Microengineered Materials with Self‐Healing Features for Soft Robotics

Recent advancements in soft robotics have led to the development of compliant robots that can exhibit complex motions driven by living cells, chemical reactions, or electronics. Further innovations are, however, needed to create the next generation of soft robots that can carry out advanced functions and exhibit complex locomotion. Material designs that incorporate “smart” functional properties can contribute to the development of robotic systems with in‐built mechanical responsiveness and functions. Herein, a simple material design that integrates stimuli‐responsive self‐healing and microarchitectural features to control locomotion of soft robots is reported. By employing these material designs along with hyperelastic soft actuators to control propellant dispersion and direction, a circuitry of pneumatic and microfluidic logic is created within a dragonfly‐shaped body that enables the robot to undergo user‐ and environment‐controlled locomotion over water surface. In addition to steering the robot to skim, the material properties are also leveraged to detect water acidification, temperature changes, and hydrophobic impurities such as oil. The design, fabrication, and integration strategies demonstrated herein pave a way for developing futuristic multifunctional soft robots, biomedical devices, and environmental monitoring probe.


Introduction
Soft bodied invertebrates and vertebrates have been an inspiration for the development of soft robots. [1,2] Compared to robots made from hard materials, soft robots made of complaint materials have the ability to carry out sophisticated and delicate tasks including probing confined spaces and complex environments. [3][4][5] Such robots can be deployed into the real world to aid humans with tasks such as monitoring, exploration, and remediation, or detection and handling of biological and chemical entities. [6,7] In doing so, these robots must exhibit complex and diverse locomotion, rapid responsiveness to environmental changes, and adaptation akin to living organisms while navigating natural environments with immense diversity. Mimicking zoomorphic motions, including bending, [8][9][10][11] crawling, [9,12,13] climbing, [14] rolling, [15] jumping, [16] and swimming, [10,[17][18][19] inspired by organisms such as snakes, octopuses, spiders, caterpillars, jellyfish, and stingrays has been an active area in soft robotic research. Although early efforts in the area of robotics have largely relied on electronics to control locomotion, [20] recent studies have successfully utilized chemical reactions [21] and self-actuating cells [22] (e.g., cardiomyocytes) to achieve zoomorphic motions and additional functions such as programmable actuation and sensing. [23] The incorporation of soft, active materials can imbue soft robots with advantages such as light weight, compactness, versatility, low cost, resilience, safe interactions with environment, and ability to navigate through confined spaces. Stimuli-responsive materials have been explored to introduce various functions to soft robots such as locomotion, grasping, or sensing. [24] However, efforts at creating an entirely soft multifunctional robot wherein multiple capabilities are enabled without the use of any hard components are still in its infancy. In this article, we describe how utilizing smart material properties such as stimuli responsiveness, self-healing, and microarchitectural features can be used to achieve autonomous control over direction, environment sensitivity, and energy efficient locomotion over water surface in an entirely soft robot.
The dragonfly body offers a unique structural frame to design a functional device that can locomote over water surface as their simple bauplan and body-water interaction at the interface DOI: 10.1002/aisy.202100005 Recent advancements in soft robotics have led to the development of compliant robots that can exhibit complex motions driven by living cells, chemical reactions, or electronics. Further innovations are, however, needed to create the next generation of soft robots that can carry out advanced functions and exhibit complex locomotion. Material designs that incorporate "smart" functional properties can contribute to the development of robotic systems with in-built mechanical responsiveness and functions. Herein, a simple material design that integrates stimuli-responsive self-healing and microarchitectural features to control locomotion of soft robots is reported. By employing these material designs along with hyperelastic soft actuators to control propellant dispersion and direction, a circuitry of pneumatic and microfluidic logic is created within a dragonfly-shaped body that enables the robot to undergo user-and environmentcontrolled locomotion over water surface. In addition to steering the robot to skim, the material properties are also leveraged to detect water acidification, temperature changes, and hydrophobic impurities such as oil. The design, fabrication, and integration strategies demonstrated herein pave a way for developing futuristic multifunctional soft robots, biomedical devices, and environmental monitoring probe. enables efficient skimming. We mimicked some of the body features contributing to the dragonfly locomotion over water surface with a main focus on propulsion, skimming, and environment responsiveness. The unique microarchitectural features (veins and membranes) in the dragonfly wings along with their hydrophobicity allow them to exhibit unique functions including the ability to maintain wing functions even during contact with water. [25,26] We have incorporated these features in a spatially defined manner into an elastomer-based, dragonfly-shaped structure to fabricate a soft robot that skims over water surface, which we call "DraBot" (Figure 1a). Specifically, the hydrophobic wings were realized by fabricating them out of silicone-based elastomers-Sylgard 184 and Ecoflex. The corrugation and heterogeneous cell structures of the wings were mimicked by incorporating microporous Sylgard 184 structures (Figure 1a). These microfeatures along with the intrinsic hydrophobicity of silicone www.advancedsciencenews.com www.advintellsyst.com elastomers ( Figure S1, Supporting Information) and flat wetted surface of the DraBot ensured seamless skimming over water surface. Though some microfeatures were incorporated into the wing-like structures of the robot, the DraBot skims over water surface by utilizing hydrophobicity-mediated floatation and air propulsion whereas dragonflies use aerodynamics to skim. Long-distance locomotion in DraBot was enabled by externalizing the propellant (compressed-air) reservoir, and user control over the speed and direction of travel was executed by a cascade of pneumatic and microfluidic logic gates. The wings were further designed such that they can be operated independently to provide maneuverability and controlled airflow to achieve propulsion and control over its direction. In addition to user control, DraBot was also designed to exhibit pH-driven control over locomotion by utilizing spatially localized pH-responsive self-healing of the wings structure. The incorporation of pH-mediated selfhealing domains also armed the robot with additional environmental sensing and adaptation capabilities. Dragonflies' ability to undergo color changes in response to variations in surrounding temperature was mimicked by incorporating thermochromic pigments. Furthermore, as a proof-of-concept, the hydrophobic microarchitectures were exploited to absorb hydrophobic moieties such as oils from water surface to demonstrate potential applications of such a soft robot.

Results and Discussion
The base of DraBot was fabricated from silicone elastomers with a multistep fabrication process as detailed in the Experimental Section (Figure 1b-f ). Briefly, Sylgard 184 was cured to achieve dragonfly-like structural features embedded with microfluidic channels (Figure 1b-d) and capped with Ecoflex membrane (Figure 1e). The forewings were embedded with pneumatic channels to drive propulsion, and the hindwings were embedded with hydraulic channels leading to microfluidic balloon actuators to regulate the "flapping" (flexing) angle at the base of the hindwings for controlling locomotion. The optimal channel dimensions for achieving the targeted flexing angle were identified from a combined finite element analysis of fluid dynamics and bending mechanics using experimentally determined material properties as input parameters (Figure 1g, Figure S2 and S3, and Movies S1 and S2, Supporting Information). The hindwings and the abdomen of DraBot were decorated with macroporous silicone structures generated from Sylgard 184 ( Figure 1e and Figure S4, Supporting Information). These porous structures offered multiple functions: flotation and motion control in combination with the balloon actuators. Finally, soft tubings for air and fluid were connected to the corresponding pneumatic and hydraulic channel inlets for user-controlled locomotion ( Figure 1e) to yield the final functioning robot (Figure 1f ). DraBot locomotion is driven by pneumatic propulsion through microchannels in the forewings by supplying compressed air via external tubing. The outlets of the pneumatic channels in the forewings face the hindwings allowing the pressurized air to exit in the backward direction, enabling the motion to be controlled by the positioning of the hindwings ( Figure 2). In the rest state, the hindwings are coplanar to the forewings, causing the porous silicone structures on the left and right hindwings to block the air outlets on the forewings. This causes the exiting airflow from the forewings to be dispersed in random directions through the microporous silicone structures, resulting in no net propulsion (Figure 2a,c). To enable forward propulsion, both hindwings were bent upward, through the use of balloon actuators, which moves the blocking microporous silicone structure out of the way of the air outlets in forewings (Movie S3, Supporting Information). The unidirectional air flow now provides a net forward propulsion to the bot, making it move in the forward direction (Figure 2b,d). The speed of the robot can then be simply controlled by the input pressure into the pneumatic channel; Figure S5, Supporting Information, shows the velocity of the DraBot as a function of input air pressure. For all locomotion studies presented, we used 30 psi as the input pressure, resulting in a velocity of %10 cm s À1 .
The bot maintains a straight path over time without any turns when undergoing forward motion (Figure 2d and Movie S4, Supporting Information). To introduce changes in the direction of motion, we used the balloon actuators to flap the hindwings, which misaligns the microporous structure and the air outlets. Specifically, to make the bot turn right, the left hind wing was flapped upward, which exposed the left air outlet, generating a net torque that caused the bot to turn rightward (Figure 2e), and vice versa, flapping the right hind wing caused the bot to turn leftward (Figure 2f ). The extent to which the bot changes direction depends on the duration of time a wing is flexed upward. A sustained left-or right-turn signal led to continuous circular motion (Movies S5 and S6, Supporting Information). The entire control system of DraBot can effectively be mapped onto a cascade of logic gates that includes 4 AND, 2 XOR, and 1 OR gates (Figure 3a). The two microporous structures in the hindwings correspond to the two XOR gates whereas the forewings represent the AND gates at the end of the circuit. The AND and the OR gates at the beginning of the circuit control which part of the downstream circuit gets activated based on the user input. The user input is comprised of three Boolean signalsone pneumatic signal (P), that diverges to both forewings, and two hydraulic signals (H L and H R ) for the balloon actuator in the left and right hindwings-which we denote using [P, H L , H R ]. There are three outputs, one corresponding to forward motion (F) and the other two corresponding to left (T L ) or right turn  (Figure 3b). Using a combination of multiple input signals in series and speed (air pressure) control, DraBot can exhibit complex motions such as slithering (enabling exploration of large areas) and avoiding obstacles through sharp turns (Figure 3c and Movie S7, Supporting Information). These controls over locomotion and direction allow the robot to be used for exploring regions of interest.
Although the aforementioned design principles can introduce long-term and diverse locomotion into a soft robot, the motion is still entirely user controlled. However, living organisms not only exhibit locomotion but also can sense their environment and adapt their motion accordingly while simultaneously carrying out other functions. We further engineered DraBot to simultaneously exhibit several of these characteristics of living systems. First, we introduced environment-adaptive motion by designing the wings on one side to be responsive to pH changes using a self-healing hydrogel. To this end, a surface cut was introduced across the right forewings and hindwings, and the "wound" was "dressed" with an acryloyl-6-aminocaproic acid (A6ACA) hydrogel coating that shows instantaneous reversible pH-responsive healing [27] (Figure 4a). At low pH, the A6ACA hydrogel welds the two left wings together, preventing the hind wing from flapping, and thereby blocking the air exit channel on the left forewing (Figure 4b and Movie S8, Supporting Information). Thus, when the bot encounters acidic conditions, it makes a continuous left turn (resulting in counterclockwise circular motion), despite receiving a user input signal for forward motion (Figure 4c and Movie S9, Supporting Information). Exposing to higher pH dissociates the welded state of the right wings and the bot resumes its original forward motion that it exhibited before encountering the acidic conditions. Therefore, through judicious incorporation of stimuli-responsive materials into DraBot, it was able to not only sense and adapt to pH changes in water but also report on such perturbations via observable changes in motion. Second, we mimicked the thermoresponsive color changes in dragonflies [28,29] by encoding DraBot's wings with a thermochromic pigment which changes color in response to temperature changes, wherein the wing color changed smoothly from red to yellow with increase in temperature (Figure 4d and Movie S10, Supporting Information). Finally, the hydrophobicity along with the large surface area of the microporous silicone structures at the abdomen and wings of DraBot was further exploited to detect and absorb hydrophobic entities such as oil from the www.advancedsciencenews.com www.advintellsyst.com surface of water (Figure 4e and Movies S11 and S12, Supporting Information). Efficient oil absorption is achieved through the coupled action of hydrophobicity and capillary forces resulting from the 3D microporous structure. Although each function has been demonstrated independently for clarity and ease of observation, all these functions were incorporated into the same DraBot using various domains-pH mediated self-healing was confined to the middle region of the right wing as it was also used to control the motion, thermochromic pigment was incorporated within both the wings, and the microporous structures were confined to the abdomen and far end of the wings.

Conclusions
In summary, we have coupled soft active materials with discrete functions in a spatially defined manner across a dragonfly-like structure to create a multifunctional robot that travels across the surface of water in a user-and environment-controlled manner. This was achieved by integrating hydraulic-pneumatic logic with environment-responsive smart materials including selfhealing hydrogels. Such smart soft robots can find multiple applications in environmental monitoring. A key design strategy to achieving such multifunctionality is the incorporation of www.advancedsciencenews.com www.advintellsyst.com multiple detection mechanisms, each confined within distinct structural segments of the robot, which enables simultaneous detection and reporting of multiple environmental stimuli or hazards. For instance, the ability of the robot to respond to pH can be leveraged to detect freshwater acidification, which is a serious environmental problem affecting several geologically sensitive regions. [30,31] Similarly, the ability of the microporous silicone structures to absorb hydrophobic moieties makes such long-distance skimming robots an ideal candidate for the early detection of oil spills. Such spills are indeed a serious environmental and economic hazard as evident from the deep horizon oil spill in 2010. [32,33] Although in the current configuration the DraBot is at most capable of detection, one can also envision using a swarm of such smart robots to cleanup oil spills. Finally, the temperature-sensitive color change of the robot's wings can be used to identify changes in the surface temperature of water associated with red tide, [34] bleaching of coral reefs, [35] and decline in the population of aquatic life. [36] The focus of this study was assessing the possibility of using material designs to achieving diverse functions within a soft robot. Although the proof-of-concept studies have shown the potential, further improvements are needed to realize the applications in the real world. For instance, the incorporation of wireless cameras and solid-state sensors, functioning in concert with stimuli-responsive materials-based sensing and changes in locomotion, will further enhance the reporting capabilities of the bot. Although the tethered nature of the bot does possess limitations such as wire entanglement in very long runs, creating an on-board propellant reservoir that would support long-term motion still remains a major challenge. In addition to exploring the environment, smart soft robots can also find applications in health care. [37]

Experimental Section
Finite Element Analyses to Determine Hydraulic Channel Dimensions for Balloon Actuators: A 3D numerical simulation was developed in COMSOL with an input of experimentally determined material properties to design the balloon actuators, which were used to control the bending angles of DraBot's hindwings. The inflation of the balloon actuator induced a pulling force on the microfluidic channel, which when strong enough can bend the channel structure and thus generate an out-of-plane motion. [38] The extent of bending achieved highly depended on the material properties of the two material layers sandwiching the channel, as also reported by previous works. [38,39] The technical details of the model have been described in the Supporting Information.
Mechanical Properties of Ecoflex: Hyperelastic constants of the Ecoflex 00-30 were determined by tensile measurements using an ElectroForce 3220 Series III (TA Instruments, Inc.). Type V (i.e, dumbbell) samples with 4 mm thickness were prepared according to ASTM (American Society for Testing Materials) standard ( Figure S2a, Supporting Information). The www.advancedsciencenews.com www.advintellsyst.com stress-strain measurements were conducted with a load cells of 225 N at 0.25 mm min À1 . The two-term Ogden model described later was used to fit the measured stress-strain curve to obtain the hyperelastic constants ( Figure S2b and S2c, Supporting Information) used in our balloon actuator simulations.
where λ is the principle stretch and α 1 , α 2 , μ 1 , and μ 2 are material constants. Experimental Validation of Theoretical Predictions: To validate the theoretical prediction from COMSOL simulations, rectangular microfluidic channels were prepared with constant channel width Â height (1.8 mm Â 0.15 mm) and varying length (4.1, 5.6, and 7.1 mm) using soft lithography. The patterned silicon wafer was made using SU-8 photolithography. Sylgard 184 was poured onto the patterned silicon surface and cured at 60 for at least 2 h. Cured Sylgard 184 was released from the wafer, and a 1 mm inlet was created using a biopsy punch (Miltex, Inc.). Each actuator was pressurized by a syringe pump (Harvard Apparatus, Inc.) at a flow rate of 2.5 μl s À1 , and the angular displacement corresponding to the injected fluid volume was measured. Three cycles for each actuator were tested for data acquisition. Channels with a dimension of 7.1 Â 1.8 Â 0.15 mm (length Â width Â height) were used for all further studies.
Soft Robot Fabrication and Assembly: The aluminum mold with dragonfly structural features (3.5 cm wingspan and 5.7 cm length) was fabricated by computer numeric control machining (Xometry, Inc.) with aluminum 6061-T6 material. Pneumatic and hydraulic channels were fabricated using soft lithography wherein patterns with a height of 150 μm were etched using an SU-8-100 photoresist (MicroChem, Inc.) on a silicon wafer (100 mm, (1 0 0), boron-doped, p-type, ID: 452, University Wafer). Detailed experimental steps for patterning of silicon wafer through baking, aligning, exposing, and developing were followed as per the specifications in the product datasheet. This patterned silicon wafer was used as a master mold for the pneumatic and hydrolytic channels. Sylgard 184 (Dow Corning, Corp.) was mixed at 10:1 ratio (base:curing agent), degassed, and poured onto the silicon wafer and aluminum mold. The mold was the inverted onto the silicon wafer to encase the Sylgard 184 between the mold and the silicon wafer and cured at 60 C for 2 h. The cured DraBot was peeled off from the mold, 1 mm diameter channel inlets (Miltex, Inc.) were punched, and the bot was attached on 0.5 mm thick Ecoflex 00-30 silicone rubber membrane (Smooth-on, Inc.) using O 2 plasma (K1050X, Quorum Technologies, Ltd.) at 50 W for 30 s. The membrane was attached to the body immediately after the plasma treatment and placed on a hotplate at 200 for 5 min followed by baking at 60 C for 2 h to ensure strong and durable bonding. Flexible silicone tubing, with an internal diameter of 0.0635 cm (McMaster-CARR, Inc.), was inserted into the chest area of the dragonfly-bot at inlets of the pneumatic and the two hydraulic channels.
Contact Angle Measurements: Hydrophobicity of Ecoflex, cured Sylgard 184, and microporous structures generated from Sylgard 184 were characterized by measuring water contact angle using a goniometer by placing a 4 μl droplet on flat samples. Then, 3-4 samples per material were characterized and the contact angles were reported as mean AE standard deviation.
Interconnected Microporous Silicone Structure: Microporous silicone structures were prepared using salt leaching method. Granular salt (NaCl) was filled in a 24 well plate and exposed to a humid environment. Upon hardening and clump formation, 10:1 Sylgard 184 mixture (base:curing agent) was poured onto wells containing NaCl clumps, and was placed in a vacuum desiccator for 4 h to allow the Sylgard 184 solution to penetrate into the salt clumps. The Sylgard 184 infiltrated clumps were allowed to cure at 60 for 2 h and then immersed in water overnight on an orbital shaker to dissolve the salt. The dissolution of the solid salt resulted in a silicone structure with interconnected pores. [40,41] The morphology of the microporous silicone was analyzed by SEM. Silicone slab and microporous silicone structures were cut into smaller pieces (%3 cm Â 3 cm Â 3 cm), and sputter coated with gold for about 90 s at 12 mA current using Denton Vacuum Desk V Standard sputter coater. The gold coated samples were imaged using FEI Apreo scanning electron microscope at 2 kV operating voltage. Images were acquired from at least three different positions ( Figure S4, Supporting Information). The microporous structure was cut into desired shapes and attached at hindwings and abdomen site of the DraBot.
Synthesis of A6ACA: A6ACA was synthesized from 6-aminocaproic acid (6ACA) (Sigma Aldrich, Inc.) as described in previous study. [42] Briefly, 6ACA (0.1 M) and NaOH (0.11 M) were dissolved in deionized water (80 mL) in an ice bath with constant stirring. Following complete dissolution, acryloyl chloride (0.11 M) (Sigma Aldrich, Inc.) in tetrahydrofuran (15 mL) was added dropwise. The pH was maintained at 7.5-7.8 by the addition of NaOH (2.5 M) solution until the completion of the reaction. The reaction mixture was then acidified to pH 3 using HCl (6 M) and extracted with ethyl acetate. The ethyl acetate layer was collected and dried over anhydrous sodium sulfate to remove any traces of water. After filtration of sodium sulfate, the solution was concentrated using a rotavapor and precipitated in hexane. Precipitation was conducted in an ice bath under constant stirring. Repeated precipitation was used to purify the product, and the product was dried in a desiccator.
Surface Functionalization of Silicone Structures: For the bonding of A6ACA hydrogel to cured silicone, surface of the silicone structures made from Sylgard 184 was functionalized with 3-(Trimethoxysilyl) propyl methacrylate (TMSPMA) (Sigma-Aldrich Inc.). The silicone slabs (wing of the DraBot) were cleaned by using sonication with isopropanol, ethanol, and deionized water for 5 min, respectively, and dried before use. The cleaned surface was subjected to oxygen plasma treatment (100 W, K1050X, Quorum Technologies, Ltd.) for 3 min, following which the structure was dipped into 2 wt% of TMSPMA in acetic acid solution (pH 3.5, dissolved in deionized water) for 2 h at room temperature. The silicone structure was then washed with ethanol, dried, and stored in low humidity conditions until used.
The functionalization was characterized by Fourier transform infrared spectroscopy (FT-IR). FTIR analyses was performed on 1 cm Â 1 cm silicone substrates with a thickness of 1 mm to examine the effect of oxygen plasma and TMSPMA modification. The infrared spectrum was detected using Thermo Electron Nicolet 8700 equipped with the attenuated total reflection module. Each sample was scanned 32 times at room temperature and atmospheric pressure. The data ( Figure S6, Supporting Information) were collected in the range of 4000-500 cm À1 and analyzed using OMNIC software (Thermo Fisher Scientific Inc.).
Coating of A6ACA Hydrogel onto Functionalized Silicone Structures: Functionalized silicone slabs were coated with a thin layer of A6ACA precursor solution and polymerized via free-radical polymerization at 37 C for 3 h. The precursor solution was prepared by mixing 0.1% N, N 0methylene bisacrylamide (Bis-Am) (Sigma-Aldrich, Inc.) with of A6ACA (1 M) monomer dissolved in NaOH (1 N). Roughly, 0.5% ammonium persulfate (APS) and 0.15% tetramethylethylenediamine (TEMED) were used as initiator and accelerator, respectively.
Incorporation of Thermochromic Ink: Thermochromic pigment (Blue-Colorless or Red-Yellow, Atlanta Chemical, Inc.) was used for temperature sensing of DraBot. During the fabrication process, the pigment was mixed with Sylgard 184 before pouring into the wing region whereas the rest of the bot was fabricated out of Sylgard 184 mixed with regular silicone colorant (red).
Controlling the DraBot Motion: The rear ends of the hydraulic tubings were connected to two 10 mL syringes controlled manually whereas the end of the air tubing was connected to the central compressed air. Thus, the span area of DraBot is πl 2 , for a tubing length of l. The exploring motions corresponding to various input pressures were recorded using a video camera and analyzed by ImageJ software (NIH). A 30 psi air pressure was used as a standard input pressure for all experiments described in the article.