Implications of treating water containing polynuclear aromatic hydrocarbons with chlorine: a gas chromatographic-mass spectrometric study.

The products of aqueous chlorination reactions of 1-methylnaphthalene, fluorene, dibenzofuran, anthracene, phenanthrene, 1-methylphenanthrene, fluoranthene, and pyrene have been determined. The conditions employed for these reactions approximated those that might be encountered in water treatment facilities. Reactions at pH greater than 6 tended to produce oxygenated products (epoxides, phenols, quinones, etc.), and reactions at pH less than 6 tended to produce both oxygenated (quinones) and chlorinated products.

The use of chlorination as the predominant technique for water renovation and disinfection has been questioned because of the reaction of active chlorine species with organic compounds present in the water to form products which may be biologically harmful (1)(2)(3). These reactions may occur at the site of chlorine addition as well as throughout a water distribution system (4). Polynuclear aromatic hydrocarbons (PAH) have been suggested as the precursors to at least a portion of the mutagens produced in some chlorination processes (4). The levels of these ubiquitous PAH compounds may be increased by the presence of coal tar coatings inside pipes and water storage tanks (5).
In order to provide further insight into the possible role of PAH in chlorine-induced mutagen formation, this laboratory has continued the study of the aqueous chlorination chemistry of PAH (6,7). The present report describes a detailed study of the product distributions of several PAH compounds which were chosen for study based on their previous identification in drinking water (8). A *Department of Chemistry, University of Minnesota, Duluth, Minnesota 55812. related study in these laboratories of the reaction kinetics and mechanisms for the aqueous chlorination of phenanthrene, fluorene, and fluoranthene will be reported elsewhere.
Experimental Procedure Twelve liters of water were treated with chlorine gas, and an appropriate amount of sodium hydroxide was added to achieve the desired pH. A solution of the PAH in acetonitrile (40-800 ml, specific conditions in Tables) was added to initiate the reaction which was monitored by HPLC. For high pH solutions, the pH was maintained by periodic addition of NaOH. To this, 2.5-3 equivalents (based on the total chlorine concentration) of dimethyl sulfoxide or sodium thiosulfate were added to terminate the reaction. In some cases, the reactant solution was acidified with sulfuric acid in order to suppress ionization of acidic products. The solution was then forced through two 7 x 50 mm stainless steel adsorption columns connected in series. The first column contained C-18 Porasil B (37-75 ,um, Waters Asso-ciates), and the second contained XAD-2 (100 ,um, Rohm and Haas). Reaction products which were adsorbed were later removed by elution of the column train with acetonitrile and then with methylene chloride. Some separation of these products was observed during the elution process. The various fractions collected were then checked for product content by HPLC and concentrated under a nitrogen stream prior to analysis by gas chromatography-mass spectrometry (GC/MS). Portions of fractions suspected of containing acidic compounds were methylated with diazomethane.

Instruments and Apparatus
The gas chromatograph-mass spectrometer (GC-MS) was a Hewlett-Packard 5993B quadrupole, EI-70 eV equipped with an Avondale B capillary inlet and a 21 MX-E computer (ANSWER software). The GC/MS interface was modified to allow the entire effluent from a fused silica capillary columns to be drawn into the mass spectrometer. The column was either a 25 m (0.32 mm ID) SE-54 (siloxane, Hewlett-Packard) or a 15m (0.32 ID) polymethyl (5% phenyl) siloxane (J & W Scientific). Splitless injections were employed with 5-8 psi head pressure and 2 ml/min He carrier flow. A typical temperature program was 70 to 310°C at 10°C/min. The mass range scanned was 50 to 400 amu with a scan time of 1.6 sec.
The reversed-phase high performance liquid chromatography (HPLC) apparatus consisted of a Perkin-Elmer Series 3 microprocessor-controlled gradient system equipped with a Rheodyne 7105 injector. The analytical column was a Perkin-Elmer C-18 reversed-phase column, 0.26 x 25 cm (P.N. 089-0716). The column was surrounded by a glass jacket and the temperature was maintained at 27°C with a refrigerated/heated bath and circulator to insure reproducibility. A Waters Associates Model 440 fixed wavelength detector and a Perkin-Elmer LC-75 variable wavelength ultraviolet detector were connected in series. Data was acquired on a Waters Associates Data Module. In order to improve the useful sensitivity of this system in an in-line trap column was used. The trap column consisted of two 60 x 0.7 cm stainless steel columns packed with 37-75 ,um C-18 Porasil B (Waters Assoc.) and was located between the water pump and the solventmixing coil. Acetonitrile was of HPLC grade (Fisher Scientific) and the water was obtained from a Millipore Corporation purification system (Milli-Q) fitted with an additional Continental Model 2021 canister. 75 Results and Discussion

Products
The products observed are listed in Tables 1-8 along with their mass spectral data. In most cases the product distributions were determined after the PAH ( -10O M) was allowed to react with -1O M chlorine for five to seven half lives (specific conditions are included in the tables). A small amount of acetonitrile (1 to 8%) was used in many reactions to provide sufficient quantities of PAH products for analysis and to minimize the formation of PAH crystals during a reaction. The use of such small amounts of this organic solvent did not appear to alter the product distribution based on chromatographic comparison with reactions in the absence of acetonitrile. In contrast, reaction solutions containing crystalline PAH material tended to result in the formation of only chloro products rather than both the chloro and oxygenated products observed in homogeneous solutions.
Reactions at high pH tended to produce oxygenated products while those at low pH tended to give both oxygenated and chlorinated products. The predominant product types included mono-and dichlorosubstituted compounds, quinones, phenols, carboxylic acids, and an epoxide. Minor compound types included chlorohydrins and dihydrodiols. A detailed mechanistic investigation ofphenanthrene suggested the formation of an arene oxide as the key intermediate in interpreting the product distribution.
Free-radical substitution on the methyl groups of 1-methylphenanthrene and 1-methylnaphthalene and the benzylic carbon of fluorene was not extensive. This type of reaction which results in the formation of benzyl halides and alcohols has previously been reported for some chlorination reactions (31). Benzyl alcohols can be distinguished from ring-substituted 'Relative retention time and mass spectrum identical to those of an authentic standard.   dRelative retention time and mass spectrum are identical to those of an authentic standard.  'HPLC retention time and A254/A280 identical to those of an authentic standard. 'Calculated from absorbance data at 254 and 280 nm.
gRelative retention time and mass spectrum are identical to those of an authentic standard. hProposed structure.
1Derivatized with diazomethane prior to GC/MS analysis.    dRelative retention time and mass spectrum are identical to those of an authentic standard.
eDetermined by HPLC utilizing an authentic compound as a standard. 7% unreacted fluoranthene.
fDerivatized with diazomethane prior to GO/MS analysis. isomers on the basis of prominent M-OH, M-1, M-2 and M-3 peaks in their mass spectra (32). In contrast, 1-chloromethylnaphthalene exhibits a mass spectrum which is nearly identical to that of its ring substituted isomer, 1-chloro-4-methylnaphthalene. Fortunately, a difference in relative gas chromatograph (GC) retention time (1.45 versus 1.38, respectively) allowed the conclusion that the former methyl-substituted halide was not produced in either 1-methylnaphthalene experiment. However, it is not possible to ascertain from the present data whether any of the other chloromethyl derivatives of naphthalene or phenanthrene were formed. A hydroxymethyl group is probably present in only one trace product of methylnaphthalene (Table 1, M+ = 222).

Biological Implications
An assessment of the biological implications of release of these chlorination products into the environment and into drinking water must, at present, depend upon scattered literature reports of various screening tests on individual compounds. These reports indicate that nearly all of the compound types produced in the chlorination reactions of PAH have the potential for causing adverse biological response. However, it is not possible to accurately predict activity of untested products because the nature of the various substituents and their locations on the aromatic ring may drastically affect the biological activity (33).
All of the nonchlorine-containing products of phenanthrene (except diphenic acid) shown in Table 5 have been tested for mutagenic activity by the reversion of histidine-dependent Salmonella typhimurium and the rec assay with Bacillus subtilis (34,35). The strongest mutagenic activity was observed with phenanthrene-9, 10-oxide and 9-phenanthrenol, but this activity was much less than that of the control, benzo[a]pyrene-4,5-oxide. Lack of significant mutagenicity and tumorigenicity of the epoxide and trans-dihydrodiol of phenanthrene was reported by other laboratories (36). The 1,6-and 1,8-pyrenediones have been reported to be mutagens and enhancing agents to other mutagens (37). Anthraquinone shows little or no activity (38), but certain derivatives (e.g., phenols) are quite mutagenic (39). Quinones of benzo(a)pyrene are mutagenic and the Ames assay without microsomal activation (40).
Phenols may be carcinogenic, mutagenic and toxic (18) and may bind to DNA, RNA and proteins (41). Chlorophenols may also exhibit similar activity (42). 2-Naphthol is known to affect the function of renal tubules (43), and 1-naphthol is a reported mutagen (38). Chlorinated PAH s (e.g., chloiinated dibenzofurans) may be extremely toxic (44,45). Chloromethyl- (46) and hydroxymethyl-substituted PAHs (47) may also have mutagenic and carcinogenic activity. Support for this research was generously provided as a cooperative agreement (CR806892) with the Health Effects Research Laboratory (HERL) -Cincinnati. We also wish to express our appreciation to Dr. F. Kopfler (HERL) for providing helpful guidance during the course of this research. It should also be noted that several preliminary experiments were also performed as part of the Master's thesis of K. J. Welch (1979).