Kinetics of the Non-Catalytic Oxidation of Ammonia : Static Experiments in an Empty Uncoated Silica Vessel' RY EDGAR R . STEPHENS AND ROBERT N. PFASF I+HNARY 28, 10,52
JZECRIYED
A study of the oxidatioii of minionia in d n uncoated silica reaction vessel indicates that the rate increases in passing froni auimonia-rich to -lean mixture, On the rich side, rates are roughly proportional to the product of ammonia and oxygeii Loncentrations, whereas on the 1 e m side the square of i h e oxygen concentration seeins to be iiivolced The ,ictivaiioii rnergp approxinintr, 45,flOO cal 111 hoth region*
In a previous paper? the results of an investigation of the oxidation of ammonia carried out in a flow systeni were reported. Two rather striking features of this reaction became evident in the course of this investigation. First, the percentage of ammonia oxidized rose rather sharply as the percentage ammonia in the mixture was decreased. Data were presented which indicated that a t low ammonia concentrations the rate was approximately zero order in ammonia. At higher animonia concentratioiis the zero order rate constant decreased, indicating either an inverse dependence of rate on ammonia concentration or a direct dependence on oxygen concentration. The second feature of this reaction which was discovered in the original flow studies was the striking role played by the surface of the reaction vessel Briefly, the approximate equality of conversions in empty and packed quartz reaction tubes indicated that the reaction was independent of the sudace. On the other hand, a coating of potassiuni chloride on either the empty or the packed reaction tube was found to suppress the reaction very markedly, indicating that surface does play '1 role. 'These
r-?
- ___
I P
-
-
-T I -
-11-I
__
+
- 1
._
-
facts were partially reconciled by assuming that reaction chains both start and stop on the walls. Since the data obtained in the flow system were not suited to a study of the kinetics of the reaction, further experiments were undertaken in a static system. Experimental Method The apparatus used was of conventional design. The reaction bulb was constructed of fused quartz in order t o operate a t temperatures up to 725'. The bulb was cleaned by washing with water. Ammonia and oxygen were introduced i n the desired .proportion, and at the proper total pressure, directly into a mixing chamber without purification. Opening a stopcock allowed this mixture t o enter the previously evacuated reaction vessel through a combination mercury manometer and cut-off. The mercury was then raised to a reference mark t o isolate the reaction vessel and the pressure at constant volume followed as a function of time. The initial partial pressures of the two gases were obtained from the ratio of their partial pressures in the mixing chamber and the initial total pressure in the reaction bulb. This latter could be obtained either from the pressure in the mixing chamber after pressure equilibration with the reactioii bulb or by extrapolation of the pressure-time curve. The temperature of the furnace which contained rhe reaction bulb was measured and controlled by the system employed in the previous work.$ Since pressure was being used as a measur'e of extent of reaction and the initial slopes of the pressure versus time curves were of interest, it was imperative that the temperature be controlled quite accurately, especially in view of the fact that only a small pressure change accompanies this reaction. It was neces.;,try to keep the capillary lead between the reaction bulb xntl the mercury manometer warm by means of an electrical heater to prevent condens:iiiori of the r a t e r formed i n t h t , cixirlat ion rwction.
Results and Discussion 'Typical pressure-time plots for arnnionia-oxygen mixtures a t 300 mm. total initial pressure and a temperature of 6%' in a water-washed silica reaction vessel are presented in Fig. 1. Thc curves separate rather sharply into two groups, rates being relatively low for the stoichiometric iriixture (1.33 NHB: 1 O?i or those w i t h ammonia I:ig. 1.---Increase of pressure o e ~ s i i time s For various mixt ureq: P T = ~ 300 i n i n , , T = 62*5":NHa:O1.ratio is given. (1) Taken from a thesis submitted by Edgar R . Stephens in partial fulfillment of the requirements for the 1'h.D degree. The work described in this paper was jointly supported by Contract NOrd-7920 with the U. S. Naval Bureau of Ordnance as coordinated by the Applied Physics Laboratory, The Johns Hopkins University, and by Contract h-6-ori-105 with the Office of Naval Research and Office of Air Research a s coordinated by Project Squid, Princeton University. Reproduction, translation, publication, use and disposal in whole or in part b y or for the United States Government is permitted. We wish to acknowledge the assistance of I h a n H u g h S. Taylor, who h a < generhl supervision of this project. ("iI:.. R . Stephens and N . N Pmw, '1"~; J ~ ) I J K > \ I 7 1 , l1%8
in excess, and relatively high with deficient ammonia. This is the type of behavior already noted in the flow experiments a t 1 The magnitude of the difference is considerable; comparison of the I : 3 and X : 1 mixtures gives a ratio of rates of about 30 to 1 , Such a wide spread rules out any simple dependence cm concentration. .in equation like accounts roughly for the over-all variation ; however there is some evidence that the conceiitratimi dependence is different in the two regions.
KINETICSOF
July 20, 1952
THE
3481
NON-CATALYTIC OXIDATION OF AMMONIA
= ~ P ~ ~ ( xrot) NH, m s a t 200,300 and 400 m. initid pressure are preseattd in Table I for each of Six mjxture corn- where XNH,and XO,are mole fractions, the prodpositions. Values of ioiW rats, times to 25% reaction, and times to 50% reaction are given. As llCts Po3 dt 0, P&, and PO& should be constant is so often h e of oxidation reactions, reproduci- for a given MH& ratio. A glance a t the last bility and consistency are not of a high order, In three columns of Table I will reveal that this is a
'
addition, pressure increases are relatively small
(one-fourth of the ammonia, or one-third of the oxygen-whichever is deficient) and run uniformly only 75 to 80% of the expected 6nal value for nitrogen and water vapor as products, As a consequence any conclusions must be somewhat provisional. In spite of the variability of the data, certain conclusions emerge fairly dearly from Table E. Thus there is the sharp change in rate as the NHs/ 0, ratio is decreased, already noted in Fig. 1. That this should occur over so small a range of mixture ratio as that represented by 4: 3 and 1: 1 mixtures admittedly is d3Ecult to accept but the data are consistent on this feature. Apparently there is an abrupt change in kinetics as the stoichiometric ratio is passed. TABLE I EFFECTOF
RATEFOR VARIOUS OXYGENMIXTURES AT 625'
PRESSURE ON
Po ,o
[dP/dllo x 108, mm./sec.
200 300 400
1.8 2.0 6.9
3700 3300 1880
200 300
2.3 4.8 5.2 6.8
7500 4400 3850
2.6 4.7 8.5
mm.
400
200 300 400 200
300
16 27
38 400
200 300 400
71 49 35 75 140
Izr,U
sec.
4.5 2.2 4.3 5.8 5.3 5.8 4.2
15 13
40 39
11.5
3400
2NH3/ 102 20000 13000 11000 8800
13.5
33 35
7900 4900 3180
4ms/30z 25700 14300 10400
6.5 5.2 5.3
16 15 13
51 43 42
7.4 9.9 7.5
290
lNHa/lOz 2000 (4750) 1090 980 940
31
1.2 1.2
130 90 65
1NH3/202 88 270 83 190 87 140
0.3 0.3 0.3
290
40 30 42
44
where XNH,and Xop are mole fractions in the initial mixture, The products of the latter for 3:1, 2 : l and 4:3 mixtures stand as 1:1.2:1.3. This is roughly the range observed for these three rich mixtures. If the relation is
the values for 1: 1, 1:2 and 1:3 mixtures would be as 1:2.4 :3.4. This again approximates the values observed for the lean mixtures. Incidentally the agreement is not as good if the denominator is omitted. Comparison with Xbp only gives ratios AMMONIA- of 1:1.8:2.3, which is a poorer representation of the data. With respect to the data on t26 and t 6 0 , it may be X?& noted that for a second-order process
3"s/102 12000 8700 5450
630 490 340
reasonably good conclusion. With respectto the dataon initial rates, the following comments are pertinent. For a second-order process
24 26 22
1.3
4.0
1.5
1.0
(14) 3.3 3.9 3.8 0.5 0.e 0.6
1NH&o2 50 65 130 125 200 O. 0*3 0.1 0.3 120 45 95 110 300 0.1 0.3 125 35 70 200 400 a Actual pressures are with a few mm. of values given. or ' / a of the theoretical tw and ta are observed values for pressure rise.
Information about kinetics in the two regimes is afforded by analysis of the effect of total pressure for the various mixture ratios. If total pressure appears as a squared term
dP/#
=
kPNas
x
Por = k[(XNH3-k 4)Po i(-Xo,
- 4Pl
+ 3) Po - 3PI
where X N H ~X ,o 2and Po refer to initial conditions. If the initial ratio is stoichiometric (4"s: 302), this becomes dP/dt = 12k[8/7Po
- PIz
With deficient oxygen (or a stoichiometric mixture)
(P - P0)tionl For 50% reaction (P
=
1
1
3 P%*= 3 xo,Po 1
- P0)So = -6 xo, Po
for 3 : 1 and 2 : 1 mixtures, and for the stoichiometric mixture ( P - P0)w =
1 14 Po
Integrating and solving for PI& gives values in the ratio 1:1.22:1.65 for 3:1, 2:l and 4 : 3 mixtures. Observed values increase in about this way, though more rapidly. A similar result is obtained for ih. If the relation for lean mixtures is dP [OZI'
27
=
1
+ [("*1/I021
a much more complicated situation is faced unless the mixture ratio in the denominator is identified with the initial ratio. Setting [OZI POP = [(-YO, f 3)Po - 3PI = [B - 3Pl A D rn - 1 0 1,2 L U - = k i=
U1
"1
1 -I- [XNHa/XOslO
dt
With ammonia now deficient (P
1
- Po)fin.sl
(P
- P8)W
2 X N HPO ~ =
1
g XNHI PO
ANTOE €3 BURG
3482
Integrating and solving for P ~ Mthe ) , values for 1:3, I : 2 and 1:1 mixture ratios stand as 1:2.0: 9.4. The Potw product thus should change much more rapidly with mixture ratio. The experimental data are again in reasonable agreement. The temperature coefficient has been determined for the 2NHS:102, and for 1NH3:2o2 mixtures (Table IIj. Suitable plots give reasonably good straight lines from which activation energies have TABLE 11 ACTIVATION ENERGY P0sn, = 200 mm., = 100 mm. 625
O C .
650
700
675
Eart.
kcal.
725
52 [dP/dt]" X IO4 69 109 213 361 tzdsec. 1 3850 1780 840 490 300 tso(scc.) 11000 5860 2600 1500 800 PONEJ
'C
= 100 mm., 526
[dP/dt]o X lo4 1% tsa
550
36 80 1650 730 4200 2100
42.2 45.6 46.9
POoz = 200 mm. 575
600
206 340 910
495 750 160 90 380 190
[COXTRIBUTIONFROM
THE
625
48.0 42.7 44.2
DEPARTMEST
OF
ITol. 74
been calculated. Results vary somewhat but indicate a value in the neighborhood of 45,000 cal., for both rich and lean mixtures. There is thus some evidence for an underlying similarity in mechanism. Without attempting to trace what the detailed process may involve, one additional fact may be added. Current experiments in a KC1-coated reaction vessel indicate that with ammonia-rich mixtures the rates are the same as given above for the uncoated vessel, and conform to the secondorder kinetics. On the lean side, the rate is relatively low in the early stages of reaction, and varies X PO,product. After a time, howwith the PNH, ever, the rate begins to increase, and thereafter roughly duplicates data for uncoated vessels. The behavior is not unlike that associated with "cool-flame" phenomena in hydrocarbon combustion, except that there appear to be no real discontinuities. Possibly this is again a case of slow accumulation of an intermediate, such accumulation being facilitated in the uncoated vessel. PRISCETOS,NEW JERSEY
CHEMISTRY, UNIVERSITY
OF
SOUTHERN CALIFORNIA]
The Mechanism of Decomposition of Borine Carbonyl BY ANTONB. BURG RECEIVED NOVEMBER 26, 1951
+
The rates of decomposition of B H C O and BDaCO, by a nearly homogeneous reaction of the type 2BHaCO i?BzH6 2C0, BHICO + BzHs CO, the BHs groups become intelligible on the assumption that the rate-determining step is BHs being furnished from the initial equilibrium BH&O 4 BHI CO. This mechanism implies a rate equation of the form
+
+
+
2kKt = -- In = f(x), which is confirmed by linear graphs of f(r)us. t for BHaCO a t 0-30" (loglo kK = 14.087a-x a-x SOOO/T) or for BDICO a t 10-26' (log,, kK = 12.352 - 5370/T). The corresponding AH values, 27.5 and 24.6 kcal., repreX
senting the sum of AH for the initial dissociation and AH for the activation of the rate-step, are rendered uncertain by a minor wall-reaction, small increases of kK with decreasing pressure, and the question of how well the first-step equilibrium is maintained. The equilibrium constants for the over-all decomposition are given by logto K,t, = 7.104 - 1998/T (AF;! = 9142 32.5T) for BHaCO, and by loglo Kat, = 7.097 - 185O/T ( A F ; = 8465 32.5T) for BDICO, both in the range 25-80'. These imply that the decomposition-rate studies were not complicated by the reverse reactions, since the runs terminated at 8-25% decomposition.
-
-
The original description of borine carbonyl' included a preliminary study of the rate of the decomposition reaction 2BH3CO -+ B& 2c0 (perfectly stoichiometric), with the suggestion that the first step might be the establishment of the equilibrium BHsCO 4 BH3 CO as a very rapid process. The further suggestion that the rate-determining step might be 2BHa -+ BzHewas not seriously inconsistent with the fairly rough original rate-data, but more recent studies have shown that a different theory of the rate-process is required. The new data become orderly if it is assumed that the second step is BHa BHJCO -+ GO, and that this step is SO slow as not to have any serious effect upon the maintenance of equilibrium in CO. Thus if K the first step, BHaCO & BHs is the equilibrium constant of the first step, and k the rate constant of the second, and it is noted that the total reaction yields two units of CO per unit of the rate-determining step, the rate equation is
+
+
+
+
+
(1) A ( i937).
B. Burp and 13'. I. Sc!ilesinger, THISJ O U R N A L , 69, 780
in which t is the time, a is the initial pressure, and x is the partial pressure of carbon monoxide (twice the observed increase of pressure, and equal to the loss of partial pressure of BHsCO). Integrating between 0 and t, and 0 and x , gives the more usable X a form 2kKt = -- log, =f ( x ) . a - x It is recognized, of course, that this equation cannot describe the initial rate, because it implies that dx/dt is receding from infinity as the decomposition begins, and it is clear that the equilibrium BH3CO tst BH3 CO could not be maintained against a very rapid drain of BH3 groups by the second step. Experimentally, it is very difficult to judge the nature of the early reaction, for the initial development of carbon monoxide is so fast even a t 0' that the pressure a, corresponding to zero time, cannot be estimated directly. However, on termination of a run by freezing-out, the value of a may be determined from the measured volume of CO
+
