Abstract
Cascade impactors, operating on the principle of inertial size separation in (ideally) laminar flow, are used to determine aerodynamic particle size distributions (APSDs) of orally inhaled product (OIP) aerosols because aerodynamic diameter can be related to respiratory tract deposition. Each stage is assumed typically to be an ideal size fractionator. Thus, all particles larger than a certain size are considered collected and all finer particles are treated as penetrating to the next stage (a step function stage efficiency curve). In reality, the collection efficiency of a stage smoothly increases with particle size as an "S-shaped" curve, from approximately 0% to 100%. Consequently, in some cases substantial overlap occurs between neighboring stages. The potential for bias associated with the step-function assumption has been explored, taking full resolution and two-stage abbreviated forms of the Andersen eight-stage nonviable impactor (ACI) and the next-generation pharmaceutical impactor (NGI) as example apparatuses. The behavior of unimodal, log-normal APSDs typical of OIP-generated aerosols has been investigated, comparing known input values to calculated values of central tendency (mass median aerodynamic diameter) and spread (geometric standard deviation, GSD). These calculations show that the error introduced by the step change assumption is larger for the ACI than for the NGI. However, the error is sufficiently small to be inconsequential unless the APSD in nearly monodisperse (GSD ≤1.2), a condition that is unlikely to occur with realistic OIPs. Account may need to be taken of this source of bias only for the most accurate work with abbreviated ACI systems.