Officials see an opportunity to further invest in AI research and development to improve semiconductor manufacturing.
The White House believes artificial intelligence could have a dramatic impact on semiconductor manufacturing, an increasingly complex and resource-hungry technology.
“We’ll see cross-optimized semiconductor materials emerge, both for existing and known materials, that address key issues like performance but also supply chain resiliency and critical sustainability. And we may also see new materials that we haven’t thought of before,” Arathi Prabhakar, director of the White House Office of Science and Technology Policy, said at a White House event on June 13.
President Biden’s “CHIPS and Science Act” prioritizes funding for semiconductor technology, which makes up key components in automobiles, consumer electronics, and even defense weapons systems. The plan aims to increase U.S. leadership in semiconductor technology production and reduce reliance on Asia.
Federal leaders discussed the potential of AI in semiconductor manufacturing at a White House event, and the effort is part of a broader plan. The longing for AI Explore how AI can be used for research and development in other industries, including health, education, and weather.
For semiconductors, a big challenge is time: It can take nearly a decade to design new materials needed for new chips, and that’s where AI could make a difference, people involved say.
“You’re never going to get there by doing experiments the way we’ve always been taught. You need something like artificial intelligence. But that’s not enough,” said Benji Maruyama, a senior materials research engineer in the Air Force Research Laboratory’s Materials and Manufacturing Division. “Not only do we need to invest in more experiments, but we need to get a better understanding and simulations to get better knowledge.”
Incorporating AI into production processes is highly technical, and manufacturers need vast amounts of data to feed or train AI models. This is because critical dimension scanning electron microscopes (CD-SEMs), the systems dedicated to measuring dimensions on semiconductor wafers, are becoming more complex and require large amounts of data to map the dimensions.
“The CD-SEM has to be digitized, and enough of the exact dimensions of the CD-SEM have to be fed back into the AI model to get accuracy,” says Melissa Grupen-Shemansky, CTO and vice president of technical communities at industry group SEMI. “The data points you actually need are one or two orders of magnitude more than that.”
Francesca Tavazza, group leader for Data and AI-Driven Materials Science at NIST, said materials and AI scientists need to think about the right features and ingredients when creating models. She said these models are not black boxes where more data is added to create better models, but rather follow scientific laws throughout the process.
“We need to understand the sources of the key elements that drive the physical phenomena, and then we input that into the model,” Tabazza said.
Dana Weinstein, White House Chief of Staff for Industrial Innovation at OSTP and Special Advisor for Research and Development at CHIPS Vision outlined Our goal is to develop a materials accelerator that can use AI to reduce the time it takes to introduce advanced materials from up to 20 years to just 1-3 years. Strong partnerships are essential to making this happen.
Incentives for labs, industry partners and startups to join the conversation could eliminate bias and other challenges from the start. Tom Kenny, a professor of engineering at Stanford University, said he learned the importance of incentives that foster collaboration between government, industry and academia during his time at DARPA.
“Oftentimes, we take a lot of capital funding from the government and matching funds and simply place that money in the middle between academia and industry,” Kenney said. “At DARPA, we primarily started our programs by identifying open spaces between communities that we were eager to attract together. The CHIPS Act is a great opportunity to create teams motivated to collaborate around ample resources.”