Date of Award
Doctor of Philosophy (PhD)
Michael P. Hannigan
Jana B. Milford
Shelley L. Miller
Daven K. Henze
Kelley C. Barsanti
The Denver Aerosol Sources and Health (DASH) study aims to identify and quantify the sources of PM2.5 that are related to negative health outcomes. The positive matrix factorization (PMF), a multivariate receptor model, was used as the primary tool for source apportionment of PM2.5 based on particulate speciation data. However, several questions need to be addressed on the receptor-based source apportionment of PM2.5.
In DASH study, 24-h PM2.5 samples were collected at one centrally located site in Denver. This raises the question of whether the heterogeneity in PM2.5 sources or source contributions across the urban area might lead to biased health effects estimation. In this work, PM2.5 samples were collected at four urban sites in Denver for one year. The carbonaceous speciation data were used as inputs for PMF analysis. The results showed that the four sampling sites have consistent source profiles and similar source distribution of elemental carbon (EC) and organic carbon (OC).
The speciation of PM2.5 in the DASH study includes inorganic ions, EC and OC, organic molecular markers (OMMs) and water soluble elements (WSEs). To evaluate the utility of different speciation data sets for source apportionment of bulk PM2.5 species, different combinations of source tracers with bulk PM2.5 species were applied for PMF analysis. The results suggested that OMMs were better source tracers for EC and OC than WSEs.
However, OMMs are mostly semi-volatile organic compounds (SVOCs), and their particle-phase fractions are impacted by gas/particle (G/P) partitioning. In this work, a 32-month series of PM2.5 speciation data was available for PMF analysis. The influence of G/P partitioning was identified by the comparison of PMF analysis of the full data set versus temperature-stratified sub-data sets. With the prediction of gas-phase SVOC concentrations by an equilibrium absorption model, the PMF analysis using total SVOC (gas + particle phase) data set showed consistent results between the full data set and temperature-stratified sub sets. A 1-year field study of both gas- and particle-phase SVOCs was conducted to verify the gas-phase SVOCs prediction. The observed G/P partitioning of SVOCs was reasonably consistent with that predicted by an equilibrium absorption model.
Xie, Mingjie, "Positive Matrix Factorization of PM2.5 -- Impacts of Spatial Variability, Speciation Data Set and Gas/Particle Partitioning of Semi-Volatile Organic Compounds" (2013). Mechanical Engineering Graduate Theses & Dissertations. 71.