ANALYST, JULY 1993, VOL. 118 Supercritical Fluid Chromatography With Electron-capture for the Determination of Agrochemicals Robert Moulder,* Keith D. Bartlet and Anthony A. Clifford School of Chemistry, University of Leeds, Leeds, UK LS2 9JT 737 Detection The scope for supercritical fluid chromatography with electron-capture detection of several test agrochemicals was investigated using a carbon dioxide mobile phase. Operation with packed capillaries was found to be preferable to conventional (large inside diameter) packed columns. The detector employed was found to be suitable for the detection of the high-melting point fungicides captan and captafol. Keywords: Supercritical fluid chromatography; electron-capture detection; agrochemicals; packed capillary column; high-melting point analyte One of the strengths of supercritical fluid chromatography (SFC) is its compatibility with a wide range of detectors in use with both liquid chromatography (LC) and gas chromato- graphy (GC).1-2 Amongst these is the electron-capture detec- tor (ECD); this has been long recognized in GC as an extremely sensitive means of identification of organic com- pounds con- taining heteroatoms,3~4 especially halogens. Previous studies have demonstrated a promising potential for the ECD as a post-SFC detector.'*2,s--7 Linear dynamic ranges up to four orders of magnitude with low- and sub-picogram detection limits have been reported for halogenated com- pounds.6 Failure to detect the high-melting point fungicides captan and captafol has been attributed to their insufficient volatility in the detector cavity.6 In this study, an ECD was used as an SFC detector for the determination of a variety of agrochemicals, including captan and captafol, using conventional packed columns, packed capillary columns and open-tubular columns.Experimental In all instances SFC was performed with the column in a Varian 3700 gas chromatograph and connected to a Varian h3Ni constant-current ECD8 operated at 300 "C. Nitrogen was passed through an oxygen trap and used as the detector make-up gas. Packed Column SFC Initial work using a 100 x 4.6 mm i.d. packed column ( 5 pm silica, LiChrosorb Si-60) was performed using a Gilson Model 303 reciprocating pump to deliver SFC-grade liquid CO2 (Air Products, Crewe, Cheshire, UK) at flow rates up to 2 cm3 min- I .The mobile phase was cooled by a refrigeration unit (Techne, Princeton, NJ, USA) to -8°C prior to pumping. A Rheodyne Model 7125 valve fitted with a 5 mm3 sample loop and maintained at 40°C was used for injection. The column eluate was split bctween an ultraviolet (UV) detector (Spectroflow 773) and the ECD with a splitting ratio of 200: 1. The flow of C 0 2 was directed through a pulse damper (Gilson Model 802C) prior to the injector. Lengths of 1/16 in stainless-steel tubing, 2.75 m X 0.762 mm i.d. and 20 m x 0.152 mm i.d., were also incorporated prior to the injector and post-UV detector, respectively, in order to effect further pulse damping. * Present address: Department of Analytical Chemistry, University of Uppsala, Sweden. To whom correspondence should be addressed.Packed Capillary and Open-tubular Column SFC Packed capillary and open-tubular column SFC were per- formed using a Carlo Erba 300 Series SFC pump (Fisons, Crawley, Sussex, UK). This was cooled by a Haake D8 refrigeration unit (Fisons) and controlled via an Innotech Systems computer (ITC-286). Injections were- initially per- formed with the Rheodyne valve (as above) fitted with a 0.17 mm3 loop of 50 pm i.d. fused-silica tubing. This was later changed in favour of the Valco CI4W injection valve (VICI, Houston, TX, USA) fitted with a 0.2 mm3 sample loop and pneumatics of a Carlo Erba 3000 Series SFC instrument. The injector was air driven and the solenoids operated from a 24 V power supply (Farnell, Wetherby, Yorkshire, UK). Lengths of 200 pm i.d.fused-silica tubing were slurry packed with 5 pm cyanopropyl-bonded silica (Spherisorb, Phase Separations, Clwyd, UK) from a 100 x 4.6mmi.d. stainless-steel reservoir. The ends of the columns were fitted with graphitized Vespel ferrules (SGE, Milton Keynes, Buckinghamshire, UK) into 1/16 in Valco low dead volume stainless-steel unions with the dead volume made up with poly(tetrafluoroethy1ene) (PTFE) spacers, and the packing was retained by 2 pm stainless-steel micro-screens (Phase Separations). The columns were connected to the injector by a 30 cm length of 75 pm i.d. deactivated fused-silica tubing. A 50 pm i.d. frit restrictor (Dionex, Camberley, Surrey, UK) was used to connect the column outlet to the ECD and trimmed to give the desired flow rate.A 10 m X 50 pm i.d. 30% biphenyl-substituted methylpoly- siloxane (0.25 ym film) open-tubular column (Dionex) was also used in this study. To decrease recorder noise, a 4700 yF capacitor was connected across the input of the chart recorder, thus increasing the time constant. Samples Solutions of heteroatom-containing agrochemicals were pre- pared in GC-grade hexane at concentrations ranging from 10 ng cm-3 to 50 yg cm-3. The following agrochemicals were studied: cypermethrin, tetrachlorvinphos, chlorthiamid, di- chlobenil, captan, captafol, linuron, diuron, p , ~ ' - D D T , -DDE and -DDD, o,p'-DDT, -DDE and -DDD, aldrin and lindane. Results Conventional Packed Column SFC-ECD Whilst operating both the packed and capillary systems, the use of a frit as opposed to a linear restrictor for the eluent split to the ECD was found to be essential.It was assumed that738 ANALYST, JULY 1993, VOL. 118 the rapid depressurization from the restrictor outlet achieved by the former more effectively 'sprayed' the solute into the detector. Optimum make-up gas flow rates and restrictor positioning were investigated. The conventional packed column system suffered greatly from baseline instability and poor day-to-day reproducibility. Sensitivities of at least a factor of 30 less than expected with conventional GC-ECD were observed when studying a series of test compounds, namely aldrin, DDT, DDE, DDD and a mixture of four geometric isomers of the pyrethroid cypermethrin. Fig. 1 is representative of the better results obtained, showing the separation of four geometric isomers of cypermethrin; detec- tion of 250 pg at a signal-to-noise ratio (S/N) marginally greater than 3.Examination of the baseline of this chromato- gram reveals a sawtooth-like effect, which is probably an effect of the pump pulses. Detection limits (DLs) of 20-40 pg (S/N = 3) were indicated for DDT, DDD and DDE isomers. Similarly, a DL near 10 pg was displayed for aldrin. These compounds are amenable to GC analysis but were chosen as simple test compounds for the compatibility of the detector with SFC. In conclusion, the main limitation of the conven- tional packed column system was a general baseline instabil- ity. With this borne in mind, it was hoped that the implemen- tation of a pulseless syringe pump would give better stability.Furthermore, it was expected that the sharper peaks asso- ciated with capillary SFC would further improve the detection capabilities. Packed Capillary and Open-tubular Column SFC-ECD The packed capillary SFC system did indeed show a greater improvement in baseline stability. In fact, for S/N = 3, DLs for DDT, DDE and DDE isomers were observed to be in the range 3-4 pg (Fig. 2). The detection system was, however, subject to long-term drift and although the detector body was well lagged with glass-wool, changes in the baseline were observed in the course of a day. Among the compounds selected for this study were captafol and captan (Fig. 3). Successful detection of levels down to 100 and 500 pg (S/N = 8), respectively, was observed. Chang and Taylor6 have previously reported a failure to detect captan and captafol using SFC-ECD.The Varian 63Ni ECD used in this study features a cell volume of 0.3 cm3,8 which compares with an effective volume of 1.1 cm3 (actual volume 2.1 cm3) of the Hewlett-Packard (HP) detector as used by Chang and Taylor.6 Adopting their argument of insufficient solute volatilization for these compounds in the HP detector, the 0 4 8 12 16 u 0 4 8 1 2 Time/m in Fig. 1 Packed column SFC-ECD separation of four geometric isomers of cypcrmethrin. Column: 100 x 4.6 mm i.d. LiChrosorb Si-60,s pm silica. Mobile phase: C 0 2 at 37.0 MPa (2 cm3 min-1, split at 200 : 1 to detector), 100 "C. Make-up gas: N2, 16 cm3 min-1. (a) and ( h ) 1.25 ng and 250 pg per component, respectively detection of these compounds may be attributed to superior volatilization in the smaller detector cavity; in our work the detector temperature was lower than that used in ref.6. Other agrochemicals included in this study were tetrachlor- vinphos, diuron, linuron, chlorthiamid and dichlobenil. Their elution and detectability were also checked using open- tubular columns (Fig. 4), but without optimization of separa- tion. The order of responses was as expected from the heteroatom content.4 These ranged from DLs of around 50 pg for tetrachlorvinphos and dichlobenil up to about 500 pg for diuron. In previous studies,lJj baseline rise during pressure pro- gramming was overcome by using higher make-up gas flow rates than those providing optimum sensitivity. When using the Varian ECD with nitrogen make-up gas it was found to be very difficult to maintain baseline stability during program- ming.This may also be attributed to the smaller volume of the detector cell. The use of a 1 : 1 split of packed capillary column effluent was seen to lead to increased baseline stability to pressure change. Optimization of the splitting ratio is expec- ted to show a potential improvement in this detection system. In the present study maximum detector response occurred with a make-up gas flow rate of about 15 cm3 min-1. The manufacturer's specified DL for lindane with this design of the Varian ECD in the GC mode is CO.1 pg (S/N = 2) with a linear range of >lo4 and a dynamic range of >105 with nitrogen make-up gas.* For SFC, DLs only down to 2.5 pg of lindane and reduced linear range and dynamic ranges of 103 and 104, respectively, were observed (Fig.5 ) . Standard deviations of approximately 8% peak height were observed for repeated injections. Although less sensitive than nitrogen, argon moderated with 5-10% of methane has the optimum carrier gas properties for ECDs for GC4 and, because of the noise level associated with the former, offers similar DLs. The I 2 3 0 2 4 6 8 1 0 1 2 1 4 Ti me/mi n Fig. 2 Packed capillary SFC-ECD o f y , p ' isomers of DDT (l), DDE (2) and DDD (3); 80,80 and 160 pgon-column, respectively. Column: 420 x 0.2 mm i.d., 5 pm cyanopropyl-bonded silica. Mobile phase: C 0 2 at 27.0 MPA and 100 "C '0 2 0 5 10 15 Time/mi n Fig. 3 Packed capillary SFC-ECD of captan (1) and captafol(2); 1.0 and 0.8 ng on-column, respectively.Column: 420 x 0.2 mm i.d.. S pm cyanopropyl-bonded silica. Mobile phase: C 0 2 at 100 "C and 37.0 MPaANALYST, JULY 1993, VOL. 118 739 Linuron CI Diuron Cl Chlorthiamid qSNH2 “‘0‘‘ Fig. 4 Open-tubular chlorthiamid ( 3 ) ; 1.6, mated). Column: 10 methylpolysiloxane (0. 80 “C 1 2 0 2 4 6 8 1 0 1 2 1 4 Ti me/m in SFC-ECD of linuron ( l ) , diuron (2) and 3.2 and 0.3 ng on-column, respectively (esti- m X 50 pm i.d. 30% biphenyl-substituted 25 pm). Mobilc phase: COz at 20.0 MPa and 4.0 r-- 1 Y 2.2 1 -I I I I 9‘ I I I I , 1.0 - 0 1.0 2.0 3.0 4.0 5.0 Log(amount on column/pg) Fig. 5 lindane. Column and conditions as in Fig. 3 Plot of log(peak height) versus log(amount on-column) for use of argon-methane (as used by Chang and Taylor,b as above) with the present system could conceivably offer further advantages over the nitrogen used here, including baseline stability whilst programming.Conclusions The use of ECD for trace analysis by SFC shows great potential. However, it would appear that a detector with optimized cell volume and geometry is required for sensitive and predictable performance. The lower volumetric flow rates associated with packed capillary and open-tubular capillary columns permit the use of pulseless syringe pumps and reduce the extent of ECD interference when compared with conven- tional-bore packed columns. We thank the Science and Engineering Research Council for financial support through a grant to R. M. References Richter, B . E., Bornhop, D. J., Swanson, J. T., Wangsgaard. J . G., and Andersen, M. R.. J . Chromatogr. Sci., 1989,27,303. Analytical Supercritical Fluid Chromatography and Extraction, eds. Lee, M. L., and Markides, K . E., Chromatography Conferences, Provo, UT, 1990. Detectors for Capillary chromatography, eds. Hill, H. H . , and McMinn, D. G., Wilcy, New York, 1902. Analysis of Pesticide Residues, ed. Moyc, H. A., Wiley-Inter- science, New York, 1981. Kennedy, S . , and Wall, R. J . , LC.GC, 1988, 6, 930. Chang, H.-C. K., and Taylor, L. T., J. Chromatogr. Sci., 1990, 28, 29. Munder, A., Christensen, R. G . , and Wise, S. A., J. Micro- column Sep., 1991, 3, 127. Varian Series 3700 Gas Chromatograph Operation and Main- tenance Manual, Publication No. 85-001139-00, Varian, Palo Alto. CA, 1977. Paper 310021 1 J Received January 13, 1993 Accepted March 19, 1993