A generic compact model for resistive memories and switches.

Area of Science:

• Materials Science
• Electrical Engineering
• Computer Science

Background:

• Physics-driven models for resistive switching are complex and computationally expensive.
• Device-specific models limit scalability in large-scale simulations.

Purpose of the Study:

• To propose a generic compact modeling methodology for diverse resistive memory technologies.
• To enable scalable and accurate simulations for neuromorphic computing.

Main Methods:

• Developed a generalized framework using common circuit elements (nonlinear resistors, capacitors, switches) in LTspice.
• Integrated key switching phenomena: filament formation, polarization-modulated tunneling, spin-dependent resistance.
• Validated models with experimental data and performed Monte Carlo simulations for variability analysis.

Main Results:

• Successfully modeled Resistive Random-Access Memory (RRAM), Thermally-Switching (TS), Ferroelectric Tunnel Junctions (FTJ), and Magnetic Random-Access Memory (MRAM).
• Framework accurately reproduces device switching characteristics and captures variability.
• Demonstrated seamless scalability from single-device to array-level simulations.

Conclusions:

• The proposed methodology offers a scalable and computationally efficient solution for modeling various resistive switching devices.
• Enables accurate performance evaluation and power consumption estimation for neuromorphic and in-memory computing architectures.
• Broad applicability for next-generation computing systems.