Niobium-silica catalysts for the selective epoxidation of cyclic alkenes: the generation of the...

13354 Phys. Chem. Chem. Phys., 2013, 15, 13354--13362 This journal is c the Owner Societies 2013

Cite this: Phys. Chem.Chem.Phys.,2013,15, 13354

Niobium–silica catalysts for the selective epoxidationof cyclic alkenes: the generation of the active site bygrafting niobocene dichloride

Cristina Tiozzo,a Chiara Bisio,ab Fabio Carniato,*b Alessandro Gallo,c

Susannah L. Scott,cd Rinaldo Psaroa and Matteo Guidotti*a

Niobium-containing silica materials obtained by deposition via liquid-phase grafting or dry impregnation

of niobocene(IV) dichloride are active and selective catalysts in the epoxidation of alkenes in the presence

of aqueous hydrogen peroxide. The generation of the catalytically-active Nb species was followed step-

by-step, and investigated using a combined DR-UV-Vis, NIR, Raman, XRD, XANES and EXAFS analyses. At

the end of the grafting procedure, the nature of the surface active species can be described as an oxo-

Nb(V) site, tripodally grafted onto the silica surface in close proximity to other Nb(V) centres. The liquid-

phase methodology provides a better dispersion of the metal sites onto the siliceous support than the

dry-impregnation approach. The niobium–silica catalysts were then tested in the epoxidation of cyclo-

hexene and 1-methylcyclohexene, as model substrates.

1. Introduction

In the quest for an efficient heterogeneous catalyst for theselective and sustainable epoxidation of bulky and richly func-tionalized alkenes of interest to the fine chemical industry,many research teams have focused their attention on meso-porous silica materials containing transition metals in highoxidation states (typically d0). Such catalytic centres do notundergo a change in their oxidation state. Rather, they are ableto induce a redox transformation in the substrate by catalysingthe action of the oxidising agent and enhancing the rate ofoxygen transfer from the oxidant to the CQC double bond.1

Hundreds of papers in the last two decades have shown howtransition metals can be added to silica materials either byisomorphous substitution into the silica matrix (i.e. through apH-adjusting one-pot procedure2–5) or by grafting a metalprecursor onto the support surface.6–9 In this area, mesoporoussilica materials (in particular, MCM-41 and SBA-15) have been

widely used as efficient supports to host a number of catalyticmetal species.10,11 Indeed, the high specific surface area andthe tuneable porosity of these solids characterised by large poreopenings (from 2 to 50 nm) has opened the way to their use inthe synthesis of different organic products, starting from bulkyand highly functionalised substrates of interest for the fine andhigh added-value chemistry.

Among the various metals widely used in heterogeneouscatalysis, Nb has attracted particular interest in recent years.Different studies demonstrated the activity of Nb-supportedAl2O3 for the transformation of alcohols and acids into ethersor esters under solvent-free conditions and microwave irradia-tion.12 The catalytic esterification of acetic acid with a variety ofalcohols was proved by using 10 wt% Nb2O5 supported onsilica–alumina.13

In the field of alkene epoxidation, as alternatives tothe traditional titanium, vanadium, tantalum or zirconium-containing silica solids,14–23 which often suffer from severemetal leaching problems,1,24,25 niobium-based silica catalystshave also been taken into account as suitable heterogeneouscatalysts. These solids have shown promising catalytic perfor-mances in water-containing liquid-phase oxidation reactions,since they are highly stable and robust towards metal leachingand hydrolysis. In this frame, mesoporous niobium–silicamaterials are widely considered effective systems for the epox-idation of cycloalkenes in the presence of aqueous hydrogenperoxide as oxidant.20,21,26

a CNR-Institute of Molecular Sciences and Technologies, Via C. Golgi 19, Milan,

Italy. E-mail: [email protected] Interdisciplinary Nano-SiSTeMI Centre, Department of Science and Technological

Innovation, University of Eastern Piedmont ‘‘Amedeo Avogadro’’, Viale T. Michel 11,

15121 Alessandria, Italy. E-mail: [email protected] Department of Chemical Engineering, University of California, Santa Barbara,

CA 93106-5080, USAd Department of Chemistry & Biochemistry, University of California, Santa Barbara,

CA 93106-5080, USA

Received 12th April 2013,Accepted 1st July 2013

DOI: 10.1039/c3cp51570b

www.rsc.org/pccp

PCCP

PAPER

This journal is c the Owner Societies 2013 Phys. Chem. Chem. Phys., 2013, 15, 13354--13362 13355

The remarkable stability of Nb centres in/on a silica matrixhas been attributed to the low surface mobility of niobiumatoms (for instance, the Tamman temperature, at which surfaceatoms begin to diffuse, is 893 K for Nb2O5 vs. 482 K for V2O5),and to the more optimal geometry of the Nb–O–Si bond anglescompared to V–O–Si.27 These factors likely combine to minimizethe propensity of Nb species to undergo hydrolysis/solvolysis andleaching in water-containing or highly protic media.

Promising results have been obtained, in previous studiesincluding those by some of us, for niobium–silica catalysts inthe epoxidation of unsaturated fatty acid methyl esters (FAMEs)and terpenes.28–31 The catalytic activity of these systems isinfluenced by the dispersion of niobium in/on the silica matrix,and by the synthesis conditions.23,32,33

In the present work, we describe a detailed spectroscopicstudy on the nature of the niobium species deposited ontomesoporous silica by two post-synthesis approaches: liquid-phase grafting and an alternative solventless synthesis protocolbased on dry impregnation (DI) of niobocene dichloride(Nb(Cp)2Cl2). Both approaches proved to be viable and versatilemethods to obtain leach-proof niobium silica catalysts. Never-theless, further investigation and analysis was necessary toshed light on the mechanism of niobocene deposition, thenature of the intermediate species present on the silica surfaceduring synthesis, and the generation of the Nb active sites.Finally, the catalytic properties of the resulting sites werestudied and compared in the epoxidation of two modelcycloalkene substrates: cyclohexene and 1-methylcyclohexene.

2. Experimental details2.1. Catalyst preparation

Niobium–silica catalysts were prepared by grafting bis(cyclo-pentadienyl)niobium(IV) dichloride (Nb(Cp)2Cl2; 95%, Aldrich)onto two silica supports: Dav (Grace Davison, Davisil SiO2

LC60A, 60–200 mm) and SBA-15. Dav silica is commerciallyavailable from the manufacturer. SBA-15 was synthesizedaccording to a literature procedure.34

All supports were pre-treated in dry air at 500 1C for 1 h, thenevacuated for 1 h at 500 1C and finally cooled to room temper-ature under vacuum.

Nb–SiO2 catalysts obtained via dry impregnation (Nb–Dav-DIand Nb–SBA-DI) were prepared as described in detail pre-viously.30 Briefly, the silica supports (Dav and SBA-15, respec-tively) were first hydrated with high purity deionized water(MilliQ Academic, Millipore, 18 MO cm) for 2 h, dried in arotary evaporator then heated in air at 300 1C for 1 h and undervacuum overnight at 300 1C. Nb(Cp)2Cl2 was finely ground andmixed, under an inert atmosphere in the solid phase, with thesilica. The mixtures were stirred overnight under vacuum atroom temperature. The resulting light brown solids werecalcined in dry oxygen at 500 1C for 2 h to obtain the finalNb(V)–silica catalysts.

Nb–SiO2 catalysts prepared via liquid-phase grafting (Nb–Dav-liq and Nb–SBA-liq) were obtained as described pre-viously.30 In this case, Nb(Cp)2Cl2 was grafted onto the surface

of the silica support via the procedure used to graft titanocenedichloride onto mesoporous silica supports.35,36

Bulk hexagonal phase Nb2O5 (Aldrich, 99.99%) was used asreference for comparison.

2.2. Catalyst characterization

The Nb content of the prepared catalysts was determined byinductively coupled plasma-optical emission spectroscopy(ICAP 6300 Duo, Thermo Fisher Scientific) after mineralizationof the samples in microwave digestion apparatus (MilestoneMLS 1200; maximum power 500 W) with a mixture of hydro-fluoric (aq. 40%) and fuming nitric acid.

Diffuse reflectance UV-Vis-NIR (DR UV-Vis-NIR) spectra werecollected using a Perkin Elmer Lambda 900 spectrometerequipped with an integrating sphere accessory and using ahand-made quartz cell that allows analysis both under vacuumconditions (residual pressure 10�5 mbar) and under a con-trolled gas atmosphere. Prior to the analysis, the samples weredispersed in anhydrous BaSO4 (10 wt%).

Raman spectra were recorded in the 4000–50 cm�1 range, usinga resolution of 4 cm�1 with 2000 scans, using a RFS 100 BrukerFT-Raman spectrometer equipped with a 1064 nm wavelengthexcitation laser. The laser power was set at 100 mW. The Ramanspectra were analyzed using OPUS 5.0 software (Bruker Optics).

X-ray diffractograms (XRD) were collected on unorientedground powders with a Thermo ARL ‘XTRA-048 diffractometerusing Cu Ka (k = 1.54 Å) radiation. Diffractograms wererecorded at room temperature with a step size of 0.021 and arate of 11 2y min�1.

N2 physisorption measurements were carried out at 77 K inthe relative pressure range from 1 � 10�6 r P/P0 r 1 using aQuantachrome Autosorb 1MP/TCD instrument. Prior to theanalysis, the samples were outgassed at 373 K for 3 h (residualpressure o10�6 mbar). Apparent surface areas were deter-mined using the Brunauer–Emmett–Teller equation, in therelative pressure range from 0.01 r P/P0 r 0.1. Pore sizedistributions were obtained by applying the NLDFT method(N2 silica kernel based on a cylindrical pore model applied tothe desorption branch).

X-ray absorption spectroscopy. XAS data for the Nb K-edgewere collected at room temperature in air at beamline 4–1 ofthe Stanford Synchrotron Radiation Lightsource (SSRL), whichoperates with an energy of 3 GeV and a current of 75–100 mA.Samples were packed into the sample holder in air, andfluorescence data were collected using a Lytle detector. In orderto improve the signal-to-noise ratio, at least three scans werecollected and averaged. Bulk Nb2O5 as well as Nb(Cp)2Cl2

diluted in BN and handled under an inert atmosphere wereanalyzed as reference compounds.

The data were analyzed using Athena and Arthemis soft-ware37 to obtain coordination numbers (N), distances of scat-tering atoms (R) and disorder factors (sDW).

2.3. Catalytic tests

Epoxidation of cyclohexene (99% Aldrich) and 1-methylcyclo-hexene (97% Aldrich) was carried out in round-bottom glass

Paper PCCP


batch reactors, immersed in an oil bath at 90 1C with magneticstirring (ca. 800 rpm) under an inert dry atmosphere. AllNb–SiO2 catalysts were pre-treated in air at 500 1C and cooledto room temperature under vacuum prior to use. Acetonitrile(Aldrich, HPLC grade) was used as solvent. Aqueous hydrogenperoxide (H2O2; 50%, Aldrich) was used as oxidant, with anoxidant to substrate molar ratio of 2 : 1. The total volume of themixture was 5 mL. Samples were analysed by GC (Agilent 6890Series; HP-5 column, 30 m � 0.25 mm; FID detector), withmesitylene (Fluka, puriss. Z99%) as the internal standard. GCpeaks were identified by comparison with the retention timesof genuine samples of reference standards, and by means ofGC-MS. Standard deviations of �2%, �4% and �2 h�1 areexpected for values of conversion, selectivity and specific activ-ity, respectively. Neither significant auto-oxidation nor support-catalyzed contributions to epoxidation were recorded withniobium-containing catalysts or the silica supports. After alltests, the presence of residual hydrogen peroxide was con-firmed by iodometric assay.

In order to check for possible leaching of Nb, the solidcatalysts were separated from the liquid mixtures by centrifuga-tion and the supernatant solutions were tested for reactivity.38

In tests of catalyst recovery, the solid was separated by filtrationand thoroughly washed with fresh acetonitrile and then methanol(Fluka, HPLC grade). The filtered solid was dried gently at 110 1C,weighed, activated again at 500 1C in dry air and reused in a newtest, as described above.

3. Results and discussion3.1. Monitoring the formation of Nb(V) sites on silica viagrafting

Each step in the liquid phase grafting of the bis(cyclo-pentadienyl)Nb(IV) precursor (Nb(Cp)2Cl2) on the silica surfacewas followed by DR UV-Vis-NIR spectroscopy. Fig. 1 shows theDR UV-Vis (A) and NIR (B) spectra of the non-ordered, Dav silicaafter grafting of Nb(Cp)2Cl2, prior to calcination.

In the UV-Vis region (Fig. 1A), the spectrum shows a broad andintense absorption at 240 nm, assigned to Nb–O charge-transfer

transitions, and a broad shoulder at ca. 270 nm.30 A weak bandcentred at 380 nm is also present. By analogy to the reportedspectrum of titanocene grafted onto mesoporous silica,39 thebands at longer wavelengths (270 and 380 nm) are ascribed tothe electronic transitions of the cyclopentadienyl ligands (Cp)bound to the metal centre.

The spectrum of the same material in the NIR region isreported in Fig. 1B. It is characterized by a weak, broad band at7300 cm�1, weak signals at 5940, 5800, and 5720 cm�1 and acomplex absorption at lower wavenumbers, in which two peaksat 4435 and 4350 cm�1 are clearly distinguishable. In addition aband at 5250 cm�1 is also observed and associated to thepresence of physisorbed water molecules. Based on previouswork, the weak, broad band at ca. 7300 cm�1 in the NIRspectrum indicates that almost all of the Si–OH groups of thesilica have reacted, either with the Nb complex, or with thetriethylamine used to promote the reaction between the surfacesilanols and the Nb complex, by forming Et3NH+Cl�.39 Theresidual intensity is due to silanols that are H-bonded totriethylamine and/or to vicinal RSiOH species. Indeed, thefreshly grafted Nb–silica materials contain some organic resi-dues which can derive from both triethylamine and triethylam-monium chloride (derived from the reaction with HClgenerated during the grafting reaction). The bands at 5800,5720, and 4350 cm�1 are the first overtones of the asymmetricand symmetric C–H stretching modes and the combination ofC–H deformation and stretching modes (both asymmetric andsymmetric) of organic species. Thorough rinsing of the solidwith chloroform, at the end of the synthesis procedure, was noteffective in completely removing these highly polar com-pounds, and they remained on the catalyst surface. The signalsat 5940 and 4435 cm�1 are the first overtone of the C–Hstretching mode and the combination of C–H deformationand stretching of the (Cp) ligands, respectively. A shoulderobserved at ca. 4500 cm�1 is ascribed to the combinationof O–H deformation and stretching modes (dOH + nOH) ofresidual SiOH species.40

3.2. Physico-chemical characterization of calcined,Nb(V)-containing silica catalysts

The textural properties of the silicas before and after Nbdeposition were investigated by N2 physisorption analysisat 77 K. The specific surface areas, estimated by the Brunauer–Emmett–Teller method (SBET), pore volumes and diameters,determined by NLDFT approach, are reported in Table 1.

For the purely siliceous SBA-15 support and the Nb–SBAmaterials prepared by liquid-phase and dry impregnation(Nb–SBA-liq and Nb–SBA-DI, respectively), the isotherms aretype IV according to the IUPAC classification, and have H1hysteresis loops representative of mesoporous cylindrical orrod-like pores (Fig. 2A). SBA-15 possesses a high specific surfacearea (760 m2 g�1), a pore volume of 0.94 cm3 g�1, and anaverage mesopore diameter of 8.5 nm (Fig. 2B).

A decrease in SBET was observed after Nb deposition via eitherliquid-phase grafting or dry impregnation. The latter led to amodest decrease in specific surface area, from 760 to 611 m2 g�1,

Fig. 1 DR-UV-Vis (A) and NIR (B) spectra collected at room temperature for bis(cyclo-pentadienyl)Nb(IV) grafted onto amorphous Dav silica via liquid-phase grafting.

PCCP Paper


whereas former gave a much larger reduction in specific surfacearea, to 437 m2 g�1. In addition, Nb–SBA-liq shows a significantdecrease in pore volume, from 0.94 to 0.79 cm3 g�1, indicatingpartial blocking of the silica pores (Fig. 2B) upon liquid-phaseimpregnation of Nb. This effect was not observed for Nb–SBA-DI,which showed a pore size distribution very similar to that ofpristine SBA-15. In this case, a large fraction of the Nb (probablyin form of Nb2O5 nano-aggregates; vide DR UV-Vis results) maybe located mainly on the external surfaces of the silica particles,instead of being inside the pores.

Different textural characteristics were observed for Nb-basednon-ordered, amorphous Dav silicas. The silica and its Nb-modifiedderivatives, prepared via both dry impregnation (Nb–Dav-DI) andliquid-phase grafting (Nb–Dav-liq), show isotherms of type IVfollowing the IUPAC classification, with H2 hysteresis loopstypical of solids with disordered structural pore size distribu-tions (Fig. 2A0). Nb deposition led to a reduction of the specificsurface area, from 580 m2 g�1 (for Dav SiO2) to 435 and 444 m2 g�1

for Nb–Dav-liq and Nb–Dav-DI, respectively (Table 1). Thepore size distribution in Dav SiO2 and the resulting catalystscover a wide range of values from 4.0 to 10.0 nm (Fig. 2B0), dueto the irregular geometry and morphology of the mesopores.

The deposition of Nb led to a reduction in pore volume, from0.97 cm3 g�1 for SiO2 to 0.78 and 0.79 cm3 g�1 for Nb–Dav-DIand Nb–Dav-liq, respectively (Table 1).

The coordination geometry and chemical environment ofthe metal sites dispersed on the different silica supports wereevaluated after calcination using DR UV-Vis spectroscopy(Fig. 3).

In general, all samples show an absorption maximum atca. 245 nm, assigned to O2� - Nb5+ charge-transfer bandsassociated with isolated Nb(V) centres (Fig. 3), although theattribution of this band remains controversial.32 In addition,the band appears to be asymmetric (this is mainly observed forthe Nb–Dav-DI sample), suggesting that Nb centres with differentcoordination geometries (e.g., tetrahedral and octahedral coordi-nation, promoted by coordination of two water molecules) couldbe present in the calcined Nb-containing materials. Furtherinterpretation of the DR UV-Vis results is informed by the XANESanalysis (see below). For all Nb–Dav materials, prepared via bothdry impregnation and liquid-phase grafting, no absorption bandswere detected above 400 nm, indicating that no large Nb2O5

domains are present on the catalyst surface (Fig. 3).33

Nevertheless, materials prepared by dry impregnation showa shoulder centred at 340 nm (most evident for Nb–SBA-DI),assigned to the presence of Nb2O5-like nano-aggregates formedduring the dry impregnation process.41

The nature of the aggregates responsible for the UV bandobserved at ca. 340 nm in the materials prepared by dry impreg-nation was investigated by XRD and Raman spectroscopy. TheX-ray diffractograms of Nb–Dav-DI and Nb–SBA-DI are reportedin Fig. 4, where they are compared with those of Nb–Dav-liq andNb–SBA-liq.

Both Nb-containing catalysts prepared via liquid-phasegrafting show X-ray diffraction patterns characterized by a wideband centred at 221 2y, typical of amorphous silica (Fig. 4,curves a and c). No peaks attributable to the presence ofcrystalline Nb2O5 aggregates (see the reference profile of bulkNb2O5, Fig. 4e) are detectable for these two samples, suggesting

Table 1 Textural features of the unmodified silicas and the Nb-based catalysts

SolidMetal loadinga

(wt%)SBET

b

(m2 g�1)PVc

(cm3 g�1) PDd (nm)

Dav SiO2 — 580 0.97 4.0–10.0SBA-15 — 760 0.94 8.5Nb–Dav-DI 1.75 444 0.78 4.0–10.0Nb–Dav-liq 0.76 435 0.79 4.0–10.0Nb–SBA-DI 1.72 611 0.93 8.0Nb–SBA-liq 1.56 437 0.79 8.0

a Nb content measured by ICP-MS. b BET specific surface area. c Specificpore volume. d Mean pore diameter.

Fig. 2 N2 adsorption–desorption isotherms at 77 K, for (A) SBA-15 (-J-),Nb–SBA-liq (-K-) and Nb–SBA-DI (-m-); and (A0) Dav SiO2 (-J-), Nb–Dav-liq(-K-) and Nb–Dav-DI (-m-). Pore size distributions obtained by NLDFT are shownfor (B) SBA-15 (-J-), Nb–SBA-liq (-K-) and Nb–SBA-DI (-m-); and (B0) Dav SiO2

(-J-), Nb–Dav-liq (-K-) and Nb–Dav-DI (-m-).

Fig. 3 DR UV-Vis spectra of Nb–Dav-liq (a), Nb–SBA-liq (b), Nb–SBA-DI (c) andNb–Dav-DI (d), all after calcination (but before catalytic testing).

Paper PCCP


that Nb ions are well-dispersed on the silica surface, in agree-ment with the DR UV-vis spectroscopic results. Differentfeatures are observed for the materials prepared by dry impreg-nation (Fig. 4b and d). In addition to the broad signal typical ofthe amorphous silica (centred at 221 2y), three weak reflectionsare visible at 22.7, 28.6 and 36.91. By comparison to thereflections of Nb2O5 nanoparticles with orthorhombic symme-try,42 and in agreement with our DR UV-Vis spectra, these peaksare ascribed to nanometer-sized Nb2O5 aggregates on the Davand SBA-15 silicas.

Raman spectroscopy was also used to investigate the natureof the metal oxide species present on the silica surfaces. Thespectra of Nb–Dav-DI and Nb–SBA-DI are shown in Fig. 5. Abulk Nb2O5 sample was also studied for comparison. Its Ramanspectrum (Fig. 5e) is characterized by a strong band centred atca. 700 cm1, associated with the stretching modes of differentniobia polyhedra, in agreement with our X-ray diffractionresults and as reported in the literature.42 This peak is completelyabsent in the Raman spectra of Nb–Dav-liq and Nb–Dav-15-liq(Fig. 5a and c), but is clearly visible (albeit with lower intensityrelative to bulk Nb2O5) in the Raman spectra of Nb-containingsilicas prepared by dry impregnation (Fig. 5b and d). The intensity

of this band is particularly pronounced for Nb–SBA-DI, whichcontains a larger amount of Nb2O5 aggregates compared toNb–Dav-DI (in agreement with our DR UV-Vis spectra, Fig. 3). Ingeneral, these results confirm that dry impregnation leads tothe formation of nano-sized Nb2O5 aggregates with ortho-rhombic structure on the surface of the silica. Such nano-domains are formed despite the high specific surface area ofthe support (i.e., they occur on Dav silica as well as on high-surface area SBA-15) and are likely due to partially physisorbedNb(Cp)2Cl2 domains that, upon calcination, generate nano-sized niobia aggregates. In contrast, liquid-phase graftingresults in better dispersion, since the niobocene is selectivelychemisorbed onto the silica surface and no physisorbedNb(Cp)2Cl2 domains are present after grafting (and before thefinal calcination) due to thorough washing with fresh chloro-form and dichloromethane.

3.3. X-ray absorption spectroscopic characterization

The normalized XANES of Nb2O5, Nb(Cp)2Cl2, Nb–Dav-DI andNb–Dav-liq are compared in Fig. 6. The spectrum of Nb–Dav-liqin the XANES region resembles that of bulk Nb2O5, but differsgreatly from that of the Nb(Cp)2Cl2 precursor. After calcinationin oxygen at 500 1C and stabilization in air, the oxidation stateof the silica-supported Nb is higher than in the nioboceneprecursor, as evidenced by the edge shift to higher energycomparable to the one of Nb2O5. The presence of a pre-edgepeak associated with 1s–nd transitions can be useful in deter-mining the geometry of transition metal ions.43 Here, theintensity of the pre-edge peak due to the 1s–4d transition ofNb discriminates between tetrahedral and octahedral geome-tries.44 When the geometry around the absorbing atom ismostly tetrahedral, the pre-edge peak is intense and distinct,while in octahedral geometry this feature is much lessprominent.

The weak pre-edge feature in the XANES region for Nb–Dav-liqis consistent with the majority of Nb sites having octahedral

Fig. 4 X-ray diffraction patterns of Nb–SBA-liq (a), Nb–SBA-DI (b), Nb–Dav-liq(c), Nb–Dav-DI (d) and bulk Nb2O5 (e).

Fig. 5 Raman spectra of Nb–SBA-liq (a), Nb–SBA-DI (b), Nb–Dav-liq (c), Nb–Dav-DI(d) and bulk Nb2O5 (e).

Fig. 6 Nb K-edge XANES of (a) Nb(Cp)2Cl2, (b) Nb2O5, (c) Nb–Dav-liq, and (d)Nb–Dav-DI.

PCCP Paper


geometry. As previously reported, a change occurs in the Nbcoordination state in Nb2O5 dispersed on silica and alumina,from tetrahedral to octahedral, upon adsorption of water mole-cules,24,45 while for dry Nb2O5 the weak pre-edge peak is due todistorted octahedral Nb sites. In Nb–Dav-liq, the coordination oftwo water molecules (derived from ambient moisture) has likelyconverted the surface Nb(V) sites from tetrahedral to octahedralgeometry, resulting in low pre-edge intensity.

There are only small differences between the XANES ofNb–Dav-liq and bulk Nb2O5, however, a slight change in thewhite-line shape is consistent with higher dispersion in thesupported sample.46 In contrast, the EXAFS clearly showsdifferences in the environment of the supported Nb sitesrelative to bulk Nb2O5, Fig. 7.

To analyse the EXAFS of the bulk material, a model wasconstructed from its crystal structure, which has several shortNb–O paths and long Nb–Nb paths. The two regions were firstanalyzed independently before optimizing the fit parameters inthe whole range. Two Nb–O paths, corresponding to bonddistances of 1.82 and 2.07 Å, resulted in a satisfactory fit inthe range from 1 to 2 Å, while a good fit in the range from 2.5 to4 Å was achieved using two different Nb–Nb paths (3.32 and3.78 Å). The results, shown in Table 2, are in good agreementwith those reported in the literature.45

In the case of Nb–Dav-liq, the region corresponding to theNb–O scattering paths is more complex and obtaining a good fitrequired three distinct Nb–O paths. Several attempts to fit twodifferent Nb–Nb paths, as in the case of Nb2O5, did not leadto good results. In Fig. 8, we show the best fit achieved forNb–Dav-liq. A Nb–Si path introduced into the model, togetherwith a longer Nb–Nb path, significantly improved the curve fit.

These results suggest that Nb in the calcined catalyst existsmainly in a pseudo-octahedral environment (N = 6.2), consis-tent with the XANES. The shortest Nb–O distance is in goodagreement with the presence of one terminal NbQO bond (1.77 Å).The longest Nb–O distance is probably associated with the coordi-nation of two water molecules (2.02 Å), while the intermediatedistance corresponds to bonding between Nb and oxygen ions ofthe support or bound to another Nb atom (1.90 Å). The twomaxima in the region 2.5–4.0 Å are contributed to by Nb–Si andNb–Nb scattering paths. The former arises from the interactions ofthe Nb ion with the oxygen of the silica support leading to welldispersed-Nb sites.47 The latter, with a lower coordination number(N = 1.3) and at a shorter distance (R = 3.60 Å) than in bulk Nb2O5

(N = 2.0; R = 3.78 Å) may be related to the presence of a minoramount of Nb–O–Nb oligomers on the silica support,45,48 ordimeric species formed upon adsorption of water (similar to thereaction of monomeric Nb species upon adsorption of ethanol49).

We also report the result of curve fitting the EXAFS data forNb–Dav-DI in Table 2. As in the case of the catalyst prepared byliquid-phase grafting, a good fit was obtained only by includinga Nb–Si scattering path in the model. The lower averagecoordination number for this scattering path may be the resultof a more heterogeneous coordination environment than wasthe case for Nb–Dav-liq, resulting from a lower dispersion of Nbsites and a higher amount of oligomeric Nb–O–Nb species.Even though the standard deviation values for N are relativelyhigh, this observation might be consistent with Nb2O5-likenanoparticles, as evidenced by XRD, Raman and DRUV-Vis.Likewise, the increase in the NbQO bond length, similar tothat of the first scattering path for bulk Nb2O5, together with adecrease in the number of coordinated water molecules, sug-gests a lower dispersion of Nb sites. A similar oxidation state asthe sample by liquid grafting was evidenced by analysis of theXANES region (Fig. 6).

3.4. Catalytic testing

Nb-grafted silica catalysts such as those described above findpractical application in the selective epoxidation of alkenes by

Fig. 7 Nb K-edge EXAFS of (a) Nb2O5, (b) Nb–Dav-liq and (c) Nb–Dav DI.

Table 2 EXAFS curve-fit results for Nb-containing silicas

Sample Bond R (Å)a Nb s2 (Å2)c

Nb2O5 Nb–O 1.82 (0.02) 0.85 (0.4) 0.004Nb–O 2.07 (0.01) 4.3 (0.3) 0.009Nb–Nb 3.32 (0.03) 1.2 (0.2) 0.008Nb–Nb 3.78 (0.05) 2.0 (0.5) 0.004

Nb–Dav-liq Nb–O 1.77 (0.04) 1.1 (0.3) 0.005Nb–O 1.90 (0.03) 3.1 (0.2) 0.004Nb–O 2.02 (0.05) 2.0 (0.3) 0.003Nb–Si 3.44 (0.07) 1.8 (0.2) 0.008Nb–Nb 3.60 (0.03) 1.3 (0.3) 0.006

Nb–Dav-DI Nb–O 1.82 (0.04) 1.1 (0.4) 0.003Nb–O 1.99 (0.02) 3.3 (0.5) 0.005Nb–O 2.16 (0.05) 1.4 (0.3) 0.003Nb–Si 3.44 (0.03) 1.5 (0.3) 0.008Nb–Nb 3.66 (0.02) 1.1 (0.3) 0.007

a Bond distance. b Coordination number. c Mean-square displacement.Fit range 1.0–4.0 Å. The values in parentheses show the errors for therelated parameters.

Fig. 8 EXAFS curve-fit for Nb–Dav-liq, in (a) k3-weighted k-space, and (b)k3-weighted R-space (solid black line: experimental data; broken grey line: curvefit).

Paper PCCP


aqueous hydrogen peroxide. Previously, this kind of catalystwas reported to be active and stable in the liquid-phase batchepoxidation of limonene (1-methyl-4-(prop-1-en-2-yl)cyclohex-1-ene)to limonene epoxide.31 Here, we directly compare activities for ourcatalysts in the epoxidation of two substrates, cyclohexene and1-methylcyclohexene. These two cyclic alkenes have different num-bers of substituents and therefore different electronic densities andsteric hindrance around the CQC bond.

All Nb–silica catalysts are active in the epoxidation of bothcyclohexene and 1-methylcyclohexene (Table 3). In general, thecatalysts derived from liquid-phase grafting showed higherspecific activities (i.e., moles of reactant converted per moleof Nb per unit time) than those prepared via dry impregnation.This behaviour is consistent with that observed for limoneneepoxidation,31 and is attributed to a higher fraction of isolatedsites and hence better availability of Nb(V) centres in catalystsprepared via liquid-phase grafting. Of the two cycloalkenes,1-methylcyclohexene is more readily epoxidised than cyclohexene.The tri-substituted double bond is more electron-rich, favouringtransfer of the electrophilic oxygen from hydrogen peroxidedespite the increased steric bulk.50–52

The selectivity to the epoxide is generally good or very goodat the beginning of the reaction (e.g., virtually quantitativeselectivity to cyclohexene epoxide at 36% conversion overNb–SBA-liq; Table 3). However, as the reaction proceeds, sinceNb(V) centres on the catalyst possess intrinsic acidic charactertoo, some epoxide undergoes acid-catalysed rearrangement orring-opening, and the epoxide yield decreases. Among thesecondary products obtained by subsequent acid-catalysedtransformations, cyclohexanone and methylcyclohexanone,cyclohexanediol and methylcyclohexanediol (derived fromcyclohexene oxide and methylcyclohexene oxide, respectively)are the most important. No appreciable yield of allylic oxidationproducts, such as cyclohexenol or methylcyclohexenol, wasdetected. This means that the oxidation reaction mainly followsa heterolytic pathway, rather than a homolytic, radical-basedone,53 and the presence of Nb sites is able to address thereaction towards a non-radical mechanism.

Finally, in terms of yield of the desired products, the mostinteresting catalyst appears to be Nb–Dav-liq, as observedpreviously in limonene epoxidation.31 This catalyst balancesgood alkene conversion and very good selectivity to the epoxide,

better than Nb–SBA-liq. The wide, non-ordered pore system ofDav silica (with mean pore sizes ranging from 4 to 10 nm) isless prone to blockage by carbonaceous deposits than theordered mesopore array of SBA-15. This interpretation is sup-ported by the amount of organic by-products found on thesurface of the spent catalyst by C,H,N elemental analysis. Theused, washed Nb–Dav-liq catalyst contained 3–4% carbon,compared to 7–8% for Nb–SBA-liq. In addition, small amountsof Nb–O–Nb oligomers, as revealed by EXAFS analysis, are notvery detrimental to the epoxidation reaction. Conversely, theycan show some epoxidation activity themselves and they do notcause extensive disproportionation of hydrogen peroxide (intests where H2O2 was not the limiting reagent).54

3.5. Generation of catalytically active sites

The experimental evidence described above confirms thatgrafting Nb(Cp)2Cl2 onto silica occurs in a manner comparableto the well-studied reaction of titanocene dichloride (Ti(Cp)2Cl2)with silica.35,55 The Nb complex undergoes nucleophilic attack bysilanol groups on the silica surface (Scheme 1). In this step, oneequivalent of cyclopentadiene and two equivalents of triethyl-ammonium chloride are produced and remain physisorbed (non-negligible amounts of nitrogen were indeed detected by elementalanalysis on the thoroughly washed grafted catalyst). The chemi-sorbed species is a tripodally-grafted Z5-cyclopentadienylniobiummoiety. The calcination step then leads to a four-coordinategrafted oxo-Nb(V) species, which, upon hydration, is convertedto a hexa-coordinate Nb(V) species. Prior to use, the catalyst can becalcined again in dry air at high temperature to remove the labilewater molecules and to obtain the coordinatively-unsaturatedpseudo-tetrahedral Nb site.

4. Conclusions

Deposition of niobocene dichloride onto mesoporous silicasfollowed by calcination gives rise to catalysts that are active andselective in the epoxidation of cycloalkenes by aqueous hydro-gen peroxide. Deposition can be achieved either by liquid-phasegrafting or by dry impregnation. The grafting of the organo-metallic precursor occurs via the formation of a cyclopentadie-nyl-niobium moiety on the silica surface. Combined DR UV-Vis,NIR, Raman, XRD, XANES and EXAFS analyses revealed the

Table 3 Catalytic performance of Nb–SiO2 in the liquid-phase epoxidation ofcyclohexene and 1-methylcyclohexene

Catalyst

Cyclohexene 1-Methylcyclohexene

Ca (%) Sb (%) SAc (h�1) Ca (%) Sb (%) SAc (h�1)

Nb–Dav-liq 53 65 71 94 87Nb–Dav-DI 53 94 28 59 85 31Nb–SBA-liq 73 78 44 83 61 49Nb–SBA-DI 36 498 20 69 69 37

Conditions: acetonitrile solvent; 100 mg catalyst; 1.0 mmol cycloalkene;2.0 mmol aq. H2O2; reflux temperature; batch reactor.a Cycloalkeneconversion. b Selectivity to cycloalkene epoxide. c SA: specific activity([molconverted alkene] [molNb h]�1). All the parameters were measuredafter 1 h of reaction.

Scheme 1

PCCP Paper


nature of the species present after calcination to be anoxo-Nb(V) site grafted onto the silica surface in close proximityto other Nb(V) centres and, especially for samples obtained bythe dry impregnation technique, to some Nb2O5-like smallaggregates. The liquid-phase methodology gives indeed thebest dispersion of the metal sites onto the silica support.

Both types of catalyst (obtained either via liquid-phasegrafting or via dry impregnation) are effective in the epoxida-tion of the model substrates cyclohexene and 1-methylcyclo-hexene. However, Nb–Dav-liq shows the best catalytic activity.

Some light has been shed on the nature of the active site.This class of grafted Nb–silica catalysts is now ready to be testedand used in the selective oxidation of a wide range of substratesthat are currently oxidised by conventional procedures relyingon economically or environmentally unfriendly technologies,such as the use of peracetic, performic or perbenzoic acids.

Acknowledgements

The authors would like to thank the Italian Ministry of Education,University and Research through the Project ‘‘ItalNanoNet’’ (ReteNazionale di Ricerca sulle Nanoscienze; Prot. No. RBPR05JH2P).A. G. is grateful to the US Department of Energy, Basic EnergySciences (DE-FG02-03ER15467) for his grant. Portions of thisresearch were carried out at the Stanford Synchrotron RadiationLightsource, a Directorate of SLAC National Accelerator Labora-tory and an Office of Science User Facility operated for the U.S.Department of Energy Office of Science by Stanford University.

Notes and references

1 R. A. Sheldon, M. Wallau, I. W. C. E. Arends andV. Schuchardt, Acc. Chem. Res., 1998, 31, 485.

2 M. Selvaraj, D.-W. Park, I. Kim, S. Kawi and C. S. Ha, DaltonTrans., 2012, 41, 9633.

3 M. Selvaraj, M. Kandaswamy, D. W. Park and C. S. Ha, Catal.Today, 2010, 158, 286.

4 M. Selvaraj, M. Kandaswamy, D. W. Park and C. S. Ha, Catal.Today, 2010, 158, 377.

5 M. Selvaraj and P. K. Sinha, New J. Chem., 2010, 34, 1921.6 C. Coperet, M. Chabanas, R. Petroff Saint-Arroman and

J. M. Basset, Angew. Chem., Int. Ed., 2003, 42, 156.7 J. M. Thomas, J. Chem. Phys., 2008, 128, 182502.8 D. Xuereb and R. Raja, Catal. Sci. Technol., 2011, 1, 517.9 V. Dal Santo, M. Guidotti, R. Psaro, L. Marchese, F. Carniato

and C. Bisio, Proc. R. Soc. A, 2012, 468, 1904.10 A. Corma, M. A. Camblor, P. Esteve, A. Martinez and

J. P. Perez Pariente, J. Catal., 1994, 145, 151.11 F. Carniato, C. Bisio, E. Boccaleri, M. Guidotti, E. Gavrilova

and L. Marchese, Chem.–Eur. J., 2008, 14, 8098.12 S. L. Barbosa, G. R. Hurtado, S. I. Klein, V. Jr. Lacerda, M. J.

Dabdoub and C. F. Guimaraes, Appl. Catal., A, 2008, 338, 9.13 V. S. Braga, I. C. L. Barros, F. A. C. Garcia, S. C. L. Dias and

J. A. Dias, Catal. Today, 2008, 133–135, 106.14 F. Quignard, A. Choplin and R. Teissier, J. Mol. Catal. A:

Chem., 1997, 120, L27.

15 E. Gianotti, M. E. Raimondi, L. Marchese, G. Martra,T. Maschmeyer, J. M. Seddon and S. Coluccia, Catal. Lett.,2001, 76, 21.

16 K. K. Kang and W. S. Ahn, J. Mol. Catal. A: Chem., 2000,159, 403.

17 W. Adam, W. Malisch, K. J. Roschmann, C. R. Saha-Moellerand W. A. Schenk, J. Organomet. Chem., 2002, 661, 3.

18 A. Held and P. Florczak, Catal. Today, 2009, 142, 329.19 Y. Liu, K. Murata and M. Inaba, Chem. Lett., 2003, 32, 992.20 F. Somma, P. Canton and G. Strukul, J. Catal., 2005, 229, 490.21 M. Trejda, A. Tuel, J. Kujawa, B. Kilos and M. Ziolek,

Microporous Mesoporous Mater., 2008, 110, 271.22 M. Selvaraj, S. Kawi, D. W. Park and C. S. Ha, J. Phys. Chem.

C, 2009, 113, 7743.23 M. Ziolek, P. Decyk, I. Sobczak, M. Trejda, J. Florek,

H. Golinska, W. Klimas and A. Wojtaszek, Appl. Catal., A,2011, 391, 194.

24 J. S. Reddy, P. Liu and A. Sayari, Appl. Catal., A, 1997,151, 409.

25 Y. Deng, C. Lettmann and W. F. Maier, Appl. Catal., A, 2001,214, 31.

26 J. V. Coelho, L. C. A. Oliveira, F. C. C. Moura, P. P. de Souza,C. A. Silva, K. Barros Batista and M. J. da Silva, Appl. Catal.,A, 2012, 419–420, 215.

27 M. Ziolek, Catal. Today, 2003, 78, 47.28 A. Feliczak, K. Walczak, A. Wawrzynczak and I. Nowak,

Catal. Today, 2009, 140, 23.29 N. Marin-Astorga, J. J. Martinez, G. Borda, J. Cubillos,

D. N. Suarez and H. Rojas, Top. Catal., 2012, 55, 620.30 C. Tiozzo, C. Bisio, F. Carniato, L. Marchese, A. Gallo,

N. Ravasio, R. Psaro and M. Guidotti, Eur. J. Lipid Sci.Technol., 2013, 115, 86.

31 A. Gallo, C. Tiozzo, R. Psaro, F. Carniato and M. Guidotti,J. Catal., 2013, 298, 77.

32 J. M. R. Gallo, H. O. Pastore and U. Schuchardt, J. Catal.,2006, 243, 57.

33 P. Carniti, A. Gervasini and M. Marzo, J. Phys. Chem. C,2008, 112, 14064.

34 D. Zhao, Q. Huo, J. Feng, B. F. Chmelka and G. D. Stucky,J. Am. Chem. Soc., 1998, 120, 6024.

35 T. Maschmeyer, F. Rey, G. Sankar and J. M. Thomas, Nature,1995, 378, 159.

36 M. Guidotti, N. Ravasio, R. Psaro, G. Ferraris and G. Moretti,J. Catal., 2003, 214(2), 242.

37 B. Ravel and M. J. J. Newville, Synchrotron Radiat., 2001,8, 322.

38 M. Guidotti and B. Lazaro, Deactivation of Molecular Sieves inthe Synthesis of Organic Chemicals, in Deactivation andregeneration of zeolite catalysts, ed. M. Guisnet andF. Ramoa-Ribeiro, Imperial College Press, London, 2011,pp. 303–334.

39 E. Gianotti, C. Bisio, L. Marchese, M. Guidotti, N. Ravasio,R. Psaro and S. Coluccia, J. Phys. Chem. C, 2007, 111, 5083.

40 L. Marchese, E. Gianotti, V. Dellarocca, T. Maschmeyer,F. Rey, S. Coluccia and J. M. Thomas, Phys. Chem. Chem.Phys., 1999, 1, 585.

Paper PCCP


41 J. M. R. Gallo, H. O. Pastore and U. Schuchardt, J. Non-Cryst.Solids, 2008, 354, 1648.

42 V. S. Braga, J. A. Dias, S. C. L. Dias and J. L. de Macedo,Chem. Mater., 2005, 17, 690.

43 J. A. Horsley, I. E. Wachs, J. M. Brown, G. H. Via and F. D.Hardcastle, J. Phys. Chem., 1987, 91, 4014.

44 S. Yoshida, T. Tanaka, T. Hanada, T. Hiraiwa, H. Kanai andT. Funabiki, Catal. Lett., 1992, 12, 277.

45 H. Yoshida, T. Tanaka, T. Yoshida, T. Funabiki andS. Yoshida, Catal. Today, 1996, 28, 79.

46 T. Tanaka, T. Yoshida, H. Yoshida, H. Aritani, T. Funabiki,S. Yoshida, J. M. Jehng and I. E. Wachs, Catal. Today, 1996,28, 71.

47 K. Asakura and Y. Iwasawa, J. Phys. Chem., 1991, 95, 1711.

48 N. Ichikuni and Y. Iwasawa, Catal. Today, 1993, 16, 427.49 N. Ichikuni and Y. Iwasawa, J. Phys. Chem., 1994, 98, 11576.50 F. Somma, A. Puppinato and G. Strukul, Appl. Catal., A,

2006, 309, 115.51 C. Cativiela, J. M. Fraile, J. I. Garcia and J. A. Mayoral, J. Mol.

Catal. A: Chem., 1996, 112, 259.52 C. Berlini, M. Guidotti, G. Moretti, R. Psaro and N. Ravasio,

Catal. Today, 2000, 60, 219.53 J. M. Fraile, J. I. Garcia, J. A. Mayoral and E. Vispe, J. Catal.,

2001, 204, 146.54 H. Shima, M. Tanaka, H. Imai, T. Yokoi, T. Tatsumi and

J. N. Kondo, J. Phys. Chem. C, 2009, 113, 21693.55 C. R. A. Catlow, S. A. French, A. A. Sokol and J. M. Thomas,

Philos. Trans. R. Soc. London, Ser. A, 2005, 363, 913.

PCCP Paper

Niobium-silica catalysts for the selective epoxidation of cyclic alkenes: the generation of the...

Documents

Transcript of Niobium-silica catalysts for the selective epoxidation of cyclic alkenes: the generation of the...