Skip to main content

Laboratory Experiments on Particle Formation and Growth

Photochemical Formation of Organic Aerosols

The oxidation of hydrocarbons has a central role in the formation of both photochemical smog and secondary particulate matter.  We have been investigating these reactions in collaboration with Professor D.R. Hastie's group.  The project consists of a smog chamber studies of gas-to-particle conversion resulting from the oxidation of organic compounds. In these experiments, a single organic compound, such as toluene or β-pinene, is oxidized by photochemically produced OH radicals. The size distribution of the product aerosol is measured as a function of time using a differential mobility analyzer (DMA). The object is to determine what fraction of the oxidation products ends up in the particles and to chemically identify those products. The methods that Bart Verheggen developed to analyze field measurements of particle formation and growth are used in analyzing the data obtained in these experiments.  Hastie's group is using tandem mass spectrometry methods to identify some of the many reaction products formed in these experiments. 

The Mozurkewich group is developing a new approach to the experimental investigation of these complex reaction mechanisms.  The analytical methods applied to the study of these reactions usually identify a subset of products and fail to account for a significant fraction of the carbon.  Many of these products are not very stable, so that suitable standards aren't available; this makes both qualitative and quantitative analysis difficult.  Time resolution is limited making it difficult to obtain quality kinetic data.  As a result, many studies only try to identify products and determine yields; the yields may be inaccurate due to reactions of the products. 

Our focus will be on kinetics, rather than stoichiometry and product identification.  As hydrocarbon oxidation takes place in the smog chamber, a sample will be continuously drawn into a direct sampling time-of-flight mass spectrometer.  The objective is to obtain a signal with good time resolution from as many components as possible in order to work out the detailed reaction mechanism. Since may different compounds will be present, this will result in extremely complex mass spectra.  We will use weighted principal component analysis to isolate the mass spectra for individual components.  This should enable us to follow all major reaction products with a time resolution sufficient for kinetic analysis.  Aerosol measurements will be made simultaneously in order to incorporate particle growth. 

For this work we are using a time-of-flight (ToF) mass spectrometric measurements equipped with a quadrupole ion trap (Paul trap) for ion storage. Sample air is mixed with He and flows into the ion trap. Some of the gas escaping from the ion trap is ionized by an electron beam; the resulting ions are then transferred back into the ion trap where they are accumulated and stored. The vast majority of these ions are quickly converted to O2+; these are stored in the trap for about 100 ms prior to injection into the ToF. During this time, the O2+ acts as a chemical ionization reagent; almost all organic molecules react rapidly with O2+. This gives a higher ionization efficiency and less fragmentation than electron impact; therefore, the spectra are simpler and signal is spread over fewer channels, which improves signal-to-noise. 600 spectra are averaged and saved to provide a time resolution of about one minute.

As the experiment proceeds and more and more products are formed, the mass spectra become very complex. Each chemical component will vary with a different time sequence, so the peaks from a given component will correlate with one another but not with peaks from other species. We will exploit this to separate components; the means of doing this is principal component analysis (PCA). Conventional PCA techniques have the effect of greatly compressing instrument dynamic range. In the course of our work on analyzing aerosol size distribution data, we have addresed this problem by developing a general scheme for weighted principal component analysis; this technique will be applied in this work. 

Updated on February 28th, 2013.