| Abstract | dc.description.abstract | A study of matrix effects arising from river water samples using C18 SPE columns in the determination of 38 pesticides having different chemical properties is presented. Recoveries from river water samples of different origin and the influence of prefiltration step, pH and ionic strength were tested. A calibration equation for the extraction method was calculated by using the statistical method of linear regression, the independent variable being the amount of analyte added and the deppendent variable, the amount of recovered analyte. Extractions were carried out in waters spiked with 65 to 1600 nh L-1 of organophosphorus and nitrogen-containing pesticides, and with 12,5 to 320 ng L-1 of halogenated pesticides. Only trifluralin, captan, aldrin, tri-allate, α-BHC, γ -BHC, and azinphos-methyl behavior could not be described through a linear relationship dueto low precision and, possibly, to interferences caused by the matrix sample. Linearity, analytical sensitivity, precision and limit of detection were established for 31 compounds. Slope values in the regression equation ranged from 0,68 to 0,98, the exception being amitraz (0.43), dieldrin (0,51), and 4,4'DDT (0,58), which accounted for losses prodeced in the extraction process. Systematic errors were detected for most of the compounds and they could be corrected by using the calibration data obtained for the whole analytical process. The detection limits established through the model in most of the cases were lower than the maximum allowed limit for waters by the European Union (100 ng L-1), ranging from 7 to 21 ng L-1 for pesticides determined by GC-ECD, and from 20 to 120 ng L-1 (Except triadimeton, 220) for those determined by GC-NPD. | en_US |
