Ultrathin bilayer biomolecular memristors based on molecular layer deposition (MLD) are fabricated to achieve high-performance neuromorphic computing. The device with a TiN/Ti-cysteine/Al-cysteine/Pt structure exhibits reproducible nonvolatile bipolar resistive switching with an on/off ratio (>102), excellent retention (>105 s), and low operating voltages (-1.28 V/+2.10 V). Improved uniformity and multistate controllability stem from a charge trapping/detrapping-assisted conductive filament mechanism, while the Al-Cys layer enhances resistive switching and retention characteristics, the Ti-Cys layer stabilizes and regulates conductive filament formation. Critically, the memristor emulates key synaptic functionalities, including long-term potentiation/depression (LTP/LTD), paired-pulse facilitation/depression (PPF/PPD), and spike-rate-dependent plasticity (SRDP). When deployed in a neural network for MNIST handwritten digit recognition, it achieves 93.7% accuracy. This work resolves limitations of single-layer biomemristors by synergistic bilayer design, advancing biohybrid electronics for energy-efficient neuromorphic hardware.
Zhu et al. (Wed,) studied this question.