- Email: [email protected]
H2O2-based selective epoxidations: Nb-silicates versus Ti-silicates
Accepted Manuscript Title: H2 O2 -based selective epoxidations: Nb-silicates versus Ti-silicates Authors: Oxana A. Kholdeeva, Irina D. Ivanchikova, Nataliya V. Maksimchuk, Igor Y. Skobelev PII: DOI: Reference:
S0920-5861(18)30375-4 https://doi.org/10.1016/j.cattod.2018.04.002 CATTOD 11358
To appear in:
Catalysis Today
Received date: Revised date: Accepted date:
18-12-2017 27-2-2018 1-4-2018
Please cite this article as: Kholdeeva OA, Ivanchikova ID, Maksimchuk NV, Skobelev IY, H2 O2 -based selective epoxidations: Nb-silicates versus Ti-silicates, Catalysis Today (2010), https://doi.org/10.1016/j.cattod.2018.04.002 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
H2O2-based selective epoxidations: Nb-silicates versus Ti-silicates Oxana A. Kholdeeva,1,2,* Irina D. Ivanchikova,1 Nataliya V. Maksimchuk,1,2 Igor Y. Skobelev1 1
Boreskov Institute of Catalysis, Lavrentieva ave. 5, Novosibirsk, 630090, Russia 2
Novosibirsk State University, Pirogova st. 2, Novosibirsk, 630090, Russia
IP T
Fax: (+7)-(383)-330-9573; phone: (+7)-(383)-326-9433; e-mail: [email protected]
CC E
PT
ED
M
A
N
U
SC R
Graphical Abstract
Highlights
A
Mesoporous Nb-silicates reveal better hydrolytic stability than Ti-silicates. Alkene epoxidation over both Nb- and Ti-silicates proceeds via Eley–Rideal mechanism. The activation energy of cyclooctene epoxidation is lower for Nb-silicate. The activation energy of hydrogen peroxide decomposition is lower for Ti-silicate. The higher epoxidation selectivity of Nb-catalysts is related to their lower activity in H2O2 decomposition.
Abstract This paper aims at a comparison of the stability and catalytic performance of niobium- and titaniumcontaining mesoporous silicates prepared by evaporation-induced self-assembly (EISA) in selective
1
oxidations with hydrogen peroxide. The catalysts comprised evenly dispersed dimeric and/or small oligomeric active sites (M = Nb(V) or Ti(IV)), and the Nb sites revealed better hydrolytic stability than the Ti ones, as demonstrated by DRS UV-vis. An attempt has been made to rationalize the differences observed in chemo- and regioselectivity of alkene epoxidation using kinetic modeling tools. The rate law established for cyclooctene epoxidation over the Nb- and Ti-catalysts is consistent with an Eley–Rideal mechanism that involves adsorption of Н2О2 on М sites, interaction between M and Н2О2 to afford a
IP T
hydroperoxo species ‘MOOH’ and water, followed by oxygen transfer from ‘MOOH’ to С=С bond, producing epoxide and regenerating M. The higher epoxidation (heterolytic pathway) selectivity of the
SC R
Nb catalysts is, at least partially, related to their lower activity in H2O2 unproductive decomposition. The apparent activation energy of H2O2 degradation is higher for Nb-silicate (20.5 vs 14.8 kcal/mol for Ti-
U
silicate), while the activation energies of epoxidation follow an opposite trend (11.9 vs 14.2 kcal/mol). The
N
heterolytic pathway selectivity of Nb catalysts can be greatly improved by decreasing the reaction
A
temperature. The difference in regioselectivity of limonene epoxidation over Nb- and Ti-silicates observed
M
in MeCN, where the nature of the catalysis is truly heterogeneous, indicates different structures of the active ‘NbOOH’ and ‘TiOOH’ species and/or different mechanisms of the oxygen transfer step. The
ED
inversion of regioselectivity observed for Nb catalyst in MeOH may be caused by Nb leaching in this solvent and/or changing the structure of ‘NbOOH’.
CC E
mechanism
PT
Keywords: alkene epoxidation; niobium-silicates; titanium-silicates; hydrogen peroxide; kinetics;
1. Introduction
The selective transformation of organic compounds into their oxygenated derivatives using
A
leaching-resistant truly heterogeneous catalysts and green oxidants is a challenging goal of oxidation catalysis [1-2]. In particular, hydrogen peroxide is a clean and atom-efficient oxidant that produces water as the sole side product [5-6]. The advent in the late 1980s of hydrophobic titanium silicalite-1 (TS-1) isostructural with ZSM-5 [8] has become a revolution in the field of heterogeneous oxidation catalysis since this zeotypes material appeared capable of heterolytically activating H2O2 and oxygenating small organic substrates [9]. The demands of the fine chemicals industry stimulated an intensive research
2
activity on the development of mesoporous titanium-silicates [10,11]. However, the presence of numerous silanol groups on the surface of mesoporous silica (silicates) favors adsorption of water and hydrogen peroxide at the expense of nonpolar organic substrates, which leads to significant unproductive decomposition of the oxidant, reduction of its utilization efficiency, and contribution of homolytic oxidation products. On the other hand, the hydrophilic environment of Ti sites in the mesoporous catalysts may offer advantages for bulky substrates with intermediate polarity, e.g. alkylated phenols [12-
IP T
13]. The potential of mesoporous titanium-silicates for the production of substituted benzoquinones using H2O2 has been recently discussed [15,16]. Although several types of hydrothermally stable and leaching-
SC R
tolerant mesoporous Ti-silicates have been developed [10,11,17-18], the stability of Ti active sites toward progressive agglomeration under turnover conditions still remains a major concern if aqueous H2O2 is
U
employed.
N
In recent years, niobium-containing silicates have attracted great attention of the selective
A
oxidation community. Various types of mesoporous Nb-silicates have been synthesized and tested in
M
oxidations with H2O2 [21-22]. Some of them turned out more selective than their Ti analogs in epoxidation of alkenes, including those containing highly reactive allylic hydrogen atoms, e.g. cyclohexene and
ED
limonene, which indicated a better ability of the Nb catalysts to accomplish selective oxidation with H2O2 via a heterolytic mechanism [25,29,30]. An increase in activity and H2O2 efficiency has been also noticed
PT
for mesostructured silicates with framework-incorporated Nb(V) compared to framework-incorporated
CC E
titanium [33]. Moreover, Nb-silicates, in contrast to their Ti counterparts, were found to be able to catalyze epoxidation of both electron-rich and electron-deficient C=C bonds and demonstrate unusual regioselectivity toward epoxidation of less substituted (less nucleophilic) double bond in limonene
A
[25,26,28].
So far, a few attempts at rationalizing the catalytic performance of Nb(V) have been reported
[30,34-35], but conclusions of these studies look somewhat contradictory. Recently, we proved that epoxidation of C=C bonds in α,β-unsaturated carbonyl compounds proceeds through the so-called Payne oxidation [38], which involves rate-limiting oxidation of the solvent molecule (MeCN) and formation of peroxycarboximidic acid Н3СC(=NH)OOH that is the real epoxidizing species in the system [30]. Weak basic
3
sites present in Nb-silicates are most likely responsible for this specific catalytic activity. Kinetic and spectroscopic studies identified a hydroperoxo niobium species (absorption band at 307 nm in DRS UVvis) as the active species that performs epoxidation of electron-rich double bonds in alkenes [30]. In the present work, we made an effort to gain further insights into the understanding of the different catalytic performance of mesoporous Nb- and Ti-silicates in H2O2-based selective oxidation
IP T
through careful evaluation of kinetics of alkene epoxidation and H2O2 decomposition over these two types of catalysts. We also provide here a comparison of the stability of the Nb- and Ti-catalysts prepared by
SC R
the same, evaporation-induced self-assembly (EISA) methodology using DR UV-vis spectroscopic technique and leaching tests.
U
2. Experimental
N
2.1. Materials and catalysts
A
Cetyltrimethylammonium bromide (CTAB, 99+%), titanium (IV) ethoxide (TEOT), acetylacetone
M
(acac, 99%), and tetraethyl orthosilicate (TEOS, 98+%) were purchased from Aldrich. Niobium(V) ethoxide (99.95%) was used as received from Acros. Acetonitrile (HPLC grade, Panreac) was dried and stored over
ED
activated 4 Å molecular sieves. Cyclohexene (CyH), limonene, and cis-cyclooctene (CyO) were purchased from Aldrich and passed through neutral alumina prior to use. All other reagents and solvents were the
PT
best reagent grade and were used without further purification. The concentration of H2O2 (30 or 50 wt%
CC E
in water) was determined iodometrically prior to use. Deionized water (EASY pure, RF, Barnsted) was used for the preparation of catalysts. A mesoporous titanium-silicate (Ti-Si) was prepared by the EISA technique following a protocol
A
described previously [20]. TEOT modified with acac was used as Ti source. A mesoporous niobium-silicate (Nb-Si) was prepared with the same EISA approach using niobium(V) ethoxide [28]. The catalysts were characterized by elemental analysis, low-temperature N2 adsorption, and DR UV–vis techniques. Before all measurements, the catalyst samples were calcined at 550 °C. 2.2.Evaluation of catalyst stability
4
To evaluate the catalyst stability toward water and H2O2, the catalyst samples were subjected to two types of treatments: 1) boiling in water for 6 h and 2) stirring with a MeCN solution of H2O2 (the latter was taken as 30% solution in water) for 1 h (100 mg of catalyst, 0.11 M H2O2, 20 mL MeCN, 25 oC). After the treatments, the samples were separated by filtration, dried in air and then calcined at 550 oC directly before DRS UV-vis measurements.
IP T
2.3. Catalytic oxidations Catalytic oxidations of cyclohexene were carried out at 30 and 50 oC in MeCN (1 mL) using the
SC R
following concentrations of the reagents: alkene 0.1 M, H2O2 0.1 M (taken as 50% solution in water), and catalyst (0.003 mmol Nb or Ti). Catalytic oxidations of limonene were performed at 50 oC in MeCN or MeOH (1 mL) using alkene 0.1 M, H2O2 0.1 M (taken as 30% solution in water), and catalyst (0.003 mmol
U
Nb).The product yields and substrate conversions were quantified by GC using biphenyl as internal
A
M
2.4. Nature of catalysis and leaching tests
N
standard.
The nature of catalysis (truly heterogeneous versus homogeneous) over Nb-silicates in different
ED
solvents (MeCN and MeOH) was verified by hot filtration tests [39]. The amount of metal leached into
PT
solution during the reaction course was determined by ICP–AES. 2.5. Hydrogen peroxide decomposition
CC E
Decomposition of H2O2 (0.4 M) was studied in the absence of organic substrate at 80 oC in MeCN
(5 mL) in the presence of either Nb-Si or Ti-Si catalyst (0.015 mmol Nb or Ti). Aliquots of 0.2 mL were taken during the reaction course, and H2O2 concentration was determined by iodometric titration. Four parallel
A
experiments were carried out. The temperature dependence of the decomposition rate was studied in the range of 40–80 ºC. 2.6. Kinetic study Kinetic experiments were performed in temperature-controlled glass vessels under vigorous stirring (600 rpm). Reactions were initiated by addition of H2O2 into a MeCN solution containing organic
5
substrate (CyO), catalyst, and internal standard for GC. The total volume of the reaction mixture was 1 mL. The reaction temperature was 50 ºC. Samples (1 μL) of the reaction mixture were withdrawn periodically during the reaction course by a syringe and analyzed. Substrate conversions and epoxide product yields were quantified using biphenyl as internal standard. Each experiment was reproduced at least 2–3 times. To rule out the possibility of evaporative losses of the substrate, blank experiments
IP T
without catalyst and oxidant were carried out at the reaction temperature using the internal standard. Reaction order in catalyst. The catalyst amount was varied in the range of 3.5–45 mg, which
SC R
corresponds to 0.0015–0.01 mmol of Nb (Ti). Concentrations of other reactants were maintained constant: substrate (CyO) 0.1 M and H2O2 0.1 M.
Reaction order in substrate. The initial substrate concentration was varied between 0.02 and 0.3
N
U
M while keeping constant concentrations of H2O2 (0.1 M) and catalyst amount (0.003 mmol Nb or Ti). Reaction order in H2O2. The initial oxidant concentration was varied in the range of 0.05–0.4 M.
A
The concentration of water in these experiments was kept constant by addition of corresponding amounts
M
of H2O. The concentrations of other reactants were as follows: CyO 0.1 M, catalyst 0.003 mmol Nb or Ti.
ED
Reaction order in H2O. The initial concentration of water was varied from 0.4 to 1.7 M. Other
PT
parameters were maintained constant: alkene 0.1 M, catalyst 0.003 mmol Nb or Ti, H2O2 0.1 M. Determination of activation energies. Temperature dependences of the reaction rate were
CC E
investigated in the range of 30–80 ºC using the following reaction conditions: CyO 0.1 M, H2O2 0.1, and catalyst 0.003 mmol Nb or Ti. Initial rate determination and evaluation of rate laws. The initial rate method was employed to
A
determine the reaction orders. Initial rates were calculated as d[CyO]/dt at t = 0. The rate law was derived by applying a steady-state approximation to concentrations of all active species. Procedures for calculation of the initial rates and fitting the rates with the derived law were similar to those described earlier [40,41]. For the detailed description of the kinetic modeling procedure and derivation of the rate law, see Supporting Information (SI).
6
2.7. Instrumentation GC analyses were performed using a gas chromatograph Chromos GH-1000 equipped with a flame ionization detector and a quartz capillary column (30 m×0.25 mm) filled with BP-5. Substrate conversions and product yields in all cases were determined by the internal standard technique. GC–MS analyses were carried out using an Agilent 7000 GC/MS system equipped with a CM-WAX fused silica capillary column
IP T
(10 m × 0.25 mm × 0.25 µm). Niobium and titanium content in the solids was determined by ICP–AES using a Thermo Scientific iCAP–6500 instrument. The state of Nb and Ti in the catalysts was characterized by DR
SC R
UV–vis spectroscopy under ambient conditions using a Shimadzu UV–VIS 2501PC spectrometer. Nitrogen adsorption measurements were performed at 77 K using a NOVA 1200 instrument (Quantachrome) within the partial pressure range 10−4–1.0. The catalysts were degassed at 150°C for 24 h before the
U
measurements. Surface areas were determined by the BET analysis. Pore size distributions were
N
calculated from the adsorption branches of the nitrogen isotherms by means of the regularization
A
procedure, using reference local isotherms calculated for a cylindrical silica pore model in the framework
M
of the density functional theory (DFT) approach. Mean pore diameters were calculated as mathematical
ED
expectation values from the pore size distributions. 3. Results and Discussion
PT
3.1. Catalysts synthesis and characterization
CC E
Evaporation-induced self-assembly (EISA) is a versatile and simple methodology for inclusion of transition metals (e.g. Ti(IV) [20,42,43], Zr(IV) [44], Nb(V) [28,45], W(VI) [46], and Mo(VI) [47]) into mesoporous silicates. In this work, we employed this technique to prepare samples of Ti- and Nb-
A
containing silicates with similar metal loadings and textural properties (Table 1). Comprehensive characterization of such materials (SEM, N2 adsorption, XRD, DR UV-vis, FT-IR, Raman, and XPS) has been reported in our previous works [20,28,30]. The state of metal in the two catalysts was similar, as verified by DR UV-vis spectroscopy (compare the corresponding spectra of the initial samples in Figures 1 and 2). The broad absorptions in the range of 225–270 nm (band edges of 318–325 nm) indicate that both Ti(IV) and Nb(V) are present mostly in the
7
form of dimers and/or small oligomers [20,28 and references therein]. Recent FT-IR spectroscopic studies using pyridine and carbon monoxide as probe molecules showed that the amount of Lewis acid sites normalized to the metal content is close for Nb-silicates prepared by EISA and containing either isolated or di(oligo)meric Nb sites [30], which implies a fairly good dispersion and accessibility of the latter. 3.2. Catalysts stability toward H2O and H2O2
IP T
Agglomeration of active metal sites under turnover conditions of liquid-phase oxidation that employs aqueous hydrogen peroxide is a vital problem of mesoporous metal-silicate catalysts [10,11].
SC R
Figures 1 and 2 show changes in the DR UVvis spectra of Nb-Si and Ti-Si catalysts after treatments with boiling water and H2O2 in MeCN solution. One can notice a significant difference: while the spectrum of Nb-Si remained practically intact after the boiling procedure, indicating excellent hydrolytic stability of
U
the Nb active sites, the corresponding spectrum of Ti-Si revealed a significant long-wave shift of the band
N
edge (ca. 29 nm), which is a manifestation of increasing degree of Ti sites aggregation and formation of
A
subnanometric TiO2 clusters [10,48]. However, the band edge in the DR UV spectrum still remained below
M
355 nm, which implies that the dimension of titania clusters was less than 1.2–1.6 nm (30–70 TiO2 units)
ED
[49]. On the other hand, both catalysts revealed a rather similar sensitivity toward a cooperative effect of water and H2O2 (treatment with aqueous H2O2 in MeCN), which is manifested by a small red shift (ca. 7
PT
nm) of the DRS UV-vis absorption edge (see Figures 1 and 2). Therefore, we may conclude that active sites in mesoporous Nb-silicates possess better hydrolytic stability than active sites in Ti-silicates prepared by
CC E
the same technique while the stability toward dilute H2O2 is comparable. 3.3. Kinetics of alkene epoxidation and rate law
A
Cyclooctene was chosen as a model substrate for kinetic studies of alkene epoxidation over Nb-Si
and Ti-Si catalysts because its epoxide is rather stable toward epoxide ring opening and rearrangements. Acetonitrile was employed as solvent since the truly heterogeneous nature of the H2O2-based oxidation catalysis over Nb- [28] and Ti-silicates [20] had been previously demonstrated by hot filtration tests [39]. The rate of CyO oxidation was negligible in the absence of any catalyst. Kinetic curves of the alkene consumption over Nb-Si and Ti-Si catalysts revealed no induction periods, and the epoxidation rates were
8
not affected by the presence of molecular oxygen or light. Conventional radical scavengers, 2,6-di-tertbutyl-4-methylphenol and hydroquinone, produced no effect on the CyO oxidation rate. The reaction rates were independent on the rate of stirring of the reaction mixture in the range of 500–1000 rpm, which implies the absence of external diffusion limitations. Figure 3 shows plots of CyO consumption over Nb-Si [30] and Ti-Si catalysts versus concentrations
IP T
of substrate, oxidant, catalyst, and water. Notwithstanding the active metal (Ti or Nb), the reaction rate exhibited saturation with increasing concentration of olefin (Figure 3a) and oxidant (Figure 3b) and was
SC R
proportional to the catalyst loading (Figure 3c). The reaction over Nb-Si was slightly inhibited by H2O while, in the case of Ti-Si, practically no effect was observed (Figure 3d). The addition of CyO epoxide (ca. 40 mol.% of the amount of alkene) produced no effect on the oxidation rate, indicating that adsorption of
U
the main reaction product on both Ti and Nb sites is negligible.
N
The set of data shown in Figure 3 demonstrate that the kinetic trends for CyO transformation over
A
Nb-Si and Ti-Si catalysts are quite similar. These regularities resemble those reported for alkene
M
epoxidation with H2O2 over TS-1 [9]. Therefore, by analogy with TS-1, we may suppose that an Eley–Rideal
ED
type mechanism operates in alkene epoxidation over mesoporous Nb- and Ti-silicates. This mechanism involves adsorption of water (Eq 1) and Н2О2 (Eq 2) on metal (M = Nb or Ti) sites (the presence of Lewis
PT
acid sites of a moderate strength in the Nb-Si and Ti-Si catalysts has been recently confirmed by FTIR spectroscopy of adsorbed CO and pyridine [30]), reversible interaction of H2O2 with M to give MOOH (Eq
CC E
3), adsorption/desorption of water on/from MOOH site (Eq 4), and electrophilic oxygen transfer from
A
MOOH to С=С bond, resulting in epoxide and regenerating the initial state of the catalyst (Eq 5).
K1 M O H H 2O
M O H H 2O
1
9
K
2
M O H + H 2O 2
2
M O H (H 2 O 2 )
M O H H 2O 2
M O O H H 2O
IP T
k3
3
4
M O O H H 2O
4
ED
M
A
M O O H H 2O
N
K
U
SC R
k 3
k5
M O H e p o x id e
5
A
CC E
PT
M O O H C yO
The rate law was derived using a steady-state approximation (Eq S1, see SI for details). After fitting
the experimental data with this rate law, parameters k-3 and K4 were found to be close to zero for Ti-Si catalyst (Table S1). Previously, the same trend was found for Nb-Si catalyst [30]. Therefore, for both catalysts a simplified rate law (see SI) can be expressed by Eq 6,
10
W0
K 2k 3 H 2O 2 n (M )0
(12),
K 2k 3 H 2O 2 V 1 K 1 H 2O k 5 C yO
(6)
where n(M)0 is the total amount of Nb or Ti sites and V is volume of the reaction mixture.
IP T
Optimal values of all parameters of Eq 6 and their standard deviations are given in Table 2. The value of K1 for Ti-Si catalyst turned out comparable with its standard deviation, which resulted in a low t-
SC R
value (0.83) and, consequently, in a low statistical significance of K1 parameter. So, K1 could also be excluded from the model, giving the final rate law for Ti-catalyst expressed by Eq 7. Optimal values of
K 2k 3 H 2O 2 V 1 k 5 C yO
N
K 2 k 3 H 2O 2 n (T i) 0
(7)
M
A
W0
U
parameters of Eq 7 are provided in Table 3.
ED
Fitting of the experimental points with Eq 6 (Nb-Si) and Eq 7 (Ti-Si), which is in effect a special case of Eq 6, is shown in Figure 3. Both equations describe well the kinetic regularities observed for the Nb-
PT
and Ti-catalyzed CyO epoxidation, which strongly supports the mechanism depicted by stages (1)-(5).
CC E
The following main conclusions could be drawn from the kinetic modeling study. First, adsorption of water on M site is less favorable for Ti(IV) than for Nb(V) (compare parameters K 1 in Table 2). Second, the formation of MOOH from MOH and H2O2 is almost irreversible for both metals (k-3 = 0). Third,
A
adsorption of water on MOOH is negligible (K4 = 0) regardless of M (Nb or Ti). Fourth, the rate of MOOH formation, which can be, in principle, characterized by parameter K2k3, seems higher for Nb-Si catalyst (18.9 vs 12.9 M-1·min-1 for Ti-Si). Finally, fifth, the rate of oxygen atom transfer from MOOH to alkene is also somewhat higher for the Nb-silicate (8.5 vs 6.6 M-1·min-1). 3.5. Alkene epoxidation vs H2O2 decomposition: comparison of activation energies
11
The rate of CyO epoxidation showed a typical Arrhenius dependence (Figure 4), which implies that there was no change in the rate-limiting step over the evaluated temperature range. The estimated values of the apparent activation energy (Ea) equal to 11.9 and 14.2 kcal/mol for Nb-Si and Ti-Si catalysts, respectively, indicating that the reaction is controlled by chemical interaction rather than diffusion (for the latter, Ea is expected to be < 4 kcal/mol [50]).
IP T
Unproductive homolytic decomposition of hydrogen peroxide usually competes with selective oxidation of organic substrate over metal-silicates, especially mesoporous ones [9,10,51]. Recently, we
SC R
found that Nb-Si catalysts are significantly less active than Ti-Si ones in hydrogen peroxide degradation in the absence of any organic substrate [30]. In this work, we first compared apparent activation energies of this reaction over Nb-Si and Ti-Si catalysts. The corresponding Arrhenius plots are shown in Figure 5.
U
Significantly, Ea established for the decomposition of H2O2 over Ti-Si catalyst turned out
N
significantly lower than Ea acquired for Nb-Si: 14.8 vs 20.5 kcal/mol. Even more important is the fact that
A
Ea for CyO epoxidation and H2O2 decomposition are quite similar for the Ti-silicate (14.2 and 14.8 kcal/mol,
M
respectively) while they differ markedly for the Nb-silicate (11.9 and 20.5 kcal/mol, respectively). This
ED
finding allowed us to suggest that, similarly to homogeneous catalysts [52], higher epoxidation selectivity of Nb-silicates can be due to higher energy cost of homolytic O–O bond breaking in NbOOH intermediate.
PT
On the basis of the obtained values of the activation energies, one may predict that epoxidation (heterolytic pathway) selectivity might be improved by decreasing the reaction temperature in the case
CC E
of Nb-Si catalyst, but not in the case of Ti-Si catalyst. To verify this hypothesis, the effect of temperature on oxidation selectivity has been evaluated for both catalysts using cyclohexene as model substrate
A
(section 3.6). CyO and CyH have a similar nucleophilicity of the C=C double bond [53] and normally reveal similar epoxidation rates [28,53]. 3.6. Effect of temperature on epoxidation of cyclohexene In contrast to CyO, cyclohexene possesses highly reactive C–H bonds in the allylic position, which readily react with radical species, leading to allylic oxidation products [54]. Therefore, CyH can be used as a test substrate that enables evaluation of contributions of heterolytic and homolytic oxidation pathways
12
into the overall oxidation process on the basis of quantification of the specific oxidation products (Scheme 1). Figure 6 shows the effect of temperature on selectivity of CyH oxidation achieved at a similar level of substrate conversion over Ti-Si and Nb-Si catalysts. One can notice that, for Nb-Si, the contribution of the heterolytic pathway oxidation products increased drastically (from 70 to 85%) upon decreasing the reaction temperature from 50 to 30 °C. On the contrary, the effect of temperature on the selectivity turned out minor in the case of Ti-Si catalyst, in accordance with what was predicted by the values of the
IP T
apparent activation energies (section 3.5). The data presented in Figure 6 also show that Nb-Si catalyst is more active than Ti-Si catalyst since the reaction time required to reach the same level of conversion is
SC R
significantly lower for Nb.
3.7. Origin of different regioselectivity in limonene oxidation over Nb-Si and Ti-Si catalysts
U
Finally, we would like to discuss possible reasons for different regioselectivities observed in
N
limonene epoxidation over Ti- and Nb-silicates. Guidotti and coworkers were the first who reported on
A
unexpected regioselectivity toward epoxidation of the less electron–rich, exocyclic C=C double bond in
M
limonene and carveol over silica-grafted Nb catalysts [25,26]. On the contrary, another group identified
ED
endocyclic, 1,2-epoxide as the main product formed over Nb-MSU–X [24]. To rationalize this seeming discrepancy, we explored the solvent effect on the epoxidation of limonene over Nb-Si catalysts and found
PT
that the regioselectivity strongly depends on the choice of solvent [28]. While exocyclic epoxide was the main product in MeCN (exo/endo = 73/27, a similar ratio was also found for silica-grafted Nb catalysts
CC E
[25]), endocyclic one predominated in MeOH (exo/endo = 13/87). Note that the 1,2-epoxide was always a predominated isomer (ca. 90%) formed over Ti-silicates [25,26].
A
To understand better the solvent-dependent regioselectivity of Nb-silicates, we performed hot
filtration tests [39] for limonene oxidation over Nb-Si catalyst in both MeCN and MeOH (Figure 7). In line with the previous results [25,28], Nb-silicate behaved as a truly heterogeneous catalyst in acetonitrile (the reaction stopped completely after removal of the catalyst), while the contribution of homogeneous catalysis manifested by continuation of the reaction in the filtrate took place in methanol (compare Figures 7a and 7b). Indeed, the elemental analysis revealed substantial leaching of Nb (32 ppm) in MeOH
13
while it did not exceed 1 ppm in MeCN. A similar trend was documented for mesoporous Ti-Si catalysts, which suffer Ti leaching in alcoholic solvents but are rather stable in acetonitrile [10,11]. Therefore, we may suppose that the inversion of regioselectivity observed in the oxidation of limonene over Nb-silicates can be related, at least partially, to the solvent-dependent stability of Nb-Si catalysts and different nature of the catalysis in MeCN and MeOH. We cannot also exclude that Nb and Ti hydroperoxo species formed on the catalyst surface in the presence of MeOH have a similar structure, thereby leading to similar
IP T
regioselectivity.
SC R
Then the question that arises in relation to the heterogeneously catalyzed oxidation of limonene in MeCN is why the less substituted (nucleophilic) C=C bond is primarily epoxidized over Nb-silicates while Ti-silicates oppositely favor epoxidation of more substituted, electron-rich C=C bond, as one would expect
U
for electrophilic oxygen transfer, typical of alkene epoxidation catalyzed by complexes of d0-metals [53].
N
Recently, we found that Nb-Si catalysts are able to perform epoxidation of electron-deficient C=C bonds
A
in α,β-unsaturated carbonyl compounds through the Payne type oxidation that involves rate-limiting
M
oxidation of MeCN to peroxycarboximidic acid (the real epoxidizing species) [38] and this ability is due to the presence of weak basic sites on the surface of Nb-silicates [30]. This oxidation route leads to the
ED
formation of acetamide in the amount close to that of epoxide. However, no acetamide was detected in the oxidation of limonene over Nb-Si catalysts, which allowed us to discard the role of Payne oxidation in
PT
the unusual regioselectivity of limonene epoxidation.
CC E
The kinetic modeling study undertaken in this work showed that the general mechanism of alkene epoxidation of Nb- and Ti-silicates is similar and can be envisaged by the sequence of similar reaction steps (1)–(5). Nowadays, it is widely accepted that hydroperoxo titanium species (TiOOH) is responsible
A
for alkene epoxidation over Ti-Si catalysts [see 9,10,51,55]. However, the coordination mode of the hydroperoxo group (η1 vs η2) in TiOOH remains under debates and the exact structure of the titanium hydroperoxo species is still unknown despite numerous spectroscopic and computational studies. Recently, we demonstrated that basic additives lead to deactivation of Nb-silicates, likewise Ti-silicates [56], producing a strong rate-retarding effect on alkene epoxidation with H2O2 [30]. This finding strongly supports a niobium hydroperoxo (i.e. protonated peroxo) species, NbOOH, as the active species operating
14
in the Nb-catalyzed epoxidation of olefins. This species is manifested by a DR UV-vis absorption band at 307 nm, which shifts to 293 nm upon addition of base [30]. The different regioselectivity of limonene epoxidation in MeCN has led us to a suggestion that the structure of niobium and titanium hydroperoxo species MOOH that participate in the oxygen transfer step (Eq 5) is, most likely, different. Guidotti and coworkers first suggested that the observed peculiarity of regioselectivity in limonene epoxidation over mesoporous Nb-silicates is to be attributed mainly to steric factors [26]. Indeed, the unusual
IP T
regioselectivity toward exocyclic limonene epoxide was no longer observed when a nonporous Nb-SiO2 catalyst, prepared by grafting of Nb(V) centers onto a pyrogenic nanosized silica (Aerosil 380), was
SC R
employed.
We may tentatively suppose that NbOOH has a η2 coordination of the hydroperoxo group while a
U
η1 coordination mode is realized in TiOOH. Alternatively, different mechanisms of the oxygen transfer step
N
(Eq 5), i.e. oxygen transfer from α or β positions of the hydroperoxo group, may account for the opposed
A
regioselectivity. Recently, unprecedented β-oxygen transfer mechanism has been established by the
M
computational technique for alkene epoxidation catalyzed by Ti-containing polyoxometalates [57].
ED
4. Conclusions
The present study demonstrated that epoxidation of electron-rich alkenes catalyzed by
PT
mesoporous Nb- and Ti-silicates involves similar stages, typical of Eley–Rideal mechanism. The higher heterolytic pathway selectivity of Nb-silicates is, at least partially, due to a higher energy cost of homolytic
CC E
O–O bond breaking in NbOOH intermediate relative to TiOOH one, which is manifested by the smaller rates and activation energy of H2O2 decomposition over Nb-silicates. A significant gap between the
A
activation energies established for alkene (cyclooctene) epoxidation and H2O2 degradation over Nbcatalysts (11.9 and 20.5 kcal/mol, respectively) makes possible substantial improvement of heterolytic pathway selectivity by decreasing the reaction temperature. On the contrary, this approach does not work with mesoporous Ti-silicates, for which the activation energy values of cyclooctene epoxidation and H2O2 decomposition are similar (14.2 and 14.8 kcal/mol, respectively). The difference in regioselectivity of limonene epoxidation over Nb- and Ti-silicates established in MeCN, where the nature of the catalysis is
15
truly heterogeneous, indicates different structures of the active ‘NbOOH’ and ‘TiOOH’ species and/or different mechanisms of the oxygen transfer step. On the other hand, the inversion of regioselectivity observed for Nb catalyst in MeOH may be caused by Nb leaching in this solvent and/or changes in the structure of Nb hydroperoxo species, making it more close to that of TiOOH, which explains similar regioselectivities of Nb and Ti catalysts documented for alcoholic solvents. We hope that further comparative spectroscopic and theoretical studies may help in understanding the structure of active Nb
IP T
and Ti hydroperoxo species and their transformations. Such investigations using Nb- and Ti-substituted polyoxometalates as tractable soluble model compounds are in progress in our group and will be
SC R
published elsewhere.
Acknowledgements. The help of Dr. M. V. Shashkov and Dr. S. A. Yashnik in GC–MS and DRS UV-vis
U
measurements, respectively, is appreciated. This work was carried out in the framework of budget project
N
No. 0303-2016-0005 for the Boreskov Institute of Catalysis and partially supported by the Russian
A
Foundation for Basic Research (grant N 16-03-00827).
M
References
ED
References
[1] Modern Heterogeneous Oxidation Catalysis: Design, Reactions and Characterization, N. Mizuno (Ed.), Wiley-VCH, Weinheim, 2009.
PT
[2 ] Liquid Phase Oxidation via Heterogeneous Catalysis: Organic Synthesis and Industrial Applications, M.G. Clerici, O.A. Kholdeeva (Eds.), Wiley, Hoboken, 2013.
CC E
[3] Handbook of Advanced Methods and Processes in Oxidation Catalysis, D. Duprez , F. Cavani (Eds.), Imperial College Press, London, 2014. [4] F. Cavani, J. H. Teles, Sustainability in Catalytic Oxidation: An Alternative Approach or a Structural Evolution? ChemSusChem 2 (2009) 508–534.
A
[5] C.W. Jones, Application of Hydrogen Peroxide and Derivatives, Royal Society of Chemistry, Cambridge, 1999. [6] G. Strukul, A. Scarso, in: M.G. Clerici, O.A. Kholdeeva (Eds.), Liquid Phase Oxidation via Heterogeneous Catalysis: Organic Synthesis and Industrial Applications; Wiley, Hoboken, 2013; Ch. 1., pp. 1–20. [7] J.M. Campos-Martin, G. Blanco-Brieva, J.L.G. Fierro, Angew. Chem. Int. Ed. 45(2006) 6962– 6984.
16
[8] M. Taramasso, G. Perego, B. Notari, Snamprogetti S.p.A. assignee. US Patent 4,410,501, 1983. [9] For recent review see M.G. Clerici, M.E. Domine, in: M.G. Clerici, O.A. Kholdeeva (Eds.), Liquid Phase Oxidation via Heterogeneous Catalysis: Organic Synthesis and Industrial Applications, Wiley, Hoboken, 2013, Ch. 2, p. 21–93.
[10] O.A. Kholdeeva, in: M.G. Clerici, O.A. Kholdeeva (Eds.), Liquid Phase Oxidation via Heterogeneous Catalysis: Organic Synthesis and Industrial Applications, Wiley, Hoboken, 2013; Ch. 4, pp. 127–219.
[12] P.T.Tanev, M. Chibwe, T. J. Pinnavaia, Nature 368 (1994) 321−323. [13] A. Sorokin, A. Tuel, Catal. Today, 57 (2000) 45−59.
IP T
[11] O.A. Kholdeeva, Catal. Sci. Technol. 4 (2014) 1869–1889.
SC R
[14] N.N.Trukhan, V.N. Romannikov, E.A. Paukshtis, A.N. Shmakov, O.A. Kholdeeva, J. Catal. 202 (2001) 110−117. [15] O.A. Kholdeeva, O.V. Zalomaeva, Coord. Chem. Rev. 306 (2016) 302−330.
N
U
[16] O.A. Kholdeeva, in: B. Cornils, W.A. Hermann, M. Beller, R. Paciello (Eds.), Applied Homogeneous Catalysis with Organometallic Compounds, 3rd ed., Wiley-VCH, 2017, pp. 545– 570.
M
A
[17] O.A. Kholdeeva, M.S. Melgunov, A.N. Shmakov, N.N. Trukhan, V.V. Kriventsov, V.I. Zaikovskii, M.E. Malyshev, V.N. Romannikov, Catal. Today, 91-92 (2004) 205–209. [18] P. Wu, T. Tatsumi, T. Komatsu, T. Yashima, Chem. Mater. 14 (2002) 1657–1664.
ED
[19] F.-S. Xiao, Top. Catal. 35 (2005) 9–24.
[20] I.D. Ivanchikova, M.K. Kovalev, M.S. Mel’gunov, A.N. Shmakov, O.A. Kholdeeva, Catal. Sci. Technol. 4 (2014) 200−207.
PT
[21] I. Nowak, B. Kilos, M. Ziolek, A. Lewandowska, Catal. Today 78 (2003) 487–498. [22] I. Nowak, M. Ziolek, Micropor. Mesopor. Mater. 78 (2005) 281–288.
CC E
[23] F. Somma, G. Strukul, Catal. Lett. 107 (2006) 73–81. [24] A. Feliczak-Guzik, I. Nowak, Catal. Today 142 (2009) 288–292.
[25] A. Gallo, C. Tiozzo, R. Psaro, F. Carniato, M. Guidotti, J. Catal. 298 (2013) 77–83.
A
[26] C. Tiozzo, C. Bisio, F. Carniato, M. Guidotti, Catal. Today 235 (2014) 49–57. [27] A. Ramanathan, H. Zhu, R. Maheswari, P.S. Thapa, B. Subramaniam, Ind. Eng. Chem. Res. 54 (2015) 4236–4242.
[28] I.D. Ivanchikova, N.V. Maksimchuk, I.Y. Skobelev, V.V. Kaichev, O.A. Kholdeeva, J. Catal. 332 (2015) 138–148. [29] N.E. Thornburg, A.B. Thompson, J.M. Notestein, ACS Catal. 5 (2015) 5077–5088.
17
[30] I.D. Ivanchikova, I. Y. Skobelev, N. V. Maksimchuk, E. A. Paukshtis, M. V. Shashkov, O. A. Kholdeeva, J. Catal. 356C (2017) 85–99. [31] N.E. Thornburg, J.M. Notestein, ChemCatChem 9 (2017) 3714–3724. [32] S. Dworakowska, C. Tiozzo, M. Niemczyk-Wrzeszcz, P. Michorczyk, N. Ravasio, R. Psaro, D. Bogdał, M. Guidotti, J. Clean. Prod. 166 (2017) 901–909. [33] M. P. Kapoor, A. Raj, Stud. Surf. Sci. Catal. 129 (2000) 327–334.
IP T
[34] M. Ziolek, I. Sobczak, P. Decyk, K. Sobanska, P. Pietrzyk, Z. Sojka, Appl. Catal. B: Environ. 164 (2015) 288–296. [35] N.E. Thornburg, S.L. Nauert, A.B. Thompson, J.M. Notestein, ACS Catal. 6 (2016) 6124– 6134.
SC R
[36] D.T. Bregante, P. Priyadarshini, D.W. Flaherty, J. Catal. 348 (2017) 75–89. [37] D.T. Bregante, D.W. Flaherty, J. Am. Chem. Soc. 139 (2017) 6888−6898. [38] G.B. Payne, P.H. Deming, P.H. Williams, J. Org. Chem. 26 (1961) 659–663.
U
[39] R.A. Sheldon, M. Wallau, I.W.C.E. Arends, U. Schuchardt, Acc. Chem. Res. 31 (1998) 485– 493
A
N
[40] I.Y. Skobelev, O.V. Zalomaeva, O.A. Kholdeeva, J.M. Poblet, J.J. Carbó, Chem. Eur. J. 21 (2015) 14496–14506.
M
[41] I.Y. Skobelev, V.Yu. Evtushok, O.A. Kholdeeva, N.V. Maksimchuk, R.I. Maksimovskaya, J. M. Ricart, J.M. Poblet, J.J. Carbó, ACS Catal. 7 (2017) 8514–8523.
ED
[42] M. Ogawa, K. Ikeue, M. Anpo, Chem. Mater. 13 (2001) 2900–2904. [43] N. Hüsing, B. Launay, D. Doshi, G. Kickelbick, Chem. Mater. 14 (2002) 2429–2432.
PT
[44] H. Zhu, R. Maheswari, A. Ramanathan, B. Subramaniam, Micropor. Mesopor. Mater. 223 (2016) 46– 52.
CC E
[45] A. Ramanathan, H. Zhu, R. Maheswari, B. Subramaniam, Micropor. Mesopor. Mater. 261 (2018) 158–163. [46] J.-F. Wu, A. Ramanathan, B. Subramaniam, J. Catal. 350 (2017) 182–188. [47] A. Ramanathan, J.-F. Wu, R. Maheswari, Y. Hu, B. Subramaniam, Micropor. Mesopor. Mater. 245 (2017) 118–125.
A
[48] X. Gao, I.E. Wachs, Catal. Today 51 (1999) 233–254. [49] B.J. Aronson, C.F. Blanford, A. Stein. Chem. Mater. 9 (1997) 2842–2851. [50] P. Atkins, J. De Paula, Elements of Physical Chemistry, sixth edition, Oxford University Press, 2013. [51] C.W. Yoon, K.F. Hirsekorn, M.L. Neidig, X. Yang, T.D. Tilley, ACS Catal. 1 (2011) 1665–1678. [52] N. Antonova, J. J. Carbo´, U. Kortz, O.A. Kholdeeva, J.M. Poblet, J. Am. Chem. Soc. 132 (2010) 7488–7497. 18
[53] B.G. Donoeva, T.A. Trubitsina, N.S. Antonova, J.J. Carbó, J.M. Poblet, G. Al Kadamany, U. Kortz, O.A. Kholdeeva, Eur. J. Inorg. Chem. (2010) 5312–5317. [54] R.A. Sheldon, J.K. Kochi, Metal-Catalyzed Oxidations of Organic Compounds, New York, Academic Press, 1981. [55] S. Bordiga, E. Groppo, G. Agostini, J.A. van Bokhoven, C. Lamberti, Chem. Rev. 113 (2013) 1736–1850. [56] M.G. Clerici, Appl. Catal. 68 (1991) 249–261.
SC R
IP T
[57] P. Jiménez-Lozano, I.Y. Skobelev, O.A. Kholdeeva, J.M. Poblet, J.J. Carbó, Inorg. Chem. 55 (2016) 6080–6084.Figure Caption
Nb-Si Nb-Si + H2O2
U
4
Nb-Si + H2O
N A
2
M
F(R)
3
ED
1
0
250
300
350
400
, nm
PT
200
Figure 1. DR UV–vis spectra of the initial Nb-Si catalyst and after treatments with aqueous H2O2 in MeCN
A
CC E
(25 oC, 1 h) and with boiled water (6 h).
19
F(R)
8
Ti-Si Ti-Si + H2O2
6
Ti-Si + H2O
4
0 200
250
300
350
SC R
, nm
400
IP T
2
Figure 2. DR UV–vis spectra of the initial Ti-Si catalyst and after treatments with aqueous H2O2 in MeCN
A
CC E
PT
ED
M
A
N
U
(25 oC, 1 h) and with boiled water (6 h).
20
2,0
3
W0x10 , M/min
0,5
0,0
0,3
2,0
5,0
1,8 W0x10 , M/min
6,0
4,0
3
3,0 2,0
3 4 5 6 7 8 3 n(TM)x10 , mmol
1,0
9 10
N
2
1,2
0,2 0,3 [H2O2], M
0,4
0,5
(d) 1,0 1,5 [H2O], M
2,0
A
1
1,4
0,1
0,8
(c) 0
1,6
U
1,0 0,0
(b)
IP T
0,2 [CyO], M
3
W0x10 , M/min
0,1
1,0
0,0
(a) 0,0
1,5
SC R
3
W0x10 , M/min
2,5
3,2 2,8 2,4 2,0 1,6 1,2 0,8 0,4 0,0
M
Figure 3. Plots of the initial rate of CyO epoxidation (MeCN, 50 oC) versus concentration of substrate (a), oxidant (b), catalyst (c), and water (d). Experimental points: (■) and (●) for Ti-Si
ED
and Nb-Si catalysts, respectively. Fitting with: (-) Eq 7 for Ti-Si and (-) Eq 6 for Nb-Si. For
A
CC E
PT
reaction conditions, see Experimental, section 2.5.
21
Ea = 14.2 ± 0.8 kcal/mol
ln W0
-5 -6 -7
Ea = 11.9 ± 0.6 kcal/mol -9 0,0028
0,0030
0,0032
SC R
1/T, K
0,0034
-1
IP T
-8
Figure 4. Arrhenius plots for CyO epoxidation over Ti-Si (■) and Nb-Si (●) catalysts. Reaction conditions:
U
CyO 0.1 M, H2O2 0.1 M, catalyst 0.003 mmol Nb or Ti, MeCN 1 mL.
Ea = 14.8 ± 1.1 kcal/mol
N
-4
A
-5
-7 -8
M
ln W0
-6
ED
-9
-10
Ea = 20.5 ± 0.3 kcal/mol 0,0029
0,0030 1/T, K
0,0031
0,0032
-1
CC E
PT
0,0028
Figure 5. Arrhenius plots for H2O2 decomposition over Ti-Si (■) and Nb-Si (●) catalysts. Reaction conditions:
A
Н2О2 0.4 М, Nb 0.02 mmol, MeCN 7 mL.
22
CyH conversion Time
Selectivity: heterolytic pathway homolytic pathway 3
60 2 40
Time, h
1 20
0
IP T
Conversion/ selectivity, %
80
0
Nb
o
Ti
Nb
o
Ti
50 C
SC R
30 C
Figure 6. Effect of temperature on selectivity of CyH oxidation over Nb-Si and Ti-Si catalysts. Reaction
A
CC E
PT
ED
M
A
N
U
conditions: CyH 0.1 mmol, H2O2 0.1 mmol, 0.003 mmol Nb or Ti, MeCN 1 mL.
23
100
Hot filtration 90
80
70 0
20
40
60
80
100
120
140
IP T
Limonene consumption , %
(a)
SC R
Time, min
U
Hot filtration
N
90
80
A
Limonene consumption , %
(b)
100
50
M
0
100
150
Time, min
ED
Figure 7. Hot catalyst filtration tests for limonene oxidation with H2O2 over Nb-Si catalyst in (a) MeCN and (b) MeOH. Reaction conditions: limonene 0.1 mmol, H2O2 0.1 mmol, catalyst 0.003 mmol Nb,
A
CC E
PT
solvent 1 mL, 50 oC.
24
IP T
heterolytic pathway
homolytic pathway
A
CC E
PT
ED
M
A
N
U
SC R
Scheme 1. Products typical of CyH oxidation via heterolytic and homolytic pathways.
25
Table 1. Metal loading and texture of Nb- and Ti-silicates prepared by EISA.
M,a
SBET,
Vp,b
Dp,c
wt.%
m2/g
cm3/g
nm
Nb-Si
1.96
1190
0.56
2.8
Ti-Si
2.35
1270
0.54
2.7
Catalyst
Based on elemental analysis data for calcined samples.
b
Mesopore volume.
c
Average pore diameter.
SC R
IP T
a
N
U
Table 2. Optimal values of parameters of Eq 6 for catalysts Nb-Si and Ti-Si. Value (standard deviation) for
A
Parameter
M
catalyst Nb-Si K1, M-1 K2k3, (M·min)-1
catalyst Ti-Si 0.15 (0.18)
18.9 (5.5)
11.0 (2.3)
8.5 (1.1)
7.9 (1.3)
CC E
PT
k5, (M·min)-1
ED
0.8 (0.4)
Value (standard deviation) for
A
Table 3. Optimal values of parameters of Eq 7. Parameter
Value (standard deviation)
K2k3, (M·min)-1
12.9 (2.4)
k5, (M·min)-1
6.6 (0.9)
26
27
A ED
PT
CC E
IP T
SC R
U
N
A
M
S0920-5861(18)30375-4 https://doi.org/10.1016/j.cattod.2018.04.002 CATTOD 11358
To appear in:
Catalysis Today
Received date: Revised date: Accepted date:
18-12-2017 27-2-2018 1-4-2018
Please cite this article as: Kholdeeva OA, Ivanchikova ID, Maksimchuk NV, Skobelev IY, H2 O2 -based selective epoxidations: Nb-silicates versus Ti-silicates, Catalysis Today (2010), https://doi.org/10.1016/j.cattod.2018.04.002 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
H2O2-based selective epoxidations: Nb-silicates versus Ti-silicates Oxana A. Kholdeeva,1,2,* Irina D. Ivanchikova,1 Nataliya V. Maksimchuk,1,2 Igor Y. Skobelev1 1
Boreskov Institute of Catalysis, Lavrentieva ave. 5, Novosibirsk, 630090, Russia 2
Novosibirsk State University, Pirogova st. 2, Novosibirsk, 630090, Russia
IP T
Fax: (+7)-(383)-330-9573; phone: (+7)-(383)-326-9433; e-mail: [email protected]
CC E
PT
ED
M
A
N
U
SC R
Graphical Abstract
Highlights
A
Mesoporous Nb-silicates reveal better hydrolytic stability than Ti-silicates. Alkene epoxidation over both Nb- and Ti-silicates proceeds via Eley–Rideal mechanism. The activation energy of cyclooctene epoxidation is lower for Nb-silicate. The activation energy of hydrogen peroxide decomposition is lower for Ti-silicate. The higher epoxidation selectivity of Nb-catalysts is related to their lower activity in H2O2 decomposition.
Abstract This paper aims at a comparison of the stability and catalytic performance of niobium- and titaniumcontaining mesoporous silicates prepared by evaporation-induced self-assembly (EISA) in selective
1
oxidations with hydrogen peroxide. The catalysts comprised evenly dispersed dimeric and/or small oligomeric active sites (M = Nb(V) or Ti(IV)), and the Nb sites revealed better hydrolytic stability than the Ti ones, as demonstrated by DRS UV-vis. An attempt has been made to rationalize the differences observed in chemo- and regioselectivity of alkene epoxidation using kinetic modeling tools. The rate law established for cyclooctene epoxidation over the Nb- and Ti-catalysts is consistent with an Eley–Rideal mechanism that involves adsorption of Н2О2 on М sites, interaction between M and Н2О2 to afford a
IP T
hydroperoxo species ‘MOOH’ and water, followed by oxygen transfer from ‘MOOH’ to С=С bond, producing epoxide and regenerating M. The higher epoxidation (heterolytic pathway) selectivity of the
SC R
Nb catalysts is, at least partially, related to their lower activity in H2O2 unproductive decomposition. The apparent activation energy of H2O2 degradation is higher for Nb-silicate (20.5 vs 14.8 kcal/mol for Ti-
U
silicate), while the activation energies of epoxidation follow an opposite trend (11.9 vs 14.2 kcal/mol). The
N
heterolytic pathway selectivity of Nb catalysts can be greatly improved by decreasing the reaction
A
temperature. The difference in regioselectivity of limonene epoxidation over Nb- and Ti-silicates observed
M
in MeCN, where the nature of the catalysis is truly heterogeneous, indicates different structures of the active ‘NbOOH’ and ‘TiOOH’ species and/or different mechanisms of the oxygen transfer step. The
ED
inversion of regioselectivity observed for Nb catalyst in MeOH may be caused by Nb leaching in this solvent and/or changing the structure of ‘NbOOH’.
CC E
mechanism
PT
Keywords: alkene epoxidation; niobium-silicates; titanium-silicates; hydrogen peroxide; kinetics;
1. Introduction
The selective transformation of organic compounds into their oxygenated derivatives using
A
leaching-resistant truly heterogeneous catalysts and green oxidants is a challenging goal of oxidation catalysis [1-2]. In particular, hydrogen peroxide is a clean and atom-efficient oxidant that produces water as the sole side product [5-6]. The advent in the late 1980s of hydrophobic titanium silicalite-1 (TS-1) isostructural with ZSM-5 [8] has become a revolution in the field of heterogeneous oxidation catalysis since this zeotypes material appeared capable of heterolytically activating H2O2 and oxygenating small organic substrates [9]. The demands of the fine chemicals industry stimulated an intensive research
2
activity on the development of mesoporous titanium-silicates [10,11]. However, the presence of numerous silanol groups on the surface of mesoporous silica (silicates) favors adsorption of water and hydrogen peroxide at the expense of nonpolar organic substrates, which leads to significant unproductive decomposition of the oxidant, reduction of its utilization efficiency, and contribution of homolytic oxidation products. On the other hand, the hydrophilic environment of Ti sites in the mesoporous catalysts may offer advantages for bulky substrates with intermediate polarity, e.g. alkylated phenols [12-
IP T
13]. The potential of mesoporous titanium-silicates for the production of substituted benzoquinones using H2O2 has been recently discussed [15,16]. Although several types of hydrothermally stable and leaching-
SC R
tolerant mesoporous Ti-silicates have been developed [10,11,17-18], the stability of Ti active sites toward progressive agglomeration under turnover conditions still remains a major concern if aqueous H2O2 is
U
employed.
N
In recent years, niobium-containing silicates have attracted great attention of the selective
A
oxidation community. Various types of mesoporous Nb-silicates have been synthesized and tested in
M
oxidations with H2O2 [21-22]. Some of them turned out more selective than their Ti analogs in epoxidation of alkenes, including those containing highly reactive allylic hydrogen atoms, e.g. cyclohexene and
ED
limonene, which indicated a better ability of the Nb catalysts to accomplish selective oxidation with H2O2 via a heterolytic mechanism [25,29,30]. An increase in activity and H2O2 efficiency has been also noticed
PT
for mesostructured silicates with framework-incorporated Nb(V) compared to framework-incorporated
CC E
titanium [33]. Moreover, Nb-silicates, in contrast to their Ti counterparts, were found to be able to catalyze epoxidation of both electron-rich and electron-deficient C=C bonds and demonstrate unusual regioselectivity toward epoxidation of less substituted (less nucleophilic) double bond in limonene
A
[25,26,28].
So far, a few attempts at rationalizing the catalytic performance of Nb(V) have been reported
[30,34-35], but conclusions of these studies look somewhat contradictory. Recently, we proved that epoxidation of C=C bonds in α,β-unsaturated carbonyl compounds proceeds through the so-called Payne oxidation [38], which involves rate-limiting oxidation of the solvent molecule (MeCN) and formation of peroxycarboximidic acid Н3СC(=NH)OOH that is the real epoxidizing species in the system [30]. Weak basic
3
sites present in Nb-silicates are most likely responsible for this specific catalytic activity. Kinetic and spectroscopic studies identified a hydroperoxo niobium species (absorption band at 307 nm in DRS UVvis) as the active species that performs epoxidation of electron-rich double bonds in alkenes [30]. In the present work, we made an effort to gain further insights into the understanding of the different catalytic performance of mesoporous Nb- and Ti-silicates in H2O2-based selective oxidation
IP T
through careful evaluation of kinetics of alkene epoxidation and H2O2 decomposition over these two types of catalysts. We also provide here a comparison of the stability of the Nb- and Ti-catalysts prepared by
SC R
the same, evaporation-induced self-assembly (EISA) methodology using DR UV-vis spectroscopic technique and leaching tests.
U
2. Experimental
N
2.1. Materials and catalysts
A
Cetyltrimethylammonium bromide (CTAB, 99+%), titanium (IV) ethoxide (TEOT), acetylacetone
M
(acac, 99%), and tetraethyl orthosilicate (TEOS, 98+%) were purchased from Aldrich. Niobium(V) ethoxide (99.95%) was used as received from Acros. Acetonitrile (HPLC grade, Panreac) was dried and stored over
ED
activated 4 Å molecular sieves. Cyclohexene (CyH), limonene, and cis-cyclooctene (CyO) were purchased from Aldrich and passed through neutral alumina prior to use. All other reagents and solvents were the
PT
best reagent grade and were used without further purification. The concentration of H2O2 (30 or 50 wt%
CC E
in water) was determined iodometrically prior to use. Deionized water (EASY pure, RF, Barnsted) was used for the preparation of catalysts. A mesoporous titanium-silicate (Ti-Si) was prepared by the EISA technique following a protocol
A
described previously [20]. TEOT modified with acac was used as Ti source. A mesoporous niobium-silicate (Nb-Si) was prepared with the same EISA approach using niobium(V) ethoxide [28]. The catalysts were characterized by elemental analysis, low-temperature N2 adsorption, and DR UV–vis techniques. Before all measurements, the catalyst samples were calcined at 550 °C. 2.2.Evaluation of catalyst stability
4
To evaluate the catalyst stability toward water and H2O2, the catalyst samples were subjected to two types of treatments: 1) boiling in water for 6 h and 2) stirring with a MeCN solution of H2O2 (the latter was taken as 30% solution in water) for 1 h (100 mg of catalyst, 0.11 M H2O2, 20 mL MeCN, 25 oC). After the treatments, the samples were separated by filtration, dried in air and then calcined at 550 oC directly before DRS UV-vis measurements.
IP T
2.3. Catalytic oxidations Catalytic oxidations of cyclohexene were carried out at 30 and 50 oC in MeCN (1 mL) using the
SC R
following concentrations of the reagents: alkene 0.1 M, H2O2 0.1 M (taken as 50% solution in water), and catalyst (0.003 mmol Nb or Ti). Catalytic oxidations of limonene were performed at 50 oC in MeCN or MeOH (1 mL) using alkene 0.1 M, H2O2 0.1 M (taken as 30% solution in water), and catalyst (0.003 mmol
U
Nb).The product yields and substrate conversions were quantified by GC using biphenyl as internal
A
M
2.4. Nature of catalysis and leaching tests
N
standard.
The nature of catalysis (truly heterogeneous versus homogeneous) over Nb-silicates in different
ED
solvents (MeCN and MeOH) was verified by hot filtration tests [39]. The amount of metal leached into
PT
solution during the reaction course was determined by ICP–AES. 2.5. Hydrogen peroxide decomposition
CC E
Decomposition of H2O2 (0.4 M) was studied in the absence of organic substrate at 80 oC in MeCN
(5 mL) in the presence of either Nb-Si or Ti-Si catalyst (0.015 mmol Nb or Ti). Aliquots of 0.2 mL were taken during the reaction course, and H2O2 concentration was determined by iodometric titration. Four parallel
A
experiments were carried out. The temperature dependence of the decomposition rate was studied in the range of 40–80 ºC. 2.6. Kinetic study Kinetic experiments were performed in temperature-controlled glass vessels under vigorous stirring (600 rpm). Reactions were initiated by addition of H2O2 into a MeCN solution containing organic
5
substrate (CyO), catalyst, and internal standard for GC. The total volume of the reaction mixture was 1 mL. The reaction temperature was 50 ºC. Samples (1 μL) of the reaction mixture were withdrawn periodically during the reaction course by a syringe and analyzed. Substrate conversions and epoxide product yields were quantified using biphenyl as internal standard. Each experiment was reproduced at least 2–3 times. To rule out the possibility of evaporative losses of the substrate, blank experiments
IP T
without catalyst and oxidant were carried out at the reaction temperature using the internal standard. Reaction order in catalyst. The catalyst amount was varied in the range of 3.5–45 mg, which
SC R
corresponds to 0.0015–0.01 mmol of Nb (Ti). Concentrations of other reactants were maintained constant: substrate (CyO) 0.1 M and H2O2 0.1 M.
Reaction order in substrate. The initial substrate concentration was varied between 0.02 and 0.3
N
U
M while keeping constant concentrations of H2O2 (0.1 M) and catalyst amount (0.003 mmol Nb or Ti). Reaction order in H2O2. The initial oxidant concentration was varied in the range of 0.05–0.4 M.
A
The concentration of water in these experiments was kept constant by addition of corresponding amounts
M
of H2O. The concentrations of other reactants were as follows: CyO 0.1 M, catalyst 0.003 mmol Nb or Ti.
ED
Reaction order in H2O. The initial concentration of water was varied from 0.4 to 1.7 M. Other
PT
parameters were maintained constant: alkene 0.1 M, catalyst 0.003 mmol Nb or Ti, H2O2 0.1 M. Determination of activation energies. Temperature dependences of the reaction rate were
CC E
investigated in the range of 30–80 ºC using the following reaction conditions: CyO 0.1 M, H2O2 0.1, and catalyst 0.003 mmol Nb or Ti. Initial rate determination and evaluation of rate laws. The initial rate method was employed to
A
determine the reaction orders. Initial rates were calculated as d[CyO]/dt at t = 0. The rate law was derived by applying a steady-state approximation to concentrations of all active species. Procedures for calculation of the initial rates and fitting the rates with the derived law were similar to those described earlier [40,41]. For the detailed description of the kinetic modeling procedure and derivation of the rate law, see Supporting Information (SI).
6
2.7. Instrumentation GC analyses were performed using a gas chromatograph Chromos GH-1000 equipped with a flame ionization detector and a quartz capillary column (30 m×0.25 mm) filled with BP-5. Substrate conversions and product yields in all cases were determined by the internal standard technique. GC–MS analyses were carried out using an Agilent 7000 GC/MS system equipped with a CM-WAX fused silica capillary column
IP T
(10 m × 0.25 mm × 0.25 µm). Niobium and titanium content in the solids was determined by ICP–AES using a Thermo Scientific iCAP–6500 instrument. The state of Nb and Ti in the catalysts was characterized by DR
SC R
UV–vis spectroscopy under ambient conditions using a Shimadzu UV–VIS 2501PC spectrometer. Nitrogen adsorption measurements were performed at 77 K using a NOVA 1200 instrument (Quantachrome) within the partial pressure range 10−4–1.0. The catalysts were degassed at 150°C for 24 h before the
U
measurements. Surface areas were determined by the BET analysis. Pore size distributions were
N
calculated from the adsorption branches of the nitrogen isotherms by means of the regularization
A
procedure, using reference local isotherms calculated for a cylindrical silica pore model in the framework
M
of the density functional theory (DFT) approach. Mean pore diameters were calculated as mathematical
ED
expectation values from the pore size distributions. 3. Results and Discussion
PT
3.1. Catalysts synthesis and characterization
CC E
Evaporation-induced self-assembly (EISA) is a versatile and simple methodology for inclusion of transition metals (e.g. Ti(IV) [20,42,43], Zr(IV) [44], Nb(V) [28,45], W(VI) [46], and Mo(VI) [47]) into mesoporous silicates. In this work, we employed this technique to prepare samples of Ti- and Nb-
A
containing silicates with similar metal loadings and textural properties (Table 1). Comprehensive characterization of such materials (SEM, N2 adsorption, XRD, DR UV-vis, FT-IR, Raman, and XPS) has been reported in our previous works [20,28,30]. The state of metal in the two catalysts was similar, as verified by DR UV-vis spectroscopy (compare the corresponding spectra of the initial samples in Figures 1 and 2). The broad absorptions in the range of 225–270 nm (band edges of 318–325 nm) indicate that both Ti(IV) and Nb(V) are present mostly in the
7
form of dimers and/or small oligomers [20,28 and references therein]. Recent FT-IR spectroscopic studies using pyridine and carbon monoxide as probe molecules showed that the amount of Lewis acid sites normalized to the metal content is close for Nb-silicates prepared by EISA and containing either isolated or di(oligo)meric Nb sites [30], which implies a fairly good dispersion and accessibility of the latter. 3.2. Catalysts stability toward H2O and H2O2
IP T
Agglomeration of active metal sites under turnover conditions of liquid-phase oxidation that employs aqueous hydrogen peroxide is a vital problem of mesoporous metal-silicate catalysts [10,11].
SC R
Figures 1 and 2 show changes in the DR UVvis spectra of Nb-Si and Ti-Si catalysts after treatments with boiling water and H2O2 in MeCN solution. One can notice a significant difference: while the spectrum of Nb-Si remained practically intact after the boiling procedure, indicating excellent hydrolytic stability of
U
the Nb active sites, the corresponding spectrum of Ti-Si revealed a significant long-wave shift of the band
N
edge (ca. 29 nm), which is a manifestation of increasing degree of Ti sites aggregation and formation of
A
subnanometric TiO2 clusters [10,48]. However, the band edge in the DR UV spectrum still remained below
M
355 nm, which implies that the dimension of titania clusters was less than 1.2–1.6 nm (30–70 TiO2 units)
ED
[49]. On the other hand, both catalysts revealed a rather similar sensitivity toward a cooperative effect of water and H2O2 (treatment with aqueous H2O2 in MeCN), which is manifested by a small red shift (ca. 7
PT
nm) of the DRS UV-vis absorption edge (see Figures 1 and 2). Therefore, we may conclude that active sites in mesoporous Nb-silicates possess better hydrolytic stability than active sites in Ti-silicates prepared by
CC E
the same technique while the stability toward dilute H2O2 is comparable. 3.3. Kinetics of alkene epoxidation and rate law
A
Cyclooctene was chosen as a model substrate for kinetic studies of alkene epoxidation over Nb-Si
and Ti-Si catalysts because its epoxide is rather stable toward epoxide ring opening and rearrangements. Acetonitrile was employed as solvent since the truly heterogeneous nature of the H2O2-based oxidation catalysis over Nb- [28] and Ti-silicates [20] had been previously demonstrated by hot filtration tests [39]. The rate of CyO oxidation was negligible in the absence of any catalyst. Kinetic curves of the alkene consumption over Nb-Si and Ti-Si catalysts revealed no induction periods, and the epoxidation rates were
8
not affected by the presence of molecular oxygen or light. Conventional radical scavengers, 2,6-di-tertbutyl-4-methylphenol and hydroquinone, produced no effect on the CyO oxidation rate. The reaction rates were independent on the rate of stirring of the reaction mixture in the range of 500–1000 rpm, which implies the absence of external diffusion limitations. Figure 3 shows plots of CyO consumption over Nb-Si [30] and Ti-Si catalysts versus concentrations
IP T
of substrate, oxidant, catalyst, and water. Notwithstanding the active metal (Ti or Nb), the reaction rate exhibited saturation with increasing concentration of olefin (Figure 3a) and oxidant (Figure 3b) and was
SC R
proportional to the catalyst loading (Figure 3c). The reaction over Nb-Si was slightly inhibited by H2O while, in the case of Ti-Si, practically no effect was observed (Figure 3d). The addition of CyO epoxide (ca. 40 mol.% of the amount of alkene) produced no effect on the oxidation rate, indicating that adsorption of
U
the main reaction product on both Ti and Nb sites is negligible.
N
The set of data shown in Figure 3 demonstrate that the kinetic trends for CyO transformation over
A
Nb-Si and Ti-Si catalysts are quite similar. These regularities resemble those reported for alkene
M
epoxidation with H2O2 over TS-1 [9]. Therefore, by analogy with TS-1, we may suppose that an Eley–Rideal
ED
type mechanism operates in alkene epoxidation over mesoporous Nb- and Ti-silicates. This mechanism involves adsorption of water (Eq 1) and Н2О2 (Eq 2) on metal (M = Nb or Ti) sites (the presence of Lewis
PT
acid sites of a moderate strength in the Nb-Si and Ti-Si catalysts has been recently confirmed by FTIR spectroscopy of adsorbed CO and pyridine [30]), reversible interaction of H2O2 with M to give MOOH (Eq
CC E
3), adsorption/desorption of water on/from MOOH site (Eq 4), and electrophilic oxygen transfer from
A
MOOH to С=С bond, resulting in epoxide and regenerating the initial state of the catalyst (Eq 5).
K1 M O H H 2O
M O H H 2O
1
9
K
2
M O H + H 2O 2
2
M O H (H 2 O 2 )
M O H H 2O 2
M O O H H 2O
IP T
k3
3
4
M O O H H 2O
4
ED
M
A
M O O H H 2O
N
K
U
SC R
k 3
k5
M O H e p o x id e
5
A
CC E
PT
M O O H C yO
The rate law was derived using a steady-state approximation (Eq S1, see SI for details). After fitting
the experimental data with this rate law, parameters k-3 and K4 were found to be close to zero for Ti-Si catalyst (Table S1). Previously, the same trend was found for Nb-Si catalyst [30]. Therefore, for both catalysts a simplified rate law (see SI) can be expressed by Eq 6,
10
W0
K 2k 3 H 2O 2 n (M )0
(12),
K 2k 3 H 2O 2 V 1 K 1 H 2O k 5 C yO
(6)
where n(M)0 is the total amount of Nb or Ti sites and V is volume of the reaction mixture.
IP T
Optimal values of all parameters of Eq 6 and their standard deviations are given in Table 2. The value of K1 for Ti-Si catalyst turned out comparable with its standard deviation, which resulted in a low t-
SC R
value (0.83) and, consequently, in a low statistical significance of K1 parameter. So, K1 could also be excluded from the model, giving the final rate law for Ti-catalyst expressed by Eq 7. Optimal values of
K 2k 3 H 2O 2 V 1 k 5 C yO
N
K 2 k 3 H 2O 2 n (T i) 0
(7)
M
A
W0
U
parameters of Eq 7 are provided in Table 3.
ED
Fitting of the experimental points with Eq 6 (Nb-Si) and Eq 7 (Ti-Si), which is in effect a special case of Eq 6, is shown in Figure 3. Both equations describe well the kinetic regularities observed for the Nb-
PT
and Ti-catalyzed CyO epoxidation, which strongly supports the mechanism depicted by stages (1)-(5).
CC E
The following main conclusions could be drawn from the kinetic modeling study. First, adsorption of water on M site is less favorable for Ti(IV) than for Nb(V) (compare parameters K 1 in Table 2). Second, the formation of MOOH from MOH and H2O2 is almost irreversible for both metals (k-3 = 0). Third,
A
adsorption of water on MOOH is negligible (K4 = 0) regardless of M (Nb or Ti). Fourth, the rate of MOOH formation, which can be, in principle, characterized by parameter K2k3, seems higher for Nb-Si catalyst (18.9 vs 12.9 M-1·min-1 for Ti-Si). Finally, fifth, the rate of oxygen atom transfer from MOOH to alkene is also somewhat higher for the Nb-silicate (8.5 vs 6.6 M-1·min-1). 3.5. Alkene epoxidation vs H2O2 decomposition: comparison of activation energies
11
The rate of CyO epoxidation showed a typical Arrhenius dependence (Figure 4), which implies that there was no change in the rate-limiting step over the evaluated temperature range. The estimated values of the apparent activation energy (Ea) equal to 11.9 and 14.2 kcal/mol for Nb-Si and Ti-Si catalysts, respectively, indicating that the reaction is controlled by chemical interaction rather than diffusion (for the latter, Ea is expected to be < 4 kcal/mol [50]).
IP T
Unproductive homolytic decomposition of hydrogen peroxide usually competes with selective oxidation of organic substrate over metal-silicates, especially mesoporous ones [9,10,51]. Recently, we
SC R
found that Nb-Si catalysts are significantly less active than Ti-Si ones in hydrogen peroxide degradation in the absence of any organic substrate [30]. In this work, we first compared apparent activation energies of this reaction over Nb-Si and Ti-Si catalysts. The corresponding Arrhenius plots are shown in Figure 5.
U
Significantly, Ea established for the decomposition of H2O2 over Ti-Si catalyst turned out
N
significantly lower than Ea acquired for Nb-Si: 14.8 vs 20.5 kcal/mol. Even more important is the fact that
A
Ea for CyO epoxidation and H2O2 decomposition are quite similar for the Ti-silicate (14.2 and 14.8 kcal/mol,
M
respectively) while they differ markedly for the Nb-silicate (11.9 and 20.5 kcal/mol, respectively). This
ED
finding allowed us to suggest that, similarly to homogeneous catalysts [52], higher epoxidation selectivity of Nb-silicates can be due to higher energy cost of homolytic O–O bond breaking in NbOOH intermediate.
PT
On the basis of the obtained values of the activation energies, one may predict that epoxidation (heterolytic pathway) selectivity might be improved by decreasing the reaction temperature in the case
CC E
of Nb-Si catalyst, but not in the case of Ti-Si catalyst. To verify this hypothesis, the effect of temperature on oxidation selectivity has been evaluated for both catalysts using cyclohexene as model substrate
A
(section 3.6). CyO and CyH have a similar nucleophilicity of the C=C double bond [53] and normally reveal similar epoxidation rates [28,53]. 3.6. Effect of temperature on epoxidation of cyclohexene In contrast to CyO, cyclohexene possesses highly reactive C–H bonds in the allylic position, which readily react with radical species, leading to allylic oxidation products [54]. Therefore, CyH can be used as a test substrate that enables evaluation of contributions of heterolytic and homolytic oxidation pathways
12
into the overall oxidation process on the basis of quantification of the specific oxidation products (Scheme 1). Figure 6 shows the effect of temperature on selectivity of CyH oxidation achieved at a similar level of substrate conversion over Ti-Si and Nb-Si catalysts. One can notice that, for Nb-Si, the contribution of the heterolytic pathway oxidation products increased drastically (from 70 to 85%) upon decreasing the reaction temperature from 50 to 30 °C. On the contrary, the effect of temperature on the selectivity turned out minor in the case of Ti-Si catalyst, in accordance with what was predicted by the values of the
IP T
apparent activation energies (section 3.5). The data presented in Figure 6 also show that Nb-Si catalyst is more active than Ti-Si catalyst since the reaction time required to reach the same level of conversion is
SC R
significantly lower for Nb.
3.7. Origin of different regioselectivity in limonene oxidation over Nb-Si and Ti-Si catalysts
U
Finally, we would like to discuss possible reasons for different regioselectivities observed in
N
limonene epoxidation over Ti- and Nb-silicates. Guidotti and coworkers were the first who reported on
A
unexpected regioselectivity toward epoxidation of the less electron–rich, exocyclic C=C double bond in
M
limonene and carveol over silica-grafted Nb catalysts [25,26]. On the contrary, another group identified
ED
endocyclic, 1,2-epoxide as the main product formed over Nb-MSU–X [24]. To rationalize this seeming discrepancy, we explored the solvent effect on the epoxidation of limonene over Nb-Si catalysts and found
PT
that the regioselectivity strongly depends on the choice of solvent [28]. While exocyclic epoxide was the main product in MeCN (exo/endo = 73/27, a similar ratio was also found for silica-grafted Nb catalysts
CC E
[25]), endocyclic one predominated in MeOH (exo/endo = 13/87). Note that the 1,2-epoxide was always a predominated isomer (ca. 90%) formed over Ti-silicates [25,26].
A
To understand better the solvent-dependent regioselectivity of Nb-silicates, we performed hot
filtration tests [39] for limonene oxidation over Nb-Si catalyst in both MeCN and MeOH (Figure 7). In line with the previous results [25,28], Nb-silicate behaved as a truly heterogeneous catalyst in acetonitrile (the reaction stopped completely after removal of the catalyst), while the contribution of homogeneous catalysis manifested by continuation of the reaction in the filtrate took place in methanol (compare Figures 7a and 7b). Indeed, the elemental analysis revealed substantial leaching of Nb (32 ppm) in MeOH
13
while it did not exceed 1 ppm in MeCN. A similar trend was documented for mesoporous Ti-Si catalysts, which suffer Ti leaching in alcoholic solvents but are rather stable in acetonitrile [10,11]. Therefore, we may suppose that the inversion of regioselectivity observed in the oxidation of limonene over Nb-silicates can be related, at least partially, to the solvent-dependent stability of Nb-Si catalysts and different nature of the catalysis in MeCN and MeOH. We cannot also exclude that Nb and Ti hydroperoxo species formed on the catalyst surface in the presence of MeOH have a similar structure, thereby leading to similar
IP T
regioselectivity.
SC R
Then the question that arises in relation to the heterogeneously catalyzed oxidation of limonene in MeCN is why the less substituted (nucleophilic) C=C bond is primarily epoxidized over Nb-silicates while Ti-silicates oppositely favor epoxidation of more substituted, electron-rich C=C bond, as one would expect
U
for electrophilic oxygen transfer, typical of alkene epoxidation catalyzed by complexes of d0-metals [53].
N
Recently, we found that Nb-Si catalysts are able to perform epoxidation of electron-deficient C=C bonds
A
in α,β-unsaturated carbonyl compounds through the Payne type oxidation that involves rate-limiting
M
oxidation of MeCN to peroxycarboximidic acid (the real epoxidizing species) [38] and this ability is due to the presence of weak basic sites on the surface of Nb-silicates [30]. This oxidation route leads to the
ED
formation of acetamide in the amount close to that of epoxide. However, no acetamide was detected in the oxidation of limonene over Nb-Si catalysts, which allowed us to discard the role of Payne oxidation in
PT
the unusual regioselectivity of limonene epoxidation.
CC E
The kinetic modeling study undertaken in this work showed that the general mechanism of alkene epoxidation of Nb- and Ti-silicates is similar and can be envisaged by the sequence of similar reaction steps (1)–(5). Nowadays, it is widely accepted that hydroperoxo titanium species (TiOOH) is responsible
A
for alkene epoxidation over Ti-Si catalysts [see 9,10,51,55]. However, the coordination mode of the hydroperoxo group (η1 vs η2) in TiOOH remains under debates and the exact structure of the titanium hydroperoxo species is still unknown despite numerous spectroscopic and computational studies. Recently, we demonstrated that basic additives lead to deactivation of Nb-silicates, likewise Ti-silicates [56], producing a strong rate-retarding effect on alkene epoxidation with H2O2 [30]. This finding strongly supports a niobium hydroperoxo (i.e. protonated peroxo) species, NbOOH, as the active species operating
14
in the Nb-catalyzed epoxidation of olefins. This species is manifested by a DR UV-vis absorption band at 307 nm, which shifts to 293 nm upon addition of base [30]. The different regioselectivity of limonene epoxidation in MeCN has led us to a suggestion that the structure of niobium and titanium hydroperoxo species MOOH that participate in the oxygen transfer step (Eq 5) is, most likely, different. Guidotti and coworkers first suggested that the observed peculiarity of regioselectivity in limonene epoxidation over mesoporous Nb-silicates is to be attributed mainly to steric factors [26]. Indeed, the unusual
IP T
regioselectivity toward exocyclic limonene epoxide was no longer observed when a nonporous Nb-SiO2 catalyst, prepared by grafting of Nb(V) centers onto a pyrogenic nanosized silica (Aerosil 380), was
SC R
employed.
We may tentatively suppose that NbOOH has a η2 coordination of the hydroperoxo group while a
U
η1 coordination mode is realized in TiOOH. Alternatively, different mechanisms of the oxygen transfer step
N
(Eq 5), i.e. oxygen transfer from α or β positions of the hydroperoxo group, may account for the opposed
A
regioselectivity. Recently, unprecedented β-oxygen transfer mechanism has been established by the
M
computational technique for alkene epoxidation catalyzed by Ti-containing polyoxometalates [57].
ED
4. Conclusions
The present study demonstrated that epoxidation of electron-rich alkenes catalyzed by
PT
mesoporous Nb- and Ti-silicates involves similar stages, typical of Eley–Rideal mechanism. The higher heterolytic pathway selectivity of Nb-silicates is, at least partially, due to a higher energy cost of homolytic
CC E
O–O bond breaking in NbOOH intermediate relative to TiOOH one, which is manifested by the smaller rates and activation energy of H2O2 decomposition over Nb-silicates. A significant gap between the
A
activation energies established for alkene (cyclooctene) epoxidation and H2O2 degradation over Nbcatalysts (11.9 and 20.5 kcal/mol, respectively) makes possible substantial improvement of heterolytic pathway selectivity by decreasing the reaction temperature. On the contrary, this approach does not work with mesoporous Ti-silicates, for which the activation energy values of cyclooctene epoxidation and H2O2 decomposition are similar (14.2 and 14.8 kcal/mol, respectively). The difference in regioselectivity of limonene epoxidation over Nb- and Ti-silicates established in MeCN, where the nature of the catalysis is
15
truly heterogeneous, indicates different structures of the active ‘NbOOH’ and ‘TiOOH’ species and/or different mechanisms of the oxygen transfer step. On the other hand, the inversion of regioselectivity observed for Nb catalyst in MeOH may be caused by Nb leaching in this solvent and/or changes in the structure of Nb hydroperoxo species, making it more close to that of TiOOH, which explains similar regioselectivities of Nb and Ti catalysts documented for alcoholic solvents. We hope that further comparative spectroscopic and theoretical studies may help in understanding the structure of active Nb
IP T
and Ti hydroperoxo species and their transformations. Such investigations using Nb- and Ti-substituted polyoxometalates as tractable soluble model compounds are in progress in our group and will be
SC R
published elsewhere.
Acknowledgements. The help of Dr. M. V. Shashkov and Dr. S. A. Yashnik in GC–MS and DRS UV-vis
U
measurements, respectively, is appreciated. This work was carried out in the framework of budget project
N
No. 0303-2016-0005 for the Boreskov Institute of Catalysis and partially supported by the Russian
A
Foundation for Basic Research (grant N 16-03-00827).
M
References
ED
References
[1] Modern Heterogeneous Oxidation Catalysis: Design, Reactions and Characterization, N. Mizuno (Ed.), Wiley-VCH, Weinheim, 2009.
PT
[2 ] Liquid Phase Oxidation via Heterogeneous Catalysis: Organic Synthesis and Industrial Applications, M.G. Clerici, O.A. Kholdeeva (Eds.), Wiley, Hoboken, 2013.
CC E
[3] Handbook of Advanced Methods and Processes in Oxidation Catalysis, D. Duprez , F. Cavani (Eds.), Imperial College Press, London, 2014. [4] F. Cavani, J. H. Teles, Sustainability in Catalytic Oxidation: An Alternative Approach or a Structural Evolution? ChemSusChem 2 (2009) 508–534.
A
[5] C.W. Jones, Application of Hydrogen Peroxide and Derivatives, Royal Society of Chemistry, Cambridge, 1999. [6] G. Strukul, A. Scarso, in: M.G. Clerici, O.A. Kholdeeva (Eds.), Liquid Phase Oxidation via Heterogeneous Catalysis: Organic Synthesis and Industrial Applications; Wiley, Hoboken, 2013; Ch. 1., pp. 1–20. [7] J.M. Campos-Martin, G. Blanco-Brieva, J.L.G. Fierro, Angew. Chem. Int. Ed. 45(2006) 6962– 6984.
16
[8] M. Taramasso, G. Perego, B. Notari, Snamprogetti S.p.A. assignee. US Patent 4,410,501, 1983. [9] For recent review see M.G. Clerici, M.E. Domine, in: M.G. Clerici, O.A. Kholdeeva (Eds.), Liquid Phase Oxidation via Heterogeneous Catalysis: Organic Synthesis and Industrial Applications, Wiley, Hoboken, 2013, Ch. 2, p. 21–93.
[10] O.A. Kholdeeva, in: M.G. Clerici, O.A. Kholdeeva (Eds.), Liquid Phase Oxidation via Heterogeneous Catalysis: Organic Synthesis and Industrial Applications, Wiley, Hoboken, 2013; Ch. 4, pp. 127–219.
[12] P.T.Tanev, M. Chibwe, T. J. Pinnavaia, Nature 368 (1994) 321−323. [13] A. Sorokin, A. Tuel, Catal. Today, 57 (2000) 45−59.
IP T
[11] O.A. Kholdeeva, Catal. Sci. Technol. 4 (2014) 1869–1889.
SC R
[14] N.N.Trukhan, V.N. Romannikov, E.A. Paukshtis, A.N. Shmakov, O.A. Kholdeeva, J. Catal. 202 (2001) 110−117. [15] O.A. Kholdeeva, O.V. Zalomaeva, Coord. Chem. Rev. 306 (2016) 302−330.
N
U
[16] O.A. Kholdeeva, in: B. Cornils, W.A. Hermann, M. Beller, R. Paciello (Eds.), Applied Homogeneous Catalysis with Organometallic Compounds, 3rd ed., Wiley-VCH, 2017, pp. 545– 570.
M
A
[17] O.A. Kholdeeva, M.S. Melgunov, A.N. Shmakov, N.N. Trukhan, V.V. Kriventsov, V.I. Zaikovskii, M.E. Malyshev, V.N. Romannikov, Catal. Today, 91-92 (2004) 205–209. [18] P. Wu, T. Tatsumi, T. Komatsu, T. Yashima, Chem. Mater. 14 (2002) 1657–1664.
ED
[19] F.-S. Xiao, Top. Catal. 35 (2005) 9–24.
[20] I.D. Ivanchikova, M.K. Kovalev, M.S. Mel’gunov, A.N. Shmakov, O.A. Kholdeeva, Catal. Sci. Technol. 4 (2014) 200−207.
PT
[21] I. Nowak, B. Kilos, M. Ziolek, A. Lewandowska, Catal. Today 78 (2003) 487–498. [22] I. Nowak, M. Ziolek, Micropor. Mesopor. Mater. 78 (2005) 281–288.
CC E
[23] F. Somma, G. Strukul, Catal. Lett. 107 (2006) 73–81. [24] A. Feliczak-Guzik, I. Nowak, Catal. Today 142 (2009) 288–292.
[25] A. Gallo, C. Tiozzo, R. Psaro, F. Carniato, M. Guidotti, J. Catal. 298 (2013) 77–83.
A
[26] C. Tiozzo, C. Bisio, F. Carniato, M. Guidotti, Catal. Today 235 (2014) 49–57. [27] A. Ramanathan, H. Zhu, R. Maheswari, P.S. Thapa, B. Subramaniam, Ind. Eng. Chem. Res. 54 (2015) 4236–4242.
[28] I.D. Ivanchikova, N.V. Maksimchuk, I.Y. Skobelev, V.V. Kaichev, O.A. Kholdeeva, J. Catal. 332 (2015) 138–148. [29] N.E. Thornburg, A.B. Thompson, J.M. Notestein, ACS Catal. 5 (2015) 5077–5088.
17
[30] I.D. Ivanchikova, I. Y. Skobelev, N. V. Maksimchuk, E. A. Paukshtis, M. V. Shashkov, O. A. Kholdeeva, J. Catal. 356C (2017) 85–99. [31] N.E. Thornburg, J.M. Notestein, ChemCatChem 9 (2017) 3714–3724. [32] S. Dworakowska, C. Tiozzo, M. Niemczyk-Wrzeszcz, P. Michorczyk, N. Ravasio, R. Psaro, D. Bogdał, M. Guidotti, J. Clean. Prod. 166 (2017) 901–909. [33] M. P. Kapoor, A. Raj, Stud. Surf. Sci. Catal. 129 (2000) 327–334.
IP T
[34] M. Ziolek, I. Sobczak, P. Decyk, K. Sobanska, P. Pietrzyk, Z. Sojka, Appl. Catal. B: Environ. 164 (2015) 288–296. [35] N.E. Thornburg, S.L. Nauert, A.B. Thompson, J.M. Notestein, ACS Catal. 6 (2016) 6124– 6134.
SC R
[36] D.T. Bregante, P. Priyadarshini, D.W. Flaherty, J. Catal. 348 (2017) 75–89. [37] D.T. Bregante, D.W. Flaherty, J. Am. Chem. Soc. 139 (2017) 6888−6898. [38] G.B. Payne, P.H. Deming, P.H. Williams, J. Org. Chem. 26 (1961) 659–663.
U
[39] R.A. Sheldon, M. Wallau, I.W.C.E. Arends, U. Schuchardt, Acc. Chem. Res. 31 (1998) 485– 493
A
N
[40] I.Y. Skobelev, O.V. Zalomaeva, O.A. Kholdeeva, J.M. Poblet, J.J. Carbó, Chem. Eur. J. 21 (2015) 14496–14506.
M
[41] I.Y. Skobelev, V.Yu. Evtushok, O.A. Kholdeeva, N.V. Maksimchuk, R.I. Maksimovskaya, J. M. Ricart, J.M. Poblet, J.J. Carbó, ACS Catal. 7 (2017) 8514–8523.
ED
[42] M. Ogawa, K. Ikeue, M. Anpo, Chem. Mater. 13 (2001) 2900–2904. [43] N. Hüsing, B. Launay, D. Doshi, G. Kickelbick, Chem. Mater. 14 (2002) 2429–2432.
PT
[44] H. Zhu, R. Maheswari, A. Ramanathan, B. Subramaniam, Micropor. Mesopor. Mater. 223 (2016) 46– 52.
CC E
[45] A. Ramanathan, H. Zhu, R. Maheswari, B. Subramaniam, Micropor. Mesopor. Mater. 261 (2018) 158–163. [46] J.-F. Wu, A. Ramanathan, B. Subramaniam, J. Catal. 350 (2017) 182–188. [47] A. Ramanathan, J.-F. Wu, R. Maheswari, Y. Hu, B. Subramaniam, Micropor. Mesopor. Mater. 245 (2017) 118–125.
A
[48] X. Gao, I.E. Wachs, Catal. Today 51 (1999) 233–254. [49] B.J. Aronson, C.F. Blanford, A. Stein. Chem. Mater. 9 (1997) 2842–2851. [50] P. Atkins, J. De Paula, Elements of Physical Chemistry, sixth edition, Oxford University Press, 2013. [51] C.W. Yoon, K.F. Hirsekorn, M.L. Neidig, X. Yang, T.D. Tilley, ACS Catal. 1 (2011) 1665–1678. [52] N. Antonova, J. J. Carbo´, U. Kortz, O.A. Kholdeeva, J.M. Poblet, J. Am. Chem. Soc. 132 (2010) 7488–7497. 18
[53] B.G. Donoeva, T.A. Trubitsina, N.S. Antonova, J.J. Carbó, J.M. Poblet, G. Al Kadamany, U. Kortz, O.A. Kholdeeva, Eur. J. Inorg. Chem. (2010) 5312–5317. [54] R.A. Sheldon, J.K. Kochi, Metal-Catalyzed Oxidations of Organic Compounds, New York, Academic Press, 1981. [55] S. Bordiga, E. Groppo, G. Agostini, J.A. van Bokhoven, C. Lamberti, Chem. Rev. 113 (2013) 1736–1850. [56] M.G. Clerici, Appl. Catal. 68 (1991) 249–261.
SC R
IP T
[57] P. Jiménez-Lozano, I.Y. Skobelev, O.A. Kholdeeva, J.M. Poblet, J.J. Carbó, Inorg. Chem. 55 (2016) 6080–6084.Figure Caption
Nb-Si Nb-Si + H2O2
U
4
Nb-Si + H2O
N A
2
M
F(R)
3
ED
1
0
250
300
350
400
, nm
PT
200
Figure 1. DR UV–vis spectra of the initial Nb-Si catalyst and after treatments with aqueous H2O2 in MeCN
A
CC E
(25 oC, 1 h) and with boiled water (6 h).
19
F(R)
8
Ti-Si Ti-Si + H2O2
6
Ti-Si + H2O
4
0 200
250
300
350
SC R
, nm
400
IP T
2
Figure 2. DR UV–vis spectra of the initial Ti-Si catalyst and after treatments with aqueous H2O2 in MeCN
A
CC E
PT
ED
M
A
N
U
(25 oC, 1 h) and with boiled water (6 h).
20
2,0
3
W0x10 , M/min
0,5
0,0
0,3
2,0
5,0
1,8 W0x10 , M/min
6,0
4,0
3
3,0 2,0
3 4 5 6 7 8 3 n(TM)x10 , mmol
1,0
9 10
N
2
1,2
0,2 0,3 [H2O2], M
0,4
0,5
(d) 1,0 1,5 [H2O], M
2,0
A
1
1,4
0,1
0,8
(c) 0
1,6
U
1,0 0,0
(b)
IP T
0,2 [CyO], M
3
W0x10 , M/min
0,1
1,0
0,0
(a) 0,0
1,5
SC R
3
W0x10 , M/min
2,5
3,2 2,8 2,4 2,0 1,6 1,2 0,8 0,4 0,0
M
Figure 3. Plots of the initial rate of CyO epoxidation (MeCN, 50 oC) versus concentration of substrate (a), oxidant (b), catalyst (c), and water (d). Experimental points: (■) and (●) for Ti-Si
ED
and Nb-Si catalysts, respectively. Fitting with: (-) Eq 7 for Ti-Si and (-) Eq 6 for Nb-Si. For
A
CC E
PT
reaction conditions, see Experimental, section 2.5.
21
Ea = 14.2 ± 0.8 kcal/mol
ln W0
-5 -6 -7
Ea = 11.9 ± 0.6 kcal/mol -9 0,0028
0,0030
0,0032
SC R
1/T, K
0,0034
-1
IP T
-8
Figure 4. Arrhenius plots for CyO epoxidation over Ti-Si (■) and Nb-Si (●) catalysts. Reaction conditions:
U
CyO 0.1 M, H2O2 0.1 M, catalyst 0.003 mmol Nb or Ti, MeCN 1 mL.
Ea = 14.8 ± 1.1 kcal/mol
N
-4
A
-5
-7 -8
M
ln W0
-6
ED
-9
-10
Ea = 20.5 ± 0.3 kcal/mol 0,0029
0,0030 1/T, K
0,0031
0,0032
-1
CC E
PT
0,0028
Figure 5. Arrhenius plots for H2O2 decomposition over Ti-Si (■) and Nb-Si (●) catalysts. Reaction conditions:
A
Н2О2 0.4 М, Nb 0.02 mmol, MeCN 7 mL.
22
CyH conversion Time
Selectivity: heterolytic pathway homolytic pathway 3
60 2 40
Time, h
1 20
0
IP T
Conversion/ selectivity, %
80
0
Nb
o
Ti
Nb
o
Ti
50 C
SC R
30 C
Figure 6. Effect of temperature on selectivity of CyH oxidation over Nb-Si and Ti-Si catalysts. Reaction
A
CC E
PT
ED
M
A
N
U
conditions: CyH 0.1 mmol, H2O2 0.1 mmol, 0.003 mmol Nb or Ti, MeCN 1 mL.
23
100
Hot filtration 90
80
70 0
20
40
60
80
100
120
140
IP T
Limonene consumption , %
(a)
SC R
Time, min
U
Hot filtration
N
90
80
A
Limonene consumption , %
(b)
100
50
M
0
100
150
Time, min
ED
Figure 7. Hot catalyst filtration tests for limonene oxidation with H2O2 over Nb-Si catalyst in (a) MeCN and (b) MeOH. Reaction conditions: limonene 0.1 mmol, H2O2 0.1 mmol, catalyst 0.003 mmol Nb,
A
CC E
PT
solvent 1 mL, 50 oC.
24
IP T
heterolytic pathway
homolytic pathway
A
CC E
PT
ED
M
A
N
U
SC R
Scheme 1. Products typical of CyH oxidation via heterolytic and homolytic pathways.
25
Table 1. Metal loading and texture of Nb- and Ti-silicates prepared by EISA.
M,a
SBET,
Vp,b
Dp,c
wt.%
m2/g
cm3/g
nm
Nb-Si
1.96
1190
0.56
2.8
Ti-Si
2.35
1270
0.54
2.7
Catalyst
Based on elemental analysis data for calcined samples.
b
Mesopore volume.
c
Average pore diameter.
SC R
IP T
a
N
U
Table 2. Optimal values of parameters of Eq 6 for catalysts Nb-Si and Ti-Si. Value (standard deviation) for
A
Parameter
M
catalyst Nb-Si K1, M-1 K2k3, (M·min)-1
catalyst Ti-Si 0.15 (0.18)
18.9 (5.5)
11.0 (2.3)
8.5 (1.1)
7.9 (1.3)
CC E
PT
k5, (M·min)-1
ED
0.8 (0.4)
Value (standard deviation) for
A
Table 3. Optimal values of parameters of Eq 7. Parameter
Value (standard deviation)
K2k3, (M·min)-1
12.9 (2.4)
k5, (M·min)-1
6.6 (0.9)
26
27
A ED
PT
CC E
IP T
SC R
U
N
A
M