Phosphate‐Terminated Nb₂C MXene as Efficient Catalyst for the Transformation of Glucose Into 5‐Hydroxymethylfurfural With 3077 h⁻¹ Turnover Frequency

Abstract

The production of 5‐hydroxymethylfurfural (HMF) from glucose is a reaction of considerable interest in biomass conversion into platform molecules. This reaction suffers, however, in general from low selectivity and the generation of undesirable polymeric byproducts (humins). Herein, we describe the preparation of phosphate‐terminated Nb₂C MXene (Nb₂C‐PO₄) as a novel heterogenous water‐tolerant acid catalyst, able to afford high yields of HMF, while minimizing humin formation. The phosphate‐modified Nb₂C has been extensively characterized (X‐ray diffractometry (XRD), high resolution transmission electron microscopy (HRTEM), scanning transmission electron microscopy (STEM‐EDS), X‐ray photoelectron spectroscopy (XPS)) in order to prove the retention of the MXene structure and the presence of phosphate groups on the surface, and the acid–base properties have been assessed by NH₃‐ and CO₂‐TPD and pyridine‐DRIFTS. Apparently, the effect of phosphate is to mask less selective acid sites, while the remaining ones are more selective to promote HMF formation from glucose. This proposal is also supported by the contrasting behavior of pyridine and 2,6‐di‐tert‐butylpyridine as quenchers of acid sites. Under optimized conditions, 64% yield of HMF can be obtained at 150°C in a monophasic THF/water mixture, using a high glucose concentration (5% w/w) and a low Nb₂C‐PO₄ loading (S/C = 10 w/w), reaching a TOF of 3077 h⁻¹.

A new phosphate‐terminated Nb₂C MXene has been developed as an efficient heterogeneous catalyst to convert glucose into 5‐hydroximethylfurfural. Characterization of the Nb₂C‐PO₄ catalyst has proven the presence of phosphate and the integrity of the MXene laminar structure. In optimal conditions in monophasic THF/H₂O mixture, formation of humins is suppressed and glucose at 5% concentration is converted with turnover frequency (TOF) of 3077 h⁻¹.

1. Introduction

The shift from fossil sources to renewable biomass feedstocks as raw materials for the obtention of fuels and chemicals is a topic of great interest in research, since it represents a viable way to reach carbon neutrality and sustainability in our current economic model [1]. The largest source of usable biomass on Earth is lignocellulose [2], the main component of plant material. Lignocellulose can be obtained from crops rich in this material or, more conveniently, from agricultural and sylvicultural wastes [3]. Lignocellulose is composed of three polymers, namely cellulose, hemicellulose, and lignin, and its chemical upgrading usually starts by depolymerization into simple molecules [4]. In the case of cellulose, the largest component of lignocellulose, it can be broken down into D‐glucose, its only monomer, using acidic hydrolysis [5]. In these reaction conditions, a variety of products can be formed, among which 5‐hydroxymethylfurfural (HMF) is the most important one as it is considered a platform molecule for further transformation into diverse valuable products [3]. Therefore, there is a growing interest in developing solid catalysts and finding conditions to obtain HMF as quantitative product from glucose. HMF can be transformed via oxidation into furan dicarboxylic acid (FDCA), which is an attractive bio‐based alternative to terephthalic acid for the synthesis of bio‐based polyesters [6]. Likewise, reductive routes from HMF can also lead to diol molecules of interest to the polymer industry [7], such as bis(hydroxymethyl)furan (BHMF), bis(hydroxymethyl)tetrahydrofuran (BHMTHF), or even 1,6‐hexanediol, as well as bioderived solvents and fuels such as dimethylfuran or dimethyl tetrahydrofuran. Therefore, the obtention of HMF quantitatively from an inexpensive substrate such as lignocellulose‐derived syrups has a great potential for its implementation in biorefineries [8].

Efficient transformation of glucose into HMF is however challenging due to the numerous competing undesirable reactions that usually take place. The conversion of glucose to HMF has been proposed to occur through a two‐step process, starting with isomerization of glucose into fructose, followed by fructose dehydration into HMF [5]. These reactions are generally accepted to be catalyzed by Lewis and Brønsted acids, respectively [9, 11]. Acid sites can also promote parallel unwanted reactions such as rehydration and decomposition of HMF into levulinic acid and formic acid, and polymerization into humins [12]. In spite of the considerable effort that has been already put into catalyst design and optimization of reaction conditions in order to minimize the occurrence of side reactions and maximize HMF selectivity, additional research is still necessary to develop better performing catalyst. Although homogeneous catalysts often show high activity and selectivity, their difficult recovery from the reaction media and the associated generation of large amounts of waste during the workup make them less attractive for scale‐up application, for which heterogeneous catalysts gather most attention [13]. Thus, a plethora of solid acid catalysts have been explored in the literature, such as metal oxides, zeolites, functionalized carbon materials, or metal‐organic frameworks (MOFs), while studying at the same time the influence of solvents on the catalyst performance [14, 15]. An aqueous medium would be the intuitive choice of solvent, since it easily dissolves the reaction components and it best aligns with the principles of green chemistry. However, the use of pure water as solvent often results in poor HMF yields and low selectivity, since it favors the aforementioned HMF degradation processes via rehydration and polymerization due to the instability of HMF in water [9, 11]. Consequently, a multitude of catalytic systems have been described in recent years, in which special importance is given to both catalyst design and choice of solvent or mixture of solvents. One common choice is the use of dimethyl sulfoxide (DMSO), which minimizes side reactions by stabilizing both furanoses and the carbonyl group in HMF through solvation [16]. Remarkable results have been obtained in DMSO or its mixtures with water or 2‐propanol, using catalysts such as SnO₂ on γ‐Al₂O₃ [16] or g‐C₃N₄/γ‐Al₂O₃ composite material [17], among many others [18]. Another frequently employed solvent system is a mixture of THF/water, since the presence of THF inhibits rehydration processes [19], either as a monophasic mixture [20, 21] or saturating the aqueous fraction with NaCl to render a biphasic system [11, 22, 24], in which HMF is extracted into the prevalent organic phase as it is formed, preventing its further degradation that would occur in the aqueous phase. Other organic solvents commonly reported for biphasic mixtures include γ‐valerolactone [25], dioxane [26], or methyl isobutyl ketone (MIBK) [27], being in all cases the aqueous phase necessary to dissolve the glucose.

In regard to catalyst design, a number of factors must be considered beyond strength and amount of acid sites [5]. According to the widely accepted mechanism, the transformation of glucose into HMF starts with isomerization of glucose into fructose. This process has been described to be catalyzed by either a Lewis acid (LA) or a Brønsted base (BB), the presence of Brønsted acid (BA) sites being detrimental for the isomerization selectivity [5, 11, 22]. The subsequent fructose dehydration step to afford HMF can be catalyzed by both BA and/or LA sites [5, 10, 28], Therefore, the possibility of gathering different types of acid or basic sites, that also show the required strength and spatial distance between them [5, 10, 11], all in a single heterogeneous catalyst is critical for an optimal performance. Some successful examples have utilized sulfonic acids on solid supports, in order to combine acid and basic functionalities. Nahavandi et al. synthesized sulfonated oxidized activated carbon that contained BA groups in the form of −SO₃H, −OH, and −COOH, and BB sites from carbonyl groups, reaching 93% HMF yield at 160°C in a biphasic THF/brine solvent system [22]. Chhabra et al. prepared sulfonated g‐C₃N₄, where the BB sites were provided by the N atoms present in g‐C₃N₄ in the form of amines, and obtained HMF from glucose in 94% yield at 200°C in water [29]. Alternatively, Kirdant et al. employed a g‐C₃N₄/γ‐Al₂O₃ composite material combining BB sites from g‐C₃N₄ and LA sites from the acidic alumina, and achieved 91% HMF from glucose in a DMSO/water mixture at 140°C [17]. In a different approach, Sharma et al. treated sporopollenin with H₃PO₄ to obtain a phosphorylated organic support that yielded 92% HMF from glucose at 180°C using water as solvent, and described a mechanism involving exclusively BA sites [30].

In this work, we report the use of MXenes as innovative heterogeneous acid catalysts for HMF production from glucose, since they can carry different types of acid/base active sites on their surface. MXenes are novel 2D nanomaterials described for the first time in 2011 [31], that combine one atom thick layer of an early transition metal (M = Ti, V, Nb, Mo, etc.) and one atom thick layer of a non‐metal (X = C, N) alternated in a defined layered structure (general formula M _n X_n−1). Although still scarcely used in catalysis, the existing limited information has demonstrated potential to exhibit outstanding properties as catalysts [32, 36]. In addition to good thermal stability, good dispersibility and reasonable chemical stability in aqueous media, these materials present tunable surface terminations [37, 38]. The metal surface of an MXene will often be at least partially oxidized after exposure to air, resulting in M‐OH and M = O groups that can act as BA and BB sites, respectively, while accessible metal centers can act as LA sites in catalysis. Particularly, Nb₂C MXene [39] caught our attention due to the presence on its surface of NbO₂ units that could exhibit water tolerant Lewis acid (WTLA) behavior based on its similarity with NbO₄ structures reported as WTLA. This could render Nb₂C an ideal catalyst for glucose dehydration into HMF in aqueous media [28]. Besides, the concentration of BA and BB sites can be tuned by replacing them by inactive surface groups, such as halides (Cl, Br) [38], and the amount of BA sites can be controlled via esterification with phosphate groups [20, 23, 28, 30, 40], thus improving the reaction selectivity toward HMF by minimizing side reactions.

Based on these premises, herein, we have prepared phosphate‐terminated Nb₂C MXene (Nb₂C‐PO₄) and found that this modified MXene exhibits a high activity and selectivity as heterogeneous catalyst for the transformation of glucose into HMF in a monophasic THF‐H₂O medium. Under optimized conditions, the formation of humins is minimized, despite employing a high glucose concentration of 5% (w/w), and a substrate/catalyst (w/w) ratio of 10 affords a good HMF yield of 64% after 22 h, while most similar systems described require a mass of catalyst similar to or even larger than the mass of glucose substrate in order to reach similar activity (Table S1). Exhaustive characterization shows that the high activity and selectivity of Nb₂C‐PO₄ MXene catalyst are not only dependant on their acid–base properties, but there are also chemical and structural parameters playing a positive role in the catalytic activity.

2. Results and Discussion

2.1. Preparation and Characterization of MXene Catalyst

Nb₂C MXene was prepared from commercial Nb₂AlC MAX phase by etching Al with a solution of NaBF₄ and concentrated HCl, as previously described [39], and fully characterized (see ESI). Using this MXene sample as starting material, the Nb₂C‐PO₄ catalyst was obtained by treating at room temperature this Nb₂C sample with 1 M aqueous H₃PO₄ solution [28]. The structure and composition of the resulting Nb₂C‐PO₄ product were ascertained via several techniques. The XRD pattern of Nb₂C‐PO₄ shows the presence of the peaks at 6.7°, 40.6°, 58.8° and 74.0° characteristic of Nb₂C MXene (Figure 1) [41]. The (002) peak of Nb₂C‐PO₄ is slightly shifted with respect to the 7.4° value of the parent Nb₂C and significantly broadened, indicating a larger interlayer distance of 13.17 Å, while the rest of the MXene peaks do not show relevant differences before and after modification of Nb₂C with phosphate (Figure 1).

FIGURE 1.

XRD patterns of Nb₂C (bottom) and Nb₂C‐PO₄ (top) samples. Peaks assigned to residual MAX phase are marked with an asterisk.

The surface structure and elemental composition of Nb₂C‐PO₄ MXene were studied by high‐resolution scanning transmission electron microscopy coupled to energy‐dispersive X‐ray spectroscopy (HRSTEM‐EDS) in order to confirm the presence and distribution of phosphorus on the surface, as well as the overall integrity of the MXene structure. The images show high crystallinity of the material (Figure 2A), with slightly darker spots that could be sign of surface defects or the presence of phosphate groups, due to the differences in contrast. The laminar structure of the MXene can be clearly appreciated as well (Figure 2B,C), confirming that the acid treatment of Nb₂C with H₃PO₄ does not appreciably degrade the MXene structure. Analysis of the elemental composition of the surface by EDS mapping reveals homogeneous distribution of phosphorus (Figure 2F), with a concentration ca 2 wt%, as determined by spectral analyses.

FIGURE 2.

HRSTEM images of Nb₂C‐PO₄ showing: (A) high surface crystallinity, (B) and (C) laminar structure, and (D–F) elemental mapping from EDS analysis.

Survey of XPS analyses of Nb₂C and Nb₂C‐PO₄ confirmed the presence of all the expected elements, accompanied by a minor percentage of residual Al remaining from the Nb₂AlC precursor. Figure 3 shows the XPS peaks for the elements found in Nb₂C‐PO₄ as well as the best deconvolution to the different individual components. Table S2 in supporting information collects the percentages of each element, together with their distribution in different chemical environments for each XPS peak obtained from the best deconvolution to individual components. Importantly, the XPS P 2p core level presents a single component appearing at a binding energy of 133.3 eV corresponding to a phosphate group, as expected for Nb₂C‐PO₄. The XPS Nb 3d peak appears as a doublet at 206.8 and 209.5 eV due the spin orbit coupling attributable to Nb(V)=O and Nb‐O‐P, accompanied by a smaller component due to Nb–C observed at 203.8 eV. The XPS C 1s signal has three components due to C–Nb (281.5 eV), C–C (284.5 eV) and C–O (288.0 eV), while the F 1s has also three types assigned to F–Nb (684.3 eV), F–Al (686.5 eV), and F–P–O (688.0 eV). Finally, XPS O 1s spectrum has also three components due to O–Nb (529.8 eV), Nb–O–P (531.1 eV), and H₂O (533.5 eV).

FIGURE 3.

Experimental high‐resolution XPS peaks of Nb₂C‐PO₄ together with their best fitting to individual components whose assignment and binding energy are indicated in the plots.

Acid–base properties of Nb₂C and Nb₂C‐PO₄ MXenes were studied by means of NH₃ and CO₂ temperature programmed desorption (TPD) in the 50–450°C temperature range. Adsorption of NH₃ gas on the surface of MXenes generally occurs on ‐OH terminations (BA sites) as well as on exposed Nb metal or surface defects, these acting as LA sites. Both Nb₂C and Nb₂C‐PO₄ MXenes showed only one peak of NH₃ desorption at 111°C and 91°C, respectively (Figures S10 and S11), which correspond to weak acid sites, as previously established in the literature [42]. In addition, the amount of acidic sites quantified from these results is very low (Table 1) for both materials, but especially for Nb₂C‐PO₄, which yields only 5.2 µmol/g of desorbed NH₃ (Table 1). It is likely that the low levels of acidity obtained in these measurements are due to the agglomeration of the sheets in the solid state, while dispersion in a solvent favoring separation of MXene sheets could result in density of exposed acidic sites somewhat larger than the value measured for solid powders [41]. Regardless of the low absolute values of acid sites, comparison of the two materials shows a nearly 9‐fold lower number of acidic sites in Nb₂C‐PO₄ respect to Nb₂C, suggesting that a significant ratio of acidic sites is blocked by the formation of surface phosphoesters [28] while the remaining ones have the adequate water‐tolerant Lewis nature and acid strength to promote HMF formation from glucose with higher selectivity. In the case of CO₂‐TPD, no gas uptake was detected upon flowing a diluted CO₂ stream at 40°C, and desorption experiments showed only the release of water or pre‐adsorbed atmospheric gases, in any case showing no relevant difference between Nb₂C and Nb₂C‐PO₄. Therefore, we concluded that no basic sites are present on the MXenes surface and the catalytic activity for glucose isomerization into glucose must arise from acid sites.

TABLE 1.

NH₃‐TPD measurements on acidity of Nb₂C and Nb₂C‐PO₄ MXenes.

MXene	T, °C	NH3 uptake/desorption, µmol/g
Nb2C	111	45.6
Nb2C‐PO4	91	5.2

In order to distinguish the Brønsted and Lewis character of the acidic sites on Nb₂C and Nb₂C‐PO₄ MXenes, pyridine‐adsorbed diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) measurements were carried out, and the peaks were assigned to either BA or LA sites based on previous studies in the literature [42]. Consistently with the results from NH₃‐TPD experiments, the intensities of all the peaks arising from the interaction of pyridine with acidic sites are clearly decreased upon heating the samples above 100°C (Figure 4). Nevertheless, significant differences can be observed between the IR spectra of adsorbed pyridine for Nb₂C (Figure 4A) and Nb₂C‐PO₄ (Figure 4B). Overall, Nb₂C shows more peaks than Nb₂C‐PO₄, and particularly LA site peaks, such as those at 1508 and 1457 cm⁻¹, show a much lower intensity in the Nb₂C‐PO₄ compared to the parent Nb₂C. However, the LA site peak at 1687 cm⁻¹ appears slightly stronger in Nb₂C‐PO₄, and it remains detectable at 400°C, proving its stability. On the other hand, BA site peaks at 1558 and 1542 cm⁻¹ in Nb₂C‐PO₄ spectrum maintain a higher intensity at higher temperatures, showing better stability compared to the same sites in Nb₂C. These results suggest that the chemical environment on the new catalytic sites generated after the formation of surface phosphoesters has an influence on catalytic performance aside from the type and strength of the acid sites.

FIGURE 4.

Adsorbed pyridine DRIFTS spectra of (A) Nb₂C and (B) Nb₂C‐PO₄.

2.2. Catalytic Experiments

As a preliminary screening, Nb₂C, Nb₂C‐PO₄, as well as other MXenes (Ti₃C₂, Mo₂C) were tested as catalyst for the dehydration of glucose into HMF in aqueous medium under similar reaction conditions as those employed by the team of Hara using Nb₂O₅·nH₂O as catalyst [28] (Table 2). However, a 10‐fold lower substrate/catalyst (S/C) ratio was employed in our reactions, since it was observed that an excessive amount of MXene catalyst resulted in a very low carbon balance, presumably due to adsorption of the glucose or the organic products on the MXene surface.

TABLE 2.

Initial screening of different MXenes as catalysts for the transformation of glucose into HMF. ^a

Entry	MXene	t b , h	Conversion, %	Yield HMF, %	Yield fructose, %	Yield others c , %	C balance d , %
1	Nb2C	4	77	16	4	8	60
2		22	83	7	2	33	60
3	Nb2C‐PO4	4	49	23	6	—	80
4		22	75	13	1	16	55
5	Ti3C2	4	49	0	7	—	68
6		22	89	0	3	—	14
7	Mo2C	4	82	10	12	13	53
8		22	93	3	5	32	44
9	None	22	0	—	—	—	100

^a Reaction conditions: MXene (40 mg), glucose (40 mg), water (4 mL), 150°C.

^b Reaction time after reaching 150°C.

^c Other reaction products detected by HPLC analyses include levulinic acid and formic acid.

^d Carbon balance calculated as the sum of molar yields of the C‐containing compounds detected by HPLC analyses.

The highest conversions were obtained with Mo₂C, reaching 82% after 4 h and 93% after 22 h (Table 2, entries 7 and 8, respectively), however, the HMF yield as well as the C balance were very low, despite reaching a fructose yield of 22% after 1 h of reaction (Figure S18). It must be noted that the missing C moles in all C balances shown are assumed to be in the form of insoluble humins, that are filtered out during the preparation of analytical samples and become deposited or adsorbed on the MXene catalyst. Interestingly, Nb₂C‐PO₄ afforded the lowest conversions of the series at 4 and 22 h, while at the same time produced the highest HMF yields and C balances (Table 2, entries 3–4) among the catalysts tested, proving to have the best selectivity toward the desired reaction. Consequently Nb₂C‐PO₄ was chosen as catalyst for the rest of the optimization process. It is worth noting that Ti₃C₂, the most commonly studied MXene, yielded negligible or undetectable amounts of HMF, although it did convert a large part of the glucose substrate, presumably into humins (Table 2, entries 5–6). Finally, it was confirmed that these reaction conditions did not result in any reactivity or degradation of glucose in the absence of a catalyst (Table 2, entry 9).

A low concentration of glucose (1%, w/w) was used in the initial catalyst screening (Table 2), since a higher concentration is generally presumed to increase side reactions, especially the formation of polymeric humins [43], but for the remaining optimization experiments, the concentration was raised to 5% (w/w) and the amount of catalyst decreased to a S/C ratio of 10, in an attempt to improve the reaction efficiency. To our delight, formation of humins appeared to be much suppressed under these conditions after 22 h of reactions, though the conversion of glucose was lower (Table 3, entry 1), as it is expected after a S/C increase. In these reaction conditions, several mixtures of water and organic solvents were tested, using Nb₂C‐PO₄ as catalyst (Table 3). The best results were obtained in a THF/H₂O 3:1 (v/v) mixture, that remained monophasic at all times. In addition to the excellent 64% yield of HMF, all other products detected via HPLC, including fructose, levulinic and formic acid, and the unreacted glucose, accounted for a quantitative molar C balance (Table 3, entry 2), showing negligible formation of humins. The kinetic profile of this reaction was plotted, measuring samples at regular intervals, and the HMF yield showed a steady increase during the whole duration of the reaction (Figure 5). It can be observed as well that the C balance starts at 54% and reaches 99% by the end of the reaction (22 h), which can be explained by an incomplete dissolution of the glucose in the THF/H₂O 3:1 solvent mixture at the beginning of the reaction, though it could be also due to the formation of oligomers that later decompose into HMF [5]. Despite showing some of the best results in the literature, in our hands the DMSO/H₂O solvent system did not show an outstanding performance when using Nb₂C‐PO₄ as catalyst (Table 3, entry 4), although it did improve significantly glucose conversion and HMF yield when compared to the use of H₂O exclusively as solvent (Table 3, entry 1). It is known that DMSO is an intercalant agent of MXene sheets [44] and this property can likely influence the interaction of acid sites with glucose. Even though the performance of the 2‐propanol/H₂O 1:1 (v/v) solvent mixture was not remarkable for the conversion of glucose into HMF (Table 3, entry 3), a small amount of sorbitol, up to 7% yield at 2 h (Figure S20) was detected during the whole course of the reaction, demonstrating a certain capacity of Nb₂C‐PO₄ to act as a transfer hydrogenation catalyst. Unfortunately, further attempts to exploit this catalytic route did not result in a significant improvement of the hydrogenative activity.

TABLE 3.

Testing of different organic solvent mixtures for the transformation of glucose into HMF using Nb₂C‐PO₄ as catalyst. ^a

Entry	Solvent, v/v ratio	Conversion, %	Yield HMF, %	Yield fructose, %	Yield others b , %	C balance c , %
1	H2O	47	16	6	13	99
2	THF/H2O, 3:1	84	64	9	10	99
3	2‐propanol/H2O, 1:1	70	26	6	7 d	77
4	DMSO/H2O, 1:1	71	40	4	11	83

^a Reaction conditions: Nb₂C‐PO₄ (20 mg), glucose (200 mg), solvents (4 mL), 150°C, 2 MPa N₂, 22 h.

^b Other reaction products detected by HPLC analyses include levulinic acid and formic acid.

^c Carbon balance calculated as the sum of molar yields of the C‐containing compounds detected by HPLC analyses.

^d Includes 2% yield of sorbitol.

FIGURE 5.

Temporal profiles of glucose conversion into HMF. Reaction conditions: Nb₂C‐PO₄ MXene (20 mg), glucose (200 mg), H₂O (1 mL), THF (3 mL), 150ºC, 2 MPa N₂ (Table 3, entry 2).

In order to optimize the THF/H₂O solvent system, different solvent ratios were tested (Table 4). The 3:1 THF/H₂O ratio (Table 4, entry 3) still demonstrated to be the best in terms of both HMF yield (64%) and C balance (99%) among the systems tested. A lower proportion of THF (1:1) still improved the HMF yield and glucose conversion respect to the values obtained in pure water, but the 33% yield (Table 4, entry 2) was far from the optimal value. Using a much larger ratio of THF (39:1) coincidentally resulted in very similar HMF yield and C balance (Table 4, entry 4) as in the 1:1 mixture, however in this case, it must be noted that the glucose could not be totally dissolved in the 39:1 THF/H₂O mixture at room temperature, and it dissolved partially later inside the reaction autoclave at 150°C, as inferred from the reaction kinetic profile (Figure S23). When pure THF was used, the glucose remained undissolved even at 150°C, making impossible to follow the reaction progress by analysis of the liquid phase, and once the reaction was ended (22 h) an insoluble caramel‐like substance was retrieved from the crude. Analysis of the supernatant revealed the usual reaction products (glucose, HMF, fructose, formic acid), recovering 31 mol% of the C initially present (Table 4, entry 5).

TABLE 4.

Activity data for the transformation of glucose into HMF using Nb₂C‐PO₄ as catalyst using different THF/water volume ratios. ^a

Entry	THF/H2O, v/v ratio	Conversion, %	Yield HMF, %	Yield fructose, %	Yield others b , %	C balance c , %
1	H2O	47	16	6	13	99
2	1:1	77	33	11	19	86
3	3:1	84	64	9	10	99
4	39:1	86	33	3	31	81
5	THF	91	4	1	17	31

^a Reaction conditions: Nb₂C‐PO₄ (20 mg), glucose (200 mg), solvent (4 mL), 150°C, 2 MPa N₂, 22 h.

^b Other reaction products detected by HPLC analyses include levulinic acid and formic acid.

^c Carbon balance calculated as the sum of molar yields of the C‐containing compounds detected by HPLC analyses.

As previously described in the literature, the surface groups of the as‐prepared Nb₂C MXene consist predominantly in F and ‐O‐ or ‐OH terminal groups [39]. Therefore, the influence of different surface terminations on their catalytic performance was studied. For this purpose, Cl‐ and Br‐terminated Nb₂C MXenes were prepared by treating Nb₂C with concentrated HCl or HBr, respectively, and their structure and composition were characterized by XRD, TEM, NH₃‐TPD and XPS (see ESI). Particularly, XPS analysis (Figures S25 and S26) shows the presence of Cl or Br replacing F in the composition of Nb₂C. The results of catalytic tests under the optimized conditions (Table 5, entries 3 and 4) show similar results to those of the Nb₂C sample etched with NaBF₄/HCl (Table 5, entry 1) regarding glucose conversion and HMF yield, staying significantly lower than the 64% yield obtained with Nb₂C‐PO₄ (Table 5, entry 2). From the number of acidic sites obtained by NH₃‐TPD measurements (Table 1 and Figure S12), turnover frequency (TOF) values were calculated for all Nb₂C‐T _x in the present conditions, resulting in a value of 3077 h⁻¹ for Nb₂C‐PO₄, which is an order of magnitude higher than the other MXenes (Table 5). The TOF values for Nb₂C‐Cl and Nb₂C‐Br are also significantly higher than Nb₂C, since they afford similar HMF yields while possessing fewer acid sites. Interestingly, Nb₂C‐Cl and Nb₂C‐Br also show a shift in selectivity, yielding larger amounts of formic acid, even at short reaction times (Figures S27 and S28), which suggest that degradation pathways via hydrolysis are favored by these catalysts. It becomes clear then that the strength and density of acid sites on the Nb₂C‐T _x surface are not the main properties of these catalysts that determine conversion and selectivity in the transformation of glucose into HMF under the conditions studied here, but there must be other features on the catalytic sites that define reaction rates and pathways. In particular, these results suggest that when using Nb₂C‐PO₄ the substrates (glucose and/or fructose) rapidly desorb from the catalytic sites, minimizing the occurrence of side reactions and affording a high TOF value.

TABLE 5.

Comparison of different surface groups on Nb₂C MXene for the transformation of glucose into HMF. ^a

Entry	MXene	Conversion, %	Yield HMF, %	Yield fructose, %	Yield others b , %	C balance c , %	TOF, h−1
1	Nb2C	97	47	3	4	57	258
2	Nb2C‐PO4	84	64	9	10	99	3077
3	Nb2C‐Cl	93	38	4	18	67	880
4	Nb2C‐Br	95	44	3	24	76	539

^a Reaction conditions: MXene (20 mg), glucose (200 mg), THF (3 mL), H₂O (1 mL), 150°C, 2 MPa N₂, 22 h.

^b Other reaction products detected by HPLC analyses include levulinic acid and formic acid.

^c Carbon balance calculated as the sum of molar yields of the C‐containing compounds detected by HPLC analyses.

Table S1 provides a summary of the state‐of‐the‐art catalysts reported in the literature in comparison with the results herein achieved for glucose conversion to HMF. When making this comparison, caution has to be taken considering the various glucose concentration, glucose‐to‐catalyst ratio, solvent, reaction temperature that have been employed. However, as Table S1 shows, the present Nb₂C‐PO₄ is among those reported with the highest glucose concentration and lower temperature, affording a comparatively high TOF value and HMF productivity of 1.60 mmol·g⁻¹·h⁻¹.

In view of these results, it became relevant to test the actual role of the acid sites on Nb₂C‐PO₄ during catalysis. Thus, the reaction in optimum conditions (Table 5, entry 2) using Nb₂C‐PO₄ as catalyst was reproduced, but adding pyridine (10 molar equivalents with respect to the acid sites present in the Nb₂C‐PO₄ catalyst) before the reaction started, which is expected to neutralize accessible Lewis and Brönsted acid sites due to its basic character. The progress of the reaction was followed, and 94% of glucose was indeed converted after 22 h of reaction (Figure S29). Isomerization of glucose into fructose is expected to occur, since it can be catalyzed by basic centers, however, the dehydration into HMF was also observed, yielding 23% of HMF after 22 h (Figure S29). There is thus a loss of selectivity toward HMF along with an increase in glucose conversion. This failure to quench acid sites by pyridine could be due to the preferential interaction of pyridine with MXene surface in the reaction conditions, resulting in an intercalation between the layers, rather than acid site neutralization. This same experiment was also carried out using encumbered 2,6‐di‐tert‐butylpyridine instead of pyridine as selective Brönsted acid site quencher. The two ortho tert‐butyl groups around the basic pyridine N atom ensure that only H⁺ can access to the N lone electron pair. The purpose was to probe a distinctive poisoning effect of Brönsted sites. The result shows 62% conversion of glucose and 43% yield of HMF after 22 h of reaction, with 96% carbon balance (Figure S30), indicating that indeed, by scavenging Brönsted acid, Lewis acid sites can still form HMF while minimizing the formation of unwanted humins. Nevertheless, the HMF yield is significantly lower in the presence of 2,6‐di‐tert‐butylpyridine (Figure 5), indicating that the cooperation of Brönsted and Lewis sites is advantageous for achieving an optimum performance of the catalytic system.

In order to probe the effect of a higher reaction temperature, the optimized reaction conditions for the dehydration of glucose into HMF with Nb₂C‐PO₄ MXene were reproduced at 180°C and the course of the reaction followed periodically (Figure 6). As expected, the yield of HMF rapidly increases at the start of the reaction, reaching a maximum of 59% after 3 h. However, after this point, the yield rapidly decreases, along with the C molar balance, as the reaction continues, thus providing slightly inferior results to those obtained at 150°C after a longer reaction time. This temporal profile indicates that HMF is unstable at 180°C.

FIGURE 6.

Kinetic profiles of glucose conversion into HMF at high temperatures. Reaction conditions: Nb₂C‐PO₄ MXene (20 mg), glucose (200 mg), H₂O (1 mL), THF (3 mL), 180°C, 2 MPa N₂.

The use of fructose as substrate was also explored, since it can be directly dehydrated by BA catalysts into HMF, saving the first isomerization step from glucose, and therefore, resulting in a faster one‐step process. In our optimal reaction conditions using Nb₂C‐PO₄ MXene as catalyst, fructose was rapidly converted, yielding a maximum 50% of HMF after 7 h of reaction (Figure 7). At this point of the reaction, however, only 8% of fructose remains, showing low selectivity to HMF and a relatively quick degradation, as evidenced by the high concentration of formic acid (Figure 7).

FIGURE 7.

Kinetic profiles of fructose conversion into HMF. Reaction conditions: Nb₂C‐PO₄ MXene (20 mg), fructose (200 mg), H₂O (1 mL), THF (3 mL), 150°C, 2 MPa N₂.

The stability of Nb₂C‐PO₄ catalyst was tested by running a series of consecutive reaction cycles (Figure 8). A loss between 14% and 8% HMF yield can be observed after each of five cycles, while glucose conversion only decreases slightly during the first 3 cycles and increases in the fourth and fifth ones (Figure 8A). However, it is important to note that the small mass of catalyst lost due to incomplete recovery after each run was not accounted for in this reusability study, and the amount of solvent and glucose was not adjusted to maintain the substrate/catalyst ratio constant or the dilution of the catalyst. This experimental procedure was preferred to avoid any significant difference in the magnetic stirring of the suspension during the reaction. However, if glucose conversion and HMF yield for each catalytic cycle is normalized per mass of Nb₂C‐PO₄ catalyst present in each run, the decrease of HMF yield in each cycle is less pronounced, as shown in Figure 8B, while the same Figure shows an evident increase of glucose conversion per catalyst mass upon reuse. This latter fact, along with the drastic loss in carbon balance after the third run (Figure 8A), suggests an alteration of the catalytic sites that leads to a loss of selectivity toward HMF, probably accompanied by an increase in the number and/or strength of acid sites on the MXene.

FIGURE 8.

Reusability tests of Nb₂C‐PO₄ catalyst. (A) Total yields and conversions obtained after each run; (B) Yields and conversions normalized per mg of Nb₂C‐PO₄ catalyst. Reaction conditions: Nb₂C‐PO₄ MXene (20 mg), glucose (200 mg), H₂O (1 mL), THF (3 mL), 150°C, 2 MPa N₂, 22 h. Carbon balance values are accounted for the yields of HMF, fructose and formic acid, and the remaining glucose.

Additionally, a series of experiments aimed to rule out the assistance of dissolved species leached from the Nb₂C‐PO₄ during the catalytic reaction were carried out. A hot filtration experiment was performed at 6 h of reaction (under optimized conditions), showing no relevant increase of HMF yield after removal of the solid Nb₂C‐PO₄ catalyst (Figure S31). However, glucose and fructose were still depleted from the reaction medium (Figure S31), which was confirmed to happen also under these reaction conditions (150°C) in the presence of formic acid that is a by‐product generated during the reaction. Besides, the reaction was performed in the presence of soluble species that could potentially leach from the Nb₂C‐PO₄ catalyst, such as H₃PO₄ and Nb salts. Thus, two different experiments containing either H₃PO₄ or Nb(V) oxalate with concentrations equivalent to the active sites present in fresh Nb₂C‐PO₄ catalyst (0.016 µmol) were carried out under optimal reaction conditions. These reactions did yield HMF after 22 h of reaction, but in much lower amount (9% and 14% HMF yield, in the presence of H₃PO₄ and Nb(V) oxalate, respectively), while degradation of glucose into formic acid and humins still took place (Figures S32 and S33). These results strongly suggest that homogeneous species do not play a relevant role in the catalytic performance of Nb₂C‐PO₄ promoting the transformation of glucose into HMF.

To gain information about the changes occurring in Nb₂C‐PO₄ during the reaction, the sample submitted to five consecutive runs was analyzed by XRD, TEM and XPS in order to ascertain its structural integrity and variations in the composition or nature of the acid sites. Powder XRD pattern (Figure 9A) did not show the characteristic peak of MXenes at 6–7°. However, we already mentioned that this interlaminar diffraction peak was already very broad and ill‐defined in Nb₂C‐PO₄. It can be expected that the exhaustive use in the reaction media will render this band even broader and less intense due to an even wider distribution of interlaminar distance of the sheets in the used sample. Nevertheless, crystallinity of the Nb₂C‐PO₄ after five consecutive runs was evident in the HRTEM images that showed a highly crystalline material (Figure 9B). Crystallinity was confirmed by selected area electron diffraction (SAED) pattern (inset in Figure 9B). The laminar structure of the MXene is also retained, as confirmed by STEM images (Figure 9C), and EDS mapping analysis shows that P is still present on the Nb₂C surface, along F and O (Figure 9C).

FIGURE 9.

(A) XRD patterns of Nb₂C‐PO₄ pristine (bottom, blue) and recovered after 5 consecutive runs of catalysis (top, red). (B) HRTEM and (C) STEM‐EDS mappings of Nb₂C‐PO₄ recovered after 5 runs of catalysis. The inset of part B corresponds to the SAED pattern of this image.

XPS analyses of Nb₂C‐PO₄ after extensive use as catalyst provided important information about the structural stability of the MXene under the reaction conditions and the main deactivation process. Figure 10 shows the high‐resolution XPS data recorded, as well as the deconvolution to individual components for each element. Thus, the Nb₂C‐PO₄ sample used in five consecutive runs was studied by XPS, observing the presence of all the elements with similar proportions as in the fresh material, except for the C 1s whose proportion increases from 28% to 46.7% from the fresh to the used sample due to the remarkable increase of the C‐O and C = O components. In accordance, also the intensity of the O 1s peak in XPS increases somewhat from 17.0% to 20.0%. Importantly, the P 2p peak was also present in the five‐times used Nb₂C‐PO₄ sample, although with a decrease from 0.8% to 0.4%. A summary of the XPS data for all the elements, indicating their distribution and proportions is also provided in Table S2. Therefore, a reasonable proposal to understand the gradual deactivation observed in Figure 8 is poisoning of the MXene surface by increasing deposition of glucose‐derived humins as the main deactivation mechanism due to the notable increase in the C peak, as well as some partial loss of phosphate. In this regard, it should be noted that XPS provides the relative percentages of the elements and not the absolute amount. Thus, the fact that C atomic percentage almost double accompanied by some O atomic percentage increase due to humin deposition must be necessarily reflected in a decrease in the percentage of the other elements without being due to their loss.

FIGURE 10.

Experimental high‐resolution XPS data recorded for five‐times used Nb₂C‐PO₄ and the best deconvolution for each element to individual components as indicated in plots.

3. Conclusion

Via a simple treatment of Nb₂C MXene with 1 M aqueous H₃PO₄, in this work we have obtained a phosphate‐functionalized MXene, Nb₂C‐PO₄, that has been thoroughly characterized, proving to maintain the laminar structure and crystallinity (XRD, HRTEM), while incorporating phosphorus‐based surface terminations (STEM‐EDS, XPS). This material has been tested as an heterogenous acid catalyst for the transformation of glucose into HMF, exhibiting good yield (64%) in a 3:1 v/v mixture of THF and water, at a moderate temperature of 150°C, in the presence of a high glucose concentration (5% w/w), and minimizing the formation of humins. The exceptional selectivity of Nb₂C‐PO₄ toward fructose and HMF has proven to not be uniquely dependant on the density and strength of acid sites, as demonstrated by NH₃‐TPD and pyridine‐adsorbed DRIFS analyses, as well as catalytic experiments in the presence of bases (pyridine and 2,6‐di‐tertbutylpyridine), and therefore, the chemical environment on the catalytic sites has a critical role in catalytic performance. In this way, even though the total number of acid sites diminish upon PO₄ anchoring on the Nb2C surface, the remaining water tolerant acid sites are more efficient and selective toward HMF formation from glucose. Reusability tests show a small decrease in HMF after 5 recycles, while the glucose conversion is increased, thus resulting in a loss of selectivity. Analysis of the recovered catalyst shows that the structure and composition of the Nb₂C‐PO₄ MXene are maintained. Next steps are the scaling up of the catalyst to perform studies with larger batches and the use of real biomass feedstocks.

Funding

This work was supported by Agencia Estatal de Investigación (CEX‐2021‐01230‐S and PID2024‐161014 NB‐I00), Generalitat Valenciana (CIPROM2024‐071) and European Commission (ERC Adv. Grant 101141466 DISCOVERY).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

The authors have cited additional references within the Supporting Information [45, 46].

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.