(442i) Investigation of Mixing and Segregation of Ordered Mixtures for DPI Formulations

Authors: 
Sarkar, S., University of Connecticut
Minatovicz, B., University of Connecticut
Thalberg, K., Astra Zeneca
Chaudhuri, B., University of Connecticut

The purpose of the current study is to guide the pharmaceutical industry towards rationally choosing blenders and processing conditions for making ordered mixtures based on relevant criteria. Ordered mixtures were formed in a high shear drum mixer (HSM) and a low shear double cone blender (DCN) . Mixing patterns in these blenders were systematically investigated using experiments and numerical models. 6 samples were taken at different spatial locations within the blenders and analysed to monitor the fines fraction (<=10μm) using a Malvern Mastersizer 2000E instrument fitted with a dry Scirocco dispersion unit operated at a pressure of 4 bars. The performance of blenders was tested with respect to the following criteria: time, demixing tendency, press-on-forces, blend tribocharging and wall sticking of the drug. In addition, process variables (rotation speed, loading configuration and fill) were investigated as a function of time. Progress of mixing within the blenders was done by plotting the Relative Standard Deviation (RSD) as a function of time. In addition, hopper discharge of the ordered mixtures were observed through hoppers in mass and funnel flow regimes. Discrete Element Method (DEM) based numerical model, which computes all forces acting on particles to capture detailed contact dynamics is used to simulate the powder flow within the blenders. The estimated cohesive and adhesive forces between particles are used in the numerical model to simulate powder flow. The cohesive and adhesive forces are included through dimensionless Bond numbers, which represent the ratio of cohesive force to particle weight.

Increased rotation speed and a central loading configuration were associated with the fastest mixing, but increased speed was also associated with a greater tendency to demix and wall adhesion of API. Ordered mixtures from DCN were formed after a much longer time and were charged more, but had a lower tendency to demix. Press-on forces of the blend from both the blenders were not differentiable. DEM simulations revealed that HSM achieved greater velocities but produced lower chaos, while the inverse was true for DCN. DCN was predicted to form adhesive mixtures quickly given the adhesion between the drug and carrier was strong. The effect of material adhesion was not pronounced for HSM. HSM was also predicted to approach the theoretical limit of ideal ordered mixture in contrast to DCN.