Evaluation of cleaning efficiency: The role of detection efficiency

Abstract

A simple model for a cleaner for removing particles from surfaces is NET = A-EFF*PRE, where NET is the net change in particle count, NET = POST-PRE; EFF is the cleaning efficiency, the efficiency of removal; PRE is the original particle count (before cleaning); A is the net expected when PRE goes to zero, a clean surface to start. EFF is determined from 1-POST/PRE or from a linear regression of NET versus PRE, using data from numerous experiments with different initial levels of cleanliness. Some pitfalls are discussed. The model is extended to take into account the imperfect counting efficiency of the particle counter used to inspect the surface before and after cleaning. If the counting efficiency is greater for particles removed than for particles that remain, and the linear regression approach is used, the cleaning efficiency will be over-estimated. A similar problem exists for 1-POST/PRE as the efficiency estimate, complicated further by any addition of particles. Methods of determining counting efficiency are presented for devices that count particles and devices that map particles.