In:
Advanced Electronic Materials, Wiley, Vol. 8, No. 12 ( 2022-12)
Abstract:
Artificial synapses have attracted extensive attention due to their capacity to circumvent the inherent restrictions of Von Neumann computing architecture. Although electrolyte‐gated synaptic transistors (EGSTs) have been enabled to emulate the essential synaptic plasticity, the corresponding neurological applications have been restricted by the lack of inhibitory synaptic response and the subsequent low asymmetric non‐linearity factor. Here, Mg‐doped SnO 2 (Mg:SnO 2 ) EGSTs via a low‐cost and facile solution method are fabricated. With both excitatory and inhibitory response modes, the improved MgSnO EGSTs display variable synaptic plasticity, with featured characteristics such as excitatory post‐synaptic current (EPSC), inhibitory post‐synaptic current (IPSC), paired pulse facilitation (PPF), paired pulse depression (PPD), spiking rate dependent plasticity (SRDP), and spike timing dependent plasticity (STDP). For the University of California at Irvine image, a simulated artificial neural network built on MgSnO EGSTs may achieve a learning accuracy of 92.3%. In addition, digital logic processing, Pavlovian dog's experiment have all been realized by the newly designed MgSnO EGSTs. These results highlight the potential of the MgSnO EGSTs for next‐generation artificial intelligence systems.
Type of Medium:
Online Resource
ISSN:
2199-160X
,
2199-160X
DOI:
10.1002/aelm.202200864
Language:
English
Publisher:
Wiley
Publication Date:
2022
detail.hit.zdb_id:
2810904-1