Nonlocal Detection of Vector Spin Accumulation in Graphene for Probabilistic Computing

The deceleration of Moore’s law has led to an increasing demand for unconventional computing approaches. Probabilistic bits (p-bits) using the Heisenberg model as the energy-based model (EBM) represent a cornerstone in the field of probabilistic computing, enabling stochastic sampling, probabilistic inference, and combinatorial optimization [1]. The Heisenberg interaction, which describes the exchange interaction between localized spins, can indirectly influence angle-dependent spin transport measurements in a non-local spin valve. However, directly observing Heisenberg interactions through these measurements requires a detailed understanding of how these interactions affect spin dynamics and spin transport properties.<br/>In this work, we consider two different orientations of spin injection into a graphene channel and analyze their interactions. We experimentally demonstrate the vector spin accumulation through 2D graphene channels using lateral spin valve by varying polarization direction of spin accumulation in ferromagnetic spin injectors. The results realize the angular control of spin and readout the interaction between two spin states [2]. This demonstration paves the way for mapping Heisenberg model with stochastic low barrier magnets as vector components for scalable probabilistic spin circuits.<br/> <br/>[1] S. Bunaiyan, S. Datta, and K. Y. Camsari, “Heisenberg machines with programmable spin-circuits.” arXiv, Dec. 03, 2023. Accessed: Dec. 05, 2023. [Online]. Available: http://arxiv.org/abs/2312.01477<br/>[2] T. Kimura, Y. Otani, and P. M. Levy, “Electrical Control of the Direction of Spin Accumulation,” <i>Phys. Rev. Lett.</i>, vol. 99, no. 16, p. 166601, Oct. 2007, doi: 10.1103/PhysRevLett.99.166601.