MECHANISM OF THE PYROLYSIS OF PROPYLENE
1133
Mechanism of the Pyrolysis of Propylene: The Formation of Allene
by Akira Amano and Masao Uchiyama The Department of Applied Chemistry, Tohoku University, Serdai, Japan
(Receiued December 2 , 1963)
~~
I3ascd on the evaluation of reaction rates of a number of related elenientary reactions, the following free-radical chain mechanism is proposed for the pyrolysis of propylene a t the conditions under which ordinary cracking reactions are practiced: + allyl 11, EI C&6 + n-propyl*, H C3H6 -t H, allyl, n-propyl* + CzH4 inethyl, riiethyl C3H6 + CH, allyl, allyl C3H6+ polymer €1, allyl + C3Hj 11, allyl methyl --t C,H,, arid allyl H + C3H6. The proposed rnechanisrii reveals the existence of two distinctly different reaction zones. A t lower temperatures and higher propylene pressures, the reaction is characterized by the formation of higher boiling niaterials and it obeys three-halves-order rate law. At higher temperatures and lower pressures, however, the formation of allene becomes important and the reaction obeys first-order rate law. Consistent results are obtained for each of the zones mentioned when existirig data are critirally analyzed for both product distributions and reaction rates in ternis of the proposed niechanism. The analysis also indicates that the value of 78 kcal./niole forrnwly assigned for D(ally1-H) may be in error by a considerable amount. The formation of allenc by a severe pyrolysis is discussed quantitatively resulting in a good agreement with observations reported recently.
+
+
+
+
Introduction Although the pyrolysis of propylene has been a subjcct of many modern investigations, a number of controversial views hare bceri raised for reaction niechanism depending on their experirnerital results obtained a t different reaction conditions. According to Szwarcl the reaction was first order and its main products were allene, ethylene, and methane; whereas according to Laidler and Wojciechowski2 it was threehalves order with no cvidencr of the allenc formation. The disagreement such as typified above is inherent in the olefin pyrolysis in which decomposition and polymerization are simultaneously taking place in varying proportions according to the conditions used. In the present paper, certain elementary free-radical reactions are selected on the light of thc values of A factor and activation energy for a number of related elementary reactions which have been discussed in our recent paper.3 Table I shows thc values of those kinetic parameters upon which the present discussion is based. The mechanism consisting of the selected elementary reactions is capable of elucidating both decomposing
+
+ +
+
+ +
+ +
and polymerizing phascs of the pyrolysis. I t thus offers in the most unified manner an explanation of the propylene pyrolysis a t cracking temperatures ranging from 550 to 900". The niechanisni is then tested for both product distributions aiid reaction rates hitherto reported with spccial reference to the formation of allcnc.
Pyrolysis Generalized Mechanism. Rased 011 the evaluation of reaction rates using those kinetic paranieters listed in Table I, the following reactions can be selected in a broad range of the ordinary cracking conditions.
-
CBHe
allyl
+H
+ C3H6 -+-n-propyl* €I + C3H6 H, + allyl n-propyl* -+ CzH1 + methyl methyl + C3H6 --+CH1 + allyl H
.--)
(1)
(5) (6)
(7) (11)
(1) M . Szwarc, J . Chem, Phys., 17, 284 (1949). (2) K. J. Laidler and R. W. Wojciechowski, Proc. Rog. SOC.(London), A259, 257 (1960). (3) A. A m m o and .M. Vchiyama, J . Phys. C h a . , 67, 1242 (1963).
Volunae 68,Number 6
M a y , 1904
AKIRAAMANOAND MASAOUCHIYAMA
1134
Table I : Kinetic Parameters of Related Elementary Reactions" Reactions
--
log Ab
+H + methyl + propyl
C& 4 allyl C3& vinyl 2CaH6 allyl
+ +
--
(1) (15) (2) (15.6) (4) (8.5)
H CaHB n-propyl H CsH6 H2 allyl n-Propyl C2H4 methyl n-propyl- C3H6 H n-Propyl + Hz allyl Methyl Ht CH4 H Methyl CsH6 CH4 allyl Methyl C3H6 butyl Allyl Hz C3He H Allyl C3H6 .-,polymer H Allyl -L C3H4 H
(6) (7) (8) (9) (10) (11) (12) (13) (14) (15)
2 allyl -., diallyl Allyl methyl -+ C4Hs Allyl H 4 CaH6
(16) (17) (18)
-
+ +
+ + + +
--
+ + +
+
-+ +
-+
+
+
+ +
(5)
E , kcal./mole
99.2 - &" 91.0 62.5 - Qc
10.1 9.7 13 4 14.1 (13) 8.7 7.8 9.0 7.9 (6) 13.2
1.5 1.5 30.4 38 4 (30) 10 7.7 8.8 7.3 (15) 50.2
(7)
(0) (0) (0)
I
(8)
(9)
+ &"
+ Qc
The values in parentheses are those which are less accurate. for unimolecular reactions and l./mole sec. for bimolecular reactions. Q denotes the stabilization energy in allyl radical. a
* Units are sec.-l
allyl
+ C3H0-+
polymer
+H
+H
allyl ---+C3H4
+ methyl +CIH, allyl + H -+ CaHe
allyl
*ooor----77 (14)
(15) (17) (18)
The asterisk indicates a hot radical. In the present case, the n-propyl radical produced by reaction 5 has an excess energy of about 35 kcal./mole corresponding to the amount of heat evolved by the reaction. The excess energy can then be utilized in part for its subsequent decomposition (reaction 7). Characteristic features of the hot n-propyl radical have been described in our recent paper.3 The proposed scheme is a freeradical chain mechanism in which the chain is initiated by reaction 1 and terminated by reaction 17 and/or reaction 18. In the above scheme, reaction 14 and reaction 15 are competing in propagating the chain. The former reaction represents the polymerization, whereas the latter the decomposition to form allene. The relatire iinportance between these two reactions will be determined by the ratio k15/k14(C3HB)and hence by both temperature and pressure a t which the pyrolysis is studied. The ratio can be calculated by using the values of kinetic parameters assigned for these reactions assuming Q = 20 kcal.;mole. As is illustrated in Fig. The Journal of Physical Chemistry
1, the two zones characterized by the polymerization and the decoinposition are separated by a smooth curve corresponding to 2k16 = k14(C3&) on the pressuretemperature diagram. The curve shifts toward the left on the diagram if smaller values are assumed for &. Included in Fig. 1 are points corresponding to the conditions under which actual runs were carried out by a number of investigators to indicate that most of the studies were confined within the low temperature polymerization zone. Low Tenzperuture Zone. The mechanism of the pyrolysis in this temperature range can be obtained if reaction 15 is ignored in relation with reaction 14 in our generalized mechanism proposed above. A similar mechanism was proposed by Laidler and Wojciechowski2 although they included reaction. 9 instead of reaction 6. These two reactions are kinetically indistinguishable. Kinetic data available to test the proposed mechanism include those reported by Ingold and S t ~ b b sLaidler ,~ and Wojciechowski,2Amano and Uchiyama, and by Szwarcl for his runs carried out at lower temperatures (Fig. 1).
-
0
..r, Polyrnerlrotlon zone 4
1
600
800
I
Dewmposit ion Zone
1000
1
fern pero t u re, *C Figure 1. Pressure-temperature diagram: solid line, 2kI6 = k14(C3H6); empty circles, Laidler and Wojciechowski; filled circles, Ingold and Stubbs; empty triangles, Amano and Uchiyama; filled triangles, Szwarc; squares, Kunichika and Sakakibara. (4) K. U. Ipgold and F. J. Stubbs, J . Chem. SOC.,1749 (1951).
1135
,MECHANISM O F T H E PYROLYSIS O F P H O P Y L E N E
Assuming the relation kb = 0.4k6 reportcd by Darwent and Roberts,s the over-all stoichiometry of the proposed schernr can be expressed by the equation
-
3.8(Cs€Ts)
0.4(H2)
Laidler and Wojcierhowski' rcportcd that thc amounts of hydrogen, methane, and cthylentl fornied wcrc in the ratio of approximately 1:2:2, which is in good agrtcnient with the product distribution cxpwtcd from the equation. The similar product distribution has been rcportcd in nunierous other works. Sirice the amount of higher boiling products in the above equation corrcsponds to 1.4 moles of Ce hydrocarbons, about 74% of the carbon atoms produced by the pyrolysis should be recovcrcd in thc higher boiling fraction. This was experimentally observrd to be t50-7070 by Ingold and Stubbs4 and 30-55% by Laidler and Wojciechonski. The discrepancy may be caused by either the subsequent decomposition of the higher boiling materials or the coke formation. Ry applying steady-state approximation to the proposed scheme, two different over-all rate cxpressions can be obtaincd depending on the modt of tcrmination. If thc chain part of the reaction is controlled by rcaction 11 and is therefore terrninated by reaction 17 (case I), thc over-all rate of propyleiie consumption can be expressed by the equation -d(C3Hs)/dt
=
a t
1.0
-
0
b
Y
- as P
0
lb
20
25
3.0
log p Figure 2. log (rate, mm./rnin.) us. log (initial pressure, mm.) relationship: a, 650'; b, MOO; c, 630'; d, 620'; e, 610'; f, GOO'; g, 590'.
(361klkllk14/35k17)"1(C3H6)1/1
If, on the other hand, propagation of the chain is controlled by reaction 5 and terminated by reaction 18 (case 11), the ratc can bc cxprcssed by the equation
- d(C3HG)/dl
1.5
+ (CHI) + (C211d + polynier
= (361kik&14/35k18) "*(C3H8)'I1
Thc reaction is thus, in agreement with Laidler and Wojciechowski,2 three-halves order independent of the mode of teimination. I n analyzing their data on the premise of first-order kinetics, Ingold and Stubbs4 stated that the reaction order gradually changed from first to second as the initial pressure was decreased. By recasting raw data reported by Ingold and Stubbs to log (rate) vs. log (initial pressure) relationship, however, parallel straight lines are obtained for six different temperatures studied. This is illustrated in Fig. 2. Since the slope of these lines corresponds to about 1.4, their results can also bc taken to indicate the three-halvcs-order reaction. Szwarc' and Amano and Uchiyama3 have also calculated rate constants a t temperatures ranging from 680 to 870" assuming thc reaction to be first order. As has already bcen discussed, however, most of these runs were carried out in the range of conditions where
the order of the reaction should cffcctivcly be threehalves. As will be seen later, an excellent fit to the Arrhenius equation is in fact obtained when the values of their first-order rate constants are readjusted for the three-halves-ordcr rate law. A further support for the proposed ratc law can be obtaincd, if kinetic data reported by Sxwarc' in his runs a t 780" are used to calculate the rate constants. The valucs calculatcd are compared with the original figures in Table 11. As was pointed out by Szwarc, there is a definite trend in the values of the Szwarc's constant (kl) with respect to the change in initial pressure. The values of three-halves-ordcr ratc constant (k*,J are more consistent. Three runs reportcd by Szwarc a t pressures below 5 nini. are interitionally omitted from the list of the table, since they belong to the borderline case. Illustrated in Fig. 3 arc the Arrhenius plots of the three-halves-order rate constants obtained from the four different sources discussed above. The solid (5) B. deB. Darwent and R. Roberts, I)iscztsawns Faraday Soc., 14, 55 (1 953).
Volume 68, hrumher 6
M a y , 1964
1136
AKIRAAMANO AND MASAO UCHIYAMA
and is characterizcd by the formation of an appreciable amount of allcnc as is indicated by the over-all stoichionictric rclation
Table I1 : Comparison txtween First-Order and Three-Halves-Order Rate Constants
Run
Pressure, mm.
Contnrt tiwe, aec.
Conver8ion,
ki X 108,
%
set.-'
6.3 8.2 8.3 11.3 14.4
0,580 0.144 0.144 0.448 0.124
0.86 0 21 0.23
0.88
14.8 14.8 15.8 19.7
0.25
19 . 9
15
9 11 14 10
kai,. l./inole sec.
t
r
I
1
-d(C3IIe)/dt
-
D
-
EoA
0.0
-
9
Y
- V 0
-8D
0
-4a
I
0.9
I
1.0
I
1.1
I/T.x 1 0 3 Figure 3. Arrhenius plots o f t,he tIirre-li:tlves-order rate constants: cnipty triangles, Amano and Uchiyama; filled triangles, Sawiirc; erript,y caircles, Laidler and Wojciechowski ; filled circles, Ingold and Stubbs.
The Journal of Phvsienl Chemistry
+ (CH,) + (C2H.t) + 1.4(CJI4)
By applying stcady-state approximation to the above mechanism, the follou-ing rate equation can be obtained.
1.41 1.25 1.53 1,52 1.21
lines in Fig. 3, corrcsporiding to the theoretical ratc equations derived from thc proposcd niechanisni, would indieatc a gcncral agrccmcrit bctwccn obscrvcd and calculatcd rates. Rceausc of a bcttcr fit to thc experiments, the casc I nicchanisni is favored a t prcsent. High Temperaturs Zone. At higher temperatures and lower pressures, unimolccular decomposition of allyl radical bccomos increasingly important. The mechanism in the deconiposition zone can be obtained when reaction 14 is ignored in relation with reaction 15 in our generalized iiiechanisni. The resulting mechanism also involvcs a free-radical chain
40r
2.4(C3Hd -'0.4(Hz)
I
=
(144kik6ki6/35k.18)"2(C3H6)
Thus thc mechanistic transition from the low to the high temperature zones is reflected in the change of the reaction order. As can be secn in Fig. 1, data available to test the proposed high temperature niechanisni are scarce. Furtherniorc such a sirnplified rricchanism is only practicable in a limited teniperaturc range. At temperatures exceeding about goo", the splitting of propylene to vinyl and methyl radicals may bccome important and a large number of sceoridary reactions will also be involved. In agreement with the above stoichiometry, Szwarc' reported that hydrogen, methane, ethylene, and allene were formed in the mole ratio of approximately 1 :2 : 2 :3 for the products obtained a t about 860' and 8 mm. In spite of this agreement, our criticism to his paper lics in the fact that Szwarc analyzed his data covering a tempcraturc as low as 680" based exclusively on the first-order rate equation. All thc experimental runs, except those a t his highest temperature mentioned abovc, were made a t conditions where the three-halvesorder ratc equations should be used. The applicability of the latter rate law has been demonstrated by the linear plots illustrated in Fig. 3. At 857", which is well within the decomposition zone, the first-order rate constant recorded by Szwarc' was 0.13 scc. - l , while the theoretical value based on our cquatiori is found to be 2.5 sec.-l. I t should be noted that the kinetic analysis given by Szwarc was based on data obtained a t contact tirncs of about 0.120.15 sec. which are not much longer than the induction period (cu.0.03 sec. at 850") observed by Amano and Uchiyama.a The omission of the induction period in Szware's treatment as well as the overestimation of Q in our calculation may be major factors of the discrepancy between observed and theoretical rate constants. By the reasons described above, the value of 78 kcal./rnole assigned by Szwarc for the bond dissociation energy between allyl radical and hydrogen atom is soniewhat doubtful. The latest figure suggested is about 84 kcal./mole.6 (6) S. W. Benuon, A. N . Bow, and P. Nangia, J. A m . Chem. Soc.. 85, 1388 (1963).
1137
MECHANISM OF THE PYROLYSIS OF PROPYLENE
Recently Kuriichika and Sakakibara' observed appreciable amounts of allene and methylacetylene in the reaction products a t 100 mm. and temperatures ranging from 800 to 1400". Although both the low and the high temperature mechanisms are working simultaneously under the given conditions, the amount of allene expected in the products can be calculated by using two stoichiometric equations and the ratio kl6/kI4(C3HB)assuming Q = 20 kcal./tnole. The amount of allene calculated in the manner described above is listed in the second column of Table I11 for three different temperatures. Those observed values, also listed in the last column of the table, are the sum of allene and methylacetylene. A reasonable agreement is observed between the observed and the calculated values a t lower temperatures a t which any serious mechanistic complications are not expected. A t the highest temperature listed, however, the amount of allene observed becomes considerably less than that calculated by the present mechanism. The discrep
Table I11 : Formation of Allene at Varying Temperatures Temp.,
Allene formed/propylene consumed, mole %
7-
Calcd.
Obsd.
800
8
900 loo0
36
12-15 29 30-34
OC.
54 ~~
ancy may be caused by the complication in the mode of initiation as well as the involvernent of secondary reactions which would consume unsaturated compounds.
Acknowledgment. The authors wish to thank Prof. S. Kunichika and Prof. Y. Sakakibara of the Institute of Chemical Research, Kyoto University, for providing their unpublished data. (7) S. Kunichika and Y. Sakakibara, 14th Annual Meeting of the Chemical Society of Japan, April, 1961 ; private communications.
Volume 68,Number 6 M a y , 1964