Realization of tunable artificial synapse and memory based on amorphous oxide semiconductor transistor

Abstract

AbstractRecently, advanced designs and materials emerge to study biologically inspired neuromorphic circuit, such as oxide semiconductor devices. The existence of mobile ions in the oxide semiconductors could be somewhat regarded to be similar with the case of the ions movements among the neurons and synapses in the brain. Most of the previous studies focus on the spike time, pulse number and material species: however, a quantitative modeling is still needed to study the voltage dependence of the relaxation process of synaptic devices. Here, the gate pulse stimulated currents of oxide semiconductor devices have been employed to mimic and investigate artificial synapses functions. The modeling for relaxation process of important synaptic behaviors, excitatory post-synaptic current (EPSC), has been updated as a stretched-exponential function with voltage factors in a more quantitative way. This quantitative modeling investigation of representative synaptic transmission bias impacts would help to better simulate, realize and thus control neuromorphic computing.