Deconvolution and Quantification of Hydrocarbon-like and Oxygenated Organic Aerosols Based on Aerosol Mass Spectrometry QI ZHANG, † M. RAMI ALFARRA, §, ⊥ DOUGLAS R. WORSNOP, | JAMES D. ALLAN, § HUGH COE, § MANJULA R. CANAGARATNA, | AND JOSE L. JIMENEZ* ,†,‡ Cooperative Institute for Research in Environmental Sciences (CIRES) and Department of Chemistry and Biochemistry, 216 UCB, University of Colorado-Boulder, Boulder, Colorado 80309-0216, School of Earth, Atmospheric and Environmental Science, Main Building, Sackville Street, University of Manchester, Manchester M60 1QD, U.K., and Aerodyne Research Inc., Billerica, Massachusetts 01821-3976 A new technique has been developed to deconvolve and quantify the mass concentrations of hydrocarbon-like and oxygenated organic aerosols (HOA and OOA) using highly time-resolved organic mass spectra obtained with an Aerodyne Aerosol Mass Spectrometer (AMS). This technique involves a series of multivariate linear regressions that use mass-to-charge ratios (m/z’s) 57 (mostly C 4 H 9 + ) and 44 (mostly CO 2 + )sthe identified AMS mass spectral tracers for HOA and OOA, respectivelysas the initial principal components. Two algorithms have been developed: algorithm 1 is based solely on m/z 44 and m/z 57, and algorithm 2 is an iterative procedure expanded from algorithm 1. This technique was applied to the AMS organic aerosol data acquired at the EPA Pittsburgh Supersite during September 2002. The reconstructed organic concentrations () HOA + OOA) agree well with the measured values (r 2 ) 0.997, slope ) 0.998), and the reconstructed organic data matrix (size ) 3199 time steps × 300 m/z’s) explains 99% of the variance in the measured time series. In addition, the extracted mass spectrum of HOA shows high similarity to those of diesel exhaust, lubricating oil, and freshly emitted traffic aerosols observed in urban areas, while the spectrum of OOA closely resembles those of aged organic aerosols sampled in rural areas and also shows similarity with the spectrum of fulvic acids a humic-like substance that is ubiquitous in the environment and has previously been used as an analogue to represent polyacid components found in highly processed and oxidized atmospheric organic aerosols. There is evidence for the presence of a third component, although its contribution to the total organic signal appears to be small in this study. The most important result is that m/z 44 and m/z 57 are reliable AMS mass spectral “markers” that provide the “first guess” for algorithm 2 which allows the quantitative description of the organic aerosol concentration and mass spectra over a period of 16 days in a major urban area and allows the extraction of mass spectra of OOA and HOA that can be interpreted chemically. These findings indicate the potential of performing organic source apportionment on the basis of total particle mass, rather than on the basis of organic tracer compounds that contribute a small fraction of this mass. Introduction Organic material comprises a significant, yet poorly char- acterized, fraction of the fine particles in the atmosphere (1-4). The number and complexity of particulate organic compounds make it a significant challenge to fully charac- terize their molecular identities (1, 2). Analysis of a large number of molecules, for example, typically accounts for only a minor fraction of the organic carbon in aerosols (5-7). As a result, there is only a limited understanding of the chemistry, sources, and processing of organic aerosols, and assessments of their impacts on climate, visibility, and human health remain notably uncertain (1, 6, 8, 9). Organic aerosols originate from many different natural and anthropogenic sources and processes. Primary organic aerosols (POA) are those emitted directly into the atmosphere in particle form, e.g., from fossil fuel and biomass combustion, while secondary organic aerosols (SOA) are formed from gaseous precursors through gas-phase (1, 2, 6), particle-phase (10), or aqueous-phase reactions (11, 12). Due to their different origins and formation mechanisms, POA and SOA usually demonstrate very different chemical and micro- physical properties (13-18). Therefore, to design effective fine particle control strategies and to better evaluate the roles of organic aerosols in regional and global climate we must understand the concentrations, properties, and sources of these two organic aerosol types (6). Three general techniques are available to estimate the relative contributions of POA and SOA to ambient particle mass. One is “source-oriented” chemical transport models (CTMs) that simulate atmospheric transport and chemical reactions (6, 19-21). Another is “source-receptor” methods that interpret ambient measurements with a mathematical model that uses distinctive tracers and previous knowledge of the tracer concentrations in all known sources (6, 22, 23). Commonly applied organic source-receptor methods include the organic molecular marker chemical mass balance (CMB) method (22-24) and the organic carbon (OC) to elemental carbon (EC) ratio method (25-29). The third is receptor- only methods, which are conceptually similar to principal component analysis (30) and use statistical multivariate techniques to extract source information from the time series of multiple simultaneous measurements. Commonly used receptor methods for air pollution studies include Positive Matrix Factorization (PMF) (31) and UNMIX (32). Unlike “source-oriented” CTMs, source-receptor and receptor methods do not require detailed information on meteorology or emission inventories but instead perform source apportionment using ambient measurements as inputs (6). So far, receptor methods have been relatively successful with POA applications but show limited capability to distinguish the sources of SOA components unless * Corresponding author phone: (303)492-3557; fax: (303)492-1149; e-mail: jose.jimenez@colorado.edu. † Cooperative Institute for Research in Environmental Sciences (CIRES), University of Colorado-Boulder. ‡ Department of Chemistry and Biochemistry, University of Colorado-Boulder. § University of Manchester. | Aerodyne Research Inc. ⊥ Present address: Paul Scherrer Institute, 5232 Villigen PSI, Switzerland. Environ. Sci. Technol. 2005, 39, 4938-4952 4938 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 13, 2005 10.1021/es048568l CCC: $30.25  2005 American Chemical Society Published on Web 05/27/2005