For fully soluble finely pulverized materials, the dissolution is relatively rapid for all components. Dissolution is much faster compared to coarse materials and wood converts to a completely solvated state. Solvation was not complete with the coarse

materials. The following dissolution and fractionation mechanism seemed to differ greatly from the pulverized materials, based on compositional analyses of isolated fractions (Table 6.2) and XRD-analysis of the sawdust fraction 1 and the original sawdust (Fig. 6.3).

For the incomplete dissolution of the coarser materials (sawdust and TMP), fraction 1 was mostly composed of the materials that remained solid (but seemingly swollen) during the whole dissolution/extraction period. This was determined to be mainly lignin and polysaccharide. Fraction 2 was determined to be mainly cellu­lose, based on Klason lignin and ATR-IR analyses (Fig. 6.4). As fraction 2 was 33 % of the original sawdust fraction, meaning that most of the cellulose is extracted from the partially soluble wood sample leaving an insoluble matrix of lignin and hemicellulose. Further evidence of this is found after XRD analysis of the regenerated fraction 2, in comparison to the starting sawdust. After extraction of cellulose, the amorphous LCC network was remaining. One should ask the ques­tions, ‘Why is there an insoluble fraction when the purified polymers are all soluble in the IL?’ and ‘Why can lignin not be separated efficiently from the polysaccha­ride, even when the finely pulverized samples are completely soluble?’.

Both questions can be addressed by the explanation of precipitation based on molecular weight. However, most solvents will preferentially precipitate one com­ponent over another and this is simply not the complete picture. Both questions can be answered if you take in to consideration that wood is suggested to contain LCCs and it is actually the LCC network that is preventing dissolution. Only material that is not bound into the LCC network is extractable, under mild dissolution conditions.

This is entirely consistent with a report by Lawoko et al. [69] showing that almost all isolatable LCCs from Norway spruce consist of lignin, which is chemically bonded with hemicelluloses. Whereas, only a minor portion of spruce LCCs have been found to contain lignin-cellulose type LCCs. With this literature confirmation it is no wonder that we can extract cellulose from an insoluble LCC matrix.

As anticipated, it was nearly impossible to derivatize and dissolve any further material from fraction 1 for SEC, due to its insolubility. In case of highly pulverized wood, physical degradation of all polymeric components seem to take place and overall polydispersity decreases. Fractions from ball milled TMP pulp further gave more evidence about the close association with lignin and carbohydrates and how these structures could control the total dissolution of wood. The FT-IR analysis offered some details about the carbohydrate compositions in isolated fractions. Neutral sugar analysis is traditionally used to characterize the carbohydrate moie­ties in lignocellulose, but in this work FT-IR was used instead as a fast, non-destructive, and semi-quantitative analytical method. IR spectra of the isolated fractions from two spruce materials, ball-milled TMP and sawdust, are presented in Fig. 6.4. When spectra of fraction 1 from the two materials are compared, the significant presence of hemicelluloses can be seen for sawdust, whereas in milled

TMP the carbohydrates seem to be mostly cellulose. The hemicelluloses are more present in the lower molecular weight fraction 2 for the case of milled TMP. Other significant differences can be found in composition and yield of the water-soluble fraction 4. For sawdust it seems that the majority of the hemicelluloses have remained totally water insoluble in fraction 1, for some yet unknown reason. For milled wood around half of the original hemicelluloses were converted to be water soluble and dissolved together with relatively large portions of lignin. No lignin was observed to be water soluble from sawdust crude fractions.

Once again, this observation could be explained by the covalent attachment of hemicelluloses that are released during the milling via the fragmentation of the supporting lignin polymers, that otherwise would prevent them from being extracted during the water washing.

Sawdust and TMP pulp preparations represent structurally quite unaltered wood. Our results suggest that swelling and dissolution of native or relatively intact fibers start from the amorphous and crystalline domains of cellulose. The solvated cellulose polymers then diffuse to the bulk solution (fraction 2) leaving behind the lignin-hemicellulose matrix that remains in a rather swollen form and is restricted from complete dissolution (fraction 1). Only minor fractions of lignin (fraction 3) or hemicelluloses (fraction 4) seem to be unbound and transfer to bulk solution. Molecular weight analysis showed that the isolated cellulose-rich fraction 2 had a significantly lower molecular weight than reported for e. g. softwood pulps.

From the dissolution treatment, it is hard to estimate if there is significant depolymerization during the 48 h dissolution period. For the sawdust fractionation, some depolymerization of the carbohydrate components during the dissolution seems evident. Molecular weight of the isolated fraction 2 is low, considering the fact that this fraction was composed mostly of cellulose, as our analysis revealed (see Table 6.2 and Fig. 6.4). In light of a recent study by Gazit and Katz [42], the depolymerization of cellulose during long dissolution periods, even in purified IL, is not surprising. Their results indicate that trace level formation of acidic by-products can take place during dissolution, even below the temperatures that were used in our work. The unfortunate fact is that technical pulps have very defined specifications, in terms of molecular weight distributions, and not only their lignin and hemicellulose contents. This means that controlling acidic and oxidative impurities during an IL-mediated fractionation will be critical in the future to obtain technically useful pulps that fit existing value-chains. In many cases even the present commercial ILs contain these impurities. More must be done to quantify and understand the effects of these impurities.

The single experiment that was performed with Eucalyptus resulted in a lignin poor fraction that was the first to precipitate from the IL-solution (fraction 1). As mentioned previously, precipitation was dependent on molecular weight for Euca­lyptus, as well as milled spruce. It remains as a topic for further studies how differences in covalent structures between lignin and polysaccharide will affect the selectivity of separation. Other ILs like [emim][OAc] have been reported to fully dissolve hardwood [8]. This may be due to chemical degradation of LCC-matrix during the treatment, as discussed earlier. It remains possible that other ILs could facilitate higher yields of cellulose-enriched materials by gradual precipitation, only if the LCC cleavage and carbohydrate depolymerization remains at a low level during the dissolution.

6.3 Conclusions

Due to the rapidly expanding field of IL mediated wood processing, our knowledge in this area has increased to a new level. Many new technical advances are apparent, including more refined ILs, electrolytes, pre-treatments and processing techniques. However, the application of fundamental knowledge related to the connectivity of wood biopolymers, wood morphology, wood ultrastructure and even the solubility of wood in ILs seems to have been largely neglected. Increasing awareness related to IL reactivity has brought both challenges and possibilities to wood fractionation. Depolymerization during fractionation can result in undesired products. This is most relevant when molecular weight distributions should be maintained, e. g. for the production of cellulosic pulps. However, in some cases degradation may be beneficial, e. g. for dissolving the LCC network or reducing the recalcitrance of wood for biofuel production.

Based on our work, with sawdust and highly pulverized spruce wood, we have demonstrated that wood is not completely soluble in [amim]Cl in its native state. This is confusing as isolated lignin, cellulose and hemcellulose preparations have been dissolved efficiently in several publications. One possible reason for this is the presence of an extended LCC matrix in wood that is simply of too high molecular weight and is too interconnected to dissolve. This property can be utilized to extract cellulose, as it is not covalently bound to the insoluble LCC matrix. Cellulose is extracted and by careful control of non-solvent addition, the insoluble lignin — hemicellulose rich fraction can be first isolated, followed by regeneration of relatively pure cellulose. This cellulose extraction procedure is not yet at a stage that would yield a technically useful pulp, due to apparent depolymerization, in comparison to technical pulps and holo-cellulose. However, the more we learn about the stability of wood and lignocellulose, in technical and pure ILs, the better are our chances of yielding close-to native polymers.

The implications for biofuels production are more straight-forward, in regard to pre-treatment mechanisms. Certain ionic liquids are excellent media for cellulose dissolution and regeneration to a state, which is easier to process. Presence of impurities or intentionally added catalysts, that may depolymerize the biopolymers during this process, are beneficial, provided the IL is somewhere between 99 and 100 % recoverable. This is a function of the high cost of ILs, at present. The method of biopolymer regeneration, to enhance separate lignin from polysaccharide, is therefore quite important. If degradation is significant enough breakage of covalent linkages between lignin and polysaccharide should facilitate this. IL recyclability is a major challenge here due to the buildup of monomers, dimers, oligomers, silicates and other inorganics. Therefore improving IL recyclability will greatly enhance the chances of success.

6.4 Experimental

6.5.1 Materials

Synthesis of the [amim]Cl was performed according to a method adapted from Wu et al. [74]. Allyl chloride (200 mL, 2.51 mol) and 1-methylimidazole (160 mL,

2.1 mol) were added to a flask under nitrogen atmosphere. The mixture was refluxed at 50 °C with stirring under positive pressure of nitrogen for 18 h. The reaction was determined to be complete by 1H NMR. The mixture was transferred under nitrogen atmosphere to a rotary evaporator, attached to a high vacuum pump. The excess of allyl chloride was removed at 50 °C. The cloudy crude product was further purified by heating, at 80 °C for 18 h, with activated charcoal (3.0 g) and water (200 mL). The mixture was then filtered through Celite in agrade-3 sinter. Water was removed at 65 °C by rotary evaporation over 18 h, under high vacuum, to yield [amim]Cl as a pale yellow viscous oil. 1H NMR (300 MHz, CDCl3) 5 3.97 (3H, s, NCH3), 4.86 (2H, d, J = 6.4 Hz, NCH2), 5.33-5.26 (2H, m, C=CH-C), 5.86 (1H, ddt, J = 16.9, 10.3,

6.5 Hz, C=CH2), 7.40 (1H, s, C=CH), 7.65 (1H, s, C=CH), 10.39(1H, s, NCHN).

Unbleached Norway spruce (Picea abies) thermomechanical pulp (TMP) was

donated from a Swedish mill. Norway spruce sawdust (particle size <0.2 mm by sieving) was prepared with a belt grinder (grade 60), in-house. Eucalyptus grandis was supplied by Novozymes, NC, USA. Prior to ball-milling treatments, Norway spruce TMP was first milled in a Wiley mill with a 20 mesh (0.84 mm) sieving screen. After Wiley milling the 20 mesh powder was extracted in a Soxhlet extractor for 48 h with acetone. A portion of this fibrous material was further sieved to pass a coarse 40 mesh (0.40 mm) sieve. The remaining extracted 20 mesh Norway spruce powder was rotary ball milled in a ceramic plated 5.5 L steel jar with 470 ceramic balls (diameter 0.9 cm) and a rotation speed 60 rpm, for 28 day period. After milling, the fine powder was dried in vacuum oven. The average particle size was determined to be less than 200 mesh (75 pm).

Eucalyptus chips were Soxhlet extracted with acetone for 48 h. Remaining tannins were removed by refluxing in 0.075 M NaOH solution (1:50 w/v ratio) for 1 h prior to milling of the dried sample. Milling was performed in a Fritsch Pulverette planetary ball-mill, with a 20 mL tungsten carbide grinding bowl and steel balls, at a rotation speed of 420 rpm for 48 h in total. The total milling time was made up of a repetitive milling cycle of 30 min milling time and 20 min brake, to avoid burning of the sample. All the wood materials were dried in vacuum oven over night at 40 °C prior to their use.