Reconfigurable Multilevel\nStorage and Neuromorphic\nComputing Based on Multilayer Phase-Change Memory

In the era of big data, the amount of global data is\nincreasing\nexponentially, and the storage and processing of massive data put\nforward higher requirements for memory. To deal with this challenge,\nhigh-density memory and neuromorphic computing have been widely investigated.\nHere, a gradient-doped multilayer phase-change memory with two-level\nstates, four-level states, and linear conductance evolution using\ndifferent pulse operations is proposed. The mechanism of multilevel\nstates is revealed through high-resolution transmission electron microscopy\n(HRTEM) and finite-element analysis (FEA), which show that the sequential\nphase change among different sublayers is realized due to the different\nphysical properties of the sublayers with different doping concentrations.\nTaking advantage of the devices’ linear conductance evolution\ncharacteristic, a handwritten digit (28 × 28 pixel) recognition\ntask is implemented with a high learning accuracy of 93.46% by building\na simulated artificial neural network made up of this gradient-doped\nmultilayer phase-change memory. It is proved that this gradient-doped\nmultilayer phase-change memory is capable of both binary multilevel\ndigital storage and brain-inspired analog in-memory computing in the\nsame device, enabling reconfigurable applications in the future.