Physics & Astronomy Colloquium – Stewart

Speaker: Derek Stewart (San Jose Research Center, Western Digital Company)

Title: Optimizing MRAM: a computational exploration of spin Hall materials and 2D half metals

Abstract: While commercial spin-transfer-torque MRAM chips are now available, there are still key challenges (device size, power, reliability) that must be overcome to move this technology into a broader market. Optimizing materials and interfaces can help in solving many of these issues. Advances in numerical algorithms and access to cheap computing clusters has led to nimble simulation tools that can provide atomistic insight into materials. Industry is increasingly relying on these approaches to help guide materials design for future devices. In this talk, I will introduce the non-volatile memory simulation effort at Western Digital and discuss our recent work on optimizing material properties for spin-orbit-torque MRAM. I will focus on our recent studies on the high spin Hall angle in β-W and our search to find the thinnest half metal possible*.
Using a spin Hall material to flip a magnetic layer could lead to low-power MRAM devices. Recent studies have shown that β-W has a very large spin Hall angle (~0.30) [1]. While this large spin Hall angle is thought to be intrinsic, a direct link between the β-W electronic structure and the high spin Hall angle has not been established. We calculate the spin hall conductivity in β-W using the Kubo formula[2] based on Berry curvature determined with a relativistic KKR Green’s function approach [3]. I will discuss our results and how the crystal structure impacts spin Hall conductivity.
Fully spin polarized metals have long been sought out as an ideal electrode for high magnetoresistance devices. While half metal Heusler alloys have shown promise, these alloys can suffer from disorder and strain effects that limit their effectiveness [4]. A 2D half metal analogue could offer key advantages due to a lack of defects and dangling bonds. To rapidly identify potential half metals, we perform a broad search[5] of ~600 2D materials in the MaterialsWeb repository [6]. I will discuss the workflow used to identify realistic 2D magnetic materials. We find that only the iron dihalides (FeCl2, FeBr2, FeI2) remain half metallic. I will discuss the magnetic properties of these 2D materials, the challenges of integrating them into devices, and how they relate to their layered metamagnetic cousins.
*These projects have benefited greatly from collaborations with the University of Florida (Prof. Hennig’s Group) and the University of Bristol (Prof. Gradhand’s Group).

References:
[1] C. H. Pai et al, Appl. Phys. Lett. 101, 122404 (2012)

[2] M. Gradhand et al., Phys. Rev. B 84, 075113 (2011)
[3] M. Gradhand et al., Phys. Rev. B 80, 224413 (2009)

[4] M. Ashton et al., Nano Letters, 17, 5251 (2017).
[5] M. R. Page et al., J. Appl. Phys., 119, 153903 (2016).

[6] https://materialsweb.org

Speaker Biography: Derek Stewart joined the WD research division in the fall of 2014. His work focuses on using ab-initio approaches to explore electronic, spin, thermal, and ionic transport in new forms of non-volatile memory (e.g. MRAM, Phase Change, RRAM, CeRAM). Prior to joining WD, Derek directed the simulation effort at the Cornell Nanoscale Facility and also served as an Adjunct Professor in Material Science at Cornell University. At Cornell, he was part of the team that developed the first ab-initio approach to accurately calculate thermal conductivity in materials and nanostructures. Derek did his Ph.D. research (Physics, Univ. of Virginia) on giant magnetoresistance with Dr. Bill Butler at ORNL. During a post-doc at Sandia, Derek helped develop a first principles non-equilibrium Green’s function code to examine electronic transport in nanoscale devices, including magnetic tunnel junctions. Derek has organized numerous simulation workshops, published 50 articles and 1 patent, and given over 40 invited talks.