31P NMR Investigation of Surface Acidity Using Adsorbed

Acid Sites on SiO2−Al2O3 Monolayer Catalysts:P NMR Probes of Strength and Accessibility. Bing Hu and Ian D. Gay. The Journal of Physical Chemistry B...
0 downloads 0 Views 349KB Size
Langmuir 1995,11, 3845-3847

3845

31PNMR Investigation of Surface Acidity Using Adsorbed Tricyclohexylphosphine as a Probe Bing Hu and Ian D. Gay* Department of Chemistry, Simon Fraser University, Burnaby, British Columbia, V5A lS6, Canada Received October 7,1994. I n Final Form: July 5, 1995@

A new probe molecule (tricyclohexylphosphine) has been used to study acidity on silica, alumina, and silica-alumina catalyst surfaces. It was found that tricyclohexylphosphine (PCy3) rotates on the Si02 surface a t room temperature, and the motion is nearly isotropic in the first adsorbed layer. Larger amounts of PCy3 on silica are solidlike. Lewis sites were detected on alumina, but not on a silica-alumina catalyst. This is probably due to steric effects. On the silica-alumina catalyst, the concentrations of Br~nstedacid sites were determined. These appeared to increase on raising the adsorption temperature. At 200 "C, the maximum concentration of PCy3H+ found is 0.26 pmol/m2.

Introduction The surface acidity of silica-alumina catalysts is required for cracking, isomerization, dehydration, and polymerization reacti0ns.l Thus it is important to count the acid sites on the catalyst surface. Solid state 31PMAS NMR has been successfully used to study adsorption sites on the surfaces of silica-alumina and zeolite catalyst^.^-^ Trimethylphosphine (TMP) has been used as an NMR probe for the titration of acid ~ i t e s . ~Baltusis a et al.4>5 also used triethylphosphine (TEP), tri-n-butylphosphine (TBP), and phosphine oxides as NMR probes. While studies of small phosphine probes by 31PNMR on oxides of aluminum and silicon have yielded information about the types of surface acid sites and the numbers of these sites, the apparent acid concentration is dependent upon the choice of probe molecules. Saturation Bransted acid concentrations were 0.23, 0.19, and 0.15 mmol H+ per gram of silica-alumina for TMP, TEP, and TBP, respect i ~ e l y .This ~ may indicate that (a) some of the Bransted sites may be accessible only to small phosphines, (b) Bransted sites may be clustered, or (c) different base strengths may have an effect. Since many catalytic reactions involve large molecules, it is interesting to study the surface sites by still larger phosphine probes. Bulky probes will provide more information about whether steric factors are important. Silica-alumina catalyst is active for many reactions, some of which involve bulky starting reagents, such as the cracking of petroleum.'j In this study, tricyclohexylphosphine (PCy3) has been chosen as a probe to study surface acidity. PCya is interesting for the following reasons: first, it is a very strong base, in solution pKa = 9.70 (compared to TMP pK, = 8.757);secondly, it has a bulky structure,* with a cone angle of 170°e7 Here, we report Abstract published in Advance A C S Abstracts, September 15, 1995. @

__

(1)Tanabe, K. Solid Acids and Bases, Academic Press: New York,

19717 ._ . (2) Lundsford, J. H.; Rothwell, W. P.; Shen, W. X. J . Am. Chem. SOC. 1986,107, 1540. (3) Sheng, T. C.;Gay, I. D. J . Catal. 1994,145, 10. (4) Baltusis, L.;Frye, J. S.; Maciel, G. E. J.Am. Chem. SOC.1986, 108, 7119. (5) Baltusis, L.;Frye, J. S.; Maciel, G . E. J . Am. Chem. SOC.1987,

109, 40. (6) Tanabe, K.;Misono, M.; Ono,Y.; Hattori, H. New SolidAcidsand Bases; Elsevier: New York, 1989. (7) Golovin, N.; Rahman, M. M.; Belmonte, J. E.; Giering, W. D. Organometallics 1985,4 , 1981. (8) Davies, J. A,; Dutremez, S.; Pinkerton, A. A. Inorg. Chem. 1991, 30,2380.

0743-7463/95/2411-3845$09.00/0

the titration of Bransted acid concentrations on the silicaalumina catalyst surface by the PCy3 probe using different adsorption temperatures.

Experimental Section The silica-alumina cracking catalyst was Davison type 980, 25%Al203,20-40 mesh. This material had a surface area of 400 m2/g. The pore size distribution, determjned by N2 adsorption, showed a maximum at a radius of 17.5 A. All pores are larger than 15 A, and 90% of the pore volume is in pores smaller than 50 A. For distinguishingdifferent specieson the catalyst surface, the following model samples were used: Davison high purity silica gel, grade 923,100-200 mesh, 484 m2/g;Harshaw y - A l 2 0 3 grade Al-3945 E, 40-60 mesh, 225 m2/g; 12 M HCl from Anachemia, AC-4955, UN-1789. Tricyclohexylphosphine was from Strem Chemicals, 97%, Grade 15-6150. The samples were prepared on a vacuum line using a Y-shaped tube. One leg of the Y was sealed to a 5 mm 0.d. by 5.5 cm long NMR tube, the other to a 2 cm 0.d. bulb; the open end was connected to the vacuum line. The oxides and phospines were put into the two ends of the tube in a glovebag under a N2 atmosphere with the phosphine in the NMR tube. The tube was then quickly connected to the vacuum line. The oxides were Torr vacuum in the bulb while treated at 450 "C for 2 h in the phosphine remained in vacuum at room temperature. Then the oxides were cooled to room temperature and the open end of the tube was sealed under vacuum. Finally the oxides and the phosphines were mixed under vacuum and sealed in the NMR tube. NMJi tubes containing mixtures oftricyclohexylphosphine and oxides were heated at 100 "C for 1 h. After heating, 31P NMR measurements were carried out at room temperature. The samples were then heated to 150 and 200 "C, followed in each case by NMR measurements at room temperature. Higher temperatures were not used, because PCy3 was found to react above 200 "C. The vapor pressure of PCy3 is not available in the literature. As a comparison, triphenylphosphine has a vapor pressure of 5 Torr at 200 "C and 0.6 Torr at 150 "C.9J0 31PMAS NMR measurements were carried out on a homemade instrument operating at 60.5 MHz for 31Pusing a previously described spinner.ll The chemical shifts are reported relative to external 85% H3P04. Quantitative intensity measurements were made using 90" pulse excitation. In all cases, protons were decoupled with a 45 kHz field. The total amount of phosphine in each sample was determined by weight measurements when the samples were prepared. The amount of phosphorus in the Brgnsted acid peak of the spectrum was determined from relative areas in the integrated spectra. In order to ensure that the repolarization time (interval between successive NMR scans) (9) Forward, M. V.;Bowden, S. T.; Jones, W. J. J . Chem. SOC.1949, 3,s121. (10)Grigorev, A.A.; Konchractev, Y. V.; Siworov, A. V. Z h . Obsch. Kh. 1984,54,1935. (11)Gay, I. D. J.Magn. Reson. 1984,58, 413.

0 1995 American Chemical Society

3846 Langmuir, Vol. 11, No. 10, 1995

Hu and Gay

100

0

50

0 ppm

-100

ppm

--

-50

Figure 1. 31PMAS NMR spectra of pure PCy3 and PCy3 on silica gel: (a) pure PCy3, spinning rate = 1.89 kHz; (b) 0.38 ,umollm2PCy3 on silica, 100 "C for 1 h, spinning rate = 1.89 kHz; (c) 0.38 ,umol/m2PCy3 on silica, 150 "C for 1h, spinning rate = 1.82 kHz; (d) 0.38 ymollm2 PCy3 on silica, 200 "C for 1 h, spinning rate = 1.82 kHz; (e) 1.68 pmollm2 PCy3 on silica, 100 "C for 1h, spinning rate = 1.80 kHz; ( f ) 1.87 ,umollm2PCy3 on silica, 100 "C for 1h, spinning rate = 1.80 kHz, and (g)6.68 ,umollm2PCy3 on silica, 100 "C for 1h, spinnning rate = 1.35 kHz.

was sufficiently long, repeat spectra were taken with at least a 2-fold change in repolarization time, to verify that peak ratios did not change significantly.

Results and Discussion Figure l a shows the spectrum ofpure solid PCy3 a t 1.89 kHz MAS speed. The center band resonance position, 7.5 ppm, is close to literature values of 7.5 and 7.0 ppm,12J3 but different from 9.28 ppm reported in ref 8. Parts b, c, and d of Figure 1are the spectra of 0.38 pmollm2 PCy3 on silica gel after treatment a t 100, 150, and 200 "C respectively. We refer to the species present as "physically adsorbed. The absence of spinning side bands suggests that the apparent chemical shift anisotropies are greatly reduced by rotation of the PCy3 molecules on the surface. From the static spectrum of pure PCy3 a t room temperature, we find u11= -30 ppm, u22 = 18 ppm, and u33 = 35 ppm in agreement with ref 13. The side band intensities in Figure l a are in agreement with calculated i n t e n ~ i t i e s ' ~ based on these principal values. I t should be noted that the rotation of PCy3 on Si02 must be nearly isotropic and not simply a rotation about the %fold axis. In the latter case, spinning sidebands corresponding to an averaged axially symmetric shielding tensor would be observed. Figure le-g shows the spectra of PCy3 at higher coverages on silica after treatment a t 100 "C for 1h. The spectra were recorded several times during 3 days following the adsorption until they did not change. Figure l e with coverage 1.68 pmol/m2 is similar to Figure l b . PCy3 appears isotropic due to the motion on the silica gel. When the coverage was increased to 1.87 pmol/m2 (10,a new peak with spinning side bands appeared at 7.5 ppm. This is the same shift as pure PCy3 and is obviously different from the peaks of submonolayer coverages (le). 1.8pmoYm2is on the order ofone monolayer, based on the dimensions of the PCy3 molecule.* When the coverage is higher than one monolayer, a nonrotating species is clearly present which might arise either from multilayer adsorption or from nonadsorbed crystalline PCy3. Coverage for Figure l g exceeds three statistical layers. (12) Bemi, L.; Clark, H. C.; Davis, J. A,; Fyfe, C. A,; Wasylishen, R. E. J . A m . Chem. SOC.1982,104, 438. (13) Penner, G. H.; Wasylishen, R. E. Can. J . Chem. 1989,67,1909. (14) Herzfeld, J.;Berger, A. E. J . Chem. Phys. 1980,73,6021.

Figure 2. 31PMAS NMR spectra of 0.31 ymollm2 PCy3 on alumina: (a) 100 "C for 1h, spinning rate = 1.89 kHz; (b) 150 "C for 1h, spinningrate = 1.87kHz; (c) 200 "C for 1h, spinning rate = 1.87 kHz, and (d) 200 "C for 1h, spinning rate = 2.42 kHz. The small peaks around 60-70 ppm in c and d arise from Cy3P=O.

100

0

-100

ppm

Figure 3. 31PMAS NMR spectra of 0.30 ,umollm2 PCy3 on silica-alumina catalyst: (a) 100 "C for 1h, (b) 150 "C for 1h, and (c) 200 "C for 1 h.

Figure 2 shows the spectra of 0.31 pmollm2 PCy3 on alumina, with different adsorption temperatures. Figure 2a is the spectrum after 100 "C adsorption. The peak around 7 ppm arises from the "physically adsorbed" species. The species changed to more stable ones after further heating a t 150 "C (Figure 2b) and 200 "C (Figure 2c,d). From ref 15, the chemical shift of the M e d P C y 3 complex is -4 ppm. So the peak around -7 ppm is probably due to Lewis-bound species. Thus in spite of its large size, PCy3 can access Lewis sites on Al2O3, which other workers16J7have found to be inaccessible to 2,6dimethylpyridine. It is clear from the spinning side bands (Figure 2c,d)that the broad peak around -7 ppm actually consists of a t least two peaks: one is around -7 ppm and the other is around -22 ppm and gives rise to the spinning side bands. The multiple peaks probably result from interaction of PCy3 with different types of Lewis site. Figure 3 shows the spectra of PCy3 at 0.30 pmoYm2 coverage adsorbed on commercial Si02-A1203 cracking catalyst after treatment at 100, 150, and 200 "C for 1h. The total phosphorus detected in each spectrum is constant within the experiment error, independent of heat treatment. We find that protonated PCy3 in concentrated HC1 resonates a t 31.3 ppm, close to 32.7 ppm reported in fluorosulfuric acid.18 So around 31 ppm are the peaks arising from B r ~ n s t e dacid species. The results of Figure 3a-c indicate that the higher the treatment temperature, the more protonated phosphine appears. The resonances of physically adsorbed species were a t 9.6, 8.9, and 8.3 ppm, respectively, which are similar to the shifts in Figure 1on silica gel. No peaks were found from -4 to -20 ppm. So the Lewis sites of the silica-alumina catalyst, which (15)Barron, A. R. J . Chem. SOC.Dalton Trans. 1988,3047. (16) Gay, I. D.; Liang, S. H. J . Catal. 1976,44, 306. (17) Benesi, H. A. J . Catal. 1973,28,176. (18)Olah, G. A,; McFarland, C. W. J . Org. Chem. 1969,34, 1832.

31PNMR Investigation of Surface Acidity

Langmuir, Vol. 11, No. 10, 1995 3847 20.1

d 1

ppm

c 0

1.50

3.00

0

PQ3

pmol/m2

Figure 4. The concentrations of PCysH+on commercial silicaalumina at various coverages and adsorption temperatures: (a) 100 "C, (b) 150 "C, and (c) 200 "C. are detected3byPMe3, are not detected by the PCy3 probe. These Lewis sites must therefore be more hindered than those found on pure A l 2 0 3 . Figure 4 shows the concentrations of PCy3H+ on commercial SiO2-Al2O3 cracking catalyst at various total phosphine coverages, for different adsorption temperatures. To measure total protonated phosphine, the fraction of the total area in the PCy3H+ peak is multiplied by the total adsorbed amount. Due to the uncertainties of integration, the concentration uncertainty is about f 1 0 15%. I t can be seen t h a t the PCy3H+concentration tends to saturate with increasing phosphine coverage. From Figure 4 the maximum concentration of PCysH+ is around 0.26 pmoYm2. This is higher than the value of 0.21 pmol/m2 found with the trimethylphosphine (TMP) probe on the same c a t a l y ~ t . ~The higher apparent Bransted acid concentration may occur for one of the following reasons: (a)since PCy3 is a stronger base than TMP,7 additional reaction with weaker sites may occur, or (b) since TMP was studied only by room temperature adsorption, there might be sites only accessible by hightemperature diffusion. To test theory b, we heated several of the samples used in ref 3 to 100 "C and found no change in PCy3H+ concentration. This result favors explanation a. Figure 4 also indicates t h a t apparent B r ~ n s t e dacid concentrations are increased by higher adsorption tem-

Figure 5. The line widths of PCy3 on silica-alumina catalyst as a function of surface coverages: (a) physisorbed peak, 150 "C; (b)physisorbed peak, 200 "C; (c)Brgnsted peak, 200 "C; and (d) Brgnsted acid peak, 150 "C. peratures. This is probably due to either (a) activated diffision or (b)activated proton transfer. The later might occur if the base cannot approach H+ closely enough on the surface. The pore size measurements on this catalyst suggest that the pores are not too small to admit PCy3, but this does not preclude activated diffusion in the adsorbed layer. A similar activated process for NH3 on the A 1 2 0 3 surface has been observed.lg The line widths of adsorbed species on silica-alumina as a function of surface coverage are summarized in Figure 5. We see that the B r ~ n s t e dacid peak is much broader than that of the physically adsorbed species. Chemical shift variations among slightly differing B r ~ n s t e dstates may account for this. It can be seen that different adsorption temperatures give essentially the same line width, probably indicating that the motional state has not changed.

Acknowledgment. This work was supported by grants from the Natural Sciences and Engineering Research Council of Canada and from Imperial Oil Ltd. We thank Dr. Bogdan Czajka for arranging measurement of the pore size distribution on the SiO2-Al203 catalyst. LA940785K (19)Medema, J.;VanBokhoven, J.J.G . M.; Kuiper, A. E.T.J.Catal. 1972,25, 238.