Paper

Non-local spin-transport structures and governing length-scales

Abstract

Non-local spin-current conduction can be used to concentrate spin-current in a nanostructure in spin-torque switchable nanomagnets in solid-state memory. This is sometimes called spin-harvesting or spin-funneling. We use a simplified building block for understanding spin-harvest transport and some associated physical length-scales, identifying key materials and interface conductance parameters. A simple rotationally symmetric non-magnetic spin-conductor is used for solving its corresponding drift-diffusion transport equation, and for investigating the role finite interface spin-RAs play in affecting the spin-harvesting length-scale. A finite element model (FEM) is used to illustrate quantitative size and contact resistance dependent spin transport, followed by discussion of a simplified sheet-resistance-limit solution for illustrating a characteristic lateral length-scale that governs the spin-harvesting distance. Two key findings from this study are: (1) the effective spin-harvesting range is bound by spin drift-diffusion length λsfNM but usually shorter, and (2) There is a strong interplay among spin-flip diffusion length λsfNM and the bulk and interface spin-conductance of the NM. The more conducting the bulk NM is, and the more resistive a bottom interface RA is for spin-conduction, the longer range one can efficiently harvest spin-current. These results provide guidance for device structure designs utilizing spin-harvesting to maximize spin-current coupling into nanomagnets for spin-torque switching in magnetic random-access memory (STT-MRAM).