The modelling of the growth of soot particles during the pyrolysis and partial oxidation of aromatic hydrocarbons

Abstract

Previous light-scattering studies on aerosols generated by hydrocarbon pyrolysis in incident shock flows have shown, given a constant particulate volume fraction, that the observed scattered-light intensities agree very closely with those predicted for the free-molecular coagulation of an aerosol having a self-preserving size distribution. One crucial obstacle to the extension of this simple model to include condensational growth has been the measurement of how the particulate volume fraction changes with time. For, not only does the condensed phase contain both soot (which absorbs infrared radiation) and molecular nuclear aromatic species (which do not), but this latter (black) material apparently condenses from the gaseous phase without change in optical absorption coefficient. The approach adopted here has therefore been to generate and test various growth models that span a wide range of assumptions about condensation and nucleation. Only two models, designated COAG and CONCO, can provide quantitative agreement between prediction and observation. Both models require that at any instant the infrared-transparent, light-absorbing polynuclear aromatic intermediates are either all in the gas phase or all in the condensed phase, the switch between the two states corresponding to an instantaneous, massive nucleation step. This interpretation is supported by the marked failure of all growth models based on different assumptions to match the observations. Extensions of the COAG and CONCO models to treat nucleation more realistically give essentially unchanged predictions over substantial domains of assumed initial values of particulate number density and volume fraction. This stability explains why the simple models can correctly describe the results obtained for many shocks spanning a range of temperature, hydrocarbon species and oxygen concentration, where substantial shock-to-shock variations in nucleation rates must exist. At critically low temperatures, with delayed and less rapid nucleation, the CONCO model and extensions thereof must be preferred to the COAG model because they give greater weight to condensational growth.