US state of Georgia, 18 December (ANI): Water has confounded scientists for many years. They have been speculating for the past 30 years or more that when water is chilled to extremely low temperatures like -100C, it can split into two liquid phases of different densities. Such oil and water, these phases don’t mix, which may help to explain some of the other odd properties of water, like how it loses density as it cools.
However, because water freezes so quickly at such low temperatures, it is almost impossible to study this phenomenon in a laboratory setting. According to recent study from the Georgia Institute of Technology that uses machine learning models, the phase transitions of water are now better known, allowing increased chances for a more theoretical understanding of different components.
According to Thomas Gartner, an assistant professor in the Georgia Institute of Technology School of Chemical and Biomolecular Engineering, “We are doing this with very detailed quantum chemistry calculations that are trying to be as close as possible to the real physics and physical chemistry of water.” “Nobody has ever been able to analyse this transition with this level of accuracy,” the researcher said.
With co-authors from Princeton University, the study was published in the paper “Liquid-Liquid Transition in Water From First Principles” in the physical review letters journal.
The researchers used molecular simulations, which Gartner compared to a virtual microscope, to better understand how water interacts.
You could zoom in all the way down to the level of the individual molecules and observe how they move and interact in real time, he continued, if you had an infinitely powerful microscope. “We’re generating practically a computational movie,” the author said.
At various water temperatures and pressures, the researchers mimicked the phase separation between the high- and low-density liquids by analysing the movement of the molecules and characterising the liquid structure. They gathered a lot of data, running some simulations for as long as a year, and kept improving their algorithms to get more precise outcomes.
Running such extensive and detailed simulations would not have been feasible even ten years ago, but machine learning today provided a workaround. The energy of how water molecules interact with one another was determined by the researchers using a machine learning technique. This model dramatically accelerated the calculation compared to conventional methods, which greatly improved the efficiency of the simulations.
Since machine learning isn’t flawless, these extensive simulations also enhanced forecast accuracy. The researchers took care to carefully evaluate their hypotheses using several simulation tools. If several simulations produced findings that were comparable, their accuracy was confirmed.
Because the topic is so difficult to analyse experimentally, Gartner noted that one of the difficulties with this work is that there isn’t much data to compare to. We’re really pushing the envelope here, so that’s another reason why it’s crucial to try to accomplish this utilising a variety of computational approaches.
Extremes that are probably not present on Earth directly but might possibly exist in other water habitats in the solar system, from the oceans of Europa to the water in the heart of comets, were some of the circumstances that the researchers evaluated. These discoveries may also aid in the development of more accurate climate models, a better understanding of the peculiar and complex physical chemistry of water, as well as its usage in industrial activities.
Gartner claims that the work is even more generalizable. Water is a well-researched topic, but this concept might be applied to other complex phenomena like chemical processes or hard-to-simulate materials like polymers.
The ability of water to make this phase transition has long been a subject of debate since it is so essential to both life and industry. If we can find a solution, that would be significant, he said. But now that we have this incredibly potent new computational tool, there is a lot of room to advance the discipline because we are still unsure of its bounds.
