Structural and Mechanistic Investigation of a Phosphate-Modified

Saifudin M. Abubakar, David M. Marcus, Jeffrey C. Lee, Justin O. Ehresmann,. Ching-Yeh Chen, Philip W. Kletnieks, Darryl R. Guenther, Miranda J. Hayma...
0 downloads 0 Views 250KB Size
Download PDF
4846

Langmuir 2006, 22, 4846-4852

Structural and Mechanistic Investigation of a Phosphate-Modified HZSM-5 Catalyst for Methanol Conversion Saifudin M. Abubakar, David M. Marcus, Jeffrey C. Lee, Justin O. Ehresmann, Ching-Yeh Chen, Philip W. Kletnieks, Darryl R. Guenther, Miranda J. Hayman, Mari Pavlova, John B. Nicholas, and James F. Haw* Department of Chemistry, UniVersity of Southern California, UniVersity Park, Los Angeles, California 90089-1661 ReceiVed December 19, 2005. In Final Form: February 23, 2006 Phosphorus modification of a HZSM-5 (MFI) zeolite by wet impregnation has long been known to decrease aromatic formation in methanol conversion chemistry. We prepared and studied a catalyst modified by introducing trimethylphosphine under reaction conditions followed by oxidation. Magic-angle spinning (MAS) NMR shows that extensive dealumination occurs, resulting in a catalyst with a much higher framework SiO2/Al2O3 ratio, as well as extraframework aluminum and approximately 1.4 equiv of entrained phosphoric acid (under working conditions) per aluminum. Upon dehydration or regeneration, the phosphoric acid is converted, reversibly, to entrained P4O10. The aromatic selectivity of the modified catalyst is significantly lower than that of an unmodified zeolite with a similar, increased framework SiO2/Al2O3 ratio. By comparing the rates of H/D exchange in propene under conditions similar to those in methanol conversion chemistry, we determined that the acid site strength is indistinguishable on modified and unmodified zeolites, and this is consistent with theoretical modeling. On the phosphorus-modified zeolite, the rate of propene oligomerization is greatly suppressed, suggesting that entrained phosphate is an impediment to sterically demanding reactions.

Introduction It has long been known that methanol reacts on zeolite HZSM-5 (MFI) to give a mixture of hydrocarbons rich in methylbenzenes, a reaction called methanol-to-hydrocarbon conversion.1-7 A number of reports have shown that the introduction of phosphorus into HZSM-5, usually by wet impregnation using any of several precursors, can provide, after calcination, a catalyst with significantly reduced aromatic selectivity.8-16 Such modified catalysts17 have been considered for use in methanol-to-olefin (MTO) conversion processes as well as hexane cracking and toluene alkylation.18 In earlier work, we modified the silicoaluminophosphate catalyst HSAPO-34 by ship-in-a-bottle synthesis of a tetramethylphosphonium cation from PH3 and methanol, followed by oxidation.19 After oxidation, this catalyst had a greatly reduced * To whom correspondence should be addressed. E-mail: [email protected]. (1) Chang, C. D.; Silvestri, A. J. J. Catal. 1977, 47, 249. (2) Chang, C. D. Catal. ReV.sSci. Eng. 1983, 25, 1. (3) Chang, C. D. Catal. ReV.sSci. Eng. 1984, 26 (3,4), 323. (4) Keil, F. J. Microporous Mesoporous Mater. 1999, 29, 49. (5) Stocker, M. Microporous Mesoporous Mater. 1999, 29, 3. (6) Chu, C. T.-W.; Chang, C. D. J. Catal. 1984, 86, 297. (7) Anderson, M. W.; Klinowski, J. J. Am. Chem. Soc. 1990, 112, 10. (8) Kaeding, W. W.; Butter; S. A. U.S. Patent 3 911 041, 1975. (9) Butter; S. A.; Kaeding, W. W. U.S. Patent 3 972 832, 1976. (10) Kaeding, W. W.; Butter, S. A. J. Catal. 1980, 61, 15. (11) Vedrine, J. C.; Auroux, A.; Dejaifve, P.; Ducarme, V.; Hoser, H.; Zhou, S. J. Catal. 1982, 73, 147. (12) Tynjala, P.; Pakkanen, T. T. Microporous Mesoporous Mater. 1998, 20, 363. (13) Tynjala, P.; Pakkanen, T. T.; Mustamaki, S. J. Phys. Chem. B. 1998, 102, 5280. (14) Rahman, A.; Adnot, A.; Lemay, G.; Kaliaguine, S.; Jean, G. Appl. Catal. 1989, 50, 131-147. (15) Rahman, A.; Lemay, G.; Adnot, A.; Kaliaguine, S. J. Catal. 1988, 112, 453. (16) Dehertog, W. J. H.; Froment, G. F. Appl. Catal. 1991, 71, 153. (17) Lischke, G.; Eckelt, R.; Jerschkewitz, H.-G.; Parlitz, B.; Schreier, E.; Storek, W.; Zibrowius, B.; Ohlmann, G. J. Catal. 1991, 132, 229. (18) Vinek, H.; Rumplmayr, G.; Lercher, J. A. J. Catal. 1989, 115, 291. (19) Song, W.; Haw, J. F. Angew. Chem., Int. Ed. 2003, 42, 892.

aromatic selectivity and hence a very high propene selectivity. Here, we report analogous work using HZSM-5 as a starting catalyst. Since HZSM-5 has a larger pore than HSAPO-34, we used the less toxic, less flammable trimethylphosphine as the modification reagent, and carried out a structural and catalytic study of this material to more fundamentally understand the effect of phosphorus modification. We note that trimethylphosphine, usually without oxidation, has previously been use to modify zeolites by titrating strong acid sites.20 X-ray powder diffraction revealed no bulk phases other than the zeolite. 27Al, 29Si, and 1H magic-angle spinning (MAS) NMR all showed extensive dealumination of the zeolite framework from an initial SiO2/Al2O3 ratio of 80 to a final value typically near 240. 31P MAS NMR showed that, in the presence of water and methanol, the phosphorus was present in the modified zeolite as phosphoric acid, but, upon dehydration or calcination, it was present as P4O10. When the modified catalyst was compared with an unmodified commercial zeolite with a similar, increased framework SiO2/ Al2O3 ratio, it was found that a reduced acid site density could account for only approximately one-half of the decrease in aromatic selectivity upon phosphorus modification. We considered the possibility that the acid strength of the modified catalyst was reduced, but the relative rates of propene H/D exchange showed no difference between our modified catalyst and an unmodified HZSM-5 with a similar acid site density. However, the rate of propene oligomerization was greatly suppressed on the phosphorus-modified catalyst. We speculate that sterically demanding reactions such as propene oligomerization and aromatization are suppressed on the modified catalyst because of the dynamic formation of bulky phosphate complexes of the Brønsted acid sites under the reaction conditions as well as extraframework aluminum species. In the former case, theoretical modeling suggested that the phosphoric acid complex (20) Rumplmayr, G.; Lercher, J. A. Zeolites 1990, 10, 283.

10.1021/la0534367 CCC: $33.50 © 2006 American Chemical Society Published on Web 04/05/2006

Methanol ConVersion of Phosphate-Modified HZSM-5

Langmuir, Vol. 22, No. 10, 2006 4847

of a framework Brønsted site would have an acid strength (deprotonation energy) almost identical to that of the uncomplexed acid site. Experimental Section Materials and Reagents. Zeolite HZSM-5 (MFI) samples (CBV 8014: SiO2/Al2O3 ) 80; CBV 28014: SiO2/Al2O3 ) 280) were purchased from Zeolyst International. Typically, zeolites were calcined at 873 K for 16 h prior to modification or use to remove the remaining templating agent and then pressed into 10-20 mesh pellets. Methanol (99.93%), methanol-13C, and trimethylphosphine (97%) were purchased from Aldrich. Zeolite Modification Procedures. Our novel method for phosphorus modification of HZSM-5 zeolite involved flowing 1 mL of an equimolar solution of trimethylphosphine and methanol at a weight-hourly space velocity (WHSV) (g feed/g catalyst/h) of 8 h-1 over 300 mg of HZSM-5 (SiO2/Al2O3 ) 80) at 773 K and 600 sccm helium. The zeolite was subsequently oxidized at 773 K under 500 sccm air for several minutes prior to raising the temperature to 823 K and holding for 1.5 h to afford the “modified catalyst”. Characterization. X-ray diffraction (XRD) patterns were acquired on a Rigaku Rotaflex RU-200 high-brilliance X-ray diffractometer with Cu K radiation. The X-ray source was an anode operating at 50 kV and 70 mA with a Cu target. XRD data were collected for 2θ between 5° and 50°. Elemental analysis was performed by Galbraith Laboratories, Inc. A Chemagnetics CMX-360 NMR spectrometer operating at 71.5 MHz for 29Si, 145.6 MHz for 31P, 93.7 MHz for 27Al, and 359.7 MHz for 1H was used for MAS NMR measurements. Tetramethylsilane (TMS) (0 ppm), NH4H2(PO4)3 (0.9 ppm), 1 M Al(NO3)3 (0 ppm), and TMS (0 ppm) were used as external chemical shift standards for 29Si, 31P, 27Al, and 1H, respectively. A Chemagneticsstyle pencil probe spun 7.5 mm zirconia rotors at 5.5-6.2 kHz with active spin speed control ((3 Hz). All spectra shown were obtained at room temperature using Bloch decay, except where otherwise stated. In most cases, NMR samples were prepared by transferring the catalyst from sealed reactors into sealed NMR rotors in a nitrogen glovebox. 29Si MAS spectra were acquired using 500 scans, a 90° flip angle, and 120 s pulse delays. 31P spectra were acquired by using 2000 scans, a 90° flip angle, and 3 s pulse delays. Samples for 27Al MAS NMR investigation were exposed to air for 3 h before transfer into the NMR rotor. 27Al MAS spectra were acquired using 10 000 scans, a 15° flip angle, and 3 s pulse delays. Samples for 1H MAS NMR were prepared in a shallow bed CAVERN apparatus.21 Typically, 300 mg of zeolite was loaded into the CAVERN, which was subsequently connected to a vacuum line. The catalyst was first heated to a final temperature of 673 K at a heating rate of 0.7 K/min. The catalyst was kept at the final temperature for 12 h under vacuum and then for 1 h at a final pressure of less than 5 × 10-5 Torr. The dried sample was loaded into a 7.5 mm zirconia rotor and capped in the CAVERN. 1H{27Al} spinecho double-resonance MAS NMR experiments were performed as described previously.22 Proton measurements were acquired using 32 scans, a 90° flip angle, and 10 s pulse delays. Each τ period was equal to one or two rotor periods in all experiments. Catalysis. We used a benchtop microreactor system for all experiments reported here. All metal components contacting the catalyst, reactants, or products were stainless steel. Helium was used as the carrier gas in all experiments. The quartz reactor (26 cm long, 0.70 cm internal diameter) was loaded with 300 mg of fresh catalyst for each experiment, which was activated in place immediately prior to use by heating at 773 K for 1 h under 600 sccm flowing helium. All tubing downstream of the chamber was heated. The experiments conducted here involved using the continuous introduction of alcohol (methanol, ethanol, or 2-propanol) using a syringe pump (Harvard Apparatus model PHD 2000). (21) Xu, T.; Haw, J. F. Top. Catal. 1997, 4, 109. (22) Beck, L. W.; White, J. L.; Haw, J. F. J. Am. Chem. Soc. 1994, 116, 9657.

Figure 1. GC-FID chromatograms showing product distributions from methanol conversion in plug flow reactors for various HZSM-5 catalysts after 90 min time of on stream. The conditions in every case included a reaction temperature of 673 K, a methanol WHSV (g feed/g catalyst/h) of 8 h-1 over a 300 mg catalyst bed, and 200 sccm helium carrier gas. (a) Using unmodified zeolite (SiO2/Al2O3 ) 80), the products were dominated by methylbenzenes and alkanes. (b) Using a catalyst prepared by the modification of HZSM-5 (SiO2/ Al2O3 ) 80) by flowing a solution of trimethylphosphine in methanol followed by oxidation (as described in the text), there was a significant suppression of aromatics and alkanes. The products were dominated by light olefins and, especially, propene. (c) Using an unmodified HZSM-5 with a lower framework acid site density (SiO2/Al2O3 ) 280), we observed a reduction in methylbenzene and alkane yields, but this shift to light olefin selectivity was only a fraction of that obtained by the phosphate modification of a zeolite with a higher acid site density (see text).

Catalytic product distributions were determined using a gas chromatograph (Agilent 6890 Series gas chromatography-mass spectrometry (GC-MS) system) equipped with both a flame ionization detector (FID) and a mass selective detector (Agilent 5973). The ionization voltage was 69.9 eV, and the source temperature was 553 K. Separations were performed on a 100 m Petrocol DH 150 fused silica capillary column (0.25 mm diameter, 1.0 µm film thickness). A temperature program maintained the oven temperature at 308 K for an initial 3 min, followed by a ramp of 20 K/min to a final temperature of 563 K.

Results Structural Characterization. Figure 1 reports representative GC-FID chromatograms comparing the product distributions from continuous methanol conversion at 673 K on the three HZSM-5 catalysts studied in this contribution. Selectivity data from these measurements are summarized in Table 1. All comparisons are made after 90 min of time on stream; in general, the catalysts showed stable activity and selectivity over a period of several hours of time on stream. The starting material for our “modified catalyst” (SiO2/Al2O3 ) 80) had a total yield of methylbenzenes of 24.7% based on moles of methanol (or carbon).

4848 Langmuir, Vol. 22, No. 10, 2006

Abubakar et al.

Table 1. Comparison for the Conversion and Selectivity of the MTO Reactions between Various HZSM-5 Catalysts at 673 K for 90 min methanol conversion (%) effective conversion (%) selectivity methane (%) C2s (%) C3s (%) C2s + C3s (%) C4s (%) C5s (%) C6s (%) C4 - C6s (%) benzene (%) toluene (%) xylenes (%) trimethylbenzenes (%) aromatics (%)

HZ-80a

PZ-80b

HZ-280c

100.0 100.0

84.3 83.3

90.5 90.0

4.1 14.8 25.4 40.2 17.0 9.3 4.7 31.0 0.2 4.5 12.6 7.4 24.7

0.0 7.3 48.0 55.3 22.3 14.9 7.0 44.2 0.0 0.3 0.2 0.0 0.5

3.2 12.8 39.0 51.8 20.6 9.6 2.6 32.8 0.5 2.9 7.2 1.6 12.2

a HZ-80 ) standard HZSM-5 (SiO2/Al2O3 ) 80). b PZ-80 ) trimethylphosphine and methanol modified HZSM-5 (SiO2/Al2O3 ) 80). c HZ-280 ) standard HZSM-5 (SiO2/Al2O3 ) 280).

The same figure of merit for propene was 25.4%. The propene/ ethylene mole ratio was 1.7:1. In contrast, our modified catalyst produced only 0.5% conversion to methylbenzenes, and the conversion of propene leaped to 48.0% (again, moles of carbon basis), while the propene/ ethylene mole ratio increased dramatically to 6.6:1. Alkanes were significantly suppressed on our modified catalyst, but C4 and C5 olefins were slightly elevated. The major emphasis of this contribution is to obtain a structural and mechanistic rationale for these large differences in selectivity. Since one of our structural findings (vide infra) is that our modification procedure increases the framework composition from SiO2/Al2O3 ) 80 to a value several times higher, we also studied a commercial sample with a framework composition of SiO2/Al2O3 ) 280 to control for the effects of acid site density in our comparisons. Briefly, Figure 1c shows that reducing the acid site density moves the catalyst selectivity in the directions obtained with the phosphorus modification procedure, but the yield of methylbenzenes is still appreciable, and other details, such as the propene/ethylene ratio, are not entirely accounted for by framework acid site density alone. The effects of temperature on the activity, selectivity, and lifetime of our modified catalyst were unremarkable. The GCFID chromatograms in Figure 2 were from experiments, all at 120 min, at various temperatures. Conversion at 648 K was only 67%, whereas an unmodified HZSM-5 catalyst bed would yield full conversion under identical conditions. The modified catalyst showed at least 85% conversion at 673 K and above. Aromatics and alkanes were suppressed over the entire temperature range studied, and, as is typical in MTO catalysis, ethylene selectivity increased slightly with temperature. The first question about the structure of our modified catalyst is whether bulk phases are formed in addition to the MFI lattice of normal HZSM-5 zeolite, as happens, for example, in the case of the “solid phosphoric acid” catalyst synthesized from silica and phosphoric acid using much higher temperatures.23 X-ray powder diffraction showed no additional phases present. We then considered the amount and nature of phosphorus incorporated into the zeolite. Elemental analysis results for various materials considered in this investigation are collected in Table (23) Krawietz, T. R.; Lin, P.; Lotterhos, K. E.; Torres, P. D.; Barich, D. H.; Clearfield, A.; Haw, J. F. J. Am. Chem. Soc. 1998, 120, 8502.

Figure 2. GC-FID chromatograms showing product distributions from methanol conversion in plug flow reactors at various temperatures for the catalyst prepared by the modification of HZSM-5 (SiO2/Al2O3 ) 80) by flowing a solution of trimethylphosphine in methanol followed by oxidation (as described in the text) after 120 min of continuous time on stream. The conditions in every case included a methanol WHSV (g feed/g catalyst/h) of 8 h-1 over a 300 mg catalyst bed, and 200 sccm helium carrier gas: (a) 648 K; (b) 673 K; (c) 698 K. Table 2. Elemental Analysis Results Performed by Galbraith Laboratories on the Trimethylphosphine and Methanol-Modified HZSM-5 (SiO2/Al2O3 ) 80) results (%)a silicon aluminum phosphorus a

40.4 0.83 1.37

Results are given in weight percent.

2. The amount of phosphorus in our modified catalyst, 1.4 wt %, is clearly significant. Figure 3 reports 31P MAS NMR spectra characterizing the chemical form of the phosphorus in our modified catalyst under various conditions. In Figure 3a, we see that the dry, calcined catalyst has a single, broad isotropic resonance at approximately -42 ppm, as well as a series of spinning sidebands, reflecting appreciable chemical shift anisotropy and a lack of motional averaging. The assignment of this resonance was initially perplexing, and we used computational modeling to consider a number of hypothetical structures with phosphorus associated with the zeolite framework, acid site, or extraframework aluminum sites. None of these hypothetical structures had theoretically predicted 31P chemical shifts in accord with the experimental spectrum. Theory did suggest a possibility that we overlooked: the observed shift is in very good agreement with the tetrahedral P4O10 molecule. Furthermore, the shoulder observed upfield of -42 ppm is entirely consistent with the previously reported homonuclear dipolar coupling between

Methanol ConVersion of Phosphate-Modified HZSM-5

Figure 3. 31P MAS NMR spectra of a sample of phosphorusmodified HZSM-5 catalyst (trimethylphosphine procedure): (a) after drying under vacuum at 673 K; (b) following subsequent exposure to laboratory air at room temperature for 3 h; (c) after a second evacuation at 673 K. We assign the broad isotropic peak at approximately -42 ppm to P4O10, which acts as a reservoir of included phosphate during calcinations. Upon hydration, this condensed species is converted reversibly to monomeric phosphate (0 ppm) and its lower oligomers (e.g., pyrophosphate, -11 ppm). The spinning speed for all experiments was set to 6.0 kHz. All spectra (2000 scans) were measured at 298 K using a 3 s pulse delay. An asterisk denotes a spinning sideband.

Langmuir, Vol. 22, No. 10, 2006 4849

Figure 4. 27Al MAS NMR spectra of various HZSM-5 (SiO2/Al2O3 ) 80) catalyst samples, all after exposure to atmospheric moisture: (a) unmodified HZSM-5 calcined at 823 K and (b) following modification using the trimethylphosphine procedure described in the text. The spinning speed for all experiments was set to 6.0 kHz. All spectra (10 000 scans) were measured at 298 K using a 15° flip angle and a 3 s pulse delay.

phosphorus nuclei in P4O10.24 This feature is powerful evidence in favor of the proposed assignment. P4O10 (commonly known as P2O5) is an exceptionally strong drying agent made by the severe dehydration of phosphoric acid. Indeed, as shown in Figure 3b, exposure of the same sample to atmospheric moisture results in a 31P spectrum dominated by phosphoric acid at 0 ppm, and perhaps some pyrophosphoric acid at -11 ppm. When this sample was evacuated at 673 K to remove water, the 31P spectrum in Figure 3c showed reversible P4O10 formation. Following methanol conversion, the 31P spectrum (not shown) is essentially unchanged from that of the wet catalyst in Figure 3b. We calculated P4O10 to have a diameter of 0.53 nm (using B3YLP/631G*). The maximum diameter of a sphere that can be included in the channel intersections of the MFI structure is 0.63 nm and the maximum sphere that can diffuse through the channels is 0.44-0.46 nm.25 Therefore, it is plausible that P4O10 can be formed and trapped in the channel intersections of ZSM-5. Other NMR studies showed that our modification procedure partially dealuminated the zeolite. The 27Al MAS NMR spectra in Figure 4 show that approximately 67% of the framework aluminum was converted to nonframework aluminum with our modification procedure. No additional phases, such as AlPO4,

were observed. Other supporting evidence for the reduction of framework acid site density and the creation of extraframework aluminum is afforded by the 29Si and 1H MAS NMR spectra reported in Figure 5. All of this evidence is consistent with the quantitative interpretation of the 29Si MAS NMR spectrum in Figure 5b that our modified catalyst has a framework SiO2/ Al2O3 ratio of approximately 240, hence our decision to compare catalyst performance using an unmodified HZSM-5 zeolite with a similar SiO2/Al2O3 ratio as one of the control samples in Figure 1. Mechanistic Studies. To understand the changes in methanolconversion product selectivity following our phosphate modification procedure, we first considered the possibility of acid strength differences. In previous studies of zeolite acid strength, we adsorbed NMR probes such as acetone-13C and interpreted changes in the isotropic shift induced by partial proton transfer.26,27 Two complications with our modified catalyst suggested that this type of study would be less straightforward here. The presence of large amounts of extraframework aluminum would, of course, complicate the spectra of adsorbed probes due to the presence of both Lewis and Brønsted absorption sites. More fundamentally, however, is the fact that the chemical form of the phosphate component of the catalyst is different between the dry state in which probe molecule studies are performed and under reaction conditions in which much water is continuously produced by methanol dehydration.

(24) Jeschke, G.; Hoffbauer, W.; Jansen, M. Chem.sEur. J. 1998, 4, 1755. (25) International Zeolite Association Structure Commission Home Page. http:// www.iza-structure.org (accessed Dec 2005).

(26) Xu, T.; Kob, N.; Drago, R. S.; Nicholas, J. B.; Haw, J. F. J. Am. Chem. Soc. 1997, 119, 12231. (27) Xu, T.; Munson, E. J.; Haw, J. F. J. Am. Chem. Soc. 1994, 116, 1962.

4850 Langmuir, Vol. 22, No. 10, 2006

Figure 5. 29Si and 1H MAS NMR spectra of HZSM-5 (SiO2/Al2O3 ) 80) before and after modification using the trimethylphosphine procedure: (a) 29Si spectrum of the unmodified HZSM-5. (b) 29Si spectrum of the modified zeolite. The 29Si spectra provide supporting evidence that the treatment procedure reduces the framework aluminum content. (c) 1H spectrum of the unmodified zeolite. (d) 1H spectrum of the modified zeolite showing a reduction in Brønsted acid sites. All spectra were measured at 298 K with spinning speeds of 5.5-6 kHz. 29Si spectra (500 scans) were measured using a 120 s pulse delay. 1H spectra (32 scans) were measured using a 10 s pulse delay.

We performed a series of experiments using a solution prepared using 1 mol of 2-propanol, 1 mol of 2-propanol-d8, 2 mol of H2O, and 2 mol of D2O. When introduced into a catalyst bed at a WHSV of 8 h-1, this solution quickly equilibrated the Brønsted acid sites to 50% H, 50% D, and produced, initially, 50% propene and 50% propene-d6 with one mole of water (nominally HOD) for every mole of carbon, which is exactly what happens under methanol-to-hydrocarbon conditions. We then monitored the distribution of propene isotopomers in the product gases by GCMS as a quantitative measure of the relative extents of reaction for propene H/D exchange. This gives us the ability to make an assessment of the relative numbers and strengths of acid sites under conditions that fairly closely simulate reaction conditions, especially since most of the reactions that lead to methylbenzene formation here involve the reactions of propene and its oligomers (vide infra). The detailed methodology for this approach with various probe molecules will be reported elsewhere, but it can be appreciated by considering the theoretical fragmentation patterns in Figure 6 for pure propene-d0, pure propene-d6, a mixture of 50% propened0 and 50% propene-d6, and, finally, a fully statistical distribution of equal numbers of hydrogen and deuterium over all isotopomers as in complete exchange. We have developed algorithms for fitting experimental isotopic distribution patterns such as those in Figure 6 to recover the detailed isotopomer distribution. The experimental and simulated results in Figure 7 compare the application of this methodology to unmodified HZSM-5 with

Abubakar et al.

Figure 6. GC-MS ion mass distributions for pure propene-d0, pure propene-d6, a mixture of 50% propene-d0 and 50% propene-d6 with no exchange, and the same mixture with a fully statistical distribution of equal numbers of hydrogen and deuterium over all isotopomers resulting from complete exchange.

SiO2/Al2O3 ratios of 80 or 280 and our modified catalyst. In each case, catalyst beds of the same mass were compared at two temperatures, 473 and 573 K. Experimental and fitted mass patterns in the molecular ion region shown in each case are the detailed isotopomer distributions producing the reported fits. Briefly, these patterns show that, at 473 K, the extent of H/D exchange on our modified catalyst is very similar to the extent observed on standard HZSM-5 with a similar framework SiO2/ Al2O3 ratio, namely, 280. It is important to note that, if the acid site strength of these materials differed even slightly, the extent of the exchange reaction would have been very different. On the unmodified zeolite with a framework SiO2/Al2O3 ratio of 80, the distribution of propene isotopomers is broader as a result of the catalyst bed containing 3.5 times the number of acid sites and, hence, catalyzing proportionately more exchange events. At 573 K, all three catalysts showed essentially complete H/D exchange to a statistical distribution of isotopomers. In summary, the exchange patterns in Figure 7 support the interpretation that the number and strength of Brønsted acid sites in our modified catalyst are very similar to that in unmodified HZSM-5 with SiO2/Al2O3 ) 280. It is important to note that our measurement directly relates to the activation of propene for reactions under acid catalysis. These and other experiments using propene as a feed suggested an alternative explanation for the selectivity changes upon phosphate modification. Figure 8 reports GC-FID chromatograms of the reaction products of 1 mol of 2-propanol/2 mol of water at a WHSV of 8 h-1 and 673 K after a 90 min reaction time. Our modified catalyst yielded far less ethylene and butenes under these conditions than did the standard HZSM-5 with SiO2/ Al2O3 ) 80. As in other experiments, some, but not all, of the difference could be made up by increasing the SiO2/Al2O3 ratio to 280. The most simple mechanism for converting propene to ethylene under acid catalysis involves multiple oligomerization and cracking cycles. While elimination of ethylene in a cracking

Methanol ConVersion of Phosphate-Modified HZSM-5

Figure 7. GC-MS ion mass distributions from propene exiting 300 mg catalyst beds of various HZSM-5 zeolites at either 473 or 573 K. A reaction mixture (mole ratio) of 1 2-propanol/1 2-propanold8/2 H2O/2 D2O was flowed at WHSV ) 8 h-1 for 10 min prior to measuring the isotopic distribution of propene under steady-state conditions. A statistical distribution of labels reflecting complete H/D exchange under these conditions would result in a product predicted to be 1.6% d0, 9.4% d1, 23.4% d2, 31.3% d3, 23.4% d4, 9.4% d5, and 1.6% d6. HZSM-5 (SiO2/Al2O3 ) 80) at 473 K generated propene with significant, but still incomplete, H/D label scrambling. HZSM-5 (SiO2/Al2O3 ) 80) at 573 K formed propene with a nearly statistical distribution of isotopomers, indicating essentially complete exchange. At 473 K, our phosphorus-modified HZSM-5 exhibited modest H/D exchange, yielding mostly propene-d0 and propene-d6. At 573 K, our phosphorus-modified HZSM-5 showed near-complete exchange. Unmodified HZSM-5 (SiO2/Al2O3 ) 280) at 473 K yielded mostly propene-d0 and propene-d6, indicating modest H/D exchange. Unmodified HZSM-5 (SiO2/Al2O3 ) 280) at 573 K formed propene with nearly complete H/D exchange. See text for interpretations.

step is unfavorable, the ethylene so formed through many cycles will tend to accumulate because it is less reactive than the larger olefins. Finally, we also studied the homologation of propene and methanol-13C on the catalysts. We flowed a 5:1 mol/mol solution of 2-propanol and methanol-13C over the catalyst beds at a temperature of 673 K and a WHSV of 8 h-1, and measured the 13C isotopic distribution in various products. The results for butene isotopomers in Table 3 show that, for our phosphate-modified catalyst, only 7% of the C4 olefins had no 13C, whereas the oligomerization and cracking of propene was much more significant on unmodified SiO2/Al2O3 ) 280 HZSM-5, and ∼24% of the C4 olefins had no 13C.

Discussion Our catalyst modification procedure, which involves flowing trimethylphosphine and methanol over a zeolite bed at high temperature, followed by oxidation, partially dealuminates the

Langmuir, Vol. 22, No. 10, 2006 4851

Figure 8. GC-FID chromatograms showing reaction products from the conversion of propene on various HZSM-5 catalyst beds at 673 K under conditions that mimic methanol conversion chemistry: (a) On our phosphorus-modified catalyst, oligomerization and cracking was suppressed, leading to very low yields of ethylene and butenes and essentially no aromatics. (b) On unmodified HZSM-5 (SiO2/ Al2O3 ) 80), the yields of ethylene and butenes were appreciable, and methylbenzenes were also formed. (c) Using unmodified HZSM-5 (SiO2/Al2O3 ) 280) the yields of ethylene, butenes, and methylbenzenes were reduced, but not to the extent achieved with phosphorus modification. All experiments were carried out using a 1 2-propanol/2 H2O (mole ratio) feed with a WHSV (g feed/g catalyst/ h) of 8 h-1 at 673 K and 200 sccm helium onto 300 mg catalyst beds to provide propene on the catalyst bed in the presence of water, as in methanol conversion. Table 3. 13C Distribution in the Products Formed by the Reaction of a 5:1 Excess of Propene and Methanol-13C on Various HZSM-5 Catalysts at 673 K 13

13

13

13

1-butene and isobutene HZSM-5 (SiO2/Al2O3 ) 80) 72.8 24.2 HZSM-5 (SiO2/Al2O3 ) 280) 21.9 68.8 PZSM5a (SiO2/Al2O3 ) 80) 7.0 84.5

3.0 8.0 6.7

0.3 1.5 1.9

0.0 0.0 0.0

2-butene (trans) HZSM-5 (SiO2/Al2O3 ) 80) 73.5 24.2 HZSM-5 (SiO2/Al2O3 ) 280) 25.1 67.3 PZSM-5a (SiO2/Al2O3 ) 80) 6.4 85.9

2.1 6.2 6.6

0.4 1.2 0.9

0.0 0.2 0.2

2-butene (cis) HZSM-5 (SiO2/Al2O3 ) 80) 69.9 26.6 HZSM-5 (SiO2/Al2O3 ) 280) 24.5 67.3 PZSM-5a (SiO2/Al2O3 ) 80) 8.3 84.3

2.9 6.8 6.1

0.6 1.0 0.9

0.0 0.4 0.3

C0

13

C1

C2

C3

C4

a The catalyst was prepared by treatment with 1:1 molar equivalent of trimethylphosphine and methanol followed by oxidation at 823 K.

framework and leaves approximately 1.4 phosphorus equivalents per aluminum, whether in or out of the zeolite framework. The chemical form of the phosphorus under methanol-conversion working conditions is phosphoric acid and lower oligomers of phosphoric acid. The reduction in framework Brønsted acid site density resulting from this treatment accounts for some, but not

4852 Langmuir, Vol. 22, No. 10, 2006

Figure 9. B3LYP/DZP-optimized structure of H3PO4 hydrogen bonded to a cluster model of the framework Brønsted acid site of zeolite HZSM-5.

all of the changes in catalyst selectivity from methylbenzenes to light olefins and, preferentially, propene. The framework Brønsted sites remaining after modification have acid strengths that are indistinguishable from those of an unmodified zeolite of approximately the same acid site density, based on the rates of propene H/D exchange at 473 K. The experiments described in Figure 8 and Table 3 are strong evidence that the rate of propene oligomerization is suppressed on the modified catalyst. With a significant reduction in the concentration of hexenes and nonenes on the catalyst, the rates of cracking to ethylene, butenes, and pentenes are inhibited, and, presumably, the rate of cyclization to methylbenzene precursors is greatly suppressed as well. The homologation of propene with methanol to form butenes, a sterically nondemanding reaction, is still reasonably facile on the modified catalyst. Our interpretation of all of the available evidence is suggested in part by the theoretical structure (B3LYP/DZP) of H3PO4 cooperatively hydrogen bonded to a framework Brønsted site (Figure 9). Other structures of interest would include complexes of phosphoric acid with extraframework aluminum sites. In both cases, the phosphate complexes would provide dynamic (dissociating) transient species that would partially block channels and channel intersections, and restrict access to sterically

Abubakar et al.

demanding transition states. Our B3LYP/DZP calculations predict that the adsorption energy of H3PO4 onto the framework Brønsted site is 27.9 kcal/mol. The experimental enthalpy of deprotonation of H3PO4 in the gas phase is 330.5 kcal/mol, a value approximately 30 kcal/mol higher (less acidic) than most estimates of the value for HZSM-5. For comparison, at B3LYP/DZP, the uncomplexed cluster used in Figure 9 has a deprotonation energy of 301.3 kcal/mol, while H3PO4 is 333.7 kcal/mol at the same level of theory. The point of this latter argument is that we expect, on the basis of these calculations, that we can neglect the Brønsted acidity of gas-phase phosphoric acid relative to the stronger uncomplexed framework site. We also computed the deprotonation energy of the complex in Figure 9 and obtained a value of 303.4 kcal/mol. This latter calculation suggests that the Brønsted acid sites in phosphate-treated HZSM-5 will have essentially identical acid strengths, regardless of whether they are complexed by phosphoric acid. In summary, the available evidence indicates that phosphoric acid molecules are created in the channels of HZSM-5 by our modification procedure and that these survive calcination as P4O10. Under reaction conditions, we expect phosphoric acid to be complexed to Brønsted sites, and probably to extraframework aluminum sites as well. We cannot differentiate between the contribution to the reduction in pore size and volume of the channel from the phosphoric acid and that from the extraframework aluminum. However, it has been reported that the dealumination of HZSM-5 (without the addition of phosphoric acid) results in increased selectivity for aromatics and more rapid deactivation compared to the parent HZSM-5.28 Indeed, it may be that a combination of the effects of both is responsible for the behavior of our modified catalyst. In any case, these complexes will provide obstacles for the bulky transition states required to oligomerize propene. Suppression of hexenes and higher propene oligomers prevents olefin equilibration by cracking as well as routes leading to methylbenzenes. We also adsorbed methanol and benzene separately onto both unmodified and modified ZSM-5, and found that methanol and benzene uptake is higher (usually ∼4-7% for methanol and ∼2-4% for benzene) on the modified ZSM-5. We suggest that this may be caused by dealumination of the material stemming from the phosphate modification. The dealumination and phosphate modification occur at the same channel intersections, but, in the active material, the phosphate is converted to phosphoric acid. We speculate that the phosphoric acid may interact via hydrogen bonding with adjacent Brønsted acid sites, resulting in the reduction of the functional channel intersections. Acknowledgment. This work was supported by the U.S. Department of Energy, Office of Basic Energy Sciences (BES) (Grant No. DE-FG03-99-ER14956). LA0534367 (28) Kubelkova, L.; Novakova, J.; Nedomova, K. J. Catal. 1990, 124, 441.