The UltraLowEMM project which was a tri-lateral collaboration between research institutions in the USA, Northern Ireland and the Republic of Ireland with a number of experimental and theoretical groups involved. The goal of this collaboration is that of achieving higher energy efficiency of novel nanoscopic magnetoelectric devices, with applications in data storage, memory and computing. As part of the theoretical effort my job was to model the magnetic interactions and describe the physical processes which result from the flow of electrical current in newly-conceived nanoscopic devices with atomically-defined chemical composition, magnetic tunnel junctions (MTJ) like this (schematic of an Fe/MgO/Mn3Ga MTJ), as candidates for novel optimised STT-MRAM cells (bits):
The operation of contemporary MRAM cells is underpinned by two basic Spintronic effects. On one hand is the tunnelling magnetoresistance effect (TMR), which allows the electrical readout of the magnetic state (i.e. the alignment of its magnetisation vector) of a thin (free) ferromagnetic (FM) layer, when compared to a reference (fixed) FM layer in a nanoscopic FM|(Insulating spacer)|FM stack. The latter is called a magnetic tunnel junction (MTJ) and forms the base of the MRAM bit. On the other hand, the MTJ bits are switched by passing electrical current which interacts with the magnetisation of the free FM layer via the spin-transfer torque (STT) effect. As opposed to the early-days magnetic-field-induced switching, the STT allows scalable miniaturisation and energy efficiency. However, this switching still requires large current densities (≥10^10 A/m^2) which generate significant energy losses due to Joule heating and hinder the usage of MRAM for embedded memories and on-chip logic.
The plan of UltraLowEMM was to address previously unexplored material systems in search for breakthrough solutions that will dramatically reduce this energy dissipation in MRAM. Using FM free layers, which carry large magnetic moments and couple to intrinsic magnetic fields, as bit encoders is seen as a major contributor for the high switching currents. Hence, devices containing instead low-moment ferrimagnets (FiM) or zero-moment antiferromagnets (AFM) are expected to achieve a large improvement in the switching dissipation. Thus, the emerging field of Antiferromagnetic Spintronics holds a great promise for further discoveries, with potential impact beyond the contemporary MRAM and magnetic logic technologies.
In the Computational Spintronics Group, led by my mentor Prof. Sanvito, we have been developing a state-of-the-art quantum transport code (SMEAGOL) to study, from first principles electronic structure theory, the electron transport in nanoscopic multi-layered spintronic devices. SMEAGOL can now calculate atomically-resolved STT in MTJ stacks from ab initio theory. We have applied the method for topical AFM-MTJs. Recently we have completed a number of large-scale calculations for STT in FiM-MTJs over 10 nm thick (the self-consistent region). Publications are in preparation.
