Atomistic Simulations of Synthetic Antiferromagnetic MRAM

Increasing demand for high quality, low power, and non-volatile memory in modern computing has positioned Spin-Transfer Torque Magnetic Random Access Memory (STT MRAM) as a promising replacement for conventional RAM in the future. STT-MRAM is composed of two ferromagnetic layers separated by a non-magnetic spacer layer that acts as a Magnetic Tunnel Junction (MTJ). This MRAM faces scalability limitations with high-density configurations suffering from magnetic “cross talk” where the stray magnetic fields from one MRAM cell influence magnetic moments in a nearby MRAM cell. Replacing the typical ferromagnetic free layer with a synthetic antiferromagnetic SAF free layer coupled through the Ruderman–Kittel-Kasuya–Yosida (RKKY) interaction. This mitigates cross talk through dipole field cancellation and accelerates switching.

This thesis investigates the SAF MRAM performance using the atomistic simulation package VAMPIRE which is a software that allows the investigation of interfacial effects such as the RKKY exchange and non-uniform thermal nucleation in nanoscale MRAM discs. The simulations conducted in this thesis establish a parameter space for the minimum requirements for reliable operation. These requirements are found by employing a parameter sweep with the Minimal Working Example (MWE), a 1 × 1 × 6 nm cell, on RKKY exchange strength, critical switching field, and the polarisation that reaches the second free layer. Different physical parameters such as anisotropy, and thermal stability are also investigated to evaluate the theoretical functionality of the SAF MRAM as an effective device. When scaling to 40 nm-diameter discs the switching moves from coherent rotation to a non-uniform nucleation-driven process, as seen for standard STT-MRAM.

In conclusion, this thesis provides the necessary atomistic framework and parameter regimes for engineering high-quality, low-power, and non-volatile STT-SAF MRAM, and demonstrates the need for non-continuum modelling methods for accurate device simulation.