| 초록 |
In this work, we presented our simulation results on the falling-time controlled for the switching mechanism which includes phase-change phenomena such as melting, quenching and annealing which are controlled by Joule-heating. The falling-time varied from 10 to 100 ns. We constructed a switching model for Phase change memory (PCM) by combining electro-thermal and phase-field models in a 2-D finite element analysis package, COMSOL Multiphysics. We analyzed the variability in resulting resistance that is caused by the stochastic nature of the nucleation. A comparative study was conducted on two cell architectures: one is heater-based and the other is self-heating. The device performance such as reset current, power consumption, on/off ratio, etc., along with cell resistance statistics and the recrystalline method by melting slow cooling pulse (MSCP) and long constant current pulse (LCCP) were compared. The scalability of PCM device was also investigated. From the obtained simulation results, it indicated that set operation could be done to transit from a-GST to fully crystalline by applied melt-slow cooling after long falling-time 100ns. SHW geometry was more beneficial 56% in power consumption than HBW. Another main point in our simulation was found that the cumulative distribution for resistance statistic is Lognormal for set state and Weibull for reset state. The istropy scalibility is more gradual and benefit than anisotropy. |
