Source apportionment by receptor modelling is instrumental in devising optimized strategies to mitigate fine particulate matter (PM_2.5) pollution. Hopanes and levoglucosan are crucial source markers for vehicular exhaust (VE) and biomass burning (BB) and commonly used in receptor models. Recent studies have shown evidence for their atmospheric instability, which would cause bias in source apportionment if their degradation is not corrected. This work focuses on devising a method to incorporate the degradation of these two markers into source apportionment of PM_2.5 using chemical mass balance (CMB) model. Chemical speciation data from 92 PM_2.5 filter samples collected in the Pearl River Delta region, China (Guangzhou and Tsuen Wan site) were analyzed. Evidence for atmospheric degradation...[ Read more ]

Source apportionment by receptor modelling is instrumental in devising optimized strategies to mitigate fine particulate matter (PM_2.5) pollution. Hopanes and levoglucosan are crucial source markers for vehicular exhaust (VE) and biomass burning (BB) and commonly used in receptor models. Recent studies have shown evidence for their atmospheric instability, which would cause bias in source apportionment if their degradation is not corrected. This work focuses on devising a method to incorporate the degradation of these two markers into source apportionment of PM_2.5 using chemical mass balance (CMB) model. Chemical speciation data from 92 PM_2.5 filter samples collected in the Pearl River Delta region, China (Guangzhou and Tsuen Wan site) were analyzed. Evidence for atmospheric degradation of the targeted markers was first presented through investigating seasonal variation of source markers ratio, ratio-ratio plots, and evaluating CMB results. A novel approach was developed for determining whether hopanes degradation exists. This approach involves plotting the ratio of hopanes abundance in ambient and source measurements, with both normalized by the least reactive hopane homologue, as a function of the carbon number of hopanes.

Atmospheric oxidation of the targeted markers considering first-order kinetics was then incorporated into CMB model using a best-fit solution approach. The results reveal that fine organic carbon associated with VE and BB increase from 1.05 ± 0.70 (average ± standard deviation) to 1.78 ± 1.01 μg/m³ (9.5 vs. 16.1% of measured mass) and from 1.60 ± 1.71 to 3.30 ± 2.81 μg/m³ (14.5 vs. 29.8%) after considering the degradation, respectively. The degradation was more significant in the warm months than in the cold months, a seasonal characteristic in accordance with photochemical reaction kinetics. As to PM_2.5, the contributions rise from 4.48 ± 2.63 to 7.70 ± 3.73 μg/m³ (9.2 vs. 14.7%) for VE, and from 3.20 ± 3.07 to 6.64 ± 5.10 μg/m³ (6.6 vs. 12.7%) for BB. This research clearly demonstrates the importance of considering molecular marker degradation when they are used as source markers in receptor modelling. The method developed in this thesis could be used to improve PM_2.5 source apportionment in other locations with significant vehicular and biomass burning emissions.