Environ. Sci. Technol. 2007, 41, 8383–8387
False Cyanide Formation during Drinking Water Sample Preservation and Storage MICHAEL F. DELANEY,* CHARLES BLODGET, CORINNA E. HOEY, NANCY E. MCSWEENEY, POLINA A. EPELMAN, AND STEVEN F. RHODE Massachusetts Water Resources Authority (MWRA), 190 Tafts Avenue, Winthrop, Massachusetts 02152
Received June 07, 2007. Revised manuscript received September 19, 2007. Accepted September 27, 2007.
Carefully controlled bench-scale and on-site experiments demonstrated that cyanide can form in the treated drinking water sample container during preservation and storage. In the bench-scale experiment, treated tap water samples were collected on 20 days over six months. The tap water samples were split and some of the splits were spiked with formaldehyde, a known ozone disinfection byproduct, held for three hours and tested for cyanide. Then they were preserved and held for 2–10 days. None of the 69 initial samples had cyanide detects, but 22 of 49 formaldehyde-spiked samples and three of the 20unspikedsamplesdevelopeddetectablecyanideconcentrations during storage. In the on-site experiment, six samples were collected at a finished water tap at an ozone/chloramination treatment plant over three days. Each sample was split, and a portion was spiked with formaldehyde. Each portion was analyzed in triplicate after three different procedures: (1) immediately distilled on-site, (2) stabilized on-site in a distillation tube and distilled back at the laboratory several days later, or (3) following the conventional procedure of preserving the sample to pH >12 in a container and distilling the sample back at the laboratory. Only the samples handled in the conventional way had detectable amounts of cyanide. Both experiments demonstrated that cyanide can form during conventional preservation and storage, and it is likely that the cyanide detected for this treated drinking water was formed in the sample container as a consequence of the preservation and storage conditions.
Introduction The Massachusetts Water Resources Authority (MWRA) supplies unfiltered surface water wholesale to 42 Eastern Massachusetts communities, including Boston. Drinking water treatment was converted from chlorination/chloramination to ozone/chloramination in July 2005. Subsequent to this cyanide has been detected above 10 µg/L in 10 out of about 50 finished water samples. All detected cyanide levels were below the regulatory drinking water maximum contaminant level (MCL) for cyanide (200 µg/L), and only one sample was above half of the MCL. The detected concentrations were mostly just above the reporting limit of 10 µg/L. * Corresponding author phone: (617) 660-7801; fax: (617) 6607960; e-mail:
[email protected]. 10.1021/es071359r CCC: $37.00
Published on Web 11/22/2007
2007 American Chemical Society
On the basis of our prior experience with testing wastewater for cyanide (1, 2), we were concerned that the cyanide detections could be an artifact of the preservation and analysis method. A comprehensive examination of cyanide in the environment, including analytical methods, has been presented by Dzombak et al. (3). We describe here bench-scale and on-site experiments conducted to distinguish between any cyanide that was present in the treated drinking water from cyanide that might have formed during preservation and storage of samples. The general experimental approach was to test fresh samples after collection and again after preservation and storage. Portions of each sample were spiked with formaldehyde, a known ozone disinfection byproduct, to simulate a key aspect of the ozonation process and to potentially stimulate cyanide formation. This design would clearly distinguish between cyanide present in the fresh sample versus cyanide that was formed during preservation and storage.
Experimental Section Source Water and Treated Drinking Water. The MWRA source water, from the Quabbin and Wachusett reservoirs, is very low in total dissolved solids, low in hardness, low in alkalinity, well-oxygenated, slightly acidic, (4) and has a total organic carbon of about 2–3 mg/L. The unfiltered surface water is treated at the John J. Carroll Water Treatment Plant (CWTP) in Marlborough, MA, first with ozone for primary disinfection. The ozone is quenched with sodium bisulfite if necessary. This is followed by sodium carbonate (soda ash) and carbon dioxide for corrosion control, hydrofluorosilicic acid for fluoridation, and sodium hypochlorite, followed by aqueous ammonia to form chloramines as a secondary disinfectant. There is ∼5 min of free chlorine contact time before the ammonia is added. The final treated water has an alkalinity of 40 mg/L and a pH of 9.3. Formaldehyde, a known ozone disinfection byproduct (5), is present in the treated drinking water at a median concentration of 15 µg/L, ranging from 12, when used, was with 4 mL of 10 M NaOH per 1 L bottle. Cyanide Analysis. All samples in the bench-scale experiment and all “conventionally handled” samples described in the on-site experiment were “midi-distilled” according to United States Environmental Protection Agency (EPA) Method 335.4 (7), using commercial midi-distillation glassware and a heated block (MIDI-VAP 2000, KONTES Glass Co., Vineland, New Jersey). All microdistillation (MICRO DIST) samples in VOL. 41, NO. 24, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8383
FIGURE 1. MICRO DIST tube assembly used to stabilize samples and to perform the cyanide distillation (reprinted with permission from Lachat) (8). the on-site experiment were distilled according to Lachat QuikChem Method 10-204-00-1-X (8). MICRO DIST is approved by EPA for cyanide analysis of drinking water and wastewater. It uses plastic tubes to distill hydrogen cyanide gas (HCN) out of samples. Figure 1 shows the MICRO DIST tube assembly. In this approach, 6 mL of dechlorinated sample is placed in the sample tube. Then 0.75 mL of 7.11 M sulfuric acid/0.79 M magnesium chloride solution is added. The sample tube is then immediately sealed to the rest of the tube assembly that has been preloaded with 1.0 M sodium hydroxide (NaOH) absorber solution. Any HCN that is liberated from the sample passes through the hydrophobic membrane, where it is trapped in the NaOH absorber solution. A heated aluminum block unit is used to conduct the distillation at 120 °C for 30 min. No other treatments for potential interferences were used in these experiments. All cyanide samples were analyzed on segmented flow analyzer (Skalar Sanplus Segmented Flow Analyzer, Skalar Analytical B.V., The Netherlands) according to EPA method 335.4 (7). All MICRO DIST samples were quantitated against MICRO DIST distilled standards handled in the same manner as the corresponding samples. Conventionally handled samples were quantitated against undistilled standards. All method blanks (laboratory reagent blanks) were less than 4 µg/L. All laboratory fortified blanks and matrix spikes (laboratory fortified matrix) had recoveries in the range of 90–110%. All matrix spike duplicates had relative percent differences from the corresponding matrix spike of 12 with NaOH, and refrigerated. These treated portions were midi-distilled and analyzed after 2–10 days of refrigerated storage. This mimics the typical storage time from the time of sampling and preservation to the time of distillation and analysis. These portions were also dechlorinated again if the total chlorine residual was found to be above 0.10 mg/L after storage. We have observed that some drinking water and wastewater samples appear to redevelop small amounts of residual chlorine after they have been properly dechlorinated at the time of sample collection. This may be attributed to the slow reaction kinetics of larger organic chloramines formed from peptides, proteins and other naturally occurring nitrogenous organic compounds (10–12). On-Site Experimental Design. The on-site experiment was designed to directly compare three sampling, preservation, and storage approaches: (1) immediately distilled onsite at CWTP in MICRO DIST tubes (alternative 1), (2) sealed in MICRO DIST tubes and distilled a few days later (alternative 2), and (3) preserved to pH >12 with NaOH and distilled a few days later (conventional procedure). The only difference between alternatives 1 and 2 is the time until distillation was initiated. In both cases, 6 mL of ascorbic acid dechlorinated sample and the magnesium chloride/sulfuric acid solution were immediately sealed into the MICRO DIST tube. For alternative 1 distillation was immediately initiated. For alternative 2, the tube was refrigerated, and distillation was conducted back at the laboratory 1-4 days later. The conventionally handled samples were rechecked for residual chlorine just before distillation and dechlorinated again with ascorbic acid if the residual chlorine was >0.1 mg/L. Six samples were collected over three days (April 2–4, 2007) from the final treated water tap location at CWTP. Each sample was dechlorinated with ascorbic acid and split. One portion was spiked with formaldehyde at 50 µg/L. The formaldehyde-spiked portion was either held for three hours (morning sample) or processed immediately (afternoon sample). Each sample was prepared in triplicate for each of the three sampling approaches for a total of 108 samples [6 samples × 2 (spiked or unspiked) × 3 (sampling approaches) × 3 (triplicates)].
Results and Discussion Bench-Scale Experiment Results. The bench-scale results (Table 1) show a clear pattern of cyanide being absent in the initial samples but developing during preservation and storage. Cyanide was not detected at 4 µg/L in any of the 69 initial samples (100%), whether spiked with formaldehyde or not and whether dechlorinated or not. When these same samples were dechlorinated, pH adjusted to pH >12, and held for 2-10 days, cyanide was frequently detected. Three out of 20 samples (12%) that were not spiked with formaldehyde and 22 out of 49 samples (45%) that were spiked with formaldehyde developed detectable cyanide during storage (Figure 2). This shows that samples that initially do not have detectable levels of cyanide can develop detectable cyanide concentrations simply by preservation and storage of the sample in the manner used for conventional samples. On-Site Experiment Results. All 36 immediately distilled samples (alternative 1) and all 36 MICRO DIST stabilized samples (alternative 2) had undetectable cyanide concentrations (
