Variable-angle scanning fluorescence spectrometry for the determination of closely overlapped pesticide mixtures

Variable-angle scanning fluorescence spectrometry for the determination of closely overlapped pesticide mixtures

Analytica Chimica Acta, 228 (1990) 293-299 Elsevier Science Publishers B.V.. Amsterdam 293 - Printed in The Netherlands Variable-angle scanning flu...

546KB Sizes 6 Downloads 24 Views

Analytica Chimica Acta, 228 (1990) 293-299 Elsevier Science Publishers B.V.. Amsterdam

293 - Printed

in The Netherlands

Variable-angle scanning fluorescence spectrometry for the determination of closely overlapped pesticide mixtures F. GARCIA

SANCHEZ

* and A.L. RAMOS

RUB10

Department of Analytical Chemistry, Faculty of Sciences, University of Mcilaga, Mblaga (Spain) V. CERDA

and M.T. OMS

Department of Chemistry, Faculty of Sciences, University of Palma de Mallorca, Palma de Mallorca (Spain) (Received

20th December

1988)

SUMMARY

The application of a program written in BASIC for an IBM/PC interfaced with a commercial spectrofluorimeter enables variable-angle scanning fluorescence spectra to be obtained, without modification of the instrument. The technique is effective for determinations of the components of mixtures of carbaryl, fuberidazol and warfarin, all of which are pesticides with intrinsic fluorescence and closely overlapping profiles. The proposed method permitted simultaneous determinations with RSD of less than 2.5% and recoveries of 99-1108.

Spectrofluorimetry has great sensitivity and only moderate selectivity. However, its performance in multi-component analysis falls considerably because overlapping spectra give rise to energy-transfer processes and have an inner filter effect that produces spurious analyte signals. Thus conventional fluorimetry needs previous time-consuming separations, because the analyses of multi-component mixtures of fluorescent compounds or of compounds in a fluorescent matrix often have large intrinsic errors. At present there are two fundamental ways to avoid this problem: analytical data sampling and analytical signal processing. The first processes the digital data signal and the other modifies the method by which the analytical signal from the sample is produced to eliminate or minimize the contributions of the non-analyte signals. Synchronous wavelength-or energy-scanning fluorimetry is a development of the latter and gives good results [ 1- 31. However, because synchronous wavelength scans only provide 45” sections, their application 0003-2670/90/$03.50

0 1990 Elsevier Science

Publishers

B.V.

is limited. Recently, variable-angle scanning fluorescence spectroscopy was tested by varying the wavelength intervals between the monochromators (i.e., using different scan speeds for the two monochromators) [4]. Continuous wavelength interval variation is produced either mechanically or digitally, by processing the stored data. Mechanical variation involves modifying the instrument so that the monochromators have different scan speeds. On the other hand, digital processing generates a theoretical pathway from data stored as a “contour map”. Unfortunately, because several contour maps are required to construct a calibration graph and the method is time consuming, the potential applications are severely limited. The other approach combines signal processing, data reduction [5] and deconvolution of the signal [6]. This involves the rapid acquisition of spectral data by multi-channel optical detectors; arrays of photodiodes 171, the silicon-intensified target vidicon [8] and charge-coupled devices [9] have

F.GARCiA SANCHEZ ET AL.

294

completely transformed the technology of spectroscopic detection systems. Easy storage and manipulation of the digitalized data permit new strategies for spectral deconvolution to be developed. Further, the use of computerized data analysis has improved the systematic generation of information. This paper reports the application of a software package, written in BASIC for the IBM-PC, that permits the instrument (Perkin-Elmer LS-5 spectrofluorimeter, or any other spectrofluorimeter interfaced with an IBM-PC) to produce variable-angle scanning spectra (v.a.s.s.). Only a few minutes suffice to obtain the spectra from the samples by following the route previously selected by inspecting the “contour map”. It should be emphasized that the v.a.s.s. are generated directly from the spectrofluorimeter input, via the microcomputercontrolled monochromator devices. Information is also given about an algorithm, included in the software package [lo], which permits the theoretical variable angle scan to be obtained from data stored in the “contour maps”. This step helps to optimize the route that will produce the best variable-angle scanning spectra (highest signal value, smallest band width at half-maximum intensity and interference-free bands). Carbaryl (1-naphthyl-N-methylcarbamate) (C) (trade-name Sevin) is widely used as an agricultural [ll] and forest [12] spray. There are reports [13] that at alkaline pH it hydrolyses to l-naphthol and N-methylcarbamic acid. Warfarin [4-hydroxy-3-(3-oxo-1-phenylbutyl)coumarin] (W) is an anticoagulant rodenticide that does not produce “bait-shyness” in rats [14]. It is used in medicine as an anti-coagulant to reduce the risk of thrombosis. Fuberidazol [2-(2’-furyl)-benzimidazole] (FBZ) is a fungicide used to protect rye and pea seeds against different fungal diseases [14]. The proposed technique is used to determine these three compounds in mixtures.

EXPERIMENTAL Apparatus

and software

Fluorescence spectra were obtained with a Perkin-Elmer LS-5 luminescence spectrophotometer

equipped with a xenon lamp (9.9 W) pulsed at line frequency. Slit widths were set at 2.5/2.5 nm. Emission spectra were not corrected for non-linear instrumental response. A Kyocera F-1000 laser printer was connected to the microcomputer via the parallel port to obtain graphical representations. The luminescence spectrophotometer was interfaced to an IBM-PC/XT microcomputer via the RS232C serial interface. The software package written in BASIC [lo] controlled the instrument, data collection and data processing. Reagents

Standard solutions of carbaryl and fuberidazol (Pestanal quality, > 99%; Riedel de Ha&n, Seelze, Hannover) and warfarin (98%; Aldrich) were dissolved in ethanol to give concentrations of 0.2 mg ml-‘. All solvents were of analytical-reagent grade (Merck). The water was distilled and dernineralized. Phosphate buffer solution (0.1 M) was prepared from potassium salts (Merck). Methodology

To obtain the analytical parameters of the method and to the determine the optimum v.a.s.s. route, a standard solution of the three pesticides in near isoemissive concentrations was prepared. Phosphate buffer (3 ml) was added and the volume was made up with deionized water to give an ethanol-to-water ratio of 1: 10 (v/v). Sequential scans of the emission spectra (50 scans) were carried out between 300 and 550 nm at different excitations ranging from 230 to 368 nm. The optimal v.a.s.s. route was determined by inspecting both the three-dimensional and the three two-dimensional contour plots. The selected routes were optimized by trial and error to give maximum resolution of the three analytes. The calibration graphs were obtained by recording the v.a.s.s. of a series of different mixtures of C, W and FBZ whose individual concentrations varied independently within three ranges of concentrations: lOO-1000,500-2100 and l-8 ng ml-‘, respectively. Fluorescence intensity was measured in the v.a.s.s. plots at 251-323, 296-407 and

VARIABLE-ANGLE

SCANNlNG

FLUORESCENCE

322-346 nm, respectively, and concentration of each pesticide.

RESULTS

AND

SPECTROMETRY

plotted

against

DISCUSSION

Carbaryl, FBZ and W are moderately soluble in water and ethanol. When excited at 280-310 nm in aqueous or ethanolic solutions, they fluoresce in the range 320-420 nm. Fluorescence emission is strongly dependent on solvent composition; W and C (strong emission at high ethanol concentrations) gave behaviour patterns that contrasted with that of FBZ (high emission at lower ethanol concentrations). Experiments with different proportions of water-ethanol mixtures led to a compromise solution in which each of the three compounds had a fluorescence emission adequate for the experiments. Ethanol-water (10 : 90, v/v) appeared most suitable for subsequent experiments.

295

The influence of pH on W and FBZ mixtures revealed that between pH 6 and 9 the fluorescence intensity remains constant and maximum. Carbaryl, however, hydrolyses at alkaline pH; below pH 8.0 the hydrolysis [15] is slow and the fluorescence emission is constant. Taking into account the behaviour of W, FBZ and C, a buffer solution of pH 6-8 was found to be satisfactory, and 3 ml of a solution of phosphate buffer in a final volume of 25 ml of ethanol-water (10: 90, v/v) give a “pH” of 7.39 + 0.04. These conditions give constant fluorescence readings. Figure 1 shows the excitation-emission plots for carbaryl (B), FBZ (D) and W (C), and a mixture of the three compounds (A). The closeness of the spectral profiles makes the determination of the individual compounds in the mixtures impossible. The spectra of W and FBZ (Fig. 1C and B, respectively) are bi-modal whereas carbaryl (Fig. 1B) displays only one peak. The broad bands

230

(nm)

30

(nm)

I

230

(nm)

Fig. 1. Three-dimensional image plot of (A) isoemissive ternary mixtures of carbaryl (500 ng ml-‘), warfarin (1000 ng ml-‘) and fuberidazol(5 ng ml-‘); (B) carbaryl(4000 ng ml-‘); (C) warfarin (4000 ng ml-‘); (D) fuberidazol (100 ng ml-‘). Response factor, 4.0; scan speed, 480 nm n-k-‘; slits, 2.5/2.5 nm.

F. GARCiA

296

550 525 500 475

325 “!!4S

273 298 323 348

A(em)hm)

h(em)hm)

A (em)hm)

A(em)

Fig. 2. Contour plots for warfarin, carbaryl, ternary mixture. Conditions as in Figure 1.

S,kNCHEZ

ET AL.

This type of mixture (Fig. 2) cannot be resolved either by conventional fluorimetry (there are no excitation-emission pairs for selective excitation) or by synchronous scanning fluorimetry because the almost complete overlap of the spectral shapes impedes the detection of interference-free signals. The proposed solution to this problem is illustrated in Fig. 3. The variable-angle scanning route was carefully determined by trial and error to traverse those parts of the three-dimensional spectral zones with the least overlap. In spite of the fact that losses in sensitivity occur because no maximum peaks are traversed, interference-free signals of the three components may be obtained from the chosen routes that scan the three-dimensional zones by skirting the slopes of the peaks and avoiding the areas of interference between the three compounds. Several three-dimensional variable angle spectra of a three-component mixture of the pesticides are shown in Fig. 4D to illustrate the potential utility of this technique for mixture analysis. The

(rim)

fuberidazol

and a

appearing at the bottom left and top right comers of Fig. 1ACD arise from scattered light. In Fig. 1A the peaks are spurious and are due to the similar luminescence behaviour of the pesticides. The contour plot in Fig. 2 was more useful for locating the position and distribution in the fluorescence map of the lumiphores C, W and FBZ. The individual contour plots showed that the main peaks were centred on their respective wavelength excitation-emission pairs 275-332, 308-390 and 306-345 nm, respectively.

6 248

298 (Ill-n)

348

Fig. 3. Contour plot of an isoemissive mixture of carbaryl, warfarin and fuberidazol, and the selected v.a.s.s. route.

VARIABLE-ANGLE

SCANNING

FLUORESCENCE

291

SPECTROMETRY

(A)

290

248

A(ex)hm)

359

2

390

300

34

CD)

(-)

Em.

(+)

I

(B)

A(em)(mln) ~300 X(exmax)=359”m hkmNmax)=420

nm

nm

s 00

325(420) Akm)

hm)

Fig. 4. V.a.s.s. of mixtures (D). Three.-dimensional spectra of an isoemissive mixture of carbaryl, warfarin and fuberidazol in concentrations as in Figs. 2 and 3. (A), (B) and (C) two-dimensional projection of several mixtures of carbatyl, warfarin and fuberidazol: (A) carbaryl(500 ng ml-‘), fuberidazol(lO30 ng ml-‘) and warfarin (500,1000,1500 and 2000 ng ml-‘); (B) carbaryl (100,250,500 and 1000 ng ml-‘), fuberidazol(lO30 ng ml-‘) and warfarin (1030 ng ml-‘); (C) carbaryl(500 ng ml-‘), fuberidazol (2, 4, 6 and 8 ng ml-‘) and warfarin (1030 ng ml-‘).

software program [lo] produces and displays the required data in 4-5 min, in comparison with periods of several hours required to obtain the three-dimensional data matrices. It may be easily adapted to work with any spectrofluorimeter with TABLE

two monochromators that can be interfaced with the RS 232C port of an IBM PC or other compatible computer. The fact that both monochromators of the spectrofluorimeter cannot be set to different scan

1

Analytical

characteristics

Compound

Calibration

graph a

z=A[X]+B

r

SA” (Pg ml-‘)

Z = 364O[X] + 1.56 Z = 3.91[X] + 2.86 Z = 11.56[X] + 2.06

1.000 0.998 0.996

0.011 0.078 d 0.028

Limit of detection (/Jg ml-‘)

Linear range

0.051 0.050 d 0.200

0.2-1.0 0.2-8.0 0.7-2.1

RSD (sg) c

(pg ml-‘) Carbaryl Fuberidazol Warfarin

d

a Z = intensity, [X] = concentration b Analytical sensitivity (= SJm). cn=5. d Concentration in ng ml-‘.

in gg ml-‘.

d

1.9 (0.5 fig ml-‘) 2.4 (5.0 ng ml-‘) 1.8(1.Opgtn-‘)

F. GARCiASANCHEZ ETAL

298 CROSS-INTERFERENCE 25. -

A Warfarin-Carbaryl A Worfarin-Fuberidazol

pg ml-’ pg ml-’

EFFECTS 0 Carbaryl-Warfarin I

0

Fuberidazol-Warfarin

(ng

ml-‘)

l

Fuberidozol-Carboryl

(ng

ml-‘)

pg

Carbaryl-Fuberidozol

ml-’ pg ml-’

-25 0

2.5 CONCENTRATION

Fig. 5. Cross-interference effects of carbaryl, warfarin and fuberidazol. Ordinate, relative error on the signal of the first component; abscissa, concentration in pg ml-’ (fuberidazol in ng ml-’ X 2). First component, analyte; second component, interferent. Concentrations of analytes: carbaryl, 0.5 pg ml-‘; warfarin, 1.0 pg ml-‘; and fuberidazol, 5 ng ml-‘.

does not matter because the program signals override and differentially vary the excitation and emission monochromator drive motors of the spectrofluorimeter by pulsing their power input. Fig. 4ABC is the three two-dimensional projections of the variable-angle scanned spectra, and Fig. 4D is the combined three-dimensional spectrum of the ternary mixtures with various concentrations of C, W and FBZ (loo-1000,500-2100 respectively). Figure 4D was and 1-8 ng ml-‘, generated and plotted by scanning the selected route showed in Fig. 3. Two-dimensional projection of several scans with various concentrations of C, FBZ and W are plotted in Fig. 4ABC. It can be seen that increasing the concentration of each compound in turn had little effect on the other signals. This effect is shown in more detail in Fig. 5, which shows the relative errors of the signals of each component produced by cross-interferences from the other compounds; these are plotted against increasing concentrations of each component of the mixture. The perturbations associated with interference effects fall within a 10% relative error for concentrations of C, FBZ and W of 1.5 pg ml-‘, 4 ng ml-’ and 1.7 pg ml-‘, respectively. The more important analytical characteristics of the methods are summarized in Table 1. Mean speeds

recoveries ( f standard deviation) for seven measurements of a mixture of amounts of C, FBZ and W of 0.5, 0.005 and 1.03 pg ml-’ were 106 f 2, 102 & 2 and 102 f 1% respectively. Other experiments which aimed to produce the non-linear generation of v.a.s.s. by following routes, or by introducing a circular, elliptical or other function into the software routine, did not improve the method because the two-dimensional profiles of the optimized routes were less pronounced when rigid geometric functions were introduced. Attempts are currently being made to apply this technique to other systems with different types of fluorescence distribution. The authors thank the Subdireccion General de Promotion de la Investigation for supporting this study (Projects PB86-0247 and PA86-0033) and Mr. David Schofield for editing and translating the manuscript.

REFERENCES 1 F. Garcia Sanchez and C. Cruces Blanco, Anal. Chem., 60 (1988) 323. 2 C.C. Harris, G. La Veck, J. Groopman, V.L. Wilson and D. Mann, Cancer Res., 46 (1986) 3249.

VARIABLE-ANGLE

SCANNING

FLUORESCENCE

SPECTROMETRY

3 E.L. Inman and J.D. Winefordner, Anal. Chem., 54 (1982) 2018. 4 B.J. Clark, A.F. Fell, K.T. Milne, D.M.G. Pattie and M.H. Williams, Anal. Chim. Acta, 170 (1985) 35. 5 I.M. Warner, in D.M. Hercules, G.M. Hieftje, L.R. Snyder and M.A. Evenson (Eds.), Contemporary Topics in Analytical and Clinical Chemistry, Vol. 4, Plenum, New York, 1982. 6 M.P. Fogarty and I.M. Warner, Anal. Chem., 53 (1981) 259. 7 D.G. Jones, Anal. Chem., 57 (1985) 1057A. 8 F.W. Olesik and J.P. Walters, Appl. Spectrosc., 38 (1984) 578. 9 P.M. Epperson, J.V. Sweedler, R.B. Bilhor, G.R. Sims and M.B. Denton, Anal. Chem., 60 (1988) 327A.

299 10 F. Garcia Sanchez, A.L. Ramos Rubio, V. Cerda and M.T. Oms, Talanta, 35 (1988) 335. 11 R. von Rumker, E.W. Lawless and A.F. Meiners, Production, Distribution, Use and Environmental Impact Potential of Selected Pesticides, Vol. 7, EPA Office of Pesticide Programs, Washington, DC, 1974, p. 133. 12 J.G. Stadley and J.G. Trial, Bull. Environ. Contam. Toxicol., 25 (1980) 771. 13 R.J. Kuhr and H. Wyman, Carbamate Insecticides: Chemistry, Biochemistry and Toxicology, CRC Press, Cleveland, OH, 1976. 14 C.R. Worthing (Ed.), The Pesticide Manual, 7th edn., British Crop Protection Council, London, 1983. 15 C. Cruces Blanc0 and F. Garcia Sanchez, J. Photochem. Photobiol., 42 (1988) 357.