Answers are grounded in this paper and cite their sources.
Demonstrates synaptic functionality in an optoelectronic synapse, suggesting improved energy efficiency for computing.
An optoelectronic synapse (OES) is a key component for brain‐inspired computing and holds great promise towards energy‐efficient computation, where energy consumption remains one of the key challenges. Herein, we report a metal (In) vacancy in Indium Selenide (In 2 Se 3 ) that induces electronic defects owing to its transition from amorphous to crystalline γ‐phase during thermal annealing. These In vacancies facilitate charge trapping and slow release, enabling persistent photoconductivity, and the corresponding OES shows a long decay time of ∼214 s. The OES also demonstrates important synaptic characteristics, for example, paired‐pulse facilitation, short‐term and long‐term memory, synaptic plasticity, and learning‐forgetting behavior. We show that due to the crystallization of In 2 Se 3 film, the electrical conductivity is enhanced by ∼10 4 times, which made it possible to realize the synaptic phenomena with a remarkably low energy consumption of ∼180 aJ per spike, well below the energy range of biological systems. Further, our device has realized an exceptional accuracy of ∼94% for the MNIST handwritten digit recognition using artificial neural network simulation, while an accuracy of 85% has also been realized for a U‐Net structure‐based network for semantic segmentation work. These findings open new pathways towards energy efficient OES for next generation neuromorphic computing and vision systems.
No takes yet. Share an insight, caveat, or question.
Dhara et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: