High-Performance Edge-Line Contact Memristors with In-Plane Solid-Liquid-Solid Grown Silicon Nanowires for...

Area of Science:

• Materials Science
• Nanotechnology
• Computer Engineering

Background:

• Memristors are crucial for neuromorphic computing due to their resistive switching properties.
• Challenges in memristor technology include device uniformity and miniaturization for large-scale applications.

Purpose of the Study:

• To demonstrate scalable production of highly uniform memristors using silicon nanowires (SiNWs).
• To investigate the performance of these SiNW memristors in neuromorphic applications.
• To address critical challenges in uniformity and miniaturization for neuromorphic hardware.

Main Methods:

• Fabrication of quasi-one-dimensional diffusive memristors using heavily doped n-type SiNWs (∼50 nm diameter) via in-plane solid-liquid-solid (IPSLS) growth.
• Utilized an edge-line contact structural design for improved control over conductive filaments (CFs).
• Developed neurons with tunable sigmoidal probabilistic activation functions.

Main Results:

• Achieved highly uniform SiNW memristors with diameters as small as ∼50 nm.
• Demonstrated excellent self-compliance threshold switching characteristics: low operating voltage (∼0.8 V, SD=0.073 V), low leakage current (1 pA), high switching ratio (>10^7), ultrafast switching speed (∼8 ns), and low switching energy (47.2 fJ).
• Developed uniform neurons that achieved 96.2% accuracy in binary tumor classification tasks.

Conclusions:

• IPSLS-fabricated SiNW memristors effectively address uniformity and miniaturization challenges in neuromorphic hardware.
• The developed SiNW-based memristors show significant potential for large-scale integration in advanced neuromorphic computing systems.
• This work highlights a viable pathway for creating high-performance, uniform memristive devices for future computing architectures.