Efficient modeling algorithm and design-space exploration for double-free-layer magnetic tunnel junctions
Phys. Rev. Applied 26, 014108 – Published 31 July, 2026
DOI: https://doi.org/10.1103/xlfp-z2hf
Abstract
The double-free-layer magnetic tunnel junction has emerged as a promising structure for high-density memory applications due to its enhanced thermal stability. However, optimizing its multiple design parameters through experimental trial and error is prohibitively time-consuming, while existing numerical tools remain inadequate for solving the coupled transport-magnetization dynamics problem. To achieve accurate and efficient optimization, an integrated coupled model combining the transport process and magnetodynamics is proposed, achieving a validated speed-up factor of 117.1 and enabling device-level parameter optimization, such as ferromagnetic coupling strength and thermal stability. The transport process is modeled based on nonequilibrium Green’s function method, accelerated by two numerical techniques: (a) a mixed sampling strategy for energy integrals involving sharp peak-shaped function and (b) a complex integral approach to bypass singularities in energy-resolved functions. For magnetization dynamics, the variable-step 1(2)-order Runge-Kutta method is employed to compute the time evolution of the free-layer magnetizations. With the coupled model, the design space is systematically explored using both single-objective optimization and Bayesian optimization. A minimum switching time of 0.846 ns is achieved at the optimal parameter setting.
