Optimizing thermal-optical methods for measuring atmospheric elemental (black) carbon: A response surface study

Abstract

The chemical, physical, and morphological complexity of atmospheric aerosol elemental carbon (EC) presents major problems in assuring measurement accuracy. Since EC and black carbon are often considered equivalent, methods based on thermal-optical analysis (TOA) are widely used for EC in ambient air samples because no prior knowledge of the aerosol's absorption coefficient is required. Nevertheless, different TOA thermal desorption protocols result in wide EC-to-total-carbon (TC) variation. We created three response surfaces with the following response variables: EC/TC, maximum laser attenuation in the He phase (Lmax), and laser attenuation at the end of the He phase (L He4). A two-level central-composite factorial design comprised of four factors considered the temperatures and durations of all desorption steps in TOA's inert (He) phase and the initial step in TOA's oxidizing (O 2-He) phase. Lmax was used to assess the positive bias caused by nonvolatile unpyrolized organic carbon (OC char) being measured as native EC. A negative bias that the attenuated laser response does not detect is caused by the loss of native EC in the He phase. LHe4 was used as a surrogate indicator for the loss of native EC in the He phase. The intersection between the Lmax and LHe4 surfaces revealed TOA conditions where both the production of OC char in the He phase was maximized and the loss of native EC in the He phase was minimized, therefore leading to an optimized thermal desorption protocol. Based on the sample types used in this study, the following are generalized optimal conditions when TOA is operated in the fixed-step-durations, laser-transmission mode (i.e., TOT): step 1 in He, 190°C for 60 s; step 2 in He, 365°C for 60 s; step 3 in He, 610°C for 60 s; step 4 in He, 835°C for 72 s. For steps 1-4 in O2-He, the conditions are 550°C for 180 s, 700°C for 60 s, 850°C for 60 s, and 900°C for 90 s to 120 s, respectively.