HR: 14:52h
AN: A23D-06    [Abstracts]
TI: Deconvolution and Quantification of Primary and Oxygenated Organic Aerosols: Technique Development and Applications to the Pittsburgh AMS Datasets
AU: * Zhang, Q
EM: zhangq@cires.colorado.edu
AF: CIRES, University of Colorado, Boulder, CO 80309 United States
AU: Jimenez, J
EM: jose.jimenez@colorado.edu
AF: CIRES, University of Colorado, Boulder, CO 80309 United States
AU: Alfarra, R
EM: rami.alfarra@umist.ac.uk
AF: SEAES, University of Manchester, Manchester, M60 1QD United Kingdom
AU: Allan, J
EM: james.allan@physics.org
AF: SEAES, University of Manchester, Manchester, M60 1QD United Kingdom
AU: Coe, H
EM: hugh.coe@umist.ac.uk
AF: SEAES, University of Manchester, Manchester, M60 1QD United Kingdom
AU: Worsnop, D
EM: worsnop@aerodyne.com
AF: ARI, Aerodyne Research Inc, Billerica, MA 01821 United States
AU: Canagaratna, M
EM: mrcana@aerodyne.com
AF: ARI, Aerodyne Research Inc, Billerica, MA 01821 United States
AB: A new technique has been developed to deconvolve and quantify the mass concentrations of primary and oxygenated organic aerosol (POA and OOA) using highly time-resolved organic mass spectral data obtained with an Aerodyne Aerosol Mass Spectrometer (AMS). OOA may comprise secondary organic aerosol (SOA) as well as oxidized POA. This technique involves a series of multivariate linear regressions that use mass-to-charge ratios (m/z's) 57 (mostly C4H9+) and 44 (mostly CO2+), the identified AMS mass spectral tracers for POA and OOA, respectively, as the initial principal components followed by an iterative algorithm to evaluate and "purify" POA and OOA mass spectral tracers. We have applied this technique to the AMS organic aerosol data acquired at the EPA Pittsburgh Supersite during September 2002 and have observed excellent agreement between the reconstructed organic concentrations (= POA + OOA) and the measured values (r2 = 0.997, slope = 0.998). The reconstructed organic data matrix (size = 3199 time steps x 300 m/z's) explains 99% of the variance in the measured time series. The extracted mass spectrum of POA shows high similarity to those of diesel exhaust sampled during a chase study, lab-measured lubricating oil, and freshly emitted traffic aerosols measured in urban environments. The spectrum of OOA closely resembles those of aged organic aerosols sampled in remote areas and also shows similarity with the spectrum of fulvic acid-a humic-like substance that is ubiquitous in the environment and has previously been used as an analogue to represent polyacid compounds found in highly processed and oxidized atmospheric organic aerosols. Organic aerosols in Pittsburgh during Sept. 2002 are mainly oxygenated, on average consisting of ~ 70% OOA (likely mainly secondary in nature). Pronounced diurnal variations in the POA/OOA contributions to organic mass have been observed, with the contribution of POA peaking in the morning rush hours while that of OOA is largest in the afternoon between 3-4 pm. We compared these results to those from the EC/OC tracer method for POA/SOA estimation, and observed qualitative agreement in estimated concentrations between the two methods. The diurnal variations in the mass concentrations and size distributions of POA and OOA will be reported. We have also performed case studies on how the chemistry of POA and OOA changes during nucleation events and acidic particle periods. From these results, implications related to roles of organics in new particle growth and acid catalyzed secondary organic aerosol formation will be discussed. Applications of this algorithm to worldwide AMS datasets will be presented as in an invited talk by Jimenez et al. (this conference).
DE: 0305 Aerosols and particles (0345, 4801)
DE: 0345 Pollution--urban and regional (0305)
DE: 0365 Troposphere--composition and chemistry
DE: 0368 Troposphere--constituent transport and chemistry
DE: 0394 Instruments and techniques
SC: Atmospheric Sciences [A]
MN: 2004 AGU Fall Meeting