Ranke and co-workers are pioneers in cytotoxic evaluation of ILs. Their first study of cytotoxicity of ILs was reported in 2002 and concerned 1-butyl-3-methylimi- dazolium chloride [164]. J774A.1 macrophage cells were used and the authors determined the LC-50 (lethal concentration leaving 50 % cells alive): 0.50 mg. mL-1 after 48 h of incubation. They also showed that an augmentation of the incubation time increases the cytotoxicity of BmimCl. These results highlighted the importance of wider evaluations of ILs cytotoxicities.

Jastorff et al. proposed a SAR study based on the alkylimidazolium scaffold. They showed that long alkyl chains increase the global toxicity of imidazoliums (evaluation on IPC-81 leukemia cell lines) [44]. The year after, the same team proposed a detailed biological study of methyl — and ethylimidazoliums on IPC-81 (leukemia cells) and C6 (glioma cells) rat cell lines [151]. Again, they observed that long alkyl chains increase toxicity against cell lines involved. In this case, the anion (here PF6~, BF4~ and Br~) seems to have a few influence whereas another study published by Stepnowski et al. showed that the anion bulk, especially NTf2~, has a preponderant role in toxicity towards HeLa cell lines [165]. These results involved that the cytotoxicity of ILs depended on several parameters: IL structure, concen­trations and cell lines To rationalize the cytotoxic behavior of alkylimidazoliums, Ranke’s team studied the cellular distribution of some ILs and correlated cytotox­icity and lipophilicity [150,162,166]. Indeed, most hydrophobic ILs are more toxic and exhibit higher cellular sorption (C8MimBF4 > C6MimBF4 > C4MimBF4). C10MimBF4 was too toxic and further evaluations were not purchased. After HPLC analysis, they determined cellular distribution (~80 % in cytosol, ~12 % in membrane and ~8 % in nucleus). It is commonly assumed that cytotoxicity of ILs is partially due to interactions between ILs and cell lipidic membranes [167].

On another hand, this same team demonstrated that lipophilic properties of the anion side chains and his chemical stability are widely involved in cytotoxicity but the anions tested do not exhibit intrinsic cytotoxicity towards IPC-81 cell lines [168].

Another group investigated cytotoxicity of ILs. Their studies based on varied ILs (pyrrolidiniums, piperidiniums, imidazoliums…) showed the same tendencies (longer alkyl chains involve higher cytotoxicities) and pyridiniums seems to be more toxic than other ILs [169]. They also showed that functionalized ILs (ethers,…) exhibit significantly lower cytotoxicities [170].

Several exhaustive reports were published in 2010 and 2011 and summarize all known studies concerning cytotoxicity of ILs towards mammalian cell lines (HeLa, CaCo-2, IPC-81, HT-29, C6, MCF-7, NCI60, V79). They confirmed the interdependence between lipophilicity, structure, concentration and cytotoxicity [110, 142, 171]. About 230 ILs with various structures were evaluated on IPC-81 (mammalian cell lines derived from a model of acute myelogenic leukemia) and a QSTR profile was established, confirming the tendencies previously mentioned [172]. Some results are summarized in the next table (see Table 12.3) [162, 172].

However, cytotoxic potential of ILs is a crucial parameter to develop industrial processes, it is not a latent obstacle. For example, some pharmaceuticals (e. g. lidocaine) have been grafted on ILs to increase solubility or biodisponibility [173, 174]. In the aim to set up industrial applications, moreover in pharmaceu­tical industry, establishing cytotoxic profile of ILs is necessary and will represent an important field of search for the future.

Karine De Oliveira Vigier and Francois Jerome

Abstract The progressive introduction of biomass in chemical processes has dramatically changed the way how we design a catalytic process. Among different strategies, assisted catalysis is expected to play a pivotal role in the future. In this context, ChCl-derived ionic liquids and deep eutectic solvents has recently emerged as promising solvents to assist a conventional catalyst in the selective conversion of biomass. In particular, their ability to disrupt the hydrogen bond network of bio­polymers, their ability to stabilize polar chemicals and their low miscibility with common low boiling point solvents open a promising route for the conversion of biomass in a more sustainable way. Beside their low price and low ecological footprint, we wish to demonstrate here that these neoteric solvents have processing advantages that no other solvent can provide in the field of biomass.

Keywords Catalysis • Bioinspired ionic liquids • Deep eutectic solvents • Biomass • Carbohydrates

Abbreviations

BMIM 1-butyl 3-methyl imidazolium

ChCl Choline chloride

DMSO Dimethylsulfoxide HMF 5-hydroxymethylfurfural

IL Ionic liquid

K. De Oliveira Vigier • F. Jerome (*)

Institut de Chimie des Milieux et Materiaux de Poitiers (IC2MP) CNRS-University of Poitiers, ENSIP, 1 rue Marcel Dore, 86022 Poitiers, France e-mail: francois. jerome@univ-poitiers. fr

Z. Fang et al. (eds.), Production of Biofuels and Chemicals with Ionic Liquids, Biofuels and Biorefineries 1, DOI 10.1007/978-94-007-7711-8_3,

© Springer Science+Business Media Dordrecht 2014

3.1 Introduction

The progressive introduction of renewably-sourced raw materials in chemical processes has dramatically changed the way how we design a catalytic reaction and catalysis is now facing to new technological and scientific challenges in this area. Beside the necessity to find innovative ways capable of selectively activating these renewable raw materials, modern catalysis has also to take into account resource management (i. e. carbon, water and metals) to ensure the sustainability of these processes. If during several years catalysis aimed at building new molecules, catalysis has now to integrate the notion of deconstruction (e. g. disassembling of (bio)polymers). Response to all these constraints is however not self-satisfied anymore and catalysis also has to provide chemicals with similar prices and even superior performances than chemicals derived from fossil reserves in order to favour their emergence on the market.

The progressive introduction of biomass, especially renewable polyols such as cellulose, hemicelluloses, monomeric carbohydrates and glycerol, in chemical processes is a clear illustration of this fundamental change that is now operating catalysis. In particular, due to the complex structure and high oxygen content of biomass, catalysis is now facing to new fundamental questions that are currently hampering the industrial emergence of bio-based derivatives such as (1) how to control the regioselectivity of reaction since the presence of numerous hydroxyl groups (and different linkages) can lead to the formation of many side products, (2) how to overcome the low accessibility of biopolymers to catalyst, a major bottleneck in the deconstruction of biomass, (3) how to activate biomass without degrading carbohydrates, (4) what is the effect of water, a contaminant of biomass, on catalyst activity, selectivity, stability, and (5) how to overcome the low solubility of biomass. The specialized literature (academic and industrial) and prospective reports from different institutions and governments estimate that more than 10 years of fundamental researches are still needed to achieve mature industrial processes based on the use of biomass.

Faced with the introduction of biomass in chemical processes, several strategies are under investigation. The first one consists in a direct transfer of actual catalytic technologies based on fossil carbon to renewable carbon. This approach is for instance efficient from vegetable oils and actually explains the large number of publications/patents dedicated to this raw material although it represents less than 5 % of the worldwide production of biomass [1]. Fatty derivatives indeed have structures close to those of hydrocarbons, thus allowing a possible rapid transfer of catalytic technology with minimal cost investments. On the other hand, glycerol, the main co-product of vegetable oils, can be used as a C3 chemical to enter the propene platform [2]. However, this approach can hardly be transposed to ligno — cellulosic biomass (95 % of the worldwide production of biomass!!) mainly because current catalytic systems are not adapted to these oxygenated raw materials that exhibit very complex structures. In this context, a second strategy is under investigation and consists in designing novel catalytic surfaces capable of

❖ ‘У

High cost investments-long term vision Use of conventional catalysts

Scheme 3.1 Heterogeneous catalysis applied to biomass processing selectively activating biopolymers [3]. This long term vision is necessary but clearly requires important cost investments. Indeed, as compared to homogeneous catalysis for which all elementary steps of the catalytic cycles are known at a molecular level, it is not the case for heterogeneous catalysis for which the design of novel solid catalysts is still empirical mostly due to the difficulty to have informa­tion on the catalytic sites at an atomistic level. Assisted catalysis is another concept which is now gaining more and more interest in the field of biomass processing. The idea consists in finding innovative ways capable of assisting or driving a conven­tional catalyst in the selective conversion biomass. For instance, physical methods such as ultrasound or ball-milling are already known techniques to help solid catalysts in the conversion of recalcitrant substrates such as cellulose or lignocel — lulose. From 2000, novel and innovative media such as bio-inspired ionic liquids and deep eutectic solvents have emerged in the current literature. These new media have processing advantages (large dissolution of polyols, insolubility with many organic solvents, tunable electrochemical window, tunable acidity, tolerance to water, etc….) that no other solvent can provide. Their abilities to deactivate water, to stabilize or destabilize reaction intermediates, to disrupt hydrogen bond networks now open efficient tools for assisting a catalyst in the selective decon­struction and conversion of structurally complex raw materials such as biopolymers to more value added chemicals (Scheme 3.1).

In this book chapter, we report the most recent advances made in the field of catalytic conversion of biomass assisted by choline-derived solvent. In particular,

their ability to decrease the crystallinity of cellulose, to assist the deconstruction of biopolymers or to promote the conversion of carbohydrates to higher value added chemicals is discussed. Additionally, at the end of the manuscript, we will discuss the contribution of these neoteric solvents for the purification of bio-based chemicals such as biodiesel and furanic esters.

In this section of the chapter we report data of the fractionation work that was performed in our laboratories. The original focus of our work was to study the mild fractionation and the molecular weight distributions of the resulting precipitated fractions. The reasoning for this was to assess whether it was possible to get technically useful fractions, with suitable molecular weight distributions, by avoiding depolymerisation. Furthermore, the interaction of wood biopolymers and how this affects fractionation, is always a fundamental question that needs answering. [amim]Cl was used as the solvating IL. It has a low tendency to react with lignin, whilst being an efficient solvent for isolated preparations of all wood components. In agreement with earlier results from our laboratory, King et al. [14], it was found that only heavily pulverized starting materials were completely soluble in the IL of choice. The use of coarse materials, such as sawdust, only offered partial solubility. The reasons behind the solubility differences will be discussed in more detail in this chapter. For the regeneration of dissolved components we have applied an alternative method for the non-solvent addition. For this we have used gradual increases in non-solvent volume instead of rapid excessive addition to the IL-solution. As a result, the regeneration event is well controlled and follows the principles of traditional molecular weight distribution-related polymer fraction­ation. By applying a derivatization procedure developed in our laboratory, Zoia et al. [59], we have been able to obtain soluble lignocellulose derivatives for size — exclusion chromatography (SEC). This has allowed us to characterize the total molecular weight distributions for majority of the precipitated fractions. Combining the molecular weight information with composition analysis (acid soluble lignin analysis and IR-spectroscopy), we have been able to observe two fundamentally different mechanisms that apply during component separation, related to the degree of interaction of the wood biopolymers. These are found to be dependent on the degree of pulverization (from extensive milling) and therefore the solubility of the resulting materials. As mentioned previously solubility of wood is dependent on the degree of pulverization. This therefore influences whether the fractionation is extraction based or solvation (and subsequent selective precipitation) based. An acetonitrile non-solvent was able to regenerate majority of the dissolved materials but additional non-solvents, such as water and methanol, allowed for further component separation. The motives for selecting this non-solvent, and more com­prehensive discussion about our data can be found from our earlier publications [23, 60]. It was also found that further purification of isolated crude fractions with water resulted in secondary separations and it was possible to recover more water-soluble materials in their own fraction. A complete flow diagram of the fractionation scheme is presented in Fig. 6.1. Selected starting materials (see Table 6.1) have undergone different mechanical pre-treatment processes. Particle sizes and proper­ties changed accordingly with the preparation method. Nearly intact fibrous struc­tures have remained after TMP pulping, but were notably fragmented during sawdust preparation. Ball-milled materials represent highly pulverized wood that has lost all fibrous characteristics.

As mentioned previously, fractions were analyzed by Klason lignin analysis, ATR-IR and the molecular weight distributions of some fractions were determined by re-dissolution into [amim]Cl, benzoylation and SEC analysis (Zoia et al. [59]). The analysis results are presented in Table 6.2.

The fractionation procedure was performed roughly as follows: Wood samples were heated with [amim]Cl for the specified period and temperature. Crude fraction 1 was precipitated from IL using acetonitrile as non-solvent and was washed with water and dried. Fraction 2 was precipitated from the residual IL solution by addition of further acetonitrile and further washed, using the same procedure as for fraction 1. Fraction 3 was precipitated from IL-solution by water addition, after the acetonitrile had been removed by evaporation. Fraction 4 was prepared from the combined aqueous extracts from fractions 1 and 2. The aqueous extracts were combined, concentrated, precipitated with methanol and dried. Fraction 5 was prepared by concentration of the remaining water solution, from fraction 3, precip­itation with methanol and drying.

6.3.1 Fractionation Based on Molecular Weight

It is well known that polymers of different molecular weight have different solu­bility in solvents. This means that controlling the precipitation of polymers, of high polydispersity, from any solution can be used to separate them into fractions of decreasing molecular weight [62]. The main components in wood have distinc­tively different average degrees of polymerization (DP). Isolated softwood

Table 6.1 Starting materials, their upper particle diameter limits, and lignin contents

celluloses have been measured to have average molecular weights from 730 kDa [63] up to 1,550 kDa [64]. Hemicelluloses are of typically lower DP than cellulose and isolated hemicelluloses consist of polymers on average from 18 to 80 kDa [65, 66], depending on the isolation method. Lignin preparations that represent as close to native lignin as we can isolate with current methods, have molecular weights between 52 and 98 kDa [67]. Differences of such magnitude, including differences in chemical composition of the polymers, should offer plenty of opportunity for separation of lignin from hemicellulose from cellulose by controlled addition of a nonsolvent into ionic liquid. Articles by Lee at al. and Lateef et al. [10, 68], have shown that mixtures of the purified polymers can be highly selectively precipitated

from IL solutions. In actual fractionation processes, using minimally treated wood, this efficiency is never observed.

If we examine at the molecular weight distributions of the fractions from the highly pulverized, 28 days rotary milled and 48 h ball-milled, samples from spruce and Eucalyptus respectively (Fig. 6.2), we can see that there is a distinct precipi­tation based on molecular weight. In both cases, the molecular weights of the fractions decrease from fractions 1 to 3. Fraction 4 overlaps with fraction 2 for spruce, due to the fact that the majority of the material in the water-soluble fraction was originally dissolved from crude fraction 2 (see Fig. 6.1). If we look at the lignin contents for the main fractions 1 and 2 for spruce (see Table 6.2 entry for milled TMP), there is very little change in the lignin contents from the native wood. Thus there is clear evidence for precipitation based on molecular weight and very poor separation of lignin from polysaccharide, contrary to previous reports on the separation of mixtures of the purified polymers [10, 68]. It seems evident that LCCs are preventing the separation of the lignin and polysaccharide portion of this fully soluble pulverized wood. Seemingly, disintegration of the LCC matrix during pulverization creates fragments that have similar molecular weights to cellulose, and that have been extensively depolymerized during milling. As a result, a mixture of similar sized LCC polymers precipitate in order of molecular weight.

Fractionated ball-milled Eucalyptus

M, |Da|

In the production of 5-HMF from carbohydrates, increasing reaction time leads to an increase in the 5-HMF yield for cases in which 5-HMF decomposition does not occur. However, the yield of 5-HMF generally exhibits a maximum, indicating that decomposition of 5-HMF occurs [18, 20, 34]. The reaction mixture changing in color from yellow to deep brown is evidence for 5-HMF decomposition [5]. Gen­erally, there are three pathways for the decomposition of 5-HMF in acid catalyzed dehydration of carbohydrates [3, 21], as depicted in Fig. 9.6. The first pathway is the rehydration of 5-HMF into levulinic acid and formic acid; the second one is the self-

polymerization between 5-HMF molecules; and the third one is the cross­polymerization between 5-HMF and monosaccharides [21, 25, 40]. In non-aqueous ionic liquid systems, 5-HMF rehydration can be suppressed since the water present is limited to that of the dehydration of the carbohydrate. Control experiments without the carbohydrates exclude the possibility of self­polymerization of 5-HMF [20, 34]. Thus, it is thought that the decomposition of 5-HMF and the formation of humins are mainly due to the polymerization between 5-HMF and carbohydrates, which consume the initial carbohydrates and the formed 5-HMF, and hence reduce the 5-HMF selectivity [20].

The catalytic system and heating method have a large effect on the optimal reaction time for the production of 5-HMF. For example, when the dehydration of fructose was performed in [EMIM][Cl]-CrCl2 system, an optimal 5-HMF yield of 83 % was obtained at 80 °C in 3 h reaction time [41], but a comparable 5-HMF yield was achieved in [BMIM][Cl]-Amberlyst 15 resin system at 80 °C in only 10 min reaction time [20]. Microwave heating was found to be able to accelerate the transformation of monosaccharides and di-/polysaccharides into 5-HMF and to shorten the reaction time. Qi et al. [19] investigated the catalytic conversion of glucose in [BMIM][Cl]-CrCl3 system, and a 5-HMF yield of 71 % was achieved in 30 s for 96 % glucose conversion with microwave heating at 140 °C. In comparison, the reaction was performed with oil-bath heating and microwave heating at iden­tical conditions. Microwave heating generally gives higher yields than with oil-bath heating, which was also reported by Li et al. who obtained a 91 % 5-HMF yield with microwave irradiation at 400 W in a reaction time of 1 min [47].

Book Series in Biofuels and Biorefineries aims at being a powerful and integrative source of information on biomass, bioenergy, biofuels, bioproducts and biorefinery. It represents leading global research advances and opinions on converting biomass to biofuels and chemicals; presents critical evidence to further explain the scientific and engineering problems in biomass production and conversion; and presents the technological advances and approaches for creating a new bio-economy and building a clean and sustainable society to industrialists and policy-makers.

Book Series in Biofuels and Biorefineries provides the readers with clear and concisely-written chapters on significant topics in biomass production, biofuels, bioproducts, chemicals, catalysts, energy policy and processing technologies. The text covers areas of plant science, green chemistry, economy, biotechnology, microbiology, chemical engineering, mechanical engineering and energy studies.

Series description

Annual global biomass production is about 220 billion dry tons or 4,500 EJ, equivalent to 8.5 times the world’s energy consumption in 2008 (532 EJ). On the other hand, the world’s proven oil reserves at the end of 2011 amounted to 1652.6 billion barrels, which can only meet 54.2 years of global production. Therefore, alternative resources are needed to both supplement and replace fossil oils as the raw material for transportation fuels, chemicals and materials in petroleum-based industries. Renewable biomass is a likely candidate, because it is prevalent throughout the world and can readily be converted to other products. Compared with coal, the advantages of using biomass are: (i) it is carbon-neutral and sustainable when properly managed; (ii) it is hydrolysable and can be converted by biological conversion (e. g., biogas, ethanol); (iii) it can be used to produce bio-oil with high yield (up to 75%) by fast pyrolysis because it contains highly volatile compounds or oxygen; (iv) biofuel is clean because it contains little sulfur and its residues are recyclable; (v) it is evenly distributed geographically and can be grown close to where it is used, and (vi) it can create jobs in growing energy crops and building conversion plants. Many researchers, governments, research institutions and industries are developing projects to convert biomass (including forest woody and herbaceous biomass) into chemicals, biofuels and materials and the race is on to create new “biorefinery” processes. The development of biorefineries will create remarkable opportunities for the forestry sector, biotech­nology, materials and the chemical processing industry, and it will stimulate advances in agriculture. It will help to create a sustainable society and industry based on renewable and carbon-neutral resources.

Zhen Fang • Richard L. Smith, Jr. • Xinhua Qi Editors

Various types of cellulose fibers are used as essential materials for our modern life. The first step in manufacturing cellulose fiber is the dissolution of cellulose in an appropriate solvent, and numerous solvent systems for such dissolution have been developed [1]. As mentioned before, ILs are now well acknowledged as cellulose dissolution agents. Although a great deal of such research has been carried out to develop a pretreatment method of cellulose for bioethanol production, ILs are now seen as attractive solvents for cellulose fiber production [1, 2]. We developed a novel amino acid ionic liquid, N-(2-methoxyethyl),N, N-diethyl, N-methylammonium alanine ([N221(ME)][Ala]), and demonstrated that it dissolved cellulose very well [24].

Rinaldi [34] reported that addition of [C4mim]Cl or [C4mim][OAc] to DMI, DMF, sulfolane, or DMSO caused effective dissolution of cellulose [34]. Some traditional solvents were composed of a highly polar molecular solvent and an appropriate salt material, such as DMAc/LiCl, DMI/LiCl, or DMSO/Bu4NF as noted earlier [7, 8]. Inspired by these results, we investigated an appropriate combination of polar solvent, such as DMSO or DMF, with an ionic liquid as a cellulose dissolution solvent. The dissolving property of a mixed solvent of DMSO and [C4mim]Cl (1:1 (w/w)) against cellulose was first investigated using micro­crystalline cellulose as a model compound. However, no dissolution of cellulose took place in the solvent. We next prepared various types of mixed solvent of DMSO with hydrophobic ILs (1:1 (w/w)): [C4mim][NTf2], [C4mim][PF6], [C4mim][C5F8] [35], [N221(ME)][NTf2], [N221^™, [N221(me)][C5F8][35], [P444(ME)][NTf2], [P444(ME)][PF6], [P444(ME)][C5F8][35], [C4Py][NTf2], [C4Py] [PF6], and [C4Py][C5F8][35]. These mixed solvents did not dissolve cellulose at all even at 100 °C. On the other hand, it was found that by switching the IL to a hydrophilic liquid like [C4mim][OAc] or [N221(ME)][OAc], the corresponding mixed solvent (DMSO : IL = 1:1 (w/w)) did slightly dissolve cellulose (5 and 7 wt% vs. solvent, respectively) at 100 °C. Since [N221ME][Ala] showed the best dissolution among amino acid ILs [24], we prepared a 1:1 mixed solvent of DMSO and [N221ME][Ala] and found that the resulting solution dissolved cellulose very well (11 wt%) after just 10 min of stirring at room temperature (25 °C); a total of 22 wt% of cellulose was dissolved in this solvent at 100 °C. Furthermore, 23 wt% of cellulose dissolved even at room temperature with 6 h stirring [36].

The solubility depended significantly on the ratio of the IL to DMSO solvent ratio as shown in Fig. 4.7: cellulose did not dissolve in pure DMSO at all, and the highest solubility was recorded for ca. a 1:1 mixture of DMSO and [N221(ME)][Ala] (IL molar ratio (xIL) was 0.25) which coincidentally was the same ratio as our initial testing solvent. The resulting solution coagulated in water or methanol to obtain a transparent regenerated cellulose in quantitative yield and XRD analysis confirmed that the cellulose regenerated from this solution was only Type II form [36].

Rinaldi also reported that p-value of the mixed solvent of DMI/[C2mim] [OAc] increased when the ratio of the IL was increased and reached the highest value at the XIL = ca, 0.1 for the DMI/[C2mim][OAc] solvent system, where xIL indicates the molar ratio of the IL in the solvent system [34]. From around xIL = 0.10, the values were identical to those of the neat IL [34]. Since xIL of 1:1 (w/w) mixture of DMSO/ [N221(ME)][Ala] was calculated as 0.25, the p-value of the solvent might be the same as [N221(ME)][Ala]. [N221(ME)][Ala] has ahigh p-value (1.041) [36], which is almost the same as that reported for [C2mim][Ala] (1.036) [27]. High hydrogen bond basicity of the mixed solvent of DMSO/[N221(ME)][Ala] might contribute to break­ing the inter or intramolecular hydrogen bonds of cellulose and causing its disso­lution in the solvent as proposed by Ohno et al. [14, 18].

It was also reported that instantaneous dissolution of the cellulose (10 wt%) took place when [C2mim][OAc] was added to DMI at a ratio of over xIL 0.4 at 100 °C [34]. We confirmed that a 1:1 mixture of DMSO and [C4mim][OAc] (xIL 0.31) caused 5.0 wt% dissolution of cellulose at 100 °C, while no dissolution of cellulose took place at room temperature in this solvent. On the other hand, the mixture of DMSO/[N221(ME)][Ala] (xIL 0.25) dissolved cellulose even at room temperature. These results clearly indicated that there was a clear contrast in the dissolution

Fig. 4.7 Change in cellulose solubility for a mixed solvent composed of [N221(ME)][Ala] and DMSO at different temperature conditions with 2 h stirring

property between [N221(ME)][OAc] and [N221(ME)][Ala]. Amino acid-based IL is obviously so effective as a co-solvent or salt that it dissolves more cellulose in DMSO than conventional acetate-based ILs [36]. We anticipate that the hydrogen bond acceptor property of [Ala] and [OAc] might be different in the mixed solvent and reflect a different solubility. As mentioned previously, that free amino group of alanine was essential to realize high cellulose dissolution. Therefore, it was sup­posed that the amino group of [Ala] may interact with a certain part of cellulose and contribute to breaking its hydrogen bond network. However, since the cellulose solubility is also modified by the cationic part of the IL, cation might play an important co-operative role in the mechanism for cellulose dissolution. Further investigation of the scope and limitation of our ionic liquid technology will make it even more beneficial in cellulose science.

4.2 Conclusion

Development of an efficient means to dissolve cellulose in a simple solvent has been a long-standing goal in cellulose chemistry. ILs are now acknowledged as cellulose dissolution agents and are also seen as the most attractive solvents for cellulose fiber production. Strong hydrogen-bonding basicity ф-value in KAT values) is now recognized as the most important property of ILs with high cellulose
dissolution. Viscosity of ILs is the second key factor causing cellulose dissolution at low temperature conditions. However, cellulose solubility was not determined only by the physical characteristics of the solvent shown as KAT values, but that affinity of a certain component of ionic liquids with cellulose was an important factor of cellulose dissolution in the ILs. We hypothesized that we might be able to obtain a hint on how to design such anion or cation from nature. Focusing on the structure of hydrolyzing enzyme of cellulose (cellulase), we found that amino acid ILs were strongly capable of dissolving cellulose: N, N-diethyl, N-methyl, N-(2-methoxy) ethylammonium alanate ([N221(ME)][Ala]) worked as an excellent solvent for cel­lulose dissolution among ILs whose anion part was natural amino acid. It should be emphasized that amino acid IL, [N221(ME)][Ala], is a halogen free and safe solvent, consisting of non-toxic ammonium cation and natural amino acid. Furthermore, the present results seem to provide an important hint to cellulose chemists to consider the mechanism of how an enzyme interacts with the cellulose surface.

Bioethanol and biobutanol are two important energy additives and chemicals, while most of them are produced so far by fermentation using sugars as carbon source. Separation of these alcohols from their fermentation broth requires up to 6 % of the energetic value of the compounds themselves [86]. Conventionally, alcohols and water were separated by distillation or membrane technology, but these technolo­gies are energetically costly, or are not mature for large scale application. The tenability of miscibility and solubility of ILs to water, and other organic compounds offers significant advantages for separation of alcohols from water [87].

In an early study [88], ILs were tested for extraction of BuOH from aqueous solutions. It was found that BuOH distribution coefficient in 1-Butyl-3-methylimi — dazolium hexafluorophosphate ([BMIM][PF6])-water system was 0.85, which was very close to the predicted value. 1-octyl-3-methylimidazolium hexafluoro­phosphate ([OMIM][PF6]) had increased molecular dimensions of the alkylimi- dazolium cation gave lower mutual solubility of the ionic liquid and water, resulting in higher extractive selectivity for BuOH. Pervaporative BuOH recovery from 1 wt% aqueous solution and [OMIM][PF6] was investigated using commercial polydimethylsiloxane membrane MEM-100. Although the viscosity of [OMIM] [PF6]-water-BuOH solution was about 100 fold higher than the viscosity of the BuOH-water mixture, the flux rate through the membrane was only 0.6 fold lower at higher selectivity. BuOH-water ratio in the permeation was close to that in ionic liquid feed, suggesting that the membrane did not improve the separation. Distilla­tion is thought to be more economical for BuOH recovery from ILs [89].

It was demonstrated that the solubility of water in the hydrophobic IL 1-alkyl-3- methylimidazolium hexafluorophosphates could be significantly increased in the presence of ethanol as a so-solute. It was found that 1-hexyl-3-methylimidazolium hexafluorophosphate and 1-octyl-3-methylimidazolium hexafluorophosphate are completely miscible with ethanol, and immiscible with water, whereas 1-butyl-3- methylimidazolium hexafluorophosphate is totally miscible with aqueous ethanol only between 0.5 and 0.9 mole fraction ethanol at 25 °C. At higher and lower mole fraction of ethanol, the aqueous and IL components were only partially miscible and a biphasic system was obtained upon mixing equal volumes of the IL and aqueous ethanol. These observations indicated that ILs may be exploited as an extraction solvent for bioethanol recovery from fermentation broth.

As the hexafluorophosphate anion are partially hydrolyzed in the presence of water and thus generating corrosive HF [90], following up research in this area was moved to ILs with anions bis(trifluoromethylsulfonyl)imide and tetracyanoborate. These ILs are more stable in water and are more hydrophobic than those of hexafluorophosphate anion based ILs. Studies have led to the development of ternary diagrams, separation coefficients of 1-hexyl-3-methylimidazolium bis (trifluoromethylsulfonyl)imide/ethanol/water, and 1-hexyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide/1-butanol/water. It was found that 1-hexyl-3- methylimidazolium bis(trifluoromethylsulfonyl)imide can be used successfully to separate 1-butanol from water. Although it can also be used for ethanol separation, the solvent/feed ratio has to be unreasonably high [91, 92].

The tetracyanoborate-based ILs, 1-hexyl-3-methylimidazolium tetracya — noborate, 1-decyl-3-methylimidazolium tetracyanoborate, and trihexyltetradecyl- phosphonium tetracyanoborate, have also been studied. A complete miscibility in the binary liquid systems of 1-butanol with these ILs was observed. The presence of imidazolium cation gave lower selectivity and distribution ratio than those with phosphonium cation. The ILs with the longer alkyl chains at the cation showed higher selectivity and distribution ratio. Regarding to performance of the imidazolium based ILs, the choice of anion was shown to have a large impact on the upper critical solution temperatures of the system. The relative alcohol affinity for the different anions was (CN)2 N > CF3SO3 > (CF3SO2)2 N > BF4 > PF6 [93, 94].

Liquid-liquid extraction of 1-butanol from water employing nonfluorinated task-specific ILs (TSILs) has also been described recently [95]. Tetraocty — lammonium 2-methyl-1-naphthoate [TOAMNaph] was identified as the best IL which had butanol distribution coefficient of 21 and selectivity of 274. These data were substantially better than those of the benchmark solvent oleyl alcohol, which had butanol distribution coefficient of 3.4 and selectivity of 192. The conceptual design study showed that butanol extraction with [TOAMNaph] requires 5.65 MJ/kg BuOH, which is 73 % less than that by conventional distillation.

To establish a more economic separation technology, studies were carried out to get in depth understanding of the vapor-liquid equilibrium of ethanol-water-ILs [96], 1-butanol-water-ILs, and high pressure CO2-induced phase changes [97]. The results suggested that ILs were capable of breaking the binary azeotrope ethanol-water, opening a new possibility as entrainer for this system, while the 1-hexyl-3-methyl imidazolium chloride moved the azeotrope composition to a smaller fraction of ethanol. In addition, hexane assisted ILs extraction of ethanol [98], and phosphonium and ammonium ionic liquid-based supported liquid mem­branes have also been investigated [99].

Ionic liquids are considered as “designer solvents” because the physical, chemical and biological properties of ionic liquids can be tuned by altering the combination of their ionic constituents. However, finding the proper combination of anions and cations and their mixtures among 1018 possibilities to yield required properties is a major challenge. Many attempts have been tried to design and synthesize “task-specific” ionic liquids in the last decade. For example, Abe et al. [122] based on their experimental observations that hydrophilic imidazolium ionic liquids having alkyl ether functionalizing sulfate salts were appropriate for lipase-catalyzed reaction have anticipated that phosphonium

salt which alkyl ether group might also be appropriate for lipase. Their anticipation relied on the fact that phosphonium salt commonly exists in living creatures. Several types of phosphonium ionic liquids have been prepared and tested for the activity of ionic liquid coating lipase. A novel ionic liquids, 2-methoxyethoxymethyl (tri-n-butyl)-phosphonium bis (trifluoromethanesulfonyl) amide ([P444MEM][Tf2N]) was successfully found to enhance the activity two times while perfectly maintaining enantioselectivity of ionic liquid — coated lipase. However, this approach is mainly based on the experimental trial-error means. More recently, many methods including ab initio calculation, molecular simulation, quantum chemistry, and correlation have been successfully applied to calculate and/or predict the physical properties of ionic liquids [123, 124]. Not only physical, but also chemical and biological properties of ionic liquids can be predicted by these methods. This opens a new path for designing task specific ionic liquids which depends less on the experimental trial-and-errors. For instance, in our studies, quantitative structure-activity relationship (QSAR) model based on the information from the structure of ionic liquids (structural molecular

descriptors derived from CODESSA program) were used to predict the activity of Candida antarctica lipase B (CALB) in the kinetic resolution of sec-phenylethanol in ionic liquids. An optimal QSAR model with 5 structural molecular descriptors was established with the correlation efficient (R2) of 0.9481 and 0.9208 for 18 training and 5 testing ionic liquids set, respectively. This indicates that the performance of enzy­matic reaction in new ionic liquids could be predicted by QSAR model based on the structural molecular descriptors of ionic liquids.

Many kinds of materials can be dissolved in ILs, such as the metal salts, gases, carbohydrates, sugar alcohols, cellulose and even the biomass [36, 84, 85]. In this chapter, cellulose dissolution with various ILs would be discussed.

Cations and anions play an important role in the cellulose dissolution process [23]. The soluble ability of cellulose in ILs can be modified by changing the cations or anions. Anions that form hydrogen bonds with hydroxyl groups are effective for cellulose dissolution, small size and alkalinity of anions promote to increase cellulose soluble ability, for example, halide, acetate, formate and dialkyl phos­phate [22, 61, 86]. Different imidazolium, pyridinium and pyrrolidinium-based cations are commonly used in cellulose dissolution together with the anions above mentioned. Cellulose soluble ability decreases in the ILs with length of the alky chain. Meanwhile, alkyl chains or anions with hydroxyl groups tend to be adverse to the cellulose dissolution in ILs due to the increase of hydrogen-bond acidity of ILs [22].

The Kamlet-Taft polarity parameters, for example, p is the hydrogen-bond basicity parameter, a is the measure of the hydrogen-bond acidity, which express the ability to donate and accept hydrogen bonds, respectively, and n* is the parameter of the interactions through dipolarity and polarizability. The parameters a and p are similar to the acid and base characteristics according to the definition, while, the a and p are not completely consistent with the acidity and alkalinity of ILs in all conditions, the a and p emphasize the acceptance ability of hydrogen — bond. As for ILs, p has been the most useful parameter in predicting the solubility of cellulose in different ILs with various anions [87, 88].

With ILs, the higher p and dipolarity caused the better ability to dissolve cellulose [22]. With the cation [Bmim]+, some anions showed different trends for dissolving cellulose due to the increasing hydrogen-bond acceptance ability, for example, the p value of some anions are in an order of OAc~ > HCOO~ > (C6H5) COO~ > H2NCH2COO~ > dca~, and the cellulose solubility was about 16 wt% in [Bmim][OAc] (P = 1.161), which was higher than those in [Bmim][HCOO] (P = 1.01), [Bmim][(C6H5)COO] (P = 0.98) [8, 89—92]. For a given cation, the effect of anions on cellulose dissolution changes greatly, this result may contribute to the formula weight of ILs, which means that the cellulose dissolution is close to the mass percent in the given ILs.

The Kamlet-Taft parameters of selected ILs and examples of used in dissolving cellulose and biomass pretreatment are as followed in Table 1.5, and the properties of ILs examples and applications in biomass are shown in Table 1.6. Common ILs used for cellulose/biomass pretreatment are shown with details in Table 1.7.

Recycling of ILs is one, if not the major technological issue that needs to be solved in order to utilize this novel class of solvents in commercialized procedures in general and for the processing of cellulose in particular. ILs can be considered as rather expensive compounds but reutilization of the solvent for multiple processing cycles decreases the impact of IL prize on the overall process costs. Moreover, preventing pollution with potentially hazardous organic compounds is a matter of general concern when developing environmentally benign processes. ILs are often considered as recyclable mostly because of their very low vapor pressure.

Nevertheless, it needs to be emphasized that efficient recycling strategies for ILs used in (1) shaping, (2) biorefinery, or (3) chemical derivatization reactions are still missing.

The major component that needs to be removed after processing of cellulose is the non-solvent (usually water or an alcohol) used to regenerate cellulose (shaping) or the polysaccharide derivative (chemical derivatization). Additional impurities derive from the chemical derivatization reaction, e. g., residual reagents, side products, and co-solvents, but also from thermal decomposition of the IL and cellulose (see Sect. 5.3.2). Volatile compounds can easily be removed by evapora­tion under reduced pressure yielding crude ILs that, in some cases, could be utilized directly for another dissolution and chemical derivatization cycle. As an example; acylation of cellulose, dissolved in an imidazolium chloride based IL, yields hydrochloric or carboxylic acid when no additional base is added. Both compounds could be removed from the IL together with the volatile precipitation agent and excess acylation reagent [12, 13]. The NMR spectra of the recycled ILs showed no residual impurities and when reused as homogeneous reaction media, results were comparable to derivatization in the initial IL.

IL recycling becomes more complex with increasing number of potential impu­rities. In particular, the removal of non-volatile compounds proved to be more challenging. If a base is applied during the derivatization, the corresponding protonated acid is usually not removable by evaporation due to its higher boiling point [35, 40]. For comparison: the boiling points of pyridine, which has been applied for acylation, tosylation, and tritylation of cellulose in ILs, and pyridinium hydrochloride are 115 and 223 °C. Recycling can be achieved by neutralization of the IL in an aqueous solution, evaporation of water and the deprotonated base, and subsequent extraction of the crude IL with chloroform, which results in precipita­tion of inorganic salts that can be removed by filtration [33]. Finally, treatment of the recycled IL with an anion exchanger might be required in case anionic species are generated that cannot be removed by other means because they are too similar to the original IL’s anion, e. g., carboxylates differing in their alkyl chain, or because their protonated form is not volatile, e. g., tosylate. NMR spectroscopy can be used to follow the individual recycling steps (Fig. 5.7).

So far, most strategies for the recycling of ILs after cellulose processing focused entirely on evaporation to remove the non-solvent, used for regeneration of the polysaccharide, and as well as side product formed upon derivatization. Up to now, it has not been studied whether this rather energy consuming approach is suitable from an economic and ecologic point of view. Thus, alternative approaches for recovery and purification of ILs are constantly studied not only in the field of polysaccharide research. Upon addition of ‘water structuring salts’, e. g., phos­phates, carbonates, and citrates, or of certain organic compounds such as carbohy­drates, amino acids, and surfactants, to an aqueous IL solution, separation into an IL-rich/water-deficient and an IL-deficient/water-rich phase occurs [105,106]. This ‘salting out’ phenomenon has been exploited for recovery of AMIMCl, used as reaction medium for acylation of cellulose, in 85 % yield [16]. 1H-NMR spectros­copy was used to confirm the purity of the recycled IL but no information on

inorganic impurities, in particular the residual amount of Na2HPO4 was provided, which was used to induce phase separation.

Alternative strategies, which are frequently discussed for recycling of ILs in general, are pervaporation, reverse osmosis, and nanofiltration [107, 108]. Only the latter has been studied already for recovery of ILs that were used as cellulose solvent but not as homogeneous reaction medium. A straight forward approach of increasing importance is the development of novel cellulose dissolving ILs that facilitate efficient recycling, e. g., by extraction or induced phase separation (see Sect. 5.3.4). Frequently, utilization of ‘distillable ILs’ such as guanidinium carbox — ylates has been proposed [109]. At high temperatures and low pressure, these ILs decompose into volatile compounds that reconstituted upon cooling. This process yields ILs of high purity but is also very energy consuming. Moreover, most of the impurities need to be removed in advance because they would evaporate prior to the IL.

When developing procedures for the chemical derivatization of cellulose in ILs, it is important to take recycling aspects into consideration as well. To give an example; homogeneous esterification of cellulose with a carboxylic acid chloride (very reactive reagent) in EMIMAc (low viscous room temperature liquid IL) will yield cellulose ester in good yield and high DS. However, purification of the IL by evaporation of volatile compounds is not feasible. Hydrochloric acid, formed as side product, will induce protonation of the less acidic carboxylate anion that is removed under reduced pressure. Thus, partial anion exchange from EMIMAc to EMIMCl will occur. A very complex mixtures of the recycled ILs can also be expected when mixed cellulose esters are prepared [58].