Advances in instrumentation and chemometric analysis for multidimensional chromatography with mass spectrometry detection

Abstract

Quantitative separations by gas chromatography (GC) or liquid chromatography (LC) with mass spectrometry (MS) detection are robust techniques used for the characterization of a broad variety of complex sample matrices. Due to the high complexity of samples, compounds of interest are rarely fully separated in one-dimensional (1D) separations, making quantification and identification a challenge. Two approaches to address this challenge includes using high resolution mass spectrometry (HRMS) for detection to gain selectivity, and using comprehensive two-dimensional chromatography for improving the characterization of complex samples. Each of these solutions involve increasingly more complex instrumentation and provide information rich data sets that require efficient and effective methods to mine the data for obtaining relevant information, as manual inspection can be tedious for hundreds to thousands of analytes. Although GC and LC are commonplace for commercial use, despite its advantages over 1D separations comprehensive multidimensional chromatography still requires optimization in instrumentation and data analysis tools to be more widely adopted. In this dissertation, developments to address challenges in LC-HRMS data processing and analysis as well as developments in instrumentation and chemometric analysis of comprehensive two-dimensional gas chromatography (GC×GC) with MS detection will be discussed. First, a workflow for processing information-rich LC-HRMS data for non-targeted analyses is introduced, with the first demonstration of tile-based Fisher ratio for LC-HRMS data. Next, a novel modulation technique termed dynamic pressure gradient modulation (DPGM) for GC×GC with time-of flight mass spectrometry (TOFMS) is studied in multiple modes. The first mode is negative pulse mode, which was demonstrated for fast separations generating narrow peaks, producing high quality data that is amenable to chemometric decomposition. Next, DPGM is demonstrated in full modulation mode for GC×GC-TOFMS, achieving improvements in sensitivity, limit of detection, and trilinear data structure. DPGM was further investigated, demonstrating the optimization of experimental parameters for obtaining high peak capacity applied to multiple sample matrices. Finally, a novel modification to tile-based F-ratio analysis to handle highly variable metabolomics-related pacu fish samples was introduced with a computational method to remove sample derivatization artifact interferences.