By Mike Fitzpatrick and Ken Goldstein, Ph.D.
Back in September, we examined how filters work and learned to recite the filtration mantra of interception, impaction and diffusion. We illustrated that particle removal by interception occurs when a section of a particle “runs into” a filter fiber and “sticks”—or, in the special case of sieving, where a particle is too large to fit between two adjacent fibers and “runs into” both.
We also discussed capture by impaction, where a particle of relatively greater mass is unable to follow the curved streamline around the fiber and, as a result of momentum, travels in a straight line into the filter fiber and “sticks.”
Lastly, we saw that diffusion capture occurs when particles leave the streamline due to random collisions with the surrounding fluid molecules and strike the fiber where they again “stick.”
All three mechanisms are constantly at work. But how well they work at removing particles depends on a number of factors, including: the internal geometry of the filter elements, the velocity of the fluid through the filter and the size or mass of the particle.
At higher velocities, particle momentum increases, resulting in improved capture by impaction. At the same time, however, increased momentum reduces the effect of the random molecular collisions, decreasing the diffusion capture efficiency. Conversely, lower velocities improve diffusion capture while decreasing the impaction capture efficiency.
In other words, we would prefer higher velocities to remove large (heavy) particles and slower velocities to remove small (light) ones. Since large heavy particles are generally absent from high-purity gases and cleanroom air, we generally tend toward the slower fluid velocities.
Particle sizes between heavy and light are the most likely to get through filters and cause damage to products. They are also the most difficult to capture.
Likewise, if we vary a particle's mass, or equivalently its size, and hold the velocity constant, we will see a similar effect. Since momentum increases with mass, particle capture by impaction improves as particle mass (size) increases. The larger mass of the particle also makes it less susceptible to being moved away from the streamline due to random molecular collisions, and diffusion capture efficiency decreases.
Conversely, diffusion capture markedly improves for small (light) particles that are less likely to be captured by impaction due to their decreased momentum. The result of all this is that small (light) particles are relatively easy to capture by diffusion while the large (heavy) particles are readily removed by impaction. It's the particles in between that give us the most trouble. Figure 1 shows this effect.
Let's explore this in detail. Large (heavy) particles don't concern us much because of capture by impaction. Similarly, small (light) particles tend to be captured by diffusion. But the particle sizes between heavy and light are the most likely to get through our filters and cause damage to our product. For these particles, neither impaction-capture nor diffusion-capture dominates, and both mechanisms operate at less than peak efficiency. We find the in-between sized particles more difficult to capture than the larger particles, which we would expect. But now consider a counter-intuitive result—these in-between particles are more difficult to capture than the small particles.
The phenomena of MPPS
For anyone who thinks of filters as no more than very small window screens, this is completely unexpected. The in-between sized particle that is most difficult to capture is referred to as the Most Penetrating Particle Size, or MPPS. The MPPS for a given filter—at a given fluid velocity—is defined as the particle size with the highest penetration, or equivalently, the lowest capture efficiency. And we find that the MPPS is not at the lowest end of the particle diameter range as intuition might lead us to expect.
Considering the mechanisms described above, an interested reader can readily determine why liquid filters are orders of magnitude worse in terms of their efficiency than gas filters. Capture efficiencies for process liquid filters seldom exceed much beyond 99+%, while efficiencies for process gas filters start at 99.9999% and are known to range up to 99.999999999999%. (Hint: consider the effect of viscosity on the three primary filtration mechanisms.)
The messy details
There is a fair bit of theory that explains how fiber filters work. Unfortunately, the theory falls well short of the real world. It assumes that all the filter fibers are the same diameter and, even more unlikely, that the fibers are all parallel to each other, equi-spaced and normal (at right angles) to the flow direction.
Why do we make these unrealistic assumptions? Because they are necessary to solve the equations. To develop the equations for a more realistic scenario would be extremely challenging, and solving them would be impossible. Quite frankly, the mathematics required to support this level of modeling just doesn't exist.
Some readers will remark that the overwhelming majority of gas and liquid filters are not fiber filters, but rather, membrane filters made of either metal (stainless steel or nickel) or thermoplastics (PTFE, PVDF, polysulfone, etc.). But if we can't model the relatively simple fiber filter in anything approaching its true complexity, we certainly can't hope to model the much greater complexity of a membrane filter.
Just because we can't realistically model filters does not mean we are totally lost. Aerosol scientists have developed numerous instruments that can be used to measure and evaluate filter performance and particle behavior—yet another case of technology racing ahead of the theory used to explain it.
Not necessarily so
We have all heard statements that a particular filter is “rated at 0.2 microns” (µm) or something similar. What does this mean? We know that filters are not absolute in their ability to capture particles but rather, they can be expected to capture a certain percentage of the total particle concentration. And there is no upper cut-off value on the particle diameters that can pass through the filter. Just because a particle is larger than some given size does not mean that it won't make it through the filter.
In other words, the expression “0.2 µm-rated filter” is not defined. It could mean that the MPPS is 0.2 micrometers. Or, it could be an implied but false claim that the filter will stop all particles that are 0.2 µm or larger. Or, it could be another bit of technical sounding jargon that sounds impressive but provides no factual information.
A final note
Some readers may be familiar with true pore filters produced by nuclear (gamma) irradiation of polycarbonate sheeting covered with a photoresist-like material. These filters work primarily on the basis of sieving by creating a window screen phenomenon on a very small scale.
True pore filters really do appear to be absolute in their capability of capturing all particles greater than their cut-off size. Admittedly, these filters are useful and commonly used in laboratory applications. But because of their pressure drop and flow characteristics, these filters are seldom used in the field where these characteristics are critically important in high-flow applications.
MICHAEL A. FITZPATRICK has participated in the design and construction of semiconductor facilities for more than 24 years and is a Senior Member of the Institute of Environmental Sciences and Technology (IEST). Mike can be reached at: [email protected] KEN GOLDSTEIN is principal of Cleanroom Consultants Inc. in Phoenix, Ariz., and is a member of the CleanRooms Editorial Advisory Board. He can be reached at: [email protected]