Comparison of pinning potentials, depinning currents and scalability for domain wall-based synapses with various pinning structures

Domain wall (DW) based magnetic tunnel junctions (MTJ) are one of the promising candidates for memristive artificial synapses. Various domain wall tracks have recently been studied, and different geometric configurations for pinning domain walls have been proposed. Geometric pinning helps to control the domain wall (DW) motion along the track and quantizes the conductance of the DW-MTJ to specific values. This work compares various geometric pinning options in terms of thermal stability, depinning currents, and programming reliability under random thermal fluctuations for scaled device dimensions. Results show that all pinning configurations can meet the thermal stability limits required for non-volatile operation if the device is designed appropriately. However, depending on the pinning geometry, depinning current requirements can vary considerably. Reliable programming with good linearity can be achieved for all of the studied configurations. Additionally, one-dimensional pinning potentials suitable for a compact model of DW motion are obtained for each pinning geometry. The results could provide valuable guidelines for designing DW tracks with scaled dimensions, thermally stable quantized conductance states, and highly reliable programming behavior.