Fully self-powered memristor crossbar array with pressure-driven multilevel switching and pattern encoding

The demand for energy-efficient data processing is driving the development of self-powered systems for next-generation electronic devices. Among these, memristors that operate without external power are especially promising for neuromorphic and edge-computing applications, because their resistive states can be controlled by mechanically harvested energy. This work demonstrates a fully self-powered memristor system that integrates a high-sensitivity triboelectric nanogenerator (TENG) with a nitrogen-doped TaO_x-based self-rectifying memristor crossbar array. The memristor shows interface-type resistive switching with a high rectification ratio (> 10⁵), stable endurance over 10⁴ cycles, and reliable 3-bit multilevel data storage. The TENG converts mechanical stimuli into electrical signals and produces sufficient voltage and current to operate the memristor without any external power source. Optimization of the external circuit allows highly reproducible, pressure-controlled multilevel resistive switching. A 6 × 6 memristor crossbar array achieves spatially resolved data encoding and pattern recognition, which demonstrates its potential for low-power neuromorphic computing. The memristor’s intrinsic self-rectifying behavior suppresses sneak currents and enables stable performance under high-density integration. This scalable, self-powered memory platform offers promising applications in artificial tactile sensing, physical AI systems, and next-generation neuromorphic hardware.