A physics-based statistical model for reliability of STT-MRAM considering oxide variability

A physics-based statistical model considering oxide thickness (T_ox) variability is proposed for evaluating the impact of time-dependent dielectric breakdown (TDDB) on the performance of spin-transfer torque magneto-resistive random access memory (STT-MRAM). The statistics of breakdown events are captured by the percolation theory, physics-based analytical model for successive break down (BD) and 1-D non-equilibrium Green´s function (NEGF). Using the proposed model, we examine T_ox-dependence of STT-MRAM performance distribution, such as tunneling magneto-resistance ratio (TMR) and critical current (I_C). Simulation results clearly show that oxide variability needs to be taken into account for better lifetime prediction. The proposed model has been validated with experimental data.