EPA-650/2-73-008
October 1973
Environmental Protection Technology Series
yMm^.
tvSww
\ «^
-------
EPA-650/2-73-008
DEVELOPMENT OF A SUPPLEMENTARY
EMISSION MEASUREMENTS
MONITORING SYSTEM
by
Paul F. Bennewitz
Thunder Scientific Corporation
623 Wyoming S. E.
Albuquerque, New Mexico 87123
Contract No. 68-02-0588
Program Element No. 1A1010
EPA Project Officer: Fredric C. Jaye
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U. S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
October 1973
-------
This report has been reviewed by the Environmental Protection Agency and
approved for publication. Approval does not signify that the contents
necessarily reflect the views and policies of the Agency, nor does
mention of trade names or commercial products constitute endorsement
or recommendation for use.
11
-------
TABLE OF CONTENTS
Abstract i
Objective ii
Scope 1
Introduction 2
Interstitial sensing 3
Technical discussion 4
Fig. 1, Density states three 5
dimensional lattice
Fig. 2, Cutaway ILS Array .... 6
Fig. 3, Frequency Vs wave vector 7
Defect Structure 8
Fig. 4, Replacement lattice ... 9
Ligand Field Theory 11
Fig. 5, Standing wave, defects . 12
Fig. 6, H20 Molecule 13
Fig. 7, Far infrared stretching 14
frequencies
CO, C02, 02, H20 15
Material Selection 16
Material Analysis 17
Group Selection, Sensor firing 18
and construction ....
Test System Configuration .... 19
Fig. 8, Block Diagram EMMS ... 20
Fig. 9, C02 Gas Sensor 21
Fig.10, Steam test 21
Fig. 11, Test Unit #7 23
Fig.12, Test Unit #5 24
Fig.13, Signal conditioning ... 25
02, H20 Sensor 26
Fig.14, Test Unit #6, calibration 27
Fig.15, Test Unit #6 28
Fig.16, Test Unit #5, steam ... 29
Fig.17, Test Unit #5, steam ... 30
Life Tests 31
System Components 32
Fig.18, WSIS-1000, BR-101R ... 33
Fig.19, Gas Sampling System ... 34
Bottom, top view
Fig.20, System, front panel and . 35
Control panel
Fig.21, Atmospheric system and . 36
Test chamber
Conclusion 38
Fig. 22,Flow Chart 39
Fig. 23, System Gas Flow 40
References 41
-------
ABSTRACT
A prototype Supplementary Emission Measurements System
was designed and developed to provide measurement of
C02, CO, H20, and C^in stack gases of stationary
emission sources.
Sensors were developed or obtained from other sources
to meet the criteria set forth in the requirement.
The system was designed to facilitate direct readout
of all channels via individual analog metering. In
addition, outputs of 0 to 1 VDC were provided for on-
line computer feed.
The system was equipped with a stack sampling probe,
filters and pump to provide full and complete stack
sampling and measurement capability.
-------
OBJECTIVES AND/OR BACKGROUND
The project for development of a Supplementary Emission
Measurements Monitoring System was organized with the
following design goals and objectives:
1) Design a prototype subsidiary emissions measurements
system to measure C02, CO, 1^0 and 02 in stack
emission gases.
Ranges Requested:
C02 0 - 20% - 0.5% minimum sensitivity
CO 0-5% - 0.1% "
H20 0 - 20% - 0.5% "
02 0 - 7%, 0 - 22%
2) To procure or develop sensor with the specified
typical minimum ranges and sensitivities.
3) To supply flow velocity of 20 to 120 feet per second.
4) To supply temperature measurement in the range of
50° to 700°F.
5) To equip the system as required with sampling probes,
filters, pump, etc.
6) To supply analog meter readouts and simultaneous
analog 0-1 volt outputs for all parameters.
7) To equip the system with suitable zero and calibra-
tion circuitry as required.
ii
-------
Report of Development of a
Supplementary Emission Measurements
Monitoring System
SCOPE
This report describes sensors and systems developed to
provide precise measurements and readout of subsidiary
emissions consisting of C02, CO, H20 and 02 as required
by the Environmental Protection Agency.
Unique methods were implemented which, due
to past developmental experience at Thunder Scientific
Corporation, were considered to be an advance of the
state-of-theart and highly applicable to the requirement,
Sensing Elements (Solid-State)
This report describes the use of various types of semi-
conductor materials for discriminatory sensing of various
pollutant gases; a superior and unique semiconductor
method was used, based upon prior experience gained in
the recent development by Thunder Scientific of the ad-
vanced state-of-the-art Brady Array bulk-effect humidity
sensor. It has been determined that the Brady Array is
most effective, accurate and reliable because of its
interstitial lattice method of sensing.
Known monocrystalline semiconductors in chip form, when
connected to appropriate circuitry via applied electrodes,
exhibit various characteristics relative to sensitivity
to the environment and various gases, particularly to H20
in various concentrations; however, not without degrada-
tion and irreversible effects to be discussed later. The
consensus of opinion in both the past and present has been
to assume that when designing a new sensor, for example a
humidity sensor, or other types, one should rely upon
various effects of different types of materials to achieve
the end result, with little thought given to designing
specifically to the requirement. Previous sensors, such
as those designed for measurement of moisture or relative
humidity, relied upon surface phenomenon to achieve a
measurement. Desiccants or hygroscopic materials have
been heavily relied upon as attractants for moisture for
-------
measurement. If observed under high magnification, all
the above configurations may be seen to be agglomerating
water or forming islands of moisture during the process
of measurement. This means loss of response time and
other characteristics leading to undesirable results.
Also, because the surfaces of these devices are limited
in area, dynamic range is very narrow.
In the case of the Brady Array, humidity sensor, a semi-
conductor bulk-effect device was constructed which could
be proven mathematically to be electrically neutral. A
method was discovered in the process of development which
allows the construction of microminiature lattice arrays
made up of many crystal semiconductor junctions, separated
by interstitial spaces. These junctions, estimated to
number a million or more, present a very large geometric
area, if all crystal faces were spread out side by side.
Because the water molecule is extremely small, and the
structure is electrically neutral, the molecules may
drift randomly and freely into and out of the structure,
causing interation within the semiconductor
structure which stresses bonds and changes the volumetric
resistance when percent relative humidity or molecular
density is changed. Because of the semiconductor nature
and the array effect of the structure, the sensor exhibits
a near one-to-one ratio in output voltage vs. input exci-
tation as the surrounding moisture or molecular density
is changed.
It is this method of approach, the interstitial lattice
sensing method (ILS) , which was used extensively through-
out the requirement.
Introduction
In the following technical discussion, reference will be
made periodically to the interstitial lattice method of
sensing as applied to a semiconductor-type bulk-effect
sensor. This method and the processes involved are pro-
prietary, with patents obtained or in process, having been
solely developed by Thunder Scientific during in-house
funded R&D programs recently completed, to conceive,
develop and construct an advanced state-of-the-art humidity^
which is a solid-state bulk-effect semiconductor
device. This sensor, designated the Brady Array, is pre-
sently being marketed along with signal conditioning
modules also developed at Thunder Scientific, with excel-
lent results .
-------
Interstitial Sensing
Interstitial lattice sensing (ILS) is a method wherein
an extremely complex crystal array is processed and con-
structed within a diffusion type furnace. Contrary to
normally known configurations of semiconductors where
the semiconductor is a monocrystalline structure or die
consisting of a single slice cut and diced from a single
zone-grown crystal, the interstitial lattice array,
measuring only 10 mils in diameter and 100 mils in length,
consists of thousands of semiconductor junctions preformed
and processed so as to provide interstitial bulk access
to all faces of the crystal lattices and junctions within
the structure. Thus, single molecules of various gases
and constituents may easily drift into and out of the
array, creating Fermi-related energy exchanges, causing
the bonds within the lattices to be stessed. Because they
are vibrating at a descrete and synonymous stretching fre-
quency with the impressed molecular field, energy is
released within the structure to the free electrons therein,
which causes a shift of the Fermi level, thus causing a
change in the volumetric resistance of the structure (see
Figure 2) .
Appropriate signal conditioning, generating an excitation
voltage in a simple bridge arrangement and a demodulator for
conversion of the signal to direct current, thus supplies
an output signal across an output termination proportional
to the phenomena, for analysis by direct reading end in-
struments or computer manipulation.
Greater Sensing Area vs. Size
The crystal lattice array, sensing by interstitial means,
provides a sensor area, when viewed at a molecular level,
hundreds of magnitudes greater than that provided by a
monocrystalline semiconductor. Rather than presenting one
or a few semiconductor junctions to the medium, which sub-
sequently will greatly limit the dynamic range of readout,
the ILS presents a large geometric area if all crystal
faces and junctions were laid out side by side.
Therefore, dynamic ranges of many orders of magnitude
become available. This is a major asset when dealing
with units of measurement such as those related to
pollution and contaminant levels, i.e., PPHM, PPM, %,
mass , etc.
3
-------
Technical Discussion
In a monocrystalllne semiconductor, great care is taken
to prevent the influx of, or contamination by, impurities,
both in the processing environment and following packaging.
For example, the use of Si02 in the field electrode of MOS
transistors has caused difficulties, since this glass
easily absorbs foreign ions that can move under field in-
fluence through the glass. Thus sodium ions can easily
move within the glass if outside fields are applied, and
therefore, a change in conductivity and dielectric constant
causes a change in impedance and capacity values. Other
impurity atoms such as CO, HCL and HF can possibly move
through the glass, causing permanent changes in impedance
and capacity.
At first glance, one would tend to assume that because of
these phenomena, monocrystalline semiconductors could be
used directly to detect these impurities; however, this is
not entirely ture. The prime reason in this instance,
whether the monocrystalline semiconductor is Si02 or Ge or
combinations or with various dopants, is that the principal
of operation must be based upon a theory of absorption or
adsorption. such as in the previously mentioned surface-
type humidity sensors, thus narrowing the dynamic range
and creating major problems of repeatability and reproduci-
bility.
Of key and significant importance and one of the most major
disadvantages which should not be overlooked in evaluation
of devices suggesting monocrystalline structures, is that
these structures, solely by the fact that they are mono-
crystalline, are highly susceptible to doping, even in
limited amounts. All the various gases concerning in this
report and most others indicated in the references are
"dopants". These dopants cause irreversible changes of
impedance, capacitance and resistance within monocrystaline
structures. These changes are a major source of inaccurate,
unreliable devices for the purpose intended.
Interstitial lattice sensing (ILS) and the crystal lattice
array represent an advancement in the "state-of-the-art",
based upon the past two years of research carried on at
Thunder Scientific Corporation. This array is a semiconduct-
ing device that does lend itself directly to sensing of the
various parameters required. The ILS device can be thought
of, in one sense, as a cross between an amorphous semicon-
ductor and a monocrystalline semiconductor, i.e., the
- 4 -
-------
N(E)
i
Ul
Valonce bond
Conduction band
FIGURE ONE
DENSITY STATES THREE DIMENSIONAL
LATTICE
-------
GAS CONSTITUENT
TUNNEL ELECTRODE
DISLOCATION
JUNCTIONS
TUNNEL ELECTRODE
DISLOCATION SEMICONDUCTING JUNCTIONS
FIGURE TWO
CUTAWAY ILS ARRAY
-------
n\
FIGURE THREE
FREQUENCY Vs WAVE VECTOR
-------
electronic properties of defects of the dislocation type
play a somewhat important part, yet the structure is poly-
crystalline while at the same time being deliberately
highly disordered.
This disordering accounts for the large interstitial areas,
most important when sensing gas constituents such as 1^0,
CO, C02 , etc. In effect, the gas tends to behave like and
may be treated as electron gas. Therefore, in analysis,
one may derive from measurements of temperature dependance,
resistance, thermo-electric power and reflectivity, that
this type of structure if far less sensitive to doping than
either amorphous or monocrystalline structures.
In any crystal, monocrystal, or the individual disordered
mode crystals of the ILS structure, two different vibra-
tional modes exist, which can be related by:
- -_-_ __ 4 sin2Kd
mi
These two vibrational modes are of great importance in real
lattices giving the value for CO at the limits (see figure
3). Since the lower function corresponds to in-phase move-
ment of nearest neighbors (off and even numbered atoms), it
is called the "acoustical" branch, because the lower fre-
quencies generated in this kind of excitation correspond
to the acoustical range for most crystals. The upper branch,
shown in figure 3, is called the optical branch, because
these frequencies belong in the electromagnetic infrared
frequency range. It is these phenomena and characteristics,
among others to be explained, which account for the prime
theory of the ILS array. The frequency vs. wave-number
functions in figure 3 can have different bands if more than
two different particle masses or more than two force con-
stants are involved. The main difference between their
frequency range remains the cophase or contraphase movement
of nearest neighbors.
Defect Structure
Contrary to the case of monocrystalline semiconductors where
it is desired to hold point defects to a bare minimum, in
the ILS structure, the opposite is the case. Defects within
the structure are of absolute necessity in conjunction with
the disordered array for operation of the system. For
example, in loo.king at the periodic structure in the simple
- 8 -
-------
VO
I
2C7
FIGURE FOUR
REPLACEMENT OF Cl~- ION BY H~- ION IN
K+C1~ -LATTICE
-------
model shown in figure 4, one may begin to understand the
effect of primitive lattice disturbances caused by vacan-
cies or interstitials as applied later to the ILS array.
Consider a displacement of one atom by either a heavier
or lighter one, or, for purposes of ease of description, a
displacement of an ion in an ionic lattice, such as KC1 by
hydrogen. In this case, the light ion (R-) within the
positive K+ cloud has its own frequency of vibration
(stretching frequency), with an optically active electric
dipole moment. Because of the very localized nature of
this defect, the additional vibrational modes decrease
rapidly with distance from the defect (actually exponenti-
ally). However, if these defects are not localized but
comprise many atomic sites as in ILS, the lattice motions
for such a lattice substructure lead to standing wave
modes, literally generating a strain field as electrons
are moved into shared positions, depending upon the density
of the molecular constituent and field applied. The
lattice motions for such a lattice substructure lead to
multiple standing wave modes because of the bonds on
either side of the sub-lattice or multiple disturbance.
By representing the substructure by a single dimensional
line leading to a standing wave of: An =A(0)e~iwt sin nkd
(figure 5), because of the limitation L of the size of the
sub-lattice, K is restricted to:
K
The solution for (A) vanishes for n-0 and n-N (fixed points),
and for points in between, yields a number K -Vr/Nd independ-
ent modes.
This analogy between defect or sub-lattice excitation and
bound lattice modes may help explain a number of the effects
to be discussed connected with infrared vibration spectra
of the ILS and the disordered system array.
Dislocations, in conjunction with junction breakdown caused
by the stressed bonds, introduce interband energy levels,
therefore allowing the electrons to move easily through a
lowered field, thus allowing interband transitions to the
conduction band. It is suspected that many dislocation
space charge pipes or elements of similar nature, perhaps
assisted by the electronic wave functions, account for the
fast response and lack of hysteresis in the ILS array.
- 10 -
-------
In further analysis, in a structure such as the ILS, a
multitude of bonds, known as dangling bonds, are thought
to exist, which are generally commonplace around and in
the space charge pipe sites. These dangling bonds are in
an excellent position throughout the structure to pick up
shared electrons, thus creating a tremendous amount of
strain or stress upon the inner bonds; therefore, the
structure will also tend to purge itself of molecules of
the gas constituent of interest when the molecular density
is reduced or removed entirely from the vicinity of the
sensor. This purging, or flyback as it may be called,
occurs within 150 milliseconds in some cases.
Selection of Semiconductor Materials
Past experience at Thunder Scientific has indicated that
transition metal oxide complexes appear to be the best for
use in the construction of the ILS arrays. Transition
elements, have partially filled d or f shells. Before
describing the method to be employed to construct the ILS
array, or configurations and transition complexes analyzed
and employed, further discussion is required to better
understand the semiconducting effects of transition metal
oxides and their particular applications to this effort.
Characteristics of the various transition metal oxides are
quite interesting, in that it becomes rapidly apparent that
various combinations display extremely varied electronic,
ESR and vibrational spectra and that by very minute variance
of the mix and process times, temperature and diffusion
gases, the stretching frequencies may be manipulated rather
precisely. Some of the various coordinate complexes can
be made to take up an additional odd or even ligand very
readily. Valence states, relative ionic radii, associated
bond lengths, and stretching frequencies are extremely
important in determining the final configuration best fitted
for monitoring each gas constituent.
Ligand Field Theory
To gain a better understanding of the effects taking place
at a molecular level, it is well to define and explore
ligand field theory as it applies to the problem at hand.
Ligand field theory is defined, in this sense, as the theory
of (1) the origin and (2) the consequences, of the splitting
of inner orbitals of ions by their chemical (gas) environ-
ments. The inner orbitals of interest in this case are the
partially filled d or f orbitals, primarily the d. In this
discussion, these splittings must be related to the forces
- 11 -
-------
NORMAL LATTICE
n= 0
\3
1
sin nkd
FIGURE FIVE
ONE DIMENSIONAL STANDING WAVE
OF n DEFECTS IN LATTICE
-------
8
FIGURE SIX
H20 MOLECULE
-------
MOO 21
F-
CO
1640 070
h~f
ISM I4»T 1369 I30O ll« lift
NO 11NO, 1Isort Isotf
IMO
FIGURE SEVEN
FAR INFRARED STRETCHING
FREQUENCIES
-------
involved. The ILS structure is not purely metal nor is
it purely oxide, but rather a structure achieved by special
processes, consisting of junctions containing both metal
ions and oxide chains that contribute to the over-all spin
and stress effects observed. (See figure 2).
Carbon Monoxide
Carbon monoxide, CO, is a pi acceptor ligand. Various pi
carbonyls may also be associated. CO is a generally inert
molecule and particularly adaptable for detection by the
ILS array. Dipole moment studies indicate that the moment
of an M-C bond is very low, about 0.5 D, which suggests
that it is close to being electrically neutral as per
Paulings law. CO has a bond length of approximately 1.128A,
The CO molecule has a stretching frequency of 2145 cm~ .
If the carbonyl molecules are taken into consideration, a
band would then exist of approximately 2125 - 1900 cm~l in
the far-infrared spectrum, providing both medium and strong
spectra for detection resonance. Many combinations of
carbonyls, and other atmospheric constituents, can be ob-
tained in reactions that take place at normal atmospheric
temperatures and pressures caused by the photochemical
process.
Carbon Dioxide
Carbon Dioxide, C02 , exhibits a stretching frequency from
approximately 1570 to 1640 cm-1. It shows some tendency
to overlap NO; therefore, some discrete manipulation of
stretching frequency was necessary to avoid confusion with
NO. It was suspected that some latitude was available for
this, due to the fact that the NO band is not continuous
across the band of 1580 to 1900.
Oxygen (02)
Oxygen, 02, has an extremely broad band, exhibiting effects
across the entire range.
Water(H20)
The water molecule has a stretching frequency of 3200 to
3600 cm-1 in the far infrared region, which are rated as
strong resonances. Other minor resonances exist even into
- 15 _
-------
the low infrared; however, these are weak to very weak
and may be ignored. To better understand the complete
theory of the ILS array, the water molecule presents an
excellent opportunity in that it is relatively simple and
readily accessible. The oxygen atom is bonded to two
hydrogens in an arrangement expressed pictorially in
figure 6. The separation of the hydrogens approximate, by
present conceptual theory, an angle of 105°. Each hydrogen
precesses at a given rate (A) and in conjunction with this
momental precession, it is believed that a complex preces-
sion exists about 0, as shown by B following through C and
so on. These perturbations may be attributable to the
anti-symmetric and symmetric 0-H stretching mode at 1600
to 1650 cm""* In addition, quite a good deal of fine
structure is apparent. Therefore, as the molecular density
or moisture content of the atmosphere in the vicinity of
the Brady Array humidity sensor changes, bonds are stressed
as previously explained, thus causing a shift in volumetric
resistance of the Array and subsequent change in readout
voltage. Further direct explanation of the Brady Array
bulk-effect semiconductor humidity sensor is contained in
a recently published paper (see Appendix).
Material Selection
As previously indicated, most, if not all, first transition
series metal oxides will, when processed and doped correct-
ly, exhibit semiconductor phenomena. With proper process-
ing, the resistance of resultant elements may be controlled
over a broad range of as much as 1014 ohms/cm or more when
using the ILS technique. Past processing indicates that
the d-block elements with partially filled shells are d
shells, 3d, 4d, or 5. These d orbitals project well out to
the periphery of the atoms and ions so that the electrons
occupying them are strongly influenced by the surroundings
(the gas constituent) and in turn are able to influence the
Fermi level and internal environment of the sensor substan-
tially. The final selection of the precise transition
elements, reactants, doping materials, processing tempera-
tures and diffusion gases were determined during the program
As in the case of the previously developed bulk-effect Brady
Array, these parameters were developed by a combination, in
sequence, of mathematical analysis and emperical. Cut and
try, X-ray and spectrum analysis of lattice and crystal
structure had previouly been performed upon the Brady Array
in previous in-house research thus giving some idea of
typical structure to be expected. The "cut and try" deals
primarily with process times, temperature cycling rates and
diffusion gas content and flow rates, following initial
determinations of sensitivity to the gas constituent of
ineterest.
- 16 -
-------
Material Analysis,
Studies, and Selection
Initially, studies were instigated to determine proper
bond lengths and radii, ionization potentials, melting
temperatures, transition points, molecular weight, etc.
of various oxides of metals to determine possible compa-
tability and/or sensitivity to the various gases desired
to be monitored. Far infrared stretching frequencies
were also taken into consideration and listed. Sources
for these data are listed in the references contained in
this report.
Based upon the above referenced studies, the following
final test groups were selected for initial testing and
analysis .
EPA Code #1 EPA Code #2
50% Fe203 50% F
30% ZnO 40% ZnO
20% Sn02 10% Sn02
EPA Code #3 EPA Code M
50% Fe203 50% Ti02
20% ZnO 50%' V205
20% Co203
5% V205
5% Sn02
EPA Code #5 EPA Code #6
50% Co203 50% Zn
50% V205 50% Sn
EPA Code #16
Ti02 +
+ Neodymium
Though a total of sixteen (16) test groups were initially
mixed, it was determined that those listed above were
most applicable based upon the previous studies.
- 17 -
-------
Group Selection
To determine the applicability of each test group for use
as a sensor, it was necessary to observe the structure of
each when exposed to various firing schedules within the
furnace. (an Astro Industries Series 1000) This was done
by placing each raw material mix within ceramic boats and
observing the effects achieved when the material was fired
at 1500°C to 1550°C. If a disordered polycrystalline
structure or tendency was noted, the material was marked
for prototype sensor construction. Of the 16 test groups,
those listed previously exhibited this tendency.
Sensor Construction
Each sensor was constructed by hand beneath a microscope upon
glass cover slips. The material was mixed with a 95% - 6%
mixture of beta terpinal and celulose acetate to form a binder.
The thus mixed material was then placed on a glass cover slip
and formed, using a scalpel until a rectangular form was
achieved measuring approximately 180 mils in length and 25
mils in width, with a depth of approximately six mils. Four
mil platinum iridium electrodes were then embedded in the
material equally spaced and the overall sensor shaped and
flattened with the scalpel until of such a dried consistency .
to be picked up off the cover glass with the scalpel.
Sensor Firing
The completed sensors were then placed upon supporting
mandrels within platinum boats and fired for thirty minutes
at 1550°C. The boats were then removed and the completed
sensor thermally lead bonded to a TO-5 type eight pin header.
Generally, it must be pointed out, at this stage in the devel-
opment that a major portion of the process is one of "cut and
try" including varying of temperature and diffussion gases.
As an example, the final configuration chosen for C02 was EPA
Code #16. Here several gases were tried looking primarily
for the largest crystal formation. All gas flows were flowed
at the rate of 100 cc/min. The final gas chosen was a mix-
ture of air and CO in a 75% air, 25% C02 mixture at atmospheric
pressure.
In the CO sensor, it was determined that it functioned selec-
tively to CO and discriminated against C02 when run at a
current of 300 m.a. driven by a constant current module.
The material mix chosen due to the need for this higher
current was EPA Code #1. This material when employed within
the sensor showed no deterioration after long term running
at this higher current.
- 18 -
-------
Sensors were constructed of the various selected groups
and designated as follows:
Test Unit #1
Test Unit #2
Test Unit #3
Test Unit #4
Test Unit #5
Test Unit #6
Test Unit #7
EPA Group //I
1st unit of EPA
2nd unit of EPA
1st unit of EPA
1st unit of EPA
Group #4
Group #4
Group #5
Group #16
2nd unit of EPA Group #1(
2nd unit of EPA Group #5
Tunneling electrodes were used to allow
bonding of the sensor into a small TO-5 can
(See photo, Figure 9)
thermal lead
type enclosure
Electrical Conf iguration
An integrated circuit amplifier was constructed as a pro-
totype to achieve zero offset and a span of 1 volt DC
output. This amplifier was inserted into the circuit as
shown in Figure 8. All sensors were initially tested with
this basic breadboard.
Initial Test Results
Of all sensors initially constructed, Test Unit
Unit #6 and Test Unit #7 indicated satisfactory
ity to C02 and were then further tested.
#5, Test
sensitiv-
Test System Configuration
Initially an all glass test system was constructed as
shown in the photograph to allow testing of the sensors
in both gas and humidity environments. In addition
functioning as a mixing manifold for all gases in a
manner, the system also provided means of injection
known amounts of moisture into the gases if so
to
precise
of
desired,
allowing the capability of achieving better than
accuracy in mixing ratio.
+ 1%
Precise accuracies of mixing ratio adjustment and flow
were controlled by Matheson flowmeters with individual
calibration curves for the various gas constituents.
Mathematical computation allowed added precision when
dealing with the various viscosities and specific gravities
of the gases. Precision differential monometers allowed
- 19 -
-------
CCR
o
I
V
MB GAS ARRAY
NOTE: CCR= CONSTANT CURRENT REGULATOR
FIGURE EIGHT
BLOCK DIAGRAM EMISSIONS MONITOR
SIGNAL CONDITIONING CHANNEL
-------
COMPLETED SOLID STATE
C02 GAS SENSOR
FIGURE 9
SENSOR IN ONE HOUR
STEAM EXPOSURE TEST
FIGURE 10
- 21 -
-------
precise balance between the mixed constituents.
In initial testing for gas sensitivity and ranges, the
test system was set up to allow use as a mixing manifold
system only.
A standard Thunder Scientific Brady Array Humidity Sensor
was first installed within the system to detect any
moisture that might be leaking into the system. First
tests indicated bad leakage, though all gases had been
previously routed through two Ca S04 dessicators columns.
Final source of leakage was traced to the tygon tubing
used in the system.
Following this discovery, all tubing was replaced with a
hard dense walled nylon tubing eliminating the problem.
The test sensor, Test Unit #7 was placed within the system
and tests begun.
Initially dry nitrogen was used as a zero gas to establish
a base. Gas flows were held to 100 cc/min. Temperature
of the test chamber was continuously monitored and taken
at each reading.
Test readout equipment consisted of a Fluke 853A Differ-
ential Multimeter, an 8300 DVM and an Esterline Angus
Recorder.
The sensor was first checked for selectivity, the results
obtained are shown in Figure 11. This recording was obtain-
ed by mixing in two Matheson type flowmeters. C02 was
first vented into the system at 100% level and the amplifier
adjusted to obtain a 1 volt DC output. The first recording
shown in Figure 11 displays the response time of the Test
Unit #7 with the chart running at 1.5" per minute. The
second portion of the recording shown in Figure 1, displays
the response to 100% C02 with the recorder running at 12"
per hour. CO and 02 were then vented into the system at
100% concentration, each time clearing the system with
nitrogen. Selectivity of the sensor to C02 may be noted as
a result.
Test Unit #5 was then installed within the system and the
recorder run at 6" per minute. (See Figure 12) Here, due
to the additional doping and material selection, it will
be noted that the response time is considerably enhanced.
Test unit #5 was then operated in ambient air in live steam
for one hour. (See Figure 16) (See Figure 9)
- 22 -
-------
1 .1
t
I
..<
1:
..
'»*.
i
i
.
i
.
N
\
'..'".
,
I-1 C
i-t
(I
-------
-------
The SC-1020M signal
conditioning module
shown with the C-3A
environmentalized
cable and Brady Array
The SC-1020M signal conditioning module with Brady Array Model
B'R-101 humidity sensor
FIGURE 13
- 25 -
-------
Following one hour in live steam, the recording shown in
Figure 17 was run to determine any shift in calibration
due to this exposure. It was noted that a slight shift
of 44 m.v. occured. Considering the stringent environment,
this small amount of deviation was not considered serious.
A calibration run was made on Test Unit #6 and the record-
ing shown in Figure 15 obtained. It will be noted that
the sensor exhibits a near linear response to CC>2 These
data are plotted and the curve obtained shown in Figure 14.
Of the test units run and tested, Test Unit #6 was selected
for use in the Emission Monitoring System.
Life Tests on The C02 Sensor
The final sensor was installed at ambient in an operational
condition for life testing. Prior to installation in the
completed system, a total of 8000 hours was accumulated
with no degradation in characteristics or output.
The CO Sensor
During tests of the various test groups, it was noted that
Test Unit #1 of EPA Group 1 exhibited characteristic
sensitivity to CO, i.e., response, etc. However, sensitiv-
ity to C02 was also noted.
The CCR, Constant Current Regulator, nominally supplying
20 m.a. to the CO sensor was changed to a variable CCR type
generator and the output current across the sensor varied
in steps as the output was evaluated.
It was found that apparent "tuning" of the sensor took
place and as the internal current thru the sensor was in-
creased, the sensitivity to C02 was eliminated, while the
sensitivity to CO content was enhanced.
Considerable difficulty was experienced in clearing the CO
sensor with nitrogen, once exposed to CO. It was found that
the sensor would clear when a standard air mix was flowed
across it. Because of this, a special "CO PURGE" valve was
incorporated in the system. Test Unit //I was selected to be
employed in the final Emissions System for sensing of CO.
-------
EPA TEST UNIT #6
GROUP 16
CALIBRATION 9/13/72
I, .'i !!ij,.! :!.'': :.:;;::
.;:t: i . :./,.. .; : I
FIGURE 14
i;-:.;-i:- i :
- 27 -
-------
!
.
Q-A&
.
h
l_nC
>-(
-------
31
*»
f}
->!
-
O
... ,
t
f -~
i
1::
!
W \
V
'
'. '
TEST UNIT #5
9/17/72
IN LIVE STE;
i
j
*
iM
;
- ]
^
-
I
t
,
L Hr .
-
U
V-
'.V
, |:
i
-
,\
flyjjl-
-
v
.
i
.
-H-
l7EFT-FI-r
.
A/^
L
'
... ,
:
FwlT1
\!/ S
- -V* -t-
'
!
.
. (
-i- -4-
..-_:_ 4 -
(
T" "r-
1
A/A : - / v\ '
\!.> yv4A
l.i._;^
i
I - i '
. . . . ,
I i
i
l
i_i oo
\*
: _ .
:
*-*»*,-*
.
"(
|
.
:
;
- -l
»v\
CO
.
!
b
i t
--i
t
i - .
! )
: j\ ift/.
A] ' \ fi-
^- ^* J
.
. - . - - r. . . . . , -,
i
i
i
j
i
L_J
-
II
-------
=fvp
_j_.
-------
The 02 Sensor
Considerable difficulty in the course of the program was
experienced in obtaining a solid state oxygen sensor. It
was finally determined that, though it appeared possible,
with time, to obtain an oxygen sensor employing similar
techniques, that a standard commercially available in-
strument would be incorporated for this measurement. A
Beckman Model 715 Oxygen Analyzer was chosen in the course
of evaluation.
The H2Q Sensor
The Brady Array Model BR-101 in conjunction with a stand-
ard signal conditioning module, Model SC-1020M-2 manufac-
tured by Thunder Scientific Corporation was employed for
measurement of water vapor present within the gas stream.
This system has been employed by many in the field for
measurement of humidity with a high degree of success over
the past three years.
The Brady Array has a specified accuracy of + 4% RH, typical
accuracy being better than +2% RH. The range covered is
0% RH to 100% RH.
Due to its specific design, the sensor can be operated at
elevated temperatures, i.e., in a heated sampling enclosure,
ranging up to 160°F. (See Figure 19)
System Flow Transducers
Because it is necessary to maintain a reasonably consistent
flow across the gas sampling sensors, a sensitive flow
sensor was incorporated. This sensor, a Thunder Scientific
Corporation Model WSIS-1000 allows measurement and precise
adjustment of the gas flow within the system to 100 to 200
cc/min. (See photo in Figure 18)
Stack Temperature
Stack temperature is monitored and displayed on an analog
meter on the front panel. A thermocouple and reference is
contained within the sampling probe for monitor of temper-
atures within the gas intake. The probe is equipped with
a back flushable micron filter with the back flush control
located on the front panel of the switching module.
- 31 -
-------
Stack Flow Rate
Initially the Model WSIS-1000 was investigated for use
in determining stack velocity, however, it was found that
the extreme temperatures that could be encountered within
the stack prevented use of this device. Though Stack
Flow Rate is displayed on the front panel, no satisfactory
means was discovered in the course of the contract to
monitor this parameter accurately.
The Em i s s i o n Measurements Moni to ring Sys tern
The Model EMMS-100 Secondary Emissions Monitoring System
was developed in the course of the contract to provide a
compact, but complete, self contained monitoring system
for monitor and recording of levels of CC>2, CO 02, H20 and
temperature in stack gases of power plants or in other
emissions generating facilities.
The Switching Panel controls all gas switching and routing
within the Gas Sampling System and provides signal condi-
tioning and control for all transducers in the system.
System Components
The system is constructed with 1/4" stainless steel tubing,
All dryers, traps, sensor mounting blocks, etc. are con-
structed of stainless steel. All valves are electrically
actuated directly from the front switch panel. Figure 19
A heated enclosure is contained in the top of the gas sam-
pling system to maintain a preset temperature upon all
critical components. Though the solid state sensors ex-
hibit no sensitivity to water vapor, it was desired to
maintain the system at a low level of relative humidity.
This is accomplished by a proportional controller designed
to allow adjustment of system heat from ambient to a max-
imum of 160°F.
Switching
All switching is accomplished from the front panel of the
switching module as shown in the photo. Figure 20
Switches are of the "push on" "push off" variety. All
switches are labeled and back lighted.
- 32 -
-------
WSIS-1QOO FLOW VELOCITY SFNSOR
LOW PROFILE BRADY ARRAY SENSOR
FIGURE 18
- 33 -
-------
GAS SAMPLING SYSTEM
BOTTOM VIEW
PUMP
SYSTEM VELOCITY
SENSOR WSIS-1000
SYSTEM FLOW VALVE
02 SENSING HEAD
FILTER FLUSH
COMPRESSED AIR
FILTER
H?0 SIGNAL
CONDITIONING
TOP VIEW
( HEAT SHROUD REMOVED )
SYSTEM CONTROL
VALVES
C02, CO, & H20
SAMPLING CHAMBERS
GLASS FIBER
INSULATION
TRF. 19
- 34 -
-------
COMPLETE SYSTEM
SWITCHING AND
CONTROL PANEL
GAS SAMPLING
SYSTEM
SYSTEM SWITCHING
AND CONTROL PANEL
ANALOG READOUTS
PROPORTIONAL
/HEAT CONTROLS
RECESSED CONTROL
PANEL
ZERO OFFSET AND
SPAN CONTROLS
SYSTEM CONTROL
CONTROL SWITCHES
SINGLE CHANNEL
CAL GAS CONTROL
SWITCHES
FIGURE 20
- 35 -
-------
ATMOSPHERIC GENERATOR SYSTEM
TEST CHAMBER
FIGURE 21
- 36 -
-------
Power
Power is controlled by a back lighted switch on the
front panel labeled "POWER"
Intake Valve
The "INTAKE VALVE" switch controls the intake valve at
the input of the gas system.
Zero Gas
The "ZERO GAS" input at the rear of the system is con-
trolled by a front panel switch labeled "ZERO GAS".
Computer oti Line
The "COMPUTER ON LINE" switch, located on the front of
the switching panel switches the 0-1 VDC outputs of
all electronic modules to a connector available at the
rear of the chassis.
CO Purge
A switch is provided to allow purging of the CO sensor
with .air following each reading.
Filter Flush
A switch is also provided for back flush of the sampling
probe and filter with compressed air.
Other control functions include system heat, pump, Cal
Gas .
Sub Panel Control Functions
Control functions within the sub panel include zero off-
set and span adjustment for all electronic modules for
calibration adjustment. In addition, the proportional
heater control, 02 analyzer controls and individual channel
Cal Gas switches are included within this panel.
- 37 -
-------
Conclus ion
It has been determined in the course of the program that
viable and useful solid state semiconductor sensors can
be constructed and tuned to accurately and selectively
measure CC>2 , CO and 1^0 percentage ranges within a closed
gas sampling system.
That a reasonably small compact gas sampling and control
system can be constructed to allow efficient use of the
concept.
- 38 -
-------
EPA SYSTEM GAS FLOW
S02 , NOX
ANALYZER
FILTER
(probe)
u>
VD
WSIS
NEEDLE
VALVE
PUMP
INTAKE
VALVE
AIR>
FILTER
COMPRESSED
AIR VALVE
GAC
co2
GAC
CO
GAC
2
GAC
PURGE
VALVE
N.
I
ZERO GAS
VALVE
WATER
TRAP
CAL GAS
VALVE
-------
H20 SENSOR
CO., SENSOR
CO SENSOR
SC-1020M-2
DRIVER
600 OHM
LINE &
METER
CCR & AMP
DRIVER
600 OHM
LINE &
METER
CCR & AMP
DRIVER
600 OHM
LINE &
METER
°2
SENSOR
SYSTEM FLOW
RATE SENSOR
SYSTEM TEMP
SENSOR
STACK TEMP
SENSOR
ELEC
METER
DRIVER
1
600 OHM
LINE
CCR & AMP
DRIVER
600 OHM
LINE &
METER
PROP
CONT
HEATERS
DRIVER
ELEC
DRIVER
i
600 OHM:
LINE &
METER
600 OHM
LINE &
METER
STACK FLOW
RATE SENSOR
CCR & AMP
FIGURE TWENTY-TWO
ELECTRONICS FLOW CHART
-------
ft o ;" o r o n c o .T
i. Xot:c, N'. ?. ; "'.dec crone in Disordered Structures,"
Advances in Phyuics (British), vol. 16, pp 49-144, 1967
2. Gubanov, A.: "Quantum Theory of Amorphous Semiconductors,"
Consultants Bureau, New York 1965
3. Ovshiaalcy , S. Iv.: "Reversible Electrical Switching
Phenomena in Disorder Structures," Pays. ReVt Letters,
vol 21 #20, ?? 1450-1453, II Nov 1968
A. Xorin I-'. J.; "aalidcs, Oxides and Sulphides oi the
Transition Metals," J. Appl. Phys. suppl vol 32 tflO,
pp 2195-2197, October 1961
5. Ferraro, J. j\. : "Low Frequency Vibrations of Inorganic
and Coordination Compounds," Plenum Press, New York, 1971
6. Szc, S. X.: "Physics of Semiconductor Devices," Wiley -
Interscience, 1969
7. i,ei£h;:on, P. A. (Stanford Department of Cheraistry):
"Phouochemiucry of Air Pollution," Academic Press,
New York, 1961
3. Sheehy, J. P., Achinger, W. C., Simon, ?v. A.: "Handbook
of Air Pollution," Environmental Health Series, U. S.
Sept. of Health, Education and Welfare National Center
for Air Pollution Control, Durham, N. C.
8. Bennewitz, Paul F'. ; "The Brady Array - A New Bulk-Effect
Humidity Sensor," Society of Automotive Engineers #730571
May, 1973
- 41 -
------- |