232
Ind. Eng. Chem. Process Des. Dev., Vol. 18, No. 2, 1979
Mechanism and Kinetics of Propane Pyrolysis Stephen K. Layokun" and David H. Slater Department
of Chemical Engineering and Chemical Technology, Imperial College, London, S. W.7., England
The pyrolytic behavior of propane is followed between 600 and 800 OC with a view to elucidating the mechanism and kinetics. This is effected by continuous sampling of the pyrolytic products through an MS10J.22 mass spectrometer and the results are compared with computer predictions. Quantification of the products is done by resolving the superimposed mass spectra with the aid of a novel matrix-inversionprogram (written by the authors). The numerical integration program employed is based on the semiimplicit trapezoidal rule. A 19 reaction step model based on generation and interaction of free radicals is proposed with the appropriate rate constants and integrated. The main products are CH4, C2H,, C3H,, H,, and C2Hs. The predictions fit well the experimental data giving an activation energy at 245 kJ/mol and an overall reaction order of unity with respect to propane concentration.
Introduction Propane is employed commercially for the large-scale production of ethylene and propylene. The mechanism of the pyrolysis reaction of propane is, however, not fully understood. Recent studies indicate that the reaction is entirely free radical in nature, but beyond this, there is little detailed agreement. This must be due to complexities associated with self-inhibition reactions (Leathard and Purnell, 1968), tendency to polymerization, neglect of the effect of trace impurities, and inadequate allowance for the involvement of surfaces (Crynes and Albright, 1969). A number of investigations (Pease, 1928; Frey and Hepp, 1933; Steacie and Puddington, 1938; Bradley, 1974) also indicated that the reaction is largely homogeneous and that the main decomposition products are H2, CH,, C2H4, and C3Hsaccording to the stoichiometry C3H8 C3Hs + H2
-
and CBH,
----*
C2H4
+ CH4
Ethane, butadiene, and heavier components are formed especially at higher conversion and higher temperatures with long reaction times. In general when considering the pyrolysis of propane and heavier hydrocarbons it is convenient to consider the reactions as occurring in two stages. (a) The first stage is designated as the primary reactions wherein the reactants are decomposed through the freeradical chain mechanisms into the principal primary products: HP,CH4, C2H4,and/or propylene, and olefins up to C,-l where n is the number of carbon atoms in the feed hydrocarbon. In the cases of propane an isobutane substantial yields of C, olefins are also realized, but from other paraffinic hydrocarbons, both straight chain and branched, the C, olefins are produced in only minor quantities. (b) The second stage encompasses secondary reactions which can be classified into three types: (1) reactions involving further pyrolysis of the olefins produced by primary reactions; (2) hydrogenation and dehydrogenation reactions where paraffins, diolefins, and acetylenes are produced from the olefins; and (3) condensation reactions wherein two or more smaller fragments combine to produce large stable structures such as cyclodiolefins and aromatics.
The two stages are successive only to the extent that the primary reactions must have progressed far enough to have produced quantities of products sufficient to initiate the secondary reactions. Thus both primary and secondary reactions are usually occurring simultaneously. A t low conversion level of the primary reactant, the secondary reactions are relatively unimportant. The olefins formed are more refractory than the feed and thus are being cracked at relatively low rates. Any condensation reaction of these olefins proceeds very slowly because of their low partial pressures. Secondary reactions only become significant as the conversion of the primary reactant is raised toward higher levels when cracking of the primary olefins takes place at an accelerated rate, acetylenes and olefins appear in increasing concentrations, and the production of benzene and other aromatics becomes substantial. Free-Radical Mechanism In a saturated hydrocarbon, initiation may occur by cleavage of either a C-C bond or a C-H bond. The strengths of C-C bonds in C2-Cs hydrocarbons lie in the range 325-350 kJ/mol, while C-H bonds have typical dissociation energies of 410-42'7, 393, and 381 kJ/mol for primary, secondary, and tertiary positions, respectively (Bradley, 1974). Even in the most unfavorable case the temperature-dependent terms will lead to rates of C-C scission a t least ten times faster than the corresponding C-H rupture at temperatures up to 1500 K. It is therefore valid to assume that the primary initiation process involves unimolecular rupture of C-C bonds in the propane molecule. The key initiation step must certainly be as follows CH?CH*CH, CH3CH2 + CH, (1)
-
Propagation includes such steps as CH,
+ C3H8
4
CH3 + C1HB
+ C,HB C2Hj + CZHs
C2Hj
+
+
----*
+ n-C,H;
CH,
+ L-CXH; C2H6 + n-C,H, C?H, + L-CBH? CH4
(24 (2b) (3a) (3b)
In addition, ethyl radicals undergo some decomposition C2Hj
+
C2H4
+H
(4)
and hydrogen atoms will engage in abstraction reactions H + C?H, Hl + n-C,H(54 +
* Direct all correspondence to this author a t the Department of Chemical Engineering, University of Ife, Ile-Ife, Nigeria. 0019-7882/79/1118-0232$01.00/0
H
+ CjHB
+
H? + L - C ~ H ~
(5b)
Hinshelwood (1956) reported that at pyrolysis temperaC 1979 American Chemical Society
Ind. Eng. Chem. Process Des. Dev., Vol. 18, No. 2, 1979 233 Nee dIe
c__
___..
Figure 1. Schematic diagram of the flow system.
proximately 1 cm surrounded by a vacuum of lo4 to torr. The reactor is coupled to its base via a Viton O-ring and the sampling probe passing through the base of the flange is adjustable to permit sampling from any zone in the reactor. The thermocouple is of chromel-alumel, and residual gases are let off into the fume chamber. B. Analysis. Instead of the usual gas chromatograph, the mass spectrometer MSlO-C:! is employed to identify and quantify the pyrolysis products. Identification is accomplished by comparing the product spectrum with calibrated spectra of pure samples. Quantitative analysis is based on the fact that Dalton’s law of partial pressures applies: in this case, that the contributions of several components to a specific mass peak are linearly additive. I.e., there is perfect linear superposition of all fragments of the same mass-to-charge ratio such that hllP1 + hl2P2 +
tures, normal and isopropyl radicals are rapidly interconverted and that the reaction scheme could therefore be simplified by writing either radical as C3H7. However, Lin and Laidler (1966) have shown that normal propyl radicals decompose mainly into ethylene and methyl radicals
+
+
The normal and isopropyl radicals are generated at rates consistent with eq 2, 3, and 5 . It should therefore suffice in any kinetic scheme to employ eq 6 and 7 as representing their decomposition behavior. It is obvious from the propagation steps that methyl and propyl radicals predominate in any propane hydrolysis reaction scheme, such that the most likely termination steps are CH3 + CH3 C2H6 (8)
.‘*
J
hnP1 + hzQz + , “’
hinlP1 + hinp, while isopropyl radicals decompose mainly into propylene and hydrogen atoms CH3CHCH3 H C3Hs (7)
f
9
+ hl,P, = H1 + hzdn = H2
+ , .’. , + hinnpn = H ,
where Pn = partial pressure component n in the inlet sample, h,, = measured height of peak a t mass m due to unit pressure of component n, and H , = measured height of peak a t mass m in the mixture mass spectrum. Here m > n, and it is reasonable to suppose that the most accurate solution would use all given data. Hence a technique involving the least-squares method is chosen (Layokun, 1975). If h , g 1 + h r g 2+ ,..., + h r d n= H,, the sum of the squares of the residuals is given by m
C ( h r 1 ~+ 1 hrG2, .*.
9
r=l
+ hrnpn - Hr)’
and if this is a minimum with respect to pi, then
-+
for i = 1, 2, ..., n. The differentiation results in a set of equations typified by in
Chr,(hrlp1+ h r g 2+ , ... , + h,,pn - H,)= 0
r=l
At high temperatures the likelihood of reactions 9b, lob, and l l b occurring looks remote. The pedominant reaction of propyl radicals will be that of decomposition.
Experimental and Analytical Setup A. The Flow Diagram. The flow diagram is shown in Figure 1. It consists of propane and nitrogen cylinders (from B.O.C. C P grade, 99.9%), flow meters, sinta glass, reactor and reactor base, and a sampling probe of very narrow diameter that conveys the pyrolysis products into the mass spectrometer. Propane and nitrogen are separately passed through a U-tube containing silica gel to remove any element of water. The sinta glass effects perfect mixing of the gases and positive flow eliminates fluctuation in flow rate. The needle valve and the tap T can also act as flame traps. The preheater is lagged with Triton kaowool. The reactor, which approximates to a plug flow, is made of quartz. It has a central bore of ap-
There are now n linear equations for the n unknown pL; Le., the normal equations are now obtained and are amenable to inversion. However, in practice, some peaks are known to be more stable than others; their peak heights H are correspondingly more reliable. This fact is taken into account in the least-squares method by first multiplying each equation by the square root of a number called the “weight” and then minimizing the weighted sum of squares of residuals. The weight u: = ( n - I ) / ? (r’: - m?z :=1
where n’ = number of readings, r, = values read, and m’ = mean of the n readings. Thus the quantity to be minimized is
2 [ f i ( h r l p l+ h r g 2 + ,..., + hmPn
-
r=l
and the normal equations are given by C[uirhri(hrlpl + h r g 2 + , ... , + h,,p, - H,)I = 0
r=l
234
Ind. Eng. Chem. Process Des. Dev., Vol. 18, No. 2 , 1979
Table I reaction 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
C,H, C2Hs + CH, CH, + C,H, + CH, + C,H, C2Hs t C,H, C2H, + C,H, C,Hs -+ C2H, + H H t C2H4 C2Hs H + C,H, --* H, + C,H, C,H, CH, + C,H, C,H, 4 H t C,H, H + C,H, C,H, CH, + C2Hs C,H, C,H, + C,H, C,H, t C,H, CH, + C,H, C,H, + CH, CH, + C,H, -, C,H,, C,H, + H, C,H, + H C2H, 4 CH, + CH, CH, t C,H, CH, + C,H, H + C2H, H, t C,H,
6 X lOI4
-+
-+
+
-+
-+
-+
-+
-+
1 . 5 X 10’’
2x 3x 7.5 x 1.8 X 4x 3.6 X 1.8 X 4.2X 1x 1x 1x 5.6 x 1.2 x 8 X
-+
C4H10
CZH,
--t C Z H S
+
CzH,
+
-+
C2HS
C4HIO
E
A
1O’O 10” 10” 10l2 1O’O
10” 10” lo1*
loLo lo’, loll
107
lo1, lo8
1.3 X 10” 8.7 x 109 1 x 10”
ref Herriot e t al. (1972) Bradley (1974) Herriot e t al. (1972) Kerr and Trotman-Dickenson (1970) Herriot et al. (1972) Kondratiev (1961) Herriot e t al. (1972) Bradley (1974) Herriot e t al. (1972) Herriot et al. (1972) Pratt (1969) Layokun (1975) Herriot e t al. (1972) Herriot e t al. (1972) Bradley (1974) Bradley (1974) Pratt (1969) Purnell and Quinn (1962) Benson and O’Neal (1970)
326 20.5 25 146.3 22.60 19.20 133 133 4.18 10 0.0 0.0 0.0 92 248 50 37.6 167.2 6.0
for i = 1, 2, 3, ..., n. Weighting may not be required if the peaks are stable, but since the sensitivity of the mass spectrometer affects the sensitivity of the peaks, it is essential to determine the sensitivity of all peaks in a pattern. The sensitivity of the largest peak in the spectrum can be taken as the sensitivity of all the peaks provided the cracking patterns are stable. Thus the set of simultaneous equations to be solved takes the form
_ 7 -
1-
m
CSihri(hrlpI+ h r g 2 + ,..., + h,,,pn - H,) = 0
r=l
where the sensitivity S is the height of the largest peak in the mass spectrum of a given compound for unit quantity of sample admitted into the mass spectrometer. Details of the inversion program (Layokun, 1975) are to be published.
Results A. Computer Prediction and Kinetics. The obtained composition--time profiles of the pyrolysis products are compared with computer predictions. Product distributions a t 700 and 750 “C are shown in Figures 2 and 3, respectively. The integration program (Kershenbaum, 1972) is based on the semiimplicit trapezoidal rule coupled with a variable step size selection. “Stiffness” is taken care of such that systems typified by autocatalytic and branched free-radical systems are easily solved. Like the Gear (1971) method, it is also without the limitations of the steady-state approximation. For a tubular plug flow reactor at constant total pressure in which essentially two moles of products are formed for each mole reacted, differential equations of the type employed by Herriot et al. (1972) describe the system
--------c i p a c o
’ no,
set,
Figure 2. Product distribution from propane pyrolysis at 700 “C. I
1
1
Figure 3. Product distribution from propane pyrolysis at 750 “C.
where i = 1, 2 , 3, ..., n, and where
The consumption of materials for the formation of free radicals is assumed to be negligible compared with that for the formation of molecular products. Thus dCi/dt are essentially equal to zero in equations for which Ci is the radical concentration. These equations then become implicit algebraic equations which are solved for the free-radical concentrations at each integration step. The
differential equations are numerically integrated. Small step size is used initially to simulate the components accurately after which it can be increased progressively according to the rate of change of concentrations of components in the system. Several values of the Arrhenius parameters-the frequency factors ( A ) and the activation energies (E)-from the literature were tried, but those which were found to best describe the system over the range of temperatures considered are given in Table I. This kinetic model was integrated using a CDC 6400 digital computer. Some detailed comparisons are shown in Tables I1 and 111.
Ind. Eng. Chem. Process Des. Dev., Vol. 18, No. 2, 1979
r-
Table 11. Comparison of Experimental Data with Computer Predictions. Composition Expressed as Mol 46 exptl comp exptl comp exptl comp 873 K t = 0.8 s t= 1.5s C3H, 97.90 98.10 95.00 94.60 1.31 1.26 0.42 C,H, 0.41 0.022 0.025 0.036 0.040 C,H, 1.48 0.576 1.47 CH, 0.47 1.25 H, 0.416 0.420 1.30 973 K t = 0.4 s t = 1.2 s C,H, 46.00 45.40 23.32 22.35 C3H, 11.50 1 1 . 5 2 16.78 16.72 1.77 1.94 0.77 C2H6 0.75 C,H4 13.50 13.75 20.24 20.77 CH, 15.00 14.50 23.32 23.10 11.00 10.60 14.34 14.74 H,
235
+l
i t
t = 2.2s 90.32 89.20 2.32 2.36 0.041 0.05 2.70 2.67 2.31 2.36
lot i -l
t = 2.0 s 11.19 18.66 2.80 23.97 27.98 15.50
10.07 18.66 2.80 24.25 27.52 15.67
comp
exptl
comp
1023 K t = 0.80 s 2.70 2.50 20.40 21.50 4.20 4.25 26.70 27.00 31.20 31.00 17.10 16.00
0.62 19.50 5.00 27.50 32.50 14.30
Figure
5.
I2
11
d ,T Determination of the overall energy of activation, E. I-
x
Table I11 exptl
C,H, C,H6 C2H, C2H, CH, H,
comp
exptl
t = 0.40 s 7.80 8.50 19.00 19.80 2.50 2.50 24.70 25.00 27.20 27.80 17.40 16.50
t = 1.8s 0.40 20.50 5.00 27.40 32.50 15.00
-*
,
900
000
700
r, j 600 "c,
r temp, "C Laidler et al. (1962) De Boodt (1962) Kershenbaum (1967) this work (1975)
Figure 4. Determination of the rate constant, K , at 750 "C.
Kinetic Model. Other reaction steps which are undoubtedly occurring, though not significantly, include reactions such as C3H6 + CH3 C4H9 C3H7
+
+ C4H8 C4H7 CzH4
C4H9
C4H8 + H
+
-
-+
-
C4H7 + C3Hs
C4H6 + H 2C
+ 2Hz
-
The soot deposited on the tips of the sampling probe is most probably obtained through the reaction CzH4 2C 2Hz and may involve acetylene as an intermediate. The Overall Order and Activation Energy of Reaction. The system fits only a first-order kinetics. Graphs of In [ao/(ao- x ) ] vs. t for different temperatures are drawn; that a t 750 "C is shown in Figure 4. The slope gives the value of K , the reaction rate constant. At 600 "C, K = 0.0384; a t 700 "C, K = 1.276; at 750 "C, K = 2.00 and log K = log A - E/2.3RT. A plot of log k vs. 1/T gives a straight line of slope -E/2.3R. This is shown in Figure 5. The slope = 1.30 X lo4; E = 2.3R X 1.30 X lo4 cal/mol = 2.3 X 1.96 X 1.30 X lo4 cal/mol = 58.6 kcal/mol = 245
+
E, kJ/g-mol 227.8 221.8 218.0 245.00
kJ/mol. This gives a value of A = 2.63 X lo4 s-l. Thus, kinetics of propane pyrolysis between 600 and 800 "C can be described by the rate expression
I suc
-Time
530-670 700-750 800-1000 600-800
rp = 2.63
X
1014 exp(245/RT
X
[C3H8])
Comparison with earlier work is shown in Table IV and the comparative value of K is illustrated in Figure 6. The temperature range 600-800 "C is the range usually encountered in commercial pyrolysis. A mechanistic study as herein described backed by adequate numerical technique would eliminate the proliferation of misleading rate constants in the literature.
Nomenclature A = frequency factor in the Arrhenius expression, s-l or cm3/ (g-mol-s) a,, = stoichiometric coefficient of the ith component in the jth reaction; a,, < 0 is a reactant; a,, > 0 is a product C = concentration, g-mol/cm3 C, = total reactant gas concentration excluding diluent, gmol/cm3 K = forward rate constant, or cm3/(g-mol.s) K' = reverse rate constant, or cm3/(g-mol.s) r = rate of jth reactions, g-mol/(cm3.s) k = fractional conversion of propane t = space time, s i = refers to the component, molecular of free radical j = refers to the reaction m = total number of reactions n = total number of components 0 = refers to reactor inlet conditions (including diluent)
238
Ind. Eng. Chem. Process Des. Dev., Vol. 18, No. 2, 1979
Literature Cited Benson, S. W., O'Neal, H. E., "Kinek Data on Gas phase Unimolecuhr Reactbns", p 385, 1970. Bradley, J. N., Proc. R . SOC. London, Ser. A , 337, 199 (1974). Crynes, B. L., Albright, L. F., Ind. Eng. Chem. Process Des. Dev., 8, 25 (1969). De Broodt, H., Ingeniersthesis, Rijks Universtsit, Gent, 1962. Frey, F. E., Hepp, H. J., Ind. Eng. Chem., 25 (4), 441 (1933). Gear, C. W., Commun. ACM, 14, 176 (1971). Herriot, G. E., Eckert, R. E., Albright, L. F., AIChE J., 18, 94 (1972). Hinshelwood, C. N., Proc. R . SOC.London, Ser. A 234, 301 (1956). Kerr, J. A., Trotrnan-Dickenson, A. F., J. Chem. SOC.,P1611 (1970). Kershenbaurn, L. S., Sena, M. P., "Kinetic Modelling of Gas Phase Reactions Involving Free Radicals", presented at 65th Annual Meeting of the AIChE, 1972.
Kershenbaurn, L. S., Martin, J. J., AIChE J., 13, 148 (1967). Kondratiev, V. N., Voevodsky, V. V., Prog. React. Kinet., 1, 41 (1961). Laidler, K. J., Proc. R . . SOC.London, Ser. A , 270, 246 (1962). Layokun, S. K., Ph.D Thesis, University of London, 1975. Leafhard, D. A., Purnell, J. H., Roc. R. SOC. London, Ser. A , 305, 517 (1968). Lin, M. C., Laidier, K. J., Can. J. Chem., 44, 2927 (1966). Pease, R. M., J. Am. Chem. SOC.,5 0 , 1779 (1928). Pratt, G. L., "Gas Kinetics", p 159, Wiley, New York, N.Y., 1969. Purnell, J. H., Quinn, C. P., R o c . R . SOC.London, Ser. A , 270, 246 (1962). Steacie, E. W. R., Puddington. I.E., Can. J. Res., Sect. E, 16, 176 (1938).
Received for review August 16, 1977 Accepted September 5 , 1978
Perturbation Pyrolysis Kinetics of Propane Stephen K. Layokun* and David H. Slater Department of Chemical Engineering and Chemical Technology, Imperial College, London, S. W.7., England
Neglect of the effect of trace impurities in pyrolytic systems must have contributed to the inability of earlier researchers to rationalize the complete product distribution especially at high conversion. The effects of traces of acetone and acetaldehyde on propane were separately measured between 600 and 750 OC in a tubular reactor system. Experimental data were compared with computer predictions. The presence of the trace impurities does not alter the reaction order of unity with respect to propane concentration, but the reaction mechanism is altered. A 21-reaction step model describes the observed kinetics in the presence of acetone, while a 24-reaction step model describes the kinetics in the presence of acetaldehyde.
Introduction The free-radical chain reaction mechanism for the thermal decomposition of a paraffin requires first a scission of a C-C bond of the paraffin to generate free radicals which are necessary for initiation. Consequently, if it were possible to introduce some radicals from another source into the reactant so that the chain mechanism could be initiated or propagated by another reaction, then the decomposition rate as well as the products spectrum should be altered. Thus any chemical impurity capable of dissociating into free radicals can be considered as undergoing initiation, propagation, and termination reactions. Both acetone and acetaldehyde have the methyl group in their respective structures and this will be available for reaction after scission. Mechanistic models for the thermal decomposition of acetaldehyde (Liu and Laidler, 1967) and of acetone (Hinshelwood et al., 1935) have been proposed. These appear to be general models which, however, can be modified to interpret low or high temperature data. Hence in building a model that describes the kinetics of propane pyrolysis in the presence of these additives, the importance of some of the elementary steps in the proposed models for acetaldehyde and acetone will be discussed. Experimental conditions will be 1 atm total pressure and temperatures between 600 and 750 "C. Data acquired will be compared with computer predictions. The Reaction System The flow system which incorporates a plug flow reactor and a mass spectrometer (Model MSIO-C2) has already been described in an earlier work (Layokun and Slater, *Direct all correspondence to this author at the Department of Chemical Engineering, University of Ife, Ile-Ife, Nigeria. '
0019-7882/79/1118-0236$01.00/0
companion article in this issue). In addition an Abingdon vaporizer was employed to furnish known volumes of additives into the propane stream. The vaporizer has a graduated knob on the front face. When this knob is in the zero position, an internal mechanism closes the vapor chamber completely but permits free flow of propane or any other feed from the inlet to the outlet. The inlet and outlet tubes were fitted with Teflon, which is resistant to acetone. Rotation of the control knob produces a gradually increasing output in accordance with the calibration marked on the dial. The calibration was carried out by opening the knob to a certain position while propane of known flow rate was passing, and analyzing the output mixture of propane and additive with the mass spectrometer. The stainless steel wicks in the Abingdon vaporizer (Longworth Scientific Instruments Co. Ltd., Abingdon, Berkshire) retained very little liquid and it was possible to recover all of the additive. I. Propane-Acetone System The amount of acetone in the propane stream varied from 0.8 to 4% a t 600-700 "C but to a maximum of 3% at 750 "C as the decay of propane was expected to be much faster a t this temperature. In every case there was an acceleration of propane decomposition, and the major products in order of magnitude are CH4,C2H4,C3H6,Hz, C2H6,and C4 fraction. Methane and ethane increased in comparison to their values in the absence of acetone. Propylene and hydrogen were slightly reduced while ethylene remained virtually constant. Some of the concentrations profiles are shown in Figures 1 and 2. Mechanism and Model. Hinshelwood and Winkler (1935) concluded that the thermal decomposition of acetone is almost completely homogeneous. The work of Allen (1936) supported this. According to these authors (Hinshelwood and Winkler, 1935; Allen, 1936),the thermal 0 1979
American Chemical Society
