Maze-like crack networks guided by wrinkles in MXene/PDMS for machine learning-assisted flexible strain sensing

Isotropic wrinkles guide the crack formation in MXene/PDMS via a “peak pinning” effect. • The maze-like crack network provides high sensitivity (518.7) and a broad linear range (62.2%). • The sensor can monitor a full spectrum of human activities, from subtle pulses to vigorous body movements. • Machine learning enables the sensor to achieve 95.0% accuracy in complex gesture recognition. Simultaneously integrating high sensitivity, a wide linear range, and robust cyclic stability into a single flexible strain sensor remains a critical yet unresolved challenge for wearable electronics. Herein, we report a facile and scalable strategy to fabricate a MXene/PDMS sensor that employs isotropic wrinkle microstructures to guide the formation of a maze-like crack network (MCN). A critical “peak pinning” effect induced by these wrinkles effectively suppresses catastrophic crack propagation while promoting the progressive generation of the MCN, which inherits the meandering characteristics of the wrinkles. By precisely tuning the MXene deposition density (0.5–4 g m −2 ), both the wrinkle and crack geometry as well as the resultant piezoresistive properties can be tailored. Consequently, the optimized sensor achieves an exceptional balance of performance metrics: a high gauge factor of 518.7, a broad linear working range of 62.2%, fast response and recovery times (13 ms and 42 ms, respectively), and robust durability exceeding 1000 cycles. The device demonstrates versatile monitoring capabilities across a full spectrum of human activities, spanning subtle physiological signals (e.g., arterial pulse waves and laryngeal vibrations) to large-amplitude joint movements. Furthermore, when integrated with an XGBoost ensemble machine learning algorithm, the sensor enables a smart sensing system that achieves 95.0% classification accuracy for complex hand gestures. Collectively, this work presents a synergistic approach combining scalable fabrication, engineered microstructures, and artificial intelligence (AI)-enhanced data processing to advance the development of next-generation human-machine interfaces. Download: Download high-res image (164KB) Download: Download full-size image