EPA-R2-72-063
October 1972
ENVIRONMENTAL PROTECTION TECHNOLOGY
Development of X-ray
Fluorescence Spectroscopy
for Elemental Analysis
of Particulate Matter
in the Atmosphere
and in Source Emissions
Office of Research and Monitoring
U.S. Environmental Protection Agency
Washington, D.C. 20460
-------
EPA-R2-72-063
Development of X-ray Fluorescence
Spectroscopy for Elemental Analysis
of Participate Matter
in the Atmosphere
and in Soured Emissions
by
L. S. Birks, J. V. Gilfrich,
and P. G. Burkhalter
Naval Research Laboratory
Washington, D.C. 20390
Interagency Agreement No. 690114
Project Officer: Dr. Jack Wagman
Division of Chemistry and Physics
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND MONITORING
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
November 1972
-------
This report has been reviewed by the Environmental Protection Agency
and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Agency,
nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
11
-------
CONTENTS
ABSTRACT 1
INTRODUCTION 2
STANDARDS 4
FILTER MATERIALS 5
EXPERIMENTAL APPROACH 7
Wavelength Dispersion 7
Energy Dispersion 9
RESULTS 10
Sensitivity 11
Detection Limit 13
Analysis of EPA Samples 16
CONCLUSIONS 22
Particle Size Effect . . 25
Data Handling for Routine Analysis 25
APPENDIX 1 A-l-1
APPENDIX 2 A-2-1
111
-------
Development of X-Ray Fluorescence Spectroscopy for Elemental
Analysis of Particulate Matter in the Atmosphere and in Source
Emissions
L. S. Birks, J. V. Gilfrich and P. G. Burkhalter
Naval Research Laboratory, Washington, D. C. 20390
ABSTRACT
The application of x-ray fluorescence to the analysis of air pollution
particulate samples was demonstrated to be a rapid and economical tech-
nique at concentrations encountered in practical situations. No sample
preparation is necessary for particulates deposited on filters, which can
be placed directly in the x-ray equipment. Because the specimens are
thin, matrix absorption and fluorescence are negligible and calibration
curves are linear. All of the elements of interest can be measured
simultaneously in 100 seconds with either multichannel x-ray crystal
spectrometers or multichannel analyzers with energy dispersion detectors.
Sensitivity and detectability were compared for four types of excitation
(x-ray tubes, fluorescers, radioisotopes and high-energy ions) and for the
two types of data acquisition (crystal spectrometers and energy dispersive
detectors). Typical detection limits for 100 seconds measurements varied
from 100 to 6000 ng/cm for isotope excitation to 10-250 ng/cm^ with
ordinary laboratory x-ray spectrometers and to 1-5 ng/cm^ with either
large production-type x-ray spectrometers or with high energy ion ex-
citation (ion excitation required special thin substrates and would not be
suited to large scale routine use).
1
-------
The x-ray analysis of real pollution samples showed concentrations
for elements of interest between 50 ng/cm and 300 jig/cm . Represented
among these samples were particulates filtered from auto exhaust, municipal
Incinerators, a power plant and a cement plant.
For routine analysis of large numbers of samples for many elements
the optimum technique requires the use of a multichannel crystal spec-
trometer instrument capable of measuring a minimum of 1Z to 14 elements
at the same time. With a minimum degree of automation such instrumentation
would be capable of analyzing 600 samples per 24 hour day with a minimum
detectable limit for elements above Na in the periodic table in the range of a
few to a few hundred nanograms per square centimeter. The addition of an
energy dispersion channel using a high resolution semicpnductor detector
would be desirable for a qualitative confirmation that the crystal spectrom-
eters are measuring the elements of prime concern in the sample.
INTRODUCTION
The elemental analysis of air pollution particulate samples from the
ambient air or from emission sources is a rather unique problem. The
total amount of material Is small, but the sample may contain a large
number of elements over a wide atomic-number range and at widely dif-
ferent concentrations. A viable technique must be capable of measuring
the elements of interest in an efficient and economical manner. It must
have good detectability because of the small amount of some elements
present in an air pollution sample.
-------
There are three instrumental analytical techniques other than x-ray
fluorescence which are used to varying degrees for the determination of
the composition of air pollution particulates; optical emission spectroscopy,
atomic absorption spectroscopy and neutron activation. Each of these
techniques has sufficient sensitivity for the problem but each also has some
specific disadvantages. Arc or spark source optical emission is only semi-
quantitative for some elements at best because of difficulty obtaining uniform
excitation. The use of microwave or radio-frequency coupled plasmas show
some promise but non-equilibrium in the plasma may affect quantitative
interpretation. Atomic absorption requires that the sample be in solution
and permits the measurement of only one element at a time. As with optical
emission, some elements (the non-metals particularly) cannot be analyzed
at all by atomic absorption. Neutron activation is perhaps the most sensitive
technique for many elements. Among the elements of interest in air pollution
particulates, Pb, Fe, and S are three for which neutron activation has poor
detection limits. A particular disadvantage of neutron activation for samples
containing more than about five elements is the necessity to perform radio-
chemical separations or to count several times over extended periods to
avoid interferences.
•*tf
X-ray fluorescence"" appears attractive for the analysis of air pollution
particulates for several reasons;
1. No specimen preparation is required for filter collections; the material
on the filter is analyzed directly.
*
For readers unfamiliar with x-ray fluorescence analysis, appendix 1
gives a brief description of the technique.
-------
2. Detectability is fairly uniform across the periodic table and all
elements from atomic number 11 (Na) upwards can be analyzed.
3. X-ray technique is non-destructive and samples can be retained
for further analysis or as legal evidence.
4. Ten or twenty elements can be analyzed in one time period with
presently available commercial equipment for a cost of a few
dollars per sample.
X-ray measurements of air pollution particulates have been reported
2-8 • • •
In the literature but these have been limited in scope and no systematic
comparison has been made of the various excitation and detection methods.
The purpose of the work reported here was to examine the various ex-
citation sources and the techniques of measuring the characteristic x-rays
so that a realistic comparison might be made.
The objective is to furnish a suitable basis for choosing the most
applicable x-ray technique for air pollution problems.
STANDARDS
The importance of proper standards for quantitative analysis cannot
be overemphasized. Accuracy of the standard concentrations will determine
the limit with which the unknowns can be analyzed. For this reason it is
desirable that the concentrations of the standards be determined gravimetrically.
Appendix 2 describes the technique used to prepare our calibration standards.
2
We have chosen to express the concentrations in fig/cm and make the analytical
determination on a per-unit-area basis. It is, of course, straightforward to
convert to a volume unit, such as ug/m simply by knowing the area of the
sample on the filter and the volume of effluent (or ambient air) sampled.
-------
FILTER MATERIAL
In any technique of micro- or trace-analysis, one of the most im-
portant questions concerns the "reagent blank". In x-ray analysis of
air pollution particulates on filter substrates the "reagent blank" consists
of only the impurities in the filter material. It is obviously desirable
that the sample be collected on the purest filter available. Although glass
fiber filters have been used extensively because of their strength, the
impurity level is very high and varies significantly from batch to batch.
Table I lists the concentration of some elements of interest in a typical
glass fiber filter and in a membrane filter (Millipore) of considerably
higher purity. These data are the results of emission spectrographic
analysis and were provided to us by the Environmental Protection Agency
(EPA). Also listed in Table I are some neutron activation results (from
reference 9) for Millipore and for Whatman filter paper. Whatman filter
paper was chosen intuitively as the best substrate on which to prepare our
standards and was checked by scanning a limited wavelength range in the
crystal spectrometer; it showed no observable peaks. The purity of this
substrate was confirmed by the neutron activation analysis. The final
three filter materials listed in Table I were tape filters being considered
for use in a tape sampler. The requirements for the filter material used
in a tape sampler to be strong mitigates against Millipore and other materials
must be considered. The three candidate materials listed were analyzed by
x-ray fluorescence giving the results shown and indicating high impurity
levels.
-------
Element
Emission Spectroscopy
(from EPA)
Glass Fiber
Al
Si
S
Cl
K
Ca
V
Fe
Co
Ni
Cu
Zn
As
Se
Br
Cd
Pb
Mass Thickness
(mg/cm )
120
7000
0.25
1. 5
~ 500
250
0.025
4
<0.08
0.024
160
0.08
0.8
8
Millipore
0.06
0. 1
0.006
0.0006
0.006
0.3
0.0001
0.03
0.00002
0.001
0.006
0.002
0.008
5
TABLE I
Impurities in Filter Materials
Concentrations (fjig/cm )
Neutron Activation
(from Ref. 6)
Millipore Whatman
0.010
1.00
0. 10
0. 37
<0. 00005
0.04
0.0001
<0.02
0.06
0.007
<0.002
109,
0. 012
0. 1
0. 015
0. 14
<0. 00003
0. 04
0.0001
<0.01
<0. 004
<0. 025
0.005
10
A
16
20
31
X
Pallflex
E-60
C
154
474
1.0
1. 7
12
29
0.9
0.2
0. 3
31
-Ray Fluorescence
(NRL Results)
Pallflex
E-70
A C
34
1.0
1. 1
11
30
0.2
1.0
0.2
1.2
30 30
Acrapor
AN800
A C
61
490** 13
669
1.
0.
0.
0.
0.
0.
1
05
3
3
8
2
0.6
0.6
For X-Ray Fluorescence: A = 71.5 mCi Cd, Si (Li) Detector
Probably Cl rather than S
*#
C = Crystal Spectrometer
-------
EXPERIMENTAL APPROACH
There are several experimental approaches to x-ray fluorescence
analysis. They may be categorized according to how the characteristic
x-rays are generated or how they are dispersed and detected. In the
work under this contract we have used and compared nearly all the known
variations which are listed below and shown schematically in Figs. 1, 2,
and 3.
Generation
1. Primary x-rays from x-ray tubes (W, Cr, Mo, Rh, targets)
2. Monochromatic x-rays from fluoreseers excited by x-ray tubes
3. Radioisotopes (109Cd, 55Fe)
4. Positive ions (protons, alpha particles)
Dispersion and Detection
1. Crystal spectrometers with proportional or scintillation detectors
2. Solid state detectors with energy discrimination of the character-
istic x-ray lines
Wavelength Dispersion
Most measurements were made with a laboratory type vacuum crystal
spectrometer, Fig. 1, and employed only x-ray tube excitation because
the other sources did not give sufficient intensity. Cr, Rh, and W target
tubes were used, all operated at 900 watts (45 kV, 20 ma). As shown in
the figure, the x-ray tube was aligned so that the primary x-ray beam did
not strike the sample holder in an area which can be viewed by the collimator,
and therefore the mass which can scatter background into the spectrometer
is limited to the sample and its substrate. The sample chamber is evacuated
-------
Figure 1. Schematic diagram of the wavelength dispersive
crystal spectrometer.
Si (Li)
DETECTOR
X-RAY
TUBE
FLUORESCER
Figure 2. Schematic diagram of the energy dispersion chamber,
shown for fluorescer excitation. A radloisotope or an
x-ray tube can be placed at the fluorescer position.
SAMPLE
IONS
BEAM
DUMP
COLLIMATOR — ^
1 4
DETECTOR
Figure 3. Schematic diagram of the van de Graaff chamber used
for ion excitation.
-------
to eliminate air scattering of the primary beam. Fine collimation
(0.072° divergence angle) was used to provide good resolution and peak
to background ratio. LiF, graphite and KAP (potassium acid phthalate)
crystals were used in appropriate wavelength ranges to disperse the
radiation. The x-ray lines were measured with a flow proportional counter
and pulse height analyzer set for the full width at half maximum of the
pulse amplitude distribution.
Energy Dispersion
For the energy dispersion measurements a Si (Li) solid state detector
having an energy resolution of about Z80 eV at 5. 9 keV and 1000 c/s was,
used. A special vacuum chamber, Fig. 2, was constructed which eliminated
air scattering and yet permitted relatively close coupling of the source,
sample and detector. The offset geometry was used to prevent the detector
from viewing that portion of the chamber wall illuminated by the source.
This chamber was used for radioisotope, fluorescer and x-ray tube excitation.
Isotope sources of ^5pe ancj 109(3^ were compared with Mn and Ag
fluorescers respectively. The -* Fe isotope was of low activity (~ 7 mCi)
and a. counting time of 2000 seconds was used to make the "Fe measure-
ments comparable to the others. A special high activity (~ 70 mCi) *"°Cd
source was loaned to us by the Atomic Energy Commission (AEC) for the
"cd measurements. The W x-ray tube exciting the fluorescers was
operated at 45 kV (c.p. ) and 20 ma. Additional fluorescers of Cu and a
composite Cr-Zr powder were also tested.
The x-ray tubes emit so much radiation that, when used for direct
excitation, energy dispersion measurements can only be made with the
-------
tube operated at the lowest stable current (3 ma). In addition, a 1/16"
pinhole was placed between the x-ray tube and the sample to prevent the
intensity from overloading the detector. An x-ray tube capable of stable
operation at lower power should achieve the same results.
Ion excitation with 5 MeV protons or alpha particles, Fig. 3, was
carried out with the cooperation of the van de Graaff Branch at the Naval
Research Laboratory (NRL) and required special standards which could
withstand the ion beam current of about 50 nA on 1-2 mm . These special
standards were prepared by evaporating Cu, Au, or KBr onto thin (10 to
20 jig/cm ) carbon or plastic films and calibrating in the x-ray spectrom-
eter by comparison with filter paper standards. The ion beam was collimated
so that it struck only the sample and substrate and the detector was shielded
from radiation originating at the beam dump. The plastic films showed
some Fe and P impurities and the carbon film contained varying amounts
of Ti, Si, and Ca plus interference from Fe and occasionally Cu in the sample
holder. It is because of the sensitivity of the ion-excitation technique that
these low impurity levels are observable.
RESULTS
Results of the measurements made under this contract are divided into
two general categories: 1.) Sensitivities and detection limits determined on
the filter paper standards, and 2. ) Concentrations of the elements in real
samples provided by EPA.
The most important goal of the experiments was to compare the various
x-ray methods in terms of sensitivity'and limit of detectability for the elements
10
-------
of interest to EPA. By sensitivity we mean the slope of the x-ray intensity
2
versus concentration curve (counts /nanogram per cm ). Sensitivity depends
on excitation- source strength, fluorescent yield, spectrometer and/or
detector geometry, and detector efficiency. For detectability we use the
definition
where B is the number of counts for the background and S is the sensitivity
(the same counting time is used for the B and S measurements).
Sensitivity
The sensitivity was determined for each element of interest from the
calibration standards. Three concentrations of each element allowed con-
struction of a calibration curve of x-ray intensity versus concentration.
The slope of this curve represents the sensitivity in counts/jig per cm^.
All counting periods, except for the low activity radioisotope source and
the ion excitation, were 100 sees. Because of the low counting rate from
the Fe, counting intervals were 2000 sees. For the ion excitation, the
counting period was between 100 and 200 sees, until 5 flC of charge was
accumulated.
Table II lists the sensitivities measured by the various x-ray techniques.
As expected, sensitivity generally varies smoothly with atomic number ex-
cept where we change from K to L lines or, for wavelength dispersion, from
one crystal to another. For x-ray tubes there is enhanced sensitivity for
elements excited by the characteristic radiation from the tube target as can
be seen in Table II for the crystal spectrometer measurements of K and Ca
11
-------
TABLE II
Sensitivities for the Various X-Ray Techniques
Wavelength Dispersion
Crystal Spectrometer
100 sees
Element Cr Tube Rh Tube W Tube
Mg 50
Al 78
Si 672
S 687
Cl 1080
K 8920
Ca 5620
V 1040
Fe 440
Co 576
Ni 422
Cu 328
Zn 361
As(K8)
Se 48
Br 109
Sr
Zr
Mo
Cd(La) 597
Pb(La) 58
Au(La)
112
400
3220
3970
2020
I960
1430
1610
2100
1220
1230
1580
1430
78
450
460
840
300
290
1540
2420
2300
2000
1700
240
600
400
228
Energy Dispersion
Radioisotopes Fluorescers X-Ray Tubes 5 MeV Ions
7mCi 55Fe 70 mCi109Cd
2000 s 100 sees
24
70
1.8
3.9
6.0
9.9
14
29
46
54
56
15
100 sees
Mn Cu Ag Cr-Zr
. 1210
2940
660
1270
2200
187
322
930
2000
3100
3300
3300
770
700
209
630
1700
100 sees
Mo W W@Ni foil
370
1150
2200
3800
5500
6200
4620
8260
10100
6400
4400
3080
4900
6600
6300
8600
9500
10600
5 \l Coul.
Protons Alphas
45000
12500
4300
3400
36000
160
360
200
-------
with the Cr tube, S and Si with the Rh tube, Fe and Co with the W tube.
For energy dispersion measurements there is enhancement of Sr and Se
with the Mo tube and again Fe with the W tube. If the characteristic lines
are filtered from the x-ray tube spectrum, the sensitivity becomes more
or less uniform over a fairly wide atomic number range as is illustrated
by the data for the energy dispersion measurements using the W tube with
a Ni filter. Radioisotopes and fluorescers are essentially monoenergetic
sources; they are efficient in exciting neighboring elements but become
less and less efficient for elements of lower atomic number as shown in the
Table. The sensitivity for the fluorescers is between'50 and 100 times as
high as for the radio isotopes illustrating the effect of the higher intensity
available from the fluorescers. Excitation with 5 MeV protons or alpha
particles is particularly effective for low atomic numbers but falls off with
increasing absorption edge energy; the decline is faster for alpha particles
than for protons. An energy of 5 MeV is not high enough energy for efficient
excitation of the higher atomic numbers and more energy is required by
alphas than by protons for comparable excitation.
Detection Limit
The detection limit might be expected to improve with sensitivity and,
to a first order approximation, this is true. However, the additional de-
pendence of the detection limit on the background intensity makes the detect-
ability fluctuate somewhat. The background varies from element to element
depending on the impurity level in the substrate, the magnitude of the primary
radiation scattered by the sample and stray radiation originating as scattering
or fluorescence from various parts of the equipment.
13 .
-------
Table III lists the 100 sec. detection limits as measured; it must
be emphasized that these are single element detection limits determined
on standards containing no interfering elements. However, the back-
ground fluctuations due to the impurity level in the substrate (quite low
for the Whatman filter paper used to prepare the standards), the scattered
primary radiation and the fluorescence from the equipment are reflected
in the results because these are unavoidable. If the background is in-
creased due to other elements present in the sample, the detection limit
will degrade by the square root of the increase in this background. Since
the magnitude of the interference is a function of the resolution (which
defines line overlap), the problem is more serious for energy dispersion
than for wavelength dispersion. The Si (Li) detector used for the energy
dispersion measurements shown in Table III had a resolution of ~ 280 eV
which is about a factor of two poorer than the best detector available now.
If a state-of-the-art detector was used, the values reported in the table
would have been better by about the square root of two. An improvement
of another factor of two might be realized with such a detector because of
increased counting rate capability compared to the detector actually used.
With a higher powered x-ray tube such as presently available, the crystal
spectrometer measurements could have detection limits better than listed
by a factor of about two also. Of the measurements made by energy dis-
persion using x-ray excitation, only those using x-ray tubes directly are
comparable over a reasonable atomic number range with the wavelength
dispersion results.
The ion excitation results show quite good detection limits over a wide
atomic number range (especially for protons) even though the sensitivity
14
-------
Wavelength Dispersion
TABLE III
Detection Limits for the Various X-Ray Techniques
?' (corrected for F. P. Absorption)
Energy Dispersion
Crystal Spectrometer
100 sees
Element Cr Tube Rh Tube W Tube
Mg 0.45
Al 0. 36
Si 0.049
S 0.052
Cl 0.087
K 0. 003
Ca 0.010
V 0.053
Fe 0. 15 •
Co 0. 11
Ni 0. 18
Cu 0. 16
Zn 0. 18
As
Se 0.82
Br 0.39
Sr
Zr
Mo
Cd(La) 0.76
Pb(L
-------
decreased significantly as shown in Table II for the heavier elements.
The detection limit for 5 MeV protons is a direct result of high sensitivity
and the low background characteristic of proton excitation. It must be
remembered, however, that these measurements were made on standards
prepared on low mass (10-20 ug/cm ) nitrocellulose or carbon substrates.
The use of standard filter substrates would require lower beam current
and therefore longer counting times to achieve similar detectabilities.
Analysis of EPA Samples
The initial samples received from EPA were of three types: 1.) auto
exhaust samples from participants in the Clean-Air Car Race: 2.) ambient
air samples taken in New York City under three different atmospheric con-
ditions; and 3.) sixteen samples consisting of approximately 100 mg each
of water and benzene extracts of auto exhaust samples. Qualitative
analysis of these three types of samples showed readily observable Pb,
and Br, and Fe in the Car Race samples and in the auto exhaust material,
and Pb and Fe in the New York City samples.
The second set of samples consisted of four collections from the Bade
County (Florida) incinerator on Millipore filter and one auto exhaust sample
on teflon impregnated fiberglass. The results are shown in Table IV. Re-
sults are missing for some elements on some samples because of deterioration
of the sample during handling.
The third and largest group of specimens analyzed under this contract
were obtained from fixed sources sampled by EPA on Millipore filters.
A portion of some of the samples was retained by EPA for examination by
Atomic Absorption Spectroscopy. Results for all these samples are shown
16
-------
Auto Exhaust
Dade County
Incinerator
#1 5. 97 ft3
#2 7. 95 ft3
#3 7. 50 ft3
#4 7. 37 ft3
TABLE IV
Analytical Results on Auto Exhaust Sample
and Dade County Incinerator Samples
Concentration
Al
0. 17
0. 31
0.43
Si
0. 09
0. 10
0. 16
S
4.9
5.6
7.2
cr:
19
20
25
K
Ca
1. 5
2.6
3. 1
V
ND
0. 05
0. 04
Fe -
1.2
0. 68
0.63
0.83
Co
ND
ND
ND
Ni
0. 06
0. 16
0. 07
Cu
0.45
0. 50
0.42
Zn
14
20
8. 7
11
As
ND
ND
ND
Se
ND
ND
ND
Br
0.23
0. 14
0. 15
Cd
ND
Pb
73
15
24
13
16
Cl may be high in these results due to contamination during handling.
-------
on Table V. Agreement among the various x-ray techniques is fair when
one considers that in some cases the samples were not very homogeneously
distributed. A case In point might be illustrated by the Cu concentration
in Chicago, N. W. Incinerator Sample # 5 which analyzed to be 24 /jg/cm
by wavelength dispersion, 3 jug/cm^ by energy dispersion and 0. 77 fig/cm^
by atomic absorption, each measurement on a different portion of the same
sample.
The atomic absorption analyses show almost universally low results
when compared to the x-ray measurements. This is characteristic of
other atomic absorption analyses within our experience. The most likely
explanation is that the particulates are highly refractory being the result
of high temperature combustion and are therefore difficult if not impossible
to dissolve. Since an atomic absorption analysis depends on the sample
being in solution, the results almost invariably will be low. X-ray meas-
urements, on the other hand, require no specimen preparation and therefore
any errors cannot be ascribed to difficulty in dissolving.
Comparison of the results of these x-ray analyses by energy dispersion
with the results by wavelength dispersion points out some of the important
facets of the two analytical techniques. The most obvious observation is
the fact that the major elements can be measured by either of the techniques
using any of the sources (the Chicago Incinerator samples were simply not
109
examined for elements other than Zn and Pb with the Cd source). In-
termediate concentrations can be analyzed by either energy dispersion or
wavelength dispersion providing x-ray tube excitation is used. Elements
present at the lowest detectable concentrations can only be measured using
18
-------
TABLE V Analysis of Fixed Source Samples
CHICAGO N. W. INCINERATOR
Concentration (ug/cm )
A = 71. 5 mCi 109Cd, Si (Li) Detector
B = W X-Ray Tube, Ni Filter, Si (Li) Detector
C = Crystal Spectrometer
D = EPA Atomic Absorption Analysis
Sample # 5
A
Al
Si
S
Cl*
K
Ca
V
Fe
Co
Ni
Cu
Zn 28
As
Se
Br
Cd
Pb 32
B
22
26
18
4
1
3
33
46
C
.59
.62
22
81
18
1.8
2.4
.25
24
26
46
D
.97
. 15
.77
26
ND*
ND
. 58
11
Sample # 6
A
205
180
B
79
58
55
16
6
13
210
230
C
2.2
2.2
56
120
40
5.8
. 25
10
. 13
.66
7.0
240
2.2
280
D
6. 5
.24
4.9
229
ND
ND
16
255
Sample # 7
A
43
43
B
25
38
22
3
6
2
53
56
C
.63
. 35
20
100
21
1.4
.05
2. 5
.30
2. 5
58
.33
59
D
1.4
. 15
.66
28
ND
ND
.80
2. 1
Sample # 8
A
190
130
B
55
58
51
12
6
10
220
5
160
C
.91
.54
41
100
37
2. 5
7.2
.24
4.4
220
1.3
.97
180
D
4.7
.38
2.8
177
ND
ND
3.7
159
Sample # 9
A
B
C
3.6
.04
.24
0. 12
D
.067
. 12
. 045
0.22
ND
ND
ND
.089
Cl may be high in the crystal spectrometer data due to contamination during handling.
o-
^ND = not detected in A. A.
Blank = not detected in the XR FA (All elements listed were measured).
-------
Table V (Continued) Analysis of Fixed Source Samples
OTHER INCINERATORS Concentration (jug/cm2)
73rd St. # 3
A B C D
Al
Si
S
Cl
K 7
Ca 6
V
Fe
Co
Ni
Cu 3
Zn 64
As
Se
Br 3
Cd
Pb 99
27
19
34
6
.8
6
60
3
98
.05
2.4
5. 1
13
0.8
1.3
. 3
4.2
57
3.3
104
1.4
ND
2.4
81
ND
ND
2.0
28
73rd St. # 6
ABC
48
15
6
148
4
100
17
14
27
8
1.4
.3
3.4
125
4
71
.3
63
70
15
2.0
2.0
. 3
5. 0
116
3.2
116
SW Bkln # 1
A B C D
4
.7
1
9
.5
4
3
4
2
1
. 3
. 3
1
12
5
. 08
3. 2
17
3. 5
. 3
. 3
. 04
.4
.8
13
. 3
4.4
.28
.008
. 11
11
ND
ND
.097
4.8
SW Bkln # 3
ABC
37
13
2
2
54
1.4
30
9 .
13
10
3
3
2
47
2
33
1. 1
14
33
10
4.9
.02
2.5
.4
1.7
37
.7
34
lanoe
Sludge # 16
ABC
1
12
.3
10
5
2
2
1
.3
.4
1
11
.3
11
.3
8.1
1.8
2.5
.4
.3
.3
.9
13
.2
11
xanoe
Sludge # 17
ABC
2
25
2
28
9
5
3
1.3
.3
1
27
2
32
1. 1
.6
20
5.3
6.7
1.5
0.8
. 5
2.0
32
1.6
29
ro
O
-------
Table V (Continued) Analysis of Fixed Source Samples Concentration (jug/cm^)
Union Power Plant
.Sample #12
A
Al
Si
S
Cl
K
Ca 6
V
Fe 17
Co
Ni
Cu
Zn
As
Se
Br
Cd
Pb
B
2
2
2
5
20
.4
. 5
.7
.4
0. 1
C
6.2
8.6
21
2.8
5.6
. 07
14
. 1
.7
1. 1
. 5
1.0
Sample #13
A
4
4
.4
B
7
2
3
.3
.3
.6
C
2. 1
2. 1
29
. 5
3.0
.06
3. 5
.3
1.0
. 3
. 7
.8
Cement Plant
1 A-B gas
5. 33 ft 3
A
B
.3
.7
C
.03
.3
.3
.06
.2
.2
.4
.8
.3
D
. 094
.055
. 016
.031
ND
ND
ND
.062
4 D-B gas
35.75 ft3
A
3
. 5
B
.2
1. 5
. 3
. 1
C
. 1
.3
.7
.2
. 5
.3
.3
1.0
. 3
.2
1 A-B oil
28.85 ft3
A
3
. 1
1.2
B
.2
.2
.2
1.6
.6
C
.2
.3
.6
.2
.9
. 1
.2
1.0
.2
.9
.6
1 A-B oil
5. 30 ft3
A
.4
B
2
.4
.6
.6
C
.5
.3
.7
.2
.8
. 1
.2
.6
.2
.2
D
.28
.039
.031
.062
ND
ND
ND
ND
-------
wavelength dispersion with a crystal spectrometer. This last observation
might seem to be at odds with the comparable single element detection
limits demonstrated for the two methods of data acquisition. However,
the results illustrate the effect on the detection limit caused by the pre-
sence of interfering elements at widely different concentrations. Ad-
mittedly, the data reduction scheme used in analyzing these results did
not involve computer stripping of the spectra which might have succeeded
in identifying one or more of the missing elements. However, the stripping
technique which determines low concentrations from the small difference
between two large numbers is only a qualitative analysis and is perhaps
suspect even for that purpose when operating near the detection limit.
CONCLUSIONS
On the basis of the results of this investigation, it can be concluded
that x-ray fluorescence analysis can measure air pollution particulate
samples for the elements of interest to EPA at the concentrations en-
countered in practical situations. It seems appropriate to make some
comments at this point on the various facets of the x-ray technique to
provide guidance for its application to specific problems.
Various excitation sources were described in the Sensitivity part of
the Results Section. Generally, the use of x-ray tubes (high powered for
wavelength dispersion, low powered for energy dispersion) will provide
the most uniform excitation over a wide atomic number range. The target
of this x-ray tube might be chosen to enhance the sensitivity for a small
range of atomic numbers if these were of particular interest. The various
22
-------
monoenergetic sources such as radioisotopes, fluorescers or transmission
target x-ray tubes suffer the disadvantage that the sensitivity falls rapidly
as the absorption edge energy of the element of interest decreases away
from the energy of the primary photons. For some situations, however,
the minimum background present with these monoenergetic sources may
recommend them.
Most of the early literature references to x-ray analysis of pollution
samples ' ' ' ' used energy dispersion because the samples contained
many elements which would make scanning with a crystal spectrometer
prohibitively time consuming. Further examination of the problem points
out two difficulties with energy dispersion which are not likely to be over-
come in the near future:
1.) The best solid-state detector (150 eV resolution) will not separate
the Ka line of one element from the Kfi line of the next lower atomic number
in the region of the first period transition metal elements. Thus all the
elements from sulfur to nickel in the list of elements of interest in the
pollution problem will require mathematical unfolding to determine the
x-ray intensities. Although unfolding is an acceptable process for inten-
sities of similar magnitude it is not adequate for the range of concentrations
present in pollution samples.
4
2.) The solid-state detectors are limited to about 10 counts per
second if advantage is to be taken of their best resolution. In the energy
dispersive mode of operation the detector receives all the radiation at
the same time, including the characteristic lines of all the elements of
the sample as well as the scattered primary radiation (which may con-
tribute 50% or more of the total). For those elements present at low
23
-------
concentration in the sample, e.g., 10 ppm, 2 X 10 other photons must be
processed for each one photon of interest. At counting rates of 10^/sec
the counting time must be long to achieve reasonable statistics.
Based on these two limitations for energy dispersion we would have
to conclude that multichannel crystal spectrometers for wavelength dis-
persion offer the best approach to large scale analysis of all the elements
of interest. There are, however, other situations as outlined below which
permit other approaches. The three types of situations into which many
pollution problems may fall are:
1.) One or a few major elements of interest present in the sample at
concentrations where interferences are negligible, for example, Fe, Pb
and Br in air near a major highway. Energy dispersive analysis with a
low power x-ray tube or high activity isotope would be adequate. If only
one element were present at a concentration much above any others and
only that one element were to be measured, a simple proportional counter
could make the measurement. For more than one element a solid-state
detector would be desirable.
2. ) One or two elements of interest in the presence of some interferences
and at concentrations near the detection limit. The best resolution solid-
state detector would be required and it would be necessary to use an efficient
monoenergetic source for high sensitivity and low background. Counting
times might have to be 10 minutes or longer.
3. ) The more typical types of air pollution particulate samples (to
which this report has addressed itself primarily) require the analysis of
a single element or many elements at widely different concentrations and
24
-------
in the presence of significant interference from neighboring elements.
This analytical effort seems to require crystal spectrometers for best
resolution and the ability to separate most of the lines from possible
interferences. For routine analysis of such samples the only practical
analytical solution is the use of multichannel wavelength spectrometers.
It also seems desirable that these multi-spectrometer machines should
have an energy dispersive channel built into them for a qualitative ex-
amination of samples to insure that the elements for which the spectrom-
eters are set include all those present in appreciable amounts.
Particle Size Effect
The total amount of material collected for x-ray analysis (up to a
mg/cm ) does not show the usual inter-element effects encountered in
bulk specimens. However, there will be a size effect for particles larger
than a few microns. Fig. 4 shows the general particle size curve. For
5 /jm particles of metallic Fe and Pb, Si in SiO2» or Ca in CaSO4, the
intensity would be reduced to 60-80% of its value for the same ng/cm^
concentration but as smaller particles. With size-fractionated samples
there is no difficulty in correcting for particle size, but in general, accuracy
would be reduced if the bulk of the particles were of large size compared
to those in the calibration standard.
Data Handling for Routine Analysis
The simple linear'relation between x-ray intensity and concentration
for pollution samples means that the x-ray intensity can be converted to
concentration electrically with a zero bias to subtract background and an
amplifier to adjust for sensitivity. Thus the printout can be ng/cm ,
25
-------
ro
CTv
OQ
e
>-t
n
^
o
W
r-+%
(?
n
rt-
o
n
t— »
n
en
r-»-
N
O
3
&
01
3
en
»-••
r*-
^<
1.0
0.8
H
(f>
S 0.6
H
Z
< 0.4
o:
x
0.2
0
5/im Pb Particle
5/im Fe Particle
Ca in 5/im
CaS04
Si in
SiO
0.01 0.1 1.0 10
PARTICLE SIZE X LINEAR ABSORPTION COEFFICIENT
-------
jLlg/cm or any concentration unit desired without recourse to computers
or elaborate data handling equipment. This reduces cost and simplifies
operation for untrained personnel.
PHASE II
This report describes Phase I of the development of x-ray fluores-
cence for elemental analysis of air pollution particulates and a laboratory
effort to define the applicability of the x-ray technique to the problem. As
was mentioned in the text, the use of x-ray fluorescence as a routine
analytical tool requires a multichannel simultaneous x-ray analyzer.
The use of separate spectrometers for each element of interest has the
unique feature that each channel can be optimized (best crystal, best
detector, etc. ) for each element. The four major x-ray equipment
manufacturers in the free world each produce such an instrument, capable
of analyzing between 14 and 24 elements in a single time interval. Phase II
of this program is now in progress and involves the evaluation of these
commercial instruments for application to routine analysis. The evaluation
consists of the determination by NRL personnel of the sensitivity and de-
tection limits for the elements of interest on standards prepared at NRL.
These parameters will be compared with those measured during the
laboratory phase and some of the real samples Analyzed at NRL will be
analyzed on the multichannel machines to judge their routine analytical
capability.
ACKNOWLEDGEMENT
The technical assistance of K. L. Dunning and A. R. Knudson of the
van de Graaff Branch, Naval Research Laboratory, must be acknowledged
for the ion excitation effort.
27;
-------
REFERENCES
L. S. Birks, J. V. Gilfrich and D. J. Nagel, "Large Scale
Monitoring of Automobile Exhaust Particulates, " Naval Research
Lab. Memo. Report 2350, Oct. 1971.
F. S. Goulding and J. M. Jaklevic, "Trace Element Analysis by
X-Ray Fluorescence, " Lawrence Berkeley Laboratory, UCRL-20625,
May 1971.
3
T. R. Dittrich and C. R. Cothern, J. Air Pollut. Control Assoc. 21,
716 (1971).
4
J. Leroux and M. Mahmud, J. Air Pollut. Control Assoc. 20,
402 (1970).
T. B. Johansson, R. Akselsson and S. A. E. Johansson, "Proton-
Induced X-Ray Emission Spectroscopy in Elemental Trace Analysis, "
Lund Institute of Technology, LUNP7109, Aug. 1971.
H. R. Bowman, J. G. Conway and F. Asaro, Envir. Sci. and
Tech.
-------
APPENDIX 1
Fundamentals of X-Ray Fluorescence Analysis
Although the basic principles of x-ray fluorescence are available
from a number of texts ~ ° ~ , it seems appropriate to give a
brief description of the technique here for the benefit of those readers
of this report who may be unfamiliar with it.
X-ray fluorescence analysis is based on the measurement of the
wavelength and intensity of the characteristic x-rays emitted by a
sample which has been excited by electrons, photons or ions. A des-
cription of the technique can be naturally divided into three areas:
1.) X-Ray Generation
2. ) Wavelength and Intensity Measurement
3.) Data Interpretation
A discussion of these three areas will point out the essential details of
x-ray fluorescence analysis.
X-Ray Generation
Generation of characteristic x-rays requires two steps: first an
atom must be ionized by removing one of its inner electrons; this can
be done by bombarding with high energy electrons, ions or photons.
Second the missing electron must be replaced by one of the outer electrons,
It is the replacement process which causes emission of a characteristic
x-ray photon. Figure A-1 which is an energy level diagram illustrates
the process schematically and shows the particular transitions for the
K and L series lines; for instance if a K-shell vacancy is filled from the
outermost L level the Ko^ line is emitted.
A-l-1
-------
3=-
I
I
ro
TO
i-l
n>
>
i
Cfl
n
r s.
j"
;• 3
3 (3
a,
M-
OTQ
3
O
O
3
en
O
N
M
L
K
"
ft
Ka
V
-------
In order to excite the atom the incident quantum must have an energy
greater than the binding energy of the inner shell electron being removed.
Electrons are the most efficient quanta for generating characteristic x-rays
but also generate relatively intense continuous radiation which exists as
background under the characteristic line, degrading the detection limit.
Electrons are used, of course, in x-ray rubes but are not commonly used
in analysis except for the electron probe micro-analyzer where a focused
electron beam allows analysis of small local areas in the sample. Ion
excitation, such as protons or alpha particles, require the use of a van
de Graaff or cyclotron in order to accelerate these particles to high enough
energy to produce effective excitation. The heavier particles generate
negligible continuum compared to electrons which results in improved de-
tection limits for those samples which lend themselves to high energy ion
excitation.
Excitation by x-ray photons is the most common technique in x-ray
fluorescence analysis. (Actually a rigid definition of the word "fluorescence"
requires that photons be used to generate photons. ) X-ray sources used to
excite samples include x-ray tubes, x-ray tube - fluoresce r combinations
and radioisotopes. The most intense of these sources is the x-ray tube;
a tungsten target tube operated at the moderate power of 45 kV (c.p. ), ZO.mA
emits about 6x10 ph/s/sr line and continuum radiation. Flourescers are
of intermediate intensity; a silver fluorescer excited by a tungsten target
tube at 45 kV, 20 mA emits about 3 X 10^ ph/sec/sr of Ag K radiation.
On the low end of the intensity scale are radioisotopes; 100 mCi of 'Cd
emits about 3 X 10° ph/sec/sr of Ag K radiation (a higher activity isotope
A-l-3
-------
would, of course, be more intense but would also be hazardous to handle).
Photon excitation does not generate any continuum; the primary source
of background with x-ray tube excitation is the scattered primary radiation.
Wavelength and Intensity Measurement
The two measurement techniques which are available are wavelength
dispersion and energy dispersion, so called because of the means used to
identify the characteristic lines of the elements. The wavelength and energy
are uniquely related as for all electromagnetic radiation according to the
equation
E = \iv
where E" is the energy, h is Planck's constant and V is the frequency. The
frequency is related to the wavelength (X) by the velocity of light. When
translated into convenient units the equation becomes
E = 12398/X
o
with X in angstroms (£ ) and E in electron volts (eV).
The basic principle of wavelength dispersion is the Bragg Equation:
nX = 2d sin 0
where n is the order of diffraction,
X, as before, is the wavelength,
d is the interplanar spacing of the crystal
and 0 is the diffraction angle.
Thus, in an instrument as illustrated by Fig. 1 in the body of this report,
the characteristic lines being emitted by the unknown sample can be
identified by the 0 angle at which they diffract. By selecting suitable
crystals it is possible to make measurements over the wavelength range
A-l-4
-------
of about 0. 25 to 150 A. The ability to separate two lines of nearly the
same energy is called the resolution. For a flat crystal spectrometer,
the resolution is a function of the collimator spacing and the rocking
curve breadth of the crystal; when combined by the rule of variance,
these two parameters define the divergence of the system, A6, in the
differential form of the Bragg Equation:
AX = 2d cos 6A9.
Since the resolution is also a function of the 8 angle, it will vary over
o
a wide range from about 0. 001 to 0. 1 \ or from a fraction of an electron
volt to several hundred electron volts. The resolution of crystal spectrom-
eters is more than adequate to distinguish each element without interference
from either the (y or j3 lines of neighboring elements.
Energy dispersion identifies the elements present in a sample directly
by their characteristic photon energies. This is done with an energy sensitive
detector which emits a pulse each time an x-ray photon is detected, and the
amplitude of this pulse Is proportional to the photon energy. (This technique
is sometimes called "non-dispersive" to differentiate it from the "dispersive"
nature of the analyzing element in a crystal spectrometer. ) The detectors
commonly used in crystal spectrometers, I.e., gas proportional counters
and scintillation counters, have been used in energy dispersion but their energy
resolution is poor compared to the solid state, Li drifted, Si detectors pre-
sently available. Virtually all of the energy dispersion measurements being
made today use the solid state detector, in geometry similar to that shown
in Fig. 2 in the body of the report. The energy dispersive system is quite
efficient because the detector can accept a large solid angle from the sample,
A-l-5 i
-------
and therefore make practical use of low intensity sources such as >
radioisotopes. The resolution of a state-of-the-art solid state detector
is about 150 eV at 5. 9 keV (Mn Key), which is poor compared to 14 eV
for a LiF crystal spectrometer at the same energy.
In either energy or wavelength dispersion a vacuum path is required
o
if wavelengths longer than about Z. 5 A are to be measured because air
absorption becomes significant. Thin detector and source windows are
o
also needed for the work beyond 2. 5 A .
Data Interpretation
Quantitative analysis of pollution samples is far easier than ordinary
x-ray fluorescence of bulk material because the amount of sample is too
A-3
small to show interelement absorption or secondary fluorescence.
Measured x-ray intensity is converted directly to grams or grams/cm
with the individual straight line calibration curves. The only foreseeable
difficulty which can occur is with particle size as described in the body of
the report.
A-l-6
-------
APPENDIX 2
Preparing Calibration Standards
Sensitivity values and therefore quantitative analysis of unknown
samples depends on calibration standards for each of the elements. Two
methods of preparing standards were employed: a.) evaporation of 0.25mj£
of a known solution of a soluble salt of the element onto 9 cm disks of
Whatman #42 filter paper, b. ) filtering 11 m& from suspension of a known
insoluble salt of the element.
Evaporation of Soluble Salts
Table A-l shows the concentrations prepared for the various elements
of interest. Since this method of preparation is a fundamental gravimetric
procedure it needs no cross-check except for uniformity of deposit and
uncertainty in starting concentration or dilution. Uniformity was measured
by cutting 10 one cm squares from the 50 ^g/cm^ Zn standard and measuring
the variation in x-ray intensity. As shown in Table A-2 the standard de-
viation in the 10 determinations is 4% relative and the maximum deviation
from average is 6% relative. Variations in preparing stock solution, dilution,
and measuring 0.25 m£ were checked by 6 separate preparations which dif-
fered by less than 3% for any element.
The evaporation method seemed ideal for most elements above atomic
number 26 (Fe) because their characteristic radiation would not be absorbed
by emergence through the filter paper. For the K lines of elements below
26 in atomic number and for Cd La there is increasing absorption and a
correction factor as tabulated in Table A-3 was necessary in order to correct
to real unknown particles which are all deposited on the surface of Millipore.
A-2-1
-------
TABLE A-l
Concentrations of Elements of Interest in a
Typical Set of Standards
Element Solution Concentration of 3 Standards
(Ug/cm2)
Na
Mg
Al
Si
S
Cl
K
Ca
V
Fe
Co
Ni
Cu
Zn
As
Se
Br
Cd
Pb
NaCl in H2O
MgO in dil. HNO3
A12(SO4)3 in H2O
Na2SiO3 in H2O
A12(S04)3 in H2O
NaCl in H2O
KBr in H2O
CaNO3 in H2O
V in dil. HNO3
FeSO4 in H2O
CoCl2 in H2O
NiSO4 in H2O
CuSO4 in H2O
ZnCl2 in H2O
As in dil. HNO3
H2SeO3 in H2O
KBr in H2O
CdCl2 in H2O
PbO in dil. HNO,
32
41
51
23
90
49
23
27
43
35
34
45
64
40
22
35
48
13
45
7.
11
12
5.
22
12
6.
6.
11
8.
8.
11
16
10
5.
8.
13
3.
12
8
8
5
8
8
6
6
8
3
3.0
4.4
5.7
2.1
10
4.6
2.4
2.6
4.4
3. 5
3.5
4.4
6.4
4.0
2.4
3.5
4.9
1.6
4. 1
A-2-2
-------
TABLE A-2
Uniformity of Deposition for a Standard on
Filter Paper
Sample No. Zn KOi Intensity
(Each 1 cm2) (c/100 sees/cm2)
1 8880
2 8850
3 9730
4 9650
5 9910
6 9370
7 9130
8 9340
9 9550
10 9320
Average 9373
Std. Dev. 350
Rel.Std.Dev. 3.7%
A-2-3
-------
TABLE A-3
Calculated Filter Paper Absorption Factors
Element Correction Factor
Na 51
Mg 31
Al 20
Si 14
S 6.3
Cl 4. 5
K 2.8
Ca 2.2
V 1.4
Cd(Lft) 3.5
A-2-4
-------
TABLE A-2
Uniformity of Deposition for a Standard on
Filter Paper
Sample No. Zn Kft Intensity
(Each 1 cm2) (c/100 sees/cm2)
1 8880
Z 8850
3 9730
4 9650
5 9910
6 9370
7 9130
8 9340
9 9550
10 9320
Average 9373
Std. Dev. 350
Rel.Std.Dev. 3.7%
A-2-3
-------
TABLE A-3
Calculated Filter Paper Absorption Factors
Element Correction Factor
Na 51
Mg 31
Al 20
Si 14
S 6.3
Cl 4. 5
K 2.8
Ca 2.2
V 1.4
Cd(Ltt) 3.5
A-2-4
-------
The correction factor relates the x-ray intensity from a given concentration
deposited through the volume of the filter paper to the intensity of the same
concentration deposited on the surface of the filter. The values listed in
Table A-3 are factors by which the sensitivity (in c/s/jUg/cm^) determined
on the filter paper standards are multiplied before comparison with data
taken on the real samples. Each correction factor is believed accurate to
+ 25% but an independent method of preparing particulate standards was
desired and led to the filtering of insoluble salts from suspension.
Filtering Insoluble Salts from Suspension
A weighed amount of an insoluble salt of the chosen element was placed
in suspension in 500 m£ of 3:1 glycerine-water mixture so it could be dis-
persed uniformly. From this, an aliquot of 1 1 ml was made by dipping a
beaker into the agitated suspension. This was diluted with 10 m£, of water
to reduce its viscosity so it could be vacuum filtered onto a Millipore substrate,
Initial tests with Fe were compared with the Fe standards prepared by
evaporation; agreement was 60% which indicated that one could prepare
standards in the*desired range but with somewhat less than the desired
accuracy. Therefore a cross check of concentrations was made by neutron
activation for the elements below Fe in atomic number. Results are shown
in Table A-4 and show some discrepancies which cannot be explained. For
Mg, Al, Ca and V for which NAA results seemed reasonable, the particulate
standards were checked against the solution standards using the absorption
correction factor described in the preceding paragraph. The results given
in Table A-5 indicate that the solution standards can be used within a 10-20 %
accuracy which seems acceptable for pollution analysis.
A-2-5
-------
TABLE A-4
Results of Neutron Activation Analysis of the
Particulate Standards
(Insoluble Salts Deposited from Suspension on Millipore)
Element
Na
Mg
Al
Si
S
Cl
Ca
V
Cd
Sample
Designation
H-2
E-2
L-2
M-2
H-l
L-2
M-2
H-l
L-2
M-2
H-2
L-l
H-2
M-l
H-2
E-2
L-2
M-2
H-2
L-2
M-2
H-2
L-l
' M-2 .
H-l
Nominal Cone.
(Ug/cm2)
160
710
4. 3
27
59
8. 1
27
62
7. 5
25
57
3.7
12
33
230
1035
5. 3
18
40
3.9
27
55
10
34
78
NAA Result
(|L£g/cm2)
210
980
14*
31
53
8.2
12
37
144*
155*
190*
4. 7
15
43
475
2285
-i* j,
N.D."--
N.D.
35
3.0
16
27
16
52
150
Results considered unacceptable.
'N. D. - Not Detectable
A-2-6
-------
TABLE A-5
Comparison of Solution Standards
After Correction for Filter Paper Absorption
With Particulate Standards for Determining
Sensitivities for Light Elements
Sensitivity (c/s/tig/cm2)
Insoluble Salts Soluble Salts on Filter Paper
Element Deposited on Millipore (Corrected for Absorption in
Filter Paper)
Mg 0.50 0.45
Al 1.2 1.6
Ca 14 16
V 16 18
A-2-7
-------
It would be desirable to pursue the preparation of particulate
standards further but it was not possible to do this within the present
contract.
A-2-8
-------
APPENDIX REFERENCES
A"1H. A. Liebhafsky, H. G. Pfeiffer, E. H. Winslow and P. D.
Zemany, X-Ray Absorption and Emission in Analytical Chemistry,
(John Wiley and Sons, N. Y. , I960).
A-2
R. Jenkins and J. L,. DeVries, Practical X-Ray Spectrometry,
(Springer-Verlag, N. Y. , 1967).
A-3
L. S. Birks, X-Ray Spectrochemical Analysis, (Wiley-Inte rscience,
N. Y., 1969), 2nd ed.
A-4
E. P. Bertin, Principles and Practice of X-Ray Spectrometric
Analysis, (Plenum Press, N. Y. , 1970).
A-2-9
------- |