With the reduction in PM emission standards for light duty vehicles to 3 mg/mi
for current Federal and California standards and subsequently to 1 mg/mi in 2025
for California, the required PM measurements are approaching the detection
limits of the gravimetric method. A “filter survey” was conducted with 11
laboratories, representing industry, agencies, research institutes, and academic
institutions to analyze the accuracy of the current gravimetric filter
measurement method under controlled conditions. The reference filter
variability, measured within a given day over periods as short as an hour,
ranged from 0.61 μg to 2 μg to 5.0 μg for the 5th, 50th, 95th percentiles
(n > 40,000 weights, 317 reference objects), with a
laboratory average of 2.5 μg. Reference filters were found to gain approximately
0.01 to 0.56 μg per day (50th percentile) and 0.5 to 1.8 μg per day (95th
percentile) with an average of 4.1 μg for the laboratories, which suggests a
gas-phase adsorption artifact because metal reference objects did not gain any
weight. Tunnel blank biases (n = 615) were much higher than the
reference filter bias and had a range from 1.1, 2.8, and 13.0 μg, for the 5th,
50th, and 95th percentiles, with an average of 4.1 μg. Robotically weighed
filters showed lower reference filter variability, but expectedly, there were no
significant advantages for weighing tunnel blanks. The higher tunnel blank
compared to the reference blank suggests that the sample collection system is a
relatively significant contamination source. The uncertainties associated with
filter weighing for tunnel blanks were generally less than the 5 μg tunnel blank
correction allowed under 40 CFR 1066.
