(103d) Comparison of Equilibrium and Non-Equilibrium Methods and Force Fields for Calculating the Thermal Conductivity of Industrial Fluids

The ability to perform accurate prediction of the thermal conductivity of materials is of interest in a variety of fields, ranging from high thermal conductivity applications in the nuclear and semiconductor industries, to lower conductivity industrial fluid systems of interest in microfluidic devices and other engineering applications. While the principles of computational methods for calculating thermal conductivity, either from equilibrium or non-equilibrium simulations, have been known for some time, widespread application has been limited in view of the high computational cost involved. With the growing availability of compute clusters with tens to thousands of CPUs, accompanied by simulation systems designed specifically to take advantage of such architectures, routine calculation of thermal conductivity is becoming increasingly feasible.

In recent work, we have used the LAMMPS simulation program [1] to investigate the thermal conductivity of a variety of materials, including non-polar hydrocarbon, moderately polar and hydrogen bonded (alcohol and diol) fluids, employing both equilibrium and non-equilibrium simulation methods. This presentation will include a comparison of the equilibrium Green-Kubo [2, 3] approach with the more recent reverse non-equilibrium molecular dynamics method in the form introduced by Muller-Plathe et al. [4]. In addition, we will discuss the accuracy of predictions made using two force fields - OPLSAA and COMPASS - which have been parameterized with an emphasis on quantitative prediction of condensed phase properties, but without any particular consideration of the thermal conductivity.

References

1. Plimpton, S.J. , J Comp Phys, 117, 1-19 (1995).
2. Green, M.S., J. Chem. Phys., 22, 398 (1954).
3. Kubo, R., J. Phys. Soc. Japan, 12, 570 (1957).
4. Muller-Plathe, F., and Reith, D., Comp. Theor. Polymer Science 9, 203 (1999).