FEASIBILITY STUDY FOR THE DEVELOPMENT
     OF A MULTIFUNCTIONAL EMISSION
  DETECTOR FOR AIR POLLUTANTS BASED
  ON HOMOGENEOUS CHEMILUMINESCENT
          GAS PHASE REACTIONS
         Final Report for the Period
     15 August 1968 - 14 September 1969 on
           Contract CPA 22-69-11
                Research Laboratories,  Inc.
            SYBRON CORPORATION
       "rinceton, New Jersey

-------
AeroChem TP-217                                  September 1969
             FEASIBILITY STUDY FOR THE DEVELOPMENT
                   OF A MULTIFUNCTIONAL EMISSION
                DETECTOR FOR AIR POLLUTANTS BASED
                ON HOMOGENEOUS CHEMILUMINESCENT
                        GAS PHASE REACTIONS
                       Final Report for the Period
                  15 August 1968 -  14 September 1969 on
                          Contract CPA 22-69-11
                             Arthur Fontijn
                           Alberto J. Sabadell
                            Richard J.  Ronco
                 AeroChem Research Laboratories, Inc.

                         Princeton,  New Jersey
                   a subsidiary of Sybron Corporation
                             Prepared for:

          NATIONAL CENTER FOR AIR POLLUTION CONTROL

                          Public Health Service
                           3820 Merton Drive
                          Raleigh,  N. C.  27609

-------
TP-217
SUMMAR Y
The r-eactions of common air- pollutants, such as NO, NOz and CO, with

cer-tain second reactants, such as ozone or- 0 atoms, r-esult in light emission.
Measur-ements of the emission intensity can be used to deter-mine the concen-
tration of the pollutants.
In a detector based on this pdnciple, ambient air- and
the second r-eactant can be continuously flown thr-ough and mixed in a r-eactor
unde r- mode r-ate vacuum.
After- calibration a continuous r-ecor-d of pollutant con-
centration can be obtained.
Specific sensitivity to a given pollutant is obtained
by a suitable choice of the second r-eactant and a light filter-.
To demonstr-ate the feasibility of the method, the detection of NO using
03 has been studied experimentally.
A linear- r-esponse from about 4 ppb NO to
at least 100 ppm NO is obtained.
NOz, COz, CO, CzH4' NH3, SOz and HzO in
concentr-ations found in polluted air- do not interfer-e with NO monitor-ing.
Based
on these r-esults and data for other- chemiluminescent r-eactions, it is shown that
homogeneous chemiluminescence monitor-s for 03' NOx = NO + NOz and CO can
probably also be developed.
A compar-ison of various methods of photomultiplie r- tube output measure-
ment has been made.
Though the ultimate sensitivity of lock-in amplification
methods may be higher, inexpensive dc measurement techniques appear adequate
for monitodng of pollutants over- the full range of concentr-ations of inter-est in
air quality control.
iii

-------
TP- 21 7
I.
II.
III.
IV.
T ABLE OF CONTENTS
SUMMARY
LISTS OF TABLES AND FIGURES
INTRODUCTION
EXPERIMENTAL
A.
B.
Flow System
Procedure
C.
D.
Light Detection System
Expe rience Gained from Preliminary Experiments
RESULTS AND DISCUSSION
A.
B.
The Reaction Between NO and 03

Linearity of Response and Limit of Sensitivity for
NO Detection
C.
Effect of Othe r Air Constituents on the Detector
Response to NO

Monitoring of 03, NOx = NO + NOz and CO
D.
CONCLUSIONS AND RECOMMENDATIONS
ACKNOW LEDGMENT
REFERENCES
v
Page
iii
vi
1
2
2
3
6
7
9
9
11
11
12
16
17
17

-------
TP-2l7
LIST OF TABLES
Table
Page
I
LACK OF IN TERFERENCE OF OTHER AIR
CONSTITUENTS WITH NITRIC OXIDE MONITORING
BY OZONE
13
LIST OF FIGURES
Figure  
1 CHEMILUMINESCENCE DETECTOR 19
2 GAS FLOW SYSTEM 20
3 EXPONENTIAL DILUTION PLOT 21
4 SPECTRAL RESPONSE OF EMI 9558 QA PHOTO- 
 MULTIPLIER TUBE 22
5 SPECTRAL DISTRIBUTION OF NO/03 EMISSION 23
6 DEPENDENCE OF THE RESPONSE OF THE CHEMI- 
 LUMINESCENCE DETECTOR ON NITRIC OXIDE 
 CONCENTRATION 24
7 EFFECT OF NOz ON DETECTOR RESPONSE TO NO 25
vi

-------
TP - 21 7
1.
INTRODUCTION
The homogeneous gas phase reactions of common air pollutants, such

as NO, NOz and CO, with certain second reactants, such as ozone or,O atoms,
are known to result in light emission.
These reactions have usually been
studied in continuous flow reactors. Measurement of the light intensity of the
reactions occurring when the pollutants are .mixed with a large excess t of the
second reactant should in principle be a suitable method for contin.uous moni-
toring of pollutants.
A priori calculations based on the published spectral
distribution of the light emitted, the rate constants for light emission and the
response characte ristics of photomultiplie r tubes, indicated that the light intensity
would be quite adequate for monitoring of pollutants Ove r the concentration
ranges of interest.
The suggested use of homogeneous gas phase chemiluminescent reactions

for monitoring purposes appears attractive for a number of reasons, particularly:
1.
The emissions are specific for the pollutant being monitored;
suitable choice of a light filter and the second reactant should allow
interference-free measurements.
2.
The chemiluminescent light intensities from homogeneous gas-ph~se
reactions in continuous flow systems are rather insens'itive to changes
in surface prope rties.
3.
A family of chemiluminescence monitors may be constructed, each
unit of which is specific for one pollutant, but all of which are simi-
la r in ope ration.
The convenience in the ope ration of monito ring
stations of families of instruments with simila r manipulation and
maintenance requirements would be conside rable.
t A large excess of second reactant is needed so that its concentration is not
measurably affected by the pollutants.
1

-------
TP-2l7
A schematic design for a chemiluminescence detector is shown in
Fig. 1.
The air to be monito red and the second reactant, e. g., ozone, ente r
the reaction vessel through separate i~lets.
Rapid mixing occurs and a chemi-
luminescent reaction takes place.
A preset flow of the gases is maintained by
a mechanical vacuum pump.
The pressure in the reaction vessel is typically
1 Torr and the size of the vessel 1 liter.
The intensity of the light emitted is
measured by a photomultiplier tube and associated read-out devices (current
meter and recorder).
Afte r calib ration with samples of known concentration,
a continuous reco rd of the concent ration of the pollutant in ai r can be obtained.
To demonstrate the feasibility of the method and to determine optimum

operation conditions, an experimental monitoring system has been b,uilt and
incorporated in an apparatus which allows for convenient change in flow rates
and pressure.
Our studies with this device have concentrated on the detection
of NO, using 03 as the second reactant; the lack of interference by other common
air constituents and pollutants with the response of the NO/03 system has been
established.
To arrive at conclusions regarding the cost of equipment neces-
sary for various monitoring requirements, a comparison has also been made
between various methods (ac and dc) of measurement of the light signals
gene rated.
II.
EXPERIMENTAL
A.
Flow System
A schematic of the Pyrex/stainless steel! coppe r flow system is shown
in Fig. 2.
Oxygen (Linde Aviators Breathing Grade) and Nitrogen (Matheson,
Pre purified), dried near atmospheric pressure by activated alumina, pass
through flow meters and needle valves into the low pressure part of the system.
On the high pres sure side the oxygen is ozonated (~ O. 5 %) t by a photolytic
t Relative concentrations in this report are expressed on a molar basis (v/v).
2

-------
TP-217
ozonator (OREC Model 03VI). t Nitrogen serves as the carrier gas for NO
and the othe r pollutants fo r which possible inte rfe rence with NO detection was
inve stigated.
The NO is introduced into a 3 liter spherical Pyrex exponential
dilution flask, containing a completely enclosed magnetically-driven stirrer.
The OZ/03 and Nz/pollutant streams are mixed in the 1 liter spherical Pyrex
reactor.
The flow is maintained with a 5 CFM Welch Duo-Seal vacuum pump.
The gas streams ente r the reaction flask through nozzle s having small
openings (1 to 2 mm diam ) in the direction perpendicular to the neck of the
flask.
The reactor has a 7.5 cm diam
flat quartz window facing the photo-
cathode of the photomultiplie r tube used for light intensity measurements.
The
reactor is coated externally with Eastman white reflectance paint (BaS04) and
is packed with MgO powder inside an aluminum box.
The use of this powder
further increases the available light intensity and provides for a light-tight
detection system.
B.
Procedure
The expe riments we re pe rformed at room tempe rature at the following
reactant concentrations (in moles liter-I): [Nz] = 2.7 x 10-5; [Oz] = 2.7 x 10-5;
[03] = 1.1 x 10-7. Total reaction pressure was held at 1 Torri ([M] = 5.4xlO-5
mole 1-1). The flow rates of Oz and Nz into the system we re 1.25 m1 (STP)
sec-I.
The ozone flow was measured by iodometric titration.
A visual check
t Higher ozone concentrations can be obtained from electrical discharges than
from common photolytic methods.
However, a photolytic ozonator is to be
prefe r red for the present application since electrical discharges tend to

produce NO, even when Nz is present only as a minor impurity in Oz'
i This pressure was chosen because at higher pressures (:::: 3 Torr) mixing

became unsatisfactory, while pressures below about O. 5 Torr could not be
maintained at convenient flow rates with the 5 CFM pump.
3

-------
TP - 21 7
of mixing conditions was pe dormed. t The results obtained in Section III. B
confirm that mixing conditions were satisfactory~
Nitric oxide concentrations we re varied ove r a range of 4 ppb to 100 ppm
of that ofNz (1.1 x 10-13 to 3 x 10-9 mole 1-1). The NOwas purified by pas-
sage over (i) activated alumina (for removal of BzO) and (ii)Ascarite (for
removalofNOz)'
Atmospheric pressure sam.ples were injected with a gas-
tight syringe through the rubber stopper of a 5 liter Pyrex predilution flask,
containing prepurified Nz at atmospheric pressure. After thorough mixing,
. .

samples containing NO in concentrations in the range of 10 to 105 ppm were taken
from this flask and injected through the rubber stopper injection port of the
exponential dilution flask of the main flow system, see Fig. 2.
An exponential
dilution flask is a stirred flask purged at i'i. constant volume flow rate.
Under
these conditions the concentration of a sample injected in the flask dec reases
exponentially 1:
C = Co exp( - Qt/V)
( A)
whe re, Co = Initial concentration
Q = Volume flow rate at the flask pressure
V = Effective volume of the dilution flask
t
= Time elapsed from start of dilution.
The pressure in the flask was maintained at 50 Torr, which gives a convenient
t For the visual tests, the OZ/03 stream was replaced by an Oz/O stream,
obtained by subjecting the 0z stream to a mic rowave discharge (powered by
a Raytheon 125 Watt mic rowave gene rator), rathe r than to the ozonator lamp.
o atoms we re used in these visualization experiments since a major portion
of the emission of the NO/O reaction is in the visible part of the spectrum,
while NO/03 emis sion is not visible, cf. Section III. A.
4

-------
TP-217
time constant (V/Q on the order of 150 see) for our mea"surements. t Some
typical dilution curves are shown in Fig. 3.
The recorde r gives a continuous
record of the system response;
these continuous records.
the points shown on the dilution plots are from
In a numbe r of expe riments the effect of the presence of othe r air con-
stituents and pollutants on the detector response to NO was investigated.
Best
regular grade gases were used.
COz, SOz, CzH4 and NH3 were taken directly
from cylinders.
CO was first passed through a liquid Nz trap for removal of
carbonyl compounds.
NOz was taken from a blackened Pyrex flask in which it
had been mixed with an equal portion of atmospheric air; this procedure was
followed to oxidize trace amounts of NO to NOz and to prevent photodissociation
of NOz.
These interference gases were injected in the same exponential dilu-
tion flask as used for NO.
Sepa rate predilution flasks we re used.
In early
experiments with NOz this gas was injected through the port marked "HZO" in
Fig. 2, using a second Nz flow system similar to that shown in Fig. 2.
This
procedure gave results identical to those obtained with the single N z flow system.
Water vapor also could be introduced with a second Nz flow system.
Up
to 75% of the total N z flow was dive rted through this system, which was designed
so that N z eithe r pas sed through a saturator or directly to the IIHzO port. II This
set-up allowed for rapid comparison between the detector response to NO with
75% saturated and dry Nz. The saturator was filled with HzO at 450C. Complete
saturation of Nz flowing through the HzO port was achieved, as evidenced by the
formation of water droplets in a room temperature trap, downstream from the
saturato r.
t The effective volume of the flask has to be determined experimentally.
In
our experiments it was found to be on the order of 2500 ml and somewhat
dependent on stirring rate.
The refore, Eq. A was useful mainly for planning
our experiments but could not be relied upon to calculate concentrations
flowing into the reactor.
See fu rthe r, Sections II. D and III. B.
5

-------
TP-217
Light Detection System
C.
The radiation from the reactor is filtered through a Corning'CS 2-60
'0
filter, cutting off radiation of ~ < 6100 A (see further Section III. A); its
intensity is measured with an EM! 9558 QA trialkali photomultiplie r tube. This
model was selected because of its relatively high sensitivity in the 6000 -
8750 A region, good signal-to-noise ratio (SiN) characteristics and its useful-
ness over a wide range of wavelengths.
The spectral response characte ristic
of the tube (S20) is shown in Fig. 4.
The tube is contained in a Products for
Research TE-104 thermoelectrically- cooled housing and maintained at about
-20oC, at which temperature an SiN some 30rders of magnitude higher than
at room tempe rature is obtained. 2 The cathode-to-anode voltage of 1400 V is
supplied by a Pacific Photometric Instruments Model 200 power supply.
The
tube output is measured by a Keithley Model 602 solid state electromete r- -to
which is added a home-built dc zero-offset device--and recorded with a Heath
EU-20B Servo-recorder.
The above direct current measurement arrangement was selected as a

result of a number of preliminary experiments in which a comparison was
made between the sensitivity of this method and that available from phase-
sensitive amplifiers.
Two such amplifiers were used: A versatile, high-
sensitivity model (Princeton Applied Research HR-8) and a relatively inexpensive
type (PAR Model 120).
To provide the modulation for these ac amplifiers, a
Brower Model 312 CM light chopper was placed in front of the photocathode.
This chopper has a 2-aperture 11 inch diam
steel blade and was used to pro-
vide a 100 Hz signal.
The reference signal is generated by a proximity switch
facing the choppe r blade.
For NO detection under the conditions described above
we found the respective noise levels to correspond approximately to:
Keithley electromete r:
PAR Model HR 8 lock-in amplifier:
4 ppb NO
1 ppb NO
PAR Model 120 lock-in amplifier
with Model 112 pre-amplifier:
10 ppb NO
Further improvement in the response of the HR-8 could probably have been
6

-------
TP- 217
achieved by the use of an ac zero-offset device (Model 123) made by PAR for
this lock- in amplifie r.
In view of the satisfactory relative response of the
electrometer there appeared to be no need for obtaining this device, although
in its absence we were unable to off-set (with the HR-8) the background signal
due to impurities in the ozone source.
The signal gene rated by this source was
the equivalent of about 5 ppb of NO.
Since we did have a ze ro off- set on the
electrometer, its use was preferred.
Moreove r, for an ultimate practical air
pollutant detector the use of a dc current measurement method implies a much
lowe r cost than achievable with a lock-in amplifie r/ chopper method. Since the
currents generated by 4 ppb NO are on the order of 10-9 A (see below), a dc
current measuring device less sensitive and less expensive than an electro-


mete r can be used in a practical unit.
Expe rience Gained from Preliminary Experiments
D.
In this section we discuss results from preliminary experiments which
aided in the design of the final experiments and may be useful for future
reference.
Reaction Vessel - Initially two Pyrex multihole nozzles were used in which the
holes had been punched with a hot tungsten wire.
03 destruction occurred at
the nozzle openings.
was therefore used.
A glass-blown single hole (z 2 mm diam ) ozone nozzle
Apparently satisfact:)ry mixing conditions in the reaction
vessel were obtained when this nozzle was combined with a multihole nozzle
for the Nz/pollutant inlet, and both nozzles faced each other in the neck of the
flask.
i. d. ,
The light intensity from the NO/03 reaction was measured in a 2. 2 cm
14 cm long cylinder (for end-on observation), a 1 liter sphere and a
5 liter sphere.
All these reaction vessels were of Pyrex glass externally
cove red by aluminum foil.
The highest SiN was obtained with the 1 lite r sphe re,
which was therefore selected.
7

-------
TP-217
Tempe rature variation over the range 20 to 2000C was investigated by
heating the reaction vessel. Heating to 500C had no appreciable effect on siN, .
but highe r temperatures adve rsely affected SiN due to inc reased radiation from
the vessel walls.
Exponential Dilution Flasks - Figure 3 show s a typical exponential dilution plot.
The concentration vs. time curves obtained with these flasks showed, independent
of the magnitude of the initial pollutant concentration, an exponential concentra-
tion decrease for about two to three orders of magnitude, after which the
concentration drops less rapidly. A similar behavior has been observed by
previous investigators. 3 This levelling-out of the decay curves can apparently
be attributed eithe r to gases absorbed on the walls that are slowly released or
gas pockets that are ineffectively stirred.
These deviations caused but minor
inconvenience since we could cove r any range of pollutant concentrations of
inte rest by using a number of injections at different initial concentrations, [NO] 0'
less than two orders of magnitude apart.
Nonetheless, we made a number of
expe riments to see whethe r the linear range of one exponential dilution run
could conveniently be inc reased.
Thus the volume of the flask was varied from
1 to 5 liter, a "Lif-O-Gen" GLX40 steel sphere was used instead of a Pyrex

sphere, several stirrers were used, the dilution pressure was changed from
10 to 100 Torr and the flasks were heated to 200oC.
None of these changes
extended the linear range significantly and the heating actually had a deleterious
effect.
Light Choppers - We originally ordered a light chopper, cooled PM housing and
lock-in amplifier from one source (Princeton Applied Research Corporation).
However, the housing obtained (made by Products for Research) did not fit over
the chopper motor.
As a result, the housing had to be moved some 10 cm away
from the reaction vessel; because of the inverse square law effect, this resulted
in unnecessary losses in light intensity.
We then obtained a Brower 312 CM
chopper which allowed the phototube housing to come within about 2 cm of the
reaction flask. Several chopper blades have been used (with the Brower unit).
8

-------
TP-217
The best results (one order of magnitude larger SiN than with the PAR chopper)
we re obtained with the 2-aperture, 11 inch diam
blade.
Ove r the range 5 to
300 Hz, no dependence of SiN on chopping frequency was found to exist (as long
as multiples of 60 Hz are avoided).
In the initial experiments with the Brower
choppe r, it was established that the exte rnal refe rence light signal used in this
unit generated negative signals which made it nearly impossible to measure NO
concentrations below about 100 ppb.
Therefore, the chopper was returned to
the factory for modification.
The refe rence signal in the modified unit is pro-
vided by a proximity switch activated by the steel chopper blade; this unit
operates completely satisfactorily, and was retained.
Photomultiplie r Tube and Tube Housing - The original socket delive red with the
cooled housing placed the photocathode 2.5 em further away than necessary from
the window of the housing, and hence the reaction flask.
A modified socket,
which placed the photocathode 2.5 em closer to the window, was used for the
final experiments.
However, there was no appreciable difference in SiN obtained
with the two sockets.
Variation of the photomultiplier tube voltage showed that the highest
values of SiN we re obtained at about 1400 V.
The manufacture r (EMI) supplie s a doughnut- shaped pe rmanent magnet
to be mounted in front of the photocathode for inc reased SiN. Z Howeve r, this
magnet actually dec reased siN.
This was due to the fact that the noise level of
the cooled phototube hardly changed, but the signal decreased appreciably
(probably due to the reduced aperture).
III.
RESULTS AND DISCUSSION
A.
The Reaction Between NO and 0.3
o.n the basis of the available lite rature, ozone was selected as a suitable
second reactant for NO. monitoring.
o.the r reactants, such as a. and H atoms,
also produce a rather intense chemiluminescence4, 5 with NO; however, they also
9

-------
TP-2l7
react rapidly6,7 with NOz, producing NO.
As a result, these atoms are suitable
for the determination of [NO] + [NOz], but not for that of [NO]. The slow
reaction of NOz with 03 generates only higher oxides (N03, NzOs). 8, 9
The reaction between NO and 03 has been investigated extensively by
Thrush et al. 10, 11 The chemiluminescence is due to:
NO + 03 - N Oz* + Oz
(la)
NOz* - NOz + hv
(lb)
The light intensity, II' is given by
II = 12 { exp(-4l80 ~ 300)/ RT } { [NO][03]/[M]} sec-l (B)

for the 6000 to 8750 A region for M = air. t The [M] appears in the denomina-
tor because the emitting species,NOz*, is quenched by M, i. e., any gaseous
species present. The relative intensity distribution of the emission spectrum
is shown in Fig. 5. It may be seen that no light is emitted below about 6000 A .
Comparison to the EMI 9558 QA phototube spectral response characte ristic,
Fig. 4, shows that the tube's peak sensitivity falls at wavelengths shorter than
6000 A. Therefore the use of the CS 2-60 filter does not appreciably interfere
with the sensitivity of the monitor for NO detection by 03, but decreases the
possibility of inte rfe rence by other pollutants. It may also be seen from Fig. 4
that the phototube response cuts off near 8750 A. Therefore, the II value given
by Eq. B is roughly that for the useful emission region of our experiments.
The rate constant for the ove rall reaction, as defined by
-~ =_!&Nol= k[NO][~]
dt dt '
(C)
is 1 x 1071 mole-l sec-l at room temperature. 10 This number is sufficiently
small to make consumption of NO by 03 in the reaction flask essentially
t The units of II obtained from Eq. B are, for example,
or quanta ml-1 see-I, depending on whether mole 1-1
ml-l concentration units are employed.
Einsteins 1-1 sec-l
or number of particles
10

-------
TP- 217
negligible (::::: 6%) under our operating conditions (residence time::::: O. 5 sec;
[03] = 1. 1 X 10-7 mole l-~). The absence of an appreciable change in reactant
concentration is a necessary prerequisite for a uniform light intensity in the
reactor, needed to obtain a linear response from the detector.
Linearity of Response and Limit of Sensitivity for NO Detection
B.
Figure 6 gives a composite plot of the detector response to NO.
It may
be seen (i) that the light intensity varies linearly with NO concentration over
the range of concentrations investigated (::::: 4 ppb to 100 ppm), and (ii) that
4 ppb represents the approximate limit of sensitivity when a dc detection
method (electrometer) is used.
The data were taken under identical operating
conditions.
Two types of data points may be distinguished, (i) those obtained
from the initial injection of NO into the dilution flask and (ii) points taken from.
exponential dilution plots, cf. Fig. 3, on the assumption that a decrease in
intensity by a factor x corresponds to a decrease in NO concentration by a factor
x.
It may be seen that both types of points fall along the same line, thus con-
firming the linearity of response with respect to NO concentration.
The scatte r
in the data of Fig. 6 increases with decreasing NO concentration.
This can
reasonably be attributed to a decreasing accuracy of sample preparation.
The range of concentrations of NO encountered in (and of inte rest to)
air quality control falls within the limitslZ 10 ppb to I ppm. The linearity of
re sponse and sensitivity of the detector thus appear quite satisfactory for its
use as a monitor of NO in air.
Effect of Othe r Air Constituents on the Detector Response to NO
C.
The possibility of interference by other commonly encountered air
pollutants/constituents was investigated by adding them to the N z/NO flow.
These constituents and the minimum concentrations in which they were to be
tested were selected in consultation with the NAPCA contracting officer, R. K.
Stevens. A typical test run is shown in Fig. 7 for the case of NOz. This figure
show s points obtained (i) in a normal NO dilution run and (ii) an NO run during
which NOz was injected.
Since the concentration of NOz decreased in the
11

-------
TP-217
dilution process (pre sumably at the same rate as that of NO) it was desirable
to repeat the NOz injections several times in the course of one NO run.
It may
be seen from this figure that the presence of 9 ppm NOz does not influence the
detector's response to NO down to about 6 ppb.
The same procedure as for
NOz was followed for COz, CO, CZH4' NH3and SOz.
The results are summarized
in Table I in terms of concentration of the constituent tested which was found
not to inte rfere with the nitric oxide signal at [NO] :S 10 ppb.
It may be seen
that these concentrations are at least equal- -and in most cases considerably
highe r- -than typical high concentrations of these compounds in polluted air.
By using concentrations of NOz, CZH4 and NH3 about two times highe r than given
in the last column of Table I, signals in excess of those obtained at 10 ppb of
NO were observed.
For the other compounds this column gives the highest
concentrations tested.
The HzO data were obtained in a different manner. They pertain to a
comparison between streams of Nz/NO (i) dried and (ii) 75% saturated with
water; no difference was observed in signals from these streams.
With the
experimental set-up employed, we could not use a 100% saturated stream
(some of the NO would have been absorbed in the water of the saturator). . How-
ever, it appears very unlikely that any major interference could occur due to
the fractional increase in [HzO] represented by the increased saturation.
D. Monitoring of 03' NOx = NO + NOz and CO
A major advantage accruing to the application of homogeneous chemi-
luminescent reactions for monitoring of air pollutants would be the use of a
set of similar instruments for a number of pollutants.
One can obtain an idea
of the likelihood of obtaining such multifunctional detectors by discussing a
few examples.
This is done here by using the experimental data obtained in the
present work and published spectral distributions and rate constants for light
emission of a number of chemiluminescent reactions.
12

-------
TP-217
TABLE I
LACK OF INTERFERENCE OF OTHER AIR CONSTITUENTS
WITH NITRIC OXIDE MONITORING BY OZONE
Constituent
Typical High
Concentration Encountered
in Air Pollution Monitoring t
Concentration Used at
which no Inte rference
Was Detected at L!i2J :::: 10 ppb

( ppm)

9
NOz
COz
( ppm )
0.5
CO
300
25
300
300
CzH4
NH3
1
0.5
5
9
SOz
1
25
-------------------------------------------------------------------------
H20
100% saturation
75% saturation
t Data from Stevens, 3 see also Tebbens. 12
13

-------
TP-217
~ - The NO/03 reaction can also be used to monitor 03, In this case one would
merely replace the Ozl03 flow line (Fig. 2) with a carrier gas/NO flow line.
The sensitivity of the method at [NO] :;; 1. 1 X 10-7 mole I-I (equal to the 03 .
concentration used above) would be the same as that found above for NO, i. e. ,
4 ppb, and interference effects would also be the same as for NO detection by
03,
The sensitivity could readily be increased by using higher NO concentra-
tions.
NOx = NO + NOz - As discussed in Section III. A, in the presence of 0 atoms
NOz readily yields NO on a 1: 1 basis.
The reaction
o + NO -+ N Oz + hv
(2)
can thus be used to determine [NOx]' The rate constant and spectral distribu-
tion of Reaction (2) have been determined by Fontijn, Meyer and Schiff.4 The
light is emitted in a continuum, stretching from 3875 A. well into the infrared.
Over the wavelength region for which the 9558 QA tube is sensitive, the rate
constant kz is about 2. 5 x 104 1 mole-I see-I. Oxygen atoms can be gene rated
at the same pressure and flow conditions as used in our NO/03 experiments
(above), by replacing the ozonator with a microwave discharge, placed on the
downstream side of the stopcock D in the Oz line, Fig. 2.
Previous work, e. g. ,
Ref. 4, indicates that under these conditions [0] in the reactor would be about
1 x 10-6 mole 1-1. Thus, the light intensity Iz = kz[ 0][ NO] :::: 2. 5 x 10- Z[ NO] see-I.
Using Eq. B of Section In. A, we obtain II = 1. 7 x 10-5 [NO] see-I, for our
NO/03 experiments. t Hence the sensitivity of the detector as an NOx detector
t These data may also be used to calculate the limit of sensitivity of the detector
for NO/03 emission in terms of quanta detected. The experimental limit of
sensitivity is 4 ppb NO, hence [NO] 1. = 4 x 10-9 x [Nz] = 1. 1 x 10-13 mole I-I
1m
and I = 1. 7 x 10-5 x 1. 1 x 10-13 Einstein 1-1 sec-I = 1. 9 x 10-18
lim
Einstein 1-1 sec-I = 1 x 106 quanta I-I see-I,
14

-------
TP- 217
using a atoms is some 3 orders of magnitude higher than it is as an NO detector
using 03,
It actually is somewhat higher yet, since the phototube sensitivity
decreases with increasing wavelengths. for the spectral region of interest in
comparisons of NO/O to NO/03 emissions, d. Fig. 4.
It must be recognized
that because of the possibility of inte de rence by emissions of reactions of
a atoms with other air pollutants, the full spectrum of the NO/O reaction may
not be available.
However, it appears that even if a relatively narrow wave-
length region is selected by the use of an appropriate filter, the sensitivity of
the detector for NOx by a would still exceed that of the NO/03 detector.
co - Carbon monoxide may be monitored by using its reaction with oxygen
atom s,
CO + a - CO2 + hv,
(3 )
which results ~n emission of the carbon monoxide flame bands.
This banded
emis sion spectrum falls in the 3000 - 5000 A region, d. Dixon13 and Pearse
and Gaydon. 14 The rate constant for light emission, k3, has been determinedt
by Clyne and Thrush15 as 1. 2 x 101 1 mole -1 see-I.
Hence at [0] = 1 x 10-6 mole
1-1, 13 = k3[ 0][ CO] = 1. 2 x 10-5 [CO] sec-l.
Comparison to our NO/03 results
gives 6 ppb as the approximate limit of sensitivity.
However, the phototube
response curve (Fig. 4) shows that in the wavelength region of the CO/O emis-
sion, the tube sensitivity is some 5 times higher than in the region of the NO/03
emission.
This then places the limit of sensitivity at about 1 ppb.
Again, inte r-
ference by other pollutants could in practice result in a higher useful limit.
t The rate constants for light emission of the NO/03 (kd and CO/O (k3) reac-
tions, quoted in this report, were measured by comparison to the NO/O
reaction in the same apparatus. The absolute values then were obtained10, 15
by using the directly measured absolute value for the NO/O reaction.4 The
fact that the intensity ratios used here depend on relative measurements made
in the same apparatus enhances their accuracy.
15

-------
TP-217
The natural atmospheric background levels of 03' NOx and CO are,
according to Tebbens, lZ probably approximately 10, 20 and 100 ppb, respec-
tively.
It thus appears highly likely th,at the homogeneous chemiluminescent
detector can be used for the monitoring of these pollutants over the full range
Of course~ experimental confirmation
of their concentrations in polluted air.
remains desirable.
Since the light-emitting reactions are of first order in the
pollutant concentration, linear responses can also be predicted.
16
IV.
CONCLUSIONS AND RECOMMENDATIONS
It may be concluded from this study that:
1.
It is possible to develop a homogeneous chemiluminescence detector
of NO in polluted air which is based on the NO/03 reaction and has
a linear response from about 4 ppb NO to at least 100 ppm NO.
2.
Other common air pollutants and constituents at the concentrations
found in polluted air do not interfere with the response of the NO
detector.
3.
It is highly likely that a homogeneous chemiluminescence detector
will be multifunctional. In addition to NO at the least 03' NOx =
(NO + NOz) and CO could be monitored with such a device.
To derive practical benefit from this study it is recommended that:
1.
An NO monitor based on the NO/03 reaction, now be made and
te sted in the field.
2.
Laboratory studies be undertaken to demonstrate the usefulness of
homogeneous chemiluminescence detectors for monitoring of other
air pollutants, such as 03' NOx and CO.
3.
Laboratory studies on these other pollutants, if successful, be
followed with development and testing of field monitors for these
pollutan ts.

-------
TP-217
ACKNOWLEDGMENT
This study has benefitted from discus sion with J. Hodgeson,
R. K. Stevens and A. E. O'Keeffe of NAPCA.
We also appreciate the advice
and assistance of several manufacturer's representatives, particularly E.
Fisher of Princeton Applied Research Corporation and M. R. Cole of Products
for Research.
REFERENCES
1.
Love1ock, J. E., "An Apparatus for the Calibration of Vapour Detectors, 11
Gas Chromatography 1960, ed. R. P. W. Scott (Butte rworths, London,
1960), p. 26.
2.
Anon., "EM! Photomultiplier Tubes, " Whittaker Corp. (Plainview, L. 1.,
N. Y.), 1967.
3.
Stevens, R. K., NAPCA, Private Communication, June 1969.
4.
Fontijn, A., Meyer, C. B. and Schiff, H. 1., "Absolute Quantum Yield
Measurements of the NO-O Reaction and Its Use as a Standard for Chemi-
luminescent Reactions, " J. Chern. Phys. 40, 64 (1964).
5.
Clyne, M. A. A. and Thrush, B. A.. "Mechanism of Chemiluminescent
Reactions Involving Nitric Oxide- -the H + NO Reaction," Discus sions
Faraday Soc. 2l, 139 (1962).
6.
Kaufman, F., "Reactions of Oxygen Atoms," Progress in Reaction
Kinetics.!., 1 (1961).
7.
Kaufman, F., l'Aeronomic Reactions Involving Hydrogen.
A Re view of
Recent Laboratory Studies, " Ann. de Geophysique~, 106 (1964).
17

-------
10.
11.
12.
13.
14.
15.
TP-217
8.
Ford, H. W., Doyle, G. J. and Endow, N., 'IRate Constants at Low Con-
centrations.
Rate of Reaction of Ozone with Nitrogen Diox~de, II J.
1.
Chem. Phys. l:!!.., 1336 (1957).
9.
Johnston, H. S. and Crosby, H. J., 'IKinetics of the Fast Gas Phase
Reaction between Ozone and Nitric Oxide, " J. Chem. Phys. ~, 689
(1954) .
Clyne, M. A. A., Thrush, B. A. and Wayne, R. P., "Kinetics of the
Chemiluminescent Reaction between Nitric Oxide and Ozone, II Trans.
Faraday Soc. 60, 359 (1964).
Clough, P. N. and Thrush, B. A., "Mechanism of Chemiluminescent
Reaction between Nitric Oxide and Ozone, 11 Trans. Faraday Soc. 63, 915
(1967).
Tebbens, B. D., IIGaseous Pollutants in Air, II Air Pollution, Vol. I, ed.
A. C. Stern (Academic Press, New York, 1968), Second Edition, Chap. 2.
Dixon, R. N., "The Carbon Monoxide Flame Bands," Proc. Roy. Soc.
(London) A275, 431 (1963).
Pearse, R. W. B. and Gaydon, A. G., The Identification of Molecular
Spectra (Chapman and Hall, London, 1963), Third Edition, p. 123.
Clyne, M. A. A. and Thrush, B. A., liThe Kinetics of the Carbon Monoxide
Flame Bands, " Ninth Symposium (International) on Combustion (Academic
Press, New York, 1963), p. 177.
18

-------
~v
~~ 'v~
~~v
~o ~
~~~ ~
~~ ~~c
~C' o~
G'~e>
r:1 ~oS' "';.

OPTICAL
WINDOW
CALIBRATED
LEAKS
REACTION VESSEL
0"

»

00
PHOTOTUBE
.-
-.0
~
PRESSURE
GAUGE
MECHANICAL
VACUUM
PUMP
F I60RE 6 I
Chemiluminescence Detector
68-89
POWER
SUPPLY
LIGHT
FILTER
READ-OUT

-------
I-
I
N
o
OXYGEN
NITROGEN
A
EXPONENTIAL
DILUTION FLASK
69-160
OZONATOR
B
POLLUTANT
INJECTION PORT
E
B
H O'
2
MAGNETIC STIRRER
THROTTLING
FIG. 2 GAS FLOW SYSTEM
A, abso rbing towe r; B, flowmete rf C, needle valve;
D, stopcock; E, manomete r

-------
TIME ELAPSED AFTER INJECTION. minutes
o 4 8 12 16 20
4
2
10-5
8
6
4
en ler6
" 8
'-
" 6
Q.
e 4

-------
N
N
28
69-156
 24 ~   
 0   
  z 2.0  
  !!l   
~  (.)   
0  iL   
 20 u..   
..  UJ 1.0  
>-    
(,)  :I   
Z  ;:::)   
lLJ  ~   
- 16 z   
~ c 0  
 ;:::)  
LL  a 0.8 1.0 1.2
LL 
w   WAVELENGTH, microns 
::E     
~     
I-     
z     

-------
> 
~ 
- 
CJ) 
z 100
I.&J
~ 
Z 
I.&J 
> 
- 
~ 
« 
..J 50
I.&J
a:: 
o
0.4
N
W
69-157
"A
V-..
0.8
1.2
2.0
1.6
2.4
2.8
3.2
WAVELENGTH, microns
FIG. 5 SPECTRAL DISTRIBUTION OF Nolo3 EMISSr"ON
Units proportional to quanta per wavelength interval per
second. Spectral distribution normalized to Intensity = 100
at A = 1.2,..... Data after Clough and Thrush, Ref. 11.

-------
10-4
10-5
 10-6
en 
cu 
... 
cu 
Q, 
e 

-------
69-153
 10-7  
 8  
(/)   
CI) 6  
'-  .c
CI) 
Co   Co
e 4  Co
~   ..
..  102 "0
...J   Z
~   ~
z 2  
C)  
-   
C/)   
 10-8  
 8  
 6  
 4  
  101 
 2  
DNO
o NO + N02
103
2
.
.
.
.
.
.
.
.
.
10-9
o
4
8
12
16
20
TIME t minutes
FIG. 7 EFFECT OF NOz ON DETECTOR RESPONSE TO NO
The arrows indicate the points at which 9 ppm NOz was injected.
25

-------