The synthesis of HMF is nowadays one of the most investigated reactions from biomass. HMF is indeed considered as a chemical platforms from which new generations of biofuels (ex: dimethylfurane) and a wide range of intermediates, monomers and many other fine chemicals can be then produced [29, 30]. This old reaction is now witnessing a sort of renaissance due to the scarcity of oils. HMF is produced through a triple acid-catalyzed dehydration of hexoses. In this reaction the nature of the solvent plays a pivotal role by ensuring (i) the dissolution of carbohy­drates, including biopolymers (ii) the dilution of released water, thus limiting side reactions such as the rehydration of HMF to levulinic and formic acids (Scheme 3.6) and (iii) determine the choice of the work-up procedure. Obviously, the solvent

should be also inert. In the current literature, several solvents have been proposed. Among them, dimethylsulfoxide (DMSO), water, mixtures of water and organic compounds, and ionic liquids have been particularly investigated.

In DMSO, yields of HMF greater than 85 % were obtained but the extraction of HMF from DMSO still remains rather complex and thus expensive. Additionally, under acidic conditions, DMSO may be decomposed leading to the formation of toxic products decreasing the sustainability/attractiveness of the process. In water, yields of HMF are rather low due to the side acid-catalyzed rehydration of HMF to levulinic and formic acids (Scheme 3.6). For this reason, water is often used in combination with organic solvents. This strategy affords higher yields but one should notice that reported yields are still lower than in the presence of DMSO.

In recent years, ILs have received considerable attention for this reaction. Yields of HMF obtained in ILs are comparable to those obtained in DMSO while HMF can be conveniently recovered by liquid-liquid phase extraction using for instance methyliso — butylketone (MIBK), tetrahydrofurane or butanol. Note that the extraction of HMF from ILs can be carried out in a continuous mode, thus allowing side reactions involving HMF to be limited at the same time. It should be noted that same strategy was employed in water (biphasic system) in order to limit the rehydration of HMF. In such case, sodium chloride is generally used in order to facilitate the extraction of HMF (salt-out effect). Unfortunately, the price and toxicity of ILs together with problems linked to their long term recycling are currently hampering their utilization at an industrial scale. Clearly, the industrial emergence of HMF requires chemists to urgently develop innovative processes to produce HMF in a more sustainable way from biomass. In this context, the above-described DESs have received more and more attention for the conversion of hexoses to HMF. In the next section, we present the most recent innovative works reported in this field of chemistry. Although this topic has emerged very recently, we wish to demonstrate here that DESs have the potential to produce HMF in a more rational way than in conventional solvents (Scheme 3.7).

In 2008, B. Han and co-workers have reported that HMF can be produced in acidic ChCl-derived DESs [31]. In particular, in a melt composed of ChCl and citric acid (a cheap and renewable carboxylic acid), authors have shown that fructose can be converted to HMF with more than 76 % yield (at 80 °C, 1 h, ratio DES/fructose = 5). Other DESs made of renewably sourced carboxylic acids such as oxalic and malonic acids have been also successfully used. Owing to the low solubility of HMF in the ChCl/citric acid DES, the process can be performed in a biphasic system using ethyl acetate as an extraction solvent. Like in the case of ILs, continuous extraction of HMF not only facilitated the isolation of HMF but also allowed the selectivity to HMF to be increased. Indeed, under such conditions, HMF was obtained with a yield as high as 91 %. After removal of the ethyl acetate phase containing HMF, the ChCl/citric acid DES was recycled. A slight decrease of the HMF yield was observed upon recycling experiments mainly due to the accumulation of water in the reaction media that affects the selectivity of the process (presumable acid catalyzed rehydra­tion of HMF to levulinic and formic acid). After drying of the used ChCl/citric acid eutectic mixture, the initial yield of HMF was recovered further demonstrating (i) the negative effect of water on the HMF selectivity and (ii) the stability of the ChCl/citric acid DES under reported conditions.

Next, the same group has transposed this work to the tandem hydrolysis/dehy — dration of inulin, a biopolymer of fructose, to HMF in the presence of acidic DESs such as ChCl/citric acid monohydrate or ChCl/oxalic acid dehydrate [35]. They have shown that, at 70 °C, the solubility of inulin was 150 and 28 mg. g-1 in ChCl/ oxalic acid and ChCl/citric acid, respectively. The tandem hydrolysis/dehydration of inulin (81.0 mg, 0.5 mmol fructose units) to HMF was performed at a temper­ature within a range of 50 and 90 °C. At 80 °C, the maximum yield of HMF was 55 % in both DESs within 2 h. Note that the conversion rate of fructose to HMF is higher in the ChCl/oxalic acid DES than in the ChCl/citric acid one due to a difference of acid strength. The authors have also demonstrated that addition of a suitable amount of water as soon as the beginning of the reaction has a positive effect on the formation of HMF. In particular, authors have shown that in ChCl/ oxalic acid and ChCl/citric acid DESs water does not affect the selectivity of the reaction as soon as the water/fructose units molar ratio remained lower than 31. When the water content was increased, secondary reactions such as rehydration of HMF became the dominant reactions. The reaction temperature is of prime impor­tance in this process and closely governs the selectivity of each elementary step. The hydrolysis of inulin to fructose was optimal at 50 °C while 80 °C was found to be necessary to dehydrate fructose to HMF. In this context, the one pot process was carried out in two steps involving (i) hydrolysis of inulin at 50 °C for 2 h and (ii) heating of the solution to 80 °C for another 2 h in order to dehydrate in-situ produced fructose to HMF. Using this procedure, the selectivity to HMF (65 %) was found to be higher. For the same reasons to those described above, the HMF production rate was found to be higher in the ChCl/oxalic acid DES than in the ChCl/citric acid one. To further increase the yield of HMF, authors have attempted the reaction in a biphasic system using acetyl acetate as an extraction solvent. In agreement with previous reports, the yield of HMF was increased from 57 to 64 %

and the extraction was found to be fully selective to HMF. After elimination of the ethyl acetate phase containing HMF, the possible recycling of the DES was investigated. Although water was unavoidably accumulated in the DES phase, at least six runs were performed without appreciable decrease of the HMF yield.

In 2009, Konig and co-workers investigated the production of HMF from a melt composed of D-fructose (40 wt%) and N, N’-dimethyl urea (DMU) heated at 110 °C for 2 h in the presence of CrCl2 and CrCl3 (10 mol% each). Unfortunately, even in biphasic conditions, (using ethyl acetate as an extraction solvent) low yields of HMF were obtained (6 and 2 % with CrCl2 and CrCl3, respectively). Other catalysts such as FeCl3 and AlCl3 gave similar results. Only Amberlyst 15, a sulfonated ion-exchange resin, provided a 27 % yield of HMF in this system. Authors have next investigated the production of HMF using different urea derivatives (urea, DMU and N, N’-tetramethyl urea (TMU)). Reactions were performed in the pres­ence of FeCl3 (10 mol%) and heated at 100 °C for 1 h. Conversely to urea and DMU, TMU-based melt gave HMF with an excellent yield of 89 %. These results tend to show that the presence of — NH — groups on urea and DMU are detrimental for the selectivity of the reaction. Although high yields have been obtained with TMU, the toxicity and problem of separation arising from the use of TMU represent two serious limitations. It is noteworthy that these results are in accordance with a previous work of B. Han and co-workers who have reported that basic DESs (ChCl/ urea) in combination with Lewis acid (ZnCl2, CrCl3) or ChCl/metal chlorides-based DES were poorly efficient in the dehydration of fructose to HMF [32]. Konig et al. have next developed novel carbohydrate-derived melts (based on choline chloride as a hydrogen bond acceptor) with low melting point, low viscosity, low toxicity and high sugar content. The pH values of the different melts are presented in Scheme 3.8. Production of HMF in these melts was tested using a content of fructose of 40 wt% and Amberlyst 15 or FeCl3 as catalysts. Reactions were carried out at 100 °C for 1 h. Among all tested melts, only ChCl/fructose DES has allowed the production of 25 and 40 % yield of HMF in the presence of A15 and FeCl3, respectively. When no ChCl was employed, levulinic acid (20 %) was detected instead of HMF further supporting the key role played by ChCl.

Based on these results, authors have screened the activity/selectivity of several homogeneous, heterogeneous, Bronsted and Lewis acid catalysts in ChCl/carbohy — drate melts. Results are reported in Table 3.4. Except in the case of ZnCl2, it was found that, for all other tested catalysts (Table 3.4), HMF can be obtained with 40-60 % yields from melts composed of ChCl and fructose or inulin. Tested solid catalysts such as Montmorillonite and Amberlyst 15 were also capable of promot­ing the dehydration of fructose to HMF (40-49 % yield). Conversely to the case of fructose, Montmorillonite was found however poorly active from inulin which was ascribed to the low efficiency of Montmorillonite in the catalytic hydrolysis of inulin to fructose, a pre-requisite step prior formation of HMF.

Conversion of glucose to HMF is more challenging and requires first an isom­erization step to fructose before dehydration to HMF. In accordance with a previous work of Zhao et al. performed in imidazolium-based ILs [33], chromium-based catalysts were found the most efficient catalysts in tested DES affording HMF with 31-45 % yield and 43-62 % yield from glucose and sucrose, respectively.

In 2012, K. De Oliveira Vigier et al. reported that betaine hydrochloride (BHC), a co-product of the sugar beet industry, can be used as a renewably sourced Bronsted acid in combination with ChCl and water for the production of HMF from fructose and inulin [34] In a ternary mixture ChCl/BHC/water (10/0.5/2), HMF was produced with 63 % yield (at 130 °C from 40 wt% of fructose). As observed by B. Han and co-workers, when the reaction was performed in a biphasic system using methylisobutylketone (MIBK) as an extraction solvent, HMF was recovered with a purity higher than 95 % (isolated yield of 84 % from 10 wt% of fructose) further demonstrating that these systems similarly behave to the

traditional imidazolium-derived ILs (Scheme 3.9). After recovery of the MIBK phase containing HMF, the ChCl/BHC/water system was successfully recycled seven times. In same work, authors have shown that the dehydration of fructose to HMF can also conveniently take place in a BHC/glycerol (1:1) mixture, a DES exclusively made of renewably-sourced chemicals. Although 51 % yield of HMF was successfully obtained at 110 °C (from 10 wt% of fructose), extraction of HMF from the BHC/glycerol medium remained very difficult due to the very high solubility of HMF in this mixture.

Later, F. Liu et al. have shown that, under exposure to CO2, fructose and inulin can be converted to HMF in a ChCl/fructose DES [35]. The pKa of carbonic acid is low enough to catalyze the dehydration of fructose to HMF as previously shown by B. Han [36]. In this work, CO2 initially reacted with water contained in fructose and in the DES resulting in the formation of carbonic acid in a sufficient amount to initiate the dehydration of fructose to HMF. After 90 min of reaction at 120 ° C under 4 MPa of CO2, a yield of 74 % of HMF was obtained from 20 wt% of fructose. Due to the high selectivity of the reaction and the insolubility of ChCl in MIBK, it was also possible to selectively extract HMF which was recovered with a purity of 98 %. Remarkably, the present system was tolerant to very high loading of fructose whereas most of reported solvents suffer from a low selectivity to HMF

when the fructose loading was higher than 20 wt%. For instance, fructose with a loading of 100 wt% was successfully dehydrated to HMF in a ChCl/CO2 system without appreciable decrease of the HMF yield (66 %). The authors have ascribed the tolerance of such system to high loading of fructose to strong interaction between produced HMF and ChCl resulting in the stabilization of HMF in the reaction media. As shown in Scheme 3.10, when neat HMF and ChCl were mixed together a melt was readily obtained at a fructose content higher than 60 wt%. Such melt might be responsible for the surprising stability of HMF in such system when using high loading of fructose. It is indeed known that when a chemical is engaged in the formation of a DES its reactivity is drastically reduced.