A Response Surface Study of Atmospheric Black Carbon: An Approach Toward Accuracy in Thermal Optical Analysis (original) (raw)

Agu Fall Meeting Abstracts, 2002

Abstract

Atmospheric Black Carbon (BC) is a ubiquitous and persistent component of atmospheric particulate matter, found at measurable levels in even the most remote locations. BC derives from combustion processes, the sources of which are highly variable spatially and temporally. Consequently, atmospheric BC has no fundamental chemical identity and therefore cannot be accurately measured using absolute methods. Because of its chemically ambiguous character, measurement of BC has long been problematic. Optical properties alone cannot, in principal, quantify both BC mass and the varying aerosol absorptivity upon which mass measurement relies. To overcome this problem, thermal optical analysis (TOA) utilizes the PM's thermal as well as optical properties. A different problem, however, confronts TOA: varying thermal desorption procedures result in different BC results for the same material. Our focus therefore, was to optimize the key TOA parameters that bias BC measurement: production of char (pyrolysis carbon) from organic carbon by the instrument, and separation of the char analytically from the BC that is native to the sample. In this work, we used response-surface methods to optimize a thermal desorption procedure for BC accuracy. To cover a range of samples, we modeled responses for two types of ambient samples (indoor laboratory air and outdoor urban air) and one carbonaceous aerosol source (forest fires). We identified four factors, associated with the temperature, duration and atmosphere (inert or oxidizing), of desorption steps that potentially contribute most to BC measurement variability. The four factors comprised a central composite factorial design for measuring BC concentration and laser response during the course of the TOA analysis. Scanning electron microscope images of particles on the TOA filters were also used to assess char optimization. The response surfaces for the BC to total carbon ratio and laser attenuation response were modeled using full second-order polynomials containing 15 terms that accounted for main effects from the factors, non-linear behavior within factors, and cross-factor interactions. Optimal conditions for BC accuracy were determined by the intersection between the laser response surfaces.

Scott Wight hasn't uploaded this paper.

Let Scott know you want this paper to be uploaded.

Ask for this paper to be uploaded.