The high performing ULPA/HEPA fiberglass filters used to isolate the interiors of cleanrooms from contaminants in the outside ambient air do an excellent job of removing aerosol particles but fail miserably in controlling airborne molecular contamination (AMC).
The concentration of AMC inside a so-called cleanroom, protected by either HEPA or ULPA filters, typically either matches that of the outside ambient air supplying the makeup air for the facility, or exceeds it when the primary sources of AMC are within the cleanroom itself.
The distinction between contaminants classified as particles and those classified as AMC is often just one of contaminant mass/size. Particles are stable molecular clusters with dimensions at least as large as 0.002 micron. Below this size, molecular clusters become unstable and undetectable by any “particle counter.” The AMCs of interest in this note are molecules or molecular clusters with dimensions less than this arbitrary value. However, the reasons that glass fibrous filters don't offer much protection against these contaminants are more subtle than the fact that the pores of the filter media are much larger than the diameter of the typical molecular contaminant — these filter pores are also much larger than the diameter of many of the aerosol particles that the filters remove with high efficiency. Fibrous filters do not remove aerosol particles just by sieving, a mechanism in which all particles larger than the nominal pore size of the filter are removed and all those smaller than the nominal pore size pass through. Other removal mechanisms dominate much of the filtration process.
Cleanroom filters capture aerosol particles primarily by one of two non-sieving mechanisms: interception or diffusion. Interception is a mechanism whereby particles in flow streamlines are sufficiently large to be brought into contact with a fiber as the streamline in which they are entrained passes around the fiber. Capture efficiency by this mechanism increases with increasing particle diameter. Diffusion refers to the random, Brownian motion of small aerosol particles that is induced by multiple collisions of the particle with the molecules of the air. This action causes particles to depart from the airflow streamlines and to increase their probability of colliding with adjacent filter fibers. This capture mechanism is also size dependent but its capture efficiency increases with decreasing particle diameter, mirroring the dependence of particle diffusion coefficients upon particle diameter.
The plot shows size dependence of particle penetration through a fibrous filter based on interception and diffusion. This plot, from researchers at the University of Minnesota,1 shows particle penetration through a fiber filter at two values of air flow. Particle interception dominates the large particle portion of the plot; diffusion, the small particle portion. Note that the left hand ordinate — the vertical axis corresponding to the smallest diameters on the plot — is 0.01 micron and that filter penetration of particles of this diameter is predicted to be vanishingly small, just like the miniscule penetration of particles in the 0.5 micron diameter and larger range.
Indeed the plot clearly indicates an intermediate particle diameter at which the filter penetration is the greatest. This diameter is called the Most Penetrating Particle Size (MPPS). It represents the cross-over between the diffusion-dominated portion of the curve and the interception-dominated portion. For both HEPA and ULPA filters at typical cleanroom vertical air speeds (~100 fpm [2 cm/s]), the MPPS values are in the 0.1 to 0.3 micron range. Not coincidentally, this range of particle sizes is that at which the filtration efficiencies of these cleanroom filters are rated: 99.97 percent particle removal efficiency* at 0.3 micron defines the minimum acceptable particle removal efficiency of a filter that can be called a HEPA filter; and 99.999 percent particle removal efficiency at 0.12 micron defines the ULPA filter. That the particle removal efficiency of both these filters continues to improve as particle size decreases below this size range implies, by extrapolation, that smaller molecular clusters, such as those making up most AMC, ought to be removed even more efficiently than the molecular clusters called particles in the plot. This doesn't happen. Small molecular clusters, monotonic and diatomic gases and vapor molecules readily penetrate both HEPA and ULPA filters.
What's missing are the relative dependencies of the adhesion forces and the thermal energies of these molecular clusters upon their dimensions. Adhesion forces, including the universal van der Waals force, typically have a linear or higher power dependence upon the diameter of a molecular cluster. In the particle size range, these forces are large enough so that particle contact with a surface, such as a fiber, implies particle capture – the sticking coefficient following particle contact is approximately unity. With molecular clusters of dimensions less than those of particles, the reduced adhesion forces often mean that the sticking coefficient is less than 1. Thermal energy of the clusters is 3/2 kT and is independent of diameter.
At some diameter below that plotted, the penetration ratio rises, not because of reduced fiber contact – the higher diffusion coefficients of the smaller clusters ensure that contact continues to increase – but because contact no longer implies capture. So the molecules of the air and the AMC bounce their way through the same fiberglass media that captures and holds particles. The physical forces that adequately capture and retain most aerosol particles no longer do the job for AMC. Control of airborne molecular contaminants requires stronger adhesion forces such as those associated with chemical bonds. Fortunately, filters employing these stronger chemical adhesion forces are becoming available and are being used increasingly in cleanroom environmental control.
Robert P. Donovan is a process engineer assigned to the Sandia National Laboratories as a contract employee by L & M Technologies Inc., Albuquerque, NM. His Sandia project work is developing technology for recycling spent rinse waters from semiconductor wet benches.
1. Liu, B. Y. H., K. L. Rubow and B. Y. H. Pui, “Performance of HEPA and ULPA Filters”, pp 25 – 28 in the 1985 Proceedings of the Institute of Environmental Sciences (IEST, 940 East Northwest Highway, Mount Prospect, IL 60056)
* Particle removal efficiency (percent) = (1 – particle penetration) X 100