To beat China, new U.S. law offers billions for microchip research and training

[beijingwalker]

To beat China, new U.S. law offers billions for microchip research and training​

Universities team up with industry to seek funding under CHIPS and Science Act​

• 6 SEP 2022
• 3:00 PM
• BYJEFFREY MERVIS

Most of the $280 billion in a new law to strengthen the U.S. semiconductor industry and keep the country ahead of China in technology is a 5-year promise, not a reality. But along with the aspirational spending, the recently passed CHIPS and Science Act commits some $13 billion right now for research and training in microelectronics. And U.S. universities are now forming large coalitions with companies and local governments in order to be ready to compete for the money as soon as a trio of federal agencies announces its plans.

“No sane university with a strong interest in microelectronics is sitting this out,” says Jesús del Alamo, professor of electrical engineering at the Massachusetts Institute of Technology (MIT), which is involved with several such partnerships.

The act, signed into law on 9 August, funnels $11 billion over 5 years to the U.S. Department of Commerce to create a National Semiconductor Technology Center (NSTC) and a national advanced packaging manufacturing program. Another $2 billion will go to the Department of Defense (DOD) for a microelectronics commons, a national network of university laboratories to develop prototypes for the next generation of semiconductor technologies.

Both initiatives “are aimed at the dearth of lab-to-fab facilities,” referring to university laboratories that develop new technologies to be incorporated into semiconductor fabrication plants, or fabs, says Philip Wong, an electrical engineer at Stanford University who heads its nanofabrication center. In addition, the National Science Foundation (NSF) is getting $200 million over 5 years for education and workforce training in microelectronics.

“The [research and training] pipeline is just as important as the fabs, although it doesn’t get nearly as much attention,” says Lisa Su, CEO of chipmaker AMD and a member of the President’s Council of Advisors on Science and Technology (PCAST). She calls the new law “a once-in-a-generation … opportunity to fill the pipeline with the next generation of semiconductor technologies.”

A PCAST report due out this month will recommend NSTC spend 30% to 50% of its budget on fundamental research across a range of fields in microelectronics, from new materials and energy efficient computing to improved security and health care applications. And “a skilled workforce is a prerequisite for everything,” says PCAST member Bill Dally, senior vice president for research at NVIDIA. “We need to do a lot of things to retain our leadership in semiconductors, and all of them require talented people.”

Two large industry-academic-government partnerships are regarded as the leading contenders for NSTC: the American Semiconductor Innovation Coalition (ASIC), spearheaded by IBM and New York’s Albany Nanotech Complex; and the Semiconductor Alliance, which features Intel, Micron, and the MITRE Corporation. Both groups boast a roster of academic heavyweights. MIT, the Georgia Institute of Technology, and Purdue University are members of ASIC, whereas Stanford and the University of California (UC), Berkeley, are closely aligned with the Alliance. Although New York, Virginia, and Texas politicians have proposed their states as hosts, insiders say the center is more likely to be a network of existing facilities spread across the country than a single edifice.

The center’s to-do list includes funding multimillion-dollar upgrades to existing fab labs at dozens of universities and providing researchers access to a kind of standard workbench to lower the cost of testing and prototyping new chip technologies. NSTC would also support startup companies that want to commercialize those technologies. In addition, the center would address the need for additional talent at all levels by funding hundreds of new faculty positions, thousands of scholarships, a uniform curriculum in microelectronics with hands-on training, and outreach to middle and high school students.

Only a small number of universities have the capacity to host that training. A white paper issued last year by del Alamo and his colleagues estimates that upgrading a university fab lab to handle 200-millimeter wafers, the size that has become standard in advanced fabs, would cost $80 million. The university would then need $80 million a year in research grants to operate the lab, which could train some 500 graduate students and postdocs. That rules out many institutions, del Alamo says. Even upgrading the select group of universities that remains would still leave the country far short of filling the 42,000 new semiconductor jobs that the CHIPS Act is expected to create, says Tsu-Jae King Liu, UC Berkeley’s dean of engineering.

So last year, King Liu spearheaded the formation of the American Semiconductor Academy (ASA), a national network for
microelectronics education. She has teamed up with the SEMI Foundation, the nonprofit arm of the industry’s trade association, to seek CHIPS funding from DOD and NSF to carry out that vision.

Attracting U.S. students into microelectronics has become a challenge over the past 30 years as semiconductor jobs moved overseas and Google and other U.S.-based companies threw money at recent graduates to write software rather than make devices, says Sanjay Banerjee, a professor of electrical and computer engineering and head of the microelectronic research center at the University of Texas (UT), Austin. UT is part of a consortium that is seeking a slice of the DOD funding to reverse that trend.

Banerjee says the DOD proposal will build on a long-running NSF-funded program, called the National Nanotechnology Coordinated Infrastructure (NNCI), which provides $84 million over 5 years to 16 universities. A DOD grant would be like “putting NNCI on steroids,” he says. “It will give us better tools, more staff, and facilities more relevant to industry.”

The CHIPS legislation aims to spread research and education opportunities both geographically and by race/ethnicity. That will require participation by more institutions not on the nation’s East and West coasts, as well as those serving large numbers of students from groups historically underrepresented in science and engineering. They include the country’s historically Black colleges and universities (HBCUs), which as a group rank low on a list of institutions getting federal research dollars.

Consortia vying for the funds must make room for those institutions, says electrical engineer Michel Kornegay of Morgan State University, an HBCU in Baltimore. “Folks talk in general terms about equity and inclusion,” says Kornegay, a member of the ASA network. “But they don’t spell out how they plan to do that, and whether the people who are going to implement those plans have a track record of success.”

The influx of CHIPS money could even widen the gap between the haves and the have-nots if those problems aren’t addressed, warns Patricia Mead, dean of engineering at Norfolk State University, an HBCU. “In America, the rich will always get richer,” she says. “So you need strong leadership that is going to make a real commitment to broaden the footprint of microelectronics training in this country.”

Science | AAAS

www.science.org