Modelling secondary organic aerosol formation : from chemical mechanistic modelling to empirical modelling

Abstract

The work presented in this thesis is primarily concerned with modelling the formation of secondary organic aerosols (SOAs). SOAs cannot easily be measured with direct analytical chemical methods; indirect methods like applying organic carbon to elemental carbon ratios and utilising computer models have been employed to provide an estimate of the SOA mass concentrations in ambient air. The five models presented in this work were either developed or assessed using environmental chamber data.

Chamber experiments were undertaken using initial isoprene concentrations in the range of 22 ppb to 343 ppb, with the reactive organic carbon (ROC) to NOx ratios in the range of 2.0 to about 18. Chamber experiments were also performed for the a-pinene / NOx system with initial a-pinene concentrations ranging from 79 ppb to 225 ppb, with ROC/NOx ratios varying from 5.5 to about 41. All of the experiments were performed without the addition of propene or seed aerosol. Background aerosol levels were very low for the experiments presented in the thesis and so homogeneous nucleation processes were considered to occur in the chamber in addition to absorption and oligomerisation formation processes. Initial nucleation events resulting from the photooxidation of isoprene could be detected once the aerosol diameter was greater than 12 nm. In the a-pinene system,new particles formed via homogeneous nucleation processes were detectable in the 100-200nm diameter range.

The models presented range in complexity from the near explicit Master Chemical Mechanism to an empirical model whose key feature is its simplicity. The mechanistic model provides an insight into the SOA formation pathways and the influence of varying the initial experimental conditions and the duration of photooxidation on the simulated SOA composition. The aim of the empirical model is to simulate the SOA mass concentration produced during a chamber experiment. The development of the model is intentionally simple so that it can be applied to any hydrocarbon and has been applied successfully to isoprene and a-pinene chamber experiments. In this way, the empirical model is presented as an alternative approach to predicting the temporal variation in SOA mass concentrations.

An analysis of the partitioning absorption models developed by Odum et al. (1996) and Hoffmann et al. (1997) has informed the development of the SOA module which has been coupled to a 3D atmospheric model. Embodied within the SOA module is the gas / aerosol partitioning theory which includes the model proposed initially by Pankow et al. (1994) and by Odum et al. (1996).