A Kinetic Study of the Decomposition of Hydrocarbons by Silica

By J. L. Franklin and D. E. Nicholson. Refining Technical and Research Divisions, Humble Oil and Refining Company, Baytown, Texas. Received June 9, 19...
0 downloads 0 Views 484KB Size
,Jan., 19,56

DECOMPOSITION OF HYDROCARBONS BY SILICA-ALUMINA CATALYSTS

50

A KINETIC STUDY OF THE DECOMPOSITION OF HYDROCARBONS BY

SILICA-ALUMINA CATALYSTS BY J. L. FRANKLIN AND D. E. NICHOLSON Refining Technical and Research Divisions, Humble Oil and Refining Company, B a y t m , T e r n Received June 9, 1966

The kinetics of the catalytic decomposition of four n-paraffins, three isoparaffins and one cycloparaffin over silica-alumina catalysts have been studied in a static system with product compositions being determined from mass spectrogra h analyses. Pressure dependence measurements indicated fractional orders for the rate of disappearance of n-butane a n a n-pentane. Rates were also followed as a function of time with constant initial pressures. Activation energies have been estimated for the eight hydrocarbons investigated and the results discussed in terms of ionic reactions on solid surfaces. The rapid decrease in activation energy with increase in molecular weight of the n-paraffins parallels the changes in ionization potentials for these compounds; however, 2,2-dimethylpropane having a lower ionization potential than either 2-methylpropane or 2-methylbutaneJ is very resistant to catalytic decomposhion. Some evidence is presented which suggests that the cracking reaction may have an induction period.

determination of rate data for catalytic cracking of Introduction From product distributions obtained in catalytic propane, n-butane, n-pentane, n-hexane, 2-methylcracking of pure hydrocarbons, it has often been in- propane, 2-methylbutane1 2,2-dimethylpropane and ferred that reaction proceeds through a series of cyclohexane to allow estimation of activation encarbonium ion intermediates. There is also the ergies. possibility that hydrocarbon molecules may be adExperimental sorbed on silica-alumina surfaces and cracking iniHydrocarbons.-Research Grade hydrocarbons obtained tiated by formation of molecule-ions by transfer of from the Phillips Petroleum Company were used for most electrons more or less completely to the solid. A kinetic measurements and had the following analyses: 100 mole %; n-CdH10,99.85mole %; n-CsHlz, 99.84 have pointed out that a CsHs, number of mole %; n-CsH14, 99.85 mole %; i-C4H10, 99.90 mole %; relationship should exist between reaction rates i-C6H12,99.80 mole %; cyclo-CeHI2, 99.95 mole %. 2,2and the ease of electron transfer from adsorbed Dimethylpropane, from the Matheson Company, was found molecules to solid catalysts if the rate-controlling to have approximately 98 mole $7, urity by mass spectrometric analysis. All samples storecfin lecture bottles under step involves partial or complete ionization of the pressure were withdrawn from the liquid phase, by inversion substrate. In catalytic cracking, it can be argued of the cylinders, and distilled at least twice in the vacuum that a correlation of activation energy with ioniza- line prior t o final va orisation into hydrocarbon storage tion potential of a series of compounds should exist, bulbs. n-Pentane, n-iexane and cyclohexane were freed of air by repeated freezing, evacuation and melting of the at constant catalyst activity, provided ionization solid. of the hydrocarbon is actually rate determining. Apparatus.-Kinetic measurements were performed in a The problem of predicting product composition static system. The Pyrex reactor of approximately 280 from both thermal and catalytic decompositions of ml. was connected to manifold tubing of 2-mm. capillar thus maintaining the dead space as a small fraction of hydrocarbons is intriguing from theoretical and ap- total volume. Gaseous hydrocarbon was admitted plied standpoints. A classical approach to inter- to thereactor reactor at a known pressure and cracking carried out pretation of product distribution from thermal under conditions of constant volume, with the pressure processes has been given by Rice.6 Modification change being followed during the course of reaction by a mermanometer. After a selected time interval, such as of the method of Rice by Greensfelder' led to a cury 800 sec., the reaction mixture could be rapidly expanded similar treatment employing relative rates of hy- into an evacuated receiver bulb, previously cooled by liquid dride ion removal, rather than relative rates of nitrogen. Reaction products were warmed to room temremoval of primary, secondary and tertiary hydro- perature and transferred to mass spectrometer sample bulbs means of a Topler pump. gen atoms. I n both methods, it was postulated byCatalysts and Pretreatment.-The silica-alumina catalyst that the hydrocarbon free radicals or carbonium chosen for this investigation was the commercially available ions produced by abstraction of hydrogen atoms or 3A material, having an initial surface area of approximately hydride ions, respectively, underwent a p-scission, 600 m.*/f Chemical analysis gave the following corn osiof t I s catalyst: SiOz, 85.6%; AlzOs, 12.9%; 820, successively yielding lower molecular weight sub- tion 1.5%. Two large samples of deactivated catalyst, having stances. It is conceivable that molecule-ions and surface areas of 81 and 303 m.Z/g., were prepared by subcarbonium ions here postulated to be active inter- jectin the original sample to steam and heat sintering. mediates in catalytic cracking undergo decomposi- Cylin8rical pellets (l/g" X I/*") were prepared using a rotary tablet machine. A standardized tapping technique served tion by mechanisms resembling those by which the to the same weight of catalyst into the reactor. same ions split under electron impact in the mass In introduce general, four to ten kinetic experiments were made with spectrometer. one loading of catalyst in the reactor. An extensive series The present investigation had as its object the of preliminary experiments indicated that assage of oxygen,

tl;

(1) G.-M. Schwab, Trans. Faraday Soc., 42, 689 (1946). (2) A. Couper and D. D. Eley, Nature, 164, 578 (1949). (3) D. A. Dowden, J . Chem. Soc., 242 (1950). (4) P. W. Reynolds, ibid., 265 (1950). (5) M . Boudart, J . Am. Chrm. Soc., 12, 1040 (1950); 74, 1531 (1952). (6) F. 0. and K. K. Rice, "The Aliphatic Free Radicals," Johns Hopkins Press, Baltimore, Md., 1935. (7) B. 8. Greensfelder, H. H. Voge and G . M. Good, Ind. Eng. Chem.. 41, 2573 (1949).

saturated with water vapor at 25', over t1e catalyst in situ at a flow rate of 500-1000 cc./min. for a period of two hours yielded reproducible catalyst activity. Regeneration was accomplished by combustion of coke deposits in a stream of oxygen and water vapor for 30 min. at temperatures of 450500 Reproducibility in catalyst activity was checked by repetition of individual rate measurements with regeneration of the catalyst between these experiments. In general, a typical run could be repeated precisely enough to yield concentrations of uncracked starting material in the product

.

,r. L. FRANKLIN AND n. E. NICHOLSON

GO

gas within fly0 of an average value. This agreement is essentially equal to the accuracy of the mass spectrometric analyses. A further check waa made to estabbh the effect of coke deposition on catalyst activity. It was found that repetitive rate measurements without regeneration of the catalyst between experiments gave identical conversions of starting materials within f1%: thus, catalyst deactivation could not be detected as resulting from coke deposition. All of the measurements here Wem made with 180 g. of catalyst in the 280-ml. P p e x reactor. In some preliminary studies, smaller quantities of catalyst were employed, and it was found that conversion of starting material d e c w ~ e din approximately a linear manner as the weight of catalyst in the reactor waa decreased. I n some related studies in which surface area of the catalyst was varied over a wide range, L e . , from about 6o-600 m.*/g., the Utivity of the catalyst, as measured by conversion of isobutsne, ww found to vary linearly with surface area. Hence, a t least to a first approximation the active centers on the two silicaalumina catalysts chosen for the present investigation are similar in nature. Tern erature Regulation.-Temperature control was etfected %y a Brown EledroniK recorder-controller in conjunction with 811 bn*onStantan thermocouPle. A Ho+ kins m d e furnace equip ed with a stainless steel block contained the %ex a n 8 Vycor reactors. All thermcouples were calibrated frequently against a 25-ohm plati-

SUMMARY OF RATEDATAFOR OC.

num resistance thermometer so that absolute temperatures are known to f0.5O a t a fixed point in the reactor. Temperature Werences of approximately 2O were observed between the ends of the reactors. Care was taken to maintain the relative geometry of the thermostat and appurtenances fixed.

Results

A pronounced characteristic of the reaction prod-

from the catalytic decomposition studies in the static system was that only small amounts of olefins were present; material balance calculations for the confirmed that compoun~sof the type CnH!n Were probably polymerizing on the catalyst and hence not escaping from the reactor. From product distribution data on cracking of cyclohexane, material balances indicated the composition of the coke deposits to be C&&,,. Interference in catalytic cracking from competitive thermal decompositions was measured for seven of the eight hydrocarbons studied. Rates of thermal cracking we= essenti&' nil compared to rates of catalytic cracking for all compounds except proUctS

TABLE I CATALYTICDECOMPOSITION OF EIGHT HYDROCARBONS

Surfaoe area of oatslyst, aq. m./g.

Temp., Compound

TEE

Vol. 60

Range of reaotion times. sec.

Initial hydrocarbon pressure. mm.

x

k

104,

aec. -1

Group I Propane

303

525.0 f 0.5 550.0 450.0 500.0 400.0 450.0 476.0 399.0 454.4 400.0 450.0

2-Methylpropane 2-Methylbutane 2,2-Dimethylpropane Cyclohexane

633 635 625 625 365 367 370 66, 325 54,182,364,634 62.5 62 .O

80 80 303

80

(11.7 f 0.4) (28.0 f 1.0) (1.25 f 0.19) (66.3 f 4.9) (3.67 f 0.08) (9.7 f 0.7) (16.7 f 0.6) (0.637 f 0.005) (6.3 f 0.7) (3.9 f 0.3) (10.0 f 0.7)

300-600(3)" 200-400 (3) 600-1200(3) 300-1200(3) 900-1600(2) 300-1200(4) 3oO-600 (3) 1200 (2) 300 (4) 300-1600 (4) 300-1200(4)

Group I1 k

n-Butane

450.0 500.0 n-Pentane 425.0 475.0 n-Hexane 321.7 346.1 378.9 0 Numerals in parentheses show the number of

x

600-1200(3) 620 624 150-400 (3) 303 293 300-900 (3) 175-4OO (4) 290 400-1200 (3) 303 99 300 (2) 95 200-600 (3) 82 experiments upon which rate constants were baaed. 303

106, rnm.-l/: geo.-l

(6.75 (2.20 (12.0 (3.25 (8.53 (17.7 (36.3

f 0.58)

f 0.14) f 0.8) f 0.34) f 1.54) 0.2) f 3.0)

*

TABLE I1 KINETICDATA-PF~ESSUBE DEPJDNDENCE STUDIES Compound Temp., OC. Resotion time, aeo. Pi. nim. SpeciEo reaction rate constant k X 101, sea.-' (1.0 order) k X 10:. mrn.-'/' (1.5 order) Compound Temp., ' C . Reaotion time, aec. P I , mm. Specific reaotion rate constant k X lO:,aea.-: (1.0order) k X 10:. rnm.-'/: (1.5 order)

74.1

-

--c4HI-

+-a.&---

526-. * - - -3 151.2

-51.1 465.2

652.1

I

~

639.0

43.9

- 2 121.3 241.8

315.7

0.614

0.698

0.814

0.800

1.326

2.41

2.67

0.0803

0.0839

0.1422

0.1279

0.1817

0.1784

0.1153

0.0955

0.0492 0.0714 0.0806

92.6

-12624

91.5

139.9

197.3

323.0

0.695

0.591

0.706

0.620

2.08

2.10

2.26

0.0206

0.0671 0.0351

0.0786 2.18

1.86

1.59

-'-CiHi-500.-

500-.

H

495.7

0.1150 0.384

-60626.0

&

361.3

0,1081 0.1060 0.0543

C

173.2

0.923

-C4Hl-

P

5 -0-.

Jan., 1956

DECOMPOSITION OF HYDROCARBONS BY SILICA-ALUMINA CATALYSTS

pane at 550'; blank runs showed that about 20% of the total conversion of propane at 550" was due t o thermal cracking. V h e rate equations for homogeneous gas reactions apply in heterogeneous catalysis under conditions where the surface is sparsely covered, as is reasonable to expect under the conditions of this investigation. These equations have been used in the present study, and their use is further justified by the constancy of the specific rates. Tables I and I1 contain summary kinetic data for rates of disappearance via catalytic cracking of the eight compounds included in this investigation. The pressure dependence studies were performed some months after completion of the other rate measurements: it will be noted that catalyst activity is slightly different from that in those rate measurements on which activation energies have been based. Within experimental accuracy, the decomposition of propane fits a first-order law: n-butane and n-pentane follow a 1.5-order law approximately. The low vapor pressure of n-hexane at laboratory temperatures prevented varying initial hydrocarbon pressure over a wide range as was done for other compounds. Reference to the data in the tables will show that 2-methylpropane, 2-methylbutane and 2,2dimethylpropane have rates of catalytic decomposition which are represented by a simple first-order law to a good approximation, i.e., with specific rates which are "constant" within about i10%. The average deviation of all specific rates determined in this study was found to be *7%. Straight lines are obtained on plotting log k vs. 1/T for nhexane and 2-methylbutaneYindicating conformity to the Arrhenius equation.

of selected olefins and aromatic compounds is shown in Table IV. Total initial hydrocarbon pressures (625 mm.) and reactor temperatures (450') were maintained constant in experiments 1-8: reaction time was 210 seconds in every case. The data in column 1 represents an average of four runs performed at intervals between the other experiments. The reproducibility of these runs showed that catalyst activjty remained essentially constant throughout the series of studies. Reference to Table IV shows that propene and butene-1 accelerate the rate of disappearance of n-butane t o an extent which is greater than the average reproducibility of the measurements. Furthermore, it will be noticed that these olefins cause substantially the same increase in rate when present at equal concentrations in the starting material. 2-Methylpropene was found t o have little or no influence on cracking of n-butane, while benzene and ethylbensene exhibited a slight inhibiting effect. TABLE I11 ACTIVATION ENEBQIES AND IONIZATION POTENTIU Compound

A E L kcal./mole

Propane

41.7

n-Butane

26.3

+Pentane

20.7

n-Hexane %Methylpropane ZMethylbutane 2,ZDimethylpropane Cyclohexane

18.4 37.1 20.5 40.0 18.5

4 = 15.4

9 = 5.6 4 = 2.3

* 1.O

0.4

2.7

2.3

2.3

22.3

27.7

28.0

17.9

The specific rate constant for 2-methylpropane a t 500" has an average value of 6.7 X set.-', when measured with initial hydrocarbon pressures of approximately 625 mm.: a sevenfold decrease in initial pressure of 2-methylpropane lowered the specific rate t o 5.9 X lo-' set.-', showing the rate of disappearance of starting material to be comparatively independent of pressure. This difference, amounting t o a 12% decrease in specific rate, is not greatly outside the mean reproducibility of the rate constants. Variation of initial pressures of 2-methylbutane from 140 to 323 mm. also revealed a slight increase in rate. Experimental evidence suggesting existence of an induction period in catalytic cracking was of two types: (1) kinetic measurements in decomposition of 2-methylpropane a t short reaction times (cu. 100 sec. at 500") yielded specific rates that were approximately lower than the steady-state rate of 6.3 X set.-'; (2) addition of trace quantities of olefins accelerated the rate of decomposition of n-butane at low conversions. The influence

I a , kcal./mole

258.5

4 = 9.4

249.1

TABLE IV INFLUENCE OF ADDED GASESON CATALYTIC %BUTANEDECOMPOSITION 1 2 3 4 5 6 ~. None l-C& l-C,H, GHI i-C4Hs CsHs

Expt. No. Added material Mole % of added material 0 Moles n-CcHlocracked/100 moles n-C4Hlocharged 19.4

61

2.8

A = 6.1 243.04 = 2.5 240.5 247.4 243.3 238.7 238.7

7 CsH&sHr 1.3

15.0

16.6

It may be noted that rates of catalytic cracking of the n-parafEns at 525O are Compound

Relative rate of deoomposition

GH8 *C&o n-CBIz n-CJL

1 7 13 20

Now, from Table I11 it can be seen that the difference in activation energies for cracking of propane and n-hexane is of the order of 24 kcal./mole, corresponding to a ratio of relative rates of decomposition at 525" of e21m c2 X Is*) = 3.3 X 10'. But the experimental ratio of rates is k(n-hexane)/k(propane) = 20/1. Obviously, the frequency factors in the Arrhenius equation differ widely for propane and n-hexane. Discussion The fractional orders observed in cracking nbutane and n-pentane, at constant reaction times but with varying initial hydrocarbon pressures, may

62

J. L. FRANKLIN AND D. E. NICHOLSON

be attributed (1) to the adsorption-desorption equilibrium between the starting materials and the catalyst, or (2) t o the nature of chain initiating and terminating steps in the reaction mechanisms. Molecular diffusion in the catalyst pores under condition such that the entire surface is not available for reaction could lead t o a 1.5-order of reaction for a process that would otherwise be second order; however, calculations based on diffusion theory, applied to constant volume systems, show that the true specific rate constant for cracking of nhexane should not be influenced by mms transfer under the experimental conditions prevailing in the present investigation.* Furthermore, the likelihood of diffusion being rate limiting is minimized by virtue of the comparatively high activation energies required for catalytic cracking, i.e., 1842 kcal./mole. On plotting ionization potential us. activation energy for the n-paraffin series a smooth curve obtains. I n addition, incremental differences in ionization potentials and activation energies are essentially equal. Since the beginning of the present investigation, evidence has appeared indicating that cracking catalysts function as Lewis acids. Haldemang concluded that the active centers on silica, alumina and silica-alumina were primarily Lewis acids at 500°, and about one-half of the active centers were Bronsted acids at 150'. Mapes and Eischens'o postulated that if the catalyst be a Bronsted acid, the chemisorbed ammonia will exhibit the characteristic infrared absorption peaks of ammonium ions. If the catalyst is a non-hydrated Lewis acid, only ammonia spectra should be seen. The spectrum of a catalyst on which ammoilia was chemisorbed showed a strong band at 3.0 p due t o ammonia: an ammonium band a t 6.9 I.C was weaker than the 3.0 p band. In view of a backgrouiid of supporting data, electron transfer would (8) 9 . Wlieeler. "Advances in Catalysis," Academic Press, Inc., Ncw York, N. Y . , 1951, pp. 300-301. (9) R . G . Haldeman, l i n i u . Pilfsbuv0h Bull., 49, No. 14, 81 (1953). (IO) J. E. Mapes and R. P. Eischens, "The Infrared Spectra of .411iinonia Chemisorbed on Cracking Catalysts," 124th Meeting of the Am. Chem. SOC..Chicago, 1953.

Vol. 60

seem possible as the ratedetermining step in cracking of the n-paraffins studied in this investigation. A hydrogen and/or charge transfer mechanism will aid jn explaining the gross trends in influence of added olefins upon rate of decomposition of n-butane, assuming the important rate step is formation of the molecule-ion C4Hlo+ through bimolecular reaction of n-butane with the molecule-ion formed from the additive. A few of the possibilities are illustrated.

+ CaHa+ +C*He+ + C3Ha + C4Hsf (from I-CdH8) CaHs + C4H9. A H R = -24 kcal./mole n-C4HlI) + C3H6' +C4H9+ + CJ37. A H R = -19 kcal./mole ~ C 4 H i o+ CeHs+ +CeHa + C&o+ n-CdHio

A H R = - 2 kcal./mole

?&4HO l

----f

+

~ H= R 79 kcal./mole

I n a preponderance of other processes which can be visualized to include ionization of the substance added in small quantities to n-butane and reaction of the ion thus produced with n-butane, the heats of reaction tend to be less endothermic for the changes involving propene and butene-1 than for those involving isobutylene, benzene and ethylbenzene. I n conclusion, the correlation of ionization potential with activation energy for catalytic decomposition of propane, n-butane, n-pentane and n-hexane is believed to be suggestive that silica-alumina catalysts may function as electron acceptors with respect t o adsorbed hydrocarbons. Formation of thealkyl carbonium ions (CnHln+ I) + is proposed to occur via molecule-ion intermediates. Moreover, there are indications that the catalytic cracking process has an induction period; this behavior appears t o be similar to that noted by Otvos, et ~ l . , " , ' ~ in the hydrogen exchange reactions of n-butane and 2-methylpropane in sulfuric acid where the induction period was shortened by addition of olefins. (11) 0. Beeck. J . W. Otvos, D. P. Stevenson and C. D. Wagner, J . Chern. P h y s . , 17, 418 (1949). (12) J. W. Otvos, 0. Beeck, D. P. Stevenson and C. D. Wagner, J , Am. Chsm. SOC.,78, 5741 (1951).