The surface of ice is a critical catalyst for environmental reactions, from here on Earth to the sky above. Present in polar stratospheric clouds — which appear as rainbow-colored, long wisps tens of thousands of feet in the sky — ice surfaces catalyze the formation of molecules that are harmful to the Earth’s ozone layer.
“The ice surface favors the chemical transformation, for example, of hydrochloric acid into molecular chlorine, which desorbs from the surface, and when it is irradiated by the sun splits into atomic chlorine,” said Davide Donadio, a UC Davis professor in the Department of Chemistry. “Atomic chlorine is extremely reactive and destroys ozone.”
While scientists have long known about correlations between the presence of polar stratospheric clouds and the depletion of the ozone layer, the chemical mechanism behind the correlation has remained elusive. After all, ice is a fickle material. Its molecular structure continuously changes with shifts in temperature. This variability poses problems for even the most advanced molecular imaging techniques.
In a new study appearing in JACS Au, Donadio’s lab and colleagues from the Lawrence Livermore National Laboratory (LLNL) harness the power of machine learning to improve the molecular interpretation of a spectroscopy technique that’s prominently used to investigate the air-ice interface, the critical boundary area for these chemical reactions.
Difficulties analyzing the ice surface
The study focused on improving a technique called vibrational sum-frequency generation (SFG) spectroscopy. This technique employs two different types of laser beams — an infrared frequency and a visible frequency — to analyze and probe vibrations at a material’s surface, revealing information about its molecular structure.
With multi-layered materials like ice, which has a different molecular and atomic structure depending on the layer analyzed and the temperature, the technique faces limitations.
“The ice surface is made of layers and it’s not clear whether only the first layer will contribute to the spectra or the second and deeper layers,” said Maggie Berrens, a UC Davis Ph.D. alum and former graduate student in Donadio’s group. “Even when the experiment is done perfectly, you have a problem of interpretation.”
Berrens, now a postdoctoral researcher with the LLNL, noted that “ice is commonly proton-disordered,” meaning that the oxygen atoms in its crystalline structure occupy precise locations, but the hydrogen atoms are usually oriented randomly, as long as they form two donors and two acceptor hydrogen bonds per molecule satisfying the “ice rule.”
“This leads to the question of whether the surface of ice is proton-disordered or not,” Donadio said.
A machine learning solution
Using machine learning potentials, the research team ran molecular dynamics simulations of ice surface models at various temperatures of interest. The use of machine learning enabled the research team to efficiently run long enough simulations needed to calculate the SFG spectra.
“We calculated SFG spectra for ice models with ordered and disordered surface structures, and comparing them to experiments, we concluded that the surface of ice is proton-ordered at low temperatures, around <180 degrees Kelvin or minus 90 degrees Celsius,” Donadio said. “Additionally, we managed to assign the specific molecular configurations contributing to the observed spectral features.”
The finding is particularly important for researchers seeking to better understand how ice surfaces affect chemical reactions, including in polar stratospheric clouds and potentially outer space ice objects like comets.
YOU MAY ALSO LIKE THESE STORIES
Ice as a Catalyst
In the lab of Professor of Chemistry Davide Donadio, physical chemistry graduate student Maggie Berrens investigates the fundamental properties of the ice surface to better understand how ice acts as a catalyst for environmental reactions.
From Viruses and Materials to Galaxies and Beyond
At the College of Letters and Science at UC Davis, researchers are using the power of machine learning to help protect us from the next pandemic, discover and build new materials, and explore the myriad galaxies in the heavens above.