Scientists observe water switching between 2 different molecular structures, boosting decades-old theory
For years, scientists have suspected that, at the molecular level, water are two different liquids – one denser and one less dense – which are constantly changing places. Actual molecular evidence of this subtle change has been difficult to capture. But now, with the help of artificial intelligenceResearchers say they have finally found it.
“It’s hard to imagine – there’s only one body of water here, right?” Said xiao cheng zenga physical chemist at the City University of Hong Kong and co-author of the new study, told Live Science while holding a water bottle in the air. That puzzle sent him digging into the scientific literature, where he found a possible explanation: the two-state hypothesis. “That caught my attention. We have literature talking about this but no evidence.”
The findings, published June 4 in the journal nature physicsNot only could this prove that this long-expected molecular change is real, but it could also help explain dozens of strange behaviors of water.
Most liquids become denser as they cool, but water behaves differently; It condenses to about 4 °C, then begins to expand, causing the ice to float. Water also resists temperature changes better than similar liquids and its viscosity decreases at certain pressures. Scientists have documented various water-related anomalies and suspect that they may be related.
The two-state model is an attempt at that unified explanation.
30 year forecast
Zeng has been studying water since his postdoc days in the late 1990s, when he worked on liquid freezing. The two-state hypothesis came onto his radar later – around 2006, when he first encountered it at scientific conferences. But for years he avoided it as too difficult to deal with directly. This changed around 2016, as researchers began reporting experimental evidence that supercooled water could split into distinct higher-density and lower-density forms.
About two and a half years ago, Zeng submitted the problem liwen liA postdoctoral researcher in his laboratory. Rather than repeat traditional approaches that other groups had already struggled with, Lee suggested the use of “unsupervised deep learning” – AI trained to detect patterns in data without telling it what to look for.
Get the world’s hottest discoveries delivered straight to your inbox.
“So A.I. [is] forced to learn – to get used to [its] “Knowledge to create, to explore,” Zeng told Live Science.
The team ran large-scale molecular dynamics simulations using GROMACS simulation package. They tracked how hundreds of thousands of water molecules move and interact and generated millions of data points.
“Traditionally, you might need a lot of students to understand it. … With computers and AI, it takes time [Li] Maybe a year and a half,” Zeng said. He estimated that without AI, the same analysis could take closer to a decade.
AI was used to study the molecular structure of water.
(Image credit: Vertigo3D via Getty Images)
The AI came back with “reaction coordinates” – a small number of variables distilled from all the molecular motion, which describe how the local arrangement of a water molecule transfers from a denser structure to a looser structure and back again. They plotted the behavior of the system along those coordinates to see the size of the transformation. This includes the number and location of energy barriers, or saddle points, that the molecules must overcome to switch.
two paths on the mountain
The team found that the path the two structures take to transform into each other changes depending on certain conditions. Most of the time, the switch occurs on what researchers call a “semi-loop” pathway, which has a single energy barrier to overcome.
But near the boundary between high-density and low-density water — the same kind of boundary where ice and liquid water coexist at 32 degrees Fahrenheit (zero degrees Celsius) — molecules can take a more circular “full-loop” path, with three different obstacles instead of one.
Zeng compared it to hiking up a mountain that is cut in half, with a gentle slope on one side and a steep cliff on the other. Most hikers stick to the slopes; That’s a semi-loop. But near the border where the two parts meet, it seems as if the mountain is becoming whole again, giving hikers the opportunity to circumnavigate the entire peak. That’s the whole loop.
Zeng and his team are now building a more rigorous machine-learning model to confirm the results. They hope to eventually link it to properties such as density, viscosity and temperature.
Confirming the structure in real water will not be easy. Zeng said this will likely require new and sensitive experimental techniques – such as those developed by laboratories like Pacific Northwest National LaboratoryIn which indirect spectroscopic evidence for the two-state behavior of water was first found.
“Once we have this…confirmed experimentally,” he said, “this model can be used [understand] How water interacts with nature.”
Since most biological and pharmaceutical processes occur in water, a better understanding of the molecular structure of water can shed light on how dissolved salts, proteins, and drug molecules interact in solution. “These interactions are important for injectable drugs and cell function,” he said, “but applying this knowledge to practical use is still a long way off.”
