A team from the Changchun Institute of Optics, Fine Mechanics and Physics, Chinese Academy of Sciences, has created a Ta₂NiSe₅/SnS₂ heterojunction-based optoelectronic synaptic transistor capable of mimicking human visual functions across ultraviolet (UV) to near-infrared (NIR) wavelengths. Published in Light: Science & Applications, their work addresses key challenges in neuromorphic computing, such as limited responsivity and data retention in NIR detection. Neuromorphic sensors aim to replicate biological vision by integrating perception, processing, and memory into a single device. However, most existing optoelectronic synapses struggle with NIR detection due to low photon energy. The team introduced a physisorption-assisted persistent photoconductivity (PAPPC) effect, where adsorbed gas molecules (e.g., O₂ and H₂O) on SnS₂ extend carrier lifetimes, enhancing NIR sensitivity. The heterojunction design further boosts performance by leveraging Ta₂NiSe₅'s narrow bandgap (0.33 eV) for NIR absorption and SnS₂'s defects for charge trapping.
The device demonstrated remarkable metrics: a responsivity of 5.6×10³ A/W (visible) and 14.4 A/W (NIR), detectivity up to 4.1×10¹⁴ Jones, and a 1.7×10⁶% external quantum efficiency. It emulated synaptic functions like excitatory postsynaptic currents (EPSC), paired-pulse facilitation (PPF), and transitions from short- to long-term plasticity (STP/LTP). Gate voltage modulation enabled emotion-like learning control, while optical/electrical pulses replicated Pavlovian associative learning. A simplified human visual system model also achieved color recognition and memory.
This breakthrough paves the way for energy-efficient, all-in-one neuromorphic sensors for applications like autonomous vehicles and biomedical imaging. Future work may explore multimodal (e.g., vision-olfaction) sensing using physisorption effects.