Introduction

Of the tens of thousands of synthetic chemicals in current production (1) or generated by their environmental transformations, which merit concern as potential environmental pollutants? To qualify as a contaminant of concern, a compound must possess three attributes: appreciable amounts must be emitted to (or generated within) the environment; it must have at least a modicum of persistence; and the compound must exert toxic or other deleterious effects (directly or indirectly) on organisms. Compounds that satisfy all three criteria, if not already regulated or viewed as priority pollutants, should be regarded as candidates for environmental monitoring studies.

Two of the most widely used classes of agricultural herbicides in the U.S. are the chloroacetamides and the triazines. For example, in 1997, approximately 113-130 million lbs active ingredient (AI) of the more common chloroacetamide herbicides (alachlor, metolachlor, acetochlor, and dimethenamid) and 98-111 million lbs AI of the triazines (atrazine, cyanazine, and simazine) were applied to crops (2). Their extensive past or present use contributes to their prevalence as environmental contaminants in groundwater and surface water (3-9).

One important difference between these two classes of herbicides is that substantially larger fractions of chloroacetamides undergo transformation in soil than is the case with the triazines; this is reflected in the environmental occurrence of parent herbicides and degradates in the U.S. Although atrazine and metolachlor have been applied in roughly comparable amounts (75-82 million lbs AI for atrazine versus 63-69 million lbs AI for metolachlor in 1997; ref 2) as pre-emergent herbicides (and they may even be introduced as combined formulations), atrazine is typically reported at higher concentrations in the environment. For example, samples of groundwater and surface water analyzed in 1998 by the U.S. Geological Survey reveal a median concentration of 3.97 g/L for atrazine and 1.73 g/L for metolachlor (6). Atrazine generates a limited number of environmental degradates, mainly the dealkylated products and hydroxy analogues (10). The atrazine degradates tend to be less frequently detected than the parent herbicide (11) and have been reported at lower concentrations (8), although the relative abundance of atrazine degradates to parent compound varies on a seasonal basis, and total concentra tions of atrazine degradates may at times exceed that of atrazine itself (10). The chloroacetamides generate a much wider array of transformation products (12-33), at least in laboratory systems; degradation frequently involves modification of the chloroacetamide functional group or the N-(alkoxy)alkyl side chain. Although precise estimates of generation of chloroacetamide degradates are not available, two commonly studied classes of chloroacetamide degradates, the ionic ethane sulfonic acid (ESA) and oxanilic acid (OA) derivatives, are typically encountered at concentrations much higher than the parent herbicides (11, 34). This indicates that a substantial fraction of the parent chloroacetamide undergoes transformation within the environ ment.

Once transported to oligotrophic environments, chloroacetamide degradates are likely to be relatively persistent. Although the parent chloroacetamides may be rapidly transformed under idealized laboratory conditions, they are not readily mineralized (16, 28, 30). Recent data indicate that the principal hydrolysis products that form themselves undergo further hydrolysis at best slowly (33).

As with many pesticide transformation products (35), some neutral chloroacetamide derivatives have been shown to possess toxic attributes (36). Two alachlor degradates, 2-hydroxy-2',6'-diethylacetanilide and 2-chloro-2',6'-diethylacetanilide, have been shown to be mutagenic (37, 38), and 2-chloro-2',6'-diethylacetanilide can form DNA adducts (39, 40). Both 2-ethyl-6-methylaniline and 2,6-diethylaniline are promutagens (41) and are more teratogenic (42) than their parent compounds (metolachlor and alachlor, respectively). Other aniline metabolites, such as 2,6-dialkylbenzoquinoneimine and 2-ethyl-6-methylbenzonquinoneimine, are viewed as genotoxic (38). Hydroxyalachlor [2-hydroxy-2',6'-diethyl-N-(methoxymethyl)acetanilide] has been shown by the Microtox assay to possess a toxicity similar to that of alachlor (43). Metolachlor has been cited as an example of an agrochemical whose degradates are more toxic in Xenopus laevis bioassays than the parent compound (42). Carcinogenicity and acute toxicity may not be the only adverse outcomes associated with chloroacetamides (or their degradates); in humans, elevated levels of mercapturate metabolites of alachlor and metolachlor in urine have been correlated with poor semen quality (44).

Not all chloroacetamide degradates are likely to represent a substantial human health risk. In particular, evidence suggests that the extensively studied ionic ESA and OA derivatives are relatively nontoxic. Alachlor ESA, for example, is poorly absorbed, undergoes minor metabolism (45), and has a significantly lower subchronic and developmental toxicity than alachlor (45, 46). The Acetochlor Registration Partnership (ARP) website maintains that both the ESA and the OA degradates of acetochlor exhibit a low degree of toxicity to mammals and humans (47).

Current U.S. regulations focus on the parent herbicides, although some herbicide degradates have recently been recognized as potential toxicants. Within the European Union (E.U.), guidelines state that the sum of a pesticide and its relevant degradates should not exceed 0.1 g/L in water used for human consumption (48). While alachlor is the only chloroacetamide herbicide currently regulated in U.S. drink ing water (49), acetochlor, metolachlor, and "acetanilide pesticide degradation products" are included on the U.S. EPA Contaminant Candidate List (CCL; ref 50) of compounds that may be subject to future regulations. In the case of the triazines, the chlorinated dealkylated degradates exhibit toxicological properties that are similar to the parent herbicides, with which they are believed to share a common mode of action (51). Although atrazine and simazine are the only triazines regulated at present in U.S. drinking water (49), the U.S. EPA is considering future regulations of chlorinated triazines (including chlorinated degradates) as a group for purposes of cumulative risk assessment (51).

Neutral chloroacetamide degradates possess all three characteristics (appreciable release to the environment, moderate persistence, and potential toxicity) that are prerequisite to their consideration as "emerging contaminants". Yet neutral chloroacetamide degradates, unlike the ionic ESA and OA derivatives, have been infrequent targets of past environmental monitoring studies. The principal impedi ments to their monitoring lie in the lack of commercially available reference materials, and of validated methods for their analysis.

To date, the limited information that exists for neutral chloroacetamide herbicide degradates is mainly restricted to those degradates originating from alachlor. Several studies have found 2,6-diethylaniline in groundwater and surface water at concentrations of approximately 10-100 ng/L (4, 52). Perhaps the most comprehensive study of neutral chloroacetamide degradates was that of Potter and Carpenter (22), who investigated alachlor degradates in groundwater collected beneath a Massachusetts corn field using reference materials they synthesized themselves. These researchers were able to detect 20 alachlor degradates, of which six could be identified conclusively. Concentrations of individual degradates ranged from 4 to 570 ng/L, and the total concentration of degradates identified exceeded that of the parent alachlor (370-1100 ng/L) by a factor in excess of 2. One limitation of this work is that it relied on liquid:liquid extraction to concentrate the analytes, an approach that suffered from poor recoveries (12-20%).

The primary objective of the present study was to develop a robust yet sensitive method for analyzing neutral chloroacetamide degradates using solid-phase extraction (SPE) and gas chromatography/mass spectrometry (GC/MS). The present investigation therefore complements our ongoing survey of neutral chloroacetamide degradates in raw and finished drinking water from the Midwestern U.S. This study emphasizes determination of SPE recoveries in deionized water and in natural water to assess the magnitude of potential interferences that might be introduced by the presence of natural organic matter. Additional objectives include an exploration of the suitability of preservatives currently being used in our survey of chloroacetamide degradates in drinking water, particularly a study of their influence on analyte recoveries and their ability to minimize changes in analyte concentration after several weeks of refrigeration.

The degradates chosen for this study were selected on the basis of their occurrence in laboratory transformation studies, although the availability of synthetic routes also played a role in target analyte selection. Seventeen of the 20 neutral chloroacetamide degradates investigated herein were synthesized specifically for this and related studies. The structures of the parent herbicides and the target degradates, along with the numbering scheme used to designate each analyte, are provided in Figure 1. Triazine herbicides (and the desethyl atrazine and desisopropyl atrazine degradates), as well as the ESA and OA chloroacetamide derivatives, are included in our ongoing work because of the wealth of prior studies of these compounds; their inclusion facilitates comparison of the present findings to previous work.

Figure 1 Structures of parent herbicides and degradates under investigation. Structures I-X represent alachlor and its degradates, structures XI-XIX represent metolachlor and its degradates, structures XX-XXIV represent acetochlor and its degradates, structures XXV-XXVIII are those degradates that can result from either metolachlor or acetochlor, structures XXIX-XXX represent dimethenamid and its degradate, and structures XXXI-XXXV are triazine herbicides and their degradates.

A final objective of the present study was to apply the method developed herein to the analysis of water samples from the upper Chesapeake Bay. This locale receives freshwater inflow from the Susquehanna River watershed, a region in which chloroacetamide and triazine herbicides are extensively used (53). Peak herbicide input from the Susquehanna River to this estuary occurs in late June and early July (54); our samples were obtained in July to coincide with the latter portion of this period. A depth profile was obtained near the Bay Bridge, a location that is known to possess a redox gradient (55) that might be reflected by changes in the distribution of degradates. The mean hydraulic residence time of freshwater in the Bay at the location sampled is approximately 20 days (56), although the actual residence time varies with depth because the system is not well-mixed vertically. The present study may also therefore provide at least a preliminary indication as to the persistence of the target analytes in coastal waters.

Experimental Section

Reference Standards. Alachlor (I) [15972-60-8; 2-chloro-2',6'-diethyl-N-(methoxymethyl)acetanilide], metolachlor (XI) [51218-45-2; 2-chloro-2'-ethyl-6'-methyl-N-(2-methoxy-1-methylethyl)acetanilide], acetochlor (XX) [34256-82-1; 2-chloro-2'-ethyl-6'-methyl-N-(ethoxymethyl)acetanilide], dimethenamid (XXIX) [87674-68-8; 2-chloro-N-(2,4-dimethyl-3-thienyl)-N-(2-methoxy-1-methylethyl)acetamide], atrazine (XXXI) [1912-24-9; 2-chloro-4-(ethylamino)-6-isopropylamino-s-triazine], desethyl atrazine (XXXII) [6190-65-4; 2-chloro-4-amino-6-isopropylamino-s-triazine], desisopropyl atrazine (XXXIII) [1007-28-9; 2-chloro-4-(ethylamino)-6-amino-s-triazine], simazine (XXXIV) [122-34-9; 2-chloro-4,6-diethylamino-s-triazine], cyanazine (XXXV) [21725-46-2; 2-chloro-4-(ethylamino)-6-methylpropionitrileamino-s-triazine], and deschloroacetylmetolachlor propanol (XVII) [2-[(2-ethyl-6-methylphenyl)amino]-1-propanol] were purchased from Chem Service (West Chester, PA). 2,6-Diethylaniline (VIII) [579-66-8] and 2-ethyl-6-methylaniline (XXVIII) [24549-06-2] were obtained from Aldrich. Alachlor OA (IX) [2-[(2,6-diethylphenyl)(methoxymethyl) amino]-2-oxoacetic acid], alachlor ESA (X) [2-[(2,6-diethylphenyl)(methoxymethyl) amino]-2-oxoethanesulfonic acid], and acetochlor OA (XXIII) [2-[(2-ethyl-6-methylphenyl)(ethoxymethyl)amino]-2-oxoacetic acid] were donated by Monsanto (St. Louis, MO). Metolachlor OA (XVIII) [2-[(2-ethyl-6-methylphenyl)(2-methoxy-1-methylethyl)amino]-2-oxoacetic acid] and metolachlor ESA (XIX) [2-[(2-ethyl-6-methylphenyl)(2-methoxy-1-methylethyl)amino]-2-oxoethanesulfonic acid] were donated by Syngenta (Greensboro, NC). Acetochlor ESA (XXIV) [2-[(2-ethyl-6-methylphenyl)(ethoxymethyl)amino]-2-oxoethanesulfonic acid] was obtained from the EPA National Standard Pesticide Repository (Ft. Meade, MD).

The following compounds were synthesized in our laboratory: hydroxyalachlor (II) [2-hydroxy-2',6'-diethyl-N-(methoxymethyl)acetanilide], deschloroalachlor (III) [2',6'-diethyl-N-(methoxymethyl)acetanilide], 2-chloro-2'-6'-diethylacetanilide (IV), 2-hydroxy-2'-6'-diethylacetanilide (V), 2-hydroxy-2'-6'-diethyl-N-methylacetanilide (VI), 2'-6'-diethylacetanilide (VII) [16665-89-7], hydroxymetolachlor (XII) [2-hydroxy-2'-ethyl-6'-methyl-N-(2-methoxy-1-methyl-ethyl)acetanilide], deschlorometolachlor (XIII) [2'-ethyl-6'-methyl-N-(2-methoxy-1-methylethyl)acetanilide], a morpholinone derivative of metolachlor (XIV) [4-(2-ethyl-6-methylphenyl)-5-methyl-3-morpholinone], metolachlor propanol (XV) [2-chloro-2'-ethyl-6'-methyl-N-(2-hydroxy-1-methyl-ethyl)acetanilide], deschloroacetylmetolachlor (XVI) [2'-ethyl-6'-methyl-N-(2-methoxy-1-methylethyl)aniline], hydroxyacetochlor (XXI) [2-hydroxy-2'-ethyl-6'-methyl-N-(ethoxymethyl)acetanilide], deschloroacetochlor (XXII) [2'-ethyl-6'-methyl-N-(ethoxymethyl)acetanilide], 2-chloro-2'-ethyl-6'-methylacetanilide (XXV), 2-hydroxy-2'-ethyl-6'-methylacetanilide (XXVI), 2'-ethyl-6'-methylacetanilide (XXVII), and deschlorodimethenamid (XXX) [N-(2,4-dimethyl-3-thienyl)-N-(2-methoxy-1-methylethyl)acetamide]. Trivial names for chloroacetamide and triazine degradates are provided herein for ease of reference; these typically indicate changes in structure relative to the parent herbicide. Synthesis procedures and both ¹H NMR (proton nuclear magnetic resonance) spectral data and electron ionization mass spectra can be found in the Supporting Information.

Figure 2 Water quality parameters determined from the upper Chesapeake Bay on July 17, 2003 between 10 am and 12 pm on the outgoing tide. Conductivity, temperature, pH, and salinity were measured using an in situ probe. Dissolved oxygen and chloride were determined by the Winkler method and by ion chromatography, respectively. Vertical dashed line represents the detection limit for dissolved oxygen analysis.

Figure 3 Depth profiles indicating concentrations of alachlor and several of its neutral degradates within the upper Chesapeake Bay. Symbols represent measured concentrations (with error bars indicating standard deviation obtained in triplicate analyses), without corrections for recoveries. Solid lines represent the expected concentrations if the shallowest sample were diluted by mixing with the appropriate amount of seawater (inferred from chloride measurements). The vertical dashed lines represent the MRLs for each compound. Open symbols represent those measurements below the MRL but above the MDL.

Surrogate standards for the neutral compounds were ring labeled ¹³C₆-metolachlor and ¹³C₃-atrazine, obtained from Cambridge Isotope Labs (Andover, MA); the surrogate for the ionic compounds was 2-benzoylbenzoic acid, obtained from Aldrich (Milwaukee, WI). Internal standards for compounds analyzed via GC/MS were acenaphthene-d₁₀, anthracene-d₁₀, and chrysene-d₁₂, obtained from Cambridge Isotope Labs. The internal standard for the ESA degradates, which were analyzed via high-pressure liquid chromatog raphy with diode array detection (HPLC-DAD), was 2,4-dichlorophenylacetic acid (Aldrich).

Initial Recovery Studies. Initial recovery studies, the purpose of which was to determine the suitability of the SPE media employed, were conducted using deionized water as a sample matrix. In these studies, water was spiked with parent herbicides, neutral degradates, and the ionic degradates (i.e., the ESA and OA derivatives), using acetone (Ultra-Resi Analyzed from J.T. Baker; Phillipsburg, NJ) as a carrier solvent. Additional samples for recovery studies were prepared in deionized water that contained preservatives. The preservatives chosen were recommended by the U.S. EPA for drinking water analysis of organic compounds on the Contaminant Candidate List, including acetochlor (57, 58). These consisted of ascorbic acid (0.57 mM; to reduce residual chlorine), ethylenediaminetetraacetic acid trisodium salt (0.98 mM; to chelate metal ions that might catalyze hydrolysis), diazolidinyl urea (3.6 mM; as a microbial inhibitor), and a pH 7 buffer of tris(hydroxymethyl)aminomethane and tris(hydroxymethyl)aminomethane hydrochloride (50 mM total). All preservatives were ACS reagent grade obtained from Aldrich. SPE recoveries of the ionic degradates were found in preliminary studies to be greatly suppressed by the presence of preservatives. Subsequent analyses of samples for ionic compounds were conducted by extracting unpreserved samples as soon as possible after collection.

Recovery and Preservative Studies in Natural Waters. Additional studies were conducted to investigate whether the more complex matrix of natural waters could affect SPE recoveries, as well as to test the efficacy of the preservatives used in samples containing neutral analytes. Water used in SPE recovery and preservative studies was obtained from a local surface water impoundment, Loch Raven Reservoir (Towson, MD), on March 23, 2003. Samples were collected in two cleaned 4-L amber bottles, one containing preserva tives and one without. The water sample containing preservatives was spiked with parent herbicides and neutral degradates. The second sample (without preservatives) was spiked with the ionic chloroacetamide degradates. Samples were then filtered through a 0.7 m glass fiber filter (Millipore; Billerica, MA). Blank samples (no added herbicides or degradates) were also filtered in an identical manner prior to SPE. Half of the herbicide-spiked water containing preservatives was extracted immediately, and the other half was maintained at 4 C for 21 days prior to SPE for reanalysis of parent compounds and neutral degradates. The samples without preservatives were extracted immediately.

Chesapeake Bay Samples. Water samples from the upper Chesapeake Bay were collected on July 17, 2003. Samples were collected at 11 different depths (2, 5, 8, 10.5, 11, 11.5, 13.5, 18, 19, 20, and 21 m) from a single location near the Bay Bridge, geographic coordinates 3859'51' 'N, 7621'44' 'W, between 10 am and 12 pm (with the outgoing tide). Sample collection began with the deepest sample (anticipated to contain relatively low concentrations of parent herbicides and degradates), and concluded with the shallowest sample. In situ measurements of temperature, pH, and electrical conductivity were obtained with a model 63 YSI meter (Yellow Springs, OH) and a 100 ft cable. Electrical conductivity profiles measured at the beginning and at the end of sampling revealed only slight changes over the 2-h sampling period. Samples for analysis of herbicides and herbicide degradates were collected in 2-L amber bottles using a Solinst (Georgetown, Ontario) model 410 peristaltic pump connected to Teflon-lined tubing; additional samples, for analysis of dissolved oxygen, were collected in 300-mL glass biological oxygen demand (BOD) bottles. Weights were added to the end of the Teflon tubing to maintain its vertical orientation. All samples were collected in-line using Teflon stopcocks and glass tubing connected to the sample bottles to prevent the samples from contacting the silicone tubing at the peristaltic pump head. The pump flow rate was approximately 200 mL/min. The BOD bottles were allowed to overflow three times before stoppering to minimize contamination with atmospheric oxygen. The 2-L bottles used for herbicide analysis did not contain sample preservatives; the samples were extracted as soon as possible (within 7 days) upon returning to the laboratory. All samples were kept on ice until returning to Johns Hopkins University and were subsequently stored in a cold room at 4 C until analyses or extractions could be completed. Dissolved oxygen was determined via the Winkler method (59). A portion of the water from the 2-L bottles was analyzed for chloride using a Dionex (Sunnyvale, CA) DX-120 Ion Chromatograph with AS14 Ion Pac column (4 × 250 mm). Dissolved oxygen analyses were completed within 8 h, and chloride analyses took place within 7 days.

Upon returning to the laboratory, surrogate standards were added to the samples to provide concentrations of 50 ng/L (for neutral surrogates) and 200 ng/L (for the ionic surrogate standard). From each 2-L bottle, the water was divided accordingly: three 300-mL samples at every depth sampled for triplicate analyses of neutral degradates; two 500-mL samples (at depths of 8, 11, 13.5, 18, 19, and 21 m) for duplicate analyses of ionic degradates; one 300-mL sample (at 2, 5, 10.5, 11.5, and 20 m) was fortified (at 200-500 ng/L) with the neutral compounds; and one 500-mL sample (at depths of 2, 5, 10.5, 11.5, and 20 m) was fortified (at 500 ng/L) with the ionic compounds. The water samples were filtered with a 0.7 m glass fiber filter after addition of surrogates or fortification but prior to extraction.

Additional quality control was maintained by analyzing laboratory blanks and laboratory fortified blanks according to procedures outlined in EPA Method 526 (60). Laboratory blanks consisted of deionized water. Laboratory fortified blanks contained deionized water spiked with neutral (5-500 ng/L) or ionic (100-500 ng/L) analytes. The laboratory blanks, laboratory fortified blanks, and laboratory fortified field samples were handled in the same manner as the field samples.

Sample Extraction and Derivatization. All solvents used in sample extraction and derivatization steps were Ultra Resi-Analyzed from J.T. Baker. All extractions were performed using a Gilson (Middleton, WI) ASPEC XL solid-phase extraction system. For the neutral analytes, extractions were performed with Oasis HLB (6 mL, 200 mg) cartridges from Waters (Milford, MA). Each cartridge was conditioned with 6 mL of methanol and 6 mL of deionized water. The samples (300 mL) were loaded onto the cartridge at 6 mL/min. After loading, the cartridges were again washed with 6 mL of deionized water and 6 mL of deionized water containing 15% methanol. Analytes were eluted successively with 3 mL of methanol and 3 mL of ethyl acetate at a flow rate of 2 mL/min. Care was taken to prevent the cartridges from going to dryness during sample loading and elution. The extracts were evaporated to incipient dryness under a gentle stream of nitrogen at ambient temperature. The final sample was brought up to 1 mL with toluene containing 50 g/L of each of the three internal standards.

For the ionic compounds, an SPE procedure outlined by Shoemaker (61) was followed. The SPE cartridges employed were Supelco Envi-Carb carbon (6 mL, 250 mg; Bellafonte, PA). Each cartridge was conditioned with 10 mL of 10 mM ammonium acetate in methanol and 25 mL of deionized water. The sample (500 mL) was loaded onto the cartridge at 6 mL/min. The cartridge was washed with 10 mL of deionized water. Ionic compounds were eluted with 6 mL of 10 mM ammonium acetate in methanol at a flow rate of 2 mL/min. Once again, care was taken to avoid allowing the cartridges to go to dryness during sample loading or elution steps. The extracts were split into two 3-mL samples. Both samples were evaporated to dryness under a gentle stream of nitrogen at ambient temperature. The first sample, for HPLC-DAD analysis of the ESA degradates, was reconstituted in 0.5 mL of 10 mM ammonium acetate in deionized water containing 1 mg/L internal standard. The other sample, for GC/MS analysis of the methyl ester derivatives of the OAs, was reconstituted into 0.5 mL of acetone for methylation using diazomethane. To the acetone fraction was added an aliquot (200 L-1 mL) of ether containing diazomethane until the yellow diazomethane color persisted for 15 min. The diazomethane was generated using an MNNG (N-methyl-N'-nitro-N-nitrosoguanidine)-diazomethane generator (Aldrich; note that diazomethane is explosive and that MNNG is mutagenic). The ether:acetone solvent was evaporated under nitrogen, and the sample was brought to 0.5 mL volume with toluene containing 50 g/L internal standards. To check for the possibility of naturally occurring methyl esters of the OA derivatives, underivatized SPE extracts were also analyzed via GC/MS; the OA methyl esters were never detected in any of our underivatized samples.

GC/MS Analysis. Injections of 1 L (splitless) or 100 L (large volume injections) were made onto a ThermoQuest (San Jose, CA) Trace 2000 gas chromatograph with a programmed temperature vaporization injector (PTV) coupled to a quadrupole mass spectrometer. A DB-35ms (Agilent; Palo Alto, CA) 30 m length × 0.25 mm ID × 0.25 m phase thickness column was used to effect separations. The GC temperature program for the neutral compounds was 90 C for 1 min, 6 C/min to 290 C, followed by a 5-min hold at 290 C; the temperature program for the OA methyl esters was similar, except that the ramp rate was increased to 10 C/min. For 1 L splitless injections, the PTV injector was maintained at 200 C. For 100 L injections, the PTV was programmed to introduce the sample at 110 C at a rate of 5 L/s. The evaporation occurred at a split flow rate of 100 mL/min for 0.5 min. The injector was programmed from 110 C (0.5 min) to 290 C (3.5 min) at 14 C/s with a splitless time of 2 min. The mass spectrometer temperature was set to 250 C, with an energy of 70 eV, and spectra were obtained in electron ionization (EI) mode with selected ion monitoring (SIM). The transfer line was maintained at 285 C. Data were collected using Xcalibur software.

	Figure 4 Depth profiles indicating concentrations of metolachlor and several of its neutral degradates within the upper Chesapeake Bay. Symbols represent measured concentrations (with error bars indicating standard deviation obtained in triplicate analyses), without corrections for recoveries. Solid lines represent the expected concentrations if the shallowest sample were diluted by mixing with the appropriate amount of seawater (inferred from chloride measurements). The vertical dashed lines represent the MRLs for each compound.
	Figure 5 Depth profiles indicating concentrations of acetochlor and two of its neutral degradates within the upper Chesapeake Bay. Symbols represent measured concentrations (with error bars indicating standard deviation obtained in triplicate analyses), without corrections for recoveries. Solid lines represent the expected concentrations if the shallowest sample were diluted by mixing with the appropriate amount of seawater (inferred from chloride measurements). The vertical dashed lines represent the MRLs for each compound. Open symbols represent those measurements below the MRL but above the MDL.
	Figure 6 Depth profiles indicating concentrations of three neutral degradates within the upper Chesapeake Bay that could be produced from either metolachlor or acetochlor. Symbols represent measured concentrations (with error bars indicating standard deviation obtained in triplicate analyses), without corrections for recoveries. Solid lines represent the expected concentrations if the shallowest sample were diluted by mixing with the appropriate amount of seawater (inferred from chloride measurements). The vertical dashed lines represent the MRLs for each compound.

HPLC-DAD Analysis. The ESAs were analyzed via HPLC-DAD following a method similar to that developed by Hostetler and Thurman (62). Injections of 100 L were made onto a Waters HPLC-DAD (1525 pump and 2996 photodiode array detector) with Empower data acquisition system. The analytical wavelength was 210 nm. The mobile phase was 60:35:05 (10 mM ammonium acetate in water:methanol:acetonitrile) with a flow rate of 0.6 mL/min. A Phenomenex (Torrance, CA) Luna C₁₈ 5 m, 250 mm × 4.6 mm column was used to separate ESA degradates. The column temper ature was set at 60 C using a Phenomenex TS-130 column heater. While complete separation of the ESA degradates is achieved with this method, the metolachlor OA peak (if present in sufficiently high concentrations; >1 g/L) overlaps with the acetochlor ESA and alachlor ESA peaks. In such a case, this method cannot be used to quantify the concentra tions of acetochlor ESA and alachlor ESA.

Method Detection Limits. The method detection limits (MDLs) were obtained using the approach outlined in EPA Method 526 (60). Instrument detection limits were obtained by injecting progressively more dilute aliquots of standard solutions until the relevant peaks could no longer be discerned from the background noise. For the neutral compounds, 14 (seven for splitless and seven for LVI) 300-mL deionized water samples (with preservatives) were fortified at a concentration (10-700 ng/L for splitless; 0.10-7 ng/L for LVI) that would give a signal of 2-5 times the instrument detection limit after accounting for the 300-fold SPE concentration. For the ionic compounds, seven 500-mL deionized water samples (without preservatives) were forti fied at a concentration (the OAs at 20 ng/L and the ESAs at 500 ng/L) that would give a signal of 2-5 times the instrument detection limit after accounting for the 500-fold SPE concentration. These samples were extracted as described above. The standard deviation of the concentration of these samples was calculated. The method detection limit (MDL) = S·t_{(n-1,1-=0.99)}, where S = standard deviation of replicate analyses, t_{(n-1,1-=0.99)} = Students t value for the 99% confidence level with n - 1 degrees of freedom, and n = number of replicates. The minimum reporting limit (MRL) was established as an analyte concentration that is 3 times the MDL.

Results and Discussion

Recovery Studies. SPE recoveries for neutral analytes, summarized in Table 1, were determined by spiking the target compounds (final concentration 3 g/L) either into deionized (DI) water containing preservatives or into Loch Raven water containing preservatives. Recoveries greater than 70% (as recommended by the U.S. EPA; ref 60) were obtained in deionized water for all analytes except desisopropyl atrazine. Recoveries in Loch Raven water (ranging between 62% and 107%), while somewhat lower for some compounds, were quite similar to those in deionized water and in most cases were not different at the 95% confidence level. Recoveries measured in fortified surface water after a 3-week storage test closely matched those obtained prior to storage, confirming the suitability of the preservatives for the neutral analytes.

For the ionic analytes, deionized water or Loch Raven water without preservatives was spiked at a final concentra tion of 3 g/L with target compounds to determine SPE recoveries; results are summarized in Table S1 in the Supporting Information. Recoveries in deionized water ranged from 76% to 96%, closely matching those obtained in Loch Raven water (78-98%). Recoveries for the ionic degradates were similar to those observed by previous researchers (61).

Method Detection Limits. Table 2 summarizes method detection limits (MDLs) for parent herbicides and neutral degradates, along with analytical parameters (retention times, quantitation ions, monitoring ions) used in GC/MS analysis. The MDLs for the ionic degradates can be found in Table S1 of the Supporting Information. Splitless injections of 1 L aliquots of SPE extracts produced MDLs of 4-14 ng/L for those neutral chloroacetamide degradates that did not possess a hydroxy substituent; MDLs for compounds with such a substituent ranged from 13 to 120 ng/L (and were generally near the upper end of this range). The increased MDLs for the hydroxy-substituted compounds reflect their poor chromatography (broader, tailing peaks) and extensive EI fragmentation.

Such problems could probably be resolved by replacing the active hydrogens through conventional derivatization techniques, at additional expense and with the risk that some derivatives might not be stable. Instead, we explored the utility of large volume injections (LVI) of 100 L in hopes of reducing the MDLs. This approach successfully reduced MDLs to 0.18-3.9 ng/L for the hydroxy-substituted degradates (0.06-0.24 ng/L for most other neutral degradates). On average, the MDL values decreased 70-fold on using 100 L injections. The two primary anilines proved too volatile for LVI using the approach employed for the other neutral analytes; very small peaks were detected when using toluene as the solvent. Hexane and ethyl acetate were also explored as solvents without success in LVI analysis of these anilines.

Analysis of the OAs by SPE followed by derivatization to the methyl esters, with separation and quantitation via GC/MS, produced MDLs of 7 ng/L (only splitless injections of 1 L were utilized). The ESA MDLs were 90-100 ng/L, more than an order of magnitude higher than for the oxanilic acids. The difference stems from the lower sensitivity of HPLC-DAD relative to GC/MS. Injections of volumes larger than 100 L were tested on the HPLC-DAD but resulted in poorer reproducibility.

Minimum reporting limits (MRLs), shown in Table 2, were established at concentrations 3 times the MDL values. These ranged from 0.17 to 12 ng/L for all compounds analyzed via LVI GC/MS. The anilines that were analyzed via splitless injections (1 L) and GC/MS had MRLs of 25-33 ng/L. The OA methyl esters yielded MRLs of 20 ng/L, and the ESAs had MRLs of approximately 300 ng/L (Table S1).

Herbicide and Herbicide Degradates in Chesapeake Bay Samples. Vertical profiles of electrical conductivity, tem perature, pH, salinity, dissolved oxygen, and chloride are shown in Figure 2; a tabular listing of the data for dissolved oxygen and chloride is also provided in Table S2 of the Supporting Information. The data reveal a gradual change in parameters within the upper 10-15 m of the water column. Below 12 m, all dissolved oxygen concentrations are less than 1 mg/L and approach zero at a depth of 20 m. Chloride concentration increases with depth through the upper 10-15 m of the water column as mixing with seawater occurs. The values of temperature, salinity, and dissolved oxygen are typical for this portion of the upper Chesapeake Bay in summer and closely match those obtained by other researchers at similar dates (55).

Depth profiles for each of the parent herbicides and selected degradates are shown in Figures 3-8; results are also summarized in Table S2 of the Supporting Information. Almost all of the analytes sought were detected in at least one sample. The sole exception was deschloroalachlor, which was never detected in Chesapeake Bay water. The primary anilines, 2,6-diethylaniline and 2-ethyl-6-methylaniline, were detected but were never present above the MRL values. None of the ESAs were detected, most likely because of their higher MRLs. Of the OAs, metolachlor OA was always above the MRL, while most measurements of alachlor OA and acetochlor OA yielded results between the MDL and the MRL values. Deschloroacetylmetolachlor, deschloroacetylmetolachlor propanol, 2'-ethyl-6'-methylacetanilide, 2-chloro-2'-ethyl-6'-methylacetanilide, 2-chloro-2'-6'-diethylacetanilide, deschlorodimethenamid, deschloroacetochlor, and hydroxyacetochlor, while recognized as chloroacetamide degradates in laboratory studies (16, 31-33), do not appear to have been previously reported in natural or affected environmental water samples.

Figure 7 Depth profiles indicating concentrations of dimethenamid and its neutral degradate within the upper Chesapeake Bay. Symbols represent measured concentrations (with error bars indicating standard deviation obtained in triplicate analyses), without cor rections for recoveries. Solid lines represent the expected concentrations if the shallowest sample were diluted by mixing with the appropriate amount of seawater (inferred from chloride measurements). The vertical dashed lines represent the MRLs for each compound.

Figure 8 Depth profiles indicating concentrations of chloro-s-triazine herbicides and selected neutral degradates within the upper Chesapeake Bay. Symbols represent measured concentrations (with error bars indicating standard deviation obtained in triplicate analyses), without corrections for recoveries. Solid lines represent the expected concentrations if the shallowest sample were diluted by mixing with the appropriate amount of seawater (inferred from chloride concentrations). The vertical dashed lines represent the MRLs for each compound.

Recoveries in laboratory fortified field samples were generally between 70% and 130% (with the exception of desisopropyl atrazine, for which recoveries were >60%) after correcting for amounts present prior to fortification. Recoveries in the fortified samples compared closely with those obtained in deionized water and Loch Raven water, indicative of the absence of a significant matrix effect. Average surrogate recovery for all samples was 92 (±3)% for ¹³C₆ metolachlor, 95 (±3)% for ¹³C₃ atrazine, and 64 (±2)% for 2-benzoylbenzoic acid.

Large volume injections proved useful in measuring compounds not otherwise detectable using splitless injec tions. Drawbacks of LVI included an elevated baseline in the Chesapeake Bay samples; we also found it necessary to run at least one blank solvent injection between samples to eliminate carryover.

Solid lines on each graph in Figures 3-8 represent the expected concentration of each analyte if mixing of the shallowest sample with additional seawater were to occur. This was computed by estimating the fraction of seawater present in each sample from the measured chloride concentration at the depth in question. Concentrations of each herbicide or herbicide degradate were assumed to be zero in 100% seawater.

Although the concentrations of most of the analytes decrease with increasing depth, the simple mixing calcula tions generally underpredict parent herbicide and degradate concentrations measured at greater depths. This could be interpreted to indicate that there is a substantial influx of these target compounds from rivers discharging farther down estuary. While the majority of herbicide input (~60%) into the Chesapeake Bay originates from the Susquehanna River (53), our sampling site lies up estuary from several other tributaries (including the Potomac and James Rivers) that also discharge significant amounts of herbicides to the Bay (53). It is conceivable that tidal action could transport herbicides and herbicide degradates to locations as distant as our sampling site.

Perhaps a more likely explanation is simply that this system is not at steady state. Our sampling probably occurred after the peak fluvial input of herbicides had passed. The relatively high concentrations we measure in deeper waters could simply reflect earlier diffusive mixing into deeper waters during the peak loading events into the Chesapeake Bay.

The relative abundances of parent herbicides and neutral degradates reveal little if any changes with depth, suggesting that the degradates are not rapidly formed in the water column within this portion of the upper Chesapeake Bay. A few degradates, such as deschloroacetylmetolachlor (Figure 4) and deschlorodimethenamid (Figure 7), appear to initially decrease in concentration with depth but then increase again in concentration at the greatest depths. Deschloroacetylmetolachlor has been previously hypothesized to result from the photolysis of metolachlor (31). Its generation by photolysis cannot explain the increase in concentration at the greater depths. Prior research has revealed that deschloroacetylmetolachlor can also form during the reaction of metolachlor with iron sulfide minerals, such as iron pyrite (33), that occur within the sediments underlying the Chesapeake Bay (63). The relative abundance of deschloroacetylmetolachlor does increase slightly relative to metolachlor and its other degradates with increasing depth. This could indicate that even though most degradates do not appear to form rapidly in situ, some compounds may be produced within the deeper portions of the water column or sediment porewaters.

Atrazine and its two chlorinated degradates were detected at each depth sampled (Figure 8). Atrazine concentrations were significantly greater than any of the chloroacetamide herbicides. Simazine and cyanazine were measured at concentrations significantly lower than atrazine (Figure 8). The two atrazine degradates, desethyl atrazine and desisopropyl atrazine (assuming that the latter is primarily gener ated from atrazine rather than from simazine), were encountered at concentrations that were lower than that of the parent herbicide.

Figure 9 summarizes the abundance of neutral degradates relative to their parent compound. Overall ratios were calculated by averaging ratios of degradate to parent measured at each depth sampled. The neutral degradates that could result from either metolachlor or acetochlor were attributed to metolachlor. Most of the chloroacetamide degradates are more abundant than the parent compounds, occurring in some cases at concentrations up to 10 times that of the parent herbicide. Total concentrations of the neutral degradates were on average 20 (alachlor and acetochlor)-30 (metolachlor) times that of the parent herbicide, with total concentrations of chlorinated degradates alone being 5 (alachlor)-15 (metolachlor) times that of the parent compounds. In contrast, the two atrazine degradates studied were less abundant than the parent compound, with the average total concentration of the measured degradates being 60% of the atrazine concentration. This may in part reflect our selection of atrazine degradates; other studies (10) have revealed that total concentrations of atrazine degradates can exceed that of the parent compound prior to the spring application period, with hydroxyatrazine (not included as one of our analytes) being the major degradate under such conditions.

Figure 9 Average ratio of concentrations of herbicide degradates to parent compounds in the upper Chesapeake Bay. Those compounds that could originate from either metolachlor or acetochlor were computed assuming metolachlor as the parent. The numbers correspond to the designation provided in Figure 1. Not included are the compounds analyzed via splitless GC/MS (2,6-diethylaniline, 2-ethyl-6-methylaniline, and the OAs) or HPLC-DAD (ESAs), for which detection limits were relatively high, or deschloroalachlor, which was never detected.

Although concentrations of individual chloroacetamide degradates measured at this site were low in absolute terms, our results suggest they could still represent contaminants of potential concern elsewhere. For example, in 1989 (prior to the introduction of acetochlor), the median concentration of alachlor in Midwestern streams was determined by USGS researchers as 1.9 g/L (8), only slightly below the MCL value of 2 g/L. Existing literature indicates that alachlor is not efficiently removed during such conventional drinking water treatment unit operations as coagulation/flocculation, sand filtration, and chlorination (64, 65). If neutral alachlor degradates were also inefficiently removed during conventional treatment processes, and if their abundance relative to alachlor were comparable to values we measure in our present study, the consequences to human health in affected regions via ingestion of drinking water could have been appreciable even if only a small fraction of the neutral degradates possessed toxicities similar to that of alachlor.

To fully assess the environmental consequences associ ated with the use of chloroacetanilide herbicides, more extensive studies of the toxicity of the neutral degradates are required. Candidates for such studies would be easier to prioritize if more information were available regarding human exposure, itself dependent on the occurrence of neutral degradates in U.S. drinking water sources as well as their ease of removal during drinking water treatment processes. The latter occurrence and treatability studies are the focus of ongoing work in our laboratory. At present, it would seem prudent to include neutral chloroacetamide degradates among those micropollutants that do indeed merit consideration as contaminants worthy of additional study.

Acknowledgment

We would like to thank Dan Carlson, Khoi Than, and Mike Blumenfeld for their help in the synthesis of the neutral chloroacetamide degradates. Sampling of the Chesapeake Bay was aided by David Cwiertny and Tamar Kohn. Josh Weiss helped with the chloride analyses. We would also like to express our appreciation to Bill Ball and Dominic Di Toro for helpful discussions. We are grateful for the thorough reviews provided by three anonymous individuals, and for their detailed suggestions for improving this paper. Funding for synthesis of some of the degradates was provided by the National Science Foundation (grant no. CHE-0089168) as part of the Collaborative Research Activities in Environmental Molecular Science in Environmental Redox-Mediated Dehalogenation Chemistry at the Johns Hopkins University; funding for synthesis of the remaining degradates and for most of the analytical method development was provided by the American Water Works Association Research Foundation/U.S. Environmental Protection Agency (contract 02903). Funding for the early stages of analytical method develop ment was provided by the U.S. Environmental Protection Agency (grant R826269-01-0). The Johns Hopkins University gratefully acknowledges that the AWWA Research Foundation is the joint owner of the technical information upon which this publication is based. The Johns Hopkins University thanks the Foundation and the U.S. government, through the Environmental Protection Agency, for its financial, technical, and administrative assistance in funding and managing the project through which this information was discovered. The comments and views detailed herein may not necessarily reflect the views of the AWWA Research Foundation, its officers, directors, affiliates, or agents. This work has not been subjected to the U.S. Environmental Protection Agency's required peer and policy review, and therefore does not necessarily reflect the views of the Agency. No official endorsement should be inferred.