- Email: [email protected]
Inhibition of the first limit of the hydrogen-oxygen reaction by ethyl bromide
INHIBITION OF THE FIRST LIMIT OF THE H Y D R O G E N - O X Y G E N REACTION BY ETHYL B R O M I D E N. ARMST]~ONG* AND 1~. F. SIMMONS
Department of Chemistry, The University of ,~Ianchester Institute of Science and Technology, Manchester, England The inhibition of the first limit of the hydrogen-oxygen reaction by ethyl bronfide has been studied between 460 ° and 520°C, over a wide range of mixture composition, using clean Pyrex and silica reaction vessels. It is shown that the inhibition arises through the reactions: H-}-C2HsBr = HBr +C2H5,
(5)
C.oHs-}-O2 = C 2H4@HO2,
(7)
The surface-destruction constants for the removal of hydrogen and oxygen atoms by the surface have been determined from a detailed analysis of the uninhibited-explosion pressures, and these values have been used to calculate a value of k5 from each inhibited-explosion pressure. These values are independent of mixture composition and vessel diameter, and an Arrhenius plot gives logk~/cm a mole-~ s-~ = (13.93±0.24) - (5470±830)/4.576T. No evidence was obtained for any reaction of hydroxyl radicals or oxygen atoms with the ethyl bromide, and the implications of these results to the problem of flame inhibition by halogen compounds is considered.
Alkyl halides are widely used as fire-extinguishing agents, and it is believed that their action is chemical in nature, in that they preferentially react with chain centers which are important for the propagation of the flame to give chain termination. I A number of studies have been made in an attempt to elucidate which chain center (or centers) is removed by the inhibitor, and arguments have been advanced for the preferential removal of both hydrogen atoms 2'a and hydroxyl radicals. 4 The aim of the present work was to compare the rates of these two types of processes directly, using ethyl bromide as inhibitor, from its effect on the first limit of the thermal reaction between,hydrogen and oxygen. Both hydrogen atoms and hydroxyl radicals are important chain centers in this reaction, so that a kinetic study of the inhibition * Present address: I . C . I . Mond Division, The Heath, Runeorn, Cheshire, England.
should show whether one or both of these chain centers react with the ethyl bromide. The inhibition of the first limit was chosen for this purpose, as a preliminary study of the inhibition of the second limit showed that ethyl bromide was decomposing, with a half-life of 15 s at 520°C, into ethylene and hydrogen bromide. The determination of the second explosion pressure requires the gases to be mixed in the reaction vessel at a pressure above the limit, and the pressure is then reduced at a controlled rate to reach the explosion boundary. Thus, a considerable proportion of the inhibitor would have decomposed during this manipulation time. In contrast, the first limit is determined by the direct admission of the premixed gases into the evacuated reaction vessel, and explosion occurs almost instantaneously, so that the thermal decomposition of the inhibitor is not a problem under these experimental conditions. It will be shown that, at the first limit,
443
444
OXIDATION AND IGNITION
Ib
ol
5
IO
|
I
15
20
MOLE FRACTION ETHYL BROMIDE ( x l O 3) FIG. 1. The variation of explosion pressure with mole fraction of ethyl bromide. 3.6-cm vessel; temp., 510°C; mole fractions of hydrogen, 0.28; mole fractions of oxygen, • 0.72; • 0.56; • 0.28; • 0.14; • 0.07.
ethyl bromide reacts almost exclusively with hydrogen atoms by Reaction (5), and the rate constant for this reaction has been obtained over the temperature range 460°-520°C. The implication of these results to the problem of flame inhibition is also considered.
Experimental The apparatus was essentially similar to that described earlier3 The first limit was obtained by the conventional method, in which a known pressure of the reactant mixture was admitted to the evacuated reaction vessel, and the occurrence or absence of explosion noted. The uninhibited limits were pale blue in color, and it was necessary to exclude stray light to obtain accurate explosion pressures; but, when ethyl bromide was present, the flashes were yellow and the explosion boundary was easily detected. The limit was usually approached from the explosive side and determined with an accuracy of =t=0.03 torr. Between each individual determination, the reaction vessel was rinsed with hydrogen and evacuated for 3 rain (to a pressure < 10~ torr). Mixtures of hydrogen, oxygen, and nitrogen
were used in this investigation, to permit the effect of an independent variation of the mole fractions of hydrogen and oxygen on the efficiency of the inhibitor to be examined. With a constant mole fraction of hydrogen (x=0.28), the mole fraction (y) of oxygen was varied from 0.72-0.07, while with a constant mole fraction of oxygen (y=0.14), the mole fraction (x) of hydrogen was varied from 0.86-0.07. Most of the determinations were made using cylindrical reaction vessels (3.6 cm i.d. by 20 cm long), but the effect of veasel diameter was examined also at one temperature by using a 1.5-cm-i.d. vessel as well. One difficulty in studying the inhibition of the first limit is that surface-destruction processes are very important at the explosion boundary, and slight changes in the surface can affect the magnitude of the explosion pressure. Thus, throughout the present work, care was taken to ensure that the uninhibited-explosion pressures were reproducible from day to day, and this reproducibility was usually within the experimental error of the determination (4-0.03 torr). If the deviation were outside this experimental error, the limit had usually changed quite markedly and the reaction vessel was replaced by a freshly washed vessel.
INHIBITION OF FIRST LIMIT
445
16
fir
O b--
12
tlJ r~ it0
tD tlJ
a~ 8 Q. Z
O tD
O X
I
I
I
I
I
2 3 4 M O L E FRACTION ETHYL BROMIDE ( x l O 3 ) FIG. 2. The variation of explosion pressllre with mole fraction of ethyl bromide. 3.6-cm vessel; temp., 510°C; mole fraction of oxygen, 0.14; mole fractions of hydrogen, • 0.86; • 0.56; • 0.28; A 0.14; ~ 0.07. Results
A systematic study of the inhibition of the first limit of the hydrogen-oxygen reaction has been made in the temperature range 460 °520°C, over the wide range of mixture eomposition indicated earlier. This range of temperatures was limited by two factors. At 520°C, induction periods of 3-5 s were observed with mixtures with a low value for the ratio [-I-I2-1/[0~-] and, at this temperature, a significant amount of inhibitor would have decomposed after this time, so that it was necessary to disregard the results obtained with such mixtures. In addition, as the temperature was reduced, the tip of the explosion peninsula was approached and only traces of ethyl bromide were required to suppress explosion. This meant that the sensible lower-temperature limit in this work was 460°C. Some typical results are shown ill Figs. ] and 2. The former shows that increasing the mole fraction of oxygen in the mixture reduces the efficiency of ethyl bromide as an inhibitor; a useful indication of its efficiency is the mole fraction required to double the uninhibitedexplosion pressure, and this quantity is almost directly proportional to the mole fraction of oxygen. In contrast, Fig. 2 shows that, even
when the mole fraction of hydrogen is increased from 0.07-0.86, the efficiency of the inhibitor is only slightly affected. These observations strongly suggest that ethyl bromide competes with oxygen in a reaction with hydrogen atoms to give chain termination, and any further consideration of these results wilt be deferred until this suggestion has been examined in more detail.
Discussion
There are two possible reactions of hydrogen atoms with ethyl bromide, namely, H + C2HsBr = H B r + Cells,
(5)
H + C2HsBr = H2+ C2H4Br.
(6)
Reaction (5) is exothermic by ~ 2 0 keal mole-I, whereas Reaction (6) is only slightly exothermic ( ~ 5 kcal mole- I ) . These values should be sufficient to produce enough difference in the activation energies of these two reactions for Reaction (5) to predominate over Reaction (6), and the ethyl radicals produced in the former reaction will react with oxygen to give
446
OXIDATION AND IGNITION TABLE I Comparison of observed and calculated uninhibited-explosionpressures
Temp., 510°C; 3.6-cm clean Pyrex vessel. P(ealc) A =0.132 z
y
O. 28
O. 72 0.56 0.44 0.28 O. 14 0.07 0.14
0.86 0.72 0.56 0.44 0.14 0.07 0.44 O. 14 0.07
P(obs)
1.59 1.69 1.85 2.15 2.84 3.95 2.65 2.52 2.57 2.61 3.42 4.20 0.56 1.56 2.09 2.65 R.m.s. deviation
C=4.0 1.55 1.69 1.84 2.17 2.88 3.98 2.65 2.58 2.58 2.65 3.42 4.20 1.55 2.05 2.57 1.5%
chain termination by Reaction (7)6; C2H5-1-02 = C2H4--~-HO2,
(7)
0H+H2= H20+H,
(1)
I-I+ 02 = OH+O,
(2)
O+H2= OH+H,
(3)
H + O 2 + M = HO2+M,
(4)
H = surface destruction,
(H)
0 = surface destruction.
(0)
If Reactions (5) and (7) are added to the mechanism of the uninhibited reaction,r i.e., Reactions (1)-(4) and the surface destruction of hydrogen and oxygen atoms, the explosion condition becomes
1-- (k4mP/2k2) = (k~/2k2yP ) + E k o / ( & z P + ] c o ) ] + (ksi/2k2y ),
(1)
where P is the explosion pressure, x, y, and i the mole fractions of hydrogen, oxygen, and ethyl bromide, respectively, and m a composite third-body term expressing the efficiency of the particular mixture in Reaction (4). [-Baldwin7 considered the destruction of H and OH at the
A =0.118 C=4.5
A =0.120
1.56 1.69 1.84 2.17 2.84 3.89 2.61 2.55 2.55 2.61 3.39 4.20 1.55 2.06 2.59 1.2%
1.57 1.70 1.85 2.18 2.86 3.92 2.63 2.56 2.57 2.63 3.42 4.19 1.56 2.07 2.60 0.9%
C=4.5
A =0.122 C=4.5 1.58 1.71 1.86 2.19 2.88 3.95 2.65 2.58 2.58 2.65 3.44 4.24 1.56 2.08 2.62 1.1%
A =0.110
C=5.0 1.58 1.72 1.86 2. t9 2.86 3.87 2.62 2.55 2.55 2.62 3.42 4.24 1.57 2.09 2.63 1.1%
surface of the vessel, but the value of ]~1 obtained from the analysis of the results is very close to that expected for k3. It seems certain, therefore, that the second species destroyed at the surface is 0 atoms, rather than OH radicals.] At first-limit pressures, the term k4mP/2k2 is relatively unimportant, but a detailed analysis of tile uninhibited explosion pressures shows that neither of the terms involving kit or ko in Eq. (1) is small enough to be neglected. As a result, there is no simple graphic way in which the validity of Eq. (1) can be verified. Instead, it has been necessary to calculate a value of k5 from each experimental explosion pressure, using the knowns ratio k4/k2, and the surfacedestruction constants for hydrogen and oxygen atoms, obtained from the variatiou of the uniuhibited limit with mixture composition. The value of ks, obtained in this way, is independent of both mixture composition and vessel diameter, which confirms the proposed mechanism. Thus, in the following sections of this paper, the uninhibited limits are considered to obtain the required surface-destruction constants, and these are then used to evaluate k5 from Eq. (1).
:4 nalysis of the Uninhibited Explosion Boundary The general magnitude of the explosion pressures suggests tile destruction of hydrogen
INHIBITION OF FIRST LI MI T
447
TABLE II Comparison of observed and calculated uninhibited-explosion pressures Diameter of reaction vessels, 3.6 cm. 460°C
480°C
P(calc) A = 0.133 x
y
P(obs)*
0.28
0.72 O. 56 0.44 0.28 0.14 0.07 0.14
1.95 2.05 2.29 2.66 3.49 4.91 3.34 3.12 -4.25 5.51 1.89 2.63 3.51
0.86 0.56 0.44 0.14 0.07 0.44 0.56 0.14 0.07 R.m.s. deviation
520°C
P(calc) A = 0.170
C=5.5
P(obs)*
1.94 2.10 2.27 2.66 3.50 4.87 3.37 3.16 -4.25 5.48 1.91 2.61 3.48 1.0%
C=5.0
P(obs)t
P(calc) A = 0.080 C=4.2
1.93
1.92
1.36
1.32
2.03 2.27 2.64 3.58 5.15 3.38 3.26 3.30 4.30 5.40
2.07 2.27 2.69 3.58 5.05 3.38 3.24 3.31 4.27 5.33
1.42 1.52 1.82 2.38 3.48 2.13 2.06 2.17 2.88 3.38
1.43 1.55 1.81 2.35 3.14 2.14 2.09 2.15 2.82 3.49
1.92
1.91
2.50
2.55
t .63
1.73
3.28
3.26
2. ~.()
2.18 3.1%
1.0%
--
--
* Pyrex vessels. t Quartz vessel.
atoms is inefficient, and it is found that the only satisfactory fit to the experimental results is obtained by assuming that this is coupled with an efficient destruction of oxygen atoms. The possibility that the removal of oxygen atoms is either semi-efficient or inefficient has been examined, but %e variation of the observed limit with mixture composition cannot then be explained. Thus, kH=eHg~/d, where d is the vessel diameter, ~ the average velocity of hydrogen atoms, and eH the efficiency of the surface for their removal (fraction of collisions with the surface leading to destruction), 9 and tco=23Do°/Pd 2, where Do° is the diffusion coefficient of oxygen atoms for the particular mixture at unit pressure? Following Baldwin, r the explosion condition at the uninhibited limit becomes l-
(
(A/uP )+ [cO~(~:P~+cC)~, (2)
A is a constant (=eegH/2k2d), and cC/P=ko/ka, with c a correction for the variawhere
tion of Do° with mixture composition. The relative importance of the two terms on the r.h.s, of Eq. (2) will vary with mixture composition and, thus, the required values of A and C can be obtained from a comparison of the experimental limits with those calculated from known values of A and C. In these calculations, the reported values 7,8 of c and k4/k2 were used. Table I shows the agreement obtained between the experimental and calculated explosion pressure at 510°C and, in addition, the way in which the calculated limits vary with the values of A and C. With A=0.120 and C=4.5, the r.m.s. deviation between the calculated and experimental explosion pressures was only 0.9%, and the difference between corresponding values was usually less than the experimental error. With C constant at 4.5, a change in A produced ~ corresponding change in the magnitude of all the calculated limits, with a corresponding increase iu the r.m.s, deviation, but Table 1 shows that the agreement was still acceptable for 0.118~A 7:0.122. An acceptable agreement could also be obtained with values of A between
448
OXIDATION AND IGNITION TABLE I I I Values of k5 calculated from Eq. (1)
Temperature 510°C; 3.6-cm clean Pyrex vessel. Mixture composition (mole fractions) y
104i
Explosion pressure (torr)
0.86
0.14
0.80 1.61 2.41 3.22 4.02
3.45 5.10 6.00 7.25 10.00
3.35 3.66 2.85 2.39 2.09
0.56
0.14
1.61 2.41 4.02
4.35 8.35 11.20
3.16 3.58 2.26
0.28
0.14
0.80 1.61 2.41 3.22 4.02
3.50 4.15 5.22 7.80 13.10
2.44 2.21 2.28 2.50 2.33
0.14
0.14
0.80 1.61 2.41 3.22
4.16 6.14 8.40 ]6.10
2.15 3.11 2.97 2.87
0.07
0.14
0.80 1.61 2.41
5.75 9.00 12.65
2.94 3.55 3.15
0.28
0.72
17.70 18.45 19.30
6.52 9.00 11.70
2.46 2.70 2.71
x
10- l~k~ (cm~ mole- i s- 1)
0.28
0.56
14.20 16.08
6.52 10.00
2.33 2.46
0.28
0.28
4.82 6.42 8.04
4.16 5.18 9.77
2.06 2.04 2.36
0.28
0.07
1.21 1.61
11.77 13.50
3.42 2.64
0.110 and 0.132, provided a corresponding change was also made in the value of C. Where the values of A and C were varied outside the limits covered in Table I, the r.m.s, deviation rose rapidly and the calculated limits for the extremes of mixture composition deviated markedly from the experimental results. I t is clear that the minimum in the plot of r.m.s, deviation against A and C is very shallow, so that it is difficult to identify precisely the required values of these constants. In practice, the values
giving the lowest r.m.s, deviation were used, and it is fortunate that the uncertainty in the precise values of A and C had little effect on the value obtained for k5 (see below). Table I I shows that a s i m i l a r satisfactory agreement was obtained between the calculated and experimental explosion pressures at the other temperatures used in this work. With these 3.6-cm vessels, the values of C vary in the expected manner with temperature, and although the values of A show some fluctuation, they all corresponded to values of e~ between 1.0X10 -4 and 2.0X10 -4. I t is interesting to note that,' even when the explosion pressures had changed very significantly, e.g., the standard mixture had changed from 2.38 to 3.33 torr at 520°C, the value of C required to explain the observed variation in the limit with mixture composition remained unchanged at 4.2. This increase in the limit was due solely to an increase in ei~ from 1.55X10 -a to 3.28X10 -t. On re-washing the reaction vessel, eH reverted to 1.63X 10-4 while C remained at 4.2, which suggests that any fluctuations in the explosion pressure were solely due to changes in ell. The situation was slightly less satisfactory with the 1.5-cm vessel, in that the efficient removal of oxygen atoms requires Cd2 to be constant and, on this basis, a value of C = 2 6 would be expected, compared with the experimental value of 7.0. With the former value, however, it is impossible to reproduce the experimental variation of explosion pressure with mixture composition.
Analysis of Inhibited Limits Equation (1) has been used to calculate values of k5 from the experimental explosion pressures, the known l°'s values of k2 and k4/k2, and the values of A and C derived from the analysis of the corresponding uninhibited limits. Table I I I lists the individual values of ks, obtained in this way, for the 3.6-cm vessel at 510°C. Although there is some scatter in the values, they are effectively independent of mixture composition, and it is particularly significant that there is no drift in the values, even when the ratio J-O2-I/[-H2~is large. This eliminates the possibility of any significant reaction between hydroxyl radicals or oxygen atoms and ethyl bromide, since both these chain centers react with hydrogen. The occurrence of either of these 1)rimary chain-termination reactions would modify the explosion condition [Eq. (1)J and, thus, the values of k5 derived from Eq. (1) would not be independent of mixture composition. Such variations should be most apparent for mixtures
INHIBITION OF FIRST LIMIT where the ratio [-O2~/[-H2~ is high, but no drift is obtained in the values of ks, even with such extreme mixture compositions. Equation (1) also requires that the values of k5 should be independent of vessel diameter, and this has been confirmed at one temperature. Table IV shows that Mmost the same value was obtained for k5 in both the 3.6- and the 1.5-em vessels, and that the small difference between the values is well within the standard deviations of the two individual values. It must be concluded, therefore, that the suggested mechanism for the inhibition is correct, and that the values of ks, which have been obtained, are for the reaction of hydrogen atoms with ethyl bromide. It was pointed out earlier that there is some uncertainty in assigning precise values to A and C, and thus the sensitivity of the derived values of k5 on the values of A and C used in their calculation has been examined. For this purpose, the calculations have been repeated for the results at 510°C, using the extreme pairs of values of A and C listed in Table I. With A = 0.11 and C=5.0, the average value of /c~ was 2.64X10 ~ cma mole-~ s-~, while with A=0.132 and C=4.0, this average had only shifted to 2.76X10 ~ ema mole-~ s-~; these two values should be compared with the average of the values listed in Table III, namely, 2.70X10 ~ ema mole-~ s-L These differences are a lot smMler than the r.m.s, deviation of the results, so that any uncertainty in the values of A and C does not lead to any serious error in the determination of/c~. The detailed results for the other temperatures show a scatter similar to those at 510°C, and Table IV lists the average values obtained from these individual determinations. In all eases, this average is from at least 18 individual determinations, and the standard deviation for each set of values is of the order of 20%. This deviation can probably be attributed to the TABLE IV Variation of k~ with temperature Avg. vMue Vessel of ka diam T No. of de- (cma mole-1 Standard deviation (cm) (°C) terminations s-1) 3.6
460 480 510 520
26 27 30 18
2.10X1012 2.21 X 1012 2.70X1012 2.85 X 101~
0.41 X 101~ 0.32X10 I~ 0.49 X 1012 0.58 X 1012
1.5
510
27
2.45 X 1012 0.34X1012
449 TABLE V
Comparison of the rate constants for the reactions of hydrogen atoms with ethyl bromide and ethane T (°C)
k5 (era3 mole-I s-1)
/c(H-bC2H6) (cm3 mole-1 s-1)
460 480 510 520
2.10X10 ~2 2.21 X 1012 2.58 X 1012 2.85 X 1012
0.17X1012 0.20 X 1012 0.26 X 1012 O. 28 X 1012
sensitivity of the individuM values to the experimentally determined explosion pressures. An analysis of all the individual values of k~ gives logks/cma mole-1 s- l = (13.93=t=0.24) -- ( 5 4 7 0 - 4 - 8 3 0 ) / 4 . 5 7 6 T ,
where the error limits are the standard deviations of the slope and intercept from a leastsquares fit of the Arrhenius plot. Although the average values of k5 lie oil a good straight line, the actual errors in the Arrhenius parameters are likely to be somewhat bigger than those quoted, since k~ has only been determined over a fairly limited range of temperature. In contrast, the actual values of ka listed in Table IV will be much more reliable. The preceding kinetic analysis has established that the inhibition arises through the reaction of hydrogen atoms with the ethyl bromide, but before these values of k5 can be firmly identified with Reaction (5), it is necessary to show that Reaction (6) is unimportant under the present conditions, otherwise the values obtained are omnibus values for a combination of both reactions. Table V lists the rate constant for the reaction of hydrogen atoms with ethane [which should be very similar to that expected for Reaction (6)], and it will be seen that it is less than one-tenth of the value of/ca at the present temperatures. As a result, Reaction (6) is kinetieally negligible, and the Arrhenins equation given above can be firmly identified with Reaction (5). These Arrhenius parameters are in reasonable agreement with those obtained for the only two other reactions of an alkyl bromide with hydrogen atoms which have been studied. The activation energy for the reaction of carbon tetrabromide 1' from molecular-beam experiments is 3.0 kcal mole-~, while Fenimore and Jones 3 have obtained the rate constant for the reaction with methyl bromide from a study in which methyl
OXIDATION AND IGNITION
450 TABLE VI
alkyl bromides on flame propagation. Wilson4 found that the pre-heat region of a lean methaneComparison of the rates of Reactions (5a) and (10) oxygen flame was extended by the presence of with (2), and the rates of Reactions (8a) and (9) methyl bromide, but that the primary reaction Concentrations of reactants are the initial con- zone was narrowed and moved to a higher terneentrations in the flame used by Wilson (Ref. 4), perature. Hydrogen bromide was detected quite early in the flame and its appearance correnamely, O2 89.6%, CH4 8.7%, and CHsBr 1.7%. sponded to the disappearance of the methyl bromide. Thus, Wilson4 suggested that, in the T(°K) early stages of the flame, the methyl bromide reacted primarily with hydrogen atoms by Relative rates 600 800 1000 Reaction (5a), but in the later stages of the primary reaction zone, where the concentration k~,[CH3Br]/k2[0.2] 80 8 2 of hydroxyl radicals greatly exceeds that of k~0[CH4]/]%[-O2] 3 1.5 1 hydrogen atoms, the main path by which methyl ks~[CH3Br]/kg[CH~] _<16 _<4 _<2 bromide is consumed is Reaction (8a). If the inhibition is due to Reaction (8a), it must predominate over Reaction (9) and Table VI lists bromide was introduced into a hydrogen- the maximum possible values for the relative nitrous oxide flame. The latter gave/c5~= 1.4)< 101~ rates of these two reactions for temperatures of em ~ mole-1 s-~ at 1900°K, while extrapolation of 600 °, 800 °, and 1000°K, assuming the initial the present data to 1900°K gives ks=2.0XlO ~a concentrations present in the flame used by em 3 mole-~ s-1. The reactions of hydrogen atoms Wilson. These ratios have been obtained using with methyl and ethyl bromides should have k8~= 10~4 cm 3 mole-1 s - ' and the known value Is very similar rate constants, but even so, the for /ca. I n this particular flame, there is very excellent agreement between these two values is little reaction below 800°K and, at this ternprobably somewhat fortuitous. perature, Reaction (8a) is only four times faster than Reaction (9) under the given condiH + CH3Br= H B r + C H 8 (5a) tions. Even at 600°K, Reaction (Sa) is only favored by a factor of 16. If, as seems likely, O H + C 2 H s B r = H20+C2H~Br (8) Reaction (8a) has a small activation energy of 2-3 kcal mole-~, Reactions (8a) and (9) have [-The alternative reactions of hydroxyl radicals comparable rates at 800°K, while if the value of with an alkyl bromide have been considered by /Cs~ derived by Wilson4 is accepted, Reaction (9) becomes even faster compared to Reaction Wilson,4 who has concluded that the abstraction (8a). I t is thus very doubtful whether Reaction of a hydrogen atom is the most likely proeess.] (8a) can be responsible for the observed inhibiThe present study shows that Reaction (8) is not kinetically important at the first limit of tion, especially since Reaction (9) is not a chain-branching reaction, but only a propagathe thermal reaction between hydrogen and tion step. Nevertheless, Wilson was almost oxygen and, at first sight, this is a surprising certainly correct in concluding that the main result, especially if flame inhibition by the alkyl bromides arises through the preferential process by which methyl bromide was removed removal of hydroxyl radicals. The maximum from the system was Reaction (Sa), since the ratio ]-OH]/[-H-] would have been very high in possible value for ks will be ~1014 em ~ mole-~ s- t (i.e., the bimolecular collision frequency), and this flame. thus it is possible to compare the maximum rate for Reaction (8) with the known rate of O H + CH3Br = H 2 0 + CH2Br (8a) Reaction (1) for the present experimental conO H + CH4 = H20-]- CH3 (9) ditions. This shows that Reaction (1) is at least ten times faster than Reaction (8), and H--~-CH4 = H2-~CH3 (10) this difference in rates is presumably suificiently large for Reaction (8) to be unimportant in the present study. The alternative suggestion2'~ that the inhibition is due to the occurrence of Reaction (5a) is much more likely. To examine this possibility, Flame Inhibition by Alkyl Bromides it has been assumed that ks~=ks, in the absence The above results have important implications of any direct information about the Arrhenius from the point of view of the inhibiting effect of parameters for Reaction (5a), but, even if this
INHIBITION OF FIRST LIMIT is not strictly correct, it is unlikely to invalidate the conclusions reached on this basis. Table VI shows that, for the flame system used b y Wilson, Reaction (5a) is eight times faster than Reaction (2) at 800°K, while at 600°K, it is 80 times faster, and it is interesting to note that Reaction (5a) is still twice as fast as Reaction (2) at 1000°K. In the normal inhibited flame the only alternative reaction for the hydrogen atoms is Reaction (10), and Table VI shows that Reactions (2) and (10) have comparable rates at 800°K. Thus, the occurrence of Reaction (Sa) can undoubtedly delay the onset of the rapid increase in rate which arises through chain branching. I t follows that the primary reaction zone of the flame will be moved to a higher temperature, as observed by Wilson, and it is not unreasonable that the reaction should then occur more rapidly than in the uninhibited flame, since it will be occurring at a higher temperature. The above approach to the problem of flame inhibition by alkyl bromides is only an approximation to the actual situation, since it neglects any change in mixture composition as the temperature rises. Nevertheless, it is clear that the primary inhibition reaction must involve the removal of hydrogen atoms through the abstraction of bromine atoms from the alkyl bromide. This, in turn, produces hydrogen bromide, which is a very effective flame inhibitor itself. A full understanding of the mechanism of flame inhibition, however, also requires information about the subsequent steps in the reaction sequence and, in particular, it is necessary to establish whether a simple delay in the chain branching is sufficient to explain the observed reduction in burning velocity, or whether a final chain-termination reaction still needs to be identified. Acknowledgment This work was sponsored by the Air Force Office of Scientific Research under Grant AF EOAR 66-31
451
through the European Office of Aerospace Research (OAP,), United States Air Force. REFERENCES 1. DIXON-LEwIS, G. AND LINNETT, J. W.: Proc. Roy. Soc. A210, 48 (1951). 2. BURDON, M., BURGOYNE, J. H., AND WEIN-
m:nG, F. J. : Fifth Symposium (International) on Combustion, p. 647, Reinhold, 1955. 3. FENIMORE, C. P. AND JONES, G. W.: Combust. Flame, 7, 323 (1963). 4. WlLSOX, W. E.: Tenth Symposium (International) on Combustion, p. 47, The Combustion Institute, 1965. 5. BALDWlX, R. R. AND SIMMONS, R. F.: Trails. Faraday Soc. 51, 680 (1955). 6. BALDWIN', R. R. AND SIMMONS, R. F.: Trans.
Faraday Soc. 53, 964 (1957). 7. BALDWIN',R. R.: Trans. Faraday Soc. 52, 1344
(1956). 8. BALDWIN, 1~. ][{., JACKSON, ])., WALKER, R. W., AND WEBSTER, S. J.: Trails. Faraday Soc. 63,
t676 (1967). 9. SEMONOV, N. N.: Acta Physicochim. URSS 18,
93 (1943). 10. BALDWIN, R. R. AND ~IELvIN, A.: J. Chem.
Soe. 1964, 1785. 11. CHOTKA, H. G., MAI~TIN, H., ~X~,'DS~IDEL, W.: Z. Phys. Chem. 65, 139 (1969). 12. BAULCH, D. L., DRYSDALE, 1). D., AND LLOYD, A. C.: High Temperature Reaction Rate Data, No. 2, p. 9, Univ. of Leeds, 1968. 13. k9 obtained by combining the results of: GREIXEn, N. R.: J. Chem. Phys. 53, 1070 (1970) ; HORNE, D. C. AND NORI~.ISH, R. G. W.: Nature 215, 1374 (1967); WESTENBERG, A. A. *N'D FRISTROM, R. M.: J. Phys. Chem. 65, 591 (1961); FENIMORE, C. P. AND Joxv;s, G. W.: J. Phys. Chem. 65, 2200 (1961); FRISTROM, I~. M.: Ninth Symposium (International) on Combustion, p. 560, Academic Press, 1963.
Department of Chemistry, The University of ,~Ianchester Institute of Science and Technology, Manchester, England The inhibition of the first limit of the hydrogen-oxygen reaction by ethyl bronfide has been studied between 460 ° and 520°C, over a wide range of mixture composition, using clean Pyrex and silica reaction vessels. It is shown that the inhibition arises through the reactions: H-}-C2HsBr = HBr +C2H5,
(5)
C.oHs-}-O2 = C 2H4@HO2,
(7)
The surface-destruction constants for the removal of hydrogen and oxygen atoms by the surface have been determined from a detailed analysis of the uninhibited-explosion pressures, and these values have been used to calculate a value of k5 from each inhibited-explosion pressure. These values are independent of mixture composition and vessel diameter, and an Arrhenius plot gives logk~/cm a mole-~ s-~ = (13.93±0.24) - (5470±830)/4.576T. No evidence was obtained for any reaction of hydroxyl radicals or oxygen atoms with the ethyl bromide, and the implications of these results to the problem of flame inhibition by halogen compounds is considered.
Alkyl halides are widely used as fire-extinguishing agents, and it is believed that their action is chemical in nature, in that they preferentially react with chain centers which are important for the propagation of the flame to give chain termination. I A number of studies have been made in an attempt to elucidate which chain center (or centers) is removed by the inhibitor, and arguments have been advanced for the preferential removal of both hydrogen atoms 2'a and hydroxyl radicals. 4 The aim of the present work was to compare the rates of these two types of processes directly, using ethyl bromide as inhibitor, from its effect on the first limit of the thermal reaction between,hydrogen and oxygen. Both hydrogen atoms and hydroxyl radicals are important chain centers in this reaction, so that a kinetic study of the inhibition * Present address: I . C . I . Mond Division, The Heath, Runeorn, Cheshire, England.
should show whether one or both of these chain centers react with the ethyl bromide. The inhibition of the first limit was chosen for this purpose, as a preliminary study of the inhibition of the second limit showed that ethyl bromide was decomposing, with a half-life of 15 s at 520°C, into ethylene and hydrogen bromide. The determination of the second explosion pressure requires the gases to be mixed in the reaction vessel at a pressure above the limit, and the pressure is then reduced at a controlled rate to reach the explosion boundary. Thus, a considerable proportion of the inhibitor would have decomposed during this manipulation time. In contrast, the first limit is determined by the direct admission of the premixed gases into the evacuated reaction vessel, and explosion occurs almost instantaneously, so that the thermal decomposition of the inhibitor is not a problem under these experimental conditions. It will be shown that, at the first limit,
443
444
OXIDATION AND IGNITION
Ib
ol
5
IO
|
I
15
20
MOLE FRACTION ETHYL BROMIDE ( x l O 3) FIG. 1. The variation of explosion pressure with mole fraction of ethyl bromide. 3.6-cm vessel; temp., 510°C; mole fractions of hydrogen, 0.28; mole fractions of oxygen, • 0.72; • 0.56; • 0.28; • 0.14; • 0.07.
ethyl bromide reacts almost exclusively with hydrogen atoms by Reaction (5), and the rate constant for this reaction has been obtained over the temperature range 460°-520°C. The implication of these results to the problem of flame inhibition is also considered.
Experimental The apparatus was essentially similar to that described earlier3 The first limit was obtained by the conventional method, in which a known pressure of the reactant mixture was admitted to the evacuated reaction vessel, and the occurrence or absence of explosion noted. The uninhibited limits were pale blue in color, and it was necessary to exclude stray light to obtain accurate explosion pressures; but, when ethyl bromide was present, the flashes were yellow and the explosion boundary was easily detected. The limit was usually approached from the explosive side and determined with an accuracy of =t=0.03 torr. Between each individual determination, the reaction vessel was rinsed with hydrogen and evacuated for 3 rain (to a pressure < 10~ torr). Mixtures of hydrogen, oxygen, and nitrogen
were used in this investigation, to permit the effect of an independent variation of the mole fractions of hydrogen and oxygen on the efficiency of the inhibitor to be examined. With a constant mole fraction of hydrogen (x=0.28), the mole fraction (y) of oxygen was varied from 0.72-0.07, while with a constant mole fraction of oxygen (y=0.14), the mole fraction (x) of hydrogen was varied from 0.86-0.07. Most of the determinations were made using cylindrical reaction vessels (3.6 cm i.d. by 20 cm long), but the effect of veasel diameter was examined also at one temperature by using a 1.5-cm-i.d. vessel as well. One difficulty in studying the inhibition of the first limit is that surface-destruction processes are very important at the explosion boundary, and slight changes in the surface can affect the magnitude of the explosion pressure. Thus, throughout the present work, care was taken to ensure that the uninhibited-explosion pressures were reproducible from day to day, and this reproducibility was usually within the experimental error of the determination (4-0.03 torr). If the deviation were outside this experimental error, the limit had usually changed quite markedly and the reaction vessel was replaced by a freshly washed vessel.
INHIBITION OF FIRST LIMIT
445
16
fir
O b--
12
tlJ r~ it0
tD tlJ
a~ 8 Q. Z
O tD
O X
I
I
I
I
I
2 3 4 M O L E FRACTION ETHYL BROMIDE ( x l O 3 ) FIG. 2. The variation of explosion pressllre with mole fraction of ethyl bromide. 3.6-cm vessel; temp., 510°C; mole fraction of oxygen, 0.14; mole fractions of hydrogen, • 0.86; • 0.56; • 0.28; A 0.14; ~ 0.07. Results
A systematic study of the inhibition of the first limit of the hydrogen-oxygen reaction has been made in the temperature range 460 °520°C, over the wide range of mixture eomposition indicated earlier. This range of temperatures was limited by two factors. At 520°C, induction periods of 3-5 s were observed with mixtures with a low value for the ratio [-I-I2-1/[0~-] and, at this temperature, a significant amount of inhibitor would have decomposed after this time, so that it was necessary to disregard the results obtained with such mixtures. In addition, as the temperature was reduced, the tip of the explosion peninsula was approached and only traces of ethyl bromide were required to suppress explosion. This meant that the sensible lower-temperature limit in this work was 460°C. Some typical results are shown ill Figs. ] and 2. The former shows that increasing the mole fraction of oxygen in the mixture reduces the efficiency of ethyl bromide as an inhibitor; a useful indication of its efficiency is the mole fraction required to double the uninhibitedexplosion pressure, and this quantity is almost directly proportional to the mole fraction of oxygen. In contrast, Fig. 2 shows that, even
when the mole fraction of hydrogen is increased from 0.07-0.86, the efficiency of the inhibitor is only slightly affected. These observations strongly suggest that ethyl bromide competes with oxygen in a reaction with hydrogen atoms to give chain termination, and any further consideration of these results wilt be deferred until this suggestion has been examined in more detail.
Discussion
There are two possible reactions of hydrogen atoms with ethyl bromide, namely, H + C2HsBr = H B r + Cells,
(5)
H + C2HsBr = H2+ C2H4Br.
(6)
Reaction (5) is exothermic by ~ 2 0 keal mole-I, whereas Reaction (6) is only slightly exothermic ( ~ 5 kcal mole- I ) . These values should be sufficient to produce enough difference in the activation energies of these two reactions for Reaction (5) to predominate over Reaction (6), and the ethyl radicals produced in the former reaction will react with oxygen to give
446
OXIDATION AND IGNITION TABLE I Comparison of observed and calculated uninhibited-explosionpressures
Temp., 510°C; 3.6-cm clean Pyrex vessel. P(ealc) A =0.132 z
y
O. 28
O. 72 0.56 0.44 0.28 O. 14 0.07 0.14
0.86 0.72 0.56 0.44 0.14 0.07 0.44 O. 14 0.07
P(obs)
1.59 1.69 1.85 2.15 2.84 3.95 2.65 2.52 2.57 2.61 3.42 4.20 0.56 1.56 2.09 2.65 R.m.s. deviation
C=4.0 1.55 1.69 1.84 2.17 2.88 3.98 2.65 2.58 2.58 2.65 3.42 4.20 1.55 2.05 2.57 1.5%
chain termination by Reaction (7)6; C2H5-1-02 = C2H4--~-HO2,
(7)
0H+H2= H20+H,
(1)
I-I+ 02 = OH+O,
(2)
O+H2= OH+H,
(3)
H + O 2 + M = HO2+M,
(4)
H = surface destruction,
(H)
0 = surface destruction.
(0)
If Reactions (5) and (7) are added to the mechanism of the uninhibited reaction,r i.e., Reactions (1)-(4) and the surface destruction of hydrogen and oxygen atoms, the explosion condition becomes
1-- (k4mP/2k2) = (k~/2k2yP ) + E k o / ( & z P + ] c o ) ] + (ksi/2k2y ),
(1)
where P is the explosion pressure, x, y, and i the mole fractions of hydrogen, oxygen, and ethyl bromide, respectively, and m a composite third-body term expressing the efficiency of the particular mixture in Reaction (4). [-Baldwin7 considered the destruction of H and OH at the
A =0.118 C=4.5
A =0.120
1.56 1.69 1.84 2.17 2.84 3.89 2.61 2.55 2.55 2.61 3.39 4.20 1.55 2.06 2.59 1.2%
1.57 1.70 1.85 2.18 2.86 3.92 2.63 2.56 2.57 2.63 3.42 4.19 1.56 2.07 2.60 0.9%
C=4.5
A =0.122 C=4.5 1.58 1.71 1.86 2.19 2.88 3.95 2.65 2.58 2.58 2.65 3.44 4.24 1.56 2.08 2.62 1.1%
A =0.110
C=5.0 1.58 1.72 1.86 2. t9 2.86 3.87 2.62 2.55 2.55 2.62 3.42 4.24 1.57 2.09 2.63 1.1%
surface of the vessel, but the value of ]~1 obtained from the analysis of the results is very close to that expected for k3. It seems certain, therefore, that the second species destroyed at the surface is 0 atoms, rather than OH radicals.] At first-limit pressures, the term k4mP/2k2 is relatively unimportant, but a detailed analysis of tile uninhibited explosion pressures shows that neither of the terms involving kit or ko in Eq. (1) is small enough to be neglected. As a result, there is no simple graphic way in which the validity of Eq. (1) can be verified. Instead, it has been necessary to calculate a value of k5 from each experimental explosion pressure, using the knowns ratio k4/k2, and the surfacedestruction constants for hydrogen and oxygen atoms, obtained from the variatiou of the uniuhibited limit with mixture composition. The value of ks, obtained in this way, is independent of both mixture composition and vessel diameter, which confirms the proposed mechanism. Thus, in the following sections of this paper, the uninhibited limits are considered to obtain the required surface-destruction constants, and these are then used to evaluate k5 from Eq. (1).
:4 nalysis of the Uninhibited Explosion Boundary The general magnitude of the explosion pressures suggests tile destruction of hydrogen
INHIBITION OF FIRST LI MI T
447
TABLE II Comparison of observed and calculated uninhibited-explosion pressures Diameter of reaction vessels, 3.6 cm. 460°C
480°C
P(calc) A = 0.133 x
y
P(obs)*
0.28
0.72 O. 56 0.44 0.28 0.14 0.07 0.14
1.95 2.05 2.29 2.66 3.49 4.91 3.34 3.12 -4.25 5.51 1.89 2.63 3.51
0.86 0.56 0.44 0.14 0.07 0.44 0.56 0.14 0.07 R.m.s. deviation
520°C
P(calc) A = 0.170
C=5.5
P(obs)*
1.94 2.10 2.27 2.66 3.50 4.87 3.37 3.16 -4.25 5.48 1.91 2.61 3.48 1.0%
C=5.0
P(obs)t
P(calc) A = 0.080 C=4.2
1.93
1.92
1.36
1.32
2.03 2.27 2.64 3.58 5.15 3.38 3.26 3.30 4.30 5.40
2.07 2.27 2.69 3.58 5.05 3.38 3.24 3.31 4.27 5.33
1.42 1.52 1.82 2.38 3.48 2.13 2.06 2.17 2.88 3.38
1.43 1.55 1.81 2.35 3.14 2.14 2.09 2.15 2.82 3.49
1.92
1.91
2.50
2.55
t .63
1.73
3.28
3.26
2. ~.()
2.18 3.1%
1.0%
--
--
* Pyrex vessels. t Quartz vessel.
atoms is inefficient, and it is found that the only satisfactory fit to the experimental results is obtained by assuming that this is coupled with an efficient destruction of oxygen atoms. The possibility that the removal of oxygen atoms is either semi-efficient or inefficient has been examined, but %e variation of the observed limit with mixture composition cannot then be explained. Thus, kH=eHg~/d, where d is the vessel diameter, ~ the average velocity of hydrogen atoms, and eH the efficiency of the surface for their removal (fraction of collisions with the surface leading to destruction), 9 and tco=23Do°/Pd 2, where Do° is the diffusion coefficient of oxygen atoms for the particular mixture at unit pressure? Following Baldwin, r the explosion condition at the uninhibited limit becomes l-
(
(A/uP )+ [cO~(~:P~+cC)~, (2)
A is a constant (=eegH/2k2d), and cC/P=ko/ka, with c a correction for the variawhere
tion of Do° with mixture composition. The relative importance of the two terms on the r.h.s, of Eq. (2) will vary with mixture composition and, thus, the required values of A and C can be obtained from a comparison of the experimental limits with those calculated from known values of A and C. In these calculations, the reported values 7,8 of c and k4/k2 were used. Table I shows the agreement obtained between the experimental and calculated explosion pressure at 510°C and, in addition, the way in which the calculated limits vary with the values of A and C. With A=0.120 and C=4.5, the r.m.s. deviation between the calculated and experimental explosion pressures was only 0.9%, and the difference between corresponding values was usually less than the experimental error. With C constant at 4.5, a change in A produced ~ corresponding change in the magnitude of all the calculated limits, with a corresponding increase iu the r.m.s, deviation, but Table 1 shows that the agreement was still acceptable for 0.118~A 7:0.122. An acceptable agreement could also be obtained with values of A between
448
OXIDATION AND IGNITION TABLE I I I Values of k5 calculated from Eq. (1)
Temperature 510°C; 3.6-cm clean Pyrex vessel. Mixture composition (mole fractions) y
104i
Explosion pressure (torr)
0.86
0.14
0.80 1.61 2.41 3.22 4.02
3.45 5.10 6.00 7.25 10.00
3.35 3.66 2.85 2.39 2.09
0.56
0.14
1.61 2.41 4.02
4.35 8.35 11.20
3.16 3.58 2.26
0.28
0.14
0.80 1.61 2.41 3.22 4.02
3.50 4.15 5.22 7.80 13.10
2.44 2.21 2.28 2.50 2.33
0.14
0.14
0.80 1.61 2.41 3.22
4.16 6.14 8.40 ]6.10
2.15 3.11 2.97 2.87
0.07
0.14
0.80 1.61 2.41
5.75 9.00 12.65
2.94 3.55 3.15
0.28
0.72
17.70 18.45 19.30
6.52 9.00 11.70
2.46 2.70 2.71
x
10- l~k~ (cm~ mole- i s- 1)
0.28
0.56
14.20 16.08
6.52 10.00
2.33 2.46
0.28
0.28
4.82 6.42 8.04
4.16 5.18 9.77
2.06 2.04 2.36
0.28
0.07
1.21 1.61
11.77 13.50
3.42 2.64
0.110 and 0.132, provided a corresponding change was also made in the value of C. Where the values of A and C were varied outside the limits covered in Table I, the r.m.s, deviation rose rapidly and the calculated limits for the extremes of mixture composition deviated markedly from the experimental results. I t is clear that the minimum in the plot of r.m.s, deviation against A and C is very shallow, so that it is difficult to identify precisely the required values of these constants. In practice, the values
giving the lowest r.m.s, deviation were used, and it is fortunate that the uncertainty in the precise values of A and C had little effect on the value obtained for k5 (see below). Table I I shows that a s i m i l a r satisfactory agreement was obtained between the calculated and experimental explosion pressures at the other temperatures used in this work. With these 3.6-cm vessels, the values of C vary in the expected manner with temperature, and although the values of A show some fluctuation, they all corresponded to values of e~ between 1.0X10 -4 and 2.0X10 -4. I t is interesting to note that,' even when the explosion pressures had changed very significantly, e.g., the standard mixture had changed from 2.38 to 3.33 torr at 520°C, the value of C required to explain the observed variation in the limit with mixture composition remained unchanged at 4.2. This increase in the limit was due solely to an increase in ei~ from 1.55X10 -a to 3.28X10 -t. On re-washing the reaction vessel, eH reverted to 1.63X 10-4 while C remained at 4.2, which suggests that any fluctuations in the explosion pressure were solely due to changes in ell. The situation was slightly less satisfactory with the 1.5-cm vessel, in that the efficient removal of oxygen atoms requires Cd2 to be constant and, on this basis, a value of C = 2 6 would be expected, compared with the experimental value of 7.0. With the former value, however, it is impossible to reproduce the experimental variation of explosion pressure with mixture composition.
Analysis of Inhibited Limits Equation (1) has been used to calculate values of k5 from the experimental explosion pressures, the known l°'s values of k2 and k4/k2, and the values of A and C derived from the analysis of the corresponding uninhibited limits. Table I I I lists the individual values of ks, obtained in this way, for the 3.6-cm vessel at 510°C. Although there is some scatter in the values, they are effectively independent of mixture composition, and it is particularly significant that there is no drift in the values, even when the ratio J-O2-I/[-H2~is large. This eliminates the possibility of any significant reaction between hydroxyl radicals or oxygen atoms and ethyl bromide, since both these chain centers react with hydrogen. The occurrence of either of these 1)rimary chain-termination reactions would modify the explosion condition [Eq. (1)J and, thus, the values of k5 derived from Eq. (1) would not be independent of mixture composition. Such variations should be most apparent for mixtures
INHIBITION OF FIRST LIMIT where the ratio [-O2~/[-H2~ is high, but no drift is obtained in the values of ks, even with such extreme mixture compositions. Equation (1) also requires that the values of k5 should be independent of vessel diameter, and this has been confirmed at one temperature. Table IV shows that Mmost the same value was obtained for k5 in both the 3.6- and the 1.5-em vessels, and that the small difference between the values is well within the standard deviations of the two individual values. It must be concluded, therefore, that the suggested mechanism for the inhibition is correct, and that the values of ks, which have been obtained, are for the reaction of hydrogen atoms with ethyl bromide. It was pointed out earlier that there is some uncertainty in assigning precise values to A and C, and thus the sensitivity of the derived values of k5 on the values of A and C used in their calculation has been examined. For this purpose, the calculations have been repeated for the results at 510°C, using the extreme pairs of values of A and C listed in Table I. With A = 0.11 and C=5.0, the average value of /c~ was 2.64X10 ~ cma mole-~ s-~, while with A=0.132 and C=4.0, this average had only shifted to 2.76X10 ~ ema mole-~ s-~; these two values should be compared with the average of the values listed in Table III, namely, 2.70X10 ~ ema mole-~ s-L These differences are a lot smMler than the r.m.s, deviation of the results, so that any uncertainty in the values of A and C does not lead to any serious error in the determination of/c~. The detailed results for the other temperatures show a scatter similar to those at 510°C, and Table IV lists the average values obtained from these individual determinations. In all eases, this average is from at least 18 individual determinations, and the standard deviation for each set of values is of the order of 20%. This deviation can probably be attributed to the TABLE IV Variation of k~ with temperature Avg. vMue Vessel of ka diam T No. of de- (cma mole-1 Standard deviation (cm) (°C) terminations s-1) 3.6
460 480 510 520
26 27 30 18
2.10X1012 2.21 X 1012 2.70X1012 2.85 X 101~
0.41 X 101~ 0.32X10 I~ 0.49 X 1012 0.58 X 1012
1.5
510
27
2.45 X 1012 0.34X1012
449 TABLE V
Comparison of the rate constants for the reactions of hydrogen atoms with ethyl bromide and ethane T (°C)
k5 (era3 mole-I s-1)
/c(H-bC2H6) (cm3 mole-1 s-1)
460 480 510 520
2.10X10 ~2 2.21 X 1012 2.58 X 1012 2.85 X 1012
0.17X1012 0.20 X 1012 0.26 X 1012 O. 28 X 1012
sensitivity of the individuM values to the experimentally determined explosion pressures. An analysis of all the individual values of k~ gives logks/cma mole-1 s- l = (13.93=t=0.24) -- ( 5 4 7 0 - 4 - 8 3 0 ) / 4 . 5 7 6 T ,
where the error limits are the standard deviations of the slope and intercept from a leastsquares fit of the Arrhenius plot. Although the average values of k5 lie oil a good straight line, the actual errors in the Arrhenius parameters are likely to be somewhat bigger than those quoted, since k~ has only been determined over a fairly limited range of temperature. In contrast, the actual values of ka listed in Table IV will be much more reliable. The preceding kinetic analysis has established that the inhibition arises through the reaction of hydrogen atoms with the ethyl bromide, but before these values of k5 can be firmly identified with Reaction (5), it is necessary to show that Reaction (6) is unimportant under the present conditions, otherwise the values obtained are omnibus values for a combination of both reactions. Table V lists the rate constant for the reaction of hydrogen atoms with ethane [which should be very similar to that expected for Reaction (6)], and it will be seen that it is less than one-tenth of the value of/ca at the present temperatures. As a result, Reaction (6) is kinetieally negligible, and the Arrhenins equation given above can be firmly identified with Reaction (5). These Arrhenius parameters are in reasonable agreement with those obtained for the only two other reactions of an alkyl bromide with hydrogen atoms which have been studied. The activation energy for the reaction of carbon tetrabromide 1' from molecular-beam experiments is 3.0 kcal mole-~, while Fenimore and Jones 3 have obtained the rate constant for the reaction with methyl bromide from a study in which methyl
OXIDATION AND IGNITION
450 TABLE VI
alkyl bromides on flame propagation. Wilson4 found that the pre-heat region of a lean methaneComparison of the rates of Reactions (5a) and (10) oxygen flame was extended by the presence of with (2), and the rates of Reactions (8a) and (9) methyl bromide, but that the primary reaction Concentrations of reactants are the initial con- zone was narrowed and moved to a higher terneentrations in the flame used by Wilson (Ref. 4), perature. Hydrogen bromide was detected quite early in the flame and its appearance correnamely, O2 89.6%, CH4 8.7%, and CHsBr 1.7%. sponded to the disappearance of the methyl bromide. Thus, Wilson4 suggested that, in the T(°K) early stages of the flame, the methyl bromide reacted primarily with hydrogen atoms by Relative rates 600 800 1000 Reaction (5a), but in the later stages of the primary reaction zone, where the concentration k~,[CH3Br]/k2[0.2] 80 8 2 of hydroxyl radicals greatly exceeds that of k~0[CH4]/]%[-O2] 3 1.5 1 hydrogen atoms, the main path by which methyl ks~[CH3Br]/kg[CH~] _<16 _<4 _<2 bromide is consumed is Reaction (8a). If the inhibition is due to Reaction (8a), it must predominate over Reaction (9) and Table VI lists bromide was introduced into a hydrogen- the maximum possible values for the relative nitrous oxide flame. The latter gave/c5~= 1.4)< 101~ rates of these two reactions for temperatures of em ~ mole-1 s-~ at 1900°K, while extrapolation of 600 °, 800 °, and 1000°K, assuming the initial the present data to 1900°K gives ks=2.0XlO ~a concentrations present in the flame used by em 3 mole-~ s-1. The reactions of hydrogen atoms Wilson. These ratios have been obtained using with methyl and ethyl bromides should have k8~= 10~4 cm 3 mole-1 s - ' and the known value Is very similar rate constants, but even so, the for /ca. I n this particular flame, there is very excellent agreement between these two values is little reaction below 800°K and, at this ternprobably somewhat fortuitous. perature, Reaction (8a) is only four times faster than Reaction (9) under the given condiH + CH3Br= H B r + C H 8 (5a) tions. Even at 600°K, Reaction (Sa) is only favored by a factor of 16. If, as seems likely, O H + C 2 H s B r = H20+C2H~Br (8) Reaction (8a) has a small activation energy of 2-3 kcal mole-~, Reactions (8a) and (9) have [-The alternative reactions of hydroxyl radicals comparable rates at 800°K, while if the value of with an alkyl bromide have been considered by /Cs~ derived by Wilson4 is accepted, Reaction (9) becomes even faster compared to Reaction Wilson,4 who has concluded that the abstraction (8a). I t is thus very doubtful whether Reaction of a hydrogen atom is the most likely proeess.] (8a) can be responsible for the observed inhibiThe present study shows that Reaction (8) is not kinetically important at the first limit of tion, especially since Reaction (9) is not a chain-branching reaction, but only a propagathe thermal reaction between hydrogen and tion step. Nevertheless, Wilson was almost oxygen and, at first sight, this is a surprising certainly correct in concluding that the main result, especially if flame inhibition by the alkyl bromides arises through the preferential process by which methyl bromide was removed removal of hydroxyl radicals. The maximum from the system was Reaction (Sa), since the ratio ]-OH]/[-H-] would have been very high in possible value for ks will be ~1014 em ~ mole-~ s- t (i.e., the bimolecular collision frequency), and this flame. thus it is possible to compare the maximum rate for Reaction (8) with the known rate of O H + CH3Br = H 2 0 + CH2Br (8a) Reaction (1) for the present experimental conO H + CH4 = H20-]- CH3 (9) ditions. This shows that Reaction (1) is at least ten times faster than Reaction (8), and H--~-CH4 = H2-~CH3 (10) this difference in rates is presumably suificiently large for Reaction (8) to be unimportant in the present study. The alternative suggestion2'~ that the inhibition is due to the occurrence of Reaction (5a) is much more likely. To examine this possibility, Flame Inhibition by Alkyl Bromides it has been assumed that ks~=ks, in the absence The above results have important implications of any direct information about the Arrhenius from the point of view of the inhibiting effect of parameters for Reaction (5a), but, even if this
INHIBITION OF FIRST LIMIT is not strictly correct, it is unlikely to invalidate the conclusions reached on this basis. Table VI shows that, for the flame system used b y Wilson, Reaction (5a) is eight times faster than Reaction (2) at 800°K, while at 600°K, it is 80 times faster, and it is interesting to note that Reaction (5a) is still twice as fast as Reaction (2) at 1000°K. In the normal inhibited flame the only alternative reaction for the hydrogen atoms is Reaction (10), and Table VI shows that Reactions (2) and (10) have comparable rates at 800°K. Thus, the occurrence of Reaction (Sa) can undoubtedly delay the onset of the rapid increase in rate which arises through chain branching. I t follows that the primary reaction zone of the flame will be moved to a higher temperature, as observed by Wilson, and it is not unreasonable that the reaction should then occur more rapidly than in the uninhibited flame, since it will be occurring at a higher temperature. The above approach to the problem of flame inhibition by alkyl bromides is only an approximation to the actual situation, since it neglects any change in mixture composition as the temperature rises. Nevertheless, it is clear that the primary inhibition reaction must involve the removal of hydrogen atoms through the abstraction of bromine atoms from the alkyl bromide. This, in turn, produces hydrogen bromide, which is a very effective flame inhibitor itself. A full understanding of the mechanism of flame inhibition, however, also requires information about the subsequent steps in the reaction sequence and, in particular, it is necessary to establish whether a simple delay in the chain branching is sufficient to explain the observed reduction in burning velocity, or whether a final chain-termination reaction still needs to be identified. Acknowledgment This work was sponsored by the Air Force Office of Scientific Research under Grant AF EOAR 66-31
451
through the European Office of Aerospace Research (OAP,), United States Air Force. REFERENCES 1. DIXON-LEwIS, G. AND LINNETT, J. W.: Proc. Roy. Soc. A210, 48 (1951). 2. BURDON, M., BURGOYNE, J. H., AND WEIN-
m:nG, F. J. : Fifth Symposium (International) on Combustion, p. 647, Reinhold, 1955. 3. FENIMORE, C. P. AND JONES, G. W.: Combust. Flame, 7, 323 (1963). 4. WlLSOX, W. E.: Tenth Symposium (International) on Combustion, p. 47, The Combustion Institute, 1965. 5. BALDWlX, R. R. AND SIMMONS, R. F.: Trails. Faraday Soc. 51, 680 (1955). 6. BALDWIN', R. R. AND SIMMONS, R. F.: Trans.
Faraday Soc. 53, 964 (1957). 7. BALDWIN',R. R.: Trans. Faraday Soc. 52, 1344
(1956). 8. BALDWIN, 1~. ][{., JACKSON, ])., WALKER, R. W., AND WEBSTER, S. J.: Trails. Faraday Soc. 63,
t676 (1967). 9. SEMONOV, N. N.: Acta Physicochim. URSS 18,
93 (1943). 10. BALDWIN, R. R. AND ~IELvIN, A.: J. Chem.
Soe. 1964, 1785. 11. CHOTKA, H. G., MAI~TIN, H., ~X~,'DS~IDEL, W.: Z. Phys. Chem. 65, 139 (1969). 12. BAULCH, D. L., DRYSDALE, 1). D., AND LLOYD, A. C.: High Temperature Reaction Rate Data, No. 2, p. 9, Univ. of Leeds, 1968. 13. k9 obtained by combining the results of: GREIXEn, N. R.: J. Chem. Phys. 53, 1070 (1970) ; HORNE, D. C. AND NORI~.ISH, R. G. W.: Nature 215, 1374 (1967); WESTENBERG, A. A. *N'D FRISTROM, R. M.: J. Phys. Chem. 65, 591 (1961); FENIMORE, C. P. AND Joxv;s, G. W.: J. Phys. Chem. 65, 2200 (1961); FRISTROM, I~. M.: Ninth Symposium (International) on Combustion, p. 560, Academic Press, 1963.