Cellulose fractionation with IONCELL-P

Accepted Manuscript Title: Cellulose Fractionation with IONCELL-P Author: A.M. Stepan A. Monshizadeh M. Hummel A. Roselli H. Sixta PII: DOI: Reference:

S0144-8617(16)30476-3 http://dx.doi.org/doi:10.1016/j.carbpol.2016.04.099 CARP 11039

To appear in: Received date: Revised date: Accepted date:

15-8-2015 12-4-2016 22-4-2016

Please cite this article as: Stepan, AM., Monshizadeh, A., Hummel, M., Roselli, A., & Sixta, H., Cellulose Fractionation with IONCELL-P.Carbohydrate Polymers http://dx.doi.org/10.1016/j.carbpol.2016.04.099 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Cellulose Fractionation with IONCELL-P

M. Stepan, A. Monshizadeh, M. Hummel, A. Roselli, H. Sixta*[email protected]

Department of Forest Product Technology, School of Chemical Technology, Aalto University, P.O. Box 16300, 00076 Aalto, Finland

* Corresponding Author Telephone: +358 50 3841746

Highlights •

Pure cellulose was fractionated using ionic liquid-water mixtures

•

Providing deeper understanding of cellulose dissolution in ionic liquid-water

•

Molecular weight is a key factor in ionic liquid-water fractionation of cellulose

•

Low molecular weight cellulose with narrow weight distribution was obtained

Abstract

IONCELL-P is a solvent fractionation process, which can separate pulps almost quantitatively into pure cellulose and hemicellulose fractions using IL-water mixtures. In this work the role of the molecular weight of cellulose on itssolubility in ionic liquid-water mixtures is studied. The aim of this study was to understand and identify the determining factors of this IONCELL-P fractionation. Cotton linters (CL) served as model cellulose substrate and was degraded by ozone treatment to adjustthemolecular weight to that of hemicelluloses and low molar mass cellulose in commercial pulps. The ozone treated CLs were subjected to the IONCELL-P process using 1-ethyl-3-methylimidazolium acetate ([emim][OAc]) and water mixtures with a water content between 13.5 and 19 wt%. Based onthe molar mass distributions of dissolved and undissolved cellulose the effect of the molecular weight of cellulose in IL-water mixture appears to be a keyfactor in the fractionation process.

Keywords: Cellulose; ionic liquid; IONCELL-P; fractionation; molecular weight Abbreviations: BH (Sodium borohydride); COS (Cello-oligosaccharides); CED (Cupriethylenediamine); CL (Cotton linters); DMAc (N,N-dimethylacetamide); [emim][OAc] (1-ethyl-3-methylimidazolium acetate);DP (degree of Polymerization); GPC (Gel permeation chromatography); IL (Ionic liquid); MMD (Molar mass distribution); Mw (Weight average molecular weight); PDI (Polydispersity index); WAXS (Wide angle X-ray scattering)

Introduction

Cellulose is a versatile molecule which bears potential for many applications in native or modified forms. Cotton has an undoubtable relevance in textile industry. However, cotton production of the world is predicted to fail to meet the natural-fiber based textile demands of the growing world population(Haemmerle, 2011). Not only is there a need to fill this appearing so called cellulose-gap, but in the mindset of environmentally sustainable agriculture, the high land, irrigation and pesticide requirements of cotton production also urges to find alternative cellulose sources for textile applications. This resulted in a number of new dissolving pulp mills to produce high purity cellulose from wood(TheFiberYear, 2015).The dominant industrialized processes to produce dissolving pulps, prehydrolysiskraft and acid sulfite pulping, have drawbacks such as severe cellulose losses and limited efficiency in hemicellulose removal(Sixta et al., 2013).Subsequent hemicellulose removal via cold caustic extraction, complexsolvents or enzymatic treatments typically do not allow for the uncompromised isolation of hemicelluloses and cause changes in the residual cellulose. The increasing environmental consciousness triggered a significant amount of research on novel, environmentally benign and economically feasible processes and solvents to replace the existing ones. Ionic liquids (ILs) are considered as a new generation of “green” solvents due to their tunable physicochemical properties, low flammability,low vapor pressure and reasonably high chemical and thermal stability and several of them have been shown to be desirable solvents for polysaccharides including cellulose(Gericke, Fardim & Heinze, 2012; Gericke, Liebert, Seoud & Heinze, 2011; Heinze, Schwikal & Barthel, 2005; Karatzos, Edye & Wellard, 2012; Peng, Ren & Sun, 2010; Peng, Ren, Zhong & Sun, 2011; Pinkert, Marsh, Pang & Staiger, 2009; Welton, 1999).Thus, ILs quickly gained popularity in the past decade in cellulose modification and processing(Abbott, Bell, Handa & Stoddart, 2005; Barthel & Heinze, 2006; Castro, Rodriguez, Arce & Soto, 2014; Gericke, Fardim & Heinze, 2012; Gericke, Liebert & Heinze, 2009; Gericke, Liebert, Seoud & Heinze, 2011; Heinze, Schwikal & Barthel, 2005; Karatzos, Edye & Wellard, 2012; Kosan, Dorn, Meister & Heinze, 2010; Liu, Zhang, Li, Yue & Sun, 2009; Pinkert, Marsh, Pang & Staiger, 2009; Welton, 1999).Several ionic liquids can also dissolve other lignocellulosic components and even wood itself(King et al., 2009; Sun, Rahman, Qin, Maxim, Rodriguez & Rogers, 2009). Different ionic liquid based fractionationschemes were suggestedinliteratureto separate the constituentsoflignocellulosicbiomass(Brandt, Graesvik,

Hallett & Welton, 2013; Sun, Rahman, Qin, Maxim, Rodriguez & Rogers, 2009). Dissolution of cellulose in IL-water or solvent-antisolvent mixtures also gained more attention recently(Hauru, Hummel, King, Kilpeläinen & Sixta, 2012; Kuzmina, Sashina, Troshenkowa & Wawro, 2010). However, the detailed mechanism of the dissolution and fractionation phenomena is not yet fully understood. Numerous studies have reported the negative effects of water on the cellulose dissolution capacity of ILs (Hauru, Hummel, King, Kilpeläinen & Sixta, 2012; Mazza, Catana, Vaca-Garcia & Cecutti, 2009; Pinkert, Marsh, Pang & Staiger, 2009; Zakrzewska, Bogel-Lukasik & Bogel-Lukasik, 2010).However, by choosing the right cosolvents/antisolventsmixture the selective dissolution of different wood components has been demonstrated allowing for thefractionation of lignocellulosics and paper pulp(Froschauer, Hummel, Iakovlev, Roselli, Schottenberger & Sixta, 2013; Hauru et al., 2013; Roselli et al., 2014; Roselli, Hummel, Monshizadeh, Maloney & Sixta, 2014).Froschauer et al. presented a new methodology for the quantitative separation of bleached paper-grade pulps into high puritycellulose and hemicellulose fractions, without significant yield losses(Froschauer, Hummel, Iakovlev, Roselli, Schottenberger & Sixta, 2013; Roselli, Hummel, Monshizadeh, Maloney & Sixta, 2014; Sixta et al., 2013). There, the [emim][OAc]-water fractionation solvent system did not facilitate cellulose solubility, while the hemicelluloses were still soluble in the mixture. The extraction efficiency and selectivity was highly influenced by the water content and the temperature of the system. Later Roselli et al. investigated the effect of different pulp sources and different IL-water mixtures on the fractionation efficiency(Roselli et al., 2014; Roselli, Hummel, Monshizadeh, Maloney & Sixta, 2014). [Emim][OAc] is one of the most studied and best understood cellulose dissolving ILs with also industrial scale availability(Gericke, Fardim & Heinze, 2012).Thus, [emim][OAc]–water was chosen as solvent systemin this workto elucidate the mechanism of the IONCELL-P fractionation.The key importance of this process lies in the production of structurally unchanged cellulose and delivery of polymeric hemicelluloses, which could be a potential new raw material for industry. This process also allows for the production of narrowly distributed high molecular weight cellulose which is very attractive for material applications and derivatizations, especially if the native cellulose I crystal structure is preserved. Low molar mass celluloses and cellooligosaccharides (COS) also have several potential applications such as prebiotics in food products(Mussatto& Mancilha, 2007; Roberfroid, 2007; Tolonen, Juvonen, Niemela, Mikkelson, Tenkanen & Sixta, 2015). The low availability of COS is one of the reasons for the lack of extensive knowledge on them. Up to today there are only a few ways of preparing lower molecular weight cellulose. COS productionrequires strong acid treatment at low temperatures, while cellulose with a DP lower than the level off DP can be produced by medium acid treatment at elevated temperature or supercritical water treatment (Tolonen, Juvonen,

Niemela, Mikkelson, Tenkanen & Sixta, 2015), which are both accompanied by severe losses. With enzymatic degradations of cellulose it is also possible to produce low molecular weight cellulose. However, below a certain DP around 100 yield losses also become more prevalent(Zhang & Lynd, 2005). Cellodextrins also have been found to have health benefits such as lowering cholesterol levels and in preventing diabetes when consumed in certain daily doses(Cummings, 1997; Wakabayashi, 1995).Miller et al. has published numerous work on cellodextrins and their preparation already half a century ago, which still give the basis of today’s processes(Miller, 1963; Miller, 1960).The most common ways of preparingcellodextrinsareHCl fuming of cellulose microcrystals combined with various methods to recover and purify the cellodextrins. Mixed acid treatments are also commonly used(Zhang & Lynd, 2003).By fractionating pure cellulosea new set of properties can see daylight simply by having cellulose available in low molecular weight cellulose IIwith a narrow distribution(Meiland, Liebert & Heinze, 2011). Understanding of the principle behavior of cellulose in ionic liquid solutions is essential to improve and to commercialize the IONCELL-P process in the near future. However, hitherto the effects of chemical and physical properties of biopolymers on the fractionation efficiency in ILwater solutions have not been evaluated.This work has several similarities to the study done by Meiland et al. and Eckelt et al. on cellulose fractionation where different molecular weight celluloses were fractionated by changing solvent properties, though their work did not address a deeper understanding of the polysaccharide fractionation as a process(Eckelt, Stryuk & Wolf, 2003; Meiland, Liebert & Heinze, 2011). The aim of this work is to understand and establish the relationship between the polysaccharides’ molecule size and the dissolution properties in [emim][OAc]-water system with different water contents. To exclude any effects of chemical variability of the fractionated molecules and, thus, study the purely physical factors governing the fractionation mechanism, pure cotton linters were used as model cellulose substrate and its molecular weight range was adjusted to mimic the size of hemicelluloses and low molecular weight celluloses. To understand the dominant mechanism behind the IONCELL-P fractionation, the newly obtained results are then reflected on the results reported earlier by Froschauer et al.about fractionated birch pulp containing both cellulose and hemicelluloses,meaning biopolymers of different chemical structures.The results of this study contribute to the understanding of the cellulose dissolution in ILs in general and will help to improve the IONCELL-P fractionation process. Materials

The cotton linter was purchased from Milouban, Israel, with cellulose purity higher than 99.5%. The cotton linter had a weight average molecular weight (Mw) of 253.77 kg/mol(with a

polydispersity index (PDI) of 2.2) which corresponds to a degree of polymerization of 1430. The intrinsic viscosity of the sample was 570 ml/g. 1-Ethyl-3-methylimidazolium acetate [emim][OAc] was purchased from BASF, Germany, and used as received.Deionized water was used to prepare the IL water mixtures. Ozone was produced by a Wedeco GSO30 device using oxygen as feed gas. The other chemicals used in this work were: KI from VWR, Finland, (26846.268), Na2S2O3from Merck, Germany, (1.06516.0500), cupriethylenediamine (CED) from VWR(5761.5000), HCl from VWR (20252.335), DMAc from Sigma-Aldrich, Finland, (270555), LiCl from Sigma (62476), acetone from VWR (20066.330), NaOH from VWR (28244.295) and NaBH4 from Sigma (45,289-0) from which adequate solutions were prepared at Aalto University.Asyringe filter from Sigma-Aldrich (Supelco-57183, polyethylene frit, with a porosity of 20 μm)was used for filtration of smaller batches and a metal mesh filterin a nitrogen pressurized steel filtration unit with a metal mesh with a cutoff size of 5-6 µm was usedfor the filtrations of bigger batches.

Methods Molecular weight adjustment of cellulose by ozone treatment

The ozonization was performed in arotary reactor where the ozone was fed through a tube directly in the mixing pulp. The cotton linter’s initial pH level was adjusted to 5±0.5 to prevent the formation of undesired byproducts and reduce the yield losses. The pH was adjusted by controlled addition of sulfuric acid at 3 wt% consistency. After the pH adjustment the excess water was removed from the sample by centrifugation to reach a consistency close to 55-60 wt%.The ozone flowrate was measured 3-4 times and after defining the average of generated ozone rate, the pH adjusted cotton linters is placed in the round bottom flask and the ozone is injected to the sample through a tube while the bottle is rotated.50 g of cotton linter wasozone treated at ambient temperature and high cellulose consistency of 55-60 wt% The obtained pulp(

Table 2, “Ozone treated CL”) was used as a reference sample and starting material for all of the fractionation experiments. To calculate the rate of produced ozone, ozone was collected by purging it into a sealed bottle containing 150 ml of 2-10 wt% KI solution for 2 minutes (the quenching systemincluded an additional similar ozone trap after the first collector).Then, the solution wastransferredinto a

300 mL Erlenmeyer and acidified by adding 10 mL of 2M HCl. The released iodine was titrated with a 0.1-0.2 N Na2S2O3 solution. The amount of ozone consumed by the pulp sample was calculated as the difference of the inlet and outlet rate.The inlet is the ozone generated and measured with an empty reactor.The outlet is the unconsumed ozonepassing the pulp-charged reactor (rotating flask) and determined by the same titrating procedure. The produced ozone rate was fixed, thus the only variable factor in the treatment was the time. To reduce possiblesystematicerrors when long time treatment is required, the treatment was dividedinto shorter (4-5 minute)stages and the consumed ozone was measured in between. Thus, with these ozonizator settings 50 g of cotton linterswasozonizated for a total of 89 minutes. Stabilization

For the viscosity measurement and the IONCELL-P fractionation the treated cotton linters needs to be stabilized by including a post treatment and areducing stage. Post treatment: The post treatment was carried out using hydrogen peroxide (H2O2) and NaOH with the ratio of 8 g/kg and 10 g/kg of cellulose, respectively. The cotton linters wastreated at 75 °C and pH 5 for 3 hours. Reducing stage: Intermediate treatment with sodium borohydride (BH) was carried out with fixed BH concentrations (0.75 g/L), a liquid-to-solid ratio of 15 ml/g at isothermal conditions of 70ºC for1 hour. The pH of the BH treatment was adjusted to 13 with NaOH, where the BH half-life of 30.3 h at 70 ºC was acceptable for the selected treatment time of 1hour. After the stabilization of the treated cotton linter, the samples were dried at room temperature overnightbefore further processing with the IL-water mixtures. IONCELL-P fractionation of cotton linter For this study, different [emim][OAc]–water solutions were prepared with water molar fractions of 0.595, 0.605, 0.625, 0.659 and 0.689 (corresponding to 13.5, 14, 15,17 and 19 wt% respectively) for performing IONCELL-P on ozone treated CL samples. The simple experimental setup of IONCELL-P process is presented in more detail by Froschaueretal. andRoselli et al.(Froschauer, Hummel, Iakovlev, Roselli, Schottenberger & Sixta, 2013; Roselli, Hummel, Monshizadeh, Maloney & Sixta, 2014).The pulp is mixed with the ionic liquid-water system and agitatedin a water bath at a moderate temperature of 60 °C for 3hours. The experiments were performed with 1 or 2 g of pulp at 5 wt% pulp consistency in 50

ml Falcon tubes. The extraction is then followed by filtration of the cellulose/IL-water mixture. For samples treated with the IL solution containing water above 15 wt% a syringe filter of 20 µm porosity is used.The samples fractionated with lower water content mixtures require higher pressure for filtration anda custom constructed nitrogen pressurized steel filtration unit with a metal mesh with a cutoff size of 5-6 µm was used.Theundissolved cellulose was then washed withfresh [emim][OAc]-water mixture of the same composition as used in the extraction step in order to replace the solvent mixture containing extracted lower molecular weight celluloses which is adhering to the undissolved pulp fibers . Subsequently, the extracted pulp was washed 3 times with hot water to remove all traces of ionic liquid. All filtrates were then combined to induce precipitation of the extracted fraction. The precipitated filtrate was collected via centrifugation and the pellets were washed 3 times with hot water. The gravimetrical yields and molar mass distribution (MMD) of both the undissolved and dissolved cellulose fractions were determined. The residual cellulose and the precipitated cellulose were dried at room temperature overnight for further GPC analysis. Analytical methods

The intrinsic viscosity measurements were performed according to the standard method SCANCM 15:99 with the cupriethylenediamine (CED) solution of the samples in a calibrated capillary viscometer. The molecular weight characterization was performed by gel permeation chromatography (GPC)with the method of Borregaetal.(Borrega, Tolonen, Bardot, Testova & Sixta, 2013).The cellulose samples were first subjected to a multistep solvent exchange sequence (from water to acetone to N,N-dimethylacetamide (DMAc))to remove all water and to activate the samples in DMAc. Then the samplesweredissolved in a90 mg/L (LiCl)/DMAc solution at room temperature under continuous stirring and after a 10 times dilution with DMActheywere filtered into vials using 0.2 µm syringe filter. The samples were then analyzed at 25ºC by means of aDionexUltiMate 3000 system with a pre-column (PLgel Mixed-A, 7.5 × 50 mm), four analytical columns (4 × PLgel Mixed-A, 7.5 × 300 mm),an RI detector (Shodex RI-101) and 9 g/L (LiCl)/DMAc solution as eluent with a flow rate of 0.75 ml/min and an injection volume of 100 µl.Pullulan standards with a molecular weight between 343 Da - 708kDa (from Polymer Standard Service GmbH, and Fluka GmbH) were used to calibrate the system. To relate the measured molar masses of the pullulan standards more accurately to those of the cellulose samples, the molar masses of the standards were converted based on the calculations suggested by Berggren et al.(Berggren, Berthold, Sjoholm & Lindstrom, 2003). The cellulose allomorphs were characterized by wide angle X-ray scattering (WAXS).

WAXS measurements were performed at the IAP Fraunhofer Institutes, Golm Germany (twocircle diffractometer D5000 (Fa. Bruker-AXS, Germany) using monochromatic Cu-Kα radiation in symmetric transmission with Ge (111) as monochromator, λ=0.15406 nm; at 30 mA and 40 kV)and at the University of Helsinki, Finland (Seifert ID 3003 X-ray generator; voltage36 kV, current 25 mA, equipped with a Cu tube,λ=0.15406 nm, a Montel multilayer monochromator, and a MAR345image plate detector),respectively.Diffractograms were recorded in a 2θ-range of 4°-104° (step with ∆2θ = 0.2°with a measurement time of 55 s/2Δθ). The sample was rotated at 15 rpm. Each scan was performed 3 times to increase the statistical significance. The scattering curves were corrected concerning absorption, polarization, Compton and parasitic scattering(Fink, Fanter & Philipp, 1985; Roeder et al., 2006).From the corrected and normalized WAXS curves, the degree of crystallinityxcwas determined. The rheological characteristics of the IL-water mixtures and the IONCELL-P permeates were assessed with an Anton Paar MCR 300 shear rheometer (plate-plate geometry, 25 mm plate diameter, 1mm gap size). The cellulose-IL-watersolutionswereanalyzed before precipitating the cellulose for recovery. All measurements were performed at 60 °C coinciding with the extraction temperature in the IONCELL-P fractionation process.The dynamic viscosities of the solution were measured in steady shear mode over a shear range of 0.01-100 s-1. The low viscosity of the IL-water mixtures were near the sensitivity limit of the rheometer which did not allow for accurate viscosity determination at low shear rates <10 s-1. However, the IL-water mixtures are essentially Newtonian fluids and as such it was possible to determine the zero shear viscosity also in the regime of 10-100 s-1. The low viscous filtrates did not show shear thinning within the assessed shear rate range, that is only the first Newtonian plateau was observed and viscosity values could be deduced also at higher shear rate where the rheometer yielded stable values. Results and Discussion

To be able to identify the main driving forces in the IONCELL-P fractionation processboth the physical and chemical properties of cellulose and hemicellulose should be considered. In this work the target was to exclude the effect of chemical differences between the cellulose and hemicelluloses and to investigate solely the effect of the molecular weight on the fractionation with IL-water mixtures. To do so, cotton linters was degraded by ozonizationto a molecular weight corresponding to the molecular weight range of native hemicelluloses and native low molecular weight celluloses occurring in wood pulps. This way the chemical uniformity of the fractionated substrate is ensured. Degradation of the cellulose by ozone resembles the degradation of cellulose by other oxidizing agents in acidic medium, since ozone reacts mainly as a radical in aqueous conditions. The

mechanisms of cellulose degradation/depolymerization triggered by ozone are discussed in detail in the literature(Sixta, 2006). The sample treated with ozone is highly sensitive to alkali conditions. This is a problem for the intrinsic viscosity measurement using CED solution, which would result in alower measured viscosity value than the sample’s true viscosity. To eliminate further degradation after ozonization under alkaline conditions, a post oxidation treatment stage (P stage)and a reducing stage (R stage) wereapplied to the treated samples(Lehtaru& Ilomets, 1996; Sixta, 2006).It is important to note that the stabilization agent can cause yield losses. The relation of the BH concentration and the cellulose yield was described byTestova et al. (Testova, Nieminen, Penttila, Serimaa, Potthast & Sixta, 2014).Severe yield loss of cellulose is only expected at concentration below 0.75 g/L; Therefore, the BH concentration was adjusted to 0.75 g/L in this study. The PDI change during degradation processescan reflect the type of degradation mechanism occurring (random or systematic). If the depolymerization occurs randomly, the resulting PDI will approach a value of 2(Flory, 1936), while a heterogeneous degradation in the accessible regions only would lead to an increase of the PDI. The ozone treated CL samples show a decrease in the PDI compared to the reference sample(

Table 2), which means that the polymer chain scissions follows a more random degradation pattern(Emsley& Heywood, 1995). The MMD profiles of the original and degraded cotton linters are plotted in Figure 1. In addition, the MMD curve of hemicelluloses as present in a bleached birch kraft pulp are shown. By degrading the cotton linters, the MMD is becoming similar to hemicelluloses as it shifts to a lower molecular weight range and the distribution becomes slightly narrower indicated by a reduced PDI.

Figure 1 Molar mass distribution of the original and degraded cotton linters.* Birch hemicellulose data from (Froschauer, Hummel, Iakovlev, Roselli, Schottenberger & Sixta, 2013)

After the stabilization of the degraded cotton linters, a series of fractionations was carried out with increasing water content in the IL-water solvent system (13.5%, 14%, 15%,17% and 19%) each as an independent experimental setup. As

shows,by increasing the amount of water the share of undissolved cellulose also increased, but more significantly, the average molecular weight was lowered for both dissolved and undissolved fractions. Experiments with 13 and 12.5 wt% were also performed on another batch of cotton linters, and the results were in line and following the presented trends and consistent with the results presented in this manuscript. However, for the sake of discussing a uniform set of data, those results are not shown. Also, due to limitations of the available fractionation equipment available,theundissolved fraction below 12.5% and the dissolved fraction above 19% water content, respectively, were not enough for thorough analysis.

Error! Reference source not found.illustrates the distinct molecular weight distributions shifts

observed as the water content of the system is increasing. This trend is in good correlation with the classic solution fractionation of synthetic polymers in general and with the study of Meilandetal. on cellulose fractionation, where upon the addition of antisolventfirst larger, then smaller molecules can be precipitated(Meiland, Liebert & Heinze, 2011; Tompa, 1956a). In the work of Eckelt et al. the solvent power was tuned and a solvent-antisolvent system was also investigated for larger scale cellulose fractionations(Eckelt, Stryuk & Wolf, 2003). Although the presented work was confined to cellulose contents of 5%, the phase diagram of cellulose in different [emim][OAc]-water systems presented by Rudaz and Budtova allows to speculate that the fractionation behavior is not only dependent on the amount of added water, but also on the concentration of the polymer(Rudaz& Budtova, 2013). The authors reported a solubility limit for cellulose with a DP of 180 in [emim][OAc] with 15% water content. This is in good agreement with our findings if we convert the Mw presented in Table 3 into DP (by dividing with the weight of the anhydroglucose unit: 162.15 g/mol): The average DP of cellulose dissolved in 14% water systems is 189, for system with 15 % water it is 151. Thus, this work confirms that the solubility limits of cellulose with a DP around 180 is in the vicinity of 15% water content. The observations reported herein are also in line with the studies of Kuzminaetal. where the size distribution of cellulose particles in IL solutions was studied as function of the water content of the system(Kuzmina, Sashina, Troshenkowa & Wawro, 2010).The authors found that increasing the water content of an IL-cellulose solution will increase thehydrodynamic volume of cellulose aggregates following a logarithmic correlation in infinitely dilute solutions.This phenomenon was explained by the competing interaction of the solvent (IL) and antisolvent (water) with the cellulose hydroxyl groups and larger molecules having larger hydrodynamic volume are more likely to form intermolecular bonds with each other upon disruption of the solvent-cellulose interaction.Since aggregate size depends also on the molar mass, there is an indirect evidence that the exclusion of solvation of colloidal particles is a factor of both water content and molar mass.In the present study the larger

molecular weight celluloses is observed in the non-dissolved fractions and with higher water content smaller molecules are also hindered from dissolution.However, it has to be emphasized that interpretations concerning dissolution processes based on precipitation studies should be treated carefully since they represent distinct processes. In the presented fractionation the focus is on the dissolution process meaning that the target is to never dissolve certain cellulose fractions, thus preserving the cellulose I form and its reactivity for further value added pulp applications. Meanwhile the dissolved fraction changes allomorphs to cellulose II going through a total dissolution and precipitation, and will have a controllable, low molecular weight distribution, which opens new gates for cellulose applications. In these lower molecular weight samples more than half of the molecules have a DP lower than 100 and a third of them lower than 50. Shorter polysaccharides and oligosaccharides are for example gaining more and more attention from the food industry as prebiotics and as gelling agents in cosmetics applications(Carvalho, de Oliva Neto, Fernandes da Silva & Pastore, 2013; Gibson & Roberfroid, 1995; Mussatto & Mancilha, 2007; Roberfroid, 2007).However, these applications have not been deeply investigated for low molecular weight cellulose in the polymerization range of the cellulose discussed in this work. During the filtration of the different fractionations it was obvious that the higher water content systems needed less pressure and shorter times for total separation of dissolved and undissolved fractions. The rheological studies conducted on the separated filtrates and the structural characterization of the corresponding undissolved cellulose samples can provide a deeper understanding of the filtration hurdles. The filtration challenges are not simply due to the decreased viscosity of a solvent system upon the addition of higher amount of water, as that influences the fractionation solvent´s viscosity only marginally.It is in particular the dissolved cellulose and their molecular weight that affects significantly the viscosity of the resulting solution.Error! Reference source not found.clearlyvisualizesthe dramatic increase of the filtrate viscosity when decreasing the water content of the fractionation system.Once a system with low water content is being used, one has to consider the additional time and energy it will take to separate the fractions. In our laboratory setup, the filtrations reached their limits around 13 % water content. However, based on this study the fractionation window could be pushed towards higher water contents, achieving lower molecular weight celluloses with even narrower distribution in the dissolved phase.

Table 4summarizes the total amountof dissolved cellulose per mole of solvent in each fraction

and the molecular weight distribution of the respective fractions. Not only the ratio of the dissolved cellulose increased, but also the MMD of the fractionated polymers was shifting to a

higher molecular weight range as the water content of the fractionating solvent system was decreased. The filtratewith 13.5 wt% water has the highest total amount of dissolved cellulose (2.83g/mol solvent) with the highest average molecular weight(Mw 37.4 kg/mol) at the same time. Consequently, the filterability of systems with low water content is notably exacerbated. The filtrate from the solvent mixture containing 19% water had the lowest total amount ofdissolved polymer (0.3g/mol solvent) with the lowest average molecular weight of the solutes(Mw 16.3 kg/mol). For the latter sample more than 60% of the solubilized molecules had a DPsmaller than100 and one third even below 50. Thesesmall molecules dissolved at low concentrationdue to their low abundance in the samples and thus facilitated the filtration notably.In the work of Lee et al. the viscosity of IL and water solutions of microcrystalline cellulose was studied at low concentrations(Le, Sescousse & Budtova, 2011). With increasing water content of the solvent mixture a subtle shear thinning of the solutions was observed, and although no undissolved particles were visible through microscope, a swollen agglomerated gel-like state was assumed to develop at medium water concentration. In the present study, due to the very small pore size filtrations being performed before the viscosity measurements the nature of the solution is assumed to be different, and not contain any form of undissolved cellulose or gel-like particles. Therefore we did not observe shear thinning within the accessed shear range. (see supplementary information)

However, the fractionation performance is defined not only by the short chain cellulose removal, but also by potential structural changes occurring in the never dissolved cellulose phase.It was found in our previous study on the IONCELL fractionation, that the never dissolved cellulose fraction remains virtually unchanged, retaining cellulose I and its level of crystallinity(Froschauer, Hummel, Iakovlev, Roselli, Schottenberger & Sixta, 2013). However, that work did not investigate solvent mixtures with significantly lower water content. The ozonizated CL used in the present study had a crystallinity of 55% before being subjected to the IONCELL fractionation procedure. When comparing the undissolved cellulose fraction after treatment with the solvent systems of the lowest and highest water content(13.5% and 19%), WAXS clearly depicts a 10% lower crystallinity for the samples processed with 13.5% water content (51% crystallinity)compared to the one fractionated with the 19%water containing [emim][OAc] (61% crystallinity)(Error! Reference source not found.). The crystallinity was increased to 61% by the higher water content fractionation possibly due to the dissolution of short chain cellulose molecules in the more amorphous regions of cellulose entity.On the other hand, when less water was present, the solvent system was strong enough to initiate swelling ofsome cellulose molecules causing amorphization of the undissolved cellulose fraction. It should be kept in mind that the MMDs of theisolatedpolymer fractions do

not depict sharp separations but have an overlapping boundary zone of the MMDs(Tompa, 1956a). The decrease in crystallinity of the undissolved cellulose from lower water content fractionations is most possibly due to the swelling of some cellulose within this critical overlapping regions of molecular weight(Tompa, 1956b). For these molecules,inter- and intramolecular bonds between the cellulose chains are disrupted temporarily just enough to cause swelling and a molecular rearrangement upon regeneration resulting in a decrease of the crystallinity(Hauru, Hummel, King, Kilpeläinen & Sixta, 2012).In these samples, which were fractionated using lower water content, the zone of separation is at higher molecular weight range where proportionally more cellulose molecules are present in this critical zone of separation and swelling, facilitating molecular rearrangements for a larger fraction of the undissolvedcelluloseandresultingina lower crystallinity after recovering.The larger overlap of the MMDs of the fractions is clearly visible inError! Reference source not found.compared to the overlap of the MMDs of fractionations using higher water content. This sheds new light upon the limitations of the fractionation andis in good agreement with the results reported by Roselli et al. where hardwood pulp fractions were more successfully fractionated to pure cellulose and hemicellulose fractions than softwoods(Roselli, Hummel, Monshizadeh, Maloney & Sixta, 2014).Wood xylans have a significantly lower molecular weight than the glucomannans, allowing for a more efficient separation of xylans from the cellulose fraction. Especially in softwood, the glucomannans have a molecular weight that is considerably overlapping with the low molecular weight celluloses, thus it is more difficult to isolate it based on a size driven fractionation(Escalante et al., 2012; Roselli et al., 2014; Zhang, Li, Lindstroem, Stepan & Gatenholm, 2013).Also when hemicellulase pretreatment was applied on the eucalyptus pulp it improved the IONCELL-P fractionation not only due to the disruption of cellulose-hemicellulose bonds but also by lowering the molecular weight specifically of the hemicelluloses and thus further decreasing the critical overlapping molecular weight region of cellulose and hemicelluloses(Roselli et al., 2014).

Conclusions

This work gave a deeper understanding to the mechanism and basics of the IONCELL-P fractionation. Solely mixtures of [emim][OAc] and water were used to fractionate ozonizated cotton linters. By fractionating pure cotton linters cellulose, the effect of molecular diversity and different chemical interactions of the solvents to solutes were excluded, leaving the molecular weight distribution of the studied polysaccharide as the only substrate variable. As it was shown that it is possible to nearly quantitatively fractionate cellulose with the IONCELL-P when in the molecular weight range of hemicelluloses, it is now confirmed that the physical property of size of the molecules is a major driving force in the fractionation process. This

finding is in line with the fractionation of wood pulp carbohydrates previously published, where the larger molecular weight glucomannans were more difficult to separate from the cellulose than the generally smaller xylans. However, since the effect of different chemical structures was not elucidated in this study, it should be considered with caution. By understanding the driving force and potential limitations of this process new approaches can take place to optimize the existing process, and new raw materials and products can be dealt with, such as narrow molecular weight distribution celluloses and low molecular weight celluloses.The IONCELL-P was initially developed for pulp fractionation and production of hemicellulose lean pulps and polymeric hemicelluloses. Thus the more detailed knowledge on the mechanism of this process can also lead to an easier planning of IL-water systems for different pulp fractionations, or it could contribute to a more conscious choice of raw materials for different targets considering the molecular weight ranges of the different hemicelluloses in pulps of different provenience.The more critical structural evaluation of the undissolved fraction also motivates further investigation of other separation technologies and warns for potential effects that were unaddressed till now such as filtration limitations.Stepwise fractionation can also be considered for speciality products produced simply by consecutive fractionation through a gradually changing water content of the systems. The competitiveness of this IONCELL-P method lies in its nearly quantitative outcome, resulting in high yields, especially compared to the degrading methods using acids. Up to today, only enzymatic treatments could deliver higher yields than the most commonly used strong acid treatments, but the end products still have size limitations. As a conclusion, with tunable IL-water solvent systems a controlled, narrow molecular weight distribution of cellulose fractions can be obtained, opening doors for targeted cellulose production as high value raw material for novel specialized products. Acknowledgement

The authors would like to express their gratitude towards PatrikAhvenainen from the University of Helsinki for his valuable assistance in WAXS analyses.

References Abbott, A. P., Bell, T. J., Handa, S., & Stoddart, B. (2005). O-Acetylation of cellulose and monosaccharides using a zinc based ionic liquid. Green Chemistry, 7(10), 705-707. Barthel, S., & Heinze, T. (2006). Acylation and carbanilation of cellulose in ionic liquids. Green Chemistry, 8(3), 301-306.

Berggren, R., Berthold, F., Sjoholm, E., & Lindstrom, M. (2003). Improved methods for evaluating the molar mass distributions of cellulose in kraft pulp. J. Appl. Polym. Sci., 88(5), 1170-1179. Borrega, M., Tolonen, L. K., Bardot, F., Testova, L., & Sixta, H. (2013). Potential of hot water extraction of birch wood to produce high-purity dissolving pulp after alkaline pulping. Bioresour. Technol., 135, 665671. Brandt, A., Graesvik, J., Hallett, J. P., & Welton, T. (2013). Deconstruction of lignocellulosic biomass with ionic liquids. Green Chem., 15(3), 550-583. Carvalho, A. F. A., de Oliva Neto, P., Fernandes da Silva, D., & Pastore, G. M. (2013). Xylooligosaccharides from lignocellulosic materials: Chemical structure, health benefits and production by chemical and enzymatic hydrolysis. Food Res. Int., 51(1), 75-85. Castro, M. C., Rodriguez, H., Arce, A., & Soto, A. (2014). Mixtures of Ethanol and the Ionic Liquid 1-Ethyl3-methylimidazolium Acetate for the Fractionated Solubility of Biopolymers of Lignocellulosic Biomass. Ind. Eng. Chem. Res., 53(29), 11850-11861. Cummings, J. H. M., G.T. . (1997). Role of intestinal bacteria in nutrient metabolism. Clin. Nutr., 16(1), 3– 11. Eckelt, J., Stryuk, S., & Wolf, B. A. (2003). Fractionation of Cellulose. Cellulose (Dordrecht, Neth.), 10(1), 5-12. Emsley, A. M., & Heywood, R. J. (1995). Computer modeling of the degradation of linear polymers. Polym. Degrad. Stab., 49(1), 145-149. Escalante, A., Goncalves, A., Bodin, A., Stepan, A., Sandstroem, C., Toriz, G., & Gatenholm, P. (2012). Flexible oxygen barrier films from spruce xylan. Carbohydr. Polym., 87(4), 2381-2387. Fink, H. P., Fanter, D., & Philipp, B. (1985). Röntgen-Weitwinkeluntersuchungen zur übermolekularen Struktur beim Cellulose-I-II-Phasenübergang. Acta Polymerica, 36(1), 1-8. Flory, P. J. (1936). Molecular-size distribution in linear condensation polymers. Journal of the American Chemical Society, 58, 1877-1885. Froschauer, C., Hummel, M., Iakovlev, M., Roselli, A., Schottenberger, H., & Sixta, H. (2013). Separation of Hemicellulose and Cellulose from Wood Pulp by Means of Ionic Liquid/Cosolvent Systems. Biomacromolecules, 14(6), 1741-1750. Gericke, M., Fardim, P., & Heinze, T. (2012). Ionic liquids - promising but challenging solvents for homogeneous derivatization of cellulose. Molecules, 17, 7458-7502. Gericke, M., Liebert, T., & Heinze, T. (2009). Interaction of Ionic Liquids with Polysaccharides, 8 – Synthesis of Cellulose Sulfates Suitable for Polyelectrolyte Complex Formation. Macromolecular Bioscience, 9(4), 343-353.

Gericke, M., Liebert, T., Seoud, O. A. E., & Heinze, T. (2011). Tailored Media for Homogeneous Cellulose Chemistry: Ionic Liquid/Co-Solvent Mixtures. Macromolecular Materials and Engineering, 296(6), 483493. Gibson, G. R., & Roberfroid, M. B. (1995). Dietary modulation of the human colonic microbiota: introducing the concept of prebiotics. J. Nutr., 125(6), 1401-1412. Haemmerle, F. M. (2011). The cellulose gap (the future of cellulose fibres). Lenzinger Ber., 89, 12-21. Hauru, L. K. J., Hummel, M., King, A. W. T., Kilpeläinen, I., & Sixta, H. (2012). Role of Solvent Parameters in the Regeneration of Cellulose from Ionic Liquid Solutions. Biomacromolecules, 13(9), 2896-2905. Hauru, L. K. J., Ma, Y., Hummel, M., Alekhina, M., King, A. W. T., Kilpelaeinen, I., Penttilae, P. A., Serimaa, R., & Sixta, H. (2013). Enhancement of ionic liquid-aided fractionation of birchwood. Part 1: autohydrolysis pretreatment. RSC Adv., 3(37), 16365-16373. Heinze, T., Schwikal, K., & Barthel, S. (2005). Ionic liquids as reaction medium in cellulose functionalization. Macromol. Biosci., 5(6), 520-525. Karatzos, S., Edye, L., & Wellard, R. (2012). The undesirable acetylation of cellulose by the acetate ion of 1-ethyl-3-methylimidazolium acetate. Cellulose, 19(1), 307-312. King, A. W. T., Zoia, L., Filpponen, I., Olszewska, A., Xie, H., Kilpelainen, I., & Argyropoulos, D. S. (2009). In situ determination of lignin phenolics and wood solubility in imidazolium chlorides using 31P NMR. J. Agric. Food Chem., 57(18), 8236-8243. Kosan, B., Dorn, S., Meister, F., & Heinze, T. (2010). Preparation and Subsequent Shaping of Cellulose Acetates Using Ionic Liquids. Macromolecular Materials and Engineering, 295(7), 676-681. Kuzmina, O., Sashina, E., Troshenkowa, S., & Wawro, D. (2010). Dissolved state of cellulose in ionic liquids - the impact of water. Fibres Text. East. Eur., 18(3), 32-37. Le, K. A., Sescousse, R., & Budtova, T. (2011). Influence of water on cellulose-EMIMAc solution properties: a viscometric study. Cellulose, 19(1), 45-54. Lehtaru, J., & Ilomets, T. (1996). Stabilization of cellulose fibers with sodium borohydride. Proc. Est. Acad. Sci., Chem., 45(3-4), 160-168. Liu, C.-F., Zhang, A.-P., Li, W.-Y., Yue, F.-X., & Sun, R.-C. (2009). Homogeneous Modification of Cellulose in Ionic Liquid with Succinic Anhydride Using N-Bromosuccinimide as a Catalyst. J. Agric. Food. Chem., 57(5), 1814-1820. Mazza, M., Catana, D.-A., Vaca-Garcia, C., & Cecutti, C. (2009). Influence of water on the dissolution of cellulose in selected ionic liquids. Cellulose (Dordrecht, Neth.), 16(2), 207-215. Meiland, M., Liebert, T., & Heinze, T. (2011). Tailoring the Degree of Polymerization of Low Molecular Weight Cellulose. Macromolecular Materials and Engineering, 296(9), 802-809.

Miller, G. L. (1963). Cellodextrins, Methods. Carbohydr. Chem., 3, 134-139. Miller, G. L. D., J.; Blum, R. . (1960). A study of methods for preparation oligosaccharides from cellulose. Arch. Biochem. Biophys., 91, 21-26. Mussatto, S. I., & Mancilha, I. M. (2007). Non-digestible oligosaccharides: A review. Carbohydrate Polymers, 68(3), 587-597. Peng, X.-w., Ren, J.-l., & Sun, R.-c. (2010). Homogeneous Esterification of Xylan-Rich Hemicelluloses with Maleic Anhydride in Ionic Liquid. Biomacromolecules, 11(12), 3519-3524. Peng, X., Ren, J., Zhong, L., & Sun, R. (2011). Homogeneous synthesis of hemicellulosic succinates with high degree of substitution in ionic liquid. Carbohydr. Polym., 86(4), 1768-1774. Pinkert, A., Marsh, K. N., Pang, S., & Staiger, M. P. (2009). Ionic liquids and their interaction with cellulose. Chem Rev, 109(12), 6712-6728. Roberfroid, M. (2007). Prebiotics: the concept revisited. J. Nutr., 137(3S-2), 830S-837S. Roeder, T., Moosbauer, J., Fasching, M., Bohn, A., Fink, H.-P., Baldinger, T., & Sixta, H. (2006). Crystallinity determination of native cellulose-comparison of analytical methods. Lenzinger Ber., 86, 8589. Roselli, A., Asikainen, S., Stepan, A., Monshizadeh, A., von Weymarn, N., Kovasin, K., Hummel, M., & Sixta, H. (2014). IONCELL-P: selective hemicellulose extraction method with ionic liquids. 13th European Workshop on Lignocellulosics and Pulp. Seville, Spain. Roselli, A., Hummel, M., Monshizadeh, A., Maloney, T., & Sixta, H. (2014). Ionic liquid extraction method for upgrading eucalyptus kraft pulp to high purity dissolving pulp. Cellulose, 21(5), 3655-3666. Rudaz, C., & Budtova, T. (2013). Rheological and hydrodynamic properties of cellulose acetate/ionic liquid solutions. Carbohydr. Polym., 92, 1966-1971. Sixta, H. (2006). Handbook of Pulp. Weinheim, Germany: Whiley-VCH. Sixta, H., Iakovlev, M., Testova, L., Roselli, A., Hummel, M., Borrega, M., Heiningen, A., Froschauer, C., & Schottenberger, H. (2013). Novel concepts of dissolving pulp production. Cellulose, 20(4), 1547-1561. Sun, N., Rahman, M., Qin, Y., Maxim, M. L., Rodriguez, H., & Rogers, R. D. (2009). Complete dissolution and partial delignification of wood in the ionic liquid 1-ethyl-3-methylimidazolium acetate. Green Chem., 11(5), 646-655. Testova, L., Nieminen, K., Penttila, P. A., Serimaa, R., Potthast, A., & Sixta, H. (2014). Cellulose degradation in alkaline media upon acidic pretreatment and stabilisation. Carbohydr. Polym., 100, 185194. TheFiberYear. (2015). The Fiber Year 2015, World Survey on Textiles & Nonwovens.

Tolonen, L. K., Juvonen, M., Niemela, K., Mikkelson, A., Tenkanen, M., & Sixta, H. (2015). Supercritical water treatment for cello-oligosaccharide production from microcrystalline cellulose. Carbohydr. Res., 401, 16-23. Tompa, H. (1956a). Polymer Solutions. (p. 219 pp.): Academic Press. Tompa, H. (1956b). Polymer Solutions. (p. 174 pp.): Academic Press. Wakabayashi, S. K., Y.; Matsuoka, A. (1995). Effect of indigestive dextrin on glucose tolerance in rats. J. Endocrinol., 144(3), 533–538. Welton, T. (1999). Room-Temperature Ionic Liquids. Solvents for Synthesis and Catalysis. Chem. Rev. (Washington, D. C.), 99(8), 2071-2083. Zakrzewska, M. E., Bogel-Lukasik, E., & Bogel-Lukasik, R. (2010). Solubility of Carbohydrates in Ionic Liquids. Energy Fuels, 24(2), 737-745. Zhang, Y.-H. P., & Lynd, L. R. (2003). Cellodextrin preparation by mixed-acid hydrolysis and chromatographic separation. Analytical Biochemistry, 322(2), 225-232. Zhang, Y., Li, J., Lindstroem, M. E., Stepan, A., & Gatenholm, P. (2013). Spruce glucomannan: preparation, structural characteristics and basic film forming ability. Nord. Pulp Pap. Res. J., 28(3), 323330. Zhang, Y. H. P., & Lynd, L. R. (2005). Determination of the Number-Average Degree of Polymerization of Cellodextrins and Cellulose with Application to Enzymatic Hydrolysis. Biomacromolecules, 6(3), 15101515.

Figure Captions

Figure 1 Molar mass distribution of the original and degraded cotton linters.* Birch hemicellulose data from [27]

Figure 2 Molar mass distribution of the cotton linter samples fractionated with different ILwater ratios

Figure 3 Viscosity of IL+water solvent mixtures and filtrates with different water content

Figure 4 WAXS of the undissolved cellulose from the IL+water fractionation containing 13,5% and 19% water

Table legends Table 1Cotton lintersozonizationconditions and yield (before P and R treatment) Consistency

Time

Produced Charged Weight* ozone ozone**

Consumed ozone ** Yield loss

wt%

min

G

g/min

(%)

(%)

%

59.31

89

50

0.073

13.14

3.89

1.22

Note:*oven dried weight, ** per gram of oven dried pulp.

Table 2 Molecular characterization of the original and degraded-stabilized pulps Sample

Viscosity DP

Mw

PDI

W(DP<50)

W(DP<100)

ml/g

(kg/mol)

(Mw/Mn)

(%)

(%)

Untreated CL

570

1430

253.77

2.73

0.007

0.018

Ozone treated CL

153

364

75.72

2.33

0.042

0.104

Table 3. IONCELL-P fractionation of ozone treated CL Undissolved Water content of solvent system

Yield

Dissolved Mw

PDI

Total yield

filtrate conc.

Yield

Mw

(wt %)

wt%

kg/mol

PDI

wt%

wt%

kg/mol

%

13.5

60.6

98.3

1.60

1.89

37.8

37.4

2.14

98.2

14

72.7

93.8

1.58

1.33

26.6

30.8

2.04

99.3

15

78.8

86.8

1.61

1.12

22.3

24.5

1.98

100

17

87.3

82.5

1.73

0.62

12.4

18.2

1.78

99.7

19

91.1

79.5

1.91

0.21

4.2

16.3

1.79

95.3

Note: (Mw)weight average molecular weight, (PDI) polydispersity index

Table 4 Properties of the IONCELL-Psolvents, filtrates and the cellulose dissolved in them [emim][OAc]-water mixture (% water in mixture)

Dissolved Cellulose Mw (kg/mol)

DP<50

DP<100

g cell/mol solvent

IL+water viscosity (mPas)

filtrate viscosity (mPas)

13.5

37.4

0.11

0.26

2.83

14.1

69.8

14

26.4

0.17

0.39

1.98

13.6

39.0

15

24.5

0.19

0.43

1.64

12.1

18.9

17

18.2

0.26

0.57

0.89

10.7

13.5

19

16.3

0.31

0.64

0.30

9.58

9.68

gr1 .

Differential mass fraction

1.4

ozonizated CL 13,5% undissolved 13,5% dissolved 14% undissolved 14% dissolved 15% undissolved 15% dissolved 17% undissolved 17% dissolved 19% undissolved 19% dissolved

1.2 1.0 0.8 0.6 0.4 0.2 0.0 3

4

5

Log molar mass

gr2 .

6

[emim][OAc]+water IONCELL-P filtrate

Viscosity [mPa s]

70 60 50 40 30 20 10 0 13

14

15

16

17

Water content (%)

gr3 .

18

19

gr4 .