T H E OXIDATIOS OF ETHAYE BY H. AUSTIN TAYLOR A N D E.
w.RIBLETTI
The more interesting case of the two types of explosive reaction recently considered from a theoretical standpoint by Semenoff2 is naturally that due to branched chain reactions. The differentiation of this type from the purely thermal explosion is not however always easy, especially where in highly exothermic reactions a thermal explosion may actually precede the chain explosion. From the evidence already accumulated on the explosions of saturated hydrocarbons in oxygen there seems every reason to suspect that these are a t least in part due to chain reactions. The succeeding work is an attempt to show that even the slow, non-explosive oxidation of ethane is probably a chain reaction and that therefore the explosive oxidation may be so considered also. The latter conclusion is verified by a study of the critical explosion pressures at different temperatures. During the progress of the work there appeared a publication by Bone and Hill3 on the same subject with conclusions similar to those just stated. Stress however had been laid particularly on the slow reaction and of this, the induction period received special attention. They showed that the induction period could be reduced b y traces of several compounds such as water, alcohol and iodine. The elimination of the induction period is extremely helpful in the study of the rate of the slow reaction but i t seems doubtful whether the mechanism of the reaction in its absence, is necessarily precisely the same as when the induction inhibitors are absent. As Semenoff has shown in the sulphur oxidation the addition of small amounts of ozone is alone required to start chains which in its absence would never occur a t the particular temperature. Bone and Hill further claim that no oxidation of the ethane occurs within the induction period and that there is an inconsiderable peroxidation if any during the slow reaction immediately following the induction period. That absolutely no reaction occurs during the induction period is difficult to maintain unless some cause for its sudden initiation is forthcoming. What would seem more plausible is that reactions do occur to extents that are not measurable analytically or alternatively from a pressure-change standpoint mutually balance each other. I n the reactions here described ethane was shown to decompose with an increase in pressure a t a rate that was much less than the oxidation rate. A simultaneous addition of oxygen to the ethane molecule with a corresponding pressure decrease might easily account for the absence of any total pressure change during the Abstract from a thesis presented in partial fulfillment of the requirements for the degree of Doctor of Philosophy at New York University. 2 Chemical Reviews, 4, 347 (1929). 3 Proc. Roy. Soc., lZPA, 434 (1930).
2668
H. AUSTIN TAYLOR A S D E. W. RIBLETT
induction period in an ethane-oxygen mixture. It is true that the temperatures of Bone and Hill's experiments were considerably lower than those used here so that the ethane decomposition would have been much reduced if it occurred at all. Against this however their induction period a t 316" C was about thirty minutes whereas the average time found by us at 46ooC was about two minutes. This ratio of thirty to two for a 44' temperature difference, corresponding to quite a plausible increase in a reaction rate, would lend credence to the above interpretation of the induction period provided the mechanism of the succeeding reactions involved chains necessitating only small quantities of material as initiators. Experimental. The ethane was prepared by the Grignard method from ethyl bromide. Dibutyl ether was used as solvent because of its high boiling point and could t'herefore be more easily removed from the ethane. The ethane was purified by passage through a coil immersed in a freezing mixture and stored over water. When required for use it was drawn through two towers of sulphuric acid. Oxygen was taken directly from cylinders of the compressed gas. The reaction system was composed of a pyrex bulb attached to a capillary manometer whereon the pressure changes during experiment could be followed. The manometer was heated throughout its length to avoid condensation of products during an experiment. The gases were admitted to the reaction system from an auxiliary holder serving also as a mixing chamber. The latter had its own manometer for use in preparing known mixture of ethane and oxygen. The reaction vessel was heated in an electric furnace, the temperature of which was measured by a platinum resistance thermometer. The evacuation of the reaction system was made by means of a mercury diffusion pump backed by a hyvac oil pump, the former being necessary for reproducibility of results. Determinations were made of the rate of reaction as measured by the rate of pressure change for various mixtures of ethane and oxygen a t a series of different temperatures. It would be impossible and useless to list here the results of the hundreds of observations made. Typical results are given therefore to illustrate the salient features only. Fig. I gives some of the results obtained for a fifty per cent mixture a t 470°C. It will be observed that reaction does not begin immediately after the admission of the gases into t'he reaction vessel, that is, no pressure change is observed. I t is this period to which the term induction period has been applied and in all the cases examined here varied up to four or five minutes being as stated on the average about two minutes. This induction period is followed by a period of slow reaction which however gradually develops in rapidity in a manner typical of an autocatalytic reaction, until completed. The duplication of the rate curves is extremely difficult since the induction period and the rate of development in the early part of the reaction seem to be excessively sensitive to traces of foreign substances including reaction products. T w o reactions may be carried out a t exactly the same pressure and temperature with the same composition of mixture and yet the time of development to rapid
THE OXIDATION OF ETHANE
FIG.I
T?lE
IN
FIG.2
PIrNNJ.
2 669
2670
H. AUSTIN TAYLOR AND E. W. RIBLETT
reaction may vary by as much as one hundred per cent if extreme care is not taken in a thorough evacuation of the reaction vessel. I t was observed however that despite this fact, the rate of the rapid reaction is almost constant. Fig. 2 will illustrate the point in that the slopes of the two curves in their steepest portions are the same, whilst the times for total reaction differ by almost fifty per cent. As Bone has shown, whatever happens during the early part of the reaction, oxidation certainly occurs during the rapid reaction and one is enabled therefore to judge comparatively of the effects of pressure, temperature and composition by a study of the rapid reaction rate. Efect of Pressure. h comparison of the slopes of the rapid portions of the rate curves for given mixtures a t fixed temperatures shows in general that the ratio of these is proportional to the ratio of a power of the pressures lying between two and three, a result which does not seem to be affected by a change in temperature or by a change in composition even if one reactant be in large excess. Thus a t 470°C. a C2H6:502mixture gave for pressures o 429,419,471 and 493 mms. slopes in the ratio of 1:1,2:1,3:1.4 whilst the ratio of the squares of the pressures is r : I , I I : I , ~ z : I . ~ ~ . K i t h regard to the effect of pressure on the total pressure change during the reaction for a given mixture the ratio of total is initial pressure is always constant. Thus in the data just given the total pressures developed during reaction were 479, j 0 4 , 5 2 8 and g j z mms. The ratio of each of these to its initial pressure is 1 . 1 2 . I t is probable then that a change of initial pressure does not affect the composition of the end products for a given mixture. I t should be mentioned here that a t no time during the experiments was a deposit of carbon observed, the reaction vessel being quite clean even after numerous runs had been made, ind-icating a clean oxidation to gaseous products which could be pumped out. Injiuence of Temperature. Assuming for the present that the reaction is a chain reaction, definite evidence will be given later, the energy of activation calculated from the observed temperature coefficient will be a composite quantity since the total rate of reaction is a function not only of the number of chains initiated but also of the chain length. The effect of pressure on the latter will depend on the cause of the breaking of chains, that is, whether it occurs chiefly in t,he gas phase or on the surface. Assuming that the majority of chains are broken on the surface an increase of pressure would be expected to increase the chain length, an increase of temperature on the other hand might shorten it, with the result that the energy of activation would decrease with an increase of pressure. Such is what is actually found here. For a C2H6:3%02mixture the energy of activation calculated from the ratio of the maximum rates is found to be at zoo mms. pressure, 43,000 calories, at 2 1 5 mms. 38,000 calories and a t 300 mms. 32,000 calories. From the general magnitude of these values it would seem probable that the actual chain length is not great. Efect of Composition. I n drawing conclusions regarding the effect of composition on the reaction rate it should be fully realized at, the outset that such will only be approximate since there is no criterion that exactly the same re-
THE OXIDATION OF ETHAKE
2671
actions are occurring under the different conditions. The fact that the total pressure change per unit initial pressure is approximately constant for different mixtures may be taken as evidence of the absence of at least drastic differences in reaction. Thus a 3CpH6:Ozmixturegives a ratio of total pressure developed to initial pressure of I. 16, a CpHs:Ozmixture gives a ratio of I .23,a C2H6:3M02 mixture a ratio of 1.22 whilst a C2H6:50~mixture gives a ratio of 1.12, all a t 4 7 O o c . The ratios are all of the same order but too much reliance cannot be placed on this fact since theoretically the complete combustion of ethane to carbon dioxide and water would yield a ratio of I . I I whilst if equal amounts of carbon monoxide and dioxide are produced along with the water the ratio is only 1.2 j. It is very possible therefore that quite different reactions may be occurring with different compositions. The specific effects of oxygen and ethane however are sufficiently different to warrant their statement even if slight differences in reaction do occur with changing composition. The general effect of oxygen appears a t first sight to be approximately proportional to its pressure when in excess. Thus at 450' C, 80 mms. C2H6 and 2 8 0 mms. 02 react in their rapid range a t 2 7 mms. per minute, whilst 80 mms. C2H6 and 400 mms. 0 2 react at 40 mms. per minute. As was pointed out previously however an increase of total pressure increases the rate proportional to the second or third power of pressure. An increase in rate proportional to the first power of the oxygen therefore actually corresponds to a retardation when oxygen is in excess. This effect parallels that of nitrogen mentioned later and may also be true of ethane when present in large excess as shown later in the study of explosion limits, The latter however is not so definite probably because of the possible pyrolysis. The accelerating effect of ethane a t constant oxygen pressure is much more pronounced, so much so that it is difficult to avoid explosion unless small pressures of oxygen are used when the reaction is extremely slow. Thus at 46ooC, 1 2 6 mms. of each gas reacted at 2 8 mms. per half minute whilst with 1 2 8 mms. of 02 and 38 j mms. ethane the reaction was extremely rapid corresponding to more than IOO mms. per half minute although no flash or sound typical of explosion could be seen or heard. Such results are typical of many that were obtained. Evidence for Chain Characteristics. Chain reactions are characterised by (I) extraordinary sensitivity to traces of foreign substances ( 2 ) marked negative wall effect (3) explosive possibilities (4) specific effects of diluent gases ( j ) induction period. Ethane oxidation shows all these characteristics. The period of induction has already been discussed in part. It seems fairly certain that it is a very definite part of the reaction although it may be markedly affected by impurities including the reaction products. Even after extreme purification of the system however it is still present. As pointed out previously the variation in the length of the period with temperature, if comparison between the present work and that of Bone and Hill is reliable, is of the same order as the increase in reaction rate with temperature as found by us. This would suggest that during this period there is a gradual accumulation of some substance which is required for the complete oxidation reaction. As will be
2672
H. AUSTIN TAYLOR AND E. W. RIBLETT
seen from the later work on the critical explosion pressure, the induction period together with some slow reaction always precedes the explosion. I n other words for explosion it is not at all necessary that the mixture react immediately on admission to the reaction vessel. Fig. 3 illustrates the sensitivity of the reaction to foreign substances and also shows the negative wall effect. I n the diagram the normal reaction a t 450' C is seen to be complete in about 8 j seconds, which means that it is just below the explosive point. With fifteen per cent nitrogen the reaction is slower taking about three minutes for completion. The addition of glass
FIG.3
powder to the reaction vessel changes the time for reaction to almost five minutes, whilst if the powdered glass is coated with potassium chloride the reaction is almost completely inhibited and a t 550' C. is still incomplete after forty minutes. I n all of these cases there is evidence from the changing total pressure increase of differences of mechanism under the different conditions, a result shown definitely by Pease' in the oxidation of propane and the butanes. An attempt to induce reaction by the addition of ozone, used by Spence and Taylor2 in the oxidation of ethylene, was not entirely successful in that explosion completely destroyed the apparatus each of the several times trials were made. The explosions were however significant. The method for the experiment was t o flow ethane and oxygen at equal rates into a Y tube and thence through a furnace a t temperatures ranging from zoo to 400'C. Under such conditions the amount of oxidation was almost negligibly small. Upon
* J. Am. Chem. Soc.,
51, 1839 (1929). J. Am. Chem. Sac., 52, 2399 (19.30).
2673
THE OXIDATION O F ETHANE
exciting the ozonizer in the oxygen line explosion occurred. Visibly, the explosion seemed to occur in the Y tubes where the gases met and would strike back away from the tube in the furnace. The actual temperature of the Y tube could not have been much above room temperature, yet the remains after explosion could not be found despite the fact that the tube in the furnace always remained intact. It seems quite definite therefore that ozone will induce explosion in ethane-oxygen mixtures at temperatures very considerably lower than those necessary in its absence. Esplosire Limits. The reaction of ethane and oxygen may lead to an explosion at any of the temperatures studied provided the initial pressure of the reactants is large enough. Semenoff, as mentioned, has suggested that the critical pressure above which a reaction is explosive is a function of the temperature given by the equation log I-, = A , T
+B
The magnitude of B depends on the composition of the mixture, dimensions of the apparatus and the presence of foreign substances. The constant A is independent of these but for its interpretation the precise type of explosion must be known. For a purely thermal explosion h is directly proportional to the energy of activation whilst for a chain explosion it is solely a function of the energy necessary to cause the branching of chains. In the determination of the minimum explosive pressures the temperature of the reaction vessel mas kept constant and a series of experiments carried out at different pressures for a given mixture. The pressures were so varied that the limits in which explosion did or did not occur gradually approached each other. Thus the minimum explosive pressure could be determined with an accuracy of from one to four millimeters. The data obtained are given in Table I.
TABLE I Temp. "C. 83.33 546 495 3 90 367
-
lllinimum Pressure for Oxygen Concentrations. 78 -
j7
334
66 67
jj
00
50
00
40
00
-
2.57
-
-
-
238
193
186
182
191
-
-
-
-
-
-
133
108
104
103
-
-
-
320
-
211
164
I45
-
I12
-
91
64
67
54
-
00
-
309 248 187
266
60
53
-
-
50
-
IIO
-
2674
H. AUSTIN TAYLOR AND E. W. RIBLETT
The values of the logarithms of these pressures when plotted against the reciprocals of the corresponding absolute temperatures all yield straight lines which are parallel giving therefore a value of A of 4 0 2 2 which is independent of the composition. To show that the value is also independent of the dimensions of the apparatus the cylindrical vessel previously used was replaced by a spherical one with a correspondingly smaller surface. The slope of the log p - I / T line was unchanged although on account of the decreased surface the actual pressures for the spherical vessel were lower than those for the cylindrical, as would be expected from the negative wall effect. The value of A interpreted as for a purely thermal reaction would yield approximately 40,000 calories for the energy of activation of the reaction. Comparison with the figures given above shows this to be of the order of magnitude actually found. The reaction however is most certainly a chain reaction and a value of A of 4 0 2 2 for a pure chain explosion would correspond with an energy of about IO,OOO calories to cause chain branching. The measured temperature coefficient of the slow reaction may possibly be a composite of this latter value and a true energy of activation which could conceivably be very much larger. The previous agreement might therefore be merely fortuitous. On the other hand the time taken for a given mixture to explode after its admission t o the reaction vessel was sometimes of the order of five or six minutes though more frequently around one or two minutes. Such a long period as five minutes would suggest that an acceleration of rate due to heat accumulation in the system was probably preceding the true explosion eventually caused by increased chain length. In such an event the value of A cannot be readily interpreted at all. In a similar study to the above by Sagulin' a value for h of 4900 was obtained] which changed abruptly to 7000 as the temperature was raised above 68ooC. The temperatures used by Sagulin were all above 60ooC. I n view of the complexity of reactions possible, as shown by Pease for propane this abrupt change is quite plausible since dissociation plays a more important part a t higher temperatures. A pure oxidation to carbon monoxide and other oxidized products is the more important reaction a t lower temperatures, whilst the formation of unsaturated compounds due to partial oxidation is intermediate between the two. I t would not be surprising then if a second change of slope should occur a t lower temperatures than the one observed by Sagulin. It is certain that the curve for a C2H6:502mixture shows a break a t about 470' C, which can be seen from the data in Table I. The high pressure necessary for explosions a t still lower temperatures, disregarding their violence, precluded the possibility of determining the slope of the new curve. Finally an examination of Table I will show that there is a t each temperature a particular composition of mixture which possesses the smallest minimum explosion pressure. The minimum is not particularly sharp when plotted but can be seen to be in the neighborhood of the fifty per cent mixture. 2. physik. Chem., l B , 27j (1928)
THE OXIDATION OF ETHANE
2675
Furthermore this value is independent of temperature over the range studied. Taken in conjunction with the fact that the ratio of pressure increase to initial pressure is at its highest value in this region it would appear that the oxidation was more efficient for such a mixture.
Summary The oxidation of ethane has been studied by a static method and shown to be a homogeneous chain reaction. The effects of pressure, temperature, composition, surface and diluents have been shown to substantiate this. The critical explosion pressures determined are shown to be in agreement with Semenoff's theory but seem best explained as chain explosions preceded by thermal accelerations. Ntchols Chemical Laboratory, 'Yew York Unauerwty, Xew York, A'.
Y.
