APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Mar. 1981, p. 646-651 0099-2240/81/030646-06$02.00/0

Vol. 41, No. 3

Effects of Activated Carbon and Bacteriostatic Filters on Microbiological Quality of Drinking Water R. S.

TOBIN,`*

D. K. SMITH,2 AND J. A. LINDSAY2

Health Protection Branch, Health and Welfare Canada, Ottawa, Ontario KIA OL2,' and Ontario Research Foundation, Sheridan Park Research Community, Mississauga, Ontraio L5K IB3,2 Canada

Three activated carbon filters for point-of-use water treatment were tested in laboratory and field studies for chemical removal and microbiological effects on water. All removed free available chlorine in municipally treated water to below the limit of detection, but removed only about 50 to 70% of the total available chlorine and 4 to 33% of the total organic carbon. Standard plate count bacteria in the effluent increased steadily with time for 3 weeks and remained elevated over the 8-week period of the study. Total coliform bacteria were found to persist and proliferate on the filters for several days after transient contamination of the influent water. Silver-containing activated carbon filters suppressed total coliform but not total bacterial growth. Pseudomonas aeruginosa was recovered from the effluents of all filters at some time during the tests.

Increased awareness among the public of the health implications of drinking water quality and the continuing demand for aesthetically pleasing water have resulted in a rapid proliferation in the types and numbers of point-of-use water treatment devices sold (24). A major proportion of all such devices are activated carbon filters, including those releasing silver to act as a bacteriostat. Some of these are claimed to remove taste and odor, sediment, chlorine, trihalamethanes, carcinogens, pesticides, and a number of other perceived problem substances. Wallis and co-workers (27) were among the first to demonstrate the tremendous proliferation of bacteria in such devices, which they considered to be a potential health hazard. In a later study, Fiore and Babineau (11) also showed high bacterial counts in the effluent after periods of stagnation, but maintained that the proliferation observed was not significantly greater than that observed in untreated tap water after the same period. Taylor et al. (24) clearly demonstrated plate counts from activated carbon filters to be greater than those from corresponding influent water, but levels were lower than some previously reported. The present study investigates the bacterial growth on four types of activated carbon filters, including a silver-releasing bacteriostatic filter. The growth of total coliform bacteria and of bacteria of potential health significance on the filters is shown. (This paper was presented in part at the Annual Meeting of the American Society for Microbiology, May 1980, in Miami Beach, Fla.).

MATERLALS AND METHODS Filters. Filter A is a compact in-line filter that is sold principally for the recreational vehicle, boat, and vacation home market. The cartridge consists of a carbon core sandwiched between thin outer and inner layers of cellulose. It is designed to remove taste, odor, and sediment. Filter B is a slightly larger unit, intended for in-line attachment to cold water lines in homes. The cartridge consists of a resin-bonded cellulose fiber filter section at the outlet, an activated carbon core, and another resin-bonded cellulose network at the outlet. It is designed to remove objectionable taste and odor, as well as dirt, rust, and sand. Unit C is a small unit that attaches to a kitchen faucet. It was tested with two types of cartridge that are currently available. One contained activated carbon, and the other contained activated carbon plus silver. The silver released from the latter did not exceed 50 ,ug/liter in the effluent. Unit C is claimed to reduce chlorine, algae, suspended particles greater than 8 u, and organic chemicals such as detergents, pesticides, chloroform, and polychlorinated biphenyls. Testing of devices. Two of each of the three devices were tested in the laboratory on a manifold system described in detail elsewhere (9). Timers were adjusted to give an "on" time of 30 s every 0.5 h, and the flow rates on filters A and B were adjusted by throttle valves to 7.6 liters/min (high flow rate) and 4.25 liters/min (low flow rate). Control counts were taken from identically valved lines which were set at the same flow rates and timing as the filters. The manifold was operated for 16 h per day and left inactive for 8 h per day. Total volumes processed were approximately 123 liters per day at the fast flow rate and 68 liters per day at the slow flow rate. Filter C units were attached to faucets on the manifold, and both were adjusted to a flow rate of 475 ml/min to give a total of 7.6 liters per day. The portion of the

646

VOL. 41, 1981

manifold to which filter C devices were attached had a bleed line which was open during the "on" cycle to ensure that this portion of the manifold did not become stagnant under the low flow conditions. This bleed line would also simulate the use of the water bypass function of filter C for nonconsumptive purposes in home use. When a coliform challenge was used, a settled sewage sample was collected and diluted in a tank to provide a challenge of 70 total coliform bacteria per 100 ml. When a challenge of bacteria was required to determine the bactericidal efficacy of silver-containing units, pigmented bacteria derived from the effluents of colonized activated carbon filters were grown in nutrient broth for 24 h at 280C. The bacteria were diluted to about 1,780 colonies per ml in tap water left to stand for 24 h in open tanks (no detectable free available chlorine residual) and used to dose the water to the test filters. Another filter C unit was placed in a home supplied by the same raw water supply but a different treatment plant from that used in the laboratory tests. It was used about two to five times a day with volumes of 0.3 to 3 liters withdrawn at a time, for an average daily use of 6 liters. Sampling. Volumes of 200 ml (filters A and B) or 100 ml (filter C) were collected in sterile flasks containing 0.1 ml of 10% (wt/vol) sterile sodium thiosulfate to neutralize any residual chlorine. When the silver-carbon flter was used, samples were collected in flasks containing 1 ml of neutralizer solution (6) (5% sodium thioglycolate plus 7.3% sodium thiosulfate) per 100 ml of sample. At least three sets of samples were taken: the initial flush after overnight stagnation, after 30 s of flow through the units, and after 4 h of operation of the daily cycle. Influent samples were taken at the same times. Bacterial enumeration and identification. Plates for standard plate counts (1) were incubated at 350C for 96 h except where noted. Total coliform counts were performed on 100-ml samples using the membrane filtration technique with incubation on mEndo LES agar (BBL) at 370C for 48 h. Pseudomonas

ACTIVATED CARBON FILTERS

647

aeruginosa was isolated by the membrane filtration technique with incubation at 350C on Pseudomonas isolation agar (Difco Laboratories), and identification was performed by means of the Minitek system (BBL Microbiology Systems) and by the scheme described by Shayegani et al. (23). Chemical analyses. Free available chlorine and total available chlorine were measured by the N,Ndiethyl-p-phenylenediamine method (1) and a Hellige color comparator disk. Total organic carbon analyses were performed with an automatic TOC analyzer (Beckman Instrument Co.).

RESULTS The ability of the carbon filters to remove the residual chlorine and total organic carbon from municipally treated tap water was measured (Table 1). All filters reduced the free available chlorine content from 0.20 mg/liter to below the limit of detection (0.05 mg/liter), although 'k to ½ of the total available chlorine remained in the effluent. In units A and B, the units run at slower flow rates were more effective in removing total available chlorine than those run at the faster flow rates. Total organic carbon removal ranged between 4 and 33% in a typical experiment (Table 1) and was not consistently related to the flow rate through the units. Filter B was clearly superior to the others in total organic carbon removal, although at least two-thirds of the material remained in the effluent. Total organic carbon removal did not vary appreciably during the period of testing these devices. The microbiological quality of the water from the filters was closely monitored during the 55day test. Standard plate counts (3500) were read at 96 h to give optimum colony size and numbers (12). Up to day 10 of the test, filters A and B removed bacteria from the water to almost undetectable levels (Fig. 1). Filter C (not shown) TABLE 1. Removal of available chlorine and total began to show appreciable colonization by day organic carbon by activated carbon filters 6. By day 25, the colonization of the filters to have stabilized in both the initial appeared Chlorine reFlow Total organic carbon after overnight stagnation and after 4 flush (0 s) rate (l- sidualtr(mg/li(mg/liter) at: Umt Unt ters h of the regular daily cycle. The geometric per day) 30 s FACa TACb means of days 25 to 55 (Table 2) indicate a high 0s degree of bacterial contamination of effluent Influent 0.20 0.30 7.3 4.5 water, particularly in the first flush after over68 NDC 0.10 6.5 (11)d 4.0 (11) A night stagnation. Whereas counts decreased 123 ND 0.15 5.5 (25) 4.3 (4) A ND 68 B 0.10 5.0 (32) 3.0 (33) markedly by 30 s and were even lower after 4 h ND B 123 0.15 5.5 (25) 3.0 (33) of cycling, they were, in all cases, considerably C 7.6 ND 0.10 6.0 (18) 3.8 (16) higher than the influent values. Two filter C devices were installed in the 7.6 ND 0.10 6.0 (18) 4.0 (11) C kitchen in homes where they were used to proa FAC, Free available chlorine. b vide water for household use. One unit, operatTAC, Total available chlorine. C ND, Not detectable (