Developing energy-efficient microelectronic devices

We are adopting a co-design approach for developing energy-efficient AI-based computing while reducing use of critical materials. Targeted applications include vehicular technologies as well as sensor-based and ​“beyond exascale” computing.  

This focus area involves efforts from materials design and synthesis to the development of novel devices, architecture design, and algorithm design, so that algorithms are suited to taking advantage of unique device features. Argonne has proven strengths in the application of AI that we can bring to bear on the innovations required for next generation microelectronics technologies. For example, the development of materials that facilitate energy-efficient memories, but potentially impose constraints on memory configuration, would require the development of algorithms that could leverage these constraints. Conversely, it is important that we identify key application areas, such as AI-enhanced science applications, that have unique requirements that will drive novel device design. We will also use machine learning approaches for modeling of large systems, devices, and novel phenomena, including parameters such as the effect of temperature.

Energy-efficient manufacturing of microelectronics

The current approaches to microelectronics manufacturing are energy- and resource-intensive. This focus area aims to create new approaches to microelectronics manufacturing, including digital twins, distributed manufacturing for soft materials, and consideration of supply-chain and lifecycle impacts on microelectronic component development. These new approaches can help industry make better informed choices about the processes they follow. Additionally, we will explore ways to extend feature sizes down the to the atomic scale in 3D, to integrate dissimilar materials, and to move to more distributed manufacturing approaches where this is appropriate.