Ultrasound-driven programmable synthetic muscle mass | Nature

Fabrication of synthetic muscle mass

The damaging patterns of the unreal muscle mass had been first designed in business digital design automation software program (as proven in Supplementary Fig. 1). The patterns had been transferred right into a photomask by a direct writing laser (DWL2000) machine in a clear room (BRNC). Then we spin-coated the damaging photoresist SU8-3025 on a 4-inch silicon wafer. Using customary lithographic fabrication, the patterns had been transferred to the photoresist by way of publicity to ultraviolet mild by the masks. After the creating course of, the damaging patterns of the microbubble arrays, that’s, micropillars, had been additive on the wafer. The peak of the micropillar is determined by the spinning velocity. Next, to boost the floor properties, a silane-based hydrophobic remedy was utilized to the 4-inch wafer with micropillars for 1 h (see fabrication circulation in Supplementary Fig. 12). The PDMS used on this course of was ready with a ten:1 ratio of the base to curing agent. Then the PDMS combination was poured onto the wafer. To guarantee a high-quality coating, the combination was degassed beneath a vacuum strain of lower than 1 mbar. After degassing, spin-coating of the PDMS was carried out on the wafer. Different spin speeds resulted in various PDMS membrane thicknesses (Supplementary Fig. 2). After spin-coating, the PDMS was vacuumed once more and cured in a sequential heating course of: 1 h at 60 °C, adopted by 1 h at 80 °C and at last 1 h at 100 °C. Finally, the PDMS gentle membrane was cured after which peeled off the wafer. This course of yielded a uniform PDMS layer appropriate to be used in synthetic muscle and gentle robotic functions. In all our experiments, every cavity persistently trapped solely a single bubble as the unreal muscle submerged into the water (Supplementary Fig. 13).

Acoustic set-up

For the microscale characterization of microbubbles, the experimental set-up was constructed on a skinny glass substrate with dimensions of 24 mm × 60 mm × 0.18 mm. As proven in Supplementary Fig. 14, a round piezoelectric transducer (27 mm × 0.54 mm, resonance frequency 4.6 kHz ± 4%, Murata 7BB-27-4L0) was affixed to the glass substrate utilizing an epoxy resin (2-Okay-Epoxidkleber, UHU Schnellfest). A sq. PDMS acoustic chamber (10 mm × 10 mm × 5 mm) was positioned in entrance of the transducer, which was stuffed with deionized water and coated with a canopy glass (22 mm × 22 mm × 0.18 mm). An synthetic muscle was suspended within the centre of the chamber with one finish clamped to the facet wall and the opposite finish left free. The substrate was then mounted on an inverted microscope (Axiovert 200M, ZEISS).

For the macroscale actuation of synthetic muscle mass by sound, the experimental set-up consisted of a plastic tank measuring 10 cm × 10 cm × 8 cm with a wall thickness of two mm. For ex vivo porcine experiments, a bigger chamber (30 cm × 15 cm × 15 cm, thickness 2 mm) was used. As proven in Supplementary Fig. 15, the round piezoelectric transducers had been affixed to the within surfaces and the underside floor of the tank utilizing the epoxy resin or immediately submerged into the liquid. An synthetic muscle was suspended contained in the chamber with one finish clamped, and three cameras had been positioned across the tank to seize the actuation of acoustic synthetic muscle mass from a number of viewing angles. In addition, a miniaturized endoscopic digicam (8 mm diameter and 1080P decision, FuanTech) was used to seize pictures contained in the porcine specimens. An digital operate generator (AFG-3011C, Tektronix) and an amplifier (0–60 V_PP, ×15 amplification, High Wave 3.2, Digitum-Elektronik) had been linked to the transducer to generate sound waves with tunable excitation frequencies and voltages. Square waves successfully drive the unreal muscle, attaining most deformation and outperforming different examined waveforms, akin to sinusoidal and triangular waveforms beneath equal excitation circumstances (Supplementary Fig. 16).

Microstreaming characterization

We evaluated the microstreaming jets generated by ultrasound-driven microbubbles embedded within the muscle utilizing 6-μm tracer particles in water and particle picture velocimetry evaluation. Three uniform-size microbubble arrays, every comprising a 4 × 4 grid of microbubbles with diameters of 40 μm, 60 μm and 80 μm (150 μm in depth), had been individually chosen and examined in separate miniaturized synthetic muscle mass (500 μm × 500 μm × 200 μm; Extended Data Fig. 3a and Supplementary Video 15). When activated at their respective resonance frequencies 76.3 kHz, 57.4 kHz and 27.6 kHz, we measured the microstreaming velocity 80 μm away from the bubble interface and noticed a quadratic relationship between the typical velocity and the excitation voltage (Extended Data Fig. 3b). The streaming velocity close to the bubble reached 2.5 mm s⁻¹ at 60 V_PP. This voltage-dependent microstreaming immediately correlates with the reverse thrust generated by the microbubble array, demonstrating that the thrust magnitude may be dynamically tuned by adjusting ultrasound excitation.

We additional investigated the selective actuation of a variable-size microbubble array of 40 μm, 60 μm and 80 μm diameter, every 150 μm in depth, built-in inside a single miniaturized synthetic muscle (500 μm × 500 μm × 200 μm) with corresponding frequencies (76.3 kHz, 57.4 kHz and 27.6 kHz, respectively). The particle picture velocimetry evaluation revealed that the microstreaming developed by the 80-μm bubbles generated a median velocity of 0.23 mm s⁻¹ at 27.6 kHz, which was markedly stronger in contrast with the velocities (<0.05 mm s⁻¹) produced by the opposite two microbubble arrays on the similar voltage (15 V_PP). Similarly, adjusting the frequency to 57.4 kHz (76.3 kHz) selectively prompts the 60 μm (40 μm) bubble array, leading to extra intense streaming at 0.174 mm s⁻¹ (0.075 mm s⁻¹), in distinction to different arrays (Extended Data Fig. 2). Additionally, making use of a sweeping frequency (10–90 kHz) over 4 s at 30 V_PP enabled wave propagation throughout the unreal muscle (Supplementary Video 16).

Control experiments on synthetic muscle deformation

To decide the important thing elements influencing muscle deformation, a set of management experiments was carried out. We first examined the streaming jets of a uniform-size microbubble-array synthetic muscle (1 cm × 0.3 cm × 80 μm) patterned with over 800 microcavities (every 40 μm in diameter and 50 μm in depth). Supplementary Video 17 reveals that a synthetic muscle with out microbubbles exhibited minor deformation, with no noticeable microstreaming observable throughout the excitation frequency sweeps from 1 kHz to 100 kHz at 60 V_PP. By distinction, the actuator exhibited pronounced deformation at an excitation frequency as little as 9.5 kHz (nicely beneath resonance), the place microbubbles generated microstreaming (roughly 0.8 mm s⁻¹), leading to considerably better deformation in contrast with the case with out microbubbles.

Repeatability and characterization of synthetic muscle deformation

We assessed the repeatability of the unreal muscle’s deformation beneath equivalent excitation circumstances, with the transducer near the microbubble-embedded facet, as proven within the left panel of Extended Data Fig. 8a. When stimulated with ultrasound pulses (80.5 kHz, 52.5 V_PP and 1-s on/off cycle), the muscle exhibited repeatable bending inside 150 cycles, with an error of ±0.8 mm, representing 2.7% of the overall beam size (Extended Data Fig. 8b). With extra excitation cycles (500 cycles) of the unreal muscle, the deformation exhibited bigger error (about 10%). After 10,000 cycles, there have been no observable microbubbles within the synthetic muscle, and the unreal muscle confirmed minor deformation. Furthermore, Extended Data Fig. 8c reveals a quadratic relationship between the utilized voltage and the imply deformation amplitude of synthetic muscle mass, every patterned with uniformly sized microbubbles of 40 μm, 60 μm or 80 μm, when pushed at their respective resonance frequencies (80.5 kHz, 62.5 kHz and 30.3 kHz). In addition, the PDMS beam, within the absence of microbubbles, exhibited restricted bending (about 7% of the 40-μm microbubble-array synthetic muscle’s deformation at 52.5 V_PP) brought on by the weak radiation drive from incident sound waves originating from the transducer.

Control experiments on stingraybot propulsion

In management experiments, a stingraybot with out microbubbles exhibited no undulatory movement alongside its fins beneath ultrasound excitation and sank with out notable lateral displacement (Supplementary Video 18). Notably, beneath steady excitation at a single frequency (examined individually at 33.2 kHz, 85.2 kHz and 96.2 kHz at 60 V_PP), concentrating on microbubble arrays with cavity diameters of 66 μm, 16 μm and 12 μm, respectively, the stingraybot exhibited solely restricted locomotion (<1 physique size). By comparability, sweeping-frequency excitation (10–100 kHz over 2 s) elicited sustained undulatory movement, permitting the stingraybot to swim a considerably better distance (>3.5 physique lengths), as proven in Supplementary Fig. 17. These outcomes recommend that the ahead movement of the stingraybot is dominated by the propulsion drive generated by the sequential undulatory movement, ensuing from the reverse thrust generated by the microbubble arrays. Moreover, enhancing the design of the stingraybot with further microbubble sizes might broaden its manoeuvrability. For occasion, integrating a navigation tail with microbubble arrays of various sizes on both facet permits directional management. When activated at their respective resonance frequencies on one facet, these arrays generate an uneven torque (Supplementary Fig. 18), enabling steering of the stingraybot by way of tail rotation. As the stingraybot is stealthy and clear, we additional envision that our stingraybot may very well be used for environmental knowledge assortment or behavioural analysis on actual organisms, for instance, detecting water high quality inside coral reefs and recording swarm interplay by mixing into colleges of fish.

Robustness analysis

To consider the robustness of our ansatz throughout fluid media, we quantified synthetic muscle deformation in 100% porcine blood, observing amplitudes of roughly 0.4 mm, 1.0 mm, 2.7 mm and 4.4 mm at 15 V_PP, 30 V_PP, 45 V_PP and 60 V_PP, respectively, beneath 96-kHz ultrasound excitation (Extended Data Fig. 9). As complementary proof, we studied the unreal muscle efficiency in numerous aqueous options (deionized water, faucet water and 25–100% glycerol options) as proven in Supplementary Fig. 19. The deformation confirmed an inverse relationship with glycerol focus, with the most important deformation of about 11.3 mm in a 25% glycerol answer, adopted by about 8.4 mm in 50% glycerol and three.7 mm in 75% glycerol. The deformation was virtually negligible in 100% glycerol. These outcomes clearly exhibit that the actuator capabilities successfully in full blood, validating its potential for in vivo functions in fluids with physiological viscosity. We subsequent evaluated synthetic muscle actuation within the presence of strong obstructions (Supplementary Fig. 20). A frontal obstruction (partially blocking ultrasound) diminished the deformation by 80–90% (0.5–1-mm tip deformation versus 4.8 mm unobstructed). A lateral placement induced reasonable attenuation (about 2.5 mm) and posterior positioning retained a greater efficiency (3.8 mm). Furthermore, experimental outcomes confirmed vital deformation of the unreal muscle behind excised porcine ribs (Supplementary Fig. 21). Thus, actuators remained practical close to obstacles however required strategic positioning to maximise deformation. Our preliminary outcomes additionally revealed negligible heating results close to the piezoelectric transducer throughout synthetic muscle and stingraybot operation (Supplementary Fig. 22), underscoring the thermally benign nature of our acoustic platform. Although frequency-dependent selectivity was achieved, some cross-excitation between microbubble arrays was noticed. This impact was mitigated beneath sweeping-frequency actuation, and temporal management over the sweep dynamics has a key function in preserving spatial selectivity and guaranteeing dependable, programmable movement. In vivo biomedical environments current further challenges akin to advanced fluid circulation, irregular geometry and variable temperature gradients, all of which can distort ultrasound propagation. Although the actuator confirmed sturdy and aggressive efficiency beneath static circumstances with different methodologies (Extended Data Fig. 10 and Supplementary Fig. 23), future work will discover flow-resilient designs, together with optimized microbubble-array geometries, versatile ultrasound configurations and real-time actuation management methods to keep up dependable efficiency in dynamic fluid environments.

Numerical simulations

Finite aspect numerical simulations had been carried out utilizing the business COMSOL Multiphysics software program (v6.1), together with simulations on the acoustic strain discipline within the small PDMS chamber, acoustic streaming generated by variable-size microbubbles within the small PDMS chamber, the acoustic strain discipline within the massive acoustic tank and the deformations of the unreal muscle. All simulations had been carried out with dimensions and materials properties per the experiments. Physics modules of simulations on acoustic strain embrace strong mechanics, electrostatics, strain acoustics fields, creeping circulation, and warmth switch in solids and fluids. Simulations on the deformations of synthetic muscle mass had been carried out utilizing the strong mechanics module with corresponding boundary circumstances and drive circumstances. The microstreaming-generated thrust was assumed to be some extent drive that’s loaded on the underside of every microcavity. In addition, numerical calculations primarily based on the theoretical mannequin had been carried out utilizing the business Matlab software program (model R2021b). See Supplementary Notes for simulation particulars.

Imaging and evaluation

The microscale characterization of microbubbles was recorded with a high-speed digicam (Chronos 1.4, Kron Technologies) connected to the inverted microscope. Recording body charges ranged from 1,069 to 32,668 frames per second. The macroscale movement of ultrasound synthetic muscle mass was recorded with a high-sensitivity digicam (Canon 6D and 24–70-mm digicam lens, Canon). The recording body price was 50 frames per second. Recorded footage was analysed in ImageJ. Statistical analyses had been carried out utilizing MATLAB (model R2021b), Originlab (model Origin 2023) and Excel (model 16.54).

Preparation of the zebrafish embryo

Zebrafish (Danio rerio) embryos from pairwise crosses of WIK wild-type fish had been raised in E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl₂, 0.33 mM MgSO₄) at 28 °C beneath a 14:10 h mild/darkish cycle. Experiments as much as 5 days submit fertilization are usually not topic to animal welfare rules. All husbandry and housing procedures had been accepted by the native authority (Kantonales Veterinäramt, TV4206).

Preparation of the porcine organs

Porcine hearts, stomachs, intestines, ribs and blood had been obtained from a licensed abattoir. As the examine concerned solely ex vivo tissues from animals slaughtered for meals manufacturing, no moral approval was required.

Reporting abstract

Further data on analysis design is accessible within the Nature Portfolio Reporting Summary linked to this text.