AI Helps Unravel Water's Freezing Mystery
Researchers at Osaka University have used artificial intelligence to model water's molecular structure and explain its unusual properties when freezing. The study offers a new framework for understanding water.
Osaka University researchers have employed artificial intelligence (AI) to tackle a century-old mystery: why water expands when it freezes. The team systematically evaluated 16 different methods for describing the structure of water molecules under varying temperatures and pressures.
Using an AI model, the researchers successfully identified the most accurate way to capture the microscopic changes in water. This breakthrough provides a unified scientific framework for understanding water's anomalous properties, such as its expansion upon freezing and the behavior of supercooled water.
While water is the most abundant liquid on Earth, it behaves unusually when it freezes, unlike most other substances. Previous theories have linked these phenomena to water's microscopic structure, which changes with temperature and pressure. However, the lack of a systematic descriptor has hindered deeper understanding.
The study, led by Associate Professor Kang Kim, meticulously compared various molecular structure descriptor methods using AI. This allowed for an understanding of the local order within water and clarified the evolution of its competing liquid states.
The findings help explain how water molecules organize into a crystalline structure when freezing. Additionally, the research sheds light on the behavior of supercooled water, where high-density and low-density liquid structures compete, influencing water's anomalous physical characteristics.