In the field of artificial intelligence (AI), researchers are continuously exploring ways to improve the efficiency and accuracy of computing systems. One promising development is the use of indium-gallium-zinc-oxide (IGZO) memory technology, which has the potential to supercharge AI applications. IGZO-based 2-transistor 1-capacitor (2T1C) technology is being explored for Analog In-Memory Computing (AIMC), which addresses the limitations of traditional digital computing [860625ba].
Traditional digital computing faces challenges such as power consumption and data transfer bottlenecks. AIMC using IGZO DRAM cells offers reduced standby power consumption and can be processed in the chip's back-end-of-line (BEOL), resulting in a denser memory array. This technology has the potential to significantly improve energy efficiency and computational density for machine learning applications [860625ba].
At the 2023 International Memory Workshop (IMW), researchers from imec discussed the challenges and optimization of retention time in IGZO-based 2T1C and 2T0C gain cells for AIMC. These cells show exceptional properties for AIMC, offering superior energy efficiency and computational density, with the 2T0C cells excelling in area efficiency [860625ba].
This development in IGZO memory technology is a significant step forward in the quest to supercharge AI applications. By addressing the limitations of traditional digital computing, AIMC using IGZO DRAM cells has the potential to revolutionize AI computing systems, improving energy efficiency and computational density for machine learning applications.
In Shanghai, researchers have made a breakthrough in chip technology by developing a semiconductor chip insulator using an artificial sapphire. This innovation could potentially lengthen the battery life of smartphones and improve the efficiency of two-dimensional circuits. Traditional insulating materials become less effective when reduced to nanometer thickness, resulting in higher power consumption and heat production. However, the atomic-thick sapphire film developed by the research team can be used to build more efficient two-dimensional circuits and is scalable. The team grew a single-crystal aluminum wafer and inserted oxygen atoms to form a single-crystalline aluminum oxide just 1.25 nanometers thick. The sapphire film was combined with a two-dimensional substrate of molybdenum disulphide to create low-power transistor arrays, meeting the standards for future low-power chips. This method is easily scalable to industrial production levels and compatible with existing silicon-based processes. The development of the artificial sapphire insulator has implications for extending smartphone battery life and supporting the development of low-power chips for artificial intelligence and the Internet of Things [673751f3].