By Gary L. Hatch, John L. Buteyn and Melissa A. Kinsey
Introduction
Membrane filtration for removal of dissolved salts, fine particles and microbes has been utilized in water treatment for many years. It is relied upon by the electronics and pharmaceutical industries to produce ultrapure water and more recently by many municipalities for purification of drinking water. The point of use (POU) water treatment industry has offered home reverse osmosis (RO) membrane systems since the 1970s; there is the perception by many that RO is the ultimate in water treatment and purification. RO membranes theoretically remove particles down to the molecular size range and therefore are presumed to easily remove all microbes, including viruses, bacteria and protozoan cysts. The filtration spectrum here shows the relative pore size of barrier filtration technologies.
RO membranes are at the far left of the spectrum, having the smallest pores. They are clearly overkill in terms of essential pore size for microbiological purification. That small pore size has a cost in low flux rate of water through the membrane, necessitating pressurized storage tanks. RO’s primary use is in regions where 80-90 percent reduction of dissolved minerals, salts and certain chemical contaminants is desirable and adequate for improving taste. However, when it comes to removing pathogenic viruses and bacteria, essentially ‘absolute’ removal is required to meet microbiological standards. Reported here is documented confirmation that home undersink RO systems do not provide absolute removal of viruses and bacteria and that a new breakthrough, high-flow ultrafilter (UF) can provide complete microbiological purification as a post-membrane filter for home RO systems.
The experiment
Some manufacturers, retailers and consumers may think that RO systems reduce microorganisms to safe levels and that RO is the pinnacle of water treatment. To expose the misconception that home undersink RO systems reliably remove bacteria and viruses, four commercially available units were obtained and tested against several organisms typically utilized to demonstrate mechanical microbiological reduction. The units were challenged with Raoultella terrigena and Brevundimonas diminuta bacteria and MS-2 and fr coliphage viruses (see Table 1).
Each system was comprised of at least four components: a prefilter of carbon block or granulated carbon, the RO membrane, a storage tank of approximately three gallons’ capacity and a postfilter of either carbon block or granulated carbon. From the postfilter, a short length of polymer tubing delivered the output water to the faucet.
The RO systems were connected according to their provided instructions. A tee valve was inserted between the tank and the post RO filter to allow sampling of permeate water directly from the membrane and/or the tank. The tanks and lines were sanitized and thoroughly rinsed before use. Results of rinse water analysis for background test organisms were negative and were negligible for heterotrophic plate count (HPC) bacteria.
The units were then challenged with RO/DI water adjusted (in mg/L) to: total dissolved solids (TDS) ~110, alkalinity ~40, hardness ~35 and turbidity ~0.2 NTU. The test stand was injected with MS-2 or fr phage at a challenge level of at least 1x10E5 PFU/mL (plaque forming units per milliliter) and with B. diminuta bacteria at a challenge level of at least 1x10E7 CFU/100mL (colony forming units per milliliter). The units were run at 60-psi inlet pressure at ~ 20ºC. The systems were continuously challenged through two tank fills. The systems’ tanks required three to five hours to fill. Samples were taken at the end of the first storage tank fill and at the end of the second tank fill. Samples from each system were collected from the faucet (treated with postfilter) and from the tee valve (tank, directly from RO membrane).
Samples for virus analysis were plated in duplicate by agar overlay method with tryptic soy agar (TSA) and E. coli (ATCC #19853 or ATCC #15597) as the host bacteria. Plaques were enumerated after 18-24 hours of incubation at 35°C and reported as PFU/mL. Samples for B. diminuta bacteria analysis were membrane-filtered onto 0.2µm filter membranes on TSA plates, incubated for 48 hours at 35°C and reported as CFU/100mL. Samples for R. terrigena bacteria analysis were membrane-filtered onto 0.45µm filter membranes on m-Endo plates, incubated for 24 hours at 35°C, and reported as CFU/100mL. R. terrigena and B. diminuta colonies were easily discernable and enumerated without interference from HPC.
Three new RO systems from different manufacturers were tested with MS-2 phage and B. diminuta bacteria. System 4 was previously tested against a three-week protocol using the fr phage and R. terrigena bacteria. 
Figure 1 shows these results for System 4 as well as the results from the two-tank fill test using MS-2 and B. diminuta. This protocol included multiple sample points with various challenge water conditions. Results remained relatively unchanged throughout the three-week test. For brevity, only the geometric means of all test results are reported in these Figures. During these tests, TDS rejection for System 4 ranged from 75-92 percent. TDS rejection for Systems 1-3 was not measured but they are presumed to have been operating according to manufacturers’ specifications (75-90 percent rejection).
Undersink ROs don’t reliably remove bacteria and viruses
Figures 1 and 2 show performance of the RO units for removing these organisms relative to the required level for microbiological purification in accordance with the U.S. EPA Guide Standard and Protocol for Microbiological Water Purifiers (1987). This standard is still used as a benchmark for testing and certifying the performance of water purifiers. It requires 6-log removal (99.9999 percent) of bacteria and 4-log removal of viruses (99.99 percent).
Figure 1 shows inconsistent performance among the systems in achieving the 6-log reduction of B. diminuta bacteria, especially within the first two tankfuls (~6 gallons) of produced water. These data suggest that the RO systems provided acceptable virus removal even though some fail to remove the required percentage of challenge bacteria. This is striking because it is counter intuitive on the basis of the particle size difference between bacteria and viruses. Data presented in 
Figure 2 give some insight into this behavior.
Figure 2 shows the results from water sampled from the tank through the tee valve (not passed through the postfilter). This allows evaluation of the quality of water transferred into the accumulator tanks directly from the RO membrane. Comparison of Figure 1 and Figure 2 reveals that the postfilter in System 1 and System 3 contributed very little reduction of bacteria that had passed the RO membrane into the storage tank. This is not surprising since these two systems use granular activated carbon as the postfilter material. Systems 2 and 4 use a molded or extruded carbon block as the postfilter. This contributes over 2-log reduction of bacteria, bringing the overall performance closer to the U.S. EPA 6-log requirement, at least for R. terrigena for System 4.
Figure 2 shows the MS-2 virus performance was generally acceptable for Systems 1-3, while the B. diminuta bacteria reduction was unacceptable. Comparison of the three new systems to the fourth previously tested system suggests that new RO membranes are better at rejecting the hydrophobic MS-2 phage than an acclimated membrane. We speculate that once the membrane has been thoroughly wetted and has been acclimated for three to four weeks, the virus rejection drops to levels comparable to the bacterial rejection. This phenomenon is the subject of further research.
Figure 2 also shows RO System 4 performs poorest of all when sampled from the storage tank. One might speculate that after a certain period of use, an RO membrane element may become in some way more compromised relative to microorganism removal. This potential phenomenon must be taken into account when test protocols are being developed for certification standards. However, Figure 1 showed System 4 to have comparable performance to the new units in the 2-tank fill test, apparently because the postfilter provided significant microbial removal performance. Other tests in our laboratory have confirmed that cyst-rated carbon block filters provide moderate bacterial and virus reduction, but only for a very low service volume (≤ 50 gals).
Summary of RO membrane antimicrobial performance
Although RO membranes have pores far smaller than the dimensions of the microbial contaminants in water, a small percentage of these contaminants pass beyond the system into the drinking water output. The sealing and gluing process for making most spiral wound RO elements and the overall integrity of the membrane are apparently inadequate to provide reliable microorganism removal. There is much debate as to which of these failure mechanisms is more responsible, but some experts believe it is due to contaminants passing directly through minor flaws in the membrane. RO membranes are integrity tested for rejection ratio of TDS, not virus or bacteria. The Purifier status for bacteria is 99.9999 percent removal while the rejection ratio expected for TDS is only about 90 percent.
Solution: New high-flow UF membrane technology
Ultrafiltration products have not gained wide use in POU or point of entry (POE) drinking water filtration because, for direct delivery of flow without a storage tank, they typically require very high membrane surface area and/or high pressure to achieve adequate flow and performance. Achieving and maintaining virus removal at acceptable flow rate and pressure drop has been a difficult task for UF membrane manufacturers. The new breakthrough UF membrane technology reported here has overcome this problem and now allows utilization of a very small hollow fiber UF membrane bundle in many POU applications. We have found that this bundle provides acceptable flow rates at the low to medium pressures typically encountered from RO storage tanks (0.5-0.8 gpm at 20-40 psi).
Figure 3 shows that after replacing the standard post-carbon filter in System 4 with the new high-flow UF post-membrane filter, the bacteria and viruses that have passed through the RO membrane can now be removed. Although System 4 was the poorest performing RO system, incorporating the UF post-RO membrane filter can provide overall Purifier performance.
The results shown in Figure 3 were from a 3-week test involving daily microbial challenges. The tank was sampled daily after overnight and weekend stagnations. The system was operated to provide at least two tank-fills per day.
The product described above was specifically designed as a post filter replacement for RO
System 4. The same performance can be achieved in any RO system by retrofitting with combination carbon/ultrafilter cartridges designed in generic and custom geometries. With substitution of the standard post-membrane carbon filter in any RO system with the new high-flow UF filter cartridge, the bacteria and viruses not removed by the RO membrane can now be removed reliably to meet current or new microbiological reduction standards. With the UF cartridge in place as the postfilter, the RO system can actually deliver the ultimate water quality that has long been implied for RO.
Future microbiological purification opportunities
This new high-flow UF technology can stand alone in applications without RO. A single combination carbon/ultrafilter cartridge can purify water of unknown microbiological safety to Purifier status. A ‘purifier filter’ will cost far less than an RO system, plus it will occupy less space, waste no water and avoid removal of healthful minerals and desirable taste. In many regions, this single filter may displace RO as a drinking water treatment unit.
The geometries available in ultrafilter-based cartridges are virtually unlimited. They can easily be made into compact designs and in-line applications. They can be applied to refrigerators, icemakers, water coolers and coffee makers. Larger throughput models can be used for POE applications; however, periodic flushing to maintain performance and maximize life would be a necessary feature in most applications.
Our lab has demonstrated that a proprietary prefiltration media integral with this high-flow UF technology (patent pending) provides unmatched throughput, reliability, simplicity and long life, plus the ultimate performance to meet current and future microbiological certification standards. This new high-flow ultrafilter membrane technology will be an integral part of many new drinking water microbiological purifiers.
About the authors
Corresponding author Dr. Gary L. Hatch is the Director of Research and Development for Pentair Filtration, Inc. of Sheboygan, Wis. He’s responsible for new technologies development and new materials qualification for use in filter products manufactured by Pentair for PENTEK®, American Plumber®, OmniFilter® and Everpure® branded products. Hatch has a doctorate degree in analytical/inorganic chemistry from Kansas State University. He can be reached at (800) 222-7558 or email: [email protected]
Dr. John L. Buteyn is a research engineer with Pentair Filtration, Inc. He has a doctorate degree in analytical chemistry from the University of Wisconsin – Milwaukee.
Melissa A. Kinsey is a microbiologist with Pentair Filtration, Inc.’s research laboratory. She has a bachelor of science degree in microbiology from the University of Wisconsin – LaCrosse.
About the company
Pentair Filtration, Inc., a division of Pentair Water, manufactures an extensive variety of housings, cartridges and systems for many different applications under the PENTEK® brand name. The Sheboygan, Wis. plant has been producing filter housings and cartridges for nearly 40 years and has its own in-house state-of-the-art testing laboratory, which is capable of testing filters for chemical and microbiological reduction capabilities in accordance with NSF DWTU standards. Visit www.pentekfiltration.com