This effort serves to improve the protection of the public from airborne PM10-2.5 by aiding in the recognition of the hazard and in its prevention by improving exposure assessment. Recognizing that the spatial variability o f ambient coarse particulate matter, PM10-2.5, has limited researcher’s ability to properly and convincingly determine its health effects with current exposure assessment capabilities, the overall goal o f this study was to create a novel approach to PM 10-2.5 sampling that would capture this important characteristic.

To achieve this goal, a passive particulate sampler, already described for use indoors, was modified for ambient use and tested to establish performance criteria.  Modifications included optical microscopy for analysis and simpler strategies to determine the values of particle shape factors that alter calculated results. The method had a CV of 11.6% in the field study. The samplers also had a Pearson correlation coefficient of 0.97 with dichotomous samplers in the field and a ratio of passive sampler to the dichotomous samplers of 1.29. The limit of detection for a seven-day sample, based on the degree of contamination on the blanks was 1.7 pg/m . Since the passive sampler was to be protected from precipitation by a shelter, wind induced turbulent deposition as a source of error was a concern. An investigation of turbulent deposition was performed by testing various shelter designs in a wind tunnel. Shelters that caused turbulence near the airflow entry point also caused the particles, especially the larger ones, to deposit at a much higher rate and overestimate PM10-2.5 mass concentration.

A shelter, with the passive sampler set between two parallel flat plates, was chosen for implementation as it had little effect on deposition at various wind speeds.  The sheltered passive samplers were used in the field to characterize PM 10-2.5 variability across a Midwestern city. PM 10-2.5 concentrations in the city were found to he spatially non-uniform, CV’s > 20%, minimum R was 0.37 and the maximum coefficient of divergence was 0.36 among the 33 sites, with notable spatial trends further supporting the need for this highly spatially resolved sampling technique.