Exceptionally Efficient and Recyclable Heterogeneous Metal–Organic Framework Catalyst for Glucose Isomerization in Water

Abstract Heterogeneous catalysts are desired for the conversion of glucose, the most abundant sugar in renewable biomass, but presently their synthesis requires highly toxic chemicals with long synthesis times. We report the conversion of glucose into fructose and 5‐hydroxymethylfurfural on a heterogeneous catalyst that is stable and selective and operates in the most environmentally benign solvent, water. We used a bifunctional solid with Lewis and Brønsted acid sites by partially replacing the organic linker of the zirconium organic framework UiO‐66 with 2‐monosulfo‐benzene‐1,4‐dicarboxylate. This catalyst showed high product selectivity (90 %) to 5‐hydroxymethylfurfural and fructose at 140 °C in water after a reaction time of 3 h. It was recyclable and showed only a minor loss in activity after the third recycle, offering a realistic solution for the bottleneck glucose isomerization reaction for scale‐up and industrial application of biomass utilization.

Heterogeneous catalysts are desired fort he conversion of glucose, the most abundant sugar in renewable biomass,b ut presently their synthesis requires highly toxic chemicals with long synthesis times. We report the conversion of glucose into fructosea nd 5-hydroxymethylfurfural on ah eterogeneous catalyst that is stable and selective ando peratesi nt he most environmentally benign solvent, water.W eu sed ab ifunctional solid with Lewis and Brønsted acids ites by partially replacing the organicl inker of the zirconium organic framework UiO-66 with 2-monosulfo-benzene-1,4-dicarboxylate. This catalyst showedh ighp roduct selectivity (90 %) to 5-hydroxymethylfurfural and fructosea t1 40 8Ci nw ater after ar eaction time of 3h.I tw as recyclable and showed only am inor loss in activity after the third recycle, offering ar ealistic solution for the bottleneck glucose isomerization reaction for scale-up and industrial applicationo fb iomass utilization.
Sustainable production of chemicals requires the utilization of renewable resources, one of the most promisingo fw hich is lignocellulosic biomass. [1,2] Biomass-derived sugars (e.g.,g lucose or fructose) can be convertedi nto platform molecules, for example, 5-hydroxymethylfurfural (HMF), which can be further processed into monomers, fuel additives, paints, and av ariety of fine chemicals envisaged in af uture biorefinery. [3,4] Although fructosec an be converted into HMF easily, [5] glucose is the main buildingb lock of lignocellulosic biomass, and its conversion remains challenging. [4] The best-performingh eterogeneous catalystf or this conversion is tin-incorporated beta zeolite (Sn-beta) with Sn 4 + occupying af raction of the tetrahedral sites in the zeolitef ramework. [6][7][8] Sn-beta can effect the isomerization of glucose to fructose in water with high selectivity (> 50 %). [7] However,S n-beta requires long crystallization times,u pt o4 0days, which is industrially unviable at high temperatures, 140 8C, and, moreover,requires the use of hydrofluoric acid, an acute poison and extremely corrosive. [7] In this work, we present ar ecyclable catalyst for glucose isomerization. It is based on modified UiO-66 (Figure 1a), [9] at hermally and hydrothermally robustm etal-organic framework (MOF), which we show matches the conversion and product selectivity of Sn-beta. The advantage of using MOFs as heterogeneousc atalysts is the potentialf or tuning the solids'p roperties by inclusion of desired functionall igands, [10] such as acids ites, [11] and at the same time by simple synthesis protocols, in this case without highly toxic and corrosiveH F, [12] in less than 24 hat120 8C.
The challenge in HMF productionf rom glucose is to achieve high product selectivity.T he reaction proceeds throughi somerization of glucoset of ructose ( Figure 1b), [13] which is the limiting step to achieve high selectivity.I tw as previously proposed that the reactioni sc atalyzed by Lewis acids, [13] which enable hydride shift between the carbon atoms of glucose, [14] and at the same time, proximal silanol groupso rB rønsted acid sites form ah ydrogen-bonding network, whichf acilitates proton mobility. [15] UiO-66 is az irconium-based MOF with benzene-1,4-dicarboxylate (BDC) linkers, and it is highly stable in air up to 500 8Ca si sh ydrothermally inert. [9] Defects in the form of coordinatively unsaturated Zr 4 + sites provide Lewis acidity. [16] We found that UiO-66 itself was active in glucose conversion ( Figure 1c)a nd showed1 6% conversion accompanied with 10 %p roduct yield at 140 8Ci n3h. However,i tl acks Brønsteda cids ites. Therefore, we synthesized ac atalyst by partially replacingt he BDC linker with 2-monosulfonated benzene-1,4-dicarboxylic acid (MSBDC), [17,18] and this catalyst showed3 1% glucose conversion under the same reaction conditions with 28 %p roduct yield ( Figure 1c). This corresponds to exceptional product selectivity of approximately 90 %, which is similar to that previously reported for Sn-beta zeolite. [7] The ratio between the BDC andM SBDC linkers is critical for the successful synthesis of as table, functionalized UiO-66 material. Higher ratios of MSBDC within the framework were already shown to decrease the stabilityU iO-66. [9,18] Indeed, we found that if only MSBDC was used as the ligand then the material subsequently collapsed upon hydrothermalt reatment ( Figure FigureS2f or all catalysts). Further,E DX analysiso ft he MSBDC-containing materials revealed the absence of sodium, supported by bulk inductively coupled plasma optical emission spectrometry (ICP-OES) analysis, consistent witht he displacement of sodium ions during synthesis to yield Brønsted acidic SO 3 Hs ites.
The incorporation of sulfonic acid groups was also confirmed through FTIR spectroscopy.N ew signals at ñ = 620, 1078, 1180, and1 223cm À1 appear in the spectra of the UiO-66-MSBDC catalysts, and their intensities increase upon increasingt he linker content ( Figure S3). These bands are attributed to the characteristica symmetric bending and symmetric and asymmetric stretching vibrations of the S=Oa nd SÀO bonds. [19,20] Elemental analyses of the fresh catalysts also show S/Zr ratios close to the expected values (Tables S1 and S2). Thermogravimetric analysis (TGA) shows extensive mass loss at approximately T = 510 8Cf or both the standard and functionalized UiO-66 materials ( Figure S4). This is consistent with the reported decomposition temperature of 540 8Cf or UiO-66 and approximately5 00 8Cf or sulfonicU iO-66 materials reported. [9,18] (Tables S3 andS 4).
Significant increases in the fructose yields combined with marginali ncreases in the HMF yields suggest that modification of UiO-66 with MSBDC could affect the Lewis acidityi nt wo ways. First, more defective materials are formed;t his is supported by an increasei nt he mesopore volumeo ftheU iO-66-MSBDC catalysts (Table S5 and Figure S6). Second, the Lewis acidity of Zr 4 + is known to be enhanced significantly by the presence of an earby electron-withdrawing group;t his was previously extensively studied in sulfated zirconiac atalysts. [21] This effect was recently reported in MOFs in the presence of electron-withdrawing functional groups such as NO 2 on the organic linker; [22] indeed, we found that the conversion over NO 2 -modified UiO-66 was higher than that overthe parent ma-  Figure S8), and so it is conceivable that the sulfonyl acid groups have as imilar effect. Clearly,a dditional work is needed to understand fully the interplayoft he acid functionalities.
The recyclability of catalysts is crucial fors cale-up and industrial applications:w es tudied this by recovering the solid catalysts by using ac entrifuge andw ashing them with water after each reaction cycle. Full recovery of the catalysts was not possible owing to the presence of small catalyst particles that remainedd ispersed in the reaction medium. These nanocrystalline catalysts have intrinsically high activity ( Figure S8).H owever,o nce the small particles weref iltered out after the first run, the catalysts could be recovered in consecutive reaction cycles (Table S6). Therefore, although ad ecrease in glucose conversion waso bserveda fter the first run, no loss in activity was observed in the following three recycles (Figure 3b), particularly for the UiO-66-MSBDC(20)c atalyst( see Ta ble S9 for product yields). The PXRD pattern of UiO-66-MSBDC(20)r ecovered after four runs showedt hat the integrity of the MOF lattice was maintained (Figure 3a). Zirconium and sulfur EDX mapping of the catalysts after four reactionc ycles further confirmed the integrity of the recycled catalysts ( Figure S2). Recycling of the UiO-66 and UiO-66-MSBDC(10) catalysts resulted in am inor loss of activity after the fourth run. This loss in activity could in partb earesult of the formation of undesired side products,s uch as humins.T hese are poorly characterizedo ligomeric speciest hat are known to be the main side products of this reaction. [3] These insoluble products can accumulate on the catalysts urface and block the active sites. Indeed, the recovered catalyst mass in the recycling test increasedo wing to the collection of inseparable side products (Table S3), which could explain why the recycled catalysth ad al ower sulfur count than the fresh catalyst, as determined by EDX analysis.
Notably,h owever,t he elemental analysis of the reaction solution after the first reactionc ycle (3 hr eaction at 140 8C) showed that only trace amounts of sulfur and zirconiumw ere present,w hichc onfirmed the stabilityo ft he catalystw ith negligible leachingd uring the reaction (Table S4). Finally,t he performances of the UiO-66 materials werec ompared to that of Sn-beta. In the literature, Sn-betai su sed as ag lucosei somerization catalystw ith aS n-to-glucose ratio of 1:50, and the catalyst weight of Sn-beta far exceeds the amount of the MOF catalyst used in this study under similarr eaction conditions, for which Sn-beta showed54% glucose conversion with 30 %fructose yield. [7] As imilar conversion (48 %) and product yield (34 %, Figure S7) were obtained upon using 40 mg of the UiO-66-MSBDC (20) catalyst, an amount that was still less than a quarter of the amount of the Sn-betac atalyst (200 mg).
Ta ilor-made MOFs with desired functionalities have made it possible to achieve exceptionally efficient catalysts for glucose isomerization in water.U iO-66-MSBDC catalysts containing dual Lewis and Brønsted acidity provided exceptional product selectivity of approximately 90 %f or the conversion of glucose into fructose and HMF,a nd this selectivity is close to that shown by Sn-beta zeolite.O ther MOF catalysts reported in the literaturef or glucose isomerization use frameworks constructed from toxic metals( e.g.,c hromium) [23][24][25][26] and/or have been used in non-aqueous solvents that are toxic or flammable (e.g.,D MSO or THF). [27] Our resultss how that UiO-66-MSBDC(y) catalysts are highly promising for scale-up because they operate under aqueous conditions and are recyclable, and furthermore,t heir synthesis is simple ands hort and does not require toxic and corrosivec onditions. Scaled-up synthesis of the MOFs by using continuous flow reactors, often using water as ar eactionm edium, makes this ar ealistic prospect. [28] Enzymes including metal centers and basic histidine moietiesp ossessing multifunctional capabilities are nature's catalysts,a nd they provide high selectivity at the expense of slow reactions and sensitive operational systems. Future work on MOF catalysts will be devoted towards better understandingo ft he active sites of these catalysts and the mechanism of their activity to optimize the product distribution and their long-term stability under industriallyr elevant flow-chemistry conditions.

Catalytic activity tests
The catalyst (10 mg) was placed in ar eaction vial (4 mL) with a magnetic stirring bar and 10 wt %a queous glucose solution was added. The vial was closed and placed in ap reheated oil bath at 140 8Cf or 3h.T he reaction was quenched at 0 8C, and the product mixture was analyzed by HPLC.

Characterizationo ft he catalysts
Powder XRD data were collected by using aP ANalytical X'Pert Pro MPD equipped with monochromatic CuK a1 radiation and aP IXcel solid-state detector.M icrographs and elemental maps were obtained by using aZ eiss Gemini scanning electron microscope with alarge area SDD EDX detector operating at 5keV.Nitrogen adsorption isotherms were measured at À196 8Cw ith aM icromeritics ASAP2020 system. The samples were outgassed at 150 8Cf or 12 h prior to the sorption measurements. Infrared spectra were recorded by using aP erkinElmer Paragon 1000 FTIR spectrometer in attenuated total reflection mode. Thermogravimetric analysis (TGA) was performed by using aM ettler To ledo Systems TGA/DSC 1i nstrument under ac onstant flow of air (50 mL min À1 ). Elemental analysis was performed by Medac Ltd (UK) for Zr and Sb yu sing ICP-OES after digestion and for CHN by using combustion. Additional experimental details can be found in the Supporting Information.