United State*
Environmental Protection
Agency
Environmental Sciences Research EPA-600/2-79-1 20
Laboratory July 1979
Research Triangle Park NC 27711
Research and Development
Mobile Source
NO. Monitor
Hydrogen-Atom
Direct
Chemiluminescence
Method
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL PROTECTION TECH-
NOLOGY series. This series describes rese3rch performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-79-120
July 1979
MOBILE SOURCE NOX MONITOR
HYDROGEN-ATOM DIRECT CHEMILUMINESCENCE METHOD
by
Arthur Fontijn, Hermann N. Volltrauer and William R. Frenchu
AEROCHEM RESEARCH LABORATORIES, INC,
PRINCETON, NEW JERSEY 08540
Contract No. 68-02-2744
Project Officer
Frank M. Black
Emissions Measurement and Characterization Division
Environmental Sciences Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, N. C. 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U, S, ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences Research
Laboratory, U.S. Environmental Protection Agency, and approved for pub-
lication. Approval does not signify that the contents necessarily reflect
the views and policies of the U.S. Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or rec-
ommendation for use.
11
-------
ABSTRACT
Work was performed to determine the feasibility of using the H/NO chemi-
luminescent reaction as the basis for a mobile engine exhaust gas N0x(= NO +
NO2) monitor which does not require a separate N0a to NO converter. This was
followed by the construction and testing of a prototype instrument, the design
of which is based on the findings of the feasibility study.
Following initial studies, a hot filament source was selected for H-atom
production. In the (feasibility study) test flow reactor a linear response
from 4 ppm to > 3000 ppm NOX in N2 was obtained, independent of the N0/N02
ratio. The test apparatus showed no interference by H20, CO, C02, C2Hf,, tolu-
ene, isopentane, NH3, HCN, CH3NH2, or H2 in the concentrations encountered in
raw exhausts. Positive interference occurred with 02; evidence was obtained
that this interference is due to wall reactions and a wall-less reactor was
therefore designed for the prototype instrument.
The use of the novel wall-less reactor made it necessary to perform ex-
tensive tests with the prototype and it was found necessary to obtain a trade-
off between sensitivity and negative and positive interferences occurring un-
der these conditions. The performance of the prototype, as delivered, is:
limit-of-sensitivity, 2 ppm NOX; linear response, to > 3000 ppm NOX; negative
interference per 1% C2H4 and per 20% 02 in the sample, 4% and 3% of NOX signal,
respectively; positive interference (in the absence of NOX) at these same con-
centrations, 6 ppm equivalent NOX, each. Graphs are provided herein, which
indicate how these factors can be varied.
As discussed, performance, though adequate for mobile source monitoring,
may be improved. Indications are that the H-atom direct chemiluminescence
method could be made useful for ambient air NOX monitoring conditions, where
potential interferences are far less severe.
This report was submitted in fulfillment of Contract No. 68-02-2744 by
AeroChem Research Laboratories, Inc., under the sponsorship of the U.S. Envi-
ronmental Protection Agency. This report covers the full technical performance
period from 7 September 1977 to 30 January 1979.
iii
-------
CONTENTS
Abstract iii
Figures vi
Tables vi
1. Introduction 1
2. Operating Condition Restraints 3
3. Feasibility Study - Experimental 5
A. Gas Handling System and Flow Conditions 5
B. Thermal H-Atom Source 8
C. Light Filter Selection 9
4. Feasibility Study - Results 11
A. Linearity of Response and Limit-of-Sensitivity 11
B. Interference Tests 11
5. Prototype - Description. 16
6. Prototype - Results 18
A. Positive 02 and C2H4 Interference 19
B. Negative 02, C2H<,, and C02 Interference 22
C. Limit-of-Sensitivity 26
D. Linearity 28
E. Selection of Operating Conditions 29
7. Discussion and Projections 31
References • • 33
-------
FIGURES
Number Page
1 H/NOX chemiluminescence analyzer test apparatus 6
2 H20 saturator 7
3 Thermal H-atom source 9
4 H/NO spectrum at 4.6 Torr 10
5 Response versus NCL. concentration 12
A
6 Schematic of prototype NOX analyzer 17
7 Positive interference by 02 versus 02 concentration in
absence of sweeper gas 19
8 Response to NOX and 02 versus sweeper flow rate 20
9 Positive interference by C2H4 versus C2H<, concentration 21
10 Negative C2H/, interference versus filament voltage 23
11 Negative C2HA interference versus sample flow rate 24
12 Negative C2Hfc interference versus C2Hi, concentration 25
13 Response to 4000 ppm NOX and limit-pf-sensitivity (left)
ordinate), and background signal (right ordinate)
versus filament voltage 27
14 Response versus concentration of N02 28
TABLES
1 Interference Investigations 13
vi
-------
SECTION 1
INTRODUCTION
Chemiluminescence monitors of NOX(= NO + N02) are currently based on the
N0/03 reaction.1'2 To obtain N02 measurements this compound is converted to
NO prior to entering the reactor. For mobile source exhaust observations
thermal/catalytic converters are used for this purpose.2'3 While these can
perform satisfactorily for all diluted engine exhausts, errors can result with
raw undiluted exhausts due to the reduction of NOX to N2. Additional errors
can result from oxidation, in the converters, of other N-containing exhaust
components (NH3, HCN) to NO. To avoid such complications it is desirable to
develop a method not requiring a separate converter. It was the goal of the
work described herein to establish such a method based on the reactions
H + N02 -*• NO + OH (1)
H + NO + M •> HNO + M (2)
H + NO -*• HNO + hv (3)
H + HNO -*• H2 + NO (4)
The emission from Reaction (3) is a series of bands between 628 and 800 nm.4"6
Reaction (3) leads to the expression for the Chemiluminescence light intensity,
I oc [H][NO] (5)
Reaction (1) is a very fast process,7'8 ki = 3 x 1010 8, mol"1 s"1, and hence
conversion of N02 to NO in the reactor is essentially instantaneous and the
equation governing the monitoring method is thus
I « [H][NOX] (6)
The approach followed in this work was to first establish feasibility
using a rack-mounted test apparatus (Sections 3 and 4) and then to build a
prototype instrument and optimize its performance (Sections 5 and 6).
An alternative approach, which had been considered prior to initiation of
this work, is the use of 0 atoms, based on the cycle
0 + N02 -> NO + 02 (7)
0 + NO + M -*• N02 + M (8)
-------
0 + NO •*• N02 + hv (9)
This approach has been used in two previous feasibility studies, i.e., those
by Snyder and Wooten9 and Black and Sigsby.10 Snyder and Wooten's work was
unsuccessful because the electrical discharge source used for 0 atoms itself
produced fluctuating quantities of NO which resulted in apparently unsolvable
background radiation problems. In the work of Black and Sigsby this problem
was solved by using thermal decomposition of Oa for the 0-atom production.
They obtained only a limited linear response range up to 50 ppm, inadequate
for the high NOX concentrations encountered in source monitoring. This
appears to be due to the low 0-atom concentrations achieved and the high
reactivity of 0 atoms toward other species present in mobile source exhaust.
By using small sample volume flow rates relative to the reagent flow rate it
might, however, be possible to extend the use of their method to the higher [NOX]
(3000 ppm) needed for exhaust monitoring. Thus this approach appears to offer
definite possibilities. However, though H-atom reaction kinetics has been less
thoroughly explored than that of 0 atoms, H atoms are a priori more attractive
since (i) the rate coefficients of reactions of H with hydrocarbons are typi-
cally an order of magnitude lower than those of 0 atoms (compare e.g., Refs. 8
and 11), (ii) H-atom chemiluminescent reactions are relatively rare (contrary
to those of 0 atoms), thus reducing the number of potential interfering emis-
sions, and (iii) [H] obtainable from thermal sources is of comparable magni-
tude to that from discharge sources.12"14 Since H2 is conveniently available
free of 02 and N2, electrical discharge methods to produce H should not lead
to the NO production problem encountered by Snyder and Wooten.
-------
SECTION 2
OPERATING CONDITION RESTRAINTS
The high concentration of various compounds in raw exhausts places a
priori restrictions on operating conditions. To evaluate these we had to
consider the contractually required maximum concentrations of each of these
compounds; these concentrations may be considered an upper limit to those
anticipated to be encountered in raw exhausts. These are: NOX (3000 ppm);
H20 (3%); 02 (20%); hydrocarbons (1%), taken as ethylene, toluene, and iso-
pentane; CO (7%); C0a (14%); nitrogen compounds (200 ppm), taken as NHa, HCN,
and CH3NH2; and H2 (2%). Here the numbers in parentheses refer to volume
concentrations in the sample gas to be considered. The maximum allowable
combined interference* with the NOX signal of all these compounds is 2%.
To determine the potential negative interference resulting from H-atom
reactions, we took the reaction conditions used (Section 3.A) with the thermal
source, i.e., P = 0.8 Torr, v" - 7.6 x 102 cm s""1 and a reagent (H/H2) to sam-
ple volume flow rate ratio of 100:1. Be moreover took the portion of the
reaction tube facing the center of the PMT as representative of what is being
viewed by the whole PMT, which: procedure, while not exact, certainly is valid
for estimating the w^giff t~infg of interferences. Thus, since the distance of
the sample inlet nozzle to the center of the 5 cm diam PMT is 4 cm, t = 5 x
10~3 s. The fastest bimolecular H-consumpt±oa reaction, with a potential inter-
ferant8 is that with ethylene, which has a rate coefficient (k) of 1 x io8 I
mol l s"1 at room temperature^ for these conditions this would result in a
decrease in [H] and hence in I, the chemiluminescence intensity, of 0.2%,
which is negligible. The only other bimolecular H—atom, reactions with the
above interf erants occurring at an appreciable rate are those with toluene
and isopentane which, however, have a lower rate coefficient than ethylene.8
Since total hydrocarbons can be assumed not to exceed 1% in the sample, the
0.2% consumption figure thus represents an upper limit. The NOx itself will
also consume H atoms. If we take all NOX to be N02 (which is not realistic
but again gives an upper limit) the H-consumption by Reaction (1) in the
100:1 diluted sample flow equals the total amount of N02 present, i.e., 3000
ppm x 10~2 = 30 ppm, which for 1% [H] in the reagent corresponds to an [H]
loss of 0.3%.
Negative interference would result from H-atom consumption (cf., Eq. (6)) or
emitter quenching, while positive interference would result from chemilumin-
escent reactions other than (3) which emit in the same wavelength region.
-------
H-atom consumption also occurs via the catalyzed termolecular cycle
represented by Reactions (2) and (4),* k = 5 x 109 £2 mol~2 s-1• H02 + M (10)
»*
H + H02 •* H2 + 02 (11)
the effective k for which cycle is17 4.2 x 1010 £2 mol~2 s"1 for M = H2. Thus
the H loss by this route is 8 x 10~2%, again negligible but the dominant cata-
lyzed H-atom loss process.^
These calculations, however, neglect the possible occurrence of emitter
quenching by some of the potential interferants. Such negative interference
could not be calculated a priori since no input data were available. The
tests of Section 6 show the negative interference by quenching to be a re-
stricting condition, but no more severe than the negative interferences by
H-consumption.
The chemiluminescent Reaction (3), though phenomenologically written as a
two-body process, is included in Reaction (2) and hence in k2 expressions.
Reaction (3) is thus actually a three-body process; however, the light emis-
sion is independent of [M] at the pressures of interest here, since the
emitter quenching is also proportional to [M].5
CO can also recombine H via a three-body process, H + CO + M -> HCO + M fol-
lowed by H + HCO -> H2 + CO. These reactions, however, lead to negligible
consumption since the governing rate coefficient8 is =^r 1 x 10~3 ki0 and
[CO] is not anticipated to exceed 7% of the sample.
-------
SECTION 3
FEASIBILITY STUDY - EXPERIMENTAL
A. Gas Handling System and Flow Conditions
The test apparatus is shown in Fig. 1. In the first part of the work H
atoms were produced by a standard (e.g., Ref. 18) 2450 MHz microwave discharge
in a 1.3 cm o.d. Vycor tube. The "reagent" H/H* from this source was found
to cause interfering emissions when hydrocarbons were present in the sample.
Later in the program therefore, the thermal H-atom source described further
in Section 3.B was used. The other parts of the test apparatus were made of
Pyrex, except for the valves, etc. and a copper tube coil between the sample
feed and inlet systems used to reduce strain on the Pyrex. The 2.2 cm i.d.
Pyrex reaction tube was poisoned with phosphoric acid to reduce H-atom wall
recombination. For convenience Dri-Film SC-87 (a silicone product obtained
from Pierce Chemical Co,) was used in the reagent feed lines. The Dri-Film,
however, deteriorated rapidly in the reaction tube causing response variations
and was therefore replaced with syrupy phosphoric acid. No wall poison was
applied to the. microwave «fi'ggl«|yg^ region, or the thermal source. The "sample"
NOX was introduced from a 2£ exponential dilution, flask, using ultrapure
grade N2 (dried by passage through activated alumina} as the diluent. The
alternate inlet and bypass line, Fig. 1, were used for the high concentrations
of NOX required for spectxometric observations (Section 3.C) and also in
selecting optimum flow conditions. The detector was a cooled («= -20°C) tri—
alkali PMT (Centronic P4283TIE); the light filter is discussed in Section 3.C»
Three right angle bends ("tees") separated the microwave discharge and thermal
H-atom source from the reaction tube. These bends served as light traps and
successfully prevented light from these sources from producing a PMT response.
Sample volume flows 1% of the reagent flows were used, cf., Section 2.
Optimum reagent flow conditions were established using an iterative procedure
by throttling the pressure line at various reagent volume flows and observing
the chemiluminescence intensities for NO volume flow «* 2 * 10~"s times the
total reagent volume flow. Using the microwave discharge, we thus selected,
except for a few preliminary experiments, a reaction zone pressure of 5.4 Torr
at 13.2 ml(STP)s~l and an average gas velocity of 250 cm s"1. For the thermal
source 0.8 Torr at 6 ml(STP)s~1 and an average gas velocity of 760 cm s"1 were
used. The distance from the sample inlet to the viewing area of the PMT was
1.5 cm. Some background emission was always present in the absence of added
NO; maximizing its intensity led to essentially the same optimum intensity con-
ditions. In experiments with the microwave discharge it was found that 10% H2
in He mixtures caused roughly a factor of 10 decrease in the background over
that observed with undiluted H2, essentially without affecting signal intensity.
-------
BYPASS LINE
COPPER COIL
3 WAY-VALVE
INJECTION PORT
O Q
NZ ALTERNATE
INLET INLET 1
EXPONENTIAL DILUTION
FLASK
-MAGNETIC STIRRER
REACTION TUBE
TO MANOMETER -^ET
O H-ATOM INLET
SAMPLE INLET
—E*T- SHUT OFF VALVE
9 METERING VALVE
TO PUMP
Figure 1. H/NOX chemiluminescence analyzer test apparatus.
ALTERNATE INLET 2
LIGHT FILTER
-------
Such mixtures were therefore used. This point was not investigated for the
thermal source and undiluted Ha was used. The 10% Ha in He mixture was pre-
purified grade having an impurity content of 1 ppm >• [02] ^ 0.3 ppm and "no"
N2, CO, G02. The undiluted Ha was ultrahigh purity grade (min. purity 99.999%)
Before passing through the H-atom generator these gases were purified further
using a MG Scientific "Oxisorb" cartridge for reduction of 02 to <: 0.1 ppm
and H20 < 0.5 ppm.
The pressure in the exponential dilution flask was maintained at •» 11
Torr for the microwave discharge experiments and at «* 3 Torr in the experi-
ments with the thermal H-atom source. Samples were injected into this flask
using syringes. NOa is partially associated to NaO* at atmospheric pressure
and*room temperature. The actual amount of N02 injected was calculated from
the volume of the NOa/NaOi, mixture, withdrawn as gas from the cylinder, using
JANAF equilibrium data.19
To check for H20 interference a room temperature saturator was used, as
shown in Fig. 2. The volume flow through the saturator was 75% of the normal
sample N2 flow and entered the reactor through alternate inlet 2, Fig. 1,
while the other 25% passed through the regular exponential dilution flask.
Thus the total level of H20 with which interference effects were checked was
TEFLON
WATER LEVEL
AIR STONE
MAGNETIC STIRREf
Figure 2. H20 saturator.
-------
2.25%.* Proper operation of the H20 saturator was ascertained by trapping
the H20 in an activated alumina trap and weighing.
B. Thermal H-Atom Source
Hot tungsten filaments offer an attractive alternative to microwave
discharges as H-atom sources. For a prototype instrument a thermal source
offered the advantage of requiring only a simple power supply, thus elimina-
ting the need to guard against electronic noise originating from the microwave
power supply. Thermal sources described in the literature (e.g., Ref. 20)
are water cooled, which did not appear practical for a prototype NOX monitor.
Thus the development of an air-cooled design was undertaken in the current
program. Since the extensive testing of the lifetime and stability of such
sources could not be done simultaneously with the investigations on the feasi-
bility of the NOX/H method (and at first appeared to represent an interesting
but non-essential part of the feasibility study), this testing was done on a
separate flow rig (available temporarily from work on another project) in
which the thermal source was mounted just upstream from a microwave discharge
cavity for ready comparisons; the reactor of this "test apparatus 2" was a
1JI stainless steel sphere coated with syrupy phosphoric acid.
Several designs were tested in the course of the program, with increasing
success. The design ultimately incorporated in the main test apparatus, Fig. 1,
for the final interference tests is shown in Fig. 3. With it dissociation was
achieved with a 5 turn 0.025 cm diam W wire operated at 14 A, 33 V ac, at a
temperature of »* 1900°C (as determined with an optical pyrometer). The fila-
ment was of ellipsoidal design to optimize the filling of the available space
of the *& 7.5 cm diam Pyrex sphere and maintain a reasonable distance from its
walls. The inner turn had a =* 2 cm diam, the outer ones «*> 1 cm. Cooling
was achieved with an air blower.
Incidental observations during the development work leading to the design
of Fig. 3 showed that it is absolutely essential to keep 02 and H20 out of the
H2 supply lines. When minor leaks occurred, these not only shortened the fila-
ment lifetimes but also led to metallic looking (W?) deposits downstream from
the source, which catalyzed H-atom recombination and hence severely affected
[NOX] measurement sensitivity. For the same reason the filament should not
be heated to higher temperatures than required for adequate sensitivity.
The thermal source of Fig. 3 operated 8 hours per day for 40 days without
a sign of deterioration and was thus considered to be a reasonably reliable
tool. However, further design improvements had to be carried out in the pro-
totype work (Section 5). The most important change required was to make a
*
This level is somewhat lower than the 3% requested, which corresponds to
saturation but could not practically be achieved in the test apparatus.
Since no interference with 2.25% was detected (Section 4.B) no significant
interference can be anticipated for 3% H20.
-------
TO REACTOR O
Figure 3. Thermal IE-atom source.
permanently sealed source rather than one having a ground glass joint. The
ground glass joint was essential for filament changing during source develop-
ment work, but became an impediment for quantitative experiments on the main
test apparatus, because of warming of the grease of the joint. The resulting
vapors, though apparently not affecting the filament directly, led to a con-
tinuous decrease in sensitivity (increase in the ppm MJ^ limit-of -sensitivity) .
Kel-F was found to be better in this respect t-ha^ Apiezon- SF.
C. Light Filter Selection
To aid in the selection of light filters for the PMT, spectra of the H/NO
emission were taken, e.g. Fig. 4. It may be seen from these that maximum sen-
sitivity can be anticipated by using X ^ 660 nm. The 762 nm peak is unsuit-
able since it is subject to interference by the peak intensity of 02 (12<»+)
emission in the presence of Oa.21'22 An available cut-off filter transparent
to X ^ 600 nm* was the only light filter used for all experiments (Section
4) , except those on 02 interference for which additional filters were used
as discussed in Section 4.B.
Little is gained by including the 600-660 nm region in which we found only
very weak H/NO emission. The easy availability of the X ^ 600 nm filter
was the main reason for its selection.
-------
660
680 700
Figure 4.
2.1 x IQ-14 mol £~l
720
740
760
780
X., nm
H/NO spectrum at 4.6 Torr.
NO
3.2 x 10~5 mol £"
v =
720 cm s'1.
H atoms produced by a microwave discharge,
-------
SECTION 4
FEASIBILITY STUDY - RESULTS
A. Linearity of Response and Limit-of-Sensitivity
Experiments were performed, using the microwave discharge source, to
check the linearity of response for NO, N02, and 50% NO-50% N02 mixtures.
Identical results were obtained, showing linear response from 2 to « 3000
ppm NOX. Hence the instrument response is independent of N0/N02 ratio. The
limit-of-sensitivity here is taken as two times the smallest signal observ-
able. A typical result is shown in Fig. 5. The solid line represents the
average of a series of three NO tests made at an earlier date than the dash-
ed line tests which pertain to the N0/N02 mixture. The magnitude of the de-
viations between the solid and dashed lines is comparable to that found in
the individual NO runs from which the solid line was obtained and appears to
represent injected sample size error and read-out uncertainty. Since the
concentration coordinate is calculated for exponential dilutions, errors in
pressure and flow rate readings will affect the slopes. For this reason the
slope of the line drawn through the initial injection points (circles) is
more reliable than those from the dilutions.
The limit-of-sensitivity indicated by these experiments is « 4 ppm
taken as twice the noise in the background signal. The best sensitivity
achieved with the thermal H-atom source was also about 4 ppm. No attempts
were made to generate plots like Fig. 5 with H from this source, mainly be-
cause of time/funding limitations and in view of the changes in response
from the ground glass joint source used, as discussed in Section 3.B.
B. Interference Tests
These tests were made by adding the potential interferants at two extreme
conditions: (i) with 3000 ppm NOx injections and (ii) with no NOX. The first
tests optimized chances for synergistic effects, the second for discovering
positive interference and influence on the background. Because of the poten-
tial for interference by the product OH of Reaction (1) , N02 was used as the
principal NOX for each interferant. Frequent checks with NO yielded indistin-
guishable results. The interference gases checked and the concentrations and
the H-atom source used are given in Table 1. For the gases checked at concen-
trations up to 200 ppm, co- injection with the NOX samples was used. The larger
concentrations used for the other interferants were obtained by substituting
the gas injection for the required fraction of the N2 dilution flask flow,
using alternate inlet 1, and comparison to 100% N2 flow through this flask.
11
-------
I02
c I01
UJ
CO
§
a.
CO
w
Of. mC
10
-I
10
,-8
1
10'
I01 I02 I03 I04
CONCENTRATION , ppm (v/v)NOx
Figure 5. Response versus
concentration.
• Average of three NO tests; --- Individual injections of 50%
NO-50% N02 mixtures; Q~ initial injection points; n~
points from exponential dilution.
12
-------
TABLE 1. INTERFERENCE INVESTIGATIONS
Interference
Interferant Concentration Tested H-Sourcea Observed
Ha 2% M no
NH3 200 ppm M no
CO 7% M no
C0a 14% M no
HCN 300 ppm M no
C2H* 200 ppm M yes
1% T no
Toluene 1% T no
jMr ^r&s
Isopentane 1% _ 3
1% T no
CH3NH2 200 ppm T no
02 20% T yes
H20 2.3% T no
a M = microwave discharge. T = thermal source.
Isopentane and toluene were tested by producing a 10% mixture in N2 in
a stirred predilution flask. These samples were then co—injected with
and compared to equivalent co-injected samples of pure N2 and NOX.
The first tests were performed with the microwave discharge. No inter-
ferences for Ha, NHs, CO, C02, or HCN were found, i.e., in the presence of
3000 ppm NOX the differences measured were well within the scatter of the
data of a series of NOX injections (« 5%), while in the absence of NOX, less
than the equivalent signal of 2 ppm NOX (in the background noise) was observed.
However, using the microwave source, positive interference with ethylene and
isopentane was observed. For either 200 ppm ethylene or 1% isopentane the
interference in the absence of NOX was the equivalent of 20 ppm NOX and in
the presence of 1000 ppm NO, the equivalent of 200 ppm NOX; i.e., the total
signal was then 1200 ppm NOX. These hydrocarbon experiments were therefore
repeated using the thermal source, which was also used for all further inter-
ference measurements. Using this source no interference with ethylene, toluene,
isopentane, CH3NH2, or H20 was observed. As discussed,the limit-of-sensitivity
13
-------
varied depending upon the conditions of the thermal source and was, for the
tests with C2Hj, < 60 ppm, toluene < 6 ppm, isopentane < 20 ppm, CH3NH2 < 0.4
ppm,* and H20 < 4 ppm NOX equivalent. (Several days elapsed between the final
tests on each of these compounds.)
Tests for 02 interference were performed with a Ditric 7415SP "cut-on"
filter (this filter has 42% transmission at 743 nm, 20% at 750 nm, and 1% at
760 nm, while the transmission from 600-730 nm ^ 75%). It thus appears high-
ly suitable for removing nearly all 762 nm 02 radiation,21'22 without affect-
ing the bulk of the remainder of the HNO radiation. The second filter in
these tests was a A ^ 680 nm transmission cut-off filter. With this filter
combination, 20% 02 caused a positive interference (tested in the absence of
added NOX) equivalent to « 800 ppm NOX; the'NOx equivalent signal was linearly
proportionally less for smaller [02]. This positive interference level was
the same regardless of whether the 02 was introduced through alternate inlet
1 or 2, i.e., through the center inlet in the reaction tube or the outer in-
let 5 cm upstream from it, Fig. 1. This is clearly unacceptable.''" The tests
were therefore repeated using a Corion 692 nm, 13.4 nm FWHM interference fil-
ter without collimator.t This reduced the 20% 02 signal to = 300 ppm NOX
equivalent. Since some 760 nm radiation could still have been transmitted
by this filter, it was then used in combination with a collimator (1 cm long
with 0.3 cm diam holes) to remove any such radiation. This reduced the 20%
Oz level to *& 100 ppm NOX equivalent. It can only be concluded that some 02
emission occurs over the full wavelength region of H/NO emission. While
Wayne et al21 only discuss a peak at 762 nm, their Fig. 2 suggests that a
weak tail of this emission extends down to at least 450 nm. We first assumed
that a tail of such low relative intensity caused the observed interference.
To discriminate against this positive 02 interference an approach similar
to that used in the 0/hydrocarbon chemiluminescence analyzer18'23'24 was at-
tempted, i.e., to subtract the output of two PMTs viewing the same section of
the reaction tube at different wavelengths. These tests were carried out using
the prototype hydrocarbon analyzer instrument18 in which the 0-atom source was
replaced by the H-atom source of Fig. 3 and two Centronics 4283 PMTs were used.
The Ditric 7415SP filter in combination with a Corning 2-64 filter was used
for one of the PMTs. Through this filter combination the PMT thus viewed the
This figure was arrived at in a test with 20,000 ppm CH3NH2, 100 times high-
er than required, during which the NOX limit-of-sensitivity was ^ 40 ppm.
These results pertain to thermal source experiments. Similar results were
obtained using the microwave discharge.
T When no collimator is used, higher light intensities at the PMT cathode are
achieved than when a collimator is used. However, without collimation,
light enters the interference filter at angles other than the perpendicular
incidence to which its nominal operation specifications apply, which results
in some transmission at wavelengths other than those specified.
14
-------
approximately 660-740 nm region. The other PMT, supplied with a Corning
7-79 filter, viewed the wavelength region from «*• 750 nm to *» 800 nm, the
approximate cut-off of the PMT. The reaction tube in this instrument was
essentially the same as in the other H/NOX tests described above and was
coated with phosphoric acid. In these two-PMT tests a sample flow rate of
0.01 ml(STP)s~1 was used; the reagent flow rate was varied from » 5-15 ml
(STP)^"1 with a concomitant change in reactor pressure from 0.8 to 2.5 Torr.
Both pure H2 and 10% H2 in He were used for the reagent flow. The results
were unexpected, i.e., for 02 as the sample gas both PMTs gave about the
same response. Since the H/NOX response by both PMTs was also approximately
equal (2.3% NOX in N2 was used for the sample in this work), it was concluded
that the two-PMT method was not useful here as it would require subtraction
of signals of comparable magnitude.
The two-PMT tests nonetheless pointed the way to a solution of the prob-
lem in that they showed the H/02 signal to be unstable and poorly reproduc-
ible—changes on the order of 50% on various days were not uncommon—while
the H/NOX response was very stable and reproducible. Such behavior is typi-
cally what may be anticipated for heterogeneous (wall) reaction. The occur-
rence of such a process can also explain why the tail of the H/02 emission
in this work is so intense relative to the peak at 762 nm. (Why a similarly
intense continuum was not observed under the conditions of Wayne et al21
still needs to be explained.) It was therefore decided to construct a "wall-
less" reactor. This reactor was incorporated in the H/NO^ prototype instru-
ment and is described in Section 5. The tests with the prototype instrument
showed a decrease in the 02/NOX signal ratio by two orders of magnitude in
the 690 mn region, cf. Section 6. This decrease reduces 02 interference to
acceptable levels and further confirms the H/02 wall chemiluminescence
hypothesis.
15
-------
SECTION 5
PROTOTYPE - DESCRIPTION
A schematic of the prototype NOX analyzer is shown in Fig. 6. The ele-
ment most different in principle from the test apparatus is the "wall-less"
reactor, which will therefore be discussed first. It consists of two 2.2 cm
i.d. lengths of reaction tubing separated by a «* 1 cm gap. These tubes are
contained in a 1£ stainless steel sphere to provide vacuum. The upstream
section of the reaction tube is constructed of Pyrex and has a side arm through
which the H/H2 enters the reactor, while the downstream tube is made of alu-
minum. The Pyrex tube is poisoned with phosphoric acid and is painted black
on the outside. Sweeper gas can be added in the upstream section of the reac-
tor. This gas flows out of the sphere through the gap between the reaction
tubes and thus can be used to prevent the bulk of the reacting mixture from
reaching the walls of the sphere. A 5 cm diam window on one side of the sphere
permits viewing the gap, which thus represents the observed reaction zone. The
bottom of the sample inlet nozzle is situated *= 0.5 cm upstream from this gap.
The general gas flow system, Fig. 6, is similar to that of the feasibility
study. H2 (ultrahigh purity or prepurified grade) flows through traps to re-
move 02 (Englehard Deoxo catalytic hydrogen purifier) and H20 (activated alu-
mina) . It is then partially dissociated in the thermal H-atom source. The
thermal source used here was redesigned from that used in Section 3.B, mainly
to eliminate the ground glass joint. The source, shown in Fig. 6, consists
of a 20 cm long, 5 cm o.d. Pyrex envelope. The active element is a «* 10 cm
long, 0.35 cm diam helical filament made of 0.025 cm diam tungsten wire. The
filament, which has a resistance of « 0.7 J2, draws « 8 A at 40 V. Relative-
ly large size tubing (1.6 cm o.d. Pyrex) is used between this source and the
reactor to minimize pressure drop. This connecting tubing is poisoned with
phosphoric acid. A 10 CFM vacuum pump was used.
The trialkali PMT (Centronics P4283) views the reaction zone through two
filters (Ditric 7415SP and Corning 2-60) with a combined bandpass from =» 640
to 740 nm. No collimator is used, which results in relatively large signal
and background response. The advantage of this is that the dark current con-
tribution to the background is negligible and hence, that no PMT cooling is
required.
16
-------
SAMPLE
OEOXO
PURIFIER
O
)•=•(
H - ATOM
SOURCE
ALUMINA
DRYING C*3
TrttAJtO '
TOWER
40 v ac
AIR
SWEEPER
FILTER
METERING VALVE
SOLENOID VALVE
SONIC ORIFICE
SPLIT REACTION TUBE
Figure 6. Schematic of prototype NOX analyzer.
-------
SECTION 6
PROTOTYPE - RESULTS
Tests were made with the prototype instrument to allow performance opti-
mization and to define the various trade-offs that can be made between level
of interference and sensitivity. While it would be useful to vary all param-
eters (e.g., flow rates, geometry, etc.) independently over a wide range of
values, time limitations precluded this. Carefully selected tests were there-
fore made which dealt mainly with (i) positive 02 and C2Hi, interference due to
chemiluminescent reactions, (ii) negative 02 and C2H* interference due to H-
atom consumption, (iii) negative C02 interference due to quenching of the HNO
emitter, (iv) sensitivity, and (v) linearity. The results of these tests make
it possible for the operator of the instrument to choose operating conditions
that best suit a particular application. Test results dealing with interfer-
ences, sensitivity, and linearity are presented in Sections 6.A-D. The final
tests and the set of operating conditions selected for the delivered prototype
(on the basis of these tests) are discussed in Section 6.E.*
The tests of Sections 6.A-D were carried out by choosing a set of "stan-
dard conditions" and varying one or two parameters while the others were at
their "standard" value. The standard conditions for these tests were: sam-
ple flow rate, 2 ml(STP)s~1; sweeper (air) flow rate, 0.4 ml(STP)s~i; hydrogen
flow rate, ** 18 ml(STP)s~1; reactor pressure, 3.5 Torr; and filament voltage,
40 V at *& 8 A. Some of these choices were based on preliminary tests (e.g.,
the 0.4 ml(STP)s~1 sweeper flow rate was found to be necessary to obtain lin-
ear response). However, the 2 ml(STP)s~l sample flow rate was chosen because
it was initially thought that adequate sensitivity could not be achieved at
the 0.1 ml(STP)s~l sample flow rate of the feasibility study. Therefore, the
original aim was for a prototype instrument with two sample flow rates: 0.2
ml(STP)s~1 for raw exhausts and « 2 ml(STP)s~l for bag samples. Since lower
flow rate performance can more reliably be predicted from higher flow rate
results than can the reverse, tests were initiated with the 2 ml(STP)s~1 rate.
The higher than expected sensitivity that was achieved (Section 6.C) made the
two sample flow rate approach unnecessary and the delivered prototype has only
one sample flow rate (0.25 ml(STP)s~1), suitable for both raw exhausts and bag
samples. The effects of the three interferants tested (C2H/,, 02, and C02)
were assumed to be additive—all tests were therefore made with N2 as sample
carrier gas.
The data of Sections 6.A-D are essentially of use only to those who might
want to change the conditions established in Section 6.E. The more general
reader may wish to proceed directly from here to Section 6.E.
18
-------
A. Positive 02 and C2H4 Interference
These tests were performed in the absence of NOX. As concluded In Section
4.B the major outstanding positive interference problem to be investigated was
that caused by 02, i.e., the magnitude of its NOX equivalent signal. Figure 7
shows that the response of the instrument toward 02 with no sweeper flow is
relatively large (30 ppm NOX equivalent for 20% 02) and nonlinear. When sweep-
er air is added, Fig. 8, the positive 02 interference drops rapidly, while the
response toward NOX drops much more slowly*; hence, the positive 02 interfer-
ence in terms of NOX equivalent signal is lowered. Because the background also
rises as the sweeper flow is increased, the sensitivity drops more rapidly than
the signal. The lowest possible sweeper flow rate should therefore be used.
The 0.4 ml(STP)s~1 flow rate decided on for further tests is also sufficient
to achieve linearity (see Section 6.D).
60
40
w
$ o
u z
ft
UJ >
u. o-
ft «
UJ _
il
Ul
8
Q.
20
I
0 5 10 15
02, % of sample
Figure 7. Positive interference by 02 versus 02 concentration
in absence of sweeper gas.
Flows in ml(STP)s~1: sample, 2.0, H2, 18; reactor
pressure, 3.5 Torr; filament voltage, 40 V.
Under some conditions the opposite occurs (cf., Fig. 14)
19
-------
-•? 4000
JO
w
O
UJ
to
z
o
0.
CO
UJ
IT
O
z
2000
16
X
O
M
O
c
8
UJ
OT
o
a.
s
oc
N
O
O.I 0.2 0.3 0.4 O.5
SWEEPER AlR FLOW RATE, ml (STP) •''
Figure 8. Response to NOX and 02 versus sweeper flow rate.
Flows in ml(STP)s~1: sample, 2.0, H2, 18;
reactor pressure, 3.5 Torr; filament voltage,
40 V;A - 4000 ppm NOX, Q - 1.8% 02,
O- 3.6% 02, O - 8^ °2-
20
-------
The positive interference by C2H4 is smaller in magnitude than that by Oz
in the absence of sweeper air. A 1% C2HA concentration results in about a 2
ppm NOX equivalent interference, well below the level detectable in the experi-
ments of Section 4.B. These results are shown in Fig. 9. The results of Sec-
tion 4.B suggest that the largest positive hydrocarbon/amine interference would
be expected for C2Hi, and this 2 ppm NOX level thus represents an upper limit
for total positive hydrocarbon/amine interference.
0.5
1.5
1.0
% of sample
Figure 9. Positive interference by C2H<, versus C2HA concentration.
Flows in ml(STP)s-1: sample, 2.0, sweeper, 0.4, H2, 18;
reactor pressure, 3.5 Torr; filament voltage, 40 V.
21
-------
B. Negative 02, C2H4, and C02 Interference
While the feasibility work (Section 4.B) showed no negative interferences,
operating conditions were, as noted, varied in the prototype work to obtain a
trade-off between sensitivity and positive and negative interferences. Reaction
time for the standard test conditions is roughly comparable to that for the
feasibility work. However, the total pressure is higher and hence the partial
pressure of the potential interfering reactants in the reaction mixture at the
same dilution ratio is also increased. This necessitated some further evalua-
tion of negative interference. C2H4 and 02 were selected for these tests be-
cause of their potential interference by H consumption, cf. Section 2. C02,
which at the 14% concentration level is the most probable interferant by quench-
ing of HNO emission,* was tested in the work of Section 6.E only.
Since under our operating conditions [H] is in excess over [C2H4], the
absolute [H] consumed by this interferant, A[H], is independent of [H]. There-
fore, the higher the [H], the smaller percentage interference to be expected.
A priori, it can also be assumed that the higher the H-source filament voltage,
the higher the [H], cf.. also, Section 6.C. Figure 10 shows the % negative inter-
ference of C2H4 as a function of filament voltage. It demonstrates that, under
the test conditions, interference by C2H4 occurs and that increasing filament
voltage serves to minimize its effects. Also given in Fig. 10 are three mea-
surements, made at an increased H2 flow rate (30 ml(STP)s~1); these show that
a decrease in C2H4 interference results. This can be attributed to a higher
[H] at the higher H2 flow rate.
The effect of sample flow rate on C2H4 interference is shown in Fig. 11,
both at 40 V and the standard H2 flow, and at 50 V and the higher H2 flow. As
expected, the interference is roughly proportional to sample flow rate and
therefore [C2H4] in the reactor. A more direct measure of C2E4 interference
as a function of its concentration in the reactor is shown in Fig. 12, where
the [C2H4] in the sample is varied from 0 to 1%. A small decrease in C2H4
interference is effected by adding sweeper gas, as shown in the figure.
02 also causes a negative interference due to H-atom consumption. A few
measurements were made which showed that the interference by 20% 02 is about
equal to that by 1% C2H4 at 2 ml(STP)s~1 sample and 0.4 mKSTPjs"1 sweeper flow
rates. The C2H4 interference tests should therefore approximate the Oz inter-
ference fairly well.
*
Polyatomic molecules such as C02 are generally better quenchants than
diatomic molecules such as 02.
22
-------
I
CM
o
60
UJ
O
z
Ul
It
Ul
u,
(T
UJ
UJ
o
UJ
z
40
20
25
30
35
40
45
50
FILAMENT VOLTAGE, V
Figure 10. Negative C2H«, interference versus filament voltage.
Flows in ml(STP)s~1: sample, 2.0, sweeper, 0.4, H2,
O- 18, O ~ 3°5 reactor pressure, Q- 3.5 Torr,
D- 6.0 Torr; filament voltage, Q - 40 v» D ~ 50 v-
23
-------
CM
u
ffl
UJ
u
UJ
ac
UJ
u.
oc
III
t a*
UJ
(9
UJ
0.5 1.0 1.5
SAMPLE FLOW RATE, ml (STP) s'1
Figure 11. Negative C2Hj, interference versus sample flow rate.
Flows in ml(STP)s~1: sweeper, 0.4, H2, Q- 18, Q- 30;
reactor pressure, Q- 3.5 Torr, Q- 6.0 Torr;
filament voltage, Q ~ 4° V, Q - 50 V.
2.0
24
-------
I
CM
o
CD
UJ
O
z
LU
ft
UJ
u.
tr
UJ
H
Z
UI
UJ
z
a
c
% of sample
Figure 12. Negative CsHu interference versus CaR* concentration.
Flows in ml(STP)s~1: sample, 2.0, sweeper, Q- no flow,
Q- 0.4, H2, 18; reactor pressure, 3.5 Torr; filament voltage, 45 V.
25
-------
C. Limit-of-Sensitivity
The parameters that most affect the sensitivity of the instrument are the
H-source filament voltage and the sample flow rate (which control the [H] and
the [NOx], respectively, in the reactor). A third parameter, which has a less-
er influence, is the Hz flow through the H source. It is more desirable to
obtain a high sensitivity by increasing the filament voltage than by increasing
the sample flow rate because the former will not cause an increase in the 02
and C2H« negative interference whereas the latter will.
To study the effect of filament temperature, the filament voltage was var-
ied by placing a variable autotransformer between a constant voltage transform-
er and the stepdown transformer which powers the thermal source. In general,
an increase in filament voltage will increase the temperature of the thermal
source and increase H-atom production. The effect is to increase both the sig-
nal obtained for a given [NOX] and the background (for reasons unknown, but not
related to light leakage). Figure 13 shows the signal (SIG) and background
(BG) readings obtained as a function of filament voltage. Also shown is the
limit-of-sensitivity, S, measured at 40 V.* The limit-of -sensitivity at other
V is estimated by scaling the measured sensitivity at 40 V, S40, according to
O - Q
S ~ S* BG
where SIG40 and BGi,o are the signal and background measured at 40 V, respective
ly. For this calculation it is assumed that the noise increases as the square
root of the background and reduces the sensitivity accordingly. Thus, even
though the signal increases rapidly up to the highest voltage measured (50 V),
little can be gained above 50 V because the background increases even faster.
Some gain beyond 50 V may, however, be obtainable when the negative 02 and
C2H/, interferences (Section 6.B) are considered. This may come about because
a lower interference at a high filament voltage would make it possible to op-
erate at a higher sample flow rate for a given interference level, when the
filament voltage is increased. However, beyond a certain voltage the filament
begins to evaporate and coat the thermal source walls with tungsten, t At that
point the sensitivity decreases sharply due to H-atom recombination on the
source walls, cf., Section 3.B.
The H2 flow rate through the thermal source has a relatively small effect
on response. Under the standard conditions, an increase of H2 flow to 30
ml(STP)s~1 results in a «= 20% increase in signal.
The limit-of-sensitivity was taken to be twice the short term peak-to-peak
noise level in ppm equivalent NOX obtained with room air as sample.
Such a destructive test was attempted using a similar H-atom source as that
used in the above tests but with a different filament. In that earlier
test a decrease in signal ultimately resulted from an increase in filament
voltage. It is estimated that the voltage at which the signal begins to
decrease would be « 65 V for the filament used in the present tests.
26
-------
4800
4000
«0
^
'c
>• 3200
o
V)
z
o
Q.
V)
UJ
a:
X
O
Z
2400
1600
800
O1-
30 40
FILAMENT VOLTAGE, V
50
Figure 13. Response to 4000 ppm NOX and limit-of-sensitivity
(left ordinate) and background signal (right ordinate)
versus filament voltage.
Flows in ml(STP)s~1: sample, 2.0, sweeper, 0.4, H2, 18;
reactor pressure, 3.5 Torr.
27
-------
D. Linearity
Linearity tests were carried out using a. 5000 ppm N02 in N2 sample, diluted
progressively further with Na. Measurements were made by comparing readings
from the H/NOX instrument to those of an AeroChem built NO/NOx/03 analyzer mod-
ified to measure high [NOX] (up to 1%). Many previous tests on these modified
instruments have shown them to be linear to well past the 1% level. As the
measurements of Fig. 14 show, the H/NOx prototype instrument also has a linear
response up to at least 5000 ppm under the conditions of these tests. Sweeper
flow is, however, required to achieve linearity. With no sweeper flow the in-
strument response was found to be highly nonlinear; the ratio R = (H/NOx) sig-
nal/ (Os/NOx) signal varied from 0.77 at 40 ppm (signals in arbitrary units but
consistent throughout the test) to 2.0 at 5000 ppm. N2 sweeper at 0.09 ml(STP)
s~l reduced the deviation, Fig. 14, with R changing from 1.46 at 35 ppm to 1.82
at 5000 ppm. Finally, with a sweeper flow rate of 0.36 ml(STP)s~1 of N2, a
Linear response to within ± 4% was obtained from the 1.09 at 17 ppm to 1.00 at
I04
X
O
Z
E
o.
Ul
0}
1
m
til
(T
ui
oc
I-
0}
Z
103
102
10'
I
I01 I02 I03
CONCENTRATION, ppm NO2
Figure 14. Response versus concentration of N0a.
Flows in ml(STP)s~1: sample, 2.0, H2, 18,
sweeper, O none,Q— — — —0.09,
A • 0.36; reactor pressure, 3.5 Torr;
filament voltage, 40 V.
28
-------
5000 ppm. The reason for the nonlinearity at low sweeper flow rates is
probably the flow disturbance caused by the gap in the split reaction tube.
E. Selection of Operating Conditions
From the tests discussed in Sections 6.A-D, tentative operating conditions
were selected for a reasonable compromise between sensitivity and interferences,
as follows: 0.4 ml(STP)s~1 sample and sweeper flow rates; 18 ml(STP)s~l H2
flow rate; 3.5 Torr reactor pressure; 45 V filament voltage at 10 A; and a
separation of 0.5 cm between the tip of the sample inlet nozzle and the down-
stream end of the upper reaction tube. However, prior to final testing, the
instrument was cleaned and carefully re-assembled. Measurements with the in-
strument thus improved showed the response towards NOX as well as the back-
ground to be much larger than before the cleaning, the negative C2H<, interfer-
ence to be absent, and the positive C2H<, interference to be very large ( « 100
ppm equivalent NOX).* These results can be attributed to a much larger [H]"^
than was obtained prior to this overhaul.f To minimize the positive C2H4 inter-
ference a number of changes in operating conditions were made, the most impor-
tant of which was to increase the distance between the tip of the sample inlet
nozzle and the observation region to 8 cm. This change reduces the positive
interference by CsKu by reactively removing this compound. § A decrease in
response to C2Hi, can also be effected by reducing [H]; the filament voltage
therefore was reduced (to 40 V). The larger H-atom consumption by both C2EU
and 02, thus resulting from the longer reaction distance and hence time, in-
evitably manifested itself in increased negative C2H<, and 02 interference.
It is probable that a behavior of the instrument very similar to that ob-
tained in the tests of Sections 6.A-D can be re-obtained by a major decrease
in filament voltage, and thus in the [H], to the levels of those tests.
Titration8 of H atoms with a 0.5% N02 in N2 sample at a 14 V filament voltage
showed the [H]/[H2] to be «* 4 x 10""*. The instrument response to NOX at
50 V is « 500 times larger than at 14 V which implies a [H]/[H2] = 0.2 at
50 V (at 12 ml(STP)s~1 H2).
^ The large positive C2H<, interference, reported in Section 4.B for experiments
with the microwave discharge H-atom source, can also be attributed to high
[H] obtained with that source.
§ At 40 V, 0.3 ml(STP)s~1 sample flow, and 0.2 ml(STP)s~1 sweeper flow, the
positive interference by 1% C2H<, dropped from 12 ppm NOX equivalent at a
nozzle-to-observation zone distance of 3 cm to 8 ppm at 5 cm and 4 ppm at
8 cm.
29
-------
This interference was therefore reduced by decreasing the sample flow rate to
0.25 ml(STP)s~x, which was found to provide a reasonable compromise between
negative interference levels and sensitivity.
The operating conditions finally used in the delivered prototype instrument
are: sample flow rate, 0.25 mlCSTPjs"1; sweeper flow rate, 0.4 ml(STP)s~1; H2
flow rate, 12 ml(STP)s~1; filament voltage and current, 40 V and 8 A; reactor
pressure, 2.5 Torr; sample nozzle-to-observation zone distance, 8 cm. Under
these conditions the performance of the instrument (again for N2 sample gas)
is the following: limit-of-sensitivity 2 ppm NOx with a 3 s electronic time
constant and 6 ppm NOx f°r a 0.3 s time constant; positive interferences for
1% C2H4 and 20% Oa 6 ppm equivalent NOx each; negative interferences for the
same quantities of C2H<, and 02, 4 and 3%, respectively; negative interference
for 14% C02, less than 3%. The response of the instrument has been found to
be linear to within ± 2% from 6 ppm to at least 3000 ppm N02. Comparison to
the work of Section 6.D suggests that linearity would be retained to at least
eight times the limit investigated there, i.e., to 4% NOX. This is so since
an eight times smaller sample flow rate is used here than that of Section 6.D,
and the H consumption by N02 is the factor determining the upper limit. Be-
cause NOX concentrations higher than 3000 ppm are not of practical interest,
this point was not re-investigated.
30
-------
SECTION 7
DISCUSSION AND PROJECTIONS
While the operating conditions arrived at from the tests have resulted in
a prototype instrument that is adequate for mobile source monitoring require-
ments, improvements can be made that will result in improved sensitivity (for
a given interference level) or lower interferences (for a given limit-of-
sensitivity). Investigation of such changes was outside the time and funding
scope of this contract but it would appear useful to summarize pertinent ideas
here. Additionally, it may be possible to improve sensitivity to the point
that an H/NOX instrument would be useful for ambient air measurements. For
this latter application trade-offs should be made that would favor sensitivity
at the expense of interferences. This is possible since of the two major inter-
ferants encountered in the present work, one, Oa, is present at a constant con-
centration in ambient air and the other, hydrocarbons, at a much lower concen-
tration than encountered in mobile source monitoring ( ^r 10 ppm as compared
to 10* ppm). Projections can thus be divided into two categories: (i) general
improvements and (ii) ambient monitoring.
General Improvements
1. The optical filter could be replaced with a wider bandpass filter
since homogeneous gas-phase emissions at 762 nm appear unlikely to
add much to positive 02 interference, which problem has been solved
in the prototype instrument.
2. A PMT with higher sensitivity than the P-4283 used in the 650 to 800
nm wavelength region would lead to increased sensitivity.
3. A larger diameter reaction tube would result in more signal. Optimiz-
ing the present reactor should provide a more intense signal as well,
e.g., the reaction zone to PMT cathode distance could be decreased.
4. An investigation of the nature of the background radiation could lead
to means for its reduction, which would result in increased sensitiv-
ity. Use of a supersonic jet reaction zone might eliminate the back-
ground altogether.
Ambient Monitoring
5. Sample flow rates comparable to the H2/(Ar, He) could be used, prob-
ably resulting in an order of magnitude improvement in limit-of-
sensitivity. While this would result in an increased H consumption
by Oa, such consumption would be constant and therefore not of
overriding importance.
31
-------
6. Since the variation in [H] decrease, i.e., negative interference due
to reaction with 02 and hydrocarbons, would essentially be eliminated,
the observation zone could be extended to include all or most of the
reaction zone. Thus use of an integrating sphere or spiral reactor
might lead to an order of magnitude improvement.
7. Increasing [H] in the observed reaction zone would result in a pro-
portional increase in light intensity, cf. Eq. (5). Moreover, such
an increase in [H] would allow higher sample flows (since negative
interferences are reduced) leading to further signal increases. The
positive ethylene interference resulting from a high [H] could be re-
duced by mixing the H/H2 and sample flows well upstream of the PMT
viewing area (cf., Section 6.E). A number of means for increasing
[H] appear available: (i) operating at a higher filament voltage
than used in the present instrument, (ii) further improvements in
the thermal source design, and (iii) reconsidering the use of an
electrical discharge method for H production.
8. Combination of items 5-7 with the suggested general improvements,
items 1-4, may result in a limit-of-sensitivity in the 1-10 ppb range
required for ambient monitoring. It is clear that the interaction of
the possible changes would have to be investigated in such a context.
32
-------
REFERENCES
1. Fontijn, A., Sabadell, A.J., and Ronco, R.J., "Homogeneous Chemilumin-
escent Measurement of Nitric Oxide with Ozone. Implications for Continu-
ous Selective Monitoring of Gaseous Air Pollutants," Anal. Chem. 42, 575
(1970). ~
>
2. Fontijn, A., "Chemiluminescence Techniques in Air Pollutant Monitoring,"
Modern Fluorescence Spectroscopy, Vol. 1, E.L. Wehry, Ed. (Plenum Press,
New York, 1976) p. 159.
3. Sigsby, J.E., Black, P.M., Bellar, T.A., and Klosterman, D.L., VChemi-
luminescent Method for Analysis of Nitrogen Compounds in Mobile Source
Emissions (NO, N02 and NH3)," Env. Sci. Techn. ]_, 51 (1973).
4. Pearse, R.W.B. and Gaydon, A.B.j The Identification of Molecular Spectra
(Chapman and Hall, London, 1963) Third Ed. p. 173.
5. Clyne, M.A.A. and Thrush, B.A., "Mechanism of Chemiluminescent Reactions
Involving Nitric Oxide - the H + NO Reaction," Disc. Faraday Soc. 33.
139 (1962).
6. Ishiwata, T., Akimoto, H., and Tanaka, I., "Chemiluminescent Spectra of
HNO and DNO in the Reaction of 0(3P)/02 with NO and Hydrocarbons or
Aldehyde," Chem. Phys. Lett. 2.L, 322 (1973).
7. Phillips, L.F. and Schiff, H.I., "Mass Spectrometric Studies of Atom Reac-
tions. III. Reactions of Hydrogen Atoms with Nitrogen Dioxide and with
Ozone," J. Chem. Phys. 37, 1233 (1962).
8. Jones, W.E., MacKnight, S.D., and Teng, C., "The Kinetics of Atomic Hydro-
gen Reactions in the Gas Phase," Chem. Rev. 73, 407 (1973).
9. Snyder, A.D. and Wooten, G.W., "Feasibility Study for the Development of
a Multifunctional Emission Detector for NO, CO and S02 ," Monsanto
Research Corp., Contract CPA 22-69-8, Final Report, October 1969.
10. Black, F.M. and Sigsby, J.E., "Chemiluminescent Method for NO and
NOX(NO + N02) Analysis," Env. Sci. Techn. J3, 149 (1974).
H. Herron, J.T. and Huie, R.E., "Rate Constants for the Reactions of Atomic
Oxygen (03P) with Organic Compounds in the Gas Phase," J, Phys. Chem.
Ref. Data 2, 467 (1973).
33
-------
12. Thrush, B.A., "Reactions of Hydrogen Atoms in the Gas Phase," Progress
in Reaction Kinetics .3» 63 (1965).
13. Fontijn, A., Hogan, J.M., and Miller, W.J., "Chemi-Ionization," Project
SQUID Semi-Annual Progress Report, October 1, 1964, p. 74.
14. Fontijn, A. and Baughman, G.L., "Chemiionization as a General Phenomenon,"
Bull. Amer. Phys.-Soc. 81, 363 (1963).
15. Clyne, M.A.A. and Thrush, B.A., "Reaction of Hydrogen Atoms with Nitric
Oxide," Trans. Faraday Soc. 5J7_, 1305 (1961).
16. Baulch, D.L., Drysdale, D.D., and Home, D.G., Evaluated Kinetic Data for
High Temperature Reactions, Vol. 2 (C.R.C. Press, Cleveland, 1973) p. 389.
17. Wong, W. and Davis, D.D., "A Flash Photolysis-Resonance Fluorescence Study
of the Reaction of Atomic Hydrogen with Molecular Oxygen H + 02 + M -*•
H02 + M," Int. J. Chem. Kin. £, 401 (1974).
18. Fontijn, A., Volltrauer, H.N., and Ellison, R., "Chemiluminescent Reactive
Hydrocarbon Analyzer for Mobile Sources," Final Report, AeroChem TP-319a,
EPA-650/2-75-069, NTIS PB 245 126, June 1975.
»
19. JANAF Thermochemical Tables, Dow Chemical Co., Midland, MI, continuously
updated.
20. Trainor, D.W., Ham, D.O., and Kaufman, F., "Gas Phase Recombination of
Hydrogen and Deuterium Atoms," J. Chem. Phys. 58, 4599 (1973).
21. Giachardi, D.J., Harris, G.W., and Wayne, R.P., "Excited State Formation
in the H + 02 System," Chem. Phys. Lett. 32, 586 (1975).
22. Hislop, J.R. and Wayne, R.P., Production of 02(1Ig+) in the H + 02 Sys-
tem," J.C.S. Faraday II 73, 506 (1977).
23. Black, P.M., High, L.E., and Fontijn, A., "Chemiluminescence Measurements
of Reactivity Weighted Ethylene-Equivalent Hydrocarbons," Env. Sci. Techn.
11, 597 (1977).
24. Fontijn, A. and Ellison, R., "Homogeneous Gas-Phase Chemiluminescence
Measurement of Reactive Hydrocarbon Air Pollutants by Reaction with Oxy-
gen Atoms," Env. Sci. Techn. 9, 1157 (1975).
34
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1 REPORT NO. 2.
EPA-600/2-79-120
4. TITLE AND SUBTITLE
MOBILE SOURCE NO MONITOR
Hydrogen-Atom Direct Chemiluminescence Method
J7. AUTHOR IS)
Arthur Fontijn, Hermann N. Volltrauer, and William
R. Prenchu
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Aero Chem Research Laboratories, Inc.
P.O. Box 12
Princeton, NJ 08540
fl2. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory-RTF, NC
Office of Research and Development
I Environmental Protection Agency
Research Trianqle Park, NC 27711
3. RECIPIENT'S ACCESSIOWNO.
5. REPORT DATE
July 1979
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
1AD71? RA-44 (pV-78)
11. CONTRACT/GRANT NO.
68-02-2744
13. TYPE OF REPORT AND PERIOD COVERED
Final 9/77-3/79
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
(NO and NO ) emissions
i hydrogen atoms. This
16. ABSTRACT
An analyzer was developed for measuring motor vehicle NO
based on the Chemiluminescence reaction of NO and NO wi:
eliminated the need for an N0_ to NO converter as required with ozone Chem-
iluminescence for NO analysis. The hydrogen-atom source is based on the thermal
dissociation of molecular hydrogen on a hot (~1900 C) tungsten filament. The
unit has linear response to NO over a concentration range from 4 ppm to greater
than 3000 ppm. No interferences were observed with HO, CO, CO , toluene, isopen-
tane, NH , HCN, CH NH , or H at concentrations encountered in raw automobile ex-
haust. Oxygen and etnylene caused minor interferences which should be inconsequent
ial in actual application.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
COS AT I Field/Group
* Air pollution
* Nitrogen oxides
Motor vehicles
Emission
* Monitors
* Chemiluminescence
Hydrogen
13B
07B
13F
07D
20F
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
••—————
EPA Form 2220-1 (9-73)
19. SECURITY CLASS (This Report)
UNCLASSIFIED
20. SECURITY CLASS (This page)
UNCLASS
22. PRICE
------- |