Supercomputer Shows Atmospheric Changes Resulting from Interaction of Snow and Phenolic Compounds

Published April 29, 2021

[Enlarge] Comet supercomputer was used to develop a multi-scale, multi-model approach for the prediction of how molecules absorb light on ice in comparison to liquid solutions, as seen here on the snapshots of the phenolic compound guaiacol on ice and also on guaiacol’s UV-visible absorption spectra in solution (left curve) and on ice (right curve). The shift in the absorption spectrum is significant since it allows for much greater degradation of guaiacol by sunlight.  Credit: Fernanda Bononi et al, University of California at Davis

By Kimberly Mann Bruch, SDSC Communications

According to the Environmental Protection Agency, phenol is typically harmless. Researchers at the University of California at Davis (UC Davis), however, have recently shown that organic compounds like phenol may not be as harmless as meets the eye, especially when it comes to snowmelt.

Using National Science Foundation (NSF) Extreme Science and Engineering Discovery Environment (XSEDE) allocations on Comet at the San Diego Supercomputer Center located at UC San Diego, the UC Davis scientists created simulations that showed organic compounds such as phenols appear to degrade faster in and on ice in a way that negatively influences the impacted snow’s chemical composition. They published their recent studies in the Journal of Physical Chemistry A and Environmental Science: Processes and Impacts.

“We used Comet to run several hundred 12- to 20-hour simulations to understand how phenol (and other organic molecules) interact with the water molecules on ice/snow and how that interaction changes their photo degradation rate when compared to solution,” said Fernanda Bononi, a graduate student at UC Davis. “Our research illustrated that organic pollutants like phenols, in or on ice crystals, may be transformed faster on ice than liquid solution, which influences the fate of chemicals in snow, and consequently, the chemical composition of the surrounding atmosphere.”

Thanks to XSEDE-allocated supercomputer modeling, the research team calculated the absorption spectra of the organic molecules to check how much of this increase in the photo degradation rate was due to a shift in the UV-visible absorption of these molecules.

“Our model combined classical and quantum simulations as well as statistical learning and took into account the relevant thermodynamic conditions and the effects of the interactions between solvent and solution,” said Bononi. “Further, with this combined model, we showed that the absorption of phenolic molecules at the air-ice interface is thermodynamically more favorable than the solvation in bulk liquid regions, although molecules tend to remain in the surface of the quasi-liquid layer on ice.”

XSEDE allocations on Comet enabled the researchers to perform long molecular dynamics simulations, starting from scratch, followed by hundreds of absorption spectra calculations in order to obtain a good spectra line shape for the half-dozen molecules, including phenol.

“The calculations to show these results required the development of a complex multi-scale, multi-model approach to thoroughly account for physico-chemical features of a complex solvation environment such as the surface of a snowflake,” said Bononi, adding that it would take longer than one month without a supercomputer to obtain the necessary calculations for a single molecule. “With Comet, we were able to obtain results for four additional molecules without the extra cost associated with acquiring new computing resources,” she said.

This work was supported by NSF (1806210). Supercomputing time on Comet was allocated via NSF’s XSEDE (TG-CHE190009).