Document Type

Article

Major

Electrical & Computer Engineering

Mentor

Prof. Ali Gokirmak, Dept. of Electrical & Computer Engineering, and Prof. Jake Scoggin, School of Computing

Disciplines

Data Storage Systems | Electrical and Computer Engineering

Abstract

This project builds a simulation foundation for selective cell heating in a phase-change memory (PCM) crossbar using Ge2Sb2Te5 (GST) as the active material. Using COMSOL Multiphysics® a 3D modeling software, couples Electric Currents, Electric Circuits, Heat Transfer in Solids, and Electromagnetic Heating for the simulation. A parameterized Tungsten (W)/GST-Amorphous/GST-Crystalline(phases) /W embedded in Silica Dioxide (SiO2) and surrounded in Silica Nitride (Si3N4) is validated at the single-cell level and scaled to small GST crossbars A terminal voltage (V_active/V_inactive, or 0 V if unselected) is applied through MOSFET and diode selector elements at the ends of each word line and bit line. These selectors control current flow and confine heating to the targeted cell while minimizing impact on neighboring cells. Simulation results show that milliamp-level current and Joule heating localize to the selected junction, raising its temperature to approximately 325 K, while surrounding cells remain near ambient with minimal thermal crosstalk. These findings confirm effective electrical steering and thermal isolation, providing a validated baseline for extending the platform to include latent-heat and crystallization kinetics, time-dependent pulse trains, and larger arrays aimed at multi-bit-per-cell operation in Artificial Intelligence hardware.

Comments

The authors thank the Holster Scholars Program, the University of Connecticut Honors Program, and the UConn Foundation for funding.

Download

Share

COinS