Bioinspired multisensory fusion using a MoTe2 optoelectronic memristor with oxygen plasma treatment for in-sensor reservoir computing

Abstract

Biological multimodal perception systems play a pivotal role in environmental interactions through the sophisticated integration of multisensory information. Inspired by this natural paradigm, we demonstrate a breakthrough two-dimensional MoTe₂-based optoelectronic memristor capable of synergistically processing infrared optical and electrical signals in a monolithic device – a critical advancement toward artificial multimodal sensing systems. The developed structure demonstrates superior biorealistic synaptic functionalities, including tunable short-term plasticity, paired-pulse facilitation, and spike-time-dependent plasticity through photoelectronic co-modulation. More significantly, we construct a multimodal reservoir computing architecture that synergistically combines optical and electrical inputs, achieving higher pattern recognition accuracy compared to the single mode. This work establishes a new dimension in neuromorphic hardware design through inherent multimodal signal fusion capabilities. Our findings provide fundamental insights into photoelectronic coupling mechanisms while demonstrating practical pathways toward high-efficiency neuromorphic computing systems with biological sensory integration.