Statistical Engineering Division, CAML
School of Mechanical Engineering, Purdue University
Fire Measure and Research Division,BFRL
As computers become faster and smaller, so has the need to upgrade clean rooms for processing semiconductor chips. Semiconductor chips are susceptible to fine pollutants, such as dusk particles. The traditional instrument for sizing aerosol particles is the Differential Mobility Analyzer (DMA). A simplistic description of the DMA is that the instrument is just two charged concentric cylinders with an inlet slit and a sampling slit. The DMA seperates particles based on their electical mobility. Aersol particles for sizing are inserted into the annular region between the two cylinders at the inlet slit. They are carried by clean air flowing through the annular region. Particles with mobilities in a certain narrow range are sampled at the sampling slit. Then, an inversion calculation is done to infer the size distribution. Of course, the sizing depends on certain adjustable parameters such as the voltage, the flow rates, etc. For sizing nanometer particles Brownian motion becomes important. The clearest analytic understanding of the working of the DMA can be found in a paper by Knutson and Whitby. Their analysis of the trajectories of aerosol particles within the DMA is based on stream functions. This approach does not work if Brownian motion is important. Our task has been to rework this paper. Our anaysis of the trajectories depends only velocity and electric field distributions. This work is part of a collaboration with George Mulholland, who is producing a small particle SRM for the semiconductor industry.
Date created: 7/20/2001