EPA-R2-73-252


September 1973
     ENVIRONMENTAL  PROTECTION TECHNOLOGY SERIES
.\v.*.v.v.v.v.v.*.*.v.v.v .*.•.*.".•.•.*,•.•.•.•.*.•.*.•.".•.•.•:
!vX*X*X*X*X*X-XvX-X'X-X-XvXvXvXvXvX'

                       IN
                              1

                        33
SB.
                                                   01
                                                   O

-------
                                    EPA-R2-73-252
           INSTRUMENTATION
      FOR  MONITORING SPECIFIC
      PARTICULATE  SUBSTANCES
IN  STATIONARY SOURCE  EMISSIONS
                      by

            John V. Pustinger, David A. Shaw,
          Paul L. Sherman, and Arthur D. Snyder
             Monsanto Research Corporation
                  Box 8, Station B
                Dayton, Ohio 45407
              Contract Number 68-02-0316
              Program Element No. 1AA010


           EPA Project Officer: John O. Burckle
            Chemistry and Physics Laboratory
          National Environmental Research Center
           Research Triangle Park, N.C. 27711

                   Prepared for

         OFFICE OF RESEARCH AND DEVELOPMENT
        U.S. ENVIRONMENTAL PROTECTION AGENCY
               WASHINGTON, B.C.  20460

                  September 1973

-------
This report has been reviewed by the Environmental Protection Agency and




approved for publication.  Approval does not signify that the contents




necessarily reflect the views and policies of the Agency, nor does




mention of trade names or commercial products constitute endorsement




or recommendation for use.
                                 ii

-------
                            ABSTRACT
     The results of a research program addressed to instrumentation
or methods for monitoring specific metals in stationary source
emissions is reported.  The program had two distinct objectives:

(a)  To produce an engineering prototype analytical and sampling
     system capable of continuously monitoring the concentration
     of beryllium and cadmium in typical stationary source emis-
     sions; and

(b)  To prepare briefing documents which identify technology
     appropriate to the analytical problem of continuous
     measurement of beryllium, cadmium, arsenic, antimony,
     barium, boron, chromium, copper, lead, manganese, mercury,
     nickel, and vanadium in stationary source emissions.

     Based upon the briefing document on analytical techniques
for beryllium and cadmium (Appendix II), two candidate continuous
monitoring techniques were chosen for application studies; radio-
frequency induced emission spectroscopy and AC arc-induced emission
spectroscopy.  A feasibility study of the radio-frequency induced
plasma system indicated that the system could not be applied to
continuously flowing sample gases containing air, even though a
detection limit of 0.4 pg Cd/m3 was obtained in an argon matrix.

     The AC arc-induced emission spectroscopy studies were con-
ducted employing an engineering prototype based on the Webb-arc
cell.  This system is potentially useful for continuous monitoring
of any metal by employing a monochrometer to select specific atomic
emission lines of the various metals.  Laboratory testing of this
system indicated detection sensitivities for beryllium at the
1 Mg/m3 level.  The corresponding sensitivity for cadmium was very
poor and time did not permit optimization of the continuous monitor
for cadmium detection.  Considerable difficulty was encountered in
limited field tests of the Webb-arc system at a coal-fired power
plant when the prototype continuous monitoring system was inter-
faced with the flue gas effluent.  In spite of these difficulties,
the Webb-arc emission spectroscopic system is considered the most
promising approach to monitoring trace metals in work environments
and stationary source effluents.  Advantages of the arc-induced
emission spectroscopy approach are:  (a) direct, real-time, con-
tinuous measurements down to low trace levels 1 yg Be/m3 without
emission signal integration; (b) high specificity, (c) moderate
volume flow rates (up to 40 1/min); and (d) relatively simple,
low-weight instrument package.  The applicability of the Webb-arc
induced emission spectroscopy techniques to direct continuous
monitoring of Be, Cd, and other potentially toxic metals requires
further study.
                               iii

-------
                           FOREWORD
     Airborne particulate matter is a major air pollutant having
significant effects on health, economics, ecology, visibility,
and aesthetics.  Effective techniques and hardware systems for
source emissions measurement are needed for application to the
various sources to achieve control of particulate emissions and
thus protect the environment.

     Available information reveals the major sources of air
pollutants containing toxic metallic species emanate from:  the
combustion of fuels containing such metals either as impurities
or as fuel additives; metal refining and metallurgical operations;
operations involving metals fabrication or machining; and waste
disposal.  Many such compounds have been found in our waters and
our air.  Historical records of analysis of surface water and
atmospheric samples indicate a trend of increasing amounts of
these contaminants.  The development of improved control tech-
nology, control practices, and emission information requires
instrumentation or methods to  measure and monitor these particular
substances.

     This contract work was sponsored by the National Environmental
Research Center at Research Triangle Park, North Carolina, to
identify promising techniques  applicable to monitoring the con-
centration of the specific trace metal components in typical
stationary source emissions and to produce and test an engineering
prototype, based on existing technology through adaptation of
existing analytical instrumentation, for measurement of beryllium
and cadmium.

     This report is included in the Environmental Protection
Technology series - the series, devoted to new or improved tech-
nology required for control and treatment of pollution sources
to meet environmental quality  goals, Includes reports of work
dealing with research, development,, and demonstration of instru-
mentation, equipment, and methodology to repair or prevent
environmental degradation from point and non-point sources of
pollution.

     The Project Officer gratefully acknowledges D. Van Lehmden,
J. Durham, R. Statwick, and especially W. Mitchell for their
participation in the review and critique of their review.
                               John 0. Burckle
                               Project Officer
                               U.S. Environmental Protection Agency
                               Office of Research and Development
                              iv

-------
                        TABLE OF CONTENTS

Section                                                      Page

   1     DISCUSSION OF REPORT FORMAT                           1

   2     EXECUTIVE SUMMARY                                     3

         2.1  Program Objectives                               3
         2.2  Program Approach                                 3
         2.3  Findings                                         8

              2.3.1  Preliminary Field Tests for Problem       8
                     Identification
              2.3.2  Laboratory Tests on Emission              9
                     Spectroscopy With Radio-Frequency
                     Induced Plasma and Webb-arc Cell
              2.3.3  Field Test of Webb-cell System at        10
                     Coal-fired Power Plant

         2.4  Conclusions                                     11

              2.4.1  Beryllium and Cadmium Monitor            11
              2.4.2  Analytical Technology - Determination    12
                     of Antimony, Arsenic, Barium, Beryllium,
                     Boron, Cadmium, Chromium, Copper, Lead,
                     Manganese, Mercury, Nickel, and Vanadium

         2.5  Recommendations                                 13

   3     LITERATURE SURVEY AND BRIEFING DOCUMENTS             14

         3.1  Summary - Analytical Technology for Continuous  16
              Monitoring of Beryllium and Cadmium
         3.2  Summary - Analytical Technology for Continuous  17
              Monitoring of Mercury
         3-3  Summary - Analytical Technology for Continuous  23
              Monitoring of Lead
         3.4  Summary - Analytical Technology for Continuous  25
              Monitoring of Arsenic and Antimony
         3.5  Summary - Analytical Technology for Continuous  27
              Monitoring of Barium, Boron, Chromium, Copper,
              Manganese, Nickel and Vanadium

   4     SUMMARY OF PRELIMINARY FIELD STUDIES                 40

         4.1  Preliminary Sampling Study - Power Station      40
         4.2  Preliminary Sampling Study - Municipal          42
              Solid Waste Incinerator
         4.3  Preliminary Field Study - Machining Operation   47
              of Beryllium Metal
                                v

-------
                  Table of Contents - (Cont'd)

Section                                                      Page

   5     INSTRUMENT DESIGN AND FEASIBILITY (LABORATORY)       48
         TESTING

         5.1  Selection of Prototype Design                   48

              5.1.1  Direction Indicated for Experimental     48
                     Plan
              5.1.2  Source Sampling Considerations           49

         5.2  Radio-Frequency Excited Optical Emission        50
              Spectroscopy

              5.2.1  Instrumentation                          50

                     5.2.1.1  Lepel Radio-Frequency           50
                              Generator System
                     5.2.1.2  Drake Radio-Transmitter         53
                              Radio-Frequency Generator
                              System
                     5.2.1.3  Monochromator and Emission      55
                              Detection System

              5.2.2  Results and Discussion                   57

                     5.2.2.1  Plasma Formation With Lepel     57
                              R.F. Generator System
                     5.2.2.2  Plasma Formation With Drake     58
                              Radio Transmitter Generator
                              System
                     5.2.2.3  Systems for Introducing Known   60
                              Quantities of Be or Cd Into
                              RF Plasma
                     5.2.2.4  Detection of Cadmium With       63
                              Radio-Frequency Plasma
                              Emission Spectroscopy

              5.2.3  Conclusions - Radio-Frequency Excited    65
                     Optical Emission Spectroscopy

         5.3  Arc Excited Optical Emission Spectroscopy       67

              5.3.1  Instrumentation                          67

                     5.3.1.1  Arc Discharge Cell              68
                     5.3.1.2  Arc Discharge Power Supply      68
                     5.3.1.3  Calibration Systems             68

              5.3.2  Laboratory Studies                       70
                               vi

-------
                  Table of Contents - (Cont'd)

Section                                                      Page

   6     FIELD TEST OF PROTOTYPE                              78

         6.1  Conclusions                                     86

   7     REFERENCES                                           89

         APPENDICES

              Appendix I                                      91

                 Sampling System Considerations and Design

              Appendix II                                    111

                 Briefing Document - Analytical Techniques
                 for Beryllium and Cadmium

              Appendix III                                   207
                 Briefing Document - Analytical Techniques
                 for Mercury

              Appendix IV                                    272

                 Briefing Document - Analytical Techniques
                 for Lead

              Appendix V                                     303
                 Briefing Document - Analytical Techniques
                 for Arsenic and Antimony

              Appendix VI                                    320

                 Briefing Document - Analytical Techniques
                 for Barium, Boron, Chromium, Copper,
                 Manganese, Nickel, and Vanadium

              Appendix VII                                   382

                 Operational and Maintenance Manual
                 RF and Arc Excited Emission
                 Spectroscopy System
                               vii

-------
                        LIST OF FIGURES

Figure

   1    Fly Ash Particle Size Distribution From              1*3
        Power Plant Test Site (February 1972)

   2    Be/Cd Sample Compartment                             51

   3    Diagram of Drake Radio-Transmitter                   54
        Radio-Frequency Generator System

   4    Diagram of Impedance Matching Network for Drake      54
        Radio-Transmitter Radio-Frequency Generator System

   5    Scan of Emission Spectrum (3100-3300 8) From         59
        Argon-Air RF Plasma

   6    Diagram of Cadmium Metal Vaporizer Probe for         62
        Injecting Cadmium Metal Vapor Into RF Plasma

   7    Diagram of Webb-Type Arc Discharge Chamber           69

   8    Emission Intensity at 3131 X for Webb-Type           71
        Arc Cell

   9    Calibration Curve - Be (3131 &) Emission From Webb   72
        Arc Cell Using Nebulized 1000 ppm Aqueous Be(N03)2
        Solution and Variable Carrier Gas Flow Rates

  10    Scan of Emission Spectrum (2000-4000 X) of           73
        Laboratory Air Using Webb-Type Arc Cell

  11    Scan of Emission Spectrum (2000-4000 8) From         74
        Webb-Type Arc Cell

  12    Scan of Emission Spectrum (2000-4000 fl) From         75
        Webb-Type Arc Cell

  13    Stack Gas Delivery System and Sampling Manifold      79

  14    Main Delivery Line and Blower Locations for          80
        Power Plant Emission Study

  15    Power Plant Test Facility                            8l

  16    Typical Pattern Webb-Type Cell Emission at           83
        Be 3131 A Line
                             viii

-------
                         LIST OF TABLES

Table                                                        Page

   I   Summary of Evaluation Parameters for Continuous        18
       Monitoring System for Beryllium

  II   Summary of Evaluation Parameters for Continuous        20
       Monitoring System for Cadmium

 III   Summary of Evaluation Parameters for Continuous        24
       Monitoring System for Mercury

  IV   Summary of Evaluation Parameters for Continuous        26
       Monitoring System for Lead

   V   Summary of Evaluation Parameters for Monitoring        28
       Systems for Arsenic and Antimony

  VI   Summary of Evaluation Parameters for Monitoring        33
       Systems for Barium

 VII   Summary of Evaluation Parameters for Monitoring        34
       Systems for Boron

VIII   Summary of Evaluation Parameters for Monitoring        35
       Systems for Chromium

  IX   Summary of Evaluation Parameters for Monitoring        36
       Systems for Copper

   X   Summary of Evaluation Parameters for Monitoring        37
       Systems for Manganese

  XI   Summary of Evaluation Parameters for Monitoring        38
       Systems for Nickel

 XII   Summary of Evaluation Parameters for Monitoring        39
       Systems for Vanadium

XIII   Beryllium and Cadmium Analyses of Particulate          41
       Collected From Power Station

 XIV   Semi-quantitative Emission Spectrographic Analysis of  44
       Particulate From 1st Stage of Brink® Cascade Impactor

  XV   Beryllium and Cadmium Content of Particulate and       46
       Aerosol Collection From Municipal Solid-Waste
       Incinerator

 XVI   Analysis by Atomic Absorption Spectrophotometry        85
       of Particulate Samples Collected at Power Plant
       for Beryllium and Total Particulate
                               ix

-------
                       ACKNOWLEDGEMENTS
     The complexity of the program required major contributions
from many technical and supporting staff members of the Dayton
Laboratory of Monsanto Research Corporation.  To provide a
mechanism for future transfer of technology related to specific
tasks in this program, special recognition of specific and
shared responsibilities is given to the following:

     Radio-Frequency Induced Plasma System (Lepel Generator)

          D. A. Shaw
          G. W. Wooten

     Drake Radio Frequency Plasma System
          G. W. Wooten
          H. C. Tucker
          C. M. Ellas

     Stationary Source Sampling System

          M. G. Konicek*
          A. D. Snyder

     AC- and DC-Arc Power Supply Systems

          D. A. Shaw
          H. C. Tucker

     MRC Signal Processor
          G. W. Wooten
          J. F. Moon

     Laboratory Tests - RF Plasma
     Laboratory and Field Tests - Arc Emission System

          D. A. Shaw
          P. L. Sherman

     Briefing Documents - J. V. Pustinger

     The efforts of Mrs. B. J. Weaver in preparing the
     manuscript are greatly appreciated.

     Of particular note was the active participation and coopera-
tion of the EPA Project Officer, Mr.  John 0. Burckle of the
National Environmental Research Center, Research Triangle Park,
North Carolina, on this program.
*Present address is Diamond Shamrock Chemical Company,
 Technical Center, Fairport Harbour, Ohio.

-------
1.   DISCUSSION OF REPORT FORMAT

     This report contains information on two interrelated, but
separate, general tasks.  Task I was to produce an engineering
prototype analytical and sampling system capable of monitoring
continuously the concentration of beryllium and cadmium in typical
stationary source emissions.  The task includes identification
and development of a workable detection concept, evaluation of
the sampling problem, fabrication of a prototype analytical and
sampling system, and laboratory and field testing of the analyti-
cal and sampling system.

     Task II involved the preparation of detailed briefing docu-
ments based on a thorough review of the literature, which identify
technology appropriate to the analytical problem of continuous
measurement of beryllium, cadmium, mercury, lead, arsenic, anti-
mony, barium, boron, chromium, copper, manganese, nickel, and
vanadium from stationary emission sources.

     To cope with the mass of data accumulated in both tasks, a
special report format was adopted.  A summary of all data and the
conclusions and recommendations derived from laboratory and field
testing of the prototype monitors and from the literature surveys
are presented in the Executive Summary (Section 2).

     Task I data are reported in Sections 3, **, 5, and 6, which
cover (a) Summary of Literature Survey and Briefing Documents,
(b) Preliminary Field Studies, (c) Instrument Design and Labora-
tory Testing, and (d) Field Testing of Prototype, and in Appen-
dices I and VII, which describe in detail (a) Sampling System
considerations and Design and (b) Operational and Maintenance
Manual - RF and Arc Excited Emission Spectroscopic Systems.
Task II data are reported in Section 3 (Summary of Literature
Survey and Briefing Documents) and Appendices II to VI, which
are detailed briefing documents on analytical techniques for
continuous monitoring of (a) Be and Cd, (b) Hg, (c) Pb, (d) As
and Sb, and (e) Ba, B, Cr, Cu, Mn, Ni and V, in stationary emis-
sion sources.
1.1  Format of Appendices

Seven appendices describe (a) detailed discussion of sampling
problems; (b) five detailed briefing documents for (1) Be and
Cd, (2) Hg, (3) Pb, (4) As and Sb, and (5) Ba, B, Cr, Cu, Mn,
Ni and V; and (c) operating manual and maintenance manual for
Be/Cd monitoring system.

     1.1.1  Briefing Document Format

     The briefing documents were prepared to show potential
methods of analysis, some of the limitations of the techniques,
possible problems in applying the procedures, and recommendations

-------
for the most acceptable approaches.  Emphasis was placed on five
prime factors, analytical reliability  (accuracy and precision),
selectivity, sensitivity, a continuous operation mode, and speed.
Other factors  (personnel, apparatus, installation, and maintenance
requirements and costs) although important, can be controlled by
proper hardware design and are emphasized to a lesser degree.

     Since some of the chemical and physical properties of the
elements and their compounds will markedly influence the selec-
tion of the analytical technique, a brief commentary on selected
properties for emissions anticipated from known sources is
included in a preface for each element.  Potential analytical
approaches, based on the chemical or physical properties are
then presented.

     Wherever specific analytical approaches have been used to
monitor an element, a detailed outline of the principle, appli-
cability, and operational parameters is tabulated.  These detailed
outlines generally highlight the merits and limitations of
potential methods.

     In several cases, candidate analytical methods have not been
applied to the element of interest.  As a consequence, specific
information regarding operational parameters and response factors
are not available.  In the briefing document, the technique is
described by indicating the principles involved, potential
applicability, and anticipated problems.

     Many manual techniques currently used for monitoring trace
metal emissions, while meeting criteria of analytical reliability,
selectivity, and sensitivity, require moderate to lengthy time
periods for sample pretreatment and are not considered or evalu-
ated in this document.  An analytical method was not eliminated
simply based on need for preliminary pretreatment, but was
screened based on whether the treatment, including chemical re-
actions, could be performed at the site, unattended, and within
a total measuring time of five to ten minutes.

     Each briefing document is assembled as a separate entity
containing its individual table of contents and sets of references.

     1.1.2  Format of Operating and Maintenance Manuals

     The operating and maintenance manuals are comprised of
manuals supplied by the manufacturers of individual sub-systems
and procedures developed under the experimental laboratory and
field testing phases of this program.

-------
2.   EXECUTIVE SUMMARY


2.1  Program Objectives

     The principal program objectives were:

(a)  To produce an engineering prototype analytical and sampling
     system capable of monitoring the concentration of the
     specific trace metal components - beryllium and cadmium -
     in typical stationary source emissions.

(b)  To prepare briefing documents which identify technology
     appropriate to the analytical problem of continuous mea-
     surement of beryllium, cadmium, mercury, lead, arsenic,
     antimony, barium, boron, chromium, copper, manganese,
     nickel and vanadium.


2.2  Program Approach

     The program consisted of the following tasks:

1.   Literature survey to identify the technology required to
     select an analytical approach and to determine potential
     problems in applying the monitor to typical Be and Cd
     stationary emission sources - power plant, incinerator,
     and refinery or machining operation.

2.   Preparation of briefing documents describing analytical
     technology related to continuous measurement of Hg, Pb,
     As, Sb, Ba, B, Cr, Cu, Mn,Ni, and V.

3.   Preliminary field studies to determine (a) the level of
     the Be and Cd emission from the power plant, incinerator,
     and refinery or machining operation; (b) physical and
     chemical nature of the emissions; and (c) potential prob-
     lems in interfacing the monitor and sampling system with
     the stack.

4.   Selection of analytical approach based on literature survey
     and evaluation of preliminary field studies.

5.   Design and fabrication of monitor and sampling system.

6.   Laboratory feasibility testing and calibration of the
     monitor and sampling systems to establish response charac-
     teristics, interferences, and sampling problems.

7.   Field testing of the prototype monitor and sampling system.

     The selection of a system for "continuous monitoring"
depends on several factors:  the nature of the source, the

-------
sampling constraints, and the availability of an analytical
technique compatible with these.  The selection of an analytical
approach is further influenced by the definition of continuous
monitoring and by the necessary requirement for a pretreatment
step.  Continuous monitoring can be defined in a very restrictive
sense as being only real-time measurement with no sample pretreat-
ment.  However, the present technology in abatement control and
detection systems permits only an approximation of continuous
sampling and measuring.  Methods considered during the selection
phase of the program included semi-automatic techniques which
required sample concentration or pretreatment of the specimen
including chemical reactions that could be performed at the site,
unattended, and within a total measuring time of five to ten
minutes.

     The following evaluation parameters were used to select the
analytical technique for the Be/Cd monitor:  Specificity, sensi-
tivity, range, accuracy, precision (repeatability), operational
mode, sampling time, analysis time, total cycle time, calibration
technique, sample form, interferences, unit output, special
equipment required, multi-element application, safety hazards,
problem areas, routine maintenance, volume flow rate, estimated
instrument cost, estimated size and weight, and compatibility
with beryllium (or cadmium).

     The selection of the proper analytical technique depends
not only on the response factors of the detection system, but
also on the physical and chemical properties of the total emis-
sions.  Consideration must be given to:

     Rate of emission of Be and Cd, weight/volume of effluent,
     and total volume rate of exhaust.

     Overall composition of emission, including potential
     interferences (gases, aerosols, and other metals).

     Temperature/pressure conditions at source emission point.

     Representative particle size distribution of particulate
     emitted.

     Distribution of Be and Cd within this size distribution.

     The process data suggests the sensitivity and resolution
requirements of the monitor, its concentration span, the poten-
tial sources of interference (chemical and physical) that it
must cope with, and the requirements of sampling and delivery
equipment to ensure that a representative sample is presented
to the analyzer.

     In evaluating the candidate analytical technique for either
real-time continuous or intermittent monitoring, the following

-------
questions related to the properties of the source emission must
be answered:

1.   Is the technique sufficiently sensitive to detect Be and Cd
     at the level of emission on an instantaneous or integrated
     sample basis?  Does a large variation in concentration ad-
     versely influence the quantitative reponse of the detection
     system?

2.   Is the effluent velocity too fast for quantitative instru-
     ment response for real-time analysis?  Is the velocity too
     fast for a filter, impinger or electrostatic collection
     system?

3.   Are the elements being measured present as particulate,
     mist or vapor?  Will the analytical technique measure total
     amount of element in solid, liquid or gas phases?

4.   Will other components of the effluent (gases, aerosols and
     metals)  interfere with the analytical measurements?

5.   Will the temperature/pressure conditions at the source
     emission point affect the operation of the analytical
     system?  Will variations in these parameters affect the
     precision of the measurement?

6.   What effect does particle size distribution have on the
     accuracy and precision of the measurement?  If sample is
     measured in real-time, are large and small particles mea-
     sured equally?  If a sample must be isolated from the
     vapor stream, will a representative specimen be obtained?
     (Note - Filtration presents special problems.  While filters
     are capable of retaining particles of all sizes, the filtra-
     tion efficiency varies with particle size and inertial
     parameter between about .5 and .05 ym.  Use of such filters
     also introduces additional matrix problems.)  Can particu-
     late, fume and vapor containing Be and/or Cd be trapped
     quantitatively with filters, impingers or by chemical
     reactions?

7.   How randomly are Be and Cd distributed within the particu-
     late?  What influence does this distribution have on the
     precision of the analytical measurement?

     The chemical and physical properties of beryllium and
cadmium and their compounds differ considerably between each
element and,  as expected, profoundly influence the selection
of methods for quantitative analysis.  Analytical techniques
that can be applied to one element and its compounds cannot
necessarily be applied to the other.

     Fundamental differences in atomic structure and nature of
compound formation for each element restrict application of

-------
certain analytical techniques based on nuclear or atomic proper-
ties to individual elements and their compounds and do not allow
application for analyses of both elements by the same technique.
Analytical techniques for trace analyses based on x-ray emission
of characteristic radiation for cadmium (atomic number ^*8) cannot
be applied to beryllium (atomic number 4).  Conversely, photo-
neutron excitation techniques which can be applied to beryllium
cannot be used with cadmium.

     Spectrographic methods have been devised for trace analyses
of both elements, beryllium and cadmium.  However, differences
in volatility and attendant response factors influence operating
conditions for attaining maximum sensitivities for each element.
Beryllium and its oxides (low fired and high fired) are less
volatile and more refractory than cadmium and cadmium oxide.

     In selecting a continuous monitoring device for both beryl-
lium and cadmium emissions from stationary sources, there are two
options.  Two analytical methods, based on measuring different
properties and requiring different measuring systems and opti-
mized for maximum response can be incorporated into one monitor.
However, the more ideal alternative would be to select one
measuring system which is highly selective and responsive to
both elements.

     At the present state-of-the art, there is no analytical
technique available for universal application as a direct,
continuous, monitor of beryllium and cadmium from all types of
stationary source emissions at the A.C.G.I.H. 1967-TLV levels
of 0.002 mg Be/m3 and 0.02 mg Cd/m3 (as soluble salts and metal
dust) or 0.1 mg Cd/m3 (as CdO fume).

     A variety of analytical techniques, including colorimetry,
fluorimetry, emission spectroscopy, atomic absorption spectro-
photometry, gas chromatography, and electrometric measurements
have been used routinely in monitoring beryllium effluents and
atmospheric pollution.  Somewhat similar approaches, i.e.,
colorimetry, emission spectroscopy, and atomic absorption spec-
trophotometry, plus polarography have been applied for measuring
cadmium in dusts and fumes.  These procedures employ standard
analytical instrumentation and require collection in an impinger,
electrostatic precipitator, or on a filter followed by low-
temperature ashing or acid digestion prior to the actual analyt-
ical measurement.  The techniques have proven useful for first
generation monitoring but require moderate to lengthy time
periods for sample pretreatment, are prone to contamination
from reagents and the digestion processes, and are not usable
for on-site, continuous monitoring of source emissions.

     The prime candidates for a continuous monitor of metal
emissions from stationary sources are based on radiant energy
emission or absorption, and measurement of nuclear properties

-------
and radioactivity.  The techniques potentially suitable for
sensors for automated, continuous analysis are:

          Emission spectroscopy
          Flame photometry
          Atomic absorption or fluorescence spectroscopy
          X-ray emission (fluorescence)
          Activation analysis
          Mass spectrography

     Conventional x-ray emission (fluorescence) techniques can
be applied to Cd, but not to Be.  Continuous monitoring devices
and manual analytical procedures based on activation analysis
have been used to determine Be, but are not applicable to Cd.
A number of problems with low ionization efficiencies and multi-
ion fragments occur when applying mass spectrography to measuring
Be and Cd.  Triply ionized aluminum (27A13+) and doubly ionized
oxygen (1802+) can interfere with the ion, 9Be+.  The specto-
scopic techniques - emission, flame photometry, and atomic
absorption or fluorescence - also require somewhat different
operating conditions to obtain maximum response for either Be
or Cd.  With the spectroscopic techniques, the problem is more
serious with the flame techniques - flame photometry and atomic
absorption or fluorescence - than with continuous, flow-through,
emission spectroscopy.

     Several continuous, flow-through, monitoring systems based
on the spectrographic measurement of arc (or spark) induced
optical emission characteristics of beryllium have been developed
and used successfully for measuring the beryllium levels (down
to <1 yg/m3 of air) in beryllium production or manufacturing
operations (See Appendix II, Sections 4.1.1 to 4.1.2).  The sensi-
tivity (detection limit 
-------
     When emphasizing current monitoring needs, we found that the
problems related to sampling and monitoring Be and Cd in emissions
of widely different physical and chemical composition require
better definition.  To better understand the sampling and moni-
toring requirements, preliminary sampling studies were performed
at a power plant, incinerator, and a beryllium machining opera-
tion.  Obviously, the problems are less acute in monitoring
beryllium or cadmium in operations where either element or its
compound, although existing at a low concentration in the stack
gas, constitutes the major portion of the particulate or fume.
The most complex sampling and analytical problems would be
encountered in the power plant or incinerator emissions where the
concentration of Be and Cd are low in a particulate matrix con-
taining relatively high amounts of other elements in refractory
compounds.

     Although Webb, et al. (ref. 1,2) demonstrated that the
flow-through emission spectroscopic technique could be applied
successfully to monitoring low levels (<1 yg Be/m3 to 100 yg
Be/m3), the data also suggested that larger amounts of energy
would be necessary to vaporize relatively large (»1 mg/m3)
quantities of refractory materials encountered in power plant
emission sources and to excite Be in these effluents to produce
optical emission spectra.  As a consequence, emphasis was placed
on developing a higher source of excitation energy with a radio-
frequency (RF) plasma.  However, in addition, an arc cell was
borrowed from Webb to establish feasibility for application to
detecting Be emissions in power plant effluents.


2.3  Findings

     2.3.1  Preliminary Field Tests for Problem Identification
            (See Section 4 and Appendix I)

     Samples of particulate were obtained with a cascade impactor
from stack emissions from three test sites - coal fired power
plant, solid waste municipal incinerator, and a beryllium alloy
machining operation.  Sampling was accomplished with the flow
rate through the nozzle of the set below isokinetic velocity to
favor collection of larger particles.  The non-isokinetic condi-
tion was established to determine if a reasonably representative
dust sample could be obtained without isokinetic sampling.
     The measurements showed that more than oQ% of the particu-
late collected from the power plant had diameters below 5 microns,
These data suggest that a reasonably representative particulate
sample could be obtained below isokinetic conditions.

     Atomic absorption spectrophotometric analyses of particulate
on individual stages of the cascade impactor indicate concentra-
tions in the flue gas of the power plant of approximately 30 yg
Be/m3 and 1.2 yg Cd/m3 in particles with diameters between 0.24

-------
micron and 7.0 microns.  The limited data suggest that the
beryllium content is lower in the smaller particles than in the
larger particles, whereas the opposite relative distribution
appears to hold for cadmium.

     Difficulties in sampling the municipal incinerator were
encountered due to high moisture content of the stack effluent.
Unless proper temperature control was maintained, most of the
material collected as a water condensate in the cyclone prior
to the separation stages.  When condensation occurred, no beryl-
lium was detected in the impactor stages which would isolate
particulate in the 0.24 micron to 7.0 micron range; however, very
small quantities of cadmium were detected in a relatively uniform
distribution in 4 of the 5 stages in the same particle size range.
When condensation was minimized, concentrations of approximately
0.4 to 2 pg Be/m3 and 40 ug Cd/m3 were collected in the particle
size range of 0.24 micron to 7-0 microns; the distribution of
beryllium appears to be approximately the same in the 4 stages
through the particle size range of 0.36 micron to 7.0 microns,
but slightly lower in the range 0.24-0.36 micron.  The distribu-
tion of cadmium in the various stages of the cascade impactor is
markedly different from that observed for beryllium; quantities
of cadmium are greater on the stages that isolate the smaller
particles 0.24 micron to 1.1 microns, than on the stages 1.1
microns to 7.0 microns.

     No particulate was isolated when sampling the beryllium
machining operation, with  filtering devices which could retain
>0.2 micron particles.  In this preliminary study, no attempt
was made to provide means for sampling vapor or "fume" in
impinger solutions or other media.

     2.3.2  Laboratory Tests on Emission Spectroscopy With
            Radio-Frequency Induced Plasma and Webb-arc Cell
            (See Sections 5.2 and 5.3 and Appendix VII)

     A feasibility study was conducted to determine if a radio-
frequency (RF) induced plasma system could be applied as a high
energy source for a continuous monitor based on emission spectro-
graphic measurements for Be and Cd.  Although a detection limit
of 0.4 yg Cd/m3 of argon was obtained in pure argon by direct
measurement, without signal integration, the system could not
be applied with continuously flowing sample gases containing air.

     This study showed that although radio-frequency induced
plasma temperatures as high as 20,000°K can be attained at
atmospheric pressure, the detection of the emission spectra of
trace metals is extremely difficult when using air or air-enriched
argon as the matrix gas.  The plasma temperatures are sufficiently
high to vaporize most refractory compounds and the excitation
energies (up to 25 electron volts) provide the necessary elec-
tronic excitation to produce excellent emission spectra with
argon as the matrix.  But with air in the plasma matrix, the

-------
plasma shape becomes distorted, increased background emission
spectra are obtained, and the line intensities of the element
being determined are diminished.  As a result direct analytical
determinations of cadmium in air matrices were not possible even
at levels in excess of 1000 yg/m3.

     No direct correlation with molecular emission bands, e.g.,
metal oxide, was  obtained under the test conditions.  No attempt
was made to pursue the potential use of molecular emission bands
in this study.

     When using the Webb-cell as the excitation source, a detec-
tion level was reported by Webb et al. (ref. 1,2) as 1 yg Be/m3
and was substantiated in a modified (AC arc, rather than pulsed
DC arc) cell on this program.  No data were reported by Webb et
al. (ref. 1,2) for cadmium, but Webb indicated in a private com-
munication that cadmium measurements should not be a problem.
However, limited laboratory studies with cadmium aerosols and
with the AC arc modification in this program show much poorer
sensitivities, which were insufficient to provide detection
limits equivalent to the TLV levels [0.02 mg Cd/m3 (as soluble
salts and metal dust) and 0.1 mg Cd/m3 (as CdO fume)].  Unfor-
tunately, no detailed study was made and data was accumulated
only with the instrument conditions used for detecting beryllium.
Optimization of conditions for detecting cadmium may provide the
needed sensitivity.  Undoubtedly, instrument conditions must be
optimized for each element.

     2.3.3  Field Test of Webb-cell System at Coal-fired Power
            Plant (See Section 6.0)

     Although limited laboratory testing showed the feasibility
of using the Webb type system for direct, continuous monitoring
of beryllium, considerable difficulty was encountered in field
tests (power plant) with a prototype, continuous monitor for
measuring beryllium directly in stack effluent.  However, the
Webb arc emission spectroscopic system still remains the most
promising approach to monitoring trace metal in work environments
and in effluent from stationary sources.   The advantages are:
(a) direct, real time, continuous measurements down to low trace
levels (1 yg Be/m3); (b) specificity; (c) moderate volume flow
rate (up to 40 1/min); and (d) relatively simple low weight
instrument package.  Lower detection limits «1 yg Be/m3 can
be obtained by integrating the emission signal for 30-60 seconds,
resulting in sampling and analysis times of approximately
30-90 seconds.   For beryllium, the analytical working range by
direct measurement without signal integration is 1 ug - 200 yg
Be/m3.  Problems to be solved include:  (a) fouling of the elec-
trodes and cell with particulate deposits when monitoring stack
effluents having high loadings of refractory (silicate) particu-
late,  and (b) absorption of moisture and other stack gases by
the asbestos cell resulting in a breakdown of electrical
insulation.
                              10

-------
     For stationary emission sources, yielding relatively high
loadings of refractory silicate particulate (>1000 yg/m3), a
much hotter vaporization system other than the Webb cell design
is needed.  In the Webb cell, the relatively small arc zone
serves to vaporize the particulate and to induce optical emis-
sion spectra but does not have sufficient energy to accomplish
both simultaneously and efficiently.  Solutions to this problem
would involve either the addition of a vaporizing system imme-
diately prior to the analytical arc discharge or the use of a
hotter analytical discharge system.  In the former case, an
auxiliary arc could be used to vaporize the particulate prior
to the analytical arc.  With the latter approach, a hotter exci-
tation system - radio-frequency induced plasma or Shumaker arc
oven - would be used.


2.*t  Conclusions

     2.^.1  Beryllium and Cadmium Monitor

     At the present state-of-the-art, there is no analytical
technique available for universal application as a direct,
continuous, monitor of beryllium and cadmium from all types of
stationary source emissions at the A.C.G.I.H. 1967-TLV levels
of 0.002 mg Be/m3 and 0.02 mg Cd/m3 (as soluble salts and metal
dust) or 0.1 mg Cd/m3 (as CdO fume).

     The Webb system (ref. 1,2) can be used for Be monitoring
in a limited number of situations, where the total particulate
loading is low (<1000 yg/m3) and where Be is present in excess
of 0.1 yg/m3.

     Although several possible solutions to the air effect problem
with RF induced plasma can be suggested, considerable development
work would be required to establish feasibility.  Also, the cost
and time factors, and the dilution effects for the actual moni-
toring measurements may become excessive.  Possible approaches
include removal of oxygen or reaction of oxygen to form relatively
stable oxide (CO, CC-2, HzO, or metal oxide) and incorporation of
additives to inhibit nitrogen-oxygen free radical reactions and
metal-oxygen compound formation, or to change the ionization
potential in selected plasma zones.  An alternate approach would
rely on spectral measurements of metal oxide emission bands or
spectral lines characteristic of more energetic transitions.

     After reviewing the results obtained with the radio-frequency
induced plasma and after discussion with Drs. V. A. Fassel (Iowa
State University), T. B. Reed (MIT), and R. Mavrodineanu  (NBS),
we have concluded that a radio-frequency induced total air plasma
or an argon plasma with moderate percentage of air which will
provide for detecting trace quantities of cadmium in air may be
attainable eventually, but not at the present state-of-the-art.
                              11

-------
     For the Webb cell system to be usable under conditions of
higher particulate loadings, particularly in systems containing
refractory silicates, an additional source of energy is needed
to vaporize the particulate.

     Possible approaches include using a DC arc or RF plasma to
preheat the sample prior to the Webb cell.  An alternate approach
would be to use a Shumaker arc chamber (see Section 6.0), which
develops high temperatures (>13,000°K), in place of the Webb
cell.

     2.4.2  Analytical Technology - Determination of Antimony,
            Arsenic, Barium, Beryllium, Boron, Cadmium, Chromium,
            Copper, Lead, Manganese, Mercury, Nickel and Vanadium

     A number of analytical approaches can be used to determine
each of the subject elements.  Most techniques, however, rely
on collecting an integrated sample by filtration or by isolation
in impinger fluids.  As a consequence, errors resulting from
losses of volatile species can be serious.  Despite difficulties
encountered in this program in attempting to measure beryllium
and cadmium emissions from a power plant source, spectroscoplc
techniques (emission and atomic absorption) remain the more
promising approaches to monitoring a wide variety of trace
metals directly.

     The Webb arc system used in Task I of this program for
detecting Be can be applied to monitoring other elements.
Limits for other elements need to be defined.  The major dis-
advantage occurs when handling stack effluents having heavy
loadings of refractory particulate.  Modification of the Webb
system to produce a hotter vaporizing source is needed.

     With selected stationary emission sources, and under the
proper environmental conditions, analytical techniques limited
to specific elements can be used.  For example, in situations
where the total particulate loading is extremely low and the
particulate consists of mostly beryllium and/or its compounds,
a continuous monitoring system for beryllium based on using a
sintered metal filter impregnated with an alpha emitter to pro-
duce the reaction

                   9Be + a  -»•  n + Y + 12C

can be used (see Appendix II, Section 4.4).   Also, with mercury
or its compounds, a mercury reduction process (chemical or pyro-
lytic) coupled with a photometric measurement can yield a rapid,
intermittent determination (see Appendix III, Section 3.1).
For organic lead compounds, reduction over hot carbon followed
by atomic absorption measurements of the atomic lead can be used
for continuous monitoring; however, application to effluent con-
taining high loadings of inorganic particulate is questionable
(see Appendix IV, Section 3.2).


                              12

-------
     By using collection devices (filters and impinger fluid),
analyses of heavy metals, but not Be and B, can be made with
XRF techniques.  The major limitations are matrix interferences,
time, and potential loss of "volatile" compounds.


2.5  Recommendations

     Based on the laboratory and field studies reported by Webb
et al. (ref. 1,2) for monitoring Be in work environments, and
the laboratory data obtained in this program, which substantiate
Webb's observations, we recommend that a program based on a
modified Webb-type approach be continued.  The applicability of
the system to direct, continuous monitoring of Be and Cd, as
well as other, potentially toxic, metals must be better defined.
The system could also be used at the effluent side of a standard
EPA sampling train to verify the efficiency of the EPA sampling
procedure.

     The following program is suggested:

1.   Test a Webb arc chamber and pulsed DC arc electronic
     system on a low level Be emission source, e.g., Be
     machining or refining operation.

2.   Modify a Webb arc chamber by incorporating auxiliary
     electrodes and additional chamber heating and determine
     feasibility of further applying the system to stack emis-
     sions containing relatively high loadings of particulate
     (power plants and incinerators).

3-   Modify electronics and optical system to provide modu-
     lation capability to enhance discrimination of analytical
     signal from background.  Wavelength modulation and use
     of lock-in amplifier system is suggested.

4.   Establish feasibility of using a hotter arc chamber of
     the Shumaker type to vaporize particulate in stack gases
     with high loadings.

5.   Determine feasibility in laboratory studies of using the
     arc excited emission spectroscopic technique as a universal,
     direct, continuous monitor for measuring Be, Cd, B, Ba,
     Cr, Cu, As, Sb, Pb, Ni, Mn, Hg, and V.

6.   Interface a Webb arc chamber with a standard EPA sampling
     train to validate the efficiency of the collection methods
     (based on filters and impingers) for total particulate,
     and for beryllium, lead, and other toxic materials.
                              13

-------
 3.   LITERATURE SURVEY AND BRIEFING DOCUMENTS

     Detailed briefing documents describing technology appropriate
 to the analytical problem of continuous measurement of beryllium,
 cadmium, mercury, lead, arsenic, antimony, barium, boron,
 chromium, copper, manganese, nickel, and vanadium are presented
 as five appendices to this report.  Data for the most promising
 techniques in these documents are summarized in Tables I-XII.
 Whenever possible, data derived from a known, specific applica-
 tion are listed.  Also, the tables are based on data for systems
 that would be practical as continuous, or moderately intermittent,
 monitors for stationary source emissions.  In some cases, lower
 detection limits are attainable with highly sophisticated labora-
 tory research instrumentation, e.g., Van de Graaff generator for
 x-ray emission (fluorescence) spectroscopy, but these devices
 are impractical for field monitoring.  Due to lack of detailed
 information appearing in the available literature some data had
 to be supplemented by crude estimates.

     At this time, there is no method available that can be imme-
 diately adapted to monitoring particulate, fume and/or gaseous
 emissions from stationary sources for trace quantities of Be,
 Cd, Hg, Pb, As, Sb, Ba, B, Cr, Cu, Mn, Ni, and V on a continuous,
 24-hour basis.  A number of techniques have been applied to
 measuring selected elements from this group in specific chemical
 or physical forms in work environment or ambient air, but no
 technique has been applied to all of the listed elements in the
 variety of chemical or physical forms that can be encountered
 from stationary source emissions.

     It is important to understand that at the current state-of-
 the-art most of the methods described in the briefing documents
 have not been applied to the continuous monitoring of trace
 elements in air or in stack gas emissions.  The methods cited
 in the briefing documents and summarized in Tables I-XII are
 presented primarily to provide background information for the
 selection of potential avenues, where R&D efforts may best be
 applied.  Some of the information listed in Tables I-XII was
 derived from the analysis of aqueous samples (atomic absorption
 spectrophotometry) and geologic specimens (alpha particle scat-
 tering measurements of moon specimens).  As a consequence, the
 detection limits quoted in the original publications and pre-
 sented in the summary tables are in a variety of units.   These
 and other techniques can be applied to determinations of trace
 elements, if specimens are collected on filters or in impinger
 fluids and analyzed directly or after suitable chemical trans-
 formation.   Since the "detection level" depends on so many
 factors including:  volume of air sampled, size or volume of the
 collection device, surface area or volume of sample exposed to
 the measuring system (XRP, alpha particle scattering),  and com-
position of sample matrix, we chose not to introduce a possible
bias by adjusting the literature detection level to a common

-------
unit.  The values reported in the summary tables are those pre-
sented in the original reports and merely serve to acquaint the
reader with the potential application of the method.

     A major concern in all of the candidate analytical approaches
is the problem related to obtaining a representative sample with-
out loss of the element sought or without producing changes in
the chemical composition which would affect the precision or
accuracy of the analytical measurement.   Techniques based on
analyzing specimens collected on filters may yield erroneous
results due to losses of sample through volatilization of par-
ticulate from the filter during the sampling processes or due
to inability to collect fume or gaseous metal compounds.

     Any technique relying on the collection of particulate on
filters from relatively large volumes of air over a lengthy time
period (1-24 hrs) risks the potential loss of volatile compounds.
This is true even with relatively nonvolatile compounds depending
on the gas temperature and composition.   BeO vaporization can be
enhanced and promoted at lower temperatures by a gas composition
containing appreciable amounts of water; Cr20a evaporation is
enhanced in an oxygen atmosphere containing water vapor; PbCl2
volatilization is increased significantly at lower temperatures
in a stream of HC1 (see Appendix I, Section 2).

     PbO vaporizes at red heat, but is stabilized in presence
of silica and/or iron oxide.  PbS volatilization occurs at normal
stack temperatures and is promoted by currents of combustion
gases in metallurgical furnaces.  Evaporation losses of mercury
compounds and arsenic trioxide can occur at ambient temperatures.

     The most attractive candidate analytical techniques are
based on spectroscopic measurements taken from the excitation
of continuous, flowing effluent (ref. 1-7).  These techniques
do not require the collecting of particulate and thus are not
affected by the problems associated with isolating specimens on
filters or in impinger fluids.  By using a spectroscopic analysis
of a continuous flowing system, measurements of total element
concentration as particulate, fume, and vapor are attainable in
sampling and analysis time periods of 60 seconds.

     Summaries of the briefing documents (Appendices II-VI)
identifying the analytical technology for continuous monitoring
of Be, Cd, Hg, Pb, As, Sb, Ba, B, Cr, Cu, Mn, Ni, and V are pre-
sented in the following sub-sections.  Tables I-XII show the
summary data for some of the analytical techniques considered
as potential candidate methods.
                              15

-------
3.1  Summary - Analytical Technology for Continuous Monitoring
     of Beryllium and Cadmium

     There is no method available at this time that can be imme-
diately adapted to monitoring particulate and/or fume emission
from stationary sources for beryllium and/or cadmium on a con-
tinuous, round-the-clock basis.  The work of Webb, et al.
(Appendix II, Section 4.1.1) suggests that a continuous monitor
can be developed based on> emission spectroscopic measurements
of a continuous flowing stream (40 liters/minute) of stack gases.
This type of emission spectroscopic measurement can provide a
means of monitoring both beryllium and cadmium (or most other
metallic elements and some non-metals) simultaneously.  The
technique has the necessary (a) selectivity, (b) speed (30 sec.)
for rapid, repeated analysis, and (c) sensitivity (VL ug/m3).
     An evaluation of the emission spectrographic approach must
be made to establish the efficiency of the excitation process
when analyzing entrained particulate of variable particle size
and when measuring stack emissions having moderate to high
loadings of entrained particulate.  The Webb system worked well
in monitoring the beryllium content in a work environment where
the particulate was principally beryllium compounds.  No data
are available regarding performance in measuring small amounts
(1-100 ppm) of Be or Cd in inorganic particulate, e.g., fly ash.

     Electrode wear and deposition of particulate from stack
gases having high loadings of particulate relative to ambient
or work-place atmospheres may be major problems in applying the
Webb technique to monitoring stack emissions.  By using an
electrodeless discharge system (radio-frequency coupled exci-
tation), the electrode wear and deposition problems could be
eliminated or minimized.

     The major problem related to using radio-frequency excitation
is the difficulty in initiating and maintaining an air plasma of
desirable geometry.  Also, adverse oxidation effects may limit
the sensitivity.  By blending mixtures of inert matrix gas (A or
He) with air containing entrained particulate and with a reducing
gas, e.g., hydrogen, a stable plasma and suitable sensitivity
may be attained.  However, by diluting the sample with an inert
or a reducing gas, some loss in the capability to detect the
lower concentrations of Be or Cd will be experienced.

     X-ray fluorescence techniques can be used for measuring
cadmium, but not for determining beryllium.  X-ray tube excita-
tion sources have been applied to analyses for cadmium, but
limited work has been performed with radioisotope induced x-ray
emission (2l*iAm).  The application of radioisotope induced x-ray
emission spectroscopy to determining cadmium in collected stack
emission particulate may require further development of new or
modified radioisotope sources, e.g., 125I.
                               16

-------
     A trade-off of performance specifications may depend on the
type of emission source being tested.  The Webb system may be
usable for low (1-100 ug/m3) Be or Cd levels in source emissions
composed of particulate or fume having Be or Cd as major compo-
nents and little, if any. other metallic elements.  Similarly,
the method based on the  Be(a,n,Y)12C reaction (Appendix II,
Section 4.4.5) can be used for measuring beryllium continuously
in atmospheres or stack gas emissions containing principally
beryllium or its compounds down to 1 ug/m3, but cannot be used
where non-beryllium particulate would be excessive.


3.2  Summary - Analytical Technology for Continuous Monitoring
     of Mercury

     At this time, there is no continuous monitoring system which
measures total mercury (vapor and particulate) directly.  Most
monitors are based on the measurements of mercury vapor alone.

     The techniques (Appendix III) which have sufficient speci-
ficity and sensitivity are:

     (a)  mercury vapor photometry (non-flame atomic absorption)

     (b)  microwave induced emission spectroscopy
     (c)  arc excited or radio-frequency induced emission
          spectroscopy.

     Mercury vapor photometry measures elemental mercury vapor
only and would require a pretreatment step to generate elemental
mercury from chemically bound mercury.  Also, the technique is
subject to interferences from organic compounds, smoke, dust,
etc.  As a consequence, a compensation technique or removal
(physical or chemical) of the interfering materials must be made.

     When using low wattage microwave induced emission spectros-
copy, a pretreatment step to vaporize particulate or to reduce
chemically bound mercury to elemental mercury vapor must be used.
The microwave energy (low power) is not sufficient to vaporize
particulate directly.

     With arc excited or radio-frequency induced emission spec-
troscopy, sufficient energy is generally available to vaporize
particulate up to particle sizes of approximately 80y at flow
rates of 40 liters/minute.

     Preheating of the chemically bound mercury to generate and
vaporize elemental mercury can be accomplished by induction or
resistance heating, oxy-hydrogen or oxy-acetylene flame, ac-dc
arc, or rf-plasma.

     For each of the techniques measuring emission spectra, suf-
ficient spectral resolution with a monochromator must be available
                               17

-------
                                                          Table I-A

                       SUMMARY OP EVALUATION PARAMETERS FOR CONTINUOUS MONITORING SYSTEM FOR BERYLLIUM
                                                (Baaed on Literature Review)
bmlsslon Spectroscooy
Evaluation
Parameter
..peclflclty
Detection Limit
Range
Accuracy
Precision
Operational Mode
Sampl Lng Time-
Analysla Time
Total Cycle Time
Calibration
Technique
Sample Form
Interferences
Unit Output
Special Equip
Multi-element
Application
Safety Hazard
Problem Areas
Routine
Maintenance
Volume Flou Rate
Est. Instrument
Cost-!
Lst Instrument
Size-
Compatibility
w/Cd
Arc 1 Spark
Lxcltatlon
High
0.5 vg/m1
0 5-100 ug/m'
Est. 5-101
5-10H
Continuous or
Intermittent
20 sec
60 sec
75 sec.
(a) std. aerosol
(b) Be-Cu
Air specimen
P^slDly slight
nU, Hg 4
particle size
ug/m'
None
Yes
High electri-
cal voltages
Electrode
deterioration
Adjust elec-
trodes every
8 hrs
10 1 /mln
$10,000-20,000
3' x 2' x 6'
100-500 Ibs
Yes
Induction Coupled
Radlofrequency
High
0.1 ppm in solid MgO
0.1-500 ppm
Est 5-10J
6 5» 8 10 ppm
Continuous or
Intermittent
20-60 sec
ca 60 sec.
Cjl 1-2 mln
Dry , powdered
beryllium car-
Partlculate (MgO)
In argon
None reported
ppm Be In MgO

Yes
High electrical
voltage I flame
Possible mass
overloading of
plasma
Very little
ca. 500 cc/min
$10,000-20,000
1> x 2' x 6'
Yes
Microwave
High
Est. 10""-10""g
io-"-io-'g
Est. 5-10K
7« 6 10'"g
40« 8 det. limit
Intermittent
anal, collected
specimen
10 mln-24 hrs
<2 mln
12 mln-2t h»s
Std samples
(solution or
aerosol)
Collected
partlculate
or solution
Matrix, total
mass & memory
effects
ug
Vacuum
Yes
Flame
Matrix; total
mass overloading,
memory effect
Clean excltor
tube every few
hours
-
(10,000-20,000
1' x 2- x 6-
Yes (Depending
on method of
vaporizing Be)
Atomic Absorotloni X-r
Conventional
Flame
High
0 03 ppm In H,0
0 03 to '1 ppm
-
2»
Generally
solution
10 mln-21 hrs
30 mln
1 hr-21) hrs
Std. solutions
Aqueous
solution
Refractory
nature of Be
ug
None
Depends on
choice of lamp
1 flame temp.
Open flame
Refractory
nature of Be
High consump-
tion of fuel
gases
-
$5,000-10,000
V x 2' x 6'
Each needs
different flame
temperature
Emls
Flameless (Fluore
High
ay
slon
scence)
9 x 10'"g
Not determined
-
>2»
Generally
solution
10 mln-2iJ hrs


30 mln
1 hr-24 hrs
Std. solutions
Aqueous
solution
Refractory
nature of Be
ug
None
Depends on
lamp choice
Moderate elec-
trical voltages
3
n
ta
M
0




Refractory
nature of Be,
coating effect
Clean tube
regularly
-
$10,000-15,000
*• x 2' x 6'
Yes
  -Atomic absorption mode could be used with RP excitation.   Multi-element  functionality would be restricted by number of
   elements In lamp or lamps.

  -If collection device Is required, sampling time depends on level  of  Be In  the  specimen and the sensitivity of the
   analytical technique.

l.iCost and size mostly dependent on electronics or dispersion system (monochromator) and can be minimized with proper
   design criteria.

  ^Requires chemical separation to eliminate Interference.

  ^General sensitivity (not detectable directly on air partlculate sample).

  ^Mounted on pick-up truck
                                                            18

-------
                                                       Table I-B
                     SUMMARY OF EVALUATION PARAMETERS FOR CONTINUOUS  HONITORIHO  SYSTEM  FOB  BERYLLIUM
Activation Analysis
Evaluation
Parameter
Specificity

Detection Limit
Range

Accuracy
Precision

Operational Mode

Sampling Time-
Analysis Time

Total Cycle Time
Calibration
Technique
Sample Form


Interferences


Unit Output
Special Equip

Multi-element
Application
Safety Hazard

Problem Areas


Routine
Maintenance
Volume Flow Rate
Est . Instrument
Cost!
Est . Instrument
Size*

Compatibility
w/Cd
Neut
Reactor
Poor*

H5 ug*
(??)

C7)
2-101

Intermittent
Anal. Filter
10 mln-21 hrs
Lengthy,
requires sep.
12-8 hrs
Std. samples

Partlculate
on filter

Elements w/hlgh
cross-sections

ug
Nuclear reactor
1 shielding
Yes with
exceptions
Radioactivity

Reactor mass
unacceptable

_

-
>*100,000

Massive


Requires chem.
separation
;ron
Radlolsotope
Poor*

115 ug1
('?)

(")
2-101

Intermittent
Anal Filter
10 mln-21 hrs
Lengthy ,
requires Sep.
12-8 hrs
Std. samples

Partlculate
on filter

Elements w/hlgh
cross-sections

ug
Shielding

Yes with
exceptions
Radioactivity

Radioactivity
leakage

Very little

-
125,000-50,000

Heavy, but
mobile!

Requires chem.
separation
Kadlolaotope
Photo-activation
High

2 ug/m1
2-25 UB/m1

(??)
><0f 8 25 ug
w/2 mln. counts
Intermittent
Anal. Filter
2 mln g 25 ug/m1
2 mln t 25 ug/m1

li mln g 25 ug/m1
Std. samples

Partlculate on
filter tape

Background from
source

ug/m1
Radioactivity
shielding
No

Radioactivity
leakage
Leakage of
source

Check covering
on source dally
79 l./mln
Parts - $5,000
Labor - (5,000
Unknown, Est.
2' x 4' x 6'

No

Alpha-Particle
Scattering
Moderate

1 ug/cm2
1-150 ug/cm1

1 atomic (
No data

Intermittent
Anal Filter
10 mln-21 hrs
20-21)53 mln

1 day
Data for pure
elements
Partlculate on
filter paper

Sample surface
thickness &
particle size
Atomic f or ug
Radioactivity
shielding I vacuum
Boron to titanium

Radioactivity
leakage
Leakage of source
and malnt. vacuum

Malnt . vacuum
and source
-
Est. (10,000-
20,000
Head - 13.3 cm x
17 1 cm x 11 7 cm
<<4 kg)
No

Mass
Spectrography
Interference from
"A1 + 1
Est 10.01 ug/m1
0.001 ug/m1 to
1)0 ug/m1
Semi -quantltat Ive
Semi-quant 1 tat Ive

Intermittent
Anal Filter
1 hr-21 hrs
2-1/2 hrs

3-21 hrs
Std. samples

Partlculate
extracted from
filter
"0-', "Al"1


VB
Vacuum

tea

Moderate elec-
trical voltages
Semi-quantitative
& 0 or Al Inter-
ferences
Malnt vacuum

-
ca. (100,000

Not mobile


Yes

 -Atomic  absorption  mode  could  be used with RF excitation.  Multi-element functionality would be restricted by number of
  elements  In lamp or  lamps.

 -If collection  device Is required,  sampling time depends on level of Be In the specimen and the sensitivity of the
  analytical  technique.

.iCost  and  size  mostly dependent on  electronics or dispersion system (monoehromator) and can be minimised with proper
  design  criteria.

 ^Requires  chemical  separation  to eliminate Interference.

 ^General sensitivity  (not detectable directly on air partlculate sample)

 -Mounted on  pick-up truck.
                                                           19

-------
                                                       Table II-A

                        SUMMARY  OP  EVALUATION  PARAMETERS FOR CONTINUOUS MONITORING SYSTEM FOR CADMIUM
                                              (Baaed on Literature Review)
Emission Spectroscopy
Evaluation
Parameter
Specificity
Detection Limit
Kangc

Accuracy


Precision

Operational Mode


Sampling Tlme-
Analysls Time
Calibration
Technique
Sample Form


Interferences
Arc 1 Spark
Induction Coupled
Excitation- Radlofrequency Microwave
High
ca_ 0 1 ug
0 1-100 ug

No data


125J t 3 ug level

Intermittent
Anal filter

10 mln-20.001S

No data
u
a
J








Continuous m Intermittent


Immediate
Immediate <
Std. solutions

Air sample


„ Anal filter
9
u 10 mln-21 hrs
' 10 mln-1 hr
Std. samples

Partlculate on
on filter










Particle size Particle Matrix, Particle
material - size
>1 micron
„ filament
a
Unit Output
Special Equip.
ug o ug or ug/ml
Special elec- 7, Vacuum system
trode system 5
nj
Multi-element
Application
Safety Hazard
Problem Areas


Routine
Maintenance

Volume Flow Rate

Est . Instrument
Coot 5
Est Instrument
Size*



Yes 3 Yes
n
li]
voltages
Handling the
filter

Adjustment and
changing of
electrodes
0.5 ft'/irln

approx $50,000

approx
*• x 21 x 6'



excitation
Mass over-
loading of
plasma
Clean excltor
tube regularly

_

no, 000-20,000

approx .
l|< x 2' X 6'



mg/m1
Propane blow
lamp
Pb
Flame

size, thickness,
Inter-element,
A, blanks
ug



Vacuum or helium
purge
Atomic number

Ml

activity
Particle size Particle Matrix, particle
size

Replace pro-
pane - 2 hrs.
batteries - 20 hrs
Comparable to
human respiration
1*5,000

(a) 10 cm x 25 cm
x 15 cm
(b) 15 cm x 27 em
x 13 cm
25 kg
size Interfere

nces

Leak check sample
chamber regularly
or replace He
_

(10,000-30,000

approx
V x 21 x 6'



cyl








Compatibility
  a/Be
                                                                                                        No
  -Based on direct analysis of filter paper with special spark electrode  system  and  direct reading Balrd-Atomlc
   spectrograph.  Continuous monitoring arc technique used for Be (Table  I)  could also  be used  for Cd.

  -Data for special continuous monitoring system

  -Atomic absorption mode could be used with RF excitation rather than conventional  flame If  lonlzatlon la not too great.

  -If collection device Is required, sampling time depends on level  of Be In the specimen and the sensitivity of the
   analytical technique.

i.iCoat and size mostly dependent on electronic or dispersion system (monochromator) and can  be minimized with proper
   design criteria.

  ^General sensitivity.  On air partlculate samples,  levels below 9.6 iig  cannot  be measured without radlochemlcal
   separation to eliminate Interferences.
                                                         20

-------
                                                      Table  II-B
                      SUMMARY OP EVALUATION PARAMETERS  FOR CONTINUOUS HONITORIHO SYSTEM FOR CADMIUM
Activation Analysis
Evaluation neutron
Parameter Reactor Radlolsotope PI;
Specificity Barely Barely
adequate adequate
Detection Limit 0 005 ug1 0 005 pg*
Range
Accuracy
Precision
Operational Mode
Sampling Time-
Analysis Time
Calibration
Technique
M H
Sample Form v w
i «
f fri
Interferences i i
Unit Output ^ ^
n a
Special Equip. " a
Multi-element | |
Application
Safety Hazard
Problem Areas
Routine
Maintenance
Volume Flow Rate
Est Instrument
Cost!
Est Instrument Heavy, but
Size* mobile
Compatibility Requires chem Requires chem.
w/Be separation separation


naaioisotope Alpha-Particle Mass
oto-actlvatlon Scattering Spectrography
Oood
0 3
ppm
>0 3 ppm
Semi-quantitative
Semi -quantitative
Intermittent
anal, filter
1 hr-24 hrs
§ § 3 hrs-21 hrs
"K "c
i f std
u o
. samples
a ° Partlculate
tt o extracted from
g 2 filter
3 % None reported
rH rt
a a ppm or ug
£ g Vacuum
Yes
Moderate elec-
trical voltages
Semi-quantitative
Maintain vacuum

Cf!
Not
Yes
-
noo.ooo
mobile

  -Based on direct analysis of filter  paper with  special  spark electrode system and direct reading Balrd-Atomlc
   spectrograph.   Continuous monitoring are technique  used  for Be  (Table I) could also be used for Cd.

  -Data for special continuous monitoring system.

  -Atomic absorption node could be used with RF excitation  rather  than conventional flame If lonlzatlon Is not too great

  -If collection  device Is required, sampling time  depends  on level of Be  In the specimen and the sensitivity of the
   analytical technique.

i.lCost and size  mostly dependent on electronic or  dispersion system  (monochromator) and can be minimized with proper
   design criteria.

  ^General sensitivity.  On air partlculate samples, levels below  9 6 ug cannot be measured without radlochemlcal
   separation to  eliminate Interferences
                                                           21

-------
to provide separation of spectral interferences which may arise
from the matrix.  The mercury vapor photometer system does not
necessarily use a monochromator, since the basis of the technique
(atomic absorption phenomenon) provides specificity for mercury.

     The need for a monochromator in the emission spectroscopy
techniques, particularly the low power microwave induced optical
emission, can be eliminated if collection-desorption as a gold
or silver amalgam is used to isolate the mercury from interfering
components of the matrix.

     Detection of mercury vapor with an argon ionization system
can be accomplished on a short-term, intermittent basis if the
bound mercury is decomposed to mercury vapor and the mercury
vapor is removed from the matrix with silver or gold amalgama-
tion.  Desorption of elemental mercury vapor into the argon
ionization detector would produce a highly responsive signal.

     Five systems can be identified for potential use as continu-
ous monitors for total mercury in stationary source emissions.
     System A

     Sample •* Thermal decomposition
     System B
                        Compensated mercury
                          vapor photometer
     Sample •* Thermal decomposition ->• Isolation as gold
          amalgams -»- Mercury vapor photometer

     System C

     Sample -»• Thermal decomposition -»• Isolation as gold
          amalgam ->• Microwave induced emission spectroscopy
     System D

     Sample •*
     System E
Radio-frequency induced or arc excited emission
 spectroscopy
     Sample •* Thermal decomposition •* Isolation on carbon or
          as gold amalgam -*• Argon ionization detection

     Although System D does not require a separate thermal
decomposition sub-system, it does require a relatively good mono-
chromator to provide spectral resolution.  However, with the
monochromator, System D provides for multi-element monitoring.
                              22

-------
     Systems A, B, and C are more sensitive than System D,  but
require high efficiencies in the thermal decomposition and  iso-
lation steps (when needed as in B and C).  No information is
available on decomposition efficiencies of refractory particulate
matrices or at modest to high flow rates.  Also, little informa-
tion is reported on amalgamation and desorption efficiencies of
mercury on gold at modest to high flow rates.


3.3  Summary - Analytical Technology for Continuous Monitoring
     of Lead

     Although several techniques have been applied to the contin-
uous monitoring of lead vapor, principally as the organo-lead
compounds, no single method has been tested completely to qualify
as a continuous monitor for total lead, i.e., vapor and particu-
late, from stationary source emissions.

     Non-flame atomic absorption spectroscopy, preceded by  a
thermal and/or chemical reaction decomposition step to reduce
airborne lead compounds to elemental lead and to maintain the
elemental lead as vapor, should be considered.  The major problems
in adapting the most effective decomposition system (passage of
air containing lead compounds over hot carbon) used for ambient
air are:  (a) efficiency of the reducing reaction at high stream
velocities normally encountered in stack emissions and at the
resulting short residence time in the decomposition zone is un-
known; (b) potential plugging by refractory particulate in  the
carbon chamber when used with stack emissions involving high non-
lead content, e.g., power plant fly ash, needs to be evaluated;
and (c) consumption of the hot carbon rods at high stack gas flow
rates, where the stack gases are predominantly oxidants, e.g.,
oxygen, etc., must be determined.

     Continuous monitoring of ambient air flowing at 1 ftVmin.
has been accomplished by emission spectroscopy with a spark dis-
charge.  The technique permits a working range of 0.003 mg  Pb/ft3
to 0.05 mg Pb/ft3, but based on newer instrumentation may allow
lower limits.  The instrumentation was applied to measuring lead
in vapor and particulate.  Other methods of excitation - rf
plasma, and intermittent arc - may also improve the efficiency
of the emission spectrographic method.

     X-ray emission (fluorescence) spectroscopy can be applied to
measuring lead in particulate collected on filters.  The major
restriction to its application for determining total lead is its
inability to analyze vapor.  Some means must be found to collect
volatile lead compounds on a filter-type media.  Some considera-
tion should be given to adapting paper tape samplers impregnated
with complexing agents, e.g., tetrahydroxy-p-benzoquinone,  used
in collecting lead fume for colorimetric analyses.
                              23

-------
                                                 . 0? EVALUATION PARAMETERS TOR COWT1HUOU3 MONITORING SYSTEM TOR MERCURY
                                                                  (Baaed on Literature Review)
bvaluailon
I'd inn tar
Speririclty

Rang*
Accuracy
Precision
Operation Mode
Sampling Tim*
Analysis Tim
Tottl Cycle Tine
Calibration
Technique
Samplr Porn
Inter Terences
Unit Output
Special Equip
Hultt-clciMnt
Application
safely Haiard
Problem Areas
Routine
maintenance
Volume Plow Rate
Eat Instrument
Coot
Eat Instrument
Sice
Mercury Vapor Photometer
(Planeleas Atonic Absorption)
Oood

Dependo on path length, etc
(Seo Appendli III)
Insufficient data
Std dev t 0 0003 ug - 371
1 0 01 vg - 3 to 81
Continuous or Intermit tan t
Continuous (no time delay)
Alc-oai Instantaneous
Almost Instantaneous
Std samples Hg vapor
blenental Hg vapor-
Organic*, ocone, partlculate.
aulfldes, magnttlc fields
ug/ffl* or ug
a
H.<
Moderate electrical voltages
Toxlclty of Hg
Spectral Interferences,
directly, magnetic inter-
ferences
very little
Ho major limitation
1500 - 15,000
Portable - approx
1' x 1> x 0 7' U7 IBS)
to aonroi 2' x *' x 6'
Flam Atomic Absorption
Absorption >i
Oood (except for high
Co BBtrlx)
(a) 1-10 ug/nl (HiO)
(») o.ooz ug/ liter or
air (carbon)
1-300 ug/nl (KiO)
Rel Error - <.10S
Coeff of variation
0 5-21
In t emit, tint
10 mln - 2 hra
Approx 15 nln
25 mln - 2-1/2 hra
Std solutions
(n) KB vapor or reac-
tive participate
In HiO
(b) HR vapor or par-
tlculate on carbon
Spectral - Co
Chemical - Ozldltlng
a reducing reagent
tig/ml or ppm
(a) -
(b) Carbon collector
I tantalum boat
Yes. with IBHP l
flame limitations
Open hot flane
Co interference
Presence of oxld
1 red agents
Distribution of Hg(I).
Hg(II) and Hg°
Replace burner gasee
-
12,000 - 110,000
2' I «' I 6*
and Fluorescence

Good
(a) 0 OB ug/ml
(dlreot]
(b) o 002 ug/ml
(additives)
•
Bel. Error - <10I
Approx aame as AA
In t trait tint
10 aln - 2 lira
Approx 15 mm
25 nln - 2-1/2 hri
Std solutions
Hg vapor or reac-
tive partioulate
In HiO
a1", AU. Pt,
Cr(VI). Ag
ug/Bl or ppm
•
Tea, with llnlta-
tlona
Open hot flame
Chenloal incor-
fereneea
Replace burner
gaaea
-
12.000 - 110,000
2' I ft' K 6'
Emission Spectroacopy
Had io- iWquenc y
or Arc Excitation"
Qood
Bat lO'Mo"'^
-
Data Inadequate
(Eat xlOS)
Data Inadequate
(Eat HOI)
Continuous or
intermittent
Approx 1 mln
Approx 1 mln
Approx 2 Bin
Std solutions
Kg* vapor »
partlculate
-
Mass or
uaa/voluae
-
Tea
Plane-like dlacharge
or high voltage
electrical discharge
Particle Bite
gas (A or He)
ca 300 ec/Bln (RP)
Up io io 1/min
(arc)
110.000 - 120.000
«' i 2' x 6'
Hlcrowava
(Low Power)
Oood
<1 i 10'"gf
Linear - 0 1-100 ng
as nethyloercurlc
chloride
Data Inadequate
(Eat tlOI)
Data Inadequate
(Eat 2101)
Intermittent
Approx 1-10 mln
Approx 2-5 mln
Approx 3-15 lain
Kg" vapor or Hg
•alta on heated
probe
Hg* vapor only
Matrix, total naaa
HE
Vacuum
Yea (with nano-
chronator)
Flame-like discharge
Must collect mer-
cury, nuflt decom-
pose paniculate
replace carrier gaa
supply
Oependa on effi-
ciency of collec-
tion system
110,000 - 120,000
4' It 2' X 6'
Radio t sot op lc Jet hod
Good
0 01 IIR/I
0 01 - 1 2 ug/1
Insufficient data
Std dev 0 001 ug/1
(range 0 01-0 02 ug/1)
and 0 075 ue/1
(range 0 2-1 ? ug/1)
Intermittent
2-20 mln
Counting tlm-100 aee
5-25 mln
Air saturated with
Kg vapor
Hg vapor
Sulflde Ion
Count a/1 00 oec con-
vertible to up or ug/1
Radioactive liotopo
and aMeldlnjr
Specific foi UK. but
absorber could be modl-
elemento having radio-
active isotopes
Hode rate radioactivity
Sulflde and depletion
of Hopcalllt
change so lull on and
Hope a me
0 5 to l 5 l/vin
$5.000 - 910,000
«• x :• x 6*
^Depends on optical path, eleetronlca. Instrument dealgn. collection medlun (if uaed). and compensation technlouo
 for background

-If decomposition atep Is used, a pretreatment can measure total Hg (vapor and partlculate).

-Thermal decomposition and Isolation atepe can be uaed to yield total Hg and minimise Interferences

-SyeteB specific for Hg, but non-flame atomic abaorptlon phenomena could be uaed on other elements

-Based on continuous flow system (Appendix III, ref  25,26,27).

^Based on data from Appendix III. ref  28   MRC preliminary data Indlaatea potential detection limit 10~"  to 10~"g

-------
     When using the most sensitive lead line (L ) for measuring
lead by x-ray emission (fluorescence) spectroscSpy, the analyst
must be aware of potential spectral interference from As (K ) in
media containing moderate amounts of arsenic.  The PbLg line1 has
been used successfully with 109Cd radiosiotope excitation.

     Techniques using y-absorptimetry are not necessarily adapt-
able to universal measurement of lead but could be used in
special circumstances where lead or its compounds would be the
only compounds emitted and could be collected on filters, etc.
The sensitivity of the method needs to be optimized and specific
detection limits established.
3.^  Summary - Analytical Technology for Continuous Monitoring
     of Arsenic and Antimony

     There is no technique available for measuring arsenic and
antimony continuously in ambient air or in stack emission.  All
techniques presently used in measuring the air pollutants -
arsenic and antimony - require the collection of particulate in
impingers or on filters.
     The low sublimation temperature of AsaOa suggests the poten-
tial loss of volatile arsenic compounds from particulate collected
on high volume filtering systems.  Analytical techniques, e.g.,
x-ray emission (fluorescence) spectroscopy, neutron activation
analysis, and mass spectroscopy, using this method of collection
may yield low results.

     Similar low results will also be observed with emission
spectrographic techniques based on analyzing particulate collected
on filters.  However, emission spectrographic techniques based on
the excitation of a continuous flowing stream of air or stack gas
through an excitation zone would not have this problem.

     Other than using flame  (acetylene-oxygen), or radio-frequency
induced excitation of flowing solutions of arsenic and antimony,
no data are available regarding spectrographic analyses of flowing
air samples.  Based on information derived from flame excitation
studies, a moderate amount of development work should be antici-
pated to establish the optimum viewing zone in the flame or
plasma.  Emission lines of elements (arsenic and antimony) of
high excitation potential and high ionization potential may not
appear in the flame mantle, but occur as intense emission lines
in the reaction zone under the proper conditions (see Appendix V,
Section 3-3).  The use of radio-frequency or microwave-induced
plasmas should be considered.  The latter when used in a low
power mode will not permit direct measurements, but will require
isolation of the arsenic or antimony in impinger solutions.
                               25

-------
                                                      Table IV
                      SUMMARY OP EVALUATION PARAMETERS FOR CONTINUOUS MONITORING SYSTEM FOR LEAD
                      	(Based on Literature Review)	^^
    Evaluation
    Parameter
 Specificity
 Detection  Limit
 Range

 Accuracy
 Precision

 Operation  Mode
 Sampling Time

 Analysis Time
     Emission Spectroscopy
	(Spark Excitation)
Good
<0.003 ng/ft1
0.003 mg/ft'-0.05 rag/ft'
No data
No data
(Est.
(Est.
Continuous
Appro*. 1 minute
(simultaneous w/analysls time)
Approx. 1 minute
(simultaneous H/sampllng time)
Total Cycle Tine    Approx. 1 minute
Calibration
  Technique
Sample Form

Interferences

Unit Output
Special Equip.
Multi-element
  Application

Safety Hazard

Problem Areas
Routine
  Maintenance
Volume Flow Rate
Est. Instrument
  Cost
Est. Instrument
  Size
Comparison of chemical analysis
for TEL In air
Vapor or partlculate

None reported

rag/ft1
None
Yes
High voltage electrical
discharge
Particle size
Replace electrodes

1 ft'/mln.
»10,000 - $20,000

4' x 2' X 6'
Non-flame Atomic Absorption
Yes
0.16 ug/rn'
0.16 ug/m'-S'l pg/m*

Insufficient data
Max. dev. from mean - 4.5X
Avg. dev. from mean - 1.5K
Continuous
Instantaneous

Instantaneous

Instantaneous
Air doped with TEL
                         Vapor;  fume; possibly
                            partlculate
                         Possibly  copper,
                            halogen organlcs
                         Mass or mass/vol.
                         Rf heated carbon
                         Could be  modified for Indi-
                         vidual  volatile elements,
                         e.g., Hg
                         Moderate  to high temper-
                         ture; rf  radiation
                         Depletion of carbon In
                         oxidizing atm., particle
                         size, residence times In
                         decomposition chamber
                         Replace carbon
                         1.2 llters/mln.
                         *10,000 - $20,000

                         4' x 2' x 6'
       X-Ray Emission
(Fluorescence) Spectroscopy
  (X-ray Tube Excitation)
Yes
0.05 Mg/m1
Dependent on matrix effects,
(1 Mg/cm1 up to 1.5 mg/cm1
on filter)
Insufficient data
No data

Intermittent (Anal, filter)
24 hours

5 minutes

Approx. 24 hours
Filters containing standard
samples
Partlculate on filter

As, possibly Br

PS
Vacuum or helium purge
Atomic number >11
                               X-rays; moderate to high
                               electrical voltages
                               Matrix, particle size
                               Interferences
                                                        Leak check sample chamber
                                                        regularly or replace He cyl.
                                                                                      $10,000 - $30,000
                                                                                      4' x 4' x 6'
                                                       26

-------
     Considerable difficulty can be experienced in attempting
arsenic analyses in high lead matrices by x-ray emission (fluores-
cence).  Not only does the high absorption characteristic of the
matrix cause problems, but direct spectral interference of Pb Lai
radiation occurs with As KCU-

     When using neutron activation analysis, particularly as a
nondestructive technique, instrumentation having high resolution
capability [Ge(Li) semiconductor detector] is necessary to mini-
mize a variety of interferences in measuring arsenic and antimony
in stack emission particulate.

     The use of an arsine generation step aids in concentrating
arsenic for atomic absorption spectrophotometry, colorimetry,
and potentially flame-, rf- or microwave-excitation of emission
spectra, but the total analysis' time requiring collection of
sample, digestion to yield trivalent arsenic, and generation of
arsine is too long for short-term, intermittent measurements.

     Although a long path absorption tube improves the sensitivity
of the atomic absorption method for arsenic, considerable inter-
ferences  (formation of arsenides, oxides) can occur, resulting in
low arsenic measurements.  High flame or non-flame temperatures
are needed.

     Considerable development work is needed to attain continuous
monitoring of arsenic and antimony from stack emissions.  There
is no method currently available that meets the requirements for
continuous monitoring of arsenic and antimony from stationary
emission  sources.  Removal of arsenic and antimony and their com-
pounds from the stack gases by a combination of filtration and
collection in impingers followed by chemical spectrophotometric,
or atom absorption spectrophotometric methods is currently the
only practical approach.  Some consideration should be given to
determining the feasibility of using continuous arc excited emis-
sion spectrographic techniques for direct measurements for As
and Sb on work environment or stack emission sources.
3.5  Summary - Analytical Technology for Continuous Monitoring
     of Barium, Boron, Chromium, Copper, Manganese, Nickel and
     Vanadium

     At this time, there is no continuous monitoring system for
stationary source emissions which measures B, Ba, Cr, Cu, Mn, Ni,
and V as individual elements or in a multi-element functional
model.  Most laboratory techniques to date have been applied to
integrated samples of particulate collected on filter paper from
ambient air.

     High sensitivities and the attendant low detection limits
of the absorption and emission spectroscopic techniques make these
                              27

-------
                                                      Table V-A

                   SUMMARY OF EVALUATION PARAMETERS FOR MONITORING SYSTEMS FOR ARSENIC AND ANTIMONY
                                             (Based on Literature Review)
Atomic Absorption Spectrophotometry
Lvaiujilon
Parameter
Specificity
Detection Limit
Hange
Accuracy
Precision
Operation Mode
Sampling Time
Analysis Time
Total Cycle Time
Calibration
Technique

Conventional A. A.
Yes
1 ppm In HjO
(0.02 ug/m! with
2000 m' air sample)
Linear up to 50 ug/ml
No data
Eat. i2X
Intermittent
Up to 24 hours
Depends on collection
method and extraction
(10 rain, to 2 hrs)
Approx 24 hrs
Known aqueous
standard solutions
Arsenic
A. A. w/Arslne Qen
Yes
0.02 ug
-
No data
Est. 12*
Intermittent
Up to 24 hours
10 mln. to 2 hrs
Approx. 24 hrs
Known aqueous
standard solutions

Absorption Tube
Yes
0.006 ug/ml
-
No data
Est i2(
Intermittent
Up to 24 hours
10 mln. to t hrs
Approx 24 hrs
Known aqueous
standard solutions
Antimony
Conventional A. A
Yes
1 ug/ml
Linear up to 10 ug/ml
No data
Est. ^2X
Intermittent
Up to 24 hours
Depends on collection,
digestion methods,
10 min. to 2 hrs
Approx 24 hrs
Known aqueous
standard solution:*
Sample Form
Interferences
Aqueous solutions of
dissolved partlculate
                      Flame background,
                      flame noise, Mg, Ca,
                      Al, Nl, Co,  Hn
                      arsenides,  Fe
Arsine gas produced
from partlculate or
fume

Oi, chromate, raolyb-
date, metavanadate,
nitrates, chlorates,
Cu, Sb, HjS
Aqueous solutions of
dissolved partlcu-
late

Same as Conventional
A.A. but background
less
                                                                                                 Aqueous solutions of
                                                                                                 dissolved partlculate
                                                                                                 Possibly Cu and Pb
Unit Output
Special Equip.
Multi-Element
Application
Safety Hazard
Problem Areas
Routine
Maintenance
Volume Flow Rate
Est. Instrument
Cost
Est. Instrument
Size
ug/ml
-
Depends on choice of
lamp and flame
Open flame
See Interferences
above
Replace fuels
-
$5,000-$10,000
4' x 21 x 6-
US
Arsine generator
Possibly Sb with
generator
Open flame, toxic
arslne gas
See Interferences
above, valence state
As must be trlvalent
Replace fuels, change
Zn and H2SO. In
generator
-
»5,000-$10,000
4- x 2- x 8'
Ug/ml
Heated 91 cm Vycor
cell
Same as Conventional
A. A.
Flame
See Interferences
above, deposition
of AzOi In tube
Replace fuels, clean
deposit from tube
-
*7, ooo-$: 5, ooo
4' x 21 x 9'
ug/ml
-
Depends on choice of
lamp and flame
Open flame
Possibly Cu and Pb
Interferences
Replace fuels
-
»5,000-$10,000
4' x 21 x 6'
                                                        28

-------
                                       Emission Spectroscopy
Arsenic
Lvaluatlon
Parametei
Specificity
Detection Limit
Rnngc
Accuracy
Precision
Operation Mode
Sampling Time
Analysis Time
Total Cycle Time
Calibration
Technique
Sample Form
Interferences
Unit Output
Special Equip.
Multi-Element
Safety Hazard
Problem Areas
Routine
Maintenance
Flame
Lxcltatlon
Yea
2 2 Pg/ml
No data
No data
No data
Intermittent
Up to 24 hrs
10 mln-24 hr
depending on
sample prep
Approx 24 hrs
Standard
solutions
Aq. solution
of dissolved
partlculate
or fume
Zn, Cd, Pos-
sibly Sn ( Nl
pg/ml
-
Depends on
excitation
potential
Flame
Rf Plasma
Yes
0.1 Pg/ml
No data
No data
No data
Intermittent
Up to 24 hrs
10 mln-24 hr
depending on
sample prep
Approx. 24 hrs
Standard
solutions
Aq. solution
of dissolved
partlculate
or fume
Cd
pg/ml
-
Yes
Flame and
rf radiation
See Interfer- Cd
ences above,
selection of
anal line ( silt
width, possibly
formation of ar-
senides In flame
Replace fuels
Replace car-
rier gas
Antimony
Flame
Excitation
Yea
1.0 Pg/ml
No data
No data
No data
Intermittent
Up to 24 hrs
10 mln-24 hr
depending on
sample prep
Approx. 24 hrs
Standard
solutions
Aq. solution
of dissolved
partlculate
or fume
As
Ug/ml
-
Depends on
excitation
potential
Flame
None reported
Replace fuels
Rf Plasma
Yes
0 2 lig/ml
No data
No data
No data
Intermittent
Up to 24 hrs
10 mln-24 hr
depending on
sample prep.
Approx 24 hra
Standard
solutions
Aq. solution
of dissolved
partlculate
or fume
None reported
PB/ml
-
Yes
Flame and
rf radiation
None reported
Replace car-
rier gas
X-ray Emission
(Fluorescence) Spectroscopy
Arsenic
Interferences (Pb,W)
1-10 Pg (Depends on
matrix)
Dependent on matrix
No data
No data
Intermittent
(Anal, filter)
Up to 24 hrs
1*5 mln
Approx 24 hrs
Filters containing
deposits of known
concentration
Partlculate on
filter
absorption
Pg
Vacuum or helium
purge
Atomic number >11
X-rays, moderate
to high electrical
voltages
Matrix (particularly
lead), particle size
effects
Leak check sample
chamber regularly or
replace He cylinder
Ant Imony
Yes
2 Pg (Depends on
matrix)
Dependent on matrix
No data
No data
Intermittent
(Anal filter)
Up to 24 hrs
^5 mln
Approx 24 hrs
Filters containing
deposits of known
concentration
Partlculate on
filter

Pg
Vacuum or helium
purge
Atomic number >11
X-rays, moderate
to high electrical
voltages
Matrix (particularly
lead alloys), par-
ticle size effects
Leak check sample
chamber regularly or
replace He cylinder
Volume Flow Rate
Est. Instrument
  Cost1
Est  Instrument
  Size
$5,000-115,000  $5,000-117.500   *5,000-415,000  t5.000-tl7.500  tlO,000-t30,000       »10,000-*30.000


II' x 2' x 6'     4'  x 2'  x  8'     4- x 2' x 61    4- x 2' x 81    I1 x I1 x 6'          4'  x 4'  j  6'
•Cost depends on resolution, detection limit, and sample  time required, which In turn depends on the sample matrix.
                                                            29

-------
techniques most attractive for continuous or short-term, Inter-
mittent monitors.  By maintaining the element sought in the stack
gas and by introducing the stack gas as part of the oxidant or
fuel in the flame or high energy discharge of the excitation source
of the spectrometer, a continuous monitoring system is possible.
With B, Ba, Cr, Cu, Mn, Ni, and V, the major problems with using
the absorption or emission spectroscopic techniques are the
inter-element or compound formation effects.

     With atomic absorption spectrophotometry, the inter-element
or compound formation effects are minimized or eliminated by using
a nitrous oxide-acetylene flame.  The best results are obtained
with a sheath gas (inert gas, e.g., nitrogen) to exclude oxygen
from the flame.  Obviously, in ambient air or stack emissions,
elimination of the oxygen is extremely difficult.  One possible
way is to pass the stack gas or air over hot carbon to produce
carbon monoxide.

     The use of an inert gas plasma as an atomization source for
atomic absorption spectrophotometry will also minimize inter-
element effects if a reducing atmosphere is maintained in the
plasma.  The reducing atmosphere can be maintained by introducing
hydrogen gas.  An additional problem related to the use of a
plasma to provide the atomization for atomic absorption measure-
ments is the possibility of overexciting the element sought to
emission, rather than to the necessary ground energy state.  As
a result, the population of ground state atoms may be too low to
attain maximum sensitivity.  Generally, the selection of the
"viewing zone" of the plasma must be optimized depending on the
element to be measured.

     Although flame photometry can be used for measuring Ba and
a number of other elements, inter-element and compound formation
effects are also a problem.  These effects can be minimized by
using a nitrous oxide-acetylene flame.

     Some inter-element effects are observed in arc- or spark-
emission spectroscopy, but these can be eliminated by using high
energy excitation, e.g., a radio-frequency plasma.  Continuous
monitoring systems for Be, Hg, and Pb, based on emission spec-
troscopy, have been developed using arc- or spark-excitation.
Similar approaches with either an arc-or-spark excitation or a
radio-frequency plasma can be applied to other elements.  With
a radio-frequency induced plasma, emission spectrographic analy-
ses for B, Ba, Cr, Cu, Mn, Ni, and V without inter-element effects
and on a continuous operational mode are possible.  Repeated
scanning of plasma induced emission lines of chromium and man-
ganese show high stability as indicated by relative standard
deviations of intensities of 1.95? and 1.7%, respectively, when
the sample is introduced continuously as an aqueous solution.
                              30

-------
     X-ray emission (fluorescence) spectrometry is not applicable
for measuring boron emissions.  With a high resolution semi-
conductor detector, x-ray emission techniques with conventional
x-ray target tubes and with radioisotope sources can be used on
an intermittent basis to determine Ba, Cr, Cu, Mn, Ni, and V in
particulate collected on filters.   Care must be taken to measure
thin film deposits so as to minimize matrix effects.  Also, the
background of secondary emission of x-rays from the filter media
can be a major interference.

     Radio-isotope sources for x-ray emission measurements pro-
vide a high degree of portability  and simplicity of ancillary
equipment.  When applied to monitoring particulate collected on
filters, good accuracy and repeatability of measurements are ob-
tained (see Appendix VI, Section 3.4, Rhodes ref. 88,89).  The
major limitations in applying the x-ray techniques are the time
to collect sufficient particulate on the filter and the potential
loss of volatile emissions containing the element sought.

     Nondestructive neutron activation analysis can be performed
on particulate for certain elements, but not all.   The major
problem is the interference from radioactive isotopes of other
elements.  Lengthy cooling times ranging from several hours to
several months are necessary to obtain sufficient sensitivity
for some elements, particularly Cr and Ni.  Manganese and vana-
dium can be measured immediately.   Depending on the detection
level required, copper can be measured immediately or, if a lower
detection level is necessary, after 20-30 hr decay time.

     By using high intensity reactor pulsing techniques, boron
can be determined by nondestructive neutron activation analysis
techniques, but not by conventional steady-state thermal neutron
irradiation.

     With fast neutrons from an isotopic source, much shorter
irradiating and cooling times are possible.  There is some sac-
rifice of sensitivity, but the technique provides for detecting
Ba, Cr, Cu, Mn, and V in approximately 30 minutes total time.
Measurement of Ni can be accomplished in approximately 2 hours.

     Continuous, on-stream (aqueous solutions), neutron acti-
vation analyses of V and Mn have been performed with isotopic
sources at the 4100 ppm level in 5-10 minute cycles.

     The chemiluminescent reaction of Ni(CO)i, with ozone in
presence of carbon monoxide deserves consideration as a possible
system for continuous monitoring of Ni(COK emissions.  Possible
applications including monitoring:  (a) effluent from the Mond
process of refining nickel, (b) formation of Ni(COK when CO in
effluent emissions passes over or through nickel products or
alloys, and (c) losses of nickel metal particulate from filter
media in particulate sampling systems.
                              31

-------
     In special cases, continuous colorimetric measures are
possible for certain boranes.  However, the application of the
techniques must be made with care to ensure absence of inter-
ferences .

     Measuring boron indirectly with mass spectrometry by deter-
mining the number of helium atoms generated in the 10B(n,a)
reaction should be further evaluated.
                              32

-------
                                                           Table VI
                               SUMMARY OF EVALUATION  PARAMETERS FOR MONITORING SYSTEMS FOR BARIUM
                                                  (Baaed on Literature Review)
Evaluation
Parameter
Specificity
Detection Limit
Range
Accuracy
Precision
Operation Mode
Sampling Time
Analysis Tine
Total Cycle Time
Calibration
Technique
Sample Form
Interferences
Unit Output
Special Equip
Multi-element
Application
Safety Hazard
Problem Areas
Routine
Maintenance
Volume FlOH Rate
Eat • Instrument
Cost
Est. Instrument
Size
Atonic Absorption
Spoctrophotometry
Yes
0.02 ug/m'2
-
No data
No data
Intermittent
21 hrs
Depends on collection
method and extraction
(10 mln to 2 hrs)
12H hrs
Known aqueous
standard solutions
Aqueous solutions of
dissolved partlculate
Catlonlc 1
anlonlc
ug/m1
Nitrous oxide-
acetylene flame
Depends on choice of
lamp and flame
Open flame
Catlonlc & anlonlc
Interferences
Replace fuels
_
»5, 000-10, 000
1' X 2' X 6'
Emission Spectroscopy
Flame
Yes
0 0035 ug/ml
-
No data
No data
Intermittent
No data


-




Catlonlc t
anlonlc
ug/ml
Nitrous oxide-
acetylene flame
Yes, for selected
elements
Open flame
Catlonlc 1
anlonlc inter-
ferences
Replace fuels
_
$5,000-10.000
»• X 2' X 6'
RP Plasma
Yes
0.0001 ug/ml
-
Ho data
No data
Intermittent
No data


-




None reported
ug/ml
Argon
Yes
Flame, high
voltage; RF
radiation
None reported
Replace Argon
-
(5,000-17,500
V x V i 8-
X-ray Emission
(Fluorescence)
Spectroscopy
Yes

No data
No data
No data
Intermittent
(Anal Filters)
21 hre
10-30 mln
24-25 hrs
Known samples on filter
Partlculate on filter
Matrix absorption,
particle size
"S
Vacuum or helium purge
Atomic number >11
X-rays, moderate to
high electrical
voltages
Matrix, Interelement, and
particle size effects,
background from filters,
resolution
Maintain vacuum or helium
purge to sample chamber,
If uaed
-
(10,000-30,000
1- X !>' X 6'
Neutron
Activation Analysis
Yes
0 02 ug
No data
No data
No data
Intermittent
(Anal. Filters)
21 hra
30 mln
21-25 hrs
Known samples
Partlculate on filter
Background from filters,
matrix background?
ug
Neutron reactor or
Isotope source
Yes, but certain elements
may require chemical
separation
Radiation
Background from filters,
some matrix background
•
-
>$100,000-
Masslve reactor-
- 2000 m* sample
- Isotoplc source may reduce price
- Isotoplc source may reduce size and weight
                                                              33

-------
                                        Table  VII

              SUMMARY OP EVALUATION PARAMETERS FOR  MONITORING  SYSTEMS  FOR  BORON
                              (Based on Literature  Review)
evaluation
Parameter
Specificity
Detection Limit
Range
Accuracy
Precision
Operation Mode
Sampling Time
Analysis Time
Total Cycle Time
Calibration
Technique
Sample Form
Interferences
Unit Output
Special Equip.
Multi-element
Application
Safety Hazard
Problem Areas
Routine
Maintenance
Volume Flow Rate
Est. Instrument
Cost
Est . Instrument
Size
Atomic Absorption
Spec trophotome try
Yes
3-6 pg/ml
Linear up to 200 ug/ml
No data
No data
Intermittent
-v-21 hrs
Depends on collection
method and extraction
-V.2IJ hrs
Known aqueous
standard solutions
Aqueous solutions of
dissolved partlculate
Matrix, oxide formation
ug/ml
Nitrous oxide-
acetylene flame
Depends on choice of
lamp and flame
Open flame
Refractory nature of
boron; needs high
temperature flame
Replace fuels
-
$5,000-10,000
ll' x 2' x 6'
RF Plasma
Emission Spectroscopy
Yes
0.03 yg/ml
-
No data
No data
Intermittent
No data
Depends on collection
method and extraction
•\-21 hrs
Known aqueous
standard solutions
Aqueous solutions of
dissolved partlculate
None reported
ug/ml
Argon
Yes
Flame; high voltage;
RF radiation
None reported
Replace Argon
-
$5,000-17,500
l|' x 1' x 8'
Neutron
Activation Analysis
Yes
1.1 ug
No data
No data
No data
Intermittent
21 hrs
30 mln
•v.24 hrs
Known standard
samples
Partlculate
Some matrix effects
Ug
Requires pulsed
neutron flux
B, Be, Li, 0, F, Mg,
and Pb In pulsed mode
Radiation
Matrix effects
-
-
>$ioo,ooo2
Massive reactor-
*
  - isotopic source may reduce cost and size.

-------
                                                         Table  viii

                            SUMMARY OF EVALUATION PARAMETERS FOR MONITORING  SYSTEMS  FOR  CHROMIUM
                                                (Based on Literature Review)
Evaluation
Parameters
Specificity
Detection Limit
Range
Accuracy
Precision
Operation Mode
Sampling Time
Analysis Time
Total Cycle Tine
Calibration
Technique
Atomic Absorption
Spectrophotometry
Yes
0 002 pg/m'
-
No data
No data
Intermittent
21 hrs
Depends on collec-
tion method and
extraction (10 mln
to 2 hrs)
124 hrs
Known aqueous
standard solutions
RF Plasma Emission
Spectroscopy
Yes
0.001 ug/ml
-
No data
No data
Intermittent
No data
Depends on collec-
tion method and
extraction (10 mln
to 2 hrs)
-
Known aqueous
standard solutions
X-ray Emission
(Fluorescence) Spectroscopy
Yes (requires proper
resolution)
0.053 dg/m1
No data
No data
No data
Intermittent (Anal filters)
24 hrs
Approximately 1 hour
121 hrs
Standard samples of
partlculate on filters
Neutron Activation
Analysis
Yes
0 02 ug
-
No data
No data
Intermittent (Anal, filters)
21 hrs
Requires 20-30 day decay
time or chemical separa-
tion
-
Standard samples of
partlculate on filters
Sample Form


Interferences



Unit Output

Special Equip.


Multi-element
Application

Safety Hazard


Problem Areas
Aqueous solutions of
dissolved partlculate
Aqueous solutions of
dissolved partlculate
Interelement effects    None reported
UB/m'

Should use nitrous
oxide-acetylene flame

Depends on choice
of lamp and flame

Open flame


Interelement effects
lig/ml

Argon


Yes
Flame; high voltage,
RF radiation

None reported
                                                                    Partlculate on filters
Filter background, matrix
elements, Inter-element
effects

UK

Vacuum or helium purge,
semiconductor detector

Atomic number >11
X-ray, moderate to high
electrical voltages

Matrix; Interelement and
particle size effects,
background from filters,
resolution
                                                                                                   Paniculate on filters
                                                                               Matrix  elements
ug
                                                       Yes


                                                       Radiation
                                                                                                   Matrix  elements which
                                                                                                   require chemical separa-
                                                                                                   tion or lengthy decay
                                                                                                   times
Routine
Maintenance
Volume Flow Rate
Est . Instrument
Cost
Est . Instrument
Size
Replace fuels
-
$5,000-10,000
4' x 2' x 6'
Replace Argon
-
15,000-17,500
1' x 4' x 8'
Maintain vacuum or helium
purge to sample chamber
-
no, 000-30, ooo
4' x I1 x 6'
~
-
>S100,000
Massive reactor
                                                           35

-------
                             Table IX

SUMMARY OF EVALUATION PARAMETERS FOR MONITORING SYSTEMS FOR COPPER
                   (Based on Literature Review)
Evaluation
Parameters
Specificity
Detection Limit
Range
Accuracy
Precision
Operation Mode
Sampling Time
Analysis Time
Total Cycle Time
Calibration
Technique
00
CT\ Sample Form
Interferences
Unit Output
Special Equip.
Multi-element
Application
Safety Hazard
Problem Areas
Routine
Maintenance
Volume Plow Rate
Est . Instrument
Cost
Est Instrument
Size
Atomic Absorption
Spectrophotometry
Yes
0.001 ug/rn'
-
No data
No data
Intermittent
24 hrs
Depends on collec-
tion method and
extraction (10 mln
to 2 hrs)
•V24 hrs
Known aqueous
standard solutions
Aqueous solutions of
dissolved partlculate
Interelement effects
lig/m1
Should use nitrous
oxide-acetylene flame
Depends on choice
of lamp and flame
Open flame
Interelement effects
Replace fuels
-
(5,000-10,000
II1 x 2- x 6'
RF Plasma Emission
Spectroscopy
Yes
Insufficient data
-
No data
No data
Intermittent
No data
Depends on collec-
tion method and
extraction (10 mln
to 2 hrs)
-
Known aqueous
standard solutions
Aqueous solutions of
dissolved partlculate
None reported
lig/ml
Argon
Yes
Flame, high voltage,
RF radiation
None reported
Replace Argon
-
*5, 000-17, 500
1- x H- x 8'
X-ray Emission
(Fluorescence) Spectroscopy
Yes, with proper resolution
0.011 iig/m*
No data
No data
No data
Intermittent (Anal, filters)
24 hrs
Approximately 1 hour
1.2 * hrs
Standard samples of
partlculate on filters
Partlculate on filters
Filter background; matrix
elements ; Inter-element
effects
UK
Vacuum or helium purge,
semiconductor detector
Atomic number >11
X-ray; moderate to high
electrical voltages
Matrix, Interelement and
particle size effects;
background from filters,
resolution
Maintain vacuum or helium
purge to sample chamber
-
$10,000-30,000
4- x H' x 6'
Neutron Activation
Analysis
Yes
0.1 WE
-
No data
No data






Intermittent (Anal, filters)
21 hrs
Requires 3 mln decay
-

time

Standard samples of
partlculate on filters
Partlculate on filters
Matrix elements
ue
-
Yes
Radiation
Requires 3 mln decay
-
-
>noo,ooo
Massive reactor





time





-------
                               Table X



SUMMARY OF EVALUATION PARAMETERS FOR MONITORING SYSTEMS FOR MANGANESE
Evaluation
Parameters
Specificity
Detection Limit
Range
Accuracy
Precision
Operation Mode
Sampling Time
Analysis Time
Total Cycle Time
Calibration
Technique
Sample Form
Interferences
Unit Output
Special Equip.
Multi-element
Application
Safety Hazard
Problem Areas
Routine
Maintenance
Volume Flow Rate
Eat . Instrument
Cost
Est. Instrument
Size
Atonic Absorption
Speetrophotometry
Yes
0.001 ug/m1
-
No data
No data
Intermittent
24 hrs
Depends on collec-
tion method and
extraction (10 mln
to 2 hrs)
i24 hrs
Known aqueous
standard solutions
Aqueous solutions of
dissolved partlculate
Interelement effects
Ug/m'
Should use nitrous
oxide-acetylene flame
Depends on choice
of lamp and flame
Open flame
Interelement effects
Replace fuels
-
$5,000-10,000
4' x 2' x 6'
RF Plasma Emission
Spectroscopy
Yes
Insufficient data
-
No data
No data
Intermittent
No data
Depends on collec-
tion method and
extraction (10 mln
to 2 hrs)
-
Known aqueous
standard solutions
Aqueous solutions of
dissolved partlculate
None reported
Ug/ml
Argon
Yes
Flame, high voltage,
RF radiation
None reported
Replace Argon
-
$5,000-17,500
4' x 1)' x 8'
X-ray Emission
(Fluorescence) Spectroscopy
Yes, with proper resolution
0.027 ug/m1
No data
No data
No data
Intermittent (Anal, filters)
24 hrs
Approximately 1 hour
•>-24 hrs
Standard samples of
partlculate on filters
Partlculate on filters
Filter background; matrix
elements ; Inter-element
effects
Pg
Vacuum or helium purge;
semiconductor detector
Atomic number >11
X-ray, moderate to high
electrical voltages
Matrix, Interelement and
particle size effects,
background from filters ,
resolution
Maintain vacuum or helium
purge to sample chamber
-
$10,000-30,000
4' x 4' x 6'
Neutron Activation
Analysis
Yes
0.003 Ug
-
No data
No data






Intermittent (Anal filters)
24 hrs
Requires 15 mln decay
-
Standard samples of
partlculate on filters
Partlculate on filters
Matrix elements
PK
-
Yes
Radiation
Requires 15 mln decay
-
-
>$100,000
Massive reactor

time








time





-------
                                                                                     Table XI
                                                       SUMMARY OF EVALUATION PARAMETERS FOR MONITORING SYSTEMS FOR NICKEL
                                                                           (Based on Literature Review)
LO
CO
Evaluation
Parameters
Specificity
Detection Limit
Range
Accuracy
Precision
Operation Mode
Sampling Time
Analysis Time
Total Cycle Time
Calibration
Technique
Sample Form
Interferences
Unit Output
Special Equip.
Mult 1 -element
Application
Safety Hazard
Problem Areas
Routine
Maintenance
Volume Flow Rate
Est . Instrument
Cost
Est. Instrument
Size
Atomic Absorption
Spec trophotomet ry
Yes
O.OQll wg/m*
-
No data
No data
Intermittent
24 hrs
Depends on collec-
tion method and
extraction (10 mln
to 2 hrs)
1.24 hrs
Known aqueous
standard solutions
Aqueous solutions of
dissolved partlculate

U6/m3
Should use nitrous
oxide-acetylene flame
Depends on choice
of lamp and flame
Open flame
Interelement effects
Replace fuels
-
$5,000-10,000
4- x 2- x 6'
RF Plasma Emission
Spectroscopy
Yes
0.006 ug/ml
-
No data
No data
Intermittent
No data
Depends on collec-
tion method and
extraction (10 mln
to 2 hrs)
-
Known aqueous
standard solutions
Aqueous solutions of
dissolved partlculate

Ug/ml
Argon
Yes
Flame, high voltage,
RF radiation
None reported
Replace Argon
-
$5,000-17,500
1' x 4- x 8'
X-ray Emission
(Fluorescence) Spectroscopy
Yes, with proper resolution
0.013 wg/m1
No data
No data
No data
Intermittent (Anal, filters)
24 hrs
Approximately 1 hour
t24 hrs
Standard samples of
partlculate on filters
Partlculate on filters
Filter background, matrix
elements ; Inter-element
effects
Ug
Vacuum or helium purge ,
semiconductor detector
Atomic number >11
X-ray, moderate to high
electrical voltages
Matrix, Interelement and
particle size effects,
background from filters,
resolution
Maintain vacuum or helium
purge to sample chamber
-
$10,000-30,000
4' x 4' x 61
Neutron Activation
Analysis
Yes
1.5 ug
-
No data
No data
Intermittent (Anal, filters)
24 hrs
Requires chemical separation
or 20-30 day decay time
-
Standard samples of
partlculate on filters
Partlculate on filters
Matrix elements
PS
-
Yes
Radiation
Requires chemical separation
or 20-30 day decay time
-
-
>tioo,ooo
Massive reactor

-------
                             Table XII

SUMMARY OP EVALUATION PARAMETERS FOR MONITORING SYSTEMS FOR VANADIUM
                   (Based on Literature Review)
Evaluat Ion
Parameters
Specificity
Detection Limit
Range
Accuracy
Precision
Operation Mode
Sampling Time
Analysis Time
Total Cycle Time
Calibration
Technique
Sample Porm
Interferences
Unit Output
Special Equip.
Multi-element
Application
Safety Hazard
Problem Areas
Routine
Maintenance
Volume Flow Rate
Eat. Instrument
Cost
Est. Instrument
Size
Atonic Absorption
Speetrophotometry
Yes
O.D1 ug/m'
-
No data
No data
Intermittent
24 hrs
Depends on collec-
tion method and
extraction (10 mln
to 2 hrs)
124 hrs
Known aqueous
standard solutions
Aqueous solutions of
dissolved partlculate
Interelement effects
Pg/m1
Should use nitrous
oxide-acetylene flame
Depends on choice
of lamp and flame
Open flame
Interelement effects
Replace fuels
-
$5,000-10,000
4- x 2- x 6'
RF Plasma Emission
Spectroscopy
Yes
0.006 ug/ml
-
No data
No data
Intermittent
No data
Depends on collec-
tion method and
extraction (10 mln
to 2 hrs)
-
Known aqueous
standard solutions
Aqueous solutions of
Slssolved partlculate
None reported
Pg/ml
Argon
Yes
Flame, high voltage,
RF radiation
None reported
Replace Argon
-
$5,000-17,500
4' x 4- x 3-
X-ray Emission
(Fluorescence) Spectroscopy
Yes, Kith proper resolution
0.007 Ug/m'
No data
No data
No data
Intermittent (Anal, filters)
24 hrs
Approximately 1 hour
1.24 hrs
Standard samples of
partlculate on filters
Partlculate on filters
Filter background; matrix
elements ; Inter-element
effects
Pg
Vacuum or helium purge;
semiconductor detector
Atomic number >11
X-ray, moderate to high
electrical voltages
Matrix, Interelement and
particle size effects;
background from filters ,
resolution
Maintain vacuum or helium
purge to sample chamber
-
$10,000-30,000
4- x 4' x 6'
Neutron Activation
Analysis
Yes
0.001 ug
-
No data
No data
Intermittent (Anal.
24 hrs
3 mln decay time
-





filters)



Standard samples of
partlculate on filters
Partlculate on filters
Matrix elements
vg
-
Yes
Radiation
3 mln decay time
-
-
>$100,000
Massive reactor











-------
1.   PRELIMINARY FIELD STUDIES

     An evaluation of the published information available for
the four stationary sources - power plant, incinerator, ore
refineries, and machining operation - identified as being major
sources of beryllium and cadmium revealed the lack of specific
process information which defines the physical and chemical
properties of the emissions.  Process data was needed to estab-
lish the sensitivity and resolution requirements of the monitor,
its concentration span, the potential sources of interference
(chemical and physical) that it must cope with, and the require-
ments of sampling and delivery equipment to ensure that a
representative sample is presented to the analyzer.

     A preliminary field test study at three field test sites -
power plant, incinerator, and machining operation - was devised
to establish the following information:

     Rate of emission of Be and Cd, weight/volume of effluent.

     Overall composition of emission, including potential
     interferences (gases, aerosols, and other metals).

     Representative particle size distribution of particulate
     emitted.

     Distribution of Be and Cd within this size distribution.

*J.l  Preliminary Sampling Study - Power Station

     Samples were collected from the power station site at a
15-ft square duct after the electrostatic precipitator and after
the I.D. Fan.  The sample was taken with a Brink® Model BMS-11
Cascade Impactor through a 3-in. diameter port.  The objective
of this effort was to determine the particle size distribution
of the fly ash and establish the sensitivity of the sampling
system to non-isokinetic sampling practices.

     The Brink® Cascade Impactor contains a glass cyclone, a
five stage impactor and a dust filter assembly.  The cyclone
removes all particles larger than 7 microns from the gas sample
with the remainder of the dust collected in the impactor according
to particle size; i.e., the largest particles collecting in the
first stage and successively smaller particles collecting on the
lower stages.  Table XIII lists the particle size range collected
in each stage of the impactor for the sampling conditions employed
in this study.

     Since the objective of this effort was to determine the
sensitivity of the Be/Cd sampling system to non-isokinetic
sampling, the flow rate through the nozzle of the probe was set
below isokinetic velocity to favor the collection of the larger

-------
                                     Table XIII
      BERYLLIUM AND CADMIUM CONTENT OF PARTICULATE COLLECTED FROM POWER STATION
Analyses of Combined Runs I &
Be
Stage
Cyclone
First
Second
Third
Fourth
Fifth
Filter
Particle
Size Range
(microns)
>7.0
1.9-7.0
1.1-1.9
0.7^-1.1
0.36-0.74
0.24-0.36
0.2-0.24
Particulate
Collected
(mg)
(Runs I & II)
0.00 0.00
6.29 7.605
2.735 2.08
0.88 0.535
0.44 0.205
0.10 0.05
0.135 0.105
Cone . as
Particulate
in Flue Gas
(ve/m3)
_
20.5
7.7
1.6
0.47
^0.10
—
Cone . in
Particulate
(ppm)
_
118
127
88
59
^53
—
II
Cd
Cone, as
Particulate
in Flue Gas
(Ug/m3)
_
0.53
0.50
N.D.
^0.11
N.D.
—
Cone . in
Particulate
(ppm)
_
3.4
9.5
N.D.
14
N.D.
—
Total
10.58  10.58
30.4
1.14
Footnotes:
   (1)  Particulate size distribution obtained with the Brinks Cascade Impactor.
   (2)  Analyses performed by Atomic Absorption Spectrophotometry-Carbon Rod Assembly.
   (3)  Sample taken from power station flue gas after electrostatic precipitator.
   (4)  N.D. - Not detectable above blank.

-------
particle sizes.  Should the mass median particle  size obtained
under these  sampling conditions be below  5 microns,  it would be
safe to assume that a reasonably representative dust sample could
be obtained  even under non-isokinetic conditions.  Table XIII
gives the weight of dust  collected in each stage  for two dif-
ferent dust  samples.

     The particle size distributions are  shown graphically in
Figure 1.  For both samples, more than 80 wt-$ of the collected
particles had diameters below 5 microns.  Since these samples
were obtained under sub-isokinetic conditions, the particle size
distribution is skewed towards the larger particle sizes.  An
isokinetic sample might therefore give an even higher percentage
of particles below 5 microns.  With this  information it is safe
to assume that a reasonably good sample from power plant flue
gas can be obtained without isokinetic sampling.

     The particulate collected in the various stages of the
Brink Cascade Impactor was digested with an acid  solution (see
Preliminary EPA Method 11.  Determination of Beryllium from
Stationary Sources) and analyzed for beryllium and cadmium by
non-flame atomic absorption spectrophotometry.  The atomic ab-
sorption measurements were made with a Perkin-Elmer Model 303
Atomic Absorption Spectrophotometer equipped with a Varian
Techtron Model 61 Carbon Rod Atomizer.

     The analyses for the power plant are reported in Table XIII.
The results for the power plant were obtained by  combining the
collected fractions for two flue gas samplings.

     Based on two 20-minute collections at a flow rate of
2 liters/min., the analyses indicate concentrations in the flue
gas of approximately 30 yg Be/m3 and 1.2 yg Cd/m3 in particles
between 0.24 micron and 7.0 microns.  Analysis of fly ash
deposited in the flues shows a concentration of 89 ppm of Be
in the ash.  The beryllium content appears to be  lower in the
smaller particles than in the larger particles, whereas the
limited data suggest the opposite for cadmium.

     The semi-quantitative emission spectrographic analysis of
particulate  (power plant) from the first stage of Brink Cascade
Impactor is shown in Table XIV.

4.2  Preliminary Sampling Study - Municipal Solid Waste
     Incinerator

     A particulate sample was collected from a port in the ex-
haust stack of a municipal incinerator after the caustic scrubber
by using a Brink® Model BMS-11 Cascade Impactor described in
Section 4.1 of this report.

-------
    10
•5!
I
ra
O_
    1.0
    0.1
           nRun #2, Table XIII
           •Run #1, Table XIII
                   Sample Port: Breeching to Stack No. 5
                   Gas Temperature: 270°F
                   Gas Velocity: ~ 3000 ft/min
                   Probe Tip Diameter:  2.0 millimeters
                   Sample Rate: ~2.0 liters/min
                   Velocity at Probe Tip: -1800 ft/min
                   Sample Time: 20 minutes
     0.01  0.5 a 1   0.5  1  2
5   10   20  30 40 50 60 70  80    90
  Cumulative Mass Percent Less Than Dp
95    98  99
99.9    99.99
            Figure 1.   Fly  Ash  Particle  Size Distribution From
                          Power Plant Test  Site  (February  1972).

-------
                   Table XIV

SEMI-QUANTITATIVE  (±20%) EMISSION SPECTROGRAPHIC
    ANALYSIS OF PARTICIPATE FROM 1st STAGE
          OF BRINK® CASCADE IMPACTOR
Al - Major Constituent
Si - Major Constituent
Fe - Minor Constituent
Mg - 0.7%
Ca - 0.6%
Ti - 0.3%
Na - 0.3%
Cu - 0.05%
Mn - 0.03%
Ni - 0.02%
Zr - 0.02%
 V - 0.02%
Be - Not Detected; if present <0.01% (<100 ppm)
Cd - Not Detected; if present <0.05% (<500 ppm)

-------
     The first attempt (Run I) to sample the incinerator by
"conventional techniques," i.e., by withdrawing a sample from
the stack through a heated line, proved to be inadequate.  The
temperature of the delivery line could not be controlled within
the narrow temperature range required to prevent condensation
of water vapor or evaporation of entrained water.  In the first
sampling test, a considerable quantity of water was collected
in the cyclone and none in the impactor.

     To eliminate the problems associated with transporting the
sample to the impactor, the sampling procedure was changed as
follows for Run II.

1.   The impactor was removed from the sample case.

2.   The sampling nozzle was fitted directly into the Impactor
     inlet.

3.   Thirty-foot lengths of rubber tubing were connected to
     the inlet and out pressure taps and vacuum outlet.

Jj.   The impactor was lowered into the stack by means of a
     suitable length of rope.

     The impactor was lowered into the stack about two feet from
the stack wall by means of an extension rod.  To prevent conden-
sation of liquid within the cold impactor, it was held in the
stack for two minutes prior to sampling.

     The gas flow rate was adjusted to 3-0 liters/minute and
maintained at this rate for 10.0 minutes.  Thus, a total of
30 liters (0.03 m3) was sampled.

     The particulate collected in the various stages of the
Brink Cascade Impactor was digested with an acid solution (see
Preliminary EPA Method 11.  Determination of Beryllium from
Stationary Sources) and analyzed for beryllium and cadmium by
non-flame atomic absorption spectrophotometry.  The atomic
absorption measurements were made with a Perkin-Elmer Model 303
Atomic Absorption Spectrophotometer equipped with a Varian
Techtron Model 61 Carbon Rod Atomizer.

     During the collection of aerosol from the incinerator (Run
II), liquid condensate was deposited in each impact compartment
of the Brink Cascade Impactor and no weights of individual frac-
tions could be obtained.  However, by allowing the moisture to
evaporate at room temperature and dissolving the residue in known
volumes of high purity acid, estimates of the distribution of
aerosol and the concentration of Be and Cd can be made.  These
results are reported in Table XV.

-------
                         Table XV
     BERYLLIUM AND CADMIUM CONTENT OP PARTICULATE AND
 AEROSOL COLLECTION FROM MUNICIPAL SOLID-WASTE INCINERATOR
Concentration of Element Collected as
Particulate or Aerosol (yg/m3)
Particle Size
(microns)
7.0 (Cyclone)
1.9-7.0
1.9-1.9
0. 74-1.1
0.36-0.71*
0.24-0.36
Total

Be
1.9
N.D.
N.D.
N.D.
N.D.
N.D.
1.9
Run I
Cd
46
0.097
0.064
0.033
N.D.
0.093
46
Run
Be
N.C.
0.083
0.10
0.13
0.073
0.033
0.42
II
Cd
N.C.
0.016
0.31
12
7.3
17
37
Footnotes:
   (1)  Particulate size distribution obtained with the
        Brinks Cascade Impactor.
   (2)  Analyses performed by Atomic Absorption
        Spectrophotometry-Carbon Rod Assembly.
   (3)  N.D. - Not detected
   (4)  N.C. - Not collected
   (5)  Approximately 2 ml of water was obtained as
        condensate in the cyclone.
                            46

-------
     Two collections of aerosols were made.  Run I was performed
with little control of inlet heat to the impactor and as a con-
sequence the water aerosol deposited principally as condensate
in the first stage (cyclone).  Better control of inlet heat on
the impactor in Run II resulted in aerosol deposition through
the train.  When condensation occurred (Run I), no beryllium was
detected in the Impactor stages which would isolate particulate
in the 0.24 micron to 7.0 micron range; however, very small quan-
tities of cadmium were detected in a relatively uniform distri-
bution in 4 of the 5 stages in the same particle size range.

     When condensation was minimized (Run  II), concentrations of
approximately 0.4 to 2 yg Be/m3 and 40 yg Cd/m3 were collected
in the particle size range of 0.24 micron to 7.0 microns.  The
distribution of beryllium appears to be approximately the same
in the four stages through the particle size range of 0.36 micron
to 7-0 microns, but slightly lower in the range of 0.24-0.36
micron.  The distribution of cadmium in the various stages of
the cascade impactor is markedly different from that observed
for beryllium; cadmium concentrations are greater on the stages
that isolate the smaller particles 0.24 micron to 1.1 microns,
than on the stages retaining particles of 1.1 microns to 7.0
microns.
4.3  Preliminary Field Study - Machining Operation of
     Beryllium Metal

     Preliminary stack emission samples were taken at a machining
operation which was being considered as a candidate field test
site.  The machining operation has three metal cutting machines
which fabricate beryllium metal parts.  On the day for sampling,
only one of the three was operating.  Visual inspection of the
air leaving the plant suggested that a high volume of sample
would be required to obtain a measurable quantity of particulate.
The sample probe was located after the fan and two absolute
filters in the dust cleaning system.  Visual inspection of the
dust cleaning system further suggested that the particles that
pass through the second absolute filter are of a very small
size.  For this reason, the cascade impactor was abandoned in
favor of a 0.2 micron filter.

     A total of 240 cu.ft was drawn through the filter over a
period of four hours (1.0 cfm).  Weighing the filter paper after
the sampling period revealed no apparent weight gain.  The
balance has a maximum accuracy of 0.05 mg.  If it Is assumed
that 0.05 mg was deposited upon the filter, then the dust con-
centration would be:

           0.05 mg      35.2 cu.ft  _  0.007 mg
          240 cu.ft         iP            iff3

     No beryllium was detected by atomic absorption spectro-
photometric analysis of the filter.

                              47

-------
5.   INSTRUMENT DESIGN AND FEASIBILITY (LABORATORY) TESTING


5.1  Selection of Prototype Design

     5.1.1  Direction Indicated for Experimental Plan

     An evaluation of the literature data and the preliminary
field studies indicated that a monitor based on continuous-flow
emission spectrographic techniques was the most promising of all
approaches currently available for monitoring Be and Cd from
stationary source emissions.  Earlier studies with arc excitation,
showing a detection limit of 0.5 to 1 ug Be/m3 at flow rates of
J<0 1/min, demonstrated the application of the technique for con-
tinuous monitoring of Be emission in work environment atmospheres
with approximately 30-60 second sampling times.  Several instru-
ments based on the emission spectroscopic concept had been used
for monitoring Be in England, Prance, Yugoslavia, Japan, and the
United States.

     In all previous applications, the emission spectrographic
technique was applied to relatively clean atmospheres containing
small quantities of particulate composed of principally beryllium
or its compounds.  The potential problems of electrode wear and
of particulate depositing on electrodes of the arc system or not
being completely vaporized or excited to generate optical emis-
sion spectra, indicated the desirability for an electrodeless,
more energetic, discharge system.  Therefore, a development pro-
gram involving an electrodeless radio-frequency induced plasma
was devised to develop an excitation system where the stack gases
could be excited to optical emission directly or after dilution
with an inert gas and where the problems of electrode wear and
particulate deposition would be eliminated.  The major problems
anticipated for the radio-frequency induced plasma were the abil-
ity to maintain a stable plasma in an air rich gas and to detect
elemental optical emission spectra which would not be affected
by oxidation phenomena in the air matrix.

     The program included three separate, but related, approaches
to generating the optical emission spectra:

1.   A commercial radio-frequency (RF) generator capable of
     variable frequencies output (15 to 50 MHz) and a maximum
     continuous power output of 2500 watts was applied to
     develop air and argon-air plasmas.

2.   An attempt was made to develop a light-weight, portable,
     radiofrequency (RF) generator system based on an amateur
     radio transmitter and a high frequency linear amplifier
     which was tunable from 21 to 21.5 MHz or 28.5 to 29 MHz
     with a continuous power output capability of 1250 watts.
     The system would provide an electrodeless RF induced plasma
     in air or air-argon mixtures.

-------
3.   The Webb cell (see Appendix VII, Section 2.3) was obtained
     from the developer for use in an AC discharge (Pt electrode)
     system.

     Initially, the discharge systems were evaluated in laboratory
tests to establish feasibility.  Finally, the Webb arc discharge
system was evaluated in field tests at a power plant.

     5.1.2  Source Sampling Considerations

     A valid stationary source monitoring system must deliver a
representative sample of the emission components to be monitored
to the analytical detector.  Appendix I of this report presents
a detailed discussion of sampling system considerations.

     Previous experience by the Environmental Protection Agency
in sampling and collection of particulate from a beryllium
machining plant using a typical EPA sampling train indicated
that beryllium was uniformly distributed throughout the train.
In general, little beryllium-containing particulate was obtained
on the filter and a majority of the metal was found on analysis
of the probe wash solution or in the impinger train.  The
beryllium found in the impinger train did not decrease progres-
sively when a number of impingers were arranged in sequence.
This slippage of material through the filter and the potential
problems in volatilization of metals in the oxidizing and reactive
gas constituents characteristic of a number of stationary sources
points to selection of a monitoring system where all of the emis-
sion components are sampled and transported to an analysis stage
as a combined vapor/aerosol mixture.  This approach in turn is
dependent upon selection of an analytical detection approach for
Be/Cd which will not be adversely affected by the presence of all
other components in the emissions matrix.

     The preliminary field studies discussed in the previous
subsection indicated that the particulate containing beryllium
and cadmium from three major sources was largely in size ranges
amenable to non-isokinetic sampling (e.g., <^4y diameter; see
Appendix I, Section 3.0).  Any industrial facility, however,
which employs a wet scrubber treatment will emit a binodal par-
ticle size distribution.  One maximum is associated with water
droplets >30y in diameter and the other maximum associated with
particulate generally below 5y in diameter.  Since the water
droplets can incorporate fine particulate containing beryllium
and cadmium, an isokinetic sampler must be employed to accommodate
a facility employing a wet scrubber stage.

     Appendix I contains a detailed examination of the require-
ments for isokinetic sampling and the design of the sample
transport system required to minimize wall loss of sampled
aerosol by gravitational settling or diffusion.  Puchs' treatise
on the mechanics of aerosols (ref. 16) was employed to conduct
calculations of flow rates, transfer line diameters, etc., which

-------
would optimize transport of the sampled aerosol from the sampling
probe to the analysis instrumentation.  This analysis showed that
aerosol loss by diffusion to the transport line walls would be
negligible for any line diameter.  This conclusion is in contrast
to the situation when sampling of pollutant gases is desired,
since these gases have diffusion coefficients several orders of
magnitude larger than even submicron particulate.

     The major considerations, therefore, for proper sampling
of particulate matter and interface with the analyzer are to
ensure isokinetic sampling conditions and to design the sample
transport lines in a manner to minimize settling of particulate
under the influence of gravity.  Based on these considerations,
a sampling system was designed which included isokinetic sampling
of a primary stream at flow rates of about 2 cu.ft./min. followed
by secondary sampling of this primary stream to isolate a con-
stant flow sample for delivery to the analytical instrumentation.
The sampling system was designed to employ an isokinetic sampling
probe system.  The secondary sampling of this stream was designed
and packaged in a Be/Cd sampling compartment where the gas
transport system is interfaced with the rf-induced emission
spectroscopy technique.  A schematic of the Be/Cd sample compart-
ment is shown in Figure 2.

     Later, field testing of the AC discharge arc emission detec-
tion approach was conducted at a power plant employing a sampling
facility  used  earlier on Contract EHSD 71-30.  This field
sampling facility is described in detail in Appendix I.


5.2  Radio-Frequency Excited Optical Emission Spectroscopy

     5.2.1  Instrumentation

     Two types of excitation sources (a) Lepel generator and
(b) Drake radio transmitter were used to generate radio-frequency
(RF) induced plasmas.  The Lepel generator, weighing approximately
270 pounds and requiring a 15-gallon heat exchanger is not port-
able, but is transportable and costs $3200.  The Drake radio
transmitter system weighs less than 100 pounds, requires minimal
water cooling (1/2 gal/min) and power, and costs $2200.

     Both systems provide a tunable frequency generator capable
of producing a radio-frequency of 30 MHz.  This frequency has
been used successfully to initiate and sustain radio-frequency
induced plasmas in a variety of spectroscopic studies (ref.
8,9,10).

          5.2.1.1  Lepel Radio-Frequency Generator System

     The Lepel generator is a high-frequency induction heating
unit (Lepel Model T-2.5-1-MC2-BW Type T-252-53, Lepel High
Frequency Laboratories, Inc., 59-21 Queens Midtown Expressway,


                              50

-------
                       Insulated  Enclosure
2mm Sample
   Tube
  Solenoid
   Valves
  Discharge
    Tube

    Quartz
    Window
     Filter
      Main Gas
  —  Sampling
        Tube
 Temp.
 Power Supply to
    Heaters
 Pressure Gauges
                                                        Control Valves
                                                          Critical Flow
                                                           Orifices
 Air Filter

Air
  Gas Pump
             Figure 2.  Be/Cd Sample  Compartment.
                                51

-------
Maspeth, New York City, New York 11378) designed and built speci-
fically for generating RF plasmas.  The unit is tunable in
frequency from 15 to 50 MHz with a maximum continuous power
output at the load of 2500 watts.

     The temperature control system supplied with the generator
was modified by adding an external heat exchanger.  A circulation
pump (minimum 2 gallons per hour at 30 psi with maximum 86°F water
temperature), 5-gallon distilled water reservoir, and 15-gallon
cooling water heat exchanger with 50 feet of one-half inch copper
tubing for the distilled water were assembled.  Cooling water
was city tap water circulated through the 15-gallon reservoir.

     Front panel terminals on the generator were provided to
attach the water-cooled load coil for generating the RF plasma.
The plasma was confined in a quartz tube, and a pressure cylinder,
regulator, and rotameter were used to admit the plasma gas to
the base of the plasma tube via a holder with tangential inlets.

     The plasma shape was very Important in order to get a gaseous
sample into a region of excitation (ref. 11,12).  A toroidal, or
donut-shaped, plasma was desirable since it allowed sample to
enter up the central cooler region for proper excitation.  An
undesirable plasma was elongated and bullet-shaped with a hot
zone up the middle and cooler at the edges.  With such a plasma
the sample would tend to be thrown out to the cooler area along
the sides of the plasma tube rather than pass through the hotter
central excitation zone.

     The tangential plasma gas inlet with resulting vortex flow
and, in later studies, coaxial flow using concentric quartz tubes
was required to form a toroidal plasma.  An excitation frequency
appropriate for each type of plasma gas and plasma tube diameter
was also necessary.

     Load coils of 3/16" and 1/V refrigeration tubing were used.
One to five turn coils were wound and used in attempts to form
desirable plasmas, but three turn coils were found to be the best
performers.  Spacing between turns was varied from 1/16" to over
1/2" with large diameter plasma tubes.  Inner diameters of the
load coils were varied to match the outer diameters of the quartz
tubes up to 1/2" or more clearance with the larger plasma tubes.
Some miscellaneous coils shaped like cones or pancakes were also
wound.

     The tangentially injected plasma gas flowed through the
discharge tube in a counterclockwise (CCW) direction.  Coils were
wound so that their ascending turns coincided with the direction
of the ascending gas.  In this configuration, the ground side of
the load coil was nearest the plasma gas inlet.  Alternatively,
the turns descended CCW opposing the CCW ascending gas, such that
the ground side of the load coil was in this case nearest the
                              52

-------
upper exit end of the plasma tube.  No effect was observed on
plasma operation.  This was contrary to results of an earlier
study (ref. 13).

     Quartz was used for plasma tubes to resist the high tempera-
tures of the plasma.  Outside diameters were 15 mm, 20 mm, 22 mm,
25 mm, 28 mm, 32 mm, 38 mm, and 48 mm.  The tubes 25 mm and
smaller had 1 mm walls, while the others had 1-1/2 or 2 mm walls.
All were approximately 10 to 14 inches long.  All sizes except
the 15 mm tube were used alone with tangential plasma gas inlet
and vortex flow.  The 20 mm and 28 mm tubes were used as outer
tubes with a concentric inner 15 mm tube to form coaxial plasma
gas flows.

     The discharge tubes were mounted in two types of holders.
Initial studies were performed with phenolic, aluminum, and brass
holders designed for 28 mm, 38 mm, and 48 mm tubes.  Later,
aluminum and stainless steel holders were constructed so that
sample gas could come in contact only with the stainless steel.
The inner tube in the concentric assembly was 15 mm, and the
outer tube was interchangeable, either 20 mm or 28 mm.

     The aluminum-stainless steel holders could be used with con-
centric tubes or disassembled for single plasma tube studies.
All holders had a provision for inserting a 6 mm diameter tube
through the bottom to inject gaseous samples at high velocity
into the plasma.  The larger holders could be used with slightly
smaller tubes wrapped with tape to provide a gas-tight fit.  Con-
centric tube holders similar in concept have been used by several
other workers (ref. 9jl3)s but not all have employed the great
versatility of interchangeable parts used here.

          5.2.1.2  Drake Radio-Transmitter Radio-Frequency
                   Generator System

     An amateur radio transmitter (Drake Model T-4XB with Drake
AC-4 power supply, R. L. Drake Company, Miamisburg, Ohio 45342)
with an input power of 200 watts tunable from 21 to 21.5 MHz and
28.5 to 29 MHz served as the RF source.  To obtain the necessary
power to sustain an RF plasma, the Drake Transmitter was used to
drive a high frequency linear amplifier (Model PA70-A, Ehrhorn
Technological Operations, Inc., Brooksville, Florida 33512) with
a continuous power output capability of 1250 watts.

     The load coil consisted of copper refrigeration tubing wound
in a helical shape to conform to the outside shape of the quartz
plasma type (20 mm O.D.).  The quartz tube merely confines the
plasma gas and substance to be excited (beryllium and cadmium in
this investigation) in a defined zone.  A block diagram serves
to illustrate this  (Figure 3).
                              53

-------
                                               Quartz Plasma Tube

RF Generator





Linear Amplifier

|

1
                                               Load Coil
                                     Plasma Gas Inlet
       Figure 3-  Diagram of Drake Radio-Transmitter
                  Radio-Frequency Generator  System.
Linear Amplifier
IIX
_L J
T ^
V "
^^B
_L
Load Coil and
Plasma Gas
System
Figure 4.  Diagram of Impedance Matching Network  for Drake
           Radio-Transmitter Radio-Frequency Generator  System.

-------
     To obtain efficient and maximum transfer of power from the
RF generator to the load coil-plasma system, an impedance matching
network is necessary.  A workable network was designed and con-
structed as shown in Figure 4.

     The output impedance of the ETO linear amplifier is approxi-
mately 50 ohms, but the input impedance of the load coil-plasma
gas system is variable and depends on a number of operational
parameters.  Most important of these are the ease and degree of
ionization of the plasma gas.

     The load coil-plasma gas system can be considered to be a
transformer with a multiturn primary winding (the excitation
coil) and a single turn secondary (the plasma gas).  Before ini-
tiating the plasma, the gas is not ionized and exhibits a high
impedance which is considerably greater than 50 ohms.  However,
immediately after the highly ionized plasma is formed, the
impedance of the load is drastically reduced to well below
50 ohms.  An efficient, adjustable, matching network is required
to follow the change in impedance to maintain the plasma.

     To form a plasma, argon gas was passed up the 20 mm O.D.
quartz tube tangentially.  A tangential flow (vs. a laminar flow)
pattern is advantageous because the cool, nonionized plasma gas
passing rapidly near the inner tube wall prevents overheating of
the quartz tube by the plasma.  The formation of a toroidal-shaped
plasma is also promoted with a tangential flow.  Tangential flow
is obtained by admitting the plasma gas through a specially
machined holder at the base of the plasma tube.  Argon was ad-
mitted at a flow rate of about 20 to 30 liters a minute via a
pressure tank, regulator, and rotameter.  The load coil  (copper
refrigeration tubing) was cooled by pumping distilled water con-
tinuously through an external circulation system (1/2 gal/min).

     Load coils were wound using 1/8", 3/16", and 1/V refrigera-
tion tubing.  There were about 1-1/2 turns to as many as 10 turns
with the spacing between turns varied from almost touching to
about 1/V.  The inner diameters of the coils were 20 mm, which
barely touched the quartz tube, to over 25 mm.

     The plasma tube was quartz, 20 mm O.D. with a 1 mm wall
thickness.  Most tubes were about ten inches in length, but their
length was not critical.  Tube holders were made of Plexiglas and
aluminum.

          5.2.1.3  Monochromator and Emission Detection System

               5.2.1.3-1  Jarrell-Ash 0.25-Meter Monochromator -
A 0.25-meter Jarrell-Ash Model 82-410 monochromator was used as
the wavelength separation device for radiation emitted from the
RF or a.c. arc excited materials.  It is a low cost, small size,
light instrument suitable for portable or field use.  The speci-
fications are as follows:


                              55

-------
     Focal Length - 0.25 meter Ebert design

     Linear Dispersion - 3.3 nm/mm with 1180 grooves/mm grating
                         1.65 nm/mm with 2360 grooves/mm grating
     Aperture Ratio (speed)  - f/3.5

     Gratings (two supplied) - Ruled area:   64 mm x 64 mm
                               Replicas:   1180 grooves/mm
                                          2360 grooves/mm
     Gratings blazed at - 300 nm and 600 nm

     Resolution (half-band   - Better than 0.3 nm in First  Order
       width at 313.1 nm)      with 2360 grooves/mm grating with
                               150y slits

     Slits - Two 150 micrometer slits,  fixed

     Wavelength Coverage - 200 to 900 nm

               5.2.1.3.2  Emission Detection System - Two RCA
1P28A photomultiplier  (PM) tubes were housed in Pacific Photo-
metrics Model 50F PM tube housings.  One PM tube served as the
sample tube and the other was the reference tube, as briefly
described under the design concept in Section 1 of Appendix VII.
A Bertran Model 602-11N power supply supplied the high voltage
to the PM tube dynode chain.

     An MRC designed and fabricated electrometer amplifier was
used to monitor the PM tube currents.  The amplifier was a dif-
ferential, dual channel design arranged to accept two PM tube
inputs, one due to a sample and one to a reference, and subtract
the reference from the sample before display on a meter or op-
tional strip chart recorder.

     The sample channel incorporated a switch selectable range
of input currents of 10~9 to 10~s amps full scale in decade steps,
Intermediate steps XI, X2, and X5 were also provided for each
decade range.  The reference channel was hard wired to 10~6 amps
full scale, and a ten-turn potentiometer at its output allowed
subtraction of any current from 0 to 10~6 amps from the sample
channel.  A manual zero control (10-turn potentiometer) allowed
adjustment of the electrometer circuitry to zero output with no
current inputs.

     In addition to real time spectral measurement capabilities,
the electrometer had a built-in integrator allowing accumulation
of PM tube currents over a manually controlled time period.

     An auto zero circuit allowed zeroing the electrometer output
in the spectral mode when observing only the background of an RF
plasma or a.c. arc discharge with both the sample and reference
PM tubes.  With no material of interest (Be or Cd) undergoing
emission, both PM tube channels should be equal and the elec-
trometer output should be zero.  When operating in the integral


                              56

-------
mode, the auto zero was set first in the spectral mode.   Then
an "integral zero" switch position allowed manual resetting of
the integrator.

     The time constant of the electrometer could be chosen and
an appropriate low leakage capacitor hard-wired into the final
amplifier stage of the electrometer.  A front-panel meter indi-
cator served to show the electrometer output, and a back-panel
recorder jack was provided to allow for permanent record keeping
using a 10-mV strip chart recorder.

     5.2.2  Results and Discussion

          5.2.2.1  Plasma Formation With Lepel R.F. Generator
                   System

     Argon, nitrogen, air, and a gas mixture (2.1% S02,  15.2$
C02, 3.1? 02, 79% Na) were used as coolant or plasma gas with
the single or concentric tube arrangement.  However, in  tubes
smaller than 28 mm in diameter, a plasma could be formed only
with argon.  The sample carrier gas was argon, nitrogen, or
air.  Oxygen was also used with success as a plasma gas  in a
single tube, and a plasma was formed using a heated injection
probe and an argon-hydrogen mixture.

     The plasma was easily initiated and sustained on pure argon
gas, but it was impossible to initiate a plasma on any other gas
or gas mixture.  After a number of attempts, a 100? nitrogen,
air, or artificial stack gas plasma could be sustained in a
single 28 mm quartz plasma tube by gradually adding these gases
to the pure argon plasma and tuning the Lepel generator  as the
ratio of argon in the plasma decreased.  Once the proper load
coil, excitation frequency, and tuning procedure was familiar,
it became an easy and rapid procedure to start an argon  plasma
and switch over to the desired plasma gas.

     A well-defined toroidal plasma could be obtained with argon,
but when other gases were added the plasma became elongated,
smaller in diameter, and more bullet-shaped.  When using air, a
long tail flame extending to the open end of the plasma  tube as
well as a "preflame" extending to the plasma tube holder formed.
A great deal of chemiluminescence was observed with the  hottest
area of the plasma being bluish-white and the preflame and tail
flame being yellowish.  With artificial stack gas, the chemi-
luminescence was again observed but there was more reddish color
apparent in the intermediate temperature zones between the bluish
and yellow zones.

     The changes in plasma shape with the introduction of air,
along with decreases in line-to-background ratio and line inten-
sities, were also observed by Truitt and Robinson (ref.  14) with
a radio-frequency (8 MHz) induced plasma at 4.0 kVA power.  Also,
                              57

-------
NOy-system bands appeared along with fairly strong, poorly devel-
oped 02 bands and moderately strong, well-developed N2 and NH
bands.  The emission intensity increased markedly with increasing
RF power from 1.2-4.0 kVA (ref.
     When operating an argon plasma, the RP power level neces-
sary to sustain the plasma was lower than with any other gas.
Also, the plasma was quieter acoustically with a lower background
intensity and there were less pronounced intensity fluctuations
with argon.

     With an air plasma, chemiluminescence was very evident.
Plasma flicker was more troublesome and the radio frequency
interference (RPI) and electromagnetic interference (EMI) radi-
ated by the load coil and picked up by the electronic measuring
circuitry was intolerable.  Apparently the greater Lepel genera-
tor RP power levels necessary to maintain an air plasma cause
excessive radiation of unwanted RF into the photomultiplier tube,
electrometer, and recorder electronic systems.  Figure 5 shows
an emission scan of an argon-air RF plasma from 3100 X to 3300 X
indicating the multitude of emission lines and noise in this
region of the spectrum.

     All grounded instrument power cords (115V, 3-wire) were iso-
lated.  The Lepel generator (requiring 230V, single phase, 3-wire
power) ground was left intact (a metal shielded power cable and
positive connection to utility ground) in order to comply with
FCC and manufacturers specifications.  The best grounding was
found by connecting the cases and shields of all instruments
(PMT high voltage power supply, monochromator-PMT housing, elec-
trometer, recorder) together via separate, heavy #12 gauge copper
wire and grounding them at one common point of utility ground.
The Lepel utility ground was not identical to the instrument
utility ground, since unsatisfactory operation resulted when the
Instruments were grounded to the Lepel generator case which was
in turn connected separately to 230V utility ground.

     The RFI and EMI radiating from the load coil (not being
absorbed to operate the plasma) were eliminated with a galva-
nized tin U-shaped shield placed around the coil.  The shield
was open at the top and bottom to admit and exhaust samples and
plasma gases and had a 1" diameter aperture in the side to allow
passage of emission light to the monochromator and photomulti-
plier tube.

          5.2.2.2  Plasma Formation With Drake Radio Transmitter
                   Generator System

     A good plasma was formed using the radio transmitter system
and argon gas.   However, careful tuning of the transmitter,
amplifier, and matching network with a fixed load coil, plasma
tube, and gas flow rate was absolutely necessary.
                              58

-------
vn
VD
                                     4~--^4	1" —-T    T  i     I
                                                   a
           Figure 5.  Scan of Emission Spectrum  (3100-3300  &)  From Argon-Air RF Plasma
                       Slits =  lOy
                       Scan Speed =  20 S/min.
                       Chart Speed =  l"/min.
Argon = 26 1/min.
Air =7.5 1/min.
1 x 10 5 amps full scale

-------
     Plasma operation could not be sustained in nitrogen gas.
Even a small percentage of nitrogen in the argon extinguished
the plasma.

     The RF generator system was not powerful enough in this
application to sustain anything except an argon plasma.  The
argon plasma was very difficult to initiate.  At one point,
voltages high enough to destroy a large variable air capacitor
in the output tuning network of the linear amplifier were real-
ized.  The matching network was perhaps less efficient with an
inductively coupled load vs. a capacitively coupled load used
successfully at low power levels by other investigators (ref.  8).
Since impedances changed so drastically from before to just after
plasma initiation, there was simply not enough reserve power to
sustain the plasma while readjusting the matching network from
the high impedance, "pre-plasma" settings to the low impedance
settings just after plasma formation.  Many arc-like, ionization
streamers appeared instead of a true, uniform plasma.  Because
of these operational difficulties, this approach was abandoned
at an early stage.

          5.2.2.3  Systems for Introducing Known Quantities of
                   Be or Cd Into RF Plasma

     Several different systems (fluidized bed, nebulizers, and
vapor generators) served to introduce known concentration of
cadmium, as aerosol or vapor, into the RP plasma.  Each system
had limitations.  The major problems were related to deposition
of particulate or condensed vapor on the walls and optics.  The
problems were greater when introducing air, than when using argon
alone, because of the different plasma shape (bullet, rather than
toroidal), length, temperature distribution down the pre-plasma
zone, RF coupling to metal probes, and higher pressures at the
base of the bullet-shaped plasma.  The most efficient method of
delivering the calibration specimen was by vaporizing cadmium
metal from a copper tipped probe which was maintained at 210-
220°C and which provided close delivery at a high injection
velocity.  Delivery of sample to the air or air-argon plasma
from other systems resulted in little penetration at the base
of the plasma and in a sample flow pattern around the periphery
of the plasma at the tube walls where the pressure was less and
where, unfortunately, the temperature was also lower.

     Typical problems and some solutions, are described in the
following sub-sections.  Cadmium was chosen for the preliminary
laboratory studies because it is less toxic than beryllium.

               5.2.2.3-1  Pneumatic Aerosol Generator - A pneu-
matic aerosol nebulizer (Bennett #2814 Twin Nebulizer, Bennett
Respiration Products, Inc., 1639 Eleventh Street, Santa Monica,
California)driven by argon gas was used with cadmium solutions
to produce a spray of known cadmium concentration.  Difficulties
in steady plasma operation were experienced when the aerosol


                              60

-------
entered the plasma.  The water droplets absorbed great amounts
of RF energy on vaporizing and disassociating in the plasma and
the greater pressures developed on heating the liquid (small
volume) droplets to disassociated atoms (large volume) caused
intolerable plasma flicker and unsteady plasma operation.

               5.2.2.3.2  Desolvatlng Apparatus - A desolvating
apparatus similar to that used earlier by Veillon (ref. 15) was
used; however, the larger droplets produced by the pneumatic
nebulizer (vs. the ultrasonic nebulizer used by Veillon) were
not as effectively desolvated.  A Beckman Type 2010 total con-
sumption type atomizer burner served as an aerosol generator,
but produced droplets larger than those of the Bennett nebulizer.
A large proportion of the solid aerosol produced by desolvating
the liquid droplets crystallized on the apparatus walls.

              5.2.2.3.3  Motor Driven Syringe - A motor driven
syringe (Model 335, Sage Instruments, Inc., White Plains, New
York) was used to inject known volumes and concentrations of
cadmium in aqueous solutions.  Difficulties were encountered
when attempts were made to get sufficiently close to the plasma
to inject the aqueous sample.  The solution vaporized unevenly
resulting in intermittent injection rates and ultimately the
stainless steel syringe needle fused from the heat of the plasma.


               5.2.2.3.4  Cadmium Metal Vaporizers - Since the
volatility of cadmium metal(and a number of cadmium and beryl-
lium salts) is rather high (cadmium b.p. 767°C and a vapor
pressure of approximately 2 x 10"1* mm @ l83°C), the possibility
of forming a metal vapor by passing a carrier gas (argon @
1 1/min.) over a heated cadmium metal was explored.  Several
configurations were tried, but the most efficient system was a
stainless steel delivery tube (1.5 mm I.D.) jacketed with con-
centric tubes of copper to permit handling of a heat exchanging
fluid.  With this probe system, the injector tip could be placed
directly into the base of the plasma, which enveloped the injector
tip.  A diagram of the probe is shown in Figure 6.

     The innermost tube which carries the sample gas is made of
3-mm stainless steel with a 1.5-nun I.D. central channel.  The
two other concentric tubes are 4.5 mm and 6 mm O.D. copper.
All tubes were silver soldered into an upper copper tip and a
copper block at the bottom for rigidity and getting coolant in
and out.  A container made of black iron pipe and fittings wrapped
with a Variac controlled heating tape and insulation was used to
generate cadmium metal vapor to inject into the plasma via the
injection probe.  Reagent grade cadmium metal filings served as
a source of cadmium vapor.

     By using glycerol (b.p. 290°C) as the heat exchanger fluid,
the probe could be maintained at 210-2iJO°C by circulating glycerol
from an external constant temperature bath.  With pot temperatures


                              61

-------
             Cadmium Vapor Out
                     t
                            —   Cooling Fluid In
                       - - | - -  Cooling Fluid Out
                      "-IT
                     I
              Cadmium Vapor In
Figure 6.  Diagram  of  Cadmium Metal Vaporizer Probe  for
           Injecting Cadmium Metal Vapor Into RP  Plasma,
                           62

-------
of 180-200°C, sufficient cadmium vapor can be obtained with no
deposition on the injector probe.

     A minor disadvantage with using glycerol was the difficult
start-up with room temperature, viscous glycerol.  The probe
passages were restricted to such an extent that viscous fluids
would pass only with difficulty, but once operating temperature
and initial circulation were achieved, the low viscosity of hot
glycerol presented no problems.

     The metal vapor injecting probe was used only with cadmium
metal; however, volatile beryllium or cadmium salts could just
as easily have been used.  By using controlled temperatures,
carefully regulated carrier gas flow rates, and careful choice
of volatile salts, beryllium or cadmium or both simultaneously
could be injected into the plasma in known amounts to calibrate
the instrument.

     Calibration of the quantity of cadmium delivered by the
probe was accomplished by atomic absorption spectrophotometric
analyses of cadmium collected in impingers over time periods
extending up to 16 hours.  At the flow rates of the plasma
gases, the cadmium generated in this system would produce con-
centrations in the plasma at the 10 yg Cd/m3 level.

     For long term operation, the trace quantities of oxygen in
the argon carrier gas should be removed to prevent oxidation of
the cadmium metal in the reservoir.  This is readily accomplished
by passing the carrier gas through freshly reduced copper turnings
maintained at 500°C.

          5.2.2.4  Detection of Cadmium With Radio-Frequency
                   Plasma Emission Spectroscopy

     Laboratory experiments using the cadmium vaporizer system
and the Lepel RF generator were conducted to determine the
feasibility of employing an RF induced plasma and emission
spectroscopy for monitoring cadmium continuously in a flowing
stream of air.  Initially, experiments were conducted with argon
and, finally, with argon-air mixtures.  Two types of plasmas
were evaluated:  (I)  induction coupled RF plasma formed with
a mixture of sample and sheath gases, and (II) induction coupled
RF plasma jet formed with sample gas' alone.  The former is pro-
duced at much higher RF power and flow rates than the latter,
but the latter is quieter and yields lower detection limits.

     All the major cadmium emission lines were observed including
2288 &, 2980 A, 3261 8, 3*»04 A, 3*»66 fl, and 3610 ft, and the
2288 A and 3261 S emission lines were found to be most sensitive
and approximately equal in intensity.
                              63

-------
               5.2.2.4.1  Cadmium Detection With Type I RF
Plasma - In an induction coupled RF plasma (Type I) with pure
argon (24 1/min), cadmium emission (3261 X) was detected photo-
electronically by measuring the emission signal directly.
Without signal integration, the photoelectric system produced
a detection limit better than 10 yg Cd/m3 of argon.  The 10 yg
Cd/m3 value was established by collecting the Cd for approxi-
mately 16 hours in impingers containing 10$ hydrochloric acid
and by performing analyses of the collected specimens with
atomic absorption spectrophotometry.

     The intensity of cadmium emission at this wavelength with
a nitrogen plasma was reduced by a factor of 5 when compared to
the emission intensity in the argon plasma.  Approximately the
same reduction in emission was observed when an oxygen plasma
was used.

     After introducing a small quantity of air (<0.5 1/min.)
into the plasma gas (total flow rate of 20 1/min.), no emission
spectrum for cadmium, introduced either using the water solution
nebulizer (set to deliver up to 0.01 g Cd/m3 of air) or as cad-
mium metal vapor (up to 0.6 g Cd/m3 of air) generated by heating
cadmium metal to its melting point (321°C), was detected photo-
electronically at 3261 X or during a scan from 2000 X to 4000 8.
The effect was also observed visually by dramatic changes in
color of the plasma from a definite blue color with a pure argon
matrix to a complete loss of the blue color with the addition
of approximately 235 air.

     Finally, several grams of cadmium oxide powder were placed
in the bottom of the plasma tube holder and blown up into the
plasma.   With an argon plasma, an intense blue color was seen
visually, but when air was added, the blue color disappeared
immediately and almost completely.  Only a very slight hint of
blue color was evident by visual examination.  A similar experi-
ment was performed with cobalt oxide and iron oxide powders and
the same result was observed visually.

     With the air addition, (a) the plasma becomes elongated,
(b) flame flutter occurs, (c) optical noise increases,
(d) turbulence is greater, (e) toroidal shape is lost,
(f) background emission is enhanced, and (g) no cadmium emis-
sion lines are observed.  In addition, a wavelength scan of
the elongated plasma at various positions in the preflame,
central flame, and tail-flame sections of the plasma yielded
neither cadmium emission lines, nor distinct line or band
spectra correlatable with CdO molecular spectral emission.

     Movement of the sample injection tube to various positions
in the plasma, i.e., from plasma base center to a position
nearer the cooler wall, but still at plasma base, or from the
plasma base and preflame zone to the tail-flame, did not help.

-------
               5.2.2.4.2  Cadmium Detection With Type II RF
Plasma - A small plasma jet flame (Type II RF Plasma) can be
produced at the tip of the sample Injection probe, when the
probe is positioned in the induction coil zone and the only
source of argon is the small argon-sample flow (2 1/min.)
through the injection probe.  The induction coil, quartz tube,
and sample injection have the same configuration as during
normal plasma operation, but the matrix gas (argon @ 20 1/min.)
is not present.

     This phenomenon has not been reported with an inductively
coupled plasma, but it is similar to that reported earlier (ref.
8) using capacitively coupled plasma probes.  The jet was about
2 or 3 mm in diameter, which was approximately the size of the
internal diameter of the central stainless steel injection probe
tube, and about 20-25 mm tall.  It appeared to be the same color
as a typical argon plasma.

     The plasma jet is produced with pure argon or with up to
105? hydrogen in argon.  However, it was not generated with nitro-
gen or air mixed with argon.   Very little RP power was necessary
to maintain the plasma jet.  It operates silently and does not
flicker, but RP interference and electromagnetic interference
are proportionally greater in the electronic measurement system
considering the very low power input to maintain the jet.

     The best limit of detection of cadmium (0.4 yg/m3) (without
signal integration) in an RF plasma was obtained by using a pure
argon plasma jet and by measuring the intensity of the cadmium
emission line at 3261 A.

     5.2.3  Conclusions - Radio-Frequency Excited Optical
            Emission Spectroscopy

     This study showed that although radio-frequency induced
plasma temperatures as high as 20,000°K can be attained at
atmospheric pressure, the detection of the emission spectra
of trace metals is extremely difficult when using air or air-
enriched argon as the matrix gas.  The plasma temperatures are
sufficiently high to vaporize most refractory compounds and the
excitation energies (up to 25 electron volts) provide the neces-
sary electronic excitation to produce excellent emission spectra
with argon as the matrix.  But with air in the plasma matrix,
the plasma shape becomes distorted, increased background emission
spectra are obtained, and the line intensities of the element
being determined are diminished.  As a result, direct analytical
determinations of cadmium in air matrices were not possible even
at levels in excess of 1000 yg/m3.

     Although several possible solutions to the air effect problem
with RF induced plasma can be suggested, considerable development
work would be required to establish feasibility.  Also, the cost
                               65

-------
and time factors, and the dilution effects for the actual moni-
toring measurements may become excessive.   Possible approaches
include removal of oxygen or reaction of oxygen to form relatively
stable oxide (CO, C02, H20, or metal oxide) and incorporation of
additives to inhibit nitrogen-oxygen free  radical reactions and
metal-oxygen compound formation, or to change the ionization
potential in selected plasma zones.

     The most promising additive approach would be to incorporate
a free radical scavenger to inhibit the nitrogen-oxygen reactions
which produce the background spectral emission due to chemilumi-
nescence.  However, so little work has been done to date, that
the potential degree of success cannot be estimated.  Also,
although the chemiluminescence may be eliminated, the other air
effects may prevent attainment of sufficient line to background
signal ratios.  Complete removal of oxygen is the only assured
successful solution to the problem of the air effects but also
is the most impractical considering the large volumes (2-40 1/min)
of sample gas that must be handled.

     After reviewing the results obtained with the radio-frequency
induced plasma and after discussion with Drs. V. A. Fassel (Iowa
State University), T. B. Reed (MIT), and R. Mavrodineanu (NBS),
we have concluded that a radio-frequency Induced total air plasma
or an argon plasma with moderate percentage of air which will
provide for detecting trace quantities of cadmium in air may be
attainable eventually, but not at the present state-of-the-art.
By using auxiliary magnetic fields to restrict the plasma size
and to control its shape a more concentrated source of optical
emission spectra may be obtained; unwanted background emission
spectra may be controlled by employing selected additives.  In
addition, use of metal oxide molecular spectra or spectral lines
characteristic of more energetic transitions, which would be
affected to a lesser degree by background interferences, may be
developed.  The discrimination of band and line spectra from the
background may be further enhanced by modulating the analytical
signal and using a synchronous detector, or lock-in amplifier,
to extract the weak analytical signals from intense spectral
background interference or from optical or electronic noise.

     Modulation can be accomplished by several techniques, in-
cluding modulation of:  (a) sample introduction, (b) wavelength,
and (c) two excitation sources, with and without analyte.

     Signal modulation of sample introduction is potentially
useful but less practical than other techniques.  By rapidly
admitting sample, then completely shutting it off (or by rapidly
alternating between sample and blank), one can synchronously
detect the difference between the emission due to the sample
and that when no sample is present.  The major problem is to
design a system with valves and other hardware to alternately,
rapidly start and stop sample introduction to the analytical
excitation device.


                              66

-------
     Sample background corrections made by wavelength modulation
techniques are most useful when the analytical line is buried in
a large amount of background or when the analytical line falls
on the shoulder of a large peak of another species present.  A
rotated or wobbled quartz plate placed just in front of the exit
slit of the monochromator refracts the light falling on the exit
slit.  The result is a rapid back and forth scan over a narrow
band of the spectrum centered on the wavelength of the line of
interest.  Rains (ref. 1?) has shown this gives the first deriva-
tive of the normal output of wavelength vs. time of a scanning
spectrometer.

     A mechanical mirror-chopper could be used to direct the
emission from two identical excitation sources, which excite
emission from stack gas and particulate and from stack gas alone.
A filter system would have to be used to remove particulate from
the reference stream, and obviously, the system would only be
used in cases where the analyte is nonvolatile or can be chemi-
cally or physically removed.  A lock-in amplifier would be used
to demodulate the resultant signal.

     The choice of modulating technique would depend on the
nature of the interference.  Wavelength modulation would be
the most useful for minimizing spectral background interferences.


5.3  Arc Excited Optical Emission Spectroscopy

     The most successful means of determining beryllium continu-
ously in a flowing stream of air (40 1/min.) from a beryllia
manufacturing facility was accomplished by Webb, et al. (Appendix
II, Section 4.1.1).  By using an intermittent DC arc, 1 yg Be/m3
was detected with sampling and analysis times of 20-60 seconds.
No significant particle size effects were observed for particles
of up to 20 microns for beryllia and up to 85 microns for the
metal.  The particulate was primarily beryllium containing mate-
rial and thus the total particulate loading in the sample stream
was low.  The common elements normally found in atmospheric dust,
namely calcium, aluminum, silicon, iron and sodium do not affect
the reading of the monitor even when they are present as a
visible cloud.

     Because of the failure to detect cadmium with radio-
frequency excited optical emission spectroscopy, a small program
was devoted to determine the feasibility of using a Webb-type
system for detecting beryllium in power plant emissions.

     5.3.1  Instrumentation

     Except for the arc discharge cell and the arc power supply,
the equipment was the same as that used with the RF excited
optical emission equipment.  However, in some tests, a second
separate monochromator was used in the reference system to aid
in identifying the background emission problems.

                              67

-------
          5.3.1.1  Arc Discharge Cell

     The discharge cell employed in the Webb-Harwell monitor
 (Appendix II, Section 4.1.1) was obtained on loan and a duplicate
was fabricated by Monsanto Research Corporation.  The original
Webb cell was machined from Sindanyo, a high density asbestos
material, and the MRC cell was fabricated from Electrobestos,
an asbestos material used in high voltage applications.  A dia-
gram of the discharge cell is shown in Figure 7.  The critical
feature of the cell design is the flow pattern to the arc dis-
charge zone.

     Several types of electrodes were tested.  Copper wire
 (l^B&S gauge) conducted the most current without heating, but
the wear rate was excessive.  Platinum electrodes could not
carry as much current before heating, and erosion was fairly
rapid, requiring one or more adjustments per day.  Platinum-
iridium electrodes were better since they could be used as much
as eight hours at low power levels before requiring adjustment.

          5.3.1.2  Arc Discharge Power Supply

     Because of time limitations, the fabrication of an inter-
mittent DC arc power supply described by Webb could not be
completed.  Tests were performed with an AC arc system based
on a luminous transformer (neon sign power supply) modified
with a 0.01 microfarad (50 KV) capacitor across the transformer
secondary to transfer maximum energy to the arc.  With 115 VAC
supplied to the transformer primary, the secondary supplies
alternating current (AC) at 15,000 volts and 60 milliamperes
 (MA), or 900 VA.  This system differs from that proposed by
Webb in that it is high voltage, low current AC arc rather
than low voltage, high current triggered direct current (DC)
arc.

          5.3.1.3  Calibration Systems

     To generate a beryllium or cadmium aerosol, electrodes
incorporating these metals were operated in the Webb cell.  With
the Webb cell placed before and in series with the MRC cell, air
drawn through both cells carried metal aerosol into the MRC cell
for excitation.  A diagram of the system is shown in Appendix VII,
Section 2.6.2.  Cadmium aerosol was generated by reagent cadmium
metal formed into a suitable electrode or by a 14-gauge silver
soldering alloy rod containing 18% cadmium, 26% copper, 35?
silver, and 21% zinc, and beryllium aerosol was generated by a
3% beryllium-975& copper alloy in 3/32" diameter or 3/16" diameter
rod form.

     In addition, the Bennett nebulizer discussed with the RP
plasma studies was used to form beryllium and cadmium aerosols
from aqueous solutions.  The heated injector probe from the RF
plasma work was found to inject insufficient cadmium into the
arc to permit detection.

                              68

-------
             -Quartz Window
                                Machined
                             Sindanyo Block
                               Electrodes
             Glass Inspection
                Window
Cone to Take
Filter Paper
Figure 7.  Diagram of Webb-Type Arc Discharge  Chamber.
                          69

-------
     5.3.2  Laboratory Studies

     Limitations of time precluded an in-depth evaluation of the
Webb type approach.  A laboratory feasibility study was conducted
to establish whether the modified Webb system could be used for
field applicationsj e.g., analyses of power plant emissions.

     Contrary to results with the RF plasma, water injected into
the arc did not interfere with its operation, and the resulting
spectrum had a low background with no major interferences.  In
the high voltage, low current, alternating current arc, beryllium
was more sensitive to analysis by emission than was cadmium.
The limit of detection for beryllium by direct spectral intensity
measurement(no integration) was the same as Webb had measured,
1 yg/m3, but that for cadmium was several times higher, >8o yg/m3.

     Figure 8 shows calibration of the Webb-type arc cell with
nebulized 1000 ppm aqueous solution of Be, as Be(NOa)2.  Figure 9
is a calibration curve for the data derived from Figure 8.  The
response is neither a linear, nor a logarithmic function of con-
centration.  These deviations indicate the possibility of quenching
or poor excitation effects and suggest the need for additional
power or an auxiliary heating arc system.  Better improvement may
be evident with the Webb intermittent DC arc, rather than the AC
arc as used in this study.

     Spectral scans of about 10 & to 20 & including all the
major beryllium and cadmium emission lines (M.I.T. Tables)
were made to determine the optimum emission wavelength.  The
wavelength positions of these have been confirmed with similar
scans using a Be or a Cd hollow cathode lamp source.  The best
beryllium emission line is the 3130-3131 A doublet.  The opti-
mum cadmium emission line is probably 2288 or 3261 X; however,
low sensitivities observed at all cadmium wavelengths prompted
devoting full efforts to the beryllium analysis.

     Figures 10, 11 and 12 show scans of the emission spectra
obtained with the Webb-type cell (2000 A to 4000 A) for lab air,
lab air plus 10,000 ppm Cd solution (nebulized into the air) and
lab air plus Be emitted from a second cell containing two 35? Be—
97J5 Cu electrodes.  Figures 9 and 10 are essentially the same,
indicating that no Cd emission is being detected at this concen-
tration (>80 yg/m3) with conditions optimized for beryllium.
Unfortunately, method development time was not available to
optimize the instrument conditions for cadmium.

     On Figure 12, the unresolved 3130-3131 & Be doublet appearing
as a single line is marked along with a pair of Cu lines.  Be
concentration is ^30 yg/m3.  A comparison of spectra of aqueous
solutions to spectra of laboratory air show no water interferences.
                               70

-------
                                                 263O-TMI-P

                               4 •       '2

                          Time, minutes
Figure 8.  Emission Intensity at 3131 A* for Webb-Type Arc Cell

           Excitation of various concentrations of Be con-
           taining aerosol in argon.  Arrows indicate flow
           rate changes.  [Aerosol produced by nebulizing
           aqueous Be(N03)2 solution.]
                               71

-------
   100


    90


    80



 1  7°
o

 §  60
 S
 ">  50
    40
 -  30
    20
    10
50
                              100
150
200
250
                             Concentration, jjg Be/m"
 Figure 9.  Calibration Curve - Be  (3131  S)  Emission From
             Webb Arc  Cell Using Nebulized 1000 ppm Aqueous
             Be(N03)z  Solution and Variable Carrier Gas
             Flow Rates.
                                 72

-------
unu g -  SufOBds
                               sduiv 9 OT  x
                        peadg UTBOS "
jo (  000t/-0002)
                                 Sujsn
                                     uofssfuig jo

-------
uiiu
                                     SI-BOS
                                            sduiv 9_OI x
                                          UBOS
            S9po.iq.oaig
pezfinqeu uoTq.nios  uidd OOO'OT  B iuoaj po  snjd JT-B q-ei  ST

                          (^ OOOty-QOO^) umaqoedg uoTSSTiug  jo u-gog
                                                                 "II

-------
- I	i	
	 •»	**	_^ . f» _ui. ;_ **=_^:^-^ •* --• - ?- ** - 'I ""^ •* •^•h^^ ft rrT^ t.-. n ;Q_ -^_ *> -' T-, i T *» '_ ~ " 3=_ « ~ |
                                        SV- o  '• ' -t o •-;—.: o
                                     -- ', ^T- g - Jr-T - Q	-f - O
   Figure  12.
Scan of Emission Spectrum (2000-4000  X)  From Webb-Type  Arc Cell.
[Sample is lab  air plus Be from a  set of 3% Be-97% Cu electrodes.]
                    25y Slits
                    50 S/Min.  Scan Speed
                    5 x 10 6 Amps  Full Scale
                                  Pt  Electrodes
                                  Variac  - 75 Volts
                                  Electrode Spacing -  2  mm

-------
     For preliminary laboratory studies, a second monochromator
and phototube system was used in conjunction with the dual
channel electrometer so that a background correction can be
obtained for fluctuations in the arc emission.  This presently
works best with the reference monochromator set on the back-
ground emission near the 3130-3131 X Be lines.

     Slit widths have been varied to determine the optimum
instrument response and this is about 5 micrometers with the
intense background observed.  With slits of 5 ym, 7.5 urn, and
10 ym, the signal to noise ratio around the 3130-3131 A Be
line degrades from about 10 to 6 to 2.5, respectively, with Be
concentrations of 9 yg/m3.

     Maximum wear occurs with large electrode gaps and/or high
input voltages from the Variac-controlled neon tube transformer.
Most experimentation has been with the electrode gap at about
3 to 6 mm and the Variac input voltage about 60 volts.

     An artificial stack gas containing carbon dioxide and sulfur
dioxide does not interfere with Be emission.  Water vapor from
droplets generated in a pneumatic nebulizer does not interfere,
and little or no OH emission occurs in the 3100 X to 3200 X
region of the spectrum with air, stack gas, or the nebulized
droplets.

     Fly ash was introduced with few adverse effects except that
in excessive amounts the electrodes seem to coat up and become
unsteady.  It was difficult to add the fly ash in an even, steady
flow.  No interferences were noted and only a small memory effect
is observed when viewing the Be 3130-3131 A line.

     A filter was employed immediately downstream from the ana-
lytical electrodes in the MRC cell, and samples collected and
analyzed by atomic absorption spectrophotometry were used to
confirm the aerosol concentrations.

     Since the sensitivity to beryllium was several times that
of cadmium and since Webb had analyzed beryllium with a similar
system, the decision was made to concentrate on the analysis of
beryllium alone.  The 3130.42-3131.07 A* doublet beryllium emis-
sion line was used for the analytical measurements.  An alternate
analytical Be emission line is 2348.61 X-  Although Webb et al.
(ref. 1,2) reported a positive magnesium interference at 3130 X,
in which 1 mg Mg/m3 gives a reading approximately equivalent to
1 yg Be/m3, this interference was not observed with the fly ash
matrix in this study.  It is notable that the nearest spectral
lines for Mg are 3104 A and 3229 X, which are well resolved from
the 3130 X Be lines with the present optical system.  The inter-
ference, if it exists, is not a direct spectral interference and
must arise from another mechanism.
                               76

-------
     Laboratory feasibility studies showed no air or water spec-
tral interferences as had been experienced with plasma operation,
but after continuous operation with aqueous aerosols, absorbed
water increased the conductivity of the asbestos and it became
more difficult to maintain a high power arc.   Electrode wear was
a problem, but this was partially solved by the use of more dur-
able platinum-iridium alloy, and there was no spectral interfer-
ence with the beryllium and cadmium.  Resistance heating of the
electrodes caused a diminished arc, but this  could be remedied
by better arc design or by using larger diameter electrodes.
                               77

-------
 6.   FIELD TEST OF PROTOTYPE

     Field studies with the arc-discharge emission spectroscopy
 system were conducted in a semi-permanent, air-conditioned shed
 with adequate power and stack sampling facilities on the roof
 of a power station.  Effluent gases from the newest and most
 modern boiler at the site were sampled.  Stack gas is delivered
 to the laboratory with a five-horsepower blower through a series
 of heated, insulated two-inch pipes.  The sample is drawn from
 a duct, which is located after the mechanical and electrical
 precipitators and before the I.D. fan.  The boiler tested sup-
 plies steam to a 1*10 megawatt generator.

     Figure 13 presents a schematic layout of the field test
 facility located on the roof of the power plant.  Figure 14
 shows the top front and side views of the delivery line and
 blower installation used to draw samples from ducting before
 the I.D. fans from either of two boilers.  The two sample lines
 joined before entering the high pressure blower, located atop
 the elevator shaft.  The blower was capable of delivering 100 cfm
 of 300°F gas at an increase of 22-inches W.G.  Since the sample
 points were at a negative 15-inch W.G., the effective pressure
 of the gas downstream of the blower was 7-inches W.G.  The 2-inch
 diameter Schedule 40 carbon steel delivery line carried the sample
 from the blower down to the power plant roof, about 32 feet below.
All sections of the delivery line and blower were heated and
 insulated.  The entire system was weatherproofed.

     A 10-ft x 14-ft prefabricated steel shed is installed on
 the roof.  Figure 15 is a picture of the power plant test facility
 showing the blower, delivery line and the exterior of the shed
with a 23,000 Btu air conditioner installed in one of the doors.

     Numerous difficulties were encountered immediately after
adding stack gas containing particulate to the cell.  The emis-
 sion intensity from the arc decreased and the arc glow was
observed to change from brilliant white light to a duller red
glow.  The red glow was not observed in the laboratory studies
with artificial stack gas or with fly ash when tested separately.
 It may occur only with both present.  A scan of the emission
 spectrum of the stack gas showed some emission bands not observed
 in the laboratory studies, but no identifications were made.

     Particulate matter in the stack gas built up rapidly on
 inner walls of the cell and the electrode holder assemblies.
Examination of the cooled electrodes showed the formation of
five to ten small glasslike fused beads on both electrodes
within a few minutes of sampling.  The AC arc does not pro-
vide sufficient energy to vaporize the fly ash.

     The emission spectrum obtained for the beryllium aerosol
generated with the Be-Cu electrodes showed an unresolved doublet
                              78

-------
                                                            {^ J-	Blower
                                                            CD
—q
vc
                                                  o OS ample  o o Ports -*o o
                  Figure  13.   Stack Gas Delivery  System and Sampling Manifold.

-------
CO
o
                           'OMQ
                                                          SCE BETAIL-S-

                                                /9-0
                  FLEXIBLE PIPE

                   GATE
                                                                                 DC TAIL "6-
                                                                                                  PROBE
                                                                                                  DUCT WITH
                                                                                                   OPENING
                                                                                                             C
                      EXI5T  1*" I BEAM
                                  "X.
                        EXIST. V STM. LINE
                   NOTE: EACH PIPE SECT
                        CONTAINS 8OOWATT
                        HTflTING WIRE
                         S  IN ALL
                                          ..r-1
                                          T-JI
        -*•"*
           *—"£T?Y,,"U  "f
                      =^HH
                                                                   o-t
                                             SEE  DETAIL
                                             C.S. .KH. 4O PIPE
                                                                    -• |
                                                                              FROM flJOVE WITH 'UNISTRUT-
                                                ".lEADEE" MEATINS WIBE
                                                 2- PIPE
                                                 FIBST LAYER IMS.
                                                 SECOND LAYER IMS.
                                                 WEATHER STRIPPING

                                                 T.C. WIRE
                                                                                 DETAIL "A"
                                      >M
—r~
-+
                                                                                                    CORPI>lt.vrilI!S
                                                                                   ^CALE   NTS
                                                                                                          ^-30
                                                   DWG
        Figure  14.   Main Delivery  Line and Blower Locations for  Power  Plant Emission Study.

-------
Figure 15-  Power Plant Test Facility,
                  81

-------
(3130.42 and 3131.0? &) easily differentiated from the back-
ground emission or from nearby emission lines.  A similar scan
at the same wavelength with a beryllium hollow cathode lamp
source showed a spectrum similar to that observed with beryllium
aerosol in the arc.  With clean analytical electrodes (Pt or
Pt-Ir) in dry air, no emission lines are observed at the beryllium
wavelengths.

     When first admitting stack gas to the analytical cell oper-
ating with freshly cleaned Pt or Pt-Ir electrodes, a resolved
doublet similar to that described above for beryllium appeared
in the scan of the emission spectrum.  However, during the
repeated scans in approximately five or six minutes, the intensity
of the doublet diminished and finally the doublet disappeared
from the spectrum.  During this time, particulate in the stack
gas coated the clean electrodes, lowered the overall intensity
of the arc (by visual check and by sound intensity of arc), and
completely masked or destroyed any beryllium optical emission
from the metal assumed to be present.

     The monochromator was set to the beryllium doublet emission
lines with a Be hollow cathode lamp or by arcing the beryllium
aerosol.  When stack gas was then admitted, a rapid decrease in
intensity resulted.  After attaining steady operation, a periodic
fluctuation in emission at the beryllium line was observed.  The
increase in intensity coincided with increased particulate concen-
tration in the stack gas.  The plant blower seems to produce a
cyclic emission of particulate.  The cycle is approximately
40 seconds.  This phenomenon was observed several times on dif-
ferent days of operation and a representative response pattern
using a direct reading mode (non-integration) is shown in
Figure 16.

     A second reference monochromator and photomultiplier tube
system was used to monitor fluctuations in the background signal
near the beryllium doublet emission lines.  By electronically
subtracting out any background fluctuations it was hoped that
an overall response indicative of the actual amount of beryllium
present could be obtained.

     The reference signal, however, was observed to fluctuate
simultaneously with and at nearly the same intensity variation
as the signal from the beryllium emission wavelength.  No im-
provement was seen by adjusting the reference monochromator to
a non-beryllium emission line in the stack gas spectrum.  Both
iron, which was present in the particulate, and platinum, the
electrode material, lines were used without success.  As a con-
sequence, the measurements taken at 3131 A could not be correlated
with the Be content in the stack gas.  This background phenomenon
may be similar to the Mg interference reported by Webb et al
(ref. 1,2).
                              82

-------
                                                                     -I .6
Wavelength changed to 3125 A
    J	I
                                                       Stack Gas On
                                                                          GO
                                                         o
                                                         'x
                                                                       .3
                                                                Arc On
                                                                       .2
                                                         Arc Off
_L     J        I                 II
  24       21
 IS
15       12
    Time, minutes
           Figure 16.   Typical  Pattern  Webb-Type  Cell
                         Emission at Be  3131 & Line.
                                 83

-------
     The two cells used in this study each had a built-in quartz
window for monitoring spectral emission photoelectronically.
The high energy arc ejected solid particulate (or electrode mate-
rial if only air was being investigated) outwardly with velocity
sufficient to impact firmly on the quartz windows.  Operation
for several hours left the windows completely opaque.

     The problem was solved by sealing the monochromator end of
the tube with a large quartz disc (held about 4 inches from the
arc) and by keeping the tube at positive pressure by pumping in
air at the rate of about 0.5 liters per minute.  This served to
dilute the stack gas somewhat (2%), but kept moisture out of the
optical emission measurement system and particulate from the
quartz windows.

     Asbestos cannot be machined with an airtight fit, and with
each disassembly for cleaning the airtight seal degrades.  With
the suction pump being used to pull stack gas through the MRC
asbestos cell, some leakage of air and displacement of sample
gas can occur.  This was not a serious problem, however.

     A great many difficulties were experienced with the asbestos
cells.  Asbestos is a satisfactory insulator when kept dry, but
if wet its conductivity makes it entirely unsuitable for a cell
material.  A symptom which appeared regularly was high voltage
arcing on the outside surface of the cell as well as on some
internal surfaces.  Also, the cell became physically hot through
the effects of resistive heating.  Purposely heating the cell
may, however, prevent moisture absorption.

     Particulate matter in the stack gas built up on the elec-
trode holder surfaces and on the inner cell walls.  It is likely
that with extended operation the particulate matter would accumu-
late to such a degree that large amounts would fall randomly
into the arc area.

     Filter samples of the particulate were collected from the
sample gas flow both after the cell (Samples #1-5) and before
the cell just after the valve at the point where the sampling
system Joined the main line (Samples #8-13).  The results are
shown in Table XVI in terms of total grams particulate per cubic
meter, yg of Be per cubic meter and part per million of Be in
the particulate.  The Be analysis was performed by atomic absorp-
tion using a Varian carbon rod atomizer.  Samples 1-4 and 8-12
were sampled at the flow rate normally employed during the tests,
28 1/min.  Samples 5 and 13 were sampled at 9-3 1/min and
4.7 1/min, respectively.  In addition, three fritted bubblers
each containing 40? HC1-8? H2SOi, were installed after the filters
on #5 and #13.  Analysis by atomic absorption showed no detection
of Be in any of the bubblers
                              84

-------
                   Table XVI


ANALYSIS BY ATOMIC ABSORPTION SPECTROPHOTOMETRY
      OP PARTICULATE SAMPLES COLLECTED AT
POWER PLANT FOR BERYLLIUM AND TOTAL PARTICULATE
Sample
1
2
3
4
5
8
9
10
11
12
13
Volume Concentration
Sampled of Particulate
(m3) (g/m3)
0.246
0.338
0.285
0.303
0.283
0.315
0.290
0.280
0.283
0.252
0.312
0.508
0.489
0.396
0.410
0.305
0.896
0.655
0.536
0.666
0.609
0.581
Concentration
of Be
(yg/m3)
20.6
18.2
16.7
10.4
6.4
45.5
28.8
17.5
22.6
34.8
18.5
Concentration
of Be in
Particulate
(ppm)
40.5
37.2
41.9
25.4
20.8
50.7
44.0
32.6
33.9
57.1
31.8
                      85

-------
     As noted in Table XVI, a high level of particulate was
being emitted from the stack.  As a consequence, the deposition
of fly ash contaminated the discharge cell, electrode mountings,
etc., forcing a termination of the program after two weeks of
intermittent operation at the power plant.  Lack of time and
funds precluded additional field testing.


6.1  Conclusions

     At the present state-of-the-art, there is no analytical
technique available for universal application as a direct,
continuous, monitor of beryllium and cadmium from all types of
stationary source emissions at the A.C.G.I.H. 1967-TLV levels
of 0.002 mg Be/m3 and 0.02 mg Cd/m3 (as soluble salts and metal
dust) or 0.1 mg Cd/m3 (as CdO fume).  A continuous monitoring
system for beryllium in work environments was developed by Webb,
et al. (ref. 1,2) based on the principle of measuring the optical
emission spectrum for beryllium generated in a pulsed DC arc.
The technique was used in several atomic energy projects for
monitoring beryllium levels in work environments and in produc-
tion facilities.

     A detection level was reported by Webb et al. as 1 yg Be/m3
and was substantiated in a modified (AC arc, rather than pulsed
DC arc) in this program.  No data were reported by Webb et al.
for cadmium, but Webb indicated in a private communication that
cadmium measurements should not be a problem.  However, limited
laboratory studies with cadmium aerosols and with the AC arc
modification in this program show much poorer sensitivities,
which were insufficient to provide detection limits equivalent
to the TLV levels.  Unfortuantely, no detailed study was made
and data was accumulated only with the instrument conditions
used for detecting beryllium.  Optimization of conditions for
detecting cadmium may provide the needed sensitivity.  Undoubt-
edly, instrument conditions must be optimized for each element.

     Although limited laboratory testing showed the feasibility
of using the Webb type system for direct, continuous monitoring
of beryllium, considerable difficulty was encountered in field
tests (power plant) with a prototype, continuous monitor for
measuring beryllium directly in stack effluent.  However, the
Webb arc emission spectroscopic system still remains the most
promising approach to monitoring trace metal in work environments
and in effluent from stationary sources.  The advantages are:
(a) direct, real time, continuous measurements down to low trace
levels (1 yg/Be/m3); (b) specificity; (c) moderate volume flow
rate (up to 40 1/min); and (d) relatively simple low weight
instrument package.  Lower detection limits «1 yg Be/m3 can
be obtained by integrating the emission signal for 30-60 seconds,
resulting in sampling and analysis times of approximately
30-90 seconds.  For beryllium, the analytical working range by
                              86

-------
direct measurement without signal integration is 1 yg - 200 pg
Be/m3.  Problems to be solved include:  (a) fouling of the elec-
trodes and cell with particulate deposits when monitoring stack
effluents having high loadings of refractory (silicate) particu-
late, and (b) absorption of moisture and other stack gases by
the asbestos cell resulting in a breakdown of electrical
insulation.

     With minor design changes, the Webb type cell system could
be used to monitor beryllium directly from emission sources
having relatively low loadings of particulate.  Applications
include beryllium machining operations and beryllium ore pro-
cessing operations.

     For stationary emission sources, yielding relatively high
loadings of refractory silicate particulate, a much hotter
vaporization system other than the Webb cell design is needed.
In the Webb cell, the relatively small arc zone serves to vaporize
the particulate and to induce optical emission spectra but does
not have sufficient energy to accomplish both simultaneously and
efficiently.  Solutions to this problem would involve either the
addition of a vaporizing system immediately prior to the analyti-
cal arc discharge or the use of a hotter analytical discharge
system.  In the former case, an auxiliary arc could be used to
vaporize the particulate prior to the analytical arc.  With the
latter approach, a hotter excitation system - radio-frequency
induced plasma or Shumaker arc oven - would be used.

     The problem of providing enough energy to vaporize sample
and to excite analyte to optical emission is the major defi-
ciency when applying the Webb arc system to source emissions
containing relatively high loadings of particulate.  Dr. R.
Mavrodineanu (ref. 18) suggested that an arc source developed
by Shumaker (ref. 19) for high temperature (in excess of
133000°K) gas studies may be applied to solve the problem.

     The Shumaker arc chamber is a tube-shaped device with quartz
windows in both ends.  Hollow graphite electrodes are placed near
the ends where argon gas is added to keep them in an inert atmos-
phere and to keep the quartz windows clean.  Sample gas enters
through the middle of the side of the tube, and it is exhausted
via ports one-fourth of the total length from either end.  The
arc length is 10 cm and 3/16-in. diameter and is viewed through
the hollow graphite electrodes.  The device is water cooled.

     In operation, a very high current DC arc is struck between
the electrodes.  Argon gas enters near the ends and sheaths the
electrodes in an inert gas atmosphere to prevent their erosion.
Sample enters the oven and is exhausted along with the argon
sheath gas.  Spectroscopic emission passing through the quartz
windows is monitored with the usual instrumentation.
                               87

-------
     Advantages of the Shumaker arc chamber are its durable
electrodes requiring infrequent adjustment, its extremely high
excitation energies, and its clean-window, low maintenance de-
sign.  Disadvantages are the high power (25 kw) necessary for
operation and the need to supply an inert gas (argon) during
operation.  Also, sample flow rates and potential deposition
problem need to be evaluated.  The device is available commer-
cially (TAFA Division, Humphreys Corporation, New Hampshire).
                               88

-------
7.    REFERENCES
 1.  Webb, R. J., M.S.W. Webb,  and P.  C.  Wildy,  "A Monitor for
     the Quantitative Determination of Beryllium in the Atmo-
     shpere," United Kingdom Atomic Energy Authority,  AERE-R
     2868 (1959).

 2.  Webb, M.S.W., R. J. Webb and P. C.  Wildy,  "Monitor for the
     Quantitative Determination of Beryllium in  the Atmosphere,"
     Journal of Sci. Instr.  37., 466-471  (I960).

 3.  Churchill, W. L. and A.H.C.P. Gillieson, "A Direct Spectro-
     graphic Method for the  Monitoring of Air for Minute Amounts
     of Beryllium and Beryllium Compounds," Spectrochimica Acta,
     1952, 5, 238-250.

 4.  Gillieson, A.H.C.P., "Monitoring for Toxic  Dusts  and Fumes,"
     Atomics 6_, 15-18 (1955).

 5.  Fromm,  D. and A. v. Oer, "Nachweis  von Beryllium  und
     emissionsspektrographischer Nachweis von Quecksilberdampf
     in Luft," Mikrochem. Acta I960, 235-244.

 6.  Aughey, H., "A Rapid Mobile Analyzer for Minute Amounts of
     Lead in Air," J. Opt. Soc. Am. 39.,  292-293  (1949).

 7.  Koppius, 0. B., "Detection of Lead  in Air With the Aid of a
     Geiger-Muller Counter," J. Opt. Soc. Am. 39., 294-297 (1949).

 8.  Mavrodineanu, R. and R. C. Hughes,  "Excitation in Radio-
     Frequency Discharges,"  Spectrochim.  Acta 19, 1309 (1962).


 9.  Greenfield, S., C. T. Berry, and L.  G. Bunch, "High-Pressure
     Plasmas as Spectroscopic Emission Sources," Analyst 89,
     713 (1964).

10.  West, C. D. and D. N. Hume, "Radiofrequency Plasma Emission
     Spectrophotometer," Anal.  Chem. 3j6_(2), 412  (1964).

11.  Dickinson, G. W., "Application of the Induction Coupled
     Plasma to Analytical Spectroscopy,"  Ph.D. Thesis, Iowa
     State University, 1969.

12.  Dickinson, G. W. and V. A. Fassel,  "Emission Spectrometric
     Determination of the Elements at the Nanogram per Milliliter
     Level Using Induction-Coupled Plasma Excitation," Anal.
     Chem. 4JL, 1021 (1969).
                               89

-------
13.  Dagnall, R. M., D. J. Smith, R.  S.  West, and S.  Greenfield,
     "Emission Spectroscopy of Trace  Impurities in Powdered
     Samples With a High-Frequency Argon Plasma Torch," Anal.
     Chim. Acta 54., 397 (1971).

14.  Truitt, D., and J. W. Robinson,  "Spectroscopic Studies of
     Radio-Frequency Induced Plasma,  Part 1.   Development and
     Characterization of Equipment,"  Anal.  Chim.  Acta 49,
     401-415 (1970).

15-  Veillon, C. and M. Margoshes, "A Pneumatic Solution Nebu-
     lization System Producing Dry Aerosol  for Spectroscopy,"
     Spectrochim. Acta 23B, 553 (1968).

16.  Fuchs, N. A., The Mechanics of Aerosols, The Macmillan
     Co., New York, N. Y.  (1964).

17.  Rains, T. C., "Flame  Emission Spectrometry With  Repetitive
     Optical Scanning in the Derivative  Mode," Anal.  Chem.  42,
     394 (1970).

18.  Mavrodineanu, R., private communication.

19-  Shumaker, J. B., "Arc Source for High  Temperature Gas
     Studies," Rev. Sci. Instr. 32_, 65-67 (1961).
                              90

-------
               APPENDIX I
SAMPLING SYSTEM CONSIDERATIONS AND DESIGN
                   91

-------
1.   INTRODUCTION

     In the beryllium-cadmium monitoring system, as in any source
emissions monitoring system, sampling must be accomplished in a
manner to assure that a representative sample is delivered to the
detector.  Design of the sample probe and sample transport sub-
systems must take into account the multiphase properties of the
emissions from a number of major industrial sources of beryllium
and cadmium.
2.   TRANSPORT OF PARTICULATE THROUGH A SAMPLING TRAIN

     Previous experience in sampling and collection of particu-
late from stationary sources has centered on the use of filter/
liquid impinger trains using the EPA sampling system.  Concern
exists regarding the efficiency of the collection systems when
collecting even relatively nonvolatile particulate.  The high
throughput of stack gases, the presence of moisture and reactive
gases (02, HC1, NO , SO , etc.), and the long sampling times
and lengthy exposure of high surface-to-volume ratio particles
may adversely affect the volatilization of substances contained
on the filter which are normally considered nonvolatile.
     During field sampling of emissions from a beryllium machining
plant using a typical EPA sampling train, a relatively uniform
distribution of beryllium throughout the sampling train was ob-
served (ref. 1).  Little particulate was obtained on the filter
with a majority found in either the sampling probe or in the
impingers.  Since the distribution of beryllium in the impingers
did not decrease progressively when a number of impingers were
arranged in sequence, slippage of beryllium was apparent.  No
specific reason can be given for this behavior although slippage
of submicron particles, formation of relatively volatile organo-
metallic compounds, and volatilization of hydrated BeO may be
contributing factors.  Deposition in the probe may result from
electrostatic effects of the probe walls on the very small
particles.

     Although little data are available that show transport of
BeO or other metal oxides at low to moderate temperature (i.e.,
<200°C), volatilization of B203 (ref. 2), BeO (ref. 3,4) and
Cr203 (ref. 5) has been reported at temperatures in the range
of 1000-111000C in the presence of water and in an oxidizing
atmosphere.  The mechanism for the vaporization promoting effect
of water has not been completely established; however, in the
case of Be and B the formation of volatile hydrates has been
proposed (ref. 2,4).  In the case of Cr203, moisture may change
the activation energy and serve as a promoter for the surface
oxidation of Cr203 to the more volatile CrOs (ref. 5).
                              92

-------
     Losses of the more volatile mercuric compounds can occur as
vaporization of the specific compound or through a disproportion-
ation reaction in which the very volatile elemental mercury is
formed.

     Volatilization of lead monoxide occurs at approximately
530°C and some indications of vaporization have been reported
at room temperature (ref. 6).  The vaporization of lead chloride
occurs at 225°C in a stream of HC1 and the volatilization of
lead sulfide vaporization by currents of combustion gases in
metallurgical furnaces is reported (ref. 5).

     Losses of elemental nickel can occur as volatile nickel
carbonyl.  Nickel carbonyl can be formed in significant amounts
(1 ppm to >1000 ppm) at 25°C, when the concentration of carbon
monoxide ranges from 100 ppm to 3 Mole % in presence of finely
divided nickel (ref. 7).  Gases containing some carbon monoxide
were found to contain nickel carbonyl after passing through
nickel-alloyed pipes (ref. 8).

     Information on the transport mechanisms of metallic elements
and their compounds is lacking, particularly with respect to
collecting submicron particulates containing microgram quantities
of the elements sought.  The problem can be compounded by the
need for processing large volumes of stack gas.

     Since beryllium emissions can be transported throughout an
EPA-approved sampling train, the use of filters and impinger
systems  is not desirable for use in the Be/Cd monitoring systems.
As stated earlier, a sampling and transport system delivering
vapor and aerosol directly to the detector was selected for this
study.


3.   SAMPLE INTAKE PHENOMENA

     To  ensure that the sample drawn through the sample tube
Is representative of the aerosol in the main gas stream, the
following conditions must be met:

     (1)  The tube must be aligned in a direction parallel
          to flow with mouth facing upstream, especially when
          particles >2 ym are present.

     (2)  The velocity of gas in the sample tube must equal
          the velocity of the main gas stream (isokinetic).

     (3)  The walls of the sample probe must be very thin
          (unobstructed flow).

     Departures from these criteria result in nonrepresentative
sampling.  The most difficult criterion to meet is isokinetic
sampling, and has been the subject of numerous technical papers.


                              93

-------
The Environmental Protection Agency (EPA) requires that all
particulate sampling systems used for emission controls be iso-
kinetic.  Under certain conditions, however, representative
sampling can be achieved without strict adherence to isokinetic
conditions.  When the size of the particles becomes exceedingly
small, the aerosol behaves as an ideal gas, and representative
samples can be obtained without Isokinetic conditions.  Figure 1
(ref. 9,10) shows the sampling efficiency obtained for various
particle sizes with a sample probe aligned parallel to main
stream flow.  It is apparent that the efficiency of collection
is nearly unity for all sampling velocities for particles below
4 micron in diameter.  Figure 2 (ref. 10) shows the effect of
sampling angle upon particle sampling efficiency.  The effi-
ciencies of collection approach unity for sampling angles <^5°.
For particle sizes <4y collected at a sampling angle 
-------
vo
\J\
               1.6
                1.4
                1.2
                1.0
                0.8
                0.6

                        Limit for very large particles
                 Alignment parallel
                 with the windstream
h-    /
   ,x
   1
                          \   \
1
J	I
I	L
                    0.5
1.0                1.5
Windspeed/Inlet Air Speed,  U0/U
                                  2.0
                    Figure 1.   Sampling Efficiency and the  Ratio of Flow Rates
                                Inside  and Outside a Sampling  Tube.

-------
CTN
                     0
  30        60        90
Angle Between Inlet Tube and Wind Direction, 6, degrees
                   Figure 2.  Sampling Efficiency and the Angle  Between the Axis
                              of the  Sampling Tube and the Flow  Direction.

-------
            =   (2y/l-y2/3 + arcsin
                                                             (1)
where
and
          y =
                    =  a dimensionless parameter,
         L  = tube length
         R  = tube radius
         V_s = particle velocity
         U  = mean horizontal velocity

     Figure 3 is a plot of particle collection efficiency in a
horizontal tube with laminar flow as a function of the dimension-
less parameter y.

     Early plans on the project included an rf excitation source
for the Be/Cd monitor at a sample flow rate of about 300 cc/min.
To determine the importance of deposition, the following assump-
tions have been made.

     (1)  Flow rate - 5 cc/sec

     (2)  Tube length in horizontal position - 100 cm

     (3)  Tube diameter - 2 cm

     (4)  Setting velocities are obtained from relaxation
          time parameters given in reference 12.
The value y can be calculated as follows:
         = 3TTgTRL
           8 Vol
                  ET = V
            wnere gi   vg,
                                         and
                                         ana
                                               _
                                             V(jl
                                                   !_
                                                   Ru
and     g = acceleration of gravity
        T = relaxation time
      Vol = volume flow rate
Particle
Diameter
  (y)

  20
  10
   4
   2
   1
   0.4
   0.2
   0.1
Relaxation Time
  Constant T
    (sec)

  1.23 x 10~3
  3.08 x 10~"
  5-03 x 10~s
  1.31 x 10 5
  3.54 x 10~6
  6.8? x 10~7
  2.28 x 10 7
  8.81 x 10~8
                                   Parameter
                                   In Eq.  (1)
                                      (y)

                                     28.4
                                      7-11
                                      1.16
                                      0.302
                                      0.082
                                      0.016
                                      0.005
                                      0.002
                                                    Collection
                                                   Efficiencies
                                                       (F)

                                                      1.00
                                                      1.00
                                                      1.00
                                                      0.44
                                                      0.15
                                                      0.03
                                                      0.01
                                                      0.004
                              97

-------
>«.
O
c
0>

'o
c
o
1.0



0.9



0.8



0.7



0.6
"C  0 5
QJ  U< "
O

o

CD

O
ro
Q_
0.4



0.3


0.2


0.1
            0.1  0.2   0.3  0.4   0.5   0.6   0.7   0.8  0.9   1.0

                    Dimensionless Parameter  j
       Figure  3.   Particle Collection  Efficiency in

                   Horizontal Tube with Laminar Flow.
                             98

-------
Thus for a 2-cm diameter tube having a 1.0 meter horizontal run
and delivering 5 cc/sec, all particles greater than 4.0 microns
will be deposited and half of the particles ^2 microns will be
deposited.

     Changing the the tube diameter from 2 cm to 2 mm has the
following effect on collection efficiency:
  Particle
  Diameter
    (y)

    20
    10
     4
     2
     1
     0.4
     0.2
     0.1
Relaxation Time
  Comstan TL
     (sec)
1
3
5
1
3
6
2
8
.23
.08
.03
.31
.54
.87
.28
.81
X
X
X
X
X
X
X
X
10
10
10
10
10
10
10
10
-3
-«f
-5
-6
-7
-7
-8
Parameter
in Eq. (1)
   (y)

   2.84
   0.771
   0.116
   0.030
   0.008
   0.002
  <0.001
  <0.001
Collection
Efficiency
   (F)

   1.00
   0.85
   0.18
   0.05
   0.016
   0.003
   0.001
  <0.001
Thus it appears that the delivery line should be kept as small
as possible to prevent particle deposition due to gravity.  With
a 2 mm tube diameter and a 300 cc/min. flow rate, the flow is
well into the laminar flow regime.

     These calculations were conducted assuming a flow rate of
300 cc/min to the detector.  At the higher flow rates actually
employed on the program, the degree of deposition would be even
smaller than that calculated above since the increased linear
velocity would decrease the value of the dimensionless param-
eter y.  In a 2-mm tube, however, a total volume flow rate in
excess of 4 liters/minute would result in turbulent flow condi-
tions which would not permit use of the laminar flow assumption.

4.2  Particle Loss by Diffusion to Walls

     When an aerosol is flowing in a tube in laminar flow,
particles can migrate to the wall by diffusion where they are
removed from the gas stream.  Gormley and Kennedy (ref. 13)
give a formula for calculating collection efficiencies in a
tube due to particle diffusion.
                   = 2.56y2/3 - 1.2y - 0.177y*/3
                                              (2)
where y is the dimensionless group DL/R2U, where D = diffusion
coefficient.

     Substituting Vol = irR2U into the expression for y, we
find that
                           y =
                TTDL
                Vol
                              99

-------
Values of D are given in reference 12.


                 Particle           Diffusion
                 Diameter          Coefficient
                   (y)              (cm2/sec)

                   20              1.38 x 10"8
                   10              2.38 x 10 8
                    4              6.10 x 10"8
                    2              1.27 x 10~7
                    1              2.7^ x 10~7
                    0.4            8.32 x 10~7
                    0.2            2.21 x 10~6


Taking 300 cc/min. as our flow rate, and 5 meters as our delivery
tube length, the following collection efficiencies will occur as
a result of particle diffusion.
          Particle
          Diameter       Dimensionless       Collection
            (y)            Parameters        Efficiency

            20            4.3 x 10~7              0
            10            7.5 x 10~6              0
             4            1.9 x 10~5              0
             2            4.0 x 10~s              0
             1            8.7 x 10~s              0
             0.4          2.6 x 10~"            M).01
             0.2          7-0 x 10'*            M).02


For our purposes, the loss of particulate matter by diffusion
is negligible.  This is true for any tube diameter, a somewhat
surprising relationship.  An increase in volume flow rate would
tend to minimize diffusional wall losses of particulate.

     A recent paper by Yamada and Charlson (ref. 1*1 ) discusses
the serious measurement errors that can result by diffusive loss
of gaseous pollutants to inlet line walls.  The fact that the
diffusion coefficients of particulate matter (l-10y) are a factor
of about 106 lower than gaseous species accounts for the negli-
gible loss of sampled particulate by this mechanism.
5.   RECOMMENDED SAMPLING SYSTEM

     Based on the considerations presented above, a sampling
system design approach was selected which included isokinetic
sampling of a primary sample stream at flow rates of about
                              100

-------
2 cu.ft/min, followed by secondary sampling of this primary
stream to Isolate a constant flow sample for delivery to the
analytical instrumentation.  Two sampling systems were designed
and/or used on the program.  The first system was designed and
packaged to supply a 300 cc/min flow to the rf induced emission
detection instrument.   The second system was that employed in
preliminary field testing of the AC discharge arc emission ap-
proach at a neighboring power plant.  These systems are described
in the following subsections.

5.1  RF Induced Emission Be/Cd Monitor Sampling System Design

     The design of the Be/Cd monitor sampling system was based'
on the following requirements:

     (1)  The Be/Cd excitation by r-f requires a constant flow
          rate of sample gas.

     (2)  The Be/Cd excitation chamber requires metered amounts
          of dilution gas and purge gas.

     (3)  The sampling system should minimize particle deposi-
          tion resulting from gravitation.

     To meet these requirements, the sampling system was designed
in two stages.  The stack effluent is first obtained via an
isokinetic sampler similar to that employed in Method 5.  This
effluent gas flow is then sampled in a second stage (2-mm diameter
tube) to extract a subsample at constant volumetric flow which
is delivered to the radiofrequency induced excitation chamber.
Some sacrifice in isokinetic conditions is inherent in the second-
stage sampling, but for power plant, incinerator and metallurgical
operations with efficient electrostatic precipitators or baghouse
control equipment, this may not be too serious a problem since
particle sizes may be largely below 4y.  As mentioned previously,
wet-scrubber controlled processes can result in metal emissions
associated with large (>30y) water droplets.  In this case it
may be feasible to sample at elevated temperature, thereby
vaporizing the droplets upstream of the secondary sampling stage.

     5.1.1  Isokinetic Sampling System

     Figure 4 shows a schematic representation of a typical iso-
kinetic sampling train.  It consists of a dual purpose sample
probe, a series of impingers, a gas pump, a wet test meter and
associated pressure gauges, thermometers, valves and tubing.
Figure 5 shows a schematic representation of the isokinetic
sampling control console.
                               101

-------
o
ro
                                                  PARTICULATE SAMPLING TRAIN
                                                                                                    THERMOMETER
                          Be-Cd SAMPLE CASE
                                                                            VACUUM GAUGE

                                                                                  \NEEOLE VALVE
                                                                            VACUUM
                                                                            TUBING
                                                                          (UMBILICAL)
TOGGLE-TYPE.
FAST-OPENING.       \     /
TO PREVENT CAVE-INS OF\	<1
FILTER, ELIMINATING   I     I
CONTAMINATION IN
THE IMPINGER IN CASE
THE FILTER BECOMES CLOGGED.
                                                                                                                  DUAL
                                                                                                                MANOMETER
                                                                                           CONTROL CONSOLE
                      Figure  4.   Schematic Representation of  Isokinetic  Sampling Train.

-------
                       Insulated Enclosure
2mm Sample
   Tube
  Solenoid
   Valves
  Discharge
    Tube

    Quartz
    Window
     Filter
                                                       0
       Main Gas
       Sampling
         Tube
  Temp.
  Power Supply to
    Heaters
 Pressure Gauges
                                                        Control Valves
                                                          Critical Flow
                                                           Orifices
 Air Filter

 Air
•-Gas Pump
               Figure 5.  Be/Cd Sample Compartment.
                                  103

-------
     5.1.2  Be/Cd Sampling Compartment

     The Be/Cd sampling compartment (omitted in Figure 4) is
shown in Figure 6.  The concept of this system is to sample a
sub-sample at constant volumetric flow rate from the isokinetic
sample delivery tube.  The sub-sample will, of necessity, be
taken non-isokinetically since a constant volume of about
300 cc/min. is required.  In a truly isokinetic sampling system,
the sampling rate varies as the velocity of gases in the stack
varies.  Furthermore, the extremely low sampling rate of
300 cc/min. would mean that the sampling tube diameter should be
about 0.5 mm to approach flow velocities normally obtained in
commercial stacks.

     Although the design does not include isokinetic sampling in
the Be/Cd sample tube, we believe the sample will still be repre-
sentative.  This belief is based upon the expectation that most
particulate discharges will have some .type of collection system,
thereby effectively removing all but the finer particles.  As
discussed previously, the adverse effects of non-isokinetic
sampling decrease with decreasing particle sizes.

     Once the sample enters the Be/Cd sampling tube, every pre-
caution is taken to prevent particle deposition.  Horizontal
runs are kept to a minimum, flow velocities are kept as high as
possible, all lines are maintained above the dew point tempera-
ture of the sample gas and obstructions to the flow are held to
a minimum.

     As shown in Figure 3, the flow rates of the dilution gas,
purge gas and total flow are measured by critical flow orifices
and controlled by fine-metering control valves.  The pressure
drop through the orifices are measured by sensitive pressure
gauges.

     A simple gas pump is used to deliver the sample gas, purge
gas and elution gas.  The gases will be discharged to a suitable
vent.

     The entire Be/Cd sampling system is housed in a temperature-
controlled heated enclosure.  A quartz window in the side of the
enclosure provides a means for observing the discharge glow.
Solenoid valves, operated by a single relay, admit either sample
gas or air to the excitation chamber.

5.2  Power Plant Facility for Testing of the AC Arc Discharge
     Emission Detector

     Late in this program preliminary field testing of the AC
arc discharge emission detection instrumentation was conducted
using a facility employed previously at a power plant (ref. 15).

-------
         UMBILICAL
        CONNECTION-
           VACUUM
UMBILICAL
CONNECTION—
PITOBE
  UMBILICAL
  ELECTRICAL
  CONNECTOR
PILOT LIGHTS


        OVEN

VACUUM PUMP

  PYROMETER

 ELECTRONIC
       TIMER

SELECTOR FOR
   PYROMETER
 TEMPERATURE
    INDICATOR

 PYROMETER
                    316 S.S.
                    CABINET
                  THROUGHOUT
 QUICK-RELEASE BYPASS
 FOR BLEED TO
 ATMOSPHERE
                   DUAL
                   INCLINE
                   MANOMETER
                                            DRY GAS
                                            METER
  VACUUM
  REGULATOR
  VALVE

              VACUUM PUMP
    MOUNTED INSIDE CABINET
    Figure 6.  Isokinetic Sampling Control Console.
                          105

-------
Figure 7 is a drawing of the stack gas delivery system and samp-
ling manifold employed in the field test facility.   Figure 8
shows the top front and side views of the delivery  line and
blower installation used to draw samples from ducting before
the I.D. fans from either of two boilers.  The two  sample lines
joined before entering the high pressure blower,  located atop
the elevator shaft.  The blower was capable of delivering
100 cfm of 300°F gas at an increase of 22-inches  W.G.  Since
the sample points were at a negative 15-inch W.G.,  the effective
pressure of the gas downstream of the blower was  7-inches W.G.
The 2-inch diameter Schedule 40 carbon steel delivery line carried
the sample from the blower down to the power plant  roof, about
32 feet below.  All sections of the delivery line and blower
were heated and insulated.  The entire system was weatherproofed.

     A 10-ft x 14-ft prefabricated steel shed is  installed on
the roof.  Figure 9 is a picture of the power plant test facility
showing the blower, delivery line and the exterior  of the shed
with a 23,000 Btu air conditioner installed in one  of the doors.
6.    REFERENCES

 1.  Burckle, J. 0., private communication, 1971.

 2.  Meschi, D. J., W. A. Chupka, and J. Berkowitz, J.  Chem.
     Phys. 33., 530 (I960).

 3.  Hutchison, C. A. and J. G. Malm, J. Am.  Chem.  Soc.  71,
     1338 (19*19).

 4.  Wartenberg, H. v., Z. anorg. Chem.  264,  226 (1951).

 5.  Caplan, D., and M. Cohen, J. Electrochem Soc.  438-442
     (1961).

 6.  Mellor, J. W., "A Comprehensive Treatise of Inorganic and
     Theoretical Chemistry," Vol. VII, Langmans, Green  &  Co.,
     New York, N. Y. (1927) .

 7.  Brief,  R. S., F. S. Venable, and R. S. Ajemian, "Nickel
     Carbonyl:  Its Detection and Potential for Formation,"
     Am. Ind. Hyg. Assoc. J. 2_6:72 (1965).

 8.  Sullivan, R. J., "Preliminary Air Pollution Survey of Nickel
     and Its Compounds," U.S. Dept. HEW, NAPCA, Raleigh,  N.C.,
     October 1969.

 9.  Fuchs,  N. A., The Mechanics of Aerosols, The  Macmillan
     Co., New York, N.Y., 1964.
                              106

-------
                                                                    Blower
H
o
                                                  o o-Sample 0 o  Ports -*o o
                    Figure 7.  Stack  Gas  Delivery System and  Sampling Manifold.

-------
o
00
                                                        .^-—SEE DETAIL'S*
                                                    HR
            — ^
                                13-0
                      EXIST l»" I BEAM
                                                 »8-0
                  FLEXIBLE PIPE

                    GATE VALVE
                                   -v
                                       *
                         E«IST. f STM. LINE
                   NOTE: EACH  PIPE  SECT.
                        CONTAINS  BOO WATT
                         HFflTING  WIRE
                         5 IN ALL
                                       vM
*-**
   ^ »MCT
                                                                   :: T
                                           «^—
                                              SEE DETAIL "A"
                                            2" C.S. .»CH. 4O PIPE
                                                     TO »«4^. SHE*
                      n-t,



                       \


                      B-t>


                       ^ 1
--(-
                                                                                   DCTAIL
                                                                                                     PROBE
                                                                                                     DUCT WITH
                                                                                                      OPENIN6
                                                                                                                Sc
                                                                                                                cz
                                 BLOWER - SUPTOBTED
                                 FROM *»VE WITH -UNISTRUT"
                                                                                                 -SEABED" HEATING  WIRE
                                                                                                  3- PIPE
                                                                                                  FIRST LAYER INS.
                                                                                                  SECOND LAXCR INS.
                                                                                                  WEATHER  STRIPPinO

                                                                                                  T.C. WIRE
                                                                                    DETAIL "A"
                                                                                               I) * ^ I UN 1 ^|*.

                                                                                                 LMYTUfX.
                                                                                     | SCALE  NTS
                                                                                                     EH SB  71-30
                                                                                                           G7QI
                                                     DWG
                                                     NO A-O//-OO7-APO-O
                    Figure 8.    Main Delivery  Line and  Blower  Locations  for  Power  Plant
                                   Stack  Emission Study.

-------
Figure 9.  Power Plant Test Facility
                 109

-------
10.  Watson, H., Amer.  Ind.  Hyg.  Ass.  Quart.  15,  21  (1954).

11.  Puchs, N.  A., The  Mechanics  of Aerosols,  The Macmillan
     Co., New York, N.Y.,  1964, p.  112.

12.  ibid., pp  71, 184.

13-  Gormley, P., and M.  Kennedy, Proc.  Roy.  Irish Acad.  52A,
     163 (19^9).

I1*.  Yamada, V. M., and P.  J.  Charlson,  Env.  Sci. Tech.  3_,
     483 (1969).

15.  Snyder, A. D., E.  C.  Eimutis,  M.  G.  Konicek, L.  Parts, and
     P.  L. Sherman, Volume  II, Final Report,  EPA  Contract
     EHSD 71-30, January 1972.
                              110

-------
                  APPENDIX II
              BRIEFING DOCUMENT -
ANALYTICAL TECHNIQUES FOR BERYLLIUM AND CADMIUM
                       111

-------
                        TABLE OF CONTENTS

Section                                                      page

   1      INTRODUCTION                                        HH

   2      SOURCES AND TYPE OP BERYLLIUM EMISSION              115

   3      SOURCES AND TYPE OF CADMIUM EMISSION                117

   4      CANDIDATE ANALYTICAL TECHNIQUES                     117

          4.1  Emission Spectrographic Methods                118

               4.1.1   Beryllium - Emission Spectrographic    124
                         Analysis With Arc  Discharge
               4.1.2   Beryllium - Emission Spectrographic    128
                         Analysis With Spark Discharge -
                         Solid Copper Electrodes
               4.1.3   Beryllium - Emission Spectrographic    131
                         Analysis With Spark Discharge -
                         Hollow Electrodes
               4.1.4   Beryllium - Emission Spectrographic    134
                         Analysis With Spark Discharge -
                         Filter Paper Tape
               4.1.5   Beryllium - Emission Spectrographic    138
                         Analyses With Spark Discharge
                         Directly on Filter Paper
               4.1.6   Beryllium - Emission Spectrographic    l4l
                         Analysis With A.C. Arc
               4.1.7   Cadmium, Chromium, Copper, Lead,  and   144
                         Nickel-Emission Spectrographic
                         Analysis With Spark Discharge
               4.1.8   Emission Spectrographic Analysis  for   148
                         Cadmium With a Microwave Induced
                         Argon Plasma
               4.1.9   Radio-Frequency Induced Plasma-        152
                         Spectroscopic Studies
               4.1.10  Emission Spectrographic Analysis  With  156
                         Low Wattage Microwave Induced Argon
                         plasmas at Atmospheric  Pressure
               4.1.11  Beryllium-RF Plasma  Torch - Emission   160
                         Spectroscopy
               4.1.12  Comparison of D.C. Arc Type Plasma     164
                         and High Frequency (RF) Induction
                         Type Plasma as Spectroscopic
                         Emission Source
                               112

-------
                   Table of Contents - Cont'd

Section                                                      Page

          4.2  Atomic Absorption, Thermal Emission, and       168
               Atomic Fluorescence Spectroscopy

               4.2.1   Cadmium Analysis - Atomic Absorption   170
                         Method for Cadmium Fumes

          4.3  X-ray Emission Spectroscopy                    173
          4.4  Activation Analysis                            176

               4.4.1   Automatic Beryllium in Air Monitor     177
                         Based on the 9Be(a,n,y)C12 Reaction
               4.4.2   Determination of Beryllium by the      180
                         Photoneutron Method With the
                         (y,n) Reaction
               4.4.3   Determination of Beryllium on Air      183
                         Filters with the (a,n) Reaction
               4.4.4   Beryllium - Determination by a Gamma   186
                         Activation-Photoneutron Method
               4.4.5   Beryllium - Continuous Monitor for     189
                         Air-borne Beryllium Based on the
                         9Be(a,n,y)12C Reaction

          4.5  Alpha Particle Scattering                      192

               4.5.1   Alpha-Scattering Technique of          192
                         Chemical Analysis
               4.5.2   Beryllium - Measurement by             196
                         Alpha-Particle Scattering

          4.6  Mass Spectrographic Analysis                   199

   5      RECOMMENDED ANALYTICAL TECHNIQUE FOR CONTINUOUS     200
          MONITORING OF BERYLLIUM AND CADMIUM

   6      REFERENCES                                          202
                               113

-------
              ANALYTICAL TECHNOLOGY FOR CONTINUOUS
            MONITORING OF STATIONARY SOURCE EMISSIONS
                    FOR BERYLLIUM AND CADMIUM
1.   INTRODUCTION

     The chemical and physical properties  of beryllium and
cadmium and their compounds, which influence the selection of
methods for quantitative analyses, differ  considerably between
each element.  Analytical techniques that  can be applied to one
element and its compounds cannot necessarily be applied to the
other.

     Fundamental differences in atomic structure and nature of
compound formation for each element restrict application of cer-
tain analytical techniques based on nuclear or atomic properties
to individual elements and their compounds and do not allow ap-
plication for analyses of both elements by the same technique.
Analytical techniques for trace analyses based on x-ray emission
of characteristic radiation for cadmium (atomic number 48) cannot
be applied to beryllium (atomic number 4).  Conversely, photo-
neutron excitation techniques which can be applied to beryllium
cannot be used with cadmium.

     Spectrographic methods have been devised for trace analyses
of both elements, beryllium and cadmium.  However, differences
in volatility and attendant response factors influence operating
conditions for attaining maximum sensitivities for each element.

     In selecting a continuous monitoring  device for both
beryllium and cadmium emissions from stationary sources, there
are two options.  Two analytical methods,  based on measuring
different properties and requiring different measuring systems
and optimized for maximum response can be  incorporated into one
monitor.  However, the more ideal alternative would be to select
one measuring system which is highly selective and responsive
to both elements.

     The following briefing document was prepared to show poten-
tial methods of analysis, some of the limitations of the tech-
niques, possible problems in applying the  procedures, and
recommendations for the most acceptable approaches.  Emphasis
was placed on five prime factors, analytical reliability (accuracy
and precision), selectivity, sensitivity,  a continuous operation
mode, and speed.  Other factors (personnel, apparatus, installa-
tion, and maintenance requirements and costs) although important,
can be controlled by proper hardware design and are emphasized to
a lesser degree.

-------
     Continuous monitoring can be defined in a very res'trictive
sense as being only real-time measurement with no sample pre-
treatment.  However, the present technology in abatement control
and detection systems may permit only an approximation of con-
tinuous sampling and measuring.

     Many manual techniques currently used for monitoring trace
metal emissions, while meeting criteria of analytical reliability,
selectivity and sensitivity, require moderate to lengthy time
periods for sample pretreatment and are not considered or evalu-
ated in this document.  An analytical method was not eliminated
simply based on need for preliminary pretreatment, but was
screened based on whether the treatment, including chemical
reactions, could be performed at the site, unattended, and
within a total measuring time of five to ten minutes.

     Since some of the chemical and physical properties of the
elements and their compounds will markedly influence the selec-
tion of the analytical technique, a brief commentary on selected
properties for beryllium or cadmium emissions anticipated from
known sources is included in a preface for each element.  Poten-
tial analytical approaches, based on the chemical or physical
properties are then presented.

     Wherever specific analytical approaches have been used to
monitor an element, a detailed outline of the principle, appli-
cability, and operational parameters is tabulated.  These
detailed outlines generally highlight the merits and limitations
of potential methods.

     In several cases, candidate analytical methods have not been
applied to the element of interest.  As a consequence, specific
information regarding operational parameters and response factors
are not available.  In the briefing document, the technique is
described by indicating the principles involved, potential
applicability, and anticipated problems.


2.   SOURCES AND TYPE OF BERYLLIUM EMISSION


     Natural,  production, and product sources of emission have
been reviewed by Durocher  (ref. 1).  The principal chemical
forms in these emissions are elemental beryllium, beryllium oxide
and mixed oxides with other elements, beryllium halides, beryl-
lium sulfate, ammonium beryllium fluoride, beryllium-copper
alloys and beryllium hydroxide.

     Beryllium and its compounds (ref. 1) are relatively invola-
tile and refractory.  The oxide exhibits different physical and
chemical properties, depending on the temperature at which it is
calcined.  High-fired (>1600°C) beryllium oxide (m.p. 2530°C)
                               115

-------
resists chemical attack and is extremely refractory.   The halides
(bromide - sublimes at ^73°C, chloride - b.p.  H88°C,  and iodide -
b.p. 488°C) are more volatile.  Beryllium nitrate (trihydrate) has
avery low boiling point of 142°C.  Beryllium metal has a boiling
point of 2970°C.

     Based on the information presented by Durocher (ref. 1), the
assumption can be made that the candidate monitoring  system must
be able to measure quantitatively,  beryllium in a variety of
chemically-bound forms which have a wide range of volatility
(b.p. <500°C to >3000°C) and chemical reactivity (ranging from
highly reactive to very inert).

     However, more descriptive of potential air concentrations
than these data are the theoretical air concentrations (Table I)
of beryllium based on vapor pressures of materials tabulated by
Durocher (ref. 1) from the original compilation by Tepper, et al.
(ref. 2).  Tepper (ref. 2) also points out that under conditions
in which air is circulated over a hot vaporizing surface, larger
concentrations of Be can be obtained in the air stream than those
indicated by vapor pressure data alone if condensation to fume
(sub-micron particulate) occurs.


                             Table I

           THEORETICAL AIR CONCENTRATIONS OF BERYLLIUM
              BASED ON VAPOR PRESSURES OF MATERIALS
      Compound
Beryllium
Beryllium oxide
Beryllium fluoride


Beryllium chloride,
   iodide, bromide

Beryllium borohydride
Temperature
   (°C)

     885
   1,060

   1,710
   1,990
     420
     525
     108
     158

      20
                                                 Beryllium
                                               Concentration
                                                  (yg/m3 )

                                                  2
                                                200

                                                  2
                                                200
                                                  2
                                                200
                                                  2
                                                200
                                                   . 35 x 106
     Although most emissions will contain beryllium as the rela-
tively nonvolatile oxide, hydroxide or metal particulate, certain
emission sources, particularly metallurgical reduction facilities,
                               116

-------
could yield the more volatile beryllium compounds in the vapor
phase.  The sampling and analytical systems used for monitoring
these emissions must be applicable to the analyses of beryllium
compounds in the form of dust, mists, and fumes.

     Data for coal-fired power plants (ref. 3,4) and incinerators
(ref. 5) show compositional analyses for particulate and flue
gases for beryllium in power plant and incinerator emissions.
Based on these data, concentrations could range from 10~'*g/m3
to 10~6g/m3.


3.   SOURCES AND TYPE OF CADMIUM EMISSION

     The principal source of cadmium is in the refining of other
metals, primarily zinc (ref. 6).  Little information is available
on the specific emissions of cadmium in these processes.  Based
on ratio analysis, cadmium sulfate, cadmium oxide, cadmium oxide-
iron oxide, and cadmium sulfide were reported in zinc refinery
dusts.

     Almost all of the oxide, sulfate, and chloride of cadmium
and most of the cadmium metal produced in the United States is
consumed by the electroplating industry.  Cadmium is alloyed
with nickel, silver-copper, silver, zinc-tin, and lead-tin, and
is used in paints (cadmium red).

     Cadmium oxide sublimes at 1770°K at atmospheric pressure and
evaporation occurs at 900° to 1000°C.  Cadmium sulfide has a vapor
pressure of 10"* mm Hg @ 540°C and 1 mm Hg @ 8?5°C.  The vapor
pressure of cadmium chloride is reported as 3 mm Hg @ 6l8°C and
316 mm @ 930°C.

     The cadmium emissions may occur principally as particulate,
but under certain circumstances the cadmium may be in the gas
phase.

     Prom refuse incinerator data (ref. 5), the rate of cadmium
emission is approximately 3.6 grams/hour in total dust loadings
of approximately 2.8 kilograms/hour.


4.   CANDIDATE ANALYTICAL TECHNIQUES

     A variety of analytical techniques, including colorimetry,
fluorimetry, emission spectroscopy, atomic absorption spectro-
photometry, gas chromatography, and electrometric measurements
have been used routinely in monitoring beryllium effluents and
atmospheric pollution.  Somewhat similar approaches, i.e.,
colorimetry, emission spectroscopy, and atomic absorption  spec-
trophotometry, plus polarography have been applied for measuring
cadmium in dusts and fumes.  These procedures employ standard


                                117

-------
analytical Instrumentation and require collection in an impinger,
electrostatic precipitator, or on a filter followed by low-
temperature ashing or acid digestion prior to the actual analyti-
cal measurement.  The techniques have proven useful for first
generation monitoring but require moderate to lengthy time periods
for sample pretreatment,  are prone to contamination from reagents
and the digestion processes, and are not usable for on-site,
continuous monitoring of  source emissions.

     The prime candidates for a continuous monitor of metal
emissions from stationary sources are based on radiant energy
emission or absorption, and measurement of nuclear properties
and radioactivity.  The techniques suitable for sensors for
automated analysis are:

         Emission spectroscopy
         Flame photometry
         Atomic absorption or fluorescence spectroscopy
         X-ray emission (fluorescence)
         Activation analysis
         Mass spectrography

     Sensitivity, speed,  and selectivity make the spectroscopic
techniques well suited for use in continuous monitorins systems
for trace metal analyses.  Several continuous monitoring systems
based on the spectrographic measurement of optical emission
characteristics of beryllium have been developed for measuring
the beryllium levels in production or manufacturing operations.

     In addition to the spectrographic methods, a photoneutron
technique involving the Be9 (a, n, y) C12 reaction has been
successfully applied for intermittent monitoring of beryllium
contamination in air.  The photoneutron technique permitted the
detection of 25 yg of Be/m3 in 30 seconds.

     These methods are described briefly in the follosing sub-
sections.

4.1  Emission Spectrographic Methods

     The major concern in applying emission spectrographic
techniques for continuous monitoring of beryllium involves the
efficiency of the excitation source to develop quantitative and
representative optical emission from beryllium particulate.
Spark and arc excitation have been used with relatively good
success and methods using these excitations are summarized in
this section.  There is some evidence for incomplete vaporization
and excitation of large particulate matter and for electrode
contamination.  It is apparent that a higher energy source,
preferably electrodeless, is desirable.
                               118

-------
     The most popular means of producing excitation for emission
spectroscopy are electrical sources (arcs and sparks).  High
excitation temperatures provide good sensitivity for most ele-
ments.  The most serious limitations, particularly with the use
of arc source, are electrical instability, fractional distil-
lation, and wandering of the arc resulting in fluctuations in
temperature at the excitation zone.  With spark excitation,
integration of several hundred sparks during an analysis produces
more reproducible results, but less sensitivity than the arc.

     By using time resolved spark spectra and signal integra-
tion, the lower detection levels can be reduced by minimizing
background.  Measurements can be started after the initial gas
spectra have faded and unwanted lines can be eliminated if they
have a higher excitation potential.  Hagenah (ref. 7) used a
two-channel (2 wavelength) recording system and integration
during the period between 29 and 44 ysec after start of each
spark to improve the determination of cadmium in zinc by a
factor of 4.

     Improvements in stability of the d.c. arc can be accom-
plished through the use of cathode excitation (ref. 8) and by
using a cathode cored with a material (BaC03 or Li2C03) of an
element with an ionization potential lower than carbon (ref. 9).
Cathode excitation offers advantages in sensitivity for some
non-refractory elements and gives greater precision for a number
of more volatile elements in a carbon matrix.  With the cored
electrode, arc wander and current and voltage fluctuations are
minimized, but no reduction of arc temperature occurs.  The
added element forms a stationary positive-ion cloud at the
cathode and acts as a ballast to electron flow.  In general,
materials with boiling points between 1000° and 3500°C show
improvement in burning characteristics.

     Improved vaporization and excitation can be attained by
using a dual source system.  A d.c. arc is used to heat and
vaporize the sample while an a.c. arc or a spark provides the
energy to more efficiently excite the optical emission spectra
(ref. 10).

     Fluorination reactions during arc discharge excitation have
been used to improve the vaporization of beryllium.  By intro-
ducing fluoro-carbon polymers (Fluoroplast) (ref. 11) or filter
paper impregnated with Li citrate plus HF (ref. 12) into the
arc, the efficiency of vaporizing beryllium was markedly in-
creased.

     By incorporating gas-sheathed (argon or helium) arc or
spark discharges, additional stabilization of the spectral
excitation and improvements in sensitivity can be attained.
                               119

-------
     In the d.c. arc plasma, the discharge may be considered as
two overlapping sources - a high-current arc at the cathode with
a temperature near 5000°K and a compressible plasma beam with a
temperature of 11,000-14,300°K.  Comparative studies of conven-
tional d.c.-arc and plasma-arc excitation show the plasma arc
to be twice as precise as the d.c. arc, but the d.c. arc is con-
siderably more sensitive than the plasma arc (ref. 13).  Matrix
effects were found to be less than other excitation methods;
the effects of hydrochloric, nitric, and sulphuric acids were
found to depend on the nature of the acid but not on the concen-
tration.  The dependency on the type of acid is less than other
techniques (ref. 14).

     By applying a high-frequency electric field for induction
heating of heat conducting gases, excitation of optical emission
spectra is obtained.  Detection limits are increased for many
elements.  Depressive interference is reduced.  The excitation
atmosphere can be neutral, reducing, or oxidizing, depending on
the gases used in the torch.  Background and banding problems
are also reduced (ref. 15).

     Mavrodineanu and Hughes (ref. 15) show spectral data obtained
in radiofrequency discharges for approximately 70 elements includ-
ing Be and Cd.  Additional spectra for carrier gases - air, N2 ,
Oa, C02, Ke, and H2 are also reported.  Excitation of the spectral
emission was accomplished at atmospheric pressure with radiofre-
quency discharges obtained with (a) 30 MHz and 250W, and (b) 2^50
MHz and 2 kW.  The flame-like discharges originate at a conical
tip at the end of a conductor forming a part of the main inductor
of the tank circuit with the 30 MHz system and at the tip of a
graphite electrode with the 2^50 MHz system.  Excellent spectral
data are displayed but limited quantitative data are presented.
Helium and hydrogen carrier gases are preferred because of low
background and non-oxidizing media.

     In a study of two types of plasma sources - d.c. arc type
and the high-frequency induction type  (36 MHz) - Greenfield et
al. (ref. 16) concluded:  The plasma source is far simpler to
operate than conventional arc and spark methods and gives the
high stability associated with the a.c. spark combined with the
sensitivity of the d.c. arc.  Particular advantages of the high
frequency plasma torch are the lack of electrodes, which gives
freedom from contamination and produces extremely low background.
Most important is the ability to inject powdered samples directly
into the plasma, over a period of time, which overcomes the
problem of sample inhomogeniety by permitting integration of
spectral emission and allows longer exposures than with usual
powder techniques, with consequent increase in sensitivity.
Also, the plasma source provides sufficient energy to overcome
depressed interference effects.caused by the formation of stable
compounds.
                               120

-------
     An extremely sensitive emission spectrography technique
with detection limits as low as 10 12 gram of metal has been
developed based on low wattage microwave induced plasmas (ref.
17).  These sensitivities are attainable based on introduction
of solid sample into an argon plasma at reduced pressure after
removal of solvent or diluent vapors.  The sensitivity is due
in part to relatively high electronic temperatures and low
spectral background, in addition to the lack of compound forma-
tion in the inert atmosphere.

     The major limitation of low wattage microwave excitation
is the problem of quenching the plasma when molecular vapors or
gases having high ionizational potentials are introduced into
the plasma.  Separation of the metal elements from the sample
matrix is required prior to introduction into the plasma.  Total
amounts of material in excess of 10 6 to 10 8 gram per second
are sufficient to extinguish an atmospheric pressure argon plasma
(argon flow rate 100 ml per minute) (ref. 17).

      Using a  laser  source,  samples can  be  analyzed  that  are
refractory, nonconductors,  or molten materials.   However, this
source  is primarily  a microsampling  system,  exciting  a  spot  of
5p  to lOOy in  diameter.  The beam  of light is  focused on a  spot
and concentrated  through a  microscope objective  onto  the sample.
Excitation of  the plume produced by  the  laser  vaporization  of
the  sample is  accomplished  by an auxiliary electrode  system
located between the  sample  and  the microscope  objective.

      In two articles, Rasberry  et  al.  (ref.  18,19)  covers the
characteristic of the source and an  investigation of  quantitative
aspects of laser-probe excitation  in spectrochemical  analysis.
Random  errors  come  largely  from variations in  laser energy  and
from photometric  errors.  Correlations  were  established  between
energy  of the  laser  beam, size  of  the pit  formed during  excita-
tion,  and  spectral  intensities.  The major problem  in applying
the  laser-probe to  quantitative analysis  is  the  lack  of  suitable
standards.

      Runge et  al.  (ref. 20)  described the  application of a  giant
pulse laser to quantitative  analysis of major  constituents  of
metals  and nonmetals by pure laser excitation.   Fairly  reproduc-
ible quantitative relationships were obtained  by superimposing
four laser-excited  exposures.   For stainless  steel  samples,  the
coefficients  of variation for Ni and Cr were  5-26%  and  3-82?,
respectively,  and were indicated as  being  on  a par  with  a.c.  and
d.c.  arc methods.

      Potentially, the laser excitation  approach  is  very  attrac-
tive,  but at  the  current state-of-the-art, this  approach is  not
practical  for  immediate application  for a  continuous, routine,
monitoring system.
                                121

-------
     Reliable comparisons of detection limits for various ana-
lytical methods are difficult, if not impossible, to make.  The
experimental data must be accumulated objectively by the same
investigator on the same type of samples without introducing
bias by employing preconcentration steps, integration or non-
integration of signals, desolvation, and other specialized
treatments on selected techniques.  There are no data which
compare response characteristics of the emission spectra gene-
rated from a variety of excitation sources for samples of air
particulate taken directly and continuously.

     Dickinson and Fassel (ref.  21) show a tabulation of detection
limits (Table II) produced by a number of excitation sources for
trace metals in aqueous solutions.  The data  summarize the values
obtained from a variety of literature sources and indicate the
sensitivities in concentration units of yg/ml.  Relative detec-
tion limits based on absolute values (e.g., yg of element) may
be quite different depending on the quantity  of solution required
in the procedure and the use of desolvation steps.  (Note -
Evaporation of solvent and the use of small volumes - 5 Ml -
in the microwave plasma or flameless atomic absorption systems
would permit detection of 10"11*  g of element).  The table is
presented to show a comparison of mass/volume sensitivities
for the excitation sources which potentially  could be used to
monitor air particulate emissions directly.

     Problems of lesser concern involve the potential inter-
ferences and the selection of the analytical  wavelengths.
Sufficient resolution is needed to separate the beryllium lines
at 3130.416 ft and 3131.074 & from the mercury line at 3131.833 A,
titanium line at 3130.800 A, the weak iron line at 3130.56? A, and
one line of the OH band, resulting from excitation of water vapor,
which could interfere with the lower wavelength beryllium line.
The beryllium line at 2348.6 A"  can also be used, but considera-
tion must be given to self-reversal phenomena.

     Technology on a variety of excitation techniques, including
the electrodeless system, is included in the  briefing document
to permit an understanding of the problems related to their use
and the possibility of applying the technique to a continuous
monitor.
                               122

-------
                                             Table II
H
ro
uo
Element

   Al
   As
   B
   Ba
   Cd
   Ce
   Co
   Cr
   Fe
   Hf
   La
   Ni
   P
   Pb
   Sb
   Sr
   Th
   V
   Zn
   Nb
   Ta
   Ti
                Wavelength
                   A**
                  3961
                  2288
                  2*197
                  2288
                  4186.6
                  3453-5
                  3578.7
                  3719.9
                  3399.8
4086
3524.5
2535.6
4057.8
2598
4077
4019
4379
                  2138.6
                  4058.9
                  3012.5
                  3349.4
                                 MEASURED DETECTION LIMITS, yg/ml
Radio-frequency
    Plasma*
   Emission	

    0.002
    0.1
    0.03
    0.0001
    0.03
    0.007
    0.003
    0.001
    0.005
    0.01
    0.003
    0.006
    0.2
    0.008
    0.2
    0.00002
    0.003
    0.006
    0.009
    0.01
    0.07
    0.003

Flame
Emission
0.01
6
0.3(BO)
0.001
2
10
0.05
0.005
0.05
75
O.l(LaO)
0.03
3(PO)
0.2
1.5
0.0002
150
0.01
50
0.6
18
0.2
Flame
Atomic
Absorption
0.02
0.2
3
0.05
0.001
95
0.005
0.002
0.002
15
2
0.005
—
0.004
0.2
0.01
—
0.02
0.0005
5
6
0.04
Stabilized
Arc or
Plasma Jet
0.1
—
0.05
0.3
0.4
_
3
0.05
0.14
—
-
1
1.1
4
-
0.07
_
0.2
0.3

_
—

Microwave
Plasma
0.03
4
0.03
1
0.5
20
—
_
0.5
—
-
0.3
100
1
0.6
_
4
0.1
0.1
_
_
_
       Concentration required to produce a line signal two times greater than background
        fluctuation.
      **Wavelengths used were not always those used for results shown in columns 4-7.

-------
     4.1.1  Beryllium - Emission Spectrographic Analysis
            With Arc Discharge

          4.1.1.1  Principle and Applicability

               4.1.1.1.1  Principle - Air is drawn through a
triggered intermittent arc (50 c/s) and the intensity of the
beryllium doublet at 3130 A" is monitored together with the ad-
jacent background and the ratio of these intensities is recorded
on a ratio recorder.

               4.1.1.1.2  Applicability - Mobile, fully automatic
prototype was tested for analyzing atmospheres in plants handling
beryllium.  Accuracy and sensitivity were not materially affected
by chemical form of the beryllium and are unaffected by sizes of
particles normally encountered in atmospheric dust .   Should be
applicable to monitoring all forms of Be emissions.

          4.1.1.2  Range and Sensitivity
          Range:  1 to 100 yg/m3

          Sensitivity:  Limit of Detection - 1 yg/m3

          4.1.1.3  Interferences

               4.1.1.3.1  Chemical - Unaffected by common ele-
ments normally found at high levels in atmospheric dust, namely
calcium, aluminum, silicon, iron and sodium.  Vanadium, titanium
and magnesium gave positive readings when present in compara-
tively high concentrations.  Magnesium has a large effect, giving
a reading equivalent to 1 yg Be/m3 when present at a concentra-
tion of 1 mg/m3 .  There is a direct spectral interference by
mercury line at 3131-5 A* but is not sensitive enough for mercury
vapor at room temperature to cause an interference.  Chemical
form of beryllium has no significant effect.  Beryllium metal,
high-fired beryllium, beryllium hydroxide, and beryllium ammonium
fluoride were studied.

     Care must be taken to use a monochromator with sufficient
dispersion (band pass of 0.1 ft) to resolve one of the beryllium
doublets at 3131.072 & and 3130.416 A from a titanium line at
3130.800 X.

               4.1.1.3.2  Physical - Independent of particle
size at least up to 85y which was limit of author's interest.

               4.1.1.3.3  Memory Effect - Negligible effect
observed up to 25 yg/m3 and 10% of previous determination at
very high concentrations (100 yg/m3).  Concentrations of 100
to 200 yg/m3 required three or four determinations to return
to unbiased readings.

-------
          4.1.1.4  Accuracy, Precision and Stability

     Insufficient data are reported to establish statistically
reliable accuracy and precision, but comparison of monitor field
trial data with conventional spectrophotometric analyses per-
formed on filter paper specimens taken at the same time as monitor
readings, shows good agreement.  An average of approximately 150
monitor readings shows a concentration of 2.6 yg Be/m3, whereas
10 "integrated" filter samples analyzed by a spectrophotometric
method indicate an average concentration of 2.5 yg Be/m3 for the
concentration range of 0.7 yg/m3 to 4.8 yg/m3 .   Monitor readings
were taken every minute, whereas filter paper specimens were
collected on 10 or 20-minute periods.  The more numerous monitor
readings show many one-minute periods where no Be was detected,
but also showed periods where concentrations reached 53 yg/m3.

          4.1.1.5  Apparatus

               4.1.1.5.1  Spectrographic source to produce 50 c/s
arc discharge with time constant such that the discharge between
the electrodes is approximately 25% of any one cycle.

               4.1.1.5.2  Electrodes - 1/4-in.  diameter copper
and tipped with platinum (negative electrode) and a piece of
14 s.w.g. platinum wire (positive electrode) arranged such that
air flow of sample is negative to positive.  (Note - Copper and
tungsten electrodes oxidized too quickly, resulting in high
instability. )

               4.1.1.5.3  Monochromator - Czerny-Turner with
30,000 lines/inch grating; 3rd order, gave reciprocal linear
dispersion of 5 S/min.j exit slit provided a 1 A band pass.

                          Detector - Two quartz-windowed E.M.I.
Ltd., photomultipliers .

               4.1.1.5.5  Ancillary Equipment - Vacuum pump
(used only to pull air sample at atmospheric pressure through
analyzer), flow controller, flow meter, filters, calibration
system, ratio recorder,  synchronous sequence timer and electronic
power supply.

          4.1.1.6  Time  Cycle for Sampling and Measuring

     At a sample flow rate of 40 liters/minute, the monitor was
set to complete one determination, based on a 20-second sampling
time, every minute.  Automatic programming provided a 25-minute
cycle involving five calibration runs (blanks and standard) and
20 sample determinations.
                               125

-------
          ij.1.1.7  Calibration Procedure

     Primary calibration and validation of the technique was
accomplished by comparing the monitor response data with analyses
on the arc chamber effluent obtained on filter paper specimens
with spectrophotometric techniques.   Aerosols for primary cali-
bration and for secondary calibration during automatic operation
were generated by passing filtered air through a spark discharge
between beryllium-copper electrodes.  The average amount of
beryllium evolved during the calibration period was 0.095 yg
(±0.01 yg) corresponding to a concentration of 7 yg Be/m3.
No drift in the amount of beryllium was noted in three months.

          4.1.1.8  Method of Sampling

     Air drawn directly into excitation chamber.

          4.1.1.9  Multi-element Application

     Could be applied to all metal elements.

          4.1.1.10  Size

     Sampling head containing arc chamber, trigger unit for arc,
monochromator and detectors measured 19-in. x 10-in. x 11-in.
high and was connected to main console containing power supplies,
vacuum pump and recorder.

          4.1.1.11  Unit Output

     Response converted to yg/m3 .

          4.1.1.12  Safety Hazards
     Moderately high electrical voltage for arc discharge.

          4.1.1.13  Maintenance

     Platinum electrodes should be adjusted or replaced every
eight hours .

          4.1.1.14  Recommendations for Method Improvement

     Use of radio-frequency induction-coupled plasma should be
considered to eliminate the need for the arc-excitation elec-
trodes and the attendant maintenance.

     Major restriction for arcs is the high erosion of permanent
electrodes.
                                126

-------
     For improved vaporization of refractory particulate or high
particulate loadings, consideration should be given to using an
auxiliary heating arc or to using a hotter arc chamber of the
Shumaker type (ref. 4-Section 4.1.1.15).

          4.1.1.15  References

1.   Webb, R. J., M.S.W. Webb, and P. C. Wildy, "A Monitor
     for the Quantitative Determination of Beryllium in the
     Atmosphere," United Kingdom Atomic Energy Authority,
     AERE-R 2868 (1959).

2.   Webb, M.S.W., R. J. Webb, and P. C. Wildy, "Monitor for the
     Quantitative Determination of Beryllium in the Atmosphere,"
     Journal of Sci. Instr. 37, pp. 466-471 (I960).

3.   Webb, M.S.W., R. J. Webb, and P. C. Wildy, "A Monitor
     for the Quantitative Determination of Beryllium in the
     Atmosphere," United Kingdom Atomic Energy Authority,
     AERE-R 3318.

4.   Shumaker, J. B., "Arc Source for High Temperature Gas
     Studies," Rev. Sci. Instr. 32., 65-67 (1961).
                               127

-------
     *1.1.2  Beryllium - Emission Spectrographic Analysis
            With Spark Discharge - Solid Copper Electrodes

          4.1.2.1  Principle and Applicability

               ty. 1.2.1.1  Principle - The contaminated air stream
is introduced by passing the air through the spark gap between
solid copper electrodes.  The intensity of beryllium spark lines
at 3130.416 ft and 3131.07*1 A" is recorded on a 24-hr circular chart
recorder and correlated with the concentration of beryllium in
the air.

               4.1.2.1.2  Applicability - A prototype of a mobile
monitor was tested briefly in a beryllia production facility and
showed moderate potential as a continuous monitoring device for
work area atmospheres.  Additional experimental data is needed
to establish application in sites containing mercury (spectral
interference) and applications to beryllium compounds other
than beryllia.  Most useful application would be as tolerance
meter to show presence to excess beryllium, e.g., >2 yg/m3.

     The instrument has been proved to respond semiquantitatively
to beryllium in the form of the oxide (as furnace fume and as
dust), of the fluoride and ammonium double fluoride and as beryl
ore.  It has been used to determine the hazard in filling mobile
containers with powdered ore, in operating a bricquetting press
for furnace charges for an ammonium beryllium fluoride dryer,
and for the furnaces for the decomposition of this fluoride
and for the final metal production.

          4.1.2.2  Range and Sensitivity

     Range at maximum sensitivity is 0.5 Mg Be/m3 to 20 yg Be/m3.
The limit of detection corresponds to 0.5 Mg Be/m3.

          4.1.2.3  Interference

               4.1.2.3.1  Chemical - Possible spectral interfer-
ence from mercury which has a fairly strong line at 3131.833 &.
Author discounts mercury interference by stating "....mercury
vapour in amounts sufficient to cause interferences are unlikely
in circumstances where beryllium was being worked with and even
if it did, would of itself present a serious toxic hazard."  A
potential spectral interference by a weak iron spark line at
3130.567 A* was shown not to exist.  No data are presented to
establish presence or lack of response variations for different
beryllium compounds.
                               128

-------
               ty.1.2.3.2  Physical - The limited experimental
data indicate that the spark source monitor records 25% lower
values of Be content for atmospheric concentrations >8 yg/m3
when compared to filter specimens measured spectrophotometri-
cally.

          *t. 1.2.4  Accuracy, Precision and Stability

     No data reported for accuracy and precision of measurement.
No information regarding instrument stability is given.

          4.1.2.5  Apparatus

               4.1.2.5.1  Spectrograph - Modified Littrow design
with grating having 14,400 lines/inch; dispersion in 1st order -
30 mm/ft.

               4.1.2.5.2  Excitation Source - Uncontrolled Hilger
spark source unit with 0.005 mfd capacity and 0.03 mH inductance;
a 1 KVA Variac auto-transformer was used to prevent high fre-
quency currents generated by spark discharge from producing
feed-back to detection system and amplifiers.

               4.1.2.5.3  Electrodes - 3-25 mm copper wire elec-
trodes of high grade electrolytical quality sleeved with quartz
tubing to within 1/8-in. of sparking surfaces.

               4.1.2.5.*!  Detection System - 1P28 photomultiplier
and Bristol Pyromaster Potentiometer Model 440M Circular Chart
Recorder (24-hr cycle) fitted with a planimeter coupled to a
cyclometer fixed behind the pen drive axle to provide mechanical
integration of area under the trace.

               ^.1.2.5.5  Ancillary Equipment - Vacuum pump,
filter, calibration system, electronic amplifiers, flow meter.

          4.1.2.6  Time Cycle for Sampling and Measuring

     Monitoring is continuous; no provision for automatic
calibration.

          4.1.2.7  Calibration Procedure

     Manual calibration was accomplished by generating a beryllia
aerosol (particles <25 microns diameter) by passing a flowing
stream of air over a bed of beryllia which is being agitated
mechanically.  The intensity of the beryllium spark line was
correlated with independent spectrophotometric analysis of par-
ticulate collected on filters.
                                129

-------
          4.1.2.8  Method of Sampling

     Air drawn directly into excitation chamber.

          4.1.2.9  Multi-element Application

     Could be applied to all metal elements.  Authors note par-
ticularly Si, Pb, Zr, V, Cr, Cd, Tl,  Ge ,  Sb , Al ,  B, and Mn in
addition to Be.

          ty. 1.2. 10  Physical Dimensions

     Mobile monitor measured 3-1/2 x  2-1/4 x 4-1/2 ft, weighed
200 Ibs and was mounted on a hospital cart .

          4.1.2.11  Unit Output
     Recorder tracing convertible by calibration standard to
jag/m3 ; planimeter-cyclometer measurement convertible to yg/hr.

          4.1.2.12  Safety Hazard

     Moderately high voltage to generate excitation discharge.

          4.1.2.13  Maintenance

     Due to high wear, electrodes should be adjusted or replaced
every 8 hours .

          4.1.2.14  Recommendation for Method Improvements

     Instrument is unsatisfactory for monitoring system as pre-
sently designed.  Operation as tolerance meter for levels
<8 yg/m3 is adequate.  However, monitoring of levels >8 yg/m3
would require additional design considerations to minimize low
readings at the higher concentrations and the memory effect
after high readings.

     The major problem was the inefficiency of the excitation
system to ignite or vaporize the beryllium containing particles.
Many large beryllium particles were not vaporized, resulting
in low values.   Also, the short period of signal integration
resulted in erratic results.

          4.1.2.15  References

     Churchill, W. L. and A.H.C.P. Gillieson, "A Direct Spectro-
     graphic Method for the Monitoring of Air for Minute Amounts
     of Beryllium and Beryllium Compounds," Spectrochimica Acta,
     1952, 5_, 238-250.

     Gillieson, A.H.C.P., "Monitoring for Toxic Dusts and Fumes,"
     Atomics 6_, 15-18 (1955).

                                130

-------
            Beryllium - Emission Spectrographic Analysis
            With Spark Discharge - Hollow Electrodes

           .1.3.1  Principle and Applicability

                          Principle - A high intensity spark
discharge is used to excite the beryllium in the air sample,
which is introduced through hollow electrodes.  Beryllium spark
excitation at 3130 to 3131 & is detected by a photomultiplier
tube and displayed on a strip chart recorder, calibrated in
pg Be/m3 .

               4.1. 3*1-2  Applicability - Instantaneous response
feature and ability to operate continuously can be useful in
site monitoring as a tolerance meter.  Quantitative response is
affected by "hash" or instantaneous noise level resulting from
intermittent spark discharge.

          JJ.1.3.2  Range and Sensitivity

               4.1.3.2.1  Range - Approximately 1 to 25 pg Be/m3

               *t. 1.3. 2. 2  Sensitivity - Detection level esti-
mated by authors as somewhat less than 1 yg Be/m3.

          4.1.3.3  Interferences

               *t. 1.3. 3.1  Chemical - None reported, but con-
sideration should be given to possibility of Hg spectral
interference .

               4.1.3.3.2  Physical - No data on particle size
effects is given, but method is dependent on particle size.
Atmospheric water vapor and particulates will influence spark
characteristics.  To reduce these effects an auxiliary spark
gap, quenched by filtered air was used.  Problems of deposits
on optical surfaces were encountered from the action of the
spark on the copper electrodes.

               *l.l.3.3.3  Memory Effect - No data reported.
Apparently, no performance tests were made.

          *<.1.3.*t  Accuracy, Precision and Stability
     Limited data reported.  For the range 1 yg/m3 to 25 yg/m3,
the 95% confidence limit for a single measurement over this
range was approximately ±1.3 Mg/m3 .  (Note - No allowance was
made in calculations for reduced "hash" level at lower beryllium
concentrations, which accounts for an estimated detection limit
which is less than the observed confidence limit.)
                                131

-------
     Operational tests revealed good long term (24 hours) sta-
bility if an auxiliary spark is used to minimize water vapor
and particulate effects and deposition on optical surfaces of
copper from electrodes is prevented.  However, intermittent
fluctuations in the spark discharge cause considerable instan-
taneous noise (hash).

          4.1.3.5  Apparatus
               4.1.3-5.1  Spectrograph - 1/2-meter,  Ebert-mounted
grating monochromator enclosed in constant  temperature box.

               4.1.3-5.2  Excitation Source - Conventional
laboratory spark source modified by adding  heavy-duty capacitors
for reliability under continuous operation  and adjusted to give
1 to 2 breakdowns per half cycle.

               4.1.3.5.3  Electrodes and Spark Chamber - Spark
chamber contains 3/8-inch copper tubing electrodes supported on
ceramic insulators.   Air sample is drawn through one electrode
into sample chamber.   (Note - 2nd electrode could be hollow
tubing or solid rod.)  Use of an open optical port was necessary
to overcome spattering from aerosol generated during spark action
on the copper electrodes.  An auxiliary spark gap (tungsten
electrodes) was used to minimize water vapor and particulate
interferences .

               4.1.3. 5. 4  Detection System  - Photomultiplier
coupled to electrometer and, in turn, to strip-chart recorder.

               4.1.3-5.5  Ancillary Equipment - Vacuum pump,
filters, calibration system, flow meter.

          4.1.3.6  Time Cycle for Sampling  and Measuring

     Monitoring of air at sampling rate of  0.088 mVmin. is
continuous .

          4.1.3.7  Calibration Procedure

     Primary, manual calibration was accomplished by aspirating
a standard beryllium nitrate solution through a flamephotometer
burner and allowing the exhaust gases to pass through the
sampling port .   From the solution concentration and flow rate
through the aspirator, the amount of beryllium present in the
exhaust gases was calculated.  A manual field calibration was
accomplished by using a beryllium-copper electrode as a
secondary standard.

          4.1.3.8  Method of Sampling

     Air drawn continuously into excitation chamber.
                               132

-------
          H.l.3.9  Multi-element Application

     Applicable to all metal elements.

          it. 1.3.10  Physical Dimensions

     None available.

          4.1.3.11  Unit Output

     Recorder tracing convertible to yg/m3.

          4.1.3.12  Safety Hazard

     Moderately high voltage to generate excitation discharge.

          4.1.3.13  Maintenance

     Copper-tubing electrodes could be used continuously for a
24-hr period before electrode gap required adjustment.  Electrode
replacement was required after several days operation.

          *t.l.3'l^  Recommendations for Method Improvement

     Needs capability to permit periodic calibration checks
during a test cycle.  Also, consideration should be given to
improved spark stability.  Questions concerning memory and
particle size effects remain to be answered.

           4.1.3.15   Reference

      Rowan,  J.  H. and W.  W.  Lee,  "An  Instantaneous  Direct-Reading
Beryllium  Air  Monitor,"  Analytical  Chemistry  in  Nuclear  Reactor
Technology,  Fifth Conference,  Gatlinburg, Tenn.,  16-21,  1961.
                                133

-------
     *J.m  Beryllium - Emission Spectrographlc Analysis
            With Spark Discharge - Filter Paper Tape

          4.1.4.1  Principle and Applicability

               *l. 1.4. 1.1  Principle - Samples of airborne beryl-
lium dust were collected automatically on filter paper tape and
subsequently analyzed by direct high voltage a.c. spark excita-
tion of the tape bearing the specimen and by conventional
measurements with a direct reading spectrograph set at 3131.07 A*.

               JJ.m.1.2  Applicability - The system can be used
for automatic analyses of ambient atmosphere for the continuous
quantitative determination of beryllium particulate matter.  The
method is rapid, resulting in quantitative analyses in 75 seconds,
When equipped with red lights and warning horn, the system has
been used as a tolerance meter when the toxic limit of 2 yg/m3
beryllium is reached.

          4.1.4.2  Range and Sensitivity
     Sensitivity of the unit for beryllium is better than
0.5 yg/cm3 for a 60-second sampling period and values of
0.1 yg/cm3 are readily obtainable.  Increased sensitivity is
possible by incorporating longer sample times in the operational
cycle.  The system was calibrated up to 20 yg/m3 .

          4.1.4.3  Interferences
               *t.l. 4.3.1  Chemical - Potential spectral inter-
ference from OH band, resulting from excitation of water used
in applying a covering tape to the sample, was reported as
minimal.  A background signal was reported at the level equiva-
lent to 0.1 yg/m .   No response differences for variation in
chemical form are shown.

               *t.l.4.3.2  Physical - No effects from particle
size differences are presented.

               4.1.4.3.3  Memory Effect - Memory effect was
minimized by incorporating a burn-off period of 18 seconds during
the 30-second sparking cycle.  This sufficed for most samples;
however, for high concentrations >15 yg/m3 a clean-up period
equivalent to three cycles or 3-5 minutes was required.

          ^.1.^.4  Accuracy, Precision and Stability
     When three or four results are averaged, a precision of
5 to 10% is readily obtainable.  Long-term laboratory-obtained
precision data are reported as about 20% at the 95% confidence
level.

-------
                   Apparatus
               4.1.4.5.1  Sampling System - An automatic filter
tape drive mechanism is used to sample the air, remove particu-
late, and advance the tape specimen into the spark gap.

               4.1.4.5.2  Spectrograph - N.S.L. 0.5 -meter plane
grating type, 55 » 000 lines/inch grating biased, 2nd order; slit
width of 40 microns (entrance) and 80 microns (exit).

                          Electrodes - Copper capped with 1/16-
inch platinum; 5/16-inch diameter (Tower) and 1/5-inch diameter
(upper).  (Note - Platinum capping minimized erosion and lowered
background. )

               4.1.4. 5. 4  Detector - Photomultiplier photocell
1P28-RCA with signal amplification by an electrometer input d.c.
amplifier.  Signal is integrated over the exposure period by a
summing condenser positioned in the feed-back circuit.

               4.1.4.5.5  Ancillary Equipment - Vacuum pump,
flow meters, filters, recorder and electronic power supplies.

          4.1.4.6  Time Cycle
     By programming overlapping events, an average time cycle
of 75 seconds is attained.

     The sequence of operations is as follows:  (1) sample
collection - 60 sec., (2) tape advancement and application of
protective coating - 5 sec., (3) sample excitation - 30 sec.,
(4) recording data - 5 sec., and (5) reset - 5 sec.  Although
the sample excitation is 30 seconds, exposure time consumes
only the first 12 seconds, whereas the last 18 seconds are used
to burn off any electrode contamination.

          4.1.4.7  Calibration
     Three methods were used:

     (a)  Beryllium oxide mixed at varying concentrations with
          graphite were weighed (1 mg) directly on the tape.

     (b)  Known amounts of beryllium oxide were dispersed in
          water by ultrasonic agitation, 10 yl aliquot s were
          pipetted on filter paper discs (0.25-in. diam.)
          backed with impervious masking tape, and dried by
          infrared heat at 29°C.

     (c)  Acetate solutions of beryllium were substituted for
          the oxide after initial calibration to provide a
          rapid routine check.
                               135

-------
          ty.l.JJ.8  Method of Sampling and Sample Preparation

     A filter tape (glass and asbestos type) system is used to
sample air at 50.7 liters/minute.  (Note - The authors believed
that a flow rate of this level was necessary to minimize the
sampling error anticipated with a random particle size distri-
bution.)  Dust is collected in a 9-mm circular area.  At the
conclusion of the sampling period, the tape is advanced auto-
matically to the spark gap for excitation.  During the advance
a layer of water-activated adhesive paper is applied to the
filter paper as a protective coating.

     The filter tape has a minimum retention of 95$ of 0.3 micron
dioctyl phthalate smoke particles.

          4.1.4.9  Mult i -element Application

     Technique can be applied to all common metal atmospheric
pollutants .

          4.1.4.10  Physical Dimensions
     Although size and weight factors are not given and the
system is considered mobile, the spectrometer used is quite
large and relatively heavy.

          4.1.4.11  Unit Output
     Data are displayed in three forms:

     (a)  A panel meter indicates the beryllium in air content
          (iag/m3) at all times during the exposure.

     (b)  A red light and warning horn operates when the toxic
          limit of 2 yg/m3 beryllium is reached.

     (c)  A strip chart recorder indicates the successively
          obtained beryllium concentrations in air for periods
          up to 8 hours.

          4.1.4.12  Safety Hazards
     Moderately high voltage required for spark discharge.

          4.1.^.13  Recommendations for Method Improvement

     To be useful as a stack monitor, the physical size must be
reduced and the monochromator system replaced, if possible, with
an interference filter system or smaller monochromator unit .

     The effect of the chemical form of beryllium needs to be
evaluated.
                               136

-------
          4.1.4.m  Reference

     Rozsa,  J.  T., J.  Stone,  0.  W.  Uguccini,  and R.  E.  Kupel,
"Beryllium Air  Monitor," Applied Spectroscopy 19, (1),  7-9
(1965).
                                137

-------
     4.1.5  Beryllium - Emission Spectrographic Analyses With
            Spark Discharge Directly on Filter Paper

          4.1.5.1  Principle and Applicability

               4. 1.5.1.1  Principle - Beryllium is determined
spectrographically on air sample filter papers by direct sparking
of the filter paper mounted on a silver disk which revolves on
a rotating platform electrode apparatus eccentric from a silver
counter electrode.  The intensities of beryllium lines are mea-
sured at 3130.42 A and 3131.0? A and compared to a molybdenum
internal standard (3132.59 &).

               4.1.5.1.2  Applicability - The method was developed
as a manual operation but is extremely fast and precise with a
minimum of sample handling.  The technique can be used with filter
samples containing as little as 1 x 10 * g of beryllium.

          4.1.5.2  Range and Sensitivity

               4.1.5.2.1  Range - Reported calibration range
1 x 10 9 g to 1 x 10"*, but can be adapted to higher levels.

               4.1.5.2.2  Sensitivity - Detection limit of
1 x 10~9 g.

          4.1.5.3  Interferences

               4.1.5.3*1  Chemical - None reported.

               4.1.5.3.2  Physical - None reported, but particle
size effects are mentioned as possible interferences.

               4.1.5.3.3  Memory Effect - If the filter paper
containing beryllium is removed from the electrode after sparking
for one revolution, beryllium "memory" lines are observed until
additional electrode sparking is continued for five revolutions.

          4.1.5.4  Accuracy, Precision and Stability

     Lack of certified standard samples precludes determination
of accuracy.  Repeatability of measurements gives 5$ standard
deviation at the 5 x 10 9 g and 1 x 10 8 g levels and 14$ at
2 x 10~9 g.

          4.1.5.5  Apparatus

               ^.1.5.5.1  Spectrograph - Jarrell-Ash Ebert,
3.4 m, 15,000 line/in, grating with slit width of 30 microns.

               4.1.5.5.2  Excitation Source - Jarrell-Ash Vari-
source, high voltage spark, five discharges per half cycle.
                               138

-------
               4.1.5-5.3  Detection System - Photographic
emulsion.

               4.1.5.5.4  Electrodes - High purity Ag disks,
1-1/8 in. diameter, 1/8-in. thick; Counter-High purity Ag rod,
5/l6-in. diameter, 2-in. length, flat topped.

               4.1.5.5.5  Air Sampler - Filtron portable air
sampler with 5 cfm blower and 1-in. diameter filter holder.

               4.1.5.5.6  Filter Paper - One-inch diameter
Whatman No. 40 filter papers were used for the analyses.  Glass
fiber filters gave poorer sensitivity, apparently due to the
fact that the paper was only partially consumed by the spark.
Millipore filters present difficulties in the addition of the
internal standard and in the preparation of standard samples.

               4.1.5.5.7  Ancillary Equipment - Rotating platform
electrode assembly, densitometer.

          4.1.5.6  Time Cycle - Sampling and Measuring

     Sampling time is variable and is limited by the time re-
quired to collect at least 1 x 10 9 g Be.  Sparking time is
60 seconds.  Mounting of filter paper on electrode would take
approximately 60-120 seconds.  Developing of the photographic
emulsion and measurement of line density would take approximately
15-30 minutes.

          4.1.5.7  Calibration Procedure

     Each filter paper was pretreated with 10 yg of molybdenum
as an internal standard.  The molybdenum is added by pipetting
100 yl of solution, containing 100 yg/ml Mo in dilute H2SOi»
onto each filter paper.  (Molybdenum solution spreads uniformly
through the filter paper.)  The paper is dried at room tem-
perature .

     Standard samples are prepared in the same fashion by adding
solutions containing beryllium and molybdenum to the filter
papers, to give 2, 5, or 10 myg Be in addition to the 10 ug Mo
per filter paper.  High purity molybdenum metal and high purity
BeO were dissolved in analytical reagent grade HaSOi, to prepare
the solutions.

          4.1.5.8  Method of Sampling and Sample Pretreatment

     Samples were taken on 1-inch diameter Whatman No. 40 filter
papers with a Filtron air sampler equipped with a 1-inch diameter
filter holder (Gelman Instr. Co.) and pumping 5 cfm.
                                139

-------
     For the spectrographic measurement, the filter paper was
mounted on the silver disk electrode with double-sided masking
tape, 1-1/2 inch width, in which a 7/8-inch diameter hole has
been punched.  The tape was placed on the disk electrode with
the hole concentric and the filter paper was pressed on the
tape.  A separate disk, freshly cleaned in concentrated reagent
grade PUSOi, at room temperature was used for each analysis.
     Sparking was accomplished on an offset rotating platform
for 60 seconds at rotating speeds of 5 rpm.  The filter paper is
consumed after one revolution, but a total of five is needed to
remove residual Be.

          4.1.5.9  Multi-element Application

     Technique could be used for most common metals found as
air pollutants .

          4.1.5.10  Physical Dimensions

     No physical dimensions are given.  However, the spectrograph
used in this study is a relatively large analytical instrument.

          4.1.5.11  Unit Output

     Calibration data yields a data output in myg which can be
changed to beryllium concentration in air by dividing by the
volume of air sampled.

          4.1.5.12  Safety Hazards

     High electrical voltages required to generate spark dis-
charge .

          4.1.5.13  Recommendation for Method Improvement

     Although the method was devised for manual operation,
suitable addition of mechanical manipulative devices could make
the technique automatic for sampling and sample preparation.

          4. 1.5.14  Reference

     Wheeler, G. V., W. A. Ryder, and K. R. Arnold, "Analysis
of Air Samples for Beryllium by the Rotating Platform Spectro-
graphic Technique," Applied Spectroscopy l6_ (1), 17-19 (1962).
                               140

-------
     4.1.6  Beryllium - Emission Spectrographic Analysis
            With A.C. Arc

          4.1.6.1  Principle and Applicability

               4.1.6.1.1  Principle - An emission spectrographic
method using a sustaining a.c. arc was developed based on excita-
tion of beryllium volatilized from a chloride matrix to attain
sensitivities as low as 0.001 microgram.  The beryllium line at
2348.6 A" was used.

               4.1.6.1.2  Applicability - Method is a manual
technique developed for analysis of beryllium in air particulate,
ores, and biological materials.  In its present form the tech-
nique is not applicable to continuous monitoring of stationary
source emissions.  However, the technology developed during the
method development can be used to increase response factors, to
better understand and minimize volatilization problems, and to
improve precision and accuracy of analyses.

          4.1.6.2  Range and Sensitivity

               4.1.6.2.1  Range - Reported as 0.001 microgram
to 0.1 microgram.

               4.1.6.2.2  Sensitivity - 0.001 micrograms.

          4.1.6.3  Interferences

               4.1.6.3.1  Chemical - A delayed volatilization
results from the presence of salts of Na and K (salts of alkali
metals above lithium).  A suppressing effect on the intensity
of trace element spectra produced in an a.c. or d.c. arc can
result from the presence of these alkali metals due to their
low ionization potentials.  The authors demonstrate the repres-
sive effect on volatilization by using a ratio of 0.1 yg Be to
4 mg of salt.  The presence of chloride ion (Li salt, but not
Na or K salt) increases volatilization.

               4.1.6.3-2  Physical - Response depends in large
part on the volatility of the beryllium compound.

               4.1.6.3-3  Memory Effects - With the d.c. arc,
complete vaporization of beryllium can take approximately
120 seconds.  By using the sustaining a.c. arc and chloride
buffer, the volatilization and excitation of 0.1 yg Be were
essentially complete in 90 seconds, and approximately 98? com-
plete in 75 seconds.
                               141

-------
          4. 1.6. 4  Accuracy, Precision and Stability

     An evaluation of the total analytical scheme, including wet
ashing, over a six-month period for samples containing 0.001 to
0.100 microgram quantities of beryllium, gave a mean coefficient
of variation of 13-9/2.

     The accuracy was evaluated by applying the total procedure
to eight beryllium ore samples analyzed previously by the Bureau
of Mines using a quinalizarin method.  For six samples the
% deviation ranged from 0.1-9.4? and for the remaining two the
% deviation varied from 13.6? to 15. &% .  The ore specimens con-
tained 0.1 to 3.1% Be and minute aliquot ranging from 1:100,000
to 1:2,500,000 were taken for spectrographic analysis.

          4.1.6.5  Apparatus

               4.1.6.5.1  Spectrograph - Bausch and Lomb large
Littrow type spectrograph with quartz optics, condensing lens
system, and a 20-micron slit.

               4.1.6.5.2  Excitation Source - NSL sustaining
a.c. arc Model PE 7090 operated at 18 amp.

                          Electrodes - Regular grade, spectro-
scopic graphite electrodes, 5/16-inch diameter.

               4.1. 6.5« 4  Detection System - Photographic plate.

               ^.1.6.5.5  Ancillary Equipment - Densitometer ,
micropipet assembly, air particulate samplers (filter, impinger
or electrostatic precipitator ) .

          4.1.6.6  Time Cycle

     A 90-second excitation period was used.  Total analysis
time is approximately 45 to 60 minutes.

          4.1.6.7  Calibration Procedure

     Beryllium standards were prepared from BeSCU^HaO  in 1:1
HC1 diluted with water.  The pH of each solution was maintained
between 1 and 2 to prevent deposition of beryllium on the walls
of the glassware.  The intensity ratio of the Be 23*18.6 A" line
and the background on the higher wavelength side of the beryllium
line were recorded and correlated with the amount of Be used for
calibration.

          4.1.6.8  Method of Sampling and Sample Pretreatment

     Samples can be obtained as residues on filters, or impinger
solutions.  No details are given for collecting air particulate


                                142

-------
specimens.  The sample pretreatment involves wet ashing, conver-
sion to chloride form, and dissolution in water.  The liquid was
deposited on waterproofed electrodes to which LiCl-graphite
buffer was added.

          4.1.6.9  Multi-element Application

     Technique can be applied to most metal elements encountered
as air pollutants.

          4.1.6.10  Physical Dimensions

     No data are given, but a large laboratory spectrograph was
used.

          4.1.6.11  Unit Output

     Data output is presented as micrograms of Be.

          4.1.6.12  Safety Hazards

     High electrical voltages are used to generate the electri-
cal arc.

          4.1.6.13  Recommendations for Method Improvement

     In its present form, the method is not applicable to con-
tinuous monitoring.  However, the technology regarding the sus-
taining a.c. arc and the improved volatilization of Be should
be useful in other spectrographic techniques.

          4.1.6.1*1  Reference

     Keenan, R. G. and J. L. Holtz, "Spectrographic Determination
of Beryllium in Air, Biological Materials, and Ores Using the
Sustaining A.C. Arc," Am. Ind. Hyg. "Assoc. J. 25., 25^-263 (1964).
                               143

-------
            Cadmium, Chromium ,  Copper ,  Lead,  and Nickel-Emission
            Spectrographic Analysis With Spark Discharge

            1.7.1  Principle and Applicability
                          Principle - A spectrographic technique
for the direct measurement of elements of airborne dirt collected
on filter paper has been developed.  The technique is not con-
tinuous, but requires separate collection and analysis systems.
The filter paper is manually transferred to the spectrograph
after folding the paper into a cylindrical form.  By using a
push-up electrode systerrij the cylindrical filter paper is forced
through a hollow graphite electrode into the spark gap where the
filter paper is consumed in an oxygen atmosphere and the metal
elements are excited to optical emission.

     Spectra are produced and recorded on a direct reading spec-
trograph interfaced with an IBM 026 card punch.

               4.1.7.1.2  Applicability - The method was devised
to monitor dirt level in manufacturing facilities for controlling
airborne contamination and to measure twelve elements which can
be related to the source of contamination in the manufacturing
atmosphere.  Correlation of the dirt level observed in manufac-
turing areas were made with work schedules, machine wear, and
air conditioning efficiency.

          4.1.7.2  Range and Sensitivity

               4.1.7.2.1  Range


           Element      Wavelength (fl)     Range (pg)

           Cadmium         2288.02          0.1-100
           Chromium        3021.56          0.1-100
           Copper          3247.54          0.1-25
           Lead            2833-07          0.1-100
           Nickel          3002.49          0.1-100


               4.1.7.2.2  Sensitivity - A detection limit of
0.1 yg is reported for each element.

          4.1.7*3  Interferences and Sources of Contamination

               4.1.7*3.1  Chemical - Analytical curves were S
shaped, showing a toe effect at the lower concentration which
is attributable to background and residual element in the filter
paper.  The shoulder at the higher concentration may be due to
incomplete sample excitation and self-absorption.  Curves are
linear over the 0.3-10 yg range.
                                144

-------
     Care to avoid contamination when preparing the filter paper
is required.  Improper manipulation of the filter can result in
considerable error-.

               4.1.7.3.2  Physical - None are mentioned directly,
although there is a suggestion that incomplete sample excitation
may occur.  No data showing possibility of particle size effects
are presented.

               4. 1.7« 3. 3  Memory Effect - None with replaceable
electrodes.

          4.1.7-4  Accuracy, Precision and Stability
     Precision is approximately ±25/8 of the amount present at
the 3 yg level.

          4.1.7-5  Apparatus

               4.1.7.5.1  Spectrograph - Baird-Atomic 3 m re-
search direct reading, model GX-1, 1200 grooves/mm, 1st order
with Baird-Atomic Spectrosource, model NB-1.

               4.1.7.5.2  Electrodes - (a) sample - 1/4 in. x
1-1/2 in. high purity graphite, drilled axially 1/8 in. diameter
through length of electrode; (b) counter - high purity graphite.

               4.1.7«5*3  Excitation - High voltage a.c. spark
with 6 breaks/half-cycle.

               4.1.7.5.4  Detector - Photomultiplier of 1960-
3300 8 range and 14 channel digital clock read-out .

               4. 1.7 -5.5  Push-up Assembly - Cardiod cam and
push rod.

               4.1.7.5.6  Oxygen Distribution Nozzle - 30 ftVhr.

               4.1.7.5.7  Sample Collector - (a) 25-mm diameter
filter paper (Schleicher and Schnell type 589-1H) in nylon filter
holder with stainless steel support screen, retaining particles
>0.5 micron; (b) membrane-type filter - Millipore Filter Corp.
or Gelman Instrument Co.  (Note - These brittle and flammable
membrane filters had to be prepared by rolling the membrane to-
gether with a filter paper or dissolving the membrane on the
paper. )

               4.1.7'5.8  Ancillary Equipment - Vacuum pump,
tubing, motor drive for push-up electrode system, computer
interfacing.

-------
          4.1.7.6  Time Cycle for Sampling and Measuring

               4.1.7.6.1  Sampling - Time adjusted to fit
calibration curves.

               4.1.7*6.2  Measuring - Excitation period
30 seconds for a total analysis time of 5 minutes.

          4.1.7.7  Calibration Procedure

     Standard filter papers were prepared by solution addition
of the elements to blank filter papers.  All elements except
silicon were combined in a single solution.  Standards for sili-
con were prepared using reagent-grade sodium metasilicate in
distilled water.  Standards were exposed in quadruplicate and
analytical curves were drawn relating Baird digital clock read-
ings to log of concentration.  Equations (5th order) of these
curves were determined and programmed into an IBM S/360 model
65 computer.

          4.1.7.8  Method of Sampling and Sample Preparation

     For routine ambient atmosphere sampling, a steam or com-
pressed air ejector (Graham Mfg. Co., Inc.) was used to provide
a vacuum of 15 in Hg and a flow rate of 0.5 ftVmin.  Flow
control was accomplished with an orifice of 0.045-inch in
diameter.  Collection efficiency was 955? or better on all par-
ticles larger than 0.5 micron.

          4.1.7.9  Multi-element Application

     Applicable to all common metal pollutants.

          4.1.7.10  Physical Dimensions

     Not given, but spectrograph is a large research laboratory
model.

          4.1.7.11  Unit Output

     Micrograms of element and a five digit profile number based
on the percentiles of certain elements in the total dirt.  Each
digit of the profile number indicates to nearest 10%, the pro-
portion of the total contributed by a particular element or
group of elements, e.g., 1st digit - iron; 2nd digit - copper;
3rd digit - zinc, tin, and lead; 4th digit - calcium, magnesium,
and silicon; and the 5th digit - chromium and nickel.

          4.1.7.12  Safety Hazard

     High voltage spark source.
                               146

-------
          4. 1.7. 13  Recommendations for Method Improvement

     As described, the technique is obviously not suitable for
continuous monitoring.  Although a mechanical means of folding
a filter and introducing the specimen through a cored electrode
may possibly be developed, this method of handling the sample
filter does not appear practical for an unattended, automatic
analyzer.

          4.1.7.14  Reference
     Lander, D. W. , R. L. Steiner, D. H. Anderson, and R. L.
     Dehm, "Spectrographic Determination of Elements in Airborne
     Dirt," Applied Spectroscopy 25.(2), 270-275 (1971).
                                147

-------
     4.1.8  Emission Spectrographic Analysis for Cadmium With
            a Microwave Induced Argon Plasma

          4.1.8.1  Principle and Applicability

               4.1.8.1.1  Principle - A non-flame cell using a
low power (20-70 W) microwave induced argon plasma at atmospheric
pressure was used for emission spectroscopic measurements.  The
method was tested on a number of elements, but principally on
cadmium (228.8 nm) .  A rapid electronic amplifier/integrator
system was used to follow the weak and transient emission signals.

               4.1.8.1.2  Applicability - The method was demon-
strated as applicable to detect zinc, antimony, selenium, arsenic,
lead, cadmium, iron, iodine, copper, boron, beryllium, and mer-
cury down to 10 9 to 10~13 gram of metal in aqueous solution.
Quantitative measurement of cadmium was performed over the range
0.01-5.00 pg Cd ml 1.  It was not possible to study concentra-
tions above 5-00 ng Cd ml"1 with the electronic integrator.
Method used minute amounts of sample (0.12
          4.1.8.2  Range and Sensitivity

               4.1.8.2.1  Range - Cadmium, 0.01-5.00 yg Cd ml"1

               4.1.8.2.2  Sensitivity


              Element           Detection Limit (g)

             Cadmium                2.0 x 10" 1 3
             Beryllium              1.2 x 10~10
             Zinc                   8.0 x 10"1 1
             Antimony               5.0 x 10" 10
             Selenium               4.0 x 10" 10
             Arsenic                4.2 x 10"1 1
             Lead                   1.2 x 10~10
             Iron                   3-0 x 10~10
             Iodine                 1.2 x 10 9
             Copper                 1.2 x 10" 10
             Boron                  1.2 x 10~10
             Mercury                1.6 x 10" ! 1


          4.1.8.3  Interferences

               4.1.8.3.1  Chemical and Spectral - Vanadium,
present in the form of ammonium metavanadate at 1000-fold molar
amounts compared to cadmium, interfered giving  a 40% reduction
in signal intensity.

     At concentrations of 1000-fold molar amounts, phosphate,
silicate and aluminum did not interfere .
                               148

-------
     When cadmium solutions containing 0.25 yg Cd ml"1 and other
elements in 1000-fold molar amounts were analyzed, only arsenic
showed moderate spectral interference from the 228.812 nm direct
line emission.  No interference was observed from either cobalt
(228.781 nm) or antimony (228.898 nm) when present in a 1000-fold
molar amount compared to cadmium.

     Most useful portion of spectrum is below 300.0 nm where
the plasma background is lowest.

     A light leak in the spectrophotometer resulted in distorted
integrated emission signals when cadmium solutions containing
sodium (>, 5000 yg/m"1) were analyzed.

               4.1.8.3-2  Physical - Arsenic evaporation from
the platinum filament occurred in several stages:  (a) at room
temperature, and (b) at various filament temperatures.  These
data suggest that variations in emission intensity could depend
on the volatility of the species.

          4.1.8.4  Accuracy, Precision and Stability

     Precision data were obtained by analyzing thirty solutions
containing 0.25 yg Cd ml"1.  The relative standard deviation
was 10%.

          4.1.8.5  Apparatus

               4.1.8.5.1  Spectrograph - Unicam SP900
spectrophotometer.

               4.1.8.5.2  Excitation System - "Microton 200"
microwave generator (2450 ± 25 MHz) with Evenson type 1/4 X
cavity producing stable plasma with powers ranging from 20-75 W
and optimized at 40 W, plasma cell, platinum or tungsten filament
for sample vaporization, and Tesla coil ignitor.

               4.1.8.5.3  Detection System - EMI 9601B photo-
multiplier with specially designed rapid amplifier/integrator
unit (see reference) and Servoscribe recorder.

               4.1.8.5.4  Ancillary Equipment - Argon supply,
flow controller and flowmeters.

          4.1.8.6  Time Cycle for Sampling and Measuring

     No data given; excitation time is <2 seconds; sample time
including evaporation of solvent is estimated as <2 minutes.
                                149

-------
          4.1.8.7  Calibration Procedure

     Standard solutions were prepared from analytical-grade
chemicals and deionized water.  Ten samples at each concentration
were vaporized into the plasma and the emission intensities were
averaged.  A log/log curve of signal vs concentration was plotted.
The normal analytical working curve was linear over the concen-
tration range 0.01-5.00 yg Cd ml"1^  It was not possible to study
concentrations above 5-00 yg Cd ml *, due to electronic satura-
tion of the integrator.

     Calibration curves constructed without using the integrator
were not reproducible and were non-linear; the signal increased
iJ-fold for a 1000-fold increase in concentration.

          *l.l.8.8  Method of Sampling and Sample Preparation

     Aqueous solutions (0.12 yl) were retained on an elliptical
platinum wire loop by a surface tension film.  The aqueous film
was evaporated first by passing a 0.1 amp d.c. current through
the filament.  Then the current was increased to 1.9 amp which
caused the compound remaining on the loop to vaporize and pass
into the plasma region.  Tungsten loops were used when a higher
volatilization temperature was required.

          4.1.8.9  Multi-element Application

     Technique can be used for all metallic and non-metallic
elements.  The principal problem is the method of sample intro-
duction and potential losses of very volatile elements or
compounds.

          4.1.8.10  Physical Dimensions

     None given, but spectrometer is medium sized laboratory
instrument.

          4.1.8.11  Unit Output

     Output expressed in weight units, grams or micrograms.
Data can be converted to mass/unit volume.

          4.1.8.12  Safety Hazard

     Moderate voltages in electronic amplifier.

          4.1.8.13  Recommendation for Method Improvement

     Amplifier/integrator system should be modified for higher
signal capacity.
                                150

-------
     Method is not directly applicable to continuous monitoring
of stack emissions.  Modifications of sample introduction system
are required.  The use of a metal frit or electrostatic collector
should be considered.

          4.1.8.3.11  Reference

     Aldous, K. M., R. M. Dagnall, B. L. 'Sharp, and T. S. West,
"A Microwave-Induced Argon Plasma System Suitable for Trace
Analysis," Anal. Chim. Acta 5l, 233-243  (1971).
                                151

-------
     4.1.9  Radio-Frequency Induced Plasma-Spectroscopic Studies

          4.1.9.1  Principle and Applicability

               4.1.9.1-1  Principle - Emission spectra of various
gases - argon, helium, nitrogen, air, water, argon and air, argon
and oxygen, argon and nitrogen, or argon and helium - were ob-
tained by radio-frequency (8 MHz) excitation of optical emission
spectra.

               4. 1.9.1.2  Applicability - Technique was not
devised for trace metal analyses, but the spectral data are
important to determine potential spectral interferences from
the sample background.

          4.1.9.2  Range and Sensitivity

     No data reported.

          4.1.9-3  Interferences

               4.1.9.3.1  Chemical - When a short coolant tube
was used, the after-flame emitted band spectra characteristic
of 02,+N2, OH and weak bands which were apparently due to NH
and N2  .  When the coolant tube was very short, the secondary
region extended beyond the end of the coolant tube Into the
atmosphere and these band systems became more prominent and the
continuum increased.  The results of this torch configuration
were quite similar to thse obtained when large air samples were
introduced into the coolant system.

     When water vapor was introduced Into either the coolant or
center streams, moderately strong OH bands (3064 ft) were emitted
from all regions of the plasma and the overall intensity of the
plasma was decreased while the amount of heat dissipated in-
creased.  Balmer-series hydrogen line were also found.  These
lines were very strong and quite broad from the core and secon-
dary region and were stronger when the sample was introduced
into the core than when the sample was introduced into the
coolant stream.  The hydrogen lines rapidly weakened in the
tail-flame and disappeared before the end of the coolant tube
was reached.  The 2800 X OH band was also found occasionally,
but was only 105? as strong as the 3064 ft band.  Low concentra-
tions (no values reported) had little effect on the intensities
of the argon lines and almost no visual effect.  The introduction
of water vapor into the core causes considerable increase in
the intensity of the continuum and thus reduced the line-to-
background ratio of the argon.  The intensity of the 3064 S OH
band was found to increase from the core to the tail-flame, while
the effect on the Balmer-series hydrogen was reversed.  The OH
                               152

-------
bands were found to increase in intensity with increases in the
radio-frequency power.

     When nitrogen was introduced, the spectra obtained from all
three regions of the plasma discharge (core, secondary or transi-
tion zone, and tail-flame) showed the second positive nitrogen
system and at higher power levels, particularly from the core and
secondary region, bands of the first negative nitrogen system
appeared, originating with Na .  Atomic nitrogen lines are prob-
ably present, but are too weak to detect.

     Introduction of oxygen, even at low levels, caused severe
reduction in the line intensities.  At higher concentrations (no
levels given), a poorly developed oxygen band system appeared.

     Introduction of helium had no effect on the spectrum as far
as the number of emission lines and bands are concerned, but did
cause a slight reduction in the intensities of the argon lines
and an increase in the intensity of the OH bands and the con-
tinuum.  No emission lines characteristic of helium were found.

     Air introduction produced an effect which was a mixture of
the effects produced by oxygen and nitrogen.  Nitrogen and oxygen
bands were prevalent as in the case of either gas and the line-
to-background ratio decreased along with a smaller decrease in
the line intensities.  The OH bands became very prominent and
the NOy-system appeared.  The effect of rf power on emission
intensity for moist nitrogen is reported graphically.

                          Physical - When nitrogen, oxygen,
helium and air replaced argon in the coolant stream, the plasma
discharge was not extinguished.  However, with some mixtures it
was necessary to increase power level to 2 kVA to maintain the
discharge.  Pure samples of air, nitrogen, oxygen and helium in
the coolant stream required approximately twice the minimum
power (ca. 1.4 kVA) used for argon alone.

     In the case of nitrogen and helium, all of the argon could
be replaced in the center tube, but in the case of air and
oxygen, pure gas samples could not be introduced into the center
stream without raising the power levels.

     In general, introduction of any of the gases into the center
stream resulted in severe contraction of the core, a reduction
in the line-to-background ratio and reduction of line and band
intensities in the spectrum obtained of the core and secondary
region and to a lesser extent from the tail-flame.  Introduction
of sample gases into the tail-flame did not affect the core or
secondary region, but drastically changed the spectrum of the
tail-flame .
                                153

-------
     Continuous operation was not possible at  power levels above
1.75 kVA with pure argon because the plasma attacked the quartz
tubing.

     Considerably more sample (up to 10 times) could be handled
in the coolant stream as in the center stream  without extinguish-
ing the plasma.  The amount of sample introducible into the
coolant stream was limited only by build-up of carbonaceous
material from samples highly loaded with organic compounds or
condensation in the torch head when liquid samples were used.
The latter problem was minimized by operating  in the discharge
downward position.

     Care must be taken when recording the emission spectrum
from the core and secondary regions to prevent saturation of
the photomultiplier.

          4.1.9.4  Accuracy, Precision and Stability

     Limited data are reported.  Continuous monitoring of the
4201 8 argon line with the instrument set to read full-scale
deflection gave a signal-to-noise ratio of 100:1 to 200:1 with
maximum variation from the mean of 2% over 3 successive 20-min.
runs produced over a period of 2 hours.  Over  short periods of
time, ca. 1 or 2 days, the system may be shut  down and re-
started and the readings reproduced within 1-2/? without adjust-
ment of equipment.

          4.1.9.5  Apparatus

               4.1.9.5.1  Spectrograph - Modified Jarrell-Ash
atomic absorption spectrophotometer, Model 82000 with a 0.5 m
Ebert mount scanning monochromator.

               4.1.9.5.2  R-F Power Unit - Lepel T-5-2-MC-J-B
with 5 kVA nominal output operated at 8 MHz.

               4.1.9.5.3  Excitation Source -  Wendt and Fassel
type plasma torch; dual-tube, laminar-flow.

               4.1.9»5.^  Ancillary Equipment  - Strip chart
recorder, gas manifold, silica-gel drying tower, ascarite tower,
argon pre-heater and oxygen stripper (3 ft x 0.5 in. copper
tubing wrapped with a 800-W heater), carbon rod igniter, external
spark coil, metering valves, and flowmeters.

          4.1.9.6  Time Cycle for Sampling and Measuring

     No data reported.

-------
          4.1.9.7  Calibration Procedure

     None given.

          4.1.9.8  Method of Sampling

     Gaseous samples were introduced through a metering valve and
rotameter into a mixing chamber where the sample was mixed with car-
rier gas or introduced directly into the torch without carrier gas.

     Samples could be introduced into the plasma in three places:
(a) center stream in which case a portion of the sample would be
introduced directly into the core of the discharge, (b) coolant
stream in which case most of the sample would pass by the core
and only limited mixing would occur, or (c) tail-flame with or
without carrier gas.

     When air was examined only 2 mg/sec could be introduced into
the center stream, representing a composition of 20% air and B0%
argon.  However, pure nitrogen (22 liter/min.) could be introduced
into the coolant stream or into the tail-flame.

     The flow conditions under which a plasma was sustainable were
critical and are reported in detail by the authors.

          4.1.9.9  Multi-element Application

     Applicable to all common metal pollutants and potentially
applicable to non-metallic elements.

          4.1.9.10  Physical Dimensions

     Not given, but spectrometer is moderately sized research model,

          4.1.9.11  Unit Output

     None reported.

          4.1.9.12  Safety Hazard

     High voltages; open flame.

          4. 1.9. 13  Recommendations for Method Improvement

     Although method was not used for metal analysis directly, the
concept and technology are useful for trace element spectrographic
measurements .

          4.1.9.14  Reference
     Truitt, D. and J. W. Robinson, "Spectroscoplc Studies of
Radio-Frequency Induced Plasma, Part I. Development and Charac-
terization of Equipment," Anal. Chim. Acta 4_£, 401-415 (1970).


                               155

-------
     4.1.10  Emission Spectrographic Analysis With Low Wattage
             Microwave Induced Argon Plasmas at Atmospheric
             Pressure

          4.1.10.1  Principle and Applicability

               ^1.1.10.1.1  Principle - Optical emission spectra,
generated by microwave(2^50 MHz) induced excitation in an argon
plasma, are used to measure metals in the_range of 10 11 to
10 12 gram on a sample size of 10~5 to 10 6 gram.  Relatively
low microwave power levels of approximately 25 watts gave maxi-
mum emission intensities for volatile metal salts or organo-
metallics.

               *t.1.10.1.2  Applicability - The technique was
applied to volatile copper, chromium, iron and silver compounds
in solution.  Results indicate that the method is extremely
sensitive with detection limits as low as 10~12 gram of metal.

     Major limitations are the upper limit of the amount of mate-
rial (10 6  to 10~5 gram per second), which can be introduced
into an atmospheric pressure argon plasma before the plasma is
extinguished, and the low thermal temgeratures of the discharge.
To use samples larger than 10 6 to 10 5 gram per second, a sepa-
ration or concentration step is required.  Measurement of solu-
tions is accomplished by evaporating the solvent from a sample
deposited in a loop of metal filament.  A secondary heating
source - resistance heating of the metal filament - is used to
vaporize the metal compound into the discharge zone.

          4.1.10.2  Range and Sensitivity

               4.1.10.2.1  Range - Approximately 10~12 gram
to 10~7 gram.

               ^.1.10.2.2  Sensitivity - Results indicate detec-
tion limits as low as 10~1Z gram of metal.

          4.1.10.3  Interferences

               ^.1.10.3.1  Chemical - The mechanism of certain
enhancement or depressive effects is not fully understood.  The
presence of another metal in the sample matrix results in
response enhancement until a level of 10~8 gram of metal being
measured is reached.  However, when glycerin was added, a depres-
sive effect on intensity was noted above 10 9 gram of total
sample.

     Presence of oxygen in the inert gas (argon) can result in
formation of relatively stable molecular species.
                               156

-------
               4.1.10.3.2  Physical - Principal factors which
affect sensitivity are microwave power, position in discharge,
sample vaporization rate, argon flow rate, sample size and com-
position, and temperature of argon.  Most of these factors can
be optimized and stabilized; many interact with each other.

     One of the most important effects is the variation of
spectral emission intensity as a function of the position in
the discharge.  (The spectral emission is a distribution func-
tion which may have an intensity maximum anywhere over a distance
of 0.1 to 5.0 cm in the discharge.)  The intensity maximum shifts
depending on the element being studied, on the matrix, and on
sample evaporation rate.  The maximum for acetylacetonates con-
taining copper, iron, cobalt or chromium may occur within the
first few millimeters of the discharge.  Copper is not carried
through the entire discharge, but plates out on the walls of
the plasma tube.  With normal sample sizes, a given tube may be
used for several days without interference.

     The position and intensity of the maxima are also affected
by the presence of a matrix.  Excitation of pure metal chelates
yield lower intensities and maxima, which occur earlier in the
discharge, than excitation of the same amount of metal chelates
in a matrix containing another metal chelate.  The analytical
significance of the position effect is that intensity observa-
tions should be made in the first few centimeters of the
discharge.

     Considerable variation in spectral intensity occurs depend-
ing on the volatility of the sample, heating rate of the filament
(for vaporizing sample) and gas flow rate.  Low quantities of
pure materials yield different emission intensities than the same
quantity of material in a sample specimen containing other metal
compounds.  (Note - Addition of chromium or other metal complexes
to a copper complex in sufficient quantity to yield a final
sample size of approximately 10 6 gram minimizes interference
phenomena.)

     A satisfactory metallic matrix to add to silver iodide or
copper chloride could not be found.  It is essential that the
evaporation rates be nearly equal so that both the matrix metal
and sample enter the discharge simultaneously.

     Since the sample is introduced into the argon stream by
evaporation from a hot filament which causes some heating of
the argon during the evaporation cycle, a lowering of the line
emission and an increase in background can occur.  This effect
can be minimized by using filaments with a diameter <0.il mm.

     Plasma discharge tube was limited to  1 mm i.d.  When a
larger tube is used, the position of the discharge within the
tube is not stable and tends to wander.


                                157

-------
               *t. 1.10.3.3  Memory Effects - Some plating out of
metals can occur on the walls of the excitation tube.

          *t. 1.10.4  Accuracy, Precision and Stability

     Precision is such that analytically useful determinations
can be made when the sample signal is ^% of the total signal
(background + sample signal).  The percent standard deviation
at the detection limit for each metal is approximately 40/8.
At 10 10 gram, the percent standard deviation is 7%.

          4.1.10.5  Apparatus

               4.1.10.5.1  Spectrograph - JACO No. 82000 scanning
Ebert 0.5 meter; 1180 line/mm, 3000 A blaze, 10 micron or 25
micron fixed slits.

               4.1.10.5.2  Excitation System - 100 watt microwave
generator (2*150 mHz), microwave coupling cavity, mount and plasma
cell.

               4.1.10.5.3  Detection System - Photomultiplier
(1P21 and 1P28), phototube power supply, amplifier, strip chart
recorder.

               4.1.10.5.4  Ancillary Equipment - Argon supply,
flow controller, flow meter, optical bench, lens and mounts,
vaporization chamber with platinum filament.

               4.1.10.5.5  Gas Purification System - Spectral
interference from nitrogen was reduced by passing argon over
granulated magnesium heated in a tube furnace to 550°C.

          4.1.10.6  Time Cycle for Sampling and Measuring

     None given; excitation time is <2 seconds.  Total time for
sample introduction, evaporation, and excitation estimated as
<2 minutes.

          4.1.10.7  Calibration Procedure

     No specific technique was reported.  Calibration assumed
based on use of solutions of known concentrations of metal com-
plexes (acetylacetonates) or salts in organic solvents or water.

          4.1.10.8  Method of Sampling and Sample Preparation

     Sample  (10~12 to 10~5 gram of solid in 10 yl of organic or
water solvent)  is deposited on a platinum filament.  After  evapo-
ration of solvent, the platinum filament is heated by passing an
electrical  current through the filament to vaporize the residual
solid into  the  microwave discharge zone.
                               158

-------
          4.1.10.9  Multi-element Application

     Technique is applicable to all metallic and non-metallic
elements but limited somewhat by the method of introducing the
sample.  Modifications would be necessary depending on the sample
volatility.  Technique was applied principally to water or
organic-based solutions.

          4.1.10.10  Physical Dimensions

     None given, but spectrometer system is medium sized labora-
tory spectrograph.

          fr.l.lO.ll  Unit Output
     Output expressed in weight units, i.e., grams or micrograms.
(Note - Since a specific volume of solution was analyzed, simple
calculations yield mass/unit volume.)

          H. 1.10. 12  Safety Hazard

     Moderate voltages in electronic amplifiers.

          ^.1.10.13  Recommendations for Method Improvement

     As described, the method is not applicable for a continuous
monitoring system.  The method of sample introduction requires
that source effluent would have to be collected in an impinger
system as a solution or dispersion in a liquor and subsequently
introduced into the microwave discharge.  Alternate possibilities
include collection of particulate on a metal filter (frit) which
could subsequently be heated to vaporize the sample.  Care would
have to be taken to prevent overloading of the discharge zone
with sample, resulting in plasma quenching or inefficient
excitation.

     An evaluation of operation at reduced pressure (1-10 torr)
should be made to determine whether improvement in response or
reduction in interferences are observed.  Argon was selected as
a plasma gas in this study because of its monatomic character
and moderate ionization potential.  Low wattage microwave exci-
tation of molecular vapors or gases with high ionization poten-
tials to form plasmas is difficult unless the gas pressure is
reduced.

          4.1.10.1*1  Reference

     Runnels, J. H. and J. H. Gibson, "Characteristics of Low
Wattage Microwave Induced Argon Plasmas in Metals Excitation,"
Anal. Chem. 39, 1398-1405 (1967).
                                159

-------
     4.1.11.  Beryllium-RF Plasma Torch - Emission Spectroscopy

          >4. 1.11.1  Principle and Applicability

               *|. 1.11. 1.1  Principle - An inductively coupled
high-frequency (36 MHz) argon plasma torch was  used to excite
emission spectra from which quantiative composition analyses of
trace metal elements were made.   Measurements were performed in
the low ultraviolet region, 23^.9 nm (beryllium)  and 2^9.7 nm
(boron) .

               fr. 1.11. 1.2  Applicability - The  system is capable
of exciting the major spectral lines of trace,  refractory-oxide-
forming elements (Be and B) present in a powdered matrix (metal
oxides).  Although the  technique was not applied  to the direct
analysis of air particulate, the solid samples  were injected
into the plasma from a fluidized bed and with suitable modifica-
tion the system may be applicable to air particulate monitoring.

          4.1.11.2  Range and Sensitivity
               4.1.11.2.1  Range

     Beryllium - 0.1-500 ppm in solid MgO.   (Note - Deviation
          from linearity begins around 150  ppm.)

     Boron - 2.5-1000 ppm in solid MgO.   (Note -  Deviation
          from linearity begins around 600  ppm.)

               4.1.11.2.2  Sensitivity - Limits of detection
(signal = 2X standard deviation of background) for beryllium
and boron were 0.1 ppm and 2.5 ppm, respectively, in solid MgO.

          ^.1.11.3  Interferences and Sources of  Contamination

                           Chemical - The effects of 10,000 ppm
of aluminum, calcium, lithium, tantalum and zinc were investi-
gated on 10 ppm of beryllium and boron in magnesium oxide, but
in no instance was the variation in emission intensity signifi-
cantly greater than that shown by the standard deviation for
beryllium or boron alone.

     Although no comments  were made by the authors on the devi-
ations from linearity of their calibration curves, the data
suggest a concentration effect occurs which could be attributable
to spectral self-absorption for the beryllium 23^-9 nm line or
incomplete vaporization and excitation of the beryllium and boron
containing particles at the higher loadings.
                               160

-------
               4.1.11.3.2  Physical - None mentioned, but see
4.1.11.3.1.

               ^.1.11.3.3  Memory Effect - None reported, appa-
rently not studied.

          4.1.11.4  Accuracy, Precision and Stability

     Relative standard deviations for 20 samples containing
10 ppm impurities added to magnesium oxide were 6.5% for beryl-
lium and 105? for boron.

     No other data are reported.

          4.1.11.5  Apparatus

               4.1.11.5-1  Spectrophotometer - Unlearn SP900A
flame spectrophotometer with entrance slit reduced to 2 mm in
height .  (Note - Instrument was screened from high frequency
pick-up.  All component packs within the spectrophotometer were
grounded and all external leads were sleeved with copper braid.)

               4.1.11.5.2  Plasma Generator - Frequency 36 MHz.
Maximum power output 2.5 kW.

               ^.1.11.5.3  Plasma Torch - Brass base with two
double 0-ring seals and B14 socket, all concentric.  Tangential
gas injection for cooling and plasma flows via two quartz tubes
(outer 25 mm and inner 21 mm internal diameter respectively with
1.5 mm nominal wall thickness).  The outer quartz tube extends
30 mm above the inner tube and the overall torch height is 18 cm.

     Gas flow rates used are:  plasma flow (argon) 7.51 min"1,
cooling flow (nitrogen) 7-51 min 1 and injector flow (argon)
0.51 min J .

               4.1.11.5.4  Working Coil - 2.5 turns of a O.l875-in
square section of 20 gauge copper tubing with its ends silver-
soldered to 0.25-in. BSP fittings for connection to cooling water
and high-frequency power supply.  The coil separation is 1.5 mm
with an internal diameter of 32 mm and an outside diameter of
4l mm.  Coil winding ascends anticlockwise when viewed from above
(i.e., from the open end of torch).  The coil is located concen-
trically 4 mm above the upper end of the inner quartz tube.

               ^.1.11.5.5  Fluidized-bed Chamber - Sintered glass
disc (2 cm diameter, porosity 1) fused inside a borosilicate
glass tube which is drawn out, 10 cm above and 2 cm below the
disc, to 0.25-in. outside diameter tubing.  The required aliquot
of powder is placed above the disc and gas is injected from
below.
                               161

-------
               4.1.11.5.6  Recording System - Standard galvan-
ometer attachment to 0-1 mV strip chart recorder; galvanometer
and recorder leads were converted to coaxial cable and an
auxiliary 6.3 V transformer provided power for the galvanometer
bulb and exterior instrument lights.

               4.1.11.5.7  Ancillary Equipment - Tubing, gas
flow regulators and meters, cylinders of argon and nitrogen gas.

          ^.1.11.6  Time Cycle for Sampling and Measuring

     Total time cycle not reported.  The emission intensity was
recorded over a period of 30 seconds.

          4.1.11.7  Calibration Procedure

     Dry, powdered beryllium carbonate or boric acid was added
to powdered magnesium oxide.  Varying concentrations w?re ob-
tained by dilution with magnesium oxide.  The calibration powder
was dispersed in the fluidized bed chamber and injected into the
plasma torch.

          4.1.11.8  Method of Sampling and Sample Preparation

     Sample was introduced into the plasma torch from a fluidized
bed chamber containing 3-5 g (±0.1 g) of powder.  Ignition of
the plasma was accomplished by inserting a carbon rod into the
torch mouth.  Pine adjustments of injector flow rate were made
to give a prearranged recorder reading with an emission line
from the matrix as an internal standard (silicon @ 251.2 nm) .

          4.1.11.9  Multi-element Application

     Applicable to all common metal pollutants.

          4.1.11.10  Physical Dimensions

     Not given, but spectrophotometer is a laboratory instrument
which could be made mobile.

          4.1.11.11  Unit Output
     Recorded as ppm of element in solid MgO.

          4.1.11.12  Safety Hazard

     High voltage power source; hot open flame.
                               162

-------
          4.1.11.13  Recommendations for Method Improvement

     Technique could be modified with suitable flow regulation
to permit direct introduction of stack gases or air for sampling.
The principal problem would be to prevent quenching of the
plasma.  Might want to use a different beryllium line to mini-
mize spectral self-absorption.

          4.1.11.14  References
     Dagnall, R. M., D. J. Smith, and T. S. West, "Emission
Spectroscopy of Trace Impurities in Powdered Samples With a
High-frequency Argon Plasma Torch," Anal. Chim. Acta 5*t, 397-
406 (1971).
                                163

-------
     ^.1.12  Comparison of D.C. Arc Type Plasma and High
             Frequency (RF) Induction Type Plasma as Spectro-
             scopic Emission Source

          4.1.12.1  Principle and Applicability

               4.1.12.1.1  Principle - The principles of two
types of high pressure plasma sources are described below.

     In the d.c.-arc plasma jet, a closed chamber contains at
one end an anode (or cathode) and at the other end a cathode (or
anode) that has a small opening in it.  A coolant fluid (argon)
is introduced tangentially through the chamber wall and swirls
around the chamber and out of the cathode hole.  When the arc is
struck the coolant fluid cools the outer layers of the arc so
giving a "thermal pinch" effect, which causes the arc column to
contract.  The resultant increase in current density gives a
higher arc temperature, and the pressure generated leads to the
ejection of extremely hot plasma through the cathode opening,
where it appears as a flame-like jet.  At higher operating cur-
rents, the arc suffers a further "pinch" effect called the
"magnetic pinch" due to the self-induced magnetic field.  For
spectroscopic purposes the substances to be analyzed is injected
into the arc column and is carried out in the plasma flame.

     In the high-frequency plasma torch, a stream of ionized gas
(argon) contained in a circular quartz tube, surrounded by a coil
carrying high-frequency alternating electric current, is heated
by induction.  Cold gas, being un-ionized, is not an electrical
conductor and therefore the plasma torch must be externally ini-
tiated, which can be achieved by holding a carbon rod in the
mouth of the quartz tube.  The high-frequency field heats the
carbon rod, which in turn heats and ionizes the argon.  Once the
main discharge has started, the carbon rod is removed and the
gas stream carries the ionized gas plasma down the tube away from
the coil, where it emerges at the tube mount as a "flame."  To
maintain the discharge, a portion of the ionized gas must be re-
cycled.  This is achieved by feeding the gas tangentially into
the tube which causes a vortex, and the partial reduction in
pressure at the center of this causes some of the ionized argon
to move back down the tube in the opposite direction to the main
gas flow.

     The quartz tube described in this paper is cooled by a
secondary stream of argon passing over it.  When fully coupled,
the high-frequency plasma can be regarded as a transformer, the
work coil being the primary and the ionized gas a one-turn
secondary.

-------
     The electron temperature of the d.c. arc plasma has been
measured as approximately 9500°K, whereas two electron tempera-
tures have been determined for the high-frequency thermal plasma
torch.  The extremely hot plasma region inside the cell has a
temperature range of 12,000 to 15,000°K, and the "tail flame"
region, where spectroscopic measurements are usually performed,
has an electron temperature of 8000°K.

               4.1.12.1.2  Applicability - Either technique can
be applied to the analysis of trace metals in samples of solids,
liquids and gases.  Both plasma sources have high stability and
sensitivity.  The major advantage of the high frequency plasma
is the lack of electrodes which minimizes contamination and back-
ground .

          4.1.12.2  Range and Sensitivity

     No absolute values of sensitivity or range are given.
Sensitivity is stated as comparable to that obtained with the
d.c. arc.

          4. 1.12. 3  Interferences

               4.1.12.3.1  Chemical - High excitation energy
available in these sources overcame depressive effects of phos-
phate and aluminum ions on calcium emission at 422.7 nm.  The
graphite electrodes in the d.c. arc plasma produce typical arc
type background; the high-frequency plasma-torch yields no in-
tense background spectrum other than the OH band system for
samples containing water.

               4.1.12.3-2  Physical - None reported or studied.

               ^.1.12.3.3  Memory Effect - No carry-over contami-
nation between samples if a 30-second wash with distilled water
was made between exposures.

          4.1.12.4  Accuracy, Precision and Stability
     Only limited data are available.  For the d.c. arc plasma,
the standard deviation for percentage transmission of aqueous
solution of 20 ppm calcium (3933 S) is ±0.12 with a coefficient
of variation of 4.5%.  Note - Considering the high level of back-
ground density in this wavelength region, the source stability
indicated by this figure is very good.  With the high-frequency
plasma torch, the standard deviation of the percentage trans-
mission for an aqueous solution containing 100 ppm calcium
(3933 ft) is ±2.3 with a coefficient of variation of 6.2%.  With
the calcium line at 4226 X, the standard deviation of the per-
centage transmission was ±1.9 with a coefficient of variation
of 4.95K.
                                165

-------
          M.I. 12. 5  Apparatus

               4.1.12.5.1  Spectrograph - (a) Hilger large-quartz
spectrograph; (b) Unicam SP900 flame spectrophotometer .

               4.1.12.5.2  D.C. Arc Plasma - Modified Margoshes
and Scribner plasma jet.

               4.1.12.5.3  High-Frequency Plasma Torch - Modified
Reed silica torch cell with 2.5 kW dielectric-heating generator
operating at 36 MHz.  Plasma torch was run at 1.5 kW output
power.

               4.1.12.5.^  Detector - Photographic plate.
               4.1.12.5.5  Ancillary Equipment - Argon gas (at
approximately 22 liters/minute), flow-meters and controlling
device, carbon and ignition.

          fr.1.12.6  Time Cycle for Sampling and Measuring

     D.C. arc plasma - 30-second exposures were taken on aqueous
samples with a pre-spray wash time of 30 seconds.  Unit was only
operated continuously up to 30 minutes.

     High frequency plasma torch - exposures were for 30 seconds.
System is operated continuously.  Test operation period was
8 hours .

          4.1.12.7  Calibration Procedure

     None given.

          *t.!.12.8  Method of Sampling and Sample Preparation

     Aqueous solutions were aspirated as an aerosol into the
plasma by a Unicam Atomizing unit.  Powder injector for high
frequency plasma torch is mentioned but  not described.

          4.1.12.9  Multi-element Application

     Applicable to all common metal pollutants.

          4.1.12.10  Physical Dimensions

     None given .

          H. 1.12. 11  Unit Output

     Calibrated for ppm metal in water.   Units depend on method
of calibration.
                               166

-------
          H. 1.12.12  Safety Hazards
     High voltage and hot open flame.
          4.1.12.13  Recommendations for Method Improvement
     Needs additional testing with solid particulate.
          1.1.12.11  Reference
     Greenfield, S., I. LI. Jones and C. T. Berry, "High-Pressure
Plasmas as Spectroscopic Emission Sources," Analyst 89, 713-720
(1961).                                             ~~
                               167

-------
4.2  Atomic Absorption, Thermal Emission, and Atomic
     Fluorescence Spectroscopy

     A number of very sensitive analytical procedures based on
digestion techniques, dissolution in water, and measurement of
atomic absorption, thermal emission, and atomic fluorescence
have been developed for the analysis of beryllium and cadmium
(ref. 22-37).  These techniques and others permit detection of
beryllium down to 0.03 ppm with repeatability of 2% of the
amount present and cadmium down to 0.001 ppm.  The maximum time
expenditure is 30 minutes.  The methodology is based on perform-
ing analyses on liquid phases.  Air particulate, collected on
filters and placed into solution after a digestion process has
been analyzed routinely for cadmium by atomic absorption pro-
cedures (ref. 30).

     The major limitation encountered when attempting to analyze
beryllium and cadmium simultaneously by conventional flame tech-
niques (i.e., non-plasma induced) is the incompatibility of the
two elements.  With atomic absorption, beryllium is best measured
in a very hot flame (acetylene-NaO reducing flame) whereas the
highest sensitivity for cadmium is attained with a cooler flame
(acetylene-air or hydrogen-air oxidizing flame).

     Although the highest sensitivity for beryllium is attained
by using the nitrous oxide-acetylene flame, Fleet, et al.
(ref. 33) observed signal depressant and enhancement effects
by a number of other metal ions in solution.  When present at
the 10,000 ppm level in water, aluminum (-43.4%), silicon
(-18.9%), and magnesium (-3.8%) reduced the signal intensity
for a 4 ppm beryllium solution.  Twenty-three other metal ele-
ments including cadmium, vanadium, iron, lead, calcium, and
potassium resulted in enhancement effects up to increases of
17%.  The addition of potassium to all samples produced more
uniform analytical results.

     Analyses of the individual elements can be improved by
using sheath gases to minimize chemical reactions in the flame
and background emission, but no single measurement with conven-
tional flames will yield both beryllium or cadmium analyses
with sufficient sensitivity.

     Most flame techniques use sample introduction methods which
provide for nebulization of a solution of the sample to form an
aerosol of sufficiently small particle size which can be atomized
in the flame.  In most cases, direct introduction of particulate
into the conventional flames results in inefficient atomization
or excitation.  White (ref. 38) devised an instrument, based on
the atomic absorption principle, for the immediate and continuous
measurement of cadmium fume under both laboratory and industrial
conditions.  The technique was used for cadmium fumes which by
                               168

-------
virtue of the low particle size (micron or submicron) would not
settle in the inlet of the burner or would be inefficiently
vaporized in the flame.  This method is reported in detail in
the following section.

     Robinson (NAPCA Contract CPA 70-176) has a development pro-
gram in progress for a flameless atomic absorption meter as a
continuous air monitor.  The program is based on earlier work
(ref. 39) and includes radio-frequency excitation of metallic
pollutants and the use of the atomic absorption phenomena to
measure selected and specific elements.
                               169

-------
     Jj.2.1  Cadmium Analysis - Atomic Absorption Method
            for Cadmium Fumes

          H.2.1.1  Principle and Applicability

               4.2.1.1.1  Principle - A portable instrument
based on the atomic absorption principle was devised which would
measure the cadmium (2288A) concentration in air.  The air
sample containing the cadmium fume was used to provide the
oxidizing medium for a propane-air flame.  Measurements were
performed continuously.

               4.2.1.1.2  Applicability - Continuous measure-
ments were performed on working atmospheres in the casting shop
of a manufacturer of cadmium copper (^0.8$ cadmium).  The method
worked well on metal fume defined as a suspension in air of
micron and sub-micron particles of metals or oxides from a con-
densed vapour.  The technique was developed for, and has been
used for, metal fumes which by virtue of the low particle size
will not settle in the inlet of the burner or will not be
inefficiently vaporized.  Apparatus was used in industrial
casting shops and foundries.  No data are reported for dust from
machining -applications.

          4.2.1.2  Range and Sensitivity

     The limit of detection was 0.005 mg/m3 for cadmium, but the
working range was generally 0.02 mg/m3 to 0.2 mg/m3.

          4.2.1.3  Interferences

               4.2.1. 3«1  Chemical - None reported.

               JJ.2.1.3.2  Physical - No specific details are
given, but particle size greater than 1 micron may settle out
or be inefficiently vaporized.  Modulation of light source is
required to minimize interference from light from the flame.

          4.2.1.*!  Accuracy, Precision and Stability
     Within the accuracy limits required for the program
(±10-15?), no change in calibration was noticed whenever hollow
cathode lamps were replaced or other adjustments were performed.

          4.2.1.5  Apparatus

               4.2.1.5.1  Source of Resonance Radiation - Hollow
cathode light source modulated by mechanical chopping or modula-
tion of power supply.
                               170

-------
               4.2.1.5.2  Burner - Modified Soudgaz cartridge
blow lamp; 5 cm long flame produced from propane-air mixture;
burner consumed air (sample) at a rate comparable to human
respiration.

               ^.2.1.5.3  Monochromator - Relatively simple
Hilger 292 grating spectrometer.

                          Photometer - A photomultiplier sensi-
tive to blue and ultraviolet region (RCA 1P28) and an a.c.
amplifier.

               ^.2.1.5.5  Readout System - Panel meter.

               4.2.1.5.6  Power Supplies -

     Photomultiplier - Ten 90 V batteries (Ever Ready V126).

     Hollow-cathode lamp - Ten 90 V batteries (Ever Ready V126)
in 2 parallel stacks giving 20 ma @ ^50 V.  With current drain
of 10 ma, a life of 20 hours is expected.

          4.2.1.6  Time Cycle for Sampling and Measuring

     Sampling and analysis are continuous.  Checks with burner
out of the optical path were made every few minutes to correct
drift of monochromator or lamp intensity.  Continuous operation
is restricted to 2 hours by propane gas cartridge and to 20 hours
by battery life.

          4.2.1.7  Calibration Procedure

     Apparatus was calibrated by introducing standard solutions
into a conventional atomic absorption nebulizer and spray chamber
to produce aerosols which were equivalent of the concentration
of cadmium in air of 0.02 mg/m3 to 0.2 mg/m3.  These aerosols
were fed to the burner and the absorption was measured on a meter
scale calibrated in terms of mg cadmium/m3 of air.

          4.2.1.8  Method of Sampling and Sample Preparation

     Air was sampled directly by the aspirating effect of the
blow lamp.

          4.2.1.9  Multi-element Application

     Instrument was applied also for measuring lead fume in air.
                                171

-------
          *l.2.1.10  Physical Dimensions

     Instrument was constructed as two units, measuring head
    cm x 25 cm x 15 cm) and battery and amplifier unit
    cm x 27 cm x 13 cm).  Total instrument weight was 25 kg.

          4.2.1.11  Unit Output

     Milligrams per cubic meter of air.

          4.2.1.12  Safety Hazard

     Open flame .

          4.2.1.13  Recommendation for Method Improvement

     Consideration should be given to applications involving
particles >1 micron.  The output, depending on particle size
and type of cadmium compound, needs to be better defined.
Potential application to other elements should be considered.
The read-out system needs improvement for unattended operation
and the gas supply volume should be increased for longer uninter-
rupted operation.  Better response could be obtained by using
signal integration system.

          4.2.1.14  Reference
     White, R. A., "Atomic Absorption Method for the Measurement
     of Metal Fumes," J. Sci. Instrum. 4£,  678-680 (196?).
                               172

-------
4.3  X-ray Emission Spectroscopy

     X-ray emission can be induced by application of a number
of excitation sources including electron beams, higher energy
x-rays, or alpha, beta, and gamma radiation from radio-isotopic
sources.  The simplest, less expensive, and mobile systems are
those based on x-ray and radio-isotopic sources.  Examples of
these techniques as applied to metal analyses of air pollutants
are included in this survey.

     In all cases, the routine application of x-ray emission
spectroscopy for monitoring air pollutants is restricted by
absorption of low energy x-rays to elements with atomic numbers
greater than 11.  The technique has greatest sensitivity for
elements in the atomic number range of 24 to 40.  Detection
limits vary for each element and are dependent on chemical and
physical properties of the matrix.  Enhancement and lowering
of intensity are common problems.

     The detection of beryllium by x-ray emission spectroscopy
at the levels encountered as an air pollutant are virtually im-
possible with current state-of-the-art techniques.  Measurement
of soft x-rays (e.g., emitted by beryllium) has been reviewed
by Holiday (ref. 40) and many of the problems and techniques
encountered in these measurements are described.  The appli-
cation of x-ray emission techniques for the direct analysis of
beryllium at the levels of stationary emission sources is not
practical.

     Alternate means for measuring beryllium indirectly have
been developed based on the quantitative reaction with a heavier
element which can be detected and measured more easily.  Pre-
cipitation of beryllium as the arsenate or phosphate and x-ray
fluorescence measurements of As or P yielded sensitivities of
1 yg to 40 yg (ref. 41).  Although the sensitivity is high, the
sample preparation techniques are too lengthy for rapid moni-
toring purposes.

     A markedly different concept for measuring light elements
by x-ray analysis has been proposed by Pichoir (ref. 42).  The
sample is analyzed by bombarding it with accelerated electrons
(other excitation sources would also be applicable), passing
the x-rays emitted from the sample through a pair of radiation
filters, and measuring the intensities of the x-rays.  The radia-
tion filters are filled with gaseous fluids whose absorption
characteristics are separated by only a very narrow wavelength
band.  During operation, the pressure of the gaseous fluids in
the filters is varied so that the intensities of the x-rays out-
side of the narrow wavelength band coincide.  This method can be
used to determine the concentration of beryllium, boron, carbon,
                               173

-------
nitrogen, oxygen and fluorine, but no specific details are given
for beryllium and no sensitivities are reported for any elements.

     Cadmium x-rays are more energetic than those of beryllium
and, therefore, can be used to measure low levels of cadmium
present in airborne dusts.  Several techniques based on x-ray
spectrographs or radio-isotopic emission have been reported for
measuring air-borne particulate collected on filters.  Although
electron beam and x-ray excitation of the characteristic cadmium
x-ray emission have been employed for analytical purposes, the
most promising excitation sources for mobile, relatively quick
and inexpensive, source emission monitoring are based on radio-
isotopes.

     X-ray emission or fluorescence dispersion techniques have
been used for measuring trace metals in air-borne dust; however,
no specific technique has been reported for cadmium.  Cares
(ref. *I3) comments briefly on a comparison of the response fac-
tors of CdKa (0.54 A), CdLa (3-96 8), and CdLB (3.74 A) and
matrix effects.  Additional general comments on the application
of dispersive x-ray emission analyses as applied to metal analy-
ses (down to 0.5 yg) air-borne dusts in industrial hygiene
studies are reported by Hirt, et al. (ref. 44).

     The cadmium L-beta (3-74 A) line is more sensitive than the
K-alpha (0.54 X), but matrix absorption, e.g., from iron, reduces
the intensity of the longer wave-length line by 405? and does not
affect the K-alpha intensity.  In addition, a critical thickness
for the L-alpha and L-beta radiations is reached at approximately
400 micrograms/cm2.  Also, higher counting losses can occur at
the higher wavelengths if particles are buried in the fibers of
the filter (ref. 43).

     Although no specific data are reported for cadmium, the
x-ray emission technique for measuring trace metals collected
on filters is characterized by problems with high blanks from
metals in the filter itself.  Ashless and membrane filters are
essentially free from contamination except for traces of iron,
potassium, and calcium.  Glass fiber filters have relatively
high iron and zinc blanks, and contain sufficient calcium,
potassium, and barium to create severe interference at or near
the principal lines of those elements.

     When using tungsten targets to excite x-ray fluorescence,
tungsten and copper reflections from the x-ray tube produce
intense interferences.  Their intensities vary depending on the
characteristics of the tube and the sample surface.  Low concen-
trations of mercury, arsenic, and selenium are difficult to
measure due to the intense tungsten reflections.
                               174

-------
      Relatively high  sensitivities  (down  to  0.5 ug) for trace
metals were  reported  (ref. 44)  on specimens  of air-borne dusts
collected  on filters.  Matrix problems related to absorption
of  secondary x-rays,  sample thickness and excessive particle
size  can affect the accuracy and precision of the measurement.
When  the total amount of sample exceeded  200 yg on a 2.4 cm
(diameter) filter, considerable reduction of the x-ray fluores-
cence was  observed.   Losses were attributed  to a "piling up
effect" of the deposit, so that reabsorption of the secondary
x-rays and lack of penetration  of the primary x-rays occurred.
To  attain  the high sensitivities, measurements had to be per-
formed in  vacuum or helium atmosphere to  minimize atmospheric
absorption losses.  (Note - Argon K-alpha may interfere with
CdL- if air path is  used.)

     A number of the problems related to  the application of x-ray
fluorescence for the quantitative measurement of trace metals in
particulate  can be minimized by proper selection of radioisotopic
excitation sources.   High resolution with semiconductor, nondis-
persive detector systems permit high sensitivity and specificity.
However, problems of interelement, sample thickness, particle
size, and spectral interferences must still  be considered.
Rhodes, et al. (ref.  45 or 46) reviews most  of the problems re-
lated to design and application of x-ray  emission analyzers using
radioisotope x-ray or gamma ray sources.

     Cadmium measurement over the concentration range of 0.001 to
1% in solution were reported by Hollstein, et al.  (ref. 4?).  By
using 2l|1Am  (10 mCi)  as the excitation source and a Si(Li)  semi-
conductor detector,  a detection limit of  0.001% of cadmium was
reached in a feasibility study.

     Under U. S.  Atomic Energy Commission Contract No.  AR-(40-1)-
4205j Rhodes (ref.  48) is developing technology related to radio-
isotopic XRP analysis of particulate air  pollutants.  Preliminary
results for cadmium show good instrumental precision, but data
are only semiquantitative as far as absolute quantities are con-
cerned.  Cadmium excitation of K x-rays by 1-125 is currently
being evaluated.   Rhodes, et al. (ref. 49,50) reported detection
limits for radioisotope induced x-ray fluorescence determinations
of 1? elements in ambient air particulate collected on cellulose
filters.   However,  no data are reported for  cadmium.
                               175

-------
*l.*l  Activation Analysis

     Activation analysis is widely used in measuring trace metals
where high sensitivity is required.  The greater proportion of
these measurements are performed with thermal neutrons because
of the availability of nuclear reactors and the large cross
sections of nuclides for thermal-neutron capture.  Sensitivities
of 10~9 are common depending on the matrix.

     Non-destructive analyses in relatively short irradiation
and counting times can be performed on a number of nuclides.
However, matrix effects can be severe and lengthy separation
processes may be required.  Severe difficulties occur when the
matrix contains substantial amounts of elements with large
thermal-neutron capture cross sections.  A form of sample
shielding results which limits the potential activation of the
elements present in the sample at lower concentrations.  In
addition, a number of elements (beryllium, carbon, nitrogen,
oxygen, iron and lead) are not highly activated with thermal
neutrons.

     A commercial activation analysis service reports general
sensitivities, based on thermal neutron activation analysis, of
15 yg for beryllium and 0.005 yg for cadmium.  However, in an
air particulate sample (14 mg) collected from urban air, beryl-
lium could not be detected and cadmium could not be measured
below 9.60 ug without radiochemical separation to eliminate
interferences.

     Thermal neutron activation analysis, based on neutron flux
from nuclear reactors is obviously not applicable to mobile
operation.  Although a mobile, power-free, maintenance-free
activation analyses system has been developed based on a 253Cf
spontaneous-fission neutron source (ref. 51), the technique is
not applicable to Be or Cd monitoring.  Interference-free analy-
ses are possible for 13 elements (0, Na, Al, Si, Mn, Fe, Ni, Cu,
Zn, Ag, Pt, Au, and Pb), but not for Be and Cd.

     Beryllium analyses with sensitivities of a few micrograms
can be performed with photon activation with gamma radiation
from radio-isotopic sources.  Photon activation analysis was
reviewed by Lutz (ref. 52), and selected techniques are evalu-
ated in the following section.

     Direct, interference-free analyses of cadmium on the average
air particulate matrices are not possible with activation analy-
sis techniques.  Lengthy separation procedures are required.
                                176

-------
     4.4.1  Automatic Beryllium In Air Monitor Based on the
            3Be(ainiY)C1't Reaction

          4.4.1.1  Principle and Applicability
               4.4.1.1.1  Principle - Beryllium-containing dust
was removed from air with a filter tape and exposed to an alpha
emitting radioisotopic source (polonium-210) to induce the
nuclear reaction, 9Be + a-»-n + Y+ 12C.  The yield of gamma
radiation is related to the beryllium content of the dust on
the filter tape.

               4.4.1.1.2  Applicability - The technique was
developed to monitor beryllium in air on a semi -continuous basis
and to yield an alarm signal when excessive concentrations were
reached.  The original specification for an alarm at 25, 2 and
0.01 pig of beryllium per cubic meter over periods of 30 seconds,
1 hour, and 24 hours, respectively, was not met.  A concentration
of 25 yg/m3 required 4 minutes for reliable alarming and a con-
centration of 2 yg/m3 required 60 minutes.

          4.4.1.2  Range and Sensitivity

     Both range and sensitivity depend largely on the time
allotted for sampling and for counting the y or neutron emission.
Four minutes were required to reliably detect 25 yg Be/m3.  At
the sampling rates used in the study, and the percentage of
beryllium actually being measured, the detection of 25 yg/m3,
is equivalent to measuring 4.3 yg of beryllium.  Statistically
less reliable measurements were obtained for absolute value of
beryllium at the 0.27 yg level.  No upper limit was defined.

          4.4.1.3  Interferences
               4.4.1.3.1  Chemical - None reported; technique
is generally considered free from chemical interferences.

               4.4.1.3.2  Physical - The y-detector must dis-
criminate against the attendant gamma radiation from the alpha
source.  Gamma ray energy from the beryllium reaction is 4.5 mev,
well above the energy of most gamma rays emitted directly from
the alpha source.  However, through the mechanism of gamma ray
"pile up," a background gamma ray counting rate is very much in
evidence (for polonium-210 sources) even though discrimination
of the detector is set to reject all energies below 4.0 mev.
The gamma ray "pile up" is dependent on both the energy and
intensity of the attendant radiation and the amount of lead
shielding employed between the reaction site and the scintilla-
tion counter.
                               177

-------
          4.*). 1.4  Accuracy, Precision and Stability

     The standard deviation at the 25 yg/m3 level is bQ% of the
beryllium count with sampling period of 2 minutes and counting
time of 2 minutes.  For the low level detection (2 yg/m3), the
standard deviation is 86% of the beryllium count with sampling
period of 2 minutes and counting time of 60 minutes.

     Long term stability is questionable since the stainless
steel foil window on the alpha source developed holes in a few
weeks time (average of 4 weeks).

          4.4.1.5  Apparatus
               *t.*l. 1.5.1  Alpha Source - Two 3-curie
Polonium-210 sources.

               4. 4.1.5. 2  Sampling System - Filter tape (Gellman
Instrument Co. plastic tape), tape drive mechanism, Vacuum pump.

               4.4.1.5.3  Detection System - Scintillation
detector, pulse height discriminator and analyzer, alarm mechan-
ism and power supplies.

          4.4.1.6  Time Cycle for Sampling and Measuring

     Time cycle is dependent on sensitivity required.  Four
minutes are required for reliable measurement of 25 yg/m3 of
beryllium and 60 minutes are necessary for measuring 2 yg/m3 .
Sampling time is generally 2 minutes.

          4.4.1.7  Calibration Procedure
     Calibration of the beryllium monitor was achieved by ana-
lyzing tape samples chemically after measuring the gamma emission
with the photo-activation system.  Ten percent beryllium oxide
in "Arizona Road Dust" was used in an aerosol generator to pre-
pare standard air samples which were sampled by the filter tape
system.

     The calibration curve was linear from 5-0 yg to 87-0 yg.

          4.JJ.1.8  Method of Sampling and Sample Preparation

     Samples of air (79 liters/minute) were drawn through a
2-inch intake onto a 2 1/2-inch wide plastic, filter tape
(Gellman Instrument Co.) for two minutes.  The spot containing
the specimen was advanced mechanically to the a-radiation and
scintillation detector for measurement.  The plastic tape samples
by two mechanisms:  (a) conventional filtering action through
small pore sizes and (b) electrostatic collection of fine par-
ticles on its surface.
                               178

-------
          4.4.1.9  Multi -element Application

     Designed only for beryllium.

          4.4.1.10  Physical Dimensions

     No specific details are given on the final prototype.
Photographs (8) are shown.

          4.4.1.11  Unit Output
     Principally designed for visual and audio alarming.  How-
ever, output can be related to yg/m3 .

          *t. 4.1.12  Safety Hazard

     Moderate voltages are used for electronic power supply.
Deterioration of the source window can result in serious radia-
tion hazard.

          4.4.1.13  Recommendation for Method Improvement

     The background of gamma rays emitted from the alpha source
must be minimized.  By using Pu-238 or one of the newer developed
Curium alpha sources, lower attendant gamma radiation could be
obtained.  Better discrimination between signal and background
would be possible.

     Better source design based on the current state of the art
should minimize the source window problem and loss of radioacti-
vity.  A safe alpha source capable of giving sensitivity in the
range of at least 30 to 60 counts per minute per microgram of
beryllium is needed.

     Improved resolution and response of the newer semi-conductor
detectors would help minimize background problems and improve
sensitivity.

     A larger source (at least 2 inches in diameter) would
improve the efficiency of excitation.  (Note - Source used in
this study irradiated only 55% of the available beryllium. )

          *J. 4. 1.1*1  References
     Braman, R. S., 1963, "Research and Development of An
     Automatic Beryllium-in-Air Monitor,"  RTD-TDR-63-1H2,
     Air Force Systems Command, Edwards AFB, California.

     Braman, R. S., 1962, "Research and Development of An
     Automatic Beryllium and Boron Monitor," Armour Research
     Foundation Final Report No. ARF 3203-3 for Air Force
     Flight Test Center, Edwards AFB, Calif.
                               179

-------
        . 2  Determination of Beryllium by the Photoneutron
            Method With the (y,n) Reaction

          4.4.2.1  Principle and Applicability

                          Principle - Solutions containing
beryllium were analyzed by (y ,n) reaction involving photodisin-
tegration of the beryllium nuclide.  The sample is irradiated
with gamma rays from 121*Sb and the photoneutrons produced in
the reaction are counted and related to the beryllium content.

               *!.*!. 2. 1.2  Applicability - The technique was
applied to solutions of mixed fluoride salts, mixed beryllium
and magnesium oxides with or without thorium oxide, and beryllium-
copper alloys.  In all cases the sample was placed into solution.

     Although the method was designed for the analysis of solu-
tions, the principle could be applied with suitable modification
to monitor beryllium deposited on a filter tape mechanism.

          4.4.2.2  Range and Sensitivity
               *t . *1 . 2 . 2 . 1  Range - 0.1-3 mg of beryllium/ml.

               4.4.2. 2. 2  Sensitivity - Minimum of 1.5 mg of
beryllium was necessary to yield a relative standard deviation
of ±1%.  Although not  reported, a lower detection limit could
easily be reached by accepting a larger standard deviation and
altering instrument  parameters.

          4.4.2.3  Interferences
                          Chemical - Potential interferences fall
into two classes:  elements having high thermal or epithermal
neutron cross sections which would tend to absorb photoneutrons,
and elements having high cross section for the (y »n) reaction
and might also release photoneutrons.  Cadmium, which has a high
thermal-neutron cross section (o = 2500 barns), can produce a
relative error of approximately 5% , depending on the Cd/Be ratio
and the volume of solution.  [Note - The interference is greater
with larger solution volumes because water moderates the photo-
neutron (20 k.e.v. average energy).]  A cadmium shield (0.1 in.)
placed between the sample and detector screens out thermal
neutrons with an attendant loss (155?) of total signal, but elimi-
nates error from neutron absorbers.  Only a few elements (Cd,
Sm, Eu, Gd) have sufficiently high cross sections to interfere.
Elements tested in a 20:1 ratio to Be and found not to interfere
were Li, K, Cu, Zn, Ba, Al, Zr, Sn, Pb, Bi, Fe, Co, Ni, Th, V,
and B.
                               180

-------
     Deuterium is the only isotope other than 9Be that is capable
of undergoing a (y»n) reaction with 12lfSb gamma irradiation.
A 24:1 weight ratio of D20 to Be would cause a 1% relative error.
(Note - A gamma source with a greater flux S6Co would increase
response to DzO by two orders of magnitude.)

               4.4.2.3.2  Physical - None reported.

          4.4.2.4  Accuracy » Precision and Stability
     Statistical data relating accuracy as measured by comparison
to other techniques (colorimetry ) are not given, but analyses are
shown to yield comparable values.

     The relative standard deviation of duplicates was ±1.156.

     Long term stability is questionable since a 400 me 12l|Sb
(123Sb irradiated) source is useful for about 2 half -lives
(120 days).

          4.4.2.5  Apparatus
               4.4.2.5.1  Isotopic Source - 40 mg of 9&% enriched
123Sb irradiated for 14 days to produce 400 me of 12"Sb.

               4.4.2.5.2  Detector - Large BFa neutron counter
tubes (2-inch), high voltage power supply, linear amplifier,
sealer, timer, and lead shielding.

               4.4.2.5.3  Ancillary Equipment - Polyethylene
moderator, sample cells.

          4.4.2.6  Time Cycle for Sampling and Measuring

     Approximately 10 minutes required for loading and obtaining
sufficient counts.

          4.4.2.7  Calibration Procedure

     Solutions of known concentration of beryllium are counted.

          4.4.2.8  Method of Sampling and Sample Preparation

     Samples are placed into solution with suitable reagents and
5-, 25-, or 50-ml aliquot s are counted in the apparatus.

          4.4.2.9  Multi -element Application

     Specific to beryllium with 12l|Sb irradiation, but can be
applied to deuterium analysis by using 56Co.
                               181

-------
          4.4.2.10  Physical Dimensions

     Electronics occupy 4-ft rack and the irradiation cell is
a 12-in. x 12-in. cylinder.

          Jl.4.2.11  Unit Output

     Output appears as sealer counts which are relatable to the
quantity of beryllium either as mass or mass/volume units.

          4.4.2.12  Safety Hazard

     Electronics require moderately high voltage.

     Radiation hazard depends on shielding.  Operator is exposed
to 100 mr/hr on hands and 50 mr/hr at head level in changing
samples, but total exposure is less than 1 hr/wk.  Since source
activity is about 20 r/hr at 1 foot, installation must be done
carefully in few seconds with 3-foot tongs.

          ^.^.2.13  Recommendations for Method Improvement

     Technique, as presented in the publication, is not directly
applicable; however, the concept can be modified to provide
intermittent, <10-minute time span, measurements.  Newer detec-
tion systems (semi-conductors or scintillation counters) could
increase sensitivity.  Consideration should be given to other
isotopic sources.  Sensitivities of a few micrograms are
possible .

          4.4.2.14  Reference
     Goldstein, G., "Determination of Beryllium by the
     Photoneutron Method," Anal. Chem. 35., 1620-1623 (1963)

     Lutz, G. J., "Photon Activation Analysis - A Review,"
     Anal. Chem. ±3_, 93-103 (1971).
                               182

-------
     4.4.3  Determination of Beryllium on Air Filters With
            the (ot,n) Reaction

          4.4.3.1  Principle and Applicability

               4.4.3.1.1  Principle - The (a,n,y) reaction was
used to measure the beryllium content (10 to 15 Wg) on impreg-
nated air filters.  Three separate detection schemes were
employed:  (1) a coincident neutron and gamma system which would,
in theory, eliminate background; (2) fast neutron detection with
plastic scintillators; and (3) thermal neutron detection using
moderated BFs tubes.

               4.4.3.1.2  Applicability - Technique was estab-
lished to measure the beryllium content on air filters.  Due to
escape of radioactivity from the source, the technique was not
completely evaluated.

     The three detection techniques were devised to minimize
interferences from background, neutron and/or gamma radiation,
signals.  It was hoped that simultaneous measurements of both
neutrons and gamma radiation produced in the (a,n,y) reaction
channeled through a coincidence gate circuit could eliminate
effects of other light elements trapped on the filter.  However,
the sensitivity for detecting the 4.4 MeV photon, which was
supposed to occur in about 60% of the reactions was insufficient.

     By mounting a second fast neutron detector in place of the
Y radiation detector, the output signals were channeled through
an anti-coincidence gate to reduce background effects.  The
system was moderately effective, but high background from the
Po-210 source limited the beryllium detection sensitivity.

     The result of the fast neutron detection system were sub-
stantiated by a thermal neutron counting system.  Although the
count rate with the thermal neutron system was higher, the
neutron interference from the Po-210 source limited the sensi-
tivity to 10 yg.

          4.4.3.2  Range and Sensitivity

     With the Po-210 source available to the authors, it was not
practical to detect <10 yg of beryllium and the desired <1 pg
was not reached.  No other data are reported.

          4.4.3.3  Interferences

               4.4.3.3-1  Chemical - Small quantities of carbon,
oxygen, aluminum, sodium, etc. impurities in the filter or the
Po-210 source produced high background of neutron flux.
                                183

-------
                          Physical - No interferences based
on physical properties of the sample are reported.
          4.4.3.4  Accuracy, Precision and Stability
     No accuracy or precision data are reported.  The seal on
the polonium source deterioriated over a period of 3 months,
resulting in escape of radioactivity.
          ty.fr. 3. 5  Apparatus
               *>.*!. 3 .5.1  Isotopic Source - Alpha source
(3.11 Ci Po-2!OT
                          Detectors - Nal crystal, photomulti-
plier tube and fast neutron detector; BF3 thermal neutron
counting system moderated with paraffin.
               **.**. 3. 5. 3  Ancillary Equipment - Sealers, shield-
ing, filtering system, electronic amplifiers and power sources.
          4.4.3.6  Time Cycle for Sampling and Measuring
     None given; estimated <10 minutes.
          4.4.3.7  Calibration Procedure
     None given .
          4.4.3.8  Method of Sampling and Sample Preparation
     Beryllium dust was collected on filter paper and exposed to
alpha radiation, but no specific details are given.
          4.4.3.9  Multi-element Application
     Principally designed for beryllium, although neutron flux
is generated by other light elements.
          4.4.3.10  Physical Dimensions
     None given .
          4.4.3.11  Unit Output
     Signal convertible to yg Be.
          Jt.fr. 3. 12  Safety Hazard
     High voltages for electronic power supplies.  High radia-
tion hazard due to leakage of Po-210 through cover seal.
                                184

-------
          4.4.3.13  Recommendation for Method Improvement

     Better source design and the use of newer detection systems
(semi-conductors) with higher resolution and sensitivity would
improve the technique and permit the possible application as a
beryllium monitor.

          4.4.3.14  Reference

     Buch, W. L. and W. G. Spear, "Measurement of Beryllium
     on Air Filters," AEC Contract No. AT(45-1)-1350, General
     Electric Co., Hanford Atomic Products Operation, Richland,
     Wash., 4 June
                                185

-------
     *l.*t.*J  Beryllium - Determination by a Gamma Activation-
            Photoneutron Method

          H.4.4.1  Principle and Applicability

               *K*I. 4.1.1  Principle - Beryllium, in beryllium
metal, beryllium oxide, and beryllium compounds, is determined
by gamma activation of aqueous solutions or powder with a
Sb-124 radio-isotopic source.  The neutrons generated in the
9Be(y,n)8Be •* 2lfHe reaction are measured with helium-3 neutron
detectors.

               it.Jj.lt.1.2  Applicability - Technique was developed
principally as a short, precise, and economical method for mea-
suring moderate amounts of beryllium as a major component.  The
technique presently is not designed for monitoring of relatively
small amounts (<100 mg) of beryllium.  However, the concept may
possibly be modified for greater sensitivity.

          fr.fr.fr. 2  Range and Sensitivity

     A linear calibration curve is shown for the working range
of 100-1600 mgs of beryllium with counts of 0.5-6 x 106 per
5 minutes.

          fr.fr.fr. 3  Interferences

               fr.fr.fr.3.1  Chemical - None reported.

               fr.fr.fr.3.2  Physical - There was no significant
bias in the analysis provided the powders had very similar com-
positions and particle sizes so that they occupied the same
volume.

          fr.fr.fr.fr  Accuracy, Precision and Stability

     Precision of a single analysis at the 95% confidence level
is ±0.25? of the value.  The analyses do not differ significantly
at the 98% beryllium content from results obtained by either a
gravimetric or volumetric method.  Source must be irradiated
every six months to generate Sb-12fr.

          fr.fr.fr. 5  Apparatus

               4.4.4.5.1  Gamma Activation Source -
500-millicurie Antimony-12fr.

               4.4.4.5.2  Analyzer - Lead pig (9" OD x 12" H)
with inside cavity (?" ID x 7-1/2" H); two He-3 neutron de-
tectors; electronic amplifiers and sealer.
                                186

-------
               *t. *<.**. 5. 3  Automatic Sample Charger coupled to
the counting equipment .

          *t.*t.*t.6  Time Cycle for Sampling and Measuring

     Counting is accomplished in five minutes for powdered
samples .

          4.4. 4.7  Calibration Procedure

     Calibration standards were prepared from beryllium metal
(NBL F-224 from AEC, New Brunswick Laboratory, N. J.) and from
beryllium oxide powder (L-33 from Kawecki Berylco Industries,
Inc., Reading, Pa.).

          *t. 4.4.8  Method of Sampling and Sample Preparation

     Samples were prepared either by dissolving the beryllium
sample in hydrofluoric acid, or by forming a compressed pellet.
With the latter technique, the powders were compressed into
stainless steel dishes (I11 OD x 5/16" H) under a pressure of
about five tons to form a pellet approximately 0.225-inch thick.

          4.4.4.9  Multi-element Application

     No others mentioned.

          ^.4.4.10  Physical Dimensions

     No overall specifications are given.  The lead pig is
9" OD x 12" H.

          4.4.4.11  Unit Output
     Milligrams of beryllium.

          4.4. 4. 12  Safety Hazard

     Moderate electrical voltages are used.  Potential radio-
activity hazard exists, but through automatic sample changing
is minimized.

          4.4.4.13  Recommendation for Method Improvement

     Technique, as currently designed, is not applicable as an
intermittent analyzer of filter paper samples of collected par-
ticulate .   It is not obvious that the instrument is operating
at maximum sensitivity.  A trade-off of precision and improvement
of the preamplifier and amplifier components may yield a more
sensitive  device which could be applied to analysis of filter
tape specimens.
                               18?

-------
     fr.fr.fr.m  Reference

Kienberger, C. A., "Determination of Beryllium by Gamma
Activation," Report Y-1733, U.  S. Atomic Energy Commission,
Union Carbide Corporation, Oak  Ridge, Tenn.
                          188

-------
     4.^.5  Beryllium - Continuous Monitor for Air-borne
            Beryllium Based on the "BeCa^n^y) 1ZC Reaction

          *!.*>. 5.1  Principle and Applicability

               4.1.5.1.1  Principle - A filter fabricated from
sintered metal and incorporating an alpha emitting material
dispersed throughout the filter is used to collect beryllium
particulate and to measure the quantity of beryllium and the
rate at which the beryllium is collected.

     The beryllium particles collected by the filter react with
alpha particles in the filter to yield neutrons and gamma radia-
tion by the reactions:

          9Be + a •*• n + 12C (excited state)

          12C (excited state) -> y + 12C (ground state)

     The yield of gamma photons and/or neutrons is measured and
related to the quantity of beryllium collected by the filter.

               *!.*>. 5. 1.2  Applicability - Technique can be used
to provide a continuous quantitative measurement for beryllium
in air or to operate an alarm mechanism to indicate an excessive
concentration of beryllium in the atmosphere.  The system Is
restricted in application as an occupational health monitor in
a relatively clean atmosphere where the expected venting of
particulate would be principally beryllium or its compounds.
Exposure of the filter to high volume of non-beryllium particu-
late would cause plugging of the filter and attendant loss of
efficiency and ultimately loss of sensitivity and response
capability.

          4.4.5.2  Range and Sensitivity
     No data reported, but sensitivity estimated as less than
  yg/m3.

          4.4.5.3  Interferences
                  . 5. 3.1  Chemical - None reported.

                  .5.3.2  Physical - Presence of excessive
amounts of non-beryllium particulate could cause plugging of
filter and subsequent loss of response.                     ,

          ^.^.5-^  Accuracy, Precision and Stability

     No data reported.
                               189

-------
                   Apparatus
               *t.*l.5.5.1  Integrated Filter and Alpha Source -
Filter made from sintered metal such as platinum or its oxide,
or from a controlled alloy or a metalloceramic .  An alpha
emitting material such as polonium-210, curium-242, radium or
Plutonium is incorporated into the filter to provide a source
of alpha particles integral with the filter and dispersed
throughout.  As an alternative, the alpha particle source may
be plated or electrodeposited upon the filter.

               *t. *l.5.5. 2  Detection System - Sodium iodide
crystal coupled with a photoelectric tube; electronic counter
to read total beryllium collected; electrical differentiating
to indicate rate at which air-borne beryllium is collected;
alarm system.

          4.4.5.6  Time Cycle for Sampling and Monitoring

     Instantaneous and continuous.

          4.4.5.7  Calibration Procedure

     No specific details are given.  Procedure would involve
sampling known quantity of beryllium as suspended particulate,
as an aerosol, or as a beryllium compound in solution.

          4.4.5.8  Method of Sampling and Sample Preparation

     Air sample is drawn directly onto the filter by means of
a blower connected to the air housing outlet.  No sample prepa-
ration is required.  Sample could be any fluid media.

          4.4.5.9  Multi-element Application
     May be used to detect lithium, boron and fluorine.

          4.4.5.10  Physical Dimensions

     No details on size or weight are given.   However, unit
should be very compact and small .

          4.4.5.11  Unit Output
     Data should be convertible to mass, mass/unit volume, and
rate of mass change.

          4.4.5.12  Safety Hazard
     Moderate electrical voltages.  Potential radiation hazard
depending on efficiencies of sealants.
                               190

-------
          *l.*t.5.13  Recommendation for Method Improvement

     Technique should be applicable to monitoring for beryllium
in a relatively clean atmosphere, but the operational lifetime
in a dirty atmosphere is questionable.  Consideration of
replaceable filter costs is paramount.

          l>.4.5.1fr  Reference
     Lamb, I. E., "Monitor for Air-borne Beryllium,"
     U.S. 3,291,896 (1966).
                                191

-------
*t.5  Alpha Particle Scattering

     Light elements can be detected by the back-scattering of
a beam of alpha particles issued from a radioactive source.  This
principle has been applied to the detection of light elements
on lunar probes and may be applied to air particulate analysis
of filter collected specimens.  A combination of radioisotopic
induced x-ray fluorescence (heavy elements) and alpha-particle
scattering (light elements) could provide a possible approach
for air particulate analyses.

     Some of the advantages and disadvantages of alpha particle
scattering measurements are shown in the following evaluation
reports .

     4.5.1  Alpha-Scattering Technique of Chemical Analysis

          4.5.1.1  Principle and Applicability
               4.5. 1.1.1  Principle - The back-scattering of
a monochromatic, collimated beam of alpha particles generated
in a thin radioactive source is used as a non-destructive ana-
lytical method to determine the elemental composition of solid
specimens or particulate.  The intensity of scattering is deter-
mined primarily by the square of the nuclear charge .   For
elements heavier than aluminum and for relatively low energy
alpha particles, the large-angle scattering is primarily
Rutherford scattering.  For light elements such as carbon and
oxygen, and particularly with rather high-energy alpha particles
(5-80 Mev from 2l"*Cm), nuclear effects enhance the scattering
above that predicted from pure Rutherford scattering.

               4.5.1.1.2  Applicability - The technique has
been applied for measuring light elements in rocks and on lunar
missions.  It is best used in a vacuum and is applicable to the
study of surfaces (1 to 100 y thick).  The energy spectra of the
elastically back-scattered alpha particle and spectra of protons
produced by a,p reactions have been used to determine the com-
position of lunar rocks.

     Since a single mass scatterer produces a continuous spec-
trum, an observed spectrum from a surface containing several
elements must be decomposed into components.  The high-energy
end points characteristic of each mass number are unaffected
by the chemical or physical state of the scatterer.

     The technique can be applied for analyzing particulate on
filter paper.  There are some uncertainties on thickness and par
ticle size effects.  Although the technique has been applied to
                               192

-------
light elements, no specific data for beryllium are reported.
There is some question concerning interacting effects of the
9Be(a,n,y) 12C reaction.

     The method has good resolution for light elements, but
resolution becomes poorer as the atomic weight increases (Fe,
Co and Ni cannot easily be resolved).

     Using practical source intensities (ca. 100 me), the rate
of analysis is slow.  A complete analysis requires about one day.

          *t.5.1.2  Range and Sensitivity

     Sensitivity is reported as high, but no data given for
beryllium.

          4.5.1.3  Interferences
               *l. 5. 1.3-1  Chemical - The continuous spectra
overlap and must be analyzed mathematically based on a library
of spectra from pure elements.  Such a library includes the
response to all elements between boron and titanium (except
neon, argon, and scandium) and to selected heavier elements
(iron, barium and gold).

               4.5.1-3-2  Physical - Response is affected by
surface roughness, sample thickness, and particle size.

          4.5.1.*t  Accuracy, Precision and Stability

     Results for most of the elements indicate absolute accu-
racies of about 1 atomic % .

     System operated very well on lunar missions and has func-
tioned on earth for two years.

          4.5.1.5  Apparatus

               4.5.1.5.1  Alpha Particle Source - 100 me, 2l*2Cm.

               4.5.1.5.2  Detector - Silicon semiconductor
detectors, and suitable electronics; pulse-light analyzers.

          4.5.1.6  Time Cycle for Sampling and Measuring

     Measurements are performed with counting times varying from
20 to 2^53 minutes.  Based on lunar mission data, total proces-
sing of sample and manipulation of mathematical treatment of data
can take one day.
                               193

-------
          4.5.1.7  Calibration Procedure

     Calibration is based on a library of data for pure
elements, mathematical adjustment of response factors, and
monitoring of a standard nuclide 25lfEs.

          4.5.1.8  Method of Sampling and Measuring

     Samples of airborne particulate are collected on filter
paper and exposed directly to alpha radiation.

          4.5.1.9  Multi-element Application

     Response is observed to all elements between boron and
titanium (except neon, argon, and scandium)  and to selected
heavier elements (iron, barium and gold).  Because of poorer
resolution with heavier elements, measurements reported for
calcium include calcium and potassium; iron  values denote ele-
ments titanium through zinc; barium analyses cover all elements
heavier than zinc.

          4.5.1.10   Physical Dimensions

     Lunar device weighed <4 kg.  The "head  unit" containing
the excitation source, detectors and part of the electronics
was 13.3 x 17.1 x 11.7 cm and was connected  by cable to an elec-
tronic package of unreported size on the Surveyor spacecraft.
(Note - Due to lack of atmosphere on the moon, no vacuum system,
which would be required for earth-bound  operation, was included.)

          4.5.1.11   Unit Output

     Signal converted to atomic %t but could be converted to
mass units, if desired.

          4.5.1.12   Safety Hazard

     Moderately high voltage for electronics.  Potential radia-
tion hazard, but system design appears to minimize the problem.

          4.5.1.13   Recommendation for Method Improvement

     If sensitivity requirements can be  met  with shorter count-
ing periods and if  the time and cost required for reducing the
data can be minimized, the technique could be used for lighter
elements.  The application to beryllium is questionable based
on potential interference from the 9Be(a,n,y)12C reaction.
                               194

-------
4.5.1.1**  References

Turkevich, A. L., "Chemical Analysis of Surfaces by Use
of Large-Angle Scattering of Heavy Charged Particles,"
Science 131*, 672-67^ (1961).

Patterson, J. H., A. L. Turkevich and E. Franzgrote,
"Analysis of Surfaces Using a-Particles," J. Geophys.
Res. 70.* 1311 (1965).

Turkevich, A. L., K. Knolle, R. A. Emmert, and W. A.
Anderson, "Instrument for Lunar Surface Chemical Analysis,"
Rev. Sci. Instr. 37., 1681-1686 (1966).

Turkevich, A. L. and K. Knolle, "Chemical Analysis Experi-
ment for the Surveyor Lunar Mission," J. Geophys. Res. 72,
831-839 (1967).                                         ~~

Economou, T. E., A. L. Turkevich, K. P. Sowinski, J. H.
Patterson, and E. J. Franzgrote, "The Alpha-Scattering
Technique of Chemical Analysis," J. Geophys. 75, 651^-6523
(1970).
                           195

-------
     *t.5.2  Beryllium - Measurement by Alpha-Particle Scattering

          4.5.2.1  Principle and Applicability

               ^.5.2.1.1  Principle - An a-particle, when de-
flected in a close encounter with a nucleus, loses kinetic
energy equal to the recoil energy of the scatterer and the loss
of energy can be related to the mass of the scatterer and the
angle through which the a-particle is deflected.  A measurement
of energy loss and deflection of particles scattered from a
sample of material provides the basis of an atomic analysis of
the material.  By using isotopic excitation sources and surface
barrier detectors, a mobile, compact apparatus was produced.

               ^.5.2.1.2  Applicability - By working in a modest
vacuum or in hydrogen at atmospheric pressure, relatively high
sensitivities can be attained for a number of elements, particu-
larly those with low atomic numbers.  Resolution of adjacent
elements is possible for some, but not all elements.  The appa-
ratus described in the report produced the resolution and
sensitivities (minimum sample size) tabulated below.

                                 Minimum Sample Size
             Element             	(yg/cm )*	

          Be                              1
          B                               2
          C                               0.2
          N, 0                            2
          P                               4
          Na                              8
          Mg, Al                          7
          Si, P, S                        6
          Cl, Ar, K, Ca                   5
          Sc to Zn                        4
          Ga-Sn                           3
          >Sn                             2
   *Target of about 4 cm
                        2
     Although the sensitivity for beryllium is extremely good,
the measuring (counting) times necessary to attain this response
are lengthy (1000 minutes) and are not applicable to a realistic
monitoring device.

          4.5.2.2  Range and Sensitivity

     No specific data are given for the range of beryllium con-
tent beyond the minimum sample size of 1 yg/cm2 (total on 4 cm2
target of 4 yg of beryllium).  Based on the limitation of sample
                               196

-------
thickness, a beryllium sample up to 150 yg/cm2 (total of 600 yg
on a target of 4 cm) would be the maximum amount which would
yield distinct peaks for adjacent elements.

          4.5.2.3  Interferences

               ^.5.2.3.1  Chemical - None reported.

               4.5.2.3.2  Physical - To resolve the backscattered
spectrum for adjacent atomic numbered elements, the sample thick-
ness must be less than 50 yg/cm2 for elements around calcium
(A=40) and less than 150 yg/cm2 for elements around oxygen (A=l6),
The spectrum from a thick target is a superposition of peaks
from each infinitesimal layer in the sample and the alpha par-
ticles scattered from progressively deeper layers are attenuated
more and more before leaving the sample.  The superposition
results in a continuous spectrum which is relatively flat with
a sharp cutoff at the characteristic energy (a step display),
whereas the spectrum from the thin film is characterized by in-
dividual peaks (a peak display).

          4.5.2.4  Accuracy, Precision and Stability
     No data reported for accuracy and precision.  Erratic
deterioration of the protective surface film on the source was
observed.  Lifetimes range from 7 days to 2 months.

          4.5.2.5  Apparatus

               4. 5. 2. 5.1  Alpha Source - 0.5 me Cm-2^2 source.

               4.5.2.5.2  Detection System - Surface barrier
detector, preamplifier, multichannel analyzer, recorder.

          4.5*2.6  Time Cycle for Sampling and Measuring

     Counting times for 1 yg/cm2 sensitivity were 1000 minutes.

          4.5.2.7  Calibration Procedure

     Standard samples were prepared from known quantities of
beryllium.  No specific details are given.

          ^.5-2.8  Method of Sampling or Sample Preparation

     No specific details are given.  Specimens were mounted on
collodion filma poly-vinylacetate-chloride and A1203-
                               197

-------
          4.5.2.9  Multl-element Applications

     Technique can be applied to light elements with unit atom
number resolution, e.g., Be, B, C, F, Na,  and to heavier ele-
ments in combinations of two or more, i.e.,  N,0; Mg,Al; Si,P,S;
Cl,Ar,K,Ca; Sc to Zn; Ga to Sn; >Sn.

          4.5.2.10  Physical Dimensions

     Specific dimensions are not given.  Source and detector
head is small, being less than 12 inches in  diameter.  Elec-
tronics would occupy approximately 4-ft standard rack.

          4.5.2.11  Unit Output

     Micrograms of beryllium.

          4.5.2.12  Safety Hazard

     Moderate electrical voltages are used in electronics.
Deterioration of protective covering on alpha source presents
potential radioactivity hazard.

          4.5-2.13  Recommendations for Method Improvement

     The lengthy counting times (1000 minutes) preclude appli-
cation of the technique in present form as a short-term,
immediate response monitoring device.  Major improvements in
the detection system and the irradiation source are necessary
to maintain high sensitivity, but allow shorter analysis times.

          4.5.2.14  Reference

     Semmler, R. A., J. P. Tribby, and J.  E. Brugger, "Applica-
     tion of Alpha-Particle Scattering to  Chemical Analysis,"
     COO-712-89, USAEC Contract AT(11-1)-712, October 1964.
                               198

-------
4.6  Mass Spectrographic Analysis

     High sensitivity and selectivity for metals are attainable
by ion- or spark-source mass spectrometric techniques.   However,
analyses must be performed in a relatively high vacuum  and,  as
a consequence, the measurements are made on a batch basis and
require considerable time between samples to attain suitable
conditions for efficient excitation.  The technique is  limited
to measuring samples collected on a filtering device and is  not
applicable to direct and continuous source monitoring.

     Additional problems related to low ionization efficiencies
and spectral interferences limit the application of mass spec-
trographic analyses for beryllium and cadmium.  Depending on the
energy of the ionization source, some difficulties in attaining
sufficiently high ionization efficiencies can be encountered.
Generally, thermal ionization sources are inadequate and an  elec-
tron impact or other ion sources are required.  With thermal
ionization, low efficiencies are encountered with cadmium,
nickel, copper, zinc, silver, lead, tin, germanium, and silicon;
whereas beryllium, phosphorus, sulphur, arsenic, selenium,
tellurium, osmium, gold, polonium and palladium may not be
observed (ref. 53).

     With spark source excitation, considerable multi-ion frag-
ments can be produced.  Triply-ionized aluminum (27A13+) and
doubly-ionized oxygen (1802+) can interfere with the beryllium
measurements.

     Brown and Vossen (ref. 54) were able to measure air pollu-
tion particles collected on special nitrocellulose filter pads
for elements in the concentration range of 0.004 yg/m3  to
40 yg/m3 by spark source mass spectrometry.  With an electrical
read-out system, a survey of all elements from atomic number
92 (uranium) to atomic number 3 (lithium) can be made in a
scanning time of 9 minutes.  The operations involved in changing
specimens and re-establishing operating conditions take approxi-
mately 10 minutes.  Combustion of the filter, addition of
internal standard (silver nitrate), and preparation of the
sample with graphite in the form of a conducting electrode takes
approximately two hours.  Results for 28 elements, but  not for
beryllium or cadmium, were reported.

     Analyses for cadmium and 61 other elements (but not beryl-
lium) were performed by Morrison and Kashuba  (ref. 55)  on
simulated lunar samples by spark source mass spectrometry.  The
technique, although lengthy, provided for the detection of
0.3 ppm cadmium.  The 9Be+ ion was obscured by  1802+ and 27A13+
and therefore could not be measured.
                                199

-------
     An ultra-sensitive (10 12 g of Be) mass spectrographic tech-
nique, based on the formation of Be ion emission from a tungsten
filament coated with the sample, was developed by McHugh and
Sheffiend (ref. 56).  Although very sensitive, the technique
lacked analytical precision and the time required for preparing
samples limits the application to very special situations.

     Relatively high sensitivity (submicrogram quantities) for
beryllium, as metal chelates, has been attained by specialized
mass spectrographic techniques.  Majer, et al. (ref. 57) measured
beryllium as the oxinate; Pagel, et al. (ref. 58) used beryllium
B-diketonates; and Hughes, et al. (ref. 59) detected beryllium
as the hexafluoroacetylacetonate.  (Note - A potential interfer-
ence from the 27Al3+-ligand can also occur.)  Although highly
sensitive, the methods are not suitable for continuous or short-
term intermittent monitoring because of the time required for
derivative formation.
5.   RECOMMENDED ANALYTICAL TECHNIQUE FOR CONTINUOUS MONITORING
     OF BERYLLIUM AND CADMIUM

     There is no method available at this time that can be imme-
diately adapted to monitoring particulate and/or fume emission
from stationary sources for beryllium and/or cadmium on a con-
tinuous, round-the-clock basis.  The work of Webb, et al. (see
Section 4.1.1) suggests that a continuous monitor can be developed
based on emission spectroscopic measurements of a continuous
flowing (40 1/min.) stream of stack gases.  This type of emission
spectroscopic measurement can provide a means of monitoring both
beryllium and cadmium (or most other metallic elements and some
non-metals) simultaneously.  The technique has the necessary
(a) selectivity, (b) speed (30 sec.) for rapid, repeated analyses,
and (c) sensitivity (^1 yg/m3).
     An evaluation of the Webb-type emission spectrographic
approach must be made to establish the efficiency of the excita-
tion process when analyzing entrained particulate of variable
particle size and when measuring stack emissions having moderate
to high loadings of entrained particulate.   The Webb system worked
well in monitoring the beryllium content in a work environment
where the particulate was principally beryllium compounds.   No
data are available regarding performance in measuring small
amounts (1-100 ppm) of Be or Cd in inorganic particulate, e.g.,
fly ash.

     Electrode wear and deposition of particulate from stack
gases having high loadings of particulate may be major problems
in applying the Webb technique to monitoring stack emissions.
By using an electrodeless discharge system (radio-frequency
coupled excitation), the electrode wear and deposition problems
could be eliminated or minimized.
                               200

-------
     The major problem related to using radio-frequency excitation
is the difficulty in initiating and maintaining an air plasma of
desirable geometry.  Also, adverse oxidation effects may limit
the sensitivity.  By blending mixtures of inert matrix gas (A or
He), with air containing entrained particulate and with a reducing
gas, e.g., hydrogen, a stable plasma and suitable sensitivity may
be attained.  However, by diluting the sample with an inert or a
reducing gas, some loss in the capability to detect the lower con-
centrations of Be or Cd will be experienced.

     A trade-off of performance specifications may depend on the
type of emission source being tested.  The Webb system may be
usable for low (1-100 ug/m3) Be or Cd levels in source emissions
composed of particulate or fume having Be or Cd as major compo-
nents and little, if any. other metallic elements.  Similarly,
the method based on the 5Be(ct,n,Y)' 2C reaction (see Section 4.4.5)
can be used for measuring beryllium continuously in atmospheres
or stack gas emissions containing principally beryllium or its
compounds down to 1 ug/m3, but cannot be used where non-beryllium
particulate would be excessive.

     X-ray fluorescence techniques can be used for measuring
cadmium, but not for determining beryllium.  X-ray tube excitation
sources have been applied to analyses for cadmium, but limited
work has been performed with radioisotope induced x-ray emission
(21>1Am).  The application of radioisotope induced x-ray emission
spectroscopy to determining cadmium in collected stack emission
particulate may require further development of new or modified
radioisotope sources, e.g., 125I.
                               201

-------
6.    REFERENCES

 1.  Durocher, N. L., "Preliminary Air Pollution Survey of
     Beryllium and Its Compounds," U.  S.  Dept.  of HEW,  NAPCA,
     N.C., October 1969.

 2.  Tepper, L. B., H. H.  Hardy and R. I. Chamberlin, The Toxicity
     of Beryllium Compounds, London: Elsevier,  1961

 3.  Cuffe, S. T., R. W. Gerstle, A. A.  Orning  and C. H.  Schwartz,
     "Air Pollutant Emissions from Coal-Fired Power Plants,  Report
     No. 1," J. Air Poll.  Control Assoc.  l4_(9), 353-362 (1964).

 4.  Gerstle, R. W., S. T. Cuffe, A. A.  Orning  and C. H.  Schwartz,
     "Air Pollutant Emissions from Coal-Fired Power Plants,  Report
     No. 2," J. Air Poll.  Control Assoc.  15_(2), 59-64  (1965).

 5.  Cross, F. L., R. J. Drago, H. E.  Francis,  "Metal and Particu-
     late Emissions from Incinerators  Burning Sewage Sludge  and
     Mixed Refuse," Proc.  1970 National  Incinerator Conference,
     ASME, New York, N.Y., 1970, pp 189-195.

 6.  Athanassiadis, Y. C., "Preliminary  Air Pollution Survey of
     Cadmium and Its Compounds," U.S.  Dept. of  HEW, NAPCA, N.C.,
     October 1969.

 7.  Hagenah, W. D., "Improvement of Trace Sensitivity  [in
     Spectrochemical Analysis] by the  Use of Time-resolved Spark
     Spectra," REv. Universelle Mines  15, 369-75 (1959).

 8.  Brooks, R. R. and C.  R. Boswell,  "A Comparison of  Cathode
     and Anode Excitation in the D.C.  Arc," Anal. Chim. Acta 32,
     339-3^5 (1965).

 9.  Mellichamp, J. W., "Cored Cathode for Stabilization of  the
     d.c. Arc," Appl. Spectry. 2_1, 23-27 (1964).

10.  Schroll, E. and D. Sauer, "Use of Large Graphite Beakers  in
     Double-Arc Analysis," Appl. Spectry. 2_0_, 404-407  (1966).

11.  Izyumova, L. G., "Use of a Fluorination Reaction for Deter-
     mining Clarke Contents of Beryllium," Spektrosk.,  Tr. Sib.
     Soveshch., 4th, 348-9 (1965).

12.  Johannes, E., "Direct Spectral Determination of Trace
     Elements in Dictyonema Shale by Burning in a Paper Wrapper,"
     Eesti NSV Tead. Akad. Toim., Keem.,  Geol.  17_, 167-74 (1968).

13.  Schrenck, W. G., Show-jy Ho, and  D.  A. Lehman, "Comparison
     of d.c. Arc and Plasma Arc Excitation to Determine Rhenium
     in Molybdenite," Appl. Spectry. 20,  241 (1966).
                               202

-------
14.   Serin, P. A. and K. H. Ashton, "Spectrochemical Analysis
     of Aqueous Solutions by the Plasma Jet," Appl.  Spectry.  18,
     1966 (1964).

15.   Mavrodineanu, R. and R. C. Hughes, "The RF Discharge at
     Atmospheric Pressure and Its Use as an Excitation Source in
     Analytical Spectroscopy," Developments in Applied Spectros-
     copy 3, 305-333 (1963).

16.   Greenfield, S., I. LI. Jones and C. T. Berry, "High Pressure
     Plasmas as Spectroscopic Emission Sources," Analyst 89,
     713-720 (1964).

17.   Runnels, J. H. and J. H. Gibson, "Characteristics of Low
     Wattage Microwave Induced Argon Plasmas in Metals Excitation,"
     Anal. Chem. 39., 1398-1405 (1967).

18.   Rasberry, S. D., B. F. Scribner and M. Margoshes, "Laser
     Probe Excitation in Spectrochemical Analysis.  I:  Charac-
     teristics of the Source," Appl. Opt. 6_, 8l (1967).

19.   Rasberry, S. D., B. F. Scribner and M. Margoshes, "Laser
     Probe Excitation in Spectrochemical Analysis.  II:  Investi-
     gation of Quantitative Aspects," Appl. Opt. £,  8l (1967).

20.   Runge, E. P., S. Bonfiglio and F. R. Byan, "Spectrochemical
     Analysis of Molten Metal Using a Pulsed Laser Source,"
     Spectrochemica Acta 22, 1678 (1966).

21.   Dickinson, G. W. and V. A. Fassel, "Emission Spectrometric
     Detection of the Elements at the Nanogram per Milliliter
     Level Using Induction-Coupled Plasma Excitation," Anal.
     Chem. 4_1, 1021-1024 (1969).

22.   Bokowski, D. L., "Rapid Determination of Beryllium by a
     Direct-Reading Atomic Absorption Spectrophotometer," Am.
     Ind. Hyg. Assoc. J. 2_9_, 474-481 (1968).

23.   Hingle, D. N., G. E. Kirkbright, and T. S. West, "The
     Determination of Beryllium by Thermo-emission and Atomic-
     fluorescence Spectroscopy in a Separated Nitrous Oxide-
     Acetylene Flame," Analyst 93., 522-527 (1968).

24.   Robinson, J. W. and C. J. Hsu, "Atomic Fluorescence of
     Beryllium," Anal. Chim. Acta 43_, 109-117 (1968).

25.   Manning, D. C., "The Nitrous Oxide-Acetylene Flame in Atomic
     Absorption Spectroscopy," Newsletter 5_, 127-134 (1966).

26.   Manning, D. C., "The Determination of Boron, Beryllium,
     Germanium and Niobium Using the Nitrous Oxide-Acetylene
     Flame," Atomic Absorption Newsletter 6_, 35-37 (1967).


                               203

-------
27.  Kirkbright, G. P., A. Semb, and T. S. West, "The Separated
     Nitrous Oxide-Acetylene Flame as an Atom Reservoir in
     Thermal Emission Spectroscopy," Spectrosc. Lett. 1, 7-11
     (1968).

28.  Bratzel, M. P., Jr., R. M. Dagnall, and J. D. Winefordner,
     "Hot Wire Loop Atomizer for Atomic Fluorescence Spec-
     trometry," Appl. Spectrosc. 2_4, 518-521 (1970).

29.  Ramakrishna, T. V., P. W. West, and J. W. Robinson, "The
     Determination of Aluminum and Beryllium by Atomic Absorp-
     tion Spectroscopy," Anal. Chim. Acta 39, 81-87 (1967).

30.  Thompson, R. J., G. B. Morgan and L. J. Purdue," Analysis
     of Selected Elements in Atmospheric Particulate Matter by
     Atomic Absorption," Atomic Absorption Newsletter 9_, 53-57
     (1970).

31.  Kahn, H. L., and J. E. Schallis, "Improvement of Detection
     Limits for Arsenic, Selenium, and Other Elements (Cadmium)
     with an Argon-Hydrogen Flame," Atomic Absorption Newsletter
     7, 5-9 (1968).

32.  Dagnall, R. M., K. C. Thompson, and T. S. West, "Some
     Experimental Parameters in Atomic Fluorescence Spectro-
     photometry," Anal. Chim. Acta 3_6_» 269-277 (1966).

33.  Fleet, B., K. V. Liberty, and T. S. West, "Study of Some
     Matrix Effects in the Determination of Beryllium by Atomic
     Absorption Spectroscopy in the Nitrous Oxide-Acetylene
     Flame," Talanta !£, 203-210 (1970).

34.  Hobbs, R. S., G. F. Kirkbright, M. Sargent, and T. S. West,
     "Spectroscopy in Separated Flames," Talanta 15, 997-1007
     (1968).

35.  Marshall, G. B., and T. S. West, "Multi-element Atomic
     Fluorescence Spectroscopy," Anal. Chim. Acta 51, 179-190
     (1970).

36.  Vickers, T. J. and R. M. Vaught, "Non-dispersive Atomic-
     Fluorescence Analysis," Anal. Chem. 4.1, 1476-1478 (1969).

37.  Breck, P., "Multi-element Approaches in Atomic Absorption
     Spectroscopy," Proc. Colloq. Spectrosc. Inst. 14th, 3_,
     1191-6 (1967).

38.  White, R. A., "Atomic Absorption Method for the Measurement
     of Metal Fumes," J. Sci. Instrum. 44_, 678-680 (1967).
                               204

-------
39.  Truitt, D., and J. W. Robinson, "Spectroscopic Studies of
     Radio-Frequency Induced Plasma, Part I.  Development and
     Characterization of Equipment," Anal. Chim.  Acta 49,
             (1970).
40.  Holiday, J. E., "Soft X-ray Emission Spectroscopy in the
     10 to 150 A Region," in E. F. Koelble (ed.), Handbook of
     X-rays for Diffraction, Emission, Absorption and Microscopy,
     pp. 38-1 to 38-41, McGraw-Hill Book Co., New York (196?).

41.  Luke, C. L., "Determination of Traces of Lithium, Beryllium
     or Phosphorus by X-ray Analysis," Analytica chim Acta 45,
     365-376 (1969).                                       ~~

42.  Pichoir, P.M. A., "Fluorescent X-ray Analysis Using Fluid
     Filters," U.S. 3,456,108, 15 July 1969.

43.  Cares, J. W., "The Quantitative Determination of Airborne
     Metallic Dusts and Fumes by X-ray Spectrometry , " Amer. Ind.
     Hyg. Assoc. J. 29_, 463-468 (1968).

44.  Hirt, R. C., W. R. Doughman, and J. B. Gislard,  "Applica-
     tion of X-ray Emission Spectrography to Air-Borne Dusts in
     Industrial Hygiene Studies," Anal. Chem. 28, 1649-1651
     (1956).                                  —

45.  Rhodes, J. R., A. Pradzynski, R. D. Sieberg, and T. Furuta,
     "Application of a Si (Li) Spectrometer to X-ray Emission
     Analysis of Thin Specimens," presented at Third  Symposium
     on Low Energy X-ray and Gamma Ray Sources and Applications,
     Boston College, June 10-12, 1970.

46.  Rhodes, J. R., "Design and Application of X-ray  Emission
     Analyzers Using Radioisotope X-ray or Gamma Ray  Sources,"
     Energy Dispersion X-ray Analysis:  X-ray and Electron Probe
     Analysis, ASTM STP 485, American Society for Testing and
     Materials, 1971, pp. 243-285.

47.  Hollstein, M. G. and DeVoe, J. R., "Analysis for Medium-
     weight Elements by a-excited X-ray Fluorescence," in
     ORNL-IIC-10, Baker, P. S. and Gerrard, M., eds.   Oak
     Ridge National Laboratory, Oak Ridge, Tenn . , Sept. 1967,
     pp. 483-502.

48.  Rhodes, J. R., "Radioisotope XRF Analysis of Particulate
     Air Pollutants," U. S. Atomic Energy Commission Contract
     No. AT-(40-1)-4205, Monthly Letter Reports 1 and 2 (1971).

49.  Rhodes, J. R., A. H. Pradzynski, and J. S. Payne, "Energy
     Dispersion X-ray Fluorescence Spectroscopy for Rapid Multi-
     element Analysis of Air Particulates," Columbia Scientific
     Industries, presented at Pittsburgh Conference,  Cleveland,
     Ohio, March 1972.

                                205

-------
50.  Rhodes, J. R., A. H. Pradzynski, and R. D. Sieberg, "Energy
     Dispersive X-ray Emission Spectrometry for Multielement
     Analysis of Air Particulates," Columbia Scientific Industries,
     presented at Instrument Society of America Symposium, San
     Francisco, Calif., May 1972.

51.  Ricci, E. and R. H. Handley, "Activation Analysis with
     Californium-252," Anal. Chem. 4^, 380-382 (1970).

52.  LutZj G. J., "Photon Activation Analysis - A Review,"
     Anal. Chem. 4_3_, 93-103 (1971).

53.  Tyrell, A. C., J. W. Roberts, and R. G. Ridley, "Electron
     Bombardment Ion Source for Isotopic Analysis of Solids,"
     J. Sci. Instrum. 4^, 806-807 (1965).

54.  Brown, R. and P.G.T. Vossen, "Spark Source Mass Spectro-
     metric Survey Analysis of Air Pollution Particulates,"
     Anal. Chem. 4_2_, 1820-1822 (1970).

55.  Morrison, G. H. and A. T. Kashuba, "Multielement Analysis
     of Basaltic Rock Using Spark Source Mass Spectrometry,"
     Anal. Chem. 4^, 1842-1846 (1969).

56.  McHugh, J. A. and J. C. Sheffield, "Mass Spectrometric
     Determination of Beryllium at the Sub-nanogram Level,"
     Anal. Chem. 39_, 377-8 (1967).

57-  Majer, J. R., M.J.A. Reade and W. I. Stephan, "Mass Spec-
     trometry of Metal Chelates-III," Talanta 15_, 373-378 (1967).

58.  Patel, K. S., K. L. Rinehart, and J. C. Bailar, "Mass-
     spectral Studies of Some 3-diketones and Their Beryllium
     Complexes," Org. Mass. Spectrom. 3_, 1239-54 (1970).

59.  Hughes, G. M., T. 0. Tiernan, W. R. Wolf, R. E. Sievers,
     and M. L. Taylor, "Determination of Chromium and Beryllium
     at the ppb Level by Gas Chromatography-Mass Spectrometry,"
     Anal. Chem. (in press).
                               206

-------
                     APPENDIX III
BRIEFING DOCUMENT - ANALYTICAL TECHNIQUES FOR MERCURY
                         207

-------
                        TABLE OF CONTENTS

Section                                                      Page

   1   INTRODUCTION                                           210

   2   SOURCES AND TYPE OP EMISSION                           210

   3   CANDIDATE ANALYTICAL TECHNIQUES                        211

       3.1  Mercury Vapor Photometers and Flameless           212
            Atomic Absorption Spectrometry

            3.1.1  Mercury Photometric Analysis Using a       215
                   Source with a Broadened Emission Line
                   to Minimize Interferences
            3.1.2  Mercury Photometric Analysis with Twin     218
                   Cell System to Minimize Interferences
            3.1.3  Measurement of Mercury Vapor Using a       221
                   Photometric System with a Mercury Reso-
                   nance Lamp as a Monochromatic Source
            3.1.4  Analyses for Mercury:   Photometric Mea-    225
                   surements Corrected with Zeeman Effect
            3.1.5  Mercury Photometric Analysis with Thermal  228
                   Decomposition and Gold Concentrator
            3.1.6  Photometric Analyses of Mercury Vapor -    231
                   Detection and Estimation of Organo-Mercury
                   Dusts and Vapors in the Atmosphere
            3.1.7  Portable Mercury Vapor Photometer          23*1

       3.2  Flame Atomic Absorption and Atomic Fluorescence   237
            Spectroscopy

            3.2.1  Determination of Mercury with Impregnated  239
                   Charcoal and Flame Atomic Absorption
                   Spectrophotometry

       3.3  Emission Spectroscopy                             242

            3.3.1  Determination of Mercury Compounds by      243
                   Emission Spectrometry in a Helium Plasma

       3.4  Formation of Colored Complexes
       3.5  Neutron Activation Analysis
       3.6  Radioisotopic Method

            3.6.1  Radiochemical Determination of Mercury     247
                   Vapor in Air

       3.7  Argon lonization Detection                        251
       3.8  Condensation Nuclei Monitor                       251
                               208

-------
                   Table of Contents - Cont'd

Section                                                      Page

   H   SAMPLE COLLECTION TECHNIQUES                           251

   5   SAMPLE PRETREATMENT                                    256

   6   TECHNIQUES FOR GENERATING KNOWN CONCENTRATIONS         258
       OF MERCURY VAPOR IN AIR

   7   PROBLEMS COMMONLY ENCOUNTERED IN MERCURY               259
       ANALYSIS

   8   SUMMARY AND RECOMMENDATIONS                            262

   9   REFERENCES                                             264
                               209

-------
                   ANALYTICAL TECHNOLOGY FOR
               CONTINUOUS MONITORING OP STATIONARY
                  SOURCE EMISSIONS FOR MERCURY


1.   INTRODUCTION

     Mercury in the atmosphere can exist in many chemical and
physical forms depending on the type of source emission.  Ele-
mental mercury, because of its high vapor pressure, is constantly
emitted by many industries as vapor.  Chemically bound mercury -
organic and inorganic compounds - can be emitted as vapor or
aerosols.  The volatile elemental mercury is readily produced by
thermal decomposition of chemically bound mercury at relatively
moderate temperatures and by a combination of reduction and dis-
proportionation reactions.

     Continuous monitoring of stationary source emissions for
mercury must be able to handle vapor and particulate, either of
which could be elemental mercury, inorganic or organic mercurials,
or mercury (or its compounds) adsorbed on particulate.

     At this time, there is no method that has been sufficiently
tested or used for continuous monitoring of total mercury emis-
sions from stationary sources.


2.   SOURCES AND TYPE OF EMISSION

     Emission sources and uses were identified in a survey by
Stahl (ref. 1).  These include ore refining, electrical apparatus,
electrolytic production of chlorine and caustic soda, paints,
Pharmaceuticals, agricultural compounds, power plants, and in-
cinerators.  The principal chemical forms are classified as ele-
mental mercury, mercuric oxides, inorganic and alkyl mercury
salts (nitrates, sulfates, chlorides, sulfides, fluorides,
bromides, iodides) and organo-mercury compounds (alkyl or aryl
mercury compounds or salts of organic acids).  Most of these
compounds are volatile or decompose into elemental mercury at
moderate temperatures.

     Mercuric oxide (500°C), mercuric oxychloride (260°C),
mercuric iodide (290-310°C) mercuric fluoride (645°C), mercuric
bromide (130-l40°C), and mercurous nitrite (100°C) decompose,
yielding principally volatile, elemental mercury (b.p. 356.6°C).
Other inorganic salts, mercurous chloride (i!00°C), mercurous
bromide (3**5°C), mercurous iodide (l40°C), and mercuric sulfide
(58i»°C) sublime at low to moderate temperatures yielding vapor
containing the individual molecular species, or a mixture con-
taining elemental mercury, the original mercury compound, and a
mercury compound in a higher oxidation state.  The mixture can
                               210

-------
be obtained for moist compounds, whereas dry materials do not
yield measurable amounts of the disproportionation products.
If not dry, mercurous chloride decomposes at the sublimation
temperature into Hg° and HgCl2.  There is also some evidence for
a similar effect during sublimation of mercurous iodide (ref. 2).

     Mercurous chloride has a vapor pressure of 0.004 mm @ 90°C
and 0.2? mm & 170°C.  Mercuric chloride volatilizes rapidly at
its melting point of 276°C and has a vapor pressure of 0.003 mm
@ 60°C and 3 mm @ 152°C.  The vapor pressure of mercuric bromide
is reported as 0.045 mm @ 90°C and 3 mm % 162°C (ref. 2).

     Organo-mercury compounds are moderately volatile to highly
volatile.  Dimethylmercury has a normal boiling point of 95°C.
Although no specific boiling point is given, methylmercuric
chloride is reported as volatile at 100°C.  Diphenylmercury sub-
limes at 121.8°C and phenylmercuric hydroxide decomposes at 200°C.


3.   CANDIDATE ANALYTICAL TECHNIQUES

     Methods of analyses for total mercury in various media have
been based on colorimetry, emission or absorption spectroscopy,
neutron activation analysis, x-ray fluorescence, isotope dilution
and electrolysis (ref. 1).  Some of the methods - colorimetry,
neutron activation, and electrolysis - require considerable sample
pretreatment and as a consequence cannot be readily adapted for
continuous or intermittent, but rapid, monitoring systems.  Many
of the techniques require the collection of mercury and mercury
compounds prior to analysis.  In certain methods, for example,
neutron activation, some sample separation is required before
the mercury can be measured.

     The high selectivity, sensitivity, and speed of the spectro-
scopic techniques make methods based on these principles most
attractive.  Several methods using the photometric measurement of
the absorption of ultra-violet radiation at 2537 A have formed the
basis of commercial monitoring detectors for mercury vapor.  The
lower detection limit of these instruments is approximately 5 to
10 yg/m3 but the instruments are subject to interference from
chemical compounds, dust, and smoke which would absorb or scatter
light at 2537 S.  To lower the detection limits of the photometers,
techniques have been devised to concentrate the mercury vapor
and/or eliminate the interferences.  Other techniques using emis-
sion spectroscopic principles are very sensitive and have potential
for a continuous or intermittent, but rapid, monitoring system.

     This document describes the analytical approaches for direct,
continuous monitoring of mercury vapor and/or mercury containing
particulate.  An attempt is made to show the advantages and limi-
tations of each method.
                               211

-------
     Also, the means for collecting or concentrating the mercury
 specimen are described for possible use in short-term, intermit-
 tent monitoring requiring high sensitivity.

     The analytical methodology is categorized as (a) Mercury
 Vapor Photometers and Flameless Atomic Absorption, (b) Flame
 Atomic Absorption and Fluorescence Spectrophotometry, (3) Emission
 Spectroscopy, (d) Formation of Colored Complexes, (e) Neutron
 Activation Analysis, (f) Radioisotopic Method, (g) Argon loniza-
 tion Detection, and (h) Condensation Nuclei Monitor.

 3.1  Mercury Vapor Photometers and Flameless Atomic Absorption
     Spectrometry

     The mercury vapor photometer and flameless atomic absorption
 Spectrophotometry are based on the principle that mercury reso-
 nance radiation at a wavelength of 2537 X is strongly absorbed
 by mercury vapor.  Both techniques measure the atomic absorption
 phenomena.

     In the simpler mercury vapor photometer, generally no mono-
 chromator is used and the source of the mercury resonance radia-
 tion is a low pressure germicidal lamp.  Isolation of the 2537 A
 line is accomplished with a narrow-band photoelectric cell.  The
 flameless atomic absorption spectrophotometer uses a hollow cath-
 ode lamp as an energy source and a monochromator to isolate the
 2537 & line.  The monochromator minimizes spectral interference,
 principally stray radiation due to (a) ambient radiation, (b) re-
 flected radiation from the exciting source, and (c) fluorescence
 of the silica cell components excited by UV radiation.  Stray
 radiation causes nonlinear or reduction of photometric response
 which is characterized by curvature and reduction in slope of the
 analytical calibration curves.

     A number of low cost, portable, continuous, mercury-vapor
 photometers are available commercially (Coleman Instruments;
 Beckman Instruments, Inc.).  The basic design was developed
 by Woodson (ref. 3).  Many other variations have been reported
 (ref. 4-17).  When used as a direct measuring device without
 pre-concentration steps, the lower sensitivity is in the range
 of 5 to 10 yg/m3.  Costs range from $300 to $1000 (without
 recorder) and the devices are approximately 13" x 12" x 8" and
 weigh between 14 to 20 Ibs.

     Flameless atomic absorption spectrophotometers are generally
more sensitive,  but are also more expensive, ranging from $2500
 to approximately $10,000 (without recorder), and weigh approxi-
mately 150 Ibs.
                               212

-------
     Although these techniques are continuous monitors and are
very sensitive, considerable interferences can occur which affect
the reliability of the measurements.  Dust, smoke, and fog can
reduce the light intensity and magnetic interferences are re-
ported.  A number of compounds. Table I (ref. 17), also absorb
ultraviolet radiation at 2537 A and could result in erroneous
high mercury values.  As shown in Table I, the spectral response
for mercury vapor is much greater than the responses for equiva-
lent amounts of the potential interfering compounds.  Relatively
high concentrations of these compounds are required before
interference becomes a problem.

     Olin Corporation markets an automatic "Mercury Monitor for
Gases" based on pyrolysis (760°C) of total mercury containing
materials - metallic, ionic, organic - followed by reduction with
stannous chloride in hydrochloric acid solution to yield elemental
mercury vapor.  The mercury vapor content is measured in an ultra-
violet spectrophotometer equipped with a one meter gas cell.  A
working concentration range of 0-0.05 mg/m3 or higher in air or
other gaseous systems is quoted by the manufacturer.  No data on
the specific, lower detection limit is given.  The analysis cycle
is approximately six minutes per analysis.  The instrument cost
is quoted as approximately $20,000.

     Various methods have been devised to prevent or compensate
for interferences.  Although developed principally for measuring
mercury in occupational environments, the techniques in modified
form can also be used to monitor mercury from stationary source
emissions.  Background compensation with a deuterium continuum
lamp, broadened emission line, twin cells and mercury removal,
Zeeman effect, and amalgamation on silver, gold or palladium have
been used.  Technology relating to some of these methods are pre-
sented in the following subsections.
                                213

-------
                      Table I
      SENSITIVITY OF MERCURY VAPOR PHOTOMETER

                                Relative Sensitivity
                                	(ppm)	
Mercury                                0.0001
Tetraethyllead                         0.13
Xylene                                 0.2
Monochlorobenzene                      0.3
Aniline                                0.3
Perchloroethylene                      0.5
Chloroprene                            0.5
Toluene                                1.0
Benzene                                1.2
Vinylacetylene                         1.6
Phosgene                               5
Acetone                                5
Ethylbenzene                           5
Pentachloroethane                      7
Hydrogen sulfide                       8
Trichloroethylene                     10
Carbon disulfide                      12
Gasoline (Blue Sunoco)                50
            Photometer Insensitive to:
Methylene chloride              Ethyl alcohol
Carbon tetrachloride            Amyl alcohol
Ethylene dichloride             Ethyl acetate
Tetrachloroethane               Ethyl Cellosolve
Chloroform                      Methyl Cellosolve
Methyl chloride                 Dowtherm A
Vinyl chloride                  Water vapor
Methyl alcohol

-------
     3.1.1  Mercury Photometric Analysis Using a Source with a
            Broadened Emission Line to Minimize Interferences

          3.1.1.1  Principle and Applicability

               3.1.1.1.1  Principle - Mercury vapor in air atmo-
spheres is measured quantitatively with a photometric analyzer
which measures the absorption of a broadened emission line (at
2537 8) emitted by a mercury vapor lamp.  The width of the emis-
sion line is greater than the width of the absorption line.
The portion of the broadened emission line that is outside the
absorption line of the mercury vapor in the sample is not absorbed
by the mercury vapor.  This portion of the emission line is used
as a reference to account for the absorption of non-mercury
species which normally interfere with the mercury photometric
measurements .

     High sensitivity is attained by using a folded path concept
based on four internal reflections of the 2537 & mercury line in
the sample cell.

               3.1.1.1.2  Applicability - System was designed for
measuring mercury emitted from solid or liquid batch samples.

     Method can be used for intermittent, but rapid, monitoring
of mercury vapor in atmospheric samples or for measuring mercury
vapor emitted from mercury-containing specimens which yield ele-
mental mercury on pretreatment with thermal or chemical reduction.
Spectral interferences common to the mercury photometric measure-
ment are minimized.

          3.1.1.2  Range and Sensitivity

     No data given, but should be able to detect at least 5 ug/m3.

          3.1.1.3  Interferences

     Compensation technique used in measuring sidebands of the
broadened emission line from the mercury vapor lamp should cor-
rect for most, if not all, interferences (dust, aromatic hydro-
carbons, etc.) encountered in the application of the mercury
vapor photometers.

          3.1.1.4  Accuracy, Precision and Stability
     No data given.

          3.1.1.5  Apparatus

               3.1.1.5.1  Mercury Vapor Lamp - Model 11SC1
(Penray Ultraviolet Products) with emission line at 2537 & having
a linewidth of about 0.1 X .
                               215

-------
               3.1.1.5.2  Sample Cell - Plane and spherical
mirrors are used internally to provide four reflections.

               3.1.1.5.3  Interference Filter - To isolate
2537 A emission.

               3.1.1.5.^  Optical Filter - Cylindrical cell con-
structed with quartz windows and containing dense mercury vapor.
Cell is used to absorb or suppress the central portion of the
emission line, so that the light passing through the optical fil-
ter is contained within the sidebands only.

               3.1.1.5.5  Source and Reference Photomultipliers -
RCA 1P28.

               3.1.1.5.6  Electronics - Preamplifiers, differen-
tial amplifier, integrator, power supplies, and sealer and
readout meter.

               3.1.1.5.7  Flow Controller - 4 liters/minute.

               3.1.1.5.8  Dilution Apparatus - To provide mercury
sample for calibration.

          3.1.1.6  Time Cycle for Sampling and Measuring

     No data given, but should be very rapid.  Time cycle is
dependent mostly on the ease by which the elemental mercury can
be obtained in the thermal decomposition step.  A sampling time
estimate would be approximately 5-10 minutes.

          3.1.1.7  Calibration Procedure

     Special dilution apparatus is provided in which mercury
vapor can be diluted with air.

          3.1.1.8  Method of Sampling and Sample Preparation

     Batch samples (solid or liquid) were heated to vaporize
elemental mercury into a stream of purified air which was passed
through the analyzer.

          3.1.1.9  Multi-element Application

     Present system is specific for mercury, but the principle
used could be applied with hardware modifications to other vola-
tile elements.

          3.1.1.10  Physical Dimensions

     None reported; probably small to moderate in size.
Estimated 4' x 2' x V .
                               216

-------
          3.1.1.11  Unit Output

     Weight units or concentration units.

          3.1.1.12  Safety Hazard

     Moderate electrical voltages.

          3.1.1.13  Recommendation for Method Improvement

     System was designed principally for measuring mercury in
batch samples.  Minor modifications would permit continuous or
semi-continuous measurement of air specimens.  The deterioration
of the mirrored surfaces in the sample cell could become a major
problem if air or stack emission samples had high loadings of
particulate.  Elimination of the mirrors and the use of a long
optical path may provide sufficient sensitivity with moderate
loadings of particulate.

     Instrument response factors would have to be determined to
establish higher flow rates than 4 liters/minute if direct stack
emissions were to be analyzed.

          3.1.1.14  References

     Barringer, A. R., "Method for Absorption Analysis Using a
Source Having a Broadened Emission Line," U.S. 3,571,589
(23 March 197D .

     Barringer, A. R., "Interference-free Spectrometer for
High-sensitivity Mercury Analyses of Soils, Rocks and Air,"
Trans. Inst. Mining & Metallurgy Sec. B, 75, Bull. 714, 120-4
(1966).
                               217

-------
     3.1.2  Mercury Photometric Analysis with Twin Cell System
            to Minimize Interferences

          3.1.2.1  Principle and Applicability

               3.1.2.1.1  Principle - Mercury vapor In air Is
measured by a photometric analyzer based on the absorption by
the mercury vapor of radiation (2537 A*) emitted by a fused silica,
low-pressure mercury vapor lamp.  Fluctuations in lamp intensity
and signal voltage are minimized by compensating with a reference
path and separate detector in a twin-cell arrangement.  Correc-
tions for components, other than mercury, which also absorb at
2537 ft are accomplished by measuring the absorption background
for a "demercurified" specimen simultaneously with a measurement
on the mercury-containing air sample.  Removal of mercury is
attained by passing one-half of the sample through a column of
glass wool impregnated with palladium chloride.

               3.1.2.1.2  Applicability - Although designed for
geochemical prospecting, the system can be used for direct mea-
surement of mercury vapor in air, as well as mercury in solid
samples - rocks and soils.  Analyses on rocks and soils are per-
formed on vapors generated by heating the samples to 600-800°C.

          3.1.2.2  Range and Sensitivity

               3.1.2.2.1  Range - No details on upper limits.

               3.1.2.2.2  Sensitivity - Present instrument
10 *yg Hg with possibility of ICT'ug Hg.

          3.1.2.3  Interferences

               3.1.2.3.1  Chemical and Spectral - The effects of
ultraviolet absorbing materials(organic compounds, ozone, etc.)
are minimized or eliminated by using the twin cell arrangement
with a reference path containing the air sample with mercury re-
moved.  (Note - The reaction of Hg with PdCl2 is specific and
takes place spontaneously in the cold and in the absence of
moisture.)

     In addition, the generation of ozone from atmospheric oxygen
by radiation having wavelengths <2000 & emitted by the mercury
lamp is eliminated by using a lamp made of slightly impure fused
silica (Vycor) which filters out the undesirable radiation.

               3.1.2.3.2  Physical - Physical scatter of ultra-
violet radiation by fume, dust, and other suspensions in the
vapor is compensated by employing the twin-cell reference path
system.
                               218

-------
     Adsorption of mercury on glass, silica gel, and other mate-
rials in the instrument can be a problem.  Degree of adsorption
seems to be affected by the presence or absence of moisture.
(Note - The effect is presented in Ref. 1, Section 3.1.2.14 as a
private communication from S. H. Williston, but no data are given
regarding whether the adsorption is enhanced or decreased by in-
creasing moisture content.)

          3.1.2.4  Accuracy, Precision and Stability

     No data given for vapor, but a reproducibility of ±15? at
the 95% confidence level over the range 1-10 ppm was reported
for solid samples heated to 600-800°C.

          3.1.2.5  Apparatus

               3.1.2.5.1  Source - Low-pressure, fused-silica
(Vycor) mercury vapor lamp.

               3.1.2.5.2  Detector - Two photocells and suitable
electrical circuit for measuring differential output.

               3.1.2.5.3  Sample Cell - Two glass columns with
quartz windows.

               3.1.2.5.4  "Demercurlfying" Agent - Palladium
chloride, prepared by evaporating a 1% aqueous solution of PdCl2
on glass wool.  (Note - Apparently removes mercury from dry air
down to concentrations below 0.01 pg/m3 Hg).

               3.1.2.5.5  Ancillary Equipment - Silica gel column
to remove water vapor; flowmeters, glass wool; vacuum pump; meter
or recorder; suitable valves and tubing; and electronic amplifiers

          3.1.2.6  Time Cycle for Sampling and Measuring

     Analyses for mercury in air should be instantaneous and
continuous.

     No data are given for heating times required to release
mercury from soil or rock samples.  Estimated times are dependent
on the method of introducing the sample into the hot zone, but
could range from 1 to 15 minutes.

          3.1.2.7  Calibration Procedure

     Apparatus was calibrated by injecting known volumes of air
saturated with mercury at different temperatures.
                               219

-------
          3.1.2.8  Method of Sampling and Sample Preparation

     Air sampling was accomplished by drawing an air specimen
through the instrument with a suction pump.

     For solid materials, the mercury content is determined by
heating (600-800°C) a known weight of sample and allowing the
vapor to pass into the instrument as done for air sampling.

          3.1.2.9  Multi-element Application

     Method is specific for mercury.

          3.1.2.10  Physical Dimensions

     No data given, but unit is mobile and was designed for
geochemical prospecting.

          3.1.2.11  Unit Output

     Voltage readings convertible to yg Hg or yg/m3.

          3.1.2.12  Safety Hazard

     Moderate electrical voltages.

          3.1.2.13  Recommendation for Method Improvement

     The adsorption of mercury on glass and materials of con-
struction in the presence of water must be evaluated.

     By performing direct heating (^800°C) of airborne particu-
late in moderate flowing air streams, an analyzer which measured
total mercury (particulate and vapor) and/or mercury vapor can
be developed.  The major concern would be the residence time in
the heated zone required to efficiently volatilize the mercury
in the particulate.  Plow rate limits have not been defined,
but thermal decomposition of organo-mercury and inorganic mercury
compounds has been accomplished in a continuous flow system at
5 liters/minute at 800°C (see ref. Leong and Ong below).

          3.1.2.14  References

     James, C. H. and J. S. Webb, "Sensitive Mercury Vapour
Meter for Use in Geochemical Prospecting," Bull. Inst. Mining
Met. 691 (6), 633-641 (1964).

     Leong, P. C. and H. P. Ong, "Determination of Mercury by
Using a Gold Trap in Samples Containing Considerable Sulfide
Minerals," Anal. Chem. 43_, 9^0-941 (1971).
                               220

-------
     3.1.3  Measurement of Mercury Vapor Using a Photometric
            System with a Mercury Resonance Lamp as a Mono-
            chromatic Source

          3.1.3.1  Principle and Applicability

               3.1.3.1.1  Principle - By coupling polychromatic
radiation from a standard low pressure arc lamp through a highly
evacuated (0.001-mm pressure) fused silica mercury resonance
lamp, monochromatic radiation can be obtained.  The resonance
lamp will emit resonance radiations 1850 X and 2537 A" .   Since
the radiation at 1850 & will be absorbed by oxygen, pure 2537 X
radiation will be obtained.  The line width (about 0.002 8) is
determined by Doppler's effect only and is a constant,  irrespec-
tive of the line width of the exciting source.

     With the resonance lamp system, no filters or narrow band
pass photoelectric detectors are necessary as required in the
standard mercury vapor photometer, and the monochromator, normally
used with atomic absorption spectroscopy, is not needed.  A
highly sensitive mercury photometer, in which ultimate spectral
sensitivity can be achieved, is possible.

               3.1.3.1.2  Applicability - The system can be used
in all applications in which the standard mercury vapor photometer
is used.  Disadvantages, such as (a) lamp instability due to tem-
perature variations in low pressure arc discharges, (b) broadened
and self-reversed emission lines, (c) reductions in spectral
intensity and non-linearity of response resulting from use of
filter or narrow band pass detectors, which are normally encoun-
tered with standard mercury photometers are eliminated.


     Because of its high sensitivity, the technique can be used
to measure mercury vapor regenerated by heat after it has been
extracted by a collector - cadmium sulfide-impregnated asbestos,
gold leaf, palladium chloride, etc.

          3.1.3.2  Range and Sensitivity

               3.1.3.2.1  Range - 0.0003 yg to 0.05 yg.

               3.1.3.2.2  Sensitivity - 0.0003 yg.

          3.1.3.3  Interferences

               3.1.3.3.1  Chemical - Organic compounds and ozone,
which also absorb radiation at 2537 &, can interfere with detec-
tion of mercury vapor.  To eliminate or compensate for these
nonatomic interferences, a modification of the mercury photometer
concept incorporating a resonance lamp filled with argon to pro-
vide a broadened resonance line was devised.  (Ref.-See Section
3.1.3.1*1.)

                               221

-------
     A shutter is used to pass radiation from twin sources through
two pathways.  Path 1 is a four-stage system having (1) a reson-
ance lamp filled with argon at one atmosphere pressure, (2) a cell
filled with nitrogen at one atmosphere and containing a drop of
mercury,  (3) a low pressure resonance lamp containing mercury
vapor at low pressure (0.001 mm), and (4) the sample absorption
cell.  Path 1 transmits a broadened resonance line, which measures
only nonatomic absorption, while path 2 transmits the sharp reson-
ance line, which measures both nonatomic and atomic absorption.
The difference (atomic absorption) between the two values is
relatable to the mercury vapor content in the sample.

               3.1.3*3.2  Physical - Dust, smoke, and fog can
interfere with the measurement by attenuating the transmission
of radiation.  Corrections can be made by using the system and
techniques cited in sub-section 3.1.3-3.1.

     Optical systems of the type used in this study, which do not
use a dispersion device (spectrograph), are subject to problems
(undesirable curvature and reduction in slope of the analytical
response curves) caused by the presence of stray radiation.  The
four sources of stray radiation are:

     1.   Ambient radiation.
     2.   Reflected radiation from the exciting source.

     3.   Fluorescence of the fused silica excited by
          UV radiation.

     4.   Fluorescence of mercury vapor in the resonance lamp.

     A detailed, quantitative examination of the stray radiation
was performed (see reference below) and a system to minimize the
effect was devised.  By using a short (3 cm) focal length
"suprasil" lens between the source and the resonance lamp, and
by mounting a Wratten 18B filter as a window on the photomulti-
plier end of the absorption tube, the stray radiation was reduced
to about 0.7%.

          3.1.3.4  Accuracy, Precision and Stability

     The percent standard deviation at 0.0003 yg is 37% and
2.9-7.7% in the range 0.01 yg to 0.04 yg.

     Long term stability was affected by the purity of the gas
in the resonance lamp.  Small amounts of foreign components,
e.g., H2 and C02 can cause quenching of the resonance radiation.
Deterioration of the re-emitted energy occurred when an adhesive
was used to cement the parts of the lamp.  However, when the lamp
was constructed by fusing components together, no loss of energy
occurred.
                               222

-------
          3.1.3*5  Apparatus

               3.1.3.5.1  Excitation Source - G.E. ozone lamp
GE OZ 4511  (selected for high stability).

               3.1.3.5.2  Mercury Resonance Lamp - Highly evacu-
ated fused silica containing mercury vapor(^0.001 mm) at room
temperature.

               3.1.3.5.3  Absorption Cell - Tube 1 cm diameter x
35 cm length fitted with two windows, one of fused silica at the
resonance lamp end, while the other, made of Wratten 18B glass
filter, is placed at the detector end.  (Note - The filter cuts
off the visible "stray" radiation produced by fluorescence of the
fused silica excited by UV radiation and fluorescence of mercury
vapor in the resonance lamp.)

               3.1.3.5.4  Photodetector - RCA 1P28 photomulti-
pliers (matched) enclosed in light-proof housing to exclude
ambient radiation.

               3.1.3.5.5  Ancillary Equipment - Vacuum pump,
tubing, flowmeters, amplifiers and recorder, lens, collimator.

          3.1.3.6  Time Cycle for Sampling and Measuring

     Two types of operational modes are possible:  (a) continuous
operation and instantaneous response of air samples, and
(b) batch analyses on "collected" mercury samples.

          3.1.3.7  Calibration Procedure

     A calibration curve was obtained with mercury content ranging
from 0.01-0.05 yg.  Known quantities of mercury, as the sulfide,
were added to a sample tube which could be vented into the absorp-
tion cell to permit diffusion of mercury into the optical path.

     The sulfide mercury solution was prepared from mercuric
chloride solution to which was added a solution of hydrated sodium
sulfide.  Suitable volumes of the solution were introduced into
sample tubes.  After evaporation to dryness in a desiccator under
reduced pressure, the tubes were connected to the absorption cell
and heated to MOO°C with an infrared lamp to promote diffusion
of mercury into the cell.

          3.1.3.8  Method of Sampling and Sample Preparation

     Two modes are possible:  (a) Continuous sampling and measure-
ment of mercury and air samples taken directly; and (b) collection
of mercury by cadmium sulfide impregnated asbestos, gold leaf,
palladium chloride, etc., followed by thermal regeneration of
mercury vapor into the photometer.


                                223

-------
          3.1.3-9  Multi-element Application

     Technique designed specifically for mercury, although general
principles could be adapted to other volatile elements.

          3.1.3.10  Physical Dimensions

     None reported other than a statement that the instrument is
compact and can be powered with batteries to allow for portable
operation.

          3.1.3.11  Unit Output

     Micrograms/m3 or mass units.

          3.1.3.12  Safety Hazard

     Moderate electrical voltages; some ozone generation.

          3.1.3«13  Recommendation for Method Improvement

     Tandem coupling to a thermal decomposition furnace should be
considered to permit analyses of organo-mercurials and possibly
mercury-containing particulate.

          3.1.3.1^  Reference

     Ling, C., "Sensitive Simple Mercury Photometer Using Mercury
Resonance Lamp as a Monochromatic Source," Anal. Chem. 39>
798-804 (1967).
                               224

-------
     3.1.4  Analyses for Mercury:  Photometric Measurements
            Corrected with Zeeman Effect

          3.1.4.1  Principle and Applicability

               3.1.4.1.1  Principle - Mercury, free or chemically
bound, is vaporized in air or oxygen in an absorption cell heated
externally by a furnace.  The quantity of mercury vapor generated
is measured with a mercury photometer.

     To compensate for non-mercury components which either
scatter or absorb the 2537 A radiation, the radiation from the
excitation source (electrodeless 199Hg lamp) is broadened by
applying a magnetic field (10 kgauss) to the source.  The broad-
ening results from hyperfine splitting (Zeeman effect) of the
2537 A line.  By using an argon filled cell containing mercury
vapor after the furnace and before the detector to absorb only
the center component of the broadened 2537 &, a comparison mea-
surement can be performed which compensates for non-mercury
absorption.  The non-mercury absorption is established from
absorption of the two components created by the Zeeman effect,
which lie above and below the 2537 & line.

               3.1.4.1.2  Applicability - System was applied to
analyzing for mercury in food, but can be used on any system for
measuring mercury vapor.  Bound mercury must be decomposed into
element mercury vapor.  Although the system was not designed for
continuous monitoring of air, suitable modifications can be made
to permit this type of measurement.  The major problem, as with
any mercury photometer measurement for continuous monitoring,
is the efficiency of the decomposition process to form elemental
mercury vapor from bound-mercury.

          3.1.4.2  Range and Sensitivity

     No details are given other than detection of 0.04 ppm in
food samples.

          3.1.4.3  Interferences

     System is designed to compensate for interferences which
normally affect the mercury photometric measurement .  Could not
be used in region of high magnetic field.

          3.1.4.4  Accuracy, Precision, and Stability
     Little data reported.  For tuna containing 0.49 and 0.24 ppm
of Hg, all measurements were within 12% of the average over the
range 5 to 35 ng of Hg.
                               225

-------
          3.1.4.4  Apparatus

               3.1.4.5.1  Excitation source - Electrodeless
199Hg lamp.

               3.1.4.5.2  10 kgauss permanent magnet .

               3.1.4.5.3  Absorption cell and automatic heating
furnace (up  to 500°C) .

               3.1.4.5.4  2537 S filter.
               3.1.4.5.5  Natural Hg cell filled with argon.

               3.1.4.5.6  Detection system - photomultipliers ,
phase-sensitive detector and digital integrator.

          3.1.4.6  Time Cycle for Sampling and Measuring

     Total time cycle is approximately one minute.

          3.1.4.7  Calibration Procedure

     Standards were prepared by mixing Hgl in carbon and HgO in
carbon and starch.

          3.1.4.8  Method of Sampling and Sample Preparation

     Known weights  of sample were introduced into the furnace.
No chemical pretreatment is necessary.

          3.1.4.9  Multi-element Application

     Potentially could be applied to any element currently mea-
sured by atomic absorption techniques.

          3.1.4.10   Physical Dimension

     No data reported.  Basic photometer and furnace should be
relatively small.  The weight of the magnet, which is unknown,
could be significant.

          3.1.4.11   Unit Output

     Nanograms of Hg.

          3.1.4.12   Safety Hazard

     Moderate voltages in electronics.
                               226

-------
          3.1.^.13  Recommendations for Method Improvement

     Although not designed for continuous monitoring of air or
stack gases, the decomposition step could be suitably modified
to provide continuous decomposition.  Again, as with any monitor
based on photometric measurement of mercury vapor, the key factor
is the efficiency of the decomposition step.

          3.1.4.1*<  Reference

     Hadeishi, T., and R. D. McLaughlin, "Hyperfine Zeeman Effect
Atomic Absorption Spectrometer for Mercury," Science,
        (1971).
                                227

-------
     3.1.5  Mercury Photometric Analysis with Thermal
            Decomposition and Gold Concentrator

          3.1.5.1  Principle and Applicability

               3.1.5.1.1  Principle - Method was devised for
measuring mercury in geologic materials, but principles can be
adapted to monitoring particulate in stack emissions.  Geologic
materials are heated to 800°C or 900°C in an induction furnace
(ref. 1) or with ^00-kilocycle radio-frequency at 500°C (ref.  2)
to decompose mercury-containing particulate and organo-mercury
compounds.  Mercury vapor, released during the decomposition,
is collected from air at 28- liters/minute on gold wool (ref.  1)
or gold foil (ref. 2) at room temperature or 215-255°C as poten-
tial interfering compounds and smoke are exhausted.  Mercury is
subsequently desorbed from the gold amalgamator by using radio-
frequency induction heating to rapidly heat the amalgam to 800-
850°C.  Mercury vapor content was measured by absorption of
radiation at 2537 A.

               3.1.5.1.2  Applicability - Technique was developed
for measuring mercury in soils, rocks, and gas on a batch sample
basis.  Consideration should be given to the potential adapta-
tion to intermittent, but short time cycle, analyses on stack
emissions.  These analyses would be performed on mercury vapor
generated by an induction (or resistance) furnace or an electrode-
less radio-frequency induced plasma used to heat stack effluent
on a continuous basis.  The major concern is the efficiency of
the decomposition process, which would be dependent on the tem-
perature of the decomposition zone, particle size, stability of
the mercury-containing compound, and residence time of mercury
compound in the decomposition zone.

          3.1.5.2  Range and Sensitivity

               3.1.5.2.1  Range - 1 x 10~9g to >300 x 10~9g.

               3.1.5.2.2  Sensitivity - Detection limit reported
as 1 x 10~9g with low cost mercury vapor meter, but has potential
to detect 1 x 10~12g.

          3.1.5.3  Interferences

     Use of gold to collect the mercury vapor permits removal of
organic vapors, sulfur contaminants, and particulate which could
interfere with the mercury photometric analyzer.  In cases where
tars and water can condense on the gold and thus lessen its
effect as an amalgam, operation of the gold collector between
25° and 259°C may be required to eliminate condensation.
                               228

-------
     During measurements on samples of galena and sphalerite,
minor amounts of lead and zinc oxides were deposited on the tube
adjoining the furnace and gold collector.  The build-up of de-
posits of this type must be considered in the design criteria
and could be a major maintenance problem in continuous, long-
term use.

     Some deterioration of the desorption efficiency (lengthening
of desorption time by several seconds) is reported with long
term use of gold wool.  Also, when amalgamation was performed
at room temperature, 3 to 6% of the mercury was retained on the
gold wool, whereas less than 2% was retained at temperatures
ranging from 215-255°C.

          3.1.5.**  Accuracy, Precision and Stability

     Accuracy was assessed by comparing the measured mercury
contents with spectrographic analyses of rock specimens and was
reported as 5-10% of the amount present depending on the level.

     Reproducibility of the method was determined by analyzing
a standard rock sample ten times and was reported as 11/8 devia-
tion from the mean.

     Although some deterioration of the gold amalgam efficiency
was noted, several hundred analyses of geological specimens were
performed with reliable results.

          3.1.5.5  Apparatus

               3.1.5'5.1  Decomposition System - Induction
furnace - Leco Model 521-500 or unidentified resistance furnace
(ref. 1); ^JOO-kilocycle radio-frequency induction heater (ref. 2);
quartz chamber.

               3.1.5.5.2  Gold Collector - (a) 300 ft of gold
(99.99% Au) wire of 0.003-in. diameter with a theoretical surface
of 3^ in.2 (ref. 1); (b) 15 g of gold squares (2 mm x 2 mm x
1/2 mm).

               3.1.5.5.3  Gold Collector Desorption System -
Induction furnace - Leco Model 521-500 at 800-900°C (ref. 1);
JJOO-kilocycle radio-frequency induction heater at 500-600°C
(ref. 2).

               3.1.5.5.**  Detector - Low cost mercury vapor
photometer (unidentified)(ref.TJ; specially designed single-
or double-cavity absorption chamber with differential amplifier,
voltage-frequency converter, electronic counter and timer
(ref. 2).
                                229

-------
          3.1.5.6  Time Cycle for Sampling and Measurement

     Approximately ^ to 8 minutes total time.

          3.1.5.7  Calibration Procedure

     Mercury from known volumes of air saturated with mercury
was collected on the gold amalgam (ref. 1) and different weighed
aliquots of a standard sample were analyzed (ref. 2).

          3.1.5.8  Method of Sampling and Sample Preparation

     Solids or gases were heated from 500-900°C for 1-2 minutes
to vaporize mercury which was collected as a gold amalgam.
Mercury concentrated as the gold amalgam was subsequently vapo-
rized into a mercury vapor photometer.  Note - Efficiency for
forming an amalgam with gold and subsequent desorption of mercury
was excellent, but much poorer desorption (^50%) efficiency was
observed with platinum.

          3.1.5.9  Multi-element Application

     Specific for mercury.

          3.1.5.10  Physical Dimension

     No data reported, but estimated as 41 x 2' x 6'.

          3.1.5.11  Unit Output

     Mass units (nanograms).

          3.1.5.12  Safety Hazard

     Normal electric voltages.

          3.1.5.13  Recommendation for Method Improvement

     Must determine efficiency of sample decomposition with
induction heating or use of electrodeless radio-frequency plasma
for continuous operation.

          3.1.5.11*  References

     (1) Azzaris, L. M., "A Method of Determining Traces of
Mercury in Geologic Materials," Can. Dept. Energy, Mines Resour.;
Geol.  Surv.  Can., Pap. No. 66-5^, pp. 13-26 (1967).

     (2)  Vaughn, W. W. and J. H. McCarthy, Jr., "An Instrumental
Technique for the Determination of Submicrogram Concentrations
of Mercury in Soils, Rocks, and Gas," Geological Survey Research,
pp. D123-D127 (1964).


                               230

-------
     3.1.6  Photometric Analyses of Mercury Vapor - Detection
            and Estimation of Organo-Mercury Dusts and Vapors
            in the Atmosphere

          3.1.6.1  Principle and Applicability

               3.1.6.1.1  Principle - Total airborne mercury
(organo and metallic) was determined by using thermal decomposi-
tion at 800°C of all mercury forms to elemental mercury followed
by analysis with a mercury vapor photometer.  (Note - Furnace
vapor is cooled to atmospheric temperature before entering pho-
tometer. )

               3.1.6.1.2  Applicability - System was designed to
monitor mercury and organo-mercury fungicides in the atmosphere
where seed treatment is carried out and where treated seed is
stored in large quantities.  Technique should be applicable in
any organo-mercury or mercury vapor atmosphere where the total
organic vapor does not exceed the combustion capacity of the
decomposition tube.  (Note - Excessive levels of hydrocarbons,
dust, ozone, etc., can interfere with the photometric analysis.
All organic material must be decomposed and oxidized prior to the
photometric measurement.)  There is not sufficient data to indi-
cate the efficiency at which the system operates for decomposition
and measurement of mercury-containing particulate.

     Organo-mercury compounds are decomposed quantitatively at
a temperature of 800°C and at a flow-rate of 5 liters/minute.

          3.1.6.2  Range and Sensitivity

               3.1.6.2.1  Range - 0-750 yg/m3.

               3.1.6.2.2  Sensitivity - Estimated as <0.01 yg/m3.

          3.1.6.3  Interferences

               3.1.6.3.1  Chemical - All organic contamination
must be eliminated in the combustion process.  The effect of
ozone, generated by the mercury lamp, should be minimized by
proper ventilation to limit accumulation of ozone in the source
chamber.  Interferences normally encountered in the mercury
photometric measurements would also be expected in the read-out
system of this procedure.  (Ref.-See Section 3.1.6.1^.)

               3.1.6.3.2  Physical - Light scattering and atten-
dant loss of radiation must be considered if the system is used
in atmospheres of high dust loadings.  The filter located after
the furnace and prior to the photometric detector should minimize
the scattering effect from particulate.
                               231

-------
          3.1.6.*t  Accuracy, Precision and Stability

     No data reported.  Should have the same accuracy and pre-
cision as the general mercury vapor photometric systems.

          3.1.6.5  Apparatus

               3.1.6.5.1  Furnace - Tube furnace (Section 1A of
CR300 - A. Gallenkamp & Co. Ltd.) with silica tube (1/2-in. bore,
20-in. length bent into U-tube of uneven length with longer arm
of 10 in.).

               3.1.6.5-2  Condenser - 5-in. double surface type.

               3.1.6.5.3  Mercury Vapor Concentration Meter -
Type £3^72 (Hendry Relays Ltd).

               3.1.6.5.^  Ancillary Equipment - Flowmeter, needle
valve, filter, tubing, vacuum pump, pyrometer, thermocouple,
energy regulator (220 V - 1,300 W), recorder (10 mv).

          3.1.6.6  Time Cycle for Sampling and Measuring

     Samples are taken and measured continuously and instan-
taneously .

          3.1.6.7  Calibration Procedure

     Calibration was performed by drawing a small flow of air
(10 to 100 ml/min) over a metallic mercury reference held at a
constant temperature until saturation of the air with mercury
is achieved.  This flow was diluted with additional air to re-
duce the mercury concentration to a measurable level at 25°C and
at the same time increasing the air flow to the required rate of
5 liters/minute.  These data were extrapolated back to a zero
flow of air over the mercury, giving the concentration at satura-
tion point.   This was repeated at several fixed temperatures and
the results were compared to published data by Jacobs (19^9)-,
Lindstrom (1958)-, and Handbook of Chemistry and Physics.
-Jacobs, M. B., Analytical Chemistry of Industrial Poisons,
 2nd Ed., p. 222 (London:  Interscience) (19^9).

^Lindstrom, 0. J. Agric.  Ed. Chem.  6_, 283 (1958).
                               232

-------
          3.1.6.8  Method of Sampling and Sample Preparation

     The air sample was drawn through the apparatus at 5 liters/
minute.  When the air flow was passed through the furnace, total
mercury (organo and metallic) was determined.  When the air flow
bypassed the furnace, metallic mercury only was read and the
organo-mercury was calculated by difference.

          3.1.6.9  Multi-element Application

     Technique, as designed, is specific for mercury.  By using
atomic absorption equipment, the concept of pretreating the
sample in a tube furnace prior to absorption measurement could
be adapted to other volatile organo-metallic compounds (e.g.,
lead).

          3.1.6.10  Physical Dimensions

     Height - 23 in.; width - 21 in.; depth - 12 in.; and
weight - 88 Ibs.

          3.1.6.11  Unit Output

     Micrograms/meter3.


          3.1.6.12  Safety Hazard

     Moderate electrical voltages.

          3.1.6.13  Recommendations for Method Improvement

     The efficiency of the decomposition technique must be estab-
lished for mercury-containing particulate before the system can
be applied as an air pollution (universal) monitor for mercury.

     The air flow rate (5 liters/minute) may be a limiting factor
in an application to stack monitoring.  Redesign of the instru-
ment (higher temperatures and/or longer heated zone or double
pass) may be necessary to attain sufficiently high decomposition
efficiencies at higher flow rates.

     Consideration should be given to maintaining the sample at
warm to moderate temperatures after the furnace to insure against
loss of mercury on the glass walls of the system after vapors
pass into and through the cooling condenser.

          3.1.6.1*1  Reference

     Hamilton, G. A. and A. D. Ruthven, "An Apparatus for the
Detection and Estimation of Organo-mercury Dusts and Vapours in
the Atmosphere," Laboratory Practice 15, 995-997 (1966).


                                233

-------
     3.1.7  Portable Mercury Vapor Photometer

          3.1.7.1  Principle and Applicability

               3.1.7.1.1  Principle - Mercury vapor concentrations
are determined by measuring change in ultraviolet radiation inten-
sity resulting from the absorption of radiation at the mercury
resonance line at 2536.5^.  The detection system responds to the
difference in output voltages for dissimilar, but balanced photo-
tubes.  One of the phototubes responds only to visible radiation,
whereas the other responds to both visible and ultraviolet radia-
tion.
               3.1.7.1.2  Applicability - System was designed as
a self-contained survey meter for mercury vapor only.  Weight and
size factors permitted portability and battery power allows use
in inaccessible sites.

     Device is not directly applicable to stack monitoring, but
with suitable modification the concept is attractive for a very
low cost, stable, mercury vapor monitor.

     The system would be subject to potential interferences from
ultraviolet absorbing materials, but would compensate, in part,
for scattering losses due to particulate.  Potential interferences
could be eliminated by using a mercury isolation-desorption step.

          3.1.7.2  Range and Sensitivity

               3.1.7.2.1  Range - 0.04 to 0.6 mg/m3.

               3.1.7.2.2  Sensitivity - Detection limit estimated
from calibration curves as approximately 0.02 to 0.04 mg/m3.

          3.1.7.3  Interferences

               3.1.7.3.1  Chemical - Organic compounds, ozone, and
other materials which absorb ultraviolet radiation.

               3.1.7.3.2  Physical - Radiation scatter effect re-
sulting from particulate, etc., would be minimized by comparison
of ultraviolet and visible radiation.

     Magnetic field effects are minimized by housing amplifier
tubes in a magnetic shield.

          3.1.7.4  Accuracy, Precision and Stability

     Instrument reproducibility at any mercury vapor concentration
is 2% of full scale.  Warm-up period of 20 minutes is required to
minimize drift.  After this period, the drift is less than one-and-
a-half meter divisions per hour.  The instrument requires minor
zero adjustment with shift in ambient temperature.


                               234

-------
     Instrument was in operation for over 500 hours without
serious malfunction.  Only maintenance was the redistribution
of mercury droplets in the mercury lamp.

          3.1.7.5  Apparatus

               3.1.7.5.1  Ultraviolet Absorption Cell - Flow-
through chamber containing phototubes, UV source lamp, calibra-
tion filter, and light baffle.

               3.1.7.5.2  Ultraviolet Light Source - GE B-H6
mercury discharge lamp selected for low current drain.  Operating
current of 0.75 milliampere provided adequate light intensity.
Temperature effect on emission intensity and ratio of ultraviolet
to visible emission was minimized by jacketing the source with a
quartz envelope.  The trapped air served as a thermal insulating
medium.  Although the warm-up time was increased, the zero shift
due to changes in ambient temperature was reduced.

               3.1.7.5.3  Detection System - A Type 93*1 phototube
which responds only to light in the visible spectrum is used as
a reference.  A Type 935 which responds to light in both the vis-
ible and ultraviolet regions is used as a signal phototube.
Metering' circuit uses two balanced CK526AX subminiature tubes to
amplify outputs of phototubes and to develop sufficient power to
operate a 20-microampere meter.  The CK525AX tubes were selected
because of low filament and power requirements, ability to with-
stand mechanical shock without shift in response, and long term
stability.

               3.1.7.5.^  Sample Blower - Low power, battery
operated (6V, 0.25 A)centrifugal blower operating at a flow rate
of 0.5 cfm.

          3.1.7.6  Time Cycle for Sampling and Measuring

     For batch samples, sampling and measuring time is five
seconds; flushing time is ten seconds.

          3.1.7.7  Calibration Procedure

     Instrument was calibrated with a mercury vapor generator
(not identified).  Field calibration is obtained by checking two
points on the response curve.  Instrument is adjusted for zero
response with no mercury vapor present.  (Note - Hopcalite filter
used if necessary.)  The second point, full scale, was provided
by using a filter made of Pyrex glass which was inserted between
lamp and signal detector to produce an output equivalent to
3.5 mg Hg/m3 of air.
                                235

-------
          3.1.7.8  Method of Sampling and Sample Preparation

     Blower pushed air through instrument at 0.5 ftVmin.  No
sample pretreatment was used.

          3.1.7'9  Multi-element Application

     Designed specifically for mercury vapor.

          3.1.7.10  Physical Dimensions

     Weight - 18 Ibs; size - occupies 0.3 ft3.

          3.1.7.11  Unit Output

     Microamps convertible to mg/m3.

          3.1.7.12  Safety Hazard

     None.

          3.1.7.13  Recommendation for Method Improvement

     Some form of compensation or sample treatment is needed to
minimize effect of other ultraviolet absorbing species.

     Placement of phototubes, lamp, and calibration filter in the
absorption cell would be critical in a stream containing high
loading of particulate.  Separate optical and sample transport
paths are needed.  Problems with particulate deposition would
also be encountered with the blower on the inlet side.

     Concept could be modified for use as the detection sub-
system in a thermal decomposition - Hg isolation and desorption -
detection system for monitoring total Hg emission.

          3.1.7.11*  Reference

     McMurray, C. S., and J. W. Redmond, "Portable Mercury Vapor
Detector," Report No. Y-1188, Union Carbide Nuclear Company,
Oak Ridge, Tennessee, Jan. 7, 1958.
                               236

-------
3.2  Flame Atomic Absorption and Atomic Fluorescence Spectroscopy

     Atomic absorption measurements of mercury with conventional
flame techniques are relatively insensitive when compared to
results from nonflame methods.  Detection limits of 10 yg/ml of
Hg in water with conventional flame techniques and 0.001 yg/ml
of Hg in water with nonflame methods are common.

     The principal problem with flame techniques is the unfavor-
able distribution of mercury in the Hg2 + , Hgf+ and Hg° states.
Greater sensitivity is obtained for mercury (I) than for mer-
cury (II), and further increase in sensitivity results when
mercury(I) is reduced to elemental mercury (ref. 19).  An advan-
tage of the nonflame technique results from the fact that the
sample is introduced into the optical path as elemental mercury
     Atomic fluorescence spectroscopy has been used to determine
mercury in aqueous media (ref. 20-22).  Elements which emit reso-
nance lines in the UV (below 3000 S) can be adapted readily to
atomic fluorescence measurements.  In addition, elements with
resonance lines in the UV and with weak absorption (Hg 2537 A*)
are particularly suited to atomic fluorescence flame spectrometry ,
This factor can produce a major advantage of atomic fluorescence
over atomic absorption and is due principally to better quantum
efficiencies (ref. 20-22).  Improvements in sensitivity can reach
a factor of 10 (0.08 yg/ml as compared to 0.8 yg/ml).

     Additional enhancement of sensitivity and attendant lowering
of the detection limit can be attained by using the reduction of
Hg to metal during atomization (ref. 23).  With the introduction
of a reducing agent, SnCl2, fluorescence intensity is increased
by a factor of 100, permitting a detection level of 0.002 yg/ml
at a confidence level of 95%.  A number of elements and ions pro-
duce interferences.  The S2~ ion must be absent; 50 yg Au, Pt,
Cr(VI), and 0.5 nig Ag all in 1 ml interfere in the determination
of 1 mg Hg/ml.

     By using an interference filter with auxiliary optics of
low /-number (ref. 24) or by using a solar-blind photomultiplier
(ref. 25) in place of a conventional monochromator , nondispersive
atomic fluorescence analyses for Hg, Zn and Cd in water are
possible.  For mercury, the limit of detection with the filter
and photomultiplier is an order of magnitude lower than that ob-
tained by using a monochromator (ref. 24).  An optimum limit of
detection of 2.5 x 10 ^yg/ml was obtained with the solar blind
photomultiplier (ref. 24).

     One of the more commonly accepted methods of monitoring
ambient air for mercury is based on collecting free and bound
mercury on activated carbon and desorbing mercury vapor with
                                237

-------
heat into the optical path of a flame atomic absorption unit.
The advantages are:  (a) concentration of mercury,  (b)  rapid
desorption as vapor, (c) minimization of interferences,
(d) conversion of all forms of mercury into elemental mercury,
and (e) high sensitivity based on the concentration effect and
desorption as a unit volume (or mass).  Technology  related to
the collection on activated carbon and desorption into  the atomic
absorption flame is reported in the following subsection.
                               238

-------
     3.2.1  Determination of Mercury with Impregnated Charcoal
            and Flame Atomic Absorption Spectrophotometry

          3.2.1.1  Principle and Applicability

               3.2.1.1.1  Principle - Mercury is desorbed from
impregnated charcoal collectors and analyzed directly for mercury
with an atomic absorption sampling boat assembly in an oxidizing
air-acetylene flame.  Measurements are made based on the absorp-
tion of the mercury resonance line at 2537 A.

               3.2.1.1.2  Applicability - Technique was devised
to measure mercury vapor, volatile mercury compounds, and mercury-
bearing dust in industrial samples.  Dusts were retained on glass
wool, and the volatile mercury and its compounds were adsorbed on
the impregnated charcoal.  Although a manual technique, the total
analysis time for one air sample is less than three minutes.

     Speed, sensitivity, collection efficiency and apparent free-
dom from interferences are major advantages.  Main disadvantages
are manual operation and open flame excitation.

          3.2.1.2  Range and Sensitivity

     Limit of Detection - 0.02 yg Hg for a 10-liter (0.01 m3)
air sample.

          3.2.1.3  Interferences

               3.2.1.3.1  Chemical - All commercially available
impregnated charcoal for mercury must be heated at 600-800°C for
one hour prior to use.  The procedure removes excess impregnant
and all interfering volatile organic solvents from the charcoal.

     Author found that levels much greater than threshold "limit
values" of benzene, toluene, acetone and carbon tetrachloride
which absorb 2537 A radiation must be present, before sufficient
amounts are adsorbed on the charcoal to interfere with the
analysis of atmospheric mercury.

               3.2.1.3.2  Physical - Tantalum boat deteriorates
after repeated use and a standard sample must be analyzed after
every 4 or 5 samples to determine loss in sensitivity.  Boat must
be replaced when analytical sensitivity falls to about 75% of
the initial value.

          3.2.1.*!  Accuracy, Precision and Stability

     No specific data are given; authors imply that method is
more precise than other ultraviolet methods at levels less than
0.02 mg/m3.  Trapping efficiencies appear to be excellent with
no apparent slippage to second unit of charcoal.


                                239

-------
          3.2.1.5  Apparatus

               3.2.1.5.1  Atomic Absorption Spectrophotometer -
Perkin-Elmer Model 303 or equivalent.

               3.2.1.5.2  Tantalum Sampling Boat System

               3.2.1.5.3  Triple Slot Burner - Oxidizing air-
acetylene flame.

               3.2.1.5.*t  Recorder Read-out - Equipped for total
system scale expansion of 5X.

               3.2.1.5.5  Impregnated Charcoal - Barnebey-Cheney
Co. Type 580-12 or 580-20 (20-40 mesh).(Note - Charcoal impreg-
nated with trace amounts of iodine and iron powder greatly
increase,adsorption efficiency of activated carbon for mercury,
without interfering with recovery of trapped mercury.)

               3.2.1.5.6  Sampling Tube - 6-in. long, 4-mm diam-
eter, packed with two one-inch sections (180 mg each) of charcoal.
The two charcoal sections are separated and retained by fiber-
glass plugs.

          3.2.1.6  Time Cycle for Sampling and Measuring

     Ten liters of air are taken for analysis, but no data on
sampling rates are reported.  The total analysis time for one
air sample is less than three minutes.

          3.2.1.7  Calibration Procedure

     Calibration curve prepared from atomic absorption standards
containing known amounts of mercury.

          3.2.1.8  Method of Sampling and Sample Preparation

     Mercury vapor and its volatile compounds trapped on impreg-
nated charcoal are desorbed from the charcoal in the air-acetylene
flame of an atomic absorption spectrophotometer.  The charcoal is
introduced into the flame in a tantalum boat (std. PE accessory).

          3.2.1.9  Multi-element Application

     By varying instrument conditions, flame temperatures, and
radiation sources, a variety of individual elements could be
determined, but not a large number simultaneously.

          3.2.1.10  Physical Dimensions

     Laboratory instrument (atomic absorption spectrophotometer)
approximately 5* x 2' x 21.


                               240

-------
          3.2.1.11  Unit Output

     Response convertible to mass (micrograms) or mass/volume.

          3.2.1.12  Safety Hazard

     Intense, open flame.

          3.2.1.13  Recommendations for Method Improvement

     As designed, the method involves manual operation and con-
sumable charcoal samplers and is not directly applicable to
continuous monitoring.  Consideration should be given to the
design of a thermal desorption system which would permit cycling
of the collection tube.  Obviously, a study of the trapping effi-
ciency deterioration during cycling is needed.  Loss of iodine
may be a major problem.  Desorption with induction heating,
rather than the air-acetylene flame, should be considered.

          3.2.1.14  Reference

     Moffitt, A. E. and R. E. Kupel, "A Rapid Method Employing
Impregnated Charcoal and Atomic Absorption Spectrophotometry
for Impregnated Charcoal and Atomic Absorption Spectrophotometry
for the Determination of Mercury in Atmospheric, Biological,
and Aquatic Samples," Atomic Absorption Newsletter, 9_, 113-118
(1970).
                               241

-------
3.3  Emission Spectroscopy

     Generally, direct application of spectrographic analysis for
mercury is reported as unsatisfactory.   However, this observation
is based on the analyses performed by excitation of batch samples
from crater electrodes.  The problem is volatilization of the
metal from the crater of the electrode prior to its activation.
This problem should not exist when the mercury containing air-
borne specimen (vapor or particulate) is passed continuously
through the excitation zone of an arc,  spark, and radiofrequency
or microwave induced plasma.

     The technique of Webb, et al (ref. 26,2?) used for monitor-
ing beryllium in air continuously, or Fromm and V. Oer (ref. 28)
used for measuring beryllium and mercury in air should be appli-
cable to monitoring mercury from stationary source emissions.

     Excitation of optical emission at 2537 & with low power
microwave radiation (ref. 29) resulted in a detection limit of
0.1 nanograms.  (Note - Lower limits are possible.)  This ap-
proach is presented in more detail in the following sub-section.
                               242

-------
     3.3.1  Determination of Mercury Compounds by Emission
            Spectrometry In a Helium Plasma

          3.3.1-1  Principle and Applicability

               3.3.1.1.1  Principle - Optical emission of the
2537 X atomic mercury line is induced with a microwave-powered
inert gas (helium) plasma generated at a pressure of 5 to 10 mm.
The intensity of the emission is related to the quantity of
organo-mercury compound present in the effluent from a gas
chromatographic system.

               3.3.1.1.2  Applicability - Technique was developed
as a gas chromatographic detection system to measure extremely low
concentrations of organo-mercury compounds which were separated
chromatographically and would be isolated in an inert gas matrix.
The detector could be applied to the measurement of mercury vapor
and volatile mercuric compounds in an inert gas matrix.

     Gas chromatographic isolation of the organo-mercury compounds
in the inert and monoatomic, helium matrix of the gas chromato-
graphic effluent is necessary to minimize the quenching effect on
the microwave excited plasma by molecular species, e.g., HaO, Oa,
Na , organic solvents, present in the original sample.

     The application to the direct analysis of total mercury in
air is doubtful unless an inert matrix (helium or argon) can be
substituted for the air.  This would be possible through a mech-
anism involving chromatographic removal of the air from an aliquot
sample, or (less likely) if the concentration of mercury and
instrument sensitivity were high, by dilution with an inert gas.

          3.3.1.2  Range and Sensitivity

     Range - Linear response from 0.1 to at least 100 nanograms
of methylmercuric chloride.

     Sensitivity - 0.1 nanogram as dimethylmercury or methyl-
mercuric chloride.

          3.3.1.3  Interferences

               3.3.1.3-1  Chemical - No interferences when used
as a chromatographic detector.  There is some question whether
excessive molecular species would quench the helium plasma in
non-chromatographic applications.

               3.3.1.3.2  Physical - Technique is restricted to
measurements involving vapor forms of mercury.  Direct analysis
of mercury particulate is doubtful unless metallic mercury could
be generated in a pre-analysis step.
                               243

-------
          3.3.1.**  Accuracy, Precision and Stability

     Data obtained for methylmercuric chloride was comparable to
that determined with an electron capture detector.

          3.3.1.5  Apparatus

               3.3.1.5.1  Discharge Tube - 1.5-mm bore, 8-mm o.d.
quartz tube.

               3.3.1.5.2  Cavity - Tapered microwave 2450 MHz
cavity fitted with two irises to facilitate tuning.

               3.3'1.5.3  Ancillary Equipment - Recorder, power
supplies, flowmeters and pressure regulators.

          3.3.1.6  Time Cycle for Sampling and Measuring

     Approximately four minutes of chromatographic time is re-
quired to isolate methylmercuric chloride in the helium matrix
for measurement with the emission detector.  The excitation and
recording of the optical emission requires less than one minute.

          3.3.1.7  Calibration Procedure

     Known amounts of organo-mercury compounds were injected into
the gas chromatograph.

          3.3.1.8  Method of Sampling and Sample Preparation

     Known volumes of liquid sample were injected into a gas
chromat ograph.

          3.3-1-9  Multi-element Application

     Could be modified for application to other volatile metal-
and nonmetal-containing compounds.

          3.3.1.10  Physical Dimensions

     No data given, but detection unit should be relatively small

          3.3.1.11  Unit Output

     Recorder (mv) readings convertible to mass units (nanograms)

          3.3.1.12  Safety Hazard

     Moderate electrical voltages.
                               244

-------
          3.3.1-13  Recommendations for Method Improvement

     Consideration should be given to use of a several stage
system to (a) thermally decompose mercury compounds and particu-
late into mercury vapor, (b) separate elemental mercury from the
air matrix (chromatographically or collection system), and
(c) use the low pressure helium plasma (microwave induced) to
measure the total amount of mercury present in air samples.
Analytical times less than 5 or 10 minutes are possible.

          3.3.1.14  References

     Bache, C. A. and D. J. Lisk, "Gas Chromatographic Deter-
mination of Organic Mercury Compounds by Emission Spectrometry
in a Helium Plasma," Anal. Chem. ^3_, 960-963 (197D-

     Bache, C. A. and D. J. Lisk, "Simple Tuning Device for
Microwave Tapered Matching Cavity," Anal. Chem. 40, 2224 (1968).

     Bache, C. A. and D. J. Lisk, "Selective Emission Spectro-
metric Determination of Nanogram Quantities of Organic Bromine,
Chlorine, Iodine, Phosphorus, and Sulfur Compounds in a Helium
Plasma," Anal. Chem. 39, 786-789 (1967).
                               245

-------
3.4  Formation of Colored Complexes

     Several methods for measuring atmospheric mercury vapor
have been developed based on the reaction of mercury with copper
diiodide to form a double Hg-Cu iodide salt (CuI-HgI2).  The
reaction has been used in several forms:  (a) Hg absorbed in I2
and KI solution (ref. 30,31), (b) CuI2 deposited on silica gel
(ref. 32), and (c) CuI2 deposited on filter paper (ref. 33,34).
Depending on the sampling time (5-20 minutes), detection of
mercury vapor in air down to <0.15 yg/1 is reported.

     The most promising of the approaches for short-term inter-
mittent monitoring would be a modified version of Palalau method
(ref. 34).  The method is based on the color formed when mercury
vapor in the atmosphere reacts with copper diiodide on filter
paper.  Color varies from cream-white to orange-yellow as a func-
tion of the mercury concentration; color intensity is measured
by determining the luminosity intensity of the paper at 510 nm
with a spectrophotometer adapted to reflectance measurements.


3.5  Neutron Activation Analysis

     The lengthy cooling times (20-30 days) required after irra-
diation and prior to counting preclude the application of neutron
activation analysis (NAA) to continuous monitoring of mercury in
complex matrices similar to typical ambient air particulate.  For
nondestructive NAA measurements (no radiochemical separation), a
commercial activation analysis service reports a detection limit
of 0.810 yg Hg for particulate matter (^14 mg) collected from
urban air.  By using radiochemical separations, the activation
analysis service coutinely determines mercury to 3 nanograms in
a 1-gram sample and to 10 parts per trillion in 50-g samples with
extended irradiations (up to 100 hrs).

     In other nondestructive neutron activation analyses of air
pollution particulates (ref. 35,36), detection limits of 0.01 pg
Hg are reported after irradiation for 2-5 hours, cooling for
20-30 days, and counting for 4000 seconds.


3.6  Radioisotoplc Method

     Magos (ref.  37), and Magos and Clarkson (ref. 38) developed
an isotopic method for determining metallic mercury in air.  The
sample is passed through a solution containing Hg(II) as a com-
plex salt or chelate, e.g., EDTA complex with 203Hg.  The 203Hg
released from the reagent into the gas stream is absorbed in
Hopcalite or KMnO^ and measured with a scintillation counter.
Good reproducibility is reported for 0.016 to 1.2 pg Hg/liter
of air.
                               246

-------
     3.6.1  Radlochemlcal Determination of Mercury Vapor in Air

          3.6.1.1  Principle and Applicability

               3.6.1.1.1  Principle - Air containing mercury is
passed through an aqueous solution of 203Hg-mereuric acetate and
KC1 to promote isotope exchange.  The issuing air contains the
same concentration of mercury, but labelled and with the same
specific activity as the reagent solution.  The 203Hg is absorbed
on Hopcalite and estimated by gamma scintillation counting.

               3.6.1.1.2  Applicability - Method can be used for
short term sampling (5 minutes) or continuous monitoring of
metallic mercury vapor in air at flow rates of 0.5 to 1.5 liters/
minute.

          3.6.1.2  Range and Sensitivity

               3.6.1.2.1  Range - Range is approximately
0.01 yg/liter to 1.2 yg/liter.

               3.6.1.2.2  Sensitivity - Detection limit depends
on the amount of sample collected, but is approximately
0.01 yg/liter.

          3.6.1.3  Interferences

               3.6.1.3.1  Chemical - No effects were observed
with test mixtures of I/? household gas (containing IQ% carbon
monoxide), 1% acetylene, 25? C02 and 0.1/1 S02.

     Effect of H2S was only determined qualitatively.  H2S pro-
duced a black precipitate in the reagent and a rapid decrease
in the efficiency of the exchange reaction.  This effect was
prevented by using a midget impinger with 15 ml 2% cadmium
chloride solution or IN NaOH before the reagent..

     The lack of interference from S02 bears noting since solu-
tions containing chloride/HgC!2 in the ratio of 2:1 can form
stable disulphitomercurate complexes (see West and Gaeke method
for absorbing S02).  Fortunately, with higher ratios, as used
in the exchange reagent, the stability of the complex decreases -
if it is formed at all.

     High concentrations of air contaminants, which react with
Hopcalite (solid mixture or copper and manganese oxide), can
exhaust the absorber in prolonged tests.

     The efficiency of the exchange reaction was not affected
by pH changes from a normal of 6.2 to 9.6 with NaOH addition
or to 1.5 with citric acid added.
                               24?

-------
               3.6.1.3.2  Physical - Temperature of the reagent
bath had no influence on the exchange reaction In the range 5°C to
30°C, but at 9°C, the efficiency decreased by approximately
     Variations in air-flow from 1.0 liter/minute to 1.5 liter/
minute produced no effect.  However, reduction in flow to
0.5 liter/minute resulted in increased efficiency of approxi-
mately 3.5%.

          3.6.1.4  Accuracy, Precision and Stability

     The standard deviation of the method is 0.004 yg/liter in
concentrations up to 0.2 yg/liter, and is 0.075 yg/liter in the
range 0.2-1.2 yg/liter.

          3.6.1.5  Apparatus

               3.6.1.5.1  Collection Train - Plow regulator and
flowmeter; 30 ml impinger containing 20 ml reagent; one or more
Hopcalite absorbers.

               3.6.1.5.2  Exchange Reagent - 2.5 x lO^M mercuric
acetate and 0.25M potassium chloride in glass distilled water
labelled with 203Hg++.  (Note - Increase in mercury concentration
caused by labelling was always less than 1.05?.)  Reagent usually
contained <2 yc/20 ml.

               3.6.1.5.3  Radlochemlcal Detector - Well-shaped
Nal (gamma) scintillation detector (Detector N664B) with a
sealer (N610A. Ekco Electronics, Ltd) having a counting effi-
ciency for 2oSHg of
               3.6.1.5.4  Absorbers - Polyallomer tubes
(5/8 x 3-in.) (Beckman Instr. Inc., Spinco Division) containing
3 g of activated Hopcalite (solid mixture of copper and man-
ganese oxide).  Inlet and outlet tubes were Sterivac polythene
cannulae (bore 2.0 mm, wall thickness 0.5 mm).

               3.6.1.5.5  Ancillary Equipment - Lead shielding
for midget impinger, suction pump, special saturation system
for calibration, recorder.

          3.6.1.6  Time Cycle for Sampling and Measuring

     Measurements can be made continuously if the absorber is
kept in the well of the gamma scintillation counter.  Short-term
sampling can be accomplished in 5-minute sampling cycles followed
by removal and replacement of the absorber tube.  Counting times
are usually 100 seconds .

     Sampling times are dependent on the analysts requirements
for precision and sensitivity and can range from 2-20 minutes.
                               248

-------
          3.6.1.7  Calibration Procedure

     Calibration was based principally on literature data for the
concentrations of mercury vapor in air at saturation at various
temperatures.  Air was saturated with mercury at various tem-
peratures and introduced directly or after dilution with known
amounts of additional air.

     A specially designed 1 yg calibrator was fabricated from a
plastic syringe.  The syringe provided a way to deliver a volume
of mercury saturated air containing 1 yg mercury without the
errors which may attend mixing two air streams.

     Calibration of the technique with the 1 yg calibrator also
provided a means for measuring the efficiency of the exchange
reaction (97-7/2) to incorporate a correction factor for the ab-
sorbed (Hopcalite) activity.

     Since the efficiency of one absorber was considered quanti-
tative (99.^/0 and within counting errors, no correction for
absorption efficiency was required.

          3.6.1.8  Method of Sampling and Sample Preparation

     Air samples containing metallic mercury vapor are drawn
through an isotope exchange solution at 1 liter/minute.  Radio-
active Hg produced in the isotope exchange is absorbed on
Hopcalite and counted with a gamma scintillation counter.

          3.6.1.9  Multi-element Application

     Methodology is specific for mercury; however, by suitable
changes in reagents and absorber, applications to other volatile
elements having radioactive isotopes are possible.

          3.6.1.10  Physical Dimensions

     No data reported; collection train and absorption system
would be relatively small.  A laboratory-type scintillation
counting system was used and undoubtedly would be the largest
component.

          3.6.1.11.  Unit Output

     Counts/100 seconds convertible to yg Hg or yg Hg/liter.

          3.6.1.12  Safety Hazard

     Moderate electrical voltages are used in conventional am-
plifiers, etc.  Radioactivity could be minor problem, but low
activity in the reagent (<2 yc/20 ml) and Hopcalite permits
disposal as normal waste.
                                249

-------
          3.6.1.13  Recommendation for Method Improvement

     Application of improved gamma radiation measurements with
semi-conductor detectors may permit shorter sampling and count-
ing times, and higher sensitivities.

          3.6.1.1**  References

     Magos, L.j "Radiochemical Determination of Metallic Mercury
Vapour in Air," Brit. J. industr. Med. 23., 230-236 (1966).

     Magos, L. and T. W. Clarkson, "Isotope Method for Deter-
mining [Metallic] Mercury in the Air," Brit. 1,130,921 (16 Oct.
1968); C.A. 70., lH229p.
                               250

-------
3.7  Argon lonization Detection

     Krestovnikov and Sheinfinkel (ref. 39) collected mercury on
activated carbon and desorbed the mercury into a stream of argon
to be detected with an argon ionization detector (three electrode
type with 10 mg 1'*7Pm radioactive source).  Results agreed well
with those of a colorimetric method and indicated a detection
threshold of 1 x 10 sg/l with collection time of 10 minutes.

     By improved detector and electronic design and by using gold
wool as a trapping system, greater sensitivity and collection ef-
ficiency can be attained which would permit shorter term cycles
for intermittent (<10 minute) monitoring of stack emissions.

3.8  Condensation Nuclei Monitor

     Environment One Corporation manufactures a Mercury Vapor
Detecting System consisting of two units, a Mercury Converter
and a Condensation Nuclei Monitor, for measuring total atmospheric
mercury.  Mercury, collected by drawing air through a cartridge
containing a plug of silver wire is desorbed with heat into a
cloud chamber.  The desorbed mercury vapor is irradiated by ultra-
violet light, forming submicroscopic particles of HgO.  Water
vapor condenses on the nuclei in the cloud chamber and droplets
are formed which may be counted by optical and electronic circuits.

     Preliminary tests at Battelle's Columbus Laboratories indi-
cate insufficient sensitivity, unsatisfactory duplication of
replicate samples and questionable short term stability (ref. 40).


*>.   SAMPLE COLLECTION TECHNIQUES

     In most cases of ambient air monitoring, some collection or
pretreatment procedure is used to concentrate the mercury or to
eliminate interfering materials.  A variety of sample pretreat-
ment or collection steps are used and representative types are
listed in Table II.  Generally, the mercury is collected as vapor
in impingers containing various fluids, on activated or modified
carbon, or on a variety of metals which form amalgams.

     As shown in Table II, many collection techniques can be used
for specific forms of mercury, but there is no universal collec-
ting media for all forms.  Two media - 0.1N iodine monochloride
in 0.5M hydrochloric acid (I), and fluidized bed of activated
carbon impregnated with iodine (II) - have been applied to a
greater variety of mercury forms than the others.  However, the
efficiency of (I) for collecting airborne mercury particulate has
not been fully tested.  By arranging a special sequence of collec-
ting media, complete collection of all forms of mercury emissions
are possible.  A particulate filter combined with either (I) or
(II) would provide a good, general collection system.
                               251

-------
                                             Table II

                          COLLECTING MEDIA FOR MERCURY AND ITS COMPOUNDS
ru
\ji
ro
           Collecting Media
Water

25% Ethanol in water

Aqueous solution sodium
   carbonate

IN Hydrochloric acid
Alkaline hyprobromite

Aqueous solution NaOCl-

Aqua regia-water (1:3) spray
Acid permanganate solution
   (several cone.)
      0.1N Permanganate

      0.5N Potassium permanganate
         in 2N sulfuric acid
      Aqueous iodine (0.25?) and
         potassium iodide (3?)
      Aqueous iodine (0.12?) in
         1.5% solution of KI
      Isopropanol
                                    Mercury Form
Mercury-bearing dusts
Mercury-bearing dusts
w/fat and oil
(a) Mercury vapor
(b) Monomethyl and
    ethyl mercury salts
Mercury vapor
Mercury vapor

Mercury vapor

Mercury vapor
(a) Mercury vapor
(b) Alkylmercurials
(c) Monomethyl & ethyl
    mercuric salts
(a) Mercury vapor
(b) Dibutylmercury

Dimethylmercury


(a) Mercury vapor       -,
(b) Mercury bearing dust
(c) Dimethylmercury     -,
(d) Diethylmercury
(a) Mercury vapor
(b) Diethylmercury
(c) Ethyl mercuric
    chloride

Dibutylmercury
                              Efficiency
Adequate
Adequate

Not retained
Quantitative

Adequate
90? (1st stage)
99-100? (2nd stage)

>99?


Adequate

Poor

Adequate
Poor
                                                          Quantitative

                                                          Negligible at
                                                           20 liters/minute

                                                          Adequate
                          Ref.
   42


   43
   44

   44

   45
46 & 4?

   68

   48

   40

41 & 50




   51



   48

-------
                                        Table II - (Cont'd)
          Collecting Media
ru
VJ1
U)
     Alcoholic iodine (0.08$)
     Cold trap

     0.1N Iodine monochloride in
        0.5M hydrochloric acid
    Mercury Form
     Solid crystal iodine


     Activated carbon

     Activated carbon w/Cl

     Fluidized bed of activated
        carbon
(a) Mercury vapor
(b) Diethylmercury
(c) Ethylmercuric
    chloride

Mercury vapor


(a) mercury vapor
(b) Dimethylmercury
(c) Diethylmercury
(d) Ethylmercury
      chloride-
(e) Methylmercury
      chloride-
(f) Phenylmercury
      chloride-
(g) Mercury oxide-
(h) N-(Ethyl mercury)
      p-toluene
      sulfonamide-
(a) Mercury vapor
(b) Diethylmercury
Mercury vapor
Diethylmercury

(a) Mercury vapor-
(b) Diethylmercury
(c) Diphenylmercury
(d) Ethylmercury
      chloride
(e) Ethylmercury
      phosphate
(f) Methylmercury
      dicyandiamide
    Efficiency
Quantitative at
  25 liters/minute


Adequate at flow rate
   <5 liters/minute
95-100%
90-100?

90-100%

100%

100%

88 to 92%
94%
Adequate
<20%
Adequate
Adequate
                                                               90-100%
  Ref.
                                                                                          51
52 & 53
   54
   55
   56

-------
          Collecting Media
Ul
     Cadmium sulfide-glass fiber
Activated Mn02
MgSOu, KI, Ia, H20, glycol,
   acetone-
Filter paper impregnated
   w/aqueous solution
-  H.6% I2 and 55% KI
Palladium chloride
Selenium sulfide coated
   filter paper at 65°C
Selenium impregnated
   filter paper
Silver foil amalgamator
Silver sponge-
Gold leaf
Gold wire (M.4 cm2) @ 600°C
Gold impregnated glass frit
   or asbestos
Gold foil
Gold wool (25°-250°C)
Copper spiral
Copper diiodide
Copper foil
   Table II - (Cont'd)
    Mercury Form	
(a) same as fluidized
    bed of carbon with
    exception of
    diethylmercury
(b) Mercury-bearing dust
Mercury vapor
Mercury vapor

Mercury vapor

Mercury vapor
Mercury vapor

Mercury vapor

Mercury vapor
Mercury vapor
Mercury vapor
Mercury vapor
Mercury vapor

Mercury vapor
Mercury vapor
Mercury vapor
Mercury vapor
Mercury vapor
                                                              Efficiency
Adequate
Adequate
Quantitative

>92%-

Quantitative
Adequate

Adequate

Good
Good
<5%
95%
Quantitative at
  2 ftVhr
Quantitative
Quantitative
Quantitative
Quantitative
Quantitative
                          Ref.
                           56
                                                                                          57
                                                                                          58
   60
   61

   62

   63
64 & 65
   45
   66
   67

   69
70,71,72
   73
   34
   74

-------
         Footnotes  - (Table  II)

         ±1  g available  Cl/liter and 100  g NaCl/liter at  pH 10.7 and 50°C.
         -Trapped as vapor or added as  solid directly to  IC1 reagent.
         -Added as  solid directly to IC1  reagent.
         -Fluidized bed  of iodized carbon would be satisfactory for mercury vapor.
         -14 g anhydrous MgSCU,  1.2 g KI, 0.1 g I2, 0.4 ml H20, 1 ml ethylene glycol
              and 12 ml  acetone.
         -Requires  humidifier prior to  filter to attain high efficiency.
         -A  pre-filter @ 800-900°C of decomposed KMnOi* (prepared @ 600°C) is used to
              remove N oxides, halogens,  S-, P-, Sb-, Se, -, As-containing compounds,
ro
ui
VJ1

-------
5.   SAMPLE PRETREATMENT

     As an alternative to a sequence of collecting media to iso-
late mercury particulate, elemental mercury vapor, and organic
or inorganic bound mercury, thermal decomposition techniques
which convert the total mercury to one form, elemental mercury,
can be used.  Decomposition of bound mercury to elemental mercury
vapor produces a form of mercury which can be easily collected
or measured directly.

     The major problems in performing quantitative thermal decom-
positions of mercury compounds are the variations in compound
stability.  Unless the heating is performed rapidly at high
temperature or a moderately long residence time is used, volatile
halides and organomercurials may not decompose but escape as
undercomposed vapor.  Very hot oxyhydrogen flames, combustion in
air in furnaces at 800-900°C, and thermal decomposition in pres-
ence of reducing media (iron powder and CO, or SnCla) have all
been used.  Representative techniques are reported in the follow-
ing summaries.

5.1  Thermal Decomposition with Fe and CO

     Mercury dust is sublimed at 700°C for 30 minutes in a spe-
cially designed quartz tube.  Iron powder is used as a reducing
agent, and CO is used as a carrier gas to expel mercury vapor.
An asbestos filter is used to retain sulfur and other components
interfering with the mercury determination.  Mercury vapor in
the CO stream is absorbed by a 0.2% solution of I2 in 2% KI and
determined colorimetrically with a sensitivity down to 0.2 ppm
of Hg in the dust (ref. 75).

5.2  Thermal Decomposition in Air, Nitrogen or Oxygen

     Sample powder containing organic or inorganic bound mercury
is burned in a 35-cm silica tube of 12-mm inner diameter in a
stream of N2, 02, or air at 15-20 ml/min. at 800-900°C.  The
evolved Hg is absorbed on Ag sponge in a 3.5-cm silica tube of
8-mm diameter.  Volatile elements (P, As, Sb, S, Se, Te, F, Cl,
Br, and I) are removed by a 12-cm section of KMnOi* decomposition
product (prepared at 600°C) placed in the combustion tube.  The
errors in determining 1%, 0.1%, 0.01$, and 0.0015? were ±0.1%,
±0.015?, ±0.0015?, and ±0.0002%, respectively (ref. 65).

5.3  Thermal Decomposition at 1200°C

     Mineral sample (85 mesh) is heated up to 1200°C for 1 to
95 hours in a closed system to release mercury vapor (ref. 52).
                               256

-------
5.4  Combustion at 850°C in Air

     Combustion of 25 to 50 mg samples of mercury-containing
materials is attained in an air stream (100 cc/min) in less
than five minutes with gas burner or electric coil heating in
a heating cycle from room temperature to 850°C (ref. 67).

5-5  Decomposition of Powdered Samples at 850°C

     Powdered samples containing 0.001-1? Hg (from the Sn indus
try) were heated at 750-850°C in a stream of air (5-7 liters/
minute) for approximately 2-3 minutes, yielding photometric
analyses with coefficient of variation of 3-4? (ref. 76).

5.6  Reaction with SnCl2 and Thermal Vaporization

     Powdered sample is mixed with SnCl2 to reduce mercury com-
pounds; electrical heating of system vaporizes the mercury for
measurement with atomic absorption techniques.  Sensitivities
reach 10 6? Hg in powdered solid material with an average rela-
tive error of 10-15? (ref. 77).

5.7  Reaction with
     Organic or inorganic sample is heated at 650°C with 12:1.6:4
Na2C03-K2C03-Na202 in an air stream to distill elemental mercury
(ref. 78).

5.8  Thermal Decomposition of Powdered Geological Samples

     Pulverized ores and geological samples (up to 150 mesh) were
heated in a tube furnace under forced gas flow.  Evolved mercury
vapor was measured by atomic absorption spectrophotometry yield-
ing analyses down to 3 x 10 6% at sample weight of 0.5-2 g in
approximately 3 minutes.  The error of a single determination
varied from 8 to 30% (ref. 79).

5.9  Thermal Decomposition - Addition of Lime

     Decomposition of sulf ide-containing dusts, soils and minerals
is accomplished with gas-fired flame in the presence of lime at
a sample to lime ratio of 1:2.  Small quantities of interfering
sulf ides are eliminated by reaction with the lime.  For 0.25 g
sample, approximately 10 minutes are required (ref. 80).

5.10  Thermal Decomposition of Organo and Metallic Mercury
      Compounds at 800°C

     Total mercury in atmospheric dusts and vapors was determined
by mercury vapor photometric readings after thermal decomposition
of organo-mercury and inorganic mercury compounds in a continuous
flow system.  Furnace temperature was 800°C and the air flow was
set at 5 liters/minute (ref. 8l).

                               257

-------
5.11  Thermal Decomposition and Removal of Sulfur Compounds
      with Na2C03

     Inorganic pigment samples were heated directly and the
amount of mercury liberated was determined by the photometric
mercury vapor procedure.  Interference due to presence of S
was avoided by igniting the sample while mixed with NazCCb and
and passing the evolved gases through a silica gel trap (ref. 82)

5.12  Decomposition in Oxyhydrogen Flame

     Mercury containing organic and inorganic compounds were
decomposed in a hot oxyhydrogen flame of burner assembly of a
flame photometer.  The exhaust gases pass through condensers
and filter towers and then finally through the cell of a mercury
vapor meter for measurement (ref. 11).

5.13  Pyrolysls at 650°C and Removal of Interfering Substances
      with a Series of Traps

     Organic and inorganic samples are pyrolyzed at 650°C, gen-
erating mercury vapor from the mercury compounds present in the
sample.  The pyrolysate is passed over heated copper oxide and
swept through a series of traps (silver vanadate, Ascarite, Dehy-
drite, and anhydrous alumina) to remove interfering substances.
The mercury content of the gas stream is determined with a commer-
cially available mercury vapor meter.  Amounts of mercury as low
as 0.01 yg can be detected (ref. 83).

5.14  Pyrolysls at 700°C and Removal of Interfering Substances
      with Silver Reaction Chamber Followed by Gold Amalgamation

     Thermal decomposition is used to convert chemically bound
mercury into mercury vapor.  Silver wire removes chlorides and
gold foil isolates the mercury from the remaining contaminants.
Mercury desorbed by heating the gold amalgam is measured with an
atomic absorption spectrophotometer (ref. 84).


6.   TECHNIQUES FOR GENERATING KNOWN CONCENTRATIONS OF
     MERCURY VAPOR IN AIR

     Current methods for producing known concentrations of mer-
cury have several deficiencies:  (1) air and mercury are usually
not sufficiently heated to produce a saturated gas stream on
cooling, (2) excessive time is required to reach equilibrium
after concentration changes have been made, and (3) large size
of condensation and mixing vessels and the need for bulky con-
stant temperature apparatus limit field use.
                               258

-------
     The mercury vapor generator developed by Nelson, et al.
(ref. 85) and Nelson (ref. 86) occupies 0.1 m2 of bench space and
yields concentrations from 0.011 to 2.5 mg/m3 at flow rates of
6 liters/minute or greater with an accuracy of ±3%.  Concentration
changes are made simply and equilibrium is reached almost
instantaneously.

     Chase, et al.  (ref 40) constructed an apparatus employing
a permeation tube to supply a constant amount of mercury vapor
in a gas stream.  The permeation tube contained 10 grams of
mercury in a 7/8-inch section of silicone rubber tubing (1/2-in.
ID, 3/l6-in. wall thickness) and was plugged on both ends with
teflon.  By maintaining the permeation tube at 31°C and by pas-
sing nitrogen gas maintained at 31°C over the tube at 220 ml/min.,
a mercury vapor flow of 15 ± 1 ng/min. was attained after an
equilibration period of 24 hours.

     Considerable differences exist in data reported for mercury
vapor pressure over the 20° to 30°C temperature range.  Nelson,
et al. (ref. 85), Mayer (ref. 87), and Ernsberger and Pitman
(ref. 88) indicate similar data, but differ markedly with data
by Hill (ref. 89,90).


7.   PROBLEMS COMMONLY ENCOUNTERED IN MERCURY ANALYSIS

     Most direct analyzers (photometers) and collection systems
for air atmospheres measure or collect elemental mercury vapor
only.  Although filter devices can collect particulate and a
variety of impinger fluids have been used for organomercurials,
no single, efficient collection system for total mercury is
available.  Most promising solution to the problem is conver-
sion of all forms of mercury to elemental vapor by thermal
decomposition.

     Particulate (smoke, dust, etc.), iodine, sulfur compounds,
organic compounds,  and ozone, interfere with the highly sensi-
tive, inexpensive mercury vapor photometric method of analysis.
Several instrumental solutions to the problem are possible by
compensating for the background signal.  Alternate means of
trapping the mercury as an amalgam with subsequent desorption
as elemental mercury vapor provide an efficient solution.

     Sulfides, naturally present in a sample stream or as decom-
position products of sulfates and sulfur compounds in a flame,
may remove elemental mercury vapor.  A solution to the problem
uses gold to amalgamate the mercury prior to any filter on which
the sulfide can be deposited.  After venting the contaminating
gases, the mercury is desorbed for analysis.
                               259

-------
     Iodides trapped in filters may also combine with mercury in
similar manner as sulfides.   Use of a collection step with gold
will also minimize the potential loss.

     Mercury losses occur during processes using sample collec-
tion because of adsorption effects, hydrolysis, and decomposition
into metallic mercury which evaporates.  The presence of adsorbed
moisture on glass surfaces can cause removal of elemental mercury
from air or gas streams.  Lindstrom (ref. 11) reports that boro-
silicate glassware which had been cleaned with sulfuric acid-
dichromate solution shows negligible adsorption.  However,
silicone-treated glassware cannot be used because of adsorption
effects (ref. 11).  (Note - Lindstrom's observations differ
somewhat with regard to observations with other metals and
undoubtedly depend on many additional factors, e.g., form of
compound, i.e., organo-mercurial, inorganic salt, etc., and
must be evaluated independently depending on the application.)

     Cellulose filter paper strongly adsorbs mercury.

     Studies using centrifugation of solutions show that mercury
from clear solutions can be removed by adsorption on invisible
dust particles.

     Decomposition of many mercury derivatives to metallic mer-
cury can result in rapid loss of the more volatile elemental
mercury.  With dilute solutions of mercury compounds, particu-
larly at a neutral pH, the loss due to decomposition into
metallic mercury, which has a low solubility and high vapor
pressure, can be extensive.

     In systems which use a collecting medium followed by reduc-
tion of Hg by SnCl2 directly in the atomizer of an absorption
flame photometer, several chemical interferences are observed.
CN~, I~, SaOa  interfere and over 1 mg As(III) or >^0.1-0.2 mg
Se(IV), Te(IV), Cr(VI) and Mn(VII) per cubic centimeters inter-
fere.

     Relatively high concentrations (100 yg/ml) of cobalt produce
a 10$ spectral absorption interference in the atomic absorption
determination of mercury.  The Co line at 2536.^9 ^> which causes
the interference when determining Hg at 2536.52 8 in a Co matrix,
arises from a metastable state.

     The application of gold wool or foil to separate mercury
from interfering components  is an excellent approach.  Although
only slight deterioration of efficiency has been observed with
continued use, some consideration should be given to a potential
problem with gold amalgams which have picked up marked quantities
of lead and tin.  When such amalgams are heated, mercury does not
evaporate.   The amalgams explode into small pieces.
                                260

-------
     Unless extensive shielding is used, the mercury vapor pho-
tometers cannot be used in the vicinity of high magnetic fields
(electrolytic preparation of chlorine and caustic soda).

     Several problems can be encountered in attempting to collect
mercury salts.  Absorption in liquids can be incomplete since
particles between 5u and 25y may pass through.  Use of a filter
coupled with an absorption device is recommended.  The absorption
device should be used to back the particle filter to minimize the
volatilization of the mercury salts due to the continuous passage
of air through the filter and over the mercury compounds, e.g.
halides,  which have relatively high vapor pressure.

     Hwang, et al. (ref. 91) thoroughly investigated analytical
variables of the technique using chemical reduction (SnCl2) of
air-borne mercury and its compound collected in KI-I2 reagent
and flameless atomic absorption measurements.  They concluded
(a) that argon, rather than air, should be used as a purge gas
to minimize oxidation of Hg; (b) low results occur when using
SnCla as a reducing agent with KI-I2 collecting media (hydrazine
hydrate should be used); (c) non-specific background absorbance
due to light scattering from aerosol droplets, etc., can be
minimized with a reference deuterium lamp, automatic background
correction; and (d) care, regarding working curve range, mixing
time, sample volume, and flow rate of flushing gas, must be taken.

     Nelson (ref.  86) encountered considerable response irregu-
larities when using grille-type instruments, as opposed to
self-sampling orifice-type mercury vapor detectors (photometers).
With the high sensitivity scale (0 to 0.1 mg/m3), and if the
photometers were zeroed on still air, any pure air flow through
the grill-type photometer caused a positive needle deflection.
The grill-type photometer must be zeroed with a flow rate of pure
air just equal to the flow rate of the mercury vapor-air mixture
being tested.  The zeroing process must be carried out immediately
before the calibration point is to be observed.  If this procedure
is used, grill-type, photometric, mercury detectors are extremely
accurate.

     No special zeroing procedure is needed with self-sampling
orifice-type instruments (ref. 86).

     Nelson (ref.  86) observed that deviations from linearity
occurred with grill-type instruments, but not with the self-
sampling orifice-type instruments, above 1.0 mg/m3.  When the
grill-type detectors were used with mercury in air concentrations
in excess of 1.0 mg/m3, the indicator needle would swing up to
a reading of 1.0 mg/m3 (on a 0-3.0 mg/m3 scale) and then de-
crease to a lower value.  As the mercury in air concentration
became higher, the reading observed became lower.
                               261

-------
8.   SUMMARY AND RECOMMENDATIONS

     At this time, there is no continuous monitoring system which
measures total mercury (vapor and particulate) directly.  Most
monitors are based on the measurements of mercury vapor alone.

     The techniques which have sufficient specificity and sensi-
tivity are:

     (a)  mercury vapor photometry (non-flame atomic absorption)
     (b)  microwave induced emission spectroscopy

     (c)  radio-frequency induced or arc excited emission
          spectroscopy

     Mercury vapor photometry measures elemental mercury vapor
only and would require a pretreatment step to generate elemental
mercury from chemically bound mercury.  Also, the technique is
subject to interferences from organic compounds, smoke, dust,
etc.  As a consequence, a compensation technique or removal
(physical or chemical) of the interfering materials must be made.

     When using low wattage microwave induced emission spectros-
copy, a pretreatment step to vaporize particulate or to reduce
chemically bound mercury to elemental mercury vapor must be used.
The microwave energy (low power) is not sufficient to vaporize
particulate directly.

     With radio-frequency induced or arc excited emission spec-
troscopy, sufficient energy is generally available to vaporize
particulate up to particle sizes of approximately 80y at flow
rates of 40 liters/minute.

     Preheating of the chemically bound mercury to generate and
vaporize elemental mercury can be accomplished by induction or
resistance heating, oxy-hydrogen or oxy-acetylene flame, ac-dc
arc, or rf-plasma.

     For each of the techniques measuring emission spectra, suf-
ficient spectral resolution with a monochromator must be available
to provide separation of spectral interferences which may arise
from the matrix.  The mercury vapor photometer system does not
necessarily use a monochromator, since the basis of the technique
(atomic absorption phenomenon) provides specificity for mercury.

     The need for a monochromator in the emission spectroscopy
techniques,  particularly the low power microwave induced optical
emission, can be eliminated if collection-desorption as a gold
or silver amalgam is used to isolate the mercury from interfering
components of the matrix.
                               262

-------
     Detection of mercury vapor with an argon ionization system
can be accomplished on a short-term, intermittent basis if the
bound mercury is decomposed to mercury vapor and the mercury
vapor is removed from the matrix with silver or gold amalgama-
tion.  Desorption of elemental mercury vapor into the argon
ionization detector would produce a highly responsive signal.

     Five systems can be identified for potential use as continu-
ous monitors for total mercury in stationary source emissions.
     System A

     Sample •* Thermal decomposition
     System B
                        Compensated mercury
                          vapor photometer
     Sample •* Thermal decomposition -»• Isolation as gold
          amalgams -»• Mercury vapor photometer

     System C

     Sample •* Thermal decomposition •*• Isolation as gold
          amalgam -»• Microwave induced emission spectroscopy
     System D

     Sample •+
     System E
Radio-frequency induced or arc excited emission
 spectroscopy
     Sample •+• Thermal decomposition -»• Isolation on carbon or
          as gold amalgam -»• Argon ionization detection

     Although System D does not require a separate thermal
decomposition sub-system, it does require a relatively good mono-
chromator to provide spectral resolution.  However, with the
monochromator, System D provides for multi-element monitoring.

     Systems A, B, and C are more sensitive than System D, but
require high efficiencies in the thermal decomposition and iso-
lation steps (when needed as in B and C).  No information is
available on decomposition efficiencies of refractory particulate
matrices or at modest to high flow rates.  Also, little informa-
tion is reported on amalgamation and desorption efficiencies of
mercury on gold at modest to high flow rates.
                                263

-------
9.    REFERENCES


 1.  Stahl, Q. R., "Preliminary Air Pollution Survey of Mercury
     and Its Compounds," U.S.  Dept. of HEW,  NAPCA,  Raleigh,  N.C.,
     October 1969.

 2.  Mellor, J. W., "A Comprehensive Treatise on Inorganic and
     Theoretical Chemistry," Vol.  IV, Longman, Green, and Co.,
     695-10*19 (1923).

 3.  Woodson, T. T.,  "Mercury Vapor Detector," Rev. Sci.  Instr.
     10., 308 (1939);  U.S. Patent 2,227,117 (Dec. 31, 19*10).

 4.  Van Suchtelen, H., N. Warmoltz and G. L. Wiggerink,  "Method
     for Determining the Mercury Content of Air," Phillips Tech.
     Rev. 11, 91-97 (19*19).

 5.  Ipatov, V. A., and L. P.  Pakhornov, "Photoelectric Apparatus
     for the Determination of Mercury Vapor in Air," Pribory i
     Tekhn. Eksperim.  No. 2, 91 (1958); CA 53, 7679c.

 6.  Trog, D. J., "Measurement of Atmospheric Pollution by
     Ultraviolet Photometry," Anal. Chem.  27, 1217  (1955).

 7.  McMurray, C. S.,  and J. W. Redmond, "Portable  Mercury Vapor
     Detector," U.S.  At. Energy Comm. Rept.  Y-1188  (1958).

 8.  Ballard, A. E.,  and C.W.D. Thornton,  "Photometric Method
     for Estimation of Minute Amounts of Mercury,"  Ind. Eng.
     Chem. Anal. Ed.  1^, 893 (19*11).

 9.  Ballard, A. E.,  D. W. Steward, W. 0.  Kamm and  C. W.  Zuehlke,
     "Photometric Mercury Analysis - Correction for Organic
     Substances," Anal. Chem.  2_6,  921 (195*0.

10.  Monkman, J. L.,  P. A. Maffett and T.  F.  Doherty, "The
     Determination of Mercury in Air Samples  and Biological
     Materials," Amer. Ind. Hyg. Assoc. Quart. 17,  *»l8 (1956).

11.  Linstrom, 0., "Rapid Microdetermination  of Mercury by
     Spectrophotometric Flame Combustion," Anal. Chem. 31, **6l
     (1959).

12.  Jacobs, M. B., S. Yamaguchi,  L. J. Goldwater and H.  Gilbert,
     "Determination of Mercury in Blood,"  Amer. Ind. Hyg. Assoc.
     J.  21, 475 (I960); "Ultramicrodetermination of Mercury  in
     Blood," Ibid. 22_, 276 (1961).

13.  Mansell, R. E.,  and E. J. Hunemorder, "Photometric Method
     for Trace Mercury Determination Using a  DU Spect"©photom-
     eter," Anal. Chem. 35_, 1981 (1963).
                               264

-------
14.   Hemeon, W.C.L., and G. F.  Haines, Jr., "Automatic  Sampling
     and Determination of Micro-quantities of Mercury Vapor,"
     Amer.  Ind.  Hyg. Assoc. J.  22, 75 (1961).

15.   Jacobs, M.  B.,  and R. Jacobs, "Photometric Determination  of
     Mercury Vapor in Air of Mines and Plants," Amer. Ind.  Hyg.
     Assoc.  J.  26^,  261-265 (1965).

16.   Schachter,  M.  M., "Apparatus for Cold Vapor Atomic Absorp-
     tion of Mercury," J. Assoc.  Official Agric.  Chem.  49,
     778-782 (1966).

17.   Pappas, E.  G.,  and L. A. Rosenberg,  "Determination of
     Submicrogram Quantities of Mercury by Cold Vapor Atomic
     Absorption Photometry," J. Assoc. Official Agric.  Chem.  49,
     782-792 (1966).

18.   Hanson, V.  P.,  "Ultraviolet  Photometry, Quantitative
     Measurement of Small Traces  of Solvent Vapor in Air,"  Ind.
     Eng. Chem., Anal. Ed. 13_,  119 (1941).

19.   Hingle, D.  N.,  G. F. Kirkbright and  T. S. West, "Some
     Observations on the Determination of Mercury by Atomic-
     absorption Spectroscopy in an Air-Acetylene Flame," Analyst
     92, 759-762 (1967).

20.   Winefordner, J. D., and R. A. Staab, "Determination of Zinc,
     Cadmium, and Mercury by Atomic Fluorescence Flame  Spectrom-
     etry,"  Anal. Chem. 36_, 165-8 (1964).

21.   Vickers, T. J., and S. P.  Merrick, "Determination  of Part-
     per Milliard Concentrations  of Mercury by Atomic-fluorescence
     Flame Spectrometry," Talanta 15., 873-876 (1968).

22.   Browner, R. P., R. M. Dagnall and T. S. West, "Studies in
     Atomic-fluorescence Spectroscopy - VIII.  Atomic Fluores-
     cence and Atomic Absorption  of Thallium and Mercury with
     Electrodeless Discharge Tubes as Sources," Talanta 16,
     75-81 (1969).

23.   Vitkun, R.  A.,  N. S. Poluektov, Yu.  V. Zelynkova,  "Atomic-
     fluorescent Determination of Mercury," Zh. Anal. Khim. 1970
     25(3),  474-8;  CA 73_, 21071w.

24.   Warr,  F. D., "Use of a Filter in Atomic-fluorescence
     Spectroscopy," Talanta 17_, 543-548 (1970).

25.   Vickers, T. J., and R. M.  Vaught, "Nondispersive Atomic
     Fluorescence Analysis," Anal. Chem.  41, 1476-1478  (1969).
                               265

-------
26.  Webb, R. J., M.S.W.  Webb,  and P.  C.  Wildy,  "A Monitor for
     the Quantitative Determination of Beryllium in the Atmo-
     sphere," United Kingdom Atomic Energy Authority,  AERE-R
     2868 (1959).

27.  Webb, M.S.W., R. J.  Webb and P. C. Wildy,  "Monitor for the
     Quantitative Determination of Beryllium in  the Atmosphere,"
     Journal of Sci. Instr.  37., 466-471 (I960).

28.  Fromm, D., and A. v. Oer,  "Nachweis  von Beryllium und
     emissionsspektrographischer Nachweis von Quecksilberdampf
     in Luft," Mikrochem. Acta I960, 235-244.

29.  Bache, C. A., and D. J. Lisk, "Gas Chromatographic Determi-
     nation of Organic Mercury Compounds  by Emission Spectrometry
     in a Helium Plasma," Anal. Chem.  4_3_, 950-952 (1971).

30.  Shkol'nikova, R. I., and V. P. Makar'ev, "Methods for
     Selecting Samples and Their Phase Analysis  While  Controlling
     Mercury Vapor Condensation," Tr.  Proekt. Nauch.-Issled.
     Inst. "Gipronikel,"  30., 57-9 (1966); CA 68^, 71920q.

31.  Levine, B. S. (ed.), "Quantitative Determination  of Mercury
     Vapor in the Air," U.S.S.R. Literature on Air Pollution and
     Related Occupational Diseases, 8_, 30-33 (1963); CFSTI
     63-11570.

32.  Gurvits, B. I., and  G.  P.  Chuklenkova, "Rapid Method for
     Determining Mercury  Vapors in the Air," Metod. Materialy
     i Nauchn. Soobshch.  Vses.  Nauchn.-Issled Inst. Zheleznodor.
     Gigieny 2_, 21-27 (1962); CA 6^, 2470h.

33.  Bregvadze, U. D., and M. N. Mirianashvili,  "Determination of
     Small Amounts of Mercury in Dust," Sb. Nauchn. Rabot  Inst.
     Okhrany Truda, Vses. Tsentr. Sov. Prof. Soyuzov 4_, 103-106
     (1961); CA 58, 7292c.

34.  Palalau, L., "The Rapid Determination of Mercury  in the
     Atmosphere by Means  of Emission Spectrophotometry," Rev.
     Chim. (Bucharest) 19, 54-58 (1968);  CA 6_9_,  29942u.

35.  Dams, R., J. A. Robbins, K. A. Rahn  and J.  W. Winchester,
     "Nondestructive Neutron Activation Analysis of Air Pollution
     Particulates," Anal. Chem. 4_2, 861-867 (1970).

36.  Rahn, K., J. J. Wesolowski, W. John  and H.  R. Ralston,
     "Diurnal Variation of Aerosol Trace  Element Concentrations
     in Livermore, California," J. Air Poll. Control Assoc. 21,
     406-409 (1971).

37.  Magos, L., "Radiochemical Determination of  Metallic Mercury
     Vapour in Air," Brit. J. industr. Med. 23_,  230-236 (1966).
                               266

-------
38.  Magos, L., and T. W.  Clarkson, "Isotope Method for Determin-
     ing [Metallic] Mercury in the Air," Brit.  1,130,921 (16 Oct.
     1968); CA ;7p_, 14229P-

39.  Krestovnikov, A. N.,  and I. G. Sheinfinkel, "Argon Detector
     for Determination of Mercury in Gases," Gaz.  Khromatogr.,
     Sb., Moscow 2_3 53-58 (1964); CA 6_5_, 1373b.

40.  Chase, D. L., F. E. Huber and W. M. Henry,  "Development and
     Evaluation of an Analytical Method for the  Determination of
     Total Atmospheric Mercury," E.P.A. Contract No.  EHSD 71-32,
     Tenth Monthly Progress Report, September 28,  1971, Battelle
     Columbus Laboratories, Columbus, Ohio.

41.  Church, F. W. (Referee), "Determination of  Mercury in Air,"
     American Conf. of Gov. Ind. Hygienists Recommended Method,
     Manual of Analytical Methods.  Recommended  for Sampling and
     Analysis of Atmospheric Components (1958).

42.  Kimura, Y., and V. L. Miller, "Vapor Phase  Separation of
     Methyl or Ethyl Mercury Compounds and Metallic Mercury,"
     Anal. Chem. 3_2_, 420 (I960).

43.  Delaughter, B., "Mercury Determination in  Industrial Plant
     Atmospheres by Atomic Absorption Spectrophotometry," At.
     Absorption Newsletter, 9_, 49-50 (1970).

44.  Milton, R. F., and W. D. Duffield, "The Estimation of Mer-
     cury Compounds in the Atmosphere," Analyst  72, 11-13 (1947).

45.  Bell, D., "Mercury Recovery from Gas Flows,"  Ger.  Offen. 1,
     945, 173, 27 Aug. 1970.

46.  Nordlander, B. W., "Quantitative Methods for  the Determina-
     tion of Mercury Vapor," IEC 1£, 522-524 (1927).

47.  Dinman, B. D., E. E.  Evans and A. L. Lynch, "Organic
     Mercury - Environmental Exposure, Excretion and Prevention
     of Intoxication in Its Manufacture," A.M.A. Arch.  Ind.
     Health 18^, 248 (1958).

48.  Sandell, E. B., "Colorimetric Determination of Traces of
     Metals," Vol. 3, 4th Ed., Interscience Publishers, New
     York (1959).

49.  Quino, E. A., "Determination of Dibutyl Mercury Vapors in
     Air," Amer. Ind. Hyg. Assoc. J. 2_3_, 231 (1962).

50.  Linch, A. L., R. F. Stalzer and D. T. Lefferts,  "Methyl and
     Ethyl Mercury Compounds - Recovery from Air and Analysis,"
     Amer. Ind. Hyg. Assoc. J. 29, 79-86 (1968).
                               267

-------
51.  Milligan, M.  P., E.  E.  Campbell,  B.  C.  Eutsler,  J.  McClelland
     and W. D. Moss, "Analytical Procedures  of the Industrial
     Hygiene Group, Los Alamos Scientific Laboratory  of  the Univ.
     of Calif. LA-1858, 2nd  Ed., Chapter  18  (1958).

52.  Tubin, A. Ya, "Fractional Determination of Mercury  Vapors
     of Mercury and of Some  Organic Mercury  Compounds in Air,"
     Nauch. Raboty Khim.  Lab.  Gor'kovsk.  Nauch-Is-sledovatel
     Inst. Gigieny Truda i Profboleznei,  Sbornik 6_,  23-8 (1957);
     CA 5_1» 12892h.

53-  Javonovic, S., and G. W.  Reed, "Mercury in Metamorphic
     Rocks," Geochim. cosmochim. Acta, 32,  3^1-346 (1968).

54.  Burke, W. J., S. Moskowitz, and B. H. Dolin,  "Estimation
     of Mercury in Air," New York State Industrial Bulletin,
     Vol. 18, No.  5 (1939).

55.  Williston, S. H., "Method of Detecting  Mercury Vapor by
     Collecting the Mercury  and Thereafter Analyzing  the Col-
     lected Mercury by Ultraviolet Analysis," U.S. 3,281,596,
     Oct. 25, 1966.

56.  Zemskov, I. P., and G.  I. Sidel'nikova, "Adsorption of
     Diethylmercury by Activated Carbon," Zh. Prikl.  Khim 35,
     466-9 (1962).

57.  Christie, A.  A., A.  J.  Dunsdon and B. S. Marshall,  "Field
     Methods for Determining Certain Organomercurial  Vapours
     in Air," Analyst £2_, 185-191 (1967).

58.  Pakter, M. K., "Determination of Mercury in the  Air,"  Gig.
     Sanit. 3JL, 70-72 (1969).

59.  Vol'berg, N.  Sh., and E.  P. Gershkovich, "Use of Water-
     soluble Solid Chemisorbents for Determining Mercury Vapors
     and Hydrogen Sulfide in Air," Gig. Sanit. 33., 60-63 (1968).

60.  Hemeon, W. C., and G. F.  Haines,  Jr.,  "Automatic Sampling
     and Determination of Micro-quantities of Mercury Vapor,"
     Am. Ind. Hyg. Assoc. J. 22, 75-79 (1961).

61.  James, C. H., and J. S. Webb, "Sensitive Hg Vapor Meter for
     Use in Geochemical Prospecting,"  Institution of  Mining &
     Metallurgy, 633 (1964).

62.  Nordlander, B. W., "Selenium Sulfide -  A New Detector for
     Mercury Vapor," I.E.G.  1£, 518-521  (1927).

63.  Stitt, P., and Y. Tomimatsu, "Sensitized Paper  for  Estima-
     tion of Mercury Vapor," Anal. Chem.  2_3, 1098-1101 (1951).
                               268

-------
64.  Kalb, G.  W.,  "Determination of Mercury in Water and  Sediment
     Samples by Flameless Atomic Absorption,"  At.  Absorption
     Newslett. £,  84-8? (1970).

65.  Pechanec, V., and J. Horacek,  "Elemental  Analysis  of Organic
     Substances Containing Mercury.  II.   Micro- and Semimicron-
     determination of Mercury,"  Collection Czech,  Chem. Commun.
     27., 232-238 (1962).

66.  Pechanec, V., "A Universal  Method for Determination  of
     Mercury in Inorganic Substances Containing 100-0.001% of
     Mercury," Collection Czech. Chem. Commun. 29, 716-721 (196^).

67.  Stepanov, I.  I., A.  A. Rudkovskii and V.  Z. Fursov,
     "Metallic Gold Sorbent for  Mercury Determination in  Geo-
     chemical Samples by  an Atomic  Absorption  Method,"  Izv. Akad.
     Nauk Kaz. SSR, Ser.  Geol.  2j6,  86-6 (1969); CA 71,  10396e.

68.  Anderson, D.  H., J.  H. Evans,  J. J.  Murphy and W.  W. White,
     "Determination of Mercury  by a Combustion Technique  Using
     Gold as a Collector," Anal. Chem. 4_3, 15H-1512 (1971).

69.  Monkman,  J. L., P. A. Moffett  and T. F. Doherty, "The
     Determination of Mercury in Air Samples and Biological
     Materials," Amer. Ind. Hyg. Assoc. Quart. 17, 4l8  (1956).

70.  Vaughn, W. W., and J. H. McCarthy, Jr., "An Instrumental
     Technique for the Determinations of Submicrogram Concentra-
     tions of Mercury in  Soils,  Rocks, and Gas," Geological
     Survey Research, pp. D123-D127 (1964).

71.  Vaughn, W. W., "A Simple Mercury Vapor Detector for  Geochemi-
     cal Prospecting," Geological Survey Circular 540,  U.S. Dept.
     Int., 1967.

72.  Azzaria,  L. M., "A Method  of Determining  Traces of Mercury
     in Geologic Materials," Can. Dept. Energy, Mines Resour;
     Geol. Surv. Can., Paper 66-54, pp. 13-26  (1967).

73.  Azzaria,  L. M., and  G. R.  Webber, "Mercury Analysis  in
     Geochemical Exploration,"  Canadian Inst.  Min. & Met. Bull.
     62, 521-530 (1969).

74.  Barakso,  J. J., and  C. Tarnocai, "A Mercury Determination
     Method and Its Use for Exploration in British Columbia,"
     Canadian Inst. Min.  & Met.  Bull. 6_3_, 501-505 (1970).

75.  Bregvadze, V. D., and M. N. Mirianashvili, "Determination
     of Small Amounts of Mercury in Dust," Sb. Nauchn.  Rabot.
     Inst. Okhrany Truda  Vses.  Tsentr. Soveta  Prof. Soyuzov,
     No. 3, 103-106 (1961); CA 58^  7292c.
                               269

-------
76.  Chabovskii, L. P., and Yu.  N.  Kuznetsov,  "Device for the
     Rapid Determination of Mercury in Powdered Samples," Zavod.
     Lab. 3i, 1549-50 (1968).

77.  Razumov, V. A., and T. P. Utkina, "Device for the Rapid
     Atomic Absorption Determination of Trace  Amounts of Mercury
     in Liquids, Powders, and  Optical Coatings," Spektrosk.,
     Tr. Sib. Soveshch., 4th 1965 (Publ. 1969), 291-293.

78.  Jerie, H., "Microdetermination of Mercury and Halogens in
     Organic and Inorganic Substances," Mikrochim. Acta 1970,
     (5), 1089-91.

79.  Kuznetsov, Yu. N., V. M.  Konovalov, I.  I. Stepanov, and
     L. P. Chabovskii, "Atomic-absorption Photometers for the
     Rapid Determination of Mercury in Ores  and Geological
     Samples," Prikl. Spektrosk. Mater. Soveshch., 16th (1965)
     1, 105-7-

80.  Leong, P. C.,  and H. P. Ong, "Determination of Mercury by
     Using a Gold Trap in Samples Containing Considerable Sulfide
     Minerals," Anal. Chem. 4^3,  940-9*11 (1971).

81.  Hamilton, G. A., and A. D.  Ruthven, "An Apparatus for the
     Detection and Estimation  of Organo-mercury Dusts and Vapours
     in the Atmosphere," Lab.  Practice 15_,  995-997 (1966).

82.  Wenninger, J.  A., and J.  H. Jones, "Determination of Sub-
     microgram Amounts of Mercury in Inorganic Pigments by the
     Photometric Mercury Vapor Procedure,"  J.  Assoc. Offie. Agr.
     Chemists 4_6_, 1018-21 (1963).

83.  Wenninger, J.  A., "Direct Microdetermination of Mercury in
     Color Additives by the Photometric-Mercury Vapor Procedure,"
     J. Assoc. Official Agric. Chem. 4j^, 820-832 (1965).

84.  Joensuu, 0. I., "Mercury-vapor Detector," Applied Spec-
     troscopy 25, 526-528 (1971).

85.  Nelson, G. 0., W. Van Sandt, and P. E.  Barry, "A Dynamic
     Method for Mercury Vapor  Detector Calibration," Am. Ind.
     Hyg. Assoc. 26., 388-393 (1965).

86.  Nelson, G. 0., "Generating Known Concentrations of Mercury
     Vapor in Air," Rev. Sci.  Instrum. 4JL,  776-7 (1970).

87.  Mayer, H., "Concerning a  New Method for the Measurement
     of the Lowest Vapor Pressures  of Mercury  and Potassium,
     Z. Physik §7.,  240 (1931).
                               270

-------
88.   Ernsberger, F. M., and H.  W.  Pitman, "New Absolute Manometer
     for Vapor Pressures in the Micron Range," Rev.  Sci. Instr.
     2i6, 584 (1955).

89.   Hill, C. P., "Measurement  of Mercury Vapor Pressures from
     0° to 35°C by the Knudsen  Pressure Gauge," Phys.  Rev.  18,
     113 (1921).

90.   Hill, C. P., "Measurement  of Mercury Vapor Pressure by Means
     of the Knudsen Pressure Gauge," Phys. Rev. 20,  259 (1922).

91.   Hwang, J. Y., P. A. Ullucci, and A. L. Malenfant, "Determina-
     tion of Mercury by a Plameless Atomic Absorption  Technique,"
     Can. Spectrosc. l£, 43-15, 53 (1971).
                               271

-------
                    APPENDIX IV
BRIEFING DOCUMENT - ANALYTICAL TECHNIQUES FOR LEAD
                        272

-------
                        TABLE OF CONTENTS


Section                                                      Page

   1      INTRODUCTION                                        274

   2      SOURCES AND TYPES OF LEAD EMISSION                  274

   3      CANDIDATE ANALYTICAL TECHNIQUES                     275
             \
          3.1  Emission Spectroscopy                          276

               3.1.1  Emission Spectrograph as Mobile         278
                      Analyzer for Lead in Air

          3.2  Atomic Absorption Spectrophotometry            281

               3.2.1  Flame Atomic Absorption                 281
                      Spectrophotometry
               3.2.2  Non-flame Atomic Absorption             282
                      Spectrophotometry

          3.3  Atomic Fluorescence Spectrophotometry          286

          3.4  X-ray Emission Spectroscopy                    286

          3.5  Nuclear Activation Analysis and                289
               y-Absorptimetry

          3.6  Colorimetric Methods                           290

          3.7  Electrometric Methods                          292

          3.8  Mass Spectrometry                              293

          3.9  Measurement of lonization Current              294

         3.10  Detection of Lead Particles - Formation        294
               of Ice Nuclei

         3.11  Gas Chromatography                             295

   4      SUMMARY AND RECOMMENDATIONS                         295

   5      REFERENCES                                          297
                                273

-------
                    ANALYTICAL TECHNOLOGY FOR
               CONTINUOUS MONITORING OF STATIONARY
                    SOURCE EMISSIONS FOR LEAD
1.
INTRODUCTION
     Lead can be present In stack emissions in a variety of
chemical forms and may be emitted as fume, dust or organic vapors.
The most stable compound and the largest tonnage material used
commercially is lead monoxide (PbO).

     Lead is well known to be a cumulative poison, with a rated
maximum allowable concentration for the industrial 8-hour working
day of 150 micrograms per cubic meter.  The most toxic compounds
are the lead alkyls.

     No monitoring system is available at this time for measuring
total lead emissions from industrial stacks.  Techniques have
been developed for measuring the volatile organo-lead compounds
in ambient air or inorganic particulate separately, but not con-
tinuously and together.  Continuous monitoring of stationary
source emissions for lead must be able to detect lead as vapor
and particulate, and as inorganic or organic compounds.


2.   SOURCES AND TYPES OF LEAD EMISSION

     Lead enters the environment through natural and man-made
processes (ref. 1).  It is used to the extent of approximately
one million tons per year in almost every type of industry and
may escape to the atmosphere as fume, dust, or organic vapors.
Industrial emissions may occur during processing of lead con-
taining raw materials, e.g., storage battery manufacture, non-
ferrous foundries, and production of pigments, coatings, etc.
Organo-lead compounds may enter the atmosphere during manufacture
of insecticides and gasoline additives.  Particulate and fume can
be generated in preparing castings, plating for sheet iron, flex-
ible tubes, cable sheathing and lead-lined fittings.

     Lead monoxide is commercially the most important lead com-
pound and is prepared by many different processes as a high
tonnage heavy chemical.  Lead monoxide, principally as aPbO,
litharge, is used in preparing storage-battery plates, in com-
pounding in rubber, as an activator in synthesizing sodium
plumbite for sulfur removal from gasoline, in formulating glasses,
glazes, and enamels, and in making many large tonnage lead com-
pounds.  The more important of the latter are basic lead carbon-
ate, basic lead sulfate, basic lead silicate, lead chromate, lead
arsenate, red lead, lead soaps, and greases.

-------
     Elemental lead has a boiling point of 17^0 ± 10°C and a
vapor pressure of 1.0 mm Hg at 987°C.  Lead monoxide melts at
888°C and has a vapor pressure of 1.79 x 10~3 mm at 670°C.
Higher oxides decompose by losing oxygen - Pb30i, (500°C), Pb203
(370°C), and Pb02 (290°C).

     Lead chlorides, bromides and iodides have boiling points
between 900-95^°C.  The sulfate and sulfite of lead are thermally
unstable; lead sulfate begins to decompose when heated in air
at about 500°C to yield a mixture of lead oxide and sulfate
(PbOPbSOi.) which subsequently decomposes between 900°C and
1000°C to yield lead oxide and sulfur oxides.  Lead sulfide
melts at lllil0C and lead carbonate decomposes to lead oxide
at 315°C.

     Organo-lead compounds are relatively volatile - tetraethyl-
lead boils at approximately 195°C with decomposition or at 91°C
at 19 mm, tetramethyllead has a boiling point of 110°C, and
tetraisopropyllead boils at 120°C at 14 mm.


3.   CANDIDATE ANALYTICAL TECHNIQUES

     Methods currently used to monitor lead levels in ambient
air are based on the principle of isolating the lead compounds
by processing large volumes of air.  The separation of the lead
compounds from air is accomplished either by physical means (fil-
ters) to remove particulate or by chemical reaction techniques,
in which lead fume, vapor or particulate react with a chemical
media to form a stable, relatively non-volatile compound.

     Various types of filters - paper, fiber glass, millipore,
membrane and microsorban - have been used.

     Iodine monochloride is the most efficient of the chemical
agents for trapping lead.  However, a variety of other materials,
including iodine in methanol, iodine in aqueous potassium iodide,
iodine in carbon tetrachloride, iodine crystals and dilute nitric
acid, have been tried.

     The analytical techniques most commonly employed for mea-
suring lead content on the filters or in the chemical trapping
solutions include colorimetry (dithizone reaction), atomic absorp-
tion spectrophotometry, emission spectroscopy, polarography, and
x-ray emission (fluorescence) spectroscopy.

     Hundreds of samples of atmospheric airborne particulate con-
taining lead have been analyzed after collecting specimens on
filters followed by suitable digestion procedures.  Measurements
were made with colorimetric techniques (dithizone) (ref. 2),
atomic absorption spectroscopy (ref. 3,^,5), mass spectroscopy
(ref. 6), x-ray spectrometry (ref. 7), and emission spectroscopy
(ref. 8).

                               275

-------
     Although the sensitivities of these methods are good, the
inordinate amount of time used in collecting and preparing the
samples precludes the direct application of methodology estab-
lished in these procedures to monitoring stationary source emis-
sions.  Some of the techniques can be modified for use in contin-
uous (or short-term intermittent) analyses of lead from stationary
source emissions.

     A number of the analytical methods have been applied to
measuring volatile organo-lead compounds (alkylleads), but not
particulate.  Others have been used to determine lead in particu-
late, but not in vapor.  No analytical technique is currently
available that can be immediately and directly applied to the
analysis of total lead, as vapor, fume, and particulate.

     The following sections contain technology related to ana-
lytical methods that can be applied to the continuous or short
term measurement of lead from stack emissions.  Although a variety
of techniques are presented, the most attractive methods for
general application (vapor and particulate) based on selectivity
and sensitivity, are absorption and emission spectroscopy.
Others, e.g., x-ray fluorescence, may find application in mea-
suring particulate and some, e.g., gas chromatography, would
provide specific analyses for alkyllead compounds.

3.1  Emission Spectroscopy

     Aughey (ref. 9) and Koppius (ref. 10) used an emission
spectrographic method to monitor lead (particulate and vapor)
continuously in air in the range 0.003 mg/ft3 to 0.05 mg/ft3.
The technique is reported in greater detail in the following
subsection and with modification can be adapted to measuring
lead emissions from stationary sources.

     Additional data relating to emission spectrographic analyses
for lead, collected on filters or in liquids are also included
to aid in understanding potential interferences and problems that
can be encountered.

     Although little information is available on the application
of electrodeless radio-frequency—  or microwave—induced emission
spectroscopy to measuring total lead, both techniques can be
used.  Technology cited in other briefing documents in this
series - (a) Beryllium and Cadmium, and (b) Mercury - can also
be applied to lead.  In applying microwave-induced emission
spectroscopy to measuring total lead, a pre-treatment step in-
volving decomposition of particulate (thermal treatment, directly
or in a reducing medium) would be necessary.
                               276

-------
     Razumov and Aidarow (ref .  11) attained absolute sensitivity
of 4 x 10 9g and relative sensitivity of 5 x 10~6$ with spectro-
graphic analyses with an a.c. arc.  Measurements were made on
the 2833.07 A lead line for lead in polluted air collected in
acetic acid or nitric acid solution.  Excitation (40 sec)  of an
aliquot (0.06 ml) after the addition of a NaCl spectral buffer
was accomplished in a carbon electrode which was previously coated
with polystyrene.  Similar methodology was applied by Yakovleva
et al. (ref. 12) in nitric acid solutions of lead collected on
an air filter.

     Trace quantities (0.15-50 yg) of lead (2823-19 &) were de-
termined by emission spectrographic analysis of atmospheric dust
particles collected on filter paper (ref. 13).  The paper was
burned in flat graphite electrodes after adding graphite powder.
By carrying out the excitation in a magnetic field, the relative
standard deviations were decreased from 
-------
     3.1.1  Emission Spectrograph as Mobile Analyzer for
            Lead in Air

          3.1.1.1  Principle and Applicability

               3.1.1.1.1  Principle - Air is drawn continuously
through a condensed spark discharge where the lead spark emission
spectrum is excited.  The lead line at 2203.5 A1 is measured with
a mobile quartz spectrograph equipped with a photographic plate
or photoelectric Geiger counter.

               3.1.1.1.2  Applicability - Instrument was applied
to monitoring lead levels in various manufacturing areas on a
semi-continuous basis.  The technique measures lead in the form
of elemental lead or inorganic or organic compounds of lead.  The
system is mobile, being mounted on wheels, and has a high degree
of specificity and sensitivity.

          3.1.1.2  Range and Sensitivity

               3.1.1.2.1  Range - Calibration curve shows a range
of 0.003 mg Pb/ft3 to 0.05 mg Pb/ft3 for a typical operation mode,

     Detection limit estimated as <0.003 mg Pb/ft3 and is deter-
mined chiefly by the erratic behavior of the spark discharge,
which emits a continuously changing background radiation.

          3.1.1.3  Interferences

     None reported.  Spectra of air matrix were minimized by
operating when a condenser across the electrodes was at a mini-
mum capacity (250 micro-microfarads).

          3.1.1.**  Accuracy, Precision and Stability

     No specific data are reported for accuracy or precision.

     Technique was not developed to a point where continuous
quantitative measurements were made over periods involving many
hours or days.  Difficulties with electrode wear and mechanical
and electrical instabilities were encountered.

          3.1.1.5  Apparatus

               3.1.1.5.1  Spectrograph - Quartz prism spectro-
graph (Hilger) with dispersion of 16 A/mm at 2200 &.

               3.1.1.5.2  Excitation Source - Flow through spark
chamber with quartz window; l/*l" lead-free copper electrodes;
12,000 volt transformer; condenser across electrodes adjusted
to 250 micro-microfarads.
                               278

-------
               3.1.1.5.3  Detector Systems - Photographic plates
or photoelectric Geiger counter.

               3.1.1.5.*t  Ancillary Equipment - Vacuum pumps,
power supplies, electronic amplifiers, etc.

          3.1.1.6  Time Cycle for Sampling and Measuring

     Sampling and analysis time was approximately one minute with
the direct reading system.  Obviously, with the photographic
plate technique, up to three hours could be required.

          3.1.1.7  Calibration Procedure

     Calibration was accomplished by measuring air containing
minute amounts of tetraethyl lead.  (Note - No specific data is
given.)  Chemical analyses of samples obtained from a common stream
at identical times to the monitor were compared to the response
of the monitor.

          3.1.1.8  Method of Sampling and Sample Preparation

     Air was drawn through Saran tubing to the spark chamber at
1 ftVmin.

          3.1.1.9  Multi-Element Application

     Could be used on all metals.  Authors specifically note Hg,
As, Ba, and Be.

          3.1.1.10  Physical Dimensions

     None reported although total monitor was mounted on a
wheeled carriage.

          3.1.1.11  Unit Output

     Milligrams Pb/ft3.

          3.1.1.12  Safety Hazard

     Potential fire hazard presented by an open spark.  However,
later model was mounted in an airtight box in which an inert
atmosphere such as nitrogen can be maintained.

          3.1.1.13  Recommendations for Method Improvement

     Newer grating spectrographs, photomultiplier tubes, and more
stable power sources could add increased response and stability.

     Use of electrodeless rf induced discharges as the excitation
source could improve long term, continuous operation.
                                279

-------
          3.1.1.12*  References

     Aughey, H., "A Rapid Mobile Analyzer for Minute Amounts of
Lead in Air," J. Opt. Soc.  Am. 39., 292-293 (19^9).

     Koppius, 0. G., "Detection of Lead in Air with the Aid of
a Geiger-Muller Counter," J. Opt. Soc. Am. 39, 294-297 (19^9).
                               280

-------
3.2  Atomic Absorption Spectrophotometry

     Measurements of lead content from ambient air, in particular
air near alkyllead production facilities, have been made on a
continuous basis by flame and non-flame atomic absorption tech-
niques.  With the flame technique, a detection level of 1 yg
Pb/m3 was reached, whereas the non-flame method yielded a detec-
tion level of approximately 0.16 yg Pb/m3.

     3.2.1  Flame Atomic Absorption Spectrophotometry

     In general, the flame techniques use a hydrogen-contaminated
air flame.  However, some success has been attained by measuring
the atomic absorption of lead in an argon 5 MHz induction plasma.

     Thilliez (ref. 17,18) reports the detection of 1 pg Pb/m3
with a flame atomic absorption system which measures the Pb atoms
in an open-ended tube after being generated in a hydrogen-
contaminated air flame.  The device is described for continuous
monitoring of lead concentrations from tetramethyllead and tetra-
ethyllead in the air of two workshops.  Measurements are made at
2833 X and the response time is less than two minutes.  Its sta-
bility is demonstrated by the fact that its initial calibration
has not required any modification during four years of operation.

     In a somewhat similar system, but not applied to continuous
monitoring of lead in air, Molden, et al. (ref. 19) attained in-
creased sensitivity (10-fold) by using the 2170 ft Pb line, rather
than the 2833 & line.  A hydrogen rich, air-hydrogen flame di-
rected into a long path, low reflecting fused silica absorption
tube was used to detect lead extracted from silicate rocks.  By
measuring the absorbance at 2170 ft and correcting for background
absorption by subtracting the absorbance at the nonabsorbing
Pb 2203.5 ft line, excellent calibration linearity was attained.

     Ramirez-Munoz and Roth (ref. 20) also reported that the best
single pass results for Pb were obtained with the 2170 ft line.
In addition, they indicated that the 26l4 ft Pb line may be useful
for the extension of the dynamic range to higher concentrations.

     Whenever the 2170 ft lead line is used, care must be taken
to ensure that interference from copper materials does not occur.
Hall and Woodward (ref. 21) have shown that erroneous results can
be obtained if lead analyses are performed on samples having copper
in the matrix.  The error is due to the absorption by copper in the
sample of copper radiation (2165.09 ft) emitted by the hollow-cathode
lamp.  The problem occurs with atomic absorption spectrophotometers
which use slitwidths of 1 mm (nominal bandpass ca. 6.5 ft).
                                281

-------
     The effect would be eliminated if narrower slits were used;
however, in many instruments the higher gain settings required
when using narrower slits result in extremely high noise levels.

     The difficulty can be eliminated by making measurements at
2833 8 or by using an electrodeless discharge tube as the energy
source when measuring at 2170 A*.

     A high-frequency (5 MHz) induction plasma (argon) was used
in a reducing mode to vaporize powdered samples and to provide
an absorption region (2 cm) for atomic absorption (ref. 22,23).
Direct measurement of powdered samples for a number of elements
including lead was accomplished.  Higher sensitivity was attained
in a reducing plasma than in a neutral plasma.  No specific data
are indicated, but good agreement with results obtained by flame
spectrometry are reported.

     The studies (ref. 22,23) demonstrate the possibility of a
considerable increase in sensitivity while operating a plasma in
the presence of reducing agents such as hydrogen.  The sensitivity
increase can be attributed to the direct reduction of metal oxides
in the plasma or a delay in the oxidation obtained directly in
the plasma by pyrolysis.

     Verkhovskii, et al. (ref. 24) constructed an atomic absorp-
tion system without a monochromator by using high intensity lamps
which emit narrow spectral lines and interference filters.   De-
tection limits of 1.0 yg/ml with relative standard deviation of
30$ for lead solutions are claimed.  No data citing presence or
absence of chemical or spectral interferences are reported.

     A general study of the flame atomic absorption spectroscopy
of lead was reported and experimental parameters were evaluated
to optimize instrumental conditions (ref. 25).  Operational con-
ditions are indicated for several matrices including lead in air
(lead trapped on iodine column and/or millipore filter).  The
most sensitive conditions were as follows:  wavelength, 2170 ft;
flame, oxy-hydrogen, with hydrogen atomizing the sample (re-
versed from normal); a "T" flame adapter.  A detection limit of
0.013 ppm of Pb in aqueous solution is reported.

     3.2.2  Non-flame Atomic Absorption Spectrophotometry

     A very sensitive method (detection limit of approximately
0.16 pg/m3) for the continuous measurement of lead in air was
developed based on the reduction of lead compounds,  particularly
organo-lead compounds, over hot carbon.  Atomic absorption mea-
surements were performed on the Pb° retained in a heated long-path
absorption tube (ref.  26).   The procedure is reported in detail
in the following sub-section.
                               282

-------
          3.2.2.1  Atomic Absorption - Continuous Determination
                   of Lead in Air

               3.2.2.1.1  Principle and Applicability

                    3.2.2.1.1.1  Principle - Ambient air samples
were analyzed continuously by drawing the air over hot carbon
rods to reduce the lead compounds to the atomic state.  The neu-
tral atoms were passed into a heated long-path absorption tube
where quantitative measurement of the Pb content was performed
by atomic absorption spectrophotometry at 2833 8.

                    3.2.2.1.1.2  Applicability - Method provided
a rapid, sensitive, and reproducible system for the continuous
determination of atmospheric lead.  The lower limit of sensitivity
was observed to be about 3 x 10 xlg of lead.  Measurements were
performed on laboratory air.  No data are reported for field oper-
ation.  Laboratory tests indicate that organic bound lead can
easily be measured.  Measurements on vapor emitted from externally
heated lead phosphate, carbonate or chloride were also performed.
However, no data are shown which define the problems related to
handling lead oxide particulate directly.

               3.2.2.1.2  Range and Sensitivity

                    3.2.2.1.2.1  Range - Approximately 0.16 yg/m3
to 5** yg/m3 in air, or as expressed on an absolute basis,
3 x 10 Tlg to 1 x 10 8g of Pb.

                    3.2.2.1.2.2  Sensitivity - Detection limit
was estimated as 0.16 yg/m3 in air or on an absolute basis
3 x lO'^g.

               3.2.2.1.3  Interferences

                    3.2.2.1.3.1  Chemical - No interferences from
most organic compounds(15 were tested) and water were observed.
Halogen-containing organic compounds caused some change in the
absorption intensity at the 100 mg/m3 (organic) level.  Apparently
some molecular absorption occurs.

     Interferences caused by incomplete reduction of molecules
to ground state atoms or the formation of refractory oxides or
anion effects (P03~ and C0§ ) commonly encountered with samples
in aqueous media were not observed.

                    3.2.2.1.3.2  Physical - Radiation interfer-
ences were eliminated by using a modulated light source.

     Particle size effects were not determined.
                               283

-------
               3.2.2.1.4  Accuracy, Precision and Stability

     For ten samples containing a concentration of 6.0 ug/m3 Pb,
the maximum deviation from the mean was 4.5/8, whereas the average
deviation from the mean was 1.5%.

     No data are reported for the length of time that the instru-
ment was operated continuously.  The consumption of carbon at
white heat in an air stream may dictate only moderate continuous
operation times.

               3.2.2.1.5  Apparatus

                    3.2.2.1.5.1  Radiation Source - Demountable
hollow cathode lamp using a shielded cathode and Jarrell-Ash
82-500 power supply.

                    3.2.2.1.5.2  Monochromator - Jarrell-Ash
82-500 1/2-meter monochromator - 100 micron fixed slits.

                    3.2.2.1.5.3  Detector - RCA 1P28 photomultiplier.

                    3.2.2.1. 5.4  Carbon Reduction System - Four-
inch graphite rods in quartz tubing heated with Lepel 5000 watt
rf generator.

                    3.2.2.1.5.5  Absorption Tube - Quartz tube
(160 cc) of unidentified length, heated by 1000-watt nichrome
resistance heaters.

                    3.2.2.1.5.6  Ancillary Equipment - Compressor
or vacuum pump, Jarrell-Ash 82-000 amplifier, Beckman Model 10005
Recorder.

               3.2.2.1.6  Time Cycle for Sampling and Measuring

     Continuous sampling and instantaneous measurement.

               3.2.2.1.7  Calibration Procedure

     Calibration of the apparatus was accomplished with a diffu-
sion apparatus which introduced tetraethyllead from tubes of
known diameter (approximately 1-10 mm) and path length.  The rate
of diffusion is determined by weight loss over a period of days
or weeks.  The response of the monitor was correlated with these
diffusion rates.

               3.2.2.1.8  Method of Sampling and Sample Preparation

     Ambient air was introduced into the analyzer on a continuous
basis with a small diaphragm compressor.  Flow rate is approxi-
mately 1.2 liter/minute.


                               284

-------
               3.2.2.1.9  Multi-element Application

     Although technique was applied only to lead measurements,
the method could be modified to handle other volatile elements,
e.g., Hg.

               3.2.2.1.10  Physical Dimensions

     No specifications reported.  Estimated at V x 2" x 6'.

               3.2.2.1.11  Unit Output

     Absorbance values convertible to mass or mass/volume units.

               3.2.2.1.12  Safety Hazard

     Moderate to high temperature; some exposure to rf radiation.

               3.2.2.1.13  Recommendations for Method Improvement

     The effectiveness for converting lead compounds present in
various particle sizes must be evaluated.  The system should work
well as an ambient air monitor, but residence times and conversion
efficiencies, plus consumption rate of carbon rod reducing system,
must be evaluated at higher air flow-rates for continuous stack
emission monitoring.

               3.2.2.1.1*1  Reference

     Loftin, H. P., C. M. Christian, and J. W. Robinson, "The
Continuous Determination of Lead in Air," Spectroscopy Letters 3.,
161-17** (1970).
                               285

-------
3.3  Atomic Fluorescence Spectrophotometry

     Although atomic fluorescence spectrophotometry has not been
applied to direct measurement of lead in air, the slightly lower
detection limit (one-half that of flame atomic absorption) attain-
able with aqueous solutions suggests the possible application to
air monitoring in selected cases.  If, in a high copper matrix,
the 217.0 nm line (Pb) cannot be used with the available instru-
mentation and, as a consequence, the sensitivity of the flame
atomic absorption is insufficient, the atomic fluorescence mea-
surements may be useful.

     The experimental conditions for detecting low concentrations
of lead by atomic fluorescence spectroscopy were studied and op-
timized (ref. 27,28).  A detection limit of 0.01 yg/ml of lead in
aqueous solution is obtained by measuring the direct-line fluores-
cence at ^05- 8 nm in an argon-oxygen-hydrogen flame.  This detec-
tion limit is an improvement as compared to 0.02 pg/ml obtained
by atomic absorption measurements at 217.0 nm (air-hydrogen flame)
and 0.2 ppm by flame photometry at *J05- 8 nm (nitrous oxide-
acetylene flame).  The effect of 100-fold excesses of 30 cations
and anions were examined, but only aluminum interfered signifi-
cantly (^3% drop in signal).
     As noted above (ref. 27,28), low detection levels are attain-
able by atomic fluorescence spectroscopy (APS)  in aqueous media.
However, consideration must be given to interatomic quenching of
fluorescence, e.g., by nitrogen, if the AFS phenomena were to be
applied to monitoring lead in air or in stack gas emissions intro-
duced directly into the flame.  Quenching of Pb fluorescence was
observed (ref. 27,38) with an air-hydrogen flame.  A 4-fold de-
crease in signal intensity was obtained with aqueous solutions
in an air-hydrogen flame, as compared to an argon-oxygen-hydrogen
flame .

3.4  X-ray Emission Spectroscopy

     A number of excitation sources ranging from a Van de Graaff
generator to the more common x-ray tube (tungsten target) and
radio-isotope sources have been used to measure lead on air-borne
particulate collected on filters.  Although high sensitivities
and attendant low detection limits result with  the Van de Graaff
system, devices using the latter excitation sources are more
practical for field applications.  In all cases, the techniques
require the collecting of particulate for measurement.  Also, the
technique is restricted to particulate unless the lead vapor can
be trapped by a complexing medium.  (Note - The same complexing
agents used in the colorimetric techniques for  collecting lead
vapor potentially could be applied to a paper tape sampling sys-
tem using x-ray fluorescence detection principles.)
                               286

-------
     Detection limits of 1 microgram in airborne dusts have been
reported (ref. 29-32) with x-ray tube excitation.  Major problems
are matrix and particle size effects.  Also, direct spectral in-
terference of the most sensitive lead lines La can occur with
AsKa fluorescent emission.

     Cares (ref. 33) used x-ray spectrometry to measure lead in
airborne metallic dusts and fumes collected on filter paper.  A
study of potential interferences, method of preparing reference
standards, and x-ray characteristics of various types of filter
media are presented.

     Leroux and Mahmud (ref. 3*O showed that air pollution samples
collected on membrane filters and weighing less than 1.5 mg/crrr
can be analyzed by x-ray emission spectrography without matrix
correction for all elements with atomic numbers greater than man-
ganese.  The method is non-destructive, reaches detection limits
to 0.05 Pg/m3, and requires about 5 minutes per element for analy-
sis.  Although glass fiber filters are not suitable for multi-
element analyses due to presence of Ba, Sr, Rb, Zn, Ni, Fe, etc.,
both glass fiber filters and membrane filters can be used in Pb
analyses.  A comparison of the x-ray fluorescence results with
atomic absorption measurements for the same samples shows a mean
deviation between the techniques of 0.26 yg/m3.

     The x-ray emission (fluorescence) spectrographic method is
rapid (5 minutes per element), but the time required to collect
sufficient sample from ambient air is very long, up to 2^1 hours.

     The analytical line (La.. - 1.1726 &) generally used for mea-
suring lead by x-ray emission or fluorescence techniques may be
interfered with by the arsenic line (Ka, - 1.173*1 A").  Corrections
can be made based on the ratio of the Ka/Kg lines of arsenic,
if tungsten LY lines are not present.  [Tungsten Ly line group
interferes with the arsenic Kg line (ref. 35).]

     Buchanan and Schroeder (ref. 36) converted a single-channel
x-ray spectrometer to a two-channel instrument to provide a means
of recording lead x-ray fluorescence and background measurements
simultaneously.  The technique was devised to improve precision
(minimize sample and electronic conditions) if long counting times
(500 seconds) were used to attain high accuracy on trace quanti-
ties of lead on particulate collected from the atmosphere.  The
420 plane of a lithium fluoride crystal was used to reflect back-
ground radiation while the 200 plane simultaneously reflected
lead fluorescent emission.  A maximum error of ±0.15 Mg or 17%
was observed at the 1 ug level.

     Johansen, et al. (ref. 37) attained extremely low detection
levels (0.3 ng) for lead from particulate collected by "dust fall"
method on carbon foil from atmospheric air.  For lead, L-radiation
was generated by irradiating the sample in vacuum with a proton
                                287

-------
beam (MeV) from a Van de Graaff generator.   Sufficient discrimina-
tion and sensitivity (counting times of 60  minutes) were obtained
with a high resolution Si(Li)-detector to detect 0.3 ng Pb in the
multi-element matrix characteristic of ambient air particulate.

     Carr-Brion (ref. 38) compared the limit of detection for
x-ray methods for on-stream analysis of a number of metals  in-
cluding lead in ores.  The limit of detection was defined as
percent element equivalent to 2 sigma background intensity with
counting times less than or equal to 100 seconds and were re-
ported as 0.001/8 for x-ray tube excitation  and 0.004J? for radio-
isotopic excitation.  (Undoubtedly, the level of detection can
be lowered at least by another order of magnitude with the use
of semiconductor detectors.)

     Enomoto (ref. 39) reported on a probe  design for the continu-
ous determination of Pb in aqueous solutions by using radioisotope
y-ray sources to induce x-ray fluorescence.  A precision of 0.2%
(absolute) was obtained for Pb in light matrices.

     Portable x-ray fluorescence analyzers  using x-rays produced
by radioisotope sources were developed to detect lead and other
elements having atomic numbers >20 (ref. 40).  Fluorescence radia-
tion is isolated by pairs of balanced absorption filters and the
abundance of the metal in the sample is estimated from the dif-
ference count rate obtained when the two filters are used sequen-
tially.  Instrument weights were only 12 Ibs and 16 Ibs.
     Burkhalter and Marr (ref. 41) report the detection limit of
40 ppm of Pb in synthetic standards and ores by using radioiso-
topic (57Co) induced x-ray fluorescence.

     Sciaraffa and Ziegler (ref.  42) sampled automobile exhaust
emissions with a Millipore filter and then used a radioisotope-
excited (109Cd) x-ray fluorescence analyzer (Panalyzer-4000) to
determine the amount of lead collected on the filter.  With
64-second counting times, measurements of the L-lines of lead
resulted in a detection limit of 5 Vg of lead.  However, in the
presence of bromine, considerable over-lapping of fluorescence
from bromine and lead was observed.  To separate the lead signal,
filters were required resulting in some loss of lead fluorescence
intensity and moreover a minimum detection limit of 15 to 20 pg
of Pb.  [Note - Availability of newer, more efficient silicon
(Li) semi-conductor detectors may permit better discrimination
of lead fluorescence without filters and attendant loss of in-
tensity. ]

     Hayashi (ref. 43) combined the use of x-ray fluorescent
analysis and x-ray diffraction analysis to determine the quantity
of lead and to establish the nature of the chemical composition
of industrial lead fume collected on filter paper.  The procedure
                               288

-------
was non-destructive.  The limit of detection for lead (La line)
was determined as 10 micrograms on a filter paper having a surface
area of 99-95 mm2.

     A reasonable linear relationship was shown for lead content
on the filter up to 50 mg of lead.  There was no apparent varia-
tion with the limited particle size (0.3y and O.ly) generated in
a laboratory fume generator.  Counting times of 180 seconds were
used (ref. 43).

     Gate (ref. 44) discusses the interference problems related
to the determination of lead in air sample filters (Whatman 41)
by isotope-excited, energy dispersive, x-ray fluorescence analysis,
Spectral interferences of As, Pb, Br and Compton scattering from
the isotope (2ltlAm) source are considered.  By using 109Cd excita-
tion, although a less efficient exciter than ^1|1Am, and a high
resolution semiconductor detector, a minimum detectable quantity
of lead in a 10-minute count of PbLg.,  was 0.072 yg Pb/cnr (equiva-
lent to 0.475 yg Pb for the 6.6 cm2 filter disk used in the test).
Cost of instrumentation was estimated by Gate (ref. 44) as
$7,000-$8,000.

     Rhodes, et al. (ref. 45,46) used 109Cd to excite Pb to emit
characteristic x-rays which were measured with a high resolution,
semi-conductor detector.  Particulate collected from ambient air
for 24 hours on 8"  x 10" Whatman 41 filter paper was analyzed for
Pb and 16 other elements by using three different radioisotope
sources.  Direct analysis without corrections for matrix effects
was possible by using thin specimens (average mass per unit area
of 370 yg/cm2).  A  detection limit of 0.11 yg Pb/cm* (0.025 yg
Pb/m3 of air) was attained with counting times of 10 minutes of
the PbLg emission.   The major problem with the procedure is the
need to equilibrate the cellulose filters for 24 hours in a con-
stant humidity box  before each weighing to minimize hygroscopic
effects.  In addition, weighing had to be performed for a fixed
time (7 minutes) after removal from the box.

     Qualitative identifications of the lead compounds were per-
formed by x-ray diffraction measurements on samples containing
1.6 mg to 149 mg of lead.  Generally,  2.8 x 4.8 cm sections were
cut from the glass-fiber filter for the x-ray diffraction analysis,

3.5  Nuclear Activation Analysis and y-Absorptlmetry

     Although activation analysis with thermal neutrons yields
poor detection limits for lead, charged particle activation and
irradiation with a  fast neutron flux can allow lower detection
limits.  With fast  neutron flux, 0.1% lead in a 100-mg sample
can be measured.
                               289

-------
     A non-destructive neutron activation analysis method (ref.
47) was developed for measuring lead in Pb compounds and ores.
The technique is applicable for emissions comprised principally
of lead compounds.  A detection of 0.1 mg of lead in a 100 mg
sample is possible by irradiation with a fast (>5 MeV) neutron
flux of 6 x 1011 neutrons/cm2-sec for 20 minutes.  After 30 minutes
of cooling, 2°'*pbin is counted with a Ge(Li) semiconductor detec-
tor.  Obviously this technique cannot be used as a field monitor,
but does have potential application for laboratory analyses.

     When lead is to be detected in an emission source which emits
predominantly lead or lead compounds as particulate, y-absorp-
timetry can be used for intermittent monitoring at moderate time
intervals.

     Strain and Leddicotte (ref. 48) measured lead in a variety
of industrial samples by y-absorptimetry with an 2lflAm source.

     Belina and Sotnikov (ref. 49) developed a y-absorption method
for automatic control of lead in ores, concentrates, and inter-
mediate products.  Since two y-ray beams with different energies
from 75Se and 60Co sources are used, the surface density need not
be determined and corrections need not be made.   The lead concen-
tration is related to the ratio of the natural logarithms of the
absorption of soft and hard y-rays.

     A rapid (3-4 min.) method for the determination of lead in
finely ground, dried samples of Pb ore, concentrate, and tailings
is reported based on y-emission.  The source of y-ray emission is
133Ba with a half-life of ^7-5 years.  The apparatus is relatively
stable, requiring recalibration once a month (ref. 50).

     By measuring g-radiation backscattering phenomenon, a rapid
(2 min.) determination of lead in lead products  can be made.  A
90Sr (0.5m Ci) provided the B-radiation.  The mean square error
was reported as <0.3% absolute (ref. 51j52).  (Note - Concentra-
tion range was not indicated.)

     In a review of photon activation analysis,  Lutz (ref. 53)
indicates the application of the technique for non-destructive
determination of lead using the reaction, 20 I|Pb(Y,n) 20 3Pb.  The
principle advantage of the technique is the avoidance of the
interference from 2l*Na production encountered with thermal-neutron
activation of materials, e.g., forensic samples, containing large
amounts of sodium.

3.6  Colorimetric Methods

     A number of manual colorimetric methods has been used to
measure airborne lead collected in various media.  Generally,
these techniques are based on the formation of a colored complex
                               290

-------
with dithizone.  High sensitivity provides a detection range of
1-500 yg Pb/m3 of air.  However, most of the methods require con-
siderable time for collecting sufficient sample to detect the
lead and additional time to prepare the specimens to develop the
color-producing reaction.

     Several relatively rapid methods using dithizone, tetrahydro-
quinone, tetrahydroxy-p-benzoquinone, and iodine have been applied
to detecting volatile lead compounds in air.  Although each may
not be universally applicable, one or another might be usable in
specific applications.

     Moss and Browett (ref. 5M developed methods for determining
particulate lead and tetra-alkyllead vapor in air by collecting
particulate on a glass fiber filter and isolating the alkyllead
in a hydrochloric acid solution of iodine monochloride.  Measure-
ments of lead-in-air concentrations down to 0.1 mg of lead per
10 cubic meters of air can be made automatically with the dithi-
zone reaction and a Technicon Auto-Analyzer after sampling periods
of 8 hours.  An alternate method based on a sampling period of
half an hour can provide a detection level of 0.3 mg of lead per
10 cubic meters.

     Several color comparison methods, based on reaction of tetra-
hydroxyquinone (ref. 55,56), have been used to determine lead
content in air.  Also, Snyder, et al. (ref. 57) developed a rapid
micromethod for field use based on a manual color development of
lead, collected in iodine-potassium iodide solution, with a dithi-
zone solution.  The Snyder, et al. method requires approximately
10 minutes and is accurate to better than one microgram of lead
per cubic foot of air.

     The colorimetric methods could be modified for automatic
colorimeters or for paper tape color comparison measurements.

     A field test system for the determination of lead fumes in
industrial atmospheres was developed based on the formation of
a purple color on filter paper impregnated with tetrahydroxy-p-
benzoquinone (ref. 58).   After collecting the fume on the filter
paper, the lead particles are dissolved in situ by means of a
spray of glacial acetic  acid and acetone.  A uniform purple
background stain is developed in a few seconds and quantitative
measurements are made by comparing the color intensity with a
standard color chart.

     The method was developed specifically for determining lead
fume (solid condensed dispersoids of particle diameter less than
one micron and composed  mainly of lead oxide particles) in the
presence of moderate amounts of lead dust.  The intent was to
determine with reasonable specificity a type of particulate lead
of well defined origin and size range, which also happens to be
a particularly toxic form.  Large particles of lead dust, which
                               291

-------
might be of relatively low toxicity, but could result in an ana-
lytical value far greater than the true lead fume concentration,
are not measured.

     The technique was applied to measuring lead fume in a lead-
refining works and in a factory manufacturing storage batteries.
Excellent agreement of the method was established with analyses
performed by the dithizone and spectrographic methods on portions
of the filter.

     The method is sensitive to about 0.05 mg Pb/m3.   Above
0.2 mg Pb/m3, the determination becomes less precise.

     Of the common metal fumes, those of cadmium and  zinc also
form coloured complexes under the conditions of the field test.
Antimony fume will form a color if more than the normal amount
of spraying is carried out.  The stains of these metals, when
present in not more than moderate amounts, can be removed by
washing with acetic acid or an aqueous solution of sodium hydro-
gen tartrate.

     Although the method was developed for manual operation, an
automatic paper tape system with photometric measurement of color
could be developed for lead fume based on this approach.

     Total airborne lead particulate collected on a paper filter
has been determined by using an aqueous solution of tartaric
acid-sodium hydrogen tartrate with glacial acetic acid and ace-
tone spray to develop a characteristic color in conjunction with
tetrahydroxy-p-benzoquinone (ref. 55).

     The amount of volatile organic lead compounds in the atmos-
phere has been determined by passing a regulated stream of air
over iodine crystals for a period of time so that at  least one-
half the iodine crystals are sublimed.  Any Pbl? compounds and
free iodine are collected on a porous substrate.  The free iodine
is removed and a yellow color on the substrate indicates the
presence of lead.  By comparing the color with colors produced
with known amounts of volatile organic lead compounds, quantita-
tive measurements can be performed.  The technique can be used
to determine tetraethyl and tetramethyl lead in the atmosphere
(ref. 59).

3.7  Electrometric Methods

     An electrometric procedure based on anodic voltammetry was
developed as a quick measurement of lead in atmospheric air
(ref. 60).  The technique is free from interferences  and permits
direct measurements of lead of the order of 0.02 yg.   The collect-
ing container is used as the electrolytic cell in the anodic
voltammetric measurement.  No data are given which would indicate
the effect of compound type or physical form.
                               292

-------
     The application of square wave polarography as an automatic
continuous analyzer of lead in the atmosphere is reported by
Yamate, et al.  (ref. 60).  Sample air is passed through an absorb-
ing bubbler containing 30 ml of 0.1N nitric acid for 46 minutes
at 22 liters/minute.  Subsequently, nitrogen is passed through
the absorbing bubbler for 3 minutes to remove oxygen from the
solution.  The  sample solution is transferred to an electrolytic
cell.  A high-sensitivity polarograph is programmed to record
square-wave polarograms from -0.6 to -0.85 V.  Wave heights of
the recorded polarograms are converted to lead concentration
(pg/m3) by using prepared lead calibration curves.  The detection
range for Pb is reported as 0-15 yg/m3.  The procedure requires
one hour.  No data are reported for potential interfering mate-
rials.

3.8  Mass Spectrometry

     Mass spectrographic techniques provide high sensitivity and
specificity for manual analyses of lead in airborne particulate
collected on filters.  However, the need for a high vacuum during
analyses and the time required to attain suitable conditions for
efficient excitation precludes the application of this technique
in a continuous monitoring system.

     Brown and  Vossen (ref. 62) used a special nitrocellulose
filter and decomposition technique to measure lead (particulate)
by spark source mass spectrometry.  A detection range of 0.004-
4 yg/m3 was reported for lead collected in a batch sampling
technique.

     Matrix effects, resulting in problems related to proper
preparation of  standards to calibrate signal response intensities,
are commonly encountered with ion- or spark-source mass spec-
trometry.  To minimize these effects, a technique called isotopic
dilution has been used.  Although the technique is not applicable
to continuous monitoring, the approach can be useful in selected
cases for monitoring batch-type samples collected on filters.

     Isotopic dilution techniques, when coupled with mass spectro-
metric measurements have been used to determine trace elements
in many matrices (ref. 63).  Generally, mass spectrometers with
thermal ionization sources are employed, but are not suitable to
simultaneous multi-element determinations.  By using high voltage
spark source excitation, all elements can be ionized (ref. 64).

     Lead is isotopically altered, separated by cation exchange,
eluted into an  electrolysis cell, electrodeposited onto gold wire
and sparked in  the mass spectrograph.  The method is somewhat in-
volved and requires too much time for use as a monitoring technique
                               293

-------
     Unlike most trace analytical methods, isotope dilution analy-
sis does not require analyzed or synthesized standards of the
same matrix material for reliable determinations.  The simplest
analytical procedure is to evaporate a portion of an isotopically
spiked solution of the sample on a suitable substrate for spark-
ing in the spectrometer.  Compared to direct sparking of the
sample, poorer limits of detection are obtained.   In addition,
unless separation techniques are used, a trace element may be
undetermined if the only isotope lines suitable for measurement
are interfered with by isotope lines of the matrix elements or
complex ionized species of the matrix and polyatomic fragments.

     Once equilibrium has been established with the added iso-
topes, the analysis is unaffected by sample losses.

     For other special analyses, mass spectrometric analysis of
volatile metal chelates can be used to detect ultra-trace (10 12g
to 10 1!*g) lead (ref. 65).  Because of the somewhat lengthy di-
gestion procedures required to form the metal chelate, the tech-
nique would only be applicable to manual, intermittent, batch
type analyses.  The mass spectrometric measurement time is short,
requiring <5 minutes.

3.9  Measurement of lonizatlon Current

     The M.S.A. Billion-Aire (ref. 66) has been used to detect
PbEti* (TEL) by forming particulate by sample pyrolysis and subse-
quent changes in current in an ionization chamber.  With air in
the ionization chamber, the finely divided particulate produces
pronounced decreases in the current because the particles promote
effective recombination of positive and negative ions by 3rd-body
collisions and decrease the mobility of the ions.  Most gaseous
additives in concentrations of several thousand ppm cause only
very small changes in the ion current.  The technique permits the
detection of alkyllead compounds down to the ppb level in air.

3.10  Detection of Lead Particles - Formation 'of Ice Nuclei

     Although the universal application of this technique is
questionable, restricted use is possible.  Morgan (ref. 6?) re-
ported on a method for monitoring lead particles in the atmosphere
by forming ice nuclei after reacting atmospheric lead with iodine.
Continuous real time measurements were performed by counting the
ice nuclei formed in the cold chamber of an acoustical counter.

     Potential interferences are silver and terpenes which can
also form ice nuclei with iodine.

     Although the technique measures lead particles, rather than
lead content, some consideration should be given to determine if
the technique could be used to measure total lead content.  By

-------
passing lead-containing air or stack gases through a thermal
reduction process (flame or furnace heating) or reduction in CO
(passage of lead-containing air over white-hot carbon), the
elemental lead generated by reduction may be measurable by the
ice-nucleating technique.

3.11  Gas Chromatography

     Several methods have been developed to detect specific alkyl-
lead compounds.  Detection of these compounds on an intermittent,
batch sampling basis is possible at the 0.02 yg/m3 level.

     Laveskog  (ref.  68) reported the separation of tetramethyllead
(TML) and tetraethyllead (TEL) from air by gas chromatographic
means and detection with a mass spectrometer.  By enriching the
air samples, analyses of street air for TML and TEL were obtained
between 0.02 and 0.2 yg/m3.  Approximately 10-15 minute sampling
times are required for street air.

     A gas chromatographic method using an electron capture de-
tector has been used to measure tetraethyllead (TEL) down to
10~I0g/ml (or 100 ppb) (ref. 69).  The method consists in passing
the air for analysis at a flow rate of 1.5 liters/minute for 10-
15 minutes through a sampling tube packed with the same material
as the chromatographic column.  The sample is subsequently de-
sorbed and injected into the gas chromatograph.  The concentration
of TEL is determined from the values of peak areas and the spe-
cific retention volumes.
     SUMMARY AND RECOMMENDATIONS
     Although several techniques have been applied to the contin-
uous monitoring of lead vapor, principally as the organo-lead
compounds, no single method has been tested completely to qualify
as a continuous monitor for total lead, i.e., vapor and particu-
late, from stationary source emissions.

     Non-flame atomic absorption spectroscopy , preceded by a
thermal and/or chemical reaction decomposition step to reduce
airborne lead compounds to elemental lead and to maintain the
elemental lead as vapor, should be considered.  The major problems
in adapting the most effective decomposition system (passage of
air containing lead compounds over hot carbon) used for ambient
air are:  (a) efficiency of the reducing reaction at high stream
velocities normally encountered in stack emissions and at the
resulting short residence time in the decomposition zone is un-
known; (b) potential plugging by refractory particulate in the
carbon chamber when used with stack emissions involving high non-
lead content, e.g., power plant fly ash, needs to be evaluated;
                                295

-------
and (c) consumption of the hot carbon rods at high stack gas flow
rates, where the stack gases are predominantly oxidants, e.g.,
oxygen, etc., must be determined.

     Continuous monitoring of ambient air flowing at 1 ftVmin.
has been accomplished by emission spectroscopy with a spark dis-
charge.  The technique permits a working range of 0.003 mg Pb/ft3
to 0.05 mg Pb/ft3, but based on newer instrumentation may allow
lower limits.  The instrumentation was applied to measuring lead
in vapor and particulate.  Other methods of excitation - rf
plasma, and intermittent arc - may also improve the efficiency
of the emission spectrographic method.

     X-ray emission (fluorescence) spectroscopy can be applied to
measuring lead in particulate collected on filters.  The major
restriction to its application for determining total lead is its
inability to analyze vapor.  Some means must be found to collect
volatile lead compounds on a filter-type media.  Some considera-
tion should be given to adapting paper tape samplers impregnated
with complexing agents, e.g., tetrahydroxy-p-benzoquinone, used
in collecting lead fume for colorimetric analyses.

     When using the most sensitive lead line (La) for measuring
lead by x-ray emission (fluorescence) spectroscopy, the analyst
must be aware of potential spectral interference from As (K ) in
media containing moderate amounts of arsenic.  The PbLg line has
been used successfully with 109Cd radiosiotope excitation.

     Techniques using y-absorptimetry are not necessarily adapt-
able to universal measurement of lead but could be used in
special circumstances where lead or its compounds would be the
only compounds emitted and could be collected on filters, etc.
The sensitivity of the method needs to be optimized and specific
detection limits established.
                               296

-------
5.    REFERENCES

 1.   Ziegfeld,  R.  L., "Importance and Uses of Lead,:  Arch.
     Environ.  Health 8, 202-12 (1964).

 2.   Ludwig, J. H.,  D.  R.  Diggs,  H.  E. Hesselberg,  and J. A.  Maga,
     "Survey of Lead in the Atmosphere of Three Urban Communities:
     A Summary," J.  Amer.  Indust. Hyg. Assoc. 26,  270-284  (1965).

 3.   Thompson,  R.  J., G. B. Morgan and L. J.  Purdue,  "Analysis  of
     Selected Elements  in Atmospheric Particulate  Matter by  Atomic
     Absorption,"  Atomic Absorption Newsletter 9_,  53-57 (1970).

 4.   Burnham,  C. D., C. E. Moore, and E.  Kanabrocki,  "Determina-
     tion of Lead  in Airborne Particulates in Chicago and Cook
     County, Illinois,  by Atomic  Absorption Spectroscopy,"
     Environ.  Sci. Technol. 3_, 472-475 (1969).

 5.   Burnham,  C. D., C. E. Moore, T. Kowalski, and J. Krasniewski,
     "A Detailed Study  of Lead Determinations in Air-Borne
     Particulates  Over  Morton Grove, Illinois by Atomic Absorption
     Spectroscopy,"  Appl.  Spectrosc. 24,  411-414 (1970).

 6.   Chow, T.  J.,  J. L. Earl, and C. P. Bennett, "Lead Aerosols
     in Marine  Atmosphere," Environ. Sci. Technol.  3_, 737-740
     (1969).

 7.   Pillay, K.K.S., and C. C. Thomas, "Determination of the
     Trace Element Levels in Atmospheric  Pollutants by Neutron
     Activation Analysis [and X-ray Spectrometry]," J. Radioanal.
     Chem. 7.,  107-118 (1971).

 8.   U.S. Public  Health Service, "Air Quality Data from the
     National Air  Sampling Networks and Contributing State  and
     Local Networks," 1966 Edition,  Durham, N.C.,  1968.

 9.   Aughey, H., "A  Rapid Mobile  Analyzer for Minute Amounts of
     Lead in Air," J. Opt. Soc. Am.  39., 292-293 (1949).

10.   Koppius,  0. G., "Detection of Lead in Air with the Aid  of
     A Geiger-Muller Counter," J. Opt. Soc. Am. 39., 294-297  (1949)

11.   Razumov,  V. A., and T. K. Aidarov, "Chemical  Spectral  Micro
     Determination of Lead and Mercury in Air of Industrial
     Buildings  and in Some Biological Objects," Tr. po Khim.  i
     Khim. Tekhnol.  3_,  397-402 (1964); C.A. 6_5_, 1025f (1966).

12.   Yakovleva, G. K.,  V.  N. Kvasova, and N.  S. Stenin,
     "Spectrographic Determination of Lead in Air," Lab. Delo.
     12, 720-723 (1966).
                               297

-------
13-  Perman, J. and E.  Perman,  "Spectrographic Determination of
     Metal Traces in the Atmosphere," Bol.  Geol.  Minero 80,
     476-84 (1969); C.A. 72., 124798x (1970).

14.  Bykhovskaya, M. S., and M.  D.  Babina,  "Spectrographic Deter-
     mination of Lead,  Zinc, and Thallium in  Air," Gigiena i
     Sanit. 7_, 26-30 (1956); APTIC  No.  28467.

15.  Petrova, A. G., V. A.  Razumov, and T.  K.  Aidarov,  "Determina-
     tion of Lead in Air and in  Biological  Materials,"  Zavodsk.
     Lab. 30., 1095-6 (1964); C.A. 62_, 30526 (1965).

16.  Lazebnaya, G. V.,  I. G. Yudelevich, L. I. Mamontova,  G. V.
     Shaidurova, and Yu. A. Lyandusova," Elimination of the  Effect
     of Sample Composition in Spectrographic  Analysis," Izv. Sib.
     Otd. Akad. Nauk SSSR,  Ser.  Khim. Nauk  5_,  48-54 (1967);
     C.A. 69., 56717f (1968).

17.  Thilliez, G., "Quantitative Analysis of  Metals by  Atomic
     Absorption in Industrial Medicine:  I. Application to
     Quantitative Analysis  of Organic Lead  in the Air of Workshop
     Environments," Arch. Maladies  Profess. Med.  Trav.  Securite
     Sociale 3_1, 133-140 (1970); NAPCA  Abstract Bulletin 71,
     No. 02, p. 46 (APTIC-19276).

18.  Thilliez, G., "Determination of Traces of Lead in  Air by
     Atomic Absorption  Spectrometry," Anal. Chem. 39, 427-432
     (1967).

19.  Moldan, B., I. Rubeska, M.  Miksovsky,  "Determination  of Lead
     in Silicate Rocks  by Atomic Absorption Spectrophotometry
     with Long-path Absorption Tubes,"  Anal.  Chim. Acta 50,
     342-5 (1970).

20.  Ramirez-Munoz, J., and M. E. Roth, "Sensitivities  in  Atomic
     Absorption Spectrometry: V. Lead, Lithium,  and Silver,"
     Flame Notes, Beckman 2_, 47-53  (1967).

21.  Hall, J. M., and Woodward,  C., "Interference in the Atomic
     Absorption Determination of Lead in Copper-based Materials,"
     Spectrosc. Letters 2_,  113-116  (1969).

22.  Biancifiori, M. A., and C.  Bordonali,  "Utilization of an
     Induction Plasma in the Analysis of Traces by Atomic  Absorp-
     tion," Com. Naz. Energ. Nucl., RT/CHI  (67) 15, 17  PP  (1967);
     C.A. 69., 7988m (1968).

23.  Bordonali, C., M.  A. Biancifiori,  "Plasma Induction in  Trace
     Analysis by Atomic Absorption," Corsi  Semin. Chim. £, 153-68
     (1968); C.A. 72, 50499b (1970).
                               298

-------
24.   Verkhovskii, B. I., V.  L.  Ginzburg, E.  E.  Maizil,  and G.  I.
     Satarina, "Atomic Absorption Spectrophotometry Using Light
     Filters in the Ultraviolet Region," Zh. Prikl. Spectrosk.
     8(1), 156-7 (1968); C.A.  69, 7990f (1968).

25.   Chakrabarti, C. L., J.  W.  Robinson and  P.  W.  West, "The
     Atomic Absorption Spectroscopy of Lead," Anal. Chim. Acta
     31, 269-277 (1966).

26.   Loftin, H. P., C. M. Christian, and J.  W.  Robinson,  "The
     Continuous Determination  of Lead in Air,"  Spectroscopy
     Letters 3_, 161-172 (1970).

27.   Browner, R. P., R. M. Dagnall, and T.  S. West, "The  Deter-
     mination of Lead by Atomic Fluorescence Spectroscopy,"
     Anal. Chim. Acta 50., 375-381 (1970).

28.   Sychra, V., and J. Matousek, "Atomic-Fluorescence  Spectros-
     copy of Lead," Talanta 17., 363-370 (1970).

29.   Hirt, R. C., W. R. Doughman, and J. B.  Giselard,  "Application
     of X-ray Emission Spectrography to Air-borne  Dusts in
     Industrial Hygiene Studies," Anal. Chem. 2^,  16*19-51 (1956).

30.   Schroeder, T.  D., "Application of X-ray Emission  Spectrometry
     and Square-wave Polarography to the Determination  of Trace
     Lead as an Air and Water  Pollutant," 1969, 128 pp.,  Univ.
     Microfilms, Ann Arbor,  Mich., Order No. 69-21, 726;  Diss.
     Abstr. Int. B 1970, 3_0.(7), 3064.

31.   Thielen, W. C., A. E. Alcocer, and H.  L. Helwig,  "X-ray
     Spectrographic Determination of Microgram  Quantities of
     Lead," Am. Chem. Soc.,  Div. Water Waste Chem., Preprints
     316-319 (1963).

32.   Mitsuzi, H., N. Takata, M. Motoyama, M. Akamatsu  and
     G. Hashizume,  "Determination of Zinc and Lead in  Suspended
     Particulates by Fluorescent X-ray Spectrometry,"  Bunseki
     Kagaku (Japan Analyst)  19_, 1383-1388 (1970);  APTIC-25015-

33.   Cares, J. W.,  "The Quantitative Determination of  Airborne
     Metallic Dusts and Fumes  by X-ray Spectrometry,"  Am. Ind.
     Hyg. Assoc. J. 2_9_, 463-8  (1968).

34.   Leroux, J., and M. Mahmud, "Flexibility of X-ray  Emission
     Spectrography as Adapted  to Microanalysis  of  Air  Pollutants,"
     J. Air Poll. Contr. Assoc. 2Q_, 402-404  (1970).

35.   Leroux, J., and M. Mahmud, "Improvement of X-ray  Spectro-
     graphic Analysis by Filtration of the L Lines from the
     Primary Beam," Can. Spectry. 13, 19-24  (1968).
                               299

-------
36.  Buchanan, Jr.,  E.  B.,  and T.  D.  Schroeder,  "Trace  Lead
     Analysis Employing a Dual-Channel Single-crystal X-ray
     Spectrometer,"  Appl. Spectry  2^,  100-103  (1970).

37.  Johansson, T.  B.,  R. Akselsson,  and  S.A.E.  Johansson^
     "X-ray Analysis:   Elemental Trace Analysis  at  the  10  12g
     Level," Nucl.  Instr. and Methods  8^,  Ill-ll3  (1970).

38.  Carr-Brion, K.  G., "X-ray Methods for On-stream Analysis,"
     Chem.  and Ind., 1753-1756 (1968).

39.  Enomoto, S., "Determination of the Heavy  Element Concentra-
     tion by X-ray  Fluorescence Analysis  Using Radioisotope
     y-ray Sources," Commis.  Energ. At. (Pr.), Rapp. 1968,
     CEA-R-3369, 68  pp; C.A.  70., 13832k (1969).

10.  Bowie, S.H.U.,  "Portable X-ray Fluorescence Analyzers  in
     the Mining Industry,"  Mining  Mag. (London)  118(1),  230-1,
     233, 235, 237,  239 (1968); C.A.  6_£,  10928q  TT9"68).

11.  Burkhalter, P.  G., and H. E.  Marr, "Detection  Limit for
     Gold [Lead, Tungsten and Uranium] by Radioisotopic  X-ray
     Analysis," Int. J. Appl. Radiat.  Isotop 2^  (7), 395-103
     (1970).

12.  Sciaraffa, P.  L.,  and  C. A. Ziegler,  "Automobile Exhaust
     Analysis for Lead  Using a Nuclear Technique,"  Isotop.  Radiat.
     Technol. 8_, 163-1  (1970).

13.  Hayashi, H., "Determination of Chemical Composition and
     Crystal Phase  of Fume  by X-ray Fluorescence and X-ray
     Diffraction Analysis," Ind. Health 8_,  66-77 (1970).

11.  Gate,  J. L., "Determination of Lead  in Air  Sample  Filters
     by X-ray Fluorescence  Analysis,"  UCRL-51038, Lawrence
     Radiation Laboratory,  Univ. of Calif., Livermore,  Calif.
     (197D.

15.  Rhodes, J. R.,  A.  H. Pradzynski,  and J. S.  Payne,  "Energy
     Dispersion X-ray Fluorescence Spectroscopy  for Rapid Multi-
     element Analysis of Air Particulates," Columbia Scientific
     Industries, presented  at Pittsburgh  Conference, Cleveland,
     Ohio,  March 1972.

16.  Rhodes, J. R.,  A.  H. Pradzynski,  and R. D.  Sieberg, "Energy
     Dispersive X-ray Emission Spectrometry for  Multielement
     Analysis of Air Particulates," Columbia Scientific  Industries,
     presented at Instrument Society  of America  Symposium,  San
     Francisco, Calif., May 1972.
                               300

-------
47 .   Filby, R.  H., K.  R. Shah, A.  I.  Davis, "Determination of
     Lead by Fast Neutron Inelastic Scattering Induced Lead-204m,"
     Radiochem. Radioanal. Lett.  5_, 9-18 (1970).

48.   Strain, J. E., and G. W. Leddicotte, "Industrial Activation
     Analysis," Proc .  Natt .  Anal.  Instr. Symp. 8, 297-302 (1962);
     C.A. 6_£, 6l97g
49.   Belina, G. L., and V. A.  Sotnikov, "y-Absorption Method for
     Control of Lead Content," Radioizotopyne Metody Automat.
     Kontrolya, Akad. Nauk Kirg.  SSR, Tr.  Rasshiren. Soveshch.
     Vses.  Seminara po Primeneniyu Radioaktivn. Izotopov v
     Izmeritel'n Tekhn. i Priborostr., 1961 2, 52-60 (1962);
     C.A.  59., 3307a (1963).

50.   Frolov, V. M., V. N. Balashev, "y-Emission Method for Rapid
     Determination of Lead Content," Tsvet . Metal.  42, 21-24
     (1969); C.A. 72., I8l72d (1970).

51.   Velyus, L. M., "Rapid Determination of Lead Content in Lead
     Products," Sb. Nauch. Tr., Magnitogorsk. Gornomet .  Inst .  45,
     64-68  (1968); C.A. 72., 139323n (1970).

52.   Hisashi Mori and Sadaichiro Taira, "The Analysis by B-ray
     Backscattering, " Nippon Isotope Kaigi Hobunshu 4_, 238-42
     (1961); C.A. 6^, 12910e (1964).

53-   Lutz,  G. J., "Photon Activation Analysis - A Review," Anal.
     Chem.  43., 93-103 (197D-

54.   Moss,  R., and E. V. Browett, " Determination of Tetra-alkyl
     Lead Vapour and Inorganic Lead Dust in Air," Analyst 91,
     428-438 (1966).

55.   Tufts, B. J., "Determination of Particulate Lead Content  in
     Air,"  Anal. Chem. 3_1, 238-241 (1959).

56.   Andur, M. 0., and L. Silverman, "Field Determination of Lead
     in Air," A.M. A. Arch. Ind. Health 1£, 159 (1954).

57-   Synder, L. J., W. R. Barnes, and J. V. Tokos,  "Determination
     of Lead in Air," Anal. Chem. 2p_, 772-776 (1948).

58.   Dixon, B. E., and P. Metson, "A Field Method for Determina-
     tion  of Lead Fume," Analyst 84^, 46-50 (1959).

59-   Walker, A. 0., "Volatile  Organic Lead Compounds in the
     Atmosphere," U.S. 3,453,081, 01 July 1969; C.A. 7_1, 4l997q.
                               301

-------
60.   Berlincioni,  M.,  "Determination of Lead  in  Atmospheric  Air
     by an Anodic  [Electrochemical]  Method,"  Boll.  Lab.  Chim.
     Prov. 20_, 479-91  (1969).

6l.   Yamate,  N.,  T.  Matsumura,  and M.  Tonomura,  "Automatic Con-
     tinuous  Analyzer  of Lead  in the Atmosphere," Eisei  Shikensho
     Hokoku 8?., 28-31  (1969);  C.A. 7i,  Il5953a.

62.   Brown, R., and  P.G.T.  Vossen, "Spark  Source Mass  Spectro-
     metric Survey Analysis of Air Pollution  Particulates,"
     Anal. Chem.  42_, 1820-1822 (1970).

63.   Webster, R.  K., "Isotope-Dilution Analysis," Advances in
     Mass Spectrometry,  J.  D.  Waldron,  Ed., Pergamon Press,
     New York, N.Y., 103-119 (1959).

64.   Alvarez, R.,  P. J.  Paulsen, and D.  E. Kelleher, "Simultaneous
     Determination of  Trace Elements in Platinum by Isotope
     Dilution and  Spark  Source Mass  Spectrometry,"  Anal.  Chem.
     41, 955-958  (1969).

65.   Kowalski, B.  R.,  T. L. Isenhour,  and  R.  E.  Sievers,  "Ultra-
     Trace Mass Spectrometric  Metal  Analysis  Using  Heptafluoro-
     dimethyloctanedione Chelates,"  Anal.  Chem.  998-1003 (1969).

66.   Strange, J.  P., K.  E.  Ball, D.  0.  Barnes,  "Continuous Parts
     Per Billion Recorder for  Air Contaminants," J. Air  Pollution
     Control  Assoc.  1(D,  423-6  (I960).

67.   Morgan,  G. M.,  "Technique for Detecting  Lead Particles  in
     Air," Nature  213, 58-59 (1967).

68.   Laveskog, A., "A  Method for Determination  of Tetramethyl
     Lead (TML) and  Tetraethyl Lead  (TEL)  in  Air,"  preprint
     International Union of Air Pollution  Prevention Assoc.,
     86 p, 1970;  APTIC-25055.

69.   Cantuti, V.,  and  G. P. Carton!, "The  Gas Chromatographic
     Determination of  Tetraethyllead in Air," J. Chromatog.  32,
     641-647  (1968).
                               302

-------
               APPENDIX V
BRIEFING DOCUMENT - ANALYTICAL TECHNIQUES
        FOR ARSENIC AND ANTIMONY
                   303

-------
                        TABLE OF CONTENTS

Section                                                      Page

   1       INTRODUCTION                                       305


   2       SOURCES AND TYPE OP EMISSION                       305

           2.1  Arsenic and Its Compounds                     305
           2.2  Antimony and Its Compounds                    306


   3       CANDIDATE ANALYTICAL TECHNIQUES                    307

           3.1  Colorimetry and Microtitration                307

                3.1.1  Arsenic                                307
                3.1.2  Antimony                               308

           3.2  Atomic Absorption Spectrophotometry           308

                3.2.1  Arsenic                                308
                3.2.2  Antimony                               309

           3-3  Emission Spectroscopy                         310
           3.4  Atomic Fluorescence Spectroscopy              311
           3.5  X-ray Emission Spectroscopy                   311

                3.5.1  Arsenic                                311
                3.5.2  Antimony                               311

           3.6  Neutron Activation Analysis                   312

                3.6.1  Arsenic                                312
                3.6.2  Antimony                               312

           3.7  Mass Spectrographic Analysis                  313


   1\       SUMMARY AND RECOMMENDATIONS                        314


   5       REFERENCES                                         316
                                304

-------
1.    INTRODUCTION

     Arsenic is emitted from stationary sources in a variety of
chemical compounds.  The most common arsenic compound is arsenic
trioxide, which can be emitted as fume, vapor, or particulate.
Smaller quantities of arsenic are emitted as the very volatile
and toxic arsine.  Arsenicals - inorganic and organic - used as
herbicides, insecticides , defoliants and desiccants are present
in stack emissions from production facilities and from plants,
e.g., cotton gins, processing agriculture products.

     No 24-hour maximum atmospheric concentration has been set
for arsenic in the United States; however, the U.S.S.R. and
Czechoslovakia use a basic 24-hour standard of 3 pg/m3.  The
American Conference of Government Hygienists and the American
Industrial Hygiene Association recommend threshold limit values
for industrial workers (8 hour/day exposures) of 0.5 mg/m3 for
arsenic (ref. 1), 0.15 mg/m3 for lead arsenate (ref. 1), and
0.2 mg/m3 for arsine (ref. 2).

     Antimony and its compounds can appear in stationary source
emissions as the element antimony, as an alloy in combination
with other metals, e.g., antimonial lead, or as the oxides,
halides, sulfates and sulfides.  The largest tonnage emissions
occur in the ore refining processes where the predominant product
is antimony trioxide.

     The threshold limit values cited by the American Conference
of Governmental Industrial Hygienists (196?) for antimony and its
compounds (as antimony) are 0.5 mg/m3.


2.    SOURCES AND TYPE OF EMISSION

2.1  Arsenic and Its Compounds

     Sullivan (ref. 3) reviewed the sources of emissions and uses
of arsenic and its compounds.  Arsenic, as arsenic trioxide, is
produced as a by-product in gold and copper smelters.  The low
temperature (193°C) at which sublimation of arsenic trioxide
occurs results in relatively wide-spread dissemination of the
As20s as fume (vapor and suspended small particles).  Arsenic
trioxide is the predominant material used in the production of
other arsenic compounds, and as a consequence, can be a major
component in emissions from chemical production facilities.

     Arsenic metal and arsenic compounds - arsenic sulfides,
arsine, arsenic(V) oxides (in presence of a reducing agent) and
organic arsenates - are readily converted by heat and oxygen to
arsenic trioxide fume.
                                305

-------
     The volatile, toxic gas, arsine, is formed in pickling, sol-
dering, etching, and plating processes where hydrogen is emitted
from the dissolution of metals in acid.  Organic and inorganic
arsenicals are produced as agricultural chemicals and significant
emissions have been reported from cotton gins.

     Processes requiring the combustion of coal can produce mea-
surable amounts of arsenic in the emission products.

     Arsenic trioxide has a sublimation temperature of 193°C-;
arsenic disulfide has a boiling point of 565°C.  The halides,
arsenic trichloride, trifluoride, and triiodide boil at 130.2°C,
63°C, and ^400°C, respectively.  Arsine has a boiling point of
-55°C, and lead arsenate decomposes above 28o°C.  Elemental ar-
senic sublimes ca. 600°C and has a vapor pressure of 1 mm at 372°C.

2.2  Antimony and Its Compounds

     The natural and industrial sources of antimony and its com-
pounds, and the chemical and physical properties of the various
antimony compounds are described in a series of review articles
(ref. 4).

     Antimony trioxide (sublimation temperature of 1550°C) is
generated as the major antimony compound in ore refinery processes.
Sublimation of the trioxide is used for separating antimony from
the ore.  Antimony trioxide is a chemical intermediate, flame-
proofing agent, and an additive in ceramic, glass and paint manu-
facturing.

     Halides - SbCl3 (b.p. 223°C), SbBr3 (b.p. 280°C), and SbI3
(b.p. 401°C) - are chemical intermediates and are used as mordants
and catalysts, and in the electroplating industry.

     Antimony trisulfate and pentasulfide decompose when heated
and are used in manufacturing explosives, fireworks and other
pyrotechnics.  The pentasulfide is also used as a pigment and in
vulcanizing rubber.  The trisulfide is a good camouflage pigment
and is used with the pentasulfide in matches.

     Antimony oxychloride decomposes at 170°C and is used as a
chemical intermediate and as a smoke-producing agent.
Footnote:

-The vapor pressure of arsenic trioxide varies considerably de-
 pending on the crystalline or amorphous form - monoclinic,
 octahedral, and glassy.  Typical vapor pressure data are:
 2.4 x 10~7 mm % 60°-6l°C; 2.5 x 10   mm & 8l°-86°C; 4.6 x 10"" mm
 % 101°-105°C; 2.2 x 10~3 mm @ 119°-126°C; and 2.6 x 10   mm %
 149°-152°C.
                               306

-------
     The very volatile and toxic stibine (SbH3 - b.p. -17°C) is
produced by the reaction of hydrogen on antimony or its compounds.

     Alloys of antimony are used as bearing metals, battery plates,
cable coverings, and castings.  The chief alloys are brittania
metal and antimonial lead.


3.   CANDIDATE ANALYTICAL TECHNIQUES

     No direct reading instruments for field application are
available for measuring arsenic and antimony in ambient air or in
stack emissions.  Laboratory methods are based on the collection
of dust or fumes with filter paper samplers, electrostatic pre-
cipitators, and/or impingers.

     Analytical techniques relying on instrumental methods have
been applied to measure arsenic and antimony in various media,
including particulate collected from ambient air.  The techniques
used predominantly are colorimetry, iodine microtitration, atomic
absorption spectrophotometry, atomic fluorescence, emission spec-
troscopy, x-ray fluorescence and neutron activation analysis.
Most of these methods are based on analyses of particulate col-
lected on filters (paper or glass fiber) and suffer from the high
probability of potential loss of vaporized As20a and other com-
pounds when large volumes of air are drawn over the collected
particulate.  Some of this loss can be minimized by a tandem
arrangement of filter and impinger containing a collecting liquid.
The vaporization losses on filters could be moderately high when
sampling heated stack gases.

3.1  Colorimetry and Microtitration

     3.1.1  Arsenic

     Determinations of arsenic are performed by reaction with
silver diethyldithiocarbamate (ref. 5,6), by iodine microtitra-
tion method using an amperometric titrator (ref. 7), and by
reaction with molybdenum blue (ref. 8).

     The Gutzeit method (reaction with mercuric chloride or bro-
mide) (ref. 6) is not recommended, except for laboratories where
a very large number of analyses is performed (ref. 6).  The
Gutzeit method is susceptible to changes in many variables.

     Conversion of the arsenic compounds to arsine is required in
the molybdenum blue, silver diethyldithiocarbamate, and Gutzeit
methods.  To obtain conversion to arsine, the arsenic must be in
the trivalent state.  Each method, therefore, incorporates a re-
duction step to ensure total conversion to the trivalent arsenic.
Antimony may result in an interference as stibene in amounts in
excess of 0.0001 g.  Hydrogen sulfide, unless removed by the lead
acetate glass wool plug in the standard train, can also interfere.


                                307

-------
     The formation of arsine is based on the reaction of trivalent
arsenic with nascent hydrogen generated by reacting zinc metal
with sulfuric acid.  Since oxygen inhibits the formation or arsine,
air should be eliminated.   The reaction is a batch process.  A
detailed study of the interferences encountered with the evolution
of arsine was reported by  Whitnack and Martens (ref. 9).  Inter-
ferences were observed when ppb quantities of chromate,  molybdate
and metavanadate ions were present.  Slower rates of hydrogen evo-
lution were observed and the formation of molybdenum and vanadium
arsenite complexes are suggested.  Additional interferences of
nitrates, chlorates, or compounds of copper have been observed
(ref. 10).

     The silver diethyldithiocarbamate is the most sensitive and
simplest of the four methods for arsenic.  By collecting a speci-
men from ambient air at 1  ft3/min. using water or decinormal
sodium hydroxide as collecting medium in a Greenburg-Smith im-
pinger, a detection limit  of approximately 0.1 to 0.2 ug arsenic
per cubic meter, in a 30 cubic foot air sample, can be attained
(ref. 5,6).

     Agranov (ref. 11) reports the development of an automatic gas
analyzer which permits the detection of arsine and arsenic at the
1 mg/m3 level (estimated).  No specific details are given, but the
general approach is to measure the change in the light absorption
in the reflection by a paper tape, on which a color reaction is
developed.

     3.1.2  Antimony

     Several volumetric methods (a) KMnOi,, (b) KBrOs, (c) I2 have
been used to titrate Sb(III) (ref. 12).  Interferences by As(III),
Pe(II), S02, V(IV) occur.   Two colorimetric methods, rhodamine B
and iodoantimonous acid, have found favor for estimating trace
quantities of antimony (ref. 12).  An, Fe(III), Ga, Tl and W can
interfere in the rhodamine B method.  Bi and Tl can interfere in
the iodoantimonous acid method.

     Traces of antimony may be detected by evolution of stibine
that forms when antimony solution is treated with zinc and hydro-
chloric or sulfuric acid,  and by measurements with one of the
colorimetric techniques indicated above (ref. 12).

3.2  Atomic Absorption Spectrophotometry

     3.2.1  Arsenic

     Arsenic has been measured with a variety of atomic absorption
techniques.  By varying flame composition - air-acetylene (ref.
13,14), argon-hydrogen (ref. 12), nitrous oxide-acetylene (ref.
15), oxy-acetylene (ref. 16) - high flame background and noise
                               308

-------
levels obtained at the resonance lines, 193.7 nm or 196.0 nm, can
be minimized.  These techniques are based on measurements with
aqueous solutions of arsenic and yield detection limits of ap-
proximately 1 ppm of arsenic in water.

     Thompson et al. (ref.  17) reported the detection limit of
0.02 yg As/m3 of air by atomic absorption spectrophotometric mea-
surement of acid solutions of particulate collected on glass fiber
filters (based on 2000 cubic-meter air sample).  Sample collection
and sample digestion and analysis require relatively lengthy time
periods.

     Ando, et al. (ref. 18) improved the sensitivity by using a
nitrogen (entrained air)-hydrogen flame in conjunction with long-
path (91 cm) Vycor cells.  A detection limit of 0.006 yg of As
per ml of solution (H20) is claimed.  Major improvements in mini-
mizing flame absorption are attained.  Interferences are many:
(a) all acids as well as ammonia affect absorption of arsenic at
1937 A1, (b) magnesium, calcium, aluminum, nickel, cobalt and
manganese reduce arsenic absorption (attributable to formation
of arsenides in the flame), and (c) increasing concentrations of
iron first increase and finally decrease arsenic absorption.  Even
in very high reducing environments (hydrogen) dissociated neutral
atoms of arsenic combined with oxygen to form white deposits of
arsenic oxide on the cell wall causing serious changes in sensi-
tivity.  Cells had to be cleaned often.

     By generating arsine,  Holak (ref. 19) was able to detect
0.04 yg of arsenic by atomic absorption spectrophotometry.

     The arsine generator technique has been used in colorimetric
methods, but is also currently being promoted as a means of in-
creasing the sensitivity of atomic absorption spectrophotometric
measurement for arsenic.  Sampling systems for generating arsine
for delivery to the atomic absorption spectrophotometer are
available commercially from Perkin-Elmer Corp. and from Fisher
Scientific Co. (Jarrell-Ash Div.).  The basic apparatus and prin-
ciples are described by Fernandez and Manning (ref. 20) and
Manning (ref. 21).  An absolute detection limit of 0.02 yg is
reported; arsine generation requires 4-5 minutes.

     Folger, et al. (ref. 22) reported the development of an
arsine generating system using concentrated hydrochloric acid,
instead of sulfuric acid, and using a large surplus of fine-
grained zinc powder to a small volume of solution.  By operating
at room temperature, 90% of the arsine [10 yg As(III)] was gene-
rated in one second and 99-8 ± 0.25? was produced in one minute.

     3.2.2  Antimony

     Yanagisawa et al. (ref. 23) studied the determination of
antimony in different flames and found that the acetylene-air
                                309

-------
flame was most suitable.  They found no interferences from other
cations or anions .

     The 2176 S wavelength is generally recommended.  A sensiti-
vity of ca. 1 yg/ml can be attained and a copper interference is
noted (ref.
     Stresko and Martiny (ref. 25) report a negative effect by Cu
and an enhancement by Pb when the most sensitive line, 2176 ft, is
used with a 7 & slit.  The lead interference is not observed with
a 2 X slit.  Apparently the interference is due to the absorption
of an antimony line at 2170.23 A" by the primary lead line at
2169. 99^ & (ref. 26).  It is worthy to note that the opposite
effect of antimony on a lead determination does not appear.

3.3  Emission Spectroscopy

     Very little data are reported for flame-, arc-, or spark-
excited emission spectrographic analysis of arsenic and antimony.
Poor detection levels are generally reported.  Dean and Carnes
(ref. 27), and Dean and Adkins (ref. 28) show that emission lines
for elements of high excitation potential and high ionization
potential, that are generally absent or very weak in the portion
of the excitation zone (e.g., mantle of a flame) generally used
for analyses, appear with unusual strength in the spectrum of the
reaction zone (e.g., rich acetylene-oxygen flame).  Emission in-
tensity can be enhanced in the flame by using organic aerosols,
and continuous background radiation from the reaction zone can
be minimized by using a light-guide.

     In measuring solutions with flame emission spectroscopy ,
detection levels of 2.2 yg As/ml and 1.0 yg Sb/ml are reported
(ref. 27).  Some spectral interferences are observed but have
been minimized by selecting other analytical lines.  The 228.8 nm
arsenic line suffers direct interference from the cadmium line at
228.8 nm and the tin line at 228.7 nm.  The tin line at 235-5 nm
and the nickel line at 23^.5 nm could interfere with the arsenic
line at 235.0 nm if the slit width ever exceeded 0.3 nun.  Concen-
trations of antimony, up to at least 500 yg/ml, did not affect
arsenic.  However, arsenic depressed the emission of antimony by
approximately 2 yg/ml (out of 80 yg Sb/ml) for each 100 yg As/ml.
Tin at concentrations of 1000 yg/ml did not affect arsenic, but
zinc repressed arsenic by approximately 1% per 100 yg Zn/ml .

     The use of radio-frequency induced emission spectroscopy
requires additional investigation.  Dickinson and Fassel (ref. 29)
show detection limits in rf plasma emission of 0.1 yg As/ml and
0.2 yg Sb/ml.

     No data are available for continuous, flow-through, emission
studies on airborne arsenic and antimony.  However, modification
of techniques reviewed in other documents in this series,


                               310

-------
(a) beryllium and cadmium, (b) mercury, and (c) lead, and reported
elsewhere (ref. 30,31,32,33,31*) should be considered.

3.^  Atomic Fluorescence Spectroscopy

     Dagnall, et al. (ref. 35) developed an analytical technique
for antimony in aqueous solutions using atomic fluorescence spec-
troscopy.  Atomic fluorescence measurements at 2176 8 permits the
determination of antimony in the range 0.1-120 ppm with a detec-
tion limit of 0.05 ppm.  With the same equipment and source, the
range of measurement for atomic absorption was 6-120 ppm and the
detection limit was 1 ppm.  No interferences were observed from
100-fold molar amounts of Cd, Cu, Co, Fe, Hg, K, Mg, Mn, Mo, Na,
NHi,, Pb and Zn or from arsenate, chloride, nitrate, phosphate and
sulfate.

3.5  X-ray Emission Spectroscopy

     Excitation of secondary emission (fluorescence) of x-rays
can be induced by a number of excitation sources including elec-
tron beams, higher energy x-rays, or alpha, beta, and gamma
radiation from radio-isotopic sources.  Very little information
on the application to determination of arsenic and antimony in
airborne particulate is available.  Analyses have been reported
in a variety of liquid and solid matrices; detection of 1-10 yg
of arsenic can be attained with conventional x-ray spectrometry
(platinum target x-ray tube).  The technique is limited to analy-
ses of particulate collected on filters and is not applicable to
continuous monitoring.  The technique is also subject to matrix
and particle size effects.

     3.5.1  Arsenic

     The measurement of trace quantities of arsenic in a matrix
containing low to moderate proportions of lead are not possible.
Two effects are noted:  (a) high absorption properties of the
matrix (moderate to high quantities of lead), and (b) spectral
interference of the Pb La: (1.173 X) line with the As K0l
(1.173 8) line.  Determinations of arsenic in antimonial lead
alloys can be made in the range of arsenic contents, 0. 055&-0.06%,
by using the weaker Kg lines for arsenic (1.057 A*) (ref. 36).
Care must also be taken to minimize the background radiation in-
terference from the x-ray tube with the arsenic analytical lines,
particularly WLy3 with As Kgj (ref. 37).  Primary beam filters are
useful to a certain degree but some attenuation of the analytical
line can be experienced (ref. 37).

     3.5.2  Antimony

     Antimony (Ka 0.^72 X) has been measured in lead alloy
matrices (antimonial lead) (ref. 36) and in presence of Sn, Se,
As, Ge, and Cu (ref. 38).  The high proportion of lead affects
                               311

-------
the absorption of both primary and secondary x-rays resulting in
only moderately low detection limits.  However, limits of detec-
tion of 2 micrograms are attainable in other matrices (ref. 38).

3.6  Neutron Activation Analysis

     Nondestructive neutron activation analysis has been used to
measure airborne arsenic and antimony collected as particulate on
filters near several metropolitan areas.  Although high sensiti-
vity can be attained, long cooling times or separation techniques
are required to minimize interferences.

     3.6.1  Arsenic

     Nondestructive measurement of arsenic in airborne particu-
late by neutron activation analysis was reported by Dams, et al.
(ref. 39).  Detection limit of 0.04 yg or 0.004 yg/m3 in a 24-hour
sample are attained.  This detection level is attained by irradi-
ating with thermal neutrons for five minutes, waiting 20-30 hours
for cooling and counting for 2000 seconds.  The long cooling time
is required to minimize interferences.

     Rudelli et al. (ref. 40) reports on a variety of interfer-
ences involved in the application of NAA to determine arsenic and
antimony.  It is important to realize that care must be taken to
have sufficient resolution and to understand the potential inter-
ferences.  Limitations for the nondestructive activation analysis
of both arsenic and antimony can be minimized by using Ge(Li)
spectrometers.  It is notable, that in some samples where the
antimony content is high, arsenic was not detected.

     A rapid and non-destructive fast-neutron activation analysis
using a semiconductor detector (ref. 4l) was developed which would
permit a sensitivity of 10-100 yg/10 counts for arsenic after
irradiation of selected samples for 400 seconds.  The method was
proposed for industrial quality control.

     3.6.2  Antimony

     Rudelli, et al. (ref. 40) applied nondestructive activation
analysis to measuring arsenic and antimony in high tin matrices.
A problem arises from the contribution to the antimony content
by the matrix, even when chemical separation has been conducted.
This is due to the presence of 125Sb produced by the reaction
12I|Sn(n,Y) 12SSn  P~ r 125Sb.  The energy of the gamma rays of
125Sb is very similar to the energy of gamma rays of izzSb and
12l*Sb, both produced by n,y reaction from natural antimony.  The
best Nal(Tl) crystals do not provide sufficient resolution.  Also,
the presence of arsenic causes a serious interference because the
photopeak of 76As at an energy of 559 keV cannot be resolved from
the 564 keV peak of 122Sb with the Nal(Tl) crystals.  In addition,
unless sufficiently long cooling times are used, interferences
                               312

-------
can also be observed for copper, iron, zinc, silver and sodium.
All interferences can be eliminated by using higher resolution
Ge(Li) spectrometers.
     Fujii et al. (ref. 41) developed a rapid and nondestructive
fast-neutron activation analysis technique for a number of ele-
ments including antimony by using a co-axial Ge(Li) semiconductor
detector.  Based on irradiation times of 400 seconds and recording
times of 500 seconds, antimony was detected with sensitivities of
<10 yg/10 counts.

     A number of neutron activation analyses of airborne particu-
late has been reported (ref. 42,43).  Detection levels of antimony
in particulate collected on filters (cellulose) in the Chicago
area range from 1.4 ng/m3 to 55 ng/m3 (ref. 43).  With decay times
of 20-30 days and collection times of 24 hours, the minimum de-
tectable concentration of antimony in urban air was 0.001 yg/m3
(ref. 39).  Haines et al. (ref. 44) used an automatic smoke
sampler and neutron activation analysis to detect airborne anti-
mony oxide .  Background interference from soot and filter paper
limited the technique to about 0.2 micrograms of antimony without
separation procedures.

3.7  Mass Spectrographlc Analysis

     The sensitivity of the mass spectrometric method is dependent
in large part on the means by which the sample is excited.  Very
poor efficiencies are encountered with thermal ionization of ar-
senic.  The needed high sensitivity and selectivity for metals
are attainable only by ion- or spark-source mass spectrometric
techniques.  However, analyses must be performed in a relatively
high vacuum and, as a consequence, the measurements are made on
a batch basis and require considerable time between samples to
attain suitable conditions for efficient excitation.  The tech-
nique is limited to measuring samples collected on a filtering
device and is not applicable to direct and continuous source
monitoring.

     Due to the relatively high volatility of As20s and other
arsenic compounds, considerable error can be encountered due to
losses with high volume air sampling filters, digestion or sample
preparation steps, and pumping down to the necessary vacuum.

     Brown and Vossen (ref. 45) report the measurement of arsenic
(0.005 yg/m3) in air pollution particulates collected on a special
nitrocellulose filter pad.  Total analysis time, including com-
bustion of the filter and preparation of the graphite electrode,
required approximately 2-1/2 hours per sample.

     By using a special modification of a thermal ionization
source which allows an electron bombardment crucible source system
to be used in thermal ionization solid source mass spectrometers,
                                313

-------
a detection level of less than 5 yg antimony in solid specimens
is possible with low memory effect (ref.


4.   SUMMARY AND RECOMMENDATIONS
     There is no technique available for measuring arsenic and
antimony continuously in ambient air or in stack emission.  All
techniques presently used in measuring the air pollutants -
arsenic and antimony - require the collection of particulate in
impingers or on filters.

     The low sublimation temperature of As20a suggests the poten-
tial loss of volatile arsenic compounds from particulate collected
on high volume filtering systems.  Analytical techniques, e.g.,
x-ray emission (fluorescence) spectroscopy , neutron activation
analysis, and mass spectroscopy, using this method of collection
may yield low results.

     Similar low results will also be observed with emission
spectrographic techniques based on analyzing particulate collected
on filters.  However, emission spectrographic techniques based
the excitation of a continuous flowing stream of air or stack gas
through an excitation zone would not have this problem.

     Other than using flame (acetylene-oxygen), or radio-frequency
induced excitation of flowing solutions of arsenic and antimony,
no data are available regarding spectrographic analyses of flowing
air samples.  Based on information derived from flame excitation
studies, a moderate amount of development work should be antici-
pated to establish the optimum viewing zone in the flame or
plasma.  Emission lines of elements (arsenic and antimony) of
high excitation potential and high ionization potential may not
appear in the flame mantle, but occur as intense emission lines
in the reaction zone under the proper conditions.  The use of
radio-frequency or microwave-induced plasmas should be considered.
The latter when used in a low power mode will not permit direct
measurments, but will require isolation of the arsenic or antimony
in impinger solutions.

     Considerable difficulty can be experienced in attempting
arsenic analyses in high lead matrices by x-ray emission (fluores-
cence).  Not only does the high absorption characteristic of the
matrix cause problems, but direct spectral interference of Pb La,
radiation occurs with As Kai.

     When using neutron activation analysis, particularly as a
nondestructive technique, instrumentation having high resolution
capability [Ge(Li) semiconductor detector] is necessary to mini-
mize a variety of interferences in measuring arsenic and antimony
in stack emission particulate.
                               314

-------
     The use of an arsine generation step aids in concentrating
arsenic for atomic absorption spectrophotometry, colorimetry,
and potentially flame-, rf- or microwave-excitation of emission
spectra, but the total analysis time requiring collection of
sample, digestion to yield trivalent arsenic, and generation of
arsine is too long for short-term, intermittent measurements.

     Although a long path absorption tube improves the sensitivity
of the atomic absorption method for arsenic, considerable inter-
ferences (formation of arsenides, oxides) can occur, resulting in
low arsenic measurements.  High flame or non-flame temperatures
are needed.

     Considerable development work is needed to attain continuous
monitoring of arsenic and antimony from stack emissions.  The
most promising approaches are based on direct measuring of emis-
sion spectra generated by an arc excitation or by a radio-frequency
plasma.
                               315

-------
5.    REFERENCES


 1.   "Arsenic  and Its Compounds,"  Hygienic  Guide  Series,  Am.  Ind.
     Hyg.  Assoc.  J.  25.,  610-613  (1964).

     "Arsine," Hygienic  Guide  Series,  Am. Ind.  Hyg.  Assoc.  J.  26,
             (1965).
 3.   Sullivan,  R.  J.,  "Preliminary  Air  Pollution  Survey  of  Arsenic
     and Its Compounds," U.S.  Dept .  Health,  Education, and  Welfare,
     NAPCA,  Raleigh,  N.C.,  Oct.  1969-

 4.   Kirk,  R. E.  and  D.  F.  Othmer (ed.),  Encyclopedia of Chemical
     Technology,  Vol.  2, pp.  50-68,  Interscience  Encyclopedia,
     Inc. ,  N.Y. (1948) .

 5.   American Conference of Government  Industrial Hygienists:
     Manual of Analytical Methods.   1014  Broadway,  Cincinnati  2,
     Ohio  (1958).

 6.   Dubois, L.,  and  J.  L.  Monkman,  "Determination of Arsenic  in
     Air and Biological  Materials,"  Amer.  Ind.  Hyg.  Assoc.  J.  22,
     292 (1961).

 7.   Monkman, J.  L.,  and L. Dubois,  "Arsenic Digestion Losses  in
     the Preparation  of  Biological  Samples," Amer.  Ind.  Hyg.
     Assoc.  J.  2_3_, 327  (1962).

 8.   Sandell, E.  B.,  "Colorimetric  Determination  of Traces  of
     Metals," 2nd Ed.,  Interscience  Publishers, Inc., N.Y.  (1950).

 9.   Whitnack,  G.  L.,  and H.  H.  Martens,  "Arsenic in Potable
     Desert Groundwater: An Analysis Problem," Science  171,
     p.  383-385 (1971).

10.   Furman, N. H. (ed.), Standard  Methods of Chemical Analysis,
     Vol.  1, Sixth Ed.  p. 103, D. Van Nostrand Co.,  Inc.,
     Princeton, N.J.  (1962).

11.   Agranov, Kh.  I.,  "Automatic Gas Analyzer for Determining
     Trace  Concentrations of Gases," Nov.  Obi.  Prom.-Sanit. Khim
     1969  (60-71); C.A.  71, 116275P  (1969).

12.   Furman, N. H. (ed.), "Standard  Methods  of Chemical  Analysis,
     Vol.  I - The Elements," Sixth  Edition,  83-105,  D. Van
     Nostrand Co., Inc., Princeton,  New Jersey (1962).

13.   Kahn,  H. L.,  and J. E. Schallis,  "Improvement of Detection
     Limits for Arsenic, Selenium,  and  Other Elements With  an
     Argon-Hydrogen Flame," At.  Absorption Newslett . 7.(D,  5-8
     (1968).
                               316

-------
14.   Kirkbright,  G.  F.,  M.  Sargent,  and T.  S.  West,  "The  Determi-
     nation of Arsenic and  Selenium  by Atomic  Absorption  Spec-
     troscopy in  a Nitrogen-Separated Air Acetylene  Flame,"  At.
     Absorption Newsletter  8_,  34 (1969).

15-   Kirkbright,  G.  P.,  and L.  Ranson, "Use of the Nitrous Oxide-
     Acetylene Flame for Determination of Arsenic and  Selenium
     by Atomic Absorption Spectrometry," Anal. Chem.  43,  1238-
     1241 (1971).

16.   Smith, K. E., and C. W. Frank,  "Characterization  of  Arsenic
     by Atomic Absorption Spectroscopy in Oxy-acetylene Flames,"
     Appl.  Spectry.  22_,  765-768 (1968).

17.   Thompson, R. J., G. B. Morgan,  and L.  J.  Purdue,  "Analysis
     of Selected  Elements in Atmospheric Particulate Matter  by
     Atomic Absorption," Atomic Absorption Newsletter  £,  53-57
     (1970).

18.   Ando,  A., M. Suzuki, K. Fuwa, and B. L. Vallee,  "Atomic
     Absorption of Arsenic  in Nitrogen (Entrained Air) -  Hydrogen
     Flames," Anal.  Chem. 4^,  1974-1979 (1969).

19.   Holak, W., "Gas-Sampling Technique for Arsenic  Determination
     by Atomic Absorption Spectrophotometry,"  Anal.  Chem. 4_1,
     1712-1713 (1969).

20.   Fernandez, F. J., and  D.  C. Manning, "The Determination of
     Arsenic at Sub-microgram Levels by Atomic Absorption Spec-
     trophotometry," Atomic Absorption Newsletter  10,  86-88
     (197D.

21.   Manning, D.  C., "A High Sensitivity Arsenic-Selenium Sampling
     System for Atomic Absorption Spectroscopy," Atomic Absorp-
     tion Newsletter 10_, 123-124 (1971).

22.   Folger, H.,  J.  V. Kratz,  and G. Herrman,  "Rapid Volatiliza-
     tion of Arsenic, Selenium, Antimony and Tellurium in Form
     of Their Hydrides," Radiochem.  Radioanal. Letters,  1/3),
     185-190 (1969).

23.   Yanagisawa,  M., M.  Suzuki, and  T. Takeuchi, "Atomic  Absorp-
     tion Spectrophotometry of Antimony," Anal. Chim.  Acta 47,
     121 (1969).

24.   Mostyn, R. A.,  and A.  F.  Cunningham, "Determination  of
     Antimony by  Atomic Absorption Spectrometry,"  Anal.  Chem.
     39_, 433-442  (1967).

25.   Stresko, V., and E. Martiny, "Determination of  Antimony in
     Geological Materials by Atomic  Absorption Spectrometry,"
     Atomic Absorption Newsletter 11, 4-6 (1972).
                                317

-------
26.  Slavin, S., and T. W. Sattur, "Spectral Interference of Lead
     on Antimony," Atomic Absorption Newsletter 7_,  99 (1968).

27.  Dean, J. A., and W. J. Carnes, "A Study of the Emission
     Spectra of Arsenic, Antimony, and Bismuth from the Reaction
     Zone of Acetylene-Oxygen Flames," Analyst 8_7_,  743-747 (1962),

28.  Dean, J. A., and J. E. Adkins, "Excitation Gradients in
     Acetylene-Oxygen Flames," Analyst 9_1,  709-713  (1966).

29.  Dickinson, G. W., and V. A. Fassel,  "Emission  Spectrometric
     Detection of the Elements at the Nanogram per  Milliliter
     Level Using Induction-Coupled Plasma Excitation," Anal.
     Chem. *a, 1021-1024 (1969).

30.  Webb, R. J., M.S.W. Webb, and P. C.  Wildy, "A  Monitor for
     the Quantitative Determination of Beryllium in the Atmo-
     sphere," United Kingdom Atomic Energy  Authority, AERE-R
     2868 (1959).

31.  Webb, M.S.W., R. J. Webb and P. C. Wildy, "Monitor for the
     Quantitative Determination of Beryllium in the Atmosphere,"
     Journal of Sci. Instr. 37., 466-471  (I960).

32.  Fromm, D., and A. v. Oer, "Nachweis  von Beryllium und
     emissionsspektrographischer Nachweis von Quecksilberdampf
     in Luft," Mikrochem. Acta I960, 235-244.

33-  Aughey, H., "A Rapid Mobile Analyzer for Minute Amounts of
     Lead in Air," J. Opt. Soc. Am. 39., 292-293 (1949).

34.  Koppius, 0. B., "Detection of Lead in  Air With the Aid of
     a Geiger-Muller Counter," J. Opt. Soc.  Am. 39, 294-297
     (1949).

35.  Dagnall, R. M., K. C. Thompson, and  T.  S. West, "Studies
     in Atomic Fluorescence Spectroscopy  -  V.  The  Fluorescence
     Characteristics and Determination of Antimony," Talanta 14,
     1151-1156 (1967).

36.  Fillmore, C. L., A. C. Eckert, and J.  V. Scholle, "Deter-
     mination of Antimony, Tin, and Arsenic  in Antimonial Lead
     Alloys by X-ray Fluorescence," Appl. Spectry.  23, 502-507
     (1969).

37.  Gilmore, J. T., "Use of Primary Beam Filter in X-ray
     Fluorescence Spectrometric Determination of Trace Arsenic,"
     Anal. Chem. 4j), 2230-2232 (1968).

38.  Wlotzka, P., "Determination of Microgram Amounts of Sb, Sn,
     Se, As, Ge, and Cu in the Presence of  Each Other by X-ray
     Fluorescence Analysis," Z. Anal. Chem.  215.(2), 8l-5 (1966).


                                318

-------
39.   Dams,  R.,  J.  A.  Robbins,  K.  A.  Rahn,  and J.  W.  Winchester,
     "Nondestructive  Neutron Activation Analysis  of  Air  Pollu-
     tion Particulates," Anal.  Chem. 4_£, 861-86?  (1970).

40.   Rudelli,  M.  D.,  H.  C.  Rocca, and G. B.  Baro,  "Nondestructive
     Activation Analysis for Arsenic and Antimony  in Soft  Solder-
     ing Alloys,"  Nat.  Bur.  Stand.  (U.S.),  Spec.  Publ. 1969,
     No. 312(1),
41.   Fujii,  I.,  T.  Inouye,  H.  Muto,  K.  Onodera,  and  A.  Tani,
     "Rapid  and  Nondestructive Fast-neutron Activation  Analysis
     for Many Elements  by Using a Semiconductor  Detector,"
     Analyst £4  (1116), 189-97 (1969).

42.   Lee, J., and R.  E. Jarvis, "Detection of Pollutants  in
     Airborne Particulates  by  Activation Analysis,"  Trans. Amer.
     Nucl. Soc.  11, 50-51 (1968).

43.   Brar, S. S., D.  M. Nelson, J. R.  Kline, P.  F. Gustafson,
     E.  L. Kanabrocki,  C. E. Moore,  and D. M. Hat tori," Instru-
     mental  Analysis  for Trace Elements Present  in Chicago Area
     Surface Air,"  J. Geophysical Research 75_, 2939-2945  (1970).

44.   Haines, G.  P., H.  Cember, W.C.L.  Hemeon, "Development of
     Tracer  Techniques  to Establish  Geographical Distribution
     of  Pollution from  a Stack Using Neutron Activation Technique,"
     (Progress Report).  Ind .  Hyg. Foundation of America, Inc.,
     Pgh, Pa , July 1955; APTIC-22733.

45.   Brown,  R.,  and P.G.T.  Vossen,  "Spark Source Mass Spectro-
     metric  Survey  Analysis of Air Pollution Particulates,"
     Anal. Chem. 42_,  1820-1822 (1970).

46.   Tyrrell, A. C.,  J. W.  Roberts,  and R. G. Ridley, "Electron
     Bombardment Ion  Source for Isotopic Analysis of Solids,"
     J.  Sci. Instrum. 4_£, 806-807 (1965).
                               319

-------
                            APPENDIX VI
         BRIEFING DOCUMENT - ANALYTICAL  TECHNIQUES  FOR
BARIUM/ BORON/  CHROMIUM/  COPPER/  MANGANESE/  NICKEL/  AND  VANADIUM
                               320

-------
                        TABLE OF CONTENTS

Section

   1     INTRODUCTION

   2     SOURCES AND TYPE OP EMISSION                         322

            2.1   Barium and Its Compounds                    322
            2.2   Boron and Its Compounds                     323
            2.3   Chromium and Its Compounds                  325
            2.4   Copper and Its Compounds                    32?
            2.5   Manganese and Its Compounds                 329
            2.6   Nickel and Its Compounds                    331
            2.7   Vanadium and Its Compounds                  333

   3     CANDIDATE ANALYTICAL TECHNIQUES                      334

            3.1   Atomic Absorption Spectrophotometry         335
            3.2   Atomic Fluorescence Spectrophotometry       341
            3-3   Emission Spectroscopy                       3*11
            3.4   X-ray Emission Spectroscopy                 345
            3.5   Neutron Activation Analysis                 354
            3.6   Colorimetry                                 358
            3.7   Polarography                                361
            3.8   Back-scattering of B-radiation              361
            3-9   Mass Spectrometry                           362
            3.10  Chemiluminescence                           364
            3.11  Coulometry                                  364
            3.12  Polymer-Radiographic Methods                365
            3.13  Gas Chromatography                          365
            3.14  Reaction of Nickel Carbonyl With            366
                  Mercuric Oxide

   4     SUMMARY AND RECOMMENDATIONS                          366

   5     REFERENCES                                           369
                                321

-------
1.    INTRODUCTION

     Information related to the technology for continuous measure-
ment of barium, boron, chromium, copper, manganese, nickel and
vanadium in emissions from stationary sources is presented in this
report.  There are no continuous monitors currently available which
will measure these elements individually in all the physical and
chemical forms found in stationary emission sources.

     Data on physical and chemical properties of each element and
its compound are reported to show potential problems regarding
the physical state and reactivity of specific emissions.   Although
little information is available regarding continuous monitoring,
selected references are summarized to indicate the potential appli-
cation of certain analytical technology and limitations of other
analytical methods for continuous measurement of the emissions
from stationary sources.


2.    SOURCES AND TYPE OF EMISSION

2.1   Barium and Its Compounds

     The major industrial sources of barium emissions are the
processes involved in the mining, refining, and production of
barium and its compounds (ref. 1).  Quantitative data on the
amount of emissions from industrial processes are not available.

     The two main barium-containing minerals are barite (barium
sulfate) and witherite (barium carbonate).  Of the two minerals,
barite is the principal mineral used in the United States.  Barium
compounds are also found in the waste (gangue) in lead and zinc
ore deposits and in most igneous rocks.

     Barium oxide is reduced with aluminum powder to produce
barium metal which is recovered from the reaction retort by dis-
tillation at 1200°C at approximately 0.1 mm.  Barium metal is
very reactive and is used to remove residual gases in radio tubes
and to produce barium-nickel and lead, barium and calcium alloys.

     Barite (barium sulfate) is used for oil-drilling muds, as
filler in paper, cloth, etc., as the white paint pigment, litho-
pone, as raw material for producing barium compounds, and as a
component in glasses, glazes and enamels.  Lithopone - 70% barium
sulfate and 30% zinc sulfide - is used as a pigment and is also
used in manufacturing of automobile tires, rubber matting, and
other rubber products.

     Barium sulfide is produced during the reduction of barium
sulfate with coal and serves as the raw material for other com-
pounds.  Barium carbonate produced by C02 or carbonate reaction
with the sulfide is an intermediate in preparing other barium
                                322

-------
compounds.  The carbonate is used In ceramic and glass industries;
in preparing case-hardened steels; and in manufacturing photo-
graphic paper.

     Barium chloride is the most important of the halides and is
prepared by reacting BaS with hydrochloric acid to yield the
chloride salt and hydrogen sulfide.  The chloride is used as a
chemical intermediate, as a filler, and in steel carburizing.
The bromide is a raw material for preparing phosphors and the
fluoride is used as a flux and opacifier in enamels and is used
in glass manufacture.

     Barium nitrate is used in pyrotechnics, in green flares, in
tracer bullets, in primers and in detonators.

     Barium oxide has a high reactivity to water and carbon di-
oxide and readily forms the hydroxide and carbonate.  The oxide
is formed by heating the carbonate with carbon in a brick kiln or
electric-arc furnace.  The oxide is used as an intermediate in pro-
ducing barium hydroxide and barium peroxide.  Other uses include
formulation of lubricating oil detergent, glass manufacturing, and
as a desiccant for refrigerants in household refrigerators.

     Barium stearate is utilized as a metallic soap and drier.
Barium acetate and barium benzene sulfonate are lubricating oil
additives.  Also, the acetate is a mordant for printing fabrics
and a catalyst for organic reactions.

     Barium chromate is a pigment in coloring glass, ceramic and
porcelain and is used in metal primers, pyrotechnics and safety
matches.

     The combustion of coal is an additional source of barium
emissions.  The barium content in coal ash ranges from 
-------
     Borax and boric acid have the greatest commercial applica-
tions of the Group 1 compounds.  Boron dusts of unknown composi-
tion (probably mixed oxides) are emitted in iron furnaces.
Limited data indicate a boron concentration of approximately 0.055?
in particulate from a gray iron furnace (ref.  2).   Boric acid is
used as a ceramic glaze, chemical intermediate, alloying material
for steel, antiseptic, flameproofing agent, cosmetic base, and
dye stabilizer, and is used in electroplating solutions, in
photography, and in latex paints.

     Borax serves in manufacture of glasses and ceramics; in
leather-tanning industry; in detergent and soap formulations; in
cosmetics.  It is also an effective preservative and fire retard-
ant; aids in controlling microorganisms in fuels and solvents;
and is an ingredient in artificial fertilizers.  Sodium tetra-
borate pentahydrate serves as a rust inhibitor in antifreeze solu-
tions, as an ingredient in liquid starches and as an intermediate
in preparing other borates.  Ammonium biborate is used in manu-
facture of organic resins; ammonium pentaborate is used in elec-
trolyte condensers.

     The boric acid esters (Group 2 compounds) are produced by
esterification of boric acid with alcohols or phenol or by reacting
boric oxide with alcohols.  The esters serve as fuel and lubricant
additives, as antiknock agents in gasoline (most important of all
uses), as growth inhibitors for microorganisms in diesel and avia-
tion fuels, as anti-icing agent, and as an anti-oxidant.  Other
uses include epoxy resin curing agent, nuclear-shielding, and
components in nuclear-detecting devices.

     The Group 3 refractory boron compounds include metal borides,
boron carbide, and boron nitride.  The metal borides generally
contain one or more of the following elements in combination with
boron:  aluminum, chromium, nickel, calcium, barium, manganese
and copper.  The borides are used primarily in rocket engines,
jet turbines, and other equipment subjected to high temperatures
and chemically reactive agents.  Boron carbide serves as an
abrasive, in jet engine parts, and as a semiconductor.  Boron
nitride is used in manufacture of heat resistant materials and
for high-voltage electrical insulation.

     The boron halides of Group 4 include the trichloride, tri-
fluoride, tribromide, and triiodide compounds.  In addition, the
trifluoride combines with the fluoride ion to form fluoborate
ion, BFi,, which, in turn, can combine with cations to yield the
fluoborates, e.g., NaBPi, .  The halides, BF3 and BC13, are useful
as catalysts; all trihalides of boron hydrolyze readily.

     Boron hydrides, boranes, which comprise the Group 5 compounds
are represented by two generic formulae:  B H  j, and B H  /..  The
most important are the compounds in the firstnseries where n = 2,
                                324

-------
5, 6, and 10, and in the second series where n = 4 and 5-  Diborane
is a toxic gas that readily ignites in moist air; diborane is used
as a polymer catalyst and in manufacture of higher boranes and
amine complexes.  Higher boranes are used as high-energy fuels.
Decaborane is the most stable borane and is used in vulcanization
of rubber and as a fuel for rockets.

     Boranes also form metal borohydride or hydroborates with
metals.  The general formula is M(BHi») , where x may vary from
1 to 4 depending on the metal atom.  Typical compounds are LiBHi,,
NaBH,,, Ee(BH,,)z) A1(BH^)3, Ga(CH3) 2-BH^ , Mg(BH,,)2, Ca(BH^)2,
V(BHUK, and Zn(BHi,)2.

     The boranes also react with amalgams of the highly electro-
positive metals to form addition products - Na3B3H6, K3Bi»Hi0,
and K2B5H9.

     Polymerization of borazole, produced by reacting borane and
ammonia, yields a long-chain polyphenyl-type product.

     Group 6 compounds are produced by replacing hydrogen atoms
of the boranes with alkyl or trimethyl amine groups.  Little
information is available regarding the commercial importance and
production figures for these compounds.

     The combustion of coal is an additional source of boron.
Concentrations of approximately 0.005 to 0.2% of boron have been
found in ash from various grades of coal.

     Boron boils at 2550°C and boron oxide has a boiling point of
1500°C.  Boron nitride sublimes at 3000°C, whereas the trihalides,
BBr3, BC13,  BF3, and BI3 boil at 90.1°C at 7^0 mm, 12.5°C, -10.°C
and 210°C, respectively.  The boranes are very volatile; typical
boiling points are:  diborane, -92.5°C; pentaborane, 0°C at 66 mm;
decaborane,  100°C at 19 mm; and trimethylborane, -20.2°C.  Tri-
methyl borate boils at 68.7°C.

     The Threshold Limit Values (A.C.G.I.H.-1967-TLV) are boron
oxide, 15 mg/m3; boron trifluoride, 3 mg/m3; diborane, 0.1 mg/m3;
pentaborane, 0.01 mg/m3; and decaborane 0.3 mg/m3.  Although no
TLV has been established for boric acid, Hyatt and Milligan
(ref. 3) suggest that the concentration not exceed 2 mg/m3.

2.3   Chromium and Its Compounds

     Chromium emissions can occur in the metallurgical and chemi-
cal industries, in chrome plating, in processing asbestos, in
manufacturing cement, and in burning coal (ref. 4).

     The only important commercial chromium mineral is chromite
which has the general chemical formula FeO'Cr203 where MgO (4
can replace some FeO (11-20?) and portions of the Cr203  (42-55?)
                                325

-------
can be replaced by A1203 (7-16%).   CaO (1-5%) and Si02 (3-8?) may
also be present.  Chromite used in the metallurgical industries
is usually converted into ferrochromium or chromium alloyed with
iron, nickel, or cobalt.  Most of the chromium is used in making
stainless steel, other specialized steels, and nonferrous alloys.

     Because of its chemical inertness and high melting point
((\-3956°F), chromium ore is used to manufacture refractory bricks
and as coatings to close pores or join furnace bricks.

     Sodium chromate, produced directly by roasting chromite ore
with soda ash or soda ash and lime, and sodium dichromate are the
principal chemicals from which other chromium compounds are pre-
pared.  The monochromate ion exists in alkaline solution and the
dichromate in acid solution.  In preparing the chromate, trivalent
chromium dusts are emitted from dry ores at the beginning of the
process and, after chromite oxidation, chromate dusts are emitted.

     The mono- and di-chromates serve as pigments, as corrosion-
inhibitors, as oxidizing agents in dyeing, tanning and synthesis
of organic and inorganic chemicals.

     The chrome plating industry accounts for the largest quanti-
ties of chromium trioxide as chromic acid, which can be emitted
as a mist during evolution of hydrogen bubbles.

     Chromates and chromic oxide,  combined with zinc, lead, iron,
barium, molybdenum, and strontium to form a variety of colors,
serve as pigments.  Primer paints  and dips containing chromates
are used in the construction steel and automobile industries.

     Bichromates are used in printing and reproducing processes
and serve as mordants in the textile industry.

     Chromium carbonyl serves as a catalyst for olefin polymeri-
zation and as a gasoline additive.

     Chromium chemicals, mostly as chromates, are used in manu-
facturing fungicides and as wood preservatives (industrial cooling
towers).  Other manufacturing uses include production of paper
matches (l^CraO?), fireworks, dry-cell batteries (lithium chro-
mate), and antistatic additives for fuels (chromium salt of
alkylated salicylic acid).

     Other emission sources are industries processing asbestos,
manufacturing cements, and burning coal.  Chromium emissions from
coal-fired power plants have been  reported as 18-500 yg/m3 after
emission controls.

     Chromium melts at 1890°C and  boils at 2480°C.  The trioxide
melts and begins to decompose at 197°C with loss of oxygen to
                               326

-------
form lower oxides.  The melting and boiling points of chromic
oxide are 1990°C and ^3000°C, respectively.

     The volatilization of chromic oxide can be enhanced by form-
ing chromium trioxide in an oxygen atmosphere.  Further enhance-
ment of volatilization occurs with presence of water in the oxygen
atmosphere and in a dynamic system with a considerable flow of
gas (ref. 5).  Chromium trioxide has been observed on the surface
of chromic oxide at relatively low process temperatures, 100-300°C
(ref. 6).  These data suggest the possibility that under certain
conditions difficulties may be encountered in attempting to col-
lect quantitatively small quantities of oxides of chromium by
using physical filtering methods and high volume flow rates for
excessively long sampling periods.

     Other chromium compounds and some selected physical con-
stants are:  chromium carbonyl - sublimes at room temperature,
sinters at 90°C, decomposes at 130°C, and explodes at 210°C;
chromic chloride - sublimes at 1300°C; chromic fluoride - b.p.
>1100°C; chromic formate - decomposes above 300°C; chromic
nitrate - decomposes above 60°C; chromous fluoride - b.p. >1300°C;
chromyl chloride - b.p. 117°C; chromyl fluoride - sublimes at
29l6°C; lead chromate - m.p. 84J|0C; potassium chromate - m.p.
975°C; potassium dichromate - m.p. 398°C, decomposes ^500°C;
sodium chromate - m.p. 20°C; sodium dichromate - m.p. 356. 7°C;
ammonium dichromate - decomposes ^l80°C, ignites readily.

     The Threshold Limit Values (A.C.G.I.H. 1967-TLV) for chromium
and its compounds are:  chromium - 1 mg/m3; chromates - 0.1 mg/m3;
chromic salts, soluble - 0.5 mg/m3; chromous salts, soluble -
0.5 mg/m3; chromium, insoluble compounds - 1 mg/m3; chromium
trioxide - 0.1 mg/m .

2.^   Copper and Its Compounds
     The ores of copper are categorized as:  (1) native ores
(pure metal), (2) oxide ores, and (3) sulfide ores.  At the
present time, only a very small proportion of U. S. copper comes
from native ore.  The original deposits of oxide ores are largely
depleted.  Currently, the sulfide ores are the most important
sources of copper (ref. 7).

     The sulfide ores are usually complex mixtures of copper and
iron sulfides associated with compounds of iron, zinc, arsenic,
antimony, bismuth, selenium, tellurium, silver, gold and platinum.
Most domestic ores contain less than 2% Cu although imported ores
contain 20, 30, and even 5055 Cu.  U. S. ores are generally treated
at the mine by a concentration process to upgrade the quality.

     The smelting process is generally carried out in a blast
furnace or a reverberatory furnace.  The flue dust will consist
of both original and decomposed, or partially decomposed, par-
ticles of ore, flux, furnace lining, and fuel.


                               327

-------
     Analyses (ref.  8) of flue dust from roasters,  reverbatory
furnaces, blast furnaces and converters have the following ranges:
Cu (5-32%), S (6-14%), Si02 (5-26%), FeO (6-35%),  CaO (0.3-6.4%),
A1203 (7-8%), Pb(6-20%), with smaller amounts of Ag and Au.

     A large percentage of the copper is used as alloys - brasses,
tin bronzes, aluminum bronzes, nickel silvers, and the cupro-
nickels.   Addition of magnesium, beryllium,  aluminum and silicon
have been used to retard oxidation of copper.  Other elements
(Sn, Ag,  Mn, P, Cd,  Sb, S, As, Ni, Au, Si,  Zn, and Bi) change the
softening temperature.  Other elements - Fe, Te, Se, Co, Pb, Cr -
can also be present  in the alloys.

     Cupric acetate  (b.p. 240°C) is a basic  raw material for in-
secticides (arsenite); basic cupric acetate  is important as cata-
lytic ingredient in  organic synthesis.  Basic cupric carbonate
serves as a raw material, a pigment (ceramics, paint and varnish
industries), a fungicide, and a pickling agent for brass.

     Cuprous chloride (b.p. 1366°C) is used  as a catalyst in
organic and inorganic chemicals manufacture  and as a condensing
agent for soaps, fats, and oils.  It is used in the petroleum
industry as a decolorizing and desulfurizing agent.  It is also
used in denitration  of nitrocellulose.  Cupric chloride [b.p.
993°C (dec.)] serves as a mordant in textile industry; it is used
in refining of copper, gold, and silver, in  photography and in
deodorization and desulfurization of petroleum distillates.
Cupric oxychlorides  are formed by exposure  of cupric chloride to
air and are used as  pigments and fungicides.

     Cuprous bromide has a boiling point of  1345°C; the cupric
bromide (m.p. 498°C) is used in photography  and organic synthesis.
Cuprous iodide boils at 1290°C.  Cuprous fluoride sublimes at
1100°C.

     Cupric nitrate, as tri- and hexa-hydrates, boils with decom-
position and loss of nitric acid at 170°C and is used in electro-
plating solutions.

     The most important oxides are cuprous  and cupric forms.
Cuprous oxide boils  with loss of oxygen at  1800°C and is manu-
factured by furnace  or electrolytic methods.  It is also produced
as a by-product of the metallurgy of copper  refining and fabri-
cation.  Cuprous oxide is used in production of copper salts, in
ceramics, in porcelain glaze, in electroplating, in paint, as
a fungicide, and as  a rectifier in the electrochemical industry.
Cupric oxide decomposes at 1026°C; reduction to copper occurs
when heated in H2, CO and many carbon compounds.  Cupric oxide is
used extensively for preparing cuprammonium  hydroxide solution in
rayon industry and as a pigment in glasses,  glazes, and enamels.
It is also used in the petroleum industry for desulfurizing oils.
                               328

-------
     Cupric hydroxide serves as a mordant and pigment.

     Cupric sulfate pentahydrate is the most important of copper
chemicals.  It decomposes to cupric oxide at 650°C and effloresces
in dry air, giving off two molecules of water.

     Copper sulfate is the starting material for the preparation
of many copper compounds, but has its greatest use as a fungicide.
It is also used as a "sweetening" agent in oil refining and as an
algaecide.  Other uses include preparation of copper ferrocyanide,
diazo colors, as a mordant, and as an electrolyte.

     Cuprous and cupric sulfides are important sources of copper.
Cuprous sulfide decomposes to cupric sulfide, cupric sulfate, and
sulfur dioxide when heated in air, but in absence of air only
copper and cupric sulfide are formed.  At red heat, hydrogen
slowly and completely reduces cuprous sulfide, but carbon and
carbon monoxide have no appreciable action.  Exposure to chlorine
results in slow decomposition, and slight decomposition occurs
with presence of CC>2 .

     Cupric sulfide boils with decomposition at 2200°C.  When
dry, cupric sulfide is stable in air, but when moist it oxidizes
to cupric sulfate on exposure to air.  In absence of air, cupric
sulfide decomposes to cuprous sulfide and sulfur when heated to
redness.  Reduction takes place in the presence of hydrogen.

     The Threshold Limit Values (A.C.G.I.H. - 196?) reported for
copper-dusts and mists, and for copper-fume are 1.0 mg/m3 and
0.1 mg/m3, respectively.

2.5   Manganese and Its Compounds

     In a literature review of manganese and its compounds,
Sullivan (ref. 9) identifies the major emission sources.  The
major emissions of manganese compounds occur in the iron and
steel industry - ferromanganese blast furnaces, electric-arc
furnaces, and basic oxygen and open-hearth furnaces.  Sullivan
(ref. 9) reports that the particulates from a ferromanganese
furnace are extremely small (0.1 to 1 micron) and contain man-
ganese (15-25%), iron (0.3-0.5%), sodium and potassium oxides
(8-15%), silica (9-19%), alumina (3-11%), calcium oxide (8-15%),
magnesium oxide (4-6%), sulfur (5-7%) and carbon (1-2%).  Some-
what similar data (ref. 9) are reported for electric-arc furnaces
where dust and fume particulate contain approximately 4% MnO with
70% of the particles less than 5y in diameter.

     Power plants and steam generator systems burning fossil
fuels - coal and residual fuel oils - emit particulate containing
approximately 60-400 yg/m3 when coal-fired and approximately
25 yg/m3 when fired with residual fuel oils (ref. 9).
                               329

-------
     Most manganese emissions from these sources are in the form
of oxides - MnO, Mn02, Mn203, and Mn30i» .  Manganese and MnO when
heated in air are readily converted to Mn203 and Mn30i», respec-
tively; Mn02 (307-600°C) and Mn203 (800-1200°C)  lose oxygen to
form Mn203 and Mn30i»,  respectively, when heated.  However, in the
presence of S02 and N02, these oxides, particularly Mn02, react
to produce sulfates, dithionates, and nitrates.

     Manganese is combined with many elements in the common man-
ganese ores.  These impurities can be classified into four general
groups:  (1) metal - iron, zinc, lead, copper, and silver;
(2) gangue - silica, alumina, lime, magnesia, and baria;
(3) volatile matter - water, carbon dioxide, organic matter;
(4) nonmetallics - sulfur and phosphorus (ref. 10).

     Manganese is used in iron, aluminum, aluminum bronze, con-
stantan, manganese bronze, Monel, nickel-chromium, nickel-silver,
and copper alloys.  A large number of manganese compounds are
produced commercially, but the major products are the permanga-
nates, manganese carbonyls, manganese dioxide, manganous acetate,
manganese chloride, manganous sulfates and manganese soaps.

     The permanganates are used commercially as oxidizing agents.
Manganese carbonyls, e.g., methylcyclopentadienyl manganese tri-
carbonyl, serve as antiknock and smoke inhibiting fuel additives.
Many uses exist for manganese dioxide, but major application is
in production of dry cell batteries.  Manganous acetate is a
mordant in dyeing, tanning and finishing leather, a component
of fertilizers, a drier in paint and varnish, and a catalyst.
Manganese chloride is a chemical intermediate; manganous sulfate
is used also as a chemical intermediate and in fertilizers or
crop sprays.  Manganese soaps are driers, lubricants (wire draw-
ing), and catalysts.

     Other sources of manganese emissions include the fertilizer
industry - manganous acetate, manganous sulfate, and manganese
ethylene—bis-dithiocarbamate; control of sulfur dioxide, mercap-
tan, and sulfide emissions; and welding shops.

     Manganese boils at 1900°C and manganese dichloride has a
boiling point of 1190°C.  Most other halides decompose when
heated.  Manganous sulfate decomposes at ^850°C, and the dioxide
(307-600°C) and sesqui-oxide (800-1200°C) lose oxygen when heated.

     Manganese organometallic compounds are very volatile.  Typi-
cal boiling points are:  164-8°C @ 10 mm (di-tert-butylcyclo-
pentadienyl manganese tricarbonyl) and 150°C @ 28-29 mm (tert-
butylcyclopentadienyl manganese tricarbonyl).

     The Threshold Limit Values (A.C.G.I.H - 196?) for manganese
are reported as a ceiling "C" limit value of 5 mg/m3.  The U.S.S.R
                                330

-------
has recommended a 24-hour limit of 10 yg/m3 and a maximum allow-
able one-time concentration of 30 ug/m3 for manganese and its
compounds (ref. 9).

2.6   Nickel and Its Compounds

     Nickel or its compounds is emitted from metallurgical plants
using nickel, nickel plating facilities, production facilities
preparing nickel compounds, power plants and steam generating
systems burning coal and oil; and incinerators (ref. 11).

     The most common nickel minerals are sulfides and oxides;
others having less importance are silicates and arsenicals.  The
silicate ore, garnierite, is refined in the United States in
electric furnaces to yield ferronickel.  Most nickel is imported
after being refined.

     The principal use of nickel is in the metallurgical industry
for preparing alloys, e.g., stainless steels and specialty alloys
containing high nickel contents.  Other uses include nickel
plating - electroplating, chemical plating, electroforming, nickel
cladding, spraying and fusing of powdered nickel, and chemical
vapor deposition by decomposition of nickel carbonyl, nickel
chloride, or nickel acetylacetonate.

     Nickel and nickel carbonyl serve as catalysts.  Thermal
decomposition of nickel carbonyl is used to produce high purity,
finely powdered nickel for powder metallurgy.  Nickel carbonyl,
cyclopentadienyl nickel complexes, nickel alkylphosphates, and
nickel chelates of salicylaldehyde nitrites and 6-diketones are
used as fuel additives.  Nickel carbonyl can be formed at 25°C by
reaction of carbon monoxide and powdered nickel or nickel pipe.

     Nickel acetate serves as a mordant and catalyst; nickel
carbonate (basic) and nickel formate are intermediates in pre-
paring finely divided nickel catalysts.  Nickel chloride is used
in electroplating and as an intermediate for preparation of nickel
catalysts.

     Nickel nitrate serves primarily as an intermediate for pre-
paring nickel catalysts, particularly impregnated nickel catalysts,

     Nickel monoxide is the most common oxide and is generally
prepared by calcining sulfide ores although hydroxide, carbonate,
formate, nitrate and other salts are also used.  A green nickel
oxide, containing 76.0% nickel, a fraction of a percent of cobalt,
0.5% iron,  0.25% copper, and 0.75% sulfur, is formed from nickel
sulfide.  The green form is only partially soluble in acids.  An
additional  calcining step with sodium carbonate yields a black
modification with lower sulfur (^0.03%) and higher solubility.
Nickel oxide is used in production of ferrous and nonferrous
alloys, as  an ingredient in enamel frits, as a pigment in glazes,
as a coloring and decoloring agent in glass manufacture, as a raw


                               331

-------
material for preparing ferrites, and as a reagent for other nickel
compounds and catalysts.

     Nickel hydroxide is  usually prepared from nickel sulfate and
sodium hydroxide.  The hydroxide is used in electroplating, in
alkaline storage batteries, and in preparing nickel catalysts and
salts of organic acids.

     Nickel sulfate, as the hexahydrate, is the major nickel salt
of commerce and is prepared from high-sulfur nickel shot and hot
concentrated sulfuric acid.  The sulfate is the most widely used
nickel compound.  It is an important constituent of nickel-plating
baths, is an intermediate for preparing nickel catalysts, and is
a reagent for preparing other nickel compounds.

     Nickel ammonium sulfate hexahydrate is a double salt of
nickel sulfate with ammonium sulfate and is useful in nickel-
plating baths.

     Nickel sulfide can exist in several forms having varying
degrees of solubility depending on the method of preparation.

     The three most common nickel soaps are the naphthenate,
oleate and stearate.  They have been used as special lubricants
in metal drawing, and as  crankcase lubricants.

     Nickel ferrites belong to a class of mixed oxides of the
type MO'PeaOa where M is  one or more divalent metals as Mg, Zn,
Ni, Co, Fe, and Mn.  These materials require substantial quanti-
ties of nickel oxide and  are formed by high temperature sintering
of mixtures of oxides under strongly oxidizing conditions.  They
possess useful magnetic properties and are used in the electronics
industry.

     Nickel forms a number of complexes with ammonia, amines,
diamines, dialkyldithiocarbamates, and oximes.  Nickel amines
are used in extracting nickel from oxidized ores; carbamates
serve as antioxidants, fungicides, and inhibitors to prevent
cracking in GR-S rubber;  oximes can be used as pigments.

     The boiling points of nickel, nickel carbonyl, and nickel
chloride are 2900°C, ^3°C, and 973°C, respectively.  Nickel sul-
fate decomposes above 840°Cj evolving SOs and yielding a residue
of nickel oxide.  Nickel  sulfide melts at 797°C and yields a
basic sulfate on heating  in air.  The nickel nitrate of commerce
is usually the hexahydrate that melts or dissolves in its own
water of crystallization  at about 55°C and with additional heating
loses water up to 105°C,  at which temperature water vapor and
nitrous oxide fumes are evolved to yield a residue of NiO.  At
about 200°C, nickel formate starts to decompose, yielding nickel,
carbon dioxide, hydrogen, water, carbon monoxide, and a trace of
methane.
                               332

-------
     Nickel chloride exists more commonly as the hexahydrate
which loses water on heating to form lower hydrates and finally
is completely dehydrated by gentle heating above 1^0°C.  The
anhydrous salt is sublimed by further heating in vacuo or in a
stream of hydrogen chloride.  The vapor pressure of the anhydrous
salt is 38.9 mm @ 5^1°C.  When sublimation occurs in air, some
decomposition occurs, forming chlorine and nickel oxide.

     Nickel is also emitted in facilities handling asbestos mate-
rial.  Chrysolite contains from approximately 0.1-0.2% nickel.
Nickel emissions as fly ash from coal-fired power plants after
abatement controls are reported as ranging from 130-690 yg/scf
(ref. 12).  Fly ash from residual fuel oil has been reported as
containing 1.8 to 13.2% nickel oxide (ref. 13).

     The Threshold Limit Values (A.C.G.I.H. - 1967) are 1 mg/m3
for nickel-metal and solid compounds and 0.007 mg/m3 for nickel
carbonyl.

2.7   Vanadium and Its Compounds

     Industrial stack emissions of vanadium occur from industries
producing the metal, its chemical compounds and alloys.  In addi-
tion, considerable amounts are emitted by power plants and utili-
ties burning coal and residual or crude oils.  Data on emission
sources and types of emissions have been reported by Athanassiadis
(ref. 14).  Because of the affinity of vanadium for oxygen, car-
bon, and nitrogen, it is difficult to produce the elemental metal.

     The most important ores are the sulfides or partially oxi-
dized sulfides, the vanadates, the carnotite-roscollite minerals -
K20'2V03'V205'3H20 and K(V,A12)(AlSi3)0i„(OH,P)2, the lead-zinc-
copper-vanadates, the chlorovanadates, the titaniferrous magne-
tites, phosphate rock, chromite ore, and ferrophosphorus ores.

     Vanadium is recovered from ores by ion exchange and solvent
extraction, by acid precipitation, and by a leach roasting pro-
cess.  The product of the processing of the ores is a sodium
and/or calcium hexavanadate and is usually expressed as vanadium
pentoxide.

     A highly dispersed aerosol of vanadium pentoxide is formed
from condensing vapor during the smelting operation.  Prior to
use in alloying, the pentoxide is reduced in a slow process to
lower oxides and finally metal.  At the temperatures (above 3000°C)
of the alloying processes, considerable amounts of volatile
pentoxide (b.p. 1750°C w/decomposition) can be generated.  Melting
of ferrovanadium results in pentoxide concentrations in working
environments of 68,000 ug/m3 of air, and concentrations of lower
oxides up to 450 ug/m3 of air (ref. 14) .
                                333

-------
     Reduction of the oxide is generally accomplished with carbon,
with silicon, or with aluminum.  The major alloys are ferro-
vanadium, vanadium-carbide, vanadium-silicon, and vanadium-
aluminum .

     A large variety of vanadium compounds with vanadium in
various valence states can be formed.  In general, the halides
are hygroscopic and hydrolyze in water.  Both the halides and
oxyhalides are volatile - VCU (b.p. l48.5°C); VF3 (sublimes at
red heat), VOC1 (b.p. 127°C); VOC13 (b.p. 126.7°C); VOF3 (b.p.
480°C); VF5 (b.p. 47.9°C).  Several oxides exist; V205 is sold
commercially and is formed when the vanadium oxides, chlorides
or oxychlorides are heated in air.  V205 dissociates into V02
and oxygen at temperatures only slightly above its melting point
(690°C).  Numerous vanadates can be formed.   Also, a variety of
sulfates have been prepared.

     Because the chemical and physical characteristics of the
vanadium compounds differ widely, many uses  exist.  In catalysis,
many different vanadium compounds have been  used.  Two of the
more important compounds are V20s (particularly in the contact
process for sulfuric acid) and VOC13 in ethylene-propylene syn-
thetic rubbers.

     Other uses include medicinals (antiseptic and inhibitor of
micro-organism growth); dye manufacture (aniline black); inks
(indelible inks and quick-drying inks promoted by ammonium meta-
vanadate); glass making (actinic glasses); ceramics; photography;
metal soaps (driers).

     Power plants and utilities using fossil fuels - residual
and  crude oils, and coals - and petroleum refineries of crude,
which burn crude oil or residuals, emit vanadium compounds as
inorganic particulate and possibly as organo-complexes.   The
concentration of vanadium from fly ash collected from coal-fired
power plants after the abatement controls was reported as
230-390 yg/m3 and 1,350-1,580 yg/m3 (ref. 12).

     The Threshold Limit Values (A.C.G.I.H.  - 196?) for vanadium
(V205) dust and vanadium (V20S) fume are 0.5 mg/m3 and 0.1 mg/m3,
where the value for dust is a ceiling "C" limit that should not
be exceeded and the value for fume is a time-weighted average
concentration for an 8-hour working day.


3.   CANDIDATE ANALYTICAL TECHNIQUES

     Except for special cases, e.g., nickel carbonyl, boranes,
and chromic acid mist, no direct reading instruments for field
and continuous or semi-continuous application are available for
measuring Ba, B, Cr, Cu, Mn, Ni, and V  in ambient air or in stack
emissions.
                                334

-------
     Laboratory methods are based on collecting dust or fumes
with filter paper samplers, electrostatic precipitators,  and/or
impingers.

     Analytical techniques relying on instrumental methods have
been applied to measure Ba, B, Cr, Cu, Mn, Ni, and V in various
media, including particulate collected from ambient air.   The
techniques  used predominantly are colorimetry, atomic absorption
spectrophotometry, atomic fluorescence, emission spectroscopy,
x-ray fluorescence, neutron activation analysis and mass  spectrom-
etry.  Most of these methods are based on analyses of particulate
only when collected on filters (paper, organic membrane or glass
fiber) and  suffer from the possibility of potential loss  by vapor-
izing volatile compounds from stack emissions when large  volumes
of air are  drawn over the collected particulate.  This effect would
be noticed  readily when collecting compounds that are obviously
volatile, but the effect could occur also in special cases, e.g.,
the volatilization of chromic oxide can be enhanced by forming
chromium trioxide in an oxygen atmosphere, particularly in a
dynamic system with a considerable flow of gas.  Some of  this loss
can be minimized by a tandem arrangement of filter and impinger
containing  a collecting liquid.  The vaporization losses  on fil-
ters could  be moderately high when sampling heated stack  gases.

     No attempt is made in this report to cover the numerous
spectrophotometric or wet chemical methods that have been applied
to the determination of the subject elements.  Information that
is pertinent to technology related to the continuous or short-term
intermittent monitoring of the seven metals is reported.   No
analytical  technique is currently available that can be immediately
and directly applied to the analysis of total B, Ba, Cr,  Cu, Mn,
Ni, and V,  as vapor, fume, and particulate.

     The following sections contain technology related to analy-
tical methods that can be applied to the continuous or short term
measurement of metals from stack emissions.  Although a variety of
techniques  are presented, the most attractive methods for general
application (vapor and particulate) based on selectivity  and
sensitivity, are absorption and emission spectroscopy.  Others,
e.g., x-ray fluorescence, may find application in measuring
particulate and some, e.g., gas chromatography, would provide
specific analyses for nickel carbonyl and organo-metallics.

3.1   Atomic Absorption Spectrophotometry

     Because of moderately high sensitivity and good specificity,
atomic absorption spectrophotometry has been used to measure trace
quantities  of metallic elements collected from ambient air after
dissolution of the particulate in acid.  The major problem in
applying the technique to continuous monitoring of stack emissions
                                335

-------
for B, Ba, Cr, Cu, Mn, Ni, and V is the inability to handle re-
fractory compounds, for example, B, and elements that readily
form stable compounds - VO - in the flame.   Serious interelement
and matrix effects are observed.

     All seven elements are affected by matrix effects and com-
pound formation, particularly in air- or oxygen-acetylene flames.
As a consequence, direct aspiration of air-contaminated gas into
the flame has not been used to any large degree for continuous
monitoring of metallic elements in air or stack emissions.

     Depending on the concentrations of the element in the stack
and the matrix, a continuous monitoring system based on direct
aspiration of stack gas or air into the acetylene fueled flame
can be developed, but the estimated detection limit would probably
not be <1 mg/m3 without a preconcentration step.

     The direct aspiration of nickel concentrated into a small
volume air stream is reported but no specific details are given.
The apparatus (Perkin-Elmer Model 303 Atomic Absorption Spectro-
photometer) modified for simultaneous determination of four
selected elements was described by Zweibaum and Moorhead (ref. 15).
The system was designed for continuous monitoring of several ele-
ments concentrated from the atmosphere into a small volume air
stream.  The air stream with enriched sample content is fed
directly and continuously into the burner nebulizer.  Applications
to nickel are reported, but no data are given.

     The highest sensitivity and minimal matrix and chemical inter-
ferences for B, Ba, Cr, Cu, Mn, Ni and V are attained with nitrous
oxide as the oxidant and with a nitrogen sheath gas to restrict
interaction of ambient air with the chemical species of the flame.

     Interelement effects can also be eliminated by using a high
energy plasma to atomize the elements being measured.  The major
problem in applying this technique is maintaining high levels of
ground state atoms, but low ionization efficiencies, in the plasma.
Energy from the plasma may ionize the element sought, thus reducing
the population of ground state atoms which must be high to attain
high sensitivities for the atomic absorption phenomenon.

     Improved atomic absorption spectrophotometric detection
limits (<1 ppm) for boron in aqueous solutions have been attained
by using the Kranz d.c. plasmajet as a source of atomic vapors
(ref. 16).  No interelement effects were found with 50 ppm of Al,
Ba, Fe, Li, K, Mn, and Zn, and no difference in the_absorption
signal was found in the presence of Cl~, SO^, or NOs ions.

     The Kranz d.c. plasmajet is a gas stabilized d.c. arc burning
between thoriated tungsten electrodes.  Nitrogen was used as the
carrier gas during the boron analysis.
                                336

-------
     The technique developed by Loften, et al. (ref. 17) to moni-
tor Pb as alkyllead compound should be considered as an approach
to continuous monitoring of other metal elements.  The method is
based on the reduction of lead in air by carbon monoxide formed
from oxygen in the air specimen as it passes through a hot bed of
carbon.  Elemental lead produced in this manner is introduced
into the long optical path of a nonflame atomic absorption spec-
trophotometer.  With other elements, data on efficiencies of the
reducing reaction and on the capability to maintain the elemental
ground state atoms in the vapor state must be obtained.

     Other possibilities include using a high-frequency (5 MHz)
induction plasma in a reducing mode (ref. 18, 19).  Little data
are available but high sensitivities are attainable for elements
in powdered samples which were introduced directly.

     Conventional atomic absorption spectrophotometry uses a mono-
chromator system to isolate the characteristic radiation.  However,
increased stability of the atomic absorption spectrophotometric
analyses for Ni can be attained by using a resonance lamp as a
monochromator (ref. 20).  The wavelength setting is stable under
the most rigorous conditions and, with provision for automatic
setting of full-scale deflection, the system requires little, if
any, adjustment.  Changes in ambient temperature or pressure, or
small mechanical vibrations have no effect.  The resonance re-
emission from Ni vapor provides a means for isolating the resonance
lines at 3^15, 3515, and 3524 it; no data are reported for the more
sensitive line at 2320 A*.  For routine determinations of Ni in
the range of 0.05-2?, the accuracy for single determinations was
±5.1? of the mean at the 955? confidence level (ref. 20).

     Also, improved sensitivity and good resolution can be ob-
tained by atomic absorption or fluorescence spectrophotometric
techniques without monochromators by using selective modulation
or resonance detectors for isolating resonance lines and by cou-
pling a solar-blind photomultiplier to the system.  The applica-
tion of solar-blind photomultipliers to measuring nickel has been
demonstrated (ref. 21).  No specific data are reported.

     A continuous measuring technique for nickel carbonyl (speci-
fic if no other reducing agents are present) was developed by
Vol'berg (ref. 22) based on the atomic absorption phenomenon but
using a low cost mercury vapor photometer.  The method determines
microquantities of nickel carbonyl in air by measuring the amount
of mercury liberated by the reaction of the nickel carbonyl with
red HgO @ 200°C.  The elemental mercury is measured by ultraviolet
absorption techniques.

     The following summaries on atomic absorption techniques for
B, Ba, Cr, Cu, Mn, Ni and V are presented to show the detection
limits and potential problems with spectral and chemical inter-
ferences.
                                337

-------
     Atomic absorption spectrometric analyses of acid extracts
of particulate collected on glass fiber filters and membrane fil-
ters have been used extensively for measuring particulate from
ambient air (ref. 23).  The sampling times and sample preparation
times are long, but relatively low detection limits are attainable.

     Care must be taken to prevent loss or contamination of the
sample.  All reagents must be distilled to remove metal contami-
nants and all glassware must be carefully cleaned (soaking in
20% nitric acid for 2 to 6 hours).

     The solids content must be maintained at <0.5%.  Also, silica
extracted from glass fiber filters and from the particulate sample
must be eliminated from the solutions (settling for 12 to 24 hours
followed by centrifugation) or significant interferences with Fe,
Ca, Mn, and Zn occur.

     Glass fiber filters yield background contamination of elements
such as Fe, Ba, Zn and others..  Membrane filters must be used to
measure these elements.

     Based on an integrated 2000 m3 air sample, the following de-
tection limits were obtained:  Ba (0.02 yg/m3), Cr (0.002 yg/m3),
Cu (0.001 yg/m3), Mn (0.001 pg/m3), Ni (0.004  g/m ), and V
(0.01 yg/m3).

     Particulate, collected from ambient air by filtering air
through a Millipore membrane for 8 hours, was analyzed for Cr,
Ni, Mn, and Cu by an atomic absorption technique (ref. 24).  The
acetate based membrane is destroyed with acetone and, after evapo-
rating the solvent, the residue is ashed in a muffle furnace at
550°C for one hour.  The ash is extracted with hot nitric acid
to prepare a solution specimen for atomic absorption analysis.

     A lean air-acetylene flame was found satisfactory for the
determinations of Cr, Ni, Mn, and Cu, although a reduction in
sensitivity was observed with Cr.

     Lean-flame conditions gave much less interelement effect in
comparison with a fuel-rich flame.  However, significant inter-
element effects are reported with the addition of one of the four
elements to the one being measured.  Anionic interferences were
also noted for HC1CK, H2SO.,, H3P04, HC1, and HN03 additions.

     Glass fiber filters were found to be contaminated with sig-
nificant amounts (6-20 yg/fliter) of Cr, Ni, Mn, and Cu, plus Ba,
Pb, Zn, Fe, Ca, etc., which not only contribute to the level of
analyte but also may cause interelement effects.  Contamination
of 0.1 to 0.5 yg of each of the four elements - Cr, Ni, Mn, and
Cu - was found in the organic membrane filters.
                               338

-------
     To minimize the interelement effects, all analyses were per-
formed by the standard addition method.

     Ambient air particulate samples collected on glass fiber
filters, ashed with a low temperature asher, and leached with
nitric acid have been analyzed by atomic absorption spectropho-
tometry for vanadium, nickel, chromium, manganese, copper and
other elements (ref. 25).  The total analytical procedure is time
consuming (sampling - 7 days @ 20 ft3/min; ashing - 2-14 hr;
extraction - approx. 3 hrs; atomic absorption measurement -
approx. 10 min.).  The following detection limits and average
precision are reported:  V - 0.091* ug/m3 (215K): Ni - 0.006 ug/m3
(1358); Cr - 0.003 ug/m3 (22%}; Mn - 0.002 yg/ms (105&); Cu -
0.003 yg/m3 (95?).

     The determination of Mn in aqueous solutions by atomic absorp-
tion spectrophotometry is improved by deaeration of the diluting
water (ref. 26).  Some compound formation apparently occurs with
dissolved oxygen.

     Iron interferences in the atomic absorption flame photometric
determinations of Co, Cr, Cu, Mn and Ni were studied by Ramirez-
Munoz and Roth (ref. 27).  With air-acetylene flames, increasing
amounts of Fe decreased sensitivity for all elements.  Fe inter-
ferences were greater with Cr, and less with Mn.  The possible
formation of Fe-Cr compounds is considered responsible for some
signal decrease.  (Note - A hotter flame, e.g., N20-C2H2, may
minimize the effect.)

     The reaction of a metal oxide species with an overexcited
molecule of a C species, e.g. C2 or CH, in a fuel-rich air-C2H2
flame is a possible cause of the decrease observed in the atomic
absorption of Fe, Cr, Co, Cu, and Ni in flames having a C2H^-air
ratio greater than a critical value which is characteristic for
each metal.  The absorption decrease was observed with SO*  and
NOa solutions of the metals, but not in Cl~ media (ref. 28).

     The determination of barium is best performed with a nitrous
oxide-acetylene flame even though the flame temperature is suffi-
cient to cause a relatively high degree of ionization and attendant
loss of ground state atoms necessary for the atomic absorption
process (ref. 29).  The nitrous oxide-acetylene flame minimizes
compound formation encountered in other flame systems.  However,
a problem due to high emissivity of the nitrous oxide-acetylene
flame at the analytical wavelength for barium (5536 X) occurs.
The cause is inadequate emission from the hollow cathode lamp
(ref. 30).

     Boron has been determined by atomic absorption spectropho-
tometry (2^98 A") with a nitrous oxide-acetylene flame.  A detection
limit of 6 yg/ml has been attained (ref. 31).
                                339

-------
     When using a graphite tube furnace (P.E.  HGA-70) in nonflame
atomic absorption spectrophotometry,  an intense C2 emission line
0.2 X away from the Ba resonance line can cause significant
interference.

     Marshall and Schrenk (ref. 32) performed  detailed evaluation
of various flames and matrices for the analysis for vanadium
(3185.4 A") by atomic absorption spectrophotometry.  Extensive
interference from Al, Bi, Cr, Fe, K,  Li, Sb, Ti, Zn, acetate and
phosphate were encountered in an oxyacetylene  flame.  Several
explanations are given including formation of  nonvolatile vanadates
or intermetallic compounds.   A detection limit of 15 yg V/ml is
reported.

     A number of elements interfere with the atomic absorption
measurement of V at 3185.4 S.  Chelation techniques and solvent
extraction prevent interferences from all elements except Co11,
Cu, Se and WVI (ref. 33).  The best results are obtained with a
nitrous oxide-acetylene flame.

     Atomic absorption and flame emission spectroscopic determina-
tions of Ba in aqueous solution were  compared  by computer techniques
(ref. 3*O«  In the absence of interferences, atomic absorption
with air-acetylene flame is  the preferred method for 10-500 yg
Ba/ml.  In the presence of interferences or at lower concentra-
tions (1-100 yg/ml), atomic  absorption with the N20-C2H2 flame
is the most convenient method.  For <1.0 yg Ba/ml, flame emission
with N20-C2H2 flame is preferred and  gives a detection limit of
0.0035 yg Ba/ml, as compared to 0.03  and 0.72  yg Ba/ml for atomic
absorption with the N20-C2H2 and air-C2H2 flames, respectively.

     A large number of elements are known to interfere with con-
ventional flame atomic absorption spectrophotometric analyses for
chromium.  A summary of these interferences and methods for over-
coming them is presented by  Hurlbut and Chriswell (ref. 35).  The
elements (Mg, Ba, Ti, Mo, W, Mn, Fe,  Co, Ni, Cu, Ag, Cd, Hg, Al,
Bi, and Ce) and the compounds (HsPOi,, H2SCU and NfUOH) interfere
with the absorption by Cr at 3579 &.   The most effective surpres-
sor of interference was Na2SOi» .

     Sachdev, Robinson, and  West (ref. 36) developed an atomic
absorption technique for measuring Mn and Ni in air particulate
collected on filter paper.  After ashing, the  residue is dissolved
in hydrochloric acid and measurements are made with atomic absorp-
tion spectrophotometric equipment modified with a quartz T-piece
adapter to provide a 15 cm optical path.  With the adapter, 10-
to 15-fold increase in sensitivities  was obtained.  Sensitivities
of 0.01 mg/1 (Mn) and 0.02 mg/1 (Ni)  for 1% absorption are ob-
served.  Interference effects were studied, but only a slight
suppression on the absorption of manganese was observed with
titanium, vanadium, and zirconium.  These interferences were
removed by complexing the ions with fluoride.
                                3^0

-------
     Some, but not all, cationic interferences in the atomic ab-
sorption spectrophotometry of chromium can be eliminated by using
a nitrous oxide-acetylene flame (ref. 37).  With an air-C2H2
flame, the atomic absorption of Cr (357-9 nm), as Cr(III) and
Cr(VI) solutions, was enhanced by Cu > Ba > Al > Mg _> Ca while
Na, K, Sr, Zn and Sn depressed the Cr absorption.  F¥ depressed
the absorption of Cr(III), but enhanced that of Cr(VI); Mn and
Ni depressed Cr(III) absorption, but gave no marked enhancement
for Cr(VI).  By using the N20-C2H2 flame, Na, K, Zn, and Pe inter-
ferences were eliminated.

3.2   Atomic Fluorescence Spectrophotometry

     Better sensitivities for some, but not all, elements are
reported for atomic fluorescence techniques when compared to
atomic absorption spectrophotometric analyses of the same elements
The technique is not applicable to determinations of all metal
elements and has limited value to special cases.

     Because of the interatomic quenching effect of fluorescence
by nitrogen, direct introduction of air or stack gas emission
into the flame is not practical.

     The following selected summaries show the effects of instru-
mental parameters in the application of trace metals in aqueous
media and show the development of a simultaneous multielement
analyzer.

     A detailed study of atomic fluorescence spectrophotometry
was performed as applied to the determinations of iron, cobalt,
and nickel (ref. 38).  The effects of source parameters, optical
arrangement, uptake of sample solution, solvent, inorganic acids
and 13 other elements were investigated.  For aqueous solutions,
a detection limit of 0.003 ppm Ni was reported; by using an ex-
traction technique 0.0001 ppm Ni can be detected.

     An instrument was designed for simultaneous multi-element
analysis using sequentially excited atomic fluorescence radiation.
The instrument was used for the simultaneous determination of Ag,
Cu, Fe and Mg in aqueous solutions and provided detection limits
of 0.002 ppm (Ag), 0.003 ppm (Cu), 0.1 ppm (Fe), and 0.001 ppm
(Mg) (ref. 39).

3.3   Emission Spectroscopy

     Although emission spectroscopic techniques have been used
extensively to measure the metal content of particulate collected
by the National Air Sampling Network (ref. 40,41), no data are
available for continuous, flow-through, emission spectroscopic
studies on airborne or stack emissions of B, Ba, Cr, Cu, Mn, Ni,
and V..  However, modification of techniques reviewed in other
documents in this series, (a) Beryllium and Cadmium, (b) Mercury,
                               341

-------
and (c) Lead, and reported elsewhere (ref. 42,43,44,45,46) should
be considered.

     The major limitations of emission spectroscopic techniques
are the chemical interferences - compound formation in the exci-
tation zone - and spectral interferences.  Compound formation can
be minimized by using a sufficiently energetic source, e.g.,
radio-frequency— or microwave—induced plasma.  Spectral interfer-
ences can be eliminated by selection of proper analytical lines
and by using a monochromator which provides sufficient resolution.

     Data relating to flame-, arc- or spark-excitation of optical
emission for B, Ba, Cr, Cu, Mn, Ni, and V are reported in the
following sections to aid in understanding potential interferences
and problems that can be encountered.

     Flame photometric analyses of boron in aqueous solutions have
been performed based on extraction into an organic solvent and
measurement of emission at 546 nm in an oxygen-hydrogen flame
(ref. 4?).  Extraction was complete for <0.4 mg B and 96% B was
recovered from solutions containing 0.4-1.0 mg B.  Presence of
Mn, Ti, Mo, and W caused spectral interference.  Fluoride in the
aqueous solution reacted with B to give EF^" which was not extract-
able .

     Ba has been measured with emission flame photometry (553-6 nm)
in a propane-butane-air flame.  Interferences from Ca, Cu, Cd, Sr,
and Mg are reported (ref. 48).

     A number of analytical methods for measuring barium in sili-
cates were evaluated with regard to speed, reliability, and sys-
tematic errors (ref. 49).  These methods included isotope dilution
mass spectrometry, flame emission and absorption spectrometry,
x-ray fluorescence spectrography, neutron activation analysis,
and classical gravimetric methods.  Emission spectrometry, with
a spark-solution technique and automated equipment produced rapid
and accurate results.  Other techniques are reported as not satis-
factory as practical routine methods for barium in silicate rocks.
In most cases, the other techniques were subject to interference
and lengthy separations were required (ref. 49).  Routine deter-
mination of barium by direct dilution x-ray methods is satisfac-
tory if adequate standards are available and if precautions are
observed (ref. 49).

     By using a nitrous oxide-acetylene flame, Fassel, et al.
(ref. 50) report the "interference-free" determinations of Cr,
Cu, Mn, Ni, and V in solutions of low and high alloy steels.
Lowest determinable concentrations of 0.0025? (Cr), 0.01% (Cu),
0.025? (Mn), 0.00552 (Ni), and 0.01% (V) are reported.

     A multichannel flame emission spectrometer (ref. 51,52,53)
was developed for the simultaneous determination of magnesium
                                342

-------
(2852 X), calcium (£22? X), copper (3242 X), manganese (4031 X),
and chromium (4254 A).  Detection limits are 1.0 ug/ml or less
for each element in aqueous solutions.  The repeatability for
each element is excellent, the coefficient of variation ranging
from 0.5 for Cr to 3-2% for Cu.  The accuracy (comparison of mean
value determined and the expected value calculated from the dilu-
tion of each original, certified sample) varied - Cr (82?),
Cu (90-93?), Mn (92-94?).

     The emissivity of chromium increases with a corresponding
increase in HC1.  Mutual interference exists between Ca and Cr;
the intensity of the Cr emission is lower in presence of Ca.  A
concentration of 300 yg K/ml will affect Mn emission intensity.
Phosphate ion decreases the intensity of the emission at 4031 X
for Mn.  No interference was observed for nitrate and sulfate
ions.  By incorporating selected additives into the solutions,
interferences can be overcome; the most efficient additive proved
to be a solvent system of half-saturated 8-hydroxyquinoline and
0.4M perchloric acid in 1M hydrochloric acid (ref. 51,52,53).
     By shielding the secondary reaction zone in an
flame with a stream of nitrogen flowing parallel to the flame to
prevent access to atmospheric oxygen to its base, higher sensi-
tivities and less interference from bonded molecular emissions
are observed with flame emission spectrophotometry (ref. 54).
The technique permitted measurement of Ba, Cr, Cu, Mn, Ni, and V
by flame emission spectroscopy .  The detection limit for all but
V was improved by a factor of 10, and V, which was previously
undetectable, could be detected to 10 ppm.

     A critique on interferences by SiO (molecular band spectra)
and Pe (2497.82 X) on the B 2497.73 & emission line is presented
by Moore (ref. 55).  With silicate-bearing rocks, interference
from Pe (2497.82 A) was judged to be of no consequence at least
to 10? FeO.  A correction method for SiO interference was developed
based on extrapolating a baseline from a line connecting tops of
SiO band peaks on the longer wavelength side of B + SiO @ 2497.73 X,
For concentrations of boron in the range from 1-520 ppm, a coeffi-
cient of variation of 17? was obtained.

     By using an argon atmosphere instead of air for determining
B in a quartz matrix in a d.c. arc discharge, the interference
due to SiO band spectra (2414-2925 X) is eliminated when B
2497.7 X is used as an analytical line.  A detection limit
^0.5 ppm of B in quartz can be determined (ref. 56).

     The suitability of measuring B in steel with the B lines
2496.8 X, 2088.9 X, 2089.6 X, 1826.4 X and 1825-9 X with various
conditions of excitation under argon was determined with instru-
ments of 4.6 or 7 X per mm dispersion.  The 2089.6 X and 1825-9 X
lines ,' although subject to considerable interference by Mo and Cu,
respectively, are the only suitable ones (ref. 57).
                                343

-------
     Manganese, chromium, and vanadium were detected in atmospheric
dust particles with emission spectroscopy by burning the particu-
late and the filter paper used to collect the specimen, in flat
graphite electrodes after adding graphite powder.   By carrying out
the excitation in a magnetic field, the relative standard devia-
tions were decreased from <15% to 
-------
     Fundamental studies using a plasma jet generator (A with
nozzle-shaped Cu anode and W rod cathode) in emission spectro-
scopic determination of Ni, Mn, and Cr in aqueous solution show
less interelement effects than with an a.c. arc discharge method.
However, when a sample is in water, the background due to H20
can be so large that the detection limit is not always higher
than that by a.c. arc excitation (ref. 65).  A simultaneous
determination of Ni, Mn, and Cr was developed which required no
internal standard and had detection limits lower than conventional
techniques (ref. 65,66).

     By measuring the emission of Cr radiation @ 4254.346 X, from
a helium plasma generated with an RP induction furnace (3 Me) a
detection limit of 10 9 g of Cr has been attained (ref. 6?).  The
system uses a graphite crucible as a receptacle for either liquid
or solid samples and can be applied to a wide range of elements.
With evaporated solution samples, the precision of the measure-
ment (peak height) is 1-4% depending on the element.  A precision
of 4-10% was attained with powdered samples.

     Chromium was measured with a high frequency (40 mHz) induc-
tion plasmatron and emission spectroscopy.  Analytical lines and
detection limits in an argon plasma are:  Cr - 3578.6Q A* (1 pg/ml);
3886.80 8 (125 yg/ml); 3919.16 X (60 yg/ml);  4254.33 A (2 yg/ml)
(ref. 68).

     SpectraMetrics Inc. (Burlington, Mass.)  has a multi-element
spectrometer (SpectraSpan) using an argon plasma (10,000°K)
based on an electrode system.  The spectrometer can be used in
an emission or atomic absorption mode.  High spectral resolution
is obtained with an echelle grating and a series of polychromator/
encoding cassettes.  By mixing air or stack gas samples with
argon, a continuous measurement of up to 10 elements is possible.
(Note - A 401 Process spectrometer is available.)  High sensi-
tivities are obtained, but no specific data on direct monitoring
of metal emissions are reported.

3.4  X-ray Emission Spectroscopy

     A number of excitation sources ranging from a Van de Graaff
generator to the more common x-ray tube (tungsten target) and
radio-isotope sources have been used to measure metal elements
in airborne particulate collected on filters.  Although high
sensitivities and attendant low detection limits result with the
Van de Graaff system, devices using the latter excitation sources
are more practical for field applications.  In all cases, the
techniques require the collecting of particulate for measurement.

     The major problems associated with the technique involve
matrix effects - loss or enhancement of the intensity of the
analytical line due to interelement interactions - particle size
effects, spectral interferences, and weight, size and power
                                345

-------
factors.  With thin films of particulate, many of these problems
can be minimized.  Spectral interferences have been reduced by
the development Si(Li) semiconductor detectors.  Instrument porta-
bility has been attained with radioisotope excitation sources.

     Cares (ref. 69) considered some of the sampling handling prob-
lems encountered in measuring airborne metallic dusts and fumes
by conventional x-ray spectrometry.   Severe interferences can be
encountered from contaminants present in the filter media.  Ash-
less and membrane filters are essentially free from contamination
except for traces of iron, potassium, and calcium.  Glass fiber
filters have relatively high iron and zinc contamination, and
contain sufficient calcium, potassium, and barium to create severe
interference at or near the principal lines of these elements.

     Additional spectral interferences (ref. 69) are also observed
for systems using tungsten x-ray target material.  Tungsten and
copper reflections are observed with varying intensity, depending
on the age of the tube and surface characteristics of the sample.
With the tungsten target x-ray tube, it is difficult to determine
low concentrations of Hg, As, or Se.

     Intensity reduction or enhancement result from matrix, sample
thickness, particle size, and interelement effects (ref. 69).
When using analytical x-ray lines of low energies (long wave
lengths), considerable counting losses can result by penetration
or burial of particles between fibers of some types of filter
paper.  Counting losses due to burial were highest in Whatman 4l
filter paper and lower, but variable, in glass fiber filters.
Absorption losses due to sample matrix are also influenced by the
wavelength of the analytical line.  Net counts for Cd Lg (3.74 X)
are reduced about 40$ from values obtained with Cd K^ (0.536 S)
in a matrix containing an equal amount of iron or silver.  Also,
although the Cd Lg line is more sensitive than the K_, a critical
sample thickness ror Cd Lg occurs at approximately 400 yg/cm .

     The principal analytical lines  (x-ray) of selected metals of
industrial hygiene significance are:  Cr (Ka - 2.29 &), Cu (Ka -
1.54 £), Mn (KQ - 2.10 ft), Ni (KQ -  1.66 fl ), and V (g  - 2.50 X).

     Problems related to background  signals from glass fiber
filters used to collect airborne particulate were observed by
Leroux and Mahmud (ref. 70).  Only lead could be measured without
interference in particulate collected on glass fiber filters.
Ba, Ni, Cu and other lements would all be subject to interfering
background signals emitted from the  glass fiber filters.  Organic
membrane filters have considerably less contaminants; data for
Mn collected on organic filters are  reported.  The authors also
conclude that the average air pollution samples weighing less
than 1.5 mg/cm2 can be analyzed without matrix correction for
all it's elements higher than manganese in the Periodic Table.
                                3^6

-------
     Hirt, et al.  (ref. 71) developed technology for collecting
particulate, for performing x-ray emission spectrographic analy-
ses, and for preparing calibration standards to detect and measure
quantitatively heavy metal elements in airborne particulate.  One
to 100 micrograms  of Co, Cr, Fe, Pb, Hg, Ni, Ft, V, and Zn were
detected and measured.  Particulate was collected on glass fiber
filters and analyzed with a Norelco x-ray spectrograph equipped
with NaCl crystal  and a helium flooded sample chamber.

     For deposits  less than 200 micrograms on a 2.4 cm filter
disk, no reabsorption of secondary x-rays and no lack of penetra-
tion of the primary x-rays occurred.  Calibration curves were
derived from standards prepared by evaporating different volumes
of solutions of known concentrations onto an aluminum foil or
filter disk made nonabsorbent with acrylic resin impregnant.

     Disadvantages of the x-ray emission (fluorescence) method
are complications  due to inter-element effects and difficulty in
preparing proper standards.

     Mathematical  equations which correct for the interference
(absorption and enhancement effects) of CrKa with MnKg line have
been developed (ref. 72).

     Numerous theoretical, mathematical, and empirical techniques
have been devised  to correct for interelement (matrix) effects.
However, a number  of deviations from the predicted intensities
occur.  Mitchell and Kellam (ref. 73) reported several unusual
matrix effects when determining combinations of 4th-period ele-
ments such as Cr,  Fe, Co and Ni.  Chromium has an absorbing effect
on NiKa and CuKa among others, whereas iron will severely absorb
NiKa, and less severely absorbs CuKa.

     Opposite and  coexistent with the absorption effect is the
enhancement of CrKa by the absorption of the iron radiation and
to a lesser degree by the presence of Co, Ni and Cu.

     Interrelated  effects do occur.  The matrix effect of iron
or nickel on CrKa  in a light matrix is an enhancement at low
levels of iron and nickel, but, as the iron or nickel reaches
a critical level,  the CrKa intensity decreases.

     Graham and Bray (ref. 7*0 observed a linear relationship
between excited x-ray intensity and element concentration for
film thicknesses up to about 0.1 mg/cm2 and concluded that in
this thickness range absorption effects have no observable in-
fluence on x-ray intensity.  Based on these observations, metal
film standards were prepared.

     A correction  factor scheme was developed to minimize inter-
elemen-t absorption effects for x-ray analyses for vanadium,
zirconium, tungsten, iron, titanium, tantalum and niobium (ref. 75)
                                3^7

-------
     A general x-ray spectrographic solution method which pro-
vided a single calibration system per element for all materials
and acid matrices was used with a x-ray Industrial Quantometer
(ref. 76).

     Data for an automatic x-ray spectrograph, Autrometer, show a
standard deviation of 0.03$ for Ni in a 10% nickel alloy (ref. 77)
Precision was poorer for trace elements because of the statisti-
cal lack of counts.  Average accuracy for W, Mn, Ni, Cr, Co, Fe,
Mo, and V determinations, between 2 and 23? concentration was
0.77/& in comparison with wet analyses.

     During x-ray fluorescence analyses of trace elements in sea
water, Morris (ref. 78) observed several spectral interferences.
Conventional x-ray fluorescence equipment - tungsten target x-ray
tube, lithium fluoride crystal, O.l6-mm collimator, vacuum sample
chamber, and gas-flow proportional counter with pulse height
selection - was used.  Unresolved first order KQl and Ka2 lines
for V, Cr, Mn, Fe, Co, Ni, Cu, and Zn were used after separating
and concentrating these elements from sea water.

     The intense first-order tungsten L , line at 42.99° and L ~
line at 43.44° emitted by the x-ray tube swamped the K , 2 emis-
sion of 5 ppm amounts of copper and nickel at 44.99° and3 48.64°,
respectively.  These interferences could be corrected by using
a titanium filter, but some reduction of the analytical line in-
tensity also occurs.  The x-ray fluorescent determination of the
transition elements in presence of one another is complicated by
the proximity of the first order Kg,  emissions to the first order
K  lines of the next element in order of increasing atomic number.
The analytical system could not resolve the Kg, line of vanadium
at 69-12° from the Ka lines of chromium at 69-29° and 69-43°.
In addition, the Kg, lines of chromium and manganese are only
partially resolved from the K  lines  of manganese and iron,
respectively.  The KR1 emission of Fe may interfere with the K , ?
emission of cobalt. p                                           '

     Spectral interferences were minimized by using calibration
curves adjusted for each elemental interference.  Detection limits
of 0.4 ug or better are claimed for 600 sec. counting times.

     Boron cannot be measured by conventional x-ray emission tech-
niques.  However, by using proton excitation, Kamada et al. (ref.
79) were able to detect down to 0.00145? B in nickel base alloy.
A 200 kV proton source and a flat crystal (lead stearate) were
used.  Considerable background interferences due to low-angle
specular reflection from the lead stearate and due to emission
lines of base material were observed.

     Because of low energy, ultrasoft boron x-rays are easily
absorbed in x-ray analytical systems  and, unless special analyzer
crystals are used, the ultrasoft radiation cannot be resolved.

                               348

-------
Film analyzers were prepared from Ba and Pb salts of stearic,
behenic, and lignoceric acids, which resolved B K  lines and
allowed the microanalysis of B with high sensitivity (ref. 80).
(Note - No specific data are reported.)

     An improved method for the determination of barium was
developed based on using chromium x-ray tube, rather than tungsten,
and BaLal, rather than BaK , (ref. 81).  The major problem with
using W x-radiation and the BaK , analytical line is the low
peak-to-background ratio resulting in large part from scattered
tungsten radiation.  By using chromium radiation and the BaL ,
line, the net intensity is approximately three times larger.0
A detection limit of 0.002? Ba was achieved.

     In geologic samples, a detection limit of 10 ppm for barium
was attained if corrections for a titanium spectral interference
can be made (ref. 82).  The first-order titanium K , line (2.7*132 X
interferes with the barium first-order L , analytical line
(2.7696 &).  Note - The barium Lg, line, which has the same inten-
sity as the L , line and no titanium interference, can be used if
cerium is absent.  CeLai ls a spectral interference for BaLg,.

     Gunn (ref. 83) and Willis, et al. (ref. 84) compared several
alternative methods for determing Ba by x-ray fluorescence tech-
niques.  The measurements can be performed by using BaKa or L
x-ray lines.  To excite BaKa emission, a higher primary x-ray
beam energy is necessary than is available on most spectrometers.
Alternatively, the much weaker BaL lines can be used with a Cr
primary radiation.

     Of the available L lines of Ba the L , is overlapped by the
TiK  line.  The Lg, line is overlapped by  CeLa line; Lg2 is free
of interference, but is very weak.  The L , line is stronger when
using a Cr primary tube at 1800 watts; however, the background
is strongly curved requiring a curvature correction.

     By using 100 KV to 150 KV excitation of W Ka line in the
primary tube, analyses based on the BaKa line can be performed.
The standard LiF monochromator crystal is not adequate to dif-
fract the BaKa line; however, sufficient resolution is obtained
with LiF crystal cut to the 220 reflection planes.  Using this
crystal, the BaKa line may be resolved from the iodine absorption
edge.  The iodine interference would occur if a Nal scintillation
detector is used.  Gunn (ref. 83) reports that, by using the BaKa
lines rather than L lines, considerable improvement is obtained.
Total analysis time per determination using BaKa line is three
minutes with a replication error of less than 2%.

     The high energy of the BaKa  radiation results in greater
penetration of samples than normally encountered with other
elements.  As a consequence, the analyst must be aware of a

-------
potential problem with sample thickness.   Excellent replication
was found in samples less than infinitely thick for the radiation
being used.

     Rhodes (ref. 85,86) reviewed the development of x-ray excita-
tion with radioisotope sources and nondispersive energy detection.
The relationships of interferences arising from absorption,
enhancement, and particle size effects were also discussed.  The
application of radioisotope sources provides portability and less
background scatter of target radiation which is characteristic of
the conventional x-ray tube excitation techniques.

     A review of energy dispersion techniques, methods of calcu-
lation, and interpretation of data from inhomogeneous samples
was prepared by Birks (ref. 87).

     Rhodes, et al.  (ref. 88,89) applied  x-ray fluorescence analy-
sis using radioisotope sources for excitation, a Si(Li) detector
for energy dispersive spectrometry, and a minicomputer for data
reduction and automatic instrument control to determining "heavy
element" constituents of airborne particulate collected on filter
paper.  The technique was used to measure 17 elements including
V, Cr, Mn, Ni, and Cu.  Based on collecting particulate from an
average air volume of i860 m3 over 24-hour periods, the following
detection limits were attained:  V - 0.007 yg/m3, Cr - 0.053 yg/m3,
Mn - 0.027 yg/m3, Ni - 0.013 yg/m3, and Cu - 0.011 yg/m3.

     Matrix effects  - absorption or enhancement by other elements
in the sample - are  reduced or eliminated by using thin specimens,
^370 yg/cm .  Also,  by using thin specimens, the ratio of fluores-
cent intensity for a given element to the intensity of backscattered
(background) radiation is much larger than for infinitely thick
specimens.

     To obtain the best excitation efficiencies, each sample was
irradiated with each of three annular radioisotope source assem-
blies.  V, Ca, and Ti were excited by 55Fe.  Cr, Mn, Fe, Co, Ni,
Cu, and Zn were irradiated with 238Pu.  Hg, Pb, As, Br, Sr, Zr,
and Mo were excited  by 109Cd.

     The automated x-ray fluorescence measurements required
approximately 0.2 hour for operator and 0.45 hour for equipment
operation.  Good precision, accuracy (data compared with atomic
absorption spectrophotometric analyses) and speed are attainable.
A major time-consuming step in the analytical procedure is the
weighing of the cellulose filter before and after collecting the
sample.  Also, due to the hygroscopic nature of the cellulose
filters, they must be stored for 24 hours in a constant humidity
box before each weighing and must be weighed a fixed time
(7 minutes) after removal from the box.  Additional operator time
is required to cut out a 47-mm diameter disc from the 8" x 10"
Whatman filter paper.
                               350

-------
     Giauque (ref. 90) developed a technique for experimentally
substracting iron interference of Mn and Co observed in non-
destructive x-ray fluorescent analyses of glasses and ceramics.
By using an analyzer with a radioisotope source-target assembly
(As K x-rays excited from an As203 target by an 2>tlAm source)
and a Si(Li) semiconductor detector, detection limits 4 100 ppm
Mn are attained.  The interfering Fe peak was subtracted by ex-
citing a pure piece of Pe metal with the analyzer in a subtract
mode.

     Frankel and Aitken (ref. 91) reviewed recent developments in
energy-dispersive x-ray emission spectroscopy with particular em-
phasis on the solid state Si(Li) semiconductor detectors.  The
improved resolution attainable with the Si(Li) provides the means
to distinguish between adjacent elements, e.g., Cr, Mn, Fe, Ni,
Cu, Zn.  Resolution attainable by crystal spectrometers equipped
with scintillation counters, proportional counters, and ionization
counters is not sufficient to provide comparable separation.

     A portable fluorescent x-ray instrument (approximately 10 Ibs
and 0.5-1 ft3) utilizing bremsstrahlung radiation from (a) tritium
absorbed in zirconium or (b) promethium-l^iy was developed by
Karttunen et al. (ref. 92,93) to measure elements in the range
Z = 16 to 35.  Analyses for macro amounts of Cr, Ni, V, Mn, and
Cu are reported for alloys and cement.  Although the resolution
was adequate for the samples, the newer semiconductor detectors
would be required to detect lower concentrations.

     Yamamoto (ref.  9*1) applied an x-ray spectrometer with a
Si(Li) semiconductor detector and an 12^I radioisotope source to
the analysis of low-weight nickel samples.  A detector limit for
Ni of 1 yg is reported.  An excellent discussion of the instrument
parameters that affect the intensity of the analytical emission
line is presented.

     A high analytical efficiency, which is independent of the
composition and density of the sample and of the measurement
geometry, can be attained by using two monochromatic radioisotope
sources in sequence (ref.  95).  The energies of the two sources
must be such that the energy of the other is slightly higher than
that needed to excite the K-radiation from the element to be
determined.  I*9V and S5Fe were used to determine V.

     Si(Li) semiconductor detectors can provide the resolution
necessary to separate adjacent atomic numbered elements, e.g.,
Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn.  However, the additional cost
of the device, and the maintenance of operation (liquid nitrogen
coolant) may limit the application as a stack monitoring device.
An alternate method of attaining suitable resolution is based
on spectrum analysis by means of filters and radioisotope excita-
tion sources (ref. 96).
                                351

-------
     The technique combines the simplicity of nondispersive
systems with the resolution of complex dispersive equipment.
Fluorescent radiation from a specimen, excited by means of a
radioactive source, tritium/Zr, is measured after transmission
through each member of a sequence of filters consisting of ele-
ments of consecutive atomic number.  This provides a scan of the
radiation by means of the absorption edges of the filters.

     Disadvantages owing to interference in certain cases can be
overcome by the use of suitable standards.  Advantages of this
nondestructive method are that it avoids complex instrumentation,
elements of adjacent atomic number can be resolved, and no inter-
ference effects owing to target material can occur.  Moreover,
source stability is accurately assessable and corrections for
lost counts are free from form factor calculations.  In addition,
the method is exempt from the effects associated with higher dif-
fracted orders.  The principle is applicable to solids and liquids,
metals, nonmetals, conductors, nonconductors, or radioactive
specimens.  It has been applied to determine Mn in aqueous solu-
tions over a concentration range of 0.7-21.0$ Mn.  A detection
limit of 0.7 ppm Mn is estimated.

     Rhodes, et al. (ref. 97) described a system using radioiso-
tope x-ray fluorescence techniques and differential thin foil
filters to isolate the characteristic radiation of Zn and Cu.
The excitation source was 109Cd and the ZnK x-rays were selected
by Cu/Ni filters, whereas the CuK x-rays were isolated with Fe/Ni
filters (Co/Ni could also be used).

     A high-efficiency x-ray spectroanalyzer is described which
is equipped with a llf'Pm 3-ray source and a Xe-CHi,-filled pro-
portional counter (ref. 98).  Unwanted scattered g-rays are
eliminated with a Be foil filter.  K x-rays for the elements
from Cr to Zn are used for analysis.

     The intensity of the Ka line of Cu, as excited by a 3H/Zr
source, was measured by means of a proportional counter.  Fe
absorbs the measured x-rays and so lowers the results.  The
interference by Fe can be corrected after measuring the inten-
sity of the Ka line of Fe or by using nomograms.  In the absence
of interfering elements and at the accuracy of the x-ray intensity
measurement of 5-10$, the Cu content of 0.15-1$ can be determined
with an absolute error of 0.05-0.1$ Cu (ref. 99).

     A radioisotope x-ray fluorescence and preferential absorp-
tion method for the analysis of copper in mineral products was
described by Ellis, et al. (ref. 100).  For the copper-iron-
lead-zinc minerals, considerable difficulty was encountered with
the Cu analyses.  The methods for compensating for changes in
matrix composition were not suitable.  The presence of zinc and
lead complicated the correction for absorption effects.  X-rays
of energy under the zinc K absorption edge minimized the effect
                               352

-------
of zinc and lead.  Changes in the iron content produced varia-
tions in the absorption of x-rays by iron and resulted in marked
changes in Cu K x-rays.  This effect was minimized by using a
detector filter of suitable thickness.  The error in the Cu
determination was further reduced (halved) by compensation using
an analysis for zinc.  A Pu-238/Ga gamma-ray excited x-ray source
and a 23 mg/cm2 Cu filter was used.

     A nondispersive two-channel x-ray fluorescence apparatus
with two filters and two proportional counter detectors was de-
veloped by means of which the ratio of intensities of characte -
istic emission in two spectral regions isolated by the channels
can be measured and the differences in intensity determined by
the method of balanced filters (ref. 101).  Fluorescence of the
samples is induced by the emission of a tritium/Zr radioisotope
source.  In determining copper in a brass matrix, a Cu foil filter
was placed in front of one of the counters and a Ni foil filter
in front of the other.  The absolute variance of determinations
of Cu was ±0.25? for measuring times of 8 minutes.

     A method was developed to minimize the matrix effects on
calibration curves (ref. 102).  The method uses measurements at
two wavelengths, one on either side of an absorption edge for the
element under study.  A development of the Beer's law expressions
through subtraction of the absorbtivities at the two wavelengths
(equal intensities are assumed for the two primary beams) leads
to a linear calibration curve.  The intercept of this curve is
related only to the density and absorption properties of the
matrix, while the slope is a linear function of the intercept
and the absorption properties of the element.  Once the slope
of the calibration curve has been determined for an element in
one matrix, a calibration curve in another matrix can be deter-
mined with only absorption measurements on the new matrix itself.

     If the density and composition of the new matrix are known,
the total calibration curve can be calculated without absorption
measurements.  The two wavelengths are obtained from Ka and Kg
radiations from an element just higher in atomic number than
the element analyzed, by using the target element or one of its
compounds in the sample position.  The Ka and Kg radiations are
resolved with an appropriate analyzing crystal.  Good agreement
was obtained between calculated and experimental results.  For
example, analysis of Ni over a concentration range 0-1.5 wt.%
had a standard deviation of about 0.03?.

     Two portable x-ray fluorescence analyzers, based on radio-
isotopic excitation and isolation of fluorescent radiation by
pairs of balanced absorption filters, have been developed.  The
instruments weigh between 12 and 16 pounds and are useful for
detecting elements having atomic number >20.  Sources and filter
pairs useful for Ti, Fe, Cu, Zn, Mo, Sn, Pb, Cr, Mn, Ni, Zr, Ba,
and W are reported (ref. 103).
                                353

-------
     Hans (ref. 104) analyzed plant stream samples automatically
with x-ray fluorescence for Cu by minimizing mutual interferences
of Co and Cu with dilution techniques and addition of concen-
trated sulfuric acid.

3.5   Neutron Activation Analysis

     Main virtues of neutron activation analysis (NAA) are sensi-
tivity and freedom from reagent contamination problems.  A typical
NAA analysis consists of four steps:  irradiation of the sample,
chemical isolation of the element sought, measurement of the
activity, and evaluation of the results.  The most time consuming
of these steps are the chemical isolation of the element to be
measured and the cooling time to minimize interfering radiation
prior to measuring activity.

     Another limitation of NAA is the massiveness of the conven-
tional neutron sources.  The limitation is somewhat alleviated
by using radioisotope sources which also permit on-stream detec-
tion of V, Mn, In, Se, Al, F, and U.

     Nondestructive neutron activation analyses of particulate
collected from ambient air have been reported for the Chicago
area (ref. 105, 106).  Particulate collected on cellulose filters
over a period of 24 hours was analyzed for manganese, vanadium,
chromium, and other elements after irradiating the specimen with
thermal neutrons.  Irradiation exposures were 1 minute and 24 hours
(for long-lived nuclides).  Cooling times, prior to counting, were
as long as 1, 2 and 9 months after irradiation.  Concentrations
of 2-120 ng V/m3 of air, 6-37 ng Cr/m3 of air, and 100-900 ng
Mn/m3 of air are reported.

     When doing neutron activation analysis of particulate col-
lected on filters, consideration must be given to the physical
properties and elemental composition of the filter itself.  A
practical aerosol filter for routine use must be easy to handle,
have high collection efficiency, and be low in price.  Commonly
used fiber glass and asbestos-base filters cannot be used because
of high background.  Millipore and Microsorban have acceptable
backgrounds but are expensive, fragile and hard to weigh due to
static charges.  Variability in background from batch to batch
is characteristic of Microsorban.  Cellulose-base filters are
easy to handle, are low in price, and have a low background.
Unfortunately, the collection efficiency is low; two filters
in series are required (ref. 106).

     A nondestructive and computer assisted neutron activation
analytical procedure for determining up to 33 elements including
Cr, Cu, Mn, Ni and V in air pollution particulates was developed
and tested in ambient air studies near the Northwest Indiana in-
dustrial area (ref. 107).  Polystyrene filters were used because
of low background signals.  Each specimen was irradiated for five
                               354

-------
minutes at a flux of 2 x 1012 neutrons/cm2sec.   After suitable
decay times, the following detection limits in absolute values
(yg) and in concentration in urban air (yg/m3)  for a 24-hour sam-
ple were obtained:  V @ 3 min decay time (0.001 yg; 0.002 yg/m3);
Cu @ 3 min decay time (0.1 yg; 0.02 yg/m3); Mn @ 15 min decay time
(0.003 yg; 0.0006 yg/m3); Cu @ 20-30 hr decay time (0.05 yg;
0.005 yg/m3); Cr g 20-30 day decay time (0.02 yg; 0.00025 yg/m3);
Ni @ 20-30 day decay time (1.5 yg; 0.02 yg/m3).

     Pillay, et al.  (ref. 108) developed a nonisolative (non-
destructive) neutron activation analysis procedure for determining
the concentration of 16 elements in airborne particulates collected
in Buffalo, N.Y. area.  The procedure involved multiple neutron
irradiation and high resolution gamma-ray spectrometry.

     Short-lived isotopes produced by thermal neutron irradiation
(5 minutes) of manganese and vanadium could be measured immediately,
After irradiating a second time for two hours and permitting a
decay for 8 to 12 hours, manganese, and copper were determined.
A third irradiation for 24 hours with a decay period of 5 to 7 days
was required to measure chromium and nickel.  Also, since the
filter paper was damaged by irradiation during the third irradia-
tion, an extraction step with nitric acid was required.

     Air particulate contains a large number of elements that
can interfere with the analyses.  However, by using activated
standards and by calibrating the detector-analyzer [semiconductor-
Ge(Li)] system with a large number of known isotopes, the inter-
ferences are  overcome (ref. 108).

     Activation analysis detection limits are generally expressed
on an interference-free basis.  Typical values on an interference-
free basis for selected elements are as follows:  B (1.1 yg),
Ba (0.02 yg), Cr (0.3 yg), Cu (0.002 yg), Mn (0.0001 yg),
Ni (0.7 yg), and V (0.002 yg).  However, during nondestructive
activation analyses, i.e., no radiochemical separation, where
interferences can occur, a commercial neutron activation analysis
service reports detection limits as follows for an air particulate
sample (14 mg) :  Cr (18.0 yg), Cu (1.10 yg), Ni (190 yg),
V (2.30 yg).  Less problems were observed with Ba and Mn.

     Wyttenbach (ref. 109) performed nondestructive NAA on
powdered rocks, cements, and meteorites for V,  Mn, and other
elements with irradiation times below one minute and counting
times below five minutes on specimens 10-40 mg.  The reproduci-
bility of V results is influenced by the Mn and_Al concentrations.
With Al, a critical weight ratio, V:A1 = 5 x 10 3 was observed.
(Undoubtedly, this effect can be alleviated somewhat by a detector
system - semiconductor, Ge(Li) - that has better resolution).
Grain size >150 mesh presented a problem since the specimens used
were inhomogeneous.
                               355

-------
     Mean concentrations of Mn (0.033 yg/m3) and V (0.015 yg/m3)
were measured by NAA and gamma-ray spectrometry in airborne par-
ticulates (ref. 110).  The technique is nondestructive, and is
most suitable for those elements which have a large neutron acti-
vation cross section for the production of isotopes which decay
with the emission of y-rays of high abundance and have convenient
half-lives (1 min.  to 1 day).  Irradiation time of 5 minutes
resulted in detection limits of Mn - 0.0003 yg/m~3 and V -
^0.001 yg/m3 for an air sample drawn through 20 cm2 of a glass
fiber filter.  The V detection limit was influenced by the masking
of the y-ray spectrum by other elements.  [Note - A Nal(Tl)
scintillation detector was used.  Better resolution and potentially
less interference would be attained with a semiconductor detector.]

     With Ge(Li) y-ray detectors better resolution is possible
than that obtained with Nal(Tl) scintillation counters previously
used for y-ray spectrometry (ref. 111).  By irradiating for
5 minutes, particulate collected on filters can be analyzed for
V, Mn, and other elements without destruction of the sample.
The best times after irradiation for counting are 0-20 minutes
for V and 1-4 hours for Mn.  Detection limits for samples (20 to
50 m3 of air) collected for 12 hours are V - 5 x lO'^yg/m3 and
Mn - 0.001 yg/m3.

     Generally neutron activation analysis (NAA) methodology uses
gamma-ray spectroscopy as the means of detecting over 70 elements
and steady state irradiation to induce activity.  However, poor
sensitivity is usually encountered with short-lived radionuclides.

     Improved sensitivity can be attained by using short-lived
radionuclides - including some pure beta emitters and some with
half-lives of one second or less - in NAA with high intensity
reactor pulses and a combination detector system, which includes
both a Nal(Tl) detector and a Cerenkov detector (ref. 112).

     Samples were irradiated with pulses of 109 watt peak power
and 16.6 milliseconds (FWHM) deviation.  Under these conditions,
the thermal- and fission-spectrum peak neutron fluxes were
1.1 x 1016n cm 2sec~1 and 2.1 x 10  n cm 2sec ', respectively.
These can be compared with thermal- and fission-spectrum neutron
fluxes in the steady-state irradiations (250 KW) of 2.5 x 1012n
cm 2sec~1 and 5-3 x 1012n cm~2sec l, respectively.

     The technique can be applied to the measurement of B, Be,
Li, 0, F, Mg and Pb.  With this sytem, the following detection
limits were attained:  B - 1.1 yg, Be - 15 yg, Li - 0.0008 yg,
0 - 370 yg, F - 0.83 yg, Mg - 160 yg and Pb - 0.4 yg.  The detec-
tion limits for B,  Be, and Li are superior to those reported by
steady state irradiation techniques whereas the detection limits
for F and Pb are comparable to, but those for 0 and Mg are poorer
                               356

-------
than previously reported limits.  The technique is rapid, but
limited in the analytical time requirements by the need of having
a weighed specimen.

     The detection of Cu by activation analysis is dependent on
other elements, particularly Na, capable of activation.  By using
computer techniques for analyzing gamma-spectra, a practical limit
of detection for copper in a sodium containing matrix corresponds
to a Na/Cu ratio of 100.

     The fact that 6l|Cu is one of the few positron-emitting
nuclides produced by thermal neutron activation has led to the
use of 180° annihilation quanta for gamma-gamma coincidence mea-
surements.  The application of this technique to a number of
matrices is shown by Michelsen and Steinner (ref. 113).  For the
concentration range, 50-650 ppm Cu, in geological samples having
a Na/Cu ratio of 3-3 to 330, the relative standard deviation
ranged from 2.0-6.4% for 10-minute counting times.  Good agreement
with other analytical techniques are reported.  The spread of
replicate determinations was reported as due to counting errors
and errors due to neutron flux inhomogeniety in the irradiation
position.  Possible interference from the nuclear reaction
  Zn(n,p)6l*Cu might be significant, but becomes important only
if the zinc content is 100 times higher than the copper content.

     Boron has been measured successfully by the 11B(p,Y)12Cu
reaction after irradiation with 900-keV protons (ref. 114).  At
this energy level the only interfering element is Li, but its
effect can be eliminated by using resonance reaction with 163-keV
protons.

     Several nondestructive neutron activation methods (ref. 115,
116) are based on the activation of 10B (natural isotopic abundance
of 19$).  By using the 10B(n,a)7Li  reaction, a particles can
be measured with an Au-Si surface barrier detector or after de-
excitation of the excited state; Li nuclei y-quanta can be mea-
sured with a Ge(Li) detector.  Irradiation with neutrons can be
accomplished with an Am-Be neutron source (ICi).

     Increased differentiation between elements by nondestructive
activation analysis can be obtained by using fast neutrons from
an isotopic source, rather than by using thermal neutrons.
Activation analysis by isotopic neutrons is suggested as being
superior in certain applications for selected elements, e.g., in
assaying geological specimens.  Ming and Wahlgren (ref. 117)
report the sensitivities (counts per gram) with fast neutrons
from an 2 
-------
     Boron and other elements, which cannot be detected by steady
state thermal neutron activation analysis, can be determined by
using half-lives between 0.55 and 873 msec with 1^-MeV neutrons.
A pulsing technique is used (ref. 118).

     Activation in linear accelerators or fixed or variable energy
cyclotron can result in the detection of 10~9g of B, C, N, 0,
and F.  Deuteron activation is used for trace amounts of B (ref.
119, 120).

     Gamma spectrometric measurement of a boron-containing sample
by a modulated neutron beam permits the determination of the
short-lived product (Th = 5 x lO'^sec) of the ' °B(n,a) 7Lim
reaction.  To obtain quantitative results, an internal standard
(samarium) is necessary to overcome variations in reactor flux
and the asymptotic relationship between concentration and count
rate for materials of high cross section (B) (ref. 121).

     The detection of traces of V in flowing aqueous systems has
been determined by continuous activation by fast  neutrons.  Large
volumes of solutions were activated and the gamma component of
induced radiation was measured.  Suggested applications include
detection of F, V, and U by on-stream monitoring  of industrial
effluent (ref. 122).

     Kartashev and Shtan (ref. 123) developed a similar on-stream,
continuous, automatic, neutron activation system  for V, Mn, In,
Se, Al, and F by using a 239Pu-Be source.

     By using a 12l*Sb-Be source for on-stream activation analysis,
V, Mn, Al, Hf, In and Ag were determined repetitively in 5-10 min
cycles at <100 ppm.  Additional data are given for Cl, Co, Mg, Na
and Cu (ref. 124).

3.6   Colorimetry

     Numerous spectrophotometric methods are available for mea-
suring B, Ba, Cr, Cu, Mn, Ni, and V as trace elements in various
media.  For measuring these elements, digestion processes (ashing,
acid extraction, and possible complexation) are necessary.
Because of the lengthy time elements involved, most spectrophoto-
metric methods are not applicable to continuous or short-term
intermittent measurements.  Information in the following subsection
is reported as related to potential problems involved in applying
spectrophotometric methods, or to methods providing high sensi-
tivity, specificity or high degree of automation.

     The spectrophotometric technique (diphenylcarbazide) for
determining chromium in air as particulate collected on cellulose
or glass-wool filters may produce erroneous results due to low
recovery of Cr(^605?).  The low values result from partial reduc-
tion of Cr(VI) to Cr(III) (ref. 125).
                                358

-------
     An automated spectrophotometric method for Cr has been
developed based on_catalysis of Cr(VI) of the reaction between
aqueous H202 and I  in HaSCU.  The time required for producing
a small fixed amount of Is is found by automated measurement of
the extinction at 365 nm and is_proportional to the Cr(VI) concen-
tration.  Na, K, Mn(II), Ni, Cl , and NOl do not interfere, but
oxidizing and reducing agents do.  From 0.6 to 3.0 yg of Cr(VI)
can be determined to within ±1 to 2% in 10 to 50 sec (ref. 126).

     An automated solvent extraction, colorimetric method for
determining Cu has been developed based on forming the diethyl-
dithiocarbamate or dibenzyldithiocarbamate and using a Technicon
Autoanalyzer.  The method enables 15 samples/hour to be handled
with precision comparable to corresponding manual methods
(ref. 127).

     A kineticjnethod for V based on the catalytic effect of V on
oxidation of I  by Br03 has been used to measure the concentration
of V within the range 0.01 to 100 ng per ml.  The rate of reaction
(measured photometrically) is directly proportional to concentra-
tion of V (ref. 128).

     Brief, et al. (ref. 129) summarize methods for collecting
and analyzing air streams for nickel carbonyl and report a de-
tailed development of the thermodynamics associated with the
formation of nickel carbonyl.  The various methods evaluated for
collecting and detecting nickel carbonyl include:

(1)  Collection - Absorption in saturated solution of sulfur
                  in trifluoroethylene
     Detection - Emission spectrography
     Detection Limit - 0.0003 ppm

(2)  Collection - Reaction with mercuric oxide to form Hg°
     Detection - Mercury vapor photometer
     Detection Limit - 0.0014 ppm

(3)  Collection - Absorption in acidified solution of
                  chloramine-B in alcohol
     Detection - Colorimetric using dimethylglyoxime
     Detection Limit - 0.01 ppm

(4)  Collection - Reaction with iodine in CCli,
     Detection - Colorimetry-dimethylglyoxime determination
     Detection Limit - 0.1 ppm

(5)  Collection - Reaction with dilute sulfuric acid
     Detection - Spectrophotometry-sodium diethyldithiocarbamate
     Detection Limit - No data

     Brief, et al. (ref. 129) propose a method for nickel carbonyl
in which air or process gas is bubbled into 10-15 ml of 3% aqueous
                               359

-------
HC1 solution in a midget impinger for 30-60 minutes at 0.1 ftVmin,
A filter assembly preceding the bubbler removes nickel solids.
After pH adjustment, color formation is developed with alpha-
furildioxime and the resultant colored complex is extracted into
chloroform to be measured spectrophotometrically.

     Other collection media for nickel carbonyl include IC1 in
acetic acid and alcoholic or ethereal iodine solution.  For con-
tinual determination of Ni(CO)i» in the gas phase, the use of
absorption spectrophotometry is reported, but no data are given
(ref. 130).

     A very sensitive, continuously recording nickel carbonyl
detector was developed based on the thermal decomposition (350°C)
of nickel carbonyl on a borosilicate glass plate.  The reflectance
of polarized light incident at the Brewsterian angle for the glass
is measured by a recording photomultiplier photometer and corre-
lated with the concentration of nickel carbonyl (ref. 131).  The
critical parameters are the rate of sample flow (500 cc/min) and
the rate of rotation of the glass disc on which the deposit is
formed.  With a 5-minute deposition tiem, a detection limit of
0.2 ppm is attainable; with a 10-minute time, a detection limit
of 0.05 ppm is observed.

     The system is also sensitive to iron carbonyl and could be
adapted to detection of alkyllead compounds and other metal-
organic s.

     A multiple dilution system for generating known concentra-
tions of nickel carbonyl is also presented (ref. 131).

     By sampling 100 cc of air or other test gases containing
nickel carbonyl @ 1 cc/sec, color development can be induced in
silica gel containing 0.5% AuCls.  Adjustment of the sample volume
permits a measuring range of 3-2000 ppm of Ni(COK (ref. 132).
Silica gel impregnated with KI03 in HaSCU has also been used to
determine Ni(COt) vapor in the atmosphere (ref. 133).

     Draeger tubes are available that measure Ni(CO)i, in the
range 0.1-1 ppm (ref. 134).

     An evaluation (ref. 135,136,137) of a number of methods for
determining chromic acid mist in air indicates that a filter paper
procedure is the most suitable for a rapid field method in prefer-
ence to methods employing sampling liquids.  By using a thick
(0.026-in.) absorbent paper, an efficiency of 99.9% was attained
by drawing air through the filter with a hand pump.  A reagent
mixture of ^% phthalic anhydride and 0.255? s-diphenylcarbazide
in 95% ethanol and 20 ml of glycerol per 100 ml of reagent is
                               360

-------
used to prepare the impregnated paper discs.  By visual compari-
son with dye standards in the field, a 600-ml air sample can
yield results within 0.5 mg/m3 for concentrations below 3.3 rng/m'*.

     A direct, specific, determination of decaborane in air can
be attained by absorbing the decaborane as a soluble red adduct
with quinoline in xylene (ref. 138).  High collection efficiencies
were obtained at gas flow rates of 0.5 1/min and concentrations
of 1 to 25 yg of decaborane per milliliter of solution can be
determined conveniently.  Neither diborane nor pentaborane inter-
fere .

     Two instruments - a portable, field model with a hand-operated
pump and a continuous recording analyzer - based on the nonspecific
reduction of triphenyltetrazolium chloride by boron hydrides have
been developed (ref. 139).  Metered air samples are passed through
filter paper or cloth tape impregnated with reagent.  The red
color produced is measured by visual comparison techniques in the
field model and by a differential reflectance photometer with the
automatic instrument.  Although relatively insensitive to diborane,
0.1 to 0.5 ppm of decaborane and pentaborane can be detected.
(Note - It is important to understand that the reaction is not
specific to boranes and that a large number of reducing agents
could interfere with the method.)

     Fristrom et al. (ref. 1^0) concluded after evaluating avail-
able analytical techniques that there is no quantitative detection
means applicable directly to all the volatile boron hydrides and
their organic derivatives in the range of their maximum allowable
concentrations.  As a consequence, a monitoring unit was developed
in which boron hydride vapors are burned in air and the solid
products (boric oxide and boric acid) are collected and analyzed
as solutions with a carmine colormetric (manual) method.  Best
colorimetric results were obtained on solutions containing 1 to
10 yg B/ml.  Sampling flow rates of 0.5 liter/min yielded 30 yg
B as BaOa (6 yg/ml HaO) in 30 minutes from air containing 1 ppm
pentaborane.

3.7   Polarography

     Particulate from air collected on Millipore filters or glass
fiber filters has been analyzed for Cu, Cr, Ni, and Mn and other
trace metals by polarographic techniques (ref. l4l).  Although
relatively high sensitivity (10 yg) and reasonable specificity is
attained, the preliminary sample preparation, which entails wet
or dry ashing (removal of organics) and acid extraction, is time-
consuming.

3.8   Back-scattering of g-radiation

     Back-scattering of B-radiation from 90Sr-90Y by Ba in ores
and concentrates has been used to measure Ba content (5-9^$) with
                               361

-------
0.2-5.0/8 relative error if samples and standards are <_20y grain
size, constant density, and constant moisture.  Interference
(soft backscattered B-radiation) by elements with atomic number
<26 was eliminated by using an Al filter (ref. 1^2).

3.9  Mass Spectrometry

     One of the major limitations of applying spark source mass
spectroscopy to particulate collected from ambient air or stack
emission is the fact that nonconducting materials cannot be used
as self-electrodes.  The nonconducting powdered sample must be
mixed with a conductor, e.g., graphite.

     Evans, et al . (ref. 1^3) demonstrated an analytical method
using a spark-source mass spectrograph equipped with an electri-
cal detection system.  With reproducible sample positioning, Cr,
Mn, V, and other elements were measured in stainless steel stan-
dards with precision of ±3% and accuracy of ±1%.  The major
problems are (a) spark-electrode instability, (b) sample or in-
strument turnaround, including mounting, chamber evacuation,
and prespark (Note - requires approximately 15 min/sample), and
(c) restrictions of mass range since only a M-2M mass range can
be covered.  (Note - Restrictions on accelerating voltage are
maintained to provide sensitivity and good ion extraction effi-
ciency.  Approximately 15 min is required to find a new mass
range, allow the magnetic field to stabilize, and set proper
peaks on the electrostatic peak switcher.

     A typical analysis for up to five elements (within M-2M) per
sample would require (a) 30-min peak setup and analysis of standard
per day and (b) 20 min/sample for a total of 21 samples and 105
determinations per 8-hr day.  For analyses requiring measurements
for a wide mass range of elements [assuming 4 groups of 5 elements
(C to Bi)], approximately 100 determinations per 8-hr day are
predicted (ref.
     Brown and Vossen (ref. 1*14) developed a special analytical
technique for collecting particulate from ambient air and for
determining all elements from atomic number 92 (uranium) to atomic
number 3 (lithium) in one scan with a spark source mass spectrom-
eter.  The mass spectrometer is fitted with an electrical detection
system which permits a measuring time of 9 minutes plus set-up
time between analyses of 10 minutes.

     Particulate is collected for 9.5 hours @ 19 liters/minute
on a nitrocellulose filter pad which is ashed to prepare a speci-
men for the mass spectrometric measurement.  The total weight of
sample retained on the pad was two milligrams and the following
concentrations of elements were reported:  B - 0.004 yg/m3,
Ba - 0.02 yg/m3, Cr - 0.30 yg/m3, V - 1.9 yg/m3, Mn - 0.0? yg/m3,
Ni - 0.32 yg/m3, and Cu - 0.26 yg/m3.
                                362

-------
     No data on repeatability, accuracy, etc., are shown.  The
work is principally a preliminary feasibility study.

     Bratten and Wood (ref. 1^5) studied the changes of relative
sensitivity coefficients for Cu, Cr, Mn, and Ni with concentra-
tion and matrix in measurements performed with spark-source mass
spectrometry.   Cr and Cu reveal little or no concentration or
matrix effects down to 0.08$ in nickel- or iron-base matrix.
However, Ni seems to produce considerably higher factors as the
concentration drops below 100 ppm.  Conversely, coefficients for
Mn decrease sharply below a few tenths of a percent.

     Reasonable agreement between spark source mass spectrometry
and neutron activation analysis for Cr, Cu, Ni and other elements
have been observed in forensic applications (ref. 1^6).  Poorer
agreement is observed for Mn, Ba, and Ti; spark source mass spec-
trometric analyses are higher.  There is the possibility of a
hydrocarbon interference for Mn, Ba and Ti in the mass spectrometer
data.

     Mass spectrometry has been used to measure boron indirectly
by determining the number of helium atoms produced in the thermal
and epithermal 10B(n,a) reaction (ref. 1^7).  The detection limit
of B varies with the type of sample; in steel, concentration below
10~5% B in samples of 0.1 to 1 g can be determined.

     Other mass spectrometric methods, based on stable isotope
dilution have been applied to the determination of trace metals.
A typical example (ref. 1^8) is the method as applied to trace
amounts of chromium.  The sample is dissolved in hydrochloric acid
containing a small portion of tracer solution, oxidized with nitric
acid and heated to fumes with sulfuric acid, and the chromium(III)
is then oxidized to chromium(VI) with cerium(IV) sulfate.  The
chromium(VI) is extracted into isobutyl methyl ketone in N hydro-
chloric acid and is then back-extracted into water.  The solution
is deposited on a rhenium filament and dried.  The deposit is
vaporized in the source of the mass spectrometer and the peak
intensities in the mass range 51 to 5*1 are measured.

     One of the major advantages of isotope dilution-mass spec-
trometry is that once isotope exchange has taken place there is
no requirement for quantitative or "clean" separations.

     The isotopic dilution method has been to the determination
of boron in silicon and zirconium (ref. 1^9).  By adding 1:1
glycerol-NaOH or 1:3 mannitol-MeOH to the isotopic-dilution
matrix, a stable, intense beam is obtained.

     The determination of Cu at ng levels, as its acetylacetonate
complex, by electron impact mass spectrometry is reported (ref.
150).  The method requires dissolution of the copper containing
                                363

-------
material, formation of the acetylacetone complex, and addition
of a completely deuterated complex as an internal standard.  The
error of the method is ^3%.

     Booker, et al.  (ref. 151) reports the quantitative measure-
ment of 0.5 to 200 ng of Cr by mass spectrometric analysis of
chromium(III) hexafluoroacetylacetonate.  The metal chelate is
prepared by digesting the specimen with perchloric acid followed
by chelation with hexafluoroacetone.   Although highly sensitive,
the analytical procedure requires over eight hours for the diges-
tion and chelation steps.

3.10  Chemiluminescence

     Although no analytical method has been reported based on
chemiluminescent reaction of nickel carbonyl with ozone, the
kinetics and mechanism studied by Morris and Niki (ref. 152)
suggest that a very sensitive, continuous, analytical method
based on this phenomenon may be attainable.  NiO emission was
observed with Ni(COK and ozone only  in the presence of carbon
monoxide.  With only ozone and nickel carbonyl, a nonluminous
reaction occurs, but the NiO emission can be generated by adding
carbonyl-free CO downstream.  Iron carbonyl may interfere but
the kinetics of the decay of FeO and  NiO emission are quite
different; also, the formation of NiO emission in ozone from
nickel carbonyl requires CO, whereas  the formation of FeO emis-
sions from iron carbonyl does not.

     Small amounts (down to 0.05-5 ng/ml) of Mn in any oxidation
state and in the presence of activators catalyze the oxidation
of luminol by Ha02.   The resulting increase in Chemiluminescence
is a measure of the Mn concentration.  Fe, Co, and Cr interfere
in high concentration, but the Fe interference can be prevented
by UV radiation of the mixture before adding luminol (ref. 153).

     Chemiluminescent reactions involving luminol-metal_(Cu)-H202
and similar reactions with V permit the detection of 10 9g of
metal (ref. 154).

3.11  Coulometry

     A continuous, automatic coulometric titration system was
devised to monitor boranes in air (ref. 155).  Boranes absorbed
from air into a sodium bicarbonate-potassium iodide electrolyte
are titrated with coulometrically generated iodine.  Detection
limits as low as 0.2 ppm by volume of diborane and decaborane
in air are attainable.  Theoretical limits are concentrations
of 0.01 ppm by volume.  All boranes react in the same way with
the iodine.  At air flow rates of ^0.420 liter/minute, collection
times of 15-20 minutes were used to measure 0.24 to 0.28 ppm
decaborane and collection times of approximately 45 minutes and
90 minutes were used to measure 1.02-1.08 ppm and 0.48 ppm of
diborane, respectively.


                               364

-------
     Acetone, peroxides, H2S, ozone, and nitrogen oxides inter-
fere; methanol, methyl borate, boric acid, benzene, butene and
hydrogen do not interfere (ref. 155).

3.12  Polymer-Radiographic Methods

     Although the practical application of an autoradiographic
method (ref. 156) for determining trace quantities of boron
(<1 ppm) in solids in collected airborne or stack emission par-
ticulate is questionable, the approach requires description.
Since cellulose acetobutyrate films will record a-particles only
up to 1.5 MeV, a moderately specific method for determining trace
quantities of boron may be devised based on this phenomenon.

     The technique is based on measuring the density of tracts in
the acetobutyrate film produced by the alpha particles generated
in the 10B(n,a)7Li reaction during irradiation of specimens with
a thermal-neutron flux.

     Too much analysis time is required to perform the steps of
irradiating, forming a replica, heat-treating, and etching to use
the technique as it exists now as a monitoring method.

     However, some small consideration should be given to the
feasibility of using polymers sensitive only to alpha particles or
proton tracks to detect B or other elements having isotopes with
large cross-sections for (n,a) or (n,p) reactions.  By using an
impact collection device with an a or proton sensitive polymer,
a portable radioisotope neutron source to promote the (n,a) or
(n,p) reaction, and a light obscuration system, a semi-continuous
monitor may be possible.  The principal limitation may be lack of
sensitivity of the polymer to the desired radiation resulting in
inordinate exposure or developing times.

3.13  Gas Chromatography

     Sunderman, et al. (ref. 157) developed a gas chromatographic
method for nickel carbonyl in blood and breath.  The nickel car-
bonyl in breath (0.5 yl/D was trapped in ethanol and subsequently
analyzed by gas chromatographic means.  The procedure has value
for measuring industrial exposures or plant emissions since the
method differentiates between Ni(CO)i» and less toxic Ni compounds.

     The major limitation is the need to collect the Ni(CO)i» in
ethanol.  A direct sampling procedure or a preconcentration step
on a section of gas chromatographic packing which could be tempera-
ture programmed for desorption of Ni(COK into the analytical
column would be a more attractive procedure for an intermittent
monitoring system to measure Ni(CO)i».
                                365

-------
3.1*J  Reaction of Nickel Carbonyl With Mercuric Oxide

     Brief et al. (ref. 158) summarized a continuous method
developed by Voll'berg (ref. 159), in which air containing nickel
carbonyl is drawn through mercuric oxide at 200°C and liberated
elemental mercury is measured spectrophotometrically .  A parallel
stream of air containing nickel carbonyl is drawn through an
oxidizing reagent to convert nickel carbonyl to carbon dioxide.
This effluent is then passed over mercuric oxide and the liberated
elemental mercury is determined spectrographically .   The differ-
ence in the amounts of mercury vapor obtained is a measure of
the nickel carbonyl in the air.  A sensitivity of O.OOHl ppm is
reported.


H.   SUMMARY AND RECOMMENDATIONS
     At this time, there is no continuous monitoring system for
stationary source emissions which measures B, Ba, Cr, Cu, Mn, Ni,
and V as individual elements or in a multi-element functional
model.  Most laboratory techniques to date have been applied to
integrated samples of particulate collected on filter paper from
ambient air.

     High sensitivities and the attendant low detection limits
of the absorption and emission spectroscopic techniques make these
techniques most attractive for continuous or short-term, inter-
mittent monitors.  By maintaining the element sought in the stack
gas and by introducing the stack gas as part of the oxidant or
fuel in the flame or high energy discharge of the excitation source
of the spectrometer, a continuous monitoring system is possible.
With B, Ba, Cr, Cu, Mn, Ni, and V, the major problems with using
the absorption or emission spectroscopic techniques are the
inter-element or compound formation effects.

     With atomic absorption spectrophotometry , the inter-element
or compound formation effects are minimized or eliminated by using
a nitrous oxide-acetylene flame.  The best results are obtained
with a sheath gas (inert gas, e.g., nitrogen) to exclude oxygen
from the flame.  Obviously, in ambient air or stack emissions,
elimination of the oxygen is extremely difficult.  One possible
way is to pass the stack gas or air over hot carbon to produce
carbon monoxide.

     The use of an inert gas plasma as an atomization source for
atomic absorption spectrophotometry will also minimize inter-
element effects if a reducing atmosphere is maintained in the
plasma.  The reducing atmosphere can be maintained by introducing
hydrogen gas.  An additional problem related to the use of a
plasma to provide the atomization for atomic absorption measure-
ments is the possibility of overexciting the element sought to
emission, rather than to the necessary ground energy state.  As
                               366

-------
a result, the population of ground state atoms may be too low to
attain maximum sensitivity.  Generally, the selection of the
"viewing zone" of the plasma must be optimized depending on the
element to be measured.

     Although flame photometry can be used for measuring Ba and
a number of other elements, inter-element and compound formation
effects are also a problem.  These effects can be minimized by
using a nitrous oxide-acetylene flame.

     Some inter-element effects are observed in arc- or spark-
emission spectroscopy, but these can be eliminated by using high
energy excitation, e.g., a radio-frequency plasma.  Continuous
monitoring systems for Be, Hg, and Pb, based on emission spec-
troscopy, have been developed using arc- or spark-excitation.
Similar approaches with either an arc-or-spark excitation or a
radio-frequency plasma can be applied to other elements.  With
a radio-frequency induced plasma, emission spectrographic analy-
ses for B, Ba, Cr, Cu, Mn, Ni, and V without inter-element effects
and on a continuous operational mode are possible.  Repeated
scanning of plasma induced emission lines of chromium and man-
ganese show high stability as indicated by relative standard
deviations of intensities of 1.9? and 1.7?, respectively, when
the sample is introduced continuously as an aqueous solution.

     X-ray emission (fluorescence) spectrometry is not applicable
for measuring boron emissions.  With a high resolution semi-
conductor detector, x-ray emission techniques with conventional
x-ray target tubes and with radioisotope sources can be used on
an intermittent basis to determine Ba, Cr, Cu, Mn, Ni, and V in
particulate collected on filters.  Care must be taken to measure
thin film deposits so as to minimize matrix effects.  Also, the
background of secondary emission of x-rays from the filter media
can be a major interference.

     Radio-isotope sources for x-ray emission measurements pro-
vide a high degree of portability and simplicity of ancillary
equipment.  When applied to monitoring particulate collected on
filters, good accuracy and repeatability of measurements are ob-
tained.  The major limitations in applying the x-ray techniques
are the time to collect sufficient particulate on the filter and
the potential loss of volatile emissions containing the element
sought.

     Nondestructive neutron activation analysis can be performed
on particulate for certain elements, but not all.  The major
problem is the interference from radioactive isotopes of other
elements.  Lengthy cooling times ranging from several hours to
several months are necessary to obtain sufficient sensitivity
for some elements, particularly Cr and Ni.  Manganese and vana-
dium can be measured immediately.  Depending on the detection
level required, copper can be measured immediately or, if a lower
detection level is necessary, after 20-30 hr decay time.


                               367

-------
     By using high intensity reactor pulsing techniques, boron
can be determined by nondestructive neutron activation analysis
techniques, but not by conventional steady-state thermal neutron
irradiation.

     With fast neutrons from an isotopic source, much shorter
irradiating and cooling times are possible.  There is some sac-
rifice of sensitivity, but the technique provides for detecting
Ba, Cr, Cu, Mn, and V in approximately 30 minutes total time.
Measurement of Ni can be accomplished in approximately 2 hours.

     Continuous, on-stream (aqueous solutions), neutron activa-
tion analyses of V and Mn have been performed with isotopic
sources at the ^100 ppm level in 5-10 minute cycles.

     The chemiluminescent reaction of Ni(CO)i» with ozone in pres-
ence of carbon monoxide deserves consideration as a possible
system for continuous monitoring of Ni(CO)if emissions.  Possible
applications including monitoring:  (a) effluent from the Mond
process of refining nickel, (b) formation of Ni(COK when CO in
effluent emissions passes over or through nickel products or
alloys, and (c) losses of nickel metal particulate from filter
media in particulate sampling systems.

     In special cases, continuous colorimetric measures are
possible for certain boranes.  However, the application of the
techniques must be made with care to ensure absence of inter-
ferences .

     Measuring boron indirectly with mass spectrometry by deter-
mining the number of helium atoms generated in the 10B(n,a)
reaction should be further evaluated.

     Any technique relying on the collection of particulate on
filters from relatively large volumes of air or stack gases risks
the potential loss of volatile compounds.  This is true even with
relatively nonvolatile compounds, for example, chromic oxide
volatilization can be enhanced by forming chromium trioxide in
an oxygen atmosphere.
                               368

-------
5.    REFERENCES


 1.  Miner, S., "Preliminary Air Pollution Survey  of  Barium  and
     Its Compounds," U.S.  Dept.  HEW,  NAPCA,  October 1969.

 2.  Durocher, N.  L., "Preliminary Air Pollution Survey  of Boron
     and Its Compounds," U.S. Dept. DEW,  NAPCA, Raleigh, N.C.,
     October 1969-

 3.  Hyatt, E. C.  and M. F.  Milligan, "Experiences With  Unusual
     Materials and Operations,"  Am. Ind.  Hyg.  Assoc.  Quart.  14,
     289-293 (1953).

 4.  Sullivan, R.  J. , "Preliminary Air Pollution Survey  of
     Chromium and  Its Compounds," U.S. Dept.  HEW,  NAPCA, Raleigh,
     N.C. ,  October 1969.

 5.  Caplan, D. and M. Cohen, "The Volatilization  of  Chromium
     Oxide," J. Electrochem. Soc. 108, 438-442 (1961).

 6.  Filimonov, V. N., "Reactions of  0 with NiO, Fe203,  and  Cr203
     Investigated  from Their Absorption Spectra in the Infrared,"
     Kinetika i Kataliz 4_(3), 367-372 (1963);  C.A. 59_, 5835g.

 7.  Kirk,  R. E. and D. F. Othmer (ed.),  Encyclopedia of Chemical
     Technology, Vol. 4, pp. 391-479, Interscience Encyclopedia,
     Inc.,  N.Y. (1949) .

 8.  Kirk,  R. E. and D. F. Othmer, Encyclopedia of Chemical
     Technology, Vol. 4, p.  417, Interscience  Encyclopedia,
     Inc.,  N.Y. (1949).

 9.  Sullivan, R.  J., "Preliminary Air Pollution Survey  of
     Manganese and Its Compounds," U.S. Dept.  HEW, NAPCA,
     Raleigh, N.C., October 1969.

10.  Kirk,  R. E. and D. F. Othmer (ed.),  Encyclopedia of Chemical
     Technology, Vol. 8, pp. 718-763, Interscience Encyclopedia,
     Inc.,  N.Y. (1952).

11.  Sullivan, R.  J., "Preliminary Air Pollution Survey  of Nickel
     and Its Compounds," U.S. Dept. HEW,  NAPCA, Raleigh, N.C.,
     October 1969.

12.  Gerstle, R. W., S. T. Cuffe, A.  A. Orning, and C. H. Schwartz,
     "Air Pollutant Emissions from Coal-Fired  Power Plants,  Report
     No. 2," J. Air Poll.  Control Assoc.  15.(2), 59-64 (1965).

13.  Smith, W. S., "Atmospheric  Emissions from Fuel Oil  Combus-
     tion - An Inventory Guide," U.S. Dept.  HEW, Public  Health
     Surv.  Publ. 999-AP-2  (1962).
                               369

-------
14.  Athanassiadis, Y.  C.,  "Preliminary Air Pollution Survey of
     Vanadium and Its Compounds," U.S.  Dept.  of HEW,  NAPCA,
     N.C.,  October 1969.

15.  Zwiebaum, F. and J.  Moorhead, "A Multi-element Atomic
     Absorption Analyzer,"  At.  Absorption Newsletter 6_(6),
     137 (1967).

16.  Malinek, M. and H. Massmann, "Use  of the Kranz Plasmajet as
     a Source of Atomic Vapor in Atomic Absorption Spectrometry,"
     Can.  Spectrosc. 15_(3), 77-79, 8*1 (1970).

17.  Loftin, H. P., C.  M. Christian,  and J. W.  Robinson, "The
     Continuous Determination of Lead in Air,"  Spectroscopy
     Letters 3_, 161-172 (1970).

18.  Biancifiori, M. A. and C.  Bordonali, "Utilization of an
     Induction Plasma in the Analysis of Traces by Atomic Absorp-
     tion," Com. Naz. Energ. Nucl., RT/CHI (67) 15, 17 pp (1967);
     C.A.  6_9_, 7988m (1968).

19.  Bordonali, C. and M. A. Biancifiori, "Plasma Induction in
     Trace  Analysis by Atomic Absorption" Corsi Semin. Chim. 9_,
     153-68 (1968); C.A.  72., 50499b (1970).

20.  Sullivan, J. V., A.  B. Timms, and  P. A.  Young, "Atomic
     Absorption Analyses for Nickel Using a Resonance Detector,"
     Australas. Inst. Mining Met. Proc. No. 226 (Pt.  1), 31-36
     (1968); C.A. 69., 831476 (1968).

21.  Larkins, P. L., R. M.  Lowe, J. V.  Sullivan, and A. Walsh,
     "The  Use of Solar-blind Photomultipliers in Flame Spec-
     troscopy," Spectrochim. Acta 21, 727 (1965).

22.  Vol'berg, N. Sh.,  "Determination of Microquantities of
     Nickel Carbonyls in Air,"  U.S.S.R. 135,684, Feb. 15, 1961;
     C.A.  55., 15230b (1961) .

23.  Thompson, R. J., G.  B. Morgan, and L. J. Purdue, "Analysis
     of Selected Elements in Atmospheric Particulate Matter by
     Atomic Absorption," Atomic Absorption Newsletter 9_, 53-57
     (1970).

24.  Hwang, J. Y. and F.  J. Feldman,  "Determination of Atmospheric
     Trace  Elements by Atomic Absorption Spectroscopy," Appl.
     Spectry. 24_, 371-374 (1970).

25.  Kneip, T. J., M. Eisenbud, C. D. Strehlow, and P. C.
     Freudenthal, "Airborne Particulates in New York City,"
     J. Air Poll. Control Assoc. 20_,  144-149 (1970).
                               370

-------
26.  Scheub, W. H. and C.  J. Stomsky, "Effect of Dissolved Air
     on Rate of Aspiration in Atomic Absorption Spectrophotom-
     etry," Atomic Absorption Newsletter 6(4), 95 (1967).

27.  Ramirez-Munoz, J. and Roth, M.  E., "Metallurgical Applica-
     tions of Atomic-absorption Flame Photometry.  I.  Interferences
     by Iron on Cobalt, Chromium, Copper, Manganese,  and Nickel,"
     Flame Notes 3_(D, 2-10 (1968).

28.  Coker, D. T. and J. M. Ottaway, "Overexcitation  Interferences
     in Atomic Absorption Spectrophotometry With an Air-acetylene
     Flame," Nature 227 (5260), 831-2 (1970).

29-  Manning, D. C. and L. Capacho-Delgado, "Dissociation and
     lonization Effects in Atomic Absorption Spectrochemical
     Analysis," Anal. Chim. Acta 36., 312 (1966).

30.  Kahn, H. L., "Determination of  Barium With Nitrous Oxide-
     Acetylene Flame," Atomic Absorption Newsletter 6(4),  96
     (1967).

31.  Manning, D. C., "The Determination of Boron, Beryllium,
     Germanium, and Niobium Using the Nitrous Oxide-Acetylene
     Flame," Atomic Absorption Newsletter 6_(3), 35-37  (1967).

32.  Marshall, D. and W. G. Schrenk, "Atomic Absorption Charac-
     teristics of Vanadium Using the Kniseley Burner," Spectroscopy
     Letters 1(2), 87-93 (1968).

33.  Nakahara, T., M. Munemori, and  S.  Musha, "Atomic-absorption
     Spectrophotometry of Vanadium (V)  [with a Nitrous Oxide-
     Acetylene Flame], J.  Chem. Soc. Japan, Pure Chem. Sect.
     90(7), 697-703 (1969).

34.  Kerber, J. D. and W.  B. Barnett, "Computer Analysis of the
     Determination of Barium by Atomic  Absorption and  Flame
     Emission Spectroscopy," Atomic  Absorption Newsletter 8^(6),
     113-116 (1969).

35.  Hurlbut, J. A. and C. D. Chriswell, "Suppression  of Inter-
     ferences by Sodium Sulfate in Trace Chromium Analyses by
     Atomic Absorption," Anal. Chem. 4_3_, 465-466 (1971).

36.  Sachdev, S. L., J. W. Robinson, and P. W. West,  "Determina-
     tion of Manganese, Iron, Cobalt, and Nickel in Air and
     Water by Atomic Absorption Spectroscopy," Anal.  Chim. Acta
     38., 499-506 (1967).

37.  Yanagisawa, M.s M. Suzuki, and  T.  Takeuchi, "Cationic
     Interferences in the Atomic Absorption Spectrophotometry
     of Chromium," Anal. Chim. Acta  5_2, 386-9 (1970).
                               371

-------
38.   Matousek,  J.  and V.  Sychra,  "Atomic Fluorescence  Study  on
     Iron,  Cobalt  and Nickel,"  Anal.  Chem.  4^(3),  518-522  (1969).

39.   Mitchell,  D.  G.  and  A.  Johansson,  "Simultaneous Multielement
     Analysis Using Sequentially  Excited Atomic Fluorescence
     Radiation," Spectrochim. Acta  25B, 175-182  (1970).

40.   Air Pollution Measurements of  the  National Air Sampling
     Network, Analyses of Suspended Particulates,  1957-1961,
     U.S. Dept.  of Health, Education, and Welfare, Public  Health
     Service Publication  No.  978, U.S.  Government  Printing
     Office, Washington,  D.C.  (1962).

41.   Air Quality Data from the  National Air Sampling Networks  and
     Contributing  State and  Local Networks, 1966 ed.,  U.S. Dept.
     of Health,  Education, and  Welfare, National Air Pollution
     Control Administration  Publication No. APTD 68-9,  U.S.
     Government  Printing  Office,  Washington,  D.C.  (1968).

42.   Webb,  R. J.,  M.S.W.  Webb,  and  P. C. Wildy,  "A Monitor for
     the Quantitative Determination of  Beryllium in the Atmo-
     sphere," United  Kingdom Atomic Energy  Authority,  AERE-R
     2868 (1959).

43.   Webb,  M.S.W., R. J.  Webb and P.  C. Wildy, "Monitor for  the
     Quantitative  Determination of  Beryllium in the Atmosphere,"
     Journal of  Sci.  Instr.  37., 466-471 (I960).

44.   Fromm, D.  and A. v.  Oer,  "Nachweis von Beryllium  und
     emissionsspektrographischer  Nachweis von Quecksilberdampf
     in Luft,"  Mikrochem.  Acta  I960,  235-244.

45.   Aughey, H., "A Rapid Mobile  Analyzer for Minute Amounts of
     Lead in Air," J. Opt. Soc. Am. 39_, 292-293  (1949).

46.   Koppius, 0. B.,  "Detection of  Lead in  Air With the Aid  of a
     Geiger-Muller Counter," J. Opt.  Soc. Am. 39,  294-297  (1949).

47.   Agazzi, E.  J., "Extraction-flame Photometric  Determination
     of Boron,"  Anal. Chem.  3£(2),  233-5  (1967).

48.   Zolotavin,  V. L., Bukreev, Yu. F., Muzgin, V. N.,  and
     Prorok, M.  M., "Effect  of  Some Elements on the Determination
     of Calcium, Strontium,  and Barium  in Vanadium-containing
     Materials," Tr.  Inst. Khim., Akad. Nauk SSSR, Ural. Filial
     1970,  No.  17, 159-161;  C.A.  7jL»  38009a.

49.   Ingamells,  C. 0., N.  H.  Suhr,  F. C. Tan, and  D. H.  Anderson,
     "Barium and Strontium in Silicates.  A Study  on the Develop-
     ment of Analytical Methods," Anal. Chim Acta  5_3(2), 345-360
     (1971).
                               372

-------
50.  Passel, V. A., R. W. Slack, and R. N. Kniseley, "Flame
     Emission Spectrometric Determination of Aluminum, Cobalt,
     Chromium, Copper, Manganese, Niobium (Nickel), and Vanadium
     in Low and High Alloy Steels," Anal. Chem. 43, 186-191
     (197D.

51.  Puwa, K. and B. L. Vallee, "Multichannel Flame Spectrometer.
     I.  Instrument Design," Anal. Biochem.  12, 444 (1966).

52.  lida, C. and K. Fuwa, "Studies With a Multichannel Flame
     Spectrometer.  II. Simultaneous Determination of Magnesium,
     Calcium, Copper, Manganese, and Chromium," Anal. Biochem.
     21, 1-8 (1967).

53-  lida, C. and K. Fuwa, "Studies With a Multichannel Flame
     Spectrometer.  III. Simultaneous Analysis of Magnesium,
     Calcium, Copper, Manganese, and Chromium:  A Method Devoid
     of Cation and Anion Interferences," Anal. Biochem. 21,
     9-21 (1967).

54.  Hobbs, R. S., G. F. Kirkbright, and T.  S. West, "Performance
     of the Separated Air-acetylene Flame in Thermal Emission
     Spectroscopy," Analyst 9_4_, 554-562 (1969).

55.  Moore, D. M., "B 2497-73 A* and Line Interference in the
     Analysis of Silicate-bearing Rocks," Appl. Spectroscopy
     23_, 278-279 (1969).

56.  Takahashi, K., T. Takahashi, and E. Sudo, "Spectrochemical
     Analysis of Boron in High Purity Quartz by d.c. Arc Discharge
     Under Argon Atmosphere," Bunko Kenkyu 18/6), 311-314 (1969);
     C.A. 73., 31242s (1970).

57-  Holler, P. and K. Slickers, "Spectroscopic Determination of
     5 to 59 ppm of Boron in Steel," Arch. Eisenhutt Wes 39(2),
     129-134 (1968).                                     ~~

58.  Perman, J. and E. Perman, "Spectrographic Determination  of
     Metal Traces in the Atmosphere," Bol. Geol.  Minera 80(5),
     476-484 (1969); C.A. 71, 124798x (1970).

59.  Schindler, R., "Determination of Vanadium in Sediments by
     Emission Spectral Analysis," Z. Angew.  Geol. 15_(6), 309-313
     (1969); C.A. 73, 31309u (1970).

60.  Snelleman, W., T. C. Rains, K. W. Yee,  H. D. Cook, and
     0. Menis, "Flame Emission Spectrometry  With Repetitive
     Optical Scanning in the Derivative Mode," Anal. Chem.  42,
     394-398 (1970).

61.  Ricard, R., "Copper Arc Spectrum Line,  A 2392.627 X,"
     Method Phys. Anal. £(2), 169 (1970); C.A. 73., 7l838x (1970).
                               373

-------
62.  Dickinson, G.  W.  and V.  A.  Fassel,  "Emission Spectrometric
     Detection of the  Elements at the Nanogram per Milliliter
     Level Using Induction-Coupled Plasma Excitation,"  Anal.
     Chem. 4_1, 1021-1024 (1969).

63.  Mavrodineanu,  R.  and R.  C.  Hughes,  "The RP Discharge at
     Atmospheric Pressure and Its Use as an Excitation  Source
     in Analytical  Spectroscopy," Developments in Applied Spec-
     troscopy, Volume  3, 305-333, Plenum Press, N.Y.  (1963).

64.  Runnels, J. H. and J. H. Gibson, "Characteristics  of Low
     Wattage Microwave Induced Argon Plasmas in Metals  Excitation,"
     Anal. Chem. 39., 1398-1405 (196?).

65.  Tsugio, T. and Shigenori, S., "Fundamental Study of Emission
     Spectroscopy Using Plasma Jet Excitation," Bunseki Kagaku
     16/9), 888-894 (1967)i C.A. 68^, 119064t (1968).

66.  Matsumoto, C., S. Taniguchi, K. Suzusho, and T.  Sakaguchi,
     "Simultaneous  Spectrographic Determination of Nickel, Man-
     ganese, and Chromium by the Plasma  Jet Arc Technique,"
     Bunko Kenkyu 17_(2), 61-71 (1968); C.A. 70., 53690b  (1969).

67.  Morrison, G. H. and Y. Talmi, "Micro-analysis of  Solids by
     Atomic Absorption and Emission Spectroscopy Using  an R.P.
     Furnace," Anal. Chem. 4^, 809-811 (1970).

68.  Bandyukova, T. N., V. K. Zakharov,  P. A. Koka, and A. P.
     Malinovskaya,  "High-frequency Induction Plasmatron and Its
     Use in Emission Spectroscopy," Primen, Plasma-trona Spektrosk.,
     Mater Vses. Simp. 1968 (Pub 1970),  191-193; C.A. 7_4_,
     1344l6b (1971).

69.  Cares, J. W.,  "The Quantitative Determination of Airborne
     Metallic Dusts and Fumes by X-ray Spectrometry," Amer. Ind.
     Hyg. Assoc. J. 2£, 463-468 (1968).

70.  Leroux, J. and M. Mahmud, "Flexibility of X-ray  Emission
     Spectrography  as  Adapted to Microanalysis of Air Pollutants,"
     J. Air Poll. Control Assoc. 20^, 402-404 (1970).

71.  Hirt, R. C., W. R. Doughman, and J. B. Gisclard, "Application
     of X-ray Emission Spectrography to  Air-borne Dusts in
     Industrial Hygiene Studies," Anal.  Chem. 2£, 1649-1651 (1956).

72.  Ananthamurthy, B. S., "Interlemental Effect of Chromium and
     Manganese in Solutions on the Intensity by X-ray Spectros-
     copy," Trans.  Indian Inst. Metals 2JL(4), 47-50 (1968).

73.  Mitchell, B. J. and J. E. Kellam, "Unusual Matrix Effects
     in X-ray Spectroscopy:  A Study of  the Range and Reversal
     of Absorption  Enhancement," Appl. Spectrosc. 22, 742-748
     (1968).


                                374

-------
74.  Graham, M. J. and C. S. Bray, "The Use of Evaporated Metal
     Film Standards in Thin Layer X-ray Fluorescence Analysis of
     Mixed Oxides," J. Sci. Instrum.  (J. Phys. E) 2_ (ser. 2)
     706-708 (1969).

75.  Mitchell,  B. J., "X-ray Spectrographic Determination of
     Zirconium, Tungsten, Vanadium, Iron, Titanium, Tantalum,
     and Niobium Oxides.  Application of the Correction Factor
     Method," Anal. Chem. 3_2_, 1652-6  (I960).

76.  Mitchell,  B. J. and H. J. O'Hear,  "General X-ray Spectro-
     graphic Solution Method for Analysis of Iron, Chromium,
     and (or) Manganese Bearing Materials," Anal. Chem. 3**,
     1620-5 (1962).

77.  Wittig, W. J., "Production Control Analyses with the
     Automatic  X-ray Spectrograph," Norelco Kept. 6_, No. 1,
     17-22 (1959).

78.  Morris, A. W., "The Simultaneous Determination of Vanadium,
     Chromium,  Manganese, Iron, Cobalt, Nickel, Copper, and Zinc
     in Sea Water by X-ray Fluorescence Spectrometry," Anal.
     Chim. Acta ^2_, 397-406 (1968).

79.  Kamada, H., R. Inoue, M. Terasawa, Y. Gohshi, H. Kamei,  and
     I. Fujii,  "The Determination of  Light Elements by Proton
     Excitation and X-ray Spectrometry," Anal. Chim. Acta 46,
     107-112 (1969).

80.  Chernoberezhskii, Yu. M., A. I.  Yanklovich, T. A. Kuz'mina,
     A. M. Solov'ev, and S. A. Toporkov, "Preparation of Films
     of Organic Materials as Analyzers  for the Ultrasoft X-ray
     Spectral Region," Vestn. Leningrad. Univ., Fiz., Khim.
     1970, (1), 163-5; C.A. 73., 19874s  (1970).

81.  Prokopvich, S. A. and E. R. McCartney, "An X-ray Spectro-
     graphic Method for Determining Barium," Analyst 92, 253-256
     (1967).

82.  Fabbi, B.  P., "X-ray Fluorescence  Determination of Barium
     and Strontium in Geologic Samples," Appl. Spectrosc. 25,
     316-318 (197D.

83.  Gunn, B. M., "The Determination  of Barium in Silicate
     Samples by X-ray Fluorescence Analysis," Can. Spectroscopy
     14_, 164-170 (1969).

84.  Willis, M. P., H. W. Fesq, E.J.D.  Kable, and G. W. Berg,
     "The Determination of Barium in  Rocks by X-ray Fluorescence
     Spectrometry," Can. Spectroscopy 14, 150-157, 163 (1969).

85.  Rhodes, J. R., "Radioisotope X-ray Spectroscopy," Analyst 91,
     683-699 (1966).


                                375

-------
86.   Rhodes,  J.  R.,  "Design and Application  of X-ray  Emission
     Analyzers Using Radioisotope or Gamma Ray Sources,"  Energy
     Dispersion  X-ray Analysis:  X-ray and Electron Probe
     Analysis, ASTM  STP 485, pp.  243-285  (1971).

87.   Birks,  L. S., "Current Trends in X-ray  Fluorescence  Spec-
     trometry,"  Appl. Spectrosc.  303-308  (1969).

88.   Rhodes,  J.  R.,  A. H.  Pradzynski, and J.  S.  Payne,  "Energy
     Dispersion  X-ray Fluorescence Spectroscopy  for Rapid Multi-
     element  Analysis of Air Particulates,"  Preprint  of paper
     to be Given at  Pittsburgh Conference, Cleveland, Ohio,
     March 1972.

89.   Rhodes,  J.  R.,  A. H.  Pradzynski, and R.  D.  Sieberg,  "Energy
     Dispersive  X-ray Emission Spectrometry  for  Multielement
     Analysis of Air Particulates," Preprint  of  paper to  be given
     at Instrument Society of America Symposium,  San  Francisco,
     Calif.,  May 1972.

90.   Giauque, R. D., "A Radioisotope Source.   Target  Assembly
     for X-ray Spectrometry," Anal. Chem. 40, 2075-7  (1968).

91.   Frankel, R. S.  and D. W. Aitken, "Energy-dispersive  X-ray
     Emission Spectroscopy," Appl. Spectrosc. 24,  557-566 (1970).

92.   Karttunen,  J. 0., H.  B. Evens, D. J. Henderson,  P. J.
     Markovich,  and  R. L.  Niemann, "A Portable Fluorescent X-ray
     Instrument  Utilizing Radioisotope Sources,"  Anal.  Chem.  36,
     1277-1282 (1964).

93.   Karttunen,  J. 0. and D. J. Henderson, "An Improved Portable
     Fluorescent X-ray Instrument Using Radioisotope  Excitation
     Sources," Anal. Chem. 3_7_, 307-309 (1965).

94.   Yamamoto, S., "A Radioisotope X-ray  Fluorescence Spectrometer
     With a High-Resolution Semiconductor Detector,"  Anal. Chem.
     41, 337-342 (1969).

95.   Bol'shakov, A.  Yu, "Use of Two Radioisotope  Sources  in X-ray
     Fluorescence Analysis of Mineral Raw Materials," At. Energ.
     25, 535-536 (1968); C.A. 70, 738l2w.

96.   Dothie,  H.  J. and B.  Gale, "Nondispersive X-ray  Fluorescence
     Absorption-edge Spectroscopy," Spectrochim.  Acta 20(11)3
     1735-1755 (1964).

97.   Rhodes,  J.  R.,  T. G.  Ahier,  and D. 0. Poole,  "Analysis of
     Zinc and Copper Ores by Radioisotope X-ray  Fluorescence,"
     At. Energy  Res. Estab. Repts. AERE-R-44?4,  21 pp (1964).
                               376

-------
 98.  Tanemura, T., "Performance of An X-ray Spectroanalyzer Using
      IIf7Pm B-ray Source," Japan J. Appl.  Phys.  5_(D,  51-58  (1966)
      (Eng).

 99.  Orlov, V. N., I.  V. Tomskii, G.  F.  Komissarzhevskaya,
      "Determination of the Copper Content in Natural  Deposits
      and Samples by a  Radioisotopic X-ray Fluorescent Method,"
      Vop. Razved. Geofiz. 1^, 47-49 (1969); C.A.  73.,  62337f
      (1970).

100.  Ellis, W. K., R.  A. Fookes, V. L. Gravitis,  and  J.  S.  Watt,
      "Radioisotope X-ray Techniques for On-Stream Analysis  of
      Slurries.  Feasibility Studies Using Solid Samples  of
      Mineral Products," Int. J. Appl. Radiat. Isotopes,  20,
      691-701 (1969).

101.  Bet in, Yu. P., M. I. Bursukova,  F.  M. Lipkin, and L. S.
      Shelkov, "Nondispersive X-ray Fluorescence Analysis by
      Means of a Two-channel Apparatus," Zavod.  Lab.  32(12),
      1531-1533 (1966); C.A. 66., 121746u (1967).

102.  Knapp, K. T., R.  H. Lindahl, and A.  J. Mabis, "X-ray
      Absorption Method for Elemental  Analysis," Advan. X-ray
      Anal. 7_, 318-24 (1964).

103-  Bowie, S.H.U., "Portable X-ray Fluorescence Analyzers  in
      the Mining Industry," Mining Mag. (London) 118_(4),  230-1,
      233, 235, 237, 239 (1968).

104.  Hans, A., "Rapid  Automatic Analysis  by X-ray Fluorescence
      of Mineral Powders and Solutions from the  Hydrometallurgical
      Plants of Union Miniere du Hant-Katanga, Chim. Anal.  42,
      282-6 (I960).

105.  Brar, S. S., D. M. Nelson, E. L. Kanabrocki, C.  E.  Moore,
      C. D. Burnham, and D. M. Hattori, "Thermal Neutron  Activa-
      tion Analysis of  Particulate Matter  in Surface Air  of  the
      Chicago Metropolitan Area:  One-minute Irradiations,"
      Environ. Sci. Technol. 4_, 50-54  (1970).

106.  Brar, S. S., D. M. Nelson, J. R. Kline, and P. F. Gustafson,
      "Instrumental Analysis for Trace Elements  Present in Chicago
      Area Surface Air," J. Geophysical Research 75,  2939-2945
      (1970).

107.  Dams, R., J. A. Robbins, K. A. Rahn, and J.  W.  Winchester,
      "Nondestructive Neutron Activation Analysis of Air  Pollu-
      tion Particulates," Anal. Chem.  42_,  861-867 (1970).

108.  Pillay, K.K.S., C. C. Thomas, and C. M. Hyche,  "Neutron
      Activation Analysis of Inorganic Constituents of Airborne
      Particulates," Nucl. Technol. 10_(2), 224-231 (1971).
                               377

-------
109-   Wyttenbach, A., "Rapid Instrumental  Nuclear Activation
      Analysis of Rocks, Cements,  and Meteorites," Helvetica
      Chimica Acta 52_, 2458-2465 (1969).

110.   Keane, J. R. and E.M.R.  Fisher, "Analysis of Trace Elements
      in Air-borne Particulates by Neutron Activation and Gamma-
      ray Spectrometry," Atmospheric Environment 2_, 603-614  (1968).

111.   Zoller, W. H. and G.  E.  Gordon, "Instrumental Neutron
      Activation Analysis of Atmospheric Pollutants Utilizing
      Ge(Li) y-ray Detectors," Anal. Chem. 4j[,  257-265 (1970).

112.   Lukens, H. R., "A Neutron Activation Analysis Method for
      the Determination of Be, Li, B, F, and Pb," J. Radioanal.
      Chem.  1, 349-354 (1968).

113.   Michelsen, 0. B. and E.  Steinnes,  "Determination of Copper
      in Geological Material by Neutron Activation and Gamma-gamma
      Coincidence Spectrometry," Talanta 15_, 574-578 (1968).

114.   Vasil'ev, S. S., G. I. Mikhailov. L. P.  Starchik, and  L.  V.
      Konanykin, "Use of the Reaction riB(p,y)12C for Determining
      Boron," Zav. Lab., 35_(3), 299-300  (1969).

115.   Jinno, K., "Indirect Activation Analysis  of Boron Based on
      the Spectrometry of Prompt Alpha Particles by Using an
      Americium-beryllium Neutron Source," Radioisotopes 19,
      210-211 (1970).

116.   Juna,  J., K. Konecny, and M. Vobecky, "Nuclear-reaction
      Method for the Determination of Boron,"  Colin. Czech,  chem.
      Commun., 34, 1605-1610 (1969).

117.   Wing,  J. and M. A. Wahlgren, "Optimal Detection Sensitivi-
      ties in Activation with Fast Neutrons from an 2I*lAm-2 **2Cm-Be
      Source," Appl. Spectroscopy 23, 5-7  (1969).

118.   Golanski, A., "The 14-MeV Neutron Activation Analysis  Using
      Half-lives Between 0.55 and 873 Milliseconds," J. Radioanal.
      Chem.  1(3-4), 161-173 (1969).

119.   Engelmann, C., "Analysis by Activation With Charged
      Particles and y Photons," Bull. Inform.  Sci. Tech No.  140,
      65-70  (1969).

120.   Engelmann, C., "Ultrasensitive Analysis  of Light Elements
      in Materials of Very High Purity by  Activation Using y
      Photons and Charged Particles," Int. J.  Appl. Radiat.
      Isotopes 18.(8), 569-578 (1967).

121.   Isenhour, T. L. and G. H. Morrison,  "Determination of Boron
      by Thermal Neutron Activation Analysis Using a Modulation
      Technique," Anal. Chem.  38., 167-169  (1966).


                               378

-------
122.  Jervis, R. E., H. Al-Shahristani,  and S.  S.  Nargolwalla,"
      Fast Neutron Continuous Activation Analysis  of Dilute
      Solutions," Nat. Bur.  Stand.  Spec. Publ.  1969, No.  312(2),
      918-924.

123-  Kartashev, E. R. and A. S.  Shtan,  "Activation Determination
      of Indium, Selenium, Fluorine, and Other  Elements in a
      Stream of Solutions,"  Vses. Nauch-Issled. Inst.  Radiat.
      Tekh., Rep. 1968, VNIIRT-2, 18 pp.; Nucl. Sci. Abstr. 1969,
      23_(l4), 26196.

124.  Downs, W. E. and M. W. Davis, "Characteristics of an
      On-stream Analysis System Using a  Multikilocurie Antimony-
      124-beryllium Neutron  Source," Nucl. Appl.  Technol.  7(5),
      466-471 (1969).

125.  Dutkiewicz, T., J. Konczalik, and  M. Przechera,  "Evaluation
      of Colorimetric Methods for the Determination of Chromium
      in Air or Urine by Comparison with the Radio-isotope
      Technique," Acta Pol.  pharm.  2j[(2), 165-172  (1969).

126.  Hadjiioannou, T. P., "Catalytic Micro-determination  of
      Chromium (VI)," Talanta 15_(6), 535-539 (1968).

127.  Carter, J. M. and G. Nickless, "Solvent-extraction Technique
      with the Technicon AutoAnalyzer,"  Analyst 95_(1127),  148-
      152 (1970).

128.  Yatsimirskii, K. B. and V.  E. Kalinia, "Kinetic  Method for
      Determination of Nanogram Amounts  of Vanadium,"  Zh.  analit.
      Khim.  2JK3), 390-394 (1969).

129.  Brief, R. S., F. S. Venable,  and R. S. Ajemian,  "Nickel
      Carbonyl:  Its Detection and  Potential for  Formation,"
      Am. Ind. Hyg. Assoc. J. 26_, 72-76  (1965).

130.  Benes, M., M. Hejtmanek, and  B. Polej , "Determination of
      Nickel Carbonyl in Fuel Gases," Sb. Vys.  Sk. Chem.-Technol.
      Praze, Technol. Paliv  17., 77-101 (1969);  C.A. 73., 111538h
      (1970).

131.  McCarley, J. E., R. S. Saltzman, and R. H.  Osborn,  "Recording
      Nickel Carbonyl Detector,"  Anal. Chem. 2^,  880-882  (1956).

132.  Yoshitaka, K., "Rapid  Method  for the Determination  of Low
      Concentrations of Nickel Carbonyl  Vapor," Yuki Gosei Kagaku
      Kyokaishi 1J5, 466-471  (1957); C.A. 5_1, l6204g (1957).

133.  Vol'berg, N. Sh and E. E. Gershkovich, "Use  of Solid Sorbents
      in Industrial Chemistry," Nov. Obi. Prom.-Sanit. Khim 1969,
      41-44; C.A. 71, 1283l8t (1969).
                               379

-------
134.  Leichnitz, K., "Draeger Tubes," Draeger-Hefte No.  271,
      20-23 (1968).

135.  Ege, J.  R. Jr. and L.  Silver-man, "Efficiency  of Filter
      Paper and Impingers for Chromic Acid Mist  in  Air," Am.
      Ind. Hyg. Assoc.  Quart. 8_,  12-13 (1947).

136.  Ege, J.  R. Jr. and L.  Silverman, "A Stable Colorimetric
      Reagent  for Chromium," Anal.  Chem.  19,  693-694 (1947).

137.  Silverman, L.  and J. E. Ege,  Jr., "A Rapid Method  for the
      Determination  of Chromic Acid Mist  in Air," J. Ind.  Hyg.
      Toxicol. 29_, 136-139 (1947).

138.  Hill, W. H. and M. S.  Johnston, "Determination of
      Decaborane," Anal. Chem. 27.,  1300-1305  (1955).

139.  Kuhns, L. J.,  R.  H. Forsyth,  and J. F.  Masi,  "Boron Hydride
      Monitoring Devices Employing  a Triphenyltetrazolium Chloride
      Reagent," Anal. Chem.  28, 1750-1752 (1956).

140.  Fristrom, G. R.,  L. Bennett,  and W. G.  Berl,  "Integrating
      Monitor  for Detecting Low Concentrations  of Gaseous Boron
      Hydrides in Air," Anal. Chem. 31, 1696-1697 (1959).

l4l.  DuBois,  L. and J. L. Monkman, "Polarographic  Determination
      of Heavy Metals in Air Samples," Am. Ind.  Hyg. Assoc. J.
      25(5), 485-491 (1964).

142.  Bochavarov, N., M. Karamanova, and  V. Kolarov, "Rapid
      Analysis for Barium by Beta Backscattering,"  Simp. Ispol'z.
      Metod. Mechenykh At. Soversh. Tekhnol.  Protsessov  Proizvod.
      Primen.  Yad.-Fiz. Metod. Anal. Sostava  Veshchestva 1968
      (Pub. 1969), 307-H.

143.  Evans, C. A.,  R.  J. Guidoboni and F. D. Leipziger, "Routine
      Analysis of Metals Using a Spark Source Mass  Spectrograph
      with Electrical Detection," Appl. Spectry. 2_4_, 85-91 (1970).

144.  Brown, R. and  P.G.T. Vossen,  "Spark Source Mass Spectro-
      metric Survey  Analysis of Air Pollution Particulates,"
      Anal. Chem. .42, 1820-1822 (1970).

1*15.  Bratton, W. D. and C.  H. Wood, "Mass Spectrometric Studies
      of Relative Sensitivity Coefficients in Nickel- and Iron-
      Base Alloys,"  Appl. Spectry.  2^, 509-513  (1970).

146.  Williamson, T. G. and W. W. Harrison, "Comparison  of
      Activation Analysis and Spark Source Mass  Spectrometry
      for Forensic Applications," Nat. Bur. Stand.  Spec. Publ.
      No. 312(1), 283-287 (1969).
                               380

-------
      Weitman, J., N.  Doverhog,  and S.  Farvolden,  "New Boron
      Analysis Method," Nucl.  Appl. Technol £(3),  408-415  (1970).

148.   Hedley, A., "The Determination of Trace Amounts  of Chromium
      by Stable Isotope Dilution-Mass Spectrometry," Analyst  93,
      289-291 (1968).

149.   Perie, M. and M. Chemla, "Mass Spectrometric Determination
      of Boron by Isotopic Dilution," Advan.  Mass  Spectrom.  3,
      585-90 (1966).

150.   Terlouw, J. K.,  J. J. DeRidder, W. Heerma,  and G.  Dijkstra,
      "Mass-spectrometric Determination of Metal  Chelates.
      I.  Quantitative Determination of Copper at  the  Nanogram
      Level," Fresenius' Z. Anal.  Chem. 249(5), 296-301  (1970).

151.   Booker, J. L., T. L. Isenhour, and R. E. Sievers,  "Rapid
      Mass-spectrometric Determination of Chromium as  Chromium(III)
      Hexafluoroacetylacetonate,"  Anal. Chem. _4l,  1705-1707  (1969).

152.   Morris, E. D. Jr. and H. Niki, "Chemiluminescent Reactions
      of Iron and Nickel Carbonyls with Ozone," J. Am. Chem.  Soc.
      92_, 5741-5742 (1970).

153-   Kalinichenko, I. E., "Determination of Nanogram  Amounts of
      Manganese by Measurement of  Chemiluminescence,"  Ukr.  khim.
      Zh. 3_5(7), 755-757 (1969).

15^.   Babko, A. K., N. M. Lukovskaya, and L.  I. Dubovenko,
      "Chemiluminescent Method of  Analysis,"  Sobrem. Metody
      Analiza, Metody  Issled.  Khim. Sostava i Stroeniya  Veshchestv,
      Akad.  Nauk SSSR, Inst.  Geokhim. i Analit. Khim.  83-95
      (1965); C.A. 65, 10397e  (1966).

155-   Braman, R. S., D. D. DeFord, T. N. Johnston, and L. J.  Kuhns,
      "A Coulometric Borane Monitor," Anal. Chem.  32,  1258-1262
      (I960).

156.   Hughes, J.D.H. and G. T. Rogers,  "High-Resolution  Auto-
      radiography of Trace Boron in Metals and Solids,"  J.  Inst.
      Metals 95_, 299-302 (1967).

157.   Sunderman, F. W., N. 0.  Roszel, and R.  J. Clark, "Gas
      Chromatography of Nickel Carbonyl in Blood  and Breath,"
      Arch.  Environ. Health 16_(6), 836-843 (1968).

158.   Brief, R. S.a F. S. Venable, and R. S.  AJemian,  "Nickel
      Carbonyl:  Its Detection and Potential  for  Formation,"
      Am. Ind. Hyg. Assoc. J.  2j6:72 (1965).

159.   Voll'berg, N. Sh., "Determination of Trace  Amounts of  Nickel
      Carbonyl in Air," U.S.S.R. Patent 1353684 (Feb.  15,  1962).
                               381

-------
                  APPENDIX VII
      OPERATIONAL AND MAINTENANCE MANUAL
RF AND ARC EXCITED EMISSION SPECTROSCOPY SYSTEM
                      382

-------
1.
DESIGN CONCEPT
     A relatively  simple  grating monochromator and optoelectronic
detection system was  coupled with two types of excitation  sources
- a radio-frequency  (RF)  generator and an AC arc discharge
system - to form two  test versions of a Be/Cd monitor based on
spectroscopic measurement of optical emission from beryllium and
cadmium.  Two approaches  were devised for producing the RF exci-
tation.  Instrumentation  for both RF systems and the AC arc
excitation mode are described in the following subsections.

     The basic configuration of the spectroscopic systems  was
the same regardless of  the excitation source, and the general
schematic for the  configuration is shown in Figure 1.
   RFor AC Arc
   Power Supply
             RF or AC Arc
            Excitation Zone
                         Reference
                                  Band Pass Filter

                                  Photomultipliers
        Figure 1.   Schematic of Be/Cd Detection  System
                   Based on Radio-frequency  Excitation.
                               383

-------
     Spectral emission from Cd and Be excited by the RP or AC
arc induced discharge is isolated by a diffraction grating mono-
chromator.  The emission intensity of the spectral lines, which
is a function of concentration, is measured by a photomultiplier
detector and ancillary electronic read-out.  A Jarrell-Ash
quarter meter monochromator is used as the dispersive element in
this arrangement (Figure 1).  To ensure a high degree of source
stability, a small fraction (^1/8) of the radiant beam emanating
from the discharge zone is split off, a spectral band isolated
by a band-pass filter, and the intensity of the band monitored
by a photomultiplier detector.  Small perturbations in the source,
arising from'discharge instability, etc., are "nulled-out"
electronically by incorporating this type of reference channel
in the electro-optical segment of the detector.  This is accom-
plished by processing the signal from the reference channel and
the analytical channel(s) through a differential amplifier.
The optical band-pass of the reference beam is selected on the
basis that it reflects a strong dependence on salient discharge
parameters such as flow, pressure, power input, etc.  Small
changes in source intensity, consequently, are detected and
subtracted from the analytical channel.


2.   INSTRUMENTATION

2.1  Lepel Radio-frequency Generator

     The Lepel generator is a vacuum tube type instrument designed
to deliver an appropriate inductively coupled load 2.5 kilowatts
(KW) of megacycle RP energy.  The following specifications apply
to the generator:

     Power Output              2.5 KW

     Operating Frequency       15-50 MHz, continuously variable
     Input Voltage             230 VAC single phase

     Full Load Current         25 amperes

     Power Input               5.75 KVA
     Cooling Water Pressure    Min. 30, Max. 60 PSI

     Cooling Water Flow        Min. 2.0, Norn. 3.0 GPM

     Cooling Water Tempera-    Max. 86°F
     ture at Inlet

     Dimensions                31-1/2" w. x 24" h. x 19-1/2" d.

     Weight                    270 pounds

-------
     The manufacturer's instruction manual Model No. T-2.5-1-
MC2-BW, Serial No. 7228J, Type No. T-252-53 is included with
hardware returned to EPA under contract terms.

2.2  Drake RF Transmitter and ETO Linear Amplifier

     The Drake RP transmitter is a typical tube-type "ham" radio
operator transmitter which offers selectable single sideband,
semi break-in CW, and controlled carrier AM transmission.   The
unit provides 200 watts PEP input on SSB and AM and 200 watts in-
put on CW, which is ample power to drive the ETO linear amplifier,
The following specifications apply to the Drake transmitter and
to the ETO linear amplifier.


     2.2.1  Drake Transmitter

     Power Output              200 watts PEP in SSB and AM modes
                               input power to final stage
     Operating Frequency        3-5- 4.0 MHz
                                7.0- 7.5 MHz
                               14.0-14.5 MHz
                               21.0-21.5 MHz
                               28.5-29.0 MHz, all crystal
                                              controlled
     Input Power               Supplied by Drake AC-4 power
                               supply  120/240 VAC  50/60  cycles
     Pull Load Current         Approximately 5 amps
     Cooling                   None required

     Dimensions                10-3/4" w. x 5-1/2" h. x 11-5/8" d
     Weight                    14 Ibs 1 oz.

     2.2.2  ETO Linear Amplifier

     Power Output              Input power to final stage  is
                               SSB mode:  3 KW PEP continuous
                               CW mode:   1.25 KW continuous

     Operating Frequency       3 to 30 MHz continuously variable
                               over six ranges

     Operating Power           115/230 VAC  50/60 Hertz
     Requirements              single phase

     Full Load Current         Approximately 15 amperes/230 VAC
                               or 30 amperes/115 VAC
     Cooling                   A built-in fan is used for  air
                               cooling

     Dimensions                17" w. x 9-1/2" h. x 18" d.
     Weight                    70 pounds

                              385

-------
     The manufacturer's instruction manuals are included with
hardware returned to EPA under contract terms.

2.3  AC Arc Discharge System

     The AC arc discharge system consists of an asbestos cell
containing the discharge electrodes powered by an external high
voltage power supply.  A circuit diagram for the AC power supply
appears in Figure 2.

     A detailed description of the power supply is as follows:

     Variac                    115 VAC input  10A max.
                               0-140 VAC output

     Luminous Tube Transformer
        Primary input          115 VAC
        Secondary output       15 KVAC
        Maximum current at     60 mA
          secondary
        Power output           0.9 KVA

     Capacitor                 0.001 yF at 50 KVDC or similar
                               with at least 25 KVDC voltage
                               rating

     The Webb cell itself appears in Figure 3-   Webb's original
cell was constructed using "Sindayno," a British trade name
for a high density asbestos.  The MRC cell patterned after the
Webb cell was made of "Electrobestos," a dense asbestos used
for high voltage industrial welding and other purposes.  Both
types of asbestos are formulated to absorb a minimum of water,
which would raise the conductivity and not permit the use of
high voltages in forming arcs between the built-in electrodes.

2.4  Radio-frequency Discharge Tube

     The RF discharge tube system is composed of three concentric
quartz tubes (coolant, plasma, and sample injection) held in spe-
cially designed metal holders machined from 316 stainless steel
and aluminum.  The system was devised to facilitate replacement
of broken or melted discharge tube and is shown pictorially in
Figure 4.  Design specifications of the holder are presented
schematically in Figure 5.  In Figure 4, the upper holder is
typical of the type used with a 38 mm coolant tube, whereas
the lower holder is used with a 20 mm coolant tube.  Both holders
use a 15 mm plasma tube and a 6 mm sample injection tube.

     The entire discharge tube assembly is of a concentric tube
design with three gas inlets.  The center tube functions as the
sample inlet.  A discharge gas (air, helium, argon, nitrogen,
etc.) is admitted tangentially into another tube surrounding
the sample inlet.  Another concentric tube surrounds both the
sample and discharge gas tubes and serves to admit a cooling


                              386

-------
115 VAC
 Line
         Variac
                                                        Electrodes
                   LuminuousTube
                     Transformer
  High
 Voltage
Capacitor
      Figure 2.  Electrical Schematic for AC Power Supply.
                               387

-------
              -Quartz Window
Sample
 I ntake
                                 Machined
                               Sindanyo Block
                                 Electrodes
              Glass Inspection
                  Window
Cone to Take
Filter Paper
   Figure 3.  Diagram of Webb-Type Arc Discharge  Chamber
                            388

-------
Figure 4.  Photograph of Radio-Frequency Discharge  Tube
                          389

-------
uo
VD
O
                            V
                                CD
                                                     »t-O Alt" 5o«J. IW"

                                                     Mkk'd
rl.
i ii
h' H
"F^ > .« ^
•» H»" N. (sf
^r—
•^ * ' *• c^-
•-i1

                                                                                                  *2-',Z MM it«»
                                           5 ',
                                                         0.150" 0.0


                                                         0.3S*- 0.0.
DtH£i«SIONS ARE IN INCHES
TOLEBANCC*
DECIMALS   FRACTIONS
MONSANTO RESEARCH CORPORATION
     DAYTON I
                                                                                    SCALE   WT
                                                                                                   SHEET  Of
                          Figure 5.   Schematic Showing Design  Specifications for
                                       Radio-frequench  Discharge Tube Holders

-------
gas to the system.  This coolant gas is also admitted in a
tangential manner.  The sample and discharge gas tubes terminate
within the main body of the discharge tube and the sample tube
is also shortened so as to terminate within a few millimeters
of the end of the discharge gas inlet tube.  A two to four turn
water-cooled induction coil surrounds the open end of the main
discharge tube and is powered by the Lepel 2.5 KW high frequency
generator operating at frequencies of 15 to 50 MHz.  Stabiliza-
tion of the discharge is attained by vortex gas flow.  The tube
radius required is highly influenced by the skin effect peculiar
to RF induction heating.  The energy is dissipated in a thin
layer so that the radius should not be much greater than the
thickness of the coupled layer.  By operating at relatively high
frequencies (^30 MHz) and controlled gas flows, a toroidal shaped
discharge may be formed.  The cooler core provided by this geom-
etry presents far less resistance to particulate injection into
the inner regions of the discharge than the more uniform discharge
geometry associated with lower frequency operation.  Figure 6
illustrates this concept.

2.5  Monochromator and Photometric Detection System

     2.5.1  Jarrell-Ash 0.25-Meter Monochromator

     A 0.25-meter Jarrell-Ash Model 82-410 monochromator was
used as the wavelength separation device for radiation emitted
from the RF or AC arc excited materials.  It is a low cost,
small size, lightweight instrument suitable for portable or
field use.  The specifications are as follows:
     Focal Length
     Linear Dispersion
     Aperture Ratio (speed)

     Gratings (two supplied)



     Gratings Blazed at
     Resolution (half-band
        width at 313.1 nm)


     Slits
     Wavelength Coverage
0.25 meter Ebert design
3-3 nm/mm with 1180 grooves/mm
   grating
1.65 nm/mm with 2360 grooves/mm
   grating

f/3-5
Ruled area - 64 mm x 64 mm
Replicas - 1180 grooves/mm
           2360 grooves/mm

300 nm and 600 nm
Better than 0.3 nm in First Order
with 2360 grooves/mm grating with
150y slits

Two 150 micrometer slits, fixed
0 to 900 nm
     The instrument manufacturer's operating manual is included
with the hardware returned to EPA under the terms of contract.
                              391

-------
                        o
                        c
                Sample
               Particles
   ~3-4MHz
 Low Frequency
  -30 MHz
High Frequency
Figure 6.   Particulate Entrainment Into
           Uniform vs. Toroidal Discharge
                  392

-------
     2.5.2  Photometric Detection System

     Two RCA 1P28A photomultiplier (PM) tubes were housed in
Pacific Photometries Model 50F PM tube housing.  One PM tube
served as the sample tube and the other was the reference tube,
as briefly described under the design concept in Section 1.
A Bertran Model 602-11N power supply supplied the high voltage
to the PM tube dynode chain.

     An MRC designed and fabricated electrometer amplifier was
used to monitor the PM tube currents.  The amplifier was a dif-
ferential, dual-channel design arranged to accept two PM tube
inputs, one due to a sample and one to a reference, and to sub-
tract the reference from the sample before display on a meter
or optional strip chart recorder.

     The sample channel incorporated a switch selectable range
of input currents of 10~9 to 10~5 amps full scale in decade
steps.  Intermediate steps XI, X2, and X5 were also provided
for each decade range.   The reference channel had a fixed range
of 0 to 10 6 amps full scale, and a 10-turn potentiometer at its
output allowed subtraction of any current from 0 to 10~6 amps
from the sample channel.  A manual zero control (10-turn
potentiometer) allowed adjustment of the electrometer circuitry
to zero output with no current inputs.

     In addition to real time spectral measurement capabilities,
the electrometer had a built-in integrator allowing accumulation
of PM tube currents over a manually controlled time period.

     An auto zero circuit allowed zeroing the electrometer out-
put in the spectral mode when observing only the background
of an RP plasma or AC arc discharge with both the sample and
reference PM tubes.  With no material of interest (Be or Cd)
undergoing emission, both PM tube channels should be equal and
the electrometer output should be zero.  When operating in the
integral mode, the auto zero was set first in the spectral mode.
Then an "integral zero" switch position allowed manual resetting
of the integrator.

     The time constant  of the electrometer could be chosen and
an appropriate low leakage capacitor used in the final amplifier
stage of the electrometer.  A front-panel meter indicator served
to show the electrometer output, and a back panel recorder jack
was provided to allow for permanent recording using a 1-mV input
strip chart recorder.
                              393

-------
          2.5.2.1  Specifications MRC Signal Processor Unit

Signal Channel Input Current Sensitivity:   For nominal 1.0 volt
full






scale signal output.
Range
10'5
10"6
ID'7
10"8
10"9
Manual Zeroing Control: ±100$

Amps, Full Scale
0-10~5
0-10~6
0-10"7
0-10~8
0-10~9
of full scale, i:
selected range.

Range Signal Channel:  (When activated.)  0—2 x 10~6 amps
full scale; independent of range selected; adjustable from
0-100/8 of total available reference signal.

Automatic Baseline Correction:  ±100/8 of selected range.

Signal Output Modes:

     Spectral - Output signal voltage is linearly proportional
to input signal current, with amplification factors of XI, X2,
X5.  Range indication is correct on XI factor, increasing by
X2 and X5 on other mode switch ranging.

     Integral - Output signal voltage is time integral of
input signal current.  Increased sensitivity factors of X2
and X5 switch selectable.

Signal Output Readout:  An adjustable voltage output signal
for connection to any external potentiometric strip chart
recorder with a span of 1-100 millivolts full scale may be
used (output signal is positive signal with respect to ground,
common).

2.6  Calibration System

     2.6.1  Cadmium Vaporizer Probe

     Calibration of the RF plasma system is based on the fact
that cadmium is a relatively volatile metal.  The vapor pressure
of cadmium is about 1 mm of mercury at its melting point, 321°C.
The metal vapor is normally generated at about l60°-200°C.
                              394

-------
     The cadmium metal is placed in a chamber constructed of
black iron pipe and fittings.  A Variac-controlled heating tape
wrapped around the chamber with sufficient asbestos tape to
insure even temperature control is monitored with a mercury
thermometer.  Inert gas, such as argon or nitrogen which contains
no oxygen, is passed over the heated cadmium metal and carries
the metal vapor out.  The flow rate of the inert carrier gas must
be low so that the metal vapor generating system is near equi-
librium.  An argon flow rate of 2.2 1/min with the cadmium metal
at 160°-200°C gives 0.38-6.45 yg/m3.  This is a reasonable amount
of cadmium vapor for calibration.

     Attached to the exit of the vapor generating chamber is a
heated injection probe.  Its innermost tube, constructed of stain-
less steel, carries the cadmium metal vapor into the center of
the plasma excitation region.  Surrounding the innermost tube
are two outer concentric copper tubes designed to circulate a
cooling fluid through the entire probe assembly.  Without the
cooling fluid, the plasma would soon melt the probe tip.

     Glycerol (glycerin, 1,2,3-propanetriol) is a satisfactory
circulating cooling fluid.  The injection probe must be main-
tained at a higher temperature than the temperature at which
the cadmium metal vapor is generated, or else condensation in
the stainless tube will soon plug it completely.  The glycerol
can be circulated at 220°-270°C with little difficulty.  The
glycerol circulation system is a heated constant temperature
bath with a stirrer, a small centrifugal pump, and the necessary
1/Jj-in. copper tubing and fittings.  Swagelok® fittings facili-
tate assembly and disassembly, and a portable hand-held heat gun
and rubber pipet suction bulb may be necessary to initiate the
flow of glycerol when starting a room-temperature system.  The
glycerol is very viscous at room temperature, but at about 150°C
or more the viscosity approaches that of water and is no longer
a problem to circulate through the small passages of the injec-
tion probe.  A schematic of the entire assembly is shown in
Figure 7 and the injection probe subsystem is shown pictorially
and diagrammatically in Figures 8 and 9, respectively.

     2.6.2  Beryllium and Cadmium AC Arc Discharge Calibration
            Assembly

     To calibrate the AC arc discharge system, a beryllium or
cadmium alloy wire is used as an electrode in a discharge cell
placed above the analytical discharge cell.  Metal vapor gener-
ated by the alloy electrode passes into the analytical discharge
cell for excitation and system calibration.  A gas filter (mil-
lipore type BDWB 0^700) below the analytical cell is used to
confirm the concentration of metal in the calibration vapor by
atomic absorption spectroscopy.  A schematic of the system is
shown in Figure 10.
                              395

-------

          _ Quartz Plasma Tube

          - RF Plasma

          _ Cadmium Metal Vapor
             Load Coil
            -Injection Tube
                                                  Constant
                                                 Temperature
                                                 Glycerol Bath
                                                 220 - 270°C
                Inert Carrier Gas In
            Cadmium Metal
Heated Chamber
Figure  7.   Diagram Showing  Complete  Cadmium Vaporizer
             System  for  Injecting  Cadmium Metal  Vapor
             Into a  Radio-frequency Plasma.
                              396

-------
Figure 8.  Photograph of Cadmium Vaporizer Probe for
           Injecting Cadmium Metal Vapor Into a
           Radio-frequency Plasma.
                         397

-------
                     Cadmium Vapor Out
  I t

:G
                                   Cooling Fluid In

                                   Cooling Fluid Out
                          t
                     Cadmium Vapor In
Figure  9.   Diagram of Cadmium Vaporizer  Probe for
            Injecting Cadmium Metal Vapor Into a
            Radio-frequency  Plasma.
                            398

-------
                              Clean Air In
          BeorCd Alloy
       Generating Electrode
    (OnevsPHror Both Alloy)
               Pt- Ir
             Analytical
             Electrodes
            Membrane Filter -
                                    Electro-optical
                                    Measurement
                                       System



Cast Pump



Dry Gas Meter



Absolute Filter
Figure  10.
Diagram of Beryllium or Cadmium Calibration System
Using  Electrical  Discharge  to Generate Aerosols.
                                 399

-------
     The beryllium electrode is 3/32-inch in diameter and is a
97% copper-35? beryllium alloy.  The cadmium electrode is a silver
brazing alloy rod #14 B&S wire gauge and containing 18% cadmium,
2656 copper, 35% silver, and 215? zinc.  The gap between generating
electrodes can be 1 mm up to 10 mm with the wider gap requiring
more voltage for initiation of the arc resulting in a hotter arc
with more metal vapor generated.  The power supply to operate
the calibration electrode system is the same as the power supply
to operate the analytical electrodes.


3.   OPERATION PROCEDURES

     Operation of individual subsystems are shown in respective
manuals included with the deliverable hardware.  Specific opera-
tional procedures are presented in the following subsections.

3.1  Method of RF Plasma Start-up

     The RF plasma is impossible to initiate unless pure argon
gas is used.  The power control knob R3 (see Lepel manual, page
2-1) is set to its minimum position.  However, it is not necessary
to start the argon flow at this time.  Push the start button,
S3, and note the grid current (Meter Ml) and plate current
(Meter M2).  When the power knob setting is slightly increased
(do not exceed 105? power), both the grid and plate currents
should increase simultaneously in a ratio of grid to plate of
about 1:3 to 1:7.  If grid and plate currents do not increase
together (the plate current overload switch may automatically
engage if current on the plate exceeds 830 ma; if it does
engage, there is a 45-second delay before the start button S3
can be reengaged), adjust the grid and/or plate tank controls
located at the right side of the generator.  If the grid current
decreases when the power is increased, turn the grid tank con-
trol to increase the grid current, or decrease the plate tank
control.  In order to operate at any specified RP frequency,
an RP receiver is necessary to monitor the grid and plate tank
control positions.

     Once the grid and tank are properly adjusted, start a flow
of argon gas of about 20 to 25 liters per minute in the plasma
tube and increase the power control to about 255? (but do not
exceed the grid or plate current maximum shown on the respective
meters, 250 and 830 ma, respectively).  The plasma will not
self-initiate.  One must use a Tesla coil held to the base of
the quartz tube, or use a carbon rod (1/4-in. diameter and
several inches long, similar to those used to make sample holders
for arc spectroscopy) suspended by dry, nonconducting string
down into the plasma tube in the region of the load coil.  Be
careful!  Do not allow the Tesla coil to come near the load coil,
and do not touch the carbon rod.  RF energy is, like electric
current, deadly if improperly respected or used.  As an added
                              1JOO

-------
precaution, do not touch any grounded objects when initiating
the RF plasma by either of these methods.

     Both methods of plasma initiation involve generation of free
electrons in the plasma gas (argon).  Once even a few free elec-
trons are generated, and provided the generator and tuning and
power controls are properly adjusted, an argon plasma is easy to
initiate.

     One must initiate the plasma with argon, but once initiated
another gas such as air, nitrogen, etc., can be gradually added
to the argon as the argon flow is decreased.  Watch the grid
and plate current meters and adjust the grid and tank controls
accordingly as the new gas is added.  Each plasma gas has a
characteristic impedance and must be individually matched to
the output impedance of the RP generator.

3.2  Procedure for Generating Cadmium Calibration Vapor

     In order to generate cadmium vapor to calibrate the RF
plasma, turn on the heating control for the vapor generating
pot and turn on the heating unit in the glycerol constant tem-
perature bath.  Do not turn on the inert carrier gas at this
time.  Once the glycerol is warm enough that its viscosity
lowers to approach that of water, turn on the circulating pump
and use the pipet suction bulb (at the exhaust end where glycerol
is returned to the constant temperature bath) to start glycerol
flowing through the injection probe.  The portable heat gun
should be used whenever necessary to warm the glycerol in the
probe, pump, or lines and reduce its viscosity.  Once glycerol
begins flowing, it will not cease unless allowed to cool until
it becomes too viscous (maintain about 150°C or more).  Once
the glycerol is above the temperature of the cadmium vapor
generating pot, the inert carrier gas may be turned on.  Main-
tain the temperature of the glycerol above that of the cadmium
metal vapor generating system or else metal will condense in
the injection tip passage and block it.

3.3  Procedure for Generating Beryllium or Cadmium Calibrating
     Aerosols With Arc Discharge System

     Attach the two asbestos cells, the membrane filter, the gas
meter, and the Gast pump as shown in Figure 10 (Section 2.6.2).
Attach the power supplies to both the aerosol generating and the
analytical electrodes.  Turn on both electrode systems and the
Gast pump simultaneously by using a multi-outlet junction box
with a single on-off switch.  Measure the volume of gas using
the gas meter and measure the amount of beryllium or cadmium
aerosol generated by independently analyzing the material trapped
on the membrane filter by atomic absorption spectroscopy.  This
calibration system should be operated for about ten minutes in
order to collect sufficient material for analysis by AAS.

-------
3.4  Operation of Monochromator and Photometric Detection System

     Refer to the Jarrell-Ash manual for details of operation of
the 0.25-meter Ebert monochromator supplied with the deliverable
hardware.  The physical placement of the emission source, signal
PM tube, reference PM tube, PM tube high voltage power supply,
monochromator, and electrometer is shown in Figure 1 (Section 1).
A detailed operations manual for the MRC designed and fabricated
signal processor unit will be found in the following sections.

     3.4.1  Signal Processor Unit

     The signal processor unit contains all of the electronic
circuitry required to process and interface the detector with
the readout device.  It consists of a basic main frame section
where all interconnections and control functions tie together
the modular, plug-in, solid-state, printed circuit cards that
accomplish the basic signal processing functions.

     The processor is fabricated in a double shielded enclosure
with adequate filtering of all input and output leads to permit
operation in high power, ambient, radiofrequency fields developed
by the RF excitation source.  This heavy shielding is required
to prevent RF saturation of highly sensitive electronic circuitry
within the unit.

     Basic circuitry to process the detector signal to a usable
readout signal is contained on plug-in printed circuit modules.
Descriptions of the five modules:

     (a)  Signal Electrometer Module (Model 9900.03)
     (b)  Reference Electrometer Module (Model 9900.0*1)
     (c)  Spectral Amplifier/Integrator and Automatic
          Baseline Correction Module (Model 8901.01)
     (d)  Amplifier Power Supply Module (Model 1100.01)
     (e)  Central Power Supply Module (Model 1200.01)

are presented in the following subsections.  Schematics showing
the electronic circuitry of the modules appear in Figures 11-16.

          3.4.1.1  Signal Electrometer Module (Model 9900.03)

     This module functionally transduces the signal detector
output current to a voltage output which can then be differen-
tially combined with three (3) other possible signals obtained
elsewhere.  These signals are the automatic baseline correction
(auto zero), the background (reference) signal, and the manual
zeroing signal (MZ).  The circuitry consists of an electrometer,
i.e., current to voltage transducer, A-200, in conjunction with
range resistors R-101 through R-105 selected by range switch
SW-101.  Full scale current sensitivities of 10~5 to 10 9 amps
                              402

-------
o
LO
          Figure  11.   Electronic  Circuit  Schematic  -  Signal  Processor Unit - Main Frame

-------
o
-t
                                      KMB-^300.03
                                        f]               J?	'
                                         ,ilrnrlrivlir
                                                                                 •HHT  OF
                                                    Si OTHERWISE SPECIFIED

                                                   DWOWONI ARI IN INCHES

                                                   TOUMANCOi

                                                    IMAUI   PHACT1OI
                                                   FINISH   _.. ._
                                               MONSANTO RESEARCH CORPORATION
                                                   DAYTON I
                                                                      «r. |WT^!
            Figure 12.
Electronic  Circuit Schematic - Signal  Electrometer Module
(Model  9900.03)

-------
o
VJ1
                                        "flrmr rirRF
_J
                                                     UNLESS OTHERWISE SMC If I CO
                                                     (NKEttSIONS »RE IN INCHES
                                                     TOLERANCES:
                                                     DECIMAL*
                                                     ATCBIAL	^	
                                                                                  SHE.ET  OF
    MONSANTO RESKARCH CORPORATION
         DAYTON I
          UuMTON. C
                                                                             T
                                                                                   SHEET  OF
        Figure 13.  Electronic  Circuit Schematic - Reference  Electrometer Module  (Model  9900.01)

-------
J5T
O
cr\
                                                             ^^    J
                                                                                 •HKTT  OF
                                                   Nstoss 'if. IN rwcvrs
                                                  TOLERANCES
                                                  DECIMALS   FRACTION*
                                                   .XX -«   *
                                                   -XXX-*   ANGLES
                                                   XXXX BA1
                                                                       MONHANTO RESEARCH CORPORATION
                                                                    SCALE :v 73 WT"
                                                                            _L
                                                                                  HET	OF
       Figure 14.
                                                  *
Electronic Circuit  Schematic - Spectral Amplifier/Integrator  and
Automatic  Baseline  Correction Module (Model  8901.01)

-------
                                                                                   SHEET   or

                                                 mwtriStCNS »RF IN INCHES
                                                 TOLERANCES
                                                  CIMAL*   FRACTION!
                                                  XX -*   *
                                                  XXX -t   ANGLE*
  -!«r=g	

Figure 15.

                                                                         MONSANTO RESKAHCH CORPORATION
                                                                    SCALE *T> WT    | 3?/-a'o" T*

                                                                                    SHEET
                                  I                       A
Electronic Circuit  Schematic  - Amplifier  Power Supply Module  (Model  1100.01)

-------
-tr
O
00
                                                            BC IN INCHES
                                                        TOLERANCES
                                                        DCCIMALS  FRACTIONS
                                                         XX - *   *
                                                         XXX - *   ANGLES
MONSANTO RESKAHCH CORPORATION
     DAVTOM LAKMtATMY
      BATTOM. OHIO
                                                                             E * r> |WT *£?
                                                                                          SHEET   OF
       Figure 16.   Electronic  Circuit  Schematic -  Central  Power  Supply  Module (Model 1200.01)

-------
per volt output are selected by the range switch.  Amplifier
A-210 provides a voltage gain >1 for the electrometer output
voltage signal.  Passive output attenuation P-103 through P-107
provide for exact calibration of each range selected.  The other
previously mentioned signals are applied differentially to A-210.
The reference signal obtained from a similar electrometer circuit
can be adjusted to cancel out continuum background where it
exists, leaving only the desired signal.  The manual zero can
be adjusted to cancel out static dark current in the photomulti-
plier detector.  The automatic baseline correction can be used
to account for shifting or drifting of any of the above signals
and can be activated at any desired time by the operator.  The
auto baseline correction (ABC) can be disabled, if so desired,
as can the reference signal by operation of switches located on
the internal subchassis.  Overall, the circuit is noninverting,
polaritywise.

          3.^.1.2  Reference Electrometer Module (Module 9900.04)

     This module functionally transduces the reference detector
output current to a voltage output which is one of the possible
differential signals applied to the signal channel electrometer
module.  It is a single range unit with individually adjustable
output voltage for each range.  This offers high versatility
in the application of this signal.

     Amplifier A-300 is the basic electrometer, cascaded with
gain block A-310.  The circuit overall is an inverting configu-
ration so that differential action can be obtained in A-210 on
the signal electrometer module.

          3.4.1.3  Spectral Amplifier/Integrator and Automatic
                   Baseline Correction Module (Model 8901.01)

     This module functionally accepts the resultant output sig-
nal from the signal electrometer module and processes it to the
best usable form depending on the type of sampling technique
used.  Continuous sampling suggests use of the spectral mode.
Batch, or spot, sampling technique can use the spectral opera-
tional mode, but a possible better mode is the integral mode
of operation.  The output signal voltage is the time integral
of the sample concentration equivalent detector current.

     Switching is incorporated to select the spectral or integral
mode.  This switching further provides gain enhancement of
amplifier/integrator Q-110/A-110, by factors of X2 or X5 over
the basic current sensitivity obtained with XI operation, either
in the spectral or integration modes.  All mode switching func-
tions are obtained by SW-102.  Relay K-102 provides for resetting
the integrator circuit to zero conditions.  The output signal can
be connected to an accessory strip chart potentiometric recorder
                              409

-------
with any full scale span of 1 to 100 millivolts.   Recorder span
is adjusted to the unit by attenuator P-118.   The output signal
is positive for current into the signal electrometer input.

     It is highly desirable to have a zero output voltage signal
with no sample at the detector.  An additional circuit contained
on this module is the automatic baseline corrector circuit, pro-
viding a zero output voltage signal with no sample present.  This
is also a requirement of the input signal for the integrator
circuit, i.e., to prevent false integrals, input  voltage other
than zero must be cancelled out.  The circuit to  accomplish this
feature consists of Q-100/A-100 forming an additional integrator
circuit.  This circuit is activated whenever panel switch SW-103
is placed in the AZ (auto zero) position, energizing relay K-101.
When K-101 is energized the ABC integrator is connected in a
loop-feedback circuit and will integrate in such  a way as to
drive its own input back to zero volts.  This redrive signal is
one of the differential signals applied to the signal electrometer
module.  The circuit will follow variations always keeping its
input at zero volts.  When SW-103 is activated to the data posi-
tion (sample also introduced) K-101 de-energizes  and the ABC
integrator holds the value of voltage required to zero the unit
in memory.  The sample signal now is connected to the spectral/
integrator amplifier input.  Only that signal voltage value over
and above background signal and dark currents is  then processed,
either in spectral or integral form.

          3.4.1.4  Amplifier Power Supply Module  (Model 1100.01)

     This module provides dual, highly regulated, voltages to
the electrometers and integrated circuit amplifiers on the other
modules.  Good regulation is required to prevent  circuit drift
and instability.

          3.4.1.5  Control Power Supply Module (Model 1200.01)

     This module provides regulated voltage for operation of the
relays in the unit.

          3.4.1.6  Performance Monitor

     A panel meter is provided to give readout of operating
voltages for the unit as well as signal output voltages.


4.   MAINTENANCE/ADJUSTMENTS/CALIBRATION OF MRC SIGNAL
     PROCESSOR UNIT

     The following section contains instructions  for unit main-
tenance.  Maintenance and adjustments for calibration should
only be performed by personnel thoroughly familiar with the
technique involved.  This requires an experienced electronic


                               410

-------
technician familiar with analog test procedures.  Adjustments
on the analytical subsystem requires personnel experienced in
photometric techniques.

     For performance to meet specifications, a logical stepwise
procedure must be employed to test or troubleshoot problem per-
formance.  The following sections are outlined in this form.

4.1  General Maintenance Suggestions
     The electronic circuitry provided has some inherent testing
capability which can be used for quick checks on portions of the
instrument operations.  These are of use both for the operator
and for the maintenance technician.  These tests can be performed
at any time.  It is good practice to record analog meter test
readings.  If the instrument parameters after initial specified
operation are confirmed, a basis is then available for future
comparisons.  As with all electronic instruments, some deviations
will occur with aging and environmental conditions such as shift-
ing ambient temperature.  Small deviations can be cancelled by
adjustments with front panel controls.  Gross deviations gener-
ally indicate a circuit failure or improper operating technique.

     Troubleshooting for circuit failures should only be per-
formed by knowledgable, experienced personnel.  Recalibration
should not be attempted except after thorough study of the
instrument .

     Repairs on the printed circuit boards require the use of
a "solder sucker" type iron to remove defective components.
The wattage rating should not exceed approximately 50 watts.
Replacement parts should be resoldered with a small tipped iron
not to exceed 50 watts dissipation.  The holes in the P.C. board
should be completely free of old solder before the new component
is installed.  Do not force leads through the holes.  Do not
overheat the component or circuit foils when resoldering.  Do
not use excessive solder.  Use only 60/^0 tin/lead rosin core
solder for repairs.  Use of heat sink clips is recommended in
all cases possible.  Residue flux can be removed with a brush
and toluene.  Avoid excessive application of toluene as compo-
nent markings may be destroyed.

     One final general word of caution:  Avoid touching the
range switch, the range resistors, and the electrometer circuit
boards, as well as the components and the P.C. board in the area
of the summing points of the signal integrator/amplifier and the
automatic baseline correction circuit.  Fingerprints or gener-
ally dirty conditions can cause these circuits to perform outside
of specifications.

-------
     Detailed troubleshooting and recalibration require the use
of external test equipment of known accuracy and precision.  The
following equipment is recommended:

1.   Digital multimeter capable of resolution to 1 mV and having
     high input impedance.  (Almost any "name brand" meter on
     the market meets these requirements).  An analog meter may
     be used, but they generally lack the degree of readout
     resolution desired.

2.   A variable "voltbox" capable of output range of 0 to
     ±1000 yV minimum range and having a resolution of less than
     100 microvolts.  Another desirable feature is the inclusion
     of a range resistor(s) for electrometer testing.

3.   Laboratory grade potentiometer with a resolution of less
     than 10 microvolts.  While not required, it is useful for
     electrometer tests especially if the "voltbox" is not
     available.

4.2  Generalized Troubleshooting Procedure

     Efficient troubleshooting of operational problems should
follow the generalized steps listed below.  No description of
troubleshooting procedures can give complete coverage of all
possible problems and tests, but they can serve as a guide to
assist maintenance and operating personnel.  Familiarity with
overall circuit functions is paramount for effective mainte-
nance.  Most malfunctions in this type of analog circuitry
manifest themselves as saturation at circuit voltage limits,
either positive or negative, or as drifts in output voltage
levels with time.

A.   Localize the Problem Area - Performance Test Meter Monitoring

     1.  Disconnect the signal and reference input cables on
         the rear panel of the processor unit.

     2.  Rotate the test meter switch through its positions
         and note the presence, absence, or deviations from
         normal readings of the power supplies.  Compare the
         readings with those obtained when unit was opera-
         tional.  Deviation from these normal readings indicate
         the possible problem area.  Troubleshoot any area
         where a malfunction is indicated.  If no malfunction
         is indicated by the meter readings, proceed to step B.

B.   Systematic Test Procedure

     This test procedure can help localize troubles in the
     signal processor unit.

-------
     1.   Set the panel controls as follows:

         a.  Data/Auto Zero         DATA
         b.  Spectral/Integral      Spectral XI
         c.  Panel Meter            B
         d.  Manual Zero            500
         e.  Range Switch           10 6

     2.   Set the internal controls as follows:

         a.  Reference Channel      OFF
         b.  ABC (AZ)               OFF

     3.   Connect test digital voltmeter to TP-1.

         a.  Reading should be 0.000 ± 0.050 volts DC

             (1)  If reading exceeds ±0.050 volts, adjust
                  P-200 on signal electrometer  for zero
                  reading - 0.000 ± 0.010 volts DC

             (2)  If reading is not adjustable  and is approxi-
                  mately ±1^1 volts DC electrometer amplifier
                  A-200 is faulty.

             (3)  If unit is in specification or can be adjusted
                  to specification, proceed to  step 4.

     4.   Connect DVM to arm of manual zero control (TP-9)  and
         adjust control for 0.000 ± 0.010 volts DC

     5.   Connect DVM to TP-2.  Reading should be 0.000 ±
         0.050 volts DC.  If reading exceeds ±0.050 volts,
         adjust P-210 on signal electrometer for 0.000 ±
         0.010 volts DC.

     6.   Set range switch to 10~9.  Reading at  TP-2 should be
         0.000 ± 0.050 volts DC.  If reading exceeds ±0.050
         volts DC, carefully readjust P-200 for 0.000 ±
         0.010 volts DC.

     7.   Set range switch to 10~6.  Connect PVM to TP-5.
         Reading at TP-5 should be 0.000 ± 0.050 volts DC.
         If reading exceeds ±0.050 volts DC, carefully
         adjust P-110 on the amplifier integrator module
         to 0.000 ± 0.005 volts DC.
8.   If all preceding tests are valid, unit should operate
         correctly.

-------
     9.  Rotate manual zero control clockwise.   Panel meter
         should go upscale.  If meter "follows" control,  unit
         is operational from electrometer differential ampli-
         fier through the output amplifier.   Reset control
         for zero output reading on panel meter and recorder.

C.    Input Signal Test

     1.  Leave DVM connected to TP-5.

     2.  Connect negative output of voltbox  or  laboratory
         potentiometer to center conductor of SO-3 through
         1.00 megohm 1% resistor.  Set output to -1.000 volts
         DC  Connect positive output to processor chassis.
         Panel meter should read 50 (1/2 full scale).  Recorder
         should read 1/2 full scale.  DVM should read
         +0.500 volts DC.

     3.  Reduce voltbox or potentiometer input  voltage to 0.000,
         Panel meter, recorder and DVM should be at 0.00.

D.    Test of Reference Electrometer

     1.  Connect DVM to TP-10, panel meter switch to "C".
         Reading should be 0.000 ± 0.050 volts  DC.

         (a)  If reading exceeds ±0.050 volts DC, adjust  P-300
              on reference module for 0.000  ± 6.010 volts DC.

         (b)  If reading is not adjustable and  is approximately
              ±14 volts DC, electrometer amplifier A-300  is
              at fault.

     2.  Connect DVM to TP-7.  Reading should be 0.000 ±  0.050
         volts DC.  If reading exceeds ±0.050 volts DC,
         adjust P-310 on reference module for 0.000 ± 0.010
         volts DC.

     3.  Leave PVM connected to TP-7.

     4.  Connect negative output of voltbox  or  laboratory -
         potentiometer at -1.000 volts DC through 1.00 megohm,
         1/&, resistor to center post of SO-2.  Connect positive
         output to processor chassis.

         (a)  DVM should read 1.000 ± 0.020  volts DC.

         (b)  Panel meter reading will depend on calibration
              setting of P-109 and setting of reference balance
              control P-101.  Reading should be maximum with
              reference balance control completely counter
              clockwise.

-------
     5.  Set reference balance at 1000 count.  Turn SW-107 so
         that reference channel is activated.  Reverse recorder
         lead wires.  With -1.00 voltbox input through 1.00
         megohm resistor still connected, advance reference
         balance control toward 000 count (CCW) .   Recorder
         should go upscale.  Meter "C" will go upscale.  If
         both respond, reference channel is operable.   Dis-
         connect input signal.  Return recorder wires  to
         original configuration.

     This completes the test procedure.  If all responses were
as described, unit is in good operating shape except for possible
calibration errors.

4.3  Unit Calibration
     Calibration requires an accurate source of voltage and the
use of precision resistors and should not be attempted without
complete understanding of the unit circuitry.  The test input
signal used in the troubleshooting procedure 4.1.C gives a par-
tial indication of calibration precision on one range only.
If total recalibration is required, it is suggested that the
user contact Monsanto Research Corporation for authorization
to return unit for calibration.

-------
 BIBLIOGRAPHIC DATA
 SHEET
1. Report No.
       EPA-R2-73-252
4. Tide and Subtitle
  Instrumentation  for Monitoring  Specific Particulate
  Substances in Stationary Source Emissions
3. Recipient's Accession No.
                                         5. Report Date
                                         September 1-973
                                        6.
7. Author(s)  John V. Pustinger, David  A.  Shaw,
         Paul L. Sherman and Arthur D. Snyder
                                        8. Performing Organization Rept.
                                          No- MRC-DA-350
9. Performing Organization Name and Address

  Monsanto Research  Corporation
  Station B, Box  8
  Dayton, Ohio  15407
                                         10. Proiect/Task/Work Unit No.
                                           P.E. No.  1AA010
                                         11. Contract/Grant No.

                                           68-02-0316
12. Sponsoring Organization N'amc and Address

   Office of Research & Development
   U.S.  Environmental Protection Agency
   Washington, D.  C.   20460
                                         13. Type of Report & Period
                                           Covered

                                           Final Report
                                         14.
15. Supplcmcntar> Not* s
16. Abstracts
   This  report describes the development of  an engineering prototype
   analytical and  sampling system capable  of monitoring  continuously
   the  concentration of beryllium and cadmium in typical  stationary
   source emissions.  Data for  laboratory  tests with radio-frequency
   and  AC arc induced emission  spectroscopic systems are  reported.
   Information derived from field testing  of the AC arc  induced emis-
   sion  spectroscopic system at  a power plant is described.   Briefing
   documents describing technology appropriate to the problem of con-
   tinuous measurement of beryllium, cadmium, mercury, lead, arsenic,
   antimony, barium, boron, chromium, copper, manganese,  nickel, and
   vanadium from stationary emission sources are also presented in
   the  appendices.
17. Ke> Words and Document Analysis  17o. Mt script


       Instrumentation
       Continuous Monitoring
       Hazardous Particulate
          Substances
       Emissions
       Emission Spectroscopy
       Radio-frequency  Plasmas


I7b. Identifiers/Open-Ended Terms
                    AC Arc Discharge
                    Beryllium
                    Cadmium
                    Mercury
                    Lead
                    Arsenic
                    Antimony
  Barium
  Boron
  Chromium
  Copper
  Manganese
  Nickel
  Vanadium
   Continuous Monitors,  Hazardous  Particulate  Substances,
   Stationary Sources
i7e. COSATI Fie id/Group Environmental Protection Technology Series
18. Availability Statement

   Release Unlimited
                               19. Security Class (This
                                 Report)
                               	UNCLASSIFIED
                                                 20. Security Class (This
                                                   Page
                                                     UNCLASSIFIED
         21. No. of Pages
            426
                                                 22. Price
FORM NTIS-39 IREV. 3-721
                                                                   USCOMM-DC !4tS2-P72

-------