Development of a Database for Factor Analysis-Derived Profiles from TAG Measurements

MICHAEL WALKER, Claire Fortenberry, Brent Williams, Washington University in St. Louis

     Abstract Number: 284
     Working Group: Source Apportionment

Abstract
Over the last 15 years, different types of thermal desorption aerosol gas chromatograph (TAG) instruments have been utilized in laboratory and field measurements to provide insight to the composition of organic gases and particles in the atmosphere by coupling in-situ collection and thermal desorption techniques with gas chromatography – mass spectrometry (GC-MS). However, the amount of data generated by these instruments and time required for processing remain significant barriers, as thousands of compounds may be identified in the tens of thousands of spectra that can be acquired hourly.

As an alternative, or complement, to peak integration methods, positive matrix factorization and related factor analysis methods can be leveraged to make the data more manageable. Depending upon the specific processing approach, the generated factor profiles can help identify chemically distinct classes of organic compounds based upon their similar mass spectral fragmentation patterns, or can help to identify compounds that covary across chromatograms, indicative of a common source. In order to further leverage the wealth of information contained within these factor profiles, a custom database has been developed to serve as a repository for existing and future datasets.

To highlight the benefits of this database, and the various methods for quantitatively comparing the profiles, factors from both indoor and outdoor measurements during the Air Composition and Reactivity from Outdoor aNd Indoor Mixing (ACRONIM) campaigns were utilized to identify related factors from a variety of laboratory and field measurements. Despite differences in operating conditions, instrumentation and measurement locations, appropriate processing of the data allows for direct comparisons of the volatility-resolved chemical information, providing the framework required for future, real-time source apportionment with TAG measurements. Furthermore, the framework of this database could readily be extended to any measurement technique that involves a separation step prior to detection with a mass spectrometer.