Anal. Chem. 1998, 65, 7773-1778
1779
Automatic Gas Chromatographic Determination of NMethylcarbamates in Milk with Electron Capture Detection Evaristo Ballesteros, Mercedes Gallego, and Miguel Valchcel' Department of Analytical Chemistry, Faculty of Sciences, University of C6rdoba, Cbrdoba, 14004 Spain
A new analytical method that combines on-line derivatization-extractionand gas-liquid chromatography for the determination of N-methylcarbamates is reported. The hydrolysis products of N-methylcarbamates(phenols)are derivatizedand extracted in a continuous fashion using pentafluoropropionic anhydride (derivatizing reagent) in n-hexaneas an extractant. The thermal instability of N-methylcarbamates is overcome by forming fluoro derivatives that can be identified and quantifiedat the nanogramper milliliter level with an ECD detector. Application of the proposed method to six mixtures of N-methylcarbamates yielded detection limits between 2 and 20 ng/mL and a relative standard deviation of 2.35-3.70 70. Average recoveriesof N-methylcarbamatesadded to cow's milk at concentrations between 50 ng/mL and 2 bg/mL ranged from 93.7 to 100.8%.
INTRODUCTION There is growing concern about the exposure of human populations to pesticides used in agriculture and household activities. Many gas chromatographic methods have been developed for carbamate determination. Some reasons for derivatizing N-methylcarbamates, other than their thermolability during direct gas-liquid chromatography (GLC) analysis, include increased detector sensitivity (particularly with the ECD), increased volatility, better chromatographic separations, the possibility of extending application to multiresidue analysis, and increased stability of the resulting compounds.' There are two generalapproachesto the analysis for N-methylcarbamates by derivatization, namely, derivatization of the intact pesticides or of a hydrolysis product, one of which is always the volatile methylamine. The reactions typically used to obtain derivatives of both intact or hydrolysis products ofN-methylcarbamates are methylation? silylation? halogenation,' acylation: and esterification.6 For analytical purposes, acetyl derivatives' and methyl derivatives8 have been prepared for use with FID, but most analysts prefer to introduce electron-capturing moieties for use with the more selective and sensitive ECD as halogen derivatives. Typical fluoro derivatizing reagents include pentafluorobenzyl bro(1)Ripley, B. D.; Chau, A. S.Y. In Analysis of Pesticides in Water. Vol. ZZZ. Nitrogen-Containing Pesticides; Chau, A. S.Y., Afghan, B. K., Eds.; CRC Press: Boca Raton, FL, 1982;Chapter 1. (2)Greenhalgh, R.; Kovacicova,J. J. Agric. Food Chem. 1976,23,325. (3)Bache, C. A.; St. John, L. E.; Liek, D. L. Anal. Chem. 1968,40, 1241-1242. (4)Gutenmann, W.H.; Liek, D. J. J. Agric. Food Chem. 1966,13, 48-50. (5)Boulton, J. J. K.; Boyce, C. B. C.; Jewess, P. J.; Jones, R. F. Pestic. Sci. 1971,2, 10-12. (6) Moye, H.A. J. Agric. Food Chem. 1976,23,415-418. (7)Coutta, R. T.; Hargesheimer, E. E.; Pasutto, F. M. J. Chromutogr. 1980,196,10E-112. (8)Ogierman, L. J.-Assoc. Off. Anal. Chem. 1982,66,1452-1456. 0003-2700/83/0385-1773$04.00/0
mide,8JO heptafluorobutyric anhydride,"JZ trifluoroacetic anhydride,lg and pentafluoropropionic anhydride.14J6 In regard to applications, GC has been used to determine N-methylcarbamates in various types of samples, namely, waters,ll vegetables and cereals,lz olives,15 fruits,16 milk,l'J8 and biological ~ample8.l~ Other chromatographic techniques frequently used for the determination of N-methylcarbamates include TLCm and HPLC.21 Flow injection analysis (FIA) has been employed for the determination of carbaryl,22 carbofuran and malaoxon,23and carbofuran, propoxur, and carbaryl,%which were monitored with a conventional or diode array%spectrophotometer. One method using continuous flow extraction for the determination of organophosphorus pesticides by HPLC with UV detection26 and MS characterization%was reported by Farran et al. Bruchet et al.n reported a method for continuous extraction and determination of different pesticides (between 10 and 100ng/L) by GC-MS using a moving needle injection technique that allowed for injection volumes of up to 50 pL. Several automated extraction-injection systems for gas chromatography have been developed to circumvent the shortcomings of manual procedures, which are tedious and prone to errors arising from contamination. Thus, Ballesteros et al.28 developed a continuous extraction-derivatization system coupled to a gas chromatograph fitted to a flame ionization detector for the determination of a variety of phenols in water samples. Anew automated extraction system combining membrane cell technology and a pneumatically (9)Tjan, G.H.; Jansen, J. T. A. J.-Assoc. Off. Anal. Chem. 1979,62, 769-773. (10)Cline, R. E.;Todd, G. D.; Ashley, D. L.; Grainger, J.; McCraw, J. M.; Alley, C.C.; Hill,R. H.J. Chromatogr. Sci. 1990,28,167-172. (ll)Nagasawa, K.; Uchiyama, H.; Ogamo, A.; Shinozuka, T. J. Chromatogr. 1977,144,77-84. (12)Lawrence, J. F.; Lewis, D. A.; McLeod, H.A. J. Chromatogr. 1977, 138,143-150. (13)Bose, R. J. J. Agric. Food Chem. 1977,25,120%1212. (14)Sherma., J.:. Shafk. T. M. Arch. Enuiron. Contam. Toricol. 1976. 3,5558. (15)Albi, T. Grasaa Aceites (Seuille) 1981,32,381-386. (16)Mattern, G.G.;Singer, G. M.; Louis, J.; Robson, M.; Roeen, J. D. J. Agric. Food Chem. 1990,38,402-407. (17)Bullock, D. In AnalyticalMethodeforPesticidesondPlant Growth Regulators: Vol. VI. Gaa ChromatographicAnalysie; Zweig, G.,Sherma, J., Eds.; Academic Press: London, 1972;pp 478-482. (18)Niehijima, 0.Noyaku-Kenaasho-Hokoku 1984,24,35-43. Hattori, H.; Liu, J.; Seno, H.; Kumazawa, T. Forensic (19)Suzuki, 0.; Sci. Int. 1990,46,169-180. (20)De la Vigne, H.; Janchen, D. J. Planar Chromatogr.-Mod. TLC 1990,3,6-9. (21)Blad, W.Fresenius' J. Anal. Chem. 1991,339,340-343. (22)Yanez-Sedeno, P.; Nova, C.; Polo Diez, L. M.Microchem. J. 1988, 38,370-375. Schmid,R. D. Anal. Chim. Acta (23)Kindervater, R.; Kuennecke, W.; 1990,234,113-117. (24) FernBndez-Band,B.; Linarea, P.;Luque de Castro, M. D.; Valdrcel, M. Anal. Chem. 1991,63,1672-1675. (25)Farran.A.: De Pablo. J.: Hernhdez, S.Anal. Chim. Acta 1988, 221,123-131. (26)Farran, A.; Cortina, J. L.; De Pablo, J.; Barcel6, D. Anal. Chim. Acta 1990,234,119-126. (27)Bruchet, A.; Cognet, L.; Mallevialle, J. ~aterRes.1984,18,14011409. (28)Ballesteros, E.;Gallego, M.; Valcbcel, M. Anal. Chem. 1990,62, 1587-1591. Q 1993 Amerlcan Chemlcal Society
1774
ANALYTICAL CHEMISTRY, VOL. 65, NO. 13, JULY 1, 1993
operated pressurized rotary gas chromatographic injection valve was developed by Melcher and Morabito28 for determination of chlorinated aromatic compounds and pesticides in waters at the part-per-trillon levels. This continuous extraction system does not permit simultaneous derivatization and consists of a number of components such as valves, solenoids, etc.; also, the chromatograph must be altered (the column is channeled through an injection port for connection to the valve), which makes it unusable for routine syringe injections. Other continuous separation techniques coupled to gas chromatography include continuous liquid-gas-phase separation combined with electron capture detection30 for the continuous generation and determination of volatile species such as chlorine and sulfur dioxide, and with a nitrogen selective detector31 for the enrichment and determination of aliphatic amines. In this work we used a continuous liquid-liquid extractionderivatization module, described elsewhere,28coupled to an electron capture detection system (not used previously with a continuous extraction-derivatization module) for the determination of N-methylcarbamate pesticides in samples with complex matrices, such as milk. The basic aim is to apply the on-line GC flow system to real samples in order to minimize sample pretreatment and accomplish the derivatizationextraction of pesticides with increased detector sensitivity. The automatic method developed requires cow's milk spiked with N-methylcarbamates to be pretreated prior to introduction into the flow derivatization-extraction module; however, the treatment is shorter than that involved in the recommended batch procedure. EXPERIMENTAL SECTION Materials. The chemicals used as analyte standards and reagents were reagent grade or better. N-Methylcarbamates were purchased from Dr. Ehrenstorfer,Augsburg,Germany. 1,2,3,4,5,6y-Hexachlorocyclohexaneand pentafluoropropionic anhydride were obtained from Fluka, and all others reagents (sodium hydroxide,acetone,n-hexane,acetonitrile,n-pentane, ethyl ether) were supplied by Merck. A standard stock solution of propoxur, carbofuran, and aminocarb (2 g/L), benthiocarb (1.4 g/L), carbaryl (1 g/L), and methiocarb (0.2 g/L) was prepared in 99.9% acetone and stored in PTFE bottle at 4 "C. Appropriate volumes of this stock solution were diluted with Milli-Q water to prepare more dilute solutions containing a few nanograms per milliter of each N-methylcarbamates in 6 X 103M sodium hydroxide (hydrolysis reagent). An n-hexane solutioncontaining 500 ng/mL 1,2,3,4,5,6y-hexachlorocyclohexane (internal standard) and 0.75% (v/v) pentafluoropropionicanhydride (derivatizationreagent)was used as extractant. Equipment. An electron capture detector ("Ni) installed in a Hewlett-Packard Model 5890 A gas chromatograph was used. Chromatographic assays were performed on a cross-linked 50% phenyl-50 % methyl polysiloxane (film thickness 2.0 pm) fusedsilica column (10m X 0.53 mm i.d.) supplied by Hewlett-Packard (HP-17). The flow rate of the carrier gas (nitrogen) was 35 mL/ min. The injector temperature was kept at 150 "C, and the detector at 300 "C throughout. The column temperature was raised from 60 "C at 11 "C/min to 105 "C (1min), at 8 "C/min to 160 O C , and then at 25 "C/min to 250 "C, for 3 min. Peak areas were measured with Hewlett-Packard 3392 A integrator. The flow extraction-derivatization system consisted of a peristaltic pump (GilsonMinipuls-a),an A-1OTsolvent segmenter(Tecator), and a custom-made phase separator furnished with a Fluoropore membrane (1.0-pm pore size, FALP; Millipore) described else(29) Melcher, R. G.; Morabito, P. L. Anal. Chern. 1990,62,2183-2188. (30) Novic, M.; Zupancic-Kralj, L.; Pihlar, B. Anal. Chirn. Acta 1991, 243, 131-137. (31) Grbnberg, L.; LGvkvist, P.; JGnsson, J. A. Chrornatographia 1992, 33.77-82.
SAMPLE
EX 1 R AC TA NT
-
PS
c3
c3
IYI
U
Flgure 1. Flow diagram of the derlvatlzation-extraction system: PS,
phase separator; DC, desiccating column; IV, injection valve; HS, heating system; IP, injection port; ECD, electron capture detector; I, integrator; S, tube stopcock: W, waste.
where.32 Poly(viny1 chloride) and Solvaflex pumping tubes for aqueous and n-hexane solutions, respectively,and Teflon tubing for coils were also used. A six-port switching valve (Knauer 6332000) was mounted over the injection port of the gas chromatograph in which the original volume of the valve (10pL) was changed using a loop of 2.5 pL. Figure 1 shows a diagram of the extracting-derivatizing analyzer combined with the electron capture detection system which allows samples to be introduced. The interface between the extraction-derivatization system and the gas chromatograph was described elsewhere.% The injection interface was constructed from a standard injection valve, originallydesigned for HPLC, which permits the injection of 5 pL (internal volume plus loop volume, 2.5 and 2.5 pL, respectively). The connection between the injector and the instrument port consisted of a 0.5 mm i.d.by 25 cm long section of stainless steel tube, with a needle soldered at the end, directly inserted into the septum of the injection port. This tube linking the valve and the injection port was heated using a wire coiled helically around a hollow ceramic tube which could be heated between 25 and 175 "C. The carrier gas inlet was split into two streams that were directly connected to the valve and the chromatograph injection port with flow rates of 25 and 10 mL/min, respectively. The inlet was shut by a stopcock so the instrument could be used for manual injections, allowing the nitrogen stream to follow its normal route through the instrument. The system used n-hexane extractant and an aqueous sample containing trace levels of the six N-methylcarbamates. Milk Pretreatment. The cow's milk sample was treated by adding different amounts of pesticides to 10 mL of milk. The sample and 5 mL of pentane were placed in a separatory funnel and shaken briefly; 20 mL of ethyl ether was added and the mixture was shaken gently for 2 min. Once the two phases were decanted, the organic layer was transferred to a 50-mL glass vessel and evaporated to dryness. The residue was dissolved in 3 mL of acetonitrile and 0.3 mL of 0.5 M NaOH and then transferred to a 25-mL calibrated flask and diluted to the mark with Milli-Q water. Derivatization-extraction was done as described below. Procedure. The flow system used in this work is shown in Figure 1. It operates as follows: an alkaline sample containing between 5 and lo00 ng/mL N-methylcarbamates (in 6 X 109 M NaOH) is continuously introduced into the system and mixed with a stream of extractant (0.75 % pentafluoropropionic anhydride and 500 ng/mL 1,2,3,4,5,6-y-hexachlorocyclohexanein n-hexane). Derivatization and extraction take place simultaneously in a knotted tube reactor of 330 cm. The pentafluoro derivatives of the N-methylcarbamates (from the membrane separator) are continously circulated through the valve. The loop contents (5 pL) are transferred to the chromatograph port by the nitrogen carrier. The section of the tube valve port is (32) Gallego, M.; Silva, M.; Valchcel, M. Anal. Chern. 1986,58,22652269.
ANALYTICAL CHEMISTRY, VOL. 65, NO. 13, JULY 1, 1993
heated at 105 O C in order to prevent adsorption of the extract on the tube walls during transport. The ECD characteristics were affected by water vapor, so a desiccating column (5cm long X 3 mm i,d., packed with sodium aluminosilicate pellets) was placed prior to the valve to prevent water from reaching the column or detector.
RESULTS AND DISCUSSION Selection and Optimization of the Reagent Concentrations. This study was done with a manifold similar to that depicted in Figure 1. Extracts of derivatized N-methylcarbamates from the membrane separator were collected in 4-mL glass vials containing anhydrous sodium sulfate, and 2-pL fractions were injected manually into the chromatograph using a syringe. The direct determination of N-methylcarbamates with ECD detection was impossible, so their halogen derivatives were obtained. We chose three derivatizing reagents, viz. trifluoroacetic anhydride, pentafluoropropionic anhydride, and heptafluorobutyric anhydride, for this purpose. The extractant was prepared by dissolving 1 mL of derivatizing reagent in 50 mL of n-hexane and the sample containing the sixN-methylcarbamates (1pg/mL propoxur, carbofuran, and aminocarb; 0.7 pg/mL benthiocarb; 0.5 pg/mL carbaryl and 0.1 pg/mL methiocarb) plus 0.05% pyridine catalystlJ5 in 0.01 M sodium hydroxide. The derivatization-extraction coil was heated a t 40 O C to favor reaction development. Aminocarb and methiocarb provided overlapped signals with trifluoroacetic anhydride, so they could not be resolved; on the other hand, heptafluorobutyric anhydride provided diminished sensitivity, so only carbofuran and aminocarb could be identified as the other pesticides were not derivatized or detected under the assayed conditions. Pentafluoropropionic anhydride was the best derivatizing reagent because it permitted the identification of all six pesticides with good sensitivity. According to the literature,' heptafluorobutyric anhydride should have resulted in the highest sensitivity; this was not the case in our experiments. We believed it to be inadvisable to conduct any further investigations with this derivatizing reagent because of its high cost, so we chose to use pentafluoropropionic anhydride instead. Four organic solvents (ethyl acetate, n-hexane, ethyl ether, dichloromethane) were assayed as extractants. n-Hexane and dichloromethane were found to be the most efficient for simultaneous acylation with pentafluoropropionic anhydride and extraction. With ethyl acetate and ethyl ether, the resulting baseline of the chromatogram was too high, probably because of the higher solubility of these two solvents in water, so separation was incomplete and the water vapor reaching the detector caused the observed perturbation. Dichloromethane was discarded since it caused the Teflon membrane of the phase separator to rapidly deteriorate, so n-hexane was finally chosen for derivatization-extraction of the pesticides. Two internal standards (aldrin, 1,2,3,4,5,6-yhexachlorocyclohexane) were selected among the most frequently used for this purpose. 1,2,3,4,5,6-y-Hexachlorocyclohexane was chosen because its peak lies nearest to those of the last pesticides. As noted earlier, pyridine is a catalyst for most acylation reactions;lJ5 it neutralizes the released acid and shifts the reaction equilibrium to product formation. We determined the optimal concentration of pyridine; for this purpose we used several samples containing different concentrations of pyridine between 0 and 0.35% (v/v) in 0.01 M NaOH. The peak areas diminished slightly with increases in the pyridine concentration (up to 0.1 9%) and then decreased dramatically. Experiments done on a sample with a lower NaOH concentration (0.005 M) provided poor results a t pyridine concentrations above 0.2%. In the manual procedure,'J5 the
4
W
0.25>
1775 1
;o.20 P
2
0.15
w 2 > L
0.10
a E
-
-2 +
.E
-I
U
4 -
5e
0.05 0.00 0
0005
0.0 1
0 015
0.02
(NaOH) , M
Figure 2. Influence of the NaOH concentration on the hydrolysls reaction: (1) methlocarb, 50 ng/mL; (2) carbaryl, 250 ng/mL; (3) aminocarb, 500 ng/mL; (4) propoxur, 500 ng/mL; (5) benthlocarb, 350 ng/mL; (6) carbofuran, 500 ng/mL. For GC condltlons, see text. 0.57
0
I
I
2
3
PENTAFLUOROPROPlONlC ANHY ORlOE , %
Figure 3. Effect of the pentafluoroproplonlc anhydride concentration (%) on the derlvatlzatlon reaction: (1) methlocarb; (2) carbaryl: (3) benthlocarb; (4) aminocarb; (5) carbofuran; (6) propoxur. Pesticlde concentrations as In Figure 2. N-methylcarbamates are hydrolyzed in an NaOH medium. Subsequently, the solution is evaporated and the residue is extracted into an organic solvent, after which pyridine and the derivatizing reagent are added-pyridine is thus required to neutralize the acid produced in the derivatization reaction. Our method requires no pyridine as the samples are aspirated into the derivatization-extraction module in an alkaline medium. For this reason, no pyridine was added to samples. The reagent concentrations (NaOH and pentafluoropropionic anhydride) were optimized using a sample containing the six N-methylcarbamates (500 ng/mL propoxur, carbofuran, and aminocarb; 350 ng/mL benthiocarb; 250 ng/mL carbaryl and 50 ng/mL methiocarb) in sodium hydroxide; the extractant used was n-hexane containing 2 % (v/v) pentafluoropropionic anhydride and 500 ng/mL 1,2,3,4,5,6-y-hexachlorocyclohexane. The effect of the NaOH concentration was studied between 0 and 0.016 M. As can be seen in Figure 2, the NaOH concentration was critical for some pesticides (methiocarb, carbaryl, benthiocarb) and less so for others. We selected a concentration of 6 X 1 PM (pH 11.8),which was the optimal compromise value for most of the pesticides (benthiocarb expected). The influence of the pentafluoropropionic anhydride concentration on the yields of the derivatives was investigated using several n-hexane solutions. The results obtained are shown in Figure 3. The chromatographic signal increased with increase in pentafluoropropionic anhydride concentration up to 0.759% ,except for propoxur, above which it decreased (however, the decrease may also be due to the fact that an increase in the pentafluoropropionic anhydride concentration also increases acidity in the reaction-extradion coil and, hence, the NaOH concentration required to neutralize it). This was continued using a sample with a higher
1776
ANALYTICAL
i
CHEMISTRY, VOL. 65, NO. 13, JULY 1, 1993
2 min
H I
8
Flguro 4. Gas chromatogramsof the pentafluoropropbnatederivatives
of Kmethyicarbamates obtained by heating the tube port at (A) 80, (B) 105,and(C)150°C. Peaks: (l)derhratizingreagent;(2)benthiocarb; (3)propoxur; (4) carbofuran;(5) amlnocarb;(6)carbaryi;(7) methlocarb; (8) internal standard. Pestlcide concentrations as in Figure 2. sodium hydroxide concentration (1 X M); the optimal concentration of derivatizing reagent was also increased. Hydrolysis and Acylation Mechanism. There is much confusion about the derivatization of N-methylcarbamates because many derivatizing reagents can act directly on the pesticides or their hydrolysis products (phenols). In fact, many direct derivatization procedures for these pesticides are questionable because few reagents or procedures are specific enough to derivatizeonly one type of functional group; attempts at derivatizing the carbamate NH function may also result in derivatization of a hydroxyl group in the molecule.' In order to check this point in our proposed method, we carried out the following experiments in quintuplicate using carbaryl and its hydrolysis product (phenol) 1-naphthol as N-methylcarbamate prototypes. The first experiment involved two samples of carbaryl and 1-naphthol with similar concentrations of each in 6 X M NaOH, a solution of 0.75 % pentafluoropropionic anhydride in n-hexane as extractant, and the manifold depicted in Figure 1; the chromatographic signals obtained, with identical retention
times for both samples, were analogous. The experiment was repeated with the sample in 1X 1o-S M NaOH; the analytical signal corresponding to the carbaryl sample was about 50% smaller while that of 1-naphthol was about 30% higher compared to that of the sample in 6 X 1o-SM NaOH. Finally, the same experiment was carried out with two samples (carbaryl and 1-naphthol) containing no NaOH (pH 6.5); no signal was obtained for the carbaryl sample, and that yielded by 1-naphthol was slightly lower than the previous one. The same experiment was also carried out with the N-methylcarbamate pesticide carbofuran and its corresponding phenol, 3-hydroxycarbofuran; similar results were obtained. From the three above experiments, we may conclude the following: (1) Derivatization with pentafluoropropionic anhydride affects the hydroxyl group rather than the amine group. Hydrolysis of the N-methylcarbamates prior to their derivatization with this reagent was thus required. (2) The derivatization of the hydrolysis products of the pesticides (phenols) was favored by alkaline media of pH lower than those for the intact pesticides. These results show that the derivatizing reaction occurs via the hydroxyl group according to following mechanism:
Optimization of t h e Flow System and Sample Introduction Device. The effect of the flow rates of sample and extractant was studied over the ranges 0.4-2.8 and 0.2-1.6 mL/min, respectively. Increasing sample flow rates (aqueous phase) at a constant organic flow rate resulted in increasing peak areas through increased preconcentration ratios; for the same reason, peak areas also obviously increased with decreasing organic phase flow rate (extractant). Sample and extractant flow rates of 2.7 and 0.7 mL/min, respectively, were selected as a compromise between adequate reproducibility, preconcentration ratio, and sampling frequency. The influence of the length of the derivatization-extraction coil was studied between 50 and 530 cm. Above 250 cm, peak areas remained constant throughout the studied interval. A coil length of 330 cm (0.5-mm i.d.1 was chosen which yielded a residence time of 12s. As some manual procedures required heating for derivatizationof pesticideswith acylation reagents, we studied the effect of the extraction coil temperature between 15 and 90 "C using a thermostated water bath. The analytical signal remained virtually constant up to 45 "C, above which it decreased dramatically, probably as a result of the derivatizing reagent decomposingabove 45 "C or heating fostering volatilization of the extractant (the boiling point of n-hexane and pentafluoropropionic anhydride is ca. 70 "C). Therefore, the extraction coil was maintained at room temperature.
ANALYTICAL CHEMISTRY, VOL. 65, NO. 13, JULY 1, 1093
Table I. Features of the Calibration Graphs and Determination of N-Methylcarbamates correlation range N-methylcarbamate regreeaion equation4 coefficient (ng/mL) benthiocarb propoxur carbofuran aminocarb carbaryl methiocarb 0
Y = 6.85 X lex + 6.47 X le2 Y = 1.91 X 1WX + 2.36 X lez Y = 4.30 X 1O-'X - 1.38 X le2 Y = 4.39 X lWX + 1.75 X Y = 1.47 X 1vX + 1.13 X lv Y = 7.81 X lvX + 3.97 X
0.995 0.999 0.998 0.999 0.994 0.999
35-700 50-800
50-1000 5 0 - ~ 26500 5-100
1777
detection
limit (ng/mL)
RSD (%)
14 20 20 20 10 2
3.68 2.75 3.70 3.35 2.35 3.35
Y,analyte/internalstandard peak area ratio; X,concentration, ng/mL.
The sample introduction device (interface unit between the extraction system and gas chromatograph) was an injection valve modified as described in a previous paperm which permits the introduction of 5 MLof the organic extracts into the injection port of the chromatograph. By using the heating system described in the Experimental Section, the tube valve port was heated between 60 and 150 "C. Figure 4 shows the chromatograms obtained by injecting extracts of the six derivatized pesticides in n-hexane. The first chromatogram (Figure 4A) was recorded with automatic injection a t 80 "C, the second (Figure 4B) at 105 OC (optimal value), and the third (Figure 4C) at 150 "C. Heating the tube valve port was necessary to avoid adsorption of the analytes on the inner walls of the tube; however,the temperature should not exceed 105 OC, otherwise the baseline increases with the analysis time. As shown in Figure 4, excess pentafluoropropionic anhydride posed no problem. Figures of Merit. The dependence of the signal on the concentrations of the six pesticides was only determined at one integrator sensitivity using the experimentally determined optimum conditions and the system depicted in Figure 1. The standard curves obtained and their figures of merit are summarized in Table I. The detection limit was calculated as the concentration yielding the minimum detectable signal in the chromatogram, and the relative standard deviations were obtained measuring 11samples containing ca. 300-400 ng of each pesticide per milliliter and 50 ng/mL methiocarb. Application to Spiked Cow's Milk Samples. We found only two previous references to the determination of N-methylcarbamates in milk samples. In one of them,l' carbaryl was determined in cow's milk by extraction with pentaneether, desiccation, and filtration of the organic phase; purification with keeper solution and evaporation to dryness, dissolution of the residue in acetonitrile and cleaning of this layer three more times with n-hexane, addition of another volume of keeper solution to the acetonitrile, and evaporation to dryness and dissolution of the residue in dichloromethane; and deriuatization with trichloroacetyl chloride and detection with a W r ECD. The derivatization step was the longest as it involved a number of steps including extraction, heating, cooling, rinsing, and evaporation; therefore, the method entails so much manipulation that ita precision and sampling frequency (a few hours per analysis) are rather poor. Recoveries from milk fortified with 5-100 ng/mL carbaryl ranged from 70 to 110%. The second procedurela involves solvent extraction, chromatography on Amberlite, cleanup on a silica cartridge, and analysis by GC with an N-P detector. In order to apply the proposed method we initially studied the influence of the sample matrix by assaying cow's milk diluted with water (in 6 X 10-8 M NaOH) as blank and n-hexane as extractant. The results obtained by using the manifold depicted in Figure 1 were satisfactory and the chromatograms were similar to those obtained by a blank of 6 X 109 M NaOH up to a milk/water ratio of 1/2.5; higher volume ratios clogged the membrane separator with the organic matter from milk. For this volume ratio, recoveries from milk fortified with 80-800 ng/mL pesticides ranged only
Table 11. Recovery of N-Methylcarbamates Added to Cow's Milk Samples concns (ng/mL) recovery sample N-methylcarbamate added" found4 (%) 1
2
3
4
4
benthiocarb propoxur carbofuran aminocarb carbaryl methiocarb benthiocarb propoxur carbofuran aminocarb carbaryl methiocarb benthiocarb propoxur carbofuran aminocarb carbaryl methiocarb benthiocarb propoxur carbofuran aminocarb carbaryl methiocarb
0.0 0.0 0.0 0.0 0.0 0.0 140 200 200 200 100 20 280 400 400 400 200 40 560 800 800 800
400 80
0.0 0.0 0.0 0.0 0.0 0.0 136.9 189.0 199.2 196.7 97.8 19.4 269.8 396.8 396.4 401.9 193.4 39.3 524.9 788.3 806.4 794.7 392.1 75.4
97.8 94.5 99.6 98.3 97.8 97.0 96.4 99.2 99.1 100.5 96.7 98.2 93.7 98.5 100.8 99.3 98.0 94.2
Concentrations referred to the diluted milk samples.
from 20 to 30%; therefore, the milk samples could not be analyzed directly; some pretreatment was required. Samples were pretreated by adding different amounts of pesticides to 10 mL of milk. The extraction in pentane-ethyl ether was made according to the recommendations for the manual method, yet there was difficulty dissolving the residue obtained upon evaporation of the pentane-ethyl ether. Two water-misciblesolvents, ethanol and acetonitrile, were assayed for this purpose. The best results were obtained with acetonitrile, but the volume usedto dissolve the residue should not be higher than 3 mL, otherwise pesticide recoveries from the spiked milk were only ca. 60%. Experiments done with aqueous N-methylcarbamate samples (25 mL) in 6 X 10-8 M NaOH and acetonitrile volumes between 0 and 8 mL provided poor results for acetonitrile volumes above 3 mL, probably because the derivatization reaction was disturbed (in the manual method, acetonitrile is evaporated prior to the derivatization step). Thus, 3 mL of acetonitrile was selected for dissolution of the residue obtained from 10 mL of milk. The applicability of the method was tested by resolving various cow's milk samples spiked with a few nanograms per milliter of the pesticides. Table I1lists the average recoveries obtained under the optimal working conditions for four milk samples containing different concentrations of the N-methylcarbamates; each sample was analyzed in quadruplicate (n = 4). Recoveries ranged between 93.7 and 100.8% in all instances. The concentrations added to the milk samples were 2.5 times higher than those listed in Table I1if one takes into account that the 10 mL of milk was diluted to a final
1778
ANALYTICAL CHEMISTRY, VOL. 65, NO. 13, JULY 1, 1993
volume of 25 mL. Finally, in order to determine whether carbaryl and naphthol in milk could be distinguished, two different experiments were performed. In the first, two milk samples containing 0.5 pg/mL naphthol or carbaryl but no NaOH (pH 6.5) were processed as described under the Experimental Section. The second sample was adjusted to pH 9.2by adding 0.5mL of pyridine. Neither sample yielded the signal for carbaryl; the prior hydrolysis required for derivatization did not take place. On the other hand, naphthol was derivatized in both instances, the maximum signal being obtained when the pyridine alkaline medium was used-the peak area was 2-fold that obtained in its absence. Therefore, the proposed method allows naphthol and carbaryl to be distinguishedsince alkaline hydrolysis identifies between the two and only naphthol is determined in the presence of pyridine.
CONCLUSIONS The proposed automatic method has some major practical implications. It allows the efficient continuousderivatization and extraction of six pesticides using an ordinary liquidliquid extractor. Fitting the flow system to a gas chromatograph is quite simple and requires no alterations, so the instrument can still be used for routine syringe injections. Compared with the manual derivatization-extraction procedure, the proposed method has several valuable advantages,
particularly reduced sample manipulation, which results in increased precision and sampling frequency. GC determinationsof N-methylcarbamateshave so far been concerned with such simple matrices as water and vegetables or fruits, which poae no special problems as the added pesticide can be readily extracted. On the other hand, more complex matrices such as milk make the process rather laborious and time-consuming. The proposed method allows the indirect determination of the pesticides in milk by simply pretreating the sample prior to introduction into the flow system. The sole pitfall of our method lies in its Sensitivity, which is somewhat lower than that reported for carbaryl" (the manual method requires 100 mL of milk containing 0.1-50 pg of insecticide that can be taken to 1 mL of n-hexane);however, the sensitivity of the proposed method can be increased 10fold by evaporating 5 mL of the organic extract from the phase separator with dry Nz (cryogenic concentration) to 0.5 mL of solventand performing a subsequent manual injection.
ACKNOWLEDGMENT This work was supported by Grant PB90-0925from the Spanish CICYT.
RECEIVED August 7,1992. Accepted November 27,1992.
