(569r) Analyzing the Data from An Anderson Cascade Impactor. A Computational Perspective

The Delivery of active ingredients through the pulmonary pathway has received a considerable amount of attention in the past 10 years. The large amount of surface area and relatively quick uptake time make the lungs a good pathway for drug delivery. In recent years researchers have focused on delivering dry powder aerosol particles to the lungs. It is well understood that in this mode of drug delivery the particle size distribution of the active aerosol determines the success or failure of the drug delivery. This creates the need for tools to accurately and quickly test the particle size distribution being emitted from a dosation device. Currently the most commonly used device to try to determine aerosol size distribution is an Anderson Cascade Impactor.

Anderson Cascade Impactors is designed to impinge air onto a metal collection plate at increasing velocities as the air travels through the impactor. In the initial stage of the impactor, where the velocity is the lowest only larger particles with more thermal momentum impact on the collection plates. As the velocity of the air impinging on the surface increases smaller and smaller particles are collected. The plates can then be weighed for mass of impacted particle or sent for HPL analysis.

Inspite of these impactors serving a relatively important roll in dry powder delivery research little work has been done to evaluate the ability of these impactors to differentiate between different particle size distributions of similar size. The goal of this work is utilize currently available computational tools to determine how much information can be be taken from the ACI devise. Special focus will be placed on what practical information can be derived from the output of the ACI test (mass on each stage).

In order to achieve this a computational model has been developed that closely matches real world ACI performance. Variable realistic PSD have been used as inputs and mass impaction is the model output. Various methods for inverting the output of the model back to an approximate PSD. These estimated PSD are then compared to the original input PSD and the method of inversion can be characterized for its accuracy.