Ternary composites for glycerol conversion: The influence of structural and textural properties on catalytic activity

Ternary composites for glycerol conversion: The influence of structural and textural properties on catalytic activity

Applied Catalysis A: General 406 (2011) 63–72 Contents lists available at ScienceDirect Applied Catalysis A: General journal homepage: www.elsevier...

2MB Sizes 0 Downloads 59 Views

Applied Catalysis A: General 406 (2011) 63–72

Contents lists available at ScienceDirect

Applied Catalysis A: General journal homepage: www.elsevier.com/locate/apcata

Ternary composites for glycerol conversion: The influence of structural and textural properties on catalytic activity Helvio S.A. de Sousa a,1 , Francisco de Assis A. Barros a,1 , Santiago J.S. Vasconcelos a,1 , Josué M. Filho b,2 , Cleanio L. Lima b,2 , Alcemira C. Oliveira b,2 , Alejandro P. Ayala b,2 , Manoel C. Junior b,2 , Alcineia C. Oliveira a,∗ a b

Universidade Federal do Ceará, Campus do Pici-Bloco 940, Departamento de Química Analítica e Físico-Química, Langmuir Lab de Adsorc¸ão e Catalise, Fortaleza, Ceará, Brazil Universidade Federal do Ceará, Campus do Pici-Bloco 922, Departamento de Física, Fortaleza, Ceará, Brazil

a r t i c l e

i n f o

Article history: Received 7 June 2011 Received in revised form 5 August 2011 Accepted 9 August 2011 Available online 16 August 2011 Keywords: Ternary composites Characterisations Activity Synthesis

a b s t r a c t Ternary composites were prepared and evaluated regarding their catalytic performance for the glycerol dehydration. The results of XRD and Raman spectroscopy revealed that the nanoparticles that were obtained contained phases such as CeO2 –ZrO2 , CeO2 , NiO, Co3 O4 , and Al2 O3 dispersed on SiO2 . The solids possessing elevated textural properties with basic and acidic sites were active in the conversion of glycerol. The mixed CeO2 –ZrO2 and Co3 O4 nanoparticles of the ternary composite was stabilised by silica and favoured glycerol conversion and selectivity to 1-hydroxyacetone and acrolein due to the stronger resistance to the phase transformations, when compared to other solids that had a high acidity and were quickly deactivated in the reaction media. Variations of the space velocity and the contact time defined the optimal conditions for glycerol conversion and acrolein selectivity. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Currently, researchers are focusing on the development of alternatives to valorise glycerol for clean processes in biodiesel plants and industries. Because of its environmentally benign and nonpetroleum-based process, gas-phase dehydration of glycerol is a suitable route for the production of valuable chemical intermediates such as acrolein and hydroxyacetone [1,2]. Various solids have been used as catalysts for the transformations of glycerol, including zeolites, sulphates, heteropolyacids, oxides and silica [1–10]. Despite disagreements over the efficiency of these solids, the general functionalities observed include the following: (i) glycerol is converted into acrolein due to Brønsted acid sites, (ii) 1-hydroxyacetone can be produced with basic sites or with Lewis acid sites, and (iii) the by-products formed are due to parallel and consecutive reactions that involve glycerol. The major disadvantages of using acid catalysts for the dehydration of glycerol are their short catalytic life, the inability to tailor the properties of the catalyst to product demand, the problems that

∗ Corresponding author. Tel.: +55 85 3366 99 82; fax: +55 85 3366 99 82. E-mail address: [email protected] (A.C. Oliveira). 1 Tel.: +55 85 3366 99 82; fax: +55 85 3366 99 82. 2 Tel.: +55 85 3366 90 08; fax: +55 85 3366 90 08. 0926-860X/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2011.08.009

arise from deactivation by coking, phase transformation and environmental issues related to disposal. Some alternative catalysts are available; however, most of them are based on the use of expensive transition metals, which require laborious synthesis processes [10,11], or they work by increasing the conversion rate without increasing the selectivity [12]. Recent reports [12,13] have shown that the active phase, such as Ce, exhibits an extremely high activity to 1-hydroxyacetone when compared to conventional acid catalysts. However, the single component of a bulky metal or their oxide nanoparticles may be unstable, and it limits the heterogeneous catalysis, especially for monoxides. Thus, the combination of metals or metal oxide nanoparticles along with Ce is essential to achieve optimal catalytic activity. The preparation of these highly dispersed transition metals or metal oxide nanoparticles is still a challenge. There has been a growing interest in the synthesis and utilisation of nanocrystals. Oxides are widely used in industrial chemical processes as catalysts, adsorbents and ion exchangers [14]. Although the conventional oxides possess structures in which the intracrystalline volume is not accessible due to the lack of channels, a reduction in the crystal size is expected to have a strong impact on their properties. Moreover, the combination of nanometric sizes with accessible intracrystalline volumes may offer numerous advantages when using the oxides in catalytic and separation processes. Therefore, nanocasted oxides would be proper in comparison with traditional oxides.


H.S.A. de Sousa et al. / Applied Catalysis A: General 406 (2011) 63–72

Nanocasted oxides have received significant attention because of their promising applications in catalysis due to their textural parameters, such as a large surface area, large pore volume, large pore diameter, and interesting morphologies and topologies [15–17]. There have been several reports on the production of nanoparticle oxides over porous supports, which can offer the high surface area and the large pore diameter that is necessary to inhibit agglomeration and enhance the stability of the metals and metal oxide nanoparticles. Among the nanoparticles, small cerium oxide nanoparticles are interesting because of their catalytic properties and potential application in several fields, such as sensors, catalysis, and biomedicine [15,18]. Ceria-based mixed oxides have shown to be versatile solid oxygen exchangers. The redox cycle Ce4+ /Ce3+ + e− facilitates oxygen storage and release from the bulk fluorite lattice, and this property makes them ideal candidates for catalytic applications such as alcohols dehydration, and steam reforming [15,19–23]. However, the surface redox chemistry of ceria is sensitive to crystal structure defects at low temperatures [15,19], which can be tuned by substituting some of the Ce cations with ions of different sizes and/or charges [24,25]. Specifically, in case of CeO2 -based catalysts, the synergistic effect of catalyst basicity and reducibility on dehydration of alcohols resulted in enhanced activities [12,19]. In the present paper, we report the synthesis of ternary CeO2 nanoparticles that were dispersed on silica to prepare the ternary CeO2 -based nanocomposite catalysts. Using glycerol as a feed, the gas-phase dehydration of glycerol to value-added chemical intermediates over the ternary CeO2 -based nanocomposite catalysts was investigated. The catalysts were characterised by XRD, TEM, Raman spectroscopy, SEM-EDX, and textural properties. For the ternary CeO2 -based nanocomposites, the effect of the structure, as well as the textural properties of the solids on their catalytic performance, is discussed in detail.

2. Experimental 2.1. Preparation of the ternary composites The oxides that were synthesised by nanocasting were CeO2 –Al2 O3 and NiO, CeO2 –ZrO2 and Co3 O4 , and CeO2 –ZrO2 and NiO, and it was used a mesoporous silica/carbon composite as a sacrificial template [17]. The solids are abbreviated as CAN, CZC, CZN. The CAN, CZC, and CZN were prepared by modifying previously described methods [13,17]. Approximately, 2 mL of an equimolar mixture of (Ce(NO3 )3 ·6H2 O Aldrich 98.5%), aluminium (Al2 (NO3 )3 ·9H2 O 98%, Aldrich), and nickel (Ni(NO3 )2 ·6H2 O 98% Isofar) aqueous solutions were added through a peristaltic pump to a bequer containing 1 g of carbon nanopowder (Carbon Aldrich 99.5%). The resulting mixture was stirred and 0.3 g of sulphuric acidic was added to the previous solution for 1 h. After drying at 40 ◦ C for 12 h, the procedure was repeated twice and the powder was obtained. Then, an equimolar mixture of water and tetrapropylammonium hydroxide (TPAOH ∼ 1.0 M, Aldrich) and tetra ethylorthosilicate (TEOS, 98%, Aldrich) was added to the powder under stirring for 2 h. Subsequently, the mixture was submitted to hydrothermal treatment at 150 ◦ C for 12 h. The resulting composite was washed, dried and calcined under nitrogen flow at 1 ◦ C min−1 until 650 ◦ C and the temperature was held for 6 h to carbonize the carbon and decompose the nitrate precursors. Subsequently, the powder was washed with a hydrofluoric solution to remove partially the silica template. To finish, the carbon was removed by heating the sample under air atmosphere to 650 ◦ C at a constant rate of 1 ◦ C min−1 and keeping the sample at that temperature for 5 h. The solid obtained was designed as CAN, CZC

and CZN were obtained by the abovementioned procedure, except with (Ce(NO3 )3 ·6H2 O, 98.5%), (Ni(NO3 )2 ·6H2 O, 98%), (ZrOCl2 ·8H2 O, 98%), and (Co(NO3 )2 ·6H2 O, 98%) as precursors. 2.2. Characterisation Powder X-ray diffraction (XRD) was carried out on an X-ray powder diffractometer Xpert MPD (Panalytical) equipped with a Co tube with Cu K␣ radiation, 40 kV, 30 mA. The measurements were taken at wide-angles (2 = 3–80◦ ). The average diameters of the oxide crystallites were estimated from the XRD patterns by applying the Scherrer equation [15]. The specific surface area and the parameters of the pore structure of the solids were determined by the adsorption isotherms of nitrogen at −196 ◦ C with an ASAP 2000 Micromeritics surface area analyser. The samples were degassed at 150 ◦ C for 24 h prior to all measurements. The BET equation was used to calculate the total specific surface area. The pore distribution was calculated by the Barret–Joyner–Halenda (BJH) method to the desorption branch of the isotherms. Inductively Coupled Plasma Optic Emission Spectroscopy (ICPOES) was performed in a Varian instrument for the determination of the chemical analysis of the calcined catalysts. Previously, the solids were digested with a mixture of hydrofluoric, nitric and chloridric acids at 90 ◦ C in a sand bath. The metallic content of the solids was determined by ICP-OES. Transmission electron microscopy (TEM) was performed on a Philips EM420 transmission electron microscope instrument at 120 kV with a resolution of 0.17 nm. Scanning electron microscopy (SEM) measurements were conducted on a JEOL LV-5800 electron microscope equipped with an EDX Link Analytical QX-20000 system coupled to the SEM microscope, using an acceleration voltage of 20 kV. Raman spectra of the catalysts were obtained on a T64000 Raman spectrometer (JobinYvon triple spectrometer) under ambient conditions. A 514.5 nm Ar laser was used as the excitation source of the sample surface at a power of 20 mW. The measurements were referenced to Si at 521 cm−1 with 16 data acquisitions in 120 s. The lens focus was 100 times. Temperature programmed reduction (TPR) analysis of the catalysts was performed using in-house designed equipment with a quartz tube reactor that had an inner diameter of 6 mm and was coupled to a TCD for monitoring the consumption of hydrogen. The mass of the catalyst was 50 mg and the experiment was performed in the range of 50–1000 ◦ C. An 8% H2 /N2 mixture was used as a reducing gas with a flow rate of 100 mL min−1 after being passed through a 13X molecular sieve trap to remove water. Before analysis, samples of ca. 0.1 g were placed in the tube reactor and heated under nitrogen to a temperature of 100 ◦ C for 2 h. Spectra from X-ray photoelectron spectroscopy were recorded using a Vacuum Generator Scientific Escalab MKII X-ray photoelectron spectroscope (XPS), and Al was used as the X-ray source at a pressure of 10−8 mbar. Carbon, as measured at 284.8 eV, was used as an internal standard. The accuracy of the BE values was ±0.2 eV. The temperature programmed desorption of the CO2 (CO2 -TPD) measurements were carried out in a Quantachrome instrument to determine the basic catalytic site amount. Prior to the CO2 adsorption, all catalysts were preheated in He flow and exposed to 5% CO2 under N2 gas flow. CO2 desorption was performed under temperatures that ranged from 50 to 1000 ◦ C. The temperature desorption of ammonia (NH3 -TPD) was performed in a Shimadzu TG/DTA 50 apparatus. For these measurements, 50 mg of the samples was pretreated at 500 ◦ C for 2 h and then cooled to 120 ◦ C under He flow. Pure ammonia was injected until the adsorption was saturated and followed by purging with He for 2 h to remove any weakly adsorbed ammonia. The temper-

H.S.A. de Sousa et al. / Applied Catalysis A: General 406 (2011) 63–72


ature was increased to 500 ◦ C at a rate of 5 ◦ C min−1 to desorb the ammonia. 2.3. Catalytic tests The gas-phase dehydration of glycerol was used to evaluate the catalytic properties of the solid at 250 ◦ C, P = 1 atm, and a space velocity of 2.6 h−1 . The catalyst (100 mg) was placed in a stainless steel tubular reactor that had quartz wool to fix the bed. The system was preheated to 300 ◦ C in an oven for 1 h under nitrogen flow at 30 mL min−1 . The reaction temperature was subsequently decreased to 250 ◦ C, and the reactant mixture, which consisted of an aqueous solution of glycerol (20 wt%) was fed through a peristaltic pump to the reactor (i.d. 8 mm and 23.0 cm of length) at atmospheric pressure. The reactant liquid flow was 0.0019 g h−1 . It was vaporised and diluted with nitrogen to a flow of 1.8 L h−1 . The products were collected in an ice trap and then dissolved in an ethanol/methanol mixture. The products of the reaction were analysed using gas chromatography (GC) connected to a flame ionisation detector (FID) and a GC 300 integrator. The products of reaction were identified by gas chromatography coupled to mass spectroscopy (GC–MS). The internal standard for the GC–MS analysis was 1-butanol. Glycerol conversion and the selectivity were obtained according to the following calculations: Glycerol conversion (%) =

Product selectivity (%)=

mols of glycerol reacted × 100 mols of glycerol in the feed


mols of carbon in specific product × 100 mols of carbon in glycerol reacted (2)

The conversion and selectivity were measured under isothermal conditions at a fixed temperature. The contact time (W/F glycerol), which was defined as the ratio between the catalyst’s weight (W) and the molar flow rate of glycerol (F glycerol), and the space velocity were also varied. 3. Results and discussion 3.1. Chemical analyses, XRD and Raman measurements Table 1 shows the composition of the solids that was determined by ICP-OES. CeO2 –ZrO2 is an active element for glycerol conversion [13]. Al2 O3 is inactive; it is only present to increase the degree of dispersion of the CeO2 –ZrO2 -containing phase [25]. Previous studies have indicated that most of the mechanical properties exhibited by the prepared catalysts are due to the presence of the inactive phase [13,24–30]. In addition, both Ni and Co are active phases for glycerol transformation [18,31–33]. The composition of the catalyst, as determined by chemical analysis, is highly correlated with the nominal composition (70:7.5:16), being CAN an exception. Some residual silica remains from the silicalite precursor with the goal to stabilise the metal particles. However, the results of chemical analyses for silicon are not accurate because

Fig. 1. XRD patterns of the solids.

the silicon had not been completely dissolved by the digestion method, before the ICP-OES analysis. The XRD patterns of the samples are shown in Fig. 1. The peak intensities of CAN were weak and broad, a common feature of nanocasted oxides. For the CZN and CZC samples, sharp peaks were found and a mixed CeO2 –ZrO2 phase was observed, which is in agreement with the JCPDS no. 38-1436 database [13,34]. The cubic structure of NiO (JCPDS no. JCPDS 4-835) and Co3 O4 (JCPDS no. 431003) phases [26,35] were detected in the diffraction pattern of CZN and CZC, respectively. In addition, CeO2 (JCPDS no. 43-1002) appeared as a segregated phase over all solids. Because the compounds could form homogeneous solid solutions and the high dispersion of the cerium and zirconium was due to the synthesis conditions, the solid cerium–zirconium solutions may not be detected by XRD. However, the existence of the CeO2 –ZrO2 , NiO, and Co3 O4 phases indicated that these oxides were dispersed on the silica, which was further detected by the EDX measurements. This indicated that these oxides could be highly

Table 1 Composition of the solids determined by ICP-OES and textural properties of the composites. Sample

Molar compositiona

Sg (m2 g−1 )

Vp meso (cm3 g−1 )

Vp micro (cm3 g−1 )

Dp (nm)


70.2:7.5:15.8 71.6:7.4:16.2 73.0:8.7:17.9

52 113 87

0.13 0.34 0.26

0.03 0.11 0.06

1.0 2.0 1.9


Me1 :Me2 :Me3 in sequential order of the samples nomenclature.


H.S.A. de Sousa et al. / Applied Catalysis A: General 406 (2011) 63–72








Raman Intensity (a.u)
























Wavenumber (cm-1) Fig. 2. Raman spectra of the composites.

dispersed or small in size and thus not detectable by XRD. This was not unexpected considering the method that the oxides are formed and the detection limit of this technique. In the case of CAN, the peaks were lesser intense than the other samples, and a few peaks, which corresponded to CeO2 and NiO were detected. Also, two peaks related to ␥-Al2 O3 (JCPDS no. 750921) were detected. The average sizes of the crystallite samples were estimated by the better defined reflections (28.7◦ for CeO2 –ZrO2 , 22.8◦ for NiO, 29.3◦ for CeO2 , and 39.3◦ for Co3 O4 ) and the application of the Scherrer equation. The particle sizes of CeO2 –ZrO2 , in CZC and CZN were estimated to be 8.2 and 6.9 nm, respectively. These values suggest that the simultaneous impregnation of Ce and Zr favours a high dispersion of either Co or Ni nitrates with a particle size that was almost not identified by XRD. As the Al2 O3 phase was suggested by the XRD pattern of CAN, this indicated that the alumina was well dispersed in the CeO2 fluorite crystal lattice of a homogeneous solid solution or in silica, as was observed in the findings [27,28]. It is plausible that the synergic stabilisation of the Al2 O3 , CeO2 and SiO2 phases (the latter is further confirmed by SEM-EDX) can hinder the CAN from sintering, as observed in Ce-based compounds [18,28]. However, the NiO diffraction pattern was detected, and the crystal size was 10.2 nm. Characteristic reflections of NiAlx Oy and NiAl2 O4 [17,30] were not detected on CAN, which suggests that these species were well dispersed on the solid. The presence of cerium could prevent the formation of spinel NiAl2 O4 or its growth during calcination. The Raman spectra of samples are shown in Fig. 2. No silica mode is observed in the samples. At lower wavenumbers, CAN nanocomposite shows weak and broad modes. The bands at 456 and 638 cm−1 were attributed to the Raman active mode of F2g symmetry allowed for perfect CeO2 based cubic fluorite lattice, as for binary CeO2 compounds [24,27]. Moreover, the findings states that F2g band could be slightly displaced to higher frequency due to the insertion of ions into the CeO2

lattice [27]; as a consequence, the formation of oxygen vacancies generated by the incorporation of ions in the structure of CeO2 is likely and it can explain the shifts in Raman modes, as compared to that of literature for CeO2 [27]. Additional bands at 180, 222, 547, and 666 cm−1 were also observed, and can tentatively be attributed to Ni2+ from either to traces of NiO or to NiAl2 O4 [29,35]. These results are in line with those obtained by XRD that suggested the formation of the abovementioned phases. The Raman bands of CZC at 150, 181, 270, 381, 463, 544, 611, 675 and 713 cm−1 were detected. The triply degenerate F2g Raman mode of CeO2 was identified at 463 cm−1 [25] and the band at 611 cm−1 was assigned to the weak Raman signal produced by the partial breaking of Raman active oscillation mode of F2g symmetry [25,27,28]. The appearance of the mode at 611 cm−1 suggests that the oxides are in close interaction, which could allow the incorporation of zirconium ions in the structure of cerium oxide, giving rise to oxygen vacancies, as for ternary cerium-based compounds [25,27]. Thus, the modes at 150, 181, 279, and 713 cm−1 can be assigned to CeO2 –ZrO2 in agreement with XRD data. Also, Raman active phones modes at 181 and 544 correspond to F2g phonon modes whereas that 675 and 713 cm−1 could be also attributed to Eg and A1g phonon, which are related to Co3 O4 vibrations [26,35]. These results are in agreement with that of XRD, which suggests the formation of the abovementioned phases. Fig. 2 also shows the Raman spectrum of CZN that was obtained in the range 100–900 cm−1 . The Raman spectrum exhibited modes at 184, 279 and 713 cm−1 were associated with tetragonal CeO2 –ZrO2 [25,34] and these bands were also observed for CZC. In addition, the strong band at 456 cm−1 was assigned to the Raman signal that was produced by the partial breaking of the active Raman oscillation mode of the F2g symmetry of CeO2 [25,28]. According to Epifani et al. [36], stoichiometric NiO has two Raman bands at 151 cm−1 and 480 cm−1 . 3.2. Morphological and textural aspects of the composites and XPS measurements Fig. 3 shows the SEM-EDX images of the composites. This figure shows spherical particles and that they were similar in appearance. The SEM-EDX analysis revealed that CZC has a uniform distribution of Ce, Ni and Zr nanoparticles, and these elements are dominant in the solid (Fig. 3a). This observation is in agreement with the Raman spectrum of CZC shows the vibrations of the phases related to the nanoparticles. In addition, the silica content was lower than 20%. For CAN, a similar spherical morphology of the particles was found. The nanoparticles of Ce, Zr and Ni were distributed on the silica, and agglomeration of certain elements in some areas is seen. The TEM micrograph of CZC exhibited a plate with nanoparticles (black dots) that were dispersed on the silica. Small particles with an average size of down to 11.5 nm which are indeed close to that predicted by XRD were present in TEM image of the CZC (Fig. 3b). Moreover, the metals particles were obviously dispersed. The particle size distribution obtained by TEM analyses of CAN and CZC ranged from 10 to 200 nm. The average particle sizes in the CAN and CZN samples were 27.4 nm, 32.3 nm, respectively. The silica content on CAN and CZN was estimated by EDX to be approximately 14%. In contrast, CZN morphology illustrated that the preparation method could induce an agglomeration of the nanoparticles by a coalescence mechanism. The final conclusion is consistent with the existence of the agglomerated particles, as seen by the SEM-EDX analysis. Another explanation may be that growth of these larger crystals may be a result of the synthesis solution on the silicalite surface, as the crystal growth within the template pores was hindered due to the limited space of the pores and chains. These findings demonstrate two mechanisms of mesoporous oxide formation: first, the metal precursor locates to the external

H.S.A. de Sousa et al. / Applied Catalysis A: General 406 (2011) 63–72


Fig. 3. (a) SEM-EDX analyses of the CZC, CAN and CZN composites. (b) TEM image of the CZC composite.

surface of carbon/silica mold where the micrograins decompose without melting, and these results in a large nonporous crystal outside of the matrix pores. Second, the metal precursor melts and decomposes after penetration into the porous system of the mold where it should give a single crystal with porous mesostructured metal oxides [37]. Both mechanisms are likely in the case of CZC and CAN; thus, the melting and decomposition of the metal precursor processes could occur simultaneously, resulting in polycrystalline porous metal oxides. The textural characteristics of the solids show that the shape of the N2 adsorption–desorption isotherms of the CZC composite (Fig. 4a) is similar to that of porous silica [17]. According to the IUPAC classification, the N2 isotherms of all of the samples were type IV, which is characteristic of mesoporous solids [17,38]. The isotherms of the samples have little hysteresis loops at relatively high pressure (P/p0 ), which indicates that the crystal sizes were small and presented a relatively high external surface area with little mesoporosity. A steep increase at relatively low pressure indicated the presence of micropores according to the IUPAC classification [38]. BET surface areas (Sg ), pore volumes (Vp ) and diameters (Dp ) of the composites are listed in Table 1. Our previous studies [13,33] revealed that the surface areas and the pore volumes of the CeO2 –ZrO2 , CeO2 –Al2 O3 and NiO–Co2 O4 composites were high because of the preparation method. Obviously, the textural parameters of CZC, CAN and CZN were lower than the binary ones; this result may be attributed to the addition of a third element in the composition of the solids. Among

them, CZC has the highest surface area and pore volume, whereas CZN has the lowest. These results agree with the XRD, SEM-EDX and TEM analyses that estimated the size of the particles for the former. In addition, the specific surface area and pore volume of CZN all markedly decreased due to the agglomeration of particles. Remarkably, the mesopores appear in the samples and the pore size distribution is broadened in the CZN sample, which indicates that the sintering of the particles during calcination contributes to the formation of micropores and mesopores. The micropore volume on CZC was ca. 0.11 cm−3 g−1 , which was superior to the other solids. This result may be because the micropores are either filled with the metal precursor or not accessible to the nitrogen molecule. In addition, we cannot exclude the participation of the residual mesoporous silica, which support the nanoparticles, in the high textural properties of the silicalite template [14,15]. This means that the ternary composites synthesised by nanocasting have large surface areas and pores volumes as well as homogenous dispersion of the CZC and CAN components. Additionally, XPS analysis of CZC shows the binding energies of the main peaks of Si2p, Zr 3d, Co 2p, O 1s and Ce 3d core electrons (Fig. 4b) and the data of XPS spectra are summarised in Table 2. The binding energy (BE) of Zr 3d5/2 core electrons for all composites is approximately 182.8 ± 0.2 eV, and this value fits well within the range for Zr4+ [22], which interacts with others species such as CeO2 . The BE (ca. 103.2 ± 0.2 eV) of the Si 2p signals remained nearly unchanged, and due to the strong interaction between SiO2 and the nanoparticles, the BE of the remaining silicalite mold


H.S.A. de Sousa et al. / Applied Catalysis A: General 406 (2011) 63–72


Adsorbed volume (cm3.g-1)

dV / d(cm 3g-1)





(b) 0


4000 Si 2p



Pore diameter (nm) Intensity (cps)

3000 2500 2000 1500 1000 500

















Binding energy (eV)

Relative pressure P/Po 1900



Intensity (cps)


5800 5700


5600 1600 5500 1500

5400 5300


5200 174













Binding energy (eV)




Binding energy(eV)

18000 16000


O 1s v


v v"

Intensity (cps)

Intensity (cps)

12000 10000 8000

v uu

u u"

6000 4000 2000 526









Binding energy (eV)









Binding energy(eV)

Fig. 4. (a) N2 adsorption–desorption isotherm of the CZC composite. The figure inset is the pore diameter representation of the solid. (b) XPS spectra of CZC.

shifted from 102.1 eV [39], which indicates that Ce may preferentially interact with Si in the presence of others metals on the surface. The binding energies at 783.2 eV and 797.5 eV were attributed to Co 2p3/2 and Co 2p1/2 [40], respectively, and the additional satellite peak suggests the existence of CoO. Therefore, the difference between the BE for (Co 2p1/2 –Co 2p3/2 ) is equal to 15.6 eV. The presence of a wide range of oxygen species associated with the species present in the oxides imposes difficulties in quantifying the

changes to the O 1s spectra for all of the species present. Therefore, no attempt was made to analyse these components individually and the BE at 533.0 eV could be attributable to oxygen from hydroxyl surface species or defects from CeO2 surface, regarding the differences in the electronegativity of the elements involved and that found of the findings [24,41]. Ce 3d XPS core level spectra from CeOx compounds show a complex shape of the spectrum that can be resolved into eight components (Table 2) by least-squares fitting according to the notation

H.S.A. de Sousa et al. / Applied Catalysis A: General 406 (2011) 63–72 Table 3 Basicity and acidity measurements of the solids.

Table 2 Binding energies of Si 2p, Ce 3d, Co 2p, Zr 3d and O 1s levels for CZC. Sample

Binding energy (eV)


Si 2p

Ce 3d

Co 2p

Zr 3d

O 1s


883.5 885.3 889.2 899.0 901.8 903.2 908.4 916.1

783.2 797.5



of Burroughs [24]: v’s represent the Ce 3d5 contribution while u’s represent the Ce 3d3 contribution. As previously shown by the findings, the peaks v (u ) have been attributed to the transitions to the final state 4fo from the initial state 4fo [41]. The v (u) notation corresponds to the transitions to the 4f1 final state from the 4f1 initial state whereas v and v (u and u ) to the 4f2 final states, which comes from the ligand-to-4f1 shake-up transitions of the 4f1 initial state [24,41]. The presence of 4f1 initial state of the formal Ce4+ could be due to partial occupancy of 4f orbital, resulting from valence mixing with the oxygen ligand. The u peak suggests a 4f configuration in the formal Ce3+ state from Ce2 O3 , although it was not observed by XRD due to its high dispersion on solid surface. From these results, it can be suggested that the coexistence of both Ce3+ and Ce4+ oxidation states is distinguishable, although the Ce4+ oxidation state is predominant at temperatures of calcination as low as 600 ◦ C, for ternary cerium-based compounds [41]. Furthermore, the signal of the components in CZN and CAN is almost invisible due to silicon interference. Thus, the surface composition of CZC revealed that the amount of CeO2 –ZrO2 and Co3 O4 oxides on the surface was higher than that obtained by chemical analysis, and this indicated that the elements were mostly dispersed on a solid surface, which is consistent with the SEM-EDX analysis. 3.3. Reducibility behaviour by TPR analysis Evaluation of the reducibility of the solids was performed to understand the ability to generate oxygen vacancies of CeO2 and to transfer oxygen to the metal particle. The H2 -TPR profiles of the composite are shown in Fig. 5.





H2 consumption (a.u)



385 502






Total amount of acidic sites (mmol NH3 g−1 )

Total amount of basic sites (mmol CO2 g−1 )


0.89 0.25 1.02

0.12 0.75 0.07

The CAN sample has a weak peak centred at 441 ◦ C, which was attributed to the reduction of free NiO to Nio [20,29], as seen in Fig. 5. The typical TPR of bulk CeO2 exhibited two broad intense peaks that were associated to the stepwise reduction of CeO2 , with the first peak at 500 ◦ C due to the surface shell reduction of ceria (i.e., the most easily reducible surface-capping oxygen of ceria) [20–23]. The second peak was attributed to the total reduction of ceria by the elimination of the O2− anions from the lattice and the formation of Ce2 O3 at 800 ◦ C [38]. The shift of the peak from 500 to 687 ◦ C may be associated to the Ce4+ /Ce3+ pair of the CeO2 nanoparticles, which are difficult to reduce due to the synergic stabilisation between the CeO2 and Al2 O3 phases that was suggested by the XRD analysis and independently confirmed [18,25]. Moreover, the peak at temperatures greater than 800 ◦ C can be associated to either the second reduction step of CeO2 or NiAl2 O4 reduction [17,30], although this phase was not detected by XRD because of the detection limit of this technique. Indeed, the presence of cerium could favour the reduction of spinel NiAl2 O4 or its growth during sample calcination [18]. The TPR profile of CZN shows a weak reduction peak at lower temperatures (e.g., 220 and 315 ◦ C) that could be attributed to NiO and CeO2 reduction [20,21,42], and these peaks indicate that the interaction among NiO, CeO2 –ZrO2 nanoparticles and silica is strong. Furthermore, the TPR profile of CZN clearly displays a big reduction peak centred at 462 ◦ C and a shoulder at 520 ◦ C that when deconvoluted, reveals the typical CeO2 –ZrO2 profile of reduction being favoured by the high oxygen mobility on the zirconia surface after the incorporation of the cerium atoms into ZrO2 , which was similar to the CeO2 –ZrO2 solid solution [18,21,25]. As expected, the peaks related to the reduction of ZrO2 were not observed. In the case of CZC, despite the broad TPR profile, which indicated that the reduction of the species occurred simultaneously, the thermogram displays peaks that were centred on 385, 502 and 784 ◦ C. Thus, the redox ability of the phases was improved due to electron transfer between the Ce4+ /Ce3+ , Zr4+ and Co2+ /Co3+ species [15,40,43,44]. Electron transfer between the metal oxides is suggested to occur over Co-containing ceria oxide [38]. The lower temperature peaks are related to the reduction of CeO2 –ZrO2 in the CZN curve. The spinel Co3 O4 is reduced at temperatures greater than 600 ◦ C [40], as seen in the TPR curve. 3.4. Acidic and basic properties

462 CZN



Temperature (ºC)



Fig. 5. Temperature programmed reduction (TPR) of the composites.

The NH3 -TPD results obtained from the spectra are summarised in Table 3 The materials have similar desorption peaks at temperatures as low as 150 ◦ C due to the physical absorption of ammonia, and these peaks originate from the weak acid sites. At temperatures greater than 400 ◦ C, chemical adsorption of ammonia confirms the presence of the acidic sites over CAN and CZN. The number of acidic sites that were titrated by ammonia to these composites was 1.02 and 0.89 mmol NH3 g−1 , respectively, and the results were superior to those of CZC, where the concentration of strong acidic sites was negligible. According to these findings, Al2 O3 possesses both Brønsted and Lewis acid sites of weak to strong strength [11,17]; however, the diminution of the strength of the Lewis acid sites could be due to the replacement of the Al3+ ions by the less elec-


H.S.A. de Sousa et al. / Applied Catalysis A: General 406 (2011) 63–72

nanoparticles. Therefore, the change in the glycerol dehydration reaction does not appear to be explained simply by these characteristics. The redox ability provided by CeO2 –ZrO2 in CZN and CZC results in a greater conversion compared to CAN after 3–6 h of reaction time. Nevertheless, the lack of a stable phase in CZN (TPR analyses) greatly degrades the conversion. The origin of this could be due to a small amount of Ni2+ present on CeO2 –ZrO2 solid, which is reduced to Ni+ species as described by the findings [20]: Ce4+ –O2− –Zr4+ ↔ Ce3+ –O− –Zr4+ 3+


Fig. 6. Conversion of glycerol (closed symbols) and the selectivity (open symbols) to acrolein over the ternary composites. Reaction conditions: temperature, 250 ◦ C; pressure, 1 atm; glycerol to water ratio, 1:3; mass of the catalyst, 100 mg.

tronegative Ni2+ ions, as was the case for the MAl2 O4 spinel oxides [17]. The adsorption of pyridine shows that on the surface of zirconia, both Brønsted and Lewis acid sites of medium to strong strength are present [1]. Thus, these assumptions explain the superior acidity of CAN and CZN generated by Al3+ , Ni2+ and Zr4+ species, whereas CZC possesses a reduced number of acid sites, which are of weak strength due to the Co3+ /Co2+ and Ce4+ /Ce3+ species. CO2 -TPD measurements revealed that CeO2 based oxides have basic sites because of the higher amount of oxygen vacancies of the CeO2 [13,23]. The basicity of these oxides were decreased when either ZrO2 or Al2 O3 are mixed with the CeO2 [13]. As suggested by the XPS analysis, the surfaces of CAN and CZN were rich on the nanoparticles of the abovementioned species. This results in a balance of the acidic and basic sites on the solids. Therefore, the total amount of basic sites in CAN and CZN were 0.07 and 0.12 mmol CO2 g−1 , respectively, were greatly decreased. However, the Co3 O4 species associated with CeO2 –ZrO2 had an increased basicity of CZC (0.75 mmol NH3 g−1 ). No influence of the residual silica on the acidity or basicity from was observed due to the inert character of the SiO2 oxide. 3.5. Catalytic results for glycerol dehydration The catalytic activity of the mesoporous composites for the gas phase dehydration of glycerol at 250 ◦ C with a glycerol to water ratio of 1:3 is shown in Fig. 6. The catalytic-induction capability was evaluated by means of the time on stream (TOS), which the catalysts need to achieve stable conversions. All catalysts show changes in the catalytic activity along the TOS. CZN and CAN composites take about 1 h to activate, and after 6–7 h, the conversion of glycerol declines, reaching CAN a conversion of ca. 35%. In contrast, CZN has 10% glycerol after a 10 h reaction. Unlike CZN and CAN, CZC reaches the steady state reaction stage after 4 h of TOS, and it converts approximately 60% of the glycerol. As manifested by the decreased glycerol conversion along the TOS, the textural properties and the acid–base features may be important features that lead to the stable performance in glycerol dehydration, as observed in the literature [9,13,27]. Thus, judging from the fact that the high surface area of CZC has a homogenous distribution of nanoparticles that were supposed to be CeO2 –ZrO2 and Co3 O4 on silica and provided basic sites of medium strength, high catalytic stability of the referred mesoporous solid was expected. More importantly, it was seen that the other solids had similar features, except for the homogenous distribution of the

+ Ni


→ Ce


+ Ni


(I) (II)

The experimental evidence that CeO2 promotes NiO reduction had been shown by TPR and confirmed by the assumption that glycerol is considered to be a strong reductant. CAN possess a stable spinel phase, like NiAl2 O4 , and dehydrates glycerol with a stable performance; however, the existence of strong acidic sites allows for an easier deactivation by coking. The nanocasted oxides in the catalytic reaction display a selectivity that is opposite the behaviour of the conversion, with CAN being an exception. Selectivity to 1-hydroxyacetone over CZC and CZN increased monotonously with an increasing TOS, whereas CAN production of 1-hydroxyacetone dropped after 10 h of reaction time. Investigation of these results also revealed that the acrolein selectivity of CAN was related to its high acidity; however, acrolein is transformed by consecutive reactions that involve glycerol, mainly due to coking and as previously observed with CeO2 based catalysts [13]. According to these findings, Lewis sites are known to interact with the ␲ electrons from C C and/or C O bonds and may hinder fast desorption of the acrolein target product, which decreases its selectivity [45] and explains the decreased selectivity of CAN. In case of CZC, the balance between acid and basic sites avoids the coking. The main by-product obtained was 1-hydroxyacetone and others products, such as acetone, ethanol, propyleneglycol, acetaldehyde, 1,3 propanediol, ethyleneglycol, acetic acid, methanol, aromatics and cyclic compounds (soketal) were also obtained. Thus, the highest acrolein selectivity was obtained with the most acidic catalyst (CAN), followed by CZN, which possesses basic and acidic sites, whereas CZC has sites that have higher basicity than CZN and is poorly selective to acrolein. Both CZC and CAN showed a high production of 1-hydroxyacetone. Because CAN and CZN performed poorly in the conversion of glycerol, the complementary studies were performed only on CZC to examine its stability. The space velocity (SV) of the mixture of glycerol and water feeds varied, as seen in Fig. 7. Fig. 7a shows the result of increasing the SV to the rate of glycerol transformation over CZC. At low SV, glycerol conversion decreased linearly to 0.0%. The rate reached a plateau above 2.6 h−1 of glycerol and water feeds, and 82% conversion was attained. A similar tendency was observed in the case of acrolein selectivity with maximum values attained at 39%. The main by-product of the reaction was 1-hydroxyacetone, as expected due to the basicity of the solid. Therefore, at a SV of 2.6 h−1 , the chemical equilibrium was achieved for both conversion and selectivity. Fig. 7b displays the effect of the contact time (W/F glycerol) on the dehydration of glycerol over the CZC catalyst. It was observed that less than 5% of the glycerol was converted by using 8.2 gcat smmol−1 . This was at the expense of 1-hydroxyacetone selectivity, which was nearly 40%. When the W/F was increased to 15.3 gcat smmol−1 , the conversion of glycerol was 30%, and 1-hydroxyacetone selectivity was enhanced to 58%. Further, increasing the W/F to 31.9 gcat smmol−1 led to the linear increase of glycerol, but above 46.4 gcat smmol−1 , the conversion was greatly diminished to 14%. The production of 1-hydroxyacetone was not favoured when the catalyst weight was increased from 8.2 gcat smmol−1 up to W/F 31.9 gcat smmol−1 due to the formation of acetaldehyde, acetone, allyl alcohol and other

H.S.A. de Sousa et al. / Applied Catalysis A: General 406 (2011) 63–72


4. Conclusions The nanocasting method produced nanoparticles of ternary CeO2 based composites. The effects of the Al, Ni, Co, or Zr promoters were studied individually and in combination in the conversion of glycerol by means of its dehydration. The high surface area of the CeO2 –ZrO2 Co3 O4 ternary composite had the CeO2 –ZrO2 phase and both basic and acidic sites of medium to strong strength were produced. This resulted in a high activity, an enhanced 1hydroxyacetone and acrolein production and glycerol conversion due to the Co3 O4 phase stabilising the nanoparticles of CeO2 –ZrO2 . Although CeO2 –NiO and Al2 O3 and CeO2 –ZrO2 and NiO are both strong acids that are active and selective in obtaining acrolein, both were deactivated after 6 h under the reaction conditions of 250 ◦ C, atmospheric pressure, glycerol/water (1:3) due to physical degradation of the catalysts. The best performance, in terms of the highest acrolein and 1-hydroxyacetone selectivities and conversion, were obtained from CeO2 –ZrO2 and Co3 O4 by varying the SV and the contact time, and the optimal conditions were T = 250 ◦ C, SV = 2.6 h−1 and to W/F of 31.9 gcat smmol−1 . Acknowledgements The authors are grateful for financial support from CNPq for this research project (CNPq/CT 574194/2008-8), as well as FUNCAP with Contract 0011-00206.01.00/09. We thank to Dr. P. Bargiela and M.G.C. Rocha at UFBA-Brazil for providing XPS results. References

Fig. 7. (a) Influence of the SV on the conversion of glycerol over the CZC catalyst. (b) Conversion of glycerol over CZC with time-on-stream at 250 ◦ C at different contact times (W/F glycerol).

compounds, including ethers and cyclic by-products. Thus, a proper amount of available CZC sites (100 mg) is required for the conversion of glycerol without varying the number of basic sites of the catalyst. Murugan and Ramaswamy [46] proposed that over a CeO2 sample enriched with surface Ce3+ –O− –Ce4+ -type defect sites, the adsorption of reactants promotes the subsurface oxygen migration and accordingly, facilitates surface reorganization by two processes: (i) the migration of sub-lattice oxygen to the surface and (ii) the replenishment of the oxygen ion vacancy from the gas phase to the bulk. From the above investigation, we support our results for the increased glycerol conversion of CZC due to the presence of the abovementioned redox and acid-base sites (e.g., CeO2 –ZrO2 ), as evidenced by Raman, XRD and TPR results. Additionally, we cannot exclude the possibility of both of the OH groups in glycerol could be interacting with the surface of CeO2 , as observed by Sato et al. for diols [19]. When the proton abstraction is in the terminal hydroxyl group from glycerol, acrolein is obtained. Then, Ce4+ can be reduced by the eliminated hydrogen radical in the catalytic cycle and both of the OH groups in glycerol could be interacted with the surface of CeO2 –ZrO2 phase. The Co3 O4 acted by increasing the basicity of the solid (TPD-CO2 measurements), being thereafter transformed. Thus, redox cycle of Ce4+ and Ce3+ on the surface would play an important role in dehydration of glycerol by activating the of triol, and acid–base features of the solid governs the product selectivity.

[1] F. Cavani, S. Guidettia, L. Marinellia, M. Piccinini, E. Ghedini, M. Signoretto, Appl. Catal. B: Environ. 100 (2010) 197–204. [2] A.S. de Oliveira, S.J.S. Vasconcelos, J.R. de Sousa, F.F. de Sousa, J.M. Filho, A.C. Oliveira, Chem. Eng. J. 168 (2011) 765–774. [3] A. Alhanash, E.F. Kozhevnikova, I.V. Kozhevnikov, Appl. Catal. A: Gen. 378 (2010) 11–18. [4] L.C. Meher, R. Gopinath, S.N. Naik, A.K. Dalai, Ind. Chem. Eng. Res. 48 (2010) 1080–1846. [5] A. Iriondo, J.F. Cambra, V.L. Barrio, M.B. Guemea, P.L. Arias, M.C. SanchezSanchez, R.M. Navarro, J.L.G. Fierro, Appl. Catal. B: Environ. 103 (2011) 83–93. [6] Y. Nakagawa, Y. Shinmi, S. Koso, K. Tomishige, J. Catal. 272 (2010) 191–194. [7] A.K. Kinage, P.P. Upare, P. Kasinathan, Y.K. Hwang, J.-S Chang, Catal. Commun. 11 (2010) 620–623. [8] L-Z. Tao, S.-H. Chai, Y. Zuo, W.-T. Zheng, Y. Liang, B.-Q. Xu, Catal. Today 158 (2010) 310–316. [9] C.-J. Jia, Y. Liu, W. Schmidt, A.-H. Lu, F. Schüth, J. Catal. 269 (2010) 71–79. [10] A. Villa, A. Gaiassi, I. Rossetti, C.L. Bianchi, K. van Benthem, G.M. Veith, L. Prati, J. Catal. 275 (2010) 108–166. [11] C.L. Lima, H.S.A. de Sousa, S.J.S. Vasconcelos, J.M. Filho, F.F. de Sousa, A.C. Oliveira, React. Kinet. Mec. Catal. 102 (2011) 487–500. [12] W. Suprun, M. Lutecki, R. Gläser, H. Papp, J. Mol. Catal. A:Chem. 342-343 (2011) 91–100. [13] S.J.S. Vasconcelos, C.L. Lima, J.M. Filho, A.C. Oliveira, E.B. Barros, F.F. de Sousa, M.G.C. Rocha, P. Bargiela, A.C. Oliveira, Chem. Eng. J. 168 (2011) 656–664. [14] A.H.M. Batista, F.F. de Sousa, S.B. Honorato, A.P. Ayala, J.M. Filho, F.W. de Sousa, A.N. Pinheiro, J.C.S. de Araujo, R.F. Nascimento, A. Valentini, A.C. Oliveira, J. Mol. Catal. A: Chem. 315 (2010) 86–98. [15] M. Abecassis-Wolfovich, M.V. Landau, A. Brenner, M. Herskowitz, J. Catal. 247 (2007) 201–213. [16] N.R. Shiju, V.V. Guliants, Appl. Catal. A: Gen. 356 (2009) 1–17. [17] A.H.M. Batista, F.S.O. Ramos, T.P. Braga, C.L. Lima, F.F. de Sousa, A.S. de Oliveira, E.B. Barros, J.R. de Souza, A. Valentini, J.M. Filho, A.C. Oliveira, Appl. Catal. A: Gen. 382 (2010) 148–157. [18] A. Iriondo, V.L. Barrio, J.F. Cambra, P.L. Arias, M.B. Guemez, M.C. SanchezSanchez, R.M. Navarro, J.L.G. Fierro, Int. J. Hydrogen Energy 35 (2010) 11622–11633. [19] S. Sato, R. Takahashi, T. Sodesawa, N. Honda, J. Mol. Catal. A: Chem. 221 (2004) 177–183. [20] D. Srinivas, C.V.V. Satyanarayan, H.S. Potdar, P. Ratnasamy, Appl. Catal. A: Gen. 246 (2003) 323–334. [21] J.A. Montoya, E. Romero-Pascual, C. Gimon, P. Del Angel, A. Monzón, Catal. Today 63 (2000) 71–85. [22] A. Kambolis, H. Matralis, A. Trovarelli, Ch. Papadopoulou, Appl. Catal. A: Gen. 377 (2010) 16–25. [23] Istadi, N.A.S. Amin, J. Mol. Catal. A: Chem. 259 (2006) 61–66. [24] J.Z. Shyu, W.H. Weber, H.S. Gandhi, J. Phys. Chem. 92 (1988) 4964–4970.


H.S.A. de Sousa et al. / Applied Catalysis A: General 406 (2011) 63–72

[25] Z. Wei, H. Li, X. Zhang, S.i. Yan, Z. Lv, Y. Chen, M. Gong, J. Alloy Compd. 455 (2008) 322–326. [26] M.J. Escudero, L. Mendoza, M. Cassir, T. Gonzalez, L. Dazaa, J. Power Source 160 (2006) 775–781. [27] G. Aguila, F. Gracia, P. Araya, Appl. Catal. A: Gen. 343 (2008) 16–24. [28] S. Boullosa-Eiras, E. Vanhaecke, T. Zhao, D. Chen, A. Holmen, Catal. Today, (2011) doi:10.1007/s11144-010-0279-7, in press. [29] A.L. Pinheiro, A.N. Pinheiro, A. Valentini, J.M. Filho, F.F. Sousa, J.R. Sousa, M.G.C. Rocha, P. Bargiela, A.C. Oliveira, Catal. Commun. 11 (2009) 11–14. [30] R.M. Freire, F.F. de Sousa, A.L. Pinheiro, E. Longhinotti, J.M. Filho, A.C. Oliveira, P.T.C. Freire, A.P. Ayala, A.C. Oliveira, Appl. Catal. A: Gen. 359 (2009) 165–179. [31] M. Akiyama, S. Sato, R. Takahashi, K. Inui, M. Yokota, Appl. Catal. A: Gen. 371 (2009) 60–66. [32] W. Yu, J. Zhao, H. Ma, H. Miao, Q. Song, J. Xu, Appl. Catal. A: Gen. 383 (2010) 73–78. [33] C.L. Lima, S.J.S. Vasconcelos, H.S.A. de Sousa, J.M. Filho, B.C.V. Viana, M.G.C. Rocha, P. Bargiela, A.C. Oliveira, Appl. Catal. A: Gen. 395 (2011) 53–63. [34] H. Vidal, J. Kaˇspar, M. Pijolat, G. Colonb, S. Bernal, A. Cordón, V. Perrichon, F. Fally, Appl. Catal. B: Environ. 27 (2000) 49–63. [35] B.H. Ryu, S.P. Yoona, J. Hana, S.W. Nam, T.-H. Lim, S.-A. Hong, K.B. Kim, Electrochim. Acta 50 (2004) 189–198.

[36] M. Epifani, J. Arbiol, E. Pellicer, N. Sergentd, T. Pagnierd, J.R. Morantec, Mater. Chem. Phys. 124 (2010) 809–815. [37] G. Satishkumar, L. Titelman, M.V. Landau, J. Solid State Chem. 182 (2009) 2822–2828. [38] L. Xue, C. Zhang, H. He, Y. Teraoka, Appl. Catal. B: Environ. 75 (2007) 167–174. [39] Z.M. Tian, S.L. Yuan, J.H. He, P. Li, S.Q. Zhang, C.H. Wang, Y.Q. Wang, S.Y. Yin, L. Liu, J. Alloy Compd. 466 (2008) 26–30. [40] N. Wang, W. Chua, T. Zhang, X.-S. Zhao, Chem. Eng. J. (2011) doi:10.1016/j.cej.2010.12.042, in press. [41] B.M. Reddy, K.N. Rao, G.K. Reddy, A. Khan, S-E. Park, J. Phys. Chem. C 111 (2007) 18751–18758. [42] Y. Wang, A. Zhu, Y. Zhang, C.T. Au, X. Yang, C. Shi, Appl. Catal. B: Environ. 81 (2008) 141. [43] E. Manova, T. Tsoncheva, C. Estournès, D. Paneva, K. Tenchev, I. Mitov, L. Petrov, Appl. Catal. A: Gen. 300 (2006) 170–180. [44] S.-Y. Lin, D.H. Kim, M.H. Engelhard, S.Y. Ha, J. Catal. 273 (2010) 229–235. [45] S. Erfle, U. Armbruster, U. Bentrup, A. Martin, A. Brückner, Appl. Catal. A: Gen. 391 (2011) 102–109. [46] B. Murugan, A.V. Ramaswamy, J. Am. Chem. Soc. 129 (2007) 3062–3063.