Enhancing Organic Characterization from TOF-ACSM Mass Spectra Using Collocated Functional Groups Measurements for ASCENT Sites

NA MAO, Manjula Canagaratna, Nga Lee Ng, Satoshi Takahama, Ruizhe Liu, Sohyeon Jeon, Theobard Habineza, Albert Presto, Seonsik Yun, Douglas A. Day, Jose Jimenez, Elena Goodspeed, Yingjie Shen, Shane Murphy, Haroula D. Baliaka, Ryan Ward, John Seinfeld, Richard Flagan, Nicolas A. Buchenau, Jason Surratt, Ann M. Dillner, University of California, Davis

     Abstract Number: 570
     Working Group: Carbonaceous Aerosols

Abstract
The Atmospheric Science and Chemistry mEasurement NeTwork (ASCENT) is a new high time resolution network that employs Time of Flight Aerosol Chemical Speciation Monitors (ToF-ACSM) to measure organics and other species in aerosol particles. Eight ASCENT sites have parallel functional group measurements from Fourier Transform Infrared Spectroscopy (FTIR) analysis of filter samples: six are Interagency Monitoring of Protected Visual Environments (IMPROVE) sites with routinely collected filters and two other sites are collecting filters for FTIR analysis. To enhance the organic characterization provided by ToF-ACSM at ASCENT sites and elsewhere, we qualitatively investigate the relationship between ACSM fragment ions and functional groups quantified by FTIR analysis, specifically carboxylic acids, alcohols, non-acid carbonyls, and alkane CH groups. In this study, we systematically explore the connections between fragment ions (m/z 29, 43, 44, 55, 57, 60, 69, 71, and 73) and their associated functional groups (carboxylic acids and alcohols) at several ASCENT sites, including Atlanta, Denver, Pittsburgh, and Yellowstone. We employ statistical methods from simple multilinear regression (ML) through more complex statistical methods like stepwise regression (SR) and Elastic Net (EN). Our research encompasses various chemical representatives, including 22 pure compounds, 20 lab-generated mixtures, and a year’s ambient ACSM data from ASCENT sites, varying in their organic carbon to oxygen (O:C) and hydrogen to carbon (H:C) ratios and sources, and covering categories such as hydrocarbon-like organic aerosol (OA), biomass-burning OA, cooking OA, and oxidized OA. The response to the alcohol group is generally somewhat linear with m/z 29, whereas the carboxylic acid functional group is nonlinear with m/z 44. Currently, the alcohol groups are predicted more accurately than carboxylic acids. Additionally, the Elastic Net model demonstrates superior predictive performance, as indicated by metrics such as R², bias, and error across all three models.