JCK(Wiley) RIGHT BATCH

Flow Reactor Studies and

Kinetic Modeling of the
H2/O2 Reaction
M. A. MUELLER, T. J. KIM, R. A. YETTER, F. L. DRYER
Department of Mechanical and Aerospace Engineering, Princeton University, Princeton, New Jersey 08536
Received 10 June 1998; accepted 20 August 1998

ABSTRACT: Profile measurements of the H2/O2 reaction have been obtained using a variable
pressure flow reactor over pressure and temperature ranges of 0.3–15.7 atm and 850–1040 K,
respectively. These data span the explosion limit behavior of the system and place significant
emphasis on HO2 and H2O2 kinetics. The explosion limits of dilute H2/O2/N2 mixtures extend
to higher pressures and temperatures than those previously observed for undiluted H2/O2
mixtures. In addition, the explosion limit data exhibit a marked transition to an extended
second limit which runs parallel to the second limit criteria calculated by assuming HO2
formation to be terminating. The experimental data and modeling results show that the ex-
tended second limit remains an important boundary in H2/O2 kinetics. Near this limit, small
increases in pressure can result in more than a two order of magnitude reduction in reaction
rate. At conditions above the extended second limit, the reaction is characterized by an overall
activation energy much higher than in the chain explosive regime.
The overall data set, consisting primarily of experimentally measured profiles of H2, O2,
H2O, and temperature, further expand the data base used for comprehensive mechanism de-
velopment for the H2/O2 and CO/H2O/O2 systems. Several rate constants recommended in an
earlier reaction mechanism have been modified using recently published rate constant data
for H ⫹ O2 (⫹ N2) ⫽ HO2 (⫹ N2), HO2 ⫹ OH ⫽ H2O ⫹ O2, and HO2 ⫹ HO2 ⫽ H2O2 ⫹ O2. When
these new rate constants are incorporated into the reaction mechanism, model predictions
are in very good agreement with the experimental data. 䉷 1999 John Wiley & Sons, Inc. Int J Chem
Kinet 31: 113– 125, 1999

INTRODUCTION extensively studied [2,3], reactions involving the for-

mation and consumption of HO2 and H2O2 are com-
The hierarchical development of detailed kinetic paratively less understood than the chain-branched re-
mechanisms for hydrocarbon oxidation and other actions which dominate at low pressures and high
complicated reacting systems necessarily begins with temperatures. In a previous paper [4], we presented a
the H2/O2 chemistry which plays a prominent role in moist CO oxidation reaction mechanism incorporating
determining the composition of the radical pool [1]. a detailed H2/O2 submechanism developed and vali-
While the kinetics of the H2/O2 systems have been dated using high-pressure (2.5 – 15.7 atm) flow reactor
data which placed significant emphasis on reactions
involving HO2 and H2O2 [5]. Since that publication,
Correspendence to: F. L. Dryer
Contract grant sponsor: Department of Energy, Office of Basic
new rate constant data have been published for several
Energy Sciences important reactions including H ⫹ O2 (⫹M) ⫽ HO2
Contract grant number: SE-FG02-86ER-13503 (⫹M) [6,7] and HO2 ⫹ OH ⫽ H2O ⫹ O2 [8]. In par-
Contract grant sponsor: Air Force Office of Scientific Research
Contract grant number: F49620-93-1-0427 ticular, the studies of Bromly et al. [6] and Davidson short
䉷 1999 John Wiley & Sons, Inc. CCC 0538-8066/99/020113-13 et al. [7] recommended rate expressions for H ⫹ O2 standard
long
JCK(Wiley) LEFT BATCH

114 MUELLER ET AL.

⫹ N2 ⫽ HO2 ⫹ N2 which, when incorporated into our dict explosion limit behavior when heterogeneous rad-
H2/O2 submechanism, adversely affected the predic- ical destruction is suppressed. In this case, HO2
tive ability of the CO/H2O/O2 mechanism at condi- consumption, primarily through HO2 ⫹ H ⫽ OH ⫹
tions where HO2/H2O2 chemistry is important. To re- OH or HO2 ⫹ HO2 ⫽ H2O2 ⫹ O2 followed by H2O2
solve this discrepancy, we recently derived new rate (⫹M) ⫽ OH ⫹ OH (⫹M), propagates chain carriers.
constant data for the low-pressure limit H ⫹ O2 ⫹ M
⫽ HO2 ⫹ M (M ⫽ N2,Ar) [9]. As discussed here, we
also obtained new experimental data on the H2/O2 re- REACTION MECHANISM
action in order to update the H2/O2 submechanism in
a manner consistent with the new rate constant data as The updated H2/O2 reaction mechanism consists of the
well as the concurrent development of H2/O2/NOX 19 reversible elementary reactions and thermochemi-
[10], H2/N2O [11], and H2/NO2 [12] reaction mecha- cal data listed in Tables I and II, respectively. Reverse
nisms. In addition, recent modifications to our exper- rate constants are computed from the forward rate con-
imental facility have extended the operating range of stants and the equilibrium constants. The thermochem-
our flow reactor to subatmospheric pressures. This ical data for each species considered in the mechanism
new capability enabled a more complete characteriza- are from Kee et al. [19] with the exception of the heat
tion of the explosion limit behavior of dilute H2/O2/N2 of formation for HO2 which is from Hills and Howard
mixtures. [20].
The entire data set of the present study spans pres- The reaction mechanism given in Table I is a re-
sure, temperature, and residence time ranges of 0.3 – vision of the H2/O2 submechanism of Kim et al. [4]
15.7 atm, 850 – 1040 K, and 0.03 – 1.4 sec, respec- which is itself a revision of the comprehensive
tively. Reaction profiles were obtained at initial con- CO/H2O/O2 reaction mechanism of Yetter et al. [21].
ditions within both the chain explosive and thermal/ The high-pressure flow reactor experiments of Kim [5]
chain explosive regimes of H2/O2 chemistry and were used to refine the rate constants of several reac-
particular emphasis was placed on exploring the ex- tions involving adducts as well as to incorporate high-
tended second explosion limit which demarcates these pressure fall-off in reactions (9) and (15). The revised
two kinetic regimes. In the chain explosive regime be- rate constants were literature values with the exception
tween the first and second explosion limits, H2/O2 ki- of that for HO2 ⫹ OH ⫽ H2O ⫹ O2 which was esti-
netics are largely controlled by the chain-branched se- mated by fitting available rate data. While our earlier
quence reaction mechanism accurately predicts the H2/O2 re-
action over a wide range of conditions, recent studies
H ⫹ O2 ⫽ OH ⫹ O (1) have refined the rate constant data for several elemen-
tary reactions involving the formation and consump-
O ⫹ H2 ⫽ H ⫹ OH (2) tion of HO2. These new data have been incorporated
into the updated reaction mechanism as follows.
OH ⫹ H2 ⫽ H2O ⫹ H (3)

and the overall reaction is characterized by an acti-

H ⴙ O2(ⴙM) ⴝ HO2(ⴙM)
vation energy approximately equal to that of the rate-
limiting reaction H ⫹ O2 ⫽ OH ⫹ O [13]. In the Although much effort has been made to determine the
thermal/chain explosive regime above the extended rate constant for this reaction, there remains consid-
second limit, H atoms are primarily consumed through erable scatter in the available data for the low-pressure
the termolecular reaction limit, k9,0, and only the study of Cobos et al. [22] has
investigated the high-pressure limit, k9,⬁. In a recent
H ⫹ O2(⫹M) ⫽ HO2(⫹M). (9) study [9], we obtained new rate constant data for
k9,0N2 at temperatures ranging from 800 – 900 K and
At conditions where the relatively unreactive HO2 rad- utilized previously reported data to derive a low-pres-
icals do not undergo further reaction or are destroyed sure limit of k9,0N2 ⫽ 3.5 ⫻ 1016 T⫺0.41exp(⫹1116/RT)
at vessel walls, reaction (9) is chain terminating and a cm6 mole⫺2 sec⫺1 over the temperature range 300 –
steady-state analysis of the radical pool gives the an- 1600 K. While the difference between this expression
alytical expression 2k1 ⫽ k9,eff commonly associated and the low-pressure limit of Slack [23] used in our
with the second limit. However, as shown by Baldwin earlier mechanisms is small over the temperature range
and co-workers [14 – 18], the consideration of gas- of 800 – 1200 K, we now determine high-pressure fall- short
phase consumption of HO2 radicals is necessary to pre- off using the Troe formulation [24] with the broad- standard
long
JCK(Wiley) RIGHT BATCH

FLOW REACTOR STUDIES AND KINETIC MODELING OF THE H2/O2 REACTION 115

Table I H 2/O2 Reaction Mechanism. Units Are cm3-mole-sec-kcal-K; k ⫽ ATn exp(⫺Ea/RT)

⌬H⬚298 A n Ea Reference
H2/O2 Chain Reactions
1. H ⫹ O2 ⫽ O ⫹ OH 16.77 1.91 ⫻ 1014 0.00 16.44 Pirraglia et al. [25]
2. O ⫹ H 2 ⫽ H ⫹ OH 1.85 5.08 ⫻ 104 2.67 6.29 Sutherland et al [55]
3. H2 ⫹ OH ⫽ H 2O ⫹ H ⫺15.01 2.16 ⫻ 108 1.51 3.43 Michael et al. [56]
4. O ⫹ H 2O ⫽ OH ⫹ OH 16.88 2.97 ⫻ 106 2.02 13.4 Sutherland et al. [57]
H2/O2 Dissociation/Recombination Reactions
5. H2 ⫹ M ⫽ H ⫹ H ⫹ Ma 104.2 4.58 ⫻ 1019 ⫺1.40 104.38 Tsang et al. [39]
H2 ⫹ Ar ⫽ H ⫹ H ⫹ Ar 104.2 5.84 ⫻ 1018 ⫺1.10 104.38 Tsang et al. [39]
6. O ⫹ O ⫹ M ⫽ O2 ⫹ Ma ⫺119.1 6.16 ⫻ 1015 ⫺0.50 0.00 Tsang et al. [39]
O ⫹ O ⫹ Ar ⫽ O2 ⫹ Ar ⫺119.1 1.89 ⫻ 1013 0.00 ⫺1.79 Tsang et al. [39]
7. O ⫹ H ⫹ M ⫽ OH ⫹ Ma ⫺102.3 4.71 ⫻ 1018 ⫺1.0 0.00 Tsang et al. [39]
8. H ⫹ OH ⫹ M ⫽ H 2O ⫹ Ma ⫺119.2 2.21 ⫻ 1022 ⫺2.00 0.00 Tsang et al. [39]
H ⫹ OH ⫹ Ar ⫽ H 2O ⫹ Ar ⫺119.2 8.41 ⫻ 1021 ⫺2.00 0.00 Tsang et al. [39]
Formation and Consumption of HO2
9. H ⫹ O2 ⫹ M ⫽ HO2 ⫹ Ma ⫺49.1 kO 3.5 ⫻ 1016 ⫺0.41 ⫺1.12 Mueller et al. [9]
H ⫹ O2 ⫹ Ar ⫽ HO2 ⫹ Ar ⫺49.1 kO 1.5 ⫻ 1015 0.00 ⫺1.00 Baulch et al. [60]
H ⫹ O2 ⫽ HO2c k⬁ 1.48 ⫻ 1012 0.60 0.00 Cobos et al. [22]
10. HO2 ⫹ H ⫽ H 2 ⫹ O2 55.1 1.66 ⫻ 1013 0.00 0.82 see text
11. HO2 ⫹ H ⫽ OH ⫹ OH ⫺36.47 7.08 ⫻ 1013 0.00 0.30 see text
12. HO2 ⫹ O ⫽ OH ⫹ O2 ⫺52.23 3.25 ⫻ 1013 0.00 0.00 Baulch et al. [30]
13. HO2 ⫹ OH ⫽ H 2O ⫹ O2 ⫺70.11 2.89 ⫻ 1013 0.00 ⫺0.50 Baulch et al. [30]
Formation and Consumption of H 2O2
14. HO2 ⫹ HO2 ⫽ H 2O2 ⫹ O2b ⫺38.53 4.20 ⫻ 1014 0.00 11.98 Hippler et al. [40]
HO2 ⫹ HO2 ⫽ H 2O2 ⫹ O2b 1.30 ⫻ 1011 0.00 ⫺1.63
15. H2O2 ⫹ M ⫽ OH ⫹ OH ⫹ Ma ⫺51.14 kO 1.20 ⫻ 1017 0.00 45.5 Warnatz [58]
H2O2 ⫹ Ar ⫽ OH ⫹ OH ⫹ Ar ⫺51.14 kO 1.90 ⫻ 1016 0.00 43.0 Brouwer et al. [59]
H2O2 ⫽ OH ⫹ OHc k⬁ 2.95 ⫻ 1014 0.00 48.4 Brouwer et al. [59]
16. H2O2 ⫹ H ⫽ H 2O ⫹ OH ⫺68.05 2.41 ⫻ 1013 0.00 3.97 Tsang et al. [39]
17. H2O2 ⫹ H ⫽ H 2 ⫹ HO2 ⫺16.57 4.82 ⫻ 1013 0.00 7.95 Tsang et al. [39]
18. H2O2 ⫹ O ⫽ OH ⫹ HO2 ⫺14.70 9.55 ⫻ 106 2.00 3.97 Tsang et al. [39]
19. H2O2 ⫹ OH ⫽ H 2O ⫹ HO2b ⫺31.58 1.00 ⫻ 1012 0.00 0.00 Hippler et al. [34]
H2O2 ⫹ OH ⫽ H 2O ⫹ HO2b 5.8 ⫻ 1014 0.00 9.56
a Efficiency factors for the collision partners of this pressure dependent reaction are: ⑀
H2O ⫽ 12.0; ⑀H2 ⫽ 2.5; and ⑀Ar ⫽ 0.75. All other
species have efficiencies equal to unity. When a rate constant is declared specifically for an Argon collision partner, the efficiency of Argon
is set to zero when determining M for the same reaction.
b Reactions 14 and 19 are expressed as the sum of the two rate expressions.

c Reaction 9 is given as a true fit with FN2 ⫽ 0.5 and FAr ⫽ 0.45. Reaction 15 is given as a true fit with F ⫽ 0.5.
c c c

Table II ⌬Hf(298.15), S(298.15), and cp (T) for Species Considered in the H2/O2 Reaction Mechanism. Units Are cal.
mole⫺1 deg⫺1 for cp and S and kcal mole⫺1 for ⌬Hf
Species ⌬Hf(298.15) S(298.15) cp(300) cp(500) cp(800) cp(1000) cp(1500) cp(2000)
H 52.103 ⫾ 0.001 27.416 ⫾ 0.004 4.97 4.97 4.97 4.97 4.97 4.97
O 59.55 ⫾ 0.024 38.49 ⫾ 0.005 5.23 5.08 5.02 5.00 4.98 4.98
OH 9.318 ⫾ 0.29 43.905 ⫾ 0.01 7.15 7.07 7.13 7.33 7.87 8.28
H2 0.0 31.232 ⫾ 0.008 6.90 7.00 7.07 7.21 7.73 8.18
O2 0.0 49.03 ⫾ 0.008 7.01 7.44 8.07 8.35 8.72 9.03
H2O ⫺57.795 ⫾ 0.01 45.13 ⫾ 0.01 8.00 8.45 9.22 9.87 11.26 12.22
HO2 3.0 ⫾ 0.4 54.75 ⫾ 0.02 8.34 9.49 10.78 11.39 12.45 13.22
H2O2 ⫺32.53 55.66 10.42 12.35 14.29 15.21 16.85 17.88
N2 0.0 45.93 ⫾ 0.005 6.95 7.08 7.50 7.83 8.32 8.60
short
Ar 0.0 37.00 ⫾ 0.001 4.97 4.97 4.97 4.97 4.97 4.97
standard
long
JCK(Wiley) LEFT BATCH

116 MUELLER ET AL.

ening factor FCN2 ⫽ 0.5 recommended by Cobos et al.

[22] instead of the Lindemann fit used previously. As
a result, the new pressure-dependent expression yields
values for k N9,eff
2 which are approximately 20 – 30%
lower at the conditions of the present study. As will
be shown in a later figure, the use of k N9,eff
2 with appro-
priate chaperone efficiencies and the k1 recommenda-
tion of Pirraglia et al. [25] results in very good agree-
ment with H2/O2 explosion limit data over the
temperature range 675 – 900 K. We note that the ex-
pression of Pirraglia et al. overpredicts recent data
[26 – 29] for H ⫹ O2 ⫽ OH ⫹ O at temperatures in
excess of 1700 K. However, extrapolation of recent k1
recommendations [30] to temperatures below 1000 K
yields values which are significantly higher than those
supported by combined (k1 ⫹ k9,0Ar[Ar]) rate data ob-
tained between 746 – 987 K [25]. This discrepancy
suggests that a non-Arrhenius temperature depen- Figure 1 Arrhenius plot of rate coefficient data for the
dency may be required to best represent k1 over the HO2 ⫹ OH ⫽ H2O ⫹ O2 reaction.
entire temperature range spanning 700 – 5300 K. We
continue to use the recommendation of Pirraglia et al.
below 1700 K and recommend additional studies to
better characterize k1 at temperatures below 1000 K.
O2 are rather sparse, particularly at temperatures of
HO2 ⴙ OH ⴝ H2O ⴙ O2 interest for combustion chemistry. Room-temperature
data [36,37] and modeling studies of H2/O2 explosion
The reaction of HO2 with OH competes with reaction
limit data in boric acid-coated vessels [18] and laminar
(11) for HO2 radicals and, in doing so, inhibits the
flame speeds of rich H2/O2/N2 mixtures [38] indicate
overall reaction rate. Due to its importance in the HOX
that reaction (11) is the dominant reaction pathway.
cycle of atmospheric chemistry, this reaction has been
The rate constants for reactions (10) and (11) used in
extensively studied and is well-characterized at low
our earlier mechanisms were from Tsang and Hamp-
temperatures [31 – 33]. The high-temperature data [34]
son [39] who based their recommendation upon the
used to develop the expression for k13 in our earlier
low temperature data of Sridharan et al. [37] and the
mechanism have been refined and additional data,
modeling results of Baldwin et al. [18].
shown in Figure 1, have been obtained by Hippler et
The rate constant data of Baldwin et al. [18], ex-
al. [35] over the temperature range 1100 – 1600 K.
pressed in the form of optimized values for the param-
This new data exhibits a highly non-Arrhenius tem-
eters k11/k1k141/2 and k10/k1k141/2 at 773 K, have been
perature dependency indicative of the formation of an
updated to incorporate the k14 recommendation of Hip-
activated complex. While a complete description of
pler et al. [40]. Furthermore, in order to provide a self-
this complex reaction requires more data over larger
consistent set of rate constants, we use the k1 expres-
ranges of pressure and temperature, it is likely that our
sion of Pirraglia et al. [25] rather than the expression
previous expression for k13 is too low at temperatures
used in the original study. The resulting rate constant
less than 1000 K. We therefore accept the recommen-
data at 773 K are combined with the room-temperature
dation of Baulch et al. [30] which is based upon the
data of Keyser [36] and Sridharan et al. [37] to derive
low-temperature work of Keyser [31]. This expression
the rate constants provided in Table I. These modified
does not capture the anomalous temperature depen-
rate constants are approximately 40% lower at 773 K
dency exhibited by the experimental data; however, at
than those recommended by Tsang and Hampson [39].
temperatures in excess of 1100 K, reaction (13) is of
We note that the reaction between HO2 radicals and
minor importance except at high pressures and fuel-
H atoms has a third possible product route, HO2 ⫹
lean conditions.
H ⫽ H2O ⫹ O, which is kinetically similar to reaction
HO2 ⴙ H ⴝ OH ⴙ OH; HO2 ⴙ H ⴝ H2 ⴙ O2 (11). The data of Keyser [36] and Sridharan et al. [37]
as well as the modeling results of Day et al. [38] in-
Rate data and product ratios for the competing reac- dicate that this reaction is much slower than reaction short
tions HO2 ⫹ H ⫽ OH ⫹ OH and HO2 ⫹ H ⫽ H2 ⫹ (11). Given the large uncertainty in its rate constant, standard
long
JCK(Wiley) RIGHT BATCH

FLOW REACTOR STUDIES AND KINETIC MODELING OF THE H2/O2 REACTION 117

we do not include HO2 ⫹ H ⫽ H2O ⫹ O in our mech-

anism.

FLOW REACTOR EXPERIMENT

A schematic of the variable pressure flow reactor Figure 3 Schematic of the fuel injector/mixer/diffuser as-
(VPFR) used in this study is shown in Figure 2. The sembly.
entire reactor is enclosed in a carbon steel pressure
vessel rated for operation from full vacuum to 30 atm.
Reactor pressures in excess of one atmosphere are chemical O2 analyzer (Infrared Industries Model
achieved by means of a back-pressure valve located 2200), and a selective thermal conductivity detector
within the reactor exhaust duct. Subatmospheric op- used to measure H2. The measurement uncertainties
eration is achieved by replacing this valve with a gas for data reported here are: H2O, ⫾10%; O2, ⫾2%; and
jet compressor which pulls a partial vacuum on the H2, ⫾5% of reading. The temperature of the reacting
entire pressure vessel. mixture is measured at the sampling location with a
Carrier gas (typically N2) is heated by a combined silica-coated type R thermocouple accurate to approx-
ferrous alloy/tungsten electrical resistance heater and imately ⫾3 K [41].
mixed with oxygen as it enters a 10.16 cm diameter The distance between the point of fuel injection and
quartz tube. The carrier gas/oxygen mixture flows sampling location is varied by moving the injector and
around a 8.9 cm baffle plate into a 0.64 cm gap serving mixer/diffuser assembly relative to the fixed sampling
as the entrance to a silica foam diffuser with a 5 degree location by means of a computer-controlled stepper
half angle (Figure 3). The hydrogen is diluted with motor. Mean velocity measurements along the center-
inert and injected radially outward into this gap where line of the reactor have been used to correlate distance
it rapidly mixes with the carrier gas and oxygen. The with residence time and experimental conditions are
reacting mixture then flows through the diffuser and chosen to produce reaction zones in which a centi-
into the test section where a hot-water cooled, stainless meter of distance corresponds to as little as 10⫺4 to
steel sampling probe continuously extracts and con- 10⫺2 seconds of reaction time. The dilute nature of the
vectively quenches a small percentage of the flow. The H2/O2/N2 mixtures used in this study ensures that the
sample gas flows via heated teflon lines to a series maximum temperature rise due to reaction is only a
analytical equipment including a Fourier transform in- small percentage (2 – 10%) of the initial gas tempera-
frared spectrometer (Nicolet Model 730), an electro- ture. Five individually controlled electric resistance
heaters maintain the local wall temperature of the
quartz tube at the initial gas temperature, thereby ap-
proximating an adiabatic reaction zone.
In the vicinity of the injection point, mixing and
diffusive effects are non-negligible and influence the
chemical induction time. High convective velocities in
the downstream test section are used to suppress spa-
tial gradients and, in general, permit neglect of the
diffusive terms in the governing equations. The ex-
perimental test section can therefore be modeled as a
zero-dimensional system using SENKIN [42] with
isobaric and adiabatic assumptions provided that there
is no memory effect of the chemical perturbations in
the mixing region. In an earlier paper [21], we have
shown that for the CO/H2O/O2 system, such pertur-
bations affect the chemical induction times and not the
reaction gradients in the test section. Similar analyses
Figure 2 Schematic of the variable pressure flow reactor of the H2/O2 system at the conditions used within the
with the gas ejector and back pressure valve which enable present study have confirmed the same behavior and,
operation at subatmospheric and superatmospheric pres- for comparative purposes, model predictions are short
sures, respectively. shifted in time to agree with the experimental data at standard
long
JCK(Wiley) LEFT BATCH

118 MUELLER ET AL.

the point of 50% fuel consumption. Under certain ex- The agreement between the experimental data and
perimental conditions where the characteristic time of model predictions shown in Figure 4 is very good with
the reaction becomes very short, large spatial gradients the exception that the model predicts a more abrupt
cannot be avoided. In these instances, the assumption transition than is observed experimentally. We also
of negligible axial diffusion is exacerbated and com- note that the measured temperature rise is approxi-
parisons between experimental data and model predic- mately 10 K lower than that predicted using the adi-
tions generated using the zero-dimensional approxi- abatic assumption. This latter discrepancy can be de-
mation must be regarded with caution. We include creased by incorporating a zero-dimensional heat loss
comments to this effect in the figure captions of the term, qLOSS ⫽ h(T ⫺ TWALL), into the energy equation
experimental data and model predictions for which solved within SENKIN. In this manner, the global heat
this cautionary note applies. transfer coefficient, h, can be adjusted to improve
agreement between the predicted and measured peak
temperatures. However, the inclusion of this heat loss
EXPERIMENTAL RESULTS AND term results in only slight changes in the predicted
DISCUSSION species profiles and, rather than introduce further com-
plexity into the model, we use the adiabatic approxi-
An example of experimental data (symbols) and mod- mation to generate the model predictions presented in
eling results (lines) obtained in this study is provided Figure 4 as well as in subsequent figures.
in Figure 4. The pressure (3.02 atm) and initial tem- The change in H2/O2 kinetics upon crossing the ex-
perature (934 K) chosen for this experiment are within tended second limit is further illustrated in Figure 5
the thermal/chain explosive regime above the explo- which shows experimental data obtained at three dif-
sion limit. The ensuing reaction begins slowly but ac- ferent pressures. The initial mixture temperature in
celerates as heat release increases the mixture temper- each of these experiments was approximately 935 K.
ature. When the reaction is approximately 25 percent At 2.55 atm, the mixture is within the chain explosive
complete, a critical state is reached, at which point the regime and, after a short induction period, a very rapid
reaction rapidly accelerates and quickly consumes the reaction completely consumes the reactants within a
remaining reactants. The transition to explosively-fast, few milliseconds. Increasing the pressure to 3.44 atm
chain-branched chemistry occurs when the tempera- moves the initial conditions into the thermal/chain re-
ture crosses the extended second limit, thereby estab- gime and the subsequent reduction in the overall
lishing an overall branching factor in excess of the branching factor results in a much slower reaction.
critical value. However, as before, the heat release of this slow re-
action is sufficient to bring the mixture to an explosive
condition. When the pressure is increased further to
6.00 atm, the heat release is no longer sufficient to
move to the mixture temperature into the chain explo-
sive regime and the reactants are consumed in a slow,
steady reaction.
Transitional chemistry of the nature shown in Fig-
ures 4 and 5(b) has been observed in the VPFR at
pressures ranging from 3.0 to 6.5 atm and provide di-
rect measurements of the extended second limit. At
higher pressures, the rate of reaction in the thermal/
chain regime increases and, as shown in Figure 6, the
transition upon crossing the explosion limit becomes
less pronounced and therefore difficult to observe ex-
perimentally. Conversely, at pressures lower than 3.0
atm, the reaction in the thermal/chain regime proceeds
so slowly as to be unobservable within the timescales
of our experiments. Under these circumstances, profile
Figure 4 Reaction profiles for the H2/O2 reaction at 3.02 measurements are unable to adequately resolve the
atm and Tin ⫽ 934 K. Symbols represent experimental data; transition from slow to fast chemistry. However, near
lines represent model predictions. Initial mole fractions are the explosion limit, the overall reaction rate becomes
listed in the Appendix. The zero-dimensional assumption extremely sensitive to small variations in temperature short
may not be valid during the rapid portion of the reaction. and, as observed in our experiments, the extent of re- standard
long
JCK(Wiley) RIGHT BATCH

FLOW REACTOR STUDIES AND KINETIC MODELING OF THE H2/O2 REACTION 119

Figure 6 Characteristic reaction times as a function of

temperature for dilute H2/O2/N2 mixtures at various pres-
sures. Maximum rates of reaction are determined using
SENKIN [38] with constant pressure and temperature as-
sumptions. Initial mole fractions are 1.0% H2, 0.5% O2, and
98.5% N2.

Figure 5 Reaction profiles of the H2/O2 reaction at: (a)

P ⫽ 2.55 atm, Tin ⫽ 935 K; (b) P ⫽ 3.44 atm, Tin ⫽ 933 K;
and (c) P ⫽ 6.0 atm, Tin ⫽ 934 K. Symbols represent ex-
perimental data; lines represent model predictions. Initial
mole fractions are listed in the Appendix. The zero-dimen-
sional assumption may not be valid for the reaction at 2.55
atm and during the rapid portion of the reaction at 3.44 atm.

action can be changed from zero to 100 percent by

increasing the initial temperature by only a few de-
grees. Explosion limit data from 0.4 to 2.5 atm could
therefore be ascertained by gradually increasing the
initial mixture temperature until a reaction was ob-
served upon injection of the hydrogen into the carrier
gas. Figure 7 Explosion limit data for stoichiometric H2/O2
The explosion limit data shown in Figure 7 were mixtures as observed in static reactor (䊊 䉮 䉫 䊊), well-
obtained using 1.0% H2/0.5% O2/N2 mixtures and rep- stirred reactor (ⵧ), and flow reactor (䉱) experiments. The
resent the minimum temperature at which explosively data from references [43–46] were obtained using undiluted
H2/O2 mixtures, those of Baldwin et al. [17] with mixtures
fast chemistry was observed at the corresponding pres-
of 28% H2, 14% O2, and 58% N2, and those of the present
sure. Also included in this figure are explosion limit
study with mixtures of 1% H2, 0.5% O2, and 98.5% N2.
data from several static reactor experiments. These lat- Relative third body efficiencies (␧H2: ␧N2: ␧O2 ⫽ 1.0: 0.43:
ter data have been modified to take into account the 0.40) are taken from von Elbe and Lewis [43] and are in-
third body efficiencies of H2 and O2 relative to N2. The corporated into pressures as follows: p ⫽ RT[M] ⌺i ␹i␧i.
error bars provided for our data are indicative of the Model predictions (M ⫽ N2) are generated using plots sim- short
reproducibility of our experimental results. ilar to those in Figure 6. standard
long
JCK(Wiley) LEFT BATCH

120 MUELLER ET AL.

As seen in Figure 7, the VPFR data extend well ond limit criteria. Furthermore, the H2/O2 mechanism
beyond the cusp formed by the convergence of the described in Tables I and II accurately predicts explo-
second and third limits observed in static reactor ex- sion limit behavior at conditions dominated by gas-
periments. At low pressures, the data are in good phase kinetics.
agreement with the classical second limit criteria of Reaction profile measurements at subatmospheric
2k1 ⫽ k9,eff. With increasing pressure, the VPFR data pressures were obtained at initial conditions within the
digress from this limit; however, rather than fold back chain explosive regime where radical losses via het-
to form a third explosion limit, the data shift to an erogeneous wall reactions are negligibly small. Fuel-
extended second limit which runs parallel to the clas- lean mixtures with initial hydrogen mole fractions of
sical second limit. These unique characteristics can be 0.5% were used to limit peak radical concentrations,
explained by considering experimental timescales and thereby reducing experimental uncertainty due to dif-
mixture dilution effects. The relatively long timescales fusive phenomena and recombination reactions within
of static reactor experiments allow the observation of the sampling probe. Figure 8 presents reaction profile
reactions with characteristic times much longer than data obtained at 0.3 atm with Tin ⫽ 880 K and an
the residence times of reasonable flow reactor exper- equivalence ratio of 0.5. Model predictions are in ex-
iments. In an adiabatic system, the heat release from cellent agreement with the experimental data and in-
these slow reactions can bring the mixture to an ex- dicate that the radical pool is largely dominated by H
plosion condition far removed from the initial condi- atoms. Because the reaction proceeds in the high-tem-
tions. The use of undiluted H2/O2 mixtures intensifies perature regime where k1 ⬎ k9,eff, the largest sensitivity
self-heating effects and, as noted in several studies, coefficients are associated with the chain-branched se-
can impose a large thermal influence upon the exper- quence consisting of H ⫹ O2 ⫽ OH ⫹ O, O ⫹ H2 ⫽
imental data in the vicinity of the third limit [46 – 48]. OH ⫹ H, and OH ⫹ H2 ⫽ H2O ⫹ H (Figure 9). How-
In fact, the data of Heiple and Lewis [46] are the initial ever, the formation and consumption of HO2 remains
pressures and temperatures for which stoichiometric an important reaction pathway for radical species. H
H2/O2 mixtures eventually reached an explosive con- atoms are primarily produced through OH ⫹ H2 ⫽
dition and do not represent the actual conditions at H2O ⫹ H and are consumed via
which the chemistry became explosively fast. In con-
trast, the dilute mixtures used in this study limit self- H ⫹ O2(⫹M) ⫽ HO2(⫹M) (44.5%) (9)
heating and allow the investigation H2/O2 kinetics
within narrow temperature ranges. As such, our high- HO2 ⫹ H ⫽ OH ⫹ OH (33.8%) (11)
pressure data are more indicative of the kinetic H ⫹ O2 ⫽ OH ⫹ O (15.7%) (1)
changes that occur when crossing the explosion limit.
The departure of the explosion limit data from the HO2 ⫹ H ⫽ H2 ⫹ O2 (5.9%) (10)
classical second limit occurs when HO2 consumption
via gas-phase reactions begins to propagate the radical
pool. As shown in Figure 7, model predictions based
solely upon gas-phase kinetics are in excellent agree-
ment with the experimental data along the extended
second limit. The accurate prediction of data along the
classical second limit requires the development of a
mathematical model with both the spatial and temporal
dependence necessary to describe the removal of HO2
radicals by heterogeneous surface kinetics or mass
transport (see for example [49,50]). At low pressures,
we attribute the loss of chain carriers to heterogeneous
reactions or convective losses in which HO2 acts as a
radical sink. The importance of these loss mechanisms
is system dependent and, as a result, the location (i.e.,
pressure and temperature) of the shift in our explosion
limit data is unique to the design and operating char-
acteristics of the VPFR. Nonetheless, the experimental Figure 8 Reaction profiles of the H2/O2 reaction at 0.3 atm
data clearly indicate that when HO2 becomes part of a and Tin ⫽ 880 K. Symbols represent experimental data; lines
propagating route, the explosion criteria shifts to lower represent model predictions. Initial mole fractions are given short
temperatures than those predicted by the classical sec- in the Appendix. standard
long
JCK(Wiley) RIGHT BATCH

FLOW REACTOR STUDIES AND KINETIC MODELING OF THE H2/O2 REACTION 121

Experimental data and model predictions for three sets

of these experiments are presented in Figure 10. The
initial conditions at 0.6 atm are within the chain ex-
plosive regime, those at 15.7 atm are within the ther-
mal/chain regime, and those at 2.55 atm are in close
proximity to the extended second limit. In each case,
the addition of excess oxygen increases the ratios
[O]/[H] and [OH]/[H]; however, inhibition of the
overall reaction rate is only observed at 2.55 atm. As
shown in Table III, H atoms are the predominant rad-
ical species at 0.6 atm for both equivalence ratios. As
a result, the flux of HO2 through reactions (12) and
(13) increases only slightly with the additional oxy-
gen. Reaction (11) remains the primary consumption
route for HO2 and the small inhibiting effect accordant
with the increased flux through reactions (12) and (13)
is more than offset by the acceleration of reactions (1)
Figure 9 Normalized sensitivity coefficients (Sij ⫽ ⭸ lnYi/
⭸ lnKj) for the H2 mole fraction profile shown in Figure 8.

where the numbers in the second column represent the

percentage of H atoms consumed by the corresponding
reaction. Due to partial equilibration of reaction (1),
reaction (9) becomes the largest consumption route for
H atoms even though k1 ⬎ k9,eff. Hydroperoxyl radicals
are rapidly consumed via reaction (11) thereby form-
ing a straight-chain sequence, consisting of reactions
(9), (11), and (3), which competes with the chain-
branched sequence.
In the chain explosive regime, reactions (10) and
(11) are the dominant consumption routes for HO2. For
the 0.3 atm experiment discussed here, these reactions
consume over 90% of the HO2 with reaction (11) alone
accounting for 77% of the total flux. With increasing
pressure, H atom concentrations are suppressed and
alternative consumption routes become increasingly
active. The importance of these other reaction path-
ways becomes evident when studying the kinetic re-
sponse of the H2/O2 reaction to changes in stoichi-
ometry. The addition of excess oxygen heightens the
role of O and OH radicals as chain carriers and con-
sequently increases the flux of HO2 through

HO2 ⫹ O ⫽ O2 ⫹ OH (12)

HO2 ⫹ OH ⫽ H2O ⫹ O2. (13)

Figure 10 Reaction profiles for stoichiometric and fuel-
The degree to which these terminating reactions in- lean H2/O2/N2 mixtures at (a) P ⫽ 0.6 atm, Tin ⬵ 897 K; (b)
hibit the overall reaction depends upon the concentra- P ⫽ 2.55 atm, Tin ⬵ 940 K; and (c) P ⫽ 15.7 atm, Tin ⫽
tions of O and OH radicals relative to those of H and 914 K. Symbols represent experimental data; lines represent
HO2. model predictions. Exact initial conditions are listed in the
The effect of oxygen addition was experimentally Appendix. The zero-dimensional assumption may not be short
investigated at pressures ranging from 0.6 to 15.7 atm. valid for the stoichiometric reaction at 2.55 atm. standard
long
JCK(Wiley) LEFT BATCH

122 MUELLER ET AL.

Table III Predicted Composition of Radical Pool for the Experiments Shown in Figure 10. Species Mole Fractions
Are Those at the Maximum Rate of Reaction and Are Given in ppm
P ⫽ 0.6 atm P ⫽ 2.55 atm P ⫽ 15.7 atm
Species ␾ ⫽ 0.75 ␾ ⫽ 0.33 ␾ ⫽ 1.0 ␾ ⫽ 0.33 ␾ ⫽ 1.0 ␾ ⫽ 0.27
H 318 215 127 6.3 0.02 0.01
O 27.2 39.1 10.9 2.7 0.002 0.004
OH 18.5 29.7 13.0 3.0 0.006 0.01
HO2 0.9 2.3 3.8 11.4 4.4 6.4
H 2O 2 0.02 0.08 0.1 3.6 12.4 22.6

and (9) which are part of the chain-branched and via HO2 ⫹ HO2 ⫽ H2O2 ⫹ O2, while negligibly small
straight-chain sequences characteristic of the chain ex- for the stoichiometric reaction, accounts for 20% of
plosive regime. the total flux when ␾ ⫽ 0.33. At high pressures and
In the vicinity of the extended second limit, the temperatures, the primary consumption route of H2O2
composition of the radical pool is highly sensitive to is through thermal decomposition; however, at 2.55
mixture stoichiometry. In particular, the reduction in atm there are sufficient H, O, and OH radicals to con-
equivalence ratio from 1.0 to 0.33 at 2.55 atm reduces sume approximately 50% of the H2O2 via reactions
the H atom mole fraction by a factor of 20. The O and (16 – 19). These reaction pathways effectively de-
OH mole fractions are also reduced, but to a much crease chain propagation and, when combined with the
lesser extent, and HO2 becomes the most abundant increased flux of HO2 through reactions (12) and (13),
radical species. Because the mole fractions of H, O, significantly slow the overall reaction rate.
and OH are of the same order of magnitude at the fuel- At 15.7 atm, the addition of excess oxygen only
lean condition, reactions (12) and (13) can compete slightly alters the radical pool and the HO2 and H2O2
effectively with reaction (11). Flux analyses of these mole fractions are several orders of magnitude larger
experiments are shown in Figure 11 and indicate that than those of H, O, and OH. At both equivalence ra-
the consumption of HO2 via reactions (12) and (13) tios, the consumption of HO2 is overwhelmingly
increases from 9% to 26% of the total flux upon ad- through HO2 ⫹ HO2 ⫽ H2O2 ⫹ O2. Due to the high
dition of the excess O2. In addition, H2O2 formation pressure, hydrogen peroxide quickly dissociates via
H2O2 (⫹M) ⫽ OH ⫹ OH (⫹M) and thereby propa-
gates the radical pool through a straight-chain se-
quence consisting of reactions (9), (14), (15), and (3).
Under these circumstances, the additional oxygen ac-
celerates the overall reaction by accelerating HO2 pro-
duction via reaction (9).
The kinetic response to changes in pressure and
stoichiometry are summarized in Figure 12 in which
the characteristic reaction rates of stoichiometric and
fuel-lean H2/O2/N2 mixtures at an initial temperature
of 935 K are plotted as a function of pressure. At con-
ditions somewhat removed from the extended second
limit, increases in pressure increase the rate of reac-
tion. However, as the second limit is crossed, the re-
duction in chain branching reduces the overall reaction
rate by over two orders of magnitude. The addition of
excess oxygen inhibits the overall reaction only in the
vicinity of the extended second limit and, at conditions
within the thermal/chain explosive and chain explo-
sive regimes, the overall reaction rate increases at the
Figure 11 Reaction flux analyses of the consumption of fuel-lean condition.
HO2 in the stoichiometric and lean H2/O2 reactions at 2.55 Neglecting the small amount of branching from short
atm shown in Figure 10. H ⫹ O2 ⫽ OH ⫹ O, the only chain sequence which standard
long
JCK(Wiley) RIGHT BATCH

FLOW REACTOR STUDIES AND KINETIC MODELING OF THE H2/O2 REACTION 123

Figure 12 Maximum reaction rate as a function of pressure

Figure 13 Reaction profiles of lean (␸ ⬇ 0.3) H2/O2/N2
for stoichiometric and fuel-lean H2/O2/N2 mixtures with re-
mixtures at 6.5 atm and varying initial temperatures. Sym-
action temperatures of 935 K. Maximum rates of reaction
bols represent experimental data; lines represent model pre-
are determined using SENKIN [42] with constant pressure
dictions. Initial mole fractions are listed in the Appendix.
and temperature assumptions. Initial mole fractions are 1.0%
H2, 0.5% O2, and 98.5% N2 for ␸ ⫽ 1 and 1.0% H2, 1.5%
O2, and 97.5% N2 for ␸ ⫽ 0.33.
k9,eff ⬎⬎ 2k1. These results support those of Baldwin
and Mayer [14] who determined an activation energy
leads to chain branching at high pressures is of 57.0 kcal/mole for the slow reaction of H2/O2
mixtures in boric-acid coated vessels. When compared
H ⫹ O2(⫹ M) ⫽ HO2(⫹ M) (9) to the overall activation energy of approximately 16

HO2 ⫹ H2 ⫽ H2O2 ⫹ H (17R)

H2O2(⫹ M) ⫽ OH ⫹ OH(⫹ M). (14)

However, this sequence is slow relative to the straight

chain sequences discussed earlier and, in the thermal/
chain regime, H2/O2 kinetics are nearly straight-chain
in nature. Figure 13 presents H2 mole fraction profiles
from a series of experiments at 6.5 atm in which the
initial temperature was varied from 884 to 934 K. With
the exception of the first few data points in the profiles
obtained at initial temperatures above 906 K, model
predictions and experimental data are in good agree-
ment. By defining the overall reaction rate as
d[H2]/dt ⫽ ⫺kov[H2][O2]1/2 where kov ⫽ Aexp(⫺Eov/
RT), these data can be used to estimate the overall
activation energy in the thermal/chain explosive re-
gime. The reaction rate data, shown in Arrhenius form
Figure 14 Arrhenius plot of overall rate constant data de-
in Figure 14, indicate an overall activation energy of rived from the experimental data shown in Figure 13 except
61 ⫾ 10 kcal/mole which is in good agreement with for those obtained at Tin ⫽ 884 K (for which there is a very
a value of Eov ⫽ E1 ⫹ E15 ⫺ E16 ⫽ 58.3 kcal/mole small extent of reaction). Symbols represent experimental
calculated using a steady state analysis of reactions data; the line is a least-squares fit of the form kov ⫽ short
(1 – 3), (9), and (14 – 16) with the assumption that Aexp(⫺Eov/RT). standard
long
JCK(Wiley) LEFT BATCH

124 MUELLER ET AL.

Appendix Initial Conditions for the Experimental This work was supported by the Department of Energy, Of-
Data Reported Here (xN2 ⫽ balance) fice of Basic Energy Sciences through Grant No. DE-FG02-
86ER-13503 and the Air Force Office of Scientific Research
mole fractions
under Grant No. F49620-93-1-0427. The authors also thank
[%]
Mr. Paul Michniewicz for his assistance in performing the
Figure P [atm] Tin [K] H2 O2 experiments.
Author’s Note: As originally referenced by Kim et al.
4 3.02 934 0.95 0.49 [4], this paper was to be submitted to this journal under the
5(a) 2.55 935 1.01 0.52 authorship of T. J. Kim, R. A. Yetter, and F. L. Dryer. The
5(b) 3.44 933 1.01 0.52 current manuscript supersedes this previous reference.
5(c) 6.00 934 1.01 0.52
8 0.30 880 0.50 0.50
10(a) 0.60 897 0.50 0.34
0.60 896 0.50 0.76
10(b) 2.55 935 1.01 0.52 REFERENCES
2.50 943 1.00 1.50
10(c) 15.70 914 1.18 0.61 1. Westbrook, C. K.; Dryer, F. L. Prog Energy Combust
15.70 914 1.18 2.21 Sci 1984, 10, 1–57.
13 6.50 884 1.29 2.19 2. von Elbe, G.; Lewis, B. Combustion, Flames, and Ex-
6.50 889 1.30 2.21 plosions of Gases, 3rd ed; Academic: Orlando, FL,
6.50 906 1.32 2.19 1987.
6.50 914 1.36 2.24 3. Dixon-Lewis, G.; Williams, D. J. In Comprehensive
6.50 934 1.36 2.24 Chemical Kinetics; Bamford, C. H.; Tipper, C.F.H.,
Eds.; Elsevier Scientific: Amsterdam, 1977.
4. Kim, T. J.; Yetter, R. A.; Dryer, F. L. Twenty-Fifth
Symposium (International) on Combustion; The Com-
kcal/mole in the chain explosive regime, the high over- bustion Institute: Pittsburgh, PA, 1994; pp 759– 766.
all activation energy of the H2/O2 reaction in the ther- 5. Kim, T. J. MSE Thesis, Department of Mechanical and
mal/chain explosive regime underscores the impor- Aerospace Engineering, Princeton University, Prince-
tance of self-heating at conditions where k9,eff ⬎ 2k1. ton, NJ, 1994.
6. Bromly, J. H.; Barnes, F. J.; Nelson, P. F.; Haynes, B.
S. Int J Chem Kinet 1995, 27, 1165–1178.
7. Davidson, D. F.; Peterson, E. L.; Röhrig, M.; Hansen,
SUMMARY AND CONCLUSIONS R. K.; Bowman, C. T. Twenty-Sixth Symposium (In-
ternational) on Combustion; The Combustion Institute:
Reaction profile measurements of dilute H2/O2/N2 Pittsburgh, PA, 1996; pp 481–488.
mixtures have been obtained at conditions ranging 8. Hippler, H.; Neunaber, H.; Troe, J. J Chem Phys 1995,
from 0.3 to 15.7 atm and 850 – 1040 K. These data, 103, 3510–3516.
particularly those obtained at subatmospheric pres- 9. Mueller, M. A.; Yetter, R. A.; Dryer, F. L. Presented at
the Twenty-Seventh Symposium (International) on
sures, represent a unique contribution to the experi-
Combustion, The Combustion Institute, Pittsburgh, PA,
mental data base available for the development of 1998.
comprehensive reaction mechanisms. The updated H2/ 10. Mueller, M. A.; Yetter, R. A.; Dryer, F. L., work in
O2 mechanism provided in Tables I and II accurately progress.
predicts the experimental data over the entire range of 11. Allen, M. T.; Yetter, R. A.; Dryer, F. L. Combust. Flame
conditions explored in this study. 1998, 112, 302–311.
New explosion limit data were also obtained for 12. Gatto, J. L. MSE Thesis, Department of Mechanical and
dilute H2/O2/N2 mixtures at pressures ranging from 0.4 Aerospace Engineering, Princeton University, Prince-
to 6.5 atm. The low-pressure data agree with the clas- ton, NJ, 1997.
sical second limit criteria of 2k1 ⫽ k9,eff, but at higher 13. Yetter, R. A.; Dryer, F. L.; Rabitz, H. Combust Sci and
pressures where gas-phase consumption of HO2 radi- Tech 1991, 79, 97–128.
14. Baldwin, R. R.; Mayor, L. Trans Faraday Soc 1960, 56,
cals becomes important, the data shift to an extended
80–92.
second limit which remains an important boundary in 15. Baldwin, R. R.; Mayor, L.; Doran, P. Trans Faraday Soc
H2/O2 kinetics. In the thermal/chain explosive regime 1960, 56, 93–102.
above the extended second limit, the reaction is nearly 16. Baldwin, R. R.; Mayor, L. Trans Faraday Soc 1960, 56,
straight chain and is characterized by an overall acti- 103–114.
vation energy significantly higher than in the chain 17. Baldwin, R. R.; Jackson, D.; Walker, R. W.; Webster, short
explosive regime. S. J. Trans Faraday Soc 1967, 63, 1676– 1686. standard
long
JCK(Wiley) RIGHT BATCH

FLOW REACTOR STUDIES AND KINETIC MODELING OF THE H2/O2 REACTION 125

18. Baldwin, R. R.; Fuller, M. E.; Hillman, J. S.; Jackson, 41. Held, T. J. Ph.D. Thesis, Department of Mechanical and
D.; Walker, R. W. Trans Faraday Soc 1974, 70, 635– Aerospace Engineering, Princeton University, Prince-
641. ton, NJ, 1993.
19. Kee, R. J.; Rupley, F. M.; Miller, J. A. Sandia National 42. Lutz, A. E.; Kee, R. J.; Miller, J. A. Sandia National
Laboratories Report No. SAND87-8215, 1987. Laboratories Report No. SAND87-8248, 1987.
20. Hills, A. J.; Howard, C. J. J Chem Phys 1984, 81, 4458– 43. von Elbe, G.; Lewis, B. J Chem Phys 1942, 10, 366–
4465. 393.
21. Yetter, R. A.; Dryer, F. L.; Rabitz, H. Combust Sci and 44. Egerton, A.; Warren, D. R. Proc Roy Soc 1950, A 204,
Tech 1991, 79, 129– 140. 465–476.
22. Cobos, C. J.; Hippler, H.; Troe, J. J Phys Chem 1985, 45. Baulch, D. L.; Griffiths, J. F.; Pappin, A. J.; Sykes, A.
89, 342– 349.
F. Combust Flame 1988, 73, 163–185.
23. Slack, M. W. Combust. Flame 1977, 28, 241–249.
46. Heiple, R.; Lewis, B. J Chem Phys 1941, 9, 584– 590.
24. Gilbert, R. G.; Luther, K.; Troe, J. Ber Bunsenges Phys
47. Foo, K. K.; Yang, C. H. Combust Flame 1971, 17, 223–
Chem 1983, 87, 169– 177.
235.
25. Pirraglia, A. N.; Michael, J. V.; Sutherland, J. W.;
Klemm, R. B. J Phys Chem 1989, 93, 282–291. 48. Kordylewski, W.; Scott, S.K. Combust Flame 1984, 57,
26. Masten, D. A.; Hanson, R. K.; Bowman, C. T. J Phys 127–139.
Chem 1990, 94, 7119– 7128. 49. Maas, U.; Warnatz, J. Combust Flame 1988, 74, 53–
27. Du, H.; Hessler, J. P. J Chem Phys 1992, 96, 1077– 69.
1092. 50. Kreutz, T. G.; Law, C. K. Combust Flame 1998, 114,
28. Yang, H.; Gardiner, W. C.; Shin, K. S.; Fujii, N. Chem 436–456.
Phys Lett 1994, 231, 449– 453. 51. Friswell, N. J.; Sutton, M. M. Chem Phys Lett 1972,
29. Ryu, S.-O.; Hwang, S. N.; Rabinowitz, M. J. J Phys 15(1), 108–112.
Chem 1995, 99, 13984– 13991. 52. DeMore, W. B. J Phys Chem 1982, 86, 121– 126.
30. Baulch, D. L.; Cobos, C. J.; Cox, R. A.; Frank, P.; Hay- 53. Peters, J.; Mahnen, G. Fourteenth Symposium (Inter-
man, G.; Just, Th.; Kerr, J. A.; Murrells, T.; Pilling, M. national) on Combustion; The Combustion Institute,
J.; Troe, J.; Walker, R. W.; Warnatz, J. J Phys Chem Pittsburgh, PA, 1972; pp 133–146.
Ref Data 1994, 23, 847– 1033. 54. Goodings, J. M.; Hayhurst, A. N. J Chem Soc, Faraday
31. Keyser, L. F. J Phys Chem 1988, 92, 1193–1200. Trans 2 1988, 84, 745–762.
32. Sirdharan, U. C.; Qiu, L. X.; Kaufman, F. J Phys Chem 55. Sutherland, J. W.; Michael, J. V.; Pirraglia, A. N.; Nes-
1984, 88, 1281– 1282. bitt, F. L.; Klemm, R. B. Twenty-First Symposium (In-
33. Lii, R. R.; Gorse, R. A.; Sauer, M. C.; Gordon, S. J Phys ternational) on Combustion; The Combustion Institute,
Chem 1980, 84, 819– 821.
Pittsburgh, PA, 1986; pp 929–941.
34. Hippler, H.; Troe, J. Chem Phys Lett 1992, 192(4), 333–
56. Michael, J. V.; Sutherland, J. W. J Phys Chem 1988,
337.
92, 3853–3857.
35. Hippler, H.; Neunaber, H.; Troe, J. J Chem Phys 1995,
57. Sutherland, J. W.; Patterson, P. M.; Klemm, R. B.
103, 3510– 3516.
36. Keyser, L. F. J Phys Chem 1986, 90, 2994–3003. Twenty-Third Symposium (International) on Combus-
37. Sridharan, U. C.; Qiu, L. X.; Kaufman, F. J Phys Chem tion; The Combustion Institute, Pittsburgh, PA, 1990;
1982, 86, 4569– 4574. pp 51–57.
38. Day, M. J.; Thompson, K.; Dixon-Lewis, G. Fourteenth 58. Warnatz, J. In Combustion Chemistry; Gardiner, W. C.,
Symposium (International) on Combustion; The Com- Ed.; Springer-Verlag: New York, NY, 1985.
bustion Institute: Pittsburgh, PA, 1972; pp 47–59. 59. Brouwer, L.; Cobos, C. J.; Troe, J.; Dubal, H. R.; Crim,
39. Tsang, W.; Hampson, R. F. J Phys Chem Ref Data 1986, F. F. J Chem Phys 1985, 86, 6171–6182.
15, 1087– 1279. 60. Baulch, D. L.; Drysdale, D. D.; Horne, D. G.; Lloyd, A.
40. Hippler, H.; Troe, J.; Willner, J. J Chem Phys 1990, 93, C. Evaluated Kinetic Data for High Temperature Re-
1755– 1760. actions, Vols. 1 and 2; Butterworths, London, 1973.

short
standard
long
JCK(Wiley) BLANK

short
standard
long