Adaptive self-healing electronic epineurium for chronic bidirectional neural interfaces

The study addresses a central challenge in peripheral nerve interfaces: chronic, safe, and stable bidirectional communication without compressive damage from mechanical mismatch between electronics and soft, dynamically moving nerves. Conventional cuff electrodes can restore function but often exert continuous compressive forces and generate shear due to modulus mismatch, leading to vascular compromise, fibrosis, and recording/stimulation failure. The authors hypothesize that an intrinsically stretchable, self-healing, stress-relaxing neural interface that autonomously adapts its mechanical modulus to nerve tissue and self-locks without sutures/glues will minimize compression- and shear-induced immune responses, enabling long-term stable recording and stimulation in vivo.

Prior peripheral nerve cuffs (e.g., spiral cuffs and flat interface nerve electrodes) have shown efficacy in human neuroprosthetic applications for stimulation and CAP recording, but chronic compression and fibrotic encapsulation remain limiting factors. Flexible cuffs alleviate early compression yet long-term shear at the electronics–nerve interface due to modulus mismatch leads to immune responses and scar tissue that degrade performance. Recent advances in soft, stretchable electronics allow modulus matching and strain-insensitive conduction, revolutionizing neural interfaces. Hydrogel-based stretchable conductors with low interfacial modulus/impedance have improved neuromodulation, underscoring the importance of mechanical matching. However, how chronic deformation and growth of peripheral nerves affect interfacial mechanics is insufficiently addressed. Strain-induced stress relaxation of extracellular matrices is relevant to cellular signaling, suggesting that materials with dynamic stress relaxation may create stress-free interfaces in vivo.

Materials and device fabrication: The A-SEE comprises a tough, dynamically crosslinked self-healing polymer (SHP; PDMS-4,4′-methylenebis(phenyl urea) (MPU)0.4-isophorone bisurea (IU)0.6) as substrate/encapsulation and an Ag flake–SHP composite (Ag:SHP = 4:1 w/w) as conductor. Layers are assembled by water-proof self-bonding (self-healing property) to form a conformal, suture-/glue-free wrap-around cuff. To enhance biocompatibility and electrical stability, a 60 nm Au nanomembrane (AuNM) is integrated with the composite by transfer: Au is e-beam evaporated onto SiO2, the Ag flake–SHP solution is cast and dried, then the AuNM-composite delaminates due to stronger Au–Ag/polymer vs Au–SiO2 interactions. A direct-deposition AuNM-composite was also tested but underperformed; the transferred AuNM-composite was used subsequently. Mechanical characterization: Dynamic mechanical analysis (DMA) measured time-dependent stress relaxation at 30% tensile strain for SHP, composite, encapsulated composite (A-SEE), and PDMS (control, 20:1 base:curing agent; Young’s modulus comparable to SHP). Tests were conducted at 25 °C and 37 °C for 60 min. Ex vivo shear stress at SHP–nerve interfaces was assessed. Finite element analysis (ANSYS Workbench) modeled a cylindrical nerve (1 mm diameter, 10 mm length) wrapped by hollow cylinders (1 mm inner diameter, 1.6 mm outer diameter, 5 mm height) of SHP or PDMS; pressure loads derived from DMA were applied to evaluate compressive stress distributions over time. Electrical/mechanical durability: Resistance–strain characteristics were measured with four-point probing under monotonic stretching (up to 100% strain) and cyclic stretching (50% strain, 1000 cycles). Self-healing was evaluated after complete cutting and thermal healing (60 °C, 1.5 h). Bending durability was tested by repetitive bending to a 0.5 mm radius (rat sciatic nerve radius) for 1000 cycles. Interconnects used Au/PI pads bonded via self-bonding, with Teflon-coated wires soldered and encapsulated. Electrochemistry: Electrochemical impedance spectroscopy (1 Hz–1 MHz, 10 mV) and cyclic voltammetry (−0.65 V to 0.8 V vs Ag/AgCl, 100 mV/s) were performed in PBS (exposed area 7.5 mm²; Pt counter electrode). Impedance stability was monitored over 10 days immersion and under strains up to 100%. Biocompatibility and ion release: In vitro cytocompatibility was tested with RAW 264.7 macrophages and C2C12 myoblasts cultured for 7 days on SHP, composite, and AuNM-composite. Ag-ion leakage was quantified by ICP-MS from media incubations (7 days) and from in vivo tissues (acid-digested samples). The effect of AuNM microcracks under strain (up to 100%) on cytotoxicity was evaluated. Animal experiments: Male Sprague–Dawley rats (~300 g) were implanted with A-SEE cuffs on the sciatic nerve using the self-locking process (<1 min). For comparative compression studies, SHP and PDMS cuffs (1 mm inner diameter) were implanted for 1 week. Histology/immunofluorescence: H&E, TUNEL assay, and immunostaining for CD68 (macrophage activation) and CTGF (pressure/fibrosis marker) were performed; fibrosis thickness quantified at 1 and 6 weeks. Western blots quantified CD68 and CTGF. Neural recording: A custom preamplifier/external module (gain 39,601; 3 dB bandwidth 425–5500 Hz) recorded CAPs from mechanoreceptor activation via brush stimuli: strong (12.5 g/cm² at 0.05 s/cm) and weak (3.1 g/cm² at 0.05 s/cm). Signals were digitized at 25 kHz and processed in LabVIEW. SNR was defined as the RMS ratio during stimulus-present vs stimulus-absent periods; denoising used maximal overlap discrete wavelet transform. Neural stimulation and kinematics: Biphasic current pulses (80 μA amplitude; 0.25 ms pulse width; 0.5 s train; 5–100 Hz repetition; current density 1.14 μC/cm²) were delivered to the sciatic nerve (AM-2200 stimulator). Muscle force was measured with a torque sensor and normalized to the 100 Hz condition. Joint kinematics during treadmill gait were recorded synchronously with neural signals. Nerve-to-nerve interfacing: Two A-SEEs were implanted bilaterally; ipsilateral recordings were processed by a controller that triggered contralateral stimulation when the rectified-bin-integrated neural signal exceeded a threshold (9 μV mapped to 50 μA stimulation), demonstrating a crossed-extensor–like response. Joint movements of the contralateral limb were quantified.

• Materials mechanics: SHP and A-SEE exhibited pronounced time-dependent stress relaxation at 30% strain, significantly exceeding PDMS; relaxation increased at 37 °C versus 25 °C, enabling adaptive modulus matching to nerve tissue. Ex vivo shear at SHP–nerve interfaces was minimized.
• Modeling: FEA showed that SHP cuffs yielded compressive stress-free nerve interfaces over time, whereas PDMS cuffs produced persistent compressive stresses and nerve deformation.
• In vivo tissue response: Compared with PDMS, SHP interfaces showed lower CTGF expression and fewer TUNEL-positive apoptotic cells after 1 week, indicating reduced compression-induced damage. Fibrotic capsule thickness around nerves was significantly reduced with SHP at 1 and 6 weeks (ANOVA with Tukey: P(Week 1)=0.0097, P(Week 6)=0.0104; n=3), demonstrating mitigated foreign body response.
• Electrode performance: The transferred AuNM–Ag flake–SHP composite showed stable resistance under stretching (0–100%) and after self-healing, and endurance for 1000 cycles at 50% strain and during repetitive bending to 0.5 mm radius. In PBS, 1 kHz impedance remained stable over 10 days and under strain (347.6 Ω at 0% strain to 526.5 Ω at 100% strain at 1 kHz). CV under 0% and 100% strain indicated electrochemical stability, with minor shifts attributed to AuNM cracking and partial Ag composite exposure.
• Biocompatibility and ion release: AuNM coating improved cell viability for RAW 264.7 and C2C12 over composite alone (one-way ANOVA with Tukey: P=0.0004 and 0.0001, respectively). ICP-MS confirmed reduced, but not eliminated, Ag ion release with AuNM. Even with microcracks under up to 100% strain, no cytotoxicity was observed. In vivo Ag leakage was measurable (e.g., 17.03 ppm after 32 weeks reported), highlighting residual release.
• Chronic recording: A-SEE recorded mechanoreceptor-evoked CAPs in awake rats for 6 weeks with distinguishable strong vs weak stimuli; median SNRs remained stable with an overall SNR around ~1.76 at 6 weeks (n=5). Denoising further clarified signals. CD68 and CTGF levels at 6 weeks were comparable to 1 week, supporting chronic stability.
• Stimulation and durability: Electrical stimulation elicited frequency-dependent muscle forces, with twitch responses below 20 Hz and tetanic contractions at ≥20 Hz. After 100,000 stimulation pulses, recording SNRs were unchanged (ANOVA P=0.7795). Sensory signals were still clearly recorded at 14 weeks; stimulation reliability persisted up to 32 weeks without device failure.
• Closed-loop demonstration: A bilateral nerve-to-nerve interface relayed ipsilateral sensory CAPs to contralateral stimulation (threshold 9 μV → 50 μA), evoking contralateral limb movements resembling a crossed extensor reflex and enabling modulation of gait phases (stance/swing) during treadmill walking.

By combining an intrinsically self-healing, dynamically stress-relaxing elastomer with a stretchable, low-impedance electrode, A-SEE addresses chronic compression and shear at the nerve–electronics interface—the main causes of fibrosis, immune activation, and device failure with traditional cuffs. Self-locking, water-proof bonding eliminates sutures and adhesives, simplifying microneurosurgery and reducing acute tissue trauma. The adaptive modulus matching and stress relaxation minimize CTGF upregulation, apoptosis, and fibrotic encapsulation in vivo, preserving ionic pathways and enabling stable recording and stimulation over weeks to months. The robust electrochemical/mechanical performance under large strains and bending, together with reduced Ag ion exposure via AuNM, underpin durable function during joint and muscle motion. The nerve-to-nerve demonstration shows feasibility for closed-loop peripheral neuroprosthetics that couple sensory inputs to motor outputs, suggesting routes to restore function after injury or disease.

This work introduces an adaptive, self-healing electronic epineurium (A-SEE) that forms compressive stress-free, conformal, and sutureless interfaces with peripheral nerves. Materials-enabled stress relaxation and self-bonding, coupled with a transferred Au nanomembrane–reinforced Ag flake–SHP electrode, yield long-term stable bidirectional interfacing in rat sciatic nerves, including weeks-to-months recording and stimulation and a proof-of-concept nerve-to-nerve reflex-like system. The platform reduces immune response and fibrosis while maintaining low impedance under large strains. Future work should further mitigate residual Ag ion release (e.g., by Au shell coating of Ag flakes), refine fully in-plane electrode designs to avoid protrusion-induced fibrosis, and pursue scaled, wireless, and human-compatible implementations for clinical electronic medicines across the peripheral nervous system.

• Residual Ag ion leakage persists despite the Au nanomembrane barrier (e.g., measurable in vivo; 17.03 ppm at 32 weeks), indicating a need for improved encapsulation (e.g., Ag flakes with Au shells).
• AuNM microcracking under strain can expose Ag composite; although no cytotoxicity was observed in vitro, it may contribute to ion release over long durations.
• Early post-implant decreases in CAP amplitude/SNR likely reflect surgical inflammation; while signals stabilized, acute responses remain.
• Protruding electrode geometries induced localized fibrosis; although an in-plane redesign mitigated this, geometry remains a design sensitivity.
• The nerve-to-nerve demonstration approximates but does not replicate the biological crossed extensor reflex pathway.
• Results are from rat sciatic nerve models with modest sample sizes; broader species, nerve types, and longer-term assessments are needed for generalizability.