Vous êtes sur la page 1sur 13

0 + NNH: A Possible New Route for NOx Formation in Flames

JOSEPH W. BOZZELLI* and ANTHONY M. DEANt


Corporate Research Labs, Exxon Research and Engineering Co., Annandale, New Jersey 08801

Abstract
We propose a new high temperature pathway for NO formation that involves the reaction of NNH with oxygen atoms. This reaction forms the HNNO* energized adduct via a rapid combination reaction; HNNO* then rapidly dissociates to NH + NO. The rate constant for 0 + NNH = NH + NO is calculated via a QRRK chemical activation analysis to be 3.3 X IOl4 exp(+510/T) cm3 mol-' s-l.This reaction sequence can be a n important or even major route to NO formation under certain combustion conditions. The presence of significant quantities of NNH results from the reaction of H with Nz. The H + Nz = NNH reaction is only ca. 6 kcal/mol endothermic with a relatively low barrier. The reverse reaction, NNH dissociation, has been reported in the literature to be enhanced by tunneling. Our analysis of NNH dissociation indicates that tunneling dominates. We report a two-term rate constant for NNH dissociation: 3.0 X 10' + [M]{1.0 X 10'3T05exp(-1540/T)} s-'. The first term accounts for pressureindependent tunneling from the ground vibrational state, while the second term accounts for collisional activation to higher vibration states from which tunneling can also occur. ([MI is the total concentration in units of mol ~ m - ~Use . ) of this dissociation rate constant and microscopic reversibility results in a large rate constant for the H + Nz reaction. As a result, we find that NNH = H + Nz can be partially equilibrated under typical combustion conditions, resulting in NNH concentrations large enough for it to be important in bimolecular reactions. Our analysis of such reactions suggests that the reaction with oxygen atoms is especially important. 0 1995 John Wiley & Sons, Inc.

Introduction
Much of the interest in the chemistry of nitrogen compounds is directly traceable to the important role of NOx in our environment. Increasingly stringent environmental regulations require more sophisticated approaches for the control of NOx during combustion. Bowman [ l l has recently presented a discussion of current control requirements as well as projected future technologies. Detailed chemical modeling plays a major role in the evaluation and implementation of such technologies. Such models usually incorporate several pathways to NOx production in combustion. These include: (1)Thermal NO2 121; (2) Prompt NO3 [31; (3) NzO mechanism [4,51; and (4) fuel nitrogen [61. In this article we introduce an additional chemical reaction path, 0 + NNH = NO + NH, to be included in NO production models. It appears to be an important contributor to NOx production in a number of cases, such as flame fronts and other regimes where relatively high concentrations of H and 0 are present. The kinetic significance of NNH was first suggested by Miller et al. [71. They proposed that it is a dissociation product from cis- or trans-HNNOH, which
*Chemical Engineering and Chemistry, New Jersey Institute of Technology t To whom correspondence should be addressed. International Journal of Chemical Kinetics, Vol. 27, 1097- 1109 (1995) CCC 0538-8066/95/111097-13 0 1995 John Wiley & Sons, Inc.

1098

BOZZELLI AND DEAN

is an isomer of the HzNNO adduct formed by combination of NH2 and NO. This suggestion sparked a considerable amount of theoretical work [8- 141. Although the dissociation reaction
(1)

NNH=Nz

+H

is exothermic by ca. 6 kcal/mol, calculations [11,12,14] indicate there is a barrier of 6-8 kcal/mol. One might expect rapid dissociation with such a low barrier. This dissociation is predicted to be further enhanced by tunneling, where calculations [121 suggest a lifetime of only ca. 3 X lop9 s for the ground vibrational state and much shorter lifetimes for vibrationally excited levels. This very short lifetime might lead one to believe that any bimolecular reactions of NNH would be unimportant relative to dissociation. However, if tunneling is important for NNH dissociation, it must also be important for the reverse reaction, H addition to N2. Given the high concentration of N2 in a typical combustion environment, this H + N2 reaction could be fast, producing a partially-equilibrated system. This is an important consideration, because, in spite of a very large dissociation rate, the NNH concentration might not become vanishingly small. For example, in air at 1800 K and 1 atm, the equilibrium ratio [NNH]/[H] is in air. As a result of At 1200 K and 1 atm, this value is ca. 1 x ca. 2 x this equilibrium, there could be conditions where [HI is high enough for it to play a role in bimolecular reactions. In general, one expects that NNH is more likely to be equilibrated in high temperature systems in the presence of air (N2). The NNH formation rate increases with temperature, since the hydrogen atom concentration increases here as well. This leads to a faster rate of approach to equilibrium as well as higher equilibrium concentrations. In this article we first estimate the NNH dissociation rate constant to obtain (via microscopic reversibility) the rate constant for addition of H to Na. Use of this value in a detailed kinetic model indicates that the reaction can be partially equilibrated under flame conditions. As a result, we explored some of the consequences of this equilibrium by examining the likely bimolecular reactions of NNH. Our conclusion is that the 0 + NNH reaction can be a significant route to NO under certain conditions.

Estimation of the Dissociation Rate Constant for NNH


We estimate the rate constant for NNH dissociation by using the barrier height and energy-dependent tunneling rate constants computed by Koizumi et al. [121. We assume two components to the rate: (1)a normal contribution due to thermal activation with a barrier height of 7.4 kcal/mol, corresponding to that calculated by Koizumi et al.; and (2) a tunneling contribution, using the rates calculated by Koizumi et al. for the various vibrational levels coupled to an assumed equilibrium distribution. The normal contribution was computed with our QRRK formalism using the parameters in Table I. The calculations of Koizumi et al. indicate tunneling is very fast. Only the ground vibrational state (000) and the first excited bend (001) and have lifetimes greater than 1ps. They estimate those lifetimes to be 3 X 2 X lo- s, respectively. A high pressure value obtained from the sum of these contributions is:

ky

4.1

109T1.13exp(-2610/T) spl

The tunneling contribution is found to dominate at all temperatures.

POSSIBLE NEW ROUTE FOR NOx FORMATION

1099

TABLE I. NNH
Reaction

N2

+H
E, (kcal/mol)
5.0, 7.4 12.0 = 3999 cm (0.044)

A (s- or cm3/mol-s)
3.2312 9.6312 v 2 = 2080 cm-l (2.10) e l k = 71.

k1 N N H = N z + H k-1 H + Nz =H NNH*: v l = 841 cm-l (degen u = 3.8 d Nz collider: ( A E ) = -290 cm-1

0.85)

v3

k l : based on A-1 and microscopic reversibility; barrier = 5.0 kcal to account or tunneling, and tunneling (cf. text). k - 1 : A based on H + CzH2; E adjusted; cf. text. * Based on 3-frequency fit to heat capacity, with the degeneracy constrained to equal 3N-6.

7.4 w/o

Analysis of the falloff behavior is complicated since there is a pressure-independent term (due to tunneling from the lowest vibrational state) that serves as a lower limit to the rate constant. The tunneling rate constants of Koizumi et al. for all levels above the ground state are larger than collisional stabilization values, even when using the strong collision assumption, for pressures below 2 atm. This suggests the system is near the low-pressure limit for many conditions of interest, and that collisional activation is rate-limiting. Assuming that the thermal activation rate can be approximated by considering activation to the first excited vibrational level (1071 cm- above the ground state), the dissociation rate constant can be written
hl =

10

+ [MI (1.0 X

1013T0.5 exp(-1540/T)} s-

In this expression, the first term is the pressure-independent term for tunneling from the ground state and the second accounts for collisional activation; [MI is the total concentration in units of moles cmp3. We tentatively suggest use of this tunneling value for inclusion into kinetic models. This expression yields a dissociation rate constant of ca. 1.2 X lo9 s- at 1 atm and 1000 K and 1.6 X lo9 s- at 2000 K. The dissociation rate constant obtained above is obviously contingent upon several assumptions. The most important involve the details of the potential energy surface for NNH. However, the very recent high-level calculations by Walch et al. [141 are in excellent agreement (within 0.1 kcal/mol) with those used here [121. We introduced additional uncertainty via use of an approximate treatment to describe the effect of pressure. In this context, we used an alternative method to estimate the dissociation rate constant to give credibility t o the tunneling value obtained above. This alternate method used a QRRK treatment of NNH dissociation, assuming that the reaction occurs via thermal activation, but with use of a lower barrier to accommodate tunneling. This QRRK approach has been applied to a variety of systems [15-191 where there were sufficient experimental results to evaluate its predictions. Its success in these applications shows that this approach provides a simple framework by which the effects of temperature and pressure can be readily understood and evaluated. The reader is referred to a review for further discussion of the QRRK method [201. We estimate the high pressure rate constant for NNH dissociation via the reverse reaction, using an A-factor for H addition to Nz equal to H addition to CzH2. A barrier of 5 kcal/mol was used to give a room temperature value for the high pressure rate constant consistent with the calculated lifetime for the ground vibrational state of NNH [121. This is lower than the 7.4 kcal/mol used for the tunneling case, but one

1100

BOZZELLI AND DEAN

might expect a lower barrier since we did not correct for tunneling. Other than the barrier height, the parameters used are the same as used above. Our QRRK results suggest that this reaction is a t the low pressure limit for conditions of interest. We estimate: (A) k = 1.3 x 1014Tp0.11ex p(-2507/T) cm3 mol-' s-l, M = N2. At 1 atm, the effective first-order rate constant is ca. 6 X lo7 s-l at 1000 K and 1 X lo8 splat 2000 K, or about a factor of ca. 20 lower than estimated using the tunneling calculations. Both approaches yield rate constants much higher than the value used by Miller and Bowman [21] ( k l = 1 x lo4 s p l , independent of temperature), who assigned this value to fit the kinetic experiments for Thermal DeNOx. In more recent modeling work, Glarborg et al. [a21 used h l = 1 X lo6 spl for the dissociation but observed some deterioration in the fit to Thermal DeNOx data. We find it difficult to reconcile the initial Miller-Bowman value, or even the recent 1 X lo6 s-l estimate, with the theoretical results on NNH. The calculated barriers for dissociation are simply too low to justify such a long lifetime. To test the effect of uncertainties in barrier height on our rate constant estimates, additional calculations were done with a barrier increased from 5 kcal/mol to 15 kcal/mol. Such a n increase is well beyond the expected uncertainty in the calculated barrier. Even with the higher barrier, the rate constant was found t o be ca. 1 X lo6 s p l at 1000 K and ca. 1 X lo7 spl a t 2000 K, still well above the Miller-Bowman value. Estimation of Bimolecular Rate Constants for Reactions of NNH One can estimate the importance of various collision partners with NNH by considering the term h x [ X ] ,where kx is the rate constant for the reaction of species X with NNH. The potential reactant in highest concentration is usually 0 2 . We used a QRRK approach to estimate this rate constant. We considered formation of the HNNOO adduct, followed by a n intramolecular hydrogen transfer to form HOONN, which can dissociate to either HO2 + N2 or N2O + OH.

NNH
(3)

0 2 5

HNNOO

NNOOH N20

OH

The HNNOO well depth is uncertain; we estimate it t o be in the range 16-22 kcal/mol. We selected a 22 kcal/mol well depth. Use of a shallower well would significantly reduce the rate of product formation. The A-factor for the since it hydrogen transfer, to form NNOOH, is estimated to be ca. 4 X involves loss of the HNN-00 rotor. We chose a barrier for isomerization to be ca. 21 kcal/mol, by analogy to that used by Lin et al. [231. Partition of NNOOH to the two product channels was estimated at 4:l favoring HO2 + N2, following our estimates of the respective transition state entropies. These calculations give the following rate constants: k2 = 1.2 x 1012~-0.34ex p ( - 7 5 / ~ ) cm3 mol-l s-l
k3 =

2.9

1011Tp0.34exp(-75/T) cm3 mol-l s-'

POSSIBLE NEW ROUTE FOR NOx FORMATION

1101

This value for the faster channel (HOz + Nz) is only ca. 9 x 1O1O cm3 mol-' s-l at 2000 K, suggesting a pseudo-first-order rate constant of ca. 1 x lo5 s-l in air at 1 atm for the reaction of NNH with 0 2 . The above analysis indicates that the rate of the overall reaction NNH + O2 = N2 + HOz is quite slow. At first glance this is surprising, given that the reaction appears to be a simple, very exothermic hydrogen abstraction. However, extensive analysis of analogous reactions suggest that this is not a direct hydrogen transfer, but rather the result of an addition/isomerization/eliminationsequence. Reactions of the form R + 0 2 = olefin + HOa have received considerable attention in the literature, and there is a growing consensus that these are not direct reactions. Perhaps the most compelling experimental evidence is that of Kaiser [241, who showed that the ethylene yield in the C2Hs + 0 2 reaction decreases dramatically with increasing pressure. He interprets this in terms of adduct formation, with the increasing pressure stabilizing the energized adduct prior to isomerization. There have been several theoretical treatments of these types of reactions [18,19,251. It has been possible to reconcile much of the available data in terms of the adduct formation pathway. It is in this context that we treat the NNH + 0 2 reaction. Even though the rate constant for formation of the HNNOO adduct is large and the pathway to HO2 + N 2 is indeed very exothermic, the rate limiting step is the isomerization reaction. In this system, the relatively shallow well for formation of the adduct puts the barrier for isomerization only 1 kcal/mol lower than the entrance channel. Since the A-factor for dissociation out the entrance channel is higher than that for isomerization, dissociation dominates, leading to the lower rate of NZ + HOz formation. Another similar reaction is that of HCO + 0 2 = CO + HO2. The H-C bond is only ca. 15 kcal/mol, such that the enthalpy change for this reaction is approximately midway between that of CzHs + 0 2 and NNH + 0 2 . In an earlier work [191, we were able to show that a QRRK analysis based on adduct formation lead to rate constants that were consistent with those measured. The rate constant was found to be higher than computed for the NNH + 0 2 case, since the HCO-02 well was deeper. Other potentially important reactants with NNH involve radicals that are formed in the flame front region. H, 0, and OH can reach mole fractions of several percent under these conditions. Thus we estimate rate constants for these reactions. The recombination of H and NNH would form HNNH. Production of NH by breaking the N-N bond is too endothermic to be important. Formation of the HNNH adduct liberates ca. 60 kcal/mol, which is probably less than the barrier for Hz elimination to form N2 and Hz. In addition, the A-factor for this concerted molecular process should be appreciably lower than for redissociation of the adduct to H + NNH. Thus, we do not expect this pathway to be significant. Another channel is the disproportionation reaction:

(4)

NNH

+ H = H2 + N2

Assuming this reaction can be treated as similar to other disproportionations, we would estimate J24 ca. 1 X 10l2cm3 mol-I s-l. This rate is the same as that suggested by Tsang and Hampson [26] for H + CzHs. That reaction is 68 kcal/mol exothermic, 1 1kcal/mol for (4).Although this is a substantial difference, both compared to 1 are so exothermic that it is not obvious that the difference is significant. If the H and T = 2000 K, the pseudo-first-order rate constant for this mole fraction were 1% reaction is ca. 6.1 X lo4 s-'.

1102

BOZZELLI AND DEAN

Recombination with OH to form HONNH is likely to form N2 elimination:

+ HzO via concerted

(5)

NNH

OH ====HONNH

N2

+ H20

This reaction occurs on the same potential energy surface as N H 2 + NO. We have analyzed this in terms of a QRRK analysis [27]. This analysis suggests: k5 = 2.4 x 1022T-2.88exp(-1230/T)cm3 mol-' sP1 At 2000 K, this value is 4 x significantly lower than the 5 x l O I 3 value used by much closer to the Miller-Bowman Miller and Bowman [21]. At 300 K, kg = 3 x value. The decline at higher temperatures is due to fall-off effects as the complex dissociates back out the entrance channel. If the OH mole fraction were 1%and T = 2000 K, the pseudo-first-order rate constant for this reaction is ca. 2.4 X lo5 s-'. There are three likely reaction channels for 0 + NNH (cf. Fig. 1):
(6)

(7)
(8)

+ OH 0 + NNH=NzO + H 0 + NNH = NH + NO
0
=N2

+ NNH

Rate constants for reactions (7) and (8) can be obtained from our analysis of the HNNO reaction system [281. That analysis showed good agreement between the predicted rate constants and the experimental data for all the reaction pathways in both the H + N20 and NH + NO reactions. Application of microscopic reversibility to our calculated rate constants k-7 and k - 8 yields: k7 = 1.4 X 1014T-0.40exp(-240/T)cm3 mol-' s-' 3.3 x 1014~-0.23ex p(+510/T) cm3 mol-' s-' In that analysis, we suggest that 0 + NNH was a plausible high temperature product of the NH + NO reaction, and that the calculated temperature dependence for this channel was consistent with the high temperature measurements of Mertens et al. [291. Specifically, Mertens et al. measured an activation energy of 13 kcal/mol for the rate constant for the overall NH + NO reaction over the temperature range 2220-3350 K. In the NH + NO reaction, there are three likely product channels: H + NzO, Nz + OH, and 0 + NNH. Only the latter is endothermic, and several analyses [28,30] suggest the other two channels have rate constants that decrease with temperature. Thus the measured temperature dependence suggests that 0 + NNH is the dominant channel in the Mertens experiments. This would imply, via microscopic reversibility, that k8 ca. 1.3 X l O I 4 cm3 mol-I s-l, a value only a factor of 2 above our estimate. Miller and Melius [30] also pointed out that this channel might be responsible for the observed temperature dependence. The Mertens et al. experiments Hcan be very rapid; it is offer convincing evidence that the reaction between 0 + " encouraging that our QRRK estimate for the rate constant is in reasonable agreement with their data. Note that all three product channels of the 0 + NNH reaction are exothermic. The unimolecular rate constants for HNNO* are sufficiently large that stabilization reactions are unimportant at typical pressures of interest. Our analysis suggests that
k8 =

POSSIBLE NEW ROUTE FOR N G FORMATION

1103

NH + NO
100

40

20

i
N2 + OH

Figure 1. Potential energy diagram for the reaction 0

+ NNH.

ks ca. 7 X 1013 at 2000 K. In our earlier work [281, we showed that one expects an unusually low A-factor for dissociation of HNNO to H + NzO since the value for the equilibrium constant connecting these species is unusually low. Much of this effect can be attributed to the low entropy of H. As a result, k7 is about an order of magnitude lower than k g , even though (7) is more exothermic. k6 is even lower, since isomerization through a tight transition state is required to access this channel. Our QRRK estimate gives:
k6 =

1.7 x ~ o ~ ~ exp T( - 2 5~ 0 / T.) cm3 ~ ~mol- sP1

At 2000 K, this value is ca. 1 X lo1, significantly lower than the other channels. We find for this system that the large A-factor associated with the N-N bond breaking step to produce NH + NO results in Reaction ( 8 ) being the fastest for this chemically-activated process, even though it is the least exothermic pathway. It is interesting to note that our calculated value for k6 is comparable to what one would have estimated by assuming this reaction as a direct disproportionation reaction. It raises a broader issue, well beyond the scope of this article, as to whether it might be useful to treat disproportionation reactions in the context of chemicallyactivated systems. Disproportionation reactions, which involve a very exothermic hydrogen transfer from one radical to another, usually have rate constants appreciably smaller than recombination reactions. This behavior is consistent with formation of an energized adduct followed by a concerted elimination reaction. Even if there is sufficient energy in the adduct for the elimination, the low A-factor associated with this reaction, as compared to redissociation, would lead to a lower rate constant. In certain cases, one might expect fast rate constants even when there is adduct formation. For example, the reaction H + HCO = Ha + CO is very fast [311, with k = 9 X 1013cm3 mol- s-l. Here the adduct would be CH20*, with ample energy to undergo Hz elimination. However, since this is a 1,l elimination, rotors are not tied up in the transition state and one might expect a higher A-factor. Even though adduct formation is involved, one might not expect to see significant pressure effects since all of the unimolecular reactions would frequently be faster than stabilization reactions.

1104

BOZZELLI AND DEAN

Of the various NNH reactions considered, it appears that the most important is + NNH = NH + NO. Assuming 1%0, the pseudo-first-order rate constant is ca. 4 X lo6 s-l at 2000 K. This value is at least an order of magnitude higher than the other reactions considered. We expect no pressure dependence for any of these radical + NNH chemically-activated reactions.

New Pathway for Fixation of Molecular Nitrogen


We have recently completed the analysis of a large number of reactions involving nitrogen compounds [27]. The approach was to attempt to specify the most plausible rate constant assignments for the reactions, without regard to whether these values gave the best agreement to experimental observations for complex systems. These assignments were used to construct a mechanism in an attempt to describe nitrogen chemistry as well as interactions between nitrogen and hydrocarbon compounds. This mechanism includes all of the reactions discussed above and is described in detail in ref. [27]. We have compared the predictions of this Bozzelli-Dean model (BD) to those obtained using the Miller-Bowman model [all (MB). The MB model has been widely used to predict nitrogen chemistry, and such a comparison serves to highlight differences between our mechanism and a conventionalone. The BD model uses the hydrocarbon mechanism of Miller and Bowman, so as to focus only on differences in the nitrogen kinetics. Figure 2 illustrates the predicted effect of temperature on NO production during the combustion of a stoichiometric mixture of methane and air using both the BD and MB mechanisms. The results are for a pressure of 1 atm, with the temperature held fixed at either 1800 K or 2400 K. Although the results of the two models are in reasonable agreement at 2400 K, the BD model predicts an order of magnitude higher concentration of NO at 1800 K. The reasons for this difference can be seen in the analysis of the NO production rates in the BD mechanism at 1800 K, as seen in Figure 3. The major contributor to NO production in the BD mechanism is the reaction 0 + NNH = NH + NO, a reaction not considered in the MB mechanism. This reaction becomes important immediately after ignition when the oxygen atom mole fraction rapidly rises to exceed 1%. In this high [OI region, the NNH = Nz + H reaction was found to be equilibrated. This equilibrium is achieved via the reverse reaction as the hydrogen atom concentration also builds up in the flame front. The equilibrium concentration of NNH is sufficiently high to undergo significant reaction with atomic oxygen. Our equilibrium calculations were based upon the NNH thermochemistry of Melius [32]. However we confirmed that these assignments are consistent with those calculated from the potential energy surface of Koizumi et al. [121. We performed several calculations to explore the effect of specific changes to the BD model. We focused on the 1800 K case since we saw a much larger difference between the BD and MB models there. Calculations using a lower value for the NNH dissociation rate constant, i.e., our QRRK estimate eq. (A), which does not include the tunneling contribution, also indicate that NNH = N z + H is equilibrated, with a predicted NO profile virtually identical to that shown in Figure 2. This insensitivity of NNH concentration to a decrease of over an order of magnitude in the rate constant emphasizes an important aspect of this reaction. It is controlled by thermodynamics, not kinetics, provided that the rate constant is reasonably large. However, use of the very small value of lo4 s- used by Miller and Bowman, more than

POSSIBLE NEW ROUTE FOR NOx FORMATION

1105

t T = l 8 0 0 K , BD t T = 2 4 0 0 K , BD +T=l800K, +T=2400K.

MB
MB

-8

0.0

0.1
t/ms

0.2

0.3

Figure 2. Predicted effect of temperature on NO profiles using the BD (this work) and MB (Miller-Bowman) models. Results are shown at 1800 K and 2400 K for a stoichiometric methane/air mixture at 1 atm which reacts in simple plug flow with the temperature held constant.

five orders of magnitude lower than that used here, eliminates this NO production route. This is illustrated in Figure 4,where the BD mechanism is used with only one change, use of the Miller/Bowman value for NNH dissociation. This calculation yields results virtually identical to the Miller-Bowman mechanism. This rate constant is sufficiently small that NNH cannot be formed rapidly enough, via the H + N2 reaction, to react with the oxygen atoms. Also in Figure 4, we show the effect of using a dissociation rate of lo7 s-'. This predicted NO profile is qualitatively similar to the BD results. In the previous section we pointed out that our estimate for several bimolecular reactions of NNH differed substantially from those of other workers. We tested the sensitivity of the calculations to these assignments. Calculations using high values for the rate constants for H + NNH = N2 + Ha and 0 2 + NNH = HO2 + N2, equal t o that used for 0 + NNH = NH + NO, yield NO concentrations virtually indistinguishable from the BD model, as shown in Figure 4 (fast abstraction case). This result can be understood in the context that the NNH concentration is continually replenished under these conditions via the very fast H + N2 reaction, even with the additional sinks for NNH included. Thus the rate of NO production via 0 + NNH is unaffected. This possibility of an equilibrium connecting NNH and Nz has some interesting consequences. It suggests a new pathway in which molecular nitrogen can be "fixed". As discussed above, NNH can react with 0 to form NO, so that NNH formation from N2 introduces a new route from N2 to NO. One might expect a contribution from this reaction under circumstances where the concentration of atomic hydrogen is only moderately high, since the rate constant for H + N2 = NNH can be quite large and

1106

BOZZELLI AND DEAN

8E-06

-&-Iota1

rate

6E-06

- U -HNO+M=H+NO+M - - *--NH+NO=NNH+O
---O--HNO+H.HP+NO --tNH+O-NO+H ---X---N+OH=NO+H

4E-06
r

- +- HNO+O-OH+NO

u)

- 2E-06

E,

E .
c

a ,

OE+OO

+
0
\

-2E-06

-4E-06
0 0.1

0.2
tlrns

0.3

Figure 3. Rate analysis using the BD mechanism (this work) for NO production at 1800 K in a stoichiometric methane/air mixture a t 1 atm. The most important contributors to NO production and/or decay are shown along with the total rate.

"21 is usually high. The consequence of this reaction is increased NO production at lower temperatures, as predicted in Figure 2. One obvious circumstance where this reaction might be especially important is in hydrogen flames, where there are no hydrocarbon radicals to form Prompt NO. One would expect increased formation of NO in H2 flames at lower temperatures than one would predict with the Zeldovich mechanism. Another consequence might be a difference in Thermal DeNOx kinetics depending on whether the experiments are carried out in helium or nitrogen as the diluent. Indeed, we have done calculations [27] that show slight differences in reactivity when nitrogen is substituted for helium. These differences are limited to a temperature near 1170 K, at the lower end of the DeNOx temperature window, where the nitrogen diluent case shows that a slightly higher temperature is needed to initiate the reaction. This behavior can be traced to the extreme sensitivity of the Thermal DeNOx system to the ratio of branching versus termination at the onset of reaction. The perturbation of the hydrogen atom concentration by the NNH = N2 + H reaction, although small, is sufficient to slow down the system. At higher temperatures the calculations indicate that the change in diluent has little effect. The net effect is that the temperature window is narrowed for the nitrogen case. For typical conditions the change is only ca. 20 K, perhaps too small to be seen experimentally. Nevertheless, the calculations suggest that a most interesting experiment would be to study the

POSSIBLE NEW ROUTE FOR N% FORMATION

1107

-4

1
-8 0.05

+ME
+BD +-ED +ED with k=iE+4for NNH diss with k=i E+7 for NNH diss.

with fast abstraction

0.10

0.15

0.20

0.25

0.30

tlms
Figure 4. Predicted effect of various model assumptions on NO profiles a t 1800 K for a stoichiometric methane/air mixture at 1 atm which reacts in simple plug flow with the temperature held constant. BD (this work); MB (Miller-Bowman); ( A ) BD model using Miller-Bowman value for NNH dissociation ( h = 1 X lo4 s-'); + BD model using k = 1 x lo7 5 - l for NNH dissociation; and ( 0 ) BD model using h = 7 X 1013cm3 mo1-l s-l for H + NNH = Nz + Hz and 0 2 + NNH = Nz + HOz.

effect of this diluent change in the temperature regime near the lower end of the Thermal DeNOx temperature window. We also considered the possibility of other radicals adding to N2. If these adducts are sufficiently stable, they can react with atomic oxygen to form NO analogous to NNH. In general, we found that the energetics of some of the most likely candidates were not favorable. For example, the NNOH adduct is ca. 50 kcal/mol less stable that OH + N2, meaning that the equilibrium concentration of NNOH would be much too small to make it a n important partner in bimolecular reactions with atomic oxygen. For CH3, even though the CH3NN adduct is only ca. 17 kcal/mol higher in energy than CH3 + N2, the equilibrium ratio [CH3NNl/[CH31 is at least three orders of magnitude lower than [NNHl/[Hl, even at 2000 K. The only other radical likely to be present in high concentrations is atomic oxygen, and we are unaware of any information on the heat of formation of the triplet NNO species. If this species were to have a heat of formation of ca. 70 kcal/mol or less (ca. 50 kcal/mol higher than ground state nitrous oxide), it would be a candidate.

Conclusion
NNH dissociation to Nz + H is expected to be rapid under all conditions. However, the hydrogen atom mole fraction is often high (>0.1%) under high temperature reaction conditions, and, because the reaction of NNH = H + NZis nearly thermoneutral, NNH formation from H + N2 is also very rapid. The result is equilibrium, or

1108

BOZZELLI AND DEAN

near-equilibrium, concentrations of NNH which are high enough to react with other important radical species. This conclusion is not contingent upon a precise assignment of the NNH dissociation rate. Use of a value over an order of magnitude smaller than that obtained from a tunneling analysis is still sufficient to permit equilibration. This suggests that bimolecular reactions of NNH should be included in combustion models. Our analysis of several NNH bimolecular reactions suggests that the most important is that of NNH with 0 atoms. The chemically-activated HNNO" adduct decomposes primarily to NH + NO before it is stabilized. Entropic effects drive this channel relative to other, lower energy, channels. This reaction provides a new pathway to NO and should be included in combustion kinetic models. We show that this channel can increase the predicted NO concentration by an order of magnitude in calculations of methane/air oxidation at 1800 K. These findings amplify the need for additional analysis of bimolecular reactions of NNH.

Acknowledgment
AMD appreciates discussions with Dr. William Green (Exxon) and Dr. Gregory Smith (SRI International) about aspects of estimating the dissociation rate in the presence of tunneling. JWB acknowledges many helpful conversations with Prof. L. Krasnoperov of NJIT.

Bibliography
[l]C. T. Bowman, Twenty-fourth Symposium (International) on Combustion, Combustion Institute, Pittsburgh, pp. 859-878, 1992. [21 Y. B. Zeldovich, Acta Physicochim. URSS, 21,577 (1946). [3] C. P. Fenimore, Thirteenth Symposium (International) on Combustion, Combustion Institute, Pittsburgh, pp. 373-380, 1970. [4] J. Wolfrum, Chemie Zngenieur Technik, 44, 656 (1972). [51 P.C. Malte and D. T. Pratt, Combust. Sci. Tech., 9, 221 (1974). [61 C.P. Fenimore, Comb. Flame, 26, 249 (1976). [71 J.A. Miller, M.C. Branch, and R. J. Kee, Combust. Flame, 43, 81 (1981). [81 C. F. Melius and J. S. Binkley, Twentieth Symposium (International) on Combustion, Combustion Institute, Pittsburgh, pp. 575-583, 1984. [91 R. G. Gilbert, A. R. Whyte, and L.F. Phillips, Znt. J. Chem. Kinet., 18,721 (1986). 1101 L.F. Phillips, Chem. Phys. Lett., 135, 269 (1987). [ l l ] S.P. Walch, R. J . Duchovic, and C.M. Rohlfing, J. Chem. Phys., 90, 3230 (1989). [121 H. Koizumi, G.C. Schatz, and S.P. Walch, J. Chem. Phys., 95, 4130 (1991). [131 S.P. Walch, J. Chem. Phys., 99, 5295 (1993). [141 S.P. Walch and H. Partridge, Chem. Phys. Lett., 233, 331 (1995). [151 P. R. Westmoreland, J. B. Howard, J . P. Longwell, and A. M. Dean, AIChE Journal, 32, 1971 (1986). [161 A. M. Dean and P. R. Westmoreland, Znt. J. Chem. Kinet., 19,207 (1987). [171 J . W . Bozzelli and A.M. Dean, J. Phys. Chem., 93, 1058 (1989). I181 J . W. Bozzelli and A.M. Dean, J. Phys. Chem., 94, 3313 (1990). [191 J . W. Bozzelli and A.M. Dean, J. Phys. Chem., 97, 4427 (1993). [201 A.M. Dean, J. W. Bozzelli, and E. R. Ritter, Combust. Sci. and Tech., 80, 63 (1991). 1211 J. A. Miller and C. T. Bowman, Prog. Energy Combust. Sci., 15,287 (1989). [221 P. Glarborg, K. Dam-Johansen, J.A. Miller, R. J. Kee, and M. E. Coltrin, Znt. J. Chem. Kinet., 26, 421 (1994). [231 M. C. Lin, Y. He, and C.F. Melius, Int. J. Chem. Kinet., 24, 489 (1992). 1241 E.W. Kaiser, J. Phys. Chem., 99, 707 (1995). [251 A. F. Wagner, I. R. Slagle, D. Sarzynski, and D. Gutman, J. Phys. Chem., 94, 1853 (1990). [261 W. Tsang and R. F. Hampson, J. Phys. Chem. Ref: Data, 15,1087 (1986).

POSSIBLE NEW ROUTE FOR NOx FORMATION

1109

[271 J. W. Bozzelli and A.M. Dean, Combustion Chemistry of Nitrogen, in Combustion Chemistry, 2nd ed., W.C. Gardiner, Jr., Ed., in press. [281 J. W. Bozzelli, A. Y. Chang, and A.M. Dean, Twenty-fifth Symposium (International) on Combustion, Combustion Institute, Pittsburgh, pp. 965-974, 1994. [291 J . D. Mertens, A.Y. Chang, R. K. Hanson, and C. T. Bowman, Int. J. Chem. Kinet., 23,173 (1991). [301 J.A. Miller and C. F. Melius, Twenty-fourth Symposium (International) on Combustion, Combustion Institute, Pittsburgh, pp. 719-726, 1992. [311 D. L. Baulch, C. J . Cobos, R. A. Cox, P. Frank, G. Hayman, T. Just, J.A. Kerr, T. Murrells, M. J. Pilling, J. Troe, R. W. Walker, and J. Warnatz, Combust. Flame, 98, 59 (1994). [32] C. F. Melius, BAC-MP4 Heats of Formation and Free Energies, 1993.

Received May 11, 1995 Accepted May 31, 1995

Vous aimerez peut-être aussi