Flexible Active Crossbar Arrays Using Amorphous Oxide Semiconductor Technology toward Artificial Neural Networks Hardware

Abstract

Memristor crossbar arrays can compose the efficient hardware for artificial intelligent applications. However, the requirements for a linear and symmetric synaptic weight update and low cycle‐to‐cycle (C2C) and device‐to‐device variability as well as the sneak‐path current issue have been delaying its further development. This study reports on a thin‐film amorphous oxide‐based 4×4 1‐transistor 1‐memristor (1T1M) crossbar. The a‐IGZO crossbar is built on a flexible polyimide substrate, enabling IoT and wearable applications. In the novel framework, the thin‐film transistor and memristor are fabricated at the same level, with the same processing steps and sharing the same materials for all layers. The 1T1M cells show linear and symmetrical plasticity characteristic with low C2C variability. The memristor performs like an analog dot product engine and vector–matrix multiplications in the 4×4 crossbars is demonstrated experimentally, in which the sneak‐path current issue is successfully suppressed, resulting in a proof‐of‐concept for a cost‐effective, flexible artificial neural networks hardware.

An Optoelectronic Synapse Based on Two‐Dimensional Violet Phosphorus Heterostructure

“…In those cases, most synaptic devices show relatively high error (over 2%) due to low DR and fewer state numbers. [25][26][27][28][29][30][31][32][33][34] By combining large DR and sufficient state numbers, our device shows negligible error from ideal cases, even in relatively complex tasks like Fashion-MNIST. This suggests that for future intelligent scenarios, increasing the DR and state numbers of synaptic devices is necessary and urgent.…”

Interface roughness effects and relaxation dynamics of an amorphous semiconductor oxide-based analog resistance switching memory

“…Few studies in the literature have reported on the analog switching (gradual resistive switching) in a-IGZO-based memristors, which is favourable for AI-related applications (refer to Table 1). 17,20,21,26,[49][50][51][52][53][54][55][56] It is found that an analog/gradual resistive switching is observed at the cost of slightly higher values of bias voltages in comparison with the filament-based switching devices. From an application point of view, endurance and retention are two main key parameters in memristors.…”

Recent progress in optoelectronic memristors for neuromorphic and in-memory computation

“…Crosstalk can be eliminated with the addition of a selector transistor in each crossbar cell. However, this approach undoubtedly increases cell area and energy consumption [24]. In fact, the biological system cannot be matched with purely voltage-assisted neuromorphic platforms.…”