Ultrastrong and Highly Sensitive Fiber Microactuators Constructed by Force‐Reeled Silks

Abstract Fiber microactuators are interesting in wide variety of emerging fields, including artificial muscles, biosensors, and wearable devices. In the present study, a robust, fast‐responsive, and humidity‐induced silk fiber microactuator is developed by integrating force‐reeling and yarn‐spinning techniques. The shape gradient, together with hierarchical rough surface, allows these silk fiber microactuators to respond rapidly to humidity. The silk fiber microactuator can reach maximum rotation speed of 6179.3° s−1 in 4.8 s. Such a response speed (1030 rotations per minute) is comparable with the most advanced microactuators. Moreover, this microactuator generates 2.1 W kg−1 of average actuation power, which is twice higher than fiber actuators constructed by cocoon silks. The actuating powers of silk fiber microactuators can be precisely programmed by controlling the number of fibers used. Lastly, theory predicts the observed performance merits of silk fiber microactuators toward inspiring the rational design of water‐induced microactuators.


DOI: 10.1002/advs.201902743
outstanding actuation characteristic pro motes the spider web to maintain its geometric configuration in widely vari able environmental humidity and even allows the spider to perceive the imposed external loads on the network. [4] How ever, although spider silks have so many advantages for actuator applications, their practical application in the industry is not feasible yet. [5,6] Because spiders are nondo mestic creatures and only produce small amounts of silk. [5] Silkworm silk fibers have interesting and unique structures and superior mechanical properties, [7] as shown in Figure 1. These fibers are the potential substitutions for spider silks for micro actuator fabrication. However, the common used cocoon silk fibers (CSFs) are usu ally inferior to spider silks in both the structural uniformity and mechanical properties. [8,9] It may be attributed to defects and irregular changes in the structure of the natural cocoon silks, which orig inates from the movement of the silkworm head ( Figure S1, Supporting Information). On the other hand, both the structure uniformity and the mechanical performance of silkworm silk fibers can be optimized through forcereeling, a technique that is applied as an alternative to the natural cocooning, where the silk fiber is directly harvested from silkworm spinneret. Indeed, the silk fibers used for the structureandproperty investigation are often collected by the forcereeling. [8][9][10][11][12] However, the prac tical application of these forcereeled silk fibers (FRSFs) has not achieved yet. Because there is still no highly efficient way to produce continuous and uniform fibers in large scales. [11,12] Herein, a speedcontrollable forcereeling strategy is devel oped to harvest the uniform silk fiber from Antheraea pernyi (A. pernyi) silkworm with the continuous spinning length larger than 1 km. Application of FRSFs exhibits a unique tradeoff in the mechanical strength and the extensibility, resulting in the mechanical toughness almost twice higher than that of CSFs, which is even comparable with that of the spider silk. The yarn spinning technique can be employed to assemble FRSFs into doublehelical microactuators with the programmable actua tion power, ranging from 0.77 to 2.1 W kg −1 . This indicates the torsional power that each kilogram of the silk fiber can provide. Moreover, the detailed actuation mechanism of the FRSF microactuators is unveiled: the rough surface and the shape gradients trigger and promote the infiltration of water on the microactuator so that they can lead to the swelling and contrac tion of FRSFs. This synergistic effect spontaneously untwists Fiber microactuators are interesting in wide variety of emerging fields, including artificial muscles, biosensors, and wearable devices. In the present study, a robust, fast-responsive, and humidity-induced silk fiber microactuator is developed by integrating force-reeling and yarn-spinning techniques. The shape gradient, together with hierarchical rough surface, allows these silk fiber microactuators to respond rapidly to humidity. The silk fiber microactuator can reach maximum rotation speed of 6179.3° s −1 in 4.8 s. Such a response speed (1030 rotations per minute) is comparable with the most advanced microactuators. Moreover, this microactuator generates 2.1 W kg −1 of average actuation power, which is twice higher than fiber actuators constructed by cocoon silks. The actuating powers of silk fiber microactuators can be precisely programmed by controlling the number of fibers used. Lastly, theory predicts the observed performance merits of silk fiber microactuators toward inspiring the rational design of water-induced microactuators.

Introduction
Welldesigned fiber microactuators are widely found in nature. For example, spider dragline silk is an extraordinary fiber that surpasses the majority of conventional materials in mechanical characteristics such as the mechanical toughness. Moreover, spider dragline silk has proven to be an outstanding torsional actuator. [1] It exhibits unique humidityinduced actuation behavior with a torsional deformation more than 300° mm −1 . This value is thousands of times higher than that of other conventional fiber microactuators, such as conducting polymer actuators (0.01° mm −1 ) and shapememory alloy fibers micro actuators (0.15° mm −1 ). [1] In fact, the humidityinduced actua tion of the spider dragline silk is even higher than that of the state of the art of the carbon nanotubes (CNTs) microactuators (250° mm −1 ), [2,3] which is powered by the electricity. This the single FRSF yarn and maintains the overall twisting bal ance of the microactuator. As a pristine protein component, FRSF microactuators have broad application prospects in bio medical and smart textile fields, including the artificial muscle, biosensor, microengine, and the smart textile. Furthermore, reviewing the literature confirms that silk fibers are favorable in supporting tissue formation in vitro and in modeling of time dependent cell responses. [13][14][15]

Force-Reeling of A. Pernyi Silkworm Silk
In the present study, the A. pernyi silkworm is selected instead of the commonly used Bombyx mori (B. mori) silkworm for the forcereeling. This selection is done in accordance with the following interpretations. First, the primary structure of the A. pernyi silk is similar to that of the spider silk. [16,17] In other words, both silk proteins consist of the highly repetitive poly (alanine) and glycinerich domains. [18] Second, A. pernyi silks are widely available, while lowcost and sustainable. [19,20] More spe cifically, the global annual production of A. pernyi cocoons is up to 60 000 tons, [9] and no specific feeding environment is required for A. pernyi silkworms. [9,19] Last but not least, in screening experiments, we found that the breeding of the A. pernyi silk worm was much easier than that of the B. mori silkworm during the forcereeling. [9] The B. mori silkworm instinctually uses facial palps to break the thread and resist the reeling force, which leads to significant fibertofiber variability. By contrast, highly uni form FRSFs can be harvested from A. pernyi silkworm through continuously reeling with a speed varying from 5 to 30 mm s −1 for more than 8 h (see Movie S1, Supporting Information). It should be indicated that the continuous FRSFs can reach up to kilometerscale at the reeling speed of 20 mm s −1 . This speed is remarkable because the obtained reeling rate is almost 2.5 times higher than the silkworm spinning speed (≈8 mm s −1 ). [21,22] The obtained fibers are highly lustrous and have uniform diameters (Figure 1b,c). Moreover, they are characterized by a strong birefringence when observed under the crosspolarized light ( Figure S2a, Supporting Information). These features are also in clear contrast with the CSFs or silk fiber spun by silkworm on a plane substrate (plane silk fiber, hereafter called the PSF). The CSF is reluster in color and has many defects on fiber surface ( Figure S2b, Supporting Information).

Force-Reeling of A. Pernyi Silkworm Silk
Mechanical characteristics are of significant importance for the practical application of an actuator. Therefore, tensile tests are initially carried out to evaluate the mechanical properties of FRSFs. Since the crosssections of FRSFs are not circular ( Figure S3, Supporting Information), the crosssectional area of the single fiber instead of the diameter is utilized to calculate the corresponding tensile stress. [23] The mechanical character istics of the FRSFs are significantly improved and maintain the elegant tradeoff in strength, modulus, and the extensibility in comparison with those of the A. pernyi CSF and the A. pernyi PSF, respectively (Figure 2a-c). For example, the strength and Young's modulus of the FRSF are 571 ± 97 MPa and 11 ± 2 GPa, respectively. These characteristics are almost 1.6 times higher than that of the CSF with strength and modulus of 359 ± 83 MPa and 7 ± 2 GPa, respectively. Furthermore, the Adv. Sci. 2020, 7,1902743  FRSF is tougher than most of the natural and synthetic mate rials and is comparable with N. clavipes spider silk with the strain of 31 ± 6% and the toughness of 131 ± 77 MJ m −3 . [18,24,25] The tradeoff of the strength, modulus, and the toughness in the FRSF is better than those for most of the synthetic fibers ( Figure 2d). For example, although the strength and the Young's modulus of Kevlar fiber approach to 3.6 and 130 GPa, the corre sponding strain to failure and the toughness are only 2.7% and 50 MJ m −3 . [26] In other words, these characteristics are almost 14 and 3 times lower than those of the FRSF, respectively.

Structural Characterization
Highresolution scanning electron microscopy (HRSEM), small and wideangle Xray scattering (SAXS/WAXS), and syn chrotron Fourier transform infrared microspectroscopy (micro SFTIR) are combined to disclose the hierarchical structures of the FRSF (Figure 3). Similar to animal silks, FRSFs at the mes oscale are composed of nanofibrils with the width of 5-200 nm (Figures 1d and 3a). These nanofibrils are highly oriented along the fiber axis. It is found that the orientation degree of nanofibrils in the FRSF (Figure 3g) is obviously higher than that of the CSF (Figure 3e). Moreover, precise investigation of longitudinalsectional HRSEM images demonstrates that the interfacial bonding and the stacking between nanofibrils in the FRSF are much stronger than those of the CSF (Figure 3b). Studies proved that the tight nanofibrils stacking is critical for improving the strength and the toughness of fibers. [27,28] This kind of selfreinforcement is achieved by the restricted nanofi bril shearing, controlled slippage, and the stress transfer.
WAXS and microSFTIR results reveal the differences between the secondary structure of the FRSF and the CSF. For example, it is found that the scattering pattern of the FRSF is stronger than that of the CSF when 2θ equals to 7.23° ( Figure S4, Sup porting Information). The diffraction peak has been assigned Stress-strain curves of the CSF and the FRSF with a tensile speed of 2 mm min −1 . The temperature and relative humidity of the test environment for the FRSF and CSF are set to 24.1 °C and 43%, and 21 °C and 42%, respectively. c) Comparison of the specific strength and the specific stiffness of the FRSF with other natural and synthetic materials. The Ashby plot was adapted from ref. [24]: Adapted with permission. [24] Copyright 2004, Taylor & Francis. d) Comparison of the density, strength, Yang's modulus, and the toughness of the FRSF with different natural and synthetic fibers. The specific values are listed in Table S1 (Supporting Information).
to (100) reflection spacing of the βsheets, implying FRSFs have higher degree of crystallinity than that of CSFs ( Figure S4c, Sup porting Information). This finding is confirmed by the synchro tron microFTIR characterization, the deconvolution of the amide III band (Figure 3i) provides an estimation of βsheet structure in the FRSFs of 35% to 38%, while that of the degummed A. pernyi CSFs are 24% to 38%. [16,17,29] Therefore, it is inferred that the high modulus and tensile strength originates from the high βsheet contents of the FRSF. Interestingly, no significant difference in the content of βsheet was detected for FRSFs that were harvested from different reeling speeds. For example, the single FRSFs that were produced from reeling speeds of 5, 11, and 30 mm s −1 have almost the same percentage of βsheet, with values of 36%, 38%, and 35%, respectively ( Figure S5, Supporting Information).

FRSF Microactuator Preparation
FRSFbased microactuators are constructed through over twisted yarnspinning techniques. [30,31] The FRSFs are initially twisted into the yarn by overtwisting the fibers and then they are folded at the middle. Then one end of the yarn is released to remove the excrescent torque and form a selftwisting struc ture (Figure 4a). A torquebalanced microactuator is formed, where the selftwisting direction is the opposite of the initial twisting direction. In order to figure out the influence of the strand usage on the fabrication process and the torsional char acteristics, a series of microactuators ranging from 2 to 10 ply fibers are made, where these microactuators are presented (Figure 4b). It is found that as the number of strands in micro actuators varies from 2 to 10, the twist angle of two single yarns linearly increases from 15° to 38°.

Actuation Performance
The actuation behaviors driven by the water fog are evaluated through a selfbuilt apparatus that couple the tensile device with a highspeed camera system (Figure 4c). During the tests, one end of the microactuator is fixed on a cantilever, while another end Adv. Sci. 2020, 7,1902743  is loaded with a paddle, whose two sides are painted with blue and white colors (Figure 4d). The paddle weight is 59.8 mg with a value that is around 200 times larger than the weight of micro actuators. A highspeed microcamera system is employed to record the actuated rotation. The quantitative data are directly extracted from the recorded video through a framebyframe analysis. Figure 4d,e presents the rotation of a 2 ply FRSF microactu ator with the load. It is observed that the time difference between the water contact with the microactuator and the fiber torsion is less than 0.8 s, exhibiting an extraordinarily rapid response for the actuation of the water fog. Therefore, the projection width variation of the paddle is calculated to measure the degree of the rotation. It should be indicated that blue and white colors in each side of the paddle provide a convenient identity of the rota tion. The rotation degree can be calculated in the form below where d(t) and d 0 are the projection width of the paddle at time t and the corresponding width of the paddle, respectively. By cal culating the angle variations over time, the correlation between Adv. Sci. 2020, 7, 1902743 time and rotation angle is established. The microactuator is accelerated to its maxrotation speed within 3.5 s, which exceeds 1488.5° s −1 (248 rotations per minute, rpm) which means that the average rotation speed is 425.3° s −1 (Figure 4e and Figure S6, Supporting Information). It is found that the maximum angular acceleration is 739.9° s −2 (12.9 rad s −2 ). As a result, the twisting moment that the microactuator can provide approaches to 0.11 Nm kg −1 , which is 1.7 times higher than that of the B. mori CSF actuators with the same structure. [32] In fact, it is even com parable with that of the graphene hydrogel fiber actuators. [33] FRSF microactuators show perfect reversible torsional deforma tions during cyclic processes. Furthermore, total actuation power of the FRSF microactuator can be programmed, because it is directly decided by the numbers of the used FRSFs. For example, an FRSF microactuator constructed by two single fibers can gen erate actuation power of 0.77 W kg −1 . As the number of FRSFs increases from 2 to 10, the acceleration time, maximum speed, total duration, and the actuation power increase from 3.4 s, 248 rpm, 11.3 s, and 0.77 W kg −1 to 4.9 s, 1030 rpm, 38.9 s, and 2.1 W kg −1 , respectively ( Figure S7, Supporting Information). This outstanding actuation performance of FRSF microactu ators converts them as ideal choices for numerous applications, including the artificial muscle, linear actuator for locks, smart fabrics, humidity sensors, and so on. [34,35] In the following section, it is intended to show how these FRSF microactuators can be designed to drive wound healing (see Figure 4f). As a prototype, an agar hydrogel was used to mimic the human skin. Then FRSF microactuators are fixed at both ends of two agar gels separated with a gap of 2 mm. Furthermore, water fog is sprayed from the FRSF microactuator side. In such a process, the water fog can trigger the torsion of FRSF micro actuators and contract the overall length, thereby driving the hydrogels to get close to each other (Figure 4g and Movie S2, Supporting Information). Microscopic images clearly show the process, where the screw pitch decreases as the number of spi rals increases, thereby shortening the microactuator as a whole (Figure 4h and Movie S3, Supporting Information).

Actuation Mechanism of the FRSF Microactuator
In order to determine the actuation mechanism of the FRSF microactuator, a highspeed microcamera system is employed to monitor the whole actuation process. As illustrated in Figure 5a and Figure S8a (Supporting Information), the water fog is initially condensed into small water droplets on the surface of the FRSF strands, which mainly consist of the hydro philic sericin. In each unit structure of the microactuator, tiny water droplets located at the smaller diameter of the micro actuator automatically move to the position with the larger diameter. During such a condensation process, the droplet sizes increase continuously. When the furthest droplet moves to the thickest end, the droplet infiltrates the whole microactuator.
The FRSF microactuator can be simplified into a double helix structure (Figure 5b). This structure is formed by two coiled FRSF yarns, while each yarn consists of K-FRSFs where K is the number of the FRSF per yarn. In a single yarn, FRSFs are twisted along the yarn's long axis with a hand of spiral opposite to the yarn's twisting direction. In other words, if the yarn is the lefthanded helix, FRSFs in the yarn must be right handed helix and vice versa. For this kind of configuration, if a torsional deformation of the microactuator happens, the syner gistic changes of the geometric structure of the FRSFs, single yarns, and the microactuator can be expressed as [36,37] cos 1 cos sin cos where δn, δr, δL denote the relative change rates of the pitch number of the single yarn, radius of the single yarn, and the FRSF length, respectively. α f and 0 α f are the twist angle of the single yarn and the corresponding initial twist angle, respec tively. It should be indicated that the twist angle is the angle between the FRSF and the fiber axis of the single yarn. There fore, when FRSFs are swelled and contracted, their long axis will be shortened so that δL < 0, the corresponding radius will be increased, i.e., δr > 0 and subsequently it is found that δn < 0. These variations are schematically presented in Figure 5c. These expressions indicate that the swelling and contraction trigger untwisting of FRSFs in the single yarn. The swelling and contraction of animal silks have been widely studied. If unre stricted animal silks are contacted with polar solvent or exposed to solvent vapor, they can contract significantly along the long axis and swell along the radial direction. [38] The length of fiber can be shortened 5-50%, depending on the species of silks. In terms of FRSFs, their contraction rate can reach 5-10%. [9] For the FRSF microactuator, both ends are fixed (Figure 5d), thus the whole pitch number per unit length of the micro actuator μ is also fixed. [36,37,39] In other words, μ is a positive constant. Therefore, the untwisting of FRSFs in the single yarn leads to an imbalance in the overall structure of the microactu ator due to the mutual restriction between the pitch number of single yarn (n) and the pitch number of the microacuator (N). This mutual restriction can be mathematically described as 2 2 a constant π π µ ( ) Equation (3) indicates that as n decreases, N must increase to compensate for the decrease of the first term. In other words, the untwisting of FRSFs in the single yarn that is triggered by the swelling and the contraction of FRSFs leads to the sponta neous rotating and twisting of the microactuator to increase the corresponding pitch numbers. However, with the progress of the water infiltration, the amount of water inhaled in micro actuator is going to be saturated; the swelling and contraction of FRSFs also will reach equilibrium. Resultantly, after a period of rotation, the rotation speed of the microactuator gradually slows down until it stops.

Influence of the Rough and the Gradient Surface on the Actuation Rate
One of the unique characteristics of the FRSF microactuator compared to other microactuators is the simultaneous presence of the rough and the gradient surface. Therefore, it is intended to investigate the influence of this surface feature on the response rate of microactuators, which is hereafter called the infiltration rate. The larger droplets are more likely to cluster on the external side of the FRSF microactuator during the infiltration ( Figure S8b, Supporting Information). It should be indicated that the external side area has the largest radius in the vertical projection direction. This indicates that the water droplets transport directionally on the micro actuator surface. This droplet characteristic can be inter preted through the combination of the Wenzel equation [40,41] and the Laplace pressure difference. [41][42][43] According to the Wenzel equation, [40,41] the surface gradient can lead to the wettability gradient, thereby driving the droplet movement to a place with higher roughness. Such a driving force can be expressed as [41,44,45] where θ A and θ R are the advancing and receding angles of the water drop on the microactuator, respectively. γ and dl denote the surface tension of the water drop and the movement distance of the water drop, respectively. l 1 and l 2 indicate the inner and external positions of the microdrive unit structure, respectively. However, the Laplace pressure difference typically causes the droplet to have a higher pressure at the slender side. Thus, the internal imbalance in the pressure forces the droplet to move from the inner position to the external position, where the movement path is presented in Figure 6a,b. This can be mathematically described as [41][42][43] where R 1 and R 2 are the radii of the inner and external parts of the microactuator, respectively, when the microactuator is approximated by a cylinder. R 0 is the radius of the water droplet.
Adv. Sci. 2020, 7, 1902743 where ϕ = 6γπNR 0 2 sin α, ε = 8πρR 0 3 lN 2 (R 2 − R 1 + R 0 ), and . ϕ and ε are parameters, which only depend on the shape gradient. δ is positively correlated with the roughness gradient so Δt is always smaller than 0. This equation revealed that the surface rough ness gradient caused by this spiral structure can substantially accelerate the infiltration of water droplets. It means that the rough and the gradient surface enable the rapid response of the microactuator to the water fog. Of note, for the sake of simplifi cation, in this derivation (Appendix 2, Supporting Information) we only consider the spreading time. Therefore, Equation (6) only describes the motion of the drops on the external part of the yarn instead of the entire infiltration process. The water infiltration also includes a process that water infiltrates into the internal yarn structure, whose dynamics are given by Washburn's law of capillary imbibition, with a strong role of viscosity in the microstructure.

Conclusion
Programmable silk fiber microactuators can be achieved by com bining the forcereeling and the yarnspinning technique. Force reeling technique allows harvesting of silkworm silk fibers with structural uniformity and mechanical performance that are com parable with spider dragline silks. On the other hand, the yarn spinning technique is applied to construct microactuators with sophisticated helical structures. In this way, humidityinduced microactuators were developed. In the obtained FRSF microac tuator, the directional transports generated by the surface gra dient, substantially promote the infiltration of water through the microactuator, thereby accelerating the microactuator response on the water fog. They exhibit a maximum rotation speed of Adv. Sci. 2020, 7, 1902743 1030 rpm, a value that is bigger than humiditydriven microac tuators made by cotton, flax wool, or Tencel [32] and even compa rable with the CNT microactuators operated by electricity. [2] The torsional deformation of the helical configuration, on the other hand, channels the actuating power of the silk fiber microactua tors. A quantitative correlation between the helical configuration and the actuating power is derived in the present study so that the required actuating power for different applications can be programmed. Benefiting from superiorities of silk fiber microac tuators in high mechanical performance, versatile processability, and excellent biocompatibility, they are expected to be applied in multiple emerging fields, such as biomedicine, smart textiles, soft robotics, and human-machine interfaces.

Experimental Section
Force-Reeling of A. Pernyi Silk Fibers: Antheraea pernyi silkworms were obtained from the wild oak forest in Dandong, Liaoning province, China. When silkworms start spinning, the heads of the silkworms were fixed to prevent it from swinging from side to side. The silk fibers were directly pulled out from the mouthpart of the silkworms with reeling speeds from 5 mm s −1 to 30 mm s −1 and collected by a rotating cylinder. The morphology characterization of FRSFs and CSFs was carried out by using polarizing optical microscope (Olympus BX51-P, Japan) and HRSEM (JEOL JSM-7800F, Tokyo, Japan) at an acceleration voltage of 5 kV. All the samples were coated with a gold layer with a spray time of 20 s to provide conductivity before observation.
Mechanical Testing: A 50 mm FRSF was cut into three segments with lengths of 10, 25, and 10 mm, respectively. Two 10 mm FRSF segments were employed to measure the cross-sectional areas of the FRSF, and the 25 mm FRSF segment was used for the tensile test. In a typical sample preparation process, a 25 mm FRSF was mounted on a cardboard frame with a base length of 20 mm and fixed with cyanoacrylate. After drying cyanoacrylate overnight, the frame was mounted in the testing machine (Instron 5966 machine, Instron, Norwood, USA) and the side supports of the frame were cut away so that the force was transmitted through the FRSF. Meanwhile, the initial length of the testing FRSF was measured at zero load point-a position in which the FRSF was tight but no force exerted on it. Further, a constant tensile speed (2 mm min −1 ) was applied to the FRSFs until they broke. All the mechanical tests were carried at 20-25 °C with the relative humidity of ≈43%. To calibrate the fibers' cross-sectional area, the reserved two FRSF segments were aligned parallelly on the cardboard with a rectangle hollow in the middle, then the cardboards were covered with polydimethylsiloxane. After vacuuming to remove the air bubbles and curing for 3 h in the drying oven at 60 °C, the cross-sectional areas of fibers were observed with SEM and were estimated with ImageJ software. The average value obtained from two segments was used as the cross-sectional area of the testing sample for the calculation of its tensile stress.
S-FTIR Microspectroscopy of Single FRSF: The experiments were performed at BL01B in the Shanghai Synchrotron Radiation Facility (SSRF), Shanghai, China. S-FTIR spectra were recorded using a Nicolet 6700 FTIR microspectroscopy. The spot size (aperture) in all experiments was fixed as 10 × 10 μm 2 . Such a square aperture, on the one hand, guaranteed enough IR light to irradiate on the sample, on the other hand, avoided light diffraction that caused by size effect of FRSF, since the size of 10 × 10 μm 2 was smaller than the dimension of single FRSF. The spectra were collected in the mid-infrared range of 800-3800 cm −1 at a resolution of 4 cm −1 with 256 coadded scans. In order to characterize the secondary structural information of FRSF, deconvolution of amide III bands was carried out using PeakFit 4.12. [16,29] The number of peaks and their positions were obtained from the second derivative spectra and fixed during the subsequent deconvolution process.
X-Ray Diffraction Experiments: WAXS and SAXS were performed to investigate the mesostructures of single FRSF. WAXS experiments were carried out at Characterization and Analysis Center of ShanghaiTech University by using Xenocs WAXS equipment, Xeuss 2.0. The diffraction patterns were collected by the detector with 172 pixels × 172 pixels of 172 μm × 172 μm area each. The wavelength and the photon flux of the X-ray source was 1.54189 Å and 4.0 × 10 7 photons s −1 , respectively. The beam size at the detector was 1.2 mm × 1.2 mm. SAXS experiments were carried out in Shanghai synchrotron source beamline BL19U2, Shanghai, China, with a wavelength of 1.03 Å, delivering a high-photon flux (5 × 10 12 photons s −1 ) onto the sample. A complementary metal-oxide-semiconductor (CMOS) hybrid pixel detector, with a total number of 172 pixels × 172 pixels, was employed to collect the diffraction patterns. The beam size at the detector was fixed as 0.33 mm (horizontal) × 0.05 mm (vertical).
Preparation of FRSF Microactuators: FRSF-based microactuators were constructed through over-twisted yarn-spinning techniques. The FRSFs were first tied between an electric motor and a movable load which was initially 40 cm away and then they were twisted into the yarn by overtwisting the fiber with the speed of 7500° min −1 for 5 min. They were folded at the middle immediately after the over-twisting process. Then one end of the yarn was released to remove the excrescent torque and form a self-twisting structure. Then a torque-balanced microactuator was formed, where the self-twisting direction was the opposite of the initial twisting direction. In order to figure out the influence of the strand usage on the fabrication process and the torsional characteristics, a series of microactuators ranging from 2 to 10 ply fibers was made.
Actuation Performance Evaluation: The actuation behavior that driven by water fog was evaluated through a self-built apparatus that coupled a tensile deceive with a high-speed camera system. During the tests, one side of the microactuator was fixed on a cantilever, and another side was loaded with a paddle (59.8 mg in weight), which was painted with blue and white color, respectively. A humidifier with a constant power was fixed in the apparatus to generate water fog for humidification of the suspended part of the FRSF microactuators. The whole actuation processing was monitored by using a telephoto lens-based high-speed camera system (i-SPEED 716, iX Camera, UK). The quantitative time-torsion angle relationship was obtained through frame-by-frame analysis of the recorded movies.
Derivation of the Driving Mechanism of FRSF Microactuator: Two relations between the length of single FRSF in yarn (L) and the length of the twisted yarn (h) were obtained from the geometrical changes shown in Figure 5c. They followed that where n is the pitch number of the single yarn, L is the length of FRSF, h is the length of single yarn, r is the radius of the single yarn, α f is the twist angle of the single yarn. By taking the second derivative of L from both sides of Equation (7), Equation (9)  LdL hdh r ndn n rdr π π = + + Substituting Equation (8) into Equation (9), Equation (10) The geometrical relation between Δh and ΔL is illustrated in Figure S9 (Supporting Information). In such a geometrical relation, sin 2 nr L f α π = . For FRSF microactuators, the radius of a single yarn was around 50 μm and the change in the twisting turns due to torsion deformation was generally smaller than 50 turns. Therefore, when the length of the single yarn was long enough, the change between twisting angle before and after torsion, i.e., 0 f f α α − , was very small. Taking this case as an example, the length of single yarn was 45 cm, which was around 9000 times larger than the radius of the single yarn, thus the angle of