Wrinkled Rhenium\nDisulfide for Anisotropic Nonvolatile\nMemory and Multiple Artificial Neuromorphic Synapses

Two-dimensional\nmaterials are emerging as potential solutions\nfor\nhigh-density nonvolatile memory and efficient neuromorphic computing.\nHowever, integrating multidimensional memory and an ideal linear weight\nupdating synapse in a simple device configuration to achieve versatile\nbiomimetic neuromorphic systems remains challenging. Here, we introduce\na wrinkled rhenium disulfide (ReS<sub>2</sub>) transistor, where the\nwrinkled structure facilitates the carrier trapping/detrapping at\nthe dielectric interface, thus enabling the fusion of nonvolatile\nmemory and both electronic and optoelectronic synaptic functionalities.\nAs a nonvolatile memory, anisotropic wrinkled ReS<sub>2</sub> can\nyield three distinct sets of data across three crystal orientations\nunder identical programming operations. Each set demonstrates exceptional\nretention and endurance properties. As a neuromorphic synapse, it\nrealizes the linear and symmetric updates of conductance states up\nto 9 bits and 8 bits, the ultra-low-energy consumption of 75 fJ and\n2.5 pJ under the electrical and optical stimuli, respectively. The\nartificial neural network (ANN) based on electronic synapses gives\na superior recognition accuracy of 92.9% for the original handwritten\ndigits. The anisotropic synaptic responses and multiwavelength sensitivities\nof optoelectronic synapses enable them to execute advanced memory\nand recognition functions for complex images that encompass a variety\nof pattern features or color information. This underscores its substantial\npotential for integration into efficient biomimetic visual systems.