Concerted attempts to dissolve cellulose do not have such a long history. The first reported study on cellulose dissolution using an ionic material was reported in 1934 by Graenacher and co-workers [5]. They used a mixture of amine and a pyridinium salt to dissolve cellulose. At this stage, pyridinium salt was not used as an IL but as an added salt. Thus, the “first” study of cellulose dissolution with ILs that was reported was by Swatloski et al. in 2002 [6]. They reported that 1-butyl-3-methylimidazolium chloride ([C4mim]Cl) dissolved pulp cellulose. This IL dissolved 3 % cellulose at 70 °C, and 10 % cellulose at 100 °C. They also clarified that the cellulose dissolving degree was improved with a combination of IL soaking and

other physicochemical treatments such as sonication or microwave irradiation. On the other hand, tetrafluoroborate-type salts and hexafluorophosphate-type salts did not dissolve cellulose unlike [C4mim]Cl. The data shown in Table 2.1 provides that the properties of ILs that deeply affect the solubility of cellulose in the corresponding ILs.

[C4mim]Cl has a high melting point (Tm) of 73 ° C and high viscosity, thus it is hard to use as a solvent at ambient condition [7, 8]. Heinze and co-workers reported that some chloride salts which have pyridinium or ammonium cations also dissolve cellulose (Scheme 2.1, Table 2.2) [8]. 3-Methyl-n-butylpyridinium chloride ([C4mpy]Cl) dissolved cellulose much better than [C4mim]Cl, and benzyldimethyl(tetradecyl)ammonium chloride (BDTAC) has a lower Tm (52 °C). Sometimes, the degree of polymerization of cellulose (DP) decreases after disso­lution in ILs (see Table 2.2). In the case of energy conversion, changes in the DP are a less important factor. Some studies also require dissolving cellulose without lowering DP from the viewpoint of cellulose application. In both cases, the disso­lution of cellulose under mild conditions is suitable considering efficient processing. Then there are some studies on reducing the Tm of these chloride salts. Physical chemistry tells us that small anions such as chloride anion interact strongly with cations due to higher charge density resulting high Tm of the salts. It is therefore important to design cations to lower the Tm. Mizumo et al. developed liquid state chloride salts using imidazolium cations having allyl group(s) [9]. Allyl group is effective to show conformational change or rotation of the group, that induces to lower the Tm of the imidazolium salt. They clarified that the adopting allyl groups into imidazoliulm cation is a valid way to lower the Tm of the chloride salts. After this report, Zhang and co-workers reported that one of room temperature ILs, 1-allyl-3-methylimidazolium chloride ([Amim]Cl), has a good ability to dis­solve cellulose [10]. [Amim]Cl dissolved no cellulose at room temperature, but it dissolved cellulose at 60 °C under stirring. With increasing temperature, cellulose could be dissolved easily in [Amim]Cl.

With an increase in the variety of chloride type cellulose-dissolving ILs (CDILs), cellulose dissolving mechanisms as well as dominant properties of the ILs for cellulose dissolution have been gradually clarified. Those studies suggest that the chloride anion works dominantly to dissolve cellulose by breaking the hydrogen bonding networks of cellulose fibrils. Remsing and co-workers have

2- Methyl-N-butylpyridinium chloride Benzyldimethyl(tetradecyl)ammonium chloride Scheme 2.1 Structure of chloride type ILs to dissolve cellulose under heating [8]

clarified that [C4mim]Cl makes hydrogen bonding between the carbohydrate hydroxyl protons and the chloride ions in a 1:1 stoichiometric ratio using 13C and 35/37Cl NMR relaxation measurements [11]. The relaxation time of the imidazolium cation and chloride anion in [C4mim]Cl was analyzed as a function of concentra­tion (wt%) of cellobiose as a model compound of cellulose. The relaxation time of the cation was almost constant regardless of cellobiose concentration. This means that there are no specific interaction between cations and cellobiose. On the other hand, there was a clear relationship between the 35/37Cl relaxation time and cello- biose concentration. This suggests that the chloride anion interacts strongly with the dissolved carbohydrate. They analyzed the interaction between the chloride anions and non-derivatized carbohydrates. This study clarified that the chloride ions interact in a 1:1 ratio with the carbohydrate hydroxyl protons.

Some simulation studies have also been reported on carbohydrate dissolution in dialkylimidazolium chloride-type ILs. Youngs and co-workers reported about the molecular dynamics simulations of glucose solvation by 1,3-dimethylimidazolium chloride [C1mim]Cl [12]. They found that the primary solvation shell around the glucose consists predominantly of chloride anions hydrogen bonding with the hydroxyl groups of glucose ring. This is the predominant interaction between glucose rings and chloride-type ILs. There is a small contribution of cations on the carbohydrate-IL interaction. Cations were however also found near the glucose, and a hydrogen at the 2-position of the imidazolium ring interacted with an oxygen atom of the secondary hydroxyl group of the glucose. A weak contribution of van der Waals force was also seen between the glucose and the cations. Even at high
glucose concentrations (16.7 wt%), the anion-cation interactions and overall liquid structure of [C1mim]Cl were found not to be significantly changed. This means that the glucose is readily solubilized by the IL even under high concentration. Gross and co-workers reported on the thermodynamics of cellulose solvation in [C4mim] Cl [13]. All-atom molecular dynamics (MD) simulations were conducted to analyze the thermodynamic driving force of the cellulose dissolving process and to clarify the role of both anions and cations in the process. They suggested that the disso­ciated cellulose has higher potential energy in water than that in [C4mim]Cl. They suggested that the cellulose insolubility in water is mostly derived from the entropy reduction of the solvent. In addition, they also suggested that both the anion and cation of the IL interact with the glucan residues. In the case of Cl_ anions, they form hydrogen bonds with the hydroxyl groups of cellulose from either equatorial or axial directions. On the other hand, for the cations, the contact with cellulose along the axial directions was closer than that along the equatorial directions. They concluded that interacting with cellulose along axial directions and disrupting the cellulose fibrils is an important step of cellulose dissolution.

The systematic preparation of ILs and low-melting organic salts is a viable approach for obtaining tailored task-specific reaction media but it is only applicable within certain limitations. As an example, the viscosities of cellulose dissolving ILs are very high, compared to molecular solvents but also other ‘low-viscous’ ILs. It is not likely that a reduction of two or three orders of magnitudes, which seems to be required for certain derivatization reactions, can be achieved by modification of the molecular structure of ILs. One main reason for the high viscosity is the strong coulomb interaction between anions and cations, which is an ‘intrinsic feature’ of ILs and salts in general. Another factor is the formation of internal hydrogen bonds. Reduction in viscosity can be achieved by decreasing the strength of hydrogen

bonding between the two ionic species. Reducing the hydrogen bond donor ability (в) of an IL, which is mainly determined by the nature of the anion, can result in a significant decrease in viscosity (Table 5.4). Imidazolium salts with more hydro­phobic cations, e. g., dicyanamide or bis(trifluoromethylsulfonyl)imide, are less viscous compared to the corresponding chlorides, carboxylates, and phosphates that are rather strong hydrogen bond acceptors. However, these low viscous ILs also lack the ability to dissolve cellulose. It has been demonstrated that cellulose is only soluble in ILs with rather high в values above >0.8, i. e., those that exhibit strong hydrogen bonding between anions and cations and consequently possess higher viscosities.

The addition of co-solvents is an efficient way to diminish ‘intrinsic limita­tions’ of ILs that are predetermined in a narrow frame by the molecular structure, such as viscosities, polarities, densities, and melting points [128]. It has been demonstrated that the viscosity of cellulose/IL solutions can be reduced drasti­cally by addition of DMSO or DMF (Fig. 5.9) [33]. The decrease proceeds nearly exponentially with the weight fraction of dipolar aprotic co-solvent in the mixtures; at a typical mixing ration of 1 g co-solvent per g cellulose/IL solution, the viscosity is two order of magnitudes lower compared to an undiluted cellu­lose/IL solution of the same polymer content. Pyridine has been utilized as well to reduce the viscosity of cellulose/IL solutions [69]. For dilute and concentrated cellulose solutions in BMIMCl and AMIMCl it was demonstrated that the conformation of the dissolved polysaccharide is not impaired upon the addition of DMSO as a co-solvent [68]. The viscosity of the IL/DMSO mixtures as a function of the mole fraction of DMSO (xDMSO) could be predicted precisely by the following simple equation; with k being a constant (0.12 for BMIMCl and

0. 15 for AMIMCl):

П xDMSO

ln = ————- (5.3)

Піь k

Co-solvents can also be applied to tailor hydrophobicity of ILs along with their hydrogen bond donor/acceptor properties [129—131]. Thus, miscibility of hydro­philic cellulose/IL solutions with hydrophobic derivatization reagents could be improved by adding nonpolar co-solvents such as chloroform [41].

The addition of a solvent to cellulose/IL solutions can result in complete miscibility but also phase separation might occur. Moreover, it might induce precipitation of cellulose. Semi-quantitative predictions on these phenomena can be gained based on the solvatochromic parameters of potential co-solvents (Table 5.5) [132]. Only compounds with a high normalized empirical polarity (Etn > 0.3) were found to be miscible with the relatively hydrophilic ILs used as cellulose solvents. Precipitation of cellulose is induced by solvents with a strong hydrogen bond donor acidity (a < 0.5). In contrast, compounds that are strong hydrogen bond donor (в > 0.4) are very efficient co-solvents for cellulose/IL solutions; the amount of co-solvent that is tolerated before precipitation increases with increasing в-value. Solvatochromic measurements have been employed to characterize IL/co-solvent mixtures of different compositions that were employed for direct dissolution of cellulose [133, 134]. It could be demonstrated that the molar fraction of ILs in the mixtures can be reduced drastically (<0.3) if suitable dipolar aprotic co-solvents such as DMSO and amides are applied. In these mix­tures, the high в-value of the IL, which seems to be important for the dissolution of cellulose, is almost unaffected even in case of an excess of co-solvent [133].

aSolvent parameters: ETN: normalized solvent polarity, a: hydrogen bond donor ability, p: hydro­gen bond acceptor ability

bAMIMCl: 1-allyl-3-methylimidazolium chloride, BMIMCl: 1-butyl-3-methylimidazolium chlo­ride, EMIMAc: 1-ethyl-3-methylimidazolium acetate

c+: miscible, —: immiscible, +/p transition from miscible to precipitation, +/g transition from miscible to gelation, +/— transition from miscible to immiscible

dAmount of co-solvent (g per g cellulose/IL solutions) that can be added without permanent

precipitation/gelation/immiscibility

eWater, methanol, ethanol, and 2-propanol

fEthyl acetate, tetrahydrofuran, dioxane, diethyl ether, toluene, and hexane

In contrast, alcohols that are used for precipitation significantly decreased the hydrogen bond acceptor basicity. For IL/water mixtures it has been demonstrated that cellulose dissolves if the ‘net basicity’ (0.35 < в — а < 0.90) of the solvent falls into a specific range [134].

IL/co-solvent mixtures can be exploited for the homogeneous preparation of highly engineered cellulose derivatives. As described above, CS and TOSC can be prepared at low temperatures (<25 °C) in a completely homogeneous reaction in ILs but utilization of co-solvents is indispensable due to the high viscosity of the undiluted cellulose solutions [33, 35]. Homogeneous preparation of hydrophobic cellulose silyl ethers was realized in IL/chloroform mixtures [41]. Another example for the beneficial use of co-solvents is the hydroxyalkylation of cellulose in EMIMAc with gaseous oxiranes is a heterogeneous process that yields higher DS values if DMSO is added to decrease viscosity and increase solubility of the etherification reagents [37]. In fact, many of the derivatization reactions reported in literature have been carried out not in pure ILs but with the aid of co-solvents (see Tables 5.1 and 5.2) although it has not been mentioned explicitly in each case. Several patented procedures that are of interest for the preparation of commercially attractive cellulose derivatives were performed in IL/co-solvent mixtures [38, 54, 135—137].

In addition to physical and dissolution properties, co-solvents can also alter the chemical behavior of IL based reaction media by either facilitating or preventing specific derivatization reactions. As an alternative to commonly applied imidazolium chlorides, 1-allyl-3-methylimidazolium fluoride has been prepared as a novel solvent for cellulose [138]. Although the IL could be used for dissolution and acetylation of cellulose, the DS values of the products obtained were low because the fluoride anion induced cleavage of the ester bond. In the presence of DMSO, however, highly substituted cellulose esters could be obtained, presumably because of strong solvation of the anion by hydrogen bonding with the co-solvent. Pyridine, on the other hand, is an attractive co-solvent for derivatization reactions that require bases, e. g., tosylation, and tritylation, because it reduces viscosity of cellulose/IL solutions and likewise promotes the chemical conversion [35, 40]. Thus, no additional bases are required because the co-solvent is applied in a slight excess. An interesting approach is to exploit the co-solvent as derivatization reagent. A series of cationic cellulose esters has been prepared in mixtures of BMIMCl with different lactames, e. g., N-methyl-2-pyrolidine and e-caprolactam, as co-solvents [135, 139]. The cyclic amides were activated with tosyl chloride and subsequently reacted with the polysaccharide backbone under ring opining.

Combination of the above described aspects, i. e., synthesis of novel ILs or low melting salts and the use of co-solvents to improve dissolution and processing of cellulose, is a logical consequence with huge potential for a broad range of applications. Interestingly this approach has already been described in the early 1930s in a number of patents [140, 141]. N-Alkylpyridinium salts have been utilized therein as cellulose solvents and pyridine was utilized to decrease their melting points of 120-130 °C and achieve dissolution at ambient temperature. Mixtures of these pyridinium salts and co-solvents have also been applied for shaping and chemical derivatization of cellulose [142, 143]. Although, these pub­lications rose little interest during their time, they were rediscovered within the frame of the increasing interest for novel polysaccharide solvents. These reports are frequently considered as the first attempts to use ILs for dissolution of cellulose which is strictly speaking incorrect regarding the fact that the high melting points of the salts applied lie above 100 °C.

It is self-evident that co-solvents represent additional components that need to be considered in the recycling process. Volatile co-solvents, e. g., pyridine and chlo­roform, can be recovered by evaporation. Subsequently, the crude IL can be purified using suitable techniques. In contrast, co-solvents with relatively low vapor pressures, in particular dipolar aprotic ones like DMF, DMSO, and DMI, will remain in the crude IL after evaporating the non-solvent used for precipitation of the cellulose derivative. After removal of impurities that are harmful for disso­lution or chemical modification of cellulose, the polysaccharide can be dissolved directly in the recycled IL/co-solvent mixture.

5.3 Conclusion

Compared to other cellulose solvents that are useful for cellulose derivatization yet restricted to small lab-scale synthesis, ILs bear huge potential for the homogeneous preparation of highly engineered cellulose derivatives on a commercially attractive scale. They can rapidly dissolve large amounts of cellulose and can be utilized as reaction media for numerous derivatization reactions. Nevertheless, it has been demonstrated in this chapter that several very specific issues that are related to the unique physical and chemical properties of ILs need to be considered. Regarding the limitations of currently applied ILs, it can be expected that a ‘next generation’ of task-specific ILs and related reaction media will be advanced in the near future, e. g., by including alternative types of cations and anions as well as IL/co-solvent systems. Another open question is IL recycling. In this context, the thermal behavior and chemical reactivity of ILs during dissolution and chemical derivati — zation of cellulose must be considered even more. If the specific side reaction of ILs are recognized and fully understood, they can be avoided or even exploited for preparing novel types of cellulose derivatives. As concluding remark it should be pointed out that research on the use of ILs for processing of cellulose will benefit a lot from interdisciplinary contributions from areas such as general organic, phys­ical, and theoretical chemistry, chemical — and process engineering, biochemistry, toxicology, material testing, and other related fields.

9.2.1 Starting Material

9.2.1.1 Fructose

Fructose is the most common studied substrate for the preparation of 5-HMF in aqueous solutions, organic solvents, and water-organic solvent mixtures. In an early work that applied molten salts to the synthesis of 5-HMF from carbohydrates, fructose was converted to 5-HMF with a high yield of 70 % in the presence of pyridinium chloride in 1983 [27]. However, this pioneering work did not stimulate wide investigations on the 5-HMF production from biomass in melt salt solutions, until the beginning of the twenty-first century. Biomass conversions in ionic liquids became a hot topic when Lansalot-Matras and Moreau reported the acid-catalyzed dehydration of fructose in 1-butyl-3-methylimidazolium tetrafluoroborate ([BMIM][BF4]) and 1-butyl-3-methylimidazolium hexafluorophosphate ([BMIM] [PF6]) with dimethyl sulfoxide (DMSO) as co-solvent in the presence of Amberlyst® 15 sulfonic ion-exchange resin as catalyst, to obtain a 5-HMF yield as high as 87 % at 80 °C for 32 h reaction time [28]. It was demonstrated that the addition of ionic liquids had a positive effect on the fructose dehydration to 5-HMF. They also studied acid-catalyzed dehydration of fructose in ionic liquid 1-H-3- methylimidazolium chloride ([HMIM][Cl]), which acted both as solvent and cata­lyst, and demonstrated a 5-HMF yield as high as 92 % [29]. According to the activation parameters calculated from the Arrhenius plots for formation and decom­position of 5-HMF, activation energies are found to be similar to those obtained in reactions catalyzed by zeolite solid catalyst. The authors attribute the high yields observed for the formation of 5-HMF in ionic liquids as solvent to be due to the differences in the preexponential factors [28]. Thereafter, many papers on the selective dehydration of fructose in ionic liquids with various catalysts, have appeared in the literature. Efficient dehydration of fructose to 5-HMF in ionic liquids was reported at much lower temperatures than in aqueous, organic solvents, and water-organic mixture systems [20, 30—34].

Qi et al. reported that 1-butyl-3-methylimidazolium chloride, [BMIM][Cl], used with a sulfonic ion-exchange resin catalyst could efficiently dehydrate fructose into 5-HMF to give a fructose conversion of 98.6 % and a 5-HMF yield of 83.3 % in 10 min reaction time at 80 °C. The reaction time could be reduced to 1 min when the temperature was increased to 120 °C which resulted in a 5-HMF yield of 82.2 % and nearly 100 % fructose conversion [20]. Comparison of the [BMIM][Cl] and [BMIM][BF4] systems that has the same sulfonic ion-exchange resin (Amberlyst®

15) as the catalyst, indicates that the higher efficiency and selectivity observed in the [BMIM][Cl] system can be attributed to a higher tendency towards concerted catalysis due to the greater hydrogen-bonding character, nucleophilicity or basicity of the chloride ion.

Shi et al. [35] used trifluoromethanesulfonic acid (TfOH), which is an interesting catalyst, to promote the conversion of fructose into 5-HMF in imidazolium ionic liquids. They studied the reaction system for different alkyl chain length ionic liquids and used various kinds of anions. In that work, yields of 5-HMF were strongly affected by aggregation of cations and the hydrogen bonds between fructose and anions of ionic liquids. Imidazolium cationic ILs with alkyl chain lengths of the cations being shorter than four carbons were found to be suitable for 5-HMF formation. They found that the anion of an IL forms strong hydrogen bonds with fructose molecules, and thus, the acid radical leads to high reaction activity. These results not only provide evidence to explain the interaction of the structure at the molecular level in 5-HMF preparation in ionic liquids, but also provide some hints on choosing suitable ILs for 5-HMF preparation [35].

Imidazolium-based ILs provide efficient dehydration of fructose into 5-HMF, however, eutectic mixtures of choline chloride with acids have been identified as a promising catalytic system for the process. Hu et al. [36] investigated the conver­sion of fructose to 5-HMF in choline chloride/citric acid at 80 °C, and obtained a 5-HMF yield of 77.8 % without in situ extraction and a yield of 91.4 % when continuous extraction with ethyl acetate was used for a 1 h reaction time. The main advantage of this process was not only high 5-HMF yields obtained but also the chemical system components (fructose, choline chloride and citric acid) used that all originate from renewable sources. Ilgen et al. studied choline chloride for the purpose of using highly concentrated mixtures of fructose (up to 50 wt%). The resulting solutions had a melting region of 79-82 °C, which is much lower than the melting point of pure choline chloride (300 °C), thus processing the solutions at low temperatures was possible. The highest 5-HMF yield obtained from the choline chloride-fructose system was 67 % for which the reaction conditions were p-TsOH as catalyst, reaction temperature of 100 °C and reaction time of 0.5 h [37]. Further­more, they made a screening study to compare the environmental impact of the choline chloride system with different conventional solvents for the conversion of carbohydrates into 5-HMF, and indicated advantages of the choline chloride sys­tems in terms of low toxicity and reduced mobility [37]. Liu et al. [38] developed a cheap and sustainable choline chloride/CO2 system for the dehydration of highly concentrated fructose solutions into 5-HMF with a yield of up to 72 %. In addition to the environmental benefits of this strategy, they found that in the presence of ChCl, 5-HMF is stabilized, probably due to hydrogen bonding that allows forma­tion of a eutectic mixture between choline chloride and 5-HMF. This aspect allows fructose can be converted with a high content (up to 100 wt%) as compared to traditional procedures where 5-HMF is obtained in yields higher than 60 % only from a fructose concentration lower than 20 wt% [20].

Recently, P. Abbott and co-workers have introduced the concept of Deep Eutectic Solvents (DES). A DES is a fluid generally composed of two or three cheap and safe components which are capable of associating each others, often through hydrogen bond interactions, to form a eutectic mixture. The resulting DES is characterized by a melting point lower than that of each individual component. Generally, DESs are characterized by a very large depression of freezing point and are liquid at temper­atures lower than 150 °C [22]. Note that most of them are liquid between room temperature and 70 °C. In most cases, a DES is obtained by mixing a quaternary

t і

I Safe hydrogen bond-donors

I (renewable carboxylic acids, urea, polyols

etc…)

Scheme 3.5 Deep Eutectic Solvents ammonium salt with metal salts or a hydrogen bond donor (HBD) that has the ability to complex the halide anion of the quaternary ammonium salt.

Owing to its low cost, biodegradabity and low toxicity, choline chloride (ChCl) and more recently glycine betaine (zwitterionic or protonic form) have been recently proposed as an organic salt to produce eutectic mixtures generally in combination with cheap and safe HBDs such as urea, renewable carboxylic acids (e. g. oxalic, citric, succinic or amino acids) or renewable polyols (e. g. glycerol, carbohydrates). As compared to the traditional ionic liquids (ILs), DESs derived from ChCl gather many advantages such as (1) low price, (2) 100 % atom-economy synthesis (no purification is required), and (3) most of them are biodegradable [23], biocompatible [24] and non-toxic [25] reinforcing the sustainability of these media. Physicochemical properties of DESs (density, viscosity, refractive index, conduc­tivity, surface tension, etc.) are very close to those of common ILs. Thereby, they have the potential to advantageously replace ILs in many applications such as metal and oxides dissolution, catalysis, electrochemistry and material preparations. Addi­tionally, through hydrogen bond interaction, DESs have the unique ability to stabilize and thus to lower the reactivity of water, opening the route to chemical transformations that are normally not feasible in hygroscopic solvents [26] (Scheme 3.5).

In 2007, Abbott and co-workers defined DESs using the general formula R1R2R3R4N+,X^Y^ [27]:

Type I DES Y = MClx, M = Zn, Sn, Fe, Al, Ga Type II DES Y = MClx. yH2O, M = Cr, Co, Cu, Ni, Fe Type III DES Y = R5 Z with Z = — CONH2, — COOH, — OH

Note that the same group also defined a fourth type of DES which is composed of metal chlorides (e. g. ZnCl2) mixed with different HBDs such as urea, ethylene glycol, acetamide or hexanediol (type IV DES).

Similar to the case of ChCl-derived bio-inspired ILs, the use of DES for dissolution of cellulose has been scarcely reported mostly due to the novelty of these systems. Although based on the state of the art, protic groups such as — OH or -COOH are clearly not favorable for the dissolution of cellulose, their involvement in the formation of a DES drastically reduces their protic nature (the — OH group being involved in hydrogen bond interaction) thus offering a better chance to achieve the dissolution of cellulose.

In 2012, Georgia Tech Research Corporation has patented the dissolution of microcrystalline cellulose (AVICEL) in various DESs made of ChCl and betaine monohydrate [28]. Neat DES made of ChCl and urea (or malonic acid or formamide) were not able to dissolve microcrystalline cellulose. When DESs were diluted with a basic solution (NaOH or NaOAc) together with a prolonged incubation time, a swelling of cellulose was however observed. By means of XRD analyses, a decrease of the crystallinity index of cellulose of 15-20 % was noticed suggesting that these systems can partly interact with the hydrogen bond network of cellulose. In agree­ment with previous results, regenerated cellulose was less recalcitrant to hydrolysis after pretreatment in basified ChCl-derived DES. Although neat DESs do not dissolve cellulose, one should however mention that their combination with a basic solution allowed avoiding the large amount of base traditionally required for the dissolution of cellulose. Next, authors highlighted the possible formation of DES from betaine monohydrate and urea. As compared to ChCl, betaine monohydrate is more attractive due to its lower cost and its direct availability from biomass (co-product of the sugar beet industry). DESs are made with more difficulty from betaine than from ChCl and only urea was found eligible as a hydrogen bond donor in such case. Such betaine derived DESs is however highly viscous. Following the same strategy than that used from ChCl authors found that a pretreatment of cellulose in the betaine/urea DES led to a decrease of 15 % in the crystallinity index of cellulose. Although these systems have allowed the crystallinity index of cellulose to be slightly decreased, DESs are however not capable of dissolving cellulose presumably because DES components are already involved in hydrogen bond interaction making difficult their interaction with the hydrogen bond network of cellulose. Additionally, removal of DESs from regenerated cellulose is not an easy task and extensive washing are required.

In the same year, M. Francisco and co-workers investigated a series of 26 dif­ferent DESs in the dissolution of cellulose, lignin and starch [14]. Since selected mixtures exhibited no melting point by differential scanning calorimetry (only glass transition), such mixtures were more considered as low transition temperature mixing (LTTM) rather than real DESs. All solubility measurements were deter­mined by using the cloud point method within a range of temperature of 60 and 100 °C. This method consists in the progressive addition of a biopolymer in LTTMs up to the observation of a turbid solution. Among them, LTTMS made of lactic acid-ChCl were found particularly efficient for lignin dissolution. A clear solubility enhancement of lignin was even observed with an increase of the lactic acid content. Reversely, LTTMs made of malic acid were found more efficient for the dissolution of starch than for the dissolution of lignin. Among tested melts, malic acid-proline melt efficiently dissolved starch and dissolution ability can be

improved by increasing of the proline ratio. Note that no clear rationalization was proposed and the search of LTTMs for the dissolution of lignin or starch still remains empirical.

In agreement with the above-described work patented by Georgia Tech Research Corporation, no significant dissolution of cellulose was observed in all tested LTTMs. Nevertheless, in the typical case of LTTMs derived from proline, turbid solution was observed with cellulose and no evidence of solid particle was further detected further supporting the superior ability of aminoacid for interacting with cellulose as described above.

Having all these results in hand, authors next checked the ability of LTTMs for the delignification of wheat straw. Using a Lactic acid-ChCl melt (2/1), 2 wt% of lignin was extracted after incubation overnight at 60 °C. Although no solubility data was provided, authors claimed that the solubility of wheat straw can be improved using a malic acid/proline melt (1/3) which is consistent with results presented in Table 3.3.

For molecular weight determination the fractionated samples were derivatized with benzoyl chloride in [amim]Cl solution following the procedure introduced by Zoia et al. [59] and analyzed using HP G1312A pump connected to Waters HR5E and HR1 columns with a Waters 484 UV-absorbance detector calibrated using polysty­rene standards. Acid insoluble (Klason) lignin and acid soluble lignin were deter­mined by method modeled from one published by Dence [75]. The acid that was used to hydrolyze the samples was diluted from conc. sulfuric acid corresponding to 72 ± 0.1 %. The lignocellulose samples were dried in a vacuum oven at 40 °C overnight. ca. 100 mg of the samples were measured accurately and mixed with sulfuric acid solution (100 mg per 2 mL) using magnetic stirring and vortex mixer. After 2 h hydrolysis at room temperature with occasional manual mixing the samples were diluted with 50 mL of deionized water and transferred into sealable bottles. The bottles were placed into a commercial pressure cooker and heated at elevated pressure for 90 min. The solid residues were filtered with a grade-3 sinter and washed with 40 mL of water. The filtrate was retained for acid soluble lignin determination. The solid residue was further washed with 60 mL of water, so that filtrate was neutral, and after air-drying the sample was placed into a vacuum oven for 20 h. Acid soluble lignin was determined spectrophotometrically from the retained filtrates. The filtrates were first diluted to precisely 100 mL and then the absorbance was measured at 205 nm wavelength in a 1 cm pathlength cuvette. The concentrations were calculated using extinction co-efficient of 110 L/g. cm.

FT-IR spectra were recorded from finely powdered samples that were dried for 20 h at 50 °C in a vacuum oven, using Perkin-Elmer Spectrum One AT-IR spectrometer. Processing was carried out using PE Spectrum One software. The spectra were processed with baseline correction, noise elimination and normalization.

In most of the reported studies for the production of 5-HMF from biomass in ionic liquids, the 5-HMF was obtained in solution, and the yield of 5-HMF was deter­mined with HPLC or GC without isolation. However, it is of great importance to develop an efficient isolation method for the synthesis of 5-HMF. Wei et al. [102] developed a novel entrainer-intensified vacuum reactive distillation process for the separation of 5-HMF from ionic liquids 1-octyl-3-methylimidazolium chloride ([OMIM][Cl]) that involves heating reaction mixtures under vacuum (ca.300 Pa) to 150-180 °C under a flow of an entrainer (nitrogen), and 95 % of the 5-HMF was recovered from the reaction mixture after 10 min at 180 °C. However, distillation is not a favorable method for the separation of 5-HMF from ionic liquids system, not only because both 5-HMF and ionic liquids have high boiling point, but also because 5-HMF is very reactive and tends to decompose at high temperatures.

Monosaccharides such as fructose and glucose have polarities similar to those of ionic liquids, while 5-HMF has different properties from ionic liquids, which allows separation of 5-HMF from ionic liquids by liquid-liquid extraction. However, an efficient separation solvent is needed to reduce the amount of solvent (low solvent/ feed ratio) required for extraction of 5-HMF from ionic liquid solutions. An ideal extraction solvent should be immiscible with ionic liquids and have a large affinity for 5-HMF. Further, it should be separable from the product, with a method that does not require a large amount of energy. Therefore, it should have a relatively low boiling point to avoid thermal decomposition or polymerization of the 5-HMF and to reduce the energy costs. Many organic solvents such as MIBK (methyl isobutyl ketone) [53], ethyl acetate [20, 63], diethyl ether [29, 45], THF [84], and acetone [60] have been reported to be efficient extraction solvents. Qi et al. [20] tested the separation of 5-HMF from [BMIM][Cl] after reaction by extracting five times with 8 ml of ethyl acetate after 0.5 g of water was added to reduce the viscosity of the ionic liquid. [BMIM][Cl] and fructose were found to be insoluble in ethyl acetate and 5-HMF was the sole product in ethyl acetate phase, and above 95 % of 5-HMF was separated from the reaction mixture after five times extraction. Hu et al. [36] presented 5-HMF synthesis from fructose in a biphasic system of ethyl acetate and choline chloride (ChoCl) with citric acid, and the continuous in-situ extraction of 5-HMF with ethyl acetate led to a 15 % enhancement in the 5-HMF yield in comparison with a neat ionic liquid. For larger-scale production, more efforts have to be made on efficient separation techniques.

There are a large number of ILs that can be produced theoretically, while the synthesized and reported ones are very limited [25]. By the end of 2009, more than 1,800 available ionic liquids which were composed of 714 different cations and 189 different anions have been reported in the book named ionic liquid: physico­chemical properties have reported [26]. According to our current ionic liquids

Tetraakylammonium

1-phenylethanoyl-3-

methylimidazolium

N-propane

sulfuricpyridiniumdihydro 1-methyl-4-(2-azidoethyl)-1, 2, 4-triazolium

database, there are more than 2,300 kinds of ILs, including varieties of 229 anions, 907 cations. In these ILs, the most common used cations are imidazolium, pyridinium, piperidinium, tetraalkylphosphonium, tetraalkylammonium, trialkylsulfonium and metal based, functional cations, the cations are listed in Table 1.1.

However, the generally used ions are either inorganic or organic, for example, hexafluorophosphate, bis (trifluoromethylsulfonyl) imide, tetrafluoroborate, trifluoro- methanesulfonate, dicyanamide, halide, formate, acetate, and alkyl-phosphate and so on (Table 1.2).

F— в—— F

F

OMe

10 Tyrosine

11 Valine

12b Tetrakis [3, 5-bis (trifluoromethyl)

phenyl] borate

13c bis(trifluoromethanesulfonyl)amide

14 Trifluoroacetate

15d AlCl4- a the methyl also could be changed into ethyl et al.

b the other borate and borane anions are fluoroacetoxyborate, bis(oxalato)borate, alkyl carborane et al. c the similar anions are bis(perfluoroethylsulfonyl)amide, 2,2,2-trifluoro-N-( trifluoromethanesulfonyl) acetamide, tris(trifluoromethanesulfonyl)methanide. d the similar metal salts based anion like FeCl4-.

In principle, cellulose, dissolved in an IL, can easily be converted also with other types of carboxylic acids, or their corresponding acid derivatives. Cellulose succi­nates and phthalates, i. e., dicarboxylic acid derivatives could be obtained by homogeneous derivatization in AMIMICl or BMIMCl with the aid of different acylation catalysts, e. g., DMAP, N-bromosuccinimide, or iodine [27—31]. Thereby, DMSO has partly been utilized as molecular co-solvent in order to guarantee homogeneous reaction conditions. The homogeneous preparation of cellulose glutarate by acylation in BMIMCl under ultrasound irradiation also has been reported as well [32]. ILs, partly in combination with DMF as co-solvent, have been utilized as reaction media for the preparation of 2-bromo and 2-chloro carboxylic acid ester of cellulose, which could be used as macro initiators for the grafting of poly(styrene) and different types of poly(methacrylate) chains onto the polysac­charide backbone [23—26].

Using AMIMCl as homogeneous reaction medium, a series of cellulose benzo­ates with high DS values in the range of 1.0-3.0 has been prepared that carried different moieties at the aromatic ring [20]. Also mixed cellulose derivatives, carrying benzoate groups (preferentially at the primary hydroxyl group) and

4- nitrobenzoate moieties (preferentially at the secondary hydroxyl groups) have been prepared in ILs by step-wise conversion with the corresponding acid chlo­rides. Highly substitute cellulose benzoates, as well as cellulose phenyl carbamates (see Sect. 5.2.3), exhibit high chiral resolving properties and could be utilized as stationary phase in liquid chromatography for separation of enantiomers [59]. Synthesis of these materials is usually performed under heterogeneous starting, e. g., by conversion of cellulose suspended in pyridine or other swelling agents [60]. However, homogeneous derivatization in ILs can facilitate preparation of tailor-made chromatography materials.

In general, cellulose esters are well soluble in ILs even at high DS values meaning that completely homogeneous esterification is feasible. However, in case of fatty acid ester with long, non-polar alkyl chains, the cellulose derivatives become increasingly hydrophobic upon advancing substitution, which renders the products insoluble in the reaction mixture. Cellulose laurates can be obtained in ILs but the derivatives rapidly precipitated upon derivatization of the polysaccharide [14]. Phase separation can be expected also for the higher homologue cellulose stearate although it has not been mentioned explicitly [19].

Compounds containing carboxyl groups are readily available but seldom used directly for esterification with polysaccharides due to their low reactivity. In case conversion into a more reactive acid chloride or anhydride is not feasible, activation of carboxyl groups with N, N’-carbonyldiimidazole (CDI) is a very versatile approach for obtaining cellulose esters under mild reaction conditions [61, 62]. Some cellulose derivatives, namely fuorates and water soluble oxy-carboxylic acid ester of cellulose with a broad range of DS from 0.1 to 3.0, have already been obtained in ILs by this procedure [21, 22]. As a first step of the homogeneous derivatization reaction, an imidazolide is formed in situ in the IL reaction medium, which reacts as active species with the polysaccharide backbone upon liberation of CO2 and imidazole as side products (Fig. 5.2). A carbodiimide has also been tested as coupling agent for homogeneous esterification of cellulose with stearic acid in an IL but only small DS values up to 0.16 could be achieved [36]. Another possibility for the activation of carboxyl groups for subsequent esterification with cellulose is the conversion with sulfonic acid chlorides (Fig. 5.2) [63, 64]. Moreover, reactive intermediates are generated by treating carboxylic acids with iminium chlorides, e. g., prepared in situ from DMF and oxalyl chloride (Fig. 5.2) [65]. The by-products liberated upon activation and subsequent esterification are either gaseous (CO, CO2, HCl) or may act as co-solvent (DMF). The last two approaches have been exploited for the

preparation of various polysaccharide esters but up to now not been adapted for the homogeneous derivatization of cellulose in ILs.

Ionic liquids were reported as early as 1914 when ethylammonium nitrate was shown to melt at 12 ° C [41]. In recent years the study of these compounds has experienced a resurgence. An ionic liquid is defined as a chemical compound that exists as an organic anion and a cation and has a melting point below 100 °C. There are many different kinds of ILs, and many more are being developed (Fig. 8.3). The most common forms are based on dialkylimidazolium, tetraalkylammonium, alkylpyridinium, or tetraalkylphosphonium cations coupled with an inorganic

Common Anions

MeSO,- CFoCOO’ CFECOO

Fig. 8.3 Common IL anions and cations anion [42]. Due to their ionic character, ILs have essentially no vapor pressure. While it has been reported that some ILs can be distilled under the right conditions [43], in general, the vapor pressure is low enough to be neglected. Because ILs are composed of discrete anions and cations, the solvent properties, such as viscosity, melting point, and miscibility with other solvents can be tuned through the right combination and design of each ion. The ability to design ILs to specific substrates, chemistries, and situations is important, because ILs are increasingly being looked to as a medium for applications, including biomass processing [44].

Additionally, most ILs display good stability under a wide range of chemical, thermal, and electrochemical conditions. Some ILs have reported thermal stability of over 300 °C, although the stability is highly dependent on the identity of the IL’s constituent ions [45]. There has been research to show that the long term thermal stability of some ILs is significantly less than that indicated by standard thermogra­vimetric analysis techniques [46]. This may be important in the development of biomass processing techniques in ILs, because one of the main advantages of ILs is the potential for the essentially complete recycling of solvents. ILs are also gener­ally assumed to have good chemical stability. In many cases, strongly acidic, reducing, or oxidizing agents can be used without degradation of the ILs [47, 48].

There are, however, some exceptions to this rule [49]. Dialkylimidazolium based ILs have a mildly acidic hydrogen that undergoes hydrogen exchange in aqueous media and can even deprotonate to form a reactive carbene under basic conditions [49, 50]. Some ILs, such as halide, acetate, or formate based ILs, can form volatile acids (such as hydrochloric, acetic, or formic acid) [43, 51]. Additionally, some ILs can be designed to be reactive with the addition of acidic moieties or metal centers [47, 52, 53]. ILs have also been looked to as media for novel electrochemistry, as some of them have a wide window of electrochemical stability [54].

ILs have found a place in a number of catalytic reactions. In many cases, the solvent properties of ILs increase reaction rate and selectivity [55—57]. Addition­ally, post reaction separations are often made easier due to immiscibility of products with the IL phase, such as in the case of esterification in acid ILs [58, 59]. ILs also give the ability to distill volatile products and reuse the IL [60]. The coordination of ILs to metal centers has also been shown to increase the activity and recyclability of some metal catalysts [48, 52]. While ILs are often designed around specific solvent properties, some ILs are designed to work as a combined solvent and catalyst. A common method for this is to attach an acid group to the end of an alkyl chain on the cation, which has been used to depolymerize cellulose and to catalyze esterification reactions [53, 61, 62].

Recently, some ILs have been shown to be able to either partially or completely dissolve cellulose, lignin, or lignocellulosic biomass. Imidazolium-based ILs seem to be especially well suited for this application. The most common solvents that are used to dissolve biomass are alkylimidazolium chlorides, acetates, and formates, although others have been investigated and used in biomass chemistry [21, 63, 64]. This property of ILs has been exploited in the production of novel materials, such as cellulosic aerogels and films in addition to being used as a solvent for catalysis of lignocellulose [65—67].

The property of these ILs that enables them to effectively dissolve biomass is their ability to function as a hydrogen bond acceptor while only having a limited ability to act as a hydrogen bond donor. In general, it is the ability of the anion to form hydrogen bonds with the hydroxyl groups of the cellulose, disrupting the hydrogen bond crosslinking of the polysaccharide, that makes these ILs effective at solubilizing biomass [22, 68] (Fig. 8.4). Dissolution of glucose in 1,3-dimethylimidazolium chloride was studied through computer modeling to analyze the IL/saccharide interactions further. This work demonstrated the almost exclusive coordination of the chloride anion to saccharides with only minimal contributions from hydrogen bonding and van der Waal forces from the imidazolium cation [69]. The dissolution process first swells the cellulose and, in the case of lignocellulose, extracts the lignin

[70] . Some ILs have even been specifically designed to dissolve carbohydrates without denaturing enzymes to allow for homogenous enzymatic catalysis ofbiomass

[71] . Work has been done to investigate ILs using solvatochromic dyes to probe the hydrogen bonding acidity and basicity, the polarity, and dispersion forces in various ILs [72]. These properties are part of what makes ILs an attractive solvent for the processing of biomass.

Due to their unique proprieties, ILs have been used for biodiesel production processes through lipase catalyzed transesterification (alcoholysis) of vegetable oils (or animal fats) (Fig. 11.2) by several research groups [27—34]. Biodiesel is a renewable diesel fuel that is also known as FAME (fatty-acid methyl esters).

The enzymatic transesterification method offers many advantages over the chemical methods, such as mild reaction conditions, low energy demand, low waste treatment, the reusability of enzymes (lipases in most cases), flexibility in choosing different enzymes for different substrates, and the fact that it allows a small amount of water to be present in substrates. Besides, in chemical processes, some oil or fats may need a pretreatment for deacidification, depending on the composition of the materials, to remove free fatty acids (FFAs), which form soap with alcohols. Lipase can convert both triglycerides and FFAs into biodiesel [2]. Some authors have even successfully used waste cooking oil to obtain enzy­matic biodiesel which may be a promising alternative for reducing the cost of biodiesel [27, 35—38].

The production yield was improved markedly when immobilized Candida antarctica lipase B (CALB) on an acrylic resin was used as a biocatalyst compared with other microbial lipases [27, 29, 30, 39, 40]. Ha et al. [30] screened 23 ionic liquids as solvents in the production of biodiesel from soybean oil using Candida antarctica lipase as catalyst. [EMIM][TfO] produced the highest biodiesel yield (80 % in 12 h of reaction), a yield that was better than for the solvent-free system and other commonly used solvents (tert-butanol). Nineteen ILs were studied to determine their effectiveness as solvents in the transesterification process using Burkholderia cepacia lipase (BSL) as catalyst [32]. The ionic liquid used have combinations of cations and anions, being the cations based mainly on imidazolium, while the anions were [NTf2] and [PF6] to get a suitable reaction media (Table 11.1).

Lipase-catalyzed methanolysis when conducted in a solvent-free medium led to the deactivation of lipase with increased molar ratio of methanol to sunflower oil >3 [45]. A similar deactivation of lipases was also observed during lipase cata­lyzed methanolysis in a biphasic oil-aqueous system for FAME production [46]. Methanolysis was conducted at different molar ratios of methanol to oil

(4:1 to 10:1) and similar results were obtained. Thus the concentration of methanol did not have a great effect on product formation in the presence of IL, which protects the lipase from methanol-induced deactivation.

Among the various types of ILs used, hydrophobic ILs were found to be the most effective for the production of biodiesel, the biodiesel yield increasing with both cation chain length and IL hydrophobicity, and decreasing when ILs with strong water miscible properties were used ([BMIM][NTf2], [EMIM][PF6], AMOENG 100, AMOENG 102, [C16MIM][NTf2] or [C18MIM][NTf2] etc). Hydrophilic ILs were not suitable as solvent in enzyme-catalyzed transesterification as only 10 %

FAME yield was obtained for [HMIM][BF4], while no FAME was observed when [BMIM][BF4] was used as the solvent [33].

This trend can be explained in terms of methanol-induced enzyme deactivation. Hydrophobic ILs protects the enzyme from such deactivation because lipase is entrapped in the IL matrix. The most notable advantages of the use of ILs in such bioconversions are that the biodiesel can be separated by simple decantation and the recovered ionic liquid/enzyme catalytic system can be re-used several times with­out loss of catalytic activity and selectivity. More recently, our group [27] used [C16MIM][NTf2] as a homogeneous reaction mixture and, when the reaction was complete, a triphasic system was created through the appearance of a FAMEs phase (upper layer), a glycerol phase (middle layer) and a lower layer with the ionic liquid containing the enzyme, which could be solidified by decreasing the reaction temperature of the media (the melting point for [C16MIM][NTf2] is 42.6 °C), in this way facilitating extraction of the biodiesel product. Furthermore, ILs provides the ideal medium for removal of the by-product glycerol, thus accounting for the increase in biodiesel yield. A promising strategy employed by Lai et al. [47] was to use cross-linked enzyme aggregates (CLEAs) of lipase from Penicillium expansum as catalyst for biodiesel production in [BMIM][PF6] from microalgal oil, with a conversion of 85.7 % in 48 h.