Caught in an acid trap! Zeolites with moderate acidity are demonstrated to be versatile and reusable solid-acid catalysts for the transformation of biomass derived carbohydrates to alkyl levulinates in alcohol solvents. In short-chained alcohols it proved possible to obtain significantly higher yields of alkyl levulinates, than what is achievable of levulinic acid in analogous reactions performed in water. The carbon-based chemicals and fuels that are necessary to meet the energy demand for our society originate presently almost exclusively from inexpensive fossil resources. The forecast of diminishing and more expensive petroleum reserves has, however, engaged the chemical industry to find new feasible routes to convert biomass into potential platform chemicals, that can replace the fossil based chemicals in order to leave the chemical supply chain unaffected (value chain).1–5 In biomass, the major source of renewable carbon is confined in lignocellulose, but selective hydrolysis combined with efficient separation of the lignocellulosic fractions; lignin, cellulose, and hemicellulose, remain a technical challenging task. In particular, the direct conversion of lignocellulosic biomass to potential chemicals with high selectivity provides a bottleneck, owing to the recalcitrant feedstock.6 Conversion of the individual monomeric components in lignocellulose to chemicals is an alternative way to make platform molecules and improve the yield of the targeted product.6c, 7 Consequently, we have focused on how levulinic acid esters can be directly derived from mono-, di-, and polysaccharides using heterogeneous catalysis. Levulinic acid has been listed as one of the twelve most important sugar-based building blocks obtainable from biomass,8 with a variety of potential applications as solvent, food flavoring agent, plasticizer, resin intermediate, and building blocks for, for example, tetrahydrofuran and succinic acid.9 To produce levulinic acid, carbohydrates are traditionally being treated with aqueous mineral acid (H2SO4 and HCl).10 However, a major drawback in this homogeneous process is tedious work-up during the separation stages. Previously, few solid-acid catalysts have been investigated for the conversion of glucose/cellulose to levulinic acid in water. Y- and Beta-type zeolites gave 10–40 % yield of levulinic acid from glucose/fructose/cellulose in water at temperatures between 110–240 °C.11 Despite the many advantages of zeolites, they are not hydrothermally stable in water during the reaction, which leads to a loss of structural integrity and can, therefore, not be recycled. Alternatively, Schraufnagel and Rase have reported that sucrose can be transformed into levulinic acid in water using acidic ion-exchange resins as catalysts at temperatures between 100 and 140 °C.12 However, owing to thermal instability of the resin at elevated temperatures, moderate yield of activity (25 % levulinic acid), and difficulty in removing humin by-products from the catalyst under mild conditions, they are unattractive for this transformation. A quantitative conversion of glucose was observed over Fe-pillared montmorillonite catalyst, but the selectivity to levulinic acid was only 20 % and a large amount of humins were also observed in this study.13 Recently, Peng et al. reported that glucose can be directly converted to ethyl levulinate (ELevu) over sulfated zirconia catalysts with a moderate yield up to 30 % at 200 °C.14 Furthermore, we found that sulfonic acid functionalized ionic liquids and solid Brønsted acids are more promising catalysts for the conversion of fructose to ELevu in ethanol than for the analogous conversion to levulinic acid in water.15 Here we introduce zeolites as catalysts for the conversion of carbohydrates to alkyl levulinates in alcohol medium to enhance the yield as well as to circumvent the drawbacks of hydrothermal and thermal instability of the catalysts. Aluminum-free Lewis acidic zeotype materials and mesoporous molecular sieves containing Sn, Ti, or Zr are capable of isomerizing hexoses, pentoses and trioses and converting them into useful chemicals.5, 16 Interestingly, we observed in the present study that glucose can efficiently isomerize to form fructose in alcohol media over zeolites without auxiliary Lewis acid metals (such as Sn, Ti, or Zr) other than aluminum. Under optimized reaction conditions, the transformation of glucose to methyl levulinate (MLevu) in methanol over various commercially available zeolites has been performed and the results are presented in Table 1. All of the experiments were performed with quantitative conversion of glucose and other mono- and disaccharides causing the reported yields and selectivities to alkyl levulinates to be identical. The H-Y zeolite experiment with a silicon to aluminum ratio of 2.6, that is, H-Y (2.6), yielded a significant amount of 32 % of MLevu. Notably, H-USY zeolites with Si/Al ratios of 6 and 30 gave 49 and 37 % yield of MLevu, respectively. These results revealed that glucose first isomerized to fructose and subsequently formed MLevu by consecutive steps, as shown in Scheme 1. The capability of H-USY (6) to isomerize glucose to fructose has been studied in more detail in alcohol and water as solvents at low temperatures, where a significant amount of fructose isomerization was found to occur in alcohols at 60–120 °C.17 Catalyst[b] MLevu yield[c] [%] ELevu yield[d] [%] Ratio of acid types (1:2)[e] H-Y (2.6) 32 15 1:0.36 H-USY (6) 49 41 1:0.81 H-USY (30) 37 26 1:0.91 H-Beta (12.5) 44 27 1:0.52 H-Beta (19) 47 28 1:0.83 H-Beta (150) 10 3 1:1.07 H-ZSM-5 (11.5) 8 2 1:1.14 no catalyst <1 0 – A plausible pathway for the formation of methyl levulinate (MLevu) from sugars. In earlier work, it has been shown that H-USY (6) contains more Lewis acid sites than H-USY (30),18 thus facilitating the isomerization of glucose to fructose and the further conversion to MLevu in relative high yield. Another common large-pore zeolite, Beta, has also been employed for this transformation and the results are shown in Table 1. The Beta zeolites with Si/Al ratio of 12.5, 19, and 150 yielded 44, 47 and 10 % of MLevu, respectively. In contrast, H-ZSM-5 (11.5) yielded only 8 % of MLevu, a yield most likely attributable to steric hindrance of the rather bulky molecules in the medium pore channels of H-ZSM-5. To recognize the role of acid sites, all the zeolites were subjected to NH3-TPD analysis and the results are compiled in Table 1 as the ratio of weak (type 1) to strong (type 2) acid sites corresponding to desorption temperatures of 100–270 °C and 270–500 °C, respectively (Figure S1 and Table S1 in the Supporting Information). It can be inferred that H-USY (6) and H-Beta (19), which have almost identical ratio of acid sites type 1 and 2 of about 1:0.8, gave the highest yields of MLevu. For H-USY (30) the acid site ratio was higher (1:0.91) than for H-USY (6) and H-Beta (19), which combined with a lower number of total acid sites gave a lower yield of MLevu (37 %). H-Beta (12.5) with acid ratio of 1:0.52 gave slightly lower yield of MLevu compared to H-USY (6) and H-Beta (19). From these results it can be deduced that an optimized yield of MLevu was obtained with catalysts having an acid site ratio between 1:0.3 and 1:1.1 along with a total acid site density of above 300 μmol g−1.19 Interestingly, H-Beta (150) yielded 33 % of methyl-D-glucopyranoside (MDGP) along with 10 % of MLevu (Table 1). This result suggested that MDGP could be an intermediate and isomerize to methyl-D-fructofuranoside (MDFF) on the pathway to forming MLevu. To confirm this transformation, a time-course study was performed over H-USY (6). The results from this study showed that the yield of MDGP and MLevu was around 16 and 23 %, respectively, after 10 min of reaction (Figure S2). Notably, the yield of MLevu increased at the expense of MDGP as the reaction time increased. To provide further prove to the suggested pathway, MDGP and ethyl-D-glucopyranoside (EDGP) were also used as substrate in experiments performed under identical reaction conditions. Here, it was observed that MDGP gave 49 % of MLevu and EDGP gave 41 % of ELevu after 20 h of reaction (Table 2). Based on the obtained results it can accordingly be established, that the formation of MLevu preferably occurred via three pathways, as shown in Scheme 1. In the first reaction pathway, glucose can isomerize to fructose and subsequently react with methanol to form MDFF followed by dehydration to form 5-(methoxymethyl)-2-furaldehyde (MMF) and rehydration to form MLevu. In the second pathway, glucose can directly react with methanol to form MDGP and then isomerize to MDFF followed by dehydration-rehydration as in the first pathway. In the third pathway, the fructose formed by isomerization can directly dehydrate to form HMF followed by etherification and rehydration to form MLevu. In previous work, it has been observed that methyl fructoside was the only product when using fructose as substrate.20 In our previous report, we confirmed the formation of 5-(ethoxymethyl)-2-furaldehyde as one of the intermediates formed when fructose was converted to ELevu in ethanol.15a Substrate MLevu yield[b] [%] ELevu yield[c] [%] glucose 49 41 fructose 51 40 mannose 53 44 sorbose 41 33 sucrose 49 35 cellobiose 53 44 maltose 51 47 inulin 50 39 starch 31 4 cellulose[d] 13 – MDGP[e] 49 – EDGP[f] – 41 In all sugar experiments with methanol, a relative large amount of humins were observed along with other intermediates and side-products such as, for example, MDGP (<20 %), HMF (<1 %), MDFF (<7 %), MMF (<3 %), 2-(dimethoxymethyl)-5-(methoxymethyl)furan (<2 %), methyl lactate (<3 %), formic acid (7–12 %), formaldehyde dimethyl acetal (<3 %), methoxyacetaldehyde dimethylacetal (<3 %), dimethyl ether, and others (<6 %). Combined the quantified products accounted generally for 50–60 % of the carbon balance, while the partly alcohol-soluble humins could not be detected by HPLC or GC-MS. Analogous ethylated intermediates and side-products were also observed in all the ethanol experiments. The conversions of other mono-, di- and, polysaccharides derived from biomass have also been studied over H-USY (6) under identical reaction conditions, and the results are summarized in Table 2. Generally, the applied mono- and disaccharides were found to yield 41–53 % MLevu or 33–47 % ELevu, respectively. Notably, cellobiose and maltose—which can be regarded as model compounds for cellulose and starch—gave relatively high yields of 53 and 51 % of MLevu, respectively. Similarly, the polysaccharide inulin yielded 50 and 39 % of MLevu and ELevu, implying that inulin was dissolved and depolymerized into fructose monomers and alkyl fructosides, before eventually being converted into alkyl levulinates. A significant yield of MLevu (31 %) was also obtained from starch, suggesting facile break down of the α(1→4) glycosidic bonds in starch. In contrast, a very low yield of MLevu (3 %) was obtained from cellulose containing β(1→4) glycosidic bonds. However, the yield of MLevu from cellulose could be enhanced to 13 %, when the reaction temperature was increased to 200 °C and the reaction time prolonged to 20 h. When changing the solvent from short alcohols to water, the yield of levulinic acid with H-USY (6) catalyst was dramatically reduced to 11 % with quantitative conversion of glucose (Table S2). Additionally, after the first reaction run in water, the formal BET surface area and pore volume of the H-USY (6) catalyst decreased significantly (Table S3), implying a detrimental loss of structural integrity of the zeolite lattice.5, 18 When using the longer alcohol solvents 1-propanol or 1-butanol the corresponding yields of propyl and butyl levulinate was also low compared to methanol or ethanol, probably due to predominant side reactions leading to dialkylethers (Table S2). Combined, the results clearly demonstrate the advantage of using a short-chained alcohol as the solvent to get alkyl levulinates—rather than using water, which would produce levulinic acid—and emphasize the ability of the alcohol solvent to preserve the catalytic activity of the catalyst. For large-scale implementation it is essential to demonstrate that catalyst reusability is feasible. Based on this view point, H-USY (6) has been tested for the transformation of glucose to MLevu and ELevu for five reaction cycles in methanol and ethanol, respectively (Figure 1). In the first through the fifth reaction run, the yields of MLevu and ELevu formed from glucose remained essentially constant at 45–50 % and 40–45 %, respectively. After each cycle, the catalyst was calcined under static air at 550 °C for 5–10 h to remove humins from the catalyst, which otherwise caused a lowering in product yield upon reuse. This was demonstrated in a separate experiment, where the product yield of MLevu was drastically reduced from 50 to 27 % if the catalyst was reused directly without pre-calcination. After the fifth reaction cycle the H-USY (6) catalyst was further subjected to nitrogen sorption analysis. Importantly, this analysis revealed that no significant changes in pore volumes and formal BET surface areas were observed (Table S3), thus confirming that the structural integrity of the H-USY (6) was indeed maintained upon recycling. Recyclability of the H-USY (6) catalyst in the transformation of glucose to methyl levulinate in methanol and ethyl levulinate in ethanol, respectively (see the Supporting Information for experimental details). In summary, we have demonstrated for the first time that unmodified zeolite H-USY (6) can directly transform biomass derived mono- and disaccharides, such as glucose and sucrose, to MLevu and ELevu in alcohols in unprecedentedly high yields (about 50 %) compared to other zeolites. In alcohol solvents the zeolite was shown to maintain its structural integrity, which further allowed the catalyst to be reused in at least five reaction cycles in the short-chained alcohols methanol and ethanol without significant changes in yields. In contrast, the catalyst provided a poor alkyl levulinate yield when used for the reaction with water as solvent, due to significant destruction of the structural characteristics of the zeolite. Additional work is currently in progress to enhance the yields of alkyl levulinates in order to improve the industrial potential of the catalytic system. The Danish Council for Independent Research—Technology and Production Sciences (project no. 10-081991) and the Catalysis for Sustainable Energy initiative funded by the Danish Ministry of Science, Technology and Innovation are acknowledged for financial support of the work. As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

	Year	Citations

References