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

-------