Abstract View
Comprehensive Detection of All Analytes in Large Chromatographic Atmospheric Dataset
SUNGWOO KIM, Gabriel Isaacman-VanWertz, Virginia Tech
Abstract Number: 395
Working Group: Instrumentation and Methods
Abstract
Ambient aerosols may contain hundreds or thousands of unique organic compounds, and even minor components may provide valuable insight into particle sources and formation chemistry. To gain insight into sources and transformations, individual components are frequently identified and quantified using gas chromatography/mass spectrometry. However, due to the complexity of this data and the highly dynamic nature of aerosol composition, data reduction is a significant bottleneck in analysis. Consequently, typically only a few dozen analytes are reported for a dataset, and a large amount of potentially useful data are discarded. We present here an automated approach of cataloging and potentially identifying all analytes in a large chromatographic dataset of ambient aerosols. We use a coupled factor analysis/decision tree approach to de-convolute peaks and comprehensively identify analytes. Positive Matrix Factorization (PMF) of small sub-sections of multiple chromatograms to extract factors that describe individual or small numbers of analytes. Potential chromatographic peaks in these factors are evaluated based on features such as peak shape, noise, and retention time. With our approach, all analytes within the small section of the chromatogram are cataloged, and the process is repeated for overlapping sections across the chromatogram, generating a complete list of the retention times and estimated mass spectra of all peaks in a dataset. We validate this approach using samples of known compounds, successfully cataloging all known analytes and resolving unknown contaminants. We demonstrate the separation of co-eluted peaks, including components with highly similar mass spectra and little-to-no chromatographic resolution, and the resolution of peaks that appear in only a fraction of chromatograms. As a case study, this method is applied to a complex real-world dataset representing months of hourly particle-phase organic composition, from which upwards of 400 analytes are resolved.