«v? •;'.
'.• .:• YV
FEASIBILITY STUDY OF REMOTE MONITORING OF GAS
POLLUTANT EMISSIONS BY RAMAN SPECTROSCOPY
FINAL REPORT
Donald A. Leonard
Contract CPA 22-69-62
November 1970
prepared for
NATIONAL AIR POLLUTION CONTROL ADMINISTRATION
DEPARTMENT OF HEALTH, EDUCATION AND WELFARE
Durham, North Carolina 27701
EVERETT RESEARCH LABORATORY
A DIVISION OF AVCO CORPORATION
-------
FEASIBILITY STUDY OF REMOTE MONITORING OF GAS
POLLUTANT EMISSIONS BY RAMAN SPECTROSCOPY
FINAL REPORT
by
Donald A. Leonard
AVCO EVERETT RESEARCH LABORATORY
a division of
AVCO CORPORATION
Everett, Massachusetts
Contract CPA 22-69-62
November 1970
prepared for
NATIONAL AIR POLLUTION CONTROL ADMINISTRATION
DEPARTMENT OF HEALTH, EDUCATION AND WELFARE
Durham, North Carolina 27701
-------
TABLE OF CONTENTS
Page
List of Illustrations v
I. INTRODUCTION 1
II. RAMAN SCATTERING THEORY 3
III. MEASUREMENT OF NO AND SO- RAMAN
SCATTERING CROSS-SECTIONS 11
IV. NUMERICAL BASIS FOR POWER PLANT
PLUME FIELD EXPERIMENTS 13
V. EXPERIMENTAL FIELD APPARATUS 21
VI. EXPERIMENTAL FIELD RESULTS 31
VII. CONCLUSIONS 39
VIII. RECOMMENDATIONS 41
REFERENCES 42
APPENDIX - MEASUREMENT OF NO AND SO2 RAMAN
SCATTERING CROSS-SECTIONS A-l
APPENDIX REFERENCES A-6
-111-
-------
LIST OF ILLUSTRATIONS
Figure Page
II-1 Potential energy diagram for diatomic molecule 4
undergoing vibrational Raman scattering.
II-2 Detailed structure of the vibrational Raman line for 7
N2 at T = 300°K showing the close overlap of the
A J = 0 transitions and the production of sidebands
by the A J = + 2 transitions.
II-3 Apparent change in signal as a function of temperature 9
with various fixed wavelength aperatures in the detection
system.
IV-1 Concentration of pollutant detectable as a function of 15
range and integration time.
IV-2 Typical Raman spectrum from combustion gases using 18
a 3371 A laser source.
V-l Optical system used in the power plant field experi- 22
ment
V-2 Transmission characteristic of the liquid filter used 24
in the optical system of the field experiment.
V-3 Schematic of the photon counting electronic detection 25
system.
V-4 Schematic of the 10-channel photon counting detection 27
system.
V-5 Plan view of the Boston Edison Mystic Station showing 28
the position of the laser Raman spectrometer van.
V-6 Photograph of spectrometer van at power plant site. 29
VI-1 Remote ambient air Raman spectrum. 32
VI-2 Raman spectrum of smoke plume showing spontaneous 33
emission lines.
VI-3 Raman spectrum of smoke plume with spontaneous 34
emission lines suppressed.
-v-
-------
Figure Page
VI-4 Plume profile showing normalized Raman scattering from 36
N2, O , CO2 and H2O.
VI-5 Water vapor to oxygen Raman ratio in ambient air. 38
A-1 Schematic of experimental arrangement for 90 cross- A-2
section measurements.
A-2 Schematic of experimental arrangement for 180 cross- A-4
section measurements.
A-3 Spectral scan of 50% NO - 50% NZ mixture showing the A-5
relative strength of the NO line at 3599 A* and the N2
line at 3658 A\
-VI-
-------
I. INTRODUCTION
This report describes work performed between 2 June 1969 and
1 August 1970 under Contract CFA22-69-62, "Feasibility Study of Remote
Monitoring of Gas Pollutant Emissions by Raman Spectroscopy, " supported
by the National Air Pollution Control Administration.
The specific objective of this contract has been to determine the
feasibility of using laser Raman scattering to remotely monitor the
emission of NO and SO? from large stationary sources such as electric
power plants. The work has been divided into two phases as discussed
below.
PHASE I - Controlled Laboratory Study: The first phase concentrated
primarily on the measurement of the Raman scattering cross-sections
for NO and SO.,, since no previous measurement of these cross-sections
was known to have been made. Knowledge of these cross-sections is
essential both to evaluate the feasibility of the method and to quantitatively
calibrate the instrumentation. The cross-sections for both NO and SO?
were satisfactorily measured in the Laboratory and the results have been
published in the Journal of Applied Physics.
PHASE II - Field Evaluation: The second phase has been devoted to an
evaluation of the feasibility of the method in the field on an actual source
of pollutant emissions. The Boston Edison, Mystic Station Electric
Power Plant, located a mile from the Avco Everett Research Laboratory,
was selected for the field study. A laser Raman spectrometer utilizing
a 3371 A pulsed nitrogen laser was installed in a van at the plant site, and
considerable data have been collected.
For the first time, Raman spectra of N , O , H?O and CO have
been obtained from a power plant plume without interference, and
detailed field procedures have been developed which permit quantitative
-------
profiles of the molecular density of these molecules in the plume to be
routinely obtained. However, NO and SO? have not yet been observed,
due primarily to the insufficient stray light rejection ratio of the
monochrometer used in the present system.
The important conclusion to date is that no evidence has yet been
obtained which shows that the method will not work. However, a conclusive
proof of the feasibility of the method has yet to be established. Improve-
ment in the monochrometer quality will allow the feasibility of the method
to be conclusively resolved.
Other improvements in S/N include the possibility of focusing the
laser to increase beam brightness at the plume and the substitution of non-
dispersive wavelength selection to reduce the loss of Raman signal by the
monochrometer slit field stop.
-2-
-------
II. RAMAN SCATTERING THEORY
A. History
In 1923, Smekal published a theoretical paper on the quantum
theory of dispersion in which he showed that for light scattering in
transparent media frequencies other than those present in the original
light might be found in the scattered radiation. The effect predicted by
Smekal was experimentally verified in liquids by Raman in 1928, and
the phenomenon is now known as the Raman effect.
For several decades Raman spectroscopy proved to be an inval-
uable tool of the physicist for studying the structure of molecules.
However, this early work was limited by the need for very long expo-
sure times with extremely high power mercury arc lamps. The advent
of the laser, a high intensity source of monochromatic light, has caused
a renewed interest in Raman spectroscopy. Instruments have been mar-
keted which combine CW lasers and scanning tandem monochromators
into systems of great practical utility for laboratory use. The intent of
this effort was to study the feasibility of extending the practical appli-
cation of Raman scattering for the remote analysis of power plant plumes
in field situations.
B. Theory
The frequencies observed in Raman scattering correspond to the
frequency of the incident light shifted by some characteristic frequency
of the scattering molecule. A potential energy diagram for a typical
molecule undergoing vibrational Raman scattering is shown in Fig. II-l.
Photons of frequency v and energy h^ are incident on the molecule.
Photons of frequency "Raman and energy n"Raman are scattered, and
the molecule is left in a higher vibrational energy state after the col-
lision than before. The difference in energy between the incident and
scattered photons is exactly equal to the energy given to the molecule,
-3-
-------
POTENTIAL
ENERGY,
V
V /_
AEV|B =
\
INTERNUCLEAR SEPARATION,
hl/RAMAN = h"o
AE
VIB
Fig. II-1 Potential energy diagram for diatomic molecule undergoing
vibrational Raman scattering.
-4-
A9692
-------
and in the case of vibrational excitation, is the vibrational energy spac-
ing, AE , . Thus, by measuring the energy of the Raman scattered
photons and by knowing the energy of the incident photons the vibrational
energy spacing of the scattering molecule may be determined. Conversely,
if the energy of Raman photons from a particular molecular species is
known, then a measurement of the Raman scattering at that photon
energy will be a means of monitoring that particular species.
The molecular species of importance for atmospheric Raman
scattering are listed in Table I. The vibrational energy spacing, the
wavelength to which the Raman scattering is shifted (assuming a 3371 A
pulsed nitrogen laser source) and the Raman scattering cross-section
for each species, normalized to nitrogen, are also shown.
The selection rules for the change in rotational-quantum number
during a vibration Raman transition are AJ = 0 and AJ = + 2. Usually
all the lines having AJ = 0 lie very close to each other in energy and
are not resolved except with very high resolution spectroscopy. The
AJ = + 2 transitions, however, are well separated in energy and appear
as side bands on either side of the intense AJ = 0 line. Figure II-2 is
a representation of the vibrational Raman scattering for molecular nitro-
gen excited with a 3371 A pulsed nitrogen laser, showing the intense
AJ = 0 line with much weaker side bands for the AJ = - 2 and AJ = + 2
rotational transitions.
This figure was drawn for a temperature of 300 K with the
assumption that the A J = 0 transition is a factor of 100 stronger than the
2
A J = _+ 2 transitions, as indicated by a theoretical calculation by Sharma.
Details such as the alternation in intensity of the A J = + 2 lines have been
neglected. See Herzberg for further information.
The side bands can be important in practice for the following two
reasons:
(1) The AJ = + 2 side bands of a strongly scattering, high con-
centration species may overlap and mask the weaker AJ = 0 scattering
from a low concentration species. This does not appear to be a problem
for the detection of NO and SO which are well separated in Raman energy
shift from N and other high concentration species. However, problems
-5-
-------
TABLE I
DATA FOR MOLECULAR SPECIES OF INTEREST
FOR ATMOSPHERIC RAMAN SCATTERING
Molecule
N2
°2
H2°
co2
CO
NO
so.,
Vibrational
Energy
Spacing
2330 cm"1
1556
3652
1285. 5
1388. 3
2143
1876
1151
Wavelength of
Raman Scattering
with 3371 A Laser
Source
3658 8.
3558
3844
3523. 7
3536. 5
3634
3599
3507
^RAMAN
CTRAMAN (N ;
1.
1.
4.
2.
2.
0.
2.
00
5*
15**
2***
2
5
4vvvv
Widhopf and Lederman (Brooklyn Polytech) see Reference 4
Cooney, NCAR - Reference 5
Derr, ESSA - Reference 6
Leonard, Avco - Reference 7 and Section III
-6-
-------
10°
AJ = 0
10
-2
ID'
AJ =
3630 3640 3650 3660 3670
WAVELENGTH (ANGSTROMS)
3680
Fig. II-2 Detailed structure of the vibrational Raman line for N2 at
T = 300°K showing the close overlap of the A J = 0 transitions
and the production of sidebands by the A J = + 2 transitions.
-7-
C2307
-------
may arise in cases such as the detection of trace amounts of CO in the
presence of high concentrations of N , as in combustion plumes. This
potential source of interference in the analysis of combustion plumes due
to side bands is discussed in more detail in Section IV.
(2) With a fixed wavelength aperture in the detection system,
temperature variations will cause changes in the amount of Raman
scattering detected from a constant density of scattering molecules. This
is illustrated by Fig. II-3 which shows the fraction of the total Raman
scattering from AJ = 0 and AJ = + 2 as a function of temperature for
various fixed wavelength aperture widths which are centered on the AJ = 0
line. This plot is calculated theoretically for the worst case when the
strength of the A J = _+ 2 transitions are assumed equal to the A J = 0.
It can be noted that for very wide apertures which include most of the
side bands there is very little temperature effect and also for very narrow
apertures there is also very little temperature effect. Precision measure-
ments at the 1% level of accuracy may have to take this effect into con-
sideration.
-8-
-------
1.0
.9
.8
Q
LU
O
jf .7
UJ
Q
O
<
cr
u_
.5
.4
.3
AX = 60 A
24
12
8
2
1
1
1
200 220 240 260 280
TEMPERATURE, °K
300
Fig. II-3 Apparent change in signal as a function of temperature with
various fixed wavelength aperatures in the detection system.
-9-
C2306
-------
III. MEASUREMENT OF NO AND SO2 RAMAN SCATTERING
CROSS-SECTIONS
The Raman scattering cross-sections for NO and SO were measured
experimentally with a 3371 A laser source as part of this feasibility study,
since no previous measurements of these cross-sections was known to have
been made.
The results of these measurements has been published in the Journal
of Applied Physics. The text of this paper is reproduced as an Appendix to
this report.
Additional information which was not included in the paper as
published:
1. The measurements were carried out at room temperature,
i. e. , approximately 300 K.
2. The instrumental bandpass was 17 A as controlled by a
1 mm slit on a 1/2-meter 1200 £/mm grating mono-
chrometer.
3. The results were corrected for spectrometer, inter-
ference filter and photomultiplier spectral response which
over the 151 & range between the SO2 line at 3507 K
and the N_ line at 3658 A* amounted to 30% or less.
4. The quoted uncertainties were the combination of the
photostatistic and spectral response uncertainties.
-11-
-------
IV. NUMERICAL BASIS FOR POWER PLANT PLUME
FIELD EXPERIMENTS
The number of Raman photoeiectrons collected by an optical
detector in an atmospheric pulsed laser backscattering system
can be expressed by the following range equation:
A
= NLASERft NSCAT aRAMAN AR — e e ^ T, T.
2 Pe UP A A_
R 2
\vhere N.—..-, = number of Raman photoeiectrons detected
N
LASER = number of laser photons per pulse
f = laser pulse repetition frequency
t = integration time
N^ . = density of molecular scatters
<7 . = Raman scattering cross-section per particle
per ster^dian
AR = range resolution
A = telescope aperture area = TiD
4
D = telescope aperture diameter
-13-
-------
R= range
e = photocathode photoelectric efficiency
e = optical system efficiency
T\ T, = two-way atmospheric transmission, T. at laser
12 1
wavelength and T, at Raman wavelength
X2
It can be seen from this equation that, if the transmitter and
receiver system parameters N .-..-,, a AMAN» ^R, tne detection solid
angle and the detection efficiencies are known, the only two quantities
remaining are the density of scattering molecules N_r- and the two-way
transmission factor T T,. to the range of interest. Thus a measure of
1 2
the Raman scattering from a single range cell can be used to determine
directly the productof the density of scatterers and the two-way transmission
to the region of interest.
If the density of one component, such as N is known, a measure of its
Raman scattering yields the transmission to the range cell of interest. Thus,
normalization of the Raman scattering from pollutants of unknown concentration
to that of N£ yields an absolute concentration ratio independent of transmission,
if the transmission at the various Raman wavelengths differs only slightly.
A calculation was carried out using the above range equation with
system parameters which assume state-of-the-art, well engineered com-
ponents. The parameters assumed are listed below together with a brief
rationale for the particular choice.
The detectable concentration of NO and SO_ as calculated using the
above range equation and system parameters is shown in Fig. IV-1 as
a function of range and integration time. The conclusion from this
figure is that NO and SO should be detectable at the 100 PPM level at
ranges up to 1000 feet with reasonable integration times the order of
a few minutes.
-14-
-------
10
cc
I io3
z
o
o
10
NO
SO,
100 1000 10,000
RANGE,FEET
Fig. IV-1 Concentration of pollutant detectable as a function of range
and integration time.
-15-
C3026
-------
System
Parameter
Value
Assumed
Rational
N
PE
NLASERf
100 photoelectrons
. 7xl017photons
sec
In a signal limited
situation, such as
during night operation,
this would produce a
photostatistical accuracy
of 10%.
This value corresponds
to 0. 1 watts of average
o
power at 337 1A in the
ultraviolet. This is
provided by the commer-
cially available Avco
Model C950 pulsed N_
laser for example.
N
SCAT
CTRAMAN (S°2)
CT RAMAN (NO)
2. 9xl019(PPMxlO"6)
4. 8xlO"29cm'd/ster
1. 0x10 cm /ster
Operation is at
approximately Standard
atmospheric density.
Using the data of
Table I and a value
of 2xlO-29cm2/ster
forCTRAMAN(N2)'
AR
10 feet
Typical power plant
plume diameter.
T
—(10) inches
Typical field optical
systems are of this size.
pe
0. 30
RCA 8850 photomultiplier
photocathode sensitivity
in the ultraviolet.
op
0. 10
1. 0
-16-
Typical transmission of
a good monochrometer
plus transfer optics.
This factor can vary from
unity to zero depending
on conditions.
-------
The experimental field apparatus described in Section V was
designed to match as closely as possible the system parameters used in
the above calculation. However, the laser output power was a factor of
4 less due to spatial and spectral filtering and there was a factor of 6
optical mismatch in the monochrometer system. This mismatch occurs
because the combination of the 4 milliradian (2 mrad 1/2 angle) laser beam
and the 60-inch focal length telescope produces a 6 mm image on the 1 mm
wide monochromator entrance slit field stop. This loss of a factor of 6
can be corrected by the use of beam expansion and focusing optics in the
laser transmitter.
The observed experimental Raman signals were a factor of 100 less
than the calculation and a factor of 4 less than the actual field apparatus
would be expected to produce. This factor of 4 discrepancy could be ex-
plained by a combination of uncertainties in cross-section absolute magni-
tude, optical transmission losses and the efficiency of the photoelectron
counting process.
In addition to the absolute Raman signal strengths, the interference
and possible overlap among the Raman scattering from the various molecu-
lar constituents of a combustion plume can also be calculated. Figure IV-2
shows the Raman spectrum expected from typical combustion gases at
300 K, including the A J = +_ 2 rotational sidebands. The A J = 0 components
have been normalized to a signal of unit strength for N_ in accordance with
their concentration and scattering cross-section.
The A J = +_ 2 components for N_ are shown in Fig. IV-2, as
previously indicated in Fig. II-2 and described in Section II. The ampli-
tude of the A J = +_ 2 components for the Raman lines of the other species
are also shown approximately two orders of magnitude weaker than the
strong A J = 0 component. The width of the various A J = _+ 2 components
is, however, scaled proportional to the 1/2 power of the rotational constant
(B in Herzberg's notation) for each specie.
The justification for this scaling can be seen by differentiating the
expression for NT the number of molecules per rotational level.
-17-
-------
X
o I0~'
2
UJ
cr
•" lO-2
CO IU
_l
<
2
W l0"3
UJ
_J
UJ
cc
I0
~4
3500
100 PPM
NO
15 PPM
CH4
25PPM
CO
10%
H20
I
3600 3700 3800
WAVELENGTH, ANGSTROMS
3900
Fig. IV-2 Typical Raman spectrum from combustion gases using a
3371 A laser source.
-18-
C2668
-------
= -- (2J
BJ(J+1) 1. 44
T
J_
Q
- BJ(J+1) 1. 44
T
2 e
1. 44B
-BJ(J+1)1. 44
T
(2J+ 1) e
and setting the derivative equal to zero and solving for J^ as a function
of B
Max
X -X ,
o J
Max
AEX2 = B (4J + 6) X2
o o
A - X T
o J.
Max
•y/BT
The widths of the A J = jf 2 sidebands were scaled according to
Bg = 2. 01, 1. 44, . 39, 9. 5 cm"1 for NZ, O2> CC>2 and H2O, respectively. 3
(SO_ was arbitrarily set equal to CO_).
In summary the amplitudes of the sidebands in Fig. IV-2 are order
of magnitude estimates whereas the widths are probably good to within a
factor of 2. A more accurate and complete recalculation of Fig. IV-2,
particularly as a function of temperature and relative concentration, would
assist in defining the limits of the Raman method. However, NO is seen
to be clearly free of interference and SO- lies more than 2 orders of magni-
tude above the sideband of CO_. The CO Raman scattering on the other
hand is less than the N_ sideband for CO concentrations below about 250
PPM.
-19-
-------
V. EXPERIMENTAL FIELD APPARATUS
This section describes the remote pulsed laser Raman spectrometer
which has been deployed and operated at the site of the Boston Edision
Mystic Station Electric Power Plant.
A. Laser
The laser used is a 3371 A ultraviolet pulsed nitrogen laser,
MOPA (jvlaster (Oscillator P_ower Amplifier). A small pulsed nitrogen
laser is used as an oscillator to provide approximately 10 kW of peak power.
The beam from the oscillator is passed through a mode control spatial filter
which produces a beam with a divergence of 2 milliradians. The divergence
controlled beam is then amplified in a second larger laser unit which is
synchronized with the oscillator to within 1 nanosecond.
The output beam from the MOPA is passed through an interference
filter designed to pass the 3371 A laser line and block by at least 3 orders
of magnitude the spontaneous emission lines in the nitrogen discharge which
occur in the wavelength region of the Raman lines of interest. Steering
mirrors are used to position the beam within the field of view of the telescope
at the range of interest.
The resulting laser MOPA unit produces an output beam with the
following characteristics:
Peak Power - 50 kilowatts
Pulse Repetition Rate - 100 pulses per second
Pulse Duration - 10 nanoseconds
Wavelength - 3371 £
Beam Divergence - 2 milliradians
B. Receiver Optics
The optical system of the receiver is shown schematically in
Fig. V-l. A 10-inch diameter, 60-inch focal length Newtonian telescope
is used to collect the Raman backscattered photons produced by the outgoing
laser pulse. Gross alignment is provided by the laser beam steering mirror.
-21-
-------
QEYEPIECE
-M
Hl-
MONOCHROMETER
LIQUID FILTER
PHOTOMULTIPLIER
Fig. V-l Optical system used in the power plant field experiment.
-22-
C2246
-------
In practice, the telescope is first aligned so that the smoke plume is in
view in the telescope eyepiece. The steering mirror is then adjusted for
maximum returned signal on a strong Raman line such as N? at 3658 A.
The photons collected by the telescope are passed through a Jarrell-
Ash 1/4 meter monochrometer. This monochrometer has a measured
resolution of 17 A per mm, a transmission of 20% in the ultraviolet and
a stray light rejection ratio reputed to be 10
A lens focuses the exit slit of the monochrometer onto a field
stop placed in front of an RCA 8850 photomultiplier. A liquid filter is
placed adjacent to the lens. The liquid filter, a water solution of 2, 7 -
dimethyl - 3, 6 - diazacyclohepta - 1,6 - diene perchlorate in a quartz
walled cell, has the property of essentially complete isotropic volume bulk
absorption of 3371 A but with nearly complete transparency at wavelengths
of 3500 A and longer. Figure V-2 is a plot of the filter's transmission vs
wavelength. This filter is essential for blocking the strong on-frequency
return at 3371 A which is reflected from the smoke plume and which the
Jarrell-Ash 1/4 meter monochrometer alone is not able to adequately
reject. It is far superior in this regard than the best interference filters
commercially available, both in transmission characteristics and in angle
requirements.
C. Basic Receiver Electronics
The block diagram of Fig. V-3 shows the receiver detection elec-
tronics. The output from the RCA 8850 photomultiplier is sent directly
to a 200 megacycle Hewlett-Packard 5248 counter. The counter discrimina-
tor threshold is set so that all photoelectron pulses produced by the photo-
multiplier are counted directly by the counter.
A synchronization pulse produced by the laser trigger circuit is
suitably delayed (100 nanoseconds = 50 feet of range) and then sent to a
gate generator which produces typically a 100 nanosecond gate pulse. This
gate pulse is introduced into the counter in a manner such that the counter
will count photoelectron pulses only during the time the gate is on. At all
other times, between the laser pulses, the counter is inactivated and does
not count the background and dark current photoelectron pulses produced.
-23-
-------
100
90
80
70
Z
o
CO
«2 60
5
-------
LASER
SYNC
DELAY
ICONS
GATE
_TL
COUNTER
Fig. V-3 Schematic of the photon counting electronic detection system.
-25-
C2241
-------
For the photomultiplier tube used, a typical dark current counting rate
when ungated is 200 counts per second. When operating at 100 laser pulses
per second, and with 100 nanosecond gate lengths the counting rate from
dark current is reduced from 200 to 200 x = 2 x 10"^ counts per
second.
D. 10-Channel Receiver
10-2
To achieve the simultaneous measurement of an entire plume profile,
as distinguished from the single point by point capability of the basic elec-
tronic system described above, a 10-channel (i. e. , range gates) photon
counting receiver was assembled and field tested satisfactorily.
As Fig. V-4 shows, the synchronization pulse from the laser trigger
circuit is suitably delayed and then sent to a set of 1 0 gate generators
which produce gate pulses, 30 nanoseconds wide and spaced 50 nanoseconds
apart over a 500 nanosecond interval. These gates are introduced into AND
circuits together with the photoelectron pulses from the photomultiplier.
When a coincidence of a gate pulse and a photoelectron pulse occurs, a
fixed increment of charge is added to the appropriate integrator. The D. C.
voltage appearing at the output of each integrator is proportional to the
number of photoelectrons received from each range cell. This device has
enabled plume profiles to be obtained with one-tenth the integration time
previously required.
E. Power Plant Site
The remote laser Raman spectrometer apparatus was placed in
a van at the site of the Boston Edison Mystic Power Plant. The spectrom-
eter van is approximately 500 feet from the base of the nearest stack, the
top of which is 317 feet above sea level. The slant range from the laser
to the stack plume was measured by the time of flight of the strong on-
frequency return to be 1250 nanoseconds or 625 feet.
The position of the laser Raman spectrometer van relative to the
power plant stacks is indicated in Fig. V-5 which is a sketch of the Mystic
Station site plan.
Figure V-6 is a photograph showing the view of the power plant
stacks from behind the spectrometer van.
-26-
-------
C ATC" in
oA 1 1 IU
JL
I
Fig. V-4 Schematic of the 10-channel photon counting detection system.
-27-
C3028
-------
RAMAN LASER
SPECTROMETER
o o
STACK STACK
6 5
STACK
4
N —
O
W
o
STACK STACK 1/2
3 (2 BOILERS)
STACKS 1/2, 3 - 260 FT HEIGHT. MOUTH DIAMETER = 16 FT
STACKS 4,5,6 - 317 FT HEIGHT, MOUTH DIAMETER =10.5 FT
Fig. V-5 Plan view of the Boston Edison Mystic Station showing the
position of the laser Raman spectrometer van.
-28-
C2244
-------
Fig. V-6 Photograph of spectrometer van at power plant site.
-29-
C5706
-------
VI. EXPERIMENTAL FIELD RESULTS
The apparatus described in Section V has been operated at the
Boston Edison site and the results to date confirm our expectations of its
capability, the current limitation being the selectivity of the 1/4-meter
Jarrell-Ash monochrometer at about the 1% level of stray light rejection.
Useful data, however, were obtained with the existing experimental
arrangement. Figures VI-1, VI-2 and VI-3 are typical spectral scans
of the backscattered photons produced by the laser pulse which have
been obtained at the power plant site. Figure VI-1 is a scan at a range
of 425 feet, which is between the spectrometer van and the stack
plume, and is a normal Raman air spectrum on a clear evening showing
the O , N and H7O Raman lines in approximately the expected ratios.
£ £ £*
This scan was obtained without an interference filter on the laser output.
Figure VI-2 is a spectral scan at a range of 625 feet, a range
coincident with the stack plume position. In this spectral scan, four new
lines have appeared at 3537 A, 3577 A , 3755 A , and 3805 R , in addition
to the O , N and H?O Raman lines at 3558 R, 3658 R and 3844 R,
L* £* £>
respectively. These new lines were identified as emission lines in the
laser discharge originating from the (1, 2), (0, 1), (1, 3) and (0, 2) transitions
of the nitrogen second positive (C TT - B TT) band system. The 3371 A laser
line originates from the (0, 0) transition of the same band system. The
appearance of these relatively weak emission lines in the Raman spectrum
when the range corresponds to the plume position indicates the enormously
larger optical reflectivity of the plume as compared to "clear" ambient
air.
The spectral scan of Fig. VI-3 resulted from the insertion of an
interference filter in the outgoing laser beam which was designed to pass
the laser line at 3371 A and attenuate by at least 3 orders of magnitude
wavelengths between 3500 R and 3900 R. Although the data of Fig. VI-3
-31-
-------
300
UJ
200
cr
UJ
o.
co
I-
z
ID
o
o
100
13 APRIL 1970
RANGE 425 FEET
(AMBIENT AIR)
0
3500
3600 3700 3800
WAVELENGTH, ANGSTROMS
3900
Fig. VI-1 Remote ambient air Raman spectrum
-32-
C2243
-------
700
600
500
UJ
5 400
cr
LU
a.
{2 300
z
o
o
200
100
15 APRIL 1970
RANGE 625 FEET
(IN SMOKE PLUME;
H,0
-»-RAMAN
— EMISSION
LINES
3500
3600 3700 3800
WAVELENGTH, ANGSTROMS
3900
Fig. VI-2 Raman spectrum of smoke plume showing spontaneous emission
lines.
-33-
C2242
-------
500
400
300
CC
UJ
O.
o
o
200
100
02
A
15 APRIL 1970
RANGE 625 FEET
(IN SMOKE PLUME)
3371 FILTER ON
LASER OUTPUT
H20
I
3500
3600 3700 3800
WAVELENGTH, ANGSTROMS
3900
Fig. VI-3 Raman spectrum of smoke plume with spontaneous emission
lines suppressed.
-34-
C2240
-------
were obtained at the same 625 feet range position as Fig. VI-Z, the
addition of the filter eliminated the four new lines leaving just the O_, N
and HO Raman lines.
Scans as a function of range were obtained on the O,,, N~ and H~O
£* L* L*
Raman lines. Data from a typical such scan are shown in Fig. VI-4 in
which the Raman backscattered return multiplied by the square of the range
is normalized to unity at a delay of 1400 nanoseconds and is plotted as a
function of delay. The plume position is at approximately 1600 nanoseconds.
Data with less delay represent scattering from in front of the plume, data
with more delay represent scattering from behind the plume.
The following conclusions and observations can be drawn from
Fig. VI-4:
(1) The ratio of the N_ (or O?) Raman signal level behind the plume
to the N (or O ) Raman signal level in front of the plume gives a direct
measure of the two-way plume transmission. A solid line is drawn through
the points for the N~ Raman signals showing a drop to 0. 2 which is a direct
measure of the two-way plume transmission. The one-way transmission
is (0. 2)1//2 or 45%.
(2) The O,, signal level at the delay time corresponding to the
plume position is less than the N_ signal level at the same delay time.
(3) The H_O signal level at the delay time corresponding to the
plume position is greater than the N_ signal level at the same delay time.
Observations (2) and (3) above show the expected decrease in O?
content of the plume and an increase in water vapor due to combustion of
the hydrocarbon fuel. The uncertainty in the data for water vapor, while
large, is still adequate to support the above conclusions. The water vapor
to N?(and O_) ratio before and after the plume is the same within the pre-
cision of the measurement.
Similar data have been obtained with the 10-channel receiver, rep-
resenting the simultaneous measurement of an entire plume profile. In
addition, with the same receiver, scattering from the plume position was
obtained at 3524 A , the Raman wavelength corresponding to CO_. This CO?
data is also shown in Fig. IV-4 normalized by setting the value at 1600 ns equal
to 1. 0 and the average of all the other range cells equal to zero. The con-
clusion is that a significant CO- signal appears only at the plume position.
-35-
-------
2.0
1200 1300 1400 1500 1600 1700
DELAY,NANOSECONDS
1800
1900
2000
Fig. VI-4 Plume profile showing normalized Raman scattering from N.
and
-36-
C2608
-------
A. Flue Gas Analytical Measurements
The Walden Research Company was engaged to make analytical
measurements by conventional methods on the power plant exhaust gases.
One such set of tests were conducted by Walden on the evening of
2 July 1970. The exhaust gases were sampled in the duct between the
electrostatic precipitator and the direct draft fan. The Orsat analysis
showed CO_ at 9. 1% and CL at 9.7% indicating the presence of about 50%
excess air. The NOV Phenal Disulfonic Acid analysis ranged from 155 to
}\
234 PPM. The SO? measurement failed to work because of a leak in the
sampling system used. It was intended that such analytical measurements
be made coincident with the laser Raman measurements to validate the
technique.
B. Remote Laser Hygrometry
Measurement of water vapor profiles in ambient air was
accomplished incidental to the plume study. Figure VI-5 shows the ratio of
the Raman signal from H_O to the Raman signal from O_ plotted as a
function of range. Each point in this case was obtained separately and the
entire set of data was obtained over a half hour period. The error bars
represent the photostatistics associated with the collection of about 100
photoelectrons for each point. The average of the data is compared with
that calculated from Weather Bureau temperature and relative humidity
data supplied from Logan Airport and the cross-sections for H?O and O?
as given in Table I. The ratio of the calculated value to the experimental
average is . 82. Similar comparisons on three other nights gave comparison
ratios of 1.06, 1.06 and 1.40. Exact agreement is not expected since the
laser and Weather Bureau measurements were not precisely coincident in
either time or space. However, the close correlation indicates that remote
laser hygrometry is feasible and might be used to improve meteorological
data acquisition.
-37-
-------
.30
.26
.22
H0
.18
.14
.10
I I I I I | I I I |
12 MAY 1970
AMBIENT AIR
BOSTON EDISON
SITE
-AVERAGE
i OF DATA
7
CALCULATED FROM LOGAN AIRPORT
WEATHER BUREAU T8R.H. AND
CROSS-SECTION DATA FOR H20 S02
i i i
400
500 600
RANGE,FEET
700
800
Fig. VI-5 Water vapor to oxygen Raman ratio in ambient air.
-38-
C3027
-------
VII. CONCLUSIONS
Calculability of Raman Returns - The cross-sections for NO
and SO? Raman scattering have been measured and are now known. The
Raman returns from plumes containing these molecules can now be calculated
with precision. The detection of Raman scattering from 100 PPM NO and
SO_ is theoretically practical with reasonable system components.
Molecular Interference - For concentrations of NO and 5O
of 100 PPM and above, it has been established theoretically that the Raman
scattering from the other molecular species in the plume (CO?1 O , N ,
£ L. L.
H?O and CO) will not interfere with the measurement if an appropriate
high performance monochrometer is used.
Raman Scattering Obtained Remotely from Plume - Raman
spectra of N7, O7, HO and CO_ have been obtained experimentally from
£ L* L. L*
a power plant plume at a range of 625 feet.
Quantitative Profiles of Plume Species - Quantitative density
profiles of molecular species in the plume can now be routinely obtained.
For example, the fact the plant was operating with 50% excess air was
detected by the ratio of the O? Raman to the N? Raman observed in the
plume and later confirmed by an Orsat analysis of the stack gases.
Plume Transmission Measured - The two-way plume transmission
was easily measured by directly comparing the ratio of the N (or O_)
Raman signal level behind the plume to the N? (or O..,) Raman signal level
in front of the plume.
Establishment of Preferred Definition of Emission Standard -
The simultaneous measurement of CO , SO? and NO can be used to unam-
C, £*
biguously obtain the ratio of SO_ to CO_ and NO to CO independent of the
optical transmission characteristics of the plume or the amount of excess
diluent air passed up the stack. Specifying the SO /CO., and NO/CO
L* L* L~
ratio might prove to be a practical, easily measureable, and legally
enforceable standard.
-39-
-------
Simultaneous Range Information Required - Approximately 10
simultaneous range data points are required to establish unambiguously
the position of plume, the transmission through the plume, and the relative
concentration of molecular constituents within the plume.
No Interference Experimentally Observed - No interference
with the molecular Raman scattering has been experimentally observed
down to an equivalent pollutant level of 10, 000 PPM.
Fluorescence and Particulate Raman Interference - The possible
interference from flourescence and Raman scattering from the solid and
liquid material in the plume, such as ash, dust, and aerosols, cannot as
yet be estimated with sufficient precision. The existance of interference
from such sources can only be determined experimentally on an actual
source of emissions such as a power plant. Reasonably sophisticated
equipment must be used.
100 PPM Feasibility Not Resolved - Raman returns from NO
and SO? have not yet been observed experimentally from the power plant
plume. The detection by Raman scattering of NO and SO_ in the 100 to
1000 PPM concentration range has therefore not yet been proven feasible.
There still exists the possibility that fluorescence and Raman scattering
from foreign materials may limit the effectiveness of the technique.
Feasibility of RamanScattering for Remote Hygrometry is
Indicated: The measurement of remote water vapor profiles in ambient
air was accomplished incidental to the plume study. The data obtained
were consistent with Weather Bureau records of the humidity in the
Boston area at the time of the tests. This technique may be of use for
meteorological support of air pollution control.
-40-
-------
VIII. RECOMMENDATIONS
Additional Field Experiments with Improved Monochrometer -
More experimental field work with an improved instrument is required
to resolve the question of the feasibility of the Raman method for measure-
ment of NO and S O2 in the 100 PPM concentration range. The work to
date utilized a standard 1/4-meter Jarreil-Ash monochrometer with a
stray light rejection ratio of 1 part in several hundred. A Spex Model
4
1702 monochrometer with a stray li^ht rejection ratio ol 1 part in ID
to 10 has been purchased by Avco for its laser Raman program. An
instrument of this caliber, used in conjunction with the laser and other
equipment assembled at the power plant site, should enable a satisfactory
resolution of the question of feasibility.
-41-
-------
REFERENCES
1. Smekel, A. , Naturwiss JJ_, 873 (1923).
2. Sharma, Ramesh, Avco Everett Research Laboratory, private
communication.
3. Herzberg, G. , "Molecular Spectra and Molecular Structure, Infrared
and Raman Spectra of Polyatomic Molecules, " D. Van Nostrand Co.
(1966).
4. Widhopf, G. F. and Lederman, S. , "Specie Concentration Measure-
ments Utilizing Raman Scattering of a Laser Beam, " Polytechnic
Institute of Brooklyn, Report No. 69-46.
5. Cooney, John, National Center for Atmospheric Research, private
communication.
6. Derr, Vernon, ESSA, Boulder, private communication.
7. Leonard, D.A., Journal of Applied Physics, 41, 4238 (1970).
-42-
-------
APPENDIX
MEASUREMENT OF NO AND SO2 RAMAN SCATTERING
CROSS-SECTIONS
The recent growing interest in air pollution control has motivated
a search for more effective means of remote monitoring of atmospheric
pollutants. One possibility is remote laser Raman scattering. The obser-
vation of Raman backscatter at ranges up to several kilometers in the
A-1 A-2
atmosphere has previously been reported for nitrogen and oxygen '
A-3, A-4
and water vapor.
The purpose of this letter is to report laboratory measurements
of Raman scattering cross-sections at 337. 1 nm for gaseous nitric oxide
(NO) and sulfur dioxide (SO_). Measurements were made at both 90 and
180 . Remote pollution monitors using laser Raman backscattering require
knowledge of these cross-sections for instrument design and calibration.
A schematic of the apparatus used for the 90 measurements is
shown in Fig. A-l. The laser source consisted of a 100 kilowatt peak power,
A- 5
10 nanosecond 337. 1 nm pulsed nitrogen laser operating at 100 pps.
The output from the laser was filtered with the interference filter, F,,
which passed the laser line at 337. 1 nm and attenuated the spontaneous
emission occurring at the Raman wavelengths in the laser discharge by
more than a factor of one thousand. After filtering, the laser beam was
focussed at the center of a test cell containing the gases of interest.
After passing through a long pass filter, F?, the focal region of
the test cell was imaged onto the slit of a half-meter monochrometer. The
filter, F_, was designed to pass the Raman lines but attenuate the 337. 1 nm
laser line by a factor greater than one thousand.
After passing through the monochrometer the signal was detected
by an RCA 7265 photomultiplier and displayed on a Textronic 585 oscil-
loscope, the latter being triggered by a sync pulse from the laser. The
trace of the oscilloscope was viewed with a 1P28 photomultiplier, the
A-l
-------
TEST
CELL
PM VIEWING
SCOPE FACE MASK
Fig. A-l Schematic of Experimental Arrangement for 90 Cross-
Section Measurements
A-2
-------
output of which went directly to a counter. A mask on the face of the oscil-
loscope prevented the 1P28 photomultiplier from seeing the baseline of the
trace.
However, an aperture in the mask placed below the trace permitted
viewing the pulses originating from the monochrometer photomultiplier
during a 100 nanosecond interval which includes the 10 nanosecond laser
pulse. These pulses are recorded as individual counts on the counter.
This method of coincidence counting is described in Ref. A-6.
The physical set-up for the 180 measurements is shown in Fig. A-2.
The additional field stop is arranged so that essentially all the light back-
scattered from the focal region within the test cell is transmitted; whereas
laser light scattered from the objective lens and the test cell windows is
strongly attenuated by the field stop.
The measurements were made with mixtures of NO with N_ and SO_
with N7 in the test cell. The desirability of using N_ Raman scattering
A-7
as a normalizing standard for atmospheric Raman spectroscopy required
the measurement of the NO and SO2 cross-sections relative to N?. No
attempt was made to measure absolute cross-sections. However, the
- 29 2 A 8
results were consistent with the value of 2 x 10 cm /ster for N?.
Typical data obtained with an equal molar mixture of NO and N? at
one atmosphere total pressure are shown in Fig. A-3 as a plot of counts
per minute as a function of wavelength. The counting rate was purposely
maintained much smaller than the 6000 pulse-per-minute repetition rate
of the laser to minimize the possibility of multiple photon events during
one gate interval. The peaks at 359. 9 and 365. 8 nm correspond to the
vibrational Raman Stokes shifts for NO and N^i respectively. Similar
data were obtained for mixtures of SO? with N.-,.
The results for both 180° and 90° are best fit by a NO to NZ Raman
cross-section ratio of 0. 5 +_ 0.1 and a SO., to N_ Raman cross-section
ratio of 2. 4 -t- 0. 3. Since this measurement includes essentially only
AJ = 0 contributions, these cross-section ratios are for the AJ = 0 transi-
tions only. A careful analysis of the wings of the Raman lines would be
required to determine the relative strengths of the AJ = +_ 2 transitions.
A-3
-------
\
(180° CROSS-SECTIONS)
TEST CELL
FIELD STOP
TO OSCILLOSCOPE
MONOCHROMETER
Fig. A-2 Schematic of Experimental Arrangement for 180 Cross-
Section Measurements
A-4
C2366
-------
120
100
z
i 80
\
V)
O
O
60
40
20
350
355 360 365
WAVELENGTH, NANOMETERS
370
Fig. A-3 Spectral Scan of 50% NO - 50% N Mixture showing the
Relative Strength of the NO Line at 3599 A" and the NZ
Line at 3658 A*
A-5
C1775
-------
APPENDIX REFERENCES
A-l. Leonard, D. A., Nature 2^6, 142(1967).
A-2. Cooney, J. A., Appl. Phys. Letters J_2, 40 (1968).
A-3. Cooney, J. A., Nature 224_, 1098 (1969).
A-4. Melfi, S. H. , Lawrence, J. D. Jr. , and McCormack, M. P. ,
Appl. Phys. Letters J_5, 295 (1969).
A-5. Leonard, D. A., Laser Focus J3, 26 (1967).
A-6. Lewis, I. A. D. , and Wells, F. H. , Millimicrosecond Pulse
Techniques, 2nd Edition (McGraw-Hill, N. Y. , 1955) p. 263.
A-l. Recommended by the Standards Committee of the Group on
Laser Atmospheric Probing, National Center for Atmospheric
Research, Boulder, Colorado.
A-8. Stansburg, E. J. , Crawford, M. F. , and Welsh, H. L. , Can.
J. Phys. 3J_> 954 (1953).
A-6
------- |