66
Ind. Eng. Chem. Prod. Res. Dev., Vol. 17, No. 1, 1978
Influence of Metal Dispersion on n-Hexane Reactions over Platinum-Alumina Catalysts E. Santacesaria and D. Gelosa Centro Nazionale Propulsion8 ed Energetics del C.N.R., Milano, ltaly
S. Carra* and 1. Adami lstituto di Chimica Fislca de/ Politecnlco di Milano, 20 133 Milano, ltaly
The n-hexane transformations over alumina-supported platinum have been investigated in order to understand the influence of metal dispersion on the catalytic processes. To obtain reproducible behavior of the catalysts a set of pretreatments was necessary. Hydrogenolysis and dehydrocyclizationrates increase with dispersion while the isomerization rate does not change. The ratios among the isomerization products (Bmethylpentane and 3methylpentane) remain constant for the examined catalysts. This finding suggests the preeminence of a mechanism which occurs through a cyclic intermediate.
Introduction It is well known that the activities and selectivities of metallic catalyzed reactions can be affected by particle sizes of the supported metal. This fact, which has been pointed out by Boudart et al. (1966),has encouraged much research on this subject. Several authors have studied, for example, the influence of platinum dispersion in hydrogenolysis and isomerization of saturated hydrocarbons. The conclusions reached in these works have been reviewed by Anderson (1973). More recently, Brunelle et al. (1976) have studied the influence of platinum particle sizes in hydrogenolysis and isomerization of n-pentane. Despite the large number of published works there are still some conflicting aspects on the subject and the role of the different reaction paths is open to discussion. In the present paper some results concerning the transformations of n hexane over supported platinum will be given. The examined process offers the possibility of deepening the effect of particle sizes, since in it both isomerization and hydrogenolysis reactions occur, according to the following scheme. Scheme I
-
bond shift MCP
*
cracking
n-H = normal hexane; 2-MP = 2-methylpentane; 3-MP = 3-methylpentane; MCP = methylcyclopentane. In previous work performed by employing Pt-black as a catalyst, Santacesaria et al. (1975) have shown that different pretreatments lead to different catalytic activities and selectivities. This behavior has been confirmed by working with supported platinum as catalyst. These findings are consistent with the observations made by Furhman and Parravano (1976) and by Hassan et al. (1976) on the behavior of supported metallic catalysts in the presence of several gaseous atmospheres. 0019-7890/78/1217-0068$01.00/0
From these facts, it follows that the study of the influence of metallic dispersion in catalytic reactions requires accurate definition of catalyst preparation and pretreatments in order to obtain reliable reproducibility. In this work, the influence of dispersion of platinum supported on alumina in the n hexane reactions has been examined after a series of pretreatments necessary for obtaining reproducibility of the kinetic data. The role of pretreatment in the attainment of stable and reproducible catalysts will be discussed in a future paper. Experimental Section Catalyst Preparation. The catalysts employed were prepared by impregnation of y-alumina powder with hexachloroplatinic acid. The y-alumina was prepared by calcinating a t 600 OC, for 6 h, aluminum hydroxide obtained by the hydrolysis of aluminum isopropoxide, according to Pines and Haag (1960). The prepared alumina had a surface area of 177 m2/g, measured with the BET method. The impregnation of the support was accomplished by using aqueous dilute solutions (from to lod4 M) of hexachloroplatinic acid, taking into account the behavior of the y alumina in the impregnation, as described by Santacesaria et al. (1977). By this method of impregnation it is also possible to obtain high dispersion catalysts for 2 wt % platinum. For catalysts containing more than 4 wt 96 platinum, it has been necessary to complete the catalyst preparation by evaporating the solutions to dryness, with continuous stirring, in order to obtain some uniformity of the catalysts. The catalysts, after filtration and drying (at 100 "C, for 12 h), were reduced with hydrogen (2.4 L h at 450 "C for 3 h). After reduction, the catalysts were submitted to a series of treatments consisting of washing for 1 h, at the same temperature, with N2 (3 L/h) and then introducing ten pulses of 0.5 cm3 of oxygen diluted in nitrogen (5 mole 96) and ten pulses of diluted hydrogen (5 mole %). Then the catalyst was submitted to the treatments suggested by Compagnon et al. (1974) for polishing and measuring the platinum surface area. These treatments consist of a prolonged wash of the catalysts with a flow stream of nitrogen (3 L/h) at 450 "C for 12 h, followed by the introduction, a t 200 "C, of several pulses of 0.5 cm3 of pure 02 in a nitrogen stream and then pulses of diluted Hz (5%in N2). We have found that the entire cycle of the 0 1978 American Chemical Society
Ind. Eng. Chem. Prod. Res. Dev., Vol. 17, No. 1, 1978
Table I. The Mean Characteristics of Experimental Catalysts. The Average Particle Diameter Has Been Evaluated As Suggested by Compagnon et al. (1974)
Catalvst
Pt. wt %
HPt
Surface area, m2/g of Pt
A B C D
0.22 0.60 1.69 4.72 9.47
0.95 0.94 1.17 0.66 0.18
268 260 324 182 51
E
Av particle diameter, 10 10 10 16 55
A
69
1
0.16
= O.l+
ISOMERIZATION
Table 11. The Kinetic Behavior of the Low Dispersed Catalvsts I
Activities (molesh m2of Pt)
I
t
Catalyst Isomerization Dehydrocyclization Hydrogenolysis D E
3.73 3.62
0.29 0.50
2.44 2.96
Table 111. Conversion % of n-Hexane Obtained for Several Catalysts
01
0
I
I
I
05
1.0
1.5
Pt
% .
Figure 1. Specific activities, expressed as mol/h g of Pt, for highly dispersed catalysts.
Conversion, 96 Catalyst A B
C D E
Isomerization Hydrogenolysis Dehydrocyclization 1.40 2.41 10.77 17.04 9.5
1.76 2.11 5.46 11.15 7.45
1.60 1.61 2.29 1.35 1.29
described treatments must be repeated a t least 3 or 4 times in order to obtain constant kinetic data even if the surface area of the catalysts does not change. Therefore, the treatments used revealed an activating and stabilizing action on the kinetic behavior of the catalysts. In fact, it must be pointed out that in the kinetic runs performed after each treatment there was a reasonable reproducibility run for run, but significant changes in kinetic data intervened after further treatment and so forth for 3 or 4 times. The n-hexane was also fed on pure alumina in the same experimental operative conditions used for the catalytic runs. In this case it was verified that there were no reactions. Therefore, intrinsic activity of the alumina used in the n hexane reactions can be excluded, a t least a t our operating conditions. The amounts of platinum in the catalysts have been determined by analyzing initial and residual solutions of hexachloroplatinic acid and then controlled by a direct chemical analysis of the catalyst. Apparatus a n d Operative Conditions. The apparatus used in the runs has been described in a previous work of Santacesaria et al. (1975). The catalytic bed consisted of 0.5 g of catalyst (40-60 mesh), diluted with 2 g of alundum granules (40-60 mesh). The isothermicity of the catalytic bed was obtained by a sand-fluidized bed. The temperature of the catalytic bed was kept, during the kinetic runs, a t 300 "C. The n-hexane was fed by a motor-driven syringe and diluted with a flow stream of hydrogen. The flow stream of n hexane was 0.00942 mol/h and that of hydrogen was 0.099 mol/h. The n-hexane used was produced by the Carlo Erba Co. a t a very high degree of purity (type RS). The analysis of the reaction products was performed on a Perkin-Elmer 880 gas chromatograph with FID detector. The gas chromatographic columns employed were 6 m in length, filled with bentone plus diisodecylphtalate (l:l),a t 10 wt %, on Chromosorb W. The
0 Pi %
Figure 2. Selectivity isomerization plus dehydrocyclization/hydrogenolysis trend as a function of wt % of Pt compared with the measured H/Pt trend. sampling valve and the exit tubes from the reactor were heated to prevent the products from condensing. Results In Table I the main properties of the experimental catalysts are reported. These catalysts can be separated into two types: those having nearly atomic dispersion, in which almost all the atoms are accessible to the reactants (A, B, C), and those a t low dispersion (D, E). By observing the data reported in Table I, we note that for the catalysts A, B, and C the ratio H P t is about 1. Ratios of H/Pt higher than 1 have already been reported in the literature, for example, by Rabo et al. (1965) and Bond (1970). An extensive analysis of the phenomenon has been reported by Karnaukhov (1971). At high dispersion of the metal, the stoichiometry of hydrogen chemisorption becomes uncertain since the value of the ratio H P t goes toward 2 instead of 1. Moreover, the spillover phenomenon occurs. For these reasons, in such conditions, the measurement of platinum surface area by chemisorption becomes less accurate. In order to avoid erroneous interpretation of the obtained data due to the difficulties of correct evaluation of the surface area for highly dispersed catalysts, we have referred the kinetic data of the catalysts A, B, and C to the amount of platinum by weight. This fact is justified if one assumes that almost all
70
Ind. Eng. Chem. Prod. Res. Dev., Vol. 17, No. 1, 1978
Table IV. Products Distributions Obtained on Several Catalysts for n-Hexane Reactions Composition of c6 hydrocarbons Catalyst A B C D E
2MP 33.6 43.3 61.5 66.3 63.3
3MP 12.4 16.5 20.8 26.3 24.4
Hydrogenolysis
MCP 53.9 40.2 17.6 7.3 12.2
Figure 3. Ratios 2MP/3MP for several catalysts using n-hexane and methylcyclopentaneas reagents. atoms of platinum of the catalysts are accessible to the reactants. In Figure 1,the isomerization, hydrogenolysis, and dehydrocyclization reaction rates as a function of percent platinum are reported. This plot is related to the high dispersion catalysts A, B, and C and therefore the reaction rates are expressed as m o l h g of platinum. As can be clearly observed, the specific rates for hydrogenolysis and dehydrocyclization increase by decreasing the platinum content, while the isomerization rate remains constant. In Table 11, the isomerization, hydrogenolysis, and dehydrocyclization reaction rates for catalysts D and E are given. They are now expressed in m o l h m2 of platinum. It can be observed that the reaction rates in these two cases are not strongly affected by the amount of platinum present in the catalysts. In Table 111, the conversion percents obtained for the examined catalysts are also reported. In Figure 2 the values of the selectivity (isomerization dehydrocyclization/hydrogenolysis)are compared with the trend of the measured ratios H/Pt as a function of percent platinum. This plot shows a structure sensitivity of the selectivity when the ratios H/Pt are next to 1 (catalysts A, B, C), and a complete insensitivity for catalysts D and E. In a previous work Santacesaria et al. (1975) observed that the value of the selectivity for Pt-black was 1.4. The comparison of this result with the ones given in Figure 2 confirms the structure insensitivity for particles having a diameter greater than 20 A. Therefore, the structure sensitivity seem to be restricted to the microclusters size. In Figure 3 the ratios (2MP/3MP) obtained for the various catalysts are reported. As can be seen, these ratios are almost constant and equal to 2.75. This value was also obtained if methylcyclopentane was fed on the catalytic bed instead of n-hexane in the same operative conditions. The products distributions for the various catalysts are reported in Table IV. As can be seen, the high dispersion does not change the distributions of the hydrogenolysis products in a significant way. For what concerns the distribution of the c6 hydrocarbons, a strong increase in the amount of MCP with dispersion can be observed. This fact confirms what has already been observed by several authors, for example, by Anderson and Avery (1966).
+
C1 18.0 19.3 23.1 20.2 19.1
c2
c3
c4
c5
16.8 18.5 16.5 16.4 17.1
32.3 33.1 28.7 33.7 34.0
16.3 17.0 15.9 14.9 16.2
16.6 12.0 15.6 13.7 13.5
Discussion a n d Conclusions On the whole, from our results it appears that the general behavior of the reaction studied does not change the product distribution appreciably by going from black metal to high dispersion supported platinum. This is consistent with the fact that pure alumina at 300 OC does not have intrinsic activity in transforming n-hexane. Therefore in our experimental conditions the bifunctionality of the catalysts considered is negligible. Some remarkable effects of platinum dispersion on nhexane reaction have been revealed in the previous section. The specific activities of hydrogenolysis and dehydrocyclization reactions, for example, increase with dispersion, while the specific activity of the isomerization reaction does not change. The selectivity expressed as (isomerization dehydrocyclization/hydrogenolysis)increases with dispersion as a consequence of the increase of dehydrocyclization reaction rate. Finally, the amounts of MCP in the reaction products strongly increase with dispersion. The ratio 2MP/3MP, on the contrary, remains unchanged for high or low dispersion. This fact seems to indicate that the dispersion of platinum does not modify the pathway of the isomerization reaction, which probably occurs through a cyclic intermediate as previously shown. According to the mechanism of dehydrocyclization, suggested by Shepard and Rooney (1964), the ring closure reaction would take place over a single atom of platinum. If we accept this mechanism, the large amounts of MCP in the reaction products obtained with highly dispersed catalysts can be explained by assuming that the dehydrocyclization reaction occurs on monoatomic sites at the corners of the platinum lattice. As has been shown by Cam$ and Ragaini (1969),these sites strongly increase with the dispersion. Therefore, the ring closure reaction rate increases with dispersion while the ring opening reaction rate remains constant. The overall effect is an increase in the amount of the intermediate product, that is, MCP. This fact also justifies the trend of the selectivity reported in Figure 3. As has been shown in Figure 1,the hydrogenolysis reaction rate increases with dispersion. Since, for this reaction, the intervention of diatomic sites seem to be more likely, as suggested by Anderson and Avery (1966), the number of these sites should also increase with dispersion. This fact is qualitatively confirmed by the statistical analysis of the probability of the formation of biatomic clusters on the surface, as suggested by Kobozev et al. (1961). However, it is necessary to emphasize, as a conclusion, the difficulty of separating the influence of metal dispersion from those of catalyst pretreatments, because these two effects strictly interact. This fact is enhanced a t high values of dispersion because of the presumably high reactivity of the atoms with low coordination degree.
+
Acknowledgments Thanks are due to C.N.R. (Consiglio Nazionale Ricerche)
Ind. Eng. Chem. Prod. Res. Dev., Vol. 17, No. 1, 1978
71
for financial support and to Professor Corrado Casci for encouragement in undertaking this work.
Hassan, S. A., Khalil, F. H., El-Gamal, F. G., J. Catal., 44, 5 (1976). Karnaukhov, A. P., Kinet, Katal., 12 (6),1345 (1971). Kobozev, N. I., Lebedev, W. P., Malzev, A. N., Z.fhys. Chem. (Leipzig),217,
Literature Cited
Pines, H., Haag, W., J. Am. Chem. SOC.,82, 2471 (1960). Rabo, J. A,, Schomaker, V., Pickert, P. E., froc. Int. Congr. Catal., 3rd, Amsterdam, 1964, 2 1264 (1965). Santacesaria,E., Gelosa, D. Carra, S., J. Catal., 39, 403 (1975). Santacesaria, E., Carra, S., Adami, I., lnd. Eng. Chem. Rod. Res. Dev., 16,41
l(1961). Anderson, J. R., Adv. Catal., 23, l(1973). Anderson, J. R., Avery, N. R., J. Catal., 5 , 446 (1966). Bond, G. C.. Proc. lnt. Congr. Catal., 4th, Moscow(1968). Boudart. M.. Benson. J. E.. Aldaa. A. W., Dougharty, . N. A,, Markins, C. G., J, Catal., 6, 92 (1966). Brunelle, J. P.. Sugier, A., Le Page, J. F., J. Catal., 43, 273 (1976). Carra, S.,Ragaini, V., Chim. lnd. (Milan), 51 (Il),1215 (1969). Compagnon, P. A., Hoang-Van, C., Teichner, S. J., Bull. Soc. Chim. f r . , 11, 23 1 1
(1977). Shepard, F. E., Rooney, J. J., J. Catal., 3, 129 (1964).
Received for review May 19,1977 Accepted November 29,1977
11974). Fuihman, Z. A., Parravano, G., froc. lnt. Congr. Catal., 6th London, (1976).
Thermal Degradation of Bitumen from the Faja of the Orinoco J. A. Fernandez Lozano,* R. Gonzalez A., and M. Fernandez C. Departamento de Ingenieria Quirnica, Universidad de Oriente, Nucleo de Anzdtegui, Puerto La Cruz, Venezuela
A kinetic study was made on the thermal degradation of bitumen from the Faja of the Orinoco. The sample was heated in a designed coking apparatus while monitoring the products of the thermal degradation. The effects of temperature and heating rate on the yields and properties of the oil, gas, and coke products were determined. Thermal degradation of natural bitumen can be used to produce high OAPl gravity, low pour point oil. The use of higher temperature and higher heating rates favors the production of oil. Nickel, vanadium, and sulfur present in the bitumen are mostly removed with the volatiles. The kinetics of thermal degradation of bitumen above 420 O C are represented by a first-order reaction and below 420 O C by an equation composed of two contributing reactions, one of zero order and the other of nth order.
Introduction Petroleum residuals and bitumen from Athabasca oil sands have been processed by thermal degradation commercially for a number of years (Blaser et al., 1971; Rionda et al., 1974; Bachman and Stormont, 1967; Kett et al., 1974). Pilot plant degradation studies have also been carried out on coke oven pitches and oil shales (Dell, 1959; Berber et al., 1968; Smith, 1962; Johnson, 1966; Whitcombe and Vawter, 1975). In this study, heavy crude from the Faja of the Orinoco has been subjected to controlled thermal degradation studies. A survey of the literature showed that most of the kinetic degradation work was devoted to pyrolysis of oil shale. Some of these studies (Hubbard and Robinson, 1960; Diricco and Barrick, 1956; Johnson, 1966) showed that the rate of decomposition of oil shale corresponds to that for a first-order reaction; it is proportional to the amount of material in the feedstock. Their rate data were adequately described by the first-order reaction eq 1. K1
= ( l / t ) In ( a ) l ( a - x )
(1)
The same workers found that the value of the rate constant, k 1, for oil shale pyrolysis may be represented as a function of temperature by an Arrhenius-type equation as 0019-7890/78/1217-0071$01.00/0
B +C T
Ink = - -
Cracking reactions of oil and oil stocks may be described by eq 1when the decomposition is limited to a low conversion per pass, and an equation similar to eq 2 may be used to express the rate constant, K (Nelson, 1968; Gunnerson, 1948; Sung, 1945). A recent thermal decomposition study of Utah and Athabasca Tar Sand bitumens showed that the kinetic data in the range from 371 to 538 "C were adequately described by a first-order reaction equation (Barbour, 1976). Studies carried out in the University of Utah and interpretation of works of previous investigators (Johnson, 1966) proposed the new kinetic model for oil shale pyrolysis represented by eq 3. Kerogene
- K,
275-365 'C
E
S
4 2 kcal
K,
bitumen
L E,
> 4 5 0 "C
350-450 "C
2
polymer
>46 kcal
oil
+ gas + c o k e
42-46 kcal K, >450 "C
1 4 kcal
(3) oil + gas
+ coke
A more recent work on shale pyrolysis (Braun and Roth0 1978 American Chemical Society
