Auristor: Self-Learning Edge States from Chirality-Breaking Growth of Graphene Nanoribbons–A Computational Study of Resonant Scattering and Memcapacitive Behavior

Neuromorphic computing needs to be revisited with the inclusion of quantum mechanics to improve the logic and memory performance, to lower energy consumption, and to better embrace the stochastic nature of neurons. We study an edge vacancy-defected armchair graphene nanoribbon’s growth one carbon at a time. We reveal its learning and memory characteristics through the resonant scattering effect using computational methods of the tight-binding model, density functional theory, and nonequilibrium Green’s functions. We found a platform that can exhibit learning and memory due to resonant scattering, and it can be used as a probabilistic bit due to the edge spin density difference varying in time. We further relate the memristive character to the memcapacitive behavior of the growth, which will make these systems more promising for low-energy-consumption applications.