Skip to main content
Dryad

Measurement and modeling of the multi-wavelength optical properties of uncoated flame-generated soot: Data from the BC2, BC3, BC3+ and BC4 studies

Data files

Feb 20, 2018 version files 3.56 MB

Abstract

Optical properties of flame-generated black carbon (BC) containing soot particles were quantified at multiple wavelengths during the BC2, BC3, BC3+ and BC4 campaigns. The particles were produced using two different flames, either an ethylene premixed flame (sampled at multiple heights above the flame front) or a methane diffusion flame. Measurements were made both for size-selected nascent soot particles and for particles that were coated (with either dioctyl sebacate, sulfuric acid or secondary organic aerosol) and then thermally denuded to remove the coating. The coating + denuding process led to a collapse of the soot particles relative to their originally lacy, fractal-like morphology. Observed optical properties for methane diffusion flame-derived particles were similar to those from the ethylene premixed flame-derived particles when sampled 20 cm above the burner. In contrast, the optical properties of particles from the ethylene premixed flame sampled at only 5 cm above the surface differed from both the ethylene premixed flame sampled at 20 cm and the methane diffusion flame. For both the ethylene premixed flame at 20 cm and the methane diffusion flame, the measured wavelength-specific mass absorption cross-sections (MAC) increased somewhat with particle size up to a volume equivalent diameter (dp,VED) of ~160 nm (corresponding to a size parameter x ~ 0.90 at λ = 532 nm), above which the MAC was approximately constant. The measured single scatter albedo (SSA) values exhibited similar behavior. Particle collapse generally led to an increase in SSA, but to negligible changes in the MAC and absorption Ångström exponent (AAE). The observations have been used to derive effective complex refractive index (RI) values for the particles and to test the ability of these theories to represent BC optical properties. Two optical theories were used: (i) spherical particle Lorenz-Mie theory (hereafter referred to as Mie theory), and (ii) Rayleigh-Debye Gans theory, which assumes the particles are aggregates of individual small spheres. Use of the derived RI values from Mie theory with larger particles (with x > 0.9) led to a systematic under-prediction of the absorption cross-sections. In contrast, it was possible to determine complex RI values for use with Rayleigh-Debye-Gans approximation that reproduced the observations for particles with x > 0.9, but over-predicted absorption cross-sections somewhat for lower x values. Although Mie theory reasonably reproduced the observed MACs for particles x < 0.9, it was likely that the increasing MAC with size in this range was due to changes in soot maturity. Overall, these results indicate that, from the point of view of light absorption, these flame-generated particles behave more like agglomerations of individual BC spherules, even after particle collapse, than like large absorbing spheres. Importantly, this implies that the use of Mie theory within air quality and climate models, as is common, may lead to under-predictions in the absorption by BC, with the extent of under-prediction depending on the assumed BC size distribution. We suggest that it is more appropriate to assume a constant, size-independent (but wavelength-specific) MAC to represent absorption by uncoated BC particles within models.