Electrospray ionization matrix effect as an uncertainty source in HPLC/ESI-MS pesticide residue analysis

The matrix effects in HPLC/electrospray ionization (ESI)-MS analysis are difficult to compensate for because of their large variability. It is, therefore, often more practical to include uncertainty due to the matrix effect into the uncertainty budget rather than try to compensate. This work presents an empirical approach--the matrix effect graph approach--for estimating the uncertainty due to the matrix effect in HPLC/ESI-MS analysis of pesticide residues in fruits and vegetables. At certain time intervals (1 month), a calibration graph using extracts of different fruits/vegetables as calibration solutions is prepared, and a regression line is fitted through these data. These fruits/vegetables may be either from the commodity group of the samples or from different commodity groups. The relative residuals of the calibration point peak areas are calculated and plotted against the measurement time. We term the resulting graph the matrix effect graph. The root mean square of the relative residuals is calculated and used as the estimate of relative uncertainty of the sample peak areas caused by the matrix effect. The matrix effect graph obtained over fruits/vegetables from different commodity groups can also be used to identify fruits/vegetables with extreme matrix effects. The matrix effect graph approach was used for determination of thiabendazole, aldicarb, imazalil, and methiocarb and was validated with tomato, cucumber, and sweet corn matrixes at the 0.5 mg/kg concentration level. When different commodity groups were used to compile the matrix effect graph, results of analysis of all samples agreed with the spiked concentrations within the expanded uncertainties (at k=2 level). When the matrix effect graph was compiled using fruits from the same commodity group as the analyzed samples (fruiting vegetables in this case), agreement was found in 98% of the cases.