We elucidate the microscopic picture of structuring in water using an assortment of data science techniques which offer us a data-centric and agnostic approach to understanding molecular environments. In particular, using Point Adaptive K-nearest neighbor estimator(PAK-NN) in conjunction with the dimensionality reduction techniques and Density Peak clustering(DPA), we proceed to make sense of the free energy landscape of a realistic water model (TIP4P/2005 ) in the temperature range of 230-300K. The environments of water are characterized using the Smooth Overlap of Atomic Positions (SOAP), a structural descriptor. Analysis of the free energy landscape in the SOAP space reveals a rough single basin in an 8-dimensional landscape at room temperature. We interpret this using traditional order parameters such as the Tetrahedrality and Local Structure Index(LSI), as well as new descriptors such as distance from hexagonal Ice in the SOAP space. A decrease in the temperature is found to result in a decline in the sensitivity of some of the traditional order parameters in predicting the position of a water environment in the free energy landscape.

Towards unraveling a microscopic description of structuring in water using a data science approach

ACS Spring Meeting 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.