EPA-600/4-77-042
October 1977
Environmental Monitoring Series
                ELECTRICAL DETECTION PROBLEMS IN
                WATER ANALYSIS BY SPARK  SOURCE
                                  MASS SPECTROMETRY
                                       Environmental Research Laboratory
                                      Office of Research and Development
                                     U.S. Environmental Protection Agency
                                             Athens, Georgia 30601

-------
                RESEARCH REPORTING SERIES

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

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

This report has been assigned to the ENVIRONMENTAL MONITORING series.
This series describes research conducted to develop new or improved methods
and instrumentation for the identification and quantification of environmental
pollutants at the lowest conceivably significant concentrations. It also includes
studies to determine the ambient concentrations of pollutants in the environment
and/or the variance of pollutants as a function of time or meteorological factors.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
                                     EPA-600/4-77-042
                                     October 1977
  ELECTRICAL  DETECTION PROBLEMS
         IN WATER ANALYSIS
BY SPARK SOURCE  MASS  SPECTROMETRY
                 by

           W.  W. Harrison

       Department of Chemistry
        University of Virginia
   Charlottesville, Virginia  22903
       Grant No.  R801829-03-0
           Project Officer

           John  M.  McGuire
   Analytical Chemistry Branch
Environmental Research Laboratory
     Athens, Georgia  30605
 ENVIRONMENTAL RESEARCH LABORATORY
 OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
      ATHENS, GEORGIA  30605

-------
                          DISCLAIMER
     This report has been reviewed by the Environmental Research
Laboratory, U.S. Environmental Protection Agency, Athens, GA,
and approved for publication.  Approval does not signify that
the contents necessarily reflect the views and policies of the
U.S. Environmental Protection Agency, nor does mention of trade
names or commercial products constitute endorsement or
recommendation  for use.
                               11

-------
                            FOREWORD

      Nearly every phase of environmental "protection depends on a
capability to identify and measure specific pollutants in the en-
vironment.  As part of this Laboratory's research on the occur-
rence, movement, transformation, impact, and control of environ-
mental contaminants, the Analytical Chemistry Branch develops and
assesses new techniques for identifying and measuring chemical
constituents of water and soil.

      Spark source mass spectrometry, a technique that has recent-
ly been applied to the elemental analysis of water and sediments,
meets many of the analytical chemist's requirements for a method
that quickly and accurately identifies trace elements in environ-
mental samples.  The research reported here describes the devel-
opment of control features to make the analytical technique more
quantitative and easier to use.
                                David W. Duttweiler
                                Director
                                Environmental Research Laboratory
                                Athens,  Georgia
                              iii

-------
                            ABSTRACT

      This report describes research on factors affecting accura-
cy in spark source mass spectrometry (SSMS), and the application
of SSMS to water-related problems.

      This project has also involved the development of instru-
mental additions to improve the operation and control of SSMS
procedures, particularly with respect to making the technique
more quantitative and easier to use.  The application of SSMS to
specific sample types is also demonstrated.

      This report was submitted in fulfillment of Grant No.
R801829-03-0 by the University of Virginia under the sponsorship
of the U.S. Environmental Protection Agency.  This report covers
the period September 1, 1971, to August 31, 1975, and work was
completed as of August 31, 1975.
                               iv

-------
                            CONTENTS

Foreword	iii
Abstract	   iv
  I  Introduction  	    1
 II  Conclusions and Recommendations	    4
III  Experimental	    5
 IV  Results and Discussion  	    8
  V  References	   38
                                v

-------
                          SECTION I

                        INTRODUCTION


     An increased recognition of the role of trace elements
in plant life,-*- animals,^ the human body^>^ industrial pro-
ducts^ »6 and processes,  etc. has made the scientific com-
munity, as well as the general public, more aware of the need
to have available dependable and comprehensive monitoring
techniques to readily determine what inorganics are present
in water, either naturally or from industrial or municipal
discharge.  Many techniques have been applied to water ana-
lysis in recent years.  Spectrophotometric methods^j^ are
quite sensitive, accurate, and inexpensive, but they are sub-
ject to interelement interferences, respond to only one element
at a time in their usual practice, and often require consider-
able premeasurement steps, such as separations.  Certain
electroanalytical methods, particularly polarography,^0 have
reported relatively simple procedures and good sensitivity.
However, they are responsive mainly to metals, and not all
of those.  Flame emissionll has been used for metal determi-
nations in water, but only a limited number of metals are
normally determined.  Atomic absorption is sensitive, simple,
accurate, and relatively interference free, but it lacks the
sensitivity necessary to examine certain elements at very low
levels.  A single ideal analytical technique to cover all of
these needs might (1) be rapid, (2) be extremely sensitive,
(3) be accurate, (4) be responsive to all elements, metals,
and nonmetals, (5) provide both a qualitative and quantitative
analysis for all species simultaneously, (6) require minimal
sample preparation,  (?) exhibit uniform sensitivity for all
elements, (8) be free from interference or matrix effects,
and (9) have a simple and convenient data acquisition readout
system.  Though spark source mass spectrometry does not meet
all of these ideal criteria, it in fact does come closer than
any other currently available trace analysis method.

-------
     Spark source mass spectrometry  (SSMS) was developed in
the 1960's, largely in industrial laboratories.  Hannay and
Ahearn^-2 at Bell Laboratories did much of the early ground-
work; Ahearnl3 has reviewed  the technique.  Basically, as
shown in Figure 1, SSMS utilizes a high voltage spark  (5-80KV)
to ionize a sample for subsequent separation in tandem electro-
static and magnetic fields,  resulting in identification and
analysis of the ions on the  basis of mass to charge.  The
energetic ion source provides essentially uniform sensitivity
for all elements in a single scan and allows analysis of re-
fractory, nonvolatile materials such as metals and alloys.
A mass range of about 6 to 250 is used routinely, covering
lithium through uranium and  thus effectively encompassing
the entire elemental range.  Sensitivities to IpP'b are obtain-
able, and all elements are detectable at sub-ppm.  Serious
interferences are few because of the polyisotopic nature of
most elements, and matrix has been shown to have relatively
little effect on elemental sensitivities.  Qualitative and
quantitative results for 30  to 5Q elements can be obtained.
Accuracy of photographic readout is + 20% to 30%.  Electrical
detection allows the more rapid acquisition of data by scann-
ing methods, but accuracy and precision are still in the 207o
to 307» range.  Peak switch integration methods provide 570 to
10% accuracy.

     However, SSMS  is not a  simple technique.  Operator skill
and experience are  important in obtaining meaningful  results.
The need for control of experimental variables is recognized.
Our research has involved the development of control features
to make SSMS more quantitative and easier to use for trace
element analysis.

-------
              No. I  Slit (Ion Acceleration)

                 No. 3 Slit (Beam Defining)

                    Y-deflector
Spark    H  I  u
Source 2-f-LLL.


        No. 2 Slit
        (Earthed)
                     Electrostatic
                     Analyzer
                                              Electrical
                                              Detectors
                           / Photographic Plate
                                                         Slit
                                 Magnetic  Analyzer
Figure  1.  Schematic diagram of MS-702 spark source mass spectrometer

-------
                         SECTION n:

               CONCLUSIONS AND RECOMMENDATIONS
     If close attention is paid to experimental variables,
spark source mass spectrometry can serve as a complementary
technique to other analytical methods for obtaining quali-
tative and quantitative information concerning trace elements
in water and related samples such as sediments and tissues.
Control features added to the spectrometer allow these
analyses to be performed more conveniently and precisely.


     A spark source mass spectrometer is an expensive and
complex unit, and thus its widespread use in water analysis
is precluded.  For large regional, state, or federal labora-
tories, however, this technique is invaluable for use on
materials that require complete elemental screening.  A
spark source mass spectrometer .should be considered for in-
clusion in such facilities.

-------
                        SECTION  III

                       EXPERIMENTAL
APPARATUS

     The basic AEI MS-702 spark source mass spectrometer and
experimental parameters have been previously described.^  The
standard AEI electrical detection systemsl5 was added and used
for all data in this study.  Appropriate magnet current,
acceleration voltage, electron multiplier voltage, and sen-
sitivity settings were determined for each isotope measured.
Modifications or additions to the standard system include:

     A spark gap monitor pickup coil mounted externally in a
source port fed an RF voltage to a Tektronix Model R564-B
oscilloscope for display.

     A modification of a Kennicott telescope mount-1-" was used
with a 30 x optical system, cross-hairs, and calibrated re-
ticle to carefully align the electrodes on the ion beam axis
and to view the discharge during sparking.  Rack and pinion
focusing of the shallow depth of field optical system allowed
measurement of spark-to-No. 1-slit distance on a calibrated
scale.

     The standard spark shield was modified to provide a light
entrance from the top (for the overhead light source) and a
2.5 cm. square front viewing window.  The latter allowed full
view of both electrode pairs and was covered with a replace-
able glass plate held in place by tabs cut into the spark
shield.

     A continuously variable (0.0.012-inch) V-shaped collector
slit was substituted for the standard two-position slit.
     The standard AEI source chamber was replaced by a larger
unit, designed and constructed in our machine shops to accom-

-------
modate a 1200 liter/sec oil diffusion pump, cold trap-baffle,
and gate valve  (all from Varian, Vacuum Division, Palo Alto,
Calif.).  The seldom-used 2000 time constant for magnet scanning
was also replaced by a more useful 150 time constant.

     A linear ratio output was obtained   for peak area com-
parison by inputting the logarithmic output of the AEI ratio
amplifier to an antilog converter consisting of an inverting
amplifier  (Analog Devices, Model 141C) and a logarithmic
function module  (Teledyne Philbrick, Model 4350) wired in the
antilog configuration.  The linear ratio was input to a V-F
converter and counter  (Heath UDI, Model EU-805) and the output
pulses were directed to a 4096 channel MCA (EDAX, Model 706)
for spectra accumulation.  Each channel of the MCA can accept
up to 10^ counts.  A visual readout module (EDAX, Model 871N),
which intensified and integrated operator-selected channel
bands of the MCA cathode ray tube display, was used to obtain
all peak areas.  A conventional oscillographic recorder was
also available.

ASHING OF BIOLOGICAL TISSUE

     The tissue is dried at 105°C in an oven for several hours
and preferably  overnight.  Vacuum drying at 100°C in a "drying
pistol" can be  used to accelerate the process and also provide
protection from contamination.  One-gram tissue samples are
taken, if  available, although samples as small as a few mg
have been  analyzed.  Large tissue sections (^ 1 g) should be
cut with a scalpel into smaller pieces to facilitate both
drying and ashing.  Plastic or Teflon vials are used for sam-
ple storage.  Quartz ashing boats are preferred to reduce
elemental pickup during HTDA and LTDA.  Quartz beakers or
flasks are used for wet ashing.

High-Temperature Dry Ashing

      The  dried  tissue  is  ashed  at 550°C in a muffle  furnace
overnight.   The quartz boats  in the furnace are protected
from  any material  falling from  the  walls  and  ceiling of  the
furnace by insertion  into 1 1/2  inch Pyrex  tubes.  The ashed
residue may  be  scraped from the quartz  boat and mixed direct-
ly with 50 mg  of graphite.   This is transferred to a plastic
vial containing a  plastic  pestle,  and shaken  for  15  minutes
on a Wig-L-Bug  to  achieve a homogeneous mixture.  Alter-
natively  (and  the method  we prefer),  the  ash  residue is
moistened with  a  few drops of  1:1 HN03  and  loosened  with a
spatula.   The  dissolved-suspended mixture is washed  from
the boat  into a Teflon beaker,  the  graphite added, mixed
to a  slurry, and  the material dried under an  IR lamp.  This


                                6

-------
residue is then treated on the Wig-L-Bug as above.  Two
electrodes are tipped out by using the standard AEI Sample
moulding disc.

Low-Temperature Dry Ashing

     The dried, chopped, tissue is placed in a quartz boat
and ashed for several hours (or overnight).'  The material may
require, depending upon effective surface area, turning or
breaking up at selected intervals.  Hair samples oxidize very
readily with no agitation; however, liver tissue may form
hard oxide encrusted particles that resist further oxidation
unless broken up.   The ash residue may be transferred to a
plastic vial and homogeneously mixed with graphite in the
Wig-L-Bug.  For very small samples, a few drops of 1:1 HNOs
is helpful in collecting the residue, with subsequent wash-
transfer into a weighed amount of graphite for drying and
homogenization.

Wet Ashing

     Although we have used HN03-HC104 wet ashing for the
analysis of many biological materials, particularly samples
of 50 mg, there are certain hazards which can make a written
"standard" procedure dangerous.  The solution must be taken
to or near dryness to evolve the excess HC104 and this can
cause difficulties if done too rapidly.  In general, we try
to use one of the dry ashing procedures for all SSMS tissue
analyses.

-------
                         SECTION IV

                  RESULTS AND DISCUSSION
INSTRUMENTAL DEVELOPMENTS

Spark Gap  Control

     One of the more difficult parameters  to maintain con-
stant is the spark gap between the electrodes.  Many in-
vestigators make no. particular effort  to monitor and control
the gap.   However, it has  long been known  that  interelectrode
gap width  can be significant.  Woolston and Honig^8 demon-
strated that the energy distribution of ions in the beam  is
affected by gap.  Bingham  and Elliott-1-5 cited gap effects
in high accuracy peak switching analysis,  as have Colby and
Morrison.    Konishi and Nakamura   have noted gap effects
on ion ratios using photographic detection.  This laboratory
has also reported effects  of gap width on  sensitivity^1-23
which showed marked changes in analytical  ion yields with
changing gap width for aluminum, copper, and steel matrices.
More recent work with compacted graphite samples indicates
an even greater dependence.  The Autospark (A.E.I.), a
standard part of the electrial detection system, is useful
for maintaining spark ignition, but its cyclical nature
continuously varies the interelectrode gap.  Manual gap
adjustment, as determined  by oscilloscopic display of the
breakdown  voltage, has been our standard procedure in those
studies calling for maximum precision, but this is tedious
and keeps  the operator constantly engaged.  To  obviate these
difficulties, a unit has been designed and constructed in our
laboratories which automatically maintains a preselected gap
width, correcting for both short term  and  long  term effects of
electrode  deterioration.   The unit described in this paper is
specifically designed to coordinate with a commercially
available  electrial detection package  currently in use in
many laboratories.
                               8

-------
     The automatic gap controller utilizes the amplitude of
the RF voltage developed between the electrodes as the con-
trolling signal.  The amplitude to which the RF builds before
breakdown is directly proportional to the spark gap width.
Our Automatic Gap Unit (AGU) continuously adjusts one elec-
trode in order to maintain a fixed RF voltage between the
electrodes, and thus a constant gap.

     The AGU was designed and constructed to be used with an
AEI MS-702 spark source mass spectrometer and to utilize cer-
tain AEI Autospark components.  The entire schematic of the
AGU and associated Autospark components is shown in Figure 2.
     Figure 2.  Schematic diagram of automatic gap unit,
                See reference 22 for details.

-------
     The AGU has proved to be extremely useful for a broad
range of sample types.  It is capable of maintaining a stable
gap with close tolerance at widely varying sparking conditions.

Voltage vs. Current Control--

     A device to maintain automatically the spark discharge for
SSMS may be current or voltage controlled.  With the current
control mode,  the current flowing in the spark circuit is
sensed and used as the governing signal while in the voltage
type, the RF breakdown voltage between the electrodes acts
as the control signal.  Figure 3 demonstrates another area
in which the AGU has proved to be helpful.  The trace shows
variations in gap width while sparking a particularly low
impurity graphite sample  (such as encountered in water ana-
lysis).  This electrode type often has a tendency to partially
fragment, producing small carbon particles which may momen-
tarily short out the electrode gap.  The Autospark (first
ten minutes of the trace) was unable to keep the spark running
continuously without operator intervention.  Points (w), (x),
(y), and (z) are such points where a manual restart of the
spark was required.  The carbon fragments drastically disturbed
the normal operating rhythm of the Autospark, because it is
actuated by current.  When the gap is shorted by a carbon
fragment, the current drops to zero.  The Autospark, however,
cannot distinguish this from an open circuit gap, because it
too results in zero current.  Thus, the Autospark must cycle
in order to reinitiate the spark.

     The AGU repsonds quickly and properly to the carbon frag-
ment phenomenona, resulting in a minimal amount of off-time.
This is shown in the last eight minutes of the scan in Figure
3.  When the gap is shorted by a fragment, the voltage be-
tween the electrodes decreases (a), but this drop is unam-
biguously interpreted as an electrode short, thus calling for
a electrode separation.  The controlled electrode backs off
and the fragment is ejected from the gap, leaving the inter-
electrode distance too large.  This results in a momentary
increase in RF  (b) above the present level.  The AGU also
senses this and rapidly decreases the gap width to the desired
maintenance level.

     Need for Gap Control—The effects which may be produced
by a variation in gap width point to the need for rigid spark
gap control.  This is illustrated in Figure 4 where the
cyclical nature of the Autospark is again evident.  Each point

                              10

-------
X
h-
g
<:
Q_
<
0
     0
24      6     8     10     12    14     16

   SPARKING  TIME   (mm)
     Figure 3.  Gap width stability comparing the Autospark  and
                AGU (poorly sparking compacted sample).   Points
                (w), (x), (y), and (z) are where manual  re-
                initiation of the spark was required.   (a) Momen-
                tary narrow gap; (b)  momentary wide  gap.  Moderate
                vibration used throughout.
                               11

-------
in the two plots, offset from each other for clarity of
graphing, is an integration of the analyte ion  (39K+1) for a
total ion beam charge accumulation of 0.1 nC.  These were
taken every ten seconds to show how the effects of gap width
can be transmitted to the final data.  The relative standard
deviation for the points in plot  (a) is + 8.9%, as opposed
to + 2.1% for those of plot (b).  Techniques and instrumen-
tation in SSMS are now to the point where gap effects of this
magnitude may contribute significantly to the overall pre-
cision and accuracy of an analysis.  Through very careful  •
sample preparation, precisions from 5 to 12%^^ on compacted
samples are possible, even utilizing photographic detection.
In peak switching electrical analysis, with precision of better
than 2% possible, stringent gap control is in order.
       30O
       2OO
      UJ
      10

      I
      LU
       100-
         0      1234
               SPARKING TIME (min)
       Figure 4.  Effect of gap width on analytical data.
                  (a) Autospark controlled gap,  (b)  AGU
                  controlled gap.  39K+1 in an ashed bio-
                  logical sample compacted in graphite.
                  All charge accumulations, 0.1 nC.
                               12

-------
Magnetic Field Monitor

     A magnetic field monitor was constructed which greatly
facilitates locating the mass peaks of interest.  An inex-
pensive Hall probe (F. W. Bell, Inc., Model BH206) provides
an output voltage analogous to the magnetic field.  No
significant hysteretic effects are noted, and peak identi-
fication is normally to within +0.1 mass unit.  The probe
is mounted in a non-magnetic shim of 1/16-in Plexiglass
(Rohm and Haas) mounted between one magnetic pole and the
side of the magnetic analyzer.  The probe is outside the
vacuum, readily accessible, and well into the magnetic field.
Figure 5a shows the position of the probe in relation to the
ion flight path.  Figure 5b shows the associated equipment
needed to construct the field monitor (power supply:  Kepco
model HB-525; DVM:  Heath Universal Digital Instrument Model
EU-805).  A reference table of mass vs.  Hall voltage was com-
piled for a specific power supply  current and ESA voltage.

     The precision of the magnetic field monitor is in part
due to the placement of the Hall probe well into the magnetic
field.  To obtain unambiguous mass definition, the probe must
be placed in the magnetic field such that it intercepts a
flux density truly representative of that which is deflecting
the ions.  Positioning of the probe outside the analyzer in
this study allowed sampling of the field at sequential lo-
cations over the magnet pole face in order to determine the
most representative flux density, thereby reducing hysteretic
effects to a very low level.  Referring to Figure 5a, the
largest error (poorest reproducibility in locating mass peaks
was observed when the probe was placed in position 1.
Hysteresis was smaller at position 2, but the measurements
were still sufficiently imprecise to preclude positive mass
identification.  In position 3, hysteresis proved to be quite
small, allowing +0.1 mass unit resolution.

     Even with a hysteresis-free probe,  accurate peak identi-
fication would still not be possible, were  it not for very
precise measurements of the associated probe parameters.  The
transverse voltage developed in the Hall probe is governed
by the following equation:

                          = k(i x B)                       (1)

where HV is the Hall voltage, k is a constant which includes


                              13

-------
     Ions
 (a)
                                                       Hall
                                                       Probe
                      MAGNETIC  SECTOR
               Magnet
(b)
             Hall
6 DIGIT
  DVM
                CONSTANT
                 CURRENT
              SUPPLY  (O.O1%)
    Figure 5.  (a)  Placement  locations  of the Hall probe in
              the magnetic field.  (b)  Hall probe measurement
              system (adapted from a drawing of F. W. Bell,
              Inc., Columbus, Ohio).
                             14

-------
geometry, B is the magnetic field strength, and i is the con-
trol current, normally considered a vector quantity by nature
of its polarity.  To obtain a reproducible Hall voltage at
a given field strength, the current must be extremely stable.
While our supply had good short term  (hours) stability of
+ 0.01%, normal long term drift necessitated measurement
"(with a six-digit DVM) and subsequent fine adjustment of the
current daily.  A very accurate measurement arid control of
the ESA voltage is also necessary to hold the mass-Hall volt-
age relationship constant.  This voltage was also precisely
monitored on the DVM and adjusted as necessary.  With these
precautions, adjustment of the magnetic field to a calibrated
Hall voltage can bring a desired mass to register on the
collector slit to within +0.1 amu.  A brief fine tuning of
the magnet current by the operator (using the log-ratio  for
signal maximization) is then usually necessary before peak
switch integrations are begun.

Magnetic Peak Switching

     The introduction of electrical detection!5,25 to spark
source mass spectrometry now allows a significant improvement
in precision and accuracy using peak switch integration tech-
niques as compared to photographic detection.  Switching from
one mass  to another can be accomplished by variation of (a)
the accelerating and electrostatic analyzer voltages or (b)
the magnetic field strength.  The former method has been gen-
erally used26,27 because of problems associated with magnet
hysteresis.  Barton, et al.^° have shown that magnetic field
control is useful for fast sweeps.

     Electrostatic peak switching may introduce several pro-
blems, including possible changes in ion extraction efficiency,
variation of energy band pass into the magnetic analyzer, and
electron multiplier gain dependence's on ion energy.  The
general electrostatic peak switch mass limitation of M to 2M
before required readjustment of the magnetic field is also
a distinct disadvantage when carrying out analysis for many
elements.

     Magnetic peak switching eliminates these difficulties
because the accelerating and ESA voltages remain virtually
unchanged.  An inexpensive magnet control unit was designed
and constructed to allow implementation of magnetic peak
switching.

                               15

-------
          The magnetic Peak Selector (MPS) interfaced to an AEI
     MS 702 enables the operator to select any ten masses (or
     more, with add on capabilities) from 6 to 250 amu and bring
     them onto the collector slit to within +0.1 amu by a
     switch selected variation of the magnetic field.  The unit
     is comprised of a magnetic field sensing Hall probe inter-
     acting with two associated circuits, one a constant current
     supply to the Hall probe and the other a magnet control
     system based on the Hall voltage.

          The circuit  (Figure 6) is built around an operational
     power supply  (KEPCO Model OPS 100-0.2B) furnishing 100 mA
     to a high sensitivity Hall probe (F. W. Bell, Model BH-206).

          Figure 7 shows the control circuit for the MPS.  The
     active elements are an integrating null detector followed by
     a high voltage driver.  When coupled to the Hall probe, these
     amplifiers form a magnetic feedback loop which drives the
     magnet to produce a Hall voltage equal to a selected reference
     voltage, but of opposite polarity.

          Internal standard analysis—An advantage of magnetic peak
     switching is the ability to compare trace constituents with
WITH RESPECT  p
  TO COM
                                                               -HALL V. OUT
              Figure 6.   Schematic of MPS Hall probe  power
                         supply.   See reference 17  for  details
                                   16

-------
internal standards using high precision integration techniques.
Internal standards in SSMS have  normally been used onJy with
photographic detection and electrical scanning because for
these techniques the accelerating voltage is left unchanged.
However, both these procedures suffer from poor precision,
(~ + 30% rel std dev).  Electrostatic peak switching shows
good precision (~ 570 rel std dev), but difficulties arise
with the use of internal standards.  Sensitivity changes
occur, associated with changing the accelerating voltage. Also,
a limitation exists in mass of M to 2M without changing the
magnetic field.  Magnetic peak switching,  however, combines
the precision of integration techniques with the advantage
of a fixed accelerating voltage to allow large numbers of
elements to be quickly measured with good accuracy through the
use of precise relative sensitivity factors (RSF's).  Table 1
    Figure 7.  -Schematic of MPS control circuit.  See

               Reference 17 for details.

                              17

-------
          TABLE  1.  RELATIVE  SENSITIVITY  FACTORS  AS  DETERMINED BY MAGNETIC PEAK



                                        SWITCHING
                Sample        55Mn   56Fe    85Cu     aaZn     75As     80Se     208Pb
           Bovine  liver No.1, 2.5o    1.56    1.50     1.57     4.60     0.679    1.80
                  day  1






£          Bovine  liver No.1, 2.40    Im57    lm4S     lmSQ     4.81     0.664    1.99

                  day  1






           Bovine  liver No.1, 2.25    1.42    1.55     1.72     4.28     0.580    1.84
                  day  2
                                                                    °'479
           Bovine  liver  No. 1,  2>45    1>56            1>8Q    g<32

           _ day 4   _


           a

           Apparent/ true,  relative  to  89Y.

-------
shows RSF's determined by magnetic peak switching.  These
data depend not only on the use of the MPS, but also on close
monitoring and control of critical experimental parameters,
including interelectrode self-shielding and spark-gap . Deter-
mination of elemental sensitivities relative to an internal
standard using magnetic peak switching has been reported by
Hull.30

     Table 2 shows an example of magnetic peak switching
analysis using an internal standard.  NBS low alloy steel


 TABLE 2.   ANALYSIS OF 15 ELEMENTS IN NBS NO.  461 BY MAGNETIC

                  PEAK SWITCHING SSMS USING A  NICKEL

                         INTERNAL STANDARD3

Element
C
P
Ti
V
Cr
Mn
Co
Cu
As
Zr
Mn
Mo
Sn
W
Pb
Certified, wt%
0.15
0.053
(0.01)
0.024
0.13
0.36
0.26
0.34
0.028
( 0.005)
0.011
0.30
0.022
0.012
(0.003)
Found, wt%
0.313
0.047
0.012
0.021
0.187
0.429
0.269
0.345
0.030
0.001
0.014
0.263
0.023
0.011
0.003

   t    +
    60Ni x at 0.40 (wt%) internal standard.
                               19

-------
No. 462 was used to calculate RSF's for 15 elemental vs.
nickel.  NBS No. 461 was then analyzed using  these RSF's
taking five 0.01 nC charge accumulations for each element.
The large discrepancy for carbon could not be explained.
Chromium and niobium are also less than satisfactory-  How-
ever, all but four of the elements in Table 2 are close to
or within the anticipated 1070 accuracy range.  RSF's may vary
somewhat with matrix.  In Table 1 for example, iron shows an
RSF of 1.54 vs. yttrium in an electrode of ashed NBS Bovine
Liver compacted in graphite.  In an electrode of ashed NBS
orchard leaves prepared in the same manner, iron shows an RSF
of 1.89.  The excellent precision (~ 5%) obtainable with
integration methods may be misleading relative to accuracy
considerations of a standard and unknown or even to the sample-
to sample use of RSF's.

Repetitive  Scan Unit

      Electrical detection methods in spark source mass spectro-
metry using both  the  peak  switch and scanning modes have been
described.  ^  Scanning methods allow qualitative analysis but
are  generally limited-to a precision on the order of ~ 35%.^5
Peak switch techniques can show precision  to + 570 by careful
control of  critical parameters'^-*- and at the expense of a
considerable  increase in analysis time per element.  For sur-
vey  analysis  of many  elements per sample,  time considerations
may  dictate use of  the scanning mode, even given its less
satisfactory  precision.

      Averaging methods have  been used to  improve spectral
signal-to-noise ratios by  means of storing and adding repeti-
tive scans.   The  background  noise is random as compared to
the  unidirectional  peak  signals, resulting in a net  improved
spectrum.   Such averaging   techniques have been applied to
mass spectrometry.  A small,time-averaging computer  has been
used with  a GC-MS system   to permit repetitive scanning
and  increased sensitivity.   Computer controlled multiple scan
averaging^ »^ was  shown to  increase mass  measuring  accuracy
 in high  resolution  spectrometry.

      The multichannel analyzer  (MCA), a versatile data  read-
out  device  available  in  many laboratories, can act as a sim-
ple  but  effective averaging unit  for spectra accumulation.
The  multiscale mode of  the MCA  has been used  to enhance the
                               20

-------
measurement precision of isotopic abundances for uranium
and plutonium^S and to determine the isotopic abundance of
europium produced by nuclear reactions.

     The fluctuating nature of the spark ion source and the
poor ion statistics associated with the SSMS scanning mode
suggested an opportune use of averaging techniques to enhance
the precision of quantitative analysis and improve the attain-
able elemental detection limits.

     Signal averaging has been shown to improve precision and
sensitivity in spark source mass spectrometry using manual
mass scan repetitions. ^  Described in this report is a ver-
satile asymmetric triangular wave form generator, designed to
permit automatic multiple scan accumulations.  A range of
scan rates is provided by the unit, which in the repetitive
mode minimizes operator intervention and error.  At fast
scan rates, this is particularly important.  The repetitive
scan unit (RSU) supplies a cyclic sawtooth type magnet
reference voltage to a MS-702 spark source mass spectrometer,
and can be set to scan all or a selected portion of the full
mass range; it also supples a trigger pulse, which synchro-
nizes the scan of the magnetic field to repetitive spectra
accumulation, using the multiscale mode of the MCA.  Scan
rate and mass region, as^well as mass range, are all con-
tinuously variable.

     The RSU schematic is shown in Figure 8.  Operational
amplifiers Al, A2, and A3 comprise a triangular waveform
generator,^? modified to produce variable asymmetric tri-
angular waves.  A4, is a high voltage operational amplifier
which steps up the triangular wave voltage to the required
magnitude to drive the magnet reference tube-type circuit.
An asymmetric wave allows variable magnet scan times, which
may be a few seconds up to several minutes, and also vari-
able magnet reset times to reinitiate the subsequent scan.

CHEMICAL STUDIES

Sediment Samples

     The application of SSMS to sediment samples allows a
rapid, elemental survey analysis of questioned materials.
Inorganic hazards, 'such as lead, arsenic, and cadmium, could
be detected to very low levels.  Nonmetals,  such as bromine
and fluorine, wlftch are nofnormally analyzed by spectro-

                               21

-------
graphic emission or atomic absorption,  show  good  sensitivity
by SSMS.

      The  14  sediment  samples  analyzed  in  this  study were
obtained from the Environmental  Research Laboratory, U.S. EPA,
Athens, Georgia.  The  samples  were  from a  series  of sediments,
each of which had been previously analyzed by neutron activation
analysis,  X-ray  fluorescence,  and spectrographic  emission.

Sample Preparation—

      The  sediment  samples,  as received at this laboratory,
were in various  states of dampness.   Four  (37913,  37914, 37916,
and 37919)  were  visibly wet, appearing  as  mud.  Others  (37934
and 37961)  were  not in mud  form  but were damp;  the particles
tended to  cling  together.  The remaining samples  were hard and
dry, either as lumps or powder.   Samples 37936  and 37950 had all
the appearances  of  sand.
                                           R2   MCA ADDRESS
                                                  RESET
II 12 I3MM5 16 17 18 1920

         R

  Figure 8.  Schematic diagram of repetitive scan unit.


                               22

-------
     Approximately 1 g of each sample was placed in a 10-ml
beaker and broken up, if necessary, with a glass rod.  The
samples were then transferred to a 100°C oven for 24 hr.

     The samples were ashed at 500°C in a muffle furnace for
24 hr to remove any organic constituents.

     Table 3 shows the loss upon drying and additional weight
loss after ashing the samples.  The drying loss values reflect
the range of sample conditions from wet to dry and sandy.
The ashing loss may indicate organic content of the sediments
or water not released at 100°c.  The sample showed little or
no color change during the drying step, but during the ashing
process, many changed in color from brown to light reddish
brown.  Other portions of these sample which lost the most
weight in the high temperature ashing  (37913, 37914, 37919)
were subsequently ashed in a low-temperature rf asher.  A color
change from brown to a lighter brown also occured with this
ashing.

     From each sample that had been high-temperature ashed,
0.200 g was weighed into a 10-ml beaker, and 0.050 g of Ultra-
Carbon graphite was added.  One milliter of a 10 ppm Y solution
(Y?0-^ dissolved in Suprapur HNO3 and diluted with distilled
deiohized water) and 2 drops of ethanol were stirred into the
slurry.  The samples were then dried in an oven at 100°C, placed
in polyethylene vials, along with a plastic ball pestle,
and mixed on a Wig-L-Bug for 10 min.  Electrodes were prepared
from this material by means of an AEI die at 14 tons pressure,
using a hydraulic press.

Experimental Conditions—

     All data were obtained using an MS-702 spark source mass
spectrometer with electrial detection in the log ratio scanning
mode.

     To enable comparison of results from one sediment to an-
other, the sparking parameters were maintained as constant as
possible from sample to sample.  The spark gap was set and main-
tained by Automatic Gap Unit.22  A gap of 100 y was used
throughout the determinations.  Electrode alignment on the ion
axis was monitored with a Kennicott telescopic viewing system.16
Spark voltage was 30 KV, at 300 pulses per second, 100 ysec.
pulse length.  The distance from the spark to the No. 1 slit
was set at 6 mm.

     Under these conditions, the ion beam current was about
7% at a ratio-amplifier gain of 10.  The flux varied from 5-
15%.
                               23

-------
     The main amplifier gain, as set by the Amplifier and Scan
Control Unit (ASCU) was 10, and the electron multiplier settings
used were 2.50, 2.85, and 3.50.  These settings were experi-
mentally determined to provide the maximum quantitative infor-
mation, given the concentration of the internal standard.  The
ratio amplifier time constant was set at 40 ms, and the chart
speed on the oscillographic recorder was 0.2 cm/sec.  The
magnet scan time constant was 200.


Table 3.  MOISTURE CONTENT OF SEDIMENT SAMPLES AS MEASURED'

          AFTER HEATING TO 100°C AND ASHING LOSS

                    AFTER HEATING TO 500°C

Sediment
sample
37913
37914
37915
37916
37919
37921
37922
37923
37933
37934
37936
37947
37950
37961
%Weight
lost at 100° C
61.0
41.6
1.15
39.0
32.9
11.5
3.92
3.91
3.08
21.9
0.417
16.8
1.86
18.9
7oWeight
lost at 500°C
19.2
13.3
5.02
8.32
12.2
5.83
5.92
4.54
9.75
8.40
1.14
8.93
1.28
7.28

 a
  Relative  to  dry weight.
                               24

-------
Data Comparison —

     Two scans were taken at each of the 3 multiplier settings.
Peak heights were obtained from the scans using a logarithmic
scale overlay-  For some elements, notably Fe, it was necessary
to use an isotope other than the major one.  57Fe was usecj for
all iron determinations.  For certain other elements, elemental
interferences necessitated the use of small isotopes.  71Ga was
used instead of ^Ga 
-------
TABLE 4.  COMPARISON OF SSMS ANALYTICAL DATA WITH XRF, FS, AND INAA FOR




                      SEDIMENT SAMPLES (in >ig/g)
               No, 37913
Element
Ti
V
Cr
Mn
Fe
Ni
Co
Gu
Zn
Ga
As
Se
Br
Rb
Sr
Sn
Sb
Ba
Pb
SSMS
339
78.5
267
1858
34. 8K
169
327
1445
62.5
8.05
3280
10.25
600
31.8
2240
29.5
117
82. 3K
	
XRF

	
	
900
39K
73
— —
1000
480
24
3000
8
510
29
640
40
	
27K
410
ES
10K
	
IK
IK
100K
IK
	
10K
IK
	
10K
	
	
' '
	
10
	
100K
IK
INAA

54
970
1100
48K
	
30
3100
7100
	
3300
	
	
	
	
	
73
48K
	
No. 37914
SSMS XRF
1297 — ^
48.5 	
483 	
1 nl 1
lull
16. IK 38K
103 53
2.78K 	
611 1300
69.3 480
21.9 26
1530 710
29.3 6
371 420
210 43
14, OK 1900
31.4 50
31.8 	
343K 77K
86.0 250
F,S
10K
	
IK
IK
100K
IK
	
10K
IK
	
10K
	
— __


	
10
	
10GK
IK
INAA

46
360
410
39K
	
34
1800
690
	
740
	
	 r


	
	
	
90K
	

-------
            TABLE 4,  Continued.
ts>
No. 37915
Element
Ti
V
Cr
Mn
Fe
Ni
Go
Cu
Zn
Ga
As
Se
Br
Rb
Sr
Sn
Sb
Ba
Pb
SSMS
5416
119
81.3
1365
24, IK
100
33.8
19.2
56.8
71.2
7.05
7,41
2.93
565
248
5,9
2.45
1207
43.2
XRF
4000
70
80
1000
32K
32
	
45
180
11
12
4
4
63
82
30
	
300
48
KS
10K
• • •
IK
IK
100K
IK
	
IK
	
	
	
	
	
	
IK
10
	
10K
IK
INAA
MMMMH««
32
150
790
4 IK
	
18
470
NB
	
14
	
	
	
	


1.9
500
	
SSMS
3239
59.8
187
437
20, 2K
121
95.5
29.6
15,4
105
4.53
10.6
1.58
256
479
2,13
0,63
No, 37916
XRF
4700
160
160
320
34K
41.
	
83
370
17
17
4
12
98
130
30
	
2.98K 600
16.8
260
FS
10K
	
IK
IK
100K
IK
—
IK
IK
—
—
—
—
—
IK
10
—
10K
IK
INAA
_im, 	 ,_..,
54
220
240
37K
	
18
650
	
	
16
	
	 	
	
	


2.6
590


-------
           TABLE 4.  Continued.
to
00
No, 37919
Element
Ti
V
Cr
Mh
Fe
Ni
Co
Cu
Zn
Ga
As
Se
Br
Rb
Sr
Sn
Sb
Ba
Pb
SSMS*
2894
786
58
1395
27K
263
65.3
80
18.9
21.2
21.6
3.08
9.93
70.3
335
79.8
1.52
543
20.4
XRF
2900
90
590
1000
98K
130
	
230
950
15
10
5
7
60
260
70


500
320
ES
10K
	
IK
IK
100K
IK
	
10K
IK
	
	


	
	
IK
10


10K
IK
INAA
	 , 	
36
700
900 '
110K
	
16
730
1500
	
20


	
	
	
	
47
1000
	
SSMS
2305
88.5
46
1367
35. 3K
85.3
6,88
39.3
25
58.3
	
12.3
8,25
185
72.3
15.8
15.2
250
	
No. 37921
XRF
3900
130
240
1800
39K
11
	
36
50
17
5
7
7
75
39
50
	
400
30
ES
10K
	
IK
IK
100K
IK
	
100
	
— —
	


	
	
IK
10


10K
100
INAA
i. _r - -
30
88
1400
22K
— —
11
300
ND
	
2.7


	
	
	
	
.91
430
	

-------
           TABLE 4.   Continued.
NJ
IO
No. 37922
Element
Ti
V
Cr
Mn
Fe
Ni
Co
Cu
Zn
Ga
As
Se
Br
Rb
Sr
Sn
Sb
Ba
TDK
SSMS
7494
167
178
155
47. 8K
198
44.8
9.13
6.68
5.65
10.9

27
46.8
457
1481
XRF
7400
190
130
150
36K
14
45
93
20
13
4
8
68
57
30
300
c:l
ES
10K
IK
IK
IK
100K
IK
100





IK
10
10K
1 nn
INAA
76
130
160
34K
15
360
ND
14




0.37
370
SSMS
2885
58,5
211
693
20. 9K
115
15.3
61.5
40
83.8
6.83
3.55
321
415
14.5
2.7
723
C QC
No. 37923
XRF
5400
130
100
1100
32K
14
29
85
13
6
4
4
76
77
30
500
OQ
FS
10K
IK
IK
100K
IK
100





IK
10
10K
1 r\n
INAA
59
80
820
25K
10
550
ND
6.1




.8
390

-------
TABLE 4.  Continued.
               No. 37933
No. 37934
Element
Ti
V
Cr
Mn
Fe
Ni
Co
Cu
Zn
Ga
As
Se
Br
Rb
Sr
Sn
Sb
Ba
Pb
SSMS
1203
72.8
231
335
21.4K
153
20.5
182
2.93
36.5
7.98
2.83
3.6
249
211
8.6
4.55
180
	
XRF
7300
200
190
1000
52K
52
	
112
370
25
6
4
7
130
100
50


700
no
ES
10K


IK
IK
100K
IK
	
10K
IK
	
	
	
	


IK
10


10K
IK
INAA

86
260
1400
58K
	
35
1000
ND
	
13
	
	






1.7
1600
	
SSMS
2361
123
268
345
25. 3K
29.9
29.8
51.5
119
356
21.8
3.63
12.3
811
780
5.95
4.23
1075
24.6
XRF
4400
150
80
220
3 OK
16
	
27
56
17
7
4
21
64
100
30


200
32
ES
10K


IK
IK
100K
IK
	
IK
IK
	
	
	
	


IK
10


10K
IK
INAA
_
79
45
390
34K
	
12
650
ND
	
13
	
	






• v
560
	

-------
TABLE 4.   Continued.
                  No.  37936
No. 37947
Element
Ti
V
Cr
Mn
Fe
Ni
Co
Cu
Zn
Ga
As
Se
Br
Rb
Sr
Sn
Sb
Ba
Pb
SSMS
4463
189
192
858
69.5K-
323
23.2
6.63
3.38
21.2
1.89
	
.963
242
164
4.8
	
494
5.8
XRF


	
NA
	
	
	
	
NA
	 •
	
	
	
NA
	
	


	
NA
	
ES
10K
	


IK
100K
	
	
100
	


	
	
	
	
IK


	
10K
100
INAA


17
26
420
84K
	
3.7
310
	


5
	
	
	




.4
260
	
SSMS
5016
131
471
1431
13 OK
358
66
70.3
34.5
69
11.6
2.85
X3.7
187
523
7.23
1563

XRF
4500
150
80
630
42K
39
	
62
130
16
7
4
110
59
260
30
600
C1
ES
100K
100
10K
10K
100K
IK
100.
10K
	
	


	
	
	
IK
10
10K
1 n
.INAA
_--j- -
73
150
400
45K
	
20
2200
	
	
12
	
	
	
	


570


-------
            TABLE 4.  Continued.
                              No. 37950
No. 37961
U)
to
Element
Ti
V
Cr
Mh
Fe
Ni
Co
Cu
Zn
Ga
As
Se
Br
Rb
Sr
Sn
Sb
Ba
•DV.
SSMS
6602
161
230
1376
28. 9K
138
775
31.8
36
62.3
61.8
3.53
18.0
709
1452
18.3
13.9
3400
OO Q
XRF
3600
60
340
390
18K
9
	
14
35
13
5
4
10
61
390
40


700
O1
ES
100K
100
100
100K
100K
100
100
IK
	
	








IK
100


10K
lr\
INAA
IM^MMMM
39
550
490
18K


6.9
2200
ND
	
4.8










1.01
1300

SSMS
6300
91
273
1084
28. 7K
231
181
78.3
73.5
'89
189


40.3
209
2521
4.6


2.75K

XRF
4000
50
80
400
35K
21
	
36
79
17
6
4
56
88
400
20


1300
OO
ES
100K
100
10K
10K
100K
IK
100
10K
	
	








IK
10


10K
1 n
INAA
.,_!.. _|
63
110
400
35K


7.8
2100
	
	
7
	








1.2
1800


-------
ments.  Also, a  large difference  in concentration exists
between the  iron and yttrium.  Even though the 57Fe isotope
was used, the yttrium peak was generally quite low in those
scans which were of sufficiently  low sensitivity to keep the
57Fe peak on scale.

     Pb was another problem element, in that  it was difficult
to verify and measure the lead peaks in several (six) of the
samples.  In all six of these samples, X-ray  fluorescence found
over 10 ppm of Pb, well within the routine detection limits
of SSMS at the gains used.  Pb is, however, less sensitive on
a weight basis than lighter elements due to the ppma-ppmw
relationship and the mass dependence of the electron multiplier
response.

     In this study, lead was not  reported unless the pattern
of the 208-207-206 isotopes was clear.  The 208 peak was used
for calculations.  If this pattern was not present in a rea-
sonably accurate distribution in  the scans, no lead data were
taken.  This was done to preclude the use of  spurious peaks.

     There is a possibility that  some Pb was  lost from the
samples during the high temperature ashing stage, in the case
of sample 37913, which showed the highest Pb  by X-ray fluores-
cence, but no detectable Pb by SSMS, further  scans were taken
at a higher gain.  These did reveal lead peaks, but at a level
down from the X-ray and spectrographic emission data by a
factor of 10.  In other work done at this laboratory, Pb was
deposited as the nitrate from solution onto graphite, and no
significant Pb loss was observed.  However, due to the nature
of the Pb in sediment type samples, some loss may have occurred
at 500°C.

     Ni was consistently higher (by a factor of 2)  by SSMS than
by X-ray fluorescence.  This may be explained by a higher sensi-
tivity  factor or by  molecular interference.   The most probable
interferents in these samples would be silicon containing
species, e.g.t Sig, SiO, SiOa.  Silicon has 3 isotopes,  thereby
providing 6 possible Sis masses,  notably 56,  57,  58,  and 59.
All of these masses were used in this study to determine, re-
spectively,  Fe,  Nif  and Co.   Ni^° was not  used because of
the possibility of Cs and SiOa species interference.   Some
measure of the possible interference on masses, 57 and
59 can be gained by comparison to the +2 lines.  The molecular
interferences do not normally doubly ionize.
                               33

-------
     The Rb and Sr values are also higher than the X-ray data.
There are possible interferents occurring at masses 85 and 88,
such as FeSi and Fe02•  The nature of molecular interferent
formation makes it difficult to predict which of the many possi-
ble species will be large enough to significantly alter the
data.

     The above discussion has pointed out individual elements
and the problems involved in their determination.  The type
of sample presented by these sediments is a subject for a more
general discussion.   The accuracy and precision of the data
depend to a large extent on the physical and chemical nature
of the electrodes.

     Perhaps the greatest concern in handling the sediments
is homogeneity.  As they were received, the samples appeared
to have gross  inhomogeneities.  The sandy samples showed grains
of various  colors  and textures.  Other samples were in lumps
of various  sizes and  degrees  of wetness.  After  drying and
crushing, many of  the samples  took  on a more even color and
texture.  The  particle size was still rather large, but only
3 or 4  samples (37915,  -21,  -36, and  -50) showed gross inhomo-
geneity at  this  point.

     The  samples were vibrated  in a Wig-L-Bug after mixing
with carbon and  deposition  of the Y solution.  Although the
plastic vial and ball pestle  are not hard enough to grind
silicate  material,  friction and  impact of the particles with
each other  should  cause  some  reduction in size.  One  sample
 (37936) required grinding  in  an agate mortar and pestle to
powder  it prior to use of' the Wig-L-Bug.  Particle size, and
the  behavior of these particles  in  the spark discharge may
significantly  affect  the analytical data.   Homogeneity of the
sample  and  a consistent  electrode  surface geometry are extremely
important for  best SSMS  results.

     SSMS has  been shown to be an acceptable survey trace ana-
lysis  technique  for bottom  sediment samples.  Using log ratio
scanning, a large  amount of  data can be rapidly  taken.  Agree-
ment of this data  with other  analytical techniques is generally
good,  provided care is taken  to standardize and  control experi-
mental  parameters.

     Sample preparation  is  important  in that the particle size
of some sediments  may require reduction for best analytical re-


                               34

-------
suits.  The use of two internal standards might also be con-
sidered if all elemental constituents, majors to trace, are to
be analyzed.  Alternatively, a multi-isotope standard element
could be employed.  For elements of particular interest, where
greater analytical precision and accuracy may be desired, peak
switch integration techniques may be used.

Biological Samples

     The electrical nature of the spark-source requires that the
sample electrodes be electrically conducting in order for the
discharge and subsequent ionization to occur.  Biological com-
pounds such as hair, nails, and tissue must be treated in such
a way that they can be made to conduct, or, rather, be placed
in a conducting sample matrix.  Even if hair or nails could
be made to spark directly in their natural state, serious pro-
blems would still result which would require sample pretreat-
ment.  The organic constituents in the biological samples
produce an exceedingly complex spectrum which would make identi-
fication and quantitation of the inorganic constituents virtu-
ally impossible.  Therefore they must be removed by one of the
ashing procedures.

     In principle the ashed samples may be mixed with any good
conductor.  However, there are certain practical considerations
that dictate matrix peoperties.

     (1) Purity.  The matrix should be of the purest grade
possible.  Since the bulk of the electrode will be made from
this material,- purity requirements are stringent.  Grades known
as "spectrographically pure" are often not satisfactory.   High-
purity silver and gold are often used, but their trace element
concentration may be unacceptable, depending on the manufacturer.
A special grade of graphites is now available which seems to
be very low in impurities and consistent from batch to batch.

     (2) Physical Nature.  Only a small amount of each electrode,
perhaps 1 mg, Is consumed during the analysis.   This puts severe
homogeneity requirements on the sample ash-matrix mixture.
Many metal powders are too coarse to allow the  proper  intimate
mixing of the ash and matrix.   Again,  the graphite is  much better
in this respect because the Ultra Superior Purity grade 1-M is  ex-
tremely finely divided (< 1  ym mesh) and provides  an excellent
homogenization character.
                              35

-------
     (3) Chemical Nature.  An ideal matrix might be monoisotopic,
low atomic weight, with no tendency to form polyatomic species
or molecular units with sample elements.  A monoisotopic matrix
removes fewer mass lines from analytical use, the low atomic
weight reduces interference from multiply charged matrix ions,
and the tendency  to maintain the atomic ionic state also re-
duces interferences.  No matrix meets all these requirements,
of course.  Graphite  interferences are not usually severe and
are centered around the polycarbon lines which must generally
be avoided.  Selection of proper spark conditions can reduce
molecular species.

Sample Ashing--

     There are three  "standard" approaches to tissue ashing
 (or oxidation) along  with many more specialized methods.  High
temperature dry ashing  (HTDA) involves simply heating the dried
tissue  in a furnace to some suitable,  temperature, perhaps 500-
to 600°C, to remove by oxidation the  organic constituents.  The
method  is simple,  effective, and relatively  inexpensive.  Some
 inorganics, such  as mercury, may also be volatilized, however.
Furthermore, the  alkali  salts of a tissue may form a hard
glossy  residue which  is  difficult  to  remove  from the ashing
container.  In the case  of quartz ashing boats, the salts will
attack  the surface and  soon ruin  the boat.  Low temperature
dry ashing  (LTDA)  uses an rf discharge  to produce atomic oxygen
which  then reacts  with dried tissue at relatively low tempera-
 tures.  It is slower  than HTDA, but the lower temperature may
be useful to prevent  volatilization.  For tissues, perhaps a
 larger  advantage  is the  physical nature of the ashed sample;
an easily handled powder ash results.  Wet ashing   relies on
 the use of oxidizing  acids, such as HN03, and HCl04, to oxi-
 dize  in solution  the  organic components of the tissue.  Low
 temperatures  (~200°C)  can be used.   For techniques which use
 solution  samples,  notably atomic absorption, wet ashing is a
very  convenient method,  although the  analyst must develop
 familiarity with  specific acids and their rates of reaction
under  controlled  conditions.  HC104 digestions can create un-
 stable  situations for the novice.  However,  a mixture of HNOs
and HClO* can be  used quite safely and efficiently if properly
monitored.42

     Each of these ashing methods  can produce a sample suitable
                                36

-------
for SSMS analysis.  The experienced analyst may exhibit a pre-
ference for a particular method because of past usage,  but in
proper hands, each can be successful.  We have used all three
methods in our laboratory and prefer the LTDA approach.  How-
ever, each technique has weaknesses that must be considered.
See the EXPERIMENTAL of this report for descriptions of the
ashing procedures.  The use of these or similar procedures has
been reported for the analysis by SSMS of tissues^ and hair^.

     Although LTDA is slower and less amenable to large numbers
of samples, we prefer it for the following reasons:

1-  An easily handled ash residue is produced
2.  The low temperatures minimize elemental volatilization
3.,  The clean oxygen environment minimizes pick up during ash-
    ing.
4.  No addition of liquid reagents (as in wet ashing) is
    necessary.
5.  The residue is in a form that readily combines with graphite
    for SSMS analysis.
                               37

-------
                          SECTION V


                          REFERENCES


 1.   G.  C.  Gerloff, A. Rev- PI. Phystol., 14, 107 (1963).

 2.   P.  W.  Durbin, Health Physics, 2_, 225 (I960).

 3.   I.  H.  Tipton and M. J. Cook, Health Physics, 9_, 103 (1963).

 4.   W.  W.  Harrison, M. G. Netsky, and M. D. Brown,  "Trace
     Elements in Human Brain:  Copper, Zinc, Iron and Magnesium",
     Clin.  Chim. Acta, 21, 55  (1968).

 5.   J.  E.  Kunzler, "Ultra High Purity Metals," American Society
     for Metals, Metals Park, Ohio, 1962.

 6.   H.  H.  Hausner and S. B. Roboff, eds., "Materials for
     Nuclear Power Reactors," Reinhold, New York, 1955.

 7.   D.  Gerstenberg, Structure and Electrical Properties of
     Evaporated and Sputtered Titanium Films",  Ann. Physik,
     11,, 354 (1963).

 8.   E.  B.  Sandell, "Colorimetric Determination of Traces  of
     Metals," 3rd Ed., Interscience, New York, 1959.

 9.   D.  F.  Boltz, "Spectrophotometric Methods of Analysis", in
     "Handbook of Analytical Chemistry," L. Mertes,  Ed., McGraw-
     Hill,  New York, 1963.

10.   J.  K.  Taylor, E. J. Marenthal and G. Marinenko, "Electro-
     chemical Methods, Trace Analysis," G. H. Morrison,  Ed.,
     Interscience, New York, 1965.

11.   J. A. Dean,  "Flame Photometry," McGraw-Hill, New York,
     1960.
                                38

-------
12.  N. B. Hannay and A. J. Ahearn  ,  "Mass Spectrographic Ana-
     lysis of Solids", Anal. Chem., 26_, 1056  (1954).

13.  A. J. Ahearn, "Mass Spectrometric Analysis of Solids,"
     Elsevier, New York, 1966.

14.  J. P. Yuracheck, G. C. Clemena, and W. W. Harrison,
     "Analysis of Human Hair by Spark Source Mass Spectro-
     metry", Anal. Chem., 41, 1666  (1969).

15.  R. A. Bingham and R. M. Elliot,  "Accuracy of Analysis
     by Electrical Detection in Spark Source Mass Spectrometry",
     Anal. Chem., 43_, 43 (1971).

16.  P. R. Kennicott, General Electric Corporation, Schenectady,
     N.Y., personal communication,  1972.

17.  C. W. Magee, Ph.t), Thesis, University of Virginia,
     Charlottesville, VA, 1975..

18.  J. R. Woolston and R. E. Honig, "Energy Distribution of
     Ions Formed in the rf Spark Source", Rev. Sci. Instrum.,
     35_, 69  (1964).

19.  B. N. Colby and G. H. Morrison, "Automatic Spark Gap Con-
     trol for Spark Source Mass Spectrometry", Anal.  Chem., 44,
     1263 (1972).

20.  F. Konishi and N. Nakamura, "Advances in Mass Spectrometry,"
     Vol. 5, A. Quale, Ed., Institute of Petroleum, London,
     1970, p. 547.

21.  C. W. Magee and W. W. Harrison, Solids Workshop of the
     19th Annual Conference on Mass Spectrometry, Atlanta, Ga.,
     May 1971; C. W.  Magee and W. W. Harrison, ASMS Solids
     Workshop, St. Louis, Mo., October 1971.

22.  C. W. Magee and W. W. Harrison, "An Automatic Gap Control
     Unit for Spark Source Mass Spectrometry", Anal.  Chem. 45,
     220 (1973).

23.  C. W. Magee and W. W.  Harrison, "Spark Gap Effects on
     Sensitivity i-n Spark Source Mass Spectrometry",  Anal. Chem.,
     45, 852 (1973).
                               39

-------
24.  G. H. Morrison and A. M. Rothenberg,  "Homogenization of
     Nonconducting Samples for Spark Source Mass Spectrometric
     Analysis", Anal. Chem., 44, 515  (1972).

25.  R. J. Conzemius and H. J. Svec, "An Electrical Detection
     System for a Spark Source Mass Spectrograph", Talanta, 16,
     365  (1969).
                                                                *
26.  C. A. Evans, R. J. Guidoboni, and F.  D. Leipziger, "Routine
     Analysis of Metals Using a Spark Source Mass Spectrograph
     with Electrical Detection", Appl. Spectrosc. 24, 85  (1970).

27.  H. J. Svec and R. J. Conzemius in "Advances in Mass Spectro-
     metry", E. Kendrich, Ed., Vol. 4, Institute of Petroleum,
     Adlard, Dorking, pp 457-64.

28.  G. W. Barton, Jr., L. F. Tolman, and  R. E. Roulette, "Fast
     Magnetic Field Sweep", Rev. Sci. Instrum., 51, 995 (I960).

29.  K. H. Krebs, "Electron Ejection From  Solids by Atomic
     Particles with Kinetic Energy", Fortschr. Phys., 16, 419
      (1968).

30.   C. W. Hull,  "A Fast Electrical Detection  Spark Source
      Spectrometer with Low Elemental Detection Limits",
      Int. J. Mass Spectrom.  Ion Phys., 3_,  293  (1969).

31.  G. H. Morrison and B. N. Colby,  "Precision of Electrical
      Detection Measurements  of Powdered  Samples in Spark  Source
     Mass Spectrometry", Anal. Chem., 44,  1206 (1972).

32.   F. J. Biros, 'Enhancement of Mass Spectral Data by Means
      of a Time Averaging Computer", Anal.  Chem., 42_, 537  (1970).

33.  R. J. Klimowski, R. Venkataraghavan,  F. W. McLafferty,
     and E. B. Delany,  "A Small On-Line  Computer System for
      High-Resolution Mass Spectrometers",  Org. Mass Spectrom.,
     4, 17  (1970).

34.  A. L. Burlingame,  D. H. Smith, T. 0.  Merren, and R. H.
     Oslen, 16th Annual Conference on Mass Spectrometry and
     Allied Topics, Pittsburg, PA, 1968.

35.  L. Mguyen, G. Goby, and B. Rosenbaum, "Electronic System
      for High Sensitivity and High Stability Mass Spectrometry,
      Int. J. Mass Spectrom.  Ion Phys., 11_, 205 (1973).

                               40

-------
36.  P. E. Moreland, Jr., C. M. Stevens, and D. B. Walling,
     "Semiautomatic Data-Collection Systems for Mass Spectro-
     meters", Rev. Sci. Instrum., 58, 760  (1967).

37.  H. V. Malmstadt, C. G. Enke, and S. R. Crouch, "Control
     of Electrical Quantities in Instrumentation," W. A.
     Benjamin, Inc., Menlo Park, CA, 1973, p. 184.

38.  R. D. Giauque, Lawrence Berkeley Laboratory, 1972, un-
     published.

39.  National Bureau of Standards Report, "interaction of
     Nitrilotriacetic Acid with Suspended and Bottom Material,"
     Program No. 16020 GFR, July 1971.

40.  R. V. Moore, "Neutron Activation Analysis of Bottom
     Sediments," Environmental Protection Agency Report No.
     EPA-R2-73-009, March, 1973.

41.  T. T. Gorsuch, >lThe Destruction of Organic Matter,
     Pergamon Press, New York, 1970.

42.  G. F. Smith, "The Wet Chemical Oxidation of Organic Com-
     positions," G. F. Smith Chemical Co., Columbus,  Ohio,
     1965.

43.  W. W. Harrison, M. A. Ryan, L. D.  Cooper, and G.  G. Clemena,
     "Determination of Trace Elements in Biological Materials
     by Spark Source Mass Spectrometry", Newer Trace  Elements
     in Nutrition, W. Mertz and W.  E. Cornatzer, Eds.,  Marcel
     Dekker,- Inc., New York, 1971,  Chapter 17.
                               41

-------
                               TECHNICAL REPORT DATA
                         (Please read Instructions on the reverse before completing)
 . REPORT NO.

  EPA-600/4-77-042
2.
                          3. RECIPIENT'S ACCESSIOI*NO.
4. TITLE AND SUBTITLE
 LECTRICAL DETECTION PROBLEMS  IN WATER ANALYSIS
BY SPARK SOURCE MASS SPECTROMETRY
                          5. REPORT DATE
                            October 1977 issuing date
                          6. PERFORMING ORGANIZATION CODE
 . AUTHOR(S)
  .W.  Harrison
                          8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Department of  Chemistry
University of  Virginia
 harlottesville, VA   22903
                          10. PROGRAM ELEMENT NO.

                             1BD713	
                          11. CONTRACT/GRANT NO.

                              R801829-03-0
 12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Research Laboratory-Athens, GA
U.S. Environmental Protection Agency
College Station Road
Athens, GA  30605
                           13. TYPE OF REPORT AND PERIOD COVERED
                              Final	
                           14. SPONSORING AGENCY CODE

                              EPA/600/01
 15. SUPPLEMENTARY NOTES
 16. ABSTRACT
       This report describes research on  factors affecting accuracy in
 park source  mass spectrometry  (SSMS), and the application of  SSMS to
vater related problems.

       This project has  also involved the development  of instrumental
additions to  improve  the operation and control of SSMS procedures, t>ar-
bicularly with respect  to making  the technique more quantitative and
easier to use.  The application of SSMS  to specific sample types is also
lemonstrated.

       This report was submitted in fulfillment of Grant No. R801829-03-0
    the University of  Virginia under the  sponsorship of the U.S.  Environ-
nental Protection Agency.  This report covers the period September 1, 197
to  August 31,  1975, and work was  completed as of August 31, 1975.
17.
                             KEY WORDS AND DOCUMENT ANALYSIS
                DESCRIPTORS
                                         b.lDENTIFIERS/OPEN ENDED TERMS
                                      c.  COS AT I Field/Group
 Water analysis
  Trace Elements
  Chemical Analysis
                Spark source mass spec-
                trometry
  07 D
  14 B
  14 D
 18. DISTRIBUTION STATEMENT
 RELEASE TO PUBLIC
               19. SECURITY CLASS (ThisReport)

                  UNCLASSIFIED
21. NO. OF PAGES
 48
                                         20. SECURITY CLASS (Thispage)
                                            UNCLASSIFIED
                                                                 22. PRICE
EPA Form 2220-1 (9-73)
                                        42
                                               it U.S. GOVERNMENT PRINTING OFFICE: 1977—757-140/6576

-------