Specifically, Lanza’s group employed a two-dimensional insulating material known as multilayer hexagonal boron nitride (about 6 nm thick) and used microchips with silicon transistors of complementary metal-oxide-semiconductor (CMOS) technology—a technology type that exists in all electronic products used, like computers, phones, automobiles, household appliances, and medical machines.
The subsequent hybrid 2D/CMOS microchips show high durability and unique electronic properties, facilitating the fabrication of artificial neural networks with very low power consumption.
They can efficiently compute spiking neural networks—electrical stresses given during very short duration—an important element of current artificial intelligence systems that are growing in demand. Most existing devices are not ideal for implementing this kind of neural network, and there is a market demand to discover new methods.
The study has gained the attention of top semiconductor companies like Taiwan Semiconductor Manufacturing Corporation (TSMC) and Advanced Semiconductor Materials Lithography (ASML) and has aided other companies in decreasing processing energy and cost.
The majority of companies in the field of artificial intelligence and microchip manufacturing are planning to develop new hardware that decreases energy consumption and data processing time but has not found an appropriate device yet.
For radio-frequency applications, IBM has tried to combine graphene into transistors, but the devices could not process or store data. On the contrary, the devices developed by Professor Lanza’s group measure only 260 nm and could be made much tinier effortlessly if more advanced microchips were available.
For the field of semiconductors and nanoelectronics, this work is thrilling because of the high performance of the circuits and devices produced and the possibility for far-reaching industry applications.
Journal Reference:
Zhu, K., et al. (2023). Hybrid 2D/CMOS microchips for memristive applications. Nature. https://doi.org/10.1038/s41586-023-05973-1.