Langmuir 1994,10, 2487-2490
2487
Notes Preparation of Bimetallic Pt-SdAlumina Catalysts by the Sol-Gel Method Krishnan Balakrishnan and Richard D. Gonzalez'
Department of Chemical Engineering, Tulane University, New Orleans, Louisiana 70118 Received January 24, 1994. In Final Form:April 11, 1994
Introduction Alumina-supported Pt-Sn catalysts are industrially important because of their potential for use in naphtha reforming. These catalysts are especially promising for low-pressure operation and in processes in which continuous regeneration is performed.' An important characteristic of these bimetallic catalysts is that they display better stability during reaction compared to monometallic Pt catalysts. The reason for this improved stability is believed to be due to a change in the catalytic selectivity that leads to a suppression of coke-forming reactions in comparison with reforming reactions.2 Pt-SdAlzO3 catalysts have been traditionally prepared by impregnation methods. In this study an attempt has been made to prepare Pt-Sdalumina catalysts by a different procedure, namely, the sol-gel method. In this method porous gels having a large surface area can be prepared. A unique characteristic of this method is that a solid bimetallic catalyst sample is prepared from a homogeneous solution that contains both the metal precursors and also the support precursor. This leads to a greater degree of control over catalyst preparation and one can "tailor make" catalysts to fit particular application^.^ In the sol-gel method, the solvent can be removed by two different processes and these result in significantly different dried gels. By definition, a xerogel is a product derived by evaporating the solvent under normal conditions and involves some shrinkage of the gel network due to capillary pressure. On the other hand, if the wet gel is dried in an autoclave under hypercritical conditions, the product is called an aerogel. Because of the nature of the drying used, there is no collapse of the gel network during solvent removal. This results in the lower density of aerogels compared to ~ e r o g e l s In . ~ this study we deal only with xerogel samples, though it would be interesting to perform similar studies with aerogel catalysts in the future. The molecular precursors of the support material used in the sol-gel process are metal alkoxides such as aluminum tri-sec-butoxide (ATB) and aluminum isopropoxide (AIP). Starting with aluminum tri-sec-butoxide (ATB), the corresponding hydrolysis and polymerization reactions are as follows:6 (1)Srinivasan, R.; Rice, L. A.; Davis, B. H. J.Catal. 1991,129,257. (2)Burch, R.J . Catal. 1981,71, 348. (3)Sanchez, C.; Livage, J. New J. Chen. 1990,14,513. (4)Hench, L. L.; West, J. K. Chem. Rev. 1990,90,33. (5)Lopez, T.;Gomez, R.; Mendez-Vivar, J.; Campero, A. Lat. Am. Res. 1990,20,167.
Hydrolysis
+
AI(0But)S H20 Polymerization PAI(OH)(OBut)2
-
-
+
AI(OH)(OBut)n ButOH
ButO-Al4-AMBut
I
OH
I
+ 2ButOH
(1)
(2)
OH
Calcination at high temperatures is then performed in order to convert the pseudo-boehmite to y-alumina.6 In this study a series of Pt-SdAlzO3 catalysts were prepared using the sol-gel method. The structure of these catalysts was studied by the physical adsorption of Nz a t liquid nitrogen temperature to determine the total surface area and pore size distribution, and by chemisorption of hydrogen at room temperature to obtain the Wpt values. Reactions between n-hexane and hydrogen were carried out a t 420 "C in order to determine the catalytic activity and also to study deactivationtrends. Catalytic selectivity was obtained a t a total n-hexane conversion of approximately 10%for the various catalysts. In this study it was hoped to learn whether the sol-gel method can be used successfully to prepare bimetallic catalysts. Once the nature of these catalysts was determined, it was hoped that future efforts could be focused toward "tailor-making" bimetallic catalysts with desirable properties using the versatility of the sol-gel method. A detailed study was previously performed on the preparation of monometallic WAlumina catalysts by the sol-gel method.' This study, on the other hand, extends the method into the realm of bimetallic catalyst preparation.
Experimental Section (a) Catalyst Preparation. The method used for catalyst preparation was adapted from the work of h o r et al.8 In the typical preparation of a Pt-Snlalumina sample (in this case 1.0 Pt-0.9 Snlalumina), 100 cm3 of 2-butanol was added to 49.8 g ofATBfollowed by thorough mixing. A second solution consisting of 0.26 gof H2PtClg.xH20 dissolved in 5 cm3 ofwarm acetone was prepared. A third solution consisting of 0.27 g of SnClc5H20 dissolved in 5 cm3of warm acetone was prepared. This solution was added to the solution containing the Pt salt followed by thorough mixing. This combined Pt and Sn salt solution was added to the ATB-alcohol mixture and the solution was stirred thoroughly. In another beaker 10.6 g of HzO was taken, 0.6 mL of HC1 was added to it, followed by the addition of 100 cm3 of methanol. The mixture was then stirred thoroughly. This solution was then added to the beaker containing the mixture of ATB and the metal salts. The resultant mixture was stirred at room temperature for 10 min in a rotavapor. It was then heated to 45 "C and maintained a t that temperature with continual stirring for 10min. The temperature was then reduced to 40 "C and stirring was continued for a further period of 1h. Solvent removal was performed using the following schedule. The procedure consisted in opening one end of the rotavapor system to the atmosphere while the sample was maintained a t 60 "Cfor 20 h under continuous stirring. The temperature was then raised to 80 "C and maintained a t that temperature for 1 h. Finally, the temperature was increased to 100 "C and the solvent allowed to evaporate over a 30-min period. The sample was then removed from the rotavapor system and dried in an oven a t 120 "C for 24 h. (6)Dwivedi, R.K.;Gowda, G. J.Muter. Sci. Lett. 1985,4,331. (7)Balakrishnan, R;Gonzalez, R. D. J. Catal. lSBS,144,395. (8)Armor, J.N.; Carlson, E. J.;Zambri, P. M.AppZ. Cutal. 1986,19, 339.
0743-7463l9412410-2487$04.50/0 0 1994 American Chemical Society
2488 Langmuir,Vol. 10,No. 7, 1994
Notes
Table 1. Catalyst Composition catalyst blank A1203 1.0 SdAlzO3 1.0 Pt/d203 1.0 Pt-0.3 sn/A1zo3 1.0Pt-0.9 Sn/Al203 1.0 Pt-1.5 Sn/AlzOa
Pt loading (designed w t %)
Sn loading (designed w t %)
Sn loading (actual wt I, ICP) 0.43
1.0 1.0 1.0 1.0 1.0
0.63 0.63 0.62 0.62
0.3 0.9
1.5
Table 2. Adsorption Behavior of Catalysts catalyst
Pt loading (actual wt I, ICP)
0.21 0.51 0.84
24 1
BET pore avpore wpt area volume (mz/g) (cm3/g) diameter (nm) ratio 1.89 5,8 2.03 598 5,g 0.70 1.99 1.84 5,9 0.57 599 0.30 2.01 4,8 0.21 2.21
The dried samples were pretreated in a flow system using the following sequence of steps. The sample was heated in flowing 5%0 2 in He to 500 "C. Heatingin this gas mixture was continued for 2 h a t 500 "C. The 5% 0 2 in He was then replaced by a pure He flow for 15 min followed by reduction in H2 at 500 "C for 3 h. The Hz was then replacedby He and the temperature increased to 520 "C. After a final treatment in He a t 500 "C for 1h, the sample was cooled t o room temperature in flowing He. The pretreatment was necessary in order to convert the dry alumina gel to active alumina and also to reduce the metallic components. Portions of the pretreated samples were sent for ICP analysis in order to determine the actual Pt and Sn loading. In this paper, catalyst loading reported as percent refer to designed metal loading (wt %). Actual metal loadings as determined by ICP are given in Table 1. The discrepancy between the designed and actual metal loadings could be due to (a) the exposure of the metal precursors to air during catalyst preparation and the likelihood of the uptake of moisture from air and (b) the loss of the metal precursors while transferring material from one container to another during catalyst preparation. (b)Adsorption Measurements. The total surface area and pore size measurements were performed on a Coulter Omnisorb instrument. Pulsed hydrogen chemisorption was used to determine the amounta of hydrogen adsorbedby the sample. Details of the experimental protocol have been described e l ~ e w h e r e . ~ (c) n-Hexane Reaction Studies. The reaction between n-hexane and hydrogen was performed over the catalysts in a continuous flow reactor system using about 0.68 g of a previously pretreated sample. Deactivation studies were performed at a reaction temperature of420 "C by allowing the reaction to proceed and collecting data points following various times on stream. The catalytic selectivity toward dehydrocyclization,isomerization, and hydrogenolysis products were obtained at a total n-hexane conversion of -10%. Further details of the reaction system have been provided e l ~ e w h e r e . ~
Results and Discussion (a) Surface Area Measurements. Surface areas of the catalysts were determined using (1) the physical adsorption of N2 a t its saturation temperature to give the total surface area, pore volume, and pore size distribution of the sample and (2)by room temperature chemisorption of H2 to obtain the Wpt ratio. The BET equation was used for surface area calculations. Nitrogen desorption isotherms were used to calculate mesopore size distribution by appIying a method based on the BJH (Barrett, Joyner, and Hallender) methodug The adsorption behavior of the catalysts is shown in Table 2. The BET surface area of the blank alumina was observed to decrease from 1008m2/gprior to pretreatment (9) Product Manual for Coulter Omnisorp 1001360 Series, Coulter Corporation, 1991.
20
I
Pore dinmiter rum)
Figure 1. Effect of platinum and tin addition on the pore size distribution of the catalysts: (a) blank &03; (b) 1.0 Pt-1.5
sd&o3.
to 503 m2/g after pretreatment. The BET surface areas of all the pretreated samples were in the 430-545 m2/g ran'ge. The totalpore volumeswere observed to be between 1.8 and 2.2 cm3/g. The average pore diameters were also observed to be similar for the various samples studied indicating that the addition of Pt and Sn does not have a significant effect on the nature ofpores in the catalysts. Figure 1 shows the pore size distribution for a typical bimetallic sample, and it is clear that the distribution looks similar to that of blank alumina. Pulse hydrogen chemisorption at room temperature was used to determine the Wpt ratio. The blank A1203 sample and the monometallic 1.0 SdAl203 sample show insignificant hydrogen chemisorption at room temperature as compared to the monometallic 1.0 WAl203 sample. The amount of adsorbed hydrogen can be expressed by the H/Pt ratio.1° It is assumed that in the case of the PtSdAI203bimetallic catalysts, hydrogen chemisorption at room temperature can be used as a probe to determine the amount of hydrogen adsorbedby Pt. Achemisorption stoichiometry of one hydrogen atom chemisorbed per surface Pt atom can be used to determine the Pt dispersion for the monometallic WAl203 sample, but in the case of the bimetallic catalysts complications may arise due to the possibility of interactions between Pt and Sn. It is clear from Table 2 that there is a decrease in the HPt ratio as the tin content in the catalyst is increased. The H/Pt ratio ofthe 1.0FWAl2O3sample was quite high (0.70) but dropped to 0.21 for the 1.0 Pt-1.5 S d A l 2 0 3 sample. Similar observations regarding the decrease in the adsorption of hydrogen on Pt by the addition of Sn have been made before.lOJ1A likely reason for this could be chemical modification of the Pt by Sn.l0 Another explanation could be the deposition of Sn on top of Pt. The Wpt ratio could also show a similar decrease if there was any significant increase in the Pt particle size with (lO)Volter,J.;Leitz,G.;Uhlemann,M.;Hermann,M.J.CatuZ. 1981, 68,42.
(11)Lieske, H.; Volter, J. J . Catul. 1984,90,96.
Notes
Langmuir, Vol. 10,No. 7, 1994 2489
Table 3. Catalytic Selectivity for the Reactions of n-Hexane at 420 "C after 0.5 h on Stream selectivity (%) conversion catalyst
(%)
1.0Pt/A1203 1.0 F't-0.3 SdAI203 1.0 F't-0.9 sd&o3 1.0Pt-1.5 Sd.&O3
84.3 68.5 57.7 34.3
0
1
2
47.7 35.3 14.5 4.1
3
______
dehydroisomhydrocyclization erization genolysis
4
20.9 44.1 71.2 88.3
5
31.4 20.6 14.3 7.6
6
7
Time on stream (hrs)
Figure 2. Change in total n-hexane conversion as a function of time on stream for the various catalysts at 420 "C:(a) 1.0 pt/A1203; (b) 1.0 Pt-0.3 sd&o3; (c) 1.0 Pt-0.9 Sn/A1&; (d) 1.0Pt-1.5 Sn/A1203.
increasing tin content in the catalyst. But it is unlikely that there is any significant increase in Pt particle size with increasing tin content based on the results ofprevious studies using XRD12 and electron micro~copy.'~ (b)CatalyticReactivity. The reaction data analysis was performed using the method recommended by Ponec and Sa~ht1er.l~ The reaction products were categorized as dehydrocyclization products (methylcyclopentaneand benzene), as isomerization products (2,2-dimethylbutane, 2-methylpentane, and 3-methylpentane), or as hydrogenolysis products (methane, ethane, propane, isobutane, butane, 2-methylbutane and n-pentane). The reactions were also studied over a blank alumina xerogel sample and a monometallic 1.0 SdAl203 sampleand the reactivity in these cases was found to be insignificant in comparison to the catalysts containing Pt. Reactions were performed at 420 "C in order to compare the catalytic activity (Table 3) and also to study the deactivation of the samples as a function of time on stream (Figure 2). Catalytic selectivity values obtained at total n-hexaneconversion values -10% are given in Table 4. (12)Berndt, H.; Mehner, H.; Volter, J.; Meisel, W. Z. Anorg. Allg. Chem. 1977,429,47. (13) Zaikovskii, V. I.; Kovalchuk,V. I.; Ryndin, Yu. A.; Plyasova, L. M.; Kuznetsov, B. N.; Yermakov, Y. I. React. Kinet. Catal. Lett. 1980, 14, 99.
(14) Ponec, V.;Sachtler, W. M. H. Inproceedingsof5thZnternationul Congress on Catalysis, Miami Beach, Flu; Hightower, J. W., Ed.; American Elsevier Company, Inc.: New York, 1973; Vol. 1, p 645.
The deactivation studies clearly show that in the case of the monometallic 1.0 WAlzO3 sample there is some deactivation, especially during the initial part of the reaction run at 420 "C. As tin is added it is evident that the amount of deactivation decreases and the 1.0 Pt-0.9 SdAl2O3 and 1.0Pt-1.5 SdAlzO3 samplesshow negligible deactivation. This activity maintenance is one of the previously documented advantages of adding a second metal such as Sn.16 It is clear that as the amount of tin in the sample was increased, there was a decrease in the total n-hexane conversion. One reason for this could be a decrease in the amount of surface Pt atoms in the catalysts with increasing tin content. Another reason could be the possibility of an interaction between the two metals when dispersed on the alumina support. A second observation is that as the amount of tin in the sample is increased, a corresponding decrease in hydrogenolysis is observed (Table 4). This is one of the beneficial effects of tin and has been observed before. Because dehydrocyclization and hydrogenolysis selectivity occur solely on the Pt metal, there is a decrease in their catalytic activity as the amount of tin in the samplesincreases. This is due to a decrease in the amount of exposed Pt as the tin content of the catalyst is increased. The isomerization selectivity on the other hand continues to increase until in the case of the 1.0 Pt-1.5 SdAl2O3 sample it reaches a value of 96.3%. This means that apparently hydrogenation-dehydrogenation does not decrease substantially. The effect of the second metal, tin, appears to be similar in many respects to the effect of a catalyst poison like sulfur. A previous study on the effect of sulfur on the reforming reactions of n-heptane over a W C catalyst had shown that dehydrocyclization can occur on the metal and does not need acid centers.16 It was also determined that if 100ppm of sulfur was added to the feed containing hydrogen and n-heptane, then the amount of cyclization was reduced to a negligibly small value but the dehydrogenation capacity of Pt was not affected.16 It is interesting to compare our results on sol-gel PtSdAl2O3 catalysts to a previous study performed on catalysts prepared by traditional methods of coimpregnation.17 In an observation similar to ours, a very high isomerization selectivitywas observed for a catalyst having a high Sn 10ading.l~It was also found that in the case of traditionally prepared samples, there was an increase in the isomerization selectivity and a decrease in the hydrogenolysis selectivity with increasing amounts of tin in the ~amp1e.l~ However, there is one important difference between the study on traditionally prepared catalysts and our work on catalysts prepared by the sol-gel method. In the case of traditionally prepared samples it was found that when the designed Pt loading was kept constant at 1wt % and the amount of tin was increased gradually, then for small amounts of tin addition (up to a designed loading of 0.5 wt %) the isomerization selectivity did not increase by a very large amount.17 When the designed loading of tin was increased beyond 0.5 wt % to 1.0 wt %, there was a large increase in the isomerization selectivity, from 39.6% to 95.5%.17 On the other hand in the case of the sol-gel samples a more gradual increase in the isomerization selectivity was observed (Table 4). A possible reason for this could be that in the case of the traditionally prepared samples, at low loadings of tin, the (15) Sachdev, A.; Schwank, J. In Proceedings of 9th International Congress on Catalysis, Calgary, Canada; Phillips, M. J., Teman, M., Eds.; Chemical Institute of Canada: Ottawa, 1988; Vol. 3, p. 1275. (16)Silvestri, A. J.; Naro, P. A.; Smith, R. L.J. Catal. 1969,14,386. (17) Sachdev, A., Ph.D. Thesis, University of Michigan, Ann Arbor, MI, 1989.
Notes
2490 Lungmuir, Vol. 10, No. 7, 1994
Table 4. Catalytic Seledvity at a total of n-Herane Convefsion -1W selectivity (%) catalyst 1.0 wAl2os
temperature ("C) 300
conversion (8) 12.2 11.4 9.1 12.6
platinum and the tin are segregated and hence the isomerization selectivity does not change very much. For larger amounts of tin, the tin is deposited closer to thet'F and has a dramatic influence on the isomerization selectivity. On the other hand, in the sol-gel samples the Sn and R interact more strongly at small tin loadings and hence the effect on the isomerization selectivity is more gradual. A possible reason for this behavior could be that because in the sol-gel method the metal precursors are mixed with the alumina precursor before the formation of active alumina, there could be more intimate contact between the Pt and Sn precursors. Another explanation could be that the interactions between the Sn and alumina may be different in the case of samples prepared by the sol-gel method as compared to the samples prepared by more traditional methods, and this could have an effect on how the R and Sn interact. Based on this preliminary study it looks as ifthe sol-gel method offers a much closer interaction between the metallic components at low loadings. This property can be used in the future to prepare better bimetallic catalysts. Of course, one cannot neglect the possibility that tin could also modify the support acidity which might result in higher isomerization selectivity and lower cracking selectivity.18 In conclusion, this work indicates that the sol-gel
dehydrocyclization
isomerization
hydrogenolysis
6.3 7.6 2.4 2.0
46.2 62.7 95.2 96.3
47.5 29.7 2.4 1.7
method offers an interesting and potentially useful method of preparing supported metallic and bimetallic samples. Future directions of research using this method should focus on studying the effects of other metals such as Re and Ir. Detailed reactivity studies should be performed using various hydrocarbons in order to find out whether the sol-gel method could be used to prepare samples whose selectivity are more desirable than those obtained with samples prepared by more traditional methods of preparation. We have already shown that the sol-gel method can be used to prepare various types of monometal@ WAl203 catalysts by changing the preparative variables such as water/alkoxide molar ratio and the type of solvent removal scheme employed.' This versatility of the sol-gel method can also be used to prepare different types of bimetallic catalysts which may demonstrate different properties due to subtle changes in the nature of metal-metal and metal-support interactions.
Acknowledgment. The authors acknowledgesupport from the U.S.Department of Energy (Grant DOE FG0286ER-1351) for this research. (18)Burch, R.;Garla, L.C. J. Catal. 1981, 71, 360.