A comparative study of the correction of systematic errors in the quantitation of pyrethroids in vegetables using calibration curves prepared using standards in pure solvent.

A comparative study of two mathematical approaches was performed in order to correct systematic errors due to the presence of the unexpected interferences which appear when the quantitation of the analyte in real samples is carried out with calibration curves built using standards in pure solvent. These methods consisted in the establishment of different mathematical expressions which transform the concentration (Cs) obtained using calibration graphs built using pure solvent into the corrected concentration (C(M)) that should be obtained if the quantitation is carried out with calibration curves built using standards dissolved in blank matrix extracts. In the two approaches the correction is performed from the results of an intermediate precision study which was carried out using both calibration graphs (prepared using pure solvent and blank matrix extract). By using ANCOVA to compare the slope of both solvent-based and matrix-matched calibration graphs, matrix effect was found in the determination of deltamethrin in tomato and acrinathrin in tomato and pepper. In these cases, both approaches led to good results.

Determination of carbendazim, thiabendazole and fuberidazole using a net analyte signal-based method.

The net analyte signal (NAS)-based method HLA/GO, modification of the original hybrid linear analysis (HLA) method, has been used to determine carbendazim, fuberidazole and thiabendazole in water samples. This approach was used after a solid-phase extraction (SPE) step, using the native fluorescence emission spectra of real samples, previously standardized by piecewise direct standardization (PDS). The results obtained show that the modification of HLA performs in a similar way that partial least-squares method (PLS-1). The NAS concept was also used to calculate multivariate analytical figures of merit such as limit of detection, selectivity, sensitivity and analytical sensitivity (gamma(-1)). With this purpose, blanks of methanol and ternary mixtures, with the target analyte at low concentration and the other two ranging according to the calibration matrix, were used, with different results. Detection limits calculated in the last way are more realistic and show the influence of the other components in the sample. Selectivity for carbendazim is higher than the corresponding values for fuberidazole and thiabendazole, whereas sensitivity, as well as the values obtained for their detection limits, are lower for carbendazim, followed by thiabendazole and fuberidazole. Results obtained by modification of HLA vary in the same way that the ones obtained by PLS-1.

Correction of predicted concentration in the use of solvent-based calibration lines for determining carbendazim, fuberidazole and thiabendazole in water after a SPE step.

This paper addresses an attempt to overcome the deviation that results from the use of a solid phase extraction (SPE) procedure for extracting trace levels of three benzimidazole pesticides (carbendazim, fuberidazole and thiabendazole) from water samples, for their subsequent quantitative determination by spectrofluorimetry, using univariate calibration. The deviation is due to an attenuation effect originating in the C(18) cartridge used in the SPE step. The approach developed is based on the calculation of a correction factor (fc) that is dependent on the signal measured after the SPE step. In order to calculate fc a study of the intermediate precision of two calibration graphs (with and without SPE) was performed. The fc was added to the predicted concentrations for the analytes using a calibration graph for pure solvent, built every time that the analysis is done. In addition, predictions were made using both average calibration graphs obtained from the intermediate precision study. In this study, the first of these three options was shown to improve the accuracy of predictions in the presence of matrix effects.

Trace determination of carbendazim, fuberidazole and thiabendazole in water by application of multivariate calibration to cross-sections of three-dimensional excitation-emission matrix fluorescence.

The simultaneous determination of carbendazim, fuberidazole and thiabendazole was accomplished by cross-section (CS) fluorimetry in combination with multivariate calibration algorithms. The total luminescence information of the compounds was used to optimise the linear trajectories of the CS. A comparison between principal component regression (PCR) and two partial least squares (PLS) algorithms, PLS-1 and PLS-2, with different pre-processing methodologies was made. The final model, which applied the PLS-1 method, built using pesticide standard and emission spectra, was successfully used for the determination of these compounds in synthetic mixtures. However, a different PLS-1 multivariate calibration model, based on CS through the total luminescence spectroscopic data, was necessary for determining the cited pesticides in water samples. Mean centring was the best pre-processing technique in both PLS-1 models. This later calibration model was built from ultra-pure water samples spiked with known carbendazim, fuberidazole and thiabendazole concentrations, after solid-phase extraction (SPE). The method, which had a precision better than 5%, was shown to be suitable for carbendazim, fuberidazole and thiabendazole monitoring in water samples at trace levels.