Epoxidation of biodiesel with hydrogen peroxide over Ti-containing silicate catalysts

Microporous and Mesoporous Materials 164 (2012) 182–189

Contents lists available at SciVerse ScienceDirect

Microporous and Mesoporous Materials journal homepage: www.elsevier.com/locate/micromeso

Epoxidation of biodiesel with hydrogen peroxide over Ti-containing silicate catalysts Nicole Wilde, Christian Worch, Wladimir Suprun, Roger Gläser ⇑ Institute of Chemical Technology, Universität Leipzig, Linnéstr. 3, 04103 Leipzig, Germany

a r t i c l e

i n f o

Article history: Available online 6 July 2012 Dedicated to Prof. Dr. Ing. Jens Weitkamp on the occasion of his 70th birthday Keywords: FAME Epoxidation Hydrogen peroxide TS-1 Microwave-assisted synthesis

a b s t r a c t The heterogeneously catalyzed epoxidation of methyl oleate with hydrogen peroxide in the liquid phase is reported using different catalysts such as TS-1, Ti-MCM-41, TiOx–SiO2, and MOx and WOx supported on Al2O3 or SiO2. At 323 K, in acetonitrile as the solvent and with an industrial TS-1 catalyst, an epoxide selectivity of 87% at 93% conversion is achieved after 24 h. Variation of the catalyst mass, particles size and the reaction temperature proove that the conversion is limited by mass transport to the outer crystal surface of the catalyst. With TS-1 as the catalyst, also the unsaturated fatty acid methyl esters (FAME) in biodiesel can be epoxidized with a conversion of 90% and selectivity of 76%. A higher TON than over TiMCM-41 or even the industrial TS-1 catalysts is reached, when TS-1 with nanoscaled particles and stacked morphology is applied as the epoxidation catalyst. The latter was synthesized by microwaveassisted synthesis and possesses Ti sites in similar coordination geometry as conventional TS-1 as shown by DR-UV–Vis spectroscopy. Ó 2012 Elsevier Inc. All rights reserved.

1. Introduction The current chemical industry has to respond to a steadily increasing demand of synthetic products on the one hand and the shortage of fossil and mineral raw materials as well as their decreasing quality on the other hand [1,2]. Therefore, renewable feedstocks have recently gained considerable interest as raw materials for chemical production. By far the largest share of utilized renewables in the chemical industry is held by fats and oils [3]. Among the applications of fats and oils, their conversion to fatty acid methyl esters (FAME) for biodiesel production is one of most prominent. FAME can be further converted to epoxidized fatty acid esters which play an important role for a broad range of large-scale industrial synthesis of chemicals and intermediates such as plasticizers and stabilizers in PVC, intermediates in the production of polyurethane polyols, components for lubricants, cosmetics or pharmaceuticals [4–6]. Currently, epoxy fatty acid compounds are mainly obtained on the industrial scale by the Prileshajew reaction, in which the unsaturated oils are converted with percarboxylic acids, such as peracetic or performic acid. This route suffers from several drawbacks: (I) in the acidic reaction media, the selectivity for epoxides is relatively low due to oxirane ring opening, (II) the handling of peracids and highly concentrated hydrogen peroxide solutions is strongly

⇑ Corresponding author. E-mail address: [email protected] (R. Gläser). 1387-1811/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.micromeso.2012.06.047

hazardous, and (III) the aqueous solutions of C1–C3 carboxylic acids formed as by products are strongly corrosive [7,8]. In view of the principles of green chemistry, catalyzed epoxidations are the preferred alternatives to conventional stoichiometric epoxidation reactions such as the Prileshajew reaction. Catalysts based on titanium, tungsten and molybdenum were investigated for epoxidations of unsaturated olefins, with high yields and/or selectivity of the target products [9–12]. Molybdenum complexes or oxides on silica and alumina supports were extensively studied in the epoxidation of cyclooctene and limonene as well as allyl alcohol [13–15]. For the epoxidation of fatty acids esters, the tungsten-based active compound ‘‘tetrakis’’, i.e., ([(C8H17)3NCH3]3+ [PO4[W(O)(O2)2]4]3 ) was reported [16]. In the presence of H2O2, acid reaction conditions result favoring ring opening of the epoxide formed. In several recent studies, catalytic epoxidations of FAME were reported using organic hydroperoxides, such as tert-butyl- or cumene hydroperoxide, as epoxidizing agents [17–19]. Typically, TiO2–SiO2 and Ti-MCM-41 were used as catalysts. However, hydrogen peroxide would be an economically and environmentally largely preferred oxidant. In fact, MCM-41 and MCM-48 with Ti sites grafted onto the surface were found to be efficient catalysts in the epoxidation of methyl oleate with aqueous hydrogen peroxide [20]. Conversions up to 96% and yields of the methyl epoxystearate up to 91% were obtained over Ti/MCM-41 after 24 h at 358 K in liquid acetonitrile. Suarez et al. [21] and Sepulveda et al. [22] reported on the performance of different aluminas as catalysts for the epoxidation of fatty acid methyl esters with anhydrous and

183

N. Wilde et al. / Microporous and Mesoporous Materials 164 (2012) 182–189

aqueous hydrogen peroxide. Over a sol–gel derived alumina, methyl oleate and soy bean oil methyl ester conversion of 95% and epoxide selectivities >95% were achieved after 24 h at 353 K in ethyl acetate as a solvent [22]. Titanosilicate molecular sieves, especially titanium silicalite-1 (TS-1), were widely studied for the selective epoxidation of a variety of organic substrates using aqueous H2O2 as the sole oxidant [23–26]. Often high selectivities up to 94% at essentially complete conversion are achieved. The present work was, thus, devoted to investigate the heterogeneously catalyzed epoxidation of unsaturated fatty acid methyl esters with aqueous hydrogen peroxide as the sole oxidant over solid catalysts. In particular, the catalytic activity, selectivity and stability of different supported transition metal oxide catalysts, e.g., TiOx, MoOx or WOx on SiO2 or Al2O3, were studied and compared to that over a commercial titanium silicalite-1 (TS-1) and Ti-MCM-41 with methyl oleate as a model substrate. After a screening of the catalysts, the influence of the solvent and the reaction temperature was studied. It was also of interest whether the catalysts can be reused and if a deactivation takes place. An additional goal was to clarify, whether the conversion of the rather bulky FAME molecules is limited by mass transport processes and, if yes, whether strategies for an improved catalyst design can be derived. Finally, it was investigated whether the knowledge obtained for methyl oleate can be transferred to the conversion of FAME from ‘‘real’’ biodiesel with aqueous hydrogen peroxide over Ti-containing silica catalysts.

heating rate of 10 K min 1, i.e., 2 h at 473 K, 2 h at 673 K and 15 h at 813 K. Titanium silicalite-1 with stacked morphology (TS-1_s) was prepared according to [23] by microwave-assisted synthesis using tetraethyl orthosilicate (TEOS, >99%, Merck), tetrapropylammonium hydroxide (TPAOH, 10 wt.% aqueous solution, Sigma Aldrich), titanium(IV) isopropoxide (TIP, 97%, Sigma Aldrich), isopropanol (99%, BDH Prolabo) and deionized water. Typically, TEOS (36.5 g), TPAOH solution (35.7 g) and deionized water (39.2 g) were mixed under stirring for 1 h. TIP (0.71 g) and Isopropanol (8.15 g) were mixed under stirring in a separate beaker for 45 min. Subsequently the solutions were added together and further stirred for 2 h at room temperature and then stirred at 353 K for 1 h to remove isopropanol. The resulting mixture was partitioned to a 100 cm3 PTFE autoclave and transferred to a microwave oven (MLS, Start 1500) for crystallization. Therefore, the synthesis mixture were heated for 5 min with the irradiation power of 600 W to 438 K and hold at this temperature under autogeneous pressure for 20 min in case of TS-1_s_20 and 60 min for TS-1_s_60. The obtained solids were recovered by centrifugation, washed five times with 25 cm3 deionized water, dried at 393 K for 6 h and calcined in air at 823 K for 6 h as well. An industrial sample of titanium silicalite-1 (TS-1, powder) was supplied by Evonik. Finally, all catalysts were pressed (6  108 N m 2), crushed and sieved to obtain a fraction with a grain size between 50 and 150 lm for use in the catalytic experiments.

2. Experimental section

2.2. Catalyst characterization

2.1. Catalyst synthesis

The catalysts were characterized by powder X-ray diffraction (XRD, Siemens D 5000) using CuKa radiation (k = 1.5418 Å) in the 2h-range of 5° and 80° with a step size of 0.05°. The specific surface area ABET were determined from N2-sorption isotherms at 77 K using an ASAP 2000 (micromeritics) apparatus. Table 1 reports the values for ABET of the samples. The metal loading of the samples was determined by elemental analysis via optical emission spectrometry with inductively coupled plasma (ICP-OES) after dissolving the solids in a mixture of HNO3, HF, and H3BO3 by chemical extraction under pressure. Diffuse reflectance UV–Vis (DR-UV–Vis) spectra were measured at room temperature on Perkin Elmer Lambda 650 S equipped with a 150 mm integrating sphere using spectralonÒ (PTFE, reflective value 99%) as a reference.

Supported MOx–Al2O3 (M: W, Mo) as well as MoOx–SiO2 catalysts were prepared by incipient wetness impregnation of c-Al2O3 (bimodal, 99.7%, Alfa Aesar, specific surface area: 80– 120 m2 g 1) or SiO2 (Aerosil 380, Degussa) with aqueous solutions of (NH4)6H2W12O40
H2O (85% WO3 basis, Aldrich) and (NH4)2MoO4 (99%, Aldrich) under constant stirring at room temperature. Typically, 5 g of c-Al2O3 or SiO2 were used. The molar ratio of metal to aluminum or silicon, respectively, for impregnation was kept constant at nM/nAl or Si = 1/10. After removal of water under UV irradiation for 180 min, the impregnated catalysts were further dried at 428 K for 16 h and calcined in air for 4 h at 823 K. TiOx–SiO2 was obtained according to [27] by grafting of titanium sites on SiO2 (Aerosil 380, Degussa) using a solution of titanium(IV) isopropoxide (TIP, 97%, Aldrich) in cyclohexanol (99%, Sigma–Aldrich). Typically, 5 g SiO2 were treated with 150 cm3 of the TIP solution (cTIP = 0.02 mol l 1) under reflux for 2 h at 433 K. After evaporation of the solvent in vacuum, the obtained solid was calcined in air for 5 h at 823 K. Ti-containing MCM-41 (Ti-MCM-41) was synthesized according to [28] starting from silica gel beads (LicrospherÒ Si 60, Merck) as the silica source, Na2Ti3O7 as the titanium source (Aldrich) and cetyltrimethylammonium hydroxide (CTMAOH) as the structuredirecting agent. CTMAOH was prepared by ion exchange of 1 g CTMABr (99%, Acros) dissolved in 42 cm3 demineralized water with 10 g Ampersep (900 OH, Fluka). The silica gel beads, a defined amount of Na2Ti3O7 (Aldrich) and a 0.08 M CTMAOH aqueous solution were mixed under stirring for 10 min to obtain the synthesis gel of the molar composition 0.012 TiO2 : SiO2 : 0.004 Na2O : 0.202 CTMAOH : 140.2 H2O. The resulting suspension was transferred to a polypropylene flask (NalgeneÒ, 60 cm3 volume) and held at 383 K for 1 day. Then, the resulting solid was removed by filtration, washed five times with 25 cm3 deionized water and once with 20 cm3 anhydrous ethanol, and dried in air at 363 K for 15 h. The obtained materials were calcined stepwise in air with a

2.3. Catalytic Experiments Catalytic experiments were carried out batchwise in a twonecked round-bottom glass reactor (V = 25 cm3) with a septum and a reflux condenser in the liquid phase at 323 K with magnetic stirring (400 rpm). Methyl oleate (MO, P99%, Sigma–Aldrich) or biodiesel (FAME, methyl oleate: 72 wt.%, methyl linoleate + methyl Table 1 Specific surface area ABET, metal content of the catalysts as well as turnover number TON and turnover frequency TOF in the epoxidation of methyl oleate with hydrogen peroxide after 24 h. Sample

Specific surface area ABET (m2 g 1)

Metal content (wt.%)

TON

TOF (h 1)

TS-1 (ind.) Ti-MCM-41 TiOx–SiO2 WOx-Al2O3 MoOx-Al2O3 MoOx-SiO2

448 994 330 131 223 154

1.5 1.2 1.8 12.5 7.8 9.1

5.9 0.6 1.4 1.4 0.7 1.0

0.24 0.02 0.06 0.06 0.03 0.04

N. Wilde et al. / Microporous and Mesoporous Materials 164 (2012) 182–189

(a) conversion XMO / %

100 80 60 40 20 0

(b) selectivity SME/ %

100 80 60 40 20 0

(c) 100 2

80

2

linolenate: 19 wt.%, rest: not determined ; JCN Neckermann-Biodiesel GmbH Halle) were used as substrates, hydrogen peroxide (73 wt.% aqueous solution, Solvay-Wolfen) as oxidant, acetonitrile (99,9%, BDH Prolabo) as a solvent and chlorobenzene (STD, 99.8%, Aldrich) as internal standard. In a typical experiment, 10 cm3 solvent was loaded into the reactor, followed by the addition of the substrate (MO or FAME, 90 mg, 0.30 mmol), hydrogen peroxide (70 mg, 1.44 mmol) and chlorobenzene (67 mg). The loaded reactor was immersed in a heated oil bath and allowed to equilibrate for 10 min to the reaction temperature of 353 ± 1 K. The reaction was started by addition of the catalyst (150 mg). A pre-drying of the catalysts (in air at 373 K) did not have an influence on the catalytic results as the catalysts are hydrophobic (vide infra). Thus, the catalysts were used without pre-treatment. Samples (0.5 cm3) were taken from the reaction mixture through to the septum via syringe after 0, 1, 3, 5 and 24 h of reaction. The catalyst was removed from the samples by centrifugation. A defined amount of the samples (0.1 cm3) was diluted in 0.5 cm3 acetonitrile and analyzed by capillary gas chromatography (Shimadzu GC 2010 equipped with a flame ionization detector) using nitrogen as the carrier gas. Product separation was achieved on a capillary column (95% dimethylpolysiloxane cross-linked with 5% diphenylpolysiloxane, Restek RtxÒ-5 MS, length 30 m, inner diameter 0.25 mm, coating thickness 0.25 lm). Reaction products were identified by co-injection of authentic samples (see ESI) and by GC–mass spectrometry (GC Varian 3800). The turnover number TON in MO conversion was calculated as moles of methyl oleate converted per mole of metal sites present in the catalysts. Accordingly, the turnover frequency TOF was obtained by dividing TON by the reaction time in hours. For the epoxidation of biodiesel (FAME), the conversion was calculated as the ratio of the mass of converted unsaturated substrates, i.e., methyl oleate, methyl linolate, and methyl linolenate, and their initial mass in the reactant mixture. The epoxide selectivity in FAME conversions is given for the cumulative mass of all epoxides (mono-, di- and tri-epoxides) formed relative to converted mass of the substrates. For calculation of the H2O2 conversion, the concentration of H2O2 in the initial reactant solution and in the product samples was determined by iodometric titration. The initial rate of the MO conversion was calculated from tangential slopes of the time dependence of the MO concentration at t = 0 using linear curve fitting between 0 and 1 h of reaction time. For reusability tests in three consecutive reactions, the catalyst was filtered off, dried at room temperature and added to a fresh reactant solution. For recycling experiments, the catalyst was removed from the reaction mixture by filtration after the third run, calcined in air 673 K for 24 h and, again, added to fresh reactant solution.

conversion XH O / %

184

60 40 20 0

3 3 iO 2 iO 2 d.) lO lO -41 -S in -xS -A 2 M -A 2 Ox O 1( C x i x O o T oO -M W M TS M Ti

Fig. 1. Conversion of methyl oleate XMO (a), epoxide selectivity SME (b) and conversion of hydrogen peroxide XH2 O2 (c) in the epoxidation of methyl oleate over different catalysts in acetonitrile at 323 K after 24 h.

3. Results and discussion 3.1. Conversion of methyl oleate 3.1.1. Catalyst screening For a screening of catalysts for the conversion of methyl oleate with aqueous hydrogen peroxide solution in the liquid phase, Ticontaining silicates as well as supported Mo- and W-oxides were selected as catalysts. On the Al2O3-supported WOx and MoOx catalysts, crystalline phases were not observed by XRD (see ESI) indicating a high metal oxide dispersion on the support with overall metal contents between 7.8 and 12.5 wt.% (Table 1). Among the Mo- and W-based catalysts, the highest conversion of methyl oleate (57%) is achieved over WOx–Al2O3 probably due to the high metal loading (Fig. 1). The Mo-based catalysts are less active,

although somewhat more selective for epoxidation than the W-based catalyst. The low activity of MoOx–SiO2 could be a result of a lower metal oxide dispersion. This is supported by the presence of reflections of crystalline MoO3 in the orthorhombic phase [29] in the XRD pattern (see ESI). Interestingly, however, H2O2 is completely converted on all three W- or Mo-based catalysts. This is consistent with earlier literature reports on oxidation reactions with H2O2 over these type of catalysts [30]. Clearly the highest MO conversion and epoxide selectivity are reached over the industrial TS-1 catalyst (Fig. 1). With >90% and 87%, respectively, these are much higher than those of the supported W- and Mo-oxides as well as TiOx–SiO2. The higher activity of the TS-1 (ind.) catalyst is also apparent when comparing the TON and TOF (Table 1). Note, however, the similar TON for

185

N. Wilde et al. / Microporous and Mesoporous Materials 164 (2012) 182–189

Fig. 2. Proposed reaction scheme for the epoxidation of methyl oleate with H2O2.

TiOx–SiO2 and WOx–Al2O3 despite the different metal loading and the higher selectivity of Ti-based catalyst for the epoxide formation. Overall, the TON over the industrial TS-1 catalysts is larger by a factor of 4–8 compared to the other catalysts studied. In order to supply a high accessible surface, Ti-MCM-41 (ABET = 994 m2 g 1, Table 1, dP,DFT=3.4 nm, VP,BJH = 0.92 cm3 g 1) was also included in the study. However, both MO conversion and ME selectivity were below 10% on this catalyst. Either, the rather bulky reactant molecules of MO do not have access to the mesopores or the Ti sites are less active than those on TS-1 (ind.) or TiOx–SiO2. The lower TON for the Ti-MCM-41 with respect to the latter two Ti-containing catalysts points towards a difference in catalytically active Ti-sites rather than in the accessibility of the pore system. This is in accordance with the results reported by Guidotti et al. [20] for the epoxidation of methyl oleate with hydrogenperoxide over Titanium-grafted MCM-41, MCM-48 and SiO2 and Rios et al. [31] for the epoxidation of methyl oleate with TBHP over Ti-MCM-41 and TiOx–SiO2. The latter authors found that conversion and selectivity are similar for the two catalyst and, thus, largely independent of the geometry of the pore system. It is also consistent with earlier reports describing the presence of two active Ti sites, i.e., the tetrahedral (SiO)4Ti (species I) and the tripodal (SiO)3Ti(OH) (species II), in Ti-containing silicates [32,33]. While species I is predominant in TS-1, species II prevails in amorphous Ti–SiO2 and Ti-MCM-41, respectively. It was found for styrene epoxidation that species I exhibits superior selectivity for the epoxide (as compared to species II) [32]. We may, therefore, assume that a larger concentration of species I in TS-1(ind.) compared to TiOx–SiO2 and Ti-MCM-41, also accounts for the observed differences in conversion, TON, and selectivity for the epoxidation of methyl oleate. The question of Ti coordination will be referred to again in a later section. Besides the major product methyl 9,10-epoxy stearate (ME), several by-products were formed, predominantly in consecutive reactions (Fig. 2). The hydrolysis product methyl 9,10-dihydroxy stearate and the products from oxidative cleavage of the epoxide were present in about equal amounts. The ketone from rearrangement is formed in small amounts only (<1%). The amount of detected cleavage products increased with reaction time. However, the exact amount of the by-products was not further quantified.

3.1.2. Influence of the solvent As opposed to organic hydroperoxides such as tert.-butyl or cumene hydroperoxide, the solubility of FAME in aqueous H2O2 solution is very limited. To obtain a homogeneous solution of the FAME/H2O2-reactant mixture, the utilization of a liquid solvent is required. The nature of the solvent is known to have a major influence on reaction kinetics and product selectivity during the oxidative conversions over TS-1 as a catalyst [34,35]. For the epoxidation of methyl oleate with H2O2 over TS-1 (ind.), the effect of polarity and the protic/aprotic nature of different solvents on initial reaction rate, conversion of MO and H2O2 and epoxide selectivity is summarized in Table 2. Generally, the initial reaction rate (determined within the first h of conversion) is well reflected in the conversion of MO after 24 h of reaction. The only exceptions are the conversions in methanol and diisopropylether where the rate was lowest and experimental error in rate determination becomes important. The activity of TS-1 is lower in methanol than in acetonitrile. This result cannot be explained only in terms of polarity, since both solvents have similar dielectric constants (emethanol = 32.7, eacetonitrile = 37.5). Likewise, it is in contrast with the reported positive effect of protic solvents on the reactivity of TS-1 [34,36]. This might be an indication that the conversion does not occur inside the micropores of the TS-1 catalysts, but at the external surface where the solvent effect might be different.

Table 2 Influence of the solvent on initial rate of reaction, conversion of methyl oleate XMO, selectivity SME and conversion of hydrogen peroxide XH2 O2 in the epoxidation of methyl oleate over TS-1 (ind.) at 323 K after 24 h. Solvent

Initial Rate r0 (mol l 1 h 1)

Conversion XMO (%)

Selectivity SME (%)

Conversion XH2 O2 (%)

Ethylacetate Acetone Acetonitrile Acetonitrile/ methanol Diglyme Methanol Diisopropylether

0.025 0.024 0.021 0.016

98 92 93 79

36 52 87 85

72 36 97 62

0.015 0.002 0.005

79 20 4

84 17 31

66 57 67

186

N. Wilde et al. / Microporous and Mesoporous Materials 164 (2012) 182–189

Scheme 1.

In general, MO conversion over 90% was achieved by using aprotic-polar solvents like ethyl acetate rather than in aprotic nonpolar solvents like diisopropylether (Table 2). The MO conversion in aprotic-polar solvents falls in the following order: ethylacetate > acetonitrile > acetone > acetonitrile/methanol > diglyme. However, epoxide selectivity in ethyl acetate and acetone were below 50%. For ethyl acetate, conversion by acid catalyzed hydrolysis leads to acetic acid and ethanol, the latter catalyzing epoxide ring opening and formation of different diols. In the case of acetone, a cyclic ketal (scheme 1) formed by the acid catalyzed reaction of aceton with the diol from hydrolysis of the epoxide ring may explain the low selectivity. The same effect, i.e., the formation of a cyclic ketal, was observed by Corma et al. [35] for epoxidation of 1-hexene with H2O2 in the presence of Ti-BETA as the catalyst. Diol formation is also the reason for reduced epoxide selectivity when methanol is added to acetonitile as the solvent. Diglyme offers lipophilic properties and has the ability to dissolve high amounts of fatty acid methyl esters [37], but, again, does not result in conversions of MO or H2O2 above 80%. Since the clearly highest values for both conversion of MO (93%) and epoxide selectivity (87%) were found for acetonitrile, this solvent was chosen for the further studies. Note that in acetonitrile, both MO and the aqueous H2O2 solution are soluble at the concentrations applied here and that, thus, a homogeneous liquid phase is in contact with the solid catalyst.

conversion X or selectivity S / %

100

XMO SME 60 0 310

320

330

340

350

T/K Fig. 3. Effect of reaction temperature on conversion of methyl oleate XMO and epoxide selectivity SME in the epoxidation of methyl oleate over TS-1 (ind.) in acetonitrile after 3 h.

but is lower at higher temperature. At these higher temperatures, the formation of by-products, predominantly the cleavage products (Fig. 2), is favored. From the initial rate data for 313–333 K, an activation energy of 21.8 kJ mol 1 is calculated (see ESI). This is considerably lower than that of typical epoxidations over TS-1. For instance, the activation energy of ethylene epoxidation with H2O2 was calclulated by DFT methods to be 98 kJ mol 1 [39]. The low activation energy observed here points to a mass-transport limitation of the MO conversion under the present reaction conditions. For further studies, the temperature was kept at 323 K as this provides the most suitable compromise between selectivity and conversion rate.

3.1.3. Influence of the temperature Generally, it is advantageous to carry out the epoxidation at lower temperatures and, thus, limit the occurrence of side reactions such as hydrolysis, etherfication and deactivation of the catalyst [38]. Here, the reaction temperature was varied in the range of 313 to 353 K. With increasing the temperature from 313 to 333 K, the conversion of MO increases almost linearly from 66% to 91%, but levels off at higher temperatures (Fig. 3). The epoxide selectivity is only slightly affected by temperature up to 333 K,

0.020

0.01

Thiele Modulus

6 0.015

0.010

4

r0

2

inital rate r0 / mol l-1 h-1

8

0.025

inital rate r0 / mol l-1 h-1

80

0.005

0.000 0

100

mcat. / mg

200

300

0 0.1

1

max. particle size / mm

Fig. 4. Initial rate r0 as a function of catalyst mass (left part) and as a function of maximum catalyst particle size (right part) for the epoxidation of methyl oleate over TS-1 (ind.) in acetonitrile at 323 K. Also the Thiele Modulus U is shown in the right part.

187

N. Wilde et al. / Microporous and Mesoporous Materials 164 (2012) 182–189

conversion X or selectivity S / %

100

80

60

40

XMO

XH O

20

0

2

XMO

2

SME 0

5

10

15

SME run 1

20

run 2

run 3

RC

t/h Fig. 5. Conversion of methyl oleate XMO, epoxide selectivity SME and conversion of hydrogen peroxide XH2 O2 as a function of reaction time in the epoxidation of methyl oleate over TS-1 (ind.) in acetonitrile at 323 K (left part) as well as after separation and reuse or regeneration (RC: regenerated catalyst, calcination in air flow for 24 h at 643 K) of the catalyst (right part, reaction time 24 h).

5

0

0

-5

-5

-10

-10

-15

-15

1.4x10

-7

1.2x10

-7

1.0x10

-7

8.0x10

-8

m/%

DTA / µV

5

-6

1.2x10

273

473

673

873 T/K

1073

1.1x10

-6

1.1x10

-6

Intensity (me = 18) / A

3.1.5. Catalyst reusability and regeneration The conversion of MO and H2O2 as well as the selectivity for the epoxide (ME) are shown as a function of reaction time for the chosen conditions, i.e., in liquid acetonitrile and at 323 K over TS-1 (ind.) as the catalyst in Fig. 5, left part. Both conversion of MO and of H2O2 rise rapidly within the first 5 h of the experiment where after the conversion increases only slightly to reach values of 87% and 93%, respectively (see also Table 2). Note, however, that H2O2 is present in a fivefold excess with respect to MO and, thus, the vast majority of H2O2 is converted by unproductive decomposition. Concomitantly, the epoxide selectivity decreases steadily from 92% to 87% at 24 h. This supports that the by-products are indeed formed via consecutive reactions from ME (see Fig. 2). Since, however, the conversion slows down strongly after the first 5 h of

reaction and since a complete conversion of MO was not reached in any of the experiment, the question of catalyst deactivation arose. In fact, it was reported that a deactivation of Ti-MCM-41 catalysts occurs during oxidation of cyclohexene with aqueous H2O2 due to adsorbed reaction residues hindering the access to the active centers and/or due to Ti leaching [41]. In the former case, the catalyst could be regenerated by calcination to eliminate adsorbed reaction residues on the active sites. If the catalyst TS-1 (ind.) is reused three times (after removal by filtration and drying in air at room temperature), the MO conversion steadily drops while the epoxide selectivity slightly increases (Fig. 5, right part). By calcination of the catalyst in air at 673 K for 24 h after the third reuse, the initial activity and selectivity of the catalyst can be completely recovered. This regeneration was further characterized by TG-DTA-MS analysis of the catalyst after the third reuse (Fig. 6). The small weight loss (2 wt.%) up to 373 K is due to the loss of physisorbed water, consistent with hydrophobic nature of the TS-1 surface. A more pronounced weight loss of 11 wt.% together with a strong exothermicity in

Intensity (me = 44) / A

3.1.4. Influence of catalyst particle size and mass To more deeply investigate whether or not the MO conversion with H2O2 over the industrial TS-1 catalyst is limited by mass transport effects, the mass and the particle size of the catalyst were varied. Upon increasing the catalyst mass from 10 to 300 mg, the inital rates increase (Fig. 4, left part). In the absence of mass transport effects, the initial rates would be expected to increase linearly with catalyst mass. However, a deviation of this expected linear increase for catalyst mass above 150 mg indicates the presence of mass-transport limitations. Additionally, the initial rates decrease with increasing catalyst particle size (Fig. 4, right part), corresponding to a decrease in MO conversion of 55% for particles <0.2 mm to 13% for particles >1.6 mm. From the decrease of the initial rates with particle size, the effective diffusion coefficient of the reactant MO was calculated applying the Thiele Modulus U (Fig. 4, right part, for calculation see ESI). The value found here, i.e., Deff = 5.7  10 11 m2 s 1 is somewhat lower than that reported by Zieverink et al. [40] for methyl oleate in the hydrogenation and isomerization over an alumina supported palladium catalyst (1.8  10 10 m2 s 1). The lower temperature and the difference in size, geometry and tortuosity of the catalyst particle may account for this lower effective diffusivity. These findings lead to the unambiguous conclusion that the epoxidation of methyl oleate with H2O2 over TS-1 is strongly limited by diffusion into the catalyst particle. As the size of the TS-1 crystallites was not altered, it is the diffusion of the reactants to the outer surface of the TS-1 crystallites that limits the conversion rate.

1273

Fig. 6. TG-DTA-profiles (top part) and MS analysis of the off-gas during thermogravimetric analysis of the catalyst TS-1 (ind.) after run 3 (see Fig. 5).

188

N. Wilde et al. / Microporous and Mesoporous Materials 164 (2012) 182–189

80

K-M-units

conversion X or selectivity S / %

100

60

XMO 40

XFAME SME

20

TS-1_s_60 TS-1_s_20

SeFAME

Ti-MCM-41 0 0

5

10

15

20

25

200

t/h

300

400

500

Fig. 7. Conversion of the substrates and selectivity for epoxidized products in the epoxidation of methyl oleate (MO) or biodiesel (FAME) with hydrogen peroxide over TS-1 (ind.) in acetonitrile at 323 K as a function of reaction time.

700

800

Fig. 8. DR-UV–Vis spectra of TS-1 with stacked morphology (TS-1_s_60, and TS1_s_20 from microwave irradiation for different times during synthesis) and of TiMCM-41.

DTA curve up to 1273 K is accompanied by the formation of carbon dioxide and water. This supports that during heating the catalyst in air, organic residues on the catalyst are removed. The deactivation of TS-1 (ind.) in the epoxidation of MO is therefore attributed to blocking of the active Ti-sites on the outer crystal surface by organic deposits, presumable formed by consecutive reactions of target products, such as dimerization or oligomerization. These can, however, be completely removed by calcination in air.

(overall fraction of mono-, di- and triunsaturated compounds: 91 wt.%). As shown in Fig. 7, the conversion of biodiesel (FAME) is only lower by ca. 10% in the first 5 h than that of pure MO as the reactant, but reaches a similar value (90% for FAME vs. 93% for MO) after 24 h. The selectivity for the formation of epoxides is also a little lower than with MO as the only reactant and amounts to 76% (vs. 87% for MO) after 24 h (Fig. 7). 3.2.2. Epoxidation of biodiesel over TS-1 with stacked morphology As an attempt to find a more active catalyst for the biodiesel epoxidation, TS-1 with stacked morphology was prepared according to [23]. These catalysts are comprised of particles with sizes in the sub-micrometer scale and, thus, a high outer crystal surface. They were previously reported to be superior catalysts for the epoxidation of linear C6–C12-olefins with the C@C-bond in terminal position with H2O2 with respect to conventional TS-1. In this study, two samples were prepared using different irradiation times in microwave-assisted synthesis. The samples show the typical XRD patterns for TS-1 (see ESI) and possess Ti contents

3.2. Conversion of biodiesel (FAME) 3.2.1. Epoxidation of biodiesel over TS-1 (ind.) In order to evaluate whether commercial biodiesel can also be epoxidized using aqueous H2O2 solution, the conversion was again carried out over the catalyst TS-1 (ind.) under the same reaction conditions as for the MO epoxidation, but with a FAME mixture derived from rapseed oil. This biodiesel consists predominantly of methyl oleate (72 wt.%) and smaller fractions of di- and triunsaturated compounds such as methyl linolate and methyl linolenate

100

10

XFAM E S eFA M E

80

8

60

6

40

4

20

2

TON

conversion X or selectivity S / %

600

Wavelength / nm

0

-M Ti

C

M

-4

1 TS

-1

d (in

.) TS

-1

_

2 s_

0 TS

-1

_

6 s_

0 -M Ti

C

M

-4

1 TS

-1

d (in

0

.) TS

-1

_

2 s_

0 TS

-1

_

6 s_

0

Fig. 9. Conversion of biodiesel XFAME and epoxide selectivity SeFAME (left part) turnover number TON (right part) in the epoxidation of biodiesel over different Ti-containing catalysts in acetonitrile at 323 K after 24 h.

N. Wilde et al. / Microporous and Mesoporous Materials 164 (2012) 182–189

of 0.5 and 0.7 wt.% and average crystallite sizes of 60 and 100 nm (see ESI for SEM micrograph) for 20 (sample TS-1_s_20) and 60 min (sample TS-1_s_60) microwave irradiation, respectively. The DR-UV–Vis spectra of these samples are typical for TS-1 and show an absorption band around 215 nm characteristic for Ti in tetrahedral coordination (Fig. 8) [24]. In sharp contrast, Ti-MCM-41 shows a broader, less intense band with maximum at 230 nm, a shoulder at 250 nm and another band with maximum at 330 nm. This indicates, probably, the presence of higher coordinated Ti species [24,41] in this sample. The conversion of biodiesel on the catalysts with the stacked morphology (TS-1_s_20 and TS-1_s_60) is clearly lower than that over commercial catalyst TS-1 (ind.), whereas the selectivity is only a little lower (Fig. 9, left part). The activity of these catalysts is much higher than that of Ti-MCM-41, as expected for the smaller fraction of active sites with high epoxidation activity, i.e., framework incorporated Ti. If, however, the TON is compared, the catalysts from microwave-assisted synthesis are clearly superior to the commercial TS-1 (ind.) (Fig. 9, right part). Especially, the sample obtained after 60 min microwave irradiation TS-1_s_60 exhibits a TON of 8.7 vs. 5.9 for TS-1 (ind.). One reason for this higher activity is certainly the higher amount of active Ti sites on the outer surface of the small crystals of TS-1_s_60. Note, however, that on the sample with smaller crystallites (TS-1_s_20), TON is lower than for TS-1_s_60. This might be explained by a higher surface hydrophobicity for the sample from microwave treatment at longer time [23]. The hydrophobicity of TS-1_s_60 might even be higher than that of TS-1 (ind) contributing to the higher TON with respect to the commercial catalyst. 4. Conclusions The heterogeneously catalyzed epoxidation of fatty acid methyl esters with aqueous hydrogen peroxide solution provides an attractive route to chemicals and intermediates for a wide range of applications such as polymer production. Using an industrial TS-1 as the catalyst, methyl oleate conversion with hydrogen peroxide yields the epoxide with a selectivity of 87% at 93% conversion after 24 h of reaction in liquid acetonitrile. Likewise, commercial biodiesel can be epoxidized over TS-1 with 76% selectivity at 90% conversion. TS-1 is superior to other Ti-containing catalysts such as Ti-MCM-41 or TiOx–SiO2 due to its high fraction of tetrahedrally coordinated Ti on framework positions as the active sites. The TS-1 catalyst is subject to a deactivation by deposition of organic compounds. The activity can, however, be completely be restored by calcination in air. Indeed, it was found out that diffusion limits the conversion of FAME over TS-1. Therefore, catalysts with small crystallite size and high outer surface area are beneficial for high activity. Here, TS-1 with crystallite sizes in the sub-micrometer scale and with stacked morphology as obtained from microwaveassisted synthesis are shown to exhibit a significantly higher turnover number than the industrial TS-1 catalyst. Further attempts to obtain even more active catalysts for the epoxidation of biodiesel with aqueous hydrogen peroxide solution should, thus, focus on the preparation of TS-1 with nanstructured crystallites and a highly accessible outer surface area, e.g., within hierarchically structured pore systems.

189

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.micromeso.2012. 06.047. References [1] G. Knothe, R.O. Dunn, M.O. Bagby, Biomass 666 (1997) 172. [2] J.O. Metzger, U. Biermann, W. Friedt, S. Lang, W. Luhs, G. Machmuller, M.R. Klaas, H.J. Schafer, M.P. Schneider, Ang. Chem. Int. Ed. 39 (2000) 2206. [3] U. Biermann, U. Bornscheuer, M.A.R. Meier, J.O. Metzger, H.J. Schafer, Ang. Chem. Int. Ed. 50 (2011) 3854. [4] DE 4201343A1, 22.07.1993, Henkel KGaA, B. Gruber. [5] F.D. Gunstone, F.B. Padley, in: F.D. Gunstone (Ed.), Lipid Technologies and Applications, Marcel Dekker, New York, 1997, p. 759. [6] DE 4202758C2, 05.08.1993, Henkel KGaA, H. Kluth, P. Daute, J. Klein, R. Grützmacher, W. Klauck. [7] A. Debal, G. Rafaralahitsimba, E. Ucciani, Fett Wissenschaft Technologie-Fat Sci. Technol. 95 (1993) 236. [8] M.A. Camblor, A. Corma, P. Esteve, A. Martinez, S. Valencia, Chem. Commun. (1997) 795. [9] G.A. Eimer, V.R. Elias, M.E. Crivello, E.R. Herrero, S.G. Casuscelli, Ind. Eng. Chem. Res. 48 (2009) 9076. [10] K. Burgess, B.S. Lane, Chem. Rev. 103 (2003) 2457. [11] A.L.V. de P, B.F. Sels, D.E. De Vos, P.A. Jacobs, J. Org. Chem. 64 (1999) 7267. [12] J.Y. Tang, L. Wang, G. Liu, Y. Liu, Y.Z. Hou, W.X. Zhang, M.J. Jia, W.R. Thiel, J. Mol. Catal. A: Chem. 313 (2009) 31. [13] S. Imamura, H. Sasaki, M. Shono, H. Kanai, J. Catal. 177 (1998) 72. [14] A.C. Gomes, S.M. Bruno, S. Gago, R.P. Lopes, D.A. Machado, A.P. Carminatti, A.A. Valente, M. Pillinger, I.S. Goncalves, J. Organomet. Chem. 696 (2011) 3543. [15] M. Abrantes, P. Neves, M.M. Antunes, S. Gago, F.A.A. Paz, A.E. Rodrigues, M. Pillinger, I.S. Goncalves, C.M. Silva, A.A. Valente, J. Mol. Catal. A: Chem. 320 (2010) 19. [16] E. Poli, J.M. Clacens, J. Barrault, Y. Pouilloux, Catal. Today 140 (2009) 19. [17] P.T. Anastas, J.C. Warner (Eds.), Green Chemistry – Theory and Practice, Oxford University Press, New York, 1998. [18] M. Guidotti, N. Ravasio, R. Psaro, E. Gianotti, S. Coluccia, L. Marchese, J. Mol. Catal. A: Chem. 250 (2006) 218. [19] M. Guidotti, R. Psaro, N. Ravasio, M. Sgobba, E. Gianotti, S. Grinberg, Catal. Lett. 122 (2008) 53. [20] M. Guidotti, E. Gavrilova, A. Galarneau, B. Coq, R. Psaroa, N. Ravasio, Green Chem. 13 (2011) 1806. [21] P.A.Z. Suarez, M.S.C. Pereira, K.M. Doll, B.K. Sharma, S.Z. Erhan, Ind. Eng. Chem. Res. 48 (2009) 3268. [22] J. Sepulveda, S. Teixeira, U. Schuchardt, Appl. Catal. A 318 (2007) 213. [23] H.L. Jin, N.Z. Jiang, S.M. Oh, S.E. Park, Top. Catal. 52 (2009) 169. [24] G.N. Vayssilov, Catal. Rev. Sci. Eng. 39 (1997) 209. [25] A. Wroblewska, J. Mol. Catal. A: Chem. 229 (2005) 207. [26] R.A. Sheldon, J. Chem. Technol. Biotechnol. 68 (1997) 381. [27] A. Campanella, M.A. Baltanas, M.C. Capel-Sanchez, J.M. Campos-Martin, J.L.G. Fierro, Green Chem. 6 (2004) 330. [28] L. Wang, J.J. Zou, M.Y. Zhang, B. Zhu, X.W. Zhang, Z.T. Mi, Catal. Lett. 124 (2008) 139. [29] Y.K. Park, S.J. Kim, N. You, J. Cho, S.J. Lee, J.H. Lee, J.K. Jeon, J. Ind. Eng. Chem. 17 (2011) 186. [30] Y. Su, Y.M. Liu, L.C. Wang, M. Chen, Y. Cao, W.L. Dai, H.Y. He, K.N. Fan, Appl. Catal. A 315 (2006) 91. [31] L.A. Rios, P. Weckes, H. Schuster, W.F. Hoelderich, J. Catal. 232 (2005) 19. [32] D. Srinivas, P. Manikandan, S.C. Laha, R. Kumar, P. Ratnasamy, J. Catal. 217 (2003) 160. [33] P. Ratnasamy, V.N. Shetti, P. Manikandan, D. Srinivas, J. Catal. 216 (2003) 461. [34] M.G. Clerici, G. Bellussi, U. Romano, J. Catal. 129 (1991) 159. [35] A. Corma, P. Esteve, A. Martinez, J. Catal. 161 (1996) 11. [36] M.G. Clerici, P. Ingallina, J. Catal. 140 (1993) 71. [37] C.Y. Lin, K.H. Wang, Fuel 83 (2004) 507. [38] A. Wroblewska, E. Lawro, E. Milchert, Ind. Eng. Chem. Res. 45 (2006) 7365. [39] E. Karlsen, K. Schoffel, Catal. Today 32 (1996) 107. [40] M.M.P. Zieverink, M.T. Kreutzer, F. Kapteijn, J.A. Moulijn, Ind. Eng. Chem. Res. 44 (2005) 9668. [41] E.R. Herrero, G.A. Eimer, S.G. Casuscelli, G.E. Ghione, M.E. Crivello, Appl. Catal. A 298 (2006) 232.