**Analysis of structural correlations in a model binary 3D liquid through
the eigenvalues and eigenvectors of the atomic stress tensors**

VA Levashov, JOURNAL OF CHEMICAL PHYSICS, 144, 094502 (2016).

DOI: 10.1063/1.4942863

It is possible to associate with every atom or molecule in a liquid its own atomic stress tensor. These atomic stress tensors can be used to describe liquids' structures and to investigate the connection between structural and dynamic properties. In particular, atomic stresses allow to address atomic scale correlations relevant to the Green-Kubo expression for viscosity. Previously correlations between the atomic stresses of different atoms were studied using the Cartesian representation of the stress tensors or the representation based on spherical harmonics. In this paper we address structural correlations in a 3D model binary liquid using the eigenvalues and eigenvectors of the atomic stress tensors. This approach allows to interpret correlations relevant to the Green-Kubo expression for viscosity in a simple geometric way. On decrease of temperature the changes in the relevant stress correlation function between different atoms are significantly more pronounced than the changes in the pair density function. We demonstrate that this behaviour originates from the orientational correlations between the eigenvectors of the atomic stress tensors. We also found correlations between the eigenvalues of the same atomic stress tensor. For the studied system, with purely repulsive interactions between the particles, the eigenvalues of every atomic stress tensor are positive and they can be ordered: lambda(1) >= lambda(2) >= lambda(3) >= 0. We found that, for the particles of a given type, the probability distributions of the ratios (lambda(2)/lambda(1)) and (lambda(3)/lambda(2)) are essentially identical to each other in the liquids state. We also found that lambda(2) tends to be equal to the geometric average of lambda(1) and lambda(3). In our view, correlations between the eigenvalues may represent "the Poisson ratio effect" at the atomic scale. (C) 2016 AIP Publishing LLC.

Return to Publications page