Abstract View
Multidimensional Chromatogram Binning – Positive Matrix Factorization Analysis for Gas Chromatography – High Resolution Mass Spectrometry Datasets
MICHAEL WALKER, Raul Martinez, David Hagan, Haofei Zhang, Lindsay Yee, Allen Goldstein, Brent Williams, Washington University in St. Louis
Abstract Number: 484
Working Group: Instrumentation and Methods
Abstract
Techniques that couple thermal desorption with gas chromatography – mass spectrometry (GC-MS) provide insight into the chemical composition of organic compounds in the atmosphere, which can aid in the identification of the sources of organic aerosol (OA). Recent advances in instrumentation, such as the high time resolution afforded by Thermal Desorption Aerosol Gas Chromatograph (TAG) instruments, or the use of high-resolution (HR) mass spectrometers can offer further information on OA composition. However, these techniques also generate large, complex sets of data that challenge traditional data processing techniques, necessitating novel analysis methods. Previous studies have successfully leveraged a chromatogram binning approach in combination with positive matrix factorization (PMF) to identify covarying compounds across multiple TAG samples, but so far these methods have not been applied to HR datasets.
Proper data alignment, along with the removal of signal related to the analytical technique, are critical aspects of the chromatogram binning process. While HR-MS data can aid in more precise removal of unwanted data features, it also presents additional challenges in the data alignment process. To demonstrate, two datasets related to the Southern Oxidant and Aerosol Study (SOAS) are considered. Despite differences in the two analytical techniques, the first a TAG-style, in-situ characterization of organic aerosol with a one-dimension separation, and the second an offline analysis of filters using a two-dimensional gas chromatography (GCxGC) separation, a generalized workflow can be applied. Dynamic mass-to-charge (m/z) calibration, retention time shift corrections, and multidimensional binning aligns the chromatograms. Measurement artifacts and internal standards are removed through a combination of peak fitting and iterative factor analysis approaches, yielding chromatograms that can be successfully analyzed by PMF to reduce the dimensionality of these complex datasets.