Biomimetic high performance artificial muscle built on sacrificial coordination network and mechanical training process

The study targets the longstanding challenge of endowing artificial muscle materials with the comprehensive mechanical and actuation characteristics of skeletal muscle: self-strengthening after mechanical training, strain-adaptive stiffening with J-shaped stress–strain behavior, and intelligent, programmable actuation in response to external stimuli. Biological load-bearing tissues such as skeletal muscle display rapidly increasing modulus with strain and strengthen after training due to fibril damage–reconstruction cycles. Many artificial muscle systems can convert heat, light, or electricity into motion but typically lack self-strengthening and strain-adaptive stiffening akin to muscle. The authors propose combining sacrificial coordination bonds with a mechanical training regimen to align polymer chains and restructure dynamic bonds, hypothesizing that this will produce both muscle-like mechanical responses and stimulus-programmable actuation in a scalable polyolefin elastomer composite.

Prior work on artificial muscles includes thermally, optically, and electrically responsive systems where actuation arises from differential thermal expansion, liquid crystal phase transitions, or field-induced changes. For example, fiber-based actuators from cyclic olefin copolymer elastomer/high-density polyethylene exploit thermal expansion mismatch for rapid actuation. However, such systems typically do not show muscle-like self-strengthening or strain-adaptive stiffening. Dynamic sacrificial bonds (e.g., metal–ligand coordination, hydrogen bonds) in biological materials like mussel byssus underpin exceptional toughness and have inspired polymer toughening; introducing Fe3+–catechol bonds greatly enhanced epoxy elastomer strength and toughness. Self-strengthening through repetitive mechanical training has been reported in hydrogels, yet these often lack precise, programmable actuation to external stimuli. Thus, integrating dynamic sacrificial bonding with training to achieve both biomimetic mechanical behavior and programmable actuation in one synthetic material remains an unmet need.

Materials and composite design: Commercial EPDM (ethylene 70 wt%, propylene 29.5 wt%, ENB 0.5 wt%) serves as the elastomeric matrix. Biomass lignin (rich in oxygenated functional groups) is incorporated as a green particulate reinforcer and ligand source. Zinc dimethacrylate (ZDMA) provides carboxylate groups that react with EPDM during vulcanization and furnish Zn2+ for coordination with lignin’s phenoxy/carboxylate groups, forming dynamic interfacial coordination bonds. Some hydrogen bonding between ZDMA-derived carboxylates and lignin also occurs. Processing: A three-step compounding in an internal mixer: (1) EPDM (100 phr) blended with lignin (e.g., 40 phr) at 50 rpm, 80 °C for 8 min; (2) add ZDMA (e.g., 12 phr) and mix 10 min; (3) add ZnO (5 phr), stearic acid (1 phr), BIPB (1 phr), TAIC (0.5 phr) and mix 5 min. Sheets are hot-pressed at 170 °C for 20 min. Samples are denoted LxZy@m% (x = lignin phr; y = ZDMA phr; m = mechanical training strain; @0% indicates no training). An electric-programmable variant substitutes half the lignin with conductive carbon black (C): L20C20Z12@300%. Mechanical training: Repetitive pre-stretching/unloading for 250–1000 cycles at fixed maximum engineering strain (200–600%) using a universal testing machine at 200 mm/min, to mimic muscle exercise-induced destruction/reconstruction of sacrificial bonds and stabilize chain orientation along the training direction. Characterization: FTIR-ATR to verify coordination (enhanced C–O stretching signals, redshift of ~1110→1097 cm⁻¹ with ZDMA). Tensile tests and cyclic hysteresis to evaluate strength, modulus evolution, energy dissipation, and self-strengthening. Stress-relaxation and DSC. In-situ XRD under fixed strains to monitor strain-induced crystallization (SIC) (PE orthorhombic (110) and (200) reflections at 2θ ≈ 20.5° and 22.8°). In-situ SAXS with 1D correlation function analysis to extract structural parameters and core crystalline layer length d0 evolution. DMA actuation tests: isoforce mode (1.0–1.3 MPa preload, −30→90 °C, 3 °C/min) for reversible strain; isostrain mode (0.01% strain, −30→90 °C) for actuation stress. Thermal actuation demonstrations using heating/cooling (hair dryer/ice). Electric actuation: L20C20Z12@300% integrated in a program-controlled circuit under 100 V DC; current (5–30 mA) modulates Joule heating; displacement and temperature monitored (IR camera). Constitutive modeling: A one-dimensional three-subnetwork model (covalent-bond, physical entanglement, and coordination-bond networks) fits engineering stress–strain curves post-training; parameters analyzed to relate training strain and network densities to mechanical response.

- Formation of dynamic interfacial Zn2+-coordination between ZDMA-grafted EPDM and lignin confirmed by FTIR (intensified C–O bands at 1276, 1110→1097, 1023 cm⁻¹ with increasing ZDMA). - Mechanical reinforcement from coordination network: tensile strength and toughness increased with ZDMA; L40Z12@0% achieved 24.8 MPa vs 14.7 MPa for L0Z12@0% (no lignin). - Self-strengthening by mechanical training: For L40Z12, increasing training strain from 0% to 600% (250 cycles) raised tensile strength from 24.8 to 30.7 MPa; stress at 200% strain increased ~2.5×. Residual strain and crystallinity increased post-training, indicating stabilized chain alignment. - Strain-adaptive stiffening (J-shaped curves) with dual-stage modulus enhancement after training. Example L40Z12@600%: elastic modulus increased 16.5×, from ~2.0 MPa (75% strain) to 33.2 MPa (150% strain). L40Z12@300% displayed two enhancement stages: first at ~75–150% strain, second at higher strains. - Mechanistic dissection: • Hysteresis tests showed sharp rise in energy dissipation between 75–150% strain (first stage), consistent with concentrated dynamic fracture of coordination bonds; FTIR supported bond changes in this region. • XRD under strain: L40Z12@300% exhibited SIC with (110)/(200) peaks intensifying with strain; relative crystallinity plateaued at 75–150% (bond fracture stage), then accelerated >150% (SIC-driven second-stage stiffening). L40Z12@0% showed weaker crystallization at same strains. • SAXS correlation analysis: core crystalline layer length d0 for L40Z12@300% remained constant at 75–100% strain (plateau) and then increased faster above 150%, corroborating SIC after bond sacrifice. Controls (no training or no ZDMA) lacked plateau and showed uniform deformation. • Controls demonstrated roles of components: without ZDMA (L40Z0@300%) the stiffening was weaker; without lignin (L0Z12@300%) the dual-stage enhancement disappeared. - Programmable thermal actuation (L40Z12@600%): • Isoforce mode (~1.3 MPa preload): reversible actuation strain up to 41% over −30→90 °C with excellent repeatability, surpassing polyolefin-based benchmarks and meeting human skeletal muscle strain (~40%). • Isostrain mode (0.01%): maximum actuation stress 1.5 MPa at 90 °C, exceeding typical LCE actuators and >4× human muscle stress (~0.35 MPa). Controls without training or missing components showed inferior/absent actuation. • Demonstrations: a 20 mg strip lifted a 205 g load (>10,000× own weight) with >30% strain; a puppet arm achieved ~30° reversible motion while lifting 200 g across cycles. - Electric-programmable actuation (L20C20Z12@300%): Under 100 V with 5–30 mA current control, produced repeatable 13% strain (ΔL ≈ 7 mm) while lifting 505 g (~2000× own weight); actuation strain increased up to ~20% when modulating 0–25 mA. Strain and temperature tracked current, enabling industrially compatible 4–20 mA-type control. - Constitutive modeling: A three-subnetwork 1D model accurately reproduced stress–strain curves at various training strains, indicating that mechanical properties can be programmed by training strain and network densities.

The results validate that combining dynamic sacrificial coordination bonds with mechanical training imparts muscle-like mechanical functionalities to a commercial polyolefin elastomer. Mechanical training orients chains and reorganizes coordination bonds along the training direction, concentrating bond rupture in a defined strain window (75–150%), which dissipates energy and triggers subsequent efficient strain-induced crystallization at higher strains. This two-step mechanism yields pronounced strain-adaptive stiffening and self-strengthening, mimicking skeletal muscle’s J-shaped mechanical response and post-exercise strengthening. The integration of stabilized chain alignment and dynamic bonds also enables large, repeatable thermal actuation strains and high actuation stresses, as well as electrically programmable actuation when conductive filler is included. Compared to human muscle benchmarks, the material meets or exceeds both strain (~41%) and stress (~1.5 MPa) requirements and can perform substantial work (lifting thousands of times its own weight). The approach relies on readily available materials (EPDM, lignin, ZDMA) and standard processing, highlighting scalability and a pathway to intelligent, programmable elastomeric systems.

The study presents a facile, scalable strategy to construct high-performance artificial muscles by integrating sacrificial Zn2+-coordination networks (via ZDMA–lignin–EPDM interfaces) with repetitive mechanical training. Post-training, the elastomer exhibits dual-stage modulus enhancement driven first by concentrated coordination-bond fracture (energy dissipation) and then by efficient strain-induced crystallization (reinforcement), yielding strong self-strengthening and strain-adaptive stiffening (modulus increase ~16.5×). The material demonstrates programmable actuation: thermal actuation strain up to 41% and actuation stress up to 1.5 MPa, plus electrically controlled actuation compatible with current-control signals. This is the first commercial polyolefin-derived artificial muscle meeting key skeletal muscle performance criteria (strain >40%, stress >0.35 MPa) while also offering training-induced strengthening and programmable actuation. Future work could explore tuning network densities and training protocols for targeted performance, extending the concept to other commodity polymers and green reinforcers, optimizing electrical/thermal efficiencies, and scaling for practical actuator applications.