(341b) Coupling Molecular Dynamics with Computational Fluid Dynamics Via Parameter Passing for the Simulation of Polymer Nanoparticle Precipitation | AIChE

(341b) Coupling Molecular Dynamics with Computational Fluid Dynamics Via Parameter Passing for the Simulation of Polymer Nanoparticle Precipitation


Di Pasquale, N. - Presenter, Politecnico di Torino
Marchisio, D. - Presenter, Politecnico di Torino
Barresi, A. - Presenter, Politecnico di Torino
Carbone, P. - Presenter, The University of Manchester

Interest in polymer nanoparticles has grown in recent years due to their unique features as nanocarriers for controlled drug delivery. A very common technique for nanoparticle production is solvent-displacement that consists in mixing a solvent (e.g., acetone) with an anti-solvent (e.g., water). The polymer (polycaprolactone, PCL) is dissolved in the solvent, which is completely miscible with the anti-solvent. The polymer is however sparingly soluble in the anti-solvent therefore as soon as solvent and anti-solvent are mixed particles are formed. Since the process is highly sensitive to mixing, very efficient micro-devices, such as confined impinging jets reactors (CIJR), are generally employed for performing this operation.

Mathematical models are very useful in this context for reactor design and optimization. One interesting approach to describe and simulate the process is based on computational fluid dynamics (CFD) via the so-called Reynolds-averaged Navier-Stokes approach (RANS). According to this approach turbulence is modeled and mixing at the molecular level (i.e., micro-mixing) is described by using a presumed probability density function approach (i.e., direct quadrature method of moments, DQMOM, coupled with the exchange and interaction with the mean, IEM). Nucleation, growth and aggregation of the nanoparticles is modeled solving the corresponding population balance equation (PBE) through the quadrature method of moments (QMOM) [1]. One very important issue is related to the identification of all the kinetic parameters and properties needed in these models. Some of these can be obtained from independent experimental measurements (e.g., equilibrium concentration and interface tension) but for some others this is not possible. One important parameter that highly affects the nucleation rate predictions (and in turn the final predictions for a key particulate characteristic such as the particle size distribution) is the molecular volume of the polymer molecules. This changes continuously from pure acetone to pure water and therefore is not costant but changes from point to point in the CIJR.

The aim of this work is to estimate this parameter for PCL by using atomistic Molecular Dynamics (MD) simulations under different conditions (i.e., PCL in pure water, pure acetone and different intermediate concentrations). This parameter is estimated from the accessible surface area (SAS) from MD simulations,  allowing to obtain the molecular volume as a function of the composition of the solvent/anti-solvent mixture. The whole model is validated by using experimental data [2]. Preliminary results show that the approach is indeed interesting, since the main trends are well described and acceptable agreement with experiments is observed.

[1] E. Gavi, D.L. Marchisio, A.A. Barresi, Olsen M.G., and Fox R.O., Turbulent precipitation in micromixers: CFD simulation and flow field validation, Chemical Engineering Research & Design, 12:1182-1193, 2010.

[2] F. Lince, D.L. Marchisio, and A.A. Barresi, Strategies to control the particle size distribution of poly-caprolactone nanoparticles for pharmaceutical application, Journal of Colloid and Interface Science, 322:505-515, 2008.