Comparison of the conventional impregnation method using ammonium heptamolybdate with a simple route to silica-supported molybdenum(VI) materials

Catalysis Communications 8 (2007) 1447–1451 www.elsevier.com/locate/catcom

Comparison of the conventional impregnation method using ammonium heptamolybdate with a simple route to silica-supported molybdenum(VI) materials Paul Ce´lestin Bakala a, Emmanuel Briot a,*, Jean-Yves Piquemal b, Jean-Marie Bre´geault a,*, Patricia Beaunier c a

Syste`mes Interfaciaux a` l’Echelle Nanome´trique, Universite´ Pierre et Marie Curie/CNRS, Case 196, 4 Place Jussieu, 75252 Paris Cedex 05, France b ITODYS, Universite´ Denis Diderot/CNRS, Case 7090, 2 Place Jussieu, 75251 Paris Cedex 05, France c Re´activite´ de Surface, UPMC/CNRS, 4 Place Jussieu, 75252 Paris Cedex 05, France Received 6 September 2006; received in revised form 15 December 2006; accepted 19 December 2006 Available online 23 December 2006

Abstract Silica-supported molybdenum oxide materials have been prepared by using the ability of molybdenum oxoperoxo complexes to interact with the surface hydroxyl groups of the support, which after calcination, have a highly dispersed MoOx/SiO2 structure, and lead to leach-resistant catalysts; with a polyoxo precursor, (NH4)6[Mo7O24] Æ 4H2O, there is a formation of MoO3 clusters and a weaker MoO3/ SiO2 interaction.  2006 Elsevier B.V. All rights reserved. Keywords: Epoxidation; Molybdenum; Mesoporous materials; H2O2

1. Introduction The field of application of nanotechnology for the development of catalysts for ‘‘Green Chemistry’’ is likely to grow rapidly during the next decade [1,2]. Surprisingly, there has been little work in which MCM-41, SBA-n, etc. have been exploited as replacability for classical industrial supports such as silicas, aluminas, silica-aluminas, etc. Are the catalytic performances of mesoporous materials superior to those of analogues obtained by transposition or conventionally prepared? Obviously, many problems concerning the catalytic target processes have to be considered [3] particularly economic factors related to: (i) the use of expensive inorganic complexes such as metal and/or silicon alkoxides and surfactants, (ii) lengthy hydrothermal syn-

*

Corresponding authors. Tel.: +33 1 44 27 36 27; fax: +33 1 44 27 36 35. E-mail addresses: [email protected] (E. Briot), bregeaul@ccr. jussieu.fr (J.-M. Bre´geault). 1566-7367/$ - see front matter  2006 Elsevier B.V. All rights reserved. doi:10.1016/j.catcom.2006.12.015

theses, (iii) template removal, (iv) poor attrition strength for industrial-scale reactors, etc. From studies on systems of aqueous [MoO4]2 or [Mo7O24]6 solutions and silicas, it is well known that molybdenum(VI) uptake by silica is relatively low over the entire pH range, except for a small increase at pH  2 or lower, owing to the formation of [SiMo12O40]4 ions, which commonly desorb into solution [4–6]. Silanols such as R3SiOH, R2Si(OH)2 and RSi(OH)3 are easily prepared and used in synthesis. The isolation of mono- or dimeric oxoperoxo compounds: [Ph3SiOMO(O2)2],[Ph2SiO2{M2O2 2 (l-O2)2(O2)2}] and of [{Ph2SiOMO(O2)2}2(l-O)]2, where M = Mo or W [7,8], led us to think that molybdenum (or tungsten) oxoperoxo species may be more suitable than [MoO4]2 or [Mo7O24]6 anions for preparing heterogeneous-Mo (or W) silica-supported catalysts. In a preceding paper [9], a simple and low-cost synthesis route was proposed to develop highly dispersed molybdenum(VI)-silica-based materials, via the so-called ‘‘oxoperoxo route’’ with transient formation of low nuclearity

P. Ce´lestin Bakala et al. / Catalysis Communications 8 (2007) 1447–1451

1448

moieties such as ”Si–O–Mo(OH)(O2)2. The g2-peroxo ligands are good leaving groups; they were decomposed during thermal treatments to give oxo groups with a nearly regular distribution of the surface species such as ‘‘(”Si– O)2–Mo(O)2’’ [10]. We showed, once a catalytic reaction has been proven using mesoporous materials (MCM-41, SBA-15), it can be developed on cheaper silica-based supports. In this note, for the first time, this ‘‘oxoperoxo route’’ is compared with the ‘‘traditional route’’ involving ammonium heptamolydate, (NH4)6[Mo7O24] Æ 4H2O. The reported detailed characterization of the materials by Raman, TEM and EDX and catalysis tests shows the interest of these techniques for checking such materials. 2. Experimental Before impregnation, SiO2 (beads or precipitated silica) were calcined in air (180 cm3 min1, 1 K min1) from ambient to 820 K (samples maintained at the final temperature for 6 h). Oxoperoxo route: wet impregnation was performed by suspending 0.950 g of the silica in a peroxidic solution of Mo(VI) initially prepared by mixing 0.060 g MoO3, 3 mL H2O and 0.5 mL H2O2 30 wt.% (entries 1, 2; Table 1) or 0.135 g MoO3, 3 mL H2O and 3 mL H2O2 30 wt.% (entry 3), stirred 30 min at 333 K in both cases, to generate the peroxo complexes. The suspension was stirred for 4 h at room temperature. The yellow solid was filtered off on a fritted glass disk, as such quickly dried over P4O10, then calcined in air (150 cm3 min1, 1 K min1) from ambient to 820 K. The solid (MoOx/ SiO2) is maintained at the final temperature for 6 h to favor the anchoring of molybdenum(VI) by ‘‘”Si–O–Mo’’ bonds. Polyoxo route: 0.058 g of (NH4)6[Mo7O24] Æ 4H2O (entry 4) or 0.098 g (entry 5) was dissolved in 3 mL H2O under stirring at 333 K. After cooling at room temperature 0.600 g (entry 4) or 0.950 g (entry 5) of the precipitated silica is added to the solution and stirred 4 h, the rest of the synthesis being identical to that for MoOx/SiO2. Samples synthesized according to the polyoxo route are denoted MoO3/SiO2. The silica is mainly prepared by precipitation (pH < 7) from a sodium silicate solution. This is followed by agglomeration of silica particles in more or less loose

aggregates in aqueous medium (sol). The precipitate is recovered, washed with distilled water and dried. See Rhoˆne-Poulenc Pat. FR 871 5275, 4 Nov. 1987. Raman spectra were recorded on a Kaiser Hololab 5000 R (excitation line, 785 nm; power source, 5–20 mW; 2–100 scans; 10 s per scan; 0.3 cm1 resolution). Elemental analyses were carried out at the Service Central d’Analyse (CNRS/Lyon) by inductive coupling plasma atomic emission spectroscopy (ICP-AES) after alkaline fusion with Li2B4O7. Diffuse reflectance IR spectra (resolution: 2 cm1) were taken on a Fourier transform apparatus (Bruker Vector 22) equipped with a Harrick diffuse reflection attachment. Adsorption and desorption isotherms for nitrogen were obtained at 77 K using a Micromeritics ASAP 2010. The samples were outgassed at 393 K and 0.1 Pa for 12 h before measurements. Transmission electron microscopy (TEM) and energy dispersive X-ray analysis (EDX) were performed using a JEOL JEM 2010 transmission electron microscope operating at 200 kV, equipped with a PGT Imix-PC system. The X-rays (Si Ka at 1.74 keV and Mo Ka at 17.48 keV) emitted from the surface (spot beam analysis) were collected in the 0– 20 keV range. Data do not include oxygen. Catalysis experiments were performed at room temperature (18–20 C); olefin (6–10 mmol) in 7 mL of pentane, catalyst (Mo/olefin = 1/100 mol/mol), 1.5 mL of 10% anhydrous t-BuOOH in decane (9 mmol of oxidant added with vigorous stirring) were placed in a home-made reactor avoiding direct contact between the catalyst and the magnetic stirring bar. The progress of the reaction was monitored by GC and the products were analysed after 24 h and quenching with MnO2. 3. Results and discussion Precipitated silica or beads (BET specific surface area . 143–223 m2 g1) and yellow molybdenum oxoperoxo species (mainly [MoO(O2)2(H2O)2] and [{MoO(O2)2(H2O)}2(l-O)]2) interact in aqueous acidic medium to form fairly stable surface oxoperoxo species (Fig. 1), with a characteristic IR band ~mOO near 870 cm1 (~mOO is usually expected to be at 830–885 cm1) [11]. In Table 1 are presented the physicochemical characterizations of pre-cat-

Table 1 Textural properties of MoOx/SiO2 or MoO3/SiO2 pre-catalysts (calcined samples after wet impregnation and drying) Entry 1 2 3 4 5

Samples MoOx/SiO2 (beads) MoOx/SiO2 (crushed beads) MoOx/SiO2 (precipitated)d MoO3/SiO2 (precipitated)d MoO3/SiO2 (precipitated)d

a b c d e

Synthesis route Oxoperoxo Oxoperoxo Oxoperoxo Polyoxoe Polyoxoe

SBETa (m2 g1) 217 223 103 123 120

b

(219) (223)b (143)b (143)b (143)b

Pore volume (cm3 g1) 1.09 1.10 0.65 0.80 0.80

b

(1.21) (1.10)b (0.84)b (0.84)b (0.84)b

Calculated with the BET model (±5 m2 g1). The corresponding values for the bare (unused) supports studied. Expected ratio of starting products. Precipitated hydroxylated silica: the density of silanol groups is about 2 or 3 Si–OH/nm2. Conventional impregnation method with (NH4)6[Mo7O24] Æ 4H2O (Section 2).

Mo (wt.%) Si (wt.%)

Si/Mo (mol/mol)

1.8 1.8 5.5 0.65 2.3

84 (50)c 84 (50)c 25 (25)c 230 (33)c 61.5 (28)c

44.1 44.1 40.1 43.5 41.5

P. Ce´lestin Bakala et al. / Catalysis Communications 8 (2007) 1447–1451

gen, as well as the molybdenum. These factors complicate the assignments of Raman bands. The Raman spectra of two thermally treated silicas with supported molybdenum(VI) oxo species (Figs. 2a–c), prepared by the oxoperoxo route, show bands which have no reference to MoO3, (NH4)6[Mo7O24] Æ 4H2O and H4[SiMo12O40] Æ aq. (Figs. 2e– g). Bands of the silica matrix are at 499–506 cm1 (associated also with Q3 species), 609–615 cm1 (attributed to strained trisiloxane rings); Si–O stretches appear at 890– 920 cm1, assigned to symmetrical stretching of the Si–O bonds of geminal silanols [10,12,13] and to the siloxane bridges (Si–O–Si). The Raman signal at 960–970 cm1 may be attributed to the symmetric stretching mode of terminal Mo@O bonds [10,12,13] with a possible contribution of Si–O stretching of SiOH  OMo ‘‘defective sites’’ and of ”Si–O–Mo moieties. (A shoulder at 1015 cm1 is tentatively assigned to Mo@O vibrations consistent with the shortest Mo@O bond length.) The spectrum (Fig. 2a) clearly indicates that MoO3 clusters are not formed in the calcined products prepared via the oxoperoxo route. On the other hand, the conventional impregnation method, using the heptamolybdate anion (Fig. 2d), leads to crystalline MoO3 [14–18] (bands at 994–997, 818–820, 666–674,

~ν

F(R), K.M.

o-o

870 cm

1449

-1

(a)

(b)

890

615

700

609

800

506

900

499

1000

381

970

(c)

a

881

1015

960

b

608

c 674

997

820

500 293

Intensity (a.u.)

994

818

d

666

290

910

847

e

f

640

970

990

704

337 379

245 252

alysts obtained after thermal treatment in air. For MoOx/ SiO2 (beads or crushed beads) and MoOx/SiO2 (precipitated), the slight decrease in specific surface area (SBET) and in pore volume (Vp) is more pronounced with high Mo content (entries 1–3); this is consistent with the initial grafting of Mo oxoperoxo species on the silica surface with the assumption of 2 or 3 OH/nm2 on SiO2 (initial Si/ Mo = 50; Entries 1, 2). It appears that for precipitated silica and for Si/Mo = 25 (entry 3), the pronounced decrease in SBET and Vp may be due to textural modifications of the support or to a slight molybdenum overloading. The overall data and the Si/Mo ratios show that by wet impregnation, anchoring of the molybdenum low-nuclearity oxoperoxo species is more easily controlled than by the polyoxo route with the bulky anion [Mo7O24]6 (entries 4, 5 – Table 1). Raman spectrometry is very useful for the characterization of anchored species formed before and after calcination in air [12,13]. However, the ~m1 stretching region for MoO4 tetrahedra or MoO6 octahedra significantly overlap with the nominal Mo@O stretching region (900– 1040 cm1). The vibrational data for the Mo@O stretch depend on the coordination of chemical species to the oxy-

824

Fig. 1. Comparison of diffuse reflectance IR spectra (Kubelka–Munk curves) at 293 K. (a) Silica (crushed beads) with anchored molybdenum oxoperoxo species; (b) pure silica; (c) difference showing the ~mOO vibration.

961

wavenumbers, cm-1

g 200

400

600

800

1000

1200

-1

Raman shift (cm ) Fig. 2. Raman spectra of calcined samples. (a) MoOx/SiO2 (precipitated); (b) MoOx/SiO2 (crushed beads); (c) MoOx/SiO2 (beads); (d) MoO3/SiO2 (polyoxo route with (NH4)6[Mo7O24] Æ 4H2O); (e) MoO3/SiO2 (mechanical mixture); (f) (NH4)6[Mo7O24] Æ 4H2O; (g) H4[SiMo12O40] Æ aq.

P. Ce´lestin Bakala et al. / Catalysis Communications 8 (2007) 1447–1451

1450

290–293 cm1), in addition to a weak band at 970 cm1 which is tentatively attributed to Mo@O bonds of twodimensional MoOx species. These results show that it is possible to achieve highly dispersed MoOx species in MoOx/SiO2 pre-catalysts by a very simple method with ‘‘Mo(VI) oxoperoxo complexes/H2O2–H2O/silica’’ systems, thus avoiding the need for expensive precursors and thermal treatments releasing acutely toxic gas. These results correlate with data obtained by diffuse reflectance UV–vis spectrometry on identically calcined samples protected from water vapour. The absorption edge is clearly shifted towards high-energy maxima (not shown here) for samples corresponding to Figs. 2a, b, while with MoO3 clusters on silica, the absorption edge is clearly shifted towards higher wavelengths (k > 400 nm). To ascertain whether the MoOx/SiO2 surface had a nearly homogeneous distribution of the molybdenum centres or not, the morphology of the particles was inspected by transmission electron microscopy (TEM) and by energy-dispersive X-ray (EDX) analyses. The composition is semi-quantitatively determined without calibration curves. Typical EDX spectra (Fig. 3 and supplementary materials, Tables 3a and b) of samples prepared by the polyoxo route with ammonium heptamolybdate and by the oxoperoxo route show signals of Si, Mo and Cu (from

Si Mo O C

Mo

Cu

(I)

Cu Si

0.0 O C

Cu

0.0

10.0

Mo

20.0

( II)

Cu

Mo 20.0

10.0

Si Cu

CO

Cu

0.0

(III)

Mo

Mo keV

20.0

10.0

Fig. 3. Energy-dispersive X-ray analyses (typical EDX patterns) of samples after calcinations: (I) and (II) MoO3/SiO2 (precipitated, polyoxo route); (III) MoOx/SiO2 (precipitated, oxoperoxo route).

the copper grids). The Mo signals reveal that the molybdenum distribution in the sample prepared by the oxoperoxo route is quasi-uniform over the SiO2 surface with (Mo/ Si)  0.13 and r = 0.02. On the other hand, samples prepared by the conventional impregnation method have a very inhomogeneous Mo distribution with MoO3 nanodomains and some zones are uncovered (supplementary materials Tables 3a and 3b) list the results of EDX experiments used to locate and quantify submicroscopic distributions of the Mo species.) These results give further support for a homogeneous distribution of the Mo sites in calcined ‘‘oxoperoxo samples’’ (quasi-ordered nanostructures on the silica support), while in a sample prepared with (NH4)6[Mo7O24] Æ 4H2O, there is a less ordered distribution between two phases with predominantly MoO3 clusters and thus phase segregation. They are in agreement with the vibrational results and our initial proposals. These Mo(VI) oxides/SiO2 materials are differentiated with simple catalysis tests: liquid-phase terpene epoxidation at room temperature. The catalytic epoxidation of (R)-(+)limonene has been the subject of many articles. Most of the publications indicate that a mixture of the mono- and di-epoxides is formed. The selective identification of a 1:1 mixture of cis/trans 1,2-epoxide was obtained at 4 C (2 h) with 94% conversion and 93% selectivity using methylrhenium oxide bis-picoline catalyst [(CH3)ReVO(pic)2] [19]. The addition of ‘‘proton sponges’’ (e.g. bipyridine or preferably 2,2 0 -bipyridyl-N,N 0 -dioxide, etc.) to ‘‘MeReO3 (MTO)/H2O2–H2O/organic solvent’’ also gives very high yields, as do tungsten(VI) catalysts (heterogeneous or phase-transfer systems) [20]. The results (Table 2) show that MoOx or MoO3/silica are less active and selective than Re(V), Re(VII) or W(VI) precursors (vide supra), the cis isomer being favoured with the molybdenum systems. MoOx/precipitated silica presents promising catalytic activity and very low leaching of active species (Table 2, run 4). For example, MoOx/SiO2 (precipitated) losses 9– 27 ppm of Mo in the first run and less than 2 ppm in a second run. The third run displays approximately the same conversion (P70%) and only traces of Mo are detected in the solution (below the detection limit, <1 ppm). The precursors (silica support, Mo(VI) complexes, run 5) and also the thermal treatments play an important role in determining the structure of Mo(VI) oxides/SiO2 catalysts and the point that no detectable Mo leaches.

Table 2 Limonene epoxidation over MoOx/SiO2 or MoO3/SiO2 catalysts after 24 h Run

Catalyst

Conv. (%)

Selec. for mono epoxides (%)

Selec. for diepoxides + diols (%)

cis/trans ratio

Mo lossa (ppm)

Mo loss (mol%)

1 2 3 4 5

MoOx/SiO2 (beads) MoOx/SiO2 (crushed beads) MoOx/SiO2 (precipitated) MoOx/SiO2 (precipitated)b MoO3/SiO2 (precipitated) 2(Si/Mo = 61.5)

40 94 66 71 49

83 86 84 92 93

17 14 10 + 6 5+3 7, no diol

3.0 3.4 2.5 1.8 4.6

38 >50 9–27 <1–2 20–40

15 – 1–3 <0.1–0.2 5–10

a b

Mo wt.% in the liquid phase at the end of the reaction with %Si in the solution in the 20–75 ppm range. Data of the second and third catalysis runs (first run: 3).

P. Ce´lestin Bakala et al. / Catalysis Communications 8 (2007) 1447–1451

In summary, with the oxoperoxo route at lower surface densities (<2–3 Mo/nm2), only two-dimensional MoOx monomers (or oligomers) are detected by Raman spectrometry, without any evidence for bands corresponding to three-dimensional MoO3 clusters (818 and 994 cm1) which are inevitably formed by the polyoxo route based on [Mo7O24]6 ions. With MoOx/SiO2, a small amount of molybdenum is leached during the first catalysis run, but at the second run these silica-supported molybdenum catalysts are effective for catalytic oxidations, even in the presence of nucleophilic reagent. The ease of preparation and the low cost of the materials make them good candidates, not only for oxidation reactions, but also for other uses such as heterogeneous catalysis with gases, under study in our group. Acknowledgements We thank Dr. F. Villain for Raman spectrometry facilities and Dr. J. Lomas for correcting the manuscript. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.catcom. 2006.12.015. References [1] S.L. Scott, C.M. Crudden, C.W. Jones (Eds.), Nanostructured Catalysts, Kluwer, Academic/Plenum Publishers, New York, 2003.

1451

[2] J. Clark, D. Macquarrie (Eds.), Handbook of Green Chemistry and Technology, Blackwell Publishing, Oxford, 2002. [3] J.-M. Bre´geault, Dalton Trans. (2003) 3289. [4] H. Jeziorowski, H. Kno¨zinger, J. Phys. Chem. 83 (1979) 1166. [5] H. Kno¨zinger, Proceedings of the Ninth International Congress on Catalysis, Calgary, Canada 5 (1988) 20. [6] R. Thouvenot, M. Fournier, C. Rocchiccioli-Deltcheft, J. Chem. Soc., Faraday Trans. 87 (1991) 2829. [7] J.-Y. Piquemal, C. Bois, J.-M. Bre´geault, Chem. Commun. (1997) 473. [8] J.-Y. Piquemal, S. Halut, J.-M. Bre´geault, Angew Chem., Int. Ed. Engl. 37 (1998) 1146. [9] P.C. Bakala, E. Briot, L. Salles, J.M. Bre´geault, Appl. Catal. A: Gen. 300 (2006) 91. [10] J. Jarupatrakorn, M.P. Coles, T.D. Tilley, Chem. Mater. 17 (2005) 1818. [11] J.-Y. Piquemal, L. Salles, G. Chottard, P. Herson, C. Ahcine, J.-M. Bre´geault, Eur. J. Inor. Chem. (2006) 939. [12] E. Briot, J.-Y. Piquemal, M. Vennat, J.-M. Bre´geault, G. Chottard, J.-M. Manoli, J. Mater. Chem. 10 (2000) 953. [13] J.-Y. Piquemal, E. Briot, G. Chottard, P. Tougne, J.-M. Manoli, J.M. Bre´geault, Micropor. Mesopor. Mater. 58 (2003) 279. [14] Y. Zhoulan, L. Xinkai, C. Qiyuan, Thermochim. Acta 352-353 (2000) 107. [15] R. Murugan, H. Chang, J. Chem. Soc., Dalton Trans. (2001) 3125. [16] M. Seman, J.N. Kondo, K. Domen, R. Radhakrishnan, S.T. Oyama, J. Phys. Chem. B 106 (2002) 12965. [17] X. Ma, J. Gong, X. Yang, S. Wang, Appl. Catal. A: Gen. 280 (2005) 215. [18] O. Collart, P. Van Der Voort, E.F. Vansant, E. Gustin, A. Bouwen, D. Schoemaker, R. Ramachandra Rao, B.M. Weckhuysen, R.A. Schoonheydt, Phys. Chem. Chem. Phys. 1 (1999) 4099. [19] A. Deloffre, S. Halut, L. Salles, J.-M. Bre´geault, J. Ribeiro Gregorio, B. Denise, H. Rudler, J. Chem. Soc., Dalton Trans. (1999) 2897. [20] L. Salles, J.-M. Bre´geault, R. Thouvenot, C.R. Acad. Sci. Paris, Se´rie IIc, Chemistry 3 (2000) 183.