Gold nanogap-based artificial synapses

Abstract

The basic building blocks for brain-inspired computing are neurons and their inter-cellular connections, called synapses. In this paper, we report artificial synapses composed of simple gold (Au) nanogaps that function using an electromigration-based method called "activation." In the activation technique, metal-atom transport is induced by a field emission current, resulting in a change in gap separation. Synaptic functionalities, including long-term potentiation, long-term depression, and spike-timing-dependent plasticity, were successfully implemented in the electromigrated Au nanogaps using activation. Furthermore, the integration of four artificial synapses in a 2 × 2 array and image memorization were achieved with the Au nanogap-based artificial synapses. These results illustrate that electromigrated Au nanogaps hold promise as synaptic devices for bio-inspired computational systems.

Wave-vector analysis of plasmon-assisted distributed nonlinear photoluminescence along Au nanowires

“…38,45 Such progressive evolution of the transport have been previously experimentally observed in wide Au nanogaps fabricated on SiO 2 and were linked to the metallic diffusion and filament growth in the gap. 45,46 We confirm the presence of metal clusters in the nanogap with scanning electron imaging of similar devices operated in air (i.e. without the PMMA layer) and taken after extensive testing with positive and negative biases.…”

“…This observation is in line with previous reports of Au/SiO 2 /Au systems. 45,46 In the present experiments, we did not apply any compliance to the electric current. Hence, complete and stable filament formation is essentially inhibited due to destructive Joule heating of the conduction path as the electric current increases.…”

“…Upon changing the polarity of the bias, diffusing species already present in the nanogap migrate in a reverse direction and leading to a dissolution of the conductive pathways previously built-up. 46 This is schematically pictured in Figure 4(c). The effective widening of the nanogap results in a presumably smaller capacitive effect on the electron concentration at the surface.…”

Bioartificial Synapses for Neuromorphic Computing

“…Various artificial synaptic devices have been manufactured in recent years, including ferroelectric devices, ,, electrochemical transistors, − and field-effect transistors (FETs). ,− These devices have successfully implemented biological synaptic functions such as short-term memory (STM), long-term memory (LTM), short-term plasticity (STP), long-term potentiation (LTP), and long-term inhibition (LTD). − However, these devices are transistors and cannot support integrated information processing and storage. As a new type of information storage device, a memristor has a structure similar to the synaptic structure and can simulate a variety of neurobehaviors of the human brain.…”

A scalable solution recipe for a Ag-based neuromorphic device

“…Such devices are expected to emulate cognitive behaviors but with modest energy consumption. Numerous materials such as the 2D materials [ 1 , 2 ], nanowire networks [ 3 – 5 ], ferroelectric materials [ 6 , 7 ], organic materials [ 8 , 9 ], metals [ 10 , 11 ], metallic oxides [ 12 – 15 ], and phase change materials [ 16 ] have been exploited as active elements to quantitatively emulate the behavioral aspects including short-term potentiation (STP), long-term potentiation (LTP), and classical conditioning. Toward this, varied device architectures based on different mechanisms such as the electrochemical metallization (ECM) [ 17 , 18 ], valence change (VCM) [ 19 , 20 ], ion migration [ 21 , 22 ], charge trapping [ 23 – 25 ], and phase change (PCM) [ 26 , 27 ] have been incorporated.…”