 B. K. Matuszewski

B. K. Matuszewski is an author of experimental research on quantitative high-performance liquid chromatography with tandem mass spectrometry. A2003 paper with collaborators, listing an affiliation at Merck Research Laboratories in West Point, Pennsylvania, develops ways to distinguish an analytical matrix effect from extraction recovery. The byline identifies the author with these initials.^[ref]

The scientific problem is that an instrument signal can change even when the amount of analyte reaching the measurement is unchanged. Other constituents of the sample matrix can suppress or enhance the response. Conversely, analyte can be lost during sample preparation without changing the response per unit amount. A low measured signal therefore need not identify which mechanism occurred.

An original three-preparation example separates these effects under a linear-response assumption. Suppose equal nominal amounts of an analyte are introduced into: a clean solvent standard; a blank sample extract after extraction; and a sample before extraction. Match the final volumes and measurement conditions. Let the resulting background-corrected signals be $A=100$, $B=80$ and $C=60$ arbitrary units, respectively. The dimensionless ratios areThe last ratio is the product of the first two. Calling the overall60percent response an extraction recovery would conflate analyte loss with ion suppression. This is an invented calculation, not a reconstruction of the paper's validation data. It assumes that the post-extraction addition represents the matrix response experienced by the extracted analyte.

Matuszewski and colleagues also compared ion-spray and heated-nebulizer interfaces while holding sample preparation and chromatographic conditions otherwise the same. For their investigated compound, a matrix effect appeared with the ion-spray interface and was absent with the heated nebulizer.^[ref] That controlled comparison shows why selectivity in mass spectrometry must be tested for the actual method and sample matrix, rather than inferred from the detector's nominal specificity.