Multi-scale analysis and simulation of powder blending in pharmaceutical manufacturing

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Chemical Engineering, September 2005."August 2005."Includes bibliographical references.A Multi-Scale Analysis methodology was developed and carried out for gaining fundamental understanding of the pharmaceutical powder blending process. Through experiment, analysis and computer simulations, microscopic particle properties were successfully linked to their macroscopic process performance. This work established this micro-to-macro approach as a valid way to study unit operations in the pharmaceutical manufacturing of solid dosage forms. The pharmaceutical materials investigated were anhydrous caffeine, lactose monohydrate and micro- crystalline cellulose (MCC). At the macroscopic level, blending experiments were conducted in mini-scale lab blenders using the Light-Induced Fluorescence (LIF) technique. Effects of operating parameters on blending kinetics were systematically evaluated. It was found that the time required to reach a homogeneous mixture (thg) increased with blender fill volume (FV) and decreased with blender rotation rate (RPM). It was also found that MCC, as an excipient, always took longer time to mix with caffeine than lactose. At the microscopic level, force interactions - cohesion/adhesion and friction - were measured directly at the single particle level with Atomic Force Microscopy (AFM).(cont.) It was found that cohesion/adhesion and friction fell into lognormal distributions. Based on AFM force maps, these distributions were attributed to the particle surface morphology. Chemically modified AFM cantilever tips were used to probe the hygroscopicity on the particle surfaces. In addition, the cohesive/adhesive forces were found to be size- dependent and thus, were converted to JKR surface energies to eliminate this dependence. Amongst the materials tested, MCC showed the strongest cohesive/adhesive and friction interactions. The AFM-measured microscopic force interactions were used to explain the blending kinetics profiles observed in the blending experiments. The longer blending time (thg) required by MCC was linked to its strong cohesive nature. In addition to these multi-scale relations, the AFM force interactions were used in Discrete Element Method (DEM) models for simulating blending processes. A two-dimensional model was used to simulate blending in a circular blender. With respect to the effect of operating parameters on blending kinetics, the simulations showed that thg increased as FV increased, RPM decreased, or when MCC as opposed to lactose was chosen as the excipient.(cont.) These trends were identical to experimental observation. A three dimensional DEM code was developed. Blending in a V-shaped blender was simulated and results were consistent with experiments, namely the flow behavior correlated well with the differences in cohesion/adhesion and friction intensities of the excipients. Through a fundamental understanding at a microscopic level, one can identify opportunities for process improvement. In this way, Multi-Scale Analysis will facilitate the ability of pharmaceutical companies in pursuing the desired quality-by-design state in manufacturing.Samuel SH Ngai.Ph.D