EFFECT OF FLUORIDE ON T-ALUMINA AND SILICA-ALUMINA
2917
The Effect of Fluoride on Surface “Acid” Sites on y- Alumina
and Silica-Alumina1 by J. B. Peri Research and Development Department, American Oil Company, Whiting, Indiana
46994
(Received February 14, 1968)
Infrared studies of adsorbed Cog and CO indicate that fluoride strengthens certain strong Lewis “acid” sites ( a sites) on y-alumina but either leaves them unchanged or eliminates them on silica-alumina. Both COz and CO are held more strongly and exhibit higher frequencies on fluoride-modified a sites than on normal a sites on alumina. Replacement of hydroxyl groups by fluoride removes Brplnsted acid sites and can indirectly eliminate “strained” M-O-M linkages. The activity for polymerization of butene was high on catalysts holding few or no hydroxyl groups, suggesting that Brplnsted acids are not catalytically important in polymerization on either fluorided alumina or fluorided silica-alumina. Strengthening of CY sites appears responsible for the polymerization activity of fluorided alumina. Lowered polymerization activity of fluorided silicaalumina partly reflects elimination of a sites but also indicates loss of p sites, A1-O-M linkages which are presumably less strained but more active.
Introduction Fluoride treatment of y - a l ~ m i n a ~and - ~ silica-aluminab often enhances their activity for acid-catalyzed reactions. Moreover, in cracking hydrocarbons, fluoride-treated silica-alumina can also form substantially less “coke” than does untreated silica-alumina.6 Although fluoride treatment is usually thought either to inductively strengthen preexisting Lewis acid sitesa or to produce strong Brpinsted acid sites,’ recent studies2J indicate that fluoride actually weakens surface acids on y-alumina. Weaker sites might give higher activity because they allow higher mobility of reactant molecules on the surface and easier desorption of products. Even if most of the acid sites on alumina are weakened by fluoride treatment, some, possibly including those of greatest catalytic importance, could be strengthened. Infrared studies have shown that certain strong “acid” sites, called a sites, exist in low concentration on dry y-alumina and silica-alumina surfaces.9 Such sites selectively adsorb COZ,olefins, and other molecules and appear to be catalytically active for isomerization or polymeriza,tion of olefins. They contain reactive oxide ions adjoining incompletely coordinated aluminum ions in the surface. On y-alumina, their properties can be modified by reaction of the surface with HCl or other chlorides,l0 apparently as a result of substitution of chloride for hydroxyl or oxide ions contained in, or closely adjoining, the a sites. The modified Q sites on alumina resemble the a sites on silicaalumina, being more acidic” than the original sites. Fluoride might be expected, like chloride, to strengthen existing a sites on y - a l ~ m i n a ,but ~ no direct evidence has shown this. Neither is there any evidence on how fluoride affects a sites on silica-alumina. Lewis acid sites on unreduced nickel-silica catalyst11 and apparently on X- and Y-type zeolites12adsorb GO, ‘(
giving infrared bands near 2200 cm-’. Similar bands from CO adsorbed on drg7 alumhala and zinc oxide have, however, been attributed to adsorption on oxide ions. l 4 Whether the GO is held by acid sites or by oxide ions, changes in the nature and extent of GO adsorption after fluoride treatment might also reflect changes in catalytically important sites. Further study of-the effect of fluoride on the surface and catalytic properties of y-alumina and silica-alumina seemed warranted. Consequently, established infrared and gravimetric techniques were used to study adsorption of COZ and CO and adsorption and polymerization of butene. For comparison, silica was studied similarly.
Experimental Section
Most of the equipment used has been d e s ~ r i b e d . ~ ~ ~ ~ ~ (1) Presented at the 165th National Meeting of the American Chemical Society, San Francisco, Calif., April 1968. (2) V. C. F. Holm and A. Clark, Ind. Eng. Chem. Prod. Res. Develop., 2, 38 (1963). (3) H. R. Gerberich, F. E. Lutinski, and W. K. Hall, J . Catalysis, 6, 209 (1966). (4) R. Covini, V. Fattore, and N. Giordano, ibid., 7, 126 (1967). (5) C. J. Plank, D. J. Sibbett, and R. B. Smith, Ind. Eng. Chem., 49, 742 (1957). (6) A. N. Webb, ibid., 49, 261 (1957). (7) I. D. Chapman and M. L. Hair, J. Catalysis, 2, 145 (1963). (8) V. A. Chernov and T . V. Antipina, Kinetika i Kataliz, 7, 739 (1966). (9) J. B. Peri, J. Phys. Chem., 70, 3168 (1966). (10) J. B. Peri, ibid., 70, 1482 (1966). (11) J. B. Peri, Discussions Faraday Soc., 41, 121 (1966). (12) C. L. Angel1 and P. C. Schaffer, J . Phys. Chem., 70, 1413 (1966). (13) L. H. Little and C. H. Amberg, Can. J. Chem., 40, 1997 (1962). (14) C. H. Amberg and D. A. Seanor, “Third Congress on Catalysis,” Val. I, North-Holland Publishing Co., Amsterdam, 1965, p 450. (15) J. B. Peri, J . Phys. Chem., 70, 2937 (1966).
Volume 72, Number 8 August 1968
2918 I n some of the work with the Beckman Model IR-9 spectrometer a very simple cell was used in which the oxide sample could be heated in a Vycor section and then cooled and slid to a section with CaFz windows for spectroscopic study. This cell was removed from the spectrometer to heat the sample. In most of the work, however, the cells used with the IR-911 and with the Perkin-Elmer 12C (CeIl D16)were as before. The alumina, si1ica,l5 and silica-aluminag samples were mostly aerogel plates prepared as previously described. The silica-alumina aerogels were SAA-2 and SAA-3, the properties of which have been described;Q they contained 32.7 and 13.6% A1z03, respectively. One type of fluorided silica-alumina was prepared by impregnating Nalco HA silica-alumina9 (26% A1203) with an aqueous solution of NH4F. The dry catalyst, which contained 4 wt % fluoride, was pressed in a 1.25in. die at 12,000 1b/ine2to form a thin, self-supporting wafer suitable for infrared study. Attempts to make clear fluorided aerogel plates by treatment of wet gel plates with aqueous NH4F or H F were unsuccessful, as was treatment of dry aerogeI plates of alumina, silica, and silica-alumina with dry HF. Treatment with NH4F vaporized in a stream of dry nitrogen at 600" did, however, give clear uncracked plates. The typical procedure for NH4F treatment of aerogel was as follows. An aerogel plate was placed in a horizontal Vycor tube (1 X 24 in.) equipped with a male ground joint at one end and wrapped in two separate sections with nichrome ribbon. The plate was heated in flowing oxygen at 600" for 2 hr. After a short flush with Nz, a small boat holding KH4F was inserted in a cool section of the tube, upstream from the aerogel, which was then dried at 600" in flowing Nz(PzOs-dried) for 2 hr. The cool section was slowly heated to volatilize the fluoride, which passed, in the Nz stream, over the heated (600") aerogel. The Vycor tube was connected through the ground joint to the simple infrared cell, which was also flushed with Nz. After the sample had been slid into the cell, the Vycor tube was replaced with a short Vycor tube closed a t one end. Subsequent treatments were usually carried out in the cell. For removal of fluoride by hydrolysis, however, the aerogel was returned to the original Vycor tube. Samples studied in the regular IR-9 cell were fluorided as described but were exposed briefly to air while being transferred to the cell. Silica-alumina aerogel plates were also fluorided by treatment at 500-600" with fluorobenzene vapor. The procedure was generally as that described for KH4F treatment, but the J Szstream was saturated with fluorobenzene at room temperature. I n one instance the plate was suspended from the quartz helix in cell D and, after calcination in 0 2 at 600" and drying by evacuation a t goo", was heated in fluorobenzene vapor (
