Background

The presence of toxic pollutants in the air has been a subject of research for many years in the United States and countries around the world. In the United States, air quality standards are governed by the “Clean Air Act” and administered by the US Environmental Protection Agency (EPA). One of the key areas of concern for the US EPA is the Suspended Particulate Matter content (SPM) of air. Historically, the measurement of SPM in air was concentrated on total suspended particulates with no preference to size selection. However, more recent research on the health effects of SPM in ambient air has focused increasingly on particles that can be inhaled into the respiratory system, i.e. particles of aerodynamic diameter of < 10 μm. These particles are referred to as PM10 (2.5 – 10 μm) and PM2.5 (< 2.5 μm). Not withstanding chemical toxicity, it is now generally recognized that these particles are a significant threat to health.
Determining the Composition of Particulates in Air Pollution

The measurement of the elemental composition of the particulate matter is a key factor in understanding the long-term health effects of pollution. Suspended particulate matter is typically pre-concentrated using high volume air samplers and collected on Teflon filters. The chemical analysis of the SPM on these air filters is traditionally performed by energy-dispersive XRF (EDXRF) using EPA method IO-3.3. EPA method IO-3.3 outlines the protocol for the analysis of 44 elements on Teflon air filters, but significant advances in the development of EDXRF instrumentation and software have occurred since this method was published. This application study demonstrates the performance of the Epsilon 5 EDXRF analyzer according to the EPA method IO-3.3, with the elemental range extended from 44 to 55 elements.
Measurement Criteria and Calibration

The air filters application used in this study was set up according to EPA method IO-3.3.The analytical measurement parameters were optimized to accommodate the technological improvements incorporated into the Epsilon 5.

The method was set up and calibrated with 59 commercially available air filter standards and a blank sample from Micromatter Co. (Eastsound, WA). The standards were composed of pure elements and compounds deposited on 40mm Nucleopore media.

The calibration was established using a single standard and a blank for each element. A Fundamental Parameter (FP) method was used to correct for the difference in sample loading when analyzing unknowns.

The measurement parameters used for this application are shown in table 1. The measurement time per condition was 100 seconds, except for the CaF2 target 600 seconds. The measurement time for each condition can be optimized according to specific needs.

Performance

The Epsilon 5 software features a very powerful deconvolution algorithm, which analyzes the sample spectrum and determines the net intensities of element peaks, even when they overlap one another. The accuracy with which this is carried out is essential to trace element analysis. Figure 1 shows a fitted spectrum of air filter standard NIST 2783 obtained with the Ge secondary target. The extremely low background is a consequence of the polarizing optical path.

Precision

The total method precision is a combination of instrument precision and stability of the sample during the measurement. The method precision can be reported for both short (repeatability) and long term (reproducibility) measurements. The repeatability of the Epsilon 5 was assessed by measuring a single filter sample (NIST 2783) 20 consecutive times in a single day. The reproducibility was determined by measuring the same sample once per day over a 10-day period.

The repeatability and reproducibility data for a selection of elements are shown in table 2. No drift correction was applied during the precision studies. The repeatability and reproducibility are both excellent and for most elements the short and long term precision are nearly identical. Comparison of the relative RMS values with the counting statistical error (theoretically, the minimum possible error) shows the excellent precision of the instrument and the non-destructive nature of the method for analyzing filter samples. Figure 2 gives a graphical representation of the short and long-term stability of Cr and Cu.