Book Description
Information storage and accessibility form the foundation of many modern electronic systems. High speed nonvolatile memory (NVM) technologies retain data without consuming power, driving the rapid growth of the portable consumer electronics market. Phase change memory (PCM) is an emerging NVM offering exceptional speed, storage density, and cycling endurance. This work uses novel multiphysics models to quantify the importance of thermal phenomena in PCM. It extends optical thermometry techniques to resolve the thermal transport physics critical for device functionality. Fully coupled finite element calculations capture the electrothermal and phase change processes in a confined cell device. The simulations demonstrate the critical role thermal boundary resistance (TBR) plays in reducing programming current. This result suggests that interface engineering can significantly reduce programming current. Compact electrothermal models use reduced cell geometries to accurately predict scaling in a variety of device geometries. These models demonstrate that the distribution of thermal resistances is the key design parameter for reducing the programming current. Nanosecond transient thermoreflectance (TTR) measurements on a variety of chalcogenide stoichiometries show that the effective thermal conductivity depends primarily on the material phase. This work extends nanosecond TTR up to 340°C to measure the thickness and temperature dependent effective thermal conductivity of GeSbTe (2:2:5) (GST) in the as-deposited, fcc, and hcp phases. Process dependent material defects, partial crystallization, and TBR all significantly alter the effective thermal conductivity. Picosecond time-domain thermoreflectance (TDTR) measurements establish the Al/TiN TBR, TiN/fcc GST TBR, and intrinsic fcc GST thermal conductivity up to 325°C. An original multi-sample, thickness-implicit data extraction technique uniquely separates the spatial distribution of thermal properties. The intrinsic conductivity increases slowly with temperature, consistent with materials with high defect and vacancy concentrations. The TiN/fcc GST TBR dominates the device thermal resistance and is the key factor determining the programming current.