| A direct thermal desorption gas chromatography/mass spectrometry (TD GC/MS) method for air particulate matter (PM) analysis of volatile and semivolatile organic compounds was investigated. This technique uses a specially designed microdesorption GC inlet utilizing an inductively heated ferromagnetic foil with a Curie point temperature suitable for desorption, which can accommodate microgram amounts of material deposited on a thin strip of quartz fiber filter. Liquid or solid samples can be rapidly desorbed within 10 s at 315°C, followed by 30–40 min of chromatography time. The results obtained by this technique were found to be statistically equivalent to those obtained by the conventional solvent extraction gas chromatography/mass spectrometry (SX GC/MS) method for analysis of aromatic and n alkane standards, single source soot particles, and PM 10 filter samples. Correlations between injecting an extract, desorbing an extract, and desorbing particles averaged R = 0.94, with a three way correlation averaging R = 0.97. High volume sampling conducted at 12 spatially distributed sites located along the US/Mexican border of the El Paso/Juarez metroplex supplied 24h PM 10 filters for an investigation combining thermal desorption with a rapid online chemical derivatization procedure, and multivariate methods of source attribution using principal component and canonical correlation analysis of the resultant chemical markers. Four major combustion related PM emission sources were revealed at these sites: automotive, waste burning, biomass burning and meat cooking. A second investigation conducted in the same area used mediumvolume sampling to collect 2 h timeresolved PM 10 receptor samples for TD GC/MS analysis. Additionally, 2 h samples for inorganic analysis, multichannel particle size distribution measurements, and meteorological data were collected enabling generation of circadian PM multicharacterization profiles. Factor analysis based receptor modeling using principal component analysis of the mixed characterization data resulted in the deconvolution of temporally overlapping PM events, trends and gradients. Results of the temporally resolved PM receptor sampling profiles confirmed the results of the spatially distributed PM receptor sampling in that the major sources were attributed to automotive traffic, biomass and waste combustion. However, in the time resolved data, urban dust events—in particular a large evening PM peak—were seen to play a more prominent role. |
