A supertough electro-tendon based on spider silk composites

Abstract

Compared to transmission systems based on shafts and gears, tendon-driven systems offer a simpler and more dexterous way to transmit actuation force in robotic hands. However, current tendon fibers have low toughness and suffer from large friction, limiting the further development of tendon-driven robotic hands. Here, we report a super tough electro-tendon based on spider silk which has a toughness of 420 MJ/m³ and conductivity of 1,077 S/cm. The electro-tendon, mechanically toughened by single-wall carbon nanotubes (SWCNTs) and electrically enhanced by PEDOT:PSS, can withstand more than 40,000 bending-stretching cycles without changes in conductivity. Because the electro-tendon can simultaneously transmit signals and force from the sensing and actuating systems, we use it to replace the single functional tendon in humanoid robotic hand to perform grasping functions without additional wiring and circuit components. This material is expected to pave the way for the development of robots and various applications in advanced manufacturing and engineering.

Fig. 1. Toughness and conductivity of spider silk composites.

a Graph shows the conductivity and toughness of different flexible materials. Green represents metals, red is PDMS-based stretchable conductors, and blue represents other special conducting materials/structures. Pink star represents S-silk@10% SWCNT composites described in this work. b Optical image of a bundle of raw dragline silk from Nephila pilipes. Inset shows a single silk fiber has a very smooth surface. c SEM image showing the wrinkled surface of a single fiber of S-silk@10% SWCNT composite formed from the intrinsic shrinkage of the spider silk after immersion in water during PEDOT:PSS and SWCNT coating. The wrinkled structure prevented any changes in the conductive path, allowing the S-silk composite to maintain its conductivity during stretching and compression. d Cross-sectional scanning electron microscopic image of a spider silk composite. The core is spider silk, and the diameter is about 3–4 μm. The outer conducting layer is about 2 μm. e Stress–strain curves of natural spider silk (S-silk), spider silk with PEDOT:PSS@0%SWCNT (S-silk composite@0%SWCNT), spider silk with PEDOT:PSS@10%SWCNT (S-silk composite@10%SWCNT). Area under gray dotted curve represents toughness of S-silk. Toughness is defined as the energy needed to break the silk. f, g Conductivity and toughness of the S-silk composite increased with increasing weight percent of SWCNT, before experiencing saturation at 12.5 wt%. The maximum conductivity and toughness achieved were 1077 S/cm and 420 MJ/m³, respectively. Our S-silk composite is more conductive and tougher than the other flexible materials shown in Fig. 1a. The error bars in f and g show standard deviations based on 50 independent samples.

Fig. 2. A typical mechanical test of the silk-SWCNT nanocomposite in DPD simulation.

a, b Show a snapshot of the coarse-grained S-silk composite. Red: crystalline structure, beta-sheet structure; green: amorphous structure (3₁-helices and beta-turns); purple: SWCNT. c–f DPD-simulated images showing the structural evolution of natural S-silk (c, d) S-silk composite@10% SWCNT (e, f). Along the x-axis with increasing strain, ɛ₁ < ɛ₂. ɛ₂ is the critical strain at which spider silk is broken. Because of the hydrophobic interactions between SWCNT and spider silk, the S-silk composite has better mechanical properties than natural S-silk. g–i Graph shows Toughness, Young’s modulus and strength increased with increasing wt% of SWCNT, indicating that SWCNT is critical for improving the mechanical properties of the S-silk composite. DPD simulation and experiments agree well. The error bars of g–i for experiments show standard deviations based on 50 independent samples. The error bars of g–i for simulations show standard deviations based on 5 independent simulations.

Fig. 3. Performance of humanoid robotic hands assembled with S-silk composite as electro-tendon.

a Photograph of a 3D-printed robotic finger with an S-silk composite@10% SWCNT electro-tendon held in place by a silicone-based extensor. b Photograph showing the index finger at resting state has an angle of about 8°. c Photograph of the finger bent to the maximum position, pulled by the tendon. d Graphs show the bending angle (red curve) of the robotic finger correlates with the change in length (black curve) of the electro-tendon. The angle at maximum bending is 73° (peak of the red curve). This whole process has taken ~1.5 s. e Cyclic bending tests of robotic fingers assembled using different materials as the tendon show tougher materials were more durable. Our S-silk@10% SWCNT could withstand ~40,000 cycles of the full bending process, nearly double that of fingers using natural S-silk and nylon fiber. Because of low toughness, fingers using steel fiber and PDMS fiber failed to complete a full bending process. The error bars of e show standard deviations based on 10 independent samples. f Graph shows the endurance of the finger using S-silk composite @10% SWCNT. g, h Lifting weight of humanoid robotic finger using different types of fibers (diameter of all fiber is 0.3 mm). S-silk composite@10%SWCNT loaded about 7.6 kg, which was comparable to steel fiber. The error bars of h show standard deviations based on 10 independent samples.

Fig. 4. Feedback processes of the humanoid robotic hand when grasping objects.

a Schematic of the actuating layout and sensing circuit for the humanoid robotic hand. S-silk composite@10% SWCNT was used as the electro-tendon to transmit force from the servo motor of a transmission system and electrical signal from the pressure sensor of the sensing system. Here U_out is the voltage of a reference resistor. Input is the program command of a software that depends on U_out. M: stepping motor; GND: ground. b Photo of the 3D-printed humanoid robotic hand assembled with an S-silk composite@10% SWCNT electro-tendon (yellow dotted rectangle) and a pressure sensor mounted on the index finger. c Graphs show the resistance (red curve) of the electro-tendon did not change when the finger was bent at different angles (green curve). d Grasping process of the robotic hand based on traditional tendon based on nylon fiber. Because nylon is an insulator, no signal is transmitted from the pressure sensor to the software, preventing the finger from catching the green balloon. TF: thumb finger; IF: index finger; MF: middle finger. e Grasping process of the robotic hand based on the S-silk composite@10% SWCNT electro-tendon. The software is programmed to stop the finger when P ≥ 170 Pa (gray area). Because the electro-tendon has high and stable conductivity, the electrical signal from the pressure sensor can be transmitted to the software, which directs the servo motor to respond by pulling the electro-tendon and bending the finger to the appropriate angle based on the pressure data to capture the green balloon.