Mesoporous TUD-1 supported indium oxide nanoparticles for epoxidation of styrene using molecular O₂

Abstract

Activation of molecular O₂ by metal or metal oxide nanoparticles is an area of recent research interest. In this work, for the first time, we report that indium oxide nanoparticles of <3 nm size dispersed on mesoporous silica (TUD-1) can activate molecular O₂ and produce styrene epoxide with a selectivity of 60% and styrene conversion around 25% under mild conditions. It is found that neither indium oxide nor TUD-1 themselves respond to the styrene epoxidation reaction. The computational studies provide evidence that an oxygen molecule is highly polarized when it is located near the interface of both surfaces. The kinetic study shows that the reaction is of pseudo-first order and that the activation energy for styrene conversion is 12.138 kJ mol⁻¹. The catalysts are recyclable for up to four regeneration steps, with the styrene conversion and styrene epoxide selectivity almost unchanged.

---

1. Introduction

Investigations into metal oxide nanoparticles¹ are motivated by their wide range of properties useful for various applications, including optics, electronics, magnetism and catalysts.² Indium oxide is an n-type semiconductor with a band gap (E_g = 3.70 eV) at room temperature. Research on indium oxide (In₂O₃) nanoparticles has attracted much attention in recent years^3,4 because of their applications in solar cells, field emission display, lithium ion batteries, nanoscale bio-sensors, gas sensors, optoelectronics and photocatalysis.^5–8 Indium in its most stable oxidation state of (+III) is a typically strong Lewis acid, possessing a vacant low energy orbital. As reported in the literature, indium (+III) is among the most commonly employed Lewis acids in organic synthesis, including that of asymmetric catalysts.⁹

Epoxides are useful and versatile intermediates for synthesis of many commodities and fine chemicals. Studies focusing on epoxidation of the C=C bond have received much attention recently.^10–12 The traditional procedure for epoxidation of alkenes using stoichiometric amounts of per-acids is not an acceptable methodology from an environmental perspective.¹³ Direct epoxidation using molecular O₂ provides an atom economic pathway to obtain epoxides under mild conditions.¹⁴ Successful production of ethylene epoxide over silver catalyst using molecular O₂ as an oxidant has resulted in many catalysts being developed in the past two decades for epoxidation of higher homologous, e.g. 1,3-butadiene, propylene etc. through the same route.^15–18

Styrene oxide is one of the most important fine chemical intermediates for producing perfume, drugs, sweeteners, epoxy resins, etc. Current commercial production mostly employs the bromohydrin method, which has serious problems in terms of equipment corrosion and environmental pollution.¹⁹ Heterogeneous catalysts, including Ti/SiO₂,²⁰ titanosilicate,²¹ mixed metal oxide,²² hydroxyapatites,^14,23 hydrotalcites²⁴ and cobalt doped mesoporous silica,²⁵ have been reported for use in this reaction. However, research continues into new catalyst systems, as, thus far, commercial production of styrene epoxidation has not been developed using heterogeneous catalysts.

Ordered mesoporous materials with their intrinsically high surface areas are excellently suited to being catalytic materials in various reactions. Compared with conventional carrier materials, ordered mesoporous solids have the advantage of stabilizing dispersed metal or metal oxide nanoparticles.²⁶ TUD-1 is a mesoporous silica matrix with a large surface area and high thermal stability. Metal ions can be incorporated into the silica matrix, making TUD-1 a potential catalyst candidate for several oxidation reactions including styrene epoxidation reactions.²⁷

In this paper, for the first time, we report that indium oxide nanoparticles 2–3 nm size dispersed over TUD-1 material can activate molecular O₂ thereby producing styrene epoxide from styrene under mild conditions. The two different indium doped TUD-1 (In/Si = 1/100 and 4/100) catalysts were prepared and characterized by N₂ adsorption–desorption isotherms, small angle X-ray scattering (SAXS), high resolution transmission electron microscopy (HRTEM), elemental mapping (STEM-EDS), Fourier transform infrared (FTIR) spectroscopy, X-ray photoelectron spectroscopy (XPS), and temperature programmed reduction (TPR) and temperature programmed oxidation (TPO) techniques. The catalytic activity was optimized by varying flow of oxygen and temperature time of the reaction. Kinetic studies were carried out to obtain the order, rate constant and activation energies of the reaction. Computational studies were carried out to support the experimental results for activation of molecular O₂. An interesting correlation is obtained between catalytic activity and catalyst characterization results as well as theoretical studies, as discussed later.

2. Materials and methods

2.1 Preparation of In-TUD-1 catalyst

In-TUD-1 catalyst was prepared by a non-hydrothermal sol–gel procedure as per the literature.^27,28 Tetraethylorthosilicate (TEOS, 98%, Acros Organics) was mixed with an aqueous solution of indium nitrate (99.9%, Aldrich) in deionized water under continuous stirring followed by dropwise addition of triethanolamine (TEA, 99%, Acros Organics). The mixture was stirred for 10 minutes, then tetraethylammonium hydroxide (TEAOH, 20% aqueous solution, Merck, Germany) was added to keep the final gel composition of the mixture in the molar ratio of TEOS : In(NO₃)₃ : TEA : H₂O : TEAOH = 1 : x : 2 : 11 : 1. (x = 0.01, 0.04). This mixture was stirred for 24 h at room temperature, then the prepared gel was dried at 110 °C for 24 h in a static air oven. Finally the dried material was taken and calcined at 700 °C for 10 h with a temperature increasing rate of 1 °C min⁻¹ in a muffle furnace (Thermcraft incorporated USA).

2.2 Catalyst characterization

The nitrogen adsorption–desorption isotherms of the prepared catalysts were measured at liquid nitrogen temperature of −196 °C with a Quantachrome NOVA 3200 (USA). The samples were pre-treated at 200 °C for 3 h under high vacuum. The surface area was calculated using the Brunauer–Emmett–Teller (BET) equation, and pore size distribution was calculated by the BJH method. Small angle X-ray scattering (SAXS) measurements were performed using a laboratory-based SAXS instrument with CuK_α X-ray radiation. The scattered intensities I(q) were recorded as a function of scattering vector q (= 4π sin θ/λ, where 2θ is the scattering angle, and λ is the X-ray wavelength). The scattered intensity did not depend on the internal structure of the domains, whether they are crystalline or amorphous.

To obtain the particle size distribution, SAXS data were analyzed in the light of a polydispersed ensemble of spherical particles. For such a case, I(q) is expressed as

where P(q,R) represents the form factor of a spherical particle of radius R, i.e.,
D(R) represents the particle size distribution, i.e., D(R)dR indicates the probability of having a particle with radius R to R + dR. In the present case, a standard log normal distribution was assumed where C is a scale factor and does not depend on q.

Powder X-ray diffraction studies were performed on a Rigaku Ultima 4 X-ray diffractometer (JAPAN) with CuK_α radiation as the X-ray source, and data were analyzed using X'pert Highscore Plus software. HRTEM images of the prepared catalysts were obtained using a JEOL JEM 2100 microscope (USA) operated at 200 KV acceleration voltage, with a lacey carbon-coated Cu grid of 300 mesh size. Image J software was used to process the TEM images in the figures. A random selection of 30 particles was used for size statics of the sample. The STEM was done using an electron microscope model x-sight (Oxford instruments). The EDX measurements were made using a Supra 55 (Zeiss, Germany) microscope equipped with Oxford instrument X-max detector and Gemini beam line attachment. The FT-IR measurements were carried out on a Perkin-Elmer spectrum two spectrophotometer (USA). The spectra were recorded in the wavelength range of 400–4000 cm⁻¹ with 4 cm⁻¹ energy resolution, and samples were prepared using KBr pellets. The XPS was carried out using a custom-built ambient pressure X-ray photoelectron spectrometer from PREVAC Inc (Poland). The spectra were collected at 50 eV pass energy using a monochromatic Al K_α X-ray source. All the binding energies of core electron were corrected using C (1s) core electron binding energy at 284.6 eV as reference. The quantification was done using CasaXPS software, taking into consideration atomic sensitivity factors of individual elements. The peaks were fitted using Shirley background correction. The TPR profiles of the catalysts were obtained using a Chemisorb 2720 (Micrometrics, USA) instrument equipped with a TCD detector. The H₂-TPR profiles were obtained by reducing the catalyst samples with 10% H₂ in Ar at a flow rate of 20 mL min⁻¹, with the temperature increasing from ambient to 800 °C at a rate of 10 °C min⁻¹. The TPO profiles of the samples were obtained using the same equipment by oxidizing the catalyst samples with a gas mixture of 4.2% O₂ in He with a flow rate of 20 mL min⁻¹, and the temperature was increased from ambient to 800 °C at a rate of 10 °C min⁻¹. First, ∼0.100 g of catalyst sample was dried with He flow in the preparation port of the instrument and the dry mass was measured. Thereafter the experiment was carried out in the reaction port with flow of the required gas mixture. To quantify the O₂ consumption, peak areas of the respective samples were compared with a calibration file using Chemisoft (Micrometrics, USA) software.

2.3 Catalyst activity studies

The In-TUD-1 catalysts were dried at room temperature in vacuum desiccators, and used for the catalytic studies with no further activation. The catalytic reaction was studied in a batch reactor in which a 50 mL two-necked, round-bottom flask was fitted with a water condenser. The reaction mixture was prepared with 10 mL DMF (99.8%, Merck, India), 6.5 mmol styrene (99%, Acros Organics) and an internal standard dodecane (0.1 mL, 99%, Acros Organics). The reaction mixture was shaken vigorously for homogenization, then 0.1 g catalyst was added. The prepared reaction mixture along with catalyst was kept in a heated silicon oil bath, which was maintained at constant temperature (130 °C). Thereafter oxygen (99.99% purity) was bubbled into the mixture at a flow rate of 10 mL min⁻¹ (flow was maintained by an Aalborg mass flow controller) under continuous stirring by a magnetic stirrer at 700 rpm rotational speed. The reaction mixture was analyzed by GC-1000 (Chemito-India) equipped with an SE-30 column and FID detector. Details of the GC conditions are provided in the ESI S1.† The TOF per hour was calculated on the basis of moles of styrene converted/mol of indium present in the catalyst as determined by EDX, and also based on oxygen uptake obtained from TPO experiments.

2.4 Kinetic analysis

COMSOL multiphysics software (India), chemical species transport and reaction engineering module were used for simulation to validate the kinetics of experimental results.

2.5 Computational details

Molecular simulation studies were performed using a Forcite module in Materials Studio 7.0 developed by Accelrys Software Inc.^29,30 The structures were optimized by energy minimization with Monte Carlo simulations using the Forcite module and the Condensed-Phase Optimized Molecular Potentials for Atomistic Simulation Studies (COMPASS) force field was used in this work as implemented in Materials Studio, as has been done previously for similar systems.^29,31 The Forcite module is an advanced classical molecular mechanics tool that allows fast energy calculations and reliable geometry optimization of molecules and periodic systems.³²

In the COMPASS force field, the total potential energy U is given by

U = U_b + U_Θ + U_Φ + U_χ + U_bb′ + U_bΘ + U_bΦ + U_ΘΦ + U_bΘΦ + U_coul + U_vdw

The terms can be divided into two categories: valence terms, which include bond-stretching (U_b), bond-angle-bending (U_Θ), torsion (U_Φ), out-of-plane angle bending (U_χ), cross-coupling terms (bond–bond (U_bb′), bond-angle (U_bΘ), bond-torsion (U_bΦ), and bond-angle–torsion (U_bΘΦ) interactions), and non-bonded interaction terms, which include the Coulombic function (U_coul) for electrostatic interactions and the Lennard-Jones 9-6 function (U_vdw) for van der Waals interactions.

To get accurate results, the crystal structures and atomic coordinates were optimized by minimizing the energy and atomic forces. Amorphous silicon oxide (SiO₂) was used as a base to impregnate indium oxide nanoparticles (cubic phase, 1 and 2 nm size). The Forcite module was used for optimization of the structural geometry. The COMPASS force field was used for the MD simulations. Interatomic potential force fields were automatically assigned for individual O and Si atoms as implemented in Material Studio, whereas for In, an “oxygen in metal oxide” force field was assigned. The COMPASS force field has been parameterized for molecules in isolation, in condensed phases, in metals and in metal oxides, to predict various properties such as dipole moment. To start the simulation, a silica unit cell was imported from the library and the structure was optimized and relaxed to minimize its energy using the Smart Minimizer method. The lower layers of silica were constrained. Indium oxide nanoparticles (Forcite, geometry optimized and fixed) were then stabilized on the silica surface along with oxygen molecules. The silica surface was annealed from 300 K to 973 K. Oxygen molecules were brought into close contact with different surfaces, for example silica, indium oxide, and the interface between silica and indium oxide. Surface composition was H₅₀₀O₁₇₂₈Si₈₆₄ for a 2 nm thickness silica slab. For a 1 nm In₂O₃ nanoparticle, surface composition was In₁₄O₂₄.

After performing Forcite geometry optimization, the charge induced on individual atoms was calculated using a charge equilibration method (QEq) in Forcite. Conditions for simulation were: force: 0.5 kcal mol⁻¹ Å⁻¹; temperature: 383 K; maximum number of iterations: 500; energy parameters – force field: COMPASS (Version 2.8); charges: force field assigned; electrostatic terms and van der Waals terms – summation method: atom-based; truncation method: cubic spline; cutoff distance: 12.5 Å; spline width: 1 Å; buffer width: 0.5 Å.

For comparison, the simulation was also carried out a using Universal Force Field (UFF), another more simplistic approach, available in Material Studio. In the UFF force field, the total potential energy U is given by

U = U_b + U_Θ + U_Φ + U_vdw

where U_b represents the bond-stretching interactions, U_Θ represents the angle-bending interactions, U_Φ represents the torsional interactions, and U_vdw represents the non-bonded interactions.

Amorphous silica surface was taken from a standard Material Studio library, SiO₂ surface was a 16 Å thick layer with composition H₁₄₀Si₄₃₆O₈₆₈ and a surface composition of H₆₉Si₅₁O_104. Silica surface and indium oxide nanoparticles (1 nm and 2 nm) were annealed from 300 to 973 K using a UFF force field. Cutoff voltage and inter atomic potentials were force field assigned as implemented in Material Studio software. For nonbonding interaction, the electrostatic interaction was based on the Ewald method and van der Waals interaction was atom-based during the calculation. Indium oxide nanoparticles (1 nm and 2 nm) were brought near SiO₂ (amorphous) surface and further optimized using a UFF method. To calculate charge on individual atoms of oxygen molecule, oxygen was brought into the vicinity of the finalized structure (indium doped on silica surface) at different locations. The charge was calculated using a charge equilibration (QEq) method. Before performing this method, the structure was neutralized.

3. Results and discussion

3.1 N₂ adsorption–desorption isotherm

The textural characteristics of TUD-1 and In-TUD-1 catalysts with different indium loading are listed in Table 1. The materials exhibited a high BET surface area, a large pore volume and a porosity exclusively of mesopores as illustrated by the nitrogen sorption isotherms (Fig. 1). From Table 1, it can be seen that with the increase in indium loading there was a decrease in surface area and pore volume, which might be caused by indium oxide nanoparticles blocking mesopores as has been reported in the literature.³³

Table 1 Surface area and pore diameter of different In-TUD-1 catalysts

Catalyst	Surface area (m^2 g^−1)	Total pore volume (cm^3 g^−1)	Pore diameter (nm)
TUD-1	452.3	0.859	7.60
In-TUD-1 (In/Si = 1/100)	688.7	1.094	8.30
In-TUD-1 (In/Si = 4/100)	663.4	0.834	7.22

---

Fig. 1 Surface area and porosity measurement of (a) TUD-1, (b) In-TUD-1 (In/Si = 1/100) and (c) In-TUD-1 (In/Si = 4/100) catalysts.

3.2 Small angle X-ray scattering (SAXS)

For a better insight into the particle size distribution of the In-TUD-1 catalysts, SAXS analysis was carried out.^34,35 The SAXS profiles of the In-TUD-1 catalysts are shown in Fig. 2. A trend towards increasing intensity at low q regime for In-TUD-1(In/Si = 1/100) and In-TUD-1 (In/Si = 4/100) catalysts was observed. Such increase in scattering at low q regime originates from density fluctuations at the larger length scale. This might be because of either the agglomeration of the primary particles or the presence of the pores. Thus, another size distribution and a power law contribution were also considered while modeling. For a sample with (In/Si = 1/00), only a power law-type increase was observed and thus estimation of larger size distribution was not possible. A plot of size distribution vs. radius of the primary particles is shown in Fig. 2. The specific surface area of TUD-1 increased with indium doping; however, at higher indium loading, specific surface area decreased (Table 2), which is in agreement with N₂-physisorption studies.

Fig. 2 Fitting of the model to the experimental SAXS data. Solid line represents the fit in each case.

Table 2 Quantitative analysis of SAXS results for In-TUD-1 catalysts

Catalyst	Average diameter (nm)	Σ (m^−1)	Calculated theoretical density of the samples (with dsilica = 2 g cm^−3 and dIn = 7.3 g cm^−3)	Σ (m^2 g^−1) [assuming the calculated theoretical density]
TUD-1	2.8	2.25 × 10^9	2.00	1090
In-TUD-1 (In/Si = 1/100)	2.2	3.13 × 10^9	2.05	1525
In-TUD-1 (In/Si = 4/100)	1.9	3.30 × 10^9	2.20	1500

---

3.3 X-ray diffraction (XRD)

The XRD pattern of In-TUD-1 (In/Si = 1/100) showed the absence of diffraction peaks for crystalline indium oxide (ESI, Fig. S2†); however, after higher loading, that is In-TUD-1 (In/Si = 4/100), typical diffraction peaks appeared at (222), (400), (440), and (622 because of the presence of cubic In₂O₃ species (JCPDS Card no. 00-006-0416).

3.4 High resolution transmission electron microscope (HRTEM) and elemental mapping by STEM-EDS analysis

HRTEM analysis was employed to study the morphology of nanoparticles, as well as to provide evidence on their distribution throughout the mesoporous silica matrix. The three dimensional sponge-like morphology¹⁷ is evident from the HRTEM image (ESI, Fig. S3†) and particle size distribution (Fig. 3) of In-TUD-1 catalysts (In/Si = 1/100, and 4/100). Nanoparticles of 2–3 nm size well dispersed in the sponge-like matrix were observed for In-TUD-1 catalysts (In/Si = 1/100, and 4/100).

Fig. 3 Particle size distribution of In-TUD-1 (a) In/Si = 1/100 and (b) In/Si = 4/100 catalysts as obtained for HRTEM analysis.

As observed in Fig. 3, the average size of the nanoparticles is bigger for In-TUD-1 (In/Si = 4/100) compared with lower indium loading (In/Si = 1/100). Nuclei formation and subsequent growth in the sol–gel synthesis largely depended on the concentration of precursor salt during synthesis.³⁶ The formation of bigger particles strengthened observations from N₂-physisorption and SAXS studies as discussed earlier in this paper.

The presence of indium in the TUD-1 matrix was confirmed by STEM analysis as shown in Fig. 4. Uniform distribution of indium in In-TUD-1 (In/Si = 1/100) also was obtained. Si and O atomic mapping confirmed the sustaining of the sponge-like matrix in In-TUD-1 (In/Si = 1/100) catalyst; however, at higher loading (In/Si = 4/100) of indium the spots were more localized.

Fig. 4 Elemental mapping by STEM analysis of (a) In-TUD-1 (In/Si = 1/100) and (b) In-TUD-1 (In/Si = 4/100) catalysts.

3.5 Fourier transform infrared (FT-IR) spectra

The FT-IR spectra of indium doped and undoped TUD-1 catalysts are shown in the ESI (Fig. S4†). A strong adsorption band was found at 960 cm⁻¹, attributed to the stretching frequency of the terminal silanol (Si–OH) group. The bands at 1090 and 804 cm⁻¹ indicated the framework vibrations, which were assigned to the asymmetric and symmetric stretching vibrations of Si–O–Si bridges, respectively.³⁷

In the case of In-TUD-1 (In/Si = 4/100), the band at 804 cm⁻¹ has the lowest intensity because of the high percentage of indium metal incorporated into the TUD-1 matrix with successive formation of the Si–O–M (M = In) moiety.²⁵ In the hydroxyl region, the weak band at 1637 cm⁻¹ and the broad band at 3470 cm⁻¹ could be attributed to the combination of stretching vibration of silanol groups or silanol “nests” with cross hydrogen-bonding interactions and the H–O–H stretching mode of physisorbed water present in the prepared catalysts.

3.6 X-ray photoelectron spectroscopy (XPS)

XPS analysis was carried out to determine the core electron binding energy (BE) values of the corresponding elements present in In-TUD-1 catalysts. The O (1s) core level XPS spectra of TUD-1 and In-TUD-1 catalysts with two different indium loadings are shown in Fig. 5. As reported in the literature, for the mixed oxides, the O (1s) peak appeared in the range 530.6–534.5 eV for the M–O–M′ bond, where M and M′ are two metals of different electronegativity.^38–40 In the present case, for two In-TUD-1 catalysts, the O (1s) peak appeared in the range 532.9–533.4 eV, as a result of the covalent oxygen present in the Si–O–Si bond. As only one O (1s) peak was observed in the present case, it can be inferred that the isolated indium species was highly dispersed on the silica matrix.

Fig. 5 O (1s) core level XPS spectra of (a) TUD-1, (b) In-TUD-1 (In/Si = 1/100) and (c) In-TUD-1 (In/Si = 4/100) catalysts.

The Si (2p) core level XPS spectra for the two different samples, TUD-1 and In-TUD-1 catalysts, are shown in Fig. 6. The peak at 104.4 eV indicates that silica is present in its +4 oxidation state¹⁸ in the In-TUD-1 (In/Si = 4/100) catalyst, which is similar to other metal-substituted TUD-1 catalysts.

Fig. 6 Si (2p) core level XPS spectra of (a) TUD-1, (b) In-TUD-1 (In/Si = 1/100) and (c) In-TUD-1 (In/Si = 4/100) catalysts.

The In (3d) core level XPS spectra of the prepared catalysts are shown in Fig. 7. The peak maximum for 3d_5/2 centered on 445 eV, implying the existence of indium in its +3 oxidation state in the In-TUD-1 catalyst. No peak around 444 eV due to metallic indium was observed by peak fitting of the XPS spectra.⁶ For the In-TUD-1 catalyst (In/Si = 4/100) the peak of the oxidized indium species shifted to a lower binding energy value. From the observed shift of Si (2p) and In (3d) binding energy values, it can be said that the nature of electronic interaction was different for In-TUD-1 (In/Si = 1/100) catalysts after higher indium loading. The quantitative XPS result (Table 3) showed that the surface concentration of indium was lower for In-TUD-1 (In/Si = 4/100) compared with In-TUD-1 (1/100) catalyst. The concentration of surface oxygen species was similar in both cases. As the activity was changed because of the change in surface indium species, it might be said that the In/Si perimeter interface was the active center for generation of electrophilic oxygen species from molecular oxygen. For the In-TUD (In/Si = 1/100) catalyst the smaller nanoparticles had more contact area in the perimeter interface, resulting in higher catalytic activity as described later.⁴¹

Fig. 7 In (3d) core level XPS spectra of (a) In-TUD-1 (In/Si = 1/100) and (b) In-TUD-1 (In/Si = 4/100) catalysts.

Table 3 Quantitative XPS data for In-TUD-1 catalysts

Catalyst	In (3d)	Si (2p)	O (1s)
	At.%	At.%	At.%
In-TUD-1 (In/Si = 1/100)	0.03	38.8	56.85
In-TUD-1 (In/Si = 4/100)	0.01	41.0	56.34

---

3.7 Temperature programmed reduction (TPR) studies

In the temperature range 450 °C to 650 °C, two main reduction peaks are seen in the TPR profile of different indium-loaded TUD-1 (Fig. 8). The appearance of the reduction peak in the temperature range 450 °C to 525 °C might be because of reduction of the highly dispersed indium species formed during preparation of In-TUD-1 catalysts from strong interaction between surface indium species and the protonic acid sites of TUD-1 material.⁴² The H₂ uptake peak around 550 °C to 650 °C could be assigned to reduction of the large-grained In₂O₃ crystallite phase.⁴³ As evident from the TPR profile, the In₂O₃ species were highly dispersed in the In-TUD-1 (In/Si = 1/100) catalyst, but in the In-TUD-1 (In/Si = 4/100) catalyst, the indium oxide species were agglomerated on the surface of the TUD-1 material.

Fig. 8 H₂-TPR profiles of (a) TUD-1, (b) In-TUD-1 (In/Si = 1/100) and (c) In-TUD-1 (In/Si = 4/100) catalysts.

3.8 Temperature programmed oxidation (TPO) studies

The TPO profile of In-TUD-1 catalysts is shown in Fig. 9. At higher loading, O₂ consumption per gram of catalyst dropped steeply and, also, the oxygen chemisorptions peak shifted to higher temperature (Table 4). The TPO observation indicated that the active sites were accessible with lower loading, whereas with higher loading the oxygen uptake was less.

Fig. 9 TPO profile of (a) In-TUD-1 (In/Si = 1/100) and (b) In-TUD-1 (In/Si = 4/100) catalysts.

Table 4 Temperature programmed oxidation (TPO) results of In-TUD-1 catalysts

Catalyst	Peak temperature (°C)	O2 uptake (μmol g^−1)
In-TUD-1 (In/Si = 1/100)	89.8	140.1
In-TUD-1 (In/Si = 4/100)	120.5	53.21

---

4. Catalytic activity studies

The catalytic activity of In-TUD-1 catalysts with two different amounts of indium loaded (In/Si = 1/100; 4/100) towards the epoxidation reaction of styrene using molecular O₂ is shown in Table 5. Maximum conversion of styrene, and also selectivity of styrene oxide, was observed in In-TUD-1 (In/Si = 1/100) catalyst compared with In-TUD-1 (In/Si = 4/100). The product was obtained only when using DMF as a solvent (ESI, Table S5†). The increased styrene conversion for In-TUD-1 (In/Si = 1/100) catalyst might be because of the homogeneous distribution of indium oxide nanoparticles over the TUD-1 matrix, which is supported by high-resolution TEM images (ESI, Fig. S3†) for In-TUD-1 (In/Si = 1/100) catalyst. The lower TOF value at higher indium loading might be caused by reduced accessibility to indium active sites, and agglomeration of the particles, as supported by HRTEM results. An interesting observation was found when TOF was calculated based on moles of indium present as well as amount of oxygen uptake. The TOF value considering indium content was higher in the case of In-TUD-1 (In/Si = 1/100) compared with In-TUD-1 (In/Si = 4/100). However, the TOF calculated based on oxygen uptake during the TPO experiment was lower for In-TUD-1 (In/Si = 1/100) with respect to In-TUD-1 (In/Si = 4/100). It is clear that oxidizing sites were not the sole active sites. The acidity of In-TUD-1 (In/Si = 4/100) was found to be higher than In-TUD-1 (In/Si = 1/100) catalyst, as found from the NH₃-TPD result (ESI, Fig. S6†). This implies that the acidic sites resulting from In(III) took part in the reaction, enhancing the hydrolysis of styrene epoxide to benzaldehyde as shown in the Table 5.

Table 5 Catalytic data for In-TUD-1 catalysts for styrene epoxidation reaction at different indium loadings^a

Catalyst	Conversion (%)	Selectivity (%)		Indium% as per EDX		O2 uptake (μmol g^−1)	TOF* (h^−1)^a	TOF* (h^−1)^b
		Styrene oxide	Benzaldehyde	Wt%	At.%
a Reaction conditions: DMF, 10 mL; styrene, 6.5 mmol; dodecane, 0.1 mL; catalyst, 0.1 g; O2 flow rate, 10 mL min^−1; reaction time, 8 h; *TOF = moles of styrene converted/moles of indium present in the catalyst per gram per hour [where a: as per indium content and b: as per oxygen uptake].
In-TUD-1 (In/Si = 1/100)	16.6	45.5	54.5	3.61	0.63	140.1	4.5	1.5
In-TUD-1 (In/Si = 4/100)	14.4	32.3	67.7	7.80	1.42	53.21	1.7	2.2

---

4.1 Effect of reaction temperature

The influence of reaction temperature on the oxidation of styrene with molecular O₂ over In-TUD-1 (In/Si = 1/100) catalyst is shown in Table 6. At the beginning, the conversion of styrene increased with increase in reaction temperature. For example, styrene conversion was 11.26% at 70 °C and it quickly rose to 24.76% at 130 °C temperature. With further increase of temperature, styrene conversion and selectivity were reduced, as shown in Table 6. As expected, TOF value increased with temperature because of greater styrene conversion. As reported in the literature,²⁵ at higher temperature the rate of the styrene epoxidation reaction was determined by diffusion of reactants to the active sites. The decrease of selectivity of styrene oxide at higher temperature might be a result of the epoxide ring opening reaction. From this experimental result it seems that an appropriate reaction temperature is essential to obtain maximum styrene oxide selectivity.

Table 6 Catalytic data for In-TUD-1 (In/Si = 1/100) catalyst for styrene epoxidation reaction at different temperatures^a

Catalyst	Temperature (°C)	Conversion (%)	Selectivity (%)		TOF* (h^−1)^a	TOF* (h^−1)^b
			Styrene oxide	Benzaldehyde
a Reaction conditions: DMF, 10 mL; styrene, 6.5 mmol; dodecane, 0.1 mL; catalyst, 0.1 g; O2 flow rate, 10 mL min^−1; reaction time, 8 h; *TOF = moles of styrene converted/moles of indium present in the catalyst per hour [where a: as per indium content and b: as per oxygen uptake].
In-TUD-1	70	11.2	32.1	67.8	3.0	0.65
In-TUD-1	90	14.6	42.1	57.8	4.0	0.85
In-TUD-1	110	16.5	45.4	54.5	4.3	1.00
In-TUD-1	130	24.7	57.0	42.9	6.4	1.43
In-TUD-1	150	23.0	54.3	45.7	6.0	1.33

---

4.2 Effect of O₂ flow rate

A significant effect was seen on catalytic performance by variation of molecular oxygen, as shown in Table 7. It was found that at 10 mL min⁻¹ O₂ flow rate, styrene conversion hardly changed but styrene oxide selectivity dropped continuously. This indicated that optimum styrene conversion was reached at 10 mL min⁻¹ O₂ flow, but selectivity to styrene oxide decreased continuously at higher O₂ flow. However, the TOF value remained constant with increase in O₂ flow.

Table 7 Catalytic data for In-TUD-1 (In/Si = 1/100) catalyst for styrene epoxidation reaction at different O₂ flow^a

Catalyst	O2 flow mL min^−1	Conversion (%)	Selectivity (%)		TOF* (h^−1)^a	TOF* (h^−1)^b
			Styrene oxide	Benzaldehyde
a Reaction conditions: DMF, 10 mL; styrene, 6.5 mmol; dodecane, 0.1 mL; catalyst, 0.1 g; reaction temperature, 130 °C; reaction time, 8 h; *TOF = moles of styrene converted/moles of indium present in the catalyst per hour [where a: as per indium content and b: as per oxygen uptake].
In-TUD-1	5	22.0	56.2	43.7	6.0	1.30
In-TUD-1	10	24.7	57.0	42.9	6.4	1.43
In-TUD-1	20	24.6	54.4	45.5	6.4	1.43
In-TUD-1	25	23.7	53.9	46.1	6.2	1.40

---

4.3 Effect of regeneration

Catalyst recyclability is an important parameter for catalyst commercialization. Deactivated In-TUD-1 (In/Si = 1/100) catalyst was regenerated by washing with acetone, then dried overnight at 80 °C. Catalytic activity remained steady for up to four regeneration cycles over In-TUD-1 (In/Si = 1/100) catalyst, as is presented in Fig. 10.

Fig. 10 Activity results of regenerated In-TUD-1 (In/Si = 1/100) catalyst for styrene epoxidation reaction.

4.4 Kinetic studies

For the kinetic studies, styrene was put into a batch reactor at 130 °C and O₂ was passed continuously at flow rate of 10 mL min⁻¹. The rate constant was calculated for different styrene reaction orders, and it was found that for order n = 1, the data obtained experimentally fit well with the calculated data. The corresponding rate constant (k) for the reaction was 0.027 h⁻¹ (ESI, Table S7 and Fig. S8†). R² values (Table S7†) and best fit (in Fig. S8†) were used as the criteria for selection of reaction order. The plot of simulated and experimental conversions showed that the best fit was obtained considering psuedo-first order kinetics for styrene conversion (Fig. 11). The apparent activation energy for the reaction was 12.138 kJ mol⁻¹ (ESI, Table S9 and Fig. S10†).

Fig. 11 Plot of simulated and experimental conversion (%) vs. time (h) for styrene epoxidation reaction in a batch reactor.

5. Theoretical calculation

To confirm the necessity of indium nanoparticles on silica surface for oxygen activation, computational studies were undertaken. The optimized structure representing In₂O₃ nanoparticles on the SiO₂ surface in the presence of oxygen molecules is given in Fig. 12. After stabilizing In₂O₃ nanoparticles on the SiO₂ surface, O₂ molecules were used as adsorbate for this new structure.

Fig. 12 Optimized structure for O₂ as adsorbate on SiO₂ surface with In₂O₃ nanoparticles, along with partial charges on the atoms of adsorbed oxygen.

Potential functions can estimate charge distribution correctly, which is a crucial indicator of molecular reactivity. The charge equilibration method (QEq) was used to calculate the charges of atoms. The charges of atoms on oxygen molecules were assigned to show that these were influenced (polarized/delocalized) on adsorbed oxygen molecules by the presence of indium oxide nanoparticles on the silica surface indicating oxygen activation. It was observed that the charges on oxygen atoms in molecules varied depending on their location. Minimum charge polarization on O₂ (δq: ≤0.05) was induced when the oxygen molecule was in close proximity to the silica surface only, and maximum charge polarization on O₂ (δq: 0.2) was induced near indium oxide. Simulation results using the COMPASS force field method were further corroborated with a more simplistic approach using the UFF force field method (ESI, Fig. S11†). It was clearly observed that very little polarization of oxygen molecules occurred on the silica surface (Δq: 0.03–0.09), while the oxygen molecules around the indium oxide nanoparticle on the silica surface were prominently polarized (Δq: 0.2–2.0). It may be concluded that the oxygen molecule was highly polarized when it was near the indium oxide stabilized on silica surface. Active oxygen species formation occurs through reductive activation of di-oxygen (generally) at catalyst surfaces; for semiconducting oxides the metal sites of oxygen activation become progressively more oxidized, consequently, the bonding interaction between the oxygen and metal site becomes progressively stronger.⁴⁴ Molecular oxygen polarization may be an indicator of its activation. Thus, our calculations indicate that surface decoration of silica with indium oxide nanoparticles resulted in oxygen activation.

6. Conclusion

In conclusion, we demonstrated a direct styrene epoxidation reaction using molecular O₂ as an oxidant over In-TUD-1 catalyst. For the first time, it was shown that indium oxide nanoparticles can activate molecular O₂ only when dispersed over silica matrix. The electronic interaction differed for the higher loading of indium as evident from XPS studies. The catalytic activity studies showed that lower loading of indium had better activity towards the styrene epoxidation reaction. The computational study showed that the molecular O₂ was polarized when indium oxide nanoparticles were dispersed over the silica surface. Little polarization of O₂ molecules was obtained on the bare silica surface. This research on the activation of molecular O₂ by indium oxide might contribute to research towards several other oxidation reactions such as photocatalytic oxidation of water for production of hydrogen.

Acknowledgements

BC would like to acknowledge DST Govt. of India for a research grant (scheme SB/S1/PC-10/2012) and DST Govt. of India for a research grant (INT/NL/FM/P-002/2013). CS and SR acknowledge UGC for a research fellowship. RK wants to acknowledge ISM for providing a research fellowship. AKS acknowledges funding under CSIR-CSC0129&0125 scheme. AR acknowledges CSIR for fellowship under PGRPE scheme.

Notes and references

1. B. M. Reddy and B. Chowdhury, Langmuir, 2001, 17, 1132–1137.
2. N. H. Chou, X. Ke, P. Schiffer and R. E. Schaak, J. Am. Chem. Soc., 2008, 130, 8140–8141.
3. C. Shifu, Y. Xiaoling, Z. Huaye and L. Wei, J. Hazard. Mater., 2010, 180, 735–740.
4. (a) D. B. Buchholz, Q. Ma, D. Alducin, A. Ponce, M. J. Yacamán, R. Khanal, J. E. Medvedeva and R. P. H. Chang, Chem. Mater., 2014, 26, 5401–5411; (b) V. D. Ashok and S. K. De, J. Phys. Chem. C, 2011, 115, 9382–9392.
5. J. Chandradass, D. S. Bae and K. H. Kim, Adv. Powder Technol., 2011, 22, 370–374.
6. Z. M. Detweiler, J. L. White, S. L. Bernasek and A. B. Bocarsly, Langmuir, 2014, 30, 7593–7600.
7. J. Du, M. Yang, S. N. Cha, D. Rhen, M. Kang and D. J. Kang, Cryst. Growth Des., 2008, 8, 2312–2317.
8. M. Kumar, V. N. Singh, B. R. Mehta and J. P. Singh, J. Phys. Chem. C, 2012, 116, 5450–5455.
9. U. Schneider and S. Kobayashi, Acc. Chem. Res., 2012, 45, 1331–1334.
10. S. Jana, B. Dutta, R. Bera and S. Koner, Langmuir, 2007, 23, 2492–2496.
11. B. Singh and A. K. Sinha, J. Mater. Chem. A, 2014, 2, 1930–1939.
12. W. Li, Y. Gao, W. Chen, P. Tang, W. Li, Z. Shi, D. Su, J. Wang and D. Ma, ACS Catal., 2014, 4, 1261–1266.
13. J. M. Judge, Organic Peroxide, ed. D. Swern, Wiley-Interscience, New York, 1971, 2, p. 963.
14. Z. Opre, T. Mallat and A. Baiker, J. Catal., 2007, 245, 482–486.
15. B. Chowdhury, J. J. Bravo Suarez, M. Date, S. Tsubota and M. Haruta, Angew. Chem., Int. Ed., 2006, 45, 412–415.
16. J. R. Monnier, Appl. Catal., A, 2001, 221, 73–91 and references therein.
17. A. K. Sinha, S. Seelan, S. Tsubota and M. Haruta, Angew. Chem., Int. Ed., 2004, 43, 1546–1548.
18. A. K. Sinha, S. Seelan, M. Okumura, T. Akita, S. Tsubota and M. Haruta, J. Phys. Chem. B, 2005, 109, 3956–3965.
19. X. Yang, S. Gao and Z. Xi, Org. Process Res. Dev., 2005, 9, 294–296.
20. Q. Yang, S. Wang, J. Lu, G. Xing, Z. Feng, Q. Xin and C. Li, Appl. Catal., A, 2000, 194, 507–514.
21. S. B. Kumar, S. P. Mirajkar, C. G. Paris, P. Kumar and R. Kumar, J. Catal., 1995, 156, 163–166.
22. N. Ma, Y. Yue, W. Hua and Z. Gao, Appl. Catal., A, 2003, 251, 39–47.
23. K. Yamaguchi, K. Ebitani and K. Kaneda, J. Org. Chem., 1999, 64, 2966–2968.
24. I. Kirm, F. Medina, X. Rodriguez, Y. Cesteros, P. Salagre and J. Sueiras, Appl. Catal., A, 2004, 272, 175–185.
25. S. Rahman, C. Santra, R. Kumar, J. Bahadur, A. Sultana, R. Schweins, D. Sen, S. Maity, S. Mazumdar and B. Chowdhury, Appl. Catal., A, 2014, 482, 61–68.
26. Y. Li, Y. Guan, R. A. van Santen, P. J. Kooyman, I. Dugulan, C. Li and E. J. M. Hensen, J. Phys. Chem. C, 2009, 113, 21831–21839.
27. S. Mandal, S. Rahman, R. Kumar, K. K. Bando and B. Chowdhury, Catal. Commun., 2014, 46, 123–127.
28. S. Mandal, A. M. Sinha, B. Rakesh, R. Kumar, A. Panda and B. Chowdhury, Catal. Commun., 2011, 12, 734–738.
29. http://accelrys.com/products/materials-studio/index.html.
30. S. Chang, T. Yoshioka, M. Kanezashi, T. Tsuru and K. Tung, Chem. Commun., 2010, 46, 9140–9142.
31. T. C. Ionescu, F. Qi, C. McCabe, A. Striolo, J. Kieffer and P. T. Cummings, J. Phys. Chem. B, 2006, 110, 2502–2510.
32. B. P. C. Hereijgers, T. M. Eggenhuisen, K. P. de Jong, H. Talsma, A. M. J. van der Eerden, A. M. Beale and B. M. Weckhuysen, J. Phys. Chem. C, 2011, 115, 15545–15554.
33. A. Ramanathan, T. Archipov, R. Maheswari, U. Hanefeld, E. Roduner and R. Roger Glaser, J. Phys. Chem. C, 2008, 112, 7468–7476.
34. D. Sen, S. Mazumder, J. S. Melo, A. Khan, S. Bhattyacharya and S. F. D'Souza, Langmuir, 2009, 25, 6690–6695.
35. D. Sen, J. Bahadur, S. Mazumder and P. S. Bhattacharya, Soft Matter, 2012, 8, 10036–10044.
36. G. J. A. A. Soler-Illia, C. Sanchez, B. Lebeau and J. Patarin, Chem. Rev., 2002, 102, 4093–4138.
37. A. Bordoloi and S. B. Halligudi, J. Catal., 2008, 257, 283–290.
38. S. Mandal, K. K. Bando, C. Santra, S. Maity, O. O. James, D. Mehta and B. Chowdhury, Appl. Catal., A, 2013, 452, 94–104.
39. B. M. Reddy, B. Chowdhury, I. Ganesh, E. P. Reddy, T. C. Rojas and A. Fernandez, J. Phys. Chem., 1998, 102, 10176–10182.
40. C. Santra, S. Rahman, S. Bojja, S. Maity, D. Sen, O. O. James, A. K. Mohanty, S. Mazumder and B. Chowdhury, Catal. Sci. Technol., 2013, 3, 360–370.
41. B. Chowdhury, J. J. Bravo-Suárez, N. Mimura, J. Lu, K. K. Bando, S. Tsubota and M. Haruta, J. Phys. Chem. B, 2006, 110(46), 22995–22999.
42. X. Zhou, X. Xu, T. Zhang and L. Lin, J. Mol. Catal. A: Chem., 1997, 122, 125–129.
43. P. W. Park, C. S. Ragle, C. L. Boyer, M. Lou Balmer, M. Engelhard and D. McCready, J. Catal., 2002, 210, 97–105.
44. R. Schlögl, Concepts in Selective Oxidation of Small Alkane Molecules, in Modern Heterogeneous Oxidation Catalysis: Design, Reactions and Characterization, ed. N. Mizuno, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, 2009, ch. 1,  DOI:10.1002/9783527627547.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra03400k

---