Parylene Based Memristive Devices with Multilevel Resistive Switching for Neuromorphic Applications

Area of Science:

• Materials Science
• Neuroscience
• Computer Science

Background:

• Memristive devices are crucial for next-generation computing.
• Organic materials offer biocompatibility and cost-effectiveness.
• Neuromorphic computing mimics the brain's structure and function.

Purpose of the Study:

• To investigate the resistive switching and neuromorphic behavior of parylene-based memristive devices.
• To evaluate the performance of these organic memristors for potential AI and biomedical applications.
• To demonstrate the feasibility of implementing neuromorphic networks using these devices.

Main Methods:

• Fabrication of Metal/Parylene/ITO sandwich structures with various metal electrodes (Ag, Al, Cu, Ti).
• Characterization of resistive switching properties, including switching voltage, OFF/ON ratio, and retention.
• Experimental demonstration of spike-timing-dependent plasticity (STDP) for device training.
• Implementation of a neuromorphic network model for classical conditioning.

Main Results:

• Parylene memristors exhibited low switching voltage (≤1 V) and high OFF/ON resistance ratios (up to 10^4).
• Devices demonstrated excellent retention (≥10^4 s) and multilevel resistance switching (≥16 states with Cu electrodes).
• Biologically inspired STDP mechanism was successfully implemented for training parylene memristive elements.
• A simple neuromorphic network model based on classical conditioning was realized.

Conclusions:

• Parylene-based organic memristors offer excellent performance and biocompatibility.
• These devices are suitable for hardware realization of spiking artificial neural networks.
• The study highlights the potential of parylene memristors for supervised/unsupervised learning and biomedical applications.