Interdependence of particle filter collection mechanisms and test methods
Determining particle-size dependent efficiency of high-efficiency filters.
By David S. Ensor and James T. Hanley, research triangle institute
Particle filters are critical to creating and maintaining clean environments. This article will develop a framework of air filter performance (based on the particle size dependent mechanisms of collection) and how this relates to filter test methods. A filter test method reflects the intended use of the instrumentation and particle generation methods available at the time of implementation. Methods to determine particle size dependent efficiency including IEST-RP-7.1 for high efficiency filters and the proposed ASHRAE 52.2 for pre-filters will be reviewed. Sources of error and quality assurance requirements will be emphasized.
The testing and evaluation of filters is a critical step in determining suitability for the application. Filters used for gas filtration are widely applied. The typical applications found in clean industries include the roughing or pre-filters for cleanrooms and the High Efficiency Particulate Air (HEPA) and Ultra Low Penetration Air (ULPA) filters used in ceilings of cleanrooms and point-of-use filters used for ultrapure gas streams. This article will review filter testing in light of the particle size dependent efficiency. The paper is divided into three parts: fundamentals, test methods and quality assurance in filter tests.
Fundamentals
Particles are captured in clean filters by reasonably well understood collection mechanisms. These mechanisms are critically dependent on the particle size, air velocity and properties of the filter such as fiber diameter and packing density. These are called the classical mechanisms. However, there are additional behaviors of particle re-entrainment and alternation of the filter performance with loading that are not well understood. These are called non-classical effects.
Classical mechanisms
The classical collection mechanisms for particle capture from fluid flowing past a fiber are illustrated in Figure 1. These five mechanisms are the major particle removal mechanisms over the entire particle size spectrum at ambient conditions. Following are brief descriptions of each mechanism in terms of collection on a single fiber.
Gravity. Large particles (>5 µm) will settle out of the gas stream. This action occurs independently of the fiber and depends on the gas velocity and the particle mass.
Direct interception. A particle following the upstream flow streamlines is collected when it collides with or touches a fiber. This mechanism can be important over the full particle size range and depends on the ratio of the particle diameter to the fiber diameter or the interception parameter (Dp/Df). For very large particles, a limiting case of interception called “sieving” occurs when the collected particles are larger than the openings in the filter.
Inertial impaction. Inertial impaction occurs when the inertia of the particle causes it to deviate from its initial streamline and collide with the collector (a fiber of the filter). Inertial impaction depends on the gas viscosity and velocity, particle density, the square of the particle diameter and the fiber diameter. Normally, inertial impaction is an important collection mechanism for particles greater than about 0.5 µm in diameter.
Diffusion. Particles >5 µm have a small enough mass that their trajectories are altered by collisions with gas molecules. This phenomena is called Brownian motion and can be described by a diffusion coefficient – the diffusional collection efficiency scales as a function of the product of the fiber diameter and particle diameter to the inverse 2/3 power. Therefore, smaller particles have higher diffusion rates and the random deviation from streamline (due to diffusion) leads to an increased probability of collection by the fiber.
Electrostatic mechanisms. Electrical charges on either the particle or the fiber, or both, create attractive electrostatic forces between the filter and the particle. Some filter media have permanent charges incorporated into the fiber. Electrostatic enhancement has also been demonstrated by applying external electric fields to the filter.
These mechanisms at the single fiber level are summed to yield a net collection efficiency and can be described in the following equation for the media:
Ef = (1 – Pt)100
Pt = exp [-4haL/(pDf)]
Where:
Ef = collection efficiency
Pt = penetration
h = single fiber efficiency, sum of individual collection mechanisms
a = packing density of the filter
Df = mean fiber diameter
Non-classical effects
The non-classical mechanisms are difficult to describe from first principles because of their complexity and random nature. Particle re-entrainment or bounce is sometimes observed in low efficiency filters filtering large dry particles. Even if the particle collides with the filter, the particle may not stick to the filter`s fibers.
Filter loading, or the deposition of solid particles in the filters with use, changes both the pressure drop and the particle size dependent efficiency of the filters. Particles collected on fibers tend to collect in structures called dendrites. The dendrites efficiently collect other particles. Ultimately a “cake” will form on the structure of the filter. Both the particle collection efficiency and the pressure drop is increased in the filter as the loading increases.
Liquid particle deposition in the filter can also produce unexpected results as reported by Bergman et al (1986) and Payet et al (1992). Repeated testing with a liquid aerosol can cause re-emission of test fluid as an aerosol from the filter and modify performance of the filter.
Most penetrating particle size
As seen in Figure 2, the net collection efficiency of a gas filter is to produce a region of minimum efficiency (or maximum penetration) called the Most Penetrating Particle Size (MPPS). The MPPS is the dominant feature of particle size dependent efficiency curves. It has been demonstrated for all gas filters no matter the nature of the filtration matrix. If the filter performance is plotted as particle penetration as a function of particle diameter for filters with the highest to the lowest efficiency, the same general form is observed. The MPPS is shifted somewhat depending on the design of the filter. An approach to predict the MPPS using the diffusion and interception mechanism was reported by Lee and Liu (1980). To illustrate this point, Figure 3 shows penetration as a function of particle diameter for filters ranging from general ventilation filters to ULPA filters. Note that these filters were not evaluated at the same air velocity but at the air flow rate appropriate for the filter. In addition, Figure 2 shows particle size efficiency curves for low and high efficiency filters illustrating the MPPS from the additive effects of the individual collection mechanisms. The low efficiency curves were measured by Hanley et al (1994) and the high efficiency filters were reported by Liu et al.(1985).
Filter parameters
The filter designer has many variables to use in the design of media (see Figure 2). Fractional penetration as a function effect of fiber diameter and gas velocity, of particle diameter showing MPPS for both theoretical equations were used to compute low and high efficiency filters. Other considerations such as dust holding, cost and pressure drop were not estimated. In Figure 3, the effect of fiber diameter on filtration efficiency is computed holding other parameters constant. In this calculation, reducing the fiber diameter increases the filtration efficiency by enhancing the interception and diffusion collection efficiencies. In Figure 4, the effects of face velocity (volumetric flow rate/filter area) are illustrated. Increasing the gas velocity reduces the effect of diffusion collection and enhances the impaction mechanism.
Standard filter test methods
Filter test methods (see Table 1) are intended to determine the performance of the filter under specific controlled conditions. Usually the tests are conducted under well-controlled laboratory conditions. However, situations exist where the filters may be tested under field conditions such as integrity testing of mounted filters for leaks.
Conceptual filter testing is simple. The upstream and downstream concentration is measured across a filter under controlled air flow conditions. However, the tests can produce artifacts if not carefully conducted. Modern filter test method approaches as described by IEST RP-7 and ASHRAE 52.2 are quite similar. The reason for this is the developers of modern standards have incorporated the ideas from RP-7. Test conditions are adjusted depending on the efficiency of the filter.
There are two general types of test standards:
Detailed specification of the method. In this standard, every “nut and bolt” is described. The expectation is that if the test equipment and procedures are exactly duplicated, the test results can be duplicated as well. Usually the difficulty is if unknown effects, such as environmental conditions or operator skill, may greatly influence the results, the data from different laboratories may not be comparable. Also, laboratories may build systems which are standard-like and may not produce similar test results.
Performance-based standard. A performance-based approach specifies data quality and uses reference standards. The advantage is that considerable latitude may be given with respect to details of how analytical equipment and test rig layout are implemented. The advantage of this approach is that new technology can be implemented without rewriting the standard. Modern filter test standards have taken aspects of this approach by developing ways to test and qualify the test system.
A key aspect to any test method is inter-laboratory or “round robin” studies. However, particle filter testing is limited by traceable standards and reference filters. A test of a filter will alter its performance in subsequent tests.
Existing test standards
Recent activities by the Department of Energy include substituting particle counters for optical photometers, Scripsick (1986). There are two concepts in these standard methods:
Evaluation at the most penetrating particle diameter used for high efficiency products such as the HEPA filters. The methods developed nationally for this test use different generators, test materials and detectors to accomplish this goal.
Challenge the filter with a standard test dust or aerosol and measure overall collection efficiency. The efficiency of the filter depends on repeatability of the test aerosol, which in the ASHRAE dust spot test depends on the atmosphere.
Generally these older standards are methodology-based.
Recommended Practice IEST-RP7.1
In the 1980s, it became evident (Cadwell, 1985) that the hot DOP (Dioctyl Phthalate) method developed to test the HEPA filter was inadequate to evaluate new filters designed to meet the needs of the electronics industry for cleaner cleanrooms. Also the use of DOP may be a concern because of potential health hazards and sensitivity of products to organic contamination. The IEST has responsibility for Federal Standard 209, an air cleanliness standard, and has an active program of developing recommended practices (RP) to provide guidance related to the standard. The scope of the RP is directed towards production testing of most penetrating particle size and pressure drop for ULPA filters using particle counters. The value of this RP is that the components of the test equipment were analyzed. Tests were proposed to qualify and calibrate the system. This document has supported movement towards performance-based test standards. The quality assurance tests will be described in the next section.
Proposed test ASHRAE 52.2
The need for the new test method was described by Ensor et al (1994). The existing ASHRAE Arrestance test was developed to provide a measure of filter protection of air conditioning coils and the atmospheric dust spot was developed to indicate how filters remove particles responsible for staining. The purpose of the proposed ASHRAE 52.2 test was to provide information which can be used to calculate the ability to remove contaminants. The test uses filtered air into which an aerosol generated from a solution of KCl is injected. Particle counters are used to measure upstream size distribution from 0.3 to 10 µm. After an initial efficiency measurement, the filter is loaded with dust in five different steps and the particle size efficiency is measured after each loading. The pressure drop across the filter is also measured in each step. The proposed test method draws heavily on and conforms to RP-7.
Quality assurance
The data quality framework for developing a filter test will be outlined through the various steps:
Establish test objectives and conditions. This includes selection of particle size range, products to be tested and measures of data quality.
Build or prepare test equipment.
Perform control tests. The control tests are described in the next section. The 100-percent penetration test is very important to detect testing artifacts.
Evaluate performance; modify if needed.
Perform filter tests.
Quality assurance testing of the aerosol analyzer
Whatever aerosol instrument is chosen, its calibration should be verified prior to use. If a large number of tests are to be performed, routine calibration checks should be performed throughout the test series. The QA checks should include:
Checking the sizing accuracy of the instrument. This is often performed with a monodisperse PSL or VOAG generated aerosol.
Checking the zero count of the instrument. This is often formed by placing a HEPA filter on the instrument`s sample inlet.
If two aerosol instruments are used to allow simultaneous sampling upstream and downstream sampling, the correlation of the two instruments must be measured frequently.
When possible, the coincidence level of the instrument for the challenge aerosol should be measured. Aerosol instruments often have an upper limit on the concentration that can be measured without significant error due to “coincidence” (i.e., the presence of more than 1 particle in the instrument`s view volume at a time). The coincidence level can be determined experimentally by controlled dilution tests or by conducting the filter tests over a range of challenge concentrations. At a minimum, the aerosol concentration should be kept well below the manufacturer`s specification (e.g., 1/10 of the specified coincidence level). Manufacturer specifications are often based on a 10-percent coincidence error. This amount of error is unacceptable for some types of filter tests. If higher concentrations must be needed for high efficiency filters then a suitable dilution system needs to be used. If the downstream concentrations are dilute, they require sufficient particle counts for a statistically valid sample.
Control testing of the test rig
The purpose of conducting control tests on the test rig is to demonstrate that the test rig and sampling procedures are capable of providing reliable fractional penetration measurements. Such tests, called for in a number of standardized test methods, are an important part of any quantitative test method. Aerosol-related qualification tests include 100-percent penetration test or determination of the correlation ratio; representativeness of the upstream sample; and representativeness of the downstream sample.
100-percent penetration tests. Unlike many other test parameters, there are no filter “standards” that can be purchased and tested to quantify the accuracy of the fractional efficiency measurement. However, the ability to measure near 0-percent penetration can be evaluated by using a HEPA or ULPA filter and the ability to measure 100-percent penetration can be evaluated by performing the fractional penetration tests with no filter in the device section.
The 100-percent penetration test is a relatively stringent test of the adequacy of the overall duct, sampling, measurement and aerosol generation system. The test is performed as a normal penetration test except that no filter is used. A perfect system would yield a measured penetration of 1 at all particle sizes. Deviations from 1 can occur due to particle losses in the duct, differences in the degree of aerosol uniformity (i.e., mixing) at the upstream and downstream probes, and differences in particle transport efficiency in the upstream and downstream sample lines. The ratio of upstream to downstream concentration as a function of particle diameter is called the correlation ratio. The correlation ratio is used to correct the filter test data for test system effects.
Representativeness of the upstream sample. In most filter tests, the upstream sample is acquired using a single center-of-duct probe located a short distance upstream of the filter. Depending upon the uniformity (i.e., well mixed) of the upstream aerosol, this may or may not be a representative sample. If it is not representative, then the upstream aerosol concentrations will be in error and can lead to an inaccurate efficiency determination.
In order for the center-of-duct probe to acquire a representative sample, the upstream aerosol must be well-mixed. When dealing with particles less than a few micrometers in diameter in relatively small diameter ducts with most of the test air being provided by the aerosol generator, well mixed conditions are not too difficult to achieve. On larger ASHRAE-type test ducts (24-by-24-inch cross section operating at up to 3,000 cfm), means must be taken to ensure mixing of the injected challenge aerosol stream with the test airflow. This is often accomplished with mixing baffles between the aerosol injection point and the test filter. Also, for particle diameters greater than a few micrometers, gravitational settling and inertial effects can lead to a non-uniform distribution across the duct and should be considered when designing the test duct. For example, if relatively low air velocities are involved, a vertically-oriented duct may be needed to prevent gravitational settling of larger-sized aerosol particles.
The uniformity of the challenge aerosol concentration across the duct can be determined by measuring the concentration over, for example, an equal area nine-point sample grid immediately upstream of the test filter.
Representativeness of the downstream sample. The main concern with the downstream sample is to ensure that all aerosol that penetrates the air cleaner (media or frame) is detected by the downstream sampler (i.e., that the downstream sample is representative). For example, when testing a high efficiency bag filter, we want to know that the downstream probe will detect a leak in a corner of one of the bags.
One means of assessing the degree to which the downstream conditions are well-mixed is by installing a high efficiency filter in the test rig. An aerosol is then injected immediately downstream of the filter at preselected injection points located around the perimeter of the test duct and one at the center of the duct. In this test, the point of aerosol injection is traversed and the downstream sampling probe remains stationary in its normal center-of-duct sampling location. The injection points should include points close to the test duct walls. This is necessary because leakage associated with the frames of air cleaners, and bypass flow in electronic air cleaners, occurs in the close-to-the-wall areas. If the downstream duct is well mixed, the downstream concentration will be independent of the injection point.
Sample line losses. One of the more important factors to consider when designing a test apparatus for aerosol filtration testing is minimizing aerosol transport losses through the test rig`s ducting and sample lines. If particle losses are high, it will be difficult to achieve a sufficient concentration of large particles for an accurate test and the differences in the upstream and downstream measurements will, for many low efficiency filters, be dominated by the test duct and sample line particle losses rather than by collection by the air cleaning device being tested.
The design of the sample lines leading from the test duct to the instrument must be considered carefully in order to avoid excessive particle loss. Particle losses associated with the sample lines include inlet losses due to non-isokinetic sampling, losses to the walls of the sample due to diffusion, electrostatic charging, gravitational settling and losses in bends resulting from centrifugal force and eddies.
Losses at the sample inlet may be minimized by sampling isokinetically. Electrostatic effects are avoided by passing the aerosol through an aerosol neutralizer and using conductive and grounded ducting and sample lines. When possible, bends should be avoided and when needed they should have a gradual curvature. Gravitational settling may be reduced by minimizing the effective horizontal length of the sample lines.
The degree of aerosol loss associated with settling, inertial losses and diffusion can be estimated by use of various equations (Fissan and Schwientek, 1987; Hanley et al, 1996). The equations fall into two categories: those applicable to laminar flow and those applicable to turbulent flows. Laminar flow is generally defined as flow having a Reynolds Number (Re) less than about 2,300. At higher Re values, the flow is generally defined as turbulent. While each sampling system is different, it is often beneficial to have the Reynolds number of the flow in the high laminar range.
To reduce the effect of sample line losses on the computed filtration efficiency, it is often beneficial to design the upstream and downstream sample lines to be identical. In this way, the fraction of particles lost in each line will be approximately the same, and thereby “cancel” each other out.
Review of test parameters
In Table 2, important considerations of filter testing are summarized for both low and high efficiency to illustrate differences between the test conditions.
Summary
The successful measurement of the aerosol filtration efficiency of a filter requires careful planning and attention to data quality concerns. In particular, the performance characteristics need to be considered when the test methodology is selected. Obtaining a valid measurement is usually not as straightforward as it first appears. A valid measurement requires consideration of the overall test objective, compatibility of the challenge aerosol with the aerosol instrument and the test filter, designing a test rig and sampling system that minimize particle losses, and conducting control tests to demonstrate the ability of the test rig to provide reliable data. CR
References
For a complete listing of references, please contact the CleanRooms editorial office at (603) 891-9429 or via fax at (603) 891-9200.
Dr. David Ensor is a senior program director at the Research Triangle Institute (Research Triangle Park, NC). He has worked in the aerosol science field for more than 25 years and in contamination control for 16 years. Dr. Ensor is a founding editor of “Aerosol Science and Technology” and a Fellow of the Institute of Environmental Sciences and Technology. He is also a member of the United States Technical Advisory Board to ISO Technical Committee 209 and is the Convenor of Working Group 7:Minienvironments and Isolators.
James T. Hanley is the manager of the Research Triangle Institute`s Aerosol Technology Program. He has worked in the field of aerosol science for more than 15 years. During the last five years, he has led tasks for the EPA and ASHRAE to quantify the fractional filtration efficiency of air cleaners. He led ASHRAE`s recent research project to define a new air cleaner test method and is a member of ASHRAE`s 52.2 Committee to standardize the method.
 |
 |
 |
 |
 |