The New Jersey Institute of Technology's
Electronic Theses & Dissertations Project

Title:	Microhydrodynamic, kinetic and thermal modeling of wet media milling for process optimization and intensification
Author:	Guner, Gulenay
View Online:	njit-etd2022-054 (xxxiii, 315 pages ~ 13.4 MB pdf)
Department:	Department of Chemical and Materials Engineering
Degree:	Doctor of Philosophy
Program:	Chemical Engineering
Document Type:	Dissertation
Advisory Committee:	Bilgili, Ecevit Atalay (Committee co-chair) Axe, Lisa (Committee co-chair) Armenante, Piero M. (Committee member) Guvendiren, Murat (Committee member) McEnnis, Kathleen (Committee member) Yao, Helen F. (Committee member)
Date:	2022-12
Keywords:	Nanosuspension Poorly water soluble drugs Process modeling Process optimization Wet stirrer media milling
Availability:	Unrestricted
Abstract:	Nanoparticle production by wet stirred media milling (WSMM) is a common method for the formulation of poorly water-soluble drugs. While most of the studies in the WSMM literature focus on the formulation aspects to overcome the stability challenges, a thorough mechanistic understanding of the process is lacking, and the process is slow, costly, and energy-intensive. This dissertation presents experimental and modeling work with the ultimate goals of (i) gaining a deeper and more mechanistic understanding of the WSMM process and breakage kinetics of the particles using a microhydrodynamic model with various improvements and advancements, (ii) examining the heat dissipation during the WSMM as a function of various process parameters, and (iii) optimizing and intensifying the WSMM using novel approaches such as bead mixtures of two different bead materials and mixtures of differently sized beads. To achieve the aforementioned goals, an nth-order breakage kinetics model is formulated to provide the best representation of the experimental median particle size evolution with time upon the milling of drug suspensions. Microhydrodynamic parameters are used to predict the breakage rate constant via a subset selection method, where the predictions are improved when the packing limit of the beads is taken into account. The analysis of heat generation–transfer experimental results suggest a significant rise in temperature during the milling, and stirrer speed is the most influential parameter followed by bead loading and bead size. An enthalpy balance model (EBM) is formulated to fit the experimental temperature profiles and determine the fraction of the mechanical power converted to heat, which is predicted using power law and machine learning approaches. As a low-fidelity alternative to the EBM, a semi-theoretical lumped-parameter model (LPM) is also formulated, which requires less experimental information though still provides a better estimation of temperature rise during WSMM as compared with the EBM. To improve the process, two novel process optimization approaches via bead mixtures are evaluated. When two bead materials, which are polystyrene and zirconia, are compared, polystyrene is found to be more efficient in terms of lower power consumption and heat generation, whereas zirconia beads are found to be better for fast breakage kinetics. Mixture of bead materials is introduced as a novel operational technique, to optimize the process from a holistic cycle time–power consumption–heat generation perspective. A decision tree for the composition of the bead mixture for various pharmaceutical application scenarios is developed. While the mixtures of polystyrene–zirconia beads help to reduce cycle time with acceptable temperature rise and power consumption, the mixtures of different bead sizes do not provide any significant benefit as compared with narrowly-sized individual beads. Overall, this dissertation addresses various process challenges of WSMM such as long cycle time and temperature rise, and formulates novel experimental solutions such as mixture of beads and predictive modeling techniques using various machine learning algorithms. Besides generating fundamental insights into the processing, the research hints at a new path to modeling the WSMM process via a combination of the microhydrodynamic model and population balance model augmented with machine learning approaches.