EPA-600/2-76-033
March 1976
X-RAY FLUORESCENCE MULTISPECTROMETER
FOR RAPID ELEMENTAL ANALYSIS
OF PARTICULATE POLLUTANTS
by
Jack Wagman, Roy L. Bennett, and Kenneth T. Knapp
Emissions Measurement and Characterization Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
LIBR/
u. s. Enviir. .' .•
EDISON, N. J. Uo.il/
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences
Research Laboratory, U. S. Environmental Protection Agency, and
approved for publication. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use,
n
-------
CONTENTS
Page
ABSTRACT iv
LIST OF FIGURES v
LIST OF TABLES vi
ACKNOWLEDGMENTS vi i
INTRODUCTION 1
SUMMARY 3
CONCLUSIONS 4
GENERAL DESCRIPTION OF INSTRUMENT 5
Spectrometer Channels 5
X-ray Tubes and Power 7
Compact Electronic Channels 7
Vacuum Chamber 8
Automatic Sample Handling 8
Computer Operation and Data Processing 8
SPECTROMETERS 11
SAMPLE CONFIGURATION AND HANDLING 19
Filters and Sample Deposits 19
Sample Handling 19
Automatic Sample Loader and Retriever 20
CALIBRATION 22
INSTRUMENT OPERATION AND DATA PROCESSING 23
Instrument Control 23
Calibration, Measurement, and Data Processing 23
INSTRUMENT PERFORMANCE 25
Sensitivities and Detection Limits 25
Precision and Accuracy Estimates 26
Analyses of Source and Ambient Air Samples 29
REFERENCES 32
-------
ABSTRACT
A multichannel wavelength x-ray fluorescence spectrometer,
specially adapted for rapid analysis of air pollution particulate
samples, is described. The system contains an array of fixed-
wavelength spectrometers optimized for simultaneous analysis of 16
preselected elements and a computer-operated scanning channel for
the determination of any number of additional elements. A loading
device permits automatic handling of batches of up to 100 frame-
mounted 47 mm filter samples. Instrument operation, data pro-
cessing, and printout of results are controlled by a minicomputer.
The system permits rapid elemental analysis at high spectral
resolution, a significant advantage with air pollution samples
which typically contain several dozen elements at a wide range of
concentrations. For samples deposited on membrane filters,
detection limits for 100-second counting times are in the range of
2 to 40 ng/cm2 for most elements of interest.
-------
LIST OF FIGURES
s
Figure P_age
1 Simultaneous X-ray Multispectrometer System .... 6
2 Monochromators in Vacuum Chamber 9
3 Automatic Sample Loader 10
4 Double Mono chroma tor 12
5 Beam Paths in a Double Monochromator 12
6 Compact Scanning Spectrometer ........... 13
7 Scanning Spectrometer Shown Schematically at Three
Wavelength Settings 14
8 Sequence of Views of Scanner at Eight
Wavelength Settings 15
9 Scanner Slit-to-Crystal Distance as a Function
of X-ray Energy 16
10 Comparison of X-ray Spectra Obtained with
Wavelength and Energy Spectrometers on a
Calcium-Potassium Sample 18
11 Separation of K Lines of Mn, Fe, and Co
by the Scanning Channel 18
12 Exploded View of a Frame-Mounted Filter 21
13 Sample Cup - Top and Side Views 21
14 Comparison of X-ray Fluorescence and Gravimetric
Determinations of I^SQ^ Deposited on
Nuclepore Filters 28
-------
LIST OF TABLES
Table Page
1 Fixed Mono chroma tors in EPA Spectrometer 7
2 Resolutions Measured with Scanning Wavelength
and Energy-Dispersive Spectrometers 17
3 EPA Multispectrometer XRF Analyzer Element
Sensitivities and Detection Limits 25
4 Precision of Replicate Analyses of Potassium. ... 26
5 Magnitude and Uniformity of Background Count
for Three Types of Membrane Filters 27
6 Comparison of XRF and AAS Analyses on Particulate
Samples from a Simulated Combustion Source ... 29
7 Typical XRF Analyses of Particulate Emissions
from 1975 Catalyst-Equipped Cars 30
8 XRF Analysis of Flyash from Coal-Fired Power
Plant 30
9 XRF Analyses of Ambient Air Particulate Samples . . 31
-------
ACKNOWLEDGMENTS
The authors are pleased to acknowledge the valuable contributions
of L. S. Birks, J. V. Gilfrich, and J. W. Criss of the Naval Research
Laboratory, Washington, D. C., in the development of specifications for
this instrument and in optimization of its performance. We also thank
R. B. Kellogg, Northrop Services, Inc., Research Triangle Park, North
Carolina, for his excellent assistance in instrument operation and data
acquisition.
vn
-------
X-RAY FLUORESCENCE MULTISPECTROMETER
FOR RAPID ELEMENTAL ANALYSIS
OF PARTICULATE POLLUTANTS
INTRODUCTION
The chemical characterization of participate pollutants in the
environment has been greatly facilitated by the development in recent
years of a number of instrumental multielement analytical procedures.
The four techniques used most widely for elemental analysis of airborne
particulate samples are optical emission spectroscopy, atomic absorption
spectroscopy, neutron activation, and x-ray fluorescence. Each of these
methods has definite advantages and disadvantages, though none is appli-
cable to all elements.
The use of x-ray fluorescence is noticeably increasing at this
time because of a number of attributes that make it especially
attractive for the analysis of airborne particulate matter. These
features include (1) the direct analysis of filter deposits with no
need for sample preparation; (2) the non-destructiveness of the method,
which permits samples to be retained for further analysis or future refer-
ence; (3) the fairly uniform detectability across the periodic table,
with the ability to analyze all elements from atomic number 9 (F) upward;
and (4) the availability of commercial instruments that permit the
analysis of samples for a large number of elements in relatively short
time intervals and at low cost.
Competition exists not only between x-ray fluorescence and other
analytical methods but also within the x-ray fluorescence method itself,
e.g., between wavelength dispersion and energy dispersion. Birks and his
-------
coworkers at the Naval Research Laboratory carried out an EPA-supported
laboratory study1'2 to develop x-ray fluorescence as a method for routine
multielement analysis of filter-deposited particulate samples. The study
compared various x-ray fluorscence techniques, including the options
available for both excitation and detection of the fluorescence. Among
the main conclusions were the following:
1. The single-element limit of detection, where interferences
are absent or neglible, is about the same for either wavelength or
energy dispersion.
2. Atmospheric particulate samples and samples from power plants,
incinerators, and other source emissions typically contain many
elements at widely different concentrations. Energy-disperse x-ray
fluorescence spectra for such samples show significant interferences
between neighboring elements, particularly the elements from sulfur to
nickel in the periodic table; and require mathematical unfolding to
determine the x-ray intensities. For such real pollution samples, the
use of wavelength dispersion spectrometers with their high-resolution
capability is a distinct advantage; their use requires considerably
less data manipulation, thus avoiding what can be in many instances a
major source of error.
3. For routine analysis of large numbers of samples, in which
elements of interest can be specified in advance, the use of multi-
channel wavelength spectrometers appears as the most practical solution,
inasmuch as these instruments combine two important features, i.e., high
spectral resolving power and simultaneous measurement of a large number
of elemental concentrations.
An instrument of this kind, adapted for analysis of filter-
deposited samples of particulate matter, has been set up and is in
routine use at the EPA Environmental Research Center in North Carolina.
Its essential features and performance are described in this report.
-------
SUMMARY
A simultaneous x-ray spectrometer, containing an array of 16 fixed
monochromators and a sequential channel, has been specially adapted and
equipped by the EPA for automated analysis of filter-deposited samples
of particulate matter. It is being used successfully for the rapid and
routine analysis of large numbers of air pollution particulate samples
for 20 to 30 elements per sample.
The multichannel wavelength spectrometer is capable of carrying
out simultaneous analysis of many elements in a sample, as is the case
for energy dispersive units; but it has in addition two important
advantages. Its crystal spectrometers can yield good detection limits,
i.e., 30 nanograms per square centimeter or less for most elements of
interest, in relatively short count times as compared to enerqy-disoer-
sive systems that are dependent on severely count-rate-limited solid
state detectors. More important, the higher resolution of the crystal
spectrometers permits analyses of air pollution samples, which typically
contain several dozen elements at a wide range of concentrations, to be
obtained with considerably less data manipulation and therefore fewer
errors in correcting for interelement interferences.
The cost of analyses carried out with the simultaneous x-ray
spectrometer has been calculated based upon a 5-year amortization of
the capital cost, typical operating expenses, and sample burdens well
within the capacity of the instrument. Estimates range from 2 to 4
dollars per sample and as low as 10 cents per element.
-------
CONCLUSIONS
The performance of an x-ray fluorescence multispectrometer system
has shown it to be a highly effective means for rapid and routine multi-
elemental analysis of large numbers of air pollution participate samples.
Rapid determinations were accomplished by sirrul taneous analyses with one
scanning and sixteen fixed spectrometers and by the relatively short
counting times possible with wavelength dispersive spectrometers.
The higher resolution obtained with crystal spectrometers comoared to
energy-dispersive instruments reduces the interferences that occur between
elements having characteristic lines of nearly the same energy. This is
especially imoortant with environmental samples that typically contain
several dozen elements, including many requiring interference corrections
that are large arid uncertain in low-resolution instruments. The use of
crystal spectro;notv.'r> wit'! :her~ high-resolution capability thus r.im'tm'zcs,
and in most case: e" ,,mi nates , the need for mathemati cal unfc'idiny procedure
that can be a significant source of error in x-ray fluorescence analysis.
The measured sensitivities and detection limits are more than adequate
for most elements of environmental concern. For 30 elements of greatest
interest the minimum detection limits for a counting time of 100 seconds
were in the range 2 to 10 ng/crrr and exceeded 30 ng/cm^ for only four of
the elements.
The precision of the analysis depends on counting statistics, instru-
ment drift, specimen variation, and operational errors.. Excellent precision
in the total counting procedure, including specimen handling and positioning,
was found in repeatability tests. Evaluation of the uniformity of various
membrane filter substrates revealed that Nuclepore was the most uniform.
Comparison of the results of x-ray fluorescence analyses with the
results obtained on the same samples by gravimetric or atomic absorotion
analysis has verified the accuracy of the method. The analytical capability
of the instrument has been demonstrated with a large variety of sample types
from mobile sources that included catalyst-equipped cars, diesel engines,
and aircraft; from stationary sources that included incinerators, power
plants, and chemical process plants; and from ambient air sampled at various
locations.
-------
GENERAL DESCRIPTION OF INSTRUMENT
The high-resolution x-ray fluorescence tnultispectrometer currently
in use at the EPA Environmental Sciences Research Laboratory was set
up to provide a means for rapid and quantitative multielement analysis
in research projects to characterize airborne particulate samples and
to demonstrate its capability for processing large numbers of air
pollution samples for routine monitoring purposes. An overall view
of the instrument is shown in Figure 1. Based upon a newly designed
Siemens (Model MRS-3) simultaneous spectrometer* adapted for use with
filter-deposited samples, the instrument includes automatic sample
handling and computer-controlled operation and data processing
capabilities.
Spectrometer Channels - The EPA simultaneous spectrometer has been
equipped with 16 fixed monochromators and 1 scanning monochromator.
The fixed channels permit each element from fluorine upward in atomic
number to be determined by an optimally designed monochromator, including
an appropriate detector for the analyte x-ray energy, and an optimum
crystal. The crystals in the fixed channels are logarithmically bent in
order to achieve a constant Bragg angle over the entire surface without
grinding.3>4 A list of the elements selected for the fixed channels is
shown in Table 1 along with the analyte lines, crystals, and detectors
employed for each channel. This list includes the nine elements with
lowest atomic numbers that can be analyzed (i.e., F, Na, Mg, Al, Si, P,
S, Cl, K) plus seven additional elements (i.e., Cr, Mn, As, Br, Cd, Hg,
Pb).
The EPA instrument is also equipped with a fully focusing curved-
crystal^ scanning spectrometer which greatly increases the flexibility
of operation and the total number of elements which may be analyzed.
Scanning is accomplished by a stepping motor which may be operated
manually or automatically according to pre-defined scan plans which have
been inserted into the minicomputer program.
*Simultaneous x-ray multispectrometers are currently manufactured also
by Applied Research Laboratories, Philips Electronic Instruments, and
Rigaku Denki.
5
-------
Figure 1. Simultaneous x-ray multispectrometer system.
-------
Table 1. FIXED MONOCHROMATORS IN EPA SPECTROMETER
Electronic
channel
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Element
Chromium
Lead
Manganese
Arsenic
Mercury
Bromine
Phosphorus
Silicon
Cadmi urn
Aluminum
Sulfur
Sodium
Fluorine
Magnesium
Potassium
Chlorine
Line
K
a
L6
K
a
K3
a
K
a
K
a
K
a
L
a
K
a
K
a
K
a
K
a
K
a
K
a
K
a
Detector
type
Scintillation
Scintillation
Scintillation
Scintillation
Scintillation
Scintillation
Flow
Flow
Flow
Flow
Flow
Flow
Flow
Flow
Flow
Flow
Crystal
LiF(200)
LiF(200)
LiF(200)
LiF(200)
LiF(200)
LiF(200)
PET
PET
PET
PET
PET
KAP
KAP
ADP
PET
PET
Window
thickness ,
yni
--
--
--
--
--
--
2
2
6
2
2
0.4
0.4
0.4
6
6
Spectrometer
position
1
1
2
3
5
9
10
2
3
4
5
7
4
8
8
9
X-ray Tubes and Power - Fluorescence excitation in the EPA spec-
trometer is accomplished by either of two interchangeable Siemens type
A661 water-cooled x-ray tubes containing chromium and rhodium targets,
respectively. The high-voltage generator for the x-ray tube, which is
located in the spectrometer cabinet at the opposite end from the sample
changer (Figure 1), has a maximum output of 4 kW. It provides for the
adjusting, reading, and stabilization of the tube voltage and current.
Analyses are generally carried out using the Cr target tube, which has
an output of 2600 watts at a maximum of 60 kV. When the Cr target tube
is used, analyses of samples for Cr are done with automatic interposi-
tion of an aluminum filter in the primary beam.
Compact Electronic Channels - The detector amplifier signals from
each of the 17 monochromators in the EPA simultaneous spectrometer are
processed by a separate compact electronic NIM module. Each of the 17
compact electronic channels consists of a linear amplifier, a discrimi-
nator, and a pulse counter. There are two types of electronic modules-
7
-------
one for scintillation detectors and the other for proportional flow
detectors. The input pulses are processed in four three-stage amplifier
sections with adjustable gain. The pulse-height discriminator allows
only those pulses that lie between the upper and lower voltages of the
"channel width" to arrive at the output. These pulses are counted by an
eight-decade capacity counter during a given, preset time interval.
Vacuum Chamber - The characteristic x-rays of light elements are sig-
nificantly reduced by absorption in air so that vacuum operation is used
to analyze the lower atomic number elements. The x-ray tube, specimen
holder, and array of 17 spectrometers are enclosed in a spectrometer tank
(Figure 2). A vacuum control system provides a fully automatic sequence
once the start button is pushed. The sample is evacuated in a forechamber;
then inserted into the evacuated spectrometer tank, counted, removed from
the spectrometer tank, and returned to atmospheric pressure as the vacuum
is released.
The entire spectrometer tank is thermally insulated and held at a con-
stant temperature (25°C) by water circulating through cooling tubes from a
thermostat. Constant temperature is necessary since the counter tubes and
the analyzer crystal lattice parameters are temperature dependent.
Automatic Sample Handling - A custom-designed automatic sample
loader (Figure 3) has been fabricated to feed filter samples to the
inlet port of the MRS-3 spectrometer and remove the samples after they
have been analyzed. The sample loader holds up to 100 filter samples
mounted in EPA-designed plastic frames. The entire sample handling
process can be controlled by the programmable minicomputer so that the
samples can be analyzed unattended once they have been loaded into the
sample tray. Fail-safe features have been incorporated into the
changer to prevent jamming.
Computer Operation and Data Processing - A Digital Equipment
PDP-11/05 minicomputer with a 20K-word memory is used to control the
entire spectrometer operation, including manipulation of the sample
charger; insertion and removal of samples; counting of samples by the
fixed-channel monochromators for programmed time intervals; programmed
operation of the scanning channel for selected elements; storing of
calibration and interference correction factors; and handling of outout
-------
data from all the fixed channels and the scanner. A DEC RT-11 real-
time magnetic tape system is used to expand the capacity of the computer.
Print-out of the results is made on a Silent 700 Texas Instrument thermal
printer.
Figure 2. Monochromators in vacuum chamber.
-------
Tigure 3. Automatic sample loader.
10
-------
SPECTROMETERS
The EPA multispectrometer accomplishes rapid high-resolution elemental
analysis of participate samples through the simultaneous use of 16 fixed
monochromators and one scanner. Each fixed channel is optimally designed
for the analysis of a particular element by the proper selection of the
analyte line, crystal, and detector. Table 1 lists the elements determined
by the fixed channels in the EPA instrument, as well as the analyte lines,
crystal, and detector used. Fixed channels for additional elements are
available and can be substituted for those listed.
The monochromators are located above the specimen and opposite the
x-ray tube within the vacuum-tight spectrometer housing. Arranged in a
semicircle about the specimen, the monochromators are mounted at ten posi-
tions. At seven of the positions, a double monochromator is used while
two single monochromators and the scanner occupy the remaining positions.
A double monochromator, shown in Figure 4, is essentially two monochromators
molded into one frame. As shown schematically in Figure 5 each monochromator
contains an inlet slit, a logarithmically bent crystal, and an outlet slit in a
focusing arrangement. Both the entrance and exit slits are adjustable in
direction and width. Fine adjustment of the reflecting angle (0) by rotation
of the crystal holder and adjustment of the 2e angle by the movement of the
exit slit is possible after the monochromators are installed, the vacuum
applied, and thermal equilibrium attained. This is accomplished by the
attachment of a remotely controlled device to two spindles which drive gears
that rotate the crystal holder and move the exit slits.
Scintillation counters are used for detection of characteristic high-
energy x-rays and flow proportional counters for detection of low-energy
radiation. The scintillation counters have 0.2-mm-thick beryllium windows,
and the flow counters have windows of different thicknesses, i.e., 6 ym or
2 ym aluminum-coated Mylar or 0.4 ym nitrocellulose (on a support grid),
according to the element being measured. For most applications, a mixture
of 90 percent argon and 10 percent methane is used in the flow counter tubes.
The scanning channel (Figure 6) is a curved-crystal spectrometer
adjustable over a 2e range of 30° to 120°, and is compactly designed so that
it occupies only one spectrometer position. A fully focusing spectrometer,
11
-------
Figure 4. Double monochromator.
1 Flow counter with preamplifier
2 Exit aperture
3 Analyzer crystal
4 Entrance aperture
5 Specimen
Figure 5. Beam paths in a double monochromator.
12
-------
Figure 6. Compact scanning spectrometer.
13
-------
the scanner is designed so that the inlet slit, crystal, and detector
slit are always on the Rowland circle as indicated in Figure 7. The
sequence of scanner photographs in Figure 8 shows that this is accomplished
by three guide arms of equal length having bearings so that they can rotate
with a common shaft at the Rowland circle center, which they connect to the
inlet slit, the crystal, and the detector slit, respectively. A guide
band ensures that the distances between the crystal and the two slits are
always eoual.
CRYSTAL
DETECTOR
SLIT
Figure 7. Scanning spectrometer shown schematically at three wavelength
settings.
Since elements of atomic numbers 9 through 19 (fluorine through potassium)
were included for measurement by fixed channels, the scanning spectrometer
was equipped with a LiF (200) crystal, thus making it effective for analysis
of all higher atomic number elements. Figure 9 shows the analytical range
14
-------
J
Figure 8. Sequence of views of scanner at eight wavelength settings
(the Rowland circle was sketched on the photographs to show its shift
with wavelength).
15
-------
KAP
PET LiF(200)
E
E
LLJ
O
CO
Q
_i
<
CO
cc
-------
Table 2. RESOLUTIONS MEASURED WITH SCANNING WAVELENGTH
AND ENERGY DISPERSIVE SPECTROMETERS
Element
line
Pb La
Zn Ka
Mn Ka
Ca Ka
29,
deg
33.93
41.78
62.97
113.09
Energy,
keV
10. 5b
8.63
5.90
3.69
Resolution, FWHM
LiF (200) crystal
spectrometer
A0, deg
0.26
0.22
0.20
0.19
AE, eV
154
87
33
6
Energy dispersive3
spectrometer
AE, eV
185
170
155
125
Princeton Gamma-Tech Model PGT-1000.
capability from two- to twentyfold greater than that of the energy-dispersive
solid state detector. Figure 10 compares the spectrum obtained for a CaCl2~
KC1 mixture by the scanning channel with that obtained on the same sample by
the energy-dispersive spectrometer. The K$ potassium line, which is easily
separated by the scanning channel, is evident in the energy-dispersive
spectrum only as a slight distortion in the right side of the calcium Ka
peak. The resolution that can be achieved by the scanner for the K lines of
Mn, Fe, and Co is shown in Figure 11.
17
-------
• • I "j I
~~ ENERGY ~ i- i^ 4~
. T : I
Figure 10. Comparison of x-ray spectra obtained with energy ana wave-
length spectrometers on a calcium-potassium sample.
1
FeKoc
6.398 Kev
CoK0
7.648 Kev
FeK0
7.057 Kev
I
MnK/3
6.489 Kev
I
Figure 11. Separation of K lines of Mn, Fe, and Co by the scanning channel
18 '
-------
SAMPLE CONFIGURATION AND HANDLING
Filters and Sample Deposits - Filters used to collect air pollu-
tion particulate samples for x-ray analysis should have the following
characteristics: (a) low mass to reduce scattering of incident radia-
tion, hence minimizing the background count; (b) low content of the
elements to be measured; and (c) minimal penetration of particles below
the surface. Also, filters composed of low atomic number elements are
desirable since these, in general , have lower attenuation coefficients
and therefore lower absorption effects when penetration of the aerosol
into the filter occurs. Filter materials which have been found most
suitable for x-ray fluorescence analysis are the thin membrane types,
e.g., Millipore, Fluoropore, and Nuclepore.
It is advantageous to use thin aerosol deposits for x-ray fluores-
cence analysis, inasmuch as matrix effects requiring corrections for
attenuation and enhancement are thereby greatly reduced, the degree to
which the matrix problem is eliminated depends on the element to be
analyzed. In general, thinner samples are required for the lower
atomic number elements. The criterion for a sufficiently thin sample as
suggested by Rhodes is:
m<
where m is mass deposit density in g/cm2 and p is the mean mass attenu-
ation coefficient in cm2/g. In a matrix of typical flyash composition,
analyses for sulfur (Ka) and iron (Ka) would require deposit densities
not exceeding about 80 and 1200 yg/cm2, respectively. However, fluorine,
which has a y value in the same matrix of about 9000 cm2/g, would require
a total deposit of approximately 10 yg/cm^ or less to meet the thin-film
criterion. When multielemental analyses are to be carried out, it is
obvious that compromises will have to be made in selecting optimal
deposit densities, and matrix correction factors will often be required
for the lighter elements.
Sample Handling - To facilitate the processing of sample deposits
in the multichannel analyzer, special two-piece plastic frame mounts
19
-------
were designed for 47-mm filters (Figure 12). The mounts consist of a
lower ring with a 50-mm outer diameter and 32-mm inner diameter, and an
expansible upper ring of slightly smaller outer diameter. The lower
ring has a lip around the outer edge which holds the filter in place and
retains the upper ring. The filter is placed in the lower ring with the
deposit side up; the upper ring is contracted with snap ring pliers
(which fit into two holes); it is then placed over the filter and allowed
to expand into the beveled edge of the lower ring.
When the sample is inserted into the specimen area of the spectrometer,
it is pressed against the underside of a gold-plated aperture plate that
has a 30-mm diameter opening. This corresponds to the maximum sample
deposit area exposed to primary x-ray excitation from the tube.
Frame-mounted filter samples are transported from the external sample
loader into the vacuum tank of the spectrometer while seated atop specially
designed sample cups, as shown in Figure 13. The sample cup serves as a
trap for the primary beam that penetrates the sample and is gold-plated to
reduce the quantity of backscattered x-rays..
Automatic Sample Loader and Retriever - A custom-designed sample
changing system has been fabricated for automatic loading of samples
mounted in the EPA filter frames. The sample changer, shown in Figure 3,
holds up to 100 specimens which slide sequentially down a chute and are
centered directly under one of three pick-up discs along the periphery of
a large circular table that turns in a programmed sequence. Each pick-up
disc has small vacuum tips of soft rubber which contact and hold the
plastic sample frame at three points near its outer edge. The table then
rotates 120° so that the sample is directly over a sample cup located at
one of four positions in a rotating carousel. The vacuum is released as
the sample is pressed into a recess at the top of the cup. The rotating
carousel transports the cup with sample into the forechamber, which is
then evacuated before the specimen is inserted into the spectrometer tank.
At the conclusion of the fluorescence measurements, the specimen is returned
to the forechamber and the vacuum released, the rotating carousel trans-
ports the sample and cup out of the instrument, where the sample is finally
removed by one of the pick-up discs in the automatic sample changer. The
automatic changer has a fail-safe feature that prevents jamming by terminating
the entire counting procedure if a sample fails to be retrieved.
The automatic changer may be operated manually but is normally controlled
by the minicomputer program.
20
-------
UPPER RING
FILTER
LOWER RING
Figure 12. Exploded view of a frame-mounted filter.
Figure 13. Sample cup - top and side views.
21
-------
CALIBRATION
Standards are required for calibration of the spectrometer for
each element. The sensitivities in terms of counts per second per
unit surface deposit concentration, i.e., cps per yg/cm2, are determined
and stored in the computer memory. As shown earlier, line interferences
are minimal with wavelength dispersion; however, in those cases where
they do occur, the interference factors are measured using pure standard
specimens and are stored in the computer memory.
Several types of thin deposited standards have been employed to
calibrate the spectrometer. These include:
1. Thin films formed from vacuum-evaporated elements and com-
pounds. These were prepared on Mylar substrate (3.8-Mm thickness) by
Micromatter Company at concentrations of about 50 yg/cm2 with an
estimated accuracy of 5 percent.
2. Aerosols generated, e.g., in a Collison atomizer, and col-
lected on membrane filters.
3. Filter deposits of materials from solutions after solvent
evaporation. This method is subject to non-uniform deposits caused by
migration of ions during evaporation unless precautions are taken.
4. Filter deposits of fine powdered materials from liquid
suspension.
22
-------
INSTRUMENT OPERATION AND DATA PROCESSING
A minicomputer, interfaced with the multichannel spectrometer,
provides automatic operation of the instrument including sample changer
operation, sample transport, and control of counting parameters (selection
of elements, counting time, etc.) for both fixed and sequential channels.
The computer also provides data processing after the calibration and
analyte specimens have been counted. The computer is a Digital Equipment
PDP-11/05 unit with 20K-word memory. It has been interfaced with a
Digital Equipment tape control and transport operated by a RT-11 real-
time software system which expands the memory capacity and permits addi-
tional data processing to be applied as needed.
Instrument Control - Operation of the fixed channels is controlled
by the selection of any one of a number of possible programs hardwired
on program cards. These permit the simultaneous analysis of up to 16
selected elements for any desired counting time interval. As counting
in the fixed channels begins, operation of the scanner also proceeds
automatically as a result of the selection of one of several scan plans
which have previously been defined and stored in the computer memory. A
given scan plan contains the elements to be determined on the sequential
channel and the counting time interval for each. The program also allows
the operator the option of selecting a maximum count (e.g., 10,000) for
each element to reduce the total count time when one or more elements are
present at high concentration. In this option, the x-ray intensity from
each element is measured automatically for either the designated time
interval or the maximum count, whichever occurs first.
Calibration, Measurement, and Data Processing - Before elemental
concentrations in particulate samples can be determined, a set of calibra-
tion stahuards must be inserted and processed in the instrument.
The computer-controlled calibration program includes measurement of
standards at the analyte lines and at other lines at which the material
in the standard may interfere. Sensitivity factors for all of the elements
and interelement line interference ratios are calculated and stored in the
computer memory.
23
-------
Analysis of a series of participate samples on a given filter sub-
strate is preceded by measurement of a corresponding filter blank for
automatic determination and storage of background count levels and
minimum detection limits for each of the elements selected for measure-
ment. Following measurement of each particulate sample, the computer
program subtracts background counts, makes line interference corrections,
2
and applies sensitivity factors to convert cps to yg/cm . The list of
elemental concentrations for each sample is stored on magnetic tape for
later reference, or is printed out at the data terminal, or both; those
element concentrations that are below detection limits are noted by
asterisks. The program also provides for the application of other
correction factors as needed, e.g., for variations in particulate
deposit area and particle size.
24
-------
INSTRUMENT PERFORMANCE
Sensitivities and Detection Limits - The sensitivity of the x-ray
spectrometer for a given element is defined as the rate of change in the
analyte-line intensity with a change in the analyte concentration.
The minimum detection limit for x-ray analysis is often defined as the
concentration of analyte corresponding to a net count equal to three
times the standard deviation, i.e., three times the square root of the
background count. The sensitivities and detection limits for 100-second
measurements on 30 elements determined in the EPA simultaneous spectro-
meter, using thin-film-on-Mylar calibration standards, are listed in
Table 3. It is noteworthy that the minimum detection limits for only
100 seconds of count time are in the range of 2 to 10 ng/cm^ for
Table 3. EPA MULTISPECTROMETER XRF ANALYZER ELEMENT SENSITIVITIES
AND DETECTION LIMITS
j Detection
Sensitivity, ' limit
; counts/ 100 sec (100 sec, 3a),
Element : yg/cm^ ng/cm^
F
Na
Mg
Al
Si
P
S
Cl
K
Ca
Ti
V
Cr
Mn
Fe
220 149
534 | 29
10280
8074
11614
13392
28013
25394
121286
87817
85635
18010
7484
17522
13300_ j
2 !
i
3
3
15
9
9
2
2
o
"T
19
14
18
i
i
Sensitivity,
counts/100 sec
Element . yg/cm2
Co
Ni
Cu
Zn
i As
Se
Br
Cd
Sn
Sb
Ba
Pt
Au
Hg
16540
Detection
limit
(100 sec, 3o),
ng/crrr
3
14504 ! 10
18880 43
21066 i 7
17125
22922
50340
17303
10
12
28
2
14800 2
31100
25000
4
7
6812 20
8498
5776
Pb 16583
91
90
30
25
-------
more than half cf the elements measured including most of the low
p
atomic number el everts. The detection limit exceeds 30 ng/cm for only
four of these elements.
Precision and Accuracy Estimates - The precision of x-ray fluorescence
analysis depends on a combination of factors, including statistical counting
error, instrument drift, specimen variation, and miscellaneous operational
errors. In order to evaluate the magnitude of these factors, the replicate
determinations listed in Table 4 were conducted. A large accumulated count
was used to minimize the relative standard deviation resulting from statistical
counting error. A single sample counted in place ten times showed a relative
standard deviation of only 0.05 percent, while the count for a sample introduced
into the instrument ten times had a relative standard deviation of 0.10 percent.
The latter demostrates the excellent repeatability of the total counting
procedure, which includes specimen handling and positioning.
Table 4. PRECISION OF REPLICATE ANALYSES OF POTASSIUM9
! . Relative standard
1 deviation
Standard deviation, (~ .x 100], %
(t, ,00),
Statistical counting error, /N~ 5.31 0.02
Observed variation of ten replicate
counts on single sample in place ; 1539 ,' 0.05
Observed variation of ten replicate
counts on a single sample in-
serted and removed ten times 2729 : 0.10
Average count, F: 2,820,092.
Another factor that must be considered in the overall precision of a
measurement is the uniformity of the filter substrate used to collect parti -
culate samples, since a blank filter count is used to correct for background.
Table 5 lists the average background counts and standard deviations for ten
LlanK specimens of each of tnree types of membrane filters frequently used.
Nuclepore filters exhibited the best overall uniformity.
26
-------
Table 5. MAGNITUDE
FOR THREE
AND UNIFORMITY OF BACKGROUND COUNT
TYPES OF MEMBRANE FILTERS
—
*
F
Na !
Mg
Si
P
S
Cl
K
Ca
Ti
V
Fe
Ni
Cu
Zn
Cd
Ba
_
Nuclepore
0.8y
(Mass =1.1
mg/cm2)
- ' - T ' ' —
fTa . Ob
1.32 0.09
0.22
0.06
0.64 | 0.08
66.22
17.77
48.54
86.49
64.48
4.43
2.91
7.95
44.28
2.10
553.46
12.14
1.87
4.44
0.62
0.82
3.18
2.67
1.17
0.56
0.80
4.21
0.36
9.03
0.81
0.08
2.35 0.35
i
Fluoropore Mi Hi pore
Type FA Type AA
(Mass = 2.7 (Mass = 5.0
mg/cm2) mg/cm2)
-— -—- -
N
_. . _.._ .._
549.78
0.15
0.75
1.25
17.63
40.21
54.44
63.49
8.85
7.67
16.10
50.41
3.51
1
567.36
26.31
2.12
5.00
a N a
_. _.— «t_— :m-^-^ z^^x-^**^— *
28.20 . 1.96
0.05 0.28
0.10
0.25
2.16
2.23
4.81
2.46
3.15
1.20
1.93
4.86
0.13
0.05
i
1.43
2.29
16.34
54.52
200.24
212.40
333.84
12.74
25.71
0.16
0.27
0.35
0.63
5.59
2.40
7.13
1.62
1.65 ,
62.51 1 4.79 !
| :
0.67
14.13
3.49
0.11
5.52
577.69
41.34
3.18
1.12 ,
21.61 •
6.82
0.17
i
0.57 8.27 I 0.94
3 N is
a is
the mean value in cps for ten blank specimens.
the standard deviation in background for ten blank specimens.
27
-------
An estimate of the accuracy of the x-ray analytical procedure, as
carried out with the simultaneous spectrometer, was determined in a
comparison with gravimetric values for the rrass of potassiurr sulfate
aerosol generated in a Collison atomizer and deposited on a series of
Nuclepore filters. The graph plotted in Figure 14 shows very good
agreement between the x-ray and gravimetric values. Another comparison,
280
240
O)
cc
O
LL
CO
CO
DC
X
CO
200
160
120
O
80
40
I
I
I
SLOPE: 1.038
INTERCEPT: -9.46
CORRELATION
COEFFICIENT: 0.993
80
120
160
200
240
280
TOTAL K2SO4 BY GRAVIMETRIC ANALYSIS, ^9
Figure 14. Comparison of x-ray fluorescence and gravimetric: determinations
of I<2S04 deposited on Nuclepore filters.
28
-------
this time with atomic absorption analyses, is shown in Table 6. The
samples consisted of particle emissions collected on high-purity quartz
filters from a controlled combustion source fueled with oil spiked
with known amounts of organometallic compounds. The calculated values
shown in the table for the five metals analyzed are based upon the
amount of spiked fuel consumed and the known fraction of total emissions
collected on the filter. The x-ray fluorescence concentrations are
in remarkably good agreement with the calculated values with the exception
of the cadmium value, which is about 10 percent low, probably because
of some attenuation of the relatively low-energy cadmium fluorescence
resulting from only partial penetration of collected aerosol into the
interstices of the quartz fiber filter.
Table 6. COMPARISON OF XRF AND AAS ANALYSES
OF PARTICULATE SAMPLES FROM A SIMULATED COMBUSTION SOURCE
i ' " _ — -. -
I Total collected, yg
Element
Pb
Mn
Co
Cd
V
i Calculated j Atomic j X-ray
i value3 absorption r fluorescence
i 202
i 182
i 182
183
; 216
; 228 i
195 j
: 193 !
1 187
159 i
212
183
183
166
230
Based on analysis of fuel oil spiked with organoraetallic compounds.
Analyses of Source and Ambient Air Samples - The EPA simultaneous
wavelength x-ray spectrometer is being used to determine the elemental
composition of large numbers of samples of particulate emissions
collected from mobile sources that include catalyst-equipped cars, diesel
engines, and aircraft; from stationary sources that include power plants,
incinerators, and chemical process plants; and from ambient air sampled at
various locations. Table 7 lists elemental concentrations found
in emissions from 1975 cars equipped with oxidation catalysts. The
results show sulfur to be the predominant element and are in good
agreement with independent measurements indicating that nearly all of
the particulate matter consists of droplets of 45 percent (w/w) sulfuric
acid in water. Table 8 shows typical analyses obtained from samples
29
-------
Table 7. TYPICAL XRF ANALYSES OF PARTICULATE EMISSIONS
FROM 1975 CATALYST-EQUIPPED CARS
Concentration, yc
Sample Sample
Element ; no. 4 no. 8
k *-=._. ,« ™-.™ - „. „.. «».-lw~^.«~_— ,»-™r^,-,-.*,l,^-r_._™__, _„ „„ J..-, — _r _w(_
Al ; 0.12
i
Si 0.40 0.02
1
P | 0.05
S
Ca
Ti
Fe
Ni
Zn
Ba
Mass loading
6.80 : 23.00
0.08 0.08
0.02 0.01
0.32 1.10
0.02 - ]
i
0.07 0.07 ;
0.01
46.40 143.60 i
I/cm2
Sample
no. 30
- — — -• -
0.05
0.11
-
40.50
0.04
0.01
1.40
0.03
0.10
0.02
270.90
Table 8. XRF ANALYSIS OF FLYASH FROM COAL-FIRED POWER PLANT
Element
F
Na
Mg
Al
Si
P
S
K
Ca
Ti
2T
Concentration, pg/cm '
Sample
no. X5
r--- , ,,r - - • - -
0.29
0.42
2.0
103.
55.
0.54
7.8
15.6
.
3.6
Sample
no. X10 :'
0.15 i
0.35 j
1.2 !
48. !
26. i
i
1.0
7.8 j
5.7 I
4.9 !
i
1.5 i
1 O"
(Concentration, pg/cm
Element
V
Cr
Fe
Ni
Zn
Br
Cd
Ba
Pb
' Sample
1 no. X5
0.55
: 23.
1 20.2
1 0.03
, 4.6
, 0.89
0.020
f
: 0.62
! 1>7
I
: Sample
> no. X10
0.25
15.
' 9.2
'
i 0.13
I 0.007
i 0.33
1 1'1
i
30
-------
of particulate emissions from a coal-fired power plant, and Table 9
lists element concentrations in urban air particulate samples
collected at different locations and times. These data illustrate the
broad range of elements and elemental concentrations present in air
pollution samples.
Table 9. XRF ANALYSIS OF AMBIENT AIR PARTICULATE SAMPLES
Element
F
Na
Mg
Al
Si
P
S
Cl
K
Ca
Ti
V
Fe
Zn
Br
Cd
Ba
Pb
C
Sample nos. 7-19
-
-
0.02
0.28
0.44
0.11
3.74
0.26
0.20
1.10
0.04
0.005
0.76
1.10
0.40
0.04
_
1.80
ioncentration,
Sample nos.
-
0.06
1.20
1.60
-
3.20
0.15
0.47
0.90
0.08
-
0.70
0.05
-
0.007
0.04
0.23
ug/cm2
7-24 Sample nos. 10-04
0.09
t 0.09
; 0.14
5.30
I 7.60
i
; 0.18
: 1.77
1 0.47
2.06
2.67
0.16
0.03
; 2.73
0.16
i 0.75
i 0.04
I 0.10
1
1 3.0
31
-------
REFERENCES
1. Birks, L.S., J.V. Gilfrich, and P.G. Burkhalter. Development
of X-ray Fluorescence Spectroscopy for Elemental Analysis of
Particulate Matter in Atmosphere and in Source Emissions.
U.S. Environmental Protection Agency, Office of Research and
Development. Publication No. EPA-R2-72-063. Research Triangle
Park, North Carolina. November 1972.
2. Birks, L.S. and J.V. Gilfrich. Development of X-ray Fluorescence
Spectroscopy for Elemental Analysis of Particulate Matter, Phase
II: Evaluation of Commercial Multiple Crystal Soectrometer
Instruments. U.S. Environmental Protection Agency, Office of
Research and Development. Publication No. EPA-650/2-73-006.
Research Triangle Park, North Carolina. June 1973.
3. Barraud, J. Monochromateur-focalisateur donnant ur, faisceau
d'ouverture notable. C.R. Acad. Science, Paris. 2]_4:795 (1942).
4. DeWolff, P.M. An adjustable curved crystal monochromator for
x-ray diffraction analysis. Appl. Sci. Res. BJ_:119 (1950).
5. Johansson, T. Ober ein Neuartiges, genau Fokussierendes
Rbntgenspektrometer. Zeitschrift fur Pnysik. 82_:507 (1933).
6. Rhodes, J.R., A. Pradzynski, R.D. Sieberg, and T. Furuta.
Application of a Si(Li) spectrometer to x-ray emission analysis
of thin specimens. In: Low-Energy X- and Gamma-Ray Sources
and Applications. C.A. Ziegler, ed. London and New York:
Gordon and Breach. 1971. p.317.
32
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1 REPORT NO.
EPA-600/2-76-033
3. RECIPIENT'S ACCESSION-NO.
4. TITLE ANDSUBTITLE
X-RAY FLUORESCENCE MULTISPECTROMETER FOR RAPID
ELEMENTAL ANALYSIS OF PARTICULATE POLLUTANTS
5. REPORT DATE
March 1976
&. PERFORMING ORGANIZATION CODE
7 AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO.
Jack Wagman, Roy L. Bennett, and Kenneth T. Knapp
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Environmental Sciences Research Laboratory
Office of Research and Development
U. S. Environmental Protection Agency
Research Triangle Park. N.C. 27711
10. PROGRAM ELEMENT NO.
1AA010
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory
Office of Research and Development
U. S. Environmental Protection Agency
Research Triangle. ParL, N.._CL _277_11_
13. TYPE OF REPORT AND PERIOD COVERED
In-house, 1 yr. ending 10/75
14. SPONSORING AGENCY CODE
EPA-ORD
15. SUPPLEMENTARY NOTES
16. ABSTRACT -~ "
A multichannel wavelength x-ray fluorescence spectrometer, specially adapted
for rapid analysis of air pollution particulate samples, is described. The system
contains an array of fixed-wavelength spectrometers optimized for simultaneous
analysis of 16 preselected elements and a computer-operated scanning channel for
the determination of any number of additional elements. A loading device permits
automatic handling of batches of up to 100 frame-mounted 47 mm filter samples.
Instrument operation, data processing, and printout of results are controlled by
a minicomputer. The system permits rapid elemental analysis at high spectral
resolution, a significant advantage with air pollution samples which typically
contain several dozen elements at a wide range of concentrations. For samples
deposited on membrane filters, 100-second detection limits are in the range of 2
to 40 ng/cm2 for most elements of interest.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTOFIS
*Chemical analysis
X-ray analysis
*X-ray fluorescence
X-ray spectrometer
Air pollution
*Aerosols
Particles
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
07D
14B
20F
13B
14G
3 DISTRIBUTION STATEMENT
19. SECURITY CLASS I Tins Report!
__ UNCLASSIFIED
21. NO. OF PAGES
44
RELEASE TO PUBLIC
20 SECURITY CLASS /This page,'
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
33
------- |