Light Absorption Measurement from Thermal/Optical Carbon Analysis: Implications for the OC/EC Split

L.-W. ANTONY CHEN (1), Xiaoliang Wang (1), Judith C. Chow (1), and John G. Watson (1)

(1) Division of Atmospheric Sciences, Desert Research Institute, Reno, NV 89512, USA

     Abstract Number: 819
     Last modified: July 28, 2010

Abstract
Thermal/optical analysis (TOA) separates organic carbon (OC) and element carbon (EC) in particulate matter collected on quartz-fiber filters through differential evaporation and oxidation. Optical (O) monitoring is introduced in some TOA methods to infer fraction of EC that results from pyrolysis of OC during the analysis. Though the pyrolyzed OC (POC) reduces filter reflectance (R) and transmittance (T) simultaneously, EC adjustment based on R and T could vary significantly due to the different response of R and T to POC on the filter surface and/or within the filter. On the other hand, the optical data before and after thermal treatment provide an alternative quantification of EC on the filter by light absorption. In this study, particle absorption (b$_(ATN)) of selected source (Fire Lab at Missoula Experiment [FLAME II]) and ambient (Interagency Monitoring of Protected Visual Environments [IMPROVE] Network) samples were calculated from T of TOA in a manner similar to an aethalometer and related to EC concentration by a 2$^(nd)-order polynomial function that takes into account the multiple scattering and shadowing effects. Better correlations (r$^2 up to 0.98) between EC and b$_(ATN) occurred for samples with lower pyrolysis, i.e. EC (R) ~ EC (T), consistent with POC being the dominant factor that influences the OC/EC split. Regression coefficients of EC versus b$_(ATN) differ between source and ambient samples, likely due to different sampling duration, face velocity, and filter properties. It is suggested that EC (O) can be inferred from optical data of a carbon analyzer with calibrations specific to aerosol type and sampling configuration. EC (R) and EC (T) tend to overestimate and underestimate EC (O), respectively, and the discrepancies can be explained by EC loading and POC/EC ratio.