5-Hydroxymethylfurfural (HMF) As a Building Block Platform.Pdf

View Online Green Chemistry Dynamic Article Links

Cite this: Green Chem., 2011, 13, 754 www.rsc.org/greenchem CRITICAL REVIEW

5-Hydroxymethylfurfural (HMF) as a building block platform: Biological properties, synthesis and synthetic applications

Andreia A. Rosatella,a Svilen P. Simeonov,a Raquel F. M. Fradea and Carlos A. M. Afonso*a,b

Received 5th August 2010, Accepted 15th December 2010 DOI: 10.1039/c0gc00401d

The bioreﬁnery is an important approach for the current needs of energy and chemical building blocks for a diverse range of applications, that gradually may replace current dependence on fossil-fuel resources. Among other primary renewable building blocks, 5-hydroxymethylfurfural (HMF) is considered an important intermediate due to its rich chemistry and potential availability from carbohydrates such as fructose, glucose, sucrose, cellulose and inulin. In recent years, considerable efforts have been made on the transformation of carbohydrates into HMF. In this critical review we provide an overview of the effects of HMF on microorganisms and humans, HMF production and functional group transformations of HMF to relevant target molecules by taking advantage of the primary hydroxyl, aldehyde and furan functionalities.

1 Introduction tion should be controlled to avoid formation of oligosaccharides and to prevent monosaccharides from reacting at the high The main source of functionalized carbon skeletons for the temperatures used.4 fine chemical industry, as well as for thermal and energy In contrast to cellulose, hemicellulose is a polymer formed by transportation, is still based on the fossil-fuel reservoir. However, different sugar units such as glucose, galactose, mannose, xylose the increasing price of oil will create new demand for molecules and arabinose, and it does not form crystalline regions, making it from renewable sources, and it seems likely that biorefineries will Downloaded by University of Oxford on 06 April 2011 more amenable to hydrolysis. Additionally, the rate of hydration 1 play a more significant role in this respect in the near future. depends on the sugar type, and decreases following the order The commercial production of wood sugars for ethanol Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D xylose > mannose > glucose. Consequently, hemicellulose is production was first considered at the beginning of the 20th hydrolysed faster than cellulose. Whereas dehydration of hexoses 2 century. Lignocellulose, a very abundant material, comprises produces HMF, pentoses can lead to production of furfural.4 important polymers (cellulose, hemicellulose and lignin), of HMF is very useful not only as intermediate for the produc- which cellulose and hemicellulose in particular are of high tion of the biofuel dimethylfuran (DMF) and other molecules, importance, since they are formed from monomers of glucose but also for important molecules such as levulinic acid, (or other types of sugar in the case of hemicellulose), and they 2,5-furandicarboxylic acid (FDA), 2,5-diformylfuran (DFF), can be used as a carbon source in fermentation processes for the dihydroxymethylfuran and 5-hydroxy-4-keto-2-pentenoic acid production of ethanol. (Scheme 1). There are already a considerable range of chemical building blocks derived from renewable resources.3 One of these, 5- hydroxymethylfurfural (HMF), plays an important role, because it can be obtained not only from fructose but also (more recently) from glucose via isomerisation to fructose, as well as directly from cellulose. Cellulose is formed by anhydro-D-glucopyranose units linked by b-1→4-glycosidic bonds, and thus hydrolytic degradation is necessary to release the sugar monomers. Hydrolytic degrada-

aCQFM, Centro de Qu´ımica-F´ısica Molecular and IN–Institute of Nanosciences and Nanotechnology, Instituto Superior Tecnico,´ 1049-001, Lisboa, Portugal. E-mail: [email protected]; Fax: + 35 1218464455/7; Tel: +35 218419785 biMed.UL, Faculdade de Farmacia´ da Universidade de Lisboa, Av. Prof. Gama Pinto, 1649-003, Lisboa, Portugal. E-mail: [email protected]; Fax: +35 1-21-7946476 Scheme 1

754 | Green Chem., 2011, 13, 754–793 This journal is © The Royal Society of Chemistry 2011 View Online

HMF was first reported at the end of the 19th century, when HMF, for instance, during heating of milk, which has a high Dull et al.5 described its synthesis by heating inulin with oxalic concentration of lactose and lysine-rich proteins.23 Under acidic acid solution under pressure. In the same year, Kiermayer6 conditions, lactulosyl-lysine can suffer 1,2-enolization via 3- reported a similar procedure for HMF synthesis, but starting deoxyosulose to form bound HMF. However, isomerisation from sugar cane. In the subsequent years, several preparation and degradation of lactose (the Lobry de Bruyn–van Ekenstein methods were reported using homogeneous and heterogeneous transformation) also accounts for the formation of HMF. acid catalysis, both in aqueous media.7 This topic was first re- Quantification of bound HMF can be used to assess the extent of viewed in 1951 by Newth et al.,8 and since then several important the Maillard reaction in foods. Morales et al. have removed the reviews have been published, including one by Moye et al.9 on free lactose from milk samples and quantified HMF released synthetic methods and industrial applications of HMF. Later, from oxalic acid degradation of lactulosyl-lysine compounds, Harris10 described the dehydration reactions of carbohydrates in using reversed-phase HPLC. This study demonstrated that this acidic and basic conditions, including their mechanisms. In 1981 method can be used to determine the extent of the Maillard two reviews where published, one covering HMF manufacture,11 reaction; however, they also showed that this reaction is a minor and other focusing on HMF chemistry.12 In 1990 and 1991, route for sugar degradation. Other techniques, such as the 2- two important reviews were published by Kuster13 and Cottier thiobarbituric acid (TBA) method, widely applied in dairies, et al.14 respectively, describing the manufacture of HMF. More can also be used to quantify HMF, but it is less suitable since recently, Lewkowski15 and Moreau et al.16 have reviewed the other aldehydes can take part in the reaction.24 synthesis and chemistry of HMF. Corma et al.3a dedicated a Many other studies have been published, but HPLC seems to chapter to the synthesis of HMF in an outstanding review be the chosen method for HMF determination.25 Solubilisation of biomass transformations. Woodley et al.17 also summarized of the ground food sample in water and use of trichloroacetic some processses for the synthesis of HMF, and Zhang et al.18 acid (as a clarifying agent), was used to eliminate interference connected biomass transformations with imidazolium salts, by during HPLC determination of HMF in cookies.25b HMF including the synthesis of HMF with ionic liquids as solvents. determination has also been used as a parameter to evaluate Some of these reviews are comprehensive, while others just heat effects during manufacture of cereal products.26 Ram´ırez– mention HMF chemistry,19 but this area has been progressing Jimenez´ et al. have reported formation of HMF during browning very fast, and over 90 articles have been reported in scientific of sliced bread, and increasing amounts were detected with journals in 2010.20 increasing heating time (14.8 mg kg-1 and 2024.8 mg kg-1 with 5 In this critical review we provide an overview of the biological or 60 min toasting time, respectively).26c Fallico et al. have also properties of HMF, recent developments in the preparation of reported the effect of the temperature in the HMF formation HMF from carbohydrates, and synthetic transformations. during the roasting of hazelnuts, and they also studied the effect of the oil in this mechanism. Defatted crushed hazelnuts produced less HMF during roasting (2.2 mg kg-1 at 150 ◦C 2 Formation of HMF during baking ◦ Downloaded by University of Oxford on 06 April 2011 for 60 min) than crushed hazelnuts (8.0 mg kg-1 at 150 Cfor In the bakery industry, the formation of dough starts with a 60 min), and addition of 10% water to the defatted crushed Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D mixture of flour, water, yeast and salt, which after fermentation hazelnuts led to an increase of HMF of approximately 32%. is subjected to high temperatures. During this baking process, Additionally, increasing the temperature to 175 ◦C produced the dough undergoes physical and chemical changes. The an increase in HMF concentration, as expected (66.5 mg kg-1 temperature leads to the evaporation of water and the formation for crushed hazelnuts and 17.9 mg kg-1 for defatted crushed of compounds that contribute to flavour and browning. These hazelnuts), even when toasted for 30 min.27 Furthermore, studies products result from Maillard reactions and caramelization. The have also demonstrated that formation of HMF decreases with first consists of a reaction between the carbonyl group of the the increase of humidity, and that fructose is more efficiently sugar and the amino group of an amino acid, and generally degraded in this furfural derivative than glucose.28 occurs at high temperatures (>50 ◦C) and acidic pH (4–7), and is favoured in foods with a high protein and carbohydrate content and intermediate moisture content.21 Caramelization is 3 Biological properties the oxidation of sugar, and needs more drastic conditions, such 3.1 Effects of HMF on the growth of microorganisms as temperatures above 120 ◦C and more extreme pH (<3or>9) and a low amount of water.21 The use of hemicellulose in fermentation as a carbon source, These reactions are frequent in bakery products, but also in and the consequent generation of HMF, has created a demand other foods subjected to high temperatures during processing. for HMF-resistant microorganism strains (Table 1). The reaction of fructose, lactose and maltose with the amino Several studied strains of Saccharomyces cerevisiae were found group of lysine to form fructosyl-lysine, lactulosyl-lysine and to be quite tolerant to HMF; however, results varied substan- maltulosyl-lysine (Amadori products) is characteristic of the tially within the studied microorganisms: 1) addition of 4 g early stages of Maillard reactions, and is responsible for L-1 of HMF to an anaerobic fermentation with Saccharomyces decreasing the available lysine and food nutritional value. Thus, cerevisiae CBS 8066 caused a decrease in the carbon dioxide evaluation of these compounds has been suggested to work evolution rate, and the growth rate was significantly affected. as control parameters for assessment of the quality of foods.22 HMF was metabolized by the yeast but this process stopped However, other products can be formed, and there are several after exhaustion of glucose, with the consequent end of ethanol examples in the literature of the degradation of the sugar in production;29 2) a lower concentration of 1.5 g L-1 HMF did

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 754–793 | 755 View Online

Table 1 Effect of HMF on the growth and/or ethanol production during fermentation using different strains of microorganisms

Microorganism HMF (g L-1) Result

Saccharomyces cerevisiae CBS 806629 4.0 Growth did not decrease, but reached a plateau after faster consumption of glucose, with consequent ceasing of ethanol production Saccharomyces cerevisiae TMB 300130 1.5 No effect on ethanol production Saccharomyces cerevisiae ATCC 21123931 <2.5 Growth was not significantly affected 3.8 Long lag phase in the growth curve of about 24 h Saccharomyces cerevisiae NRRL Y-1263231 2.5 Growth greatly affected 3.8 No growth Saccharomyces cerevisiae TMB 340023b 7.6 Decreased glucose consumption and production rate of ethanol Pichia stipitis NRRL-Y-712431 2.5 Growth not significantly affected 3.8 No growth Rhodosporidium toruloides Y433 1.9 Growth was not significantly affected

not have any effect on ethanol production during the anaerobic that HMF can be toxic if administered at doses of 75 mg fermentation of xylose by S. cerevisiae TMB 3001;30 3) studies kg-1 body weight.19d Consequently, several studies have been with S. cerevisiae ATCC 211239 demonstrated that cell growth conducted in an attempt to investigate the effect of HMF in was not significantly affected at a HMF concentration below humans. 2.5 g L-1,butat3.8gL-1, a long lag phase in the growth To assess the effect of HMF in humans, several in vitro and curve (approximately 24 h) appeared;31 4) S. cerevisiae NRRL in vivo assays have been performed. The mutagenic effect has Y-12632 was greatly affected at a concentration of 2.5 g L-1,and been assessed by the Ames test, which studies the potential no growth was detected at a concentration of 3.8 g L-1;31 and of the compound to make possible the growth of histidine- 5) larger amounts of HMF (7.6 g L-1) were also tested in an deficient bacterial strains plated without a histidine supplement. anaerobic fermentation with S. cerevisiae TMB 3400, and led In these tests, HMF was found to be not mutagenic or only to a 50% decrease of glucose concentration within 24 h weakly mutagenic.40 Brands et al. have tested heated mixtures (compared to the control, in which the glucose had already been of sugar-casein using the Ames test, and have concluded used up), with a consequent decrease of the production rate of that mutagenicity is related to the extent of the Maillard ethanol.23b reaction, and varied with the type of sugar, fructose being Adifferentyeast–Pichia stipitis NRRL-Y-7124 – was also more mutagenic than glucose (the reason being the different tested, and growth was not significantly changed in the presence reaction mechanisms).41 Furthermore, disaccharides were less of2.5gL-1 HMF; however, it was impaired at a concentration mutagenic than monosaccharides, because the first induced less of 3.8 g L-1.31 A different study performed with this last strain mutagenic compounds.41 However, the compounds responsible

Downloaded by University of Oxford on 06 April 2011 revealed that tolerance to HMF improved in stationary phase for this mutagenicity were not identiﬁed, but results were weak cultures and was greater in the presence of glucose rather compared to the chemical mutagen 4-nitroquinoline-N-oxide,

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D than xylose and, regardless of the carbon source, amino acid used as positive control.41 Additionally, the viability of the enrichment of the culture medium enhanced the ability of cells human hepatocyte cell line-HepG2 in the presence of HMF to resist HMF exposure.32 Rhodosporidium toruloides Y4 has was not significantly affected (a concentration of 38 mM was also been investigated, and addition of 1.9 g L-1 HMF was necessary to reduce viability in about 50%), and induction of demonstrated not to change significantly substrate consump- micronucleus formation in this cell line was not detected either.40a tion, biomass concentration and lipid content. This yeast strain Moreover, the presence of HMF protected the human liver can accumulate intracellular lipids as high as 60% of its cell dry cell line-LO2 against exposure to hydrogen peroxide, because weight in the presence of glucose, and the corresponding fatty it prevented nitric oxide production, caspase-3 activation and acids are similar to those of vegetable oil, making it an alternative arrest of the cells in the S-phase of the cell cycle.42 for production of biodiesel.33 Two strains of Escherichia coli In accordance with this data, HMF was present in processed (LY01 and KO11) have also been studied, and 4.0 g L-1 HMF ‘Fructus Corni’ used by the Chinese to invigorate the liver and terminated the growth of both within 24 h.34 Additionally, HMF kidney,42 and the same compound was also detected in processed at a concentration of 0.71 g L-1 was added to a culture of steamed ‘Rehmanniae Radix’, a natural remedy in Chinese Trichosporon cutaneum 2.1374, but it did not produce an obvious medicine used in several diseases such as anemia and diabetes.43 inhibitory effect on cell growth and lipid production.35 HMF has also been reported to be a promising candidate for therapy of sickle cell disease, since it binds efficiently to sickle haemoglobin, inhibiting sickling of red blood cells.44 3.2 Effects of HMF in humans On the other hand, contradictory results have been obtained As mentioned previously, some food can contain considerable in other experiments. The human colon cancer cell line CaCo- amounts of HMF, and some examples are dried fruits,36 2, the human epithelial kidney cell line HEK 293, the mouse coffee,36 cereals26d,36b and baking products.26b,26d,36–37 Addition- lymphoma cell line L5178Y, the Chinese hamster cell line ally, HMF has also been detected in medicinal fluids adminis- V79 and the human sulfotransferase SULT1A1 expressing V79 tered intravenously.38 Due to the daily consumption of these displayed DNA damage in the presence of HMF.45 Additionally, foods, the estimated daily intake of HMF is approximately a derived V79 cell line (V79-hCYP2E1-hSULT1A1)had a higher 30–150 mg per person.39 Studies with rats and dogs showed frequency of sister chromatin exchange (SCE) in the presence of

756 | Green Chem., 2011, 13, 754–793 This journal is © The Royal Society of Chemistry 2011 View Online

HMF.46 However, other reports demonstrated HMF-induced ∑ Problems in HMF synthesis. HMF is synthesized mainly DNA damage just at high concentrations and fail to correlate by the dehydration of monosaccharides, requiring the loss of 5-HMF-induced DNA damage with sulfotransferase-SULT1A1 three water molecules. Disaccharides or polysaccharides, such activity. Furthermore, HMF at a concentration of 23.71 mgmL-1 as sucrose, cellobiose, inulin or cellulose, can be used as starting induced 50% mortality of nauplii in the brine shrimp bioassay,47 materials, but hydrolysis is necessary for depolymerisation. while thermolyzed sucrose (which contains 1% HMF), when Sucrose hydrolysis is more efficiently catalyzed by a base; administered to female rats treated 1 week previously with the however, dehydration of the monomers is catalyzed by acids. colon carcinogen azoxymethane (AOM), enhanced the growth This introduces a problem, namely that the formation of HMF of the colonic aberrant crypt foci.48 by dehydration is a very complex process due to the possibility One of the hypotheses attempting to explain these differ- of side-reactions. Antal et al.56 reported the possible side- ent effects is the possibility that HMF is metabolized to a products formed by decomposition of fructose in water at high more harmful molecule such as 5-sulfooxymethylfurfural (5- temperatures, being products of isomerisation, dehydration, SMF),49 which can be produced through HMF sulfonation by fragmentation and condensation. The mechanism for fructose sulfotransferases.50 The fact that the Ames test for 5-SMF gives dehydration reaction is not clear, and two different pathways a positive result for Salmonella thyphimurium TA10049 seems have been proposed for the formation of HMF (Scheme 2).15,56 to give strength to this idea. 5-SMF was also demonstrated to exhibit a higher skin tumor initiating activity than HMF after its application on mouse skin.49,51 More recently, 5-SMF was quantified in vivo after intravenous injection of HMF in the mouse.52 A cytotoxic effect of 5-SMF was also reported in recombinant embryonic kidney cells: 5-SMF was shown to be a substrate for the organic anion transporters OAT1 and OAT3, and to decrease by 80% and 40% (respectively) the uptake of the substrates p-aminohippurate and estrone sulfate at a concentration of 1 mM, which indicates that 5-SMF can interfere with the transport of organic anions into renal proximal tubule cells, leading to kidney damage.53 Moreover, other studies did not demonstrate the presence of the sulfate metabolite in the urine of male F344 rats and B6C3F1 mice after administration of HMF (5, 10, 100, 500 mg kg-1), being about 60–80% of HMF excreted in the urine.54 None of this metabolite was detected in human subjects after consumption of dried plums and/or dried Scheme 2 15. Downloaded by University of Oxford on 06 April 2011 plum juice, although four other metabolites were detected: N-(5- hydroxymethyl-2-furoyl)glycine, 5-hydroxymethyl-2-furoic acid, ∑ Glucose vs. Fructose. Glucose (aldose) reactivity is lower Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D (5-carboxylic acid-2-furoyl)glycine and (5-carboxylic acid-2- than fructose (ketose), and this fact has been explained by the furoyl)aminomethane.36b In addition, HMF and 5-SMF were much lower abundance of acyclic glucose compared to acyclic bothtestedinMin/+ mice (heterozygous for a mutation in the fructose.13,57 Glucose can form a very stable ring structure, so tumor suppressor gene Apc), and despite increasing the number the enolisation rate in solution is lower than fructose, which of adenomas in the small intestine, they had no effect on their forms less stable ring structures.13 Since enolisation is the size, compared with the control mice, and thus were classified as rate-determining step for HMF formation, fructose will react weak intestinal carcinogens.55 much faster than glucose. On the other hand, fructose forms equilibrium mixtures of difructose and dianhydrides, and thus the most reactive groups are internally blocked, forming smaller 13 4 HMF synthesis amounts of by-products. Glucose forms true oligosaccharides which still contain reactive reducing groups, resulting in a greater This section will focus on the manufacture of HMF, taking risk of cross-polymerisation with reactive intermediates and into account reaction conditions, such as solvents, substrates HMF.13 and their concentrations, as well as catalysts and their reuse. ∑ HMF isolation methods. In most of the reported studies of Additionally, the mechanisms of the different synthetic method- HMF synthesis, the HMF was obtained in solution, and the yield ologies will be discussed. determined by HPLC or GC. However, it is important not only Several catalysts have been reported for the dehydration of to optimize the synthesis of this compound, but also to develop carbohydrates, and Cottier et al.14 organize them into five an efficient isolation method. HMF is not easy to extract from groups: organic acids, inorganic acids, salts, Lewis acids, and aqueous phase, since the distribution coefficient between the others. In recent years, carbohydrate dehydration catalysts have organic and the aqueous phase is not favourable.13,58 However, undergone a remarkable process of evolution, and several this problem has been overcome by the use of organic solvents new catalysts have been reported. Here, we group dehydration such as MIBK (methyl isobutyl ketone),59,60 DCM,60c ethyl reactions by the catalysis type: acid catalysis (homogeneous acetate,61 THF,62 diethyl ether,63 and acetone,64 which have been liquid, heterogeneous liquid–liquid, solid-liquid and gas-liquid) reported to be efficient extraction solvents. These could improve and metal catalysis. the synthesis of HMF, since they may avoid the formation of

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 754–793 | 757 View Online

Table 2 Conversion of carbohydrates to HMF using homogeneous catalysis by mineral or organic acidsa

Reaction conditions Post-reaction details Entry Biomass Temp. Conver- HMF Isolation/determi- Catalyst, reaction (ref.) source Solvent Catalyst (◦C) Time sion (%) selectivity (%) nation method medium reuse

57 1 Fructose H2OPTSA883.3h—~20 HPLC — 57 2 Fructose 1:1 H2O–PEG 4000 PTSA 88 3.3 h — ~45 HPLC — 57 3 Fructose H2O–CrCl3 PTSA 88 3.3 h — ~20 HPLC — 469 Fructose 1:1 Fructose–PEG HCl 180 10 s — 65 — — 6000 70 5 Fructose Ethylene glycol H2SO4 200 3.3 h 100 70.0 GC Solvent reused dimethyl ether 71 6 Fructose H2O HCl 200 1 s 52 63 HPLC — (27% aq. (MW) sol.) 71 7 Fructose H2O HCl 200 60 s 95 55 HPLC — (27% aq. (MW) sol.) 72 8 Fructose [BMIM][Cl] H2SO4 120 4 h 100 85 HPLC — 75 9 Fructose H2OHCl200 1 min 97 59 HPLC — (microreactor) (17 bar) 75 10 Fructose H2OHCl185 1 min 71 75 HPLC — (microreactor) (17 bar) 75 11 Fructose 1:2 H2O–DMSO / HCl 185 1 min 100 72 HPLC — (10 wt.%) MIBK–2-butanol (microreactor) (17 bar) 75 12 Fructose 1:5 H2O–DMSO / HCl 185 1 min 98 85 HPLC — (30 wt.%) MIBK–2-butanol (microreactor) (20 bar) 75 13 Fructose 1:5 H2O–DMSO / HCl 185 1 min 98 81 HPLC — (50 wt.%) MIBK–2-butanol (microreactor) (20 bar) 1476 Fructose 4:6 PTSA 100 30 min — 67 EtOAc — Fructose–choline extraction/HPLC chloride 1576 Inulin 5:5 PTSA 90 1 h — 57 EtOAc — Fructose–choline extraction/HPLC

Downloaded by University of Oxford on 06 April 2011 chloride 76 16 Glucose 4:6 CrCl2 110 30 min — 45 EtOAc —

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D Fructose–choline extraction/HPLC chloride 76 17 Sucrose 5:5 CrCl2 100 1 h — 62 EtOAc — Fructose–choline extraction/HPLC chloride

a PTSA, p-toluenesulfonic acid; DMSO, dimethyl sulfoxide; [BMIM][Cl], 1-butyl-3-methylimidazolium chloride; MIBK, methyl isobutyl ketone; MW, microwave irradiation.

by-products, such as soluble polymers or humins, among decreases, due to the formation of larger amounts of humins, others.65 Polar organic solvents, such as DMSO or DMF,65a,66 likely a result of reactions with HMF, fructose and their have a high boiling point, and due to the reactive nature of HMF intermediates (Table 2, entry 1). The addition of metal chlorides at high temperatures,60c,67 distillation is undesirable. Recently, it (Cr(III)orAl(III)) to the HCl-catalyzed dehydration improved was possible to isolate HMF by extraction with a low-boiling- the yield of HMF (Table 2, entry 3), but HMF rehydration point solvent in the presence of ionic liquids.61c,68 In the last was also enhanced.57 The formation of HMF was also affected few years, several improvements have been achieved in this field, by the pH, or the nature of the acid, but on the other hand, but more efficient separation techniques need to be developed the rehydration of HMF was not, so an increase of the acid in order to make synthesis economically viable for larger-scale concentration led to an increase in yield. The influence of water production. was also studied, performing the reaction in PEG. This was not an ideal solvent due to the possibility of the formation of HMF– PEG ethers that induce the formation of levulinic acid, although 4.1 Acid-based catalysis the yield of HMF could be improved by 45%. 4.1.1 Homogeneous catalysis. In 1986 Van Bekkum et al.57 HMF synthesis has already been reported using PEG 6000 studied the dehydration of fructose to HMF in acidic medium, as solvent.69 A mixture of PEG and fructose (1:1 w/w) became and observed that the carbohydrate concentration affected the homogeneous after heating and addition of a small amount HMF yield. For higher fructose concentrations the HMF yield of acid. Passage of this mixture through a tubular reactor, at

758 | Green Chem., 2011, 13, 754–793 This journal is © The Royal Society of Chemistry 2011 View Online

high temperatures (120–200 ◦C), led to reasonable HMF yields the main product was HMF (Table 2, entry 8), with 85% (Table 2, entry 4) with shorter reaction times, but ethers from yield. The authors reported that a similar result was obtained

HMF and PG-600 reaction were also obtained. The isolation when the reaction was carried out without H2SO4. For glucose of the product from this solvent was a drawback, due to the and mannose, although the conversion was almost complete, instability of HMF at high temperatures. the main product formed was not HMF, conﬁrming that One synthesis of HMF involved dissolving 1,2:4,5-di-o- dehydration of ketoses is quicker than aldoses.15 HMF stability

isopropylidene-b-D-fructopyranose in ethylene glycol dimethyl in [BMIM][Cl]/H2SO4 was studied, resulting in almost complete ether (EGDE) containing water and sulfuric acid as a catalyst.70 recovery of HMF (7% conversion and 1% solid residues As shown in Scheme 3, the ﬁrst step consists of the transforma- formation). Other stability studies of HMF in [BMIM][Cl] under tion of fructose into a fructose acetonide derivative, followed by various reaction conditions were performed,73 also leading to rapid dehydration to give HMF. The main advantages of this almost complete HMF recovery, showing that HMF is stable method is that high reactant concentrations can be achieved in [BMIM][Cl]. When HMF and glucose mixtures were tested ◦ using cheap and easily regenerated solvents, and the reactive ([BMIM][Cl]/H2SO4 at 120 C after 4 h), an increase of HMF hydroxyl groups of fructose which induce HMF instability are conversion (48%) and almost complete glucose conversion (96%) blocked at an earlier stage of the dehydration. However, for were observed. An increase of the solid residues was also noticed, economic reasons, it would be preferable to use a method that compared with the solid formed in the presence of just glucose. directly uses a biomass feedstock rather than another substrate This indicates that HMF in these conditions can react with that needs derivatization. monosaccharides or monosaccharide degradation products. The use of microreactors can have advantages when compared with conventional batch reactions, including better control of reaction conditions (temperature, pressure and residence time), improved safety, and portability.74 HCl-catalyzed dehydration of fructose in pure aqueous solutions was conducted in a 75 Scheme 3 70. continuous microreactor process (Table 2, entries 9–13). When this process was compared with an HCl-catalyzed dehydration Having three main goals in mind – an acid catalysed reaction of fructose in aqueous solution with microwave heating (Table 2, in 100% water, with HCl as the catalyst, and a feedstock of highly entries 6 and 7, 95% conv. with 55% HMF selectivity), at ◦ concentrated aqueous fructose – Hansen et al.71 reported the 200 C the HMF selectivity and fructose conversion were slightly microwave-assisted dehydration of fructose to HMF (Scheme 4). improved (Table 2, entry 9, 97% conv. 59% HMF selectivity, ◦ In this work, aqueous fructose (27 wt%) was irradiated with 200 C). This result was further improved by decreasing the ◦ microwaves for 1 s (200 ◦C), producing a conversion of 52% temperature to 185 C, which led to HMF with 75% selectivity with an HMF selectivity of 63%. For longer irradiation periods and 71% fructose conversion (Table 2, entry 10). To further

Downloaded by University of Oxford on 06 April 2011 (60 s), 95% conversion was achieved, but with a lower HMF improve the HMF selectivity the dehydration of fructose was selectivity, 55%, (Table 2, entries 6 and 7). Consequently, a slight carried out in aqueous solutions using DMSO as co-solvent Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D improvement was achieved compared to conventional heating and methyl isobutyl ketone/2-butanol as the extraction agent. (for example Table 3, entry 2160b). A higher fructose conversion (up to 98%) was achieved as well as a higher HMF selectivity of 85% (Table 2, entry 12). A high fructose concentration of 50 wt.% resulted in 98% conversion, with an 81% HMF selectivity (Table 2, entry 13). Highly concentrated melt systems consisting of choline chloride (ChCl) and up to 50 wt% of carbohydrates were tested in the dehydration reaction with different catalysts.76 For fructose and inulin the best catalyst was PTSA (p-toluenesulfonic acid) (Table 2, entries 14 and 15), and for glucose and sucrose the best

catalyst was CrCl2 with HMF yields of 45 and 62% respectively (Table 2, entries 16 and 17). Although some of the reactions tested were analysed by HPLC to determine the HMF yield, a method of extraction and evaporation with ethyl acetate was reported. A preliminary ecological evaluation was made and the recyclability of the process is being studied. Transformation of D-glucose was followed in a ﬂow reactor, and in high-pressure water and D-fructose, 5-HMF, furfural and 1,2,4-benzenetriol (BTO) yields were quantiﬁed.77 Glucose

71 conversion increased with the temperature (being about 100% Scheme 4 . ◦ at 400 C) and with the pressure. However, the yield of D- Almost complete conversion of fructose, glucose and mannose fructose was higher at 350 ◦C than 400 ◦C, and a decrease

were observed in the presence of a Brønsted acid, H2SO4 at of the pressure enhanced its yield. Longer residence times led 120 ◦C within 4 h, in an ionic liquid [BMIM][Cl] (1-butyl- to higher 5-HMF yield, but the effect was more evident at 3-methylimidazolium chloride).72 In the presence of fructose, 350 ◦C than at 400 ◦C. As a result, the highest 5-HMF yield

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 754–793 | 759 View Online

Table 3 Conversion of carbohydrates to HMF in heterogeneous (liquid–liquid) catalysis by mineral or organic acidsa

Reaction conditions Post-reaction details Entry Temp. Conver- HMF selec- Isolation/deter- (Ref.) Biomass source Solvent Catalyst (◦C) Time sion (%) tivity (%) mination method

160d Fructose (30 wt.%) 7:3 (8:2 HCl 200 3 min 89 85 MIBK–2-butanol

H2O–DMSO)–PVP / 7:3 (7:3) extraction MIBK–2-butanol 260d Fructose (50 wt.%) 7:3 (8:2 HCl 200 3 min 92 77 MIBK–2-butanol

H2O–DMSO)–PVP / 7:3 (7:3) extraction MIBK–2-butanol 60c 3 Glucose (10 wt.%) 4:6 H2O–DMSO / 7:3 HCl 170 10 min 43 53 MIBK–2-butanol MIBK–2-butanol extraction 60c 4 Glucose (10 wt.%) 5:5 H2O–DMSO / 7:3 HCl 170 17 min 50 47 MIBK–2-butanol MIBK–2-butanol extraction 60c 5 Glucose (10 wt.%) 3:7 H2O–DMSO / DCM — 140 4.5 h 62 48 DCM extraction 60c 6 Fructose (10 wt.%) 5:5 H2O–DMSO / 7:3 HCl 170 4 min 95 89 MIBK–2-butanol MIBK–2-butanol extraction 60c 7 Fructose (10 wt.%) 3:7 H2O–DMSO / DCM — 140 2 h 100 87 DCM extraction 60c 8 Inulin (10 wt.%) 5:5 H2O–DMSO / 7:3 HCl 170 5 min 98 77 MIBK–2-butanol MIBK–2-butanol extraction 60c 9 Inulin (10 wt.%) 3:7 H2O–DMSO / DCM — 140 2.5 h 100 70 DCM extraction 60c 10 Sucrose (10 wt.%) 4:6 H2O–DMSO / 7:3 HCl 170 5 min 65 77 MIBK–2-butanol MIBK–2-butanol extraction 60c 11 Sucrose (10 wt.%) 3:7 H2O–DMSO / DCM — 140 4.5 h 82 62 DCM extraction 60c 12 Cellobiose (10 wt.%) 4:6 H2O–DMSO / 7:3 HCl 170 10 min 52 52 MIBK–2-butanol MIBK–2-butanol extraction 60c 13 Cellobiose (10 wt.%) 3:7 H2O–DMSO / DCM — 140 9.5 h 85 45 DCM extraction 60c 14 Starch (10 wt.%) 4:6 H2O–DMSO / 7:3 HCl 170 11 min 61 43 MIBK–2-butanol MIBK–2-butanol extraction 60c 15 Starch (10 wt.%) 3:7 H2O–DMSO / DCM — 140 11 h 91 40 DCM extraction 1660b Fructose (30 wt.%) 35% aq. NaCl–1-butanol HCl 180 — 64 84 1-Butanol extraction 1760b Fructose (30 wt.%) 35% aq. NaCl–2-butanol HCl 180 — 71 79 2-Butanol extraction 1860b Fructose (30 wt.%) 35% aq. NaCl–1-hexanol HCl 180 — 78 72 1-Hexanol extraction 1960b Fructose (30 wt.%) 35% aq. NaCl–MIBK HCl 180 — 72 77 MIBK extraction 2060b Fructose (30 wt.%) 35% aq. NaCl / HCl 180 — 74 88 Toluene–2-butanol toluene–2-butanol extraction 2160b Fructose (30 wt.%) 35% aq. NaCl HCl 180 — 59 57 HPLC 2278 Fructose (30 wt.%) Sat. aq. NaCl–1-butanol HCl 180 35 min 87 82 HPLC

Downloaded by University of Oxford on 06 April 2011 2378 Fructose (30 wt.%) Sat. aq. KCl–1-butanol HCl 180 15 min 89 84 HPLC 2478 Fructose (30 wt.%) Sat. aq. CsCl–1-butanol HCl 180 15 min 92 80 HPLC 78

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D 25 Fructose (30 wt.%) 1-Butanol HCl 150 35 min 93 69 HPLC 2678 Fructose (30 wt.%) Sat. aq. NaCl–1-pentanol HCl 150 35 min 75 77 HPLC 2778 Fructose (30 wt.%) Sat. aq. NaCl–2-propanol HCl 150 35 min 39 80 HPLC 2878 Fructose (30 wt.%) Sat. aq. NaCl–2-butanol HCl 150 35 min 67 85 HPLC 2978 Fructose (30 wt.%) Sat. aq. NaCl–2-pentanol HCl 150 35 min 83 82 HPLC

3078 Fructose (30 wt.%) Sat. aq. NaCl–2-butanone HCl 150 35 min 84 82 HPLC 3178 Fructose (30 wt.%) 2-Butanone HCl 150 35 min 92 73 HPLC 3278 Fructose (30 wt.%) Sat. aq. NaCl–THF HCl 150 65 min 53 83 HPLC 3378 Fructose (30 wt.%) THF HCl 150 35 min 95 71 HPLC 3478 Fructose (30 wt.%) Sat. aq. NaCl–THF HCl 160 50 min 88 89 HPLC 80 35 Glucose H2O 1. Glucose 190 45 min 88.2 63.3 HPLC isomerase– sodium tetraborate; 2. HCl–NaCl– 1-butanol

a MIBK, methyl isobutyl ketone; PVP, poly(1-vinyl-2-pyrrolidinone); THF, tetrahydrofuran; DMSO, dimethyl sulfoxide.

(7%) was achieved at 350 ◦C, 80 MPa and 1.6 s residence time. and long residence times favoured the production of BTO and The yield of other products (furfural and BTO) also increased furfural. The authors suggested another pathway for formation with the temperature, pressure and residence times. The increase of furfural that is not derived from 5-HMF, since both are of solution density increased the rate of dehydration to 5- increased when the residence time increases.77 HMF and hydrolysis reaction of 5-HMF. High temperatures, high solution densities and short residence times seemed to be 4.1.2 Heterogeneous catalysis – liquid–liquid. Roman– advantageous for the selective synthesis of 5-HMF, preventing Leshkov et al.60d described an improved method of fructose formation of BTO. High temperatures, high solution density dehydration at high concentrations (30–50 wt.%), with an acid

760 | Green Chem., 2011, 13, 754–793 This journal is © The Royal Society of Chemistry 2011 View Online

catalyst, involving the addition of modiﬁers in both reaction to recycle the water, NaCl, a fraction of 1-butanol, and 58% of phases (Fig. 1). As reported before,57 high concentrations of the HCl. The authors did not report the isolation of HMF from fructose increase the amount of side-products. To overcome the extraction solvent. this problem, the authors added to the aqueous phase a polar To improve this biphasic reaction system with a salting-out

aprotic solvent (DMSO or 1-methyl-2-pyrrolidinone (NMP)) effect, different classes of C3–C6 extraction solvents, such as and/or a hydrophilic polymer (poly(1-vinyl-2-pyrrolidinone) aliphatic alcohols, ketones, and ethers were tested.78 Solvents

(PVP)), improving the HMF selectivity (Table 3, entries 1 and with four carbon atoms (C4) generated the highest HMF 2). In this work methyl isobutyl ketone (MIBK) was used as selectivity values within each solvent class. These solvents the extraction solvent, and the addition of these modifiers to showed the highest affinity for HMF, coupled with low water the aqueous phase increased HMF solubility in the aqueous miscibility at the reaction temperature (Table 3, entries 28 phase, hampering the extraction process with methyl isobutyl and 30–32). The increase of temperature induces higher HMF ketone (MIBK). However, adding 2-butanol to the organic layer selectivity, but on the other hand, the reaction temperature has raised HMF solubility, improving the extraction process. Several to be sufficiently low to avoid solvent degradation reactions. ◦ mineral acids were tested (H2SO4,H3PO4 and HCl), HCl having Thus, the reactions were performed at 180 C. Using 1-butanol the best HMF selectivity. as the extraction solvent, the effect of different salts on the dehydration reaction was studied. It was shown that KCl and NaCl generated the best combination of extracting power and high HMF selectivity (Table 3, entries 22–24).78 The authors showed that HMF selectivity could be improved using saturated NaCl solutions, but the conversion of fructose was higher when no salt was added to the aqueous phase (Table 3, entries 22 vs.25). However, the authors did not provide details about the isolation of the final product. One of the main routes for the transformation of glucose Fig. 1 to HMF involves an isomerisation step to fructose, followed by fast dehydration.60a,63,66,79 In view of this, Huang et al.80 Further work was carried out60c by the same authors, in reported the synthesis of HMF using an enzymatic glucose which they optimized the method for dehydration of glucose, isomerisation to fructose with a borate-assisted isomerase, achieving up to 53% HMF selectivity (Table 3, entries 3 and 4). followed by dehydration in an acidic medium (Scheme 5) with 1- In this work HCl was used as the acid catalyst, in an aqueous butanol as the extraction solvent. With this system, a 88.2% phase with DMSO as co-solvent, and an extracting phase with sugar conversion and 63.3% HMF selectivity were obtained MIBK–2-butanol, or dichloromethane. The best conditions within 45 min (Table 3, entry 35). A low percentage of soluble

Downloaded by University of Oxford on 06 April 2011 achieved for the dehydration of glucose were applied to other by-products (levulinic and formic acids) was formed in this case, saccharides such as inulin (a polyfructan), starch (a polyglucan), but for high reaction times, the HMF yield started to decrease, Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D cellobiose (a glucose dimer) and sucrose (a disaccharide of suggesting the formation of humins. glucose and fructose), just by adjusting the pH and DMSO content (Table 3, entries 6–15). The main problem observed was the carry-over of DMSO to the organic phase, and therefore HMF separation from the DMSO at the end of the process was necessary. This is very complicated due to the reactive nature of HMF at high temperatures,60c,67 meaning that high-temperature distillation is not possible. Therefore, separation techniques need to be developed to make these efficient methods of synthesis economically viable for larger-scale production. Roman–Leshkov et al.60b reported the acid-catalysed dehydra- Scheme 5 80. tion of fructose in a biphasic reactor, using different extraction A patent published in 200981 claims that a biphasic reaction solvents (Table 3, entries 16–21). Since HMF selectivity increases with dioxane as extraction solvent decreases the process time. with the efficiency of the extraction solvent, NaCl was added The reactive aqueous solution uses sulfuric acid as catalyst and to the aqueous phase to increase the extraction efficiency by fructose as substrate. a salting-out effect, and 1-butanol was used as the extracting

solvent. The advantages of this method are that no DMSO 4.1.3 Heterogeneous catalysis – gas–liquid. The CO2–water

was added to the aqueous phase, because the addition of system can replace conventional acids such as HCl and H2SO4 the salt prevents the system from becoming single-phase, as for the catalysis of some chemical reactions, with the advantage happened when 1-butanol was the extraction solvent in a system of the solution being able to be neutralized by depressurization without NaCl. On the other hand, 1-butanol is a biorenewable without requiring salt disposal.82 The main reason that it works

solvent that can be obtained by biomass-derived carbohydrate is that CO2 in aqueous solution can generate carbonic acid fermentation.60b The fructose concentrations were higher than in situ, which acts as the catalyst for the reaction. There is a 60c previous results, but the conversion and HMF selectivity were correlation between the pH solution and CO2 pressure, lower

not so good (Table 3, entries 6 vs.16). With this system is possible pH values resulting from higher CO2 pressures. The optimum

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 754–793 | 761 View Online

Table 4 Conversion of carbohydrates to HMF catalyzed by mineral or organic acidsa

Reaction conditions Post-reaction details Entry Conver- HMF Isolation/deter- (ref.) Biomass source Solvent Catalyst Temp. (◦C) Time sion (%) selectivity (%) mination method

82 1 Inulin H2O6MPaCO2 200 45 min 100 53 HPLC 83 2 Fructose (0.05 M) Subcritical H2O — 240 120 s — — HPLC HCl 240 120 s — 44.7 HPLC

H2SO4 240 120 s — 40.3 HPLC H3PO4 240 120 s — 65.3 HPLC Citric acid 240 120 s — 49.3 HPLC Maleic 240 120 s — 60.0 HPLC acid PTSA 240 120 s — 37.0 HPLC Oxalic acid 240 120 s — 17.4 HPLC 83 3 L-Sorbose (0.05 M) Subcritical H2OH3PO4 240 120 s — 50.0 HPLC 83 4 D-Mannose (0.05 M) Subcritical H2OH3PO4 240 120 s — 31.1 HPLC 83 5 D-Galactose (0.05 M) Subcritical H2OH3PO4 240 120 s — 27.3 HPLC 83 6 D-Glucose (0.05 M) Subcritical H2OH3PO4 240 120 s — 30.0 HPLC 83 7 Sucrose (0.05 M) Subcritical H2OH3PO4 240 120 s — 40.1 HPLC 83 8 Cellobiose (0.05 M) Subcritical H2OH3PO4 240 120 s — 27.2 HPLC 84 9 Fructose Supercritical H2SO4 180 (20 MPa) 120 s — 77 HPLC acetone–H2O (90:10) 84 10 Glucose Supercritical H2SO4 180 (20 MPa) 120 s — 48 HPLC acetone–H2O (90:10) 84 11 Sucrose Supercritical H2SO4 180 (20 MPa) 120 s — 56 HPLC acetone–H2O (90:10) 84 12 Inulin Supercritical H2SO4 180 (20 MPa) 120 s — 78 HPLC

acetone–H2O (90:10)

a PTSA, p-toluenesulfonic acid.

Downloaded by University of Oxford on 06 April 2011 CO2 pressure was achieved at 6 MPa, resulting in a maximum HMF yield for a variety of temperatures, which indicates that an

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D optimum reaction pH is obtained with this pressure. Also, the effects of temperature, time reaction and initial concentration of inulin on the HMF yield were studied by Han et al.82 Similar behaviour was observed when these conditions were varied: the HMF yield increased until it reached a maximum value, after which it started decreasing. This was explained by the high reactivity of HMF,which results in the formation of by-products.

Optimal conditions were obtained with a CO2 pressure of 6 MPa 83 at 200 ◦C and a reaction time of 0.75 h, which led to a 53% yield Scheme 6 . of HMF (Table 4, entry 1). pH the formation of soluble polymers is favoured. In this study, 4.1.4 Subcritical or supercritical solvents. Without catalyst, hydrochloric, sulfuric, phosphoric, oxalic, citric, maleic, and p- using a temperature in the range 473–593 K, and a residence toluenesulfonic acids were tested as catalysts for the dehydration time up to 900 s, many compounds were produced (furfural, of fructose in subcritical water (Table 4, entry 2). Phosphoric acid humin, soluble polymers, aldehydes, ketones, monosaccharides was the best catalyst used at lower pH and the best condition and organic acids, Scheme 6) during production of HMF from was achieved at pH = 2 (HMF yield = 65.3%), but at higher fructose in subcritical water.83 With increasing temperature, the pH HCl was the best (33.4% HMF yield at pH 3.0); however, amount of by-products increased, especially formic and lactic pH = 2 was shown to lead to the best result. Under subcritical acids. The amount of HMF produced increased until 530 K, water conditions, the HMF yield was shown to decrease with the decreasing for higher temperatures with signiﬁcant production initial amount of fructose. Low fructose concentrations (0.05 M) of formic, lactic and acetic acids. The reaction time favoured were used, and several other mono- and di-saccharides were also HMF production until approximately 200 s, and after this time tested (Table 4, entries 3–8). the HMF was slowly converted to the organic acids. Asghari Different parameters, such as temperature, pressure and et al.83 have also reported that at lower pH the rehydration of reaction time, were studied by Bicker et al.84 for the fructose HMF to levulinic and formic acids occurs, whereas for higher dehydration reaction in a supercritical acetone–water mixture,

762 | Green Chem., 2011, 13, 754–793 This journal is © The Royal Society of Chemistry 2011 View Online

and with sulfuric acid as catalyst. Fructose is a carbohydrate with In 200789 Dumesic et al. reported the fructose dehydration on low solubility in acetone and therefore, water was added to the solvent systems containing NMP (1-methyl-2-pyrrolidinone) as system to enhance the fructose concentration. A similar system additive in the aqueous phase. MIBK (methyl isobutyl ketone) butinsubcriticalwater83 gave a lower HMF selectivity, especially or dichloromethane were used as extraction solvents. Different with glucose and sucrose as substrates (Table 4, entries 9–12 substrates, such as fructose, inulin and sucrose were studied using vs. 2, 6, 7). There were no solids (humins) produced, although an ion exchange resin as catalyst. For inulin, the dehydration other by-products such as furfural, glucose, methylglyoxal, reaction was complete with an HMF selectivity of 69% (Table 5, dihydroxyacetone and levulinic acid were formed, but with a entry 13). Using sucrose as starting material under the same yield lower than 6%. Years later, this group reported85 the reaction conditions (Table 5, entry 15), only the monomer study of the same reaction conducted in sub- and supercritical fructose reacted, providing a conversion of 60%, with an HMF methanol and subcritical acetic acid, obtaining furfural-ether selectivity of 74%. The dehydration reaction occurred even in (5-methoxymethylfurfural with 79% selectivity and 99% con- the absence of the resin catalyst, with similar HMF selectivity, version) and furfural-ester (5-acetoxymethylfurfural with 38% although the presence of the catalyst allowed a decrease in selectivity and 98% conversion). temperature from 120 to 90 ◦C. The authors studied the DMSO content in the aqueous phase, and concluded that the increase 4.1.5 Heterogeneous catalysis – solid–liquid. of the amount of DMSO also increased the HMF selectivity. However, for higher amounts of DMSO the carry-over to the Ion exchange resins. Vinke et al.86 reported fructose dehydra- organic phase increased, complicating the HMF separation tion using a dehydration set-up consisting of a column with an procedure. The same behaviour was observed when NMP ion exchange resin as catalyst and a separate loop for adsorption was used as additive in the aqueous phase. Dichloromethane of HMF onto the activated carbon, to avoid the formation of (DCM) was tested as extraction solvent (Table 5, entries 12, side-products. Therefore, the HMF is selectively adsorbed in the 14 and 16), providing good reaction rates and selectivity. carbon during the reaction, and then is extracted with organic However the DMSO carry-over in DCM is much higher than solvents. Thus, the HMF recovery is influenced by the presence for MIBK, complicating once again the HMF separation of catalytic sites on the carbon and the pH of the extracting procedure. solution. Although the reaction temperature was not so high, It has been reported that organic solvents such as DMSO can the reaction time was 48 h, but still 77% HMF selectivity could improve fructose dehydration, avoiding the formation of by- be achieved with this system (Table 5, entry 1). products65a,87b such as levulinic acid and humins from HMF, but Improvement of HMF selectivity has been achieved by this solvent has the disadvantage of difficult separation from the carrying out fructose dehydration in organic solvents instead final product. To overcome this problem, an acetone–DMSO of aqueous solutions, DMSO being the solvent selected for this (7:3) solvent system in the presence of a strong acid cation- transformation.65a,65c,87 Halliday et al.87b reported the one-pot exchange resin catalyst was used , with microwave irradiation

Downloaded by University of Oxford on 06 April 2011 synthesis of 2,5-diformylfuran (DFF), via fructose dehydration as the heating source.65c The authors claimed that the use of to HMF. The authors claimed that they tried to reproduce acetone is beneficial due to the its low boiling point, which makes Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D several reported methods for HMF synthesis, and they did not separation of the final product easier. Due to the low solubility obtain the expected yields. Consequently, they reported HMF of fructose in acetone, DMSO was used as co-solvent. Two synthesis with an ion-exchange resin in DMSO, for 5–25 h, at different fructose concentrations were tested, and for the same 100 ◦C, with high HMF selectivity (Table 5, entries 2 and 3). reaction conditions the HMF selectivity for the higher fructose In 2009 Shimizu et al.88 reported fructose dehydration in concentration decreased, although not significantly (Table 5, DMSO and tested several heterogeneous catalysts (heteropoly entries 17 and 18). It was possible to recycle the catalyst for at acid, zeolite and acidic resin). In order to extract the water least 5 cycles without loss of selectivity or efficiency. formed during the reaction, a mild evacuation method was In 2008, Watanabe and co-workers reported fructose dehy- developed by performing the reaction under vacuum (0.97 ¥ dration starting from different substrate concentrations.90 They 105 Pa). Thereby not only the fructose conversion was improved started the study by heating an aqueous solution of fructose to 100%, but also the HMF yield was increased to 97% with a microwave irradiation (150 ◦C) with a resin as catalyst,

(FePW12O40 Table 5, entry 8). The authors proved that this mild and although they obtained a good conversion (82.6%), the evacuation method gives better results than using molecular HMF selectivity was very low (Table 5, entry 19). To attain a sieves as water evacuation agents (Table 5, entries 8, 9 and 10). better yield, acetone was chosen as co-solvent, since it has a low Another extraordinary result was achieved when the Amberlyst boiling point (56 ◦C), and fructose has been shown to rearrange 15 catalyst was reduced to a powder (0.15–0.053 mm). For to the furanoid form in acetone–water mixtures, favouring this situation, complete transformation of fructose into HMF HMF formation.90 Using the same reaction conditions, a slight was observed (100% conversion with 100% selectivity) even at decrease of HMF selectivity was obtained for an increased 50 wt.% fructose solutions in DMSO (Table 5, entries 6 and amount of fructose (Table 5, entries 20–23). When the same 7). This result was achieved even without water evacuation. The reaction was heated in a sand bath, the fructose conversion and authors proposed that reduction of the particle size enhances the HMF yield decreased to 22.1% and 13.7%, respectively, while removal of adsorbed water from the surface and near-surface of the corresponding value for microwave heating was 91.7% and the catalyst.88 This result appears to be the best one achieved so 70.3%, respectively (Table 5, entry 20). It was possible to recycle far, but unfortunately the presence of DMSO makes isolation of the catalyst for at least 5 cycles without the lost of selectivity or the final product difficult. efficiency.

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 754–793 | 763 View Online

Table 5 Conversion of carbohydrates to HMF under heterogeneous conditionsa

Reaction conditions Post-reaction details Catalyst, Isolation/ reaction Entry Biomass Temp. Conver- HMF selec- determination medium (ref.) source Solvent Catalyst (◦C) Time sion (%) tivity (%) method reuse

86 1 Fructose H2O Ion-exchange 90 48 h — 77 HPLC — resin/activated carbon 287b Fructose DMSO Dowex-type 110 5 h 100 85 GC–MS — ion-exchange resin 387b Fructose DMSO Dowex-type 80 25.5 h 100 77 GC–MS — ion-exchange resin 488 Fructose DMSO Amberlyst 15 120 2 h 100 92 HPLC — pellets with evacuation (0.97 ¥ 105 Pa) 588 Fructose DMSO Amberlyst 15 120 2 h 100 76 HPLC — Pellets (0.71– 0.5 mm) 688 Fructose DMSO Amberlyst 15 120 2 h 100 100 HPLC — Powder (0.15–0.053 mm) 788 Fructose DMSO Amberlyst 15 120 2 h 100 100 HPLC — (50 wt.%) Powder (0.15–0.053 mm) 88 8 Fructose DMSO FePW12O40 with 120 2 h 100 97 HPLC — evacuation (0.97 ¥ 105 Pa) 88 9 Fructose DMSO FePW12O40 120 2 h 100 49 HPLC — 88 10 Fructose DMSO FePW12O40 with 120 2 h 100 69 HPLC — evacuation (4 A˚ sieves ) 89 11 Fructose 4:6 H2O–NMP / Ion exchange 90 18 h 98 85 MIBK — (10 wt.%) MIBK resin (DiaionR ) extraction 89 12 Fructose 5:5 H2O–DMSO / — 120 5.5 h 92 80 DCM — Downloaded by University of Oxford on 06 April 2011 (10 wt.%) DCM extraction 89 13 Inulin 4:6 H2O–NMP / Ion exchange 90 21 h 100 69 MIBK — R

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D (10 wt.%) MIBK resin (Diaion ) extraction 89 14 Inulin 5:5 H2O–DMSO / — 120 6.5 h 100 61 DCM — (10 wt.%) DCM extraction 89 a 15 Sucrose 5:5 H2O–NMP / Ion exchange 90 21 h 58 74 MIBK — (10 wt.%) MIBK resin (DiaionR ) extraction 89 a 16 Sucrose 5:5 H2O–DMSO / — 120 6.5 h 60 69 DCM — (10 wt.%) DCM extraction 1765c Fructose Acetone–DMSO Dowex-type 150 (MW) 5 min 88.2 89.6 HPLC Ion (2 wt.%) (7:3) ion-exchange 20 min 99.0 88.3 exchange resin resin recycled for 5 cycles 1865c Fructose Acetone–DMSO Dowex-type 150 (MW) 20 min 99.0 84.1 HPLC — (10 wt.%) (7:3) ion-exchange 30 min 99.4 82.1 resin 90 19 Fructose H2O Dowex-type 150 60 min 82.6 34 HPLC — (2 wt.%) ion-exchange resin 90 20 Fructose Acetone–H2O Dowex-type 150 (MW) 10 min 91.7 70.3 HPLC Catalysts (2 wt.%) (70:30 w/w) ion-exchange recycled resin for at least 5 cycles 90 21 Fructose Acetone–H2O Dowex-type 150 (MW) 10 min 98.6 66.6 HPLC — (5 wt.%) (70:30 w/w) ion-exchange resin 90 22 Fructose Acetone–H2O Dowex-type 150 (MW) 10 min 99.6 52.7 HPLC — (10 wt.%) (70:30 w/w) ion-exchange resin

764 | Green Chem., 2011, 13, 754–793 This journal is © The Royal Society of Chemistry 2011 View Online

Table 5 (Contd.)

90 23 Fructose Acetone–H2O Dowex-type 150 10 min 98.1 51.5 HPLC — (20 wt.%) (70:30 w/w) ion-exchange resin (MW) 90 24 Fructose Acetone–H2O Dowex-type 150 10 min 22.1 13.7 HPLC — (2 wt.%) (70:30 w/w) ion-exchange resin 2573b Fructose [BMIM][Cl] Amberlyst 15 80 10 min 98.6 83.3 HPLC Catalyst/solvent (20 wt.%) recycling for at least 7 cycles 2673b Fructose [BMIM][Cl] Amberlyst 15 120 1 min 99.3 82.2 HPLC — (20 wt.%) 2773a Fructose [BMIM][Cl]– Amberlyst 15 25 6 h 90.3 86.5 HPLC — (20 wt.%) acetone (RT) 87a 28 Fructose [BMIM][BF4]– Amberlyst 15 80 32 h — 87 HPLC/UV — DMSO (5:3) 87a 29 Fructose [BMIM][BF4]– PTSA 80 32 h — 68 HPLC/UV — DMSO (5:3) 87a 30 Fructose [BMIM][PF6]– Amberlyst 15 80 24 h — 80 HPLC/UV — DMSO (5:3) 87a 31 Fructose [BMIM][PF6]– PTSA 80 20 h — 75 HPLC/UV — DMSO (5:3) 87a 32 Fructose [BMIM][BF4] Amberlyst 15 80 3 h — 52 HPLC/UV — 3366 Fructose DMF HT/Amberlyst 15 100 3 h 99 76 HPLC — 3466 Glucose DMF Amberlyst 15 100 3 h 69 0 HPLC — 3566 Glucose DMF HT/Amberlyst 15 80 9 h 73 58 HPLC Catalysts recycled for at least 3 cycles 3666 Sucrose DMF HT/Amberlyst 15 120 3 h 58 93 HPLC — 3766 Cellobiose DMF HT/Amberlyst 15 120 3 h 52 67 HPLC — 59 38 Fructose H2O–MIBK (1:5) Dealuminated 165 2 93 73 MIBK — H-form mordenites extraction/HPLC 61d 39 Glucose [EMIM][Cl] H2SO4 120 3 h 93 66 EtOAc extraction — 61d 40 Glucose [EMIM][Cl] CF3SO3H 120 3 h 87 46 EtOAc extraction — 61d 41 Glucose [EMIM][Cl] HNO3 120 3 h 56 77 EtOAc extraction —

Downloaded by University of Oxford on 06 April 2011 61d 42 Glucose [EMIM][Cl] CF3COOH 120 3 h 58 75 EtOAc extraction — 4361d Glucose [EMIM][Cl] HCl 120 3 h 53 62 EtOAc extraction — 61d

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D 44 Glucose [EMIM][Cl] CH3SO3H 120 3 h 73 58 EtOAc extraction — 61d 45 Glucose [EMIM][Cl] H3PO4 120 3 h 17 95 EtOAc extraction — 4661d Glucose [EMIM][Cl] 12-TPA 120 3 h 82 81 EtOAc extraction — 4761d Glucose [EMIM][Cl] 12-MPA 120 3 h 71 89 EtOAc extraction — 4861d Glucose [EMIM][Cl] 12-TSA 120 3 h 69 82 EtOAc extraction — 4961d Glucose [BMIM][Cl] 12-MPA 120 3 h 71 89 EtOAc extraction — 5061d Glucose [BDMIM][Cl] 12-MPA 120 3 h 57 88 EtOAc extraction — 5161d Glucose [BMPy][Cl] 12-MPA 120 3 h 52 87 EtOAc extraction — 5261d Glucose [EMIM][Cl]– 12-MPA 120 3 h 99 98 EtOAc extraction — acetonitrile 5361d Glucose [BMIM][Cl]– 12-MPA 120 3 h 99 98 EtOAc extraction — acetonitrile

a HMF selectivity is based on fructose content. [BMIM][Cl], 1-ethyl-3-methylimidazolium chloride; [BMIM][BF4], 1-butyl 3-methyl imidazolium tetraﬂuoroborate; [BMIM][PF6], 1-butyl 3-methyl imidazolium hexaﬂuorophosphate; DMSO, dimethyl sulfoxide; DMF, dimethylformamide, NMP, 1-methyl-2-pyrrolidinone; MIBK, methyl isobutyl ketone; DCM, dichloromethane; HT, Mg–Al hydrotalcite; TPA, 12-tungstophosphoric

acid (H3PW12O40); MPA, 12-molybdophosphoric acid (H3PMo12O40); TSA, 12-tungstosilicic acid (H3SiW12O40); MSA, 12-molybdosilicic acid

(H3SiMo12O40); [BDMIM][Cl], 1-butyl-2,3-dimethylimidazolium chloride; [BMPy][Cl], 1-butyl-3-methylpyridinium chloride; [EMIM][Cl], 1-ethyl-3- methylimidazolium chloride.

A dehydration of fructose using the ionic liquid [BMIM][Cl] account, and for water contents above 5 wt.% the conversion with the sulfonic ion-exchange resin, Amberlyst 15, as catalyst yield and HMF selectivity decreased. Although there were no was developed by Qi et al.73b Several mineral and Lewis acids, products of HMF self-polymerization, other by-products were and solid acid anion-exchange resins were tested, the best found by HPLC, such as glucose, levulinic acid and formic acid, catalyst being Amberlyst 15. This catalytic system resulted in a but in very low yields. The authors tested the HMF stability 98.6% fructose conversion with a selectivity of 83.3% for HMF in this catalytic system by adding an HMF sample under the at 80 ◦C and a reaction time of 10 min (Table 5, entries 25 and same reaction conditions, recovering 99.8% after the reaction 26). The water content of the ionic liquid was also taken into time. The HMF was extracted with ethyl acetate, and therefore

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 754–793 | 765 View Online

the ionic liquid and the catalyst could be recycled for at least 7 cycles.73b The same authors73a reported the conversion of fructose into HMF,again with Amberlyst 15 as catalyst and using [BMIM][Cl] as the solvent, but at room temperature (Table 5, entry 27). The viscosity of [BMIM][Cl] at room temperature is high, so it was not possible to stir the reaction solution without the addition of a co-solvent. Small amounts of different co-solvents were added to the reaction mixture, such as DMSO, methanol, ethanol, ethyl acetate, supercritical carbon dioxide, and acetone, the latter being the most efficient. The conversion and HMF yield were above 80%, independent of the organic solvent used. To improve the reaction efficiency, fructose had to be pre-dissolved in the Scheme 7 66. ionic liquid in a water bath at 80 ◦C for 20 min. The time for 73b reaction was higher than in the previous work (6 h vs. 10 min) different Si/Al ratios. The selectivity decreased when increasing but this method has the advantage of being performed at room the Si/Al ratio, i.e. by increasing the acidic properties of the temperature. By-products were also formed, but in yields lower catalysts. The optimum Si/Al ratio achieved was l : 1, and the 73a than 2%. high selectivity obtained (Table 5, entry 38) was correlated Two ionic liquids, one hydrophobic and one hy- with the shape selectivity properties of H-mordenites (bidimen- drophilic (1-butyl-3-methyl imidazolium hexafluorophosphate sional structure), and particularly with the absence of cavities [BMIM][PF6], and 1-butyl-3-methylimidazolium tetrafluorob- within the structure allowing further formation of secondary orate [BMIM][BF4]), were tested as solvents for fructose de- products. hydration with Amberlyst 15 or PTSA as catalysts.87a The Several liquid (H2SO4,CF3SO3H, CH3SO3H, CF3COOH, addition of DMSO as co-solvent increased the HMF selectivity, HNO3,HClandH3PO4) and solid acids, [12-tungstophosphoric mainly due to an increase on fructose solubility. The fructose acid (12-TPA (H3PW12O40)), 12-molybdophosphoric acid dehydration reaction in [BMIM][Cl],73a at the same temperature, (12-MPA (H3PMo12O40)), 12-tungstosilicic acid (12-TSA was much faster and more selective than when performed with (H3SiW12O40)), and 12-molybdosilicic acid (12-MSA 87a [BMIM][BF4] (Table 5, entries 25 vs. 32). This may be due (H3SiMo12O40))] were tested as catalysts for the glucose to fructose solubility differences between [BMIM][BF4]and dehydration to HMF, with [EMIM][Cl] as solvent (Table 5, [BMIM][Cl]. The addition of DMSO to [BMIM][BF4]increased entries 39–53).61d With all of these catalysts, 4 to 20% of the HMF yield to a higher value than obtained with [BMIM][Cl] humins and others by-products were formed. 12-MPA was (Table 5, entries 28 vs. 25). chosen for further studies due to the best performance and Downloaded by University of Oxford on 06 April 2011 The ‘site isolation’ concept of supported reagents allows the selectivity (Table 5, entry 47, 71% glucose conversion with 89% simultaneous use of otherwise incompatible reactive function- HMF selectivity). With this catalyst several other ionic liquids, Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D 66 alities, such acid/base pairs. Takagaki et al. reported glucose such as [EMIM][Cl] (1-ethyl-3-methylimidazolium chloride), dehydration to HMF catalyzed by a solid acid/base catalyst [BDMIM][Cl] (1-butyl-2,3-dimethylimidazolium chloride) via one-pot reaction under mild conditions. The base catalyst and [BMPy][Cl] (1-butyl-3-methylpyridinium chloride) were is necessary for isomerisation of glucose to fructose, and the tested (Table 5, entries 49–51). There was a lower activity of acidic catalyst will catalyze the dehydration reaction (Scheme 7). 12-MPA in the ionic liquids [BDMIM][Cl] and [BMPy][Cl] Firstly, two individual reactions were screened to identify the than the other two. With [BDMIM][Cl] the lower activity may best solid catalyst for both reactions. The best base catalyst be due the acidic proton lost from the imidazolium cation. for the glucose isomerisation was Mg–Al hydrotalcite (HT), So the loss of the acidity in the solvent (in this case, the ionic consisting of layered clays with HCO3 groups on the surface. liquid) reduces the glucose conversion and HMF selectivity. For the fructose dehydration Amberlyst 15 was chosen as the The addition of acetonitrile to the ionic liquid as co-solvent acid catalyst. The combination of these two solid catalysts enhances the glucose conversion up to 99%, with a 98% HMF improved the glucose transformation from zero selectivity to selectivity, with [BMIM][Cl] or [EMIM][Cl] as solvent, and no 76% (Table 5, entries 34 vs. 33), improving the conversion formation of humins (Table 5, entries 52 and 53). A glucose as well. Other substrates were tested also with high HMF dehydration mechanism by 12-MPA was proposed, where yields (Scheme 7, Table 5, entries 36 and 37). The reaction the authors believed that a key intermediate in the reaction solvent was DMF (N,N-dimethylformamide), but others, such pathway to HMF was a 1,2-enediol. The high selectivity of as DMSO and acetonitrile, were also tested with good results. heteropoly acids was attributable to stabilization of the reaction However,it is important to highlight again that the use of DMSO intermediates involved in formation of HMF. In the absence of makes HMF separation difficult, and is not ideal for industrial acetonitrile as a co-solvent, moderate amounts of humins were production. formed.61d Other solid catalysis. Dehydration of fructose to HMF was studied in a batch mode in the presence of dealuminated H- 4.1.6 Solvents as reaction promoters. In 1987 Musau form mordenites as catalysts, at 165 ◦C,andinasolventmixture et al.65a demonstrated that fructose can be converted into consisting of water and methyl isobutyl ketone (1:5 by volume).59 HMF in the presence of DMSO as solvent at 150 ◦C, with The HMF selectivity was optimized testing H-mordenite with no catalyst added (Table 6, entry 1). They tested different

766 | Green Chem., 2011, 13, 754–793 This journal is © The Royal Society of Chemistry 2011 View Online

Table 6 Conversion of carbohydrates to HMF using solvents as reaction promotersa

Reaction conditions Post-reaction details Entry Biomass Temp. Conver- HMF selec- Isolation/ Isolated Catalyst, reaction (ref.) source Reaction medium Catalyst (◦C) Time sion (%) tivity (%) determination method yield (%) medium reuse

165a Fructose DMSO — 150 2 h — 92 EtOAc —— extraction/silica gel chromatography with DCM 268 Fructose 1-H-3-methyl — 90 45 min 100 92 Diethyl ether extraction 86 Solvent/catalyst imidazolium chloride reused for 5 cycles 394 Fructose [ASBI][Tf] / DMSO — 100 6 min 98 80 HPLC — — (MW) 494 Fructose [ASCBI][Tf] / DMSO — 100 6 min 100 84 HPLC — — (MW) 94 5 Fructose ILIS–SO3H / DMSO — 100 4 min 100 70.1 HPLC — — (MW) 94 6 Fructose ILIS–SO2C l/ DMSO — 100 4 min 100 67.2 HPLC — — (MW) 761a Fructose Choline — 80 1 h 91.1 83.8 EtOAc extraction 72.2 Ionic liquid chloride–citric acid reused for 8 cycles 861a Fructose Choline —801h92<20 EtOAc extraction — —

chloride–CrCl3 961a Fructose Choline —801h25<7 EtOAc extraction — —

chloride–ZnCl2 1061c Inulin Choline — 80 2 h 100 56 EtOAc extraction — Ionic liquid chloride–oxalic acid reused for 6 cycles 1161c Inulin Choline — 80 2 h 100 64 EtOAc —— chloride–oxalic acid / extraction/HPLC ethyl acetate 1261c Inulin Choline — 50+80 2+2 h 88 65 EtOAc —— chloride–citric acid extraction/HPLC 1395 Fructose Pyridinium chloride — 120 30 min — 70 Silica gel 70 — chromatography (ethyl acetate–petroleum ether) 1495 Inulin Pyridinium chloride — 120 30 min — 60 — — — 1595 Levan Pyridinium chloride — 120 30 min — 60 — — — 1695 Glucose Pyridinium chloride — 120 30 min — 5 — — —

Downloaded by University of Oxford on 06 April 2011 1795 Sucrose Pyridinium chloride — 120 30 min — 30 Silica gel 30 — chromatography (ethyl

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D acetate–petroleum ether)

a [ASBI][Tf], 3-allyl-1-(4-sulfobutyl)imidazolium triﬂuoromethanesulfonate; [ASCBI][Tf], 3-allyl-1-(4-(sulfurylchloride)butyl)imidazolium triﬂuo- romethanesulfonate; DMSO, dimethyl sulfoxide; [EMIM][Cl], 1-ethyl-3-methylimidazolium chloride; DCM, dichloromethane; ILIS, ionic liquid immobilized on silica gel.

fructose/DMSO molar ratios and found that an optimum conversion was attained for a ratio of 0.8. The authors suggested that DMSO associates initially with only D-fructose at the start of the dehydration reaction, after which the generated water associates with DMSO, reducing the amount of DMSO available to D-fructose. Consequently, DMSO had to be sufﬁciently in excess to associate with all the water released at the end of the reaction. Amarasekara et al.91 studied the mechanism of the dehydration of fructose to HMF in DMSO at 150 ◦C without any added mineral or Lewis acid catalyst by monitoring the reaction by NMR spectroscopy (Scheme 8). It was possible to identify an intermediate (4R,5R)-4-hydroxy-5-hydroxymethyl- 4,5-dihydrofuran-2-carbaldehyde using a combination of 1Hand Scheme 8 91. 13C NMR spectra. Ionic liquids have been a promising solvent for carbohydrate Furthermore, they have been shown not only to act as solvents, transformations.92 These solvents can dissolve carbohydrates, but also as reaction promoters for carbohydrate dehydration even at high concentrations,93 and can be easily recycled. reactions.68,79

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 754–793 | 767 View Online

In 2006 HMF formation starting from fructose and sucrose, liquid–acid mixture reactive phase, the product formed was in an ionic liquid used as reaction solvent and reaction pro- extracted without any cross-contamination (Fig. 2). This ionic moter was reported.68 [HMIM][Cl] (1-H-3-methylimidazolium liquid system has the advantage of being biodegradable and chloride) is a protic ionic liquid that could convert completely non-toxic.61a fructose with 92% HMF selectivity (Table 6, entry 2). The influence of the ionic liquid as both solvent and catalyst is more important for the formation of HMF than for its decomposition. Kinetic studies showed that the energies of activation for the formation and decomposition of HMF are similar to those reported for the reaction catalyzed by solid catalysts.68 The dehydration reaction was also tested with sucrose as starting material, and a rapid cleavage of sucrose into glucose and fructose was observed, but as reported before for other conditions,7e the glucose moiety did not react, although the fructose conversion was complete. After HMF extraction with Fig. 2 diethyl ether, it was possible to recycle the ionic liquid at least 5 times. Recently the same authors reported61c one-pot inulin hydrol- The use of ionic liquids as solvents and reaction pro- ysis and fructose dehydration with moderate HMF selectivity moters was also reported by Yokoyama et al.94 In this ◦ (Table 6, entries 10–12) in choline-based ionic liquids at 80 C. work microwave irradiation was used to heat the fruc- The acidic CholCl–oxalic acid and CholCl–citric acid ionic tose dehydration reaction in a Lewis acidic ionic liquid, liquid acted as solvent and catalyst, and it was possible to [ASCBI][OTf] (3-allyl-1-(4-(sulfurylchloride)butyl)imidazolium recycle the CholCl–oxalic acid system for at least six cycles, trifluoromethanesulfonate), and a Brønsted-acidic ionic liq- just by extracting the product with ethyl acetate. A biphasic uid [ASBI][OTf] (3-allyl-1-(4-sulfobutyl)imidazolium trifluo- reaction system with ethyl acetate as extraction solvent provided romethanesulfonate) and their silica gel immobilized coun- an improved HMF selectivity (Table 6, entry 11). terparts (Scheme 9).94 The two ionic liquids with DMSO as Fayet et al. reported in 198395 HMF synthesis from fructose, co-solvent, converted fructose in very good yields, and good glucose, sucrose, inulin, and levan (fructose polymer). Different selectivity for HMF (Table 6, entries 3 and 4). The Lewis acidic pyridinium salts were tested as promoters for HMF synthesis. ionic liquid was a better reaction medium than the Brønsted For fructose, inulin and levan, a moderate HMF yield was acidic ionic liquid. These ionic liquids, when immobilized ◦ achieved with pyridinium chloride, at 120 C for 30 min (Table 6, on silica, converted fructose with 100% conversion, but with entries 13, 14 and 15). The same did not occur for glucose or medium selectivity for HMF (Table 6, entries 5 and 6). sucrose, for which the HMF yield was very low (Table 6, entries Downloaded by University of Oxford on 06 April 2011 16 and 17). A patent was published in 200896 in which the authors claimed Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D that, by heating a carbohydrate in an ionic liquid, and using extraction solvents, it was possible to obtain pure HMF.

4.2 Metal catalysis

4.2.1 Chromium-based catalysts. Zhang et al. reported79 the synthesis of HMF starting from fructose or glucose with very good selectivity. They studied the effect of different ionic liquids in the fructose dehydration. After choosing [EMIM][Cl] (1-ethyl- 3-methylimidazolium chloride) as solvent, they tested different Scheme 9 94. catalysts, such as several metal chlorides, and mineral and Lewis acids. In the absence of catalyst and at 120 ◦C, they obtained Several ionic liquids and pyridinium salts were tested, in- a fructose conversion of almost 100%, with approximately 70% cluding Brønsted-acidic ionic liquids, Lewis-acidic solvents selectivity (Table 7, entry 1). This system was not so efﬁcient for (e.g. ChoCl–metal chlorides), basic solvents (e.g. ChoCl–urea, glucose, since even with an increase of the temperature to 180 ◦C 1,1,3,3-tetramethylguanidinium triﬂuoroacetate and lactate) only 40% conversion was obtained, with less than 5% selectivity and choline chloride (ChoCl)-based deep eutectic mixtures. (Table 7, entry 2). However glucose conversion increased to near Without adding any catalyst, these solvents were tested for the 70%, and the selectivity was approximately 90% when a catalytic ◦ conversion of fructose to HMF at 80 Cfor1h.61a The Lewis amount of chromium(II) chloride was added (Table 7, entry 3).

acids ZnCl2 and CrCl3 in ChoCl–metal chloride produced less The authors proposed that the catalyst [EMIM][Cl]/CrCl2 was than 20% of HMF (Table 6, entries 8 and 9). The most efﬁcient inducing glucose isomerisation to fructose and then, fructose solvent–catalyst tested was choline chloride–citric acid, which was rapidly converted to HMF (Scheme 10).79 The HMF led to 91.1% conversion with 83.8% HMF selectivity (Table 6, stability was studied by heating pure HMF at 100 ◦Cfor3hin

entry 7). Ethyl acetate was reported as extraction solvent and [EMIM][Cl] in the presence of a catalytic amount of CrCl2 and

showed good efﬁciency. Due to the immiscibility with the ionic as a result, 98% of the HMF was recovered. When no CrCl2 was

768 | Green Chem., 2011, 13, 754–793 This journal is © The Royal Society of Chemistry 2011 View Online

it is more stable and easily handled under air, much cheaper and it is very likely that Cr2+ is oxidized to Cr3+ in the IL system containing dissolved air and water.60a The system with

[BMIM][Cl]/CrCl3 without an extraction solvent resulted in a 81% HMF yield, which was improved when toluene was added to the reaction system as an extraction solvent (Table 7, entry 11 vs. 13). This result is comparable with the reported99 glucose

dehydration in [BMIM][Cl]/CrCl3 with microwave irradiation as a heating source, for only one minute, for which the isolated HMF yield was 91% (Table 7, entry 18). Other substrates were Scheme 10 79. tested, such as inulin, sucrose, and cellobiose (Table 7, entries 14, 15 and 16, respectively), and cellulose (Table 7, entry 17). added to the system, only 28% of the initial HMF was recovered. Although high HMF selectivity was achieved for inulin and Similar studies were done with other metal halides with high sucrose, the same did not happen with cellobiose or cellulose,

HMF recovery: CuCl2,VCl4,andH2SO4 with 85, 86 and 98% even when adding H2SO4 into the reaction system for cellulose. recovery, respectively. It may be assumed that a catalytic amount In 2009, Li et al.99 reported the transformation of glucose and of some metal chlorides can not only catalyze the dehydration cellulose in [BMIM][Cl] (1-butyl-3-methylimidazolium chloride)

reaction, but also stabilize the ﬁnal product. This may be one of with CrCl3 as catalyst, affording HMF yields of 91 and 61%, the main reasons for not observing polymeric by-products, and respectively when subjected to microwave irradiation of 400 W only a negligible amount of levulinic acid being formed. This for only one and two minutes, respectively (Table 7, entries 18 method was published in 2008 in a patent.97 and 19). Different cellulose samples were tested reaching 53–62% Using [BMIM][Cl] (1-butyl-3-methylimidazolium chloride) HMF yields, indicating that this method is affected neither by ◦ as solvent (100 C, 6 h), several NHC/metal (N-heterocyclic the cellulose type nor the polymerization degree. The high yields carbene ligand) complexes were tested as catalysts for the obtained from cellulose are explained by the complete cellulose dehydration of fructose and glucose (Scheme 11).98 The authors dissolution in the ionic liquids, leaving cellulose chains accessible concluded that bulky NHC ligands protect the Cr center from to chemical transformations, and also because [BMIM][Cl] has reacting with [BMIM][Cl] and form a sterically crowded metal excellent dielectric properties for transformation microwaves center, therefore providing a higher catalytic efﬁciency. A good into heat. Although Zhang et al.79 reported smaller HMF yields

HMF selectivity was achieved, using these NHC/Cr complexes for glucose transformation with CrCl3 (Table 7, entry 4 vs. 18) in as catalysts, 96% and 81% for fructose and glucose respectively [EMIM][Cl], the authors believe that the microwave irradiation (Table 7, entries 7 and 5). The HMF yield was conﬁrmed by can improve the catalyst behaviour. The mechanism for this GC, but it was possible to separate the HMF product from transformation still remains unknown, although the authors

Downloaded by University of Oxford on 06 April 2011 the reaction medium by a simple ether extraction. After that, pursued a pathway to glucose isomerisation to fructose. the catalyst and the ionic liquid could be recycled for at least An extension of this work was made recently by the same Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D three cycles, with some loss of selectivity being observed when authors, where they have tested this microwave-assisted trans- glucose was used as substrate. A higher fructose and glucose formation with other biomass resources, such as corn stalk, rice 100 concentration was tested (20 wt.%) and no decrease of selectivity straw and pine wood. In this work CrCl3·6H2O was used as was observed (Table 7, entries 6 and 8). catalyst in [BMIM][Cl], for 2–3 min under 400 W microwave irradiation. HMF yield improvement when using microwave heating was again conﬁrmed (Table 7, entry 23 vs. 24). This reaction was also tested with [BMIM][Br] with similar results, as with [BMIM][Cl] (Table 7, entries 24 vs. 25). [BMIM][Br] can also dissolve lignocellulosic biomass, and the authors reported that the reaction medium has little effect on the dehydration efﬁciency as long as the solvent dissolves lignocellulosic biomass. Chen et al.64 studied the cellulose conversion into HMF using ◦ a ionic liquid–water mixture with CrCl2 as catalyst. At 120 C,

with 10 mol% of CrCl2 in [EMIM][Cl] (no water added) 89% HMF conversion was observed (Table 7, entry 26). This high 98 Scheme 11 . HMF yield implies that not only glucose was converted to HMF, but also other reducing sugars present after cellulose hydrolysis.

HMF conversion using [EMIM][HSO4] and [BMIM][Cl] with To enhance the cellulose hydrolysis and dehydration, water was different substrates, with or without extraction solvents (toluene added to this catalytic system, but this resulted in a decrease in 60a or MIBK) was studied. [EMIM][HSO4] with toluene or MIBK the HMF yield, even at higher temperatures (Table 7, entry 24 as extraction solvents completely converted fructose in 79 and vs. 28). 88% HMF yield, respectively (Table 7, entry 9 and 10). This Recently Zhang et al.101 studied cellulose transformation to system was not so efﬁcient with glucose, and therefore, the HMF in the ionic liquid [EMIM][Cl] using a pair of two metal

authors changed the ionic liquid to [BMIM][Cl] with CrCl3 catalysts, CrCl2 and CuCl2, with a HMF yield of 55.4% (Table 7,

as catalyst. This catalyst was chosen instead of CrCl2,since entry 29). In this work a study was carried out to achieve the

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 754–793 | 769 View Online

Table 7 Conversion of carbohydrates to HMF by chromium-based catalystsa

Reaction conditions Post-reaction details Isolation/ Entry Biomass Temp. Conver- HMF selec- determination Isolated Catalyst, reaction (ref.) source Solvent Catalyst (◦C) Time sion (%) tivity (%) method yield (%) medium reuse

179 Fructose [EMIM][Cl] — 120 3 h 100 70 HPLC — — (10 wt.%) 279 Glucose [EMIM][Cl] — 180 3 h 40 <5 HPLC — — (10 wt.%) 79 3 Glucose [EMIM][Cl] CrCl2 100 3 h 70 90 HPLC — — (10 wt.%) 79 4 Glucose [EMIM][Cl] CrCl2 100 3 h 43 70 HPLC — — (10 wt.%) 98 5 Glucose [BMIM][Cl] NHC–CrCl2 100 6 h — 81 Diethyl ether — Catalyst/solvent (10 wt.%) extraction/GC recycling for at least 3 cycles 98 6 Glucose [BMIM][Cl] NHC–CrCl2 100 6 h — 80 GC — — (20 wt.%) 98 7 Fructose [BMIM][Cl] NHC–CrCl2 100 6 h — 96 Diethyl ether — Catalyst/solvent (10 wt.%) extraction/GC recycling for at least 3 cycles 98 8 Fructose [BMIM][Cl] NHC–CrCl2 100 6 h — 96 GC — — (20 wt.%) 960a Fructose [EMIM][HSO4] — 100 30 min 100 79 Toluene —— / toluene extraction/HPLC 60a 10 Fructose [EMIM][HSO4]/ — 100 30 min 100 88 MIBK —— MIBK extraction/HPLC 60a 11 Glucose [BMIM][Cl] / CrCl3 100 4 h 91 91 HPLC — — toluene 60a 12 Glucose [BMIM][Cl] / CrCl3 100 4 h 79 79 MIBK —— MIBK extraction/HPLC 60a 13 Glucose [BMIM][Cl] CrCl3 100 4 h 83 81 HPLC — — 60a 14 Inulin [EMIM][HSO4]/ — 100 30 min — 73 MIBK —— MIBK extraction/HPLC 60a 15 Sucrose [BMIM][Cl] / CrCl3 100 4 h — 73 MIBK —— MIBK extraction/HPLC 60a 16 Cellobiose [BMIM][Cl] / CrCl3 100 4 h — 37 MIBK —— MIBK extraction/HPLC Downloaded by University of Oxford on 06 April 2011 1760a Cellulose [BMIM][Cl] / CrCl3– 100 4 h — 9 MIBK —— MIBK H2SO4 extraction/HPLC 99

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D 18 Glucose [BMIM][Cl] CrCl3 ~200 1 min — 91 Silica gel 91 — (MW chromatography 400 W) (ethyl acetate–petroleum ether) 99 19 Cellulose [BMIM][Cl] CrCl3 ~200 1 min — 61 Silica gel 61 — (MW chromatography 400 W) (ethyl acetate–petroleum ether) 100 20 Cellulose [BMIM][Cl] CrCl3 ~200 2.5 min — 62 HPLC (MW 400 W) 100 21 Corn [BMIM][Cl] CrCl3 ~200 3 min — 45 HPLC stalk (MW 400 W) 100 22 Rice [BMIM][Cl] CrCl3 ~200 3 min — 47 HPLC straw (MW 400 W) 100 23 Pine [BMIM][Cl] CrCl3 ~200 3 min — 52 HPLC wood (MW 400 W) 100 24 Pine [BMIM][Cl] CrCl3 100 (oil 60 min — 6.4 HPLC wood bath) 100 25 Pine [BMIM][Br] CrCl3 ~200 3 min — 44 HPLC wood (MW 400 W) 64 26 Cellulose [EMIM][Cl] CrCl2 120 6 h — 89 Acetone extraction/HPLC

770 | Green Chem., 2011, 13, 754–793 This journal is © The Royal Society of Chemistry 2011 View Online

Table 7 (Contd.)

2764 Cellulose [EMIM][Cl]– CrCl2 120 12 h — 13 Acetone H2O extraction/HPLC 64 28 Cellulose [EMIM][Cl]– CrCl2 140 2 h — 40 Acetone H2O extraction/HPLC 101 29 Cellulose [EMIM][Cl] CrCl2– 120 8 h — 57.5 HPLC (10 wt.%) CuCl2 102 30 Sucrose [OMIM][Cl] HCl–CrCl2 120 30 min — 82.0 HPLC (20% w/v)

31102 Sucrose [OMIM][Cl] HCl–CrCl2 120 60 min — 67.7 HPLC (30% w/v) 102 32 Sucrose [OMIM][Cl] HCl–CrCl2 120 60 min — 53.2 HPLC (50% w/v) 103 33 Fructose DMA–LiCl H2SO4 100 5 h — 63 HPLC — — (10 wt.%) 103 34 Fructose DMA– H2SO4 100 2 h — 84 HPLC — — (10 wt.%) [EMIM][Cl] 103 35 Fructose DMA–LiF H2SO4 80 2 h — 0 HPLC — — (10 wt.%) 103 36 Fructose DMA–LiBr H2SO4 100 4 h — 92 HPLC — — (10 wt.%) NaBr 2 h 93 LiI 6 h 89 NaI 5 h 91 103 37 Glucose DMA–LiCl CrCl2 100 5 h — 60 HPLC — — (10 wt.%) 103 38 Glucose DMA–LiCl– CrCl2 100 6 h — 62 HPLC — — (10 wt.%) [EMIM][Cl] 103 39 Glucose DMA–LiI CrCl2 100 4 h — 54 HPLC — — (10 wt.%) 103 40 Glucose DMA–LiBr CrBr2 100 6 h — 80 HPLC — — (10 wt.%) 103 41 Cellulose DMA–LiCl– CrCl2–HCl 140 2 h — 54 HPLC — — [EMIM][Cl] 103 42 Cellulose [EMIM][Cl] CrCl2–HCl 140 1 h — 53 HPLC — —

Downloaded by University of Oxford on 06 April 2011 103 43 Cellulose DMA–LiI CrCl2–HCl 140 3 h — <1 HPLC — — LiBr <1 103

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D 44 Corn stover DMA–LiCl– CrCl2–HCl 140 2 h — 48 HPLC — — [EMIM][Cl]

a MIBK, methyl isobutyl ketone; MW, microwave irradiation; [BMIM][Cl], 1-ethyl-3-methylimidazolium chloride; [EMIM][Cl], 1-ethyl- 3-methylimidazolium chloride; [OMIM][Cl], 1-octyl-3-methylimidazolium chloride; NHC, N-heterocyclic carbene ligand; DMA, N,N- dimethylacetamide.

best molar ratio for the two catalysts CrCl2 and CuCl2,andit

was observed that as little as 3 mol% of CrCl2 in the paired metal

chlorides was sufﬁcient to activate the CuCl2-dominant catalyst in the [EMIM]Cl solvent. The catalyst and the solvent could be recycled for three times using MIBK as extraction solvent. A mechanism for this transformation is being studied by the authors. Studies on sucrose hydrolysis and further dehydration to Scheme 12 102. HMF were performed by Chung,102 using a different ionic liquid, 1-octyl-3-methylimidazolium chloride ([OMIM][Cl]) as solvent

and two metal chlorides as catalysts (CrCl2 or ZnCl2)inan disappearance, probably due to the strong chemical reactivity acidic medium, HCl (Scheme 12). According to the authors, for isomerisation, dehydration, fragmentation or condensation the acidic medium improves the sucrose hydrolysis, and the reactions. The authors did not quantify the by-products formed.

catalysts CrCl2 or ZnCl2 catalyze further glucose dehydration. When the catalyst CrCl2 was added to the system, an HMF yield Firstly, the sucrose hydrolysis to glucose and fructose monomers improvement was observed. Depending on HCl and sucrose ◦ was studied in [OMIM][Cl] solutions at 120 C. Even without concentrations the time for reaction was adjusted (Table 7, addition of acid, the sucrose hydrolysis occurred in [OMIM][Cl], entries 30–32). The best HMF achieved was 82% for a sucrose

although with longer reaction times. With the acid addition, the concentration of 20% with 0.5 M HCl and CrCl2 as catalyst hydrolysis was faster, but it was also possible to see fast fructose (Table 7, entry 30). A 50% w/v sucrose concentration was

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 754–793 | 771 View Online

also tested, resulting in a decrease of the HMF yield to 53.2% experiments, in situ X-ray absorption spectroscopy (XAS), and (Table 7, entry 32). density functional theory (DFT) calculations. The key reaction 103 79 Binder and Raines reported the use of authentic ligno- of the catalytic system [EMIM][Cl]–CrCl2, as reported before, cellulosic biomass as starting material for HMF production, is the isomerisation of glucose to fructose. In this work the using DMA–LiCl (N,N-dimethylacetamide–lithium chloride) authors proposed a mechanism based on that suggested for as solvent. They initiated the study with fructose as starting the enzymes. In this chemical system the highly concentrated

material, with H2SO4 as catalyst and DMA–lithium salts as and mobile chloride anions from the ionic liquid promote solvent, and tested several additives, such as [EMIM][Cl] (1- various (de)protonation reactions important for the glucose ethyl-3-methylimidazolium chloride) and other ionic liquids. isomerisation (Scheme 14). In enzymes, such transformations When different lithium salts were tested, it was observed that are catalyzed by basic amino acid residues at the active site. The ﬂuoride ions were completely ineffective for HMF synthesis unique transient self-organization of Cr2+ dimers to facilitate (Table 7, entry 35), although bromide and iodide ions, which the rate-controlling H shift in glucose isomerisation is possible tend to be less ion-paired than ﬂuoride and chloride,103 achieved as a result of the dynamic nature of the Cr complexes and the HMF yields up to 92% (Table 7, entry 36). Based on these presence of moderately basic sites in the ionic liquid. and other experimental results, the authors proposed a reaction mechanism involving an oxocarbenium ion that is attacked by the halide ion (Scheme 13). HMF production starting

from glucose was also tested with CrCl2,CrCl3 or CrBr3 as catalyst with DMA–LiCl (or other salts such as LiBr, LiI) as solvent (Table 7, entries 37–40). The halide effect was also very pronounced with this substrate. With chloride anions present

in the reaction mixture (CrCl2 as catalyst and DMA–LiCl as solvent, Table 7, entry 37) HMF yields up to 60% were obtained, and they were enhanced to 62% with the addition of [EMIM][Cl] as additive (Table 7, entry 38). Although the

addition of iodide ions to the reaction mixture with CrCl2 as catalyst did not improve the HMF yields, the same did not happen with bromide ions, which improved HMF yields up to 80% (Table 7, entry 40). Cellulose was also tested as an HMF

source with CrCl2 and HCl as catalysts. Dissolution of puriﬁed cellulose in a mixture of DMA–LiCl and [EMIM][Cl] and Scheme 14 105. addition of CrCl2 or CrCl3 produced HMF from cellulose in up ◦ Downloaded by University of Oxford on 06 April 2011 to 54% yield within 2 h at 140 C (Table 7, entries 41, 42). Due to cellulose insolubility, neither lithium iodide nor lithium bromide 4.2.2 Zirconium and titanium catalysts. Watanabe studied Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D produced high yields of HMF (Table 7, entry 43). Finally, the glucose and fructose reactivity in hot compressed water with

synthesis of HMF from lignocellulosic biomass was studied. A homogeneous or heterogeneous acidic (H2SO4 or TiO2) or alkali 106 48% HMF yield (based on cellulose content of the biomass) additives (NaOH or ZrO2). They observed that isomerisation was achieved with this substrate under similar conditions as between glucose and fructose is catalyzed by alkali, and fructose for cellulose (Table 7, entry 44). The authors proposed that dehydration is promoted by acidic conditions. It seems that the the formation of HMF from cellulose in DMA–LiCl occurs equilibrium favours fructose formation in hot compressed water via sacchariﬁcation followed by isomerisation of the glucose because the rate of isomerisation of fructose into glucose is monomers into fructose and dehydration of fructose to form negligible compared to that of glucose into fructose. Zirconia

HMF. It was possible to separate the HMF formed by an ion- (ZrO2) is a base catalyst that promotes the glucose isomerisation.

exclusion chromatographic separation, where over 75% of HMF On the other hand, anatase (TiO2) was found to act as an acid was recovered. Binder and Raines are the authors of a patent catalyst to promote formation of HMF. that describes this technology with the aim of 2,5-dimethylfuran In 2000 zirconium and titanium hydrogenphosphates in the a manufacture.104 and g structural arrangements were reported as catalysts for fructose and inulin dehydration to HMF.60e These reactions were carried out in aqueous media, but even so no appreciable subsequent rehydration to levulinic and formic acids occurred. Among the investigated catalysts, both surface Brønsted and Lewis acid sites were present, and experimental results showed that both acid sites may be involved in the catalytic process. However, as the Lewis acid site strength increased, a correspond- Scheme 13 103. ing enhancement of HMF yield is obtained. For example, the lower strength of Lewis acid sites present on the external crystal 105 In 2009 Pidko et al. studied the reactivity of chromium(II) surface of c-TiP2O7 with respect to those on c-ZrP2O7 decreased

chloride towards selective glucose dehydration in an ionic the performance of c-TiP2O7 catalyst (Table 8, entry 4 vs. 5or6). liquid medium, combining different methods, such kinetic Inulin showed a similar reactivity to that observed for fructose

772 | Green Chem., 2011, 13, 754–793 This journal is © The Royal Society of Chemistry 2011 View Online

Table 8 Conversion of carbohydrates to HMF catalyzed by Zr and Ti catalystsa

Reaction conditions Post-reaction details Entry Biomass Temp. Conversion HMF Isolation/ Catalyst, reaction (ref.) source Solvent Catalyst (◦C) Time (%) selectivity (%) determination method medium reuse

60e 1 Fructose H2O a-Titanium 100 0.5 h 29.1 98.3 GC — phosphate (a -TiP) 1 h 33.4 83.5 2 h 34.1 75.4 60e 2 Fructose H2O g-Titanium 100 0.5 h 36.7 96.1 MIBK extraction/GC Catalyst and phosphate substrate (g -TiP) solution reused for 2 cycles 1 h 46.8 88.6 2 h 56.6 68.7 60e 3 Inulin H2O g-Titanium 100 0.5 h 31.8 98.1 MIBK extraction/GC Catalyst and phosphate substrate (g -TiP) solution reused for 2 cycles 1 h 44.3 94.0 2 h 91.9 70.7 60e 4 Fructose H2O Cubic titanium 100 0.5 h 24.8 98.7 GC — pyrophosphate

(c-TiP2O7) 1 h 29.3 90. 2 h 38.7 72.3 60e 5 Fructose H2O Cubic 100 0.5 h 44.4 99.8 MIBK extraction/GC Catalyst and zirconium- substrate pyrophosphate solution reused

(c-ZrP2O7) for 2 cycles 1 h 52.2 86.0 2 h 52.8 81.4 60e 6 Inulin H2O Cubic 100 0.5 h 26.4 97.8 MIBK extraction/GC Catalyst and zirconium- substrate pyrophosphate solution reused

(c-ZrP2O7) for 2 cycles 1 h 38.9 89.4 2 h 50.2 72.3

Downloaded by University of Oxford on 06 April 2011 107 7 Fructose H2OZrO2 200 5 min 65.3 30.6 HPLC — (2 wt.%) (MW) 107

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D 8 Fructose H2OTiO2 200 5 min 83.6 38.1 HPLC — (2 wt.%) (MW) 107 9 Glucose H2OZrO2 200 5 min 56.7 10.0 HPLC — (2 wt.%) (MW) 107 10 Glucose H2OTiO2 200 5 min 63.8 18.6 HPLC — (2 wt.%) (MW) 108 2- 11 Fructose H2OSO4 –ZrO2 200 5 min 88.7 37.4 HPLC — (2 wt.%) (MW) 108 2- 12 Fructose Acetone– SO4 –ZrO2 180 20 93.6 72.8 HPLC — (2 wt.%) DMSO (7:3 (MW) min w/w) 1363 Glucose DMSO — 130 4 h 94 4.3 HPLC — (7.6 wt.%) 63 2- 14 Glucose DMSO SO4 –ZrO2 130 4 h 95.2 19.2 Et2O — (7.6 wt.%) extraction/HPLC 1563 Glucose DMSO CSZA-3 130 4 h 99.1 48.0 HPLC Catalyst reused (3.9 wt.%) for at least 5 cycles 1663 Glucose DMSO CSZA-3 130 4 h 98.1 39.2 HPLC — (20 wt.%) 1763 Fructose DMSO — 130 4 h 99.6 71.9 HPLC — (20 wt.%) 63 2- 18 Fructose DMSO SO4 –ZrO2 130 4 h 99.8 67.7 HPLC — (7.6 wt.%) 1963 Fructose DMSO CSZA-3 130 4 h 99.4 56.6 HPLC — (7.6 wt.%)

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 754–793 | 773 View Online

Table 8 (Contd.)

109 20 Cellulose Hot compressed ZrO2–TiO2 250 5 min 70 13 HPLC — H2O 109 21 Glucose Hot compressed ZrO2–TiO2 250 5 min 80 28 HPLC —

H2O 110 22 Fructose Subcritical H2O ZrP 240 120 s 80.6 61.3 HPLC Reused for 6–7 (3.35 cycles MPa) 110 23 Glucose Subcritical H2O ZrP 240 180 s 53.1 39.0 HPLC Reused for 6–7 (3.35 cycles MPa)

a 2- CSZA-3,SO4 /ZrO–Al2O3; DMSO, dimethyl sulfoxide; MIBK, methyl isobutyl ketone; MW, microwave irradiation.

(Table 8, entries 3 and 6). Cubic zirconium pyrophosphate and 19). The authors suggested that the production of HMF in this g-titanium phosphate offered the best performances in terms of system may not be mainly via glucose isomerisation–dehydration activity and selectivity (Table 8, entries 5 and 2). process. The catalyst CSZA-3 was recycled for at least Watanabe et al.107 studied the HMF formation from fructose 5cycles.

and glucose with TiO2 and ZrO2 ascatalystsinwaterandwitha With the objective of coupling in one-pot the hydrolysis and microwave irradiation (200 ◦C) as heating source. The advantage dehydration reactions to produce HMF from lignocellulosic of using these heterogeneous catalysts when compared with biomasses (i.e. sugarcane bagasse, rice husk and corn cob),

homogeneous catalysts (such as HCl or H2SO4) is the low conditions employing heterogeneous catalysts TiO2, ZrO2 and

corrosion and easy separation. With the reported method, good mixed-oxide TiO2–ZrO2 under hot compressed water (HCW) conversion yields were achieved, but selectivity for HMF was low were applied.109 It was found that the catalyst preparation (Table 8, entries 7–10). The zirconium or titanium phosphates procedure affected its reactivity, and with different Ti/Zr catalysts used by Benvenuti60e had a better performance than the ratios and different calcination temperatures the catalyst acid-

TiO2 and ZrO2 described in this work. ity/basicity was different. Although these catalysts resulted in In continuation of the heterogeneous catalysis study, these good conversion (70–80% to glucose and cellulose, Table 8, 108 Downloaded by University of Oxford on 06 April 2011 authors reported in 2009 the behaviour of the solid acid entries 20 and 21), several other products were formed. The catalyst sulfated zirconia in the dehydration of fructose to HMF. HMF yield was approximately 28% for glucose and 13% for

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D In this study they also used microwave heating. The catalyst was cellulose.109 characterized, and in aqueous solutions the HMF selectivity was Asghari et al.110 reported the fructose and glucose dehydration low (37.4%, Table 8, entry 11); this may be due to the deactivation reaction with zirconium phosphate as catalyst, in subcritical of the active acid sites on the catalyst by water. Therefore, as water. The authors claimed that only a few solid catalysts were before,90 the authors changed the solvent to an acetone–DMSO acceptable for reactions in aqueous solutions (especially water at mixture. The results were very satisfactory, with 93.6% fructose high temperatures and pressure or subcritical water) in terms of conversion and 72.8% HMF selectivity being achieved in the their activity, stability and insolubility. It was found that zirco- 2- ◦ presence of SO4 –ZrO2 as catalyst for 20 min at 180 C(Table8, nium phosphates were stable in these conditions. They induced entry 12). The authors are working on in situ removal of water a moderate selectivity in the presence of fructose, comparable generated in the reaction to avoid catalyst deactivation. with that obtained with zirconium pyrophosphate,60e but led In the same year another group reported63 the glucose dehy- to higher conversions (Table 8, entry 22 vs. 6). This catalyst 2- 2- dration with SO4 –ZrO2 (CSZ) and SO4 /ZrO–Al2O3 (CSZA- was not the best for conversion of glucose into HMF (Table 8, 1–5, depending on the Zr–Al mole ratio) catalysts. CSZA-1–5 entry 23). have acidic and basic active sites, such that increasing the Al For this group of catalysts based on zirconium or titanium ratio increased the number of basic sites. When these catalysts metals, although good conversions could be obtained for were tested for glucose dehydration, the authors expected fructose, the HMF selectivity was low, and several other by- that by increasing the basic sites on the catalyst, the glucose products were formed. For glucose conversion, the obtained isomerisation would also be increased, resulting in higher HMF conversion yields were in general moderate, and the HMF yields. However, this was not observed, and in fact the catalyst selectivity was also low. Thus, despite the advantage of using with higher acidity and moderate basicity was better for the a heterogeneous catalysis due to the ease of catalyst recycling formation of the target product (Table 8, entries 14 vs. 15). and lower corrosion effects, the catalyst and/or the reaction Another important observation is the fact that the acid sites on conditions had to be improved for better HMF selectivity. the CZA or CSZA catalysts exhibited no catalytic improvement for the conversion of fructose to HMF when compared with the 4.2.3 Lanthanides as catalysts. Ishida et al.111 shown that same conditions without a catalyst (Table 8, entry 17 vs.18 or lanthanide ions can catalyse glucose dehydration to HMF. No

774 | Green Chem., 2011, 13, 754–793 This journal is © The Royal Society of Chemistry 2011 View Online

a Table 9 Conversion of glucose to HMF catalyzed by YbCl3

Reaction conditions Post-reaction details Entry Biomass HMF Isolation/ (ref.) source Solvent Catalyst Temp. (◦C) Time Conversion (%) selectivity (%) determination method

112 1 Glucose [EMIM][Cl] YbCl3 160 1 h — 8 HPLC 112 2 Glucose [BMIM][Cl] YbCl3 160 1 h — 20 HPLC 112 3 Glucose [HexMIM][Cl] YbCl3 160 1 h — 19 HPLC 112 4 Glucose [OMIM][Cl] YbCl3 160 1 h — 22.5 HPLC

a [BMIM][Cl], 1-ethyl-3-methylimidazolium chloride; [EMIM][Cl], 1-ethyl-3-methylimidazolium chloride; [OMIM][Cl], 1-octyl-3-methylimidazolium chloride; [HexMIM][Cl], 1-hexyl-3-methylimidazolium chloride.

further decomposition in the ﬁrst 15 min was observed, but for A glucose transformation to HMF using a niobium catalyst higher reaction times the HMF selectivity started to decrease. was published in a patent in 2009.114

Several lanthanide(III) chlorides (DyCl3,YbCl3,LaCl3, NdCl3, Although the main objective was to synthesize 2,5- ◦ 65b EuCl3) were tested in water, at 140 C and for 15 min reaction. furandicarboxylic acid (FDA), Ribeiro et al. reported an The ﬁnal product was extracted with benzene. Lanthanide(III) initial reaction in which fructose was dehydrated to HMF chlorides were preferred compared to typical transition metals with moderate fructose conversion yields and good HMF

because they are cheaper and less toxic, although they are not selectivity. The compounds cobalt acetylacetone (Co(acac)3),

the ideal catalysts for glucose dehydration (Table 9, entries 1- SiO2-gel, and Co(acac)3 encapsulated in sol–gel (Co-gel) were 4). Since lanthanide ions have a high afﬁnity for oxygen atoms, tested as catalysts. A triphasic reaction was performed, fructose the authors believe that they coordinate with glucose and act as was dissolved in water with MIBK as extraction solvent,

Lewis acid catalysts. and a heterogeneous catalyst was used. SiO2-gel give the best Recently,112 lanthanide(III) chlorides were again tested as HMF selectivity (Table 10, entries 5 vs. 6). These results were catalysts for glucose dehydration to HMF, but this time using improved to 100% selectivity despite moderate conversion, when ionic liquids as solvents. Firstly, the HMF stability was tested the reaction was made in water in an autoclave (Table 10, in different ionic liquids at 100 ◦C, and all the ionic liquids entry 7). tested showed some HMF degradation. The imidazolium-based Vanadyl phosphate (VOP) was used as an acid catalyst ionic liquids with halides as anion showed the lowest degree in the dehydration of fructose aqueous solutions to HMF.115 of degradation. Secondly, several lanthanide chlorides were This catalyst showed low selectivity and moderate fructose tested as catalysts for glucose dehydration in two different ionic conversion (Table 10, entry 8). It was reported115 that acidity of

Downloaded by University of Oxford on 06 April 2011 liquids, [BMIM][Cl] and [EMIM][Cl], and an improvement was VOP can be modiﬁed by isomorphous substitution of some VO3+ observed when lanthanide chlorides was present as catalysts. The groups with trivalent metals M3+ such as Fe3+,Cr3+,Ga3+,Mn3+

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D 3+ strongest Lewis acid, YbCl3, gave the highest yield, although and Al . These new catalytic systems were tested in aqueous still moderate HMF yields (Table 9, entries 1–4). The authors solutions (Table 10, entries 10–12). FeVOP showed the best suggested that this reaction takes place by a different mechanism performance, even at high fructose concentrations (Table 10, from the chromium chloride catalytic system, because the yield entries 9–12), without the formation of insoluble polymeric by- was favoured in hydrophobic ionic liquids, while the yield of the products or HMF rehydration compounds. Similar activity and Cr-catalysed version decreased with the hydrophobicity of the selectivity results were obtained with inulin as substrate, and ionic liquids. FeVOP as catalyst (Table 10, entry 13). Several metal chlorides were screened in a ionic liquid 4.2.4 Other metal catalysts. Armaroli et al.113 reported the ([BMIM][Cl], 1-butyl-3-methylimidazolium chloride), with the use of commercial niobium phosphates, or niobium catalysts objective of ﬁnding the best catalyst for fructose dehydration at prepared by treatment of niobic acid with phosphoric acid, room temperature.62 Tungsten salts provided the best HMF yield for catalysis of sugar dehydration reactions. Substrates such (Table 10, entries 17 and 18). Several ionic-liquid-immiscible and as fructose, sucrose and inulin were tested in aqueous media. low-boiling organic solvents were tested as extraction solvents It was observed that the HMF selectivity was very high for (Table 10, entries 18–20). THF provided a biphasic reaction lower reaction times, but sugar conversion rates were low with higher HMF yields than the ionic liquid system (Table 10, (Table 10, entries 1–4). For higher reaction times, although the entry 17 vs. 18). It was possible to isolate the HMF product with sugar conversions increased up to 65.5%, the HMF selectivity THF as the extraction solvent, and the ionic liquid–catalyst decreased, due to the formation of polymeric by-products.113 could be recycled. Also a continuous batch process for the To overcome this problem, the authors reported an extraction conversion of fructose to HMF in a THF–[BMIM][Cl] biphasic process with MIBK as extraction solvent, where it was possible system was developed (Fig. 3), and tested with a bigger amount to recycle both the residual aqueous substrate solution and the of fructose (10 g) as starting material.62 solid catalyst, with fructose and inulin substrates (Table 10, A screening was performed in DMSO with different metal entries 3 and 4). The conversion values increased to 60–75% catalysts (100 ◦C, 3 h) for glucose dehydration.61b The best and the HMF selectivity continued high, up to 98% (Table 10, were chromium, aluminium and tin chlorides, the latter entries 3 and 4). being the most efﬁcient. With this catalyst several ionic

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 754–793 | 775 View Online

Table 10 Conversion of carbohydrates to HMF promoted by miscellaneous catalystsa

Reaction conditions Post-reaction details Isolation/ Entry Conversion HMF determination Catalyst, reaction (ref.) Biomass source Solvent Catalyst Temp. (◦C) Time (%) selectivity (%) method medium reuse

113 1 Fructose (6 wt.%) H2OH3PO4-treated 100 0.5 h 31.2 93.3 GC — niobic acid (P/N1) 1 h 33.3 30.3 2 h 61.5 12.4 113 2 Fructose (6 wt.%) H2O Niobium 100 0.5 h 28.8 100.0 GC — phosphate (NP2) 1 h 29.3 85.2 2 h 33.3 71.9 113a 3 Fructose (6 wt.%) H2O Niobium 100 0.5 h 33.6 98.3 GC Catalyst recycled phosphate with MIBK (NP2) extraction 1 h 75.8 97.8 113a 4 Inulin (6 wt.%) H2O Niobium 100 0.5 h 25.2 77.5 GC Catalyst recycled phosphate with mibk (NP2) extraction 1 h 48.7 74.0 1.5 h 76.3 72.0 65b 5 Fructose H2O/MIBK Co-gel 88 8 h — 46 HPLC/RI — 65b 6 Fructose H2O/MIBK SiO2-gel 88 8 h — 47 HPLC/RI — 65b 7 Fructose H2OSiO2-gel 160 1.1 h 52 100 HPLC/RI — (20 bar) 115 8 Fructose (30 wt.%) H2O VOP 80 0.5 h 45.1 32.9 GC–MS — 2 h 65.2 35.8 115 9 Fructose (6 wt.%) H2O FeVOP 80 2 h 48.1 42.2 GC–MS — 115 10 Fructose (10 wt.%) H2O FeVOP 80 2 h 60.9 51.2 GC–MS — 11115 Fructose (30 wt.%) H2O FeVOP 80 1 h 70.8 59.6 GC–MS — 115 12 Fructose (40 wt.%) H2O FeVOP 80 0.5 h 57.7 87.3 GC–MS — 115 13 Inulin (6 wt.%) H2O FeVOP 80 2 h 41.8 82.7 GC–MS — 14115 Fructose (6 wt.%) H2O CrVOP 80 1 h 60.0 48.5 — — 115 15 Fructose (6 wt.%) H2O AlVOP 80 1 h 75.9 57.6 — — 115 16 Fructose H2O VOP–TiO2 80 0.5 h 35.5 93.2 GC–MS — Downloaded by University of Oxford on 06 April 2011 1 h 39.7 87.4 62 17 Fructose (20 wt.%) [BMIM][Cl] WCl6 50 4 h — 63 THF extraction Catalyst/solvent

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D recycling 62 18 Fructose (20 wt.%) [BMIM][Cl] / WCl6 50 4 h — 72 THF extraction Catalyst/solvent THF recycling

1962 Fructose (20 wt.%) [BMIM][Cl] / WCl6 50 4 h — 61 MIBK extraction Catalyst/solvent MIBK recycling 62 20 Fructose (20 wt.%) [BMIM][Cl] / WCl6 50 4 h — 59 EtOAc extraction Catalyst/solvent EtOAc recycling 61b 21 Glucose (17 wt.%) [EMIM][BF4] SnCl4 100 3 h 98 62 EtOAc extraction Catalyst/solvent recycling for at least 4 cycles (20 wt.%) 99 61 (23 wt.%) 100 61 (26 wt.%) 99 58 61b 22 Fructose [EMIM][BF4] SnCl4 100 3 h 100 62 EtOAc extraction — 61b 23 Sucrose (17 wt%) [EMIM][BF4] SnCl4 100 3 h 100 65 EtOAc extraction — 61b 24 Cellobiose [EMIM][BF4] SnCl4 100 3 h 100 57 EtOAc extraction — 61b 25 Starch [EMIM][BF4] SnCl4 100 3 h 100 47 EtOAc extraction —

a FeVOP, iron vanadyl phosphate [BMIM][Cl], 1-ethyl-3-methylimidazolium chloride; [EMIM][Cl], 1-ethyl-3-methylimidazolium chloride; DMSO, dimethyl sulfoxide; MIBK, methyl isobutyl ketone; Co-gel, cobalt acetylacetonate encapsulated in sol–gel silica.

liquids were tested, and it was observed that for ionic liquids best selectivity was the [EMIM][BF4] (Table 10, entry 21). based on anions having coordination abilities, such as chlo- Based on this and other experimental evidence, the authors

ride, bis(trifluoromethane)sulfonimide (NTf2), trifluoroacetate proposed a mechanism involving a five- or six-membered ring (TFA), trifluoromethylsulfonate (OTf) or saccharin (SAC), the chelate complex of the Sn atom and glucose (Scheme 15). Other

HMF yields were lower than with other type of anions (e.g. BF4 saccharides were also tested, such as sucrose, cellobiose, inulin – tetraﬂuoroborate). The authors suggested that these anions and starch, providing reasonable HMF selectivity (Table 10, could compete with the interaction of glucose and the Sn entries 22–25).61b After product extraction with ethyl acetate, it

atom, inhibiting the HMF formation. The ionic liquid with was possible to recycle the SnCl4–[EMIM][BF4] catalytic system.

776 | Green Chem., 2011, 13, 754–793 This journal is © The Royal Society of Chemistry 2011 View Online

Scheme 17

oxides118 under basic conditions. Gorbanev et al.119 reported Fig. 3 the formation of HMFCA as an intermediate product during the aerobic oxidation of HMF to FDA with Au/TiO2 catalyst in basic aqueous solution at room temperature. They studied the relationship between the formed products and the amount of

base or applied O2 pressure, and observed that lower pressures or low concentrations of base afforded more of the intermediate oxidation product HMFCA compared to FDA. Casanova et al.120 also observed the formation of HMFCA as a intermediate product during gold-nanopracticle-catalyzed aerobic oxidation of HMF. They described that selective oxidation to HMFCA took place at 25 ◦C after 4 h and reported 100% yield. Very recently, Davis et al.121 described 92–93% selectivity towards HMFCA with 100% conversion of HMF promoted by Au/C 122 and Au/TiO2 in basic conditions. Van Deurzen et al. oxidized

HMF with H2O2 and chloroperoxidase (CPO), which is an enzyme known to be an effective catalyst for various oxidation

reactions with H2O2. They observed formation of DFF as a major product and unexpected formation of HMFCA as a minor product in up to 40% yield.

61b Scheme 15 . 5.1.2 Selective oxidation of the hydroxyl group. The selec- 5 Synthetic applications of HMF tive oxidation of hydroxyl group of HMF leads to the formation of DFF, which is an important monomer for industry.123 The structural motifs present in HMF, namely furan, pri- Numerous examples in the literature have described the selective

Downloaded by University of Oxford on 06 April 2011 mary hydroxyl and formyl functionalities, allows synthetic oxidation of the hydroxyl group of HMF to DFF using diverse transformation to other target molecules using the following oxidants. Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D main transformations: selective oxidation and reduction of the Reijendam et al. (Table 11, entry 1) obtained DFF in 37% yield formyl, hydroxyl groups and furan ring; carbonyl and hydroxyl by using lead diacetate in pyridine. Morikawa oxidized HMF homologation; and whole-skeleton transformations. with a variety of oxidants and reported higher yields (Table 11, entries 2–5). Cottier et al. obtained DFF in 58% yield by using 5.1 Oxidation a DMSO–potassium dichromate oxidative complex at 100 ◦C. The oxidation of HMF can be performed selectively to They observed that the application of ultrasonic irradiation to the formyl or hydroxyl groups to form 5-hydroxymethyl-2- the reaction mixture afforded DFF in higher yield 75% (Table 11, furancarboxylic acid (HMFCA) and 2,5-diformylfuran (DFF) entry 7). The trimethylammonium chlorochomate (TMACC)– respectively, or can involve both groups to give 2,5- Al2O3 oxidative system was also tested in conventional and furandicarboxylic acid (FDA), which are compounds of con- under sonochemical conditions, providing DFF in 75% and 72% siderable interest as well as starting materials for further respectively (Table 11, entry 8). The same authors performed ox- transformations and chemical building blocks for industry15,116 idation of HMF adsorbed together with pyridinium chlorochro- (Scheme 16). mate (PCC) on Al2O3, and obtained DFF in 58% yield (Table 11, entry 9). McDermott and Stockman also performed oxidation

with PCC in CH2Cl2 and reported slightly higher yield (Table 11, entry 10). Quantitative yield was obtained by Mehdi et al.,who

performed oxidation of HMF with (NH4)2[Ce(NO3)6](CAN) in the ionic liquid [EMIM][OTf] (1-ethyl-3-methylimidazolium triﬂuoromethylsulfonate) as a solvent (Table 11, entry 11). DFF Scheme 16 was obtained via CPO-catalyzed oxidation of HMF with H2O2. Optimum activity for the oxidation of HMF was observed at pH 5.1.1 Selective oxidation of the formyl group. There are 5, providing 89% conversion and 59% selectivity. The highest several examples in the literature for the selective oxidation selectivity 74% was observed at pH 3 with 25% conversion of the formyl group of HMF to HMFCA (Scheme 17) by of HMF (Table 11, entry 12). Cotier et al.124 (Table 11, using silver oxide 7c,117 or mixture of silver and copper(II) entry 13) reported the oxidation of HMF with 4-substituted

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 754–793 | 777 View Online

Table 11 Examples of oxidation of HMF to DFF using different efﬁciently catalyzed air oxidation of HMF to DFF. This kind of oxidants V catalyst was found to be active for air-oxidation of not only Entry Reaction conditions Yield (%) pure HMF but also HMF produced via dehydration of fructose. It was observed that the least expensive and most readily 125 1 Pb(OAc)2, pyridine 37 available catalyst that was used, V2O5, exhibited one of the 126,127 2 CrO3, pyridine 73, 68 highest efﬁciencies for the process, providing DFF in 58% yield 128 3 Ac2O, DMSO 76 128 calculated on HMF and 43% calculated on fructose. The best 4 HNO3, DMSO 31–67 128 · 5 N2O4,DMSO 76 result was observed for VOHPO4 0.5H2O (61% and 45% yield 117 6 BaMnO4 93 respectively) (Table 12, entry 6). Very detailed studies on air or 129 7 K2Cr2O7, DMSO, ultrasonic irradiation 75 129 O2 oxidation of HMF promoted by supported platinum catalysts 8 TMACC, Al2O3, ultrasonic irradiation 72 15,130 in aqueous solutions were recently reported by Lilga et al. They 9 PCC, Al2O3, ultrasonic irradiation 58 131 10 PCC, CH2Cl2 65 observed good conversion to DFF by using Pt/SiO2 catalyst ◦ 11132 [EMIM][OTf], CAN, 100 C 100 and air as oxidant in neutral solution. Similar conversion and 122 a 12 CPO/H2O2 124 selectivity were achieved with Pt–ZrO2 and air in acidic solution 13 4-Benzoyloxy-TEMPO, Ca(ClO)2 81 14133 Dess–Martin periodinane 74 (Table 12, entries 7 and 8). The electrochemical oxidation of HMF carried out in a a 89% conversion of HMF and 59% selectivity. divided cell at a platinum anode in a biphasic H2O–CH2Cl2 system was reported by Skowronski´ et al.140 Various salts were 2,2,6,6-tetramethylpiperidine-1-oxide (TEMPO) free radicals tested as supporting electrolytes. The best yield was 68%,

and supporting co-oxidants. The best co-oxidant was found to be obtained by using Na2HPO4 for 7 h. calcium hypochlorite in the presence of 4-benzoyloxy-TEMPO, The indirect oxidation of HMF providing DFF in 91% providing 81% yield. Dess–Martin oxidation of HMF was also yield has been performed via initial protection of the hydroxyl found to be effective, providing DFF in 74% yield (Table 11, group with 5-tert-butyldimethylsilyl (1a) and 5-trimethylsilyl entry 14). (1b) groups, followed by oxidation with N-bromosuccinimide Several authors have investigated the conversion of HMF (NBS) in the presence of azobisisobutyronitrile (AIBN)141 to DFF with oxygen, air or other economical and environ- (Scheme 18). mentally friendly oxidants using various metal-based catalysts. Partenheimer and Grushin134 reported the oxidation of HMF to DFF by using Co/Mn/Zr/Br or Co/Mn/Br catalysts and air as oxidant. They observed that Co/Mn/Zr/Br was the more active, providing higher conversion and selectivity. As expected, under comparable reaction conditions the conversion increased Scheme 18 Downloaded by University of Oxford on 06 April 2011 with temperature. (Table 12, entry 1). Carlini et al. (Table 12, entry 2) oxidized HMF to DFF in a biphasic water–methyl

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D 5.1.3 Oxidation of the formyl and hydroxyl groups. HMF isobutyl ketone (MIBK) system or in pure organic solvents using is a widely exploited precursor for the synthesis of FDA various metal-modiﬁed unsupported or SiO -supported vanadyl 2 (Scheme 19), which is a potential biorenewable replacement phosphate (VOP) catalysts under O or air pressure. The authors 2 monomer for terephthalic acid in polyethylene terephthalate reported up to 10% conversion and selectivity of 60–100% plastics, and has been described as one of the building blocks of when water–MIBK was the reaction medium. Higher conversion the future.116 rates were obtained in only MIBK as a solvent but with lower selectivity (98% conversion and 50% selectivity). DMF was found to be the best solvent for this transformation, providing up to 84% conversion and 97% selectivity. Amarasekara et al. reported the conversion of HMF to DFF at room temperature without formation of FDA using NaClO as oxidant and

catalyzed by Mn(III)-salen catalysts. Oxygen and H2O2 were also tested as more economical oxidants but both failed to give DFF in the presence of Mn(III)-salen catalyst (Table 12, entry 3). Cu and V catalysts supported in poly(4-vinylpyridine) crosslinked with 33% divinylbenzene (PVP) were tested for the heterogeneous catalytic aerobic selective oxidation of HMF. They provided higher activity and better chemoselectivity than the corresponding homogeneous catalysts, when the appropriate solvent was used for the reaction. The authors also observed Scheme 19 that V-containing polymeric catalysts were more active than those containing Cu (Table 12, entry 4). Lilga et al. patented 128 a method based on activated MnO2 oxidation of HMF to Morikawa oxidised HMF to FDA using N2O4 in DMSO DFF in good yields (Table 12, entry 5). Various inorganic andnitricacidinDMSO.El-Hajjet al.117 used nitric acid for this

vanadium compounds, such as V2O5 and VOHPO4·0.5H2O, transformation and obtained FDA in 24% yield. The authors

778 | Green Chem., 2011, 13, 754–793 This journal is © The Royal Society of Chemistry 2011 View Online

Table 12 Oxidation of HMF to DFF using metal catalysts

Entry (ref.) Reaction conditions HMF conversion (%) DFF selectivity (%)

1134 Co/Mn/Br/Zr or Co/Mn/Br, air (70 bar), 50–75 ◦C, 2 h 60–99 38–73 135 ◦ 2 VOP or MVOP, O2 or air, 80–150 C Upto98 Upto99 3136 Mn(III)-salen–NaClO, RT 89a 100 4137 Cu- or V-based catalysts, DMSO, air, 130–160 ◦C. Up to 85 Up to >99 138 5 MnO2,CH2Cl2, reﬂux, 8 h 80 100 87b ◦ a 6 VOHPO4·0.5H2O, DMSO, air (1 atm), 150 C61— 139 ◦ 7 5% Pt–SiO2, air (150 psi), 60–100 C6070 139 ◦ 8 5% Pt–ZrO2, air (150 psi), 100 C, 40% AcOH 50 70 a DFF yield.

reported higher yields by using Ag2O and HNO3 or KMnO4 conditions (85% selectivity and 100% conversion) was achieved 15,130 121 as oxidants (47% and 70% respectively). Cottier et al. also with O2 and Pt–ZrO2. Very recently, Davis et al. described O2 used nitric acid as oxidant and observed the formation of FDA oxidation of HMF to FDA promoted by supported Pt, Pd and and 5-formyl-2-furancarboxylic acid, which was resistant to Au catalysts under basic aqueous conditions. They observed that further oxidation under these conditions. The ratio between Pt/C and Pd/C were more selective towards FDA compared to

the products was found to be dependent on the reaction Au/C and Au/Ti2O under identical conditions, providing 79% conditions. and 71% selectivity respectively with 100% conversion of HMF

Several authors have published and patented methods for after 6 h. Higher pressures of O2 and concentrations of base the oxidation of HMF to FDA using more economical and were required for Au catalysts, resulting in up to 80% selectivity environmentally friendly oxidants and heterogeneous metal and 100% conversion after 22 h. catalysts. Vinke et al.142 reported the oxidation of HMF to FDA A supported gold nanoparticle catalyst was also found in near-quantitative yield under basic reaction conditions using to be effective for aerobic oxidative esteriﬁcation of HMF ◦ 143 Pt/Al2O3 as catalyst at 60 C. Air oxidation of HMF catalyzed (Scheme 19). Taarning et al. reported the formation of by Co/Mn/Br/Zr or Co/Mn/Br resulted in the formation of dimethyl furan-2,5-dicarboxylate (DFD) in excellent yield at ◦ FDA in up to 61% yield and 2-carboxy-5-formylfuran as a minor 130 C in MeOH and in the presence of an Au/TiO2 catalyst product in up to 3% yield.134 It was observed that the yield and basic conditions. When the reaction was carried out at room increased with the catalyst concentration and temperature, but temperature, the oxidation took place only at the formyl group, not with the addition of Zr to the Co/Mn/Br. Lew118 patented and 5-hydroxymethyl methylfuroate (HMMF) was obtained in an oxidation method using a platinum catalyst adsorbed on excellent yield. 144 Downloaded by University of Oxford on 06 April 2011 activated charcoal in basic aqueous solutions and bubbling More recently, Casanova et al. described a one-pot base-free oxygen through the solution, providing the corresponding acid aerobic oxidative esteriﬁcation of a methanol solution of HMF

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D 138 (FDA) in 95% yield. Lilga et al. patented a method for to dimethyl furan-2,5-dicarboxylate (DFD) by using a Au–CeO2

the synthesis of FDA from HMF using Pt–ZrO2 and air, catalyst, which could be recovered and reused with small loss of claiming 100% conversion and 98% selectivity.Gorbanev et al.119 activity but maintaining high selectivity. It was observed that oxidized HMF to FDA in up to 71% yield using commercial the temperature and the substrate-to-catalyst ratio affect the

heterogeneous Au/TiO2 nanoparticle catalyst in aq. NaOH at 20 reaction rate, but nevertheless a quantitative yield of DFD was

bar O2 and ambient temperature. The authors described also the always obtained. influence of oxygen pressure and the amount of hydroxide based 5.1.4 Oxidation of the furan ring. Oxidation of the furan on the selectivity and yield. Casanova et al.120 carried out this ring of HMF can take place under photo-oxygenation con- transformation using gold nanoparticle catalysts with various ditions. When alcohol was used as a reaction medium, the supports in basic aqueous conditions. The oxidative pathway oxidation takes place via the formation of an endoperoxide starts with the fast oxidation of HMF into HMFCA, and the followed by the attack of an alcohol on the formyl group or rate-limiting reaction step was the oxidation of HMFCA into on the 5-position of the furan ring, leading respectively to FDA. Au–CeO2 and Au–TiO2 catalysts were found to be the > hydroxybutenolide 2 (Scheme 20, route a) as a major product or most active, providing FDA in 99% yield. Under optimized 145 ◦ alkoxybutenolide 3 (Scheme 20, route b) as a minor product. reaction conditions (10 bar of O , 130 C and an NaOH/HMF 2 In addition, Marisa et al.146 reported the photochemical molar ratio of 4), it was shown that Au–CeO provides higher 2 oxidation of HMF in water providing 5-hydroxy-4-keto-2- activity and selectivity for FDA. Reduced substrate degradation pentenoic acid 4 (Scheme 21), which is a possible intermediate and increased lifetime of the catalyst was observed by perform- ◦ or monomer for the chemical industry. ing the reaction using a two-step procedure – first at 25 Cfor4 hfollowedby130◦C for 3 h. Screening of supported platinum 5.2 Reduction catalysts at different pH in a flow reactor was performed by Lilga et al.139 They obtained nearly quantitative yields of FDA using 5.2.1 Reduction of the furan ring and/or formyl group. stoichiometric aqueous Na2CO3, with air or O2 over Pt/C or Selective reduction of the formyl group of HMF leads to for- Pt/Al2O3, and 98% selectivity with 100% conversion of HMF mation of 2,5-bis(hydroxymethyl)furan 5, which is an important over Pt–ZrO2 at neutral pH with air. The best result under acidic chemical building block used in the production of polymers and

This journal is © The Royal Society of Chemistry 2011 Green Chem., 2011, 13, 754–793 | 779 View Online

copper/ruthenium (CuRu/C) catalyst, and obtained 7 in very good yields (76–79%). Two years later, Binder et al.103 reported the hydrogenolysis of crude HMF from corn stover in the presence of CuRu/C, providing 7 in 49% yield. Liujkx et al.151 described in the same year the formation of 7 by the hydrogenation of HMF in the presence of palladium catalyst. Very recently, Chidambaram and Bell61d reported the hydrogenation of either neat HMF or HMF obtained by dehydration of fructose in a mixture of 1-ethyl-3-methylimidazolium chloride Scheme 20 (EMIMCl) and acetonitrile promoted by carbon-supported transition metals. They observed the formation of a series of products, and in particular 2,5-dimethylfuran. A Pd/C catalyst was found to be the most active, providing 7 in 16% yield with 47% HMF conversion (Scheme 23).

Scheme 21

polyurethane foams.16 Several reports described the reduction of HMF to 5 with sodium borohydride in high yields.147 Turner et al.148 reported the synthesis of 5 in 76.9% yield by using Scheme 23 formalin and aq. NaOH. Nickel, copper chromite, platinum oxide, cobalt oxide, molybdenum oxide and sodium amalgam 5.3 Transformations of the formyl group catalysts were found to be effective for this transformation.12,15 Hydrogenation of HMF in aqueous media in the presence of 5.3.1 Reductive amination. Villard et al.152 reported a nickel, copper, platinum, palladium or ruthenium catalysts have method for the reductive amination of HMF with L-alanine been studied.149 2,5-Bis(hydroxymethyl)furan was obtained as or D-alanine in aq. NaOH in the presence of Raney R nickel. The the main product when copper and platinum catalysts were resulting (R)- or (S)-N-(1-carboxyethyl)-2-(hydroxymethyl)-5- used. 100% conversion and selectivity towards 5 was observed by (methylamino)furan 8 was isolated in 38% yield as the ammo- 149 using Pt/C, PtO2 or 2CuO·Cr2O3, while the presence of Pd/C nium salt (Scheme 24). or RaneyR nickel148–150 catalysts resulted in hydrogenation of the furan ring, 2,5-bis(hydroxymethyl)tetrahydrofuran 6 being

Downloaded by University of Oxford on 06 April 2011 formed as the major product in high yield (Scheme 22). Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D

Scheme 24

The synthesis of 2-(hydroxymethyl)-5-(aminomethyl)-furan 9 in 72% yield as an intermediate for further transformations was

performed by the reductive amination of HMF and liq. NH3 in presence of Raney nickel.153 Using the same catalyst but

switching from liq. NH3 to aq. MeNH2, 2-(hydroxymethyl)-5- (methylaminomethyl)furan 10 was obtained in an excellent yield of 91%153b (Scheme 25).

Scheme 22

Scheme 25 5.2.2 Reduction of the formyl and hydroxyl groups. Re- duction of both the formyl and hydroxyl groups in HMF Cukalovic and Stevens58 recently reported a procedure for is one of the synthetic pathways for the synthesis of 2,5- the synthesis of several 5-aminomethyl-2-furfuryl alcohols in dimethylfuran 7, which is of particular interest because of its very good yields starting from HMF and aromatic or aliphatic high energy content and potential use as a biofuel.60b Roman– primary amines. The reaction was performed by the in situ 60b Leshkov et al. recently reported a two-step process for the reduction with NaBH4 of the initially yielded aldimines. Water production of 7. They subjected HMF (obtained from D-fructose and bio-based solvents such as methanol and ethanol were in biphasic reactor) to hydrogenation over a carbon-supported tested as reaction media, as well as conventional and microwave

780 | Green Chem., 2011, 13, 754–793 This journal is © The Royal Society of Chemistry 2011 View Online

heating, which resulted in the increase of the reaction rate compared to the room-temperature reactions. (Scheme 26).

Scheme 30

Scheme 31

Scheme 26 Goodman and Jacobsen158 performed the 1,8- diazabicyclo[5.4.0]undec-7-ene (DBU)-mediated Horner– Kojiri et al.154 patented a method where they claimed the Wadsworth–Emmons reaction of HMF with phosphonate synthesis of novel indolopyrrolocarbazole derivatives and their imide 15. The reaction was carried out in THF, providing studies as antitumor agents. The HMF derivative 11 was N-[3-(5-hydroxymethylfuran-2-yl)acryloyl]benzamide 16 in achieved via two-step reductive amination (Scheme 27). very good yield (87%). Water was also described as a solvent, obviating any need to run the reaction under an inert atmosphere (Scheme 32).

Scheme 32

Scheme 27 Another Horner–Wadsworth–Emmons reaction of HMF as a part of a synthetic procedure for the synthesis of 3(5)-substituted Resin-bound compound 12 was obtained via reductive ami- pyrazoles was performed in one step using NaH without 155 nation of HMF by Sun and Murray and used for subsequent isolation of the intermediate a,b-unsaturated tosylhydrazone Diels–Alder transformations (Scheme 28). N-sodium salt 17 before the cyclization step. The ﬁnal product, 18, was isolated in 60% yield159(Scheme 33). Downloaded by University of Oxford on 06 April 2011 Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D

Scheme 28

5.3.2 Wittig-type reactions. A one-pot ﬂuoride-promoted Scheme 33 Wittig reaction of HMF was reported by Fumagalli et al.156 The corresponding ethyl 3-(5-(hydroxymethyl)furan-2-yl)acrylate 13 5.3.3 Baylis–Hillman reaction. The Baylis–Hillman reac- was obtained in 81% yield and 85% diastereoselectivity for the tion of HMF with methyl acrylate using stoichiometric base and E isomer (Scheme 29). aqueous medium was reported by Yu et al.160 The corresponding product 19 was obtained in 62% yield after 36 h using 1,4- diazabicyclo[2.2.2]octane (DABCO) as catalyst. One year later, Yu and Hu 160b described the formation of compound 20 in 61% yield after 48 h using the same reaction conditions and acryl Scheme 29 amide (Scheme 34). The same compound was synthesized in 90% yield (and the E conﬁguration), via a Wittig–Horner reaction with ethyl 2- (diethoxyphosphoryl)acetate157 (Scheme 30). HMF has been subjected to Taylor’s tandem oxidation–Wittig procedure by McDermott and Stockman,131 providing diester 14 in 87% yield in one step after 4 days as a 6.6:1 mixture of E,E and E,Z isomers (Scheme 31). Scheme 34

5.3.4 Acetal formation. Cottier et al.161 reported the reaction of HMF with trimethyl orthoformate in the presence of ytterbium sulfate supported on Amberlite 15, providing the corresponding [5-(dimethoxymethyl)-2-furyl]methanol 21 in 80% isolated yield. A higher yield (96%) was achieved by the condensation reaction of HMF and MeOH catalyzed by calcined and dehydrated Al-b-zeolite (Si/Al = 12.5:1 molar ratio, CP806)144 (Scheme 35). Scheme 38 Hanefeld et al.,167 who described a method for the synthesis of rhodanine derivative 26 in 73% yield (Scheme 39).

Scheme 35

Scheme 39 The synthesis of 5-hydroxymethyl-2-furaldehyde bis(5- formylfurfuryl) acetal 22 by using a strong-acid cation-exchange Shinobu et al.168 reported the synthesis of 3-((5- 162 resin as catalyst was patented by Terada et al. The authors (hydroxymethyl)furan-2-yl) methylene)-N-acetyl-2-oxoindoline claimed 2.3% yield of 22 and its application for the preparation 27 by using the reaction between HMF and N-acetyloxindole of ﬂavor-improving agents (Scheme 36). catalyzed by piperidine (Scheme 40).

Scheme 36 Scheme 40 The formation of the cyclic acetal 23 was achieved by 163 Downloaded by University of Oxford on 06 April 2011 Urashima et al. by a condensation reaction of levogalactosan The aldol reaction of HMF under microwave irradiation ◦ and HMF at 100 C (Scheme 37). in the presence of KF/Al2O3 was described by Suryawanshi 169 Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D et al. They obtained chalcone 28 in 76% yield and studied its antileishmanial activity (Scheme 41).

Scheme 37

5.3.5 Aldol condensations. Several authors reported the Scheme 41 aldol condensation reactions of HMF as a synthetic strategy for the synthesis of biologically active compounds and useful The synthesis and photochemistry of HMF chromone intermediates for the synthesis of biofuels.164 derivative 30 was investigated.170 The compound 30 was The aldol condensation reaction between HMF and acetophe- obtained by using a piperidine-catalyzed aldol condensation none was performed in water or methanol in the presence of of 1-(2-hydroxyphenyl)-3-(3,4,5-trimethoxyphenyl)propane-

base, providing 5-(hydroxymethyl)furfurylidene acetophenone 1,3-dione 29 with HMF followed by SeO2 oxidation; a in 80 and 82% yield respectively.165 DBU-induced condensation was also studied as a simpler and The synthesis in 91% yield of naturally occurring furan more efficient pathway (Scheme 42). derivative rehmanone C 24, which has displayed significant The synthesis and cytotoxicity studies of two HMF curcumin biological activity, was described by Quiroz-Florentino et al.166 analogues, 32 and 33, was reported.171 Boric anhydride was using a base-catalyzed aldol condensation with 2 equivalents of first added to the reactions in order to form a complex with acetone and HMF. The same reaction conditions, but using 0.5 2,4-pentanedione or compound 31 in order to protect the C- equivalents of acetone, provided the bis-derivative 25 in 60% 3 position from Knoevenagel condensation in such way that yield within 2 h (Scheme 38). aldol condensation takes place only at the terminal carbons. Another aldol condensation reaction of HMF towards the The target compounds 32 and 33 were obtained in 12% and 32% synthesis of biologically active compounds was reported by yield respectively (Scheme 43).

Scheme 46

The reactions of HMF with 1,2-aminothiols leads to the formation of new heterocyclic systems. Undheim et al.180 reported the synthesis of thiazolidine derivative 40 in 90% yield, by reaction of L-cysteine methyl ester and HMF in the presence of potassium acetate. Benzothiazole derivative 41 was obtained in quantitative yield from HMF and 2-aminobenzenethiol in the presence of acetic acid181 (Scheme 47).

Scheme 42

Scheme 47

The synthesis and studies of insecticidal activities of neon- Scheme 43 icotinoid 42 was reported from Shao et al.182 The ﬁnal prod- 5.3.6 Other reactions. There are a considerable number of uct was isolated as its hydrochloric acid salt in 72% yield literature examples of the reaction of the formyl group of HMF (Scheme 48). with amine-based compounds. Formation of arylhydrazones 34,172 semioxamazone 35,173 semicarbazone 36174 and thiosemi- Downloaded by University of Oxford on 06 April 2011 carbazone 37175 have been reported (Scheme 44). Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D

Scheme 48

183 Scheme 44 Karaguni et al. reported a novel HMF indene derivative 44 with anti-proliferative activity, obtained in 45% yield via a The reaction of HMF with aromatic amines leads to forma- one-step condensation protocol using 5-ﬂuoro-2-methylindene- tion of Schiff bases such as b-naphthylamine 38,176 Schiff bases 3-acetic acid 43 (Scheme 49). and azomethine salts177 (Scheme 45).

Scheme 49

Scheme 45 An adenosine receptor (A2A) antagonist, HMF derivative 45, was obtained by a three-component reaction of HMF184 The conversion of HMF to its oxime derivative 39178 (Scheme 50). (Scheme 46) in 95% yield has been described, and some biolog- The condensation reaction between HMF and 2,3,4,5- ically active HMF proline-oxime-containing peptide derivatives tetrahydropyridine was reported by Miller, providing the E- were also reported and studied.179 conﬁgured derivative 46 in 64% isolated yield185 (Scheme 51).

Scheme 50

Scheme 51 Scheme 54 Baliani et al.186 described a method for the conversion of the aldehyde function of HMF into nitrile 47 inaverygoodyield of 80% by using iodine in aqueous ammonia (Scheme 52).

Scheme 55

Scheme 52 organic solvents led to the formation of 51 in moderate to very Ramonczai and Vargha187 performed the reaction of HMF good yields (Table 13, entry 1). Sanda et al. reported a method for and diazomethane, providing 5-hydroxymethyl-2-acetofuran 48 the synthesis of 51 with similar yields using chlorotrimethylsilane

in 40% yield (Scheme 53). and CHCl3 or DMSO–Et2O as solvents, but CHCl3 was found to be the best (Table 13, entry 2). Very detailed studies on the Vilsmeier reaction as a synthetic pathway for the synthesis of 51 was reported by Sanda et al. Various reaction conditions and activating reagents were tested (Table 13, entries 3–5). DMF was found to be the best solvent for this reaction. The authors also reported preparative-scale experiments, and studied the Scheme 53 inﬂuence of different co-solvents, HMF concentrations and the Downloaded by University of Oxford on 06 April 2011 HMF was selectively carbonylated to 5-formylfuran-2-acetic rate of the POCl3 addition. Screening of SOBr2,PBr3 and PBr5 reagents for the synthesis

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D acid 49 in acidic aqueous media using a water-soluble palladium complex of trisulfonated triphenylphosphine (TPPTS) as of 5-bromomethyfurfural 52 was performed by Sanda et al., catalyst.188 The only observed byproduct was 5-methylfurfural and it was obtained in moderate yields (Table 14, entries 1–4). 50 formed from the reduction of HMF. The activity and Excellent yields were achieved by the reaction of HMF with selectivity of the carbonylation was found to be influenced by Me3SiBr, using CHCl3 or 1,1,2-trichloroethane as a solvent the Pd/TPPTS molar ratio. The best efficiency was observed (Table 14, entries 5 and 6). The treatment of HMF with solution for Pd/TPPTS = 6, which gave 90% conversion and 71.6% of HBr in Et2O or aq. HBr in CCl4 resulted in the formation of selectivity. The relationship between the selectivity and the 52 in moderate yields (Table 14, entries 7 and 8). nature of the anion of the acid component was also studied. 5.4.2 Esterification. The hydroxyl group of HMF can Acids of weakly or non-coordinating anions (such as phospho- undergo esterification reactions as a normal alcohol. There ric, trifluoroacetic, 4-toluenesulfonic and sulfuric acids) favor are several examples in the literature describing the formation carbonylation, giving 49 as the main product, while the acids of various aromatic HMF esters using the reaction of HMF of strongly coordinating anions (such as HBr and HI) decrease with the corresponding aromatic acid chlorides under basic the selectivity. In agreement with this, 50 was the only observed conditions. Jogia et al.193 reported the formation of HMF product when HI was used (Scheme 54).

5.4 Reactions of the hydroxyl group Table 13 Conversion of HMF into 51

5.4.1 Formation of halides. Halogen substitution of the Entry (ref.) Reaction conditions Yield (%) hydroxyl group of HMF can be easily performed, resulting 17d,189 Gaseous or 36% aq. HCl 64–87 in the formation of 5-halomethylfurfurals (Scheme 55), useful 189 2 Me3SiCl, CHCl3,6h 92 intermediates for the synthesis of HMF derivatives due to their 190 ◦ 3 SO2Cl2,DMF,50 C, 8 h 86 190 ◦ high reactivity. 4 POCl3,DMF,5 C, 5 h 92 190 ◦ Several synthetic protocols for the synthesis of 5- 5 MeSO2Cl, DMF, 65 C, 8 h 84 189 chloromethyfurfural 51 have been described in the literature. 6 SOCl2 53 7189 SOCl + pyridine 71 Treatment of HMF with gaseous or 36% aq. HCl in various 2

Table 14 Conversion of HMF into 52 several experiments in order to obtain 56 in high yields. Different solvents and acidic catalysts were tested, using a Dean–Stark Entry (ref.) Reaction conditions Yield (%) trap. The highest reported yield was 76% with 4-toluensulfonic

189 acid as catalyst in the presence of P2O5. Formation of 56 in 38% 1 SOBr2 63 189 145c 2 SOBr2 + pyridine 75 yield was reported by Cottier et al. by reﬂuxing HMF using 189 3 PBr3 +Et3N29a Dean–Stark trap in benzene in the presence of ion-exchange 189 4 PBr5 +CaCO3 62 + 189,191 189 resin IR 120 (H ) (Scheme 59). 5 Me3SiBr, CHCl3 98, 88 189 6 Me3SiBr, CHCl2CH2Cl 99 189,192 189 7 HBr, Et2O 64, 40 189 8 47% HBr, CCl4 70

ester derivatives 54a–c using reaction of HMF with aromatic acid chlorides 53a–c in pyridine in moderate yields (30–66%) (Scheme 56). Scheme 59

The 5-(methoxymethyl)furan-2-carbaldehyde 57 and derivative 58 were obtained in 50% and 24% yield respectively145a by using the reactions of HMF with MeOH in the presence of Amberlite IR 120 H+ for 57, and ethylene glycol with Py·HCl catalyst for 58 (Scheme 60).

Scheme 56

Compound 54a was obtained in 84% yield by Bognar et al.194 using the same reaction at room temperature. The reaction of acetic anhydride with HMF in the presence of NaOAc, leading to formation of 5-acetoxymethylfurfural in 81% yield, was reported by Cottier et al.145c The synthesis of 5-propionoxymethylfurfural 55 in 66% yield resulted from the reaction of propionic anhydride and HMF. Compound 55 is an important fungicide,15 and its synthesis was patented by Cope195 (Scheme 57). Scheme 60 Downloaded by University of Oxford on 06 April 2011 Oikawa et al.198 reported the synthesis of MPEG derivative of

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D HMF 59, which was used for further transformations in tandem Ugi/Diels–Alder reactions. The reaction was carried out with Scheme 57 iodine monochloride or MeOTf in the presence of molecular The author also claimed that the conversion can be modiﬁed sieves, providing 88% or 75% yield respectively (Scheme 61). by reacting HMF with propionic acid catalyzed by a small amount of strong acid. The widely exploited DIC/DMAP procedure was used by Gupta et al.196 for the esteriﬁcation of HMF and Sieber amide resin loaded with aliphatic dicarboxylic acids (Scheme 58). The resulting solid-phase-supported HMF esters were used for the combinatorial synthesis of furan-based Scheme 61 libraries of compounds. El-Hajj et al.199 reported the reaction between HMF and di- hydropyran catalyzed by pyridinium p-toluenesulfonate (PPTS), providing 5-(2-tetrahydropyranyl)oxymethyl furfural 60 in 72% yield (Scheme 62).

Scheme 58

5.4.3 Formation of ethers. The condensation reaction of Scheme 62 HMF with alcohols is a synthetic pathway which provides HMF ether derivatives. Timko and Cram147c described the synthesis of The Lewis acid-catalysed rearrangement of glycals in the 5,5¢-diformylfurfuryl ether 56 from HMF in 44% yield using presence of alcohols, known as the Ferrier reaction, was azeotropic distillation of water and toluene in the presence of used by Filho et al.200 to obtain a 2,3-unsaturated glyco- 4-toluenesulfonic acid. Chundury and Szmant197 carried out side ether derivative of HMF 61. Various catalytic systems

were tested, and the best yield of 93% was achieved using a mixture of lithium tetraﬂuoroborate and tin(II) chloride. Both a-andb-anomers were formed, in a ratio of 85:15 (Scheme 63).

Scheme 67

The FeCl3-catalyzed Friedel–Crafts reaction of HMF with o- xylene, providing 37% yield and 62% regioselectivity for the 4-alkylated product 73, was reported by Iovel et al.205 They explained the moderate yield (compared to the much higher yields when other benzyl alcohols were used) due to the “self- Scheme 63 arylation” of HMF during the reaction (Scheme 68). The synthesis of a b-glucopyranoside ether derivative of HMF (63b) in 32% yield after column chromatography was reported using the reaction between 62a/b and HMF in the presence of 161 BF3·Et2O. (Scheme 64). Scheme 68

5.5 Furan ring reactions Several synthetic transformations involving the furan ring in Scheme 64 HMF have been reported. Oxidation and reduction reactions have already been described in Sections 1.4 and 2.1. Several examples were reported for the formation of ether derivatives of HMF using the widely exploited Williamson 5.5.1 Hydrolysis. It is known that cleavage of the furan ring synthesis, as presented in Table 15. Other examples for the of HMF takes place under acidic conditions.206 This process was protection of the hydroxyl group of HMF are obtained as its found to be very important, especially when starting directly tert-butyldimethylsilyl ether141,161,201 or trimethylsilyl ether141 in from biomass due to the formation of levulinic acid (LA) DMF using imidazole as catalyst. as the ﬁnal product. LA, together with its derivatives, are important chemical building blocks with various applications 5.4.4 Other reactions. The reaction of HMF with N,O- such as production of fuels, fuel additives, and polymers.207 (bisphenoxycarbonyl)hydroxylamine 68 under Mitsunobu con- Two possible pathways were proposed by Horvat et al.208 for ditions, providing hydroxyurea derivative 69, was reported by this transformation. Pathway A goes via 2,3 water addition to Downloaded by University of Oxford on 06 April 2011 Lewis et al.203 (Scheme 65). HMF and leads to polymerization, while pathway B proceeds

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D by 4,5 addition of water, resulting in the formation of 2,5-dioxo- 3-hexenal 74, which fragments to levulinic 75 and formic acid (Scheme 69).

Scheme 65

Dow et al.204 described the reaction of HMF and diethylazodi- carboxylate (DEAD) using Mitsunobu-type reaction conditions in the absence of other nucleophiles. Hydrazine derivative 70 was obtained in 12% yield (Scheme 66).

Scheme 69

A number of studies on the kinetics of acid-catalyzed HMF Scheme 66 degradation to LA have been reported using different acid catalysts, acid concentrations and temperature ranges.209 More Cotier et al.145c reported the reaction of HMF and ben- recently, very detailed kinetic studies on this process were zonitrile or acetonitrile catalyzed by triﬂuoromethanesulfonic reported by Heeres et al.210 The experiments were carried out acid resulted in 5-benzamidomethyl-2-furfural 71 and 5- with various acid catalysts and acid concentrations between acetamidomethyl-2-furfural 72 in 48% and 50% yield respec- 0.05–0.1 M in a temperature window of 98–181 ◦C. The effect tively (Scheme 67). of the initial concentrations of HMF was also studied in the

Table 15 Williamson synthesis of HMF ester derivatives

Entry (ref.) Reaction conditions Product Yield

1166 MeI, NaH, THF, RT, 16 h 94%

145c 2 Benzyl bromide, Ag2O, DMF, RT, 53 h 72%

3161 11%

CH2Cl2,Ag2O, RT, 5 h 202 a 4 Ph3CCl, pyridine, 40 min 4.5 g

a Starting from 7.8 g of HMF.

range 0.1–1 M. The LA was obtained in yields up to 94% using by Pachmayr, and later by Koch, but the yields were not sulfuric acid as the catalyst. provided (Table 16, entry 4). The synthesis and taste-enhancing The hydrothermolysis of HMF at 27.5 MPa and 290–400 ◦C activity of compound 80 were reported by Ottinger et al.213 was performed by Luijkx et al.,211 and resulted in 1,2,4- (Table 16, entry 5). The enantiomer (+)-(S)-80 was found to benezenetriol 76 as the major product in up to 46% yield and be the physiologically active one, whereas (-)-(R)-80 did not 50% HMF conversion. The authors also described a possible affect sweetness perception at all. Racemization was observed pathway for this transformation (Scheme 70). during the synthesis of betaine (+)-(S)-80 by reaction between HMF and L-alanine under alkaline conditions, resulting 152

Downloaded by University of Oxford on 06 April 2011 in lower taste-enhancing activity. Villard et al. (Table 16, entry 6) reported an alternative two step synthetic protocol

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D for the preparation of enantiopure ﬁnal products although in lower yields. Soldo and Hofmann extended these investigations by the synthesis and screening of the bitterness-suppressing properties of pyridinium betaines 81a–c 214 (Table 16, entry 7).

5.6 Synthesis of heteromacrocycles

Scheme 70 The hydroxyl and aldehyde functional groups present in HMF are appealing structural motifs for the synthesis of heteromacrocycles, which are of considerable interest due to their biological 5.5.2 Synthesis of betaine salts. The synthesis of betaine activity and complexation properties. salts from HMF with primary amines or amino acids is Heteromacrocyclic compounds 84 and 85 were prepared important, because they seem to be promising targets for further from 2,5-disubstituted furans 82 and 83 via ring-closing research due to their taste-modulatory activity.212 N-Methyl- metathesis (RCM) catalyzed by commercially available benzyli- 3-oxidopyridinium betaine 77 was obtained via the one-step dene bis(tricyclohexylphosphine)ruthenium dichloride (Grubbs 147a reaction of HMF and MeNH2 in low yield (Table 16, entry 1). catalyst). The formation of 84 instead of 86 was explained by Much higher yield was achieved by Muller¨ et al.,153b who the authors as due to possible conformational constraints in the reported a two-step protocol for the synthesis of 77. First, they original substrate (Scheme 71). performed reductive amination to obtain 2-(hydroxymethyl)-5- Another attempt to obtain 86 in two steps from HMF chloro (aminomethyl)furan 10, which was in turn exposed to bromine alcohol 87 was also unsuccessful, the macrocycle 84 again being in water to give 77 in good yield (Table 16, entry 2). The the only product formed (35% yield) (Scheme 72). synthesis of 78 in moderate yields was performed in one step Heteromacrocycles 88–91 were obtained starting from in basic conditions (Table 16, entry 3). The betaine salt 79 HMF in low to moderate yields.217 These kind of com- was obtained by the reaction of HMF and N-acetyllysine pounds are themselves hosts for binding organic and

Table 16 Examples on the transformation of HMF to 2-hydroxymethyl-pyridinium derivatives

Entry (ref.) Starting compound Reaction conditions Product Yield (%)

215 a 1 HMF MeNH2,EtOH–H2O 10

153b ◦ b 2 Br2,H2O, 0 C 78

215a,216 3 HMF 1-Propylamine, H2O–EtOH, NaOH, pH 9.4, 43–45 reﬂux, 3 days.

4215 HMF N-Acetyllysine, EtOH, NaOH —

213 5 HMF Alanine, NaOH, H2O–EtOH, pH 9.4, reﬂux, 48 h 51

6152 HMF (1) L-orD-Alanine, water, aq. NaOH (32%), pH 13b ◦ 8.5, Ni, H2, RT, 5 bar, 48 h; (2) water, 0 C, Br2–MeOH (0.5 h), RT (1 h) Downloaded by University of Oxford on 06 April 2011

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D 7214 HMF Glycine, b-alanine, or g-aminobutyric acid, 22 (81a),

H2O–EtOH, NaOH, pH 9.4, RT (1.5 h), reﬂux 12 (81b), (24 h) 5(81c)

a 215b 215a Pachmayr et al. reported 10% in the presence of AcOH while Koch et al. used HMF with aq. MeNH2 for 3 days under reﬂux and used the crude product directly for further transformation. b Yield from the second step.

inorganic cations. More importantly, they can serve as A macrocyclic fluorescent receptor 95 was synthesised from starting materials for preparing host compounds whose pe- HMF, and binding studies with three different types of dicar- riphery is lined with a variety of binding and shaping units boxylic acids were performed147d (Scheme 75). (Scheme 73). 218 Waddell et al. developed a method for the synthe- 6 Conclusions sis of heteromacrocyclic derivatives of HMF, 93 and 94, which have the same substituents on the ‘eastern’ side as Recently, considerable efforts have been made in order to the erythromycin-derived azalide antibiotics 9-deoxo-9a-aza- achieve more efficient integrated processes for the transforma- 9a-methyl-8a-homoerythromycin A and 9-deoxo-8a-aza-8a- tion of carbohydrates into HMF. Considerable improvement methyl-8a-homoerythromycin A, but more functionalized ‘west- has been reported for the conversion of fructose to HMF, ern’ sides, due to the introduction of a tetrahydrofuran ring whereas the transformation of glucose, sucrose and cellulose derived from HMF. Compounds 93 and 94 were prepared in remains difficult. The main drawback for the transformation of several steps from erythromycin-derived acyclic fragment 92 glucose-based carbohydrates to HMF is the isomerisation to and HMF protected as its 5-tert-butyldimethylsilyl ether, 4a fructose, which requires different conditions from the fructose (Scheme 74). dehydration step. As a result, an overall process based on

Scheme 73

Scheme 71 Downloaded by University of Oxford on 06 April 2011 Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D Scheme 74

Scheme 72

two independent steps is more desirable, and has been already explored by combining base/acid66 or enzyme/acid80 catalytic systems. More efﬁcient reaction conditions (lower temperature, and higher carbohydrate initial concentration), higher conversions and HMF selectivity are desirable, and these processes have to be environmentally friendly. The presence of two functional groups in HMF, combined with the furan ring, makes it an appealing starting material for various chemical transformations. Several transformations of HMF as a substrate involving formyl or hydroxyl groups (or both) have been reported in the literature. Serious attention was paid to the oxidation and reduction, because they provide Scheme 75 convenient synthetic pathways for the production of chemical building blocks for the polymer industry and biofuels starting temperatures) are required for this transformation, and future from renewable materials. Heterogeneous metal catalysts and air investigations will focus on the resolution of these issues. Ionic as the oxidant is the modern approach for performing selective liquids seem to be a promising medium for the oxidation, but oxidation of HMF and synthesis of FDA and DFF. A lot of there are still few examples of their application, especially for research in this direction has already been done, and some really the synthesis of DFF. good results have been achieved for the oxidation of HMF to The conversion of 2,5-bis(hydroxymethyl)furan (which is FDA in water – an economical and environmentally friendly already in use for the production of polyurethane foams) solvent. Nevertheless, considerable amounts of base (and high to 2,5-dimethylfuran (a compound of considerable interest

because of its potential as biofuel and fuel additive) has been 11 A. Gaset, J. P. Gorrichon and E. Truchot, Inf. Chim., 1981, 212, carried out by hydrogenolysis. Very good results have been 179. 12 A. Faury, A. Gaset and J. P. Gorrichon, Inf. Chim., 1981, 214, 203. reported for the synthesis of 2,5-bis(hydroxymethyl)furan by 13 B. F. M. Kuster, Starch/Staerke, 1990, 42, 314–321. hydrogenolysis promoted by Pt, while the reduction of HMF 14 L. Cottier and G. Descotes, Trends Heterocycl. Chem., 1991, 2, 233. to 2,5-dimethylfuran has resulted in low to moderate yields 15 J. Lewkowski, Arkivoc, 2001, 2, 17–54. and selectivity. Screening of more effective catalysts for this 16 C. Moreau, M. N. Belgacem and A. Gandini, Top. Catal., 2004, 27, 11–30. transformation needs to be considered. 17 A. Boisen, T. B. Christensen, W. Fu, Y. Y. Gorbanev, T. S. Hansen, Some catalysts were found to be effective for the transfor- J. S. Jensen, S. K. Klitgaard, S. Pedersen, A. Riisager, T. Stahlberg mations of not only neat HMF, but also crude HMF resulting and J. M. Woodley, Chem. Eng. Res. Des., 2009, 87, 1318–1327. from the dehydration of carbohydrates. This approach leads to 18 Y. G. Zhang and J. Y. G. Chan, Energy Environ. Sci., 2010, 3, 408– 417. reduced reaction costs, and it is important for industry that the 19 (a) F. W. Lichtenthaler and S. Peters, C. R. Chim., 2004, 7, 65–90; search for new catalysts leading to high yields and selectivity (b)X.L.Tong,Y.MaandY.D.Li,Appl. Catal., A, 2010, 385, 1–13; continues. (c) A. Gandini, Polym. Chem., 2010, 1, 245–251; (d) R. J. Ulbricht, S. J. Northup and J. A. Thomas, Fundam. Appl. Toxicol., 1984, 4, In view of the reported data, many questions about HMF 843–853. and its derivative product(s) remain to be answered. There are 20 Search made in ISI Web of Knowledge with topic ‘5- a number of reports that do not agree about the toxicology of hydroxymethylfurfural’. this compound to humans, and therefore more experimental 21 L. W. Kroh, Food Chem., 1994, 51, 341–416. 22 H. F. Erbersdobler and A. Hupe, Z. Ernaehrungswiss., 1991, 30, work needs to be developed – for instance, in the study of the 46–49. toxicology of derivatives of HMF. As far as its effect on the 23 (a) H. E. Berg and M. A. J. S. Van Boekel, Netherlands Milk Dairy wider environment is concerned, it is probably not a problem, J., 1994, 48, 157–175; (b) C. Martin and L. J. Jonsson, Enzyme Microb. Technol., 2003, 32, 386–395. because HMF is mainly produced during food processing and 24 F.J.Morales, C. Romero and S. Jimenez-P´ erez,´ J. Agric. Food Chem., from there proceeds directly to the human food chain, so it 1997, 45, 1570. is unlikely to enter in the environment in amounts to cause 25 (a) A. Ramirez-Jimenez, E. Guerra-Hernandez and B. Garcia- concern. The only exception might be in the case of industrial Villanova, J. Agric. Food Chem., 2000, 48, 4176–4181; (b)L.A. Ameur, G. Trystram and I. Birlouez-aragon, Food Chem., 2005, 98, synthesis of HMF,using natural resources (or otherwise), as well 790–796. as its derivatization to form new important molecules, including 26 (a) P. Fernandez-Artigas, E. Guerra-Hernandez and B. Garcia- biofuels. Experimental studies should be carried out in order Villanova, J. Agric. Food Chem., 1999, 47, 2872–2878; (b)A. to ascertain that such processes will not generate toxic waste Ram´ırez-Jimenez,´ B. Garc´ıa-Villanova and E. Guerra-Hernandez, Food Res. Int., 2000, 33, 833–838; (c) A. Ramirez-Jimenez, B. products. Biodegradability studies of HMF derivatives should Garc´ıa-Villanova and E. Guerra-Hernandez, J. Sci. Food Agric., be performed in order to investigate how much greener this new 2001, 81, 513–518; (d)J.A.Rufian-Henares,´ C. Delgado-Andrade source of energy would be if applied on a large scale. The use and F. J. Morales, J. Cereal Sci., 2006, 43, 63–69. 27 B. Fallico, E. Arena and M. Zappala,` Food Chem., 2003, 81, 569– of lignocelluloses for biofuel production would be of particular 573. Downloaded by University of Oxford on 06 April 2011 importance, since this is a cheap and readily available natural 28 L. Ait Ameur, O. Mathieu, V.Lalanne, G. Trystram and I. Birlouez- source that would decrease the amounts of petroleum-derived Aragon, Food Chem., 2007, 101, 1407–1416. Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D pollutants in the air, thus contributing to a greener environment. 29 M. J. Taherzadeh, L. Gustafsson, C. Niklasson and G. Liden, Appl. Microbiol. Biotechnol., 2000, 53, 701–708. 30 C. F. Wahlbom and B. Hahn-Hagerdal, Biotechnol. Bioeng., 2001, 7 Acknowledgements 78, 172–178. 31 Z. L. Liu, P. J. Slininger, B. S. Dien, M. A. Berhow, C. P. Kurtzman We thank the Fundaçao˜ para a Cienciaˆ e Tecnologia and S. W. Gorsich, J. Ind. Microbiol. Biotechnol., 2004, 31, 345– (POCI 2010) and FEDER (Ref.:SFRH/BD/28242/2006, 352. Biotechnol. Bioeng. PTDC/QUI/66695/2006, PTDC/QUI/73061/2006, 32 P. J. Slininger, S. W. Gorsich and Z. L. Liu, , 2009, 102, 778–790. PTDC/QUI/71331/2006) for financial support. 33 C. Hu, X. Zhao, J. Zhao, S. Wu and Z. K. Zhao, Bioresour. Technol., 2009, 100, 4843–4847. 8 References 34 J. Zaldivar, A. Martinez and L. O. Ingram, Biotechnol. Bioeng., 1999, 65, 24–33. 1 B. Kamm, Angew. Chem., Int. Ed., 2007, 46, 5056–5058. 35 X. Chen, Z. H. Li, X. X. Zhang, F. X. Hu, D. D. Y.Ryu and J. Bao, 2 D. Brownlie, J. Soc. Chem. Ind., 1940, 59, 671–675. Appl. Biochem. Biotechnol., 2009, 159, 591–604. 3(a) A. Corma, S. Iborra and A. Velty, Chem. Rev., 2007, 107, 2411– 36 (a) M. Murkovic and N. Pichler, Mol. Nutr. Food Res., 2006, 50, 2502; (b) C. H. Christensen, J. Rass-Hansen, C. C. Marsden, E. 842–846; (b) R. L. Prior, X. Wu and L. Gu, J. Agric. Food Chem., Taarning and K. Egeblad, ChemSusChem, 2008, 1, 283–289. 2006, 54, 3744–3749. 4 J. F. Saeman, Ind. Eng. Chem., 1945, 37, 43–52. 37 T. Husoy, M. Haugen, M. Murkovic, D. Jobstl, L. H. Stolen, T. 5 G. Dull, Chem. Ztg., 1895, 216. Bjellaas, C. Ronningborg, H. Glatt and J. Alexander, Food Chem. 6 J. Kiermayer, Chem. Ztg., 1895, 19, 1003. Toxicol., 2008, 46, 3697–3702. 7(a) H. J. H. Fenton and M. Gostling, J. Chem. Soc., 1901, 79, 807– 38 (a) C. L. Hryncewicz, M. Koberda and M. S. Konkowski, J. Pharm. 816; (b) H. J. H. Fenton and F. Robinson, J. Chem. Soc., 1909, 95, Biomed. Anal., 1996, 14, 429–434; (b) E. Jellum, H. C. Borresen and 1334–1340; (c) T. Reichstein, Helv. Chim. Acta, 1926, 9, 1066–1068; L. Eldjarn, Clin. Chim. Acta, 1973, 47, 191–201; (c) A. P.Wieslander, (d) T. Reichstein and H. Zschokke, Helv. Chim. Acta, 1932, 15, 249– A. H. Andren, C. Nilsson-Thorell, N. Muscalu, P.T.Kjellstrand and 253; (e) W. N. Haworth and W. G. M. Jones, J. Chem. Soc., 1944, B. Rippe, Peritoneal Dialysis Int., 1995, 15, 348–352. 667–670. 39 C. Janzowski, V. Glaab, E. Samimi, J. Schlatter and G. Eisenbrand, 8F.H.Newth,Adv. Carbohydr. Chem., 1951, 6, 83–106. Food Chem. Toxicol., 2000, 38, 801–809. 9 C. J. Moye and Z. S. Krzeminski, Aust. J. Chem., 1963, 16, 258. 40 (a) I. Severin, C. Dumont, A. Jondeau-Cabaton, V.Graillot and M. 10 M. S. Feather and J. F. Harris, in Dehydration Reactions of C. Chagnon, Toxicol. Lett., 2010, 192, 189–194; (b) K. H. Wagner, Carbohydrates, ed. R. S. Tipson and H. Derek, Academic Press, S. Reichhold, K. Koschutnig, S. Cheriot and C. Billaud, Mol. Nutr. 1973. Food Res., 2007, 51, 496–504.

41 C. M. Brands, G. M. Alink, M. A. van Boekel and W. M. Jongen, 73 (a) X. H. Qi, M. Watanabe, T. M. Aida and R. L. Smith, J. Agric. Food Chem., 2000, 48, 2271–2275. ChemSusChem, 2009, 2, 944–946; (b)X.H.Qi,M.Watanabe,T. 42 X. Ding, M. Y. Wang, Y. X. Yao, G. Y. Li and B. C. Cai, J. M. Aida and R. L. Smith, Green Chem., 2009, 11, 1327–1331. Ethnopharmacol., 2010, 128, 373–376. 74 D. M. Roberge, L. Ducry, N. Bieler, P.Cretton and B. Zimmermann, 43A.S.Lin,K.Qian,Y.Usami,L.Lin,H.Itokawa,C.Hsu,S.L. Chem. Eng. Technol., 2005, 28, 318–323. Morris-Natschke and K. H. Lee, J. Nat. Med., 2008, 62, 164–167. 75 T. Tuercke, S. Panic and S. Loebbecke, Chem. Eng. Technol., 2009, 44 O. Abdulmalik, M. K. Safo, Q. Chen, J. Yang, C. Brugnara, K. 32, 1815–1822. Ohene-Frempong, D. J. Abraham and T. Asakura, Br. J. Haematol., 76 F. Ilgen, D. Ott, D. Kralisch, C. Reil, A. Palmberger and B. Konig, 2005, 128, 552–561. Green Chem., 2009, 11, 1948–1954. 45 L. J. Durling, L. Busk and B. E. Hellman, Food Chem. Toxicol., 77 T.M. Aida, Y.Sato, M. Watanabe, K. Tajima, T.Nonaka, H. Hattori 2009, 47, 880–884. and K. Arai, J. Supercrit. Fluids, 2007, 40, 381–388. 46 H. Glatt, H. Schneider and Y. Liu, Mutat. Res., 2005, 580, 41–52. 78 Y. Roman-Leshkov and J. A. Dumesic, Top. Catal., 2009, 52, 297– 47 P. Khondkar, M. M. Rahman and A. Islam, Phytother. Res., 2005, 303. 19, 816–817. 79 H. B. Zhao, J. E. Holladay, H. Brown and Z. C. Zhang, Science, 48 X. M. Zhang, C. C. Chan, D. Stamp, S. Minkin, M. C. Archer and 2007, 316, 1597–1600. W. R. Bruce, Carcinogenesis, 1993, 14, 773–775. 80 R. L. Huang, W.Qi, R. X. Su and Z. M. He, Chem. Commun., 2010, 49 Y. J. Surh, A. Liem, J. A. Miller and S. R. Tannenbaum, Carcino- 46, 1115–1117. genesis, 1994, 15, 2375–2377. 81 M. Y.Chernyak, M. A. Smirnova and V.E. Tarabanko, RU2363698- 50 W. Teubner, W. Meinl, S. Florian, M. Kretzschmar and H. Glatt, C1, 2009. Biochem. J., 2007, 404, 207–215. 82 S. Wu, H. Fan, Y. Xie, Y. Cheng, Q. Wang, Z. Zhang and B. Han, 51 Y. C. Lee, M. Shlyankevich, H. K. Jeong, J. S. Douglas and Y. J. Green Chem., 2010, 12, 1215–1219. Surh, Biochem. Biophys. Res. Commun., 1995, 209, 996–1002. 83 F. S. Asghari and H. Yoshida, Ind. Eng. Chem. Res., 2006, 45, 2163– 52 B. H. Monien, H. Frank, A. Seidel and H. Glattt, Chem. Res. 2173. Toxicol., 2009, 22, 1123–1128. 84 M. Bicker, J. Hirth and H. Vogel, Green Chem., 2003, 5, 280–284. 53 N. Bakhiya, B. Monien, H. Frank, A. Seidel and H. Glatt, Biochem. 85 M. Bicker, D. Kaiser, L. Ott and H. Vogel, J. Supercrit. Fluids, 2005, Pharmacol., 2009, 78, 414–419. 36, 118–126. 54 V. B. Godfrey, L. J. Chen, R. J. Grifﬁn, E. H. Lebetkin and L. T. 86 P. Vinke and H. Vanbekkum, Starch/Staerke, 1992, 44, 90–96. Burka, J. Toxicol. Environ. Health, Part A, 1999, 57, 199–210. 87 (a) C. Lansalot-Matras and C. Moreau, Catal. Commun., 2003, 4, 55 C. Svendsen, T. Husoy, H. Glatt, J. E. Paulsen and J. Alexander, 517–520; (b) G. A. Halliday, R. J. Young and V. V. Grushin, Org. Anticancer Res., 2009, 29, 1921–1926. Lett., 2003, 5, 2003–2005. 56 M. J. Antal, W. S. L. Mok and G. N. Richards, Carbohydr. Res., 88 K. Shimizu, R. Uozumi and A. Satsuma, Catal. Commun., 2009, 1990, 199, 91–109. 10, 1849–1853. 57 H. E. Vandam, A. P. G. Kieboom and H. Vanbekkum, 89 J. N. Chheda and J. A. Dumesic, Catal. Today, 2007, 123, 59–70. Starch/Staerke, 1986, 38, 95–101. 90 X. H. Qi, M. Watanabe, T. M. Aida and R. L. Smith, Green Chem., 58 A. Cukalovic and C. V.Stevens, Green Chem., 2010, 12, 1201–1206. 2008, 10, 799–805. 59 C. Moreau, R. Durand, S. Razigade, J. Duhamet, P. Faugeras, P. 91 A. S. Amarasekara, L. D. Williams and C. C. Ebede, Carbohydr. Rivalier, P.Ros and G. Avignon, Appl. Catal., A, 1996, 145, 211–224. Res., 2008, 343, 3021–3024. 60 (a) S. Lima, P. Neves, M. M. Antunes, M. Pillinger, N. Ignatyev 92 S. Murugesan and R. J. Linhardt, Curr. Org. Synth., 2005, 2, 437– and A. A. Valente, Appl. Catal., A, 2009, 363, 93–99; (b) Y.Roman- 451. Leshkov, C. J. Barrett, Z. Y. Liu and J. A. Dumesic, Nature, 2007, 93 (a) Q. B. Liu, M. H. A. Janssen, F.van Rantwijk and R. A. Sheldon, 447, 982 -U985; (c) J. N. Chheda, Y. Roman-Leshkov and J. A. Green Chem., 2005, 7, 39–42; (b)A.A.Rosatella,L.C.Brancoand Downloaded by University of Oxford on 06 April 2011 Dumesic, Green Chem., 2007, 9, 342–350; (d) Y.Roman-Leshkov, J. C. A. M. Afonso, Green Chem., 2009, 11, 1406–1413. N. Chheda and J. A. Dumesic, Science, 2006, 312, 1933–1937; (e)F. 94 Q. X. Bao, K. Qiao, D. Tomida and C. Yokoyama, Catal. Commun.,

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D Benvenuti, C. Carlini, P. Patrono, A. M. R. Galletti, G. Sbrana, M. 2008, 9, 1383–1388. A. Massucci and P. Galli, Appl. Catal., A, 2000, 193, 147–153. 95 C. Fayet and J. Gelas, Carbohydr. Res., 1983, 122, 59–68. 61 (a) S. Q. Hu, Z. F. Zhang, Y. X. Zhou, B. X. Han, H. L. Fan, W. 96 (a) J. Lifka, B. Ondruschka and A. Stark, DE102008009933-A1, J. Li, J. L. Song and Y. Xie, Green Chem., 2008, 10, 1280–1283; DE102008009933-A1, 2009; C07D-307/50 200956; (b) J. Lifka, B. (b) S. Q. Hu, Z. F. Zhang, J. L. Song, Y. X. Zhou and B. X. Han, Ondruschka and A. Stark, DE102008009933-A1, 2009. Green Chem., 2009, 11, 1746–1749; (c) S. Q. Hu, Z. F. Zhang, Y. X. 97 H. Zhao, J. E. Holladay and Z. C. Zhang, US2008033187-A1; Zhou, J. L. Song, H. L. Fan and B. X. Han, Green Chem., 2009, 11, WO2008019219-A1; EP2054400-A1, 2008. 873–877; (d) M. Chidambaram and A. Bell, Green Chem., 2010, 12, 98G.Yong,Y.G.ZhangandJ.Y.Ying,Angew. Chem., Int. Ed., 2008, 1253–1262. 47, 9345–9348. 62 J. Y. G. Chan and Y. G. Zhang, ChemSusChem, 2009, 2, 731–734. 99C.Z.Li,Z.H.ZhangandZ.B.K.Zhao,Tetrahedron Lett., 2009, 63 H. P. Yan, Y. Yang, D. M. Tong, X. Xiang and C. W. Hu, Catal. 50, 5403–5405. Commun., 2009, 10, 1558–1563. 100 Z. H. Zhang and Z. B. K. Zhao, Bioresour. Technol., 2010, 101, 64 Y.T. Zhang, H. B. Du, X. H. Qian and E. Y.X. Chen, Energy Fuels, 1111–1114. 2010, 24, 2410–2417. 101 S. Yu, H. M. Brown, X. W.Huang, X. D. Zhou, J. E. Amonette and 65 (a)R.M.MusauandR.M.Munavu,Biomass, 1987, 13, 67–74; Z. C. Zhang, Appl. Catal., A, 2009, 361, 117–122. (b) M. L. Ribeiro and U. Schuchardt, Catal. Commun., 2003, 4, 102 J. A. Chun, J. W.Lee, Y.B. Yi, S. S. Hong and C. H. Chung, Korean 83–86; (c) X. H. Qi, M. Watanabe, T. M. Aida and R. L. Smith, Ind. J. Chem. Eng., 2010, 27, 930–935. Eng. Chem. Res., 2008, 47, 9234–9239. 103 J. B. Binder and R. T. Raines, J. Am. Chem. Soc., 2009, 131, 1979– 66 A. Takagaki, M. Ohara, S. Nishimura and K. Ebitani, Chem. 1985. Commun., 2009, 6276–6278. 104 (a) J. B. Binder and R. T. Raines, WO2009155297-A1; 67 D. W. Brown, A. J. Floyd, R. G. Kinsman and Y. Roshanali, J. US2010004437-A1; (b) J. B. Binder and R. T. Raines, Chem. Technol. Biotechnol., 1982, 32, 920–924. WO2009155297-A1; US2010004437-A1, 2009. 68 C. Moreau, A. Finiels and L. Vanoye, J. Mol. Catal. A: Chem., 105 E. A. Pidko, V.Degirmenci, R. A. van Santen and E. J. M. Hensen, 2006, 253, 165–169. Angew. Chem., Int. Ed., 2010, 49, 2530–2534. 69 B. F. M. Kuster and J. Laurens, Starch/Staerke, 1977, 29, 172– 106 M. Watanabe, Y. Aizawa, T. Iida, T. M. Aida, C. Levy, K. Sue and 176. H. Inomata, Carbohydr. Res., 2005, 340, 1925–1930. 70 J. D. Chen, B. F.M. Kuster and K. Vanderwiele, Biomass Bioenergy, 107 X. H. Qi, M. Watanabe, T. M. Aida and R. L. Smith, Catal. 1991, 1, 217–223. Commun., 2008, 9, 2244–2249. 71 T. S. Hansen, J. M. Woodley and A. Riisager, Carbohydr. Res., 2009, 108 X. H. Qi, M. Watanabe, T. M. Aida and R. L. Smith, Catal. 344, 2568–2572. Commun., 2009, 10, 1771–1775. 72C.Sievers,I.Musin,T.Marzialetti,M.B.V.Olarte,P.K.Agrawal 109 A. Chareonlimkun, V. Champreda, A. Shotipruk and N. and C. W. Jones, ChemSusChem, 2009, 2, 665–671. Laosiripojana, Bioresour. Technol., 2010, 101, 4179–4186.

110 F. S. Asghari and H. Yoshida, Carbohydr. Res., 2006, 341, 2379– 146 C. Marisa, D. Ilaria, R. Marotta, A. Roberto and C. Vincenzo, J. 2387. Photochem. Photobiol., A, 2010, 210, 69–76. 111 (a) H. Ishida and K.-i. Seri, J. Mol. Catal. A: Chem., 1996, 112, 147 (a) L. Cottier, G. R. Descotes and Y. Soro, Synth. Commun., 2003, L163–L165; (b) K. Seri, Y. Inoue and H. Ishida, Bull. Chem. Soc. 33, 4285–4295; (b)C.J.Moye,Pure. Appl. Chem., 1964, 14, 161; Jpn., 2001, 74, 1145–1150. (c)J.M.TimkoandD.J.Cram,J. Am. Chem. Soc., 1974, 96, 7159– 112 T. Stahlberg, M. G. Sorensen and A. Riisager, Green Chem., 2010, 7160; (d) S. Goswami, S. Dey and S. Jana, Tetrahedron, 2008, 64, 12, 321–325. 6358–6363; (e) F. W. Lichtenthaler, A. Brust and E. Cuny, Green 113 (a)C.Carlini,M.Giuttari,A.M.R.Galletti,G.Sbrana,T.Armaroli Chem., 2001, 3, 201–209. and G. Busca, Appl. Catal., A, 1999, 183, 295–302; (b) T. Armaroli, 148 J. H. Turner, P. A. Rebers, P. L. Barrick and R. H. Cotton, Anal. G. Busca, C. Carlini, M. Giuttari, A. M. R. Galletti and G. Sbrana, Chem., 1954, 26, 898–901. J. Mol. Catal. A: Chem., 2000, 151, 233–243. 149 V. Schiavo, G. Descotes and J. Mentech, Bull. Soc. Chim. Fr., 1991, 114 M. Hara, K. Nakajima and S. Yamashita, JP2009215172-A, 2009. 704–711. 115 C. Carlini, P.Patrono, A. M. R. Galletti and G. Sbrana, Appl. Catal., 150 (a) W. N. Haworth, W. G. M. Jones and L. F. Wiggins, J. Chem. A, 2004, 275, 111–118. Soc., 1945, 1–4; (b) A. C. Cope and W. N. Baxter, J. Am. Chem. 116 T.Werpy and G. Petersen, Top ValueAdded Chemicals from Biomass, Soc., 1955, 77, 393–396. US Dept. of Energy, 2004, vol. 1, pp. 26–28. 151 G. C. A. Luijkx, N. P. M. Huck, F. van Rantwijk, L. Maat and H. 117 T. Elhajj, A. Masroua, J. C. Martin and G. Descotes, Bull. Soc. van Bekkum, Heterocycles, 2009, 77, 1037–1044. Chim. Fr., 1987, 855–860. 152 R. Villard, F. Robert, I. Blank, G. Bernardinelli, T. Soldo and T. 118 B. W. Lew, U. S. Pat. 3326944, Chem. Abstr., 1968, 68, P49434n. Hofmann, J. Agric. Food Chem., 2003, 51, 4040–4045. 119 Y. Y. Gorbanev, S. K. Klitgaard, J. M. Woodley, C. H. Christensen 153 (a) N. Elming and N. Clausonkaas, Acta Chem. Scand., 1956, and A. Riisager, ChemSusChem, 2009, 2, 672–675. 10, 1603–1605; (b) C. Muller, V. Diehl and F. W. Lichtenthaler, 120 O. Casanova, S. Iborra and A. Corma, ChemSusChem, 2009, 2, Tetrahedron, 1998, 54, 10703–10712. 1138–1144. 154 K. Kojiri, H. Kondo, H. Arakawa, M. Ohkubo and H. Suda, 121 S. E. Davis, L. R. Houkb, E. C. Tamargoa, A. K. Datyeb and R. J. US6703373-B1. Davis, Catal. Today, 2010, DOI: 10.1016/j.cattod.2010.1006.1004. 155 S. G. Sun and W. V. Murray, J. Org. Chem., 1999, 64, 5941–5945. 122 M. P. J. vanDeurzen, F. vanRantwijk and R. A. Sheldon, J. 156 T. Fumagalli, G. Sello and F. Orsini, Synth. Commun., 2009, 39, Carbohydr. Chem., 1997, 16, 299–309. 2178–2195. 123 A. Gandini and M. N. Belgacem, Prog. Polym. Sci., 1997, 22, 1203– 157 Z. Mouloungui, M. Delmas and A. Gaset, Synth. Commun., 1984, 1379. 14, 701–705. 124 L. Cottier, G. Descotes, J.Lewkowski, R. Skowronski and E. Viollet, 158 S. N. Goodman and E. N. Jacobsen, Adv. Synth. Catal., 2002, 344, J. Heterocycl. Chem., 1995, 32, 927–930. 953–956. 125 J. W. Van Reijendam, G. J. Heeres and M. J. Janssen, Tetrahedron, 159 N. Almirante, A. Cerri, G. Fedrizzi, G. Marazzi and M. San- 1970, 26, 1291. tagostino, Tetrahedron Lett., 1998, 39, 3287–3290. 126 S. Morikawa, Noguchi Kenkyusho Jiho, 1978, 21, 25; S. Morikawa, 160 (a) C. Z. Yu, B. Liu and L. Q. Hu, J. Org. Chem., 2001, 66, 5413– Chem. Abstr., 1979, 90, 103740d. 5418; (b) C. Z. Yu and L. Q. Hu, J. Org. Chem., 2002, 67, 219– 127 E. L. Clennan and M. E. Mehrsheikh-Mohammadi, J. Am. Chem. 223. Soc., 1984, 106, 7112–7118. 161 L. Cottier, G. Descotes and Y.Soro, J. Carbohydr. Chem., 2005, 24, 128 S. Morikawa, Noguchi Kenkyusho Jiho, 1979, 22, 20; S. Morikawa, 55–71. Chem. Abstr., 1980, 90, 198181a. 162 I. Terada, T. Takeda, T. Kobayashi, T. Hiramoto and K. 129 L. Cottier, G. Descotes, J.Lewkowski and R. Skowronski, Org. Prep. Tsuyoshi, US2010029786-A1 WO2008044784-A1, 2008; C07D- Proced. Int., 1995, 27, 564–566. 307/00 200863, 2008. Downloaded by University of Oxford on 06 April 2011 130 L. Cottier, G. Descotes, J. Lewkowski and R. Skowronski, Pol. J. 163 T. Urashima, K. Suyama and S. Adachi, Carbohydr. Res., 1985, 135, Chem., 1994, 68, 693–698. 324–329.

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D 131 P. J. McDermott and R. A. Stockman, Org. Lett., 2005, 7, 27–29. 164 J. A. Dumesis, G. W. Huber, J. N. Chheda, C. J. Bar- 132 H. Mehdi, A. Bodor, D. Lantos, I. T. Horvath, D. E. De Vos and rett and J. A. Dumesic, WO2007103858-A2; WO2007103858- K. Binnemans, J. Org. Chem., 2007, 72, 517–524. A3; US2008058563-A1; US7671246-B2, WO2007103858-A2, 2007; 133 R. Davies, U. Hedebrant, I. Athanassiadis, P. Rydberg and M. C07C-001/00 200821. Tornqvist, Food Chem. Toxicol., 2009, 47, 1950–1957. 165 R. Skowronski, G. Grabowski, J. Lewkowski, G. Descotes, L. 134 W. Partenheimer and V. V. Grushin, Adv. Synth. Catal., 2001, 343, Cottier and C. Neyret, Org. Prep. Proced. Int., 1993, 25, 353–355. 102–111. 166 H. Quiroz-Florentino, R. Aguilar, B. M. Santoyo, F. Diaz and J. 135 C. Carlini, P. Patrono, A. M. R. Galletti, G. Sbrana and V. Zima, Tamariz, Synthesis, 2008, 1023–1028. Appl. Catal., A, 2005, 289, 197–204. 167 W. Hanefeld, M. Schlitzer, N. Debski and H. Euler, J. Heterocycl. 136 A. S. Amarasekara, D. Green and E. McMillan, Catal. Commun., Chem., 1996, 33, 1143–1146. 2008, 9, 286–288. 168 N. Shinobu, J. Shao, M. Kobayashi, T. Mori, K. Masataka, S. 137 O. C. Navarro, A. C. Canos and S. I. Chornet, Top. Catal., 2009, Noriaki and M. Takao, WO2008056634-A1; CN101535302- 52, 304–314. A; EP2130829-A1; JP2008543069-X; US2010076049-A1, 138 M. A. Lilga, R. T. Hallen, J. Hu, J. F. White and M. J. Gray, WO2008056634-A1, 2008; C07D-405/00 200867. US2008103318-A1, 2007. 169 S. N. Suryawanshi, N. Chandra, P. Kumar, J. Porwal and S. Gupta, 139 M. A. Lilga, R. T. Hallen and M. Gray, Top. Catal., 2010, 53, Eur. J. Med. Chem., 2008, 43, 2473–2478. 1264–1269. 170 R. T. Cummings, J. P. Dizio and G. A. Krafft, Tetrahedron Lett., 140 R. Skowronski, L. Cottier, G. Descotes and J.Lewkowski, Synthesis, 1988, 29, 69–72. 1996, 1291–1292. 171 L. Lin, Q. Shi, A. K. Nyarko, K. F. Bastow, C. C. Wu, C. Y. Su, C. 141 L. Cottier, G. Descotes and J. Lewkowski, Synth. Commun., 1994, C. Y. Shih and K. H. Lee, J. Med. Chem., 2006, 49, 3963–3972. 24, 939–944. 172 (a) A. Muther and B. Tollens, Ber. Dtsch. Chem. Ges., 1904, 37, 306– 142 P. Vinke, W.van der Poel and H. v. Bekkum, Stud. Surf. Sci. Catal., 311; (b) A. Wahhab, J. Am. Chem. Soc., 1948, 70, 3580–3582; (c)W. 1991, 59. Volksen, Arch. Pharm. Ber. Dtsch. Pharm. Ges., 1954, 287, 459–462; 143 E. Taarning, I. S. Nielsen, K. Egeblad, R. Madsen and C. H. (d)H.Kato,Bull. Agric. Chem. Soc. Jpn., 1959, 23, 551–554; (e)K. Christensen, ChemSusChem, 2008, 1, 75–78. Michail, V. Matzi, A. Maier, R. Herwig, J. Greilberger, H. Juan, 144 O. Casanova, S. Iborra and A. Corma, J. Catal., 2009, 265, 109– O. Kunert and R. Wintersteiger, Anal. Bioanal. Chem., 2007, 387, 116. 2801–2814. 145 (a) L. Cottier, G. Descotes, H. Nigay, J. C. Parron and V. Gregoire, 173 E. Erdmann, Ber. Dtsch. Chem. Ges., 1910, 43, 2391–2398. Bull. Soc. Chim. Fr., 1986, 844–850; (b) R. Alibes, J. Font, A. Mula 174 E. Erdmann and C. Schaefer, Ber. Dtsch. Chem. Ges., 1910, 43, and R. M. Ortuno, Synth. Commun., 1990, 20, 2607–2615; (c)L. 2398–2406. Cottier, G. Descotes, L. Eymard and K. Rapp, Synthesis, 1995, 175 T. S. Gardner, F. A. Smith, E. Wenis and J. Lee, J. Org. Chem., 1951, 303–306. 16, 1121–1125.

176 W. F. Cooper and W. H. Nuttall, J. Chem. Soc., 1912, 101, 1074– 200 J. R. de Freitas Filho, R. M. Srivastava, Y. Soro, L. Cottier and G. 1081. Descotes, J. Carbohydr. Chem., 2001, 20, 561–568. 177 G. Kallinich, Arch. Pharm. Ber. Dtsch. Pharm. Ges., 1958, 291/63, 201 D. Schinzer, E. Bourguet and S. Ducki, Chem.Eur.J., 2004, 10, 274–277. 3217–3224. 178 A. S. Amarasekara, O. Edigin and W.Hernandez, Lett. Org. Chem., 202 H. Bredereck, Ber. Dtsch. Chem. Ges. B, 1932, 65, 1833–1838. 2007, 4, 306–308. 203 T. A. Lewis, L. Bayless, J. B. Eckman, J. L. Ellis, G. Grewal, L. 179 F. Liu, A. G. Stephen, R. J. Fisher and T. R. Burke Jr., Bioorg. Med. Libertine, J. Marie Nicolas, R. T. Scannell, B. F. Wels, K. Wenberg Chem. Lett., 2008, 18, 1096–1101. and D. M. Wypij, Bioorg. Med. Chem. Lett., 2004, 14, 2265– 180 K. Undheim, J. Roe and T. Greibrok, Acta Chem. Scand., 1969, 23, 2268. 2501. 204 R. L. Dow, R. C. Kelly, I. Schletter and W. Wierenga, Synth. 181 L. Sattler, F. W. Zerban, G. L. Clark and C. C. Chu, J. Am. Chem. Commun., 1981, 11, 43–53. Soc., 1951, 73, 5908–5910. 205 I. Iovel, K. Mertins, J. Kischel, A. Zapf and M. Beller, Angew. 182 X. Shao, Z. Li, X. Qian and X. Xu, J. Agric. Food Chem., 2009, 57, Chem., Int. Ed., 2005, 44, 3913–3917. 951–957. 206 M. Cunningham and C. Doree, Biochem. J., 1914, 8, 438–447. 183 I. M. Karaguni, K. H. Glusenkamp, A. Langerak, C. Geisen, V. 207 (a) J. J. Bozell, L. Moens, D. C. Elliott, Y. Wang, G. G. Neuen- Ullrich, G. Winde, T. Moroy and O. Muller, Bioorg. Med. Chem. scwander, S. W. Fitzpatrick, R. J. Bilski and J. L. Jarnefeld, Lett., 2002, 12, 709–713. Resour., Conserv. Recycl., 2000, 28, 227–239; (b)D.M.Alonso, 184 J. J. Matasi, J. P. Caldwell, J. Hao, B. Neustadt, L. Arik, C. J. Foster, J. Q. Bond and J. A. Dumesic, Green Chem., 2010, 12, 1493– J. Lachowicz and D. B. Tulshian, Bioorg. Med. Chem. Lett., 2005, 1513. 15, 1333–1336. 208 J. Horvat, B. Klaic, B. Metelko and V. Sunjic, Tetrahedron Lett., 185 R. Miller, Acta Chem. Scand., Ser. B, 1987, 41, 208–209. 1985, 26, 2111–2114. 186 A. Baliani, G. J. Bueno, M. L. Stewart, V. Yardley, R. Brun, M. P. 209 (a) K. D. Baugh and P. L. Mccarty, Biotechnol. Bioeng., 1988, 31, Barrett and I. H. Gilbert, J. Med. Chem., 2005, 48, 5570–5579. 50–61; (b) B. F. M. Kuster and H. S. Vanderbaan, Carbohydr. Res., 187 J. Ramonczai and L. Vargha, J. Am. Chem. Soc., 1950, 72, 2737– 1977, 54, 165–176; (c) K. R. Heimlich and A. N. Martin, J. Am. 2737. Pharm. Assoc., 1960, 49, 592–597; (d) H. P. Teunissen, Recl. Trav. 188 G. Papadogianakis, L. Maat and R. A. Sheldon, J. Chem. Soc., Chim. Pays-Bas, 1930, 49, 784–826; (e) S. Mckibbins, J. F. Harris Chem. Commun., 1994, 2659–2660. and J. F. Saeman, J. Chromatogr., A, 1961, 5, 207. 189 K. Sanda, L. Rigal and A. Gaset, Carbohydr. Res., 1989, 187, 15–23. 210 B. Girisuta, L. P. B. M. Janssen and H. J. Heeres, Green Chem., 190 K. Sanda, L. Rigal, M. Delmas and A. Gaset, Synthesis, 1992, 2006, 8, 701–709. 541–542. 211 G. C. A. Luijkx, F. Vanrantwijk and H. Vanbekkum, Carbohydr. 191 P. Villain-Guillot, M. Gualtieri, L. Bastide, F. Roquet, J. Martinez, Res., 1993, 242, 131–139. M. Amblard, M. Pugniere and J. P. Leonetti, J. Med. Chem., 2007, 212 T. Soldo, I. Blank and T. Hofmann, Chem. Senses, 2003, 28, 371– 50, 4195–4204. 379. 192 F. H. Newth and L. F. Wiggins, J. Chem. Soc., 1947, 396–398. 213 H. Ottinger, T. Soldo and T. Hofmann, J. Agric. Food Chem., 2003, 193 M. K. Jogia, V.Vakamoce and R. T. Weavers, Aust. J. Chem., 1985, 51, 1035–1041. 38, 1009–1016. 214 T. Soldo and T. Hofmann, J. Agric. Food Chem., 2005, 53, 9165– 194 R. Bognar, P. Herczegh, M. Zsely and G. Batta, Carbohydr. Res., 9171. 1987, 164, 465–469. 215 (a) J. Koch, M. Pischetsrieder, K. Polborn and T. Severin, 195 A. C. Cope, 1963. Carbohydr. Res., 1998, 313, 117–123; (b) O. Pachmayr, F. Ledl 196 P. Gupta, S. K. Singh, A. Pathak and B. Kundu, Tetrahedron, 2002, and T. Severin, Z. Lebensm.-Unters. Forsch., 1986, 182, 294– 58, 10469–10474. 297. Downloaded by University of Oxford on 06 April 2011 197 D. Chundury and H. H. Szmant, Ind. Eng. Chem. Prod. Res. Dev., 216 O. Frank, H. Ottinger and T. Hofmann, J. Agric. Food Chem., 2001, 1981, 20, 158–163. 49, 231–238.

Published on 28 February 2011 http://pubs.rsc.org | doi:10.1039/C0GC00401D 198 M. Oikawa, M. Ikoma and M. Sasaki, Tetrahedron Lett., 2005, 46, 217 J. M. Timko and D. J. Cram, J. Am. Chem. Soc., 1974, 96, 7159– 415–418. 7160. 199 T. Elhajj, J. C. Martin and G. Descotes, J. Heterocycl. Chem., 1983, 218 S. T. Waddell, J. M. Eckert and T. A. Blizzard, Heterocycles, 1996, 20, 233–235. 43, 2325–2332.