We present a computational framework that combines dynamical mean-field theory (DMFT) with density functional theory (DFT) and the nonequilibrium Green's function (NEGF) technique to study the steady-state transport properties of magnetic tunnel junctions (MTJs). The objective of our calculations is then to understand the impact of dynamical electron correlations on the Fe 3𝑑 orbitals of an Fe/MgO/Fe system. By applying the rigid shift approximation, we study both the zero- and finite-bias properties in a simple and computationally efficient manner, obtaining the bias-dependent electronic structure, the current-versus-voltage characteristic curve in both the parallel and antiparallel configurations, and consequently, the tunneling magnetoresistance (TMR) ratio. We find that dynamical electron correlation manifests as a reduction in the spin splitting of the Fe 3𝑑𝑧2 state compared to DFT predictions and introduces a finite relaxation time. The impact of these effects on transport, however, varies significantly between magnetic configurations and applied bias voltages. In the parallel configuration, the characteristic curves obtained with DFT and DMFT are similar up to large biases, as the transport is mostly due to the coherent transmission of spin-up electrons through the MgO barrier. Conversely, in the antiparallel configuration, correlation effects become more significant as the bias increases, with DMFT predicting a sharp current increase due to bias-driven inelastic electron-electron scattering. At zero bias, both DMFT and DFT similarly overestimate the TMR ratio compared to experiments. However, at finite bias, DMFT predicts a lower bias threshold for the suppression of the TMR relative to DFT, improving the agreement with experimental data and underscoring the importance of dynamical correlation effects on finite-bias behavior. While these conclusions are specific to Fe/MgO/Fe MTJs, our computational approach can be applied to other MTJs as well, thereby advancing the use of DMFT in spintronics.
Effect of dynamical electron correlations on the tunnelling magnetoresistance of Fe/MgO/Fe(001) junctions
Droghetti, Andrea
2025-01-01
Abstract
We present a computational framework that combines dynamical mean-field theory (DMFT) with density functional theory (DFT) and the nonequilibrium Green's function (NEGF) technique to study the steady-state transport properties of magnetic tunnel junctions (MTJs). The objective of our calculations is then to understand the impact of dynamical electron correlations on the Fe 3𝑑 orbitals of an Fe/MgO/Fe system. By applying the rigid shift approximation, we study both the zero- and finite-bias properties in a simple and computationally efficient manner, obtaining the bias-dependent electronic structure, the current-versus-voltage characteristic curve in both the parallel and antiparallel configurations, and consequently, the tunneling magnetoresistance (TMR) ratio. We find that dynamical electron correlation manifests as a reduction in the spin splitting of the Fe 3𝑑𝑧2 state compared to DFT predictions and introduces a finite relaxation time. The impact of these effects on transport, however, varies significantly between magnetic configurations and applied bias voltages. In the parallel configuration, the characteristic curves obtained with DFT and DMFT are similar up to large biases, as the transport is mostly due to the coherent transmission of spin-up electrons through the MgO barrier. Conversely, in the antiparallel configuration, correlation effects become more significant as the bias increases, with DMFT predicting a sharp current increase due to bias-driven inelastic electron-electron scattering. At zero bias, both DMFT and DFT similarly overestimate the TMR ratio compared to experiments. However, at finite bias, DMFT predicts a lower bias threshold for the suppression of the TMR relative to DFT, improving the agreement with experimental data and underscoring the importance of dynamical correlation effects on finite-bias behavior. While these conclusions are specific to Fe/MgO/Fe MTJs, our computational approach can be applied to other MTJs as well, thereby advancing the use of DMFT in spintronics.
I documenti in ARCA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
