Ultralow Power Optoelectronic Reconfigurable Hf0.2Zr0.8Ox-Based Antiferroelectric Device for Adaptive Image
Chen Lu1, Pei Liu1, Kangli Xu1
1College of Integrated Circuits & Micro-Nano Electronics, School of Microelectronics, Nano Institute of Fudan University, Fudan University, Shanghai 200433, China.
ACS Applied Materials & Interfaces
|May 5, 2026
Summary
Researchers developed reconfigurable artificial neurons and synapses using antiferroelectric thin films. This innovation enables faster, more efficient neuromorphic systems for tasks like image recognition with reduced power consumption.
Area of Science:
- Materials Science
- Neuroscience
- Electrical Engineering
Background:
- Reconfigurable electronics are key for compact neuromorphic systems.
- Current devices face challenges in complex switching between neurons and synapses.
Purpose of the Study:
- To demonstrate on-demand creation of artificial neurons, synapses, and memory capacitors.
- To utilize antiferroelectric Hf0.2Zr0.8Ox thin films for reconfigurable devices.
- To enable reversible reconfiguration via optical/electrical hybrid stimuli.
Main Methods:
- Fabrication of artificial neurons, synapses, and memory capacitors using Hf0.2Zr0.8Ox thin films.
- Application of optical/electrical hybrid stimuli for device reconfiguration.
- Integration of reconfigurable components for adaptive image recognition.
Main Results:
- Demonstrated on-demand, reversible reconfiguration of device components without delay.
- Achieved low energy consumption (90 aJ/spike), high endurance (>1011), uniformity, and stability for neurons.
- Adaptive image recognition using reconfigurable components outperformed existing methods.
- Reduced on-chip power consumption by 42.1% without accuracy loss.
Conclusions:
- Antiferroelectric Hf0.2Zr0.8Ox thin films enable efficient, reconfigurable neuromorphic hardware.
- Hybrid stimuli offer a promising approach for advanced neuromorphic systems.
- This technology paves the way for energy-efficient neuromorphic computing.
