Perchlorate in Sodium Hypochlorite

emerging issues

BY PETER GREINER, CLIF MCLELLAN,

DALE BENNETT, AND ANGIE EWING

Occurrence of perchlorate in sodium hypochlorite

RESPONDING TO THE DETECTION erchlorate is both a synthetic and a naturally occurring chemical. Most of the perchlorate that is manufactured in the United States OF PERCHLORATE IN SODIUM is used as the primary ingredient of solid rocket propellant. Wastes HYPOCHLORITE BY THE from the manufacture and improper disposal of perchlorate-con- taining chemicals are increasingly being discovered in soil and water MASSACHUSETTS DEPARTMENT P (USEPA, 2007). OF ENVIRONMENTAL An additional source of perchlorate in drinking water has been found to occur through the use of sodium hypochlorite. The Massachusetts Department PROTECTION, of Environmental Protection (MDEP) has reported that significant levels of NSF INTERNATIONAL SURVEYED perchlorate can be detected in sodium hypochlorite samples that have aged for a few weeks (MDEP, 2005). Sodium hypochlorite as delivered to one utility FOR THE CONTAMINANT had a perchlorate concentration of 0.2 µg/L in the product, but the level of IN DRINKING WATER TREATMENT perchlorate rose to 6,750 µg/L after the product had aged for 26 days.

CHEMICALS FROM PRODUCTION INVESTIGATION OF WATER TREATMENT CHEMICALS BEGAN IN 2005 FACILITIES ACROSS In 2005 NSF International began analyzing samples of drinking water treat- ment chemicals for the contaminant perchlorate. These samples were collected THE UNITED STATES as part of the annual testing requirement to support NSF certification of the AND CANADA. treatment chemical to NSF/American National Standards Institute Standard 60: Drinking Water Treatment Chemicals—Health Effects (NSF/ANSI, 2005). Samples collected included not only sodium hypochlorite but other types of chemicals as well. NSF 60 currently requires testing of sodium hypochlorite samples for regulated metals, volatile organic compounds, and bromate. NSF continued the investigation of sodium hypochlorite through July 2006, resulting in the analysis of more than 67% of NSF-certified manufac- turers across North America. The levels of perchlorate reported here reflect potential at-the-tap concentrations calculated in accordance with the proce-

68 NOVEMBER 2008 | JOURNAL AWWA • 100:11 | PEER-REVIEWED | GREINER ET AL

2008 © American Water Works Association Aliquots of the sodium hypochlorite samples collected at manufacturers’ facilities were placed in 40-mL amber glass vials and stored in the dark prior to testing.

Perchlorate concentrations were determined by a liquid chromatography/mass spectrometry technique based on US Environmental Protection Agency method 331.0.

duresdures in NNSFSF 60. TheseThese “normal-“normal- izization”ation” ccalculationsalculations pprojectroject popotentialtential at-tat-the-taphe-tap conceconcentrationsntrations by assumingassuming thethe treatment chemi-chemi- cacall isis dosed aatt tthehe mmaximumaximum use llevelevel (MUL)(MUL) fforor wwhichhich it was certicerti-- ffied.ied. TypicallyTypically thethe MUL forfor so- diumdium hhypochloriteypochlorite productsproducts is equivalentequivalent to ddosingosing 10 mg/L ooff totaltotal chlorinechlorine into water. AAlthoughlthough this concentration is significantlysignificantly aboveb the th US USE Environmental i t lP Pro- tection Agency (USEPA) maxi- mum residual disinfectant level goal of 4.0 mg/L, it provides a worst-case evaluation of the so - dium hypochlorite by accounting for other potential uses such as prechlorination during water treatment and use during shock chlorination of water systems. Perchlorate health effects. Per- chlorate affects the ability of the thyroid gland to take up iodine (ATSDR, 2005). Iodine is needed to make thyroid hormones that are released into the blood and regu-

GREINER ET AL | PEER-REVIEWED | 100:11 • JOURNAL AWWA | NOVEMBER 2008 69

2008 © American Water Works Association late many body functions. Perchlo- final determination for perchlorate Laboratory analysis. The analysis rate is considered harmful to health after a 30-day public comment period. for perchlorate was performed when its inhibition of iodine uptake The agency also intends to issue a according to a modified USEPA is great enough to affect the thyroid. health advisory at the time it issues method 331.0, Determination of There is concern that hu man expo- the final regulatory determination in Perchlorate in Drinking Water by sure to higher amounts of perchlo- order to assist states with their local Liquid Chromatography Electro- rate for a long time may lower the response for perchlorate. spray Ionization Mass Spectrometry level of thyroid activity and lead to At the state level, perchlorate (USEPA, 2005). Method 331.0 is a hypothyroidism. Low levels of thy- guidance criteria of 14 µg/L in Ari- method for analyzing drinking roid hormones in the blood may zona, 5 µg/L in New York, and 1 water. All method requirements rel- adversely affect the skin, cardiovas- µg/L in Maryland and New Mexico evant to the analysis of sodium cular system, pulmonary system, have been adopted, along with hypochlorite rather than drinking kidneys, gastrointestinal tract, liver, action levels of 18 µg/L in New York water were included; the modifica- blood, neuromuscular system, ner- and Nevada and 4 µg/L in Texas tion of this method at NSF related to modification of the quality con- trol requirements. Method 331.0 allows for identi- fication by either tandem mass spec- trometry mode or single ion moni- erchlorate is considered harmful to health when its inhibition toring mode using dual ions (masses P 99 and 101). In this research, quan- of iodine uptake is great enough to affect the thyroid. tification was performed by internal standard calibration using the mass 101 ion. Results were reported in µg/L for liquid samples. In sodium hypochlorite, the average detection level for perchlorate was 250 µg/L. vous system, skeleton, male and (Bull et al, 2004). California has Approximately one third of the female reproductive systems, and established a perchlorate maximum samples tested were additionally numerous endocrine organs. Studies contaminant level (MCL) of 0.006 tested on multiple days to determine in animals have shown that the thy- mg/L (CDPH, 2007), and Massa- the rate of change in perchlorate con- roid gland is the main target of per- chusetts has established a perchlo- centration as the sodium hypochlo- chlorate toxicity. Animal studies rate MCL of 0.002 mg/L (MDEP, rite aged. Samples were maintained provided inconclusive results regard- 2006). For the purposes of estimat- in the dark and at room temperature ing effects of perchlorate on the ing the effect of perchlorate con- between analysis days. immune system. Perchlorate did not tamination, the current research affect reproduction in rats, accord- used the lowest of these values, in 164 CHEMICAL SAMPLES TESTED ing to one study. other words, 1 µg/L. Through July 2006, perchlorate Perchlorate regulation and guidance testing was performed on 164 sam- criteria. In October 2008 the USEPA SAMPLES NORMALLY COLLECTED ples of drinking water treatment announced a preliminary determina- DURING UNANNOUNCED AUDITS chemicals collected from 102 manu- tion on the regulation of perchlorate. TESTED FOR PERCHLORATE facturing locations. Of the 37 types After conducting an ex tensive review As part of NSF’s certification of chemicals tested, perchlorate was of scientific data re lated to the health program for drinking water treat- detected in only two: sodium effects of exposure to perchlorate ment chemicals, unannounced hydroxide and sodium hypochlorite from drinking water and other audits of manufacturing sites are (Table 1). sources, USEPA “. . . found that in performed annually, and samples of Of the 27 sodium hydroxide sam- over 99% of public drinking water certified treatment chemicals are ples, 22 (81%) had perchlorate levels systems, perchlorate was not at levels taken from recent production or reported as nondetectable; in the of public health concern. Therefore, retains. NSF used portions of these remaining five samples, perchlorate based on the Safe Drinking Water Act normally collected samples for this concentrations ranged from 0.01 to criteria, the agency determined there research on perchlorate. Once the 0.12 µg/L (Table 2). is not a ‘meaningful opportunity for samples were received at NSF, ali- The occurrence of perchlorate in health risk reduction’ through a quots were placed in 40-mL amber sodium hypochlorite was a more national drinking water regulation” glass vials and stored in the dark at common finding. Perchlorate was (USEPA, 2008). USEPA will make a room temperature before testing. detected in more than 91% of the

70 NOVEMBER 2008 | JOURNAL AWWA • 100:11 | PEER-REVIEWED | GREINER ET AL

2008 © American Water Works Association samples tested, at levels ranging from 0.03 to 29 µg/L. Table 3 TABLE 1 Summary of samples tested by chemical type groups the results by concentration range, including a running average Samples With of samples containing perchlorate at Samples Perchlorate levels less than or equal to the level With Perchlorate Detected Chemical Samples—n Detected—n % of perchlorate in the range. Of greater significance was the Aluminum chloride 1 correlation between the age of the Aluminum sulfate 2 sodium hypochlorite and the level of Ammonium hydroxide 3 perchlorate detected. Figure 1 shows Bentonite 1 the results of testing on samples Calcium hydroxide 1 with a known date of manufacture. Calcium hypochlorite 2 Results, plotted by sample age at the Calcium oxide 2 time of analysis, clearly demon- Carbon dioxide 1 strated a trend of increasing per- Copper sulfate 2 chlorate concentration as the hypo- Ferric chloride 2 chlorite product aged. Ferric sulfate 2 Three of the samples tested Ferrous chloride 1 yielded perchlorate concentrations Ferrous sulfate 1 of 8.8, 11, and 29 µg/L, significantly Fluorosilicic acid 1 greater than the levels found in Fluosilicic acid 1 other samples; the 29-µg/L value Hydrochloric acid 1 does not appear in Figure 1 because Hydrofluosilicic acid 1 the date of manufacture had not Hydrogen peroxide 1 been established. Because these con- Phosphoric acid 3 centrations were significantly out- Polyaluminum silicate sulfate 1 side the observed levels of perchlo- Potassium carbonate 1 rate formation in the other sodium Sodium bicarbonate 1 hypochlorite samples tested, the Sodium bisulfite 2 authors believe that contamination Sodium carbonate 1 of one of the component materials Sodium chloride 2 used to manufacture the sodium Sodium chlorite 1 hypochlorite may be the primary Sodium fluoride 1 perchlorate source. Sodium hexametaphosphate 1 Table 4 summarizes the occur- Sodium hydroxide 27 5 19 rences of perchlorate by sodium hy - Sodium hypochlorite 82 75 91 pochlorite age range. All of the sam- Sodium polyphosphates, glassy 3 ples tested within the first 30 days of Sodium silicate 5 production had a normalized perchlo- Sodium trimetaphosphate 1 rate concentration below 1 µg/L. Of Sulfuric acid 2 those samples tested between 30 and Trichloroisocyanuric acid 1 45 days after production, 97% had Zinc chloride 1 perchlorate concentrations below 1 Zinc orthophosphate 2 µg/L and just 3% had levels exceed- Total 164 ing that value. Between 45 and 60 days after production, however, 8% of samples tested showed perchlorate Samples were maintained in the tested over time were diluted at a concentrations exceeding 1 µg/L, and dark and at room temperature ratio of 1:2 with deionized water by 90 days after production, perchlo- between analyses. Results of the and also tested over time to deter- rate levels in 84% of samples ex - “over time” analysis are shown in mine whether the rate of perchlo- ceeded 1 µg/L. Figure 2. The plots demonstrated a rate formation was significantly Twenty-three of the samples consistent rate of increase across different in diluted form. As shown tested were analyzed for perchlorate multiple sample sources. in Figure 3, a comparison of the content on multiple days to provide Portions of three of the sodium full-strength and diluted samples insight into the rate of increase. hypochlorite samples that were found that the full-strength sodium

GREINER ET AL | PEER-REVIEWED | 100:11 • JOURNAL AWWA | NOVEMBER 2008 71

2008 © American Water Works Association hypochlorite generated perchlorate TABLE 2 Perchlorate occurrences in sodium hydroxide samples at a rate six to nine times faster than the same product diluted to Perchlorate Perchlorate Samples—n in Chemical—µg/kg At the Tap—µg/L half strength. Three of the sodium hypochlo- 22 ND (250) ND (0.03–0.05) rite samples were also evaluated 1 700 0.07 over time to determine whether the 1 900 0.09 level of bromate, chlorate, or chlo- 1 600 0.12 rite also changed with age. No sig- 1 160 0.03 nificant trend was noted for in - 1 110 0.01 creasing or decreasing bromate n—number, ND—not detected levels. This was expected because ND results were below the detection level of the analytical procedure as identified in the parentheses. For almost all of the bromine in chlo- calculation of the values in column 3, the level of perchlorate found in the chemical was multiplied by the maximum use level (MUL) certified for the individual chemical. Not all sodium hydroxides have the rine and the bromide in sodium same certified MUL. hydroxide—the primary ingredi- ents in sodium hypo chlorite—are TABLE 3 Perchlorate concentration range in sodium hypochlorite samples* quickly converted to bromate at the pH of sodium hypochlorite

Concentration Samples— (Chlorine Institute, 2004). The lev- Range—µg/L Samples—n Samples—% Running % els of chlorate and chlorite gener-

ND 7 9 9 ally increased with age, but sepa- > ND–1.0 42 51 60 rate research is needed to better > 1–2 9 11 71 quantify that behavior. > 2–3 15 18 89 Several factors were identified > 3–4 4 5 94 as contributing to variability in > 4–5 2 2 96 these results. > 5–6 0 0 96 • Composite samples were col- > 6–7 0 0 96 lected from manufacturers across > 7–8 0 0 96 one or more days of the manufac- > 8 3 4 100 turer’s production retains. For the Total 82 100 purposes of this study, the “date of manufacture” corresponding to n—number, ND—nondetected these samples was the date of the *At-the-tap in µg/L earliest retain of the composite sample. This practice particularly af fected the precise correlation be - tween age of the sodium hypo- FIGURE 1 Perchlorate in sodium hypochlorite (normalized to at-the-tap values) chlorite and the corresponding per- chlorate level. • The way the samples were 12 stored and shipped to NSF prior to storage and analysis at NSF also 10 added to the variability, given that both temperature and light have 8 been reported to affect the rate of perchlorate formation. 6 • Results were normalized to the maximum use level (MUL) for the 4 chemical in the NSF listing. The MULs were not necessarily propor- 2 tional to the strength of the sodium Normalized Perchlorate Result— µg/L 1 µg/L hypochlorite nor were they directly 0 associated with the level of chlorate. 0 15 30 45 60 75 90 105 120 135 150 165 180 195 210 The levels of perchlorate in this study Age of Sodium Hypochlorite—days have been presented as potential at- the-tap levels because this was the

72 NOVEMBER 2008 | JOURNAL AWWA • 100:11 | PEER-REVIEWED | GREINER ET AL

2008 © American Water Works Association primary concern being addressed TABLE 4 Perchlorate summary by age of NaOCl through NSF 60 evaluations. Perchlorate

SUMMARY AND CONCLUSION Age of NaOCl at Testing < 1 µg/L > 1 µg/L Testing affirmed the recurrent days after manufacture Analysis—n > 1 µg/L—n % % presence of perchlorate in sodium Յ 30 53 0 100 0 hypochlorite. This appeared to be > 30 to Յ 45 32 1 97 3 associated with the natural forma- > 45 to Յ 60 25 2 92 8 tion of perchlorate from chlorate, > 60 to Յ 90 24 4 83 17 but results suggested there may also > 90 32 27 16 84 be occurrences of perchlorate attrib- Total 166 utable to contamination from com- n—number, NaOCl—sodium hypochlorite ponent ingredients or manufactur- ing processes. The data compiled by NSF to hypochlorite for longer periods or tion, the data suggested a need for date supported the data previously have residual levels of aged chemical expiration dates on all sodium collected by MDEP on perchlorate in storage tanks that may contami- hypochlorite shipments to water occurrence in sodium hypochlorite. nate new shipments. utilities as well as on small contain- The data also supported the MDEP’s The data further indicated that ers of bleach that may be used by conclusion that the perchlorate lev- NSF 60 should address perchlorate small systems. els were probably not a concern for contamination. Perchlorate should For utilities that routinely use most water utilities that use sodium be a required parameter for all sodium hypochlorite supplies hypochlorite within a few weeks of sodium hypochlorite products, and within 45 days of manufacture, the production. However, perchlorate a single product allowable concen- contribution of perchlorate is likely occurrence may be a concern for tration for perchlorate needs to be to be negligible unless there is some water systems that store sodium established in the standard. In addi- contamination of the original ingre-

FIGURE 2 Perchlorate levels in sodium hypochlorite (normalized to at-the-tap values)

4.00 Sample number 1 2 3 4 5 6 7 8 9 3.50 10 11 12 13 14 15 16 17 18 3.00 19 20 21 22 23

2.50

2.00 Perchlorate— µg/L

1.50

1.00

0.50

0.00 0 15 30 45 60 75 90 105 120 135 Age of Sodium Hypochlorite—days

GREINER ET AL | PEER-REVIEWED | 100:11 • JOURNAL AWWA | NOVEMBER 2008 73

2008 © American Water Works Association ABOUT THE AUTHORS FIGURE 3 Comparison of perchlorate formation rates for full-strength and Peter Greiner (to diluted (1:2 ratio) sodium hypochlorite whom correspon- Sample number dence should be 8,000 21 (full strength) addressed) is a 21 (1:2 dilution) technical manager 7,000 22 (full strength) with NSF Interna- 22 (1:2 dilution) tional, 789 N. 6,000 23 (full strength) 23 (1:2 dilution) Dixboro Rd., Ann 5,000 Arbor, MI 48105; greinerp@nsf. org. He received a BS degree in 4,000 aquatic biology from Eastern Mich- 3,000 igan University in Ypsilanti. Since Split— µg/L Greiner has more than 28 years of 2,000 experience in product certification

1,000 and is currently a member of the Drinking Water Additives Joint

Change in Perchlorate Sodium Hypochlorite 0 Committee charged with oversight 0 5 10 15 20 25 30 35 40 45 of NSF International/American Days Since Sample Split National Standards Institute Stan- dard 60. Clif McLellan is the direc- tor of toxicology services, Dale dients. Utilities or small systems emptied of aged material and Bennett is an operations manager, that store sodium hypochlorite for flushed to minimize the potential and Angie Ewing is a senior toxi- longer periods may encounter sig- for contamination. cologist with NSF International. nificant levels of perchlorate in the finished drinking water. To mini- ACKNOWLEDGMENT Date of submission: 06/20/07 mize the perchlorate risk, sodium The authors acknowledge the Date of acceptance: 01/24/08 hypochlorite should be stored in work performed by the state of Mas- the dark at cool temperatures, sachusetts and thank the staff of the If you have a comment about this diluted if possible, and used within California Department of Health article, please contact us at a few weeks of manufacture. Stor- Services for first bringing the issue [email protected]. age tanks and piping should also be to their attention.

REFERENCES ton, Va., http://www.chlorineinstitute. Water Treatment Chemicals—Health org/files/FileDownloads/ Effects. NSF Intl., Ann Arbor, Mich. ATSDR (Agency for Toxic Substances and BromateinNaOCl-PotableWater Disease Registry), 2005. Perchlo- USEPA (US Environmental Protection Treatment.pdf. rates. Div. of Toxicology & Envir. Med. Agency), 2008. Regulatory Determi- ToxFAQsTM, September. http://www. MDEP (Massachusetts Department of nations for Priority Contaminants on atsdr.cdc.gov/tfacts162.pdf. Environmental Protection), 2006. the Second Drinking Water Contami- 7/28/06 Revisions to 310 CMR 22.00. nant Candidate List. October 2008 Bull, R.J.; Chang, A.C.; Cranor, C.F.; Shank, Inorganic Chemical Maximum Con- Preliminary Regulatory Determina- R.C.; & Trussell, R., 2004. Perchlorate taminant Levels, Monitoring Require- tion for Perchlorate. http://www.epa. in Drinking Water: A Science and Pol- ments, and Analytical Methods. gov/safewater/ccl/reg_determine2. icy Review. Urban Water Research http://www.mass.gov/dep/ html-perchlorate. Center, University of California, water/laws/perchlorate- Irvine. USEPA, 2007. Drinking Water Contami- 310CMR22-07282006.doc. nants. Perchlorate. http://www.epa. CDPH (California Department of Public MDEP (Massachusetts Department of gov/safewater/contaminants/ Health), 2007. Perchlorate in Drinking Environmental Protection), 2005. The unregulated/perchlorate.html. Water. http://www.cdph.ca.gov/ Occurrence and Sources of Perchlo- USEPA, 2005. Method 331.0—Determina- certlic/drinkingwater/Pages/ rate in Massachusetts. Draft Report, tion of Perchlorate in Drinking Water Perchlorate.aspx., accessed http://www.mass.gov/dep/cleanup/ by Liquid Chromatography Electro- Nov. 17, 2007. sites/percsour.pdf. spray Ionization Mass Spectrometry. Chlorine Institute, 2004. Bromate in So- NSF/ANSI (NSF International/American EPA 815-R-05-007, Cincinnati. http:// dium Hypochlorite Potable Water National Standards Institute), 2005. www.epa.gov/safewater/methods/ Treatment. Chlorine Institute, Arling- NSF/ANSI Standard 60. Drinking pdfs/methods/met331_0.pdf.

74 NOVEMBER 2008 | JOURNAL AWWA • 100:11 | PEER-REVIEWED | GREINER ET AL

2008 © American Water Works Association