Environ. Sci. Technol. 2003, 37, 3104-3110
Bromate in Chlorinated Drinking Waters: Occurrence and Implications for Future Regulation HOWARD S. WEINBERG,* CARRIE A. DELCOMYN,‡ AND VASU UNNAM§ Department of Environmental Sciences and Engineering, University of North Carolina at Chapel Hill, Rosenau Hall, Chapel Hill, North Carolina 27599-7431
Bromate is a contaminant of commercially produced solutions of sodium hypochlorite used for disinfection of drinking water. However, no methodical approach has been carried out in U.S. drinking waters to determine the impact of such contamination on drinking water quality. This study utilized a recently developed method for quantitation of bromate down to 0.05 µg/L to determine the concentration of bromate present in finished waters that had been chlorinated using hypochlorite. Forty treatment plants throughout the United States using hypochlorite in the disinfection step were selected and the levels of bromate in the water both prior to and following the addition of hypochlorite were measured. The levels of bromate in the hypochlorite feedstock were also measured and together with the dosage information provided by the plants and the amount of free chlorine in the feedstock, it was possible to calculate the theoretical level of bromate that would be imparted to the water. A mass balance was performed to compare the level of bromate in finished drinking water samples to that found in the corresponding hypochlorite solution used for treatment. Additional confirmation of the source of elevated bromate levels was provided by monitoring for an increase in the level of chlorate, a cocontaminant of hypochlorite, at the same point in the treatment plant where bromate was elevated. This study showed that bromate in hypochlorite-treated finished waters varies across the United States based on the source of the chemical feedstock, which can add as much as 3 µg/L bromate into drinking water. Although this is within the current negotiated industry standard that allows up to 50% of the maximum contaminant level (MCL) for bromate in drinking water to be contributed by hypochlorite, it would be a challenge to meet a tighter standard. Given that distribution costs encourage utilities to purchase chemical feedstocks from local suppliers, utilities in certain regions of the United States may be put at a distinct disadvantage if future lower regulations on bromate levels in finished drinking water are put into place. Moreover, with these contaminant levels it would be almost impossible to lower the maximum permissible contribution to bromate in finished water from hypochlorite to 10% of the MCL, which * Corresponding author telephone: (919)966-3859; fax: (919)9667911; e-mail:
[email protected]. ‡ Present address: Applied Research Associates, Inc., 139 Barnes Dr., Tyndall AFB, FL 32403. § Present address: Santa Clara Valley Water District, 5750 Almaden Expressway, San Jose, CA 95118-3614. 3104
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 14, 2003
is the norm for other treatment chemicals. Until this issue is resolved, it will be difficult to justify a lowering of the bromate MCL from its current level of 10 to 5 µg/L or lower.
Introduction Drinking water disinfected with chlorine produces a number of undesirable and potentially cancer-causing organic disinfection byproducts (DBPs) such as trihalomethanes and haloacetic acids. To minimize the production of these compounds, water treatment facilities often employ ozone followed by chloramination for disinfection. Ozone is a powerful oxidant but unstable in water, requiring a terminal disinfectant such as chloramine for the purpose of maintaining a disinfectant residual in water distribution systems. However, bromate (BrO3-) is a major DBP generated from bromide-containing source waters that undergo ozonation. It is a category I group B2 carcinogen (1) and is currently regulated in treated drinking water under the Stage 1 D/DBP Rule of the Safe Drinking Water Act (2) at a maximum contaminant level (MCL) of 10 µg/L. This MCL in drinking water was established based on an estimated 10-4 excess lifetime cancer risk from exposure to bromate in drinking water at 5 µg/L. A linear extrapolation down to a 10-6 cancer risk level is a longer term goal (3) but would require a practical measurement of bromate at 0.05 µg/L. A treatment plant using sufficient ozone to inactivate Cryptosporidium in a bromide-containing water would generate bromate that in all likelihood would exceed the lower MCL, and there is currently no economically feasible technology to remove bromate once formed. Moreover, it would have been premature to lower the MCL while accurate and reliable analytical methods were not readily accessible to measure below 10 µg/L. Recent developments now permit quantitation at 5 µg/L and below, but discoveries of bromate in chlorinated drinking waters that had not been treated with ozone (4) raised concerns that another source of this carcinogen existed. Most small utilities (serving less than 10 000 consumers) and many large utilities utilize liquid sodium hypochlorite or solid calcium hypochlorite for drinking water disinfection. These chemicals can be handled without a high level of expertise and with minimal hazard as compared to the alternative, gaseous chlorine. Indeed, many larger utilities are switching from gas to liquid for both safety and regulatory reasons and use doses in the range 1-20 mg/L as chlorine (5). Even plants that use chloramine for final disinfection are likely to use up to 5 mg/L hypochlorite in its production (6). For a short time, it was even considered a possibility that chloramine could generate bromate in finished waters through reaction with bromide (7), but this hypothesis was quickly dispelled (8). Another study (4) suggested that chlorine could generate bromate from reaction with bromide in waters in the pH range of 6-7 but that this mechanism was only significant at temperatures exceeding 30 °C. The discovery of bromate in hypochlorite-treated drinking waters led The Chlorine Institute to evaluate the source and fate of bromide in the chlor-alkali and bleach production processes (9). The levels of bromate contamination were shown to vary according to manufacturer and were attributed to different electrolysis cells and sources of salt used for manufacturing hypochlorite solutions. Bromide naturally coexists with chloride in salt and may undergo oxidation to bromate in the manufacturing process of hypochlorite feedstock solutions (9): 10.1021/es026400z CCC: $25.00
2003 American Chemical Society Published on Web 06/17/2003
Br- + 3OCl- f BrO3- + 3ClBecause different types of electrolysis cells are used in the manufacturing of hypochlorite, the level of bromide partitioning from each chemical component in the process may vary. Use of diaphragm cells in the manufacturing process has shown 80% bromide partitioning in the caustic soda versus 20% in chlorine. Moreover, it was demonstrated that all bromide entering the process, whether from chlorine or caustic soda, was completely oxidized to bromate, which was first discovered in hypochlorite solutions in the range of 99%) and sodium chlorite (∼80%) were from Fluka (Ronkonkoma, NY), and potassium iodate (certified ACS grade) was from Fisher Scientific (Fair Lawn, NJ). Stock solutions of these target anions were prepared at 1 g/L in deionized (DI) water, stored at 4 °C in glass amber bottles sealed with Teflon septa, and replaced every 6 months. DI water for preparation of stock solutions of standards, dilutions
of feedstocks, and preparation of mobile phases was supplied from a mixed-bed ion-exchange resin and carbon polisher (Virginia Water Systems, Inc., Richmond, VA). Reagents for the ion chromatography mobile phase (sodium carbonate), regenerant (sulfuric acid), and postcolumn reaction (sodium bromide and sodium nitrite) were supplied as ACS reagent grade from Fisher Scientific (Fair Lawn, NJ). A standard solution of 10-13% free available chlorine was supplied by Sigma Chemical Company (St Louis, MO). Analytical Methods. Free chlorine was measured by iodometric titration using Standard Method 4500-Cl B (15) on a 1:10 dilution in DI water of the supplied feedstock. The ion chromatographic (IC) method employed for the measurement of bromate in both hypochlorite feedstocks and finished waters used a post-column reaction (PCR) that generates stoichiometric amounts of tribromide proportional to the concentration of target oxyhalide, which is sensitively detected at 267 nm by UV (16). The filtered 9.0 mM Na2CO3 mobile phase was sparged prior to use before being placed under pressure with ultrahigh purity (UHP) helium (Holox, Norcross, GA). 25 mM H2SO4 was used as the regenerant in this process. For the PCR, 0.75 M H2SO4 and 0.145 mM NaNO2 were employed to generate in-situ nitrous acid, which together with 2M NaBr generated the tribromide ion in the presence of the eluting oxyhalide. The IC (Dionex, Sunnyvale, CA) was equipped with the following modules: AI-450 v.3.32 and PeakNet v.4.30 computer software with an RS232 advanced computer interface, ASM-3 automated sampler, GPM-2 gradient pump, LCM-3 conductivity detector, a VDM-2 variable wavelength detector, and an Eluant Degas module. The equipment used for the PCR setup included a quick pump (DQP-1), post-column pneumatic controller, and column heater (CH-1). The mobile phase passing through the selected analytical column (AS9HC) entered first through the conductivity detector (used for bromide, chlorate, and chloride detection) and then joined with the PCR mixture before entering the UV detector (for bromate, chlorite, and iodate detection). Sample injection volume was 150 µL, which was supplied through a six-port pressure-operated slider valve. The practical quantitation limit for bromate using this procedure is 0.05 µg/L (17). Quality control criteria applied with this method included the average of triplicate analyses having a coefficient of variation less than 10% and spiked recoveries of the target analyte at approximately two times its concentration in the matrix in the range of 85-110%. Field Samples. Preliminary operating and analytical data were gathered from each utility participating in the study, and a schematic of the treatment process was supplied. Sample points were then selected that captured chlorate and bromate levels in the treatment train before and after addition of hypochlorite. Appropriately labeled clean acid-washed 40mL VOA vials (Laboratory Supply Distributors, Mt. Laurel, NJ) were prepared for drinking water sample collection. Each sample vial contained 20 µL of a 100 mg/mL ethylenediamine (EDA) solution to quench residual chlorine and to stabilize the oxyhalides in samples prior to analysis (18). The vials were then secured with packing material and dry ice packs in a cooler for shipment to the corresponding facility. Travel and field blanks as well as sealed empty vials for collection of a hypochlorite feedstock sample were also included in each shipment. An information and instruction sheet was sent along in the cooler with the vials to collect relevant plant operation data. Once the samples were received back in the UNC laboratory, they were placed in a 4 °C refrigerator until analysis could be performed (within 14 d). Both hypochlorite feedstocks and drinking water samples were collected in duplicate 40-mL VOA vials with zero headspace from each drinking water treatment facility that participated in the sampling study. VOL. 37, NO. 14, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3105
FIGURE 1. Ion chromatograms for analysis of a 1:10 000 dilution of sodium hypochlorite (10-13% available chlorine): (A) by conductivity (13 mg/L chloride and 3.2 mg/L chlorate; (B) by PC-UV (14 µg/L chlorite, 15.4 µg/L bromate, and 3.2 mg/L chlorate).
TABLE 1. Oxyhalide Analysis through an Ozone-Chloramine Treatment Plant
TABLE 2. Oxyhalide Analysis in an Ozone-Chlorine Plant concentration (µg/L)
concentration (µg/L) sample location
raw water
