Abstract View
Analysis of the Unresolved Complex Mixture of Intermediate Volatile Organic Compounds in Gas Chromatograph-Mass Spectrum Data using Positive Matrix Factorization
Quanyang Lu, Christopher Hennigan, Albert Presto, Yunliang Zhao, Neil Donahue, ALLEN ROBINSON, Carnegie Mellon University
Abstract Number: 268
Working Group: Instrumentation and Methods
Abstract
We present a new approach based on Positive Matrix Factorization (PMF) to better characterize the unresolved complex mixture (UCMs) in gas-chromatograph–mass-spectrometer (GC-MS) data. The goal of this work is to recover information on the chemical composition of the intermediate volatility organic compounds (IVOC) UCM for use in source apportionment and for simulation of secondary organic aerosol formation. We identified the chemical character of the PMF factors by comparison with reference compounds in the NIST mass spectral database. This technique can recover more detailed compositional information than traditional UCM assignments. PMF analysis of mobile source samples (gasoline, diesel and aircraft) show the effects of emissions control technologies, fuel composition and engine load on IVOC composition. For example, gasoline vehicle sources show increasing fraction of alkanes and oxygenates and decreasing fraction of single-ring aromatics (SRAs) from small off-road engine (SORE) to newer on-road vehicles. IVOC emissions from biomass burning show very different composition than mobile sources, with a large fraction (>50%) of IVOC oxygenates. Combing PMF technique with chemical mass balance model (CMB) on samples collected in a highway tunnel, we show that even though diesel vehicles represented only 20% of fuel consumption, they contributed around 70% of mobile source IVOCs. PMF analysis on ambient samples in Pasadena, CA show very high fractions (>70%) of oxygenates, where minor hydrocarbon IVOC peaks in early afternoon suggest contributions from local evaporative emissions. SOA modelling on mobile source emissions under low- and high-NOx conditions highlighting the importance of IVOC chemical information in chemical transport models. This technique can be applied to both archival and future GC-MS data analysis.