Oxide-Based Optical Synapses with Low Consumption for In-Sensor Reservoir Computing

In this work, we propose a two-terminal ITO/NiO/HfO 2 /Pd optical synapse that has tunable temporal dynamics under light stimuli. The optical synapse exhibits light intensity and frequency-dependent nonlinear synaptic short-term plasticity (STP) at a low read voltage of 0.05 V, indicating the potent...

Full description

Saved in:
Bibliographic Details
Published in2025 9th IEEE Electron Devices Technology & Manufacturing Conference (EDTM) pp. 1 - 3
Main Authors Zhou, Jingyi, Jiang, Biyi, Tang, Huansong, Wang, Zhen
Format Conference Proceeding
LanguageEnglish
Published IEEE 09.03.2025
Subjects
Online AccessGet full text
DOI10.1109/EDTM61175.2025.11040532

Cover

Loading…
More Information
Summary:In this work, we propose a two-terminal ITO/NiO/HfO 2 /Pd optical synapse that has tunable temporal dynamics under light stimuli. The optical synapse exhibits light intensity and frequency-dependent nonlinear synaptic short-term plasticity (STP) at a low read voltage of 0.05 V, indicating the potential for achieving in-sensor reservoir computing (RC) with reduced power consumption. The RC system, utilizing our optical synapses, can operate at an ultra-low energy consumption level of 15 fJ while achieving a remarkably high recognition accuracy of 95.95 %.
DOI:10.1109/EDTM61175.2025.11040532