Simultaneous Determination of Imidazolinone Herbicides from Soil

water and 9-13 ng/L for river water. (b) Soil sample analysis utilized combined soil column extraction (SCE) and off-line solid phase extraction (SPE)...
0 downloads 0 Views 292KB Size
Download PDF
Anal. Chem. 1998, 70, 121-130

Simultaneous Determination of Imidazolinone Herbicides from Soil and Natural Waters Using Soil Column Extraction and Off-Line Solid-Phase Extraction Followed by Liquid Chromatography with UV Detection or Liquid Chromatography/ Electrospray Mass Spectroscopy Aldo Lagana`,* Giovanna Fago, and Aldo Marino

Department of Chemistry, “La Sapienza” University, Ple. Aldo Moro 5, 00185 Roma, Italy

This paper describes the simultaneous quantification of the imidazolinone herbicides (IMIs) imazapyr, m-imazamethabenz, p-imazamethabenz, m,p-imazamethabenzmethyl, imazethapyr, and imazaquin in two types of samples. (a) Groundwater, lake water, and river water samples were enriched by off-line solid-phase extraction with a Carbograph-1 cartridge and analyzed by reversed-phase liquid chromatography using a UV detector (λ ) 240 nm). The overall recoveries of IMIs extracted from 1 L of groundwater (fortified with 500-100 ng/L), 0.5 L of lake water (fortified with 500-100 ng/L), and 0.5 L of river water (fortified with 1000-200 ng/L) samples were not lower than 89%. The mean relative standard deviation (RSD) was 5.1% (ranging from 4.1% to 6.8%) in natural water. The detection limits were 30-39 ng/L in groundwater, 43-51 ng/L in lake water, and 55-67 ng/L in river water. The method involves confirmatory analysis by LC/ES-MS in full-scan mode. The dependence of the ion signal intensities on proton concentration in the mobile phase was investigated with a view to optimizing the sensitivity of the ES-MS detector. When LC/ES-MS was used, the limit of detection, calculated from extractedion current profiles (EICPs), was 4-7 ng/L for groundwater and 9-13 ng/L for river water. (b) Soil sample analysis utilized combined soil column extraction (SCE) and off-line solid phase extraction (SPE) for sample preparation, analyzing with LC/ES-MS under selected ion monitoring (SIM). Several different extractants were evaluated for the purpose of SCE optimization. The system that best optimizes the extractability IMIs from the soil was found to be the mixture CH3OH/(NH4)2CO3 (0.1 M, 50:50 v/v). The effect of IMI concentration in the matrix on recovery was evaluated. The total recovery of each IMI from soil at each of the two levels investigated ranged from 87% to 95%. Under three ion SIM conditions, the limit of detection (S/N ) 3) was 0.1-0.05 ng/g in soil samples. S0003-2700(97)00749-X CCC: $14.00 Published on Web 01/01/1998

© 1997 American Chemical Society

The imidazolinones (IMIs) are a relatively new class of herbicides used to control a wide spectrum of broad-leafed weeds and grasses in a variety of agricultural commodities.1 These herbicides are very potent weed killers and are used in doses that are substantially lower than those of conventional herbicides. As shown in Figure 1, the members of this class of herbicides have similar structural features centered around the imidazolinone ring and an attached aromatic system bearing a carboxylic acid moiety. IMIs have an excellent activity against annual and perennial grasses and broad-leafed weeds when applied either pre- or postemergence. They function by inhibiting acetohydroxy acid synthase, the feedback enzyme in the biosynthesis of branchedchain essential acids.2-4 This enzyme is not present in animals. The imidazolinone ring of herbicides is amphoteric and can behave as a weak base or a weak acid. The movement of the acid imidazolinones in the soil can be strongly influenced by many soil properties, the most important of which are pH, organic matter, and clay content. Binding of the acid imidazolinones increases as pH decreases. Basic herbicides protonate and are adsorbed on negatively charged soil colloids. Acidic herbicide anions also become protonated as pH decreases, reducing the repulsive forces present when the molecule is dissociated, thus increasing molecular adsorption.5-9 The typically low application rates used for IMI herbicides make their chemical analysis difficult. A number of methods exist for determining IMI residues in both water and soil samples. Current analytical methods for the (1) The Imidazolinone Herbicides; Shaner, D. L., O’Connor, S. L., Eds.; CRC Press: Boca Raton, FL, 1991. (2) Shaner, D. L.; Anderson, P. C.; Stidham, M. A. Plant Physiol. 1984, 76, 545-546. (3) Anderson, P. C.; Hibbert, K. A. Weed Sci. 1985, 33, 479-483. (4) Shaner, D. L.; Millipudi, N. M. In The Imidazolinone Herbicides; Shaner, D. L., O’Connor, S. L., Eds.; CRC Press: Boca Raton, FL, 1991. (5) Wehtje, G. R.; Dickens, R.; Wilcut, J. W.; Hajek, B. F. Weed Sci. 1987, 35, 858-563. (6) Stougaard, R. N.; Shea, P. J.; Martin, A. R. Weed Sci. 1990, 38 (1), 67-73. (7) Che, M.; Loux, M. M.; Traina, S. J.; Logan, T. J. J. Environ. Qual. 1992, 21, 698-704. (8) Loux, M. M.; Liebl R. A.; Slife F. W. Weed Sci. 1989, 37 (4), 712-718. (9) Renner, K. A.; Meggit, W. F.; Penner, D. Weed Sci. 1988, 36 (1), 78-83.

Analytical Chemistry, Vol. 70, No. 1, January 1, 1998 121

Figure 1. Chemical structures and common names of the imidazolinone herbicides.

determination of IMIs in water are targeted at individual members of the group. For the determination of IMI in water at 1 µg/L levels, the authors10,11 isolated these compounds by means of a solid phase extraction (SPE) cartridge and solvent-partitioning steps with final analysis by HPLC/UV detection or GC/N-P detection.10,12 One attempt at the direct determination of imazapyr in water by LC/UV detection gave a detection limit of 10 µg/L.13 More recently, Stout et al.,14 by combining liquid chromatography with electrospray ionization mass spectrometry or electrospray ionization tandem mass spectrometry, determined the IMI at 1 µg/L levels with only one filtration step prior to analysis. Imidazolinone residues must be monitored at low parts-perbillion levels in soil when performing soil rate-of-dissipation (ROD) studies to determine the compound’s persistence in the environment. Current analytical methodology for determining the imidazolinones in soil generally follows a similar route for all members of this herbicide class. This laborious, time-consuming approach starts with a 1 h extraction of 50 g of soil with several (10) Devine, J. M. In The Imidazolinone Herbicides; Shaner, D. L., O’Connor, S. L., Eds.; CRC Press: Boca Raton, FL, 1991. (11) Wells, M. J. M.; Michael, J. L. J. Chromatogr. Sci. 1987, 25, 345-450. (12) Mortimer, R. D.; Weber, D. F. J. AOAC Int. 1993, 76, 377-381. (13) Liu, W.; Pusino, A.; Gessa, C. Sci. Total Environ. 1992, 123-124, 39-43. (14) Stout, S. J.; daCunha, A. R.; Picard, G. L.; Safarpour, M. M. J. Agric. Food Chem. 1996, 44, 2182-2186.

122

Analytical Chemistry, Vol. 70, No. 1, January 1, 1998

hundred millimeters of 0.5 N NaOH. Next, 20-25 gramequivalents of soil are processed through a series of precipitation and centrifugation procedures, followed by partitioning with several hundred millimeters of CH2Cl2. After evaporating the CH2Cl2, final cleanup is accomplished with solid phase extraction using a strong cation exchange (SCX) column and an alkyl (C8 or C18) column.10,15,16 After this extensive cleanup procedure, determination by LC/UV yields a limit of quantitation (LOQ) of 5 ng/g. Estimates of sample preparation time are ∼12 h for a set of six samples.17 A method was recently developed18 for determining imidazolinone residues in soil at the nanogramsper-gram level. The authors, using imazethapyr, the most widely utilized member of the class, as a representative, achieved not only a limit of quantitation of 1 ng/g but also developed a cleanup procedure by combining microwave-assisted extraction (MAE) with gas chromatography/electron capture negative chemical ionization mass spectrometry. Quantification analyses are preceded by sample preparation to extract the analyte compounds. This step is more difficult for complex samples, such as soil, than for water samples, so the selection of the extraction process can be one of the most important factors in the optimization of IMI analysis. In recent years, the need for more efficient and more economical methods of sample preparation has generated considerable interest in methods in which extraction isolation and preconcentration of organic compounds are combined. The aim of this work was to investigate two procedures: (a) The optimization of an analytical procedure based on an enrichment step using a Carbograph-1 cartridge, a graphitized carbon black, followed by screening LC/UV analysis for different natural water samples. The second objective was to evaluate the potential of LC/ES-MS under full-scan mode as a confirmatory procedure. (b) The potential of combination of soil column extraction (SCE) and solid phase extraction for sample preparation of soil samples coupled with LC/ES-MS under selected ion monitoring (SIM). EXPERIMENTAL SECTION Chemicals. Authentic imidazolinones (IMIs), namely imazapyr (2-[4,5-dihydro-4-methyl-4-(1-methylethyl)-5-oxo-1H-imidazol2-yl]-3-pyridinecarboxylic acid), m,p-imazamethabenz-methyl (methyl (()-2-[4,5-dihydro-4-methyl-4-(1-methylethyl)-5-oxo-1H-imidazol2-yl]-5-methylbenzoic methyl ester and methyl (()-2-[4,5-dihydro4-methyl-4-(1-methylethyl)-5-oxo-1H-imidazol-2-yl]-4-methylbenzoic methyl ester) (mixture of 20% meta and 80% para); imazethapyr ((()-2-[4,5-dihydro-4-methyl-4-(1-methylethyl)-5-oxo1H-imidazol-2-yl]-5-ethyl-3-pyridinecarboxylic acid), imazaquin ((()2-[4,5-dihydro-4-methyl-4-(1-methylethyl)-5-oxo-1H-imidazol-2-yl]3-quinolinecarboxylic acid), were purchased from LabService (Bologna, Italy). m- and p-imazamethabenz were obtained by hydrolysis from the methyl ester. For HPLC, distilled water was further purified by passing it through the Milli-Q RG apparatus (Millipore, Bedford, MA). Acetonitrile “plus” and methanol “plus” of LC gradient grade were from Carlo Erba (Milan, Italy). Formic acid was purchased from Merck (Darmstadt, Germany). All other (15) Curran, W. S.; Liebl, R. A.; Simmons, F. W. Weed Sci. 1992, 40, 482-489. (16) Loux, M. M.; Reese, K. D. Weed Sci. 1992, 40, 490-496. (17) Reddy, K. N.; Locke, M. A. Weed Sci. 1994, 42, 249-253. (18) Stout, S. J.; daCunha, A. R.; Allardice, D. G. Anal. Chem. 1996, 68, 653658.

Figure 2. HPLC/UV chromatograms obtained by analyzing (a) 1 L of groundwater, spiked at 500 ng/L (m-imazamethabenz at 100 ng/L); (b) 0.5 L of lake water, spiked at 500 ng/L (m-imazamethabenz at 100 ng/L); and (c) 0.5 L of river water, spiked at 1000 ng/L (m-imazamethabenz at 200 ng/L). For the chromatographic conditions, see Experimental Section. Peaks: 1, imazapyr; 2, m-imazamethabenz; 3, p-imazamethabenz; 4, m,p-imazamethabenz-methyl; 5, imazethapyr; 6, imazaquin.

Table 1. Effect of Increasing Proton Concentration in Mobile Phase on Signal Intensities of IMI Pesticides relative response (%) no.

compound

I

II

III

IV

1 2 3 4 5 6

imazapyr m-imazamethabenz p-imazamethabenz m,p-imazamethabenz-methyl imazethapyr imazaquin

50 58 56 48 55 42

100 100 100 100 100 100

91 94 96 82 89 83

87 93 88 79 87 78

a Mean values from three determinations relatively calculated. Mobile phases: I, pump A, H2O; pump B, CH3CN; II, pump A, H2O, 30 mM HCOOH; pump B, CH3CN, 5 mM HCOOH; III, pump A, H2O, 60 mM HCOOH; pump B, CH3CN, 10 mM HCOOH; and IV, pump A, H2O, 180 mM HCOOH; pump B, CH3CN, 30 mM HCOOH. For the gradient conditions, see Experimental Section.

solvents were reagent grade (Carlo Erba) and were used as received. Carbograph-1 was supplied by Alltech (Deerfield, IL). The particle size range was 37-150 µm. The Carbograph-1 extraction cartridges were prepared by filling large-diameter (6 cm × 1.3 cm i.d.) syringe-like polypropylene tubes (Supelco, Bellefonte, PA) with 0.5 g of adsorbing material. Polyethylene frits were placed above and below the sorbent bed. Before processing the samples, the cartridge was washed with 10 mL of methylene chloride/methanol (80:20 v/v) acidified with formic acid (50 mmol/L), 5 mL of methanol, and 20 mL of water acidified with hydrochloric acid (pH ) 2), respectively. Hydrolysis and Standard Preparation. To obtain the mand p-imazamethabenz from the corresponding methyl ester, it was endeavored to establish the most suitable hydrolysis conditions ensuring the best yield without any loss of analyte or its transformation into unwanted products. We focused our attention on studying the reaction in basic media. In a glass vial, we added 50 mg of the esters to 100 mL of deionized water basified with NaHCO3 (50 mmol/L). Complete hydrolysis was obtained within 36 h at 40 °C in a water bath. We obtained confirmation of

complete conversion of ester to acid by injecting a small volume of the alkaline solution into the LC column until the complete disappearance of the peak having the same retention time as for the ester and the simultaneous appearance of a peak having the same retention time as the corresponding acid compound. The primary standard solutions at a concentration of 0.5 mg/mL were prepared by dissolving 50 mg each of IMI in 100 mL of CH3CN and storing them at 4 °C. Appropriate volumes of this solution were used to prepare the working standards. Sampling and Sample Preparation. Water samples were collected near Rome, Italy (1 L glass bottles) and stored at 4 °C until analysis. In particular, for the river water (Tiber) and groundwater, a portable, automated sampler (4900 Priority Contaminant Sampler, Manning Products, TX) was used, whereas lake water samples were taken at depths of about 5 m using a Niskin bottle. Prior to the analysis, the water was allowed to reach room temperature. The river samples were filtered through 0.45 µm pore size Whatman GF/C glass fiber pads (Maidstone, UK). For recovery studies, water samples were spiked with an appropriate working standard solution and then shaken vigorously and set aside for 2 h. The soil used for this experiments was a sandy loam (pH ) 6.5, 58% sand, 15% silt, 27% clay, 1.3% organic matter) from Cadriano, Bologna, Italy, sieved to pass 2 mm to remove stones and plant material. Soil samples were taken at depths of 0-30 cm. Soil was stored in a freezer prior to use. Freshly fortified samples were prepared by adding an appropriate volume of standard working solution to 5 g of dried homogenized soil samples. Acetone was added until the solvent completely covered the soil particles. The bulk of the solvent was slowly evaporated to air-dried level. The mixture was then thoroughly mixed for 1 h in a mechanical shaker. Procedure. Extraction of Water Samples. Water samples were artificially contaminated by spiking them with a suitable volume of the composite working standard solution. After being stirred for ∼2 min, samples were subjected to the extraction Analytical Chemistry, Vol. 70, No. 1, January 1, 1998

123

Table 2. Typical Fragment Ions and Their Relative Abundances for IMI Herbicides at Different Cone Voltages skimmer cone voltage (V) no.

a

compound

1

imazapyr

2

m-imazamethabenz

3

p-imazamethabenz

4

m,p-imazamethabenz-methyl

5

imazethapyr

6

imazaquin

Mna

40

50

60

261

262 (100)

274

275 (100)

274

275 (100)

288

289 (100)

289

290 (100)

311

312 (100)

262 (100) 234 (34) 217 (59) 275 (100) 257 (24) 229 (52) 275 (100) 257 (24) 229 (52) 289 (100) 257 (45) 229 (55) 290 (100) 262 (21) 245 (36) 312 (100) 284 (15) 267 (32)

262 (35) 234 (31) 217 (100) 275 (31) 257 (21) 229 (100) 275 (31) 257 (21) 229 (100) 289 (41) 257 (33) 229 (100) 290 (100) 262 (33) 245 (86) 312 (80) 284 (23) 267 (100)

Nominal mass, Da.

procedure. A Carbograph-1 cartridge was fitted into a side-arm filtering flask. Liquids were forced through the extraction device under vacuum from a water pump. Water was forced through the cartridge at flow rates of 30-50 mL/min. When samples were passing through the cartridge, the vacuum was reduced to a minimum. Most of the water remaining in the cartridge was expelled under vacuum for ∼5 min. The cartridge was then washed with 10 mL of water, and the residual water content further decreased by slowly passing 1 mL of CH3OH through the cartridge. After the cartridge had been air-dried for 1 min, a suitably drilled cylindrical Teflon piston with one conical indented base and a Luer tip was forced into the cartridge until it reached the upper frit. The cartridge was turned upside down, a 1.4 cm i.d. glass vial with a conical bottom was placed below it, and analytes were back-eluted by passing through the cartridge 10 mL of CH2Cl2/CH3OH (80:20 v/v) acidified with formic acid (50 mmol/L) at a flow rate of ∼5 mL/min. The mixture was dried at 40 °C under a stream of nitrogen. The precaution was taken not to allow the residue-containing vial to stay in the water bath for more than a few minutes after the solvent appeared to have been completely removed. The residue was reconstituted with 250 µL of CH3CN/H2O (8:92 v/v) acidified with 30 mmol/L HCOOH. Then, 50 µL of the final extract was injected into the LC column. For recovery studies, the concentrations of the analytes were calculated by measuring peak areas and comparing them with those obtained from standard solutions. These were prepared by dissolving suitable known volumes of the working standard solution in the eluent phase used for eluting analytes from the Carbograph-1 cartridge and then following the rest of the procedure described above. Extraction of Soil Samples. A 5 g portion of soil was weighed onto aluminum paper and transferred quantitatively into a 6 cm × 1.3 cm cartridge (Supelco). The IMI herbicides were extracted from the soil column by passing through it 25 mL of methanol/ammonium carbonate (0.1 mol/L, pH ) 8.2, 50:50 v/v). The extract was brought to 25 mL in a volumetric flask, and then 12.5 mL of the extract was diluted to 250 mL with water and passed 124

Analytical Chemistry, Vol. 70, No. 1, January 1, 1998

through a previously treated 0.5 g Carbograph-1 cartridge. It follows the same procedure above reported for the water samples cleanup. The concentration of the IMIs in the samples was calculated by comparing the individual peak heights with an external calibration. The calibration was obtained by injecting standard solution into the HPLC column. Standard solutions were prepared by dissolving suitable known volumes of the working standard solution in the eluent phase used for eluting IMIs from the cartridge and then following the rest of the procedure described in the preceding section. Instrumental Conditions. Liquid chromatography was carried out using a Perkin-Elmer Series 400 instrument (PerkinElmer, Norwalk, CT) equipped with a Rheodyne 7125 injector with a 50 µL loop. The analytes were chromatographed on a 25 cm × 4.6 mm i.d. column filled with 5 µm (average particle size) Alltima LC-18 packing (Alltech) and a Supelguard 2 cm × 4.6 mm i.d. precolumn supplied by Supelco. The UV detector was set at a wavelength of 240 nm. Data were acquired using the LCI-100 laboratory computing integrator from Perkin Elmer. Analysis was carried out using a gradient solvent program. Phase A was water acidified with 30 mmol/L HCOOH, and phase B was acetonitrile acidified with 5 mmol/L HCOOH. The initial composition of the mobile phase was 8% B, which was increased linearly to 60% in 25 min, followed by 5 min at 100% B. The flow rate of the mobile phase was 1 mL/min. For LC/ES-MS analysis, the liquid chromatograph, column, and mobile phase were as reported above. A 30 µL/min portion of the column effluent was diverted into the ES source. A Navigator (Finnigan, Manchester, UK) benchtop mass spectrometer consisting of a pneumatically assisted ES interface and a single quadrupole was used for detecting and quantifying target compounds in the LC column effluent. This was introduced into the ES interface through a 40 cm length of 50 µm diameter PEEK capillary tube. The MS was operated in the positive-ion mode by applying a voltage of 4 kV to the capillary. The source temperature was maintained at 130 °C. Under these conditions, no

Figure 3. Product ion spectrum from CID process for imazapyr and global schema for the possible fragments.

thermal degradation of the targeted IMIs was observed. Ions were generated using extremely pure nitrogen as drying and nebulizing gas at flow rates of 260 L/h and 13 mL/min, respectively. Fullscan LC/ES-MS chromatograms were obtained for the aqueous samples by scanning the quadrupole from 100 to 350 m/z with a 1 s scan after setting the skimmer cone voltage at 50 V. For the soil samples, LC/MS chromatograms were obtained by operating in the time-scheduled selected ion monitoring acquisition mode. Time-scheduled SIM conditions were as follows: LC time 15-19 min, m/z 234, 257, 262, and 275; LC time 19-24 min, m/z 229, 245, 289, and 290; LC time 24-28 min, m/z 267, 284, and 312. Dwell time was 0.5 s, and the span was 0.2 amu for each selected m/z. Fragment ions were obtained by controlling the potential difference between sample and skimmer cones. This provided structurally significant fragment ions in the intermediate ion transport region by collision-induced dissociation (CID) of protonated adduct ions with background gas. The instrument control and data processing utilities included Megalynx software. RESULTS AND DISCUSSION Water Matrix Studies by LC/UV. It is essential to perform a more complete mapping of these pesticides in natural water in

order to get a complete environmental toxicity figure. Therefore, in this paper we set out to focus on the simultaneous determination of six different imidazolinones. All six components in the synthetic mixture of IMIs were initially separated by HPLC within about 25 min. Several silica-based reversed-phase (C8, C18, and Supelcosil LC-ABZ14,19) analytical columns were investigated for optimal chromatographic resolution. The gradient and the effect of varying the proton concentration (HCOOH) in the LC mobile phase were optimized to provide the maximum separation possible in a minimum time period. Whenever high concentrations of HCOOH were used in the LC mobile phase, the m- and pimazamethabenz isomers were not fully resolved. The optimum column, eluent, and gradient conditions used are detailed in the Experimental Section. These conditions ensured the full resolution of the m- and the p-imazamethabenz. This is an improvement over previous data,14 since no analytical method had previously been available for the simultaneous determination of the isomers of the above-mentioned compound. Furthermore, total separation of imazethapyr and m,p-imazamethabenz-methyl was also obtained. (19) Nova´kova´, O. Chromatographia 1994, 39 (1/2), 62-66.

Analytical Chemistry, Vol. 70, No. 1, January 1, 1998

125

Figure 4. TIC chromatogram obtained by analyzing 1 L of groundwater fortified with six imidazolinones at individual level of 250 ng/L (mimazamethabenz, 50 ng/L) (A) and 0.5 L of river water, fortified at level of 500 ng/L (m-imazamethabenz at 100 ng/L) (B). Peaks: 1, imazapyr; 2, m-imazamethabenz; 3, p-imazamethabenz; 4, m,p-imazamethabenz-methyl; 5, imazethapyr; 6, imazaquin.

In order to analyze trace levels of IMIs, trace enrichment of larger volumes of water samples is necessary. When adopting the SPE technique using Carbograph-1 cartridge for preconcentrating trace IMI pesticides in natural waters, both the matrix and the volume to be sampled are important parameters to be considered in order to obtain a high extraction efficiency. The analytes were adsorbed and enriched on the cartridge, and their characterization was achieved by breakthrough experiments under the conditions described in the Experimental Section. The breakthrough volume is measured as the volume at which the analyte is no longer adsorbed on the Carbograph-1 cartridge due to saturation. Breakthrough volume > 1 L was obtained for groundwater, while loss of IMIs occurred when they were extracted from 1 L of river and lake water. Nevertheless, it can be safely assumed that the complete retention of the IMI herbicides should be achieved by passing through the cartridge only 0.5 L of any river and lake water sample, as reported in the Experimental Section. The characteristics which account for the excellent capacity of Carbograph-1 as extraction and cleanup medium for acid substances have been amply described in previous work.20-21 Briefly, the presence of some active centers bearing a positive charge enable Carbograph-1 to behave as both nonspecific and anion exchange sorbent. It follows that anion organic compounds are specifically adsorbed on to the Carbograph-1 surface via 126 Analytical Chemistry, Vol. 70, No. 1, January 1, 1998

electrostatic forces and can be desorbed only by adding a displacing agent to an organic solution. Initially, as reported in a previous work,21 after 1 L of groundwater had passed through the cartridge, the compounds were eluted, and 60-70% of the least acid IMIs were recovered, while the figure for m,p-imazamethabenz-methyl was 0%, the remaining quantity being found in the neutral eluent phase designed to purify the extract. This incomplete recovery can probably be accounted for by the fact that the humic acids and other acid species present in natural water compete with the IMIs for adsorption on to the active sites located on the surface of the carbon. The compound m,p-imazamethabenz-methyl had already been eluted during the neutral washing phase owing to the absence of any acid functional group. As, unlike the preceding work, the aim was the complete recovery of the analytes investigated in a single fraction, washing with neutral eluent phase was omitted, and only subsequently were the IMIs eluted by “back-flushing” with a mixture of 10 mL of methylene cloride/ methanol (80:20 v/v) acidified with HCOOH (50 mmol/L). Quantitative results were obtained. Repeatability was calculated by spiking 1 L of groundwater and 0.5 L of river and lake water, respectively, with 500-100, 1000-200, and 500-100 ng/L of the (20) Di Corcia, A.; Marchese, S.; Samperi, R. J. Chromatogr. 1993, 642, 163164. (21) Lagana`, A.; Fago, G.; Marino, A. J. Chromatogr. In press.

Figure 5. Extracted ion current profiles of the imidazolinones from LC/ES-MS of 500 ng/L (m-imazamethabenz 100 ng/L) fortified river water. Peaks: 1, imazapyr; 2, m-imazamethabenz; 3, p-imazamethabenz; 4, m,p-imazamethabenz-methyl; 5, imazethapyr; 6, imazaquin.

seven IMIs and then analyzing six times each type of water sample. In fortified groundwater, the recoveries of the method were between 91% and 96% (RSD between 4.6% and 6.1%), in river water between 89% and 93% (RSD between 5.2% and 6.8%), and in lake water between 90% and 95% (RSD between 4.8% and 6.6%) for all imidazolinones. The limits of detection (LODs) were calculated by using a signal-to-noise ratio of 3. For groundwater, the LOD was from 30 to 39 ng/L, for river water from 55 to 67 ng/L, and for lake water from 43 to 51 ng/L. The corresponding LC/UV chromatograms, obtained by analyzing IMIs-fortified groundwater (A), lake water (B), and river water (C), are shown in Figure 2. The gradient HPLC retention times obtained from 20 replicated LC/UV analyses of the target IMIs in groundwater extracts were checked. The general criterion used for the identification of target compounds was that the HPLC chromatographic retention time should agree within 2% with the retention time for a reference standard of the same compound analyzed under the same conditions.22 The results of the reproducibility of the retention time obtained on the same day (intraday) and during 1 month (interday) show a fluctuation over a range of 0.2-1.1% (RSD) and do not appear to be much greater over 1 month than over 1 day. Furthermore, the gradient HPLC retention times for each analyte are well within the (2% criteria and are often within (0.2 min retention time in this ∼25 min run. LC/UV vs LC/ES-MS Analysis. The method presented here involves confirmatory analysis by LC/ES-MS instrumentation. To obtain the highest quantitative specificity and sensitivity possible, ion signal optimization was evaluated. The formation of various molecular ion species can be used as support in IMI determina(22) Li, L. Y. T.; Campbell, D. A.; Bennett, P. K.; Henion, J. Anal. Chem. 1996, 68, 3397-3404.

tion. Their nature and intensity ratios provide some structural information that is particularly useful for target compound confirmation. Addition to the mobile phase of ions with a high affinity for the analyte, however, can significantly enhance signal intensity. We therefore investigated the response of [M + H]+ to increasing formic acid in the LC mobile phase. A standard working mixture containing pesticides at the level of 100-20 ng was flow injected into different LC mobile phases mixture containing increasing quantities of formic acid. The results of three separate determinations are shown in Table 1. For each compound, the parameter setting giving rise to the highest signalto-noise ratio was set as 100%, and the responses at other settings were calculated with reference to the highest response value. Addition of formic acid enhanced the [M + H]+ ion signal for all six compounds. As can be seen, the intensity reached a maximum at an HCOOH concentration of 30 mM in phase A and 5 mM in phase B and decreased as formic acid continued to be added. As previously reported also by Zhou and Hamburger,23 the optimum concentration of modifier does not appear to be particularly compound-specific. The best ES conditions, however, may not be optimal from a chromatographic point of view, and a compromise between LC separation efficiency and ES-MS response has to be made. The compromise was made using the conditions described in the Experimental Section. Solution chemistry is probably not the only factor governing the ES gas-phase ionization process. The molecular ions ultimately detected could also result from reactions in the ion transfer region.23-26 When performing trace analysis of pesticides in complex aqueous matrices by LC/MS instrumentation, the presence in the spectra of peaks for characteristic fragment ions is of paramount importance in order to avoid false positives and analyte overestimation. In addition, it has to be considered that legal criteria for testing contaminants in a variety of matrices usually accept spectra displaying the molecular ion species plus two typical daughter ions. Fragment ions can be generated by the CID process of protonated adduct ions with residual drying gas molecules present in the ES-MS transport region. Experimental evidence was obtained for these ion-molecule reactions. For the six targeted IMIs, the effect of increasing the potential difference in the desolvation chamber by increasing the skimmer cone voltage from 40 to 60 V on both the response of the ES-MS detector and the production of fragment ions was evaluated. This experiment was conducted by injecting a working standard solution of IMIs containing quantities of 100-20 ng into an LC column and calculating S/N values at each extraction cone voltage. Measurements were made in triplicate. The positive-ion electrospray spectra of IMI were acquired by scanning the mass range, m/z 100-350, at a rate of 1 scan/s. The resulting spectral data for each scan were stored as an averaged spectrum. Ion signal intensities were calculated by measuring the peak areas for each analyte from the total ion current (TIC) chromatograms. Ions having abundances lower than 10% were neglected. The results (23) Zhou, S.; Hamburger, M. Rapid Commun. Mass Spectrom. 1995, 9, 15161521. (24) Ikonomou, M. G.; Blades, A. T.; Kebarle, P. Anal Chem. 1990, 62, 957967. (25) Kebarle, P.; Tang, L. Anal Chem. 1993, 65, 972A-986A. (26) Le Blanc, J. C. Y.; Wang, J.; Guevremont, R.; Siu, K. W. M. Org. Mass Spectrom. 1994, 29, 587-293.

Analytical Chemistry, Vol. 70, No. 1, January 1, 1998

127

Table 3. Recoverya of IMIs from Soilb with Different Extractants recovery (%) ( RSDc no.

compound

NH4Ac/NH4OH, 0.1 M, pH ) 10

(NH4)2CO3, 0.1 M, pH ) 8.2

CH3OH

(NH4)2CO3/CH3OH, 0.1 M, pH ) 8.2 (50/50 v/v)

1 2 3 4 5 6

imazapyr m-imazamethabenz p-imazamethabenz m,p-imazamethabenz-methyl imazethapyr imazaquin

83 ( 3.8 79 ( 3.9 81 ( 4.3 87 ( 5.7 80 ( 4.7 86 ( 5.1

88 ( 4.2 84 ( 3.9 80 ( 3.7 82 ( 4.5 89 ( 5.0 85 ( 4.3

71 ( 3.5 78 ( 3.6 75 ( 3.8 79 ( 4.2 70 ( 4.2 74 ( 5.0

92 ( 4.7 94 ( 4.0 92 ( 4.3 90 ( 5.3 95 ( 4.6 91 ( 4.3

a

The averages of six samples processed through the procedure.

of these experiments are shown in Table 2. They show that increasing skimmer cone voltage did not exert a great influence on signal value. At lower cone voltages, the positive-ion electrospray spectrum of IMI consists predominantly of protonated molecule ions [M + H]+. Progressively higher cone voltages result in a simultaneous diminution of the [M + H]+ population, with a concomitant increase in the abundance of the diagnostically useful fragment ions. Interestingly, the imazamethabenz isomers exhibited the same quasi molecular ion, [M + H]+, under LC/ ES-MS, but their determination was made possible by the fact that they eluted from the LC column with different retention times. In view of the results reported in these experiments, all the work was carried out at a cone voltage of 50 V, which gave enough structural information and the best sensitivity for the compounds investigated. A graphical presentation of the results of the product ion spectra from the CID process of the [M + H]+ ion of imazapyr and a global schema of the possible fragments are shown in Figure 3. Imazapyr shows a typical fragmentation pattern for IMI compounds. In this case, the CID process (cone voltage 50 V) generates ions that initially result in a loss of “N” lactame (-14 m/z, corresponding to the fragment ion at 248 m/z). At a high cone voltage, daughter ions are produced at m/z 234 and 217, corresponding to the loss of “CO” lactame in the case of the fragment at m/z 234 and carboxylic “CO2” in the fragment at m/z 217. Therefore, to validate this method, 1 L of groundwater and 0.5 L of river water, fortified respectively with 250-50 and 500100 ng/L of IMIs, were enriched following the described techniques and analyzed by LC/ES-MS. For the purpose of quantification, the concentration of the analytes was calculated by measuring the peak areas from extracted-ion current profiles (EICPs) and comparing them with those obtained from standard solutions. For any analyte, the EICP selected was that from the most abundant ion, which in all cases was the [M + H]+ ion. Typical TIC mass chromatograms obtained by analyzing groundwater (A) and river water (B) are shown in Figure 4. As can be seen in river water, at the spiked concentrations, distinct peaks were obtained for all the analytes included in this study except m-imazamethabenz. Well-defined peaks for six IMI compounds were obtained by EICPs from the total ion current chromatograms. Figure 5 shows extracted ion chromatographic profiles for selected analytes obtained by analyzing a river extract spiked with the IMI at the level of 500-100 ng/L. The limits of detection for the samples of natural water were calculated by using a signal-to-noise ratio of 3. By extracting chromatograms for selected m/z values corresponding to selected 128 Analytical Chemistry, Vol. 70, No. 1, January 1, 1998

b

Spiked at 10 ppb.

c

Relative standard deviation.

Table 4. Recoverya of IMIs from Soil recovery ( RSD (%) no.

compound

1 ng/g

20 ng/g

1 2 3 4 5 6

imazapyr m-imazamethabenz p-imazamethabenz m,p-imazamethabenz-methyl imazethapyr imazaquin

87 ( 5.0 90 ( 4.7 89 ( 4.9 92 ( 5.2 91 ( 5.8 86 ( 4.6

92 ( 4.7 94 ( 4.0 92 ( 4.3 90 ( 5.3 95 ( 4.6 91 ( 4.3

a

Mean values from five determinations.

Figure 6. Rappresentative calibration curve for ([) imazapyr and (b) imazaquin in soil. The calibration curve was obtained by the determination of six levels in duplicate to produce a calibration curve that ranged from 0.1 to 10 ng.

analytes, LODs ranged from 4 to 7 ng/L for groundwater and from 9 to 13 ng/L for river water. Soil Matrix Studies by LC/ES-MS. Owing to the complexity of the soil matrix and for the purpose of determining IMIs at concentrations in the order of fractions of a nanogram per gram of soil, the use of LC/UV was not considered to be a sufficient

Figure 7. Time-scheduled SIM LC/MS chromatogram obtained by analyzing a soil extract fortified with imidazolinones at the individual level of 1 ng/g (m-imazamethabenz, 0.2 ng/g). Peaks: 1, imazapyr; 2, m-imazamethabenz; 3, p-imazamethabenz; 4, m,p-imazamethabenz-methyl; 5, imazethapyr; 6, imazaquin.

guarantee of sensitivity and specificity. From a practical point of view, for the purpose of quantification we adopted a more selective and sensitive approach, LC/ES-MS, in our investigation. Recently,27 after proposing a new technique for sample extraction for the determination of aryloxyphenoxypropionates (ArPP) in soil, which in addition to its excellent analytical performance proved to be quite economical, we applied the extraction system to IMIs extraction. Recovery experiments were performed by spiking the soil samples with the pesticides being studied. It is known that spiked samples may not always represent the extractability of “realworld” materials.27,28 However, since we spiked the sample in the same way for all the extractants, the comparison between the extracting systems used should be valid. To establish the most suitable extraction conditions, several solvents or solvent mixtures were investigated. The solvents chosen were those normally used in conventional methods such as microwave-assisted extraction or sonication. The experiments were repeated six times for each sample. The results are summarized in Table 3. Initially, the mixture NH4OAc/NH4OH (0.1 M, pH ) 10) was used as extracting system for SCE as, in a preceding work,18 it had proved to be the best extracting system. This mixture does not quantitatively recover the analytes investigated. To provide improved recoveries considering both that the imidazolinone ring of these herbicides is amphoteric and can behave as a weak acid or a weak base and that the soil function as cation decreases and, therefore, binds the IMIs with variable intensity mainly by means of electrostatic interaction, we focused our efforts on the use of (27) Lagana`, A.; Fago, G.; Marino, A. Submitted to Anal Chem. (28) Steinwandter, H. Fresenius Z. Anal. Chem. 1987, 327, 309-311.

(NH4)2CO3 (0.1 M). As can be seen, although in this case the efficiency of extraction is comparable with that of the mixture NH4OAc/NH4OH (0.1 M, pH ) 10), the phase containing the carbonate solution is to be preferred, as it is more specific. This can, presumably, be accounted for by the fact that the pH of the carbonate phase is lower than that of the extracting phase chosen at the beginning; consequently, the humic and fulvic acids are only partially eluted by the soil column. Quantitative recoveries were obtained only when the mixture CH3OH/(NH4)2CO3 (0.1 M, 50:50 v/v) was used. In addition, the effect of the concentration of IMIs in the matrix on recovery was evaluated. Soil samples were spiked with the appropriate standard solution to give fortification samples at both low (1 ng/g) and high (10 ng/g) concentration levels. Five replicates of each IMI were performed at each of the levels indicated. The results obtained are shown in Table 4. The overall recovery of each IMI from soil at each of the two levels ranged from 87% to 94%. A representative calibration curve for the range from 0.5 to 10 ng/g levels of imazapyr and imazaquin in soil is shown in Figure 6. Each level was analyzed in duplicate, and the resulting standard curve was linear through this range. Samples with higher levels were diluted to fit on the standard curve. Under multiple-ion SIM conditions, the limit of detection for an analyte is determined by the parent of the fragment ion giving the worst S/N value. Using this criterion and setting the S/N threshold at 3, the LOD for IMIs ranged from 0.1 (imazapyr) to 0.05 ng/g (imazaquin). From these data and considering the sample preparation procedure involved in this method, it may be Analytical Chemistry, Vol. 70, No. 1, January 1, 1998

129

inferred that this method can provide definitive and unambiguous confirmation of the presence of the six targeted IMIs at a few nanograms per gram in soil samples. The SIM mass chromatogram obtained following this procedure for a sample of soil spiked with IMI pesticide concentrations at the level of 1-0.2 ng/g is shown in Figure 7. CONCLUSION In this article, a new trace analysis method for detecting imidazolinone herbicides has been developed. This investigation has shown the ability to detect the IMIs in soil and natural water samples. An interesting feature of the present method is that it can be performed under LC/UV conditions as a “warning signal” of pesticides in natural waters. It is clear that this technique can be applied to obtain a definitive and unambiguous confirmation of the presence of the six targeted IMIs in natural water.

130

Analytical Chemistry, Vol. 70, No. 1, January 1, 1998

In soil, the combination of SCE technique and SPE also proved to be a simple and robust sample preparation method with considerable potential for application in the field. As a result, the methods not only can be used as a routine screening tool for the assessment of some of the most widely used postemergence herbicides but also have proved rugged and sensitive enough to study their fate and behavior in various kinds of soil and water samples.

Received for review July 14, 1997. Accepted September 30, 1997.X AC9707491

X

Abstract published in Advance ACS Abstracts, November 15, 1997.