Final Technical Report
Development and Application
of Tunable Diode Lasers
to the Detection
and Quantitative Evaluation
of Pollutant Gases
30 September 1971
Prepared for Environmental Protection Agency
under Electronic Systems Division Contract F19628-70-C-0230 by
Lincoln Laboratory
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
Lexington, Massachusetts
-------
45
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
LINCOLN LABORATORY
DEVELOPMENT AND APPLICATION
OF TUNABLE DIODE LASERS
TO THE DETECTION AND QUANTITATIVE EVALUATION
OF POLLUTANT GASES
E. DAVID HINKLEY
Principal Investigator
FINAL TECHNICAL REPORT
1 OCTOBER 1970 - 30 SEPTEMBER 1971
ISSUED 31 MARCH 1972
Prepared for
Environmental Protection Agency
National Environmental Research Center
Division of Chemistry and Physics
Research Triangle Park, North Carolina
EPA Project Officer: John S. Nader
LEXINGTON MASSACHUSETTS
-------
The work reported in this document was performed at Lincoln
Laboratory, a center for research operated by Massachusetts
Institute of Technology, with the support of the Environmental
Protection Agency under Air Force Contract F19628-70-C-0230.
-------
CONTENTS
Section Page
Summary v
1. INTRODUCTION 1
2. DIODE LASER FABRICATION 3
3. LASER SPECTROSCOPY OF SULFUR DIOXIDE 5
4. REMOTE HETERODYNE DETECTION 9
4.1 Heterodyne Theory and Calculations 10
4.2 Heterodyne Detection of Thermal Radiation
Test of Theory 12
4.3 Local Oscillator Power Requirement 15
4.4 Heterodyne Detection of Ethylene 15
5. EXPERIMENTS RELATED TO MONITORING
APPLICATIONS OF TUNABLE LASERS 18
5.1 Point Sampling Sensitivity Measurement 18
5.2 Pressure, Temperature, and Interference Effects 19
5.3 Long-Path Transmission Experiment 22
Acknowledgments 23
References 24
Appendix A: Tunable Laser Spectra of the v ^ Band of Sulfur Dioxide 25
Appendix B: Tabulation of Sulfur Dioxide Spectral Data , 37
Appendix C: NH3 Calibration Lines for the 1100-1200 cm Region 42
Appendix D: Detection of Air Pollutants with Tunable Diode Lasers 43
References 51
Appendix E: Tunable Laser Spectroscopy of the v, Band of SO, 53
References 57
-------
The work reported here was performed by:
E. D. Hinkley
J. O. Sample
T. E. Stack
A. Wilson
J. H. Boghos*
S. Duda*
W. F. McBride*
J. N. McMillan*
W. H. Laswell*
R. J. Woods*
* Denotes part time.
-------
SUMMARY
The purpose of this program is to develop and apply tunable semiconductor diode
lasers to problems associated with the monitoring of pollutant gases. During the ini-
tial phase of the program, reported herein, lasers were developed to operate in the
8.7-|im region where sulfur dioxide (SO,) has a strong infrared absorption band. Be-
cause of the potential usefulness of this band for various infrared techniques for the
detection of SO?, ultra-high-resolution laser scans were used to catalog several hun-
dred of the spectral lines at reduced pressure. A theoretical spectrum for this band
of SO,, derived elsewhere on the basis of accurate microwave spectra, was found to
be in near perfect agreement with the observed laser spectra once the band center
used in the theory was properly adjusted on the basis of the laser data. We can now
predict most of the strong SO., lines in this region, and construct with reasonable ac-
curacy the absorption or emission spectrum of this gas corresponding to anticipated
conditions of temperature, pressure, and instrumental resolution. Such information
can be used, for example, to determine the optimum wavelength for the detection of
SO2 by across-the-stack absorption of laser radiation, over a wide temperature
range. If such data were known for other constituent gases as well, particularly in
the desired spectral region, a complete model could be used to facilitate development
of suitable monitoring instrumentation. However, no such theories are available as
yet.
Although the above theory contains the temperature dependence explicitly, it
must rely on independent measurements for predictions involving pressure broadening
of the lines. The rate of pressure broadening of SO, lines by air was measured by
laser spectroscopy and found to be 12MHz/Torr for the full width at half-maximum
(FWHM); this yields an atmospheric-pressure-broadened linewidth of 0.30 cm . Sim-
ilar measurements for ammonia (NH,) in the 10-^m region produce values of 6. 3 MHz/
-1
Torr and 0.16 cm , respectively.
Experiments were performed to verify the fundamental theory related to remote
heterodyne detection of atmospheric pollutant gases. Application of this technique
would permit single-ended, passive detection from both stationary and mobile sources.
It is shown that, based on information obtained here on the SO., emission band at
8.7|j.m, concentrations of the gas as low as lOppm can be detected, essentially inde-
pendent of range, if the gas temperature is 100°C. For the same gas at 50°C, the
minimum detectable concentration is about BOppm, and the measurements become
more susceptible to changes in the background. Unfortunately, the present diode laser
power capability is not sufficient to attain the ultimate remote heterodyne detection
sensitivity, although progress is being made in both laser and detector technology.
Several other applications were studied during the first phase of this program.
1. Point Sampling. As has been demonstrated earlier, this technique using
tunable lasers has the advantage of very high specificity by permitting testing at re-
duced pressure where overlap between adjacent absorption lines can be eliminated in
most cases. With a 10-cm-long cell containing an NH^air mixture, and a tunable
diode laser with lOjiW output power, the minimum detectable concentration was found
-------
to be 28ppm using a line having an absorption constant per unit pressure of 1 cm /
Torr and an operating pressure of 4.5 Torr. For this test, second-derivative detec-
tion was used, which effectively enhances the desired signal.
2. "White" Cell Operation. Construction of a "white" cell permitted a path length
of 7.3 meters to be used for point-sampling applications. Comparison of the signal-to-
noise ratio using a 10-cm-long cell with that achieved by the "white" cell for identical
gaseous concentrations showed the predicted enhancement; consequently, the minimum
detectable concentration for the "white" cell is found to be about 0.3ppm. Since these
minimum detectable values are expected to be inversely related to laser power, it is
clear that the use of tunable diode lasers in the milliwatt power range can yield much
more sensitive point-sampling limits in the low ppb region.
3. Long-Path Transmission. Diode laser radiation was collimated with a lens
and transmitted over a 75-meter path from one building to another. Although the ex-
periment was preliminary, and no attempt was made to tune through any atmospheric
SO, lines, a strong signal was received, indicating good potential for long-path trans-
mission techniques using these devices.
4. Interferences. Some studies were made to find possible interferences be-
tween the spectral bands of certain other gases which may be found in the atmosphere
and the 8.7-^m band of SO, by using the diode laser to scan portions of this band.
Since NH, is used as the primary calibration standard for wavelength in this region,
it can under certain conditions produce interference; however, there are only approx-
imately 5 percent of the number of NH, lines as SO, lines in this region. A broader
test was to place a quantity of automobile exhaust in the test cell, at 1 atmosphere
pressure, and tune the diode laser over a region constituting strong SO, absorption.
No interference was found, despite the many gaseous components in automobile
exhaust.
-------
DEVELOPMENT AND APPLICATION OF TUNABLE DIODE LASERS
TO THE DETECTION AND QUANTITATIVE EVALUATION OF POLLUTANT GASES
1. INTRODUCTION
Tunable semiconductor diode lasers have been shown to have important potential application
in remote or long-range sensing and point sampling of molecular pollutant gases. ' Remote
sensing can be accomplished by means of measurements of the infrared absorption or emission
lines characteristic of a particular pollutant at atmospheric pressure. Point sampling permits
gas pressures to be reduced until the Doppler-broadened infrared absorption structure is re-
vealed, thereby making possible very high specificity. Since one measurement technique is
used for all the pollutant gases, operational requirements are simplified. Moreover, the es-
sentially instantaneous response is advantageous for a number of applications.
TABLE I
SELECTED WAVELENGTHS
FOR DETECTING ATMOSPHERIC POLLUTANT GASES
H C 3-4 urn
x y ^
CO 4.6
NO 5.2
C3H8 6.8
CH4 7.7
N2O 7.7
H2S 7.8
NO2 7.9
SO2 8.7
* Peroxyacetyl nitrate.
Infrared spectroscopy is considered an important technique since each of the major mo-
lecular pollutants contains at least one strong band in the 3- to 15-^m "fingerprint" region of
the infrared, as indicated in Table I. Dispersive infrared instruments now being used in the
field usually have resolutions that are much poorer than the linewidths of typical gases at at-
mospheric pressure, and thus potentially useful information is lost. Even when high-quality
spectrometers are used, the resolving power must be reduced so that a useful signal-to-noise
ratio can be maintained with the relatively low power per unit spectral range available from
incoherent sources. The use of gas lasers as nondispersive instruments, on the other hand, is
restricted by the fact that these lasers cannot be tuned appreciably from their nominal wave-
lengths, and the match to pollutant gas absorption lines is seldom ideal.
This program is concerned with the development of tunable semiconductor diode lasers
and their potential application toward solving some of the air pollution monitoring problems.
In Fig. 1-1 are shown wavelengths over which semiconductor lasers can be tuned by compositional
CH,O
2
PAN*
Q
3
C2H4
NH3
C3H6
C2H40
C2H6
CO,
8.7
8.7
9.5
10.5
10.5
11.0
11.5
12.0
13.9
-------
WAVENUM8ER (cm)"
(8-5-3!46-4|
pbi-»cd»s
WAVELENGTH (pm)
Fig. 1-1. Wavelength ranges which can be covered with alloyed semiconductor
lasers. The arrows indicate possible further extension beyond present limits.
variations, from 0.5 fxm in the visible to 32 jj.m in the infrared. It should be noted that it is pos-
sible to cover the entire 3- to 15-jim region of interest for monitoring gaseous pollutants with
three semiconductor materials: PbCdS, PbSSe, and PbSnTe.
During the past year, PbSnTe diode lasers were developed for the 8.7-^m region in order
to perform ultra-high-resolution spectroscopy on SO,. Prior to this, PbSnTe lasers had been
developed for operation in the 10-^j.m region, and had been used to study the spectral properties
of ethylene (C.,H4), ammonia (NH,), and sulfur hexafluoride (SF/), and to demonstrate the spec-
ificity, sensitivity, and reproducibility of point-sampling techniques using these lasers. More
recently, advances have been made in the development of tunable semiconductor lasers in new
wavelength regions for the detection of other pollutant gases: for example, PbSSe at 4.6 |o.m for
4 5
carbon monoxide (CO) detection and at 5.2|im for nitric oxide (NO) detection, and PbCdS at
3.4(j.m for detection of the C-H bond present in all hydrocarbons. These advances were not
made under EPA-sponsored programs, but are mentioned for completeness with regard to the
current state of the art. The tunable diode lasers developed under this program are described
in Section 2.
Section 3 contains some of the spectroscopic information obtained from the laser scans of
portions of the v^ band of SO, between 1100 and 1200 cm~ , with special attention to comparison
between experiment and theory. Some of the actual laser scans are shown in Appendix A. In
Appendix B are tabulated the line positions and strengths on the basis of these scans, as well
as their identification from theory. Appendix C lists the NH, lines used for calibration. Sec-
tion 3 in the main body of this report describes the linewidth and pressure-broadening studies
as well.
Initial experiments confirming the theory for remote heterodyne detection are described in
Section 4, and spectral emission lines from heated C~H.. were detected by this technique. At
the present time, infrared laser and detector technology does not permit optimum use of this
technique; however, calculations are presented which show the minimum detectable concentra-
tions of various pollutant gases which might be measured by remote heterodyne detection after
the state of the art has been suitably advanced.
-------
Section 5 consists of experiments related to potential monitoring applications involving
tunable lasers. The use of first and second derivative techniques in point -sampling applications
is compared, and an experiment showing present limits of detectability is described. Prelim-
inary studies of interference effects, temperature and pressure effects, and long-path trans-
mission of diode laser radiation are also included in this section.
Appendices A, B, and C contain the actual SO2 spectra, line tabulation, and NH3 calibra-
tion line tabulation, respectively, as mentioned above. Appendix D contains a journal article
published during the past year, and describing general applications of tunable diode lasers to
air pollution monitoring. Appendix E contains an article submitted for publication, which es-
sentially summarizes the laser spectroscopic measurements on
^-
2. DIODE LASER FABRICATION
In order to investigate the Doppler-limited high-resolution infrared absorption of the v^
band of SO2 between 1100 and 1200cm"1 (centered around 8.7^m), Pb.^Sn^e single crystals
of the proper chemical composition (see Fig. 2-1) had to be prepared. Although tunable
Snr
r, oo
U.oo
Te diode lasers had been used for spectroscopic measurements of NH,, C..H.
/ o £ *
SF,, and PF5 in the 10-nm region, modifications in the crystal growth and laser fabrication
techniques were needed for those tailored to the 8.7-|j.m region.
0 0.1 0.2 0.3
x, MOLE FRACTION OF SnTe
Fig. 2-1. Variation of Pb. Sn Te energy gap with mole fraction of SnTe
at 12 °K and 77 °K. x x
In order to develop the new fabrication procedures, several crystals of Pb,_ Sn Te were
grown from the vapor phase by two techniques: (1) a closed-tube vertical process, and (2) a
recently developed closed-tube horizontal process. In each case the source consisted of a
metal-rich, two-phase ingot of (Pb. _ Sn )._ Te , with y « 0.49. Using a crystal growth tem-
i x x i y y
perature of approximately 800°C, most of the as-grown and air-cooled crystals were found to
be n-type at room temperature for x < 0.08, and p-type for x> 0.08. For 0.08 < x < 0.20, an
n-type surface layer is formed during the air-cooling process, producing p-n junctions for diode
lasers operating in the 9- to 16-|o.m range. In order to obtain lasers for the SO, absorption
IL,
measurements around 8.7(j.m, x is approximately 0.07. Using the above procedure, the as-
18 - 3
grown crystals are n-type with carrier concentrations of about 10 cm .
-------
The best PbQ 8gSnQ 12Te diode lasers in the 10-|im region were made from crystals which
were vapor-grown, annealed, and diffused (p-n junction formation) without ever having to open
the growth ampoule. It is believed that elimination of the handling of these crystals by this
procedure maintains the cleanest natural crystal surfaces possible, and is largely responsible
for their superior performance over diodes prepared by conventional techniques. This proce-
dure requires that the as-grown crystals are p-type, which, as mentioned above, is not the
case for those where x - 0.07, where the solidus field shifts toward the metal-rich side of stoi-
chiometry. In order to counter this effect, thallium metal was added to the two-phase ingot
serving as the source for the vapor responsible for the crystal growth. (Thallium is a p-type
impurity in PbSnTe, and therefore serves to compensate the donors produced by the shift in
the solidus field.) In the vicinity of the p-n crossover temperature, the donor or defect con-
centration for metal-saturated crystals increases as temperature decreases. Consequently,
by carefully controlling the thallium concentration it is possible to grow crystals, for which
x - 0.07, at 800°C having p-type bulk conductivity, with an n-type skin formed on cooling.
Table II lists properties of the continuous-wave (CW), cur rent-tunable PbSnTe diode lasers
developed and used during the first year of this program. Some of these devices were contin-
-1
uously tunable over more than 1 cm and produced some of the widest SCX scans shown in
Appendix A. The single-mode output power from these devices is not yet sufficient for optimum
heterodyne detection, as discussed in Section 4; the crystal growth parameters and laser fabri-
cation techniques are being systematically studied in order to increase substantially the output
power, so that heterodyne experiments using these lasers as tunable local oscillators will be
feasible.
TABLE II
PROPERTIES OF CW
DIODE LASERS AT 10°K
Threshold Current
Diode
284
286
301
302
315
317
326
328
329
338
x
0.0700
0.0700
0.0700
0.0700
0.0700
0.0700
0.0575
0.0575
0.0575
0.0745
Density (A/cm )
360 (1.05A)
400 (0.75A)
490 (0.91A)
495 (0.91 A)
714 (1.52A)
930 (1.80A)
332 (0.60A)
211 (0.34A)
162 (0.23A)
241 (0.60A)
Maximum CW
Output Power (ja.W)
at I (amps)
20 (2.2A)
22 (2.0A)
6 (2.0A)
21 (2.0A)
44 (2.4A)
44 (2.0A)
16 (0.9A)
4 (2.0A)
18 (0.8A)
12 (l.OA)
Wavenumber
at Threshold
(cm )
1142
1160
1128
1132
1150
1148
1179
1200
1206
1104
-------
3. LASER SPECTROSCOPY OF SULFUR DIOXIDE
Three major absorption bands of SO2 in the 2- to 15-jim region are shown in the grating
spectrometer scans of Fig. 3-1, adapted from Pierson, Fletcher, and St. Clair Gantz. Centered
at 8.7 urn is the v . vibration-rotation band, at 7.3 urn is the v band, and at 4^m is the v + v
r- ^ j ->
combination band. At the start of this program there were no tunable semiconductor lasers in
the 4 fim region. Such devices have since been developed by Nill, Blum, Calawa, and Ha r man,
and by Melngailis and Harman.8 Although the v3 band is somewhat stronger than the v^ band,
it is in a region of strong water vapor absorption, which may produce interference effects during
atmospheric transmission over long paths. A secondary reason for selecting the longer wave-
length is that it provides greater sensitivity for remote heterodyne detection, as will be illus-
trated in Section 4.
Figure 3-2 is a grating spectrometer scan of the v±, 8.7-jjim band of SO2, as recorded by
Shelton, Nielsen, and Fletcher;9 it represents most of the 1100- to 1200-cm"1 region of interest
for the present study. There are several thousand lines in this band which overlap in varying
eo
60
40
20
0
4000
25
00
IV
90S
"*=
1
WAVENUMBER (cm"1)
500 1300 1200 1100 1000
-^
1
|
n\
\\
'
I
B
!J
ft
'
I
-A
900
850 800
|-5-3616|
750 700
' A 704 Torr
B 68 Torr
C 20 Torr
(L= 10cm)
23 4 5 6 7 8 9 10 11 12 13 14 15
WAVELENGTH (urn)
Fig. 3-1. Grating spectrometer scan of SO2 absorption at room temperature, between
2andl5fj.m, from Pierson, Fletcher, and St. Clair Gantz.7 The curves labeled A, B,
and C represent gas pressures of 704, 68, and 20 Torr, and slit-width settings of
0.927, 0.900, and 0.900mm, respectively, for the Perkin-Elmer Model 21 infrared
spectrophotometer used. Cell length (L) is 10cm.
1150
WAVENUMBER (cm-
Fig. 3-2. Grating spectrometer scan of the v^ vibration-rotation band of SO£
at room temperature, after Shelton, Nielsen, and Fletcher^ Effective path
length (L) is 20 meters, and instrumental resolution 0.3cm . Although not
explicitly stated by the authors, SO? pressure appears to be approximately
ITorr.
-------
EXPERIMENT
(from dlodo loser scon)
-5-3610-lj
\J
THEORY
1147.1
WAVENUMBER (cm'')
Fig. 3-3. Comparison between measured and calculated absorption coefficients
and positions for some lines in the v, band of SC^. These scans were computer-
generated using Lorentzian profiles. Lines designated as a, b, and c represent
the transitions (29,6,24)(28,7,21), (35,7,29)(34,8,26), and (26,4,22)(27, 3,25),
respectively. Line d is not predicted by the present theoretical model.
CELL EVACUATED
939.18 939.20 939.22
WAVENUMBER (cm'1)
Fig. 3-4. Absorption line profile of aQ(9,3) line of NH3, at 2 Torr pressure,
with different partial pressures of air, from zero to 1 atmosphere (760 Torr).
L = 30cm, T = 296°K (23°C).
-------
degrees at atmospheric pressure to produce a spectrum quite similar to that of Fig. 3-2 (limited
by the resolution of the spectrometer). These lines can be revealed individually at reduced
pressure using the ultra-high-resolution capability of tunable semiconductor diode lasers, with
demonstrated linewidths of 3 X 10 cm or less.
The SO? was obtained from the Matheson Company, Inc. of East Rutherford, New Jersey,
and had a quoted purity of 99.98 percent. Using the tunable PbSnTe lasers described in Sec-
tion 2, several hundred of the SO, absorption lines were measured in various segments of the
v band. Wavelength calibration was obtained by placing NH in the cell after each SO scan.
11
The NH, lines have been precisely measured and are normally used as wavelength calibra-
tion standards in this region of the infrared. The length of the gas cell was 30 cm, and a spec-
trometer using a Ge:Cu detector was used for approximate determination of laser wavelength
and to limit the detected signal to a single laser mode. Some of-the actual laser scans are
shown in Appendix A, and in Appendix B the fundamental absorption constants (per unit pres-
sure) of these lines are tabulated. For reference, the NH, calibration lines used are listed in
Appendix C.
A theoretical spectrum for the v. band of SO, has been calculated by S. A. Clough using
14
the method described in Ref. 13 based on accurate microwave values of the rotation constants
for both the ground and first excited state of this vibration-rotation mode. When the experi-
mental and theoretical values for the absorption constants are compared, as in Fig. 3-3 for the
1147-cm region, the SO., transitions can be identified. For example, the large absorption
-1
line at 1146.966cm is produced by the transition (14,1,13)-*-(14,2,12), where the three quan-
tum numbers are (J, K., K,,,). On the basis of these experiments, the value of the frequency
-1-1
of the v , band center was found to be 1151.71 cm ± 0.01 cm , which can be compared with
-1
the experimental value previously measured of 1151.38 cm (Ref. 9). Lines a, b, and c are
shifted slightly from their predicted positions, and are described more fully in Appendix E.
All the lines predicted by theory have been observed experimentally; however, a few lines which
have been observed are not predicted, such as line d in Fig. 3-3, which may represent a "hot"
band of SO., (involving the first excited vibrational state of the bending (v ) mode, which was
not included in the theory), another isotope, or a weaker transition not included in the theory.
There is always the chance that these extraneous lines may represent impurities in the SO,,
£
but their infrequent occurrence makes them relatively unimportant at present.
The above theoretical model for the v^ band of SO, now permits us to predict the structure
of either absorption or emission spectra under any conditions of gas temperature, pressure, or
instrumental resolution. This knowledge is essential for any spectroscopic technique of mon-
itoring SO,, and is presently being used to determine the optimum wavelength for detection by
infrared absorption at atmospheric pressure and elevated temperatures. A similar theoretical
model has already been established for ozone (O,) in the 9.5-(im region; ultimately, it would
be very useful to have a complete theoretical model of a polluted atmosphere.
The rates of pressure broadening of the SO, lines were measured and compared with those
for NH3 in the 10-|j.m region. The effect of increased air pressure on the shape of an NH, ab-
sorption line is shown in Fig. 3-4, with the corresponding effect on the absorption coefficient
illustrated by Fig. 3-5 for both NH,:air (where the absorption coefficient decreases linearly
with increasing pressure above about lOTorr) and NH,:NH, (where the absorption coefficient
reaches a steady value in the pressure-broadened regime). The spectral lines were fitted with
-------
1
o
c
0.001
-
-
X5
E /
. <
°
i 0 °
>\
"\
\
i iiii.ii
|l8-5-3?49-?|
Pt - PfllR
(PNH = 2 Torr)
^0 8
\ |
\P °.
\ »
\
I 1 1 1 1 !*>!
1 10 100
PARTIAL PRESSURE P, (Torr)
Fig. 3-5. Pressure dependence of absorption coefficient of aQ(9,3)
line for both self- and air-broadening.
-------
Fig. 3-7. Collision-broadening (FWHM) of
two SO2 lines at 296°K: (15,4,12) (15,3,13)
at 1163.297cm"1, and (8,0,8)(8,1,7) at
1148.810cm"1. Doppler-limited linewidth
(FWHM) is 53 MHz (0.0018cm"1).
SELF-BROADENED T/
(40MHz/Torr) r
'AIR-BROADENED
(l2MHz/Torr)
10 100
PARTIAL PRESSURE (Torr)
theoretical Lorentzian and Gaussian curves (where appropriate) in order to accurately deter-
mine the effective linewidths at the various pressures. The full linewidth at half-maximum in-
tensity (FWHM) for NH3 is shown in Fig. 3-6, and for SO, in Fig. 3-7. With air-broadened
FWHM values of 6.3 and 12MHz/Torr for NH, and SO,, their respective linewidths at atmos-
-1 -1
pheric pressure are 0.16 cm and 0.30cm . Some workers refer to the half-width at half-
-1 -1
maximum intensity (HWHM), which would be 0.08 cm and 0.15 cm at atmospheric pressure
for NH, and SO,, respectively.
4. REMOTE HETERODYNE DETECTION
Heterodyne detection of characteristic infrared emission from atmospheric pollutant gases
may represent the ultimate remote-sensing technique for source monitoring. Its operation is
monostatic (single-ended), passive, and essentially range-invariant. Heterodyne detection in-
volves the same emission lines monitored by passive incoherent infrared systems, but without
restrictions such as loss of resolution due to instrumental broadening of the lines (reduced
specificity) and non-ideal detector-noise-limited operation (reduced sensitivity). Some coin-
cidences between fixed-frequency gas laser wavelengths and those of atmospheric-pressure-
broadened pollutant emission lines suggest their application as local oscillators (L. O.) in the
heterodyne scheme. It is clear, however, that the ability to "tune" the L. O. wavelength ideally
to the center of the emission line as well as to a region where there is no characteristic radia-
tion is important for the elimination of background effects, and thus provides better ultimate
sensitivity. Fixed-frequency infrared gas lasers cannot normally be tuned more than the 50-
to 100-MHz (1.8 - 3.6 X 10 cm ) Doppler width, whereas the required shift for gases in the
atmosphere is at least 10,000MHz (0.33cm ), which can easily be achieved with tunable semi-
conductor diode lasers.
Calculations were made of the sensitivity limits for remote heterodyne detection of several
important pollutant gases at different temperatures. The instrumentation, theoretical analysis,
and numerical predictions are contained in Section 4.1. Quantitative experimental verification
of the basic heterodyne theory has been achieved for radiation from a thermal source (analogous
to an optically thick gas at line center) and is described in Section 4.2.
At the present time, tunable diode lasers, even with single mode power levels of tens
of milliwatts, cannot provide sufficient L. O. power for optimum heterodyne detection using
-------
state-of-the-art low-noise amplifiers and infrared detectors. In Section 4.3 this will be dis-
cussed in detail. In order to circumvent the tunable L. O. problem temporarily and permit the
detection of actual emission lines from a heated gas, a fixed-frequency CO_ laser was used as
the L. O. in an experiment to detect heated ethylene (C-jH^) gas at low pressure. Although a
tunable L. O. is usually necessary to obtain spectral information, electronic scanning of the
wideband (500-MHz detector/amplifier) signal with a narrowband (34-MHz) filter simulated the
ultimate system. This experiment is described in Section 4.4.
4.1 Heterodyne Theory and Calculations
Figure 4-1 illustrates schematically the operation of remote heterodyne detection. The
telescope receiving optics collects characteristic radiation from the gases in the stack plume
and images it onto the infrared detector. By means of a beam splitter, tunable laser radiation
also impinges on the detector, and power at the difference (heterodyne) frequency is measured
electronically. If the solid angle subtended by the source at the detector is larger than the
diffraction-limited solid angular resolution of the collection optics - a situation usually valid
for smokestack plumes at ranges up to 1 km and with collection optics as small as 1 cm - the
signal-to-noise voltage ratio (after square-law detection) is given by
S/N =
-e
-a' cL
c
\/ 1
' \eh"/kTs _ 4
(4-1)
where I represents the fraction of emission energy absorbed by the detector, 7) is the detector
quantum efficiency, a' is the absorption constant of the particular spectral line per ppm, c is
the actual concentration in ppm, B is the intermediate-frequency (IF) bandwidth of the detection
system, r the post-detection integration time, hv the infrared photon energy, L the thickness
of the plume and T its temperature, and e, is the emissivity of the background at temperature
S D
T^. The following assumptions have been made: (a) that the IF bandwidth is less than the emis-
sion linewidth; (b) that the emission and absorption due to the pollutant in the ambient atmos-
phere are not important because of its low concentration in the atmosphere relative to that in
the plume; (c) that the background attenuation from the wings of other molecular absorption
lines is negligible; (d) that the local oscillator has sufficient power to overcome the other sources
of noise (this will be discussed in detail in Section 4.3); and (e) that a photovoltaic detector is
being used. (If a photoconductive detector is used, S/N is reduced by a factor of two.)
Fig. 4-1. Configuration for remote,
single-ended, heterodyne detection
of pollutant gases from emission
sources.
-BEAM-SPLITTER
10
-------
Since e, is often much smaller than unity (about 0.1 in the 10-jj.m region of the infrared
when there are no clouds in the background), the source does not have to be hotter than the
background; in fact, it could be colder, as will be demonstrated. Using Eq. (4-1) for the pol-
lutant concentration c in ppm, we obtain the following expression:
c = -
a' L
c
In 1 -
(S/N)
1/2
<1+PLO/PLO>
(4-2)
where £ is the bracketed Bose-Einstein term in Eq. (4-1), PLO is the L. O. power, P^O is the
L. O. power required to generate noise in the detector which just equals that produced by other
sources [in assumption (d) above, the L. O. power swamped the other noise contributions]. If
we further assume that PLQ = P£Q, that S/N = 4- and that »J = £ = 1 (corresponding to an ideal
photovoltaic detector and efficient collection optics), the above equation reduces to
c =
a' L
c
In 1 -
(4-3)
Diode laser scans of the spectral lines of CO, NO, NH,, C,H., and SF, indicate approxi-
C _ 4
mately equal strengths for the strongest lines of these gases of a' = 10~ cm /ppm at atmos-
pheric pressure. (The effective strength of SO2 at 8.8jim is an order of magnitude smaller.)
If we let B = 500MHz, r = 10 sec, L = 10 meters (reasonable for a power-plant stack diameter),
T^ = 293°K, and e^ = 0.2, the minimum detectable concentrations at different temperatures are
those shown in Fig. 4-2 as a function of wavelength. Note that sensitivity is greater at longer
B'SOOMHz
T= 10 sec
L« 10m
dc = 1 xlO cm /ppm
S/N =4
P,
II
WAVELENGTH (pm)
Fig. 4-2. Calculated wavelength dependence for pollutant monitoring using
the heterodyne technique, for various gas temperatures, and background
emissivities of 0.2 and 1.0. Useful wavelengths for some pollutant gases
are indicated on the abscissa. Other parameters are indicated in the figure.
11
-------
wavelengths, which is a major consideration to be made in selecting optimum emission lines
for heterodyne detection. Note also that, even at temperatures as low as 50°C, 2 ppm of C^H.
and NH, can be detected at 10.6p.m. The wavelength dependence of sensitivity is particularly
noticeable at low gas temperatures. For example, for a gas at 50° C it will not be possible to
obtain any useful sensitivity with this technique at wavelengths shorter than 5 jim; however, at
higher temperatures, even CO and NO in the effluent of stationary or mobile sources can be de-
tected adequately.
Although SO, emits strongly around 4jj.m, in addition to the 8.8-|j.m band indicated in
L*
Fig. 4-2, the sensitivity (for lines of equal strength) at the latter wavelength is one or two orders
of magnitude better, depending on the gas temperature. In order to determine the sensitivity
for remote heterodyne detection of SO, in the 8.8-(im region, we use the parameters assumed
AH
above, but substitute the proper value of «' = 10 cm /ppm which we have measured for SO,
C £
at atmospheric pressure. The theoretical curves of Fig. 4-3 illustrate how the ultimate sen-
sitivity depends on temperature and background emissivity. An increase in the L. O. power be-
yond the value Po Pr°duces at best, in accordance with Eq. 4-2, only a factor-of-two improve-
ment in ultimate sensitivity. (This is one reason why PT
= P
is a good design parameter
-t -l .
to use.) It should also be noted that the minimum detectable concentration ranges from 2 to
50 ppm for SO2 gas temperatures from 300 to 50°C, respectively.
B = 500MHz
T=IOsec
L=10m
a^- 1 x 10"6 cnrT'/ppm (S02)
S/N"4
P -p°
PLO PLO
X'S.ey^mlSOj)
Tb = 293°K (20"C)
eh= O.Sand 1
cb=0.2
Fig. 4-3. Theoretical temperature
dependence for heterodyne detection
of SO.,, for a signal-to-(rms)-noise
ratio of 4, and background emissiv-
ities of 0.2 and 1.0. Other param-
eters are indicated in the figure.
100 200
GAS TEMPERATURE (°CI
4.2 Heterodyne Detection of Thermal Radiation - Test of Theory
In order to show G-R noise-limited operation, the experimental setup of Fig. 4-4 was
used, which involved heterodyne detection of infrared radiation in the lO-jim region from a
globar at about 1500°C (1773°K), using a CO, laser as fixed-frequency local oscillator.
12
-------
|l8-5-33?3-l|
SPECTROMETER
C02 LASER
CHOPPER
BEAMSPLITTER
DETECTOR
LASER BLOCKED
(ILO=0)
GLOBAR BLOCKED
(ILO = 40mA)
|-5-33?6-ll
100
TIME (sec)
Fig. 4-4. Experimental setup for in-
laboratory detection of thermal radia-
tion using the infrared heterodyne
technique. The lenses are BaF2, and
the beam-splitter is coated Irtran IV.
Fig. 4-5. Heterodyne signal between CO2
laser and thermal source at a temperature
of 1500°C. System bandwidth is 450MHz,
and post-detection integration time 0.1 sec.
Origin of the time axis is arbitrary.
Realizing that gas emission lines at atmospheric pressure are several GHz wide (~5GHz
for NH3 at 10 \un, ~9 GHz for SO2 at 8.8(im), it is important to use the widest IF bandwidth
possible consistent with a low amplifier noise figure and the frequency limitations of the in-
frared detector. At present, 450 MHz is the best value, with available amplifiers having noise
figures in the 2.4- to 4.0-dB range. This experiment was performed with an amplifier having
an average noise figure of 4.0dB over 450 MHz, yielding an equivalent noise temperature of
TA = 438°K. At the 10.6-(im wavelength of the CO2 laser, hv/e = 0.12V. The infrared detector
is a liquid-helium-cooled GerCu photoconductor, with r? = 0.5 and a gain G of 0.07 at an applied
bias of 22.5V. The load resistance RT was 25 ohms.
LJ
In Fig. 4-5 is shown an actual recorder trace of the heterodyne signal between the globar
source and the CO2 laser, for which the local oscillator power is greater than that needed to
overcome amplifier noise. The signal vanishes when either the CO laser or the globar is
blocked, and in the latter case the observed G-R noise is within a factor of two of the predicted
value for the 40-mA detector L. O. current. When the laser itself is blocked, the small residual
noise is produced by the amplifier. Since the G-R noise has overcome the amplifier noise, the
ultimate heterodyne signal-to-noise is being achieved, subject to optical transmission losses
and other losses present in the globar light path and other factors to be discussed later.
Quantitative verification of Eq. (4-1) was obtained by replacing the globar with a calibrated
600°C thermal source and comparing the predicted and observed signal-to-noise ratio. Using
a 2.4-dB (noise-figure) amplifier and a detector bias of 40V, G-R noise-limited operation was
again established. The transmission of the Irtran II beam splitter (0.5), BaF? lens (0.85),
13
-------
BaF dewar window (0.85), and detector front surface (0.64) produced a value for the transmis-
sion factor '| of 0.23. If we let Tg = 872° K (600° C), Tfe = 292° K (20° C), efe = i, B = 450 MHz,
and r = 0.3 sec, Eq. (4-1) becomes
S_
N
(0.23) (0.5)
(0.292 - 0.012) V450 X 10 X 0. 3 = 187
The measured ratio of signal voltage to rms noise voltage (output voltage is proportional to in-
put power because of the rf detector) is 52, implying a mixing efficiency of 28 percent. (By
comparison, we have observed heterodyne mixing efficiencies of 12 percent for gas laser/diode
laser experiments, and as high as 75 percent for diode laser/diode laser beating experiments,
so that the result appears to show reasonably good agreement with theory.) It should ultimately
be possible to approach 100 percent efficiency by proper optical phase matching.
There are several steps which can be taken to enhance the signal-to-noise ratio from its
predicted value of 187 for the above experiment. Such improvement centers on reducing the
optical transmission losses and use of a different type of infrared detector. This will be dis-
cussed in Section 4.3.
Since reasonable agreement between theory and experiment was obtained for both the noise
voltage and the signal-to-noise ratio at a fixed temperature, it remained to verify the tempera-
ture dependence of Eq. (4-1). Since PN is not a function of the source temperature Tg, there
should be a linear dependence between the square-law-detected signal voltage (proportional to
P ) and the bracketed term y, defined earlier. This dependence is followed, as shown in
Fig. 4-6 for source temperatures between 700 and 1200°K (428 to 928°C). Although not shown
here, heterodyne detection was observed for source temperatures as low as 200°C using a
4-dB (N. F.) amplifier.
TBB(°K)
eoo 900 iQoo HOP 1200
8-5-3?75-?l
Fig. 4-6. Heterodyne signal voltage as
a function of Planck function for several
source temperatures from 700 to 1200°K
(973-1473°C). CO2 laser wavelength
is 10.64 jim, and system bandwidth is
450MHz.
'-[exp(h»/kT300.K)-l]
14
-------
4.3 Local Oscillator Power Requirement
The local oscillator power necessary to produce a noise level equal to that of the pream-
plifier is defined as follows:
(kT./e) (hv/e)
where T. is the noise temperature of the amplifier, G the infrared detector gain, R, the load
resistance, and e the electronic charge. For a 2.4-dB amplifier (T. = 240°K), TJ =0.5,
G = 0.12, RT =50 ohms, and A = c/v = lOiim,
l~i
o (0.018)(012) w = 12mw .
L'U (0.5) (O.lZr (25)
This calculation illustrates why the Pb. _ Sn Te diode laser power (we have measured 1 mW in
1 ~X X
a single laser mode for the best case) is not yet sufficient to achieve ultimate sensitivity in re-
mote heterodyne detection using present amplifiers and infrared detectors.
In heterodyne detection the L. O. power can, in principle, be increased sufficiently for
G-R noise to swamp other noise components. Consequently, infrared detectors with very high
detectivities are not required, and there can be definite advantages to replacing the photocon-
ductor with a photodiode. A photodiode does not sacrifice gain in order to obtain high frequency
response; the gain, in fact, is usually assumed to be unity, and the quantum efficiency J? char-
acterizes the performance of the detector. Although presently available wideband photodiodes
have quantum efficiencies of not more than 0.2, improvements are continually being made. If
we assume f) = 1, an inherent detector gain of unity, and neglect optical losses, Eq. (4-4) yields
PLO = 86 jiW, which has been obtained in a single mode for tunable semiconductor diode lasers
in this wavelength region. Thus, improvements in both areas of technology higher tunable
laser power and more efficient photodiodes would advance the achievement of remote hetero-
dyne detection.
4.4 Heterodyne Detection of Ethylene
In Fig. 4-7 is shown a heterodyne-calibrated absorption spectrum of SF, around the P(14)
CO? laser line, as well as a similar scan of C-H, at room temperature and lOOmTorr pressure.
Within 450MHz (the bandwidth of our system) of the P(14) line, there are two distinct C,H. ab-
Lf 4
sorption lines. These become characteristic emission lines when focused onto a cold detector,
as in Fig. 4-8, and the emitted signal increases according to Eq. (4-1) as the gas temperature
is raised.
The experimental schematic of Fig. 4-9 was needed in order to obtain spectral information
during the heterodyne detection experiment, since a fixed-frequency laser not a tunable laser -
was used as the local oscillator. The gas and laser beams are combined at the detector D.
The load resistance R of 50 ohms terminates the line to the amplifier (reducing electrical re-
flections between detector and amplifier), and the capacitor C blocks the DC bias from this
path. The signal is fed into a low-noise (2.5dB), wideband (450MHz) amplifier, then to a mixer
for combination with a sweep generator signal (0 to 450 MHz). The narrowband amplifier at the
output of the mixer passes only that electrical information within 17 MHz of the sweep generator
frequency. This signal is then rectified by the RF detector into a steady voltage (except for the
15
-------
-1200 -800 -400 0 400 800 1200
FREQUENCY RELATIVE TO P (14) C02 LASER LINE (MHz)
Fig. 4-7. Heterodyne-calibrated scan of SFz (upper) at 30mTorr pressure,
referenced to P(14) CO2 laser line. The lower scan represents C^K^. at
lOOmTorr pressure over the same tuning range. L = 30cm, T = 296°K.
SPECTROMETER
MIRROR
CHOPPER
BEAM SPLITTER
Fig. 4-8. Experimental configuration
for heterodyne detection of infrared
emission lines. Windows on the
cell are made of BaF.,.
DETECTOR
GAS CELL
6-5-33?4-l|
16
-------
LASER
Fig. 4-9. Schematic for heterodyne emission spectroscopy with fixed-frequency
laser local oscillator.
modulation caused by the mechanical chopper in the beam from the gas cell), and then synchro-
nously detected by the lock-in amplifier and displayed on an X-Y recorder. The resulting signal
is shown in Fig. 4-10 for C,H. pressures of 0, i, and 3Torr at a temperature of 600°C in a
1 -meter-long cell. Although the signal-to-noise ratios for these scans do not represent the
ultimate values (caused in part by absorption by the C2H4 near the relatively cool cell window),
and a time constant of 30 seconds was required, this experiment represents the first remote
heterodyne detection of thermal emission lines. When improved infrared detectors in conjunc-
tion with stronger tunable lasers permit super-heterodyne techniques to be used, the narrow-
band amplifier and mixer of Fig. 4-9 will become unnecessary, and further improvement in the
signal-to-noise ratio by signal-averaging of repeated laser scans through the emission line will
be possible.
|-5-33?7-l|
3 TORR
200 300 400 500
FREQUENCY RELATIVE TO P(14) C02 LASER LINE (MHz)
Fig. 4-10. Heterodyne detection of emission from C^R^. at 600°C, using
the P(14) CC>2 laser line as fixed-frequency local oscillator. System
bandwidth is 450 MHz, effective scanning bandwidth is 17 MHz. The only
difference between curves is the indicated change in pressure of the gas.
17
-------
5. EXPERIMENTS RELATED TO MONITORING APPLICATIONS
OF TUNABLE LASERS
Several experiments were performed to provide some insight into the use of tunable lasers -
particularly tunable diode lasers - in monitoring applications. In Section 5.1 a sensitivity meas-
urement for point-sampling is described, involving a tunable diode laser and infrared detector.
Section 5.2 considers effects produced on the SO., absorption spectrum by changes in gas tem-
perature and increases in pressure to one atmosphere. In another experiment, no interference
was produced by a sample of polluted air at atmospheric pressure for the wavelength range used.
In Section 5.3 a long-path atmospheric transmission experiment of diode laser radiation is
described.
5.1 Point Sampling Sensitivity Measurement
We have shown earlier that in point-sampling applications it is possible to eliminate (in
most cases) any interference effects between different gaseous components of a specimen to be
analyzed by reducing the net gas pressure to the point where the characteristic absorption
spectra are revealed and overlap caused by pressure-broadening is negligible. In the experi-
ment to be described, the diode laser frequency was modulated at 500 Hz by a small (~i mA)
current superimposed on the steady diode current. When the diode laser emission scans an
absorption line, a first derivative of the absorption line profile is produced if the synchronous
amplifier is tuned to 500 Hz (this procedure was used in Ref. i to detect C^H.^ in automobile ex-
haust). If a reference signal of 1 kHz is applied to the synchronous amplifier, and the amplifier
set to that frequency, the second derivative of the absorption line profile is produced. Results
of such an experiment, involving the detection of an NH, line of strength 1 cm /Torr, are
shown in Fig. 5-1. Trace A represents the second derivative signal produced in a 10-cm-long
A NH3 (lOmTorr)
B lOmTorr NH3: 4.5 Torr AIR 80uV
C DILUTED NH3 :AIR (4.5Torr)
Fig. 5-1. Point-sampling sensitivity measurement using a 10-cm-long cell and Ge:Cu
infrared detector. (A) Second-derivative trace of NH3 line at lOmTorr pressure;
(B) Second-derivative trace of the same line when 4. 5 Torr air is added; (C) Second-
derivative trace of (B) after sequential dilutions with air until the signal at 4.5-Torr
pressure was comparable to the detector noise level, which represents approximately
28ppm peak-to-peak. Curve (D) is a first-derivative trace of (C), illustrating long-
term fluctuation effects not present in the second-derivative scans shown above.
18
-------
absorption cell by NH3 at 10-mTorr pressure; the peak-to-peak voltage is 100 jiV. Trace B
shows the effect produced by the addition of 4.5 Torr of air: slight broadening of the signal (see
also Fig. 3-6) with corresponding reduction in signal to about 80p.V. In an. attempt to determine
the minimum detectable concentration of NH3 using this point-samp ling technique, several dilu-
tions of the gas were made until Trace C was obtained, showing a peak-to-peak noise of ap-
proximately i \iV. Since 80 (iV represented a partial NH, pressure of lOmTorr, and the signal
is directly proportional to gas concentration in this region, 1 p.V corresponds to 0.125 mTorr
NH3, or 28ppm in 4.5 Torr of air. Since the noise level represents detector noise which is not
caused by the laser, it is possible to increase sensitivity by using a laser having more than the
10-p.W power of the device used for Fig. 5-1. (At some point, however, the detector noise
would increase also, due to photon noise, as the square root of the laser power; however, the
signal would increase linearly, resulting in still better sensitivity, but at a slower rate of im-
provement.) Trace D is the first derivative signal, exhibiting nearly the same signal-to-noise
as that of Trace C under the same conditions, but with the added complication of a slow change
in signal strength over the width of the scan. This signal variation probably represents the
change in laser power over this region.
Further improvement in point-sampling sensitivity can be obtained by increasing the path
* 18
length with a multireflection "white" cell. During the past year, a "white" cell with an ef-
fective path length of 730cm was constructed in accordance with the design suggestions of
P. L. Hanst. The 8-pass unit consists of two spherical mirrors, each with a focal length of
45.7cm, placed 91.4cm apart. Two partial quadrants were removed from one of the mirrors
to make room for the entrance and exit beams. The other mirror was cut into semicircular
halves to permit beam direction within the cell. When the beam from a tunable diode laser was
directed through the multipass cell, the first-derivative signal obtained by scanning through an
NH3 absorption line was 73 times greater than that obtained while scanning the same line using
a 10-cm-long cell, as expected. Since the detector noise level remains the same, we conclude
that the use of this "white" cell makes it possible to detect concentrations 73 times smaller
than the noise level of Fig. 5-1 (28ppm), or a few hundred ppb using a 10-(j.W diode laser.
5.2 Pressure, Temperature, and Interference Effects
In certain applications of infrared spectroscopy to pollutant monitoring, such as long-path
absorption or emission in the atmosphere, it is not possible to control the experimental condi-
tions as well as can be done in point sampling. The work of Burch, Pembrook, and Gryvnak,
19
using a high-resolution grating spectrometer, showed that for temperatures between 23°C and
300°C the net absorption strength of the v. SO, band is constant, although there are changes in
the relative strengths of individual lines within the band. They further show that the charac-
teristic spectrum of SO, in the atmosphere is essentially independent of temperature in the
-1
1127- and 1180-cm regions of the v. vibration-rotation band. A study was made of the tem-
perature dependence of isolated lines at reduced (1-Torr) pressure in order to check the the-
oretical prediction and thus confirm the ability of the model of Section 3 to predict spectra of
SO, under varying atmospheric conditions. The temperature dependence of the absorption co-
efficient per unit pressure (in a closed-off cell where the number of molecules remains constant)
is shown in Fig. 5-2 for three lines in the 1117-cm region, for a temperature range from 23
to 200°C. According to theory, the temperature dependence of the absorption constant of a
single line is determined by three parameters: (a) the thermal population (or Boltzmann)
19
-------
function exp[W/kT], where W is the internal energy of the lower state, k is the Boltzmann
constant, and T the absolute temperature; (b) the partition function, which varies approximately
as T ' ; (c) the linewidth, which varies as T ' for pure Doppler broadening, but is more
complicated in the 1-Torr region (where the data were taken) because of pressure-broadening
*
effects. If we assume that pressure (hence, pressure broadening) increases linearly with in-
creasing temperature, and that the Doppler- and pressure-broadened linewidth contributions
add quadratically, then the solid curves of Fig. 5-2 represent the best fit to the data for the
(18,6,12) *- (19,7,13) and (13,7,7) «- (14,8,6) lines when an arbitrary scale is used. In order to ob-
tain the fit to the points denoting the (8,8,0) *- (9,9,1) transition shown by the dashed line, the scale
factor for this curve had to be increased by 15 percent over that used for the other two curves.
TEMPERATURE (°K)
300 350 400 450 500
TRANSITION
D (8,8,0119,9,1) 1117.177
O (13,7,7 ) (14,8,6) 1117.128
A (18,6,121(19,7,13) 1117.038
0 50 100 150 200
TEMPERATURE (°C)
Fig. 5-2. Temperature dependence of the absorption coefficient of three
SO2 lines (identified in the legend) from 23 to 200°C. Gas pressure at
room temperature was 1 Torr in each case. The solid lines represent
theoretical fits to the two lower transitions; the dashed line is a the-
oretical fit to the third line when the vertical scale factor is increased
by 15 percent (see text).
20
-------
19
1165.0 1165.2
WAVENUMBER (cm-')
_ A
Fig. 5-3. Diode laser scans of the 1164-cm region of SO2 absorption for gas
pressures of from 0.25 to 32 Torr. L = 30cm, T = 296°K. (Zero-transmission
lines for the different pressures were shifted vertically for clarity.)
The effect produced on spectral resolution by increasing SO, pressure is shown in Fig. 5-3
for pressures from 0.25 to 32 Torr. At pressures above 2 Torr, the highest resolution begins
to be lost, although recognizable structure is apparent even for 16 Torr pressure. It should be
noted, however, that at 16 Torr the atmospheric concentration of SO- would be over 20,000ppm,
which would probably not be encountered in the usual monitoring applications. The self-
broadening of SO., lines can usually be neglected a conclusion also arrived at by Burch, et al.,
on the basis of their measurements.
Since air broadening of spectral absorption lines is usually the dominant mechanism af-
fecting resolution and sensitivity of spectroscopic techniques, it is illustrative to consider the
effect on a single absorption line of increasing air pressure to 1 atmosphere (760 Torr). An
experiment was performed in which the relative concentration of NH, in air was maintained at
lOOOppm, and the absorption coefficient measured as a function of air pressure, as shown in
Fig. 5-4. At pressures below 10 Torr, the absorption coefficient increases linearly with pres-
sure, corresponding to the larger number of NH, molecules in the absorption cell in direct
proportion to the number of air molecules (to maintain 1000 ppm). At air pressures above this
value, the NH, line broadens in direct proportion to the pressure at the rate of 6.3MHz/Torr
shown in Section 3. In the pressure-broadened regime, the absorption coefficient at a frequency
v away from the line center v may be approximated by the Lorentzian shape:
a(v,
(5-1)
where c^ is proportional to the fundamental line strength and Av is the full-width of the line at
half-maximum (FWHM). In the i
which yields, at the line center,
half-maximum (FWHM). In the air-broadened case, Av = c,p . (where c, = 6.3 MHz/Torr),
Ci 3.11*
c NH3
p
2 air
(5-2)
21
-------
ALTITUDE (km)
PNH /pnir 10 (1000ppm)
3 /
/
/
/o
1000
AIR PRESSURE (Torr)
Fig. 5-4. Measured variation in the absorption coefficient of a single NH3 line with air
pressure, maintaining a fixed relative concentration of 10~3 (lOOOppm). Also indicated
is a scale of altitude above sea level for corresponding values of air pressure.
Thus the absorption coefficient is constant for a fixed relative pollutant concentration. Also
indicated in Fig. 5-4 is a scale relating the air pressure to altitude above the Earth's surface,
showing the onset of full pressure-broadening effects at altitudes below 30km.
The effect of air broadening on SO, lines around 1118cm was also studied with tunable
L,
diode laser scans. At atmospheric pressure the average absorption in this region was 30 per-
cent for 20 Torr SO, (in 1 atmosphere of air) in a 30-cm-long cell. This produced a value for
7 4
net line strength per unit concentration of SO, of 4.3 X 10 cm /ppm, which is in good agree-
19
ment with that value which can be derived from the grating spectrometer data. In the some-
what stronger 1127-cm region, the strength is expected to be approximately 1X10 cm /ppm.
In an attempt to uncover any possible interferences between SO, absorption lines in the
-\
1200-cm region and those of other pollutant gases which may be usually present in the atmos-
phere, automobile exhaust at atmospheric pressure was introduced into the gas cell. No ab-
sorption lines were detected using a diode laser to scan the wavelength region containing several
SO, lines, indicating that over this region the pollutant gases present in the exhaust specimen
(including water vapor) do not interfere with the SO., determination.
5.3 Long-Path Transmission Experiment
In order to evaluate qualitatively the characteristics of diode laser transmission over rel-
atively long atmospheric paths, collimated radiation from the laser was directed outside the
laboratory to the rooftop of another building 38 meters away. A mirror on the other building
re-directed the beam toward the source, where 3.8-cm-diameter collecting optics focused the
radiation onto a Ge:Cu detector. Over 50 percent of the emitted laser radiation was collected
after traversing the 76-meter two-way path through the atmosphere. (The experiment was pre-
liminary, and no attempt was made to achieve ultimate optimization or to scan through regions
of SO, absorption.) It was clear, however, that the stability of the return signal indicated neg-
ligible atmospheric scattering, with a signal-to-noise ratio of over 50:1 for the relatively low
20jxW of laser power. As has been pointed out by Kildal and Byer, high-power lasers are not
necessary for sensitive, long-path monitoring of the atmosphere by "resonance absorption."
22
-------
ACKNOWLEDGMENTS
During the performance of the tasks described in this report,
assistance was received from many individuals not directly sup-
ported by the program or involved in its implementation. Their
unselfish generosity of time, knowledge, and even personal com-
mitment helped immeasurably in meeting the program objectives.
In the semiconductor diode laser development, the contribu-
tions of A. R. Calawa are of particular importance in that he sup-
plied the devices used during this period. Most of the crystal-
growing techniques have been developed and are still being ad-
vanced by T. C. Harman.
Initial encouragement for the application of tunable diode
lasers to gaseous pollutant detection came from P. L. Kelley, who
has continued to give valuable advice on all aspects of the program.
Helpful comments and critical appraisals were also provided by
J. O. Dimmock, I. Melngailis, R. H. Rediker, and R. H. Kingston
(RHK first proposed remote heterodyne detection). General en-
couragement and advice were also received from many other
members of the Lincoln Laboratory staff.
Appreciation is expressed to S. A. Clough of the Air Force
Cambridge Research Laboratories for providing theoretical band
structure information on SC>2 and collaborating in the interpreta-
tion of the measured ultra-high-re solution data. D. I. Underwood
of Lincoln Laboratory has performed further computer analysis
of Clough1 s theoretical data.
For supplying recent unpublished data of the positions and
strengths of NH3 lines used for calibrating the SC>2 laser-produced
spectra, we are deeply indebted to John Curtis, Peter Yin, and
K. Narahari Rao of Ohio State University
23
-------
REFERENCES
1. Detection of Air Pollutants with Tunable Diode Lasers. E. D. Hinkley and
P. L. Kelley, Science 171. 635 (1971).
2. The Use of Lasers in Pollution Monitoring, I. Melngailis, IEEE Transactions
on Geoscience Electronics (to be published).
3. Infrared Spectroscopy and Infrared Lasers in Air Pollution Research and Mon-
itoring, P. L. Hanst, Applied Spectroscopy 24, 161 (1970).
4. Infrared Spectroscopy of CO Using a Tunable PbSSe Diode Laser. K. W. Mil,
F.A.. Blum, A. R. Calawa, and T. C. Harman, Appl. Phys. Lett. 19, 79 (1971).
5. Observation of Lambda Doubling and Zeeman Splitting in the Fundamental In-
frared Absorption Band of Nitric Oxide. K. W. Mil, F. A. Blum, A. R. Calawa,
and T. C. Harman (submitted for publication).
6. Crystal Growth. Annealing, and Diffusion of Lead-Tin Chalcogenides,
A. R. Calawa, T. C. Harman, M. Finn, and P. Youtz, Trans. Met. Soc. Amer.
Inst. Mining Eng. 242, 374 (1968).
7. Catalog of Infrared Spectra for Qualitative Analysis of Gases. R. H. Pierson,
A.N. Fletcher, and E. St. Clair Gantz, Anal. Chem. 28, 1218 (1956).
8. I. Melngailis and T.C. Harman (private communication); see also Ref. 2.
9. The Infrared Spectrum and Molecular Constants of Sulfur Dioxide. R. D. Shelton,
A. H. Nielsen, and W. H. Fletcher, J, Chem. Phys. 2J, 2178 (1953).
10. Direct Observation of the Lorentzian Line Shape as Limited by Quantum Phase
Noise in a Laser above Threshold. E. D. Hinkley and Charles Freed, Phys.
Rev. Lett. 23, 277 (1969).
11. Vibration-Rotation Structure in Absorption Bands for the Calibration of Spec-
trometers from 2 to 16 Microns, E. K. Plyler, A. Danti, L. R. Elaine, arid
E.D. Tidwell, NBS Monograph 16, 21 June I960. The Low-Frequency Vibra-
tion Rotation Bands of the Ammonia Molecule. J. S. Garing, H. H. Nielsen,
and K. N. Rao, J. Molec. Spectros. ^, 496 (1959). Joan Curtis, Peter Yin,
and K. N. Rao (private communication).
12. S.A. Clough (private communication).
13. Microwave Spectrum and Rotational Constants for the Ground State of DgO,
W. S. Benedict, S.A. Clough, L. Frenkel, and T. E. Sullivan, J. Chem. Phys. 53,
2565 (1970).
14. G. Steenbeckeliers, Ann. Soc. Sci., Bruxelles 82, 331 (1968).
15. Ozone Absorption in the 9.0 Micron Region. AFCRL Report 65-862, Physical
Sciences Research Paper No. 170 (1965), S.A. Clough and F.X. Kneizys;
Coriolis Interaction in the v. and v> Fundamentals of Ozone. S.A. Clough
and F.X. Kneizys, J. Chem. Phys. 44, 1855 (1966).
16. Use of CO and CO2 Lasers to Detect Pollutants in the Atmosphere. R. T. Menzies,
Appl. Optics 10, 1532 (1971).
17. Low-Level Coherent and Incoherent Detection in the Infrared. R. J. Keyes and
T. M. Quist, Chapter 8 of "Semiconductors and Semimetals," edited by
R. K. Willardson and A. C. Beer (Academic Press, New York, 1970).
18. J. U. White, J. Opt. Soc. Amer. 32, 285 (1942).
19. Absorption and Emmision by SOg between 1050 and 1400 cm"1 (9.5-7.1 urn).
D. E. Burch, J. D. Pembrook, and David A. Gryvnak, Final Technical Report
to the Environmental Protection Agency under Contract No. 68-02-0013,
July 1971.
2 0. Comparison of Laser Methods for the Remote Detection of Atmospheric Pol-
lutants, H. Kildal and R. L. Byer, Proc. IEEE 59, 1644 (1971).
24
-------
APPENDIX A
TUNABLE LASER SPECTRA OF THE v^ BAND OF SULFUR DIOXIDE
Shown here are some of the actual diode laser scans of portions of the v^ absorption band
of SO2 at room temperature. The NHj lines used for wavelength calibration are also shown,
and are tabulated in Appendix C. The SO2 scans were usually taken at three different pressures:
0.1, 1, and 10 Torr. In most instances the 1-Torr curves were used to determine the funda-
mental line intensities. These are tabulated in Appendix B.
Experimentally, emission from the tunable diode laser was collimated by a BaF2 lens 3.8cm
in diameter and having a 10.6-cm focal length. The collimated radiation passed through a 30-cm-
long stainless steel or pyrex cell with BaF~ windows, thence to another lens for focusing onto the
entrance slit of a grating spectrometer. The spectrometer served a twofold purpose: (1) to
provide approximate wavelength information, and (2) to limit, even with wide slit positions, the
passband of the optical radiation from the laser so that power in any other modes would not be
detected. The detector was liquid-helium-cooled Ge:Cu. The SO2 was of 99.98 percent purity,
obtained from the Matheson Company, Inc. of East Rutherford, New Jersey. Evacuation of gas
cells, detector Dewars, and laser Dewars was achieved with a turbomolecular pump. Gas pres-
sure was measured with a capacitance manometer.
Figure A-1 shows diode laser scans for SO2 in the region from 1114.5 to 1115.2 cm" and
can be used to illustrate the procedure that was followed. The top curve corresponds to the
evacuated-cell case (p = 0), and the relatively noisy horizontal line slightly above the abscissa
shows the detector noise even when the laser beam is blocked, and serves as the reference for
zero transmission. The bell-shaped profile of the p = 0 curve is caused mainly by the change in
laser power as the diode is tuned across this mode it is not usually caused by the spectrometer
slit function, except over very wide tuning ranges. It may be noted that there appears to be a
"ripple" on even the p = 0 curve; this is produced by a Fabry-Perot effect between some re-
flecting (or partially reflecting) surfaces illuminated by the laser beam. Although attempts are
usually made to reduce these oscillations, they can be used to check linearity of tuning since
they are periodic in wavelength. After the p = 0 curve has been taken, 0.1 Torr of SO2 is placed
in the cell and a second curve produced. Sequentially, the 1-Torr and 10-Torr curves for SO,
LJ
are recorded, followed by the NH3 calibration trace.
Laser spectroscopy permits the fundamental absorption profile of a spectral line to be ob-
served, and its basic strength determined in accordance with the Beer-Lambert equation,
aP=-pTln(I/y ' 'A'1'
Here, I/I is the fractional transmission at the line center, p the gas pressure, L the cell
length, and a' the absorption coefficient per unit pressure (one measure of line "intensity").
Because of a relatively high rate of self-broadening (40MHz/Torr for FWHM)coupled with
a relatively narrow Doppler width as a result of a large molecular weight, pressure broadening
of SO2 spectral lines is noticeable even at pressures as low as 1 Torr. The Doppler width can
be calculated from the equation
, (A-2)
where, if T is in °K, the molecular weight in grams, and \ in jim, the Doppler width is in units
of MHz. For SO2 at room temperature, the Doppler width (FWHM) is 53 MHz at 8.7 |im. At a
25
-------
self-broadening rate of 40MHz/Torr, noticeable increases in the linewidths will occur for SO-
pressure above about 0.15 Torr, becoming approximately 66 MHz at 1 Torr, or 1.24 times
larger than the Doppler width. Since the absorption constant (per unit pressure) of Eq. (A.-1)
varies inversely with linewidth, values arrived at on the basis of measurements made at 1-Torr
pressure must be increased by 1.24 in order to give data corresponding to the Doppler limit.
This adjustment was made in reducing the data from Figs. A-1 through A-20 to the form shown
in Appendix B.
26
-------
S02 ABSORPTION AT 0.1, 1, 10TORR
ONE NH3 LINE AT 1114.71 CM"'
CELL LENGTH: 30 CM
I
I
I
EVACUATED CELL
(p = 0)
1114.6 1114.7 1114.8 1114.9
WAVENUMBER (cm"')
Fig.A-1.
1115.1 1115.2
S02 ABSORPTION AT 0.1, 1, 10 TORR
CELL LENGTH: 30 CM
o
o
I
I
1114.6 1114.7
1114.8 1114.9 1115.0
WAVENUMBER (cm"')
1115.2 1115.3
Fig. A-2.
27
-------
p= 0
S02 ABSORPTION AT 0.1, 1, 10 TORR
CELL LENGTH: 30 CM
z
o
me.7 me.B
WAVENUMBER (cm"1)
Fig. A-3.
S02 ABSORPTION AT 0.1, 1, 10 TORR
CELL LENGTH: 30 CM
o
2
z
o
LASER BLOCKED
WAVENUMBER (cm"')
Fig. A-4.
28
-------
S02 ABSORPTION AT 0.1, 0.2, 0.5 TORR
CELL LENGTH: 30 CM
o
o
o
v>
I
I
1137.6 1137.7
WAVENUMBER (cm"')
Fig. A-5.
S02 ABSORPTION AT 0.1 AND 0.2 TORR
ONE NH3 LINE AT 1139.463 CM"1
CELL LENGTH: 30 CM
p = 0
z
<
UI
>
RECORDER "OFF"
I
I
I
1139.1 1139.2 1139.3
WAVENUMBER (cm"')
1139.4 1139.5 1139.6
Fig. A-6.
29
-------
SO, ABSORPTION AT 0.2, 0.5, 1 TORR
1139.8 1139.9
WAVENUMBER (cm"1)
Fig. A-7.
NO NH, LINES IN THIS REGION
CELL LENGTH: 30 CM
I-
z
IE
UJ
0.
1141.e
WAVENUMBER (cm"')
Fig. A-8.
30
-------
S02 ABSORPTION AT 0.1, 1, 10 TORR
CELL LENGTH: 30CM
p = 0
I
I
I
1144.6 1144.7 1144.8 1144.9
WAVENUM8ER (cm"1)
Fig. A-9.
z
o
S02 ABSORPTION AT 0.1, 1, 10 TORR
CELL LENGTH: 30CM
p = 0
I
I
1147.0 1147.1
WAVENUMBER (cm"')
Fig. A-10.
31
-------
S02 ABSORPTION AT 0.1, 1, 10TORR
NO NH3 LINES IN THIS REGION
CELL LENGTH: 30CM
RECORDER "OFF"
J_
1148.8
WAVENUMBER (cm"')
Fig. A-11.
S02 ABSORPTION AT 0.1, 1, 10TORR
CELL LENGTH: 30 CM
p = 0
LASER BLOCKED,
1149.1 1149.2
WAVENUMBER (cm'1)
Fig. A-12.
32
-------
S02 ABSORPTION AT 0.1, 1, 10TORR
ONE NH3LINE AT 1152.850 CM"1
CELL LENGTH: 30 CM
z
o
I
I
1152.9 1153.0
WAVENUMBER (cm'1)
Fig. A-13.
S02 ABSORPTION AT 0.1, 1, 10 TORR
CELL LENGTH: 30 CM
p=0
_L
WAVENUMBER (cm"')
Fig. A.-14.
33
-------
S02 ABSORPTION AT 0.1, 1, 10 TORR
TWO NH3 LINES AT 115e
CELL LENGTH: 30 CM
z
o
I
I
1156.1 1156.1
WAVENUMBER (cm"1)
Fig. A-15.
p =0,0.1 TORR
I
I
1156.8 1157.0
WAVENUMBER (cm"1)
Fig. A-16.
34
-------
S02ABSORPTION AT 0.1, 1, 10 TORR (lower scons)
EIGHT IMH3 LINES (up|
CELL LENGTH: 30 CM
EIGHT NH3 LINES (upper scons)
NH, (0,
z
o
50h
WAVENUMBER (cm"1)
Fig. A-17.
z
o
S02 ABSORPTION AT 0.1, 1, 10 TORR
THREE NHjLINES
CELL LENGTH: 30 CM
NH, LASER BLOCKED
WAVENUMBER (cm"')
Fig. A-18.
35
-------
1164.0 1164.2
WAVENUMBER (ctri"')
Fig. A-19.
S02 ABSORPTION AT 0.1, 1, 10 TORR
NO NH3 LINES IN THIS REGION
CELL LENGTH: 30 CM
z
o
z
<
z
UJ
p=0
WAVENUMBER (cm"')
Fig. A-20.
36
-------
APPENDIX B
TABULATION OF SULFUR DIOXIDE SPECTRAL DATA
On the following pages are shown the positions, intensities, and identifications of the SO_
lines displayed in the tunable diode laser scans of Appendix A.
The wavenumber assigned to each line corresponds to its theoretical value based on a band
center of 1151.71 cm" . The position of the band center was deduced from diode laser scans of
regions containing both SO_ and NH, absorption lines and should be accurate to ±0.01 cm" .
The intensity of each line is listed in units of cm /Torr and represents the fundamental
value in the Doppler-broadened regime.
Identification of each line is provided by the upper and lower energy level quantum numbers
(j; KA, Ky and (Jj'KA> KjL,), respectively.
Transition*
,)*»(J, KA, K£ ) Wavenumber
(J|K
(12,08,04) (13,09,05)
(35,10,26) (35,11,25)
(34,10,24)(34,11,23)
unknown
(33,10,24) (33,11,23)
unknown
unknown
unknown
(32,10,22) (32,11,21)
(24,15,09) (23,16,08)
(31,10,22)-(31,11,21)
(26,05,21) (27,06,22)1
(21,06,16) (22,07,15)1
unknown
(30,10,20) (30,11,19)
(16,07,09) (17,08,10)
(29,10,20) (29,11,19)
(11,08,04) (12,09,03)
(28,10,18) (28,11,17)
(27,10,18) (27,11,17)
(25,09,17)
(24,09,15)
(25,04,22)
(23,09,15)
(22,09,13)
(26,04,22)
(21,09,13)
(20,05,15)
(20,09
(15,06
(19,09,11)
(10,07,03)
(18,09,09)
(17,09,09)
(16,09,07)
11)-
10)-
(25,10,16)
(24,10,14)
-(26,05,21)
(23,10,14)
(22,10,12)
(27,05,23)
(21,10,12)
(21,06,16)
(20,10,10)
(16,07,09)
(19,10,10)
(11,08,04)
(18,10,08)
(17,10,08)
(16,10,06)
1114.489 cm
.550
.650
.670
.749
.787
.793
.836
.845
.864
.939
.954
.982
1115.031
.073
.120
.166
.207
.291
1118.741
.815
.838
.887
.956
1119.013
.022
.042
.086
.102
.148
.156
.206
.262
.315
-1
Intensity
0.024 cm /Torr
0.008
0.007
0.003
0.007
0.002
0.001
0.003
0.008
0.001
0.011
0.022
0.004
0.010
0.022
0.009
0.021
0.014
0.021
0.007
0.010
0.009
0.009
0.010
0.008
0.010
0.012
0.010
0.014
0.009
0.017
0.009
0.008
0.007
* Following conventional spectroscopic notation where the upper state is listed first, () rep-
resents the direction for emission, and () represents absorption.
37
-------
Transition*
Wavenumber
Intensity
(19,02,18) (20,03,17)
(37,01,37) (38,00,38)
(29,12,18) (28,13,15)
(36,01,35) (37,02,36)
(17,01,17) (18,02,16)
(34,02,32) (35,03,33)
(08,05,03) (09,06,04)
(13,04,10) (14,05,09)
unknown
unknown
(08,02,06) (09,03,07)
(30,01,29) (30,02,28)
(25,01,25) (25,02,24)
(18,01,17) (19,02,18)
(29,04,26) (30,03,27)
(27,03,25) (27,04,24)
(13,06,08) -(12,07,05)
(27,02,26) (27,03,25)
(40,03,37) (40,04,36)
(34,04,30) (34,05,29)
(19,03,17)-
(28,01,27)-
(14,01,13)-
(38,03,35)-
(19,01,19)-
(23,03,21)-
(05,02,04)-
(17,03,15)-
(15,03,13)-
(13,03,11)-
(11,03,09)-
(12,03,09)-
(18,00,18)-
(10,03,07)-
(09,03,07)-
(14,03,11)-
(23,02,22)-
(08,03,05)-
(07,03,05)-
(06,03,03)-
(28,08,20)-
(05,03,03)-
(16,03,13)-
(04,03,01)-
(19,02,18)-
(18,03,15)-
(12,01,11)-
(21,01,21)-
(04,02,02)-
(32,02,30)-
(27,04,24)-
(20,03,17)-
(22,00,22)-
(36,03,33)-
-(19,04,16)
-(28,02,26)
-(15,02,14)
-(38,04,34)
-(20,00,20)
-(24,02,22)
-(06,03,03)
-(17,04,14)
-(15,04,12)
-(13,04,10)
-(11,04,08)
-(12,04,08)
-(19,01,19)
-(10,04,06)
-(09,04,06)
-(14,04,10)
-(23,03,21)
-(08,04,04)
-(07,0.4,04)
-(06,04,02)
-(27,09,19)
-(05,04,02)
-(16,04,12)
-(04,04,00)
-(20,01,19)
-(18,04,14)
-(13,02,12)
-(21,02,20)
-(05,03,03)
-(32,03,29)
-(28,03,25)
-(20,04,16)
-(22,01,21)
-(36,04,32)
1127.108 cm
.129
.1511
.1521
.166
.182
.183
.205
.293
.310
1137.481
.505
.537
.591
.628
.651
.715
.757
.810
.828
1139.147
.162
.169
.179
.215
.265
.293
.321
.447
.539
.609
.648
.653
.663
.664
.669
.685
.695
.709
.730
.744
.745
.748
.760
.853
.910
.948
.970
.973
1140.095
.125
.164
.192
.248
-1
0. 008 cm
0.020
0.017
0.002
0.014
0.023
0.020
0.003
0.004
0.029
0.026
0.021
0.040
0.015
0.034
0.005
0.029
0.018
0.023
0.041
0.024
0.031
0.018
0.065
0.021
0.021
0.039
0.039
0.027
0.022
0.028
0.060
0.040
0.033
0.022
0.017
0.014
0.016
0.016
0.045
0.010
0.029
0.029
0.020
0.013
0.015
0.018
0.009
0.034
0.018
0.020
-1
/Torr
* Following conventional spectroscopic notation where the upper state is listed first, () rep-
resents the direction for emission, and () represents absorption.
38
-------
Transition*
(J',K'
K»0
Wavenumber
(17,02,16)-
(08,01,07)-
(14,05,09)-
(20,00,20)-
(17,02,16)-
(31,05,27)-
(20,06,14)-
(15,01,15)-
(14,00,14)-
(18,02,16)-
(22,02,20)-
(20,02,18)-
(16,00,16)-
(11,01,11)-
(11,01,11)-
(02,01,01)-
(08,03,05)-
(25,06,20)-
(31,07,25)-
(23,04,20)-
(20,01,19)-
(14,04,10)-
(08,00,08)-
(17,04,14)-
(14,01,13)-
(41,08,34)-
(06,01,05)-
(23,05,19)-
(26,04,22)-
(21,04,18)-
(12,00,12)-
(08,01,07)-
(35,07,29)-
(29,06,24)-
(04,00,04)-
(12,01,11)-
(28,05,23)-
(11,02,10)-
(10,01,09)-
(22,04,18)-
(30,06,24)-
(08,00,08)-
(20,04,16)-
(38,07,31)-
(26,05,21)-
(32,06,26)-
(09,02,08)-
(15,03,13)-
(19,04,16)-
(34,07,27)-
(29,06,24)-
(05,01,05)-
(21,04,18)-
(06,00,06)-
(00,00,00)-
-(18,01,17)
-(09,02,08)
(13,06,08)
-(20,01,19)
(17,03,15)
(32,04,28)
(19,07,13)
(16,00,16)
(15,01,15)
(18,03,15)
(22,03,19)
(20,03,17)
(16,01,15)
(11,02,10)
(12,00,12)
(03,02,02)
(07,04,04)
(24,07,17)
(30,08,22)
(24,03,21)
(20,02,18)
(13,05,09)
(09,01,09)
(16,05,11)
(14,02,12)
(42,07,35)
(06,02,04)
(22,06,16)
(27,03,25)
(22,03,19)
(12,01,11)
(08,02,06)
(34,08,26)
(28,07,21)
(05,01,05)
(12,02,10)
(29,04,26)
(12,01,11)
(10,02,08)
(23,03,21)
(31,05,27)
(08,01,07)
(19,05,15)
(37,08,30)
(25,06,20)
(31,07,25)
(10,01,09)
(14,04,10)
(20,03,17)
(35,06,30)
-(30,05,25)
(06,00,06)
-(20,05,15)
-(06,01,05)
-(01,01,01)
1141.634 cm
.639
.681
.689
.727
.807
.872
.909
1142.005
1144.362
.403
.477
.617
.672
.735
.742
.8051
.8061
.841
.890
.904
1145.084
.144
1146.899
.956
.9641
.9651
1147.073
.074
.080
.110
.128
.129
.151
.153
.177
.194
.212
.218
1148.736
.768
.810
.835
.873
.942
.957
1149.034
.046
.107
.151
.241
.242
.310
.346
.389
-1
Intensity
0.032 cm /Torr
0.022
0.006
0.024
0.040
0.008
0.009
0.067
0.062
0.025
0.022
0.025
0.014
0.012
0.030
0.010
0.008
' 0.004
0.006
0.031
0.005
0.020
0.008
0.051
0.024
0.011
0.010
0.038
0.034
0.005
0.020
0.047
0.006
0.015
0.030
0.009
0.005
0.044
0.011
0.004
0.008
0.005
0.011
0.009
0.008
0.002
0.016
0.005
0.015
0.005
* Following conventional spectroscopic notation where the upper state is listed first, () rep-
resents the direction for emission, and () represents absorption.
39
-------
Transition*
(JJ
K"
Wavenumber
Intensity
(20,05
(32,05
(33,05
(14,02
(39,06
(26,04
(29,07
(23,03
(20,03
15) (21
unknown
27) (31
unknown
29) (32
,12) (13
34)-(38
22) (25
23) (30
21) (22
17) (19
,04,18)
,06,26)
,06,26)
,03,11)
,07,31)
,05,21)
,06,24)
,04,18)
,04,16)
(10,03,07)-
(09,03,07)-
(24,06,18)-
(19,05,15)-
(02,01,01)-
(33,08,26)-
(14,02,12)-
(12,01,11)-
(10,02,08)-
(16,02,14)-
(08,02,06)-
(36,05,31)-
(05,01,05)-
(20,06,14)-
(30,04,26)-
(18,02,16)-
(04,02,02)-
(18,02,16)-
(24,03,21)-
(14,01,13)-
(10,00,10)-
(07,01,07)-
(05,02,04)-
(20,02,18)-
(22,03,19)-
(24,03,21)-
(14,01,13)-
(07,02,06)-
(20,03,17)-
(26,03,23)-
(11,01,11)-
(30,04,26)-
(13,02,12)-
(14,03,11)-
(36,04,32)-
(30,03,27)-
(16,01,15)-
(18,01,17)-
(24,02,22)-
(11,02,10)
-(10,02,08)
-(25,05,21)
-(20,04,16)
-(02,00,02)
-(34,07,27)
-(14,01,13)
-(12,00,12)
-(10,01,09)
-(16,01,15)
(08,01,07)
-(35,06,30)
-(04,00,04)
-(21,05,17)
-(29,05,25)
-(18,01,17)
-(04,01,03)
-(17,03,15)
-(23,04,20)
-(14,00,14)
-(09,01,09)
-(06,00,06)
-(05,01,05)
-(20,01,19)
-(22,02,20)
-(24,02,22)
-(13,02,12)
(07,01,07)
-(20,02,18)
-(26,02,24)
-(10,00,10)
-(30,03,27)
-(13,01,13)
(14,02,12)
-(36,03,33)
-(30,02,28)
(15,02,14)
(18,00,18)
(24,01,23)
1152.850 cm
.900
.931
.938
.962
.984
1153.004
.069
.086
.111
.123
1153.299
.363
.371
.450
.489
.557
1155.813
.837
.867
1156.058
.078
.079
.195
.337
.521
.523
1156.580
.616
.780
.842
.911
1157.142
.183
.222
.442
ASS
.479
.481
.643
.717
1158.924
.931
.954
.955
1159.013
.041
1159.170
.224
.258
-1
0.009 cm
0.006
0.006
0.005
0.005
0.011
0.003
0.008
0.005
0.009
0.010
0.007
0.010
0.006
0.008
0.018
0.005
0.044
0.029
0.031
0.042
0.027
0.012
0.020
0.023
0.007
0.008
0.004
0.014
0.015
0.010
0.006
0.022
0.029
0.027
0.022
0.030
0.020
0.046
0.029
0.056
0.020
0.025
0.026
0.022
0.027
-1
/Torr
* Following conventional spectroscopic notation where the upper state is listed first, () rep-
resents the direction for emission, and () represents absorption.
40
-------
Transition*
A' C Wavenumber Intensity
(38,03,35) (38,02,36) 1163.988cm" 0.007 cm /Torr
(25,02,24) (25,01,25) .996 0.011
(06,03,03) (05,02,04) 1164.050 0.012
(26,01,25) (26,00,26) .118 0.008
(22,00,22) (21,01,21) .140 0.038
(17,02,16) (16,01,15) .154 0.021
(31,04,28) (31,03,29) .194 0.013
(29,03,27) (29,02,28) .198 0.014
(26,02,24) (25,03,23) .263 0.013
(28,05,23) (28,04,24) .413]
(32,02,30) (32,01,31) .413 f °'°23
(28,01,27) (27,02,26) 1167.575 0.037
(39,06,24)(39,05,35) .6051 . - n
(13,03,11) (12,02,10) .6061
unknown .611 0.022
(30,02,28) (29,03,27) .674 0.021
(27,02,26) (26,01,25) .681 0.039
(37,03,35) (37,02,36) .751 0.010
(46,07,39) (46,06,40) .755 0.003
(29,01,29)(28,00,28) .786 0.041
(37,06,32) (37,05,33) .802 0.010
* Following conventional spectroscopic notation where the upper state is listed first, () rep-
resents the direction for emission, and () represents absorption.
41
-------
APPENDIX C.
NH CALIBRATION LINES FOR THE 1100-1200 CM'1 REGION
The table below lists some of the NH, lines used to calibrate the wavelength scale for the
tunable diode laser scans of the SO- absorption lines in the v . band. Where possible, the actual
lines have been identified and their fundamental (Doppler-limited) intensities measured.
Line"
aR(8,4)
-1
-1
Wavenumber (cm~ ) Intensity (cm" /Torr)
1114.71
0.25
aR(9,2)
aR(9,D
aR(10,10)
1137.596
1138.082
1139.463
0.09
0.005
aR(10,9)
aR(10,8)
aR(10,7)
aR(10,6)
aR(10,5)
aR(10,4)
aR(10,3)
aR(10,2)
Identification
not established
at this time
(from laser scans)
1143.905
1147.538
1150.460
1152.850
1154.734
1156.219
1156.562
1157.288
1158.044
1158.859
.882
.927
.975
1159.027
.048
.055
0.10
0.03
0.01
0.05
0.15
0.13
0.08
0.08
0.09
0.08
0.17
(Line identification follows.that of Ref. 11 (G, N,&R) of text.)
42
-------
APPENDIX D
DETECTION OF AIR POLLUTANTS WITH TUNABLE DIODE LASERS*
E. D. Hinkley and P. L. Kelley
Sensitive, specific detection of pollutant gases can be obtained
with tunable semiconductor lasers.
In this paper we describe the use of tunable Pb . Sn Te semiconductor diode lasers in
remote or long-range sensing and point sampling of molecular pollutant gases. Remote sensing
can be accomplished by measurements of the infrared absorption or emission lines character-
istic of a particular pollutant at atmospheric pressure. Point sampling permits gas pressures
to be reduced until the Doppler-broadened infrared absorption structure is revealed, thereby
making possible very good specificity. Since one measurement technique is used for all the
pollutant gases, the operational requirements are simplified. Moreover, the essentially instan-
taneous response is advantageous for a number of applications.
Several optical techniques for the detection of pollutant gases are currently being explored
in various laboratories. They range from classical methods using selective filters, spectre-
graphs, interferometers, and optical-correlation instruments, to the use of lasers for direct
23 4
absorption and Raman scattering. One advantage of optical sensing over conventional wet
chemical methods is that the measurements can be performed in situ by measuring emission
spectra, detecting scattered laser radiation, or by transmission measurements over relatively
long atmospheric paths. Infrared spectroscopy is considered an important technique since each
of the major molecular pollutants contains at least one strong band in the 3-15 (jan region. Dis-
persive infrared instruments currently used in the field usually have resolutions which are much
poorer than the linewidths of typical gases at atmospheric pressure, and potentially useful infor-
mation is lost. Even when high-quality spectrometers are used, the resolving power must usu-
ally be reduced in order to maintain a reasonable signal-to-noise ratio with the relatively low
power per unit spectral range available from incoherent sources.
The use of gas lasers as nondispersive instruments, pn the other hand, is severely restricted
by the inability to "tune" these lasers appreciably from their nominal wavelengths and the
match to pollutant gas absorption lines is seldom ideal.
Diode lasers of Pb. Sn Te can be "tailored" chemically, by adjusting the composition fac-
1 "X X
tor "x," to emit anywhere in the wavelength range from 6.5 to 32 \an (Ref. 5), thereby permitting
an ideal match to the strong infrared absorption lines of most of the molecular pollutant gases.
Tuning of individual lead-salt lasers can be achieved by application of pressure, an external
7 8
magnetic field, or by varying the laser temperature. In a simple manifestation of temperature-
tuning, a laser can be tuned over a range of 40 cm" quasi-continuously (from one continuously
tunable mode to another) by changing the diode current. This tuning range is sufficient to identify
the spectral signature of a given pollutant even in the presence of other molecular species, and
to permit a quantitative determination of its concentration. These features overcome the major
limitations of conventional gas lasers in monitoring applications.
* Published in Science 171, 635 (1971).
43
-------
Other tunable laser sources may also be useful for air pollution monitoring. These include
9
the spin-flip Raman lasers, organic dye lasers, and systems involving parametric oscillators
or difference frequency generators. At the present time these devices are much more complex
than the tunable semiconductor diode lasers; nevertheless, their wide tuning ranges should soon
find many applications in the more conventional forms of spectroscopy.
TUNABLE DIODE LASERS
The diode lasers are fabricated from vapor-grown single crystals of Pb. Sn Te semi-
10 i-x x
conducting material by cleaving the crystals into rectangular parallelepipeds (each containing
a p-n junction) of approximate overall dimensions 0.12 x 0.05 X 0.03cm. Low-resistance ohmic
contacts are formed, and the laser is then mounted onto the cold-finger of a cryogenic Dewar.
Liquid helium temperatures are presently required for continuous (cw) laser operation, with
pulsed operation possible at liquid nitrogen temperature (77° K). The infrared emission fre-
quency of the laser is, within the spontaneous emission bandwidth, determined by the refractive
index of the semiconductor and the physical length of the cavity. This frequency can be fine-
tuned by a small change in the d.c. diode current, which alters the refractive index of this
mixed-crystal semiconductor (which has low thermal conductivity) through heating. A single
diode laser can be "tuned" in this manner over 40cm in continuous bands up to 2 cm wide;
but because of mode jumps at the ends of these bands, emission occurs at only about one-half
the frequencies in this tuning range. Although continuous tuning is essential for fundamental
high-resolution spectroscopic studies, the broader, quasi-continuous tuning is quite adequate
for application to pollutant gas detection. That the diode laser provides effectively infinite res-
olution for these applications was demonstrated by a heterodyne experiment involving a CO-
13
laser, where we showed that the linewidth of a 0.24 mW Pbn 00Sn_ .0Te diode laser at 10.6 urn
/. U.oo U.1^
was 54kHz (1.8 x 10 cm ). By comparison, the narrowest Doppler-broadened linewidths of
even the heaviest molecular pollutants are tens of MHz at room temperature.
APPLICATION TO POLLUTANT DETECTION
In a typical monitoring application, several Pb . Sn Te diode lasers would be used, each
i ~X X
tailored to emit in a strongly absorbing infrared region of one of the pollutant gases. This is
necessary because of the limited cur rent-tuning range of these lasers, and desirable because
it provides for increased specificity. The small size of these lasers permits several to be
mounted into the same Dewar. Specific applications to pollutant detection can be divided into
three categories: point-sampling, long-path atmospheric transmission, and remote emission
sensing.
Point Sampling. This technique permits the absorption spectra to be measured at reduced
pressure where possible overlap with neighboring spectral lines of the gas (or with other gases
in the atmospheric sample) is minimized. Several experiments have been performed with
Pb. 00Sn,, .0Te diode lasers in the 10.6 urn region to illustrate their usefulness in such an
U.oo U.1£
application. Figure D-l shows absorption profiles of the strong sP(l,0) line of ammonia (NH )
at several pressures. At low pressure the full linewidth at half peak absorption is seen to
approximate the Doppler-limited value of 85 MHz. As the NH, pressure is increased toward
0.5 Torr, the 10-cm long cell becomes essentially opaque to radiation ne.^r the line center.
The great improvement in sensitivity of this non-dispersive technique can be appreciated if a
comparison is made between this spectrum and one obtained with a grating spectrometer having
44
-------
5 Torr
|U-5-3T4T|
10-cm Cell
T = 300°K
948.25
WAVENUMBER (cm )
Fig. D-l. Absorption spectra of sP(l,0) line of NH^ taken at various
gas pressures with a tunable Pbg 88^nO IZ^6 diode laser. Resolu-
tion (diode laser linewidth) is better than 0.1 MHz (3.3 X 10~6 cm'1).
-1 14
0.1-0.2 cm resolution. In the latter experiment an NH_ pressure of 5 Torr in a 200-cm long
cell produced the same maximum absorption shown here for the 50 mTorr curve where the
amount of NH, in the optical path is 2000 times smaller.
Figure D-2 shows the v vibration-rotation band of sulfur hexafluoride (SF, ) obtained with
A C
a high-quality grating spectrometer, as well as a segment of this band taken with a tunable
Pb_ 00Snn .0Te diode laser. Since SF, is not usually present in the atmosphere, except near
U. oo ".lc b
leakage sources such as high-voltage transformers, it can be used as a trace gas in certain
monitoring applications. Sulfur hexafluoride has an exceedingly complex spectrum in the
infrared which, as shown in Fig. 2, cannot be resolved in the 950 cm" region even with a high-
quality grating spectrometer, but is resolved by the diode laser scan. The widths of these lines
correspond to the 29 MHz value predicted for Doppler broadening of SF, at room temperature,
compared with 85 MHz for NH-. Since a determination of mass can be made from the Doppler
width, we can use tunable lasers to distinguish between gaseous species on the basis of their
molecular weights, and thereby confirm pollutant identification by point-sampling at reduced
pressure. The abscissa of Fig. 2 was precisely calibrated relative to the P(16) CO, line by the
i 7
heterodyne/absorption technique described earlier.
. Ethylene (C.H ) is one of the many hydrocarbons present in automobile exhaust. It is not
usually monitored separately, but is lumped with the others in a single measurement of total
hydrocarbons. In Fig. D-3(a) is shown the absorption spectrum of C_H near the Q-branch of
\
the v- vibration-rotation band. The 25% drop in laser intensity near 949.24cm is produced
17
by a shift of laser power from this longitudinal mode. By modulating the diode current with
a small (0.1%) superimposed sinusoidal current at a frequency of a few hundred Hz, the laser
emission frequency is modulated; and, by synchronous detection, the derivative of the absorp-
tion spectrum is obtained. This is shown in Fig. D-3(b) where the top trace is a calibration
spectrum of 1000 parts per million (ppm) C->H. in N . A total gas pressure of 5 Torr produces
a maximum derivative signal for C,H , representing a balance between pressure-broadening
45
-------
Diode Laser Scan
SFg Pressure
0.1 Torr
Cell Length: 10cm
Resolution: 3 x 10~ cm"
SF6 Pressure: 0.1 Torr
Cell Length: 25cm
940 945
WAVENUMBER (cm'1)
950
Fig. D-2. (Bottom) Grating spectrometer absorption curve of v$ band
of SF^ at room temperature. After Brunet and Perez.15 (Top) Diode
laser transmission scan of a segment of this band near the P(16) CC>2
laser line.
46
-------
2
in
in
in
UJ
K
C2H4 Pressure: 25 mTorr
Cell Length: 30cm
949.18
949.20
949.22
Evacuated Cell
949.24
949.26
WAVENUMBER (cm"1)
Fig. D-3(a). Absorption spectrum near the Q branch of
pressure. See Ref. (17) for experimental procedure.
at 25 mTorr
13.5mV(1000ppm C2H4 in N2)
Total Prmura: STorr
Cell Length: 30cm
949.18
949.20 949.22
WAVENUMBER (cm'1)
949.24
949.26
Fig. D-3(b). Derivative spectra of 02^ calibration sample of 1000 ppm
in J^2> an<^ three automobile exhaust samples at different amplifier gain
settings.
47
-------
(upper limit) and reduced total number of absorbing molecules (lower limit). The lower three
traces represent spectra obtained from automobile exhaust specimens, also taken at 5 Torr
pressure, but with increased amplifier gain. By measuring the signal amplitudes and com-
paring them with the calibration curve, C,H. concentrations of from 22 to 74ppm are deduced.
It is unmistakably clear that each automobile exhaust sample contains C_H., and that there is
no interference from water vapor or any other molecular constituents present. Moreover, the
point at which the diode laser power changed abruptly is easily identifiable, and affects neither
the qualitative nor the quantitative comparisons.
There is no question that this technique is very specific at low pressures where Doppler-
7
broadening is predominant and collision-broadening negligible. Here each line is about 6x10 Hz
-3 -1
(2 x 10 cm ) wide. The total (compositional) tuning range of the Pb. Sn Te diode lasers is
13 1 -x x
about 3 x 10 Hz. Assuming 50 gaseous molecular species in the atmosphere, each with 2000
reasonably strong lines in this range, the separation between individual lines should average
about five times the linewidth.
The maximum absorption constant per unit pressure of the strongest lines of NH,, SF,, and
C H approximate 2 reciprocal centimeters per Torr. A detection limit of 0.3ppm is expected
LJ 4
for such lines in a 1-meter long cell on the basis of these initial experiments; but this limit
should improve to the low parts-per-billion (ppb) range with the use of longer path-length, mul-
tiply reflecting cells, and improved electronics. Moreover, noting that the strengths of strong
absorption lines of these three quite different gases are nearly equal, we expect those for a num-
ber of pollutant gases to have the same general values, yielding ultimate detection limits which
are also in the low parts-per-billion range.
Long-Path Atmospheric Transmission. Even at atmospheric pressure the collision-
broadened widths of spectral lines for typical gases are well below the resolution of infrared
spectrometers presently available for field use. If a tunable diode laser is used, the resolution
limit is imposed by the pressure-broadened linewidth of the gas itself, yielding all the available
optical information within the tuning range. Monitoring of pollutant gases in the atmosphere by
long-path transmission has the advantage over point sampling in that it produces an average
value for the pollutant gas concentration over the path, which may be of greater overall signifi-
cance in ambient air analysis.
If we assume that the pressure-broadened linewidth of the pollutant gas equals the Doppler
width at about 5 Torr, and if we use a Doppler-broadened absorption constant per unit pressure
of 2 reciprocal centimeters per Torr, the absorption constant (in reciprocal centimeters) at
the center of a strong line at atmospheric pressure is
a » 10"5c cm"1 , (1)
where c is the concentration in ppm. Over a one kilometer path, a pollutant gas with an aver-
age concentration of 1 ppb will produce a 0.1% change in transmission as we tune through this
absorption line.
The collection efficiency will now be considered. The emitting area of the diode laser
(which operates in a low-order transverse mode) is 500 pm x 40 |jm. If a 20X beam expander
is used, a laser beam near 10 prn wavelength diverges to an area of about 10 cm one kilo-
meter from the source. For a collection area of 100 cm and a 1 W pulsed diode laser (tuned
by varying either the pulse width or repetition rate), the collected power in the absence of
attenuation is 1 mW. For an average pollutant concentration of 1 ppb over this path, a signal
48
-------
change of 1 jiW is produced this can be measured easily using synchronous detection techniques.
A corner reflector (of the order of 100 cm or greater area to avoid diffraction losses in the
return beam) situated remotely would eliminate the requirement of having instrumentation at
two locations. By positioning a number of corner reflectors at various points around the source,
many radial measurements could be made from the same monitoring station. It should also be
pointed out that heterodyne detection of absorption lines, using an offset local oscillator, should
yield an improved signal-to-noise ratio because of the narrow emission linewidth of the laser.
Transmission experiments should be performed with these diode lasers, operating both
pulsed and cw, to test the feasibility of long-path monitoring, and to show whether or not speci-
ficity in identifying a particular gaseous pollutant can be retained in the presence of absorption
lines from other molecular constituents. As is well known, H,O and CO_ have high abundances
L* Li
in the atmosphere and can cause serious interference with infrared spectroscopy. We, there-
fore, restrict ourselves to considering the 8-13 jim atmospheric "window," which is approxi-
mately 500cm wide. For example, lasers tailored to emit in the 8.6 jim region are better
suited to sulfur dioxide (SO,) detection than those tailored to the (somewhat stronger) 7.5 |o.m
Li
absorption band. Even in the "window" region, the wings of H_O and CO- lines may produce
greater absorption than that corresponding to pollutants in small concentrations; however, this
is somewhat compensated by the fact that the slopes of these wings are relatively flat compared
with the slope of a pollutant absorption line near its peak. Derivative spectroscopy, similar to
that shown in Fig. D-3(b), will improve detection of these small peaks relative to the slowly
varying Lorentzian wings of the dominant atmospheric absorption lines.
We now consider the interference between pollutant molecules at atmospheric pressure,
and start with the assumption that there are 25 molecular species present which absorb in the
9
8-13 jim "window." At atmospheric pressure the lines are broadened to about 6 X 10 Hz, and
most of the structure within individual J values of the P and R branches disappears the Q
branch, if present, also loses most of its structure since closely spaced strong lines overlap
and weak lines become lost in the tails of strong lines. We estimate, therefore, that at atmos-
pheric pressure there are only 100 resolvable lines per molecule, which implies that, on the
average, the peaks of the lines are separated by one linewidth. Assuming each line is resolvable
Q
if a frequency interval three linewidths wide (~18 x 10 Hz) is not occupied by an absorption line
from another molecule, and a uniform probability distribution, there should be an average of
five resolvable lines per molecular specie.
Remote Emission Detection. One of the more difficult problems in air pollution monitoring
is gaining access to industrial smokestacks. Optical remote-sens ing techniques can be extremely
useful for such applications.
Using the Pb, Sn Te diode laser as a tunable local oscillator, infrared heterodyne detec-
18 i-x x
tion of spectral emission lines from pollutants present in smokestack effluent may be per-
19
formed. (The technique is similar to that used by radio-astronomers to detect weak emissions
from distant sources.) For heterodyne detection with an ideal photon counter (unit quantum effi-
ciency and zero excess noise), the signal divided by the local-oscillator-generated noise is
where 8P /<)v is the received signal power per unit frequency range in a single spatial mode,
s
hv the infrared photon energy, B the intermediate-frequency (IF) bandwidth of the detection
49
-------
system, and T the post-detection integration time. For the case where the solid angle sub-
tended by the source at the detector is larger than the diffraction-limited solid angular resolu-
tion of the collection optics a situation usually valid for smokestack plumes at ranges up to
1km with optics as small as 1cm the signal-to-noise ratio becomes
hf/kT hK/kT.
e S-1 e D -1
(3)
where L is the thickness of the plume and T its temperature, e, the emissivity of the back-
ground at temperature T, , and k is Boltzmann's constant. We have assumed (a) that the IF
bandwidth is less than the emission linewidth, (b) that the emission and absorption due to the
pollutant in the ambient atmosphere is not important because of its low concentration relative
to that in the plume, (c) negligible background attenuation from the wings of other molecular
absorption lines, (d) that the local oscillator has sufficient power to overcome the other sources
of noise (this is readily satisfied with a 1 mW diode laser). Note that since e, is often much
s D
smaller than unity (about 0.1 in the 10 pm region of the infrared when there are no clouds in the
background) the source does not have to be hotter than the background; in fact, it could be colder.
Q
We choose the following representative values for the parameters: B = 10 Hz, r = Isec, a =
-5 -1
10 c cm (where c is the concentration in ppm), L = 1 meter, hv = 0.125eV, T0 = 400°K,
S
T, = 300°K, and e, = 1, to find
b b
J-0.2C .
It should be possible, therefore, to remotely detect pollutant molecules such as SO, in smoke-
stack plumes in concentrations greater than 10ppm. In order to make absolute measurements
of concentration, aL, must be less than unity and the gas temperature must be known. If aL
exceeds unity, so that the plume is optically dense, then a weaker emission line can be selected.
Gas temperature can be measured in several ways: by heterodyne detection of one of the stack
gases at two wavelengths; by performing multicolor broadband radiometric measurements; or
by direct measurement at the top of the stack.
Atmospheric turbulence may cause fluctuations in the received signal. This can be over-
come by alternately tuning the local oscillator between the peak emission wavelength of the
pollutant gas and one of negligible emission in a time short enough so that turbulent effects are
negligible (~1 millisecond), and then summing the excess pollutant signal for a large number of
measurement intervals. Two local oscillators could also be used simultaneously one operating
at the emission peak, and the other offset from it.
GENERAL CONSIDERATIONS
The absorption bands of a few of the major molecular pollutant gases (such as CO and NO)
are beyond the compositional tuning range of Pb1_xSnxTe diode lasers. Development of other
semiconducting materials, for example, PbS . Se , which has produced laser emission under
21 i-x x
electron-beam excitation, may yield tunable diode lasers useful for detecting these gases as
well. Semiconductor lasers which operate in the 3-6.5 fim region should have the added advantage
of a higher ultimate operating temperature.
Although temperatures below 20° K are presently required for continuous operation, pulsed
laser emission can be obtained at liquid nitrogen temperature (77° K). With the availability of
50
-------
relatively inexpensive cryogenic coolers, fast on-line monitoring of gases present in automobile
exhausts or in chemical processing plants appears possible by direct absorption techniques
using diode lasers, each of which is tailored to a strong absorption line of one of the gases to
be monitored. It is expected that further improvements in materials fabrication will eventually
permit cw operation at this temperature as well.
SUMMARY
Preliminary experiments indicate that tunable Pb . Sn Te diode lasers will be useful in the
identification and sensitive detection of most of the atmospheric pollutant gases. For point-
sampling applications, concentrations in the ppb range should be measurable with unsurpassed
specificity. For long-range atmospheric transmission techniques, the improved resolution
capability and tunability of these diode lasers make them attractive replacements for spectrom-
eters and fixed-frequency laser sources where cryogenic temperature operation is not a
serious impediment. By using these lasers as tunable local oscillators in the infrared hetero-
dyne configuration, remote passive detection of gases present in smokestack effluent appears
possible. Finally, pulsed operation at temperatures available with simple cryogenic coolers
permits immediate application to fast detection of gases present in automobile exhaust and in
chemical processing plants.
REFERENCES
1. A.R. Barringer, J. Opt. Soc. Amer. 60, 729 (1970); C. B. Ludwig,
M. Griggs, and E.R. Bartle, ibid. 731 (1970).
2. H. J. Gerritsen, Trans. Amer. Inst. Mining Eng. 235. 428 (1966);
G. B. Jacobs and L. R. Snowman, Inst. Elec. Electron. Eng. J. Quantum
Electron. QE-3, 603 (1967); D. N. Jaynes and B. H: Beam, Appl. Opt.
8, 1741 (191^17 P. L. Hanst, Appl. Spectrosc. 24, 161 (1970); R. T.
Menzies, N. George, and M. L. Bhaumik, Inst. Elec. Electron. Eng. J.
Quantum Electron. QE-6, 800 (1970).
3. H. Inaba and T. Kobayasi, Nature 224, 170 (1969); T. Kobayasi and
H. Inaba, Appl. Phys. Lett. 17, 139 (1970).
4. Microwave spectroscopy has many of the advantages of optical spectro-
scopy for the detection of molecular pollutants, but with the following
limitations: (i) insufficient resolution at atmospheric pressure; (ii) very
low pressures required for high-resolution studies in the Doppler-limited
regime; and (iii) molecules such as CC>2- ^2^4' anc* ^6^6' wnicn lack a
dipole moment in the ground state, cannot be detected.
5. J. O. Dimmock, I. Melngailis, and A. J. Strauss, Phys. Rev. Lett. 16, 1193
(1966); J. F. Butler, A.R. Calawa, and T. C. Barman, Appl. Phys. Lett.
9, 427 (1966); J. F. Butler and T. C. Harman, ibid. 12, 347 (1.968); for a
review, see I. Melngailis, J. Physique, Colloque C-4, Suppl. au no. 11-12
(1968), pp. C4-84 to C4-94 and T. C. Harman, J. Phys. Chem. Sol., in
press.
6. J. M. Besson, J. F. Butler, A.R. Calawa, W. Paul, and R. H. Rediker,
Appl. Phys. Lett. 7, 206 (1965); J. M. Besson, W. Paul, and A. R. Calawa,
Phys. Rev. 173, 699 (1968).
7. J. F. Butler and A. R. Calawa, Physics of Quantum Electronics, edited by
P. L. Kelley, B. Lax, and P. E. Tannenwald (McGraw-Hill, New York,
1966), p. 458; A.R. Calawa, J. O. Dimmock, T.C. Harman, and
I. Melngailis, Phys. Rev. Lett. 23, 7 (1969).
51
-------
8. E. D. Hinkley, M.I.T. Lincoln Laboratory Solid State Research Report,
' 1968:3, p. 19; T. C. Harman, A.R; Calawa, I. Melngailis, and
J. O. Dimmock, Appl. Phys. Lett. 14, 333 (1969).
9. C. K. N. Patel and E. D. Shaw, Phys. Rev. Lett. 24, 451 (1970); A. Mooradian,
S. R. J. Brueck, and F. A. Blum, Appl. Phys. Lett. 17, 481 (1970).
10. A.R. Calawa, T. C. Harman, M. Finn, and P. Youtz, Trans. Met. Soc.
Amer. Inst. Mining Eng. 242, 374 (1968).
11. E. D. Hinkley, T. C. Harman, and C. Freed, Appl. Phys. Lett. 13, 49
(1968).
12. E. D. Hinkley, Appl. Phys. Lett. 16, 351 (1970).
13. E. D. Hinkley and C. Freed, Phys. Rev. Lett. 23, 277 (1969).
14. J. S. Caring, H. H. Nielsen, and K. Narahari Rao, J. Mol. Spectrosc. 3,
496 (1959).
15. H. Brunet and M. Perez, ibid. ^9, 472 (1969).
16. A. Turk, S. M. Edmonds, H. L. Mark, and G. F. Collins, Environ. Sci.
Technol. 2, 44 (1968).
17. A spectrometer has been inserted in front of the detector for the direct
absorption experiments in order to measure the laser wavelength and
to serve as a band-pass filter permitting radiation in a single longitudinal
mode to be transmitted, while excluding that which may be simultaneously
present in adjacent modes. In a field instrument this refinement is not
necessary.
18. M.C. Teich, Proc. Inst. Elec. Electron. Eng. 56, 37 (1968).
19. R.H. Kingston, personal communication.
20. Although compositions with "x" approaching unity should produce lasers
that emit near 3.7 micrometers at liquid helium temperatures, high
carrier concentrations have, so far, thwarted attempts to produce such
lasers (I. Melngailis, personal communication).
21. L. N. Kurbatov, A. D. Britov, I. S. Aver'yanov, V. E. Mashchenko,
N. N. Mochalkin, and A.I. Dirochka, Soviet Physics Semiconductors,
2, 1008 (1969).
22. We gratefully acknowledge the many technical contributions and continued
encouragement of R. H. Kingston. Appreciation is expressed to T. C.
Harman and A.R. Calawa for providing the Pbi_xSnxTe material and many
of the diode lasers used in this work. We thank J. O. Sample and
F. H. Caswell for laser fabrication and general technical assistance.
52
-------
APPENDIX E
TUNABLE LASER SPECTROSCOPY OF THE i^ BAND OF SO^
E. D. Hinkley, A. R. Calawa, P. L. Kelley, andS.A. Clought
Individual transitions of the v. vibration-rotation band of
SO- in the 8.7-pm wavelength region have been observed using
tunable semiconductor diode lasers, and compared with theo-
retical calculations-based on pure-rotation data from microwave
absorption measurements. A new value of 1151.71 ± 0.01 cm
has been found for the v. band center. Pressure-broadening
coefficients and intensities of individual lines were determined
from the high-resolution data, and a value was obtained for total
band intensity.
The v band of SO, is composed of approximately 3000 relatively strong lines in the 1100-
-1
to 1200-cm (9.1-to 8.3-nm) region. Using thehigh-resolution capability of tunable Pb. Sn Te
C. _ A ~
diode lasers (linewidth less than 10 cm~ ), absorption spectra were obtained for 10 percent of
these lines over various segments of the band. The measured line positions and intensities are
compared with those calculated using rotational constants of Steenbeckeliers derived for both
the (000) and (100) vibrational states from accurate microwave absorption data. In the calcula-
tion, energy levels and eigenvectors were determined for rotational quantum numbers J to 60,
and the rotational partition sum was evaluated and found to be QR = 5805 at 296°K. Transition
frequencies and strengths corresponding to AJ = 0, ±1 and AK. = ±1, ±3 were then evaluated,
yielding positive identification of over 95 percent of the observed lines.
In order to produce diode lasers which emit between 1100 and 1200 cm" , single crystals
of Pb Sn Te were grown at three different compositions: x = 0.0575, 0.0700, and 0.0745.
The crystals were cleaved into rectangular parallelepipeds (each containing a p-n junction) of
average size 0.084 x 0.025 X 0.020cm. Fine tuning of the wavelength within a single longitudinal
mode was achieved by changing the diode current, which alters the refractive index of the semi-
conductor through heating of the junction region. Each of the six diode lasers used for this
study exhibited continuous power in the 6- to 44-|j.W range at 500A/cm . Absorption spectro-
scopy was performed by transmitting the tunable laser radiation through a 30-cm-long cell
containing SO,, using techniques described earlier. The SO, was obtained from the Matheson
Company, Inc. of East Rutherford, New Jersey, with a quoted purity of 99.98 percent. Gas
pressure was measured with a capacitance manometer below 1 Torr, and with a Wallace-
Tiernan gauge at higher pressures.
-1 -1
Figure E-l shows 1.4-cm -wide scans, between 1156.3 and 1157.7 cm , of SO, at pres-
sures of 0.1, 1, and 10 Torr, in addition to one of NH. which yielded two lines for wavelength
calibration. The 10-Torr curve shows definite pressure-broadening and overlap of closely
* Submitted for publication.
TAir Force Cambridge Research Laboratories.
Jin earlier work,1 a single PbSnTe diode laser was temperature-tuned over the entire v. band
of SO,; however, individual transitions were not observed.
53
-------
DIODE CURRENT (A)
2.0 2.1
1157.0 1157.2 1157.4 11576
WAVENUMBER (cm )
Fig.E-1. Diode laser scans of SC>2 at pressures of 0, 0.1, 1, and 10 Torr at
room temperature with a 30-cm cell. The NHj lines were taken at 1-Torr
pressure. Diode laser excitation (and tuning) current is shown by the upper
horizontal scale. Eighteen of these SC>2 lines have been identified: some of
the strongest are 182 16 ^ 18d 17 (1156.523 cm"1);
cm"1); 22
3
2Q
(1157.442 cm"1); 24
100,10~91,9
3,21
24
"1
2
(1157.455 cm").
spaced lines which are resolved at lower pressures. (Absorption at 0.1 Torr is difficult to see
because of the high noise level of this particular set of scans.) Individual SCL transitions were
identified by comparing their relative positions and intensities with theory. Using the latest
values for NH, line positions in this region to calibrate the more than twenty segments covered
-1 -1
between 1110 and 1170 cm , we obtain a value of 1151.71 ± 0.01 cm for the v. band center
frequency of S O,, which can be compared with 1151.38 cm of Shelton, Nielsen, and
/ ^
Fletcher.
Figure E-2 represents a detailed comparison in the 1147-cm~ region between results of a
diode laser scan and the theoretical intensities and positions of lines in this region. The curves
are Lorentzian fits to the intensities, incorporating a linewidth which best reproduces the shape,
of the experimental curve at 1-Torr pressure. (At this pressure, the SO line profile is actually
a convolution of Gaussian and Lorentzian lines whose widths are comparable.) Minor differences
between the two curves are indicated at a, b, and c, where transitions involving higher values
of J and Kfl (relative to adjacent lines) are noticeably shifted from their calculated positions.
A
At d is a line which is not predicted; this may be due to a "hot" band transition, other isotope,
or weaker transition not included in the calculation.
The very high resolution made possible by these tunable lasers permits essentially undis-
torted measurements of individual line profiles, so that pressure-broadening studies can be
carried out by direct linewidth measurements. Figure E-3 illustrates the effects of increased
pressure on the widths of two SO, lines. At pressures above a few Torr, self-broadening of the
15
4,12
15, ,, line occurs at a rate of 20 ± 3MHz/Torr for the half-width at half-maximum.
54
-------
EXPERIMENT
(from diode loser scan)
JUL
THEORY
1147.1
WAVENUMBER (cm"')
Fig. E-2. Comparison between measured and calculated absorption coefficients
and positions for some lines in the i^ band of SO2. These scans were computer-
generated using Lorentzian profiles. Lines designated as a, b, and c represent
the transitions 29, ,. 28_ - ., 35? _g 34g _,, and 26. 22 2?3 25> respec-
tively. Line d is not predicted by theory.
K
10 L
1 1 1 1 1 II
SELF-BROADENED
(20± 3MH2/Torr)
|H-5-3609-4|
AIR-BROADENED
(6.0±0.6MHz/Torr)
80,8-8,,7
10
PRESSURE (Torr)
Fig. E-3. Self-broadened and air-broadened dependence of the half-width at
half-maximum intensity for two lines at room temperature: 15. ,- 15, .,
at 1163.297 cm"1, and 8- Q ^8. _ at 1148.8-10 cm'1. 4'1/J J'1:S
U, o 1, i
55
-------
The effect of air-broadening is shown for the 8_ 0 8. _ transition, where the rate is 6.0 ±
U, o 1,I
0.6 MHz/Torr, or 3.3 ± 0.6 times smaller than for self-broadening of the other transition.
Similar measurements on other lines, do not show any large deviation from the rates shown
here. Our measured ratio between self-broadening and air-broadening rates can be compared
with a value of 5 ± 1 obtained by Burch, et al., using a broadband technique and dry nitrogen
instead of air.
The intensities of several rather isolated SO_ lines were determined from their peak ab-
Li
sorption constants and linewidths at a pressure of 10 Torr. Table I lists the transitions, their
measured intensities, and projected values for the total v band intensity obtained by dividing
the measured intensity of each line by its calculated fractional contribution to the band. Aver-
aging these values yields a total v band intensity of 358 ± 20 X 10~ cm"1 molecule" cm ,
which is comparable with a value of 371 ± 20 X 10 cm molecule cm recently obtained
by Burch, et al, and within the range quoted by earlier workers, ~ employing the integrated
band technique. Using our result, the effective vibrational transition moment for the v. fun-
damental is found to be
| (j^. | = 0.086 ± 0.003 Debye
TABLE I
MEASURED INTENSITIES OF 32S16O2 TRANSITIONS
AND PROJECTED v BAND INTENSITY
Transition*
T 1 T II
If ' T^ ' T^" l» If "
KA'KC *~ KA'KC
30
23
18
13
22
80
60
21
51
14
10,20 *~ 3011,19
5,19 ~246,18
6,12 ~197,13
7,7~148,6
9,13 ~~2210,12
,8*~81,7
,6*~61,5
,1-20,2
,5 *~ 0,4
14.
A Al x^n A A
Frequency
(cm"1)
1115.031
1116.962
1117.038
1117.128
1118.956
1148.810
1149.346
1153.488
1156.195
1156.842
Measured
Intensity
- 1 - 1 2
(cm molec cm )
0.66 x 10"21
1.44
1.67
1.96
1.04
2.96
2.74
1.43
1.86
2.51
Projected Total
Band Intensity
(cm molec cm )
351 x 10
392
356
367
354
332
336
368
384
336
-20
* Conventional spectroscopic notation is followed, where the upper state is
listed first.
56
-------
ACKNOWLEDGMENTS
We are pleased to acknowledge the help of W. F. McBride
and A. Wilson in diode laser fabrication, J. O. Sample and
T. E. Stack for performing the many spectroscopic meas-
urements. Appreciation is also expressed to Professor
K.N. Rao and his associates at the Ohio State University
for prepublication values of the NH^ line positions.
REFERENCES
1. P. Norton, P. Chia, T. Braggins, and H. Levinstein, Appl. Phys. Lett.
18, 158 (1971).
2. G. Steenbeckeliers, Ann. Soc.. Sci., Bruxelles 82, 331 (1968). Calcula-
tions were performed in the manner of W. S. Benedict, S. A. Clough,
L. Frenkel, and T. E. Sullivan, J. Chem. Phys. 53, 2565 (1970).
3. A.R. Calawa, T. C. Harman, M. Finn, and P. Youtz, Trans. Met. Soc.
AIME 242, 374 (1968).
4. E. D. Hinkley, Appl. Phys. Lett. 16, 351 (1970). See also E. D. Hinkley
and P.L. Kelley, Science 171, 635 (1971).
5. K.N. Rao, J. B. Curtis, and P. Yin (private communication). The NHj
transitions are identified using the notation of J. S. Garing, H. H. Nielsen,
and K.N. Rao, J. Molec. Spectry. 3, 496 (1959).
6. R.D. Shelton, A.H. Nielsen,,.and W. H. Fletcher, J. Chem. Phys. 2_1,
2178 (1953).
7. D. E. Burch, J. D. Pembrook, and D. A. Gryvnak (prive communication).
8. J. Morcillo and J. Herranz, Anales -real Soc. espan. fis. y quim. A 52,
207 (1956). '
9. J. E. Mayhood, Can. J. Phys. 35, 954 (1957).
10. D. F. Eggers, Jr. and E. D. Schmid, J. Phys. Chem. 64, 279 (I960).
57
------- |