Researchers indicated a way to solve the demand for high complexity computing processes using mobility enhanced FETs.
Machine learning is a trending technology by which computers can learn on its own without being explicitly programmed. This application of Artificial Intelligence (AI) gives machines the ability to automatically learn and improve from experience. For that, it generates new programs by integrating previous inputs and outputs. The need for machine learning is increasing day by day and with this rapid expansion, increases the demand of the complex computing processes.
To overcome these challenges, researchers from the University of Tokyo have taken a significant step to improve the efficiency of the data-intensive computing system. The researchers integrated a mobility-enhanced field-effect transistor (FET) and a ferroelectric capacitor (FE-CAP) to bring the memory system into the proximity of a microprocessor. The study was presented at the 2021 Symposium on VLSI Technology.
A memory element consists of a memory unity and access transistors. The access transistors are used to read and write the memory array and are generally silicon-metal-oxide semiconductor FETs. They are fabricated in the ‘front end of line (FEOL)’ layers of integrated circuits. The memory elements, on the other hand, are formed in the ‘back end of line (BEOL)’ layers. This layout does not efficiently utilize the space.Â
The researchers suggest that the oxide semiconductors such as indium gallium zinc oxide (IGZO) can be included in BEOL layers as they can be processed at low temperatures. While the access transistors and memory elements both can be fabricated into a single monolith in the BEOL. High-density, energy-efficient embedded memory can be achieved directly on a microprocessor by including access transistors and memory elements in the close proximity.
IGZO doped with tin (IGZTO) demonstrates high mobility and reliability. Therefore, the researchers used IGZTO for both the oxide semiconductor FET and ferroelectric capacitor (FE-cap) to create 3-D embedded memory. They observed that both the electron mobility and the current driving capability of the IGZTO FET were twice those of the IGZO FET.
“The proximity achieved with our design will significantly reduce the distance that signals must travel, which will speed up learning and inference processes in AI computing, making them more energy efficient,” says Masaharu Kobayashi, the author of this study. “We believe our findings provide another step towards hardware systems that can support future AI applications of higher complexity.”
