Kinetics of the O(3P)+N2O reaction. 1. Direct measurements at intermediate temperatures

Article abstract of DOI:10.1021/jp994470v

Rate coefficients for the consumption of O atoms by their reaction with N₂O have been measured, at pressures from 130 to 500 mbar, using the high-temperature photochemistry technique. These represent the first direct measurements of k values of the reaction. The ground-state oxygen atoms were produced by laser photolysis of SO₂, or by flash photolysis of either SO₂ or O₂, and monitored by time-resolved resonance fluorescence. The results yield k(1075-1140 K) = 3.2×10^-11 exp(-9686K/T) cm³ molecule^-1 s^-1 with 2σ precision limits of ±12% and corresponding 2σ accuracy limits of ±26%. Results from several sources in the literature indicate a high sensitivity of the O+N₂O reaction system to traces of H₂O, which increases the rates if present as a contaminant. For this reason, possible effects of traces of H₂O on the results were modeled. Simulated decay curves with a hypothetical H₂O contaminant were used as a test of the experimental data reduction procedures. Although the concentration of H₂O needed to significantly affect the results is small, the amount that could have been present is even less and is shown to have had negligible effects. The results are in qualitative agreement with a recent T≥1680 K shock tube study in that extrapolation of their results to the present temperatures indicates rate coefficients much larger than had been previously thought. However, though the results agree within error limits for such a long extrapolation, the present results are about a factor of 4 smaller. Combined with the results of the companion paper by Meagher and Anderson (ref 6, following paper in this issue), in which the prior literature is critically reevaluated, it is found that the O₂+N₂ product channel dominates at the present temperatures.

Full text of DOI:10.1021/jp994470v

J. Phys. Chem. A 2000, 104, 6003-6012
6003
3
Kinetics of the O( P) + N2O Reaction. 1. Direct Measurements at Intermediate
Temperatures
†
Arthur Fontijn,* Abdellatif Goumri, and Abel Fernandez
High-Temperature Reaction Kinetics Laboratory, The Isermann Department of Chemical Engineering,
Rensselaer Polytechnic Institute, Troy, New York 12180-3590
William R. Anderson* and Nancy E. Meagher‡
U.S. Army Research Laboratory, Aberdeen ProVing Grounds, Maryland 21005-5066
ReceiVed: December 23, 1999
2
Rate coefficients for the consumption of O atoms by their reaction with N O have been measured, at pressures
from 130 to 500 mbar, using the high-temperature photochemistry technique. These represent the first direct
measurements of k values of the reaction. The ground-state oxygen atoms were produced by laser photolysis
of SO
The results yield k(1075-1140 K) ) 3.2 × 10
limits of (12% and corresponding 2σ accuracy limits of (26%. Results from several sources in the literature
indicate a high sensitivity of the O + N O reaction system to traces of H O, which increases the rates if
present as a contaminant. For this reason, possible effects of traces of H O on the results were modeled.
Simulated decay curves with a hypothetical H O contaminant were used as a test of the experimental data
reduction procedures. Although the concentration of H O needed to significantly affect the results is small,
2 2 2
, or by flash photolysis of either SO or O , and monitored by time-resolved resonance fluorescence.
-
11
3
-1 -1
exp(-9686K/T) cm molecule
s
with 2σ precision
2
2
2
2
2
the amount that could have been present is even less and is shown to have had negligible effects. The results
are in qualitative agreement with a recent T g 1680 K shock tube study (D. F. Davidson, M. D. DiRosa, A.
Y. Chang, and R. K. Hanson, ref 9) in that extrapolation of their results to the present temperatures indicates
rate coefficients much larger than had been previously thought. However, though the results agree within
error limits for such a long extrapolation, the present results are about a factor of 4 smaller. Combined with
the results of the companion paper by Meagher and Anderson (ref 6, following paper in this issue), in which
the prior literature is critically reevaluated, it is found that the O
present temperatures.
2
+ N
2
product channel dominates at the
I. Introduction
O + N O f NO + NO
∆H ) -36 kcal/mol (R1)
2
0
The role of nitrous oxide (N2O) in combustion and its thermal
decomposition have been studied extensively. Motivation for
the present investigation stems from the fact that N2O is a major
intermediate oxidizer species formed during the combustion of
many types of solid propellants and their energetic raw
O + N O f O + N
∆H ) -79 kcal/mol (R2)
2
2
2
0
5
An early critical review (BDH73) of the extensive literature
published through 1973 on O + N O and other reactions
concluded that the two channels have identical rate coefficients:
2
1
materials. Therefore, it is important for the development of
-
10
3
solid propellant models to understand the role of this oxidizer
in the combustion process. N2O also plays a major role in the
k (T) ) k (T) ) 1.7 × 10 exp(-14100K/T) cm
1
2
molecule-1 s-1 (1200-2000 K) (1)
2
formation of NOx pollutants during combustion. In addition,
dissociation of nitrous oxide offers the possibility of initiating
Modeling of some of the experiments utilized to determine
the k1 recommendation of BDH73 has recently been performed.
3
combustion reactions by the atomic oxygen produced, and has
4
been used as a test case for unimolecular reaction theories.
This work is discussed in detail in the following paper in this
6
For the preceding reasons, it is important to understand the
reactions of N2O with the three primary combustion radicals,
H, O, and OH. The reaction of N2O with O atoms, which plays
major roles in both the combustion and the thermal decomposi-
tion of N2O, has two significant product channels:
issue. The study has revealed that assumptions about the
ancillary chemistry utilized in complex modeling to extract k1,
or its reverse, from the data in some of the most important of
the works cited in BDH73 are incorrect, invalidating both those
results and the BDH73 recommendation. It is also shown in ref
6
that at temperatures below about 1700 K most of the currently
†
available studies for k1 and all for k2 are invalid. The k1
expression from BDH73 strongly affects most of the later
recommendations. Subsequent reviews have accepted the sound-
ness of the choices made in BDH73 regarding the best values
from early works. The expression given in eq 1 was used as
Current address: Department of Chemistry, University of North Texas,
Denton, TX 76203.
‡
Contribution performed in part as an ASEE Postdoctoral Research
Associate at the U.S. Army Research Laboratory. Current address: Depart-
ment of Chemistry and Physics, Texas Woman’s University, Denton, TX
6204.
7
1
0.1021/jp994470v CCC: $19.00 © 2000 American Chemical Society
Published on Web 06/02/2000

6004 J. Phys. Chem. A, Vol. 104, No. 25, 2000
Fontijn et al.
part of the fitted data to obtain a k1 recommendation in ref 7
HS85). Recommendations from HS85 for the two channels are
significantly affect the results. Although the necessary amounts
are very small, the effects of H2O are proven to be insignificant.
Results have been obtained for the total reaction rate coefficients,
ktot ) k1 + k2, in the temperature range 1076-1276 K.
(
k1(T) ) 1.2 × 10 exp(-13400K/T) cm molecule s-1
-
10
3
-1
6
In the companion paper, an evaluation of all literature
(
1200-4100 K) (2)
relevant to the title reaction and new recommendations for k1
and k2 are presented, utilizing all retained data to provide fits.
The present experimental results are that ktot is indeed much
larger at intermediate temperatures than had been previously
believed, but not as large as extrapolation of DDCH92 would
suggest.
k2(T) ) 1.7 × 10 exp(-14100K/T) cm molecule s-1
-
10
3
-1
(
1200-3200 K) (3)
These expressions only differ modestly from those of BDH73.
The recommendations of HS85 were subsequently accepted in
8
a review pertinent to propellant chemistry modeling. However,
9
a recent high-temperature shock tube determination (DDCH92),
II. Experimental Procedures
which involved direct UV absorption measurements of the O2
and NO products, yielded a very different result:
The older style HTP reactor13 was used. Briefly, it consists
of an alumina reaction tube (5.1 cm i.d.) which is surrounded
by helical SiC heating elements and insulation and enclosed in
a water-cooled, steel vacuum chamber. To achieve the desired
temperatures without decomposition of the reactant gases, a
movable, air-cooled inlet is used. After emerging, the gases are
mixed with the heated Ar bath gas. The residence time needed
for the mixture to reach the reaction zone is adjusted so that
the reactant gases are at least 95% mixed with Ar.14 The
temperature of the reaction zone was measured before and after
each experiment by a Pt-Pt/13% Rh thermocouple, which was
doubly shielded to minimize radiation effects. An FPI pressure
transducer was used. The gas flow rates were controlled by
Teledyne-Hasting flow controllers.
k1(T) ) 4.8 × 10 exp(-11650K/T) cm molecule s-1
-
11
3
-1
(1680-2430 K) (4)
k2(T) ) 2.3 × 10 exp(-5440K/T) cm molecule s-1
-
12
3
-1
(1940-3340 K) (5)
k1 from DDCH92 agrees very well with that from HS85. But,
although the magnitude of k2 is similar to that of previous
expressions at ∼2000 K, the slope of the Arrhenius plot is very
different. The activation energy, Ea, corresponding to k2 is only
1
1 kcal/mol as opposed to ∼28 kcal/mol from prior sources.
Thus, extrapolation of the k2 expression to lower temperatures,
where few data are available, would lead to much larger total
rate coefficients than yielded by the earlier expressions. The
two sets of expressions from HS85 and DDCH92 were recently
tested by comparison of flow system data on CO/N2O/H2O/N2
mixtures at 950-1123 K vs modeling results using a complex
mechanism.10 That work concludes that the data favor adoption
of expressions from HS85 rather than DDCH92. However, rate
coefficients of magnitude between those of HS85 and DDCH92
were not tested. The expressions from DDCH92 are recom-
mended in a recent critical review1 with an admonition that
there is still “appreciable uncertainty” regarding k2.
3
Ground-state O( P) atoms were generated by laser photolysis
of SO2 at 193 nm or by flash photolysis of either SO2 (through
a Suprasil Quartz window) or O2 (through a MgF2 window).
3
The decrease in the relative concentration of O( P) atoms was
monitored by fluorescence of the 130.2-130.6 nm resonance
triplet. The fluorescence was induced by a cw microwave
discharge lamp, through which flowed 99.999% He which
contains O2 as an impurity. A CaF2 lens was used to focus the
O atom radiation and filter Lyman R-radiation. The O atom
fluorescence was detected by a solar blind photomultiplier tube
1
(PMT) and fed to a multichannel scaler signal averager.
The experiments were carried out under pseudo-first-order
The O + N2O reaction is important at all temperatures for
propellant chemistry. Temperatures between 1200 and 1600 K
are especially pertinent to modeling the dark zone structure of
solid propellants (nonluminous gaseous region between the
conditions, where [O] , [N2O] , [Ar]. Under these conditions,
the fluorescence intensity I, which is proportional to [O], is
expressed by
1
a
condensed-phase propellant and luminous flame zone ). The
uncertainty regarding the extrapolation to lower temperatures
and the lack of any prior measurements under isolated conditions
indicated a need for a direct determination of the reaction rate
coefficients at temperatures below 1700 K. The analysis errors
in prior works indicated a need for a thorough reassessment of
the literature regarding this reaction. Note in this and the
I ) I exp(-kps1t) + B
(6)
o
where kps1 ) ktot[N2O] is the pseudo-first-order rate coefficient,
Io + B the intensity at time t ) 0, and B the background, due
mainly to scattered light. The values of kps1 were obtained by a
weighted fit of the observed I vs t profiles to eq 6.15 Typically,
five or six kps1 measurements at varying [N2O], with the
minimum [N2O] a factor of about 5-10 lower than the
maximum values listed in Table 1, were used to obtain ktot at
the temperature and pressure of the experiment.
6
companion paper, “intermediate temperatures” refer to the
approximate range 1000-1700 K and “high temperatures” to
higher values.
The first measurements of the title reaction under conditions
in which it is clearly isolated from the effects of other reactions
The exponentiality of the I vs t plots was tested by a two-
stage residual plot analysis consisting of, first, a visual inspection
of the plotted residuals (experimental minus fitted values) and,
(
that is, complex modeling with heavy reliance on ancillary
kinetic data is not required to extract the rate coefficients) are
reported in this paper. Measurements of the total rate of O atom
disappearance in an excess of known concentrations of N2O
were performed at the Rensselaer Polytechnic Institute (RPI)
using the HTP (high-temperature photochemistry) technique.
Modeling of the possible effects of H2O as a contaminant, whose
16
second, a runs test. The data must pass both tests. An example
of an experimental [O] decay, its fit to eq 6, and the associated
residuals plot are given in Figure 1. The residuals consist of
both positive and negative values. In Figure 1, these occur in a
band centered about zero; thus, that decay passes the visual
inspection test. For a given decay, the runs test gives a measure
of how “random” the residuals are by comparing the actual
number of runs (the number of groups of adjacent residuals
1
2
effects have been suggested in another recent N2O study, was
done at the U.S. Army Research Laboratory. The modeling
provides an estimate of the concentrations of H2O necessary to

3
Kinetics of the O( P) + N2O Reaction
J. Phys. Chem. A, Vol. 104, No. 25, 2000 6005
2
O Reaction
TABLE 1: Summary of Rate Coefficient Measurements on the O + N
Ta
P
[M]
[photolyte]
F
(mJ)
[N
O]max
z
Vj
2
i
k ( σki
(1018 cm-3
)
(10 cm
14
-3
)
(10 cm-3
15
)
(cm)
(cm s-1)
(10-14 cm molecule s-1
)
3
-1
(K)
(mbar)
1076
1076
1097
1099
1101
1121
1132
1141
1142
1142
1178
1178
1179
1179
1202
1219
1230
1230
1251
1272
1273
1276
413
492
375
283
197
228
320
355
133
324
496
549
495
556
148
328
475
526
176
400
345
403
2.78
3.31
2.47
1.87
1.30
1.47
2.04
2.24
0.84
2.05
3.05
3.45
3.10
3.40
0.89
1.95
2.80
3.14
1.00
2.30
1.97
2.28
1.72
4.84
2.35
1.60
40
27
33
40
4.00
3.00
4.30
3.70
3.42
2.90
3.12
4.50
3.10
4.32
4.40
4.50
3.40
4.00
2.11
2.50
5.80
7.10
2.40
4.07
4.72
4.48
17
10
10
10
8
8
10
8
10
10
10
16
5
10
10
17
10
10
8
12
7
9
14
14
11
13
6
16
8
7
7
9
8
0.36 ( 0.02
0.45 ( 0.01
0.45 ( 0.04
0.40 ( 0.01
0.48 ( 0.06
0.55 ( 0.03
0.71 ( 0.06
0.62 ( 0.14
0.75 ( 0.06
0.53 ( 0.07
1.12 ( 0.12
1.01 ( 0.08
1.55 ( 0.14
1.57 ( 0.11
1.40 ( 0.15
1.29 ( 0.07
2.02 ( 0.20
1.38 ( 0.06
2.61 ( 0.14
3.06 ( 0.13
2.96 ( 0.12
3.02 ( 0.18
c
d
1.59
9.60
13
22
b
d
5.01
c
d
7.94
c
2.50
c
3.24
4.15
4.70
2.58
4.85
40
27
13
53
c
d
3.66
50
16
9
8
c
d
9.80
22
5.50
6.12
8.00
47
13
9
7
b
d
14
10
10
10
2.80
4.1
2.52
20
40
33
10
10
6
a
b
window. c SO
σ
T
/T ) (2%.
O
2
was photolyzed by the flash lamp through a MgF
2
2
was photolyzed by the flash lamp through a Suprasil
d
window. For these flash-lamp experiments these energies are in joules; all of the other experiments used laser photolysis of SO
2
, with energies
expressed in millijoules.
deviation of u from the mean is given by the unit normal deviate,
1
6,17
Z.
Z gives a quantitative measure to decide whether the
residuals are randomly distributed about the fitted curve. We
have found that a stringent and adequate criterion is |Z| e 1.5;
this corresponds to a rejection region in the 6.68% tails of the
normal distribution.
The possible effects of contaminants, especially H2O, on the
N2O system reactions are of considerable concern. Normal
precautions in performing the experiments are described below
to emphasize the validity of the data presented herein. All supply
lines leading to the reactor are kept at pressures above
atmospheric up to the flow controllers which are in line just
prior to the reactor. All parts of the apparatus, especially the
flow lines at subatmospheric pressure, were regularly checked
with a helium leak detector. The reactor, as a matter of course,
is continuously baked out at high temperature; thus, it cannot
be a source of significant H2O. Therefore, there are two major
contaminant sources to consider, the supply gases and air leaks.
Only high-purity gases were used (see below). It is only
important to consider the Ar carrier gas since the others are
present in much smaller amounts and were passed through
drying towers. The carrier could have up to 20 ppm impurity.
However, the carrier is stored as a liquid and vaporized as
needed during the experiments. Thus, any impurities in the
carrier must have a very low boiling point to pass into the test
mixture. H2O from this source is clearly excluded. Air leaks
would lead to addition of, mainly, N2, O2, Ar, and H2O to the
test mixture. The first three would have no effect on the present
experiments. Since the modeling indicates a very high sensitivity
to H2O as a contaminant, a measurement of the air leak rate
was performed at operating temperatures. From this, it was
inferred that the maximum concentration of H2O which could
possibly be present in the reactor is 0.2 ppm, assuming a
laboratory relative humidity of 100%.
Figure 1. Example experimental [O] decay and residuals plot at 1142
1
5
-3
K, total pressure 133 mbar, and [N
, kps1, and B values are 1088.7, 34.67 s , and 1300, respectively. Z
-0.014. Contribution to kps1 due to diffusion found by fitting the
2
O] ) 1.1 × 10 cm . The fitted
-1
I
o
)
-
1
(
2 d
[N O], kps1) data, k ) 26 s . Upper panel: measured decay (points)
and fit to eq 6 (curve). Lower panel: residuals plot; a line is drawn at
zero in the lower panel as an aid for visual inspection.
The gases used were N2O (99.99%) from Matheson, 5% O2
(99.99%) in Ar (99.998%) from Scott, 1% SO2 (99.98%) in Ar
(99.998%) from Matheson, and Ar (99.998%) from the liquid
(Praxair).
having the same sign), u, to the mean number of such runs
expected for the number of data points used in the fit. Assuming
that the number of runs follows a normal distribution, the

6006 J. Phys. Chem. A, Vol. 104, No. 25, 2000
Fontijn et al.
TABLE 2: Subset of Important Reactions Extracted from the Detailed Mechanism Used for Modeling the Effects of
Hypothetical H
2
O Contaminanta
reaction
O f NO + NO
no.
A
n
E
a
/R
ref
21
R1
R2
R3
O + N
O + N
2
see the text
see the text
2
O f O + N
+ O (+M)
2
2
2
N O (+M) f N
2
k
k
0
9.91E-10
1.26E+12
0.00
0.00
28 510
31 510
∞
efficiencies, η
i
: /Ar 0.67/N O 5.0/ from ref 5
2
O 7.5/O 0.82/ from ref 22
2
/
H
2
R4
H + N
2
2
O f N
2
+ OH
3.70E-10
.20E-14
5.85E-08
9.96E-16
4.10E+04
7.60E+18
0.00
0.00
-0.70
1.30
-3.37
-1.27
8 430b
2 290b
8 590
0
37 640
36 880
14
4
R5
R6
R7
H + O
f O + OH
23
2
8
2
OH + OH f O + H O
NO
2
(+M) f NO + O (+M)
k
k
0
∞
efficiencies, η
i
: /Ar 0.71/N
2 2
O 1.5/H O 4.4/ from ref 5, reverse reaction
R8
NO
2
+ O f NO + O
2
6.48E-12 0.00
-120
8
a
3
n
Units are cm , molecule, s, K. The parameters A, n, and E
a
/R are for rate coefficient expressions in the form k ) AT exp(-E
a
/RT). For the
are the mole
N
falloff reactions, R3 and R7, the overall collider efficiencies are given by C
fraction and the collider efficiency, respectively, of species i. The expressions are appropriate for an N
low pressure A factors having been made where necessary. Rate coefficients of the reactions are then given using k ) Fk
M
) [P/RT]∑
X
i
η
i
, where P is pressure and X
i
and η
i
i)1
2
collider efficiency of 1.0, adjustments to
, where k ) k /(k
)] }. For reaction R3, the simple Lindemann form, F ) 1.0, is used. For reaction R7, F
For reaction R4, the rate coefficient is computed as the sum of the two exponential expressions.
∞
21
k
L
L
0
C
M
∞
2
+
0
k
0
C
M
) and log F ) log F
C
/{1 + [log (k
0
C
M
/k
∞
C
)
-
4
8
b
.95-1.0 × 10 T is used.
III. Consideration of H2O as Hypothetical Contaminant
Prior to modeling the experiments with hypothetical traces
of H2O contaminant present, a few cases representative of the
typical conditions (see Table 1) were modeled without H2O
added to the initial mixture. Most of the modeling was
performed with a “standard” mixture of 0.001 77 mole fraction
N2O and 0.000 004 4 mole fraction O2, the remainder being Ar,
and pressure 405 mbar, which is representative of experimental
conditions. It was found that the modeling of [O] must be done
utilizing a two-step calculation, representing thermal reaction
followed by photolysis. This holds true whether H2O is included
in the initial mixtures or not. The calculation is illustrated in
Figure 2. The k1 and k2 expressions used are the final
As previously mentioned, the authors of ref 12 have shown
that the thermal decomposition of N2O can be substantially
accelerated by traces of H2O, even at temperatures above 1700
K. Further discussions about its pronounced effect at lower
temperatures are presented in ref 18, which contains calculations
for a presumed 30 ppm H2O as contaminant, and in ref 19, a
combined experimental flow reactor/modeling study in which
H2O was deliberately added at a known concentration of 560
ppm. The prior observations of strong effects of H2O, and the
fact that the present results for ktot are much larger than had
previously been believed,5 raised concerns about the sensitiv-
ity of the present experimental results to the presence of trace
amounts of H2O. For this reason, the experiments were modeled
using a state-of-the-art detailed chemical mechanism for model-
ing the dark zones of solid propellants.20 The mechanism was
developed for mixtures of H2/H2O/CO/CO2/NO/N2/N2O/HCN
at temperatures between ∼1000 and ∼3000 K. It is based on
,7,8
6
recommendations made in the following paper in this issue; at
1
076 K, these yield ktot equal to that of the present measure-
ments. The first step of the calculation represents the thermal
decomposition of a minute fraction of the N2O in the hot test
mixture as it travels convectively through the reactor tube
between the injection point and the photolysis/probe volume.
This results in formation of a constant, trace, steady-state [O],
hereafter referred to as [O]SS.
2
the mechanism of Miller and Bowman for modeling of nitrogen
chemistry in combustion. Modifications include updated rate
coefficient values, thermodynamics, and additional reactions.
Reverses of all reactions are automatically included via calcula-
tion of equilibrium constants from thermodynamic data. Ar was
readily added. For the present purposes, removal of C atom
species has allowed reduction to ∼110 reactions. The full
mechanism is too long to present. However, the subset of
reactions involving only nitrogen- and oxygen-containing species
-7
The observation of a steady-state [O], 5 × 10 mole fraction
in Figure 2, is caused by the formation of O atoms in reaction
R3 and their destruction in reactions R1 and R2 (see Table 2).
Assuming only reactions R1, R2, and R3 have appreciable rates
and [N O] is nearly constant, the resulting differential equation
2
for [O] can be solved analytically. The general solution is
[O] ) k3,uni/k + C exp(-kps1t)
(7)
tot
6
is presented in the companion paper. Furthermore, sensitivity
and rate analysis of the results show that, for conditions of the
present experiments, only a few of the reactions have much
significance. These reactions are shown in Table 2. Additional
comments regarding our recent development of the subset of
reactions used for modeling of H2/N2O and H2/N2O/NH3 flames
may be found in ref 24. The SENKIN time-dependent,
homogeneous chemical reactor code25 was utilized for the
calculations. Temperature and pressure were assumed constant.
Implicit assumptions made in the application of this model are
that mixing at the inlet is instantaneous and plug flow is
appropriate for the reactor; that is, diffusion is ignored. Possible
effects of these assumptions are discussed later.
where k3,uni is the apparent unimolecular rate coefficient of
reaction R3 taking falloff into account and C is a constant of
the integration to be determined using the proper [O] , the initial
o
[O]. k3,uni/ktot is recognized as the [O] . The criterion for the
SS
[O]SS to be observed is k [N O] . k3,uni. This inequality is
tot
2
fulfilled over the range of conditions utilized in the experiments.
Using eq 7 and the proper initial conditions, it is readily shown
that the asymptotic approach of [O] to [O]SS after the injection
of N2O into the reactor exhibits a [1 - exp(-kps1t)] dependence,
while the [O] decays after photolysis have an exponential form
similar to eq 6. The exponential forms for each have the same
pseudo-first-order rate coefficient, kps1, as may be observed in

3
Kinetics of the O( P) + N2O Reaction
J. Phys. Chem. A, Vol. 104, No. 25, 2000 6007
N2O decomposition at the higher temperatures. Because this
decomposition leads to NO and NO2 formation, the -R7 and
R8 reaction sequence may be important at the highest temper-
atures utilized (-R7 indicates the reverse of reaction R7). The
consequences will be discussed later.
When hypothetical traces of H2O were added to the modeling
mixtures, both [O] and [OH] during the thermal decomposition
step increase to nearly constant plateau levels after a small
fraction of the residence time. Between the inlet tube and the
photolysis zone, the [OH] builds to a trace value due to N2O +
M f N2 + O + M (R3) and O + H2O f OH + OH (-R6).
After the photolysis pulse, in addition to reactions R1-R3, the
reactions H + N2O f OH + N2 (R4) and O + OH f O2 + H
(
-R5) are predicted to have the largest effects, accelerating the
O atom decays. Note that reaction R4 combined with reaction
R5 represents a cycle catalyzed by H and OH. Therefore, even
-
though only small traces of H and OH could be present, these
species would not be consumed during the [O] decrease. This
fact explains why the effects of even a few parts per million
H2O as a contaminant could significantly alter the resulting
apparent rate coefficient.
For the present experimental conditions, modeling indicates
H2O would increase both the rates of [O] growth to constant
[O]SS levels during the thermal decomposition step and the decay
rates to these levels in the photolysis step by equal amounts.
Plots of the predicted [O] decreases with [H2O] in the mixtures
appeared to be exponential upon visual examination. However,
fits of these decays to eq 6 revealed such mixtures would
actually produce small deviations from exponential behavior.
For example, the differences between predicted [O] decreases
and exponential fits were about 5% relative to [O]o - [O]SS for
5-20 ppm H2O.
Parametric studies of the fitted decays have revealed two
additional features of the chemical system. These features would
complicate detection of the effects of H2O. First, the predicted
order of the rate in [N2O] is fairly close to unity, for the range
of [N2O] used in the experiments, if all other variables are held
constant. Second, the order in [H2O] is much less than unity
Figure 2. O atom profile for an example modeling run. The example
used the standard mixture (no H O present) and pressure described in
2
the text, and T ) 1076 K. The ktot value used is taken from ref 6. The
portion from 0 to 1.0 s, a typical residence time, indicates exponential
[O] increase to a constant, steady-state level due to thermal reaction of
a small fraction of the N O during that period. Photolysis was assumed
2
to occur at 1.0 s. The exponential decay after photolysis represents the
experimentally measured quantity.
Figure 2. These curious characteristics result from decomposition
producing the reactant radical and would occur in other, similar
cases. Because the [O]SS is constant, its presence does not
adversely affect the extraction of kps1 via fits to eq 6. The value
of the [O]SS formed during the thermal reaction phase, and to
which [O] decays after the photolysis, is independent of [N2O]
as long as an appreciable [N2O] is present. Following these
predictions, experiments have shown, by observing background
variation vs reactor temperature, that the [O]SS can contribute
substantially to the background constant, B, in eq 6.
The concentrations of all species except O atoms at the end
of the thermal reaction step are then used as initial conditions
for the second, photolysis step of the calculation. The calcula-
tions indicate that if the thermal reaction step is ignored and
the modeling is initiated with an assumed [O] formed by
photolysis smaller than that found at the steady state, then the
(0.33 at 1076 K; 0.4 at 1276 K). These characteristics make
ruling out possible systematic errors due to the presence of H2O
nontrivial because (1) small deviations from exponential
behavior are difficult to see in real data with noise and (2) the
[N2O] order being near unity means plots of predicted kps1 vs
[N2O] exhibit approximately linear behavior despite the assumed
[H2O]. Also, the standard procedure followed in the experiments
to reveal the presence of contaminants or impurities is to widely
vary reactant and carrier gas partial pressures. Were a contami-
nant present, this procedure would yield differing reactant-to-
contaminant ratios. It is expected that varying these ratios will
reveal the presence of appreciable contaminant effects because
extracted rate coefficients will exhibit systematic trends. How-
ever, since the predicted order in [H2O] is small, the effect of
varying H2O contaminant-to-reactant ratios would not be as large
as one expects for a contaminant of higher reaction order.
[O] is predicted to increase to the [O]SS after the photolysis
pulse. Growth in [O] after the pulse contradicts the experimental
observations, proving the necessity of the thermal decomposition
calculations.
An example of a predicted [O] decrease with significant
[H2O] present is shown in Figure 3, upper panel. In this example,
the standard mixture with 5 ppm H2O added was used with the
standard pressure and an assumed temperature of 1076 K. The
HS85 k1 and k2 expressions were used. Also shown in Figure
3, upper panel, is a least-squares fit of the exponential function
of eq 6 to the simulated decay curve, the time intervals in the
latter having been first interpolated to match the data sampling
rate of the experiment. As can be seen, although the predicted
decay curve looks nearly exponential, the exponential fit is not
For the photolysis step, the initial [O] is increased above the
[O]SS, and the calculation is reinitiated. The [O] then decays
back to the baseline [O]SS (see Figure 2). Fits of the simulated
decays to eq 6 are excellent. These fitted kps1 values almost
exactly reflect the assumed values of ktot and initial [N2O], as
expected, at the lowest temperatures used. However, possible
deviations were predicted due to contributions from significant

6008 J. Phys. Chem. A, Vol. 104, No. 25, 2000
Fontijn et al.
Figure 4. Predicted O atom decay curve with random noise added
Figure 3. Predicted O atom decay curve (upper panel) and residual
plot (lower panel) with H O as a hypothetical contaminant at 5 ppm.
In the upper panel, the solid line represents the [O] profile, while the
dotted line shows the least-squares fit to an exponential function, eq 6,
for the oxygen atom decay. An enlargement of the early-time [O] curve
(upper panel) and corresponding residuals plot (lower panel). Assumed
2
conditions are the same as for Figure 3, except that noise has been
added to simulate a S/Sc (signal-to-scatter ratio) of 5. The solid line in
the upper panel is the fit of the simulated data points to eq 6. The
noise-free residuals plot from Figure 3, lower panel, is repeated as the
solid curve in the lower panel here to demonstrate that the scatter
envelope is centered about the noise-free plot, as expected.
is presented in the inset in the upper panel, revealing the deviation
7
from exponential behavior. k
1
and k
2
from HS85 were used. The
assumed conditions were identical to those in Figure 2.
as one would expect. The unusual departures at short times from
the expected band centered about zero would have been detected
even by visual inspection during data reduction, if present in
the actual experiments.
perfect; the differences are undoubtedly Very much larger than
the numerical error of the SENKIN code. At short times, the
exponential fit is slightly higher than the predicted “experimen-
tal” decay curve, crosses it at 0.014, 0.075, and 0.373 s, and
The simulated case in Figure 4, upper panel, with S/Sc ) 5
and Z ) -1.28 shows that such a low S/Sc would not cause
the runs test to lead to data rejection. However, the unusual
shape of the residuals plot shown in the lower panel would be
the basis for rejection of the data using the visual inspection
method. A similar example was tested with S/Sc ) 10, which
resulted in Z ) -2.25 and an even more unusual residuals plot;
the result is rejection of the data by either method. The
nonexponentiality of the simulated [O] decreases in the presence
of 5 ppm H2O becomes more obvious as the S/Sc ratio increases.
In the experiments, attempts to increase S/Sc by averaging
results from more photolysis flashes would have been made in
such a case. However, such efforts could only have resulted in
an increased awareness of the nonexponentiality of the decays.
Many of the experimental decays have S/Sc g 10. All the
experimental [O] decays, such as in Figure 1 with S/Sc ≈ 6,
pass both tests of exponentiality, thus yielding strong eVidence
that [H2O] cannot be as high as 5 ppm in the actual
experiments.
finally approaches the same baseline asymptotically at long
times. A plot of the residuals16 of the fitting procedure is shown
in Figure 3, lower panel. If the predicted curve were exponential,
the residuals would be zero at all times. However, the residuals
plot has a strange shape, reflective of the slight difference from
exponential decay shown in the upper panel, with zeros at the
points where the upper panel curves cross.
The following question naturally arises: If the HS85 recom-
mended rate coefficients were correct, and if enough H2O had
been present in the current experiments to account for the
measured decays, would the experimental data reduction
procedures have revealed the deviation from exponentiality and
led to rejection of the data? To answer this question, simulated
decay curves were prepared. Enough constant-level white noise
was added, using a random number generator, to mimic the S/Sc
ratios observed in typical experiments. Here, S/Sc is defined as
the signal-to-scatter ratio, where S is equivalent to Io, and Sc is
the full width of scatter at the baseline. One such curve, with
exponential fit, is shown in Figure 4, upper panel. The curve
has S/Sc = 5; larger S/Sc values were achieved in most of the
actual experiments. The corresponding residuals plot is shown
in Figure 4, lower panel, which demonstrates that the band of
residuals from noisy “data” is centered about the noise-free plot
IV. Results
The determinations of ktot are summarized in Table 1. The
experimental parameters P, total pressure; [M], the total gas

3
Kinetics of the O( P) + N2O Reaction
J. Phys. Chem. A, Vol. 104, No. 25, 2000 6009
-
9
3
ktot(1075-1275 K) ) 2.7 × 10 exp(-14580K/T) cm
molecule- s-1 (8)
1
2
2
2
The variances and covariance are σA ) 1.02 A , σE ) 1.41 ×
6
3
1
0 , and σAE ) 1.2 × 10 A. The variances and covariance are
26
combined by the method of Wentworth, which yields a
precision at the 2σ level of (12% to (22%, depending upon
temperature. Allowing (20% for any unrecognized systematic
errors then leads to a 2σ accuracy level of (29%.
In modeling the present experiments, computations for the
highest temperatures used predict that during its travel down
the reactor tube a substantial portion of the N2O could
decompose, producing, among other species, N2, NO, and NO2.
Predictions indicate formation of NO and NO2 could substan-
tially increase the observed rate of O atom decay by catalysis
via the -R7 and R8 reaction sequence. Simulation indicates
that if this were to occur, the [O] decays would still be closely
exponential; pseudo-first-order kinetics are obtained because
[N2O], [NO], and [NO2] are nearly constant on the time scale
of the decays. However, the situation is complicated because
this effect would be compensated by a decrease in the decay
rate due to a substantial decomposition of the N2O. The situation
is too sensitive to key rate parameters for reliable quantitative
prediction and, indeed, even to be sure whether the measured
ktot values would be low or high. Calculations indicate the N2O
decomposition and NO/NO2 catalytic effect are altogether
negligible at the lower temperatures used. However, it cannot
be accurately assessed at what temperature these effects become
important. At high enough temperatures, reactant decomposition
generally results in a large amount of scatter in measured rate
coefficients; this is what was observed above 1280 K. Although
there does appear to be slightly higher scatter in the 1178-
1276 K portion of the data, the difference vs scatter at the lower
temperatures is not large. In addition, the predicted temperature
at which N2O decomposition becomes important might be ∼50-
1
00 K different from experimental observations for several
reasons: (1) the amount of decomposition predicted depends
strongly on assumed rate coefficients and their uncertainty limits,
especially k3, (2) during the first few centimeters of travel
through the tube past the injection point, the N2O must mix
with hot gases and reach thermal equilibrium before much
reaction commences, and (3) during travel through the tube,
reactant gases come into contact with the wall; for O-atoms,
this results in a constant, diffusion-related loss (wall recombina-
tion), reducing the background [O]SS discussed in the previous
section. There is no simple way to include these effects in the
modeling. Since the rate constant for loss due to diffusion is
essentially constant as [N2O] is varied, it does not affect the
reliability of the predictions. Therefore, diffusion was ignored
in the computations. However, the effect could reduce the
amount of N2O decomposition in the experiments, as compared
to the model, because the [O]SS and, hence, the rates of reactions
R1 and R2 are reduced.
Since the 1178-1276 K data have survived the exponentiality
tests and do not exhibit great scatter, they have been presented
in Table 1. Nevertheless, there is slightly larger scatter in that
temperature range as opposed to the lower temperature portion
of the data. Also, there appears to be a trend toward slightly
larger Z values in the higher temperature portion of the data set
(not shown), though the values meet standards normally used
(|Z| e 1.5). When all 22 of the present data points are included
in the fit to obtain recommendations for k1 and k2 expressions
(the present data set plus nine from the literature; see ref 6), a
skewing of the slope of ktot from the overall fit vs the trend of
concentration; z, the distance from the top of the cooled inlet
to the center of the reaction zone; Vj , the average linear gas
velocity; [photolyte], the concentration of the O atom precursor;
and F, the energy of the laser or the flash lamp, were varied
over wide ranges. To demonstrate that the rate coefficients do
not depend on any of these parameters, plots of [k(T) - ki]/
k(T) vs these parameters were made (not shown), where k(T)
represents the rate coefficients obtained from the fit expression
given below and ki are the individual ktot rate coefficients
measured. These residuals plots are independent of these
parameters. To avoid systematic errors due to the nature of the
O atom precursor or the photolysis source, two precursors (SO2
or O2) and laser photolysis or flash photolysis were used. No
dependence on these factors is found, either. Below 1076 K
the reaction became too slow for accurate measurements.
Attempts at measuring ktot at temperatures above 1276 K
exhibited increasing scatter and a dependence on residence
times, indicating significant decomposition of the N2O. These
measurements were disregarded.
Figure 5 shows the present data fitted to the form A exp(-
Ea/RT) where (σk and (σT/T ) 2% contribute to the weighting
i
of each point. The fitted expression is

6010 J. Phys. Chem. A, Vol. 104, No. 25, 2000
Fontijn et al.
the present full data set results, Figure 5. A linear fit of the 12
to temperatures of the present study. The HTP data fall between
the two earlier results. The present data agree with DDCH92
in that ktot is much larger at low temperatures than had been
thought previously (e.g., HS85). At the low-temperature end
of the current data, the results are a factor of 5 larger than the
HS85 recommendation, well outside tolerances (∼50%) implied
by a short extrapolation of the recommendation. They are a
factor of 4 smaller than an extrapolation of the DDCH92 results.
Considering the error limits in the k and k expressions from
highest temperature points has a suspiciously high A factor of
-
9
3
-1 -1
1
× 10 cm molecule s , as does the fit to all 22 points,
eq 8. Further, this 12-point fit has a slope about 1.4 times that
of the 10 lowest temperature points. This slope change seems
large for such a short temperature range. These observations
may be indicative of an underlying systematic effect. These
considerations, coupled with the modeling prediction that
decomposition likely increases ktot values toward the high-
temperature end of the data set, have led us to retain only the
lower temperature portion of the data in the fit for recommended
values. As can be seen in Figure 5, when only that portion is
used, the overall fit passes through those 10 points with excellent
agreement in slope. Fitting those 10 points at lowest tempera-
tures alone, the recommended result of the present work is
1
2
DDCH92, the present results are within tolerances of the
extrapolation of the DDCH92 results, although such a long
extrapolation is a suspect procedure.
The present data are further compared with recent data on
the title reaction in Figure 6. The solid curve is the recommended
6
k1(T) + k2(T) from the following paper in this issue. The high-
temperature portion of this fit is in fair agreement with prior
-11
3
5,7,8,11
ktot(1075-1140 K) ) 3.2 × 10 exp(-9686K/T) cm
recommendations.
Also shown at high temperatures is the
molecule-1 s-1 (9)
result for k2 from DDCH92. As can be seen, at high tempera-
tures, k2 from DDCH92 is much smaller than ktot; their measured
k1 and k2 values indicate that k1 (not shown) is considerably
larger than k2 above ∼2050 K and that k1 and k2 are equal at
(not shown). The corresponding 2σ precision and accuracy limits
are (12% and (26%, respectively.
∼
2050 K. Their measurements on k2 extended to a lower limit
of 1940 K. At this temperature, because of the error limits in
V. Discussion
the reported expressions, one cannot conclude directly that k2
The assumption of 5-10 ppm H2O used for much of the
modeling in section III is of special interest. A 5 ppm
concentration is required at the highest temperatures and 10 ppm
at the lowest temperatures used in the experiments to explain
the observations if the HS85 rate coefficient expressions were
correct and conjectural H2O contamination caused the fast
observed [O] decreases. The computed decay in Figure 2, which
was obtained with ktot values equal to those measured in the
present experiments at 1076 K and no H2O in the modeling
mixture, has a 0.037 s half-life, which is typical of the
experiments. If instead the HS85 k1 and k2 values are used for
the same conditions, the computed half-life is 0.134 s. Retaining
the HS85 k values, but including 5 ppm H2O in the initial
mixture, as in Figure 3, upper panel, the half-life is 0.047 s.
With 10 ppm H2O and the HS85 k values, the half-life is 0.037
s (not shown). These results demonstrate the potentially large
effect of H2O as a contaminant. However, the modeling results
also clearly indicate that if 5 ppm or more H2O had been present,
the experimental residual analysis procedures would have
revealed an underlying problem. Such is not the case, leading
to the conclusion that the HS85 Values are too small for the
present temperature range. Predictions were also made assuming
concentrations of H2O less than 5 ppm. At [H2O] e 1.5 ppm,
it is uncertain whether the residuals analysis would have detected
the nonexponentiality of the decays. However, as discussed in
detail in the Experimental Section, the [H2O] could not have
been higher than 0.2 ppm. This value is a very conservative
estimate. Therefore, calculations were performed with and
without 0.2 ppm H2O in the initial mixture, using either the
HS85 rate coefficients or those measured in the present work.
The difference in decay rates caused by such a trace amount of
H2O is calculated to be only a few percent, which is negligible
considering the experimental tolerances. Thus, it has been shown
on the basis of two lines of reasoning that H2O contamination
in the experiments cannot account for the difference between
the present rate coefficients and the smaller ones from preVious
recommendations.
>
k1. However, on the basis of their fitted expressions, the
authors of DDCH92 suggested k2 dominates below ∼2050 K.
6
Estimates for the k1 expression of the critical review, obtained
using the retained measurements of k1 from the literature prior
to fitting, indicated the proper k1 rate coefficients would be much
smaller at the temperatures of the present study than the
presently observed ktot values. Thus, the present experimental
6
ktot results, combined with the retained k1 Values, yield the
conclusion that O2 + N2 is the primary intermediate temperature
product channel, confirming the DDCH92 suggestion.
Also shown in Figure 6 are the results of earlier intermediate
temperature measurements. The recent upper limit ktot values
2
7
from Ross et al. disagree rather strongly with the present
results. Their upper limit data points result from complex
modeling of shock tube experiments designed primarily to obtain
rate coefficients for N2O + M. It is theoretically difficult to
reconcile such small intermediate temperature rate coefficient
values with the well-established, near-linear region of the high-
temperature results (about 1675-4080 K), whether the notion
is accepted that reaction R1 dominates at high temperature or
not. A fit of the high-temperature data together with the ref 27
data would produce a ktot Arrhenius plot with a very pronounced
downward curvature. On the other hand, an Arrhenius plot with
upward curvature, such as the recommendation from the
6
companion paper shown in Figure 6, is easily explained by
the occurrence of the two reaction channels. Additionally, k1
6
could be inferred via studies on the reverse reaction at
intermediate temperatures as low as 1370 K, as discussed in
detail in ref 6. These rate coefficient measurements survived
critical tests and are used in the overall fit for recommended
expressions. The intermediate temperature rate coefficients agree
well with the high-temperature data for k1. The k1 expression
from these studies of the reverse reaction is much larger at
intermediate temperatures than the results of ref 27 would
indicate. Thus, the ref 27 results disagree with not only the data
presented herein but all pertinent prior results.
In Figure 6, the present retained data are compared to other
Qualitative support for the notion that reaction R2 dominates
the reaction at intermediate temperatures, and that ktot is larger
than most prior works indicate, is provided by preliminary
7
results of interest. First, consider the HS85 recommendation
9
and the DDCH92 result for ktot (obtained in each case by adding
2
8
k1(T) and k2(T) expressions). Note both have been extrapolated
results of Lin and Tsay for k2 (see Figure 6). Comparison of

3
Kinetics of the O( P) + N2O Reaction
J. Phys. Chem. A, Vol. 104, No. 25, 2000 6011
the present ktot values directly to k2 from Lin and Tsay is
reasonable because the companion paper concludes that k1 only
obtain the result, were reported. The data, combined with other
works, support the suggestion of a recent high-temperature shock
tube study, DDCH92, that the two product channels do not
6
9
contributes ∼10% to ktot at the present temperatures. The Lin
and Tsay experiments utilized a static reactor with mixtures of
N2O/Ar. Complex modeling was necessary to extract the rate
coefficients. Their result agrees well with an extrapolation of
the DDCH92 results for k2 (or ktot); it is larger than the present
direct measurements indicate. Note, though, that the slope of
Lin et al.’s k2 agrees quite well with the present results for ktot.
Exact conditions and error limits for their results were not
available at the time of writing of this paper.
have equal rate coefficient expressions, as previously thought.
Rather, the O2 + N2 channel dominates at intermediate
temperatures. The new data provide a much more precise
intermediate temperature determination of the total rate coef-
ficients of the O + N2O reaction than can be obtained by
extrapolation of high-temperature results. In the companion
6
paper, the HTP data are considered along with earlier studies,
and recommended rate coefficients for the two channels are
presented.
VI. Conclusions
Acknowledgment. The work at RPI was supported by ARO
under Grants DAAH04-95-1-0098 and DAAG55-98-1-0183. We
thank W. F. Flaherty for his valuable technical assistance and
M. Leanovich for his help with some of the measurements.
The first measurements on the title reaction performed under
conditions in which it is well isolated from the effects of other
reactions, i.e., complex chemical modeling was not required to

6012 J. Phys. Chem. A, Vol. 104, No. 25, 2000
Fontijn et al.
References and Notes
(15) Marshall, P. Comput. Chem. 1987, 11, 219.
(
(
16) Ko, T.; Adusei, G. Y.; Fontijn, A. J. Phys. Chem. 1991, 95, 8745.
17) Draper, N. R.; Smith, H. Applied Regression Analysis; Wiley: New
(1) (a) Fifer, R. A. In Fundamentals of Solid-Propellant Combustion;
Kuo, K. K., Summerfield, M., Eds.; Progress in Astronautics and Aero-
nautics, Vol. 90, American Institute of Aeronautics and Astronautics: New
York, 1984; Chapter 4. (b) Melius, C. F. In Chemistry and Physics of
Energetic Materials; Bulusu, S. N., Ed.; Kluwer: Dordrecht, The Nether-
lands, 1990; Chapter 4.
York, 1981; Chapter 3.
18) Johnsson, J. E.; Glarborg, P.; Dam-Johansen, K. Twenty-Fourth
(
Symposium (International) on Combustion; The Combustion Institute:
Pittsburgh, 1992; p 917.
(
19) Allen, M. T.; Yetter, R. A.; Dryer, F. L. Int. J. Chem. Kinet. 1995,
7, 883.
20) (a) Anderson, W. R.; Ilincic, N.; Meagher, N. E.; Seshadri, K.;
Vanderhoff, J. A. 32nd JANNAF Combustion Subcommittee Meeting and
995 Propulsion Systems Hazards Subcommittee Meeting, Joint Sessions;
(
2) Miller, J. A.; Bowman, C. T. Prog. Energy Combust. Sci. 1989,
5, 287.
3) Dove, J. E.; Nip, W. S.; Teitelbaum, H. Fifteenth Symposium
International) on Combustion; The Combustion Institute: Pittsburgh, 1975;
p 903.
2
1
(
(
(
1
CPIA Publication 638; 1995; Vol. 1, p 197. (b) Anderson, W. R.; Meagher,
N. E.; Ilincic, N.; Seshadri, K.; Vanderhoff, J. A. Manuscript in preparation.
(4) Troe, J.; Wagner, H. Gg. In Physical Chemistry of Fast Reactions;
Levitt, B. P., Ed.; Plenum: London, 1973; Chapter 1.
(
21) R o¨ hrig, M.; Petersen, E. L.; Davidson, D. F.; Hanson, R. K. Int. J.
Chem. Kinet. 1996, 28, 599.
22) Glarborg, P.; Johnsson, J. E.; Dam-Johansen, K. Combust. Flame
994, 99, 523.
23) Masten, D. A.; Hanson, R. K.; Bowman, C. T. J. Phys. Chem. 1990,
4, 7119.
24) (a) Sausa, R. C.; Anderson, W. R.; Dayton, D. C.; Faust, C. M.;
(
5) Baulch, D. L.; Drysdale, D. D.; Horne, D. G. EValuated Kinetics
Data for High-Temperature Reactions. Vol. 2. Homogeneous Gas-Phase
(
Reactions of the H2-N2-O2 System; Butterworths: London, 1973.
1
(
6) Meagher, N. E.; Anderson, W. R. J. Phys. Chem. A 2000, 104,
013.
7) Hanson, R. K.; Salimian, S. In Combustion Chemistry; Gardiner,
W. C., Ed.; Springer: New York, 1985; Chapter 6.
(
6
9
(
(
Howard, S. L. Combust. Flame 1993, 94, 407. (b) Dayton, D. C.; Faust, C.
M.; Anderson, W. R.; Sausa, R. C. Combust. Flame 1994, 99, 323. (c)
Sausa, R. C.; Singh, G.; Lemire, G. W.; Anderson, W. R. Twenty-Sixth
Symposium (International) on Combustion; The Combustion Institute:
Pittsburgh, 1996; p 1043.
(
8) Tsang, W.; Herron, J. T. J. Phys. Chem. Ref. Data 1991, 20, 609.
(9) Davidson, D. F.; DiRosa, M. D.; Chang, A. Y.; Hanson, R. K. In
Shock WaVes; Takayama, K., Ed.; Springer: Berlin, 1992; Vol. 2, p 813.
10) Allen, M. T.; Yetter, R. A.; Dryer, F. L. Combust. Flame 1997,
09, 449.
11) Dean, A. M.; Bozzelli, J. W. Combustion Chemistry of Nitrogen.
(
1
(25) Lutz, A. E.; Kee, R. J.; Miller, J. A. Sandia National Laboratories
(
Technical Report SAND87-8248; October 1988.
(26) Wentworth, W. E. J. Chem. Educ. 1965, 42, 96, 162.
(27) Ross, S. K.; Sutherland, J. W.; Kuo, S. C.; Klemm, R. B. J. Phys.
Chem. A 1997, 101, 1104.
In Gas-Phase Combustion Chemistry; Gardiner, W. C., Jr., Ed.; Springer-
Verlag: New York, 2000; Chapter 2.
(
12) Zuev, A. P.; Starikovskii, A. Y. Khim. Fiz. 1991, 10, 179; SoV. J.
Chem. Phys. (Engl. Transl.) 1992, 10, 255.
(
(
13) Ko, T.; Marshall, P.; Fontijn, A. J. Phys. Chem. 1990, 94, 1401.
14) Marshall, P.; Ko, T.; Fontijn, A. J. Phys. Chem. 1989, 93, 1922.
(28) (a) Lin, M. C. Private communication. (b) Tsay, T. S. Master’s
Thesis, Emory University, 1994.