Operation-robust SiO₂-based memristive application in air.

Area of Science:

• Materials Science
• Computer Engineering
• Neuroscience

Background:

• Neuromorphic computing offers high parallelism for post-Moore era efficiency.
• Synaptic devices are crucial for hardware neuromorphic systems.
• Silicon dioxide (SiO2) trapping memristive devices are CMOS-compatible but unreliable in air due to oxygen's electrochemical activity.

Purpose of the Study:

• To develop a robust SiO2 trapping-based memristive device for operation in ambient air.
• To investigate the role of ultrathin silicon nitride (Si3N4) as an isolation layer.
• To optimize Si3N4 thickness for reliable synaptic device programming and functionality.

Main Methods:

• Fabrication of SiO2 trapping-based memristive devices with ultrathin Si3N4 isolation layers.
• Systematic study of Si3N4 thickness dependence on device performance.
• Analysis of the isolation mechanism concerning electron and oxygen molecule transport.
• Mimicking synaptic plasticity modes using the optimized device.

Main Results:

• A robust memristive device based on SiO2 trapping was successfully operated in air using a Si3N4 isolation layer.
• 7 nm Si3N4 thickness was identified as optimal for reliable and flexible programming.
• An inherent isolation mechanism selectively blocks oxygen molecules while allowing electron transport.
• Various synaptic plasticity modes were successfully mimicked using the 7 nm Si3N4 device.

Conclusions:

• The integration of Si3N4 as a physical isolation layer significantly enhances the robustness of SiO2 trapping memristive devices in ambient air.
• The optimized 7 nm Si3N4 layer provides a selective barrier against oxygen, ensuring device stability.
• These findings expand the application prospects of SiO2 trapping memristors in neuromorphic computing and other fields requiring air-stable operation.