Standard addition method based on four-way PARAFAC decomposition to solve the matrix interferences in the determination of carbamate pesticides in lettuce using excitation–emission fluorescence data

•Determination of carbamate pesticides in iceberg lettuce by PARAFAC and EEM data.•The second-order advantage is useful to resolve highly overlapped spectral profiles.•EEM data measured at different dilutions of the extract has provided four-way data.•The quadrilinear PARAFAC model enabled to identify the matrix in each dilution.•A D-optimal design was used to select concentrations for the standard addition method.

The simultaneous determination of two carbamate pesticides (carbaryl and carbendazim) and of the degradation product of carbaryl (1-naphthol) in iceberg lettuce was achieved by means of PARAFAC decomposition and excitation–emission fluorescence matrices. A standard addition method for a calibration based on four-way data was applied using different dilutions of the extract from iceberg lettuce as a fourth way that provided the enough variation of the matrix to carry out the four-way analysis. A high fluorescent overlapping existed between the three analytes and the fluorophores of the matrix. The identification of two fluorescent matrix constituents through the four-way model enabled to know the matrix contribution in each dilution of the extract. This contribution was subtracted from the previous signals and a subsequent three-way analysis was carried out with the tensors corresponding to each dilution. The PARAFAC decomposition of these resulting tensors showed a CORCONDIA index equal to 99%. For the identification of the analytes, the correlation between the PARAFAC spectral loadings and the reference spectra has been used. The trueness of the method, in the concentration range studied, was guaranteed because there was neither constant nor proportional bias according to the appropriate hypothesis tests. The best recovery percentages were obtained with the data from the most diluted extract, being the results: 127.6% for carbaryl, 125.55% for carbendazim and 87.6% for 1-naphthol. When the solvent calibration was performed, the decision limit (CCα) and the capability of detection (CCβ) values, in x0=0, were 2.21 and 4.38 μg L−1 for carbaryl, 4.87 and 9.64 μg L−1 for carbendazim; and 3.22 and 6.38 μg L−1 for 1-naphthol, respectively, for probabilities of false positive and false negative fixed at 0.05. However, these values were 5.30 and 10.49 μg L−1 for carbaryl, 18.05 and 35.73 μg L−1 for carbendazim; and 1.92 and 3.79 μg L−1 for 1-naphthol, respectively, when the matrix-matched calibration using the most diluted extract was carried out in the recovery study.

Graphical abstractFigure optionsDownload full-size imageDownload as PowerPoint slide