EPA-650/2-74-020
January 1974
Environmental Protection Technology Series
llliiit
.;.;.;.
-------
EPA-650/2-74-020
FEASIBILITY STUDY OF THE USE
OF RESONANCE SCATTERING
FOR THE REMOTE DETECTION OF S02
Prepared by:
Michael C. Fowler and Paul J. Berger
United Aircraft Research Laboratories
East Hartford, Connecticut 06108
Contract No. 68-02-0656
Program Element No. A11010
Project Officer: William F. Herget
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, N. C. 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
January 1974
-------
This report has been reviewed by the Environmental Protection Agency
and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Agency,
nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
11
-------
ABSTRACT
FEASIBILITY STUDY OF THE USE OF RESONANCE
SCATTERING FOR THE REMOTE DETECTION OF SO
An analytical and experimental investigation has been carried out to
determine the feasibility of using the scattering of ultraviolet radiation
by SO as a probe of the concentration of that molecule in stationary
source emissions. Both ordinary fl rescence and resonant Raman scattering
were considered and experimentally it was found that the latter component
was present in the scattered radiation with sufficient magnitude to reduce
significantly the degrading effect that ordinary fluorescent quenching has
on this scattering technique. Further analysis revealed that current state-
of-the-art dye lasers deliver sufficient ultraviolet pulse energy to permit
SO concentration determination in practical situations but that fluorescent
scattering from particulates presents a possible constraint to the validity
of this technique. A field program is recommended to investigate the latter.
iii
-------
TABLE OF CONTENTS
Page
ABSTRACT [[[ ii:L
LIST OF ILLUSTRATIONS ........................................ v
LIST OF TABLES .............................................. vi
SUMMARY [[[ vii
SECTION I INTRODUCTION .................................... 1
SECTION II THEORETICAL BACKGROUND ........................... 3
A . Introduction ....................................... 3
B. Detection of SO "by Ultraviolet Fluorescence ....... 3
C. Raman Scattering
D. Analysis of Scattering Containing Both Raman and Fluorescence
Components ......................................... 15
SECTION III EXPERIMENTAL RESULTS ............................ 19
A . Apparatus ........................................... 19
B. Experimental Procedure and Results ................ 25
SECTION IV CONCLUSIONS AND RECOMMENDATIONS ................. 33
A . Introduction ...................................... 33
B. Lasers For SO- Detection .......................... 33
C. The Problem of Quenching ............................ 35
D. The Problem of Background Radiation ................. 38
E. Scattering from Parti culates ........................ 39
F. Field Evaluation .................................... ^0
-------
ILLUSTRATIONS
Page
1. Wavelength Dependence of the Extinction Coefficient of SO for
the Wavelength Region between 200A and Ij-OOOA k
2. Vibrational Structure of Electronic Transitions 6
3. The Normal Vibrations of SO 8
k-. Wavelength Dependence of Fluorescence of SO for the wavelength
region between 2900$ and 5000$ 10
5. Resonant Raman Scattering and Fluorescence in Liquid Solutions 16
6. Optical Train Used in SO Light Scattering Measurements 20
T. 1.2 W. Flashlamp Pumped Dye Laser 21
8. Electronics Used in SO Light Scattering Measurements 2k
9. Gas Handling System Used in SO Light Scattering Measurements 26
10. Pressure Dependence of the Apparent Rate Coefficient for
Quenching of S0_ Fluorescence at 3105A by N 28
11. Pressure and Laser Wavelength Dependence of X 30
a
12. Laser Pulse Energy Needed to Detect SO as Function of
Distance L from sample 36
-------
TABLE
1. Quenching Rates of SO Ultraviolet Radiation
vi
-------
SUMMARY
An investigation has been carried out to determine the feasibility of the
use of light scattering for the detection of SO in stationary source emissions.
An important question concerning the validity of this detection scheme as an SO
concentration probe concerns the possibility of whether the detected scattered
signal is affected by the concentration of components in the mixture other than
SO . Fluorescent scattering is particularly vulnerable to this, due to quenching
through intermolecular collisions. In contrast, Raman scattering, though ordinarly
small in magnitude, is absolutely free of quenching effects and may be enhanced
by the occurrence of the phenomenon known as resonant Raman scattering. Measure-
ments vere taken at room temperature of the intensity of the light scattered
o
inelastically from an incident laser beam of wavelength 3000 A, coincident with
an S0o absorption peak. The pressure dependence of this signal indicated the
presence of an unquenched component of magnitude sufficient to reduce significantly
the composition dependence of the SO concentration inferred from a light scattering
measurement, as well as to increase the magnitude of the radiation scattered at
o
one atmosphere pressure. Measurements taken at 200 C also indicated the presence
of the Raman component, but no information was attained as to whether the magnitude
the component depended on the temperature; nor was any information attained as to
to whether the magnitude of the Raman component displays a significant sensitivity
to which RO absorption peak is coincident with the incident laser radiation
wavelength. However, the most serious limitation on this method of SO
2
concentration measurement may be broadband fluorescent scattering from particulate:-
present in the plume, giving rise to the need offer direct field evaluation of
thin technique.
vii
-------
SECTION I INTRODUCTION
This study is concerned with the feasibility of using the intensity of light
scattered inelastically from a laser beam incident on a gas sample containing SO
as a valid indicator of the concentration of SO in that gas sample. Particular
focus is placed upon the effects of quenching collisions upon the magnitude of
the scattered signal. Fluorescent scattering is particularly vulnerable to
quenching, and the SO concentration inferred from the scattered light may be sig-
nificantly influenced by the relative concentrations of the other components in
the mixture. In contrast, Raman scattering is immune to quenching effects, but
is generally much smaller in magnitude than fluorescence. However, if the wave-
length of the incident light lies near an SO absorption line, the magnitude of
the Raman scattering cross section may be considerably enhanced. Accordingly,
effort was concentrated upon determining whether a Raman component of significant
magnitude appeared in the scattered radiation for a laser beam of wavelength iden-
tical to that of the SO G line.
Section II presents a brief theoretical background on the phenomena of
fluorescent and Raman scattering. Particular emphasis is placed upon the influence
of quenching on fluorescent scattering and upon the resonant Raman effect. An
expression is developed for the influence of quenching gas pressure on a scattered
signal containing both fluorescent and Raman components.
In Section III the experimental apparatus and procedure are described, and
the measurements of the dependence of the measured scattered intensity on quenching
gas pressure and laser line wavelength are presented and discussed. These measure-
ments indicate the presence in the scattered signal of a Raman component of magnitude
sufficient to significantly decrease the dependence of the scattered signal to the
quenching gas mixture composition, and increase the magnitude of the signal at
atmospheric pressure.
-1-
-------
Section IV contains recommendations, based on the results of this study, on
the influence of SO concentration and sample to laser distance upon the magnitude
of the laser pulse needed to detect that concentration. In addition, areas in
vhich further laboratory scale studies are necessary are enumerated, and a field
evaluation program is described.
-2-
-------
SECTION II THEORETICAL BACKGROUND
A. Introduction
In this section, the theory underlying the detection of SO by light scattering
techniques in the ultraviolet region of the spectrum is briefly reviewed. Part B
of this section deals with the detection of SO through fluorescence, a two-photon
process wherein the SO molecule first absorbs a photon from the incident laser
radiation, and after a short but finite time interval, emits a second photon at
a wavelength not necessarily identical to that of the incident radiation. The
effect of intermolecular collisions which act to quench or diminish the efficiency
of this process is discussed and the resulting limitations on the accuracy of SO
concentration determination imposed by quenching are noted. In Part C, detection
of SO by Raman scattering, a single photon process free of quenching effects, is
discussed with special emphasis placed on enhancement of this process for laser
radiation whose wavelength coincides with a maximum in the SO absorption spectrum.
Finally, Part D presents the development of an expression, to be used subsequently
in analyzing the experimental results, for the pressure variation of the apparent
fluorescence quenching rate observed when both fluorescence and Raman scattering
are present.
B. Detection of 50^ by Ultraviolet Fluorescence
Absorption Coefficient
The wavelength dependence of the absorption coefficient of SO in the wave-
o o
length region between 2000 A and ^000 A is shown in Fig. 1 (Ref. 1). Three distinct
o o
absorption systems are evident. In the wavelength region between 3500 A and kOQO A,
the absorption is due to excitation of the molecule from the ground or X state to
r+j **-*
the a state. The latter is a triplet electron spin state whereas the X state is
singlet, and radiative transitions between the two are thus rather weak as reflected
by the small value of the absorption coefficient. Stronger absorption systems
-3-
-------
o 0.3
u
I 0.2
LL
•.
:
;
x 0.1
0.0
2000
2500 3000
WAVELENGTH (A)
3500
FIGURE 1. WAVELENGTH DEPENDENCE OF THE EXTINCTION COEFFICIENT OF SO2
FOR THE WAVELENGTH REGION BETWEEN 2000 A AND 4000 A
-k-
-------
between the ground state and the singlet spin states designated A and C occur
o o o o
in the wavelength regions between 2600 A and 3*4-00 A, and 1800 A and 2*4-00 A, respec-
tively, and the former of these two systems is the one of interest to this study.
The absorption coefficient for all three of these transitions exhibits considerable
variation in magnitude. In particular, the X -» A absorption is seen to consist of
many maxima, each of which is several angstrom units wide. The cause of this
structure may be seen by examination of Figure 2(a), which presents the variation
of the internuclear potential energy V(r), on the internuclear distance r for two
electronic energy states of a simple diatomic molecule. The two states are labeled
X and A, and the potential energy minimum of the upper state is seen to occur at
a larger value of r than that of the lower state. The vibrational levels of the
upper and lower states are labelel v' and v'1, respectively. The probability of
a transition occurring between the v1 ' vibrational level of the lower state to
the v1 level of the upper with absorption of radiation is proportional to the Franck-
Condon factor given by
dr'2
where i|f , and \|r M are the vibrational wave functions of the upper and lower states,
respectively. Generally, q , ,, is largest for those values of v1 and v" for which
i|r , and $ ,, both have maxima at the same value of r (Ref. 2). For the lowest vib-
rational level of any given electronic state, the maximum falls at the value of r
for which the potential energy curve for the electronic state is at its minimum.
But as v' increases, the maxima in \|r tend increasingly to occur at the values of
r for which the potential energy curve equals the energy of the vth level. During
a transition between two electronic states, the value of r is constant, since the
time elapsed during the transition is much shorter than the vibrational period of
rw
the two nuclei. A transition from v" = 0 in the X state is thus shown as a vertical
arrow in Figure 2 and, from the argument given above, the frequency dependence of
the absorption coefficient for transitions originating from this level will consist
-5-
-------
i v
:50
V
(a) ABSORPTION
(b) EMISSION
FIGURE 2. VIBRATIONAL STRUCTURE OF ELECTRONIC TRANSITIONS
-------
of a series of maxima the largest of -which lies at a frequency equal to E_ /h,
where h is Planck's constant, and E is, as indicated in the figure, the energy
50
of the level v1 = 5 relative to that for v11 =0. The frequency between the maxima
will be approximately AE'/h where AS1 is the vibrational energy constant of the
upper level. Since the diatomic molecule also has one rotational degree of freedom,
each of the maxima mentioned above will display rotational fine structure whose
~ r*J
character will depend on the nature of the electronic states X and A (Ref. 3).
In contrast to the simple situation depicted in Figure 2, the portrayal of the
vibrational potential energy of each electronic state of SO is much more compli-
cated. As shown in Figure 3» SO has three normal vibrations, and the vibrational
potential energy of any given electronic state is seen to be dependent on the sulphur -
oxgyen bond distances, as well as the angle between these bonds. The structure
r^t «-w
shown in Figure 1 for the X -» A transition has been attributed to transitions from
the lowest energy levels of the first two normal modes shown in Figure 3 to excited
levels of these modes in the A electronic state (Ref. U). In addition, it is noted
that SO has three rotational degrees of freedom, giving rise to a very dense and
**** l*Nrf
complicated rotational fine structure in the X -» A absorption spectrum which in
turn causes the observed width of several angstrom units in each of the maxima
shown in Figure 1.
Unquenched Fluorescent Emission
Examination of Figure 2(b) reveals how the frequency distribution of fluores-
cent emission is determined by the properties of the electronic molecular states
involved. Generally, photon emission occurs prior to vibrational relaxation of
the molecule in the excited state (Ref. 5)- Accordingly, for the case shown in
Figure 2, the molecule, once excited to v' = 5 by absorption of a photon of energy
E will most probably emit a photon with energy E or E , and transitions at
sU s'J Pi
energies of E ^ and E 0 will be only slightly less probable than that with energy
56 5o
E . Therefore, the frequency distribution of fluorescent intensity observed from
a sample of gas irradiated with photons of energy E_ will exhibit one maximum at
-7-
-------
(1,0,0)
(0,1,0)
1150.5CM
~1
'X~1 = 524.5 CM"1
(0,0,1)
X~1 =1336.0 CM-1
FIGURE 3. THE NORMAL VIBRATIONS OF SO2
-------
E /h, corresponding to resonance fluorescence, and other maxima centered around
E /h. Figure k illustrates the wavelength distribution for fluorescence from. SO
pressure at 0.01 torr, and the fluorescence intensity distribution exhibits prop-
erties qualitatively similar to those described above for the simple example shown
o
in Figure 2. The wavelengths for the two maxima observed below 3800 A are unshifted
as the SO pressure was increased from 0.01 torr to 0.5 torr, indicating that vib-
»KJ
rational relaxation of the A state prior to photon emission is unimportant and that
the observed fluorescent intensity distribution is determined by spectroscopic
^s
rather than gas kinetic properties of the SO molecule in the A state in this pres-
sure range (Ref. 5).
Quenched Fluorescent Emission
r^t i*-*
However, the absolute intensity of the A -* X fluorescence was observed to
decrease as the SO pressure was increased to 0.5 torr, and fluorescence from the
~ °
a state at wavelengths greater than 3900 A was observed as indicated in Figure 4.
These facts indicate that quenching of the A state by SO occurs, leaving the mole-
n*N^
cule in the a state, for SO pressures comparable to those found in coal-fired
power plant effluents. Other molecules are also effective in quenching both the
A and a states of SO . The quenching rate coefficients for some of these are given
(•*_/ <-w
in Table 1 along with radiative decay rates for the A and a state.
The results given in Table 1 reveal that the rate of quenching of SO at the
A" state is quite sensitive to the identity of the quenching molecule and raises
the question as to whether, for a fixed quantity of SO in a gaseous mixture, the
observed fluorescent signal is sensitive to the relative amounts of the other
components in the mixture. In the following discussion the expression is developed
for the magnitude of the observed signal from a fixed pressure of SO gas as a
function of the amount of quenching gas added to the sample. It is assumed that
the pressure of SO in the sample is low enough that the degree of attenuation
of the incident laser radiation is negligible. For the situation where the rate
of destruction of excited state SO by radiative decay and quenching collisions
-9-
-------
C/3
2
til
h-
lil
UJ
cc
1.0
0.5
0.0
SO2 EXCITED BY 3020A RADIATION
—— 0.5TORRSO2
— ~" — 0.01 TORR SO2
I
3000
3500 4000 4500
WAVELENGTH (ANGSTROMS)
FIGURE 4. WAVELENGTH DEPENDENCE OF FLUORESCENCE OF SO2 FOR THE
WAVELENGTH REGION BETWEEN 2900 A AND 5000 A
-10-
-------
TABLE 1. QUENCHING RATES OF SO2 ULTRAVIOLET RADIATION
(A ^X)
( RADIATIVE DECAY RATE = 2.4x104 SEC"1 )
QUENCHING MOLECULE
C02
NO
N2
CO
D20
°2
SO 2
QUENCH.NG RATE REFERENCE
(SEC"1 TORR-')
8.0x1 05 8
7.1 x106 6
5.5x1 06 6
2.7x1 06 6
3.1 x106 6
1.8x107 6
2.5x1 06 6
8.7x1 06 6
5.1 x106 8
-11-
-------
is fast compared to the temporal variation of the incident laser intensity, the
rate of production of excited state SO is balanced by the rate of decay of SO
from the excited state
0oaPS02 - N (? Vi + rr) ' °
where 0 is the incident laser flux, a is the SO absorption coefficient, assumed
0 *
to be unaffected by pressure, p the SO pressure, N the SO excited state
SOp 2 2
number density, k the quenching coefficient of component i at pressure p , and
r is the radiative decay rate of SO . The observed fluorescence intensity, I ,
r *
is given by the product of K, the experimental detection efficiency, and hvfN r
where hv is the energy per photon of the fluorescence. The ratio of the fluores-
cence intensity, I , observed in the absence of all quenching gases other than SO .
O 2
to that observed vhen a quenching gas is present at pressure p is given by
q
V PSO? + Vfl + "r
I/I - — -- - - (2)
and a plot of I /I is linear with p with a slope m equal to
o q
m =
If the quantities k and r are knovn, the quantity k can be obtained from a
SOp r 9
plot of I /I vs p . Conversely, if all the quenching coefficients k are known,
. o
it is possible to calculate the sensitivity of I/I , the fluorescent efficiency
or fraction of absorbed radiation emitted as fluorescence,
o rr + E k p
o i
to the composition of the gas mixture.
-12-
-------
Using the information summarized in Table 1, the calculated value of I/I is
-6 °
6.2 x 10 for a trace of SO in a one atmosphere mixture containing 77$ N , 12$
CO , 8$ HO and 3$ 0 , a mixture typical of the exhaust from an oil flame. In
contrast, the fluorescence efficiency is decreased by 30$ from the above amount
if the water fraction is increased to 20$ and the fractions of the other components
relative to one another are held constant. Thus, the detected SO fluorescence
' 2
signal can be significantly influenced by the composition of the mixture sampled.
In addition it is noted that the results presented in Table I are for a mixture
temperature of 300 K. Whether the fluorescent efficiency of a given gas mixture
is sensitive to gas temperature can therefore not be determined from the data in
Table I and remains a serious question as to the validity of fluorescent emission
as a means of detecting SO .
C. Raman Scattering
In the previous section the process of fluorescent scattering of radiation
was briefly reviewed. It was seen that fluorescent scattering is essentially a
tvo-photon process, involving first the excitation of the molecule from the ground
to the excited state by absorption of a photon from the incident radiation field,
folloved, after a time interval determined by the radiative characteristics of the
excited state, by photon emission, leaving the molecule in a lover energy state.
It vas seen that due to the finite time interval spent in the excited state, col-
lisional quenching becomes a significant alternate to photon emission as a decay
channel from the excited state and thereby compromises the validity of fluorescent
scattering as an accurate measure of SO concentration. In this section it will
be shown briefly that in contrast to fluorescent scattering, the Raman effect occurs
only while the molecule is in the electric field of the incident photon, thereby
being a one-photon process and free of quenching effects. It will also be indi-
cated how the magnitude of the Raman scattered signal can become greatly enhanced
if the incident radiation wavelength coincides with that of a maximum in the
absorption spectrum of the molecule.
-13-
-------
Nature of the Eaman Effect
The nature of the Faman effect can be simply understood by considering the
oscillating electric field associated with a photon flux- incident on a molecule
and the perturbing effect that this field has upon the vave functions associated
•with molecule energy levels (Ref. 7). As a result of the interaction between the
electric field and the molecule, photons are scattered inelastically out of the
incident flux and their frequency of oscillation either decreased, if energy is
transferred to the molecule, or increased if energy is transferred from the mole-
cule to the field. The intensity of radiation scattered per unit beam path, per
molecule at frequency v is given in terms of the incident intensity, I , by the
s ' o
following expression
128 n5 I v k
where
M , ,, (B'M , , ,) (E-M , ,, ) M ,,
n'n" v n'n'' n'n" n"n
n'n ni t v ,, - v v ,, , + v
n n"n n"n'
where v is the frequency of the incident radiation, n and n1 represent the initial
and final states of the molecule, n1' represents states other than the initial and
A A
final ones, and E, and E are, respectively, unit vectors for the polarization state
a1 and the incident electric field. In the case of interest here the states
denoted by n and n1 are the lowest and first excited vibrational levels, respectively,
of the ground electronic state of the molecule in which case v is given by v-E /h
S *
where E is the vibrational energy difference between the states n and n1. The
states n" are the vibrational levels of the excited electronic states of the mole-
cule, the quantity M ,, is the electric dipole operator matrix element and hv ,,
is the energy difference between the levels n'' and n. Prom the above expression
it is seen that the fraction of the beam scattered with frequency v is related
s
to the sum of the strengths of all optical transitions involving the states n1 and n.
-------
In particular, when the value of v Is nearly identical to that of an optically
allowed transition the phenomenon is referred to as resonance Raman scattering
and the scattered intensity is given by
128 n5 iv k |M , ,,|2 |M ,, |2
_ o s ' n'n' ' ' ' n'n/ ....
•
where d . , is the line width of the allowed absorption line. Behringer (Ref, 9)
n' '
points out the near equality between the above expression and that for the intensity
of radiation absorbed in an allowed optical transition provided that d is deter-
n1 '
mined by the radiative decay rate from the level designated by n'1. For this situ-
ation the magnitude of I would equal the fluorescent intensity in the absence
res
of quenching collisions in the hypothetical case where only resonant fluorescence
could occur. In .practice I is not so large as predicted due most probably to
res
the fact that d is determined by factors other than the radiative decay rate
n1 '
and to the breakdown in the validity of the time independent perturbation theory
which was used to derive Equation 6 (Ref. 10). Still, resonance Raman scattering
has been observed to be quite intense in liquid solution (Ref. ll) as shown in
Figure 5. The large number of overtone and combinational lines is characteristic
of resonant Raman scattering. The sharpness of the resonant Raman lines in compari-
son to the rather broad fluorescence is a result of the fact that the Raman effect
is a direct scattering process and involves no intermediate energy level from which
the absorbed energy can be quenched or reemitted over a wide wavelength interval.
D. Analysis of Scattering Containing Both Raman and Fluorescence Components
In the experiments to be described in Section III, the magnitude of the incident
laser signal scattered by a gas sample containing SO and N was measured as a
function of the amount of N in the sample. Measurements were taken of the scat-
tered signal intensity at a light frequency which was displaced from that of the
incident laser radiation by an amount corresponding to the energy necessary to
excite the symmetric stretch vibrational mode of SO , 1151 cm~ . The measured
-15-
-------
RESONANCE RAMAN SCATTERING FROM 4-NITRO-4-
DIMETHYLAMINOST1BLE IN BENZENE [AFTER (11)]
55
m
A
' N
RR
I
24,000 22,000 20,000 18,000 16,000
v ( CM"1 )
RESONANCE RAMAN SCATTERING FROM
DIPHENLYDECAPENTENE IN ACETONE [AFTER (11)]
LU
H
2
4000
5000
\
6000
7000
RESONANCE RAMAN SCATTERING FROM
DIPHENYLDODECAHEXENE IN ACETONE [AFTER (11)]
CO
LU
RR
4000
5000
6000
7000
X
FIGURE 5. RESONANT RAMAN SCATTERING AND FLUORESCENCE
IN LIQUID SOLUTIONS
-16-
-------
signal is therefore expected to contain both a fluorescent component which, as
indicated in Section IIB, is quenched by the addition of N and a Raman component
whose intensity is unaltered by N addition. In the following paragraph the expected
pressure variation of such a signal is discussed, and it is shown that the observed
pressure variation indicates the relative magnitudes of the fluorescent and Raman
components of the signal.
Pressure Variation of Scattered Signal
In Section IIB it was shown that the magnitude of the scattered intensity
due to fluorescent scattering decreases upon the addition of a gas which quenches
the fluorescence, and in Section IIC the immunity of the Raman effect to quenching
effects was described. Therefore, the pressure dependence of the observed scat-
tered intensity I, which contains both fluorescence and Raman components, is given
by the following expression
I =
l+XPq
where L, is the magnitude of the fluorescence component in the absence of the
quenching gas, p is the quenching gas pressure, ]L is the magnitude of the Raman
q R
component, and X, the quenching coefficient, is given in terms of quantities defined
in Section IIB.
k
9
If the Raman scattering component is negligible for all values of p , the quenching
coefficient can be calculated from experimentally measured quantities directly from
Equation 7 with T set equal to zero, and the result is
X=-^r— /P_. (9)
-17-
-------
If T is indeed zero, X is of course independent of p . If, however, 3L is not
negligible, the experimentally derived value of X^exhibits a pressure dependence
and is no longer defined by the simple expression given .in Eq. 8. The pressure
dependence of the apparent quenching coefficient X , derived as above under the
ct
assumption that I is zero when in fact it is not, is given by the following
expression.
.. if + L 3L p
X -1 = -£ - -£ + iJL (10)
a
It is seen that a plot of X against p will yield a straight line whose slope
a q
is equal to the ratio of the Raman component of the scattered intensity to that
of the fluorescence component unquenched by foreign gas addition. The value of
the slope of the curve can then be combined with the value of X at zero quenching
a
pressure to yield the value of X which in turn is used to obtain k if k , p
q SO,-, SOp
and r are known. The magnitude of the Raman component relative to that of the
r
fluorescence, I, at any value of p is given by
F
(ii>
If the value of Ip/IL is large compared to unity at atmospheric pressure, then
the scattered light intensity is made up mostly of the Raman component, is free
from quenching effects and is thus a mixture composition and perhaps temperature
independent indicator of the SO concentration in the mixture.
-18-
-------
SECTION III Experimental Results
A. Apparatus
Radiation Source and Optical Train
The laser used in these experiments was the UARL vortex stabilized flashlamp
pumped dye laser vhich has been described in detail elsewhere (Ref. 12). The dye
used in these experiments vas Rhodamine 6G at a concentration of iko p, mole/liter
and the broadband output of the laser -with a fresh dye solution was typically 150
mj contained in a pulse with a half-width at half-height of 0.75 M-sec. Tuning of
the device was accomplished by insertion into the cavity of two etalons. One of
these etalons was a piece of quartz, one of whose surfaces was coated with two
highly reflecting layers separated by k ^m of dielectric. Its free spectral range
o o
was 9 A at a wavelength of 6000 A. With these etalons in place, the laser output
could be confined to a single line several tenths of an anstrom unit wide over a
o o
wavelength range which extended from 5800 A to 6150 A. The laser output was
focussed onto a crystal of ADP immersed in cylohexane to protect against desication
of the crystal. The diverging beam, a portion of which was now at one half the
wavelength of the original laser output, was refocussed in the middle of the fluo-
o
rescence cell, the temperature of which could be varied between ambient and 200 C.
after passing through the cell the beam was incident on an RCA 935 photodiode
o
equipped with a low pass filter with a cut off wavelength of ^100 A. The output
of the photodiode was amplified and used to trigger the electronics which will be
described later. The entire optical train is drawn schematically in Figure 6,
and a detail of the flashlamp-dye cell unit is shown in Figure 7 (Ref. 12). The
light scattered by the SO in the gas cell was focussed by a quartz lens onto the
entrance slit of a Jarrell Ash 75-150 instrument which could be used either as a
o o
monochrometer or a spectrograph with a dispersion of about 8 A/mm at 3000 A. The
image of the scattering region was rotated by placing an appropriate mirror assembly
between the quartz lens and the slit. This was done in order to render the image
-19-
-------
i
LENS
ADP
CRYSTAL
QUARTZ
LENS
TURNING
PRISM
GAS CELL
OUTPUT
MIRROR
TUNING
ETALONS
FLASHLAMP
DYE CELL
ASSEMBLY
LASER OUTPUT
AT 6000°A
FRACTION OF
OUTPUT CONVERTED
TO 3000 A
RCA
935
ELECTRONICS
SCATTERED
RADIATION
QUARTZ
LENS
MONOCHROMATOR
FIGURE 6. OPTICAL TRAIN USED IN SO2 LIGHT SCATTERING MEASUREMENTS
-------
OUTPUT MIRROR
DYE
SOLUTION
DYE
SOLUTION
DYE
SOLUTION
STAINLESS
STEEL
TUBING
GROUND
RETURN
SUPPORT FIN
QUARTZ
LAMP
ENVELOPE
DYE
SOLUTION
COAXIAL CAPACITOR
GROUNDED TERMINAL
CO AXIAL CAPACITOR
HV TERMINAL
FIGURE 7.1.2-W FLASHlAMP PUMPED DYE LASER
-21-
-------
and the slit parallel, thus Increasing the magnitude of the scattering volume
accepted by the monochrometer compared to the volume accepted were the image and
slit perpendicular to one another. During fluorescence measurements, the monochro-
meter entrance and exit slits were opened to 3 mm and the throughput of the instru-
o
ment thus lay over a spectral region 2*4- A vide. This throughput impinged on an
RCA 85T5 photomultiplier equipped with a low-pass filter with a cut-off wavelength
o
of 4100 A. The photomultiplier output vas analyzed in a manner to be described
later.
To analyze the spectral properties of the laser output, the latter was deflected
onto a strip of scotch tape placed in front of the spectrograph entrance slit, the
tape acted as a radiation diffuser and provided uniform illumination of the slit
whose width was set to about 0.05 nun for this measurement. The spectrograph through-
put was collected on Polaroid Type 52 film and the spectrum from a low pressure mercury
vapor germicidal lamp was superimposed on the photograph of the laser output spectrum.
The film containing the superimposed spectra vas analyzed using a microscope equipped
with a movable stage whose position was measureable to within 0.0001 inch. Under
magnification, spectral lines appeared as nearly gradient-free grey areas on the
photographs. The center wavelength for a given line was associated with the center
of the corresponding grey area. The effective dispersion for each spectrogram vas
o
calculated from the measured separation between the mercury lines at 2967.28 A and
o
3021.50 A, and the value of this quantity was reproducable to about 0.5$. The
chief cause of error in such a measurement is the contribution to the measured
o
width of the mercury line at 3021.50 A by intensity from a neighboring but less
o
intense, by a factor of 3, line at 3023.^8 A. The wavelength of the laser line
vas then calculated by combining the measured dispersion of the spectrogram vith
o
the measured distance between the laser line and the line at 3021.50 A. The error
o
in the position of the line at 3021.5 A is thus seen to affect both the calculated
dispersion and the measured position of the laser line wavelength relative to the
o
3021.5 A mercury line. Routine calculation indicates that the calculated value of
o
the laser line wavelength is at worst 0.5 A too high as a result of this source of
-22-
-------
error. The half width of the frequency doubled laser line at one percent of the
o
peak intensity was estimated by these measurements to be about 0.^8 A, which
o
infers a half-width at half-height of 0.22 A if the line profile is Gaussian.
Electronics
In practice, the intensity of the scattered laser radiation was measured by
averaging, over a number of laser pulses, the magnitude of scattered light signal
received by the photomultiplier attached to the monochrometer. The following para-
graph describes the electronic apparatus employed in obtaining this average.
The electronic instrumentation used in processing the photomultiplier output
current consisted of a Burr-Brown SHM 40 sample and hold circuit, a Dymec 2^01 A
integrating digital voltmeter, which was used as a voltage-to-frequency converter,
and a Hewlett Packard 5280 A reversible counter connected as shown in Figure 8.
In principle, the sample and hold circuit attains the voltage at its input during
the 2 p,sec aperture time which is defined by the trigger signal generator which
in turn is triggered by the laser pulse. The output of the sample and hold circuit
is maintained at the voltage reached by the end of the 2 usec aperture period, and
this voltage is fed to the voltage-to-frequency converter, the output of which is
an oscillating voltage whose frequency is proportional to the voltage input. During
its 20 msec aperture time, the counter registers the number of times that this
oscillating voltage passes through zero and displays this number as its output.
The output of the counter is summed over 100 laser pulses and is taken as the mea-
sure of the average magnitude of the laser radiation scattered by the SO in the
sample cell into the wavelength range as defined by the monochrometer. Typically
the maximum output of the photomultiplier was of the order of a volt when no
quenching gas was present in the cell and decreased to several millivolts when the
quenching gas pressure was one atmosphere. In practice, the sample and hold circuit
functions best at voltages of more than 0.1 V, and so provision was made for the
insertion of preamplifiers between the photomultiplier tube and the sample and
hold circuit. The linearity of the entire electronics assembly was checked and
found to be satisfactory.
-23-
-------
RCA 8575
PREAMPLIFIER
SAMPLE AND
HOLD C
IRCUIT
VOLTAGE TO FREQUENCY
CONVERTER
COUNTER
RCA 935
2 MSEC
GATE
20 /XSEC
GATE
TRIGGER GENERATOR
FIGURE 8. ELECTRONICS USED IN SO2 LIGHT SCATTERING EXPERIMENTS
-------
The maximum photomultiplier anode current vas typically less than 100 na,
and sufficient capacitance was added between the last four dynodes to guarantee
against any nonlinearity of photomultiplier response to the scattered light inten-
sity. After extensive efforts had succeeded in eliminating extraneous electrical
noise generated by the flashlamp and its attendant circuitry, it was found that
the pulses constituting the photomultiplier dark current were the main source of
background noise and the photomultiplier was operated at 2300 V cathode voltage
at which the signal to dark current noise was found to be maximum. The photomulti-
plier was not cooled although in practice doing so would further lower the dark
current.
Gas Handling System
A schematic diagram of the gas handling system used in these experiments is
shown in Figure 9. Using the mechanical pump, a base vacuum of 0.02 torr was
obtained as indicated by the Hastings-Raydiest OV6M thermocouple gauge. The pres-
sure of SO added to the cell was measured using a Wallace and Tiernan Mechanical
Gauge, labeled #1, in Figure 9, and this pressure was typically less than 0.5 torr.
The pressure of N added to the cell was read using the mechanical gauges. In
practice, after sufficient SO had been added to the cell, valves 1 and 5 were
closed and valves 2 and 3 opened, a vacuum of about 0.03 torr attained, valves
2 and 3 closed and 1 reopened. This procedure was carried out to guarantee against
leakage of additional SO into the scattering cell during the course of an experi-
ment.
B. Experimental Procedure and Results
Procedure
The procedure followed in obtaining data was as follows: with the tuning
etalons adjusted so as to attain approximately the desired laser output wavelength,
6000 A, the laser output was maximized by adjusting the orientation and position,
relative to the first lens, of the ADP crystal. As a result the frequency doubled
o
output from the crystal was now centered at approximately 3000 A, nearly coincident
-25-
-------
SCATTERING
CELL
MECHANICAL
GAUGE #1
oil-20 TORR
LIQUID
NITROGEN
COLD
TRAP
MECHANICAL
GAUGE #2
10-760 TORR
THERMOCOUPLE
VACUUM
GAUGE
-•-TO MECHANICAL PUMP
FIGURE 9. GAS HANDLING SYSTEM USED IN SO2 LIGHT SCATTERING EXPERIMENTS
-------
with the wavelength of the G line in the X -» A absorption system of SO . Spectral
o
scans of the intensity of the radiation scattered from an incident beam at 3000 A
o
disclosed a sharp maximum at approximately 3105 A followed by a much broader aiaxi-
o
mum at 34-00 A. The Stokes component for the symmetric stretch mode of SOO appears
.o o
at 3105 A for Raman scattered radiation whose original wavelength is 3000 A.
o
Accordingly, the monochrometer vas set at 3105 A with the exit and entrance slits
o
opened so as to permit radiation 12 A to either side of this wavelength to pass
through the instrument and impinge on the photomultiplier. Sulphur dioxide was
admitted to the cell to a pressure between 0.3 and 0.5 torr and the wavelength
of radiation incident on the cell was tuned by suitably adjusting the intracavity
etalons and the ADP crystal orientation. The degree of coincidence between the
wavelength of the incident radiation and the maximum in the SO G line absorption
coefficient was judged by the magnitude of the photomultiplier output signal. Runs
were taken with varying degrees of coincidence. With the tuning of the laser line
completed, measurements were made of the average scattered radiation signal, as
well as the average background signal. Both measurements were made for one hundred
laser pulses. The average laser pulse output signal was also measured in the same
way by measuring the output of the RCA 935 photodiode. The scattered intensity at
o
3105 + 12 A, I , normalized with respect to the laser output is given by
.
Jj
where I1, I and I are the measured scattered, background and laser signal, res-
o B L
pectively. Nitrogen quenching gas was added to the cell to the desired pressure,
the above process was repeated after a fifteen minute pause to allow complete
mixing of the N and SO , and the expression in Eq. 12 was used to calculate the
value of I to be used in Eq. 9. Generally, a complete experimental run consisted
of measurements at two or more N pressures, as well as one at zero N pressure.
At the end of a run, the wavelength of the laser radiation was determined in the
manner described in Section IIIB.
-27-
-------
I
ro
o
UJ
I
cc
DC
O
CO
o
X
fa
200
400
(TORR)
600
800
FIGURE 10. PRESSURE DEPENDENCE OF THE APPARENT RATE COEFFICIENT FOR QUENCHING
OF SO2FLUORESCENCE AT 3105 A BY N2
-------
The value of k was evaluated at each pressure using Equations 8 and 9 and
N2
the rate coefficients given in Table 1. All of such calculated values of k are
N2
shovn in Fig. 10 as a function of the added nitrogen pressure. That k is pres-
-1 2
sure dependent is immediately apparent and X is plotted against p in Fig. 11
a Np
for those runs in which results were obtained at more than two pressures. Assuming
that Fig. 11 indicates the validity of the model from which Eq. 11 was derived,
these results show that the ratio of the Raman scattered component intensity to
that of the zero N pressure fluorescence scattered component to be independent
o ° , °
of the laser line wavelength in the region between 2997.8 A and 3004 A. The value
of this ratio is seen to be 2.k + 0.9 x 10 and the value of k . calculated from
— N2 _g _1
the zero pressure intercepts of these curves, is found to be 3.8 + 0.6 x 10~ sec
torr , which is to be compared to Metee's value of 2.7 x 10 sec torr . Mea-
surements were also taken with the fluorescence cell heated to 200 C as measured
by a chrome1-Alumel thermocouple. As at room temperature, the quenching coefficient
calculated by Eq. 9 exhibited a dependence on the nitrogen pressure. However, the
scatter in the data vas too great to draw any useful information from it (Ref. 13).
Discussion
The results obtained above can be used to calculate an upper bound for the
magnitude of the Raman scattering cross-section for radiation initially of wave-
o o
length 3000 A scattered by SO at wavelength 3105 A, From the absorption coefficient
2 o
data given in Fig. 1, the absorption cross section at 3000 A, a , is calculated to
-19 2 S
be 5.9 x 10 cm . Were all absorbed energy to be re-emitted unquenched as fluo-
o
rescence at 3105 A, the cross section per steradian for fluorescent scattering at
o _PQ 2 -1
3105 A would be U.7 x 10 cm sr . As indicated by Fig. 2, only a fraction of
the energy is emitted at this wavelength, and this cross section must therefore
be considered an upper bound to the true value. The experimentally measured ratio
o _o
of Raman scattering to unquenched fluorescence at 3105 A was 2.U + 0.9 x 10 .
Proceeding as in Section IHJ, the value of this ratio can be expressed as follows
-29-
-------
10
DC
DC
O
I 4
CO
X
A,, = 2997.8
A,, = 3004.0
Aj, = 2999.0
A,, = 3000.0
200
400
(TORR)
600
800
FIGURE 11. PRESSURE AND LASER WAVELENGTH DEPENDENCE OF Xa~1
-------
- *—*- (13)
a r
Where o_ is the Raman scattering cross section, and the other quantities have been
R
defined previously. For the data presented in Fig. 11, the value of p was
typically OA torr, and the calculated value of CT is therefore seen to be 7.7 x
-25 2-1
10 cm Sr . The magnitude of this cross section can be calculated, ignoring
all resonance effects, from the cross section measured for argon ion laser radiation
o k
at 5145 A (Ref. 1*0 by simply assuming that the cross section increases as v
-29 2 -1
The result of this calculation is 1.2 x 10 cm sr , inferring a factor of about
6 x 10 as the upper bound for the degree of enhancement of the cross section as
a result of resonance effects. As mentioned in Section IIC, the magnitude of
enhancement can be as high as 10 , and, considering the weakness of the assumption
concerning the size of the fluorescence scattering cross section, the magnitude
of the enhancement inferred from the measurement is not unreasonable.
The impact of the experimental results upon the feasibility of using resonance
scattering for the remote detection of SO in stationary source emissions is con-
siderable. The experimental results indicate that, in the absence of foreign gas
o
quenching effects, a fraction of the incident laser intensity at 3000 A
o
is scattered by the Raman effect at 3105 A. For Q.k torr SO , the size of the
i-3 2 o
fraction is 2.14- x 10 when the measurements are made over a 2k A wide interval
and is expected to increase as this interval size is decreased. In addition, the
size of this fraction exhibits negligible dependence on the incident laser wave-
o
length over a six angstrom unit range about the SO absorption line at 3000 A.
The size of this fraction increases with pressure as the fluorescence component
becomes quenched. The ratio of Raman signal to fluorescence is about 1.8 at
atmospheric pressure, causing the scattered signal to be nearly three times as
large as that expected in the absence of Raman scattering. As shown in Section
IIA, an error as large as 30$ in the measured scattered intensity can occur because
-31-
-------
of composition variation in the gas mixture being analyzed if the intensity is
due completely to fluorescence. The presence of the Raman component reduces this
error to about 11$, and, as mentioned above, this error .is decreased still further
as the wavelength interval of the measurement is decreased. Therefore, the presence
of the Raman component significantly increases the magnitude of the scattered signal
and also its validity as a measure of the SO concentration in the gas being analyzed.
Several questions do remain with respect to the validity of the light scattering
technique as an SO concentration probe. Although some measurements were taken with
the gas sample at 200 C, no information was gained with respect to the effect of
"temperature upon the Raman to unquenched fluorescence ratio. Temperature effects,
if any, are likely to arise from the temperature dependence of the quantity d f,
in Eq. 6, and these effects should be investigated. Furthermore, the problems con-
cerning the signal scattered inelastically by soot, fly ash, or other components
present in a smoke stack plume have not been addressed in this study.
-32-
-------
Section IV Conclusions and Recommendations
A. Introduction
In this section the experimental results described in Section III are
utilized to evaluate the feasibility of using resonance scattering at 3000A
from SO for the remote detection of that molecule. In particular, the following
subjects are addressed: 1) The experimental results are utilized to calculate
E , the laser pulse energy necessary to make possible the detection of X parts
•v
per million SO in a volume J, cm long at a distance L km from the laser-detector
unit, and it is found that a laser delivering one mJ at 3000$ in a pulse one
microsecond long is sufficient to allow detection of 100 ppm SO distributed
over a meter vide stack from a distance of 0.9 km. This pulse energy is veil
vithin present laser technology. 2) The extent to -which quenching of ordinary
fluorescent scattering degrades the accuracy of the measurement is revieved
and laboratory scale programs auggested to ascertain vhether under field
conditions the magnitude of the Raman signal is so much larger than the
fluorescence signal that fluorescence quenching in no vay degrades the accuracy
of the measurement. 3) By considering the solar radiance present at 3000A
it is found that the interference to the measurement caused by scattered
sunlight is negligible, k) Some recently available results concerning
fluorescent and Raman scattering by particulates are discussed, and these
results are found to infer that such scattering may be a limitation to the
usefulness of resonance scattering as an SO concentration probe, thereby
giving rise to the need for field tests in vhich the degree of scattering
at 3000A from plumes free of SO is determined.
B. Lasers For 50^. Detection
£HB^^^^-«^^«WW
The results reported above may be utilized to calculate E the laser pulse
\f
energy necessary to make possible the detection of X parts per million SO
in a volume t. cm long at a distance L km from the laser detector unit. This
-33-
-------
quantity is expressed simply for suitably short pulses of width T and low
\>
background signal level by the following equation which expresses the condition
of equivalence between the magnitude of the laser power scattered by the SO
sample and returned incident to the detector and the detector's noise- equivalent
power, P , the power incident on the detector necessary for unity signal-to-
noise ratio:
-V T \\ V \ V (1V + V w
V
For the phot ©multiplier, detector and laser pulse width used in this study
-12
the value of P is calculated to be 3.5 x 10 W(Ref. 15). In Eq. 14,
nep
the quantity n is the number of pulses involved in the measurement, and the
various 71 's are efficiencies and are defined as follows: The quantities
Up and Up' are the fractions of the incident laser radiation and the detected
scattered radiation respectively transmitted through the atmosphere and are
given by exp (-Q? L) and exp (-« L) where a. and a are the atmospheric attenuation,
coefficients at the incident and scattered wavelengths (Ref. 16). The ratio
of the scattered detected photon energy to that of the incident photon is given
by 7| . The fraction of the k rr,. tteradian solid angle about the irradiated
sample which is subtended by the detector is given by the ratio of the radiation
2 2
collector area, assumed to be 0.1 m , to k TT L and is represented in equation
Ik by Tl • The transmittance of the detector and collecting optics is given
'c
by Tl j. and assumed to be 50$ in the results given below. If the irradiated
'opt
sample is not optically thick at the laser wave length, the absorption and
fluorescent scattering efficiency is given by
(15)
where N is the gas density in the sample, a the SO. absorption cross section
a d
7) the fluorescent efficiency given by equation k, and T^ "the fraction of the
FE X
fluorescence which falls within the wavelength limits of the detector, assumed
to be 2$ here. Finally TL,,"fche Raman scattering efficiency, is given by
-------
vhere CT Is the Raman scattering cross section and it is assumed that the
R
wavelength limits of the detector are wider than the laser line-width. Figure
12 presents E as a function of L and the product of X and £ for a measurement
^ o
involving 100 laser pulses with fi calculated using the 300 K data given
in Table 1. It is seen that a laser pulse of one raj at 3000A is sufficient
to allow detection of 100 ppra distributed over a one meter wide stack at a
distance of 0.9 km. For a pulsed dye laser, a second harmonic conversion
efficiency of as high as 10 percent (Ref. 17) has been obtained, and between
5700A5 and 630oX more than a Joule of dye laser output can be achieved (Ref. 18)
making the above-mentioned millijoule of laser output at 3000A within
the state of the art in tunable laser sources.
C. The Problem of Quenching
In Section II B the effect that fluorescence quenching has on the accuracy
of the determination of SO concentration from detected fluorescent scattering
was considered in some detail. It was found that an error as large as 30$
was generated by stack-to-stack variation in the concentration of effluents
other than SO . In addition it was pointed out that the SO fluorescent
quenching rate coefficients can be expected to depend on the temperature so
that stack-to-stack variation in the latter quantity provides another source
of error. In Section II C, the properties of the Raman effect were discussed,
one of these properties being immunity to quenching, and the experimental
results of Section III revealed the presence of a large resonance Raman
scattering component capable of reducing the above mentioned sensitivity to
gas composition by at least a factor of three. This sensitivity is obviously
proportional to the ratio of the fluorescence scattering component to the
Raman component,
CTa ^FE VaR (LT)
-35-
-------
10-2
10-3
10-4
10~5
10-6
PPM-CM SO2
103 104 105
I I I
10-2
10
1-1
100
101
L (km)
FIGURE 12. LASER PULSE ENERGY NEEDED TO DETECT SO2 AS FUNCTION
OF DISTANCE L FROM SAMPLE
-36-
-------
and is smallest under conditions -where the fluorescent efficiency is smallest
or the Raman scattering cross s-ection is largest. If it were known that, for
the temperatures typically found in stationary sources, the fluorescent efficiency
vere substantially smaller than the value calculated from the 300 K data
shown in Table I, the composition and temperature sensitivity described above
may well be negligible. On the other hand, if the Raman scattering cross
section can be increased substantially by, for instance, tuning the wavelength
of the incident laser radiation to a different maximum in the SO absorption
spectrum this sensitivity may again by made negligible. For this reason,
the temperature sensitivity of TL, and the wavelength sensitivity of cr
are the subjects of discussion of this section.
As shown by Eq. k, the value of -n is determined by the magnitude of
r Ci
the fluorescence quenching rate coefficients k each of which can be thought
of as the average value of the product of the relative speed v between the
SO molecule and the quenching species and the quenching cross section a
^ 81
for that species, the average being taken over all values of the relative
speed. At higher temperature the probability of collisions involving larger
values of vr increases so that, were a to be independant of v , the value of
k would be expected to increase as the square root of the temperature
(Ref. 19). However such behavior is never observed as the value of cr
q
often exhibits a quite complicated dependence on v . As an example, consider
the quenching of k.$ ^m emission from vibrations excited CO by N (Ref. 20).
The quenching is accomplished by either of two mechanisms. The first involves
the transfer of vibrational energy from CO to N , and the second involves
the redistribution of vibrational energy within the CO molecule as a result
of the collision with the Ng molecule. The simple treatment described
above predicts a 40$ increase in the quenching rate coefficient as the gas
temperature is increased from 300°K to 600°K. In practice, however, the ^.3 ^m
quenching rate is observed to increase by order of magnitude for the latter
of the two mechanisms described above and to decrease by 30$ for the former.
There is no reason to believe that the temperature behavior of the quenching
-37-
-------
rate for electronically excited SO is any less complicated, a fact which points
out the necessity of actually measuring these rates rather than estimating
them from simple physical arguments.
As discussed above, an alternative to determining the magnitude of 71
at elevated temperatures is to investigate whether o is substantially
R
larger at some maximum in the SO absorption spectrum other than the G line
on which measurements vere made in Section III. As was the case for calculating
the temperature dependence, making an a priori estimate of a and its
R
variation from one SO absorption line to another is extremely difficult.
As indicated from equation 6, such a calculation requires detailed information
of the shape of the potential energy curves for the upper and lower electronic
levels, and this information is itself not easily obtained. Instead, the value
of a must be obtained experimentally for several S0_ absorption lines in
R 2
order to ascertain whether its magnitude varies significantly from one line
to another. In addition, reference to figure 5 reveals that in the resonant
Raman effect scattering often occurs at frequency displacements other than
that corresponding to the ordinary Raman effect. Thus, the most comprehensive
laboratory measurement involves determining Ita/fL at aeveral frequency
displacements from the incident laser frequency as well as several incident
laser frequencies.
D. The Problem of Background Radiation
In practice, determination of the SO concentration in stack plumes will
occur with the sky being present as a background to the plume and thus
acting as the main source of background radiation to the detector. The
properties of the sky as a radiance source are presented by Platt (Ref. 21)
where it is seen that the radiance, N of the sky on a day of excellent
-62 o
visibility is about 10" W/(cm p,m steradian) at 3100A. In a manner similar
to that used in Section IV B, the background power incident on the photo-
cathode of the detector can be calculated from the expression
-38-
-------
PB= VT Vn'eV AX
where AX is the active wavelength interval of the detection system assumed
-k
here to be 10 ^m. The quantity n' A ' is the product of the solid angle
c p
subtended by the radiation collecting optic .about the volume being analyzed,
i
n , and the area of the image of the active detector surface projected on
that volume A '. This product is equal to the product n A where n is the
P c p c
solid angle subtended by the collecting optic about the detector and is, for
an f/10 optic, equal to 7-8 x lo"3 steradian, and A is the active detector
2 p
surface, assumed to be one cm . The value of 71 .is taken, as before, to be
'opt
0.5 and T| "to be unity, rendering the calculated value of P an upper bound
-13
on the actual value. The calculated value of P is 3.9 x 10 watt which
B
is an order of magnitude below the value of P so that background
nep
radiation is not expected to be a problem at 3000A.
E. Scattering From Particulates
It is unrealistic to expect that a typical smoke-stack under analysis
will contain only molecular species. In practice it will also contain fly ash
or other particulate species, and the question arises as to whether these
particulates will scatter the incident laser radiation in such a way as to
interfere with the determination of the SO concentration. The light
scattering properties of partLculates found in stationary sources is the
subject of a recently published report (Ref. 22) which considers both Raman
and fluorescent scattering. Of the many particulates considered, only CaSO,
_1 ^
displayed a frequency displacement, 1151 cm , exactly equal to that of SO
although many other sulfates had displacements of the approximately the same
size. Of perhaps greater interest was the finding of considerable broadband
fluorescence from a number of particulate materials. This result indicates
that scattering from particulates may indeed be an interference to the deter-
mination of the S02 concentration, but lack of detailed quantitative information
-39-
-------
in reference 22 with respect to the magnitude of the fluorescent scattering
cross sections prevents estimate of the magnitude of the problem.
F. Field Evaluation
The purpose here is to recommend a program for field evaluation of the
resonant scattering technique at 30008 as an SO concentration probe. In
Section IV C, it was pointed out that laboratory evaluation of this technique
is still incomplete inasmuch as the temperature dependence of the fluorescent
efficiency and the wavelength dependence of the resonant Raman cross section
are yet to be determined. However, the results discussed in Sections IV B and
E certainly indicate that the magnitude of the scattered signal is not a
problem to the measurement but that scattering from particulates may be of
sufficient importance to render any information obtained from further
laboratory evaluation irrelevant. Therefore the most effective program
of further investigation would seem to be actual field testing of this
technique using a stationary source capable of emitting plumes whose composition
can be controlled over a suitably wide spectrum. In such a program, the first
step is to construct a laser and detection system capable of performing as
predicted in Section IV B. Of particular importance in doing so is to ensure
that the laser output be centered on the SO absorption line and that the
_1
detector sensitivity by maximum at a wavelength exactly 1150 cm displaced
from the laser line. As the SO G line is at 3000$ this may be accomplished
by placing a narrowpass filter centered at 6000A in front of the laser, and
tuning the laser to maximize the power transmitted by this filter. A narrow-
pass filter centered at 3^05A would then ensure that the laser line center
and detector sensitivity maximum differ by the desired amount. The actual
field evaluation would involve first a series of tests at relatively short
range in which the SO concentration inferred from the scattering technique
is compared with that measured by a standard technique for a wide range of
effluent composition and temperature with special attention being paid to
-------
the effect of the presence of particulates on the measurement. A second
series of tests at longer ranges vould then serve to verify the results of
Sections IV B and E concerning the magnitude of the background signal and
the dependance of the scattered signal upon the distance from the source.
-in-
-------
REFERENCES
1. S. J. Strickler and D. B. Howell, J. Chem. Phys., 49 1947 (1968).
2. G. Herzberg: Molecular Spectra and Molecular Structure, (D. Van Nostrand
Company, Inc., Princeton, New Jersey, 1950), Vol. I, pp. 194-202,
3. Ibid., pp. 212-280.
4. G. Herzberg, Op. Cit., Vol. Ill, pp. 511-512.
5. H. D. Metee, J. Chem. Phys., 49 178^ (196^).
6. H. D. Metee, J. Phys. Chem., 73 1071 (1969).
7. H. Eyring, J. Walter, and G. E. Kimball, Quantum Chemistry, (John Wiley and
Sons, Inc., New York, 1944), pp. 118-123.
8. K. F. Greenough and A. B. F. Duncan, J. Am. Chem., 83 p. 555 (1961).
9- J. Behringer, pp. 168-223, Raman Spectres copy, H. A. Szymanski, ed. Plenum
Press, New York (1967).
10. S. H. Autler and C. H. Townes, Phys. Rev, 100 703 (1955) R. Gush and H. P.
Gush, Phys. Rev. 6A 129, (1972).
11. P. P. Shorygin and T. M. Ivanova, Soviet Phys.-Dokl. 3 764 (1958).
12. M. E. Mack, Appl. Opt. 13 46 (1974).
o
13. It is noted that the scattered intensity was measured over a 24 A wavelength
interval, but that raman component occurs over an interval ~ 0.5 A wide
corresponding to the full width of the laser line. Therefore, the value
of the Raman to fluorescence intensity ratio over the smaller wavelength
interval might be higher.
14. D. G. Gouche and R. K. Chang, Appl. Phys. Letters 18 579 (1971).
15. H. Kildal and R. L. Byer, Proc. IEEE ^9 1644 (1971).
16. W. A. Baum and L. Dunkelman, J. Opt. Soc. Amer 45 166 (1955).
-42-
-------
17. D. J. Bradley, J. V. Nicholas and J. R. D. Shav, Appl. Phys. Letters
19, 172 (1971)
18. C. M. Ferrar, Rev. Sci. Inst. 4o, 1437 (1969)
19. R. D. Present, "Kinetic Theory of Gases" (McGraw-Hill Book Company, Inc.
New York, 1958) pp 1^3-1
20. R. L. Taylor and S. Bitterman, Rev. Mod. Phys. kl, 26 (1969)
21. W. K. Platt, "Laser Communication Systems" (John Wiley and sons, New
York 1969) p. 121
22. M. L. Wright and K. S. Krishnan, "Feasibility Study of In-Situ Source
Monitoring of Particulate Composition by Raman or Fluorescence Scatter"
Environmental Protection Agency Report EPA-R2-73-219, June 1973.
-------
BIBLIOGRAPHIC DATA
SHEET
Report No.
EPA-650/2-7^-020
3. Recipient's Accession No.
Feasibility Study of the use of Resonance Scattering
the Remote Detection of SCL
foi
3. Report Date
January 19Jk
6.
7. Author(s)
Michael C. Fowler and Paul J. Berger
8> Performing Organization Kept,
No- N921U80-18
9. Performing Organization Name and Address
United Aircraft Research Laboratories
East Hartford, CT 06108
10. Project/Task/Work Unit No.
II. Contract/Grant No.
EPA 68-02-0656
12. Sponsoring Organization Name and Address
Environmental Protection Agency
NERC, CPL
Research Triangle Park, NC 27711
13. Type of Report It Period
Covered
Final. Report
14.
IS. Supplementary Notes
16. Abstracts
An analytical and experimental investigation has been carried out to determine the
feasibility of using the scattering of ultraviolet radiation by SO as a probe of
the concentration of that molecule in stationary source emissions. Both ordinary
fluorescence and resonant Raman scattering were considered and experimentally it
was found that the latter component was present in the scattered radiation with
sufficient magnitude to reduce significantly the degrading effect that ordinary
fluorescent quenching has on this scattering technique. Further analysis revealed
that current state-of-the-art dye lasers deliver sufficient ultraviolet pulse energy
to permit SO concentration determination in practical situations but that fluorescent
scattering from partlculates presents a possible constraint to the'validity of this
technique. A field program is recommended to investigate the latter.
17. Key Words and Document Analysis. 17o. Descriptors
Sulfur Dioxide
Lasers
Light Scattering
Raman Spectres copy
Resonance Absorption
Resonance Scattering
Fluorescence
Air Pollution
Airborne Wastes
17b. Identifiers/Open-Ended Terms
Resonant Raman Scattering
Dye Lasers
Tunable Lasers
17e. COSATI Field/Group 7B 20E
Exhaust Emissions
Optical Detection
Ultraviolet Detection
Remote Sensing
Signatures
18. Availability Statement
19.. Security Class (This
Report)
UNCL
20. Security Class
Page
UNCLASSIFIED
21. No/of Pages
51
22. Price
FORM NTIS-Se (10-70)
USCOMMOC 40S29-P71
------- |