Issue



IEST-RP-CC034.2: HEPA- and ULPA-filter leak tests


03/01/2006







R. Vijayakumar, Ph.D., Director of Market Development at TSI Incorporated, and Chair of the IEST Working Group CC034, HEPA and ULPA Filter Leak Tests

The performance of any cleanroom and its ability to achieve and maintain the designed cleanliness class are critically dependent on the performance of the air filters used in its construction. Good practice dictates testing the performance of all filters for their overall efficiency, integrity, or absence of leaks. Last month’s column, “IEST-RP-CC001.4 HEPA and ULPA filters,” discussed filter classifications and their respective test methods. The article mentioned that filters are typically tested both as manufactured and after installation. In addition, many cleanrooms in regulated industries such as the pharmaceutical or nuclear facilities require these filters to be certified periodically as installed to ensure acceptable performance during their service life.

This recommended practice (RP) provides guidance for filter integrity testing, both with automated test equipment at the manufacturing facility and for manual testing as installed, thereby ensuring that the filters meet design specifications.

History

Leak-testing is relevant to individual filters as well as to installations such as cleanrooms, biosafety cabinets, and lab benches. Recommended procedures for leak-testing were traditionally part of various IEST RPs dealing with filters or their use. These RPs were revised at different times, resulting in different recommended methods for leak-testing filters. In an attempt to eliminate confusion arising from differing recommended leak-testing practices, one leak-testing practice common to all relevant RPs was published in the mid-1990s.

Procedures for leak- or integrity-testing of HEPA and ULPA filters with both photometers and discrete particle counters are provided in this RP. In fact, many regulatory and standards-setting bodies have incorporated elements of IEST-RP-CC034 into their efforts. Although the procedures in the RP may be independently followed, the intention is for them to be used as a complement to other IEST RPs dealing with filters and filter applications.

What’s new (and improved)

Expanded guidance on the selection of leak-test methods, based on filter performance and application, is provided. All the new filter types defined in IEST-RP-CC001.4 (HEPA and ULPA Filters) are covered. The RP includes suggested choice of photometers or discrete particle counters for the tests. (See Table 1 for guideline information.)


Table 1: Guideline for selecting leak-test methods
Click here to enlarge image

As with the IEST-RP-CC001.4 discussed last month, the difference between count and mass mean diameter of aerosols is clearly defined. The count mean diameter (CMD) is the average particle size of the number distribution of the aerosol. The mass mean diameter (MMD) of the aerosol is the average particle size of the mass distribution of the aerosol. Typically, since the mass of a particle varies with the cube of its diameter, most of the mass of an aerosol tends to be in the larger sizes, resulting in the mass mean being larger than its count or number mean. Understanding the difference is significant, since filter efficiency is very size-dependent.

Leak-testing requires a uniform distribution of the challenge aerosol to ensure that local variations do not adversely affect the determination of a leak in the filter. Calculation of the spatial and temporal uniformity is addressed in the RP with illustrations. Example calculations are also provided to help in estimating uniformity. The nominal acceptable criteria for the uniformity are as follows:

  • Relative standard deviation less than 20 percent.
  • Maximum relative deviation of any single point (50 percent).
  • Ratio of concentration at the representative upstream sample port to average concentration between 0.75 and 1.25.

Dead-air spaces (areas of no airflow) pose a significant concern in leak-testing filters and their surrounding support structure. This subject is addressed in some detail in the RP. The testing of these areas is necessary to detect air bypass or leaks in the filter frame, gasket, or seals, and the filter support structure. The problem arises due to the lack of HEPA-filtered dilution air in the dead-air space. A small leak will expand throughout the dead-air space and, over time, the aerosol concentration will appear to be much greater than its actual percentage of penetration. This buildup will prevent the tester from isolating the actual source of the leak unless additional actions are taken. One action is to redirect HEPA-filtered air into the dead-air spaces using clipboards, metal plates, laminated sheets, and so forth. The HEPA-filtered air will wash out the aerosol buildup and allow the tester to find the leak source.

In recent years, there have been an increasing number of cases of originally tested and certified HEPA filters that fail during routine leak-testing and recertification in the field. In some cases, the photometer used in field leak-testing results in a continuous high reading, indicating failure. This phenomenon is referred to as bleed-through, or excessive widespread nonsite-specific penetration in leak-testing.

Leak-testing by scanning the filter measures the local performance of the filter at the current position of the scanning head. A typical HEPA filter with a face area of 4 ft.2 is constructed with over a 100 ft.2 of media. In comparison, the cross section of a scan head is about 0.01 ft.2 and samples approximately 0.25 ft.2 of media in the filter. Thus, when you scan the filter, the local performance as well as the performance over a defect is measured. Because the penetration of any filter media used in HEPA filters varies within allowable limits, scanning the face of the filter will measure this variability in the media.

In contrast, the overall efficiency of the filter measures the average or nominal penetration of the media or the filter. Bleed-through is typically believed to occur when filters are scanned for leakage with the criteria of a leak set too close to the overall performance of the filter or tested at the same mean particle size, or both. This is the allowable variability in the media, which results in the filter exceeding the leak criteria at numerous locations during scanning and, consequently, in the perceived bleed-through or excessive widespread nonsite-specific penetration.

This issue has especially become critical when the Type C filter, as defined by IEST-RP-CC001.4, is not used. That filter has a minimum efficiency of 99.99 percent when tested with a thermally generated 0.3 μm MMD (CMD < 0.2 μm) aerosol. Its leak criteria is also 99.99% (0.01% penetration), but it is recommended that the leak test be conducted using an aerosol with a larger particle size. Currently, filter manufacturers, because of market and business reasons, supply filters that will not strictly meet the Type C requirements. These filters are usually tested and classified to different test standards or use test practices modified from those originally included when earlier versions of IEST RPs were established.

Further, currently available devices for generating aerosols with larger particle sizes-typically Laskin Nozzle generators-produce aerosols in amounts sufficient for testing only small cleanrooms or biosafety cabinets. For larger cleanrooms, it is a common field-testing practice to use the vapor condensation (thermally-generated) aerosols with the mean size of particles closer to the MPPS of filters. These two current practices often result in filters tested for efficiency at particle sizes that are close to those used for leak-testing in the field. This leads to a leak criteria almost identical to that of the nominal filter efficiency and results in bleed-through. This problem is expected to be somewhat alleviated by inclusion of alternate filter classes in the newly published IEST-RP-CC001.4, which was discussed in last month’s column.

If a traditional Type C filter is not available, or if sufficient amounts of large particles cannot be generated, the occurrence of bleed-through may be minimized by increasing the local performance of the media-that is, the noise from the allowable local variability in the media is reduced, allowing only genuine defects in the filter to be detected. This can be achieved by choosing a filter with a higher nominal efficiency such that, even at the limits of the allowable variability, the local performance of the media is well within the leak criteria and, therefore, not detected as bleed-through.

Since leak-testing is a key subject to many in the contamination-control industry, IEST welcomes input and suggestions on handling this vexing problem. Or, better yet, come join the Working Group (WG) deliberations, and add your insights into this and other key topics addressed by the many RPs published by the IEST.

Acknowledgements:

Without the input, the urgency, and the lively debate among the members of the Working Group, neither the timeliness nor the quality of this RP would be possible. As chair of the WG, I thank them all for their support. Jim Wagner’s quick review of this article is also much appreciated.

Dr. Vijayakumar holds a Ph.D. in particle technology from the University of Minnesota and has over 20 years of experience in filtration, filter testing, and air quality. He has held senior technical and marketing positions in leading filter, filter media, and particle instrument companies, where he has assisted customers worldwide with their filtration and filter-testing problems. He is an instructor at the University of Minnesota’s Short Course on Filtration, and has also lectured worldwide on the subject. Currently, he is the director of market development at TSI Incorporated and is responsible for developing new markets and applications for the company’s test instruments. Dr. Vijayakumar chairs IEST’s working groups 001 and 034, and is acting chair of several other working groups. He also held the position of Chair of Standards and Practices from 1997 to 2001. He can be reached at vijay@tsi.com.