Carnegie Mellon Logo

Amrinder Nain — 2002-03 Fellow

Amrinder Nain photo

There is a growing need for a nonvolatile memory technology for high-density stand-alone and embedded CMOS applications with faster write speed and higher endurance. Ovonic Unified Memory (OUM) utilizes a reversible structural phase change between amorphous and polycrystalline states in a GeSbTe chalcogenide alloy material. The transition is accomplished by heating a small volume of the material with a current pulse and results in a considerable change in alloy resistivity. Data thus stored on the basis of different resistivity values is read back by measuring the resistance of each cell. OUM is the most promising futuristic memory solution for portable devices (cell phones, mobile PC's, etc.) and promises low cost, low power solutions and is non volatile and easy to integrate in the existing CMOS process. OUM has demonstrated considerable advantages over competing technologies such as DRAM, SRAM, FLASH, FERAM and MTJ-RAM. Scaling is the key potential advantage of OUM, as write speed and write energy both scale with programmed media volume. Hence, OUM cell size scaling is limited predominantly by column and row pitch lithography. Presently, OUM test structures have been built using the 0.18 ?m lithography process and industry focus is to scale it down to 0.13 ?m and below. Scaling to those limits and multi-bit data storage efforts require management of proximity heating with declining cell space and accurate and quick estimation of electrical resistance with decreased power.

The goal of this project is pioneering a simulation tool, which gives a measure of the electrical resitivity of OUM cell elements by analyzing the heat transfer mechanisms and crystallization kinetics during read/write strategies. This project deals with developing analytical model and simulation software for determining the electrical resistivity of the OUM cell by heat transfer mechanisms and crystallization kinetics. This will be achieved by (i) analytical and numerical simulations of heat transfer in the amorphous/crystalline structure, (ii) analytical and numerical simulations for crystalline kinetics of nucleation and growth using the Johnson-Mehl-Avrami technique, (iii) foregoing two will be supported by experimental work, which (a) will provide thermal/material properties and (b) will validate analytical/numerical results obtained. Our ultimate goal is to pioneer reliable and rapid electrical/thermal simulation software, which will be useful towards successful implementation of this upcoming technology.