A Robust and Efficient Compact Model for Phase-Change Memory Circuit Simulations

A phase-change memory (PCM) model for robust and efficient simulations of circuits including neuromorphic ones is reported in this work. The features of a hysteretic dynamic resistance in the voltage domain, and the incubation in the crystallization, are covered in the model. The Landau-Khalatnikov (LK)-type equation for ferroelectric is used to develop the PCM hysteresis module. A voltage-controlled relaxation oscillation is successfully simulated for the Ge2Sb2Te5 (GST) PCM. A technique of direct evaluation (DE) is then developed to reformulate the PCM model without any internal node. A significant enhancement of simulation efficiency is achieved compared with the traditional approach without sacrificing the accuracy. The functional correctness of the PCM device model and the acceleration effect in circuit simulations are verified.

Automated summary

This research presents a PCM model for robust and efficient simulations of circuits including neuromorphic ones, utilizing the LK equation to develop the PCM hysteresis module and successfully simulating voltage-controlled relaxation oscillation for GST PCM. The technique of DE is developed to reformulate the PCM model without sacrificing accuracy, resulting in a significant enhancement of simulation efficiency. The functional correctness of the PCM device model and the acceleration effect in circuit simulations have been verified.