Identification of Halohydrins as Potential Disinfection By-Products in Treated DrinkingWater

1 Department of Chemistry and Chemical Biology, McMaster University, 1280 Main Street West, Hamilton, ON, Canada L8S 4M1 2 Laboratory Services Branch, Ministry of the Environment (MOE), 125 Resources Road, Toronto, ON, Canada M9P 3V6 3 School of Life Sciences, University of Bradford, Richmond Road, Bradford BD7 1DP, UK 4 Environmental Innovations Branch, Ministry of the Enviroment 135 Street Clair Avenue West, 11th Floor, Toronto, ON, Canada M4V 1P5


Introduction
Since its inception in the late 19th century, drinking water disinfection has been one of the most important advancements for public health.While disinfectants such as chlorine, ozone, chloramines, and chlorine dioxide are used to kill harmful microorganisms, an unintended consequence is the formation of the so-called disinfection by-products (DBPs), which arise from the degradation of natural organic matter by the disinfectants [1,2].It is now widely known that DBPs, such as the two most common classes of DBPs, the trihalomethanes (THMs) and the haloacetic acids (HAAs), are associated with long-term health risks [3,4].
The Ontario Ministry of the Environment (MOE) has been monitoring raw and treated drinking water as part of the Drinking Water Surveillance Program (DWSP) since 1986.Target compound analyses include THMs and HAAs.
To complement these target compound analyses, gas chromatography-mass spectrometry (GC-MS) is routinely used to characterize a broad range of organic compounds, including DBPs, which may also be present in drinking water but whose identity has not yet been established.
More than ten years ago, two unexpected disinfection byproducts, a chloro compound labelled DBP-A, and a bromo analogue DBP-B, were being detected.Their electron ionization (EI) mass spectra were not available in the NIST98 database.On the basis of various mass spectrometric experiments [5], DBP-A and DBP-B were tentatively identified as the halogenated aminoxyalcohols of Scheme 1.However, the dearth of information [6] available on the dissociation characteristics of ionized aminoxyalcohols made such a proposal rather speculative, especially since the proposed structures would not be expected to be stable.This prompted our subsequent experimental and computational studies [7,8] of International Journal of Spectroscopy Br Cl 1-Aminoxy-1-chlorobutan-2-ol 1-Aminoxy-1-bromobutan-2-ol Scheme 1: Tentative structure proposals from [5] for DBP-A and DBP-B.
the dissociation characteristics of ionized 2-aminoxyethanol and its hydroxyamino isomer, HOCH 2 CH 2 ONH 2 and HOCH 2 CH 2 NHOH, respectively.Following the study of [5], the updated NIST02 database was released.This version of the database included the EI mass spectra of the halohydrins 4-chloro-2-methylbutan-2-ol and 3-bromo-2-methylbutan-2-ol shown in Scheme 2. The spectrum of the chloro compound appeared to be very close to that of DBP-A, while that of DBP-B could be that of 3-bromo-2-methylbutan-2-ol-a bromohydrin detected in ozonated waters as reported in [9,10].
A major challenge associated with the interpretation of the EI mass spectra of DBP-A and DBP-B is that the spectra do not display a molecular ion.In the study of [5], complementary chemical ionization (CI) experiments, using NH 3 as the reagent gas, yielded peaks at m/z 140 and m/z 184 for DBP-A and DBP-B, respectively.These were assigned to the formation of protonated molecules [M + H + ].This seems to be at odds with the presence of the above halohydrins, whose [M + H + ] ions correspond with peaks at m/z 123 and m/z 167, respectively.
In this study, we aimed to determine whether the above DBPs have an "aminoxy" or a "halohydrin" structure by performing additional GC-MS experiments on treated water samples and by analyzing authentic samples of the halohydrins prepared via unambiguous synthetic procedures based upon the early study of [11].

Experimental
Each 800 mL aliquot of water was spiked with the internal standard, d 10 -phenanthrene, and the surrogates, d 6 -N-nitrosodimethylamine (NDMA) and d 3 -2,4-dichlorophenol, used to monitor recoveries.The pH was adjusted to approximately 12 with NaOH solution, and the sample was serially extracted with CH 2 Cl 2 .HPLC-grade CH 2 Cl 2 was supplied by Caledon Laboratories (Georgetown, ON, Canada).According to the manufacturer's specification, this solvent is stabilized with ∼50 ppm of 2-methyl-2-butene (amylene).The pH of the aqueous phase was adjusted to approximately 2 with H 2 SO 4 , and the acidified water was serially extracted with CH 2 Cl 2 .Both sets of extracts were combined, dried over Na 2 SO 4 , concentrated, transferred to a 2 mL autosampler vial, and concentrated again to approximately 200 µL for GC-MS analysis.
The EI and CI GC-MS analysis of the water samples was performed using an Agilent 6890 Series GC coupled to a Micromass GCT Time-of-Flight Mass Spectrometer at McMaster University.For the CI experiments, ammonia was used as the reagent gas.Selected EI experiments were also Authentic samples of the halohydrins 3-and 4-chloro-2methylbutan-2-ol, 3-and 4-bromo-2-methylbutan-2-ol, and 1-chloro-2-methylbutan-2-ol were analyzed on the above instruments and also on the Varian 920FT GC-FTICR mass spectrometer at the MOE and the McMaster University ZAB-R instrument [12].
In general, the Grignard reactions were performed by adding an ethereal solution of the ketone or ester to a solution of the relevant alkylmagnesium halide under a nitrogen atmosphere at such a rate as to induce gentle reflux.After stirring overnight, sufficient saturated aqueous ammonium chloride (150 mL/mol of alkylmagnesium halide) was added dropwise to cause a clear solution to develop.The resultant solution was decanted, and the residual organic solids were extracted with diethyl ether (200 mL/mol alkylmagnesium halide).The combined organic ethereal solution was dried with MgSO 4 and then evaporated at reduced pressure to give the crude halohydrin as an amber oil.The halohydrins were purified by distillation at reduced pressure (∼35 mmHg).

DBP
Formula Mass (Calc.)Mass (GCT) [a]  Dev. [c]  Mass (FTICR) [b]  Dev.Note.Measurements obtained using [a] treated drinking water samples; [b] authentic standards of the halohydrins; the ion-guide of the FT-ICR instrument only permits ions with m/z > 65 to be transmitted to the cell; [c] the deviation from the calculated value is given in parts per million. 1H NMR indicated that the above materials were essentially pure.The EI mass spectrum of (DBP-A) see Figure 2(a), resembles that of 4-chloro-2-methylbutan-2-ol from the NIST02 database.Nevertheless, the aminoxy compound of Scheme 1 could yield a similar mass spectrum.This is because the spectrum of DBP-A does not display a molecular ion peak, but rather peaks at m/z 107 and 109, which could arise from either loss of CH 3 • from the proposed halohydrin (M = 122/124) or loss of NH 2 O • from the aminoxy structure (M = 139/141).In both scenarios, the elemental composition of the m/z 107 ion is expected to be C 4 H 8 ClO + , whose calculated mass (107.0258)matches the measured mass of Table 1.

Identification of the Disinfection By-Products
The conclusion that DBP-A is a halohydrin, rather than an aminoxy compound, follows from the ammonia CI mass spectrum of Figure 2(d).The spectrum shows sizeable peaks at m/z 140 and 142, which at first glance could be ascribed to the [M + H + ] ions of the aminoxy compound.However, ammonia chemical ionization of secondary and tertiary alcohols may also lead to the formation of [M + NH 4 + ] adduct ions [14].Therefore, the CI mass spectrum is also consistent with DBP-A having a molecular mass that matches that of the halohydrin 4-chloro-2-methylbutan-2-ol (M = 122/124).
The elemental compositions of the ammoniated halohydrin and the protonated aminoxy compound, that is, C 5 H 15 NOCl and C 4 H 11 NO 2 Cl, respectively, differ in mass by 0.0358 Da.Therefore a mass spectrometer operating with a minimum resolution of 4000 could establish the elemental composition of DBP-A.An instrument capable of such accurate mass measurements was not available in the MOE laboratory at the time [5] appeared.This prompted us to reexamine the putative DBPs using the Micromass GCT (time-of-flight) instrument at McMaster University, which is capable of a maximum resolution of 5000.
As shown in  Table 2: Accurate mass measurements (NH 3 CI mode) of the DBPs using GCT and FTICR instruments.DBP Formula Mass (calculated) Mass (GCT) [a]  Dev. [c]  Mass (FTICR) [b]  Dev. [ Note.Measurements obtained using [a] treated drinking water samples; [b] authentic standards of the halohydrins; [c] the deviation from the calculated value is given in parts per million.
The EI mass spectrum of Figure 2(a) is very close to the database spectrum of 4-chloro-2-methylbutan-2-ol, whose molecular ion, (CH 3 ) 2 C(OH)CH 2 CH 2 Cl •+ , readily dissociates into protonated acetone m/z 59 ions, CH 3 C(OH)CH 3 + , by direct bond cleavage.The prominent m/z 43 peak undoubtedly represents the acetyl cation, CH 3 C=O + [15].Tandem mass spectrometry experiments [15] performed using the ZAB-R instrument [12] indicate that this ion is likely formed by the route depicted in Scheme 3.
The other C 5 H 11 OCl isomer expected to show similar dissociation characteristics is the halohydrin 3-chloro-2methylbutan-2-ol of Scheme 2. It is also expected to dis-sociate into protonated acetone (m/z 59) ions by direct bond cleavage.Indeed, our EI mass spectra of the two isomers appear to be very similar.However, the 3-and 4-chlorohydrins have very different retention times (8.03 versus 10.15 minutes), and this makes their differentiation by GC-MS quite easy.These results leave little doubt that DBP-A is 3chloro-2-methylbutan-2-ol.
The halohydrin 3-bromo-2-methylbutan-2-ol yields the best library match to the EI spectrum of DBP-B shown in Figure 2(b).Here too, an ammonia CI experiment, (see Table 2) confirms that we are dealing with a halohydrin rather than an aminoxy compound.Authentic samples of the  International Journal of Spectroscopy Recognizing the need for accurate mass measurements for the structure analysis of unknown compounds, the Ministry of the Environment has acquired a Fourier transform ion cyclotron resonance (FTICR) mass spectrometer capable of a mass resolution of 100,000 on the capillary GC time scale.Tables 1 and 2 show results for authentic standards of DBP-A, B, and C obtained using the FTICR instrument.It is seen that the results obtained from both the GCT and FTICR instruments are in excellent agreement.

The Halohydrins May Not be Genuine DBPs, but Rather Laboratory
Artefacts.An important clue that the halohydrins reported in the study of [5] may not be genuine DBPs was the observation that the acid fraction of the liquid-liquid extraction procedure described in Section 2 contained the highest concentrations of these DBPs.This is contrary to the expectation that the halohydrins, which are not appreciably water soluble and neither acidic nor basic, would be more concentrated in the neutral fraction.Indeed, a small quantity (∼0.1 mg) of the labelled halohydrin (CD 3 ) 2 C(OH)CHClCH 3 , spiked into an 800 mL water sample, is not detected in the acid fraction but rather in the combined base/neutral fraction.This result implies that the generation of the purported DBPs is promoted by acid conditions and occurs during the extraction procedure [9,16].
As shown in Scheme 4, we propose that DBP-A is generated in the reaction of residual chlorine in the water sample (in the form of hypochlorous acid) with 2-methyl-2-butene, which is used as the stabilizer of the CH 2 Cl 2 used for extraction.The bromo analogue DBP-B may arise from a Cl/Br exchange reaction of DBP-A with naturally occurring Br − ions in the water [9] or by oxidation of bromide to bromine by excess chlorine and subsequent reaction of bromine and water with 2-methyl-2-butene.
To support this proposal, a control experiment was performed with a stock solution of CH 2 Cl 2 spiked with 50 ppm of the labelled 2-methyl-2-butene (CD 3 ) 2 C=CHCH 3 .Analysis of a water sample extracted with this spiked solvent indeed showed GC-MS signals of comparable intensity for labelled and unlabelled DBP-A as well as labelled and unlabelled DBP-B.
A complementary experiment, in which 5 ppm of NaOCl was added to the water sample, showed a 100-fold increase in yield of the chlorohydrins DBP-A and DBP-C.In contrast, the yield of DBPs was reduced 100-fold when 200 ppm of sodium thiosulfate was added to the treated drinking water sample to reduce any residual chlorine.This shows that the better part of the halohydrins in our samples are laboratory artefacts, but trace quantities may be genuine DBPs [9].

Conclusions
The present study leaves little doubt that DBP-A does not have the previously proposed "aminoxy" structure.It shows that DBP-A is the halohydrin 3-chloro-2-methylbutan-2-ol and that DBP-B is its bromo analogue.The EI mass spectra of the 3-chloro and 4-chloro-isomers and their bromo analogues are closely similar, but their retention times are very different.We propose that DBP-C is the isomeric halohydrin 1-chloro-2-methylbutan-2-ol.This study also shows that halohydrins are not necessarily genuine disinfection by-products.They can also be laboratory artefacts generated by the reaction of residual chlorine in the water with the 2-methyl-2-butene, a stabilizer in the CH 2 Cl 2 extraction solvent.The interference of these artefacts can be minimized by adding sodium thiosulfate to the aliquots of drinking-water that are being investigated in the monitoring and testing process.
DBP-A, DBP-B, and DBP-C.The total ion chromatogram (TIC) of the treated drinking water sample (Figure1) shows the disinfection byproduct peak DBP-A eluting at a retention time of 8.03 minutes.Its bromo analogue (DBP-B), along with several other halogenated species (DBP-C to DBP-H), is also present at levels ranging from ca. 10-100 ppb.In this study, we will focus on the structure analysis of DBPs A, B, and C.

Figure 1 :
Figure 1: Total ion chromatogram (TIC) of a treated drinking water extract.

Figure 2 :
Figure 2: EI mass spectra of (a) DBP-A, (b) DBP-B, (c) DBP-C, and (d) the CI mass spectrum of DBP-A obtained on the Micromass GCT instrument.

Scheme 4 :
Scheme 4: The halohydrins are generated in the reaction of residual chlorine in the water sample with the stabilizer (2-methyl-2-butene) of the CH 2 Cl 2 .

Table 1 :
Accurate mass measurements (EI mode) of the DBPs using GCT and FTICR instruments.

Table 2 ,
the measured mass of 140.08465 is consistent with ammoniated halohydrin ion C 5 H 15 NO 35 Cl but not with the protonated aminoxy compound C 4 H 11 NO 2 35 Cl.The consecutive losses of H 2 O and NH 3 from the m/z 140 ion account for the presence of the fairly intense m/z 105 peak in the CI mass spectrum.These results indicate that DBP-A is a halohydrin with a molecular mass of 122/124 and the elemental composition C 5 H 11 OCl.