

nloaded from