Exploring aerosol-specific calibration and performance of three direct-reading photometers.

Abstract

Laser photometers provide real-time data on airborne aerosols. They are a valuable tool for assessing task exposures, as well as process and environmental changes. However, their performance compared to the validated National Institute of Occupational Health and Safety (NIOSH) method Particulates Not Otherwise Regulated, Respirable 0600 gravitational method is uncertain. NIOSH has established a criterion for sampling and analytical methods to be within 25% of the 'true' concentration. Manufacturers and research scientists cite the importance of using an aerosol-specific calibration factor to improve instrument correlation with the gravimetric method. Field data from three photometers are presented to illustrate instrument performance variability and evaluate single and averaged aerosol-specific calibration factors. Respirable particulate and respirable crystalline silica (RCS) were simultaneously measured ten times in an operating rock crushing facility using the NIOSH methods 0600 and 7500 Silica, Crystalline, by XRD (filter redeposition) and three factory calibrated photometers. Ten aerosol-specific calibration factors were calculated for each photometer and used to determine single and averaged aerosol-specific calibration factors. Single and averaged aerosol-specific calibration factors were mathematically applied to "correct" the factory calibrated instrument measurements. Performance was evaluated using absolute relative error. With the factory calibration, the average absolute relative error for each instrument exceeded 25%. A single-event aerosol-specific calibration factor reduced the average absolute relative error for all instruments, bringing it below 25% for one of the three photometers. A 3-run average aerosol-specific calibration factor reduced the average absolute relative error below 25% for all instruments. Further averaging of calibration factor provided no significant advantage. The 95th percentile of absolute error fell below 25% for one of the tested instruments when applying both a single and averaged calibration factor but remained above 25% for the other two instruments. Field testing of the single-run, three-run average and ten-run average calibration factors revealed that the absolute relative error exceeded 25% in at least one of the three CF-field tests for each instrument. The average absolute relative error in estimates of RCS varied from 7 to 38%.