<>EPA
             United States
             Environmental Protection
             Agency
             Environmental Sciences Research
             Laboratory
             Research Triangle Park NC 27711
EPA-600/2-78-213
November 1978
             Research and Development
Proton Scattering
Analysis for Light
Elements  in Air
Particulate  Matter

-------
                RESEARCH REPORTING SERIES

Research reports of the Office of Research and Development. U S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:

      1.  Environmental Health  Effects Research
      2.  Environmental Protection Technology
      3.  Ecological Research
      4.  Environmental Monitoring
      5.  Socioeconomic Environmental Studies
      6.  Scientific and Technical Assessment Reports (STAR)
      7.  Interagency Energy-Environment Research and Development
      8.  "Special" Reports
      9.  Miscellaneous Reports

This report has  been assigned  to the ENVIRONMENTAL PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
                                                 EPA-600/2-78-213
                                                 November 1978
    PROTON SCATTERING ANALYSIS FOR LIGHT ELEMENTS IN
                 AIR PARTICULATE MATTER
                           by
        J. William Nelson, Principal Investigator
G. M. Hudson, H.C. Kaufmann, W. J. Courtney, I. Williams
      K. R. Akselsson, D. Meinert, J. W. Winchester
                 Forida State University
               Tallahassee, Florida  32306
                    Grant R - 802913
                     Project Officer

                Thomas G. Dzubay  (MD-47)
       Atmospheric Chemistry and  Physics Division
       Environmental Sciences Research  Laboratory
      Research Triangle Park, North Carolina   27711
       ENVIRONMENTAL  SCIENCES RESEARCH  LABORATORY
           OFFICE OF  RESEARCH AND DEVELOPMENT
          U.S. ENVIRONMENTAL PROTECTION AGENCY
      RESEARCH TRIANGLE  PARK, NORTH  CAROLINA  27711

-------
                          DISCLAIMER








        This report has been reviewed by the Environmental



Sciences Research Laboratory, U.S. Environmental Protection



Agency, and approved for publication.  Approval does not signify



that the contents necessarily reflect the views and policies of



the U.S. Environmental Protection Agency, nor does mention of



trade names or commercial products constitute endorsement or



recommendation for use.

-------
                           PREFACE





        The intent of this report is to summarize our research



on proton scattering as it applies to the analysis of samples



of particulate matter found in the atmosphere.  The most



important details (such as kinematics and angular distributions)



are treated in the reprints which form the appendices.  While



the scattering literature of nuclear physics is vast, only



articles of direct relevance to the subject matter of this



report are included in the bibliography.  Included as appendix



B is a reprint which describes the computer code REX which



serves as the model for computer code SCAT.
                             111

-------
                           ABSTRACT








        Proton scattering analysis has been developed to provide



light element analyses as a complementary method to x-ray anal-



ysis.  Samples of air particulate matter from both filter and



impaction devices have been analyzed and time sequence and par-



ticle size information obtained.  In combination with x-ray



analysis, total mass may be determined by summation of elemental



masses; however, further investigation is needed to compare with



the gravimetric method.  Like x-ray analysis, it has a broad



range, and is rapid, subject to automation, and non-destructive



of the prepared sample.  Analyses of typical atmospheric samples



may be performed in five minutes or less.  With further develop-



ment, the method can readily be advanced from experimental to



routine usage.





        This report was submitted in fulfillment of Grant



Number R-802913 by Florida State University under the sponsor-



ship of the U.S. Environmental Protection Agency.  This report



covers a period from May 1, 1974 to December 31, 1977, and



work was completed as of January 16, 1978.
                              IV

-------
                           CONTENTS



Preface. .	     iii

Abstract	o.. „	      iv

Figures	     vii

Tables	      ix

  1.  Introduction	       1

  2.  Conclusions	       5

  3.  Recommendations.. „	       6

  4.  Proton Scattering by Particle Analysis -
      The Choice	       8

  5.  Air Particulate Analyses	„	      11

  6.  Computer Code for Data Evaluation - SCAT	      15

  7.  Total Mass Determination	      19

  8.  Carbon Determinations	      23

References	      28

Bibliography	      29

Appendices	....	      30

     A.  "Proton Scattering for Analyses of
         Atmospheric Particulate Matter" by
         K. R. Akselsson, J. W. Nelson and
         J. W. Winchester  (Reprint from NBS
         SP 425, Vol. A, 1975)....	      3d

     B.  "Rex-A Computer Programms for PIXE
         Analysis" by Henry C. Kaufmann,
         K. Roland Akselsson and William J.
         Courtney  (Reprint from NUCL. Inst.
         and Meth 142, 251 (1977)	      34

                               v

-------
C.  "Light Element Analysis by Proton
    Scattering" by J. W. Nelson and W. J.
    Courtney (Reprint from Nucl. Inst. and
    Meth 142, 127 (1977)	    41
                          VI

-------
                             FIGURES
Number
 1.   Schematic representation of proton scattering

     by 12C and 28Si,..................................       4,

 2o   Proton scattering spectrum for an impactor

     sample...........................................        4

 3.   a)  Inelastic scattering y~raY spectrum for filter

         aerosol sample. . ,	        9

     b)  Elastic and inelastic particle spectrum for

         the same sample..............................        9

 4.   PESA spectrum for an aerosol sample from St. Louis,

     Missouri o ..................	       11

 5.   Time plot of PESA data from the General Motors

     Sulfate Dispersion Experiment	       12

 6,   Combined PESA-PIXE analysis of an aerosol sample

     taken with an impactor	       13

 7.   Particle size plot from combined PESA-PIXE

     analyses.........................................       14

 8.   Calibration spectrum using the plastic Kapton....       17

 9,   Calibration spectrum from an aluminum foil.......       17

10.   Scattering spectrum for an indoor aerosol

     sample...........................................       2.0
                               VII

-------
Number                                                      Page




11.  Background scattering spectrum taken a few



     millimeters distant from the spectrum of



     Figure 10	       20



12.  PESA spectra for Nuclepore and Millipore



     filters	       25



13.  PESA spectra for possible impactor surfaces-



     Nylon and Mylar	       25



14.  PESA spectrum for an impactor sample which allows



     quantitative determination of aerosol carbon....       26



15.  Spectrum for Nuclepore filter sample of sufficient



     thickness to permit carbon content



     determination	       27
                              Vlll

-------
                            TABLES

Number                                                    Page

1.  Approximate detection limits for light element
    PESA determinations on typical aerosol field
    samples taken on Nuclepore filter except for
    carbon which is reported for a polystyrene backed
    impactor sample 	     2
    Masses of the light elements determined by PESA
    for an indoor industrial filter sample	     22
                              IX

-------
                           SECTION 1



                         INTRODUCTION



        The scattering of protons by the nuclei of elements has



been successfully applied to the quantitative analysis of the



light elements.  The method is broad-range, rapid, subject to



automation and non-destructive of the prepared sample.  As such,



it furnishes an ideal complement to x-ray analysis especially



since the necessary sample requirements are almost identical.



Absorption of x-rays by the sample itself places a limit on the



lightest elements measurable by detection of x-rays.  For



samples thicker than about 1 mg/cm2  (or particle  sizes



about 10 ym) x-ray absorption corrections become difficult for



elements of atomic number Z = 13 (aluminum) and below.  A proton



scattering system was developed to observe the light elements



and overlap the lower range of the x-ray technique.



        Proton scattering is essentially a form of nuclear mass



spectroscopy which may be carried out in an efficient, broad-



range fashion utilizing commercially available silicon surface



barrier detectors.  Figure 1 pictoralizes the elastic scattering


                         12      28
of 16 MeV protons by the   C and   Si nuclei for three angles.



The energies of the scattered protons are predicted using the



laws of conservation of mass-energy and conservation of linear



momentum.  Our earliest exploratory measurements on a size

-------
 fractioned  aerosol  sample  from  Tallahassee, Florida  is  shown  in



 Figure  2.   Resolution  is sufficient  to permit  analyses  up  to  and



 including Z =  17, chlorine.   Following this initial  success,



 efforts were directed  to the  design,  construction  and test of an



 efficient system suitable  for both filter  and  impactor  aerosol



 samples.  This report  details the results  obtained on samples



 from field  projects, progress on creating  a computer code  for



 evaluation  of  such  data and results  of supportive  calculations



 (kinematics) and measurements (angular distributions and excita-



 tion functions)  needed to  fully evaluate and define  the limits



 of the  system.







TABLE 1  APPROXIMATE DETECTION LIMITS FOR LIGHT ELEMENT PESA



         DETERMINATION ON TYPICAL AEROSOL FIELD SAMPLES TAKEN ON



         NUCLEPORE FILTER EXCEPT FOR CARBON WHICH IS REPORTED



         FOR POLYSTYRENE BACKED IMPACTOR SAMPLES.
Element
C
N
0
F and Mg
Na thru Ca
3a Detection Limit
0.2
0.2
1.4
2.0
0.5
(yg/m3)






-------
        Using actual aerosol field samples, typical 30 detection



limits for light elements have been calculated and are shown in



Table 1.  As will be detailed in this report, these limits allow



significant measurements on atmospheric particulate matter with



time resolution as short as two hours.  Carbon determinations



are treated in detail in Section 8.  All spectra presented in



this report are the raw spectra obtained by multichannel analy-



sis with no background subtraction or other treatment.

-------
                    PROTON   ELASTIC  SCATTERING
                         12.00
                          I35°\
13.52
 I90'
 15.23
.'45°
                                   »*
              16 MeV
                         14.15
                        I35°\
 14.89
  190°
   15.67
   ' 45°

                16 MeV

                  P
                                    28Si
                                                                 12
Figure  1.   Schematic representation of proton scattering by   C
            and 28Si.


900

400
g
§ 3°°
1 o
200
100

n
	 . 	 —
Ep- 15.98 MeV PROTON SCATTERING AIR PARTICULATE- TLH
6L'I35°
_ Q -4/iC 0 2^i _
I2C 1 1
\ \

—
-
1
1

JJ60keV
I3C *?
•• '
*'->Kn ' '. ..
• v •' •, • t
_.,-,.,-.-...-.,-,- 	 : n|^wy,«JL/*v---';

"No
i
J
If.-. V




-
.
—
• Ji "f .
vl J
.. /i |!"J'


i s. 	 i ^ ^j»^*y i 	 i f i
i rwp-pf .|i£*._..^.
                          I2X>0     12.50          13.35
                           SCATTERED PROTON ENERGY (MeV)
Figure 2.  Proton scattering spectrum  for an impactor sample.

-------
                           SECTION 2



                          CONCLUSIONS



        Proton elastic scattering analysis for light elements



was developed as the only choice for a broad-range complement



to x-ray analysis.  Like x-ray analysis it is broad-range, rap-



id, subject to automation and non-destructive of the prepared



sample.  Furthermore, it's sample requirements are nearly iden-



tical to those of x-ray analysis.



        For air particulate matter it has been demonstrated to



be effective for both filter and impactor samples to produce



both time sequence and particle size information with excellent



resolution.  In combination with x-ray analysis total mass may



be determined by summation of elemental masses.  Carbon deter-



minations represent a special problem for any method, in that



most of the desirable backing materials contain carbon.  This



problem appears surmountable by use of thinner filters and im-



paction surfaces.

-------
                           SECTION 3



                        RECOMMENDATIONS



        In light of the reported findings proton scattering is



recommended for further development in the following aspects:



        1.  develop more precise determination of carbon content



            through the use of Teflon filters.



        2.  explore use of thinner filters to facilitate deter-



            mination of carbon content of the aerosol while



            maintaining compatibility with x-ray analysis and



            excellent time resolution.



        3.  pursue advanced programming methods for spectral



            evaluation with emphasis on non-Gaussian peak



            shapes.



        4.  move angle of observation from 120 to 135 degrees



            to enhance sensitivity of other elements relative



            to carbon.



        5.  continue to refine the method by application to



            field samples.



        6.  add detectors for hydrogen determination.



        7.  use PESA to better understand the attenuation cor-



            rections needed to x-ray analysis.



        8.  confirm the efficacy of the method for nitrogen



            determination for aerosol samples.

-------
9.   examine and define volatility limits for aerosol



    samples under analysis conditions.

-------
                           SECTION 4



     PROTON SCATTERING BY PARTICLE DETECTION - THE CHOICE



        A variety of nuclear reactions may be used for quanti-



tative measurements of the light elements.  Proton scattering



was chosen primarily for two reasons.  Firstly, its broad-range



characteristic enables measurement of all light elements simul-



taneously in a single bombardment with a single detector.



Secondly, proton scattering cross sections are generally higher



in value than those of other nuclear reactions.  Still another



desirable characteristic is the appearance of both the elastic



and one or more inelastic peaks in the spectra which provides a



definitive "signature" for each element.



        Proton inelastic scattering by gamma-ray detection is



also a method which was examined.  This method has been studied



in considerable detail by B. L. Cohen ejt al  with samples of



about 200 mg.  In order to maximize sensitivity over a range of



elements multiple detectors of different sizes should be used



depending upon the energy of the emitted gamma ray.  Further



improvement in sensitivity and selectivity would necessitate



bombardments at different energies for different elements on



particular substrates.  In addition, Germanium detectors for



gamma-rays are considerably more expensive than silicon detec-



tors for charged particle observation.





                               8

-------
a)
   5000
  4000-
  3000
D
D
O
  2000
   1000
     0
 b)
  10000 -
   1000
c
3   100
     10
                              Inelastic  Scattering  y-raySpectrum
                              Air Particulate - Nuclepore
          100  200   300  400  500  600  700  800   900
                        Channel  Number
          Proton Scattering-Particle Spectrum
           Air Part culate - Nuclepore
      |QJ	L.
          100  200  300  400  500  600  700  800  900
                       Channel Number
Figure 3   a.

            b.
          Inelastic scattering  j-ray spectrum for a  filter
          aerosol sample.
          Elastic and  inelastic  particle  spectrum for the
          same  sample  and  same  integrated charge.

-------
        Our goal was finding a light element measurement scheme



suitable for use on the same samples as used for x-ray analysis.



The spectra resulting from bombardment of such an air partic-



ulate sample on Nuclepore filter substrate are shown in Figure



3.   The inelastic scattering gamma-ray spectrum of Figure 3A


                                             12      32
reveals thei peaks due to scattering from the   C and   S first



excited states.  This spectrum was measured at a proton bombard-



ing energy of 7.0 MeV an 8% efficient, 3.5 keV FWHM lithium



drifted Germanium detector.  Higher energy peaks due to   0 are



excluded due to insufficient bombardment energy, however, if



included a considerable increase in background would result.



By contrast, Figure 3B shows the particle spectrum from bombard-



ment of the same sample.  Carbon, nitrogen, oxygen, silicon,



sulfur and potassium-calcium peaks are observed.  The potassium-



calcium (mass numbers 39-40) are not resolved but could readily



be resolved and quantitatively analyzed by the x-ray emission



analysis.   The above considerations led to the choice of devel-



oping the particle scattering method as the most rapid and the



only truly broad-range method for analyzing light elements in



aerosol samples.
                              10

-------
                            SECTION  5


                    AIR PARTICULATE  ANALYSES


         An essential component in the  development of a useful


 analysis tool is the demonstration  of  the applicability to real-


 istic field samples.  Two examples  of  the application of PESA to


 typical aerosol samples are given in this section.  Figure 4


 shows the proton scattering spectrum for an 0.3m  air particulate


 filter (Nuclepore) sample taken  in  St. Louis, Missouri.  A total


 of eight elements is observed in a  spectrum quite suitable for


 overlapping with x-ray analysis.
  200
CO
I-
Z


0IOO
     0
   PESA

Ep = 15.990

0=120°
Target STL'WU

Streaker 0.3m
                                                       4C
                                                       . Ca
                                                           35
                                                23
                                                       2Q0.
                                                        Si ,„ -.
                                                             f|

                                                             U '
                                                 Na   C.   ;•'••'
            j_
_L
                                                    J_
                    12.420
                    12.879
                                  13.232
             14.018   14.351 I4.547J 14.796

                       14.665 14.824
  Figure  4.
            Scattered  Proton  Energy  (MeV)

      PESA spectrum for an aerosol sample  from St.  Louis
      Missouri.
                                 11

-------
         From similar  spectra, the time plots (with two hour



resolution)  w,ere obtained and a,r-e  shown in Figure 5,
    ro
        2.0
         0
         1.0
         o

        2.5
     o
     ,*   o
     LJ
     o
     z
     o
     o
         1.0
          0
             GM  SULFATE  EXPERIMENT     PROTON

                                          SCATTERING
                                             Ca-K
 I	I
               10
                                             Si-AI
                      N
12   13   14   15  16   Oct.

 TIME
Figure 5.  Time plot of PESA data  from  the General Motors

           Sulfate Dispersion Experiment.
                            12

-------
          For particle size analysis, impactor collected

 particulate matter may be analyzed as shown in Figure 6.

 Spectra such as these may be used to construct particle size

 plots as shown in Figure 7.

COUNTS
0 0
no OJ
10°

I04
10 3
t-
z
n
O 2
o 10
10°
BERMUDA AEROSOL 1-3 NOV. 73
II2-SI STAGE 4 23.1 HOURS
.•ill i 1 1 1 1 I
.'•'2c PROTON SCATTERING (PESA) -
Ep = 16.0 MeV
9L =120°
. ' Q =33^0
. . ;'. I6o
"f -vs
.- ?. ' "< « : • ""ft ? 'i
• • f.. -. • y N . -i ' •
•i ... >,,. ..-.•--, : :• r. ;r
-,...,-:. \^.. .. ..-..i... . • - -. •_.- -. ; /!
-.• ..:.:.'/ \ • •••-.• . > \'- -.-/.X /' ,» .-.- *.-.. J P.- ^.J /
MeV 12.4277 12.8840 14.3028 14.5112 14.8
12.6735 13.2401 14.3601 14.556
SCATTERED PROTON ENERGY (MeV)
I 1 1 jl 1 I I II 1
. Si -.S /Co
.» .'',ci '' X- RAY (PIXE)
•'. .'•'. '• r Ep=4.0 MeV
•. '. •" / Fe 9L = I35°
"." '-'•*• Ti - Q = 1.96 p-C
* ,
'"' ,"> '-p
\
*• * . • • •
"Aj^c . ' Zn
*•'..•;.. •' • •
''••"' "V" -_'.'• '•".*•«•.. .'v'.f.-fJ. \r'.. ".'...- -/••.'.'••••• .;'•
1 1. 1 1 1 1 1 _ 1 1 I
2.62 3.69 4.51 ' 6.40



337


                        X-RAY ENERGY (keV)


Figure 6.  Combined PESA-PIXE analysis of an aerosol sample
           taken with an impactor.
                             .13

-------
            10
         ro
            10
          en
          c


          c
          o
          c 10
          o
          c
          o
          CJ
             10
Bermuda Aerosol
        1973
Oct
                                       PESA
                                       PIXE
                           4321

                         Impactor Stage
    Figure 7.  Particle size plot from combined PESA-PIXE

              analyses.



        These examples demonstrate the practical usefulness


of PESA technique applied to both filter and impactor  samples.
                              14

-------
                           SECTION 6


           COMPUTER CODE FOR DATA EVALUATION - SCAT


        Central to any rapid, automated spectral analysis method


is a peak fitting and data evaluation computer code.  The codes


SCAT and REX have been simultaneously developed to provide the


best possible analyses for PESA and PIXE respectively.  Charac-


teristics of the programs are virtually identical with the ex-


ception of the shape of the continuum which underlies the peaks.


During this grant period more experience has been gained with


REX since it was in routine analytic use while SCAT was applied


to developmental spectra.


        These programs fit spectra containing peaks due to up to


50 elements.  Nonlinear least squares Gaussian fits proceed from


a library of parameters which describe the spectrum of each ele-


ment in terms of the peak energies, full widths at half maxima


and relative ratios of peak values.  For example, the carbon


spectrum consists of three peaks while the more complex Al


spectrum requires 15 to define it.  Following fitting, the areal

                2
density (mass/cm ) is calculated from cross sections determined


from standard (usually gravimetric) samples.  Since this method


is absolute, it need not be completely calibrated for all ele-


ments on every unknown determination but only checked with a


few standards at the beginning and end of an extended run.



                               15

-------
         An unusual advantage of this type of program is its


use to synthesize spectra and thus determine sensitivity


limits for assumed loadings on a particular type of backing.


This specific sensitivity determination is the only really


meaningful one since it accounts for both the underlying


continuum and elemental peak interferences.


         Typical standard spectra are shown in Figures 8 and


9.  Figure 8 shows peaks due to carbon, nitrogen and oxygen


obtained by bombardment of the plastic, Kapton.  Figure 9


shows the considerably more complex spectrum due to bombard-


ment of a thin aluminum foil.


         Five non-linear parameters allow necessary small


shifts in the energy calibration and full width at half


maximum necessary when fitting from a predetermined spectral


library.  An additional ten linear parameters are employed to


achieve the fit to the continuum lying beneath the peaks.


These parameters were developed by trial and error during


the extended period over which the code was being developed.


Three or four iterations   produce convergence in times of

                            2
one minute with an overall x  in the range 0.5 to 1.1.


         Output of the calculations is available from a line


printer or an x - y point plotter.  Aerosol concentrations


are printed for all elements in the library with 3a detection


limits indicated.  In the plot mode the raw data, fitted curve


and continuum background curve are all displayed.


         User input to operate the program is kept as simple
                             .16

-------
           RUN  1643 KRPTON
                 N
      '0    100   200  300   400   500  600  700   800   900  1000  1400
                             CHANNEL NUMBER
Figure  8.   Calibration spectrum using the plastic Kapton.
          RUN 3124 flL  FOIL
           100   200   300   400  500  600  700   800   900   1000  1100
                             CHANNEL NUMBER
Figure  9.   Calibration  spectrum from an aluminum foil.
                                17

-------
as possible.  It consists of only two cards to define starting



values of the non-linear parameters, one card for the run



numbers to be analyzed and cards to select which elements are



to be used from the library.



         With insights gained during the development of SCAT,



initial thoughts have developed on an even more powerful



approach.  In order to allow variation in peak shapes (due to



the detector or samples which are thick or non-uniform)  it



appears feasible to integrate the Gaussian distribution over



the line shape of the data instead of fitting the Gaussians



to the spectrum.  With this new approach a 6-function line



shape results in a Gaussian peak but a rectangular or trape-



zodial line results in distorted peaks resembling those due



to non-uniform or very thick samples.  This method has been



incorporated into an experimental program and initial results



are most promising.
                             18

-------
                           SECTION 7



                   TOTAL MASS DETERMINATION



        Elements measurable by PESA represent most of the mass



of a typical aerosol sample.  PESA was used to determine the



content of a filter-type sample and the total mass was compared



to that determined gravimetrically.



        Nuclepore filters 25 mm in diameter of 0.4 ym pore size



were used.  From 4 to 7 liters/minute was drawn through a cir-



cular spot on the filter 8 mm in diameter for 24 hours.  Flow



rates were measured every hour and the total volume of air



sampled was calculated.



        The filters were weighed on a Mettler balance accurate



to ± 5 yg in a controlled  (relative humidity 50%) atmosphere be-



fore and after the samples were taken.  Before each weighing



the samples were allowed to equilibrate for at least 12 hours in



the controlled atmosphere.  The total mass of particulate matter



collected was then obtained by subtraction.



        The samples were placed in a vacuum chamber and irradi-



ated with 16 MeV protons.  The scattered protons were detected



at 120° laboratory angle.  The resultant spectra were fitted



using an interactive program which used a straight line for the



background and assumed the peaks to be Gaussian.  Yields were



obtained in this manner for spectra taken on and off the spot
                               19

-------
                                                Total Mass Sample (3)
                                                3/iC  16 MeV protons
                                                Indoor,0.4/x Nuclepore
  10       100   200   300   400   500   600   700   800   900   1000
                                    CHANNEL NUMBER

 Figure 10.	Scattering  spectrum  for an indoor aerosol s_aj:tipl.e_*
  10

LU

lio3
o
CO

I"2
O


  to1
  10
                                               Total Mass Blank (3)
                                               3/xC  16 MeV protons
                                           C   Indoor 0.4/z Nuclepore
            100   200   300   400    500    600   700   800   900    1000
                                    CHflNNEL NUMBER
 Figure  11.   Background scattering spectrum taken  a  few milli-
              meters  distant from the spectrum of Figure 10.
                                  20

-------
containing the collected particulate matter.  The difference

thus represents yields from the aerosol particles only.  A

library of absolute cross section given in terms of counts per
     2
yg/cm  was used to calculate the cross sectional density.

        Scattering spectra for a 24 hours indoor air sample from

the Physics Department machine shop are shown in Figure 10  (beam

on sample) and Figure 11 (beam on filter immediately adjacent to

sample).   Large peaks are seen corresponding to elastic and in-

elastic scattering   C and   0.  In addition peaks due to   N,
9*7    OQ   "31: 07    o Q      Af\
*'A1,  °S,   '° Cl,   K and  uCa may be quantitatively evaluated.

The total mass  (see Table 2) observed by proton scattering was

647 yg while gravimetrically 998 yg was measured.  From past
   I
experience we would estimate that the unobserved elements

(hydrogen and elements above calcium) would contribute about

86 yg to make a total of 733 yg by summing weights of individual

elements.  The differences of 998-733 yg may be attributable to

water loss when the sample is placed in vacuum, but further work

is needed to clarify this situation.

        Comparison of the weight of a sample before and after

bombardment was hampered by the necessity of gluing the filter

to a holder before irradiating it.  Further experiments might be

able to circumvent this difficulty.
                               21

-------
           TABLE 2.  MASSES OF LIGHT ELEMENTS FOR AN




                     INDOOR INDUSTRIAL FILTER SAMPLE

ELEMENTS
C
N
0
Al-Si
S
Cl
K-Ca
MASS
(yg)
330±2
36±0.5
109±3
143±4
14±0.5
5±0.2
10±0.3
*
45
5
15
19
2
<1
1
NET
YIELD
125,833
6,266
8,834
1,137
837
556
1,233
Estimated mass of



unobserved elements
86
12
                               22

-------
                           SECTION 8



                     CARBON DETERMINATION



        Determination of the amount of carbon in a sample pre-



sents special problems.  Most filters and backing materials used



in air sampling work contain carbon as a major constituent.  In



order to minimize the error due to background subtraction,



either the filter must be very heavily loaded or the filter it-



self must be made thinner.  Heavy loading is an unattractive



alternative since in x-ray analysis it makes self absorption cor-



rections difficult to impossible for softer x-rays.  On the



other hand, thinner filters are within the means of present



technology and may even be purchased on special order.  Still



another approach would be the use of special low carbon content



filters for separate carbon determinations; however, the addition



of a separate system for a single element appears unwarranted



if the thinner filter method is developed.



        Proton scattering spectra for two types of thin com-



mercially available filters are shown in Figure 12.  Both spectra



were measured for the same number of incident protons  (Q = 2yC)



and show that the Millipore contains less carbon than the Nucle-



pore.  On the other hand, its use in aerosol analyses is made



far less attractive by its considerable nitrogen content.  A



similar situation is also found for possible impaction surface






                               23

-------
Nylon.  As shown in Figure 13 the Nylon contains sufficient ni-


trogen as to make it an unattractive substrate when nitrogen


determinations are desired.


        Examples of carbon measurements on aerosol samples are


shown in Figures 14 and 15.  Figure 14 is the scattering spec-


trum for 16 MeV protons incident upon a single-orifice impactor


sample (particles > 4jam aerodynamic diameter) for l.lm  of air


from the coastal region north of Los Angeles, California.  The


numbers above each of the peaks are the masses of the elements


in micrograms per cubic meter of air.  In this case, with a

                                                2
paraffin coated polystyrene backing  (z 100 yg/cm ), the carbon


in the sample represents 26% of the carbon present.  The amount


of carbon in the backing was measured by bombardment of the


backing at a location about 2mm away from the sample.


        A spectrum resulting in determination from bombardment


of a deposit on a carbon Nuclepore filter is shown in Figure 15.


In this case analysis of a six-hour summertime sample from


downtown St. Louis, Missouri, yields a carbon value of 16% of


the total carbon present and a ±14% statistical error.  Another


feature of interest in this figure is the overall broadening of


all peaks which is a measure of total mass loading.
                              24

-------
Figure 12.

PESA spectra
for Nuclepore
and Millipore
filters.
                       1000

                       500
§ 100
o
o
   50
                        10
                       500
                       100
                     O
                     O
                        50
                        10
                     PROTON  SCATTERING
                     8L-I20°  Ep-l6MeV
                     Q • 2fj.C         NUCLEPORE
                                                                  r•<•>'•••
                                                          L. 1 •• *i
                                                  500

                                               CHANNEL  NUMBER
                                                              600
                                                                          700
                                       25

-------
1 1685 Lfl ST-1 (PESO)




_j
UJ
2
Z
a
i
o
X
1—
2
O
Q







4000F

3500
3000

2500




2000
-
1500

-
m

m




-
X2 I2C
\ (total 2.3)










I2c'
* ,1
', 1
1 1
1000
500


o-
i j
1 /
/
W

i 	 r




ri. i
L J *
A. ^
sample 0.6


I6Q
15.

t




I3C
28^.1

I
TN
/ t^i
/ V\
















27Ai+28Si
14.
t

r
23Na
No 55*
* A/
s m**** i>* *•' 9^%h***M*i*^M\^ N*/^
-------
           IO4
           io
        LU
LU
Q_
in io4
           'O
        CJ
           io
           io
              500
               «
                             I6f)
                          /
                      c
                      .
Proton  Scattering
     Ep= 15.980 MeV  eL=l20°
            0=30-0
                                     Nuclepore - Blank
              600       700       800
                  CHANNEL NUMBER
               900
1000
Figure 15.
    Spectrum for Nuclepore filter  sample of sufficient
    thickness to permit carbon content determination.
                                27

-------
                          REFERENCES




1.   Cohen, B.  L.,  K.  C.  Chan,  L.  Shabson,  G.  Wedberg and



    H.  Rudolph.   Trace Element Studies at University of



    Pittsburgh.   In:  Proceedings  of the Third Conference on



    Application of Small Accelerators, Vol.  1, ERDA CONF-



    741040-P1, NTIS Springfield,  Virginia, 157-162 pp.
                             28

-------
                        BIBLIOGRAPHY



Anttila, Al. et al. 1972.  Use of Backscattering in Elemental



        Analysis.  Int. Jour. Appl. Rad and Isotopes 23, 315.



Cohen, L. and R. A. Moyer.  1971. Analysis for Impurities by



        Nuclear Scattering.  Anal. Chem 4_3, 123.



Jolly, R. K. and H. B. White, Jr. 1971.  Elemental Analysis



        by Elastic Scattering.  Nucl. Inst. and Meth. 97, 299.



Krivan, V. 1972.  Analytische Charakterisierung von Stoffen



        durch Strevung von geladenen Terchen.  Z. Anal,  Chem.



        262, 1.



Nelson, J. W. 1977.  Proton Induced Aerosol Analyses:  Methods



        and Samples in X-ray Fluorescence Analyses of Environ-



        mental Samples edited by Thomas G. Dzubay, Ann Arbor



        Science Publishers, Ann Arbor, Michigan.



Nelson, J. W. and D. L. Meinert. 1975. Proton Elastic Scatter-



        ing Analyses - A Complement to Proton Induced X-ray



        Emission Analyses of Aerosols in Advances in X-ray



        Analysis Volume 18 /  Plenum Press,  New York,  N.  Y.




Williams, I. E.  1973.  Quantitative Analysis of Elements by



        Proton Scattering.  Florida State  University Thesis




         (unpublished).
                             29

-------
                        PROTON SCATTERING FOR ANALYSIS OF ATMOSPHERIC PARTICULAR MATTER*
                               K. R. Akselsson, J. w. nelson, and J. W. Winchester
                                        The Florida State University
                                        Tallahassee, Florida   32306

                 Proton scattering has been applied to the problem of elemental quantitative analy-
            sis of air particulate matter.  Elements up through chlorine may bo resolved using 16
            MeV protons incident upon targets up to about 1 mg/cm2 in thickness.  Using the FSU
            Super FH Tandem Accelerator and a large area solid state proton detector, an analysis
            can be performed in several minutes.  Combination of this technique with proton induced
            X-ray emission analysis provides a means of quantitative analysis for all elements.
            These accelerator based methods are being applied to studies of the composition of air
            particulata matter in diverse locations such as St. Louis, Mo.; Los Angeles, Ca.;
            several cities in Florida; and Bermuda.
            "This research was supported in part by the U.S. Environmental Protection Agency,
            Grants R-803913 and R-802132, and the National Science Foundation for accelerator
            operations support.

                       (proton scattering; quantitative analysis; air particulate matter)
                     Introduction

     Direct measurement of the most abundant elemental
constituents of atmospheric aerosol particle size frac-
tions is important in relating aerosol particle physics
theory to the variation of chemical composition with
particle size.  The former depends on physical interac-
tions leading to particle coagulation and removal from
the atmosphere, which are functions of particle mass,1
and the latter is important in describing the transport
of specific chemical substances in the atmosphere, many
of which are of practical public health importance.2
The direct measurement of the most abundant constituents,
carbon, nitrogen, and oxygen, however, is difficult by
most analytical techniques and in practice is almost
never attempted in atmospheric research requiring large
numbers of separate elemental analyses.  Therefore,
full advantage has not been taken of our present under-
standing of the physics of aerosols in accounting for
the occurrence of chemical components of aerosol par-
ticles.

     The present research is directed toward developing
• sensitive and rapid technique for determining the
light elements up to sulfur and chlorine in non-vola-
tile, components of aerosol particle size fractions by
proton elastic scattering analysis, PESA.3"5  Elements
sulfur and heavier can be determined by proton induced
X-ray emission, PIXE,6"8 and the two methods in combi-
nation should provide a determination of all elemental
constituents which contribute significantly to Che
total aerosol particle mass.  The present paper pre-
sents evidence that this approach is feasible, and the
two techniques can be applied to the same atmospheric
samples by proton bombardment in a Van de Graaff accel-
erator.

                       Experimental

     The experimental arrangement is similar to that
used in our previous reports.3"5  A collimated proton
beam of 16 MeV is used as an optimum of high enough
energy to provide resolution of protons elastically
scattered by isotopes of adjacent elements up to chlorine
and of low enough energy to be stopped completely in the
available Si detector of 1500 micron depletion d«pth.
The detector used in the present study is 6 :rro x 50 cm
active area collimated by a 1 rm elliptical curve run-
ning the length of the detector, so as to accept a
narrow range of baekscattered proton energies from each
target nuclide at the 120° detector angle and still
have a large enough effective detector surface are« to



3SC3
3809

J3Z3


ZCCC

1503


ice;
«.

.
1 66









'Y

it
r:
it
|i
i t
\ / 1
! .' 1
W '>,
tee
i LH 5T-I (PESaj
X2 :













("a]
**^»!WW V»~S*S— -xA*^A
230 3aa »oe s:o =o

ZC '
I (roioi 23)
• sample 0.6
ij

"0
! 15.
! i
I 3.
bi!
jl
| ^1^'
r >„
! ^3ci
H j [if^ 40
' j'n? k "Na : '• \ 2.7 C0
!.Ay ? -r^o 3.5: -;S>,«
^^V Uv7^ ^'v
e 7ce ace see uoo u
                                                    CHANNEL NunBER
fig. 1  Proton scattering spectrum for 16 MeV protons incident upon an air particulate sample from the coast
north of Los Angeles, California.  This sample was obtained from 1.1 m3 of air using a single orifice cascade
impactor.  The numbers above each peak are the mass of that element in micrograjns per cubic meter of air.

-------
assure good counting efficiency for small aerosol par-
ticle samples in the target position.  The detector is
located outside the vacuum system of the scattering
chamber, and scattered protons pass through a 1/4-mil
({.25 |ffl) Mylar window and about 2 mm of air before
entering the detector, causing negligible loss of reso-
lution in comparison with other factors in the arrange-
ment.

                       Results

     Figure 1 presents the elastically scattered pro-
ton spectrum from a sample of aerosol particles
greater than 4 urn aerodynamic diameter collected during
« 16 hour period along a coastal location near Los
Angeles, September 1974.  The sample is stage 1 of a
cascade impactor9 operating at a 1 liter/min air  flow
rate, representing 1.1 m-3 total air volume, and the
•ample is supported by polystyrene film of thickness
•100 ug/on2,  figure 2 is an X-ray spectrum produced
by bombardment of the same sample with 3.75 KeV pro-
ton* and detection in a Si(Li) detector.  At the pre-
sent time the PIXE procedure is precisely calibrated8
permitting routine quantitative analysis, and the  PESA
procedure is undergoing calibration verification.  The
sample from stage 1 is large and has a diameter of
about 2 am and the beam was colligated to about 1.5mm.
Thus, not all of the sample was analyzed. The numbers
                            I
                             S
                               10
                                            Bermuda Aerosol
                                                let. 1973
                                                      PESA
                                       54321
                                          Impactor Stage
                   Fig. 3  Size fraction analysis of a Bermuda aerosol sara-
                   ple using proton elastic scattering analysis (PESA) for
                   C,N,O, and S, and proton induced x-ray emission analysis
                   (PIXE) for S,C1, and Ca.  Impactor stages S through 1
                   represent particles of equivalent aerodynamic diameters
                   0.25-0.5,0.5-1,1-2,2-4, and >4 urn, respectively.
                   30908
                   JS999
                   iseea
                             1473  LO ST-1
                          CI3.2        :»•
                                   Pe
                                  1.87
 Zn      Br
.038   .OS4
                           Hi  Ti   5  Cul   PbLa/
                            !  ;OI6  !|  CM.]   .202  PtiLp
                                          380
                                                      see   see
                                                    CMONNEL
                                                                   709    889    908   1998   11(10
 rig, 2  Proton induced X-ray emission analysis of the same air paniculate sample as shown in fig. 1.  Numbers
 above each peak are the mass of that element in mierograms per cubic meter of air.
 of mierograms of each element given in Fig. 1 have
 been calculated after normalizing the PESA Ca +• K
 value to that of PIXE, supposing the elements have
 the same distribution in the sample.  It ia seen that
 the sample size for PESA analysis is sufficient for
 detection of most elements from carbon to calcium and
 •till small enough not to cause proton energy disper-
 sion and loss of resolution.  The sample size is also
 adequate for detection of about ten elements by PIXE.

      Figure 3 presents the distribution of seven ele-
 ments with particle size for a sample collected during
 • 72 hour period in October 1973 by cascade iapactor
 from a tower in Bermuda.-0  This analysis, performed
 before the absolute PESA calibration vas quantified,
 represents concentrations of S, Cl, and Ca by PIXE and
 concentrations of C, tt, 0, and S by PESA with norma-
 lisation of S to the PIXE value all stages. Carbon
 values are approximate owing to uncertainties in poly-
 styrene backing thickness.  The particle size distri-
                   butions  show high  large particle abundances  for  Na.Cl,
                   and Ca,  which may  originate  from sea  water dispersion,
                   and high small particle abundances  for much  of the  N,
                   S, and 0, where composition  of  (NH^SOj  is  expected,
                   and for  some of the  carbon.  The combination of  PESA
                   and PIXE appears to  be valuable for presenting
                   enough elemental data to suggest chemical composition
                   relationships such as these.  PIXE  alone  would be
                   insufficient.

                         Because of the  special  interest  in measurements
                   of nitrogen and sulfur related  to the atmospheric
                   chemistry of these elements, test bombardments have
                   been  carried out on  targets  which could serve as
                   standards.  Figure 4 shows the  scattering spectrum for
                   Kapton film, a material with a  precisely  known atomic
                   ratio 0/N * 2.5.  The spectrum is  simple and the ele-
                   ments are clearly  resolved and  can  be readily analyzed
                   by suitable computer programs.  This  material provides
                   a convenient means to determine the relative N/0

-------
  IT
 Fig.  4  Proton  scattering  spectrum for  16 HeV proton*
 incident upon a 7.5  urn thick  Kapton film.
fig-  5   Proton scattering spectrum  for  16 MeV protons
incident upon a sulfur target.  The target was prepared
by vacuum evaporation of elemental  sulfur onto a Mylar
becking  and  then over sprayed with  Krylon to stabilize
the sulfur to bombardment in vacuum.

scattering cross sections.  Figure  S shows the scat-
tering spectrum for elemental sulfur on Mylar, with
Krylon binder.  The elastic scattering  cross section
for sulfur is dominant over lower energy inelastic pro-
ton groups in the spectrum.  Our identification of
peaks between channels 100 and 600  is not yet complete,
but a few inelastic sulfur scattering peaks are appar-
ently  present and may be useful in quantitative reso-
lution of spectra from complex mixtures as may be en-
countered in analysis of environmental  samples. Current
emphasis on high quality analyses of particulace sulfur
in atmospheric samples, because of  its  pollution signi-
ficance,  makes it desirable to have alternate methods
for sulfur determination.  FESA and FIXE can both be
applied  to sulfur determination in  the  same samples.

      High quality elemental analyses may be produced
providing the proton scattering cross section is not
highly sensitive to variables in a routine procedure
end providing that variations in saiipla composition and
thickness do not degrade the spectrum to the point of
not being.able to resolve the elemental constituents.
As exemplified by Fig. 1, impactor sampling time may
reedily be chosen for obtaining a sample of size large
enough for detection of elements by PESA and small
enough to permit their resolution.  It  should be pointed
out that  thick samples tend to cause proton energy dis-
persion,  not proton particle loss, and  some dispersion
can be tolerated if adjacent energy peaks can be resol-
ved,  tn contrast,  PIXE and any other X-ray methods for
elemental analysis  are vulnerable to x-ray attenuation
et low x-ray energies, and this makes determination of
elements lighter than sulfur uncertain if the required
 sample  self-absorption corrections  are  large.   Thus,  in
 many practical cases,  determinations  of elements in the
 region  of  sulfur  by both PIXE and PESA  offer a decided
 advantage  over determinations by one  method alone.

     In order to  define the  sensitivity of the PESA
 method  to  instrumental variables, measurements of effec-
 tive cross section with angle and proton energy have
 been made.  Figure 6 shows how the  relative cross sec-
 tion varies with  scattering  angle for 12C,160,  and  32S.
 (These  are relative values only.  The variation of effec-
 tive cross section at  120° with mass  number of nuclides
 ?Li to  *0ca has been given in Fig.  9  of reference 5.)
 Figure  6 indicates that the  sulfur  cross section varies
 imperceptibly over a broad angle interval from 100O to
 135O, oxygen has  a gentle minimum centered about 113°,
 and carbon decreases almot monotonically with  angle
 from 100°  to 135°, being relatively flatter around  12QO
 than angles greater or less  thin this.   For convenience
 in our  experimental arrangement a.id for the least prac-
 tical sensitivity to variations in  scattering  angle,  we
 have chosen 120°  for most of our further calibration
 experiments.   This choice of angle  also affords a rela-
 tively  high carbon cross section and  permits precise
 carbon  measurements where subtraction of backing mate-
 rial contribution is required.  We  also note that more •
 forward angles suffer  from apparently less satisfactory
 peak to background ratios, and the  peaks are kir.'emati-
 cally more closely spaced.   These and other practical
 considerations for the resolution of  elemental  consti-
 tuents  in  environmental samples indicate that  the opti-
 mum angle  chosen  for analysis is not  a  trivial  problem
 end should be explored with  great care.   At present,
 our choice of 120° appears to be a  good one.
        2 15
        I
          10
        r
                     Proton Scattering
                         Ep« 15.930 MeV
             100     HO      120      130
                   Scattering Angle (dig)
                                             140
Fig. 6  Differential cross section in arbitrary units
versus scattering angle for three isotopes.  The lines
through the points are guides for the eye.

     Figure 7 shows relations calculated which influence
the resolution which can be achieved for 12c and J5ci.
The angle subtended by the detector is determined by the
width of the elliptical curve which fonr.s she collimatcr
of our detector face, currently about 1°.  since the
intrinsic detector resolution is in the region of 40 keV.
it is desirable to keep additional energy dispersion due
to detector angle well below this value.  This is espe-
cially true if resolution of :7A1 and J8Si is to be
achieved, as we consider necessary in atmospheric
aerosol studies.
                                                  32

-------
  ISO
  160
• 140
JI2C
Joo
Jeo

!<°
s
                Proton Scottering
                AEviAS
                E?'I6 MaV
          '"01     2     3     4     5
                    Angli Subtendsd by Ottectcr
                         A9 (dsg)
Fig. 7  Variations  in  scattered  proton  energy versus
angle subtended by  the proton  detector.  In the  analy-
sis system at FSU,  angles  between  1 and 2  degrees are
employed.
        10
      - 4
      s  z
      5
              doYdfl vs Ep
                                   23,
                                     'No
         15.90   15.92
                        I5S4    15.96    15.98
                      Incident Prolon Energy (MeVI
                                               16.00
Fig. 8  Excitation  functions  (or  the ground  and  first
excited states of '^Na over tne range Of  interest  for
the FSO analysis system.  The  lack of structure  is
desirable in that quantitative analyses do not depend
upon sample thickness.  The line  indicates 30 keV,
the maximum sample  thickness  analyzable with good
resolution.

     Finally, Fig.  8 shows the differential  scattering
cross section as a  function of proton energy for i3Na
as measured from a  60 ug/cm*  Ma Cl sample. The uncer-
tainty in the repeatability of the proton energy scale
was ±10 keV.  Data  for the two major peaks observed in
the sodium spectrum, elastic  scattering from the ground
state and inelastic scattering to form the first exci-
ted state, are given for a range of energy.  This range
is several times greater than  we anticipate  will be
caused by extreme variations  in sample thickness which
can be accepted and still give the energy resolution
required for a successful analysis for individual
elements.  Normally, no thickness greater than that
equivalent to a 30  kaV energy  loss would  be  accepted,
and neither sodium  cross section is found to vary over
this range.  Also, it is seen  from fig. a  that tedious
energy calibration  procedures  of the incident proton
energy are  not necessary for sodium measurements.
Clearly, such tests must be carried out for  every
 element for which analysis is attempted.  For those
 we have completed to date, similar results have been
 found, indicating no difficulties are foreseen due to
 cross section sensitivity to this effect.  It is also
 fortunate that the ratio of cross sections for the
 elastic and inelastic groups of scattered protons from
 sodium is constant with energy, because then both
 groups can be used in the quantitative analysis for
 sodium in complex mixtures by straightforward computer
 fitting programs.

                      Conclusions

     Experiments performed to date indicate that proton
 elastic scattering analysis can be a practical, rapid,
 and sensitive means of elemental analysis when applied
 to atmospheric aerosol samples.  It has inherent sim-
 plicity, relative freedom from interference between
 elements present in the sample, and potential for
 automated nondestructive analysis.  In combination
 with PIXE, PESA has the capability for determining all
 elemental constituents of aerosol samples that contri-
 bute significantly to the total aerosol mass, a capa-
 bility which is unique in contrast to alternative ana-
 lytical techniques.  Future research and development
 of PESA should be directed to detailed cross section
 measurements of the nuclides of interest over the
 range of variables of energy, angle, and other para-
 meters encountered in practical elemental analysis. An
 effort should be made to develop computer hardware and
 software which will automate the procedure so that ana-
 lyses can be carried out in minutes or less of time
 and data can be handled readily for interpretation.
 Finally, the PESA technique should be field tested
 extensively so that relationships in elemental compo-
 sition of environmental samples revealed by the tech-
 nique can be evaluated.  Such evaluations may indicate
 directions for further improvement of the technique.

                      References

 1C. E. Junge, Air Chemistry and Radioactivity, Academic
 Press, Hew York, 1963.
 'T. B. Johansson and J. w. Winchester, Proc. 2nd Int.
 Conf. Duel. Meth. Environ. Res., Columbia, Mo. July
 1974 (In press. Tech. Info. Center, USAEC, Oak Ridge,
 Tenn.).
 3j. w. Nelson et al., IEEE Trans. Hucl. Sci. NS21.618
 (1974).
 4j. w. Nelson and 0. L. Meinert, Advances in x-Ray
 Analysis 18, (197S) (in press,Plenum).
 5j. w. Nelson et al., Proc. 3rd Conf. on Applications
 of Small Accelerators, Denton, Texas, Oct. 1974 (in
 press, USAEC Tech. Info. Center).
 «T. 8. Johansson et al., Nucl. Instr. £ Meth. 84
 (1970) 141.
 ?T. B. Johansson et al., Advances in X-Ray Analysis JL£
 p. 373 (Plenum Press, 1972).
 8T. B. Johansson et al,, (accepted Anal. Chen.).
 9R. I. Mitchell and J. M. Pilcher, Indus, and Engr.
 Chem. 5.1, 1039 (1959).
100. L. Meinert, U.S. thesis, The Florida State univer-
 sity, June 1974 (unpublished).
                                                         33

-------
NUCLEAR INSTRUMENTS  AND METHODS 142  (1977)  251-257 ;  ©  NORTH-HOLLAND  PUBLISHING CO.



     REX - A COMPUTER PROGRAMME KOR P1XE ANALYSIS*



     HENRY C. KAUFMANN

     Department of Physics. Florida Slate University. Tallahassee. Fla 32306. U.S.A.

     K. ROLAND AKSELSSON

     Department of Environmental Health, University of Lund. 22362 Lund, Sweden

     and

     WILLIAM J. COURTNEY

     Department of Physics, Florida A&M University. Tallahassee, Fla 32307, U.S.A.

     The implementation of a physics-based model for  non-linear least-squares analysis of proton-induced X-ray emission
     spectra via a Fortran programme  REX is discussed. The modelling of distinct  spectral components and of the physical
     effects involved is briefly discussed. Results are presented in graphical and tabular form. The limitations of the present
     model are discussed and future refinements indicated.
1. Introduction
  Proton-induced X-ray  emission analysis (PIXE)
has been shown to be a  fast, inexpensive, reliable
and convenient method  for routine multi-elemen-
tal trace analysis'). The main divisions in such an
analysis are: sample preparation and bombardment
and decomposition of the measured  pulse-height
spectra. Intense work  is going on  in all these fields
to find suitable, time-saving sample preparation
techniques, to build bombardment and data-taking
systems  which are fast, reliable and automated
and to develop inexpensive,  reliable and automat-
ed  computer programmes  for  resolving   PIXE
spectra.  Kaufmann  and Akselsson2) and  Kauf-
mann et al.3)  have earlier described  versions of a
computer code,  REX, based on a model  of the
physics in PIXE-analysis. This report  describes the
current  version  of REX, some  results  obtained
with it  and  a  discussion of further improvements
which  can be made.
  The  parameters  of  the model are found using a
least-squares minimization based on a technique
developed by  Kaufmann4).

2. The model
  The  radiation  environment due to the photons
present  during proton bombardment  of a sample
may be deduced from the pulse-height spectra ob-
tained from a  Si(Li) detector. The photon  radiation
spectrum consists  of  discrete quanta  due to char-

• Research supported in pan by US Environmental Protection
  Agency, Grants R-803913 and R-803887.
acteristic X-rays and a continuum. The continuum
proper has its origin in two phenomena: (1) brems-
strahlung of secondary electrons and of protons2'5)
and (2) Compton-scattered electrons from  high-en-
ergy X-rays and y-rays which originate in or  near
the sensitive volume of the detector. For typical
samples, the bremsstrahlung component  is  domi-
nant: however, the sample  thickness and compo-
sition  and the  amount of absorptive material be-
tween  the sample and  the  detector can alter the
relative  contribution  of  the  Compton  electron
component. To  the  first order, both phenomena
scale with sample thickness  and, while the brems-
strahlung component will  be  attenuated by  any
absorptive material present,  the Compton electron
component  will  not.
  The putative  Compton electron component is
modelled  in a fashion which implicitly allows for
phenomena such as  bremsstrahlung escape  and
backscattered photoelectrons. Both of these  con-
tributions to the continuum  are more properly  mo-
delled  in the detection process.  Preliminary at-
tempts at implementing a more precise model for
these components will be deferred to a later  sec-
tion. The model  discussed  here does not include
the silicon escape peaks for the characteristic ra-
diation: inclusion of these in the  model  presents
some difficulty as will  be discussed later.
  For  the case in which the size of a thin sample
is smaller than the area of  the proton beam,  the
yield of a characteristic K. X-ray transition is given
by
I = nNacokTi T2sfl/(4n),                      (1)

             VI. DETECTORS AND ELECTRONICS
                                              34

-------
                                      H. C. KAUFMANN el al.
where n  is the proton flux in particles per cm2, N
is the number of atoms of the elements, a is the
ionization cross  section for the  K.-shell, w  is the
fluorescence yield for the K-shell, k is the branching
ratio for the emission of the characteristic X-ray, r, rs
the transmission between the sample and the de-
tector, TI is the  mean  transmission  for X-rays
through  the  sample itself, e is  the detector effi-
ciency and Q is the solid angle subtended  at the
detector  by the  collimator.
  The proton flux, n,  is deduced from the dead-
time-corrected integrated charge and the values for
er, w, and It are taken from refs.  6, 7 and 7 re-
spectively.
  The factors JT,, T2 and £ in eq. (1) depend on the
X-ray energy as  do the continuum components to
be discussed  later. Thus, it is most convenient to
express the model in terms of "equivalent" X-ray
energy rather than as  pulse-heights. The "equiv-
alent" energy corresponding to a particular  pulse-
height is taken  here  to  be  the energy of  fictive
monoenergetic X-rays which would give their full-
energy peak  at that pulse-height.  The parameters
of  the  transformation   of  the  spectra   from
pulses/channel to pulses/energy  interval are in-
cluded in the least-squares fit.
  The form  of the transformation depends on the
characteristics of the pulse-height analysis system.
For optimum speed, a linear calibration  is crucial
and  in this case the transformation is given by
£ =  p, + p2(x-;0),                          (2)
where p, and p2 are parameters,  x is the channel
number and j0 is a fixed channel number chosen
to optimize  the  calculational procedures.
  When  applied to a thin sample on a thin back-
ing,  the  model gives the following expression for
the yield of  pulses per energy unit:
f(P, L, E) = CON 1 (P, E) + T, (P, E) • T2 (P, E) •

         • CON2(P, £)  + £ L(I) • T, (P,  £(/, J)) •

         •T2(P,  £(/,J))-PEAK(P, £(/,./)),     (3)
where / is the number of pulses per energy unit,
P is  an array of  non-linear parameters, L is  an ar-
ray of linear parameters,  T\(P, E)  is the  trans-
mission  through  absorbers   external  to  sample,
Ti(P. E)  is the  mean  transmission through  the
sample material, CONl(/>, £)  is  the  continuum
unaffected to the first  order by  absorption effects
such  as,  e.g., from Compton scattered electrons,
CON2(/>, E) is the continuum affected by absorp-
tion  such as, e.g.,  bremsstrahlung from secondary
electrons and PEAK(/>, £(/, J)) is the characteristic
X-ray  line shape convoluted with  a detector  re-
sponse-function.
  The indices / and  J refer to the atomic number
of the element, e.g.  Fe, and  to  specific character-
istic X-ray transitions of the element, e.g. K.,,, re-
spectively. The  parameter arrays,  P and  L,  are
found  by minimizing the function
                                             (4)
where  y,  is  the number  of pulses  in  channel
/, w, = 1/y, if y,>0, otherwise y, = 1 and d£/d* is
the derivative of the transformation from channel-
space to  energy-space. In  the  case of a  linear
transformation  [eq.  (2)], the component  d£/dx is
folded  into the  linear  parameters and is thus  in-
cluded in /.
  It is the simultaneous fitting to all the model
parameters describing the entire pulse-height spec-
trum  which  distinguishes   the  present  approach
from others.  As is illustrated by  the fitted  data
shown later, this approach provides a versatile  so-
lution to such common problems as element inter-
ferences  and the use of unorthodox absorption
schemes.
3. Discussion of the components in the model

3.1. CHARACTERISTIC X-RAY TRANSITIONS
  Although  the  detector  response to a  X-ray
transition is better  described  as  the  sum  of a
Gaussian and  a  low-energy electron  backscatter
tail8),  to  minimize computing-time, we have chos-
en to  approximate the system response by a Gaus-
sian without tails.  The  full  width at  half maxi-
mum  (fwhm) of this distribution is dependent on
the energy of the detected radiation. In REX, this
dependence is taken to be of the form
fwhm-p, [£/£«]",                           (5)
where p} and p4 are suitable parameters and £0 a
reference energy  which serves the same purpose
as y» in eq.  (2).
  The intensity of the X-rays from each element
are parameterized with a single amplitude and the
different transitions for each element are scaled to
this parameter using known  relative  intensities.
This procedure lessens  the  ambiguity  in dealing
with  interfering  transitions  from  different  ele-
ments. Transitions  from a particular element are
included individually if they are separated by more
than 50 eV.
                                           35

-------
                                                REX
  Obtaining reliable measurements of the elemen-
tal masses based on the yield of K X-rays depends
mainly  on  the  accuracy of  the  elemental stan-
dards. For L X-rays, the relative intensities are sig-
nificantly dependent on  the proton  energy.  Thus,
elemental standards for  L X-rays  have to be ana-
lysed at the same energy as the samples. For thick
samples, these variations in  the relative transition
probabilities give rise to complications which can-
not yet be  resolved using REX.

3.2.  CONTINUUM
  The continuum component not  affected by first-
order absorption,   CON1,  is  mainly  caused  by
Compton scattered electrons passing  through the
sensitive volume of the detector. Other processes
are,  however, also  involved  and it is  not straight-
forward to model this component from basic phy-
sics of the process.  The following empirical expres-
sion has been found to give good  fits:
CON1  = p11/E + pl2-r-p13z + p14z2 + p15z3,   (6)
wherez = x-y'i,x  is the channel number and J,  a
reference channel chosen to give a suitable range
for the numbers in the calculations.  In spectra
having no peaks  above 15 keV, the parameters pti,
PH and p]S are held fixed at zero.
  The bremsstrahlung  component has been  thor-
oughly  investigated5). However, to save  comput-
ing-time, we have used the approximation
CON2 = exp(-pt,z)-(p7 + p8z + p9r2 + p10z3),   (7)
where z=x-jt,x is the channel number and y'2  a
reference  channel. Empirically this  expression is
found to give  good fits.

3.3. TRANSMISSION
   In earlier versions of REX2-3) four nonlinear pa-
rameters were used  to describe the  transmission
through the sample itself and through external ab-
sorbers. Since each nonlinear  parameter adds sig-
nificantly to the computing-time and since  three
of these parameters did not vary between  runs, we
have remodelled the transmission factors.
                           Blank Nuclepore
                           I mg/cm*.0.4^.
                           Absorber: None
                           Hole:-
   ' 0      100     200     300     100      SOO
                      CXI1KNEL NUMBER
                             Blonk Nuclepore
                             Img/cm1.0.4^4
                    Ft        Absorber :350/im Mylar
                             Hole: None
                                                                                   \,	 .
                                                                                             V-V:^STI
 10 0      10o     200      300     UOO      500
                     CHANNEL NUMBER
                           Blank Nuclepore
                           I mo/cm* ,0.4 fj.
                           Absorber :350|im Mylar
                           Hole: 9%
                  200     300      KOO
                     CHPNKEL NUMBER
                                        SOO
Fig. 1. Fits to proton induced X-ray emission spectra obtained from bombarding nominally blank Nuclepore niters using a variety
of detection schemes. Note in particular the fit to the spectrum obtained by using an absorber with hole (bottom).

                                                                 VI. DETECTORS AND ELECTRONICS
                                               36

-------
                                       H. C. KAUFMANN et al.
  The external absorber is  a  multi-component ab-
 sorber consisting  of foils to  maintain  vacuum in
 the  sample chamber and the detector housing, of
 the  gold layer and the dead-layer of the detector,
 of air and of an optional  absorber. The "trans-
 mission"  through this system  is modelled  by
 T, = £TOP,  = exP(fl£1') [1 - exp(C£')]  Top,,       (8)

 where E  is the  detector efficiency and  7"opl the
 transmission  through  an  optional  absorber.  The
 contants a, b, c, and d in the expression for the de-
' lector efficiency9) are  initially obtained by a least-
 squares fit and are  then held fixed.
  The mean  transmission through  the sample it-
 self is modelled  by
T2 = [1 - «p{-ji(E)ps}]/&i(E)p,],            (9)

 where //(£) is the  absorption coefficient of the
 sample matrix. For certain standard matrices, /j(£)
 is tabulated in REX.
  The optional  absorber may  be  a  solid  Mylar
 sheet. However,  in the usual  case with  samples
predominantly composed of low  Z elements but
with traces  of high  Z elements, an absorber with
a small hole may be an optimum. This type of se-
lective absorption  is easily introduced without ad-
ding more parameters by writing
ropl = (l-W)exp{-/((£)rf} + H,             (10)

where  H is  the fraction  of the solid angle Q sub-
tended by the hole, n(E) the absorption coefficient
as a function  of  the  X-ray energy E and  d the
thickness  of the absorber.

4. Results
  The  coding  of  the  model in  the  Fortran pro-
gramme REX has been  undertaken with speed of
calculation and memory  requirements being of pri-
mary concern.  At  present the programme is com-
piled  in 42k  of memory  in  a Datacraft 6024/3
computer. Execution time per spectrum evaluation
varies  between 20 and  180s  depending on  the
quality of the starting  parameters  and the number
of fixed parameters.
10S

u
10

a
103
102

..1
Nelson Streoker
c. Absorber: None
« Hole:-
* j A
T A ( .
few ,
J' --U' '
7 -V p
' ^..
'^ LA A <»>,,„»,,,
' "^"f-^-s^^'i^
10S

u
d
i
U 3
COUNTS/
5M °c

,
ca
I

:|V^
V '
J


Nelson Streaker

'• Absorber : 350 urn Mylar
J! Hole: None
,|
W ' ™
"^^y '. w
\^ /; [; tm
^5" i '; A '':
'V1 'lI'lV*1"
XiB i to (* .*,
x^/uV<






bH
Zr
^ **


 5  3
 5 to3
  10 o
          100     ZOO    300    HOO     SOO    600
                      CHANNEL NUMBER
                                                                       CHANNEL NUKBER
                            Nelson Streaker
                            Absorber: 350/j.m Mylar
                            Hole: 9%
                       300    100
                      CHANNEL NUMBER
 Fig. 2. Fits to proton induced X-ray emission spectra obtained from bombarding an aerosol sample obtained with the Nelson
 Streaker. For quantitative comparison see table 1.
                                           37

-------
                                               REX


TABLE I
Comparison of results from runs using different absorbers. The sample analyzed is from 2 mj air and the count rate in all three
analyses was lUOOcps analyzed for 500s each.
Element
Al
Si
P
S
Cl
K
Ca
Ti
V
Mn
Fe
Ni
Cu
Zn
Pb
Br
Sr
Zr
Absorber: None
Hole: —
Amount Error MDL"
(ng/cm2) (%) (ng/cm2)
1914
6743
426
1633
233
557
3300
133

24
1404
23
36
74
556
120
41
45
53
39
40
32
23
20
19
17

20
14
19
15
15
16
14
19
21
205
53
36
21
18
13
12
11
11
11
7
7
6
6
27
6
10
15
Absorber: 350 //m
Hole: None
Amount Error
(ng/cm2) ('*)



543
3486
165
12
33
1625
10
28
69
693
119
44
26



20
19
17
25
16
14
17
13
14
16
14
14
15
Mylar
MDL"
(ng/cm2)


1030
32
13
6
5
3
2
2
2
2
5
1
2
3
Absorber: 350 tim
Hole: 8.9'Ai
Amount Error
(ng/cm2) (%)
2013
5740
203
1314
90
504
2893
135

26
1338
9
27
61
573
110
42
30
37
32
36
30
24
19
19
16

16
14
19
14
14
16
14
14
15
Mylar
MDL"
(ng/cm2)
231
59
40
24
20
14
11
7
7
5
3
3
2
2
8
2
2
4
  Minimum detection limit.
  The quality of the present model  may best be
assessed by examining fitted spectra from a varie-
ty of detection and sample conditions. In this sec-
tion,  we present such an assessment.
  The continuum model  was tested  by obtaining
photon spectra from a nominally  blank hydrocar-
bon filter material such as Nuclepore. In fig. 1  fit-
ted spectra for data obtained using the three com-
mon  absorption schemes in the aerosol research
programme  at FSU are  shown. The continuum
model provides  a  good description of the data in
all three cases.
  A test  of the complete model  on real data is
presented in fig. 2. The  data are from  the proton
bombardment of an aerosol sample obtained with
a  continuous  filter sampler10). Once  again  the
spectra were obtained using  the three common  ab-
sorption  schemes  and  are from bombardments of
the same spot on the  filter.  As a  further illustra-
tion of the overall versatility and reliability  of  the
analysis, the masses obtained for each element in
the three analyses are tabulated in  table 1.
  Careful study of the figures and table  1  indi-
cates  some of  the inadequacies  of the present
model. The silicon K-shell edge absorption affects
the measured values of P, S, and Cl significantly.
The fitted  spectra deviate  from  the  functional
form of the data at both the lowest and the high-
est energies in the spectra. However, the analytical
errors associated with these effects are comparable
to the instrumental systematic errors.
  The ease  with which  elemental  interferences are
handled is illustrated in fig. 3. In the  upper part
of the figure is displayed  the result of a  fit to  a
PIXE analysis of an aerosol sample from  Chacal-
taya, Brazil") in which As has been omitted from
the element request list. The fit for the Pb X-rays
is seen to be very  poor. The lower figure includes
both As and Pb in the request list and  the result-
ing fit is excellent. The  element library and the si-
multaneous fitting of all model parameters greatly
facilitates  the correct accounting  of element  inter-
ferences.

5. Discussion
  We have  presented a  status  report  describing
the continuing  development of  a programme  to
obtain reliable quantitative measurements  of trace
elements  using proton  induced  X-ray  emission.
The model used is based on the  physics of X-ray
             VI. DETECTORS AND ELECTRONICS
                                             38

-------
                                        H. C. KAUFMANN  et al.
          100    200    300    <400     500    600    700
                               Chocoltoyo .Bolivia
                               Atsi»be.:J50,mM,lar
                                  Hol<:9%
          100    200    300    100     SOO    600    700
                     CHBNNEl. NUHBER
Fig. 3. Fits  to a spectrum from an aerosol sample without (up-
per) and  with (lower) arsenic in the request  list.

production  and detection. The data  presented at-
tests  to the overall  suitability of the  model and of
its  implementation.
  Pointers to future refinements of the model are
shown  by the inadequate accounting of the  con-
tinuum as  can be seen in  the fitted data  in  figs.
1 and  2. At  present, the  errors  associated  with
these errors are comparable  to the errors associated
with  the instrumentation  and  do  not  therefore
greatly  lessen  the accuracy of the analysis.
  Some of the difficult problems  which must  be
solved are displayed in fig. 4. The upper figure is
a spectrum from a  long  accumulation  of the X-
rays from a 55FeCl2 source.  The spectral properties
which we do  not explicitly include  are (a) the Si
K X-ray  escape  peaks and (b)  possible radiative
Auger emission12). Although we can, in principle,
include the escape peaks  in the  present model, in
practice the inclusion is made difficult by  the ex-
cessive width of the escape peak compared to that
of a  full  energy  X-ray of  the same energy.  The
radiative  Auger transition is very weak [a 0.5% of
the  intensity  of the ^-transition12))  and   thus
should  only present  an  analytical problem when
small amounts of  a lower Z element  are to  be
                                                           55FeCl2 Source
                                                           RATiRodKJtive Auge' Transition
                                MnK.
                                                        1 0    100   200   300    UOO   SOO    600   700   800
                                                                          CWWNEL KIM8ER
      Lanthanum Standard
      UaorOa-.U ea fm
                               LoK.
                                                       10
                 300   WOO   500
                     CHANNEL NUMBER
Fig. 4. Two spectra showing features which the current ver-
sion of REX is unable to fit. These are silicon escape  peaks
(ESC) and radiative Auger emission (RAT) in the  upper spec-
trum and the low-energy shoulders on the peaks and the  broad
bump between channels 150 and 500 in the other spectrum.

measured in the presence of an abundant interfer-
ing element  of higher Z.
  The lower part of fig. 4 shows a spectrum taken
during proton bombardment of a La standard. The
La L X-rays have been absorbed so that the detec-
tor response to the K X-rays  can clearly  be  seen.
In this particular case, the causes of the shoulders
on  the  low energy sides  of the  peaks and of the
rise  of the continuum  at energies below  about
channel  450 are not unambiguously known.
  We hope to resolve  some of these questions by
measuring the detector response  to  monoenergetic
photons  obtained by single-crystal diffraction.
References
 ') T. B. Johansson. R. E. Van Grieken, 1. W. Nelson and }. W.
   Winchester. Anal. Chem.  47 (1975) 855.
 2) H. C. Kaufmatm and K. R. Akselsson, Advan. X-ray Anal. 18
   (1975)353.
 3) H. C. Kaufmann. K. R. Akselsson and W. ). Courtney, Ad-
   van. X-ray Anal. 19(1976)355.
 4) H. C. Kaufmann, to be published.
                                               39

-------
                                                      REX


5) F. Folkmann.J. Borggrenand A. Kjeldgaard, Nucl. Instr. and    ') H. M. Schupferling, Nucl. Instr. and Meth. 123(1975)67.
  Meth. 119(1974) 117.                                      ') W. J. Gallagher and S. J. Cipolla. Nucl. Instr. and Meth. 122
6) K. R. Akselsson and T. B. Johansson, Z. Physik 266 (1974)      (1974) 405.
  245.                                                    I0) i. W. Nelson, B. Jensen, G. G. Desacdeleer. K. R. Akselsson
7) W. Bambynek, B. Crasemann, R. W. Fink, H.-U. Freund, H.      and J. W. Winchester, Advan. X-ray Anal. 19 (1976) 403.
  Mark, C. D. Swifi. R. E. Price and P. Venugopala Rao, Rev.   ") F. Adams, private communication.
  Mod. Phys. 44 (1972) 716.                                 12) G. Presser, Phys. Lett. S6A (1976) 273.
                                                                          VI. DETECTORS  AND  ELECTRONICS
                                                     40

-------
NUCLEAR  INSTRUMENTS AND METHODS  142 (1977)  127-132  ; ©  NORTH-HOLLAND  PUBLISHING CO.



      LIGHT ELEMENT ANALYSIS BY PROTON SCATTERING



      J.  W.  NELSON*

      Department of Physics. Florida Slate University, Tallahassee, Florida 32306, U.S.A.

      and

      W. J.  COURTNEY'

      Department of Physics. Florida A & M University, Tallahassee,  Florida 32306.  U.S.A.

      A  system for quantitative elemental analysis by proton scattering at 16 MeV has been developed. Samples of thicknesses
      up to  1 mg/cm2 may be analyzed for the light elements up  to Cl. Examples of spectra for atmospheric paniculate matter
      and biological specimens are shown.
1. Introduction
  Proton induced  X-ray emission analysis (PIXE)
may be used  to perform quantitative analyses  of
any element on many types of samples.  No single
system design is optimum for measurement of  all
elements in samples ranging from very thin to in-
finitely  thick  to the proton beam.  Using  proton
beams of  the  Florida  State  University  Tandem
Van de Graaff accelerator, we  have developed  an
energy dispersive  PIXE  analysis  system1) for  thin
solid  samples (thicknesses  up to  approximately
1 mg/cm2). Attenuation of the  softer X-rays  in
both the sample and windows, limits the useful-
ness of our system to analyses  of elements having
Z>13.  Proton elastic scattering analysis  (PESA)
has been developed for light elements in order to
complement our PIXE technique.
  While  quantitative PIXE analysis may  be  per-
formed for even the lightest elements, absorption
by the sample itself imposes severe limits on the
useable thicknesses of solid samples.  In planning
our system primarily for determinations on air par-
ticulate matter,  individual  particle  linear  dimen-
sions up to 10-50 nm were of interest.  For selected
elements the thicknesses needed  to  reduce the in-
tensity of a beam  of X-rays by one-half as a func-
tion of photon energy are shown  in fig. 1. For par-
ticles in the 10/im region attenuation is seen to  be
a major correction in quantitative  measurements.
For this reason proton scattering was  investigated
as a complementary  method for analysis of  light
elements.
• Research supported in part by EPA Grant R802913030.
2. Proton scattering analysis
  A light element method to  supplement  PIXE
analysis should possess similar characteristics and
sample requirements as PIXE. Proton elastic scat-
tering analysis has properties  which make it quite
complementary. It is absolute, multi-element, rap-
id,  subject to automation  and non-destructive of
the prepared sample.  Sample thicknesses up  to
about 1 mg/cm2  may  be  used.  Unlike the  X-ray
analysis in which too thick  a  sample attenuates
but does not degrade the quality of the spectrum,
PESA spectra exhibit loss of resolution which  im-
mediately signal that sample thickness is too large.
Although nuclear  scattering  cross  sections  are
several orders of  magnitude smaller than  those for
X-ray production, comparable analysis times (min-
   100
 in
 c/)
 LU
 (J
 I
 <
 I
    10
                                                      0.1
      I         2.3        4       5

               PHOTON ENERGY (KeV)
Fig. 1. Thicknesses needed to reduce the intensity of a beam
of X-rays to half value as a function of photon energy for
selected elements.

                        IV.  USE OF  HEAVY  IONS
                                               41

-------
                                     J.  W.  NELSON  AND  W.  J.  COURTNEY
    10'
      4
     10'
                    K X-RAYS
                    Ep 5MeY
L X-RAYS
Ep SMeV
                                                                                          M X-RAYS
                                                                                          Ep SMeV   '
                     Elastic  Protons
                     Ep  16 M«V
                     6L  IZO-
     IO°LI  I  I  I  I  I  I  I  I  I  I  I   I  I  I	I  1  I  I  I  I  I  I  I   I  I  I  I  I  I  I  \-\
         2    6    10   14   18   22   26  30   34   38   42
          B«   0    Ma   S
                                        ATOMIC  NUMBER
         50M58626670747882
                Br                            n
Fig. 2. Overall efficiency of the Florida State  University combined PESA-PIXE quantitative analysis system.
    16
 I
 u  ,4
 §
 513
                                           PROTON SCATTERING
                                                                          Ep'ICMeV
         I3S-
                CSkiV
                                                                                            16
                                                                                            14
                                                                                            13
      0      20     4O      6O      80      KDO      120      140      160      ISO     ZOO
                                     SCATTERING  ANGLE (DEC)

Fig. 3.  Kinematics of 16 MeV elastic proton scattering for some of the light elements.
                                                      42

-------
                                    LIGHT  ELEMENT ANALYSIS
utes) are achieved on air paniculate samples. This
is in part due to the presence of larger amounts of
many light elements. In fig. 2 the overall efficien-
cy of the  FSU system2) is  shown. Only  those ele-
ments for which the production cross section has
been determined for our system are shown. The
ordinate is not the cross section as usually defined
but  the readily useable number of outgoing parti-
cles per //C  of incoming  protons  per ^g/cm2 of
sample.
  Values  of elastic scattering  cross sections are
generally an  order  of magnitude higher than the
inelastic cross  sections. Thus elastically scattered
protons were  chosen  for  analysis rather  than
inelastic groups or  the gamma rays  accompanying
inelastic scattering.  Further choices of the incident
proton energy and angle of detection were dictated
by other considerations. An incident proton ener-
gy  16 MeV was selected on the basis of depend-
able accelerator operation (1 MV below  maximum
rated voltage), greater  energy difference between
adjacent mass  nuclei, and  smallest  energy  loss in
the  sample.  In fig. 3 the  kinematics of  16 MeV
elastic proton scattering are presented. Selection of
a rearward scattering angle- is necessary  for a suf-
ficient energy  difference  to  permit resolution  of
target mass 27 from mass 28. Assuming  an overall
system  resolution of 30-40 keV is  maintained, a
scattering angle of 135° will result in a 65 keV sep-
aration  to allow determination of both  aluminum
and  silicon. From fig. 3 protons scattered  at 135°
from elements carbon through  chlorine  will have
energies between  12  and  15 MeV and may  be
  100
>
«
i 80

o:
UJ
z
UJ
z 60
? 40
<
5 20
Ep=l&OOMeV
8
      01234
      ANGLE SUBTENDED BY DETECTOR(DEC)
Fig. 4. Energy variation of scattered protons versus the angle
subtended by the detector.
stopped in large area (300 mm2) 1500 yum depletion
depth detectors.
  In the desire to attain the highest counting rate
consistent with 30-40 keV  overall resolution,  the
energy broadening contribution due  to  the angle
subtended by the detector  was calculated  and is
shown in fig. 4. At a scattering angle of 135° we
utilize detector collimators which subtend l°-2° in
order to constrain this source of loss of resolution
           RUN 1643 KRPTON
       0   100  200  300  400  SOO  600  700   800   800  1000 1100
                             CHRNNEL NUMBER

Fig. S. PESA spectrum Tor Kapton plastic.
                                                                           IV.  USE OF  HEAVY IONS
                                                  43

-------
                                J. W.  NKLSON AND  W. J.  COURTNEY
to about 20 keV. A further practical consideration
is the mechanical  problem  of milling  elliptically
shaped collimators for large area detectors which
subtend  a significant fraction of the cone of equal
energy outgoing  particles.  From the standpoint  of
the kinematics alone, the  most extreme backward
scattering angle approaching 180° is desirable.  Me-
chanical problems  of target  chamber construction
with  detector outside the  vacuum chamber influ-
enced us not to use  this  approach.  On the other
hand the 90° scattering angle offers the possibility
of the use  of linear  shaped collimators but  at  a
cost of less  kinematic energy  separation of adja-
cent  masses (see fig. 3).
3. A particular PKSA system
   One  system has  been developed  at  FSU using
 16MeV incident protons and  detecting  scattered
 protons at 120° with respect to  the incident proton
 direction. Typical PESA spectra are  shown  in figs.
 5 and 6.  In  fig.  5 the spectrum of  Kapton plastic
 is shown. Referring to fig.  2 it may be  observed
 that carbon  and  nitrogen are accentuated relative
 to oxygen  at  this  scattering angle and incident
 energy. Thus a suitable evaluation computer code
 must  be developed  to  account for the relative
 cross  sections. Such  a  program  is in  the  final
 stages of development and is closely patterned  af-
 ter our PIXE  analysis program  REX3). For exam-
           100   200   300  UOO  SOO   600   700  800  900  1000  1100
                             CHPNNEL  NUMBER

Fig. 6. PESA spectrum for the element Al. This is the most complex spectrum among those of the elements lighter than calcium.
  62500
  40000
 ; 22500
  j 10000
   2500
          GM- STREAKER 15-D
                                            xs
       0   100  200   300   MOO   SOO   £00  700  800  900  1000 1100
                             CHANNEL NUMBER

Fig. 7. PESA spectrum obtained by bombarding a Nuclepore niter paniculate sample of relatively clean winter air from rural
Michigan.
                                            44

-------
                                    LIGHT  ELEMENT ANALYSIS
pie I2C  is characterized by both the elastic peak la-
belled in fig. 5 and the first inelastic group just be-
low  channel number  100. Similarly  UN  is charac-
terized by the labelled elastic peak and that due to
the proton group to the second excited state which
occurs just  above channel  number 200.  Although
the majority of light elements  have spectra which
are simply described, the most complex case is de-
picted  in  fig.  6. The 27A1 spectrum requires  about
IS Gaussian peaks for its description. This is simi-
lar to the situation  with  the L and  M X-ray lines
of Pb which need 17 Gaussian peaks for their de-
scription in our PIXE analysis program. It should
be  noted that in cases such as  aerosol  analysis
only the dominant- elastic "Al peak  will be  sig-
nificant.

4. Applications
  The PESA system was  designed primarily for
analysis of air paniculate samples. An example of
this application  is shown as  fig.  7 which is  the
scattering spectrum obtained by bombarding a Nu-
clepore  filter paniculate sample of relatively clean
winter air from a  rural area of southern  Michigan.
           1685 LA ST-1  (PESO)
                                             total 23)
                                             sample 0.6
                                   see    see   7ee   see    see   leee   nee
                                 CHONNEL NUMBER
Fig. 8. PESA spectrum of a size fraction of paniculate matter from a site near Los Angeles, Ca. The numbers above the peaks
are the mass of that element in micrograms per cubic meter of air.
      0    100   200   300    <400    SOO    600    700   600   900   1000   MOO
                                 CHHNNEL NUMBER

Fig. 9.  PESA spectrum resulting from the bombardment of 2 mm of the root end of a human hair.


                                                                             IV. USE OF  HEAVY  IONS
                                                  45

-------
                               J.  W.  NELSON AND  W. J.  COURTNKY
  10 0    100   200   300   tOO   500  600   700   800   900  ~1000  1100
                            CHRNNEL NUMBER

Fig. 10. Clam heart spectrum. Sample prepared  by flattening between  glass plates  and freeze drying.
In this spectrum the I2C and "0 peaks are dom-
inated by those isotopes contained  in the filter it-
self. By contrast the  spectrum4) shown in  fig.  8
was  taken by bombarding  a   thin  polystyrene
backed (100 #g/cm2) sample in which case subtrac-
tion for the I2C in the  backing is feasible. The
numbers above the peaks are the mass of that ele-
ment in micrograms per cubic meter of air.
  Although  suitably thin  biological samples  are
usually  difficult to  prepare, results  have been
achieved in several cases.  Fig. 9 is the spectrum
resulting from the bombardment of 2 mm  of  the
root end of a  single human hair.  The somewhat
broad peak labelled 32S contains  lesser amounts of
P and Cl which were detected by  P1XE analysis.
For hair specimens of thicker cross  section the res-
olution of adjacent peaks is not as  complete as for
those in fig. 9  but indications are that they  can be
evaluated with the aid of a computer code. Ano-
ther example of a biological sample analysis is  the
clam heart spectrum shown as fig. 10. Clam hearts
are normally too thick to  permit resolution of ad-
jacent peaks. This difficulty was overcome by flat-
tening the  heart between two glass plates.

5. 'Conclusions
   A quantitative elemental analysis system utiliz-
ing proton elastic scattering analysis  for light ele-
ment determinations has been developed. For thin
samples (1 mg/cm2 or less) the method is absolute,
multi-element,  rapid  (minutes)  and  non-destruc-
tive of the prepared sample. In conjunction  with
proton induced  X-ray emission analysis, it  forms a
system capable of  measurement  of  all elements
within its detection  limits. Extensive application to
the analysis of atmosphere particulate matter have
begun and feasibility studies for biological  samples
are being explored.
References
') T. B. Johansson, R. E. Van Grieken, J. W. Nelson and J.
  W. Winchester, Anal. Chem. 47 (1975) 855.
2) J. William Nelson, Proc. EPA Symp. and Workshop on X-
  ray fluorescence oj environmental samples. Chapel Hill, N.C.
  (1976) (in  press. Ann Arbor Science Publ.).
3) H. C. Kaufmann, K. R. Akselsson and W. J. Courtney. Ad-
  van. X-ray Anal.  19 (1975) 355.
4) K. R. Akselsson. J. W. Nelson and J. W. Winchester. Null.
  Bu. of Stds.  Special Publ. 425,  U.S. Govt. Printing Office
  (1975).
5) E. C. Henley. J. W. Nelson and M. E. Kassouny, Proc. 10th
  Annual Conf. on Trace substances in environmental health (Ed.
  D. D. Hamphill; in press USAEC).
                                              46

-------
                                   TECHNICAL REPORT DATA
                            (Please read Instructions on the reverse before completing)
1. REPORT NO.
 EPA-6QO/2-78-213
                                                           3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE

  PROTON SCATTERING  ANALYSIS FOR LIGHT ELEMENTS IN
                 AIR PARTICULATE MATTER
                                                           5. REPORT DATE
                        1Q7R
             6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
  J.  Nelson. G. Hudson,  H. Kaufmann, W.  Courtney,
  I.  Williams, K. Akselsson, D. Meinert, J.  Wincnester
                                                           8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS

  Physics Department
  Florida State University
  Tallahassee, Florida   32306
                                                           10. PROGRAM ELEMENT NO.
             1AD712  BB-38 (FY 781
             11. CONTRACT/GRANT NO.
                                                             R - 802913
12. SPONSORING AGENCY NAME AND ADDRESS
  Environmental Protection Agency - RTP, NC
  Atmospheric Chemistry and Physics Division
  Environmental Sciences Research Laboratory
  Research Triangle  Park. NC  27711	
                                                           13. TYPE OF REPORT AND PERIOD COVERED
                                                             Final    5/74 - 7/78
             14. SPONSORING AGENCY CODE
               EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT

  Proton scattering  analysis has been developed to provide analyses  of elements
  with atomic numbers  ranging from 6 to  20  as a complementary method to X-ray
  analysis.  Samples of air particulate  matter from both filter  and  impaction
  devices have been  analyzed and time sequence and particle size information
  obtained.  In  combination with X-ray analysis, total mass may  be determined by
  summation of elemental masses; however, further investigation  is needed to
  compare with the gravimetric method.   Like X-ray analysis, it  has  a broad range,
  and is rapid,  subject to automation, and  non-destructive of the prepared sample.
  Analyses of typical  atmospheric samples may be performed in five minutes or
  less.  With further  development, the method can readily be advanced from experimental
  to routine usage.
17.
                                KEY WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
                                              b.lDENTIFIERS/OPEN ENDED TERMS
                           c. COSATI Field/Group
  * Air pollution
  * Particles
  * Chemical Analysis
  *Proton Scattering
  * Chemical Elements
   Light weight elements
13B
07D
2 OH
07A
18. DISTRIBUTION STATEMENT
  RELEASE TO PUBLIC
                                              19. SECURITY CLASS (ThisReport)
                                                 UNCT.ASSTFTF.n	
                           21. NO. OF PAGES
                                57
20. SECURITY CLASS (Thispage)
  UNCLASSIFIED
                                                                         22. PRICE
EPA Form 2220-1 (9-73)
                                            47

-------