Investigation of\nthe Versatile Utilization of Three-Dimensional\nVertical Resistive Random-Access Memory in Neuromorphic Computing

The continuous advancement of computing technologies\nsuch as the\nInternet of Things and artificial intelligence emphasizes the need\nfor innovative approaches to data processing. Faced with limitations\nin processing speed due to the exponentially increasing volume of\ndata, conventional von-Neumann computing systems are transforming\ninto a new paradigm called neuromorphic computing, inspired by the\nefficiency of the human brain. We fabricate three-dimensional vertical\nresistive random-access memory (VRRAM), which is highly suited for\nneuromorphic computing, and demonstrate its value as an artificial\nsynapse. Beyond simulating simple synaptic functionalities such as\nspike-rate-dependent plasticity, spike-timing-dependent plasticity,\nand paired-pulse facilitation, we propose specific applications and\nexperimentally implement them. In pattern recognition simulations\nbased on the weight update characteristics of the fabricated VRRAM,\nthe accuracy achieved in pattern recognition using appropriate pulse\nschemes reaches 90.4%. Additionally, we demonstrate adaptive learning\nbehavior on the device by mimicking Pavlov’s dog experiment\nwith combinations of applied voltage pulses. Finally, we employ suitable\nwrite/erase pulse trains to implement binary representations for decimal\nnumbers ranging from 0 to 15, thereby illustrating the significant\npotential of local devices for edge computing applications.