EPA-600/2-76-277
November 1976
Environmental Protection Technology Series
REMOTE MONITORING OF NITRIC OXIDE BY
GAS-FILTER CORRELATION TECHNIQUES
Environmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into five series. These five broad
categories were established to facilitate further development and application of
environmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to develop and
demonstrate instrumentation, equipment, and methodology to repair or prevent
environmental degradation from point and non-point sources of pollution. This
work provides the new or improved technology required for the control and
treatment of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-76-277
November 1967
REMOTE MONITORING OF NITRIC OXIDE BY
GAS-FILTER CORRELATION TECHNIQUES
by
Darrell E. Burch and David A. Gryvnak
Ford Aerospace & Communications
Corporation Formerly Known As
Aeronutronic Ford Corporation
Aeronutronic Division
Newport Beach, California 92663
Contract No.
68-02-0766
Project Officer
William F. Herget
Emission Measurement and Characterization Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences Research
Laboratory, 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 con-
stitute endorsement or recommendation for use.
11
-------
ABSTRACT
The feasibility of remotely monitoring the concentration of Nitric Oxide (NO)
in the effluent of industrial stacks has been investigated analytically and
experimentally in the laboratory. The type of instrument considered employs
two or more gas-filter cells that contain different amounts of NO. Radiant
energy emitted by the hot gas in the effluent is measured after it has
passed either through one of the gas-filter cells or through a neutral
density filter. By comparing the amounts of energy received through each
of the filters, it is possible to determine the concentration of NO in the
presence of a moderate amount of continuum-emitting material such as small
particles. A simple, single-line spectral model served as the basis for
the analytical work. Heated cells containing NO + N£ or B^jO + N2 simulated an
industrial stack for the laboratory experiments. Interference by hot H_0 in
the effluent and cold H^O in the atmospheric path causes the most serious
uncertainties in the measurements for many types of stacks.
iii
-------
CONTENTS
Page
Abstract ±±±
Figures vi
Abbreviations viii
I Introduction 1
II Summary 3
III Conclusions 5
IV Recommendations 7
V Illustration of the Spectroscopic Principles of
Remote Sensing by use of a Simple Analytical Model. . . 9
VI Experimental Procedures 36
VII Results of Laboratory Measurements 52
References 69
v
-------
FIGURES
Number gage
1 Optical Schematic Diagram of an Instrument of the Type on Which
the Calculations of Section V are based 10
2 Plot of Transmittances of GFC and Neutral-Density Filter Combina-
tions 12
•Q
3 Plots of Nv es Tatt vs (v - vo) for Two Samples with eg in the
Non-Linear Region Near the Line Center 14
•n
4 Plots of Nv eg Tatt vs (v - VQ) for Two Samples with es Near the
Linear Region. 16
•n
5 Spectral Plots of Nv es Tj vs (v - VQ) for the Attenuator and
Two GFC's 18
T>
6 Spectral Plots of NV sg TQ^J_ vs (v - VQ) for Two Temperatures. . 19
13
7 Spectral Plots of N e T, vs (v - v ) for Two Temperatures. . . 20
v S L O
8 Semi-Logarithmic Plots of Y. vs w for Various Samples 21
9 Plots of Z. vs u for Various GFC's "3
j
10 Plots of Z. vs Temperature for Various GFC's 24
11 Semi-Logarithmic Plots of Y-j vs w for Two Small Samples at Differ-
ent Temperatures with the Same Value of Yatt 26
•D
12 Spectral Plots of Nv eg Tatt for Three Samples at 450 K 27
13 Plots of Z, vs u for Five Different Values of ec 29
14 Plots of ZQ.I vs u for Five Different Values of ec 3Q
15 Plots of ec vs u for Five Different Values of Z-, 3^
16 Plots of ec vs u for Three Different Values of ZQ ]_ 32
17 Plots of 0S vs u for Five Different Values of ec
vi
-------
FIGURES (CONTINUED)
Number Page
18 Optical Diagram of the Apparatus Used With a 1.42-cm Sample Cell . 37
19 Optical Diagram of the Apparatus Used With a 200-ctn Sample Cell. . 41
20 Transmission Spectra of H20 and NO and of the Bandpass of the
Grating Assembly < 43
21 Logarithmic Plots of the Average Absorptance of NO + N£ Mixtures
vs u for Two Temperatures 53
22 Logarithmic Plots of Y, vs u for NO + N2 Mixtures at 450 K . . . . 55
23 Semi-Logarithmic Plots of Z= vs u for Four Sample Temperatures . . 57
24 Logarithmic Plots of Yj vs u for Samples of NO + N2 with Additional
Continuum Emission 60
25 Semi-Logarithmic Plots of Z* vs u for Samples of NO + N2 with Addi-
tional Continuum Emission. . . . 61
vii
-------
ABBREVIATIONS AND SYMBOLS
x --species of gas being measured
GFC --gas-filter cell that contains gas of species x
C --concentration of gas species x (ppm)
p --partial pressure of a gas species (atm)
P --total pressure of a gas mixture (atm)
•I --length of optical path through a gas
0 --temperature of gas (K) . A subscript s denotes the temperature
of the sample
u --absorber thickness of gas of species x in the sample (atm cm or
ppm meters), u = p£ . 1 atm cm = 10 ppm meters
w --absorber thickness of gas of species x in a gas-filter cell at
300 K (atm cm) . w = p£ .
v, VQ --wavenumber (cm ) . VQ denotes the center of an absorption line
T --transmittance that would be observed with infinite resolving
power. Subscripts denote different gases or optical components
as follows: s, sample that may include gas species x as well
as other gas species. s,x gas species x in sample, s ,y gas
species other than x in sample. s,c continuum in sample.
--The index letter j denotes a GFC. A specific GFC may be identi-
fied by a letter b, c, etc., or by a number such as 1.0, or 0.1
that indicates the value of w for that particular. GFC. For
example, T^, is the transmittance of GFC-b, (gas-filter cell b)
including its associated neutral density filter. ^b,g anc* Tb n
represent the transmittances of the gas only, and the neutral-
density filter only, respectively, of GFC-b. Tb = T^ n Tb g.
Tatt, transmittance of the attenuator not associated with a
particular GFC. Tf-Q, denotes the transmittance of the filter,
_ or combination of filters, that determines the spectral bandpass.
T --average transmittance over the spectral interval of interest
A --absorptance (1-T)
e --emissivity. The subscripts used with T also apply for ^f, A, A
B e, and e
Nv --spectral radiance of a blackbody at the temperature of the hot
sample jjj,watts/(cnr cm"1 ster)]
k --absorption coefficient due to a single line of gas species x.
k - S
n 9 2 u
(v - v ) + or
S --[ kdv, strength of the absorption line (atm cm)"1 cm"1
a --half-width of absorption line (cm"1), a = a°p (300/0)1/2
viii
-------
ABBREVIATIONS AND SYMBOLS (Cont'd)
E --radiant power incident on the detector from the sample. A sub-
script denotes the GFC through which the radiant energy passes.
Yj ~-JNv es Tj dv, a quantity proportional to the radiant power
(gwatts cm" 2 ster"^-) reaching the detector from the sample through
a GFC indicated by the subscript index j. Substitution of a,
b or 0.1, etc. for j denotes a specific GFC; substitution of
att for j denotes the attenuator that is not associated with a
particular GFC (see Equation (7)).
M --Ej/Yj. A factor that accounts for losses, for the length of the optical
path between the emitting gas and the detectorj and for the sizes
of the detector and the collecting optics (see Equation (7)).
D- --detector signal resulting from the chopped energy beam
R --responsivity of the detector. R = Dj/AEj, where AE* is the dif-
ference between the radiant power levels on the detector during
the two halves of the chopper cycle.
Zj "(Yatt - Yj)/Yatt (see Ecluation
ix
-------
SECTION I
INTRODUCTION
Full-time monitoring of all potentially polluting industrial stacks is obviously
a costly task that would require thousands of instruments. Many stacks operate
most of the time with the amount of polluting gas in the effluent well within
EPA or local standards. Therefore, continuous monitoring of all of these stacks
would be quite inefficient. For many inspection purposes, it would be adequate
to spot check these stacks periodically to see that they meet standards. An at-
tractive approach to spot-inspection involves a remote, passive monitor that could
operate from a road-side or from a parking lot at distances as large as 200-300 m
from a stack. Such an instrument could be operated at any time of day by in-
spectors with no cooperation required on the part of the management of the plant
in which the stack is located. The work reported here has dealt mostly with the
remote monitoring of NO, one of the major pollutant gases in the effluent from
stationary sources.
A class of instruments that employ gas-filter cells (GFC's) as the key component
have demonstrated good performance as monitors of several gases, including NO,
that exhibit sharp structure in their infrared spectra. In these instruments
the sample gas absorbs energy emitted by a radiant energy source that is a part
of the instrument. The GFC contains the gas species that is to be monitored in
a sample that may contain several other gas species and particulate matter. The
good sensitivity and good discrimination possible with these relatively simple
instruments result from the correlation between the spectral structure of the
gas in the GFC and that of the gas species to be measured in the sample. Strong
positive correlation exists if there is sharp structure in the spectra because
both gases are the same species. Without the need for complicated,inefficient
dispersing instruments, the GFC provides a simple means of comparing spectral
structures with very high effective resolution; the spectral resolution corres-
ponds approximately to the widths of the separate spectral intervals that are
opaque near each absorption line of the gas in the GFC. Most instruments em-
ploying GFC's do not require the beam to be well collimated nor to be passed
through a narrow slit; consequently, the "throughput" of radiant energy can be
quite high.
The purpose of the analytical and laboratory work reported here has been to in-
vestigate the feasibility of extending gas-filter correlation techniques to pas-
sive, remote sensing of pollutant gases in the effluent from industrial stacks.
Although many of the principles and findings apply to any gas species, the em-
phasis has been on NO. Field instruments of the type studied receive infra-
red radiant energy emitted by the hot gas being investigated. The concentra-
tion of NO is to be determined by measuring the different amounts of infrared
1
-------
energy received from the hot gas through a series of filters. Two or more of
the filters contain NO, or whatever other gas species is being measured. Another
filter may simply be a neutral-density attenuator. At least in principle, the
measurements through the attenuator and two or more GFC's provide enough infor-
mation that the NO concentration in the stack effluent can be determined without
knowing the gas temperature or the amount of continuum energy emitted by parti-
culate matter that may also be in the effluent.
Under a previous contract with with EPA, we designed and built an across-the-
stack instrument to measure the concentration of either NO, CO, SO?, HCl or HF
in the effluent from stationary emission sources. Although the sample gas is
studied in absorption in this previously built instrument, many of the principles
are the same as in a passive instrument. Thus, many of the findings of the pre-
vious work have been carried over to the present investigation. The most im-
portant part of the design of the NO channel of the across-the-stack instrument
was not in obtaining sensitivity but in minimizing interference by I^O vapor.
The hot H20 vapor is present in most stack effluent and has several absorption
lines in the only infrared region where NO absorbs or emits significantly. The
research performed as part of the previous contract led to the conclusion that
the least interference by H90 vapor was realized if the instrument operated in
the spectral band from approximately 1896 cm~l to 1907 cm"-*-. Most of the lab-
oratory tests of the present contract were performed with the instrument adjusted
to this ll-cm~l wide spectral interval.
1. Burch, D. E. and D. A. Gryvnak. Infrared Gas Filter Correlation Instrument
for In-Situ Measurement of Gaseous Pollutants. EPA-650/2-74-094, Environ-
mental Protection Agency, Washington, B.C. Prepared by Aeronutronic Ford
Corp., under Contract No. 68-02-0575, December 1974. Also, Burch, D. E.
and D. A. Gryvnak, "Cross-Stack Measurement of Pollutant Concentrations
Using Gas-Cell Correlation Spectroscopy," Chapter 10 of Analytical Methods
Ap^lied_to_Air Pollution Measurements, R. K. Stevens and W. F. Herget (Eds.)
Ann Arbor Science Publishers, Ann Arbor, Michigan, 1974, pp 193-231.
-------
SECTION II
SUMMARY
The optical principles and the problems involved in the passive, remote sensing
of NO in the effluent from stationary sources have been investigated analytic-
ally and experimentally. The type of instrument considered employs gas-cell
correlation techniques with one or more gas-filter cells that contain different
amounts of NO. Each GFC acts as a highly selective filter that absorbs strongly
near the center of each NO absorption line. The widths of the spectral inter-
vals of strong absorption are different for each GFC because of the different
amounts of NO.
The hot NO in the effluent gas under study also emits strongly, according to
Kirchoff's law, at the same wavenumbers where the GFC's absorb strongly. The
instrument compares the relative amounts of energy emitted by the hot NO that
are transmitted through each of the GFC's as well as through an attenuator that
has constant transmittance over the spectral interval passed. From the compari-
son, it is possible to obtain information that is related to the spectral shape
of the NO emission near each line. From this information, it is possible to
determine the product C^ of the NO in the hot gas without knowing the gas tem-
perature or the amount of emitted energy that is due to continuum emission from
particles or other gases. C is the concentration, and -t is the optical path
length through the sample.
The analytical part of the investigation is discussed in Section V and is based
on a simple spectral model that consists of a single spectral line. The strength
and width parameters assigned to the model line are similar to those of a typical
NO line in the spectral interval being considered for a field instrument. A
simple computer program has been used to make very precise calculations of the
emitted and absorbed energy based on the one—line model. It is assumed that
energy emitted by the hot gas is transmitted through either one of the GFC's or
the attenuator to a detector. The spectral radiance of the hot gas and the
transmitted power are calculated for a very narrow spectral interval over which
the spectral radiance is nearly constant. Each such calculation is repeated
for adjacent narrow intervals; the results for all intervals are then summed to
integrate over the spectral interval from the center of the single model line
to a point midway to where the next line center would be. The calculations have
been repeated for a variety of hot sample temperatures and gas concentrations
and with different amounts of continuum emission contained in the hot sample.
Several figures illustrate the spectroscopic principles of detection and the
effects of changing the different parameters: the amount of emitting gas, sample
temperature, continuum emissivity, and amount of gas in a GFC. Three separate
-------
measurements of the apparent spectral radiance of a hot sample as observed
separately through an attenuator and two properly selected GFC's are sufficient
for the determination of the gas concentration without a-priori knowledge of
the gas temperature or of the amount of continuum emission.
Section VI describes the laboratory apparatus and procedures used to make a
series of measurements under simulated field conditions. Either one of two
heated cells contained samples of NO + N£ or of 1^0 + N£ to simulate the ef-
fluent from a stack. The shorter, 1.42-cm, cell was employed for several mix-
tures of different NO concentrations between 17= and 1007= with sample temper-
atures between approximately 410 K and 450 K. A sapphire disk was placed next
to the hot sample cell for a few of the measurements to investigate the effect
of continuum emission in addition to the emission by NO. The continuum emis-
sion simulated emission by particulate matter in typical stack effluent.
A double-pass cell with a 200-cm optical path-length was heated to 445 K to con-
tain samples of 1^0 + N£. Transmission spectra of the hot H^O were obtained,
and the interference caused by emission from this hot gas was investigated.
Even in the optimum spectral interval for NO monitoring, the hot NO in a typical
stack of interest probably emits less infrared energy than the hot H20. Thus,
the monitoring instrument must discriminate well against 1^0 to avoid having
this gas produce serious errors.
A small grating assembly that resembles a grating monochromator served as a
narrow bandpass filter. Most of the measurements were made with the grating
assembly adjusted to pass the 1896 - 1907 cm"l interval, which was selected be-
cause the H£0 absorption is less than in any other interval of similar width
in the strong part of the NO band. A few data on K^O interference were also
obtained with the spectral interval shifted a few cm"*- in each direction.
Two GFC's, each 1 cm long, were employed in the laboratory apparatus; one con-
tained 1 atm of NO, and the other, 0.1 atm. The two GFC's and an attenuator
were mounted on a sliding assembly that allowed either of the three components
to be moved easily into the beam of energy emitted by the hot sample under study.
The energy beam was chopped at 450 Hz and was detected by an InSb detector
cooled by liquid nitrogen. The detector signal was processed by a synchronous
demodulator and amplifier to produce a dc voltage proportional to the chopped
energy on the detector. Three separate signal measurements were made for each
sample, one with each GFC and one with the attenuator in the beam. Convenient
parameters based on the relative values of the three signals were related to
sample conditions. The results have been presented in graphical and tabular
form. These results provide most of the information on the spectral properties
of hot NO and on the characteristics of the GFC's containing NO that is re-
quired to estimate the sensitivity of a field instrument of the type considered
here. The data on H20 interference also make it possible to estimate the per-
formance limitations imposed by this gas.
-------
SECTION III
CONCLUSIONS
Gas-filter correlation techniques have been shown to be quite adaptable to the
remote sensing of NO because of the sharp spectral structure due to the strong,
well-separated absorption lines of this gas. In addition, the strong vibration-
rotation lines of NO occur in the spectral region (near 1900 cm"1) for which
there are sensitive detectors and where the spectral radiance is relatively high
for a blackbody near the temperature of stack effluent.
The number and types of radiometric measurements required to determine the NO
concentration in a sample plume depend on the nature of the source of emitted
energy and on the amount of information that can be determined by other methods.
In all cases, we assume: that the NO concentration and the temperature are uni-
form over the optical path through the plume; that the length of the optical
path through the plume can be determined by other methods; the plume is at 1 atm
pressure; the temperature and amount of the emitting NO are high enough that the
emitted power can be measured without serious errors due to detector noise;
emission by the chopper and other instrument components can be accounted for;
interference by clouds and the atmosphere beyond the plume can be accounted for;
the instrument is sensitive only to energy in one selected spectral interval;
and the instrument has been calibrated so that detector signals can be related
directly to the amount of radiant power on the detector.
In the simplest case, the sample plume temperature is known and the plume con-
tains no material, other than NO, that emits infrared energy in the spectral
interval of interest. In this case, the concentration of NO can be determined
from a single radiometric measurement without a gas-filter cell. In a less-
simple case, the temperature is known, but other gases and/or particulate matter
in the plume emits an unknown amount of continuum energy; we assume that the
emissivity ec of the continuum is the same for all wavenumbers. Two radiometric
measurements are then required to determine the concentration of NO in such a
plume. One radiometric measurement is made directly, or with the energy passing
through a neutral-density attenuator; the other is made through a properly chosen
GFC filled with NO. If the temperature is unknown and there is continuum emis-
sion, three radiometric measurements are required: one directly or through a
neutral-density attenuator, and one each through two separate GFC's that con-
tain different amounts of NO.
Interference by HoO probably imposes the most serious limitation on the per-
formance of a remote NO monitor of the type considered here. Absorption by H20
in the atmospheric path between the plume and the instrument causes some unavoid-
able interference; this interference can probably be accounted for reasonably
-------
well for most field conditions. The most serious interference results from the
emission by 1^0 in the hot plume being monitored. The spectrum of the energy
emitted by the HoO contains spectral structure that makes it impossible to treat
the H20 emission as continuum emission. In order to accurately account for the
H20 emission, it is necessary to determine ahead of time the response of the
instrument to emission by a variety of H20 samples at different temperatures.
The H20 concentration in the plume must then be estimated by other methods so
that the radiometric measurements can be adjusted to account for the 1^0 inter-
ference .
The spectral interval between 1896 and 1907 cm"1 is probably about the best com-
promise based on the amount of radiant energy emitted by NO and on the minimum
interference by HoO. Wider spectral intervals would increase the amount of
energy received, making it possible to use detectors of lower detectivity and/or
smaller collecting optics. However, the interference by H20 increases rapidly
as the interval is widened or shifted. Use of a spectral interval significantly
narrower than the recommended 11-cm"1 wide one would decrease the amount of the
radiant power received to the point that detector noise might become a limiting
factor in the measurement accuracy. In addition, interference filters for this
spectral region ( « 1900 cm"1) with a bandpass narrower than approximately
11 cm are much more difficult to fabricate than ones with a wider bandpass.
A field instrument would require an interference filter along with an optical
system with a greater "throughput" than is possible with the grating assembly
used in the laboratory tests.
The minimum detectable thickness of NO for a GFC field instrument monitoring
stacks of interest would probably vary fron less than 0.005 atm cm to more than
0.1 atm cm. The smaller value corresponds to a stack of known temperature with
little or no 1^0 in its effluent. The larger value corresponds to large stacks
with 100 atm cm of 1^0 in the plume at an unknown temperature. These estimates
are based on a GFC instrument operating in a single spectral interval and located
between approximately 100 m and 200 m from the plume. It is also assumed that
the H£0 content of the effluent can be estimated by other methods and that no
effluent gases other than 1^0 produce any significant interference. In addition,
it is assumed that the continuum emissivity due to particulate matter is less
than about 0.1. Performances somewhat better than those indicated above could
probably be realized by making similar sets of measurements in two or more spec-
tral intervals in which the relative emissions by NO, H20, and the continuum
are different.
-------
SECTION IV
RECOMMENDATIONS
Although the ultimate accuracy of a remote, passive NO monitor of the type con-
sidered here may be limited seriously under certain conditions, its application
to less severe conditions should be investigated further. For many applications
of remote sensing, errors as large as + 20% may be acceptable. Gas-filter corre-
lation instruments appear to be capable of + 20% accuracy. Furthermore, no other
type of simple NO monitor appears to have more promise for the near future than
does the GFC type described in this report. Recommendations for additional in-
vestigations and further development of a GFC instrument are listed below under
two sub-headings: (1) analytical, and (2) laboratory and field tests.
ANALYTICAL
A field instrument that contains the same basic components as the one employed
in the laboratory experiments of this investigation should serve as the basis
for several calculations. A preliminary design of the instrument should be made
with mirrors, choppers, apertures and detectors of the sizes and in arrangements
that would be practical for a field instrument. Two 1-cm long GFC's should be
assumed with 1 atm of NO in one cell and 0.1 atm of NO in the other. The spectral
filter should pass the spectral interval between 1896 and 1907 cm~l; the average
transmittance assumed for the filter should be consistent with multi-layer inter-
ference filters that can be fabricated for this spectral interval. Such an in-
terference filter would necessarily replace the complex grating assembly employed
in the work reported in Section VII. The detector employed in the design should
be liquid-nitrogen cooled InSb with a detectivity within the present state-of-the-
art. The minimum detectable amount of NO as imposed by the detector noise should
be calculated for the instrument for various samples of NO 4- N£ at temperatures
between approximately 400 K and 460 K.
Values of radiance for samples consisting of different amounts of NO and at dif-
ferent temperatures can be obtained from the experimental data presented in Sec-
tion VII. The values of the minimum detectable amount of NO calculated in this
manner do not account for interference by H20 and continuum emission and, there-
fore, represent the most optimistic instrument performance. In order to estimate
more realistic performance, typical concentrations of H^jO and typical emissivities
of particulate matter should be assumed. Data given in Section VII on the inter-
ference by continuum emission and by 1^0 should then be used with the typical
amounts of interfering materials to estimate the degradation in the performance
of the assumed field instrument. By comparing the estimated performance with the
-------
EPA requirements, it would be possible to determine the types of stacks and
situations for which the instrument could be applied. The estimated performances
should be compared with those for other methods of remote monitoring that are
available, or might be available in the foreseeable future, to the EPA.
A few modifications to the basic instrument could also be considered for addi-
tional analytical work. One possible modification involves replacing the opaque,
multi-blade chopper with a combination of a GFC and rotating mirrored chopper.
This could be adjusted to make the instrument sensitive only to radiant energy
of wavelengths near the strong NO lines. The other two GFC's would still be em-
ployed in the same manner, but the amounts of NO in each would need to be changed.
Improved discrimination against interference by 1^0 and continuum emission could
probably be achieved by this more complex scheme.
A different possible modification involves using two different spectral bandpasses.
The relative sensitivities of the instrument in the two different bandpasses to
temperature, H20 concentration and NO concentration would be different. Thus, the
two sets of measurements could improve the accuracy to which the interferences
could be accounted for and the NO concentration could be determined.
LABORATORY AND FIELD TESTS
It is probable that the performance of a remote sensor for NO of the type discussed
here is limited by interferences. The amount of the interference can not be de-
termined accurately in the laboratory or analytically. Therefore, a few field
tests should be performed on typical stacks to obtain additional information on
the magnitude of the interferences and to gain practical experience in accounting
for them. The first set of recommended field tests should be performed with a
simple prototype instrument of the type recommended above for further analytical
work. The instrument should employ an interference filter that passes a single
spectral interval, approximately 1896 to 1907 cm~l. Two GFC's, with approximately
0.1 atm cm of NO in one cell and 1 atm cm in the other, should be incorporated
along with an attenuator on a mechanism that periodically alternates each of these
three components into the monitoring beam. The detector signal should be processed,
as suggested in Section V, in such a way as to compare the three levels of energy
reaching the detector during the times that the two different cells or attenuator
are in the beam.
The first field tests should be performed on a typical stack that can simultan-
eously be monitored by other methods to obtain comparison data on gas temperature,
emissivity due to continuum, 1^0 concentration and NO concentration. Interpreta-
tion of the data would be simplified if different ones of these parameters could
be varied independently and if the instrument could be located within a few meters
of the stack to reduce absorption by 1^0 in the atmospheric path. A series of
measurements made at different distances from the same stack would also provide
information on interference by atmospheric HoO.
Additional laboratory tests made under controlled conditions can provide more in-
formation that would be valuable in designing a field instrument. The emphasis
for laboratory tests should be on ways to minimize interference and on ways to
account for it. The results of the recommended analytical work and preliminary
fLeld tests should provide the basis for the additional laboratory work.
-------
SECTION V
ILLUSTRATION OF THE SPECTROSCOPIC PRINCIPLES OF REMOTE SENSING
BY USE OF A SIMPLE ANALYTICAL MODEL
INTRODUCTION
This section deals with some of the fundamental spectroscopic and radiometric
principles involved in passive, infrared methods of remote sensing of pollutant
gases in the hot effluent from stationary sources. The discussion is limited
to samples of unknown temperature that may include emission by an unknown amount
of continuum in addition to the emission by gas species x. All of the results
presented in this section have been calculated on the basis of a simple, single-
line spectral model that is similar to a very narrow portion of the infrared
spectrum of NO. A real instrument would pass a spectral band wide enough to
contain several NO lines; however, the simpler model used here is adequate to
illustrate some of the problems involved and some of the limitations. The
mathematics involved in the single-line calculations is easy to follow, and
many of the conclusions reached with the single-line model can be applied dir-
ectly to the real spectrum.
In addition to illustrating the principles, this section also provides the basis
for the experimental measurements and the methods of data reduction that are
discussed in subsequent sections of this report. Because of the many parameters
involved in relating a series of simple measurements to the concentration of
the pollutant gas, the principles can be described best by holding several of
the parameters constant while systematically varying a few of the others. Ana-
lysis of the data involves accurate determination of relatively small differ-
ences between two or more measurable quantities. By employing an analytical
model, the small differences can be calculated precisely, making it possible to
observe their influences en the values of concentration that are determined.
The response of an instrument to a constant-pressure sample with optical path
length £ of gas species x at concentration C is a function of the temperature
and the product C£. We assume that £ can be determined by other methods and
that the temperature of the sample is uniform over the volume being observed.
The instrument being considered, and on which the calculations are based, is
illustrated conceptually in Figure 1. The instrument employs a series of one
or more GFC's used as spectral filters; each GFC contains a different absorber
thickness, w, of gas species x. Collecting optics that are normally a part of
an instrument of this type are not included in the conceptual diagram of Figure 1.
The discussion of Section V does not include problems related to: interference
from emission by gases other than species x in the sample, absorption by the
atmospheric path between the sample and the observer, or emission by atmospheric
gases, clouds or other objects in the background. It is assumed that the
-------
Attenuator
Tatt = 0.7415
GFC's (4)
Detector
Bandpass Filter
Tn , = 0.7630
U. 1 ,n
^^ ii r
Neutral-Density Filters ^-—
^y
Figure 1.
Optical schematic diagram of an instrument of the type on which the calculations
of Section V are based. The GFC's are each 1 cm long and are identified by the
pressure of pure gas of species x contained in them. The attenuator and the GFC's
with their associated neutral-density filters are mounted on a movable assembly
so that either GFC or the attenuator can be inserted into the beam.
-------
spectral bandpass is fixed and that the instrument has been calibrated; i.e.,
the relationship between detector response and the power incident on the de-
tector is known.
SPECTRAL MODEL AND INSTRUMENT PARAMETERS
Figure 2 shows the spectral plots of transmittance of each combination GFC and
filter. We assume that the bandpass filter limits the spectral band from the
center of the line to a point 1.5 cm~l from the line center. Each line is
symmetrical about its center; therefore, the opposite side of the line need
not be considered when calculating the average transmittance or the average
emissivity. It also follows that calculations of average transmittance, or
average emissivity, based on this simple model would also apply under some con-
ditions for a spectral band consisting of many identical lines with the center
of each line separated by 3 cm"l from the adjacent ones. The simple, single-
line model obviously cannot be applied to a many-line band when the pressure or
absorber thickness is high enough that there is significant overlapping of ad-
jacent lines. The restriction of little or no overlapping of lines is not
particularly severe for the present application because there is little over-
lapping of adjacent NO lines for the spectral intervals and absorber thicknesses
of interest.
The GFC shown in Figure 1 with the most gas contains one atm cm of gas species
x; the transmittance curve of the gas is illustrated by curve D of Figure 2.
This GFC has no neutral-density filter associated with it so that Tj_ n * 1.
(The subscript 1 identifies the GFC by the absorber thickness, w, of'the gas,
and the n refers to the neutral-density filter for that GFC. The subscript j
is used as a general index to indicate the GFC identified by that letter. The
subscript g is used below to indicate the transmittance of the gas in a parti-
cular GFC.) The filters associated with each of the other three GFC's are ad-
justed to provide the same average transmittance of the gas-plus-filter combi-
nation as the GFC containing 1 atm cm of gas. The attenuator also has this
same transmittance. Thus each correlation cell, or the attenuator, transmits
the same energy from a blackbody source to the detector.
The following equations show this relationship and define the various trans-
mittances.
Tatt = Tl = T0.5 = T0.2 = T0.1 = °'7415>
where
T0.5 = T0.5,gT0.5,n. •
^v o ~~ •!•/"» 1 ~ ''TV O —
0.2 0.2,g 0.2,n
T0.1 = T0.1,gT0.1,n '
Using these adjusted filters with the three GFC's simplifies the analysis of
11
-------
0.9717
0.9441
0.8641
0.7415
Figure 2. Plot of transmittances of GFC and neutral-density filter combinations.
Each combination has the same average transmittance, 0.7415.
-------
the data. In a real instrument they would not necessarily be equal because a
different gain factor could be used when reducing the data for each GFC. How-
ever, the gas concentration is related directly to the differences between the
detector signals, so there are some instrumental advantages in keeping the dif-
ferences small when there is no absorbing gas in the sample.
The intensity, S, and half-width, a, of the model line used in the calculations
a ve»«
are:
dv = TT (atm cm)" cm" at 300 K
= 300 TT/e (atm cm)"1 cm"1 at other 6 (5)
0.030 TT/e (ppm • m)"1 cm"1,
where k is the absorption coefficient for the single line.
<* = 0.06 P (300/0)0'5 cm"1 (6)
where P is the total pressure in atm.
The value of TT was chosen for the intensity S because it simplified some of the
calculations and is similar to that of the strongest NO lines. The results can
be applied to other values of S by adjusting the values of absorber thickness
because the absorption is a function of the product of line intensity and ab-
sorber thickness.
We note from the curves in Figure 2 that all of the GFC's are opaque at the
center of the NO line. However, the GFC with 0.1 atm cm is strongly absorbing
only over a region very near the line center. The absorption by. the gas in-
creases as the amount of gas in the GFC is increased. The transmittance of the
filter, indicated by T^ n is necessarily smallest for the correlation cells con-
taining the least amount of gas. Essentially all of the absorption indicated
by curve A for (v - vo) greater than approximately 0.1 cm"-*- is due to the neu-
tral-density filter associated with the 0.1 atm cm correlation cell.
SPECTRAL EMISSIVITY AND RESPONSE FOR SAMPLES WITH NO CONTINUUM
Figure 3 shows plots of the spectral radiance of two representative samples at
two different temperatures, 450 and 410 K. The value of u for the 410 K sample
has been adjusted so that the total radiance over the 1.5 cm"1 interval is the
same for both samples. Values of N§, the spectral radiance of a blackbody at
the sample temperature, are based on 1900 cm"1 and are assumed to be constant
over the narrow interval used in the calculations. Because of the strong de-
pendence of blackbody radiance on the temperature for the temperatures and wave-
numbers of interest, a considerably larger value of u is required for the lower
temperature sample.
The values of u for these two samples are probably as large as, or larger than,
the largest values of NO that would be measured with the proposed instrument.
13
-------
i—i—i—i
PQ
10
u
(atm cm)
0.1500
0.3545
v."
s
(K)
450
410
cmz cnfister Jlcm2 ster
0.5
V0 (cm"1)
1.0
1.5
Figure 3
Plots of Nv es Tatt vs (v - vo) for two samples with es in
the non- linear region near the line center. The emissivity
ec of the continuum in the sample is assumed to equal zero.
Tatt = 0.7415.
14
-------
The values plotted in Figure 3 represent the beam after it has passed through
the attenuator The spectral radiances of the samples are equal to the values
plotted divided by 0.7415, the transmittance of the attenuator.
The power incident on the detector when GFC-j is in the beam is given by the
following equation if there is no absorption in the intervening atmospheric
path.
Ej = M -f Nv es TJ dv = M Yj (7)
The "geometrical" factor M accounts for the length of the optical path between
the emitting gas and the detector and for the sizes of the detector and the
collecting optics. In most of the discussion that follows, it can be assumed
that M remains the same so that Ej is proportional to the calculable Y..
The area under each curve of Figure 3, J N§ eg Tatt dv = Yatt» is proportional
to the energy incident on the detector for the corresponding sample of emitting
gas. As explained above, eg for the 450 K sample is made smaller than for the
410 K sample by decreasing u to account for the higher value of N^ for the hotter
sample. It follows that a simple radiance measurement cannot distinguish be-
tween these two samples, although their concentrations differ by more than a
factor of two. If the temperature were known, and if there were no emission by
other gases, the NO concentration could, in principle, be determined from the
simple radiance measurement. The following discussion explains how, and under
what conditions, the GFC's can use the difference in the shapes of the two curves
of Figure 3 to determine the concentration without the temperature being known.
Figure 4 shows a pair of curves similar to those in Figure 3 but for much smaller
samples. The value of u for the 410 K sample was adjusted so that j N^ eg Tatt dv
= Yatt is the same for both samples. An instrument would frequently be required
to measure thicknesses of NO as small as, or smaller than, those indicated in
Figure 4. We note that the two curves in Figure 4 are much more alike than are
the two curves in Figure 3. Curve B lies slightly above curve A for (v - v0)
greater than about 0.05 cm"l, but the small difference can not be seen in the
figure. The samples represented by Figure 4 are sufficiently small that the
emissivity near the center of the line is much less than unity and is approxi-
mately proportional to the absorber thickness u. In contrast, the samples repre-
sented in Figure 3 are in the non-linear region so that increases in u do not
substantially increase the emissivity near the line centers. Thus, as the value
of u is increased to account for decreasing blackbody radiance associated with
the decreasing temperature, the spectral radiance increases in the wings of the
line at the expense of decreasing radiance near the line center. Because of the
similarity of the two curves in Figure 4, it follows that any instrument depend-
ing on the spectral characteristics of the radiant energy cannot easily distin-
guish between the two samples, although they correspond to quite different con-
centrations of NO.
The emissivities at the center of the line for the four samples represented in
Figures 3 and 4 are as follows:
u (atm cm)
e (K)
es (at v0)
15
0.1500
450
0.871
0.3545
410
0.994
0.0150
450
0.184
0.0257
410
0.308
-------
cmz cm"1 ster
450 18.82
410 10.40
Figure 4. Plots of Nv es Tatt vs (v - v0) for two samples with es near the
linear region. The continuum sc = 0. Tatt = 0.7415.
16
-------
Several of the curves that follow will help in understanding the spectroscopic
principles of sensing and of determining concentration for conditions under
which the emission by the hot gas is in the non- linear region.
The four curves of Figure 5 present the spectral distribution of the radiant
energy reaching the detector for one of the samples listed above. Curve A cor-
responds to having no GFC or attenuator in the beam; curve B corresponds to the
attenuator. The other two curves correspond to the GFC and filter combinations
indicated by the value of w. The value of Yj is proportional to the detector
signal for each situation. Because of the correlation between the position of
the emitting line and the absorbing line, the GFC with 1 atm cm transmits less
than one-fourth of the energy incident upon it. Recall that its average trans-
mittance to blackbody radiation is 0.7415. Note also that the 0.1 atm cm GFC
and its associated filter absorb a large fraction of the energy for (v - VQ)
less than 0.05 cm'1. Curves for the other two GFC's (0.2 and 0.5 atm cm) are
not shown, but their positions can be estimated on the basis of the curves in
Figure 2. It is apparent that each GFC has a strong influence over a different
portion of the spectrum; therefore, measurements made with the series of GFC's
can provide information about the shape of the emission line.
Figures 6 and 7 show quantities proportional to the spectral radiance for the
two samples illustrated in Figure 3 that have the same value of Yatt. Figure 6
represents the radiant energy at the detector after it has passed through the
0.1 atm cm GFC. Similarly, the curves in Figure 7 correspond to the 1 atm cm
GFC. Values of Y^, which are proportional to the areas under the curves, are
tabulated in the captions of the figures. Note that either GFC absorbs a bigger
fraction of the energy emitted by the source at higher temperature and smaller
u than of the energy emitted by the low- temperature source.
Figure 8 shows a plot of Y^ vs w, the absorber thickness of gas species x in
the GFC's. Curve A corresponds to the 410 K sample represented in the previous
figures. The point at w = 0 corresponds to the detector signal when the atten-
uator is in the beam. The other 4 points on the same curve correspond to Yj,
which is proportional to the detector signal when the four GFC's shown in Figure 1
are in the beam. Curve E corresponds to the 450 K sample represented in previous
figures. The other curves correspond to other samples at the temperatures indi-
cated. In every case the value of u has been adjusted to produce a constant
value of
Recall from the discussions of Figures 1 and 2 that Yj would be the same for
all values of w if the emitted radiation were continuous in nature. The differ
ence between Y tt and Y^ for any one of the GFC's is a measure of the structure
in the spectrum of the emitted energy that is correlated with the structure in
the spectrum of the gas in the GFC.
We first consider measuring the value of Yatt and assume that for each value of
Yatt we have available a family of curves similar to those shown in Figure 8.
Next, the value of YI is measured. It is possible, in principle, to then de-
termine a unique value of u by interpolating between the curves of the appro-
priate figure corresponding to Figure 8. It is not necessary to have the com-
plete set of curves of Yj vs w; only a series of values corresponding to the
value of w used is required. Thus, when all of the sample emission is by gas
17
-------
10
T T
1 T
= 0
= 0.3545 atm cm
- 410 K
= 10.40
jowatts
2 cm-1 ster
T.
Cliwatts \
cm2 ster I
1 1.919
0.7415 1.423
0.7630 1.168
1 0.394
0.5 1.0
v - v (cm" )
1.5
Figure 5. Spectral plots of Nv eg Tj vs (v - VQ) for the attenuator
and two GFC's. Curve A represents the corresponding
quantity with neither an attenuator nor a GFC in the beam.
18
-------
CO
1
-------
NJ
O
I
-------
I
0)
4J
03
0.2
0.3545
0.2756
0.2193
0.1798
0.1500
0.1272
0.1230
0.1049
0.0902
0.0780
0.0680
0.0595
0.5 ;;;i:EE
0.5
w (atm cm)
1.0
Figure 8. Semi-logarithmic plots of Y. vs w for various samples.e = 0. The points
represent the values of w that correspond to the four GFC's illustrated in
Figure 1.
-------
species x, the value of u can be determined with only the attenuator and one
correlation cell. Measurements made with the other correlation cells would
provide additional data and tend to improve the accuracy of the measurement.
However, for samples of pure gas of species x with no continuum emission or
emission by interfering gases, the value of u can be determined with exact
measurements of Yatt and Yj for one of the GFC's. If a single GFC is to be
used, it follows that an appropriate value of w must be used for the range of
values of u to be measured. The advantage of using more than one GFC will be-
come apparent in later discussions of samples containing continuum emission.
/
In the discussion of the method of determining concentrations from two measure-
ments, we assumed that we had available a very large number of curves corres-
ponding to those in Figure 8. Each set of curves represented a different value
of Yatt. The value of Yatt is expected to vary over wide ranges, making it
necessary to have many sets of calibration curves corresponding to many discrete
values of Yatt. It would be more convenient to plot the quantity
Y - Y.
Z = -^ 1 (8)
J att
The ratio represented by Z ^ has a much weaker dependence on the sample temper-
ature or absorber thickness; therefore, fewer sets of calibration curves of this
quantity would be required. This ratio is also a measure of the correlation be-
tween the spectral structure of the emitting sample and that of the gas in the
GFC. For example, if the sample emits continuum energy with no spectral struc-
ture, then Yatt = Yj» and Z. = 0. In the other extreme, if the GFC is opaque
at all wavenumbers where the sample emits, then Y^ = 0, and Z equals the maxi-
mum value of unity.
Figure 9 shows the data from Figure 8 plotted in a different manner. The ordi-
nant is the ratio above, and the absorber thickness of the sample is the abscissa.
Each curve corresponds to a different GFC, each containing one of the different
values of w indicated. A set of curves similar to those in Figure 9 would be re-
quired for several discrete values of Yfltt. Interpolation would be performed for
values of Yatt between those values for which a family of curves is available.
Sample temperatures required to maintain the fixed value of Y tt are higher than
those of interest for values of u less than those represented.
Figure 10 shows another plot for the same data shown in Figure 8. In this figure
the sample temperature is the abscissa. Recall that the value of u depends on
the temperature because Yatt is maintained constant. Thus, from a measurement
of Yatt and a measurement of Yj for a properly chosen GFC, we can determine the
sample absorber thickness and its temperature. Of course, this is based on the
unrealistic assumptions that the sample gas is at a uniform temperature, there
is no emission by an interfering gas, there is no continuum emission by particles,
there is no interference by the atmosphere, and the measurements can be made with
good precision. It is further assumed that the absorber concentration in the
sample is sufficiently high that the emissivity near the center of the line is
high enough to have a non-linear relationship with the concentration. The re-
mainder of this section deals with conditions under which one or more of these
assumptions is not valid.
22
-------
ro
u>
-2 -1
Y • 1.423 gwatts cm ster
att
Figure 9.
Plots of Z
maintain Ygtt
. vs u for various GFC's. The sample temperatures have been adjusted to
att ~ 1-423 p,watts cm"2 star"1, e = 0. Values of w are in atm cm.
-------
>>
I
4J
4J
4-1
400
410
420
430
440
450
460
470
8 (K)
Figure
10. Plots of Z. vs temperature for various GFC's. Values of u
have been adjusted to maintain constant Yatt = 1-423 M-watts/cm2 ster
e B 0. Values of w are in atm cm.
24
-------
Figure 11 shows semi-logarithmic plots of Y, vs w for the relatively small samples
represented by Figure 4. The two samples are at 410 and 450 K with the values
of u ad justed to provide the same value of Yatt for both samples. The two curves
of Figure 11 differ by only a small amount; however, the values of u differ by
approximately a ratio of 1.6 to 1. Therefore Yatt and Y, for one of the GFC's
will both have to be measured quite accurately in order for one to determine u
from a family of curves such as the two represented in Figure 11. The relatively
weak dependence of the two curves on u could, of course, be predicted from the
two curves of spectral radiance in Figure 4. Because of the small difference in
shapes of the two curves of spectral radiance, it is to be expected that the rad-
iance measured through any correlation cell would be similar for both samples.
The calculated data presented in this subsection are based on samples with no
continuum emission nor emission by gases other than species x. The amount of
radiant energy emitted is therefore dependent only on the temperature and the
absorber thickness of gas species x. For samples with large enough absorber
thickness that the emissivity has a non-linear relationship with absorber thick-
ness, the shapes of the spectral curves of emission are different, making it
possible to determine u and 8 from two separate measurements. The quantities
measured are proportional to Yatt and Y*, where Y. is measured with a GFC con-
taining an appropriate amount of gas species x. when considering samples that
contain continuum emission, the radiance depends upon three quantities: the
temperature, u and e . Therefore, in order to determine u, it is necessary to
make at least three independent measurements of three different quantities, each
of which has a different dependence on the three parameters. If the temperature
could be determined by some other method, u could be determined, at least in
principle, from only two measurements. For the present study, it is assumed
that no other method can be depended upon to determine the gas temperature. The
following subsection deals with hot gas samples containing continuum emission.
EMISSION BY SAMPLES WITH CONTINUUM
Figure 12 shows three plots of Ny eg Tatt vs (v - v ) for three different samples.
Sample A is one of those represented in Figure 3 and is used as a basis for some
of the calculations in the other figures. Samples B and C are at the same tem-
perature, 450 K, but also contain enough particles or other source of continuum
emission to produce continuum emissivities of ec = 0.02 and 0.05, respectively.
The value of u has been varied to make Yatt constant for all three samples.
Samples such as those represented in Figure 12 form the basis for several of the
following curves used to demonstrate a procedure by which the absorber thickness,
or concentration of gas species x, can be determined from a set of three or more
measurements.
We assume that Yatt is measured first and that a complete set of calibration
charts of Z* vs u are available for many different values of Yatt- Interpolation
would be required for values of Yatt for which no calibration curves are available.
In practice, the many sets of calibration curves would probably be replaced by
data stored on a simple computer. However, it is better for our present purposes
to discuss the analysis in terms of the graphs so that the relationships between
the parameters are more apparent. By using the graphs, it is possible to get a
better understanding of the accuracy required for each of the measurements in
order to obtain the required accuracy in the concentration measurement.
25
-------
0.3
0)
•U
CO
csj
I
e
o
CO
cd
1
0.1
0.05
0.02
= 0.2041 pwatts cm ster
(K) (atm cm)
A 410 0.0257
B 450 0.0150
EtE
w (atm cm)
Figure 11. Semi-logarithmic plots of Y vs w for two small samples at different
temperatures with the same value of Yatt.
-------
0.1500
0.0950
0.0495
v - VQ (cm" )
JB
Figure 12. Spectral plots of Nv es Tatt for three samples at 450 K.
The emissivity of the continuum is as indicated for each
curve. The value of u has been adjusted to make Yatt; =
1.423 M-watts cm ster
27
-------
Figures 13 and 14 illustrate the method by which measured values of Yat,-, Zp
and Z0.i can be used to determine unique values of ec and u under certain re-
stricted conditions without an independent measurement of the temperature. The
two figures are based on Yatt = 1.423 y-watts cm"Z ster"1, the value that corres-
ponds to the samples illustrated in Figure 12 and in a few of the other figures
shown above. Figure 13 represents measurements made with the GFC containing
1.0 atm cm of gas; Figure 14 shows corresponding data for the GFC containing
0.1 atm cm of gas. Each curve in the two figures corresponds to the value of
continuum emissivity indicated. As an example, if Ya^t = 1.423 p,watts cm'2 ster"1,
Zj_ = 0.835 and Z0.i = 0.302, we can see from Figures 13 and 14 that ec = 0 and
u = 0.1 atm cm. As another example, assume that Yatt = 1.423 p-watts cm"2 ster"1,
Z, = 0.70 and ZQ i = 0.208. The only pair of values that fits the curves of both
figures is ec = 6.01 and u = 0.2 atm cm. The examples given were chosen so that
the values of ec fall on one of the curves. In most cases, it would be necessary
to interpolate between the curves to find a value of ec and a value of u that
would produce the two observed values of Z =. In interpreting the curves of
Figures 13 and 14, it is important to recall that each point on a given curve
corresponds to a different sample temperature. The ends of the curves that
represent small values of u have been omitted because they correspond to temper-
atures higher than those of interest.
The results corresponding to the other two GFC's (w = 0.5 and w = 0.2 atm cm)
are not included. As discussed above, only three independent measurements are
required. Therefore, Yatt, Z-^, and ZQ_^ should be adequate. Inclusion of
measurements with the other two GFC's could possibly reduce error, but in prin-
ciple they are not required.
The data in Figures 13 and 14 have been cross-plotted in Figures 15 and 16. The
latter two figures provide a different method for determining the values of u
and eQ from the data. Each curve represents a given value of Z =. By interpolat-
int between the curves for values of Z^ not plotted and by comparing Figures 15
and 16, it is possible to determine a unique set of values of ec and u. It is
apparent from the curves that a small error in Z. for either GFC can produce a
large error in u. After ec and u have been determined, 8S can be determined
from a set of curves similar to those in Figure 17.
In order to illustrate the errors in u that would result from errors in Z-, we
have made several calculations covering a variety of conditions. Some of the
results are summarized in Table 1. In the example represented by the first line,
Yatt = °-2041 M,watts/(cm2 ster), Z0.i = 0.247 and Zj_ = 0.625. From plots similar
to those of Figures 13 and 14 that were drawn for Yatt = 0.2041 u,watts/(cm2 ster),
we have determined that u = 0.020 atm cm and ec = 0.005. We then assume that l*±
was read correctly, but ZQ -^ was erroneously read as 0.259 instead of 0.247,
(see line 2). This realistic 5% error in ZQ^ causes the value of u determined
from the calibration curves to decrease by 29% from 0.0200 to 0.0142. The third
line corresponds to an error of 0.01 (1.5%) in Zj_; 0.635 was read instead of 0.625
the correct value* The first line of each group of three lines represents the
correct values. The second and third lines illustrate the errors produced by the
erroneous readings as underlined.
The errors assumed in the example above, 0.012 for Z and 0.010 for Z , are
about equally probable. Note that the error in Z .produces a larger error
28
-------
to
VO
cm ster
w =1.0 atm cm
u (atm cm)
Figure 13. Plots of Z- vs u for five different values of e
-------
"0.1
u (atm cm)
Figure 14. Plots of Zfi .. vs u for five different values of e .
-------
0.05
0.04
1.423 M-watts f-T-J
rn:
0.2 0.3
u (atm cm)
Figure 15. Plots of e vs u for five different values of Z...
-------
N>
0.04
0.2 0.3
u (atm cm)
Figure 16. Plots of &c vs u for three different values of ZQ
-------
LO
600 fer
500
CO
CD
400 ?
300
u (atm cm)
Figure 17. Plots of Bs vs u for five different values of ec. The curves are
based on the data in Figures 15 and 16.
-------
TABLE 1
ERRORS IN u DUE TO INCORRECT VALUES OF Z
r
Z0.1
0.247
0.259
0.247
0.261
0.274
0.261
0.255
0.268
0.255
z.
0. 1
0.207
0.217
0.207
0.208
0.218
0.208
0.187
0.197
0-187
V
Yatt
Zl
0.625
0.625
0.635
0.710
0.710
0.720
0.735
0.735
0.745
v
Yatt
z
1
0.543
0.543
0.553
0.592
0.592
0.602
0.631
0.631
0.641
— no n/i i
~ U./.UfJL
u
(atm cm)
0.020
0.0142
0.022
0.040
0.021
0.0455
0.060
0.029
0.0685
^^ 1 / O O /_
u
(atm cm)
0.070
0.055
0.076
0.100
0.0835
0.1065
0.200
0.171
0.209
|j,watts
cm2 ster
e
c
0.005
0.0031
0.00505
0.005
0.0031
0.00495
0.005
0.0031
0.0043
Uwatts
cm2 ster
e
c
0.02
0.015
0.0205
0.02
0.0148
0.0196
0.02
0.0195
0.0183
0s
(K)
404
428
400
374
407
370
358
392
376
0
s
(K)
457
483
459
446
462
447
420
427
420
i
Error in u :
297= low
107= high
487= low
117= high
517= low
117= high
Error in u
j
1
\
217= low
97= high
i
177= low
6.57= high
157= low
4.57= high
34
-------
in u than does the error in Zj; the two errors are also in the opposite direction.
The error in ZQ_^ also produces large errors in the values determined for ec and
9g. The value assumed for u (0.02 atm cm) in the first example is low enough
that the emissivity at the line center is still near the linear region; i.e.,
the emissivity is nearly proportional to u. The emissivity is definitely too
high to be in the linear region for the larger values of u represented in the
table.
Only small values of ec (0.005) are assumed for the upper portion of the table,
which corresponds to low values of Yatt. If larger values of ec had been assumed
for the same values of u and Yatt, the resulting temperature would have been
well below the temperature range of interest. As expected, percentage errors
in u are lower for the larger values of u represented in the lower portion of
the table.
35
-------
SECTION VI
EXPERIMENTAL PROCEDURES
OPTICAL APPARATUS
Figure 18 shows a schematic diagram of the optical apparatus used for most of
the emission measurements. The emitting gas being studied (called the sample)
is contained in a 1.42-cm long sample cell, which consists of a stainless steel
body with NaCl windows. Silicone rubber 0-rings provide a good vacuum seal
between the windows and the cell body. The diameter of the opening through
the windows is approximately 1.8 cm. The sample cell is supported inside a
piece of ceramic tubing with approximately 7 cm inside diameter. The ceramic
tubing forms the core of an electric furnace that is manually controlled to
the desired temperature. All samples studied were at 1 atm total pressure.
The cell can be evacuated or filled with non-emitting N2 in order to obtain
background data corresponding to no sample gas. Two gas lines connected to
the cell make it possible to flush a sample through the cell continuously to
avoid errors that might arise from leakage in the lines or from adsorption of
some of the sample gas on the walls of the tubing or of the sample cell. When
a measurement is being made, the gas flow is either stopped or adjusted low
enough so that the gas in the cell has time to reach equilibrium temperature.
The sample can be measured either in emission or in absorption. When measur-
ing the absorption, a Nernst glower serves as a source of radiant energy that
is chopped at 450 Hz by chopper 1. The energy beam is directed through the
sample cell with an image of the Nernst glower formed near the center of the
sample cell. Chopper 2 is stopped in the open position so that another image
of the Nernst is formed on aperture Ap 3. From there, the beam travels through
the grating assembly to the liquid-nitrogen-cooled InSb detector. (The grating
assembly serves as a spectral filter and is discussed in the following sub-
section.) The detector signal is processed by a synchronous demodulator and
amplifier that produces a dc signal that is proportional to the 450 Hz compo-
nent of the energy incident on the detector. Radiant energy emitted by the
sample cell is not chopped and therefore does not produce an output signal.
The average transmittance of a sample is measured by comparing the signal out-
put observed with the sample in the cell to that observed with the sample cell
evacuated. The ratio of these two signals corresponds to the average trans-
mittance of the sample over the narrow spectral interval passed by the combina-
tion grating assembly and filter. The filter placed immediately in front of the
detector passes a spectral band that includes the narrow interval transmitted
by the grating assembly and blocks higher orders of shorter wavelength energy
that are also passed by the grating assembly.
36
-------
SHUTTER
UJ
DETECTOR
\
SAMPLE CELL^LJc=^BLACKBOOY
M7
Figure 18. Optical diagram of the apparatus used with a 1.42-cm sample cell.
-------
When the gas is being studied in emission, the shutter near mirror T3 is moved
into the beam to block energy from the Nernst. Chopper 1 is turned off and
chopper 2, which also operates at 450 Hz, is turned on. The blackened blades
of chopper 2 are at room temperature; therefore, the signal output of the syn-
chronous demodulator is proportional to the radiance of the sample cell plus
the room-temperature background objects minus the radiance of the black chopper.
The spectral radiance of a blackbody at typical sample temperatures in the
spectral interval of interest is from 10 to 25 times as much as the spectral
radiance of a blackbody at room temperature. Therefore, the signal is nearly
proportional to the absolute spectral radiance of the sample cell. The shutter
near mirror T3 is also maintained at room temperature so that no signal is pro-
duced by background objects when the hot sample cell is removed from the opti-
cal path.
Apertures Ap 1, Ap 2, and Ap 3 limit the effective field-of-view. Aperture Ap 1
limits the height of the image formed on the window of the sample cell
so that none of the light is blocked by the walls of the cell. This makes it
convenient to use the visible light from the Nernst to align all of the optics
between the sample cell and the detector and to determine the field-of-view.
In aligning the cell, the image formed at the sample cell is reimaged on a
small aperture, Ap 3, which is slightly smaller than the image formed on it.
Aperture Ap 3 is then imaged on the entrance slit of the grating assembly.
Both dimensions of the slit are smaller than the corresponding dimensions of
the image formed, so that the slit is completely illuminated and the amount of
signal is not subject to slight movement of the image on the slit. Aperture
Ap 3 is not required to limit the size of the image, but it is useful in re-
ducing the amount of modulated energy reaching the detector from the blades of
the rotating chopper 2. Aperture Ap 2 is sufficiently small that it deter-
mines the angular divergence of the beam that reaches the detector. All of the
succeeding mirrors and components in the grating assembly, with the exception
of the slit,which is at an image, are underfilled. The sensitive area of the
detector is larger than the image of the exit slit formed on it. Therefore,
the field-of-view is limited by the entrance slit of the grating assembly and
aperture Ap 2. No energy emitted by the core of the furnace, except for a
very small amount that is scattered by the windows of the sample cell, can
reach the detector. Light shields not shown in the figure limit the stray
radiant energy.
The component labeled as a blackbody in the furnace is a piece of black-ano-
dized aluminum that can be remotely moved into or out of the path from which
energy can reach the detector. This blackbody is at nearly the same temper-
ature as the sample and provides a continuous source of radiant energy over the
spectral interval of interest so that the average transmittances of attenuators
and GFC's can be balanced as suggested in Section V. Although the emissivity
of this blackbody is somewhat less than unity, the energy reaching the detector
from its surface is essentially that of a blackbody at its temperature because
any energy reflected by the surface originated from the inside surface of the
furnace core. The core is also at nearly the same temperature as the sample
cell. Small errors in the effective spectral radiance of the blackbody do not
cause any significant problem in the reduction of the data. A polished sapphire
window placed adjacent to the sample cell for a few measurements simulated con-
tinuum emission with emissivity of 0.06.
38
-------
The assembly that contains the attenuator (Att) and the two gas-filter cells
(GFC-b and GFC-c) corresponds to the alternator in Figure 1. The attenautor
and the two GFC s are mounted on a single assembly that can be moved manually
so that either of the three components, or none of them, is in position for
the energy beam to pass through it. The sliding assembly has a guide and
stops so that the positions of the components in the beam can be reproduced
accurately. Each GFC is one cm long with an effective diameter of approxi-
mately 2 cm. GFC-b is filled with pure NO to 1 atm pressure; GFC-c is also
filled with pure NO, but to a total pressure of only 0.1 atm. The corres-
ponding absorber thicknesses, w, for cells GFC-b and GFC-c are 1.0 and 0.1 atm cm,
respectively. The GFC's and the attenuator are at room temperature. The at-
tenuator consists of two sapphire windows that are similar to the ones on the
GFC's. Both surfaces of each window are anti-reflection coated to increase
the amount of transmitted energy and to avoid potentially troublesome "fringes"
that can result from interference between the energy reflected from the two
surfaces of each window.
As discussed in Section V, a field instrument operating on the principles of
the instrument used for these laboratory tests would probably contain an alter-
nator that rapidly moves either one of the GFC's or the attenuator into the
beam at a frequency between 1 Hz and 30 Hz. The detector signal would then be
processed in such a way as to accurately measure the small differences between
the amounts of energy passing through each of the three components. Further-
more, each of the GFC's would have associated with it a neutral-density atten-
uator so that the three components on the alternator would be optically bal-
anced when the source of radiant energy was a blackbody near the temperature
of the samples to be studied. This means that the amount of radiant energy
passing through each of the three components would be the same. Of course,
the addition of hot NO to the sample would cause a misbalance in the attenuator
because of the selective absorption characteristics of the NO in the GFC's.
In the laboratory instrument illustrated in Figure 18, it was not necessary to
incorporate the accurately adjusted attenuators in series with the GFC's. The
signal passing through each GFC, or the attenuator, was measured by manually
sliding the assembly until the appropriate component was in the energy beam;
the average of the signal output as displayed on the strip chart recorder was
measured over a period of 10 to 30 seconds. An appropriate "correction factor"
was determined for each of the components to account for the differences be-
tween the output signals observed through each of the components when the
blackbody was being used. The appropriate correction factors were then applied
to the output signal observed when an NO sample was being investigated.
Typically, the minimum detectable difference between two signals observed
through a GFC and an attenuator was approximately 0.2%. Smaller fractional
differences could, of course, be observed if the components of the alternator
were moved into and out of the beam rapidly and the small differences were
measured directly. This smaller fractional difference in the minimum detect-
able observed signal would correspond to smaller minimum detectable concentra-
tions of NO. The method used here in the laboratory instrument was easier to
incorporate and was adequate for the purposes of the investigation because the
fractional modulation of smaller samples can be determined by extrapolating
from the data obtained.
39
-------
The receiver portion of the apparatus shown in Figure 19 is the same as the
corresponding portion of the instrument in Figure 18. The sample cell illus-
trated in Figure 19 has a base length of 1 meter and uses a mirror inside of
the sample cell to produce an optical path-length of 2 meters. The arrange-
ment illustrated in Figure 19 was used to obtain some additional data on
samples of NO, but it was used primarily to investigate the interference by
emission by hot H20. All of the H20 data were obtained with samples near 440 K;
all samples were at a total pressure of 1 atm and were introduced into the sample
cell through gas lines that were heated to avoid condensation. Mixtures of H20
+ N2 were made by first introducing the H20 into the cell to the desired pres-
sure; the valve at the cell was then closed and the lines were evacuated. The
N2 was then added to approximately 1 atm pressure. The sample cell was evac-
uated to obtain background data corresponding to no NO or H20 in the sample.
The optical path through the atmosphere is approximately 8 m long and typi-
cally contained approximately 40% relative humidity at 296 K.
The double-pass sample cell has an inside diameter of approximately 10 cm and
is wrapped with heating wire and insulation capable of maintaining a constant
and uniform temperature up to approximately 470 K. The box that contains part
of the insulation extends beyond the cell to minimize cooling of the end of
the cell on which the windows are mounted. Five thermocouples mounted at dif-
ferent positions on the cell and on one of the windows are used to monitor the
temperature at the various points. The temperatures are typically the same to
within + 3°C. The emissivities of the two windows and the mirror MC in the
sample cell are such that the energy received from the evacuated cell corres-
ponds to a gray—body with approximately 8% emissivity at the temperature of the
cell. This relatively small amount of emission can be accounted for in order
to determine the signals that would be observed with isolated gas samples cor-
responding to ones in the sample cells. Mirror MC and the windows of the sample
cell are sufficiently over-sized that the field-of-view is determined by the
entrance slit of the grating assembly and aperture Ap 2. As with the appara-
tus illustrated in Figure 18, the gas sample can be studied in absorption by
stopping chopper 2 in the open position and turning on chopper 1 and the Nernst
source. The shutter near mirror Tl is moved into the beam during emission
studies and is out of the beam during absorption studies.
SELECTION OF SPECTRAL INTERVAL
Nitric oxide is a diatomic molecule and therefore has only one strong, funda-
mental vibration-rotation absorption band in the infrared. Thus, any remote
sensing system involving infrared energy must make use of this band, which is
centered near 1876 cm'-'-. The band contains many strong, well-separated absorp-
tion lines, making it a good candidate for gas-filter correlation techniques.
Undoubtedly, the most difficult problem involved in the remote sensing of NO
by GFC techniques results from interference by H20, which also absorbs and
emits throughout a spectral region that includes the NO band. The proper
selection of the spectral interval is very important, more because of the need
to reduce the possible interference by H20 than to optimize the sensitivity of
the instrument. In previous investigations, we have found that interference
by H20 cannot be completely avoided, but it can be greatly reduced by careful
40
-------
MB
DETECTOR
\
GFC-b
/ L.-J--GFC-C
CHOPPER 2
T2
Figure 19. Optical diagram of the apparatus used with a 200-cm sample cell.
-------
selection of the narrow spectral interval, or intervals, to be used. It is ap-
parent that the spectral interval must be sufficiently narrow and positioned so
as to include a minimum of H20 lines and not include excessive continuum emis-
sion or absorption by the extreme wings of strong I^O lines centered outside of
the interval. The minimum practical width of the interval is limited by the
minimum amount of energy that must be received in order that the observed sig-
nals are large compared to the detector noise.
Figure 20 shows transmission spectra of NO and K^O over the region that con-
tains most of the R-branch of the NO band. The need for care in minimizing
the interference by H20 is apparent from the amount of absorption and emission
by H20. The absorption by 1^0 is even stronger in the P-branch of the NO band,
which occurs at lower wavenumbers than those included in the figure.
An optical arrangement similar to that shown in Figure 19 was employed to ob-
tain the data illustrated in Figure 20. The Nernst glower was used along with
chopper 1, making it possible to measure the transmittance of samples contained
in either the heated 200-cm sample cell or in one of the GFC's. The beam of
radiant energy was intercepted by a mirror placed just ahead of mirror M6
(Figure 19) to divert the beam to a small grating monochromator that is not
shown in the figure. Panel III of Figure 20 shows the spectrum of the H/jO in
the approximately 8-meter air path of the energy beam. Panels I and II con-
tain spectra of NO in addition to the H20 in the air path. Most of the structure
in the spectra due to NO can be observed by comparing either Panel I or II with
Panel III to account for the H20 absorption. The 445 K sample represented in
Panel I consists of a 1 atm mixture of 0.1% NO in N2 in the 200-cm cell. The
296 K sample represented in Panel II consists of a 1 atm mixture of 20% NO in
N2 in one of the 1-cm long GFO's.
The influence of increased temperature on the absorption characteristics of NO
can be seen by carefully comparing a few of the NO absorption lines in Panels
I and II. For example, the pair of barely resolved lines very near 1900 cm-1
absorb more in the 296 K sample than in the higher-temperature sample. The
relative amounts of absorption are reversed for the higher wavenumber lines in
the H20 vapor window between approximately 1924 and 1940 cm"-'-. Near 1900 cm~l,
the absorption is greater by the low-temperature sample because it contains
more NO molecules per unit cross-section of the energy beam. In addition, the
intensities of the NO lines in this spectral region are slightly lower at the
higher temperature than at 296 K. On the other hand, the increase in the pop-
ulations of the energy levels involved in the transitions that produce the
higher wavenumber lines cause the intensities of these lines to increase rapidly
with increasing temperature.
Panel IV of Figure 20 shows three spectra of H20 and 1^0 + N2 mixtures at 445 K
in the 200-cm cell. The mixtures of 10% 1^0 and of 30% t^O are at a total pres-
sure of 1 atm. The 100% H20 mixture is at approximately 0.95 atm. These three
spectra of hot 1^0 illustrate the spectral features of typical absorber thick-
nesses of H20 that might be present in the exhaust from stacks of interest.
Interference in the measurement of NO by t^O samples similar to those repre-
sented in Panel IV has been investigated and is discussed in Section VII.
42
-------
g
GO
•z.
<
o:
0
1950
WAVE NUMBER (cm'1) 1900
5.15
5.20 5.25
WAVELENGTH (jxm)
5.30
Figure 20. Transmission spectra of 1^0 and NO and
of the bandpass of the grating assembly.
The H^O in the 8-meter air path contributes
to the absorption observed for all of the
samples. The precentages of H-O in the
H_0 + N_ mixtures represented by Panel IV
are indicated.
43
-------
Curves A, B, and C in Panel III of Figure 20 illustrate the three spectral
bandpasses employed. The transmittance scale for these three curves is in
arbitrary units. The shapes and positions of the curves were determined by
passing a beam of energy through the grating assembly to a grating monochro-
mator that was used to scan the spectra. All of the data, except for some of
the data on l^O interference, presented in Section VII were obtained while
using the spectral interval corresponding to Curve B. This interval was sel-
ected to include a minimum of H20 absorption while containing four pairs of
strong NO lines. The lines passed correspond to 2 m = 13, 15, 17, and 19.
A pair of adjacent lines exist for each of these four values of 2 m. One line
of each pair corresponds to the subband of 0, = + 1/2; the other corresponds to
fi = 3/2. The higher wavenumbers correspond to the large values of 2 m. The
spectral slitwidth used in obtaining the spectra of Figure 20 was narrow enough
that the pairs of lines corresponding to 2 m = 17 and 19 are barely resolved;
whereas those corresponding to 2 m = 13 and 15 are not resolved. The positions,
intensities, and half-widths of the NO lines-have been reported by a number of
workers, including Shaw, and Abels and Shaw .
The selection of the spectral interval is based on the previous studies carried
out in our laboratory ' . It represents an optimization of minimum interference
by H20, sensitivity to NO, energy throughput, and the possibility of matching it
with an interference filter. A small, fixed-position grating assembly is used
in the present instrument to produce the accurately defined spectral interval.
The grating assembly is similar to those described in previous reports by us '
and has the important advantage that the spectral interval can be carefully
tailored during the course of the experiment. A grating assembly of this type
would probably not be used in a field instrument because of its complexity and
because of the relatively small energy throughput to which it is limited.
However, it is likely that a small interference filter could be designed and
built to pass a spectral interval very close to the one passed by the labora-
tory instrument. The energy throughput of an instrument using such an inter-
ference filter could be several times as great as that of the laboratory in-
strument, making it possible to investigate more distant sources, or smaller
sources, than could be investigated with the laboratory instrument.
Near the end of the measurements, one of the mirrors in the grating assembly
was readjusted to shift the spectral interval to either of the two positions
indicated by curves A and C in the Panel III of Figure 20. A few measurements
were made at each of these latter two intervals in order to investigate the
effect of shifting or widening the spectral interval so that it contains more
of the strong H^O absorption.
2. Shaw, J. H. Nitric Oxide Fundamental. J. Chem. Phys. 24:399-402, 1956.
3. Abels, L. L. and J. H. Shaw. Width and Strengths of Vibration-Rotation Lines
in the Fundamental Band of Nitric Oxide. Journ. Molecular Spectroscopy
20:11-28, 1966.
4. Gryvnak, D. A. and D. E. Burch. Monitoring NO and CO in Aircraft Jet Exhaust
by a Gas-Filter Correlation Technique. AFAPL-TR-75-101, Air Force Wright
Aeronautical Laboratories, Wright-Patterson Air Force Base, Ohio. Prepared
by Aeronutronic Ford Corp., under Contract No. F33615-75-C-2038, Jan. 1976.
44
-------
RADIOMETRIC CALIBRATION AND CORRECTION FOR CHOPPER EMISSION
The radiant power reaching the detector from the sample when the chopper is in
the open position and there is no attenuator or GFC in the beam is given by
E = M J NB es dv (9)
(see Equation (7)). The integration covers the spectral interval passed by the
grating assembly, and M is a constant for a given optical apparatus. When the
attenuator is moved into the beam, the incident power is reduced by a factor
equal to T the transmittance of the attenuator. Thus
ciC U
E = M T N e dv
att att v s
I N
J
The detector signal is processed by a synchronous demodulator, making the effective
detector signal, D , proportional to the difference between the power levels
incident on the detector during the two halves of the chopper cycle indicated by
(open) and (closed). The responsivity R of the detector is defined by
Eatt (°pen) ' Eatt (closed) '
D *.*. = R
att
Calibration of the instrument is complicated by the emission by the chopper blade,
which can be assumed to be a blackbody at room temperature (296 K), and by the
room-temperature walls of the laboratory. We are concerned only with radiant
energy from paths that change from one half of the chopper cycle to the other.
Energy emitted by components between the chopper and the detector is not chopped;
therefore it is not detected. When the chopper is open
\N
|_J
E (open) = M T I I NB (sample temp.) e dv +
att att I i v °
(12)
?B (296 K)(l-e ) dv
v v ' s
•>
NB (sample temp.) and NB (296 K) represent the spectral radiances of blackbodies
at the sample temperature and at 296 K, respectively. The transmittance, Tg, of
the sample is equal to (1 - e- ). The first term in the equation corresponds to
the power emitted by the sample through the attenuator and other optical com-
ponents to the detector. The second term represents the power emitted by the
wall, or by other room-temperature objects, that passes through the sample cell
to the detector. The corresponding equation for Eatt (closed) gives the power
emitted by the chopper blade,
Eatt (closed) = MTattfNv (296 K) dV> (13>
Combining Equations (11), (12), and (13) gives
D = RM T ^
att att
fNB (sample temp.) - NB(296 K)J jeg dv. (14)
45
-------
The values of N^ are calculated, and Tatt is measured. The spectral bandpass
can be approximated by an 11 cm"1 interval of constant transmittance. By mea-
suring Datt for a sample of known average emissivity eg, the product of the two
constants R and M is determined. It is not necessary to determine the values
of R and M individually; as long as their product, RM, is known, the detector
signals can be related directly to values of Y. (see Equation (7)).
Values of blackbody spectral radiance are listed in Table 2 for several temper-
atures of interest. The values at 1901 cm"1 correspond closely to the center
of the spectral band passed by the grating assembly. The ends of the interval
are near 1896 cm"1 and 1906 cm"1, the other two wavenumbers for which the spec-
tral radiance is tabulated. The ratio of the radiances near the end-points
indicates the slight change in the slope of the spectral radiance curve as the
temperature changes. Because N^ for any typical stack temperature is nearly
constant over the effective spectral interval, the values of the quantities in
Equations (12) and (13) are changed very little by removing Ny from under the
integral sign and using the mean value of this quantity for a fixed temperature
as a constant factor.
The radiometric calibration was performed by using the known average emissivity
of the 100% NO samples at 1 atm. The average absorptance, which is equivalent
to the average emissivity, was measured by using chopper 1 with the Nernst
glower source. By comparing the detector signal observed with the sample cell
evacuated to that observed with it filled with gas, the average absorptance
was measured quite accurately. At the four temperatures of interest, the aver-
age absorptances (emissivity) for the 1 atm sample in the 1.42-cm cell are:
406 K, 0.364; 423 K, 0.352; 431 K, 0.347; 450 K, 0.337. The appropriate value
of average emissivity was used in Equation (14) to determine the value of the
product RM. It was necessary to adjust the values of Datt slightly to account
for the small amount of energy emitted by the cell windows. The method used to
account for the window emission is discussed in the next sub-section of this
report.
Errors in the value used for RM do not change the values determined for Z^ and
ZQ.I (see Equation (8)) as long as the instrument is stable and the same value
of RM is used for all data in one set. One set consists of all of the values
of detector signal observed through each GFC and the attenuator for all of the
different gas concentrations at a single temperature. The accuracy of the
value of Yatt attributed to a given sample is, of course, directly dependent on
the accuracy with which RM is determined.
We had originally intended to use the block of black-anodized aluminum mounted
on a hinge near the 1.42-cm sample cell as a blackbody (eg = 1) of known tem-
perature to determine RM. However, during the tests we found that the surface
of the aluminum block apparently had a significant reflectivity. This allowed
a small amount of energy emitted by the core of the furnace to be scattered
into the beam received by the detector. Because of a difference between the
temperatures of the anodized aluminum block and the furnace core, the energy
emitted by the core changed the effective radiance temperature. In addition,
the temperature of the aluminum block varied after it was raised into the beam
received by the detector. The uncertainty in the effective radiation temper-
ature of the aluminum block made it unsatisfactory as a means of determining
46
-------
TABLE 2
SPECTRAL RADIANCE OF A BLACKBODY FOR TEMPERATURES
AND WAVENUMBERS OF INTEREST
INV
Temp.
(K)
V
1896
cm-1 star/ NB (lgQ6
(cm"1)
1901 1906
N® (1896
cm )
cm )
296
390
400
410
420
430
440
450
460
2000
0.8070
0.7944
0.7819
7.447
8.871
10.48
12.28
14.29
16.51
18.95
21.63
7.372
8.786
10.38
12.17
14.17
16.37
18.81
21.47
7.297
8.701
10.29
12.06
14.05
16.24
18.66
21.31
2788.
2797.
2805.
0.9689
0.9799
0.9808
0.9819
0.9821
0.9832
0.9837
0.9847
0.9852
1.0063
47
-------
the constant that related output signal to source radiance. This block was ade-
quate, however, as a continuous energy source near the sample temperature to
determine the relative average transmittances of the attenuator and the two GFC1
s,
As is evident from the theoretical data of Section V, the accuracy of a field
instrument of the type considered here depends strongly on the ability to mea-
sure accurately a small difference between two power levels. The ratio 2. *
(Equation (8)) is proportional to the difference between the radiant power trans-
mited through two different optical components, the attenuator and a GFC. The
instrument is designed so that this difference is to be adjusted to zero when
the source of radiant energy is a blackbody at the same temperature as the sample
gas. This adjustment is referred to as balancing the alternator and amounts to
"zeroing" the instrument. (The alternator consists of the attenuator and the
GFC's.) The temperature of the gas in the effluent from a stack is ordinarily
not known; therefore, it is important to know how strongly the balance of the
attenuator depends on the temperature of the blackbody used as the energy source.
With this problem in mind, we carefully measured the effective average transmit-
tance of the two GFC's with continuous energy sources at three different temper-
atures. One of the sources was the black-anodized aluminum block labeled as a
blackbody in Figure 18. This source was at approximately 450 K. The optical
apparatus illustrated in Figure 18 was employed, and the average transmittance
was determined by measuring the detector signal with the GFC evacuated and with
it filled with pure NO. One cell was filled with 0.1 atm of NO and the other
with 1 atm. The same gas-sampling procedure was followed while using the Nernst
glower as a source. One set of measurements was made with the Nernst glower
operating at approximately 2300 K; another set was made with the current adjusted
much lower so that the Nernst was only at about 1300 K. No significant differ-
ence was observed in the average transmittance for the same amount of NO. A
difference as small as + 0.002 in transmittance would have probably been obser-
vab le.
The right-hand column of Table 2 indicates that the slope of the spectral rad-
iance curves would be different for the three different temperatures. Thus,
the very slight dependence, if there is any dependence at all, of the average
transmittance on source temperature indicates that the NO transmittance is ap-
proximately symmetrical within the spectral bandpass. For example, if most of
the NO absorption occurred on the high-wavenumber side of the interval, the
average transmittance would be expected to decrease with increasing temperature.
The results indicate that the alternator balance would be very insensitive
to changes in source temperature. Therefore, no significant error in the lab-
oratory measurements are expected to result because of the slight difference
between the temperatures of the sample gas and the blackbody used to balance
the attenuator.
CORRECTION FOR WINDOW EMISSION
By comparing the radiance of the evacuated 1.42-cm sample cell to the radiance
of the sample cell filled with 1 atm of pure NO, we determined that the effec-
tive emissivity of the cell windows was approximately 0.03. This effective
emissivity is higher than was expected; the bulk emissivity of NaCl is known
48
-------
to be very low at the wavenumbers (near 1900 cm"1) and temperatures (390 - 460 K)
of interest. The observed apparent emissivity of the windows has been attributed
to scattering by the surfaces of the windows, which were slightly "foggy" because
of previous contact with humid air. A small amount of energy emitted by the in-
side wall of the furnace core apparently was scattered by the window surfaces
into the beam of energy received by the detector. It is not necessary that the
continuum emission by the windows be known accurately, or that its origin be
completely understood, in order to account for it adequately in determining the
contribution to the emission by the hot NO.
If the emitting NO and the windows that produce the continuum emission are at
the same temperature, the emissivity e (gas + window continuum) for the combina-
tion is related to the separate emissivities by
e (gas + window continuum) = e (gas) + e (window continuum)
- e (gas) • e (window continuum).
From this expression it follows that the value of Y^ (gas only) that would be
observed for the gas only is given by
Y? (gas + window continuum) - Y* (window continuum)
Y. (gas only) = — * (16")
1 - e (window continuum)
c
The numerator represents the difference in YJ produced by adding the NO + N2
gas mixture to the sample cell. The denominator — 1 - 0.03 = 0.97. It is ap-
parent that a small error in e (window continuum) will not cause a sizeable
error in the corrected value of Yj (gas only). The accuracy of Yj (gas only)
for mixtures of low NO concentration depends strongly on the accuracy with which
the small difference corresponding to the numerator of Equation (16) can be
measured. Equation (16) was used in reducing the emission data to determine
values of Yatt, YI} and YQ i that correspond to the gas in the sample cell.
DETERMINATION OF CORRELATION FROM TRANSMITTANCES
The apparent radiance of a source of hot NO depends on whether it is viewed
through an attenuator or through a GFC containing NO. The differences in the
apparent radiance result from the strong correlation between the spectral
structures of the hot emitting gas and the gas in the GFC. By making a few
simple assumptions that are essentially valid for the instrument described here,
this correlation can, at least in principle, be determined from a series of
transmission measurements. Values of Z, that would be determined by emission
measurements can therefore be calculated from the results of the transmission
measurements. The more accurate of the two methods depends on a combination
of several parameters, including sample temperature and sample emissivity.
The mathematical equivalence of the two methods is illustrated by the deriva-
tions and suggested measurements given below. It is assumed that the following
49
-------
three quantities are essentially constant over the narrow spectral interval
passed by the instrument: N^, the spectral radiance of a blackbody at either
the temperature of the sample or of the Nernst glower used as an energy source
for the transmission measurements; e(Nernst), the emissivity of the Nernst
glower; and Tatt, the transmittance of the attenuator. Recall that Ts = 1 - es,
and that the instrument is balanced so that ,f T-^ dv = J TQ>1 dv = J Tatt dv.
Consider the situation in which the Nernst glower and chopper 1 are being used
to measure sample transmittances. The detector signal observed with no gas in
the sample cell and with the attenuator in the beam is given by
f*
D°(att) = (Constant) N® e(Nernst) j Tafct dv. (17)
•n
NV e(Nernst) is much greater than the corresponding quantity for the chopper,
which can therefore be ignored. When a sample of transmittance T is added,
the detector signal is reduced to
T, (»
D (att) = (Constant) N" e(Nernst) ] T T dv. (18)
The ratio, D/D , of these two quantities is determined by measuring the detector
signals with the sample cell empty and with the sample gas in the cell,
Pi T dv T f
s att _ ^s ^_
ratt dV *att
D (att) _ J B att " _ s att _ - (19)
D°(att) IT dv f S
The corresponding ratios are measured, first with GFC-b, then with GFC-c re-
placing the attenuator. These two GFC's contain w = 1 atm cm and 0.1 atm cm,
respectively. From these two measurements, we obtain
D (w = 1)
D° (w = 1)
,
and (20)
D (w - 0.1) = Ts 0.1
D° (w = 0.1) - * (21)
T0.1
The expression given by Equation (19) reduces to Ts because Tatt = Tatt is
constant. However, a similar simplification can not be made to Equations (20)
and (21) because neither Ts, T^, nor TQ^ is constant.
We now consider the emission measurements of vatt, YI and YQ -^ made for the
same sample by the methods described previously in this report. In this case
NV refers to the spectral radiance of a blackbody at the temperature of the
sample, and eg is written as 1 - Tg.
50
-------
att
L0.1
r r
I ! T
L
-------
SECTION VII
RESULTS OF LABORATORY MEASUREMENTS
RADIANCE OF NO SAMPLES
The average absorptance was measured for samples with the same parameters as
those to be studied in emission. All of the samples consisted of NO + N2 mix-
tures contained in the 1.42-cm long sample cell at a total pressure of 1 atm.
The NO concentrations for the various mixtures were: 1%, 2,5%, 57=, 1070, 207o,
40% and 1007». Figure 21 shows logarithmic plots of the average absorptance,
Ag, of the different mixtures as a function of the sample absorber thickness u.
The two temperatures, 410 K and 450 K, represent the lowest and highest temper-
atures at which emission data were obtained. Data points corresponding to in-
termediate temperatures were not plotted in order to avoid crowding, but they
would fall, as expected, between the two curves shown.
Chopper 1 interrupted the beam of energy from the Nernst glower before it en-
tered the sample cell so that the detector did not respond to energy emitted by
the sample cell. The average absorptance was determined by comparing the de-
tector signal observed while the sample was in the sample cell to the signal
observed with the cell evacuated. The quantity measured corresponds to the
average absorptance over the spectral interval indicated by curve B in panel III
of Figure 20. This same spectral bandpass was employed for all of the absorption
and emission data described in this section with the exception of a few 1^0 in-
terference data discussed below.
As discussed in the previous section, the average absorptance of an NO sample
measured over this narrow spectral interval is essentially independent of the
temperature of the continuous source used. This is true because the spectral
interval is so narrow that the spectral radiance of the blackbody is nearly_con-
stant over the entire interval. Because of this^ the average absorptance, Ag,
is very nearly equal to the average emissivity, £„, of the same gas mixture.
Thus, a logarithmic plot of sample radiance vs u produces a curve with essen-
jrially the same shape as the curves shown in Figure 21. The average absorptance,
Ag, and thus Sg>of small samples can be measured more accurately in absorption
than in emission because the emission by the windows of the sample cell is auto-
matically accounted for by the "sample in-sample outf'method used in the absorp-
tion measurement. In addition, the spectral radiance of the Nernst glower is
many times higher than that of a blackbody near the sample temperature. Therefore,
the detector signal observed in absorption is larger and can be'measured more ac-
curately. Detector noise limits the accuracy to which small detector signals can
be measured.
52
-------
0.01
0.01
0.05
0.5
1.0
u (atm cm)
Figure 21. Logarithmic plots of the average absorptance of NO + N£
mixtures vs u for two temperatures. Sample cell length,
1.42 cm.
53
-------
When u is less than approximately 0.06 atm cm, the slopes of the curves in
Figure 21 are approximately equal to unity. _This indicates a near-linear rela-
tionship between the absorber thickness and Ag. For values of u greater than
approximately 0.06 atm cm, the individual absorption lines in the sample are
nearly opaque near the line centers. As the absorber thickness increases, the
only increase in the average absorptance is due to increasing absorption between
the lines.
Because of the temperature difference of the two sets of samples represented in
Figure 21, a 450 K sample contains approximately 10% fewer molecules/cm2 than a
410 K sample with the same value of u. By carefully comparing the two curves,
we can see that the value of u in the 450 K sample must be approximately 207=
greater than the corresponding value to produce the same Ag in the 410 K sample.
Thus, approximately 107, more molecules/cm2 are required to produce the same Ag at
the higher temperature as at the lower temperature. As the temperature of the
sample increases, the relative populations of the different vibration-rotation
energy levels change in such a way as to reduce the intensities of the lines of
NO that occur within the spectral interval passed by the grating assembly.
This decrease in the line intensities, along with a slight decrease in the line
widths, accounts for the decrease in the absorption cross-section of each mole-
cule as the temperature increases.
EMISSION BY GAS SAMPLES WITHOUT CONTINUUM
The logarithmic plots of Y^ shown in Figure 22 represent the emission data ob-
tained for gas samples at 450 K. The raw data have been adjusted by the pro-
cedure discussed in the previous section to account for the continuum emission
by the windows of the sample cell. The data represented by the figure corres-
ponds to isolated samples of NO + N2.
The curve labeled Yatt represents the apparent radiance of the samples as they
are viewed through the attenuator. The radiance of the samples can be deter-
mined by dividing a value from the curve by 0.594., the transmittance Ta^t of the
attenuator. The average emissivity of the 1.42 atm cm sample of pure NO at 450 K
was previously determined from absorption measurements to be equal to 0.330.
This known value of emissivity for this sample was then used by the procedure
described in the previous section to determine the constant that relates detector
signal to sample radiance. This constant, which is the product KM, (Equation
(14)) was then used to relate the detector signal to radiance for all of the
other samples represented in Figure 22.
The curve labeled Y-^ represents the apparent radiance of the source as it is
viewed through the GFC that contains 1 atm cm of pure NO. Recall from the pre-
vious discussions that the detector signals have been adjusted so that Yi = Y
when the source of energy is continuous. Approximately 8070 of the energy emit-
ted by the samples with u less than 0.04 atm cm is absorbed by the NO in the GFC
with w = 1 atm cm. This efficient absorption occurs becuase the gas in the GFC ab-
sorbs very strongly at exactly the same wavenumbers where the hot NO in the
sample cell emits.
The YQ.I curve corresponds to the energy transmitted through the GFC with 0.1
atm cm of NO. As expected, this curve lies between the other two curves.
54
-------
100
0)
U
01
CM
I
E
o
CO
•U
Figure 22. Logarithmic plots of Yj vs u for NO + N2 mixtures at 450 K.
The continuum emission has been accounted for. Sample cell
length, 1.42 cm.
55
-------
Because of the correlation between the positions of the absorbing lines of the
GFC and the emitting lines of the sample, the gas in the GFC is expected to
absorb more of the energy from the hot gas than does the attenuator. However,
the relatively small amount of gas in this GFC is opaque only over a very narrow
spectral region near the center of each line. Therefore, this GFC does not
absorb as much of the energy as the one that contains more NO. The absorption
characteristics of these two GFC's correspond closely to the curves drawn in
Figure 2 for the same values of w.
The measured values of Y^ may be in error by as much as 10 - 20%, for the two
smaller values of u (0.014 atm cm and 0.036 atm cm). This relatively large un-
certainty is due to the small change in detector signal that results when one of
these samples is added to the cell and to errors in correcting for the window
emission. The percentage error is, of course, larger for the values of Y]_ than
for the other two values of Yatt and YQ ^ because the signal being measured is
smaller. The percentage uncertainty in the values of Y« decreases with increas-
ing absorber thickness to approximately _+ 5% for the sample of pure NO.
The concentrations of NO in the effluent of most stacks of interest are probably
such that the u is between approximately 0.005 and 0-5 atm cm. If the stack
diameter is 2.5 m, this would correspond to concentrations between 20 ppm and
2000 ppm of NO. These values of absorber thickness correspond to the lower por-
tions of the curves of Figure 22 and to values lower than those plotted. For
values of u lower than those plotted, Y* can be assumed to be proportional to u.
It is unlikely that many sources of interest will be large enough or contain
enough NO to correspond to the larger values of u (between 0.5 and 2 atm cm)
represented in Figure 22.
Most sources of NO produce plumes of the effluent that are more than 1 m in dia-
meter. The 1.42-cm long sample cell used in this experiment can simulate sources
for which the source dimension, £, is much greater, but the concentration C is
much less. To a good approximation, the emission of a sample at a given temper-
ature and total pressure is a function of the product C^- Slight deviations
from this can be expected to occur if the short cell contains samples with a NO
concentration greater than approximately 30%. In this case, self-broadening of
the NO emission lines causes their widths to be slightly different from those
corresponding to a dilute mixture of NO in N2 at the same total pressure.
Data similar to those represented in Figure 22 were also obtained for the same
gas mixtures at three other temperatures; 431 K, 423 K and 406 K. Logarithmic
plots of the data obtained at the other 3 temperatures have the same shapes as
the corresponding curves in Figure 22. Of course, the values of Y; for a given
NO + N2 mixture are lower for the lower temperatures, primarily because of the
rapid decrease in the spectral radiance of a blackbody as the temperature de-
creases.
The values of Yj for the different gas mixtures and different sample temper-
atures have been substituted into Equation (8) in order to determine the corres-
ponding values of Z± and Z0>1. These values have been plotted in Figure 23 for
all four temperatures at which measurements were made. The values of Z, are
related to the amount of correlation between the spectral structures of the
-------
Ul
--J
0.2
0
0.01
0.05
0.1
0.5
u (a tin cm)
Figure 23.
Semi-logarithmic plots of Zj vs u for four sample temperatures. The
continuum emission has been accounted for. The data points are
based on measurements of Y. similar to those represented by
Figure 22.
-------
emitting gas and the GFC. Zj varies from 0 for an emitting sample that con-
tains only continuum emission with no spectral structure, to a value of 1 for
a case in which the GFC absorbs all of the energy emitted by the sample gas.
By referring to the curves in Figure 2, we can see that Z. is related to the
fraction of the emitted energy that occurs within the narrow interval near each
line center where the GFC absorbs strongly.
The data points in Figure 23 that represent 406 K samples with u < 0.1 atm cm
fall well below the points corresponding to the same mixtures at higher temper-
atures. These unusually low values of Z, are. undoubtedly a result of some sys-
tematic error in measuring the corresponding Yj's. There is no physical reason
for Z? to be lower for the smaller values of u than for larger values of u at
the same temperature. As discussed above, the uncertainty in the measured values
of Y^ is greatest for the small samples at the lowest temperatures.
A single curve represents, reasonably well, all of the data for a given GFC ob-
tained at all four temperatures. This weak temperature dependence, along with
a relatively weak dependence of Z^ on u, makes Zj a convenient parameter to use
along with measured values of Yatt to determine the NO concentration in an un-
known sample. Low values of Z^ result when a large fraction of the emission is
by continuum or by gases with spectral structure that has little or no correla-
tion with the spectral structure of the gas species under study.
Although the data illustrated in Figure 23, with the exception of the erratic
data corresponding to 406 K, show no apparent dependence on temperature, it is
likely that a slight dependence could be observed if precise data were obtained
over a wider temperature range. The shapes and strengths of spectral lines are
known to change with increasing temperature; therefore, a slight dependence of
Z on the temperature is probable.
The values of Z^ observed when using a hot NO source must be quite different
from each other if the two measurements are to provide information about the
absorption in different parts of the spectral line. The GFC with the most gas
should absorb a large fraction of the energy emitted by a sample of hot gas only.
For NO absorber thicknesses of most interest, Z^ is greater than 0.75, indicating
that 1 atm cm of NO is enough to make the GFC an efficient absorber of the energy
emitted by the hot NO. Slightly larger values of Z- could be obtained by using
more NO in the GFC; however, this increase in efficiency would probably be more
than offset by a loss in the discrimination against other gases.
The value of Zj for GFC-c is strongly dependent on the emission within approxi-
mately 0.03 cm~l of each line center and nearly independent of the emission out-
side of these very narrow spectral intervals. On the other hand, the value of
Zj for GFC-b, which contains more NO, is about_equally sensitive to all of the
emission any place within approximately 0.2 cm of each line center.
It follows that Zj for GFC-c must be much less than Zj for GFC-b when the emit-
ting source is NO. This requirement seems to be satisfied; ZQ l ^ Q.3 and Z-^ =
0.8 for values of u less than approximately 0.3 atm cm. If the amount of NO in
GFC-c were reduced to less than O.Latm cm, Zc would decrease and the sensitivity
to emission for (v - VQ) < 0.03 cm relative to that for 0.03 < (v - v )
< 0.3 cm might increase slightly. This could slightly increase the
amount of information provided by the two values of Z.. However, a
58
-------
practical lower limit on the amount of NO in GFC-c is that required to make the
gas essentially opaque over an interval within about 0.02 - 0.03 cnf1 of the
center of each strong NO line.
Although the amounts of NO in the two GFC's is not critical, the amounts used
(1 atm cm and 0.1 atm cm) are probably about optimum for the spectral interval
used. It is unlikely that the amount in GFC-c should be increased above the
0.1 atm cm being used; if any change is in order, the amount should probably be
reduced. If the spectral interval were to be widened to include weaker NO ab-
sorption lines, a slight increase in the amount of gas in GFC-b could possibly
improve performance, but it should not be increased significantly for the pre-
sent spectral interval.
EMISSION BY SAMPLES CONTAINING NO + CONTINUUM EMISSION
The influence of adding continuum emission to the sample was observed by placing
a 1 mm thick sapphire window adjacent to the 1.42-cm sample cell. The sapphire
was placed in the beam on the side of the sample cell away from mirror Ml and
was at essentially the same temperature as the sample cell. The sapphire was
carefully placed with its surface perpendicular to the central portion of the
beam of energy accepted by the instrument; therefore, there was no significant
energy emitted by the hot furnace that was reflected from the surfaces of the
sapphire window into the instrument. In the spectral interval of interest, the
1 mm thick sapphire at 450 K has an emissivity slightly greater than 0.06. The
combination of this window and the windows of the sample cell produce an effec-
tive emissivity of 0.096.
The values of Y- observed with this arrangement with different amounts of NO in
the sample cell are plotted in Figure 24. These curves correspond to the curves
of Figure 22, which represent samples of NO without any continuum emission. It
is obvious that the values of Y- should approach zero as u approaches zero when
there is no continuum emission. However, as indicated by Figure 24, all of the
values of Yj approach a common constant level corresponding to the continuum
emission as u becomes very small. When u = 0, Yatt = YQ.I = Yj_ because of the
method by which the alternator is balanced with no NO in the radiant energy
source. All of the values of Ys plotted in Figure 24 increase as u increases,
and as in Figure 22, Yatt. > YQ.I > Y].. As is to be expected, YI increases very
slowly for values of u less than approximately 0.1 atm cm. Most of the
additional radiant energy emitted by the hot NO in the sample is absorbed by the
GFC with 1 atm cm of NO.
Values of Zj based on the data shown in Figure 24 have been plotted in Figure 25.
The influence of continuum emission can be seen by comparing the curves of Figure
25 with the corresponding ones in Figure 23, which represent samples without
continuum emission. It follows that Zl and ZQ^ should approach zero as u be-
comes smaller.
The values of Z^ plotted in Figure 25 reach a maximum of 0.5 at u = 0.6 atm cm.
For larger values of u, Zi decreases as it does for increasing values of u when
there is no continuum emission present. The maximum value of Z-L is less than
59
-------
ON
O
CO
100
50
10
u (atm cm)
Figure 24. Logarithmic plots of Yj vs u for samples of NO + N2 with additional
continuum emission. ec = 0.096. Sample temperature, 450 K. Sample
cell length, 1.42 cm.
-------
0.6
0.4
0.2
0.01
Figure 25. Semi-logarithmic plots of Zs vs u for samples of NO + N2 with
additional continuum emission. The curves are based on the data
shown in Figure 24.
-------
the corresponding value of 0.73 for the same absorbe£_thickness if no continuum is
present. From the curves of Figure 21, we see that eg — 0.22 for u = 0.6 atm cm,
the amount corresponding to the maximum value of Z^ in Figure 25. As expected,
the value of ZQ.I increases more slowly than Z^ as u increases. The estimated
uncertainty in the values of Zj is between 0.02 and 0.04. The shape of the curve
corresponding to ZQ^ cannot be determined accurately, but the decrease with in-
creasing values of u greater than approximately 0.6 atm cm is significant.
Values of Y^ and Z. corresponding to those in Figures 24 and 25 can be calculated
for other values of continuum emissivity from the data presented in Figures 21,
22 and 23 for samples in which all of the emission is by NO. If the gas and the
material producing the emission continuum are at the same temperature, e (gas +
continuum), the emissivity of the combination of gas and continuum, is related
to eg, the emissivity of the gas only, and ec, the emissivity of the continuum
only, by
e (gas + continuum) = e + e - e e . (27)
g c g c
It is assumed that ec is constant over the spectral interval of interest; it
therefore follows that the corresponding values of Yj are related by
Y. (gas + continuum) = (1 - e ) Y. (gas only) + Y. (continuum). (28)
Values of Yj (continuum) can be calculated for a known ec and sample temperature.
Values of Sg can be obtained from the curves in Figure 21, making it possible to
calculate Yatt (Sas only). The values of Y^ and YQ ^ obtained from the curves of
Figure 23 are valid for all temperatures between approximately 400 K and 450 K.
These values make it possible to calculate Yj_ and YQ ^ from the values of Yatt
(gas only). Note that the only assumptions that have been made in the derivation
of Equations (27) and (28) are: the emitting gas and continuum are at the same
temperature; the spectral radiance of a blackbody is constant over the entire
spectral interval; and the emissivity of the continuum is assumed to be constant
over the entire spectral interval. Because of the narrow spectral interval used,
these assumptions can be made without introducing significant error to the data.
The experimental values of Yj and Z- shown in Figures 24 and 25 agree within the
expected uncertainty with calculated values based on the known continuum emissivity
(ec = 0.096) and the data for NO emission given in Figures 21, 22 and 23.
HLO INTERFERENCE
Because of the potentially troublesome interference by H20 in the remote sensing
of NO, we have performed a series of measurements on the absorption and emission
by H20 samples that are representative of the 1^0 in the effluent from a stack.
The samples were contained in the 200-cm sample cell illustrated in Figure 19.
All of the samples were at 445 K and consisted of either pure HoO or of H20 + No
at a total pressure of 1 atm. The results are summarized in Table 3.
62
-------
TABLE 3
SUMMARY OF H00 INTERFERENCE DATA
Interval
1896.0-1907.0 cm
-1
1.4%
(Air)
30%
1007=
1888.5-1899.5 cm
-1
107o
30%
1900.5-1911.5 cm
-1
10% 30%
s att =-
0.983 0.672 0.152 0.774 0.461 0.826 0.557
Latt
T T-
s 1
0.984 0.639 0.136 0.764 0.459
0.799 0.506
T T
s Vl
0.981 0.672 0.148 0.773 0.462 0.826 0.547
L0.1
(a)
att
L0.1
3.06 59.04 156
3.36 63.7 158
3.02 58.4 155
40.7 97.0
41.2 100.6
39.1 96.5
31.3 79.7
35.5 88.0
31.3 79.0
(b)
Z.. (transmission)
Z . (transmission)
-0.101 -0.019 -0.044 -0.037 -0.155 -0.115
0 -0.005 -0.004 -0.002 0 -0.023
Z (emission)(b) -0.098 -0.079 -0.013 -0.012 -0.037 -0.133 -0.104
Z. (emission) 0.013 0.011 0.006 0.039 0.005 0.001 0.009
vJ • JL
(R\ -2 -1
w Values of Y are in p-watts cm ster .
(b) Values of Z are based on the transmission data or emission data,
as indicated.
63
-------
Three spectral intervals were employed, corresponding to curves A, B, and C in
Panel III of Figure 20. The 1896.0 - 1907-0 cm"1 interval is represented by
curve B and is the one used to obtain the NO data presented previously. A few
data on H20 interference were obtained in the other two intervals to determine
the influence of relatively strong 1^0 absorption lines that occur just outside
of the 1896.0 - 1907.0 cm"* interval. The first column of data listed in Table 3
correspond to a sample of 1 atm of air let into the sample cell. The H20 con-
tent was not measured, but it was probably near 1.4%. Each of the other columns
represents samples of either H20 or H/jO + N2 with the t^O concentrations indi-
cated.
The first line of Table 3 gives the average transmittances of the samples as
they were measured with the monitoring beam passing through the attenuator. Be-
cause Tatt is constant over the spectral interval passed by the instrument, this
quantity represents the average transmittance of the sample. All of the tabu-
lated values of transmittance are somewhat greater than they would be if the ap-
proximately 9-meter optical path through the air were free of 1^0. The air con-
tained approximately 1.4% 1^0 and absorbed some of the energy near the H^O ab-
sorption lines. Although the atmospheric 1^0 absorbed when the H20 sample was
in the cell as well as when the cell was empty, the atmospheric H20 still in-
creases the average transmittance that is measured. The primary purpose of these
measurements was to provide data from which the potential interference of H20 in
a field instrument can be estimated. Elimination of the H20 in the atmospheric
path would probably not significantly change the estimated interference. Further
more, a field instrument will be required to operate through much longer atmos-
pheric paths, so that the effort required to eliminate the H^O from the optical
path during the laboratory measurements was not justified.
The quantities tabulated in lines 2 and 3 of Table 3 were obtained in the same
manner as the values in line 1, except that the attenuator was replaced in the
monitoring beam by either of the two GFC's. The method of obtaining the data
and the definitions of the terms are given at the end of Section VI. The rela-
tive values of the three transmittance ratios listed in lines 1, 2 and 3 for a
given sample are probably accurate to approximately + 0.002. Uncertainties in
the content of the mixtures may cause the transmittance ratios to be in error
by as much as 0.02. This relatively large uncertainty is unimportant for the
present purposes because the potential interference by K^O depends mostly on
the differences between the transmittance ratios tabulated.
The H20 absorber thickness of the 30% samples is 60 atm cm, the amount of
in a 6-meter diameter stack with 10% H20. This is probably typical of some of
the larger stacks of interest, although some stacks may be even larger. The H20
absorber thickness (20 atm cm) of the 10% samples is probably more representative
of the average stacks to be investigated. Note that the average transmittance
of the 100% sample is only approximately 0.15. This corresponds to an average
emissivity of 0.85, which is sufficiently high to make it difficult to measure
the emission by NO, even if there were no interference caused by correlation be-
tween the spectral structures of the IO and the NO.
It is well known that the collision half-widths of H20 absorption lines for a
sample at 1 atm total pressure are much greater when the sample consists of pure
64
-------
H20 than when the sample consists of H20 in a dilute mixture with N2. Collisions
of the absorbing H20 molecules with other H20 molecules are more effective in
broadening the lines than are collisions of the absorbing H20 molecules with N2.
Because of this, the lines corresponding to 100% H20 sample are probably about
five times as wide as if the sample were a dilute mixture of H20 in N2. Thus,
the results tabulated for the 100% sample cannot be applied directly to a stack
for which the H20 absorber thickness is the same, 200 atm cm. The average trans-
mittance of a dilute mixture of 200 atm cm of H20 in N2 would probably be between
0.25 and 0.30. Self-broadening of the H20 lines also has an effect on the 30%
mixture, but the effect is much less because of the lower H20 concentration.
The procedure discussed at the end of Section VI has been applied to the trans-
mittance ratios listed in Table 3 to calculate the values of Z\ and ZQ.I that
appear in the 7th and 8th lines of the table. No values of Z are given for the
1.4% H20 mixture because its absorption is so low that the small difference be-
tween the measured transmittance ratios is no more than the experimental uncer-
tainty. All of the values of Z^ are seen to be negative, which indicates that
there is a slight negative correlation between the positions of the H20 lines
and the NO lines in the spectral interval passed by the grating assembly.
A review of Figure 2 can be helpful in understanding the basis for this negative
correlation. Consider curve D of the figure, which closely resembles a plot of
the transmittance of a single NO line for the GFC that contains 1 atm cm of NO.
An H20 line would contribute to positive interference and have a positive corre-
lation with the NO if the H20 line occurred where curve D of Figure 2 lies below
the average transmittance of the attenuator, which for the example in Figure 2
is 0.7415. On the other hand, if the H20 line occurs more than approximately
0.45 cm'1 from the center of the NO line, it will produce a negative correla-
tion because the gas in the GFC transmits more than the attenuator in this spec-
tral interval. The usually good discrimination of a GFC instrument depends on
a near-random relationship between the positions of the absorption lines of the
gas to be measured and the lines of other gases to be discriminated against.
If the lines of the interfering gas are spaced so that those that produce a nega-
tive correlation exactly cancel those that produce a positive correlation, there
will be no interference in the measurement as long as the average attenuation of
the potentially interfering gas is accounted for. If the absorption by another
gas produces constant attenuation across the spectral interval, it will behave
as a continuum absorption or emission and will not contribute to Z.
Because of the consistent negative values of Z^, we conclude that the strongest
H20 emission lines occur sufficiently far from the strong NO lines that they
produce enough negative interference to more than offset the contribution by the
lines that produce positive interference. The amount of correlation is obviously
dependent on the spectral bandpass and is difficult to calculate accurately on
the basis of known parameters of the NO and H20 lines. However, the negative
correlation could probably be predicted from a careful examination of spectra of
H20 and NO such as those shown in Figure 20. None of the prominent H20 lines
occur within a few tenths of a cm"1 of the centers of the strong NO lines.
Because of the very narrow interval near each line center over which there is
strong absorption by the GFC containing 0.1 atm cm, the value of Z0.l is much
65
-------
smaller than that observed with the other GFC. The values of Zg.i based on the
transmission measurements are also negative; however, their magnitude is approxi-
mately equal to the uncertainty in the measurements.
Comparison of the values of Ts of the 30% H20 mixtures for the three different
spectral intervals illustrates the importance of choosing the proper spectral
interval. The 1896.0 - 1907.0 cm"1 interval was chosen to minimize H20 absorp-
tion and emission. The increased average absorptance in the other two intervals,
represented by curves A and C in Panel III of Figure 20, is predictable from the
H20 spectra of Figure 20. It is interesting to note that although the H20 ab-
sorption is greater in these two "shifted" spectral intervals, the absolute value
of Z-^ for the 30% mixtures is not increased appreciably. In fact, the absolute
value of Zi based on the transmission data for the 30% sample in the 1888.5 -
1899.5 cnf! interval is less than the corresponding value for the higher wave-
number interval with less average absorption.
The values of Yatt, YI and YQ.I for H20 listed in lines 4, 5 and 6 of Table 3
were obtained by measuring the emitted energy by the same method used for the NO.
The continuum emission, by mirror MC and the windows of the 2-meter cell, was
accounted for by the same procedure used to account for the continuum emission
by the windows of the 1.42 cm cell. The combined emissivity of mirror MC and the
2 windows was approximately 0.08. The values of Z^ based on the emission data
are in fair agreement with the corresponding values based on the transmission
data. The main contributors to the differences in the two values are probably
errors in accounting for the continuum emission and to instabilities in the opti-
cal components, including variations in the amount of H20 in the atmospheric path
through which the radiant energy beam passed. Some of the difference may be due
to the extra length (approximately 1 m) of the atmospheric path through which
the radiant energy passes for the transmission measurement.
The absolute values of ZQ ^ based on the emission data are in most cases too
small to be significant. The relatively large value of 0.039 for the 10% mix-
ture with the 1888.5 - 1899.5 cm'1 interval is the only value that is signifi-
cantly greater than the estimated uncertainty. There is no apparent explanation
for this unusually high value.
The interference by H20 can be discussed more easily if we treat the H20 emission
as if it consisted of two. separate parts, the non-correlated part and the corre-
lated part. The non-correlated part can be treated as if it were continuum emis-
sion. This non-correlated portion of the emitted energy would contribute exactly
the same amount to the values of Yatt, Yj^ and Y0>1; thus Z± and ZQ ^ would equal
zero. It is not necessary that the non-correlated portion be constant over the
entire spectral interval in order for Z-^ and Zg -, to be zero. It is only neces-
sary that there be no correlation, either positive or negative, between the spec-
tral structures of this portion of the emitted energy and the absorption by the
GFC's. The correlated portion can be thought of as the energy emitted by a given
amount of NO. If this correlated portion contributes to positive values of Z^
and ZQ.I, it is equivalent to a positive amount of NO.
Consider the results obtained in emission in the 1896.0 - 1907.0 cm"1 region
(Table 3) for the 60 atm cm of H20 sample that consists of 30% H20 + 70% N2.
66
-------
Note that Z^- -0.079; the negative value indicates that the correlated portion
corresponds to & negative amount of NO. As is seen in Table 3, it is possible
for Z± to be negative while ZQ>1 is positive; it would also be possible for the
signs to be reversed for a different interfering gas. Because of the difference
in the signs ( + or - ) of Z]_ and ZQ^, it is apparent that what has been defined
as the ncn-correlated part of the emission depends on which GFC is being used.
Thus, the concept of interfering emission t>eing composed of a correlated part
plus a non-correlated part is limited in its value and can not be dealt with in
a simple algebraic manner. Nevertheless, the concept is useful in discussing
the different ways in which the H20 emission interferes with NO measurements
and how it can be accounted for.
It is apparent from the data in Table 3 that hot H20 in the stack effluent being
monitored for NO can produce serious interference. Let us assume for the sake
of discussion that Z± = ZQ ^ = 0 for the 30% H20 mixture. This is equivalent to
assuming that all of the I^O emission is non-correlated and would minimize the
interference problem to that of dealing with continuum emission for which ec =
(1 - TR20) = (1 - 0.672) = 0.328. Even with this simplified problem, it is dif-
ficult to estimate the accuracy to which the concentration of NO could be mea-
sured in a stack containing this much E^O. The value of ZQ ••, which is propor-
tional to the difference between Yatt and YQ ^, is very small and must be mea-
sured quite accurately when ec is this large! The value of Z^ is larger than
ZQ_;L, and can be measured somewhat more accurately. When the effluent temper-
ature is unknown, both Z± and ZQ ^ must be measured accurately in order to de-
termine the NO concentration. If the temperature could be determined by some
other method, an accurate measurement of ZQ -^ would not be required, and the NO
concentration could be determined from measurements of Y tt and Yp It is un-
likely that a field instrument would be capable of measuring 1\ with an uncer-
tainty of less than 0.01 under the best conditions. This would correspond to
%0 — 0.004 when "e^O = 0.328. By extrapolating the 450 K curve of Figure 21,
we see that this corresponds to a minimum detectable NO absorber thickness u of
approximately 0.005 atm cm (50 ppm - m), which would be quite adequate for many
applications.
If the gas temperature is not known, as is usually the case, the minimum detect-
able u would probably increase by at least a factor of 4 to 10 because of the
reliance on an accurate measurement of ZQ ]_. This minimum detectable u, 0.02 -
0.05 atm cm, is probably too large for most applications. Of course, the efflu-
ent of many stacks contains much less H20 than is used in this example; as the
amount of H20 decreases, the minimum detectable u for NO will also decrease ac-
cordingly.
The correlated part of the interfering H20 emission further complicates the re-
duction of the data and adds to the errors. If, for example, emission from the
30% mixture were treated as if all of the H20 emission were non-correlated, the
Z, = -0.079 would lead to an NO absorber thickness of approximately -0.02 atm cm.
A negative absorber thickness has no physical meaning, but if NO were also pre-
sent in the emitting sample, the measured NO absorber thickness would be too low
by 0.02 atm cm. Many gas-filter correlation instruments of various types are
troubled by interference from gas species other than the one being measured. By
previously measuring the amount of interference by known quantities of the
67
-------
interfering gas, it is possible to account for the interfering gas and reduce
the error caused by it. This method can also be used, but to a lesser degree,
on a field instrument of the type considered here. The H20 interference could
be measured for a variety of H20 samples of known concentration; the data could
then be used along with estimates of the amount of H20 in the stack to calcu-
late the interference. This procedure is much more complex for this type of
field instrument than for a laboratory instrument that operates under stable
conditions with a known sample temperature. It is almost certain that the amount
of interference by a known amount of H20 depends on its temperature, which is
usually unknown.
Absorption by the H20 in the atmospheric path also complicates the interference
problem. Much of the error due to H20 absorption can be accounted for by bal-
ancing the three components (attenuator, GFC-b and GFC-c) of the alternator while
using a continuous emitter located near the stack as the energy source. The hot
surface of the stack just below the top might serve for this purpose. By balan-
cing the alternator in this manner, the correlated portion of the H20 absorption
will not produce errors as it would if the atmospheric paths were quite different
when measurements were made than when the alternator was balanced. Possible
errors due to the atmospheric H20 absorption are not completely eliminated in
this manner because of the strong correlation between the spectra of the emitting
H20 and the absorbing H20. The absorbing H20 changes the spectral character-
istics of the energy emitted by the hot H20 that reaches the instrument. Thus,
the correction that would have to be made for interference by emitting H20 de-
pends on the amount of H20 in the absorbing path as well as on the temperature
and amount of hot H20 in the emitting gas.
Still another mechanism for possible H20 interference exists because of emission
by atmospheric H20. The amount of radiant energy involved in this process is
low because of the low spectral radiance of a blackbody at typical atmospheric
temperatures. The H20 involved in this process includes not only that between
the hot gas source and the instrument, but also the atmospheric H20 in the line-
of-sight beyond the hot gas source. Particularly on a cloud-free day, the energy
emitted by the distant atmosphere has emission maxima due to H20 emission. The
correlated part of the structure in this background emission could also produce
small errors if it is not accounted for. Fortunately, the interference by this
background emission can probably be measured by pointing the receiver off to the
side of the stack gas so it is observing an atmospheric path very similar to the
one in the field-of-view during the measurements.
A carefully designed GFC field instrument of the type considered here with a pre-
sent day state-of-the-art interference filter and a liquid-nitrogen-cooled de-
tector would probably be limited in its accuracy by H20 interference. The mini-
mum detectable thickness of NO for stacks of interest would probably vary from
less than 0.005 atm cm to more than 0.1 atm cm. The minimum value corresponds
to a stack with little H20 in its effluent if the temperature can be determined
by another independent method. The larger value corresponds to larger stacks
with more H20 and with no a-priori knowledge of the temperature. These estimates
are based on an atmospheric path between approximately 100 m and 200 m and on the
assumption that the H20 content of the effluent can be estimated by other methods.
It is also assumed that the continuum emissivity due to particulate matter is less
than about 0.1 and that no effluent gases other than H20 produce any significant
interference.
68
-------
REFERENCES
1. Burch, D. E. and D. A. Gryvnak. Infrared Gas Filter Correlation
Instrument for In-Situ Measurement of Gaseous Pollutants.
EPA-650/2-74-094, Environmental Protection Agency, Washington, D.C,
Prepared by Aeronutronic Ford Corp., under Contract No. 68-02-0575,
December 1974. Also, Burch, D. E. and D. A. Gryvnak, "Cross-Stack
Measurement of Pollutant Concentrations Using Gas-Cell Correlation
Spectroscopy", Chapter 10 of Analytical Methods Applied to Air
Pollution Measurements. R. K. Stevens and W. F. Herget (Eds.) Ann Arbor
Science Publishers, Ann Arbor, Michigan, 1974, pp 193-231.
2. Shaw, J. H. Nitric Oxide Fundamental. J. Chem. Phys. 24:399-402, 1956.
3. Abels, L. L. and J. H. Shaw. Width and Strengths of Vibration-Rotation
Lines in the Fundamental Band of Nitric Oxide. Journ. Molecular
Spectroscopy 20:11-28, 1966.
4. Gryvnak, D. A. and D. E. Burch. Monitoring NO and CO in Aircraft Jet
Exhaust by a Gas-Filter Correlation Technique. AFAPL-TR-75-101, Air
Force Wright Aeronautical Laboratories, Wright-Patterson Air Force
Base, Ohio. Prepared by Aeronutronic Ford Corp., under Contract No.
F33615-75-C-2038, Jan. 1976.
69
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-76-277
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
REMOTE MONITORING OF NITRIC OXIDE BY GAS-FILTER
CORRELATION TECHNIQUES
5. REPORT DATE
November 1976
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Darrell E. Burch and David A. Gryvnak
8. PERFORMING ORGANIZATION REPORT NO.
U-6252
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Aeronutronic
Aeronutronic
Ford Road
Newport Beach,
Ford Corporation
Division
10. PROGRAM ELEMENT NO.
1AD712 (1AA010)
11. CONTRACT/GRANT NO.
California 92663
68-02-0766
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory
Office of Research and Development
U. S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final. 6/73-6/76
14. SPONSORING AGENCY CODE
EPA-ORD
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The feasibility of remotely monitoring the concentration of Nitric Oxide (NO) in the
effluent of industrial stacks has been investigated analytically and experimentally
in the laboratory. The type of instrument considered employs two or more gas-filter
cells that contain different amounts of NO. Radiant energy emitted by the hot gas
in the effluent is measured after it has passed either through one of the gas-filter
cells or through a neutral density filter. By comparing the amounts of energy
received through each of the filters, it is possible to determine the concentration
of NO in the presence of a moderate amount of continuum-emitting material such as
small particles. A simple, single-line spectral model served as the basis for the
analytical work. Heated cells containing NO + N2 or H20 + N2 simulated an industrial
stack for the laboratory experiments. Interference by hot H20 in the effluent and
cold H20 in the atmospheric path causes the most serious uncertainties in the measure-
ments for many types of stacks.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
COSATl Field/Group
*Air Pollution
*Nitric Oxide
*Remote Sensing
Monitors
Gas-Filter Correlation
13B
07B
14B
8. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (ThisReport)
UNCLASSIFIED
21. NO. OF PAGES
80
20. SECURITY CLASS (This page)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
70
------- |