Flexible Zn‐TCPP Nanosheet‐Based Memristor for Ultralow‐Power Biomimetic Sensing System and High‐Precision Gesture Recognition

• Published in: Advanced functional materials Vol. 34; no. 26
• Main Authors: Wang, Yilong, Su, Jie, Ouyang, Guoyao, Geng, Sunyingyue, Ren, Mengchen, Pan, Weiliang, Bian, Jing, Cao, Minghui
• Format: Journal Article
• Language: English
• Published: Hoboken Wiley Subscription Services, Inc 01-06-2024
• Subjects: Artificial intelligence; Bending; biomimetic sensing system; Biomimetics; Bionics; Cognition & reasoning; Cotton fibers; flexibility; Gesture recognition; Learning; Low voltage; low‐power consumption; Memristors; Neural networks; neuromorphic computing; Porphyrins; Power consumption; Random noise; Stimuli; Voltage pulses; Zn‐TCPP nanosheets
• ISSN: 1616-301X; 1616-3028
• DOI: 10.1002/adfm.202316397

The flexible biomimetic sensory system inspired by biology exhibits learning, memory, and cognitive behavior toward external stimuli, providing a promising direction for the future development of the artificial intelligence industry. In this work, a Zn‐TCPP (TCPP: tetrakis (4‐carboxyphenyl) porphyrin) based flexible memristor with ultra‐low both operating voltage (≈80 mV) and power consumption (0.39 nW) that simulates typical synaptic plasticities, under continuously adjustable ultra‐low voltage pulses (50 mV). The synaptic properties are well maintained even when bending 1000 times at a radius of 5 mm. Furthermore, the flexible bionic sensing system integrated with Zn‐TCPP based memristor and cotton fibre piezoresistive sensor can remember pressure and deformation current, thus simulate the learning‐forgetting‐relearning characteristics under mechanical stimuli (power supply = 100 mV). Especially, the system achieves a high recognition rate of 97% for gestures through self‐built datasets and neural network calculations and remains at a high level under the influence of 10% Gaussian noise (80%) and 5 mm bending state (91%). Consequently, the ultralow‐power flexible biomimetic sensing system shows great potential in the field of integrated artificial intelligence with multiple modules, paving the way for the development of low‐power biomimetic robots in the future.
The Zn‐TCPP nanosheet‐based flexible memristor with ultra‐low both operating voltage and power consumption by a simple spin coating process. Combining various characterization techniques and data analysis, the memristor and the cotton fiber piezoresistance sensor can be used to integrate a flexible biomimetic sensing system, which can remember and recognize the gestures.