Studies in Surface Science and Catalysis, volume 158 J. (~ejka, N. Zilkov~iand P. Nachtigall (Editors) 9 2005 Elsevier B.V. All rights reserved.
Beckmann rearrangement on microporous and mesoporous silica R. Palkovits a, Y. Ilhan a, W. Schmidt a, C.M. Yang c, A. Erdem-Sentalar b, F. Schiith a aMax-Planck-Institut fi~r Kohlenforschung, 45468 Mtilheim an der Ruhr, Germany bIstanbul Technical University, Department of Chemical Engineering, Istanbul, Turkey CNational Tsing Hua University, Department of Chemistry, Hsinchu, Taiwan Crosslinked Silicalite-1 and SBA-15 have been investigated as solid acid catalysts in the vapour phase Beckmann rearrangement reaction of cyclohexanone oxime to c-caprolactam. The catalytic activity of crosslinked colloidal Silicalite-1 is strongly dependent on the used amount of linker. With increasing amount of linker the external surface area increases as well as the catalytic activity. Surface modified SBA-15 was investigated to study the influence of pore dimensions, acid sites and catalytic activity in micro- and mesopores. A combination of these studies allows making suggestions concerning the location of the catalytic reaction. 1. I N T R O D U C T I O N On industrial scale the Beckmann rearrangement of cyclohexanone oxime to e-caprolactam, an important intermediate in the production of Nylon-6, is carried out in a liquid phase reaction. The process is catalyzed with concentrated sulphuric acid resulting in ammonium sulphate as by-product in amounts as high as 1.8 t / t e-caprolactam [ 1]. Together with the high corrosion potential of the reaction media these are the main drawbacks of the industrial process. In the last decade research focused on alternative solid acid catalysts such as zeolites with the perspective of replacing the industrial liquid phase reaction by a vapour phase process. Beside studies on the influence of acid strength and on the optimization of reaction conditions, a key subject of research has been the location of the reaction sites in the Beckmann rearrangement. Although Sato et al.  found evidence for a reaction on the outer surface, the selectivity for caprolactam formation was found to be dependent on the pore size of the zeolites . H61derich et al. [4,5] reported an increasing catalytic activity with decreasing crystal size indicating a reaction mainly occurring on or close to the outer surface. Nevertheless it has been shown in sorption experiments  that both cyclohexanone oxime and caprolactam can diffuse into the pore system of MFI-type materials. In spite of these investigations, the location of the Beckmann rearrangement is still a matter of debate. In this study, we use Silicalite-1 and SBA-15 in the Beckmann rearrangement of cyclohexanone oxime to e-caprolactam. Combining the idea of a reaction mainly occurring on or close to the outer surface  with the proposal of Martens et al.  of the existence of zeolite "nanoslabs" in solution having a maximum external surface area, colloidal Silicalite-1 crystals were crosslinked with a silicone linker to form a network with the Silicalite-1 crystals. This helps to avoid agglomeration of the particles during the isolation step from solution while a high external surface area is maintained . The modification allows us to investigate the catalytic activity depending on the external surface area. The results of this study induced us to further
1256 investigate the location of the catalytically active sites. Recently Yang et al. presented a method [8,9] for template removal from ordered mesoporous silica (SBA-15) via ether cleavage which allows a selective opening of micro- and mesopores. Consequentially the influence of the pore types on the catalytic activity can be studied separately. Thus the modified materials were investigated concerning the dependence of the catalytic activity on external surface area and pore dimensions. 2. E X P E R I M E N T A L
2.1. Crosslinked Silicalite-I Colloidal Silicalite-1 was synthesized from a clear solution with a molar composition of 7.8 TPAOH : 21.4 SiO2 : 390 H 2 0 : 85.6 EtOH following a method described by Ravishankar et al. . The solution was prepared starting with the hydrolysis of TEOS (tetraethoxysilane) in a concentrated aqueous TPAOH solution (tetrapropyl ammonium hydroxide, 40%) after 30 min, 9 g distilled water were added and the solution was stirred for another 24 h at RT. The samples were directly dried after aging at RT and 363 K, respectively, skipping the extraction step described in , followed by calcination at 823 K for 5 h with a heating rate of 1 K/min. In order to crosslink the zeolite particles 1,7-dichloro-octamethyl-tetrasiloxane, 0.1 mL (0.26 mmol), 0.25 mL (0.65 mmol), 0.5 mL (1.3 mmol) or 1 mL (2.6 mmol), were added to 4 mL of the synthesis solution described above which had previously been aged for 24 h. After stirring for 5 h, the samples were aged without stirring at RT or 363 K for 24 h, dried at 363 K for 24 h, and finally calcined at 823 K as described above. The samples were acid treated with 0.2 M HNO3 (100 mL/g sample). The solution was stirred at RT for 3 h, filtered, washed with 500 mL of water and 50 mL acetone, dried at RT overnight and calcined again at 823 K with a heating rate of 1 K/min. 2.2. Characterization of Silicalite-1 Crosslinked Silicalite-1 samples with different amounts of linker were characterized by nitrogen sorption with a Micromeritics ASAP 2010 at liquid nitrogen temperature. Prior to the measurements, samples were pretreated overnight under vacuum at 473 K. 29Si MAS NMR measurements were carried out with a Bruker Avance 500WB spectrometer at a spinning rate of 10 kHz. XRD measurements were performed on a STOE STADI P transmission diffractometer with a position sensitive detector. Selected samples were analyzed by transmission electron microscopy on a Hitachi HF 2000 microscope equipped with a cold field emitter gun. 2.3. Mesoporous Silica (SBA-15) Ordered mesoporous silica (SBA-15) has been synthesized with triblock poly(ethylene oxide)-poly(propylene oxide)-poly(ethylene oxide) (EOEoPO70EO20) copolymer Pluronic P 123 as structure-directing agent, following the procedure described by Kleitz et al. . The synthesis solution had a molar composition of 0.16 P123 : 9.54 SiO2 : 950 H20 : 5.12 HC1. TEOS was added under stirring at 308 K, resulting in a molar ratio of SiO2 : P123 of 60. The mixture was stirred at 308 K for 24 h, and subsequently aged for 24 h at 333 K or 373 K, respectively. The precipitate was filtered and dried without washing at 363 K for 24 h. To facilitate the template removal 10 g solid were stirred in 500 mL EtOH under addition of 0.01 mole HC1 for 30 min, followed by filtration, washing with acetone and drying at 363 K for 24 h. For cleaving the template to generate exclusively mesopores, 10 g solid were mixed with 300 g 96 % H2SO4 solution and 300 mL H20, refluxed at 368 K for 18 hours and filtered. The
1257 solid was washed with water until the eluent became neutral, then washed with acetone and dried at 363 K for 24 hours. To generate micropores, the acid-treated sample was heated to 573 K in air for 3 h with a heating rate of 2 K/min . To hydrophobize the mesopore surface with trimethylchlorosilane (TMCS), 1.5 g of the acid treated sample were stirred in 100 mL toluene for 30 min, 8 mL of TMCS were added. The sample was stirred further for 24 h, filtered, washed and calcined at 573 K to remove the template from the micropores as described above. Additionally, as-synthesized material was heated to 453 K with 5 K/min, kept there for 3 h, followed by heating to 823 K with 1 K/min and calcined for 6 h.
2.4. Characterization of SBA-15 The SBA-15 samples were characterized by N2 sorption with a Quantachrome Nova 3200E at liquid nitrogen temperature. Prior to the measurements, samples were pre-treated overnight under vacuum at 473K. 29Si MAS NMR measurements were carried out with a Bruker Avance 500WB spectrometer at a spinning rate of 10 kHz. XRD measurements were performed on a STOE STADI P transmission diffractometer with a position sensitive detector. 2.5. Beckmann Rearrangement The catalytic activity of the different materials was tested in the vapour phase Beckmann rearrangement reaction of cyclohexanone oxime to e-caprolactam. The reaction was carried out at atmospheric pressure and 553 K. The continuous flow reactor was packed with 200 mg of catalyst together with 2 g quartz sand. The samples were pelletized and ground to a particle size fraction of 125-250 ~tm. The samples were activated overnight in air flow at 573 K. Before reaction, the reactor was cooled down to 553 K. Cyclohexanone oxime was dissolved in toluene in a molar ratio of 26.5 and fed to the reactor by a HPLC-pump along with N2 as carrier gas. The reaction was carried out under atmospheric pressure with WHSV = 1.95 goxime g'lcat h "l, thus adjusting the vapour pressure of oxime to 2.2 kPa. The product was recovered by condensation of the effluent, collecting fractions for 5 min each. For analysis, three samples collected after 1 h (55, 60, 65 min) and 4 h (275, 280, 285 min) were chosen. The condensate was analyzed by gas chromatography by a flame ionization detector, Agilent Technology 6890N. The deactivation is defined as decrease in reaction rate between 1 and 4 h. 3. R E S U L T S AND D I S C U S S I O N
3.1. Textural properties of crosslinked Silicalite-1 Colloidal Silicalite-1 was reacted with a silicone linker to form a network with surface silanols of the zeolite crystals and thus avoid agglomeration to keep a maximum of external surface area . The process is schematically shown in Fig. 1. Nitrogen sorption shows a Type I isotherm for unlinked Silicalite-1, meeting the expectations for a zeolite material. For ~'~ -OH
~ -OH ~
Me Me Me M e a l l 11" Cl--Si-O-Si-O-Si-O-Si-CI I I I I Me Me Me Me
Fig. 1. Proposed crosslinking reaction for Silicalite-1
1258 crosslinked samples an additional hysteresis loop at higher pressure can be found indicating a secondary porosity. This can be attributed to the expected network formation, and thus confirms that the proposed crosslinking reaction proceeded. The effects of different amounts of linker are different, depending on the aging temperature. Samples aged at 363 K (they are already crystalline as shown by XRD) show decreasing microporosity with increasing linker P,,..i
ui E 200
9~ 300 O m
| 100 E
pip0 3 P/POp. Fig. 2. Adsorption isotherm of S'ilicalite-1 samples aged at 363 K (right, a, b shifted up 50 cm/g, ST ) and RT (left, b, c shifted up 100 cm3/g, STP): a) tmlinked b) 0.65 mmol c) 1.3 mmol
amounts (Fig. 2, tight) which corresponds to the overall decreasing fraction of Silicalite-1 in the samples. The mesopore area changes from 127 m2/g for unlinked to 114 mZ/g and 196 mZ/g for 0.65 mmol and 1.3 mmol of linker. Crosslinking of Silicalite-1 aged at RT (amorphous according to XRD up to aging times of three weeks) results also in decreasing microporosity with increasing amounts of linker, consistent with the results found for samples aged at 363 K. The mesopore surface increases strongly by the crosslinking process (Table 1) giving additional evidence of a network formed by crosslinking. Table 1 Textural properties of calcined Silicalite-1 samples a[ged at RT (analysed by DFT) Linker Amount Micropore Volume, cm3/g Mesopore Area, m2/g Unlinked 0.146 70 0.65 mmol 0.044 531 1.3 mmol 0.134 252 2.6 mmol 0.070 282 The non-continuous development of micropore volume and mesopore area is somewhat surprising, but was reproduced for several series of samples. It might be due to pore blocking effects and linker molecules just coveting the single particles and not connecting to a network. 3.2. Connectivity of crosslinked Silicalite-1
29Si MAS NMR spectroscopy provides information about the condensation degree of the Silicalite-1 samples and allows distinguishing the amounts of different silicon species present. For the unlinked samples three broad resonances at -88, -98 and -109 ppm can be observed, corresponding to Q2, Q3 and Q4 silicon atoms (Fig.3,a). In the crosslinked samples the signals corresponding to Q2 and Q3 silicone decrease in intensity with increasing amount of linker. The relative intensity of the Q3 signal first increases due to Q2 groups reacting with the linker and thus transforming to Q3 groups, but they decrease, too, for higher amounts of linker which
Fig. 3.29Si MAS NMR spectra of Silicalite-1 samples aged at RT (left) a) unlinked b) 0.26 mmol c) 0.65 mmol d) 1.3 mmol e) 2.6 mmol of linker and Silicalite-1 aged at 363 K (right) a) unlinked b) 0.26 mmol c) 2.6 mmol of linker.
which is clearly visible for Silicalite-1 aged at RT with 1.3 and 2.6 mmol of linker. In agreement the intensity of the Q4 signal increases with decreasing relative intensity of Q2 and Q3 silicon signals. Additionally two new signals become visible at -18 and-22 ppm. They can be attributed to the silicon atoms in the silicone linker. These observations support the notion that there is indeed a crosslinking reaction, consuming a part of the silanol groups on the one hand, but creating mesoporosity on the other hand.
3.3. Catalytic performance of crosslinked Silicalite-1 The aging temperature has a strong influence on the catalytic activity of both the unlinked and the crosslinked Silicalite-1 samples. Overall a remarkable coherence between the surface area and the catalytic activity of the Silicalite-1 samples can be observed. Samples aged at RT show a maximum in mesopore area for an amount of crosslinker of 0.65 mmol. The catalytic activity reflects this. These materials exhibit the highest reaction rate for the sample with the highest mesopore area (Fig. 4, left). Silicalite-1 aged at 363 K shows already 600 c~
0 200 0 u~ 100
1.3 m m o l
p, .10 r O
5 0 ,..., 0.65 m m o l '
1.3 m m o l
Fig. 4. Relation between catalytic activity and extemal surface of crosslinked Silicalite-1 aged at RT (left) and Silicalite-1 aged at 363 K (fight) for different amounts of linker. unlinked a significant catalytic activity which further increases with increasing amounts of linker (Fig.4, right) as the mesopore area does. Crosslinking of the colloidal Silicalite-1 crystals enhances clearly the catalytic activity. The results suggest that the reaction is at least somewhat diffusion limited, and therefore probably takes place on or close to the extemal
1260 surface of the crystallites. These findings induced us to further investigate the location of the catalytic reaction in the vapour phase Beckmann rearrangement of cyclohexanone oxime to caprolactam, and in this regard especially to take materials with combined micro- and mesopore structure into consideration. 3.4. Characteristics of ordered mesoporous silica (SBA-15) Ordered mesoporous materials such as SBA-15 have a combined micro- and mesopore system. Different aging temperature and the stepwise template removal via ether cleavage  allow tuning of the pore systems and selective opening and functionalization of only the mesopore surface. For samples aged at 60~ higher micropore volumes can be measured while samples aged at 100~ show larger mesopore areas together with larger pore dimensions. In Fig. 5 the sorption isotherms of differently pretreated samples are compared. The sample directly calcined at 550~ has smaller pores and a lower mesopore volume compared to the samples which had been pretreated with sulphuric acid and the subjected to ,-, 160
t~ 8o ~70
E IO0 .u. "o ,,Q
"0 a) 50 .Q
0.8 P/Po 1
0.8 P/Po 1
Fig. 5. N2 sorption isotherms of SBA-15 aged at 60~ (fight, b shifted with 200 cma/g, c with 100 cm3/g, STP) or 100~ (left, b shifted with 400 cm3/g, c with 200 cm3/g, STP), a) calc. 550~ b) H2SO4-treated, calc. 300~ c) H2SO4-treated, mesopores TMCS covered, calc. 300~ low temperature calcination. This can be attributed to the additional condensation during the acid treatment which stabilizes the network against shrinkage. In addition, the lower temperature of calcination helps to maintain the pore system. SBA-15 with hydrophobized mesopore surface also exhibits a sharp hysteresis at high relative pressure together with high adsorbed N2 volumes, confirming the absence of pore blocking in the mesopores. 3.5. M A S N M R Analysis of SBA-15
29Si MAS NMR analysis corroborates the picture of a better condensed network in the case of acid treated samples calcined at low temperature. For all samples three broad resonances at -88, -98 and -109 ppm can be observed, corresponding to Q2, Q3 and Q4 silicon atoms. Acid treated, low temperature calcined SBA-15 shows almost no intensity for Q2 groups and intense signals for Q3 and Q4 groups whereas SBA-15 calcined at 550~ exhibits a broad signal covering the range of Q2, Q3 and Q4 groups. Samples with hydrophobized mesopore surface have decreased intensity for the Q2 and Q3 group signals, corresponding to their consumption in the hydrophobization reaction with TMCS. Due to the anchored trimethylsilane a new silicon resonance arises around -18 ppm (Fig.6).
Fig. 6. SBA-15 aged at 100~ (left) and 60~ (right), a) calc. 550~ b) H2SO4-treated, calc. 300~ c) H2SOa-treated, mesopores TMCS covered, calc. 300~ 3.6. Catalytic performance of SBA-15 The catalytic activity of SBA-15 in the Beckmann rearrangement of Cyclohexanone oxime to e-caprolactam does not seem to be connected to the aging temperature of the samples. For both aging temperatures, sulphuric acid treated and low temperature calcined samples show a superior activity compared to samples calcined at 550~ (Table 2). SBA-15 with hydrophobized mesopores surface exhibits a lower catalytic activity. None the less, the reduced activity indicates that the catalytic reaction takes place in both meso- and micropores. The NMR analysis allows determining the number of silanol groups which are supposed to be the catalytically active acid centers . Table 2 gives an overview over the intensity distribution and the silanol concentrations in the different samples. Table 2 Concentration of silanol groups from 29SiMAS NMR and catalytic activity SBA- 15 aged at 100~
Q2 Q3 Q4 T [%] [%] [%] [%]
Silanol conc. [mmol/g] ,
H2SO4, calc.300~ H2804, TMCS, calc.300~ SBA- 15 aged at 60~ calc.550~
H2SO4, calc.300~ H2SO4, TMCS, calc.300~
5 5 4
32 38 22
63 57 64
6.59 7.46 4.63
2.01E-01 3.33E-01 1.01E-01
Interestingly, the silanol concentrations in the samples correspond very well with the catalytic activity. However, it seems that the activity per silanol for the trimethylsilylated samples is lower than for the unmodified materials. Since the trimethylsilylation predominantly consumes the silanols in the mesopores (the micropores are still filled with template during the modification), one may speculate that in SBA-15 the highest contribution to the activity comes from the mesopore silanols. Whether this is due to different intrinsic activity of these silanols, mass transfer limitations for the reaction in the micropores, or lower concentrations of reagents in the pore systems due to the hydrophobization is unclear and more work is necessary to elucidate the decisive factors.
1262 4. CONCLUSION The investigation of crosslinked colloidal Silicalite-1 crystals in the Beckmann rearrangement of cyclohexanone oxime to e-caprolactam showed the advantages arising from an additional pore system. The results indicate a connection between mesopore area and catalytic activity of the Silicalite-1 samples. For Silicalite-1 aged at RT crosslinking results in increasing mesopore areas together with increasing catalytic activity up to a certain amount of linker. At higher linker concentrations, the pore volume decreases again, together with the catalytic activity. Silicalite-1 aged at 363 K is already crystalline and has appreciable catalytic activity even before crosslinking. Increasing amounts of linker result in increased mesopore area and an increased catalytic activity. SBA-15 has a combined meso- and micropore system. The catalytic activities of samples differently pretreated and with different types of pore systems confirm the advantages of a meso- and micropore system. Independent of the pretreatment, all samples show remarkably high catalytic activity, significant differences related to the aging temperature can not be found. SBA-15 treated with sulphuric acid and calcined at 300~ exhibits the highest activities corresponding to the high total silanol concentration. Samples calcined at 550~ are slightly less active which can be attributed to a smaller amount of silanol groups in the samples. The catalytic activity of samples with hydrophobized mesopore surface emphasises that both pore systems are involved in the catalysis, possibly with a higher contribution from acid sites in the mesopores. The catalytic activities correspond well with the silanol concentrations in the samples, giving evidence for them being the catalytically active species regardless of their location in meso- or micropores. Nevertheless, the location of the silanol groups may have an influence in terms of coordination of the silanol groups towards one another, resulting in silanol species of different acidity and steric environment. Further investigations are necessary to distinguish differently active silanol species, and to obtain a closer insight into the special nature of the catalytically active silanol species concerning acidity and geometric restrictions. ACKNOWLEDGEMENTS We would like to thank B. Zibrowius for NMR and E. Htibinger for GC measurements. REFERENCES
[ 1]     
H. Kath, R. Gl~iser, J. Weitkamp, Chem. Eng. Technol. 24 (2001) 150. H. Sato, K. Hirose, M. Kitamura, Y. Nakamura, Stud. Surf. Sci. Catal. 49 (1989) 1213. T. Yashima, N. Oka, T. Komatsu, Catal. Today. 38 (1997) 249. W.F. HSlderich, J. Rtiseler, G. Heitmann, A.T. Liebens, Catal. Today. 37 (1997) 353. G. Dahlhoff, J.P.M. Niederer, W.F. HSlderich, Catal. Rev. 43 (2001) 381. R. Ravishankar, C.E.A. Kirschhock, B.J. Schoeman, D. Devos, P.J. Grobet, P.A. Jacobs, J.A. Martens, Proceedings of the 12th Zeolite Conference MRS, PA (1999) 1825.  P. Agren, S. Thomson, Y. Ilhan, B. Zibrowius, W. Schmidt, F. Schtith, Stud. Surf. Sci. Catal. 142 (2002) 159.  CM. Yang, B. Zibrowius, W. Schmidt, F. Schtith, Chem. Mater. 15 (2003) 3739.  CM. Yang, B. Zibrowius, W. Schmidt, F. Schtith, Chem. Mater. 16 (2004) 2918.  M. Choi, W. Heo, F. Kleitz, R. Ryoo, Chem Comm. (2003) 1340.  G.P. Heitmann, G. Dahlhoff, W.F. H/51derich, J. Catal. 186 (1999) 12.