EPA-R2-73-218
May 1973 Environmental Protection Technology Series
Development of In-Situ
Prototype Diode Laser System
to Monitor SC>2
Across the Stack
Office of Research and Monitoring
U.S. Environmental Protection Agency
Washington. DC 20460
-------
EPA R2 73 218
Development of In-Situ
Prototype Diode Laser System
to Monitor SC>2
Across the Stack
by
E. D. Hinkley
Massachusetts Institute of Technology
Lincoln Laboratory
Lexington, Massachusetts
Contract No. 68-02-0569
Program Element No. 1A1010
EPA Project Officer: John S . Nader
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND MONITORING
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
May 1973
-------
This report has been reviewed by the Environmental Protection Agency and
approved for publication. Approval does not signify that the contents
necessarily reflect the views and policies of the Agency, nor does
mention of trade names or commercial products constitute endorsement
or recommendation for use.
11
-------
CONTENTS
Section Page
1 INTRODUCTION AND SUMMARY 1
1.1 Background 1
1.2 Present Effort 2
2 SYSTEM DESIGN 3
2.1 Sulfur Dioxide Absorption Spectrum 3
2.2 Infrared Detector and Filter Assembly 5
2.3 System Operation 7
3 DIODE LASER DEVELOPMENT 9
3.1 Crystal Growth 9
3.2 Electrical Contacts 10
3.3 Preparation of Laser Cavity 10
3.4 Summary 10
4 CHARACTERISTICS OF DIODE LASERS
USED IN MONITORING SYSTEM 11
4.1 Possible Monitoring Techniques 11
4.2 Properties of Diode Lasers Used in System 12
5 LABORATORY TESTS 17
5.1 Sensitivity 17
5.2 Linearity 19
5.3 Interferences Due to Smoke, Water Vapor, and Steam 19
5.4 Interferences Due to Other Gases: NH3, C2H4 23
6 FIELD TESTS 24
6.1 Helium-Neon Laser Transmission Across
an Oil-Burn ing Stack 24
6.2 In-Stack Measurements (Oil/Natural Gas)
Using Diode Laser 27
6.3 On-Line SO2 and Transmission Measurements
at a Coal-Burning Power Plant 31
7 RECOMMENDATIONS FOR FUTURE STUDIES 33
7.1 Temperature Dependence 33
7.2 Interferences from Other Gases 33
7.3 Laser Tuning and Directionality 34
Acknowledgments 35
References 35
Appendix A: Operating and Circuit Details 37
Appendix B: Tunable Infrared Lasers and Their Applications to Air
Pollution Measurements 54
111
-------
The work reported here was performed by:
E. D. Hinkley
J. O. Sample
L. B. McCullough
A. R. Calawa*
T. C. Harman*
J. P. McVittie*
J. N. Walpole*
L. J. Belanger''
A. E. Paladino*
J. H. Boghos*
S. Duda*
J. N. McMillan*
* Denotes part time.
IV
-------
DEVELOPMENT OF IN SITU PROTOTYPE DIODE LASER SYSTEM
TO MONITOR SO2 ACROSS THE STACK
1. INTRODUCTION AND SUMMARY
EJventual application of tunable lasers in air pollution monitoring has been significantly
advanced by recent developments both in laser technology and in the realization that certain re-
mote sensing techniques can become more sensitive and specific when tunable lasers are used.
Semiconductor diode lasers have been in the forefront of these advances because of their ability
to match strong infrared spectral lines of the known pollutant gases, their simplicity of design
and operation, and prospects for low ultimate cost.
This program is an outgrowth of an earlier experimental effort which investigated the po-
tential of tunable diode lasers for point sampling, in situ source monitoring, ambient air moni-
toring, and single-ended, passive heterodyne detection. In this report are described the develop-
ment and operation of an experimental diode laser system for the monitoring of sulfur dioxide
(SO2) in power plants by across-the-stack in situ measurements. Data related to the particulate
loading are recording simultaneously. The infrared laser and detector are cooled with liquid
helium for this research prototype. Analog readout of the SO- concentration and percent trans-
mission of laser energy across the stack is recorded on a strip chart; digital readout of time,
SO2 ppm, and percent transmission is performed by a digital clock, panel meter, and printer.
1.1 Background
Semiconductor diode lasers can be tailored to emit in desired wavelength regions by control
of the energy gap - a feature made possible by the development of ternary semiconductor com-
pounds of adjustable chemical composition. The different materials from which semiconductor
lasers have been made are shown in Fig. 1. Complete coverage of the region between 0.63 and
34 nm is now possible. The dashed lines indicate possible future extension of the present limits
(solid lines). Most of the tunable diode laser work has been performed with the lead chalco-
genides Pb1_xSnxTe (between 6.5 and 34 nm), PbSi_xSex (between 4.0 and 8.5 (Jim), and Pbj_xCdxS
WAVENUMBER (cm)''
I 1 1 1 [ 1 1 1
|l|-S-J24fi-ll
In, Ga As >.
cisBSi,.t i— »
i i i 1 i i 1 1
M I 1 | 1 1
Pb
Pb Sn
PbV,Si. "
HCl
CH, .
M ,1,
1
in St r-
^—^^*
Pbi-.c*,s
IMNO,.
" N02'
J-MO|
1 I ill I
1 1 1 1 | I 1 1
G,r,
MjS.NjO
1 1 1 1 1 1 t 1 1 1
WAVELENGTH (
Fig. 1. Wavelength ranges which can be covered with alloyed semiconductor
lasers. The arrows indicate possible further extension beyond present lim-
its. Some of the gases of environmental concern are also shown.
-------
(between about 2.5 and 4 jun). Principal absorbing wavelengths of some of the important pollutant
gases are also indicated, at the bottom of Fig. 1. It should be noted that with these three semi-
conducting compounds, nearly all of the important pollutants can be detected.
Figure 2 is a drawing of a semiconductor diode lasei in its standard package, approximately
the same size as an average transistor. The diode crystal, nominally 0.12 X 0.05 X 0.03 cm, is
mounted on a copper stud, which serves as one electrical contact. A silver ribbon serves as the
other contact. Laser emission is produced by passing a current through the diode; this current
can be supplied by a small battery or DC power supply for CW operation, or by a pulser for
pulsed operation. There are several ways to "tune" a diode laser, the simplest being to change
COPPER STUD
CLEAVED FACE
SILVER RIBBON
DIODE LASER
COPPER STUD
Fig. 2. Semiconductor diode laser mounted in its standard package.
Over-all package size is approximately 1 cm,
tl.u magnitude of an applied direct bias current, which changes the junction temperature through
heating. Since the refractive index within the laser cavity is temperature-dependent, the laser
wavelength changes. Although this type of tuning is thermal, it is still relatively fast because
of the limited volume involved. Modulation frequencies of several hundred Hertz can be applied
before thermal inertia becomes very noticeable, and useful experiments have been carried out
with frequencies as high as 10 kHz.
1.2 Present Effort
This program is a culmination of the previous year's effort of broader scope, during which
point sampling, source monitoring, and ambient air monitoring by resonance absorption were
investigated, as well as remote heterodyne detection.3 Diode lasers were developed to operate
in the 1100-1200 cm"1 region corresponding to the t^ band of SO2; its fundamental infrared band
structure parameters were measured and the results published.
For this year's effort, new lasers were fabricated for a selected narrow wavelength region
(1120-1130 cm'1) appropriate for source monitoring of SC)^, i.e., where the line strengths are
strong, but relatively independent of temperature. Improvements during the year resulted in
diode lasers with power levels more than ten times greater than the best achieved earlier. Var-
inu.s monitoring schemes were considered, and one was selected which involved a single laser
-------
operating pulsed at two different wavelengths approximately i cm apart, where the tuning of
alternate pulses was accomplished by a third controllable current pulse of relatively long dura-
tion. System linearity was confirmed over a wide range of SO, concentrations, and the system
response was also checked for day-to-day reproducibility. Measurements of interference rejec-
tion were made by using varying amounts of smoke, water vapor, steam, ammonia (NH3), and
ethylene (C.H.). As the program progressed, in-stack experiments were performed several
times at an oil-burning power plant in order to produce information useful for final system de-
sign. At the end of the program, measurements were made at a coal-burning power plant, and
the resulting SO, determination was compared with data obtained with a point-sampling monitor.
3. SYSTEM DESIGN
2.1 Sulfur Dioxide Absorption Spectrum
The band model and spectral line broadening parameters determined by diode laser spec-
troscopy in last year's program ' were used to generate theoretical spectra for the entire v^
band of SO, in order to determine the optimum wavelength region for detection at temperatures
to be expected in a smokestack. In Fig. 3 is a set of computer-generated scans incorporating
nearly 1000 fundamental lines of SO,, each of which had a Lorentzian profile with 0.28 cm"
full-width-at-half-maximum intensity, which corresponds to atmospheric SO,.
a
o
z
o
t-
Q.
o:
O
(n
CD
o
UJ
_i
o
0,100,200,
300 and 400"C
WAVENUMBER (cm"1)
Fig. 3. Computer-generated atmospheric SO2 absorption spectrum
of a portion of the v^ band. The assumed linewidth was 0.28 cm~l
for all temperatures.
In the region of 1126 cm there is a crossover in the temperature dependence of the absorp-
tion coefficient per molecule, so that for monitoring applications the relative temperature inde-
pendence makes the 1125.8 to 1126.8 cm" wavenumber difference from "on" to "off" absorption
an attractive choice. Not only would the actual in-stack measurements be relatively independent
of gas temperature fluctuations with time, but of temperature variations across the stack as well.
-------
WAVELENGTH
S 5
20 Torr S02
IN I ATMOSPHERE AIR
~27,000ppm)
!0-cm CELL
SPECTROMETER RESOLUTION
02cm'1
1200 1150
WAVENUMBER (cm"')
Fig. 4. Passband of infrared detector/filter combination used in stack-monitoring system.
PM- 10 X 10*'» (815 - 909 urn)
ELECTRICAL BANDWIDTH lOHl-IMHl
NET LOAD IMPEDANCE 212 kohm
t e 10 12
DETECTOR BIAS (V)
Fig. 5. Variation in infrared detector response with bias voltage.
-------
Detection by resonance absorption is based upon the Beer-Lambert equation
p r PO exp[-a^cL] exp[-0L] , (1)
where pQ and p are the transmitted and received laser power, respectively, cv' is the absorption
coefficient per ppm SO2> c is the average SO. concentration along the path L, and /3 is a term
defining extinction caused by turbulence, dust particles, and other substances which may be
present.
If PJ is the received power at wavelength \^, and p2 the power at X-, then the ratio of trans-
mitted power at the two wavelengths becomes
p.j/P2 = expf-faj - a2) cL] , (2)
as long as the extinction parameter 0 is independent of wavelength between X. and A,.
Since the infrared detector (to be described in the next section) is a photon counter, its out-
put voltage is proportional to laser power, i.e., v./v- = p./p.. If the ratio signal is analyzed
by a logarithmic converter, then the output voltage is proportional to the pollutant concentration:
Vg = Iog(vt/v2) = a'cL , (3)
where a' (= a'^ - «2) is the effective absorption coefficient for SO,.
2.2 Infrared Detector and Filter Assembly
The Ge:Cu detector of dimensions 2 mm x 2 mm x 3 mm was attached to the "cold-finger"
of a liquid helium Dewar. In order to reduce the unwanted background radiation, a narrow-band
filter was inserted in front of the detector. It is at about the same temperature as the detector
(100K).
The 1-inch-diameter filter has 85^ transmission between 8.112 and 9.032 |om (1107-1023 cm ),
and less than 0.1% transmission outside this region (to 13 jim). Characteristics of this detector/
filter combination are shown in Fig. 4, which is a spectrometer scan over the region of interest.
The passband is seen to be approximately 1 \an. For reference, a 10-cm-long cell containing
27,000 ppm SO, in one atmosphere of air was introduced in front of the spectrometer. This pro-
duced the SO2 absorption spectrum noted in the figure.
Detector responsivity measurements were made with a calibrated blackbody source at 600°C,
0
producing 3.0 x 10 W within the passband of the infrared filter. Using a net external load of
212 kilohms (including the 100-kilohm detector impedance, the total load is 68 kilohms), the sig-
nal voltage, noise voltage, and signal-to-noise (S/N) ratio were measured, as shown in Fig. 5.
Optimum operation is seen to occur at a detector bias of around 10 volts, where S/N is a max-
imum. The responsivity at 10 volts bias is calculated to be 7.3 x 10 V/W. Since the total load
is 68 kilohms, this may also be written as 1.1 A/W. These values for responsivity can be used
to calculate the infrared power just outside the entrance window of the detector Dewar, within
the passband of the filter. In detector noise calculations where power at the detector surface
must be calculated, the responaivity is increased in inverse ratio of the products of the trans-
mission of the optical elements in the Dewar: namely, the BaF, window (0.8) and infrared filter
c
(0.85), producing an effective "internal" responsivity of 1.07 x 10 V/W (1.6 A/W). There is also
a cold aperture in the Dewar which restricts the field of view to about 18°, or f/3.2 optics. At
a detector bias of 10 volts, the noise voltage in a 1-MHz electrical bandwidth is approximately
- 9
0.5 mV. Thus, the noise-equivalent-power is 6.8 x 10 W.
-------
|l8-5-3770-l|
R«tror»fl«ctor (rr)
Calibration Cell
Smoktstock
i Shutter
Beora-Splittsr
I I
I I Shutter
I I
I I
(movable)
___/lu^ Infrared
\Is Detector
Fig. 6. Schematic of diode laser system for across-the-stack monitoring.
SYNC
CURRENT
PROGRAMMER
[_._.._
i
I
--- ----- [
INTEGRATE- |_
4NO-
HOLD CIRCUITS {•
rV-
LASER
Fig. 7. Block diagram of stack-monitor operating components.
-------
2. 3 System Operation
In Fig. 6 is shown a schematic diagram of the diode-laser stack monitor. Operationally, a
short (~2 usec) pulse of current is supplied to the diode laser from the C.'urrent Programmer,
causing the laser to emit radiation "on" an absorbing region of SO.,. The radiation passes
through the beam-splitter and smokestack to a retroreflector situated on the opposite side of the
stack; it is then reflected backward toward the beam-splitter, and onto the infrared detector.
As shown in Fig. 7, the signal from the infrared detector is integrated and maintained constant
so that a steady signal is fed into one side of the Logarithmic Converter. A few milliseconds
after the first pulse, a low-amplitude (less than 500 mA) "tuning" current is applied to heat the
junction region of the diode laser and shift its wavelength "off" the SO- absorption region. At
the end of this "tuning" current, a second pulse is applied to the diode, whose beam traverses
the same optical path as the first one. This detector signal is measured by a second Integrate-
and-Hold circuit, which represents the second input to the Logarithmic Converter. The diode
laser Current Programmer, Integrate-and-Hold circuits, and Logarithmic Converter have been
constructed using conventional operational amplifiers, and are described in detail in Appendix A.
Figure 8 is a photograph of the entire system, consisting of the main optical table (an tripod),
electronics console, analog recorder, and retroreflector (on second tripod). The tripods can be
raised to a height of 12 feet above floor level. In Fig. 9 is shown a close-up view of the optical
table with its plexiglass protective cover removed. Clearly visible are the tail of the diode laser
Dewar and its collimating lens of BaF., an IRTRAN IV beam-splitter (80:20), the infrared de-
tector Dewar and its associated lens and preamplifier, and a 10-cm-long reference cell with
HORIZONTAL SCALE: 02 V/div
VERTICAL SCALE 50 ma/div (0.05 mA/div expanded )
Fig. 8. Photograph of diode laser stack monitor, consisting of optical table
(on tripod), electronics console, analog recorder, and retroreflector.
-------
Fig. 9. Detailed view of optical table, consisting of laser and detector dewars,
He-Ne alignment laser, calibration cell, beam-splitter, shutters, retroreflectors,
and amplifier.
-------
BaF_ windows, the He-Ne alignment laser, the beam-blocking shutters, and the retroreflector
motor. The positions of the shutters and retroreflector are controllable from the main console.
The operating procedure is given in detail in Section I of Appendix A. Also contained in
Appendix A are the Electronic Circuits and Timing Sequence (Section II), Electronic Alignment
Procedure (Section III), and Cryogenic Dewar Preparation (Section IV).
3. DIODE LASER DEVELOPMENT
Single crystal Pb, Sn Te material with composition x near 0.075 was required for diode
lasers emitting in the 1127 cm region where the absorption spectrum of SO- is relatively inde-
pendent of temperature. We discuss here various aspects of the crystal growth, material proc-
essing, and device fabrication employed during the year for optimizing the performance of
Pb. Sn Te diodes near this composition.
3.1 Crystal Growth
Most of the growth runs were carried out using a new closed-tube, horizontal vapor growth
technique. The method involves vapor growth from a metal-rich source material on Fbj_xSnxTe,
rather than on quartz walls as in our previously used methods. This growth technique has pro-
duced high-quality, mm-size crystals of low dislocation density. The technique (as with our
earlier methods) allows growth to occur under conditions which produce p-type material with a
concentration which varies with the growth temperature for temperatures above 775°C; however,
because of the large diffusion coefficients at these high temperatures, the crystals grown from
a metal-rich source even at 800°C are converted to n-type on cooling. By raising the growth
temperature to 825-8SO°C, the p-type concentration of the as-grown crystals is increased and
n-p junctions are formed on cooling. For growth temperatures less than 825°C, it is necessary
to dope the source material with thallium (TJ ), an acceptor, in order to obtain p-type material
during growth and form the junction on cooling.
Although variation of growth temperature allows some control of the bulk p-type concentra-
tion, the concentration profile in the junction region is not accurately known and perhaps not as
controllable as might be desired. Using junction capacitance measurements, we have found that
diodes with lower threshold current densities have smaller capacitances than diodes with high
thresholds. These results suggest that lower carrier concentrations near the junction region
are advantageous.
In order to determine more accurately the effects of carrier density on laser performance,
some annealings and subsequent diffusions of junctions at lower temperatures after the growth
have been employed. We have found that the p-type concentrations which can be obtained by
isothermal annealing on the Te-rich side of the stability range are apparently too large.
These concentrations are generally >1019 cm"3. In order to obtain lower p-type concentra-
tions, a two-zone annealing technique is required. This technique involves varying the vapor
pressure of Te over the heated sample by placing Te in a second-temperature zone. Two-zone
annealing has been used for all of the binary lead salts, but the time and temperature param-
eters were not known for the ternary alloys. Several annealing experiments were performed on
Pb. Q25Snn 075Te crystals. and tne results showed that p-type carrier concentrations in the
range of 2.3 x 1018 cm"3 to 1019 cm"3 can readily be achieved. However, lasers have not yet
been made from two-zone annealed material.
-------
Fabrication of lasers has continued during these annealing experiments using the junctions
formed on cooling from the growth temperatures. The first lasers were made using Tt doping
J rt - 3
of the source {~10 cm ) and relatively low growth temperatures (~750°C). Threshold current
densities obtained were quite good, typically 200 mA/cm . However, the output power was
18 -3
relatively low - only a few microwatts. Since the presence of approximately 10 cm of a
foreign impurity such as Tt may adversely affect laser performance, we have in recent months
worked with undoped crystals.
3.2 Electrical Contacts
In addition to preparing semiconductor material for laser fabrication, an effort was made to
reduce the contact resistance between the laser chip and the metal layers for contacts. This
contact resistance was approximately three times that of the semiconductor itself (0.1 ohm,
compared with 0.03 ohm for the average diode laser), and resulted in a significant power loss in
the form of heat. Although some of this heat may be useful for current-tuning, it must be con-
trolled to prevent thermal destruction of the laser. Most of the resistance is due to the contact
to the p-type semiconductor material, for which both gold and thallium metals have been used.
The lowest contact resistance (approximately 0.01 ohm) was achieved by alloying thallium to the
p-type semiconductor; however, reproducibility was poor. Alloyed or plated gold contacts pro-
duce the most consistent ohmic contact to p-type FbSnTe, for which the lowest contact resistance
has been about 0.05 ohm. Because of simplicity and reproducibility, plated gold contacts were
used for lasers in this program.
It is important to realize that the various factors discussed here are interrelated. Alloy
composition, contacting procedures, and carrier density are related in two ways, in particular,
which are very important. First, variations in contacts and carrier density may affect heat pro-
duction and/or heat sinking. These in turn affect the operating temperature of the device and
consequently may require a slight adjustment in alloy composition to achieve the precise laser
wavelength desired. Second, as the p-type concentration is lowered to obtain better thresholds,
acceptably low electrical contact resistance becomes increasingly difficult to obtain.
Because of these problems, often diodes with the lowest thresholds do not emit at the opti-
mum laser frequency or they may be limited in maximum output power by excessive heating at
the contact. We believe this major difficulty can now be avoided in diodes of composition near
x = 0.075 by employing a technique recently developed in diodes with higher SnTe composition.
This technique involves diffusion after growth of a thin p layer at the surface, which greatly re-
duces contact resistance without affecting the carrier concentration near the junction region.
3.3 Preparation of Laser Cavity
Another important advance in the fabrication technique made during the last weeks of this
contract year involves polishing, rather than cleaving, the reflecting faces of the diode laser to
obtain a much improved mirror quality. Up to an order of magnitude improvement in pulsed
output power has been obtained by this method in otherwise identical lasers. The polishing tech-
nique has only been used on the last few devices supplied for the program.
3.4 Summary
Some test results on diodes used in the SO- monitoring system are summarized in the follow-
ing section. These devices reflect the improvements in maximum output power which have been
10
-------
obtained at the optimum laser wavelength for SO- absorption. Advances in fabrication techniques
recently developed for other Pb|_ Sn Te compositions should result in further improvements in
laser performance at this composition as well.
4. CHARACTERISTICS OF DIODE LASERS USED IN MONITORING SYSTEM
The versatility of semiconductor diode lasers permits several techniques to be considered
for pollutant monitoring applications. Some of these are listed below, with comments on their
suitability for in-stack monitoring.
4.1 Possible Monitoring Techniques
4.1.1 CW Laser Operation Using First-Derivative Detection
This technique is identical to thatused for spectroscopic analysis of point samples at reduced
- 4 -1
pressure, with ultimate resolution limited only by the laser linewidth of around 1 x 10 cm
However, the power dissipation for continuously operating diodes is normally several hundred
milliwatts, which produces a substantial liquid helium "boil-off," and concomitant reduction in
Dewar hold time.
4.1.2 Pulsed Operation with Two Diode Lasers
Pulsed techniques with average power levels two or more orders of magnitude below those
for CW operation can reduce considerably the thermal drain on the liquid helium reservoir. Di-
ode currents of several amperes can be used to produce high peak powers. One laser is tuned
to a strongly absorbing region - the other to a region having little SO- absorption. Although this
approach would be simple to instrument electronically, and the two diodes could be situated close
together in the same Dewar, the optical alignment of the two laser beams onto a single detector
would be somewhat difficult.
4.1.3 Pulsed/Cur rent-Tuned Operation with a Single Diode Laser
This technique employs two high-current pulses separated by a tuning current, and is the
one selected for the research prototype developed in this program. Initial experiments showed
that the wavelengths of diode lasers under pulsed operation could be tuned at the same rate as
for CW operation by the application of a steady current, or a pulse of long duration below thresh-
old for laser action. For the stack-monitoring system, laser emission occurs when either of the
high-amplitude pulses I or I- is applied. Between the two pulses is a "tuning" current I of
smaller amplitude (usually below threshold) but longer in duration. The sequence repeats after
pulse 2, during which time the diode temperature reverts to its original value. The sequence of
pulses is shown in Figs. A-5 and A-7.
The wavelength A. of radiation emitted by pulse 1 (I. ) is determined by the ambient temper-
ature of the diode laser, its chemical composition, and physical length. The emission at pulse 2
(I_) occurs at a somewhat shorter wavelength \2> where the difference is a function of the diode
temperature reached upon application of the tuning current. There is some shift in wavelength
during I. and I_ at a rate determined by the constant resistance, bulk electrical resistance, and
thermal conduction of the interfaces between the laser and the liquid helium. Pulses are there-
fore kept as short as possible (less than 2 usec), consistent with a reasonable signal from the
infrared detector circuit, which has a time constant of 18 (isec.
11
-------
Measurements were made of the pulse-to-pulse (p-p) reproducibility of laser output power
by employing a current generator with 0.1% p-p repeatability. In general, the output power from
the diode lasers was found to have better than 1% p-p reproducibility, which should be quite
adequate for the stack-monitoring system, particularly when signals from several pulses are
averaged.
4.2 Properties of Diode Lasers Used in System
Four diode lasers were incorporated into the laser Dewar, each emitting from one of four
mutually orthogonal ports. This redundancy provides for three spares and also permits com-
parative measurements to be made. Some of their properties are given in Table I.
TABLE I
Diode
355
376
377
379
Threshold
Current
(mA)
800
1300
1200
860
Nominal IR
Frequency
(cm'1)
1121
1121
1122
1130
Total Output
Power
(mW)*
0.2
1.5
0.6
0.7
Absorption Due
to 400 Torr-cm SO2
at It = Of
(percent)
40
32
28
28
* Thermopile measurements at 2 Amp currents.
t20 Torr SO2 in 740 Torr air (26,300 ppm); 20-cm path length.
Three of the above devices, 376, 377, and 379, were fabricated using the new polished-
cavity technique described in the preceding Section 3.3. The end faces of laser 355 were cleaved.
Since the laser is at a higher temperature for pulse 2 than for pulse 1, its emitted power
during pulse 2 is generally less when I. = I,. This is evident in Fig. 10 for these four lasers,
where the output power corresponding to pulses 1 and 2 are plotted as a function of the magnitude
of the tuning current. As expected, the power emitted by I. is relatively independent of tuning
current, since the laser had been able to return to its original temperature beforehand. This is
clearly not the case for I_; consequently, there must be a provision for adjustment of the pulse
amplitudes or widths in order to have identical "untuned" and "tuned" transmissions in the ab-
sence of SO-. For this prototype system, a potentiometer on the front panel controls the dura-
tion of li to permit equalization of signals.
While Fig. 10 represents the output power of the diode lasers for a range of tuning currents,
the ratio of the output powers at pulses 1 and 2 is used to measure the SO, concentration, pro-
vided that the equalizing adjustment has been made at a particular value of L. The effective ab-
sorption coefficient per ppm SO-, a', was determined by alternately filling the 10-cm calibration
cell with 20 Torr SO, in 740 Torr air, and evacuating the cell: The results are shown in
Figs. 11 (a) through (d) as a function of the tuning current. It is important to note that there is a
systematic and reproducible dependence of a' on tuning current from day to day, depending on
the manner in which the laser Dewar was filled with liquid helium (see Appendix A, Section IV).
This difference in "tuning" response between "direct" fill and "precooled" fill is particularly
evident in Fig. ll(b) for diode 376, and presumably arises from a small change in thermal con-
duction between the liquid helium and the Dewar tail using the two techniques of filling. Even
12
-------
Fig. 10. Variation of output laser power with tuning current for the four diodes
used in the stack-monitoring system, p and p2 refer to emitted power by cur-
rents I. and I, (tuned), respectively.
when the same fill procedure is used, there are minor differences in "tuning" response caused
by changes in the thermal properties of the diodes under temperature cycling. Therefore, re-
calibration is necessary each time the laser is cooled from room temperature to its operating
temperature. If the low temperature is maintained, however, the "tuning" response is constant
and no recalibration should be necessary.
From Eq. (3) it can be shown that the effective absorption coefficient per ppm SO. can be de-
duced from the expression
V
S
GcL
(4)
where a', and al correspond to absorption at pulses 1 and 2, respectively, and Vg is the output
voltage from the ratio gain control circuit, as monitored on the analog recorder or digital panel
meter. G is the voltage gain beyond the log converter, c is the SO? concentration in ppm, and
L is the path length in cm.
In selecting an appropriate tuning current for a particular diode laser, a high a' is desired
at the lowest usable tuning current (in order to minimize liquid helium boil-off). There is
another consideration to be made, however, related to the laser power itself. If detector noise
is the determining factor for ultimate sensitivity, as it is in the laboratory, then for two lasers
with the same a', that one with higher power will have a greater detection capability. Recall
from Eqs. (3) and (4) that the output voltage can be written as
Vs,Gln^ij =
C5)
-------
Fig. ll(a-d). Effective SC>2 absorption coefficient vs tuning current for the four system
diode lasers, illustrating variations caused by different LHe fill procedures.
14
-------
LN PRECOOLING
12/14/72
DIRECT LHe FILL
H/30/72,RUN 2
(c)
(d)
DIRECT LH« FILL
11/17/72
LN PRECOOLING
11/16/72
....mlllintliimnn
DIRECT LHe FILL
n/13/72
LN PRECOOLING
II/IO/72
100 200
I, [mil
Fig. 11. Continued.
-------
where V. and V- are the infrared detector voltages due to
detector noise component present in both V^ and V^, and "V
then
V2lV2
and I,, respectively. If V is a
. the output noise due to this term,
K»Vsv
evf)
(6)
Consequently, for a particular concentration c of SO,, a fixed pathlength L, and a detector
noise voltage V , the output signal-to-noise ratio becomes
cL
'N
\-j2V
(7)
where V. _ a V. or V,.
It is clear from Eq. (7) that it is the product a' p which determines
the signal-to-noise ratio under these conditions, since V. , is proportional to the laser power
p . Consequently, that diode having the largest a' p product will have the best SO- detection
capability.
Figure 12 supports the above analysis. In this figure are plotted the signal-to-noise ratios
for six different diode lasers for the detection of 20 Torr SO, in 740 Torr air over a 20-cm path.
The signal-to-noise ratios range from 2.5 (for diode 375 at L = 84 mA)to 108 (for diode 379 at
i i i i i i r
so
10
1 1
i i
~i—rr
/
/
O 376
O 355
A 377
O 375
• 379
• 378
OJ
i i 1
11I-5-43M |
j—i i i i
Fig. 12. Signal-to-noise ratios for six different diode lasers in the detection
of 20 Torr SO2 in 740 Torr air over a 20-cm path. The noise values used
were peak-to-peak fluctuations with 0.3-sec averaging.
16
-------
66 mA). The noise represents peak-to-peak fluctuations with 0. 3-second averaging. If the
rms value had been used, the signal-to-noise values would be 2 •v/T times larger, ranging from
7 to 305.
For actual operation in the field, other (external) factors will probably limit sensitivity of
the system. If the noise term in this case is defined as N ., then the signal-to-noise ratio
becomes
VS Ga'cL , .
V~~ = 1? - • ( '
VN Wext
The laser power is unimportant as long as N „. is substantially larger than the detector noise
contribution, and the absorption coefficient a' is paramount. In other words, lasers with higher
output power levels would not improve the detection sensitivity.
Other considerations, such as possible interferences with other constituents, and linearity
under different conditions, will be considered in Section 5.
5. LABORATORY TESTS
Tests were performed in the laboratory in order to determine sensitivity, linearity, and
interference characteristics of the diode laser system operated according to the procedure de-
scribed in Section 2 and Appendix A.
5.1 Sensitivity
Figure 12 of the preceding section was used to illustrate the laser parameters important
for sensitive detection of SO-. The signal-to-noise data were for the detection of 20 Torr of
SO2 in 740 Torr air (26,300 ppm) over a pathlength of 20 cm. For comparison purposes, the
equivalent concentration for two-way transmission across a smokestack 5 meters in diameter is
526 ppm. For diode 379, the signal-to- noise ratio for detection of this equivalent concentration
is approximately 100 for a 0. 3-second averaging time. Recall that the noise represents a peak-
to-peak variation; consequently, if we assume that a signal-to- noise ratio of 2 is appropriate
for defining the ultimate detection sensitivity (this would correspond to S/N = 5.6 for rms noise),
the minimum detectable concentration of SO, is around 10 ppm for this particular diode, with
0. 3-second averaging. Mathematically, this result may be written as
where T is the integration (or averaging) time in seconds, as set by the selector switch on the
front panel of the electronics console. If very fast response is not needed, a time constant of
10 seconds can be used, such that the minimum detectable concentration becomes 1.7 ppm.
Some final observations should be made with regard to the above analysis. First, under
actual stack-monitoring conditions, fluctuations in the transmitted power at wavelengths \^ and
X_ will occur as a result of turbulence and particulate scattering within the stack. Although such
effects are partially canceled by the ratio technique, they are not eliminated entirely. If the
purpose of this system were to provide the long-term, steady monitoring capability of an instru-
ment ready for the production line, then time constants of many minutes would have been incor-
porated. However, since its main purpose is to investigate a principle involving infrared laser
technique for in situ monitoring, the shorter time constants provide the means of investigating
rapid fluctuations, which may be caused in part by temporal variations in the SO, concentration
17
-------
10,000 20.000
CjH^ CONCENTRATION Ippm)
30,000
Fig. 13. Linearity check of integrator-log converter combination, using a 10.6-nm
diode laser (single-wavelength) and varying concentrations of atmospheric C-H..
18
-------
as well as by the other factors mentioned above. The extent to which Eq. (9) holds for the field
measurements will depend upon these factors, and it must be recognized as an upper limit of
sensitivity for that particular diode laser. The other lasers will have minimum detection limits
proportional to the signal-to-noise values in Fig. 12.
5.2 Linearity
5.2.1 Confirmation of the Beer-Lambert Equation
During the early stages of this program, measurements were made of the pulsed-laser
transmission through a 30-cm-long cell containing ethylene (C-H.) of varying concentrations in
one atmosphere of air, using an available diode laser operating in the 10-pin region. The pur-
pose of this experiment was to verify the equation, p = p exp[-a'cL], introduced earlier, over
a wide range of concentrations, and thereby validate our assumptions on which electronic design
of the prototype SO2 monitoring system was to be based.
Figure 13 shows the relative transmission (I/I ) as a function of C?H4 concentrations from
zero to 32,000 ppm. The straight line is a theoretical fit to the data corresponding to an a' of
a 3.86 x 10 cm /ppm. Using a 0.1-second post-detection time constant, the effective noise
level was found to be 19 ppm for this experiment with a 30-cm cell, and was caused by the detec-
tor noise only.
5.2.2 Linearity for the Two-Pulse (Ratio) Technique
In order to determine linearity of the technique incorporating the ratio of transmissions cor-
responding to two different wavelengths, AI and X.,, of the laser radiation, for concentrations to
be expected in a working smokestack, a diode laser emitting around 8.9 fim was adjusted for op-
timal values of pulse and tuning currents. The effective absorption coefficient a' (= a'. - a' )
- 7 -1
was 2. 3 x 10 cm /ppm SO . Concentrations for a 10-meter-diameter stack were simulated
in a 1-meter cell by increasing the SO- values tenfold.
The result of this study is shown in Fig. 14 for SO. concentrations ranging from 100 to
2000 ppm. The output voltage from the logarithmic converter is seen to be essentially linear
with respect to the SO, concentration, as expected.
In order to demonstrate linearity of the system at a variety of tuning currents for a particu-
lar diode, the above experiment was repeated for laser 379 at tuning currents of 40, 100, 140,
170, 200, and 230 mA, each of which represents a local peak or valley for absorption. The data
for this diode are plotted in Fig. 15 for SO_ concentrations to 500 ppm (along a 10-meter effective
path), and show good linearity at each value of tuning current.
5.3 Interferences Due to Smoke, Water Vapor, and Steam
A specimen of gas from a nearby oil-fired stack was obtained and analyzed in the 730-cm-
long White cell constructed during last year's program. The object was to determine the SO,
concentration and observe any interference effects which may be revealed by a high-resolution
laser scan of the stack gas at reduced pressure. By placing a small modulation current on the
steady diode current, derivative spectra were obtained. The stack-gas scan is shown in Fig. 16,
together with one for pure SO,. By comparing the two traces, we determined that the stack
gas contained 670 ppm SO,. Moreover, it is clear from the spectra that there are no strong in-
terferences over this wavelength region due to any other gases. Although these particular SO.
lines are about 20 times weaker than the strongest lines observed, the noise level is adequate
19
-------
1000
CONCENTRATION
10,000
(ppm)
Fig. 14. Linearity check of ratio (two-pulse) technique for the detection of SO?
in the 8.9-jun region. Equivalent concentrations for a 10-m path were achieved
by a tenfold increase of the SC>2 partial pressure in the 1-m path used for this test.
zo
-------
400
CONCENTRATION Ippm]
Fig. 15. Linearity check of ratio (two-wavelength) technique for the detection of SC>2
at various diode laser tuning currents, from 100 to 230mA, The output voltage is
positive or negative, depending on the position of the nominal (untuned) wavelength
relative to the SO, absorption lines.
21
-------
CALIBRATION TRACE
(0.2 Torr
STACK GAS SAMPLE
at p - 5 Torr
(670 ppm S02)
L = 730 cm
a'p - 0.004 cm'V Torr
Fig. 16. Derivative detection scan for SOj in a specimen of stack gas. Laser
modulation current was approximately 1 mA peak-to-peak.
o
in
"» 60
ATTENUATION DUE
TO SMOKE
-TEST CELL FILLED
(2OOO ppm
CELL EVACUATED -
TIME (mm)
Fig. 17. Interference effect of smoke on 50% determination. The lower
curve is the SO_ (ratio) signal, on an arbitrary linear scale.
22
-------
at 20 ppm. (Had the laser wavelength been in the region of the strongest SO_ absorption line,
the equivalent noise level would be 1 ppm under the same conditions.) The periodic, slowly
varying oscillations throughout the stack-gas scan are caused by reflections between the entrance
and exit windows of the White cell (Fabry-Perot effect).
The lack of interfering gases shown by the above measurement suggest s that water vapor
will probably be the major contributor to absorption and scattering losses, particularly when
scrubbers are being used. Although the 1127 cm region is well-placed between relatively weak
water vapor lines, it is known that at atmospheric pressure the "wings" of such lines can extend
unusually far from the line centers. In order to observe the effects of smoke, water vapor, and
steam on the system, a 20-cm-long pyrex tube (open at both ends) was placed in the diode laser
beam. (For the steam test, the walls were heated above 100°C.) The tube could be filled by a
port at the center, with discharge from both ends. Also along the laser beam was placed a closed
10-cm-long cell filled with a mixture of SO, and air at atmospheric pressure. The effect of
smoke attenuation is illustrated in Fig. 17. The upper trace corresponds to transmission of
pulse 2, whereas the lower trace is the ratio of transmissions at X. and \». No noticeable change
in the SO- reading is produced for up to 60% attenuation of the laser beam.
a.
a.
O
Q
UJ
IE
sf
BOO
700
600
5OO
i
400
300
ZOO
400
1 1" I 1 1 1 1 1 1
~~l'l-5-"'*-^l
_
T-
AO *$ l\
rg a o o ae a a °° . e-s-°^ Ji
0 D 0 Q
-
-
A Smoke
o Water Vopor (T<100«C)
o Water Vapor (T>100*C)
1 1 1 1 1 J 1 1 -L
10 20 30 <0 30 60 70 80
PERCENT TRANSMISSION LOSS
Fig. 18. Interference effects of smoke, water vapor, and steam on SO£.
Attenuation due to any of the above sources can be as high as 65% before there is a significant
change in the SO_ reading. This is illustrated in Fig. 18 where in one case the absorption ranged
from zero to 98%. It is evident that the water vapor interferes as an aerosol (similar to the
smoke), while the steam interference is probably due to infrared absorption.
5.4 Interferences Due to Other Gases: NH,, C_H.
Each of the four diode lasers incorporated into the final system was checked for responsivity
H3 and C,H., in addition to SO,. Since NH, is used as •<
the 8.9-nm region, some interfering regions were expected.
to NH3 and C,H., in addition to SO-. Since NH, is used as a wavelength calibration standard in
23
-------
The results of the gaseous interference tests are shown in Figs, 19(a) through (d). In each
case, the effective absorption coefficient (per ppm) for SO- is indicated by the solid line, for
NH, by the plus signs, and for C-H^ by the dotted line. Regions of tuning current for sensitive
detection of a particular gas are readily apparent. For example, for diode 355 a tuning current
of 18 mA produces an a' of 1 x 10~ cm" /ppm for SO2> with no interference from either NH3
or C,H.. Diode 376 shows better response to NH, than to SO, or C,H. over most of its tuning
Z 4 _ 7 _ .i ^
range, with an absorption coefficient of 8 x 10 cm /ppm at a tuning current of 40 mA. It is
clear that for the conditions under which this diode was operated (pulse width of 3 jisec), SO-
could not be detected in the presence of NH,, unless correlation readings were taken at two or
more values of tuning current. (The NH, interference is positive below 100 mA, and negative
above. ) For diode 377, Fig. 19(c) indicates specific SO2 detection at around 30 mA. Diode 379
is sensitive only to SO, for tuning currents to 140 mA, with an average absorption coefficient of
- 7 — \
2X10 cm /ppm. As indicated in Section 5.1, the combination of high power and strong ab-
sorption coefficient makes this diode the best of the four for monitoring SO2>
The interferences shown in Figs. 19(a) through (d) can be minimized if the laser current
pulse widths are reduced below the value used for these tests (3 jjisec). The amount of laser fre-
quency "chirping" is approximately 0.1 cm per ^isec pulse width, although the exact variation
depends on the particular laser. Ideally, the pulse width should be less than 0.1 (o.sec to detect
atmospheric gases; however, since the detector time constant is 18 jisec, such short pulses pro-
duce very weak signals. In order to confirm this relationship between pulse width and specificity,
these interference experiments should be repeated using different laser current pulse widths.
Interferences from many other gaseous constituents present in smokestack effluent, such
as CO and NO, can be ruled out because they do not absorb around 8.9 fxm. Although the tuning
range of the two-wavelength technique is less than 0.1% of the nominal infrared frequency, there
may be interferences from other constituents; this will have to be checked by further studies.
Ultimately, high-resolution laser scans of stack-gas samples in regions covered by the pulsed
lasers during field operation would afford the best assurance that there are no significant
interferences.
6. FIELD TESTS
6.1 Helium-Neon Laser Transmission Across an Oil-Burning Stack
In order to design an optical system for in-stack monitoring of SO_, it was deemed important
to measure the transmission properties of a visible laser beam through a typical stack (broaden-
ing of image due to thermal convection currents, translation of image position, scattering, ab-
sorption, etc.). A 1-mW He-Ne laser, operating in the visible at 0.6328 (xm, was used to perform
this experiment. The laser beam was directed through a stack at the Air Force Cambridge Re-
search Laboratories having an inside diameter of 2.45 meters. The distance from laser to
ground-glass/Polaroid film camera detector system was 3.2 meters. Photographs were made to
show the image size, degree of sharpness, and position during both quiescent and stack-purging
conditions. The photographs of Fig, 20 were taken with Polaroid 3000-speed film at a camera
setting of f/16 for 0.1 second.
As shown in Fig. 20(a), where the laser beam was transmitted over a 3.2-meter path in the
laboratory, the undisturbed image is sharp, with a diameter of 0.32 ± 0.02 cm. During quiescent
(normal) operation of the stack, it is clear from Fig. 20(b) that some blurring of the image has
occurred (presumably due to temporal variations in refractive index caused by convective currents
Z4
-------
DIODE 355
TUNING CUHREMT (mAI
(a)
DIODE 376
TUNING CURRENT (ma)
(b)
Fig. 19(a-d). Interference of other gases: Sensitivity of four system diode lasers
to SO^, NHj, C^H^ for various tuning currents.
25
-------
,00 200
TUNING CURRENT (mA)
(C)
g
"o -z
DIODE 379
100 ZOO
TUNING CURRENT (mA)
(d)
Fig. 19. Continued.
26
-------
IN LABORATORY
Fig. 20. Photographs of He-He laser
beam after being transmitted across
a smokestack 2.45-m wide. Total
beam path was 3.2 m.
ACROSS STACK
(quiescent)
ACROSS STACK
(purging)
in the effluent), and its diameter has increased by 0.11 cm to 0.43 ± 0.03 cm. When the stack
was purged to release collected particulates from the boiler, no significant change was observed
in either the intensity or the size of the laser image, although there was an apparent translation
of the beam by approximately 0.2 cm, as shown in Fig. 20(c). (The "translation" may have been
partially caused by deflections of the walls on which the laser and camera were mounted.)
If we extrapolate these results for a 2.45-meter-diameter stack to one whose diameter is
10 meters, the size increase and position shift for a one-way pass will be about four times larger,
or 0.44 cm and 0.8 cm, respectively. (Since scattering should be significantly less at 8.7 jim
than at the 0.6328-\an wavelength used in this experiment, the effect here is probably larger than
it will actually be.) Assuming such a 0.8-cm translation in the beam (1.6 cm for a two-way path),
the effective deviation in direction from the undistorted beam is approximately 0.05°. A 3.8-cm-
diameter collecting lens with a focal length of 10.6 cm, together with an infrared detector of
0.05-cm diameter, yielding a collecting angle of 0.3°, should be adequate.
6.2 In-Stack Measurements (Oil/Natural Gas) Using Diode Laser
Using a FbQ gjSn. 0?Te diode laser emitting around 8.9 (im, tests were again made at the
AFCRL stack to obtain operational information prior to final system design. The ambient tem-
perature outside the stack where the electronics system was located was high (115 to 120°F), and
the relative humidity was 35%. The fuel was No. 6 grade oil containing 1.43% sulfur, consumed
at a rate of 430 to 460 gal/hr. Two boilers were operational, and the stack-gas temperature
was approximately 470°F (243°C). During the experiments, conversion was made from natural
gas to oil, and back to gas.
27
-------
Single-pass attenuation of the diode laser power by the stack gases and particulates was less
than 10%. However, with the infrared detector situated near one of the ports, infrared emission
from the hot gases produced "noise" several times larger than the inherent detector noise. The
detector voltage shown in Fig. 21 corresponds to three conditions: (a) detector looking through
the stack at the "cold" laser (diode current zero); (b) same conditions as (a), except that the
laser Dewar was blocked from the detector field of view; (c) stack blocked from detector field
of view (detector noise limit). By comparing Figs. 21 (a) and (b), it is clear that the noise contri-
bution due to random fluctuations of the position of the "cold" object (laser package) was small.
Most of the noise was caused by the stack gases themselves, the dominant frequency being ap-
proximately 50 Hz. Because of the infrared filter, the detector responds only to radiation be-
tween 1100 and 1227 cm" (see Fig. 4); consequently, the wideband thermal radiation responsible
for the noise in Figs. 2d(a) and (b) must be within this wavelength region. Since, ideally, the de-
tector noise of Fig. 21(c) should represent the limiting sensitivity, steps had to be taken to reduce
la I THROUGH STACK TO LASER (diode current off")
(c) DETECTOR NOISE ONLY (field of view blocked)
VERTICAL SCALE OSmV/div
HORIZONTAL SCALE' SOmsec'div
BANDWIDTH 8O Hz -40kHz
(3dB points)
Fig. 21. Infrared detector signal during stack monitoring under different
conditions: (a) Through stack to laser, with laser current "off"; (b) through
stack, but laser image blocked; (c) detector blocked so thatit does not "see"
the stack at all. Vertical scale: 0.5 mV/div; horizontal scale: 20 msec/div.
Bandwidth (3-dB points): 80 Hz to 40 kHz.
28
-------
stack-gas "noise" from its value of 1 mV (peak-to-peak) to below the 0.25-mV value of the de-
tector itself. There were several ways to accomplish this:
(1) Reduce optical bandwidth even further.
(2) Use a high-pass electrical filter to eliminate the low-frequency "noise."
(3) Increase system integration time.
(4) Utilize unsymmetrical transmission/reflection characteristics of beam-splitter so
that only a small fraction of the stack-gas emission is collected by the infrared
detector (this is possible for the two-way transmission technique shown in Fig. 6).
The following corrective action was taken to reduce the fluctuations in the infrared detector
signal induced by the stack gases.
(1) A second infrared filter was ordered, which would block transmission above
1142 cm . In conjunction with the 1100-1227 cm" filter already installed, the
combined effect would be to reduce "noise" due to stack gases at 250°C by a factor
of 3.6, and the laser signal by 25%. However, because insertion of this filter in-
volved redesign of the detector package, and because of the uncertainties associated
with transmission through two filters in series, we decided not to incorporate this
new filter unless absolutely necessary.
(2) The coupling capacitor between the infrared detector and preamplifier was reduced
from 1 microfarad to 0.001 microfarad, raising the minimum detectable frequency
to 140 Hz. Since the dominant frequency of the stack-gas-induced "noise" was 50 Hz,
this modification resulted in a 70% reduction; with no effect on the 2-nsec-wide la-
ser pulses.
(3) New circuits were designed and incorporated into the system to permit the selection
of integration times from 0.1 second to 10 seconds at the output stages of both
"transmission" and "SO," readings.
Ci
(4) The optical table was designed to permit the return signal for two-way transmission
to reflect off (rather than be transmitted through) the beam-splitter. As a result,
only 20% of the stack-gas radiation is collected by the infrared detector, with no ef-
fect on the laser pulse.
During these initial field measurements, conversion was made from oil to natural gas. Meas-
urements made during this conversion are shown in Fig. 22, where the upper trace represents the
transmission across the stack, which increased by a few percent after conversion. Although the
diode laser was not optimally tuned to monitor the SO_ concentration during this experiment,
there was a detectable reduction in the SO,, content during the shift from oil to natural gas. The
large spikes probably stem from higher SO. concentrations produced when the gas flame oxidized
unburned sulfur which had collected on the walls of the furnaces.
After making these improvements, the system was assembled for two-pass operation at the
same facility. The optical table was placed on the floor (rather than on a tall tripod at the access
port) and two plane front-surface mirrors were used to direct the laser beam to the port, approx-
imately 12 feet high. Another mirror was positioned at the other side of the stack, and was also
supported by a tripod.
As a result of the improvements made since the first field test with a diode laser, the opti-
cal alignment procedure was simpler and "noise" induced by the stack gases was absent. Without
29
-------
ioo r
S02 CONCENTRATION (arbitrory scale)
Fig. 22. Measurements of transmission and SOj concentration made during
conversion of power plant from oil to natural gas. Stack diameter: 2.45 m.
any adjustments, internal calibration at the stack site agreed with that obtained earlier in the
laboratory for the 1000-ppm internal "calibration" cell, although the "zero" setting was 60 ppm
high. It should be noted that during the one and a half hour period between laboratory checkout
and assembly at the field site, the laser and detector Dewars were dismantled, electronics de-
activated, and transport of the components carried out.
With the function switch set to the "Monitor" position, the laser beam was directed from
floor level to the mirror at the access port to the stack (a distance of 12 feet), through the
2.45-meter (8-foot) stack to the reflecting mirror, back through the stack, and down to floor
level again. The net transmission over the two-way path was 80%, in agreement with earlier
measurements which indicated 90% transmission for the one-way configuration. Unfortunately,
the return signal suffered fluctuations of as much as 90%, caused by deflection of the beam by
the turbulent gases in the stack and the long moment arm of this arrangement. Using a 3-second
post-detection time constant, the fluctuations in the transmitted signal were reduced so that
transmission could be read to ±1.5%.
Since the SO, signal is determined by the ratio of adjacent transmitted pulses, the large sig-
nal fluctuations made the SO^ reading very noisy, and difficult to interpret. (Post-detection fil-
tering of the ratio signal comes after the logarithmic converter, so that all of the transmission
noise is at its input. ) As a calibration check, a 30-cm-long cell containing 3120 ppm SO_ was in-
serted into the path of the laser beam. The SO^ reading was found to increase by 3200 ppm. When
the calibration cell was removed, we noted that the average SO? signal varied over a range of
250 ppm every few seconds. Although this variation probably corresponds to changes in the SO-
content of the stack gas, there was uncertainty as to the actual location of zero ppm on the re-
corder, so that the absolute value could not be obtained.
30
-------
6.3 On-Line SO, and Transmission Measurements at a Coal-Burning Power Plant
Final measurements were performed in Worcester, Massachusetts at the Webster Street
Station of the Massachusetts Electric Company. The full load capacity of the power station was
35 megawatts, produced primarily by bituminous coal (16-17 tons per hour), although some oil
was being used to reduce particulate emissions. The coal contained 0.8^ sulfur, and air atom-
ization was used in the furnace. An Aerotech Cyclone dust precipitator was in operation.
The diode laser system was transported to the site and positioned so that the beam traversed
a 5-meter-wide horizontal breech between the precipitator and fans which forced the combustion
products into the stack. Caps were removed from 4-inch-diameter pipes to permit optical trans-
mission directly across the breech; and, because the apparatus was located before the exhaust
fans, a negative pressure existed which eliminated the need for windows. Another port was
available for the insertion of a Dynascience SO2/NO point sampling monitor, operated by
Charles Rodes of the Environmental Protection Agency.
The high particulate loading of this stack (2.5 grains per cubic foot) together with the long
optical path attenuated the laser signal by 99c7r when the two-way configuration shown in Fig. 6
was used. Under these conditions, the signal was only a few times larger than the detector noise,
so that the ratio signal proportional to the SO_ content was too noisy to be meaningful. By plac-
ing the detector Dewar on the opposite side of the stack, for single-path transmission, the trans-
mitted signal was an order of magnitude higher, and it was possible to make measurements of the
SO. concentration.
IE
STACK DIAMETER 5m
FUEL;COAL/OIL
INTEGRATION TIME- Isec
ASH REMOVAL
O
to
5
14:30 14:40 14.50 15 OO
15:10 15:20
TIME (hr)
15 JO
1540 1550 1600
Fig, 23. Transmission of diode laser radiation across 5-meter-wide coal-burning
stack in Worcester, Massachusetts (Webster Street Generating Station).
In Fig. 23, transmission of the laser beam is shown for the operating stack over a 1.5-hour
period. The 45-second oscillations are real, as they also showed up at times on the point-
sampled data, but their origin is unknown. They could be caused by fluctuations in the coal feed
rate, or resonances between the two exhaust fans. During removal of ashes, the particulate
concentration increased abruptly, as indicated by a drop in transmission at 3; 25 p.m. (15:25).
31
-------
-5-4U3-1
STACK DIAMETER 5m
FUEL COAL/OIL
INTEGRATION TIME I0s«c
11/3/72
12 o
TIME Ihr
Fig. 24. Transmission and SO2 measurements at Webster Street Generating
Station using diode laser monitoring system. A one-way path was used be-
cause of the high attenuation due to particulates.
700
E
a
_a
O
UJ
(J
O
500
too
300
200
100
WEBSTER STREET STATION
FUEL ; COAL/OIL
FULL LOAD 35 MW
11/2/72
SO,
MILL ADDED
CHANGED BOILER
AIR FLOW
SO,
12 20 IZ 30
1240 1250 1300
TIME (hr)
1310 1320
Fig. 25. Measurements of SO2 and total NOX at Webster Street facility
using a Dynascience sampling monitor.
32
-------
At times, the SO? reading also exhibited oscillations having a periodicity of 45 seconds, but
such oscillatory behavior was not usually present. Figure 24 shows transmission and SO- values
over a 19-minute period. In each case the integration time constant is 10 seconds. The mean
SO- value is around 400 ppm, with excursions between 200 and 600 ppm.
Using the Dynascience sampler, the SO- concentration was found to vary between 400 and
500 ppm. This range was found to hold from day to day, and was relatively constant even when
the power load was reduced from 35 1VTW to 18 MW. (A lower NO value was measured during
reduction of the load, as expected.) Results of on-line sampling of the SO- and NO content of
c* X
this stack during the day preceding the diode laser mea.surements are shown in Fig. 25. This
particular time interval was selected to show that the 45-second oscillations which were detect-
able in transmission across the stack can show up in the SO- measurement as well. As men-
tioned earlier, however, such oscillations in the SO- content are not usually present.
Because of a scheduling conflict, no simultaneous measurements involving both monitoring
techniques were carried out. Nevertheless, some conclusions can be drawn on the basis of these
measurements.
(1 ) Periodic (45-second) oscillations in the transmission of the laser beam across the
stack appeared, at times, in the SO- readings of the Dynascience point-sampling
monitor. Is the point sampler responding to fluctuations in the particulate
concentrations?
(2) The average concentration of SO- in the stack gas, as determined by the diode laser
technique, was approximately 150 ppm below that measured by the Dynascience
sampler. Does this reveal a fundamental operating difference between the two
methods (e.g., free SO- molecules vs those absorbed on particles), or does it indi-
cate the difference between sampling at one point in the stream and obtaining an
average value across the stream''
Experiments directed toward answering the questions raised by these measurements can be
performed under laboratory conditions. The coni;lusions can have an important bearing on the
future course of source-monitoring instrumentation.
7. RECOMMENDATIONS FOR FITTURK STUDIKS
With the system now operational, we recommend that .several studies be performed in the
laboratory before its application in the field. These are described as follows.
7.1 Temperature Dependence
Although the region for SO? detection was selected by virtue of its relatively temperature-
independent absorption coefficient, the temperatures within a stack can fluctuate so strongly
that the system response should be checked as a function of SO- gas temperature. The test is
relatively simple, and would yield valuable information with regard to the analysis of any
in situ data.
7.2 Interferences from Other Gases
As shown in Section 5.4, interferences due to Nil, and C.H. can occur at certain wavelengths.
NO and CO are also present in large quantity, but they should not interfere because of a lack of
spectral lines beyond approximately 5 microns. However, a study should be made of other gases
33
-------
which may be present to see if their wavelengths for absorption are near those for the diodes of
the present system; if there are some, an interference measurement should be made.
7.3 Laser Tuning and Directionality
There is evidence that some of the diode lasers undergo a change in the direction of emission
when the "tuned" pulse is triggered. When this is the case, interpretation of the data can be
difficult. It is possible, however, to overcome this problem by initially setting up the system
with the tuning pulse set equal to zero. With the "zero" retroreflector and across-the-stack
retroreflector identically aligned, the presence of the tuning pulse will shift the direction such
that it can be monitored on the "zero" retroreflector, for which the "zero" control is then ad-
justed to produce a reading of 0 ppm. By sliding the "zero" retroreflector out of the way, the
across-the-stack transmission will yield directly the ppm SO, in the stack.
The cause of the above directionality is not understood at the present time, although new
fabrication techniques, such as the use of stripe-geometry contacts, are expected to improve the
characteristics. For very sensitive applications, two different lasers can be used to produce
the electromagnetic energy simultaneously at two wavelengths for ratio analysis.
34
-------
ACKNOWLEDGMENTS
We acknowledge the generous cooperation of personnel at Hanscom Field
in Bedford, Massachusetts in permitting several preliminary measure-
ments to be made at the base heating plant facility; and of officials of the
Massachusetts Electric Company for assistance in the final system meas-
urements at their coal-burning power plant in Worcester.
Appreciation is also expressed to Charles E. Rodes of the Stationary
Source Emissions Measurement Methods Branch of the Environmental
Protection Agency, for performing correlative point-sampling meas-
urements at the Worcester power plant, and for assisting in the data
analysis.
REFERENCES
1. For a general review, including extensive references, see Tunable Infra-
red Lasers and Their Applications to Air Pollution Measurements.
E. D. Hinkley, J. Opto-Electronics 4, 69 (1972); also included as Appen-
dix B of this report.
2. Laser-Raman Radar, H. Inaba and T. Kobayasi, ibid. ±, 101 (1972); Mie
Scattering Techniques for Air Pollution Measurement with Lasers.
R. T. H. Collis and R. E. Uthe, ibid. 4, 87 (1972); Laser Raman Radar
Detection Based on a Sampling Technique. L. R. Lidhold, ibid. 4, 133
(1972); A Study of Tunable Laser Techniques for Remote Mapping of Spe-
cific Gaseous Constituents of the Atmosphere, Raymond M. Measures
and Gilles Pilon, ibid. 4, 141 (1972).
3. Development and Application of Tunable Diode Lasers to the Detection
and Quantitative Evaluation of Pollutant Gases, Final Technical Report
prepared for the Environmental Protection Agency under Electronic Sys-
tems Division Contract F19628-70-C-0230 by M.I.T., Lincoln Laboratory,
30 September 1971.
4. 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).
5. Tunable-Laser Spectroscopy of the n Band of SO?, E. D. Hinkley,
A. R. Calawa, P. L. Kelley, and S. A. Clough, J. Appl. Phys. .43, 3222
(1972).
6. We are indebted to S. A. Clough of the Air Force Cambridge Research
Laboratories for making his computer line listing for SOj available to us.
35
-------
APPENDIX A
OPERATING AND CIRCUIT DETAILS
This Appendix contains operating details and a description of the electronic components of
the Diode Laser Monitoring System. It is divided into four main sections:
I. Operating Procedure
II. Electronic Circuits and Timing Sequence
III. Electronic Alignment Procedure
IV. Cryogenic Dewar Preparation
I. OPERATING PROCEDURE
Figure A-l shows the front panel of the electronics control console. Figure A-2 is a top
view of this cabinet with the cover removed. The system is made operational (except for the
diode laser power) by placing the main power toggle switch of Fig. A-l in the "on" position. A
Fig. A-l. Front view of electronics control console.
yellow light to the left of this switch indicates the application of main power, and the fuse to the
right will light if it has blown. As soon as the main power has been turned on, the " Laser Ampl"
toggle switch can be raised to energize the power supply to the laser drive circuit. Because
there can be damaging currents supplied to the diode laser if the main power is turned on or off
while the "Laser Ampl" switch is "on," a time-delay relay is incorporated. This relay, des-
ignated as Kl in Fig. A-2, shorts out the diode laser for a period of 10 seconds after the main
power is applied, allowing time for power-supply-induced transients to disappear. In the event
of a power failure to the entire system, the relay will also protect the diode.
Figure A-3 is a close-up view of the optics table. Initial alignment of the beam-splitter,
calibration cell, and infrared detector should be performed in the laboratory so that no further
adjustment is necessary at the field site. Optical alignment in the field then requires the rotation
-------
' with
DETECTOR 9
DEWAR
PREAMPLIFIER
"FLIP"- MIRROB
Fig.A-3. Detailed view of optics table, illustrating various components.
-------
of the laser Dewar for maximum transmission signal in both the "zero" and "calibrate" positions
of the front panel selector switch, followed by an orientation of the entire assembly for maximum
transmission across the stack to the retroreflector, and back. The optical and electrical adjust-
ments are as follows.
(1) Set instrument panel function switch alternately to "zero" and to "calibrate," rotating
laser Dewar until both transmission signals, as indicated by the analog recorder, are the same
as those achieved previously in the laboratory. This ensures that the laser Dewar is oriented
properly, (Since the detector Dewar cannot be moved, no adjustment of its collecting lens is
warranted.)
(2) Position the retroreflector on the opposite side of'the stack. Set function switch to
"monitor," and with the aid of the visible helium-neon laser (previously aligned collinearly with
the diode laser beam), orient the assembly until the laser beam hits the retroreflector. With
the He-Ne laser "flip" mirror removed, correct for any slight alignment difference by a small
rotation or tilt of the table for maximum "transmission" signal. The system is now aligned
optically.
(3) Return function switch to "zero." This causes the movable retroreflector to inter-
cept the beam before it can enter the stack. By adjusting the "zero" control knob, set the SO,
analog recorder pen to zero volts. The digital panel meter will also read zero if the panel toggle
switch is down (in the "monitor" position). Both the tuned and untuned laser pulses now produce
equal detector output in the absence of any SO,.
(4) Set function switch to "calibrate," thereby directing the laser beam through the
10-cm-long (calibration) cell containing a known quantity of SO,. The ratio of calibration cell
length to smokestack diameter is multiplied by the gas concentration in the cell to yield a value
for the comparable £©2 concentration in the stack. For example, if the calibration cell contains
20,000 ppm SO,, the equivalent concentration for a smokestack 10 meters in diameter is 0.1/10 x
20,000 = 200 ppm. The monitor "calibrate" control knob would then be adjusted so that the dig-
ital panel meter reads 0200.
(5) Set function switch to "monitor," and select appropriate time constants for "SO2"
and "transmission" readings between 0.1 and 10 seconds. The system is now operational.
39
-------
STACK
| 1 OPTICS JABLE_ J
''° .2! "I 13
Fig.A-4. Block diagram of electronic modules for diode laser stack-monitoring system.
(a) _p~-
i
i
T| = 8 msec -y- — Tz » 8 msec —
1 u«ec (pulse No 1 trigger)
i
i
(c) — j l—~ 20>j«c (No 1 dump pulse)
1 ]
(d) |^f
i i
D _n
i i
ID j-u
1
(») _l
(h)
(i )
p — — 25jjsec (No 1 gate pulse)
Pulse No 1 delay (set to coincide with end
[— l-3usec (Pulse No 1 to AZ)
| Turing pulse delay
!
(_ ] Tuning. Dulse to A2
• |- 1 jjsec (pulse No 2
(j) Pulse No 2 delay |~~| (set to coincide with
(K)
(1)
(m)
i J
— ^1 |
h
h
ih
of 1 | dump pulse)
j
u
1 1
1
1
trigger) (]
i
end of dump pulse) |~~|
1 {
[— {|— l-3psec (Pulse No 2 to AZ) i (J
-^ \- 20 usec (No Z dump pulse) (~~|
— -j f--~Z5usec (No
2 gate pulse) | |
Ad|. on A4
Adj. A, an AS
Widrh 3d) by "zero"
knob on control panel
Adi A2 on A3
Ad| S2 or A3
No adj
Adj Aj or A3
Width adj by S,
on A3
I Adj on AS
Fig. A-5. System timing sequence, indicating locations for making adjustments.
All pulses are 3-5 volts in amplitude.
40
-------
II. ELECTRONIC CIRCUITS AND TIMING SEQUENCE
The block diagram of Fig. A-4 shows interconnection of the various electronic modular
components, each of which is contained on a circuit board within the main console and can be
seen in Fig. A-2. (The Preamp is located in a shielded box on the optics table near the detector.)
The timing sequence is depicted in Fig. A-5 with individual pulses denoted by (a) through (m).
The Master Clock (A3) contains an oscillator (a) that feeds the logic generating circuits.
These circuits control the Laser Driver (A2), which cause the diode laser to emit pulses of in-
frared energy alternately "on" and "off" an SO2 absorption line. A short pulse It is fed into the
laser at a time determined by (e), with a width determined by (f), causing it to emit at a certain
wavelength fixed by the laser crystal composition, its length, and the ambient temperature. The
laser is then heated slightly with a relatively long "tuning" pulse, L, of duration determined by
(g) and (h), and of adjustable amplitude (to 500 mA) usually below threshold for laser emission.
The heating effect shifts the laser emission, \^, to another wavelength, \,. Radiation at A, is
emitted when the laser is driven by a second short pulse, I,, whose location and width are deter-
mined by (j) and (k),
Prior to the current pulse, Ij, Gated Integrator 1 (A4) receives a trigger pulse (b) which
causes it to "dump" its old signal in an interval (c) and prepare to receive new information during
a gate time set by (d). This information is held until a new cycle of operation is started. At the
end of the tuning pulse L.. a trigger (i) of duration (1) is sent to Gated Integrator 2 (AS) for the
"dump" process. The gate to Integrator 2 is turned on by (m), during which time the signal from
the preamplifier is stored. If there is SO2 gas between the laser and detector, the amplitude of
the transmitted signals at the "untuned" and "tuned" wavelengths will be different, causing the
two integrators to store different voltage levels.
Each of the integrators is an input to the Log Converter (A6) whose output is proportional to
the logarithm of the ratio of the two input voltages, V. and V,, such that
Vout = log- in volts .
In the event that the untuned pulse is in a region of small absorption compared to that of the tuned
pulse, a miniature toggle switch on the Log Converter card (A6) may be switched to the opposite
position, so that the output remains positive.
From the Log Converter the steady state signal is processed by the Ratio Gain circuit (A7),
which is capable of changing the output voltage to correspond to a particular reading on the
10-volt-full-scale digital panel meter. For example, if 200 ppm SO, is used for calibration, the
variable gain control of A7 is adjusted for a reading of 0200 on the meter, or 0.2 volt. The Ratio
Gain circuit also provides for filtering of noise, with adjustable time constants from under
0.1 second to 10 seconds.
The Transmission Gain circuit receives the larger (in absolute value) of the two inputs to
the Log Converter, and processes it in a manner similar to that of the Ratio Gain circuit. For
example, if the percent transmission is to be set at 100, then the gain control is adjusted for a
panel meter reading (with the toggle switch set to "transmission") of 0100, or 0.1 volt.
The infrared detector is Ge:Cu, of dimensions 3 mm x 2 mm x 2 mm. It is attached to the
"cold finger" of a liquid helium Dewar according to the schematic of Fig. A-6, and has a resist-
ance of approximately 100 kilohms when cooled. Optical baffling is provided by 2-mm and 5-mm
diameter apertures which restrict the field of view to 30° (f/1.9). The infrared filter, from
41
-------
G« Cu DETECTOR
(2«2i3mm]
Fig.A-6. Cross section of infrared
detector and filter assembly.
10 20 30 40 50 02 4
(0)
Fig.A-7. Oscillograms of important voltages and currents of the monitoring
system, (a) and (b) show the laser current pulses; (c) and (d) the preamplifier
signal; and (e) and (f) the output from one of the gated integrators.
,
-------
Optical Coating Laboratory, Inc., permits radiation only between 1100 and 1227 cm" to reach
the detector. A room-temperature BaF2 window serves as the entrance port to the detector.
In order to reduce the capacitive loading, as well as noise, the preamplifier shown in Figs. A-3
and A-4 is located very near the detector Dewar, so that the lead length is less than 20 cm.
The preamplifier is a low-noise circuit (detector-noise-limited) with a gain adjustable from
5X to 30X by an attached potentiometer. The unit saturates when the output pulse reaches
4.0 volts amplitude and provision has been made to ensure that the preamplifier is within its op-
erating range. This is performed by a measurement of the output voltage from the Transmis-
sion Gain circuit under the condition of minimum gain (i.e., with the "Net Transmission — Calib."
knob fully counterclockwise). As long as the output of the preamplifier is less than 4 volts, the
voltage recorded by the "transmission" pen of the analog recorder, or that indicated on the dig-
ital panel meter (for net transmission) will be less than 0.1 volt. If the final voltage is greater
than 0.1 volt, the preamplifier gain should be reduced. If a further reduction in signal is still
necessary, then a smaller laser current amplitude or width can be used.
Typical oscillograms of voltage and current waveforms under operating conditions are shown
in Fig. A-7. In (a) is shown the laser current pulse 1 of 2 amperes, with a full width at half-
maximum of 3 usec. In (b) the time scale is compressed to illustrate the entire sequence of cur-
rents applied to the diode laser: pulses 1 and 2 of nearly equal magnitude and width, and the tun-
ing pulse I{ which extends from 3 to 8 msec, and tunes the laser so that the wavelength emitted
at pulse 2 is shifted from that of pulse 1. For 8 msec after pulse 2 there is no current applied
to the laser, during which time the junction temperature reverts to its original value before
pulse 1 is again applied. A 3.4-volt signal from the preamplifier is shown in (c); and in (d) the
preamplifier outputs due to both pulses are shown in the same time scale as (b). The gated in-
tegrator is normally at zero volts, as indicated in (e) for time before 24 (isec; but beyond this
point the integrator responds to the preamplifier signal, reaching a value of —0.9 volt, and hold-
ing this value until the dump pulse occurs, as shown in (f). Integrators 1 and 2 are identical in
operation, except for their being gated to pulses 1 and 2, respectively. Their output voltages
are negative and serve as comparative inputs to the logarithmic converter, whose output is posi-
tive and constant for steady-state conditions. The output from the logarithmic converter is fur-
ther amplified by the Ratio Gain circuit, which also permits post-detection filtering, and displayed
on the analog recorder and digital panel meter as "SO, concentration." The larger (in absolute
value) of the two outputs from the gated integrators goes to the Transmission Gain circuit as the
"transmission" signal, which is displayed in the same manner as the ratio signal.
The current-voltage characteristics of the diode lasers at liquid helium temperature are
shown in Fig. A-8. (At room temperature there is no strong rectification because of the narrow
energy gap of the semiconductor material.) In cases where degradation of the electrical contact
of a laser is suspected to have occurred, new curve-tracer scans should be compared with these.
However, in no case should an ohmmeter be used to check for rectification because the standard
internal 1.5-volt battery will cause the diode junction region to "break down," resulting in perma-
nent damage to the laser.
Each of the circuits described above is shown in detail in Figs. A-9 to A-15, which constitute
the remainder of Section II of Appendix A.
43
-------
Fig. A-8. Current-voltage characteristics of the system diode lasers
at liquid helium temperature. Horizontal scale: 0.2 V/div; vertical
scale: 50 mA/div and 0.05 mA/div (for the expanded reverse traces).
Voltages and currents are zero at the center of each oscillogram.
I i
-------
~
flOJUST T2
2 B
9601
Zl
T
8
9601
22
T
AUJUji ii_ «UJUb r ^_
r $ f-^W -X\^-l 0*5V •—K »-^WV ^AA. 1—0 + 5V
I
Fig,A-9. Master Clock (A3).
-------
TUNING PULSE
°~~|| LASER I
-5VU
0 11— I 1 25»
[I USER PULSE 2 M9j
LJ
NOTES
220B - AMLOG DEVICES
*-H£AT SINK REQUIRED
Fig. A-10. Laser Driver (A2).
OOOVF, 100V
~ ARCO
/\ , .1 >^ 400
r~
OIMF:: (
\. '*
• 1% . • ^^
^
5k
11%
*20 25V \ \l
) -=- 13 each Moiiory) ? ' '°
T TR-I35-R
^4^ ± r
i
INPUTS
NOTES
P50I -ANALOG DEVICES |_
'*** ""I |ll-S-41TO|
400 I 2k
*+^ OUTPUT
f 1 TO GATED
>\^ 1 | INTEGRATOR
^^^ [^ 51 ,— . .
J.
FSI25
COM 0
1
1
E* 0 r E- O
E- O INPUTS < OUT O
o- o-
OUT o i ^ RFF o
FS125 - COMPUTER LABS
ALL BNCs INSULATED
1 __
BOTTOM VIEWS
Fig. A-ll. Detector Bias Supply and Preamplifier.
46
-------
23IK -ANALOG DEVICES
CiG-30- TELEDfNt
CRVSTALON CS
9601 - FS.RCHi^O SEMI
Fig.A-12. Gated Integrator (A4, A5).
47
-------
O - IN
O +• IN
B* 0
COM 0
B" 0
OUT 0
TRIM O
IM 2k, 10 TURN
A/W iHWV I i
OUTPUT
NOTES
ALL RESISTORS 1%
75IP - ANALOG DEVICES
118 - ANALOG DEVICES
CAPACITANCE IN >.F
UNLESS OTHERWISE
NOTED
I IB (bottom view)
.Fig. A-13. Log Converter (A6).
48
-------
.«,! TIME
I3I CONSTANT
Fig. A-14. Ratio Gain Control circuit (A7).
INPUT
TIME
CONSTANT
Fig.A-15. Transmission Gain Control circuit (A8).
49
-------
III. ELECTRONIC ALIGNMENT PROCEDURE
Section III is separated into three parts, each describing a particular aspect of the electronic
alignment procedure. Part A details the adjustments to be made immediately prior to monitoring.
In Part B are shown adjustments which affect the timing and sequencing of circuits, but are not
usually adjusted. Part C concentrates on trim controls for the operational amplifiers.
A. Adjustments for Initial Setup and Monitoring
Several adjustments for the laser current pulses, preamplifier gain, and output signal level
are described below.
(1) Laser Driver A2: Tuning Current Amplitude I is set for the most optimum value
for each laser (i.e., maximum absorption coefficient for SO2); it should usually be
under 100 MA to minimize LHe boil-off.
The other two adjustment potentiometers at the top of the Laser Driver card are
for I.j and I.,, which are usually kept at their maximum values of 2 amperes.
(2) Master Clock A3: The only adjustment which should ever need to be made on this
card is S3, which controls the width of current pulse 2. Only if the signal due to
the pulse saturates the detector preamplifier should the width be reduced. The
width of pulse I is controlled by "zero" knob on the front panel and is adjusted such
that the signals due to both pulses are equal in the absence of SO2.
(3) Preamplifier Gain Control: On the preamplifier is a variable-gain control. It should
be set at a maximum (fully clockwise) except when its output approaches 4 volts,
where saturation occurs. In order to ensure that the preamplifier output is under
4 volts in the field, where an oscilloscope may not be available, the transmission
signal, as measured by the panel DVM, should read less than 0.100 volt with the
transmission gain knob (on front panel) at its minimum value (fully CCW).
(4) Panel Controls
(a) "Zero" adjusts width of current pulse 1 so that signals due to 1 and 2. are equal
in absence of SO,.
(b) "SO., Calibration" knob adjusts final ratio amplifier gain (circuit A7) to read
ppm SO., directly.
(c) "Transmission Calibration" knob adjusts final transmission gain (circuit A8)
to read directly in percent transmission.
(d) "Time Constants" for SO2 and transmission readings are independently adjust-
able on front panel.
(5) Polarity Switch
On the "Logarithmic Converter" Card (A6) is a miniature toggle switch which re-
verses the two input voltages to this circuit, so that the output voltage remains
positive. The position of this switch depends upon the initial wavelength of the diode
laser (whether it is in a peak or valley of the SO., spectrum), which may vary from
one laser to another.
50
-------
B. Potentiometers Not to Be Adjusted
The following labeled potentiometers do not ordinarily need adjustment, as they control the
timing sequence and some noncritical parameters:
(1) On Laser Driver A2, current amplitudes I. and I_ are set at maximum values of
2A. Fine adjustment of current amplitude is difficult, so the laser pulse width
controls are normally used to adjust detector output signal.
(2) On Master Clock A3, the basic timing sequence is determined by T, and T-, and
the delays for pulse 1, 2, and I by A., A3, and AZ< respectively. These should
not be adjusted, nor should the width of I_, governed by potentiometer S^.
(3) On the two Gated Integrator circuits, A4 and A5, there are two potentiometers
marked "B" and "T," which do not normally need adjustment. "B" can be used to
set the output voltage to zero in the absence of any input signal, and "T" adjusts
the time during which the integrator is "on." It is important that the outputs of A4
and A5 are equal if the inputs are the same; and IJie adjustment "T" can be used to
accomplish this.
C. Operational Amplifier Trim Adjustments
Sixteen operational amplifiers are employed in the system on the various functional circuit
boards as well as in the detector preamplifier." Each OP-AMP has been adjusted by an attached
potentiometer so that the output voltage is zero when the voltage to its input resistor is zero.
Since this trim adjustment has already been made, it should not need to be rechecked unless an
OP-AMP is replaced.
51
-------
IV. CRYOGENIC DEWAR PREPARATION
This section describes evacuation and cool-down procedures for both the laser Dewar and
the detector Dewar. The laser Dewar requires liquid nitrogen (LN) in its outer jacket, with
liquid helium (LHe) in the center, whereas the Dewar housing the infrared detector requires
only LHe.
A. Laser Dewar (Supairco)
Should be pumped to approximately 3X10 Torr just prior to filling with LN and LHe, ac-
cording to procedure (1) or (2) below. If evacuation is not possible (e.g., at field site), procedure
(2) must be used to permit LHe cryopumping to take place during LN transfer.
Cooling Procedures:
(1) Pre-Cooled (requires 4-5 liters of LHe)
Connect rubber hose between LN tank and inlet (larger diameter) to external jacket.
Connect a rubber tube from the outlet (smaller diameter) to center section of Dewar.
Keep safety pressure relief disk and Dewar valve warm by using a heat gun. Fill
until LN spills from inner section. Connect split hose to outer jacket, and stopper
inner section. After 15 minutes, use gaseous He to force LN from inner section.
Purge LHe transfer tube with He gas, and immediately fill inner section of laser
Dewar, using approximately 4 psi helium gas pressure. LHe fill takes approx-
imately 4 minutes and is noted by a marked increase in size of the water vapor vent
cloud.
(2) Direct LHe Fill (requires 6-7 liters of LHe}
Connect rubber hose from LN tank to inlet (larger diameter) of external jacket,
leaving outlet (smaller diameter) venting to atmosphere. Keep inner section purged
with a small flow of helium gas to prevent condensation. Transfer LN to outer
jacket for a few minutes to allow some pre-cooling of the Dewar. If the outside of
the Dewar becomes cold, or after the outer jacket has been filled, remove helium
gas purge tube and transfer LHe. Because of the relatively warm temperature of
the Dewar, there will be "puffs" of water vapor visible for several minutes after
transfer, as the LHe continues to cool it. The Dewar should be "topped" off with
both LHe and LN about one hour later if a test of 4-5 hours is planned.
(3) "Topping Off" Dewar
In order not to blow out all the LHe in the Dewar during this procedure, the transfer
tube is first inserted into the LHe storage Dewar, and 2-4 psi of He gas pressure
applied until LHe emanates from the other end of the tube, at which point it is in-
serted into the laser Dewar.
B. Infrared Detector Dewar (Santa Barbara)
This is a LHe-only Dewar, and does not require LN during operation. It should be pumped
to approximately 3X10 Torr every two weeks, but may need more frequent pumping if LN
precooling is used. However, because of the small difference in LHe required, direct LHe
transfer is usually adequate, simpler, and avoids any chance that LN is still in Dewar before
LHe transfer.
52
-------
Cooling Procedures:
(1) Direct Transfer (requires 3-4 liters of LHe)
Purge LHe transfer tube with He gas and insert into storage and detector Dewars.
Apply 2-3 psi He gas and transfer until vapor plume increases noticeably (2-3 min-
utes). Reduce He gas pressure, remove transfer tube, and fully screw on cap to
detector Dewar; then turn cap counterclockwise 1 full turn. Make sure rubber
stopper is on securely (this can be held in place with tape).
(2) Pre-cooling (requires 2-3 liters of LHe)
Fill Dewar with LN, replace cap, and allow to sit for 10-15 minutes. Remove cap,
invert Dewar, and "shake" out the LN. Transfer LHe immediately according to
the procedures described in (1).
C. LHe Transfer Tube (Janis Research Co.)
Should be evacuated to pressures less than 10 Torr every 2-3 months. This is not critical
because of cryopumping of LHe during transfer, but a better vacuum reduces the amount of LHe
required.
53
-------
APPENDIX B
TUNABLE INFRARED LASERS AND THEIR APPLICATIONS
TO AIR POLLUTION MEASUREMENTS"
E. D. Hinkley
This paper describes a number of tunable infrared lasers and techniques
employing them for the detection and monitoring of gaseous air pollutants.
Recent progress in the development of lasers that can be matched to char-
acteristic infrared absorption or emission lines of certain pollutants sug-
gests wide potential application for sensitive, specific monitoring. Ex-
amples to be described include highly specific point-sampling, in situ
source monitoring, ambient air monitoring, resonance fluorescence, and
remote heterodyne detection.
\. INTRODUCTION
The use of fixed-frequency gas lasers for air pollution detection has been suggested and
advanced by several workers over the last few years. By selecting appropriate laser transi-
tions which overlap those of the gaseous pollutants, their concentrations can be measured by di-
rect absorption. However, the match is seldom ideal - and for some important pollutant gases,
such as SO-, there appear to be no entirelv satisfactory laser lines. Even when overlaps are
found, there may be other reasons why a selected gas laser line is not suitable - such as absorp-
tion by other components of the normal atmosphere, or coincidence (interference) with spectral
lines of other pollutants which may be present.
During the past two years substantial progress has been made in the development of several
types of tunable lasers, and systems utilizing these devices are being built for point-sampling
and in situ source monitoring. Tunable infrared lasers have also been proposed for ambient
air monitoring (both at ground level and in the upper atmosphere) and for single-ended, remote
heterodyne detection of emissions from stationary or mobile sources.
A brief survey will be given of the various types of tunable lasers, followed by a discussion
of monitoring techniques which can be used. Most of the experimental results represent work
performed in our Laboratory using current-tuned semiconductor diode lasers, although work
performed by others will also be described, for completeness. Two recently published articles
4
can provide general background information for laser techniques. Kildal and Hyer compare in
detail laser Raman scattering, resonance fluorescence, and tunable laser absorption techniques.
Melngailis covers in more detail the direct absorption techniques and tunable semiconductor
lasers.
2. TUNABLE LASERS
Table I lists the types of tunable lasers which are presently available or under development.
We consider only primary tunable laser sources, although other wavelengths are possible by
mixing radiation from a tunable laser with that from a fixed-frequency laser.
^Published in .1. Opto-Electronics 4, 69 (1972), as an invited review paper.
54
-------
TABLE I
TUNABLE LASERS
Organic Dye Lasers
Parametric Oscillators
Semiconductor Diode Lasers
Spin-Flip Raman Lasers
Bulk Semiconductor Lasers
(Optically pumped)
High-Pressure Gas Lasers 4.8 - 8.5*(CO)
(Electron-beam-pumped) 9.1 — 11.3*(CO,)
Approximate Coverage
(jim)
0.34 -
0.5 -
0.63-
5.3 -
9.2 -
0.32 -
1.2
3.75
34
6.2 (CO)
14 (C02)
34 *
Single- Mode Pulsed
CW Output Output
(W) (W)
5 x 10"2 107
3 X 10"3 105
ID'3 102
1 103
10-3
* Predicted
2.1 Organic Dye Lasers
Tunable organic dye lasers are available commercially, covering the wavelength range from
the near ultraviolet to slightly over 1 (jm in the infrared. They operate by optical pumping from
the ground state to an excited singlet electronic state of high rotational and vibrational energy.
Subsequently the vibration-rotation energy is thermalized by very rapid nonradiative relaxation.
A population inversion then exists, and stimulated emission occurs between low vibration-rotation
energy levels of the first excited singlet electronic state and higher vibration-rotation levels of
the ground electronic state (according to the Franck-Condon principle). Due to the high density
of vibration-rotation levels of the large dye molecules and the large widths of these levels in
solution, very broad, continuous bands of radiation are emitted (typically several hundred ang-
stroms wide). Narrow-linewidth lasing, tunable within the broad emission band, is achieved by
the use of highly frequency-selective optical resonators.
Lasers have been made from dyes of the oxazole, xanthene, anthracene, coumarin, acridine,
azine, pthalocyanine and polymethine families. Using various dyes, pulsed operation has been
obtained from 0.34 \jm to 1.2 ^m, with both flash lamps and lasers (notably N, and ruby) used
8
as pumps I.inewidths as narrow as 7 MHz have been obtained with an external Fabry-Perot-
Peak powers of 10 W, and average powers of the order of 1 W (Ref. 8a) have been observed.
2.2 Optical Parametric Oscillators
Commercial versions of optical parametric oscillators surpass the dye laser in infrared
tunability, extending the available range to almost 4 ^m. An optical parametric oscillator re-
quires a laser pump source impinging onto a nonlinear crystal. The resulting radiation can be
Q
tuned over a wide wavelength range by changing the temperature or orientation of the crystal.
A doubled Nd: YAG-laser-pumped oscillator using I.iNbO, as the nonlinear element can tune
— \ -1
from 0.55 to 3.75 \im with a bandwidth of 1 cm" in the visible and less than 0.25 cm" in the
infrared; with single pulse energy of 1 m.I and repetition rates up to 1000 pps. I'sing a ruby
laser pump source, a spectral region from about 1 pm to 4 (jm can be covered with pulse energies
55
-------
of approximately 0.1 J. The spectral width typically 0.25 cm" is determined by the crystal
length, but it can be reduced by using an intracavity etalon. A bandwidth of 30 MHz (0.001 cm )
has been recently reported in the vicinity of 2.5 ^m for a pulsed LiNbO3 doubled Nd:YAG oscil-
lator using a novel interferometer mode-selector scheme. Recently, parametric oscillation
in the 10-|ini region has been achieved in the laboratory, indicating that this tunable source can
potentially cover all of the important infrared "fingerprint" region between 3 and 15 |jm.
2.3 Semiconductor Diode Lasers
Semiconductor diode lasers can be tailored to emit in desired wavelength regions by control
of the energy gap — a feature made possible by the development of ternary semiconductor com-
pounds of adjustable chemical composition. The different materials from which semiconductor
lasers have been made are shown in Fig. B-l. Complete coverage of the region between 0.63 and
34 pm is now possible. The dashed lines indicate possible future extension of the present limits
WAVENUMBER (cm)*'
IOS
1 1 1 1 1
|l<-9-3246-3J
In
AI.G
Go/>
CUS,Si
i i
*0 103 K
O1.|Al h— 1
*(-,p. 1 — 1
1-1 l~~l
1 1 I 1 1
1 1 1! | 1
Pbl-i
_|J L
_BS
5n.1
i—
-i 1
i Se ,_
K ^
1
5b, Cd S
1-1 i
1 1 1 1 | 1
HNOj, H2S, N20
N02, PAN, S03
S02, 03, NH3
fl lll.li L__! 1 1 1 I_L11
01
10
WAVELENGTH (pm)
too
Fig. B-l. Wavelength ranges for semiconductor lasers made from different
alloys and compositions. Also shown are some strongly absorbing regions
for several common atmospheric pollutants.
(solid lines). Most of the tunable diode laser work has been performed with the lead chalco-
genides Pb. Sn Te (between 6.5 and 34 ^m), PbS. Se (between 4.0 and 8.5 jim), and pbi_xcdxs
(between about 2.5 and 4 (j.m). Principal absorbing wavelengths of some of the important pol-
lutant gases are indicated at the bottom of this figure. It should be noted that with these three
semiconducting compounds, nearly all of the important pollutants can be detected.
Figure B-2 shows a semiconductor diode laser in its standard package, which is 1 cm in
over-all length. (The entire unit is approximately the size of an average transistor.) The diode
crystal is mounted on a copper stud which serves as one electrical contact. A silver ribbon
serves as the other contact. Laser emission is produced by passing a current through the diode;
this current can be supplied by a small battery or DC power supply for CW operation, or by a
pulser for pulsed operation. There are several ways to "tune" a diode laser, the simplest being
to change the magnitude of an applied direct bias current, which changes the junction temperature
through heating. Since the refractive index within the laser cavity is temperature-dependent,
56
-------
COPPER STUD
CLEAVED FACE
COHERENT
RADIATION
(OUT)
COPPER STUD
CERAMIC
SILVER RIBBON
DIODE LASER
Fig. B-2. Semiconductor diode laser in a standard package having
an over-all length of \ cm. The laser itself has approximate di-
mensions of 0.12 x 0.05 x 0.03 cm. Tuning is accomplished by
changing the direct current applied to the diode.
InSb Sample
5-6/1 CO Laser
CW Tunable
Raman Laser
Output
Solenoid
Fig. B-3. Experimental arrangement for a CW spin-flip Raman laser consisting
of an InSb semiconductor crystal pumped with a CO gas laser. The magnetic
field produced by the solenoid controls the electronic energy levels within the
semiconductor, permitting wavelength-tuning.
57
-------
the laser wavelength changes. Although this type of tuning is thermal, it is still relatively fast
because of the limited volume involved. Modulation frequencies of several hundred Hertz can be
applied before thermal inertia becomes very noticeable, and useful experiments have been car-
ried out with frequencies as high as 10 kHz.
Current-tuned semiconductor diode lasers have been used to perform Doppler-limited spec-
14 2
troscopy on several gases: SF,, NH,, CjH^ in the 10-(im region, ' , NO, CO, and water vapor
around 5 (im (Ref. 15), and an extensive study of the v band of SO- in the 8.7-^im region.
At present, the greatest obstacle to broader application of semiconductor lasers is their
requirement for cryogenic cooling. Large improvements in efficiency are expected to relax
these cooling requirements to the point where pulsed operation even at wavelengths as long as
10 \im will be possible at temperatures above 100°K.
2.4 Spin-Flip Raman Laser
The spin-flip Raman laser is a device which uses a fixed-frequency laser (at present, a CO
or CO£ gas laser) to pump a semiconductor crystal at cryogenic temperatures and in a magnetic
field.17 The pump laser photons lose energy when they collide with electrons in the crystal and
flip their spin. The down-shifted Raman photon is separated in energy from the pump photon by
the magnitude of the electron spin energy g/3H, where g is the conduction electron g-factor, ft the
Rohr magneton, and H the magnetic field strength. Consequently, the output frequency depends
on the magnetic field.
A schematic drawing of a spin-flip Raman laser using a CO laser pump and InSb semicon-
ductor is shown in Fig. B-3. A linewidth of 1 kHz for the tunable laser radiation has recently
A Q
been reported by Patel on the basis of heterodyne measurements. The spin-flip Raman laser
19
has been used to obtain spectroscopic data of NH3 in the 10-)im region and of NO around 5 >jm
(Hef. 20).
2.5 Optically Pumped Semiconductor Lasers
A Q-switched Nd:YAG laser has been used to pump samples of the ternary compound semi-
conductor InxGa,j_xAs at room temperature, producing laser radiation at 1.09 and 1.12 (im
(Ref. 21). Recently, Melngailis achieved CW laser emission from a Hquid-helium-cooled
^O 88^nO 12^e crvstal pumped with a GaAs diode laser. It appears that, by using commer-
cially available GaAs diode lasers as pump sources, tunable semiconductor lasers can be made
which have some of the advantages of gas lasers, such as high output power and good beam qual-
ity, in addition to the relative simplicity, ruggedness, and small size characteristic of diode
lasers.
2.6 High-Pressure Gas Laser
At pressures of 10-15 atmospheres, adjacent vibration-rotation lines, which are responsible
for emission in gas lasers, overlap such that lasing can occur at intermediate wavelengths as
well. At present, electron-beam excitation is required for this potentially powerful tunable
source of infrared radiation, and several laboratories are investigating its potential for spec-
troscopic applications.
3. MONITORING TECHNIQUES
Comparison is often made between tunable laser techniques and those involving conventional
infrared instrumentation for air pollution monitoring. The laser system usually has better
58
-------
sensitivity and specificity because of its higher usable power in a narrow spectral bandwidth,
compared with that available from thermal sources. A comparison of resolution is illustrated
in Fig. B-4, where a diode laser beam at 10 ^.m was transmitted through a laboratory-quality
grating spectrometer with narrow (80-jj.m-wide) slits. The smooth bell-shaped curve represents
the spectrometer slit function, with a half-width (resolution) of 0.18 cm . The narrow absorp-
tion lines were produced by introducing 0.5 Torr of ethylene into a 30-cm-long gas cell in front
of the spectrometer. These lines are predominantly Doppler-broadened to a width of 66 MHz
(0.0022 cm" ). The laser linewidth itself is two or three orders of magnitude narrower than
this. It is clear that any attempt to use this spectrometer to evaluate the high-resolution spec-
tra of ethylene will suffer, both in terms of sensitivity and specificity, because of overlap and
a severe reduction in intensity produced by the relatively wide spectral interval accepted by the
instrument.
For actual in situ measurements,, spectral lines of pollutant gases which are this narrow
would only occur at very high altitudes — in and above the stratosphere. Nevertheless, even at
ground level where atmospheric pressure broadens these ethylene lines to a value of approxi-
mately 0.2 cm~ , some instrumental broadening will be produced by even this spectrometer,
and substantially more by one which can be used in the field.
Three fundamental monitoring techniques employing tunable lasers are: remote heterodyne
detection, resonance fluorescence, and direct absorption. Each has its particular advantages
and applications, as described below.
3.1 Remote Heterodyne Detection
This technique represents a completely passive, single-ended laser system for the detection
of gaseous pollutants from stationary or mobile sources. Characteristic infrared emission lines
from a pollutant gas are detected by heterodyning them with tunable laser radiation of the same
wavelength, using the configuration of Fig. B-5. The laser radiation and that from the source
are directed onto a wideband infrared detector cooled to liquid helium temperatures. By scan-
ning the laser wavelength through that of the pollutant emission line, a beat frequency is produced
whenever the difference frequency between the two infrared signals is within the bandpass of the
detector/amplifier system; the amplitude of the signal is related to pollutant concentration. The
signal-to-noise ratio is given by
| = [i-exp(-0;.cL)l Lp(nt,AT )_1 'expdWkT ) - ll 'BT'1/2 ' <1)
18 b J
where a' is the absorption coefficient of the line per ppm of gas concentration (c), L is the thick-
ness of the plume, T is its temperature, f, is the emissivity of the background at temperature
T. , v is the infrared frequency, h is Planck's constant, k is Boltzmann's constant, B the
system bandwidth and r the post-detection integration time. Equation (1) is also based on the
following assumptions: (a) that the [F 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 a low concentration there relative to that in the plume; (c) that the background attenuation
from the wings of other molecular absorption lines is negligible; and (d) that the local oscillator
has sufficient power to overcome the other sources of noise. Equation (1) holds, regardless of
range, as long as the field of view of the collecting telescope is fully subtended by the plume; in
the infrared, with collection optics only a few centimeters in diameter, it should be possible to
detect pollutants one kilometer away.
59
-------
CO
CO
2
CO
z
<
or
UJ
tr
Spectrometer
Slit Function
(P = 0)
945.1 945.2 945.3 945.4
WAVENUMBER (cm'1)
945.5
Fig. B-4. Transmission of tunable Pbrj.88SnO \"i^e diode laser
radiation through a grating spectrometer with 80-jj.m slits,
showing the bell-shaped slit function. The narrow absorption
lines are produced when 0.5 Torr of C2H4 was introduced into
a 30-cm-long cell in front of the spectrometer.
•BEAM-SPLITTER
DETECTOR
Fig B-5. Configuration for remote heterodyne detection of pollutant gases
from a smokestack using a tunable-diode-laser local oscillator
60
-------
B'SOOMHz
10»ec
L - !0m
* 1«10"5 cm'/pp
S/N '4
-------
liquid helium boil-off. When wideband (>500 MHz) photodiodes having nearly unit quantum ef-
ficiency become available, the local oscillator power requirement will be less than 100 jiW - a
more realistic value. Experimentally, heterodyne theory has been confirmed by measurements
of thermal radiation from a calibrated source; and lines from hot ethylene (600°K) near 10 ^im
were detected using a CO- laser as local oscillator and a Ge:Cu photoconductor as infrared
detector.
3.2 Resonance Fluorescence
Absorption of tunable laser light, and detection of subsequent re-radiation, can be used for
remote single-ended, detection. Because the laser radiation is directed into the atmosphere,
this technique is active, rather than passive. Fluorescence can be excited by pumping either
at vibrational frequencies of the molecules, which require lasers at infrared wavelengths, or at
electronic transition frequencies, which fall in the visible or ultraviolet part of the spectrum.
Several gases, such as SO2, O,, NO, NO2, NH3, HCOH (formaldehyde), and C^H^ (benzene),
have absorption bands in the ultraviolet. However, there is a potential problem due to overlap
between individual bands, some of which are quite broad. Moreover, atmospheric transparency
becomes poor as we approach the vacuum ultraviolet. The electronic transitions should be im-
portant for detecting metal vapors such as As, Be, Cu, Zn, Na, and Hg. Atomic sodium has
been detected at altitudes of about 90 km, in concentrations of the order of 10 /cm using a dye
laser tuned to the D lines. For resonance fluorescence in the infrared, Kildal and Byer pre-
dict that with a tunable optical parametric oscillator at 4.7 ^m, emitting 1 mJ pulses, 100 nsec
wide, with a
100 meters.
wide, with a 0.1 cm" spectral width, it should be possible to detect 2 ppm of CO at a range of
3.3 Absorption
Direct absorption of tunable laser radiation offers the. greatest versatility of the three
schemes discussed. Point sampling at reduced pressure permits very high specificity as the
infrared "signatures" of pollutant gases become observable. Accordingly, this technique can
also be used to calibrate other instrumentation used for point-sampling. In situ source monitor-
ing can be performed by transmitting the laser radiation across the effluent stream, leaving the
stream itself essentially undisturbed. Transmission of tunable laser radiation over long at-
mospheric paths can be used to attain the high sensitivity required for ambient-air monitoring.
3.3.1 Point Sampling
In point-sampling applications the total gas pressure can be lowered to reduce or eliminate
completely any overlap between adjacent vibration-rotation lines; consequently, very high spec-
ificity is possible. One example of this is shown in Fig. B-7, containing tunable PbSn /nSeg JQ
diode laser scans of NO in the 5.2-fim region. The characteristic Lambda-doubling of the
R(21/2),j /2 line at 1912.1 cm in the upper trace is an unmistakable "signature" for this gas
Recently, Brueck, Johnson, and Mooradian have observed Lambda-doubling of the Rd/2).,.,
-1 20
line of NO at 1881.0 cm using a CO-pumped spin-flip Raman laser
Kreuzer and Patel have used a 50-m\V spin-flip Raman laser in conjunction with an opto-
acoustic cell (spectrophone) to detect NO in both the ambient air and in samples of automobile
exhaust. By using a mechanical chopper to modulate the intensity of the laser beam, acoustic
pulses produced in the cell when the laser is tuned to an absorption line can be detected with a
6i
-------
489
491
605
493
495
497
499
1 Torr
RI21/2I,
607 609
DIODE CURRENT (mAI
613
-\
Fig. B-7. Tunable PbSQ £0Se0 40 diode laser scans of two NO lines in the 1912.1 cm
region, Lambda-doubling of the RI21/2)./, line is 300 MHz wide (0,01cm'1), After
Kill, et al.*5 1/Z
63
-------
capacitive microphone, analogous to the operation of a Golay cell. Some of their measurements
are shown in Fig. B-8. Trace A is a calibration scan of 20 ppm NO in NZ at a total pressure of
300 Torr. The lines designated as 1, 5, 6, 8, and 11 are due to NO, while the others represent
water vapor. Trace B shows the noise level with the laser radiation blocked, which is primarily
Johnson noise in the first amplifier stage. The concentration of NO in a sample of ambient air
is estimated to be approximately 0.1 ppm on the basis of Trace C. A sample from a busy road.
Trace D, shows a somewhat higher NO content of 2 ppm. In trace E the amount of NO in an ex-
haust specimen from an automobile is seen to be over 50 ppm.
The spectrophone is useful for detecting absorbed tunable laser radiation because it is sen-
sitive (for tunable lasers with moderate power levels), independent of wavelength, and operates
at room temperature. An experiment was performed to compare the absorption signal from a
spectrophone with the simultaneously transmitted signal through the cell, monitored with a
liquid-helium-cooled Ge:Cu infrared detector. A tunable Pb- ggsn0 <2Te ^i°de laser was used,
and the radiation was chopped mechanically before entering the cell. The lower trace of Fig. B-9
is the spectrophone signal proportional to the absorbed laser power p ; the upper trace repre-
sents the infrared detector signal, proportional to transmitted power p., where the dashed line
corresponds to 100% transmission. The traces are seen to resemble the expected relation to
each other, according to the expression p = 1 — p .
I vL
By frequency-modulating the laser emission, rather than using amplitude modulation, the
signals for transmission and absorption should be identical, except for phase. Derivative de-
tection was used to compare the minimum detectable concentrations of the two techniques. Using
a 10-|xm Pb_ ggSn,) ^p"1"6 diode laser having a CW power level of 6 ^W, and a 10-cm long cell,
the detection limit for C2H4 using the spectrophone was a few hundred ppm, whereas that em-
ploying the liquid-helium-cooled infrared detector was 1 ppm under the same conditions. The
absorption coefficient for the C.H. line was approximately 10~ cm" /ppm, yeilding a total ab-
sorbed power of P o' cL - 6 x 10'^W, which is close to the noise-equivalent-power of the Ge:Cu
infrared detector. Conversely, the Johnson noise limit for the spectrophone imposed by the high-
— 8
impedance circuitry is approximately 10" W (Ref. 26), making the spectrophone less appropriate
for systems involving tunable laser sources of relatively low power.
Frequency modulation of the laser emission eliminates the need for a mechanical chopper.
For a semiconductor diode laser, this is accomplished by superimposing a small (~1 mA) sinu-
soidal current upon the steady current, and detecting the first derivative at the modulation fre-
quency, or the second derivative at the first harmonic.
Figures B-10 and B-ll illustrate application of this technique to the point-sampling of auto-
mobile and smokestack effluents. In Fig. B-10 is a set of first-derivative scans for C,H4 near
near the Q branch of the ^_ vibration-rotation band at 10.6 \im. The upper trace represents a
2000-ppm mixture of C^H. i° N?' at a tota^ pressure of 5 Torr, and is used for calibration. The
three lower traces were obtained for different samples of raw automobile exhaust, for which the
total pressure was also reduced to 5 Torr. The presence of C^H. is unmistakable because of the
identical signatures; and by comparing their amplitudes with that of the calibrated sample, a
quantitative determination can be made of the C2H4 content of each. There is no noticeable inter-
ference by any of the other components such as water vapor, CO, NO, and other hydrocarbons,
which are also present in large quantities.
Figure B-ll shows results for a similar study of SO, in a sample of stack gas from an oil-
fired power plant. Using a tunable PbQ Q3Sn0 0?Te diode laser in the 8.7-^m region of the i^band,
64
-------
182O
t
s
IBIS
J L
J i L_L
I I I I
I I
J I
ig?o IBIS
FREQUENCY |cm~')
Fig. B-8. Detection of NO with an opto-acoustic detector using a tunable spin-flip Raman
laser; (A) calibration scan of 20 ppm NO in N2; (B) noise level with laser blocked; (C),
(D), and (E) are scans for samples of laboratory atmosphere, ambient air near a busy
highway, and exhaust gas from an automobile, respectively. All measurements were
performed at approximately 300 Torr total pressure. Lines 1, 5, 6, 8, and 11 are due to
NO — the rest are caused by water vapor. Reprinted with permission of the American
Association for the Advancement of Science. After Kreuzer and
65
-------
so
(a) T • 0 03 §«e
£
3 I0
I ' I
(b) r • 3 sec
[• 0.1cm"1 1
IOSO 1100 1120 1I4O
DIODE CURRENT (mA)
Fig. B-9. (a) Transmission scan for 02^4 in air using a Pb^_xSnxTe diode laser
in the iO-jim region and a liquid-helium-cooled Ge:Cu detector, (b) Absorption
scan over the same spectral region using opto-acoustic (spectrophone) detection.
Cell length is 10 cm, C2^4 pressure is 5 Torr, and air pressure is 15 Torr.
66
-------
Calibration
Exhaust 1
V.
13.5 mV (2000 ppm C2H4in N2)
1.0 mV (148 ppm)
I
I
949.18
949.20
949.22
949.24
949.26
WAVENUMBER (cm"1}
Fig. B-10. First-derivative spectra, taken with a Pbg ggSng ^Te diode laser, of:
-------
the derivative scan yields a value of 670 ppm for the concentration of SO, in the specimen. As
with the automobile exhaust measurements, there appear to be no interferences from other gas-
eous components which are present. There is, however, a periodic background fluctuation caused
by Fabry-Perot-type feedback from windows of the sample cell, which can be eliminated by proper
design. It should be noted that these lines of SO, are not the strongest in the v band, and that
the derivative signal would be 20 times larger if a more appropriate wavelength region were
i * j16
selected.
3.3.2 In Situ Source Monitoring
A tunable-diode-laser system for across-the-stack monitoring of SO, is presently under
construction at our Laboratory, The system, which is illustrated in Fig. B-i 2, yields an average
value for the SO^ concentration across the stack. This should be more representative than point-
sampled values for predicting the total amount of SO, emitted into the environment. Preliminary
Retro-reflector Irr)
Smokestack
Shutter
Digital
Printer
Detector
Electronics
KJ V
en in r
Fig. B-12. Diode laser system for across-the-track monitoring
of pollutant gases in a smokestack.
work involved an extensive study of the 8.7-(im SO2 band; the locations and strengths of over
200 lines were made by tunable-diode-laser spectroscopy, and compared with a theoretical
model. Other parameters, such as line-broadening coefficients, were also measured, and for
SO2 in the atmosphere the full linewidth at half-maximum intensity was found to be 0.3 cm"1.
For the detection of pollutant gases in the atmosphere, where linewidths are relatively wide,
pulsed-diode laser techniques can be used. By using pulses of current, rather than steady val-
ues, the cryogenic-cooling requirements are less stringent, and higher output laser powers can
be achieved since larger injection currents are possible. The diode laser can still be tuned by
a superimposed direct current. The power p. received at the detector is related to the power p
from the laser by the Beer-Lambert equation
i'cL] , (3)
68
-------
where the other symbols were defined earlier. With the aid of an integrate-and-hold circuit in
conjunction with a logarithmic detector, as shown in the schematic of Fig. B-13, the recorded
voltage is directly proportional to the pollutant concentration, c.
Application of this technique to on-line automobile exhaust monitoring was demonstrated by
measurements made using a 1 .1 5-meter-Long optical path. Exhaust gas from an automobile was
passed through a windowless tube through which the laser beam was also directed. Test results
for a 1972 8-cylinder station wagon are shown in Fig. B-14 for two starting cycles: (a) a normal
start where the accelerator pedal was not depressed; (b) a "rich" start. The measurement is
essentially instantaneous, limited only by the rate of change of the C2H4 content in the exhaust.
The C2H4 concentration of around 300 ppm during idle was confirmed by taking a sample and
using the first-derivative technique at reduced pressure, described earlier.
3.3.3 Ambient-Air Monitoring
For ambient-air monitoring, where pollutant levels are usually much lower than they are
near sources, sensitive detection can be achieved by long-path transmission or with the aid of
multireflection cells. Although multireflection cell techniques are very susceptible to mirror
contamination, tunable lasers can, by scanning alternately "on" and "off" an absorption line,
eliminate most of the adverse effects caused by such variations in reflectivity. Of course, con-
sideration must be given to the atmospheric "windows" when selecting appropriate wavelength
regions.
At low pollutant concentrations, Eq. (3) can be approximated as follows:
Pt=Po(l-a^cmL) =po-ap , (4)
where 6 is the minimum detectable change in power. Solving for the minimum detectable pol-
lutant concentration, c , we obtain
m
(5)
Under laboratory conditions, presently available infrared detectors can detect power variations
of less than 10 W. In actual application, however, fluctuations in received laser power im-
posed by atmospheric turbulence, scattering, and equipment vibration may degrade this value.
-9
If we assume that the minimum detectable power 6 is 10 W, that the laser power p is 100 ^W
-5-1
--
a' = 10 cm /ppm, and a path length L of 500 meters, the minimum detectable concentration,
cm, is 0.02 ppb.
We have recently transmitted 10.6-nm radiation from a PbQ 8gSn0 12Te diode laser over a
500-meter path, with a collection efficiency of 33 percent. The system used the electronic
schematic shown in Fig. B-13. With the beam directed horizontally about 4.5 meters above a
parking lot, an average increase of several ppm of C,H4 was detected during the late afternoon
exodus of automobiles. The noise level for this test, with 2 (i\V of received laser power, was
120 ppb, caused primarily by relative vibrations in the optics.
4. CONCLUSION
There appear to be several important uses of tunable lasers for the monitoring of atmos-
pheric pollutants, and some systems are already under construction. This field is very new,
-------
1
1
Current
Programme
Sync
I
i
Intt graft -
ond-
hold Circuit
[ Reference
Veltage
Later
Detector
Fig. B-13. Operational schematic for general application of pulsed semiconductor
diode lasers to the monitoring of atmospheric pollutants at emission sources or in
the ambient air.
Fig. B-14. Results of an on-line emissions test for C2H4 in a 1972 station wagon,
using a pulsed Pbg ggSng ^Te diode laser operating at 10 57 pm (a) represents
a normal start of a warm engine, with ignition at time t = 0. For (b) the acceler-
ator pedal was depressed during ignition, to produce a "rich" start, with signif-
icantly higher 02^4 emission.
70
-------
less than two years old for the most part, and although requirements of low-temperature opera-
tion limit widespread use of some tunable lasers at present, improvements in both laser tech-
nology and advances in cryogenic cooler development should eventually eliminate this drawback
for future monitoring applications.
ACKNOWLEDGMENTS
The author would like to thank H. A. Pike for permission to use
pre-publication results of his spectrophone measurements and on-line
automobile emissions data, R. S. Sinclair for designing the electronic
circuitry for pulsed-laser monitoring, A. R. Calawa and T. C. Harman
for developing the diode lasers, and K. W. Nill and F. A. Blum for pro-
viding their nitric oxide spectra. Appreciation is also expressed to
J.O. Sample, T. E. Stack, and L.B. McCullough for very competent
technical assistance.
71
-------
REFERENCES FOR APPENDIX B
1. For a comprehensive review, including an extensive set of references to earlier
work, see P. L. Hanst, "Advances in Environmental Science and Technology,"
Volume 2 (John Wiley & Sons, Inc., New York, 1971, edited by James N. Pitts, Jr.
and Robert L. Metcalf), Chapter 4.
2. E.D. Hinkley and P. L. Kelley, Science l_7i (1971) 635-639.
3. E.D. Hinkley and R.H. Kingston, Proc. Joint Conference on Sensing of Environ-
mental Pollutants, Palo Alto, California, November 8-10, 1971.
4. H. Kildal and R. L. Byer, Proc. IEEE 59 (1971) 1644-1663.
5. I. Melngailis, IEEE Tcans. on Geoscience Electronics GE-10 <1972) 7-17.
6. C.F. Dewey, Jr., and L.O. Hocker, Appl. Phys. Lett. 1_8 (1971) 58-60.
7. D.J.Bradley, Proc. Electro-Optical Systems Conference, Brighton, England,
March, 1971; B.B. Snavely, Proc. IEEE 57 (1969) 1374-1390.
8. T.W. Hansch, I. S. Shahin, A. L. Schawlow, Phys. Rev. Lett. 27 (1971) 707-710.
8a. S. A. Tuccio and F. C. Strome, Jr., Appl. Optics 11 (1972) 64-73; A. Dienes,
E. P. Ippen, and C. V. Shank, IEEE J. Quantum Electron. 8 (1972) 388.
9. S. E. Harris, Proc. IEEE 57 (1969) 2096-2113; R.G. Smith, "Optical Parametric
Oscillators," in Laser Handbook (North-Holland Publishing Company, Amsterdam),
edited by F.T. Arrechi and E. O. Shultz-DuBois (to be published).
10. J. Pinard and J.F. Young, Optics Comm. 4(1972) 425-427.
11. R. L. Byer, VII International Quantum Electronics Conference, Montreal, Canada,
May 8-11, 1972.
12. For a general review, with extensive references, see T.C. Harman, "The Physics
of Semimetals and Narrow-Gap Semiconductors," proceedings of the Conference
held in Dallas, Texas, 1970 (Pergamon Press, New York, edited by D. L. Carter
and R. T. Bate); also, T.C. Harman, J. Phys. Chem. Solids Supplement 12 (1971)
363-382.
13. E.D. Hinkley, T.C. Harman, and C. Freed, Appl. Phys. Lett. 13 (1968) 49-51.
14. E.D. Hinkley, Appl. Phys. Lett. 16 (1970) 351-354.
15. K. W. Nill, F.A. Blum, A. R. Calawa, and T.C. Harman, Appl. Phys. Lett. 1^
(1971) 79-82; , Chem. Phys. Lett, (to be published).
16. E. D. Hinkley, A. R. Calawa, P. L. Kelley, and S. A. Clough, J. Appl. Phys. 43
3222 (1972).
17. C.K.N. Patel and E.D. Shaw, Phys. Rev. Lett. 24 (1970) 383-385; A. Mooradian,
S. R. J. Brueck, and F.A. Blum, Appl. Phys. Lett. 17 (1970) 481-483.
18. C.K.N. Patel, Phys. Rev. Lett. 28 (1972) 649-652.
19. C.K.N Patel, E.D. Shaw, and R. J. Kerl, Phys. Rev. Lett. 25 (1970) 8-11.
20. L.B. Kreuzer and C.K.N. Patel, Science HI (1971) 45-47; R.A. Wood,
R. B. Dennis, and J. W. Smith, Optics Comm. _4 (1972) 383-387; S.R.J. Brueck,
E. J. Johnson, and A. Mooradian (private communication).
21. J. A. Rossi, S. R. Chinn, and A. Mooradian, Appl. Phys. Lett. 20 (1972) 84-86.
21 a. I. Melngailis (private communication).
22. V. N. Bagratashvili, I. N. Knyazev, and V. S. Letokhov, Optics Comm. 4^(1971)
154-156; N.G. Basov, E. M. Belenov, V. A. Danilychev, and A. F. Suchov,
Quantum Electron. (USSR) N3 (1971) 121-122; C. A. Fenstermacher, M. J. Nutter,
W. T. Leland, and K. Boyer, Appl. Phys. Lett. 20 (1972) 56-60.
23. E.D. Hinkley and C. Freed, Phys. Rev. Lett. 23 (1969) 277-280.
24. R. J. Keyes and T. M. Quist, "Semiconductors and Semimetals" (Academic Press,
New York, 1970, edited by R. K. Willardson and A. C. Beer), Chapter 8.
25. M.R. Bowman, A.J, Gibson, and M.C.W. Sandford, Nature 21 (1969) 456-457.
26. L.B. Kreuzer (private communication).
72
-------
BIBLIOGRAPHIC DATA
SHEET
1. Report No.
EPA-R2-73-218
3. Recipient's Accession No.
Title and Subtitle
Development of In Situ Prototype Diode Laser System
to Monitor SCL Across the Stack
5. Report Date
May 1973
6.
. Author(s)
E. David Hinkley
8> Performing Organization Rept.
No.
Performing Organization Name and Address
Lincoln Laboratory
Massachusetts Institute of Technology
P. O. Box 73
Lexington, Massachusetts 02173
10. Project/Task/Work Unit No.
11. Contract/Grant No.
68-02-0569
12. Sponsoring Organization Name and Address
Environmental Protection Agency
Division of Chemistry and Physics
Research Triangle Park, N.C. 27711
13. Type of Report Si Period
Covered
Final, 11/71-3/73
14.
15. Supplementary Notes
16. Abstracts
This report describes the development and testing of a semiconductor diode laser
system to monitor sulfur dioxide by differential absorption of infrared radiation. Laser
material was prepared and diodes fabricated which would operate in a temperature-
independent region of SC>2 absorption. Data concerning sensitivity and interferences
from aerosols and other gases were recorded in the laboratory. Field tests were then
performed at an operating coal-burning power generating station, with the results com-
pared with SO™ measurements taken with a conventional chemical monitor.
17. Key Words and Document Analysis. 17o. Descriptors
tunable infrared laser
differential absorption
infrared detection
in situ monitoring
17b. Identifiers/Open-Ended Terms
17c. COSATI Field/Group
18. Availability Statement
Release unlimited
19, Security Class (This
Report)
UNCLASSIFIED
20. Security Class (This
Page
• UNCLASSIFIED
21. No. of Pages
78
22. Price
FORM NTH-II 110-70)
USCOMM-DC 40329-P71
73
------- |