Professional Documents
Culture Documents
Miller Et Al - 72
Miller Et Al - 72
00 + 50
Pnnted in Great Britain. All rights reserved. ( 1989 Pergamon Press pLc
Abstract--Our current understanding of the mechanisms and rate parameters for the gas-phase reactions
of nitrogen compounds that are applicable to combustion-generated air pollution is discussed and illus-
trated by comparison of results from detailed kinetics calculations with experimental data. In particular,
the mechanisms and rate parameters for thermal and prompt NO formation, for fuel nitrogen conversion,
for the Thermal De-NO~ and RAPRENO, processes, and for NO2 and N20 formation and removal
processes are considered. Sensitivity and rate-of-production analyses are applied in the calculations to
determine which elementary reactions are .of greatest importance in the nitrogen conversion process.
Available information on the rate parameters for these important elementary reactions has been surveyed.
and recommendations for the rate coefficients for these reactions are provided. The principal areas of
uncertainty in nitrogen reaction mechanisms and rate parameters are outlined.
CONTENTS
1. Introduction 287
2. Computational Methods and Reaction Mechanisms 288
3. Thermal NO Mechanism 289
4. Prompt NO Mechanism 290
5. NO Production from Fuel Nitrogen 298
5.1. The oxidation of hydrogen cyanide 298
5.2. The NO --. HCN --* N: mechanism 301
5.3. The oxidation of NH 3 305
5.4. Fuel-nitrogen conversion in well-stirred reactors 309
6. Thermal De-NO, and RAPRENOx Mechanisms 312
6.1. Thermal De-NOt mechanism 312
6.2. RAPRENO,. mechanism 319
7. Nitrogen Dioxide Mechanism 323
8. Nitrous Oxide Mechanism 324
9. Concluding Remarks 327
Acknowledgements 327
References 327
Appendix A 331
Appendix B 337
surveyed, and recommendations for the rate coef- central role in the discussions that follow. The
ficients for these reactions are provided. Finally, the methods used in performing the sensitivity analyses
principal areas of uncertainty in nitrogen reaction have been described in detail in previous articles. 9=t: In
mechanisms and rate parameters are outlined. the flame and homogeneous reaction calculations, the
In the following sections, the gas-phase reaction sensitivity coefficients are displayed as
mechanisms and rate parameters for thermal and exk
A
prompt NO formation, for fuel nitrogen conversion, : /ax, (x),
for the Thermal De-NO, and RAPRENO~ processes,
and for NO, and N,.O formation and removal where [3~k(x) is the sensitivity coefficient for change in
processes are discussed separately; however, as will be the mole fraction of the k'" species, Xk, due to a small
seen, there are instances where particular reactions change in the temperature-independent factor of the
are important in several of the reaction mechanisms. /tit reaction rate coefficient, A , The superscript on
For example, there are many common features in the X2krnax~ indicates the maximum value of Xk that
reaction mechanisms for prompt NO formation and occurred in the calculation: x is position above the
fuel nitrogen conversion. burner surface (for a homogeneous reaction x is
In preparing this paper, we were fortunate to be replaced by time, t). For a PSR calculation, there is no
able to draw upon a large body of published work, spatial or time dependence, and X k = X2ma~.
including several extensive compilations of elemen- Otherwise, fl~k is specified in the same way as in the
tary rate coefficients, t'2Ar°'169']7° As appropriate, we flame and homogeneous reaction calculations. Con-
direct the reader to these sources. tribution factors obtained from rate-of-production
(or reaction path) analysis are displayed as
negligible, the most important one being temperature recommendation of Baulch e t al. for the
forward reaction. The expression for kA233is the same
ICH: + H: , • CH~ + H. (All0) as that recommended in Ref. 1. The expression for
kA234is given for the reaction, N + OH --* NO + H,
In the following sections the triplet state is written as discussed by Miller e t al. ~°
CH2 and the singlet state is written 'CH2. In addition, Reactions (A232)-(A234) involve the O and OH
we have introduced more accurate formulations of the radicals, which also play an important role in the fuel
pressure dependence of the reactions oxidation mechanism. Hence, in general, it is
necessary to couple the thermal NO reactions to the
CH3 + H ( + M ) . ' CH,(+M) (A2)
reaction sequence describing oxidation of the fuel.
and Since the overall rate of NO formation by the thermal
mechanism generally is slow compared to the fuel
oxidation reactions, it often is assumed, following the
CH3 + C H 3 ( + M ) ' C2Hr(+M). (AI)
suggestion of Zeldovich, ~9 that the thermal NO
For reaction (AI) we have adopted the expression formation reactions can be decoupled from the fuel
given by Wagner and Wardlaw, '6 and for reaction oxidation process. In this situation, NO formation
(A2) our rate coefficient is essentially the same as that rates are calculated assuming equilibrium values of
of Stewart, Smith, and Golden. 17 For the other temperature and concentrations of 02, N2, O, and
important pressure-dependent dissociation/recom- OH (the N-atom concentration is calculated from a
bination reactions, steady-state approximation applied to reactions
(A232)-(A234)).
H + C2H4(+M ) . " C,,Hs(+M) (A73) The errors that may be introduced by this approxi-
mation are illustrated in Fig. 1. Here, the maximum
and (initial) NO formation rates calculated from the
Zeldovich model are compared with the maximum
H + C2H:(+M) . ' C2H3(+M), (A79) NO formation rates calculated from a detailed
kinetics model for oxidation of methane-air for a
we continue to describe the pressure dependence in fuel-rich mixture (see Appendix A for a complete
terms of the Lindemann form. ~4'~7Such a description tabulation of the reactions and rate parameters used
appears to be adequate for the present purposes. in the detailed calculation). In this comparison, an
isothermal reaction was assumed (to allow indepen-
3. T H E R M A L NO MECHANISM dent variation of temperature) and the prompt NO
initiation reactions, discussed in Section 4, were
Nitric oxide (NO) is an important nitrogen oxide deleted from the mechanism so that the observed
emitted from combustion sources. In the combustion differences in NO formation rates are due entirely to
of clean fuels (fuels not containing nitrogen
compounds), oxidation of atmospheric nitrogen (N:) 6 ,,, ...... I .... I .... I .... 110
J
by the "thermal' mechanism is a major source of NOx
-
and Z F 4
~' F 4
o~ 2 b ~_0.2
N + OH . • NO + H. (A234)
1F . . . . ~ , ,i,,, ,]o,
There have been indirect and direct measurements of 1600 1800 2000 2200 2400 2600
the rate coefficients of these three reactions, and these Temperature - K
data have been critically evaluated by Baulch e t al. ~
and Hanson and Salimian.: The expressions for the FIo. l, The effect of superequilibrium radical concentrations
rate coefficients for reactions (A232)-(A234) used in on NO formation rates in the isothermal reaction of 13%
methane in air (~b = 1.37). The upper curve is the ratio of
the present calculations are given in Appendix A, and the maximum NO f'~rmation rate calculated using the
are the same as those given and discussed by Miller e t reaction mechanism of Appendix A (with prompt NO in-
a l / ° The expression for k.~.,3, is given for the reaction, itiation reactions deleted) to the initial NO formation rate
N + NO ~ O + N.,, and was derived from the calculated using the Zeldovich model. The lower curve is the
ratio of the NO concentration at the time of the maximum
high-temperature data of Ref. 18 for the reverse NO formation rate calculated using the detailed reaction
reaction (using the equilibrium constant) and the low- mechanism to the equilibrium NO concentration,
J~ECS 15:4-B
290 J.A. MILLERand C.T. BOWMAN
nonequilibrium radical concentrations during the convenient to consider the reaction in reverse. By
oxidation process. There is a noticeable acceleration comparing reaction (A153) with BAC-MP4 (bond
of the maximum NO formation rate above that cal- additivity corrected M611er Plesset fourth-order per-
culated using the Zeldovich model during the initial turbation theory) calculations for the isoelectronic
stages of reaction due to nonequilibrium effects, with reaction O + HCN, 3°a~ we conclude that the
the departures from the Zeldovich model results de- addition complex for adding NH to the nitrogen end
creasing with increasing temperature. While none- of HCN lies about 20 kcal/mol above NH + HCN.
quilibrium effects on the NO formation rate are To form CH2 + N, tkiis complex must undergo a 1,3
evident over a wide ~temperature range, the ac- hydrogen shift from the nitrogen to the carbon atom.
celerated rates are sufficiently low so that very little as shown in the diagram. Such hydrogen transfers in
NO is formed by the accelerated reaction. This result open-shell systems, where the hydrogen atom is
is illustrated by the lower curve in Fig. 1, where the accepted at a free valence site, normally have intrinsic
NO concentration at the time of maximum rate of ~nergy barriers (in the exothermic direction) of ap-
formation calculated from the detailed kinetics model proximately 40kcal/mol. This information and tae
is compared with the equilibrium NO concentration. endothermicity of the reaction suggest an activation
energy for reaction (A153) of approximately 85 kcal/
tool, a very high value. In addition, the transition state
4. P R O M P T N O M E C H A N I S M
for the reaction corresponds to that for the 1,3
hydrogen transfer. Such transition states normally are
Nitric oxide formation rates in combustion of
very tight and lead to low A-factors, probably about
hydrocarbon fuels can exceed those attributable to
101Zcm3/mol-sec at high temperature. The rate coef-
direct oxidation of molecular nitrogen by the thermal
ficient expression used in the present calculations,
mechanism discussed above, especially for fuel-rich
kAi53 = 1013exp (-- 7 4 0 0 0 / R T ) cm3/mol-sec, is taken
conditions. This rapidly formed NO was t e r m e d
from Glarborg et al. H and represents a reasonable
'prompt N O ' by Fenimore 2° since the rapid NO
upper limit for the rate coefficient; a value of
formation was confined to regions near the flame
kAm = 1012exp(-85000/RT)cm3/mol-sec appears
zone. Typical levels of prompt NO range from a few
more likely. These values of kA153 appear to be consis-
parts per million by volume to more than 100 ppw.v.
tent with Benson's 32 estimate, using current thermo-
Available data indicate that some prompt NO
chemistry, of an activation energy exceeding 70 kcal/
formation occurs due to superequilibrium concen-
tool and a low A-factor.
trations of O and OH, which accelerate the rates of
The present estimate of kA~3 differs markedly from
reactions (A232)-(A234). However, numerous studies
the similar analysis of Sanders et al. 33 The source of
have shown that prompt NO in hydrocarbon flames is
the discrepancy is the incorrect assumption in the
formed primarily by a reaction sequence that is
latter paper that the four-centered transition state,
initiated by the rapid reaction of hydrocarbon
discussed above, is analogous to the transition state
radicals with molecular nitrogen, leading to
for an oxygen atom adding to the carbon atom in
formation of amines or cyano compounds that subse-
HCN. This assumption leads to an estimate of the
quently react to form N O . 20"29
activation energy that is too low. With the rate coef-
What hydrocarbon radical is responsible for the
ficient given in Ref. 33, reaction (A153) could be an
prompt NO in hydrocarbon flames? A number of
important contributor to prompt NO. However, even
species have been suggested, e.g. CH, CH2, C,., C2 H,
with a rate coefficient much faster than used here,
and C, 2°-29 although the experiments of Blauwens et
al. :6 suggest that CH and CH: are the major contri-
butors. The mechanism used here explicitly considers
3CH 2 ÷ N2 ~ PRODUCTS
the following reactions:
~ 15oi
CHz + N2 .':'" HzCN + N (-A154)
I HCN ÷ NH
and
reaction (A153) is an insignificant contributor to The most direct determination of kA~sL is the shock
prompt NO in the present calculations. tube study of Dean et al. TM in the temperature range
The other reaction channel involving methylene 2700-3700K. As can be seen in Fig. 4, there are
and molecular nitrogen, significant differences among the various determi-
nations of kp,~5~, especially in the temperature range of
CH: + N: . - H2CN + N ( - A 1 5 4 ) interest in combustion. However, most of the deter-
minations should be assigned relatively large uncer-
is endothermic by 79kcal/mol and is isoelectronic tainties.
with the thermal-NO reaction, O + N2 ~- NO + N Theoretical estimates of kAj_~l are complicated by
( - A232). On this basis we have estimated kAj54to be the failure of tiae reaction to conserve electron spin.
essentially the same as kA,.3:. This reaction also is an Figure 3 shows that the intersystem crossing occurs at
insignificant contributor to prompt NO in our cal- the high point along the reaction path and thus plays
culations. the deciding role in determining kAjs~. In order to
A reaction that does lead to prompt NO is (A151), obtain rate parameters for the reaction, we fixed the
activation energy at a value of 13.6kcal/mol, as
CH + N2 . ' " ~ HCN + N, (AI51) inferred from the flame data of Matsui and
Nomaguchi, 2~ and adjusted the A-factor to fit the
originally suggested by Fenimore. '° A reaction coor- well-stirred reactor data of Bartok et al. 36 discussed
dinate diagram for reaction (AI51) is shown in Fig. 3. below. The reaction coordinate diagram of Fig. 3 has
The first step in the process is the formation ofmetast- been constructed accordingly.
able HCN2*. According to the ab initio electronic The A-factor deduced for reaction (AI51),
structure calculations of Bair, 34 this addition process 3 × 10 j~cm3/mol-sec, is very small, approximately a
has no energy barrier and results in a bond of factor of 100 smaller than one might anticipate for a
18.3kcal/mol. Berman and Lin :5 have measured the similar reaction that conserved electron spin. This
rate coefficient for the pressure-dependent reaction, result implies that the intersystem crossing actually
occurs only 1% of the time that the critical configur-
CH + N : ( + M ) . ' H C N z ( + M ) ,
ation is obtained. It should be noted, however, that
the present determination of kAjsj is not very sensitive
in the temperature range 297-675 K. As expected for
to the activation energy and A-factor independently,
such reactions, the rate coefficient for this reaction at
but only to the value of kA~5~in a narrow temperature
any given pressure decreases rapidly with increasing
range centered at about 2000K. The kA~5~ which we
temperature, and under most combustion conditions
have inferred is shown in Fig. 4 and in Table 2. Our
the pressure-independent channel,
kAzsj, which is a factor of 1.5 larger than Glarborg et
CH + N, . • HCN + N, (A151) a l . f is smaller than most of the other determinations
at 2000K. However, our kAl~ value at 2000K is
is dominant. consistent with the high-temperature data of Dean et
There have been several indirect inferences of kAlsl al. ~TJ if an activation energy of 26 kcal/mol is assumed
from flame, shock tube and stirred reactor data and for the reaction. The present value of kAtsl accurately
one direct shock tube measurement. Available rate predicts the NO produced in the stirred-reactor ex-
data for reaction (A151) are summarized in Fig. 4. periments of Duterque eta/., j79 as well as the HCN
CH + N 2 ~ HCN + N
1ooi
14o!oHI2,,,
120~
H N~N
, \/
C
REACTION COORDINATE
g O
model exper,ment
..'.
•" ". CH xlO 4 O--CH xl0 =
(x ~ Temperature
. '. ..~229o~. n-c.2 ,.oS
~ <,,.o"w°'£',g,, Z
0
O o.
Z
Cl -
g
I CH2Ox!O=
.,:': O
';.
b
og
~c
t,
2
JS
~J
0
I-1
0 NOxl0 I"
lexDerlmentl
OxlO
(model)
[ ] (experPment)
g
9
§
Eo-
o
g.
o
/./~,
0.2 04 0.6 08
o
C' X (cm)
2 3
X Icm)
F]G. 6. Companson of model predictions using the mechan-
FIG. 5. Comparison of model predictions using the mechan- ism of Appendix A with the 'flame I' data of Blauwens et al.26
ism of Appendix A with the "flame I' data of Blauwens et al? 6 for CH2, CH, and C.
for O and NO. The assumed temperature profile and experi- '
mental data are also shown. The two different NO predic-
tions correspond to two different values of k^m ((A151)---- Figure 7 shows a comparison of the present model
CH + N:~ HCN + N), that of the present kinetic model predictions with the well-stirred reactor experiments
and that deduced by Blauwens et al. of Bartok e t al. 36 Following Glarborg, 4° we have made
the comparison for the 3 msec residence time data.
explanation for the discrepancies between calculated This is the only residence time for which temperatures
and measured NO profiles for distances less than I cm were measured at equivalence ratios richer than ~toic-
is the presence of formaldehyde, which could con- hiometric. Use of the experimental temperatures in
tribute to the m / e = 30 peak in the mass spectro- the calculation eliminates an important uncertainty in
meter. The calculations. Fig. 5, indicate the presence the calculation.
of formaldehyde at distances up to 0.7 cm above the Several important results are readily apparent from
burner surface, although Blauwens e t al. imply that Fig. 7. First, thermal NO, computed by deleting from
interference due to formaldehyde was negligible the complete mechanism the prompt-NO initiation
downstream of the visible reaction zone ( > 0.5cm). reactions discussed above, is insufficient to account
In summary, the present kinetics model is able to for the experimentally observed NO above an equiva-
match all of the reported species profiles, including lence ratio of approximately 4~ = 0.9. At an equiv-
CH. for 'flame I" of Ref. 26 except for NO. It was not alence ratio of q5 = 1.2, hydrogen cyanide begins to
possible to match the entire NO profile in this flame make a significant contribution to the total fixed
using either the present value of kA~s~ or the value nitrogen (TFN = NO + HCN + NH,) in the
reported in Ref. 26. Hence, we assign a relatively high reactor. The HCN concentration peaks at (or near)
uncertainty to the kAlS] determination of Ref. 26. q$ = 1.4 and drops off quickly beyond that. The cal-
Clearly. further progress in modeling prompt NO in culations predict the experimentally observed NO up
flames requires a direct and unambiguous determi- to ~b = 1.4, but at higher equivalence ratios the model
nation of kA]<~ at flame temperatures. predicts smaller NO concentrations than are observed
The last prompt-NO reaction considered explicitly experimentally•
in the mechanism is
175
t 3 0 m l l l c ~nlodell
C + N, . " CN + N. (-A152) 0 3 0 m s e c (experiment)
150 - • T h e rr~8 I NO (model)
. . . . . HCN (mociel) . .
~00
CN + N ---, N, + C
75
Z
is very fast both at room temperature 37 and under
shock tube conditions:~83QDean e t a l ) 73 have reported 5o-
Interpretation of the computational results is (3) the rates of interconversion among fixed nitrogen
facilitated by the use of sensitivity analysis and rate- fragments.
of-production (or reaction path) analyses. Sensitivity As can be seen from the sensitivity analysis,
coefficients for NO, HCN, and TFN, corresponding reaction (A151) is an important rate-limiting step for
to Fig. 7 are shown in Figs 8-10. The most important NO, HCN, and TFN production under all conditions
reaction paths in prompt-NO formation and fuel investigated; the reaction has sensitivity coefficients
nitrogen conversion, which have many common for these species approaching unity in rich mixtures,
features, are shown diagramatically in Fig. 11. and even under lean conditions (~b < 0.8) it plays a
Prompt-NO formation involves three separate major role in NO formation. The thermal-NO
kinetic issues: mechanism is the dominant source of NO only in the
(1) the CH concentration and how it is established; equivalence-ratio range ~b = 0.8-1.0. For ¢ < 0.8,
(2) the rate of molecular nitrogen fixation: i.e. value the temperature in the reactor is sufficiently low that
of kk~sl; the sequence,
and
A9 CH3+O=CH20 +H A9 CH3+O=CH20+H
it
A,-~T AIO CH2OH+H--CH3+OH AIO CH2OH+H--CH3+OH
A161 A12 CH3+OH--CH2+H20 A12 CH3+OH--CH2+H20
A15 CH2OH+MICH20+H + M A15 CH2OH+M=CH20+H+M
A24 CH2+HICH+H2 oo A24 CH2+HICH+H2
A25 CH2+OH--CH+H20 A25 CH2÷OH--CH+H20
A26 CH2+OHICH20÷H A26 CH2+OH~CH20+H
A27 CH+O2--HCO+O A32 CH+H20=CH20+H
o A32 CH+H20=CH20+H A132 O+OH=O2+H
A39 C+OHICO+H A151 CH+N2~HCN+N
A132 O+OH--O2+H A202 NH+NO=N20+H
A151 CH+N2IHCN+N A204 N204 H--N2"+'OH
O A156 C+NO=CN+O A205 N20+M--N2+O+M
>- A176 HCN+OINCO+H >- A232 N+NO=N2+O
I-- I,--
p-O I--
Z Z
IJ.J LLJ
(,,3O
Z Z °
LL
~-~ t--
AIo AI,
o,
O o,.
05 075 1 125 15 175 0.5 075 1 125 15 175
EQUIVALENCERATIO EQUIVALENCERATIO
FIG. 9. Sensitivity plot of selected reactions for HCN for the FIG.10. Sensitivity plot of selected reactions for TFN for the
conditions of Fig. 7. conditions of Fig. 7.
Nitrogen chemistry in combustion 295
NH 3
HOCN +H ~" H N C O . ~v'
FN,~
HC N2
CH + N2
FIG. 11. Reaction path diagram illustrating the major steps in prompt NO formation and conversion of
fuel nitrogen (FN) to NO. The bold lines represent the most important reaction paths.
In the present calculations there is the additional CH: Beyond ~b = 1.2, several factors combine to cause the
source involving the singlet species, NO concentration to decrease relative to HCN:
(a) the conversion o f H C N to NO by the above
CH 3 + H . ~ ICHz + H, (-All0) mechanism is no longer rapid:
(b) the 'recycle' of NO to HCN by the mechanism
~CH2 + M • CH2 + M, (AI06) discussed below begins to inhibit NO production;
(c) the reaction (A232), N + N O , " N : + O,
which is more important in the calculations than the shifts directions from reverse to forward.
direct abstraction (A13). Only for ~b I> 1.5 does the A decrease in the O-atom concentration contributes
indirect route to CH., through the C: hydrocarbons to all three of these occurrences. But, in particular,
come into play. under rich conditions, reactions that convert CH3 to
The most important CH-consuming steps in the CH: are better able to compete with reaction (A9),
calculations are the reactions CH3 + O ~ C H 2 0 + H, and allow the recycle
reactions,
CH + H _ ..~ C + H:, (A31)
C + NO . ' CN + O (A156)
CH + HzO . • CH:O + H, (A32)
CH + NO . • HCN + O (A157)
and
CH 2 + NO _ ' HCNO + H (A158)
CH + CO., . " HCO + CO. (A30)
to become more effective.
Reaction (A32) was not included in the previous cal-
At ~ = 1.4 the peaks in the CH and T F N con-
culations of Glarborg e t al. H The present rate coeffi-
centrations occur simultaneously. Beyond tp = 1.4
cient expression for k~3-. is based on the experiments
one enters a regime in which the fixed nitrogen con-
of Zabarnick e t al. 43 in the temperature range
centration is limited by the availability of the chain
297K < T < 670K. However, a T" function was
carriers (H and OH) required for producing CH from
used to extrapolate their results to high temperature,
methane, as evidenced by the increased sensitivity of
rather than the Arrhenius expression given in Ref. 43.
the T F N to chain branching processes, i.e.
Up to an equivalence ratio of approximately
¢ = 1,2 the HCN and N produced by reaction
(A151) are converted rapidly to NO, principally by H + O: . ' OH + O (-A132)
the mechanism discussed by Miller e t a l . , ~°
OH + CH3 ~ • CH2OH + H ( - A l 0 )
HCN + O = NCO + H (A176)
CH:OH + M , • CH:O + H. (A15)
NCO + H -~. NH + CO (A192)
This regime is also characterized by the appearance of
NH + H ~ " N + Hz (A211) reaction (A 168),
296 J.A. MILLERand C.T. BOWMAN
CH 3 + N . " H2CN + H, (A168) In Fig. 13, sensitivity coefficients for NO are shown
for 1800K and 2500K. At low temperatures
as a competitor for nitrogen atoms, competing (T < 2000 K), the rate of NO formation is dominated
favorably with reaction (A234), by the hydrocarbon-nitrogen prompt-NO channel.
As the temperature increases, the relative importance
N + OH . " NO + H. (A234) of the hydrocarbon-nitrogen prompt-NO channel
decreases, so that for temperatures above 2500 K, NO
The dissociation of H2CN following reaction (A168) formation is controlled mainly by the thermal
results in the formation of hydrogen cyanide, rather mechanism. The contribution of the hydrocarbon-
than nitric oxide, from the N atom. nitrogen channel to NO formation may be quantified
Since the measured NO levels for ~ > 1.6 exceed by comparing the calculated NO concentration at the
the predicted TFN concentrations, it seems unlikely
that the discrepancy between the measured and (a)
predicted NO concentrations under very rich ~o
A1 2CH3(+M)--C2H6{+M)
A3 CH4+O2-CH3+HO2
conditions is due to failure of the parts of the A4 CH4+H--CH3+H2
A8 CH3+O2--CH30+O
mechanism that result in interconversion of HCN and A9 CH3+O--CH20+N
A14 CH30+M~CH20+H+~
NO. More likely, the discrepancy is due either to AS4 CH20+O--HCO+OH
A56 HCO+M"H+CO+M
experimental error, to an additional source of CH or A60 HCO+O2--HO2+CO
A66 C2H6+H--Cg'H5+H2
an additional nitrogen fixing reaction for ~b > 1.4, or A74 C2H5+ H--2CH3
A132 O+OH--O2+H
to an inadequate description of chain-branching/ A133 O+H2'~H+H
A151 CH+N2mHCN+N
chain-terminating processes in this regime.
To illustrate the relative contributions of the
thermal and hydrocarbon-nitrogen mechanisms to
NO formation, we have performed a series of calcu- L~
A4 CH4+H-,CH3+H2
20 ,,,.i,,,~[ .... ~ .... t .... 1.50 A8 CH3+O2--CH30+O
A9 CH3+O-CH20+H
¢~ A25 CH2+OH,-CH+H20
O A26 CH2+OH--CH20+H /
1.25 A27 Ctq+O2--HCO+O ~ /
A31 CH+H-C+H2 ~x /
~i" A38 C+02-CO+O / \ /
15 - A39 C+OHmCO+H / \ i
A43 CH2+O-CO+2H I \ /
1.00 A44 CH2+O--CO+H2 I \ /
o t~ A46 CH2+O2--CH20+O I \ /
Z
O - A57 HCO+H--CO+H2 I \
A74 C2HS-i-H--2CH3 / ~x/
~I0 0.75 z A132 O+OH-O2+H I ~ / ~
A133 O+H2-OH+H ~ / /
o ~ A151
CJ- CH+N2-HCN+N ~ t ~ -
z : AlS2 CN+N~C+N2 II I - J ~
Z 0.50 O - ; A232 N+NO--N2+O ~
5
0.25
Z
time of maximum NO formation rate with equi- fraction is about a factor of 3.4 higher than the
librium NO concentrations (Fig. 12). At tow tem- adiabatic equilibrium value.
peratures, this channel produces NO levels in excess The mechanism of CH formation in the flame cal-
of equilibrium levels, while, at temperatures above culations is very similar to that in the well-stirred
2500K, this channel contributes only slightly to reactor calculations. The principal difference is that.
overall NO formation. for any equivalence ratio, a somewhat greater portion
The temperature, O-atom, NO, and H C N flame of the CH comes from CH2 that has been formed via
profiles calculated for the same methane/air mixture the C2 hydrocarbon route in the flame than in the
as in the flow-reactor calculations, are shown in Fig. well-stirred reactor. Figure 15 shows the NO sensi-
14. Under these conditions, all the NO formed is from tivity coefficients for the flame of Fig. 14. Large
the hydrocarbon-nitrogen mechanism. Even though positive sensitivities occur for reaction (AI51),
the O-atom concentration exceeds the final equi- CH + N2 ~ HCN + N, and for reactions that
librium value by a factor of 700, the temperatures in produce CH from CH2. Large negative sensitivities
the flame where the superequilibrium O-atoms are arise for reactions that channel methyl through non-
achieved are so low that the thermal mechanism is CH-producing sequences and from the reaction of
ineffective in producing prompt NO. This appears to CH with water,
be a general observation for flames--superequili-
brium O-atom concentrations are not a very CH + H20 . ' CH20 + H, (A32)
important source of prompt NO inflames because of
the low temperatures at which they occur. which is an important sink for CH in the flame.
In Fig. 14 the peak HCN mole fraction in the flame To illustrate the transition from the prompt (hydro-
is more than half of the value of the final NO mole carbon-nitrogen) NO mechanism to the thermal
fraction farther downstream (approximately 15 ppm mechanism in premixed flames as conditions become
HCN compared to 27 ppm NO). This occurs because leaner, NO profiles are plotted in Fig. 16 for a series
HCN is produced both directly by reaction (AI51), of methane/air flames. The conditions shown range
CH + N, --* HCN + N, and indirectly by reaction from 13% methane ( 4 ) = 1.37) to 10% methane
of nitrogen atoms with methyl radicals, (4) = 1.06). Over this range of equivalence ratios, the
NO formation mechanism changes from completely
N + CH3 . ' H:CN + H (A168) hydrocarbon-nitrogen to predominantly thermal.
The origin of the term 'prompt NO' is most obvious
HzCN + M . • HCN + H + M . ( A I 5 5 ) in the 11% methane case, where there clearly are two
time (or length) scales for NO formation. Approxi-
Only when the methyl concentration drops off suf- mately 10 ppm are formed very quickly in the reaction
ficiently is the nitrogen atom oxidized directly to NO. zone: the remainder is formed much more slowly
Consequently, most of the NO formed comes from an downstream. An examination of the NO sensitivity
HCN intermediate. Under relatively rich conditions, plot for this flame in Fig. 17 shows that the prompt
the HCN --* NO conversion is not very rapid and the NO formation arises from the CH + N2 --* HCN +
resulting NO is not very 'prompt,' even though the N reaction, whereas the slower increase in NO later in
hydrocarbon-nitrogen mechanism is the dominant
source of NO. It is also worth noting that the rela- A 9 CH3+O=CH20 +H
A 24 CH2+H=CH+H2
tively constant downstream value of the NO mole 25 CH2+OH=CH+H20
26 CH2+OH=CH20 +H
o A 31 CH+H~C+H2
A151 A 32 CH+H20=CH20 +H
/~- A 46 CH2+O2=CH20+O
A 132 O+OH=O2+H
/ A 151 CH+N2=HCN+N
f
/
/
f
S;
- >- / L
// 0,1o3 NOxlO4 ~ -
A24
, I / ~ )'~
'I ,I ' J =
Z ~ ~ A 1 3 2
,/ "Ag" ~o,
C, ©~ 02 03 04 05
X(CM)
OO 01 0.2 0.3 04 05
FIG. 14. Calculated temperature. O-atom, NO and HCN
X(CM)
profiles for an atmospheric-pressure flame of 13% methane
in air (~ = 1.37). FIG. 15. Sensitivity plot for NO for the flame of Fig. 14.
JPECS ]5:4-C
298 J,A. MILLERand C.T. BOWMAN
o _ L
~,15~ , A~I5 ~ 9 CH3+O=CH20+H
/7]-~, 24 CH2+H--CH+H2
// | A 25 CH2+OH=CH+H20
125%CH4 ....._i..j¢.--.
~ ~
/ I A 27 CH+O2=HCO+O
- - - - -
~ J A 31 CH+H=C+H2
/ / A 32 CH+H20=CH20+H
~ J A 46 CH2+O2=CH20+O
,_ _ 12%C ~ l A 56 HCO+M=H+CO+M
- ~ I A 132 O+OH=O2+H
-_ A232 -J~, 151 CH÷N2=HCN+N
232N+NO=N2+O
Z 0
o o~
0.0 01 02 0.3 0.4 05 Z
X(CM)
FIG. 16. Nitric oxide mole fraction profiles as a function A32
of equivalence ratio for atmospheric pressure flames of
methane in air. A132
o H* _ C N ~ ÷O N2
, o.,+o//" I
NOx10~
HCN" *0 - - N H .a~H N .N
NCO NO
0 05 : 15 2 25 0 A 131 OH+H2=H20+H
I
I
X{CM) A 132 O+OH=O2+H ]
A 133 O+H2--0H+H i
o
g- A 134H+O2(+M)--HO2{+M) ~__~...-~---...~z~3J~q3
A 176 HCN+O=NCO+H ~ L~q~
FIG. 18. Comparison of model predictions with the ex- A 177 HCN+O=NH+CO ~ l
perimental data of Miller et al.~° for a lean (th = 0.5) H2/O~/ A 182 CN+OH=NCO V
Ar flame (P = 25 torr) to which 1% HCN was added.
A3:
H ---
o ,
i o_
Z
!=< 9
u,..
u~
=, HCNxI0 •
o IAITe
o ! 1
0.5 1.5
X(CM)
0 0.5 1 15 2 FIG. 22. Sensitivity plot for HCN in the rich flame of Fig. 20.
X(CM)
i
©
=c=,,o\ /- =o,, z
i o o o ~
Q-
9-
C 05 !.5 2 25
X(CM)
O
0 0.5 . i 1'5
X(CM)
Flo. 20. Comparison of model predictions with the ex-
perimental data of Miller et aL L° for a rich (~ = 1.5) H 2/O:/
Ar flame (P = 25 tort) to which 2% HCN was added. FIG. 23. Sensitivity plot for NO in the rich flame of Fig. 20.
300 J.A. MILLERand C.T. BOWMAN
and NH with hydrogen atoms, producing N, are rela- based on BAC-MP4 calculations for the two
tively fast and have little influence on the observed processes are shown in Figs 24 and 25. Miller e t ' a l . 3~
species profiles. The distribution of NO and N: in the and Miller and Melius :3 have used the BAC-MP4
flame is governed by the N-atom reactions, results and statistical-theoretical methods, modeling
the intramolecular dynamics as stochastic processes.
N + OH . • NO + H (A234) to predict rate coefficients for the various elementary
steps of interest. For O + HCN the steps are
N + NO . " N: + O. (A232) reactions (A176) and (A177); for OH + HCN three
channels are possible,
In studying hydrogen cyanide decay in the
postflame gases of very rich atmospheric-pressure OH + HCN . • HOCN + H (A172)
flames, Haynes, 49 Morley 5°'5i and Fenimore 5:
concluded that H C N is removed primarily by reaction OH + HCN . • HNCO + H (A173)
with OH. Two channels were suggested by these in-
vestigators, one that is first order in OH, and
and one that is second order, For O + H C N a comparison of the total rate
coefficient predicted by the theory with the ex-
OH + H C N . • CN + H20 (A171) perimental results of Davies and Thrush, 54 Perry and
Melius, 3° and Roth e t al. 55 is shown in Fig. 26. The
CN + OH . • NCO + H, (A182) agreement is excellent, particularly when a minor
modification is made to the barrier height for adding
under the partially equilibrated conditions prevalent
in the postflame gases of atmospheric-pressure flames.
Reaction of HCN + OH
Because of the equilibration of the H C N --* CN con-
version, the rate of the second-order process is
HCNOH #3 #5 #2
governed by reaction (A182). Haynes concluded that
the first of these channels was dominant below -~
,¢ 6o N ~ HOCN + H
2300 K, the latter above 2300 K. It is of interest to '~ 40 CN + H20
note that the O + H C N reactions, (A176) and I HC NCO + H2
(A177), are also second order in the OH concen- HNCO + H
tration under these conditions because of the relation-
ship between O and OH in partial equilibrium. wo (.)
Several theoretical and experimental investiga-
0 H
tions 3°'31'53"55-58have helped to reconcile the seemingly "CN ~H
disparate conclusions of Miller e t al. and those of -2o - ('")
Haynes, Morley, and Fenimore. The reactions of Reaction Coordinate Diagram
interest, the first-order O + H C N and the
FIG. 25. Reaction coordinate diagram from BAC-MP4 cal-
OH + H C N reactions, are complicated processes culations for the reaction OH + H C N ---} products.
involving multiple potential wells and multiple
product channels. Reaction coordinate diagrams T (K)
2000 1000 750 SO0
10"~o
140 ,
Reaction of HCN + O O+HCN'-~ productl
10""
1
Oav~es-Thrush
O
120
# # 6 ~
I HCO+N
CN + OH
"~_~ 1°""1
o i0-,3~ °:!iT;i
E loo
10"~ \
$0 . = NH ÷ CO 10-,61
0.0 Eo ~Lo ,~o 2o.o 25.0
40 lIT "I0"'
Reaction Coardlnmte Diagram
FIG. 26. Comparison with experiment of the theoretical
FIG. 24. Reaction coordinate diagram from BAC-MP4 cal- prediction of Miller et al. 31 for the total rate coefficient,
culations for the reaction O + HCN ---,products. kAi76 + kAi77, of the reaction HCN + O --, products.
Nitrogen chemistry in combustion 301
O to HCN. 31 F o r the specific channels, the theory normal combustion conditions. In contrast, however,
accurately predicts the experimental results of Louge reaction (AI71) is relatively fast at combustion tem-
and Hanson 56 for kAi76and those of Szekely et al. 57 for peratures. Figure 27 shows a comparison of the
kay77. The rate coefficient expressions in the theoretical prediction for k^~7~ of Miller and Melius
mechanism are consistent with both the theoretical with the experimental expression given by Szekely et
and experimental results. al. 5s (used in the present mechanism), which is based
For the OH + HCN reaction there is very little on their experimental data for the reverse reaction at
experimental data available. In fact, the only relevant high temperature (2460 K < T < 2840 K) and the
data are from the flame experiments discussed above. lower-temperature (500K < T < 700K) data for
Figure 27 compares the theoretical prediction for the the forward reaction of Fritz et al. 59 Although the
total rate coefficient, k T = kA]72 -F kA173 + kA174, agreement between theory and experiment is not as
with the experimental results, which are sensitive only good as one might hope, both indicate that the ab-
to this sum. The agreement between theory and ex- straction is the fastest of the OH + HCN channels.
periment is remarkably good. The rate expressions in This result limits the conditions under which the other
the present model are based on the theoretical results, OH + H C N steps, reactions (AI72)-(A174), can be
which show kA172 > kA173 > kA174 under most com- effective in removing H C N to those where the ab-
bustion conditions. straction reaction (A171) is nearly equilibrated.
For kA~s: (CN + OH ~ NCO + H (A182)), The present kinetics model incorporates the ex-
which controls the rate of the process that is second perimental observations of Miller et al. and those of
order in OH, we have adopted the Haynes49/Morley 5° Haynes, Morley, and Fenimore about H C N removal.
rate coefficient without modification. However. it is What conditions favor the different mechanisms? Ex-
possible that their determination of this rate coef- perience in modeling a variety of flame, stirred
ficient was hampered to some extent by interference reactor, and flow reactor conditions indicates that the
from the O + HCN reaction, which would also be O + H C N sequence shown in Fig. 21 almost always
second order in OH under their experimental plays the major role. The OH + HCN reactions
conditions. normally come into play only under conditions that
Both the O + HCN and the OH + HCN are both rich and nearly equilibrated. If HCN conver-
reactions have direct hydrogen abstraction channels sion takes place in a highly nonequilibrium reaction
that should be considered, zone, even under very rich conditions, the O + HCN
sequence still is likely to be dominant.
0 + HCN . " CN + OH (A178)
1
pressure premixed flame experiments of Thorne et
al.12 In these experiments, the flames are nominally the
rich, ¢ = 1.5, H:/O2/Ar flame of Miller et al., ~° to
which small quantities of acetylene, hydrogen
10-,'1 i cyanide, and nitric oxide were added in various com-
oo 50 lo.o ls.o 20.0 25.0
1/T "10" binations. Acetylene was used as the hydrocarbon
because of the ease with which it yields low molecular
FIG. 27. Comparison with experiment of the theoretical weight hydrocarbon free radicals through the
prediction of Miller and Melius $3 for the total rate coef- O + C : H : reaction, z:
ficient, kr = kAir: + k~l~3 + kAl,.4, of the reaction Figure 28 shows a comparison of the model predic-
OH + HCN --, products. Also shown are predictions and
tions with the experimental data of Thorne et al. for
experiments for reaction (A171), OH + HCN ---,
CN + H,O. the nitrogen-containing species in the flame to which
302 J.A. MILLERand C.T. BOWMAN
e~
CN
o
×
° /
,..1
_ HCNx102
o
N~xI02 Z
• m (.,,)
HCN+OH==CN+H20
o-
/ ...C.N..+.H..2..--H.CN+ H......
HCN+O~CN+OH
o ~ ~ c~ ~
o
0~ oB 08 i ,2 ;.* ~'a ;. 0.5 1 1.5
X(CM) X(CM)
FIG. 29. Degree of equilibration of the reactions F]o. 31. Comparison of model predictions with the experi-
HCN + Y ~ C N + HY, where Yis H, OH, or O. The ments ofThorne et aL i 2 for a rich (@ = 1.5) H2/O2/Ar flame
function f,(x) is defined in Eq. (5.1). at 25torr to which 2.9% C2H2 and 1.4% NO were added.
Nitrogen chemistry in combustion 303
were converted to number densities using the calibra- HCN a substantial fraction of the time (more than
tion factor determined above. Again the agreement half). The sensitivity analysis shows that the CN is
.between model prediction and experiment is excellent. produced and destroyed in this region of the flame by
Both show the initial conversion of the NO to HCN the sequence
and CN and the subsequent formation of N2 on a
slower time scale. C + NO , " CN + O (A156)
Sensitivity profiles for NO, CN and HCN in the
flame of Fig. 30 are shown in Figs 32-34; sensitivity CN + H 2 . " HCN + H. (A179)
profiles for the other flame are similar. The most
surprising feature of these flames is the sharp peak in The main reaction paths involved in converting NO
the CN number density at about 4 m m above the to N: through HCN and CN are shown in Fig. 35.
burner. This peak is reproduced by the model only if The improvement in the agreement between the
the NO conversion results in CN as a precursor to model calculations and the experimental data, par-
ticularly in the CN profile of Fig. 31, over that
reported by Thorne e t al. is due to the increased
A g CH3+O=CH20+H
importance in the present model of the
A 24 CH2+H=CH+H2 C 2 H : ~ H C C O path, followed by the sequence,
A 27 CH+O2=HCO+O
A 38 C+02=CO+O
r~ A 76 C2H2+O=CH2+CO
o A 7"7 C2H2+O=HCCO+H
A 79 H+C2H2(+M)=C2H3(+M)
A 110 CH2(S)+H2--CH3+H t~
A 132 O+OH=O2+H
A 1330+H2=OH+H
A 156 C+NO--CN+O A 27 CH+O2=HCO+O
o A 157 CH+NO=HCN+O A 38 C+O2,=CO+0
I>- A 226N+NO=N2+O A 132 O+OH--O2+H
F-- A 228N+OH=NO+H A 133 O+H2=OH+H
__E A 156 C+NO-CN+O
A 157 CH+NO-'HCN+O
A 176 HCN+OmNCO+H
A 179 CN+H2"=HCN+H
iii A 181 CN+O2=NCO+O
u3
©
z A~B
¢.+ AI~-
o A,~7
O z( , . ) ~ A157
"3" o - A~e
0 05 1 15 2
X(CM)
FIG. 32. Sensitivity plot for NO for the flame of Fig. 30.
A 24. CH2+H=CH+H2 0 05 1 !5 2
A 31 CH+H=C+H2 X(CM)
A 38 C+O2=C0+0
co A 76 C2H2+O=CH2+CO
o- A 79 H+C2H2(+M)=C2H3(+M)
f '~ A 1320+OH--O2+H
FIG. 34. Sensitivity plot of selected reactions for HCN for the
J ~ A 1330+H2=OH+H flame of Fig. 30.
/ ~ A 156C+NO=CN+O
o- ] ~ A 179 CN+H2=HCN+H
/ -~- A 181 CN+O2=NCO~O
+O
C2H 2 m= H C C O
A~61 , L--
.+M,+/cH2
At=3
zO-
CO 3-C1"12~" N+" ' ~ ' ~ ' ~ HCNO ÷H
ew +No~ / pH
o- CH v HON~..-+O /
, " .co
,
C +NO v"- O N
r>.,+, +O 2
,A+I+
O
05 1 1.5 2 FIG. 35. Reaction path diagram illustrating the
X(CM) NO ---, HCN --, N 2 conversion mechanism in the NO-added
flames of Thorne et al. t: The bold lines indicate the most
FIG. 33. Sensitivity plot for CN for the flame of Fig. 30. important reaction paths.
304 J.A. MILLERand C.T. BOWMAN
H + HCCO . ~ tCH2 + CO (AI01) possibilities for the early CN formation in Fig. 30,
C + NO . • CN + O (A156)
tCH, + H: . • CH3 + H. (All0)
and
Reaction ( A l l 0 ) is very fast 66 and inhibits the
formation of CH and C (and thus CN) early in the CH + NO . ' CN + OH.
flame. In addition to the sequence involving carbon
atoms, HCN is produced directly by the reaction, although the model includes only reaction (A156).
The temperatures and pressure in the flame experi-
CH + NO . • HCN + O. (A157) ments ofThorne et al. are lower than are encountered
in typical reburning situations, and it is possible that
Hydrogen cyanide is partially converted to N: in the reaction channels other than those discussed above
flame by the reaction sequence discussed above, play a role in NO removal by reburning. Roby and
Bowman ~74have shown that in atmospheric-pressure.
O + HCN , ~ NCO + H (A176) fuel-rich H2/O:/Ar flames to which small amounts of
nitric oxide were added, significant removal of the NO
NCO + H .~, • NH + CO (A192) ( 2 40%) occurs by the reaction sequence,
o
and
00 C' I C' i' 0 3 04 O5
X. (cm)
NNH ~ N, + H (B43/A221)*
FIG. 37. Calculated temperature, NO and HCN profiles for
an atmospheric-pressure, 13% methane in air flame or
(~ = 1.37) to which 1000ppmv NO was added.
N N H + NO ~ N: + HNO.
o A156 C'~N0-CN+O
~34~6,157 CH+NO=HCN+O
(B44/A222)
- --A158 CH2+NO'HCNO+ N
A161 HCCO+NOIHCNO+CO
o A228 HNO+M'H"NO'~ M This reaction sequence, involving N2H ~ species,
o A232 N+NO--N2+O
A234 N+OH=NO+H occurs even in moderately lean ammonia flames.
However, its importance appears to be limited to
>,-
I--o
8
o
/ flames for which elemental nitrogen is a major
component of the fuel. Consequently, in most
practical situations, where the fuel contains low levels
of nitrogen, NH, + NHj (i, j = 0, 1, 2) reactions are
z
negligible.
U,J
U30 Jl The important features of the ammonia oxidation
o c mechanism, Fig. 39, can be summarized as follows.
z
o
Ammonia is converted to NH2 by hydrogen abstrac-
0 tion. For most conditions, the primary abstraction
reaction is
o
NH 3 + OH ~ NH: + H20.
o
o (B40/A218)
O0 01 0.2 0.3 04 05
X (cm) However, for lean conditions, reaction with oxygen
atoms and for rich conditions, reaction with hydrogen
FIG. 38. Sensitivity plot for NO for the flame of Fig. 37.
(B50/A228)
N + 02 . • NO + O (B57/A233)
HNO + OH . • NO + H20
N + OH • NO + H. (B58/A234)
(a51/A229)
Since the appearance of Ref. 15, several improve-
HNO + NH2 " NO + NH3. ments in the kinetics model have been brought about
by more reliable determinations of important rate
(B53/A230) parameters, and some rate-parameter estimates and
The formation of N., from NO by reaction with NH2 conjectures have been placed on firmer ground. Most
occurs either directly, notably, the rate coefficients of the reactions of
ammonia with O H , 73-75 H, 76-78 and 0 73'79.80 have now
NH, + NO . " N 2 + H20, (B39) been determined reliably over broad temperature
ranges. Hack, Kurzke, and Wagner81have determined
or through the NNH intermediate, the rate coefficient of reaction (B22)
NH2 + NO . • N N H + OH NH + 02 . • NO + OH
(B38/A216) (B22/A201)
Nitrogen chemistry in combustion 307
up to T = 543 K and have confirmed the products a factor of two between calculated and reported
inferred theoretically from BAC-MP4 calculations,g2 hydrogen atom concentrations: however, uncertain-
This same group 8~ has also measured the rate coeffi- ties in the measurement of hydrogen atoms by
cient of reaction (B61), molecular beam mass spectrometry are of this order.
The model calculations and the experiment agree
NH, + NH . ' N:H., + H, (B61)
£
at room temperature. Their k-value is consistent with o
:t
the estimate made theoretically by Miller e t al. ]5"69 and
values subsequently inferred at high temperature
from shock tube 7-~and flame 71 experiments.
z~ o
To demonstrate the predictive capability of the
-- o oC~bo
ammonia oxidation mechanism of Appendix B, we
have made comparisons of model predictions with the ~_ o
o o .o.e " " ".;
results of the low-pressure flame experiment reported No . ...,o> ° A.i...x, ° o~
recently by Bian, Vandooren, and Van Tiggelen.7° Ooj ,~ /" \ ~ t
0 ~ •
Comparisons of model predictions with other
ammonia flame data were reported previously in Ref.
o
o
;
15. ....... oo°°ooo
The comparisons between the model predictions o 02 04 0.6 08
and the experimental results are shown in Figs 40--44. X(CM)
Overall, the agreement between model and experi-
FIG. 42. Comparison of profiles of NH, H, and O calculated
ment is good. There is a discrepancy of approximately using the mechanism of Appendix B with the data of Bian
et al. 7°
f-f >.,o
c
(")
o
o e/,
q ~ i ' " "i," ~ o,
Z
__Q®
bo ° -
<
/ % ~ N H 7 ~ 10
~c ,. \ t,, ¢
- \0 ~-,a ~ m ~ t3 a
\ ~ D r" L-"e--
o
:~os -
0 x 10
Z
~.--o
~c
< 5 k,-
: ........ :......:...:...'Z...7. ..... .2,,=
--e
S / • .... C' O
iii
S f
c
c
c
o
, . !"
02
•
•
04
o o~
06
O
0
08
Q HN
ok
X(CM) 0 0.2 04 06 D8
X(CM)
FIG. 41. Comparison of profiles of NO and N,O calculated
using the mechanism of Appendix B with the data of Bian FIG 44 Comparison of the HNO profiles calculated using
et al. 7° the mechanism of Appendix B with the data of Bian et a l 7°
308 J.A. MILLERand C.T. BOWMAN
somewhat better for O and OH. Such differences are NH: + O . • HNO + H,
probably within the experimental uncertainty for
molecular beam/mass spectrometric measurements of (B34/A212)
these free radicals. The agreement of model and ex-
and the N-atom reactions themselves.
periment for the major stable species (NH3, O:, N,,
and H20) and for the oxides of nitrogen (NO and
N + O2 . " NO + O
N20) is almost perfect. The amine free radical con-
centrations (NH2 and NH) also are predicted very (B56/A233)
well by the model; a discrepancy of only 30% occurs
for the maximum NH concentration, and the N H : N + OH . " NO + H.
profile is predicted almost perfectly everywhere. The
most serious discrepancies occur for the H N O and the (B58/A234)
nitrogen atom. Figure 44 shows that the model
Early in the flame, the NO concentration also is influ-
predicts about an order of magnitude more H N O
enced by the rate of HNO dissociation and by the
than is observed experimentally, and our prediction of
chain branching step,
3 x 10 -4 for the maximum nitrogen atom mole
fraction is significantly higher than the experimentally
H + O., . • OH + O.
determined upper limit of I x 10 -5. However, there
appears to be no reasonable way in which the ( - B 3 / - A132)
mechanism can be modified to improve the overall
agreement. The intermediate species concentrations The NO is removed primarily by reaction with
(free radicals, N20, and NO) are so strongly coupled nitrogen atoms,
that a change in the mechanism that has a large effect
on one species profile normally has a large effect on N + NO . " N: + O,
several of the others.
Our main interest in this flame is the NO and N 2 0 (B57/A232)
profiles. Sensitivity plots for these species are with only secondary contributions from the
presented in Figs 45 and 46. The NO in this flame is N H : + NO and NH + NO reactions.
formed both through the nitroxyl route and by N- The N~O is formed by reaction (B23/A202)
atom reactions. The main rate-controlling reactions
are the formation of nitroxyl via reactions (B29) and
NH + NO . ' N20 + H
(B34),
NH + OH . • HNO + H (B23/A202)
¢,)
0
B 30+OH--O2+H B 30+OH--O2+H
] B 23 NH+NO--N20+H B 23 NH+NO--N20+H
J B 29 NH+OH--HNO+H B 24 N20+H=N2+OH
/ B 32 NH+H,-,N+H2 B 29 NH+OH--HNO eH
_ _ J_,a 35 NH2+O--HNO+H B 32 NH+H=N+H2
~- ~ Fr~B39 NH2+NO--N2+H20 B 35 NH2+O--HNO+H
J B ,tO NH3+OH'-NH2+H20 O~
(~_ LI: B 36 NH2+OH=NH+H20
~ 50 HNO(+M)--H+NO(+M) B 37 NH2+H--NH+H2
/ ~ 7 '~---, ~ 56 N+NO--N2+O B 39 NH2+NO=N2+H20
//( ~6" ] - B 57 N+O2--NO+O B 49 NNH*O--N20+H
~ . ~ ~ ~ B 58 N+OH--NO+H B 50 HNO(+M)--H+NO(+M)
>-- 0 - B 51 HNO+OH--NO+H20
B 56 N+NO--N2+O
B 61 NH2+NH--N2H2+H
B 69 N2H2+NO=N20+NH2
~-. 04 , ',,,
i e0,
0
ua z
J
r~
,?.-
o-
? 9
0 0.25 0.50 0.75 I 0 0125 o15o o175
X(CM) X(CM)
FIG. 45. Sensitivity plot for NO for the flame conditions of FIG.46. Sensitivity plot for N20 for the flame conditions of"
Bian et al. 7° Bian et al. 7°
Nitrogen chemistry in combustion 309
N.,O + H , ' N 2 + OH. the stable hydrocarbon products (as well as CO and
CO2) and the stable nitrogenous species as a function
(B24/A204) of equivalence ratio (from ~ = 1 to ¢~ = 1.9) for a
In addition to these two reactions, Fig. 46 shows that fixed temperature and residence time (T = 1750K
the N.~O concentration is sensitive to reactions that and r = 7msec). They have compared their results
establish the N H and NO concentrations (and to a with PSR (perfectly stirred reactor) model predictions
based on the mechanism of Glarborg, Miller, and
lesser extent to reactions that determine the H-atom
concentration). Since the N H : (from which N H is Kee) ~The agreement is very good for both the hydro-
carbons 86 and the nitrogenous species. We include
formed), NH, and NO profiles are predicted accu-
here only a discussion of the nitrogen results in com-
rately, we are confident that we have correctly iden-
parison with the PSR predictions based on the
tified the N : O formation and destruction mechanism
mechanism of Appendix A, which is a slight extension
in this and in other ammonia flames.
and improvement over the Glarborg-Miller-Kee
The rate coefficients used in the present model for
mechanism used by the M.I.T. workers.
reactions (B23/A202) and (B24/A204) are discussed in
The M.I.T. experiments are particularly valuable in
more detail in Section 8. However, it should be noted
that they provide information not only about the NO
that reducing kB23 any further compromises the
produced, but also about the other principal stable
agreement between the predicted and experimental
fixed-nitrogen compounds, HCN and NH3. The com-
N , O profiles, as one might expect from the sensitivity
parisons between the model predictions and the ex-
analysis. On the other hand, increasing ka23 a t high
perimental results for these species are shown in Figs
temperature is easily compensated for by increasing
47--49. A comparison for the total fixed nitrogen
kBz~,
(TFN = NO + HCN + NH3) is shown in Fig. 50.
N : O + OH . ' N, + H e : , The agreement between the model and experiment is
excellent for NO and HCN, less good (but still quite
(B28/A203) acceptable) for NH3. The comparison for the T F N
0 HCN {experiment)
bility of increased values of kB.,3and kB28should not be HCN (model}
overlooked, s-~ O
X
5.4. F u e l - N i t r o g e n Conversion in W e l l - S t i r r e d
)--
Reactors (..)
<
=,
Longwell and coworkers s6"87at M.I.T. have recently o
performed a set of experiments in a jet-stirred reactor 0 0 C
C NH3 (exper~mentb
NH2+NH3 {model)
× O L
Z 7.
0
X
C
~ c ~J
<
2>
,,== O
© O
o
c C O O © 0 0 0 0
!2 14 16 18
FuedA~r Equwalence Ratio
14 16 !8
Fuel/Air Equivalence Ratio
FIG. 47. C o m p a r i s o n of the model predictions of N O con-
centrations with the well-stirred reactor data of Sun et a l ? ~
Fuel nitrogen = NH~ (~-2600ppmv); T = 1750K, resi- FIG. 49. C o m p a r i s o n of the model predictions o f N H 3 con-
dence time = 7msec. centrations with the well-stirred reactor data of Sun et al. ~"
310 J.A. MILLERand C.T. BOWMAN
(a)
J
16 A24 CH2~H=CH+ H2
A31 CH+HnC+H2
A32 CH+H20--CH20+H
A39 C+OH~CO+H
A77 C2H2+ O~FICCO+FI
1412 AlOl H+HCCO=ICH2+CO
x A132 O+OH--O2+H
© A156 C+NO=CN+O
Alal HCCO+NOnHCNO+CO
A163 HCNO+H'HCN÷OH
A166 CO2+N=NO+CO
A168 CH3+N=H2CN+H
=< 6-
A232
A233
N+NO--N2+O
N+O2--NO+O
=, A234 N+OH=NO+H
© 4- NO ADDED
Z a24
O
1.2 14 16 1'8
Fuel/Air Equivalence Ratio
(b)
25-
20-
HCN ADDED , . , ' " " " 1 125 1.50 1.75 2
x
z '"'*" NH3 ADDED
EQUIVALENCE RATIO
~.)
.-r
15-
z
_o F]G. 53. Sensitivity plot for NO for the conditions of the
L,) stirred-reactor experiments of Sun e t al. ~
< 10-
© A9 CH3+O=CH20÷H
5- °1_ A24 CH2+ H~CH+H2
A31 CH+H--C+H2
A32 CH+H20=CH20+H
A77 C2H2+O=HCCO+H
0 A101 H+HCCOslCH2÷CO
12 14 16 18 O 6
A! J_ A156 C+NOsCN+O
Fuel/Air E q u i v a l e n c e R a t i o A157 CH+NO-NCN+O
A161 HCCO+NO=HCNO+CO
A168 CH3+N--H2CN+H
A176 HCN+O--NCO+H
(c) A232 N+NO=N2+O
14
1.2
LU
x i
z
..r
Z
Z 0.8-
~-)
< 0.6-
C 1 1 25 1.50 "1.75
',.2 14 16 18
Fuel/Air E q u i v a l e n c e Ratio
EQUIVALENCE RATIO
FIG. 54. Sensitivity plot of selected reactions for HCNT for the
FIG. 52. Calculated (a) NO. (b) HCN, and (c) NH3 con- conditions of the stirred-reactor experiments of Sun e t al. ~7
centrations as a function of equivalence ratio for the three
different fuel-nitrogen sources. All calculations were per-
formed for the conditions of the stirred-reactor experiments CH + NO . • HCN + O (A157)
of Sun e t al. s~
begins to compete with the N + NO reaction, leading
to hydrogen cyanide formation. For this regime (and
The agreement between model and experiment in this for all 4) t> 1.3), the mechanism is much more com-
regime is excellent, reflecting a precise knowledge of plicated than for leaner conditions, as evidenced by
these rate coefficients. several reactions having comparable sensitivities. The
Starting roughly at ~ = 1.3, reaction of NO with most important direct (i.e. involving nitrogen
hydrocarbon free radicals (C and CH), compounds directly) steps determining the TFN and
its speciation for 1.3 < ¢ < 1.5 are conversion of
C + NO . • CN + O (A156) NO to HCN via reactions (A156) and (A157),
312 J.A. MILLERand C.T. BOWMAN
+02
+M,÷OH,+NH 2
NH *O2'÷OH="= H N O -'~ N O
•
.~-NH~ ~+NO
W
NNfi +NO
NH 3 ÷OH,.O ~ N2
(÷OH)
~NO
FIG. 56. Reaction path diagram for the Thermal De-NO~ process.
Lyon and Hardy. ~ Figure 58 shows the effect of ~5o ~o ~o~o ,5o ~25o ,3~o ~5o
adding NH~ in quantities that exceed the O2 and the T e m p e r a t u r e (K)
,~ECS IS:4-D
314 J.A. MILLERand C.T. BOWMAN
100 -
• date =el 2. NH3
(B44/A222)
• date set 3. NH$
50 - 0.0% H~O (model)
.... . .w.~...;~o.. Lm.o.d.:!L
HNO + M . • H + NO + M.
0
350 950 1060 ,,5o ,25o ,35o 1450
(B50/A228)
Temperature{K)
The H atom produced in reaction (B50) reacts with
FIG. 60. The effect of water on the NH 3 concentration at a molecular oxygen to produce OH and O via
• reaction time of t = 0.1 sec. Same conditions as Fig. 59.
H + 02 . • OH + O.
past few years, and some of these are reliable only in
narrow temperature ranges. Among the important ( - B 3 / - A 132)
parameters are the rate coefficients for reaction of Since reaction (B42) converts NH 3 to NH2 without
ammonia with OH, O, and H, the rate coefficient and losing a chain carrier, the reaction sequence (B40),
product distribution for the NH2 + O reaction, s3 (B38), (B44), (B50), (B3), (1342) results in a net gain of
three chain carriers per cycle. In the presence of water
NH: + O . • HNO + H
the oxygen atom reacts with H20 ,
(B34/A212)
O + H20 . • OH + OH,
NH 2 + O . • NH + OH ( - B 9 / - A138)
(B35/A213) and there is still a net gain of three chain carriers (OH)
per cycle. As long as chain termination processes do
and the rate coefficient and products of the NH + O: not interrupt this pattern a majority of the time, i.e. as
reaction,S l long as the branching sequence occurs at least a
quarter of the time, the overall reaction is self-sustain-
NH + 02 . • NO + OH. ing. The important chain termination reactions in the
present mechanism are
(B22[A201)
NH2 + NO • N2 + H20,
By far the most important aspect of the Thermal
De-NOx mechanism is the identity of the products of (B39/A217)
the NH2 + NO reaction. However, a discussion of
this point is deferred until later in this section. NH: + HNO • NH 3 + NO,
Thermal De-NO~ works because the NH: + NO
reaction has a significant (perhaps dominant) chain- (B53/A231)
branching component, allowing the overall reaction
and
to be self-sustainingJ3"1°7The ammonia is converted
to NH2 principally by reaction with OH, OH + HNO . • H:O + NO.
NH3 + OH . • NH2 + H20,
(B51/A229)
(B40/A218)
Reaction (B39), although it converts NO to N 2
but also in the absence of water vapor by reaction directly, is detrimental to the NO reduction scheme
with oxygen atoms, because it is chain terminating.
At the low-temperature end of the effective tem-
NH3 + O . • NH2 + OH.
perature range, the NO reduction is limited princi-
(B42/A220) pally by the rates of the chain termination reactions
that compete with the branching sequence described
In order to be self-sustaining the NH2 + NO reaction above. As temperature increases, the growth in chain-
must directly or indirectly regenerate OH and O to carrier concentrations becomes so rapid that event-
continue the NH3 ~ NH., conversion. In the model ually, in the absence of water, reaction (B34),
Nitrogen chemistry in combustion 315
(B36/A214) L~
and
NH + O2 . ' NO + OH
× H3
r~
(B22/A201)
or =<
LL N
NO
NH + 02 " HNO + O ©
(B2]/A200)
o
followed by HNO dissociation. 0 00 0 02 0 04 006 0,'08 0~10 O 12 0'14 016
To illustrate some of these points quantitatively, TIME (sec)
the calculated NO and NH3 time histories for the
1150 K and 1350 K cases (no H2) of Fig. 57 are plotted FIG. 62. NH 3 and NO time histories. Initial conditions cor-
in Figs 61 and 62 and the corresponding NO sensi- respond to data set 1 (see Fig. 57), T = 1350K.
tivity profiles are shown in Figs 63 and 64. For the
low-temperature case, increasing the reaction time
beyond t = 0.1 sec increases the NO reduction.
However, for the high-temperature case the reaction
is complete in about 0.03 sec. The high-temperature
case also exhibits the interesting behavior that the NO
>-
NH3/' "~
z
× LU
03--
Z ©
0
z
B5 H+O2+M--HO2+M \ 7B~_~
B38 N H 2 + N O ' N ~l'"t~"~;Iq ~ -7-
B39 NH2+NO--N2+H20 ~ |
i')40 NH3+OH--NH2+H20 B3e /
v?_ B43
B44
NNH--N2+H
NNH+NO==N2+HNO ~3e
BSO HNO+M--H+NO+M -T
B51 HNO+OH--NO+H20
B53 HNO+NH2==NH3+NO
0.00 0.02 004 0 06 0.08 010 0.12 014 0.16
TIME (sec) 0.0000.025 0.050 o.o7~ o.ioo 0125 0.150
TIME (sec)
F}G. 61. NH~ and NO time histories. Initial conditions cor-
respond to data set 1 (see Fig. 57), T = I150K. FIG. 63. Sensitivity plot for NO for the conditions of Fig. 61.
316 J.A. MILLERand C.T. BOWMAN
AT ~Xr~o
i and
fl(Ar) = X~'" ?A r
> ~tNO .
.... • ...
",,,
'\
o-
t
T-
Z ' zO
~u -
O-
',. Z ' :i
"'.-.•..
~(AT) "'"--... ........ ~(A T)
..P.(~L.. .. P3~L..
r
0.00 0102 oio4 Cos olo. o,o o,2 or," 036 000 oo2 00a o06 010s 0~0 0~2 0.14 0.16
TIME (sec) TIME (sec)
FIG. 65. Sensitivity of N O to the total rate coefficient and FIG. 66. Sensitivity o f N O to the total rate coefficient and
branching fraction of the reaction N H 2 + N O --, products branching fraction of the reaction N H : + N O --- products
for the conditions o f Fig. 57 (no H2), T = l 1 5 0 K . #(Ar) for the conditions o f Fig. 57 (no H,_), T = 1350K. fl(Ar)
a n d / / ( a ) are defined in the text. and fl(~) are defined in the text.
Nitrogen chemistry in combustion 317
not identical, effect--hydrogen peroxide dissociates fraction to produce OH: the results vary from less
into two hydroxyl radicals, thus increasing the chain than 15% ~°9 to greater than 65%. TM
Neither of the
carrier concentration. Our calculations with H202/ two groups that attempted to detect hydrogen atoms
NH3 addition show the effective temperature window were successful. This latter result implies that the
to be both broader and shifted to lower temperature radical products, if they occur at room temperature,
when compared with calculations in which NH3 is are N N H + OH and not N: + H + OH. The
added alone. A discussion of the effect of hydrogen situation has been further complicated by the sugges-
peroxide on the Thermal De-NO~ process, based on a tion "5 that the OH detected in the discharge-flow
reaction mechanism very similar to ours, is given by experiments TM was produced by the reaction
Cooper. ~72 sequence,
The self-inhibiting effect of ammonia is shown in
Fig. 58. For temperatures up to about 1275K, 3% F + NH 3 . " NH2 + HF
NH 3 addition produces less NO removal than when
385 ppm NH3 is added (compare Fig. 57). This effect F + NH2 . • NH + H F
occurs as a result of reaction (B41), H + NH 3
NH2 + H:, beginning to compete with reaction NH + NO . • N2 + OH,
( - B 3 ) , H + 02 --, OH + O, for hydrogen atoms.
Consequently, the rate of chain carrier growth is as a result of not having the NH 3 sufficiently in excess
reduced. At temperatures above 1275 K, NO removal to prevent the occurrence of the F + NH: reaction.
is enhanced by the large quantity of added ammonia However, this suggestion does not seem to have merit.
as a consequence of NH3 inhibiting chain carrier Although it is plausible that NH is produced by F-
growth, which enhances NO removal at higher tem- atom reactions in the discharge flow reactor experi-
perature. ments, it is unlikely that the NH + NO reaction
The 6ffect of adding water is a relatively minor one, produces appreciable OH. BAC-MP4 electronic
as shown in Figs 59 and 60. Nevertheless, this effect is structure calculations "6 (supported by the flame
predicted accurately by the mechanism. The principal modeling work discussed here) strongly suggest that
role of water is to compete for oxygen atoms through the only significant products from the NH + NO
reaction ( - B 9 ) , O + H: O --* OH + OH. At lower reaction are N:O + H.
temperatures this reaction competes with reaction Considerable insight into the NH: + NO reaction
(B42), NH3 4- O --, NH: + OH, and inhibits NO has been gained from ab initio electronic structure
removal. At higher temperatures it competes with calculations. "6-n9 Figure 67 shows a reaction coor-
reaction (B34), NH: + O --* HNO + H, and dinate diagram, constructed from BAC-MP4 calcu-
inhibits NO formation. The result is a shift of the lations, for this reaction. Products are formed as a
optimum temperature for the process to a higher consequence of a series of rearrangements of the inter-
value. mediate N H : . . . NO complex. These rearrange-
The key to the Thermal De-NOx process is the ments are surprisingly facile. 116A17 The products
NH., + NO reaction. As discussed above, for obtained are determined principally by the fate of the
Thermal De-NOx to work this reaction must produce complex,
free radicals at least a quarter of the time at tem-
peratures roughly above 900 K. The identity of these H
XN=N~ ' (II)
radicals is not completely clear at the present time. OH
Subsequent to the appearance of Refs 13 and 107,
several groups ~°9-~' have tried to determine directly
the branching fraction of the reaction to free radicals.
The strategy in all the experiments to date has been to Reaction of NH 2 + NO
try to detect the appearance of OH, either directly or HNNO + H
indirectly, as a product at room temperature. No ex-
periments have been conducted at temperatures N~NO J ~...__ N2H + OH
directly applicable to Thermal De-NOx. Two of the _¢ N2+H+OH
groups have also looked for hydrogen atoms in an ~, 40-
attempt to distinguish between the reactions
H2NNO H\NN\oH HtNN~oH~..~H2* NNO
NH: + NO . • NNH + OH
w
(B38/A2 ] 6) -40
N2 + H20
and -80
Reaction Coordinate Diagram
NH: + NO • N: 4- H + OH.
FIG. 67. Reaction coordinate diagram from BAC-MP4 cal-
In these studies there is no consensus on the branching culations for the reaction NH: -4- NO --, products.
318 J.A. MILLER a nd C.T. BOWMAN
formed by 1,3 hydrogen transfer from the initial Early in the reaction H N O dissociates,
addition complex,
HNO + M . ' H + NO + M,
H
"N--N----O (I) (B50/A228)
H
but as the free radical concentrations increase, the
If Complex (II) undergoes c i s - t r a n s isomerization, N: HNO more frequently reacts with OH or NH2,
and H : O are the products, as shown ifi the figure.
However, Complex (II) also may dissociate into N N H HNO + OH . • NO + H : O
and OH. The BAC-MP4 calculations indicate that the
N2 + H 2 0 channel is energetically favored but that (B51/A229)
NH2 + NO ~ N N H + OH also is possible as an
exothermic reactiori. HNO + NH: . " NH3 + NO,
An important consideration in establishing the
(B53/A231)
relative importance of the various channels of the
NH~ + NO reaction is the lifetime of the N N H resulting in chain termination. Consequently, with
radical. This molecule dissociates exothermically to N N H + OH as the radical products of the
N2 + H with a barrier to dissociation of unknown NHz + NO reaction, the reaction limits itself and a
height (BAC-MP4 calculations indicate only 6kcal/ larger branching fraction is required than if the
mol). Using the low energy barrier, Curtiss e t al. its products are N: + H + OH (i.e. if N N H has a
have estimated the lifetime of N N H with respect to vanishingly small lifetime).
tunneling to N: + H and have obtained a value of To support the conjecture that modeling Thermal
approximately 10 -H sec, a very small lifetime. This De-NO~ requires a long lifetime for NNH, a series of
result implies that the radical-producing channel of calculations was performed with the lifetime of N N H
the NH2 + NO reaction could be written as reduced to rNNH = 10-" sec, in which the branching
fraction, ~, of the NH: + NO reaction to free radicals
NH2 + NO . • N2 + H + OH was varied to find an optimum value. The results of
these calculations are compared with the experimental
equally as well as data of Fig. 57 (no Hz) in Fig. 68. F o r a 25%
branching fraction (~ = 0.25), no significant NO
N H , + NO . • N N H + OH.
removal is predicted to occur in 0.1sec. For
This result directly contradicts the experimental ob- = 0.508 (the value used in the preferred model),
servations discussed above that show OH, but not H, NO is removed at temperatures well below 900K,
to be a direct product of the NH2 + NO reaction. illustrating the importance of the self-limiting
There are several factors in the calculation that could character of the mechanism in which N N H has a long
cause it to be biased toward a short lifetime. These lifetime. As ~ is reduced below 0.508, the optimum
include the small barrier to dissociation, the one- temperature for NO reduction increases and the
dimensionality of the dynamical calculation effective temperature window gets nar.rower. Overall,
(including tunneling), the assumption that the the best agreement is obtained with ~ ~ 0.30. Even
potential energy along the dissociation coordinate can for this case the agreement with experiment is not as
be described as a parabola, and simplifying assump- good as for the preferred model, but the combination
tions about the internal dynamics of the N N H
2/5O
molecule. To what extent improving these aspects of
the calculation will. increase the lifetime is not clear.
Another factor that favors a long lifetime for N N H 200- ~ \ , / ~
is the modeling of experiments associated with i
Thermal De-NOx. The radical-producing reaction
NH2 + NO ~ N N H + OH has a self-inhibiting
effect on the NO reduction process, whereas
z
o-,-
1 o o - .... •=.-o.25
i~ !\ . / / .....'/
NH2 + NO ~ N, + H + OH does not. This self- ~--. ~ // // •
inhibition arises in the model because N N H does not .........~'~3° .... ~' Z /// "
~o- ~..... / \ "/ •
always dissociate; if its lifetime is sufficiently long, it .....
may undergo subsequent reactions that lead to chain
termination. In the present model these subsequent o
850 950 10;O 1150 1250 13;0 1450
reactions are initiated by Temperature (K)
of possible uncertainty in the experimental data and promoting H + O: recombination suppresses chain-
uncertainty in other rate parameters in the mechanism carrier growth and inhibits NO removal in the
may make this level of agreement acceptable. In Figs 1100-1200 K temperature range. This effect does not
69 and 70 the a = 0.30 model predictions are occur significantly for the long NNH-lifetime model
compared with the experimental data of Fig. 59 (4% because a greater portion of the chain-carrier growth
02, with and without added water). F o r the case comes directly from the NH 2 + NO reaction. For the
without water the agreement again is not as good as short NNH-lifetime model to be consistent with the
for the preferred model, but it may be adequate. The Thermal De-NO, experiments, it is necessary for the
real discrepancy arises for the case with 10% added effectiveness of water in promoting reaction (B5) to be
H20, or perhaps, -note correctly, in the differences essentially the same as that of a noble gas, a require-
predicted with and without H20. The experimental ment that is contradicted by a relatively large body of
results in Fig. 70 at T = 1200 K show that approxim- experimental data. ~
ately two-thirds of the NO is removed in 0.1 sec, The previous discussion notwithstanding, model
whereas the model predicts that virtually none is predictions for Thermal De-NO~ are insensitive to
removed. The experiments clearly show that the effect many details of the mechanism. For example, the
of adding water between l l 0 0 K and 1200K is self-limiting aspect of the long NNH-lifetime model
minimal, an effect predicted by the preferred (long could result directly from reactions of N N H with OH
N N H lifetime) mechanism. In contrast, the short and NH2, rather than through the HNO reaction
NNH-lifetime model leads to the conclusion that described above. There may be other possibilities as
water significantly inhibits the NO removal process well. In addition, one may exchange increases in ct for
between l l 0 0 K and 1200K (compare Figs 69 and decreases in N N H lifetime over a range of conditions
70), contrary to the experimental results. without substantially modifying the model predic-
The origin of the difference between the predictions tions. A question that arises in the long NNH-lifetime
of the short and long NNH-lifetime models is interest- model is why NO, is not formed through the
ing. In forcing the lifetime of N N H to be vanishingly sequence~3"20
small, it was necessary to reduce the value of ~ from
0.508 to 0.30 in order to optimize the agreement with N N H + O, . ' N2 + HO2
the Thermal De-NOx data (no H20). In doing so, a
greater burden for producing chain-carrier growth is HO: + NO . ' NO: + OH.
placed on the reaction
(B59/A188)
H + O: . " OH + - O ,
The failure of experiments to detect NO: in Thermal
( B 3 / - A132)
-
De-NO~ places a severe restriction on the rate of the
first of these steps.~3'~:° Even though our knowledge of
which at the low end of the temperature window is in many details in the mechanism is lacking, the process
competition with the chain termination step, can be described accurately by models such as the one
discussed above.
H + O., + M . ' HO: + M.
6.2. RAPRENO~ Mechanism
(B5/A134)
In the short NNH-lifetime model, when water is Another promising NOx abatement scheme based
added, the increased effectiveness of H 2 0 in on aftertreatment of combustor exhaust products is
250 250"
2o01 2oo-
,oo!
150 "-E 150"
R !
d
Z 100
50 50~
FIG. 69. Comparison of the small ZNNHmodel (a = 0.30) FIG. 70. Comparison of the small rNNH model (~ = 0.30)
with the NO experimental data of Fig. 59 for the case with with the NO experimental data of Fig. 59 for the case with
no added water. I0% added water.
320 J.A. M]LLFA~and C.T. BOWMAN
RAPRENOx. 96'97 In RAPRENOx, cyanuric acid, a decreases with increasing temperature. Coincident
commerically available chemical, is injected into an with the onset of NO reduction is an increase in the
exhaust stream containing NO. When heated, N : O concentration in the gas stream. At 1150 K, the
cyanuric acid sublimes and decomposes to form temperature giving maximum NO reduction, the N, O
isocyanic acid, concentration reaches a maximum value, correspond-
ing approximately to one mole of N , O formed for
(HOCN)3 ----* 3HNCO. each mole of NO removed.
At sufficiently high temperatures, the isocyanic acid The data in Fig. 71a present an interesting
will decompose either in the gas-phase or on surfaces challenge to the combustion modeler. There is not
and initiate a sequence of reactions leading to NO nearly so much known about the reaction kinetics
removal. Perry and Siebers97 reported NO removal in underlying the RAPRENOx process as there is about
simulated and real exhaust gas streams at tem- those underlying Thermal De-NOx. Perry and
peratures as low as 600 K in a stainless steel flow Siebers97 proposed the following mechanism to
reactor. Later experimental work by Siebers and describe the R A P R E N O , process:
Caton, m by Lyon and Cole, t61 and by Heap e t al. 162
indicated that at temperatures below 1000 K surface HNCO . " NH + CO l initiation
decomposition of H N C O initiates the RAPRENO.,
process and that NO reduction does not occur at low NH + NO . • N:O + H )
temperatures in the absence of such surfaces. In recent
experiments in a quartz flow reactor, 163-~65Siebers and H + HNCO . • NH: + CO
Caton found that there is a temperature window for
the RAPRENOx process and that the effectiveness of N: + H20
NH2 + NO . *
the process depends on the composition of the gas N N H + OH
stream. In addition, under some conditions, relatively
high concentrations of N 2 0 are formed as a by- NNH . • N: + H
product of the process. Figure 71a, taken from Ref.
165, shows the concentrations of NO, HNCO, CO,
CO + OH . • C02 + H.
CO:, and NzO in a simulated exhaust gas stream
measured at the exit of the quartz flow reactor as a Hence, it appears that the gas-phase RAPRENOx
function of reactor temperature. F o r the conditions of mechanism comprises reactions important in the
this experiment, significant NO removal begins to Thermal De-NOx process, reactions involving
occur at a temperature of 1000K, coincident with HNCO, and reactions important in the oxidation of
significant reductions in the concentrations of CO and moist CO. In order to model the data in Fig. 71a, we
H N C O and an increase in the CO2 concentration. As have combined the Thermal De-NOx mechanism of
the temperature increases to 1200 K, nearly all of the Appendix B with the moist CO oxidation reactions
NO present initially in the gas stream is removed. F o r from Appendix A (reactions (A61)-(A64);
temperatures above 1200K, the NO reduction (A131)-(A150)), the H N C O reactions shown in Table
'(a) 3, and reactions (A192)-(A198) involving NCO.
1500 . ' ' ' ' 1 .... I .... I'' 'el ' ' ,l. ' I '~_~_.1
HNCO = ~ e (b)
1500
g
:~
1000
750
,%
~t
L l E
1250
o. 1000
500
_~
750
NO
o
250 ~ , 500
2bO
800 900 1000 11O0 1200 1300 1400
Reactor Temperature - K
C¢
800 900 1000 1I00 1200 1300 1400
FtG. 71(a). Measured species concentrations at the exit of a
quartz flow reactor as a function of reactor temperature for Reactor TemiDerature - K
fixed initial mole fractions of NO (330ppmv), CO
(1260ppmv), O, (12.3%), H_,O (4.5%) and cyanuric acid FIG. 71(b). Calculated species concentrations at the exit of an
(470 ppmv, corresponding to 1410 ppmv HNCO at complete isothermal flow reactor as a function of reactor temperature
conversion), with the balance N:. The residence time in the for the initial conditions of Fig. 71(a). Reaction
reactor is approximately 0.8 sec. The data are from Ref. 165. time = 0.8 sec.
Nitrogen chemistry in combustion 321
Reaction A /~ E Ref.
Rate coefficients are in the form kf = ATOp(-E/RT).Units are moles, cubic centimeters, seconds, kelvins and calories/
mole.
1NH2+NO=NNH+OH
2 NH2+NO=N2+H20
3 NNH+NO=N2+HNO
4 HNO+OH=NO+H20
5 NCO+NO=N20+CO
% 6 SUMMATION
o
o
E
_5 -2 x 1 0 "8
o
u_
t-
O
-4 x 1 0 .8
E
0
o
_6=xlO-S ~ ~ = ~ i ¢ = , , I . . . . I . . . . I ,
0.05 0.10 0.15 0.20
Time [sec]
FIG. 72. Contribution factor plot for NO from the present RAPRENOx reaction mechanism for the initial
mole fraction of NO. CO, O,, H:O, and HNCO of Fig. 71 at 1050K.
322 J.A. MILLERand C.T. BOWMAN
f lj I NH+NO=N20+H
2 NCO+NO=N20+CO
3 SUMMATION
" • 4 x 10"8
3 x 1 0 "8
o
E
"5 2 x 1 0 8
u_
._~
~ 1 x10 8
o
0
0 [] []
0.05 0.10 0.15 0.20
Time [sec]
FIG. 73. Contribution factor plot for N20 from the present RAPRENO~ reaction mechanism for the
conditions of Fig. 72.
1H÷HNCO=NH2+CO
2 HNCO+OH=NCO+H20
3 HNCO+O=NH+CO2
40+OH=O2+H
5 H+O2+M=HO2+M
6 NCO+NO=N20+CO
7 CO+OH-CO2+H
0
c; -2
(/)
-4
-6
, L j , I ~ H I . . . . I--; , , , I ,
FIG. 74. Sensitivity plot for NO from the present RAPRENO~ reaction mechanism for the conditions of
Fig. 72.
reactions, ~A67 and the rate parameters for the overall • NCO + H20(AH = - 4 kcal/mol)
OH + H N C O .
reactions are reasonably well known. However, at the NH2 + CO2 (AH = - 33 kcal/mol).
present time, no information is available on the
relative importance of the variou.s product channels of We assumed that the abstraction channel is the major
these reactions. The most likely product channels for product channel at elevated temperatures, although
the H + HNCO reaction are, some contribution from the NH2-producing channel
cannot be ruled out at this time. The sensitivity plot,
Fig. 74, indicates that the reaction of O-atoms with
H + HNCO. ~NH2 + CO (AH = - 8 . 2 kcal/mol) HNCO plays only a minor role in the RAPRENOx
H2 + NCO (AH = 10.9kcal/moi).
mechanism.
In addition to the NO-removal channels involving
In our calculations, we assumed that the NH 2 channel NCO and NH2, there also is an NO-formation
is the primary product channel, in accordance with sequence involving NH2,
Ref. 166. Tully e t al. 167 have investigated the kinetics
HNO + H (B34/A212)
of the OH + HNCO reaction and suggest the NH2 + O . "
following product channels, NH + OH (B35/A213)
Nitrogen chemistry in combustion 323
NO
NCO ~ NzO
/ ,
I
HNCO
t ~ Na
Z
NH; l
No~~NNH
%%% A 188HO2+NO-NO2+OH
A 189 N O 2 + H = N O + O H
NH,HNO ~ NO A 190 N O 2 + O - - N O + O 2
(A 188). Since the rate of this reaction depends on the oil-fired furnaces, while the measured N : O levels in
HO, concentration, NO: formation also is sensitive to the stack gases of gas-fired combustors are low.~4~ 146
reactions forming and removing HO2. Of particular Recently, it has been suggested that a sampling
importance is artifact is responsible for the high N20 levels reported
in coal and oil flames and that the N:O levels in these
H + O2 + M . • HO2 + M, (A134) flames are similar to those measured in gas flames. :s
As noted earlier, a significant amount of N20 is
which is the principal HO2 formation reaction in the
formed as a by-product of the RAPRENO~ process
low-temperature region of the flame. The H-atoms
and N : O is an important nitrogen oxide under fuel-
needed for this reaction are transported from the
lean conditions. Hence, the mechanism for N:O
high-temperature regions by molecular diffusion.
formation and removal is of interest.
Because of the role that H-atoms play in HO2
Studies in laminar premixed flames ~47indicate that
lbrmation, NO2 formation also is sensitive to reaction
N : O is a very short-lived species in hot combustion
(A132). The principal NO2 removal step in this flame
gases and that the principal N20 formation reactions
is reaction (A189). Of much less importance for NO2
involve NO and various nitrogen-containing radicals.
removal in this flame is reaction (AI90).
From the above discussion, it is clear that the
NH + NO . " N20 + H (A202)
accuracy of predictions of NO2 concentrations in
flames is dependent both on the accuracy of the
kinetics parameters of reactions (A188) and (A189) NCO + NO . " N20 + CO. (A197)
and on the accuracy of the species diffusional
The N20 formed in these reactions rapidly reacts to
transport calculations, especially for NO and H-
form N2, principally by
atoms. The kinetics parameters for the principal NO2
formation reaction, (A188), are well known in the N:O + H • N, + OH. (A204)
temperature range of 230-1760 K, with an estimated
uncertainty of + 30% from the expression tabulated These conclusions are illustrated in Figs 46 and
in Appendix A. 2 There are no high-temperature data 78-80, which show results of N20 sensitivity and
for either of the two important NO2 removal rate-of-production analyses for three different flames
channels. However, both of these reactions have a and a well-stirred reactor. Figure 46 shows the
very small temperature dependence. Hence, ex- principal N20 formation and removal channels in a
trapolation of the low-temperature data to flame low-pressure, lean (~b = 0.71) NH3 - O5 flame] ° As
conditions can be done with confidence. Estimated discussed in Section 5.3, the principal N20 formation
uncertainties in the rate coefficients for both reactions reaction is (A202), and the primary N20 removal
are + 30% in the temperature range 298-650 K. 2 channel is (A204). Figure 78 shows the principal N20
Perhaps the largest kinetic uncertainty is introduced formation and removal channels in an atmospheric-
through uncertainties in the third body efficiencies for pressure, rich (~b = 1.37) methane-air flame with
reaction (A134), particularly for collision partners 1300ppm of added NO. In this flame, the
such as CH4 and H20. As noted above, diffusional NH + NO --, NzO + H channel is somewhat more
transport plays a major role in the NO2 formation and important than NCO + NO --* N.~O + CO for
removal process in flames. Uncertainties in multicom-
ponent diffusion coefficients for species such as NO,
9
NO2, HO:, and H-atom will t:esult in uncertainties in r~ A 197NCO+NO-N20+CO I
® ~ A 20~NH+NOIN20+H I
calculated NOz profiles in flames that may exceed
those introduced by kinetic uncertainties. While ad-
ditional high-temperature rate data on reactions
(A188, A189) are desirable, the reaction mechanism
and rate parameters for NO2 formation and removal
are adequate for combustion calculations. ~ 'A
A204
204~
At flame temperatures NO2 can exist only as a -~,
transient species. If NO2 is to persist in the combus-
tion products, then there must be quenching of the
NO2 formed in the flame. This quenching might occur
in turbulent flames by rapid mixing of hot and cold z
fluid elements, which serves to quench the NO2-
removing reactions by reduction in radical concen-
trations, t36,138-t4o
A197 NCO+NO~N20+CO
A202NH+NO--N20+H
A204N20+H--N2+OH
A2OSN20+M~N2+O+M
A207N20+O=2NO
A
aO
o
x
)7 g- ~0~ ¸
m ~ .
uJ
a: LU
n"
z
o
z
A 202NH+NO~N20+H ~ L A204j
A 204N20+H=N2+OH
i
O5 1 ~5 2 25 05o0175 1 125 1~0 175
X(CM)
EOUIVALENCE RATIO
FIG. 79. Contribution factor plot for N , O in a low-pressure FIG. 80. Contribution factor plot for N 2 0 in a well-stirred
(25 torr) rich ('-h = 1.5) H 2 - O : - A r flame to which 2% H C N reactor for the experimental conditions of Bartok et al. ~6
was added.
N~O formation. As for the ammonia flame, the only reaction NH + NO ~ products. However, there is
significant N, O removal channel is Considerable disagreement among the high-
N~O + H --, N: + OH. From this figure it can be
seen that the N , O formed by reactions (A202) and ' ' ' ' I . . . . I . . . . I . . . .
H a r r i s o n et at [152]
(A197) is rapidly and quantitatively removed by Gordon e t al. [148] "]1-- --
Miller et al, [15] %. _ .-J. . . .
reaction (A204). Figure 79 shows results for a low- " ~ , . . - - , ~ - ~- ~- - ~ ~ - - Hansen el al. [149]
13 ' ~ ' ~~ ~ ~ ~'~This Daper Cox e t a l [ 151 ] -
pressure (25torr) rich (05 = 1.5) H2-O2-Ar flame . ~ "~ Morley 1j~[51
with 2% added HCN. ]° Here, the principal N : O "6 ~' - ~ ~ " ~ Peterson [154]
formation channel is NCO + NO --* N 2 0 + CO,
although NH + NO --* N.,O + H makes a signifi- { . \
cant contribution to N.,O formation. As before, the
principal N:O removal channel is N 2 0 + H - - , 11
Roose el aL [155] ~ ~ ~ "
N, + OH. Figure 80 shows the principal N:O
oF ~-- %,o,,
formation and removal channels in a well-stirred
t L i ~ k ~ t i i 1 k i i ~ J i J i
reactor for methane and air. 36 For lean and moder- 10
2 3
ately rich fuel-air mixtures (05 < 1.2), the principal
1000/T(K)
N:O formation reaction is
FIG. 81. Compilation of rate coefficients for NH t NO --,
O + N,+ M ..... " N : O + M. products, and an energy diagram for various intermediates
and products from Ref. 116.
( - A205)
14.0
' ' ' ' I . . . . I . . . . r . . . . J
F
For richer mixtures, reactions (A197) and (A202) -t
become the principal N:O-formation steps. For lean
mixtures, the most important N20-removal reaction 13.5
Cookson et al. [150 I
is (A204), forming N:, witti a small contribution from _m
Peru/[156] ~
respectively. There is excellent agreement a m o n g the FIG. 82. Compilation of rate coefficients for
lou-temperature studies for the rate of the overall NCO + NO ---, products.
326 J.A. MILLERand C.T. BOWMAN
temperature studies, all of which involved an indirect However, Perry notes that other exothermic channels,
determination of the reaction rate. In none of the such as
studies published to date has a direct measurement
been made of the products of this reaction. The results NCO + NO . " N, + CO,.
of a BAC-MP4 calculation, H6 shown as an inset in
Fig. 81, indicate that the NH + NO reaction (AH = - 153 kcal/mol)
proceeds through a stable HNNO intermediate to and
three possible sets of products. The energy barriers
associated with the subsequent reactions of H N N O NCO + NO . " N: + CO + 0
indicate that the primary product channel at tem-
peratures of interest in combustion is (AH = -26kcal/mol)
grounds that the most likely channel is the exothermic 0.5 1.0 1.5 2.0 2.5 3.0
reaction, IO00FF(K)
NCO + NO . ' N20 + CO FIG. 83. Compilation of rate coefficients for N:O + H --)
products, and an energy diagram for various intermediates
(AH = - 65kcal/mol). and products from Ref. 159,
Nitrogen chemistry in combustion 327
peratures. The reaction exhibits distinct curvature on extent to which reactions such as NH + NO
an Arrhenius diagram below 750 K, which has been products and N C O + NO --. products produce N : O
attributed to tunneling) 59 Results from BAC-MP4 is not yet firmly established. Lastly, the role played by
calculations for the reaction '59 are shown as an inset tCH2 in producing low-molecular-weight hydrocar-
on Fig. 83. These calculations indicate that the bon free radicals needs further clarification. Very little
dominant channel at combustion temperatures is information is available on reactions of ~CH2 at any
temperature, particularly at temperatures of interest
N20 + H ~ ... ---* N: + OH. in combustion.
At 2000 K, the rate of the NO-producing channel It is unlikely that the mechanism presented here is
the final word on the subject. However, future modifi-
N:O + H . . . . . NH + NO cations are more likely to be quantitative rather than
qualitative, and diagrams such as Fig. 11 are likely to
is estimated to be less than 7% of the overall reaction
remain relatively unaltered. We hope that the present
rate. In the present model calculations, we have used article (and the reaction mechansim) will be a useful
the rate expression in the Leeds compilation, which is point of departure for future discussions of nitrogen
in close agreement with the direct measurements at chemistry in combustion.
temperatures above 800 K. In the temperature range
800-2000K, the estimated uncertainty of the rate
Acknowledgements--We would like to thank Fran Rupley
coefficient expression is + 50%. for preparing the figures for this paper. In addition, we
acknowledge John Kramlich, Dick Lyon, Dennis Siebers,
and Mark Zahniser for providing copies of some of their
9. CONCLUDING REMARKS papers prior to publication. JAM is grateful to Bob Kee
(Sandia), Peter Glarborg (Technical University of
Denmark), and Carl Melius (Sandia) for their past and
In this paper we have tried to synthesize an continuing influence on his thinking about modeling of
extensive amount of information into a consistent and nitrogen chemistry in combustion. We are also grateful to
coherent description of the mechanism by which Peter Glarborg for doing the initial calculations that resulted
nitrogen compounds react under conditions typically in Figs 5 and 6. CTB would like to thank Dan Seery, Brian
Haynes, and Ron Hanson for many stimulating discussions
found in combustion systems. While the reaction of the kinetics of nitrogen species in flames.
mechanism has been tested against experimental data
over a relatively broad range of temperatures,
pressures, stoichiometries, and fuel types, caution REFERENCES
should be exercised in applying the mechanism
outside of the range of conditions considered in this l. BAULCH,D. L., DRYSDALE,D. D., HORNE, D. G. and
LLOYD, A. C. Evaluated Kinetic Data .for High Tem-
paper.
perature Reactions. Vol. 1 and Vol. 2, CRC Press,
In the process of surveying rate data on elementary Cleveland (1973).
reactions in the mechanism and comparing model 2. HANSON, R. K. and SAL1MIAN, S. Survey of rate
predictions with experimental data, we identified constants in the N/H/O system, Combustion Chemistry,
several areas where additional kinetic information is W.C. Gardiner, Jr (Ed.), Springer, New York (1984).
3. gEE, R. J., MILLER, J. A. and JEFFERSON, T. H.
needed. Notable among these are the reactions CHEMKIN: A general-purpose, problem-indepen-
between carbon-containing radicals and N: and NO, dent, transportable, Fortran chemical kinetics code
particularly package, Sandia Rep. SAND80-8003 (1980).
4. gEE, R. J., RUPLEY, F. M. and MILLER, J. A. The
CH + N., . • HCN + N chemkin thermodynamic data base, Sandia Rep.
SAND87-8215 (1987).
5. gEE, R. J., DIXON-LEwIS,G., WARNATZ,J., COLTRIN,
and M. E. and MILLER,J. A. A Fortran computer package
for the evaluation of gas-phase, multicomponent
CH + NO . • HCN + 0 transport properties, Sandia Rep. SAND87-8246
(1986).
6. gEE, R. J., GRCAR, J. F., SMOOKE, M. D. and MILLER,
C + NO , • CN + O J. A. A Fortran program for modeling steady laminar
one-dimensional premixed flames, Sandia Rep.
CH: + NO . " HCNO + H SAND85-8240 (1985).
7. GLARBORG,P., gEE, R. J., GRCAR, J. F. and MILLER, J.
A. PSR: A Fortran program for modeling well-stirred
H C C O + NO . • H C N O + CO. reactors, Sandia Rep. SAND 86-8209 (1986).
8. LUTZ,A. E., KEL R. J. and MILLER, J. A. SENKIN: A
The branching fraction of the NH: + NO --. Fortran program for predicting homogeneous gas
products reaction, particularly at high temperature, phase chemical kinetics with sensitivity analysis. Sandia
Rep. SAND87-8248 (1988).
and the lifetime of N N H are still uncertain and need 9. GRCAR, J. F., KEE, R. J., SMOOKE, M. D. and MILLER,
to be determined accurately for a truly fundamental J. A. 21st Syrup. :Int.) Combust., pp. 1773-1782. The
understanding of NO aftertreatment schemes. The Combustion Institute, Pittsburgh (1987).
328 J.A. MILLERand C.T. BOWMAN
10. MILLER, J. A., BRANCH, M. C., MCLEAN, W. J., 37. WHYTE,A. R. and PHILLIPS,L. F. Chem. Phys. Lett. 98,
CHANDLER,D. W., SMOOKE,M. D. and KEE, R. J. 20th 590 (1983).
Syrup. (Int.) Combust., pp. 673-684, The Combustion 38. SLACK, M. W. J. Chem. Phys. 64, 228 (1976).
Institute, Pittsburgh (1985). 39. NATARAJAN, K. and Roan, P. 21st Syrup. (Int.)
1I. GLARBORG,P., MILLER,J. A. and KEE, g. J. Combust. Combust., pp. 729-737, The Combustion Institute,
Flame 65, 177 (1986). Pittsburgh (1986).
12. THORNE, L. R., BRANCH, M. C., CHANDLER, D. W., 40. GLARaORG, P. Kinetisk modellering of dannelse og
KEE, R. J. and MILLER, J. A. 21st Symp. (Int.) nedbrydning af NOx ved forbraending af simple
Combust., pp. 965-977, The Combustion Institute, kulbrinter, Doctoral thesis, Technical University of
Pittsburgh (1986). Denmark (1987) (in Danish).
13. MILLER,J. A., BRANCH,M. C. and KEE, R. J. Combust. 41. MALI'E,P. C. and PRATT,D. T. Combust. Sci. Technol.
Flame 43, 81 (1981). 9, 221 (1974).
14. MILLER, J. A., MITCHELL, R. E., SMOOKE,M. D. and 42. MAL't'E, P. C. and PRAtt, D. T. 15th Symp. (Int.)
KEE, R. J. 19th Syrup. (Int.) Combust., pp. 181-196, Combust., pp. 1061-1070, The Combustion Institute,
The Combustion Institute, Pittsburgh (1982). Pittsburgh (1975).
15. MILLER,J. A., SMOOKE,M. D., GREEN, R. M. and KEE, 43. ZABARNICK,S.. FLEMING,J. W. and LIN, M. C. 21st
R. J., Combust. Sci. Technol. 34, 149 (1983). Syrup. (Int.) Combust., pp. 713-719, The Combustion
16. WAGNER,A. F. and WARDLAW,O. M. J. Phys. Chem. Institute, Pittsburgh (1986).
92, 2462 (1988). 44. FENIMORE,C. P. Combust. Flame 19, 289 (1972).
17. STEWART,P. H., SMITH,G. P. and GOLDEN, D. M. The 45. AXWORTHY, A. E. and DAYAN, V. H. Chemical
pressure and temperature dependence of methane de- reactions in the conversion of fuel nitrogen to NOr:
composition, Int. J. Chem. Kinet., in press. Fuel pyrolysis studies, 2nd EPA Stationary Source
18. MONAT,J. P., HANSON,R. K. and KRUGER,C. H. 17th Combust. Syrup., September (1977).
Syrup. (Int.) Combust., pp. 543-552, The Combustion 46. FENIMORE,C. P. Combust. Flame 26, 249 (1976).
Institute, Pittsburgh (1979). 47. DE SOETE, G. G. 15th Syrup. (Int.) Combust.,
19. ZELDOVICH,YA. B., SADOVNIKOV,P. YA. and FRANK- pp. 1093-1102, The Combustion Institute, Pittsburgh
KAMENETSKII, D. A. Oxidation of nitrogen in (1975).
combustion, translated by M. Shetef, Academy of 48. I'--I-ouSER,T. J., HULL, i . , ALWAY,R. M. and BIFTU,T.
Sciences of the USSR, Moscow (1947). (See also Int. J. Chem. Kinet. 12, 579 (1980).
ZELDOVICH,Ya. B., BARENBLATT,G. I., L1BROVICH,V. 49. HAYNES,B. S. Combust. Flame 28, 113 (1977).
B. and MAKHVILADZE,G. i . The Mathematical Theory 50. MORLEY,C. Combust. Flame 27, 189 (1976).
of Combustion and Explosions, pp. 30-36, Consultants 51. MORLEY, C. 18th Syrup. (Int.) Combust., pp. 23-32,
Bureau, New York (1985).) The Combustion Institute, Pittsburgh (1981).
20. FENIMORE, C. P. 13th Syrup. (Int.) Combust., 52. FENIMORE, C. P. 17th Syrup. (Int.) Combust.,
pp. 373-379, The Combustion Institute, Pittsburgh pp. 661-670, The Combustion Institute, Pittsburgh
(1971). (1979).
21. IVERACH, D., KIROV, N. K. and HAYNES, B. S., 53. MILLER, J. A. and MELIUS, C. F. 21st Syrup. (Int.)
Combust. Sci. Technol. 8, 139 (1973). Combust., pp. 919-927, The Combustion Institute,
22. BACHMEIER,F., EBERIUS,K. H. and JUST,TH. Combust. Pittsburgh (1986).
Sci. Technol. 7, 77 (1973). 54. DAVIES,P. B. and Trmusn, B. A. Trans. Faraday Soc.
23. HAYHURST,A. N. and McLEAN, H. G., Nature 251, 303 64, 1836 (1968).
(1974). 55. ROTH, P., LOHR, R. and HERMANNS, H. D. Ber.
24. HAYHURST, A. N. and VINCE, I. M. Prog. Energy Bunsenges. Phys. Chem. 84, 835 (1980).
Combust. Sci. 6, 35 (1980). 56. LOUGE, M. Y. and HANSON, R. K. 20th Syrup. (Int.)
25. IVERACH,D., BASDEN, K. S. and Kmov, N. Y. 14th Combust., pp. 665-672, The Combustion Institute,
Symp. (Int.) Combust., pp. 767-775, The Combustion Pittsburgh (1984).
Institute, Pittsburgh (1973). 57. SZEKELY,A., HANSON,R. K. and BOWMAN,C. T. 20th
26. BLAUWENS,J., SMETS, B. and PEETEP,S, J. 16th Syrup. Symp. (Int.) Combust.. pp. 647-654, The Combustion
(Int.) Combust., pp. 1055-1064, The Combustion Institute, Pittsburgh (1984).
Institute, Pittsburgh (1977). 58. SZEI~LY, A., HANSON, R. K. and BOWMAN,C. T. Int.
27. MIYAUCm,T., MORI, Y. and IMAmJkX, A. 16th Syrup. J. Chem. Kinet. 16, 1609 (1984).
(Int.) Combust., pp. 1073-1082, The Combustion 59. FRITZ, B., LORErCrZ,K., S~INERT, W. and ZELLNER,R.
Institute, Pittsburgh (1977). Oxidation Commun. 6, 363 (1984).
28. MATSUI,Y. and NOMAGUCHI,T. Combust. Flame 32, 60. MVERSON,A. L., TAYLOR,F. R. and FAUNCE,B. G. 6th
205 (1978). Syrup. (Int.) Combust.. pp. 154--163, Reinhold, NY
29. MA~UI, Y. and YtmKl, A. Jap. J. Appl. Phys. 24, 598 (1957).
(1985). 61. WENDT,J. O. L., STERNLING,C. V. and MATOViCH,M.
30. PERRY, R. A. and MELIUS, C. F. 20th Syrup. (Int.) A. 14th Syrup. (Int.) Combust., pp. 897-904, The
Combust., pp. 639-646, The Combustion Institute, Combustion Institute, Pittsburgh (1973).
Pittsburgh (1984). 62. MYERSON, A. L. 15th Syrup. (Int.) Combust.,
31. MILLER,J. A.. PARRISH.C. and BROWN, N. J. J. Phys. pp. 1085-1092, The Combustion Institute, Pittsburgh
Chem. 90, 3339 (1986). (1975).
32. BENSON,S.W. 16th Syrup. (Int.) Combust., p. 1062, The 63. SONG, Y. H., BLAIR, D. W., SIMINSKI, V. J. and
Combustion Institute, Pittsburgh (1977). BARTOK, W. 18th Syrup. (Int.) Combust., pp. 53-63,
33. SANDERS,W. A., Lls, C. Y. and LIN, M. C. Combust. The Combustion Institute, Pittsburgh (1981).
Sci. Technol. 51, 103 (1987). 64. CFIEN,S. L., MCCARTHY,J. M., CLARK,W. D., HEAP,
34. BAIR, R. A. Annual Report, Theoretical Chemistry M. P., SEEKER,W. R. and PERSmNG, D. W. 21st Syrup.
Group (Chemistry Division), Argonne National (Int.) Combust.. pp. 1159-1169, The Combustion
Laboratory, September (1985). Institute, Pittsburgh (1986).
35. BERMAN,i . R. and LIN, M. C. J. Phys. Chem. 87, 3933 65. MIYA~, S., lrdmE, H. and MAKINO, K. Evaluation of
(1983). in-furnace NOv reduction, Proc. 1985 Joint Syrup.
36. BARTOK,W., ENGLEMAN,V. S., GOLDS'rEIN,R. and DEL Stationary Combustion NO V Control (1986).
VALLE, E. G. AIChE Syrup., Ser. 126 68, 30 (1972). 66. LANGFORD, A. O., PETEK. H. and MOORE, C. B. J.
Nitrogen chemistry in combustion 329
JPECS lS:4-E
330 J.A. MILLERand C.T. BOWMAN
Ass. 35, 231 (1985). 156. PERRY. R. A. J. Chem. Phys. 82, 5485 (1985).
125. MERRYMAN, E. L. and LEVY, A. 15th Syrup. (Int.) 157. ZArtNISER, M. S., MCCURDY. K. and KOLB. C. E.
Combust., pp. 1073-1083, The Combustion Institute, Chem. Phys. Lett., to be published.
Pittsburgh (1975). 158. ALBERS. E. A.. HOYERMANN, K., SCHACKE. H.,
126. FENIMORE,C. P. Combust. Flame 25, 85 (1975). SCHMATJKO, K. J. and WAGNER. G. GG. 15th Symp.
127. HARGKEAVES,K. J. A., HARVEY,R., ROPER, F. G. and (Int.J Combust., pp. 765-773, The Combustion
SMITH, D. B. 18th Syrup. (Int.) Combust., pp. 133-142, Institute, Pittsburgh (1975).
The Combustion Institute, Pittsburgh (1981). 159. MARSHALL.P., FONTIJN,A. and MELIUS,C. F. J. Chem.
128. CERANSKY,N. P. and SAWYER,R. F. 15th Syrup. (Int.) Phys. 86, 5540 (1987).
Combust., pp. 1039-1048, The Combustion Institute, 160. LESCLAUX,R. Rev. Chem. Intermed. 5, 347 (1984).
Pittsburgh (1975). 161. LYON, R. K. and COLE, J. A. A reexamination of the
129. SCI-mFER,R. W. and SAWYER,R. F. 16th Syrup. (Int.) RAPRENO~ process, Combust. Flame. submitted for
Combust., pp. 119-134, The Combustion Institute, publication.
Pittsburgh (1977). 162. HEAP, M. P., C~EN, S. L., MCCARTHY, J. M.,
130. OVEN, M. J., GOULDIN, F. C. and MCLEAN, W. J. 17th KRAMLICH, J. C. and PERSHING,D. W. An advanced
Syrup. (Int.) Combust., pp. 363-374, The Combustion selective reduction process for NO~ control, Nature.
Institute (1979). submitted for publication.
131. DRAKE,M. C., CORREA,S. M., PITZ, R. W., SHYY, W. 163. SIEBERS, D. L. and CATON, J. A. Removal of nitric
and FENIMORE,C. P. Combust. Flame 69, 347 (1987). oxide from exhaust gas with cyanuric acid, Western
132. DUTERQUE,J., AVEZARD,N. and BORGHI,R. Combust. States Sect./Combust. Inst. Meet.. Pap. 88-66 (1988).
Sci. Technol. 25, 85 (1981). t64. CATON,J. A. and SIEBERS,n. L. Reduction of nitrogen
133. ALLEN,J. D. Combust. Flame 24, 133 (1975). oxides in engine exhaust gases by the addition of
134. JOHNSON,G. M., SMITH, M. Y. and MULCAHY,M. R. cyanuric acid, Sandia Rep. SAND88-8693 (1988). Also
F. 17th Symp. (Int.) Combust., pp. 647-660, The A S M E J. Eng. Gas Turbines Pwr, submitted for pu-
Combustion Institute, Pittsburgh (1979). blication.
135. SAND,T. Combust. Sci. Technol. 29, 261 (1982). 165. CATON,J. A. and SmBERS,D. L. Comparison of nitric
136. SAND,T. Combust. Sci. Technol. 38, 129 (1984). oxide removal by cyanuric acid and by ammonia,
137. SAND,T. Combust. Sci. Technol. 43, 259 (1985). Western States Sect./Combust. Inst. Meet., Pap. 88-67
138. McLEAN, W. J. 17th Symp. (Int.) Combust., (1988).
pp. 659-660, The Combustion Institute, Pittsburgh 166. MERTENS, J. D., CHANG, A. Y., HANSON, R. K. and
(1979). BOWMAN, C. T. Decomposition kinetics of HNCO at
139. HoP,i, M. 21st Syrup. (Int.) Combust., pp. 1181-1188, high temperatures, Western States Sect./Combust. Inst.
The Combustion Institute, Pittsburgh (1986). Meet.. Pap. 88-64 (1988).
140. LEUNG, S. W., STRAND, M., SAWYER, R. F. and 167. TULLY, F. P., PERRY, R. A., THORr,m, L. R. and
KOSHLAND, C. P. The formation and destruction of ALLENOORF,M. A. Free-radical oxidation of isocyanic
nitrogen dioxide, Proc. Joint Conf. Western States Jap. acid, presented at the 22nd Symp. (Int.) Combust.,
Sect. Combust. Inst., 11-13 (1987). (1988).
141. PmROTTI,D. and RASMUSSEN,R. A. Geophys. Res. Lett. 168. ROTH, P. and IBREIGHITH,M. Combust. Flame 55, 279
3, 265 (1976). (1984).
142. WEISS,R. F. and CRAIG, H. Geophys. Res. Lett. 3, 751 169. WARNATZ,J. Rate coefficients in the C/H/O System,
(1976). Combustion Chemistry, W. C. Gardiner, Jr (Ed.),
143. HAD, W. M., WOFSY,S. C., MCELROY, M. B., BEER,J. Springer, New York, (1984).
M. and TOQAN,M. A. J. Geophys. Res. 92, 3098 (1987). 170. TSANG, W. and HAMPSON, R. F. J. Phys. Chem. Ref
144. KRAMLICH, J. C., NIHART, R. K., CriEN, S. L., Data 15, 1087 (1986).
PERSHING,D. W. and HEAP, M. P., Combust. Flame 48, 171. DEAN, A. J., DAVIDSON, D. F., HANSON, R. K. and
I01 (1982). BOWMAN, C. T. Development and application of CH
145. KRAMLICH, J. C., COLE, J. A., MCCARTHY, J. M., laser absorption diagnostic for shock tube kinetic
LANIER, W. S. and MCSORLEY,J. A. Combust. Flame, studies, Western States Sect./Combust. Inst. Meet.,
in press. Pap. 88-91 (1988).
146. HAD, W. M., WOFSY, S. C., MCELROY, M. B., 172. COOPER, D. A. J. Inst. Energy 61, 78 (1988).
FARMAYAN, W. F., TOQAN, M. A., BEER, J. M., 173. DEAN, A. J., DAVIDSON, D. F. and HANSON, R. K.
ZAHNISER, M. S., SILVER, J. A. and KOLB, C. E. C-atom ARAS diagnostic for shock tube studies of
Combust. Sci. Technol. 55, 23 (1987). C + N, -, CN + N, Proc. 17th Int. Symp. Shock
147. ROBY, R. J. and BOWMAN,C. T. Combust. Flame 70, Waves "Shock Tubes (1989).
119 (1987). 174. ROBY, R. J. and BOWMAN,C. T. Effects of hydrocar-
148. GORDON, S., MULAC, W. and NANGIA, P. J. Phys. bons on nitric oxide removal in rich, premixed hyd-
Chem. 75, 2087 (1971). rogen-oxygen flames, Joint Meet. Western States
149. HANSEN,I., HOINGHAUS,K., ZETZSCH, C. and STUHL, Canadian Sect. Combust. Inst., Pap. 86-8 (1986).
F. Chem. Phys. Lett. 42, 370 (1976). 175. SEERY, D. J. and ZABIELSKI,M. F. 18th Syrup. (Int.)
150. COOKSON,J. L., HANCOCK,G. and MCKENDRICK, K. Combust.. pp. 397--404, The Combustion Institute,
G. Ber. Bunsenges. Phys. Chem. 89, 335 (1985). Pittsburgh (1981).
151. Cox, J. W., NELSON, H. H. and McDONALD, J. R. 176. CATTOLICA, R. J. 22nd Symp. (Int.) Combust.,
Chem. Phys. 96, 175 (1985). pp. 1165-1173, The Combustion Institute, Pittsburgh
152. HARRISON,J. A., WHYa'E, A. R. and PHILLIPS, L. F. (1989).
Chem. Phys. Lett. 129, 346 (1986). 177. SARV, H. and CERNANSKY,N. P. Combust. Flame 76,
153. NIP, W. S., Ph.D. Thesis, University of Toronto (1974). 265 (1989).
154. PE'rERSON, R. C., Ph.D. Thesis, Purdue University 178. i u z l o , L. J. and KRAMLICH,J. C. Geophys. Res. Lett.
(1981). 15, 1369 (1988).
155. RoosE, T. R., HANSON,R. K. and KRUGER,C. H. 18th 179. DUTERQUE,J., AVEZARD,N. and BORGHI, R. Combust.
Symp. (Int.) Combust., pp. 853-862, The Combustion Sci. Technol. 25, 85 (1981).
Institute, Pittsburgh ( 1981).
Nitrogen chemistry in combustion 331
A P P E N D I X A.
Reaction Mechanism Rate Coefficients In F o r m k! = A T ~ e x p ( - E / R T ) .
Units are moles, cubic centimeters, seconds, Kelvins and calories/mole.
REACTION A B E
I. 2CH~(+M) = C, H e ( + M )
High pressure limit: 9.03e+16 - 1.2 654.
Low pressure limit: 3.18e+41 -7.0 2762.
Troe parameters:* a = 6.04c-01, T ' = 6927.0, T ' " = 132.
Enhanced third-body efl~ciencies:
H 2 = 2.0, C O = 2.0, C O s = 3.0, H 2 0 = 5.0
2. CH3 + H ( + M ) ~- C H , ( + M )
High pressure limit: 6 . 0 0 e + 16 - 1.0 O.
Low pressure limit: 8.00e+26 -3.0 O.
SRI parameters: 2 4.50e-01 797.0 979. 1.
Enhanced third-body efllciencies:
H 2 = 2.0, CO= 2.0, CO~= 3.0, H~O= 5.0
3. CH4 + 02 ~ O H 3 + H 0 2 7.90e+13 0.000 56000.
4. CH4 + H 4~- CH3 + H2 2.20e+04 3.000 8750.
5. CH, + O H ~- C H 3 + H 2 0 1.60e+06 2.100 2460.
5a. CH4 + 0 ~ OH3 + O H 1.02e+09 1.500 8604.
6. CH4 + H O 2 ~- CH: + H202 1.80e+ 11 0.000 18700.
7. C H 3 + H 0 2 ~- C H 3 0 + O H 2.00e -- 13 0.000 0.
8. C H 3 + 02 ~ C H 2 0 + O 2.05e--19 -1.570 29229.
9. CH3 + O ~ CH~O + H 8.00c-- 13 0.000 0.
I0. C H 2 O H + H ~ CH~ + O H 1.OOe-- 14 0.000 0.
11. CH:O+H ~ CH2+OH 1.OOe-- 14 0.000 0.
12. CH3 + O H "2- C H ~ + H 2 0 7.50e- -06 2.000 5000.
13. CH~ + H ~ C H 2 + H2 9 . 0 0 e - - 13 0.000 15100.
14. CH~O + M ~ CH~O + H + M 1.00e- - 14 0.000 25000.
15. C H ~ O H + M ~- C H 2 0 + H + M 1 . 0 0 e - - 14 0.000 25000.
16. C H ~ O + H --2- O H i O + H 2 2.00e- -13 0.000 O.
17. C H 2 O H + H ~- C H ~ O + H~ 2.00e--13 0.000 O.
18. C H 3 0 + O H ~- C H 2 0 + H ~ O l.OOe -- 13 0.000 O.
19. C H 2 0 H + O H ~- C H ~ O + H ~ O 1.00e--13 0.000 O.
20. CH~O + O ~ CH~O + OH 1.00e+ 13 0.000 O.
21. CH~OH + O ~ CH~O + OH 1.00e+13 0.000 O.
22. C H 3 0 + 02 ~- C H 2 0 + H02 6.30e+10 0.000 2600.
23. C H 2 0 H + 02 ~- C H 2 0 + H 0 2 1.48e+13 0.000 1500.
24. CH2 + H ~- C H + H2 1.00e+ 18 -1.560 O.
25. CH2 + O H . m - C H + H ~ O 1.13e+07 2.000 3000.
26. CH2 + O H ~ CH~O + H 2.50e+ 13 0.000 O.
27. C H + 02 ~ H C O + O 3 . 3 0 e + 13 0.000 O.
28. CH+O~ CO+H 5.70e+13 0.000 0.
29. C H + O H ~- H C O + H 3.00e+13 0.000 O.
30. C H + CO~ ~- H C O + C O 3 . 4 0 e + 12 0.000 690.
31. CH+H~- C+H2 1.50e+14 0.000 O.
32. C H + H 2 0 +m_ C H 2 0 + H 1.17e+15 -0.750 0.
33. C H + C H ~ O ~- C H ~ C O + H 9.46e+13 0.000 -515.
34. CH+C~H~ ~- C ~ H = + H 1.00e+14 0.000 o.
35. CH+CH2 ~ C2H2+H 4.00e+ 13 0.000 0.
36. C H + CH~ ~ C~H~ + H 3 . 0 0 e + 13 0.000 O.
37. C H + CH4 ~- C2Hq + H 6 . 0 0 e + 13 0.000 O.
38. C + 0 2 ~- C O + O 2.00e+ 13 0.000 O.
39. C + O H ~- CO + H 5.OOe + 13 0.000 0.
40. C + C H ~ ~-- C~H2 + H 5.00e+13 0.000 0.
A P P E N D I X A. (continued)
Reaction Mechanism Rate Coefficients In Form k," = AT~ e x p ( - E / R T ) .
Units are moles, cubic centimeters, seconds, Kelvins and calories/mole.
REACTION A fl E
A P P E N D I X A. ( c o n t i n u e d )
Reaction M e c h a n i s m R a t e Coefficients In F o r m k! = A T 0 e x p ( - E / R T ) .
U n i t s are moles, cubic c e n t i m e t e r s , s e c o n d s , K e l v i n s a n d calories/mole.
REACTION A ~ E
A P P E N D I X A. (continued)
Reaction Mechanism Rate Coefficients In Form k t = A T a e x p ( - E / R T ) .
Units are moles, cubic centimeters, seconds, Kelvins and calories/mole.
REACTION A fl E
A P P E N D I X A. ( c o n t i n u e d )
Reaction M e c h a n i s m R a t e Coefficients In F o r m k! = A T ~ e x p ( - E / R T ) .
U n i t s are moles, cubic c e n t i m e t e r s , seconds, K e l v i n s a n d c a l o r i e s / m o l e .
REACTION A /~ E
A P P E N D I X A. (continued)
Reaction Mechanism Rate Coefficients In F o r m k/ = A T a e x p ( - E / R T ) .
Units are moles, cubic centimeters, seconds, Kelvins and calories/mole.
REACTION A ~ E
A P P E N D I X B.
Reaction Mechanism Rate Coefficients In Form ki = A T ~ e x p ( - E / R T ) .
• Units are moles+ cubic centimeters, seconds, Kelvins and calories/mole.
REACTION A /3 E
A P P E N D I X B. ( c o n t i n u e d )
Reaction M e c h a n i s m Rate Coefficients In F o r m k I = A T a e x p ( - E / R T ) .
U n i t s are moles, cubic c e n t i m e t e r s , seconds, Kelvins and calories/mole.
REACTION A fl E