THERMAL DECOMPOSITION RATES OF SATURATED CYCLIC HYDROCARBONS B E L A
M. FABUSS, RALPH
K A F E S J I A N ,
AND
J OH N 0 . S M I T H
.llonsarito Research Carp., Eit?ett, .tfass.
C HA R L ES N
.
SA TT E R F I E LD
, 2tfassachu.relts Institute of
I'eLhnoloqy, Cambrzdge, .\iu~s.
A general correlation between decomposition rate and molecular structure has been developed from measurements at elevated pressures of thermal decomposition rates of 28 compounds, including cyclohexane, decahydronaphthalene, hydrindan, and their derivatives. Present and previous rate data on cyclohexane and decahydronaphthalene are discussed in detail. Some activation energies found are cyclohexane and derivatives, 61 to 63 kcal. per gram-mole; decahydronophthalene, 64 kcal. per gram-mole; and methylhydrindan, 60 kcal. per gram-mole.
A
of technical applications requires the use of fluids thermally stable a t high temperatures. In man)' of these applications, saturated cyclic hydrocarbons lvould be particularly appropriate. T h e objective of this study \vas to develop a general correlation between the rate of decomposition a t elevated pressures and molecular structure from measurements of decomposition rates of various pure cyclic saturated hydrocarbons chosen to represent the major structural groups, ioith varying degrees of substitution. VARIETY
Experimental
"Pure" hydrocarbons, as received, were passed through fresh silica gel andl'or activated coconut charcoal and filtered
Table 1.
through 0.45-micron hlillipore filters. T h e purity of the hydrocarbons was judged primarily from vapor phase chrornatograms. I n several cases \vhen impurities were not renioved by the adsorption treatment, the material was fractionated. Final samples were stored in glass containers near 0 ' F. Table I lists the compounds studied and the estimated minimum purity. In some cases, several ihorners Lvere present, and the minimurn purity \vas taken to be equivalent to the sum of the areas of the several neighboring peaks, each presumably representing one isomer. Table I lists the number of coniponents present, as indicated by the number of peaks in the chroniatogram, a n d also the area of the main peak as a percentage of the total. T h e critical temperature and critical pressure for
Boiling Points, Purity, and Critical Data of Selected Hydrocarbons Vapor Phuse Chromatogram, E ~ t d .Wzn. A i l a m Peak 'YO. of Cr1t. B.P., O F . . Purzty, 0 ' Aiea, c; Components Temp., ' F . ~
Compound
Cyclohexane
Cyclohexane, methylCyclohexane, ethylCyclohexane, n-propylCyclohexane. n-butylCyclohexane, isopropylCyclohexane. iert-butyllohexane, 1,4-ethylmethylclotiexane, diethylclohrxane, 1,3,5-trinicthylclohrxane, 1.2,4,5-tetrarnethylI~ecahyd~onaphthalerie diinethylI~ccatiydrona~hthalene, Dc.cal!ydlonaphthaleiie, l-ethylr)ecaliydronaphthalene, isopropylDpcaliydronaphthaleiie, tert-butyltIydrindan
Hydriiidan, mrthylHyiilindan. ethylHydrindan, isopropylCyrloprntane, n-propylCyclopentane, n-butylBizyclopcntyl Bicyclohexyl Biryclohrxyl, isopropyl-
hlcthaiir, dicyclohexyl1,3-dicyclohexyl-
Hutant.,
I'c.r~iy[ilo-')-ethylanthracene n
248
Or
177.3 213.6 267 310 353 306 337 298 346 281 331 386 435 438 466 484 332 358 396 428 266 310 372 542 486 574 573
99.9-t 99.9+ 99.9+
99.5+ 99.9 98.4+ 99.9
99.8 99.8 99.8 99.8 94.9 99.99t 98.7 99.9-b 99.4 99.8 95.9 99.4 96
+
99 99.99+
98.9 99.2 99.6 99.98 95 8 99
inr sf boiling range. I&EC
PROCESS D E S I G N A N D DEVELOPMENT
99,9+
100 100 99.5
2 1 1
44 ..
3
614 656 766
6 2 3 3
918 872 933 942
19.9 23 1 17.3 19.1
3 6
63 51 81 58.8 64.5 52.2 78.8 87 45 51 56 99
100 98.9 99 45 99,9
95.8 99
Atm.
38.16 34 1 30 2 28 0 25.4 28.4 25.5 27.8 25.0 27.6 24.9 28.7 23.4 24.9 22.3 20.3 30.7 27.6 25.1 23.1 29.8 26.9 28,7
99.9
98.4 100 77.6
1 1
537 570 625 664 703 668 706 647 693 624 674 784 809 813 836 827
Crtt. Press.,
. .
rach compound were estimated by the method of Lyderson: described by Sher\i.ood and Reid ( 2 0 ) . I n most cases. the rcaction conditions i n this study are in the supercritical region.
;IS
Procedure. 'Thermal decomposition rates were measured primarily iii a static test apparatus. From 1.26 to 1.5 ml. of thc h ~ - d r o c a r h o n\cas placed in a borosilicate glass tube, the q \ i , i i i t i t y of ti>-drocarbonbeing adjusted to constitute in each ~ ' a s c19.Sc4 of the tube volume a t room temperature. All samples \verp degassed by alternate freezing and thawing under V ~ ~ ~ L I I I I T and II. the tubes were then sealed under vacuum. 'rlicy i \ c w prcheatcd I O .500' I?. in a n oven and then immediJtcIy t r m f r r r e d to a large aluminum block furnacr and held ,at the reaction temperature (7000.750'. 800'. or 850' F.) Toi. 1)i:riods from 1 to 24 hours. Complete temperature equilibrium \cas reached in less than 15 minutes. I n severd ins1ancc-s. reaction t i m e up to IO0 hours \cere used. ;Iflei- rapid cooling. each tube was opened, gas formed by rcaction \\as allo\ccd to escape. and the volume of liquid remaining mas m e a s u r ~ d . T h e residual concentration of unreacted material in the liquid was determined by vapor phase rlirc~matograph>-. and the fraction of starting material remaining \\.as ca!culated from the ratio of final to initial volumes of liciuid. Slight chaiigos in density associated \\ith reaction products rrmaining i n the liquid phase lvere presumed to be iicgligible. A n F Rr M Model 124 high-temperature vapor phase chromatograph with a conventional thrrmal conductivity detcctor unit waFt used with a 12-foot column packed with hromosorh ,\, :\pie;ron grease being the immobile phase. he column \cas maintained a t 200' to 600' F. depending upon the hydrocarbon used. Injection sample volume was 5 11. 'l'he helium pressure was 1.5 p.s.i.g. to give a flo~crate of 75 cc. per minute. A siibstantial number of runs was made a t 800' F'. \+.ith each h y d u x a r b o n to cover a range of conversions. Some hydrocarbons \cere studied a t other temperatures to dcterminc the effrctive activation energy. Some studies u e r e made in a stainless steel (T)-pe 304) highpressure' isoteniscope [ 9) to determine the effect of pressure on decomposition. 1-urthcr studies \yere also made a t 500 p.s.i.g. in a microflow reactor (6)that consisted of a hypodermic tube 45 inches long a n d 0.02 inch in inside diametrr.
Table ll.
T h e tube was coilrd. heatrd electricall!-. and held inside B Deicar flask for heat insulation. A11 riins wrre made a t one fixed inlet liquid flow r a t r . Results and Discussion
Hydrocarbon cracking reactions, particdarly a t elrvated precsures. are highly complrx kinetically. 'Thr drcompo4tion process usually followr a first-order relationship to a first approximation. but "self-inhibition" (decrease in calculated first-order rate constant Lcith incrrasing coilversion, is frequently observed. I n a batch reactor, on thc other hand. i n Fchich the total pressure increasrs with increasing conversion, t h e calculated first-order rate constant sometimes increases ivith incrrased conversion. T h e first-ordrr rat? constant was calculated for each of our runs from the expression : k
=
~~
1 In
1
(1)
I--x
t
\vhere t is the reaction time, and x i s the fraction of original material reacted. ,\s thus defined. the valiic of i- varied sonit'what with conversion for a subsrantial iiiimbcr of the compounds studied. In order to have a simplc method of ir:prcsenting the data, on? which permits the reactivities of different compounds to be compared easily. the first-ordcr kinetic expression was retained. ' I h e n a statistical approach was used to determine for each compound studird ichethrr k was brttcr represented as a true constant or instead by a liricar function of conversion. Table TI summarizes the experimental rr the iiryt-order initial rate constant doubled for a pressure increase of about 340 p.s.i. A batch reactor such as a sealed tube is useful and reliable in measuring reaction rates Lvhen the rate of reaction is sufficiently slow that temperature and other transients a t the beginning and end of the run are relativrly unimportant. In contraSt to the situation in flow reactors. all the material present is kno\iri to be subjected to the reaction tc-mperature for the same time. and gmrrally isothermal conditions can br. achieved.
1000
900
800
-
Id
LX
Id
10.2
055
060
0.65
0.70
075
RECIPROCAL TEMPERATURE
+x
080
I~~(*R:~I
Figure 5. Arrhenius plot for decahydronaphthalene Conversion, Press., Atm. Wt.
%
0 Tilicheev (23)
A A
SundgrCn ( 2 2 ) Bachman et a/.( I ) 0 Present flow experiments 3 Present static test d a t a (overage of 15 experiments) 3 Malinovskii and Stoyanovskoya ( 1 5 )
8-96 18-99
ca. A0
10-32
33-A7
15-30
35
24-62 12
A0 1
1
$-k :.
z- 2 0.3
Q-
02
3%
sg
$Y O2
0.1 007
005
0
200
400
600
800
1000
1200
1400
PRESSURE, RS.1 A
Figure 4. Effect of pressure on rate of decomposition of tert-butylcyclohexane at 800" F. Horizontal arrows indicate initial and final pressure in isoteniscope experiments
252
l&EC PROCESS DESIGN A N D DEVELOPMENT
IO 12 14 16 18 2 0 2 2 24 26 28 30 32 34 36 38 CHARACTERIZATION NUMBER, n
Figure 6. Rate constants for decomposition of naphthenic hydrocarbons a t 800" F.
b i t h a n extremely small diameter (0.02-inch inner diameter). T h e second uncertainty concerns the flow ar,d mixing pattern in the reactor. In a tubular reactor, plug flow is generally assumed although. in fact, major departures from this assumption frequently occur. T h e effect was analyzed for our reactor using the recently published correlations of Levenspiel ( / . I ) , C n d e r typical reaction conditions, the Reynolds n u m ber here is about 100. a n d the Schmidt number about 2. T h e intensity of dispersion ((reciprocal of the axial Peclet n u m b e r ) , D ' u d . is about 1 (where D is the diffusivity, u the linear velocity, a n d d the tube diameter), a n d the reactor dispersion number, (Dl'ud) (d,'L),, where L is reactor length, is about 0.00044. From the correlations of Levenspiel, such a reactor would be expected to behave in identical fashion to a plug flow reactor. I n qualitative terms, the plug flow behavior of our microreactor is due to the low Reynolds number a n d the extremely high length-to-diameter ratio. T h e third possible uncertainty occurs if there is a change in n u m b e r of moles on reaction. To determine true residence time, it is then necessary to have a complete analysis of reaction products as a function of degree of converaion. Decahydronaphthaleiie was studied in the microflow reactor at a series of temperatures between 1000" a n d l l O O o F. a n d a t a pressure of 500 p.s,i.g. to obtain kinetic data on a hydrocarbon over a wider temperature range. T h e results are shown in the form of an ,4rrhenius plot in Figure 5, which also includes d a t a from the static tests a n d other sources. Tilicheev (1'3)measured the rate of decomposition in the 425' to 500' C. (797" to 932' F.) temperature range; one experimental point is given by Malinovskii and Stoyanovskaya (75'1, b u t their residence time is not clearly defined. A number of rate constants can be calculated from Sundgrtn's work (22). b u t he was not primarily interested in obtaining kinetic data, and there is some uncertainty in OUI: interpretation of his work. Recently reported d a t a on cracking of a mixture of 56.97, cis- and 41.27, trans-decahydronaphthalene by Bachman a n d coworkers ( 7) are also included. T h e rate constant of decomposition c a n be expressed as : log k = 18.649 - 25,298 (1,'T)
(7)
where 7 is expressed in R . a n d the activation energy is 64,300 cal. per gram-mole. As with cyclohexane, the first-order rate constant appears to increase with pressure although the effect cannot be evaluated quantitatively from the available d a t a . T h e agreement between the two different methods of measuring kinetic constants !gives confidence in the validity of both method and d a t a . T h e constant pressure of 500 p.s.i.g. used in the flow reactor is close to the average occurring in a typical batch study. T h e studies with decahydronaphthalene correspond to conversions u p to 3OyG, so no correction was m a d e in analyzing these data for the change in number of mole: on reaction. T h e decahydronaphthalene. like its 1-ethyl derivative discussed above, consists of a mixture of the cis- and transisomers. Thus, the cdculated first-order rate constant decreases with increased conversion because of the accompanying isomerization reaction. Estimation of Rate Constant for Decomposition
l'able I1 yhows that the rates of decomposition vary widely. T h e ino.>t stable compounds are the nonsubstituted rings ; within each set of derivatives the rate4 increase with number and vile of substituent groups. A simple empirical group contribution method was developed to correlate the decomposition rate d a t a and permit estimation of rates for unknown cyclic saturated hydrocarbons. T h e basic assumption is that the
rate is determined by the kind of bonds present a n d that it is proportional to the number of bonds of a given type, Each molecular structure is represented by a characterization n u m ber. n; this consists of the characterization number of the basic ring compound minus one for each substituent that replaces a C-H bond in the unsubstituted ring, plus the sum of the characterization numbers for a n y side chains present. From the above sums, a slight correction is subtracted for bicyclics a n d tricyclics. T h e characterization number of the basic ring compounds is taken equal to the number of C-H bonds, except for decahydronaphthalene, for which the n value was empirically adjusted to 14. These values are listed below: n 10
Cyclopentane Cyclohexane H ydrindan Decahydronaphthalene Perhydroanthracene
12 16 14 24
T h e characterization numbers for side chains a n d corrections for separated polycyclics are as follows : n
CHB CH2 CH C C (in ring) Bicyclics (separated) Tricyclics (separated)
$2 $4
+6 +4 +4 -1 -2
Examples of the calculation of the characterization number are given below: n
n-Propylcyclohexane Characterization no. for substituted ring = 12 - 1 2 C H ? groups = 2 X 4 C H I group
11 +8
+2 n =
21
n =
13 +6 +4 23
Isopropyldecahydronaphthalene Characterization number for the substituted ring = 14 - 1
C H group 2 CH3 groups
=
2 X 2
Dicyclohexylmethane n for two rings = 2 X 1 2 CH, group Correction
-
2
22 4
=
-1 n =
25
A plot of the average rate constant from Table I1 as a function of the characterization number is shown in Figure 6. Only very few compounds deviate from the line, the equation of which is : k
=
0.044 - 0.114n
+ 0.0008n2
(8)
Since the characterization number increases by four units for the next member in a series of homologous compounds, the precision of this graph is always within this range. This means that the error of a n estimated rate constant is smaller than the difference between the rate constants for two consecutive members of the given series. Figure 6 should be useful for estimating rate data for a variety of naphthenes and other saturated cyclic hydrocarbons. I t does not apply to paraffins a n d probably becomes less reliable for alkyl side chains exceeding C, in length. literature Cited (1) Bachman. K. C.. Matthews, E. K., Zudkevitch, D.. Directorate of Materials and Processes. Arronautical Systems Division. Air Force Systems Command, IVright-Patterson Air Force Rase, Ohio, T e c h . Doc. Kept. ASD-TDR-62-254 (1962). VOL. 3
NO. 3
JULY
1964
253
(2) Dintses. A. I., Compt. Rend. Acad. Sci. C'.S.S.R. 2, 153 (1933). (3) Dintses, .A. I., C'spPkhi Khim. 7 , 404 (1939). (4) Dintses, A. I.. Frost, A . V., J . Gen. Chem. V.S.S.R. 3, 747 (1933). (5) Dixon, \V. J., Massey, F. J., Jr., "Introduction to Statistical Analysis," McGraw-Hill. New York, 1957. (6) Fabuss. B. M., Smith, J. O., Lait, R . I., Borsanyi, A. S., Satterfield, C. N., IND.ENG.CHEM.PROCESS DESIGN DEVELOP. 1, 293 (1962). (7) Frey, F. E., Ind. Eng. Chem. 26, 198 (1929). (8) Frost, A. V., J . Phyr. Chem. U.S.S.R. 8, 290 (1936). 191 Johns. I. B.. McElhill. E. X.. Smith. J. 0.. IND.ENG.CHEM. PROD.RES.DEVELOP. 1,'2 (1962). (10) .Jost, LV., Mufflng, L., Z . Elektrochem. 47, 766 (1941). (11) Kasansky, B., Plate, A. F., Ber. 6 7 , 1023 (1934). (12) Konovalov, D. S., Migotina, E. N., Zh. Prikl. Khim. 26, 332 (1953). (13) Kuchler. L.. Trans. Faraday Soc. 35, 874 (1939). (141 Levenspiel. 0.. "Chemical Reaction Engineering," LYiley, h e w York. 1962. (15) MalinoLskii, M . S.. Stoyanovskaya, Ya. I., J . Appl. Chem. l'.S.S'.R. 29, 1369 (1956). \
I
(16) Morris, H . E., Fabuss. B. M.: Smith. J. O., Satterfield, C. N., in "hdvances in Petroleum Chemistry and Refining," J. J. McKetta, ed., Vol. IX, \Vile!.. N c w York. in press. (17) Pease, R . N., Morton, J. M.. J . d m . Chrm. Soc. 5 5 , 3190 (1934). (18) Rosen, R., Oil Gas J . 39, Fcb. 13. p. 49; Feb. 20, p. 45 (1941). (19) Schultze, G. R., LVassermann, G.. Z. Ekektrochem. 47, 774 (1941). (20) Sherwood, T. K . , Reid, R . C.. '.The Propertirs of Gases and Liquids," McGraw-Hill, New York; 19%. (21) Steacie, E. I V . R., "Atomic and Free Radical Reactions," p. 122, Reinhold, New York, 1954. (22) SundgrCn..'i.,.lnn. Ofice .Vat/. Combustibles L q u i d e s 5,44 (1930). (23) Tilicheev, M . D., Zh. Przk/. K h m . 12, 735 (1939). RECEIVED for review July 25, 1963 ACCEPTED November 4. 1963 \Vork performed under the direction of Materials Central, Directorate of Advanced Svstems Technolow. LVricht .\ir Development Dibision, \Vright-Pa;terson Air For& Base," Ohlo, on Contract AF 33(657)-8193.
THERMAL DEMETHYLATION OF TOLUENE C
.
C
.
ZIM M ERM AN
A ND R0 BE RT Y 0 R K
, Cornr// Iniae?sztj. Ithaca, S.Y .
The thermal demethylation of toluene at 700" to 950" C. and atmospheric pressure has been studied in a continuous flow reactor. The reaction was found to be first-order with respect to toluene and half-order with respect to hydrogen. The rate constants for the decomposition of toluene and formation of benzene are identical within experimental error and have values in units of (moles/liter)-"2 (sec.)-l of k , = 3.5 X 10'" exp (-50,90O/RT) and kh = 3.0 1 O'O exp (-50,90O/RT). The yield of by-products follows the
x
approximate empirical relation: W = 1 .O X 1 O'O O/b exp (-51,00O/RT), where W = the yield of b y products, weight per cent of feed, b = hydrogen-toluene rotio, and 0 = actual gas residence time, seconds. The reaction of toluene to form benzene i s increased b y the presence of large amounts of n-heptane, but i s unaffected b y thiophene, pyridine, or carbon monoxide. A possible reaction mechanism is discussed.
I
past few )-ears there has been much interest in utilizing the large supply of petrochemical toluene as a raw material for chemical synthesis rather than as a fuel. At elevated temperatures, toluene will react with hydrogen to produce benzene and methane as the main products. N T m
CHa
Processes for both the thermal and catalytic demethylation of toluene have appeared in the recent literature (3:4 ) . Silsby and Sawyer (7) have proposed a reaction mechanism and presented some kinetic d a t a o n the thermal demethylation reaction. Several other articles on the thermal demethylation reaction ( 7 . 5. 8) have also appeared. However, few pertinent data of use to the chemical engineer have been published. 'This study was undertaken to obtain kinetic a n d engineering d a t a on the thermal demethylation reaction. Of particular interest \vas the determination of the yield of benzene as a function of temperature, residence time, and the hydrogentbluene ratio of the feed. Some of the problems of commercial application were investigated by determining the effects of various impurities in the toluene and hydrogen feed streams. 1 Present address. Denver Research Center. Marathon Oil Co., Littleton. Colo.
254
I&EC PROCESS DESIGN A N D DEVELOPMENT
Experimental
Reactor System. T h e flow reactor is shown schematically in Figure 1. T h e main part of the reaction system was a vaporizer, a preheater, and a reactor. T h e vaporizer consisted of 10 feet of '/,-inch stainless steel tubing immersed in a molten salt bath. T h e vaporized toluene, after mixing \vith hydrogen, passed into the preheater tube made of 11 feet of Vycor tubing formed into a coil. T h e preheater section was designed for rapid heat transfer and a short residence time. T h e reactor, or soaker, section was also constructed of Vycor glass tubing and provided the required residence time for conversion to take place. T h e reactor section was 24 inches long and had a volume of 2.25 liters. Temperature was measured with five Chromel-.4lumel thermocouples located a t intervals inside the reaction chamber. Residence time in the reactor was varied by changing the reactant feed rates. T h e reaction product was quenched in a short. double-pipe heat exchanger which was cooled with circulating cold water. T h e reaction product passed through the annular space between the thermocouple lead protection tube and the inner wall of the condenser. T h e condenser effluent, in the form of a dense aromatic fog, was then passed through a glass x.001 filter to separate the gas and liquid. T h e gas was then passed through a cold trap (acetone and dry ice) where additional liquid product condensed. Toluene was fed into the reaction system by a metering pump, and the hydrogen flow was regulated by means of a calibrated rotameter. T h e liquid product from the reactor was determined by weighing the filter collection flask and cold trap aftrr each experimental r u n . Gaseous products were metered by a wet-test meter prior to being vented to a flare.
