Scientists have long suspected that water exists as two distinct liquid phases at the molecular level, and new research using artificial intelligence has now provided direct evidence of this phenomenon. Researchers deployed machine learning algorithms to analyze water's behavior at extreme conditions, observing molecular interactions that reveal water transitioning between two different liquid states.
The study builds on decades of theoretical work suggesting water's anomalous properties stem from its ability to shift between a low-density liquid phase and a high-density liquid phase. These phases differ in how water molecules arrange and interact with one another. Previous experiments hinted at this behavior indirectly through measurements of water's density, viscosity, and other physical properties at varying temperatures and pressures. The new AI approach allows scientists to observe the actual molecular rearrangement in real time.
Water exhibits unusual characteristics compared to most liquids. Its density increases as it cools from room temperature until reaching 4 degrees Celsius, then becomes less dense as it approaches its freezing point. This anomaly explains why ice floats on liquid water, a property essential for aquatic life. The two-liquid hypothesis suggests these peculiarities result from water molecules organizing themselves differently depending on conditions.
The AI-powered analysis examined how water molecules form and break hydrogen bonds during phase transitions. Machine learning models trained on molecular dynamics simulations identified distinct structural patterns corresponding to the two liquid phases. This computational approach circumvents limitations of traditional experimental methods that struggle to capture rapid molecular events at the conditions where the transitions occur.
Understanding water as a two-liquid system has implications beyond basic science. It could help explain water's role in biological processes, inform industrial applications requiring precise control of water's properties, and refine climate models that depend on accurate representations of water behavior. The research also demonstrates AI's growing utility in analyzing complex molecular phenomena that resist direct observation.
The findings remain theoretical pending experimental confirmation under the extreme conditions where the phase transition occurs. Researchers