Skip to content Skip to navigation

Reconfigurable 2D Materials with Neuromorphic Functionality

Solid-state electronics and advanced computation has spurred significant interest in artificial intelligence and neuromorphic (i.e., brain-like) computing. However, the deterministic correlations between input and action in conventional silicon microelectronics are not well-matched to information processing in biological systems. In particular, neuronal networks blur the lines between inputs and actions in that the extremely large number of connections (synapses) between neurons (>1000 synapses to 1 neuron) rather than the density of neurons establish the hierarchy of perception and action. By exploiting the two-dimensional (2D) geometry and rapid defect motion in monolayer polycrystalline MoS2, the Northwestern University MRSEC has realized the first experimental demonstration of a multi-terminal hybrid memristor and transistor (i.e., memtransistor). The seamless integration of a memristor and transistor into one multi-terminal device has the potential to simplify artificial neuronal network architectures in addition to providing opportunities for studying the unique physics of defect kinetics in 2D materials.

Nature, 554, 500-504 (2018).