Mesoporous tantalum phosphates: preparation, acidity and catalytic performance for xylose dehydration to produce furfural

Abstract

Mesoporous tantalum phosphates (TaOPO₄-m) with varying P/Ta molar ratios (m = 0.41–0.89) were prepared, comprehensively characterized by ICP-AES, N₂ physisorption, small-angle XRD, TEM, Raman, FT-IR, NH₃-TPD and IR of pyridine adsorption and employed to catalyze the dehydration of xylose to produce furfural in a biphasic batch reactor. The physicochemical properties of these TaOPO₄-m samples were affected significantly by variation of m. More ordered mesopores were formed in the sample with a higher m. On the other hand, the density of acidity decreased but the ratio of Brønsted acidity to Lewis acidity (B/L) increased with the increase in m. TaOPO₄-0.84, which showed adequate mesoporosity and a high B/L ratio, was identified as the best performing catalyst among these TaOPO₄-m catalysts in terms of high furfural selectivity (ca. 72 mol%). Correlating the catalyst performance with its acid property showed that the xylose consumption rate decreased with the increasing B/L ratio, while furfural selectivity showed a volcano-type dependence on the B/L ratio. Besides, the huge decrease in the furfural selectivity after poisoning the Brønsted acid sites by adding 2,6-dimethyl pyridine revealed a kind of Brønsted acid catalysis for selective furfural production.

---

1. Introduction

The selective transformation of biomass-derivatives to value-added chemicals has attracted more and more attention during the last few decades, because biomass is one of the most promising alternatives to fossil fuels for the production of chemicals in terms of its huge reserves and being carbon neutral.^1,2 Carbohydrates, which account for 75% of the annual production of biomass, contain two types of sugars: hexoses and pentose.³ Xylose is the most common pentose derived from xylan hemicelluloses and the second most abundant sugar.⁴ Catalytic dehydration of xylose to furfural could be an important route in the utilization of xylose, as furfural is an important raw material in production of various value-added chemicals, such as solvents, pharmaceuticals and synthetic resins.⁵

The commercial production of furfural involves dehydration of xylose in the presence of mineral acids such as sulfuric acid, acetic acid.^6,7 These mineral acids are flawed by their corrosiveness and toxicity. On the other hand, the separation of homogeneous catalysts from products is energetically intensive and economically unfeasible. Alternatively, solid acid catalyst has emerged as an ideal catalyst for the production of furfural, because they are more environment-friendly and easier to separate. Various solid acid catalysts have been used to catalyze the dehydration of xylose to furfural, including zeolites,^8–11 ion-exchange resins,^12,13 bulk heteropolyacids,¹⁴ metal oxides,^15,16 exfoliated titanate, niobate and titanoniobate nanosheets.¹⁷ Furfural formation is always accompanied by the formation of humins, as depicted in Scheme 1. Choudhary et al. studied the conversion of xylose using HCl as a Brønsted acid catalyst and CrCl₃ as Lewis acid catalyst.¹⁸ It is shown that xylose could undergo a triple dehydration reaction to furfural directly catalyzed by Brønsted acid sites (I), and that it could also be isomerized to xylulose catalyzed by Lewis acid sites (II), and then dehydrate to furfural rapidly catalyzed by Brønsted acid sites (III). Using the combination of Lewis and Brønsted acids, a much higher furfural yield can be obtained than using HCl alone. Furfural can further polymerize to humins catalyzed by Lewis and Brønsted acid sites (IV) and react with xylose or xylulose to form humins (V and VI) catalyzed by Lewis acids.¹⁹ It seems generally understood that Brønsted acid catalysts are more selective for furfural production than Lewis acid catalysts.¹⁹

Scheme 1 Liquid-phase dehydration of xylose and its competing reactions.^18,19

Besides the acidic property of catalyst, the solution of the reaction also plays an important role in furfural selectivity. A much higher furfural selectivity can always be obtained in a biphasic system (water phase + organic phase), because the formed furfural can be extracted from the aqueous phase by an organic phase to minimize the condensation between xylose and furfural.^20,21 Various organic solvents were applied, such as DMSO, 1-butanol, toluene, and methyl isobutyl ketone,^{8,11,13–15,17} among which toluene is found to be the best solvent for furfural production.^{11,13–15,17}

As water is always used as solvent in this reaction, it is crucial to develop water-tolerant heterogeneous catalysts. The group VB (i.e. V, Nb and Ta) pentoxides and phosphates are very attractive due to their strongly acidic and highly water-resistant property.^15,22–24 However, the catalytic activity of these group VB catalysts is limited by their relatively low specific surface area. As is well-known, synthesis of mesoporous is one of the most important way to increase the surface area and improve the diffusion for reactants and products. Thus, various mesoporous niobium phosphate and mesoporous tantalum phosphate were prepared and used as catalysts for many reactions.^25–29 Mesoporous tantalum phosphate is known as a strongly acidic solid catalyst with high surface area and large pore size, but its potential for the liquid-phase dehydration of xylose has been rarely explored.

In this paper, a series of mesoporous tantalum phosphate with varying molar P/Ta ratios (m) were prepared to regulate surface Lewis and Brønsted acidity and used to catalyze the dehydration of xylose in a biphasic batch reactor. Discussion of the correlation between the ratio of Brønsted acidity to Lewis acidity (B/L) and xylose consumption as well as furfural selectivity is also presented.

2. Experimental

2.1 Sample preparation

Mesoporous tantalum phosphate samples were prepared by hydrothermal method described in references.^25,26 In a typical synthesis procedure, 0.012 mol tantalum penta-ethoxide (Ta(OC₂H₅)₅, 99.9+%, Ningxia Orient Tantalum Industry Co., Ltd) and 0.036 mol tartaric acid (C₄H₆O₆, 99.5%, J&K Scientific Co., Ltd) were dissolved in 50 mL anhydrous ethanol to produce tantalum tartrate complex solution as the tantalum precursor. Excess ethanol was carefully removed by heating the solution at 80 °C under vigorous stirring. Then 38 mL deionized water was added to the solution under continuous stirring until a homogeneous solution was obtained. The solution was further mixed with 14 mL (NH₄)₂HPO₄ (98+%, Beijing Modern Eastern Fine Chemical Co., Ltd) solution containing the desired amount of P/Ta ratios (m). 15 mL of 0.256 M hexadecyltrimethylammonium bromide (CTAB, 99+%, J&K Scientific Co., Ltd) solution in deionized water was added dropwise (0.5 mL min⁻¹) to the mixture under vigorous stirring at 35 °C and stirred for another 60 min. The final mixture was transferred to a 25 mL Teflon lined autoclave and annealed at 130 °C for 24 h. The obtained precipitates were washed with deionized water for 3 times and then anhydrous ethanol for 3 times. The resulted precipitates were dried at 60 °C for 12 h and then calcined at 550 °C in flowing air for 6 h to obtain TaOPO₄-m samples (the molar P/Ta ratios in the calcined samples were m = 0.41, 0.61, 0.75, 0.84, and 0.89, as determined by inductively coupled plasma-atomic emission spectroscopy (ICP-AES)).

Hydrated tantalum oxide samples were prepared by precipitation of tantalum penta-ethoxide (Ta(OC₂H₅)₅, 99.9+%, Ningxia Orient Tantalum Industry Co., Ltd) in ethanol solution with ammonium hydroxide solution (NH₃·H₂O, 7 M, Beijing Modern Eastern Fine Chemical Co., Ltd) at room temperature. 15 g tantalum penta-ethoxide was first dissolved in 120 mL anhydrous ethanol, then 90 mL ammonium hydroxide solution (7 M NH₃·H₂O) was added dropwise (1 mL min⁻¹) to the solution under vigorous stirring. The resulted suspension was refluxed at 80 °C under vigorous stirring for 2 h and then aged for 3 h. The precipitate was then filtered, and washed with deionized water until conductivity of the filtrate was less than 10 μS cm⁻¹ (detected by DDS-307 W conductivity meter). The washed precipitate was dried at 110 °C overnight and calcined at different temperature in flowing air for 5 h. The resulted samples were denoted as Ta₂O₅-T, where T is the calcination temperature (T = 300, 500 °C).

2.2 Sample characterizations

The actual P/Ta ratios (m) for TaOPO₄-m samples were determined by ICP-AES analysis on a Perkin Elmer Optima 3300 RL instrument. The samples were dissolved by hydrofluoric acid (HF, 48–51%, J&K Scientific Co., Ltd) for the ICP-AES tests. Transmission electron microscopy (TEM) images of TaOPO₄-m samples were recorded on a JEOL JEM-2010 system operated at 120 kV. The small-angle XRD patterns for TaOPO₄-m samples were recorded on a D-Max-2200PC diffractometer at a scanning rate of 0.5° min⁻¹ in the range of 0.6–10° using Cu Kα radiation (λ = 0.15418 nm) as the light source at 40 kV and 30 mA. The surface area, pore volume, and average pore diameter data were obtained from nitrogen adsorption–desorption isotherms at 77 K on a Micromeritics TriStar II 3020 instrument. Before the isotherm measurements, the samples were degassed under vacuum at 90 °C for 2 h and then 200 °C for another 2 h. Raman spectra were recorded at room temperature on a Jobin-Yvon T64000 triple-stage spectrometer at a resolution of 2 cm⁻¹ with a 325 nm He–Cd laser at output power of 25 mW. Fourier transform infrared (FT-IR) spectra were recorded at room temperature on a Perkin-Elmer FT-IR System 2000 spectrophotometer at a resolution of 4 cm⁻¹ in transmission mode using self-supported wafers of the catalyst samples.

Temperature programmed desorption of NH₃ and IR of pyridine adsorption were measured to evaluate the acidic properties of the catalyst surfaces. NH₃-TPD profiles were recorded on Cat-Lab (BEL JAPAN, INC.) apparatus equipped with an on-line QIC-200 quadrupole mass spectrometer (Inprocess Instruments, GAM 200) as detector, as detailed previously.³⁰ In order to prevent a possible intervention of H₂O desorption, the signal of m/z = 15 was recorded to measure the NH₃ desorption profiles, rather than m/z = 16 or 17. Relative populations of Lewis and Brønsted acid sites on the catalyst surfaces were measured by IR of adsorbed pyridine carried out on the Perkin-Elmer FT-IR System 2000 spectrophotometer in transmission mode. A quartz-lined stainless steel IR cell with optical path length less than 5 mm was used for the measurements. The self-supported sample wafers (15 mg cm⁻²) were pretreated in a flowing of 20 vol% O₂/Ar (40 mL min⁻¹) at 300 °C for 2 h. Adsorption of pyridine was done at 100 °C, by switching the sample to a pyridine-containing Ar (10 000 ppm pyridine, 40 mL min⁻¹) for 30 min. The samples were purged in flowing Ar (40 mL min⁻¹) at 100 °C for 1 h before recording the IR spectra to remove the reversibly adsorbed pyridine. Each spectrum was recorded for 20 scans at a resolution of 4 cm⁻¹.

2.3 Catalytic reaction and product analysis

Biphasic batch reactions of xylose dehydration were conducted in a 25 mL stainless steel autoclave at 140 °C with magnetic stirring. In a typical reaction, 300 mg xylose, 100 mg catalyst, 6 mL deionized water and 14 mL toluene were loaded into the reactor at room temperature. After purging with nitrogen for 6 times, the reactor was pressurized to 0.8 MPa with nitrogen. The magnetic stirrer (900 rpm) was switched on when the reaction temperature (140 °C) was reached. The reaction was performed within kinetic controlling regime as no diffusion limitation was observed in our preliminary experiments conducted at different stirring speeds (600–1100 rpm). Termination of the reaction was done by switching off the stirrer and immediately quenching the reactor in an ice-water bath. The selective poisoning experiments were performed by adding different amount of 2,6-dimethyl pyridine (99%, J&K Scientific Co., Ltd) when loading the reactor.

The aqueous phase and organic phase products were collected by filtration and phase separation. The separated liquid products were analyzed by a LC-10 HPLC (Jiangshen, Dalian, China) equipped with an ultraviolet (280 nm) detector (UVD) as well as a refractive index detector (RID). Furfural in both organic phase and aqueous phase were quantified by using a SB-C18 column (Agilent Zorbax) at 40 °C with a 70 vol% methanol/water (flow rate 0.7 mL min⁻¹) eluent. For analysis of xylose in the aqueous phase, a P-OA 2000-0 column (Benson Polymeric Inc., USA) operated at 40 °C was used with 5 mM H₂SO₄ solution (flow rate 0.7 mL min⁻¹) as eluent.

The xylose conversion, product selectivity and yield were defined as follows:

The catalytic activity was expressed as area-specific xylose rxn rate according to the consumption of xylose, which were obtained using the following equation:

3. Results and discussion

3.1 Physicochemical properties

The P/Ta molar ratios (m) in TaOPO₄-m samples were determined by ICP-AES and the results are summarized in Table 1. The P/Ta molar ratios (m) of the samples increased with increasing the starting ((NH₄)₂HPO₄/Ta(OC₂H₅)₅) ratio in the preparation solution. However, the P/Ta ratio determined by ICP-AES was lower than the theoretical P/Ta ratio in the starting solution. Similar deviation was also observed in the literature.²⁹ The loss of P should be attributed to equilibrium during ligand exchange between tantalum tartrate complex and (NH₄)₂HPO₄. The nearly invariant m for samples with theoretical P/Ta ratio in starting solution higher than 2.5 suggested a saturation in the ligand exchange reaction.

Table 1 Composition, texture properties, and acidity of the catalyst samples

Sample	m		SA^b (m^2 g^−1)	PV^c (cm^3 g^−1)	PD^d (nm)		Area-specific acidity (μmol m^−2)			B/L^h
	Theoretical^a	ICP			Adsorption	Desorption	Total^e	BAS^f	LAS^g
a The starting (NH4)2HPO4/Ta(OC2H5)5 ratio in the preparation solution.b BET surface area.c Pore volume measured at P/P0 = 0.991.d Average pore diameter measured from the adsorption and desorption branch of isotherms according to BJH method.e Measured by NH3-TPD.f Calculated from the bands at 1540 cm^−1.g Calculated from the bands at 1445 cm^−1.h Measured by pyridine-IR and calculated using equations proposed by Emeis, assuming the integrated molar extinction coefficients are 2.22 and 1.67 cm μmol^−1 for Lewis and Brønsted acid sites, respectively.^31
TaOPO4-0.41	0.5	0.41	110	0.10	4.52	4.83	3.036	0.509	2.527	0.20
TaOPO4-0.61	1.0	0.61	156	0.13	3.96	4.37	2.769	0.756	2.013	0.38
TaOPO4-0.75	1.5	0.75	218	0.17	3.14	3.65	2.197	0.748	1.450	0.51
TaOPO4-0.84	2.5	0.84	275	0.22	2.96	3.25	2.138	1.055	1.084	0.98
TaOPO4-0.89	3.5	0.89	298	0.18	2.56	2.60	1.846	0.970	0.876	1.11
Ta2O5-500	—	—	98	0.12	4.12	4.36	1.710	—	1.710	—
Ta2O5-300	—	—	157	0.19	4.06	4.27	2.231	—	2.231	—

---

N₂ adsorption–desorption isotherms of the TaOPO₄-m samples are shown in Fig. 1. For TaOPO₄-0.41 and TaOPO₄-0.61 samples, their isotherms resembled with type II isotherms defined by IUPAC with no distinct hysteresis loops. For samples with m = 0.75, 0.84 and 0.89, the isotherms were typical type IV sorption isotherms with hysteresis loops in the range of 0.4 < P/P₀ < 1.0, which were characteristic for mesoporous materials. TaOPO₄-0.75 and TaOPO₄-0.84 samples displayed type H4 hysteresis loops at relative nitrogen pressure (P/P₀) below 0.80 and type H3 hysteresis loops at higher pressures. The Type H4 loop involves in narrow slit-like pores while the type H3 loop characterizes slit-shaped pores formed from aggregates of plate-like particles.³² TaOPO₄-0.89 sample had a type H2 hysteresis loop, which was associated with complex interconnected networks of pores with narrow necks and wide bodies (i.e., ink bottle pores).³² The change of hysteresis loop type could be due to the fast precipitation of tantalum phosphate. The pH of the starting solution was too high during the preparation of TaOPO₄-0.89 as (NH₄)₂HPO₄ was an alkaline compound. Higher pH resulted in the fast precipitation of tantalum phosphate without sufficient interaction with surfactant.²⁹

Fig. 1 N₂ adsorption–desorption isotherms of the TaOPO₄-m samples with different molar P/Ta ratio m.

Textural properties derived from the N₂ physisorption isotherms are also summarized in Table 1. Increasing m resulted in continuous increase in the specific surface area from 110 m² g⁻¹ for TaOPO₄-0.41 to 298 m² g⁻¹ for TaOPO₄-0.89. The average pore diameter decreased from 4.83 nm to 2.60 nm at the same time, which could be due to the increasing number of pore walls with increasing phosphate content. As more ordered mesopores formed, the pore volume increased from 0.10 cm³ g⁻¹ to 0.20 ± 0.02 cm³ g⁻¹ with m increasing from 0.41 to 0.89.

The effect of P/Ta molar ratios (m) on the mesoporosity of the TaOPO₄-m samples was investigated using small-angle XRD, as shown in Fig. 2. TaOPO₄-0.41 and TaOPO₄-0.61 samples appeared to be disordered materials, with no distinct X-ray diffraction peak in the small-angle region detected (0.6–10°). For the TaOPO₄-m samples of m ≥ 0.75, a broad peak centered at 2θ = 2.4° appeared, indicating a worm-like pore structure.³³ The average inter-pore distance in TaOPO₄-0.75 sample was calculated to be 3.68 nm according to the position of the peak. The shift of diffraction peak to lower angle with increasing the P/Ta molar ratio (m) suggested a larger inter-pore distance at higher P/Ta ratio. No peak appeared at higher 2θ angles, suggesting that all of these samples showed no long-range order. These observations were further supported by TEM results as shown in Fig. 3. It can be seen that the worm-like pores were not obvious for TaOPO₄-0.41 and TaOPO₄-0.61 samples, while they were much more apparent for the TaOPO₄-m samples of m ≥ 0.75, which was consistent with N₂ adsorption–desorption isotherms and small-angle XRD patterns. Thus, more phosphate contained in the framework of TaOPO₄ samples made mesoporous structure more stable.

Fig. 2 Small-angle XRD patterns for the TaOPO₄-m samples of (a) m = 0.41, (b) m = 0.61, (c) m = 0.75, (d) m = 0.84, (e) m = 0.89.

Fig. 3 TEM images of the TaOPO₄-m samples of (a) m = 0.41, (b) m = 0.61, (c) m = 0.75, (d) m = 0.84, (e) m = 0.89.

Surface acidity of solid acid catalyst is the key to the catalytic performance for dehydration reactions. The Brønsted acid site of TaOPO₄-m is due to the presence of P–OH, which was investigated by Raman and FT-IR. In order to compare the relative intensity of P–O bond among these TaOPO₄-m samples, the Raman and FT-IR spectra was normalized by the height of Ta–O vibration peak at 660 and 655 cm⁻¹, respectively.

Raman spectra of TaOPO₄-m samples are shown in Fig. 4A. As a comparison, the spectrum of Ta₂O₅-300 is also presented. For the Ta₂O₅-300 sample, a strong Raman band appeared at 660 cm⁻¹, which was characteristic for Ta–O stretching vibration mode.²⁶ Similar features were also observed in the TaOPO₄-m samples. Besides, a distinct band centered at 1013 cm⁻¹, which was attributed to the P–O asymmetric stretching vibrations of phosphate species, can be found for all TaOPO₄-m samples, suggesting that Ta–O–P bond was formed in mesoporous TaOPO₄ samples.²⁶ The band observed between 200 and 400 cm⁻¹ can be assigned to O–Ta–O bending modes for the Ta₂O₅-300 sample and to various O–P–O and O–Ta–O bending modes for TaOPO₄-m samples.³⁴ Furthermore, with increasing m, the area ratio of the peak assigned to P–O to the peak assigned to Ta–O increased.

Fig. 4 (A) Raman spectra and (B) FT-IR patterns of (a) Ta₂O₅-300, the TaOPO₄-m samples of (b) m = 0.41, (c) m = 0.61, (d) m = 0.75, (e) m = 0.84, (f) m = 0.89.

The FT-IR spectra for TaOPO₄-m and Ta₂O₅-300 samples are presented in Fig. 4B. The broad and intense band at 655 cm⁻¹ in both Ta₂O₅-300 and TaOPO₄-m samples could be assigned to the Ta–O stretching vibration mode.³⁵ The 1050 cm⁻¹ band only present in TaOPO₄-m samples corresponded to the P–O asymmetric vibration stretching mode of the phosphate groups.³⁶ The shoulder at about 1200 cm⁻¹ was also assigned to δ_P–OH of uncoordinated phosphate groups.²⁶ The band at 1620 cm⁻¹ assigned to H–O–H bending mode was associated to molecularly adsorbed water. As similar to the Raman data, intensity of peak (1050 cm⁻¹) associated with P–O bond increased with increasing m.

In order to understand the effect of m on the type and number of acidic sites on the TaOPO₄-m samples, NH₃-TPD and IR spectra of pyridine adsorption were measured, respectively. NH₃-TPD profiles of TaOPO₄-m and Ta₂O₅-T samples are shown in Fig. 5 and the corresponding quantitative results are given in Table 1. NH₃-TPD profiles for TaOPO₄-m samples showed a broad peak (130–500 °C) with maximum at 180–350 °C. The profiles could be divided into two desorption peaks (peak at 220 and 320 °C), which was attributed to the weak acid sites and strong acid sites.³⁷ The high-temperature peak became more distinct and lager amount of desorbed ammonia was detected with increasing m, which could be explained by more P–OH groups introduced with more phosphate content. However, for sample TaOPO₄-0.89, the desorption peak remained almost unchanged compared with sample TaOPO₄-0.84, which was consistent with the almost unchanged phosphate content. The profiles for Ta₂O₅-T (T = 300 °C, 500 °C) samples were different from that for mesoporous TaOPO₄-m samples. The samples of T = 300 °C produced two desorption peaks, featuring medium strong acid sites and strong acid sites, while high-temperature peak became almost invisible and the low-temperature peak reduced to a lower one on the profiles for Ta₂O₅-500.³⁷ The area specific acid density was obtained via NH₃-TPD profiles of TaOPO₄-m and Ta₂O₅-T. The density of acid sites decreased from 2.231 μmol m⁻² to 1.710 μmol m⁻², when the T of Ta₂O₅-T increased from 300 °C to 500 °C. The area specific acidity of TaOPO₄-m, normalized to the surface area, decreased from 3.036 μmol m⁻² to 2.769, 2.197, 2.138, and 1.846 μmol m⁻², when the m increased from 0.41 to 0.61, 0.75, 0.84, and 0.89, respectively.

Fig. 5 NH₃-TPD profiles for the Ta₂O₅-T samples of (a) Ta₂O₅-500, (b) Ta₂O₅-300 and the TaOPO₄-m samples of (c) m = 0.41, (d) m = 0.61, (e) m = 0.75, (f) m = 0.84, (g) m = 0.89.

Pyridine-IR spectra are shown in Fig. 6. All the spectra showed the bands centered at 1445 and 1610 cm⁻¹, which featured adsorbed pyridine on Lewis acid sites. These Lewis acid sites could be due to the existence of octahedral TaO₆ and tetrahedral TaO₄.³⁸ For TaOPO₄-m samples, the new bands at 1540 and 1638 cm⁻¹ are the characteristic bands of Brønsted acid sites due to the presence of OH groups associated with P on the surface.³⁹ While Ta₂O₅-T samples showed no detectable band at ca. 1540 cm⁻¹, indicating that Ta₂O₅-T samples possessed almost no Brønsted acidity. This result was consistent with our previous report on the acidity of hydrated tantalum oxide though TaCl₅ was used as the source of Ta.³⁰ Meanwhile the peak area ratio of 1540 cm⁻¹ to 1450 cm⁻¹ increased with increasing m, illustrating that the higher phosphate content, the more Brønsted acid sites were formed. Quantification of Lewis and Brønsted acid sites was obtained according to the integrated areas of the two peaks at 1540 and 1450 cm⁻¹, respectively, calculated through the basis of equations proposed by Emeis, given in Table 1.³¹

Fig. 6 Pyridine-IR spectra (A) for the TaOPO₄-m samples of (a) m = 0.41, (b) m = 0.61, (c) m = 0.75, (d) m = 0.84, (e) m = 0.89; pyridine-IR spectra (B) for the tantalum compound samples of (a) TaOPO₄-0.84, (b) Ta₂O₅-500, (c) Ta₂O₅-300.

3.2 Catalytic performance

The xylose dehydration reaction was conducted in batch operation, using a water–toluene solvent mixture at 140 °C. Xylose is dissolved and the catalyst is suspended in aqueous phase, and the formed furfural can be extracted from water phase to toluene phase. Fig. 7 shows xylose conversion and furfural selectivity as a function of reaction time for these TaOPO₄-m catalysts. Xylose conversion increased almost lineally at the initial 60 min, on the other hand, the furfural selectivity increased at the initial 120 min, then decreased with further reaction time. As shown in Scheme 1, the formed furfural could be further converted to humins, thus the furfural selectivity decreased in the last two hour due to the secondary reactions.¹⁹

Fig. 7 Performance of the TaOPO₄-m catalysts with different molar P/Ta ratio m by the time courses of (A) xylose conversion, (B) furfural selectivity: m = 0.41 (■), m = 0.61 (○), m = 0.75 (▲), m = 0.84 (◆), and m = 0.89 (▽).

The catalytic reaction data obtained at 2 h are given in Table 2 to show the catalytic activity and product selectivity over these mesoporous TaOPO₄-m catalysts. The main product was furfural (60–72 mol%), which could be accurately quantified by HPLC. However, lots of solid products called humins were also produced, which was difficult to quantify and denoted as others in the Table 2. Humins were highly polymerized insoluble carbonaceous species formed by the reaction of furfural with itself or between furfural and xylose.¹⁹ The consumption rate of xylose at the reaction time of 20 min decreased from 0.164 mmol h⁻¹ m⁻² to 0.083 mmol h⁻¹ m⁻² with the increase in m from 0.41 to 0.89, while the furfural selectivity increased from ca. 60 mol% to 72 mol% when m increased from 0.49 to 0.84, but decreased to ca. 60 mol% when m was further increased to 0.89. Thus, TaOPO₄-0.84 was identified as the best performing catalyst for furfural production, which gave the highest furfural selectivity (ca.72 mol%) and yield (ca. 35 mol%). The reaction rate is competitive compared to other catalyst systems in the reference (Table S1†) and the selectivity of furfural over the TaOPO₄-0.84 catalyst is higher compared with those obtained over the tantalum-based and niobium-based catalyst under the similar reaction conditions.^15,19,24

Table 2 Effect of molar P/Ta ratio m on xylose conversion and product distribution of catalytic xylose dehydration over theTaOPO₄-m catalysts^a

Catalyst	Xylose conv. (%)	Rxn rate^b (mmol h^−1 m^−2)	Product selectivity (mol%)		Furfural yield (%)
			Furfural	Others
a Rxn conditions: 100 mg catalyst; 300 mg xylose; 140 °C; rxn solvent, 6 mL of water and 14 mL of toluene; rxn time, 2 h.b Rxn rate after reaction for 20 min.
TaOPO4-0.41	33.3	0.164	59.6	40.4	19.9
TaOPO4-0.61	40.9	0.154	63.3	36.7	25.9
TaOPO4-0.75	50.4	0.150	65.7	34.3	33.1
TaOPO4-0.84	49.0	0.122	72.1	28.0	35.1
TaOPO4-0.89	49.8	0.083	60.4	39.6	30.1

---

The catalytic performance of TaOPO₄-0.84 catalysts was also compared with that of Ta₂O₅-300 and Ta₂O₅-500, as shown in Fig. 8. Ta₂O₅-300 and Ta₂O₅-500 exhibited a higher xylose conversion than TaOPO₄-0.84 in the initial 2 h. However, it should be noted that when the duration lasted to 240 min, TaOPO₄-0.84 showed the highest xylose conversion, indicating higher durability of TaOPO₄-0.84 compared to Ta₂O₅-T. Ta₂O₅-300 and Ta₂O₅-500 produced much lower furfural selectivity (<55 mol%) at the whole reaction (4 h). Thus, hydrated tantalum oxides are poorer than mesoporous tantalum phosphates for the dehydration of xylose.

Fig. 8 Performance of the Ta₂O₅-T and TaOPO₄-m catalysts by the time courses of xylose conversion (A), furfural selectivity (B): Ta₂O₅-300 (□), Ta₂O₅-500 (●), TaOPO₄-0.84 (▲).

3.3 The stability and regenerability of catalysts

In order to evaluate the stability of the spent TaOPO₄-0.84 catalyst, N₂ physisorption, small-angle XRD, TEM, and NH₃-TPD of the spent TaOPO₄-0.84 sample were performed. N₂ adsorption–desorption isotherms of the spent TaOPO₄-0.84 sample was shown in Fig. S1.† For the spent TaOPO₄-0.84 sample, the hysteresis loop was not as obvious as the fresh one. The small-angle XRD pattern of the spent catalyst was shown in Fig. S2† and no distinct X-ray diffraction peak in the small-angle region was detected. The TEM result was shown in Fig. S3.† It can be seen that the worm-like pores were not obvious for the spent catalyst and the carbon deposition was observed in the image. NH₃-TPD profile was shown in Fig. S4.† Less desorbed ammonia was detected in spent catalyst compared with the fresh one, which indicated the decreasing of surface acidity. These observations can indicate that solid humins deposited on the catalyst surface and blocked the mesoporosity of the catalyst. Leaching of tantalum phosphate from the solid catalyst to the aqueous phase was measured by ICP-AES. No tantalum was detected and the concentration of phosphate in solution was 2.507 μg mL⁻¹, which represented only 0.17 wt% of the phosphate initially present in the solid catalyst. This demonstrated that no leaching process existed in the whole reaction. The reusability of the TaOPO₄-0.84 catalyst was tested with the catalyst separated by filtration from the aqueous phase. The spent catalyst was thermally treated in air at 550 °C for 4 h before reuse. The results in Fig. 9 indicate that the thermal treatment regenerated the catalyst performance as the xylose conversion and furfural yield did not decreased a lot. The mesoporous tantalum phosphate exhibits fairly good stability when used in the xylose dehydration reaction.

Fig. 9 Reuse study for the TaOPO₄-0.84 catalyst at 140 °C after 2 h reaction. Rxn conditions: 100 mg catalyst; 300 mg xylose; rxn solvent, 6 mL of water and 14 mL of toluene; rxn time, 2 h. For the second and third run, the catalyst is regenerated at 550 °C for 4 h in air.

3.4 Correlation between the acid property and catalytic performance of the TaOPO₄-m catalysts

It has been know that the catalytic performance of xylose dehydration is dependent on the catalyst surface acid property. Lewis acid site can isomerize xylose to xylulose and Brønsted acid site can dehydrate xylulose to furfural, indicating that both Lewis and Brønsted acid sites are important for furfural production.^9,18,19 Therefore, we correlated xylose consumption rate and furfural selectivity of the TaOPO₄-m catalysts with the ratio of Brønsted acidity to Lewis acidity (B/L), as shown in Fig. 10. It was clear that the consumption rate of xylose decreased with increasing the B/L ratio, which agreed well with earlier reports.^18,19 Various solid acids were investigated to catalyze the aqueous-phase dehydration of xylose, which also showed that increasing the ratio of B/L decreased the catalytic activity.¹⁹ On the other hand, furfural selectivity showed a volcano-type dependence on the B/L ratio; the highest furfural selectivity (ca. 72 mol%) was obtained over TaOPO₄-0.84 with a B/L ratio of 0.98. Although the TaOPO₄-0.89 catalyst showed the highest B/L ratio (ca. 1.1), the furfural selectivity was only ca. 60 mol%. As shown in Scheme 1, Lewis acid sites can reduce furfural selectivity by catalyzing more side reactions to form humins. Weingarten et al. also found that high furfural selectivity was usually produced over the catalyst with a high B/L ratio; however, HY zeolite showed a low furfural selectivity due to strong irreversible adsorption of the furfural in the micro-pores, causing an increase in the rate of humins formation, although it has a high B/L ratio.¹⁹ The N₂ adsorption–desorption isotherms showed that TaOPO₄-0.89 had complex interconnected networks of pores with narrow necks, small average pore diameter and pore volume, which could be responsible for the low selectivity for furfural. According to the reference, the selectivity of furfural was restricted by furfural resinification and condensation reactions.⁶ On the furfural resinification reaction, the Brønsted acid shows similar catalytic activity compared with Lewis acid.¹⁹ However, on the furfural condensation reaction, xylulose isomerized by Lewis acid site is more reactive and can be converted to polymeric side-products faster than xylose, which could explain that Lewis acid sites catalyze more side reactions.⁴⁰

Fig. 10 Effect of molar Brønsted/Lewis ratio on the catalytic performance of furfural selectivity.

The results in Fig. 10 reveal that the nature of acid sites or B/L ratio is the key factor affecting the catalytic activity for the TaOPO₄-m catalysts and furfural selectivity. However, besides the B/L ratio, the texture properties (surface areas, pore size and volume) also changed with different P/Ta ratio, which also played an important role in the catalytic performance. Thus, we used TaOPO₄-0.84 as the only catalyst and regulated its B/L ratio by using 2,6-dimethyl pyridine to poison the Brønsted acid sites. The selective and permanent poisoning of protons with 2,6-dimethyl pyridine has been widely used to distinguish whether Brønsted acid sites are the active sites for the reaction.^41,42 Table 3 presents the catalytic data over the TaOPO₄-0.84 catalyst when various amount of 2,6-dimethyl pyridine was added to the reaction system. The furfural selectivity decreased but the xylose conversion and the selectivity for humins increased with increasing the amount of 2,6-dimethyl pyridine added in the reactor, as more Brønsted acid sites were poisoned. These results further identify Brønsted acid sites are more selective for furfural production than Lewis acid sites.

Table 3 Effect of poisoning reagent 2,6-DMP on xylose conversion and product distribution of catalytic xylose dehydration over the TaOPO₄-0.84 catalyst^a

Ratio^b	Xylose conv. (%)	Product selectivity (mol%)		Furfural yield (%)
		Furfural	Others
a Rxn conditions: 100 mg catalyst; 300 mg xylose; 140 °C; rxn solvent, 6 mL of water and 14 mL of toluene; rxn time, 2 h.b Refers to the molar ratio of 2,6-DMP added to the reaction solution to the total acid sites of the catalyst.
0	49.0	72.1	27.9	35.3
2 : 1	52.2	56.2	43.8	29.3
4 : 1	53.6	54.7	45.3	29.3
10 : 1	54.1	50.3	49.7	27.2
15 : 1	55.6	43.9	56.1	24.4
50 : 1	58.6	26.2	73.8	15.3

---

4. Conclusions

This work provides the first investigation on the potential of mesoporous tantalum phosphate (TaOPO₄-m) for the liquid-phase dehydration of xylose to produce furfural. The P/Ta ratio (m) was found to be the key to the physicochemical properties and catalytic performance. Higher m resulted in more mesoporosity and higher ratio of Brønsted acidity to Lewis acidity (B/L) for the TaOPO₄-m samples. TaOPO₄-0.84 was identified as the most selective catalyst for the furfural production (72 mol%). Correlations between the catalytic performance and acidic property showed that the consumption rate of xylose decreased with increasing the B/L ratio, while higher furfural selectivity was obtained over the catalyst with a higher B/L ratio. Selective poison of Brønsted acid sites by 2,6-dimethyl pyridine revealed that the dehydration of xylose to furfural would be a kind of Brønsted acid-catalyzed reaction. On the other hand, pores with small pore volume were found to be adverse to the production of furfural due to a high selectivity for humins.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (grant: 21221062) and the National Basic Research Program of China (Grant: 2013CB933103).

References

1. S. Adhikari, C. Chandrapal, N. Murali and S. Fernando, Energy Fuels, 2006, 20, 1727–1737.
2. G. W. Huber, S. Iborra and A. Corma, Chem. Rev., 2006, 106, 4044–4098.
3. M. Watanabe, T. M. Aida, R. L. Smith and X. Qi, Bioresour. Technol., 2012, 109, 224–228.
4. S. Iborra, A. Velty and A. Corma, Chem. Rev., 2007, 107, 2411–2502.
5. L. Hu, G. Zhao, W. W. Hao, X. Tang, Y. Sun, S. J. Liu and L. Lin, RSC Adv., 2012, 2, 11184–11206.
6. K. J. Zeitsch, Sugar Ser., 2000, 13, 1–22.
7. E. I. Gürbüz, J. M. R. Gallo, D. M. Alonso, S. G. Wettstein, W. Y. Lim and J. A. Dumesic, Angew. Chem., Int. Ed., 2013, 52, 1270–1274.
8. A. S. Dias, M. Pillinger and A. A. Valente, J. Catal., 2005, 229, 414–423.
9. V. Choudhary, A. B. Pinar, S. I. Sandler, D. G. Vlachos and R. F. Lobo, ACS Catal., 2011, 1, 1724–1728.
10. S. B. Kim, S. J. You, Y. T. Kim, S. M. Lee, H. Lee, K. Park and E. D. Park, Korean J. Chem. Eng., 2011, 28(3), 710–716.
11. S. Lima, A. Fernandes, M. M. Antunes, M. Pillinger, F. Ribeiro and A. A. Valente, Catal. Lett., 2010, 135, 41–47.
12. J. Franco, L. Reina, C. Tabarez, G. Moyna, P. Moyna and V. Heguaburu, Catal. Commun., 2012, 27, 88–91.
13. A. Larreategui, J. Requies, M. B. Güemez, P. L. Arias and I. Agirrezabal-Telleria, Bioresour. Technol., 2011, 102, 7478–7485.
14. A. S. Dias, S. Lima, M. Pillinger and A. A. Valente, Carbohydr. Res., 2006, 341, 946–2953.
15. C. García-Sancho, I. Sádaba, R. Moreno-Tost, J. Mérida-Robles, J. Santamaría-González, M. López-Granados and P. Maireles-Torres, ChemSusChem, 2013, 6, 635–642.
16. A. Chareonlimkun, V. Champreda, A. Shotipruk and N. Laosiripojana, Bioresour. Technol., 2010, 101, 4179–4186.
17. A. S. Dias, S. Lima, D. Carriazo, V. Rives, M. Pillinger and A. A. Valente, J. Catal., 2006, 244, 230–237.
18. V. Choudhary, S. I. Sandler and D. G. Vlachos, ACS Catal., 2012, 2, 2022–2028.
19. R. Weingarten, G. A. Tompsett, W. C. Conner Jr and G. W. Huber, J. Catal., 2011, 279, 174–182.
20. Y. Román-Leshkov, J. N. Chheda and J. A. Dumesic, Science, 2006, 312, 1933–1937.
21. J. N. Chheda, Y. Román-Leshkov and J. A. Dumesic, Green Chem., 2007, 9, 342–350.
22. C. García-Sancho, M. B. Güemez, P. Maireles-Torres and I. Agirrezabal-Telleria, Appl. Catal., B, 2014, 152–153, 1–10.
23. C. García-Sancho, P. Maireles-Torres, P. L. Arias and I. Agirrezabal-Telleria, Chin. J. Catal., 2013, 34, 1402–1406.
24. X. L. Li, T. Pan, J. Deng, Y. Fu and H. J. Xu, RSC Adv., 2015, 5, 70139–70146.
25. A. Sarkar and P. Pramanik, J. Non-Cryst. Solids, 2010, 356, 2709–2713.
26. I. Jiménez-Morales, J. Santamaría-González, P. Maireles-Torres and A. Jiménez-López, Appl. Catal., B, 2012, 123–124, 316–323.
27. I. Jiménez-Morales, A. Teckchandani-Ortiz, J. Santamaría-González, P. Maireles-Torres and A. Jiménez-López, Appl. Catal., B, 2014, 144, 22–28.
28. A. Sarkar and P. Pramanik, Microporous Mesoporous Mater., 2009, 117, 580–585.
29. Y. Zhang, J. J. Wang, X. H. Liu, X. C. Li, Y. J. Xia, G. Z. Lu, J. W. Ren and Y. Q. Wang, Catal. Sci. Technol., 2012, 2, 2485–2491.
30. L. Z. Tao, B. Yan, Y. Liang and B. Q. Xu, Green Chem., 2013, 3, 696–705.
31. C. A. Emeis, J. Catal., 1993, 141, 347–354.
32. D. H. Everett, R. A. W. Haul, L. Moscou, R. A. Pierotti, J. Rouquérol, T. Siemieniewska and K. S. W. Sing, Pure Appl. Chem., 1985, 57, 603–619.
33. N. K. Mal, M. Fujiwara and A. Bhaumik, J. Non-Cryst. Solids, 2007, 353, 4116–4120.
34. G. T. Stranford and R. A. Condrate, J. Solid State Chem., 1990, 85, 326–331.
35. I. W. Boyd, M. B. Mooney, P. K. Hurley, J. T. Beechinor, B. J. O'Sullivan, P. V. Kelly, G. M. Crean, J.-P. Senateur, C. Jimenez, M. Paillous and J. Y. Zhang, Appl. Phys. A: Mater. Sci. Process., 2000, 70, 647–649.
36. V. V. Ordomsky, V. L. Sushkevich, J. C. Schouten, J. van der Schaaf and T. A. Nijhuis, J. Catal., 2013, 300, 37–46.
37. V. Hulea and E. Dumitriu, J. Catal., 2003, 218, 249–257.
38. K. Nakajima, Y. Baba, R. Noma, M. Kitano, J. N. Kondo, S. Hayashi and M. Hara, J. Am. Chem. Soc., 2011, 133, 4224–4227.
39. L. C. P. Almeida, R. Landers, R. C. G. Vinhas, F. J. Luna and M. K. de Pietre, React. Kinet., Mech. Catal., 2010, 99, 269–280.
40. O. Ershova, J. Kanervo, S. Hellsten and H. Sixta, RSC Adv., 2015, 5, 66727–66737.
41. H. C. Liu, N. Bayat and E. Iglesia, Angew. Chem., Int. Ed., 2003, 42, 5072–5075.
42. L. Oliviero, J. C. Lavalley, F. R. Sarria, M. Gaillard, F. Maugé and A. Vimont, Phys. Chem. Chem. Phys., 2005, 7, 1861–1869.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra07830c

---