Low, high and very-high density forms of liquid water revealed by a medium-range order descriptor

TL;DR

This paper introduces a graph-theory based order parameter from molecular dynamics simulations that characterizes various density phases of liquid water, revealing new high-density clusters under moderate pressures.

Contribution

A novel medium-range order descriptor, node total communicability, is proposed to identify different liquid water phases and detect very high density water clusters.

Findings

Identifies high, low, and very-high density water phases.

Detects very high density water clusters under moderate pressures.

Provides a new computational approach for analyzing liquid water structure.

Abstract

We present in this paper a computational approach based on molecular dynamics simulations and graph theory to characterize the structure of liquid water considering not only the local structural arrangement within the first (or second) hydration shell, but also the medium- to long-range order. In particular, a new order parameter borrowed from the graph-theory framework, i.e. the node total communicability (NTC ), is introduced to analyze the dynamic network of water molecules in the liquid phase. This order parameter is able not only to accurately report on the different high-density-liquid (HDL) and low-density-liquid (LDL) water phases postulated in the liquid-liquid phase transition hypothesis, but also to unveil the presence of very high density liquid (VHDL) clusters, both under pressure and at ambient conditions. To the best of our knowledge, VHDL water patches under moderate pressures were not observed before.