Nonbiodegradable “forever chemicals” like perfluoroalkyl and polyfluoroalkyl substances (collectively, PFASs) were invented in the 1930s as a way to advance nonstick and waterproof materials. Unfortunately, as the chemicals began to accumulate in living bodies over time, researchers started to uncover links between PFASs and a range of diseases, including kidney and testicular cancer. Recently, researchers from the University of California Riverside (UCR) applied supercomputing power at the San Diego Supercomputer Center to investigate methods of removing these dangerous chemicals from drinking waters – one of their primary methods of ingress to the human body.
Specifically, these researchers focused on the ability of light to degrade the carbon-fluorine bond in PFASs. They ran a series of simulations, bombarding virtual PFASs dissolved into virtual water molecules with virtual light.
“Recent interest has focused on processes driven by lasers/light for directly decomposing PFAS contaminants, and the large-scale simulations used in our research shed crucial insight into these photo-induced degradation processes,” explained Bryan Wong, a professor of chemical and environmental engineering at UCR, in an interview with SDSC’s Kimberly Mann Bruch. “The photo-induced mechanisms of PFAS degradation are not well understood at all, and the supercomputer-enabled simulations used in our research help us understand this important process at a quantum-mechanical level of detail so that we can work on creating approaches for directly treating PFASs.”
To run the simulations, they used SDSC’s Comet supercomputer, a 2.76-peak petaflops system powered by 1,944 Intel Haswell CPU nodes and 72 Nvidia GPU nodes, which the researchers accessed through the Extreme Science and Engineering Discovery Environment (XSEDE). “We used more than 570,000 CPU hours, and Mahidhar Tatineni from SDSC helped us efficiently compile our code on Comet,” Wong said. “As someone that has been using XSEDE resources for almost 20 years, I am grateful to the National Science Foundation for these allocations as they allow us to push the boundaries of what we can simulate – carrying out calculations to a level that would otherwise be impossible to do.”
The iterative simulations allowed the researchers to tune the frequency and intensity of the light to specifically break the bonds in PFASs. Next, the researchers are planning to continue their work. “Thanks to allocations from [XSEDE], we will continue expanding this work even further with SDSC’s newest supercomputer Expanse,” Wong said. Expanse, which delivers 5.16 peak petaflops, employs 728 AMD-powered compute nodes and 52 Nvidia-powered GPU nodes.