Energy Dispersive X-Ray Fluorescence Analysis of Air Particulates in Texas John R. Rhodes,' Andrzej H. Pradzynski, and Colin B. Hunter Columbia Scientific Industries, P.O. Box 6190, Austin, Tex. 78762 Jimmie S. Payne and James L. Lindgren Texas State Department of Health, Air Pollution Control Services, Austin, Tex. 78751
m Suspended particulate matter samples were collected in 38 Air Quality Control Network Stations in Texas using high-volume air samplers with 8 X 10 in. filter paper sheets, Whatman Type 41. Specimen discs 2 in. in diam were cut out from the sheets and measured in a n automatic, energy dispersive X-ray fluorescence spectrometer using the radioisotope sources Fe-55, Pu-238, and Cd-109 for X-ray excitation. The following elements were determined: Ca, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Hg, Pb, As, Br, Sr, Zr, and Mo. Standards made up from calibrated solutions deposited on filter paper and dried were measured along with the specimens and blanks. Analytical results were compared with those obtained by atomic absorption spectrophotometry. Air particulate pollution data from a statewide survey in Texas were compared with other available data. The method and equipment has proved itself capable of use for air particulate survey measurements and for pollution source location.
E
nvironmental specialists are concerned not only with the total amount of particulates in the air but also with their chemical composition. Multielement analysis is necessary for revealing and identifying air particulate pollution sources. For routine, multielement analysis of large numbers of air particulate samples the analytical technique must be rapid and inexpensive. If such a technique were available, surveillance could be established over wide areas where air pollution is a problem. Conventional analytical techniques such as gravimetry, colorimetry, emission spectroscopy, or atomic absorption spectrophotometry can generally provide the required sensitivity for the detection of metallic element air pollutants, but they are time- and labor-consuming. They require a n experienced analyst to perform the determinations even when done routinely. They do not allow preservation of the sample for repeat analyses, since it must be dissolved. Furthermore, the necessary wet chemical sample preparation is an additional source of errors. Emission spectroscopy (Morrow and Brief, 1971) and atomic absorption spectrophotometry (Thompson et al., 1970) are at present the most widely used methods in air pollution control laboratories. Emission spectroscopy is a multielement technique but requires careful sample pretreatment and is not considered to be very precise. Also, preparation of standards requires considerable care and effort. Atomic absorption spectrophotometry is strictly a singleelement analysis technique, but modern instruments have interchangeable lamps for sequential analysis of up to six
To whom correspondence should be addressed. 922 Environmental Science & Technology
elements per sample. A standardized 8 X 10 in. air particulate filter may not be large enough for sequential analyses of many elements, especially when the more accurate method of standard addition is used. The sensitivities of both of these methods vary significantly from element to element. In fact, emission spectroscopy has insufficient sensitivity for such important potential pollutants as Se, Hg, As, Cd, and atomic absorption for As, Hg, Se. Several papers have been published on the application of neutron activation for the analysis of air particulates-e.g., Dams et al., 1971 ; Brar et al., 1970; Zoller and Gordon, 1970. The method is very sensitive for some elements while quite insensitive for others-e.g., lead. To cover the whole range of elements of interest to air pollution control, access to a nuclear reactor is necessary and a long time between sample irradiation and measurement must be allowed [from several hours up to a few weeks for some elements, according to Dams et al. (1971)l. Neutron activation analysis can be used in special cases but is rather impractical for routine application. X-ray fluorescence spectroscopy has been used for air particulate analysis by a few workers (Hirt et al., 1956; Bumsted, 1964). However, the conventional method, using a crystal spectrometer and sequential counting of each element has not found broad application in routine air particulate analysis. One reason could be the high capital cost and sophistication of the instrumentation. The automatic, energy dispersive system with a Si(Li) detector and radioisotope sources, used in this work, has all the features required for routine, multielement air particulate analysis of large numbers of samples. No sample preparation is necessary, beyond cutting a 2-in.-diam piece out of the filter. The measurements are simultaneous for all the characteristic X-rays excited, and the method is nondestructive. Once loaded with samples, the apparatus can operate unattended. Experimental Sampling Procedure. The particulate samples obtained for this study were collected on 8 X 10 in. Whatman 41 filter paper. This was found to be essentially free of metallic elements whereas the standard glass fiber filters contain appreciable quantities of zinc, iron, barium, potassium, calcium, and strontium, rendering them quite unsuitable for the direct analysis methods used here. To eliminate errors due to hygroscopicity of the cellulose filters, they were stored for 24 hr in a constanthumidity box before each weighing, and weighed a fixed time (7 min) after removal from the box. Two sets of samples were collected, the handling procedure being the same in each case. The filters were numbered, stored in the humidity box, weighed, and mailed to the sampling stations. The samples were collected on the specified day and the filters mailed back, stored in the constant-humidity box, reweighed and analyzed for 17 elements using the automatic energy dispersive X-ray spectrometer.
For comparison, 12 of the samples were also analyzed for six elements each by atomic absorption spectrophotometry. The first set of 38 samples was collected on the same day in June 1971 at the stations indicated on the map of Texas in Figure 1. The second set of 51 samples was collected at one city, Corpus Christi, in three batches of 17 on three specified days in July 1971. X-Ray Fluorescence Analysis. The analytical system used in this work is shown schematically in Figure 2 and consists of the following main components: automatic sample and source changer with capacities of 30 samples and four sources, respectively; three annular radioisotope source assemblies : Fe-55, 120 mCi: Pu-238, 400 mCi; and Cd-109, 12 mCi (Amersham Searle, Models I E C - K - ~ PPC-5, ~~, and CUC-3, respectively); one 80 mm2 X 4 mm thick Si(Li) detector and FET preamplifier with pulsed optical feedback, cooled by liquid nitrogen and having a resolution of 180 eV (FWHM) at 5.9 keV (Kevex Corp., Model 3000P) ; electronic circuitry detailed below; and a minicomputer (Nova with software supplied by Applied Computer Systems Division of Columbia Scientific Industries), with teletype terminal. The electronic circuitry consisted of the following modules : high-voltage supply and amplifier with baseline restorer and pile-up rejector (Kevex Corp., Model 4000); spectrum stabilizer (Canberra Industries, Model 1520), set on a reference peak located at the high-energy end of the spectrum range; scaler with preset count (Mechtronics Nuclear, Model 701); analog to digital converter and 1024-channel analyzer with a scope display (Northern Scientific, Model NS-636). The baseline restorer with pile-up rejector is needed to preserve the spectrometer resolution at count rates above 1000 counts/sec (cps). However, the resultant high dead time causes the percentage of lost counts to vary significantly with the total detected count rate. These losses are not corrected for by the live time meter of the multichannel analyzer. T o provide accurate timing at high count rates, the peak from the reference source is monitored by the scaler. The accumulation of the X-ray spectrum in the multichannel analyzer is stopped by the scaler after a preset count is obtained in the reference peak. The radioisotope used in the reference source is Am-241 which has a half-life of 450 years.
METHOD.Specimens in the form of discs of 47 mm diam were cut out of the original filter papers and measured in batches of 27 unknowns and three standards. T o obtain the best excitation efficiencies, the 17 elements to be determined were divided into three groups, each excited with a different source, namely: Ca, Ti, V with Fe-55; Cr, Mn, Fe, Co, Ni, Cu, Zn with Pu-238; and Hg, Pb, As, Br, Sr, Zr, Mo with Cd-109. Each batch of specimens was counted with each of the three sources. The counting period was that required to accumulate 200,000 counts in the reference peak, which amounted to about 10 min/specimen with Pu-238 and Cd-109, and 5 min with Fe-55. A typical spectrum of an air particulate sample is shown in Figure 3. The computer program was designed for automatic operation of the sample and source changer and the spectroxeter system. Groups of channels were preset for Ka and Ks or La and L, peaks of each element to be determined and for backgrounds between peaks. The data-processing part of the program was designed to integrate peak and background areas, subtract backgrounds from total peaks, subtract superimposed a- or P-peaks of adjacent elements using KJK0 and La,ILp ratios predetermined from multielement standards, and calculate element concentrations using calibration factors obSIGNALS FROM SAMPLE AND R E F S O U R C E
COOLED DETECTOR AN0 PREAMP
AMPLIFIER, PILE UP REJECTOR AN0 BASE L I N E RESTORER
~
H.V. BIAS SUPPLY
STABILIZER
REF. PEAK SCALER
-
MULTl CHANNEL ANALYZER AN0
..
SCOPE DISPLAY
Figure 2. Schematic of radioisotope X-ray fluorescence analytical system
EXCiTATION SOURCE Cd- 109 COUNTlkG T I M E : 10 MINUTES AIR VOLUME' 1590 M' n L T E R PAPER W H A T M L N 41 TOTAL DEPOSIT 401 p g / d
40
80
120
160
200
240
CHANNEL
Figure 1. Locations of sampling stations in statewide survey
280
320
BO
400
440
480
NUMBER
Figure 3. X-ray fluorescence spectrum of air particulate sample from west Texas Volume 6 , Number 10, October 1972 923
~
~~
Location Amarillo Beaumont Clute Corpus Christi El Paso Fort Worth Harlingen Lubbock Dallas Houston Matagorda San Antonio
~~
Ca 2.7 18.3 29.9 21.9 13.7 17.7 11.8 8.5 4.8 6.6 0.8 3.5
~
Ti 0.34 0 10 0.22 0.29 0 34 0.20 0 20 1.33 0.08 0.11 0.04 0.12
V
