Report

# Metal Chlorides in Ionic Liquid Solvents Convert Sugars to 5-Hydroxymethylfurfural

See allHide authors and affiliations

Science  15 Jun 2007:
Vol. 316, Issue 5831, pp. 1597-1600
DOI: 10.1126/science.1141199

## Abstract

Replacing petroleum feedstocks by biomass requires efficient methods to convert carbohydrates to a variety of chemical compounds. We report the catalytic conversion of sugars giving high yield to 5-hydroxymethylfurfural (HMF), a versatile intermediate. Metal halides in 1-alkyl-3-methylimidazolium chloride are catalysts, among which chromium (II) chloride is found to be uniquely effective, leading to the conversion of glucose to HMF with a yield near 70%. A wide range of metal halides is found to catalyze the conversion of fructose to HMF. Only a negligible amount of levulinic acid is formed in these reactions.

A sustainable future for the chemical industry requires feedstocks based on renewable rather than steadily depleting sources. Inability to effectively transform five- and six-carbon carbohydrate building blocks derived from nature is a major barrier toward this challenging goal. Glucose and fructose, two abundant six-carbon sugar molecules, are potential feedstocks for this purpose, and recent efforts have focused on converting them to 5-hydroxymethylfurfural (HMF) (1), a versatile intermediate between biomass-based carbohydrate chemistry and petroleum-based industrial organic chemistry (2). HMF and its derivatives could potentially replace voluminously consumed petroleum-based building blocks (3), which are currently used to make plastics and fine chemicals. Recently, Dumesic and co-workers introduced the idea of using HMF as a key intermediate to produce liquid alkanes from renewable biomass resources (4). High production cost currently limits the availability and use of HMF industrially.

A process to produce pure HMF from abundant renewable carbohydrates in high yield at low energy cost must be developed before a biorefinery platform can be built on the basis of this substrate. Current processes to produce HMF involve the use of acid catalysts and are mainly limited to fructose as feed (57). A drawback with acid catalysts is that they cause various side reactions, significantly increasing the cost of product purification. For example, in water under acidic conditions, HMF decomposes to levulinic acid and formic acid. Levulinic acid is particularly difficult to separate from HMF. Substituting glucose as a feed substantially reduces HMF yields and produces additional by-products (8, 9).

A study by Antal and co-workers suggested that HMF is formed from dehydration of fructose in its furanose form (10) and occurs through a series of cyclic furan intermediates (11). Others have suggested HMF is formed through an acyclic mechanism proceeding through an enediol pathway (1114). The enediol is proposed as an intermediate in the isomerization of glucose to fructose. Glucose has competing reaction pathways that lead to formation of by-products. In one pathway, dehydration forms nonfuran cyclic ethers; in another, C-C bond scission occurs through reverse aldol condensation (14). To obtain high HMF yields from glucose thus requires effective methods for selective in situ isomerization to fructose.

HMF yield has been shown to increase significantly in systems that partition HMF from H2O. Dumesic and co-workers, building on the work of several earlier researchers, demonstrated high yields from fructose by using strong polar organic solvents, such as dimethylsulfoxide (DMSO), in aqueous-organic reaction media (7). Other solvent systems have also shown promising results. HMF can be formed in high yields from fructose in 1-H-3-methylimidazolium chloride solvent, which also acts as an acidic catalyst (15, 16). In sugar-solubilizing 1-alkyl-3-methylimidazolium chloride solvents, water is not needed as part of the solvent system, and the actual amount of H2O present is reduced to the water formed during dehydration. By minimizing HMF exposure to acidic aqueous solutions at elevated temperatures, HMF yield loss to levulinic acid is kept very low.

In our own work, we have built upon this concept by using sugar-solubilizing high-purity 1-alkyl-3-methylimidazolium chloride, [AMIM]Cl, as a solvent class. Our method is distinguished from previous reports in that we observe high yields of HMF from fructose without added acid. Even more importantly, one of these solvent-catalyst systems is able to produce HMF in high yields from glucose, the first step in our ultimate goal of developing a system to generate HMF from complex biomass such as cellulose.

We tested the reactivity of fructose in three [AMIM]Cl solvents, where A represents octyl, butyl, or ethyl (17). Because [EMIM]Cl (E is ethyl) was equivalent or better than the other two solvents, we report results in this system (Fig. 1). Figure 2 shows the results of simply heating fructose and glucose in high-purity (99.5%) [EMIM]Cl (18). At sufficiently high-temperatures, fructose was converted to HMF, but the yield dropped substantially between 120° and 80°C. In contrast, glucose did not produce any substantial amount of HMF even at 180°C. When water was added to the solvent ([EMIM]Cl:H2O = 5:1), glucose was effectively inert.

We were able to catalyze the dehydration of fructose at 80°C by addition of a catalytic amount of a number of metal halides (Fig. 1A). For example, HMF yields ranging from 63 to 83% were achieved in 3 hours when using 6 mole percent (mol %) loading (based on sugar) of CrCl2, CrCl3, FeCl2, FeCl3, CuCl, CuCl2, VCl3, MoCl3, PdCl2, PtCl2, PtCl4, RuCl3, or RhCl3 (fig. S1). The product mixtures were very clean: Yields of levulinic acid and α-angelicalactone were less than 0.08%. Not all metal halides were effective; for example, the alkali chlorides, LaCl3 and MnCl2, did not work.

We also looked at mineral and Lewis acid catalysts. Mineral acids were effective as expected. An 80% HMF yield was achieved when 18 mol % H2SO4 (relative to fructose) was used. A lower acid loading (1.8 mol %) gave 75% yield. In contrast, the widely studied AlCl3–Lewis acid was not effective at molar ratios between 0.5 and 2 (19).

We repeated these studies with use of glucose as feed but raised the temperature to 100°C because of its lower reactivity (Fig. 1B). Twelve of the metal halides tested showed 40% conversion of glucose, but only one catalyst, CrCl2, gave HMF in high yield (Fig. 3). HMF yield in [EMIM]Cl containing sulfuric acid or AlCl3 was only 10%. The results were reproduced at least 20 times, and HMF yields for systems that did not contain CrClx were consistently 10% or less, whereas CrCl2 afforded HMF yields of 68 to 70%, a previously elusive efficiency from glucose. The products from the other catalysts included sugars such as mannose, dehydration products such as 1,6-anhydroglucose, and poorly characterized polymeric products [determined by 13C nuclear magnetic resonance (NMR) spectroscopy].

For many of the catalysts, glucose conversion was high even though HMF yields were low (Fig. 2). We did a number of control experiments to demonstrate that the low HMF yield in these instances was not the result of HMF instability under the reaction conditions. After heating pure HMF at 100°C for 3 hours in the presence of CrCl2, 98% was recovered. When CrCl2 was not added to [EMIM]Cl, HMF recovery was only 28%. Similar studies were done with other metal halides (HMF recoveries noted in parentheses): CuCl2 (85%), VCl4 (86%), and H2SO4 (98%). Interestingly, catalytic amounts of certain metal chlorides appear to play a role in stabilizing HMF. Catalysts usually enhance reactions; the concept of a catalytic amount of a substance, in less than stoichiometric quantities, blocking or quenching a reaction is most unusual.

In a second study, we examined HMF stability in the presence of sugar and catalyst. In these tests, xylose, a five-carbon sugar that cannot form HMF, was used. Fifty mg of a 1:1 mixture of HMF and xylose were added to 500 mg of the appropriate [EMIM]Cl-catalyst system and heated to 100°C for 3 hours. HMF recovery was high (recoveries given in parentheses): CrCl2 (83%), CuCl2 (90%), and VCl4 (83%). HMF once again was more stable in the presence of metal halide; HMF recovery in [EMIM]Cl-xylose without catalyst was 66%. The data show that the low HMF yield cannot be accounted for by product instability under reaction conditions. Instead metal halides, such as CuCl2 and VCl4, catalyze undesired reaction pathways.

The singular effectiveness of catalytic amounts of CrCl2 in [EMIM]Cl for the conversion of glucose to HMF was unanticipated. In an effort to understand the results, we turned to spectroscopy. Our NMR study showed that the glucose starting material, when dissolved in the [EMIM]Cl solvent, is predominantly an α-anomer. Solvation of sugars occurs through hydrogen bonding of chloride ions of the solvent with the carbohydrate hydroxy groups (20). However, this interaction is insufficient to cause mutarotation (that is, α- to β-anomer conversion; Fig. 4). Little interconversion of the α- and β-anomers occurred in [EMIM]Cl, even after several hours at 80°C. However, in the presence of a catalytic amount of CuCl2 or CrCl2, mutarotation leading to an equilibrium mixture of anomers was rapid (figs. S2 and S3). In the 1H NMR spectrum, the six –OH resonances were sharp. In the presence of catalytic amounts of CuCl2 the –OH resonances shifted upfield and were very broad, indicating exchange through interactions with the metal (21).

[AMIM]Cl structures are known to be weakly coordinating (22) and do not compete with sugar for the binding of the metal chlorides. Our hypothesis is that sugar-metal coordination is responsible for the catalysis. To characterize the prevailing coordination bonding motif, we examined the effect of adding stoichiometric glycerol or glyceraldehyde to glucose solutions. Glucose can be thought of as a glycerol molecule attached to a glyceraldehyde molecule. By using NMR spectroscopy, we confirmed that glyceraldehyde exists as a hemiacetal dimer in [EMIM]Cl, which makes it a very good mimic for glucopyranose. In the competition reactions, glycerol had no impact on the catalysis, and 70% yield of HMF was achieved from glucose. Glyceraldehyde, however, did affect the chemistry: Reaction inhibition was observed. HMF yield was reduced to less than 20%, and glucose conversion was reduced to ∼60%. In addition, we evaluated 2,2′-bipyridine as a strongly coordinating ligand (5:1 molar ratio to Cr). In the presence of the strongly coordinating ligand, the reaction essentially shut down: HMF yield was less than 2%, and glucose recovery was 90%. The results of the glycerol and glyceraldehyde competition studies show that the metal interacts with the hemiacetal portion of glucopyranose but that there is little interaction with the polyalcohol portion of the sugar.

Although we lack a clear picture of why CrCl2 is a singularly effective catalyst in [EMIM]Cl solvent, we are able to offer some insights into the mechanism. We studied the kinetic behavior of CrCl2, CuCl2, and FeCl2, which show dramatic differences in their reaction pathways (fig. S4). The rate of glucose conversion was highest with CrCl2; CuCl2 was reactive, but mainly gave condensation products; and FeCl2 showed no reaction. With CuCl2, multiple products were formed, including mannose, HMF, and 1,6-anhydroglucose. This diverse product mix suggests that CuCl2 coordination with the sugar is different from that of CrCl2.

To explain the results with CrCl2, we show a mechanism consistent with the data (Fig. 4). Because only 0.5% by weight of CrCl2 was added to the solvents, the plausible formation of [EMIM]+CrCl 3 would consume only an equimolar amount of [EMIM]Cl with respect to CrCl2, according to Eq. 1: $Math$(1)

We propose that the CrCl 3 anion plays a role in proton transfer, facilitating mutarotation of glucose in [EMIM]. A critical role of CrCl 3 is to effect a formal hydride transfer, leading to isomerization of glucose to fructose. As discussed above, all other tested metal chlorides failed to convert glucose to fructose in the [EMIM]Cl solvent. A chromium enolate may be the key intermediate (23). Once fructose is formed, dehydration of fructofuranose is rapid in the presence of the catalyst in the solvent. Lowering the dielectric constant of the media by addition of organic solvents (10:1 glycerol to [EMIM]Cl) results in loss of catalytic activity. Other metal halides also bind to glucose. However, they promote alternative reaction paths that do not lead to the desired products.

Supporting Online Material

Materials and Methods

Figs. S1 to S4

View Abstract