Recently, research efforts have been devoted to the production of biofuels and biomass-derived chemicals from nonedible lignocellulosic biomass owing to the increasing awareness of energy exhaustion and environmental concerns [1, 2]. Furfural (FUR), which is mainly produced from the acidic hydrolysis of hemicellulose and accounts for 25%–35% of the lignocellulosic biomass, has been selected as one of the top 30 biomass-derived platform chemicals and employed as the feedstock for the sustainable production of biofuels and value-added chemicals [3-6]. The catalytic hydrogenation of furfural has been extensively investigated, which can be transformed to furfural alcohol (FA) and tetrahydrofurfuryl alcohol (THFA) [7-10]. Hydrogenolysis may also occur during the hydrogenation process which can produce 2-methylfuran, 2-methyltetrahydrofuran, cyclopentanone, cyclopentanol and polyols, such as 1, 5-pentanediol and 1, 2-pentanediol [11-14].

THFA is widely used as green solvent, for the synthesis of special chemicals, such as dihydropyran, and has also been proposed for use as a biofuel or as a fuel additive because of its similar properties to kerosene [4, 15]. THFA is usually produced by the hydrogenation of FA derived from FUR, and it can be obtained by the direct hydrogenation of furfural in the presence of noble or non-noble metals [16-19]. The production of THFA from furfural has been investigated extensively. Biradar et al. [9] reported an effective complete hydrogenation of furfural using a 3% Pd/MFI catalyst and achieved the highest conversion and THFA selectivity in the range of 93%-100% and 67%-95%, respectively, under optimized conditions of 220 ℃ and 3.5 MPa. Nakagawa et al. [18] prepared a Ni/SiO₂ catalyst with Ni particle size < 4 nm and achieved a maximum THFA yield of 94% for the gas-phase hydrogenation of furfural. They found that the conversion of FA intermediate to THFA was strongly structure sensitive and the turnover frequency (TOF) decreased with increasing metal particle size. The same group also obtained a 94% yield of THFA catalyzed by a large amount of Pd-Ir/SiO₂ (Pd/Ir = 1) bimetallic catalysts in the liquid-phase process and suggested that a high hydrogen pressure (8 MPa) and low reaction temperature (2 ℃) were useful to suppress side reactions [20]. However, harsh conditions, such as high hydrogen pressure and high reaction temperature, are usually required for the conversion of FUR to THFA. The design of novel and environmentally-friendly catalysts that can achieve a high THFA selectivity under mild conditions in a green process is of great importance.

Metal organic frameworks (MOFs), which have emerged as a new class of porous materials with diverse properties such as high surface area, permanent porosity and easy functionalization by post-synthetic modification or direct synthesis, have been widely studied and applied in catalysis, in particular, biomass catalysis [21-24]. For example, Zeolitic imidazolate frameworks (ZIFs) were used for the transformation of sugars to lactic acid derivatives with a high conversion and yield [25]. A MOF-based polyoxometalate [Cu-BTC][HPM] showed good catalytic activity in the conversion of 5-hydroxymethylfurfural (5-HMF) [26]. The metal nanoparticles supported on Zr-MOFs were proven to be highly efficient catalysts for biomass refining [27, 28]. MIL-101, one of the most stable MOF structures, possesses a high surface area, large porosity, numerous coordinatively unsaturated metal sites and can be subjected to diverse functionalization or guest species encapsulation, and it has been widely used for biomass catalysis [29-32]. The sulfonic-acid-functionalized MIL-101(Cr) [MIL-101(Cr)-SO₃H] was investigated as a solid acid for the catalytic conversion of glucose to 5-HMF [33]. Noble metal nanoparticles, such as those containing Pd or Ru, incorporated within MIL-101 or organic-functionalized (–SO₃H, –NH₂) MIL-101 exhibited a high activity and selectivity in the hydrodeoxygenation or selective hydrogenation of biomass compounds [31, 34, 35]. It has been demonstrated that the presence of free amine groups in the MOF plays a key role on the formation of uniform, well-dispersed and leaching resistant metal nanoparticles within the MOF host. In addition, the nitrogen-containing support may have an effect on the selectivity towards the target products in hydrogenation reactions [36, 37].

Herein, we reported the direct synthesis of amine-functionalized MIL-101(Cr) [MIL-101(Cr)-NH₂], which exhibits excellent stability to moisture and acid compared with Fe-, Al-and V-MIL-101. The palladium nanoparticles were loaded into the MOF matrix through a direct anionic exchange approach followed by hydrogen reduction [38, 39]. The resulting Pd@MIL-101(Cr)-NH₂ was found to be an efficient multifunctional catalyst for the aqueous selective hydrogenation of FUR to THFA with a selectivity of nearly 100% under mild conditions.

Amine-functionalized MIL-101(Cr) was hydrothermally synthesized by the direct reaction of Cr(Ⅲ) and 2-aminoterephthalic acid with assistance from hydroxide based on the previous literature with a slight modification [38]. To be specific, 2-aminoterephthalic acid (0.18 g, 1 mmol) and sodium hydroxide (0.1 g, 2.5 mmol) were dissolved in de-ionized water (7.5 mL) by ultrasonication, where the OH^- ions could promote the dissolution of organic acid. Then, chromic nitrate hydrate (0.4 g, 1 mmol) was dispersed into the former clear aqueous solution. After ultrasonication for 5 min, the suspension was transferred to a Teflon-lined autoclave and heated at 150 ℃ for 18 h in a convection oven. After cooling to room temperature naturally, the resulting green precipitate was collected by centrifugation and washed sequentially with de-ionized water, DMF and ethanol several times to remove the excess reagents. The sample was then soaked in hot ethanol with continued heating and stirring at 100 ℃ for 24 h for further purification, after which the product was dried at 100 ℃ under vacuum for 12 h.

Pd nanoparticles were successfully immobilized to MIL-101(Cr)-NH₂ by a direct anionic exchange and subsequent H₂ reduction. Taking the synthesis of 3.0 wt% Pd@MIL-101(Cr)-NH₂ as an example, activated MIL-101(Cr)-NH₂ (0.5 g) was first dispersed in deionized water (30 mL) by ultrasonication and treated with a suitable amount of diluted HCl to adjust the pH to approximately 4. Then a solution of H₂PdCl₄ (containing ca. 3.0 wt% of Pd) was added dropwise to the above slurry under vigorous stirring, and the mixture was further stirred for another 24 h. The solid was separated by centrifugation, repeatedly washed with deionized water, followed by drying at 100 ℃ under vacuum for 12 h. The resulting [MIL-101(Cr)-NH₃⁺]₂[PdCl₄]^2- sample was reduced in a H₂/Ar (H₂/Ar = 20/40 mL min^–1) flow at 200 ℃ for 4 h to yield Pd@MIL-101(Cr)-NH₂. The synthetic procedure is shown in Fig. S1.

MIL-101(Cr) was synthesized according to our previous report [40], then an impregnation procedure similar to the above-described was adopted for the incorporation of Pd within MIL-101(Cr) except with an adjusting of the pH.

The powder X-ray diffraction (XRD) patterns of the samples were recorded on a D/MAX-2400 diffractometer with Cu K_α radiation (λ = 1.5418 Å) at 40 kV and 100 mA. The Pd contents in MIL-101(Cr)-NH₂ were quantitatively determined by inductively coupled plasma-atomic emission spectroscopy (ICP-AES) using a Perkin-Elmer Optima 2000 DV. Thermogravimetric (TG) experiments were performed on a Mettler Toledo TGA/SDTA851e thermogravimetric analyzer with a heating procedure from 25 to 800 ℃ at a rate of 10 ℃ min^–1 under a nitrogen atmosphere. The nitrogen adsorption–desorption isotherms were performed at -196 ℃ on a Quantachrome Autosorb-IQ apparatus. The samples were degassed under vacuum at 150 ℃ for 10 h before the adsorption measurements. Fourier-transform infrared spectroscopy (FT-IR) characterization was performed on a Thermo fisher Nicolet 6700 spectrometer with a resolution of 0.09 cm^–1 at room temperature. The particle size distributions of the catalysts were analyzed by transmission electron microscopy (TEM) measurement using a JEM-2000EX instrument at 120 kV. Powder samples were ultrasonicated in ethanol and dispersed on TEM copper grids.

The selective hydrogenation of FUR was carried out in a 50 mL stainless steel autoclave equipped with a magnetic stirrer and an electrical heating jacket. The catalysts were reduced once again by H₂/Ar (H₂/Ar = 20/40 mL min^–1) at 200 ℃ for 2 h, and passivated under Ar overnight before use. In a typical run, catalyst (0.05 g), FUR (2.1 mmol), and water (20 mL) as a green solvent were added into the autoclave and purged with H₂ three times at room temperature. The autoclave was heated to 40 ℃ and then pressurized with H₂ to 2 MPa. After the reaction, the autoclave was cooled naturally, and the liquid products were separated by centrifugation, analyzed by gas chromatography (GC-7890F, FID, FFAP column 30 m × 0.32 mm × 0.5 μm) and identified by gas chromatography-mass spectrometry (Agilent 7890B-5977A GC/MSD). The quantitative analysis was performed by an internal standard method with propylene glycol added to the solution after the reaction as the internal standard.

The XRD patterns of MIL-101(Cr)-NH₂ and Pd@MIL-101(Cr)-NH₂ with various Pd loadings are shown in Fig. S2. The as-synthesized MIL-101(Cr)-NH₂ exhibited a typical diffraction pattern of MIL-101(Cr), which indicated a successful formation of the MIL-101(Cr)-NH₂ crystalline structure. After Pd loading, no obvious structural changes were observed, which suggested a robust MIL-101(Cr)-NH₂ structure as a host for the Pd nanoparticles [38]. It should be noted that the diffraction peaks derived from Pd NPs were barely observed in the wide-angle XRD patterns (Fig. S3) until the Pd content reached 5.4 wt%, which arose from the low Pd loading [41].

The thermal stability of MIL-101(Cr)-NH₂ and Pd@MIL-101(Cr)-NH₂ were examined and are shown in Fig. S4. TG analysis showed that both MIL-101(Cr)-NH₂ and Pd@MIL-101(Cr)-NH₂ were stable up to 350 ℃. The weight losses under 140 ℃ were ascribed to desorption of the adsorptive and coordinated water molecules and other residuals remained in the MOF cavities [29].

Fig. 1 presents the N₂ adsorption–desorption property of the synthesized MIL-101(Cr)-NH₂ materials. The bare MIL-101(Cr)-NH₂ exhibited a BET surface area of 1669 m² g^–1 and a total pore volume of 1.35 cm³ g^–1, which was in agreement with the data reported previously [42]. The sharp uptake under low pressure (P/P₀ = 10^–6 to 0.1) and the pore size distribution centered at 1.4 and 1.8 nm demonstrated the microporous feature of the materials. The increased N₂ uptake near P/P₀ = 1.0 arose from the textural pores created by nanoparticle aggregation [43]. After Pd loading, an obvious decrease of the micropore area and pore volume were observed (Table S1), which was attributed to the occupation or blocking of cavities by the deposited Pd nanoparticles or the partial collapse of the framework.

The MIL-101 samples were further characterized by TEM as shown in Fig. 2. Pd NPs were uniformly dispersed on MIL-101(Cr)-NH₂ for the 3.0 wt% Pd@MIL-101(Cr)-NH₂ sample with an average particle size of 3.5 nm (Fig. 2(a)). However, an excessive Pd loading (5.4 wt%) resulted in the formation of much larger nanoparticles with an average size of 4.4 nm (Fig. 2(b)). For the purpose of comparison, MIL-101(Cr) was used to load Pd NPs (2.7 wt%) as shown in Fig. 2(d). A wide range of particle size distribution and larger particles were observed owing to the agglomeration of Pd NPs, which inferred that the presence of amine groups within the frameworks arose from the strengthened adsorption force between NH₂ groups and Pd precursors, which led to the formation of uniform and well-dispersed Pd nanoparticles within the frameworks [44, 45].

The characterization of MIL-101(Cr) samples with IR spectroscopy are presented in Fig. S5. Compared with the unmodified MIL-101 (Cr), the two bands that appeared at 3490 and 3380 cm^–1 for MIL-101(Cr)-NH₂ were ascribed to the N–H bond stretching vibration of the aromatic primary amine in the crystal skeleton. The observed band at 1624 cm^–1 was associated with the bending vibration of N–H groups, and the bands at 1340 and 1256 cm^–1 were assigned to the stretching vibration of C–N bonds within aromatic amines [46].

The catalytic performance of the as-prepared Pd@MOF materials was evaluated for the catalytic selective hydrogenation of biomass-based FUR in aqueous media. The C=O group of FUR was more easily hydrogenated to form FA, owing to the lower bond energy than the C=C in the furan ring. Further hydrogenation of FA would facilitate the production of THFA. Fig. 3(a) shows the evolution of the reactants conversion and the products selectivity for the hydrogenation of FUR as a function of reaction time over the 3.0 wt% Pd@MIL-101(Cr)-NH₂ catalyst.

Nearly 100% conversion of FUR was achieved in the first 2 h, and the selectivity of THFA was up to 53%, which was accompanied by FA as an intermediate product. A full transformation of FA to THFA could be realized by prolonging the reaction time to 6 h. In addition, the FA could be fully converted to THFA within 2 h when the same amount of FA as the reactants was added into the reaction system. No further hydrogenolysis formed were produced with an extended reaction time (Fig. 3(b)). The above results suggested that THFA could be generated with high selectivity from FUR or FA under the current catalytic system. Comparing the Pd catalysts with other supports in the literature, such as Pd/MFI [9], Pd/Al₂O₃ [47] and Pd/SiO₂ [20], our Pd@MIL-101(Cr)-NH₂ catalyst could obtain highly monodispersed Pd nanoparticles owing to the existence of amine groups on the framework and the abundant pore-structure of MIL-101(Cr)-NH₂. The smaller Pd nanoparticles could promote the production of THFA with high yield under mild conditions and prevent the side reactions, which require harsh reaction conditions.

The effect of H₂ pressure on the hydrogenation of furfural is shown in Fig. 4. It can be clearly seen that with the increase of H₂ pressure from 0.5 to 1 MPa, the hydrogenation rate for FUR to FA was significantly increased. FUR was fully converted at 1.0 MPa and the selectivity to THFA reached 67.8% simultaneously under the investigated conditions. The intermediate FA reached a maximum selectivity at 0.5 MPa and was hydrogenated to THFA when the pressure was gradually raised to 3 MPa. However, the generation rate of THFA showed a slow improvement with the increased H₂ pressure, which indicated that the pressure had no dramatic effect on THFA generation under the current catalytic reaction system.

The temperature dependence of the product distribution was investigated from 30 to 100 ℃ (Fig. 5(a)) with 100% conversion of FUR. A complete hydrogenation saturation of FUR to THFA could be achieved when the temperature was increased to 40 ℃. The selectivity of THFA decreased when the reaction temperature was increased gradually. Cyclopentanone (CPONE), which originated from the hydrogenation rearrangement of FUR and FA in water media, was generated and became the main product at 100 ℃. At the same time, a small amount of 2-methyltetrahydrofuran and 5-hydroxy-2-pentanone as well as other unidentified substances were found in the products. The results suggested that temperature had a significant effect on the product distribution for the hydrogenation of FUR. A high reaction temperature would promote the rearrangement reaction to form CPONE, which hindered the transformation of FA to THFA (Fig. 5(b)) [14, 48]. However, we were unable to obtain more CPONE by further increasing the temperature owing to the aggravated polymerization of FUR and FA under the corresponding catalytic conditions [49]. The target product THFA could be obtained with high selectivity under a mild temperature of 40 ℃ over the Pd@MIL-101(Cr)-NH₂.

The higher activity and selectivity of Pd@MIL-101(Cr)-NH₂ compared with the other catalysts in the literatures [8, 17, 19] were mainly attributed to the existence of highly dispersed small Pd nanoparticles on MIL-101(Cr)-NH₂, which was more beneficial for the excitation of reactants. Meanwhile, there were strong host-guest interactions between the framework and metal nanoparticles through coordination and π-π forces, which could also enhance the catalytic activity [50, 51].

Table 1 lists the reaction results over Pd@MIL-101(Cr)-NH₂ with different Pd contents and Pd@MIL-101(Cr) at 40 ℃ and H₂ pressure of 2 MPa. No products were detected at the corresponding reaction time when MIL-101(Cr)-NH₂ was used as catalyst. However, THFA could be obtained as a final product with a selectivity > 99.9% through the direct hydrogenation of FUR when using Pd@MIL-101(Cr)-NH₂ with a Pd content of 3.0 wt%. A higher Pd content of 5.4 wt% had no effect on FUR hydrogenation but only slightly reduced the completion time of the reaction, which could be attributed to the greater number of active Pd sites. For comparison, the Pd@MIL-101(Cr) catalyst, which was prepared through a similar impregnation method, could only yield THFA with a selectivity of 53.2% under the same conditions. The higher catalytic performance for the hydrogenation saturation of FUR to THFA over Pd@MIL-101(Cr)-NH₂ compared with that over Pd@MIL-101(Cr) arose from the existence of free amine moieties within the frameworks, which could enhance the hydrophilic nature of the support and the formation of highly dispersed Pd NPs. Furthermore, the amine groups could improve the hydrogen bonding interactions between FA and the MOF matrix, which thus promoted a further hydrogenation of FA to THFA in cooperation with the metallic sites [36]. Therefore, the multifunctional Pd@MIL-101(Cr)-NH₂ resulted in a higher selectivity of THFA for FUR hydrogenation.

The recyclability test was performed with Pd@MIL-101(Cr)-NH₂ under the same reaction conditions to evaluate the catalyst stability. After each cycle of the reaction, the catalyst was separated and washed thoroughly with water and ethanol, and reduced again at 200 ℃ for 2 h under an atmosphere of Ar:H₂ = 2:1. In Fig. 6, the conversion rate of FUR exhibited a decrease after the first cycle, which indicated a slight decay of the catalytic activity. However, the FUR could be fully consumed with prolongation of the reaction time, and the THFA could still achieve a high selectivity above 90%. It should be noted that the activity of the catalyst became steady during the next three-cycle experiments. The XRD pattern of the Pd@MIL-101(Cr)-NH₂ catalyst after the reaction showed some decrease in crystallinity compared with the fresh catalyst (Fig. S2), which indicated partial destruction of the crystal framework of the MOF support after several cycles of reaction. The N₂ adsorption experiment showed a slight decrease of the BET surface area for the used Pd@MIL-101(Cr)-NH₂, while the significantly reduced micropore area and volume suggested the collapse of the partial microporous structure (Fig. 1). Meanwhile, the insoluble polymer derived from the polymerization of FA also affected the adsorption of N₂ [11]. Undesirably, the TEM image (Fig. 2) displayed a growth of the particle size from an average diameter size of 3.5 nm for the fresh Pd@MIL-101(Cr)-NH₂ to 5.2 nm for the used Pd@MIL-101(Cr)-NH₂, which suggested that the Pd nanoparticles could not be well stabilized by the MIL-101(Cr)-NH₂ support, which might be the main reason for the slight decrease in the activity of the as-prepared catalyst. To confirm that the decrease of catalytic activity mainly arose from the growth of the Pd rather than the Pd leaching, ICP experiments with the used Pd@MIL-101(Cr)-NH₂ and reaction liquid were performed. The results showed that the Pd content in Pd@MIL-101(Cr)-NH₂ was higher than 2.9 wt% and no Pd in the reaction liquid was detected.

Palladium nanoparticles were successfully incorporated into MIL-101(Cr)-NH₂ by a direct anionic exchange approach followed by hydrogen reduction. The presence of amino groups within the frameworks plays a key role in the formation of uniform and highly dispersed Pd nanoparticles on the support. Pd@MIL-101(Cr)-NH₂ has been demonstrated to be an efficient and reusable heterogeneous catalyst in the aqueous phase selective hydrogenation of the biomass platform compound FUR to THFA under the mild conditions of 40 ℃ and 2 MPa of H₂. The high activity and selectivity toward THFA benefit from the cooperation between the highly dispersed Pd nanoparticles and amino groups in the framework of MIL-101(Cr)-NH₂. The present results provide the possibility to further extend the applications of Metal@MOF composites to biomass usage.