TiO₂-supported heteropoly acid catalysts for dehydration of methanol to dimethyl ether: relevance of dispersion and support interaction

Abstract

Two heteropoly acids (HPAs) with Keggin structures, namely, H₃PW₁₂O₄₀ (HPW) and H₄SiW₁₂O₄₀ (HSiW), have been deposited on TiO₂ and used as catalysts for the production of dimethyl ether (DME) from methanol. The catalysts have been prepared by incipient wetness impregnation of HPW and HSiW on TiO₂ with HPA loadings ranging between 0.9 and 9.0 Keggin units (KU) per square nanometer. The structure and acid properties of the final catalysts have been thoroughly characterized by N₂ adsorption–desorption isotherms, TGA, XRD, Raman spectroscopy, XPS, NH₃ adsorption isotherms, DRIFT and ¹H NMR. All catalysts exhibit very high DME productivities and high methanol conversion rates at temperatures as low as 413 K. The effect of the HPA loading on TiO₂ for the production of DME has been correlated with the structure and acid properties of the final catalyst. We find an optimum loading for both TiO₂-supported HPW and HSiW of 2.3 KU nm⁻². At this level, both HPW and HSiW are well dispersed onto the support, thus permitting the access of methanol to the active acid sites. On the contrary, higher HPA loadings on TiO₂ result in the formation of larger HPA units that prevent the access of methanol to the inner acid sites within the HPA. On the other hand, HPA loadings below 2.3 KU nm⁻² result in a strong interaction between the acid protons of the HPAs and the support, hence preventing the participation of such protons in the methanol dehydration reaction, which results in lower methanol conversion rates.

---

Introduction

Methanol is amongst the top 10 commodity chemicals produced globally, having applications within different sectors, including chemical intermediates and fuels. Methanol can be used as fuel, either directly or blended with other petroleum products. It can also be used to produce fuels via the formation of olefins by means of the so-called methanol-to-olefin and methanol-to-gasoline processes. It is also a raw material for different chemicals, including formaldehyde, dimethyl ether (DME), methyl tert-butyl ether (MTBE) and acetic acid.¹ Amongst these, DME is gaining a great deal of attention. Currently, DME is mainly used as an aerosol propellant or, due to its similar properties to liquefied petroleum gas (LPG), as a domestic fuel blended with LPG, particularly in Asia.^2–4 However, due to the good ignition, burning properties and high cetane number of DME, the possibility of its use as fuel in diesel engines instead of the conventional diesel has triggered the interest in DME for the transportation sector. In fact, DME combustion leads to low emission levels of nitrous oxides and particulate matter compared to diesel combustion.⁵ Moreover, the possibility of obtaining DME from lignocellulosic biomass via thermochemical treatments such as pyrolysis and/or gasification, and as a consequence its qualification as a second-generation biofuel, has triggered governmental interest in the production and use of DME from biomass-derived syngas.⁶

The main route for the synthesis of DME is the dehydration of methanol (2CH₃OH ↔ CH₃OCH₃ + H₂O), which is a gas-phase exothermic reaction, conducted typically at 1.0–2.0 MPa, 493–523 K, using solid acid catalysts such as γ-Al₂O₃.^4,7 Recently, the direct synthesis of DME from syngas (3CO + 3H₂ ↔ CH₃OCH₃ + CO₂ and 4H₂ + 2CO ↔ CH₃OCH₃ + H₂O) has been proposed. This process is carried out at 4.0–5.0 MPa, 473–623 K, using hybrid catalysts within a single reactor combining two functionalities: a Cu/ZnO/Al₂O₃-based catalyst for the synthesis of methanol and an acid solid catalyst for methanol dehydration.⁸ Also, the use of CO₂ as a raw material for DME production has been proposed recently.⁹

The most typical acid catalysts for the dehydration of methanol to DME include γ-Al₂O₃, zeolites and mixed/supported oxides.^10–18 These catalysts show high methanol conversion per pass; however, due to the high operation temperatures, the formation of undesired by-products such as light hydrocarbons and coke is favoured resulting in a decrease of the productivity of DME along with a higher rate of catalyst deactivation. Recently, it has been reported that heteropoly acids (HPAs) exhibit very high catalytic activities for the dehydration of alcohols.^19–25 In fact, some HPAs exhibit higher reaction rates than conventional solid acid catalysts.^11,20,24 In a recent study,²³ we reported that TiO₂-supported tungstosilicic acid (H₄SiW₁₂O₄₀) is a very active catalyst for methanol dehydration even at temperatures as low as 413 K, reaching conversions of ca. 80% at 453 K. In fact, γ-Al₂O₃ requires much higher temperatures, above 523 K, to reach the same methanol conversion value.

HPAs are a class of solid acids constituting polyoxometalate (POM) units, i.e., metal–oxygen cluster anions, which may exhibit different structures, with the most typical ones being the Keggin and Dawson structures. HPAs exhibiting the Keggin structure are the most stable ones and have found industrial application as acid catalysts.²⁶ They are formed by a central tetrahedron (XO₄) surrounded by four octahedral (MO₆) units. This anion is stabilized by protons, which are responsible for the high Brønsted acidity of these materials.^27–29 The most typical HPAs used for alcohol dehydration reactions are the phosphotungstic (H₃PW₁₂O₄₀) and the tungstosilicic acids (H₄SiW₁₂O₄₀).

One of the main drawbacks of the HPAs when it comes to their behaviour as heterogeneous catalysts is their very low surface area, typically below 10 m² g⁻¹. One strategy to overcome this issue is to deposit/disperse HPAs onto high surface area solids such as SiO₂, TiO₂, ZrO₂ or zeolites.^{20,24,30–32} However, their structure and, as a consequence, the acid properties of the HPAs may be altered upon the deposition process. For instance, decreasing acid strength of H₃PW₁₂O₄₀ and Cs_1.5H_0.5PW₁₂O₄₀ after supporting has been reported.^33,34 It should be noted, however, that the catalytic performance of supported HPAs is associated not only with their acid strength. A recent study indicates that despite the lower acid strength of TiO₂-supported HPW or HSiW, these catalysts are more active and resistant than the corresponding bulk HPAs for the dehydration of methanol to DME.²³ This is because the decrease in the acid strength weakens the interaction between H₂O molecules and the acid sites of the HPA, hence facilitating the access of methanol to the active acid sites and leading to higher turnover frequencies (TOF). This proposal also explains the higher catalytic activity of supported HPAs as compared with bulk HPAs observed for methanol and ethanol dehydration reaction reported by other groups.^20,30

The aim of this work is to study the effect of the HPA loading for the methanol dehydration reaction with TiO₂-supported H₃PW₁₂O₄₀ (HPW) or H₄SiW₁₂O₄₀ (HSiW). The catalysts have been thoroughly characterized in order to understand the relationship between the catalytic activity for the production of DME and the evolution of the structure and acid properties of the catalysts with the HPA loading.

Experimental

Preparation of HPA/TiO₂ catalysts

Supported heteropoly acids were prepared by incipient wetness impregnation. The appropriate amounts of phosphotungstic acid (H₃PW₁₂O₄₀·xH₂O, Sigma-Aldrich, reagent grade) and tungstosilicic acid (H₄SiW₁₂O₄₀·xH₂O, Sigma-Aldrich, ≥99.9%) were dissolved in ethanol and added dropwise to TiO₂ powder (Degussa P25, 20% rutile, 80% anatase, 53 m² g⁻¹). After impregnation, the samples were held at room temperature for 24 h to facilitate the uniform distribution of the HPA clusters over TiO₂, and subsequently, they were dried in static air at 333 K for 24 h. The heteropoly acid surface density of the prepared catalysts ranges between 0.9 and 9.0 Keggin units (or polyoxometalates) per square nanometre of the support (KU nm⁻²). Taking into account that the size of the Keggin unit ranges between 0.9 and 1.5 nm (ref. 30 and references therein) an HPA loading between 1.23 and 0.5 KU nm⁻² would, in principle, result in a monolayer coverage of HPA on TiO₂. The catalysts are named xHPW/TiO₂ and xHSiW/TiO₂, where x denotes the surface density of the HPA (KU nm⁻²). Bulk phosphotungstic acid (HPW) and tungstosilicic acid (HSiW) were dried at 333 K and used as reference materials.

Characterization techniques

The Brunauer–Emmett–Teller (BET) specific surface areas were determined from N₂ adsorption–desorption isotherms at 77 K collected using a Micromeritics ASAP 2010. Before recording the isotherms, the samples were degassed at 413 K for 16 h.

Thermogravimetric analysis (TGA) was carried out using a TGA/SDTA851 (Mettler Toledo) instrument, with a heating rate of 10 K min⁻¹, from room temperature up to 1250 K, under synthetic air flow (20 vol.% O₂ and 80 vol.% N₂).

The X-ray diffraction (XRD) patterns (2θ = 5°–90°) were recorded at room temperature with a Seiffert 3000 Xpert X-ray diffractometer using Cu Kα radiation. The phase identification was made by comparison with patterns from the International Centre for Diffraction Data (ICDD) database.

A Renishaw inVia Raman microscope spectrometer equipped with a laser beam emitting at 532 nm and 100 mW output power was used to collect the Raman spectra. The photons scattered by the sample were dispersed with a grating monochromator at 1800 lines per millimeter and simultaneously collected using a CCD camera; the collection optics was set at 50× objective. The spectral resolution is 1 cm⁻¹.

The X-ray photoelectron spectra (XPS) of the samples were obtained with a VG Escalab 200R spectrometer equipped with a hemispherical electron analyzer using a Mg Kα (hν = 1253.6 eV) X-ray source. The XPS data were signal averaged for at least 200 scans and were taken at intervals of 0.1 eV. The charging effects were corrected by setting the Ti 2p_3/2 peak as internal reference at 459.0 eV. High-resolution spectral envelopes were obtained with curve-fitting peak components using the XPS peak software. The raw data were used with no preliminary smoothing. The curve fitting was performed using Newton's iteration methods and 200 iterations. Symmetric Gaussian–Lorentzian product functions were used to approximate the line shapes of the fitting components. Chi-squared minimum values were used to mathematically ensure and obtain the optimized fitting parameters.

The amount of ammonia adsorbed by the samples was determined from NH₃ adsorption isotherms which were recorded at 373 K with ASAP 2010/2000 Micromeritics equipment. Before NH₃ adsorption, the samples were treated at 473 K under vacuum for 12 h. The irreversible ammonia uptake was calculated as a difference between the total uptake of ammonia, detected in the first isotherm, and the amount of ammonia released upon the subsequent desorption stage, which was determined by a second isotherm

DRIFT spectra of adsorbed ammonia were obtained with a FTIR-6300 JASCO Fourier transform spectrophotometer equipped with a Harrick diffuse reflectance accessory (HVC-DRP cell). Before the ammonia adsorption, the catalysts were treated at 493 K under He flow (20 mL min⁻¹) for 30 min. Subsequently, the NH₃ was adsorbed on the catalysts by flowing 5 vol.% NH₃ in He at 20 mL min⁻¹ for 15 min. Then, the DRIFT cell was purged under He flow (20 mL min⁻¹) for 30 min, and the adsorbed NH₃ spectra were recorded at increasing temperature (from 373 to 773 K, 15 K min⁻¹).

Solid-state NMR measurements were carried out with a Bruker AV-400-WB spectrometer at a ¹H NMR frequency of 400.13 MHz, using a 2.5 mm double-resonance MAS probe. Spinning speed was set at 25 kHz. Single-pulse experiments used a 3 μs π/2 pulse, a spectral width of 35 kHz and a relaxation delay of 2 s. All measurements were taken at room temperature using H₂O as secondary external reference at 4.77 ppm relative to TMS as primary external reference. All samples were kept at 373 K before the NMR experiments and rapidly transferred to the analysis chamber to avoid further hydration.

Methanol dehydration reactions

Conversion of methanol to DME was measured in a fixed-bed reactor. The catalytic bed was prepared by diluting 0.2 g of the catalyst (0.25–0.30 mm) with 2.0 g of SiC (0.25–0.30 mm) to avoid hot spots. The catalysts were pretreated at 493 K under N₂ flow for 1 h. The reaction was running at atmospheric pressure with N₂ as carrier gas (16.7 vol.% methanol; total space velocity, 14 400 cm³ h⁻¹ g_cat⁻¹) at 413, 433 and 453 K (measured directly in the catalytic bed). Reactants and products were analysed online with a gas chromatograph (Varian CP-3800) equipped with a compacted column (Carboxen 1000, 4.6 m × 0.32 cm) connected to a TCD detector, where inorganic gases (H₂, N₂, CO and CO₂) were measured, and with a capillary column (SPB-5, 60 m × 0.53 mm) connected to a flame ionization detector (FID) to analyze organic compounds (CH₃OH, CH₃OCH₃, and light hydrocarbons). Methanol conversion (X_CH3OH) and product selectivity (S_i) were calculated as follows:

Results and discussion

Table 1 shows the composition and BET surface areas of xHPW/TiO₂ and xHSiW/TiO₂, respectively. As expected, the specific surface area of the HPAs/TiO₂ decreases with the increasing loading of HPA. However, even for the catalyst containing the highest HPA loading in the series (x = 9.0 KU nm⁻²), the surface area value is approximately four times higher than that of the bulk HPA.

Table 1 Physicochemical properties and DME production of the xHPW/TiO₂ and xHSiW/TiO₂ catalysts. Data for bulk HPW and HSiW are included for comparison

Catalysts	Nominal HPA content (wt.%)	Surface area (m^2 g^−1)	NH3 chemisorbed (molNH3 molHPA^−1)	DME production^a (millimoles DME per acid site per hour)
a The number of acid sites was calculated from NH3 isotherms. Reaction conditions: 453 K, 1 bar, 16.7 vol.% methanol, and 14 400 cm^3 h^−1 gcat^−1.
0.9HPW/TiO2	19	56	4.7	27.5
2.3HPW/TiO2	37	48	3.4	39.4
4.5HPW/TiO2	53	40	3.5	27.7
9.0HPW/TiO2	70	31	2.9	20.9
HPW	100	8	2.9	7.3
0.9HSiW/TiO2	19	59	5.0	32.1
2.3HSiW/TiO2	37	49	4.0	39.6
4.5HSiW/TiO2	53	44	3.8	30.5
9.0HSiW/TiO2	70	37	3.8	21.9
HSiW	100	8	3.5	10.7

---

The XRD patterns for xHPW/TiO₂ and xHSiW/TiO₂ along with those of the corresponding bulk HPAs are shown in Fig. 1a and b, respectively. All diffractograms of xHPA/TiO₂ show a set of reflections at 2θ = 25.3°, 36.9°, 37.8° and 38.5° and at 2θ = 27.4° and 36.1° attributed to the rutile (JCPDS# 00-001-1292) and anatase (JCPDS# 00-001-0562) phases of TiO₂, respectively. In fact, these reflections are the only ones observed in the diffractograms for the catalysts with the lowest HPA loadings, 0.9HPW/TiO₂ and 0.9HSiW/TiO₂. The lack of diffraction lines characteristic of the HPAs in these diffractograms is indicative of the high dispersion of the HPA on the support. This result indicates that monolayer coverage is not reached for the catalysts with HPA with loadings of up to 0.9 KU nm². For the catalysts with higher HPA loadings, reflections attributed to H₃PW₁₂O₄₀·6H₂O^35–37 and H₄SiW₁₂O₄₀·xH₂O (x > 6)^23,38 are observed. Note that the hydration level of the HSiW-based catalysts is higher than that of the HPW-based ones irrespective of the actual HPA loading. These results confirm that the Keggin structure is preserved upon the impregnation of the HPAs onto TiO₂, at least when the HPA loading is ≥2.3 KU nm⁻².

Fig. 1 X-ray diffraction patterns of HPW and xHPW/TiO₂ (a) and HSiW and xHSiW/TiO₂ (b). Reflections of TiO₂ (*), H₃PW₁₂O₄₀·6H₂O (1) and highly hydrated H₄SiW₁₂O₄₀·xH₂O (2) are indicated.

Raman spectroscopy was used to elucidate the molecular structure of the supported and bulk HPAs. The Raman spectra shown in Fig. 2a and b were recorded under ambient conditions. The Raman bands of the dehydrated heteropoly acids tend to become less intense and broader than those recorded for hydrated HPAs.²³ As a consequence, the assignment of the Raman bands in the spectra of the dehydrated supported catalysts containing low amounts of HPA is not straightforward. The Raman spectrum of the bulk HPW shows two intense bands at 1007 and 991 cm⁻¹, which are attributed to the symmetric and asymmetric stretching modes of terminal W=O_t. Two less intense bands at 982 and 902 cm⁻¹ are also observed and ascribed to the stretching vibration of P–O_a and to the bridging vibration of W–O_b–W bonds, respectively.^39–42 Similarly, bulk HSiW exhibits two intense Raman bands at 999 and 975 cm⁻¹ assigned to the symmetric and asymmetric ν(W=O_t) and two less intense bands at 925 and 883 cm⁻¹ due to the ν(Si–O_a) and ν(W–O_b–W) modes.⁴² The spectra of the supported catalysts show the same bands as described above for the corresponding bulk HPAs, although their intensity decreases visibly with the decreasing HPA loading. The presence of the same Raman bands in the spectra for the xHPA/TiO₂ and for the bulk HPAs indicates that the Keggin structure of the HPA remains stable upon impregnation on TiO₂, even for the catalyst with the lowest HPA loading of 0.9 KU nm⁻². The preservation of the structure of the HPAs is further supported by the absence of bands of the individual oxides, i.e., from the lack of bands for WO₃, which appear upon the decomposition of the Keggin structure at ca. 796 cm⁻¹.^14,23

Fig. 2 Raman spectra of HPW and xHPW/TiO₂ (a) and HSiW and xHSiW/TiO₂ (b).

The surface composition of the catalysts was determined by XPS. The W 4f core-level region of the spectra of the bulk and supported HPW and HSiW catalysts is shown in the ESI† (Fig. S1). All catalysts show a W 4f doublet at 35.2–35.8 eV, in good agreement with previous results for bulk and supported HPAs.^42–44 The position of the W 4f peak does not change significantly with the HPA loading. Remarkably, the bulk HPAs show an additional W 4f doublet at 37.2 and 38.2 eV for HPW and HSiW, respectively, which are attributed to the presence of H₂O around the Keggin units.^42,43 The contribution of the W 4f doublet ascribed to the hydrated HPA is more significant for HSiW than for HPW, suggesting, in line with the XRD results discussed above, a higher hydration level of the HSiW-based HPAs. A further peak at 37.4 eV assigned to the Ti 3p level of Ti⁴⁺ (TiO₂) is also observed for the supported samples. The evolution of the surface atomic W/Ti ratios calculated from the area of the XPS peaks, which was previously normalized to their corresponding atomic sensitivity factors, and the nominal W/Ti atomic ratios as a function of HPA loading are shown in Fig. 3. As observed, the surface atomic W/Ti ratio increases with increasing HPA loading, reaching a plateau at HPA loadings ≥2.3 KU nm⁻² for the HPW catalysts and at loadings ≥4.5 KU nm⁻² for the HSiW samples. These results suggest that HPAs remain well dispersed on TiO₂ for HPW loadings ≤2.3 KU nm⁻² and HSiW loadings ≤4.5 KU nm⁻², forming 3-dimensional clusters in the catalysts with higher HPA loadings. This observation agrees well with the lack of diffraction peaks of the HPAs in the samples with low HPA loadings.

Fig. 3 Variation of surface and nominal W/Ti atomic ratios as a function of the HPA loading for xHPW/TiO₂ and xHSiW/TiO₂ catalysts.

The thermal stability of the xHPA/TiO₂ catalysts was studied by thermogravimetric analyses recorded under air flow from room temperature to 1250 K (ESI,† Fig. S2 and S3). All catalysts show a weight loss at temperatures below 373 K, which is attributed to the removal of physisorbed water. This process leads to the formation of the relatively stable hexahydrated structures, H₃PW₁₂O₄₀·6H₂O and H₄SiW₁₂O₄₀·6H₂O, in both the bulk and the supported HPAs.^23,45,46 In such hexahydrated structures, the water of crystallization is located between the Keggin units, forming hydrogen bonds with the acid protons in the form of [H₂O⋯H⁺⋯OH₂] ions.^45,47 For loadings higher than 2.3 KU nm⁻², a second weight loss is observed between 432 and 470 K, which corresponds to the formation of anhydrous H₃PW₁₂O₄₀ and H₄SiW₁₂O₄₀ species upon removal of the water of crystallization molecules.²³ This weight loss is not observed for the catalysts with HPA loading ≤0.9 KU nm⁻², probably due to the detection limit of the thermobalance equipment. At higher temperatures (T >700 K), a small weight loss is observed for the bulk and highly loaded catalysts. This process corresponds to the loss of constitutional water, and consequently, it leads to the complete deprotonation of the HPAs to form PW₁₂O_38.5 and SiW₁₂O₃₈.^45,46 The complete decomposition of the HPA takes place at higher temperatures without a recordable weight loss process.²³ The thermal analyses indicate that the Keggin structure in both bulk and supported HPAs is stable at temperatures below 723 K.

The acid properties of the HPA-based catalysts have been assessed by using NH₃ as a probe molecule. NH₃ was selected as a probe molecule because it allows for the evaluation of both the surface and the bulk acidity of the HPAs.⁴⁸ The density of acid sites has been determined by recording NH₃ adsorption isotherms at 373 K. Previously, the catalysts were thermally treated at 473 K under vacuum overnight. Then, the first NH₃ adsorption isotherm, which includes both chemisorbed and physisorbed NH₃, was registered. Afterwards, the sample was evacuated at 373 K under vacuum for 4 h in order to remove physisorbed NH₃. Subsequently, a second adsorption isotherm was taken under the same conditions. The amount of NH₃ adsorbed in the second isotherm corresponds to physisorbed NH₃ species. Therefore, the amount of chemisorbed NH₃ on the catalyst surface can be calculated from the difference between the first and the second NH₃ isotherms. The amount of chemisorbed NH₃ can be considered as a measure of the number of acid sites on the catalyst, inclusive of the bulk and surface acid sites.⁴⁹ The amount of chemisorbed NH₃ for all catalysts is shown in Fig. 4. Clearly, depositing HPW or HSiW on TiO₂ results in a decrease in the total number of acid sites per gram of the catalyst. Remarkably, the observed decrease is linear with the amount of HPA (expressed as weight percentage) in the catalysts. The amounts of chemisorbed NH₃ per Keggin unit for HPW- and HSiW-based catalysts are shown in Table 1. The amounts of ammonia chemisorbed by the bulk HPW and HSiW per Keggin unit are 2.9 and 3.5 mol_NH3 mol_HPA⁻¹, respectively.

Fig. 4 Chemisorbed NH₃ determined from the adsorption isotherms recorded at 373 K with HPW- and HSiW-based catalysts.

These values are slightly lower than the expected stoichiometric ones (3 and 4 mol_NH3 mol_HPA⁻¹, respectively). It has been reported²² that the adsorption of ammonia in bulk HPAs is limited to the protons located at the surface layers because of their low specific surface area. On the other hand, the TiO₂-supported HPAs show higher NH₃ uptakes per mole of HPA than that observed for the unsupported HPAs, suggesting a higher accessibility of the NH₃ molecules to the protons in the supported HPAs.²³ The number of acid sites per mole of HPA increases with the decrease in the HPA loading in the catalyst, reaching the highest value in the series for the catalysts with the lowest HPA loading, 0.9 KU nm⁻². This observation in fact suggests that the accessibility of the NH₃ molecules to the protons of the HPA increases with the increasing dispersion of the HPA onto the support. The catalysts containing HPW loadings ≤4.5 KU nm⁻² and HSiW loadings ≤2.3 KU nm⁻² exceed the stoichiometric value of chemisorbed NH₃. This trend has been reported previously and ascribed to the formation of [N₂H₇]⁺ adducts when the NH₃ molecules interact via hydrogen bonds with already existing NH₄⁺ cations.²²

The nature of the acid sites in the bulk and supported HPAs has been further investigated by DRIFT spectroscopy using NH₃ as a probe molecule. The spectra of adsorbed NH₃ at 373 K in the region between 3500 and 500 cm⁻¹ for HPW- and HSiW-based catalysts are shown in Fig. 5. The spectrum of NH₃ adsorbed on TiO₂ is also shown for comparison.

Fig. 5 DRIFT spectra of chemisorbed NH₃ at 373 K on TiO₂ and HPW-based (a) and HSiW-based (b) catalysts.

As observed, the spectrum of TiO₂ exhibits intense structural bands between 1200 and 700 cm⁻¹, which overlap with the fingerprint region of the Keggin structure.⁵⁰ As a consequence, the identification of the vibrations associated with the Keggin anions is not straightforward for the catalysts with HPA loadings <4.5 KU nm⁻². All HPA catalysts show a set of bands attributed to the symmetric and asymmetric deformation modes of protonated ammonia on Brønsted acid sites, δ_s(NH₄⁺) and δ_as(NH₄⁺), at 1425 and 1682–1664 cm⁻¹, respectively.^51–54 These bands, due to protonated ammonia, are not observed in the spectrum of TiO₂, suggesting that the Brønsted acid sites are exclusively formed on the catalysts containing heteropoly acids. On the other hand, the spectra of the supported catalysts show bands at 1235–1243 cm⁻¹ and 1600–1602 cm⁻¹, which are ascribed to the symmetric and asymmetric deformation modes of coordinatively adsorbed NH₃ on Lewis acid sites, δ_s(NH₃) and δ_as(NH₃), respectively.^51–53 Lewis acid sites have been observed previously in TiO₂-supported HPW catalysts and assigned to coordinately unsaturated Ti⁴⁺ species,⁵⁵ which agrees with the absence of these bands in the bulk HPW and HSiW spectra. The δ_s(NH₃) band appears at 1151 cm⁻¹ in the spectrum of TiO₂. The shifting of the δ_s(NH₃) band to higher wavenumbers in the spectra for HPA/TiO₂ may be indicative of an increase of the acid strength of the Lewis acid sites on the supported catalysts.⁵²

Solid ¹H NMR has been used widely to evaluate the nature of the protons in acid solids such as zeolites.⁵⁶ However, the high mobility of the protons in the Keggin structure prevents establishing unequivocal correlations between the chemical shift of the protons in HPA and their acid strength.⁵⁷ The ¹H NMR spectra of bulk and supported HPAs are shown in Fig. 6. For loadings above 2.3 KU nm⁻², a strong peak at ca. 9.0 ppm is observed and assigned to protons in anhydrous heteropoly acids, which are usually referred to as “isolated acidic protons”.^42,58 An additional signal can be observed at 7.8 and 6.5 ppm for 2.3HPA/TiO₂ and 0.9HPA/TiO₂, respectively. It has been reported that protons in SiO₂-supported HPW with low HPW loadings are likely to be located on the neighbouring oxygen atoms of the support which are more basic than the outer oxygen atoms of the Keggin anion.^34,59,60 The interaction between the HPA protons and the oxygen atoms of the support could be responsible for the shifting to lower chemical shifts of the ¹H NMR peaks in the spectra of 0.9HPW/TiO₂ and 0.9HSiW/TiO₂, indicating that the interaction between HPA and the support is higher in the catalysts with low HPA loadings.

Fig. 6 ¹H NMR spectra of bulk and supported HPW and HSiW.

The catalytic performance for the dehydration of methanol to dimethyl ether with all HPW- and HSiW-based catalysts has been tested in a fixed-bed tubular reactor; the conversion values and the methanol conversion rates recorded at different temperatures for HPW- and HSiW-based catalysts, respectively, are shown in Fig. 7a and b. Both the bulk and the supported HPA catalysts under study are very active for the methanol conversion reaction, exhibiting measurable conversion rates even at temperatures as low as 413 K. On the other hand, TiO₂ is not active for the methanol dehydration under the studied reaction conditions. As expected, increasing the reaction temperature leads to an increase in the rate of methanol conversion with all catalysts. In all cases, dimethyl ether is the only product detected during the reaction irrespective of the reaction temperature. In line with previous reports, the xHPA/TiO₂ catalysts with x ≥2.3 KU nm⁻² record higher catalytic activities than bulk HPAs.^20,23 On the other hand, the catalytic performance of 0.9HPA/TiO₂ is similar to that recorded for the bulk HPAs. In our previous work,²³ we show that the higher activity of supported HPAs towards DME production from methanol arises from the higher accessibility of methanol to the active acid sites of the supported HPAs in comparison with the bulk HPAs. This behaviour is due to the weakening of the adsorption strength between water molecules and the acid sites of the Keggin units upon deposition on the support. This weakening results in a more facile replacement of water by methanol molecules, thus leading to higher methanol conversion rates.

Fig. 7 Methanol conversion vs. temperature for HPW-based catalysts (a) and HSiW-based catalysts (b) at 1 bar, 16.7 vol.% methanol, space velocity = 14 400 cm³ h⁻¹ g_cat⁻¹.

Fig. 7a and b show that the loading of the HPA does not affect the methanol conversion rate per gram of xHPA/TiO₂ when x ranges between 2.3 and 9.0 KU nm⁻². On the contrary, methanol conversion falls for the catalysts containing 0.9 KU nm⁻²; in fact, 0.9HSiW/TiO₂ exhibits methanol conversion values below those recorded with HSiW. This effect can be explained by taking into account the strong interaction between the acid protons in the 0.9HPA/TiO₂ series and the oxygen atoms of the support as observed by ¹H NMR (see above), which results in a decreasing number of acid sites available for the methanol dehydration reaction.

The production of dimethyl ether per acid site has been calculated by considering the amount of chemisorbed NH₃ as a measurement of the number of acid sites; the values are collected in Table 1. The DME production per acid site increases as the HPA loading increases, reaching a maximum for the catalyst containing 2.3 KU nm⁻² (note that this loading is close to the theoretical monolayer's coverage value). On the contrary, DME production decreases progressively with the higher HPA loadings. The existence of an optimum HPA loading, i.e., of a volcano plot for the methanol conversion rate vs. HPA loading, indicates that the HPA loading affects the total number of active sites in two opposite ways. On the one hand, the decreasing activity recorded for xHPA/TiO₂, where x ≥ 2.3 KU nm⁻², i.e., for the catalysts with HPA loadings slightly above the monolayer's coverage, is a result of the limited accessibility of methanol to the acid sites located within the bulk of the HPA. The fraction of such bulk acid sites increases with the decreasing dispersion of HPA on the support. This effect, i.e., the lower accessibility of bulk acid sites, has been in fact observed during the measurement of the NH₃ isotherms. On the other hand, despite the accessibility of methanol to the acid protons present in the xHPA/TiO₂, where x < 2.3 KU nm⁻² should be high, these acid sites are not accessible for the adsorption (and dehydration) of methanol probably due to the strong interaction between such protons and the oxygen atoms in the support. Therefore, the optimum HPA loading for TiO₂-supported HPA catalysts will be that at which the HPA is well dispersed on the support without the formation of 3-dimensional clusters while exhibiting a moderate interaction between the HPA and the support. For both HPW- and HSiW-based catalysts, this loading is at 2.3 KU nm⁻².

Conclusions

Supported HPW or HSiW on TiO₂ results in very active catalysts for the dehydration of methanol to DME. However, the catalytic performance of the TiO₂-supported HPA is strongly influenced by the HPA loading on TiO₂. It has been found that for HPA loadings ≤2.3 KU nm⁻², the HPA remains well dispersed on TiO₂, while for higher loadings, 3-dimensional clusters are formed. At low HPA loadings, an interaction between HPA and support has been observed by the appearance of stronger Lewis acid sites assigned to coordinately unsaturated Ti⁴⁺ species and by the shift in the ¹H-NMR spectra of the signal assigned to the protons in the HPA. On the other hand, the NH₃ adsorption isotherm experiments showed that the higher the HPA dispersion is, the more accessible the protons of the HPA are. Nevertheless, these protons are strongly interacting with the oxygens of the support, which prevent their participation for the methanol dehydration, as observed during the catalytic testing. It has been found that the optimum HPA loading for methanol dehydration reaction is 2.3 KU nm⁻² for the two studied heteropoly acids, HPW and HSiW.

Acknowledgements

R. Ladera gratefully acknowledges the Regional Government of Madrid (Spain) for a contract (CPI/0173/2008).

Notes and references

1. T. Mokrani and M. Scurell, Catal. Rev.: Sci. Eng., 2009, 51, 1–145.
2. M. Marchionna, R. Patrini, D. Sanfilippo and G. Migliavacca, Fuel Process. Technol., 2008, 89, 1255–1261.
3. D. A. Wood, C. Nwaoha and B. F. Towler, J. Nat. Gas Sci. Eng., 2012, 9, 196–208.
4. T. H. Fleisch, A. Basu and R. A. Sills, J. Nat. Gas Sci. Eng., 2012, 9, 94–107.
5. G. Thomas, B. Feng, A. Veeraragavan, M. J. Cleary and N. Drinnan, Fuel Process. Technol., 2014, 119, 286–304.
6. C. Panoutsou, A. Bauen and J. Duffield, Biofuels, Bioprod. Biorefin., 2013, 7, 685–701.
7. F. Raoof, M. Taghizadeh, A. Eliassi and F. Yaripour, Fuel, 2008, 87, 2967–2971.
8. R. Ahma, D. Schrempp, S. Behrens, J. Sauer, M. Doring and U. Arnold, Fuel Process. Technol., 2014, 121, 38–46.
9. G. Bonura, M. Cordaro, C. Cannilla, A. Mezzapica, L. Spadaro, F. Arena and F. Frusteri, Catal. Today, 2014, 228, 51–57.
10. Y. Tavan, M. R. Khosravi Nikou and A. Shariati, J. Ind. Eng. Chem., 2014, 20, 668–673.
11. S. S. Akarmazyan, P. Panagiotopoulou, A. Kambolis, C. Papadopoulou and D. I. Kondarides, Appl. Catal., B, 2014, 145, 136–148.
12. B. Sabour, M. H. Peyrovi, T. Hamoule and M. Rashidzadeh, J. Ind. Eng. Chem., 2014, 20, 222–227.
13. Y. Sang, H. Li, M. Zhu, K. Ma, Q. Jiao and Q. Wu, J. Porous Mater., 2013, 20, 1509–1518.
14. R. Ladera, E. Finocchio, S. Rojas, G. Busca, J. L. G. Fierro and M. Ojeda, Fuel, 2013, 113, 1–9.
15. R. Ladera, E. Finocchio, S. Rojas, J. L. G. Fierro and M. Ojeda, Catal. Today, 2012, 192, 136–143.
16. K. Lertjiamratn, P. Praserthdam, M. Arai and J. Panpranot, Appl. Catal., A, 2010, 378, 119–123.
17. V. Vishwanathan, H. S. Roh, J. W. Kim and K. W. Jun, Catal. Lett., 2004, 96, 23–28.
18. K. C. Tokay, T. Dogu and G. Dogu, Chem. Eng. J., 2012, 184, 278–285.
19. D. Varisli, T. Dogu and G. Dogu, Chem. Eng. Sci., 2007, 62, 5349–5352.
20. D. Varisli, C. K. Tokay, A. Ciftci, T. Dogu and G. Dogu, Turk. J. Chem., 2009, 33, 355–366.
21. R. T. Carr, M. Neurock and E. Iglesia, J. Catal., 2011, 278, 78–93.
22. L. Matachowski, A. Drelinkiewicz, E. Lalik, M. Ruggiero-Mikolajczyk, D. Mucha and J. Krysciak-Czerwenka, Appl. Catal., A, 2014, 469, 290–299.
23. R. M. Ladera, J. L. G. Fierro, M. Ojeda and S. Rojas, J. Catal., 2014, 312, 195–203.
24. D. Varisli, T. Dogu and G. Dogu, Ind. Eng. Chem. Res., 2008, 47, 4071–4076.
25. A. Ciftci, D. Varisli, C. K. Tokay, N. A. Sezgi and T. Dogu, Chem. Eng. J., 2012, 207–208, 85–93.
26. A. Bielański and A. Micek-Ilnicka, Inorg. Chim. Acta, 2010, 363, 4158–4162.
27. I. V. Kozhevnikov, Chem. Rev., 1998, 98, 171–198.
28. N. Mizuno and M. Misono, Chem. Rev., 1998, 98, 199–217.
29. I. V. Kozhevnikov, J. Mol. Catal. A: Chem., 2007, 262, 86–92.
30. A. Micek-Ilnicka, E. Bielanska, L. Litynska-Dobrzynska and A. Bielanski, Appl. Catal., A, 2012, 421–422, 91–98.
31. L. R. Pizzio, P. G. Vázquez, C. V. Cáceres and M. N. Blanco, Appl. Catal., A, 2003, 256, 125–139.
32. S. Zhu, Y. Zhu, X. Gao, T. Mo, Y. Zhu and Y. S. Li, Bioresour. Technol., 2013, 130, 45–51.
33. A. M. Alsalme, P. V. Wiper, Y. Z. Khimyak, E. F. Kozhevnikova and I. V. Kozhevnikov, J. Catal., 2010, 276, 181–189.
34. G. C. Bond, S. J. Frodsham, P. Jubb, E. F. Kozhevnikova and I. V. Kozhevnikov, J. Catal., 2012, 293, 158–164.
35. K. Okumura, K. Yamashita, K. Yamada and M. Niwa, J. Catal., 2007, 245, 75–83.
36. P. Madhusudhan Rao, A. Wolfson, S. Kababya, S. Vega and M. V. Landau, J. Catal., 2005, 232, 210–225.
37. L. Marosi, E. Escalona Platero, J. Cifre and C. Otero Areán, J. Mater. Chem., 2000, 10, 1949–1955.
38. F. J. Berry, G. R. Derrick and M. Mortimer, Polyhedron, 2014, 68, 17–22.
39. R. Thouvenot, M. Fournier, R. Franck and C. Rocchiccioli-Deltcheff, Inorg. Chem., 1984, 23, 598–605.
40. S. R. Matkovic, L. E. Briand and M. A. Bañares, Mater. Res. Bull., 2011, 46, 1946–1948.
41. L. Nakka, E. Molinari and I. E. Wachs, J. Am. Chem. Soc., 2009, 131, 15544–15554.
42. N. Legagneux, J. M. Basset, A. Thomas, F. Lefebvre, A. Goguet, J. Sa and C. Hardacre, Dalton Trans., 2009, 2235–2240.
43. P. A. Jalil, M. Faiz, N. Tabet, N. M. Hamdan and Z. Hussain, J. Catal., 2003, 217, 292–297.
44. R. Dziembaj, A. Malecka, Z. Piwowarska and A. Bielanski, J. Mol. Catal. A: Chem., 1996, 112, 423–430.
45. I. V. Kozhevnikov, J. Mol. Catal. A: Chem., 2009, 305, 104–111.
46. L. R. Pizzio and M. N. Blanco, Microporous Mesoporous Mater., 2007, 103, 40–47.
47. K. Y. Lee, N. Mizuno, T. Okuhara and M. Misono, Bull. Chem. Soc. Jpn., 1989, 62, 1731–1739.
48. Y. Kamiya, Y. Ooka, C. Obara, R. Ohnishi, T. Fujita, Y. Kurata, K. Tsuji, T. Nakajyo and T. Okuhara, J. Mol. Catal. A: Chem., 2007, 262, 77–85.
49. Y. Kamiya, T. Okuhara, M. Misono, A. Miyaji, K. Tsuji and T. Nakajyo, Catal. Surv. Asia, 2008, 12, 101–113.
50. E. Caliman, J. A. Dias, S. C. L. Dias and A. G. S. Prado, Catal. Today, 2005, 107–108, 816–825.
51. G. Busca, Catal. Today, 1998, 41, 191–206.
52. A. Gutiérrez-Alejandre, P. Castillo, J. Ramírez, G. Ramis and G. Busca, Appl. Catal., A, 2001, 216, 181–194.
53. M. Tamura, K. Shimizu and A. Satsuma, Appl. Catal., A, 2012, 433–434, 135–145.
54. S. M. Jung and P. Grange, Appl. Catal., B, 2002, 36, 325–332.
55. S. M. Kumbar and S. B. Halligudi, Catal. Commun., 2007, 8, 800–806.
56. H. Pfeifer, D. Freude and M. Hunger, Zeolites, 1985, 5, 274–286.
57. T. Baba and Y. Ono, Appl. Catal., A, 1999, 181, 227–238.
58. S. Uchida, K. Inumari and M. Misono, J. Phys. Chem. B, 2000, 104, 8108–8115.
59. V. M. Mastikhin, S. M. Kulikov, A. V. Nosov, I. V. Kozhevnikov, I. L. Mudrakovsky and M. N. Timofeeva, J. Mol. Catal., 1990, 60, 65–70.
60. V. M. Mastikhin, V. V. Terskikh, M. N. Timofeeva and O. P. Krivoruchko, J. Mol. Catal. A: Chem., 1995, 95, 135–140.

---

Footnote

† Electronic supplementary information (ESI) available: XPS, ATG and NH₃ isotherms. See DOI: 10.1039/c4cy00998c

---