Role of Surface Area in Dehydrocvclization Catalysis J
J
J
ALLEN S. RUSSELL AND JOHU J. STOKES, JR. Aluminum Research Laboratories, Xew Kensington, Pa.
T
H E deposition of active -4ctivated aluminas were impregnated with varying (2) which supplies feed from another tube, D, a t high rates. quantities of molybdena and their catalytic activities agents onto adsorbents has been employed widely t o determined for dehydrocyclization of n-heptane to toluene Pyrex reaction bulbs E, of 100-ml. catalyst chamber volprovide catalysts of increased at atmospheric pressure. Activity increased nearly linesctivity, improved stability, ume, fit by standard taper arly w-ith added molj-bdena until a concentration degreater selectivity, more conpendent on the alumina area was attained. Excess molybjoints t o the feed supply and renient physical form, and dena beyond this value did not further increase activity. condenser. Catalyst is preioR-er cost (11). The developActivated aluminas of varying area were impregnated with vented from falling out of ment of a simple method for the bulb by a plug of Pyrex constant amounts of molybdena and tested. -4ctivity increased linearly with surface area to a value dependent on measuring surface areas of wool. The feed is preheated porous materials ( 1 ) provided by passing through the furthe molybdena concentration. Further increase of area a basis for correlating infornace in a small tube before produced m u c h less rapid increase of activity. Comparientering the catalyst chammation on the action of adson of the calculated molybdena area w-ith the experimental value for maximum activity shows that the alumber. The sealing lubricant sorbents in this application. This work was undertaken t o ina surface is covered with a monolayer of molybdena at for joints and stopcocks is maximum activity. Activity was relatively insensitive to examine the relation of ada mannitol-dextrin-glycerol change in particle size from 2 to 20 mesh. mixture insoluble in hydrosorbent area t o catalyst activity for molybdena impregcarbons (IS). Heat is supnated onto activated alumina plied t o the reaction bulb by as a catalyst for the dehydrocyclization of n-heptane t,o toluene. a well insulated furnace F containing three separately controlled This reaction was reported with a variety of catalysts prepared coils. Temperatures are measured by chromel-alumel couples in diverse manners on a number of supports (4, 5, 7 , 9, 14, 15). held by a four-hole porcelain tube in a n axial well of the reaction Investigation of the reaction mechanism indicates t h a t a n initial bulb. One couple is set at the geometric renter of each of the dehydrogenation is followed by cyclization and further dehydroheating coils. The product of the reaction is cooled by water genation (8, 17, 18). Carbon skeleton rearrangement can take condenser G and ice condenser J. The liquid is collected in d a c e on a chromia-alumina catalyst (10;. calibrated tube H . Gas volume is measured by a Sargent wettest meter, ;If, and molecular weight is determined in an Edwards MATERIALS, APPARATUS, AND ASALYSIS gas density balance, K and L. The liquid product from n-heptane dehydrocyclization was The activated aluminas were products of the Aluminum Ore assumed to be toluene, unreacted heptane, and small quantities Company. The activated alumina F series analyzed: 2y0 loss of heptenes (8, 14). Toluene was determined by density or rem ignition, 0.1% iYazO, 0.1% Sios, 0.05% FenOs, and 0.77, C1. fractive index on the assumption t h a t the heptenes present have The activated alumina H series analyzed: 4% loss on ignition, the density or index of n-heptane. Liquid olefinic hydrocarbons,' 0.1% hTa20,7% SiOz, 0.27, FenOs,0.3% CaO, 0.2'3 MgO, and calculated as heptenes, were determined by titrating a 2-ml. @.3%SOo. .Ictivated aluminas of the H series have one and a sample in 10 ml. of 10% sulfuric acid a t 18-20" C. with 0.5 N half t o three times the surface area and pore volume, and more potassium bromide-potassium bromate, adding an 0.4ml. increuniform pore widths as determined by sorption, than do those ment which was not decolorized, shaking 2 minutes, and backof the F aeries. Nolybdena was obtained as Baker's Analyzed titrating with 0.2 A- sodium thiosulfate. These analyses were smmonium molybdate: 83y0 MOOS, O . O l ~ OSO,, all other reconsistent with toluene plus olefinic hydrocarbons; the latter ported impurities less than 0.01%. ;immonia n-as Baker's were determined by the volume change in a 10-ml, sample shaken .Inalyzed (meeting A.C.S. standards). The n-heptane, supplied with 30 ml. of a solution made by dissolving 100 grams of phosby Keatvaco Chlorine Products Company, had boiling point phorous pentoxide in 300 ml. of 95% sulfuric acid. 98.42' C., freezing point -90.65" C., diO 0.68371 gram per The amount of coke deposited on the catalyst was determined ml., and n2: 1.38766. Tank hydrogen passed over hot copper as the sum of carbon plus hydrogen measured by combustion of snd dried with activated alumina was used for catalyst reduction. a representative sample of the catalyst after the heptane run. Catalysts were prepared as follon-s: Ammonium paramolybAnalyses showed the coke t o have the approximate composition date was dissolved in distilled m-ater (0.5 nil. per gram of alumina CH. This value of coke deposition was found t o agree well with in the completed catalyst), and the quantity of concentrated the weight increase of the catalyst during the run corrected for ammonia necessary t o convert the paraniolybdate t o t h e normal the weight loss of molybdena on reduction. For'the standard molybdak was added. The solution was boiled and the alumina catalyst this reduction correction was found t o be 1.0 gram. added iuimediately with vigorous stirring. The excess water was removed by drying a t 70-90' C. under overhead heat, with EXPERIMENTAL PROCEDURE occasional stirring. When the alumina appeared dry, in 4-6 hours, the sample was calcined 16 hours a t 500" C. Unless a statement is made to the contrary, all the catalysts Figure 1 shon-s the catalytic apparatus. Constant-speed were run as follows: Catalyst sample, screened t o 4-8 mesh, was geared motor A loxers plunger B into a closely fitted glass tube, heated 16 to 20 hours at 475-500" C. in a furwce with air circulaC, containing feed for low feed rates, or actuates a bellows pump tion, transferred quickly t o a reaction bulb snd weighed. The
1071
INDUSTRIAL AND ENGINEERING CHEMISTRY
1072 Figure 1.
Catalytic Apparatus
L-
JUM
Val. 38, No. 10
catalyst when each catalyst produces the same TO, which in this paper stands for the integral percentage of toluene plus olefinic hydrocarbons in the liquid reaction product. The catalyst selectd as stnndard is an activated alumina F of area 0.38 m.b./g. impregnated to a n atoniic ratio of InOlgbdeIIum to aluminum of 0.05. To determine activity, it is convenient first to measure and graph TO produccd by standard catalyst as a function of feed rate. TO from the test catalyst at a givcn feed rate is measured, and the feed ratc at which standard catalyst produces this TO is determined from the graph. Activity is computed as test catalyst few1 rate divided by the standard catalyst Iced rate. This concept must br conditionccl t o tlw fact that TO changes rapidly as t l i e run progresses (16). I n establisliing the graph of 2'0 against feed rate, tlic s(aiiclnrd catalyst was run a t each ferd rate for such a duration that its coke deposition was 4.0 grams per 100 ml. of catalyst. The test catalyst \vas run at a convenient feed rate and temperRtiirc for the duration that would givt. 4.0 grams of coke per 100 ml. of standard catalyst. Usually 100-mi. catalyst volume, 2.3-nil. liquid feed per hour, 497" C., and 1.5-hour duration were employed for the test, catalyst. The yields from a run are defined as 1007, times the milliliters of heptane reacted t o form the given product, divided by the total milliliters of heptanc reacted. RESULTS WITH STANDARD CATILYST
bulb was connected t o the catalytic system, heated in 0.75 hour to 500" C., and held at 500" for 1.5 hours while hydrogen was *passed through tlie catalyst at 1.5 to 2 cubic fcet per hour. At the end of the reduction period the hydrogen \\-as turned off and the heptane flow st,arted. The furnace temperature was controlled either manually or by instrument during the run PO t h a t each of the thermocouples was as close as possible to the desired temperature. Temperature readings a t the start of the run sometimes varied 15' C. from the setting but usually averaged within 1' of the desired temperature. The molecular m i g h t of the gas stream was determined during the last 0.25 hour of the run. The system was flushed with hydrogen after the run, and air was excluded from the bulb while i t was weighed. Surface areas were measured by the method of Brunauer, Emmett, and Teller (I), employing the sorption of n-butane a t 0' C. Results are expressed as the millimoles of n-butane which just cover the area of 1 gram of alumina in the sample (m.b./g.). Values of area in square meters per gram are obtained by multiplying the foregoing values by J93 if the area of the n-butane molecule is taken as 32 square A. ( 3 ) . In the range of molybdena concentrations and temperatures employed, the prcscnce of reduced molybdena on the alumina does not change the area of the latter appreciably. CALCULATIONS
The activity of a test catalyst is defined as the ratio of n-heptane feed rate for the test catalyst to the feed rate for a standard
Although the ability of the c:Ltalyst, to cause reaction diminishes as the reaction proceeds, the catalj-st can be restored to its original activity by buriiing off the coke. Used standard catalyst heated 16 hours in air at, 360-400" C. was gray in color and shonetl activity, after standard reduction. ouly 80% t h a t of the origirial samplethat is, activity 0.8. Heating 16 hours in air at 500" C. produced a light ycllow catalyst of activity 1.0 after the same reduction. Prolonged heating in air or oxygcn a t 500" C. did not further increase activity. If the standard catalyst is properly burned but is not reduced in hydrogen before the run, its activity is about 0.7. Activity increases with increased reduction temperature to 500 C. (hydrogen flow rate 2 cubic feet per hour, duration 1.5 hours), with hydrogen flow rate t o 1 cubic foot per hour (temperature 500" C., duration 1.5 hours), and with duration of reduction to I hour (temperature 500" C., flow rate 2 cubic feet per hour). At standard reduction of 2 cubic feet per hour of hydrogen for 1.5 h o u k at 500" C. the activity is 1.0. More vigorous reduction produces relatively smaller changes in activity. Thcse findings agree with the statement t h a t actiT'ity is greater, the more complete the reduction ( 1 2 ) . Itesults of five tests x i t h standard catalyst show rcproducibility within 37, TO and 2% yield. These data are collected in Table I. Runs with nonstandard catalysts are usually reproducible to 0.1 activity unit and 270 yield. The dependence of TO on feed rate is shown in Figure 2. The reaction has a high temperature coefficient, TO increasing roughly 0.870 per, ' C. at a flow rate of 25 ml. per hour. The sum of toluene plus olefinic hydrocarbon yields is O
October, 1946
,
INDUSTRIAL AND ENGINEERING CHEMISTRY
1073
Pa 4
ii
80
n U=
zb >24 0
> L
d
> c
t
w
+
" 6
20
0' 0
10
20
30
40
50
60
70
80
90
100
110
fEED RATE, ML./HR
Figure 2.
Effect of Feed Rate and Temperature Liquid Composition
011
Standard catalyst
MO/AI
Figure 3.
nearly independent of feed rate, although a t high feed rates olcfinic hydrocarbon yield increases a t the expenx of toluene yicld
Effect of SIolybdena Concentration on Activity
Activated alumina F, surface area 0.47 millimole n-butane per gram
RELATION O F AREA TO h l 0 L Y B D E S . i COSCESTR 4 T I C S
-
I00
The results of Figure 3 show t h a t actiyity of impregnated activated alumina F, area 0.47 m.b /g., increases linearly to 1.18 with increasing molybdena from atomic ratio N o !AI 0.000 to 0.052 and, a t a higher ratio, activity is constant. Thus, to attain maximum activity, an alumina area of 0.090 m.b./g. is required for each increment of 0.01 in the atomic ratio on these samples. Yields change u i t h molybdenum to aluminum ratio up to 0.06 and tliercafter remain constant (Figure 41, in approximate agrecmcnt with the activity cliange,.. A series of activated aluminas of type F having areas up to 1.03 m.h./g. was iniprrgnated t o No/.il 0.05 and tested. Figure 5 s h o w that activity increases linearly to 1.15 as area increases to 0.43 m.h. ' g , and thereafter increases much less rapidly to activity 1.31 as area increases to 1.03 m.b. 'E. - UP . to the uoint where activity ceascs to increase directly x i t h increase in area, each lfo/Al0.01 requires a n area of 0.086 ni.b./g., in good agreement ,r.itll tile value of o,090 m,b./g, from Figure 3. Tields to o.4-o.5 m,b,/p. and thereafter remain change ,Titll area constant (Figure 6 ) . Activity increases with molybdena concentration on activated alumina II a t nearly tile Same rate as for the F samples, This is siyllificant in ~ e L of v the marked differences betLTeen these al,lniinas. But sample the activity does not cease a t lIo/Akl o.052 but continues at least to lfo/Adlo,15. Since the area of this alumina is 1.4 m.b./g., an area no greater than 0.09 m.b./g. is required to accommodate each Xlo/Al 0.01 on activated aliiniina H . DISCUSSION
I
A
80 o -1 Y
% CASEWS
-t
~ E o k% COKE
60
OLEFINIC
'
8 40
-
HYDRO CARBONS % TOLUENE
20
0 0
Figure 4.
.04
.I2
.OB Mo/AI
.I6
Effect of 3Iolybdena Concentration on Yields
Activated alumina F, 491' C.. 23 ml. per hour feed rate
other gases. Furthermore, the a w m t ) t i o n that molyMens spreads over the surface as lIo03 is o w n to question. Thus, the agreement may be fortuitous betwxli the cnlculated arra, 0.094 m.b./g., required to accommodate JIo/.\l 0.01 in a I I I J ~ I O law, and the ~ 1 u i . 0.086 s and 0.090 m.h./g. found experimmtally t o be required for maximum cfficicncy with this amount of molybdena. Nonetheless. it appears probablts that niaximum efficjency is achieved at approximately a m o n o l w r 01' molybdena On
If maximum activity is achieved a t a monolayer of molybdena, it is to be expected that activity increases lineady witli impregnated area. "impregnated area" means the s n i d e r of the two factors, alumina area (m.b./g.) or the amount. of area which can be covered by the available molybdena-namely, 9 hlojA1 (m.b./g.). Accordingly, activated aluminas of differing surface areas and amounts of molybdena were tested and their activities plotted against impregnated area (Figure 7 ) . The linear rela-
T o explain the surface area-molybdena concentration relations, the molybdena required to cover an alumina surface was calculated by the procedureoof Emmett and Brunauer ( 3 ) . The area of RIo03 is 154square A. by the formula, Area = l.53(rn/d)2!3 where m = molecular weight of 144 TABLE I. REPRODUCIBILITY O F RUNS WITH STANDARI) CATALYST d = density of 4.5 grams per ml. (Temperature, 497' C.; run duration,,l.50 hr.; feed rate 22.6 ml,/hr.: catalyst. vol. The Emmett and Bruniuer value for n-butane is 32 square A. One gram of alumina impregnated t o Mo/Al 0.01 contains 0.196 millimole of Moo3 which would just cover an area of 0.09! m.b./g. There is some doubt that 32 square A. is the correct area for the n-butane molecule. Harkins and Jura (6) found that values from 37-52 square 1. give best agreement with results for
ml., oxidized at. 95 g r a m )
Gas
a
Mol. 'Cu. Toluene, Olefins, Prodat. ft.O Vol. % Voi. % uct, M I . 6.9 0.693 64.3 2.9 17.6 7.5 0.736 69.3 2.9 17.2 0.754 70.5 2.6 17.0 6.8 7.0 0.727 69.6 3.3 17.0 7.0 0.746 68.6 3.0 17.0 250 c., 72 preesure.
Coke, G. 3.9 4.2 4.0 4.1 4.1
Toluene 55 56 57 55 54
Yield, % Olefins Coke 2 24 2 25 2 23 2 . 25 2 24
100
Gas 20 23 20 20 21
i ~
io74
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 38, No. 10
100
I
CASEOUS
80
HYDROCARBONS COKE
60
OLEFINIC HYDROCARBONS
_I 0
w 40 x
TOLUENE
20
n
0
.3
.6
.9
1.2
SURFACE AREA (UILLIHOLES N - B U T A N E PER GRAM)
Figure 6.
n
activated alumina F, 497’ C.. 13 m l . per hour feed rate
0 .3 .6 .3 ‘.Z SURFACE AREA [MILLIMOiES N - B U T A N E PER GRAM)
Figure 5 ,
Effect of Surface Area an Activitj-
Activated alumina F. Mo/Al
=
have been ubserved with increase of area. The linear increase in activity with impregnated area implies that diffusion is of minor importance in establishing the dehydrocyclization rat,e for the samples tested. Further evidence for this is the relatively small influence of particlc size on activity (Table 11). In the extreme case of rcaction limited t o the geometric surface, or for high values of Thiele’s modulus (19),the reaction rate might increase fivefold on decreasing part’icle size from 2-4 t o 14-28 mesh. The observed change in activity v a s relat,ively small (Table 11) and there is evidence that it may not be due to change in diffusion path, since a low area (\Tide pore) alumina, in which diffusion rates should be of lesser influence, shows as great change in activity with particle size as does the narrow pore alumina.
0.05
f--r 24
Effect of Surface Area on lieldq
I
V ”
I
1
ACKNOWLEDGMEST
The authors are indebted t o V. 11. Stowe for helpful suggestions, especially the method of calculating activity, and for the preparation of many of the catalysts. LITERATURE CITED
0
.3
.6
.9
IMPREGNATtD AREA (MILLIMOLES
Figure 7 .
I. 2 1.5 N-BUTANE PER CRAM)
Effect of Impregnated Area on Activity
rion indicates impregnated area as tile most important factor changing the activit,y of these samples. The samples employed in this graph had a Tide range of chemical impurities and physical structures. Hovever, i t is not to be expected that this same relation rvill hold for all possible modifications of this catalyst. The act,irated aluminas of t,he F series here employed are characterized by constant pore volume so that, as their surface arem increase, the pore widths diminish by the same ratio. If diffusion of reactants and products controlled the dehydrocyclizaiion rate, a decrease rather than an increase in activity might
OF CATALYST PARTICLE SIZEON ACTIVITY TABLE11. EFFECT
Activated Alumina
R.I.B./G.
Srea,
Particle Size, Mesh
F
0.47
2-4 4-8 14-28
F H
.
0.27
2-4
1.41
4-8 4-8 8-14
Activity 0.70 1.18 1.19 0.43 0.82
Brunauer, S., Ltmnett, P. H., and Teller, E., J . Am. Chem. SOC., 60, 309-19 (1938). Corson, €3. B., and Cerveny, W. J., IND.ENQ. CHEM.,. ~ N A L . ED., 14,899 (1942). Emmett, P. H., and Brunauer, S., J . Am. Chem. SOC.,59, 156364 (1937). Green, S. J., J . Inst. Petroleum, 28, 179-208 (1942). Grosse, A. V., AIorrell, J. C., and Mattox, W. J., IND. ENQ. CHEU.,32, 528-31 (1940). Harkins, 7V. D., and Jura, G., J . Am. Chem. Soc., 66, 1366-73 (1944). Heard, L., G. S.Patent Reissue 22,196 (Oct. 6, 1942). Hoog, H., S’erheus, J., and Zuiderweg, F. J., Trans. Faraday SOC.,35, 993-1006 (1939). Kasanskii, B. A , Sergienko, S.R., and Zelinskii, S . D., Compt. rend. uced. sei. (U.R.S.S.), 27, 664-9 (1940). Koinarewksy, 1‘. I., and Shand, W. C., J . A m . Chena. Soc., 66, 1118-20 (1944). Krczil, F., “Technische Adsorptionsstoffe in der Kontaktkatalyse”, Leipzig, 1938; Ann Arbor, Edwards Brothers, Inc., 1943. blaslyanskii, G. N., and Shenderovich, F. S.,J . Phys. Chem. (U.S.S.R.), 14, 1301-7 (1940). hieloche, C . C., and Fredrick, W. G., J . Am. Chem. SOC.,54, 3264-66 (1932). Xloldavskii, B. L., and Kamusher, H., Compt. rend. acad. sci. (U.R.S.S.), [W.S.] 1, 355-9 (1936). Plare, A. F., Uspekhi Khim., 9, 1301-32 (1940). Smith, D. J., and Moore, L. W., Chem. & M e t . Enu., 48, No. 4 , 77-9 (1941). Steiner, H., J . Am. Chem. Soc., 67, 2052-4 (1945). Taylor, H. S., and Turkevich, J., Trans. Faradup Soc., 35, 92134 (1939). Thiele, E. W., IND.GNQ.CHEM.,31, 916-20 (1939).
1.12
1.35
PRESESTED before the Division of Phyaical and Inorganic Chemistry at t h e 109th Neering of the AMERICAN CEEXICALSOCIETY, dtlantic City, N. J.
