Better understand how water behaves with machine learning – Zoo House News
Water has occupied scientists for decades. For the past 30 years, they have theorized that when water is cooled to a very low temperature, such as -100°C, it might be able to separate into two liquid phases of different densities. Like oil and water, these phases don’t mix and may help explain some of water’s other strange behaviors, such as how it becomes less dense as it cools.
However, it is almost impossible to study this phenomenon in the laboratory because water crystallizes so quickly at such low temperatures. Now, new research from the Georgia Institute of Technology uses machine learning models to better understand the phase changes of water, opening up more opportunities for a better theoretical understanding of different substances. Using this technique, the researchers found strong computational evidence supporting the liquid-liquid transition of water that can be applied to real-world systems that use water to operate.
“We do this with very detailed quantum chemical calculations that try to get as close as possible to the real physics and physical chemistry of water,” said Thomas Gartner, an assistant professor in Georgia Tech’s School of Chemical and Biomolecular Engineering. “This is the first time anyone has been able to study this transition with this level of detail.”
The research was featured in the article “Liquid-Liquid Transition in Water From First Principles” in the journal Physical Review Letters with co-authors from Princeton University.
To better understand how water interacts, the researchers ran molecular simulations on supercomputers, which Gartner compared to a virtual microscope.
‘If you had an infinitely powerful microscope, you could zoom in down to the level of individual molecules and watch them move and interact in real time,’ he said. “That’s what we’re doing, almost creating a computer film.”
Researchers analyzed how the molecules move and characterized the liquid structure at different water temperatures and pressures, mimicking the phase separation between the high and low density liquids. They collected extensive data – ran some simulations for up to a year – and continued to refine their algorithms to produce more accurate results.
Running simulations this long and detailed would not have been possible just a decade ago, but machine learning now offers a shortcut. The researchers used a machine learning algorithm that calculated the energy with which water molecules interact with each other. This model performed the computation much faster than traditional techniques, allowing the simulations to proceed much more efficiently.
Machine learning isn’t perfect, so these long simulations also improved the accuracy of the predictions. Researchers made sure to test their predictions with different types of simulation algorithms. When multiple simulations gave similar results, their accuracy was validated.
“One of the challenges with this work is that because this is an issue that is almost impossible to study experimentally, there isn’t a lot of data to compare against,” Gartner said. “We’re really pushing the envelope here, which is another reason why it’s so important that we try to do this with several different computational techniques.”
Some of the conditions tested by the researchers were extremes that are unlikely to exist directly on Earth but could potentially be present in various aquatic environments of the Solar System, from the oceans of Europe to the waters at the core of comets. However, these findings could also help researchers better explain and predict the strange and complex physical chemistry of water, inform the use of water in industrial processes, develop better climate models, and more.
The work is even more generalisable, according to Gartner. Water is a well-studied area of research, but this methodology could be extended to other difficult-to-simulate materials such as polymers or complex phenomena such as chemical reactions.
“Water is so central to life and industry, so this particular question of whether water can go through this phase transition is a long-standing issue, and if we can get close to an answer, that’s important,” he said. “But now we have this really powerful new computational technique, but we don’t yet know what the limits are and there’s a lot of room to push the field forward.”