Abstract Number: 421 Working Group: Aerosol Chemistry
Abstract The molecular diversity of compounds in atmospheric organic aerosol (OA) poses challenges for measurement and modeling of these complex mixtures. Common methods to interpret OA composition in atmospheric or laboratory measurements and numerical simulations include speciated molecule concentrations and atomic ratios. Functional group (FG) abundance in OA provides an intermediate chemical resolution that retains source class fingerprints, capability for OA mass estimation, and physicochemical properties such as hygroscopicity, volatility, and reactivity. However, extracting FG content from measurements such as infrared (IR) absorption spectra or numerical models have been impeded by our lack of computational tools required to harvest this information. We therefore present two sets of tools we have recently developed to accomplish model and measurement interpretation through FGs.
First, we describe tools based on chemoinformatic algorithms for extracting FG abundances from molecularly speciated measurements and simulation. We demonstrate its application to GC-MS measurements and simulations of secondary OA formation using the Master Chemical Mechanism v3.2 with gas/particle partitioning. These FG abundances are compared against observed FG abundances from IR spectroscopy.
We further present chemometric tools for FG estimation of OA using IR spectroscopy. While the sample collection and signal acquisition can be relatively simple by this technique, the processing of the IR spectrum requires several algorithms to correct for substrate interferences and relate the remaining analyte absorbance to abundances of vibrational modes present in the organic mixture. Judicious selection of algorithms permit not only quantification of FG abundance, but statistical identification for the most relevant absorption bands related to concentrations of collocated measurements (such as organic and elemental carbon measured by thermal optical methods). We demonstrate its application to samples routinely collected in the IMPROVE network.