ML-PCM: Machine Learning Technique for Write Optimization in Phase Change Memory (PCM)

Desai, Mahek; Quinn, Rowena; Asadinia, Marjan

Phase-change memory (PCM) materials are being developed for faster, non-volatile & high-density memory that can facilitate more efficient computation as well as data storage. The materials used for these PCM devices are usually chalcogenides that can be switched between their amorphous and crystalline phases thus producing orders of magnitude difference in the electrical resistivity [1, 2]. The operation of such devices is limited by elemental segregation and void formation, which occurs as a result of the extensive cycling. After crystallization, the structure gradually transforms from fcc to hexagonal. In the present work, we are studying these different phase changes in-situ as they occur in PCM materials basically using TEM imaging. The aim is to correlate device modeling and electrical characterization in order to improve the models and enable accurate, predictive simulations. The thin film materials and devices can be directly deposited onto Protochips devices, allowing controlled temperature changes while imaging in the TEM. Although the temperature change rate achievable is too slow as compared to the fastest PCM-device operation, these rates can provides valuable insights into the various property changes in the material and phase transformations as well. Both a Cs-image corrected Titan ETEM and a Tecnai F30 have been used for this study. The ETEM is equipped with a K2 direct electron detector camera allowing high-speed video recording of these PCM materials.

More Like this