svEPA
United States
Environmental Protection
Agency
Industrial Environmental Research
Laboratory
Research Triangle Park NC 27711
EPA-600/7-80-091
May 1980
Investigation of CARS
and Laser-induced
Saturated Fluorescence
for Practical Combustion
Diagnosis
Interagency
Energy/Environment
R&D Program Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments of, and development of, control technologies for energy
systems; and integrated assessments of a wide range of energy-related environ-
mental issues.
EPA REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved
for publication. Approval does not signify that the contents necessarily reflect
the views and policies of the Government, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/7-80-091
May 1980
Investigation of CARS and Laser-induced
Saturated Fluorescence for
Practical Combustion Diagnosis
by
A.C. Eckbreth, P.A. Bonczyk, and J.F. Verdieck
United Technologies Research Center
East Hartford, Connecticut 06108
Contract No. 68-02-3105
Program Element No. INE623
EPA Project Officer: William B. Kuykendal
Industrial Environmental Research Laboratory
Office of Environmental Engineering and Technology
Research Triangle Park, NC 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
DISCLAIMER
This report has been reviewed by the Industrial Environmental Research
Laboratory, U.S. Environmental Protection Agency, and approved for publication.
Approval does not signify that the contents necessarily reflect the views and
policies of the U.S. Environmental Protection Agency, nor does mention of
trade names or commercial products constitute endorsement or recommendation for
use.
11
-------
FOREWORD
A broad based continuing research program is being sponsored by the
Industrial Environmental Research Laboratory at Research Triangle Park to
reduce emissions from stationary sources. A part of this research program is
concerned with reducing air emissions via modification of the combustion
process. In order to better achieve this goal, it is necessary to understand
on a fundamental basis the combustion process. Combustion diagnostic measure-
ments play a key role in developing this understanding.
The work reported herein covers research on two key measurements of
significant interest in combustion research: combustion temperature measurements
and specie concentration of CH, CN, and NO.
William B. Kuykendal
Process Measurements Branch
Industrial Environmental Research
Laboratory
Research Triangle Park
111
-------
ABSTRACT
Under Contract 68-02-3105 sponsored by the Environmental Protection
Agency, the United Technologies Research Center (UTRC) has conducted experi-
mental investigations aimed at developing nonperturbing, spatially precise,
in-situ diagnostic techniques to measure species composition and temperature
in flames. The contract continued the development initiated under Contract
68-02-2176 of coherent anti-Stokes Raman spectroscopy (CARS) and laser-in-
duced saturated fluorescence. The program was divided into two main, concur-
rent tasks. In Task I, Optical Thermometry, the practical feasibility of CARS
has been demonstrated in a program of research scale combustor testing. In
Task II, Optical Composition, laser-induced saturated fluorescence has been
examined in regard to its capability for measuring CH, CN, and NO concentra-
tions in flames. Saturation of the fluorescence in CH and CN was achieved
and considerable insight into the physics of saturated fluorescence has been
obtained. Promising, initial results of NO fluorescence in flames are de-
scribed but saturation was not observed for laser spectral intensities up to
6(106) W/cm2 cm'1.
IV
-------
Investigations of CARS and Laser-Induced
Saturated Fluorescence for Practical Combustion Diagnosis
TABLE OF CONTENTS
Page
DISCLAIMER ii
FOREWORD iii
ABSTRACT iv
TABLE OF CONTENTS v
LIST OF FIGURES vii
SUMMARY 1
INTRODUCTION 4
CONCLUSIONS AND RECOMMENDATIONS . . 6
TASK I OPTICAL THERMOMETRY
CARS INVESTIGATIONS IN PRACTICAL COMBUSTORS 8
Introduction 8
Combustion Facility 10
Portable CAPS Instrument 10
Experimental Program 21
TASK II OPTICAL COMPOSITION
INVESTIGATIONS OF LASER-INDUCED SATURATED
FLUORESCENCE FOR TRACE SPECIES DETERMINATIONS 37
Introduction 37
Laser-Induced Saturated Fluorescence Theory 40
-------
TABLE OF CONTENTS (Cont'd)
gage
Experimental Investigations 47
General Discussion 86
REFERENCES 88
VI
-------
LIST OF FIGURES
No^ Title Pages
1 Low Pressure Fundamental Combustion Facility 11
2 Combustors Used in CARS Demonstrations 12
3 Postrun Condition of Combustion Tunnel Windows 14
4 Schematic Layout of Portable CARS Apparatus 16
5 BOXCARS Setup in Jet Burner Stand 17
6 Portable CARS Instrument in Jet Burner Stand 18
7 Horizontal Dispersion Output Optical Arrangement 19
g Temperature Variation of ^ CARS Spectrum 22
9 Thermometry Algorithm for No CARS Spectra 24
10 CARS Transmission Through a 60M Dia , 20m Long,
Optical Fiber 25
11 Temperature Variation with Stoichiometry in
Model Combustor 26
12 Spatial Variation of Temperature in Model Combustor
with Jet A Fuel 28
13 Stoichiometric CARS Temperature Variation in Swirl
Burner with Refractory Back Wall 30
14 Single Pulse CARS Thermometry in Swirl Burner with
Refractory Back Wall 32
15 Operational Temperature Variation in JT-12 Exhaust 33
16 BOXCARS Temperature Profile of JT-12 Exhaust 35
17 Single Pulse BOXCARS Thermometry in JT-12 Exhaust 36
18 Modified Two-Level Model for Saturated Fluorescence
with Rotational Redistribution 44
Vll
-------
LIST OF FIGURES (Cont'd)
Figure No. Title Pages
••• • ' •- -"~" • — -Jfc •.•
19 CH and CN Energy Level Diagrams ,.
20 NO Energy Level Diagram
21 Laser Induced Fluorescence Apparatus ,.„
22 Laser Induced Fluorescence Apparatus Photograph „
23 Stilbene Dye Laser Pulse „
24 Rayleigh Pulse 57
25 CH Spectra 5g
26 Principal Laser-Excited CH Lines -Q
27 Laser Excited CH Flame Fluorescence ,,
28 CH Concentration Data Reduction g-
29 Energy Level Diagram for CN g.
30 CN Saturated Fluorescence Experimental Arrangement g«-
31 Experimental Arrangement for CN Saturated
Fluorescence ^j
32 Emission Spectrum of CN Violet System
(B2S—X22) 69
33 Laser-Induced Fluorescence of CN Radical ^Q
34 Laser-Induced Fluorescence of CN Radical -,-,
35 Laser-Induced Fluorescence of CN Radical 72
36 Laser-Induced Fluorescence of CN Radical 73
37 Laser-Induced CN FluorescencerSaturation Behavior c
viii
-------
FigureNo,
LIST OF FIGURES (Cont'd)
Title Pages
33 Saturated Fluorescence Power Versus Inverse Laser
Spectral Irradiance 76
39 Horizontal Scan of Laser-Induced Fluorescence
in CN Slot Burner 78
40 Laser Induced UV Fluorescence Apparatus 79
4! NO Spectra 82
42 NO Fluorescence Pulses 83
43 Laser Excited NO Flame Fluorescence 85
IX
-------
SUMMARY
Under Contract 68-02-3105, sponsored by the Environmental Protection
Agency, the United Technologies Research Center (UTRC) has conducted experi-
mental investigations aimed at developing non-perturbing, spatially precise,
in-situ diagnostic techniques to measure species composition and temperature
in flames. The contract continued the development initiated under Contract
68-02-2176 of coherent anti-Stokes Raman spectroscopy (CARS) and laser-induced
saturated fluorescence. The program was divided into two main, concurrent
tasks. In Task I, Optical Thermometry, the practical feasibility of CARS has
been evaluated in a testing program in a research scale combustor. In Task II,
Optical Composition, laser-induced saturated fluorescence has been examined in
regard to its capability for measuring CH, CN, and NO concentrations in flames.
The feasibility of coherent anti-Stokes Raman spectroscopy for remote,
spatially and temporally resolved, measurements in practical combustion systems
has been demonstrated. Experiments were performed in a 50-cm dia. combustion
tunnel with generous optical access, located in UTRC's Jet Burner Test Stand.
The tunnel could be fitted with either a swirl burner or shrouded JT-12 can.
The swirl burner could be fired with either gaseous or liquid fuels and per-
mitted access to primary and secondary combustion zones. The JT-12 can was
shrouded, restricting measurements to its exhaust region. The CARS instrument
consisted essentially of transmitter and receiver optical assemblies and detec-
tion instrumentation. The CARS transmitter comprised the two laser sources
required to generate CARS, and the requisite optics and transmitter lens to
focus and cross the mixing optical beams in the combustor. The transmitter was
arranged on a 3' x 6' optical pallet mounted on a portable cart. The pallet
could be translated to allow measurements at various radial positions in the
tunnel. The receiver optics were mounted on rails and linked to the transmit-
ter permitting tandem translation of the optics. The CARS signal emerging from
the combustor was processed by the receiver and piped out of the burner stand
test cell employing 20 m long, 60|x dia. fiber optic guides. The fiber optic
conduit delivered the CARS signal to the detection equipment, a 0.6 m spectro-
eraph fitted with an optical multichannel analyzer, located in the control
room, a more benign environment for delicate optical equipment. The optical
multichannel analyzer permitted the capture of single 10 sec CARS spectra and
time averaging of the spectra from regions of the combustion tunnel where the
temporal fluctuations were modest. Thermometry is performed from the shape of
the CARS signatures; in the tests reported herein CARS was generated always
from No. Crossed-beam phase matching or BOXCARS was employed in all of the
tests to remove any potential spatial resolution ambiguity. This also provided
a more stringent test of the practical feasibility of CARS than the more
conventional collinear phase matching.
-------
Successful tests were conducted first in the swirl burner with gaseous
propane as the fuel at several stoichiometries. Then Jet A fuel was run and,
since no problems were encountered, all subsequent testing was with the liquid
fuel. The Jet A fueled tests were quite impressive in that the combustion
luminosity was extremely intense, difficult to view for any extended period.
With the Jet A fueled swirl burner, high quality averaged and single pulse CARS
spectra were obtained at several different spatial locations and stoichiome-
tries, including the spray zone. In most cases, the spectra rival the quality
of those obtained in laboratory studies. To simulate a furnace more closely,
a refractory back wall was placed flush with the exit of the swirl burner.
This caused no particular problem and good spectra were again obtained. Mea-
surements were also made in the exhaust of the Jet A fueled JT-12 can. High
quality averaged and single pulse CARS spectra were obtained at several sets of
conditions. The temperature profile of the exhaust was obtained near the exit
at the so-called cruise condition. CARS temperature measurements further down-
stream in the exhaust agreed quite well with those determined by an aspirating,
thermocouple probe. All in all, these experiments constituted a fairly rigorous
examination of the feasibility of CARS for practical combustion diagnosis.
Under Task II, Optical Composition, the technique of laser-induced
saturated fluorescence was investigated for measurement of CH, CN, and NO
radical concentrations in laboratory flames. These species were chosen because
they are of extreme importance in combustion chemistry and pollutant formation.
The laser-induced saturated fluorescence method is important for two reasons:
the more important is the elimination of quenching corrections in determining
concentrations; the second is that the maximum possible signal is obtained from
a given set of experimental conditions. These two factors taken together
contribute to a measurement technique of good precision and high sensitivity.
The CH, CN, and NO radicals absorb light in the blue (4300 A), near UV
(3880 A), and far UV (2260 A), respectively. Correspondingly, dye laser
systems were constructed to provide short pulse length, 0(10" sec), narrow
spectral width, 0(0.1 cm" ), radiation at the appropriate wavelengths. Each
of these systems was optically pumped with a 10 Hz repetition rate pulsed
neodymium.'YAG laser; either the second harmonic (2xNd:YAG = 5320 A) or
third harmonic (3xNd:YAG * 3533 A) was employed. The resulting dye laser
radiation was used directly to excite the radical fluorescence (in the case of
CH), or mixed with the 1.06u. neodymium fundamental (lxNd:YAG) to generate UV
radiation (for CN). For the excitation of NO fluorescence, both a frequency
doubling of the dye laser radiation and subsequent sum-frequency mixing with
1.06^ was required. The resultant laser intensity was low, insufficient to
saturate the NO fluorescence.
Laser-induced fluorescence of CH and CN was observed from the same premixed
slot-burner using acetylene-oxygen, and acetylene-nitrous oxide, respectively.
Prior to determining saturation curves for these two species, high resolution
-------
fluorescence spectra were obtained by slowly scanning the appropriate regions.
It is necessary to determine the laser-induced fluorescence spectrum of a
particular species for several reasons. Probably, the most important reason is
that examination of the spectrum provides some assessment of rotational relaxa-
tion, an assessment essential to proper data reduction. For experimental
purposes, the spectrum also suggests the best combination of transition to
excite and fluorescing transitions from which to collect signal (to minimize
interference and maximize signal). Moreover, the actual fraction of fluores-
cence collected to total fluorescence must be determined from the laser-induced
spectrum.
Saturation curves, plots of laser-induced fluorescence signal against
laser intensity, were obtained by attenuating the laser beam with neutral
density filters. The CN radical fluorescence was more easily saturated than
that of the CH fluorescence, as expected. Concentrations of these species were
determined from the fluorescence power data according to a modified two-level
saturated fluorescence model. The model, developed in the report, presents the
effects of rotational relaxation in a simple, direct manner. Using a frozen
rotation data reduction scheme, concentration measurements found in these
investigations agreed to within a factor of 2 to 3 with previous absorption
measurements. Quenching rates were also obtained from the fluorescence
power data.
Laser-induced NO fluorescence spectra were obtained from NO-doped methane-
air flames. However, saturated fluorescence was not observed because the
incident spectral irradiance fell short of the saturation values predicted by
calculation. It is estimated that with the use of commercially available dye
lasers, the saturated regime for the NO fluorescence can be attained.
-------
INTRODUCTION
With the increasing availability of laser sources, light scattering and
wave mixing spectroscopic techniques are assuming an ever-increasing role in a
broad spectrum of physical investigations. Of particular importance is the
application of laser spectroscopy to the hostile, yet easily perturbed, envi-
ronments in which combustion occurs. Nonintrusive laser diagnostics should
facilitate greatly improved understanding of a variety of combustion processes
which, in turn, should lead to enhanced efficiencies and cleanliness in energy
propulsion, and waste disposal systems.
Under the sponsorship of the Environmental Protection Agency, the United
Technologies Research Center (UTRC) has been conducting analytical and experi-
mental investigations aimed at developing nonperturbing, spatially precise, in
situ diagnostic techniques to measure species concentrations and temperature in
flames. The present contract (68-02-3105) continued the development initiated
under Contract 68-02-2176 of coherent anti-Stokes Raman spectroscopy (CARS) and
laser-induced saturated fluorescence. The program was divided into two main,
concurrent investigations, the results of which are reported herein. Under
Task I, Optical Thermometry, the practical feasibility of CARS has been assessed
in a program of research scale combustor testing. In Task II, Optical Composi-
tion, laser-induced saturated fluorescence has been examined in regard to
its capability for measuring CH, CN, and NO concentrations in flames.
By way of perspective, it is instructive to review the results of the
first contract (68-02-2176) which was divided into two serial tasks. In Task
I, a comprehensive review (Refs. 1-3) was conducted of potential, unobtrusive,
in situ techniques for chemical composition and temperature measurements in
flames with particular emphasis on hostile environments such as research scale
furnaces. The review focussed upon four laser techniques which appeared most
promising, namely, spontaneous and near-resonant Raman scattering, laser-induced
fluorescence, and CARS. Although they have received considerable attention for
combustion diagnostics (Refs. 4, 5), spontaneous and near-resonant Raman
scattering appeared to possess a low probability of successful application to
highly luminous, particle-laden flames even with advanced state-of-the-art
laser sources. Laser fluorescence appeared capable of measuring species
concentrations to tens of ppm in practical flames and tenths of ppm in clean
flames for selected molecules whose absorptions could be saturated. In this
saturation approach, fluorescence magnitudes do not depend upon or can be
experimentally corrected for quenching effects. The technique is applicable to
several molecules of combustion interest, primarily radicals such as NO, OH,
CH, CN, 0.2, and NH. CARS was perceived capable of successful thermometry and
major species measurements in practical environments, although potential
jeopardies such as laser-soot interaction effects had then to be addressed. it
should be noted that the two techniques, for the most part, are complementary
in regard to their capabilities.
-------
In Task II, experimental investigations of saturated laser-excited
molecular fluorescence and CARS were conducted and are reported in Ref. 6.
Saturated laser-induced fluorescence measurements using a tunable, flashlamp-
pumped dye laser, were performed on the radicals CH and CN in atmospheric
pressure flames (Ref. 7). Partial saturation of the fluorescence was achieved
for both molecules with readily achieved laser spectral intensities in the
(• £ O — 1
10 to 10 Watts/cm cm range. In order to test the validity of the
fluorescence results, absorption measurements were made to independently
determine the concentrations. The concentrations measured by absorption were
found to be larger by about a factor of two for CH and five for CN than the
saturated fluorescence values. It was believed that the fluorescence results
were low, due to an overestimate of the fluorescence (i.e., species) sample
volume arising from strong spatial gradients of concentration in the small,
high pressure flames studied. Nevertheless, the experiments demonstrated the
viability of saturated, laser-induced fluorescence for trace species detection
of molecular radicals. Based upon these studies, it was proposed to refine the
experimental approach, reexamine CH and CN and then proceed to attempt satur-
ated fluorescence detection of NO. The experimental refinement consisted
primarily of replacing the flashlamp-pumped dye laser with a dye laser pumped
by an appropriate harmonic of a Q-switched neodymiunr.YAG laser. Such dye
lasers typically possess far superior beam quality (i.e., focusability),
narrower linewidths, higher intensities, and higher pulse repetition rates
than flashlamp-pumped lasers.
The CARS investigations conducted demonstrated its promising potential
for temperature and species probing of practical combustion environments. CARS
spectra from N£ over premixed flames displayed excellent agreement with
computer generated model predictions. Single pulse (10 seconds) CARS
spectra were obtained in flames and displayed the capability of CARS for
"instantaneous" temperature measurements. Quite importantly, CARS was success-
fully performed from N^ in highly sooting flames indicating that CARS should
be applicable to practical systems. Although thermometry received the major
emphasis in the CARS investigations, CO species sensitivity was also examined.
From these investigations, a logical next step in the development of CARS was
to examine its practical utility in tests in actual research scale combustors.
Based upon this foundation, the present contract was structured into the
two parallel investigations mentioned earlier. The next section summarizes
the conclusions reached during these investigations and contains recommendations
for the future development of these techniques which are a logical extension to
the investigations described herein. Then, the Task I Optical Thermometry
investigations are described. Therein, the successful application of CARS
thermometry, both averaged and single pulse, in liquid-fueled practical combustors,
is reported. The last section contains the results generated under the Task II,
Optical Composition investigations. With the improved experimental setups, the
measurements of CH and CN are now closer to the previous values determined by
absorption and considerable insight into saturated fluorescence physics has
been obtained. Promising, initial results of NO fluorescence studies are also
described.
-------
CONCLUSIONS AND RECOMMENDATIONS
Coherent Anti-Stokes Raman Spectroscopy
The feasibility of coherent anti-Stokes Raman spectroscopy (CARS) for
remote, spatially and temporally resolved,, measurements in practical combustion
systems has been demonstrated. N2 CARS temperature measurements are reported
at atmospheric pressure for two different liquid-fueled combustors, a swirl
burner, and a JT-12 combustor can, situated in a 50-cm dia. tunnel. Crossed-
beam CARS, i.e., BOXCARS, was employed to ensure good spatial precision.
Delicate instrumentation was housed in a control room adjacent to the burner
test cell, and the CARS signals were piped out employing 60P dia. fiber optic
guides. Average temperature measurements and single pulse thermometry were
performed. Based upon these and other investigations, CARS may be anticipated
to see widespread applicability to a variety of practical combustion devices
and problems.
In attempting to assess the potential for the routine application of CARS
thermometry to combustion research, several points are worth bearing in mind.
Expense: The capital equipment required for a complete CARS system including
computer controlled data acquisition and reduction would be on the order of
$150K to $200K. For an automated system, considerable software development
would be required, perhaps two man-years worth. CARS computer codes would need
to be developed, two man-years, or purchased. To operate the system, an engineer
and technician would be required. Operating expenses, other than wages, would
not be large, perhaps $10K/year. The major hurdle, financially, would be the
initial outlay for equipment. Personnel: Initially, a fairly experienced sci-
entist would be required to bring into operation and maintain the system. After
a period of time, the technology could be transferred to lesser experienced (in
CARS) personnel with periodic supervision or consulting with someone knowledga-
ble in the technique and, preferably, near-by or in house. Presently, scien-
tists/engineers experienced in CARS with application to combustion research
are scarce. Applications; No use of CARS at the present time can be called
routine in the sense that a low temperature (£, 1200 °K) thermocouple measure-
ment might be called routine. For fundamental flame studies in the laboratory
it can become "routine" to an experienced scientist in a relatively short time.
Practical applications, certainly initially, tend to be distinctly unique,
requiring new approaches and twists. For example, the step from measurements
in a one atmosphere, 50 cm dia. tunnel well equipped with optical access, to a
25-40 atm, annular, combustor can perhaps twice as large, or a 3m dia. furnace
is neither easy or straightforward. On the other hand, it is not impossible,
-------
and one should not view such applications as improbable. To date, CARS has
been successfully demonstrated in an internal combustion engine and simulations
of furnaces, gas turbine combustors and other turbomachines. Based on these
recent advances, one may anticipate that CARS will provide for considerable
advances in combustion science in the coming years.
Laser-Induced Saturated Fluorescence
The application of the laser-induced saturated fluorescence technique
for spatially-resolved, remote measurement of radical concentrations in a flame
was confirmed by this investigation, although data reduction questions still
need to be resolved. Resolution of these issues would greatly enhance measurement
accuracy. In particular, CN and CH radical concentrations were measured; both
species were readily saturated with laser-pumped dye-lasers of 10 nanosecond
pulse duration, narrowband ( ~ 0.1 cm~ ) spectral width, and reasonable beam
divergence (2-3 milliradians). The extrapolated detectivity limits for concen-
tration measurements based upon the dye lasers used in this study for a clean
flame environment are: CN ~ 2 ppm; CH ~14 ppm. These estimates are based
upon a minimum detectable number of photons equal to 100 and a quantum efficiency
of 0.25.
Laser-induced fluorescence spectra of NO were obtained from an NO-doped
flame in this investigation, but the saturated fluorescence regime could not
be attained because of insufficient laser power at 2260 A. Because of the
extreme importance of nitric oxide as a pollutant, further investigation of the
saturated fluorescence measurement of NO, using a more powerful dye laser, is
certainly warranted.
Finally, it is recommended, that, in order to truly understand how
rotational relaxation affects the interpretation of saturated fluorescence
data, computer-generated, time-dependent solutions to the rate equations be
generated and studied. Time-dependent solutions for some general cases have
already been obtained by R. J. Hall of this laboratory. Such computer solu-
tions, with time-dependent variation of the pertinent rates, could help greatly in
understanding the data reduction problem, particularly if compared with well-
designed experiments. Rates for rotational energy transfer can be measured
experimentally by observing the time dependence of the rotational structure of
the laser-induced fluorescence spectra.
-------
TASK I - OPTICAL THERMOMETRY
CARS INVESTIGATIONS IN PRACTICAL COMBUSTORS
Introduction
Coherent anti-Stokes Raman spectroscopy, CARS, (Refs. 2, 8-10 and the
references therein) has received considerable attention in the last several
years for remote combustion diagnosis because of its potential applicability
to practical systems, e.g.,furnaces, gas turbine combustors, etc.. Spontane-
ous Raman scattering has also been widely investigated in this regard (Refs.4
5) but, due to its weak signal strength and incoherent character, is generally
limited to use in relatively clean flames. In practical combustion environments
laser induced incandescences from soot (Refs. 2, 11, 12) or fluorescences from
fuel fragments (Ref. 13) can mask detection of the Raman signals, often by
several orders of magnitude.
CARS is a nonlinear, light wave mixing process capable of both high
spatial and temporal resolution and is described in detail in the aforemen-
tioned references and the reports (Refs. 1, 6) prepared under the preceed-
ing EPA Contract 68-02-2176. Very briefly, laser beams at frequencies w an(j
u>2» termed the pump and Stokes frequencies respectively, with a frequency
difference appropriate to the molecular species being probed, are focussed and
"mixed" by overlapping to generate a laser-like CARS signal beam. Temperature
measurements derive from the spectral distribution of the CARS radiation (Ref.
14), concentration measurements from the signal intensity (Ref. 15) or, in
certain ranges, also from spectral shapes (Ref. 14). The practical potential
of CARS has been exhibited in several investigations and recently was displayed
when CARS was employed to map the temperature field throughout a highly sooting
laminar propane diffusion flame (Ref. 16), and for single pulse measurements,
in a small highly turbulent flame (Ref. 17). Despite these advances, concern
over practical feasibility has been exhibited in regards to the possibly
deleterious effects of large scale turbulence, turbulence over a large physical
dimension and the presence of liquid fuel droplets. Large scale turbulence
i.e., scales on the order of the laser beam diameter, and strong spatial
gradients can cause steering of the laser beams and deterioration of their
focussing properties (Ref. 18). Small scale turbulence also degrades phasefront
coherence and, when integrated over a substantial path, results in reduced
focal volume intensities. Fuel droplets can attenuate, scatter and refract th
incident laser beams. These effects can seriously degrade CARS signal generatio
efficiency. There may exist situations where these distortions are so severe
that CARS or optical probing of any sort will be precluded. To address these
points, CARS investigations in practical combustion environments were performed
under Task I of the contract. Specifically, this section of the report describe
-------
successful CARS measurement demonstrations, both averaged and single pulse, in
the primary zone of a highly swirled burner fueled with either propane (g) or
Jet A, and in the exhaust of a JT-12 can burning Jet A. These combustors were
selected to evaluate CARS in several different types of combustion regions.
CARS has been recently demonstrated in an internal combustion engine
(Ref. 19)> a bluff-body stabilized, gaseous propane diffusion flame in a
combustion tunnel (Ref. 20), and in the exit plane of a kerosene-fueled burner
(Ref. 21). Subsequent to the investigations described herein, the combustion
tunnel CARS experiments reported in Ref. 20 were extended to liquid fuels as
well (Ref. 22). Due to the scale of a typical internal combustion engine and
its repetitive behavior, a CARS approach different than that required for
probing large volume, nonstationary sources can be employed. In nonstationary
systems, spectrally broadband Stokes lasers must be used to generate and
capture the entire CARS spectrum with each pulse. In a cyclical process, a
narrowband Stokes laser can be scanned synchronously with the engine cycle to
generate the CARS spectrum piecewise but, nevertheless, temporally resolved
vis-a-vis the engine cycle. The magnitude of turbulent fluctuations at a given
point in the cycle cannot be obtained, however, unless the broadband approach
is used. For small-scale devices with modest radial gradients, collinear
mixing of the input beams can probably be employed without loss of spatial
resolution, although the validity of this approach should be verified.
The experiments reported here differ from the tunnel work in Refs. 20, 22
in several respects. Here crossed-beam phase matching or BOXCARS (Ref. 23)
was adopted over collinear mixing. By crossing the incident laser beams, in
a manner similar to laser velocimetry, CARS is generated only at the beam
intersection. This manner of phase matching yields fine and, perhaps more
importantly, unambiguous spatial resolution. In turbulent combustion environ-
ments, this approach provides a stringent test for CARS since the crossing
beams traverse different optical paths. Further, longer optical paths were
examined in this work since the tunnel used was, by chance, 50 cm in diameter
in contrast to the 25 cm tunnel employed in the (Refs. 20, 22) studies.
Lastly, delicate instruments, such as the spectrometer and optical multichannel
analyzer, were not located in the burner stand test cell, but rather in the
adiacent control room. The CARS signals upon exiting from the combustor were
focused into a 20-meter long, 60-udia., optical fiber and transmitted to the
remotely located instrumentation. Remote detection of CARS signals using fiber
ootics can be highly useful, particularly for diagnostic application in intense
vibratory environments (Ref. 24).
In the next section, the combustors and combustion tunnel employed in
these experiments will be described. The CARS instrument assembled for these
tests will then be discussed followed by a presentation of the experimental
results.
-------
Combustion Facility
The CARS experiments were performed in the combustion tunnel schematically
illustrated in Fig. 1. The tunnel was fitted with either a swirl burner or
shrouded JT-12 corabustor can shown in Figs. 2a and 2b, respectively.
Characteristics of the swirl burner and combustion tunnel are summarized in
Table 1. The tunnel is of double-wall, stainless steel construction and
water-cooled. It is 50-cm in internal diameter and 150-cm long and contains
four rows of rectangular quartz windows, 6.4 cm x 12.8 cm, every 90° about the
tunnel axis. When the windows are not in use, steel insert plates are used
to blank off the tunnel. Each row contains eight window locations, numbered
sequentially beginning upstream. Due to the expense of large quartz windows
5 cm dia. windows in a specially constructed blank were used instead for the
CARS apertures. The windows were offset from the combustor outer wall by a
cylindrical standoff purged with dry nitrogen. The unobstructed window aper-
ture was approximately 3.8 cm. In Jet A fueled tests, the windows would coat
with soot on the time scale of an hour. Lesser deposition occurred where the
high laser intensities passed, as seen in Fig. 3; in effect the laser beams
were self cleaning. A similar observation was made in Ref. 19. Although the
tunnel can be operated at pressures up to four atmospheres, all of the tests
reported here were at a constant pressure of one atmosphere.
The swirl burner, Fig. 2a, is modeled after those developed at the
International Flame Research Foundation (Ref. 25) and is of stainless steel
construction. It contains a movable block swirl generator with an adjustable
swirl number range of 0 to 2.5. The fuel nozzle tube is 1.5 cm dia. inside
a 6.6 cm dia. air duct which diverges to 12.7 cm dia. at its exit plane, as
shown. The quarl exit plane is approximately 21 cm from back wall of the
tunnel and 8 cm upstream of the center of window #1. The JT-12 combustor
can, shown unshrouded, is pictured in Fig. 2b. Its exit diameter is compar-
able to that of the swirl burner, and it protrudes into the tunnel 30 cm from
the back wall. The can is welded into a cylindrical shroud to duct the air
into and around the combustor. In the experiments to be described the air
mass flow was held approximately constant at 0.15 Ib/sec for the swirl burner
and about 1 Ib/sec for the JT-12 can. The swirl burner swirl number was held
constant at approximately 1.2. Air preheat was used for the JT-12, but not for
the swirl burner.
Portable CARS Instrument
Figures 4 and 5 illustrate the optical layout of the portable CARS
instrument and its arrangement about the combustion tunnel. Two different
output orientations, nearly optically identical, were used, a horizontal
arrangement shown in Fig. 4 and the vertical setup depicted in Fig. 5.
A photograph of the CARS instrument in the burner stand test cell is shown in
10
-------
FIGURE 1. ATMOSPHERIC PRESSURE COMBUSTION FACILITY
• TRANSITION DUCT
AIR
SWIRL BURNER
•4 ROWS OF 8 QUARTZ WINDOWS
WATER-COOLED, DOUBLE-WALLED, CONSTRUCTION
•COMBUSTOR SECTION (60 IN. LONG X 20 IN. DIA)-
ID
I
O
I
01
-------
FIGURE 2. COMBUSTORS USED IN CARS DEMONSTRATIONS
SWIRL VANE ADJUSTMENT INDICATOR
(a)
QUARL NECK
v SWIRL
I I k VANES
AERODYNAMIC j j
PROBE - J
(b)
I
AIR
STATIC PRESSURE
FUEL
COOLING
WATER
79-09-!.
18
L2
-------
TABLE 1
UTRC ALTERNATIVE FUELS COMBUSTION FACILITY
General Characteristics and Specifications
Burner :
Type
Construct ion
Swirl generator
Swirl number range
Fuel nozzle diameter
Air annulus inside diameter
Air annulus outside diameters
Exit quarl diameter
Fuel types
Air preheat temperature
Operating pressure range
Nominal air flowrate
Fuel flowrate range
Swirl stabilized, air/fuel annular burner
Stainless steel
Movable block
Adjustable from 0 to 2.5
1.5 cm (0.60 inch) max.
1.9 cm (0.75 inch)
6.6 cm (2.6 inch), 5.1 cm
(2.0 inch), 4.1 cm (l.b inch)
12.7 cm (5.0 inch)
Gaseous, liquid, or pulverized solids
400 °C (750 °F) max.
1 to 4 atm.
240 kg/hr (525 Ib/hr)
8 kg/hr (17.6 Ib/hr) to 14.3 kg/hr
(31.5 Ib/hr)
Firing rate range
3000,000 Btu/hr to 600,000 Btu/hr
Cotnbustor
Type
Construction
Inside diameter
Length
Optical windows
Double walled, water cooled can combustor
Stainless Steel
50 cm (20 inch)
150 cm (60 inch)
4 rows of 8 quartz windows each
located at 0", 90°, 180°, 270° around
combustor
Window size
Wall cooling features
6.4 cm x 12.8 cm (2.5 inch x 5.0 inch)
rectangle
20 uniformly distributed wall segments,
each individually controllable
Operating pressure range
1 to 4 atm.
13
-------
•-
I
FIGURE 3. POSTRUN CONDITION OF COMBUSTION TUNNEL WINDOWS
-:
O
--
K
M
U
-------
Fig. 6. The horizontal output arrangement is shown in the Fig. 7 photo-
graph. All of the CARS components were mounted on a portable 3' x 6' Newport
Research Corporation optical pallet which rode on pillow blocks and rails
permitting the radial measurement location to be varied. The rails were
attached to an angle iron frame fitted with heavy duty pneumatic tire casters
and four permanently attached scissors jacks to raise the instrument to the
proper height. The size of the pallet was dictated primarily by the size of
the Quanta-Ray frequency-doubled neodymium:YAG laser. For simplicity, and
since size was not a major problem, more sophisticated packaging arrangements
were not investigated.
o
The laser emits two beams at 5320 A by sequentially doubling the fundamental
1.06U and residual 1.06^ from the first doubler. The primary beam energy is
about 200 mJ/pulse, 10" sec pulse duration at 10 Hz. The secondary is about
an order of magnitude lower. Referring to Fig. 4, the secondary green is
dispersed from the remaining 1.06P by a series of prisms and is used to
optically pump the Stokes dye oscillator. The Stokes oscillator output is
amplified in a dye cell in flow series with the oscillator dye cell. The am-
plifier is pumped by a fraction of the primary split off at a 33 percent beam
splitter. The Stokes laser dye spectrum was centered at 6067 A, midway be-
tween the ground and first vibrational state Raman shifts of N2, the molecular
species used for all of the thermometry experiments reported herein. Nn is
ideal for CARS thermometry investigations in airfed combustion processes, be-
cause it is the dominant molecular constituent and is present everywhere in
high concentration independent of the extent of the chemical reaction. The dye
spectrum was centered in comparison with a neon lamp spectrum by adjusting the
concentration of the rhodamine 640 dye (Exciton Chemical Co.) dissolved in
ethanol. The dye spectrum was approximately 170 cm'1 (63 A) wide FWHH.
The Stokes laser output is magnified by a factor of two or three, passes
through an optical flat which can be rotated to vary the BOXCARS phase-matching
angle (Ref. 23) and reflects off the dichroic D to the beam crossing and
focussing lens. The remaining primary beam is split into two components in the
manner shown; these components are aligned parallel and sent to the crossing/
focussing lens after transmission through the dichroic. Based upon the size
of the combustor and the desire to be able to radially vary the measurement
location, long focal length crossing/focussing lenses, i.e., 480 mm, were
employed. This diminishes the CARS generation efficiency, e.g., relative to
a laboratory situation, due to the strong dependence of the signal on laser
intensity. The pump beams were crossed at a half angle of 1.1" or 0.77° de-
pending on when the tests were performed. With the latter configuration, the
spatial resolution was experimentally measured by generating CARS from within
a translatable 1 mm thick microscope slide. Ninety-percent of the spatially
integrated CARS was generated over a 1.2 cm axial extent. The measurement
volume is a cylinder approximately 0.5 mm in diameter by 1.2 cm long oriented
15
-------
FIGURE 4. SCHEMATIC LAYOUT OF PORTABLE CARS APPARATUS
2xNd: YAG LASER
'!! \ LL
OF
L L
^
1 . W
BACK H |
WALL H
(OPTIONAL)
FUEL-C^Z
COMBUSTION
TUNNEL
TO SPECTROGRAPH
16
-------
FIGURE 5. BOXCARS SETUP IN JET BURNER STAND
PORTABLE CARS
INSTRUMENT
ID
o
a>
I
M
-------
-
.
FIGURE 6. PORTABLE CARS INSTRUMENT IN JET BURNER STAND
Kl
'
I
-------
:-
-
FIGURE 7. HORIZONTAL DISPERSION OUTPUT OPTICAL ARRANGEMENT
0
-
-
-------
transverse to the flow axis. The axial resolution would be slightly better at
the larger crossing angle,
-------
the requirement for transfer optics (Ref. 24). The CARS transmission effi-
ciency through the spectrometer was approximately 30 percent, the loss occur-
ring since the f number of the spectrometer ( ^6.5) exceeded that of the optical
fiber ' 3-6. Except for small input and output coupling losses, about 10
ercent collectively, all of the CARS was transmitted through the low loss optical
fiber (< 10 dB/km). The spectral resolution with this arrangement at the ^
TARS frequencies was checked using a xenon lamp and found to be slightly better
than 2 cm .
Experimental Program
Cal_ibrat_ip_n
Prior to transport to the jet burner test stand, the CARS system was
alibrated by performing temperature measurements in a premixed CH^/air
flame. Previous studies at UTRC (Refs. 14, 27) indicated that CARS tempera-
ture measurements from NZ were accurate to within 40K at flame temperatures,
sing constant Raman linewidth computer codes. With improved codes, which
alculate the Raman linewidth as a function of temperature and rotational
uantum number, similar accuracies are also attained (Ref. 28). These refine-
ents occurred subsequent to the conduct of the investigations described.
Temperature is deduced from the shape of the CARS signature which is very
temperature sensitive as shown in Fig. 8. There, as an example, is displayed
rhe calculated temperature dependence of the ^ CARS spectrum at a resolution
f 1 cm""*- At t°w temperatures, one sees the v = 0 * 1 band containing low J
lue Q branch transitions, i.e. AJ = 0, where J is the rotational quantum
mber. As is apparent the low J Q branches are unresolved. As the temperature
' creases the band broadens as the rotational population distribution spreads
t due to the vibration-rotation interaction, <*QJ(J + 1). At high tempera-
es the spreading of the band is sufficiently large to permit the resolution
the individual even J Q transitions, ranging from Q(20) to Q(40). The odd Q
ansitions, which have a nuclear spin weighting equal to half of the even
bered transitions are reduced in intensity by a factor of four and do not
A out. At intermediate temperatures, few Q branch lines are resolvable.
r Q branch transitions beyond Q(40), overlap with the v = 1 -*• 2 band tran-
. • s OCcurs giving rise to two prominent peaks in the "hot" band, so termed
ause it appears only at higher temperatures. At the lower spectral resolu-
s 2 cm , employed in the experiments here, the fine structure shown in
*8 is not as pronounced. This is actually advantageous since the full
*dth of the ground state band at half peak height (FWHH) exhibits a smoother
W rature variation than at high resolution. Temperatures are deduced from
h N CARS spectra from the full width at half maximum of the ground state
d (FWHH) and the ratio of the hot band to ground state band peak heights
( hr) at elevated temperatures where a hot band is exhibited.
21
-------
FIGURE 8. TEMPERATURE VARIATION OF N2 CARS SPECTRUM
=0.8CM~1
SLIT = 0.75 CM-1
1.0-r
V)
z
Q
LU
N
OC
o
z
TEMPERATURE-DEG K
0.44-
21070
21090
FREQUENCY-CM-1
22
-------
Computer code calculations of N2 CARS spectra were performed in 100
K increments from 300°K to 2400°K for parameters corresponding to the
portable CARS instrument, i.e., Au^ » Q.8 cm"1, Ao>2 = 170 cm"*, slit = 2cm"1.
From these spectra, the FWHH and phr were obtained and are shown plotted as a
function of temperature in Fig. 9. These curves serve as a simple algorithm
to extract gas temperature from the N2 CARS spectra. At this resolution, the
FWHH variation is relatively smooth. The slight undulation at high temperatures
's caused by some Q transition fine structure. In all of the data to be
subsequently displayed, temperatures were deduced from the Fig. 9 curves.
Figure 10 displays a calibration point in a methane-air flame at 2110°K
sustained on a 2.5 cm dia. array of hexagonally arranged, stainless steel
hvpo tubes. At the flow rates employed, the temperature was measured using a
radiation-corrected, coated fine wire, thermocouple. The radiation correction
experimentally determined from sodium line reversal temperature measurements.
The spectrum exhibits the features displayed in Fig. 8 but at slightly lower
esolution. The dispersion on the optical multichannel analyzer was determined
sing a xenon lamp which has a multiplicity of transitions in the N2 CARS
avelength region. The phr of 0.33 yields a temperature of 2140°K and the
FWHH of 14.1 cm"1 indicates 2125°K in good agreement with the classically
sured temperature of 2110°K. Later in the report, comparisons are presented
f CARS temperatures and specially designed thermocouple probe measurements in
the JT-12 can exhaust.
Swirl Burner Measurement Results
The initial measurements on the feasibility of CARS in practical environments
conducted in the swirl burner operating on gaseous propane. The propane
iniected radially outward from the central fuel tube into the swirling,
nnular air stream to promote rapid fuel/air mixing. In these tests, attention
focussed primarily on the feasibility and quality of CARS generation rather
than on measurement acuracy per se.
In Fig. 11 are displayed CARS signatures of N2 for various stoichiometries
indicated by the overall equivalence ratio at the window #1 axial location
f s 6 cm downstream of the quarl exit plane) and slightly off-axis. The CARS
cctra were averaged on the optical multichannel analyzer (OMA) for 10-15
onds corresponding to 100-150 laser pulses. In situations where turbulent
oerature fluctuations are large, it is not valid to average the CARS spectra
extract average medium properties due to the nonlinear behavior of the CARS
. i level with temperature and density. In examining the real time display
the OMA, the spectra exhibited minor fluctuations and the averaged spectra
obably representative of the mean temperature. Single pulse spectra
23
-------
FIGURE 9. THERMOMETRY ALGORITHM FOR N2 CARS SPECTRA
s
2
300
600
900 1200 1500
TEMPERATURE - DEG. KELVIN
1800
2100
0.45
- 0.40
2400
-------
FIGURE 10. CARS TRANSMISSION THROUGH A 60p DIA., 20m LONG, OPTICAL FIBER
SPECTRUM FROM FLAME N2AT~2110 K
4/12/79-5a
-
2
> w
t- z
(75 °
m O
^ £
— o
:
•-
FREQUENCY-0.586 CM^1/DOT
c
.-
_
:
-------
-
ID
o
01
FIGURE 11. TEMPERATURE VARIATION WITH STOICHIOMETRY IN MODEL COMBUSTOR
AIR 0.15 Ib/sec, FUEL C3Hg(g); i = 5cm, x = 6cm; X1212
= 0;T = 350°K
1.0; 1500°K
5/23/79-9b
-3b
FREQUENCY
0.8; 1 750°K
1.3;1200°K
-0.586 cm"1/DOT
-6b
-5b
-------
will be displayed later. The Fig. 11 spectra are quite typical of N2
ctra as described earlier and seen in a variety of other flames (Refs. 14,
1 ft 26 27). For the high temperature data, the temperatures deduced from both
the FWHH and phr were in very good agreement, i.e. within 50K. Shown for
eference purposes in Fig. 11 is the CARS spectrum from flowing air just
fter combustor shutdown. The rounded and sloping nature of the lower right
d side of the Q-branch ground state bandhead is due to coma in the O.b m
r erny Turner monochromator and does not greatly affect data reduction.
-ause of the combustion tunnel volume and water-cooled walls, the peak
eratures produced are well below the nominal adiabatic flame temperature.
Th highest temperature recorded at the specific measurement location was at an
rail equivalence ratio, of 0.8 and was 1750°K. With increases in fuel
flow the temperature decreased to 1500°K at * = 1.0 and to 1200° at 4> = 1.3
probably to a shift in the peak heat release location farther downstream in
the tunnel.
The propane flame exhibited only a slight degree of sooting in the
urement location at which the Fig. 11 data were taken. Thus potential soot
nuation effects are likely to be negligible. This permits an assessment of
ffects of turbulence on CARS generation, at least in this instance.
toring out signal variations due to the density changes at the various
^ratures, the Fig. 11 spectra were of comparable magnitude over the range
f toichiometries examined. Furthermore, the magnitude of the ground state
d oeak scaled in accordance with computer code predictions (Ref. 14) when
flame was extinguished, i.e., when comparing the flame and cold gas signal
Is In addition, the cold gas signal showed little effect of turning the
flow on and off. Thus the presence of combustion and flow had little
fl ence on the efficiency of CARS generation other than that anticipated.
ianal levels of the Fig. 11 data are low due to optical fiber damage,
• u. resulted in over an order of magnitude attenuation in signal transmission
2h the fiber. The fiber was replaced prior to the Jet A fuel tests which
described below and to which a discussion of absolute signal levels is
deferred.
In Fig. 12 are shown data taken with Jet A fuel at two different axial
tions in the model combustor at an overall equivalence ratio of 0.8. With
A fuel the combustion process was extremely luminous and highly sooting.
i 1v it was difficult to view for any extended period. With the degree of
• 1 employe<^» the flame, as perceived from the soot luminosity, expanded
'Hi and appeared to fill most of the tunnel. At the window #1 location,
p^surements were made through the spray 11 cm downstream of the fuel
CAiv j mt; c* ^
1 and the temperature was found to be about 900 K. Moving downstream to
jf\ where the flame was very luminous and turbulent, the temperature
window ii J y
ed to 1500 K. The hot spectrum displayed in Fig. 12 is of very high
i • t- and as fine as those typically produced in laboratory experiments, e.g.
qU ,Q it was obtained after averaging for 10 sec (100 pulses) and the per
27
-------
?
FIGURE 12. SPATIAL VARIATION OF TEMPERATURE IN MODEL COMBUSTOR WITH JET A FUEL
AIR 0.15 Ib/sec, 0 = 0.8 r = 5mm
x = 6cm T = 900°K
5/25/79-2b
= 38cm T = 1500°K
-10b
•
FREQUENCY-
-»- 0.586 cm-1 /DOT
0)
-------
Ise peak height count is about 350 on the optical multichannel analyzer,
hich corresponds to about 4400 photons in a thin spectral slice about the
u This is about a factor of eight below the intensity one would calculate
f r the experimental parameters employed (Ref. 2), but quite typical of the
al magnitude agreement between pulsed CARS calculations and experiments,
n in well characterized situations (Refs. 9, 10). The strength of the
1SOO°K spectrum also scales appropriately, to within a factor of two, from the
Id gas signal levels. Furthermore, the signals were comparable to those
htained in laboratory simulations in flat flames. All of this is indicative
h t at this equivalence ratio, 0.8, the combined effects of soot and turbulence
rhis case were not very severe. Single shot measurements in the spray,
at x = 6 cm, indicated fairly large fluctuations in temperature, and,
averaging of the spectra to extract a mean temperature there is suspect.
The temperature listed is a density weighted average, probably to the low side
f the true mean temperature, and likely to be in error by 100K. or more. There
also evidence that the efficiency of CARS generation was lower in the region
f the spray, perhaps by an order of magnitude. A more thorough investigation
Id be required to quantize these effects rigorously. Nevertheless, these
• 'rial investigations of CARS generation through the spray are encouraging.
"'rh proper window purging, operation with Jet A fuel posed no particular
blems relative to gaseous fuels except for the increased window sooting
dency. Due to this fact, all further testing was conducted using the liquid
fuel.
In an attempt to simulate a furnace more closely, a refractory back wall
fabricated from asbestos millboard and placed flush with the burner exit
i ne as sketched in Fig. 4. This alteration posed no particular problems
the CARS measurements. CARS measurements were again made at window location
-i and are displayed in Fig. 13 for three different stoich iome tr ies with Jet
t ei The small temperature changes which occur with variation in the
ichiometry are indicative of the temperature sensitivity of the CARS spectra.
before the temperatures are believed accurate to about 50K. At the overall
uivalence ratio of 0.8, the signal level as before, is consistent with
ectations. With increases in fuel flow to overall equivalence ratios of 1.0
d 1 2, the CARS signals decreased by factors of about three and ten, respectively,
least half of these decreases are attributable to increased window fouling,
• e when the fuel was quickly leaned out, the signal returned to only half of
former level. Despite the attenuation experienced at the higher soot
dines, laser-induced soot interferences, sometimes encountered in highly
ting flames (Ref. 17) were not seen. For thermometry, which derives from
hape of the spectrum, the attenuation by soot should cause no particular
hlem as long as the CARS signals remain acquirable and with good photon
. t£cs For density measurements, which depend upon the absolute signal
. this will be more problematical. The attenuation will have to be
'ed and then corrected for, a procedure likely to degrade measurement
^curacy- One approach would be to place parallel CARS reference cells (Ref.
tsTboth'before and after the test region; these would not only normalize
ulse-to-pulse signal variations in the usual manner, but permit estimation of
the^oot extinction encountered.
e
-------
FIGURE 13. STOICHIOMETRIC CARS TEMPERATURE VARIATION IN SWIRL BURNER WITH REFRACTORY BACK WALL
JET A FUEL, AIR FLOW 0.15 Ib/sec; x= 38cm;
OVERALL EQUIVALENCE RATIO, TEMPERATURE
0.8, 1450°K 1.0, 1540°K
1.2, 1400°K
0, 300°K
7/9/79-10
7/9/79-12
7/9/79-15
7/9/79-20
O
VJ
8
FREQUENCY-
0.576 cm~1/DOT
-------
As alluded to earlier, in highly turbulent environments characterized
bv large fluctuations in density and temperature, the CARS spectrum must be
aptured and temperature information extracted with each pulse. By repeating
these measurements a statistically significant number of times, one is able,
barring sampling biasing errors, to obtain the temperature probability distri-
b tion function, pdf. From the pdf, the average temperature at any one location
nd the magnitude of the turbulent temperature fluctuations there can be
scertained. In Fig. 14 is shown a single 10 nanosecond CARS spectrum of
at window #3 in the swirl burner fueled with Jet A. A single pulse was
ntured from the 10 Hz train using the electronic shutter with an aperture
a ration of ~75 millisec. This permits capture, with high probability, of only
single pulse. Also shown for comparison purposes, is the CARS spectrum
erased for 13 seconds (~130 pulses). The single pulse spectrum is of
liehtly lower quality due to the weakness of the signal (~500 counts at peak)
a the high dark current on the DMA vidicon. The latter leads to poor noise
btraction over the ten cycle scan used to "read" the tube. Some irregularity
the single pulse spectrum also occurs due to spectral structure on the
ctokes dye laser, i.e., it is not perfectly smooth. These dye spectrum irre-
larities average out in time. The data displayed in Fig. 14 demonstrate
he feasibility of single pulse CARS thermometry and, when coupled with computer
trolled data acquisition, will permit pdfs to be acquired in a reasonable
unt of time, e.g. 1-2 minutes. The single pulse spectral temperature at any
tant should not necessarily agree with the average temperature and depends
the scale of the turbulent fluctuations. The single pulse temperature was
1A85°K from phr and 1550°K from FWHH. The hot band contains about 50 counts at
, an(j thus is susceptible to a statistical variation of + 14%, i.e.,
P-l/2 or a phr temperature uncertainty of + 60 K. The single pulse
surement error, conservatively, may be as high as ± 100 K. In assembling
obability distribution function, these statistical uncertainties would tend
verage out and the pdf would be more accurate than the individual single
shot measurement accuracy however.
12 combustor Exhaust Measurements
CARS measurements were also performed in the exhaust of a shrouded JT-12
stor can burning Jet A fuel. The air flow rate of about 1 Ib/sec is about
times that employed for the previously described swirl burner. The can
ses an exit diameter of 12.5 cm and CARS measurements were made 13 and 82
P a wnstream of the can exit. CARS signatures at the former location are
ed in Fig. 15. CARS measurements were made at the latter location
temperature probe data was available there for several operational
(Ref. 29). The probe used was a water-cooled, aspirating thermo-
_
(Aero Research Instruments, Model T-1006-6). In Table 2 is presented
C rison of the CARS and thermocouple measurements at several operational
together with the air and fuel flowrates and air inlet temperature for
sett ings &
31
-------
c
~-
I
FIGURE 14. SINGLE PULSE CARS THERMOMETRY IN SWIRL BURNER WITH REFRACTORY BACK WALL
-------
FIGURE 15. OPERATIONAL TEMPERATURE VARIATION IN JT-12 EXHAUST
CARS MEASUREMENT : 13.2 CM DOWNSTREAM OF CAN EXIT PLANE
FLOWRATES:
IDLE
CRUISE
AIR
1.07 LB/SEC
1.02 LB/SEC
FUEL (JET A)
40.9 LB/HR
52.5 LB/HR
TIN
333° K
513° K
AIR, 300°K
IDLE, 620°K
-.
-
Q
.
'
[
-
!
CRUISE, 985°K
6/29/79-1
6/28/79-15
FREQUENCY
0.574 cm-1 /dot
-------
those conditions. The variation on the thermocouple measurements is believed
to be representative of the resetability of combustion conditions in the JT-12
over a period of time. This is appropriate in the present context since the
probe and CARS measurements were made several weeks apart. Similar variations
in combustor operation were also found during a given run from the CARS measure-
ments. As can be seen, considering the circumstances, the agreement between the
two sets of data is fairly good and lends credibility to the CARS measurements.
With the horizontal dispersing arrangement (Fig. 4), the CARS measurement
location could be radially translated by about +_ 6 cm. In Fig. 16 is shown
the radial temperature profile from the JT-12 can at cruise conditions 13 cm
downstream from the can exit. As might well be expected, the temperature
profile is relatively flat. The temperature variation at the centerline is due
to variations in the fuel and air settings over a period of time. Similar
temperature variations could also be expected at the other radial locations if
more data had been taken. Also shown is a centerline measurement at the idle
condition.
In Fig. 17 is shown a comparison of a single CARS pulse and an averaged
CARS spectrum in the JT-12 exhaust at cruise. Due to the lower temperatures in
the JT-12 exhaust relative to the swirl burner and, hence, higher gas densities
the single pulse spectra are of higher quality and nearly identical qualitative!
to the average as can be seen. The" single pulse and average temperatures
deduced from FWHH are 980 and 1050°K, respectively. Here the discrepancy
results probably from some single pulse fine structure on the Stokes laser
which is evident in the single pulse CARS spectrum. Such effects could most
likely be minimized by ascertaining the temperature from a spectral fitting
routine, e.g., least mean squares. For all of the averaged JT-12 measurements
displayed (Fig. 15), the CARS signal to the OMA was attenuated by an order of
magnitude with a neutral density filter to prevent the vidicon from saturating
TABLE 2
COMPARISON OF CARS AND THERMOCOUPLE MEASUREMENTS
IN JT-12 COMBUSTOR EXHAUST
Operating Condition
Air Flow (Ib/sec)
Jet A Flow (Ib/hr)
Tinlet(°K)
TCARK>
Idle
1.07
40.9
333
58°
590 ± 50
Cruise
1.02
52.5
513
875
900 +
50
Max. Continuous
1.00
54.9
522
875
920 ± 50
34
-------
FIGURE 16. BOXCARS TEMPERATURE PROFILE OF JT-12 EXHAUST
1 3.2 CM DOWNSTREAM OF CAN EXIT PLANE
l-n
r
o
r
1200
1100
1000
z
> 900
ID
O
I
LLJ
tr
QC
111
1
UJ
800
700
600
500
400
300
-6
-4
A IDLE
A
_L
-2 0
RADIAL POSITION - CM
CRUISE
CONDITIONS
-------
FIGURE 17. SINGLE PULSE BOXCARS THERMOMETRY IN JT-12 EXHAUST
CRUISE CONDITION, P = 1 ATM. TCARS~1050°K
SINGLE PULSE (1C)-8SEC)
100 PULSE AVERAGE (10 SEC)
6/28/79-10a
6/28/79-9
l
i
f
-
f
FREQUENCY
0.574 cm"1/dot
-------
TASK II-OPTICAL COMPOSITION
INVESTIGATIONS OF LASER INDUCED SATURATED FLUORESCENCE
FOR TRACE SPECIES DETERMINATIONS
Introduce ion
There is an obvious need for diagnostic techniques which measure atomic
and molecular species concentrations in combustion media. Such techniques
will facilitate a better understanding of the kinetics of combustion processes
which, m turn, could lead to a more optimum utilization of alternate fuels
nd perhaps, to a reduction in pollutant emission levels. Two different
approaches to species measurement are probe and optical sampling. Probe
sampling suffers the disadvantages of physical intrusion into the medium under
study, and sample change by reaction in lines before analysis, for example, by
a mass spectrometer. Optical techniques have the advantage that they are
nonperturbing and capable of in-situ measurements. Of these, optical absorption,
laser-induced fluorescence spectroscopy (LIFS), and coherent anti-Stokes Raman
spectroscopy (CARS) are being used for species identification and for tempera-
ture measurement. Optical absorption is a well-established technique, but
inherently yields poor spatial resolution. The measured species concentration
•g in fact averaged over the optical absorption path length. For an inhomogeneous
edium, this average may differ significantly from the actual concentration at
any one point. In contrast, LIFS is a spatially precise technique which is
very sensitive and capable of trace species detection. Basically, the concen-
tration is found from the intensity of the LIFS emission. Historically, a
roblem associated with fluorescence techniques has been the need to make
uenching corrections to the measured fluorescence intensity. This is difficult
do accurately and, indeed, impossible unless constituent partial pressures
nd quenching rates are known a priori. Fortunately, the quenching problem can
h avoided if one employs optical saturation of the electronic transition
ducing tne LIFS emission. For this case quenching rate corrections are not
eded, or can be extracted from the saturated fluorescence data along with the
ecies concentration. Also, under saturation conditions the fluorescence
• enal is a maximum; hence, species detectivities are increased, particularly
instrumentally hostile environments such as flames or discharges. The CARS
chnique is capable of high spatial precision, and is regarded as complementary
LIFS in that is is applicable to temperature measurements and to major
ecies determination (concentration of 0.1% or higher). This is a most
rtunate coincidence because the species measurable with CARS, i.e., No,
H 0, CO, and C02, do not have strong electronic absorptions in the
ible-near ultraviolet spectral region and are precluded from LIFS measurements.
LIFS is not applicable to all atomic and molecular species since there
specific criteria which must be met. First, the molecule, upon excita-
• to a higher energy state by absorption of laser radiation, must not
• sociate prior to emission; obviously, in such a case, the molecule would
37
-------
have no emission spectrum. Unfortunately, this is the case with many poly-
atomic species. Second, the molecular absorption wavelength must be accessi-
ble to a tunable laser source in the UV-visible region. Third, it is necessary
to know the Einstein coefficient for spontaneous emission of the emitting
level. Finally, the uncertainties introduced by quenching of the fluoresence
must be dealt with. As indicated above, a way to avoid the quenching problem
is to saturate the absorbing transition with a laser having high spectral
intensity. The additional requirement for saturation of the laser-excited
transition restricts the number of molecular species which are viable candidates
for quantitative LIFS measurements. Section VII of Ref. 1 describes how
the choice can be made; Section III of Ref. 1 provides the necessary spec-
troscopic data for making the choice.
To date, saturated LIFS concentration measurements have been made for
a very limited number of molecular species. The earliest case is that of
C2 (Refs. 30, 31). CH and CN have been measured in flames in this
laboratory as described below. Much work has been performed on the OH radical
because of its easily resolved spectrum and ease of production (a moderately
abundant hydrocarbon flame component), but only one case of saturation of OH
radical fluorescence has been reported (Ref. 32). As regards other molecular
species, Pasternack, Baronavski, and McDonald (Ref. 33) recently have demon-
strated saturated LIFS for MgO. Recent lifetime measurements of the excited
(A2n+) SH radical by Becker and Haaks (Ref. 34) indicate that this
species may be a candidate as well. The SH radical is important in various
fast reactions which couple sulfur compounds in flames. Recent work along
these lines by MUller, Schofield, Steinberg, and Broida (Ref. 35) indicates
that in addition to SH, the radicals 82 and SO might be amenable to saturated
LIFS detection. With regard to atomic species, saturated LIFS has been demon-
strated for Na by Daily and Chan (Ref. 36), and by Smith, Winefordner and
Omenetto (Ref. 37), for Mg by Kuhl and Spitschan (Ref. 38), and for Tl,
Cr, Sr and Zr by Omenetto et. al., (Ref. 39). Atomic species serve as ideal
cases for saturation because the number of quantum states is small and the
electronic transitions are typically strong.
Under Contract 68-02-2176, sponsored by the Environmental Protection
Agency and carried out at UTRC, the laser-induced saturated fluorescence tech-
nique has been applied successfully to the measurement of CH and CN radical
species concentrations in laboratory-scale flames. The results of this work
are contained in the Task II Technical Report under the above Contract (Ref.
6), and have been reported in the open literature (Ref. 7). The saturated LIFS
measurements were performed on CH and CN in atmospheric pressure acetylene
flames with a tunable, low pulse rate, flashlamg-pumped dye laser. Fluorescence
was observed from the A A state of CH at 4315 A and from the B 2£ state
of CN at 3880 A. For both radicals, the dependence of fluorescence intensity
on laser spectral intensity exhibited a pronounced departure from linearity at
high laser spectral intensity, indicating saturation. These were the first
38
-------
observations of saturated fluorescence for these radicals, and the data permitted
evaluation of the species concentration for both CH and CN. Since the saturated
fluorescence technique was new and subject to some uncertainty, it was important
to have an independent measurement of radical concentration in order to test
the validity of the saturated fluorescence results. To this end, optical
absorption measurements were performed in a single pass on a specially constructed
slot burner operating at atmospheric pressure. The CH concentration was about
60 ppm in an oxy/acetylene flame, and the CN concentration approximately 150
ppm in a nitrous oxide/acetylene flame, both measured by absorption. The
concentrations measured by absorption were larger by about a factor of two for
CH and five for CN than the values determined by saturated fluorescence. A
possible explanation for the discrepancy suggested an overestimate of the
fluorescence sample volume size, which led to an underestimate of the species
concentration. That this overestimate may occur is due to the size of the
smallest possible focussed laser spot relative to the concentration gradients
known to exist across the radial dimension in small, high pressure flames. For
accurate saturated fluorescence results, the characteristic fluorescence sample
volume dimension should be considerably smaller than that of the flame spatial
gradient. Significantly, in the experiments reported in Refs. 6, 7, the laser
focal diameter was limited by the poor beam quality of the flashlamp-pumped dye
laser. With laser-pumped dye lasers, the beam quality is generally much
better, permitting finer scale probing. Such a laser system contributed
significantly to removing this ambiguity in the CH and CN saturated fluorescence
results reported herein.
The principal objectives of the present program were to remeasure CH
and CN by means of saturated fluorescence with improved precision and further,
to attempt to apply the saturated fluorescence technique to NO. To improve the
CH and CN measurements, a repetitively-pulsed, laser-pumped tunable dye laser
was employed. Specifically, in the work reported below, harmonics of a Nd:YAG
laser are used as dye laser pumping sources. Such a laser system offers a
number of advantages. The beam quality of the dye laser is good, which implies
a small beam divergence and a correspondingly small focussed laser spot. A
beam divergence of a few milliradians is readily achievable, which for a 10 cm
focal length lens yields a 100 to 200 micron diameter. Typically, this is an
order of magnitude smaller than the spot size reported in Ref. 6 and, therefore,
offers the possibility of eliminating the source of discrepancy mentioned
above. The smaller spot size also produces higher spectral intensities. The
laser-pumped dye laser may be made to operate in a narrow spectral bandwidth
which allows for very precise excitation of molecular transitions and equally
importantly, contributes to a high laser spectral intensity which is essential
to saturation. Since the laser spectral intensity is inversely proportional to
the duration of the dye laser pulse, the (5-10) nsec time of laser-pumped dye
laser pulses is distinctly superior to the (200-300) nsec associated with
flashlamp-pumped dye laser systems. The shorter pulse durations also minimize
species loss due to laser-enhanced excited-state chemistry and quenching to
39
-------
slowly-relaxing vibrational states. Finally, a repetitively pulsed laser permits
signal averaging of fluorescence emission to improve the signal-to-noise ratio
(S/N), and to scan the entire LIFS spectrum. Examples of the latter are given
in the results below. The optical absorption results for CH and CN given in Refs
6, 7 are again used for comparison. Based upon determinations of the radical
spatial gradient scale and the size of the probe beam used to make the absorption
measurements, the absorption measurements are suspected to be low. The CH and CN
fluorescence data were taken with the same burner used for the earlier optical
absorption measurements, and the exact operating conditions of the burner were
precisely repeated. The motivation for selecting NO for LIFS study stems from it
importance as a major pollutant and, experimentally, from its stability relative t
most radical species. NO may be contained indefinitely in a cell provided that
other reacting species, especially oxygen, are not present. Consequently, saturated
LIFS measurements may be carried out where the NO pressure is accurately known.
This permits a well-calibrated test of the saturated fluorescence results for NO
concentration. Another practical advantage with NO is that it may be doped into a
flame in known quantities which also calibrates saturated fluorescence results for
NO at flame temperatures.
Following this introduction, the general theory of laser-induced saturated
fluorescence is given, including a discussion of the characteristic time to
saturate a transition, and a development of a model which treats the importance
of rotational energy transfer in analyzing fluorescence data. This is followed
by three sections devoted separately to tesults for CH, CN and NO. Finally a
general discussion and evaluation of the saturated fluorescence results is
given.
Laser-Induced Saturated Fluorescence Theory
The original saturated fluorescence measurements on CH and CN performed
at UTRC (Ref. 6) used a flashlamp-pumped dye laser with a pulse duration
of 150-300 nanoseconds. A laser-pumped dye laser was used for the measure-
ments described in this report because, among other reasons, the shorter pulse
duration, 10 nanoseconds, yields much higher peak laser power. However, a
question which arises is whether steady state saturation of the fluorescence
transition can be achieved within the duration of a pulse of 0(10 sec).
Characteristic Time to Saturate
A straightforward analysis of the rate equations, under conditions of
saturation, provides the answer to this question of laser pulse length. The
complete rate equations for the upper and lower level populations N2 and
N,, respectively, are
40
-------
(i:
dt
dNi
f r>\
(2)
B, 2 and B2, are the Einstein coefficients for induced absorption and
induced emission, respectively, A2^ is the Einstein coefficient for spontaneous
emission, and Q2j is the electronic quenching rate. I is the laser spectral
irradiance and c the speed of light. If saturation occurs, then the condition
(B12 + B21)l/c » Q21 + A21 (3)
holds. Transitions between N^ and N2 are determined solely by the laser.
If the conservation equation Nj + N2 = Ntotai = constant is applied, then
Eq. 1 may be written as
^2= "tot8!!1 _ N2(B12 H. hl^ (4)
dt c c
may be integrated readily to
|(B19 + B91)lt
- Lf £i I . (5)
The constants C, and C2 are determined from initial (t < 0) and steady state
(t * °°) conditions:
for t < 0, N2 = 0, then Cj = -C,
NiB12
for t * ", N2 = -i-i±- = C1
B21
41
-------
Recall also Chat g^B^ * g2B21» where the 8 8 are the degeneracies of the
respective states; then the final solution is
- expj - (1 + 82/8i>B12It/c
Clearly, steady state is reached in a time Tgat $ c/B,,!
tgat can be evaluated for typical values of Bj2 and the laser spectral
irradiance I in order to find the characteristic time to saturate the transi-
tion. If the inequality (3) is true, then saturation occurs. A typical
value of the spontaneous emission rate, &2l» *s 10 sec . If Q~, = 10^ A =
1, then for the inequality (3) to hold, B12I/c •/« lO^sec"1. Thus,
Thus, a
time of a few picoseconds is sufficient to saturate transitions corresponding
to A21 * 108sec"1. For weaker transitions, A21 is of OClO^sec'1), the charac-
teristic time to saturate would be of 0(10 "sec). This latter figure is
still much shorter than a typical Q-switched laser pulse, 0(10~8sec). Hence
the ten nanosecond pulse used for the studies reported here is many times the
characteristic time to saturate. Times of the order of microseconds, often
used for saturated fluorescence studies, are not required, nor necessarily
desirable because laser-induced chemistry may occur on such a time scale. A
theoretical and experimental study of the time dependence of saturated fluores-
cence in atoms has been given by Olivares and Hieftje (Ref. 40). Their
conclusions regarding the time required to saturate are in general agreement
with those given here.
Rotational Energy Transfer Effects
Another question concerns the validity of the simple two-level, steady
state model employed by many authors. The two-level model is based upon an
atomic or molecular system which is essentially electronic in nature and has no
vibrational-rotational structure (nor spin). Daily (Ref. 41) has treated
the two-level and three-level cases. As would be expected, the two-level model
describes well the saturation behavior of atoms and some molecular cases. In
general, a molecular system is not a simple two-level system because of the
vibrational-rotational partitioning of the electronic levels. Vibrational-
rotational energy transfer, during the period of laser irradiation, must be
considered for the laser-induced saturated fluorescence of any molecular
system. For most molecules, vibrational relaxation can be neglected as being
slow compared to typical laser pulse lengths. This is not true for rotational
relaxation, which may have rates of 0(10* sec"*) or even higher.
One of the first treatments of laser-induced saturated fluorescence which
includes rotational relaxation is that of Lucht and Laurendeau (Ref. 42).
These authors treat two sets of rotational levels, of which selected levels are
42
-------
connected by laser excitation. A steady-state is assumed, and the resultant
algebraic equations are solved with a computer. The rotational relaxation rate
was varied relative to the electronic quenching rate. The significant result
of this study was that complete Boltzmann equilibrium of rotational levels is
not achieved unless rotational relaxation exceeds electronic quenching by a
factor of 0(10 ). A second important result is that a linear relationship
between the fluorescence power and the inverse of the laser spectral intensity,
similar to the two-level case, is found. An important modification is a
population factor which accounts for rotational level population redistribution
caused by collisional energy transfer. As defined in Ref. 42, the popula-
tion factor is the ratio of molecules involved in the laser-induced transition
to the total number of molecules in the electronic state of interest. In the
limit of very rapid rotational relaxation, the population factor goes over to
unity, that is, the entire electronic state participates in the transition.
For the other extreme limit of no rotational energy transfer during the laser
pulse, there is only one rotational state involved in the transition; and the
population factor is the inverse of the Boltzmann occupation probability
described in more detail below. In a simplification of the Lucht-Laurendeau
model, a similar relationship was found by Berg and Shackleford (Ref. 43)
wno, in one limit, considered the case of very fast rotational relaxation.
Again, a two-level model result is obtained with a population factor which is
essentially the entire population of the ground electronic state.
In order to apply the results of these models to the interpretation of
experimental data, it would be necessary to have data comparing rotational re-
laxation rates with electronic quenching rates. Unfortunately, these relation-
ships are not known for most radical species, certainly not for CH, CN, and
NO.
i fied Two-Level Model with Rotational Redistribution
The basic two-level model for saturated fluorescence can be extended in
a tractable manner as follows. Consider two energy level groups, shown as
boxes in Fig. 18, which represent the rotational levels of two vibrational-
electronic states. The upper group is empty, prior to excitation. The lower
group has a total population, Nj, of which some fraction f is excited by laser
radiation, IvL at an excitation rate B12 \\J^' Tne iaser radiation is assumed
narrowband, and selectively pumps only a particular rotational state of the
lower group. However, because of rotational energy exchange, nearby rotational
levels will feed into this selected level and participate in the transition.
The fraction of N, that participates is denoted by the factor f.
In the limit of no rotational relaxation, only one rotational level is
excited; and the factor f is the Boltzmann factor,
(2J + 1) exp(-Bhc J(J*
Pj = _ kT _ ' (7)
43
-------
FIGURE 18. MODIFIED TWO-LEVEL MODEL FOR SATURATED FLUORESCENCE WITH
ROTATIONAL REDISTRIBUTION
I
I
I
B12I/c
B21I/C
21
> ,
79-09-63-13
-------
where J is rotational quantum number and Q^Qf the rotational partition function,
For this limit, f is a small number, ~0.03, at flame temperatures for a mole-
cule with a rotational constant, BRQT ^,2 cm (e.g., N2, CN, C,,) . If some
rotational relaxation occurs during the laser pulse, f is larger than the
Boltzmann factor, Pj. f approaches unity for the other limiting case of very
fast rotational relaxation. In this limit then, the entire lower group of
rotational levels participate in the excitation process. N2 is the total num-
ber of molecules in the upper group. With this preface, the rate equation
for N9 is stated and solved in terms of the total population.
dN,
dt
= fN,
+ A
21
(8)
Assuming the steady-state approximation, the derivative is set equal to zero.
Using the conservation relation, NTota^ = NT = Nj + N
terms
of
N = N
T 2
1 +
!21
Recall
= g2B21, then
NT = N2(l + gj/fg2)
Inverting this equation for N
1 -t-
fB12I/c
21
one can solve for NT in
(9)
gl/g2f) B]2I
(10)
fNn
Fluorescence Power
(Q,
21
(f
A21)c
-1
(11)
Given N2, Eq. (11), the experimentally observed quantity, SF, the
fluorescence power, can be written
he
S = N,,
V21
(12)
45
-------
where h is Planck's constant; c is the speed of light; AF, is the fluorescence
wavelength; n is the light collection solid angle; and V is the volume of
sample fluorescing. e is an efficiency factor which accounts for spectrometer
throughput, lens losses, etc. This expression is substituted in Eq. (11)•
at the same time the term in the brackets is expanded as unity plus a small
term, because B12I/c » (Q2i + A2j). Hence, the fluorescence power is
S
F
\f 4TT (f + gj/gj)
1 -
A
21
(f + gl/g2)B12I
(13)
In the limit of very fast rotational relaxation, f -*. 1, and this model coincides
with the two-level model result of Baranovski and McDonald (Ref. 30). The
limit f = 1 corresponds to complete rotational relaxation, that is, the entire
lower group of levels participates in the laser-induced transition. Therefore
the entire lower group population can be measured. For f < 1, only a fraction'
of the total population is measured. In general, this fraction is unknown be-
cause the rotational relaxation rates are not available. For the extreme case
of very slow relaxation, only a single rotational state is excited and f becomes
the inverse of the Boltzmann factor, Pj, defined above. Sp plotted against the
reciprocal laser spectral intensity will yield a straight line, a procedure
first suggested by Baronavski and McDonald (Ref. 30). From the intercept
Sp, the total population NTQT can be found, if f is known or can be reasonably
est imated .
For the foregoing discussion, it is assumed that f is constant in time
which is true if the steady-state condition is true. But the steady-state
approximation may not be valid for real laser pulse shapes, for rotational
relaxation rates comparable to the induced absorption/emission rates, or for
quenching to slowly relaxing vibrational states. Because steady-state models
may fail to completely describe real molecular systems in saturated fluorescence
another approach is suggested. Time-dependent solutions to the population rate
equations should be examined and compared with steady-state solutions. Since
the rate equations are linear coupled differential equations, they are easily
solved through use of a computer. The computer can generate sets of solutions
incorporating a wide variation of parameters such as the rotational relaxation
rates, electronic radiative and non-radiative rates, as well as laser pulse
properties. In this manner, one can eventually determine the extent of rota-
tional relaxation for a particular situation, in essence, determine the f-factor
One direct way to accomplish this is to compare computer generated fluorescence
spectra with experimental spectra. Moreover, the availability of such synthetic
46
-------
spectra could recommend the most efficient manner of performing experiments
a particular molecular species in a given environment.
on
It should be stressed that it is the total fluorescence, N2A2,, which is
considered in the modified two-level model presented above. Figure 18 at-
tempts to indicate both radiative and nonradiative transitions, A2, and Q2i>
from N2 to states within and without the group fN^. The experimentalist has
the option, of course, of measuring all or some fraction of the total fluores-
cence, A21N2. If Rayleigh or Mie scattering is large, then the resonant
fluorescence region would likely be excluded.
Focal Beam Profile
It is assumed that in the excitation of the fluorescence power, uniform
illumination of the interaction volume is achieved. That is, the laser beam
has a "top hat" or rectangular spatial distribution. This is not a realistic
distribution for dye lasers, nor is the Gaussian distribution, often assumed.
However, a distribution with low intensity wings, like a Gaussian, is probably
not unreasonable. For such a case, if saturation occurs at a point in space
which is coincident with the peak of the Gaussian distribution, it may or may
not occur in the far wings of the distribution. If the laser spot size is
significantly smaller than the characteristic flame dimensions, some molecules
are saturated and others are not; this leads to errors in the measured species
concentration. Detailed considerations along these lines have been given by
Daily (Ref. 44). Pasternack, Baronavski, and McDonald (Ref. 33) have
analyzed saturated fluorescence data for Na using three different beam profiles
and observed the influence of this variation on calculated species concentra-
tion; these distributions were rectangular, Gaussian, and truncated Gaussian.
The Gaussian profile was shown to be unsatisfactory for making concentration
measurements, while the rectangular shape gave a result for Na concentration
more nearly equivalent to that determined from optical absorption than did a
truncated Gaussian. In light of this result and, in addition, the fact that
beam profiles in the present work are not Gaussian, the rectangular beam
approximation is used below. This simplified approach does not alter the fact
that a satisfactory and consistent procedure is needed to account for intensity
distribution, so as to make saturated fluorescence and optical absorption
methods agree as regards species concentration.
Experimental Investigations
In this section, saturated laser fluorescence experiments are described
for the radicals CH, CN and NO. As mentioned above, in order to apply the
saturated LIFS technique to these species, a radiation source is required which
mav be tuned to an absorption band of these molecules. For CH and CN, strong
47
-------
absorptions, which occur between the electronic ground state and an excited
state pertinent to this work, are at 4315 A and 3883 A, respectively.
Energy level diagrams for CH and CN are given in Fig. 19. CN differs from CH
in that a third electronic level,A w, is present. Each electronic level, in
the case of both CH and CN, has a vibrational-rotational structure as indicated.
For CH and CN excitation, a laser is needed which has 4315 and 3883 A nominal
operating wavelengths, respectively, and which may be tuned to individual
vibrational-rotational levels. An energy level diagram for NO is given in Fig.
20, where, as indicated, a laser at 2260 A is required. Unfortunately, it is
not possible presently to configure a single tunable dye laser which, without
modification, would be applicable to all three molecules. For CH at 4315 A,
a pulsed 3xNd:YAG laser at 3550 A is used to pump a dye laser employing a
stilbene/methanol solution. The laser wavelength is tuned by using a diffraction
grating in the laser cavity. When pumped as indicated, the wavelength at which
peak lasing energy occurs, coincides well with CH absorption lines. For CN,
the stilbene is not useful since when pumped as indicated it does not emit at
wavelengths suitable for CN excitation at 3883 A. Although an alternate dye
may be used, a significant reduction in laser pulse energy is likely to occur.
Indeed, even for 3xNd:YAG pumping of stilbene, the pulse energies are not
comparable to those obtained for 2xNd:YAG pumping of rhodamine 6G dye. Radi-
ation at 3883 A is obtained by using a 2xNd:YAG pump laser at 5320 A to
excite a rhodamine 640 dye which lases near 6100 A. This output is then
combined in a nonlinear type I KD*P mixing crystal with the fundamental 1.06/u
Nd:YAG output to yield frequency up-converted radiation at 3883 A. This is
an efficient process, despite its relative complexity, owing to the sizeable
energies available at 6100 A and 1.06/J.
In the case of NO, an approach is required which is different than that for
both CH and CN. Similar to the case for CN, a 2xNd:YAG pump laser is used to
excite, in this case, a rhodamine 590 dye which lases near 5740 A. This output is
then frequency doubled in a KD*P crystal to yield radiation at 2870 A, which in
turn is combined with 1.06 /x radiation in a second nonlinear mixing crystal to
obtain the required radiation at 2260 A. This is a more complex process than
those above and is capable of only moderate energy conversion efficiency. There is
an alternative approach to obtaining 2260 A radiation. One may elect to use the
3xNd:YAG pump to excite one of several coumarin dyes and thereby obtain lasing near
4520 A. Then, the output would be frequency doubled in a KPB (potassium pentabor-
ate) crystal to yield the desired wavelength. There were two reasons for not
selecting this approach in this work, although it might be worth evaluating experi-
mentally. First, blue dyes degrade rapidly when subject to laser pumping which
results in a sharp reduction in laser energy. This was a significant but not a
limiting difficulty for the present CH work with stilbene. Secondly, the frequency-
doubling efficiencies in the 4520 A region are known to be small, only a few
percent or less as verified by Dewey (Ref. 45). More detailed descriptions of the
laser sources for CH, CN and NO are given below.
48
-------
FIGURE 19.CH AND CN ENERGY LEVEL DIAGRAMS
25
CO
b
5
£
UJ
CD
r>
2°
15
I 10
LU
g
1
01
•A2A
,X2TT
CH
79-09-1-2
49
-------
FIGURE 20. NO ENERGY LEVEL DIAGRAM
8
CO
CO
a:
LLJ
oa
LLJ
I
O
DC
LLJ
2
LU
o
o
cc
LLJ
LU
OL-
2,3/2
79-09-1 -3
50
-------
CH Investigations
The apparatus used for the CH measurements is shown schematically in Fig.
21. Figure 22 is a photograph of the apparatus used for the NO measurements;
however, the layout of dye laser components in this case did not differ in any
significant way than for CH. The CH apparatus may be regarded as consisting of
a laser-pumped, pulsed tunable dye laser suitable for producing radiation at
4300 A, a burner for CH production, laser focussing and light collection
optics, and signal recording instrumentation.
The pump laser in Fig. 21 is a Quanta-Ray 1.06 u Nd:YAG laser with
a 10 nsec pulse width and a 10 Hz pulse repetition frequency. It has an
oscillator and a single amplifier stage, and the combination yields upwards of
700 mJ of energy at 1.06-microns. As part of this system, frequency doubling
and wave mixing crystals are incorporated so that 200 mJ and 100 mJ of energy
are available at 5320 and 3550 A, respectively. A Brewster prism, P, is
used to turn the 3550 A radiation by 90 degrees, and a half-wave plate,,
PL, rotates the linear, horizontally polarized pump radiation by 90 degrees to
the vertical direction. For this polarization direction, about 10 percent of
the pump radiation is reflected in turn by two quartz windows, W, while the
remaining, transmitted pump light strikes two mirrors, M, at close to 45° angle
of incidence resulting in near longitudinal excitation of a power amplifier dye
cell, DC. Light reflected from the first window, W, is focussed by two cylindri-
cal lenses, L, to form a thin line image close to the near inner surface of the
first dye cell, DC. This excites a flowing stilbene/methanol dye solution and
lasing occurs for the optical cavity defined by a grating and mirror, M. The
four prisms, P, serve as a beam expander to spread light across the grating
grooves. The increased dispersion narrows the spectral output from the oscil-
lator section of the dye laser. The oscillator output is sent through a
second dye cell which serves as a preamplifier excited in a manner identical to
that in the oscillator section. The dye laser has a second or power amplifier
stage pumped in a near longitudinal configuration. In order to match the size
Of the laser beam exiting the preamplifier with the pump beam, a beam expander
is inserted between the amplifier dye cells, DC, which consists of two lenses,
L forming a Galilean-type telescope arrangement. The stilbene dye laser
output energy was 10 mJ for 100 mJ of pump energy dt 3550 A, which represents
an overall 10 percent energy conversion efficiency. The optical gain for the
first amplifier was 4-5x and 3x for the second amplifier. Accordingly, the
energy of t*16 oscillator alone was in the range of 0.6 to 0.8 mJ. The laser
energy peaked near 4270 A for roughly 200 mg of stilbene in li of methanol,
while the spectral bandwidth was about 0.3 A as measured with a 1/2-meter
Jarrell-Ash spectrometer having a reciprocal linear dispersion of 16 A/mm.
The optimum resolution of the spectrometer at 4358 A (Hg resonance line) was
n 2 A, which, in this case, was adequate for a reasonably accurate determina-
tion of the laser spectral width. The 4270 A laser wavelength was particularly
uitable for CH excitation. A temporal display of the stilbene dye laser pulse
-------
FIGURE 21. LASER INDUCED FLUORESCENCE APPARATUS
Ul
I
§
ND YAG
LASER
TRIGGER FROM LASER
BOXCAR
AVERAGER
SPECTROMETER
-------
FIGURE 22. LASER INDUCED FLUORESCENCE APPARATUS
79-09-1-20
5 <
-------
is given in Fig. 23. The photo exposure time was 10 sec so that a hundred or
so pulses overlap,from which the pulse jitter may be estimated. This is a few
nsec for a nominal pulse width of (5-10) nsec.
The burner in Fig. 21 is the same as used in previous CH measurements
carried out at UTRC (Ref. 6). It is an oxy/acetylene welding torch modified
to have a longer active path length for absorption measurements, which
are described in Ref. 6. The flame emanates from a 0.2 mm x 2.4 cm slot,
and an oxygen/acetylene mixture was burned for the CH measurements. Care was
exercised to ensure that the flame operating conditions were the same as those
for the previous optical absorption measurements in order that meaningful
comparisons could be made as regards species concentration. As pointed out in
Ref. 6, use of the slot burner has the disadvantage that sharp spatial
gradients occur in the transverse flame direction which may lead to erroneous
results for species concentration. However, it was not possible to find an
adequate substitute for the slot burner which would yield both (10-100) ppm of
CH and possess spatial homogeneity. In this regard, a Wolfhard-Parker type
burner was constructed which is known to have a broader flame reaction zone
(Ref. 46). It was not possible to achieve satisfactory stable operation of
this burner with an oxygen/acetylene mixture. A methane/oxygen mixture burned
well at atmospheric pressure, but CH flame emission intensities were several
orders-of-magnitude smaller than those from the slot burner and laser induced
fluorescence signals were not observed. That small quantities of CH were
present was confirmed by computer-code evaluation of mole fractions of various
species present in the flame for prescribed thermodynamic flame properties such
as pressure, temperature, etc. In view of the foregoing difficulties, the slot
burner was used for the present CH measurements.
A 14.5 cm focal length lens is used to focus the dye laser beam into the
flame of the burner, B. A trap, T, helps prevent stray, background laser light
from entering the spectrometer. The fluorescence is viewed at a right angle
with respect to the direction of laser propagation. It is collected and
focussed into the entrance slit of a 1/2-meter Jarrell-Ash spectrometer using
four lenses, L, and a dove prism, P. The first two lenses in combination are
such that the flame is imaged without magnification and the light collection
corresponds to f/8 optics. A small opaque disk is placed directly in front of
the first lens in order to limit the depth-of-field of the viewing optics. The
dove prism, P, rotates the image formed by the first two lenses by 90-degrees.
Consequently, a horizontal laser/flame interaction length is imaged vertically
onto the spectrometer slit and limited to 1 cm by adjusting the slit height.
The image rotation is used to increase the fluorescence intensity collected by
the spectrometer while maintaining good spectral resolution and, thereby,
improving the signal/noise ratio. The lenses, L, adjacent to the prism are
there to ensure that collimated light enters the prism, since this is required
for its proper operation, and, secondly, that f/8 optics are maintained to
match that of the spectrometer.
54
-------
FIGURE 23. STILBENE DYE LASER PULSE
X =-4262 A
LASER
TIME 10NSEC/DIV
79-09-1-6
-------
The signal recording electronics shown in Fig. 21 consists of two RCA
Type 8575 photomultiplier tubes and a boxcar averager. These are supplemented
by a fast oscilloscope, not shown in Fig. 21, in order to view laser and
fluorescence pulses directly. One PMT records reference laser pulses, while
the other, attached to the spectrometer, records fluorescence. The boxcar
averager permits signal averaging which results in a considerable S/N enhancement,
Data
In order to evaluate species concentration from fluorescence intensity,
it is necessary to know the optical collection efficiency of the apparatus,
which from Eq. (12) is the parameter, e . This factor may be determined by
observing Rayleigh scattering of light since the Kayleigh signal intensity is
directly proportional to e. A typical Rayleigh signal used to evaluate e for
the present experiments is shown in Fig. 24. Details concerning the evalua-
tion of e from the signal intensity have been given previously in Ref. 6. It
is evident from Eq. (12) above, that the sample volume size, V, must be
known in order to compute species concentration from fluorescence power. V=A1
where A and 1 are sample cross-sectional area and length, respectively. 1 is
the spectrometer slit height. A, which is the area of the focussed laser spot
must be evaluated. The Rayleigh scattering signal is also useful for this
calibration. By measuring Rayleigh intensity as a function of spectrometer
slit width, the focussed laser spot diameter may be measured. Following this
procedure, the spot diameter was measured to be 270-microns; the diameter was
taken as twice that distance from the center of the spot at which the peak
_o
intensity was reduced by a factor of e . A 14.5 cm focal length lens was
used to focus the laser radiation. The spot diameter, U, is the product, fe,
where f is lens focal length and G is laser beam divergence. A beam divergence
of 0 = 2 x 10 radians is calculated which is fairly good.
Figure 25 displays CH spectra of ordinary flame emission and the laser-
excited fluorescence. The pulsed dye laser and boxcar averager make it
possible to scan the entire CH laser-induced fluorescence spectrum. It is
worth stressing that this was not possible for the previous CH measurements
reported in Ref. 6. The very large signal at the excitation wavelength
near 4262 5 A is due to Rayleigh scattering. The LIFS spectrum is signifi-
cantly different from that observed in normal flame emission, indicating a
non-Boltzmann distribution of population among rotational levels. The principal
characteristics of Fig. 25b are readily understood. The three strong emis-
sions near 4262, 4308, and 4359 A are the R-,Q~, and P- branch lines, respec-
o
tively, which originate from the rotational level in the upper, A, electronic
state excited by the laser. The R(8) resonance fluorescence line is heavily
masked by the Rayleigh scattering signal at the laser wavelength. The preceding
three lines are precisely identified in Fig. 26. The laser excites the
transition K"=8 •* K'=9, where K" and K1 are rotational quantum numbers for the
56
-------
FIGURE 24. RAYLEIGH PULSE
RAY
TIME 20 NSEC/DIV
:•
79-09_1_7
-------
FIGURE 25. CH SPECTRA
a) FLAME EMISSION
Ui
C3
f
g
*+
i
b) LASER-EXCITED FLAME EMISSION
I I I I I
4250
4300
WAVELENGTH — ANGSTROMS
P(10)
P-BRANCH
4350
-------
FIGURE 26. PRINCIPAL LASER-EXCITED CH LINES
i 4262 X [2TT(V" = 0), K" = 8-»2A (V* = 0), K' = 9]
'EXC
1
*-E:
I
(C
R
1
Q
8)
t
1
1
1
1
P(10)
I
9) |
I
I
I
I
I
I
I
I
I
IK'
•10
59
79-09-1-g
-------
f\ f\
ground electronic IT, v"=0 and excited A, v'=0 levels, respectively.
Since the selection rules for emission are AK=0, + 1, the lines P(10),
Q(9), and R(8) are observed in Fig. 25b. Other, weaker P-and R-branch lines
clearly visible in Fig. 25b result as a consequence of rotational energy
transfer out of K'»9 to other nearby rotational levels due to molecular colli-
sions with subsequent emissions according to the preceding selection rules.
From the rotationally nonequilibrated spectrum, one could conclude that the
rate for rotational relaxation, R, does not greatly exceed the electronic
quenching rate, Q, following the results of Lucht and Laurendeau (Ref. 42).
However R » Q but R < I/TL is also possible as described earlier. Smith,
Crosley and Davis (Ref. 47) have made a detailed study of rotational non-
equilibrium of laser-excited OH in an atmospheric presure flame. The ratio of
rotational to electronic quenching was measured and found to vary with the
rotational quantum number of the excited level, and the rates were not greatly
different for large rotational quantum numbers. The rotational rate exceeded
the quenching rate in all cases. A comparable, numerical analysis was not
carried out in the present work owing to the small amount of available data and
the relatively low spectral resolution present in Fig. 25b. On the other
hand, since the rotational level spacings in OH and CH are approximately equal,
the principal conclusions of Smith, Crosley, and Davis for OH relaxation may
apply to CH as well.
Data for CH fluorescence as a function of laser energy are displayed in
Fig. 27. Fluorescence was observed very near 4308 A, which corresponds,
principally, to laser-excited Q(9) emission at this wavelength. The departure
from linearity between 0.01 and 0.1 millijoules, is indicative of saturation.
Of particular significance is the very high S/N accompanying these measurements.
It contrasts very favorably with earlier and similar data given in Fig. 14 of
Ref. 6. The results of Fig. 27 are used below to compute species concen-
tration and excited-state quenching.
CH_R£Sult_s
Following the procedure developed by Baronavski and McDonald (Refs. 30, 31)
and the model developed in this report under Saturated Fluorescence
Theory, the concentration and quenching for CH may be evaluated. Figure 28
is derived from Fig. 27; fluorescence intensity is plotted versus the in-
verse of the laser energy for data points corresponding to the highest laser
energies in Fig. 27. The fluorescence power corresponding to the highest
laser power used is equivalent to about 700 photons. As noted in Fig. 28,
the intercept, SF°, of the straight line yields CH concentration, and the
slope yields the excited state quenching rate, each modified by the f-factor
which follows also from Eq. (13). A significant difference between the
present approach and that used in Ref. 6 is the evaluation of the spec-
tral fraction of fluorescence collected from the laser-excited CH fluorescence
rather than from flame emission spectra. The latter procedure is based on the
60
-------
100
FIGURE 27. LASER INDUCED CH FLAME FLUORESCENCE
XEXC = 4262A[2TT(V"= 0). K"= 8
= 0), K'= 9]
O>
m
DC
10
OJ
z
LU
UJ
O
LU
tr
O
ID
VI
(O
O
ID
0.1
0.001
J I
J I
I
J I I
0.01 0.1 1
LASER ENERGY — MILLIJOULES
10
-------
FIGURE 28. CH CONCENTRATION DATA REDUCTION
• CH CONCENTRATION FROM INTERCEPT
t
CO
I 5
CD
-SLOPE YIELDS QUENCHING RATE
co
z
LLJ
I 3
LU
LU
DC
S 2
nl I I I I I I I
1 2
(LASER ENERGY)'1 — (MILLIJOULES)'1
62
79-09-1 -11
-------
assumption of a Boltzmann equilibrium among rotational levels which was found
to be approximately the case for the longer 300 nsec pulses previously used.
With the shorter laser pulses employed here, rotational equilibration does not
occur. Consequently, the fluorescence spectrum needs to be accurately and com-
pletely determined. The data point at 2.5 (mJ)'1 in Fig. 28 corresponds
to an energy for which the approximate expression for Sf in Eq. 13 is no
longer valid. Table 3 gives the results for CH using the Boltzmann factor
for f in Eq. 13; previous fluorescence as well as absorption results are in-
cluded as well for comparison. The new fluorescence results may be high since
some rotational redistribution occurs during the time of the pulse, thus the
fraction of molecules actually pumped, f, exceeds the Boltzmann fraction, fg,
used in Eq. 13. However, the saturated fluorescence result from this investi-
gation for the concentration agrees, within experimental error, with the
previous absorption measurements, (Ref. 6). It should be noted that the
improvements in the experimental apparatus have little effect on the experimental
errors assigned which are fundamental to an absolute intensity measurement.
The apparatus improvements affect the ability to saturate the transition, to
excite a particular line, to minimize laser chemistry and to signal average.
TABLE 3
LASER-INDUCED SATURATED FLUORESCENCE
MEASUREMENTS SUMMARY
CH
Absorption^1'
Sat. Fluor
Sat. Fluor
(1)
(2)
PPM
57 •* 20
23 _+ 10
103 + 50
N(cm~3)
QCSecT1)
1.6 +_ 0.5 x 101"
7.1 _+ 3.1 x 1013 3 x 109
3.3 + 1.6 x 1014 8 x 1011
CN
Absorption^
Sat. Fluor
Sat. Fluor
(1)
(2)
(1) Ref. 6
(2) This work
131 _+ 34
25 _+ 11
380 + 150
3.8 + 1 x 10
14
8 + 3 x 10
13
2 x 109
1.1 + .44 x 1015 2 x 1013
63
-------
FIGURE 29. ENERGY LEVEL DIAGRAM FOR CN
30,000
20,000
o
DC
UJ
z
UJ
10,000
64
-------
CN Investigations
Theo£e^_icaJL Consider at ions _f or _the_C^_ __^
The cyano radical, CN, differs considerably from the methyne radical,
CH, in two important ways. First, CN has a much smaller rotational constant,
1.9 cm" , compared to 14 cm for CH. Apart from dipole moment considerations,
this implies that the CN radical should have a much larger rotational relaxation
rate than the CH radical. A second difference is that CN has three electronic
states, X, A, and B, the ground, first, and second excited states respectively,
any two of which are connected by radiative transitions. CH likewise has 3
important states to consider; however, the A and B states are not connected by
a strong radiative transition. It is also noted that the CN B-+-X transition is
approximately 10 times stronger than A-»X in CH. Accordingly, CN should be
easier to saturate than the CH radical.
A detailed energy level diagram for CN is shown in Fig. 29. The transition
B-»X, the "CN violet bands" chosen for this saturated fluorescence study are
shown. The 0-0, 1-1, 2-2, etc. sequences are easily observed indicating a
narrow Franck-Condon parabola, that is, the position of the potential minimum
in the B state has shifted very little from that of the X state.
Experimental Consideration^ for_ CN
For the CN saturated fluorescence studies, it was decided to evaluate a
different type of dye laser configuration than that used for the CH work.
Rather than the prism beam expander-grating configuration, a grazing-incidence
grating dye laser was employed. Both the double-grating configuration of
Littmann (Ref. 48) and the grating-mirror arrangement (Refs. 49, 50)
were tried. The former arrangement generated a time-averaged linewidth over
several laser pulses no smaller than 0.1 cm as measured with a Fabry-Perot
interferometer. The grating-mirror combination yielded a similar linewidth at
slightly higher power and appeared to be more stable. Hence, the grating-
mirror configuration was employed for these studies.
The experimental layout is shown in the photograph (Fig. 30) and the
schematic diagram (Fig. 31). In brigf, a frequency-doubled neodymium:YAG
(henceforth 2xNd:YAG) laser at 5320 A pumps t^e grazing incidence rhodamine
640 dye laser oscillator operating near 6100 A. The oscillator is
followed by a preamplifier; both of these laser stages are transversely pumped.
A third, power amplifier stage is longitudinally pumped. The orange dye-laser
radiation is mixed with residual (from the frequency doubler) infrared radia-
tion at 1.06/n in a potassium dideuterium phospate (KD*P) nonlinear optical
crystal to generate the sum frequency in the ultraviolet in the range 3860
A to 3880 A. Emerging from the KD*P crystal are all three frequen-
cies
• the ultraviolet is separated with a Brewster angle 90° deviation
65
-------
FIGURE 30. CN SATURATED FLUORESCENCE EXPERIMENTAL ARRANGEMENT
66
79-08-
63-9
-------
FIGURE 31. EXPERIMENTAL ARRANGEMENT FOR CN SATURATED FLUORESCENCE
OSC AMP1 AMP2
SPECTROMETER AND
PHOTOMULTIPLIER
SLOT BURNER
TRAP
o
03
-------
o
dispersing prism. The 6100 A and 1.06w beams are then trapped as shown in
Fig. 31. The ultraviolet beam is focussed with a simple lens across the same
slot burner as used for CH. In order to generate an easily observable
amount of CN, the previously successful combination of acelytene and nitrous
oxide was used.
The laser-induced fluorescence (and Rayleigh scattering) was collected
with the two-lens system shown. The collection f no. was f/8 to match the
spectrometer numerical aperture. Glass lenses, quite adequate for this range
of the UV were employed throughout. The spectrometer was placed on its side so
that the slit was horizontal in order to maximize the amount of fluorescence
captured from the horizontal, focussed laser beam, while at the same time
maintaining good spectral resolution.
Expe r ime nt a 1 Sj:ud_i ess
Prior to performing laser-induced fluorescence measurements of the CN
radicaj in the acetylene/nitrous oxide slot burner, a moderate resolution
(0.16 A) emission spectrum of CN was recorded. This is shown in Fig. 32.
The 0-0, 1-1, and 2-2 band heads are indicated as is the origin of the 0-0 band.
The origin is the place where the P-and R-branch meet (originate). It is the
place where the Q-branch would be if it were present. Also indicated is J-20
of the R-branch of the 0-0 band which was chosen for the saturated fluorescence
studies. The B * X transition in CN (the violet bands) is a J>^Z transition
with no Q-branch. Laser-induced fluorescence spectra shown in Fig. 33 through
Fig. 36 demonstrate the effect of excitation into different rotational levels.
It is evident that, in contrast to CH, that many rotational levels are popula-
ted because lines are observed up to J-30 and beyond. Presumably, the CN radical
represents a case where the rotational relaxation rate is more pronounced, in
contrast to the case of CH. This difference was mentioned earlier. These
laser-induced fluorescence spectra were recorded at moderately high resolution
using 10 u slits equivalent to a spectral bandwidth of 0.16 A, as was the
flame emission spectrum, Fig. 32. Note that even though the narrowband
laser excitation supposedly excites only a single rotational state of the ground
vibrational level of the ground electronic state, the 1-1 bandhead appears in
the fluorescence spectrum (and perhaps the 2-2 bandhead as well). This indi-
cates either vibrational level communication or excitation overlap (in spite of
the narrow laser linewidth).
The saturation behavior of the laser-induced fluorescence of CN was determin H
by reducing the laser input power by insertion of neutral density filters into
the laser beam. As the laser input energy was so reduced, neutral density
filters were removed from the collection optical train in order to keep the
signal strength at the boxcar signal averager at a reasonable level. For these
68
-------
FIGURE 32. EMISSION SPECTRUM OF CN VIOLET SYSTEM (B 2S-
ACETYLENE-NITROUS OXIDE SLOT BURNER FLAME
X2S)
R-BRANCH OF 0-0 BAND
o>
u
3850
3860
3870
WAVELENGTH, A
3880
3890
-------
FIGURE 33. LASER-INDUCED FLUORESCENCE OF CN RADICAL
EXCITATION INTO RQ(22)
2Qft SLITS (= 0.32A) BY 0.2cm LENGTH
O.SmJ LASER ENERGY
LOWER TRACE IS RAYLEIGH SCATTERING
(FLAME OFF)
0-0
BANDHEAD
J = 22
ATTENUATION OD = 1.0
3840
3850
3860
3870
3880
f
WAVELENGTH, A
-------
FIGURE 34. LASER INDUCED FLUORESCENCE OF CN RADICAL
EXCITATION INTO Ro(14)
10/1 SLITS (= 0.16A) BY 0.2cm LENGTH
0.2mJ LASER ENERGY
J = 14
0-0
BANDHEAD
I
0>
u
3850
3860
3870
WAVELENGTH, A
3880
3890
-------
FIGURE 35. LASER-INDUCED FLUORESCENCE OF CN RADICAL
EXCITATION INTO Ro(3)
T
o
-------
FIGURE 36. LASER-INDUCED FLUORESCENCE OF CN RADICAL
EXCITATION INTO Ro(20)
oo
BANDHEAD
SLIT WIDTH: 10(1, BY 02cm LENGTH
0.85mJ LASER ENERGY
OJ
AM/IMAAAAAAA
J = 20
ID
i
00
u
CD
3850
3860
3870
3880
WAVELENGTH, A
-------
saturation measurements the spectromgter slits were opened to 300 u and the
spectrometer wavelength set to 3880 A, the maximum of the 0-0 P-branch
For this case, the spectral slit width of the spectrometer is nearly 5A.
Because the P-branch forms a head, a large fraction of the total laser-induced
fluorescence, s 402, is captured by means of this arrangement. Excitation
occurred at R(20) which is the value of J for the maximum in the R-branch (very
broad) at 2800°C.
The laser-induced fluorescence signal as a function of laser energy is
shown in Fig. 37. At the lowest laser energies the curve is linear with unity
slope but begins to roll over at about 10 millijoules. Thereafter a second
linear region of slope 0.33 is attained. The second limiting linear region is
in general agreement with the treatment of Daily (Ref. 44). For a Gaussian
laser intensity distribution, he obtained a limiting slope of approximately
0.26. Another interpretation of this second linear region is due to Van Calcar
and colleagues (Ref. 51) who note that for fluorescence signals integrated over
the entire duration of the laser pulse, a limiting slope of 0.34 is reached.
These results were obtained from experimental measurements on sodium. In
summary then, the data indicate that the CN radical is easily partially saturated-
however, full saturation is difficult to achieve.
The laser-induced fluorescence data can be treated according to the
Baronavski-McDonald (Ref. 30) technique to extract the concentration and
quenching rate. However, the fluorescence power intercept is now taken to be
SS = fNT consistent with the model developed in this report. A plot of
the absolute laser-induced fluorescence signal strength (in Watts) as a function
of the reciprocal laser spectral irradiance, Jem is shown in Fig. 38.
This plot is linear at high laser energies but rapidly departs from linearity
as the laser energy falls off. This behavior is consistent with the "partially-
saturated" model of Ref. 30. The value for Sp is 7 x 10~7 W. This signal
corresponds to approximately 14,000 photons.
Discussion
The number density which is obtained must be interpreted with reference
to the modified two-level model developed earlier in this report. The number
density, according to this model, can be found from the relation NT
-------
FIGURE 37. LASER INDUCED CN FLUORESCENCE: SATURATION BEHAVIOR
XEXC = 3857A[2Z(V" = 0)J = 21-~2£(V' = 0) J = 20]
SLIT WIDTH 300/1 BY 0.2 CM LENGTH, SPECTROMETER AT 3883 °A
§
O>
u
I
1000
CO
z
ID
DC
-------
FIGURE 38. SATURATED FLUORESCENCE POWER FOR CN VERSUS
INVERSE LASER SPECTRAL IRRADIANCE
h-
O
X
CO
CO
100
1/lvL — cm2/J
150
g
M
-------
The Boltzmann factor for J = 20 of CN at 2800° C is 0.03. This gives a
1 5 —3
value of NT from the saturated fluorescence data of 1.1 x 10 cm . The
i y _ o
previously measured absorption value (Ref. 6) was 3.8 ± 0.8 x 10 cm
for the same conditions and same position in the flame. These results are
shown in Table 3. The fluorescence measurement is high by a factor of
nearly three compared to the absorption measurement. This difference can be
accounted for qualitatively by the two-level model if f is interpreted to be
greater than ^Boltzmann' ^ quantitative value for f cannot be assigned
because of lack of data about rotational relaxation rates. The direct compari-
son suggests a factor of 3 which must multiply the Boltzmann factor of 0.03.
However, the accuracy of the absorption measurement is open to some question
because of spatial averaging over the flame by the absorption beam. The
distribution of CN in the flame in the horizontal direction, perpendicular to
the slot direction, was mapped out by the fluorescence signal. The flame
holder was translated relative to the laser beam and the relative fluorescence
signal recorded as a function of travel. The double-humped distribution, shown
in Fig. 39, was found. This distribution could make the absorption measure-
ment low relative to the fluorescence determination, because the latter was
performed on the maximum in the distribution toward the collecting lens. Hence
the absorption results are considered a lower limit.
The quenching rate constant was found from the slope of Fig. 38, assuming
that Q2i » A21' The va^ue found is 1.9 x 10 sec , uncorrected for the
f-f actor discussed above. The effect of this factor on Q is small, a factor of two
at most .
The experimental error in the determination of the number density Ni could
be found by means of the standard propagation of error technique; however, it
is simpler to estimate the error in the intercept, Sp° of the graph in Fig. 38.
The extreme values are approximately 5.5 and 8.5 x 10~ W. Hence the error about
the chosen value 7 x 10 W is ± 20%. This may be a conservative estimate.
Further error occurs in converting Sp° to number density because of uncertainties
in the laser beam area, A, and the transmission factor, e . The error in A should
not be much larger than + 10% as two different methods gave good agreement.
The error in E is estimated to be ±30%, because several factors are involved.
The most probable experimental error in NT is then about + 40%. Note
that this is the experimental error and does not include uncertainties in A2i
nor the f-factor of the modified two-level model introduced earlier.
NO Invest igat ions
The apparatus used to measure laser-excited NO spectra is shown schematically
in Fig. 40. It consists of a 2xNd:YAG laser-pumped tunable dye laser with
77
-------
FIGURE 39. HORIZONTAL SCAN OF LASER-INDUCED FLUORESCENCE
IN CN SLOT BURNER
(C2H2 — N20)
0.060
FLAME HOLDER TRAVEL, CM
79-08-63-6
78
-------
FIGURE 40. LASER INDUCED UV FLUORESCENCE APPARATUS
L_h
BOXCAR AVERAGER
TRIGGER FROM LASER
7
8
-------
frequency doubling followed by subsequent mixing of the output with 1.06U
radiation to generate UV wavelengths. The UV output of the laser interacts
with NO in a cell or burner. Light collection optics similar to that used
for CH completes the system.
The laser consists of a laser-pumped tunable dye laser operating near 5740 A
which is frequency-doubled to 2870 A and mixed with 1.06^ radiation to yield
frequgncy up-converted radiation near 2260 A suitable for NO excitation. The
5740 A dye laser in Fig. 40 is very similar to the stilbene dye laser in
Fig. 21. The principal difference is that 2xNd:YAG pump radiation at 5320 A
is used to excite a rhodamine 590 dye as opposed to 3550 A 3xNd:YAG excita-
tion of stilbene. The laser output at 5740 A is frequency-doubled in a KD*P
Type I crystal to produce 2870 A radiation. In order to better match the
size of the 2870 A beam to that at 1.06p, it is beam-expanded by a factor of
1.7. A dichroic, oriented at 45-deg, reflects the 2870 A beam in the direc-
tion of the frequency up-converting crystal, FM, also Type I KD*P, where it com-
bines with the 1.06p radiation passed unattenuated by the dichroic filter. Th
emerges from the crystal, FM, radiation at 2870, 10,600 and 2260 A. In order
to select 2260 A, a dispersing quartz prism, P, and ^ight trap, T, are used.
The energy of the dye laser at 5740, 2870, and 2260 A was (30-40) mJ, (2-3) mj
and (.2-.3) mJ, respectively, for approximately 175 mJ of pump radiation at
5320 A. The laser pulse widths were 0(5-10) nsec and the laser bandwidth at
2260 A was 0.1 A. The estimated beam divergence of the UV dye laser was a
factor of two larger than for CH or, in this case, equal to 4 mrad.
The UV beam is focussed into a burner, B, (or cell) with a lens having
typically, a 10 cm focal length. A meter, PM, measures the energy of the
transmitted beam. The light collection optics are very similar to that for CH
above. In this case, however, an image-rotating dove prism is not used in
front of the spectrometer entrance slit. Instead, as the case for CN, the
spectrometer is rotated 90-deg, that is, placed on its side so as to make the
slit horizontal, duplicating £he function of the dove prism. In order to
enhance observation of 2260 A emission, a UV-sensitive Hamamatsu R456
photomultiplier tube is used in conjunction with a Jobin-Yvon 0.6 meter Type
HRP-1 spgctrometer equipped with a 2400 groove/mm hologrgphic grating, capable
of 0.08 A resolution and with peak efficiency near 3000 A.
The burner and cell used for NO measurements are both simple in design.
The cell consists of a standard pyrex cross with quartz entrance and exit
windows; it can be evacuated to about 10 Torr pressure and filled with a
known quantity of NO as measured accurately with a calibrated capacitance
manometer. The optical absorbing path of the 2260 A focussed laser
radiation was 10 cm. The burner design is such that premixed fuel/air emerges
from a 1-inch diameter, stacked honeycomb arrangement of stainless steel hypo
tubing. This gave a flat, stable CH^/air flame to which NO was added by in-
jection into the premix chamber. The flame was operated at near stoichiometri
80
-------
mditions, and the fraction of NO in the CH^/air/NO mixture was 3.6%. Gas
flows were measured with calibrated flow meters.
Nitric oxide fluorescence observations were carried out successfully only
in the case of the burner referred to above; difficulties were encountered with
the cell measurements. Over a wide range of NO pressure, clean laser-excited
fluorescence signals were not observed from the cell, because of excessive
absorption of the focussed UV beam in the long path length (10 cm) cell. This
was confirmed by a calculation of the absorption expected, which showed that
the cell would be opaque to the UV radiation even for the lowest NO pressures
attainable within limitations imposed by gas impurities and cell evacuation.
Since any observation of NO saturation in the cell would not definitively
determine the laser spectral intensities required to saturate NO in a flame,
owing to radically different quenching conditions, the cell measurements were
terminated in favor of laser-excited NO flame fluorescence measurements.
Studies in specially-designed cells of NO fluorescence in various controlled
gas mixtures could be quite valuable and should be seriously considered for
future studies.
A laser-excited NO fluorescence spectrum observed in a premixed CH^/air
flame doped with NO is shown in Fig. 41. The wavelength region (2240-2280) A
is shown, gcanned with and without NO doping of the flame. The laser wavelength was
near 2258 A, clgse to the (Rn + Q21) = 13 1/2 (=J") line (Ref. 52). The
signal at 2258 A in both a) and b) is due to Rayleigh scattering with scattering
from the burner face superimposed. The latter occurred since the strongest NO
signals were observed very near the burner face. By removing the burner, the
Rayleigh signal could be observed alone, and its intensity was about the same
as the strongest NO signals. The amplitude of the NO fluorescence pulses was
very erratic, much more so than for CH. This was due probably to laser amplitude
variations induced by frequency instabilities at 2258 A. This jitter was
evident in oscilloscope traces of the fluorescence pulses at 2265 A and,
indirectly, from Fabry-Perot analysis of the laser mode structure at 5740
A which demonstrated considerable instability. NO fluorescence pulses
are shown in Fig. 42. About 100 pulses are superimposed since the photographic
exposure time was 10 sec and the laser repetition rate was 10 Hz. Considerable
amplitude variation is evident; in isolated cases, the location of the pulse in
time shifted a full 10 nsec pulse length. The pulse duration may not be taken
as true since the rise time of the photomult iplier is 3 nsec. The spectrum in
pig. 41 was taken with a 40 p. spectrometer slit width for which the measured
spectrometer resolution is of order 0.25 A. This precluded very high
resolution of rotational spectra, but the 40 /u slit width was required in order
to observe emission with good S/N. In spite of the foregoing, it was possible
to observe distinctly the band heads identified in Fig. 41, and thereby confirm
observation of NO emission.
81
-------
FIGURE 41. NO SPECTRA
a
s
I
_*
/
01
a) CH4/AIR FLAME EMISSION
b) CH4/AIR/NO LASER-EXCITED FLAME EMISSION
VV
12
P22 + Q12
2240
2250 2260 2270
WAVELENGTH — ANGSTROMS
2280
-------
FIGURE 42. NO FLUORESCENCE PULSES
X =-2265 A
DBS
TIME 20 NSEC/DIV
I
79-09-1-13
-------
o
NO emission at 2265 A was monitored as a function of laser energy
to look for NO saturation. Saturation of the fluorescence was not observed
as may be seen from Fig. 43. As in previous experiments, fluorescence
was observed at a wavelength shifted from that of the laser in order to avoid
Rayleigh scattering. The dependence of fluorescence intensity on laser energy
was strictly linear, and this was confirmed by repetition of the measurement.
The highest laser energy in Fig. 43 corresponds to a 6.4 x 10 W/ctn
cm laser spectral intensity incident on the CH^/air/NO flame. In Table
XXIII of Ref. 1, it was estimated that the laser spectral intensity required
to saturate NO in a flame is 4x10 W/cm cm . This estimate is consistent
with present observations.
Since saturation of the NO fluorescence was not observed, it is not possible
to determine the flame NO concentration employing the saturated fluorescence
data reduction approach described earlier. On the other hand, the data given
in Fig. 43 does permit evaluation of the fluorescence efficiency or Stern-
Voimer factor, A2, /(Q2^+A2^) , provided that the NO concentration is
known and, secondly, Rayleigh data are used to evaluate the apparatus factor, e.
Since a calibrated flowmeter is used for doping the CH^/air flame, the NO
concentration corresponding to the Fig. 43 data can be estimated. In
experiments at UTRC, Seery has found that 25-50% of NO doped into low pressure,
premixed flames can be lost at equivalence ratios of 0.9 and above. Only if
the flame is very lean is the NO likely to be conserved. Since the NO fluores-
cence was strongest at the burner surface, it is clear that some NO loss
occurred in the flame used for the fluorescence experiments. If it is assumed
that no NO loss occurred, then the estimates of required saturation spectral
intensity will err on the conservative side.. The Stern-Volmer factor measures
the degree to which fluorescence intensity is diminished by collisional quenching
of the excited electronic state. Evaluation of this factor is important because
the ratio, IH/B^*^^/^!* A2p» a measure of saturation, can then be com-
puted. For the NO study, this factor is apparently small since saturation was
not observed. Evaluation of the preceding ratio makes it possible to predict the
increase in I,v required for saturation to occur. The Stern-Volmer factor was
calculated to be A21/(Q2j+A2j) = 4.4x10 , and the saturation parameter
ILV(B12+B2i)/(Q2i+A21) = 2.5x10 . By contrast, the saturation parameter for
CH corresponding to the highest laser energy was 2.5. The foregoing indicates
that IT v for the NO measurements must be increased by about a factor of 100
for application of the saturated fluorescence technique to species measurement
consistent with the earlier estimates (Ref. 1). Similarly, it is instructive
to note that the highest value of I.v achieved at 2258 A would correspond in
the case of CH (see Fig. 27) to measurements at laser energies of less than
0.04 mJ, in the linear region. Such a comparison is reasonable in that Q,
for NO and CH are both, approximately, 10 sec .
84
-------
10
9
8
7
FIGURE 43. LASER EXCITED NO FLAME FLUORESCENCE
\EXC~2258A[2TT(V" = 0)-» 2£(V=0]
" 2265 *
C/5
CD 4
CC
>
CO
LU
o
Ul
LU
CC
O
lvl_SAT~108 W/cm2-cm'1
\v |_ = 6 X 106, W/cm2 — cm'1
0.06
0.08 0.1
LASER ENERGY — MILLIJOULES
0.2
79-09-1-14
85
-------
As indicated above, an increase in ILv by a factor of 100 is required
to saturate NO. It appears that such an increase is feasible if the present
breadboard laser is replaced by a commercial dye laser system. The laser
spectral intensity is given by ILv • (E/i) (irD2/4)~1(Av)"1, where E and
T are the pulse energy and length, respectively; D is the focussed beam diameter-
and Av is the laser bandwidth. The time, T, is not expected to differ for the
two systems since the pump lasers are identical. Of the remaining factors,
numerical values for the commercial dye laser system at 2260 A. for 200 mJ of
5320 A pump energy, are: E » 2mJ; D = 100 microns; Av = 1 cm~* (Ref. 53).
This represents a lOx increase in E, a 4x reduction in D, and a 2x reduction in
Av when compared with our system, which implies a 320x increase in I, Conse-
quently, saturated NO measurements appear feasible with an improved laser pumped
dye laser system.
General Discussion
The most significant result of the saturated fluorescence research is the
relatively good agreement of the saturated fluorescence measurements with the
absorption measurements as tabulated in Table 3. The agreement is roughly a
factor of two for the CH radical and a factor of three for CN. These measurements
were made on extremely reactive radical species in an atmospheric pressure
flame at temperatures of 0(2500 °C> by means of a still-developing technique;
for these reasons, the agreement should be considered quite satisfactory. In the
only other molecular case known to the authors where a comparison between
absorption measurements and saturated fluorescence was made (Ref. 33), the
agreement was good only to a factor of ten. There is, of course, uncertainty
in the f-factor which accounts for the extent of rotational redistribution. AS
mentioned above, this factor cannot be determined until rotational relaxation
effects are modeled and compared with experiment. The importance of solving
this particular problem is discussed below.
The three radical species chosen for this study form an interesting series with
respect to ease of saturating the fluorescence transitions. Both CH and CN
could be saturated with attainable laser energies, CN being easier to saturate
than CH because of the larger spontaneous emission coefficient, A2i, (hence,
larger B^ coefficient) for CN. The NO radical represents a particularly
difficult species to saturate for two reasons: the A to X gamma bands are not
strong transitions, and they lie nearly in the vacuum UV region of the spectrum.
Although presently available commercial dye laser systems attain reasonable
power levels in this region by wave-mixing processes, these techniques were
untried when this program was initiated. Saturation of the NO radical fluores-
cence was not achieved in this investigation mainly because of low laser energy
at 2260 A, ^ 0.1 millijoule. However, laser-induced fluorescence spectra
were obtained from NO doped flames, and it appears that saturation of NO fluo-
rescence could be achieved with commercial dye laser systems.
86
-------
From a purely experimental viewpoint, the choice of the short-pulse Nd:Y.AG
laser pumped dye laser systems was a good one. Dye lasers operating in the orange,
yellow, and blue were constructed which performed well with good beam quality
and narrow spectral width, both essential properties for obtaining high laser
spectral irradiance. Doubling and mixing of these dye-laser beams to generate
UV radiation was achieved. The less than 10 nanosecond pulse duration of these
lasers, in addition to generating high peak powers, ensures that very little
laser-induced chemistry (characteristic time scale of microseconds) will take
place. If chemistry occurs during the fluorescence observation period, molecules
are lost from the system leading to erroneous measurements. During a 10 nano-
second pulse, this effect is negligible.
Through the use of the boxcar averager, high-resolution scans of the
entire fluorescence spectrum were made for all species. It should be stressed
that these spectra represent 10 nanosecond "snapshots" of the fluorescence.
Contained in these spectra is a wealth of information regarding the relation-
ship of rotational relaxation rates to quenching rates, spontaneous emission
rates, and the length of the laser pulse. It is precisely this relationship
which must be understood in order to properly interpret the saturated fluores-
cence data in terms of a model which incorporates rotational energy transfer.
The modified two-level model introduced in this report exhibits the proper
limiting behavior for both fast and slow extremes of rotational energy transfer.
For intermediate rotational rates the f-factor cannot be determined without
detailed knowledge of all of the rate processes involved. Computer modeling of
the time-dependent solutions of the rate equations, along with experimental
measurements, will be required to obtain this knowledge.
The detectivity limits for the CH and CN radicals can be found by calculat-
ing the number of photons corresponding to the fluorescence power induced by
the maximum laser spectral irradiance employed. Based upon a shot-noise limit
analysis and choosing a signal-to-noise ratio of 5, these limits are:
for CH, 14 ppm
CN, 2 ppm
It must be stressed that these limits apply to the conditions used in this
investigation, that is, for the specified values of laser power, fluores-
cence fraction collected, rotational level excited, etc. These limits could
he improved upon with increased laser power and more efficient fluorescence
collect*00 and other modifications. For example, changing the collection
optics from f/8 to f/3 would improve the signal by a factor of seven. The
spectrometer could be replaced with a bandpass filter, once the proper spectral
region for collection was determined. This could improve signal strength by a
factor of 2-3 or more.
87
-------
REFERENCES
1. Eckbreth, A. C., P. A. Bonczyk, and J. F. Verdieck: Review of Laser Raman
and Fluorescence Techniques for Practical Combustion Diagnostics, Report
EPA-600/7-77-066, June 1977.
2. Eckbreth, A. C., P. A. Bonczyk, and J. F. Verdieck: Laser Raman and
Fluorescence Techniques for Practical Combustion Diagnostics, Appl. Spect.
Rev., Vol. 13, pp. 15-164, 1978.
3. Under corporate sponsorship, an updated and revised version of Reference
2 was prepared and appeared in Progress in Energy and Combustion
Science under the title "Combustion Diagnostics by Laser Raman and Fluo-
rescence Techniques", Vol. 5, pp. 253-322, 1979.
4. Lapp, M., and C. M. Penney: Laser Raman Gas Diagnostics, Plenum Press,
New York, 1974.
5. Lederman, S.: The Use of Laser Raman Diagnostics in Flow Fields and
Combustion, Prog. Energy Comb. Sci., Vol. 3, pp. 1-34, 1977.
6. Eckbreth, A. C., P. A. Bonczyk, and J. A. Shirley: Investigation of
Saturated Laser Fluorescence and CARS Spectroscopic Techniques for
Combustion Diagnostics, Report EPA-600/7-78-104, June 1978.
7. Bonczyk, P. A., and J. A. Shirley: Measurement of CH and CN Concentration
in Flames by Laser-Induced Saturated Fluorescence, Combust. Flame, Vol. 34,
pp. 253-264, 1979.
8. Nibler, J. W., W. M. Shaub, J. R. McDonald and A. B. Harvey: Coherent
Anti-Stokes Raman Spectroscopy, in Vol. 6 Vibrational Spectra and Structure,
J. R. Durig Ed., Elsevier, Amsterdam, 1977.
9. Druet, S., and J. P. Taran: Coherent Anti-Stokes Raman Spectroscopy, in
Chemical and Biological Applications of Lasers, C. B. Moore, Ed., Academic
Press, New York, 1979.
10. Nibler, J. W., and G. V. Knighten: Coherent Anti-Stokes Raman Spectroscopy,
in Raman Spectroscopy of Gases and Liquids, A. Weber, Ed., Springer-Verlag,
Berlin, 1979.
11. Zinn, B. T., Ed.: Experimental Diagnostics in Gas Phase Combustion Systems.
AIAA, New York, 1977.
88
-------
REFERENCES (Cont'd)
12. Eckbreth, A. C.: Effects of Laser Modulated Particulate Incandescence
on Raman Scattering Diagnostics, J. Appl. Phys., Vol. 48, pp. 4473-4479,
1977.
13. Leonard, D. A.: Field Tests of a Laser Raman Measurement System for
Aircraft Engine Exhaust Emissions, AFAPL-TR-74-100, 1974.
14. Hall, R. J.: CARS Spectra of Combustion Gases. Comb. Flame, Vol. 35,
pp. 47-60, 1979.
15. Roh, W, B., and P. W. Schreiber: Pressure Dependence of Integrated CARS
Power, Appl. Opt., Vol. 17, pp. 1418-1424, 1978.
16. Eckbreth, A. C. and R. J. Hall: CARS Thermometry in a Sooting Flame,
Comb. Flame, Vol. 36, pp. 87-98, 1979.
17. Eckbreth, A. C,: CARS Investigations in Sooting and Turbulent Flames,
Project SQUID Report UTRC-5-PU, 1979.
18. Hong, N. S., A. R. Jones and F. J. Weinberg: Doppler Velocimetry
Within Turbulent Phase Boundaries, Proc. Roy. Soc. Lond. A, Vol 53,
pp. 77-85, 1977.
19. Stenhouse, I. A., D. R. Williams, J. B. Cole and M. D. Swords: CARS
in an Internal Combustion Engine, Appl. Opt., Vol. 18, pp. 3819-3825, 1979.
20. Switzer, G. L., W. M. Roquemore, R. P. Bradley, P. W. Schreiber and
W. B. Roh: CARS Measurements in a Bluff-Body Stabilized Diffusion Flame,
Appl. Opt., Vol. 18, pp. 2343-2345, 1979.
21. Attal, B., M. Pealat and J. P. Taran: CARS Diagnostics of Combustion,
AIAA Paper No. 80-0282, 1980.
22. Switzer, G. L., L. P. Goss, W. M. Roquemore, R. P. Bradley, P. W. Schreiber
and W. B. Roh: The Application of CARS to Simulated Practical Combustion
Systems, AIAA Paper No. 80-0353, 1980.
23. Eckbreth, A. C.: BOXCARS: Crossed-beam Phase-matched CARS Generation
in Gases, Appl. Phys. Lett., Vol. 32, pp. 421-423, 1978.
24. Eckbreth, A. C.: Remote Detection of CARS Employing Fiber Optic Guides,
Appl. Opt., Vol. 18, pp. 3215-3216, 1979.
25. Beer, J. M.,and N. A. Chigier: Combust ion Aerodynamics, John Wiley, New
York, 1972.
89
-------
REFERENCES (Cont'd)
26. Eckbreth, A. C.,: R. J. Hall and J. A. Shirley: Investigations of Coherent
Anti-Stokes Raman Spectroscopy (CARS) for Combustion Diagnostics, AIAA
Paper 79-0083, New Orleans, La., 1979.
27. Eckbreth, A. C.,: and R. J. Hall: CARS Diagnostic Investigations of Flames,
Proceedings of the NBS 10th Materials Research Symposium on Characterization
of High Temperature Vapors and Gases, Gaithersburg, Md., 1978.
28. Hall, R. J.: Pressure-broadened Linewidths for N2 CARS Thermoraetry.
Submitted to Applied Spectroscopy, January 1980.
29. Colket, M. B., et. al.: NO Measurement Study; Probe Measurements, Task
II Report under Contract DOT-FA77WA-4081, United Technologies Research
Center Report R79-994150-2, 1979.
30. Baronavski, A. P. and J. R. McDonald: Measurement of $2 Concentrations
in an Oxygen-Acetylene Flame: An Application of Saturation Spectroscopy.
J. Chem. Phys., Vol. 66, pp. 3300-3301, 1977.
31 Baronavski, A. P. and J. R. McDonald: Application of Saturation Spectro-
scopy to the Measurement of Co, *u Concentrations in Oxygen-Acetylene
Flames. Appl. Opt., Vol. 16, pp. 1897-1901, 1977.
32. Lucht, R. P., Private Communication.
33. Pasternack, L., A. P. Baronavski, and J. R. McDonald: Application of
Saturation Spectroscopy for Measurement of Atomic Na and MgO in Acety-
lene Flames. J. Chem. Phys., Vol. 69, pp. 4830-4387, 1978.
34. Becker, K. H. and D. Haaks: Lifetime Measurements on Excited SH (A2 &)
Radicals. Journal of Photochemistry, Vol. 1, pp. 177-179, 1972/73.
35. Muller, C. H. Ill, K. Schofield, M. Steinberg, and H. P. Broida: Sulfur
Chemistry in Flames (Preprint).
36. Daily, J. W. and C. Chan: Laser-Induced Fluorescence Measurement of
Sodium in Flames. Comb, and Flame, Vol. 33, pp. 47-53, 1978.
37. Smith, B., J. D. Winefordner, and N. Omenetto: Atomic Fluorescence of
Sodium under Continuous-Wave Laser Excitation. J. Appl. Phys., Vol.
48, pp. 2676-2680, 1977.
38. Kuhl, J. and H. Spitschan: Flame-Fluorescence Detection of Mg, Ni, and
Pb with a Frequency-Doubled Dye Laser as Excitation Source. Opt. Comm.,
Vol. 7, pp. 256-259, 1973.
90
-------
REFERENCES (Cont'd)
39. Omenetto, N. , P. Benetti, L. P. Hart, J. D. Winefordner, and
C. Th. J. Alketnade: Non-Linear Optical Behavior in Atomic Fluorescence
Flame Spectrometry. Spectrochimica Acta, Vol. 28B, pp. 289-300, 1973.
40. Olivares, D. R. and G. M. Hieftje: Saturation of Energy Levels in
Analytical Atomic Fluorescence Spectrometry-I. Theory. Spectrochimica
Acta, Vol. 33B, pp. 79-99, 1977.
41. Daily, J. W.: Saturation Effects in Laser Induced Fluorescence Spectros-
copy. Appl. Opt., Vol. 16, pp. 568-571, 1977.
42. Lucht, R. P. and N. M. Laurendeau: Two-Level Model for Near Saturated
Fluorescence in Diatomic Molecules. Appl. Opt., Vol. 18, pp. 856-861,
1979.
43. Berg, J. 0. and W. L. Shackleford: Rotational Redistribution Effect on
Saturated Laser-Induced Fluorescence. Appl. Opt., Vol. 18, pp. 2093-2094,
1979.
44. Daily, J. W.: Saturation of Fluorescence in Flames with a Gaussian
Laser Beam. Appl. Opt., Vol. 17, pp. 225-229, 1978.
45. Dewey, H. J.: Second Harmonic Generation in KBrOo * 4H20 from 217.1 to
315.0 nm. IEEE J. Q. E., Vol. QE-12, pp. 303-306, 1976.
46. Wolfhard, H. G. and W. G. Parker: A New Technique for the Spectroscopic
Examination of Flames at Normal Pressures. Proc. Phys. Soc., Vol. 62,
pp. 722-730, (1949).
47. Smith, G. P., D. R. Crosley, and L. W. Davis: Rotational Populations
Distributions in Laser-Excited OH in an Atmospheric Pressure Flame.
Paper presented at Eastern State Division of the Combustion Insti-
tute, Atlanta, GA, November 1979.
48. Littman, M. G.: Single-Mode Operation of Grazing Incidence Pulsed Dye
Laser, Optics Lett., Vol. 3, pp. 138-140, 1978.
49. Littman, M. G., and H. F. Metcalf: Spectrally Narrow Pulsed Dye Laser
Without Beam Expander, Appl. Optics, Vol. 17, pp. 2224-2227, 1978.
50. Shoshan, I., N. N. Danon, and U. P. Oppenheim: Narrowband Operation of
a Pulsed Dye Laser Without Intracavity Beam Expansion. J. Appl. Phys.,
Vol. 48, pp. 4495-4497, 1977.
91
-------
REFERENCES (Cont'd)
51. Van Calcar, R. A., M. J. M. Van de Ven, B. K. Van Vitert, K. J. Biewenga,
T. Hollander and C. Th. J. Alkemade: Saturation of Sodium Fluorescnce
in a Flame with a Pulsed Tunable Dye Laser, J. Quant. Spectres. Rad.
Trans., Vol. 21, pp. 11-18, 1979.
52. Zacharias, H., A. Anders, J. B. Halpern, and K. H. Welge: Frequency
Doubling and Tuning with KBcOg^H^O and Application to NO A 2£+ Exci-
tation. Opt. Coram., Vol. 19, pp. 116-119, 1976.
53. Private communication with S. Brosnar of Quanta-Ray, Inc.
92
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO. 2.
EPA-600/7- 80-091
4. TITLE AND SUBTITLE
Investigation of CARS and Laser-induced Saturated
Fluorescence for Practical Combustion Diagnosis
7. AUTHOR(S)
A.C.Eckbreth, P.A.Bonczyk, and J.F. Verdieck
9. PERFORMING ORGANIZATION NAME AND ADDRESS
United Technologies Research Center
East Hartford, Connecticut 06108
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
3. RECIPIENT'S ACCESSION NO.
6. REPORT DATE
Mav 1980
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
INE623
11. CONTRACT/GRANT NO.
68-02-3105
13. TYPE OF REPORT AND PERIOD COVERED
Final; 8/78-8/79
14. SPONSORING AGENCY CODE
EPA/600/13
,5. SUPPLEMENTARY NOTES IERL_RTP project officer is William B. Kuykendal, Mail Drop
62, 919/541-2557. EPA-600/7-77-066 and EPA-600/7-78-104 are related reports.
6. ABSTRACT
The report gives results of experimental investigations aimed at develop-
ing nonperturbing, spatially precise, in-situ diagnostic techniques to measure spe-
cies composition and temperature in flames. The investigations continued earlier
development of coherent anti-Stokes Raman spectroscopy (CARS) and laser-induced
saturated fluorescence. The program included two main, concurrent tasks. In Task
1, optical thermometry, the practical feasibility of CARS was demonstrated in a
program of research-scale combustor testing (results agreed to within 5% for com-
bustion zone temperature measurements made with CARS and with shielded thermo-
couples). In Task 2, optical composition, laser-induced saturated fluorescence was
examined in regard to its capability for measuring CH, CN, and NO concentrations
in flames. Saturation of the fluorescence in CH and CN was achieved and consider-
able insight into the physics of saturated fluorescence was obtained. Promising init-
ial results of NO fluorescence in flames are described, but saturation was not ob-
served for laser spectral intensities up to 6 million W/sq cm/cm.
17. KEY WORDS AND DOCUMENT ANALYSIS
a DESCRIPTORS
Pollution Spectroscopy
Combustion Fluorescence
Flames Lasers
Measurement
Optical Tests
Temperature
13. DISTRIBUTION STATEMENT
Release to Public
b.lDENTIFIERS/OPEN ENDED TERMS
Pollution Control
Stationary Sources
Optical Composition
Optical Thermometry
CARS
19. SECURITY CLASS (This Report)
Unclassified
20. SECURITY CLASS (This page)
Unclassified
c. COSATI Field/Group
13B
21B 20F
20E
14B
102
22. PRICE
EPA Form 2220-1 (t-73)
93
------- |