On the Role of Intermediate States for Artificial Spin Ice Inspired Computation

Abstract Conventional approaches to computing with nanomagnets require an initial state, and a final state corresponding to an expected outcome. Assuming a single spin-flip regime, an energetic relaxation from the initialized state to that of a final low energy state will include intermediate states with variable magnetostatic energies, which affect the probability of an outcome. Here, we investigate intermediate states in a simple nanomagnet toy model consisting of four nanomagnets arranged onto a square plaquette. In doing so, we are able to systematically demonstrate design approaches to mitigate (or leverage) computational “errors” in artificial spin ice inspired computational logic. We reveal an explicit relation between geometry and energetics by way of multipolar analysis, and comment specifically on the nature of energy relaxation pathway with direct consequences to designing logic units for conventional and unconventional computing applications. We support our theoretical models with experimental results on field-induced relaxation using Magnetic Force Microscopy measurements. Our work provides an intuitive way to design logic elements for future computing applications with artificial spin ice structures.