Thermodynamic extrapolations predict fluid structural properties at many conditions

Researchers simulate the microscopic structure of a fluid to calculate the spatial arrangement of different molecules, a polymer chain’s degree of expansion, or the size distribution of micelles suspended in solution. With that information, they learn how the fluid structure influences macroscopic properties such as pressure, density or viscosity. This then allows tailoring of the fluid to produce desired thermodynamic or dynamic behavior.

Traditionally, each simulation only provides information about the structural property of interest for a given set of conditions — temperature and chemical potential, for example — for which it was performed. This means that a complete description of the fluid structure requires many simulations over a broad range of thermodynamic states. Now a team of researchers has identified a way to extrapolate these structural properties over a continuous range of states.

Reporting in The Journal of Chemical Physics, the researchers first calculated a certain structural property of the fluid, such as its radial distribution function, radius of gyration, or cluster size distribution, at a few specific temperatures or chemical potentials. By measuring fluctuations in observable properties, the mathematical tool of a Taylor series can describe how changes in temperature, for example, affect these structural properties in a continuous way.

These Taylor series relationships enable extrapolation of a particular structural property to new thermodynamic conditions without having to perform any new simulations. A similar approach has previously been used to extrapolate thermodynamic, but not structural, properties of fluids.

This approach could improve the computational efficiency of structural property calculations, as a few simulations can now provide a large amount of data over a wide range of conditions. That data can then be utilized into systems powered by machine learning or artificial intelligence.

Source: “Predicting structural properties of fluids by thermodynamic extrapolation,” by Nathan A. Mahynski, Sally Jiao, Harold W. Hatch, Marco A. Blanco, and Vincent K. Shen, The Journal of Chemical Physics (2018). The article can be accessed at https://doi.org/10.1063/1.5026493 .