Mechanically Durable Intrinsically Stretchable Neuromorphic Devices via Molecular Microstructure Design.

Intrinsically stretchable neuromorphic devices (ISNDs) have been widely investigated for intelligent wearable on-device computing. However, conventional material design strategies that soften the polymer conjugated moiety to impart stretchability have shown limited mechanical durability, typically 103 cycles at 50% strain, with severe electrical degradation. Here, we present a highly durable ISND that maintains stable electrical performance for up to 105 cycles at 50% strain, enabled by molecularly controlling chain stacking of the semiconducting polymer. This is achieved by incorporating a microstructure-controlling moiety into the polymer backbone, which modulates the chain packing from a bundle-like to a mesh-like structure. The resulting mesh-like morphology forms robust and long-range percolation networks that preserve charge transport pathways and structural integrity under mechanical deformation. Utilizing this material, we fabricate ISNDs that exhibit device-level stretchability of up to 150% and exceptional cyclic stability, with less than 15% variation in output current after 105 cycles at 50% strain. Furthermore, we demonstrate reliable on-device artificial intelligence using reservoir computing, with consistent classification accuracy maintained even after 105 mechanical cycling at 50% strain. This work offers a molecular design strategy for tuning semiconductor film morphology, achieving mechanical reliability in stretchable neuromorphic electronics for future wearable and biomedical systems.