Biorealistic response in a technology-compatible graphene synaptic transistor

TL;DR

This paper introduces a graphene-based synaptic transistor that mimics biological synapses with highly tunable, biorealistic behavior, suitable for neuromorphic computing systems, demonstrating multiple conductance states and dynamic synaptic functions.

Contribution

It presents a novel, all-technology-compatible graphene synaptic transistor with biorealistic dynamics and tunable conductance, advancing hardware neuromorphic computing.

Findings

Device exhibits a continuous range of conductance levels

Emulates biological synaptic functions like facilitation and plasticity

Achieves high tunability and low power consumption

Abstract

Artificial synapse is a key element of future brain-inspired neuromorphic computing systems implemented in hardware. This work presents a graphene synaptic transistor based on all-technology-compatible materials that exhibits highly tunable biorealistic behavior. It is shown that the device geometry and interface properties can be designed to maximize the memory window and minimize power consumption. The device exhibits a virtually continuous range of multiple conductance levels, similar to synaptic weighting, which is achieved by gradual injection/emission of electrons into the floating gate and interface traps under the influence of an external electric field. Similar to the biological synapse, the transistor has short-term intrinsic dynamics that affect the long-term state. The temporal injection/emission dynamics of an electronic synapse closely resemble those of its biological counterpart and is exploited to emulate biorealistic behavior using a number of synaptic functions, including paired-pulse facilitation/depression, spike-rate-dependent plasticity, and others. Such a synaptic transistor can serve as a building block in hardware artificial networks for advanced information processing and storage.