We are performing a multi-component isotopic dilution assay by LC-MS/MS where occasionally, we get a field sample where one particular analyte does not give linear dilution results within the calibration curve range. The problem came to light when a routine dilution of a sample that was just above the top calibrator gave odd results upon dilution. The difference between results for unfiluted, 5x, 10x, 15x dilutions varied by 50-70%, even though they were all within the original calibration curve.

The method uses a matching 13C-labeled internal standard, so matrix effects should (?) be corrected. There is minimal matrix suppression/enhancement of the internal standard raw area counts. The calibration curves for all of the transitions are run within the same analysis batch are linear, weighted 1/x and meet routine QC technical guidelines. The remainder of the analytes behave as expected during dilution.

A full-scan MS analysis of the problematic undiluted sample didn't show any large interference peaks, only a slightly larger amount of in-source fragmentation of the target compound compared to the standards.

Not quite sure where to start with this troubleshooting, since it only occurs with unknown field samples. Cleaning of the source, checking the instrument tune and settings didn't change the result on sample reinjection.