Authors
Wenhao Wang, Zhengnan Fang, Yingying Zheng, Wei Lin, Chengyang Ying, Haoliang Lou, Bo Yao, Qinyong Dai, Zebo Fang
Published in
Materials horizons. Nov 12, 2026. Epub Nov 12, 2026.
Abstract
Neuromorphic systems are increasingly moving toward integrating multiple operating modes into a single device. However, existing oxide-based optoelectronic synapses typically rely on pulse strength to switch from short-term plasticity (STP) to long-term plasticity (LTP), which limits the adaptability of a single device to different computational tasks. Here, high-performance HfLaO (HLO)/InGaCdO (IGCO) optoelectronic synaptic transistors are fabricated via full magnetron sputtering, with a steep subthreshold swing (SS) of 96 mV decade-1 and an ultrahigh on/off ratio of 2 × 108. Reversible switching between STP and LTP is realized through gate bias tuning under fixed optical excitation without electrical performance degradation. Stable discrete multilevel conductance states are acquired under an ultralow read voltage of 1 mV and 0.0645 mW cm-2 optical pulses, which supports low-power fully connected neural networks for high-accuracy MNIST handwritten digit recognition. Negative gate bias reproduces typical STP characteristics including paired-pulse facilitation (PPF), and gate-adjustable STP relaxation dynamics construct in-sensor physical reservoirs for temporal signal analysis and four-frame human motion recognition. This all-inorganic oxide transistor exhibits exceptionally high electrical performance and operates at low power, while integrating dual synaptic modes to perform trajectory recognition and speech command recognition. It overcomes the limitations of conventional single-modulation strategies and provides practical hardware solutions for low-power multimodal devices.
PMID:
43016364
Bibliographic data and abstract were imported from PubMed on 12 Nov 2026.
Read full publication at:
Please sign in
to see all details.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 5
- Comments 0