Article pubs.acs.org/est
Interlab Comparison of Elemental Analysis for Low Ambient Urban PM2.5 Levels Choong-Min Kang,* Souzana Achilleos, Joy Lawrence, Jack M. Wolfson, and Petros Koutrakis Exposure, Epidemiology, and Risk Program, Department of Environmental Health, Harvard School of Public Health, Boston, Massachusetts 02215, United States S Supporting Information *
ABSTRACT: There is growing concern about the accuracy of trace elemental analysis of ambient particulate matter (PM) samples. This has become important because ambient PM concentrations have decreased over the years, and the lower filter loadings result in difficulties in accurate analysis. The performance of energy-dispersive X-ray reflectance spectrometry was evaluated at Harvard School of Public Health using several methodologies, including intercomparison between two other laboratories. In reanalysis of standard films as unknown samples following calibration, the HSPH ED XRF measurements represented good performance: 2% errors in precision and 4% errors in accuracy. Replicate analysis of ambient air filters with low PM2.5 levels indicated that S, K, Fe, and Ca showed excellent reproducibility, most other quantifiable elements were below 15% error, and the elements with larger percent of flagged measurements had less in precision. Results from the interlaboratory comparison demonstrated that most quantifiable elements, except Na and Al, were quite comparable for the three laboratories. Na performance could be validated from the stoichiometry of Na to Cl of indoor PM2.5 filter samples.
1. INTRODUCTION A number of epidemiological studies have reported significant associations between increases in levels of ambient particles and excess daily morbidity and mortality from respiratory and cardiovascular causes.1−3 The associations have been observed even when the concentrations are below the U.S. National Ambient Air Quality Standards (NAAQS), which might imply that chemical characteristics of particles should also be considered as important factors in assessing health effects. Elemental composition is typically one of the important components of PM2.5 (i.e., particles less than 2.5 μm in diameter), together with carbonaceous species, that may play a key role in adverse health effects of exposure because some elements present in ambient particles are understood to be toxic.4−6 For this reason, elemental composition has widely been used to identify the origins of ambient particles,7,8 and assess environmental exposure and health effects in diverse studies.9,10 A variety of analytical methods have been used to measure elemental composition of particles on filter media, and have also been evaluated through intermethod comparison studies.11,12 Energy dispersive X-ray fluorescence (EDXRF) spectroscopy is a nondestructive analysis, does not require sample pretreatments such as acid digestion, and is capable of analyzing multiple elements simultaneously. No additional sample preparation is needed, making it possible to do relatively rapid analyses using automated sampler techniques. And, because the method is nondestructive, it is possible to have © 2014 American Chemical Society
repeated measurements, both within the same laboratory and between laboratories, of the same filter samples. Because of these strengths EDXRF has been used widely in various studies. The U.S. Environmental Protection Agency (EPA) has also applied EDXRF multielement analysis of aerosols in a variety of source apportionment and modeling studies.13 However, as more elements are included in XRF analysis, the uncertainties of individual element concentrations increase due to the increased number of overlapping spectra.14 Recent studies have examined the stability, the sources of uncertainty, and the matrix effects of collocated samples in EDXRF analysis for the U.S. EPA monitoring network data set.15−17 In addition to these uncertainties, ambient PM2.5 levels have indeed decreased over the years,18,19 which results in lower particle loadings on sampled filters making accurate analysis more difficult. Furthermore, a variety of EDXRF spectrometers are commercially available, which differ with respect to radiation sources, excitation conditions/energies, source/detector geometry, and instrumental configurations. This can result in different determinations for the same sample. Due to the variability in EDXRF analytical quality, it is necessary to determine the acceptable precision and accuracy of measurements using typical ambient PM2.5 samples, including Received: Revised: Accepted: Published: 12150
June 19, 2014 August 7, 2014 September 24, 2014 September 24, 2014 dx.doi.org/10.1021/es502989j | Environ. Sci. Technol. 2014, 48, 12150−12156
Environmental Science & Technology
Article
Figure 1. Histogram and boxplot of PM2.5 filter concentrations used for this comparison.
errors could result in distorted conclusion; (2) the sample concentrations of the entire data set were grouped into ranges in increments of 5 μg/m3; and (3) the filters were chosen randomly but proportionally with respect to the range fractions obtained from the entire data set. As a consequence of the selection criteria, the average PM2.5 concentration of 50 filters was 11.9 μg/m3, ranging from 3.2 to 40.4 μg/m3, while the average of the entire set of 2323 filters collected during 2003− 2009 was 9.3 μg/m3, ranging from 0.3 to 52.7 μg/m3. 2.2. XRF Analysis and Instrumentation. Energy dispersive X-ray fluorescence (EDXRF) spectrometry is a technique widely applied to the elemental analysis of ambient particles collected on filter media. EPA method IO-3.3 specifies the protocol and the calibration procedure for analyzing elements on Teflon filters.13 Briefly, spectra are acquired for each sample. Elemental intensities in the sample spectra are determined by spectral deconvolution, with a least-squares algorithm. The least-squares algorithm synthesizes the spectrum of the sample under analysis by taking a linear combination of elemental shape spectra along with the background shape spectrum. The coefficients of the linear combination of elemental shapes and background spectra are scaling factors, which are determined by minimizing chi-square to produce the best fit to the sample spectrum. Elemental analysis at HSPH was conducted using an Epsilon 5 EDXRF spectrometer (PANalytical, The Netherlands) which utilizes secondary excitation from 10 secondary selectable targets. The spectrometer employs a 600-W dual (scandium/ tungsten, Sc/W) anode X-ray tube, a 100-kV generator, and a solid state germanium (Ge) detector. A total of 49 MicroMatter XRF calibration standard polycarbonate films (Micromatter Co., Vancouver, Canada) were used for calibration of 48 elements ranging in atomic number from 11 (Na) to 82 (Pb). We also used the U.S. National Institute of Standards and Technology (NIST) standard reference material (SRM) 2783, which is to simulate ambient PM2.5 on filter media, for quality control of the analytical procedure.
an assessment of interlaboratory comparison results. Nevertheless, to our knowledge there are very few, if any, studies of the interlab comparability for EDXRF analysis between commercially available laboratories. This paper thus addresses the quality controls and validation of EDXRD analysis for ambient filter samples with, in particular, low PM2.5 loadings at Harvard School of Public Health (HSPH) using several methodologies. Interlaboratory comparison was conducted with two other laboratories that have been frequently available for EDXRF analysis of aerosols on filter media in the United States.
2. MATERIALS AND METHODS 2.1. Measurement and Selection of Ambient PM2.5 Filters. Ambient PM2.5 filters used for the intercomparison were collected at the Harvard supersite in downtown Boston. The supersite is located on the roof of the Countway Library at the Harvard Medical School. For daily 24-h PM2.5 filter measurements, ambient air is drawn through a Harvard Impactor (HI) sampler20 with a 2.5-μm cut point at a flow rate of 10 L/min, and PM2.5 is collected on a 37-mm Teflon filter. The filters are pre- and postweighed using an electronic microbalance (Mettler MT-5, Columbus, OH) at controlled temperature and relative humidity to obtain PM2.5 mass concentration, and then analyzed for elemental composition using EDXRF spectrometers. A total of 50 PM2.5 filters were used for interlab comparison by different XRF methods. A limited number of filters were used to control analytical cost. The filters were chosen randomly from archived filters collected daily at the Harvard supersite between 2003 and 2009. To choose relatively representative samples (1) the samples with very low concentrations
