THE MECHANISM OF POTASSIUM PROMOTION OF CHROMIA

THE MECHANISM OF POTASSIUM PROMOTION OF CHROMIA-ALUMINA DEHYDROGENATION CATALYSTS. Joanne M. Bridges, G. T. Rymer, and D. S. ...
1 downloads 0 Views 1MB Size
n b y , 1962

POTASSIUM PRO3I/roTIOS

OF

CHROMIA-ALUMIXA DEHYDROGENATIOS CATALYSTS

87 1

THE MECHANISM OF POTASSIUl!t PROMOTIOX OF CHRO&lIA-AL’C’MIXA DEHYDROGENATION CATALYSTS BY JOANNE M. BRIDGES, G. T. RYMER,AND D. S. MACIVER Gulf Research & Development Go., Pittsburgh, Pa. Received October 36, 1861

A study has been made of the mechanism of the alkali promotion of chromia-alumina dehydrogenation catalysts by investigating the reaction of cyclohexane over a variety of promoted and unpromoted chromia catalysts. At the same time the magnetic and adsorptive properties of these materials have been determined and correlated with catalyst activity and selectivity. From the results obtained it was shown that there are, in general, two types of active sites on chromia-alumina, a dehydrogenation type associated with the chromia portion of the surface and an acid type associated with the alumina surface. The acid sites catalyze the formation of methylcyclopentane, which acts as a poison for the chromia dehydrogenation sites. The addition of potassium as a promoter ‘heutralizes” these acid sites, thereby preventing the formation of methylcyclopentane and hence increasing the dehydrogenation activity. Potassium in excess of that required for this “neutralization” is associated with the chromia portion of the surface in such a way that it blocks some of the dehydrogenation sites and thus lowers the dehydrogenation activity. The interaction of the potassium with the chromia apparently results in the formation of a Dotassium-chromium comalex with the chromium in a f 6 oxidation state, which under reaction conditions reduces to a +3 state.

Introduction Chromia-alumina has been widely used as a catalyst for a variety of reactions, notably the dehydrocyclization and dehydrogenation of alkanes and naphthenes,l and it has been well established1,2 that both catalyst activity and selectivit>yoften are improved by adding a salt of an alkali metal such as potassium nitrate to the catalyst. Several workers have studied the influence of potassium on the chemical and physical characteristics of chromia-alumina and have attempted to relate their findings to variations in catalytic properties. Voltz and Weller3 reported that potassium tended to partially stabilize the catalyst against reduction and to decrease the surface acidity; these latter results were later confirmed by Voltz, Hirschler, and Smith.4 It ,also was found3 that potassium decreased the activity for ethylene hydrogenation, cyclohexane dehydrogenation, double bond isomerization, and carbon monoxide oxidation, while it increased the activity for hydrogen peroxide decomposition. More recently, the structures of chromia-alumina-potassium oxide catalysts have been studied, and the results related to activity and selectivity in decomposing isopropyl alcohol. 5 To date, however, information is lacking on the precise mechanism by which alkali functions as a promoting agent, especially with respect to hydrocarbon reactions. For this reason, an investigation has been carried out in this Laboratory on the physical properties of alkali promoted chromiaalumina catalysts and the role of these properties in determining the manner in which various hydrocarbons react over the catalysts. The present paper reports the effect of potassium on the dehydrogenation of cyclohexane. Experimental Materials.-Hydrogen, helium, and prepurified nitrogen were obtained from the Air Reduction Company; oxygen (1) F. G. Ciapetta, R. M. Dobres, and R. W. Baker, “Catalysis,” Vol. VI, ed. b y P. H. Emmett, Reinhold Publ. Carp., New York, N. Y.,1958, Chapter 6. ( 2 ) R. C. Archihald and E. 8. Greensfelder, fnd. Eng. Chem., 37,356 (1945). (5) S. E. Voltz and S. W. Weller, J . Phys. Chem., 69, 569 (1955). (4) S. E. Voltz, A. E. IIirechler, and A. Smith, ibid., 64,1594 (1960). ( 5 ) A. H.Rubinstein, N. A. Pribytkova, V. A. Afanasyev, and A. A. Slinkin, “The Second Iiiternational Congresii on Catalysis,” Paris. France, July 4-9, 1960,Paper No. 100.

and argon from the Linde Company. The hydrogen used in the kinetic studies was purified by passage, first through a “Deoxo” unit, then through magnesium perchlorate, over platinized asbestos at 300’, through magnesium perchlorate again, and finally through a -78” trap. The argon was purified by passing i t over reduced copper at 400” and then through magnesium perchlorate. The treatment of the gases used in the magnetic and adsorption studies has been described elsewhere.6 Cyclohexane (Phillips Research Grade) and methylcyclopentane (Phillips Pure Grade) were used as received. Chromia-alumina catalysts were prepared by impregnating Davison ~-A1203with reagent grade chromic acid. Emission analysis of the original alumina indicated the presence of Si, Cu, and Ca to the extent of about 0.03% each, all other contaminants being present in amounts less than 0.01%. Ferromagnetic impurities were virtually absent as determined by magnetic susceptibility. The alumina, ground to 50-140 mesh, was contacted with chromic acid solutions of various concentrations for 24 hr. The liquid phase then was removed by filtration, and the solid was dried at 100’ for 24 hr. and calcined a t 500’ for another 24 hr. The calcined catalysts were subjected to three cyclic treatments a t 500°, first in hydrogen for 1 hr. and then in oxygen for 1 hr. A blank alumina sample was prepared in exactly the same manner except that distilled water was used in place of the chromic acid solution. Samples of chromia-alumina (10.5 wt. % Cr) obtained as indicated and a sample of the original alumina were promoted with potassium by impregnating with potassium nitrate solutions and treating in the same manner as described above for the chromia impregnation. Again, a blank chromia-alumina was made by substituting water for the potassium nitrate solution. Pure chromia was obtained by a method described earlier,6 and a portion was promoted with potassium in the same fashion as the chromia-alumina samples. Apparatus.-The kinetic measurements were made a t atmospheric pressure using an all-glass gas flow system. The flow rates of the carrier gases, argon and hydrogen, were regulated to within 1% in the fashion described by Bailey.’ These gases were saturated with cyclohexane a t 15” and passed through the catalyst reactor. Periodic mass spectrometric analysis indicated that the partial pressure of cyclohexane in the reactant mixture was 57 f. 1 mm. throughout the course of the work. For certain experiments it was necessary to add an arbitrary amount of methylcyclopentane to the reactant stream. This was accomplished by passing the stream through a U-tube containing a small amount of liquid methylcyclopentane at room temperature. The amount of methylcyclopentane so introduced was not sufficient t o appreciably change the partial pressure of cyclohexane in the reactant stream. When not in use the U-tube was isolated from the rest of the system. The reactor consisted of a 70 cm. long glass tube having a diameter of 12 mm. The catalyst bed, which was 10 cm. long, was located in the center of this tube, the remainder of the tube being packed with Pyrex wool. The entire reactor was enclosed in (6) D. S. Maclver and H. H. Tobin, J . Phys. Chern., 64,451 (1960). (7) S. W.Bailey, J . Sci. Instr., 31, 93 (1954).

872

J. M. BRIDGES, G. T. RYMER, A N D D. S. MACTVER

a tube furnace whose temperature was regulated by an automatic temperature controller. The temperature gradient across the bed was 0.5', and the maximum tem erature fluctuation during the course of a run was k0.5'. Sroduct analyses were made chromatographically. The product gases passed directly from the reactor to a manually operated Aerograph 6-way samplin valve which was used in conjunction piith a Fisher-Gulf Fartitioner chromatographic unit, The chromatographic column was 12 feet long and contained 10% 8-N-8 Flexol supported on an inert carrier. Magnetic susceptibility measurements were made over the temperature range from -78 to 200" using a Faraday type susceptibility balance previously described.8 The -78" temperature was attained by a Dry Ice-acetone bath. Temperatures above room temperature were maintained to within f O . l O , as measured with a calibrated copper-constantan thermocouple, by use of a platinum heating coil actuated by an automatic temperature control device. Adsorption studies were carried out gravimetrically using a quartz spiral balance described in an earlier publication.e Liquid nitrogen and Dry Ice-acetone were used to maintain temperatures of - 195 and -78", while elevated temperatures were obtained with a small, automatically controlled, ceramic furnace. Temperature measurements were made using nitrogen or carbon dioxide vapor pressure thermometers and with calibrated thermocouples. Procedure.-In making the kinetic measurements 4.6 g. of catalyst (except in the case of unsupported chromia where 1.3-g. samples diluted to a 5 cc. volume with either 50-140 mesh Pyrex chips or alumina were used) were reduced in situ with hydrogen a t 500" and a gaseous hourly space velocity of 5000 for 16 hr. The catalyst then was cooled to the initial reaction temperature of 416" in the hydrogen stream, and the reactant charge was passed over the catalyst. In the majority of the experiments this charge consisted of argon, hydrogen, and cyclohexane in a molar ratio of 6:6:1, although a few runs were made using an argon-cyclohexane charge in which the molar ratio of argon to cyclohexane was 12 : 1. In both cases the argon was used as a "tracer" to permit conversions to be calculated from the chromatograms according to the method described by Hall, MacIver, and Weber.'O Conversions were measured at 416' as a function of feed rate in the order 40, 30, 20, 15, 80, and 40 cc. (STP)/min., the identical rates at the beginning and end serving as a check of catalyst stability. The catalyst was allowed to equilibrate a t least 30 min. a t each feed rate before analyzing the product gas. At the end of this run the tem erature was lowered to 400°, and the catalyst was allowefto remain overnight at this temperature in the presence of the reactant stream. The following day, rate tsudies were made as a function of space velocity a t 400,380, and 355". The 416" run then was re eated in order to determine whether any catadyst aging h a i occurred. Susceptibility measurements were made on chromiaalumina samples in both the oxidized and reduced state. Ten-mg. samples, contained in a spherical quartz bucket, were treated, in situ, with a stream of oxygen or hydrogen a t 500' for 6 hr. and then evacuated for 16 hr. at this same temperature. During the course of the susceptibility measurements, the samples were maintained in a helium atmosphere. At each temperature susceptibilities were determined a t five field strengths ranging from 4200 to 7760 oersteds. The apparent paramagnetic susceptibilities at a particular temperature, after correction for the diamagnetic contributions of the quartz sample container and the alumina support, were plotted as a function of the reciprocal field. The hPld plots a t the various temperatures for any one sample were parallel with small positive slopes. True aramagnetic siisceptibilities were obtained from these plots %yextrapolation to infinite field. Surface areas of the catalysts were determined with nitrogen a t -195' by the BET method using a value of 16.2 A.a for the cross sectional area of the adsorbed nitrogen molecule. Oxygen chemisorption on chromia-alumina at - 195" was measured following the general procedure of Bridges, MacIver, and Tobin." ( 8 ) J. R. Tomlinson. R. 0. Keeling, Jr., G. T. Rymer, and Joanne M . Bridges, "The Second International Congress on Catalysis," Paris, France, July 4-9, 1960, Paper No. 90. (9) D. S. MacIver and H. H. Tobin, J . Phys. Chenz., 66, 1665 (1961). (10) W. Keith Hall, D. S. MacIver, and H. P. Weber, Ind. Enu. Chem., 62, 421 (1960).

1'01. 66

Results The surface areas of the catalysts used in this work are presented in Table I along with oxygen chemisorption values for several of the chromiaalumina samples. The oxygen adsorption at -195" is defined in the manner described by Bridges, MacIver, and Tobin,ll namely, as the amount of oxygen retained after the reduced catalyst has been exposed to oxygen at - 195" and then evacuated for 1 hr. at -78". In order to put these data on a comparable basis for the various catalysts the oxygen chemisorption values in column 6 of Table I have been divided by the BET nitrogen monolayer values; these ratios are given in column 7. TABLE I CATALYST SURFACE PROPERTIES'

141 32.2 .. .. 2.30 139 31.7 .. 0 2 0 3 ... 0 17 3.8 1.8 0.47 Cr203 ... 5.26 6.4 1 . 5 .. .. CrzO3/AI2O3 2.11 0 126 28.6 .. CrzO3/Al203 4.35 0 128 29.2 .. .. Cr203/Alz03 6.10 0 132 30 1 .. .. Crz03/A1203 10 60 0 125 28.5 2.7 0.10 Cr2O3/Alz03 11.05 0.20 123 28.1 2 . 8 .10 CrzO3/AlzO3 10.47 .36 122 27.8 2.5 .09 Cr203/A1203 10.57 .86 113 25.8 2.0 .08 Crz03/A1203 10.04 1.25 111 25.3 2 . 1 .08 CrZ03/A1,03 10.34 2.25 92 20.9 1 . 5 .07 a Ti,(Nn) is nitrogen monolayer value (BET) a t -195'. V(O2)is volume of oxygen chemisorbed at -195". A1203 A1203

. . I

...

0

..

..

The magnetic susceptibilities of the potassium promoted chromia-alumina samples were found to obey the Curie-Weiss law, and, hence, from plots of the susceptibility vs. reciprocal temperature one could obtain the effective magnetic moments, peff,and the Weiss constants, A. These quantities are presented in Table I1 for the various catalysts studied. The precision of measurement was rt0.02 Bohr magneton for the moments and *5"K. for the Weiss constants. Also included in Table I1 are the ratios ( p e g / p ~ r s a ) where 2 pCrCS is the theoretical spin-only moment of Cr+3 (ie., pcr+a = 3.87). This ratio also is equal to the fraction of chromium in the +3 state assuming that this is the TABLE I1 MAGNETIC PROPERTIES OF CHROMIA-ALUMINA CATALYSTS AS A FUNCTION OF PRETREATMENT Bohr-magnetons OxiRedised duced

--Lceff,

Wt. % Cr

W&%

10.60 0 11.05 0.20 .36 10.47 .86 10.57 10.04 1.25 10.34 2.25

3.23 3.23 3.04 2.98 2.92 2.69

3.41 3.39 3.52 3.48 3.39 3.43

--A8

OK.--

Oxidized

Reduced

220 220 160 160 180 170

200 210 210 210 220 240

-(

sff/p+cr3)+

bxi-

dized

Reduced

0.696 0.776 ,696 .767 .827 .617 .593 .808 ,569 .767 .483 .785

(11) Joanne M. Bridges, D. 9. MacIver, and H. H. Tobin, "The Second International Congress on Catalysis," Paris, France, July 4-9, 1960, Paper No. 110.

May, 1862

POTASSIUM PROMOTIOX OF

CHROMIA-ALUMISA DEHYDROGENATION CATALYSTS

only paramagnetic species present and that each Crf3 ion realizes its full theoretical mqment of 3.87. Preliminary istudies in which an argon-cyclohexane charge was passed over an unpromoted chromia-alumina catalyst (10 wt. % Cr) showed that, the products of the reaction were cyclohexene, benzene, and a small amount of methylcyclopentane. Traces of 3-methylcyclopentene, which mould have appeared as part of the cyclohexane peak on the chromatogram, could have been present, although mlore detailed chromatographic m d infrared analyses indicated that it was not a significant product. The 1-methyl and 2-methyl isomers were not found. Unfortunately, the catalyst aged so rapidly during the course of the run that reliable kinetic measurements could not be made, and it was necessary to add hydrogen to the charge in order to obtain stable catalyst activities. Under these conditions the activity was essentially constant over a 48-hr. period and, consequently, all subsequent experiments were carried out using the argon-hydrogen-cyclohexane charge. I n addition to stabilizing the catalyst t,he hydrogen also changed the product distribution by increasing the amount of methylcyclopentane formed. Again, while small qua,ntities of the methylcyclopentenes may have been present, they were formed only in trace amounts at best. I n view of the observed product distribution, the rates which were measured for the present study were the rate of cyclohexane reaction ( T C H ) and the rates of methylcyclopentane (TMCP), benzene ( T B ) , and cyclohexene ( Y C H ~ ) production. I n general, the conversion of cyclohexane varied from 0.5 to 96.0% depending upon temperature, space velocity, and catalyst; the amounts of benzene, cyclohexene, and methylcyclopentane in the product strezm ranged from 0.4 to 96.0,O.l to 0.6, and 0.5 to 4.0Yc, respectively. To determine the rate of cyclohexane reaction both an integral and differential method of treating the data were tried. For the case of a flow reactor the rate of cycbohexane reaction may be expressed by the steady st ate mass balance equation T c a d A =:

FdfcH

(1)

where PCH

F A ~ C

= rate of cyclohexane reaction (moles/rnin.-m.2)

feed rate (molrs cyclohexane/min.) = surface area (m.2) H= fraction of cyclohexane reacted =

Therefore, if the conversion, ~ C H is , plotted against the quantity, A/F,then the differential reaction rate, ~ C H ,a t any particular conversion is given by the slope of the curve a t that conversion. I n the present work rates mere obtained by measuring the slopes of such curves with a tangentimeter. As discussed e l ~ e w h e r eit, ~often ~ ~ ~is~ useful to determine differential reaction rates a t very low conversions (ie., infinite space velocity) where the effect of reaction products upon the kinetics is minimal. Obviously, the rate a t infinite space velocity, YCK', may be obtained directly by measuring the initial slope of the conversion us. A / F (12) C. D. Prater and R. M. Lago, Advances in Cutalyais, 8 , 293 (1956).

(13) P, B. Weisz and C . P, Prrtter, i b i d . , 6, 143 (1954).

873

Fig. 1.-Log of the rate of cyclohexane reaction a t 416' obtained from conversion-reciprocal space velocity plots us. log of the partial pressure of cyclohexane: 0 , 10.47 wt. yo Cr, 0.36 wt. yo K; 9, 10.34 wt. % Cr, 2.25 wt. Yo K; 0, 6.10 wt. % Cr, 0.0 wt. % K.

curve. Such a procedure, however, places considerable weight on the extrapolated portion of the curve and for this reason an alternative method was used to obtain YCHO. Weller14 has indicated that for many heterogeneous reactions the rate may be expressed as a simple power dependency of rate on concentration. In the present case of cyclohexane conversion, therefore, the rate was assumed to be given by TCH

=

(2)

bCHn

where IC

constant partial pressure of cyclohexane = reaction order =

pCH =

n

and plots were made of log TCH z's. log ~ C H . Examples of such plots are shown in Fig. 1. In all cases these plots were linear, and values of n could be obtzined from the slopes; these values were than substituted in eq. 2 and by setting pc15 = pCKo,the initial partial pressure of cyclohexane, velues of rc11.O were calculated. If eq. 2 is valid over the entire range of conversions studied and assuming for the moment that n = 1, then combining eq. 1 and 2 gives, upon integration a t constant total pressure ( A P ) = - [(I SI/) In (1 - SCH) 6I/fCH] (3) where

+

+

6 = increase in the no. of moles of the reacting system/ mole of reactant converted (i. e., 6 = 3 for the dehydrogenation of cyclohexane to benzene) I/ =

mole fraction of cyclohexane in the entering feed

The fit of the data obtained in this work to eq. 3 is illustrated in Fig. 2 for several chromia-alumina catalysts. The slopes of these lines give kpCHO = RCH,the integral reaction rates. In Table 111, the values of n, TCHO, and RcH at 416" are given for the various chromia-alumina catalysts tested. From this it may be seen that n. is very close to unity for all the catalysts, indicating that the rate of cyclohexane reaction is approximately first order, and that, within experimental error, RCH= YCH. These facts justified the use of the integral method represented by eq. 3 and, be(14) S. Weller, A.I.Ch,E. Jozcmul, 2, 59

(1956),

J. M. BRIDGES, G. T. RYMER, A U D D. S.RIACIVER

Yol. 66

TABLEI11 RATEO F CPCLOHEXASE REACTION OVER CHROYIA-AILMISA AT 416” TCH

Reaction Wt. % Cr Wt. 70 K order, n

2 4 6 10

11 35 10 60 11 05 10 47 10 57 10 04 10 34 a RCH=

0 0 0 0 0 20 36 86 1 25 2 25 activation

x

108,

moles/ min.-m.z

RCH

x

108

moles) min.-rn.z

EcrI,a kcal./ mole

1 11 1 18 1 17 31 0 95 3 56 3 64 30 1 15 7 56 7 60 30 1 05 12 6 12 9 31 1 00 11 6 11 9 30 1 00 31 5 32 0 31 0 99 32 0 32 7 30 1 14 22 3 23 2 30 1 12 14 6 14 3 31 energy for Cyclohexane reaction.

to the ratio of the two rates a t this conversion. Typical examples of such plots are shown in Fig. 3 0 4 8 I2 for benzene and methylcyclopentane. The linearx lo-6 (*). ity of these curves indicates that the rate of formation of both products was first order with reFig. 2.-First-order rate plots for the reaction of cyclohexane at 416’: 0 , 10.47 11%. % Cr, 0.36 wt. % IC; 9, spect to cyclohexane, and, hence, it was possible to 10.34 wt. 7 0 Cr, 2.25 tvt. yo K ; 0, 6.10 wt. % Cr, 0 wt. obtain the initial rates rBn and T M C ~ O by taking the 9% K. product of the slopes and TCH. I n Table IV are summarized values of TB and ?“MCpn at 416’ for all the chromia-alumina catalysts studied.

+

r

u)

i

!i

W

0

I-

8

0.03

TABLEI V RATESOF BENZENE AKD METHYLCYCLOPENTANE FORMATION OVER CHROMIL-ALUMIOA AT 416’

0 C

z 0.50

W

z

+

5 0.02

W

z

::

N

z m

>

w

-I

2.

E

@4

0.25

2

0.01

V

z

2

0

0 K

L

-

n 0.25

n

0.W

0 75 FRACTION CYCLOHEXANE IN PRODUCT STREAM.

100

Fig. 3.-Frartion methylcyclopentane and benzene us. fraction cyclohexane in the product, stream j n the case of 6.10 wt. 70 Cr&-A1203 a t 416: 0, fraction methylcyclopentane; 0 , fraction benzene.

cause of its convenience, this method was used, generally, to obtain the rate of cyclohexane reaction ( r c ~ O ) . For the case of the rate of formation of a product an equation analogous to eq. 1 may be written as r,dA = F dj,

(4)

where rP = rate of formation of product p (moles p/min.-m.2) jp = fraction of reactant converted to p

Equations 1 and 4 may be combined to give and, therefore, iff, is plotted against J(cH, the slope of the curve a t a particular conversion level is equal

rBo

Wt. % Cr

Wt. % K

X lo8, moles/ min.-m.Z

kcal./ mole

rxcpo X lo8, moles,/ min.-m.%

EICPb

kcal./ mole

2 11 0 0 8G 35 0 31 23 4 35 3 11 35 38 24 0 6 10 7 20 31 27 20 0 10 60 12 9 31 26 27 0 11 9 30 Small 11 05 0 20 10 47 36 32 0 30 0 10 57 86 32 7 30 0 10 04 1 25 23 2 30 0 10 34 2 25 14 3 30 0 a EB = activation energy for benzene formation. I,E ~ P = activation energy for methylcyclopentane formation.

Under the reaction conditions employed in the present work only trace quantities of cyclohexene were produced. At cvclohexane conversions of less than 60% only 0.5 i 0.1% of t)he cyclohexane was converted to cyclohexenc>and, as near as could be determined at those low levels, this conversion was independent of cyclohexane concentration. In other words, the cyclohexene was present in steady state concentrations ( i e . , TCHs = 0 ) . Above 60% cyclohexane conversion, the amount of cyclohexene in the product dropped off; the analytical data, however, were insufficient to permit expression in the form of a rate equation. In all cases the amount of cyclohexene found in the product stream was considerably less than the equilibrium value. Activation energies for the various reactions were computed in the usual fashion from the Arrhenius equation using the differential rates at infinite space velocities. The Arrhenius plots are illustrated for several catalysts in Fig. 4 and the

energy values obtained are presented in Tables I11 and IV for all the chromia-alumina catalysts investigated. It should be mentioned that, under the experimental conditions employed, diffusion effects did not have any appreciable influence upon the observed rate constants. Using the procedure described by Wheeler,16 an effectiveness factor of >0.99 was calculated for each of the several catalysts employed, indicating that the entire internal surface area of each catalyst was effective in the reaction. I n order to obtain information on the respective roles of alumina and chromia in the reaction mechanism the dehydrogenation of cyclohexane was studied over pure chromia, pure alumina, and mechanical mixtures thereof. The results at 416" are summarized in Table V. Only in the case of pure chromia was an activation energy for the cyclohexane reaction measured, and a value was found of 40 f 3 kcal./mole. It is interesting to compare the activity of pure chromia with the chromia mixed with alumina. In the former case about 30% of the cyclohexane was converted to benzene at 416" and a feed rate of 40 cc. (STP)/ min.; this catdyst activity level was constant throughout the length of the run (3 hr.). I n the case of the mechanical mixture a complete mass balance was not possible because of extensive cracking. Under the same conditions as above, however, 6% of the cyclohexane was converted to benzene andl methylcyclopentane at the beginning of the experiment; after 3 hr. on-stream, the conversion had dropped to 3%. When the alumina, prior to mixing, was treated with potassium, the mixed catalyst behaved essentially the same as pure chromia. The pure alumina did not catalyze the conversion to either cyclohexene or benzene ; instead, a large amount of cracking took place. TABLE V REACTION RATESAT 416" OVER MIXEDCATALYSTS Catalyst

rcrlo X 106, moles/ min.-m.z

05 I45

I40

+

I50 Y

I55

I60

IO~,(OK:$,

Fig, 4.-Arrhenius plots: 0 , cyclohexane reaction and benzene production over 10.47 wt. % Cr, 0.36 wt. 70IC; 8 , cyclohexane reaction and benzene production over 10.34 wt. % Cr, 2.25 wt. % K; 0,cyclohexane reaction over 6.10 wt. % Cr; 0, benzene production over 6.10 wt. % Cr; ii, methylcyclopentane production over 6.10 wt. % Cr.

Discussion In general, chromia-alumina may be considered as a dual functional catalyst,l having an acid function which catalyzes reactions such as isomerization and a hydrogenation-dehydrogenation function. It has been proposedl that cyclohexane reacts over a catalyst of this type according to the scheme

0

D

-e-

+ H

C I

C I

?Bo

X 105, moles/ min.-m.2

rktcpo

moles/ min.-m.z

a A1203 0 0 Cr,Oa 1 88 1 88 0 Cr20a(5.2 wt. % K ) 0 06 0 03 0 a h b Cr203 A1203 CrzOa (2.3 wt. % K) 1.18 1 18 0 a Extensive cracking of cyclohexane. Conversion data too erratic to permit calculation of specific rate. Product was obtained but the conversion data was too erratic to permit calculation of specific rate.

+ +

A final experiment consisted of saturating the argon-hydrogen-cyclohexane charge with methylcyclopentane al 25" and passing it over a potassium promoted chromia-alumina catalyst (10.5 wt. % Cr and 0.86 wt. % K). Before addition of the methylcyclopentane 76% of the cyclohexane was dehydrogenated to benzene at 42 l o and a feed rate of 40 cc. (STP)/min. ; in the presence of methylcyclopentane only 50% of the cyclohexane was converted. The poisoning was irreversible. (15) A. Wheeler, "Catalysis," Vol. 11, ed. b y P. H. Emmett, Reinhold Publ. Corp., New York, N. Y., 1955, Chapter 2,

where H-D represents reaction over hydrogenation-dehydrogenation sites and A, reaction over acid sites. The data obtained during the present work is consistent with this mechanism in that, using an unpromoted chromia-alumina, the cyclohexane reacted to give benzene, cyclohexene, and methylcyclopentane. Because of this distribution of products it was possible to consider the promoting action of potassium in terms of both activity and selectivity. The variation of activity with potassium content may be seen from Fig. 5, where the rate of cyclohexane reaction at 416' has been plotted as a function of the amount of potassium in the catalyst. It will be noted that the first increment of potassium added to the catalyst did not change the activity; beyond this point, however, the activity increased abruptly and then declined as additional potassium was added. As may be seen from Table IV, a change in the selectivity of the reaction accompanied this sudden rise in activity, in that, at this point, the methylcyclopentane disappeared from the product stream. That these two phe-

J. M. BRIDGES, G. T. RYMER, AND D. S. h h C T V E R

876

In I

3.0

1.0

I 0

Fig. 5.--Rate

I 1.0 WEIGHT PER CENT POTASSIUM.

1

I

I

20

of cyclohexane reaction a t 416’ vs. weight

% potassium for those catalysts containing approximately 10.5 weight % ’ chromium. nomena were related was apparent from the experiment in which methylcyclopentane was deliberately added to the charge over a promoted chromia-alumina catalyst; the resulting irreversible decrease in catalyst activity strongly suggested that the methylcyclopentane was acting as a poison or a precursor to a poison and that the promoting role of the potassium was one of suppressing the formation of methylcyclopentane. According to the reaction scheme given above this latter compound is formed by the isomerization of cyclohexene over acid sites, and it is significant that potassium has been found to decrease the surface acidity of chromia-alumina c a t a l y ~ t s . ~ ? ~ Further evidence bearing on this point is provided by the data in Table V on the activity of mechanical mixtures of chromia and alumina. It is to be noted that no methylcyclopentane was formed over pure chromia and that the catalyst was quite stable. A mixture of alumina and chromia, on the other hand, produced methylcyclopentane, and, in this case, the catalyst aged quite rapidly. As indicated earlier, a mixture of potassium-treated alumina and pure chromia behaved essentially the same as chromia alone. Thus it would appear that, under the conditions employed, chromia did not cause the C6-to C6-ring isomerization and that the acid sites of an alumina surface were required for the reaction. The potassium apparently serves to poison these sites, thereby destroying the acid function associated with the alumina and hence preventing the formation of the methylcyclopentane which deactivates the chromia dehydrogenation sites. The acid nature of alumina and the effect of alkali thereon has been demonstrated recently by the studies of Pines and Haag’c and their findings are generally consistent with those reported above. It is interesting to observe that, in the present work, the Cg- to Crring isomerization did not take place over pure alumina. This demonstrates the necessity of the cyclic.olefin intermediate, this latter being present only when the dehydrogenation component was included in the catalyst bed. (16) H. Pines and W.0. Haag, J . Am. Chem. SOC.,84,2471 (1960).

Vol. 66

It is believed that the results just discussed for the mked catalysts are relevant also to the impregnated chromia-alumina catalysts. Earlier workll has indicated that a chromia-alumina catalyst prepared in the manner indicated has only about 20% of its surface covered with chromia, the chromia probably being present as crystallites on the alumina.’’ Thus, one can consider that in such a catalyst a sizable portion of the surface is similar in chemical properties to a pure alumina surface. I n other words, a chromia-alumina surface may, for certain purposes, be regarded simply as a mixture of chromia and alumina surfaces. On this basis, treating the chromia-alumina catalyst with potassium poisons the acid sites associated with the alumina portion of the surface, thereby preventing the formation of methylcyclopentane and, hence, promotes the dehydrogenation activity. The views expressed above concerning the role of methylcyclopentane in the poisoning of the chromia dehydrogenation sites are quite similar to the conclusions reached by Myers, Lang, and Weisz18who found that, using a platinum re-forming catalyst, cyclopentane reacted to give a poisonous intermediate, possibly cyclopentadiene, which acted as a catalyst deactivator. They postulated that this intermediate was adsorbed on the catalyst surface, where it underwent a polymerization to form “coke.” The poisonous nature of Cs-ring compounds probably is related to the tendency of such compounds to polymerize when in contact with an acid catalyst19and to their ability to form complexes with transition metals and transition metal ions.Z0 As in the case of platinum re-foTming catalysts, the presence of hydrogen in the charge to the chromia-alumina catalysts reduced the aging rate, presumably by hydrogenating the catalyst poisons. As shown in Fig. 5 the cyclohexane reaction rate passed through a maximum as potassium was added to the catalyst, and it remains to explain the deactivating influence of excess potassium. The most obvious explanation is that this potassium decreases the number of dehydrogenation sites by causing a decrease in the chromia surface area (is.,sintering), by chemically combining with the dehydrogenation sites, or by simply physically blocking or covering a portion of the chromia surface. The magnetic data in Table I1 indicate that the first increment of potassium added did not have any great influence on the magnetic properties of the catalyst, probably because, at this point, the potassium was associated primarily with the alumina portion of the surface. Additional potassium, however, was directly associated with the chromia, as is indicated by the decrease of peff in the oxidized state. It is apparent that when the catalyst is oxidized the potassium is involved with (17) R. P. Eischens and P. W. Selwood, ibid., 69, 1590 (1947). (18) C. G. Myers, W. E.Lang, and P. B. Weisz, Ind. Eng. Cham., 13, 299

(1961).

(19) C . Ksmball and J. J. Rooney, Proc. Roy. Soc. (London), 267A, 132 (1960). (20) L. E. Orgel, “An Introduction to Transition-Metal Chemistry: Ligand-Field Theory,” Methuen and Co. Ltd., London, 1960, Chapter 10.

May, 1962

PRESSURE OF GaaO OVER GALLIUM-G~~O~ MIXTURES

the chromia in such a way as to stabilize the Cr+6 electron configuration; this is consistent with the observation of Volta and Weller3 that potassium increases the surface oxidation state of a chromiaalumina catal;rst, possibly via the formation of potassium chromate or dichromate. The potassium-chromium complex, however, does not appear to be stable against reduction since, as shown in Table 11, treatment with hydrogen restored the effective moments of the promoted samples to that of the unpromoted catalyst. Since activities were determined with the catalysts in the reduced state it would not appear that the decrease in activity came about simply because of a change in average oxidation state of the chromia. However, despite the fact that the potassium-chromium complex is reduced, the po1,assium presumably remains at least physically associated with the chromia in some way since reoxidation of a promoted catalyst gave the same effective moment as was observed prior to the reduction. This potassium may exist in some form such as 820 on the surface of the reduced chromia and in this way block some of the dehydrogenation sites. In line with this it will be observed from the data of Table V that treating pure chromia with potassium decreased the reaction rate per. unit area, suggesting that! the same effect is possible when the chromia is supported on an alumina surface. Another possible explanation for the deactivation is that the potassium causes a sintering or clumping of the chromia crystallites on the alumina. It is interesting to note that pure chromia decreased in area when treated with potassium (Table I) and tha’. the Weiss constants of the reduced potassium promoted chromia-alumina catalysts increased slightly with potassium (Table 11), as would be expected if the chromia was becoming less well dispersed over the alumina surface. In any case, it appeared that the deactivation could best be explained on the basis of a decrease in available chromia area either by blocking or sintering.

a77

Earlier work by Bridges, MacIver, and Tobin’l has shown that the amount of oxygen chemisorbed by a reduced chromia-alumina catalyst at - 195’ is an approximate measure of the extent of chromia surface, that is, the portion of the total surface which is contributed by the chromia component. Similar oxygen chemisorption measurements made on the present series of promoted chromia-alumina catalysts are reported in Table I, where it may be seen from the data in column 7 that as the potassium concentration increases, the amount of oxygen chemisorbed per unit catalyst area decreases. This would seem to suggest that the potassium causes a decrease in the available chromia area which, as discussed above, is suflicieiit to explain the decline in dehydrogenation activity of the catalysts at high potassium concentrations. It is of interest to note that the activation energy for the cyclohexane reaction was independent of both the chromium concentration and the level of potassium promotion (Table 111), suggesting that there is no change in the energetics of the reactantcatalyst interaction as these two factors are varied and that the promotion does not take place via some “electronic” factor entering into the activation energy. Using pure chromia, however, the activation energy was significantly greater (10 kcal./mole) than that over the supported chromia catalysts, which may be indicative of a basic chemical or physical difference between supported and unsupported chromia. A change in the lattice constants of the chromia when it is placed on a support, for example, possibly would be significant since the dehydrogenation reaction is thought to involve a dual site inechanism,21 which could be energetically dependent upon the distance between active sites on the catalyst ~urface.22~23 (21) E. F. G. Herington and E. K. Rideal, Proc. Roy. Soc. (London) 190A,289 (1947). 122) A. Sherman and H. Eyring, J. Am. Chem. Soc., 54, 2061 (1932). (23) A. M. Rubinshtein, S. G. Kulikov, and N. A. Pribytkova, Doklady Akad. N a u k S.S.S.R., 85, 121 (1952).

THE PRESSURE OF GazO OVER GALLIUM-Ga203 MIXTURES BY C. J. FROSCH AJSD C. D. THURJ~OND Bell Telephone Laboratories, Inc., Murray Hill,New Jersey Received October 88, 1961

The pressure of GaeO over a mixture of Ga(1) and GazOs(s)has been measured over the temperature range of 800 to 1000° by the transport method. A heat of formation at 298°K. for GasO(g) of -20.7 kcal. has been obtained.

We have measured the pressure of Ga20 over mixtures of Ca and Ca20sin the temperature range 800 t o 1000’ b y the transport method. The loss in weight of the mixture was measured as a function of time, teimperature, and flow rate. A pestled mixture of 5 moles of Ga to 1 mole of Gaz08 was placed in a fused silica boat which was heated in a fused silica tube under a helium gas stream. When this mixture was heated for a sufficient length of time it reached constant weight and all of the oxide had disappeared. The weight of Ga left was that calculated for the evaporation of GaaO.

Brukl’ has shown that the solid phase condensed from the vapor phase obtained by heating GaGa203mixtures has the composition GazO. Mass spectrometric evidence for the vapor species GaaO has been obtained by hiitkin and Dibelerz and others.3 The experimental results are shown in Fig. 1 where the logarithm of the ratio W / V hits been (1) D. A. Brukl. Z. anorg. allgem. Chem., 203,23 (1932) (2) S. Antkin and V. H. Dibeler, J . Chem. Phys., 21, 1890 (1953). (3) W. S. Chupka, J. Berkonitz, C . F. Giese, and M. G Inghram, J. Phys. Chem., 62, 611 (1958).