Researchers create a high-performance “brain” for AI using OLED TV materials

Against conventional architecture, the novel device enables processing to happen within the storage of a system.

Jijo Malayil
Researchers create a high-performance “brain” for AI using OLED TV materials
Close up of motherboard

sankai/iStock 

The advent of ChatGPT has made Artificial Intelligence (AI) coupled with a language model accessible and practical for daily tasks, including paper writing, translation, coding, and more, all through question-and-answer-based interactions. 

The disruptive launch of OpenAI’s product has created ripples in the industry, with technology giants like Google and Microsoft scrambling to maintain their position in the market by launching their half-baked AI-powered models. 

The AI systems rely on its deep learning capabilities, requiring extensive training to eliminate errors, resulting in frequent data transfers between memory and processors. Restricting further advancements, the conventional computer architecture separates the computation of information from its storage, leading to higher power usage and lag in AI computations.

To address this challenge, a team of researchers at Pohang University of Science and Technology (POSTECH) in South Korea has utilized semiconductor technologies to develop a high-performance device made of indium gallium zinc oxide (IGZO), an oxide semiconductor widely used in OLED displays. The device is claimed to offer high performance and power efficiency. 

The study regarding the research was published in the journal Advanced Electronic Materials.

The material offers computations in the memory responsible for data storage

The requirement of highly demanding AI tasks for processing to happen within the storage is fulfilled by utilizing IGZO. According to the team, earlier semiconductor systems were limited in meeting all the requirements, such as linear and symmetric programming and uniformity, to improve AI accuracy.

 An extensive study by the team proved that when IGZO was used as a key material for AI computations, it provided uniformity, durability, and computing accuracy. IGZO consists of “four atoms in a fixed ratio of indium, gallium, zinc, and oxygen and has excellent electron mobility and leakage current properties, which have made it a backplane of the OLED display,” said a media release. 

A representational image of the system

 The material has enabled researchers to develop a novel synapse device composed of two transistors interconnected through a storage node. According to the team, controlling the node’s charging and discharging rate has enabled the AI semiconductor to meet the diverse performance metrics required for high-level performance. 

In addition, including such devices in a large-scale AI system requires the output current of synaptic devices to be minimized. The team confirmed the “possibility of utilizing the ultra-thin film insulators inside the transistors to control the current, making them suitable for large-scale AI.”

Tests with synaptic device provided an accuracy of 98%

An AI-language model was run using the device to train and classify handwritten data, achieving an accuracy rate of over 98%, proving its capability of powering high-accuracy AI systems in the future.

“The significance of my research team’s achievement is that we overcame the limitations of conventional AI semiconductor technologies that focused solely on material development,” said Professor Yoonyoung Chung, the study’s lead author. The team could obtain linear and symmetrical programming characteristics by employing its novel structure for a synaptic device. “Thus, our successful development and application of this new AI semiconductor technology show great potential to improve the efficiency and accuracy of AI,” said Yoonyoung.

Abstract

This work presents an analog neuromorphic synapse device consisting of two oxide semiconductor transistors for high-precision neural networks. One of the two transistors controls the synaptic weight by charging or discharging the storage node, which leads to a conductance change in the other transistor. The programmed weight maintains for more than 300 s as electrons in the storage node are well preserved due to the extremely low off current of the oxide transistor. Ideal synaptic behaviors are achieved by utilizing superior properties of oxide transistors such as a high on/off ratio, low off current, and large-area uniformity. To further improve the synaptic performance, self-assembled monolayer treatment is applied for reducing the transistor conductance. The reduction of on current reduces the power consumption, and the reduced off current improves the retention characteristics. There is no noticeable decrease in simulated neural network accuracy even when the measured device-to-device variation is intentionally increased by 200%, indicating the possibility of large-array operation with the synapse device.