Modeling thermal effects in STT-MRAM

Abstract

We employ the stochastic Landau–Lifschitz–Gilbert (sLLG) equation to explore thermal effects on switching in the spin-transfer torque magnetoresistive random access memory (STT-MRAM). The distribution of the switching times depends on the meshes used for the discretization within the finite element method (FEM) implementation and we introduce an effective temperature scaling in the thermal field calculation to mitigate the switching time distribution dependencies on the element size. Furthermore, we investigate the switching statistics of the STT-MRAM at different temperatures and show that the switching time distribution has a lower mean value, but possesses a longer tail of long switching times for the higher operating temperatures. As a result, the STT-MRAM switching with a fixed voltage pulse duration becomes more error-prone at elevated temperatures.