Roughly the size of a fingernail, semiconductor chips breathe life into modern technologies, including artificial intelligence, smartphones, cars and national security systems.
They’re also projected to wreak irrecoverable damage on the planet.
Energy consumption from semiconductor computing is doubling every three years compared to the low annual 2% growth in world energy production. And as global demand for data storage skyrockets, the world’s supply of silicon — a key material in the chips — may be left depleted by 2040, according to the 2020 Decadal Plan for Semiconductors.
Water usage has also increased fivefold in the industry since 2014, followed by escalating chemical dangers and greenhouse gas emissions, according to the Department of Energy EES2 roadmap published last year.
U.S.-based experts are exploring alternative methods to semiconductor expansion while minimizing environmental harm, and UF researchers play a vital role in finding those solutions.
Digital twins
CHIPS for America, a Biden-Harris Administration act, provided $285 million to kick off the SMART USA Institute in 2024, aiming to bolster domestic semiconductor chip production using AI. The institute will span seven bases across the country, with one encompassing the Florida/Caribbean region.
The goal is to reduce dependency on overseas chip producers after the COVID-19 pandemic cut off access to vital supply points, a standstill that threatened to leave U.S. technology — like national security systems — in peril.
Taiwan, a small, independent island in Asia, is a semiconductor stronghold manufacturing over 90% of the world’s most advanced chips, according to the U.S. International Trade Commission. It’s also caught in the crossfire of decades-old tension with a neighboring global superpower, China, which still claims the island as its territory. Other countries, including the U.S., now fear economic retribution from China if they recognize Taiwan as self governing.
“If China would make a move, then our entire semiconductor supply chain would be disrupted,” said Volker Sorger, the Florida/Caribbean Hub director and SMART USA Institute deputy chief digital officer.
The escalating international situation prompted the U.S.’ race to develop a self-reliant way to produce semiconductors domestically.
As a key player in state research, UF collected $20 million of SMART USA federal funding for the university’s supercomputer, HiPerGator, to create AI-generated copies of real tools and machines used to build semiconductors. These models, known as digital twins, allow scientists in the Florida/Caribbean Hub to test virtual scenarios and streamline manufacturing before expending resources to rebuild physically.
Sorger, who is also a UF Rhines endowed semiconductor photonics professor, said the SMART USA Institute will reduce manufacturing costs and time to nearly half, increase the number of usable chips in each batch by 40% and cut greenhouse gas emission by 30% within five years.
If environmental concerns are ignored, Sorger said, the industry’s unsustainable trajectory could lead to worse challenges producing chips and the demise of services reliant on modern technology.
“If we had no semiconductors anymore from one day to the next, essentially 90-some-% of all our businesses would be stopping,” he said.
The institute will finalize sustainability preparations by the middle of this year with a focus on cutting costs and saving resources through digital twins.
The Florida/Caribbean Hub might also collaborate with the existing Florida Semiconductor Institute, which houses UF and other state researchers in computing, chemistry, physics and more, working to apply their expertise to developing the industry domestically, according to Sorger.
‘Brute force’ computation
There’s one question Juan Manuel Restrepo-Flórez, a new addition to the Florida Semiconductor Institute, first asks himself before designing and deploying chemical procedures: “Is my process sustainable or not?”
The UF chemical engineering assistant professor proposed generating AI calculations to test the environmental impacts of chip production. Restrepo-Flórez and his team look at multiple factors beyond greenhouse gas emissions, which he said is often a sole focus in the industry.
Using HiPerGator for “brute force” computation and machine learning, Restrepo-Flórez and his team can test millions — or even billions — of hypothetical situations efficiently and sustainably. They examine the semiconductor manufacturing process through the lens of emissions, the capacity of ecosystems and limits to natural resources – a wider approach than most research.
“If you explore these alternatives by hand, it will take forever,” he said. “Forever, literally.”
The results may reveal a select few successes in each test, which he said could be later considered in chip manufacturing improvements.
‘Needle’ in a chemical haystack
Joshua Moon, a UF chemical engineering assistant professor and new member of the Florida Semiconductor Institute, is an experimentalist on the flip side of Restropo-Flórez’s digital simulation coin. The subject of Moon’s intrigue: forever chemicals leaching into semiconductor wastewater, yet another domino effect of the production process.
Per- and polyfluoroalkyl substances, or PFAS, have found their way into American homes in the form of non-stick cookware, water-resistant clothing and cleaning products, among other common items. These synthetic chemicals are also used in semiconductors and can cause a cocktail of long-term health impacts like cancer, liver and thyroid issues, weakened immune response and infertility, according to the Environmental Protection Agency.
The substances are slow to degrade, also harming the environment through soil, water and air contamination. Out of thousands of PFAS, six are monitored in drinking water by the EPA.
The problematic components of PFAS often linger in a much larger sea of harmless substances, which Moon said makes them challenging to pinpoint.
“It’s sort of like trying to pull a needle out of a haystack, essentially, in trying to pull these forever chemicals out of water,” he said.
Moon, who specializes in finding materials that can capture excess carbon, also studies how to solve the semiconductor chemical issue. PFAS seep into wastewater and stick to it, but he said certain water-swellable gels may have the capacity to absorb them.
Materials used to tackle the job traditionally contain a chemical called fluorine to grab the similarly flourinated PFAS, but that strategy could also cause more chemicals to detach into the water. Moon said he keeps his experiments fluorine free, opting to use methods comparable to a Brita filter’s chemistry to accomplish the same goal.
The chip production process is broken into several segments, and some can produce streams of wastewater that hide contrasting sets of PFAS. The range of variants would call for special absorbent materials to clean every stream, with some completely incompatible toward others, he added.
Moon said he hopes to collect and test wastewater samples directly from semiconductor manufacturing.
“I would say that we’re not quite there in terms of finding a perfect solution yet,” he said, a statement indicative of the industry’s uncharted road ahead.
Contact Rylan DiGiacomo-Rapp at rdigiacomo-rapp@alligator.org. Follow her on X @rylan_digirapp.
Rylan DiGiacomo-Rapp is a third year journalism and environmental science major and the Fall 2024 Enterprise Environmental Reporter. Outside of the newsroom, you can usually find her haunting local music venues.
