Prediction of Viscosity for Siloxane Fluids Using United Atom Molecular Dynamics Simulations

Siloxanes are polydisperse mixtures of linear and/or cyclic oligomers where properties such as density and viscosity are influenced by the composition of those components. The ability to reformulate products with a different mix of components is advantageous in many circumstances but inevitably leads to changes in important properties, such as viscosity and density. An associated challenge is that the reformulation process is often empirical, typically requiring tedious Design of Experiments (DOEs) to achieve siloxane property targets. There is a need for computational models that can predict properties such as density and viscosity as a function of composition of siloxanes. To this end, we use molecular dynamics (MD) simulations with a United Atom (UA) force field for the prediction of density and viscosity for siloxane fluids. Focusing on siloxanes consisting of silicon atoms bonded to either two methyl groups (designated as D unit) or one methyl group and one phenyl group (designated as DPh unit), we find that MD simulations predict viscosity and density reasonably well for linear siloxanes and their mixtures over a wide range of kinematic viscosity values up to 500 cSt. However, in the case of cyclic siloxanes and mixtures of cyclic siloxanes, we find significant deviations between predicted values from simulations vs reported values. We discuss the limitations of the model in viscosity prediction of cyclic siloxanes and share learnings about potential model reparametrizations needed to predict viscosity for cyclic cases.