&EPA
           United States
           Environmental Protection
           Agency
           Industrial Environmental Research
           •Laboratory
           Cincinnati OH 45268
EPA 600 7 V) 1
Juris 19ftO
           Research and Development
On-Line Zeeman
Atomic Absorption
Spectroscopy for
Mercury Analysis in
Oil Shale Gases

Interagency
Energy/Environment
R&D Program
Report

-------
                RESEARCH REPORTING SERIES

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

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

This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series.  Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the  transport of energy-related pollutants and their health and ecological
effects; assessments of, and development of, control  technologies for energy
systems; and integrated assessments of a wide range of energy-related environ-
mental issues.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
                              EPA-600/7-80-130
                      On-Line Zeeman Atomic Absorption
                     Spectroscopy for Mercury Analysis
                              in Oil Shale Gases
                                 D. C. Girvin
                                  J. P. Fox
                                August 1, 1979
                                    ERRATA
p. vi     Line 21:  Change "45" to "42".
p.  1     Paragraph 2, line 5:  Change "interface" to "interfere".
p. 10     In Figure 2, units to the left of 0 on the horizontal axis should
          be positive; units to the right of 0 should be negative.
p. 17     Line 4:  Change "iron-water" to "iron water".
p. 17     Line 5:  Change "lamp-water" to "lamp water".
p. 17     Line 9:  Change "magnet-light source" to "magnet light-source".
p. 33     Line 5:  Change "ouptut" to "output".
p. 44     Line 1:  Change "copper wire-tubing circuit" to "copper-wire,
                   tubing circuit".
p. 44.    Line 9 of first new paragraph:  Change "voltage-limited" to
                   "voltage limited".
p. 48     Line 9 of first new paragraph:  Change "standard tables" to
                   "standard tables12".
p. 50     Equation (1) should read:
               fto m _ / T W760
               p    P \273
p. 50     Equation (2) should read:
                o
                             760
p. 52     First equation at top of page should read:
               P° = P
               PC   "

p. 52     Second equation from top of page, change "H " to "Hg",
p. 55     Line 6 from bottom of page:  Change "±6% and 7%" to "±6% and ±7%'
                                (over)

-------
                                       -2-
p. 60     The equation in the first new paragraph should read:

               V    «
                ZAA

p. 68     In Figure 30:  Change "dilution devise" to "dilution device".
p. 68     Last line of figure caption:  Change "B  is the temperature B was
               varied" to "B as the temperature of B was varied."
p. 70     In Figure 31:  Change "DMV" to "DVM".
p. 71     Line 4 of figure caption:  Change "4500 scc/min and 700 scc/min" to
               "4500 scc/min and 250 scc/min,  respectively".
p. 72     Line 2 of first new paragraph:  Change "DMV" to "DVM".
p. 73     Line 1 of second new paragraph:   Change "austentic" to "ausenitic".
p. 94     Reference 4:  Change "LERC" to "LETC".
p. 95  ...  Reference 14:  Change "HANAJ" to "JANAF".

-------
                                            EPA-600/7-80-130
                                            June 1980
      ON-LINE ZEEM\N ATOMIC ABSORPTION
     SPECTROSCOPY FOR MERCURY ANALYSIS
             IN OIL SHALE GASES
                     BY
         D. C. Girvin and J. P. Fox
      Energy and Environmental Division
          University of California
        Lawrence Berkeley Laboratory
            Berkeley, CA  94720
          Contract No. 68-03-2667
               Project Officer

                Paul E. Mills
          Program Operations Office
Industrial Environmental Research Laboratory
             Cincinnati, OH  45268
INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY
     OFFICE OF RESEARCH AND DEVELOPMENT
    U, S. ENVIRONMENTAL PROTECTION AGENCY
          CINCINNATI, OH  45268

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

-------
                                   FOREWORD
    When energy and material resources are extracted, processed, converted,
and used, the related pollutional impacts on our environment and even on
our health often require that new and increasingly more efficient pollu-
tion control methods be used.  The Industrial Environmental Research
Laboratory-Cincinnati (lERL-Ci) assists in developing and demonstrating new
and improved methodologies that will meet these needs both efficiently and
economically.
    This publication describes the development and initial testing of
instrumentation for continuous on-line analytical measurement of mercury
concentrations in complex gas streams or in ambient air.  The mercury
monitor described is not susceptible to interferences which plague other
methods and thus may be used to characterize mercury emissions on a real-
time basis.  This mercury monitor will find immediate application for the
characterization of synfuel and other industrial emissions, mobile source
identification, and environmental health monitoring.  For further
information the Quality Assurance Branch of the Industrial Environmental
Research Laboratory-Ci should be contacted.
                                                David G. Stephan
                                Industrial Environmental Research Laboratory
                                                   Cincinnati
                                     111

-------
                                   ABSTRACT
    This report describes an instrumental  technique to continuously measure
total mercury in gas streams on a real-time basis.  The technique utilizes
Zeeman atomic absorption spectroscopy (ZAA) for on-line measurement of
mercury in the presence of smoke, organic vapors, and oil mist which are
typically present in offgases from oil shale processing plants.  The
accuracy of the ZAA background correction technique enables analytical
                                                             o
measurement of mercury with up to 95% attenuation of the 2537A analytical
line by broadband UV absorption.
    The ZAA mercury gas monitor built for use in field settings consists of
a ZAA spectrometer and a sample gas handling and metering system.  The
spectrometer incorporates a new thermally stabilized light source which
reduces instrumental baseline drift to >0.5 ppb for ambient temperature
variations between 13°C and 35°C.
    The spectrometer utilizes a new flow-through stainless steel (SS)
furnace maintained at 900°C by joule heating.  Corrosion of the furnace by
^S in the sample gas is minimized by diffusion of Al into the surface of
the SS and subsequent oxidation of this layer to form a protective coating
of alumina (A^CO.   Corrosion tests on furnaces treated in this manner
were conducted for 2% (v/v) H2S at 1093°C.  Estimated furnace lifetime
for continuous use under these conditions was three days.
    Furnaces with optical absorption tubes of different lengths are used
depending upon the mercury concentration.  Between 5 and 250 ppb (nanomoles
                                      iv

-------
Hg/mole of gas) of mercury in an 18 cm furnace is used.  The instrumental



response with this furnace is characterized by a detection limit (DL) of



2 ppb, a linear response up to 100 ppb, and a precision of ±7% or better.



I'n the 50 ppb to 1.6 ppm range a furnace with a 5 cm optical absorption



tube yields a DL of 10 ppb, a linear response up to 800 ppb, and a



precision of ±10% or better.  Sample gas flow rates can be varied between



400 and 4000 sec/min for either furnace.



    The spectrometer is calibrated using a dynamic calibration device which



generates a known concentration of saturated mercury vapor in a stream of



carrier gas.  Concentrations in the range of 1 ppb (0.01 mg/m ) to 2 ppm



(20 mg/m ) are obtained by variable dilution of this calibration gas.



    This report was submitted in fulfillment of Contract W7405-EN6-48



(Radian Corp. PO 15861) by The University of California Lawrence Berkeley



Laboratory under the partial sponsorship of the U. S. Environmental



Protection Agency.  This report covers a period from August 8, 1978 to



July 6, 1979 and was completed as of August 1, 1979.

-------
                                   CONTENTS
Foreword	iii
Abstract	-jv
Figures	ix
Tables	xi
Abbreviations and Symbols	X1--j
Acknowledgment 	  xiv
    1.   Introduction  	   1
    2.   Conclusions   	   4
    3.   Recommendations 	   5
    4.   Description of ZAA Spectrometer 	   7
              Theory of Operation  	   7
              Description of Components  	  16
              Light Source Assembly	16
              Variable Phase Retardation Plate 	  17
              Furnace Assembly	19
              Detector Assembly ' 	  21
              Electronics	22
    5.   Description of Mercury Gas Monitor System 	  39
              Functional Description 	  39
              Description of Components  	  42
              Heated Sample Delivery System  	  45
              Calculation of Hg Densities for Calibration  	  50

                                     vii

-------
    6.   Preliminary Evaluation and Performance of Mercury
         Gas Monitor System	   54
              ZAA Calibration	   54
              Light Source Temperature Stability and Control 	   62
              Determination of Heated Sample Transport Line Operating
              Temperature	   67
              Flow Controller	   69
              Corrosion Test and Estimate of Furnace Lifetime  ....   72
    7.   Operating Instruction 	   81
              ZAA Spectrometer	   81
              Light Source	   81
              Variable Phase Retardation Plate (Squeezer)   	   82
              Electronics	   84
              Furnace	   86
              Gas System	   87
    8.   Methods Comparison  	   89
    9.   Potential Applications  	   92
References	   94
                                    viii

-------
                                 FIGURES
1.  Electro-optical components of a Zeeman atomic absorption
    spectrometer 	   8
2.  Comparison of the emission lines from a 204Hg discharge lamp
    in a 15 kG magnetic field with the absorption profile (data
    points) of natural mercury at 1 atm of N2	10
3.  Schematic representation of ZAA depicting TT (probe)  beam and  a
    (reference) beam switching 	   12
4.  Diagram of the current-controlled variable phase retardation
    plate	13
5.  Signal  processing electronics	15
6.  Light source water jacket assembly 	   18
7.  Eighteen centimeter ZM furnace	20
8.  Square  wave generator circuit diagram	23
9.  ZAA wiring diagram	24
10. Audio frequency generator-amplifier circuit diagram	26
11. DC clamp-mixer circuit diagram 	   27
12. Block diagram of squeezer and squeezer electronics 	   28
13. Log amplifier circuit diagram	30
14. Lock-in amplifier circuit diagram	32
15. PMT high voltage supply circuit diagram	-. .   35
16. Output  current controller circuit diagram for 11 kilowatt
    power supply	36
17. Circuit diagram of 11 kilowatt power supply	37
18. Schematic of gas handling system for ZAA mercury monitor	40
19. Heated  sample probe	43

-------
20. Flow controller wiring diagram 	  47

21. Schematic of mercury calibration system	49

22. ZAA calibration curve obtained for 18 cm furnace
    (0-500 ppb)	56

23. ZAA calibration curve for the 18 cm furnace (0-250 ppb)	57

24. ZAA calibration curve for 18 cm furnace (0-100 ppb)	58

25. ZAA calibration curve for 5 cm furnace (0-1.16 ppm)	59

26. ZAA calibration and furnace test	61

27. Change in intensity of gaseous mercury discharge lamp
    with temperature	63

28. Temperature dependence of ZAA response to a constant
    concentration of mercury 	  64

29. Change in ZAA output voltage due to self-reversal in the
    absence of mercury in the sample gas	55

30. Heated transport line test arrangement 	  68

31. Flow controller test arrangement	70

32. Comparison of flow controller and wet test meter (WTM) flow
    readings	j-\

33a. Cross section of Corrosion Test sample:   magnification
     x320	75

33b. Enlargement (x800) of corrosion test sample 	  75

34.  Corrosion test phase stability diagram at 1093°C	78

35a. Example of Lock-in amplifier mixer output if phase adjust-
     ment has been improperly made	85

35b. Lock-in amplifier mixer output if phase is properly
     adjusted	85
                                      X

-------
                                   TABLES
1.  Required measurements for gas handling system 	  41
2.  Heated transport line experimental data 	  69
3.  Comparison between EPA reference method and ZAA method  	  90
                                      XI

-------
                      LIST OF ABBREVIATIONS AND SYMBOLS

ABBREVIATIONS

AD         — absorption units
L          — liters
ppb        — parts per billion nanomoles Hg/mole gas
ppm        — parts per million micromoles Hg/mole gas
atm        — atmosphere
&          — length of absorption light path
T          — temperature
P          — total pressure
p          — partial pressure
PO         — density
p°         — density converted to standard conditions
              (standard conditions:  Pressure 760 mmHg, Temperature 0°C)
qo         — volumetric flow rate
q°         — volumetric flow rate converted to standard conditions
sec        — cubic centimeters converted to standard conditions
mg/m^      — milligrams per cubic meter
m          — meters
Kcal       — kilocalorie
ZAA        — Zeeman Atomic absorption
AA         — atomic absorption
LETC       — Laramie Energy Technology Center
VPRP       — variable phase retardation plate (squeezer)
PMT        — photomultiplier tube
LIA        — lock-in amplifier
EDL        — electrodeless discharge lamp
PLL        — pen light lamp
SS         — stainless steel
NIM rack   — nuclear instrumentation module electronics rack
SW6        — square wave generator module
LG         — inductor-capacitor circuit
DVM        — digital volt meter
WTM        — wet test meter
NBS        — National Bureau of Standards
AF6        — audio frequency generator
DL         — detection limit
                                     xxi

-------
SYMBOLS
Al            — aluminum                  Fe         — iron
AT 2^3         — alumina                   Cr         — chromium
A1C1          — aluminide                 Ni         — nickel
$2            — sulfur                    H2$        — hydrogen sulfide
02            — oxygen                    CrS        — chromium sulfide
H2            — hydrogen                  FeS        — iron sulfide
                                           NiS        — nickel sulfide
                                           NiO        — nickel oxide
                                           Fe304      — iron oxide
0                   i                       IC1        — iodine monochloride
A             — Angstrom
ir             — emission or analytical  line
CT             — reference or background correction  line

   Hg         — mercury isotope 204
R             — universal gas constant
In K          — natural log of equilibrium constant
AGf           — Gibbs standard free-energy of formation
                                     xiii

-------
                               ACKNOWLEDGMENTS
    The authors thank Dr. Tetsuo Hadeishi for his continued interest in and
contributions to this particular application of Zeeman spectroscopy.  We
also thank all the individuals who contributed to the progress and success-
ful completion of this work, in particular Al Hodgson, Suzanne Doyle, and
Lucy Pacas.  Special appreciation is expressed to Paul Mills and Bob
Thurnau of EPA and to Bob Magee of the Radian Corporation for their advice
and support.
                                     xiv

-------
                                  SECTION 1
                                 INTRODUCTION
    Preliminary investigations of pilot-scale oil shale processing
      1-3
plants    indicate that the level of mercury in offgases may be signifi-
cant.  Extrapolation of these results to field conditions suggests that a
100,000 barrel-per day oil shale plant processing 100 L/ton (24 gal/ton)
oil shale with an average mercury content of 0.86 ppm  (milligrams
Hg/gram shale) could release approximately 33 tons of mercury per year to
the atmosphere.  In contrast, the amount of mercury released from world
coal consumption in 1967 was estimated to be 18 tons. '   These data
suggest that mercury emissions from oil shale plants may be of future
environmental concern and that these plants may require control technology
to reduce mercury to acceptable  levels.  Reliable techniques will be
required to measure the mercury  in these gases and thus evaluate the extent
and type of control technology needed.
    Reliable and representative measurements of mercury in gases from
                                 i
in-situ shale plants are difficult to obtain.  Conventional mercury gas
stack sampling techniques such as gold bead absorption tubes or impinger
trains are of limited use because high concentrations of organic and sulfur
compounds in the oil shale offgas interface with the collection and
subsequent analysis of mercury.
    Direct on-line Zeeman atomic absorption (ZAA) measurements of mercury
in an oil shale offgas stream were made during a complete burn of the

-------
Laramie Energy Technology Center  (LETC) 20-Kg retort.   These measurements
were made using a prototype instrument originally designed for analysis of
mercury in discrete liquid and solid samples.  Despite the broad band UV
absorption caused by the high concentration of organic and sulfur compounds
in the off-gas, this prototype ZAA was capable of making analytical mercury
measurements.  This capability is due to the unique ZAA background
correction system.  Conventional atomic absorption spectroscopy cannot
perform such measurements.  The LETC experiment also showed that the mercury
concentrations in the gas stream varied from 10 ppb (parts per billion
nanomoles Hg/mole gas) to 8 ppm (parts per million micromoles Hg/mole gas)
during the retorting process.   If these variations are typical, frequent
measurement must be made throughout the course of the retorting process to
obtain representative mercury emission values.
    It was clear from the LETC experiment that a number of improvements and
design changes would better facilitate continuous on-line ZAA analysis of
mercury in gas streams.  This  final report presents the results of a six-
month project to: (1) make these improvements and design changes, and (2) to
build and test a field version of a Zeeman Mercury Monitoring System for
continuous on-line analysis of mercury in oil shale offgases.  These changes
and improvements include design of a thermally stable light source, design
of a heated sample probe, material selection to minimize corrosion, redesign
of furnace and electronics to  facilitate parts replacement and improvement
of electronic stability for long-term use under field conditions.  The
report contains: (1) a discussion of the theory of ZAA operation, (2) a
description of the Zeeman atomic absorption spectrometer, (3) a description
of the mercury gas monitoring system, (4) an initial laboratory evaluation

-------
of the system, and (5) operating instructions.   Subsequent field testing,



evaluation and development of the monitoring system is continuing under



separate contract.

-------
                                  SECTION 2
                                 CONCLUSIONS
    A new Zeeman atomic absorption  (ZAA)  spectrometer was designed and
built to make continuous on-line measurements of mercury in the off-gas
from oil shale retorts.  Design changes  in ZAA spectrometer, gas sampling
system, and the calibration device  suggested by initial experiments with a
prototype ZAA at LETC were made, and a field ZAA mercury gas monitor system
was built which incorporates these  changes.  Laboratory testing of the new
monitor system demonstrated significant  improvements in the performance and
stability of the ZAA spectrometer,  gas sampling and metering techniques,
instrumental calibration techniques, and  extension of the ZAA furnace
lifetimes.  The system is ready for large-scale field testing.
                                      4

-------
                                  SECTION  3
                               RECOMMENDATIONS
    On the basis of the earlier experiments at LETC and the more recent
laboratory testing of the new monitoring system the following
recommendations are offered.
    1.  The new ZAA mercury monitoring system should be tested during field
retorting experiments in order to evaluate the system and the sampling and
calibration strategies under actual field conditions.
    2.  Improvements in instrumentation and changes in sampling and
calibration strategies suggested by field testing should be implemented.
    3.  The concentration of the saturated mercury vapor obtained with the
dynamic calibration device should be compared against independent standards
to determine its accuracy as a function of temperature and carrier gas flow
rate.
    4.  A large amount of raw data will be generated when continuous
monitoring is performed during retorting experiments which last days to
weeks.  An on-line microprocessor-based data collection and analysis
system, capable of recording and'permanently storing the 20-odd parameters
and data streams associated with the system, should be interfaced to the
ZAA monitor.  Off-  or on-line microcomputer access and analysis of data
would reduce data reduction time from weeks to days.   During field use of
the ZAA monitor, this rapid data analysis and summary capability would be

-------
invaluable in evaluating monitoring strategies under rapidly changing field
conditions before the experiment ends.
    5.  The ZAA mercury monitor system should be applied to the characteri-
zation of synfuel, geothermal, and other industrial emissions in addition
to environmental health monitoring.
    6.  The ZAA monitoring concept developed here for mercury should be
extended to other volatile trace elements including Cd, Se, and As.

-------
                                  SECTION 4
                       DESCRIPTION OF ZAA SPECTROMETER
THEORY OF OPERATION
    Zeeman atomic absorption spectroscopy (ZAA) is an analytical technique
developed at Lawrence Berkeley Laboratory by Tetsuo Hadeishi.   ZAA is
similar to conventional atomic absorption spectroscopy; however, it differs
principally in that the light source is placed in a magnetic field.  The
magnetic field splits the 2537A mercury resonance emission line into
linearly (IT) and circularly (a) polarized Zeeman components.  This split-
ting is referred to as the Zeeman effect.  The ir component is used to
detect the presence of mercury, and the two a components, which are shifted
in wavelength, are used to monitor light scattering by smoke.  A unique
electro-optical switching device distinguishes between the ir and a com-
ponents.  These components are then alternately passed through the sample
vapor, and the difference in absorption of these two components is used as
a measure of the amount of mercuryspresent.  Since the spatial and temporal
variations in the IT and a components are virtually identical, background
correction capabilities are vastly superior to those obtainable with con-
ventional AA techniques.  As a result, mercury can be measured in the
presence of large quantities of smoke, organic molecules and other
interfering substances.
    The spectrometer consists of three major components (Fig. 1):  a light
source which provides a 2537A mercury emission line (n) and reference lines

                                      7

-------
oo
        PLUG-IN LIGHT SOURCE,
        R.F. EXCITER
        COMBINATION
MAGNETIC FIELD
15-25 KG.
                                     VARIABLE  PHASE-RETARDATION
                                     PLATE ASSEMBLY


                                      FUSED SILICA SLAB
                                 LINEAR
POLAR.ZER    ^AggPgggN TUBE
                  COMPACT LIGHT
                  SOURCE AND
                  WAVE LENGTH
                  MODULATOR
                                                                            PHOTOTUBE
                                            SAMPLE
                                             INLET
                                 -TYPICAL ATOMIC ABSORPTION SPECTROMETER-
                                                                                    XBL 793-8742
                 Figure 1.   Electro-optical components of a Zeeman atomic  absorption spectrometer.8

-------
(a) for background correction; a furnace-absorption tube assembly where



vapors from thermally decomposed samples are swept Into the path of the



emission and reference beams; and a detector which converts changes in the



intensity of the transmitted emission and reference beams into an AC



voltage for signal processing.



    The key to the ZAA technique lies in the mode by which the emission



and reference lines are generated and subsequently distinguished from one



another.  Both the emission and reference lines are simultaneously



generated by a single mercury discharge lamp operated in a 15-kilogauss

                                                                        o

magnetic field.  The Zeeman effect is the splitting of the original 2537A



resonance line, in the presence of a magnetic field, into its three Zeeman



components:  a a~ component shifted to a longer wavelength, a cr



component shifted to a shorter wavelength,  and an unshifted ir component.


                              ?04
These Zeeman components for a    Hg lamp are shown in Fig. 2.



    The mercury present in the furnace-absorption tube consists of a



naturally occurring mixture of several stable isotopes at one atmosphere.



Thus the absorption lines of each isotope are pressure-broadened.  The



resulting total absorption profile due to naturally occurring mercury  (at



1 atm of ^) is superimposed upon the Zeeman-split emission spectrum



(Fig. 2).  Note that the IT component coincides with the peak of the absorp-



tion profile for natural mercury, while the o components are both on the



outer edges of the profile.  Therefore, the difference in absorption of the



ir and cr components may be used as a measure of the quantity of mercury



present in the absorption tube.  Here the ir component becomes the mercury



probe beam and the a components taken together become the reference beam.



The Zeeman splitting also provides a means  of distinguishing between the ir

-------
     CL
     h-
     co

     OJ

     c •
     c
     ON
1.0
	1	
 Emission line
 X=2537A
                 204Hgcr-(l5KG)f
                                       204Hgo-+(15KG)
             0  -0.114  -0.076  -0.038   0    0.038  0.076
                               0.058A
                                 A\(A)
                                                      XBL73I-I05B
Figure  2.  Comparison of the emission lines from a    ^g  discharge lamp in a
          15 kG magnetic field  with the absorption profile (data points) of
          natural  mercury at 1  atm of ^\2'
                                  10

-------
and a components.  When the light source is viewed perpendicular to the
applied magnetic field, that is, along the optical axis of the instrument,
both a components are linearly polarized perpendicular to the field, while
the ir component is linearly polarized parallel to the field (Figs. 1 and 3).
    Alternate selection of the IT and a components before transmission
through the absorption tube is achieved by using a variable phase retarda-
tion plate (VPRP) and a simple linear polarizer (Figs. 1 and 3).  The
linear polarizer is oriented with its polarization axis parallel to the
light source magnetic field (Fig 3.).  The VPRP (Fig. 4) consists of a slab
of fused quartz to which a sinusoidally varying stress is applied.  This
quartz slab is mounted in the optical light path with the stress axis
oriented at an angle of 45° to the light source magnetic field.
    The polarization axis of the incident linearly polarized light is
rotated 90" as the light passes through the stressed quartz.  This rotation
is due to the difference in the propagation velocities for those components
of polarized light which are parallel and perpendicular to the quartz
stress axis.  The amount of rotation is controlled by appropriate selection
of current to the driver coil and the optical path length of the quartz.
As seen in Fig. 3, when the current applied to the driver coil is zero (no
                               i
stress), only the ir component is transmitted by the linear polarizer.  When
the driver coil current is adjusted so that the quartz is a halfwave plate
(maximum stress), both ir and a components are rotated by 90°, and the
linear polarizer passes only to the a components.
    The gaseous sample to be analyzed for mercury enters the furnace-
absorption tube assembly where it is heated to 900°C which well exceeds the
vaporization temperature of mercury.  Individual free atoms of mercury and
                                      11

-------
tsi.
                 Light source

                       Magnetic
                           field
         Light path
                           7T.
         Light path
                    Phase  retardation
                    plate (squeezer)
                    No stress
                    (no rotation)
                                       Max stress
                                       (90° rotation)
Linear polarizer 8t furnace     Detector
                                                                                                Phototube
                                                                               S      '
                                                                                                Phototube
                                                                                              XBL 793-789
Figure 3.  Schematic representation  of  ZAA depicting
           (reference) beam  switching.
                                                                           (probe) beam  and  a

-------
                                           XBL748-3969B
Figure 4.   Diagram of the  current-controlled variable phase retardation
           plate,   (a)  Plate  of fused quartz; (b) laminated pulse transformer
           core;  (c)  0.5 mm gap;  (d) drive coils; (e) stiffener plates.  The
           long arrow in the  center of the quartz represents the stress axis,
           while  the  double arrows depict the linear polarization  axes of the
           TT and  a beams. ^
                                     13

-------
decomposition products are then swept by the gas stream into the light path


of the absorption tube.  Oxygen is introduced into the furnace chamber to


promote combustion of organics and thus reduce smoke.  The IT component is


attenuated due to absorption by mercury atoms in their ground state and


scattering and absorption by decomposition products and smoke.  The magni-


tude of this attenuation is exponentially dependent upon the product


£-(P|^_ + Pscat)> where 2. is the length of the absorption light path,


p^q is the density of mercury atoms, and Psca|- is the density of

scattering atoms and molecules.  The a component is attenuated due to


scattering and smoke only.  Thus, the magnitude of the attenuation of the a


component depends exponentially only on £'Pscaf  The detector consists
                                                                        o
of an interference filter which passes all Zeeman components of the 2537A


line equally well but blocks light of other wavelengths from striking the


cathode of the photomultiplier tube (PMT).  The PMT generates an output


voltage proportional to the intensity of the sum of the ir and a com-


ponents.  If no mercury is present in the absorption tube, the probe and


reference beams are absorbed and scattered identically by nonmercury back-


ground.  Hence, as they alternately fall upon the PMT, the light intensity


does not change, and the PMT output voltage remains constant.  In the


presence of mercury, however, the probe component will be more strongly


absorbed than the reference component, and the PMT output will vary at the


audio frequency at which the switching from one beam to the other takes


place (Fig. 5).


    The PMT output together with an audio reference signal from the


oscillator driving the magnetic clamp, are fed into the lock-in-amplifier


(LIA) (Fig. 5).  The tuned amplifier in the front end of the lock-in
                                      14

-------
W3

0>
k.

CO
                                           Max stress






\

Time
Zero
stress
                      Alternates
                      TT-CT Signal
                                              Reference
                                              from squeezer

                                                     TfSignal
                                      Lock-in-amp
C71
        
-------
amplifier accepts only those signals having the  same frequency as that
used by the VPRP to switch between n and a beams.  This amplifier first
recognizes and then takes the difference between ir and 0 components in the
audio portion of the PMT output.  It supplies a  DC voltage which is
proportional  to this difference and thus which is proportional to the
density of mercury in the absorption tube.
    The accuracy of the background correction obtained by ZAA through the
use of spatially and temporally coherent ir and a beams, synchronous beam
switching, and electronic signal processing techniques results in a signif-
icant advance beyond conventional AA background  correction.  As a result,
ZAA is capable of performing accurately with up  to 95% attenuation (from
smoke) of the ir and a components.  Thus, ZAA is  uniquely suited for direct
analysis of mercury in most gas, liquid, and solid samples without prior
chemical treatment.  This direct analysis capability is the major advantage
of ZAA over conventional AA for on-line field measurements of mercury in
gas streams.
DESCRIPTION OF COMPONENTS
    A ZAA spectrometer has been designed and built which is capable of
continuously measuring mercury concentration in  offgas streams on a
rea'i-time basis.  Specifically, a new light source, furnace assembly, and
electronics package have been developed to accommodate gas sampling and
mercury analysis under field conditions.  These  and other ZAA components
are described below.
Light Source  Assembly
    A new low-pressure mercury gaseous discharge lamp has been built and
tested which  will replace the radio frequency excited electrodeless
                                      16

-------
discharge lamp (EDL) previously in use.    This "pen light lamp" (PLL)
consists of a U-shaped quartz tube containing argon and a small quantity of
mercury.  Minute electrodes are sealed in each end of the tube.  The outer
diameter of the tube is 7 mm.  The lamp is surrounded by a soft iron-water
jacket fitted with a quartz window (Fig. 6).  The lamp-water jacket
assembly fits between the pole tips of the permanent magnet which produces
the Zeeman splitting of the resonance lines.  The argon plasma and mercury
resonance lines are produced by a 700 Hz high voltage driver described
below (pg. 22).  The magneto-light source assembly is mounted so that  it may
be adjusted with respect to the optical axis of the instrument.
Variable Phase Retardation Plate
    The variable phase retardation plate or squeezer is a slab of fused
quartz mounted inside a pulse-transformer core which has drive coils on
either side of a 0.5 mm air gap in the core (Fig. 4).  The quartz is
approximately 2.5 cm long, 1.2 cm wide, and 1.2 cm thick.  The drive coils
are 200 turns each.  With the passage of current through the coils, the
core acts as a magnetic clamp, applying a stress to the quartz.  The  light
beam traverses the quartz near its center, with the stress axis making an
angle of 45° to the axes of polarization of the light beam.
    The rotation of the polarization axes is achieved by driving the
magnetic clamp with an AC current superimposed upon a DC bias current.  By
adjusting the magnitude of the DC current and the amplitude of the AC
current, the stress at one extreme is adequate to make the quartz function
as a halfwave plate (90° rotation of polarization axis) while, at the other
extreme, the stress is near zero.  It is essential that the stress does
not pass through zero; otherwise the core and quartz may actually lose

                                     17

-------
                                                            CBB 793-3220
Figure 6.  Light source water jacket assembly.  The low pressure mercury
           light source screws into the upper right hand port and is flush
           with the flat end of the cone shaped soft iron pole tips.  The
           flange cap shows the pole tip.  The quartz window is mounted in
           the threaded hole lower right.  Water temperature is measured by
           the thermocouple (lower left port).  Water from constant
           temperature bath enters through tube upper left and exits from
           tube on right.
                                     18

-------
contact.  This condition produces an audible chattering and damage can

result if allowed to persist.  The electronic circuit used to drive the

squeezer is described below.

Furnace Assembly

    A new furnace for continuous on-line analysis of mercury in gas streams

has been constructed and successfully operated at 900*C for extended

periods.  The furnace (Fig. 7) is constructed of 1.25 cm OD, 0.12 cm thick

wall, 321 stainless steel  (SS) tubing welded into a tee.  Incoming gases

fir^st pass through the atomization-cornbustion chamber which is maintained

at 900° by joule heating.  This chamber  is filled with ceramic beads to

break up the gas flow and  increase the thermal contact area.  The gases

then pass through a small  opening into the absorption chamber which is

aligned along the optical  path of the spectrometer.  The temperature in

this chamber is lower, approximately 500°C,  as the current  in each leg is

one-half of that flowing through the atomization chamber.   Quartz windows
                                                   o
at the ends of the absorption chamber pass the 2537A mercury resonance

lines while isolating the  hot sample gases from the ambient air.  Gases

exit the furnace through tubes located near  each end of the absorption

chamber.

    Current and mounting support for the furnace are supplied via variable

cross-section strips of 304 SS welded to the tubing.  When  the furnace is

at operating temperature,  the outer ends of  the strips are  cool, thus

preventing the buildup of  resistive oxide layers on the power connector

surfaces.  The entire furnace assembly is mounted on an adjustable platform

which permits the precise  optical alignment  of the absorption tube.
                                      19

-------
ro
                                                                                        XBL792-48I
        Figure 7. Eighteen cm ZAA furnace.  The furnace is supported by flexible SS sheet  held  in
                  position by copper blocks which also act as a power bus.  The section  of the  furnace
                  where the gas enters is filled with ceramic chips.  At  right angles  to this is the
                  absorption chamber, through which the Hg light beam passes.  Sample  gas  leaves the
                  furnace through small tubes located just behind the quartz window and  flexible support.

-------
    As noted above (pg. 14) the attenuation of the ir beam, and thus the
instrumental voltage response to a given concentration of mercury atoms
(PH ), varies exponentially with the length, £, of the absorption light
path.  The length of the furnace absorption tube determines £.  Furnaces
with long absorption tubes are used to measure sample gases with low
mercury concentrations.  The instrumental response with these long tubes is
linear at low concentrations.  However, with a significant increase in con-
centration, the instrumental response becomes non-linear, eventually
flattening, unless the absorption path length is decreased.  For this
reason, furnaces with absorption tubes of different lengths are used,
depending upon the mercury concentration to be measured.  A high sensi-
tivity furnace with an absorption tube 18 cm long is used in the five to
250 ppb (nanomoles Hg per mole of gas) range.  A low sensitivity furnace
with an absorption tube 5 cm in length is used in the 50 ppb to 1.6 ppm
mercury concentration range.  These furnaces have identical mounting
brackets and are easily interchangeable.  Details of the ZAA response
versus mercury concentration for each furnace are discussed below (pgs. 54
and 55).
Detector Assembly
    The detector consists of a narrow bandpass filter, a photomultiplier
tube and a light-tight PMT housing.  The Oriel narrow bandpass interference
filter has a 120A transmission window centered at approximately 2537A.  The
filter is mounted in the input window of the PMT housing.  The PMT is a
ninestage Hammatsu R928.  The PMT high voltage power supply is located in
the nuclear instrumentation module (NIM) electronics rack located below the
ZAA.
                                     21

-------
Electronics
    Light Source Driver
                                                    i
    The newly designed square wave generator module (SWG) is the first
stage of the light source power supply.  It provides a 200 volt peak-to-
peak, 700 Hz waveform to the primary of a saturable core step-up trans-
former.  The transformer output signal (2000 V) is sufficient to excite the
argon plasma discharge in the PLL.  The transformer also acts as an impe-
dance matching device capable of responding to the complex input impedance
characteristics of the PLL.  This is essential if sufficient power is to be
coupled in a controlled way into the PLL to excite the plasma but not burn
out the electrodes in the PLL.  The transformer will henceforth be referred
to as the coupling transformer.  A circuit diagram of the SWG is shown in
Fig.  8.
    The SWG module is located in the NIM electronics rack (Fig. 9), and the
coupling transformer is located next to the permanent magnet in the light
source section of the spectrometer (Fig. 1).  The front panel of the SWG
consists of a power switch and a run light.  The SWG output is located on
the rear panel BNC connector J-l.  A switch on the coupling transformer
enables remote on-off switching of the PLL when the PLL is removed from the
pole-tip water-jacket assembly.
Audio Frequency Generator/Amplifier
    The newly designed audio frequency generator-power amplifier supplies
the sinusoidal voltage used to drive the pulse transformer coils of the
variable phase retardation plate.  It consists of two main components:  a'
function generator with variable frequency apd amplitude controls which
provide a low voltage sinusoidal waveform and a 75-watt audio amplifier.
                                      22

-------
N)
                                                        IT,) t	°V]f*"- Si.
                                                           (	1  AIM   A
                                                                        JHt23f
                                                                  lltL
                                                                   Tir






f-



r-st


(ft rt
iv ir



i
la T

'(J— 5
*



i
1
                                                                                               XBL 7912-13748
                               Figure 8.  Square wave generator circuit  diagram.

-------
                                                                                         C -.OUTRO- firry.  E-HS
                         WATER


                      LUMP IN  J4
to
                      S8UEEZER J3

                      PMT OUT  J2

                      PMT H.V.  n
208VM
w
s
=?.
Jl :nr.t- CUPFLY
J3 WEF OUT
J2 fluDIO PWR
J4 HUDIOFWP IN
J5 SQUEEZER SUPPLY
J6 PMTSI9 IN
J7 PUT MDN.
J8 LOS OUT
J9 LOS IH
JIOKEF ill
jil oc oor
Jit
je
JM

3QOAI?E" WAV£ 5UPFLY
OUDIf) KRE'J SEN. & flMp.
29- 600 J8S
ClflMP/MKER
i 3- 600534
LOS BMP
as - eoosas
LOCK IN BMP.
?9 • toOOSSt*
FUKNflCB CONrKOL
29-60Or97
OOX FURNftCE COHT.
Z9-600SSB
PMT VOLTKSE CONT.
29-600J89

                                                                                            XBL 7912-13755
                 Figure 9.
ZM wiring diagram.   The interconnections are  shown  between the ZAA
spectromter, the  electronics modules located in the  NIM rack and
the furnace and heated  transport power lines.

-------
In response to the sinusoidal waveform, the quartz slab varies between its
zero wave plate (n beam) and halfwave plate (a beam) conditions.  A circuit
diagram is shown in Fig. 10.
    The module is located in the NIM electronics rack (Fig. 9).  The
amplitude and frequency of the function generator signal are selected using
front panel potentiometers.  The function generator output signal is pro-
vided on the front panel BNC connector labeled "Audio Ref." and on the rear
panel BNC connector labeled "
-------
                                       TEST JUMPER
                                        DURIN* IBST
ts)
                                                                                                           J2.
                                                                                               XBL 7912-13753
                        Figure 10.  Audio frequency  generator-amplifier circuit diagram.

-------



«.p. i ! I'O1 5
fiKft 1 p-L " \ \J



! 	 J
F, P.


C
I
i





KC
*.&,,•*
tt




R.P.
n-
'"> ' rt. M"*- v* . sw^
I ] i i/rsww i f»X £C rz
U 1 | 	 ^ 	 1 1 55^ /ou,
T< l-t
:i f\/ \ tLS I e-/,C*


&LK frTt

\~
1
1
C3 Cf \VH
1MI* -i
IHI , -1
"J. «•!« I
1



1
1
l_
                                                                               R.P
                                                                              JV
                                                                                     ic Anr
                                                                                          OUT Pur
f.P.  « lOCfTCV OH Ff'HT
                                                                                 XBL 7912-13751
                      Figure 11.  DC  clamp-mixer circuit diagram.

-------
                       BLOCK DIAGRAM  of  SQUEEZER  and  SQUEEZER ELECTRONICS
CO
              Audio frequency
              generator amplifier
                   DC isolation
                   AC blocking
                   choke

o
o
o
o
          DC and audio
          power mixer
                               Squeezer - driver coils
                                     and
                               pulse transformer core
               DC clamp
               power supply
                                                                                   XBL 801-77
                      Figure 12.  Block diagram of squeezer  and squeezer electronics.

-------
circuit.  At this resonant frequency, the power absorbed by the squeezer,
and thus the stress developed in the quartz slab, is a maximum.
    The DC clamp/mixer module is located in the NIM electronics rack
(Fig. 9).  The front panel controls include a DC clamp power switch with
indicator light and a one-turn potentiometer controlling the magnitude of
the DC voltage.  The audio power input from the frequency generator
amplifier and the mixer output are located on rear panel BNC connectors J-4
and J-5, respectively.  A five-amp fuse for the DC clamp supply is also
located on the rear panel.
Log Amplifier
    The purpose of the log amplifier is to convert the PMT signal, which is
an exponential function of mercury density, into a signal which varies
linearly with mercury density.  A circuit diagram of the log amplifier is
shown in Fig. 13.  The module is located in the NIM electronics rack
(Fig. 9).  The PMT input to the log amplifier can be monitored from the
front panel BNC connector.  The PMT input signal, a second PMT monitor
signal and the log output signal are located on rear panel BNC connectors
J-6, J-7, and J-8, respectively.
Lock-in Amplifier
    A new lock-in amplifier was designed for the field mercury monitor.
The purpose of the lock-in amplifier (LIA) is to extract and amplify the
audio frequency signal contained in the output signal of the log ampli-
fier.  This audio component is due to the presence of mercury in the sample
gas.  The amplifier alternately measures the signal level corresponding to
mercury plus background absorption (TT beam) and the signal level which
corresponds to background absorption (a beam).  The output is a voltage
                                      29

-------
JS

PMTO-
SI6
       «l
       IOM;
              Ml

              AD503K
                                             M2.
A~D755P
      ANTILOG
      ELEMENT
                                              rm/_ _-it,»uo _  _	|
                                                                        OUT
                                                                XBL 7912-13749
                    Figure 13. Log amplifier circuit diagram.

-------
which is proportional to the difference between the two and, thus, is
proportional to the density of mercury.  Since the background and the
mercury plus background signals are presented to the LIA alternately at
the same terminal, the amplifier must also be provided with information as
to when each signal arrives.  This synchronous information is provided by
the reference signal from the audio oscillator.  This signal, after
amplification, is the same signal used to drive the squeezer.
    A circuit diagram of the LIA is shown in Fig. 14.  The module is
located in the NIM electronics rack (Fig. 9).  The front panel of the LIA
contains several controls which are described immediately below.
    As soon as the signal enters the LIA, it encounters a tunable audio
frequency amplifier.  This circuit selectively amplifies the harmonics of
the input signal which have the same frequency as the reference signal and
suppresses components of the input signal which differ substantially from
this frequency.  The potentiometer, marked "frequency," is used to select
the amplification frequency.  The potentiometer, labeled "gain," provides
continuous selection of amplifier gain.  With the "mixer-frequency" toggle
switch in the frequency position, the output of the tuned amplifier is
available for tuning purposes on the front panel BNC connector labeled
                                \
"mixer/frequency."
    For a variety of reasons, the stress on the quartz plate may not peak
at exactly the same time as the audio reference signal reaches its maximum
value; that is, there may be a phase difference between the two.  To
correct for this, the phase of the audio reference signal may be shifted
electronically relative to the input signal by adjusting the phase shift
knob on the front panel.
                                      31

-------
(fcHH
    NOTES:
      1. fllL «SES  tV -»  tllVPC   /»;«/
                -V —»  - li YCC   ^/A/
      a. flU RES. VALUES IK  ONM5
      3. fllL tflP, V«LUES IN  MICRO FBRIDS
      4. M I, Z, «,5; fcx % II  BRE LMT1ICH
        Ml, 7         ARE LM3ION
        MS, 10         HRE LM3IIN
T/VO
                                                                                                               /r«eu
                                                       XBL 7912-13747
                              Figure  14.  Lock-in  amplifier circuit diagram.

-------
    The circuit that takes the difference between the * and the a signals
is called the mixer.  It is this circuit that requires the audio reference
signal  in order to discriminate between the two signals.  When the fre-
quency and phase adjustments are properly made, this circuit provides a DC
ouptut voltage which is proportional to the difference.  The mixer output
is available for tuning purposes on the "mixer/frequency" front panel BNC
connector when the "mixer-frequency" toggle switch is in the mixer position.
    A fixed-gain DC amplifier  immediately follows the mixer circuitry.
When the difference between the IT and a signals is small, the voltage
output from the mixer and, hence, the DC amplifier may fluctuate due to
random instabilities in the system.  Since only the average voltage output
of the amplifier is significant, means are provided to electronically
average (damp) the output.  The "time constant" toggle switch on the front
panel provides a choice of one- or  two-second  averaging times.  This
time-averaged output is then presented to the  meter at the upper left and
to the BNC connector marked "DC out" (J-ll) on the rear panel.   It is this
voltage which serves as the measure of the quantity of mercury  in the
absorption tube at any given time.
    The voltage presented to the meter and to  the output connector may not
be zero in the absence of mercury,in the absorption tube.  One  reason for
this is the difference in the  intensity of the n- beam and 0 beams due to
self absorption of the IT beam in the light source.  The LIA output voltage
may be zeroed by adding or subtracting a constant offset voltage using the
"zero adjust" control in the lower  right corner of the front  panel.  The
log amplifier input (J-9) and  the reference input (J-10) BNC  connectors  are
located on the rear panel.
                                      33

-------
PMT High Voltage Supply
    The PMT supply provides the high voltage to the dynodes of the
photomultiplier for the initial amplification of the PMT signal.  A circuit
diagram of the high voltage supply is shown in Fig. 15.  This supply module
is located in the NIM rack (Fig. 9).  The potentiometer on the front panel
provides continuous adjustment of the output voltage from 0 to 1000 volts
and is calibrated to read in volts.  A special high voltage output
connector is located on the rear panel and is labeled J-14.
Furnace and Auxiliary Furnace Control
    The outputs of the 11-kilowatt power supplies used to heat the ZAA
furnace and heated sample transport  line are controlled by identical
control modules located in the NIM electronics rack (Fig. 9).  The
potentiometer on the front panel of  each module is used to vary the reactor
current in the power supplies and thus their output.  A circuit diagram of
these controllers is shown in Fig. 16.  A BNC connector on the rear panel
of each controller provides the control signal voltage.  These are labeled
J-12 and 3-13.
Furnace and Heated Transport Line Power Supplies
    Each power supply can deliver up to 11 kilowatts of AC power.  These
supplies are designed to deliver high currents (650 amps maximum) at low
voltages (17.5 volts maximum).  The  high currents are required to provide
the joule heating in the furnace and sample transport lines.  These
supplies require a single-phase 208-volt, 30-amp power source.  A circuit
diagram of the supply is shown  in Fig. 17.  The BNC connector for the out-
put control signal is located on the rear panel along with the output power
bus.  The main circuit breaker  is located on the front panel and is labeled
                                      34

-------
             PIN
+I2V
                                   \   f
                              R2
                        R3
                        2.*
Ul
                           rfn
                                                                    PSJ.
                                                                                       _L
                                                                                                 tjl
                                                                                            >O/*^
                                                                         7777




                                                                          XBL 7912-13750
                             Figure  15. PMT  high voltage  supply circuit diagram.

-------
oo
en
               o
u
5
^
•<
* *' :
S ">x <
5
o
o
GIW
Bin ^JV J_
/
e
c
7777
X
^ . + /A
> 3 /•
L x-x
>/?z
d
?fp
/vrf/i
p^/^i/^c (
f ../ 03-1^^
.«/ -r:
-r- A A A , i ^
'^X^ V.?^ ff
j ^
r>
T" "1
A /ox?
/777
/?6
A A A
OKANC?
BIN*I 
-------
    FRONT  PftNEL


c
— (
fTVl1
~\^>

	 f
t-A- ^
-1—
1
>l 0 	



T4
(AM) 3408
, ^ _, CONTACTOR Z20 .I ==
o i o • 2 fc
! .^^ I95^llf

                                                                                 °J
                                                                                 oj
                                                                                 o
       p rv N e -
PRIME t Tw/feK  INPUT
209 Vrf-'  S04f

EPITH
 CIRCUIT BKBftKER
OUTPUT
VOLTflSE
                                                                                    O0|

                                                                                   ~ool
  OUTPUT
                                                                                       Jl  DC CONTROL
                                                                                          VOlTflS-E
                                                                                         (JNBlOS
                                                                                         VOLTflSE
                                                                                         OUT
                                                                                    XBL  7912-13754
       Figure 17. Circuit diagram of 11  kilowatt power supply.   This power supply  is
                   used to heat  the ZAA furnace and  heated sample transport line.

-------
"Prime Power Circuit Breaker."  A front panel toggle switch selects
"standby" or "operate" modes.  An output current meter and the reactor
current meter are also located on the front panel.  The reactor current
provides a measure of the control signal from the remote control module
described above.
    A diagram showing the NIM electronics rack, the ZAA, the power supplies
and the connections between these units is provided in Fig. 9.
                                      38

-------
                                  SECTION 5



                  DESCRIPTION OF MERCURY MONITOR GAS SYSTEM



    A gas sampling and metering system has been designed and built which



will enable the ZAA spectrometer to be located from 20 to 30 meters from
                   \

the sampling point without loss of mercury.  This sampling and metering



system and the dynamic calibration system are described below.



OVERALL FUNCTIONAL DESCRIPTION



    The gas sampling and metering system is shown schematically in



Fig. 18.  Sample gas, e.g., retort offgas, of a given temperature, pressure



and mercury density (T, P, p, respectively) enters the heated sample tran-



sport line at a volumetric flow rate, q.  The heated transport line is



maintained at approximately 310°C by joule heating.  Oxygen (Tn ,


                                                               .
Pn , qn ) is continuously introduced at point A (Fig. 18) and mixed
 U2   U2


with the sample gas.  The oxygen is introduced just upstream of the ZAA



furnace to promote combustion of organics in the combustion chamber of the



furnace which is maintained at 900°C.  During calibration, mercury vapor in



a carrier gas (T-, P~, q~, p-) is introduced into the sample line



at point A.  Dilution gas (Tp, PQ, qQ) can also be introduced at  the



same point.



    The sample and calibration gases enter a gas mixer located just



downstream of point A and then pass into the furnace where they are heated



to 900°C to atomize the mercury.   The density of mercury atoms in the



furnace is then measured and converted into a voltage response as described
                                     39

-------
                                                       Q (Retort offgos flow rate)
                Oxygen (Tfl2,P02,q02)-^
                   Meter
                   run
                                        ^Heoted somple tube
                                              (200°c)

                                          Point A
                                                                 gas pipe
                    Calibration gas
                             Cooling coil



                            Vane pump
      '•Dilution gas


      Hg light beam


Zeeman spectrometer furnace (Tf'.



     -Servometering valve
                                                       XBL792-480
Figure 18.  Schematic  of gas  handling system for ZAA  mercury monitor.
             Temperature T, pressure P, mercury  density p,  and flow rate q are
             shown for  the various  gas streams,   pf  and qf  refer to the
             mixture of gases  flowing towards the furnace.
                                           40

-------
above.  Since the density of the sample gas and, thus, the voltage response


of the ZAA to a given concentration of mercury vary inversely with

temperature, it is essential that the temperature of the furnace be

maintained within ±25°C between calibration runs.


    The volumetric flow rate through the furnace (q°M) is metered by the

use of a flow controller.  The controller consists of a flow sensor,


solenoid valve, and electronics package.  Periodic calibration of the flow


controller is necessary since the measurement of flow by the sensor depends


upon the specific heat of the sample gas, which varies during the course of
                            i
a retort run.


    Table 1 summarizes the gas parameters which must be monitored during

the analysis of mercury.
Table 1.  Required measurements for gas handling system.

Off gas
Oxygen
Calibration gas
Dilution gas
Meter run
Ambient conditions
Temperature
T
TO
TC
TD
TM
Troom
Pressure Flow
P
P0 %
PC ic
PD %
PM q°M
Patm
                                      41

-------
DESCRIPTION OF INDIVIDUAL COMPONENTS
Heated Sample Delivery System
    The sample delivery system consists of a probe which is inserted into
the retort offgas pipe and a transport tube between the probe and the ZAA
furnace.  Both are heated to prevent the loss of mercury during sampling.
Heated Sample Probe
    The heated sample probe and probe gate valve assembly are shown in
Fig. 19.  The probe consists of two concentric 316 SS tubes insulated from
each other by Teflon spacers.  The gas sample passes through the inner tube
while the outer tube provides support.  The ODs and wall thicknesses of
the inner (sample) and outer (support) tubes are 0.635 x 0.089 cm and
1.905 x 0.089 cm, respectively.  When inserted into the offgas stream, the
probe is supported and aligned by the Teflon bushings on either side of the
gate valve.  The probe is held in position by the lock nut on the valve
      \
assembly which compresses an 0-ring against the outer support tube.  The
loss of retort offgas to the atmosphere is prevented by this locking 0-ring
and the 0-rings in the Teflon bushings and spacers between the sample and
support tubes.  The gate valve and a ball valve on the outer end of the
sample tube allow the probe to be inserted or withdrawn during retorting
operations without leakage from the offgas line.
    The sample tube inside the probe is maintained at a temperature of
300°C by joule heating using a 60-amp, seven-volt AC power supply.  A
No. 10 Formvar-coated copper wire passes through the spacers separating the
sample and support tubes and makes electrical contact with the sample tube
near its upstream end.  The spacers insulate the wire from the support tube
and the retort plumbing system.  The stainless steel sample tube forms the
                                      42

-------
                       -PTFE NOSE PLUG
         SAMPLE TUBE-
         OFF GAS
          (3 PSIG)
                PTFE PLUG

           -'O' RING
              LOCK NUT
                T.C. NsZ-
SUPPORT TUBE—

 SPACER (3)
Crl
     BUSH/NG te)

      VALVE

/-/FA TED  SA MPL E

        FJG. /9
                                                   TO CURRENT
                                                   CONTROL
                            FORMVAR COAT£D.
                                                              GAS  SAMPLE
                                                              'TO  ZAA
                                 -BALL VALVE
                                TC. Ma 2
                                1C. Nd 2.
                                                                                            XBL 798-11134
                 Figure 19. Heated  sample probe.  The probe can be  inserted into or withdrawn
                            from the  offgas plumbing system by using the gate valve and  small
                            ball valve.  The sample tube is heated  by passing current through
                            the tube.  The tube is insulated from the retort plumbing system by
                            teflon  bushings and supports.

-------
high resistance leg of the copper wire-tubing circuit.  Temperatures of
300"C have been maintained for three days without failure of the 0-ring
seals.  Two thermocouples are mounted on the probe assembly.  One is
spot-welded to the sample tube and measures sample tube temperature.  The
other is located at the tip of the probe and measures the temperature of
the offgas at the point of sampling.
Heated Sample Transport Line
    The 304 SS transport tube runs between the probe and the mixer and is
heated by passing an AC current through the walls of the tube.  The tubing
is 0.953 cm OD with a wall thickness of 0.165 cm.  One of the 11-kilowatt
power supplies described above is used to heat the transport tube.  The
supply's maximum current and voltage outputs are 650 amps and 17.5 volts,
respectively.  The tubing diameter and wall thickness (i.e., resistance/
meter) were chosen so that the power supply can deliver adequate heating
current (approximately 130 amps) to a 7.5 meter section of tubing before
the power supply output becomes voltage-limited.  With the tubing insulated
by a 7.6 cm diameter fiberglass sleeve, a 7.5 meter section of tubing can
be maintained at 450°C with ambient temperatures as low as 6°C and four
liters/minute of air passing through the tube.  The output of the power
supply under these conditions would be approximately 130 amps at 17 volts.
The power supply is capable of keeping at least 30 meters of insulated
tubing at approximately 450°C (ambient 6"C) by paralleling the electrical
connections of four 7.5 meter sections.  When operating at lower ambient
temperatures or where wind chill factors are significant, either the length
of the heated sample transport tube would have to be reduced or the output
of the power supply increased to achieve tubing temperatures of 450°C.

-------
Heated Gas Mixer Assembly


    Oxygen and mercury calibration gas are introduced into the sample gas

stream just ahead of a small mixing chamber.   This chamber is located


approximately one meter upstream of the ZAA furnace and insures that a

uniform mixture of gases enters the combustion chamber of the furnace.  The


mixing chamber is a stainless steel cylinder, 5 cm in diameter, with


flanged ends.  The 0.953 cm OD heated transport tube fits into the stain-


less steel tube connectors welded into the flanged ends.  The chamber is


13 cm long and filled with Pyrex glass beads which introduce turbulence

and, thus, promote mixing.  The same current which flows through the heated


transport line also flows through the mixer.  However, this current is not

adequate to heat the mixer to a high enough temperature (350°C) to prevent


loss of mercury.  A nichrome wire heating coil is mounted around the mixer


to increase the temperature.  Using a 3000-watt Variac, mixer temperatures


in excess of 500°C can be obtained.


Gas Metering System


    The flow controller measures and controls the gas flow rate and


provides a digital readout of the flow rate in cubic centimeters per minute

at standard conditions (scc/min).  The control range is 100 scc/min to
                               !

5000 scc/min.  The controller consists of four units:  an in-line mass flow


sensor, an in-line servo metering valve, an electronics module, and a


digital voltmeter (DVM).  The sensor and servo metering valve are located


downstream of the furnace.  Those portions of the sensor and valve

assemblies which are in direct contact with the offgas stream are con-


structed of 316 SS or Teflon to minimize corrosion.  An in-line paper


cartridge filter is used upstream of the sensor and the valve units to
                                      45

-------
reduce clogging by participates.  The electronics module is mounted in the
NIM rack with the DVM in an adjacent panel.  A wiring diagram is shown in
Fig. 20.
    The command potentiometer on the front panel of the control  module
adjusts the flow rate.  The actual flow, as measured by the sensor, is
continuously compared to the desired flow set by the command potentio-
meter.  The difference signal drives the servo valve in the appropriate
direction until the difference signal is reduced to zero and the desired
flow is achieved.  The command potentiometer setting corresponds to twice
the actual flow in scc/min of air; that is, a command potentiometer setting
of 6400 corresponds to a flow of 3200 scc/min.  As the flow ranges from 100
to 5000 scc/min, the DC output signal from the controller varies linearly
from 0 to 5.00 volts.
    The function switch on the front panel of the control module selects
the command potentiometer setting (position A) or the actual flow
(position B) for the DVM display when the DVM is set at 20 V DC full
scale.  During normal operation, the function switch is in position B.
Position A is used for adjustment of controller electronics (see
Section 7).  If the system flow rate is less than the corresponding command
potentiometer setting, the servo valve will be held in a "wide open"
position and the DVM will provide a reading of the actual uncontrolled flow
rate.
    The flow controller maintains a flow which is dependent upon the
specific heat of the gas.  When retort offgas is being metered, the DVM
flow rate reading can be set to the correct value by adjusting the cali-
bration control on the front panel so that the reading agrees with values
                                      46

-------
                           FLOW CONTROLLER WIRING DIAGRAM
         Signal from sensor
          •Signal to valve
Controller
circuit
Command pot
                                 3O
                                 AO
                           (Command pot signal)
                                Function
                                select
                                switch

                                                               Calibration pot
             (Flow signal)
                                                             X2
                                                    To DVM
                                                           Non - inverting amp

Figure 20. Flow controller  wiring diagram.  The command pot  located  on the front panel of the
           controller module  determines the gas flow through the  ZAA spectrometer.  With the
           function  switch  in position A the DVM readout equals the  pot setting which is approx-
           imately twice  the  flow.   In position B the DVM readout is adjusted by the calibration
           pot to be equal  to volumetric flow rate at standard conditions.

-------
obtained using a Wet Test Meter  (WTM).   When  air  is  being metered, the
calibration control dial should  be  set  to  approximately  5000.  The exact
value should then be determined  using a WTM.
Calibration System
    A dynamic calibration system which  generates  known concentrations of
mercury vapor in a carrier gas  (air) is  used  to calibrate the gas monitor.
This system, which is shown  in Fig. 21,  is  based  upon the apparatus
described by Nelson.    Heated air  impinges on the surface of a pool of
mercury warmed to about 60°C.  The  mercury  laden  gas proceeds through two
successive equilibration vessels; excess mercury  condenses, and the gas
leaves the vessel saturated  with mercury at an accurately determined
temperature.  The saturation mercury vapor  pressure  is obtained from
standard tables as a function of the measured temperature.  The saturated
gas is then diluted with mercury-free gas  (air) and  introduced as the
calibration gas into the sample  line.
    The individual components of the calibration  device  are constructed of
Pyrex glass and are joined using Teflon  unions.   The measurement of the
saturated mercury vapor temperature is  made with  a precision mercury
immersion thermometer and a  T-type  thermocouple located  directly in the
exit gas stream.  Both devices are  calibrated against National Bureau of
Standards (NBS) traceable platinum  thermometers and  are  accurate to within
±0.1 between 15°C and 45°C.  To  maintain a  relatively constant mercury
concentration at the outlet, the entire  system is mounted inside a box
filled with vermiculite for  thermal insulation.   Rotameters used for gas
metering (Fig. 21) are equipped  with high  precision  metering valves and are
                                      48

-------
Heating  X
mantle
Jf
Flow— ' •
meters
(Rotamete




f
rs)

s*




f


                             Equilibriu m
                             vessels
                                                      Dilution gas
                                                      (qD,TD, PD)
                                                         Mixing
                                                         chamber
                                                Carrier gas saturated
                                                 with Hg  vapor
                                                 C'qC> TC' PCJ
                                                  XBL793-792
Figure 21. Schematic of mercury calibration system.

-------
calibrated using a WTM and bubble meters at constant rotameter  inlet
pressures.
    With this system,  it is possible to obtain calibration  gas  with mercury
concentrations which range from 0.01 mg/m  (1 ppb)  to approximately
       o
20 mg/m  (2 ppm) by varying the flows of mercury saturated  gas  to
dilution gas.
CALCULATION OF Hg DENSITY FOR CALIBRATION
    To calibrate the ZAA, the voltage response of the instrument must be
related to a known density of Hg atoms (PF) in the  furnace.  To calculate
Pp, it cannot be assumed that the sample and calibration  gases  will be at
the same initial temperature.  Therefore, the Hg densities  in both sample
and calibration gases  must be corrected for temperature differences.  In
addition, the amount of dilution of sample gas by oxygen  and calibration
gas must be determined.  To calculate PF, we have assumed that  the ideal
gas law adequately describes changes of state in sample and calibration
gases.  Density and volumetric flow can then be converted to standard
conditions (760 mm Hg, 0°C) using equations (1) and (2):
                  / T \   /760
                  (m)   rr
                 /273\   / P \
                 VTV   VTW
where T is temperature in degrees Kelvin,  P is pressure  in  mm Hg, p is
the density of mercury in mg Hg/m ,  q is the volumetric  flow rate in
m3/min, and the superscript zero designates standard conditions.
                                      50

-------
    The density P|_ entering the furnace is the sum of the flow-weighted
mercury densities in the sample and calibration gas streams.
                                                                (3)
where

         qF ~ q    qC   qD   q02

The measured quantities qc, qQ and qQ , defined in Section 5,
page 39, are converted to standard conditions using equation (1) and
their corresponding measured temperatures and pressures.  The standard
flow rate of sample gas, q , is not determined directly but is
obtained by the difference between q:. and q^ + % + qO  '  Jt 1S assumed
here that the reaction of CL with the offgas in the furnace does not
significantly affect the accuracy of this indirect measurement of q .  An
attempt will be made to test the validity of this assumption during future
field testing.
    With the calibration system turned on and the sample gas diverted, the
density of mercury entering the furnace becomes, from equation (3),
                                                                 (4)
                                                                 l  J
The mercury density  in the calibration gas at standard conditions,  PC,  is
calculated using equation.(1);
                                      51

-------
           0
          PC =  PC   \T73
where
v  = (3.22xlOu)   LliiM
                V *C  / \
                                   ™
The measured temperature of the calibration gas  is T~, and P(Hg) is the
vapor pressure of mercury at TC obtained from  standard tables.
    The relative error in determining  pp,  as defined by equation 4, is
±4% over the calibration range specified above.  This is based upon actual
conditions where T~ is measured to within ±0.1"C, other temperatures are
measured to within ±0.2°C, and both pressure and flow are measured to ±1%
or better.
    As noted above, a calibration curve can be obtained by varying the
mercury calibration gas and- dilution gas ratio and recording the ZAA
voltage response.  When the calibration system is turned off and the sample
gas is reintroduced, the calibration curve is  used to determine the unknown
mercury density, p , in the sample gas.
    However, during analysis of the sample gas,  the mercury density
entering the furnace (pH) must be corrected for  dilution caused by the
introduction of CL at point A.  From equation  (3), we have:
                                       52

-------
    An alternate calibration procedure is to inject  the calibration gas
directly into the sample gas.  The concentration P^  in this case is
given by equation (3).  If matrix effects are significant,  this method of
addition can be used to determine the unknown concentration in the sample
gas.  The standard addition curve can then be used after a  dilution
correction as a calibration curve for subsequent measurements if the gas
composition, matrix effects, furnace temperature, and the sample gas flow
rate remain constant with time.
                                       53

-------
                                  SECTION 6



     PRELIMINARY EVALUATION AND PERFORMANCE  OF MERCURY GAS MONITOR SYSTEM



                               ZAA CALIBRATION



    The preliminary tests which were used to  determine the  linearity,



precision, and  accuracy of the mercury ZAA spectrometer and calibration



system are described.  These tests were conducted  in  the  laboratory with



ambient temperatures of 20° to 25°C and furnace temperatures of 900° to



950°C.  During  any single day of testing, the furnace temperature remained



constant within ±10"C.



    Compressed  air was used as the carrier gas for the calibration system,



as dilution gas and as the test (sample gas)  during instrumental evalua-



tion.  The level of mercury in the compressed air  was below the ZAA



detection limit of 2 ppb.  A constant quantity of  oxygen  (60 scc/min) was



added to the test gas.  Test and calibration  gases were fed directly into a



heated (300°C)  5 cm length of stainless steel tube connected to the input



end of the ZAA furnace.  Rotameters calibrated with a WTM or bubble meters



were used to measure gas flow rates.



    Mercury concentrations between 0 and 600  ppb were obtained by varying



the flow of carrier gas through the calibration system while maintaining a



constant test gas flow rate.  Concentrations  above 600 ppb were obtained by



varying the carrier gas flow rate and decreasing the  test gas flow rate.



    As noted in Section 2, ZAA response begins to  saturate above 400 ppb



when the 18 cm  (absorption tube length) furnace is used.  This is shown" by





                                      54

-------
the curve in Fig. 22 which was obtained for a test gas flow rate of
800 scc/min, calibration gas temperature and mercury saturation vapor
pressure of 25.0"C and 1.83 x 10~3 mm Hg, respectively, and calibration
gas flow rates varying between 10 and 275 scc/min.  The response is linear
between 0 and 100 ppb but becomes non-linear between 100 and 200 ppb.
Non-linearity increases above 200 ppb and the response becomes flat, within
the precision of the measurement, between 450 and 500 ppb.
    The precision of the instrument with the 18 cm furnace installed was
determined by making 10 replicate analyses of air containing 215 ppb
mercury.  This was accomplished by repeatedly turning off the calibration
gas and then resetting the flow to obtain the calculated 215 ppb of
mercury.  These measurements were made over a four-hour period.  The
average ZAA response and the standard deviation for these ten replicates
was 53±3 absorption units (All), with a range of 50 to 57 AU.  These
arbitrary units are proportional to the voltage response of the ZAA.
    The 18 cm furnace provides an excellent response below 250 ppb as
calibration curves in Figs. 23 and 24 demonstrate.  These were obtained for
test gas flow rates of 2000 scc/min and 3000 scc/min, respectively.  Ten
replicate samples containing 75 ppb (3000 scc/min) and 150 ppb
(2000 scc/min) were analyzed as described above, and the precisions
(standard deviation) were found to be ±6% and 7%, respectively.
    Above 250 ppb of mercury, a 5 cm furnace should be used.  A calibration
curve obtained with this furnace is shown in Fig. 25.  The voltage response
is linear up to 0.8 parts per million.  Ten replicate measurements at
0.80 ppm and 1.34 ppm yielded precisions of ±8% and ±10%, respectively.
This calibration curve was obtained by varying the dilution gas flow rates
                                      55

-------
                         CALIBRATION  CURVE  FOR 18cm FURNACE
en
              80-
           CO  60-
           H
           Z
           ID

           Z
           o
           Q_
           a:
           o
           c/>
           CD
              40-
               20-
                          DILUTION GAS FLOW RATE qD= 800scc/min

                          OXYGEN FLOW RATE q0 = 60scc/min.

                          FURNACE TEMPERATURE = 930°C

                          LIA GAIN = 0.1
                 0
100          200          300

   MERCURY CONCENTRATION (ppb)
400
500
                                                                              XBL 801-83
         Figure 22. ZAA calibration curve obtained for 18 cm furnace.  For mercury concentrations in
                  excess of 450 ppb the curve is flat within experimental errors.

-------
                           ZAA CALIBRATION CURVE - 18cm FURNACE
en
            50-1
            40-1
        C/)

        2
O
I-
        o
        to
        CD
            20i
             101
                             DILUTION GAS FLOW q = 2000scc/min.
                             OXYGEN FLOW q  = 60scc/min.
                             FURNACE TEMPERATURE = 9IO°C
                             LlA GAIN = 0.10
                             50
                                     100
150
200
250
                                       MERCURY CONCENTRATION (ppb)
                                                                                     XBL 801-82
           Figure 23. ZAA calibration curve obtained for the 18 cm furnace at a sample gas flow rate
                     of 2000  scc/min.   The curve is linear below 200 ppb Hg,

-------
Ui
CO
                    10
                           ZAA  CALIBRATION  CURVE - 18cm FURNACE
                                 DILUTION GAS FLOW RATE qD = 3000 scc/mia
                                 OXYGEN FLOW RATE q0z= 60scc/min.
                                 FURNACE TEMPERATURE = 920°C
                                 LIA GAIN = 0.50
20      30      40      50      60

     MERCURY  CONCENTRATION (ppb)
70
80
90
                                                                                     XBL 801-81
     Figure 24. ZAA calibration curve obtained for  18 cm furnace at a sample gas flow rate of 3000 scc/min,

-------
          lOOn
           80-
                  ZAA CALIBRATION  CURVE- 5cm FURNACE
Cn
           60-
O
|_
Q_
o:
O
O)
m
<
                                                     FURNACE TEMR935°C
                                                     LI A GAIN  0.10
                    0.20    0.40     0.60     0.80     1.00     1.20


                              MERCURY CONCENTRATION (ppm)




                       Figure 25. ZAA calibration curve for 5 cm furnace.
                                                          1.40     1.60
                                                          XBL 801 - 80A

-------
between 100 and 825 sec/min and the calibration gas flow rate between 25
and 250 scc/min.
    A subsequent test using the 18 cm furnace was made to determine:
(1) whether the density of mercury in the calibration gas was independent
of the gas flow rate through the calibration device and (2) whether the
residence time of the mercury in the atomization chamber in the furnace
affected the ZAA response to a fixed mercury concentration.  It is useful,
in order to understand this test, to keep the following relation  in mind:
                 /   qc   \
         VZAA    (qp-n^J  PHg
                   u
Here V^,,,, is the ZAA voltage response recorded  in  absorption  units; qp
and qp are the calibration and dilution gas flow rates;  and p..   is the
calibration gas mercury density.  In Fig. 26 the results of the  test are
compared to a calibration curve obtained using  a dilution  gas flow of
2000 scc/min.
    After the calibration curve was obtained, the  dilution gas flow rate
was set at 3000 scc/min and the calibration gas flow  rate  was set at
250 scc/min.  The response at these conditions  corresponds to point A in
Fig. 26 which, within the errors of measurement, falls  on  the calibration
curve.  Next, the calibration gas flow rate was set at  50  scc/.min.  The
response at these conditions corresponds to point  B which  within the
experimental errors also falls on the calibration  curve.   The fact that
points A and B lie on the calibration curve  implies that the  calibration!
gas is saturated with mercury even at flows  as  high as  300 scc/min.
                                       60

-------
                               ZAA CALIBRATION 8 FURNACE  TEST
                                            18cm FURNACE
                50-
CTl
•z.
o
Q_

-------
    Finally, the dilution gas flow rate was reduced to 600 scc/min, thus
returning the mercury concentration entering the ZAA to the value it had at
point A.  The response at this condition corresponds to point C in
Fig. 26.  In going from condition A to B, there was essentially no change
in the residence time of the sample gas in the furnace combustion chamber.
However, the change from B to C increased the residence time from 0.2 sec
to 2.0 sec.  If the atomization of mercury at A were incomplete, point C
would not coincide with A in Fig. 26.  In fact, points A and C do coincide,
within experimental errors, indicating that the variation in flows during
routine calibration have no observable effects upon the ZAA response.
    These results indicate that the design and temperature of the
combustion chamber are adequate to insure the complete atomization of
mercury.  Nevertheless, it is possible, although unlikely, that under-
saturation or oversaturation of the calibration gas combined with
incomplete atomization would yield the results just described.
LIGHT SOURCE
    By far the most temperature-sensitive component of the spectrometer is
the light source.  This sensitivity can be a problem in field applications
where significant temperature fluctuations are likely to occur.  The change
    ;                        o
in the intensity of the 2537A line with variations in the light source
temperature is shown in Fig. 27.  Between 12°C and 31°C the intensity
increases by a factor of three (0.8 to 3.0 volts) due to an increase in the
mercury vapor pressure within the lamp.  However, the ZAA response to a
constant mercury concentration remains stable within measurement errors
over this temperature range (Fig. 28).  Stability is achieved by routing
the PMT signal through a log amplifier before  it enters the tuned amplifier
                                      62

-------
CM
       B   3-°
CD
JT  2.0

"§   1.5

2   L°
°-  0.5
             0
                   Relative intensity change in 2537A Hg line
                           vs. temperature of Hg lamp
                      I    I   I    I    I
              468
                                    I    I    I   I    I    I
                  I    i   i    i	i   i    i    i   i
j	i
                  10  12  14   16  18  20 22  24 26  28  30 32
                    Hg lamp temperature  (°C)
                                                            XBL 793-869
            Figure 27. Change in intensity of gaseous mercury discharge lamp with temperature.

-------
                        ZAA signal vs. temperature for

                  .22mgHg/m3 (I22ppb) in sample gas
X—*
£
o
%i
"o
c
o>
•«v
_5
j?"
o
1"
o
c

o
0
_J



0.9

0.8

0.7
0.6

0.5

0.3

0.2

O.I

0
i i i i i i i i i I i 1 i I
_ 	 ,

_ _

— _
• * • *
X

_ / _
i /
Corrected for baseline shift -

— —

— —

i i i i i i i i i i i i ii

4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
                      Hg lamp temperature  (°C)
mi 793-870
Figure 28. Temperature dependence of ZAA response to a constant concentration of mercury.
         The decrease in response at 6°C is an experimental anomaly caused by the defocusing
         of.the light beam by water droplets condensed out on the outside of the light
         source window at this temperature.

-------
section of the lock-in amplifier.  This electronic processing effectively
filters out the effect of light intensity changes due to changes in
temperature.
    However, there is another temperature effect which is not filtered out
by the electronics.  The relative intensity of the •* and o lines is
altered by self-reversal or self-absorption of these lines within the
plasma of the  lamp.  This effect, which increases with temperature, mani-
fests itself as a change in instrumental baseline voltage and, thus, is
indistinguishable from the signal produced by mercury in the sample gas.
The magnitude  of this effect is shown  in Fig. 29.  In the absence of
mercury, a 12-31°C change in temperature produces a 220 millivolt ZAA
voltage as shown by the lower curve in Fig. 29.  The upper curve shows this
change in parts per billion (ppb) of mercury.  If the lamp is operated at
25°C, a variation of ±2°C produces a ±6 ppb error, which is significant
when measurements are below 60 ppm.  Temperature control of the light
source was not possible with the old EDL.  However, with the new PLL, the
problem has been eliminated by enclosing the  lamp in the water-jacket
assembly described above and coupling  it to a small constant-temperature
bath mounted within the instrument.  This arrangement allows the temper-
ature of the PLL to be controlled to within ±0.2°C, which is equivalent to
an error of only ±0.5 ppb of mercury.
    The PLL source offers additional advantages.  The maximum intensity of
the 2537A analytical  line obtained with the PLL  source is approximately 50%
greater than that obtained with the EDL source.  This enhances the  instru-
ment's ability to measure mercury in the presence of large amounts  of
smoke.  Another advantage of the PLL is the absence of radio frequency
                                      65

-------
en
CTl
                        Shift in baseline vs. temperature of Hg lamp
                         For ZAAgainto
                         detect Hg in 5 to
                         200 ppb range
                     6   8
10  12  14  16  18  20 22 24  26 28 30 32
  Hg lamp temperature (°C)
                                                                      XBL 793-868
         Figure 29. Change in ZAA output voltage due to temperature-induced self-reversal  in the
                  absence of mercury in the sample gas.  The lower curve shows this effect in terms
                  of baseline voltage.  The upper curve shows this effect in terms of apparent
                  mercury concentration.  Both curves are normalized to 6°C.

-------
pickup in adjacent instruments, such as thermocouples and pressure

transducers.

DETERMINATION OF HEATED SAMPLE TRANSPORT LINE OPERATING TEMPERATURE

    The experimental setup shown in Fig. 30 was used to establish the

temperature above which the heated transport line must be maintained to

prevent a loss of mercury to the tubing walls.  The test gas (calibration

plus dilution gases) could be routed directly into the ZAA (Path A) or

through a heated 7.5 m stainless steel transport  line before entering

the ZAA (Path B).  Fiberglass insulated 304 SS tubing (0.95 cm OD,
                         I
0.165 cm wall) was used for A and  B.  A 0.7 m  length of the same tubing,

maintained  at 310°C, was used for  Path A.  Two heated stainless steel ball

valves were used for switching between A and B.   The average temperature of

the transport line (Tg) was determined using three thermocouples.

    ZAA responses for paths A and  B were compared  as a function of the

heated transport line temperature.  The data are  summarized in Table 2.

For each run the ZAA response shown in the table  is  an eight- to ten-minute

time average.  Runs one  through seven  show no difference in ZAA response;

thus, maintaining the transport line at 440°C prevents a loss of elemental

mercury.  Runs eight through twelve show that as  the temperature of the

line was decreased, the  ZAA response also decreased.  At TB = 310°C, a 7%

decrease in the  ZAA response had occurred.  This  is  approximately equal to

the precision of the measurement.  Therefore, the  transport line should be

operated at a temperature xS
                                      67

-------
         HEATED TRANSPORT LINE TEST ARRANGEMENT
    Calibration
      and
    dilution devise
                        «
A
t
                             Path "A"
                             0.7m @ 310° C
          -
                                                      Path "B"
                                                      7.5m heated and
                                                      insulated tube
                   ZAA
                             Thermocouple

                             Stainless
                             steel
                             ball valve
                                                        XBL801 -55
Figure 30.  Heated transport line test arrangement.  The ZAA response to
           a fixed concentration of mercury was compared for path A and
           B is the temperature B was varied.
                                 68

-------
Table 2.  Heated transport line experimental  data.
                            Hg             TB          ZAA response (All)
Run    Total flow      concentration
No.    (scc/min)           (ppm)           (°C)        Path A       Path B
1
2
3
4
5
6
7
8
9
10
11
12
200
200
200
200
925
925
925
200
200
200
200
200
1.21
1.21
1.21
1.21
0.27
0.27
0.27
1.21
1.21
1.21
1.21
1.21
440
440
440
440
440
400
400
466
375
310
300
208
67
66
68
67
31
31
33
69
68
69
68
68
69
68
67
68
32
30
32
68
67
64
63
60
FLOW CONTROLLER
    The linearity  and accuracy of the flow controller was determined using
the experimental setup  shown  in Fig. 31.  A constant back pressure of
ten psig was maintained over  the entire range of  volumetric flow rates.
The flow controller  digital voltmeter  (DVM) reads  directly  in cubic centi-
meters of flow per minute  at  standard conditions  (scc/min).  These readings
were compared with volumetric flows measured with  a WTM corrected to
standard conditions.  The  calibration control potentiometer (see Section  5)
was set at 4.93  to correct the DVM reading so that it agreed with the WTM
at a flow of 2500  scc/min.  The command potentiometer was used to set the
flow.  At each setting, three WTM readings were made.  The results are
shown  in Fig. 32.
                                      69

-------
                               FLOW CONTROLLER TEST ARRANGEMENT
               Pressure
               gauge
Compressed air and
pressure regulator
                                                               Wet test
                                                               meter
                                                                      XBL801 -54
              Figure 31.   Flow  controller test  arrangement.
                                        70

-------
       COMPARISON OF FLOW CONTROLLER
               AND WET TEST METER
              FLOW CONTROLLER SET TO GIVE
               ZERO ERROR AT 2500 sec /min
               CALIBRATION CONTROL = 4.93
i- OVJWW
"^
0
0
^4000
CJ>
%
o
£ 3000
£
o
x«^
H^
*- onnn
CD *-UUU
"55
> 1000"
^
•—
o o










->-


/













f
i i i i /
/L-
/VHr
yO.2%low
/
/
o
/

/
2% high/
/
/
/
' -i i i i
Figure 32.
0      1000    2000   3000   4000  5000

      Wet test meter flow (scc/min)

                                   XBL801 -57

Comparison of flow controller and wet test meter (WTM) flow
readings.  With the calibration pot set to agree with WTM at
2500 scc/min the error in controller reading was 0.2% low and
2% high at 4500 scc/min and 700 scc/min.  The correct (WTM)
values at these two points are indicated by the dotted lines.
                         71

-------
    The controller is linear over its entire range with a slope that is
approximately unity.  The error at 2500 scc/min was defined as zero by
using the calibration pot.  Away from this point, error increases.  At
250 scc/min the DVM reading is 2% low, and at 4500 scc/min the DVM reading
is 0.2% high.
    Greater accuracy (±0.2%) over a narrow range can be achieved by setting
the DMV to the WTM reading with the calibration pot.  The calibration pot
has a wide range of adjustments so that it can be used to correct for
changes in gas composition.
CORROSION TEST AND ESTIMATE OF FURNACE LIFETIME
    The hydrogen sulfide and oxygen concentrations, or more precisely the
chemical activity (partial pressures) of S~ and Op in oil shale offgas,
determine the extent of the potential corrosion problem.  The other gaseous
constituents are not directly involved.  Stainless steel is usually
oxidation-resistant due to the formation of a protective oxide layer of
CrpO,.  However, at low-oxygen partial-pressure and especially when the
environment also has a significant sulfur partial pressure, Cr^Oo is
less protective.  Sulfidation may take place simultaneously with oxide
formation and, since sulfide growth rates are generally orders of magnitude
greater than those of the corresponding oxides, corrosion results.
    Oil shale offgas may seriously corrode the high-temperature portion of
the stainless steel  furnace due to the presence of HLjS and the consequent
sulfidation reactions.  Sulfidation reactions refer to the formation of
chromium-rich sulfides and at higher temperatures the formation of nickel-
nickel sulfide and iron-iron sulfide eutectic liquids referred to as slag.
                                      72

-------
With the formation of these eutectic liquids, corrosive attack is generally
catastrophic.
    Alumina, A1203, offers a more effective protective oxide.  It forms
more easily than Cr203 when the oxygen partial-pressure is low, is more
stable than CrJD., at higher temperatures, is more resistant to inward
              <_ 0
sulfur diffusion.  Thus, to extend the lifetime of the 321 SS furnance,
aluminum has been diffused into the surface of the SS tubing by a process
termed alonization.  The resulting microlayer of aluminum, which is sub-
sequently converted to alumina, is effective in reducing the rate of
                                                                  I O
corrosive attack on stainless steels due to sulfidation reactions.
    The 321 SS is an austentic solid solution whose elemental composition
(% weight)  is as follows: iron 66-71%, chromium 17-19%, nickel 9-12%,
manganese 2%, silicon 1%, and carbon <1%.  The tubing was alonized by
packing it  in a bed of powdered Al and Al-CU plus an activator and
                                         C- O
heating the bed to 950°C for 15 hours in a reducing atmosphere of H2.
The aluminum diffusion layer, which is highly enriched with  aluminum as an
aluminide (e.g., A1C1), is approximately ten microns deep on both inside
and outside surfaces.  All 321 SS tubing used in the following test and for
furnace construction was from the same manufacturer's lot.
    The purpose of the laboratory test was:  (1) to determine the resistance
of alonized 321 SS to sulfidation, and (2) to obtain a qualitative estimate
of the lifetime of an alonized furnace.
    The test was conducted by hanging segments of alonized tubing in a
ceramic chamber maintained at 1093°C for a 65-hour exposure  time.  The
segments were 1.25 cm lengths of tube which had been cut in  half length-
                                              \
wise.  The  composition of the input gas  (% volume) was: H2 83.6%, H2S

                                     73

-------
1.8%, H20 1.8%, and argon 12.8%.  The flow rate of this mixture was
200 scc/min.  The gas was preheated to 1093°C before entering the chamber
which was maintained at one atmosphere total gas pressure.
    After termination of the experiment, the segments were cut, mounted  and
polished, then subjected to metallographic examination.  Figs. 33a and 33b
show a cross-section of alonized tubing magnified 320 and 800 times,
respectively.  The two curved surfaces received the alonization treatment,
whereas the cut end did not.  It is evident that the alonization treatment
effectively inhibited slag formation.  However, slag formation at the end
of the tube was most certainly accelerated due to edge effects.
    The chemical activities of S2 and 02 at 1093°C must be determined
in order to thermodynamically predict the chemical compounds of metals at
the surface of the test sample.  Assuming that equilibrium conditions
                                                    14
prevail, the Gibbs standard free-energy of formation   for the reaction
H2 + 1/2 S - H2S is

         AG° = -RTlnK = -5530 Kcal/mole
where
               InK = In
                        PH  JPS
                         H2 V   2
Here PH <., p., , and p<-  are the equilibrium partial  pressures of H2S, H2,
and S2 respectively.  The ratio of pH s:pH  is equal to  the  volume percent
ratio of these two input gases.  Using this relationship we  can solve for
p<-  to obtain pc  = 10"    atm.  Similarly, the chemical activity
 5             ^
                                      74

-------
                                                            CBB 797-9668


Figure 33a.  Cross section of corrosion test sample:  magnification x320.
             The top and bottom surfaces are alonized.  The left end,
             obscured by slag, was not alonized.  Slag formation has
             advanced from the untreated end into the alonized region.
             On the upper right surface slag formation is slight while on
             the lower right surface no slag formation has occurred.
                                   75

-------
                                                         CBB 797-9667
Figure 33b.   Enlargement (x800) of slag nodule located on upper right
             hand surface of tube shown in Figure 33a.
                                 76

-------
of 02 is obtained from the reaction H2 + 1/2 02 » H20 where *G° =
-41125 Kcal/mole. 4  Solving for pn  yields a value of 1(T16'5 atm.
                                  U2
    A phase stability diagram for Fe, Cr, Ni, and Al as functions of
PS  and P0  at 1093°c 1s shown in F""g- 34.  This diagram provides a
theoretical indication of the surface metal containing phases which are in
equilibrium with the experimental gas mixture.  However, it does not
necessarily apply to the interior portion of the scale.
    The phase stability diagram  in the absence of Al models the situation
for the non-alonized end of the  sample.  The potential phases are Cr203,
CrS, FeS, and NiS.  The Cr2CL forms a thin protective surface layer.
However, this protective oxide layer  is apparently destroyed under experi-
mental conditions.  There are several likely reasons.  First, this initial
formation of a O^Cu layer and the formation of CrS depletes the Cr
concentration just below the surface.  Since 321 SS has barely enough Cr to
form the initial oxide layer, the protective layer would be thin and
perhaps patchy.  Thus, nickel and sulfur would be able to penetrate through
or around the protective oxide layer  and react above 645°C to form a
nickel-nickel sulfide eutectic liquid.  This eutectic slagging, which is
capable of mechanically lifting  the oxide  layer, would expose new
chromium-depleted surfaces to additional slagging.  Above 900°C, iron and
iron sulfide also form eutectic  slags which would contribute further to the
destruction of the original protective oxide layer.
    Once the protective layer is disrupted, slag formation would dominate
the oxide formation due to the Cr deficiency at or just below the surface.
The slagging boundary would thus penetrate deeper and deeper into the
                                      77

-------
                              PHASE  STABILITY  DIAGRAM AT  I093°C
       P, (ATM)
CO
                                                                                       -NOT TO SCALE
                                                                                       -FOR PURE METALS
                                                          10"
                                                      PQ  (ATM)
10''
                                                                                        XBL  7912-13744
         Figure 34.
Phase stability diagram at 1093°C.  Phase boundaries  are shown by solid lines, i.e.
the boundary  between Ni metal  and NiO occurs at Po2 = 10~10'8 atm.when Ps2  <  10~Gl1*
atm.  The test conditions are  shown by the hexagon and dotted lines.  For these
conditions  the dominant metal  phases are A1203, FeS,  Cr203 and NiS.  The diagram
was constructed for pure metals  and the phase boundaries are not drawn to scale.

-------
stainless steel, depending upon the abundance of Cr and the S2 and 02

activities.


    The slag formation of the non-alonized end of the tubing shown in


Fig. 33a is extensive.  The slag formation propagates laterally away from


the end due to penetration under the alonized surface layer.  This con-


trasts with the adjacent alonized surface which is still intact.  No


electron microprobe analysis has as yet been done on these slags to


identify their elemental composition; however, the above thermodynamic


arguments would favor nickel and iron sulfides.


    The alonized surface is highly enriched in Al.  Since Al is a reactive


element, the lateral rate at which the alumina (A1203) protective layer


is formed is extremely rapid.  Even at the extremely low oxygen partial


pressures of this experiment, Al^O, is the thermodynamically stable


phase, and Cr2CU will not form to any extent.  Due to the thickness and


integrity of the exterior Al-Oo layer and the high concentration of Al


below it, inward diffusion of S? and the outward diffusion of Ni and Fe


do not occur to any significant extent.  Thus, eutectic slagging is


effectively prevented.  The central alonized surface in Fig. 33a appears


relatively untouched in comparison with the non-alonized surface; however,


some minor attack is evident under higher magnification as seen in Fig. 33b.


    The effectiveness of the alumina coating can be destroyed due to


surface spallation  if extensive temperature cycling occurs.  Spallation is


the flaking of the oxide scales due to stresses induced during repeated


heating and cooling.  For this reason, it is recommended that furnace


temperature not be reduced during the course of an experiment by more than
               tr

200°C once the operating temperature has been reached.
                                      79

-------
    Any prediction of furnace lifetime under field conditions would be
risky on the basis of this single test.  However, the alonized 321 SS
surface, away from the cut edge, remained intact for 65 hours.  On that
basis we can venture to estimate that an alonized furnace could be used for
at least three days under conditions similar to those used in the present
test, which did not include temperature cycling.  The finite lifetime of
the alonized furnace does not limit the application of the ZAA monitor
since these furnaces are inexpensive and new ones are easily installed.
                                      80

-------
                                  SECTION 7

                            OPERATING  INSTRUCTIONS

ZAA SPECTROMETER


    It is assumed in this section that the instrument does not require

extensive optical alignment or electronic tuneup.  Power requirements are

115 VAC, ten-amp service for the instrument and 208 VAC, 30-amp service for

each furnace power supply.  An oscilloscope is required at several points

during the checkout procedure and during actual operation.  The scope must

be of the DC type, so that  it can display constant as well as time-varying

voltages.

Light Source


    First, turn on the power to the NIM electronics rack using the switch

located on the right front  of the rack.  Next, turn on the square wave

generator and the DC clamp/mixer located in the NIM rack.  Make sure that

the light source coupling transformer, located behind the permanent magnet

in the spectrometer, is also switched on.  Operation of the source can be

checked by placing a small  mirror between the  light source and the
                       o
furnace.  The blue 5461A mercury line should be clearly visible.  Turn on

the light source temperature controller and adjust to 25°C.  The light

source temperature is indicated on channel 4 of the digital trendicator

located below the spectrometer.  Allow 10-20 minutes for the lamp

temperature to stabilize.   Set oscilloscope channels A and B to DC mode.

Connect the BNC output labeled PMT monitor on the front panel of the log



                                      81

-------
amp module to channel A.  Set the potentiometer on the front panel of the
PMT high-voltage supply module to 7.00.  This setting is approximately
700 volts.  With a scope sweep rate of five milliseconds, the PMT output
will appear as an envelope or band on the oscilloscope screen.  The
discrete pulses of the square wave generator firing the light source at
700 Hz can be seen within this envelope.
    After the light source temperature has stabilized, adjust the PMT
high-voltage control so that the maximum amplitude within the PMT output
signal envelope is -0.6 volts.  This voltage should drop to zero when the
light path is blocked.  Two conditions should be noted.  First, the
potentiometer setting necessary to obtain the -0.6 volt PMT output will
depend upon the operating temperature of the light source.  Second, when
the instrument is properly tuned, the envelope of the PMT output is
modulated at approximately 49 Hz.
    Optically align the furnace-absorption tube assembly by making the
necessary horizontal, vertical,  and axial adjustments of the furnace
assembly so that the peak PMT is maximized.  Next, remove the quartz
windows on the end of the furnace-absorption tube assembly.  If there is
more than a 10-20 milli-volt increase in the negative PMT output signal,
the quartz windows are probably dirty and should be cleaned with ethanol
and lens tissue.  If a significant increase in PMT signal occurs in either
of these steps, readjust the PMT high-voltage control to obtain an output
of -0.6 volts and lock the potentiometer dial.
VariablePhase Retardation Plate(Squeezer)
    Using a X10 attenuation scope probe, connect channel B of the
oscilloscope to the BNC connector located on the squeezer inside the
                                     82

-------
spectrometer housing.  Switch the scope display mode to chop in order to



display both the PMT output on channel A and the squeezer input signal on



channel B.  Use the reference signal from the front of the audio frequency



generator (AFG) to trigger the scope trace.  Adjust the DC clamp control to



0.55, then tune the AFG by setting the frequency potentiometer to approxi-



mately 7.65.  Rotate the AFG amplifier knob counterclockwise until  it



stops.  Then turn it approximately 1/16 of a turn clockwise.



    Fine tuning is achieved by adjusting the DC clamp and AFG amplitude



control until a smooth sinusoidal modulation of the envelope, the PMT



output trace (channel A) is obtained.  The modulation frequency should be



between 45 and 55 Hz.  Flat spots in this modulated envelope may occur due



to excessive DC current or AFG amplification.  Flat spots should be elimi-



nated by decreasing  these settings  until a smooth trace just starts to



appear.  Adjust the  AFG audio frequency control until the amplitude of the



squeezer input signal (pure sine wave) is maximized.  The frequency of this



squeezer input signal should be  approximately 49 Hz.  Then  decrease the



frequency setting on the potentiometer by  approximately 0.20 units  and lock



the potentiometer.



    Improper tuning  of the squeezer can occur  in two ways.  First,  if the



magnitude of the input voltage used to drive the squeezer is too high, a



loud hum or chatter  will occur.  To correct this condition, the magnitude



of the DC voltage and/or the amplitude of the AC voltage should be



decreased.  Second,  the squeezer can  be tuned to its second harmonic;



however, this condition should be avoided  by decreasing the AFG "frequency"




setting.
                                      83

-------
Electronics
    Connect the oscilloscope to the BNC connector labeled mixer/frequency
on the bottom of the lock-in amplifier (LIA) module.  Trigger the scope on
the AFG output as before.  Set the frequency-mixer toggle switch, next to
the meter, to "frequency."  The output of the tuned amplifier section of
the LIA is now displayed on the oscilloscope.  Adjust the potentiometer
marked "frequency" to obtain a maximum peak-to-peak voltage on the scope.
This adjusts the frequency response of the LIA so that it is the same as
the variable phase retardation plate modulation frequency.  This is the
synchronous detection condition.  Next, switch the frequency-mixer toggle
switch to mixer-  Set the LIA gain to approximately 0.10.  The pattern
displayed on the scope is the output of the mixer or difference amplifier
and should look like Fig. 35b.  If it does not, adjust the phase shift
control located on the LIA front panel until it does.
    A DC signal proportional to the peak amplitude of the mixer signal
is obtained by integrating and amplifying the mixer signal.  This is
accomplished in the output stage of the LIA.  Selection of the integration
time constant is made using the switch next to the LIA meter.  The DC
output voltage is proportional to the density of the mercury in the
absorption tube and is displayed on the front panel meter wlrkh reads in
volts, ±1 volt full scale.  To zero the DC output in the absence of
mercury, adjust the "zero" potentiometer until the meter indicates zero
volts.  It should be noted that the zero setting depends upon the gain,
i.e., each time the gain is changed, the instrument must be re-zeroed.
    The instrument's DC output signal is available on BNC pin J-ll located
on the rear panel of the LIA and may be connected to a dual-pen strip chart
                                      84

-------
 Amplitude
                                                        Time
Figure 35a.  Example of lock-in-amplifier mixer output if phase adjustment
            has  been improperly made.
 Amplitude
                                                                B
                                                        Time
                                                          XBL801-56
Figure 35b.  Lock-in-amplifier mixer output' if phase  is properly adjusted.
                                85

-------
recorder, data logger, or other device.  The second channel of the dual-pen
strip chart recorder should be used to monitor the* PMT voltage by connec-
ting it to the BNC connector on the Log Amplifier module.  Observation of
the total transmitted light intensity along with the atomic absorption
signal  (DC signal) provides a diagnostic tool for identification of smoke
problems.  A 90 to 95% drop in PMT signal due to excessive smoke may
produce an erroneous absorption signal.
Furnace
    Before turning on the furnace power supply adjust the furnace control
module potentiometer to zero.  With the "operate switch" on the front panel
of the furnace power supply in the "standby" position, switch the "prime
power" circuit breaker, also on the front panel, to the "on" position.  To
supply power to the furnace assembly, place the "operate switch" to the
"on" position.  It should be noted that in this state approximately 25 amps
at a few tenths of a volt will be flowing through the furnace circuit.
Adjust the furnace control potentiometer to the desired setting to increase
the current.
    The furnace control potentiometer is set at approximately 2.30 to
obtain a 900°C operation temperature with the seven-inch furnace.  At this
setting,  the reactor current and output current meters on the power supply
front panel  should read approximately 1 amp and 320 amps, respectively.
The furnace temperatures are indicated on the digital trendicator located
below the spectrometer.  Trendicator channels 1, 2 and 3 give the external
furnace temperatures at the midpoint of the combustion chamber, at the
junction of the absorption and combustion chambers and at the (right or
left) quartz window seal, respectively.  The relative temperatures at these
                                      86

-------
three points for a given potentiometer setting will vary with the volu-
metric flow rate and the composition of the gas flowing through the furnace.
    Thermal expansion of the furnace assembly may alter the optical
alignment of the absorption tube.  If a significant change in PMT voltage
occurs, realign the furnace as described above.
    A note of caution:  The furnace power supply is capable of delivering
up to 650 amps at 17.5 volts.  Because of this low voltage, serious injury
due to electrical shock will not occur under normal conditions.  However,
an accidental short across the power supply leads or accidental grounding
of the furnace, which is floating, will cause arcing and, possibly, burns.
In addition, shorting or grounding may cause major damage to the power
supply.
Gas System
    The mercury monitor gas system is shown in Fig. 18.  To activate the
system, first turn the switch on the front panel of the flow controller
module to "on" and adjust the "command" potentiometer to midrange.  Next,
open the sample line ball valve.  Turn on the oxygen supply and adjust the
rotameter to obtain the desired 02 volumetric flow rate.  Use the minimum
flow of 0? required to alleviate smoke due to incomplete combustion of
the sample gas.  The flow of Op to accomplish this may vary between
50 and 200 scc/min.  If the mode of operation calls for the addition of
dilution or calibration gases to the sample stream, open the calibration-
dilution line ball valve and adjust these flows to the desired values.  If
dilution or calibration gases are not required, make sure this ball valve
is closed.
                                      87

-------
    Connect the WTM to the gas line downstream of the flow controller in
order to calibrate the flow controller for the actual combination of
sample, oxygen, dilution, and calibration gases flowing through the
furnace.  Furnace temperature should be stabilized at the desired operating
temperature before this calibration begins.  Adjust the command potentio-
meter to 5.00, which corresponds to approximately 2500 scc/min total flow.
Using the WTM, make several  measurements of the actual flow.  The WTM
should be equilibrated for a few minutes prior to making the measurements
in order to saturate the water with the gas mixture.  Connect the DVM to
BNC connector 5 on the rear panel of the flow controller module and set the
DVM to the 20 volt DC range.   Next, adjust the calibration potentiometer so
that the DVM readout equals the measured WTM flow rate at standard condi-
tions.  Repeat this procedure at several flow settings, bracketing the
desired total flow rate.  It  should not be necessary to readjust the cali-
bration potentiometer at these settings.  The volumetric flow rate of the
sample gas is obtained by subtracting the sum of the oxygen, dilution, and
calibration gas volumetric flow rates, measured with calibrated rotameters
and converted to standard conditions, from the total flow rate measured
with the flow controller.
    Calibration of the flow controller using the WTM should be repeated as
the composition of the sample gas changes.  This is necessary as the flow
measurement depends upon the  specific heat of the gas flowing through the
flow sensor.
                                      88

-------
                                  SECTION 8



                              METHODS COMPARISON



    The use of ZAA spectrometry will extend existing capabilities for the



analysis of mercury in gas streams to include real-time continuous analysis



of complex gaseous mixtures.  This section qualitatively compares the



standard EPA mercury stack gas methods 101 and 102 with the ZAA method


                         15
described in this report.



    The EPA reference methods 101 and 102 for determining gaseous and



particulate mercury emissions are batch sampling methods based on wet



chemical principles.  Mercury is collected in acidic iodine monochloride



(IC1) and reduced to elemental mercury by hydroxylamine sulfate.  The



mercury is aerated from solution and analyzed spectrophotometrically.  This



method is only applicable when the carrier gas stream is principally air



and is therefore not suited for complex gaseous mixtures produced by many



industries.  In addition, the sampling and analytical steps are separated



in time; and the method only yields a time-averaged mercury concentration.



The ZAA method, on the other hand, is capable of direct on-line measurement



of mercury in complex gaseous mixtures.  The gas stream is passed through a



heated sample tube where the mercury is atomized and directly measured by



Zeeman atomic absorption spectroscopy.  The mercury stream is not separated



from the gas stream but is directly analyzed in the gas stream in real



time.  No chemical processing is required and the time lag between sampling



and analysis is on the order of a few seconds rather than hours to days.
                                     89

-------
 Because of the superior background correcting ability of ZAA spectrometry

 (see Section 2), the method may be used to directly measure mercury in

 complex gas streams such as occur in synfuel processes and other industries.

    The major differences between the EPA reference method and the ZAA

 method for measuring mercury in gases are presented in Table 3.  In

 summary, the principal advantages of the ZAA method over the EPA reference
Table 3.  Comparison between EPA reference method and ZAA method.
     EPA Reference Methods 101 and 102
   ZAA Gas Monitor
Applicable only when carrier gas stream
is principally air

Capable only of batch, time-averaged
operation

Limited by chemical interferences from
other components in gas stream

Mercury must be separated from gas stream
for analysis
Time lag between sampling and analysis
on the order of hours to days
Mercury can be concentrated, thus
improving detection limit over ZAA
method
Requires simple and cheap equipment and
supplies; may be operated by low-skill
technicians
Applicable in complex
gaseous mixtures

Capable of continuous
real-time operation

Largely free of chemical
interferences

Mercury may be directly
measured; no separation
required

Time lag between sampling
and analysis on the order
of seconds

Mercury is diluted by
carrier gas; may have a
higher detection limit than
EPA reference method

Requires complex equipment;
must be operated by skilled
technicians
                                     90

-------
methods are that the ZAA method is applicable to complex gaseous mixtures, is
capable of continuous real-time operation, is largely free of chemical
interferences, requires no separation of mercury from the carrier gas, and has
a lag time on the order of seconds between sampling and analysis.  As
presently practiced,the principal disadvantages of this method are that it
requires equipment not readily available to all potential users, it requires
skilled personnel for its operation, and since no concentration step  is
involved, it may have a higher detection  limit  (2 ppb) than the EPA reference
methods 101 and  102.
                                       91

-------
                                    SECTION  9
                             POTENTIAL APPLICATIONS
    The ZAA gas monitor has a wide range of potential applications, including
conventional uses and new uses made possible by the unique characteristics of
this monitor; that is, low interference operation in complex gaseous environ-
ments on a real-time basis.  It is anticipated that the monitor will find
immediate application for the characterization of synfuel and other industrial
emissions, mobile source identification, environmental health monitoring, and
mineral prospecting.
    Mercury emissions from many industries have not been characterized or
quantified due to the lack of adequate analytical methods or due to the
complexity of existing methods.  The emerging synfuel technologies, such as
coal gasification and liquefaction and oil shale retorting, produce large
quantities of highly complex gases.  A number of inorganic and organic consti-
tuents, including many sulfur compounds, in their emissions are known to
interfere with the EPA reference methods 101 and 102 for gaseous and par-
ticulate mercury and with other conventional methods for mercury such as gold
and silver bead absorption tube techniques.  The ZAA method described here can
overcome those interferences and may be used to characterize emissions on a
real-time basis.  This information may then be used to develop control
technology, to determine emission  limits and to design a suitable monitoring
program that is consistent with observed temporal variations.
                                     92

-------
    The instrumentation may also be used for a wide range of routine
monitoring applications where real-time analysis would be advantageous.  These
applications would include environmental health monitoring in workplaces where
mercury and its products are produced or handled, prospecting for mercury and
other ores, and monitoring mobile sources of air pollution.
                                       93

-------
                                 REFERENCES
 1. J. P. Fox, J. J. Duvall, K. K.  Mason, R.  D. Mclaughlin, T. C. Bartke
    and R. E. Poulson,  "Mercury Emissions from a Simulated In-Situ Oil
    Shale Retort,"  Proceedings of llth Oil Shale Symposium, Golden,
    Colorado, April  1978.

 2. J. S. Fruchter,  J.  C. Laul, M.  R. Petersen and P. W.  Ryan, "High
    Precision Trace  Element and Organic Constituent Analysis of Oil Shale
    and Solvent Refined Coal Materials,"  Symposium on Analytical Chemistry
    of Tar Sands and Oil Shale, ACS,  New Orleans.  March 1977.

 3. J. P. Fox, R. D. Mclaughlin, J. F. Thomas and R. E. Poulson, "The
    Partitioning of  As, Cd, Cu, Hg, Pb and Zn During Simulated In-Situ Oil
    Shale Retorting,"  Tenth Oil Shale Symposium Proceedings, Golden,
    Colorado, 1977.

 4. R. E. Poulson, J. W. Smith, N.  B. Young,  W. A. Robb and T. J. Spedding,
    Minor Elements in Oil Shale and Oil-Shale Products. LERC RI 77-1,
    (1977).

 5. K. K. Bertine, and E. D. Goldberg, "Fossil Fuel Combustion and the
    Major Sedimentary Cycle,"  Science, 173,  223 (1971).

 6. D. H. Klein, A.  W.  Andren,  J. A.  Carter et al.» "Pathways of
    Thirty-Seven Trace Elements Through Coal-Fired Power Plant,"  Env. Sci.
    and Tech., £, 973 (1975).

 7. T. Hadeishi and  R.  D. Mclaughlin, "Hyperfine Zeeman Effect Atomic
    Absorption Spectrometer for Mercury," Science, 174, 404  (1971).

 8. T. Hadeishi and R.  D. Mclaughlin, "Isotope Zeeman Atomic Absorption; A
    New Approach to  Chemical Analysis," American laboratory  (August 1975).

 9. T. Hadeishi, "Isotope-Shift Zeeman Effect for Trace-Element Detection:
    An Application of Atomic Physics to Environmental Problems," Appl.
    Phys. Lett., 21, 438 (1972).

10. T. Hadeishi and  R.  Mclaughlin,  Zeeman Atomic Absorption Spectroscope,
    Lawrence Berkeley Laboratory Report LBL-8031 (1978)."~"

11. G. 0. Nelson, "Simplified Method for Generating Known Concentration of
    Mercury Vapor in A->," Rev. Sci. Instr.,  41, 776 (1970).


                                      94

-------
12. Chemical Rubber Company,  Handbook of Chemistry and  Physics,  46th
    Edition, Cleveland (1962)"!

13. R. A. Perkins, "Design of Corrosion Resistant Alloys and Coatings  for
    Coal Conversion Systems," Proceedings of Corrosion,  Erosion  of Coal
    Conversion Systems Materials Conference, Berkeley,  California, 1979,
    NACE Publication (1979).

14. D. R. Stall and H. Prophet, HANAF Thermochemical  Tables, Second
    Edition, NSRDS-NBS37, National Bureau of Standards,  Washington, D. C.
    (1971).

15. Code of Federal Regulations, Title 40 - Protection  of the Environment.
    Chapter 1, "National Emission Standards for Hazardous Air Pollutants,
    Appendix B, Test Methods 101 and 102." (Revised July 1,  1977).
                                     95

-------
                                    TECHNICAL REPORT DATA
                             (Please read Instructions on the reverse before completing)
1. REPORT NO.
   EPA-600/7-80-130
                                                             3. RECIPIENT'S ACCESSION NO.
4, TITLE AND SUBTITLE
     On-Line Zeeman Atomic  Absorption Spectroscopy for
   Mercury Analysis in  Oil Shale Gases
              5. REPORT DATE
                  JUNE  1980
                                                                            ISSUING DATE.
              6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
   D.  C.  Girvin and J. P.  Fox
   Lawrence Berkeley Laboratory
                                                             8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
                                                             10. PROGRAM ELEMENT NO.
   Radian,  Corp.
   P.O.  Box 99^8
   8500  Shoal Creek Blvd.
   Austin,  TX  78766
               1NE623
              11. CONTRACT/GRANT NO.
               68-03-2667
12. SPONSORING AGENCY NAME AND ADDRESS
                                                             13. TYPE OF REPORT AND PERIOD COVERED
  Industrial Environmental Research Laboratory
  Office of Researcn and Development
  U.S.  Environmental Protection Agency
  Cincinnati, OH  45268
              14. SPONSORING AGENCY CODE


                     EPA/600/12
15. SUPPLEMENTARY NOTES
16. ABSTRACT
        This  publication describes the development  and initial testing of instrumenta-
   tion for continuous on-line  analytical measurement of mercury  concentrations  in
   complex gas  streams or in ambient air, in the  presence of smoke,  organic vapors,  and
   oil mist from oil shale processing plants.   The  mercury monitor described is  not
   susceptible  to interferences which plague other  methods and thus  may be used  to
   characterize mercury emissions  on a realtime basis/  This mercury monitor will find
   immediate  application for the characterization of synfuel and  other industrial
   emissions, mobile source identification, and environmental health monitoring.
17.
                                KEY WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS  C. COSATI Field/Group
   Measuring instruments
   Calibrating
   Absorption spectra
   Mercury lamps
   Gases
   Vapor
Zeeman atomic absorption
Mercury monitor
On-line measurement
Oil  shale
18. DISTRIBUTION STATEMENT
                                               19. SECURITY CLASS (ThisReport)
                            21. NO. OF PAGES
                                110
                                               20. SECURITY CLASS (Thispage)
                            22. PRICE
EPA Form 2220-1 (Rev. 4-77)    PREVIOUS EDITION is OBSOLETE
                                                           •fr U.S. GOVERNMENT PRINTING OFFICE: 1980--657-165/OOZO
                                              96

-------