Niobium-Enhanced Kinetics of Tantalum Phosphate in Catalytic Glucose Dehydration to 5-Hydroxymethylfurfural

Abstract

The application of water-tolerant bifunctional solid acids with high kinetic performance in converting glucose to 5-hydroxymethylfurfural (HMF) presents a number of challenges. In this study, the effect of doping a hierarchically porous tantalum phosphate monolith with transition metal ions (Nb⁵⁺, V⁵⁺, Zr⁴⁺, Ni²⁺, and Zn²⁺) was explored in delivering superior glucose dehydration kinetics and stability. Doping with Nb⁵⁺ (x%Nb-TaP) resulted in the best catalytic performance with enhanced tantalum phosphate stability. The incorporation of Nb⁵⁺ ions inhibited tantalum phosphate crystallization, increased the density of Lewis acid and Brønsted acid sites and average mesopore size, with consequently enhanced kinetics, enabling the reaction kinetics of fructose to approach a steady state. Application of a 25% mol/mol Nb (25%Nb-TaP) activated at 600 °C to convert 1.0 wt.% glucose in a water/methyl isobutyl ketone (MIBK) biphasic system delivered an HMF yield and selectivity of 72.6% and 95.6%, respectively. Moreover, an HMF productivity of 0.11 mol·h⁻¹·kg-solution⁻¹ was achieved by treating a 15.0 wt.% glucose feed at 170 °C in a water/MIBK biphasic system at a catalyst loading of 10.0 wt.%. The 25%Nb-TaP catalyst largely retained its initial activity after three recycles in the water/MIBK biphasic system, generating an HMF yield of 54.1% and selectivity of 87.0%. The results of this study demonstrate the significant potential of Nb-TaP for industrial-scale HMF production.

1. Introduction

As the most abundant form of biomass, cellulose can be hydrolyzed into glucose and then transformed to other chemicals [1,2,3]. A number of chemical routes have been explored in converting glucose to commercially valuable chemicals and compounds. In these catalytic routes, 5-hydroxymethylfurfural (HMF) is a critical intermediate, serving as a platform chemical. Transforming glucose to HMF involves two reaction steps, i.e., glucose is first isomerized to fructose with subsequent dehydration to produce HMF. Currently, processes that employ fructose as a substrate utilize homogeneous acid catalysts, including organic and inorganic acid and metal salts (in biphasic reaction systems), generating a high HMF yield and selectivity [4,5,6,7,8,9,10,11,12,13]. The high cost of fructose and the organic solvent and the complex separation ultimately contribute to high HMF prices, limiting consumer end-product viability and market expansion. Extensive efforts have been directed at HMF production from inexpensive and readily available glucose. A range of homogeneous catalysts and reaction systems have been explored for the production of HMF, obtaining a high yield as ca. 60% [14,15,16,17,18]. The operation of homogeneous reaction systems presents a series of difficulties, including equipment corrosion, catalyst recovery, and the high costs associated with solid waste treatment. The viability of homogeneous catalysis for practical industrial production remains a subject of debate.

The application of solid acid catalysis is currently considered a promising process to produce HMF. Metal oxides, zeolite-based solid acids, and metal phosphates have been the most investigated solid acid catalysts. However, solid acids have generated relatively low yields of HMF from glucose in water. Lew et al. reported that ZrO₂ produced a ~17% yield of HMF in water at 250 °C where the catalyst loading was 100% [19]. Kawamura and co-workers have recorded an HMF yield of 12.3% for the reaction of glucose over ZrO₂ in dimethyl sulfoxide (DMSO) [20]. The ZrO₂ surface is generally considered as neutral or weakly acidic in water [21], which is insufficient to promote the isomerization of glucose or the dehydration of fructose. Various bifunctional solid acids bearing Lewis and Brønsted acid sites have been designed and employed to increase catalytic activity, including metal oxides, metal-organic frameworks (MOFs), metal-loaded zeolites, and metal phosphates [22,23,24,25,26,27,28,29,30,31,32,33,34,35]. Metal oxide and zeolite catalysts such as hybrid-TiO₂, W-containing oxides, Sn-Beta, and Nb-BEA catalysts can afford an HMF yield of 20–45% in an aqueous solution, whereas the HMF yield can be improved to be 70–80% in NaCl-water/THF biphasic systems. However, ion leaching, occludation of pores with humic substances, and carbon deposition resulted in the rapid deactivation of these catalysts [31,32,33,34,35].

A number of studies have established enhanced performance of transition metal phosphates in the catalytic transformation of saccharidic biomass that includes hydrolysis, dehydration and C-C bond cleavage [36,37,38]. The use of tin phosphate has produced an HMF yield of 50–61% in NaCl-containing water/THF or water/MIBK biphasic systems [39,40]. Manganese phosphate and zirconium phosphate catalysts afforded a suboptimal 45–60% yield of HMF in a water/THF biphasic system [41,42,43]. Morales prepared a meso-porous tantalum phosphate (TaOPO₄, TaP) by precipitation and recorded a 33% HMF yield in water/MIBK [44]. Although Sn⁴⁺, Zr⁴⁺, and Mn³⁺ phosphates exhibited superior catalytic performance relative to TaP, appreciable metal ion (Sn⁴⁺, Zr⁴⁺, and Mn³⁺) and PO₄³⁻ leaching into solution during reaction is a serious drawback, and TaP showed greater stability than the other metal phosphates.

In previous work, we examined the kinetics of disaccharide hydrolysis–aldose isomerization–ketose dehydration, accompanied by decomposition in a continuous four-step cascade reaction over a hierarchical porous niobium phosphate [45]. It was demonstrated that the porous structure influenced the macroscopic reaction kinetics, and a high reaction efficiency was achieved using the catalyst with a regular co-continuous macro-porous structure. We have extended that work to prepare a hierarchically porous tantalum phosphate with a regular co-continuous macro-porous structure by sol-gel accompanied by phase separation [46]. The resultant catalyst has delivered superior performance in glucose dehydration to HMF, achieving 65% yield and 96% selectivity in a water/MIBK biphasic system. The TaP catalyst exhibited a solvent dependence where a 3/7 (w/w) water/MIBK combination resulted in the highest reaction efficiency in terms of HMF yield and selectivity when compared with DMSO, alcohols and THF as solvents or extraction phases. Kinetic analysis has revealed a high reaction rate for side reactions that limit the ultimate HMF yield, indicating an insufficient rate for the main glucose isomerization and fructose dehydration reactions. This response is attributed to a low density of acid sites required for the main reaction, a deviation of the fructose intermediate reaction from a steady-state form, and the relatively low stability of the TaOPO₄ catalyst. In this study, we explore the effect of doping the hierarchically porous tantalum phosphate with transition metal ions (V⁵⁺, Nb⁵⁺, Zr⁴⁺, Cd²⁺, Zn²⁺, and Ni²⁺) as a means of increasing acid site density, apparent reaction kinetics, and surface stability. The experimental results have shown that doping with Nb⁵⁺ ions significantly improved the catalytic performance and stability of the tantalum phosphate, demonstrating a high HMF yield (72.6%, mol/mol) and selectivity (95.6%, mol/mol) in the conversion of glucose in a water/MIBK biphasic system, displaying significant potential in the practical production of HMF.

2. Results and Discussion

2.1. Synthesis of Metal-Doped Tantalum Phosphate and Catalytic Performance Screening

Microporous tantalum phosphates doped with different transition metal ions were synthesized using a sol-gel method accompanied by phase separation, generating a molar ratio of dopant to Ta⁵⁺ (M:Ta) as 5%, where M represents the transition metal ion (V⁵⁺, Nb⁵⁺, Zr⁴⁺, Cd²⁺, Zn²⁺, and Ni²⁺). A representative scanning electron microscopy (SEM) image and photo of 25%Nb-TaP xerogel are shown in Figure 1a,b. The 25%Nb-TaP xerogel has a co-continuous microporous structure with a skeleton diameter of ca. 0.5 μm. In the sol-gel process, TaCl₅ and the transition metal salt were dissolved in ethanol and formed Ta⁵⁺ and M metal ethoxide as metal precursors. The two metal ethoxides were hydrolyzed to (CH₃CH₂O)_xTa(OH)_5-x and (CH₃CH₂O)_xM(OH)_n-x and consecutively condensed by dehydration to form metal-O-metal bonds, and then they were gelatinized. In this process, polyethylene oxide (PEO) was added to induce phase separation and form co-continuous macropores, with the addition of citric acid to control the rate of gelation, producing mesopores following calcination [46].

The effects of doping on the catalytic dehydration of glucose to HMF was assessed in a preliminary screening of 5%M-TaP using a 1.0 wt.% aqueous glucose solution (Figure 1c). Doping with Nb⁵⁺ delivered the best performance in terms of HMF yield and selectivity, whereas doping with V⁵⁺, Zr⁴⁺, Zn²⁺, and Ni²⁺ delivered a relatively low HMF yield. It should be noted that doping with Cd²⁺ served to lower HMF selectivity. Consequently, we will focus our research on the structural analysis and modulation of niobium-doped tantalum phosphate catalysts in subsequent tests.

2.2. Structure and Surface Properties of Niobium-Doped Tantalum Phosphate

In order to systematically investigate the effect of niobium doping on the catalytic performance of tantalum phosphate, the molar ratio of Nb to Ta was adjusted in the 5–30% (mol/mol) range. We have previously reported that tantalum phosphate contained a residual PEO and citric acid content, which also applies to the M-TaP materials [46]. The TG-DTG profile (Figure 2a) for 25%Nb-TaP exhibits a ca. 8.0% weight loss <200 °C, which can be attributed to a residual solvent [46]. A weight loss of ca. 3.2% between 200 and 800 °C corresponds to the removal of organic components. The N₂ adsorption–desorption isotherms for 25%Nb-TaP at different calcination temperatures are shown in Figure 2b; the associated pore characteristics are presented in

Table 1. Pore characteristics of monolithic 25%Nb-TaP samples calcined at different temperatures.

Temperature (°C)	SBET (m2·g−1)	VP (cm3·g−1)	DP (nm)
400	240	0.16	2.6
500	225	0.18	3.2
600	101	0.14	5.7
700	55	0.12	8.6
800	54	0.056	4.1

. The isotherms for 25%Nb-TaP calcined at 400–700 °C are consistent with type IV isotherms with hysteresis loops between 0.45 and 1.0 (p/p₀), confirming the presence of mesopores. The rapid adsorption of N₂ at a low relative pressure can be attributed to the formation of micropores primarily associated with the combustion of PEO and citric acid [45,46]. Following calcination at 400 and 500 °C, the samples retained surface areas (S_BET) >200 m²·g⁻¹. However, this temperature range is insufficient for the complete removal of organic species, resulting in the formation of black carbon residues. Calcination at 600 °C generated a S_BET of ca. 100 m²·g⁻¹ with a V_P of 0.14 m²·g⁻¹ and DP of 5.66 nm. Increasing the calcination temperature to 800 °C resulted in a decrease in S_BET and V_P to ca. 54 m²·g⁻¹ and 0.056 m²·g⁻¹, respectively. The decrease in S_BET can be attributed to particle growth [11,46]. The XRD patterns of the TaP samples with 0–30%Nb⁵⁺ (Nb⁵⁺/Ta⁵⁺, mol/mol) following calcination at 600 °C for 5 h and 25%Nb-TaP after calcination at different temperatures are shown in Figure 3. The 5%Nb-TaP sample exhibited significantly reduced diffraction peak intensities at 21.2°, 27.3°, 33.1°, 34.8°, and 38.4°, with a decrease in crystallinity by ca. 55% relative to the starting tantalum phosphate (TaP). At 10%Nb⁵⁺ doping, the sample exhibited weaker diffraction peaks with a broader full-width-at-half-maximum (FWHM), indicating a predominantly amorphous structure. As the Nb⁵⁺ content increased, the diffraction peak intensities progressively diminished. This response may be attributed to the difference in the Nb⁵⁺ and Ta⁵⁺ ionic radii, where doping-induced lattice distortion alters local atomic configurations and changes the crystallization kinetics [47]. The XRD patterns of 25%Nb-TaP following calcination at 400–800 °C are shown in Figure 3b, confirming that the samples are amorphous in this temperature range. An increase in calcination temperature was accompanied by an increased intensity of the diffraction peak at 23.2°, indicating that the material begins to crystallize gradually.

The FT-IR and Raman spectra of 25%Nb-TaP calcined at 600 °C are shown in Figure 4. The FT-IR spectrum exhibits a distinct peak at ca. 1652 cm⁻¹ and a broad band around 3445 cm⁻¹ that can be attributed to hydroxyl (-OH) bending and asymmetric stretching vibrations, respectively, associated with adsorbed water [48]. These hydroxyl groups may influence catalyst performance by affecting H⁺ release that modulates Brønsted acidity [45]. The sharp peak at 1015 cm⁻¹ corresponds to the asymmetric stretching vibration of phosphate P-O groups [49]. The peak at 415 cm⁻¹ is due to PO₄ bending vibrations, and the peak at 2308 cm⁻¹ is assigned to Ta-O-Ta stretching vibrations [50]. In addition, FTIR absorption at 643 cm⁻¹ is associated with Nb-O stretching modes [51]. The two strong Raman bands observed at 1019 cm⁻¹ and 259 cm⁻¹ are attributed to phosphate P-O asymmetric stretching and O-Ta-O bending vibrations, respectively [52]. The peak at 1214 cm⁻¹ is assigned to PO₃ symmetric stretching vibrations [49]. Moreover, the band at 802 cm⁻¹ is attributed to Nb-O stretching vibrations [53], whereas the peak at 630 cm⁻¹ corresponds to Ta-O stretching modes [52].

The surface acid characteristics of 5–30%Nb-TaP were determined using NH₃-TPD and pyridine-IR (Figure 5). The NH₃-TPD profiles of the x%Nb-TaP samples exhibit broad bands in the 100–500 °C range, which corresponds to the desorption of NH₃ from weak and medium-strength acid sites. As the Nb doping content was increased, the NH₃-TPD peak areas exhibited a progressive increase, indicating significantly enhanced acidity. The temperature profiles presented in Figure 5 demonstrate that the total acidity and acid strength associated with x%Nb-TaP increase systematically with higher Nb doping levels. The total acidity increased from 0.17 mmol NH₃·g⁻¹ to 0.94 mmol NH₃·g⁻¹ with an increased Nb⁵⁺ content (from 5% to 30%). In our previous work, we established that weak and medium-strength acid sites are associated with P-O-H groups and defects on the x%Nb-TaP surfaces [11,46]. The temperature-dependent pyridine FT-IR spectrum for 25%Nb-TaP is presented in Figure 5b. The sharp peak at 1490 cm⁻¹ arises from a synergism involving Brønsted and Lewis acid sites. The weaker peaks at 1450 cm⁻¹ and 1640 cm⁻¹ correspond to pyridine adsorbed on Lewis acid sites. The characteristic peak at 1540 cm⁻¹ is associated with Brønsted acid sites. The amount of pyridine desorbed from 25%Nb-TaP increased with increasing desorption temperatures. However, the total remaining pyridine was greater than that adsorbed on the tantalum phosphate host. The results indicate that doping with niobium significantly enhanced the acidity of tantalum phosphate [46], which is consistent with the NH₃-TPD results.

2.3. Catalytic Performance of Nb-Doped TaP in Glucose Dehydration

The effects of the level of Nb doping and calcination temperature were first evaluated to optimize the catalyst. The catalytic performance of 5–30%Nb-TaP calcined at 600 °C in the conversion of glucose to HMF is shown in Figure 6. The HMF yield was improved by increasing Nb doping from 5% to 25% (mol/mol), achieving an optimal HMF yield of 29.4%, approximately 8% higher than that obtained with undoped tantalum phosphate. A further increase in Nb⁵⁺ doping to 30% served to decrease both the yield and selectivity of HMF, suggesting that a decomposition of substrate and HMF was promoted at a high Nb⁵⁺ content. Wrigstedt et al. investigated the dehydration of glucose to HMF catalyzed by CrCl₃·6H₂O, where the introduction of Brønsted acidity served to accelerate the fructose dehydration step but led to a significant decrease in HMF yield [54]. This was primarily attributed to the formation of a chromium-glucose chelate, which led to a decrease in the effective concentration of Cr³⁺ and, therefore, suppressed the isomerization rate of glucose to fructose. Surface acidity plays a crucial role in catalytic systems. Accordingly, 25%Nb was identified as the most suitable to modify the host tantalum phosphate catalyst. The effect of calcination temperature (400–800 °C) on the catalytic performance of 25%Nb-TaP in glucose dehydration to HMF is illustrated in Figure S1. An optimal HMF yield of 29.4% was achieved by treating 1.0 wt.% aqueous glucose with 25%Nb-TaP calcined at 600 °C. Increasing the calcination temperature to 700 and 800 °C resulted in a slight decrease in HMF yield and selectivity.

The impact of reaction conditions, including reaction time, reaction temperature, catalyst, and initial glucose concentration, was focused on the catalytic performance of 25%Nb-TaP calcined at 600 °C. The temporal variation of glucose consumption and HMF formation over 25%Nb-TaP at 170 °C with a catalyst loading of 1:10 (w/w) is shown in Figure 7a. An optimal HMF yield of 29.4% with a selectivity of 54.7% was obtained at a reaction time of 3 h. Extending the reaction time resulted in a decrease in yield and selectivity, achieving a 24.2% HMF yield and 31.6% selectivity at a reaction time of 6 h. This decline is attributed to the decomposition of HMF and glucose, with an increased production of fructose derived from glucose isomerization. The insoluble carbon species can occlude the catalyst mesopores, leading to deactivation. A thermal decomposition of both substrate and HMF contributed to the observed decreased yield and selectivity. However, when compared with tantalum phosphate, the HMF yield increased by 5% and the corresponding reaction time reduced from 4 h to 3 h. Delidovich et al. used commercial hydrotalcite as a catalyst for the conversion of glucose to HMF and achieved an HMF yield of 27% after 24 h at 90 °C at the initial glucose concentration of 1.0 wt.% [55]. Candu et al. employed a Nb(0.05)-β18 catalyst in water and recorded a 16.2% HMF yield after 24 h at 180 °C with an initial glucose concentration of 3.6 wt.% [30]. The catalyst system developed in this study delivered a higher HMF yield at a lower reaction temperature.

The effect of reaction temperature (150–180 °C) on the catalytic performance of 25%Nb-TaP at a reaction time of 3 h is shown in Figure 7b. Glucose conversion increased with increasing reaction temperature, achieving a maximum HMF yield and selectivity at 170 °C. Increasing the reaction temperature to 180 °C served to decrease both yield and selectivity, indicating a contribution due to the decomposition of glucose and HMF. Girisuta and Ordomsky et al. demonstrated that the decomposition and condensation of HMF generates humins in homogeneous and heterogeneous acid media at high temperatures [56,57]. In our previous work, we also observed that HMF underwent decomposition and condensation under acidic conditions at temperatures above 140 °C in fructose dehydration and the conversion of glucose to HMF over metal phosphate and metal oxide catalysts [11,45,46].

Minimizing catalyst loading is an important consideration in lowering process costs and equipment investment, where catalyst loading is defined as the weight ratio of 25%Nb-TaP to the substrate. The effect of catalyst loading was examined in the 1:20–1:2 range by treating 1.0 wt.% aqueous glucose at a reaction time of 3 h (Figure 7c). Increasing the loading from 1:20 to 1:10 increased both the yield and selectivity of HMF by ca. 5%. A further increase to 1:5 and 1:2 served to increase glucose conversion, but with a decrease in HMF yield and selectivity. This suggests that a high catalyst loading promotes reactant and target product decomposition. Higher catalyst loading results in stronger acidic conditions, which promotes HMF decomposition and condensation to generate humins.

A high reaction efficiency is required at high initial substrate concentrations in order to minimize reactor size in practical manufacturing. The catalytic performance of 25%Nb-TaP at an initial glucose concentration ranging from 1.0 to 20.0 wt.% is shown in Figure 7d for a catalyst loading of 1:10 at 170 °C and 3 h. A higher initial glucose concentration was accompanied by a lower HMF yield and selectivity. An increase in initial glucose concentration from 1.0 wt.% to 10.0 wt.% resulted in a decrease in HMF yield and selectivity from 29.4% and 54.7% to 14.5% and 27.3%, respectively. A further decrease to 10.2% and 20.3% was observed when increasing the initial glucose concentration to 20.0 wt.%. The decline in catalytic performance can be attributed to catalyst deactivation caused by carbon deposition on catalytic sites and the occlusion of meso- and macro-pores [11,46]. Rao and Guo investigated the catalytic performance of 15P-TiO₂ and mesoporous Nb₂O₅ in the conversion of high-concentration glucose to HMF [25,58]. They demonstrated that the HMF product underwent condensation with glucose and other intermediates, generating humin-like byproducts. The accumulation of huminic compounds results in catalyst deactivation during the reaction.

2.4. Dehydration of Glucose in a Water/MIBK Biphasic System

In addressing the decline of catalytic performance at a high initial glucose concentration, Ordomsky and Gao et al. investigated the effects of organic solvents on HMF formation over metal oxides and tantalum phosphate in homogeneous water-organic solvent mixed solution and water-organic solvent biphasic systems, including DMSO, ethanol, n-propanol, and MIBK [11,46,59]. The results have shown that the HMF yield and selectivity were significantly increased using a water/MIBK biphasic system with a weight ratio of water/MIBK equal to 3/7. In this study, the dehydration of glucose to HMF over 25%Nb-TaP was investigated in a water/MIBK biphasic system (water/MIBK = 3/7, w/w) with an initial glucose concentration in the 1.0–20.0 wt.% range (Figure 7e). When compared with the results generated in water, utilization of the water/MIBK biphasic system significantly increased reaction efficiency in terms of the HMF yield and selectivity. An HMF yield of 72.6% with a selectivity of 95.6% was obtained in the conversion of 1.0 wt.% glucose using a water/MIBK biphasic system. An increase in the initial glucose concentration resulted in a lower process efficiency. However, the HMF yield and selectivity in the water/MIBK biphasic system were appreciably higher than the values recorded in aqueous solutions. In terms of the practical production of HMF, productivity can be quantified as the amount of HMF produced by treating 1.0 wt.% glucose solution in 1.0 h. For instance, a 45.3% HMF yield with 82.2% selectivity obtained in the conversion of 15.0 wt.% glucose represents an HMF productivity of 0.107 mol·kg-solution⁻¹·h⁻¹. The productivity of 25%Nb-TaP exceeded that of the starting tantalum phosphate catalyst (0.057 mol·kg-solution⁻¹·h⁻¹) under the same reaction conditions, with a corresponding HMF yield of 27.5% [46]. Lanziano et al. prepared a bifunctional hybrid-TiO₂, which generated an HMF yield of ca. 45% from a 2.0 wt.% glucose aqueous solution; however, the catalyst rapidly deactivated and could not be regenerated by calcination because it contained an organic component [22]. Wiesfeld et al. reported a high catalytic activity for a tungsten-based oxide in glucose dehydration at a low initial substrate concentration in an aqueous solution [23,24,25,26,27]. A Nb₄W₄ mixed oxide catalyst produced an HMF yield of ca. 20% in water, whereas Ta-W/Nb-W oxides delivered higher HMF yields (55–60%) in a water/THF biphasic system at an initial glucose concentration of 1.0 wt.%. The application of an Al₂O₃-TiO₂-W catalyst in NaCl-water/THF resulted in a HMF yield approaching 70% at low glucose concentrations [28]. Tungsten-based catalysts often suffer from ion leaching and even dissolution in water, which lowers selective catalytic activity while promoting the formation of humic substances with severe carbon deposition on the catalyst surface. It should be noted that Sn-Beta and Nb-BEA catalysts have achieved HMF yields of 56.9% and 81.5%, respectively, when processing 10.0 wt.% and 3.6 wt.% glucose in a NaCl-containing aqueous/organic biphasic system (water/THF or water/MIBK) where the aqueous phase pH was adjusted to 1.0 using HCl [29,30]. Oozeerally et al. investigated the catalytic performance of transition metal (Sn, Ga, and Cr)-modified and sulfated zeolites (zeolite Y, MOR, and ZSM-5) in glucose dehydration. They reported superior catalytic activity in homogeneous water-organic solvent solutions and two-phase systems relative to glucose aqueous solutions. The zeolite-based catalysts were prone to deactivation by carbon deposition that occluded the micropores [31,32,33,34,35]. Jiménez-Morales et al. prepared mesoporous aluminum-doped MCM-41 silica catalysts (mesoporous Al-MCM-41) using a solgel method with dodecylammonium chloride as the surfactant [60]. The resultant Al-MCM-41 catalyst achieved an HMF yield of 63% after 30 min in an MIBK/water biphasic system at 195 °C with a catalyst loading ratio of 1:3. In comparison, although the reaction time for 25%Nb-TaP was six times longer, the proposed system may present some advantages in terms of reaction temperature, catalyst loading, and HMF yield.

2.5. Kinetic Analysis of Glucose Dehydration over 25%Nb-TaP

The macroscopic kinetic analysis was conducted according to our previously proposed irreversible series-parallel reaction model, as shown in Scheme 1 [45,46]. Glucose is first isomerized to fructose at Lewis acid sites and then dehydrated at Brønsted acid sites to produce HMF with the simultaneous decomposition of glucose, fructose, and HMF. It is not possible to estimate all the reaction rate constants shown in Scheme 1. However, it was observed that the HMF selectivity exceeded 95.5% during the first 3 h of reaction in the water/MIBK biphasic system, when the initial glucose concentration was 1.0 wt.%. Therefore, the reaction rate constants, k_Glc→D, k_Fru→D, and k_HMF→D, can be ignored to simplify the kinetic analysis [46], and we can focus on the catalytic response of the Lewis and Brønsted acid sites.

The kinetic analysis of the irreversibly continuous glucose transformation can be expressed by Equations (1)–(3), assuming a reaction order of 1.0 with respect to C_Glc, C_Fru, C_HMF, C_L, and C_B, as shown below:

$$\mathrm{d}{C}_{\mathrm{Glc}}/\mathrm{d}t=-k_{\mathrm{Glc}\to \mathrm{Fru}}{C}_{\mathrm{Glc}}{C}_{\mathrm{L}}$$

(1)

$$\mathrm{d}{C}_{\mathrm{Fru}}/\mathrm{d}t=k_{\mathrm{Glc}\to \mathrm{Fru}}{C}_{\mathrm{Glc}}{C}_{\mathrm{L}}-k_{\mathrm{Fru}\to \mathrm{HMF}}{C}_{\mathrm{HMF}}{C}_{\mathrm{B}}$$

(2)

$$\mathrm{d}{C}_{\mathrm{HMF}}/\mathrm{d}t=k_{\mathrm{Fru}\to \mathrm{HMF}}{C}_{\mathrm{HMF}}{C}_{\mathrm{B}}$$

(3)

These equations can be integrated to give

$$C_{\mathrm{Glc}}=\exp (-k_{\mathrm{Glc}\to \mathrm{Fru}}{C}_{\mathrm{L}}t)$$

(4)

$$C_{\mathrm{Glc}}={\displaystyle \frac{k_{\mathrm{Glc}\to \mathrm{Fru}}{C}_{\mathrm{L}}}{k_{\mathrm{Glc}\to \mathrm{Fru}}{C}_{\mathrm{L}}-k_{\mathrm{Fru}\to \mathrm{HMF}}{C}_{\mathrm{B}}}}(\exp (-k_{\mathrm{Fru}\to \mathrm{HMF}}{C}_{\mathrm{B}}t)-\exp (-k_{\mathrm{Glc}\to \mathrm{Fru}}{C}_{\mathrm{L}}t)$$

(5)

$$C_{\mathrm{HMF}}=1+{\displaystyle \frac{k_{\mathrm{Fru}\to \mathrm{HMF}}{C}_{\mathrm{B}}\exp (-k_{\mathrm{Glc}\to \mathrm{Fru}}{C}_{\mathrm{L}}t)-k_{\mathrm{Glc}\to \mathrm{Fru}}{C}_{\mathrm{L}}\exp (-k_{\mathrm{Fru}\to \mathrm{HMF}}{C}_{\mathrm{B}}t)}{k_{\mathrm{Glc}\to \mathrm{Fru}}{C}_{\mathrm{L}}-k_{\mathrm{Fru}\to \mathrm{HMF}}{C}_{\mathrm{B}}}}$$

(6)

where C_Glc, C_Fru, C_HMF, C_L, and C_B represent the concentration of glucose, fructose, HMF, catalyst Lewis acid sites, and catalyst Brønsted acid sites on the catalyst, respectively [46]. The activation energy of each reaction step was estimated by using the Arrhenius plot of the reaction rate constants of each reaction step (Equation (7)).

$$\ln k_i=\ln k_{i0}-{\displaystyle \frac{E_{ia}}{RT}}$$

(7)

where k_i represents the reaction rate constants for each reaction step (i = Glc→Fru and Fru→HMF), k_i0 denotes the corresponding frequency factor, E_ia is the activation energy, T is the absolute temperature, and R represents the gas constant.

The concentrations of Lewis (C_L) and Brønsted (C_B) acid sites are 0.03 mmol g⁻¹ and 0.74 mmol·g⁻¹, respectively, based on the NH₃-TPD and pyridine-IR analysis. The reaction rate constants (k_Glc→Fru and k_Fru→HMF) at different temperatures were estimated by minimizing the sum of the squares between the experimental and calculated data using the Solver in Microsoft Excel 2021. The lines presented in Figure 8a–c were generated using the estimated k_i values according to Equations (4)–(6), which fitted the experimental data well, confirming the validity of the proposed kinetic model. It was realized that the concentration of fructose was detected as <0.05%, indicating that the reaction kinetics of fructose were in near-steady-state and the thermal decomposition was significantly suppressed, leading to enhanced HMF yields. The associated Arrhenius plots for a reaction over 25%Nb-TaP can be compared with the host tantalum phosphate catalyst in Figure 8d. The values of k_Glc→Fru for 25%Nb-TaP are much higher than k_Fru→HMF. The calculated activation energies are 112 and 78 kJ·mol⁻¹ for glucose isomerization (E_a,Glc→Fru) and fructose dehydration (E_a,Fruc→HMF), respectively, indicating that increasing the reaction temperature favors glucose isomerization [45,46]. When compared with the host tantalum phosphate catalyst, the values of k_Glc→Fru and k_Fru→HMF were decreased after doping Nb⁵⁺ ions. Moreover, the activation energy of fructose dehydration was lowered, demonstrating that doping with Nb⁵⁺ promotes fructose dehydration [46].

2.6. Deactivation and Reuse of the 25%Nb-TaP Catalyst

The deactivation and potential reuse of the 25%Nb-TaP catalyst were investigated using a 1.0 wt.% glucose solution in both water and a water/MIBK biphasic system (water/MIBK = 3/7, w/w) at 170 °C for 3 h, with a catalyst loading of 1:10 (Figure 9). Run 1 with 25%Nb-TaP was followed by Runs 2 and 3, where the catalyst was recovered by filtration and centrifugation, thoroughly washed with deionized water and ethanol, and dried at 80 °C before reuse. In Run 3, the catalytic performance had significantly declined, with glucose conversion and HMF yield decreasing by ca. 20% and 14%, respectively, indicating that only ca. 50% of the initial activity was retained. This loss of activity can be primarily attributed to deactivation caused by chemical coke retention, where it was observed that 15.7 wt.% carbon was deposited on the catalyst surface after Run 1. Moreover, we realized that the oxidation states of Nb and Ta were not changed after a reaction with XPS analyses, indicating that Nb and Ta were stable during the reactions.

Regenerating the catalyst by calcining at 600 °C for 4 h was carried out to recover the catalytic activity where the yield and selectivity of HMF obtained in the aqueous solution only decreased by ca. 2.5% compared with that obtained with the fresh catalyst (Run 4). Continuous regeneration via calcination resulted in a gradual deactivation of the catalyst, generating an HMF yield of 22.0% with a selectivity of 50.3% in the fourth use (Run 6). Inductively coupled plasma (ICP) analysis detected a significant amount of leached PO₄³⁻ ions (~35.2 ppm) with trace leaching of Ta⁵⁺ ions (<0.05 ppm) and Nb⁵⁺ ions (~3.0 ppm). These results suggest that the observed catalyst deactivation is due to ion leaching and coke retention. Catalyst deactivation was also a feature of the water/MIBK biphasic system despite the high yield and selectivity of HMF (Runs 7–11). The yield of HMF was decreased by ca. 18% in the fifth use, i.e., only 69% of the initial activity was retained (Run 11). The catalyst was more prone to deactivation in the biphasic system relative to aqueous solutions. This may be ascribed to increased catalyst–substrate interactions in the biphasic system that facilitate the loss of surface ions with a consequent decrease in catalytic activity. The 25%Nb-TaP catalyst exhibited a greater stability than the starting tantalum phosphate catalyst in both water and water/MIBK [46], as evidenced by a higher residual activity.

3. Materials and Methods

3.1. Materials

Tantalum chloride (TaCl₅, 99.99%), niobium chloride (NbCl₅, 99.9% metal bases), ammonium metavanadate (NH₄VO₃, 99.95% metal bases), zirconyl chloride hydrate (ZrOCl₂·nH₂O), cadmium nitrate tetrahydrate (Cd(NO₃)₂·4H₂O, 99.99% metal bases), zinc chloride (ZnCl₂, 99.95% metal bases), nickel chloride hexahydrate (NiCl₂·6H₂O), glucose (>99%), fructose (>99%), 5-hydroxymethylfurfural (HMF, >95%), methanol, ethanol (>99.7%), n-hexane and methyl isobutyl ketone (MIBK, GC grade) were obtained from Shanghai Macklin Biochemical Co., Ltd., Shanghai, China. Concentrated phosphoric acid (>85.0 wt.%) was purchased from Shanghai Hushi Laboratorial Co., Ltd., Shanghai, China. Polyethylene oxide (PEO, M_w = 3.5 × 10⁴) and citric acid were purchased from Sigma-Aldrich, St. Louis, MO, USA.

3.2. Synthesis of the Hierarchically Porous Tantalum Phosphate and Transition Metal-Doped Tantalum Phosphate

Tantalum phosphate and transition metal-doped tantalum phosphate monolithic materials with a co-continuous microporous structure were synthesized using a sol-gel method accompanied by phase separation using polyethylene oxide (PEO) as the phase separation inducer, according to our previously reported method [46]. Typically, for the synthesis of Nb⁵⁺-doped tantalum phosphate with a molar ratio of Nb:Ta = 0.25, 7.16 g of TaCl₅ was dissolved in 10 mL of anhydrous ethanol with vigorous stirring. The obtained solution was kept in an 80 °C oven overnight to ensure a complete reaction of tantalum chloride with ethanol, producing an ethanol solution of tantalum ethoxide (Ta-EtOH solution). The resulting colorless and transparent solution was used as the Ta⁵⁺ precursor in subsequent experiments. Then, 1.0 mL of the Ta-EtOH solution was taken into a 10 mL glass vial containing a magnetic stir bar, with 0.6 mL of 1.0 M HCl solution, 0.135 g NbCl₅, 3.0 g of citric acid, and 0.450 g of PEO being sequentially added and with the mixture being stirred at 80 °C until the solid compounds were completely dissolved. After cooling the solution to room temperature, 2.8 mL of concentrated phosphoric acid was added and vigorously stirred for 3 min, and the solution was gradually gelatinized. The wet gel was further aged for 24 h for full gelation. The obtained wet gel was subjected to successive solvent exchanges using methanol, methanol/n-hexane mixture (60/40, v/v), and n-hexane at room temperature for 48 h, respectively; then, it was vacuum-dried to obtain a monolithic xerogel. The obtained 25%Nb-TaP xerogel was calcined at 400–800 °C for activation.

For the recycling of the catalyst, the catalyst was separated from the reactant after the reaction by centrifugation. The catalyst separated from the reactant was firstly washed five times with deionized water and then five times with ethanol to remove physically adsorbed substances. After that, it was calcined at 600 °C in an air atmosphere for 4 h to eliminate chemically adsorbed organic species and reactivate the catalyst.

3.3. Structure and Surface Property of the Ion-Doped Tantalum Phosphate

Macroscopic morphology of the dried gel was observed by using scanning electron microscopy (SEM, JSM-6060S, JEOL, Ltd., Tokyo, Japan, with Pt coating). The micro- and mesopore distributions were measured with a N₂ adsorption–desorption instrument (BET, ASAP2010C, Micromeritics, Norcross, GA, USA). The organic content in the prepared samples was measured by simultaneously recorded thermogravimetry-differential thermal analysis (TG, STD Q600, TA Co, New Castle, DE, USA) at a heating rate of 5 °C·min⁻¹ with an airflow at 100 mL·min⁻¹. The specific surface area was calculated by the Brunauer–Emmett–Teller (BET) method whilst the pore-size distribution was assessed by Barett–Joyner–Halenda (BJH) method using the adsorption branch of the N₂ sorption isotherm. The crystal structure analyses were carried by using powder X-ray diffraction (XRD, D/max 2500 PC, Rigaku, Tokyo, Japan) with Cu Kα (λ = 0.154 nm) as an incident beam, with a scanning range of 10–80°. Infrared absorption was measured using a Fourier transform-infrared spectrometer (FT-IR, Nicolet iS50, Thermo Fisher, Waltham, MA, USA) in a 400–4000 cm⁻¹ spectral range. Raman scattering analysis was performed in a back-scattering configuration at room temperature using a triple monochromator at 532 nm (XpoloRA PLUS, Horiba/Jobin-Yvon, Kyoto, Japan).

The total acidity of the samples was determined by NH₃ temperature-programmed desorption (NH₃-TPD, Micromeritics AutoChem II 2920, Norcross, GA, USA), and clarification of acid sites was performed using a pyridine FT-IR with a Bruker Tensor 27 spectrometer, with a wavenumber range of 1000–2000 cm⁻¹, according to our previously reported methods, which was used to analyze the surface acid properties of host tantalum phosphate [46]. The ion leaching from the catalyst into the reactant was measured by inductively coupled plasma (ICP, Avio 550 MAX, PerkinElmer, Waltham, MA, USA). XPS analyses were performed on an AXIS ultra spectrometer (Kratos, Manchester, UK), with an Al anode (Al Kα, hν = 1486.6 eV). The measuring chamber was operated at a pressure of <5 × 10⁻⁹ Torr. The binding energies of Nb and Ta were calibrated with reference to the C1s peak at 284.5 eV.

3.4. Catalytic Performance and Component Analysis

Glucose dehydration was carried out under various reaction conditions to study the catalytic performance of the transition ion-doped tantalum phosphate catalysts, according to our previously reported work [46]. The doped tantalum phosphate catalysts were calcined at 400–800 °C and cracked before evaluation. The aqueous system reactions were performed in a 5 mL pressure-resistant glass reactor equipped with a magnetic stir bar. The glass reactor was immersed in a temperature-controlled silicone oil bath (DF-101S, Kenuo, Tianjin, China) for reaction. After the reaction, the glass reactor was immediately cooled by being immersed into a water bath to quench the reaction. Glucose dehydration in a water/MIBK biphasic system was conducted in a 25 mL reactor (YZPR-25, Yanzheng Instrument, Shanghai, China) with the reaction temperature being in the range of 150–170 °C. After the reaction, the reactor was placed in a cold water bath in order to quench the reaction as soon as possible. The reactant was centrifuged at 10,000 rpm for 10 min, removed using a syringe, and filtered into a 10 mL centrifuge tube by a 0.22 μm filter to separate the solid catalyst particles, preventing the clogging of the high-performance liquid chromatography (HPLC) column during component analysis. Aqueous samples were analyzed by using a ChromCore Sugar-10Ca column (6 μm, 7.8 × 300 mm, NanoChrom Technologies, Suzhou, China) with a refractive index detector (RI), using ultrapure water as the mobile phase at a flow rate of 0.6 mL min⁻¹. The column was kept at 80 °C in a column oven (Waters, Milford, MA, USA). The content of HMF in the MIKB phase was analyzed using a ChromCore 120 C18 column (5 μm, 4.6 × 250 mm, NanoChrom Technologies) with a UV detector at 284 nm and the mobile phase of a mixture of methanol and water (methanol/water = 75/25, v/v) at a flow rate of 1.0 mL·min⁻¹.

4. Conclusions

A water-tolerant and bifunctional niobium-doped tantalum phosphate with a co-continuous macro-porous structure was synthesized using a sol-gel method accompanied by phase separation and used to promote glucose dehydration to HMF. Doping with niobium inhibited crystallization of the host tantalum phosphate and increased surface acid site density, resulting in superior catalytic activity. Under optimum reaction conditions, an HMF productivity of 0.11 mol·h⁻¹·kg-solution⁻¹ was achieved by treating a 15.0 wt.% glucose feed in a water/MIBK biphasic system with a 25%Nb-TaP loading of 10.0 wt.%. Kinetic analysis has shown that the reaction rate constant for glucose isomerization to fructose (k_Glc→Fru) was higher than fructose dehydration (k_Fru→HMF). The associated activation energy (E_a,Glc→Fru) was also higher than E_a,Fruc→HMF, suggesting that increasing the reaction temperature favored glucose isomerization. The 25%Nb-TaP catalyst exhibited clear deactivation, mainly due to PO₄³⁻ leaching and chemical carbon deposition, which was partly reversed by calcination. The development of industrially applicable tantalum-based catalysts for glucose dehydration to HMF still faces several challenges, requiring a fine tuning of reaction kinetics and improved catalyst stability.