Sterilizing Filters: Right Flow, Right Size Critical

Sept. 9, 2005
Filtration can be the quickest, most cost-effective means to achieve sterilization for large volumes of simple buffers or aqueous solutions. However, finding the right flow rate and filter sizing is essential to meeting FDA’s aseptic guidelines.
By Jennifer Maynard, Six Sigma Black Belt, Baxter BioScience; Theodore H. Meltzer, Capitola Consultancy; Maik W. Jornitz, vice president, Product Management, Sartorius North America, Inc.; and Paul M. Priebe, head of Product Management, Process Filtration, Sartorius North America, Inc.Sterilizing grade, 0.2-micron-rated membrane filters are used widely within the biopharmaceutical industry. They were also given more prominence by FDA’s Aseptic Guidelines, which specified filtrative sterilization as a key criterion for process validation ( used in cases where thermal sterilization would degrade product, the filters have been redesigned and optimized to improve flow. They are now recognized as the quickest, most cost-effective means to achieve sterilization for large volumes of simple buffers or aqueous solutions.Since such fluids typically contain only very low levels of contaminants, a sterilizing filter’s throughput is far less important than its speed in transferring fluid from a mixing vessel to either holding vessels or disposable bag containment. The longer the transfer takes, the longer equipment will be idle, directly affecting the manufacturing facility’s capacity.Consider, for example, a common 0.2-micron-rated sterilizing-grade filter, which commonly achieves a flow rate of 2,500 L/hour/14.5 psi. It would take this filter 48 minutes to process a 2,000-L volume. In contrast, an optimized high-flow-rate filter with a flow rate of 6,000 L/hour/14.5 psi would require only 20 minutes, doubling equipment availability. In cases where that speed would be too high, it could be adjusted by reducing the filter’s effective filtration area (EFA).Where, in the past, sterilizing-grade filters were designed to be used broadly for a multitude of applications and fluids, they are now customized for specific applications, some of which require a high flow rate through the filter at low differential pressure. This article will discuss the importance of flow within these applications, focusing on how best to test the filters for flow and how to size them accordingly.Improved designsRedesigning the filters’ membranes and cartridges allowed them to achieve higher flow rates. Otherwise, the only way to improve flow rate would have been to raise the differential pressure or increase surface area, neither of which was practical: higher differential pressures would mean a significant increase in energy consumption or could risk exceeding the filter’s operating pressure capacity. Larger filtration surfaces would mean an increase in consumable and capital investment costs.
Figure 1. Membrane breaks at the pleat edges.

MembranesNew filter membranes feature a high pore volume and are most often asymmetric and highly pleatable. The membrane polymer selected determines many of the membrane’s performance parameters, but other variables are involved as well. For example, if the membrane is not pleatable, the filter’s EFA will be low, and the pleat edges won’t be strong enough to withstand pressure pulsation (Figure 1, right).High flow applications can create water hammer or excessive pressure pulsation. As a consequence, these membranes and filters require high mechanical stability (Figure 2, below). In other instances, the filters undergo multiple steam or sanitization cycles, dictating thermal stability. If the membrane casting does not allow for pleatability, the filter will be damaged during the filtration or sterilization process and jeopardize the sterility of the filtered fluid.Cartridges and fleecesFilter cartridge design is also a key element to optimizing flow rate. A single-layer membrane construction will usually achieve higher flow rates than a membrane double-layer combination, especially when this combination is a homogenous (e.g., 0.2/0.2 micron) configuration. The flow restriction of a homogenous double-layer design can be so high that a single-layer membrane filter of a smaller pore size, e.g. 0.1-micron-rated, might reach a comparable flow rate (Figure 3, below ).Furthermore, the support fleece and pleat densities must be well balanced in order to avoid a limited EFA or uneven flow distribution within the membrane pleat pack. A support fleece with a small fiber diameter and high density might be so compact that it could end up resisting flow, eventually causing flow to drop due to a decrease in pressure over the fleece material. Optimizing filter flow rate requires thorough study, and balancing all of the parameters above.Putting filters to the testDespite the importance of filter flow, pharmaceutical manufacturers tend to take it for granted and to rely on data provided by filter manufacturers, instead of testing the filter within its future application. And even if manufacturers do their own testing, they may use inappropriate filter samples — for example, samples of the membrane alone, rather than a composite sample representative of the entire filter cartridge construction.Another common mistake is using 47-mm discs for testing. Tests for flow rate using 47-mm discs are meaningless, since they can evaluate only the porosity and thickness of the membrane itself. Critical and beneficial parameters of the true filter element, which will be used within the process, are not evaluated. Side-by-side trials employing 47-mm discs cannot determine the true flow rate performance of the filter within the production process.Figure 4 depicts true flow rate measurements using 10-inch filter elements and tests performed with 47-mm discs. The tests have been performed at a set differential pressure of 0.5 bar with water at 20°C. The flow rate results of the 10-in. elements and 47-mm discs were calculated to L/m²/hour to compare the flow rates of the disc and the true flow of process elements.As Figure 4 shows, the 47-mm flow rate results differ greatly from those of the 10-inch element flows — i.e., 47-mm test discs are inappropriate for use in determining an appropriate filter type and scale. Such tests are only time consuming and are not of true value. Only large or real facility scale trials can determine the best flow rate filter. More likely than not, the effect of the flow restrictions not related to the membrane will become more pronounced in larger filter formats. The scaleup becomes less and less reliable from a scale of 1-by-10-in. to 1-by-30-in., and further still to multiple 30-in. filters. This is exceptionally important, as selection trials for flow rate are typically employed for very large volumes beyond 20,000 liters, where multiple 30-in. filters will be used to achieve acceptable processing times.Apples to applesOptimizing filtration processes requires tests using comparable filter elements, commonly 10-in. filter cartridges. A good side-by-side trial can only be performed by using such comparable elements, in order to take into account the entire design of the filter, filter housing and membrane design, as well as the effective filtration area, flow distribution due to pleat densities, and fleece thickness.Tests must performed under the required or specified process conditions, such as using a set inlet pressure and temperature while measuring the time to filter the fixed fluid volume. Such a test setup can work for different filter types.However, it is important to remember that process parameters must remain constant. For example, buffer composition, pressure and temperature settings must be the same to achieve meaningful results that will help in determining the optimal filter type or combination of types.Water can be used as a test medium, for comparing flow rate performances of multiple, commercially available filters. But, in order to size the filter system correctly, the actual product has to be used under process conditions with a full-scale pleated filter element. Only these test and equipment conditions can be validated and used in production.The perfect filter?There is no such thing as the perfect membrane filter for every application. In the past, membrane filters were used universally. This has changed as different applications require very specific filtration properties and filter designs. One of the properties is flow and the need for high flow rates within specific applications like buffer filtration. Experimental trials must be conducted to establish the high flow filter or filter combination for the particular application and process.Such tests and test procedures require purposeful design and consideration; otherwise, the trial results might lead to improper filter selection, resulting in increased downtime or even contaminated batches. While filter manufacturers offer technical service to support users in determining the experimental design, performing the tests, and writing the appropriate test reports, in-house staff at a facility can and should be trained to design and run the experiments as well.Figure 2. Pulsation resistance of different high-flow filter cartridges using polyethersulfone at 5 bar differential pressure (*double-layer filter).

Figure 3. Flow rate comparison of different 10-inch membrane filter cartridge configurations.

Figure 4. Flow rate comparisons of 47-mm discs and 10-in. elements (extrapolated to one m2) and actual 10-in. cartridge flow (*double-layer filter), all with polyethersulfones.

About the AuthorsJennifer Maynard is a Six Sigma Black Belt for Baxter BioScience in Thousand Oaks, Calif., currently working within purchasing supplier management. She has extensive knowledge in supply chain management and process improvement within cell culture manufacturing.Theodore H. Meltzer, Ph.D., is a private consultant with Capitola Consultancy in Bethesda, Md., with over 40 years experience in the separation technology and ultrapure water industry. He is the author and co-author of multiple books, book chapters and over 100 scientific papers.Paul M. Priebe is head of product management — process filtration, for Sartorius North America (Edgewood, N.Y.) He has more than 8 years experience in process filtration. He has previously held positions in manufacturing, engineering, sales and product management.Maik W. Jornitz is VP of product management at Sartorius. Jornitz supports the biopharmaceutical industry with close to 20 years of experience in separation technologies. He is author and co-author of multiple books, book chapters and scientific publications.References1. ASTM Committee F-21. Standard Test Methods for determining Retention of Membrane Filters Utilized for Liquid Filtration. Annual Book ASTM Stand., Philadelphia, Pa., 1998; 790–795.2. FDA, CDER. Guidance for Industry, Changes to an Approved NDA or ANDA, CMC, 1999.3. HIMA. Microbial Evaluation of Filters for Sterilizing Liquids.Health Industry Manufacturers Association, Washington, D.C., 1982; Document No. 3, Vol. 4.4. Jornitz, M.W.; Meltzer, T.H. Flow and Pressure, Flow Decay, and Filter Sizing, and Cartridges, Cartridge Holders, and Their Care. In Sterile Filtration – A Practical Approach, Jornitz, M.W.; Meltzer, T.H., Eds; Marcel Dekker, New York, N.Y., 2001; Chapters 4 and 8.5. Levy, R.V. Sterile Filtration of Liquids and Gasses. In Disinfection, Sterilization, and Preservation; Block, S.S., Ed.; Lippincott Williams & Wilkins, Philadelphia, Pa., 2001; Chapter 40.6. PDA. Sterilizing Filtration of Liquids, Technical Report #26, PDA J. Pharm. Sci. Technol. Suppl. 1998; Vol. 52, No. S1.7. Soelkner. P.; Rupp. J. Cartridge Filters. In Filtration on the Biopharmaceutical Industry. Meltzer, T.H., Jornitz, M.W., Eds. Marcel Dekker, New York, N.Y., 1998.8. Priebe, P.M.; Jornitz, M.W.; Meltzer, T.H. Making an Informed Membrane Filter Choice – Criteria to Consider. Bioprocess Int. 2003, 10, 64–66.9. Jornitz, M.W.; Soelkner, P.G.; Meltzer, T.H. The Economics of Modern Sterile Filtration. Pharm. Technol. U.S. 2003, 27(3), 156–166.10. Jornitz, M.W.; Meltzer, T.H., Bromm, H. and Priebe, P.M. Choosing the Appropriate Membrane Filter - Test Requirements, PDA Journal, Vol. 59, No. 2, 2005, 96 - 102