Article Cite This: Environ. Sci. Technol. XXXX, XXX, XXX−XXX
pubs.acs.org/est
Evaluating the Comparative Toxicity of DBP Mixtures from Different Disinfection Scenarios: A New Approach by Combining FreezeDrying or Rotoevaporation with a Marine Polychaete Bioassay Jiarui Han and Xiangru Zhang* Department of Civil and Environmental Engineering, Hong Kong University of Science and Technology, Hong Kong, China
Environ. Sci. Technol. Downloaded from pubs.acs.org by EASTERN KENTUCKY UNIV on 09/02/18. For personal use only.
S Supporting Information *
ABSTRACT: The unintended formation of disinfection byproducts (DBPs) may compromise the safety of drinking water. Since no specified DBPs have been found to be responsible for the overall adverse effects and over half of total organic halogen (TOX) remains unidentified, DBP mixture toxicity is gaining increasing interest as a potential indicator of how risky drinking water might be. In this study, a new approach to evaluating the toxicity of drinking water DBP mixtures was developed by combining freeze-drying or rotoevaporation pretreatment with an in vivo high-salinity-tolerance bioassay with the embryos of a marine polychaete Platynereis dumerilii. The DBP recoveries by freeze-drying or rotoevaporation were compared with those by commonly applied liquid−liquid-extraction (LLE). For drinking water subjected to typical disinfection processes (i.e., chlorination, chloramination, chlorine dioxide treatment, and ozonation with or without postchlorination), LLE led to the lowest TOX recovery (11−18%) and the loss of all inorganic DBPs, while freeze-drying and rotoevaporation recovered 28−58% and 35−61% of TOX, respectively, and effectively recovered 81− 99% and 85−104% of inorganic DBPs, respectively. Thus, LLE caused an underestimation of the toxicity of DBP mixtures compared with freeze-drying and rotoevaporation. Besides, the comparative toxicity varied significantly for water samples pretreated with different methods due to the effect of inorganic DBPs and a synergistic effect of organic and inorganic DBPs. The new approach revealed that the bromide-rich source water disinfected with ozone caused the highest developmental toxicity, followed by those disinfected with chlorine, chlorine dioxide, and chloramine in that order.
■
INTRODUCTION Humans typically consume approximately 50 000 L of water over a lifetime, raising concerns about long-term exposure to pathogens and hazardous substances in drinking water.1−3 Disinfection reduces the likelihood of waterborne disease outbreaks and improves the quality of drinking water. Chlorine (Cl2), chloramine (NH2Cl), chlorine dioxide (ClO2), and ozone (O3) are widely used drinking water disinfectants. In the disinfection process, however, disinfection byproducts (DBPs) are produced when these disinfectants react with natural organic matter (NOM), bromide, and iodide.3−17 Epidemiological studies have suggested an association between consuming drinking water containing DBPs and increased adverse reproductive and carcinogenic effects.4,18 To date, as many as 700 DBPs have been identified in drinking water.19−21 Most laboratory-based toxicological studies have focused on individual DBPs.6,14,22−24 Meanwhile, identification and quantification of DBPs are largely uncompleted due to the inherent complexity of NOM and the limitations of detection methods for identifying highly polar DBPs and high-molecularweight DBPs.2,25,26 Moreover, source waters impacted by upstream wastewater discharges contain effluent organic matter and contaminants that may also react with disinfectants © XXXX American Chemical Society
to form DBPs, which makes the issue even more complicated.2,27,28 Studies have shown that identified DBPs only accounted for 10−50% of total organic halogen (TOX, a collective parameter for all organic halogenated DBPs) produced by Cl2, NH2Cl, ClO2, and O3.29,30 Even for the DBPs that have been identified, only around 100 of them have been subjected to toxicological investigation.4,14 Although the toxicological importance of emerging DBPs (e.g., nitrogenous DBPs and iodinated DBPs) has been well acknowledged,4,27,31,32 the toxicity based on individual DBPs may not fully interpret the epidemiological effects due to the potential interactions among DBPs.33 There is an increasing interest to evaluate the toxicity of DBP mixtures containing both identified and unidentified species considering that one is exposed to a mix of all kinds of DBPs when consuming drinking water.22,34−36 However, most studies about DBP mixture toxicity focused on bioassays of organic DBPs formed in one or two disinfection processes.36−38 By evaluating the Received: Revised: Accepted: Published: A
April 17, 2018 August 7, 2018 August 20, 2018 August 20, 2018 DOI: 10.1021/acs.est.8b02054 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
combined the freeze-drying or rotoevaporation pretreatment with the P. dumerilii bioassay and (ii) to evaluate the efficiencies of pretreatment methods in retaining organic and inorganic DBPs. Since LLE shows extraction efficiencies similar to SPE for various DBPs,53−55 the efficiency of LLE in retaining halogenated organic DBPs was evaluated and compared with that of freeze-drying or rotary evaporation. The effect of pretreatment methods on the toxicity assessment of a DBP mixture (including both organic and inorganic DBPs) was also investigated.
toxicity of DBP mixtures in drinking water disinfected with Cl2, NH2Cl, ClO2, or O3, a relatively safe disinfectant could be identified. Since most DBPs are present at submicrogram to microgram per liter levels, pretreatment is required to concentrate DBPs in drinking water before bioassays. Liquid−liquid extraction (LLE) and solid-phase extraction (SPE) are the most commonly employed methods for preparing concentrated water prior to biological assays.19,38−40 Drinking waters are complicated mixtures containing compounds with various properties. SPE and LLE are basically processes of distributing a certain compound between an organic solvent/sorbent and water, during which inorganic DBPs including bromate, chlorite, and chlorate are entirely lost. Bromate and chlorite are the key DBPs in ozonation and ClO2 disinfection, respectively, with 10 μg/L bromate and 1.0 mg/L chlorite being the maximum levels allowed in drinking water by the U.S. Environmental Protection Agency (USEPA).1 Additionally, (highly) polar organic DBPs may not be effectively extracted during SPE or LLE, and volatile DBPs are entirely lost during organic solvent removal for bioassays. These losses of DBPs mean that the adverse impacts of DBP mixtures are often underestimated. We thus propose to use a simple and solvent-free pretreatment method to concentrate water samples. The method involves removing the water content while retaining most of the organic DBPs, inorganic DBPs, total dissolved salts, and disinfected water matrix through freeze-drying or rotary evaporation. Freeze-drying has been applied in source water pretreatment to concentrate NOM,41 and it has recently been used to concentrate disinfected water.38 Rotary evaporation is commonly used to remove organic solvents, but it can be extended to NOM enrichment by removing liters of water.42 Freeze-drying and rotary evaporation can theoretically retain all nonvolatile organic and inorganic DBPs, but their efficiencies in retaining organic and inorganic DBPs still require investigation. Through pretreatment by freeze-drying or rotary evaporation, a drinking water concentrate can be obtained without the volatile fraction. For such a concentrate, regularly used bacterial or mammalian-cell assays may not work because the total dissolved salts are also retained and concentrated, and the high salinity would damage the bacteria and mammalian cells.43,44 For instance, the average level of total dissolved salts in surface water is approximately 120 mg/L.45 Once the water is concentrated by 100 times for biological analysis, the salinity of the concentrated water becomes 12 000 mg/L, while the growth inhibition of Salmonella typhimurium occurs at a salinity over 5000 mg/L.43 Recently, the embryos of a cosmopolitan marine polychaete, Platynereis dumerilii, have been used successfully for studying the developmental toxicity of individual DBPs and DBP mixtures.46−51 The in vivo bioassay is highly sensitive and reproducible, and more importantly the marine polychaete can tolerate a high salinity. Thus, in this study we adopted this bioassay to evaluate the developmental toxicity of different drinking water DBP mixtures. Notably, because it has been proved that the volatile fraction of a DBP mixture has little contribution to the overall developmental toxicity,52 this bioassay could still reveal the toxic potency of the DBP mixtures despite the loss of volatile DBPs during the sample pretreatment. The objectives of this study were (i) to evaluate the comparative toxicity of DBP mixtures of different disinfectants (Cl2, NH2Cl, ClO2, or O3) using a new approach that
■
MATERIALS AND METHODS Chemicals, Reagents, and Seawater. Suwannee River NOM (SRNOM) was purchased from the International Humic Substances Society. Ultrapure water (18.2 MΩ·cm) was obtained from a water purification system (Cascada I, Pall, USA). Seawater was collected locally, filtered with a 0.45 μm membrane, autoclaved at 121 °C for 30 min, cooled to room temperature, and aerated for 1 h before it was used for polychaete culturing and the developmental toxicity bioassay. A stock solution of NaOCl was obtained from Sigma-Aldrich and calibrated using the DPD ferrous titrimetric method.56 A NH2Cl solution was freshly prepared before use by gradually adding NaOCl to an NH4Cl solution to a molar ratio of 0.8:1 (Cl:N), and the concentration was determined using the DPD ferrous titrimetric method.56 ClO2 was generated in the laboratory as described in Standard Method 4500-ClO2 B.56 Briefly, 20 mL of H2SO4 (10%, v/v) was slowly added to 500 mL of a NaClO2 solution (0.5 M). The generated ClO2 gas passed through a scrubber containing 600 mL of a saturated NaClO2 solution, and then ClO2 was absorbed in ultrapure water. The prepared ClO2 stock solution was found to be chlorine-free using the iodometric method with dimethyl sulfoxide as described elsewhere.56,57 The pure ClO2 stock was standardized spectrophotometrically with the molar absorption ratio of 1150 M−1 cm−1 at 360 nm.29 The pure ClO2 stock was kept in amber glass vials without headspace at 4 °C until use. O3 was generated in the laboratory with an ozone generator (10K-2U, Enaly, China) and absorbed in a 2 L glass reactor containing ultrapure water. The concentration of dissolved ozone was measured using the indigo colorimetric method.56 Disinfection of a Simulated Raw Water. A simulated raw water was prepared with ultrapure water containing 3 mg/ L SRNOM as C, 80 mg/L NaHCO3 as CaCO3, 2.0 mg/L NaBr as Br−, and 0.20 mg/L KI as I−. The relatively high levels of bromide and iodide as reported in bromide- and iodide-rich source waters54,58 were used to amplify the formation of brominated and iodinated DBPs since brominated and iodinated DBPs are much more toxic than their chlorinated analogues.4,46,59 Doses and contact times were determined according to previous studies on comparing DBP formation of different disinfectants.29,60,61 Two contact times, 30 min and 48 h, were applied to simulate disinfection in drinking water treatment plants and distribution systems, respectively. The doses of ClO2 and O3 were 1.5 mg/L as ClO2 and 3 mg/L as O3, respectively. Preliminary tests were conducted after a contact time of 48 h using the DPD ferrous titrimetric method to determine the doses of Cl2 and NH2Cl required to achieve residual targets of 0.5 mg/L as Cl2 for chlorination and 1.5 mg/L as Cl2 for chloramination.56 All water samples were adjusted to pH 7.5 with 0.1 M HCl. The same set of disinfection conditions was applied to raw waters for both B
DOI: 10.1021/acs.est.8b02054 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
300 mL of ultrapure water for TOX analysis following the same procedure. Notably, for the raw water disinfected with ClO2 or O3 (+ Cl2) before and after the pretreatment with freeze-drying or rotary evaporation, 105% of the stoichiometric amount of Na2S2O3 was added to reduce inorganic DBPs (including chlorite, chlorate, and bromate) before pH adjustment during TOX analysis. For TOX analysis, a sample was first adjusted to pH 2 with concentrated nitric acid before adsorption onto activated carbon columns. Under acidic conditions, chlorite, chlorate, and bromate may react with NOM, the formed DBPs, or the activated carbon and thus affect the TOX analysis.56,63,67 The removal of these inorganic DBPs enables a more accurate comparison of the TOX recoveries for the three pretreatment methods. Recoveries of inorganic DBPs, i.e., chlorite and chlorate in ClO2 disinfection and bromate in ozonation, were evaluated by comparing the inorganic DBP concentrations in samples without pretreatment and in the redissolved samples after freeze-drying or rotary evaporation. These inorganic DBPs were measured with an ion chromatograph (ICS-3000, Dionex, USA). TOX and inorganic DBPs analyses were performed in duplicate. Comparative Developmental Toxicity Bioassay with the Embryos of P. dumerilii. P. dumerilii was cultured and the developmental toxicity bioassay was conducted as described previously and detailed in the Supporting Information.46 Toxicity test solutions were obtained by dissolving solid concentrates of the disinfected water samples pretreated by LLE, freeze-drying, or rotary evaporation. Since dissolved salts were also retained during freeze-drying and rotary evaporation, the tolerance of the embryos of P. dumerilii to salinity was tested (see the Supporting Information for details). The salinity of toxicity test solutions was controlled by dissolving the sample concentrates using seawater or seawater diluted with ultrapure water to eliminate potential negative effect of salinity. Notably, toxicity test solutions from freezedrying and rotary evaporation had a pH higher than 10. The high pH negatively affected the accuracy of the bioassay because it exceeded the pH tolerance of the embryos68 and resulted in precipitation which interfered with the observation. Accordingly, the toxicity test solutions were adjusted to pH 7.0−8.5 with 0.5 M HCl. By doing so, these solutions showed almost no precipitates, and the adverse effect induced by pH was eliminated.68 Embryos at 12 h postfertilization were exposed to water samples that were disinfected with Cl2, NH2Cl, ClO2, or O3 (+ Cl2) and pretreated by one of the three methods. Normal embryos develop to the first larval stage by 24 h postfertilization. The normal development percentage was calculated as the ratio of normally developed embryos to total embryos. Preliminary tests were conducted by dissolving the solid concentrate of a water sample to different volumes to determine the critical concentration factor for toxicity tests. The higher the normal development percentage for a water sample at the same concentration factor, the lower the toxicity. Additionally, the higher the toxicity of a sample pretreated by a certain method, the higher the efficiency of that method in retaining and concentrating toxicologically related DBPs in the sample. Control samples were prepared by exposing embryos to seawater for 12 h. For the developmental toxicity bioassay, all samples were analyzed in duplicate. Combination Effect of Organic and Inorganic DBPs on the Development of the P. dumerilii Embryos. As discussed in Results and Discussion, with the pretreatment of
contact times. Disinfection with Cl2, NH2Cl, or ClO2 was performed by adding a predetermined volume of a disinfectant stock solution to a 2 L chlorine demand-free glass bottle containing the simulated raw water. Ozonation was performed by adding the ozone-containing solution to a predetermined volume of 50× concentrated simulated raw water (i.e., 150 mg/L SRNOM as C, 4.0 g/L NaHCO3 as CaCO3, 100 mg/L NaBr as Br−, and 10 mg/L KI as I−). Upon addition of the ozone-containing solution, the desired ozone dose was achieved for the simulated raw water. In the case of disinfection for 48 h, postchlorination was performed on the ozonated water sample after the first 30 min. All disinfection samples were stored without headspace at room temperature in the dark. After the given contact time, 105% of the stoichiometric amount of Na2S2O3 was added to quench the residual disinfectant.38 Figure S1 shows the scheme of the raw water disinfection, pretreatment, and chemical and biological analyses. Sample Pretreatment. Each disinfected water sample was concentrated in three different ways, i.e., by LLE, freeze-drying, or rotary evaporation. LLE with methyl tert-butyl ether (MtBE) as the extracting solvent was adopted since it has been widely used in chemical and biological analyses of DBPs.19,48−50 The procedure of LLE is described elsewhere.48−50,62 In brief, a 1 L aliquot was acidified to pH 0.5 with H2SO4 (70%, v/v) and saturated with 100 g of Na2SO4. The aliquot was then extracted with 100 mL of MtBE. The MtBE layer collected was transferred and concentrated to 0.5 mL by rotary evaporation. The 0.5 mL concentrate was dried to solid with a gentle nitrogen gas and kept at 4 °C until analysis. Freeze-drying was performed in a freeze-dryer (FD1A-50, BiLon, China). A 1 L aliquot was transferred onto five glass plates (200 mm × 30 mm) and prefrozen first at −18 °C for 8 h and then at −36 °C for 12 h to avoid the glass breakage, followed by freeze-drying under a condenser temperature of −50 °C and an inside pressure below 0.1 mbar. The solid concentrates on the five plates were transferred to a 40 mL amber vial. Each glass plate was then rinsed three times with ultrapure water, which was also transferred to the vial. The concentrates and rinsing solution were dried to solid following the above prefreezing and freeze-drying procedure. Rotary evaporation was conducted in a rotary evaporator (Laborota 4000, Heidolph, Germany). A 1 L aliquot was transferred equally into five 500 mL round-bottom flasks, and each was concentrated to 1 mL with the rotary evaporator with the water-bath temperature set at 30 °C and the rotary speed at 180 rpm. The relatively low temperature was selected to prevent DBPs from thermally decomposing. Each 1 mL concentrate in the five flasks and the ultrapure water used for rinsing each flask were transferred to a 25 mL round-bottom flask and rotary-evaporated to dryness. Vacuum grease was applied to the joints of the evaporation system to prevent the failure of air-tightness. Analyses of TOX and Inorganic DBPs. The measurement of three components of TOXtotal organic chlorine (TOCl), total organic bromine (TOBr), and total organic iodine (TOI)is detailed in the Supporting Information.63−66 TOX recoveries for different pretreatment methods were evaluated by comparing the TOX in a disinfected sample and that retained after pretreatment. Briefly, a 300 mL disinfected sample aliquot was directly used for TOX analysis. Another 300 mL aliquot was concentrated to solid by LLE, freezedrying, or rotary evaporation. The solid was then redissolved in C
DOI: 10.1021/acs.est.8b02054 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 1. TOX concentrations (dark blue ■, TOCl; pink ■, TOBr; light blue ■, TOI) for different disinfectants after the contact times of (a) 30 min and (b) 48 h and TOX recoveries for different pretreatment approaches after the contact times of (c) 30 min and (d) 48 h.
after the contact times of 30 min and 48 h. Among the samples disinfected for 30 min (Figure 1a), Cl2 generated the most TOX, while ClO2 generated the least (ca. 11% of TOX formed during chlorination). Similar levels of TOX were generated by NH2Cl and O3, while the compositions were significantly different. TOBr was the dominant component of TOX in chlorination and ozonation due to the relatively high bromide level and the fast substitution of NOM by bromine, which was formed in the oxidative reaction between Br− and Cl2/O3.12,70 TOI accounted for about half of TOX in chloramination. Cl2 and O3 have been reported to effectively oxidize I− to HOI/ OI− and further to IO3−, while ClO2 and NH2Cl can only oxidize I− to HOI/OI− which further reacts with NOM to form I-DBPs.71,72 In this study, I-DBPs were the sink of approximately 7% of iodide for the ozonated sample and over 81% of iodide for the chloraminated sample. Cl2 and ClO2 formed the same levels of TOI, but the ratio of TOI to TOX in the sample disinfected with ClO2 was 33%, much higher than the 5% for the chlorinated sample. As the contact time increased from 30 min to 48 h (Figure 1b), the TOBr level doubled while the TOI level decreased by 60% for the chlorinated sample. The TOBr increase and the TOI decrease in chlorination were primarily attributed to the reactions of HOCl/OCl− and HOBr/OBr− with intermediate I-DBPs in the presence of chlorine residual.12 When Cl2 served as a secondary disinfectant following ozonation, TOX was reduced by 35% as compared to that for chlorination alone because O3 could change the characteristics of NOM and reduce the precursors of halogenated DBPs,73,74 while ratios of the three components to TOX remained relatively stable. For 48 h chloramination, TOI reduced slightly and TOX increased
rotary evaporation or freeze-drying, the embryos in the bioassay were coexposed to organic and inorganic DBPs in a DBP mixture. Due to different mechanisms by which organic or inorganic DBPs could affect the development of the embryos, their combination could produce a synergistic, additive, or antagonistic effect.69 To investigate the combination effect of organic and inorganic DBPs, the developmental toxicity bioassay was conducted by exposing the embryos to the combined organic and inorganic DBPs, the organic DBPs alone (by removing inorganic DBPs in the DBP mixture, detailed in the Supporting Information), and the inorganic DBPs alone of the water that was disinfected with O3 or ClO2 and pretreated with freeze-drying. As described by Chou,69 a combination index (CI) can be calculated to evaluate the combination effect of the components in a mixture using the following equation: CI =
d dA + B DA DB
(1)
where dA and dB are the doses of mixture components A and B at which a specific adverse effect is induced by their combination and DA and DB are the doses of A and B at which the same effect is induced by them separately. According to Chou,69 CIs in the ranges of less than 0.9, 0.9−1.1, and over 1.1 correspond to synergism, addition, and antagonism, respectively. In this study, we defined A as organic DBPs and B as inorganic DBPs in a drinking water DBP mixture.
■
RESULTS AND DISCUSSION TOX Formation in Different Disinfection Scenarios. TOX formation was analyzed for various disinfection scenarios D
DOI: 10.1021/acs.est.8b02054 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Table 1. Chlorite and Chlorate Detected in the ClO2-Disinfected Samples and Bromate Detected in the O3-Disinfected Samples without Pretreatment and with Rotoevaporation and Freeze-Drying Pretreatment chlorite (μg/L) without pretreatment rotary evaporation freeze-drying
chlorate (μg/L)
bromate (μg/L)
30 min
48 h
30 min
48 h
30 min
48 h
905 ± 28 847 ± 12 797 ± 31
951 ± 30 808 ± 31 771 ± 47
99 ± 4 97 ± 3 98 ± 3
96 ± 5 95 ± 2 94 ± 5
112 ± 4 117 ± 3 108 ± 2
117 ± 2 115 ± 5 113 ± 4
molecular size.73,76 Under high-vacuum conditions, halogenated DBPs that had low molecular weights and thus were more volatile than high-molecular-weight DBPs were more likely to be lost during freeze-drying and rotary evaporation. Such loss was more significant for freeze-drying since the pressure in the freeze-dryer (
