Nanostructured Materials and Architectures for Advanced Optoelectronic Synaptic Devices

Abstract

Neuromorphic photonics system based on the principle of biological brain is emerging as one of the potential solutions to the bottleneck inherent in classical von Neumann computing system. Optoelectronic synaptic devices, used to mimic the visual function of bio-synapse by adapting synaptic weights, can construct a highly efficient brain-inspired computing system, in which the nanostructured materials and device architectures are attracting extensive interests, giving many potential benefits in confined light-matter interaction, fast carrier dynamics, and photocarriers trapping. Moreover, the conjunction of traditional nanostructured photodetectors and newly realized electronic synapses also exhibit appealing performance for many practical applications including information processing and computing. This review focuses and summarizes on recent achievement in developing advanced optoelectronic synaptic devices based on nanostructured materials as 0D (quantum dots), 1D, and 2D materials, as well as, the rapidly evolving hybrid heterostructures. In addition, challenges and promising prospects in this research field are also discussed.

Intrinsically stretchable photonic synaptic transistors for retina-like visual image systems

“…Different from simple photodetectors, photonic synapses can effortlessly integrate perceiving, pre-processing, and remembering functions into a unified device rather than detecting light alone. [16][17][18] Such devices only occupy one physical space to respond to and process parallel visual information, effectively simplifying the preparation process and eliminating unnecessary consumption. Additionally, photonic synapses based on transistor configuration also offer prominent superiorities of signal amplification, 19 stable operation, 20 and visual information reception and reading simultaneously.…”

“…21–23 Currently, although the reported photonic synaptic transistors can mimic the neuromorphic function of the retina, most of them are made of inorganic 24–27 or non-stretchable organic materials. 1–30 This makes it impossible to resist dynamic deformation and conform well to a three-dimensional (3D) hemispherical surface like the soft and elastic retina. Consequently, this issue inevitably impedes the development of the bio-inspired neuromorphic visual system and its further application in the future for visual prosthetics and artificial intelligence.…”

Flexible Optical Synapses Based on In2Se3/MoS2 Heterojunctions for Artificial Vision Systems in the Near-Infrared Range

“…Flexible optical synaptic devices with NIR sensitivity, biomimetic plasticity, and flexibility have great application prospects in the fields of artificial visual systems, autopilot, the military system, free space communication, and robots, which have been attracting more and more attention in recent years. − Compared with traditional bulk materials, two-dimensional (2D) materials show great application prospects in the field of flexible NIR optical synapses because of their unique optical, electronic, and good mechanical flexibility. , The 2D materials include insulating hexagonal boron nitride (h-BN), graphene, transition-metal chalcogenides, , and main group metal chalcogenides. , However, due to the large band gap or no gap characteristics of most 2D materials, the realization of NIR light response is a dilemma at present. Moreover, synaptic plasticity and large-scale flexible wearable devices based on 2D materials under NIR illumination are difficult to realize, which hinders the development of flexible artificial visual systems .…”

“…13−20 Compared with traditional bulk materials, twodimensional (2D) materials show great application prospects in the field of flexible NIR optical synapses because of their unique optical, electronic, and good mechanical flexibility. 7,21 The 2D materials include insulating hexagonal boron nitride (h-BN), 22 graphene, 23 transition-metal chalcogenides, 24,25 and main group metal chalcogenides. 26,27 However, due to the large band gap or no gap characteristics of most 2D materials, the realization of NIR light response is a dilemma at present.…”

A Modified SiO2-Based Memristor with Reliable Switching and Multifunctional Synaptic Behaviors

“…Recently, resistance random access memory (RRAM), one of the most promising candidates in emerging memory technologies, has attracted much attention by the silicon microelectronics industry due to several advantages: fast switching response and high scalability in per-bit energy consumption, excellent endurance, and retention during switching operations. − Furthermore, the simple two-terminal design of “metal/solid-electrolyte/metal” is also applied to a 3D stackable high-density data storage architecture . These features make it possible to employ the RRAM for many potential applications, such as computing-in-memory, data storage, and memory logic units. , Different metal oxide materials, including TiO x , , WO x , , NiO, , CuO, Ta 2 O 5 , ZrO, , etc., have extensively been investigated as the solid electrolyte based on the mechanisms of resistive switching (RS) in memory devices.…”