ited States
Environmental Protection
Agency
Industrial Environmental Research
Laboratory
Research Triangle Park NC 27711
EPA-600/7-78-104
June 1978
Investigation of
Saturated Laser
Fluorescence and
CARS Spectroscopic
Techniques for
Combustion
Diagnostics
Interagency
Energy/Environment
R&D Program Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental Protec-
tion Agency, have been grouped into nine series. These nine broad categories were
established to facilitate further development and application of environmental tech-
nology. 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 ENVIRONMENTAL PROTECTION TECHNOLOGY
series. This series describes research performed to develop and demonstrate instrumen-
tation, equipment, and methodology to repair or prevent environmental degradation from
point and non-point sources of pollution. This work provides the new or improved tech-
nology required for the control and treatment of pollution sources to meet environmental
quality standards.
REVIEW NOTICE
This report has been reviewed by the U.S. Environmental
Protection Agency, and approved for publication. Approval
does not signify that the contents necessarily reflect the
views and policy of the Agency, 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-600/7-78-104
June 1978
Investigation of Saturated Laser
Fluorescence and CARS Spectroscopic
Techniques for Combustion Diagnostics
by
A.C. Eckbreth, P.A. Bonczyk, and J.A. Shirley
United Technologies Research Center
East Hartford, Connecticut 06108
Contract No. 68-02-2176
Program Element No. EHE624
EPA Project Officer: William B. Kuykendal
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, NC 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
EPA - RTF LIBRARY
-------
DISCLAIMER
This report has been reviewed by the Industrial Environmental Eesearch
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.
-------
FOREWORD
Under Contract 68-02-2176 sponsored by the Environmental Protection
Agency, the United Technologies Research Center (UTRC) is conducting analytical
and experimental investigation of non-perturbing, spatially precise, in-situ
diagnostic techniques to measure species concentrations and temperature in flames.
Under Task I, a comprehensive review (EPA-600/7-77-066) was conducted of potential
techniques to measure chemical composition and temperature in flames with parti-
cular emphasis on instrumentally hostile environments such as research scale
furnaces. In Task II, laboratory investigations of the most promising diagnostic
techniques emerging from the Task I study have been conducted and are reported
herein. The Task II experimental studies focussed upon two diagnostic approaches,
saturated laser-excited molecular fluorescence and CARS (coherent anti-Stok.es
Raman spectroscopy).
111
-------
ABSTRACT
The report gives results of comparisons of saturated laser-excited molecular
fluorescence measurements of CH and CH in atmospheric pressure acetylene flames
with absorption measurements of these flame radicals. It was found possible to
saturate the fluorescence intensity of both CH and CN with readily achieved
levels of laser spectral intensity (lOp to 106 watts/cm2™"1). Coherent anti-
Stokes Raman Spectroscopy (CARS) thermometry investigations were conducted on
flame nitrogen in a variety of flames, including highly sooting propane diffusion
flames. CARS species sensitivity was addressed in a study of CO detectability
limits.
iv
-------
TABLE OF CONTENTS
Section
SUMMARY 1
INTRODUCTION 5
I SATURATED LASER FLUORESCENCE INVESTIGATIONS OF CH AND CN
SPECIES DETECTION 9
Introduction 9
Experimental Apparatus 1^
Absorption Measurements of CH an,d CN Radicals in Flames .... 20
Fluorescence Measurements 3^-
Experimental Results ^7
Discussion U9
II CARS INVESTIGATIONS IN FLAMES 57
Introduction 57
Experimental Approach 60
BOXCARS: Crossed-Beam Phase Matching 68
Flat Flame Thermometry 80
Sooting Flame Temperature Measurements 89
CO Species Concentration Measurements 97
III CONCLUSIONS AND RECOMMENDATIONS 107
Saturated Laser Fluorescence Investigations 107
CARS Investigations in Flames . 109
IV REFERENCES 113
APPENDIX I - CARS Spectra of Combustion Gases 1-1
-------
LIST OF FIGURES
Figure Page
1 CH and CN Energy Level Diagrams J-3
2 Schematic of Laser Induced Fluorescence Apparatus 15
3 Saturated Laser Fluorescence Experiment 16
U Slot Burner and Optical Geometry 21
5 Flame Absorption Measurements Schematic 23
6 Absorption Spectrum of CH in Oxy-Acetylene Slot Burner 25
7 Absorption Spectrum of CN in Nitrous Oxide-Acetylene Slot Burner. . . 27
8 Rotational Temperature Plot for CH (X2!!) v" = 0 . . . . 28
9 Rotational Temperature Plot for CN (X2£+) v" = 0 30
10 CH Emission Spectrum 35
11 CN Emission Spectrum 36
12 OMA Measurement of Dye Laser Spectral Output 38
13 Laser and CH Fluorescence Pulses 39
ill- Laser Excited CH Flame Fluorescence Uo
15 Laser Excited Anti-Stokes CH Flame Fluorescence ii-2
16 Laser Excited CN Flame Fluorescence U3
17 Laser Excited CH Fluorescence Spectrum U6
18 CH Concentration Data Reduction 50
19 CN Concentration Data Reduction 51
20 CARS - Coherent Anti-Stokes Raman Spectroscopy 58
-------
LIST OF FIGURES (Cont'd)
Figure Page
21 UTRC CARS Experimental Arrangement 62
22 CARS Experimental Apparatus 63
23 CARS Generation From Air at V/33A5 66
2h Broadband Dye Laser Spectral Stability 6?
25 CARS Phase-Matching Approaches 69
26 Collinear CARS Curves of Growth 71
27 BOXCARS Experimental Arrangement 75
28 BOXCARS Intensity Variation With Angular Detuning 77
29 BOXCARS Spatial Resolution 78
30 Computer Generated CARS Spectra 82
31 Experimental BOXCARS Spectrum of Flame N2 8k
32 Computed CARS Spectrum for Flame W2 86
33 Single Pulse CARS Spectrum of Flame N2 88
3^ Low Resolution CARS Spectra 90
35 BOXCARS Spectrum From Laminar, CoHg Diffusion Flame 92
36 Laser Induced Soot Emissions 9^
37 BOXCARS Spectrum of N2 in a Highly Sooting Laminar Propane
Diffusion Flame 96
38 Computed Flame CO Spectra 99
39 Crossed Beam, Phase Mismatched CARS Spatial Resolution 103
UO Crossed Beam, Phase Mismatched, CARS Spectrum of Flame CO 10^
iH CARS Spectrum of CO in CH^ Rich Flame . . . . „ 106
VI1
-------
LIST OF TABLES
Table
I Summary of Absorption Data 33
II Numerical Values Used in Sample Calculation of W1?
N and Q for CH 52
III Summary of Results of Fluorescence Measurements for
CH and CN 52
IV Collinear Phase-Matched CARS Probing Volume 70
V BOXCARS Phase-Matching Angles 73
Vlll
-------
SUMMARY
Under Contract 68-02-2176 sponsored by the Environmental Protection Agency,
the United Technologies Research Center (UTRC) is conducting analytical and experi-
mental investigations of non-perturbing, spatially precise, in-situ diagnostic
techniques to measure species concentrations and temperature in flames. Under
Task I, a comprehensive review (EPA-600/7-77-o66) was conducted of potential
techniques to measure chemical composition and temperature In flames with particular
emphasis on instrumentally hostile environments such as research scale furnaces.
In Task II, laboratory investigations of the most promising diagnostic techniques
emerging from the Task I study have been conducted and are reported herein. The
Task II experimental studies focussed upon two diagnostic approaches, saturated
laser-excited molecular fluorescence and CARS (coherent anti-Stokes Raman spectroscopy),
Saturated laser-excited molecular fluorescence measurements, using a tunable,
flashlamp-pumped dye laser, were performed on CH and CW in atmospheric pressure,
acetylene flames. The fluorescence was observed from the A2/\ of CH at U315A an(i the
B S state of CW at 3880A. The fluorescence intensity dependence on laser intensity
was observed to depart markedly from linearity at high laser spectral intensities
for both CH and CN indicative of saturation. Quite importantly, saturation occurred
at readily achieved laser spectral intensities in the range of KK to 10 Watts/
cmcm~^ for both CH and CW. These are the first observations of saturated fluores-
cence from these important flame radicals. The saturated fluorescence data permitted
evaluation of the species concentration and total excited-state foreign species
quenching rate for both CH and CW. In order to test the validity of the saturated
fluorescence results, absorption measurements of these flame radicals were made to
independently determine the concentrations. The absorption measurements were per-
1
-------
formed in a single pass on a specially constructed slot burner operating at atmos-
pheric pressure. Absorption indicated a CH concentration of about 60 ppm in an
oxy-acetylene flame and approximately 150 ppm of CH in a nitrous oxide-acetylene
flame. The concentrations measured by absorption were larger by about a factor of
two for CH and four for CN than the values determined by saturated fluorescence.
Various potential sources of error are analyzed. It is believed that the saturated
laser fluorescence measurements are low, most probably due to overestimating the
fluorescence sample volume due to the sharp spatial gradients of concentration pro-
bably present in these small, high pressure flames. This is not a fundamental
limitation however. These experiments have demonstrated the viability of saturated
laser fluorescence for trace species detection of important molecular radicals in
flames. Although problem areas remain, it is believed that saturated laser excited
fluorescence will become an important diagnostic in combustion applications.
CARS offers very promising potential for the temperature and species probing
of practical combustion environments due to its large signal conversion efficiencies
and coherent signal nature. Although CARS thermometry received the major emphasis
in the investigations reported herein, CO species sensitivity was also examined.
The CARS spectra are produced by mixing a 10 pps, frequency-doubled neodymium "pump"
laser with a spectrally broadband, laser-pumped "Stokes" dye laser. In this approach
which avoids the requirement to frequency scan the dye laser, the entire CARS spectrum
is generated in a single pulse permitting "instantaneous" measurements of medium
properties.
A new crossed-beam phase-matching technique, termed BOXCARS, has been demonstrated
for the first time and is described. This approach, which will be extremely useful
in studies of stratified flames, leads to greatly enhanced and unambiguous spatial
resolution in contrast to the conventional collinear phase-matching approaches.
Using this technique, moderate resolution (~ 1.25 cm" ) CARS spectra from hot Np,
obtained by scanning the spectrum in premixed laminar flames, show excellent agree-
ment with computer generated model predictions. Lower resolution (~ 2.7 cm"1)
collinear phase-matched CARS spectra of flame N2 have been obtained in a single
-------
10 nanosecond pulse using an optical multichannel analyzer. These single pulse
spectra also display good agreement with predicted spectra and demonstrate the
feasibility of single pulse thermometry. Measurements in a highly sooting., lami-
nar propane diffusion flame revealed the existence of a coherent spectral inter-
ference arising from electronically, resonantly-enhanced CARS generation from C^.
The Cg is produced by the laser vaporization of soot which occurs even on a ten
nanosecond time scale as illustrated by a study of the incoherent C Swan emissions.
The CARS from C2 occurs within the N2 spectrum which is generally employed for
thermometry. Reduction of the Stokes laser bandwidth and use of polarization fil-
ters permitted low distortion W2 CARS spectra to be obtained. These spectra, when
computer fitted, permitted determination of the temperature in the highly sooting
flame marking the first measurement in such flames by a remote, spatially precise
diagnostic technique. These sooting flame measurements are highly indicative that
CARS will be applicable to the probing of practical combustion environments.
The species sensitivity of CARS was examined in an investigation of CO detecta-
bility levels. Very good agreement between CARS CO spectra and computer modelling
was obtained at the k percent CO level in studies over a premixed flat flame. These
spectra display very interesting features such as the appearance of destructive
interferences not normally observed in linear spectroscopic processes. With the
fluctuations in the experimental apparatus it was difficult to detect the presence
of CO below the 1-2 percent level. The computer calculations indicate that CO
would be barely detectable to about 0.5 percent practically using conventional CARS
approaches.
In Section I the laser excited saturated molecular fluorescence investigations
of CW and CH are described together with the absorption measurements of these flame
radicals. Section II details the CARS investigations. Section III contains the
conclusions of these investigations and recommendations for future research and
development efforts in these areas. References are contained in Section IV.
Appendix I contains a description of the CARS computer code.
-------
-------
INTRODUCTION
With the advent of laser light sources, light scattering spectroscopic
diagnostic techniques are assuming an ever-increasing role in a broad spectrum of
physical investigations. Of particular importance is the potential application of
laser spectroscopy to the hostile, yet sensitive, environments characteristic of
those in which combustion occurs. Laser spectroscopic 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 Contract 68-02-2176 sponsored by the Environmental Protection Agency,
the United Technologies Research Center (UTRC) is conducting analytical and
experimental investigations aimed at developing non-perturbing, spatially precise,
in situ diagnostic techniques to measure species concentrations and temperature in
flames. The contract is divided into two tasks. Under Task I, a comprehensive
review (Ref. l) has been conducted of potential unobtrusive, in-situ techniques
to measure chemical composition and temperature in flames with particular emphasis
on hostile environments such as research scale furnaces. In Task II, laboratory
development of the most promising diagnostic techniques emerging from the Task I
studies has been investigated and these investigations are reported herein.
The Task I technical report entitled "Review of Laser Raman and Fluorescence
Techniques for Practical Combustion Diagnostics" (Refs. 1, 2) focussed upon four
general laser diagnostic techniques, namely: spontaneous and near-resonant Raman
scattering, laser fluorescence and coherent anti-Stokes Raman spectroscopy (CARS).
For diagnosis of highly luminous, particle laden flames, spontaneous and near-
resonant Raman scattering appeared to possess a low probability of successful
application even with advanced state-of-the-art laser sources. For diagnosis of
clean flames, i.e., low to moderate particle loadings, spontaneous Raman scattering
is the diagnostic of first resort because of its simplicity, high level of theo-
retical and experimental understanding and advanced state of development. 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 can be saturated. In this approach, depending on the degree of
saturation, fluorescence magnitudes do not depend upon or can be corrected experi-
mentally for quenching effects. The technique applies to a half dozen or so mole-
cules of intense combustion interest such as NO, OH, CH, CN, C2 and NH. CARS was
perceived capable of successful thermometry and major species measurements in
practical environments, although some potential jeopardies such as soot inter-
action effects had yet to be addressed. CARS sensitivity for most molecules is
limited to about 0.1 to 1 percent although sophisticated variants of CARS may lower
the sensitivity limit by one to two orders of magnitude.
For the Task II experimental investigations, two specific avenues of study
were of interest in view of the foregoing perspective, namely, saturated laser
fluorescence and CARS. For combustion applications, saturated laser fluorescence
is still in a very early stage of development. The success of the technique will
reside in the ability to achieve laser spectral intensities high enough to saturate
the molecular absorption transitions. Calculations of saturation intensities are
often questionable due to the lack of or impreciseness of the fundamental rates
required for such computations. The capability to minimize the effects of or to
correct for quenching in saturated fluorescence depends on the applicability of
simple, two level models. For the Task II experimental program, saturated laser
fluorescence was investigated via an examination of its feasibility to the measure-
ment of the molecules CH and CN which are of great interest in regard to NO for-
mation. These radicals can be probed directly with flashlamp-p'umped dye lasers
without the requirement to frequency double the dye laser and can be partially
saturated with such lasers. These two molecules are also of interest in regard to
the physics of saturated laser fluorescence, since CH is anticipated to exhibit two
level behavior while CN may display potential three level character. Although
CARS is further advanced at this stage than saturated fluorescence, questions remain
in regard to its ultimate capabilities and limitations. To probe these questions
-------
further, various CAES investigations were carried out. Major emphasis in these
studies was placed on the utility of CARS for thermometry in a variety of flames
including highly sooting diffusion flames. Species sensitivity was also addressed
in a study of CO detectability limits.
Section I of this report, which follows, details the saturated laser fluores-
cence investigations of CH and CN. Also described are concentration measurements
of these radicals via absorption which serve as a basis of comparison for the
fluorescence measurements. Section II details the CARS investigations in flames
conducted during the Task II program. Section III contains the conclusions reached
during these investigations and recommendations for future studies which are a
logical extension of the investigations presented herein. The references are
compiled in Section IV.
-------
8
-------
SECTION I
SATURATED LASER FLUORESCENCE INVESTIGATIONS
OF CH AND CN SPECIES DETECTION
Introduction
Saturated laser-excited molecular fluorescence is a promising technique for
the measurement of ppm level species concentrations in combustion environments. The
importance of saturated fluorescence to species determination was first realized
by Piepmeier (Ref. 3). Further theoretical developments were made by Daily (Ref. k)
and successful application of the technique to C^ measurement in a laboratory
flame has been made by Baronavski and McDonald (Ref. 5). The technique has several
important and distinguishing characteristics. One is that concentration may be
determined with high spatial resolution. The laser output is focused to a small
cross-sectional area, and appropriate optics collect emitted fluorescence from a
sample volume of order 1 mnP or less. Saturated fluorescence, which occurs only
in the presence of high laser spectral intensity, permits determination of concen-
tration without explicit knowledge of the details of fluorescence quenching due to
foreign species present. This latter feature is very important. In conventional
non-saturated fluorescence, the rate of quenching, including its temperature depen-
dence, must be known beforehand. This information is generally not available for
all foreign species present. This is particularly true of combustion applications,
and consequently, has the effect of precluding concentration measurement in such
environments.
The way in which saturation eliminates complications arising from fluorescence
quenching may be explained within the framework of a simple two level quantum
mechanical system. From rate equations for a two-level system, and in the steady-
state approximation, it may be shown that the number density in the upper level,
N2, is related to that in the lower level, N-^ by
-------
21
LLV the total rate of quenching of level 2; and A21, the spon-
taneous radiative decay rate of level 2. The fluorescence power, S-p, is propor-
tional to N2, and is given in general by
SF = (hcAF)(A21An)QcVc(N2/R1)N1. (2)
where h is Planck's constant; c, the speed of light; XF, the fluorescence wavelength;
Qc and Vc, light collection solid angle and sample volume size, respectively. In
the limit of small laser spectral intensity where I-^ (B^2 -t- B2]_)/(Q.21 + A21) <<: ^->
it follows from Eqs. (l) and (2) that
SF(1) = (hc/XF)(B12ATT)QcVc[A21/(Q21+A21)]ILv 1^. (3)
In Eq_. (3), it is evident that the fluorescence power depends on quenching through
the Stern-Vollmer factor [A2]_/(Q,p^+A2-|_)] (Ref. 6), and that SF depends linearly on
On the other hand, if the laser spectral intensity is large so that
^ >> L> lt follo'ws also from ®3-s- (-1) and (2) that
= (hc/XF) (A21An)QcVc[l + (g^g;,)]'1 N13 (4)
where use has been made of B21/B12 = g1/g2, in which g1 and g2 are level degenera-
cies. The power S-p corresponds to saturated fluorescence emission; it is inde-
pendent of ILV and Q21. This defines saturation for a two-level system. From the
latter inequality, saturation occurs when the rate for stimulated emission of
radiation significantly exceeds the combined rates for electronic quenching and
radiative decay. In practice, it may be very difficult to saturate fully since the
rate of quenching is generally large in a flame. However, in this case as well
provided that partial saturation is achieved, it has been shown by Baronavski and
McDonald (Ref. 5) that species concentration can be determined without reference to
10
-------
separately determined quenching data, and indeed that the total quenching rate is
determined together with the concentration. This case of partial saturation is
pertinent to the work reported herein; the related theory is somewhat more complex
than that above and is deferred to a later section of this report.
The saturation fluorescence technique is not applicable to all molecular
species. There are specific criteria which must be met. The molecule must first
of all have a known emission spectrum. This is not always the case since a mole-
cule upon optical excitation to a higher energy state may, in fact, dissociate prior
to emission, which precludes application of the technique. Secondly, the molecular
absorption wavelength must be accessible to a tunable laser source. In effect,
this means that the absorption wavelength must occur in the interval 2000$ to
1.5 microns. Thirdly, the rate of radiative decay of the emitting level must be
known. This is evident from Eq. (2) above. Finally, a means is required for handling
the uncertainties introduced by foreign gas quenching of the fluorescence intensity.
As indicated above, one way to achieve this end is to use a laser source with high
spectral intensity in order to saturate the absorbing transition. Upon application
of the above criteria only certain molecules are amenable to saturated fluorescence
detection. These are listed in Table VII of the Task I report (Ref. l) and include
C2, CH, CN, CS, NH, NO, OH, CHgO, HCN, NHg, and S02. The five molecules selected
for closer scrutiny in the Task I review were: CH, CN, NH, OH and WO. The latter
three molecules were not considered in the Task II experimental investigations due
primarily to wavelength considerations. NH, OH and NO have absorption wavelengths
of 3360, 306^ and 2074$, respectively. To achieve these wavelengths with the
available flashlamp-pumped, tunable dye laser requires frequency doubling of the
laser output. Although this may be done in principle, there results in practice
a reduction in laser intensity, which makes It more difficult to saturate the
molecular transition. Since it was anticipated beforehand that saturation might be
difficult to achieve, and hence, that maximum available laser spectral intensity
would be required, losses due to frequency doubling would be limiting and serious.
Moreover, the reduction is particularly significant for a flashlamp-pumped dye laser
11
-------
with its relatively poor output beam quality. Since CH and CN have absorption
wavelengths at 4315 and 3883^, respectively, and these wavelengths may be achieved
without frequency doubling, they were selected as candidates for detailed study.
There is an additional reason for selecting CN and CH. The developed theory of
saturated fluorescence is appropriate to a two-level system. The molecule CH,
apart from its rotational structure, is a two-level system with respect to elec-
tronic excitation as seen in Fig. 1 (Ref. 7). The CN molecule, on the other hand,
has in addition to its blue-violet B2Z+ -» X2S+ emission, strong red emission which
corresponds to the transition A2n -» X2Z+. The state B2S+ lies above ATI in energy
as shown in Fig. 1 and B2S+ -> A2!! emission also occurs. Since B £+ may de-excite
by these two channels, CN approximates a three-level system with regard to elec-
tronic excitation, and thereby distinguishes itself from CH in an important way.
By studying both CH and CN then, a test of the applicability of the two-level theory
to more complex systems is possible.
Since saturated fluorescence is a new technique with its related theory not
yet fully developed, it is important to have at hand a second, independent and
reliable reference technique. For this reason, it was decided to supplement the
fluorescence studies with absorption measurements. For absorption, the theory
and interpretation of spectra are well understood. Accordingly, comparison of
the two serves as a useful test of the accuracy of fluorescence measurements;
hence, CH and CN concentrations are determined in this work from fluorescence and
from absorption as well. There is an important difference between fluorescence
and absorption which must not be overlooked. Fluorescence is a point measurement
technique, while absorption is a line-of-sight measurement. Accordingly, only if
the flame is spatially homogeneous with respect to concentration is it expected
that fluorescence and absorption results would be the same and, indeed, that absorp-
tion results have any useful meaning. This, then, underscores the importance of
fluorescence in that the technique is applicable to inhomogeneous media while absorp-
tion is not. In this work, a homogeneous flame was created for the explicit purpose
of making meaningful comparisons of absorption and fluorescence results.
12
-------
CH AND CN ENERGY LEVEL DIAGRAMS
00
o
10
25
CO
I
o
X
to
tr.
LU
CD
5
ID
z
LU
>
cc
LU
y
^L
O
cr
o
20
15
10
A2A
~ 4315 A
3883 A
A27T
CH
CN
-------
The description of the experimental apparatus employed in the saturated
fluorescence investigations is given immediately below. This is followed by a
self-contained discussion of the absorption measurements of CH and CT. The
fluorescence measurements, which include the observation of saturation for both
CH and CN, are then summarized. Finally, species concentration and quenching rates
are evaluated and the results discussed.
Experimental Apparatus
The saturated laser fluorescence experimental layout is shown schematically
in Fig. 2 and photographically in Fig. 3. A wavelength-tunable pulsed dye laser
provides the exciting radiation which is focused by a lens into a small laboratory
flame which serves as the source of CH or CN. Appropriately placed lenses collect
the fluorescence at 90° to the exciting laser direction and focus it onto the slit
of a spectrometer to isolate the species fluorescence from other unwanted radiations.
Laser
The source of exciting radiation in these experiments is a Phase-R model 2100C
flashlamp-pumped dye laser employing a coaxial optical pumping configuration. A
xenon gas discharge lamp provides the pumping radiation. The lamp is capable of
absorbing about l60J of electrical energy. Although in principle the laser may be
repetitively pulsed, in practice it performs best on a single-shot basis. A Fabry-
Perot optical cavity is employed consisting of a flat output mirror and a grating
to spectrally condense and tune the emitted radiation. Laser energy is coupled out
of the cavity through the flat mirror which has a reflectance at the laser wave-
length of typically 25-50 percent. Depending on the operating wavelength, and the
degree to which the flashlamp is driven, the laser delivers (20-6o) mJ of spectrally
narrow, tunable energy in a pulse length of 150-300 nsec with a laser beam diameter
of 11 mm.
For the CH measurements, the laser was made to operate near 14-315$ which
corresponds to A2A -»X2I1, v = 0 -» v1 = 0 emission. The laser oscillated near this
-------
SCHEMATIC OF LASER INDUCED FLUORESCENCE APPARATUS
A, APERTURE
F, FLAME
G, HEAT ABSORBING GLASS
L, CONVERGING LENS
S, GLASS MICROSCOPE SLIDE
O, OBSCURATION DISK
SPECTROMETER
LASER
TO OSCILLOSCOPE
CHANNEL#2
TO OSCILLOSCOPE
CHANNEL #1
o
N>
I
U
f
-------
CO
en
01
O
SATURATED LASER FLUORESCENCE EXPERIMENT
cr\
01
o
CD
(O
I
CJ
p
W
-------
wavelength with Exciton Wh23 laser-grade dye dissolved In methanol. Typical dye
concentrations were 0.2 gm dissolved in 6 liters of methanol (1.3 x 10"^ molar).
When the laser cavity consisted of a 100 percent reflecting flat mirror and a
50 percent reflecting, flat output mirror, the laser energy was 200-300 mJ. In
order to narrow and tune the laser spectral output, the 100 percent reflecting
mirror was replaced by a PTR Optics TF-R2 echelle-type diffraction grating. This
grating has high dispersion since it is designed to operate in high order (n = 8-10)
at a relatively large angle of incidence, 50-60°. With insertion of the grating
into the cavity, the laser energy dropped to 50-60 mJ maximum. The energy measure-
ments were made with a Scientech model, S-3620, power meter positioned as shown in
Figs. 2 and 3. Also shown in Figs. 2 and 3 is a Princeton Applied Research optical
multi-channel analyzer (OMA). This instrument was used in order to determine the
wavelength, spectral width and reproducibility of the laser emission. In this way,
it was determined that laser output could be obtained in the interval ^210-^330A
by angle-tuning the echelle grating, and that the spectral width of the laser output
over this entire range was 2A. Peak laser emission occurred near U270A. Accordingly,
this wavelength was used to excite CH since the energy associated with it was high,
and the wavelength given coincides with an R-branch rotational transition in the
A2A •* X2H, v = 0 -» v' = 0 spectrum.
It was not possible to use the same dye solvent combination for CR measurements
as for GH. For ON emission corresponding to B2£+ -» X2£+, v = 0 -» v' =0, laser
oscillation near 3883$ is required. Exciton BBQ, laser-grade dye dissolved in 6 liters
of solvent (3.0 x 10 molar) gave optimum energy. In this latter case, it was not
possible to achieve energies near 3883$ as high as were achieved near I^IJA. With
two flat mirrors comprising the laser cavity, the energy was typically 60-100 mJ
for a 50 percent output coupling mirror. This represents a reduction in energy by
a factor of three from that associated with LDU23/methanol lasing. Not surprisingly,
then, the energy with the grating in the cavity was only 12-20 mJ. There was a
further acute difficulty with the BB^/p-dioxane combination. After only 100 shots
or so, the laser performance, as regards energy, degraded significantly. Consequently,
17
-------
in order to avoid operation at even further reduced energies, very frequent dye/
solvent changes were required. This problem was present for LDU23/methanol as well,
but in this case the higher, nominal pulse energies made losses somewhat more tol-
erable. The OMA measurements in the CN case displayed spectral condensation similar
to that above. The laser energy peaked near 38^-OA, and this, as above, served to
dictate an R-branch line in the CN spectrum as the exciting line for fluorescence
measurements.
Burners
For the CH measurements, two different types of burners were used. The first
of these was a small, portable oxy-acetylene welding torch. For this flame, the
oxygen and acetylene were premixed. Although this torch proved to be perfectly
adequate for fluorescence measurements, it was not possible to use it successfully
to make absorption measurements due to the relatively short absorption path length
associated with it. Accordingly, a similar slot-type burner with a considerably
longer active path length was constructed and used. A description of this burner
is given below as part of the discussion of absorption measurements. The slot
burner is evident in the apparatus photograph, Fig. 3. CN absorption and fluores-
cence measurements were made with the slot burner alone. To go from CH to CN with
the slot burner, it was merely necessary to replace oxygen by nitrous oxide as a
premixed component of the flame. The presence of CH and CN in the flames of these
torches was verified by viewing the corresponding emissions with a spectrometer and
identifying characteristic band spectra.
Optics
As may be seen in Figs. 2 and 3, a single lens was used to focus the output
of the laser into the flame. The lens is plano-convex and anti-reflection coated
and has a 50 mm focal length. Since the beam quality of the laser was poor it was
found useful to aperture the output beam prior to its incidence upon the lens. The
size of the focused spot for the CH/CN measurements was typically 1-2 mm dia. Further
18
-------
reduction of spot size by increased aperturing of the beam resulted in intolerable
energy reduction.
As shown in Fig. 2, the fluorescence intensity was viewed at right angles to
the direction of laser propagation. The light was collected by an imaging system
which consisted of two lenses and a 0.5 meter Jarrell-Ash spectrometer with l6$/mm
linear exit slit dispersion. The lenses and their spatial positioning were such
that the optics were f/8 and the magnification of the source at the entrance slit
of the spectrometer was near unity. The f/8 optics were selected to match those
of the spectrometer. The lens nearest the flame has an obscuration disk in front
of it. This disk sets a limit to the effective depth-of-field of the light col-
lecting optics (Ref. 8). The combination of lenses and spectrometer were made
light-tight to prevent stray light incursion.
Pulse Recording Instrumentation
The apparatus in Figs. 2 and 3 includes two RCA model 8575B photomultiplier
tubes. One is used to monitor the laser pulse energy, and the other, which is
attached to the spectrometer, records the fluorescence pulses. The 8575B photo-
multiplier has near peak spectral sensitivity at the CH and CN wavelengths of
interest, and has a 3 nsec response time. The laser and fluorescence pulses are
observed simultaneously by using a fast-responding, hOO MHz, Tektronix model 78^
dual-beam oscilloscope. For each firing of the laser, a photograph is taken of the
oscilloscope trace, which is then used for data analysis and processing. The laser
photomultiplier serves two purposes. It gives a shot-to-shot record of the laser
pulse amplitude and length. Secondly, it measured laser pulse energies below a
few mJ or so since this latter energy represents the lower limit of reliable measure-
ment of the Scientech energy meter. That is, at laser energies near a few mJ, the
photomultiplier is calibrated against the Scientech, and then the photomultiplier
is used exclusively for measurement of progressively smaller energies.
19
-------
Absorption Measurements of CH and CN Radicals in Flames
In order to check the accuracy of the fluorescence measurements, the con-
centration of CH and CN species has been determined by absorption in atmospheric
pressure flames. Measurements were made sequentially by absorption and fluorescence
techniques using the same burner operating under identical gas flow conditions.
Measurements have been made previously of absorption by CH and CN radical
species in flames. Results have been reported for CH in pre-mixed oxy-acetylene
flames by several investigators (Refs. 9-13). In contrast only one previous
determination (Ref. lU) of CN by absorption in a flame is known. That study dif-
fered from the present one in that CN was produced by the combustion of C2N2, 02
and H2 by Bulewicz et. al. whereas the combustion of C2H2 in N20 was observed in
the present investigation. Exemplary of the CH investigations is the work of
Bulewicz et. al. (Ref. 13) and Bleekrode (Ref. ll). In these studies absorption
was observed in low pressure (~ 5 Torr) slot burner flames using a multipath cell.
Up to 30 traversals of a 20 cm wide burner have been used. The results of the
present study will be compared to these studies later in this section.
It was desired to make the measurements with an atmospheric pressure burner
to minimize problems of adapting the fluorescence apparatus to the burner system.
This imposes a limitation on the accuracy of the absorption measurement because of
the short lifetime of the radical species. For example, Bulewicz has measured CH
concentration profiles above a low pressure (5 Torr) oxy-acetylene slot burner and
has determined under certain conditions that the concentration reaches a maximum
1 cm from the burner and that the concentration is 50 percent of maximum value at
1/U cm and again at approximately 3 cm. At atmospheric pressure it is expected
that the CH species will be confined to a smaller region near the flame zone in
inverse proportion to the ratio of the pressures. The effect of nonuniform flame
properties can be understood from Fig. h. Figure h illustrates the high aspect
ratio slot burner used in the present experiments. Premixed fuel and oxidizer exit
from a narrow slot and burn in a stable flame attached to the burner surface. An
optical probe beam propagating parallel to the horizontal burner surface and the slot
20
-------
SLOT BURNER AND OPTICAL GEOMETRY
o
10
I
o
I
-------
Is brought to a focus at a point in the center of the burner. Details of the bur-
ner are presented later in this section. The optical path depth varies from ray
to ray in the probe beam light pencil because the concentration in general varies
along the pencil. The probe beam aperture has been kept small to minimize this
effect. In the data reduction the optical path depth is assumed to be equal to the
width of the flame. The concentrations determined in this manner must be inter-
preted to be minimum values.
The apparatus used for the absorption measurements is shown schematically in
Fig. 5. Light from a high pressure xenon arc lamp is collimated and brought to
a focus above the slot burner. The light is recollimated and focused onto the
entrance slit of a 0.5 m monochromator. Radiation is detected at the exit sli.t by
a photomultiplier tube. In the absorption mode of operation a chopper between the
first focusing lens L2 and the flame modulates the xenon lamp radiation, thereby
providing discrimination against the emission from CH species excited in the flame.
In the emission mode the chopper is located between I/I- and the monochromator and
the xenon lamp is turned off or blocked. In the measurements reported here a point
1.5 mm from the burner exit is imaged onto the entrance slit. The entrance slit
height is 1 mm in all cases and the slit width is approximately 20 ^m. The focal
length of lenses L3 and Ik are nearly equal so that the magnification is nearly
unity.
The largest source of error In these experiments is the instability of the
arc source. The light output stability was 3 percent peak-to-peak. Some effort
was expended to Improve this since some absorptions are expected to be near this
level. A Hg-Xe lamp was found to have the same stability performance, and the light
of one Hg-Xe lamp was found to drift 10 percent after warm-up. A tungsten strip
lamp had much better stability; however, the signal intensity was an order of mag-
nitude lower. This caused problems because the low ratio of lamp signal to the
emission signal (in the bandpass of the tuned amplifier) caused an overload condition
in the tuned amplifier. With the xenon lamp there were no problems with the emission
from the flame.
22
-------
FLAME ABSORPTION MEASUREMENTS SCHEMATIC
o
XENON LAMP
AND HOUSING
SYNC
u>
APERTURE
L1
\
\
M1
4
CHOPPER
TUNED AMPLIFIER
CHART
RECORDER
PMT
I
0.5 METER
MONOCHROMETER
L2
SLOT BURNER
L3
L4
P
Ul
-------
Another effect which is a potential problem is beam steering by the strong
thermal gradients in the flame. The magnitude of beam steering was checked by
tuning away from the absorption region of the spectrum and observing the effect of
the presence of the flame. Beam steering effects were found to be small probably
because in the geometry of these experiments the light propagated perpendicular
to the temperature gradients.
The burner consists of two water-cooled, copper blocks which are joined to a
standard welding torch to eliminate flashback. The copper blocks are separated by
a piece of brass shim stock which determines the thickness of the gap between the
copper blocks. A shim thickness of 0.2 mm was used here. The flame length was
2.1+ cm. The flame was found to be stable and very uniform after the gap had been
cleaned. The oxy-acetylene flame was adjusted to eliminate the "feather." The
^2!")~<-'2H2 £^-ame was adjusted to have a small zone of CW red emission. Flowmeters
were used on the gas supplies to aid in resetting the flow rates.
A sample spectral scan of an oxy-acetylene flame in the CH (X%I -* A^A) region
is shown in Fig. 6. Emission and absorption spectra are superimposed on the same
plot. The emission spectra baseline has been displaced but the zero for the absorp-
tion spectra is as shown. The vertical scale for the emission spectra represents
the PMT signal voltage with the lamp off and the chopper positioned just before the
monochromator. Amplifier gain was a factor of 25 higher for the emission data. The
positions of several R and Q, branch lines in the v' = 0 «~ v" = 0 band have been
marked. The P branch is weaker and has not been used in this study. The presence
of absorption features is readily apparent. The peak absorption [(l -l)/I ]
where I is the lamp intensity at the signal minimum, and IQ is the unabsorbed in-
tensity corresponding to that wavelength, is obtained from this spectrum and from
a spectral scan of the lamp with the flame off. In practice little difference has
been observed between the flame off scan and signal level in the absorption scan in
regions of the spectrum where there is no absorption. Therefore, IQ in practice is
obtained from an I line fitted to the absorption scan.
-------
ABSORPTION SPECTRUM OF CH IN OXY-ACETYLENE SLOT BURNER
rv>
VJl
00
cc
ID
O
oc 3
LU
X
R9
III!
R8
R7
I HI
R6
ABSORPTION
III
R5
III
R4
I I
I
Q30
Q25
Q20
I I
R22 R.,2 R21
I I
Q15
I I
I I
Q10
II I I I
Q5
I I t I I
I I I I I I I I I I I 1 I I I I I I I I I I I I I I I I I I I I 1 I 1 I I I I 1 I I 1 I I .. I J , I I I I. 1 I I ,
4250
4260
4270
4280
4290
4300
4310
WAVELENGTH (ANGSTROMS)
01
2]
-------
The maximum absorption in the CH band is 6 percent. The resolution is insuf-
ficent to fully resolve all the components of the lines. In CH spin-orbit splitting
is large at low K and decreases as K increases, where K is the nuclear rotational
quantum number. Alternately, splitting due to X doubling is small at small K and
increases with K. The lowest spin-orbit components are resolved but overlapped
with the Q, branch. At the highest K values measured in these investigations X
doubling is almost resolved.
Figure 7 shows the absorption and emission spectra of the CN violet band
(X E+ -» B2£+) of a nitrous oxide-acetylene flame. Again the true zero is shown
only for the absorption spectrum. Several Av=0 transitions are overlapped with the
strongest v' = 0 =» v" = 0 transition. The Q, branch is very weak and the P branch
forms a head. The piling up in the P branch of the upper Av=0 transitions is great
enough to produce sizable absorptions which are superimposed on the absorption of
the R branch of the (0,0) band. The resolution (~ 2 cm~-^) is insufficient to
resolve any of the spin splitting of the lines. The maximum absorption is l6 per-
cent, but absorptions are generally less than 8 percent.
The rotational temperature T-n can be obtained by an appropriate normalization
of the data, accordingly
Iabg = const SK,K,r v exp (-Er/kTR,r) (5)
where SJM" is the rotational linestrength, v is the transition frequency expressed
in wavenumbers (cm ) , E^n is the energy level of the Kth state and k is the
Boltzmann constant. Figure 8 shows a semi-log plot of the CH absorption data nor-
malized by Sj£f£", where Sgig» is the average rotational linestrength for the two R-,
and R2 components. Rotational linestrengths have been taken from Beenakker et. al.
(Ref. 15)- The v factor has been neglected in this plot as only a small (l percent)
error is involved. The straight line shown has been least squares fit to the
data. P branch lines which are strongly overlapped with the Q branch have been
neglected in this fitting. The best fit line corresponds to 2600°K. Several runs
under identical conditions have been plotted on the figure.
26
-------
ABSORPTION SPECTRUM OF CIS! IN NITROUS OXIDE-ACETYLENE SLOT BURNER
10
00
o
ro
en
01
I
w
i 7
CO
cc
I-
=>
Q.
O
CC
I
Q.
01
X
3830
ABSORPTION
i 1 I I i i 1 I I I I I I I I I I I I i i I
I I i i I I I I I In n
40
35
30 25 20
(0,0) R-BRANCH (IM")
15 10 5 0
EMISSION
(3,3)
I
3840 3850 3860
WAVELENGTH (ANGSTROMS)
3870
(0,0)
3880
-------
ROTATIONAL TEMPERATURE PLOT FOR CH (X2IT) V" = 0
FIG. 8
^
L:
in
01
Q_
10~
D
D
- A A
3 I I I I I I I I I I I I I I I I I I I
1000
2000
EK»(CM
1-11
28
78-02-165-3
-------
Figure 9 shows a rotational temperature plot of the CN (0,0) band data.
Interferences due to overlapping with (l,l), (2,2), (3,3) P branches are clearly
visible. A complete treatment of this data would fit the absorption spectra with
computer generated simulated spectra which account for the overlapping with the
higher Av=0 P branches. This is beyond the scope of the present investigation.
A straight line has been faired through the data in such a way as to attempt to
compensate for the overlap effects. This straight line corresponds to 2900°K.
It may be noted that the one previous investigation didn't mention these diffi-
culties. This may be due to the lower temperature observed in that study.
The integrated absorption coefficient corresponding to a single transition
n", v", J' (or K") -» n1 , v1, J1 (or K' ) is given by
J\dv - (STTvV1 | Vn"Vv"
VV'J"
where g i and gn" are the degeneracies of the upper and lower electronic states
respectively, An,ntt is the electronic transition probability, AV«V» is the Franck-
Condon factor, and the other quantities have their usual meanings. The population
density of the state denoted by n", v", J" is given by
N
= £ (2-6. . )(2J+1) exp [-(Tn + GV f Fj) hc/kTj (?)
t u A
T . .
n v J tj, u , A
where the partition function Q is given by
Q = QeQvQR = S (2-6 )(2S'+1) exp (-hcTn/kT) • Qy • ^ (8)
nAs
and
Qv = Z exp [-G0(v)hc/kT]. (9)
v=0
The factor (2-6 ) accounts for A-type doubling wherein the electron spin (magni-
tude S) couples^ith the magnetic field produced by the electron angular momentum
A and splits the spin components.
29
-------
FIG. 9
ROTATIONAL TEMPERATURE PLOT4FQR CN (X2£+) V"=0
10
,-6
-O
° O
£>
LU
Q.
10-7
10-8
u
1.1
2.2 3.3 4.4 5.5
P-BRANCH HEAD INTER PER ENCES
I ' ' ' I I I I I I ' I t I I 1 I 1 J 1 I I I I I I I I I I
I I
1000
2000
3000
'
-------
V' =llfA' = o
(10)
= 0 if A1 ^ 0.
Tn, Gv and Fj are the electronic, vibrational and rotational term values. For
the temperatures of interest here hcTn/kT » 1 and
Combining these equations the integrated absorption for a single line is given by
• exp [-(Tn + Gy i- Fj) hc/kT]
When the absorption is small and the probe radiation bandwidth is greater than
the absorption line, the integrated absorption is related to the measured intensi-
ties by
"f
where AV is the spectral slitwidth (cm"-'-) and Jl is the absorbing path length. The
measured absorption depends on the resolution. The resolution in the present
investigations is moderate. Observations were made in first order with slits
approximately 20 (j,m. The resolution determined from pairs of just resolved lines
in the CH spectrum is 2.1 ± 0.3 cm"1. This resolution is not sufficient to resolve
the spin split components of either radical. Since Equation (13) applies to a
single transition the data reduction must be suitably modified.
The integrated absorption for two spin components, R-j_ and R2 of CH, for example,
is , Rl R2 v
i gv,t (.S,T',T" + bJ'J"' W hcB
= (Sm^c)"1
exp [-(Tn + Gy + Fj) hc/kT]
31
-------
If the two lines were exactly coincident, the measured absorption would be given
by
The data has been analyzed in this manner to account for the moderate resolution.
•p "P
For CH (2S"+l) = 2, therefore, (SjL,, + S 2 )/(2S"+l) = S „.
d d d d d d
Equations (l^) and (15) can be solved for the radical number density
^ Av 8TTV2
i —
'
I0 'peak I Vn"Vv" SJ'J" kcB (!6)
• exp [(Tn + Gy + Fj) hc/kT]
The data for the (0,0) transition with the lower state being the electronic ground
state has been analyzed for CH (A2A H> X2H) and CN (B2£+ -» X2E+), therefore, Tn =
G = 0 and equation (l6) reduces to
N = (——) — C1 -^— exp [Fjhff/kT] . (17)
I0 * sj.j" hcB
where C' = 8nv2c (A , iiA , n). Therefore, except for the v factor the radical
density is given by the intercept of the rotational temperature plot.
The data obtained here and the constant C' are summarized in Table I. The
uncertainties arising from several possible sources have been estimated. The
largest source of uncertainty is associated with data scatter which is typically
± 30 percent. The spectral bandwidth is undertain to approximately ± 15 percent.
As discussed previously, the uncertainty in assigning the optical path length may
be considerable. Although the visible extent of the flame is measurable to ± 5 per-
cent, the actual optical path accounting for the extent of the radical species in
the flame and the finite probe beam aperture is impossible to reliably assess. As
pointed out before, however, the actual optical path depth is only likely to be
less than the physical flame dimension taken here. Therefore, the radical densities
determined from the absorption studies must be interpreted as minimum values.
-------
The uncertainties assigned in Table I don't include uncertainties in the constant
C', which would be principally due to errors in the radiative lifetime measurement.
Historically the lifetime values have changed 33 percent, however, the current
uncertainties in this number are not likely to be so high. The uncertainties
assigned in Table I also don't include the uncertainty due to beam steering effects.
It is estimated from the shift in the baseline of the xenon lamp intensity with the
flame on and off that the maximum beam steering effect is to reduce the measured
radical density 15 percent for CN and 25 percent for CH.
TABLE I
Summary of Absorption Data
Species Intercept Temp (°K) C' (cm"1) N (cm"3)
CH (6.311.2) x 10~3 2600 5.08 x 10l6 (l.6±0.5) x lO1^
CN (3±0.5) x 10~7 2800±200 3-39 x 1013 (3.8+1.0) x 10ll+
For CH the electronic transition probability has been taken from Hesser and
Lutz (Ref. 16) who found a radiative lifetime of hjO ns. The Franck-Condon factor
(=0.993) was taken from Liszt and Smith (Refs. !?)• Transition frequencies from
the Moore and Broida (Ref. 18) compilation were used.
For CN the electronic transition probability was taken from Luk and Bersohn
(Ref. 19), who found the radiative lifetime to be 60.8 nsec. The Franck-Condon
factor computed by Wicholls (Ref. 20) (AW = ,909) was used. Transition frequencies
were taken from Jevons (Ref. 2l) or computed from spectroscopic constants found in
Suchard (Ref. 22) using formulae from Jevons (Ref. 23).
The results shown in Table I for CH correspond to 57 +_ ^ PPm which includes
the uncertainty due to beam steering. This result agrees quite well with other
investigators. Bulewicz et. al. (Ref. 13) measured ^ ppm and a rotational tem-
perature of 2250°K. Bleekrode determined the radical concentration to be k6 ppm
and TP = 2200°K. These measurements used higher spectral resolution than the present
33
-------
study and low pressure multipath optics. Both these measurements used earlier
radiative lifetime measurements (Ref. 2k] corresponding to 560 nsec. Using the
more recent data used in the present study would decrease the previously measured
radical concentrations by 16 percent.
+ 39
The results shown in Table I for CN correspond to 150 , ppm. Again beam
steering is reflected in the uncertainty but the finite aperture effect is not taken
into account.
There is no previous data with which to directly compare the CN data. Bulewicz
et. al. (Ref. lU) used different reactants. Her results correspond to 30 ppm and
2000 K. Again slightly different constants were used. The only known data for
the flame used here is in Gaydon (Ref. 25). There he reports that Gaydon and
Wolfhard measured the effective rotational temperatures of WH in emission in
several compositions of a C^/NgO flame. They found TR = 3100-3800°K depending
on the stoichiometry.
Fluorescence Measurements
Flame and Fluorescence Spectra
The flame spectra of both CH and CN were observed first without the presence
of the exciting laser radiation. Recording these spectra was important since this
served as a guide in selecting the particular transition to excite with the laser
and, equally important, the wavelength at which to observe the emitted fluorescence.
In this regard, it is important to point out that the measurements reported here
are not resonance fluorescence measurements. The wavelength of the exciting laser
radiation did not coincide with the fluorescence wavelength. This was done delib-
erately in order to preclude the potential interference of flame particulate Mie
scattering and Rayleigh scattering with fluorescence, which may occur when the
laser and fluorescence wavelengths coincide. Flame spectra of both CH and CN are
presented in Figs. 10 and 11, respectively, There are several distinct differences
to be noted. In the case of CH, the (0,0) band is relatively well displaced from
2 2
other vibrational bands of the A^& -» X II spectrum; accordingly, it is the only band
-------
CH EMISSION SPECTRUM
uo
4250
4300
WAVELENGTH - ANGSTROMS
4350
01
01
p
o
-------
CN EMISSION SPECTRUM
(0,0)
P_BRANCH (0,0)
CTN
(1,1)
R-BRANCH (0,0) AND (1,1)
OJ
O
jo
u
O
J_
3820
3840 3860
WAVELENGTH - ANGSTROMS
3880
-------
shown in Fig. 10. In the case of CN there is a clustering of the bands (0,0),
(1,1), (2,2), etc., within a relative^ narrow wavelength region with considerable
overlap among them. In both cases, the (0,0) band is of prime interest. The
presence of other bands complicates the identification of individual rotational
lines in the (0,0) band. This identification is important since it is necessary
to know precisely which transition (or transitions) the laser excites in order to
determine the total number density of ground state molecules. A second difference
between CH and CN, which is evident, is that the rotational structure in the case
of CN is not resolved, whereas it is for CH. This occurs since the rotational
constant of CN is about a factqr of seven smaller than it is for CH. The signi-
ficance of this difference is that it is easier to excite more than a single
rotational line in the CN case than it is ,f or CH. This latter possibility depends,
of course, on the relation between the spectral width of the laser and the spacing
between individual rotational lines.
The laser spectral widths corresponding to both the CH and CN measurements
as measured with the DMA are given in Fig. 12. As seen the laser spectral width
in both cases is about 2$. Representative traces of the laser and fluorescence
pulses are given in Fig. 13. The pulse lengths are both about 200 nsec. This
temporal similarity indicates that the quenched lifetime of the fluorescence is at
least as short as the duration of the laser pulse. Since the R-branch rotational
lines in CH are separated by about 108 near U273^5 the laser excites a single
rotational line. For CH, the line is split by both lambda and spin doubling. These
splittings, however, are much smaller in magnitude than the spectral width of the
laser pulse; in effect then, one excites all the molecules which correspond to the
given state of rotation.
Laser-excited fluorescence data for CH are shown in Fig. lU. These data were
taken with the oxy-acetylene welding torch. Each data point in Fig. ik corresponds
to a single laser shot. The laser was tuned to k27$ since, at this wavelength,
R-branch spectra are resolved, resulting in the excitation of a single rotational
line, and equally important, at ^73# the laser energy was near maximum. Fluorescence
37
-------
FIG. 12
OMA MEASUREMENT OF DYE LASER SPECTRAL OUTPUT
CH
CN
78-02-130-6
38
-------
LASER AND CH FLUORESCENCE PULSES
FIG.13
LASER
FLUORESCENCE
TIME 200 NSEC/DIV
78-02-130-5
39
-------
LASER EXCITED CH FLAME FLUORESCENCE
FIG. 14
XEXC~4273 A
= 0)I
100 i
z
D
QQ
< 101
h-
en
O
D
0.1
0.1
X •
•**•
1 10
LASER ENERGY-MILLIJOULES
100
78-02-36-3
ho
-------
was observed at 1+315# since this wavelength was sufficiently removed from 1+2738
in order to preclude Mie and Rayleigh scattering interferences, and the overlap of
Q-branch lines near ^315$ enhanced the fluorescence intensity. The dependence of
the fluorescence on laser energy clearly departs from linearity in Fig. lU and
indicates the onset of saturation. This makes it possible to determine the CH
concentration without prior knowledge of quenching. This is the first demonstration
of saturated CH fluorescence.
Data for laser excited anti-Stokes CH flame fluorescence are given in Fig. 15.
Again, the fluorescence departure from linearity is apparent. In this case, the
laser wavelength occurs in a region of the CH spectrum where there are many over-
lapping, predominantly Q-branch lines. This makes it more difficult to know the
fractional population excited by the laser, but is not a limiting factor in prin-
ciple. For the present, only the more straightforward Stokes-type data in Fig. l.k
are used.
Data for CN are given in Fig. l6. Much of the commentary given for CH above
applies to CW as well. The observation of saturation for CW, as for CH, has not
been reported previously. A photograph of the OMA spectral width measurement of
the laser pulse used to excite the CW spectra is given in Fig. 12. Since the
rotational spacing near the 381+2.2$ laser wavelength is of order 1A and the laser
spectral width was 2$, it was not possible to excite a single CN rotational line,
whereas this presented no difficulty in the case of CH.
Calibration Measurement^
For species determination, it is necessary to know the absolute magnitude of
the fluorescence power. Accordingly, calibration of the photomultiplier which
records the fluorescence signals is required. The light source used to calibrate
the RCA 8575B photomultiplier was a heated tungsten filament. The brightness tem-
perature of this filament was measured with an optical pyrometer. Since the emis-
sivity of the filament was known, the true temperature was calculated from the
emissivity and the measured brightness temperature. By using an optical filter at
1+1
-------
LASER EXCITED ANTI-STOKES CH FLAME FLUORESCENCE
100
[2!I(v=o)
XQBS ~ 4273 A, R(6)
QBS
CO
CD
tr
10
CO
2
LU
h-
LU
O
z
LU
u
CO
LU
EC
O
•••
to
I
0.1
0.1
1 10
LASER ENERGY - Ml LLIJOULES
100
p
at
-------
LASER EXCITED CN FLAME FLUORESCENCE
1.0
Aexc^3842 A [X2S (v =
(v = 0)]
FIG. 16
t
z
=>
CD
CC
<
Z
LLJ
LU
U
Z
LU
to
111
cr
O
0.1
0.01
/
0.001
0.001
0.01 0.1
LASER ENERGY-MILLIJOULES
1-0
10.0
78-05-51-1
-------
U33^ center wavelength and 100$ bandwidth positioned between the filament and
:hotomultiplier, the calibration was made at the CH wavelength. In addition to
the ortical filter for center wavelength and bandwidth definition, two spatially
separated pinholes were used in order to define the solid angle and filament area
for which the radiated power was observed. From Planck's formula for blackbody
radiation and accounting for the emissivity of the filament, the radiated power
striking the photomultiplier was computed and compared with the photomultiplier
output voltage. The tube calibration was performed only for the CH spectral region.
A separate calibration was not performed for CN since there is only a slight dif-
ference in the sensitivity of the photomultiplier at the CH and CN wavelengths of
^315 and 38808, respectively.
The laser-induced fluorescence power is proportional to QVe , where: Q is
the light collection solid angle; V is the sample volume size, and equal to the
product of sample length, i, and laser beam cross sectional area A; e is the spec-
trometer efficiency. The product Q^£e is required in order to evaluate the con-
centration from fluorescence power. A useful means to this end involves observation
of Rayleigh scattered light for the purpose of determining the product Qle , and an
independent measurement of laser beam area, A. The Rayleigh scattering is governed
by
Qr = Qi n()(njae), (18)
where Qr is the scattered Rayleigh intensity; Q^, the intensity of the incident
laser pulse; n, the number density of Rayleigh scatters; and — , the Rayleigh dif-
sn
ferential scattering cross- section. The Rayleigh scattering is observed from room
air. For this case, both n and — are known. By measuring both Q and Qj_, where
use is made of the results of the photomultiplier calibration in order to evaluate
^r, it is possible to evaluate (fife ) from Eq. (l8). Rayleigh data were taken at
laser wavelengths corresponding to both CH and CN excitation. In the case of CH,
the Rayleigh signal exceeded the maximum fluorescence signal by a factor of almost
seven. On the other hand, for CH the Rayleigh signal was about half of the
-------
fluorescence signal. A substantial part of this difference is due to the fact
that fluorescence power is proportional to the Einstein coefficient, A , as
in Eqs. (2, 3 and U). A21 for CK exceeds that for CH by about a factor of ten,
whereas the Rayleigh signal is about the same at the CH and CN wavelengths.
The area of the focused laser beam, A, which is required for the fluorescence
measurements, was determined by measuring the laser energy through a series of
apertures of well known size. The diameter of the beam was taken as equal to that
particular aperture which passed 80 percent of the laser energy incident upon it.
The minimum attainable laser spot size at the CH wavelength was about 1 mm diameter.
The spot size actually used for the CH measurements was 2.5 mm diameter. For high
laser spectral intensity and saturation, minimum spot size is desired. On the other
hand, since fluorescence power is proportional to laser beam area, small spot size
is not desirable from this point of view. The spot size chosen for the measurements
represents a compromise between these two considerations.
As part of the required calibration measurements, it is necessary to determine
what fraction of the emitted fluorescence is measured by the spectrometer. The
spectrometer has a linear dispersion of l6A/mm. Since the slitwidth for all fluores-
cence measurements was 0.^ mm, the spectrometer bandwidth was of order 6A. On the
other hand, the fluorescence is expected to be emitted over a much broader band-
width since rapid rotational equilibration occurs in the excited state prior to
fluorescence. In order to determine the fluorescence bandwidth for CH, the laser
was set to operate at k290$ and the fluorescence was measured as a function of
wavelength by tuning the spectrometer. These results are given in Fig. 17- By
comparison of Fig. 17 with Fig. 10, it is clear that the laser-induced fluorescence
spectrum approximates the flame spectrum. The exception is, of course, the Rayleigh
signal present in Fig. 10. Because of the foregoing similarity and the extreme
difficulty of taking data pulse by pulse as in Fig. 17 with the Phase-R laser, the
bandwidth factor or fraction of the total fluorescence actually observed, was deter-
mined in the case of both CH and CN by referring to flame spectra.
-------
LASER EXCITED CH FLUORESCENCE SPECTRUM
FIG.17
-4290 A
4300 4350
WAVELENGTH-ANGSTROMS
78-02-172-1
-------
Experimental Results
Data Analysis Procedure
In order to evaluate species concentration from the measured fluorescence
power, a procedure is used which was first described by Baronavski and McDonald
(Ref. 5)- This procedure is applicable in the important, practical situation where
it is possible to saturate a transition only partially. For a two level system,
the fluorescence power in exact form is given by
-1
SF =
Q21 + A21
+
(Blp + Bpl)
ZL<
(19)
where the spectrometer efficiency and bandwidth factors e and v, respectively, have
been included. Appropriate to conditions of near saturation where (lrv)~ is small,
Sp is developed in a Taylor series expansion about (l ) =0. The two leading terms
in this expansion are
N, (QP1 + AP1)
-
In order that this approximate expression for S-p be valid, it is necessary that
the
\-l
coefficient of (lLv ) be less than unity, i.e., (Q,2l+A2lV(B2i+Bi2-) < Lm Thls
means physically that the combined quenching and radiative decay rates may not
exceed those associated with stimulated emission and absorption of radiation. It
is evident from Eq. (20) that SF has a linear dependence on (ILV ) . In a plot of
SF versus (l^)"1, the intercept on the SF axis gives R-p and the negative slope
of this straight line is proportional to (QpjN^ permitting separate determination
of both I^ and Q,21. Aside from the fundamental constants h and c and the fluores-
cence wavelength \F, it is necessary to know Agl and the product (QCV"C) in order to
evaluate Wj_ by the intercept method. The value for A21 is taken from the literature
values; it may not be determined by the techniques described In this work. The
Quantity (n V ) follows from apparatus calibration via Rayleigh scattering of light
C C
and from laser beam cross-sectional area determination via insertion of calibrated
-------
apertures. The presence of the term (B2l + B^) in Eq. (20) causes no additional
uncertainty. This term may be written in the form
B2i Si
^ 21 12 12 ^12 ^2
where g1 and gg are the degeneracies of levels 1 and 2. Since B12 = (X /8
it follows that
Vv, (1 + —) (22)
Accordingly, the term involving stimulated rates has been expressed in terms of a
radiative rate and level degeneracies.
The concentration NX in Eq. (20) is the population of the particular rotational
level excited by the laser and not the total species concentration. It Is related
to the total species concentration, N, by appropriate Boltzmann factors. Boltzmann
factors applicable to CH and CN are those which account for both lambda and spin
doubling in the case of CH and spin doubling alone for CN. Spin doubling occurs
r) 2
since the ground states of CH and CN, as given in Fig. 1, are II and £, respectively,
where in both cases 2S' +1 = 2 or S1 = 1/2. Lambda doubling is present for the
CH n-state but it does not occur for the CN L-state. The most general Boltzmann
factor, applicable to a molecule with both lambda and spin doubling is given in
Eq. (7). An approximate expression for the total partition function, Q, in Eq. (8)
is given by
Q« (2-60jA)(2S'+l)(kT/hcBr)(l-e"hcU)/kT). (23)
If Eq. (23) is substituted in Eq. (7), the fractional, Boltzmann population, with
insertion of appropriate expressions for G and Fj, is given by
Mv,J _ (hcBr/kT) (2J+1) , -hccu/kT, -J(j+l)hcB /kT
-W (2S +1)
This is the fractional population in each of the two spin doublet states appro-
priate to a given state of rotation, J. Since the factor (2-6 ) factors out of
the final expression, Eq. (2k) is appropriate for the analysis of both CH and CN.
-------
Calculation of Concentration and Quenching
In order to explain the procedure used to evaluate ^ and Q, a representative
example is treated in detail. The example chosen is appropriate to CH measurement
with the oxy-acetylene welding torch for which saturated fluorescence data are
given in Fig. Lk. In order to apply the Baronavski-McDonald procedure (Ref. 5)
it is necessary to plot the inverse of the laser energy versus the fluorescence
intensity. This is done for only those laser energies which correspond to near
saturation. Such a plot is given for CH in Fig. 18 and for CN in Fig. 19. The
results in Fig. 19 were obtained from the data given in Fig. 16 for CN. In Fig. 18,
the least mean squares fit to the data results in a straight line with a negative
slope as would be expected from Eq. (20). The slope and intercept in Fig. 18 are
required in order to compute N-^ and Q. In order to evaluate the species concen-
tration, N, Eq. (2k) is required. Table II lists the numerical values of the
quantities required to calculate N^, N and Q, from Eqs. (20) and (2k). The parameters
AVj T and EL are related to the laser spectral intensity by ILv = (EL/r)/AAv.
Results for N-,, N and Q, for both types of burners and for both CH and CN are
summarized in Table III. There are no essential differences between the calculational
procedures for CH and CN.
Discussion
The fluorescence results given in Table III for species concentration from
slot-burner measurements may be compared directly with the corresponding values deter-
mined from absorption. The concentrations measured by absorption exceed those from
fluorescence by about a factor of two for CH and four for CN. Although the precision
of the absorption results is quite good, it is not possible to assign similar, small
uncertainties to the fluorescence results. The principal sources contributing to
the uncertainty in concentration determined from fluorescence are as follows. The
calibration factor (Qle), is determined with an uncertainty of 25-50 percent. This
is due to a lack of reproducibility in the Rayleigh scattering energy for a given
laser pulse energy. The explanation for this lack of reproducibility may be related
-------
CO
I-
00
a:
FIG. 18
CH CONCENTRATION DATA REDUCTION
30 i
-CH CONCENTRATION FROM INTERCEPT
20
CO
-z.
LU
I-
z
UJ
u
I 10|
co
LU
CC
O
•^
SLOPE YIELDS
QUENCHING RATE
I I I I
LEAST MEAN
SQUARES FIT
0.05
0.10
(LASER ENERGYH-dVULLIJOULES)-1
78-02-36-4
-------
CM CONCENTRATION DATA REDUCTION
CO
ID
DO
QC
CN CONCENTRATION
FROM INTERCEPT
v/i
LU
LU
CJ
o
to
LU
DC
O
SLOPE YIELDS-
QUENCHING RATE
LEAST MEAN-
SQUARES FIT
u_
03
O
M
I
O
CO
0.2
0.4
(LASER ENERGY)
1
-1
0.6
- (MILLIJOULES)-I
0.8
1.0
P
CD
-------
TABLE II
numerical Values Used in Sample Calculation of N-p N and Q for CH
O
extrapolated fluorescence power (Fig. l8 intercept) 1.37 x 10" Watts
\r-, fluorescence wavelength 4315-A
^21' spontaneous decay rate 1-97 x 10 sec
p o
A, laser beam area 4.9 x 10~ cm
v, bandwidth factor 0.14
(fUe), calibration factor (from Rayleigh data) 2.3 x 10
g-,, degeneracy of lower level ' 13
gp, degeneracy of upper level 15
Av, laser spectral width 10.74 cm
T, laser pulse length 280 x 10~y sec
Ey, laser energy at saturation 50 mJ
T, rotational temperature (from Table l) 2600°K
a'1
w, vibrational constant 2862 cm"
B , rotational constant lU.^57 cm"
TABLE III
Summary of Results of Fluorescence Measurements for CH and CR
M1(cm"3) N (cm-3) Q (sec"1)
CH (welding torch) 2.k x 1012 k.Q x lO1^ l x 10^
CH (slot burner) 4.2 x 1012 7.1 x 1013 3 x 109
CN (slot burner) 1.2 x 1012 8.1 x 1013 2 x 109
52
-------
to laser mode switching from pulse to pulse which results in a varying laser
intensity at the Rayleigh focal volume. Since the calibrations were performed in
room air, there may also be a varying Mie contribution from dust particles. With
improved experimental procedures, this uncertainty can be substantially reduced.
The beam quality of the flashlamp-pumped dye laser is relatively poor; it is
uadoubtedly not Gaussian and, indeed, its intensity distribution is unknown and,
most likely, varies from pulse to pulse. This introduces an uncertainty in the
measurement of the focused beam diameter. Indeed even in the case of a Gaussian
beam, Daily (Ref. 26) has shown recently, for example, that the absence of a con-
stant intensity distribution across the focal volume leads to anomalous apparent
saturation intensities. It is estimated that the uncertainty in beam cross-sectional
area could be as high as 50 percent, which implies a 25 percent uncertainty in
beam diameter. As mentioned above, it was very difficult to determine the spectral
distribution of laser excited fluorescence for both CH and CN since it was not
possible in practice to repetitively pulse the dye laser for the dye/solvent com-
binations which were used. For CW, the bandwidth factor, v, was estimated from
CN flame emission; however, because of the band overlaps in Fig. 11, this was an
imprecise procedure. Accordingly, the uncertainty in v for CH is of order 25 percent
and for CW of order 50 percent. If the above uncertainties are added to those
associated with photomultiplier calibration, selection of an appropriate value of
A21, and determination of extrapolated fluorescence power from curves such as are
given in Figs. 18 and 19, one concludes that there is a factor of 2 uncertainty in
the fluorescence results for CH with the CW results being less certain by a factor
of about 3. These uncertainties can go either way and in summary then, it appears
that there is a definite discrepancy between the absorption and fluorescence results.
There is one very plausible explanation for the concentrations measured by
fluorescence to be lower than those from absorption. The fluorescence power is
Proportional to sample volume size. In this work the sample volume is taken as the
product of laser beam cross-sectional area and spectrometer slit width. The laser
beam is focused to near its minimum spot size which is comparable to the radial
53
-------
flame dimension. If there is a concentration gradient across this dimension, which
is likely in these small, high pressure flames, the sample volume size is in effect
overestimated which leads to an underestimate of the concentration. There is no
information available regarding these gradients. However, CH flame emission data
for the welding torch taken with a narrow 10-20 micron spectrometer slit width
indicated a very pronounced dependence of the emission intensity with only slight
horizontal displacements of the burner. This suggests that such concentration
gradients are indeed present, and these may well be the root of the present
discrepancy.
Clearly, for accurate saturated fluorescence measurements, the measurement
focal volume should be considerably smaller than the characteristic spatial gradient
scale. However, since the fluorescence signal level is proportional to the sample
volume size, detection considerations may limit the gradient scale to which satu-
rated laser fluorescence is applicable. In the experiments reported here, the
laser focal diameter was limited by the poor beam quality of the flashlamp-pumped
dye laser. With laser pumped dye lasers the quality is substantially better per-
mitting finer scale probing. Within beam quality and signal detection limitations,
an alternative for fundamental studies of fluorescence involves the use of burners
with small gradients, e.g. a flat flame at reduced pressure.
In the fluorescence section of the Task I technical report (Ref. l) estimates
were made of the laser spectral intensities required for saturation of CH in
Table XVII and the detection limits for species measurements were given in Table XX.
It is of interest here to compare these latter estimates with results actually
obtained in this work. For the CH welding torch measurements reported herein, the
number of measured photons, based upon the photomultiplier tube calibrations, is
calculated to be 8.3 x 10 . When the number of photons calculated in Table XX is
adjusted for the concentration and calibration factors, 4.5 x 10^ photons are pre-
dicted. It is quite gratifying that this agrees with the above measured number
within a factor of two. The laser spectral intensity required to saturate CH in
this work was ILV = 3. It x 105 Watts/cm2™"1. This is about two orders-of-magnitude
-------
smaller than the value h x 107 Watts/cm2 cm-1 given in Table XVII of Ref. 1.
This difference may be explained by the fact that the quenching rate of CH in Ref. 1
12 -1
was taken as 1 x 10 sec , whereas the measured rate is 3 x 109 sec"1. The
smaller, actual quenching rate makes saturation correspondingly easier to achieve.
The rate in Ref. 1 for CH was taken as identical with that for the quenching of
excited WO Y-band emission by water vapor. This procedure was adopted since no
data for CH were available at the time. In retrospect then, this procedure over-
estimated the laser spectral intensities required to achieve saturation. In the
case of CN as well, the laser spectral intensity experimentally required to saturate
was 1 x ICr Watts/cm cm~ , again two orders-of-magnitude smaller than that esti-
mated in Table XVII of Ref. 1 due to the measured quenching rate being much smaller
than the values assumed earlier.
In brief summary, the CH and CN investigations conducted during the Task II
program yielded encouraging results in regard to the feasibility of saturated
laser fluorescence for radical determinations in flames. It was found possible to
saturate the fluorescence intensity of both CH and CN with readily achieved levels
of laser spectral intensity, in the range of 10^ to 10 Watts/cm2 cm" . The achieve-
ment of partial saturation permitted the determination of the CH and CN concentrations
in the flame as well as the excited state quenching rates via the procedure des-
cribed first by Baronavski and McDonald (Ref. 5). As far as can be ascertained at
this time, use of a simple two level model appears to be applicable as well to
species measurements of multi-leveled systems, e.g. CN. This is consistent with
the findings of Baronavski and McDonald in regard to Cg, which is not a simple two
level system either. The fluorescence measurements of these radicals was probably
low, based on the comparison with the absorption measurements, due most probably to
sharp spatial gradients of the species in the small, high pressure flames examined
here. In Section III, the conclusions of the saturated laser fluorescence investi-
gations are discussed and recommendations for future research indicated.
55
-------
-------
SECTION II
CARS INVESTIGATIONS IN FLAMES
Introduction
Coherent anti-Stokes Raman spectroscopy (CARS) has recently come to promi-
nence for combustion diagnostics based upon the pioneering investigations of Taran
and his coworkers (Refs. 27, 28-31) at ONERA in France. The effect was originally
discovered in the early sixties by Maker and Terhune (Ref. 32) and remained essen-
tially in the province of nonlinear optics until Taran's application of it for gas
phase diagnostics. In the United States, Harvey and his coworkers (Refs. 33-37)
have conducted numerous investigations into the technique. Barrett has demonstrated
both cw CARS generation (Ref. 38) and pure rotational CARS (Ref. 39). Broadband
CARS generation in a single pulse has also been obtained (Ref. ho). All of the
experimental work to be reported herein has been broadband or multiplex CARS for
reasons to be elaborated upon later. Publications describing investigations into
CARS are appearing at an ever increasing rate and the technique apparently will have
a major impact in molecular structure'and biological studies. Several very good
reviews of CARS have appeared recently (Refs. 35, 37, ^l)-
CARS is probably best understood by reference to Fig. 20. Incident laser
beams at frequencies u>i and o>0 (often termed the pump and Stokes beams respectively)
(^}
interact through the third order nonlinear susceptibility x (-uiy ^1" ^1' ~W2'
to generate a polarization field which produces coherent radiation at frequency
uu = 2tu -uo . When the frequency difference o^-u^ is close to the frequency of a
Raman active resonance, the magnitude of the radiation at uy then at the anti-
Stokes frequency relative to u^, can become very large. Large enough, for example,
that with the experimental arrangement described herein, the CARS signals from room
air N2 are readily visible. The incident beams, however, must be so aligned that
the three wave mixing process is properly phased. For gases which are nearly dis-
persionless, phase-matching occurs when the beams are mixed collinearly, i.e., aligned.
57
-------
CARS - COHERENT ANTI-STOKES RAMAN SPECTROSCOPY
APPROACH
CO.,
CARS, OJ3
CO
• ENERGY LEVEL DIAGRAM
VIRTUAL STATES-
,--
"X"
•X*
1
k
^
I"X
'"•M
Si**
/
f
y
1
•X1
1
>.
^
v
0)3, CARS
1
^
MOLECULAR ENERGY STATES
T
vl
00
SPECTRUM
CARS, CO3
P
ro
o
-------
Although easy to implement, collinear phase-matching possesses a drawback from a
diagnostic standpoint, namely, potentially poor and ambiguous spatial resolution.
A technique to circumvent this difficulty, which employs a crossed-triple-beam,
phase-matching scheme, was developed during the course of the contract and will be
described shortly. Assuming a fixed narrowband pump frequency, w , the CARS spec-
trum at u^ can be mapped out piecewise by scanning a variable frequency, narrow-
band laser source at u^. Or, as depicted, if a broadband source at w is employed,
the entire CARS spectrum can be generated simultaneously permitting fast, time-
resolved measurements of flucutating phenomena.
CARS offers very promising potential for the diagnostic probing of high inter-
ference environments such as those typical of combustion processes for two reasons.
First, in contrast to spontaneous Raman phenomena, CARS is a fairly strong process
leading to signal levels typically many orders of magnitude larger than those from
Raman scattering. Second, the CARS signals are coherent. Consequently, all of the
CARS radiation can be collected. Contrast this with the situation pertaining in
the normal Raman process where photons are scattered over 4rr sr and are collected
only over a limited solid angle, n. For f/3 optics, only 1 percent of the signal
is collected in an isotropic, incoherent scattering process. Furthermore, since
the CARS radiation can be collected in an extremely small solid angle, discrimination
against background luminosity and laser induced particulate interferences, e.g.
incandescences (Ref. ij-2), fluorescences, is greatly facilitated. Thus, CARS is
expected to offer signal to interference ratio (S/l) improvements of many orders
of magnitude over spontaneous Raman scattering. Based upon calculations presented
in the Task I review report (Ref. l) CARS appears capable of probing practical com-
bustion environments successfully over a broad range of operating conditions.
Before CARS can be practically implemented, however, a number of questions
needed to be addressed such as spatial resolution, computer synthesis of CARS flame
spectra, feasibility and practicality of single pulse thermometry, the effects of
soot particulates and species sensitivity limitations. The investigations conducted
during the Task II experimental program and reported herein sought to answer these
59
-------
questions and to provide an experimental foundation on which to base future avenues
of relevant development. In the section which follows, the experimental apparatus
employed in these investigations will be described. Then BOXCARS, a nomen for a
crossed-beam, phase-matching technique will be described. BOXCARS provides fine
and unambiguous spatial resolution for CARS diagnostics of spatially inhomogeneous
media. Thermometry investigations, scanned and single pulse, in premixed flat
flames are described. Then thermometry investigations in a highly sooting diffusion
flame are described. The section concludes with an initial examination of CARS CO
detection sensitivity in flames.
Experimental Approach
Although CARS has no threshold per se and can be generated with cw laser
sources (Ref. 38), high intensity pulsed laser sources are required for the probing
of high interference environments to generate CARS signals well in excess of the
various sources of interference. Because CARS is highly nonlinear in its temperature
and density dependences, signal averaging in temporally fluctuating media will
obscure and distort the CARS spectrum rendering it of little utility or leading to
measurement errors. Thus, measurements in fluctuating environments require the
CARS spectrum to be generated and captured with each laser pulse. Hence, the
individual laser pulses must be energetic enough to provide a statistically large
number of CARS signal photons in each spectral detection interval. So restricted,
laser selection narrows to a choice between ruby and frequency-doubled neodymium
(2xNd) with the latter probably preferable.
2xNd lasers can be operated at a repetition rate generally an order of magni-
tude higher than ruby, at least 10 pps versus 1 pps. In an instrumental application,
this obviously expedites data collection. In the laboratory, it permits the use of
boxcar averagers and spectral scanning techniques (when applicable). At 10 pps or
better, the experiment behaves much like a cw experiment permitting "tweaking" of
adjustments, while the experiment is running, an important feature for the critical
alignment requirements of CARS experiments. If a portion of the pump laser is split
off to pump a Stokes beam dye laser, 2xNd lasers at 5320$ can pump very efficient
60
-------
dyes in the 5500-65OQ# region of the spectrum, while ruby lasers must pump lower
efficiency near ir dyes. Furthermore, the CARS radiation from 2xNd resides in a
region of higher photomultiplier tube quantum efficiency than does the CARS from
6^3#, a small advantage. The ruby laser wavelength disadvantages are mitigated
by frequency doubling to 3^71$. Whether one would encounter desirable or undesirable
(from background constituents) two-photon resonant-enhancement effects for combus-
tion diagnosis is not known at this time. This is also true of using the third
harmonic of neodymium at 353:$. As an aside, one interesting feature of operating
in this region is that the CARS signals would be about ikk times stronger from 2x
ruby relative to ruby, and 19.5 times stronger for 3xNd than 2xNd on the basis of
wavelength scaling alone. Considerations of harmonic generation efficiencies would
diminish these gains however. For the investigations reported herein, a frequency-
doubled neodymium laser was selected based on the foregoing rationale. In some early
investigations of CARS (Ref. 27), the Stokes beam was generated via stimulated Raman
scattering of the pump beam. Such an approach is limited to probing only stable
species from which stimulated Raman is readily produced, is not suitable for broad-
band CARS generation, and thus, is not very versatile. A more flexible scheme is
to employ a tunable dye laser, generally pumped by splitting off a portion of the
pump laser. CARS configurations employing 2xNd lasers and dye lasers pumped by a
fraction of the 2xWd output are also being used at, among other places, the Naval
Research Laboratory, Stanford University and Oregon State University.
In Fig. 21 is shown a schematic of the CARS setup used for the Task II experi-
mental investigations. Various configurations were used during the course of the
investigations as will be apparent but the major features remained the same. In
Fig. 22, a photograph of the experimental set up is shown. Referring to Fig. 21,
the output of a Quanta-Ray neodymium laser (Model DCR-lA) is frequency doubled to
generate a horizontally polarized, "primary" green beam at 53208. The primary green
and residual 1.06^ are separated in a splitter section and the residual I.o6p, doubled
to generate a "secondary" green beam. Depending on the condition of the flashlamps
in the laser and the frequency doublers, the primary green is typically between 1.5
61
-------
UTRC CARS EXPERIMENTAL ARRANGEMENT
• QUANTA RAY NEODYMIUM LASER
m
o
P.A.R.C. OMA
1 m J-Y DOUB'LE MONOCHROMATOR (RAMANOR)
OR 1 /4 m J-Y SINGLE MONOCHROMATOR
Tl
P
ro
-------
CARS EXPERIMENTAL APPARATUS
FIG. 22
78-65-A
78-02-189-2
63
-------
to 2.5 W (150 to 250 mJ pulses, 10 pps, 10"8 sec pulse duration). As seen in the
figure, part of the primary is split off by the beamsplitter, BS, focussed., and
reflected by the dichroic, D, to axially pump a flowing dye cell oscillator. The
output from the dye laser is amplified in a flowing dye cell axially pumped by the
secondary green. The secondary green passes through a piece of KG3 air-cooled,
Schott glass placed at the Brewster angle to absorb any remaining 1.06^. The
secondary green is focussed off the same dichroic as the primary and pumps the dye
in the amplifier cell. The oscillator and amplifier cells are in flow series. A
small, magnetically coupled, stainless steel gear pump circulates the dye from a
1.5 liter reservoir through a 0.6|_i filter to remove and prevent large air bubbles.
The dye cells are oriented at Brewster's angle and produce a horizontally polarized
output; actual oscillator efficiencies vary between 20-30 percent while the amplifier
operates between ko and 50 percent efficiency. One problem with coaxially pumped
configurations is the existence of 5320A focal points or high intensity regions on
the various optical elements. Furthermore, many of the optical elements require
dichroic coatings. Although much of the work to be reported herein employed coaxial
pumping schemes, the system has recently been converted to off-axis, dye laser
pumping, as seen in Fig. 22, with comparable or slightly improved efficiencies. The
pump and Stokes beams are combined at a second dichroic and collinearly focussed
into the region under study to generate the CARS radiation. Optical delay can be
inserted into the primary pump leg to ensure that the pump and Stokes laser pulses
are temporally overlapped. The Stokes, pump and CARS beams are recollimated and
then dispersed with a double extra dense flint prism. The pump and Stokes beams
are trapped, while the CARS passes through cutoff filters prior • to analysis by a
1-m double monochromator (Ramanor, HG2S, Jobin Yvon) or to an optical multichannel
analyzer, OMA (Princeton Applied Research) fitted to a 1/k m spectrograph equipped
with either a 1200 or 2^00 gr/mm holographic grating. Recently the 1 m double mono-
chromator has been modified to accept the OMA as well. The experimental configuration
displayed is capable of very intense CARS generation. With collinear phase-
matching, CARS generation from N2 in room air is readily visible and easily phot-
-------
graphed as shown in Fig. 23. The CARS beam is annular as shown and is an artifact
of the unstable or diffraction-coupled resonator employed on the neodymium laser.
The laser output is an annulus with outer to inner radius ratio (magnification)
of about 3-2.
To produce the broadband Stokes laser output desired for single pulse CARS
work, a variety of dye oscillator cavity configurations were tested. Bandwidths on
the order of 100 cm"1 FWHH are desired since the vibrational band separation in
JT2 is about 30 cm . A 1200 gr/mm grating at low dye concentrations (~ 5 x 10"5 M)
resulted in a bandwidth of only ~ 15 cnf . As the dispersion in the cavity was
reduced by employing successively less dispersion, i.e., 600 gr/mm or 150 gr/mm
gratings or various density prisms, the bandwidth increased. With the low dis-
persion required to obtain adequately large bandwidth, however, control of the
laser center frequency was lost, i.e., the laser would oscillate off-axis at the
wavelength corresponding to maximum gain. Consequently, no dispersion was employed
at all. Rather, a flat-flat Fabry-Perot oscillator arrangement was adopted and
the bandwidth appropriately centered by selection of appropriate dyes and dye con-
centration (Kef. ifO). Small adjustments in the Stokes laser center frequency are
also possible by varying the cavity mirror separation. Bandwidths vary from 100 cm
to 200 cm depending on oscillator pump energy and whether the dye amplifier is
used. By using mixtures of two different dyes, very broad bandwidths are possible,
i.e., 250 to 300 cm"1. For N thermometry, which requires the Stokes laser to be
c
centered at 6073^, Rhodamine 6^0 (Kef. 1*3) at a concentration of about 5 (10 )M in
ethanol is employed.
For single pulse CARS diagnostics or for spectral scanning of laminar (i.e.
steady) situations it is important that the dye laser spectrum be smooth and repro-
ducible from pulse to pulse. In Fig. 2k are shown five OMA traces of the broadband
dye laser spectrum. Each spectrum corresponds to a single laser pulse and was ob-
tained by gating the intensifier stage of the OMA with a ~ lOOnanosecond pulse.
Each channel corresponds to 1.^. As seen the spectra are fairly smooth and repro-
from pulse to pulse.
-------
FIG. 23
o
CARS GENERATION FROM AIR AT 4733A
78-02-108-2
66
-------
FIG. 24
BROAD BAND DYE LASER SPECTRAL STABILITY
1.4 A/CHANNEL
78-02-108-1
61
-------
In broadband CARS, either the linewidth of the pump laser or the resolution
of the monochroraator determine the ultimate resolution of the spectrum. The 1-m
double monochromator has a limiting resolution of about 0.5 cm in the visible.
The 2xNd laser has a linewidth of about 1.2 cm'1 and thus limited the resolution
of the CARS spectra to around this value. However, as will be seen, this moderate
resolution is more than sufficient. The large pump laser linewidth is detrimental,
however, in regard to the strength of the CARS radiation. In this regard it would
be desirable to have the pump linewidth comparable to the Raman linewidths of the
hot flame gases, ~ 0.1 cm
BOXCARS: Crossed- Beam Phase Matching
As mentioned previously, the incident pump and Stokes laser beams must be
aligned in a precise manner so that the CARS generation process is properly phased.
The general phase-matching diagram for three wave mixing is shown in Fig. 25a and
-»-»-*->
requires that 2k-^ = k? + k.,. k. is the wave vector at frequency UK with absolute
magnitude equal to cu^n^/c, where c is the speed of light and n.j_, the refractive
index at frequency u). . Since gases are virtually dispersionless, i.e., the refrac-
tive index is nearly invariant with frequency, the energy conservation condition
uj.-, - 2uo, - CD indicates that phase-matching occurs when the input laser beams are
aligned parallel to each other, Fig. 25b. Collinearity, however, possesses a
problem in regard to spatial resolution. Since the CARS signal is coherent and
undergoes an integrative growth process, the spatial resolution cannot be well de-
fined by imaging techniques such as those successfully employed in spontaneous Raman
approaches to yield fine resolution. Since the CARS signal generation scales as
the intensity product Ij_2I2 "the incident laser beams are generally tightly focussed
for diagnostic purposes when collinear phase -matching Is employed. For diffraction-
limited beams, the interaction is assumed to occur primarily within a cylindrical
volume of diameter cp and length i given by (Ref. 30 )
68
-------
CARS PHASE-MATCHING APPROACHES
FIG. 25
a) GENERAL
b) COLLINEAR
c) BOXCARS
78-02-108-6
-------
where f is the focussing lens focal length; D, the beam aperture incident on the
lens; and X, the wavelength, In Table IV, the probe volume focal diameter, cross-
sectional area and length are tabulated for various focal length lenses for a 1 cm
diameter beam at 5320A.
TABLE IV
Collinear Phase-Matched CARS Probing Volume
Focal length (cm) Diameter (cm) Area (cm2) Length (cm)
10 6.11 (-4) 2.93 (-7) L22 (-2)
20 1.22 (-3) 1.17 (-6) ^.88 (-2)
50 3.06 (-3) 7.35 (-6) 3.05 (-1)
100 6.11 (-3) 2.93 (-5) 1.22
Depending on the specific diagnostic circumstance, i.e., the focussing lens to
measurement point separation, the spatial resolution may be less than that which
is desired. Although the resolution is very good for short focal length lenses,
gas breakdown may limit the input beam intensities leading to diminished CARS signal
levels. Many laser beams are not diffract!on-limited resulting in much poorer
spatial resolution than that tabulated here. For example, for a "three times dif-
fraction limited" beam divergence angle, the linear resolution would be about an
order of magnitude poorer. Specifically, in Fig. 26, CARS radiation curves of
growth are displayed for beam divergence angles of 1 and 2 milliradians and various
practical focal length lenses. The calculations were made using a plane wave
analysis applicable to the parameter range selected (Ref. kh) corresponding to so-
called "loose focussing." As is readily apparent in Fig. 26 significant CARS
generation occurs well before the focal region in the prefocal regions. If the
coherence length is sufficiently large, significant growth could occur beyond the
focal region as well. To make the situation worse, in the presence of density
gradients the resolution further degrades since CARS scales as the square of the
70
-------
§
o
cc
O
co
DC
<
O
Q
LU
N
QC
O
COLLINEAR CARS CURVES OF GROWTH
FIG. 26
0.05
0.02
0.01
BEAM DIVERGENCE
2 mrad
-•• — i— — 1 mrad
0.2 0.4 0.6 0.8
NORMALIZED DISTANCE TO FOCUS
78-02-108-4
71
-------
gas density. For probing of a hot flame operating at atmospheric pressure,
significant contributions to the CARS signal may originate from the cool, high
density regions adjacent to the flame. For example, in the experimental setup
used here, initial attempts to record flame spectra with ^0 cm focal length lenses
on a 7.5 cm dia burner were unsuccessful due to a dominance of room air W2 contri-
butions. With a piece of Schott glass GG^95, which transmits the laser and Stokes
beams but absorbs the CARS beam, it was determined that about 10-15 percent of the
total room air CARS signal was generated in the first 32 cm from the lens. With
the flame ignited, the CARS signal from the hot, low density flame gases was sub-
stantially less than the cold, room air contribution.
CARS signal contributions may derive also from the various elements in the
optical train, e.g., lenses, when collinear phase-matching is used. These could be
significant when low gas densities or weak resonances are being probed. Clearly it
would be desirable to avoid beam overlap and potential three wave mixing in all
regions except the desired measurement location. In an effort to avoid collinearity,
one could attempt to introduce the pump and Stokes beams at a slight angle to one
another. As phase mismatch is deliberately introduced in this manner, the CARS
signal generating efficiency will diminish. At Ak/ « 3, where Ak is the magnitude
i ~* ~* •* i
of the phase mismatch, i.e., 12k-j_ - k2 - k^], the CARS efficiency will have decreased
by an order of magnitude (Ref. 35). Accepting this loss of power for the moment
and if a 0.1 cm spatial resolution is desired, then a Ak of 30 cm"1 would be toler-
able. At wavelengths of 5320$ (u^) and 6073$ (the N2 ground vibrational state
Stokes wavelength), a 30 cm phase mismatch would be produced at an angular
separation of only 1 assuming no dispersion. Although one could operate in this
manner, it is clearly inefficient and the actual spatial resolution will depend very
critically on the precise angular separation. During the course of the Task II
experimental investigations, CARS has been generated using crossed <^ and uu beams
_i_ £_
as will be described in the CO studies, but no systematic study of the scaling of
the signal generation efficiency or spatial resolution was attempted.
72
-------
A method which permits large angular separation of the input frequencies,
while still satisfying the phase matching requirement is depicted in Fig. 25c
(Kef. 1*5). m this approach the u^ pump beam is split into two components which
are crossed at a half angle of a. The u^ Stokes beam is introduced at angle 9
producing phase-matched CARS at angle cp. Based upon the shape of the phase-matching
diagram, Fig. 25c, this technique has been termed BOXCARS. The appropriate phase-
matching angles are readily related from simple geometric considerations as follows:
n2u)2 sin 6 = ruiUo sin cp (26)
n2®2 cos 9 + n au cos cp = 2n uu cos a (27)
Neglecting dispersion, Table V lists several examples of phase-matching angles for
BOXCARS generation from N2 assuming a pump wavelength of 5320A
TABLE V
BOXCARS Phase-Matching Angles
9_ a_ eg i a eg
5
10
15
20
25
30
ko
50
1^.38
8.81
13.2
17.6
21.9
26.2
3^.7
lj-2.9
3.89
7.78
11.6
15.5
19.2
22.9
30.1
36.7
60
70
80
90
120
150
180
50.7
57-8
6U.1
69. ^
78.7
82.1
82.9
1|2 . 5
1^7 > J_
50.1
51.2
^2.5
22.9
0
The very large angles shown lead to interesting phase-matching configurations which
may be of utility in certain laboratory studies. In general, the smaller angles
^e of most interest for diagnostic applications where input and output optical
apertures are limited.
73
-------
It is also interesting to note that the angular separation 6- beam on the focussing/crossing lens permitting the
phase-matching angle 9 to be varied. After passing through the crossing point, the
four beams, i.e., CARS @ UK* u>j_, u^, m are recollimated by a second lens, generally
the same focal length as the focussing lens. Two of the components, ux and o^, are
trapped although they could be sent to a reference leg if desired to generate a
normalizing signal. The CARS radiation and the remaining UL are dispersed with a
prism and the CARS sent to appropriate instrumentation. It should be noted that
generally at small angles (a < 10°), the u^ and uu CARS radiation, although angularly
-------
BOXCARS EXPERIMENTAL ARRANGEMENT
FIG. 27
ROTATABLE
OPTICAL FLAT
78-02-153-1
T5
-------
separated, are not necessarily spatially separated which depends on the beam
diameters used. In all of the work reported here, this was always the case and a
prism was used to separate the u)n and ub radiations.
In the initial demonstrations of BOXCARS, an 89 mm diameter, ^83 mm focal
length lens was employed and CARS was generated from room air N2- a was selected
to be 3.1°; Q and cp were 3.6° and 2.8° respectively. That the detected signal was
indeed CARS was verified by separately blocking either of the two u^ or u>2 components
which led to the expected disappearance of the signal. Most convincing was the
effect on the signal of rotating the optical flat through which uv, was transmitted,
which as previously mentioned varies the phase-matching angle 9. In Fig. 28, the
variation of the CARS signal with deviation in 6 from the optimum phase-matching
angle is shown. The angular mismatch was calculated based upon the measured angle
of the optical flat, its thickness and assuming its refractive index to be 1.5.
Roughly, each 0.1 degree of angular deviation corresponds to a calculated phase
mismatch JAk) of about 20 cm" . The CARS signal displays little variation in
magnitude for misalignments up to 0.1°. Beyond that the signal declines fairly
slowly with increasing phase mismatch. The variation in signal is not symmetric
about the optimum angle. The slower decrease in signal with increasing negative
angular misalignment is probably due to a larger interaction length as ujg draws
closer to UL than when its separation from uu, increases.
In the initial experiments just reported, the spatial resolution was not very
good. The resolution was measured by examining the CARS signal generated from a
translatable 1 mm thick microscope slide. All of the signal generation occurred
within a 1.3 cm extent. Integrating the respective piecewise contributions revealed
that the total signal grew from 10 to 90 percent over a distance of about k mm.
For the sooting flame studies to be reported later, slightly larger angles were
employed and the resolution improved. In these studies a 305 mm focal length lens
was used and the angles were 01 = 5.0, 0 = 5.7° and cp = k.k° The resolution measured
with a 0.15 mm slide cover is shown in Fig. 29. Signal generation occurred within
a 2.5 mm extent and the 10 to 90 percent growth in the integral occurred over an
76
-------
FIG. 28
BOXCARS INTENSITY VARIATION WITH ANGULAR DETUNING
-0.8
-0.6
-0.4 -0.2 0 0.2 0.4
ANGULAR MISMATCH-DEC
78-02-108-7
11
-------
FIG. 29
BOXCARS SPATIAL RESOLUTION
f = 305 mm
0 = 5.7° a = 5.0° 0=4.4°
POINT CONTRIBUTIONS
-.5 0 .5
DISTANCE FROM FOCUS-MM
1.5
78
78-02-111-1
-------
a
were
extent of 0.9 mm. The focal beam diameters were measured by translating
piahole across the attenuated focal spots. The diameters at the 1/e2 points
found to be 380^ for the u^ and l£fy for the uv> beams. These measurements imply
beaa divergences of 1.6 milliradians for the dye laser (after expansion in the 2x
telescope) and 0.9 milliradians for the primary u^ component (before contraction
at the 0.7x telescope). The spatial resolution although certainly not poor could
be improved by using larger intersection angles and smaller focal waists. The
latter can be achieved by either beam expansion prior to focussing, use of a smaller
f number crossing lens or both.
At larger angles and for the longer focal lengths required in practical appli-
cations, a single conventional focussing lens generally would be quite heavy and
expensive. One can then first cross the beams and use individal lenses on each
leg. An alternative would be to investigate the use of large, corrected, acrylic
Fresnel lenses.
An interesting aspect of BOXCARS that has not yet been explored concerns the
employment of crossed polarizations (Ref. k6). In isotropic media such as gases,
the third order nonlinear resonant susceptibility can be expressed as (Ref. ^7)
"ijki <-V V V -fe) • W (l-2')6ij61a + p(6ikV + 6iiV] (28)
where the indices i, j, k, 1 refer to the polarization orientation of the frequencies
BU, UL, u) and w . p is the depolarization of the Raman mode and 6^, the Kroneker
delta (-1, i=j; =o, i/j). In Ng, p is extremely small and equal to 0.022 (Ref. 1$).
Hence
X - X 66., (29)
ijkl xxxx ij kl
With conventional CARS, where only one u^ input pump beam is employed, J = k,
and hence, for CARS to be observed, the Stokes polarization must be parallel to the
polarization, k = 1. The CARS radiation will then exhibit the same polarization,
1 = J, as the pump and Stokes beams and is generated via - Wlth BOXCARS,
however, one has independent control of the j and k polarizations. If j and k are
79
-------
parallel, then the Stokes beam must have the same polarization for CARS to be
generated, as in conventional CARS. However if j and k are orthogonally polarized,
and the Stokes beam is aligned along one of the pump components, then the CARS
radiation will emerge orthogonally polarized to the Stokes beam.
This gives rise to an interesting instrumental approach to thermometry. One
way to measure temperature (Appendix l) is to ratio the CARS from a hot band, e.g.,
v = 1, to CARS from the ground state band, v = 0. If the Stokes beam contains two
orthogonally polarized components corresponding respectively to wavelengths neces-
sary to generate CARS from the hot and ground state band, then the CARS so generated
from each band would be orthogonally polarized. The two CARS components could be
separated with a polarization splitter and temperature ascertained from the ratio
of the two. In contrast to the broadband techniques to be subsequently described,
such an approach offers backend (detection) simplicity. It obviously would be more
complicated on the front end (laser) side than broadband CARS approach. Hence, in
addition to obvious spatial resolution advantages, BOXCARS may offer other instru-
mental benefits as well.
Flat Flame Thermometry
Unlike spontaneous Raman spectra which depend linearly on and, hence, mirror
quantum state population distributions, CARS spectra are much more complex represen-
tative of the nonlinear nature of the three wave mixing process. Although some
simple schemes were first proposed to derive temperature information from CARS
spectra, there seems to be widespread consensus that computer modelling will be
required for accurate temperature measurements. In fact as confidence develops in
the analytical modelling procedures, simplified data reduction schemes will most
likely emerge. A computer model is an important adjunct to experimental studies of
CARS spectra. Initially, the model predictions guide the experimenter in judging
the validity of the experimental approach. This is particularly true of CARS research
in flames where signal contributions may emanate from other than the desired region
under study. As confidence grows in the experimental technique and the results
80
-------
therefore, the model can be revised and updated. Ultimately, if necessary, the
model will provide a data redaction capability, preferably in real time. During
the Task II experimental program, the close interaction between the experimental
results and the model predictions was quite evident.
Computer Modelling of CARS Spectra
Under Corporate funding, a computer model to synthesize CARS spectra as a
function of density, temperature and background was formulated by E. J. Hall of
UTRC. The model is currently capable of generating R2 and CO spectra and is des-
cribed in detail in Appendix I. A block diagram of the model is shown in Fig. 30.
Briefly, upon inputting selection of the molecular species of interest together with
density and temperature, the model calculates the state populations ri j and energies.
hoi j, where h is Planck's constant and v, J the vibrational and rotational quantum
numbers for the state of interest. Next the program computes the complex third
order nonlinear susceptibility. Recall that the CARS radiation is proportional to
the square of the absolute value of the susceptibility. The third order nonlinear
susceptibility may be written as
x(3) = z fc' + ix"^ + xnr (30)
0
where (x ' + ix"). is the resonant susceptibility associated with transition j and
J
Xnr is the nonresonant susceptibility contribution of the electrons and remote
resonances. The resonant contribution may be expressed as
where the detuning frequency AU>. = u>. - (OL-UV,) has been introduced. UK is the
-------
COMPUTER GENERATED CARS SPECTRA
MOLECULAR CONSTANTS,.
TEMPERATURE, DENSITY
OD
BOLTZMANN
DISTRIBUTION
CALCULATION
nv,J
1
wv.J
THIRD ORDER
NONLINEAR
SUSCEPTIBILITY
CALCULATION
LINE SHAPES
g-lUo.,), g2(co2)
1
X(3) (co,-w2)
1C? II
X1 (EVALUATION,
CONVOLUTION OVER
91-92
,(3)
Q,0,S
SLITWIDTH
CONVOLUTION OVER
SPECTROMETER
SLIT FUNCTION
oo
o
10
Tl
P
co
o
-------
A
nAJ gj j r. (32)
J
where h is Planck's constant divided by 2n; n, the total species number density;
A., the normalized population difference between the levels involved in the transi-
tion; g.j, linestrength factor equal to (v.j+1); and (Ba/Bfi)^ the Raman cross section
for the transition. The susceptibility is calculated by summing over all appro-
priately allowed transitions and selecting a value for the nonresonant susceptibi-
lity. The susceptibility emerges from this subroutine as a function of uu -u> . Next
the square of the absolute value of the susceptibility is calculated together with
a convolution over the function g^) g((«2), where gfu^) is the laser lineshape at
frequency tu±. This accounts for the scaling of the CARS radiation as I-,2! where
Ii is the laser intensity at frequency cu. . In the program, Gaussian fits to the
laser lineshapes are used. The u^ halfwidth is 1.2 cm"1 and u> linewidth appro-
priate to the experimental value, typically 150 to 300 cm" . Finally the CARS
spectra is convoluted over a monochromator slit function, generally taken to be
triangular. At the display terminal, plots are available of the susceptibility,
and the CARS spectrum before and after the monochromator slit convolution.
Scanned Flame Spectra
CARS spectra from flame N2 have been studied on premixed flat flames sustained
on a 7.5 cm dia burner consisting of either a water cooled bundle of hypotubes or
a sintered porous plug. Early spectra (Ref. ^9) were obtained using collinear phase
matching and Ar shield gas between the focussing lens and the burner, and between
the burner and the recollimating lens. In this manner, room air W2 contributions
were suppressed. Improved quality spectra ^re subsequently obtained using BOXCARS
at small crossing angles. Such a spectrum is displayed in Fig. 31. The experi-
mental spectrum was obtained using 100^ entrance and exit slits on the monochromator,
corresponding to a monochromator resolution of about 1 cm" . The output of the
monochromator photomultiplier is fed to a two channel boxcar averager (PAR Model
162) fitted with gated integrators (PAR Model 1#0. For broadband CARS spectra
83
-------
EXPERIMENTAL BOXCARS SPECTRUM OF FLAME IM2
TEST ~ 1700°K
1 CIVH RESOLUTION
CX>
oo
o
to
CD
CC
t/
LU
I-
21080
21090
21100 21110
WAVENUMBER-CM-1
21120
21130
Tl
P
CO
-------
from laminar flames, the monochromator signal can be normalized to the total inte-
grated CARS signal. The latter is obtained by splitting off a small fraction of
the CARS signal which is monitored by a second photomultiplier fitted with a narrow-
band interference filter to ensure that only CARS is collected. For these flames
at low air and CH^ flowrates, radiation corrected thermocouple measurements (Ref.
50) indicated a temperature just below 1700°K. Earlier, spontaneous Raman measure-
ments (Ref. 50) gave temperatures between 1625 and l675°K.
In Fig. 32 is the computed CARS spectrum which produced the best fit to the
experimental spectrum. The computation was performed at l650°K and 1.25 cm'1 slit
width. As seen, excellent agreement is obtained except for some features in the
hot band. In the ground state vibrational band (v=0) the low lying Q, branch tran-
sitions are unresolved and pile up to form a broadened band. At higher rotational
quantum numbers, due to the rotational-vibrational interaction, the even Q branch
peaks are clearly seen. The odd rotational quantum number transitions, which have
a nuclear spin weighting equal to half of the even numbered Q branch transitions,
do not stand out. The first hot band is also clearly visible. Initially in com-
paring the experimental and analytical spectra it was noticed that the various even
4 branch peaks did not coincide. However, when the rotational constant was updated
from the value in Herzberg's treatise, coincidence was obtained. The experimental
peaks in the hot band are reproducible and are believed to be valid. Depending on
the exact value of the vibrational frequency selected, these peaks can be duplicated
in the model predictions. The analytical curve shown corresponds to the set of
molecular constants given in Ref. 51. As more refined values of the molecular con-
stants become available, agreement of the analytical and experimental spectra can
be anticipated to improve. In the meantime, empirical procedures can be adopted
if desired. Also uncertain in the model are the Raman linewidths. As discussed
in Appendix I, linewidth variations have been studied. Good agreement is obtained
assuming constant values for the linewidths throughout and the model sensitivity is
«* very acute to the selection. Better fine structure agreement may also require
accurate knowledge of the Raman linewidths and their scaling behavior. For
85
-------
COMPUTED CARS SPECTRUM FOR FLAME N2
T = 1650°K 1.25 CM~1 RESOLUTION
1.0
CD
Q.8
LLJ
I-
0.0
21080 31090 21180 21110 21120 21130 21140
o
WAVENUMBER -
P
co
(O
-------
the most part, modelling is fairly well in hand and the use of CARS for flame
thermometry should pose no unusual problems.
Single Pulse Thermometry
As mentioned earlier, for turbulent flame work, the CARS spectrum must be
captured in a single pulse due to the highly nonlinear dependence of CARS on tem-
perature and density. Scanning the spectrum over a period of time would result in
an accumulation of terms involving self- and cross-correlations between the density
and temperature fluctuations. This obviously would render the spectrum ambiguous
from a data reduction standpoint. Furthermore, single shot thermometry leads to
a determination of the temperature probability distribution function from which
the magnitude of turbulent temperature fluctuations can be obtained.
Single pulse, CARS spectra of flame N2 were obtained using an optical multi-
channel analyzer (OMA, PARC) fitted to a 1/U m monochromator containing a 2^00 gr/mm
holographic grating. A single CARS pulse was captured using a photographic shutter.
Since the laser operated at 10 pps, setting the shutter at a 90 millisecond duration
guarantees capturing only one pulse and with high probability. In Fig. 33 is shown
the single pulse CARS spectrum of flame I\L from the previously described flat flame.
Collinear phase-matching was employed since these particular tests preceded the
demonstration of crossed-beam phase-matching. Each point on the spectrum corresponds
to 0.39$ (l.7l| cm'1). Also shown as the thin solid line is the computed CARS
spectrum at 1700°K but now convolved over a 2.7 cm"1 slit function corresponding
to the combined monochromator-OMA resolution. This was determined by examining the
spectrum of a thin atomic line. As seen the agreement between the OMA trace and
the model prediction is quite good. The data presented in Fig. 33 thus demonstrate
the feasibility of single pulse CARS thermometry.
In subsequent experiments with the OMA it was found that the OMA vidicon tube
was extremely vibration sensitive. In the single pulse thermometry experiments, the
mechanical optical shutter was mounted directly in front of the entrance slit, and
attached to the monochromator. The slight undulation on the OMA trace apparent in
87
-------
FIG. 33
SINGLE PULSE CARS SPECTRUM OF FLAME N2
0.4APER CHANNEL
88
-------
the hot band and baseline is probably due to shutter induced vibrations. Vidicon
tubes, due to this vibration sensitivity, are probably unsuited for operation in
high vibration environments such as combustor test cells. One possible solution
which exploits the coherence of the CARS signal is to locate the OMA remotely from
the test cell itself, possibly in the control room and transmit the CARS signal to
the detector. For thermometry only the relative shape of the spectrum is important
and small misalignment jitter may be tolerable. An alternate course is to choose a
solid state diode array (e.g. reticon) for multichannel detection.
At the lower spectral resolution currently encountered with state-of-the-art
multichannel detectors, the CARS spectra still exhibit fine temperature sensitivity
as shown in the model calculations presented in Fig. 3*1. As temperature increases,
the ground state band broadens substantially and the hot band increases as shown.
Temperature may be obtained by detailed curve fitting, by the breadth and slope of
the ground state band possibly, or by the ratio of the hot to ground state peaks.
In fact, qualitatively, low resolution CARS spectra resemble spontaneous Raman
spectra and, hence, should be amenable to the same mechanics used to reduce spon-
taneous Raman data. The low resolution spectra are believed to display enough tem-
perature sensitivity to perform temperature measurements with a resolution of at
least 50°K.
Sooting Flame Temperature Measurements
Based upon the Task I review (Ref. l) together with experience gathered while
attempting spontaneous Raman measurements in simulated combustors (Ref. 52), CARS
is of great interest because of its potential for operation in the highly particle
laden environments typical of practical devices. To investigate this potential CARS
generation was studied in a laminar, sooting, propane diffusion flame. Such a flame
was studied extensively earlier in conjunction with studies of laser modulated par-
ticulate incandescence interferences in Raman scattering diagnostics (Ref. te).
The flame is generally sustained on a stainless steel tube, 0.6 cm i.d., and esta-
blished by merely flowing propane through the tube. The flame resembles that from
89
-------
FIG. 34
LOW RESOLUTION CARS SPECTRA
SLIT WIDTH 2.7 CM
-1
to
LU
I-
2
CC
CJ
Q
QC
o
1.0
0.8
0.6
0.4
0.2
21080
21100
21120
21140
WAVENUMBER - CM'
77-11-92-1
90
-------
a candle and is highly sooting. Previous Mie scattering measurements (Ref. 50)
indicated a soot number density on the order of 1010 cm"3 with an average particle
diameter of lKX$. This is a very high soot level producing an attenuation in trans-
mitted light of tenths of percent per mm pathlength. The requisite spatial resolution
was achieved via the previously described BOXCARS approach using an 89 mm dia,
305 mm focal length lens. The phase-matching angles of o> = 5.7° and 9 = 6.5° gave
the resolution shown in Fig. 29, approximately O.h mm in diameter by 1 mm long.
With a two dye component mixture in the dye laser, which produces very broad
bandwidths (~ 270 cm" ), the BOXCARS signals were weak necessitating an increase
in the monochromator slit width to UoO|j, , or a resolution of k cm"1. With k cm"1
resolution, the calculated CARS spectra are not too dissimilar qualitatively from
the 2.7 cm" resolution curves shown in Fig. 3k. The broadened ground vibrational
state and first hot band are readily apparent and the resolution is sufficient for
temperature determinations.
In Fig. 35 is shown one of the first BOXCARS spectra from the highly sooting,
laminar propane diffusion flame. As is apparent this spectrum bears no resemblance
to the hot N2 spectra calculated in Fig. 3^-- The strong spectral feature seen
arises from C? which has a major Swan band transition at U737A midway between the
N2 bands at ^733$ (v=0) and ^7^0$ (v=l). Swan bands are known to be degraded to
the violet (higher wavenumbers) in contrast to the W2 CARS spectrum. The signal in
Fig. 35 consists of both incoherent and, unfortunately, coherent features. The
incoherent contribution is ascertained by blocking one of the o^ or ^ beams, or
rotating the optical flat in the i»2 Stokes beam to phase mismatch the three wave
mixing process. Incoherent interferences are generally suppressible in coherent
spectroscopy. Coherent features are obviously more problematical. Here the
*It was subsequently discovered that the photomultiplier tube used for the sooting
flame measurements was fairly noisy which restricted the tube voltage and, hence,
the tube gain, at which good results could be obtained. With the better tube now
in use, smaller slit widths probably could have been used.
-------
FIG, 35
BOXCARS SPECTRUM FROM LAMINAR, C3H8/DIFFUSION FLAME
21,110
21,120
FREQUENCY-CM"1
21,130
21,140
92
77-12-64-1
-------
coherent feature arises from three wave mixing in laser produced CL. At the high
laser intensities used to produce CARS, significant soot vaporization will occur
(Ref. te), even on a nanosecond time scale. A major product of vaporizing soot is
C2. CARS can be electronically resonantly enhanced when either the CARS radiation
itself or the pump component resides near an electronic transition (Ref. 53). The
CARS generation in the laser produced C2 vapor is resonantly enhanced due to the
coincidence of the CARS and C2 electronic transition frequencies. In Fig. 35, the
coherent contribution was approximately 25 percent of the total interference and
roughly comparable to the CARS N2 ground state peak. The shoulder in the slope of
the Swan spectrum arises from the CARS from the N2 ground state. Although this
result may at first seem discouraging, it is important to bear in mind that the
flame was highly sooting and the signal/coherent interference is of order unity.
In fact this situation was readily improved. How the improvement came about will
be more understandable if reference is made to studies performed in regard to the
scaling of the incoherent interferences.
An optical collection system was set up at right angles to the central BOXCARS
axis and laser induced emissions examined through blocking and narrowband inter-
ference filters. In Fig. 36, the temporal variation of the emissions is shown
together with the laser pulse signature for reference. The laser pulse is approxi-
mately 10 nsec FWHH in reality, but stretched somewhat in the 3100QA photomultiplier
(RCA) used here due to its 3 nsec rise and fall time. As seen the emissions at
^733, 5050, and 686^ exhibit risetimes comparable to the laser pulse. This is due
to the fact that there are-no effective soot heat transfer processes other than
vaporization (Ref. 1*2) resulting in flashed vaporization of the soot surface. The
emissions at 5050 and 686$ are laser modulated soot incandescences while that at
^338 is incandescence and primarily, Swan emission at k737$ within the filter band-
Pass. The 6861*8 trace displays a long tail probably due to the shifting graybody
radiation distribution as the particles cool, or possibly, but less likely, some
fluorescence contribution.
93
-------
LASER INDUCED SOOT EMISSIONS
LASER
PULSE
4733 A,
10 A FWHH
5050 A,
50 A FWHH
6864 A,
10 A FWHH
INTEIMSITY
(ARBITRARY UNITS)
TIME-
20 NANOSECONDS/DIVISION
IO
CO
0>
-------
Quantitatively the emissions are quite interesting. By comparing those at
117338 and 5050$, the Swan emissions are found to be about an order of magnitude
larger than the incandescence contribution. Furthermore, the two emissions display
substantially different response to the exciting laser wavelength. For example,
the two high intensity u^ (5320$) beams produce slightly more incandescence than
does the lower intensity oo2 beam (6073^). This is expected due to the saturation
behavior of the incandescence described in Ref. !|2. The Swan emissions at k737%
are much more pronounced with the 6073A excitation, however, than with the 5320$,
probably due to absorption of the broadband 6073$ by three Swan bands residing
within the dye laser profile. The Stokes dye laser width of 270 cm centered at
16^66 cm encompassed three strong Swan transitions at 16,333, 16,502, and 16,653
cm" . Similar results were found for the incoherent interference component when
recording the BOXCARS spectrum. Whether the enhanced Swan emission at ^733A by
the Stokes beam (6073^0 is via a direct vapor excitation process or vapor phase
absorption leading to increased heating of the laser irradiated soot particle has
not been determined.
The result did suggest decreasing the dye laser spectral bandwidth to decrease
the C2 Swan absorptions and to increase the CARS signal strength. Returning to
single dye component mixtures decreased the Stokes laser bandwidth by about two to
150 cm"1 (still quite sufficient for single pulse multiplex CARS) and quite signi-
ficantly resulted in about a factor of two increase in dye laser energy as well.
Furthermore, since the CARS signal from N2 using horizontally polarised laser
sources is horizontally polarized, Eq. (29), an additional factor of two discrimination
against incoherent interferences is obtained by insertion of a polarization filter
in the CARS detection system. In so doing, the spectrum shown in Fig. 37 was ob-
tained. Also shown for comparison is the computed CARS spectrum at 2300°K for a
^ cm'1 slit width resolution and 0.1 cm"1 assumed Raman linewidths. As seen the
computer model prediction fits the experimental spectrum fairly well. The adiabatic
flame temperature for propane-air is 2250°K (Ref. 5*0- The rough character of the
95
-------
BOXCARS SPECTRUM OF N2 IN A HIGHLY SOOTING LAMINAR PROPANE DIFFUSION FLAME
VQ
LOCATION OF C2 SWAN PEAK
DEGRADES
21070
21080
21090
21100
21110
FREQUENCY-CM"1
21120
21130
P
CO
-------
experimental spectral trace reflects the compromise made in the number of pulses
averaged over, the desire to record the spectrum in a reasonable period of time
(10 cm'Vmin) and use of a very short output time constant on the averager. Also
shown is the location of the beginning of the Swan band head at 21,110 cm'1 which
peaks at 21,115 (Fig. 35) and degrades to higher wavenumbers. A small amount of
electronically resonantly-enhanced CARS from laser produced C2 is probably present
but its contribution is quite small. This result is very encouraging since the
soot densities in the laminar propane flame are very high which bodes well for the
practical utility of CARS diagnostics.
CO Species Concentration Measurements
Species detection sensitivity is limited for the conventional CARS approach,
not due to an inadequacy of signal as is often the case in spontaneous Raman scat-
tering, but due to the presence of the background nonresonant susceptibility. The
nonresonant susceptibility is essentially an electronic contribution from all of
the molecules present. Species sensitivity limitation is best illustrated by
considering the case of a single resonance in the presence of background. The
susceptibility may be expressed as before
X = X1 + ix" * X (33)
where x', x" and xnr are respectively the resonant and nonresonant components of
the susceptibility. The square of the absolute value of the susceptibility is
or in the case of a very weak resonance
|x|2~xnr2+2X'xnr (35)
Since x' varies from negative to positive about line center, the presence of
* species in low concentrations is apparent by a similar modulation in the CARS
background susceptibility profile. In broadband CARS, the background nonresonant
97
-------
susceptibility profile merely mirrors the Stokes laser profile. In narrowband,
scanned CARS, it reflects the tuning power variation of the Stokes laser. When
the species concentration becomes very low, i.e., X1 «X , the "signal" is lost
in the background susceptibility profile.
Model C aIc olatIons
The foregoing situation is quite evident in the detailed computer calculations
of flame CO, some of which are summarized in Fig. 38. There, computer calculations
are displayed as a function of CO concentration in the post flame region of an
atmospheric pressure C%-air flame at 1700 K. The background susceptibility was
calculated from the values at 69^3^ in Ref. 55 assuming a postflame composition of
70% U0, 2CP/. 11^0 and 1.0% COo. Any variation in the nonresonant susceptibility with
wavelength was not accounted for. In the absence of data on the nonresonant
susceptibility of IL^O vapor it was assumed to be that of methane. Since lipO is a
major constituent resulting from hydrocarbon combustion, knowledge of its non-
resonant susceptibility would be quite valuable. Changes in the composition of
flame products which result when varying amounts of CO are produced are not accounted
for in the model. The spectra in Fig. 38 have not been convoluted with a spectrom-
eter slit function, but do reflect the limited resolution attainable with a 1.2 cm
pump laser line. Also note that the ordinate scale is not the same from plot to
plot. In all cases shown, the destructive Interference dip caused by the resonant
portion of the susceptibility going negative Is apparent, but clearly less so as
the concentration decreases. At the larger concentrations shown, the CO spectrum
clearly rises out of the nonresonant background and displays a pronounced hot band.
The computer calculations display vividly the species sensitivity limitations of
CARS. At the 0.5 percent CO level, the destructive interference causes the spectrum
to dip by only about 12 percent and at 0.1 percent CO only by 2 percent. For
scanned, laminar flame studies this places very stringent requirements on the pulse-
to-pulse stability of the CARS signal. For single pulse work, it places high reso-
lution requirements on the detection system. For the nonresonant susceptibility
-------
vo
MD
0 56-
0 45
T
DENSY
CHl«< =.
0
SEND
0.1%
E-IS
COMPUTED FLAME CO SPECTRA
T = 170O°K
28686
29908
28920
2094M
0.5%
23888
03
I
O
00
oo
-------
selected for these calculations, it would be difficult to measure GO levels below
a few tenths of a percent. If the nonresonant susceptibility were larger, the
detection sensitivity would be even less. The calculated spectra are quite sensi-
tive to the nonresonant susceptibility value selected. As an example, the 4 percent
CO spectrum shown in Fig. 38 but convolved over a 2 cm" slit function was compared
with a similar U percent CO spectrum but with double the nonresonant susceptibility
used for the Fig. 38 calculations. For double the nonresonant susceptibility, the
destructive interference dip from the nonresonant profile decreased to a 33 percent
drop compared with 55 percent previously, the ground state band peak to destructive
dip ratio decreased from k to 2, and the hot band peak to ground state peak height
ratio increased to 0.86 from 0.79-
Experimental Approach
Unlike the thermometry studies of flame Wgj one must be quite careful experi-
mentally with species concentration measurements when using collinear phase-matching.
For thermometry, one need only eliminate the potential resonant contributions from
cool N2 adjacent to the flame zone. Argon can be conveniently used as a purge gas.
In attempting CO measurements, wherein the CO is modulating a non-resonant back-
ground, purge gas selection is quite important. Argon, for example, has a slightly
larger nonresonant susceptibility than Np (Ref. 55 )> while Helium with a suscepti-
bility more than an order of magnitude lower appears to be the best choice.
With a short focal length lens, f = 20 cm, the cool regions adjacent to the
7.5 cm dia burner apparently did not pose a problem. For example, spectra showed
virtually no variation between the situation when room air was adjacent to the
burner regions through which CJD-, and ua passed, and when Helium shields were used.
The problem encountered with collinear phase-matching, however, was apparently a
lack of CO concentration profile uniformity across the "flat" flame. To produce
large levels of CO, the flat flame porous plug burner was operated fuel rich, either
with excess C% or CO. At these high equivalence ratios, the flat flame exhibited
some diffusion flame behavior, i.e., a blue conical outer secondary flame zone was
clearly evident. Based on subsequent experience with improved spatial resolution
100
-------
approaches, It appears that the collinear phase-matched CO spectra were spatially
averaged. This appears to be the case even when measurements are made 7 mm above
the burner surface. Initial dissatisfaction with collinear phase-matching arose
due to lack of agreement between the experimental and computed spectra which led to
a suspicion that the results were spatially averaged. Subsequent experience indi-
cates that: (l) the nonresonant susceptibility appears to be quite sensitive to
flame conditions and is probably difficult to calculate a priori and (2) the model
calculations are fairly sensitive to the magnitude of the nonresonant susceptibility.
BOXCARS was attempted in an effort to improve the spatial resolution. Even
at small crossing angles, however, the resolution was apparently too fine and the
CARS signals from the flame were very weak. Eecall that the CARS radiation strength
scales as the square of the interaction length. It should be pointed out the weak
CARS signals encountered at high resolution are due in part to the large pump laser
bandwidth, 1.2 cm"1, currently existing in the experimental setup. If the laser could
be spectrally condensed with no loss of energy to 0.4 cm or 0.1 cm" , the CARS
signal would increase by one and two orders of magnitude respectively. The signals
would also increase inversely with reductions in the Stokes laser bandwidth. For
broadband CARS, It is desirable to keep the Stokes laser bandwidth as small as
possible, but yet broad enough to avoid .severe distortion of the CARS spectrum.
Based upon experimentation with low angle BOXCARS, it was discovered that
CARS generation between the adjacent u^ and tu components (Fig. 27), although not
perfectly phase matched, could be substantial. Upon attempting crossed-beam,
phase-mismatched CARS, the'nonresonant CARS signal from the flame was found to be
within the detection sensitivity of the apparatus. Consequently, the CO studies
reported here employed this crossed-beam, phase-mismatched approach. A 305 mm
focal length lens was used and the pump and Stokes lasers were crossed at an angle
of 1.8°. At this angular separation, the beams intersected only over the burner
and near its centerline. Hence no contributions from cool regions outside the burner
were possible. The spatial resolution was checked in the usual way by monitoring
the CARS signal generated from a translatable microscope slide and is shown in
101
-------
Fig. 39' As seen, all of the signal is generated within a 2.5 cm extent but the
CARS originates predominantly within ±0.5 cm of the crossing point. .
An interesting aspect of crossed-beam, phase-mismatched GARS is the direction
of the emergent CARS signal radiation. One possibility is for the CARS beam to
emanate in the direction suggested by Fig. 25a. This was found not to be the case.
Rather the CARS emerged between the crossed Stokes and pump beams indicating that
the CARS generation occurs along the nominal phase-matching direction in the medium,
despite the fact that oo and u) are angled.
To measure the amount of CO produced in the flame, an uncooled quartz micro-
probe was used to sample the centerline CO concentration the level of which was
indicated by a Beckman Model 86U ndir analyzer. Sampling with uncooled probes in
high pressure flames is subject to some uncertainty, and the microprobe results
may only be roughly indicative of the actual CO levels (Ref. 56).
Measurements
In Fig. hO is shown the CARS spectrum recorded for an ndir reading of 3-9
percent CO in a CO-CH^-air "flat" flame. The spectrum was obtained 7 nun above
the burner surface . Due to various fluctuations in the laser, flame, etc.,
even the averaged CARS signal is subject to fluctuations on the order of about
± 10 percent. In the spectrum shown, lOOy, slits corresponding to a nominal 1 cm"
resolution were employed. Also shown as the series of points is the k percent CO
model calculation in Fig. 38 convolved over a 1 cm" slit function. As seen the
agreement between the experimental spectrum and the model calculation is quite good.
The quality of the spectra degraded rapidly, however, at lower CO concen-
trations indicated by the quartz microprobe sampling system. For given CH. -air
settings, the CO spectra obtained at lower CO flame levels as the CO doping level
was reduced did not behave as those shown in Fig. 38- Modulation of the CARS
spectrum was apparent at the 2 percent level and to the trained eye at 1 percent
in the presence of the random spectral fuctuations. The spectra had the qualitative
appearance of CO levels within a higher nonresonant susceptibility. It is not
102
-------
FIG. 39
CROSSED-BEAM, PHASE MISMATCHED CARS SPATIAL RESOLUTION
1.8° CROSSING ANGLE, 305MM FOCAL LENGTH LENS
FOCUSING LENS
»- RECOLLIMATING LENS
-1 0 +1
DISTANCE FROM FOCUS-CM
103
78-02-195-3
-------
CROSSED-BEAM, PHASE MISMATCHED, CARS SPECTRUM OF FLAME CO
• COMPUTER MODEL
T = 1 700°K
Xnf =9,2 (10-19)CM3/ERG
CO = 4%
MEASURED CO (NDIRl 3.9%
O
-p-
20,980
20,900 20,920
WAVENUMBER - CM"1
20,940
20,960
P
4^
O
-------
beyond the realjn of possibility, however (Ref. 56), that the CO in the flame is
actually lower than that indicated by the sampling system due to C02 reduction in
the high pressure (~ 680 torr), hot quartz microprobe and sampling lines. CO
measurements with a water-cooled quartz microprobe indicated a CO level comparable
to the uncooled probe in CH^-air mixtures, however.
CARS spectral differences were noted at the nominal 1+ percent level depending
upon how the CO was produced. In Fig. 1+1 is shown the CARS CO spectrum obtained
in a rich CH^-air flame with no added CO. Assuming for the moment that the micro-
probe sample was correct, this spectrum differs markedly from that shown in Fig. Uo
in which large amounts of CO were added to the flame. One explanation for the
Fig. hi spectrum is that the nonresonant susceptibility is larger than before. CH^
has a nonresonant susceptibility over twice that of N2 and to produce the 1+ percent
CO level, over twice the CH^ flow relative to that required to sustain the flame
was employed. It is also possible, however, that under these conditions the CO
level in the flame was lower than that indicated by the ndir analyzer as previously
mentioned. Incidentally, the large peak in the center of the ground vibrational
state band is real and reproducible and does not result from CO. Its assignment is
unknown, perhaps originating from a hydrocarbon fragment.
CARS measurements of CO species concentration in the presence of an uncertain
nonresonant susceptibility requires a three parameter model fit involving the
concentration, temperature and nonresonant susceptibility. An important question
yet to be thoroughly studied concerns the uniqueness of a three parameter fit to
species concentration data. Clearly species concentration measurements would not
be possible in an unknown, nonresonant background'if uniqueness does not exist.
Experimentally, measurements were possible only to the 1-2 percent level at best.
Analytically species concentration measurements would be practically possible to
the 0.5 percent level or so (Fig. 38) for the nonresonant susceptibility assumed.
Clearly CARS approaches which lead to a cancellation of the background nonresonant
susceptibility, some of which are reviewed in Ref. 1, would seem desirable for
species concentration measurements.
Conclusions based upon all of the foregoing CARS investigations are summarized
"i Section III which follows, together with recommendations for future developmental
efforts.
105
-------
CARS SPECTRUM OF CO IN CH4 RICH FLAME
MEASURED CO (NDIR)~4%
O
ON
oo
I
o
to
I
20,880
20,900
20,920
WAVENUMBER -CM"1
20,940
20,960
-------
SECTION III
CONCLUSIONS AND RECOMMENDATIONS
Saturated Laser Fluorescence Investigations
The Task II experimental investigations of laser fluorescence have shown that
it is possible to saturate laser excited molecular fluorescence from the flame
radicals CH and CN. Accordingly, this permits determination of these radicals with-
out the inherent imprecision arising from uncertain analytical quenching corrections
in unsaturated fluorescence. The laser spectral intensities required to achieve
saturation are readily attainable and in the 10^ to 10° Watts/cm cm range for
CH and CN. Fluorescence signals were obtained in very hot flames from concentrations,
as determined by absorption, of ~ 60 ppm for CH and 150 ppm for CN, thus confirming
the sensitivity of the saturated fluorescence technique. Based upon the signal
levels experienced, saturated fluorescence should be capable of detecting minority
flame species at the ppm concentration level. Combining the saturated fluorescence
investigations with absorption serves to assess rigorously the validity of the
saturated fluorescence approach. The saturated fluorescence results were lower by a
factor of two for CH and four for CN than the species concentrations determined by
absorption. This discrepancy is believed due to the possibility that the region
of CH and CN production was smaller than the fluorescence sample volume. In effect
the sample volume is overestimated, leading to an underestimate of the species con-
centration required to produce a certain fluorescence intensity level. Furthermore,
the results of this work indicate that the two-level saturated fluorescence theory
appears to be applicable to molecules, such as CN, which, in reality, are considerably
more complex. CN, for example, approximates a three level system from an electronic
energy level viewpoint and is even more complex when the various vibrational and
rotational level are accounted for. This is consistent with findings of Baronavski
and McDonald who successfully applied the two level model to Cg which is also con-
siderably more complex than a two-level system.
107
-------
It would be desirable to repeat the CH and CN experiments reported here with
an improved experimental approach. In this way, the optimistic conclusions reached
in these first experiments concerning the viability of saturated laser fluorescence
detection could be more definitively checked. Key to an improved experiment is an
improved beam quality, narrower linewidth, repetitively pulsed dye laser. Improved
beam quality would permit greater experimental versatility and permit much smaller
focal diameters to be achieved. This would eliminate the previously supposed
problem of a measurement volume larger than the radical, production region. Narrower
linewidth than that employed here would permit higher spectral intensities
to be attained and allow more selective excitation of the fluorescence spectrum.
Repetitive pulsing at 10 to 20 pps would permit the employment of boxcar averaging
techniques for greatly improved signal to noise ratio and permit the fluorescence
spectrum to be continuously scanned. Laser pumped dye lasers could provide the
improved beam quality, narrower linewidth, and repetition rate required to upgrade
the fluorescence experiments. A suitable drive laser would be a pulsed neodymium
laser system similar to the CARS driver operating at either the second or third
harmonic. For CH and CN the third harmonic at 3533^ would be used to pump suitable
dyes directly to probe these radicals at ^315 and 3883$, respectively. For NO, the
third harmonic would pump a tunable dye laser which in turn would be frequency-
doubled to probe the NO Y bands at 2265$. For OH at 306)4% and NH at 3350$, the
second harmonic at 5320A would pump a tunable dye laser which would be frequency
doubled to the appropriate absorptions. For C? no frequency doubling of the second
harmonic pumped dye laser is necessary. The 10 nanosecond pulse length of the Q-
switched neodymium laser ensures high intensities which result in efficient dye
laser and various harmonic generation efficiencies. The 10-20 pps repetition rate
of such a laser permits the use of commercially available boxcar averagers. Laser
pumped dye lasers can also be spectrally condensed much more efficiently and to
fairly narrow linewidths, ~ O.lA. Quite interesting, with such a system, it would
be possible to perform both the absorption and the fluorescence measurements simul-
108
-------
taneously using the spectrally narrow dye laser. By using the same laser source
for absorption and fluorescence, one is assured that the interaction region is the
same for both measurements. To reduce the concentration gradients in the flame,
it would be desirable to perform the initial experiments on a low pressure flat
flame, multipass if necessary.
Subsequent to the investigations of CH and CN, it is recommended that the
applicability of saturated fluorescence to other molecules be explored. Due to its
extreme importance, WO would be the most logical candidate molecule to investigate.
This is quite convenient experimentally, in that the neodymium third harmonic
pumped dye laser used for the CH work would be tuned to a slightly higher wavelength
and then frequency doubled. The high sensitivity, in-situ, spatially precise
detection of WO via saturated laser fluorescence is a very exciting possibility and
should be explored in the near future. Due to its importance in combustion, inves-
tigations of OH hydroxyl radical detection would also be desirable.
CARS Investigations in Flames
Coherent anti-Stokes Raman spectroscopy is capable of providing both tempera-
ture and species concentration measurements in flames and the Task II experimental
program in this area mirrored this duality. Although thermometry was accorded the
major emphasis, much of the experience gained during the CARS investigations of
flame N2 is applicable to species detection applications, e.g. spatial resolution
enhancement. During the Task II investigations, a crossed-beam phase-matching
technique, termed BOXCARS, was devised which greatlv improves upon and eliminates
any ambiguity in regard to the spatial resolution of CARS. Conventional, collinear
Phase-matching can possess poor and ambiguous resolution particularly in the presence
of density gradients. This was a major problem area for the diagnostic application
of CARS which has now been overcome. The moderate (~ 1 cm"1) resolution spectral
scans of CARS from flame N2 show excellent agreement with the predictions of the
computer code and demonstrate the suitability of CARS for thermometry. Single
, lower resolution CARS spectra from flame W2 obtained with an optical multi-
109
-------
channel analyzer (OMA) also show good agreement with the code predictions and
demonstrate the feasibility of single pulse thermometry. Such a capability will
permit temperature probability distribution functions to be measured in turbulent
combustion processes from which the magnitude of turbulent temperature fluctuations
can be ascertained. Although coherent spectral interferences have been encountered
in thermometry investigations in highly sooting flames, these seem to be well under-
stood, not too serious and suppressible. With appropriate Stokes laser bandwidth
selection and detection technique, CARS spectra of N^ in highly sooting propane dif-
fusion flames have been obtained permitting temperature determinations in such
flames. This is believed to be the first temperature measurement in such flames
with a remote, spatially precise technique. CARS is of intense interest because of
its very promising potential for probing practical combustion environments. Based
upon the Task II experimental investigations, which have certainly demonstrated
this potential in laboratory studies, it is recommended that CARS thermometry now
be attempted in a practical combustion facility, e.g., research scale furnace. Such
attempts would serve to uncover new and unanticipated difficulties, if any, and
provide the opportunity to explore the operational limits for CARS diagnostics in
practical environments.
CARS CO spectra have been obtained at the U percent level which show quite
good agreement with the model code predictions. The spectra are quite interesting,
particularly with respect to the appearance of destructive interference effects.
These occur because of the modulation of the nonresonant susceptibility profile by
the real part of the resonant susceptibility. As the CO concentration level is
reduced, this modulation of the nonresonant susceptibility background is reduced,
and the "signal" disappears into the background profile. The CO detectability
studies have demonstrated both experimentally and analytically that for typical com-
bustion molecules such as CO, conventional CARS approaches are limited in species
sensitivity to about the 0.5 percent level. Detectivity is quite sensitive to
flame conditions. Conventional CARS measurements will require a three parameter fit
to the spectrum to ascertain the concentration and further study is required to
110
-------
are
demonstrate the uniqueness of such fitting procedures. If indeed the spectra
unique, then species concentrations measurements can be made based upon the shape
of the spectrum and not necessarily from the magnitude of the signal which would
be quite advantageous. Unfortunately, conventional CARS permits measurements
generally only to the several tenths of a percent level. To increase the concen-
tration sensitivity of CARS, the relative contribution to the CARS signal from the
nonresonant susceptibility must be reduced. There are several approaches to accom-
plish this end as reviewed in Ref. 1 such as resonance enhancement, double resonance
exploitation, and polarization orientation approaches. The latter two have only
been demonstrated in liquids. Resonance enhancement has been demonstrated in Ip,
but is not widely applicable since many molecules of combustion interest have
resonances too far into the ultraviolet to be easily accessible. The most promising
near term approach involves polarization orientation. With conventional phase-
matching this involves the use of "three colors." Three beams at frequencies u)Q,
ID, and (Up are used and by proper arrangement of the polarization directions, it is
possible to greatly reduce or eliminate the nonresonant susceptibility contribution.
With crossed-beam phase-matching (BOXCARS) it is not necessary to employ a third
frequency since the pump beam is already split into two components and angled. It
is recommended that polarization orientation BOXCARS investigations be pursued as
an avenue toward enhancing the species sensitivity capabilities of CARS.
Ill
-------
112
-------
1.
SECTION IV
REFERENCES
Eckbreth, A. C., P. A. Bonczyk and J. F. Verdieck: Review of Laser Raman
and Fluorescence Techniques for Practical Combustion Diagnostics. EPA Report
R77-952665-6, February 1977.
2. Eckbreth, A. C., P. A. Bonczyk and J. F. Verdieck: Laser Raman and Fluores-
cence Techniques for Practical Combustion Diagnostics. Appi Spect Rev
Vol. 13, PP. 15-16U, January 1978. ' ''
3. Piepmeier, E. H.: Theory of Laser Saturated Atomic Resonance Fluorescence.
Spectrochimica Acta, Vol. 27B, pp. 431-1^3 (1972).
4. Daily, J. W. : Saturation Effects in Laser Induced Fluorescence Spectroscopy.
Appl. Opt., Vol. 16, pp. 568-571, March 1977.
5. Baronavski, A. P. and J. R. McDonald: Measurement of C2 Concentrations in an
Oxygen-Acetylene Flame: An Application of Saturation Spectroscopy. J. Chem.
Phys., Vol. 66, pp. 3300-3301, April 1, 1977.
6. Mitchell, A0 C. G. and M. W. Zemansky: Resonance Radiation and Excited Atoms
Cambridge University Press (1971).
7. Pearse, R. W. B. and A. G. Gaydon: The Identification of Molecular Spectra
Fourth Ed., Chapman and Hall, London (1976).
8. Eckbreth, A. C. and J. W. Davis: Spatial Resolution Enhancement in Coaxial
Light Scattering Geometries. Appl. Opt., Vol. 16, pp. 80^-8o6, April 1977-
I Bleekrode, R. and W, C. Nieuwpoort: Absorption and Emission Measurements of
C2 and CH Electronic Bands in Low-Pressure Oxyacetylene Flames. J. Chem.
Phys., Vol. U3, pp. 3680-3687, November 15, 1965-
10. Jessen, P. F. and A. G. Gaydon: Study of the Absorption Spectra of Free
Radicals in Flames. Comb, and Flame, Vol. 11, pp. H-l6, February 1967-
U. Bleekrode, R.: Absorption and Emission Spectroscopy of C2, CH and OH in Low-
Pressure Osyacetylene Flames. Philips Res. Repts. Suppl. No. 7, PP. 1-6U U967J
12. Jessen, P. F. and A. G. Gaydon: Estimation of Carbon Radical
in Fuel Rich Acetylene-Oxygen Flames * Abj«*cn
posium (International) on Combustion, pp. 401-4oy,
113
-------
REFERENCES (Cont'd)
13. Bulewicz, E. M., P. J. Padley and R. E. Smith: Spectroscopic Studies of C2,
CH and OH Radicals in Low Pressure Acetylene and Oxygen Flames. Proc. Roy.
Soc., Vol. A315, pp. 129-1^8 (1970).
1U. Bulewicz, E. M., P. J. Padley and R. E. Smith: Elementary Combustion Pro-
cesses in Cyanogen and Oxygen and Hydrogen Flames: Spectroscopic Studies.
Fourteenth Symposium (international) on Combustion, pp. 329-3^1, The Com-
bustion Institute (1973).
15. Beenakker, C. I. M., P. J. F. Verbeck, G. R. MotiUnann and F. J. de Heer:
The Intensity Distribution in the CH (A2A - X2n) Spectrum Produced by Elec-
tron Impact on Acetylene. J. Quant. Spectres. Radiat. Transfer, Vol. 15,
PP. 33.3-3^0 (1975).
16. Hesser, J. E. and B. L. Lutz : Probabilities for Radiation and Pre dissociation
II. The Excited States of CH, CD and CH+ and Some Astrophysical Implications.
Astrophys. J., Vol. 159, PP- 703-718, February 1970.
17. Liszt, H. S. and W. H. Smith: RKR Franck-Condon Factors for Blue and Ultra-
violet Transitions of Some Molecules of Astrophysical Interest and Some
Comments on the Interstellar Abundance of CH, CH+ and SiH. J. Quant. Spectros,
Radiat. Transfer, Vol. 12, pp. 9^7-958 0-972).
18. Moore, C. E. and H. P. Broida: CH in the Solar Spectrum. J. Res. Natl. Bur.
Std., Vol. 63A, pp. 19-53, August 1959-
19. Luk, C. K. and R. Bersohn: Time Dependence of the Fluorescence of the B
State of CN. J. Chem. Phys., Vol. 58, pp. 2153-2163, March 1973.
20. NichollSj R. W. : Franck-Condon Factors to High Vibrational Quantum Numbers
III. CN. J. Res. Natl. Bur. Std., Vol. 68A, pp. 75-78, Jan. -Feb.,
21. Jevons, W. : The More Refrangible Band System of Cyanogen as Developed in
Active Nitrogen. Proc. Roy. Soc. (London), Vol. 112A, pp. l^-l^i (1926).
22. Suchard, S. N. : Spectroscopic Data, Vol. I. Plenum Press (New York) (197*0.
23. Jevons, W.: Report on Band-Spectra of Diatomic Molecules. The Physical
Society (London) (1932).
24. Bennett, R. G. and F. W. Dalby: Experimental Oscillator Strengths of CN and
NH. J. Chem. Phys., Vol. 32, pp. 1716-1719, June 1960.
-------
REFERENCES (Cont'd)
25- Gaydon, A. G. : The Spectroscopv nf TnomQn PndTW+i™ T TT-I
(New York) (l97^ - - ^^' Edition, J. Wiley and Sons
26. Daily, J. W. : Saturation of Fluorescence in Flames with a Gaussian laser
Beam. Appl. Opt., Vol. 17, pp. 225-229, January 15, 1978.
27. Regnier, P. R. and J. P. E. Taran: On the Possibility of Measuring Gas Con-
centrations by Stimulated Anti-Stokes Scattering. Appl. Phys. Lett., Vol. 23
pp. 2kQ-2k2, September 1973.
28. Regnier, P. R., F. Moya and J. P. E. Taran: Gas Concentration Measurement by
Coherent Raman Anti-Stokes Scattering. AIAA Paper 73-702, AIM 6th Fluid and
Plasma Dynamics Conference, Palm Springs, CA, July 1973.
29. Moya, F. S., S. A. J. Druet and J. P. E. Taran: Gas Spectroscopy and Tem-
perature Measurement by Coherent Raman Anti-Stokes Scattering. Opt. Comm.,
Vol. 13, pp. -169-17^, February 1975.
30. Regnier, P. R. , F. Moya and J. P. E. Taran: Gas Concentration Measurement by
Coherent Raman Anti-Stokes Scattering. AIAA J., Vol. 12, pp. 826-831, June
31. Moya, F., S. Druet, M. Pealat and J. P. Taran: Flame Investigation by Coherent
Anti-Stokes Raman Scattering, pp. 5^9-575 in B. T. Zinn, Ed., Experimental
Diagnostics in Gas Phase Combustion Systems, AIAA, New York, NY, 1977.
32. Maker, P. D. and R. W. Terhune : Study of Optical Effects Due to an Induced
Polarization Third Order in Electric Field Strength. Phys. Rev., Vol. 137,
pp. A801-A818, February 1965-
33. Begley, R. F., A. B. Harvey and R. L. Byer: Coherent Anti-Stokes Raman
Spectroscopy. Appl. Phys. Letts., Vol. 25, pp. 387-390, October 197^
34. Har-vey, A. B., J. R. McDonald and W. M. Tolles: Analytical Applications of
a New Spectroscopic Tool: Coherent Anti-Stokes Raman Spectroscopy (CARS) in
Progress in Analytical Chemistry, Plenum Press.
35- Tolles, W. M., J. W. Nibler, J. R. McDomld and A. B. Harvey: A Re view of
the Theory and Application of Coherent Anti-Stokes Raman Spectroscopy (CARS).
Appl. Spect., Vol. 31, PP. 253-272 (1977).
115
-------
REFERENCES (Cont'd)
36. Ilibler, J. W., J. R. McDonald and A. B. Harvey: CARS Measurement of Vibra-
tional Temperature In Electric Discharges. Opt. Comm., Vol. 18, pp. 371-373,
August 1976.
37. liibler, J. W., W. M. Shaub, J. R. McDonald and A. B. Harvey: Coherent Anti-
Stokes Raman Spectroscopy. pp. 173-225 in J. R. Durig, Ed., Vibrational
Spectra and Structure, Vol. 6, Elsevier, Amsterdam (1977).
38. Barrett, J. J. and R. F. Begley: Low-Power CW Generation of Coherent Anti-
Stokes Raman Radiation in CH^ Gas. Appl. Phys. Letts., Vol. 27, pp. 129-131,
(1975).
39. Barrett, J. J.: Generation of Coherent Anti-Stokes Rotational Raman Radiation
in Hydrogen Gas. Appl. Phys. Letts., Vol. 29, pp. 722-724, December 1976.
40. Roh, W. B., P. W. Schreiber and J. P. E. Taran: Single-Pulse Coherent Anti-
Stokes Raman Scattering. Appl. Phys. Letts., Vol. 29, pp. 174-176, August 1976.
4l. Nlbler, J. W. and G. V. Knighten: Coherent Anti-Stokes Raman Spectroscopy in
A. Weber, Ed., Topics in Current Physics, Chapter 7, Springer Verlag, Stuttgart
(1977).
42. Eckbreth, A. C.: Effect of Laser-Modulated Particulate Incandescence on Raman
Scattering Diagnostics. J. Appl. Phys., Vol. 48, pp. 4473-4479, November 1977.
43- Exciton Chemical Company, Dayton, OH.
44. Shaub, W. M., A. B. Harvey and G. C. Bjorklund: Power Generation in Coherent
Anti-Stokes Raman Spectroscopy with Focused Laser Beams. J. Chem. Phys., to
be published.
45. Eckbreth, A. C.: BOXCARS: Crossed-Beam Phase-Matched CARS Generation in
Gases. Appl. Phys. Lett., March 1978.
46. Rahn, L., Sandia Laboratories, Livermore, CA: private communication.
47. Hellwarth, R. W.: Third-Order Optical Susceptibilities of Liquids and Solids,
Pergamon Press, Oxford (1977).
48. Penney, C. M., L. M. Goldman and M. Lapp: Raman Scattering Cross Sections.
Nat. Phys. Sci., Vol. 235, pp. 110-112, February 1972.
116
-------
REFERENCES (Cont'd)
10. Bonczyk P. A A. C. Eckbreth and j. A. Shirley: Species Composition and
Temperature Measurements in Flames. Monthly Progress Report for the Period
September 27, 1977 to October 26, 1977 for EPA Contract 68-02-21765 UTRC
Report R77-952665-1U, November 1977.
50. Eckbreth, A. C.: Applicability of Laser Raman Scattering Diagnostic Tech-
niques to Practical Combustion Systems. Project SQUID Technical Report UTRC-
IJ.-RJ, October 1976.
51. Suchard, S. N.: Spectroscopic Constants for Selected Diatomic Molecules.
Air Force Report No. SAMSO-TR-7^-82, Aerospace Corp., Los Angeles, CA (197^).
52. Eckbreth, A. C.: Laser Raman Thermometry Experiments in Simulated Combustor
Environments, pp. 517-5^7 in B. T. Zinn, Ed., Experimental Diagnostics in
Gas Phase Combustion Systems, AIAA New York, NY (1977).
53. Hudson, B., W. Hetherington III, S. Cramer, I. Chabay and G. K. Klauminzer:
Resonance Enhanced Coherent Anti-Stokes Raman Scattering. Proc. Natl. Acad.
Sci. U.S.A., Vol. 73, PP. 3798-3802, November 1976.
54. Kanury, A. M. : Introduction to Combustion Phenomena, Gordon and Breach,
New York, p. 131 (1975)•
55. Rado, W. G.: The Nonlinear Third Order Dielectric Susceptibility Coefficients
of Gases and Optical Third Harmonic Generation. Appl. Phys. Letts., Vol. 11,
pp. 123-125, August 1967.
56. Samuelson, G. S. and J. N. Harman, III: Chemical Transformations of Nitrogen
Oxides While Sampling Combustion Products. J. Air. Poll. Cont. Asst., Vol. 27,
pp. 648-655, July 1977.
117
-------
118
-------
CARS Spectra of Combustion Gases
by
R. J. Hall
United Technologies Research Center
East Hartford, Connecticut 06108
ABSTRACT
Computer-generated CARS spectra, necessary for temperature and species con-
centration measurements in combustion environments, are presented. The third-
order, nonlinear electric susceptibility governing CARS generation from Np and
CO has been programmed in terms of a Boltzmann distribution of vibration-rotation
states and the normal Raman cross-sections. Finite pump laser widths and instru-
mental effects are accounted for by appropriate convolutions. Examples of flame
temperature (Np) and minority species concentration (CO) measurements using these
theoretical spectra are presented.
AI-1
-------
-iact ion
Coherer/. Anti-Stokes Raman Spectroscopy (CARS) is a nonlinear, three-wave
optical mixing technique with considerable promise for combustion diagnostics
(Refs. 1-7). Its high potential conversion efficiency and coherence make it well
suited for thermometry and species concentration measurements in combustion environ-
ments; relative to normal Raman spectroscopy, CARS offers superior discrimination
against incoherent fluorescence, background luminosity and particulate incandescence.
The advantages and disadvantages of CARS for combustion diagnostics are discussed
in Ref. (7).
Data reduction in CARS is less straightforward than in conventional Raman
spectroscopy, however, a complication arising from spectral interference effects in
CARS. Signal strength in normal vibrational Raman spectroscopy is a simple sum of
contributions from neighboring transitions, resulting in a signal directly proportional
to vibrational state population. A CARS signal, however, is affected by interfer-
ences between neighboring transitions and between resonant and nonresonant contribu-
tions to the electric susceptibility. These interference effects complicate CARS
data reduction, and generally require that temperature or concentration measurements
be made by fitting computer-generated theoretical spectra to experiment. This paper
describes a model which generates synthetic CARS spectra, and gives examples of
thermometry and minority species concentration measurement in an air-methane, atmospheri
flame.
AI-2
-------
Theory
The basic theory of the CARS effect has been presented extensively elsewhere
(Refs. 1-6 ). Thus, only the important results needed for calculations are
summarized herein. For monochromatic, phase-matched pump lasers (powers B, and
P2) the CAHB generated" power at uuo = 2u>,-ui-> is proportional to:
pi% M2 (i)
where X is the complex, third order electric susceptibility. When the frequency
difference u>]_-U)2 is tuned to a Raman-active molecular transition, resonance contri
butions to X become important and we may write
If homogeneously broadened transitions are assumed the resonant contributions to
X can be expressed as a sum over neighboring transitions:
XR = XR + iX^ = S K- . (3)
K K n . 3
where the j summation is over all Q, 0, and S (AJ = 0, + 2) vibration-rotation
transitions in the vicinity of u^-u^, YJ is the homogeneous linewidth (FWHM), and
Aui^ is the detuning
-------
A-, nuclear spin statistics are accounted for, and a Dunham expansion has been
emrloyed in the calculation of vibration/rotation energy levels. Mutual equili-
brium between vibrational and rotational modes has been assumed at temperature T.
For forward scattering, the spontaneous Raman cross-section can be expressed in
terms of the components of the derived polarizability tensor:
da, ,^2^ h r 2 7 , J 2-, , ..
— = (—) [a * T-T b_v ] (v-t-1)
ds.Q c 21-l^Q 45 J
(5)
do, /^x Jl /7 UJ 2. , >
where :•' and o> are the reduced mass and angular frequency of the molecular oscil-
j
lator, the bTT are the Plac?e?:- Teller coefficients (Ref. fa ) and cc and y are "the
j
derivatives (with respect to internuclear coordinate ) of the mean molecular polari-
zacility andanisotrocy . v is the vibrational quantum number of the initial le'vel,
and the factor (v+l) is contributed by the vibrational matrix element.
The foregoing development is applicable to narrow band pump lasers. In
]. ractice, a broadband ?2 can be employed to generate a complete spectrum in a
single shot, and P-j_ may have an appreciable width. The frequency distribution of
the scattered power for finite laser line widths can be derived from the macroscopic,
third order polarization, whose frequency components are given by
00 CO
P3(uO = j' do?' E^UJ') J do/' X (uj'-a-11) E (u;-u)W)
where EI and E2 are the Fourier components of the pump fields. The solution of
Maxwell's Eq'oations with (6) as source term gives a CARS intensity proportional to
(7)
AI-4
-------
For a monochromatic P.^ use of a broadband Pg generates the same spectrum as that
obtained by tuning a monochromatic Pg. Because the CARS signal is linear in P
the signal can be represented as a superposition of the individual spectral elements
of Pg. The achievable resolution is, therefore, dependent on the width of P,.
There are consequently three medium parameters (N, T, \m) which go into the
calculation of a CARS spectrum, so that in principle a temperature or concentration
measurement entails a three-parameter fit. In addition, the homogeneous linewidths
Y are not well known, requiring an examination of the sensitivity of CARS spectra
to reasonable variations in these widths.
A computer program based on Equations (2-6) has been written to generate CARS
spectra numerically. The numerical scheme selects an appropriate spectral range
uv-uVj and divides it into a number of discrete intervals. Based on the Boltzmann
populations for the assumed T and N, the program calculates a value \((^-^) from
Equations (2-5) and performs the convolution integral (6) numerically.
In this analysis we have ignored population perturbations due to the pump
lasers, as well as stimulated Raman effects. It has also been assumed that the
fields E(uO and E(u)2) have the same polarization directions (Polarization Condition
(1) of Ref .( 9 )).
Flame Thermometry and Concentration Measurement
We now consider the application of the foregoing theory to the problems of N2
theracmetry and CO concentration measurement in an air-methane, atmospheric flame.
These calculations were performed in conjunction with the experiments of Ref. (lO),
in which vibration-rotation transitions were probed with a broed-band Pg. In these
AI-5
-------
experiments, a pulsed, frequency-doubled Nd:YAG laser and a dye oscillator pumped
by a portion of the Nd laser output provided the two excitation beams. Details of
the optical apparatus and flat flame burner are given in Ref. (lo); pump laser
parameters characteristic of the experiments are given in Table 1. Because the flame
was probed in the postflame region, a nominal gas concentration of 70$ N2, 20$ HpO
and 10$ COo has been assumed in these calculations. The calculations are applicable
to both scanned and single pulse experiments.
Temperature measurement is best performed using a dominant molecular species
such as Np. In an atmospheric flame, the resonant CARS contribution from N2 will
be so much stronger than the nonresonant contribution (|XjJ » XNR) that XJJR &nd N
become relatively insignificant parameters. In the absence of significant inter-
ference with the nonresonant background, the temperature can be deduced from the
qualitative shape of the CARS spectrum.
With T thus established, the interference between the CARS signal for a
minority species such as CO and the background offers the promise of concentration
measurements without having to make an absolute intensity measurement. This possi-
bility arises because the CO resonant susceptibility modulates an approximately
known background susceptibility. In the limit of dominant Xj™, the susceptibility
takes on the dispersive profile of Xp.
For the two molecules of interest in this study, Ng and CO, the molecular polari-
zability parameters were calculated from the spontaneous cross-section and depolari-
zation data of Ref. (ll). For the homogeneous linewidths there is pure rotational
data (Ref. 12 ) for both molecules, and some data for the 0 -» 1 band Q-branches of
NP (Ref. 13 ). However, it is fair to state that the magnitudes, J-dependence, and
AI-6
-------
temperature dependence of the widths are uncertain; consequently, we have assumed a
constant nominal value of .1 cm" for both molecules. Nonresonant background
susceptibilities have been taken from Ref. (lk). Lacking an experimental value for
the HgO susceptibility, a value equal to that for CH^ has been assumed. For the
assumed postflame composition, Xj^ is calculated to have an STP value of 5.25 x lO"1^
cm3/erg, a value about 30 percent larger than that for pure Ng (Ref. Ik). Because
the gas composition is not precisely known, and the EpO nonresonant susceptibility
is uncertain, the sensitivity of the predicted spectra to variations in the background
susceptibility is examined.
$2 Thermometry-
Rotational temperatures can be most accurately inferred by probing the 0 -» 1
Q branches with nearly monochromatic P, and P , making it possible to dispense with
a monochromator. For a broadband P2, high resolution can still be achieved if Pj_ is
not too broad. These calculations are concerned with experiments in which a broad-
band Pp and narrowband P-j_ are employed, but in which the monochromator resolution
may not be sufficient to resolve individual Q-branches. In the low resolution case,
inference of a vibrational-rotational temperature can utilize the relative strengths
of the 0-1 and 1-2 band envelopes.
Figure 1 shows how the calculated % spectrum at T = 1700°K evolves from the
monochromatic spectrum, through the correction for finite pump laser widths, and
through a final convolution with a triangular instrumental slit function. In the
0-1 band of the monochromatic spectrum, two distinct envelopes corresponding to the
even and odd numbered Q branches are evident. The even rotational quantum number
transitions are stronger because of a higher nuclear spin statistical weighting.
AI-7
-------
In the convolved spectrum, the low-lying Q-branches are unresolved, but at the
higher quantum numbers the spacing between transitions due to vibration-rotation
interaction (ae ) becomes sufficient to permit the resolution of even numbered Q
branches. Also of interest is the disturbance of the 1-2 band by the 0-1 band that
is evident in the spectra.
In Figure 2, the N profile is compared for T = 1200°K and 2000°K. Of note
here are the rapidly increasing modulation depth and 1-2 band strength for increasing
temperature. The shift of the spectrum maximum evident in Figure 2 is due to the
changes in the rotational Boltzmann distribution. These moderate resolution N~
CARS spectra contain prominent spectral features which vary strongly with temperature,
offering the potential for sensitive temperature determinations.
For low monochromator resolution, many of the spectral features cannot be
resolved, and the "hot band" can be used for temperature determination. Figure 3
displays the calculated low-resolution spectra for a range of temperatures. For
display purposes, the maxima of the different curves have been brought into coin-
cidence. While the width of the 0-1 band broadens substantially with increasing
temperature, the hot band intensity ratio is a much more sensitive function of
temperature. This peak height ratio is shown as a function of T in Figure k.
With regard to simplified data reduction, the expression k exp (-2^oiv/kT)
does not provide a useful relationship between temperature and peak height ratio,
and it appears that interference effects limit the utility of measuring the slope
of the 0-1 band on its hot side (Ref. ( 3 )). The best results will be obtained by
analyzing experimental data with the aid of a computer model.
AI-8
-------
Because of the uncertainty in the Q-branch linewidth data, the sensitivity of
the predicted N2 CARS spectra to reasonable variations in these widths has been
examined. Figure 5 displays the variation of peak height with y for T = 1300°K,
1700 K, 2000°K. As seen, for linewidths greater than .1 cm"1, the sensitivity is
not great. For values less than .1 cm~ , the sensitivity is somewhat greater. These
results suggest that knowledge of the homogeneous linewidths is not critical for CARS
thermometry.
Uncertainties in the value of x.^ do not appear to have much impact on the
predicted N2 CARS spectra; varying XJ^R bY ± 33 percent at T = 1700°K changes the
predicted peak height by less than 10 percent.
A comparison of the theory and experimental data is shown in Figure 6.
Radiation-corrected thermocouple and spontaneous Raman measurements had previously
established a temperature in the range l625-1700°K (Ref. 15). The CARS prediction
for T = l650°K is seen to be in quite good agreement with the corresponding experi-
mental trace; a particularly good fit is obtained for the 0-1 band, where the moderate
monochromator resolution permits the observation of individual Q-branches (Q(28)~
QfrO)). This fit was obtained using a value of .0177 cm"1 for the rotational con-
stant a (Ref. 16); use of the Ref. (17) value (.0187 cm"1) in the theoretical
program resulted in predicted wavelength assignments for the large J Q-branches that
were incorrect by about 1 cm"1. While the fit to the 0-1 band is excellent, and
the overall height and width of the 1-2 band are well accounted for, the theory
does not reproduce the finer features of the 1-2 hot band as well. In particular,
there is a discrepancy concerning the prominent peak on the hot band. The predicted
1-2 features prove to be quite sensitive to the choice of vibrational anharmonic
AI-9
-------
constants; thus, more exact spectroscopic constants or linewidths may be needed
for the theory to reproduce hot band fine structure accurately. If individual Q-
branches can be resolved, however, it should suffic* to fit the theory to the 0-1
band.
AI-10
-------
CO Concentrati_on_Measurement
It has been shown that N2 thermometry essentially requires a one-parameter
fitting of theoretical and experimental spectra. Measuring small CO concentrations
0 (1$) is more complicated, because the background susceptibility sets a lower
limit on detectivity, and because the fitting may involve more than one parameter.
A sample calculated CARS spectrum is shown in Figure 7 for an assumed flame
temperature of 1700°K and CO mole fraction of .04. The "theoretical" and convolved
spectra display the expected destructive interference dip and a pronounced contri-
bution from the v = 1 -» 2 hot band. This hot band contribution suggests that the
CO CARS spectra may contain useful information about temperature as well as concen-
tration; temperature variations show that the 0 -» 1 and 1 -» 2 bands in the CO spectra
undergo changes similar to those in N2 (Figure 3)- The asymptotic limits of the
:onvolved spectra follow the profile of P2.
Figure 8 displays the effect of CO concentration variation at fixed T and
*NR' As tne co concentration increases, its spectrum emerges from the nonresonant
background, with the interference minimum and hot band features becoming more pro-
nounced. For the assumed XNR, which is about 25 percent larger than the pure N2
ralue (Ref. lU), a lower limit on CO detectivity of slightly less than 5 x 10"3 is
implied. In reality, lower CO mole fractions could be accompanied by changed con-
centrations of gases like C^ which possess relatively high background susceptibility
(Ref. ), thus changing detectivity. The achievable detectivity in any particular
experiment will depend strongly on the background gas concentration through XTO.
Because the moduli K. are inversely proportional to the homogeneous line-
Vidths, Yj (Equation 1*) it might be expected that the latter would be important
AI-H
-------
parameters in minority species CARS spectra. The variation of the predicted CO
spectrum with y ^ exhibited in Figure 9 shows that knowledge of the homogeneous
J
linewidths is not critical for minority species concentration measurements. The
predicted sensitivity is, however, large enough to suggest a need for measurements
of these linewidths for very accurate work. As the Y^ increase, the contribution
J
of the resonant susceptibility, and thus the detectivity, are slightly reduced.
The agreement of these calculations with the experimental spectra of Refs.
( 10 ) is good, as shown in Figure 10, which compares a calculated spectrum for a
CO mole fraction of . 04 to an experimental trace. Simultaneous quartz microprobe/
ndir analyzer measurements had established a CO mole fraction of about .04. The
temperature of 1700 K employed in the calculations is consistent with the ^ ther-
mometry performed'in the same experiment, and the fact that the ratio of spectrum
extrema is well reproduced indicates that the background susceptibility is nearly
correct. While the calculated spectra are sensitive to the value assumed for XJJT>,
it was not found necessary to change this parameter from the nominal value assumed
throughout these calculations. Reproduction of the slight asymmetry in the experi-
mental spectrum did require an adjustment in the center frequency of Pp.
Parametric variations with N and X^p show, however, that these solutions are
not unique, and that the inferred CO mole fraction is subject to an uncertainty
determined by the uncertainty in *JJR. In the limit of small N, Equation (2) gives
M2~ *NR C*NR + 2XR)
AI-12
-------
Thus, there will be a family of solutions determined by N/Xjjp = const, that have
the same dispersive profile. A calculation with the computer model confirms that
the experimental spectrum of Figure 10 can be fit with a CO mole fraction of .03 if
X.rn is taken to be .75 times its nominal value. This result demonstrates the need
for a measurement of the HgO nonresonant susceptibility, and suggests that CARS
minority species measurements would be most accurately performed in conjunction with
computer modelling of the majority species concentration profiles.
Conclusions
A computer model for CARS generation from N2 and CO has been described. The
model has been applied to the practical problems of No thermometry and CO concen-
tration measurement in an air-fed methane flame. In both cases, the CARS measurements
obtained by fitting theoretical and experimental spectra are consistent with measure-
ments obtained by other techniques. For experimental conditions in which individual
Q-branches of the v = 0-1 band of Np could be resolved, a very precise fit of theory
to experiment has been achieved. Under low resolution conditions, the N2 hot band
spectrum can be employed for temperature determination. An examination of the factors
influencing predicted CARS spectra indicates that neither type of measurement is
critically dependent on the values assumed for the homogeneous linewidths, with CO
concentration measurement somewhat more sensitive than N0 thermometry. The accuracy
of CO mole fraction measurement is shown to be subject to an uncertainty determined
by knowledge of the nonresonant susceptibility.
Acknowledgement
The author is greatly indebted to Dr. Alan Eckbreth for supplying the experimental
spectra used in this paper and for many informative discussions of this subject.
AI-13
-------
References
1. N. Bloembergen, Nonlinear Optics, Benjamin, New York (1965).
2. P. D. Maker and R. W. Terhune, Phys. Rev. 137, A801 (1965).
3. F. Moya, et. al., 'Flame Investigation by Coherent Anti-Stokes Raman Scattering,"
in Experimental Diagnostics_in Gas Phase Combustion Systems, B. T. Zinn,
Editor, AIAA, New York (1977)-
k. W. M. Tolles, et. al., Applied Spectroscopy ^l, 253 (1977).
5. J. W. Nibler and G. V. Knighten, "Coherent Anti-Stokes Raman Spectroscopy,"
in Topics in Current Physics, A. Weber, Editor, Springer Verlag, Freiburg
(1977).
6. R. N. DeWitt, A. B. Harvey and W. M. Tolles, Theoretical Development of Third-
Order Susceptibility as Related to Coherent Anti-Stokes Raman Spectroscopy
(CARS), Naval Research Laboratory Memorandum Report 3260 (1976).
7. A. C. Eckbreth, P. A. Bonczyk, and J. F. Verdieck, "Review of Laser Raman and
Fluorescence Techniques for Practical Combustion Diagnostics," Applied
Spectroscopy Reviews 13, 15 (1978).
C. G. Placaek and E. Teller, Z. Physik 8l, 209 0-933)-
9. M. D. Levenson and N. Bloembergen, Physical Review B 10, W+7 (197^).
10. A. C. Eckbreth, Appl. Phys. Lett. _3_2, April 1, 1978.
11. C. M. Penney. L. M. Goldman and M. Lapp. Nature Phys. Sci. 235, HO (1972).
12. K. S. Jammu, G. E. St. John and H. L. Welsh, Can. J. Phys. IfU, 797 (1966).
13. W. H. Fletcher, private communication, October 1977.
14. W. G. Rado, Appl. Phys. Lett. 11, 123 (1967).
15. A. C. Eckbreth, Applicability of Laser Raman Scattering Diagnostic Techniques
to Practical Combustion Systems, Project SQUID Technical Report UTRC-U-PU
October 1976.
16. W. Benesch, et. al., Astrophysical Journal 1^2, 1227 (1965).
17. G. Herzberg, Molecular Spectra and Molecular Structure. I. Spectra of Diatomic
Molecules, D. Van Nostrand Co., Inc., Princeton, NJ (1950).
AI-14
-------
Figure Captions
Fig. 1 Calculated CARS spectrum for flame N2, T = 1700°K.
(a) CARS signal for monochromatic P-^, P2.
(b) Signal for finite width, P-^ P2 (Table l).
(c) Convolution of (b) with triangular slit function, resolution 1 cm"1.
Fig. 2 Computed N2 CARS spectra for T = 1200, 2000°K. Slit width 1 cm"1.
Fig. 3 Low resolution Ng CARS spectra for temperature range 1000-2000°K. Slit
width 2.7 cm"1.
Fig. U Variation of hot band intensity ratio with temperature for flame Ng.
Slit width 2.7 cm"1.
Fig. 5 Sensitivity of predicted hot band intensity to assumed homogeneous line-
width. Flame N2, slit width 2.7 cm"1.
Fig. 6 Comparison of theoretical and experimental N2 CARS signal for T = l650°K.
- experimental, ...... theoretical. Slit width 1.25 cm~l, a^'^' =
165UO cm"1, Au> = 20° cm"1.
Fig. 7 Calculated CARS spectrum for k% CO, T = 1700% XNR = 9-2 x 10"19 cm3/erg
(a) CARS signal for monochromatic Pj_, P2.
(b) Signal for finite width PI? P2 (Table l).
(c) Convolution of (b) with triangular slit function, slit width 2 cm"1.
Fig. 8 Variation of predicted CARS spectra with CO mole fraction. All other
parameters as in Fig. 7. Instrumental convolution has not been performed.
Fig. 9 Variation of predicted CO CARS spectra with homogeneous linewidth.
All other parameters as in Fig. 7.
Fig. 10 Comparison of theoretical and experimental CO CARS spectra. — experi-
mental, ...... theoretical. Slit width 1 cm"1, u^0' = 16640 cm"x. All
other parameters as in Fig. 7.
AJ-15
-------
Table 1
Nominal Pump Laser Parameters
= 18797 cm"1 u>2 = l6if81 c™"1 (N2^
= 1665*1 cm"1 (CO)
Gaussian Widths
= 1.2 cm"1 Acu = 150 cm
AI-16
-------
150
10CH
Ixl
z
LU
h-
CO
Z
LU
I-
D
LU
N
DC
O
Z
50H
2240
0.25
0.20
0.15-
0.10-
0.05^
o.oo-
21070
2280 2320
WAVENUMBER-CIVr1
2360
»t^JT T~~»^T 7 T V^ 1 ¥ [ »~»~~r~TT~¥ T T^»T~T T T
21090 21110 21130
WAVENUMBER-CM
-1
21090 21110 21130
WAVENUMBER-CM
-1
78-01-176-1
AI-17
-------
0
21070 21080 21090 21100 21110 21120 21130 21140
WAVENUMBER- CM
• 1
78-02-133-1
AI-18
-------
TEMPERATURE°K
21080
21100 21120
WAVENUMBER-CM -1
21140
AI-19
78-01-176-2
-------
o
I-
cc
to
I
1000 1250
1500 1750 2000
TEMPER ATURE-°K
2250 2500
••8-01-176-3
AI-20
-------
0.3,
0.2
2000 °K
1700
0.1
g 0.09
< °-08
oc 0.07
H 0.06
-------
21070
,.,., I.... I.... I.... I
21080 21090 21100 21110 21120 21130 21140
WAVENUMBER, CM
-1
78-03-165-1
AI-22
-------
0.4
0.3-
0.2-
0.1-
0.0
2060 2100 2140 2180
WAVENUMBER-CM-1
WAVENUMBER-CM
0.2
20880 20920
WAVENUMBER-CM
20960
-1
78-02-202-1
AI-23
-------
X10-3
Q.1%
\J,\JiJ .
• DU ~
Oca -
.DO .
Ocr» -
.ou
O^m -
.*+o
£ 0.40-
/
/
/
w 20,880
z
UJ
Z X10-3
y
/
'
X^
20,920
1%
^^s
\
\
.20,960
0.
20,880 20,920
,960
X10-3
0.5%
Ofi C '
.DOT
Ocn '
.DO:
.55.
o.so ;
.45 .
n AnJ
/
/
/
r
s \
^ ^
V
\
\
20,880
WAVENUMBER, CM
-1
20,960
20,880 20,920 20,960
78-O3-97-01
AI-24
-------
0.05 CM-
z 0
UJ
0.5
Q
ID
1.75
20880 20900 20920 20940 20960 20980
7j = 0.20 CM-1
Q 1.50-4-
1.25
1.00
0.75
0.50
20900
20920 20940
WAVENUMBER, CM~1
20960 20980
.78-03-169-1
AI-25
-------
Ul
K
COMPUTER MODEL, 4% CO
T= 1700°K
20,880
20,900
20,920
20,940
WAVENUMBER - CM
-1
20,960
AI-26
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
. REPORT NO.
EPA-600/7-7 8-104
3. RECIPIENT'S ACCESSION- NO.
TITLE AMD SUBTITLE
Investigation of Saturated Laser Fluorescence and
CARS Spectrescopic Techniques for Combustion
Diagnostics
5. REPORT DATE
June 1978
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
A.C.Eckbreth, P.A.Bonczyk, and J.A.Shirley
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
United Technologies Research Center
ast Hartford, Connecticut 06108
10. PROGRAM ELEMENT NO.
EHE624
11. CONTRACT/GRANT NO.
68-02-2176
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final; 10/76-1/78
14. SPONSORING AGENCY CODE
EPA/600/13
15 SUPPLEMENTARY NOTES jERL-RTP project officer is William B. Kuykendal, Mail Drop 62
919/541-2557.
is. ABSTRACT
repOrt gives results of comparisons of saturated laser-excited
molecular fluorescence measurements of CH and CN in atmospheric pressure
acetylene flames with absorption measurements of these flame radicals. It was
found possible to saturate the fluorescence intensity of both radicals with readily
achieved levels of laser spectral intensity (100,000 to 1 million watts per square
centimeter-reciprocal centimeter). Coherent Anti-Stokes Raman Spectres copy
(CARS) thermometry investigations were conducted on flame nitrogen in a variety of
flames, including highly sooting propane diffusion flames. CARS species sensitivity
was addressed in a study of CO detectability.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDEDTERMS
c. COSATI Field/Croup
Pollution Propane
Combustion Acetylene
Diagnosis Nitrogen
Raman Spectres copy Cyanogen
Fluorescence Carbon Monoxide
Lasers Stokes Law
Pollution Control
Stationary Sources
CH
CARS
Anti-Stokes
13B
2 IB
06E
14B
20F
20E
07C
07B
20D
3. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report)
Unclassified
20. SECURITY CLASS {Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
------- |