ELECTRON MICROPROBE ANALYSIS OF  ATMOSPHERIC
AEROSOLS

D.  K.  Landstrom,  et al

Battelle Memorial  Institute
Columbis, Ohio

31 December  1969
                            DISTRIBUTED BY:
                            Klin
                           National Technical Information Service
                           U. S. DEPARTMENT  OF  COMMERCE
                           5285 Port Royal Road, Springfield Va. 22151
            This document has been approved for public release and sale.

-------
                                           ' -'-^l'. •' ..}„:,» v
'.. '





,-fT.'
:
        1$'&
$g*   "  !il
       m
                             •• --  •
                                   v  ^ ; >



              RESEARCH  REPORT
                                              •'. *<,
                                      1
                               ANALYSIS
                       MICROPROBE
                ELECTRON
                             AEROSOLS
                    ATMOSPHERIC
                NATIONAL AIR POLLUTION CONTROL


                      ADMINISTRATION
                     December
     ' i*

    "

H .
                                              •••&
                                              . -f 'TV v-
                                        -
        •*i.'.V
        V^ .^
        -;•••-
.
.

'
                  ••, ,
                • .  • ' .*.' •.-,••••


                     -•
                 t;.  ••-.•
                              SJSst'luV:
                                 "<,.,  '.
                                • -   : •••. -,1,.,..'.  •  .   ^ •
                           • ;,  •'/- . -   .           vft

                                              - ^
                                  '   ' '     .,-..*•!» ^ilStai.





     BATTELLE  MEMORIAL  INSTITU^""'
       '.    • -  '    •  •     ' .    .•-',*•-  * ^ (•*:•'.. i" .•.'•
                                               ?^*.
                                             '^::


                                              •'^.. -

                                    •
                           . v   •- >      V '     •     " "* •

                . :.    .    •   ?-,'.    '.'*•.• '-•:,•'          '• ' ».V •






       .'-••••                        '              '

-------
           FINAL REPORT
                 on
 ELECTRON MICROPROBE ANALYSIS
    OF ATMOSPHERIC AEROSOLS
                 to
NATIONAL AIR POLLUTION CONTROL
          ADMINISTRATION
        Contract CPA 22-69-33

          December 31, 1969
                 by
          D. K. Landstrom
            Doyle Kohler
BATTELLE MEMORIAL INSTITUTE
      Columbus Laboratories
         505 King Avenue
      Columbus, Ohio  43201

-------
                              MANAGEMENT SUMMARY
Contract No.  CPA 22-69-33

      "Electron Microprobe Analysis of Atmospheric Aerosols"


Background

      This research program is designed to investigate the use of the electron micro-
probe for the  determination of the fine structure and composition of airborne particulates
of significance in air pollution studies.  The results  of this study should give NAPCA and
others interested in air pollution a good overall picture of the applicability of the electron
microprobe to small-particle analysis and characterization.
Significance of Research Results

      The value of the electron microprobe for small-particle analysis has been demon-
strated,  and it has been shown that sample-collection techniques are presently the limit-
ing factors in the amount of compositional information that can be obtained from small
particles.  A  relationship between lead, bromine,  and chlorine in automobile engine
exhaust and atmospheric samples has been demonstrated and recommendations have been
made regarding the type of instrumentation best suited for small-particle analysis.


Applications of Research Results

      The data generated in the program show the many aspects where the electron
microprobe can aid in small-particle analysis. The need for research in specimen
preparation is necessary so that a sample is suitable for the particular requirements
of the microprobe and other associated techniques.   Further investigation of the chem-
ical relationships of elements in air pollution particles is  necessary to determine their
effects on the  environment.

-------
                              TABLE OF CONTENTS

                                                                           Page

SUMMARY	    1

INTRODUCTION	    1

OBJECTIVES	2

EXPERIMENTAL WORK	2

      Test Aerosols	2
      Source Aerosols	8
           Vehicle Exhaust	8
                 Automobile  Exhaust, Stage 5, Beryllium Disk	8
                 Automobile  Exhaust, Stage 4, Beryllium Disk	12
                 Automobile  Exhaust, Stage 5, Electron Microscope Grid.  .   .   12
           Atmospheric Samples	15
                 Atmospheric Samples, Stage 4, Beryllium Disk	15
                 Atmospheric Samples, Stage 5, Beryllium Disk	15
                 Atmospheric Samples, Stage 2, Beryllium Disk	23
                 Atmospheric Samples, Stage 3, Beryllium Disk	27
      Evaluation of Experimental Data	28
      Energy Dispersive Analysis	29

CONCLUSIONS	32

RECOMMENDATIONS FOR FUTURE WORK	34

                                   APPENDIX A

GENERAL OUTLINE OF THE ELECTRON MICROPROBE POTENTIAL AND
 APPLICATIONS TO AIR  POLLUTION RESEARCH	A-l

-------
                         ELECTRON MICRO PROBE ANALYSIS
                             OF ATMOSPHERIC AEROSOLS

                                           by

                          D. K.  Landstrom and Ooyle Kohler



                                      SUMMARY
      This report describes a research effort to determine the feasibility and capability
 of electron microprobe analysis as a method for obtaining detailed compositional and
 structural information about pollutant aerosol particles in the  atmosphere.  Extensive
 data is presented from known test aerosols, automobile engine exhaust, and atmospheric
 samples.  It was determined that the sampling method is presently the limiting factor in
 the detection and analysis of small particles,  and,  by preparing  samples to minimize
 X-ray background, particles as small as 0. 1 micron can be successfully analyzed.  The
 theoretical limitations of the microprobe for both wavelength-dispersive and energy-
 dispersive spectrometers are  discussed, and recommendations are made about analysis
 systems for small particles.
                                   INTRODUCTION
      The problems associated with air pollution are well known and are becoming in-
creasingly more important as the population and industrial activities expand.  The effects
of suspended particulates in the air can be observed every day and include soiling,
deterioration of materials, corrosion, smog, and physiological reactions.  Other long-
term effects not  so readily observable also are  a direct result of particulate contamina-
tion. Actually, very little is known about the composition and fine structure of individual
airborne  particles, especially those in the size  range from about 1 micron in diameter
and below.

      Some of the classical methods of analysis  such as dustfall estimates, total sus-
pended matter,  particle size,  or gross  chemical analysis often do not provide enough
information about individual particles.  Even such valuable  techniques as X-ray diffrac-
tion, X-ray fluorescence, electron microscopy,  light microscopy,  mass spectrometry,
and many others do not provide compositional data on very small individual particles
except in a very  few special cases. It is the analysis of single selected particles that
presents both an opportunity and a challenge for expanding the information available
about the effects of air pollution. Very little is  known about the effects of individual
particles on people, plants,  animals,  or objects.  Often, particles of interest may be
present in specific particle size ranges, but very little is known  about what particle size
range can be expected for a given composition.  In many cases,  a single particle of
interest may be lost among thousands of particles that may be harmless or of no particu-
lar interest.  In  atomic energy operations, the detection of just one particle of fissionable
material among many other particles could be of major importance, especially from the
health physics standpoint.  The electron microprobe analyzer appears to have great
promise as a method for the analysis of individual particles in the micron range.

-------
      A discussion of the problems and theoretical considerations of the electron m
probe as applied to small-particle analysis is important to the work described here.
Because this discussion is so extensive, it has been written separately as an Appendix.
                                     OBJECTIVES
      The research program described herein was planned to achieve the objectives set
forth by the National Center for Air Pollution Control to determine the capacity of elec-
tron microprobe analysis as  a method for obtaining detailed compositional and structural
information about pollutant aerosol particles in the urban atmosphere and from specific
sources.  More specifically, the major objectives of the program were to perform,
through electron microprobe analysis, detailed nondestructive chemical analysis of aero-
sol particles collected from the ambient atmosphere  and from specific sources,  and
thereby define the types of quantitative and qualitative information about aerosol fine
structure and composition obtainable with the present state of the art.  The  study should
demonstrate the sensitivity and the limitations of the method, provide information on the
importance  of instrumental design parameters, and outline future capabilities as pro-
jected from the present state of the art.
                               EXPERIMENTAL, WORK
      This program on "Electron Microprobe Analysis of Atmosphere Aerosols" was
expected to begin on January 1, 1969, and continue for 6 months.  Due to delays princi-
pally in receiving samples appropriate for the study, no technical work other than that
of a survey nature was performed on this contract until April, 1969, and the overall
time period of the program was extended to 12 months.

      The first samples were delivered on March 27,  1969,  and consisted of known test
aerosols as will be described below.  Results were  reported in a letter report dated
July 1,  1969,  and are  summarized below.
                                    Test Aerosols
      Two sets of samples collected from Stages 2, 3,  4,  and 5 of a Battelle cascade
impactor were received and consisted of aerosols of NaCl and Zn(NH4)2(SO4)2'6H2O,
respectively.  Two types of collections were received for each aerosol, one using
polished beryllium disks for a substrate and the other using Formvar-coated electron
microscope grids.

      Results  obtained on the  NaCl beryllium-disk samples were generally good, and
pictures showing the location  and elemental distributions were obtained for all four
cascade impactor stages.  These samples also allowed a comparison of a new-type
combination scanning electron microscope/microprobe (Materials Analysis  Company
Model 400S) with an older, more conventional microprobe (Materials Analysis Company
Model 400).  The superior performance of the 400S is clearly indicated by a direct

-------
 comparison of photographs obtained of identical sample areas in the two instruments.
 Figure 1 is a back-scattered electron (BSE) image taken on the older microprobe show-
 ing several particles from Stage 2 of the cascade impactor.  Figures 2 and 3 are-X-ray
 distribution photographs of the sodium and chlorine, respectively,  in the same area as
 shown in Figure 1.  These figures should be compared with Figures 4 and 5, which show
 the BSE and the chlorine X-ray images respectively, of the same sample.  (Note the
 slightly lower magnification of Figures 4 and 5.) Comparison of these figures  clearly
 show the superior resolution obtained in the new microprobe. Note also the better X-ray
 resolution shown in the chlorine picture.

      As is mentioned in the general discussion on microprobe analysis  (Appendix)
 instrumental  conditions are important  for maximum particle resolution.  Figures 6  and
 7 illustrate the effect of beam current  on the resolution.   (Higher beam currents cause
 the beam to expand in size.)  Figure 6  is a BSE image taken at a specimen current of
 0.009 ua.  Note that Figure 6 represents the same area as in Figures 1  and 4,  except at
 higher magnification. Figure 7 is a similar photograph taken at 0. 03 ^a (normal operat-
 ing conditions for X-ray analysis).  It  is clear  that there is a loss of resolution in
 Figure 7 as well as a smearing effect that makes the particles appear larger than they
 actually are.  A rate-meter scan of this same area for chlorine using the same instru-
 ment conditions as for Figure 7 is shown in Figure 8,

      A useful limit of magnification for this type of scanning electron microprobe where
 the beam diameter is on the order of 0. 1 to 0. 2 micron is approximately 1000-3000X.
 Figure 9 shows three of the particles from Stage 2 of the cascade impactor at 3000X.
 At this magnification, very little structure is observable, and the  surface  detail of this
 type of specimen would have to be studied in a scanning electron microscope for proper
 resolution. The X-ray sensitivity is still good at this magnification; Figures  10 and  11
 show the distribution of Na and Cl, respectively.  Note the higher noise level (back-
 ground counts) in the Na distribution,  due to a less favorable peak-to-background ratio
 for this element in our microprobe.  Figures 9, 10,  and 11 represent a practical limit
 for this type of specimen preparation and our particular commercial microprobe.

      In general, the samples obtained on electron micro-scope grids were not as satis-
 factory as  those on beryllium disks.  The instrument conditions necessary for  X-ray
 analysis usually caused the support film to buckle or melt.  Figure  12 is a BSE image
 obtained from the Stage 2 cascade impactor sample.  It is interesting to note that the
 NaCl cubic shape ia visible on these samples,  whereas those on the beryllium-disk
 samples are generally spherical or oval in shape.  If the beryllium-disk had been
 exposed to more atmospheric moisture than the grid samples, the shape differences
 could be explained as due to the increase in moisture.

      Particles  were detected and analyzed from Stages 3, 4,  and  5 on the beryllium
 disks; but the picture quality of these was progressively worse from those presented
 for Stage 2 and, because of the difficulty of reproduction,  micrographs are not included
 here.  Sodium and chlorine could be detected in particles that were not visible  optically
 but which could be detected by the scanning electron  beam system of the  microprobe.
 There was some agglomeration  of particles on Stages 3, 4,  and 5 that gave relatively
high Na and Cl counts, while the smallest particles  (approximately 0. 1 to 0.2 ju in
 diameter) gave only e. few  counts above the background level.

      Figure  13 illustrates one method of detecting particles that is  often more sensitive
than the standard BSE image. This method of data presentation is called oblique pro-
jection (OP) and consists simply of modulating the y-deflection of the oscilliscope with

-------
           43 OX
FIGURE 1.  BACK-SCATTERED ELECTRON IMAGE OF SAMPLE FROM STAGE
NO. 2 OF CASCADE IMPACTOR (NaCl Be DISK) MODEL 400 MICROPROBE
           43 OX
FIGURE 2.  SODIUM DISTRIBUTION IN THE SAME AREA AS FIGURE 1
          Rate Meter Mode, Model 400 Microprobe
            43 OX
FIGURE 3.  CHLORINE DISTRIBUTION IN THE SAME AREA AS FIGURE 1
          AND 2
          Pulse Mode,  Model 400 Microprobe
           300X
FIGURE 4.  BACK-SCATTERED ELECTRON IMAGE OF SAN  1C FRO*
STAGE NO.  2 OF CASCADE IMPACTOR (NaCl Be DI? '<) N'  >'. L 4u
SCANNING ELECTRON MICROPROBE

-------
          300X
FIGURE 5.  CHLORINE DISTRIBUTION IN THE SAME AREA AS FIGURE 4
          Rate Meter Mode,  Model 400S Scanning Electron Microprobe
           700X
FIGURE 7.  SAME AS FIGURE 6 EXCEPT TAKEN AT 0. 03 UAMP
          SPECIMEN CURRENT
         700X
FIGURE 6.  BACK-SCATTERED ELECTRON IMAGE OF SAMPLE FROM
STAGE NO.  2 OF CASCADE IMPACTOR (NaCl Be DISK) MODEL 400S
Taken at 0.009 y AMP Specimen Current
          700X
FIGURE 8.  CHLORINE DISTRIBUTION IN THE SAME AREA AS
          FIGURE 7
          Rate Meter Mode,  Model 400S

-------
          3000X
FIGURE 9. BACK-SCATTERED ELECTRON IMAGE OF SAMPLE FROM STAGE
         NO. 2 OF CASCADE IMPACTOR (NaCl Be DISK) MODEL 400S
           3000X
FIGURE 11.  CHLORINE DISTRIBUTION IN THE SAME AREA AS FIGURE 9
           AND 10
                                                                                    .r.»^  -.       .
                                                                                            «.-.*••   •- •  *•
                                                                                                   xv-
           3000X
FIGURE 10.  SODIUM DISTRIBUTION IN THE SAME AREA AS FIGURE 9

           Rate Meter Mode, Model 400S
a**
          20 OCX
FIGURE 12. BACK-SCATTERED ELECTRON IMAGE OF SAMPLE
          STAGE NO.  2 OF CASCADE IMPACTO*
           Rate Meter Mode,  Model 400S
               uectron

-------
1000X
   FIGURE 13.
OBLIQUE PROJECTION MADE OF
SAMPLE FROM STAGE NO. 2 OF
CASCADE IMPACTOR (Nad Be DISK)
  200X
      200X
FIGURE 14. BACK-SCATTERED ELECTRON IMAGE OF SAMPLE FROM
          STAGE NO. 4 OF CASCADE IMPACTION (AUTO EX-
          HAUST. Be DISK) SHOWING PARTICLES IN SCRATCH
       FIGURE 15. LEAD DISTRIBUTION IN THE
                 SAME AREA AS FIGURE 14
                                               200X
                                                   FIGURE 16.
                     CHLORINE DISTRIBUTION IN THE
                     SAME AREA AS FIGURE 14

-------
 the BSE signal or the specimen current.  The ordinary BSE image is produced by i.
 lating the  Z axis (spot intensity).  Thus, in the oblique projection mode, very  small
 changes in the back scattered signal that might not be  observable in the regular image can
 be easily observed.  Topographical detail is also enhanced, although interpretation   the
 image is often difficult.

      The Zn(NH4)2(SO4)2-6H2O samples were all unsatisfactory for probe analysis.  The
 sample area could not be located in the light microscope or by a scanning electron image.
 Some isolated clumps of material were found, but these appeared to be debris  or contam-
 ination on the beryllium surface.  Consequently,  no further work was carried  out on
 these samples.
                                   Source Aerosols
 Vehicle Exhaust
      In April,  1969,  some automobile engine exhaust samples were received, and early
 rmcroprobe results on them were not encouraging. In these early samples,  only gross
 quantities of carbon could be detected, with traces of Si, Ca,  Fe,  and Cl.  No lead was
 detected, which was quite surprising to us since engine exhaust samples we had prepared
 here under another program all showed easily detectable lead.  At first, it was thought
 that the particle collection time was not sufficient, and, indeed, this seemed to be part
 of the problem; however,  when later samples were received (June, 1969) the same lack
 of definitive particles was noted.  It was evident that there was a definite sampling prob-
 lem; the beryllium disks simply were not retaining the impinging particles as they should,
 and this  was partially confirmed by one of the samples that was received at this time.

      A beryllium disk from Stage 4 of the Battelle cascade impactor was examined and
 very few particles  were found,  with the exception of one area where a large amount of
 lead was detected.  This area is shown in Figure 14,  which is a BSE image of this area.
 Close inspection of this region showed that this area was a surface scratch on the beryl-
 lium plate and that this was the reason that the particles were trapped.  Chart scans and
 photographs were obtained of this area, with Figures  15 and 16 showing the distribution
 of lead and chlorine,  respectively, in the same area as of Figure  14.  Photographs were
 also obtained showing the Si,  Fe,  Cr, Mn, and Ni  distribution in this area, but they are
 not included in this report.  Many Si particles were observed as well as  many Fe
 particles.   One fairly large particle (= 20)u, near arrow in  Figure  14) was positively
 identified by quantitative analysis to be type 304 stainless steel.

      In October,  1969, the first automobile exhaust  samples were received that were
 useful for extensive investigation.  Acting on our suggestion, NAPCA had put a very thin
 coating of silicone  grease on the surface of the  beryllium disks to aid in particle capture
 and, in addition, had used a  longer sampling time.  These samples were very good for
 microprobe analysis,  and the results are  reported below.

      Automobile Exhaust, Stage 5,  Beryllium Disk.   The deposit from the cascade im-
 pactor was  contained in one small spot near the center of the beryllium disk.  Unlike
 previous samples,  no  difficulty was  experienced in locating the depop.t area, which
 appeared to be a thick agglomeration of particles that thinned out toward  the edges of the
 deposit.  Near the edges and surrounding area, individual particles could be  detected,
 although  many of the larger ones appeared to be made up of clumps of many smaller
particles.

-------
TABLE 1.  RELATIVE CONCENTRATIONS OF ELEMENTS DETECTED IN AUTOMOBILE EXHAUST SAMPLES
           H = high,  M = medium,  L = low, T = trace, ND = not detected.
Element
Sample Description
Stage 5, Be disk,
random area
Stage 4, Be disk,
random area
Stage 4, Be disk,
high iron
Stage 4, Be disk,
high zinc, type 1
Stage 4, Be disk,
high zinc, type 2
Stage 4, Be disk,
high lead
Stage 5, EM grid,
random area
Pb
H

H

L

M

L

H

H

Cl
H

M

L

M

L

H

L

Br
H

M

T

L

L

H

H

Al
M

M

ND

ND

ND

ND

ND

Zn
M

L

T

H

H

T

T

Fe
M

M

H

L

L

L

L

S
L

L

T

T

T

T

L

P
L

L

L

T

T

T

L

Si(a)
M

H

H

L

L

H

ND

Ca
M

L

L

L

L

L

L

Cu
ND

ND

ND

H

T

ND

H
-------
                                            10

      In order to obtain a general idea of the deposit composition, the electron mi< •
probe beam was positioned in the densest portion of the deposit and a wavelength scan
for all detectable elements was run. * Table 1  gives a summary of the elements detected
in all of the automobile exhaust samples and presents a very rough estimation of th»
relative concentration of the various  elements.

      A BSE image of the deposit area was obtained (Figure 17) and individual X-ray line
scans for Pb,  Si, S,  Fe,  Ca, Al, Zn, P, Br,  and Cl were made  along the line visible in
Figure  17.   These  profiles were taken directly from the  oscilliscope screen of the micro-
probe at various  scale factors, so the intensities (peak heights) of the scans  are not
directly comparable.  The  scans do provide a general idea of the elemental  distributions,
and it is easy  to see,  for example,  the obvious segregation of individual particles con-
taining  iron.   The lead, bromine,  and chlorine are all fairly evenly distributed in the
deposit; the Ca, Si, Al, S and,  to a lesser extent, the Zn and P show some evidence  of
segregation into individual  particles.

      Chart scans for each of these elements along the scan line of Figure  17 were also
obtained which provide greater details than it is possible to obtain with an oscilliscope
profile. The chart scans confirm the indications shown on the X-ray line scans and also
show the deposit to be less dense at the center; i.e., the deposit is somewhat ring-
shaped. Due to the difficulty of reproduction, the chart scans are not included in this
report but are available for study if desired. **

      An area near the edge of the deposit was  located and BSE pictures at 1000X of
some scattered individual particles were obtained.  X-ray distribution photographs were
obtained for Pb,  Cl,  Br,  Zn, Fe,  P, and Si.  No photographs for S and Ca were obtained,
since these elements were  found only in heavier regions of the deposit.  These photo-
graphs  show that areas high in lead are also high in chlorine and bromine.   Phosphorus,
zinc, and aluminum also  appear to be associated with these areas. The calcium, iron,
and silicon seemed much more segregated and not necessarily associated with high-lead
areas.

      An attempt was  made to obtain more nearly quantitative data from this sample;
Table 2 shows the results for two of the stages.  Obviously,  these values will not total
100 percent, since  there  are large quantities of hydrogen,  nitrogen,  and oxygen associ-
ated with these deposits and these were not determined.  The carbon value is probably
low due  to large absorption effects  from the lead in the deposit.

             TABLE 2. SEMIQUANTITATIVE  ANALYSIS OF AUTOMOBILE
                         EXHAUST SAMPLES ON BERYLLIUM DISKS
Sample
Description
Stage 5,
Stage 4,
Be disk
Be disk
Pb
17.6
14.3
Cl
2.7
1.0
Br
4.3
1.7
Element (weight percent)
Al
1.0
1.2
Zn
3.5
Trace
Fe^a) S P
0.2 0.3 0.5
0.8 0.7 0.3
SiO>)
4.1
8.1
Ca
0.2
1.0
C
11.
10.

5
0
   (a) Highly segregated -higher values obtainable on individual particles.
   (b) Background from sllicone grease subtracted.
 •Due to the difficulty of reproducing wavelength chart scans (which may be 4 to 6 feet long), none of the charts are included
  in this report. Results will be summarized, and reference to the original charts can be made if questions arise.
••Chart scans were obtained for all of the samples documented in this report   None are included for the reasons mentioned above
  but are on hand for stuoy if desired. In general, the chart scans provide greater resolution and greater accuracy in locating
  specific panicles, hence were mainly used to pick specific particles for analysis

-------
                                      11
                    tsox
                          B.S.E. Image Showing Scan Line
FIGURE 17.  X-RAY SCANS ALONG LINE SHOWN IN PHOTOGRAPH, STAGE 5,
             Be DISK,  AUTOMOBILE EXHAUST

-------
                                          12


      Comparisons between the two stages  are quite interesting,  and it is easy to see
that the larger,  more segregated particles (Fe,  Si,  and Ca) are higher on Stage 4 than
on Stage 5; the Pb, Cl, Br, and Zn, which appears to be related to smaller particle size,
are higher on the Stage 5 sample.  It would be interesting to see what type of elemental
separation (by particle size) would be possible using the other stages of the cascade
impactor.  Unfortunately, no other stages were  supplied,  so complete comparisons are
impossible.  It could be conjectured that the cascade impact or might be useful for sepa-
ration purposes, not only on a particle-size basis,  but on an elemental basis if specific
elements are present in certain size ranges. This could be quite important in some
analytical techniques where, for maximum sensitivity, elements  of little  or no interest
could be eliminated or greatly reduced in the analyzed sample.

      Automobile Exhaust, Stage 4, Beryllium Disk.  The deposit area on this sample
appeared slightly less dense than that obtained on Stage 5.  A greater number of individ-
ual particles were  also visible in this sample.  As in the case of Stage  5, a wavelength
scan for all detectable elements was run on a representative area,  and the  results are
similar.  Relative  concentrations for Stage 4 can be found in Table 1.

      A BSE image of the deposit was obtained (Figure 18) and individual  X-ray line
scans  for Pb, Fe,  Al, Br, Zn, S, Si, Cl, P, and Ca were made  along the line visible
in the  photograph.  Again, the heights of the traces are not directly comparable,  but
provide only a relative elemental distribution along the scan line.  Elemental segregation
is easier to see in this sample than in the Stage  5 samples.

      Chart scans along the scan line of Figure  18 were also obtained but are not included
in this report.  Segregation of Fe,  Si,  Ca, Zn, and, to a less extent, Pb, Cl, Br, and S
can be clearly shown in these charts.

      An area near the edge of the  deposit was located and BSE and elemental distribu-
tion pictures were  obtained for Pb,  Br, Si, Fe,  Cl, Zn,  and Cr.  The S, P, and Al were
not intense enough to provide a good picture.  These pictures are not included in this
report.

      Since individual particles could be  found easily in this sample, particles with high
values of various elements were located  and individually analyzed with a  wavelength scan.
The results are summarized in Table 1 and include particles having high iron, high zinc,
and high lead content.

      Several particles containing high sulfur content were found in this sample and a
sulfur wavelength-shift measurement was made. Figure  19 shows  the sulfur spectrum
from the sample compared with the spectrum from PbS and CaSO4. In general,  Fig-
ure 19 shows that most of the sulfur present in this sample is in  the sulfate form. There
is a reproducible double peak in the sample spectrum that  might  indicate a  small per-
centage of elemental sulfur; however, this could not be confirmed without extensive
additional work with better standards and spectrometer alignment.

      Semi quantitative data for this sample are given in Table 2.

      Automobile Exhaust, Stage 5, Electron Microscope Grid.   The deposit on the
microscope grid was quite thick and had  the same ring structure as was visible on the
Stage 5 beryllium-disk sample.  A wavelength scan on the  thick deposit area was obtained,
with the  relative concentrations for this sample  summarized in Table  1.

-------
                                    13
                    I50X
                         B.S.E. Image Showing Scan Lines
FIGURE  18.  X-RAY SCANS ALONG LINE SHOWN IN PHOTOGRAPH, STAGE 4,
             Be DISK, AUTOMOBILE EXHAUST

-------
          (A
                                                                    CaS04  standard
           Sample auto

           exhaust stage

           no. 4
                                                                        PbS standard
                                           5.3736
5.3728
                 \
5.3720           A decreasing
FIGURE 19.  ELECTRON MICROPROBE ANALYSIS OF AUTOMOBILE EXHAUST - SULFUR WAVELENGTH SHIFT

-------
                                           15

      A BSE image of the deposit was obtained (Figure 20} and individual X-ray line
 scans for Cl, Pb,  Br,  Al, S,  Fe,  Zn, Si,  Ca, and P were made along a line which is
 visible in the photograph.  Segregation of some of the elements is clearly visible.

      Areas near the edge of the deposit areas were examined for individual particles,
 but there was no area defined clearly enough to obtain suitable photographs.


 Atmospheric Samples


      Atmospheric Samples,  Stage 4, Beryllium Disk.  This sample deposit was  much
 lighter than those obtained for the automobile exhaust samples, but it still was quite
 adequate for analysis.  Most of the  samples was concentrated in a half circle and con-
 tained many individual  particles scattered  over the beryllium surface.

      Initially, the beam was positioned in the densest portion of the deposit and a wave-
 length scan for all detectable elements was run.  Table 3 is a summary of the elements
 detected in  this sample and all of the other atmospheric samples.  No attempt was
 made to obtain quantitative data in other  than  relative terms.

      A BSE image of the deposit area was obtained (Figure 21) along with individual
 X-ray pulse mode pictures for Pb, Si, Br, Zn, Fe, Mn,  Ca,  S,  Cl,  K, and Al (Fig-
 ure 22 through 32, respectively). Since  these pictures are typical of those obtained for
 all the other samples,  only this set is reproduced; however,  Simular photographs were
 obtained on all of the atmospheric samples.  Segregation  of some of the elements is quite
 evident.

      Individual particles with high iron, high zinc, high lead,  and high calcium contents,
 respectively, were easily located in this sample, and wavelength scans were obtained.
 The data are summarized in Table 3.  BSE images and X-ray distribution photos  were
 also obtained for each of these particles  but are not included in this report.  The  high
 calcium particle  (Stage 4, Table 3) is interesting to examine since the wavelength scan
 of this particle shows trace amounts of tin, titanium, and chromium. It is important
 to realize that, unless  specific elements are  sought or particle analysis is performed
 on an individual and systematic basis, elements which are present in only a few particles
 can easily be missed in an analysis.  This  aspect is covered more fully in the Appendix.

      This atmospheric sample was extensively examined for lead particles that had no
 bromine or  chlorine associated with them,  but none could be found; i.e.,  all particles
 containing lead also contained bromine and chlorine,  as was also true of the automobile
 exhaust samples.

      Many  particles were found in which no elements could be detected, and these were
 probably hydrocarbon deposits.

      Atmospheric Samples, Stage 5,  Beryllium Disk.  The deposit  area of this sample
appeared to be a uniform circle slightly less dense in the  center.  The beam was focused
on a  dense,  representative area of the deposit and a wavelength scan was made.  The
 results  are  show in Table 3.  It is interesting that this random area contained  fewer
detectable elements than  did the sample from Stage 4 (random area, Table 3).

-------
                                     16
                    I50X
                         B.S.E. Image Showing Scan Line
FIGURE 20.  X-RAY SCANS ALONG LINE SHOWN IN PHOTOGRAPH, STAGE 5,
             E.M. GRID,  AUTOMOBILE EXHAUST

-------
                                                   17
  TABLE 3.  RELATIVE CONCENTRATIONS OF ELEMENTS DETECTED IN ATMOSPHERIC SAMPLES

             H = high,  M = medium, L = low, T = trace, ND = not detected.
                                                           Element
Sample Descpntion
Pb   Cl   Br   Al   Zn   Fe   Mn
                                                                      Si*
Ca   Cu   K   Ng   Mg
                  (b)
(d)
Stage 4, Be disk,
   random area
Stage 4, Be disk,
   high iron
Stage 4, Be disk,
   high zinc
Stage 4, Be disk,
   high lead
Stage 4, Be disk,'
   high calcium
Stage 5, Be disk,
   random area
Stage 5, Be disk,
   high sulfur
Stage 5, Be disk, 
   high iron
Stage 5, Be disk,
   high lead
Stage 2, Be disk,'
   high copper
Stage 2, Be disk,""
   high titanium
Stage 2, Be disk, 'f>
   high lead
Stage 2, Be disk,
   high sodium
Stage 2, Be disk,
   high iron
Stage 2, Be disk, (g>
   high calcium
Stage 3, Be disk,
   high iron
Stage 3, Be disk,
   high calcium
Stage 3, Be disk,
   high zinc
Stage 3, Be disk
   high lead
Stage 3, Be disk,
   high sulfur
Stage 3, Be disk,
   high aluminum
 L     TT    T    LH    MLNDH

 T    ND  ND  ND   T    H    M   M  ND   L

ND    L   ND  ND   H    M   L   L  ND   L

 H     HH    LNDMNDHTH

ND    T   ND   L    L    M   T   H  ND   H

 L    ND  ND   T   ND   T    T   M  ND   H

 L    ND  ND  ND  ND   T    ND   H  ND   M

ND   ND  ND  M   ND   H    ND   H   T    H

 H     MHNDTTNDMLH

ND    M   ND    H   ND   T    ND   M  ND   H

ND   ND  ND   ND  ND   T    ND   ND  ND   H

 H     LLHNDHLLTH

 ND   ND  ND   ND  ND  ND   ND   L  ND   L

 ND   ND  ND   ND   L    H    ND   M  ND   H

 ND    T   ND   T   ND   T    ND   L  ND   H

 ND   ND  ND  ND  ND   H     L
-------
150X
       FIGURE 21.  B.S.E. IMAGE STAGE 4, Be DISK,
                  ATMOSPHERIC SAMPLE
150X

 FIGURE 23.  Si DISTRIBUTION IN AREA SHOWN IN FIGURE 20
150X

 FIGURE 22.  Pb DISTRIBUTION IN AREA SHOWN IN FIGURE 20
                                                                                                                                           oo
150X

  FIGURE 24. Br DISTRIBUTION IN AREA SHOWN IN FIGURE 2'

-------
 150X
 FIGURE 25.  Zn DISTRIBUTION IN AREA SHOWN IN FIGURE 20
150X
 FIGURE 21.  Mn DISTRIBUTION IN AREA SHOWN IN FIGURE 20
150X





 FIGURE 26.  Fe DISTRIBUTION IN AREA SHOWN IN FIGURE 20
                                                                    150X
                                                                     FIGURE 28. Ca DISTRIBUTION IN AREA SHOWN IN FIGURE 20

-------
150X



 FIGURE 29.  S DISTRIBUTION IN AREA SHOWN IN FIGURE 20
150X
 FIGURE 31.  K DISTRIBUTION IN AREA SHOWN IN FIGURE 20
150X



 FIGURE 30. O DISTRIBUTION IN AREA SHOWN IN FIGURE 20
                                                                     150X
 FIGURE 32.  Al DISTRIBUTION IN AREA SHOWN IN FIGURf 20

-------
                                   21
                 I SOX
                     B.S.E.  Image Showing Scan Line
       Pb
       Br
Al
FIGURE 33.  X-RAY SCANS ALONG LINE SHOWN IN PHTOGRAPH, STAGE 5,
             Be DISK,  ATMOSPHERIC SAMPLE

-------
          to
          o
          i_
          o
          v>
          c
          0>
                                            PbS standard
                                                                       Sample atmospheric

                                                                       stage no. 5 Be disc
                                          5.3736
5.3728
5.3720    X decreasing
FIGURE 34.  ELECTRON MICROPROBE ANALYSIS OF ATMOSPHERIC SAMPLE - SULFUR WAVELENGTH SHIFT

-------
                                          Z3

      A BSE image of the deposit was obtained (Figure  33) and individual line scans
 for Pb, K,  Si, Si,  Br, and Al were made along the line visible in the photograph.

      A BSE image and X-ray distribution photos were  obtained for this sample; however,
 since they are basically similar to those already presented for the Stage 4 sample,  they
 are not included in this report.

      Individual particles with high specific-element content were located and wavelength
 scans for several of these particles are summarized in Table 3.  Particles run include
 high sulfur, high iron, and high lead.  BSE image and X-ray distribution photos  were
 obtained for each of the particles examined.  As in the  previous samples, particles having
 a high lead content also contain bromine  and chlorine.

      A sulfur-wavelength-shift measurement was made (Figure 33), similar to  that run
 for the  automobile exhaust  sample (Figure 19).  Figure 34 shows the sulfur spectrum
 from the sample compared with the spectrum  from PbS.  As in the case of the automobile
 exhaust, the majority of the sulfur present in  the sample  is in the sulfate form.  The
 double peak again might indicate the presence of elemental sulfur,  but this would have to
 be confirmed by additional  measurements.

      Atmospheric Samples, Stage 2,  Beryllium Disk.  The sample density was  lower on
 this  sample than any previous  samples, but many large individual particles were found
 scattered over a large area.  Since the deposit was so widely and thinly dispersed on
 the beryllium disk, a  random-area wavelength scan was not made.  A BSE image
 (Figure 35) was obtained as well as X-ray distribution photographs for  Ca,  Ti,  Fe,  Ca,
 Zn,  Pb, Br, Al, Si, S, Cl, K,  and Mg.  The X-ray distribution photos are not included
 in this report,  since they are similar to those already shown for the Stage 4 sample.
 The  chart scans  obtained for this sample confirmed the presence of many individual
 large particles.
                      150X

                FIGURE 35.  BSE IMAGE SHOWING SAMPLE DEPOSIT,
                             STAGE 2, Be  DISK, ATMOSPHERIC SAMPLE

      Wavelength scans for particles high in specific elements were obtained, with the
results  summarized in Table 3.  Particles run include high copper,  high titanium, high
lead,  high sodium,  high iron, and high calcium.

-------
                           iii
                                                                                               40
                                                                                               30
                                                                                               20
in
at
u



I
<«-
o
w


f
                                                                                               10
                                       12   II   10   9876    54    32   10

                                            Co/Si  Ratio
FIGURE 36.  FREQUENCY OF INTENSITY RATIOS FOR CALCIUM-SILICON (100 PARTICLES MEASURED)

-------
          i   i  i  i  i  i   i  i
n r
                                                                                                30
                                                                                                  M
                                                                                                  w

                                                                                                20o
                                                                                                  4-
                                                                                                  k-


                                                                                                  s.
                                                                                                  V


                                                                                                •of
                           30   28   26   24  22   20    18   16   14   12   10   8   6    4   2   O


                                              Ca/AI Ratio
                                                                                                                m
FIGURE 37.  FREQUENCY OF INTENSITY RATIOS FOR CALCIUM-ALUMINUM (100 PARTICLES MEASURED)

-------
                                                                                                     3O
                                                                                                        M
                                                                                                        01
                                                                                                     201
                                                                                                        0»

                                                                                                     101
     20   19   18   17   16   15   14    13   12   II    10   9    6    7   6    5    4    3   2     10

                                                Si/AI Ratio
                                                                                                                       Cv)
FIGURE 38.  FREQUENCY OF INTENSITY RATIOS FOR SILICON-ALUMINUM (100 PARTICLES MEASURED)

-------
                                           27


      For this sample, it was decided to see if there were  many particles with similar
 elemental compositions or if the compositions varied greatly from particle  to particle.
 Consequently, the method of White,  Denny, and Irving(1) was used to analyze 100 of the
 calcium-containing  particles.  This  method recognizes that for small particles,  the
 particle-to-particle variation in absolute X-ray intensities for a given element may vary
 by over an order of magnitude,  depending on the fraction of the incident electron beam
 absorbed by the particle.  They found that the intensity ratio for two elements is
 essentially independent of particle size and shape in  a mixture of particles of the same
 composition.  Thus, by measuring intensity ratios and plotting against known composi-
 tion,  one can determine quite accurately the composition of unknown particles.

      The procedure followed in making the ratio measurement was to select a particle
 from the  X-ray image; then the electron beam was positioned directly on the particle
 and both elements were counted simultaneously for 10 seconds.  Duplicate intensity
 readings  were made for each particle.  There is a very definite advantage in the sim-
 ultaneous measurement of the two elements, since an error in beam positioning is
 cancelled out as are errors in resetting the spectrometers.  Of course, there is a cor-
 responding savings in time for  each  analysis.

      Figure 36 is a graph showing the  relative  frequency of intensity ratios for Ca/Si
 on 100 measured calcium-containing particles in this  sample.  Figure 37 is a similar
 graph showing the Ca/Al intensity ratio, while Figure 38 shows the Si/Al  ratio.  The
 wide  variations in intensity  ratios is apparent from these figures,  and we can thus con-
 clude that there is a wide  variation in the  composition of individual particles.   Extensive
 statistical analysis involving several thousand particles would be necessary to  make
 more definite statements.   For this type  of analysis, a probe with automated readout is
 nearly a necessity,  as operator time becomes excessive.

      Atmospheric Samples, Stage 3, Beryllium Disk.  The particle density of this
 sample was greater than that of the previous sample,  and individual particle could easily
be seen.   Figure 39 is a BSE image  of this sample showing the particle distribution.
                    150X

                FIGURE 39.  BSE IMAGE SHOWING SAMPLE DEPOSIT,
                             STAGE 3, Be DISK, ATMOSPHERIC
                             SAMPLE

-------
                                           28

      X-ray distribution pictures were also taken for Ca, Si, Al, S, Cl, Fe, K,  Ph, "r
 and Mg, but these again are similar to those previously presented and are not  included
 in this  report.

      Wavelength scans for particles high in specific elements were obtained,  with
 results shown in Table 3.  Particles run include high iron, high calcium, high zinc,
 high lead, high sulfur, and high aluminum.
                             Evaluation of Experimental Data


      When the rough data regarding particles containing high lead in the automobile-
 exhaust samples and in the atmospheric  samples are tabulated, it is easy to see the
 correlation of the lead with chlorine and bromine (see Table 4).   Study of this table
 points the way for an interesting set of experiments on the lead-chlorine-bromine re-
 lationship in atmospheric samples.  Intensity ratios for these elements should be mea-
 sured in many particles; the  results should prove a definite correlation. It is unfortu-
 nate that under the scope of this program this could not be done; however,  since the
 qualitative relationship has been demonstrated, the quantitative measurement should be
 a routine matter.  If the  chlorine and bromine are  examined in Table 4, it can be seen
 that there is an increase in the relative amounts from Stage 2 to Stage 5 (Stage 2 trapping
 the larger particles and Stage 5 the smallest).  The lead concentrations for these stages
 are simply listed as "high",  since at the scale factors used to detect; the other elements,
 the lead was off scale in all cases.   Closer investigation has shown that the concentra-
 tion of lead in the individual particles increases as the particle size decreases and that
 there is a definite corresponding correlation for the chlorine and bromine values.  Dis-
 tributions on the cascade impactor stages for several other elements seem highly de-
 pendent on the particle size,  specifically iron, aluminum, and manganese, which de-
 crease with particle size.  Specifically,  it seems that the  high-lead-content particles
 located on Stage 5 of the cascade impactor have higher lead, chlorine, and bromine  than
 those from the other stages (larger particle size).  The larger particles contain higher
quantities of Al, Fe,  and Mn.  Thus, there seems  to be a  definite separation of the ele-
ments by particle size.

       TABLE 4.  RELATIVE  CONCENTRATIONS OF ELEMENTS DETECTED
                   IN PARTICLES OF HIGH LEAD CONTENT FOR AUTOMO-
                   BILE  EXHAUST AND ATMOSPHERlCiSAMPLES
                          Data Condensed From Tables 1  and 3.
Element
Sample Description
Stage 4,
Stage 2,
Stage 3,
Stage 4,
Stage 5,
Be disk,
Be disk,
Be disk,
Be disk,
Be disk,
auto exhaust
atmospheric
atmospheric
atmospheric
atmospheric
Pb
H
H
H
H
H
Cl
H
L
T
H
M
Br
H
L
M
H
H
Al
ND
H
.L
L
ND
Zn
-•T
ND
ND
ND
T
Fe
L
H
L
M
T
Mn
ND
L
T
ND
ND
S
T
L
T
H
M
P
T
T
ND
T
L
Si
H
H
L
H
H
Ca
L
T
L
T
T
K
ND
L
L
H
L
H = high, M = medium. L = low, T = trace. ND = not detected.

-------
                      21   current  bestsellers
       on   the   energy  crisis  from   NTIS
          Special  Discount
    ]  The U.S. Energy Problem
               — in three volumes —
    only $20 instead of $29.85 if purchased as set.
                  PB-207-517-SET
[J The U.S. Energy Problem—Volume I, by 6. C.  Szego
   Intel-Technology Corp., Warrenton, Va., Nov.  1971, 83p,
   $4.85
   PB-207 517
   O Volume II, Appendices—Part A, 744p, $12.50
      PB-207 518
   [J Appendices—Part B. 685p, $12.50
      PB-207 519
  The Growing Demand for  Energy, by R. 0.  Doctor, Rand
  Corp., Santa Monica, Calif., Jan. 1972, 29p, $3.00
  AD-742 382
  Electrical Power Supply and Demand Forecasts for the United
  States Through 2050. Hittman  Associates, Inc., Columbia,
  Md., Feb. 1972, 57p, $4.50
  PB-209 266
  A Survey  of Fuel  and Energy Information Sources, Vol. I.
  Mitre Corp., McLean, Va., Nov.  1970, 298p, $3.00
  PB-197 386
  [J Volume II,  Federal Questionnaires. 63Ip, $9.00
     PB-197 387
  Energy Research  Needs, Resources  for  the Future,  Inc.,
  Washington, D.C.,  Oct. 1971, 893p,  $13.50
  PB-207 516
  Nuclear Energy tar a New Town. Rhode Island Univ., Kings-
  ston, Water Resources Center, Aug. 1972, 64p,  $3.00
  PB-207 300
  Energy Consumption for Transportation In the U.S. Oak Ridge
  National Lab.,  Tenn., March 1972, 38p,  $3.00
  ORNL-NSF-EP-15
                  [J Survey of Nuclear Power Supply Prospects. Hittman Associ-
                     ates, Inc., Columbia, Md., Feb.  1972, 99p, $4.85
                     PB-209  283
                  Q Energy,  Air Quality and the System Approach, by Guy Black,
                     George  Washington  Univ., Wash., D.C., Program  of Policy
                     Studies  in Science and  Technology, July 1970, 92p, $3.00
                     PB-195  797
                  fj Conservation and Better Utilization of Electric  Power by
                     Means of Thermal  Energy Storage and Solar Heating, by
                     Manfred Altman, Towne School of Civil and Mechanical Engi-
                     neering, Philadelphia, Pa. Oct. 1971, 265p, $6.75
                     PB-210  359
                  [] Use of Steam-Electric  Power Plants to Provide Thermal
                     Energy to Urban Areas. Oak Ridge National Lab, Tenn.,  Jan.
                     1971. 195p.  $3.00
                     ORNL-HUD-14
                  G Survey of Thermal Research Programs Sponsored by Federal,
                     State, and Private Agencies (1970). Oak Ridge National  Lab,
                     Tenn., March 1971, $3.00
                     ORNL 4645
                  fj Solar Energy In Developing Countries: Perspectives and Pros-
                     pects.  National  Academy  of  Sciences,  Washington,  D.C.,
                     March 1972, 60p, $4.50
                     PB-208  550
                  fj Study of the Future  Supply of Low Sulfur Oil tar Electrical
                     Utilities. Hittman Associates. Inc.. Columbia, Md., Feb. 1972.
                     76p, $4.85
                     PB-209  257
                  Q Systems Evaluation of Refuse as a Low Sulfur Fuel—Vol. I.
                     Envirogenics  Co.,  El  Monte, Calif., Nov.  1971, 311p. $9.00
                     PB-209 271
                     D  Volume II, 301p, $9.00
                         PB-209 272
                  Q Study of the Future Supply  of  Natural Gas for  Electrical
                     Utilities. Hittman Associates, Inc., Columbia, Md., Feb. 1972,
                     44p, $3.75
                     PB-209 285
                  fj The Potential of Nuclear MHD Electric  Power Systems by
                     G. R. Seikel  and L. D. Nichols, National Aeronautics  and
                     Space Administration Lewis Research Center, Cleveland, Ohio,
                     1971, 12p, $3.00
                     N71-24578
                             All books available In microfiche.  Check box below.
 Fill out and return now to:  National Technical Information Service
                            U.S. Department of Commerce, Springfield,  Va.  22151
 D  Send  the titles I  have indicated above  in microfiche.
     95* each.
 O  Send  the titles I have  indicated  above in paper copy.
 D  Send  me your free brochure NTIS  INFORMATION
     SERVICES.* NTIS-PR-73-00
                 Charge my NTIS Account No.	
                 Bill  me (Not applicable to  foreign customers).
                 Add 50# per title.
                 Hire is my check for
 NAME

 TITLE
ORGANIZATION
 ADDRESS
 CITY, STATE, ZIP

-------
                                          29

                              Energy Dispersive Analysis
      In order to test the relative sensitivity of energy-dispersive and crystal-dispersive
 systems, a Nuclear Diodes lithium-drifted silicon semiconductor detector was fitted to
 a Cambridge Scanning Electron Microscope.

      Since a  scanning electron microscope (SEM) is quite similar in principle to the
 microprobe, it is quite easily adapted for energy-dispersive analysis.

      The normal-wave length dispersive spectrometer such as is used in the micro-
 probe analyzes X-rays by diffracting them off a crystal and  observing the angle (wave-
 length)  where the Bragg  equation is  satisfied (NX = 2d sin 9).  This type of detection is
 capable of X-ray resolution of a few electron volts, hence it is no problem to separate
 the wavelengths of adjacent elements.  The wavelength dispersive spectrometer is,
 however, a very low-efficiency device,  and it has the disadvantage of being a "single
 channel" device in that it can examine only one wavelength at a given time.  Also,  at the
 very low beam currents  used in a SEM, the intensity of X-rays produced is very much
 less than is that produced in a microprobe, and, also, very long  analysis times are
 necessary.

      An energy-dispersive spectrometer has the advantage of recording, essentially
 simultaneously, the full  spectrum of the X-rays present.  This detector gives a signal
 that is proportional to the energy of the X-ray photons and,  thus,  the X-ray spectrum
 is dispersed according to its energy.  This type of detector  is very efficient, since it
 can be placed close to the sample and can view a significant solid angle.  Furthermore,
 since the entire spectrum can be viewed and recorded in a very few seconds, the total
 analysis time is generally short.  The major fault associated with energy-dispersive
 devices is their lack of resolution (about 200 ev),  which is considerably poorer than the
 crystal spectrometer. In many cases, however, this resolution is sufficient for prac-
 tical detection of elements from about oxygen and higher in the periodic table.

      The Stage 5 automobile exhaust sample collected on an electron microscope grid
 was mounted and placed  in the SEM,  Since the electron microprobe had previously de-
 tected particles with high concentrations of calcium in this sample, the pulse height
 analyzers of the Nuclear Diode analyzer was set to accept only the calcium peak,  and
 a scan at 1000X was made over  several areas of the sample.  The SEM can be modified
 to produce X-ray pulse pictures similar to those obtained on the microprobe, and by
this method an area was  located that seemed to have a particle with calcium concen-
 trated in it.  Figure 40 is a high-resolution, back scattered electron image of the se-
 lected area, while Figure 41 is  the X-ray distribution for calcium obtained on the SEM.
 Figure  42 shows a BSE image of the same area  taken on the electron microprobe; the
 superior resolution of the SEM photograph (Figure 40)  is evident.  Figure 43 is the
 X-ray distribution for calcium obtained in the microprobe,  and, in this case, it is easy
 to see the superior X-ray peak-to-background resolution of  the microprobe when com-
 pared with the SEM X-ray distribution (Figure 41).

      The electron beam of the SEM was positioned on the particle showing the high
 calcium, and the spectrum was  recorded by the energy-dispersive spectrometer
 (Figure 44).  Elements that can positively be identified from this  spectrum are Cu
 (from the electron microscope grid), Al, Cl,  Ca,  Pb,  and Br.

-------
1000X
  FIGURE 40. SCANNING ELECTRON MICROGRAPH OF AREA
             FROM STAGE 5, AUTO EXHAUST. E. M. GRID
 1000X
   FIGURE 42. B. S.E. IMAGE OF SAME AREA AS FIGURE 39
              TAKEN ON ELECTRON M1CROPROBE
    1000X
FIGURE 41.  X-RAY DISTRIBUTION FOR Ca IN SAME AREA AS FIGURE 39

      Energy Dispersive Spectrometer Mounted on S. E. M.
     1000X
       FIGURE 43.  X-RAY DISTRIBUTION FOR Ca IN SAME AR A
                  AS FIGURES 39 AND 41
       Electron Microprove Wavelength Dispersive Spectrometer

-------
                                    31
   10
   8
2  6
x
in

I  5
o  a
       CuL£
                               CuKa
                                6      10      12
                                 Channel Number
14      16      18     20
FIGURE 44.  SPECTRUM OBTAINED BY ENERGY-DISPERSIVE SPECTROMETER,
            STAGE 5, AUTOMOBILE EXHAUST, E. M. GRID, HIGH CALCIUM
            PARTICLE

-------
                                          32

      This same particle was run in the electron microprobe, and the same elements
were found (considerably higher peak-to-background ratio), as well as Si, S, Zn,
and Fe.

      In this case,  the energy dispersion gave a spectrum of the major elements in about
36 seconds, compared with about  55 minutes for the wavelength scan on the micropiobe.
It did not detect a few elements that were found in trace amounts by the microprobe.

      The question must be asked, "Just what information is desired in particle
analysis? "  If one were seeking only the major components  of many particles, then the
energy-dispersive  system is  clearly superior.  On the other hand, if one is interested
in trace components in particles containing a specific element,  then the wavelength
spectrometer is superior because the wavelength spectrometer  can be set to a specific
element and can find a particle containing that element just as fast and with greater
sensitivity than the energy-dispersive spectrometer.

      In this particular example,  if sulfur had been present in more than trace amounts,
it would have interfered with the chlorine peak,  and the energy-dispersive  spectrome-
ter could not have  separated the true peaks.  The  same is true  of the silicon-aluminum
pair.   Perhaps  a computer could  smooth out the curves, but the time element for analy-
sis then becomes important.  The wavelength dispersive  spectrometer has no difficulty
in separating these adjacent-element peaks and  also has the advantage of providing
quantitative data if desired.  The energy-dispersive system is semiquantitative at best.
                                     CONCLUSIONS


      The efforts on this program demonstrate clearly that the electron microprobe
analyzer  is a useful and proper instrument for the examination of air pollution particu-
late matter.  The present limitations are not so much that of the instrument but rather
the type and method of specimen collection.  During this project, it became apparent
that, given a suitable  sample, a great deal of information can be obtained on particles
down to about 0. 1 \JL in diameter.   At the present state of the art,  particles smaller than
this cannot be examined because the X-ray count ratios are too low, and the particles
are almost invisible on the electron-scanning image and are not visible at all on the
light optical system.  It seems clear that the minimum size of particle permitting ac-
curate  quantitative  analysis lies between 1 and 5 microns.  It  is possible  to obtain in-
formation on the composition of smaller particles, but such information is generally
qualitative at best,  especially in the situation where X-rays may be excited simultan-
eously  from the particle and a substrate material; once the characteristic radiation
from the particle is overwhelmed by the background radiation from the substrate, even
qualitative analysis of particles much smaller than 0. 5 ju is impossible.

      If the specimen collecting conditions are directed to producing dispersed parti-
cles on a substrate producing no X-ray background (such as beryllium or carbon films),
interference from the substrate is avoided and particles much smaller than 1 micron
can be  analyzed qualitatively with a fair degree of certainty.   If two  or more elements
are present, some  degree of quantification is possible using the ratio method described
previously.

-------
                                          33

      The importance of using the electron microprobe in conjunction with other analy-
 sis techniques cannot be overemphasized.  Use of electron microprobe analysis, com-
 plemented by scanning electron microscopy, X-ray diffraction,  electron microscopy,
 optical microscopy, mass spectrometry, and chemical analysis becomes a powerful
 system for characterizing particles.  The practical usefulness of combining  these tech-
 niques is particularly evident in a laboratory where all of the instruments are available
 and there is  constant and close cooperation among those  responsible for using the vari-
 ous methods.

      The question as to which instrument is best suited  for particle analysis has no
 real answer; it all depends on what information is  desired.  The scanning electron mi-
 croscope is indispensable  if surface-morphology information is desired; no present
 commercial microprobe can supply the necessary resolution to provide surface detail
 on 0. \-fji particles.  The compromises necessary to provide this resolution detract
 from the X-ray and/or light optical performance of the basic microprobe instrument.
 If the problem is to detect one particle in several thousand with a specific elemental
 composition, an automated microprobe may best fit the requirements.  If a fast general
 analysis of the major elements  in many particles is desired,  energy-dispersive analy-
 sis on a  scanning  electron microscope or conventional electron microprobe is probably
 more suitable.

      Ideally, one should have at least three separate instruments - the electron mi-
 croprobe, the scanning electron microscope,  and the  transmission electron microscope,
 with techniques developed  so that the specimens  can be interchanged easily between
 these instruments and the  same area and/or particle easily located.  This is not as dif-
 ficult as it may sound, and only by a combination of these techniques can the maximum
 amount of information be obtained about a particle or  particles.

      In commercial  instrumentation, the electron microprobes currently available do
 provide much better scanning facilities than those  available just a few years  ago, but
 this scanning facility is approaching the limit that  can be expected without degrading
X-ray  performance.  Instruments incorporating  microanalysis and transmission elec-
 tron microscopy have been built, but at this time are  not commercially available in a
 useful  form.   Energy-dispersive systems should be watched closely,  as improvements
 in this field have been very rapid, and, if the sample  is designed around the  peculiar
 limitations of the  energy-dispersive system, a lot of useful information may be obtained
 in a very short time.

      In general, at the present state of the art,  the most valuable single instrument
 commercially available would be an automated,  programmable scanning electron mi-
 croprobe with both wavelength and energy-dispersive  spectrometers.   Such an instru-
ment is available from several manufacturers at about $125, 000 to $150, 000. If some
 compromises are  made, an adequate instrument for many applications of particle
analysis  could be purchased for about $60,000 to $80, 000.

-------
                                          34

                       RECOMMENDATIONS FOR FUTURE WORK
      It must be emphasized that present methods of collecting air pollution particles
are not adequate to provide the maximum amount of information from single particles,
and an experimental program to develop suitable techniques is necessary.  For ex-
ample,  it is  often easier to disperse particles from a liquid medium; collection tech-
niques involving electrostatic precipitation into a liquid media should be investigated.
There are also many transmission electron microscopy techniques that are applicable
to microprobe analysis and which could be  easily utilized.

      The aspect of suitably moving specimens from instrument to instrument, yet
keeping track of individual particles, has been mentioned. This technique development
is necessary to provide the maximum information from small particles.

      An investigation of the lead-chlorine-bromine correlation in atmospheric and
automobile exhaust samples by the use of intensity ratio techniques would provide very
useful information and should be undertaken as a separate program.

-------
                         APPENDIX
     GENERAL OUTLINE OF THE ELECTRON MICROPROBE
POTENTIAL AND APPLICATIONS TO AIR POLLUTION RESEARCH

-------
                                          A-l

                                      APPENDIX


                GENERAL OUTLINE OF THE ELECTRON MICROPROBE
           POTENTIAL AND APPLICATIONS TO AIR POLLUTION RESEARCH


       The utility of the electron microprobe method in the chemical analysis of micron-
 size volumes is well known.  The fields of study to which the method has been used
 encompass metallurgy, ceramics, geology,  biology, electronics, and others.  To date,
 most applications have been predominantly  concerned with the in-situ analysis of
 microscopic regions in solid materials where the specimen is essentially of infinite
 size compared with that of the electron-bombarded region.   Very limited applications
 involving the analysis of individual particles in the micron-size range have also been
 carried out.

      One type of application to which the microprobe method has not been extensively
 extended  but from which very useful chemical information can be derived is the analysis
 of particles of microscopic dimensions.   The complexity - and hence  the practical
 realization - of such an application is  critically dependent on the characteristics of the
 particles and the information desired.  The principal factors determining this com-
 plexity include the following:

      •  Specimen collection and preparation requirements

      •  The size range of the particle of interest

      •  The number of elements requiring  analysis

      •  The total number of particles requiring analysis

      •  The maximum number of particles which can be analyzed simultaneously

      •  The need for correlating X-ray data with individual particles

      •  The volume and element analysis resolution required

      •  The degree of accuracy required in the chemical analysis.

      Thus, in view of the variability to which these factors would be restrictive for a
given problem,  a method by which particle analysis could be carried out cannot be
stated in a simple manner.  An understanding of the restrictive nature of each of these
factors is necessary to determine  the procedures required to obtain the maximum
amount of information from a given situation.

      This section is directed to the extension of the microprobe method to the specific
application of particle analysis.  The procedures and instrumentation required for this
purpose are discussed.  In so doing, the limitations of the microprobe method - in terms
of the basic physics of the technique as well as instrument restraints - are cited.

      In spite of all the work being carried out related to air pollution, very little is
known about the composition and fine structure of the individual airborne  particles that

-------
                                          A-2

 seem to present major problems.   This is especially true of the particles in the size
 range  from about 1 micron and below where classical methods of identification such as
 chemical analysis, light microscopy, electron microscopy,  and X-ray diffraction are
 severely limited.  Considering the  possible high rate of reactivity because of high
 specific surface area and the degree of pulmonary retention of this size particles, it
 seems imperative that their characterization be attempted.  This information is desir-
 able, not only from the basic aspect of identification but also to  relate atmospheric
 particulates with specific source effluents, to understand more fully the basic aerosol
 mechanisms, and to assess the effects on materials, animals, and humans.

      One of the most powerful tools for  single-particle identification is the electron
 microprobe analyzer.  In this device small areas from a few tenths of a micron to
 several  hundred microns can be examined  nondestructively in several different modes
 to establish the spatial distribution and the identification of any element from atomic
 number  5 (boron) through 92 (uranium).  Simultaneous with elemental identification, the
 topography of a sample can be photographed to help  relate the probe information to light
 or scanning electron microscopy examinations.  Most elements  can be detected in a
 homogeneous sample at levels as low as  100 ppm, while isolated particles of an element
 can be identified if as little as 1 x 10"15  grams is present in a small area.

      There are many problems associated with the analysis of particles below 1 micron
 in size in the electron microprobe analyzer.   First  and foremost is the fact that most
 commercial microprobes have a beam size (and thus imaging resolution) on the order of
 1 micron, making imaging by the conventional techniques of back-scattered electron or
 specimen current images of little value.  The problem,  of course,  is that if the particle
 cannot be found by the light microscope incorporated in the analyzer or by one of the
 conventional scanning modes,  it is  impossible to determine  in advance where a particle
 of interest is located.  Even though the X-ray system may be sensitive to as  little as
 1 x 10-15 gram, it may be impossible to locate and  analyze  a particle with sufficient
 mass using  conventional techniques.

      It is generally agreed that there is little problem in the qualitative and quantitative
 analysis of particles down to a size range of 2 to 5 microns, depending on their density
 and conductivity.  Below this size,  the X-ray  intensity falls  off,  though smaller sam-
 ples can be  analyzed by ratios or by a known size correction.  Nonconductive particles
 present problems of their own because of defocusing and deflection of the electron beam
 due to  the charge buildup on the particle.   This effect often can be minimized by coating
the particles with a thin layer  of carbon or other suitable material.

      The qualitative and quantitative determinations with an electron microprobe de-
pend upon the quality and versatility of the  visual optical system, the resolution and
efficiency of the electron optics, and the  detection capability of the X-ray spectrometers.
Essentially  all electron microprobes available today can be expected to provide about
the same quality of information about a particulate sample.  Some microprobes are
more versatile than others  and some have  special additional features that make them
attractive; however,  at present,  all have virtually the same  electron-imaging resolving
power  and X-ray sensitivity.  The successful examination of particles of 1 micron in
 size  and below presently is limited  by the characteristics of the  current electron micro-
probes and the experience and skill of the operator.

     The ability to obtain specific chemical information on small particles with the
electron microprobe  is strongly dependent  on  certain physical limits.  These limits  are

-------
                                          A-3

 somewhat variable but are basically determined by instrument design and specimen
 characteristics as well as basic theoretical considerations.  In any practical situation,
 knowledge of these various limitations is necessary to provide a sensible approach to
 the problem of particle analysis.

      Some of the major factors which affect the chemical analysis  capability of the
 electron microprobe are:

      (1)   The size (volume) of material from which a chemical analysis can
           be made

      (2)   The precision and accuracy of the analysis

      (3)   The detection sensitivity for various elements

      (4)   The degree of specimen alterations due to exposure to the electron
          beam and vacuum environment.

 These factors, dependent on the physical processes of X-ray excitation and the large
 variation in specimen types, all affect the minimum analyzable volume, which can vary
 from less than 0. 01 cubic micron to greater than 1000 cubic microns.  The error in
 the analysis  can  range from less than 1 percent to greater than 50 percent, the limit
 of sensitivity to element detection can vary from 10 to 1000 parts per million, and
 specimen alterations can vary from little or negligible change to almost complete de-
 terioration.  Some of the principal variables are discussed in greater detail below.


 Incident Electron Energy

      The energy of the bombarding electrons  normally is,  for various practical and
 theoretical reasons,  held constant for a given probe analysis.  The  choice of beam
 accelerating voltage is dependent upon the  spectral line analyzed and, in the case of
 simultaneous multielement analysis, is  determined by the spectral line of shortest
 wavelength to be  analyzed.  The incident electron energy  is variable in most electron
 microprobes from about 2 to 50 kV and its magnitude has a  primary influence on the
 following:
      Electron Excitation Volume.  The electron excited volume can be roughly ex-
pressed in terms of beam diameter and the distance penetrated by electrons along the
incident as well as lateral directions.

      The size of the excited volume is a function of the range of the primary electrons
and hence their accelerating potential,  and also the critical excitation potential of the
particular characteristic X-ray line utilized for the analysis.  Two major dimensions,
the depth and width, of the X-ray-producing volume have been measured experimentally
as a function of accelerating potential and found to vary according to the well-established
electron range equation:(2)

-------
                                         A-4

where R is the electron range expressed as a distance times the sample's'density an:'.
Eo is the primary accelerating potential.  This range equation, suitably modified to ta«.e
into account the electron energy required to produce the characteristic X-ray under
investigation, has been used to calculate the volume contributing to the analysis througi.
direct electron excitation.  The important fact is that the volume contributing to the
analysis is a rapidly changing function of accelerating potential and that the analysis of
low-density specimens, with good spatial resolution, such as some types of air pollution
particles, requires  the use of low accelerating potentials.  At an accelerating potential
of 5 kV, the analytical volume in such a specimen is approximately three orders of
magnitude smaller than that which would result from the use of an accelerating potential
of 20  kV. Reducing the diameter of the electron probe at the low accelerating potential
will further reduce the size of the analyzed volume and improve spatial resolution,  al-
though not nearly so substantially so as obtained by the reduction of the primary
accelerating potential.

      The above  paragraph should be  modified, since this may not be necessarily true
for all preparations of air pollution particles.  A more general statement would be  that
in order to preserve spatial resolution (see sections on microprobe analysis  of parti-
cles), one must either use a short-range, low-voltage probe or study thin or widely
dispersed specimens of small size.  The use of low voltages for analyzers has been
covered by Anderson(2) and by Anderson and Hasler(3), which, for certain specimens,
have very high sensitivity.  Hall(4), on the other  hand,  has some very convincing argu-
ments in favor of high-voltage operation when examining thin sections  of biological
material.  The important point is that, by proper specimen collection and preparation
methods, samples that are similar  to thin biological specimens can be produced from
air pollution collections.   This is precisely the case if one  is interested in very small
(less  than 1 /LI) particles and it is possible to disperse these particles (for example,  in
a thin carbon film) so that X-ray background from the substitute does not interfere  with
the analysis.

      The effect  of the  critical excitation potential of the characteristic X-ray line of
the element under study on the volume analyzed is important also and it can be shown
that large increases in spatial resolution can be attained with respect to a particular
element by using an accelerating potential close to its critical excitation potential.   Al-
though this technique produces very small analytical volumes for the element in question,
it also produces  the generally undesirable consequence that the characteristic X-ray
emission of other elements of lower critical excitation potential studied in connection
with the first element will come  from comparatively large volumes resulting in the re-
ported elemental analyses not representing the same volume.  For example^), at
11 kV a zinc analysis using the ZnK&  line (Ec = 9- 66 kV) originates from a volume of
about 2 ju3, but at the same kV sodium analysis using its Ka line (£c =  1. 07 kV) originates
from  a volume of 75 /j3.  This situation can be improved by using the La line of zinc
(Ec = 1. 02 kV) and reducing the  accelerating potential still  further.  At 5 kV the volumes
contributing to the analyses of zinc  and sodium as well as all the lighter elements will
be about 2 jiP.

      From the above discussions it is clear that, generally, to attain the best possible
spatial resolution in a bulk specimen  while remaining within the bounds of X-ray emis-
sion microanalysis as commonly employed and yet maintaining the equality of the
spatial relationships of the elemental analyses, the longest wavelength characteristic
X-ray lines of the elements which are detectable  should be  used at the  lowest accelerating

-------
                                         A-5

potential compatible with the other basic aims of the analysis.  This consideration is
especially important when dealing with the light elements (Z <12) for which the critical
excitation potential is  low (of the order of 1 kV or less).

      Investigations of the effects of such parameters as accelerating potential and
critical excitation potential are necessary to demonstrate the maximum sensitivity ob-
tainable  when air pollution particles are  examined.


      Target Absorption.  The amount of absorption of X-rays  escaping from the sample
increases as the accelerating voltage increases.  Primarily, this is due to the greater
depth of  electron penetration, and,  consequently, the correction factors required for
quantitative analysis to account for  absorption are also increased,  thus decreasing the
correction's accuracy.


      Secondary Emission (Fluorescence).  The percentage of  secondary emission
(caused by the continuum) as compared with that of primary emission increases with the
incident  electron beam energy,  thus increasing the total X-ray excitation volume and
decreasing the accuracy of corrected intensity data for  quantitative analysis.
      Signal-to-Background Ratio.  The signal-to-background ratio, which affects the
minimum, level of element detectability, initially increases as the accelerating voltage,
EQ,  excee'ds the critical excitation energy, Ec,  of a given spectral line, and then de-
creases for large values of the ratio, EO/EC.
      Electron Background.  Total electron background (single or multiple scattered
electrons) increases as the accelerating voltage increases.  Since it contributes to the
measured background of the X-ray detection system, the minimum level of detectability
is reduced.
      Specimen Surface Conditions.  A very complete understanding of the nature of the
surface of a specimen is necessary for the proper interpretation of measured data.
High accelerating potentials cause greater electron penetration and decrease the sen-
sitivity of the analysis as far as surface conditions are concerned.  This decreased
sensitivity is,  of course, highly dependent on the magnitude of absorption which the
excited X-rays undergo and can be negligible in the case of a very-high-absorption
situation.
Specimen Characteristics

      Characteristics of the specimen may be the most important controlling factor(s)
in the analytical capabilities of the microprobe.  Some of the important characteristics
include:
      Density.  For a given accelerating voltage and analyzed spectral line wavelength,
the extent of electron penetration is determined by the density of the bombarded region.

-------
                                          A-6

      Conductivity.   The thermal and electrical conductivity of the specimen determir
 the stability of the electron beam and the ability to withstand the total energy in the
 incident beam without deterioration.
      Distribution of Element Analyzed.  A large concentration of the element being
 analyzed in regions surrounding the bombarded region is a very significant factor
 affecting the minimum detectable limit.
      Spectral Interferences.  The presence of other elements, which produce  spectral
 lines that interfere with the desired element analysis, is a very significant factor
 affecting the minimum limit of detectability.  Misidentification can occur when lines
 from two different elements in the sample appear at the same wavelength position in the
 spectrum.  One example of this is the near  coincidence of Ti K£ and V K
-------
                                         A-7

      Focused Beam Size.  The minimum usable diameter of the focused beam for X-
 ray analysis is limited by mechanical design compromises necessary to incorporate
 X-ray spectrometers and light microscope systems for the analysis of opaque speci-
 mens.  The minimum diameter is about 0. 1 to 0. 5 micron at 20 kV.  At low accelerating
 voltages, below approximately 5 kV, the thermal spread of the thermionic emitting
 source greatly affects the minimum beam diameter because  of the chromatic aberration
 sensitivity of magnetic lens focusing, and very small beam sizes generally cannot be
 obtained.
      Probe Current Density.  The current density of the focused beam is determined
by the brightness of the electron source, the stability of the lens and accelerating-
voltage power supplies, the  spherical aberration of the final demagnifying Lens, and
the spread in thermally emitted electrons from the source. The limiting factor at the
present time, from a practical point of view, is the brightness of the emitter source.
      X-Ray Analysis System.  Dispersive Type.  The efficiency of X-ray detection and
selection varies with the spectral line analyzed and is dependent upon the mechanical
precision of the spectrometer, the perfection and efficiency of the analyzing crystal, the
type of spectrometer, and the efficiency of the X-ray detector.  The basis for dispersive
spectrometer design is a practical compromise for the requirements of efficient X-ray
detection for the total spectrum of X-rays generally analyzed,  i. e. ,  1 to  100A.
      Energy-Dispersive Type.  Energy-dispersive X-ray analysis can be carried out
using solid-state lithium-drifted silicon detectors in conjunction with multichannel pulse-
height analyzers.  The performance of this type of X-ray spectrum analyzer is limited
by the energy resolution of the detector,  the maximum counting rate of the detector/
electronics system, and the minimum photon energy it is capable of detecting.  It may
be used for high concentration levels of element analysis, but its poor signal-to-
background performance limits its usefulness for low-concentration analyses.   How-
ever, its higher efficiency for photon collection and its ability to analyze the complete
spectrum (lower limit of about 4A) simultaneously can,  in certain cases, substantially
decrease the time of analysis.  This efficiency also can be an overriding factor in de-
termining which type of detection system to employ (dispersive or energy-dispersive)
in the measurement of low signal levels where background is not predominantly of the
X-ray type  (see following item).   This is an important consideration in the examination
of air pollution particles and may point the way to use of a scanning electron microscope
equipped with an energy-dispersive detector as a way of investigating very small parti-
cles.  This will be discussed in detail in a later section.  Use of energy-dispersive
detectors of the gas-flow proportional type can also be used in a similar manner for
low-energy photon detection that is inaccessible to the solid-state detector, i. e. ,
greater than 4A.


      Noise Level.  In the  analysis of low X-ray intensity signals, two factors  which can
restrict measurement precision  - and which are not related to the physical processes
involved in the electron beam/target interactions - are  electronic noise  and cosmic
radiation.  Electronic noise may be related to random pulses caused by limitations in
the X-ray detector/preamplifier system.  The degree to which cosmic radiation inter-
feres is dependent upon the amount of shielding present.   With little  shielding,  a

-------
                                          A-8

count-rate level of approximately 0. 3 cps  can be experienced.  Instrument location ir.
the center of a large building (preferably,  the basement),  could reduce this level to
0. 1 cps or less.
                            Microprobe Analysis of Particles
      From the preceding discussion,  it is easily seen that spatial resolution and detec-
tion sensitivity are extremely important topics, considering that the basic goal of
microanalysis  is chemical analysis of the smallest possible volume with the greatest
possible sensitivity.  The pursuit of this ideal is important, especially, in the field of
small air pollution particles where the microstructures of interest are generally much
smaller and their spatial interrelationships more complex than those most commonly
investigated in the other fields of electron probe application.  Spatial resolution(2) is
determined by  the volume of the specimen which,  through excitation by the primary
electron beam, contributes to the analysis.   Detection sensitivity is commonly con-
sidered in two  categories.  The  absolute detection limit gives the smallest number of
atoms,  commonly expressed in grams, necessary for the analysis.  The weight frac-
tion or  relative detection limit describes the smallest fraction by weight detectable in
the sample irrespective of the volume analyzed.

      The weight-fraction detection sensitivity of an element is determined by the inten-
sity of the characteristic X-ray  line of the element utilized for the analysis and the
intensity of the background radiation associated with the characteristic line.  These  in-
tensities integrated over time are dependent, as mentioned previously, on the accel-
erating potential used,  the primary electron beam current,  the efficiency  of the X-ray
crystal and detector used in the  X-ray detection system, and the matrix in which the
element is found.  The weight fraction detection limit is defined here as the  concentra-
tion of the element that is required to  produce an X-ray intensity greater than the in-
tensity of the background at the position  of the analytical line by an amount equal to
three times the standard deviation of the background intensity.  The absolute detection
limit is determined by the weight-fraction detection sensitivity and the mass of material
excited.

      In light of the previously stated ideal of detecting the smallest number of atoms
of an element in the smallest possible volume, it is of primary importance-.to consider
the absolute detection limit that  is the analytical definition of this ideal.  The absolute
detection limit is the smallest number of atoms of the element required for the analysis
and is calculated by multiplying  the weight-fraction detection sensitivity by the product
of the excited volume of the sample and the  sample's density.  The size of the excited
analytical volume, which is a function of the range of the electrons and therefore of the
primary accelerating potential as discussed above, plays  a dominant role  in determining
the number of atoms detected, with the result that those general conditions that produce
the best spatial resolution also contribute to the best absolute detection limits.

      The  choice as to which of the characteristic X-ray lines of an element to be em-
ployed in a particular investigation should be carefully considered in light of the detec-
tion sensitivities achievable with each line.   Each characteristic line of an element will
in principle have a different detection  sensitivity as the accelerating potential,  sample
composition, and efficiency of the detection system for each line are changed.

-------
                                         A-9

      Two basic questions may be asked in an elemental analysis.  The first question
 ie qualitative and seeks the elements present.  The second question, predicated on the
 first, is quantitative and seeks how much of each of the elements is present.  A qualita-
 tive analysis by X-ray emission, therefore, depends  on correctly identifying the Lines
 of the characteristic X-ray spectrum as to their source elements,  while a quantitative
 analysis depends on an accurate measurement of the  X-ray line intensities as related
 to a reference standard of known composition.  A qualitative analysis is,  therefore,
 accomplished by examining the X-ray spectrum of the specimen and comparing the wave-
 lengths of all the characteristics lines present with one of the several available  refer-
 ence compilations of X-ray wavelength versus elemental atomic number.  Since the
 microprobe has crystal spectrometers which through suitable gearing give the wave-
 length directly in angstrom units, or in any case, since the wavelength is easily deriv-
 able from the & angle,  this task becomes one of simply comparing the wavelength read-
 ing of the peak of the spectral line with the reference list in order to identify the element
 that is responsible for  that particular line.  In most cases this is rather easy, consider-
 ing that the X-ray spectra of the inner shell transitions are comparatively uncomplicated.

      In this way, information about a number of elements suspected of playing an im-
 portant role in the chemistry of the sample can be quickly obtained.  In the general case,
 when a complete qualitative analysis is desired,  the X-ray spectrometers are run
 through the complete wavelength range so as to detect as many of the elements present
 in the sample as possible.  The practice of surveying as much of the wavelength spec-
 trum as possible is  a good one to adopt even for so-called known samples.  Once the
 complete  spectrum has been recorded the problem of identifying the source  elements
 may be attacked by identification of the individual lines.


                         General Aspects of Particle Analysis


      The microprobe is capable of providing  chemical information from particles which
 in many cases is beyond the limitations of other analytical methods.  For example, this
 situation may be true when

      •  The sample size is very small

      •  The analysis of individual particles is desired

      •  Nondestructive analysis is required

      •  Increased sensitivity and accuracy in the detection and analysis of
        minor  constitutes is necessary

      •  The valence state of the elements,  i. e., the state of chemical com-
        bination, is important

      • Improved quantitative analysis is required.

      The capability of most overriding importance is that of mass sensitivity, i. e. ,
the ability of derive useful information from material of microgram quantities or less.
 This capability provides the basis for identifying constituents  in a specimen which,  in a

-------
                                         A-10

macroaample, occur in very low average concentrations but, in a microsample,  occur
in a percentage amount which fluctuates from high to low values. The points to be con-
sidered for detecting such "needles in the haystack" and characterizing them within the
limits of the microprobe method are discussed below.  In this discussion,  the particles
being sought are assumed,  for the sake of simplicity, to be  composed of a single ele-
ment.  The arguments  presented can be easily extended to the multielement case using
the weight fraction of the element being analyzed as a modifying factor.
Definition of Analysis Sample

      The first point to be determined is the sample size (as distinguished from parti-
cle size), which must undergo analysis to provide the desired information.  This deter-
mination  requires a definition of the minimum number of particles which will give a
satisfactory statistical representation of the total particle aggregate.  The definition
must take into account the relative occurrence of the particle types of interest.  In this
respect, the minimum sample then will be defined by the particle type which occurs
least frequently and should include a number of such particles.   Arbitrarily, for the
purposes  of discussion here and below, ten such particles will be considered sufficient
for statistical purposes.  Thus, if the least frequent particle type of interest occurs
once in n  particles, then the minimum number of particles requiring analysis is lOn.
If the frequency of occurrence of a given particle type is unknown and is the informa-
tion being sought, then an assumed minimum level that is  based on  some practical
criteria must be  used. In so doing,  the criteria used must recognize the  complexity
and  cost of the analysis.  In the event that the number of particles is limited,  i. e. ,
below the lOn required,  then the statistical characterization of the aggregate will of
necessity be determined by the total number available.

      The above comments  point out one of the major problems  associated with particle
analysis;  that is, the detection of a minor constitutent requires  the  analysis of a very
large number of particles.  For example, if a constituent  is being sought  which occurs
once in every 10,000 particles, the analysis of 100,000  particles is required.  Although
individual analysis of all  particles is not necessary, the analysis will in any event be
very repetitive in nature, thus making it advisable to pursue automated procedures.
The  possibility of sample treatments that will concentrate the particle types of interest
should also be evaluated thoroughly in the event that analysis cost is excessive.  Also,
the  possible use of other  techniques for pinpointing certain particle  types  should be
pursued,  e.g., detection by luminescence behavior under  electron bombardment (visible,
infrared,  ultraviolet).
Specimen Preparation

      The preparation of particles for microprobe analysis can be carried out in several
different ways; for example:

      •  Metallographic mounting, in which the particles are embedded in a
         resin binder material such as bakelite or epoxy (limited  to fairly
         large particles)

-------
                                         A-ll

      •  Briquetting,  in which the particles are compressed to yield a dense
         composite (when a large number of particles are available)

      •  Substrate mounting, in which the particles are placed on a substrate
         and affixed to it by the use of collodion or a  vacuum evaporated
         coating

      •  Carbon-film embedding, in which the particles are placed on a near-
         dry collodion-amyl acetate film, carbon coated,  and the collodion
         film dis solved.

Of these methods,  metallographic mounting is best from an analysis standpoint but is
limited to particles of a few microns in size or larger and materials that are not soluble
in resins.  Where this mounting method is applicable, quantitative chemical analysis
can be successfully carried out using standard correction procedures.  Briquetting is
useful for submicron-size particle aggregates where  quantitative analysis on a micron
scale is desired, sufficient specimen material is available, and the analysis of con-
stituents of very low concentration is not required. Substrate mounting or carbon-
film embedding is necessary when the amount of sample material is limited,  the mech-
anical alteration of the particles is not allowed, the identification and analysis of
individual submicron particles is required, or when only qualitative or semiquantitative
chemical analysis is needed.  Carbon-film embedding may be necessary to provide
good thermal contact for dissipation of heat generated by the beam, and,  in addition, it
provides maximum sensitivity to the particle analysis through the reduction of X-ray
background and electron diffusion.  The factors  affecting the particle density per unit
area for optimizing the efficiency of the analysis problem are discussed further below.


Single -Particle Detection Limits

      One aspect of particular concern in particle  analysis is the  size of the smallest
particle that  can be detected as well as the ease with  which the detection can be ac-
complished.   Many factors  must be considered to answer this question including the
photon excitation rate produced by the probe, the composition of the particle in ques-
tion,  the photon-detection efficiency of the spectrometer,  the background level, and the
statistical character of photon production and measurement.  The discussion in the
remainder of this  section will cover these points,  with the exception that only pure-
element particles will be considered.
      Typical Instrument Performance. ^) Consider a target of copper bombarded by a
30 keV electron beam of a current density equal to 100 amperes /cm .   This current
density is equivalent to 0. 1 jua in a 1 -micron-diameter beam and is representative of
typical microprobe operation.   (This current density can be increased by a factor of
about 5 if all conditions are optimized. )  Under these conditions, a Cu Ka counting
rate of about 100, 000 cps can be measured by a focusing -type dispersive spectrometer
with a peak to background of about 1000.  Assuming the excited volume is about 3 cubic
microns,  then the counting  rate that would be measured from a cubic particle of 0. 1-
micron dimension will be 30 cps.   In other words, if a copper particle of this volume
were bathed in an electron beam under the conditions given and no X-ray background
was  produced by the electrons not striking the particle, the counting rate of 30 cps would

-------
                                         A-12

 be recorded by the spectrometer.  A copper particle of this volume represents apprc :-. i
 mately 10-14 grams.  This calculation provides an indication of the practicality of de-
 tecting particles in the submicron-size  range.  Thus, the experimentally measurable
 counting rate per picogram of copper, using a dispersive focusing spectrometer for  the
 conditions stated, is approximately 3000 cps.  The dependence of counting rate per
 picogram on critical excitation and accelerating voltages,  spectral series,  and specimen
 density has already been discussed.
      Probe Diameter and Current Density. (°) The question of the improvement that can
 be obtained by reducing the probe diameter deserves  comment at this point.  In this
 respect, it should be noted that, for a fixed electron-source brightness and spherical
 aberration constant of the final probe forming lens, the maximum current obtainable in
 the focused probe is proportional to the 8/3 power of  the probe diameter.  Thus, the
 current density varies directly with the 2/3 power of  the probe diameter.  This cir-
 cumstance thus will decrease the estimated counting  rate given above if the probe
 diameter is decreased and will increase it if the reverse is true.  Three  additional
 factors affecting the most useful probe size are (1) the ability of the prepared specimen
 to dissipate the heat generated by the beam,  (2) the X-ray background generated by
 electrons not striking the particle of interest, and (3) the real particle density of the
 prepared specimen.  All of these factors are controllable, at least to some extent, by
 specimen preparation procedures.  It should be noted that,  although the probe current
 density varies directly with the 2/3 power of the probe diameter, the 8/3 -power de-
 pendence of the total current on the probe diameter will cause the X-ray background  -
 from other than the particle of interest  - to increase  in the same manner, i. e. , 8/3
 power.   Thus, any benefit from increasing the current density can be realized only if
 conditions for minimizing this background are rigorously adhered to.  Finally,  it should
 be mentioned that improvement in the typical brightness of the electron source and
 spherical aberration of the final probe forming lens are possible but  involve major de-
 sign innovations in current microprobe instrumentation.


      Energy -Dispersive X-ray Detection. W  In the above discussion, the X-ray detec-
 tion system was assumed to be  a dispersive -focusing- spectrometer type.   The use of an
 energy-dispersive X-ray detection system  could, in certain cases, allow an increase in
 the efficiency of total X-ray counting by a factor of 10 or more. Energy-dispersive
 systems using solid-state,  lithium-drifted  silicon detectors are limited in their total
 count- rate capacity (energy resolution decreases with counting rates  above about
 25,000  cps).  In addition, their peak-to-background performance in the measurement
 of X-rays from pure elements  is quite low,  varying from  5 to 25 in the wavelength
 range 1 to 4A. Recent availability of solid-state detectors, with a factor -of -2 im-
 provement in energy resolution, should  increase these peak-to-background values also
 by a factor of 2.  Energy-dispersive detectors of the  gas -flow proportional type have
 poorer  photon-energy resolution above 3 keV (4A).  Below this energy,  however, their
 resolution can be comparable or better.  The relative mechanical simplicity of the gas-
 flow type compared  with the solid-state type is also a favorable practical  characteristic.

      It should be noted that the peak-to-background characteristic of the  dispersive
 focusing spectrometer may not be applicable at low counting rates.  Under conditions
 where the X-ray signal is  10 cps, a calculated background -count level of 0. 01 cps
would be obtained if a 1000 to 1  peak-to-background ratio applied.  Since the cosmic
 radiation background can be 0.3 cps if little shielding  is present, the  effective peak to

-------
                                   A-13 and A-14

background is reduced to 33, thus making the dispersive and nondisperaive systems
comparable in this respect.  The higher photon-collection efficiency of the energy-
dispersive system may thus be quite practical in certain aspects of particle analysis.

      Hall' ' haa also commented on the possibility of favorable energy-dispersive
detection using very small probe diameters (0. 1/u to 0. 03/u) and thin samples.  This
could be the case if a scanning electron microscope equipped with an energy-dispersive
detector were used on an air pollution sample suitably dispersed on or in a material of
low X-ray background.   The lower  elemental resolution would  still be a major limita-
tion,  however.
                                   REFERENCES


(1)  White, E.  W. , Denny,  P.  J. , and Irving,  S.  M. , "Quantitative Microprobe Analysis
    of Micr©crystalline Powders", in The Electron Microprobe, proceedings of the
    symposium sponsored by the Electrochemical Society,  Washington, D. C.,
    October, 1964.  (Editors:  T. D. McKinley, K. F. J.  Heinrich, and D. B.  Wittry;
    John Wiley and Sons, 1966, pp 795-796).

(2)  Anderson,  'C. A. , "An Introduction to the Electron Probe Microanalyzer and Its
    Application to Biochemistry", chapter in Methods of Biochemical Analysis,  Vol XV,
    pp 147-270, John Wiley, Interscience,  1967.

(3)  Anderson,  C. A. , and Hasler in X-Ray Optics and Microanalysis (Eds. , Castaing,
    Descamps, and Philibert), Hermann,  Paris,  1966, p 310.

(4)  Hall,  T. , in Quantitative Electron Probe Microanalysis (Ed. , K. F. J. Heinrich),
    U. S. Department of Commerce, National Bureau of Standards, Special Publication
    298, October, 1968, pp 286-287.

(5)  Macres,  V. C. , Materials Analysis Company,  private communication.

(6)  Ibid.

(7)  Ibid.

(8)  Hall, T. , op cit. ,  pp 295-296.

-------