Identification of Ionic Chloroacetanilide−Herbicide Metabolites in

Zengru Wu, Wenqing Gao, Mitch A. Phelps, Di Wu, Duane D. Miller, and James T. .... Sônia Alvim Veiga Pileggi , Itamar Soares de Melo , Marcos Pileggi...
0 downloads 0 Views 172KB Size
Anal. Chem. 1997, 69, 4547-4553

Identification of Ionic Chloroacetanilide-Herbicide Metabolites in Surface Water and Groundwater by HPLC/MS Using Negative Ion Spray Imma Ferrer,† E. M. Thurman,‡ and Damia` Barcelo´*,†

Deptartment of Environmental Chemistry, CID CSIC, Jordi Girona 18-26, 08034 Barcelona, Spain, and U.S. Geological Survey, 4821 Quail Crest Place, Lawrence, Kansas 66049

Solid-phase extraction (SPE) was combined with highperformance liquid chromatography/high-flow pneumatically assisted electrospray mass spectrometry (HPLC/ ESP/MS) for the trace analysis of oxanilic and sulfonic acids of acetochlor, alachlor, and metolachlor. The isolation procedure separated the chloroacetanilide metabolites from the parent herbicides during the elution from C18 cartridges using ethyl acetate for parent compounds, followed by methanol for the anionic metabolites. The metabolites were separated chromatographically using reversed-phase HPLC and analyzed by negative-ion MS using electrospray ionization in selected ion mode. Quantitation limits were 0.01 µg/L for both the oxanilic and sulfonic acids based on a 100-mL water sample. This combination of methods represents an important advance in environmental analysis of chloroacetanilide-herbicide metabolites in surface water and groundwater for two reasons. First, anionic chloroacetanilide metabolites are a major class of degradation products that are readily leached to groundwater in agricultural areas. Second, anionic metabolites, which are not able to be analyzed by conventional methods such as liquid extraction and gas chromatography/mass spectrometry, are effectively analyzed by SPE and high-flow pneumatically assisted electrospray mass spectrometry. This paper reports the first HPLC/MS identification of these metabolites in surface water and groundwater. The chloroacetanilide herbicides, acetochlor, alachlor, and metolachlor, are a major class of herbicides used in the United States. Together with triazine herbicides, chloroacetanilide herbicides account for the majority of pesticides applied to farmland in the Midwestern United States for weed control on row crops, such as corn and soybeans.1 Alachlor and metolachlor have been used for more than 20 years, whereas acetochlor was used extensively for the first time in March of 1994.2 Chloroacetanilide herbicides are known to degrade more quickly in soil than triazine herbicides, and typical half-lives of the chloroaceta†

Department of Environmental Chemistry, CID CSIC. U.S. Geological Survey. (1) Gianessi, L. P.; Puffer, C. M. Use of Selected Pesticides for Agricultural Crop Production in the United States; NTIS: Springfield, VA, 1982-1985; p 490. (2) Kolpin, D. W.; Nations, B. K.; Goolsby, D. A.; Thurman, E. M. Environ. Sci. Technol. 1996, 30, 1459-1464. (3) Leonard, R. A. In Environmental Chemistry of Herbicides; Grover, R., Ed.; CRC: Boca Raton FL, 1988; pp 45-88. ‡

S0003-2700(97)00467-8 CCC: $14.00

© 1997 American Chemical Society

nilide herbicides are from 15 to 30 days,3 compared to 30-60 days for triazine herbicides. Several recent studies have pointed out the occurrence of chloroacetanilide metabolites in surface water and groundwater.4-6 Sometimes, concentrations of metabolites may equal to or even exceed concentrations of parent compounds. For example, Kolpin et al.4 found that metabolite concentrations in groundwater often exceed parent compound concentrations for both chloroacetanilide and triazine herbicides, whereas in surface water the parent compound is most abundant after application of herbicide in the spring and is replaced gradually with metabolites throughout the growing season. In the fall, the metabolite concentrations may exceed concentrations of the parent compound. Studies of metabolites of parent herbicides are critical to understanding the fate and transport of herbicides applied to soil. Toxicological studies of metabolites are underway to assess whether the presence of these compounds is important to the total burden of pesticides in surface water and groundwater. For these reasons, continued development of reliable and sensitive methods of analysis for metabolites are important for studies of water quality. One relatively new method of analysis of herbicides in water is the use of high-performance liquid chromatography (HPLC) coupled with mass spectrometry (MS) by means of electrospray ionization (ESP).7 ESP provides an ionization of the molecules, caused by ion evaporation, with a sensitivity sufficient for the trace analysis of herbicide metabolites in natural waters, where concentrations are frequently below the microgram-per-liter level. Commonly, HPLC is needed because the metabolites are ionic compounds; therefore, they cannot be routinely analyzed by gas chromatography/mass spectrometry (GC/MS), where derivatization is required.8 Thus, HPLC is a simple and efficient method, especially when coupled with MS. This paper addresses the development of an analytical method for the determination of the oxanilic and sulfonic acids of the chloroacetanilide herbicides in surface water and groundwater. The purpose of the research was (1) to develop a generic method coupling solid-phase extraction (4) Kolpin, D. W.; Thurman, E. M.; Goolsby, D. A. Environ. Sci. Technol. 1996, 30, 335-340. (5) Thurman, E. M.; Goolsby, D. A.; Aga, D. S.; Pomes, M. L.; Meyer, M. T. Environ. Sci. Technol. 1996, 30, 569-574. (6) Aga, D. S.; Thurman, E. M.; Yockel, M. E.; Zimmerman, L. R.; Williams, T. D. Environ. Sci. Technol. 1996, 30, 592-597. (7) Barcelo´, D. Applications of LC-MS in Environmental Chemistry; Elsevier: Amsterdam, 1996; p 543. (8) Zimmerman, L. R.; Thurman, E. M.; Gilliom, R. J. Abstracts of Papers, 209th National Meeting of the American Chemical Society, Anaheim, CA, Spring1995; American Chemical Society: Washington, DC, 1995; p 218.

Analytical Chemistry, Vol. 69, No. 22, November 15, 1997 4547

(SPE) for anionic pesticide metabolites with high-performance liquid chromatography with mass spectrometry using high-flow pneumatically assisted electrospray and negative ion detection, (2) to demonstrate the method for several types of metabolites, the oxanilic and sulfonic acids of the major chloroacetanilide herbicides, and (3) to apply the method to several surface water and groundwater samples to show the relative importance of these compounds in the aquatic environment. EXPERIMENTAL SECTION Reagents. HPLC-grade solvents acetonitrile, methanol, and water were purchased from Merck (Darmstadt, Germany). Acetic acid was also purchased from Merck. For the pesticide standards, acetochlor sulfonic acid was obtained from Zeneca Agrochemicals (Jealott’s Hill, UK), acetochlor oxanilic acid and alachlor oxanilic acid from Monsanto Chemical Co. (St. Louis, MO), and alachlor sulfonic acid from the EPA Repository (Cincinnati, OH); metolachlor sulfonic acid was synthesized in the U.S. Geological Survey laboratory in Lawrence, KS, by D. S. Aga,6 and metolachlor oxanilic acid was obtained from Bob Zablowitcz in the Agricultural Research Service (Stoueville, MS). The SPE cartridges that were used (Sep-Pak from Waters-Millipore, Milford, MA) contained 360 mg of 40-µm C18-bonded silica. Standard stock solutions were prepared in methanol. Sampling. Surface water samples were collected from several streams in the Midwestern United States. Samples were collected with a depth-integrating technique at three or more locations across each stream. The herbicide samples were collected in glass or Teflon bottles, composited in large containers, and filtered through 1-µm-pore diameter glass fiber filters into baked-glass bottles prior to shipment to the laboratory. All sampling equipment was cleaned with non-phosphate detergent, rinsed thoroughly with tap water, and then rinsed with distilled water and deionized water, followed by a final rinse with a 50% solution of methanol and organic-free water. Groundwater samples were collected during the spring and summer of 1995 from 303 wells penetrating near-surface, unconsolidated, and bedrock aquifers of the Midcontinental United States.9 From these wells, six samples were selected for analysis by HPLC/ESP/MS. Water samples collected for herbicide analysis were filtered through a 1.0-µm glass fiber filter into amber, baked-glass bottles and chilled immediately. Sample Preparation. The SPE procedure was automated with a Millipore Workstation (Waters, Milford, MA) as described previously.10 The C18 cartridges were washed sequentially with 2 mL of methanol, 6 mL of ethyl acetate, 2 mL of methanol, and 2 mL of distilled water. A 100-mL aliquot of sample was passed through the cartridge at a flow rate of 10 mL/min. Excess water was forced from the cartridge with air. The cartridge was eluted with 3 mL of ethyl acetate, followed by a transfer step to remove the ethyl acetate (top layer) containing the parent herbicides from the residual water (bottom layer) in the eluate. Then, the cartridge was eluted with methanol to remove the herbicide metabolites. Methanol extracts were evaporated to dryness under nitrogen at 45 °C using a Turbovap (Zymark, Palo Alto, CA). The (9) Kolpin, D. W.; Burkart, M. R. Work Plan for Regional Reconnaissance for Selected Herbicides and Nitrate in Ground Water of the Mid-Continental United States; U.S. Geol. Survey, Open-File Report 91-59; U.S. Geological Survey: Iowa City, IA, 1991, p 18. (10) Thurman, E. M.; Meyer, M. T.; Pomes, M. L.; Perry, C. A.; Schwab, P. Anal. Chem. 1990, 62, 2043-2048.

4548 Analytical Chemistry, Vol. 69, No. 22, November 15, 1997

Table 1. Chromatographic Conditions and Operational Parameters Used in the Analysis of the Water Samples ions monitored under SIM conditions cone voltage ESP voltage source temperature flow rate analytical column column temperature mobile phase

146, 160, 206, 264, 278, 314, 328 40 V 3.1 kV 150 °C 0.3 mL/min 25 cm × 3 mm, 5 µm, C18 60 °C 40% ACN, 24% MeOH, 35.7% H2O, 0.3% acetic acid

extracts were reconstituted with 75 µL of 20/80 phosphate buffer/ methanol mixture. Finally, a volume of 20 µL of the sample extracts was injected into the mass spectrometer system. HPLC/ESP/MS. The eluent was delivered by a gradient system from Waters Model 616 pumps controlled by a Waters Model 600S Controller (Waters-Millipore). The mobile phase consisted of 24% methanol, 35.7% water, 40% acetonitrile, and 0.3% acetic acid at a flow rate of 0.3 mL/min. A 25-cm × 3-mm-i.d. column packed with 5 µm of C18 was used. The analytical column was set at 60 °C in order to achieve a better separation between the compounds analyzed. This HPLC system was connected to a VG Platform system (Micromass, Manchester, UK) equipped with a Megaflow ESP probe. The design of the ESP consists of a coaxial flow probe. After the HPLC separation, the sample was introduced into the ESP source together with a nebulizing gas, which flowed directly through the probe tip, maximizing the efficiency of the nebulization. A drying gas was added to flush out any solvent that may have entered the gas line by capillary action. The high-flow pneumatically assisted electrospray,7 using a VG Platform instrument, was used at a flow rate of 0.3 mL/min and at a source temperature of 150 °C. The drying-gas flow rate was kept at 350 L/h, and the ESP voltage was 3.1 kV. The cone voltage was varied between 20 and 60 V in order to study the fragmentation of the different metabolites. Generally, a cone voltage of 40 V was used in the analysis of the water samples owing to the enhanced fragmentation as compared with lower cone voltages. The instrument control and data-processing utilities included the use of the MassLynx application software installed in a Digital DEC PC 466. In Table 1 are summarized the experimental conditions and operational parameters used for the characterization of the analytes. RESULTS AND DISCUSSION Generic Solid-Phase Extraction. The parent compounds acetochlor, alachlor, and metolachlor were separated from the oxanilic and sulfonic acids of each compound by selective elution. The method was originally developed for the separation of alachlor and alachlor sulfonic acid.11 The method consists of sorption of the alachlor and alachlor sulfonic acid metabolite onto the C18 sorbent at pH 7. At this pH, the metabolite is in the anionic form. However, the metabolite is isolated at the surface of the C18 as a surfactant with sodium or calcium as the counterion. The elution consists of washing the column with ethyl acetate. The ethyl acetate removes the parent compounds but does not solubilize (11) Aga, D. S.; Thurman, E. M.; Pomes, M. L. Anal. Chem. 1994, 66, 14951499.

Table 2. Solid-Phase Extraction and Recovery of Anionic Compounds Using Sequential Elution with Ethyl Acetate and Methanola compound

% sorbed

% recovery in ethyl acetate

% recovery in methanol

acetochlor ESA acetochlor oxanilic acid alachlor ESA alachlor oxanilic acid metolachlor oxanilic acid metolachlor ESA 2,4-dichlorophenoxyacetic acid

98 98 98 98 98 98 5

10 10 10 10 12 12

88 88 88 88 86 86

a The SPE cartridge material was a Waters C 18 monofunctional sorbent with 125-Å pore diameter and 60-µm particles. The C18 loading was 12% and was end-capped with trimethylsilane; spiked level ) 1 µg/L; CV ) (5 (n ) 3).

the anionic metabolites. Next, the cartridge is eluted with methanol. The methanol solubilizes the sulfonic acids. Furthermore, the isolation by C18 is a generic separation, which was not realized in the earlier work.11 In this study, it was found that the oxanilic acids of each of these herbicides are also effectively sorbed and eluted in the methanol fraction (see recovery in Table 2). The recovery for alachlor sulfonic acid was reported to be approximately 95%;11 in this study, we obtained 88% recovery (Table 2). Although the oxanilic acid contains one less carbon atom, it is effectively recovered also. As the anion loses carbon atoms, its ability to sorb decreases until it is not possible to isolate them at neutral pH. Results of this extraction process show that isolation is possible down to a value of ∼10 carbon atoms/ functional group. For example, the herbicide 2,4-D, which contains eight atoms of carbon, was not isolated or sorbed by the C18 at neutral pH. The sorbent used for this work was a monofunctional, end-capped C18 with a carbon loading of 12%. The low carbon loading may be involved in the sorption and recovery of the anions since there is considerably more exposed silica relative to the heavier C18 loadings of 17-18% carbon.12 Below this value of 10 carbon atoms/functional group, C18 is not able to sorb efficiently enough for the SPE method. Thus, the SPE method is generic in that it isolates both oxanilic and sulfonic acids in the range of >10 carbon atoms/functional group. The power of the method is that a fraction is generated that contains anionic molecules when the generic elution is used (ethyl acetate followed by methanol). Furthermore, this anionic fraction should be easily ionized by electrospray mass spectrometry using negative ion detection. Mass Spectrometry. Figure 1 shows the mass spectrum and fragmentation ions for acetochlor oxanilic acid at a cone voltage of 40 V. The molecule minus a proton, [M - H]-, has a mass of 264. The 265 ion is the C13 isotope of the [M - H]- ion. The 324 ion is formed from the addition of acetic acid to the molecular ion [M - H‚CH3COOH]-. After LC separation, the sample is vaporized from the solution and introduced into the source with a nebulizing gas that maximizes the efficiency of the nebulization. In the cone region, a skimmer device is used for sampling for the fully or partly desolvated ions. In this region there exists a high pressure that can lead to CID (collisionally induce dissocia(12) Zief, M.; Kiser, R. Sorbent extraction for sample preparation; J.T. Baker: Phillipsburg, NJ, 1988.

tion)-type reactions and that will favor the formation of [M H‚CH3COOH]- adducts. This reaction takes place up to 60 V and does not occur if acetic acid is not used in the mobile phase, but in that case poor LC separation is achieved, as is discussed in the next section. The base peak at 146 is formed via the breaking of the carbon-nitrogen bond between the aromatic ring and the amide nitrogen (Figure 1). A proton is transferred from the alkyl group of the aromatic ring to the nitrogen, which forms the 146 ion. The 146 base peak ion occurs at a cone voltage of 40 V. However, at a cone voltage of 20 V, the base peak ion is the molecular ion. Therefore, a cone voltage of 40 V was chosen for sample analysis in order to distinguish acetochlor oxanilic acid from alachlor oxanilic acid, which both have the same molecular ion, [M - H]-, at 264. Figure 2 shows the mass spectrum and fragmentation ions for alachlor oxanilic acid at a cone voltage of 40 V. The molecule minus a proton, [M - H]-, has a mass of 264. The 265 ion is the C13 isotope of the [M - H]- ion. Once again, the 324 ion is formed from the addition of acetic acid to the molecular ion [M H‚CH3COOH]-. The base peak ion for alachlor oxanilic acid is 160. This ion is formed by the transfer of either of the ethyl groups from the aromatic ring to the amide nitrogen. Figure 3 shows the mass spectrum and fragmentation ions for metolachlor oxanilic acid at a cone voltage of 40 V. The molecular ion is 278, and it is referred to the [M - H]- ion. The ion corresponding to the adduct with acetic acid is also observed at 338. In the case of acetochlor and alachlor, the base peak ions originate from the fragmentation between the aromatic ring and the amide nitrogen, but in the case of metolachlor, the fragmentation occurs between the alkyl side chain and the amide nitrogen, which gives the base peak of mass 206 (Figure 3). The negative ion spectra in electrospray for the three sulfonic acid metabolites gave only the molecular ion (Table 3). Thus, acetochlor and alachlor sulfonic acids have the same molecular ion at 314. Metolachlor sulfonic acid has its molecular ion at 328. These spectra were obtained at 20 and 40 V. Nevertheless, minor fragmentation was seen for both acetochlor and alachlor sulfonic acids at a cone voltage of 60 V. Unfortunately, their sensitivity is low at this cone voltage and it is not suitable for routine analysis. Therefore, chromatographic separation is required for identification of acetochlor and alachlor sulfonic acids. Finally, Table 3 shows the summary of mass spectral conditions for the identification of the three oxanilic and the three sulfonic acid metabolites. Chromatographic Conditions. In our previous work, using HPLC with diode array detection, a 0.010 M potassium phosphate buffer at pH 7 (40% methanol, 10% acetonitrile, 5% water) was used in order to separate the three sulfonic acids of the chloroacetanilide herbicides. The use of potassium as a counterion for the sulfonic acid resulted in a nearly baseline separation when the column was heated to 60 °C. The column heater was required to separate the rotational isomers of acetochlor from alachlor sulfonic acids. Unfortunately, potassium cannot be used as a modifier for HPLC/MS because of its low volatility. Therefore, chromatographic conditions were optimized for two other buffers systems, ammonium acetate and acetic acid. The chromatography for the oxanilic and sulfonic acids using the ammonium acetate buffer resulted in broad peaks with poor resolution. This result was caused by the association of the ammonium sulfonic acid ion pair with both the C18 stationary phase and ammonium ion associated with the silica matrix of the Analytical Chemistry, Vol. 69, No. 22, November 15, 1997

4549

Figure 1. Mass spectrum and proposed fragmentation ions for acetochlor oxanilic acid at a cone voltage of 40 V. Carrier stream: acetonitrilewater-methanol containing 0.3% acetic acid at a flow rate of 0.3 mL/min.

Figure 2. Mass spectrum and proposed fragmentation ions for alachlor oxanilic acid at a cone voltage of 40 V. Carrier stream: acetonitrilewater-methanol containing 0.3% acetic acid at a flow rate of 0.3 mL/min.

column. This result was not observed for the potassium phosphate buffer. We hypothesize that the ammonium ion is capable of hydrogen bonding to the silica matrix, whereas the potassium is not. The adsorbed ammonium ion is also capable of interacting with the sulfonic acid metabolites; therefore, the chromatography of sulfonic acids has two mechanisms of interaction that broaden the peak. The elution order of the compounds was that the oxanilic acids eluted before the sulfonic acids, which is consistent 4550

Analytical Chemistry, Vol. 69, No. 22, November 15, 1997

with the hydrophobicity of metabolites. Acetochlor and alachlor oxanilic acids coeluted before metolachlor oxanilic acid, whereas acetochlor and alachlor sulfonic acids coeluted before metolachlor sulfonic acid. The chromatography of the oxanilic and sulfonic acids was considerably better with the acetic acid buffer at 0.3%. At this concentration of acetic acid, the oxanilic acids are partially protonated, which results in improved chromatography and in a

Figure 3. Mass spectrum and proposed fragmentation ions for metolachlor oxanilic acid at a cone voltage of 40 V. Carrier stream: acetonitrilewater-methanol containing 0.3% acetic acid at a flow rate of 0.3 mL/min.

Table 3. Typical Fragment Ions and Relative Abundances (RA) of the Herbicide Metabolites in HPLC/ESP/MS in Negative Ion Mode of Operationa 20 V

40 V

Mn

m/z

RA

Mn

m/z

RA

acetochlor ESA acetochlor oxanilic acid

314 264 314 264

metolachlor ESA metolachlor oxanilic acid

328 278

2,4-D

220

100 100 90 100 100 30 10 100 100 10 100 40

314 264

alachlor ESA alachlor oxanilic acid

314 264 146 314 264 160 192 328 278 206 219 161

314 146 264 314 160 264 158 328 206 278 161 219

100 100 70 100 100 60 10 100 100 80 100 15

compound

314 264 328 278 220

a Cone set at 20 and 40 V and corona at 3.1 kV. Carrier stream: acetonitrile-water-methanol containing 0.3% acetic acid at a flow rate of 0.3 mL/min. Mn) nominal mass.

reversal of retention, with the most hydrophobic oxanilic acid, metolachlor, eluting before the acetochlor and alachlor oxanilic acids. With this buffer system, there was no separation of the acetochlor and alachlor oxanilic acids. Apparently, the pKa’s and chemical structures of these compounds are so similar that separation is not possible. However, the base peak fragmentation ions for acetochlor and alachlor oxanilic acids are different; thus, these two compounds can be distinguished on the basis of the mass spectra (see discussion in earlier section on mass spectrometry). On the other hand, the chromatography of acetochlor and alachlor sulfonic acids presented the same problem. Both sulfonic acids have nearly the same retention time and the same molecular ion, so it was not possible to quantify them unequivo-

cally. Nevertheless, there was a slight difference in retention time between the two compounds; thus, it was possible to detect the presence of both compounds in sample extracts, but it was not possible to quantify them separately. More chromatographic separations are under consideration to separate completely the acetochlor and alachlor sulfonic acids. A polymeric column and triethylammonium salt as an ion pair reagent were tried, but neither method provided successful separation. Thus, the quantitation of acetochlor and alachlor sulfonic acids is not possible at this time. Metolachlor sulfonic acid, although not completely separated from acetochlor and alachlor sulfonic acid, may be quantified by the 328 ion. Finally, in Table 1, the chromatographic conditions and operational parameters used in the analysis of the water samples are summarized. In general terms, HPLC/MS was better for the detection of oxanilic acids than for the detection of sulfonic acids because two ions were obtained on mass spectra for selected ion monitoring (SIM) identification instead of only one ion, as in the case of sulfonic acids. Sample Analysis. Table 4 shows the results obtained for selected surface and groundwater samples after the analysis by negative electrospray under SIM conditions. The fragmentation ion of each compound that was used for quantitation is also shown in this table. The coefficient of variation varied between 15 and 35% (n ) 5) due to the fact that sensitivity decreased as the ion source was plugged by salt present in the samples. One of the improvements achieved with HPLC/ESP/MS as compared to conventional HPLC/DAD is that a better sensitivity is obtained when working under SIM conditions. Furthermore, the broad matrix peak corresponding to the humic and fulvic substances of natural waters is artificially avoided, which gives a clearer baseline. In this respect, the data of the samples reported in Table 4 correspond to surface water and groundwaters with different Analytical Chemistry, Vol. 69, No. 22, November 15, 1997

4551

Table 4. Concentration (µg/L) of the Metabolites Studied, in Surface Water and Groundwater Samples, after the Analysis by HPLC/ESP/MSa

Samples

acetochlor alachlor metolachlor metolachlor oxanilic acid, oxanilic acid, oxanilic acid, ESA, m/z ) 146 m/z ) 160 m/z ) 206 m/z ) 146

Cedar Riverb Iowa Riverb Wapsipiniconb Old Man’s Creekb Skunk Riverb plot 1c plot 1c plot 1c plot 1c plot 1c plot 1c plot 2c plot 2c

nde nd 0.15 0.12 nd 0.03 0.04 0.03 0.08 0.07 nd nd 0.02

Surface Water nd nd 0.16 0.21 nd nd 0.02 0.01 0.13 0.17 nd 0.07 0.05

nd nd 0.23 0.29 0.25 nd 0.01 nd 0.11 0.22 nd 0.29 0.08

0.74 0.33 0.75 1.82 0.75 0.14 0.05 0.07 0.15 0.47 0.09 0.42 0.11

Iowa-1d Iowa-2d Iowa-3d Iowa-4d Iowa-5d Iowa-6d

0.17 0.02 0.09 0.06 nd nd

Groundwater 0.24 0.03 0.09 0.14 0.02 1.66

0.91 0.07 0.13 0.22 0.03 0.59

1.83 0.37 0.14 0.95 0.10 0.99

a CV varied between 25 and 35% for surface water samples and 15 and 20% for groundwater samples. b Surface water samples were collected in Iowa. c Plot 1 and plot 2 correspond to several runoff water samples from a field plot carried out in Topeka, KS. d Groundwater samples were collected in Iowa. e Not detected.

amounts of humic substances. It has been reported11 that surface water samples had higher levels of humic and fulvic substances as compared to groundwater samples, so the coefficients of variation (CV) varied between 25 and 35% in surface water, whereas in groundwater samples CV varied between 15 and 20%. Another improvement of HPLC/MS versus HPLC/DAD is the selectivity obtained between coeluting compounds in HPLC, such as the case of acetochlor and alachlor oxanilic acids. It has been demonstrated that, although these oxanilic acids have the same molecular mass, it is possible to separate them using the detection of their fragmentation ions. They both have nearly identical DAD spectra and cannot be differentiated. The presence of all the metabolites of acetochlor, alachlor, and metolachlor was confirmed with the HPLC/ESP/MS technique (see Table 4), although it was not possible to identify the presence of acetochlor ESA due to the coelution with alachlor ESA. For that reason, the quantitation of alachlor ESA was not carried out, even though it is present at high concentrations in surface water and groundwater samples (see Figure 4). However, separation of the acetochlor and alachlor sulfonic acids was possible using a potassium phosphate buffer and HPLC/DAD. Unfortunately, the conditions of HPLC/MS do not allow the use of nonvolatile salts. Thus, the separation of these compounds warrants further investigation. However, we are able to collect fractions from a chromatographic analysis and re-inject them into the HPLC/MS for identification, although this is a cumbersome method. Acetochlor oxanilic acid was detected considerably less frequently than the other acids. The much lower concentrations for acetochlor metabolites likely reflect the substantially lower amounts of acetochlor used at this time compared to other herbicides examined. Previous research has shown that chemical use is an important factor in the transport of pesticides to streams across

4552 Analytical Chemistry, Vol. 69, No. 22, November 15, 1997

Figure 4. LC/ESP/MS chromatogram in negative ion (NI) mode of operation and under SIM conditions corresponding to a groundwater sample. LC conditions are as described in the Experimental Section (Table 1).

the Midwestern United States.13 Alachlor sulfonic acid, although not quantified, was found in all the samples analyzed. Alachlor sulfonic acid was much more abundant than the parent compound alachlor, with peak concentrations occurring in the late summer and early fall.5 Metolachlor sulfonic acid is a major metabolite of metolachlor on the basis of the concentrations that were found and the percentage of detections in all the samples. Finally, isotopically labeled standards would be useful to improve the precision of measurements because of reproducibility problems with HPLC/ESP/MS. Variations in the solvent pumps and plugging problems encountered in the source are the main causes of the lack of reproducibility achieved in the measurements. Constant comparison of standards to samples plus a onepoint standard curve were needed to obtain semiquantitative values for water samples ((30% coefficient of variation). The method was improved using the 2,4-D as an internal standard (results not shown here) for sample analysis, and the CVs obtained were much lower (5-8%). Also, more precise HPLC pumps for microliter flow rates would improve precision, and these additions are being considered for future work. However, in spite of these problems, the power of the generic methodsisolation by SPE, followed by negative electrospray HPLC/MSsis exciting in that these metabolites can be measured with great sensitivity, with detection levels of 0.01 µg/L for a 100-mL sample, which means a total mass detected of ∼200 pg. The detection levels of 0.01 µg/L were estimated on the basis of instrumental signal-to-noise ratio of 3-4 (the ratio between the peak intensity and the noise) obtained under SIM conditions. A small sample volume of 100 mL improves recovery by SPE and avoids interferences, such as the humic and fulvic acids that occur naturally in surface water and groundwater. When using only HPLC analysis, the humic and fulvic acids are a huge interference peak in the chromatogram. With electrospray using negative ion detection, we can “see through” the humic material, and no interference occurs. This result enhances the detection limit of the method. This is yet another major advantage of using the generic method of HPLC/ ESP/MS for the analysis of trace anionic metabolites and herbicides in groundwater, and this method should prove useful for other anionic pesticide metabolites. (13) Thurman, E. M.; Goolsby, D. A.; Meyer, M. T.; Mills, M. S.; Pomes, M. L.; Kolpin, D. W. Environ. Sci. Technol. 1992, 26, 2440-2447.

ACKNOWLEDGMENT This work has been supported by the Commission of the European Communities, Environment & Climate Program 199498 (ENV4-CT96-0333) and CICYT (AMB97-1597-CE). We thank David French of Zeneca Agrochemicals (Jealott’s Hill, UK) for partial support of this project. Roser Chaler and Rosi Alonso are gratefully acknowledged for HPLC/MS technical support. The use of trade, firm, or brand names in this paper is for identification purposes only and does not constitute endorsement by the U.S.

Government. This work was a joint research effort between the Consejo Superior de Investigaciones Cientificas (CSIC) and the U.S. Geological Survey (USGS). Received for review May 7, 1997. Accepted August 19, 1997.X AC9704671 X

Abstract published in Advance ACS Abstracts, October 1, 1997.

Analytical Chemistry, Vol. 69, No. 22, November 15, 1997

4553