Broadband photodetection and artificial visual synapses enabled by the photovoltaic and photoconductive effects of 2H-MoTe2/WSe2 heterojunction

Abstract

Neuromorphic computing, which simulates biological synaptic plasticity to achieve efficient information processing, is seen as a key solution to the computational power and energy efficiency limitations of traditional von Neumann architecture. However, existing synaptic devices face major challenges like high energy consumption, unstable non-volatile storage, and limited multi-modal response capabilities, severely restricting their practical application. Two-dimensional materials with atomic-scale thickness, high carrier mobility, and excellent responsiveness offer a new model for developing low-power, high-performance, and adaptive synaptic devices, while their heterostructures can synergistically process multi-modal signals like light and electricity to enable precise modulation of brain-like plasticity. This work proposes a self-powered photodetection and optoelectronic synaptic device based on 2H-MoTe₂/WSe₂ heterojunctions, which operates in the visible to near-infrared spectrum (405–1550 nm). An ultra-low optical signal with an optical power density of 10 μW cm⁻² can be detected in the near-infrared light at 1064 nm with a high responsivity (R) of 11.74 mA W⁻¹. In addition, the R of the heterojunction under 405, 660, 808, and 1550 nm light are 13.73, 24.03, 7.57, and 6.65 × 10⁻⁴ mA W⁻¹, respectively. Moreover, the heterojunction exhibits broadband synaptic properties and a minimum power consumption of 90 fJ for one spike. The paired-pulse facilitation (PPF) index of 52.6% is achieved with two consecutive optical pulse stimulations (0.4 s interval). The excellent performance of this heterojunction provides an innovative solution for next-generation low-power visual sensing systems and artificial neural networks.