Elemental Speciation by Parallel Elemental and Molecular Mass Spectrometry and Peak Profile Matching

Abstract

Trace elemental speciation in complex, real-world matrixes is a daunting task because of the low concentration of metals/metalloids and the correspondingly high molecular chemical noise. We constructed a liquid chromatography parallel elemental and molecular mass spectrometry (PEMMS) system and evaluated the use of peak elution profiles to identify trace molecular species containing specific heteroatoms, using the case of Se in yeast. We demonstrate that it is possible to use the HPLC-inductively coupled plasma (ICP)MS peak profile (retention time, width) to identify candidate ions with matching peak profiles in the molecular MS data. Proof of principle was demonstrated by C18 separation of three Se-amino acid standards (0.005-15 ppm as Se). The molecular MS (atmospheric pressure chemical ionization time-of-flight, APCI-TOF-MS) data set was converted into selected ion chromatograms of 0.05 Th width. ICPMS and APCI-TOF-MS ion chromatograms were fit by the Haarhoff-VanderLinde function using the following parameters: area, retention time, width, and skew. The ICPMS fit parameters were more reproducible than the APCI-TOF-MS fit parameters from run to run, and the APCI-TOF-MS signal was expected to limit correlation in most circumstances. Retention time and width were found to correlate well between the two MS systems for APCI-TOF-MS peaks with signal-to-fit-error (S/FE) of >25. Correction factors for differences in flow path length and peak broadening were required. The normalized correction factors were species and concentration independent and were stable from run to run. The skew parameter was found to be highly susceptible to noise and was not generally useful in matching ICPMS and APCI-TOF-MS peaks. An artificially noisy sample was generated by spiking 30 ppb Se-methionine (SeMet) and 5 ppb Se-methylselenocysteine (SeMSC) with unselenized yeast extract and run by PEMMS. The PEMMS software was able to detect four molecular MS peaks associated with SeMet and two for SeMSC, while filtering out >40 coeluting spectral peaks associated with chemical noise in each sample. In summary, we have demonstrated that correlated information in peak shape between parallel detectors can facilitate detection of trace elemental species in complex matrixes.