The Many Phases of Water

Despite its ubiquitous presence in the everyday environment, our understanding of water has only slowly improved over the years. You may remember the scene in the 1993 movie Jurrasic Park where chaos theory is explained using a dripping water faucet. This scene is based on real science. Certain properties of water remain notoriously difficult to predict but with recent advances in cyberinfrastructure and advanced resources like NCSA’s Delta GPU-based supercomputer, simulations of even the most difficult models can be attempted.

Machines aren’t the only things that have advanced far enough to make this breakthrough possible. The methodology has improved in the last three decades as well. Utilizing a method called first principles data-driven quantum simulations, the research team led by Francesco Paesani, chemistry and biochemistry professor at UC San Diego, was able to create a highly detailed many-body model of water called MB-pol. Their approach involved using the fundamental laws of quantum mechanics to predict the behavior of water molecules.

“Water’s simple formula belies its complex behavior,” Paesani said. “Using MB-pol, we’ve been able to model water across a wide range of temperatures and pressures, providing insights into how factors such as enthalpic, entropic and nuclear quantum effects shape its free-energy landscape. This work illustrates how recent advancements in first-principles, data-driven simulations have opened the door to realistic computational studies of complex molecular systems.”

Detailed in their study published in Nature Communications, the team’s breakthrough involved first focusing on constructing an accurate phase diagram for water. This phase diagram shows how water behaves at different temperatures and pressures, revealing the various forms water can take, solid, liquid and gas. They then used the phase diagram to inform simulations performed on supercomputers. To test the accuracy of the simulation, they compared it to real-world observations of water’s phase transitions. Their simulations matched the real-world observations very closely, suggesting that their model is highly accurate and reliable.

Sigbjørn Bore, a postdoctoral researcher who worked with Paesani on the project, spoke highly of the resources used in their research. “Given the computational demands of our extensive simulations, it is important to underscore that our comprehensive explorations of water’s phase diagram, spanning wide temperature and pressure ranges over long timescales, would not have been possible without the availability of high-performance computing resources,” he said.

Such complex computer simulations as those done by the research team would have taken significantly longer without the computing power provided by NCSA Delta and Expanse, at the San Diego Supercomputer Center. Delta is one of the newest supercomputers in the NSF portfolio. The choice to create Delta as a GPU-heavy supercomputer is already paying dividends, as Delta was able to perform a large proportion of the computational work to complete the simulations.

A bonus of being able to run simulations on how water behaves: You can put water under much more extreme conditions than is easily replicable, or even possible, with a real-world test. The simulation can put water under the immense pressure of the deep ocean, for instance, or show what happens to ice in the vacuum of space.

Studies like these show the importance and power of using data-driven quantum simulations. By utilizing advanced computational methods, scientists can study complex systems like water with remarkable accuracy and detail in a highly customizable environment allowing for a greater understanding of complex systems. But there’s also a more immediate impact to this research. Using the MB-pol model, scientists could study how water might behave in a deep sea station or even on Mars. If humans ever plan to inhabit other planets, it’ll be simulations like these that ensure the fundamental necessities of life are taken care of for future explorers. 

Delta is an advanced computing and data resource funded by the National Science Foundation through its Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS) program.

You can read more about this story here: Groundbreaking Simulations of Water’s Phase Diagram Advance Study of Complex Molecular Systems