Researchers exhibited monolithic 3D integration of layered 2D material into innovative artificial intelligence computing hardware. The novel method offers a material-level solution for fully integrating multiple functionalities into a single, compact electronic chip, paving the path for powerful AI computing.
With integrated sensors, processors, memory, and other specialized components, multifunctional computer chips have evolved to accomplish more. However, as chip sizes have increased, so has the time required to transfer data between functional components.
“Think of it like building a house,” said Sang-Hoon Bae, an assistant professor of mechanical engineering and materials science at the McKelvey School of Engineering at Washington University in St. Louis. “You build out laterally and up vertically to get more function, more room to do more specialized activities, but then you have to spend more time moving or communicating between rooms.”
Monolithic 3D integration has the potential to reshape the entire electronics and computing industry by enabling the development of more compact, powerful and energy-efficient devices.
Sang-Hoon Bae
To address this challenge, Bae and a group of international collaborators, including researchers from the Massachusetts Institute of Technology, Yonsei University, Inha University, Georgia Institute of Technology, and the University of Notre Dame, demonstrated monolithic 3D integration of layered 2D material into novel processing hardware for artificial intelligence (AI) computing. They hope that their innovative approach will not only give a material-level answer for fully integrating multiple functionalities into a single, compact electronic chip, but will also open the path for breakthrough AI computing. Their study was featured on the front cover of Nature Materials on November 27.
The team’s monolithic 3D-integrated device outperforms current laterally integrated computer chips. The device is made up of six atomically thin 2D layers, each with its own function, and has a greatly reduced processing time, power consumption, latency, and footprint. This is accomplished by stacking the processing layers firmly to guarantee dense interlayer connection. As a result, the hardware provides previously unrivaled efficiency and performance in AI computing activities.
This finding provides a fresh way to integrate electronics and ushers in a new era of versatile computing hardware. With ultimate parallelism at its foundation, this technology has the potential to significantly enhance the capabilities of AI systems, allowing them to execute complicated jobs with lightning speed and outstanding precision, according to Bae.
“Monolithic 3D integration has the potential to reshape the entire electronics and computing industry by enabling the development of more compact, powerful, and energy-efficient devices,” Bae stated in a press release. “Atomically thin 2D materials are ideal for this, and my collaborators and I will continue improving this material until we can ultimately integrate all functional layers on a single chip.”
According to Bae, these devices are also more adaptable and functional, making them suited for a wider range of applications.
“From autonomous vehicles to medical diagnostics and data centers, the applications of this monolithic 3D integration technology are potentially boundless,” he told the audience. “For example, in-sensor computing combines sensor and computer functions in a single device, rather than a sensor gathering data and transferring it to a computer.” This allows us to acquire a signal and directly compute data, resulting in faster processing, lower energy consumption, and improved security because data is not transported.”
