ANALYST, JANUARY 1993, VOL. 118 Extraction of Organophosphorus Pesticides From Soil by Off -line Supercritical Fluid Extraction Klaus Wuchner, Rudy T. Ghijsen and Udo A. Th. Brinkman Department of Analytical Chemistry, Free University, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands Robert Grob and Jacques Mathieu ENSCT, Laboratoire de Chimie Analytique, 3 18 Route de Narbonne, 31077 Toulouse Cedex, France Supercritical fluid extraction (SFE) conditions were optimized for the isolation of organophosphorus pesticides (OPPs) from soil. Results using pure carbon dioxide and carbon dioxide modified with acetone, ethyl acetate or methanol were compared in terms of recoveries, extraction rates and matrix effects. Despite the good solubility of the OPPs in pure C02, the addition of methanol was necessary t o achieve high recoveries (390%) when spiked soil was extracted.The extent t o which matrix effects, which probably cause the decrease in recoveries, occur depends on both the analyte polarity and the spiking method used. Similar extraction efficiencies were achieved by adding microlitre amounts of the modifier directly t o the soil in the extraction cartridge and by using a pre-mixed gas cylinder. SFE of OPPs in soil was compared with solvent extraction. Keywords: Supercritical fluid extraction; organophosphorus pesticides; soil; optimization; matrix effects The growing interest in supercritical fluid extraction (SFE) as a sample preparation method in analytical chemistry is reflected by the large number of papers and comprehensive reviewsl-3 published and by the many contributions to symposia4.5 in recent years.SFE is increasingly employed as an alternative to the more tedious classical extraction proce- dures, such as Soxhlet and manual solvent extractions with organic solvents. These classical extraction techniques are often the limiting step in the determination of organic pollutants in environmental solids and cause a considerable contribution to waste production in the laboratory. On the other hand, SFE has been shown to be an efficient and rapid technique for the isolation of polychlorinated biphenyls (PCBs),h-") polycyclic aromatic hydrocarbons (PAHs), chlorinated dibenzodioxins,lJ halocarbons15 and other organic pollutantslcp18 from sediments and soils. The widespread use of agricultural chemicals, with more than 1000 pesticides in common use,lg demands efficient and practical analytical methods for the inventory and assessment of the spread of these hazardous chemicals in the environ- ment.There is particular interest in the monitoring of pesticide residues in soil. The utility of SFE for the determi- nation of pesticides has been demonstrated for urea herbi- cides,20-22 chlorinated insecticides,'"24 triazines25 and phen- oxyace tic esters .Zh An important group of pesticides, the organophosphorus pesticides (OPPs), have been the subject of only a few SFE studies, despite their worldwide application in agriculture and their high mammalian toxicity. Ethion was extracted from a freeze-dried grape sample with pure C 0 2 to demonstrate the potential of an on-line coupling of SFE and GC.27 Cortes et a1.28 used methanol-modified C 0 2 for the determination of chlorpyrifos in grass samples.The pesticide was quantitatively recovered and interferences could be eliminated by coupling SFE with micro-LC/GC. Bergna et ~ 1 . 2 9 showed that the extraction of disulfoton and tolclofos methyl from soil was feasible by on-line coupling of SFE and GC. The extractability of a larger number of OPPs with pure and modified C 0 2 under various extraction conditions was dis- cussed by Lopez-Avila rt al.23 Dimethoate, azinphos methyl and coumaphos could not be recovered by pure C02; diazinon was not even extractable with 10% methanol-modified C02. The analytes were added to sand, which has been proposed as a non-sorptive matrix in method development of SFE.30 Other studies showed a decrease in the extraction efficiency due to solute-matrix interactions when relevant environmen- tal samples, such as soil or sediments, were extracted.'".l"26." Obviously, there still is a need for an SFE procedure that permits the efficient isolation of a large number of OPPs from soil.The main objective of this study was the development of an efficient SFE method for the rapid determination of OPPs of varying polarity in soil. The selected set of OPPs covers a wide range of polarity, as indicated by their octanol-water partition coefficient, Kow; log KO, ranges from 0.5 for dimethoate to 5.1 for carbofenthion.32 The extraction conditions, such as pressure and the use of modifiers, were optimized for different matrices (quartz wool, sand, soil) in order to achieve high recoveries (>SO%) in less than 1 h.Two different spiking methods were tested to study analyte-matrix interactions. The results were mainly discussed under the aspect of matrix effects. Practical aspects, such as the method and conditions of modifier addition, were emphasized. No attempt has been made to appreciate fundamental thermodynamic and kinetic SFE parameters, for which some basic treatments and models are available in the literature. Finally, the SFE results were compared with those obtained using a classical solvent extraction method. Experimental Chemicals Diazinon, disulfoton, dimethoate, malathion, parathion ethyl, carbofenthion, azinphos methyl and coumaphos of 9649% purity were obtained from various sources. Ethyl acetate, acetone (both of analytical-reagent grade), methanol (HPLC grade) and anhydrous sodium sulphate (analytical- reagent grade) were purchased from J .T. Baker (Deventer, The Netherlands) and used without further purification. Carbon dioxide (purity 299.99%) and pre-mixed methanol- modified C 0 2 (2% m/m) were supplied by Hoek Loos (Schiedam, The Netherlands) and Intermar (Breda, The Netherlands), respectively. Calibration standards were prepared in ethyl acetate by serial dilution of a stock solution of the OPPs in acetone (0.5 mg ml- 1). The stock solution was used to spike the various matrices. All solutions were stored at 4°C in the dark.12 ANALYST, JANUARY 1993, VOL. 118 Matrices and Spiking Methods Soxhlet-extracted glass-wool was used as an inert matrix.A purified seasand, purchased from J. T. Baker, was also studied. A dried soil with a particle size distribution of <2 pm 1.6%, 2-38 pm 3.1%, 338 pm 88% was used throughout this study. It contained 3.3% of total organic carbon. Two methods were used to prepare a contaminated soil at the 4 and 40 ppm spiking levels. In the spot method, approximately 250 mg of soil were weighed into the extraction cartridge. The soil was then fortified with a known amount of the analytes by adding 20 p1 of the stock solution directly to the soil. In the slurry method, 10-20 g of soil were weighed into a 50 ml flask. After the addition of 20 ml of acetone and an appropriate volume (80 p.1-1.6 ml) of the spiking mixture in acetone, the flask was sealed and its contents were stirred for 3 h.With both methods, the solvent was evaporated to dryness at ambient temperature in a hood for at least 20 h. The completeness of the drying step was controlled by weighing the sample. SFE System All extractions were performed on a laboratory-built SFE system, represented in Fig. 1. The pump heads of the syringe pump (1) (Chrompack, Middelburg, The Netherlands) are cooled to 5 "C. The extracting phase is preheated in a coil (2) and enters the extraction cartridge (4), a 20 X 4 mm i.d. HPLC precolumn (Chrompack), via a stop valve (3) (Alltech, Deerfield, IL, USA). The stop valve is closed during pressurization of the pump and decompression of the extrac- tion vessel. The six-port valve (5) (Valco, Schenkon, Switzer- land) permits selection between two restrictors (6) or the flow of the supercritical fluid to be stopped during static extraction.When a dynamic extraction is performed, the supercritical pressure in the extraction cartridge is maintained by using fused-silica capillaries. The dimensions of the capillaries, which determine the flow rate at a given pressure and temperature, were 15-25 cm x 25 pm i.d. (Chrompack) and 80-100 cm x 50 pm i.d. (SGE, Ringwood, Australia). 9 GC oven 8 Fig. 1 Schematic diagram of the off-line SFE system The restrictors were introduced via a septum into a vial containing 2 ml of ethyl acetate, which contained phorate as an internal standard. The internal standard is used to correct for the loss of collection solvent during extraction.Solvent losses were minimized by condensation of the vapours in a reflux cooler (7). The extraction temperature was regulated by a GC oven (8) (HP 5840A, Hewlett-Packard, Palo Alto, CA, USA). A short 30 cm X 0.25 mm i.d. PEEK capillary (9), connected to a 500 pl syringe (Hamilton, Reno, NV, USA), permitted 50 pl aliquots of the trapping solvent to be taken during extraction. In this way extraction profiles, i . e . , plots of percentage recovery versus extraction time or volume of C02, could be easily monitored. SFE Procedures Extractions were generally performed at 50°C and 250 bar, i.e., at a density of 0.84 g ml-1 for pure C02. The collection solvent and extraction cartridge were heated for 5 min at SO "C before the extraction was started. Dynamic extractions with pure or pre-mixed methanol-modified C02 were carried out at about 0.8 ml min-1 after a static equilibrium period of 1 min.Steady-state extractions were performed by closing the outlet valve (5) for 2 min. The extracted analytes were then transported dynamically to the collection solvent with 1 ml of C02. This procedure was repeated 2-5 times. The effect of modifier on the recovery was determined by adding 5-70 pl of the modifier with a syringe directly to the matrix in the cartridge. In that event, only steady-state extractions were carried out to permit the dissolution of the modifier in the supercritical fluid. The volume and the flow rate of the supercritical fluid were read at the pump and correspond to its fluid state. Conventional Solvent Extraction The spiked soil samples were also extracted by a conventional solvent extraction method, currently used in a laboratory in The Netherlands for the determination of OPPs in A portion of 5 g of soil was placed in a 50 ml Erlenmeyer flask and 20 ml of ethyl acetate were added.The flask was sealed with a septum-lined cap and shaken for 90 min. After decantation, the organic layer was transferred into a closed centrifuge tube. A second portion of 15 ml of ethyl acetate was then introduced into the Erlenmeyer flask. The suspension was shaken for another 90 min and equilibrated overnight. The combined extracts were centrifuged for 10 min at 2000 rpm . GC Analysis The SFE collection solvent and extracts from the conventional extraction technique were dried with anhydrous sodium sulfate and analysed by GC with flame ionization detection (FID) (PU 4450, Pye Unicam, Cambridge, UK), with on-column injection.A 1 pl volume was injected into a 2 m x 0.32 mm i.d. diphenyltetramethyldisilazane (DPTMDS)- deactivated retention gap (B. Schilling, Zurich, Switzerland), connected to a 15 m X 0.25 mm i d . DB-1701 fused-silica column (J & W, Folsom, CA, USA) via a press-fit connector. A Hewlett-Packard 5890 Series I1 GC with a pressure programmable on-column injector was used for thermionic detection. Results and Discussion In the development of an SFE method, several steps can be discerned. First, the pressure, temperature and supercritical fluid composition have to be optimized. To that end, the analytes should be extracted from a simple inert matrix such as glass-wool, filter-paper or sand,3(J to prevent matrix effects.ANALYST, JANUARY 1993, VOL.118 13 These experiments are also necessary for the evaluation of the trapping efficiency of the collection system. Second, a spiked matrix of interest should be extracted in order to reveal the influence of the matrix on the extraction efficiency and to detect interfering co-extractives. It may well be that the extraction conditions will have to be adjusted to overcome matrix effects. Finally, the reliability of the SFE method can be tested by extracting a certified reference material. Extraction of OPPs From Inert Matrices The extractability of the OPPs of interest by pure C 0 2 was determined by dynamically extracting an inert matrix, glass- wool, at 50 "C using pressures ranging from 100 to 250 bar.The spiking solvent was evaporated prior to SFE extraction to avoid changes in the solubility characteristics of the super- critical C02. All OPPs were quantitatively recovered (295%) with only 1.5 ml of liquid CO2 at 250 bar. The analytes could also be dissolved in less dense C02 (50°C' 100 bar), but the extraction rate was lower and at least 3.5 ml of C02 were required. No attempt was made to study the influence of temperature on the extraction efficiency. First, the recoveries obtained by Lopez-Avila et al.23 at 60 and 70 "C were not higher than those achieved at 50°C in our study. Moreover, an increase in temperature will cause a greater loss of collection solvent during extraction, which may adversely affect the trapping efficiency.The high recoveries from the inert matrix demonstrate that ethyl acetate (2 ml) is an efficient trapping liquid. Flow rates between 0.3 and 1 ml min-1 could be applied without any noticeable loss (22%) of the extracted pesticides. In order to compare the performance of our system with that of Lopez-Avila et aZ.,23 a spiked sand sample was extracted dynamically with pure CO2 at SO "C and 250 bar. The recoveries, shown in Fig. 2, are comparable (394%) to those achieved in the experiment with glass-wool. Apparently, the investigated sand sample does not influence the analyte recovery. Better results were found in our study, especially for dimethoate, diazinon, azinphos methyl and coumaphos, which could not be recovered at all with pure C02 in the study of Lopez-Avila et al.The differences in recoveries may be caused by the different nature of the sand samples or more likely the poorer trapping qualities of hexane. 1 nn 1 2 3 4 5 6 7 8 OPP sand, spot spiked soil, spot spiked soil, slurry spiked Fig. 2 Influence of matrix and spiking method on OPP recovery (n = 3). Extraction conditions: 250-300 mg of the spiked material (40 ppm; approximately 10 pg of each compound) were extracted dynamically with pure C02 at 50 "C and 250 bar (25 MPa) at a flow rate of 0.7-0.9 ml min-l. The samples were extracted 24 h after spiking. OPPs: 1, diazinon; 2, disulfoton; 3, dimethoate; 4, malathion; 5 . parathion ethyl; 6, carbofenthion; 7, azinphos methyl; and 8, coumaphos SFE of Spiked Soil Samples, Matrix Effects Spiked soil samples were extracted dynamically with pure C02 at 50 "C and 250 bar to determine the extraction efficiency for a relevant environmental sample.Two spiking methods were used. Although the spot method is frequently used in SFE studies,13,16Jg the less commonly used slurry method143'7 should provide a more realistic evaluation of the influence of solute-matrix interactions on extraction efficiency. In the spot method, a few microlitres of the spiking solution are added directly to the soil in the extraction vessel; the analytes are present in a narrow region, a spot. In our work, the acetone used as the spiking solvent had completely disappeared after 1-2 h. A further distribution of the solutes over the soil sample can only be due to volatilization from the soil surface.However, migration of semi-volatile chemicals, such as OPPs, by volatilization from a dry soil is a very slow process .34 In the slurry method, a suspension of about 10 g of soil in 20 ml of a dilute spiking solution is stirred for 3 h before evaporation of the solvent. Stirring the slurry causes the compounds to spread all over the soil and to interact with the total surface of the soil particles. Further, the analytes can partition between the liquid phase and the whole wetted soil surface during a much longer period of time than in the spot method (about 20 versus 2 h). It can therefore be expected that the analytes will interact with essentially all active sites of the matrix. Active matrix sites for OPPs are localized in the organic matter (e.g., humic material) and clay fraction of the soil .35 Adsorption of solutes on soil involves interaction forces ranging from van der Waals-London interactions to chemical bonding .34 Analyte recoveries after extraction with pure C 0 2 from spot-spiked soil are depicted in Fig.2. The extraction yields were as good as those obtained with the sand sample, except for dimethoate. Its recovery decreased from 94% (sand) to 71% (soil). Apparently no significant interaction of the soil matrix occurs with any of the other OPPs. However, when a slurry-spiked soil was extracted under the same conditions, the extraction was distinctly less efficient. As can be seen from Fig. 2, the extraction yields of disulfoton and azinphos methyl were about 20% lower than for the spot-spiked soil and only 41% of dimethoate was extracted.The recoveries of the remaining OPPs decreased by &13%. Analyte losses due to the evaporation of the larger spiking solvent volume were estimated by conducting a 'blank' slurry-spiking experiment, i . e . , without soil. The observed evaporation losses of 2-8%, which represent an extreme situation, because no particulate matter was present during the solvent elimination step, cannot explain the substantial I C - / A o B 0 1 2 3 4 5 6 CO$m I Fig. 3 OPP recovery (%) versus volume of liquid C02 used in SFE. SFE conditions as in Fig. 2. The slurry-spiked soil was extracted 5 d after spiking. Dimethoate; A, spot-spiked and B, slurry-spiked; azinphos methyl: C, spot-spiked and D, slurry-spiked soil14 ANALYST, JANUARY 1993, VOL.118 decrease in the recoveries of azinphos methyl, disulfoton and dime thoate. An attempt t o improve the OPP recoveries by carrying out SFE at higher pressure (300 bar) was not successful. As can be seen from the extraction profiles of dimethoate and azinphos methyl in Fig. 3, a plateau was rapidly reached (after about 3 ml of C02) when a spot- and a slurry-spiked soil were extracted with pure C02. As expected, increasing the volume of extractant to 9 ml of C 0 2 did not cause a noticeable increase in the recoveries of the OPPs, and neither did exhaustive steady-state extraction with pure C02. The decreased recovery of the OPPs and especially of dimethoate from a slurry-spiked soil implies that part of the analytes is strongly bound to the soil.Despite the good analyte solubility in pure C02, the apolar supercritical C 0 2 cannot compete efficiently with the active matrix sites to displace the sorbed analytes. The plateau of the extraction profile in Fig. 3 indicates the area in which the extraction becomes sorption and/or diffusion limited. The explanation is further supported by the fact that dimethoate, which obviously is the analyte most sensitive to sorption interactions, showed the largest decrease in recovery when testing the slurry-spiking proce- dure, with which interaction with active sites will be most prominent. It was also observed that the C02-extractable percentage of the OPPs decreased with the storage time of the spiked soil. For instance, only 66% of azinphos methyl was recovered after 5 d (cf., Fig.3), compared with 75% after 24 h (cf., Fig. 2). As another example, compare the 40 ppm spiked data in Fig. 2 with those in Table 1. Such an increase of the strongly bound portion of an analyte with time is well known from the literature dealing with sorption processes in soil .34,35 The low extraction yields obtained with the slurry-spiked soil suggest that poor results may also be expected when the extraction of OPPs from field samples is attempted with pure CO?. SFE With Modified C02 The negative influence of solute-matrix interactions on analyte recovery from real samples has been reported in several studies. Incomplete extraction with pure C 0 2 , despite good analyte solubility, has been observed by Mulcahey et uf.1 0 for PCBs, by Onuska and Terry14 for tetrachlorodibenzo- p-dioxins and by Hawthorne et uf.3' for PAHs from sediments and by Engelhardt et al.Z6 for explosives from soil. The extraction efficiency can be increased by using a supercritical fluid having a dipole moment ( N 2 0 , CHCIFS) or by adding an organic solvent as modifier to C 0 2 . Because of practical aspects ( N 2 0 is explosive, CHCIF3 very expensive and NH3 Table 1 Cumulative OPP recoveries (%) as a function of organic modifier used Modifier added to the samplc? Prc- extraction Ethyl Analyte with C03* acetate Acetone Methanol Diazinon Disulfoton Dimethoate Malathion Parathion Carbofenthion Azinphos methyl Coumaphos 72 62 37 72 71 75 67 73 82 71 55 81 80 84 85 80 82 70 61 82 81 80 88 86 82 70 83 84 82 81 99 90 * Slurry-spiked soil (40 ppm) dynamically extracted with 5.5 ml of pure CO? at 50°C and 250 bar.11 d after spiking. i- Subsequent addition of 35 pl of modifier to the C02-pre-extracted soil in the cartridge; 2 min static extraction, followed by 1.0 ml ofCOZ. The procedure was repeated once. toxic), modifiers were used to enhance the extraction effi- ciency. Acetone, ethyl acetate and methanol were selected as modifiers, because they are used in conventional techniques for the extraction of OPPs from soil and sediment.35--37 Table 1 lists the cumulative OPP recovery for a slurry-spiked soil, using these modifiers. Before adding modifier, the soil was exhaustively pre-extracted with pure C02. That is, the increased recoveries observed on subsequent SFE in the presence of modifier reflect the successful competition for the analytes bound to the active matrix sites.As can be seen from Table 1, the recoveries improved with all threc modifiers. Significant mutual differences were observed for the pesti- cides for which the recovery was most seriously affected by the matrix effect, viz., azinphos methyl and dimethoate. With both analytes methanol was the most effective displacer. The effect of the addition of methanol to a C02-pre-extracted soil sample is clearly demonstrated in Fig. 4. Fig. 4(a) shows the gas chromatogram with nitrogen-phosphorus specific detec- tion (NPD) of an SFE extract with pure COT. In Fig. 4(b), the chromatogram obtained after a subsequent extraction of the same sample in the presence of 70 PI of methanol is shown. Fresh trapping solvent was used for the methanol-modified extraction.The distinctly increased recoveries for dimethoate (63%) and azinphos methyl (26%) are clearly observed. In order to study the extraction yield, the analyte recovery was monitored as a function of the number of methanol- modified extraction cycles. Some typical results are presented in Fig. 5 . A plateau is reached for diazinon after a single extraction cycle. The same behaviour was shown by disulfo- ton, malathion, parathion and carbofenthion. Two cycles were required to obtain an extraction efficiency of at least 80% for azinphos methyl, dimethoate and coumaphos (data not shown). With this group, a further increase of 5 4 % could t - m c 0, m .- IS 0 5 10 15 20 Retention time/min Fig.4 GC-NPD traces of SFE extracts from a slurryspiked soil. ( u ) Spiked soil (4 ppm) extracted with 5 ml of pure C 0 2 at 50 "C and 250 bar (25 MPa); ( h ) same soil subsequently extracted with methanol- modified C 0 2 (addition of 70 pl to the soil). For compounds, see Fig. 2. GC conditions: 1 pl injected on-column into a 2 m X 0.32 mm i.d. retention gap, coupled to a 25 m x 0.32 mm i.d. DB-5 column. Temperature programme: 80 "C for 1.5 min. then increase to 280 "C at 10°C min-1. Carrier gas: He at 80 kPa head pressureANALYST, JANUARY 1993, VOL. 118 15 100 s - 80 L- > 60 I 0 1 2 3 4 5 Number of cycles Fig. 5 Plot of analyte recovery versus number of SFE cycles for three OPPs using methanol-modified C02. A 35 yl volume of methanol was added to the soil bcforc steady-state extraction cycle.One cycle corresponds to approximately 1.4 ml of C02. Extraction conditions: 2 rnin static; dynamic purging with 1.0 ml of C 0 2 , 50°C; 250 bar (25 MPa). decompression of the extraction cell for 3 rnin before addition of the next portion of modifier. The spiked soil (40 ppm) was extracted 7 d after slurry-spiking Table 2 OPP recovery using methanol-modified steady-state SFE of soil. Slurry-spiked soil extracted 24 h after spiking. Same conditions as in Fig. 5 cxcept that four cycles were performed ( n = 3). Analysis by GC-FID (40 ppm) and by GC-NPD (4 ppm) Spiking level 4 PPm 40ppm Analyte Diazinon Disulfoton Dimethoate Malathion Parat hion Carbofenthion Azinphos methyl Coumaphos Recovery 95 70 100 96 97 95 103 99 ("/.1 RSD 2 2 3 1 2 2 4 3 ( Y o ) Recovery 89 75 94 92 90 89 94 90 (%) RSD 6 6 5 4 8 6 9 6 (Yo ) be achieved if five instead of two cycles were performed. The slight improvement certainly does not justify the increase in the extraction time from 22 to 55 min. The extraction yields during one cycle could not be further improved by extending the dynamic purging, i.e., by using more than 1 ml of CO-,. Probably the organic modifier is rapidly swept out of the extraction vessel. The static period was varied from 1 to 4 min; 2 min were sufficient for equilibration, which agrees with the results of Onuska and Terry.9 All of the OPPs were extracted efficiently from a slurry- spiked soil at 40 ppm by performing two methanol-modified steady-state extractions.With the exception of disulfoton (see below), the recoveries of all analytes were at least 80%. The adsorption of the OPPs on the soil probably is due to hydrogen bonding or ligand exchange.35 This would explain why the most polar modifier tested, methanol, is the most effective in overcoming the matrix effects. Different extraction yields were obtained using methanol-modified C02 when soil was extracted 1 d after spiking (see Table 2) or 11 d later (see Table 1). The slight decrease in analyte recovery (about 5%) with storage time may be due either to stronger bonding of OPPs to matrix sites or to degradation. In order to evaluate whether the extraction procedure can also be used at lower concentration levels, a soil was slurry Table 3 OPP recovery (% t SD) from spiked soil using SFE or solvent extraction.Slurry-spiked soil (4 ppm) extracted 7-9 d after spiking ( n = 3) Methanol- modified Pre-mixed steady-state modified C02 Solvent Anal yte SFE" dynamic SFEt extraction$ Diazinon Disulfoton Dimethoate Malathion Parathion Carbofenthion Azinphos methyl Coumaphos 92 k 3 63 f 3 111 + 4 101 * 4 95 -r- 3 96 k 5 103 f 5 96 f 2 91 rt 2 66 f 5 104 f 4 97 f 4 98 f 2 92 + 4 102 f 4 93 ?E 5 90 -e 4 65 2 2 75 + 6 90+ 4 96 f 8 93 f 8 95 f 8 91 * 5 * Conditions as in Fig. 5 ; four cycles. cylinder (2% m/m) at 50°C and 250 bar. $ See Experimental for conditions. Dynamic extraction with 3-5 ml of methanol-pre-mixed C 0 2 spiked at the 4 ppm level. The typically 90-95% analyte recoveries obtained at both spiking levels (Table 2) suggest that the present procedure will be applicable over a wider range of concentrations. The relatively low RSD values obtained at the lower spiking level may well reflect the inherently greater selectivity of NPD as compared with FTD. At both spiking levels it was necessary to add at least 30-40 p1 of methanol per cycle to obtain the recoveries quoted above.The addition of only 5 or 15 p1 of modifier yielded lower recoveries, especially with dimethoate and azinphos methyl. On the other hand, the use of 70 instead of 35 pl did not improve results at all. The low recovery of disulfoton is probably caused by degradation. We observed that solutions of disulfoton were not stable when they were exposed to daylight and ambient temperature for longer periods of time. Disulfoton contains a thioether group.Degradation studies on phosphorothiolo- thionates in soil indicate a rapid oxidation to their sulfone and sulfoxide form .35 These degradation products are more polar than the parent compound and, hence, are less extractable. On the other hand, it cannot be excluded that disulfoton interacts strongly with the soil matrix and that none of the tested modifiers could displace the bound pesticide. Performing methanol-modified static extractions takes approximately 11 rnin per cycle (addition of solvent, installa- tion of the cartridge, static period, dynamic purging and decompression) or 20-25 rnin for the two cycles required. On the other hand, the exhaustive dynamic extraction of the 40 ppm spiked soil with pure C02 was complete after 12 min. Consequently, the use of a pre-mixed methanol-modified CO-, cylinder was also studied.Table 3 lists the OPP recoveries obtained with a methanol- modified C02 cylinder at 2% m/m. Dynamic extraction with the pre-mixed phase is seen to be as efficient as methanol- modified steady-state extraction. The high results obtained for dimethoate (104-111%) were due to peak tailing and integra- tion, problems occurring with this analyte during this part of our research. The extraction of all solutes was completed when using 3 ml of premixed C 0 2 . The total time of extraction, including installation o f the cartridge, pre-heating and static equilibra- tion, was only 12 min. Hencc t h c use of pre-mixed modified phases is attractive for routine analysis, because sample handling is minimized.Still, the addition of a modifier to the soil is more appropriate for SFE method development, because different modifiers can easily be tested. Finally, SFE was compared with conventional solvent extraction of a slurry-spiked sample. As can be seen from Table 3, the several sets of data show good mutual agreement, with the precision of the SFE methods being better than that16 ANALYST, JANUARY 1993, VOL. 118 of solvent extraction. For the rest, SFE was more efficient in removing the polar dimethoate from the soil sample. It should be noted, however, that the solvent extraction was not optimized for dimethoate. A major advantage of SFE over solvent extraction, often cited in the literature,*Jg is its speed. This is also true for the extraction of OPPs from soil.Depending on the mode of modifier addition, SFE took 12 or 25 min compared with 3-4 h for solvent extraction (even apart from the overnight equilibration). Conclusions Under optimized conditions, SFE permits the quantitative recovery (290%) of most OPPs from soil in less than 15 min. These high recoveries are obtained by adding methanol to the soil or by using premixed methanol-modified C 0 2 . All OPPs except dimethoate can be quantitatively recovered from inert matrices, such as glass-wool and sand, and from spot-spiked soil samples using pure C02 at 50 "C and 250 bar. However, the extraction yields with pure C02 decrease significantly when the analytes are added to soil by means of the slurry-spiking method. Solute-matrix interac- tions apparently prevent the desorption of part of the spiked analytes, as was especially manifest for dimethoate and azinphos methyl.This confirms that good analyte solubility in the supercritical phase (in our case C02) does not guarantee an efficient extraction from environmental samples. Matrix effects have to be considered as the major factor to be understood, for a successful application of SFE in the field of environmental analysis. The use of a polar modifier, methanol being more suitable than either ethyl acetate o r acetone, completely eliminates these problems. The optimized SFE method is slightly more precise than, and as efficient as, the conventional solvent extraction. Its main advantage however, is the dramatic gain in the time required for sample handling.Future work will focus on the influence of different soil characteristics and of water on solute-matrix interactions and on SFE efficiency. The authors thank Chrompack (Middelburg, The Nether- lands) for the loan of the syringe pump and RIVM (Bilthoven, The Netherlands) for the gift of the soil sample. References Hawthorne, S. B., Anal. Chem., 1990, 62, 633A. Vannoort. R. W., Chervct, J.-P., Lingeman, H . , De Jong, G. J., and Brinkman, U. A. Th., J . Chromatogr., 1990, 505, 45. Analyticul Supercritical Fluid Chromatography and Extraction, eds. Lee, M. L., and Markides, K. E . , Chromatography Conferences, Provo, UT. 1990. Abstructs of the International Symposium on Supercritical Fluid Chromatography and Extraction, Park City, Utah, USA, 1991.Proceedings ,for the European Symposium on Supercritical Fluid Chromatography and Extruction, Wiesbaden, Germany, 1991. Hawthorne, S . B., and Miller, D. J., J. Chromatogr., 1987,403, 63. Schantz, M. M., and Chesler. S . N., J. Chromutogr., 1986,363, 397. 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 Lohlcit, M., Hillmann, R., and Bachmann, K . , Fresenius' J. Anal. Chem., 1991. 339, 470. Onuska, F. I., and Terry, K. A., J. High Resolut. Chromatogr., 1989, 12, 527. Mulcahey, L. J., Hedrick, J. L., and Taylor, L. T., Anal. Chem., 1991, 63, 2225. Hawthorne, S. B . , Miller, D. J . , and Langenfeld, J. J . , J. Chromatogr. Sci., 1990, 28, 2. Hawthorne, S . B., and Miller, D. J . , Anal. Chem., 1987, 59, 1705. McNair, H. M., and Frazicr, J .O., fnt. Lub., 1991, 21, 33. Onsuka, F. I . , and Terry, K. A., J . High Resolut. Chromutogr., 1989, 12, 357. Levy, J . M., and Rosselli, A. C., Chromatographia, 1989, 28, 613. Richards, M., and Campbell, R. M., LCGC Int., 1991, 4, 33. Hawthorne, S. B., Miller, D J.. Walker, D . D., Whittington, D. E., and Moore, B. L., J. Chromatogr.. 1991, 541, 185. Spall, W. D., Martinez, A. A.. and Smith, B. F., in Abstracts of the International Symposium on Supercritical Fluid Chromato- graphy and Extraction, Park City, Utah, USA, 1991, p. 11 I. Pesticide Analysis, ed. Gas, K. D., Marcel Dekkcr, New York. 1981. McNally, M. E. P., and Wheeler, J. R., J. Chromatogr.. 1988, 435, 63. McNally, M. E. P., and Wheeler, J. R., J. Chromutogr., 1988, 447, 53. McNally, M. E. P., and Wheeler, J . R., J. Chromatogr. Sci., 1989. 27, 534. Lopez-Avila, V., Dodhiwala, N. S . , and Beckert, W. F., J. Chromatogr. Sci., 1990, 28, 468. Engelhardt, H., and Gross. A., J. High Resolut. Chromatogr. Chromatogr. Commun., 1988, 11, 726. Janda, V., Steenbekc, G . , and Sandra, P., J. Chromatogr., 1989,479, 200. Engelhardt, H., Zapp, J. and Kolla, P., Chromutographia, 1991, 32. 527. Mapelli, G., Pigozzo. F., Raynor, M. W., and Trestianu. S . , in Proceedings of the 13th Internutionul Symposium on Capillary Chromatogr.. ed. Sandra, P., Huthig. Heidelbcrg, 1991, p. 489. Cortes, H. J. Green. L. S . , and Campbell, R. M., Anal. Chem., 1991, 63, 2719. Bergna, M., Banfi, S . , and Cobelli, L., in Proceedings o f t h e 13th International Symposium on Capillary Chromatography, ed. Sandra, P., Huthig, Heidelberg, 1991, p. 300. Pipkin, W., LCGC Int., 1992, 5, 8. Hawthorne, S. B., Miller, D. J . , and Langenfeld, J . J., in Abstracts of the In ternationul Symposium on Supercritical Fluid Chromatography and Extraction, Park City, Utah, USA, 1991, Borburgh, H. J.,Applicution Note, J . T. Baker, Deventer, 1990. M. deWilde, CIVO-TNO, Scist, The Netherlands, personal communication. Organic Chemicals in the Soil Environment, eds. Goring, C. A. 1.. and Hamaker, J . W., Marcel Dekker, New York, 1972. Khan, S . U.. Pesticides in the Soil Environment, Elscvicr, Amsterdam, 1980. Durand. G., and Barcelo, D.. Anal. Chim. Acta, 1991,243,259. Kjslholt. J . , J. Chromutogr., 1985, 325, 231. Bartle. K . D.. and Clifford, A. A., LCGC Int., 1991, 4, 10. p. 91. Paper 2/01 993 K Received April 16, 1992 Accepted September 30, 1992