ARTICLE pubs.acs.org/Biomac
Dissolution Behavior of Different Celluloses Ute Henniges,† Mirjana Kostic,‡ Andrea Borgards,§ Thomas Rosenau,† and Antje Potthast*,† †
BOKU - University of Natural Resources and Life Sciences, Vienna, Department of Chemistry & Christian-Doppler Laboratory “Advanced Cellulose Chemistry and Analytics“, Muthgasse 18, A-1190 Vienna, Austria ‡ Department of Textile Engineering, Faculty of Technology and Metallurgy, University of Belgrade, Karnegijeva 4, 11000 Belgrade, Serbia § Lenzing AG, Research & Development, A-4860 Lenzing, Austria
bS Supporting Information ABSTRACT: Celluloses from different origins were dissolved stepwise in N,N-dimethylacetamide/lithium chloride (9% v/w; DMAc/LiCl) with the aim to study the time course of the dissolution process, completeness of dissolution in the dissolved fractions, possible discrimination effects, and differences between the celluloses. Cellulosic pulps from both annual plants and different wood species were analyzed. The obtained fractions were subject to gel permeation chromatography (GPC) with multiple detection to monitor the development of molecular mass distribution (MMD), molecular mass, and recovered mass. The dissolution behavior of accompanying xylans was followed by quantitative analysis of the uronic acids by fluorescence labeling - GPC. The morphological changes at the remaining fibers in the stepwise dissolution were addressed by SEM. The time needed to dissolve completely the cellulosic pulp differed from species to species, mainly between pulps from annual plants and pulps from wood. Annual plants generally needed much longer to dissolve completely. In the beginning of the dissolution, the dissolved fractions of annual plants showed a distinct discrimination effect because they were enriched in hemicellulose. By contrast, wood pulps dissolve fast and without distinct changes in the MMD of the dissolved fractions over time. Bagasse pulp is an exception to the observation for annual plants and rather resembled the behavior of wood celluloses. Prolonged dissolution times, as often practiced in cellulose GPC, do not lead to any improvements regarding the determination of molecular mass, MMD, and recovered mass of injected sample, so that the dissolution times required for reliable GPC analysis can be significantly shortened, which will be important for biorefinery analytics with high numbers of samples.
’ INTRODUCTION Cellulose, as the most important sustainable material nowadays, will undergo an even more intense utilization in the future. Fast and reliable methods to characterize the polymer will increasingly be necessary for quality and process control. This holds true for celluloses extracted from wood but more and more also for cellulose from annual plants, especially in the context of biorefinery scenarios that will increasingly produce celluloses from new and less common sources. To optimize the corresponding analytical techniques, most importantly, the determination of the molecular mass distribution (MMD) by gel permeation chromatography (GPC), it is necessary to better understand the solution process and its influencing parameters. With the high numbers of samples, especially in biorefinery analytics, the time required for reliable dissolution is a crucial issue to speed up the time for analysis and processing in higherthroughput settings. Cellulose is insoluble in most common solvents because of its peculiar structure and its particularly strong hydrogen bond network. For GPC analysis, which requires the dissolved state, either derivatization techniques or specific cellulose solvents r 2011 American Chemical Society
have to be used. Because derivatization, even if carried out under mild conditions, might impair cellulose integrity, most current approaches to MMD determination rather rely on the direct dissolution of the polymer. The solvent of choice for cellulose analysis by GPC is the N,N-dimethylacetamide/lithium chloride system (DMAc/LiCl).1,2 It is generally agreed that this direct cellulose solvent does not cause any degradation of cellulose and forms a stable solution when used under appropriate conditions.3,4 Prior to cellulose dissolution, several steps to condition the cellulose toward dissolution, commonly called “activation”, have to be performed. There are many protocols on how to dissolve cellulose in DMAc/LiCl, mainly regarding activation procedures, concentration of lithium chloride, and duration of the solution process. The latter aspect seems to differ most significantly in the literature. The suggested times range from a few hours for softwood sulphite dissolving pulp5 and 8-18 h for bleached hardwood dissolving Received: August 23, 2010 Revised: February 13, 2011 Published: March 10, 2011 871
dx.doi.org/10.1021/bm101555q | Biomacromolecules 2011, 12, 871–879
Biomacromolecules
ARTICLE
pulp6 over 48 h for aged and nonaged Whatman filter paper7,8 up to 5 days for bleached softwood pulp and historic cotton fiber paper.9 Keeping in mind that analytics accompanying biorefinery research mostly have to deal with dozens of samples in one batch, a fast and robust GPC analysis of cellulose is required, and such analysis times of days for a single sample would appear prohibitively long. In this study, the process of cellulose dissolution was studied over time, evaluating already dissolved and not-yet-dissolved fractions in terms of changes in the MMD, weight-average molecular mass (Mw), recovered mass, and uronic acid content originating from accompanying hemicellulose constituents (xylans). It was of special interest whether the composition (MMD) of dissolved fraction would stay the same over the whole dissolution process (i.e., the cellulose dissolves uniformly without any constituents being preferred or discriminated) or whether the MMD of the dissolved fractions would change over time (i.e., certain molecular mass fractions or constituents are preferentially dissolved and others follow in later stages of the dissolution). Nine cellulosic pulps were chosen that reflect the requirements of real-world cellulose GPC analysis reasonably well: samples from different origin and samples containing hemicelluloses as well. In the course of the solution process, samples were withdrawn after several time intervals to monitor the process of dissolution and were compared with each other. “Solvent peeling” allows separating the initially dissolved material from the bulk cellulose and provides further insight into the fiber morphology. When combined with fluorescence labeling of uronic acid groups in hemicellulose, the experimental setup also allowed us to get a deeper insight into the internal layered structure of different pulp fibers. Thus we aimed also at a better understanding of reactivity and accessibility of pulps that are determined by structure and morphology of the fiber.10
Solvent Peeling. For each pulp, 10 subsamples (each 20 mg, airdry) were prepared as follows: the subsample was suspended in 200 mL of demineralized water and shortly (2 times 10 s) disintegrated in a kitchen blender. Excess water was removed by suction filtration; the subsample was shortly washed with ethanol, and then placed in a dry 4 mL glass vial with a tight screw cap. The vial containing the pulp was either left in 4 mL of DMAc overnight for solvent exchange or labeled in DMAc/FDAM for 7 days at 40 °C. The excess DMAc was removed. Residual FDAM was removed by washing with DMAc. Each subsample was placed in a fresh 8 mL glass vial, and 2 mL of DMAc/LiCl 9% (v/w) was added. The vials were placed on a laboratory shaker. After certain time intervals (5, 20, 40, 60, 90, 120, 240, 480, 1440, and 7200 min), the dissolution process was stopped by the addition of pure DMAc (final ratio DMAc/LiCl 9%:pure DMAc 2:3). (For the EWKP, more dilute samples were prepared with a ratio of 1: 3 to improve separation in the chromatographic columns.) Directly before GPC measurement, the dissolved fractions were filtered through a 0.45 μm PTFE filter. Instrumentation. GPC measurements used the following components: online degasser, Dionex DG-2410; Kontron 420 pump, pulse damper; auto sampler, HP 1100; column oven, Gynkotek STH 585; multiple-angle laser light scattering (MALLS) detector, Wyatt Dawn DSP with argon ion laser (λ0 = 488 nm); fluorescence detector, Shimadzu RF 535 (λex: 280 nm, λem: 312 nm); and refractive index (RI) detector, Shodex RI-71. To avoid fluorescence originating from pulp components, half of the MALLS detectors are equipped with interference filters (488 ( 10 nm) that were used when appropriate. Data evaluation was performed with standard Astra, GRAMS/32, Chromeleon, and Origin software. Some samples (cotton linters and Whatman filter paper: 5 and 10 min and EWKP) were treated by adjacent averaging to smooth the MMD. The following parameters were used in the GPC measurements: flow: 1.00 mL min-1; columns: four PL gel mixedA LS, 20 μm, 7.5 300 mm; injection volume: 100 μL; run time: 45 min; and N,N-dimethylacetamide/lithium chloride (0.9% v/w), filtered through a 0.02 μm filter, was used as mobile phase. The amount of dissolved material was determined from the RI signal using a dn/dc of 0.136 mL/g and a detector constant of 5.3200 10-5 V-1; both values were determined by the authors. Microscopy. To prepare cellulose samples for SEM analysis, we sputtered samples with gold for 60 s. Instrumentation: Hitachi S-4000 at 10 kV accelerating voltage. For light microscopy (Kruess, Germany), pulp samples were prepared by solvent exchange to activate the cellulose. Dissolution of pulp samples was observed in transmitted light, and photomicrographs were taken. For each pulp, three specimens were taken. Solid-State 13C NMR Spectroscopy. Wetted pulp samples were tightly fitted into 7 mm zirconium oxide MAS rotors for 13C CP-MAS NMR analysis. All spectra were acquired on a Bruker Avance DPX300 NMR spectrometer, operating at 75.46 MHz 13C resonance frequency. The spectrometer was equipped with a 7 mm 1H/BB MAS probe, the sample spinning speed was 4 kHz, and the 1H decoupling field was 50 kHz. NMR measurements were performed at room temperature (T = 26 °C) with the following acquisition parameters: acquisition time 41 ms, cross-polarization (CP) contact time 1 ms, repetition interval 3 s, 1024-2048 accumulations. Chemical shifts were referenced to δTMS = 0 ppm using adamantane as external reference. All acquired FIDs were multiplied with an exponential window function using a line broadening of 10 Hz prior to Fourier transformation. Zero-order/first-order phase correction and sinusoidal baseline correction yielded the final spectra.
’ MATERIALS AND METHODS Chemicals. Chemicals were of the highest purity grade available and were used as received. DMAc was obtained from LGC Promochem, Germany. Lithium chloride was purchased from BDH Prolabo by VWR International, Germany. Demineralized water was prepared by reverse osmosis. Pulps and Papers. Different pulps were submitted to solvent peeling (Tables 1 and 2 of the Supporting Information). Whatman filter paper No. 1 and cotton linters (Buckeye) were chosen to study the dissolution process of samples that consist of almost 100% cellulose and of only minor amounts of hemicelluloses, extractives and lignin. Contrary to the rather pure cellulose samples, totally chlorine-free (TCF) bleached beech sulphite pulp (BWSP) and elemental-chlorinefree (ECF) bleached paper-grade eucalyptus kraft pulp (EWKP) still contain hemicelluloses and minute amounts of residual lignin. Also, two softwood kraft pulps were included into the study: spruce/pine (80%/ 20%) softwood kraft pulp (SWKP) and another softwood sulphite pulp (SWSP) produced from a mixture of spruce and pine (60%/ 40%) with a very low lignin content (0.6%). This pulp was TCF bleached. Three pulps from annual plants were chosen to complete this study: bagasse, flax, and hemp pulp. All of them were nonblended, pure pulps that were ECF bleached. Labeling of Uronic Acids. The labeling of the uronic acid groups was performed according to the method described by Bohrn.21 Fluorenyl diazomethan (FDAM) was designed as the group-selective label, which allows for quantification of uronic acid groups in relation to the MMD after GPC separation. The marker needs to be freshly synthesized before labeling.
’ RESULTS AND DISCUSSION Molecular Mass Distribution in the Course of Cellulose Dissolution - Annual Plant Pulps. At first, cotton pulp samples
were studied (Figure 1). The first two dissolution time intervals 872
dx.doi.org/10.1021/bm101555q |Biomacromolecules 2011, 12, 871–879
Biomacromolecules
ARTICLE
Figure 1. Molecular mass distributions of two cellulose samples after different time intervals for dissolution. Left: Cotton linters. Right: Whatman filter paper.
Figure 2. Molecular mass distributions of two cellulose samples after different time intervals for dissolution. The samples originate from annual plants and have a significant hemicellulose content. Left: flax. Right: hemp.
Thus, fiber morphology and cellulose allomorphism in cotton remain as general influences regarding the rate-determining step for dissolution. This assumption is further supported by swelling tests shown below (Fiber Morphology section). The same observations as those for the nearly neat cellulose samples hold partially true for hemicellulose-containing annual plant celluloses, such as hemp and flax. A low-molecular-mass shoulder is clearly detected within the first 20 min of dissolution, especially in the flax pulp samples (Figure 2). After 40 min, only a slight distortion in the symmetry of the MMD has remained. In flax pulp, the contribution of the low-molecular mass fractions to the overall MMD continuously decreased from 34 to 5% in the course of dissolution. The portions of high- and low-molecularmass fractions were calculated from the bimodal distribution. As a borderline between low-molecular and high-molecularmass area, the minimum was selected, and the ratio of the two areas was expressed in percent. In the case of hemp samples, 1 h is enough for a sufficiently good dissolution of the pulp. After that time, the MMD does not change anymore, even if the dissolution time is extended up to 5 days. Flax samples require
(5 and 20 min) were generally too short to achieve a complete dissolution. The signal was still very noisy, indicating that no sufficient amount of sample was dissolved. Within these short time intervals, also the MMD was still subject to strong changes. The presence of short-chain cellulose and minor amounts of hemicelluloses dominate the MMD in this early stage of dissolution. Trace amounts of hemicelluloses present in cotton linters and short chain cellulose fractions were only prominent within the first period of cellulose dissolution because they dissolved first. After 120 min, however, these traces of lower molecular mass fractions disappear, and the typical MMD profile of cotton linters with a low polydispersity can be seen. For complete dissolution, these samples require at least 240 min. Both pulps in Figure 1, Whatman filter paper and cotton linters, do not contain significant amounts of hemicelluloses. Their dissolution can, in principle, be influenced by fiber morphology, the higher content of cellulose Iβ as compared with wood pulp (cf. Table 2 of the Supporting Information), or the extractive content. Both celluloses are purified from extractives, so the extractive content cannot be the decisive factor. 873
dx.doi.org/10.1021/bm101555q |Biomacromolecules 2011, 12, 871–879
Biomacromolecules
ARTICLE
Figure 4. Molecular mass distributions of BWSP after different time intervals for dissolution.
Figure 3. Elution profile of Bagasse polysaccharides (cellulose and hemicelluloses) pulp after solvent peeling.
∼10 h for good dissolution. They still change their MMD within the first hours of dissolution; however, there is no difference between 24 h and 5 days. Another annual plant cellulose studied was bagasse pulp. Unlike other annual plant celluloses (cotton linters, hemp, and flax), bagasse dissolved almost spontaneously. To observe the development of dissolution better and to avoid fitting errors, we plotted the elution profiles without conversion into the MMD (Figure 3). Because of the presence of high amounts of hemicelluloses, it is difficult to fit correctly the data (retention time vs log molecular mass): the fit refers to either the cellulose part of the sample or the hemicellulose part that is different in the slope. For Mw determination (see later in the text), the bagasse pulp was fitted for its cellulose fraction. After 5 min of dissolution time, 32% of the mass is concentrated in the low-molecular-mass shoulder, and 68% is concentrated in the high-molecular-mass shoulder; this ratio changes only insignificantly to 29 versus 71% after 60 min. Bagasse samples require 60 min for good dissolution. There is no difference in the MMD between 1 h and 5 days. Molecular Weight Distribution in the Course of Cellulose Dissolution-Wood Pulps. Bleached BWSP is a dissolving pulp that was designed for high reactivity in chemical derivatization reactions. A complete dissolution of this sample was achieved already within the first 5 min (Figure 4). No additional changes of the MMD were observed for all other time intervals up to 1440 min. Therefore, a few minutes in the solvent is sufficient to shape the characteristic MMD of bleached BWSP and to guarantee a good dissolution. Paper-grade pulps are often more difficult to dissolve than dissolving pulps (“nomen est omen” for the latter). In general, paper-grade pulps show distinct peaks from the hemicellulose fraction. An illustrative example is bleached EWKP (Figure 5). In the kraft cooking process, the hemicellulose fraction is initially dissolved under the strong alkaline conditions applied (pH 14), but progression of the cook leads to a decrease in pH to ∼12, which causes reprecipitation of the hemicelluloses into and onto the cellulose fibers. This hemicellulose “deposit” could hamper the dissolution process in DMAc/LiCl causing some oscillation of the MMD (Figure 5). The higher dilution applied for this pulp improves the separation of the hemicellulose peak but causes some aggregation in the high-molecular mass region at the same time, visible as a shoulder.
Figure 5. Molecular mass distributions of bleached EWKP after different time intervals for dissolution. Samples prepared with higher dilution (1 part sample/3 parts DMAc) after smoothing procedure.
After the initial period of 60 min with strong changes in the hemicellulose region, the MMD does not change significantly anymore. Times of 120 min dissolution time are sufficient for this type of sample to obtain a representative shape of MMD. Finally, two softwood pulps were subjected to dissolution analysis (Figure 6). As already observed for BWSP, the MMD of both pulps did not change over time. Thus, dissolution was already complete within the first 5 min time interval, and the MMD did not change further afterward. From visual analysis of the dissolved sample, it was evident, however, that the SWKP pulp solution in DMAc/LiCl still contained fibers and did not appear to be completely dissolved. Even after the longest interval, some residue still remained undissolved. No attempts were made to quantify the amount of that undissolved fraction of SWKP. Dissolution Process in Terms of Mw and Calculated Mass Recovery. In the following section, the Mw and the calculated injected mass recovery of all samples are discussed, divided into celluloses from wood and from annual plants. The fastest pulp to dissolve was bleached BWSP, resulting in a constant Mw after only a few minutes (Figure 7, left). SWSP and SWKP pulps also 874
dx.doi.org/10.1021/bm101555q |Biomacromolecules 2011, 12, 871–879
Biomacromolecules
ARTICLE
Figure 6. Molecular mass distributions of two softwood pulp samples after different time intervals for dissolution. Left: SWSP pulp, no visible undissolved fractions. Right: SWKP pulp, undissolved fraction still visible.
Figure 7. Development of Mw of the dissolved fraction in the course of dissolution over 1440 min. Left: three wood celluloses. Right: four annual plant celluloses.
quickly dissolve. In Figure 7 (right), the selected pulps from annual plants are compared. They also reach a constant level of Mw after 1440 min. Note that pulps from annual plants are more difficult to dissolve than pulps derived from wood. In the earlier phase of dissolution, that is, between 5 and 120 min, the Mw still increases. This increase is also observed between 120 and 1440 min, however, it slows down. Therefore, dissolution should take up to 1440 min. This is faster than often recommended in sample preparation procedures. The time needed to achieve a constant Mw is generally reflected in the time needed to establish a constant calculated mass recovery (Figure 8). According to dilution protocol in the sample preparation, the recovered mass should range between 250 and 300 μg in theory. This theoretical amount is only reached by cotton linters and bleached BWSP. The absolute value of the recovered mass is influenced by the dn/dc (refractive index increment), the RI detector constants, and the accuracy of the injector. Hence, data given in Figure 8 are mainly reflecting how fast a sample reaches its final dissolution state with maximum solubility. With the only exception of cotton linters, severely prolonged dissolution times, however, did not improve the mass recovery.
When comparing pulp from wood with that of annual plants, there are two differences: celluloses from annual plants generally need more time to establish a constant calculated mass than the counterparts pulp from wood, and the total mass recovery is often higher for wood than for annual plant celluloses. Most pulps from wood have reached a constant calculated mass recovery after 120 min. The total mass recovery, however, differed a lot between the pulps: from SWKP pulp, only ∼100 μg is found, whereas for SWSP pulp, >300 μg is recovered. Some pulps from annual plants (flax, hemp, and bagasse) also reach the maximum mass recovery after roughly 120 min, in the case of bagasse even very fast after ∼5 min. In other cases, such as in the case of cotton linters and Whatman filter paper, the amount of recovered mass still increased, even after 24 h. State of Dissolution - Conformation Plots. In Tables 3 and 4 of the Supporting Information, the exponent ν derived from the conformation plots, log molecular mass versus log rms (root-mean-square radius), are shown. In his renormalization group theory, Freed predicted the value of the exponent ν to be 0.588. Schulz et al.15 interpreted values of ν slightly higher than that value as an indication of increasing chain stiffness or, for even higher values of ν > 0.8, as aggregates of cellulose chains. 875
dx.doi.org/10.1021/bm101555q |Biomacromolecules 2011, 12, 871–879
Biomacromolecules
ARTICLE
Figure 8. Calculated mass recovery of celluloses upon GPC analysis. Left: four wood pulps. Right: five annual plant pulps.
Figure 9. Comparison of uronic acid contents, indicative of hemicelluloses, in the dissolved fraction over dissolution time. Left: Comparison between EWKP, flax, and bagasse. Right: Zoom into the data of bagasse pulp.
In Table 3 of the Supporting Information, there is no ν value higher than 0.58 for wood pulp samples, with the exception of the diluted EWKP samples with ν values between 0.59 and 0.61, resulting from aggregation due to higher dilution (cf. also Figure 5). The higher dilution was done to obtain a better separation of hemicellulose and cellulose fraction, at the expense of slight aggregation in the high-molecular-mass part. SWKP and BWSP samples, however, have a ν value below 0.5 that indicates a slightly more compact structure. The ν values for annual plant cellulose samples are generally higher than those from wood (Table 4 of the Supporting Information). Only cotton linters ranged between 0.5 to 0.6. Hemp, flax, and bagasse pulp have slightly elevated ν values between 0.56 and 0.65, indicating higher chain stiffness. Whatman filter paper samples have strikingly high ν values, especially after short dissolution intervals of several minutes. The value of 0.88 at the beginning of the dissolution process (after 5 min) indicates that in this stage dissolution is still largely incomplete. The ν values for Whatman in later stages of dissolution are still clearly higher than the corresponding values of the other annual plants.
Uronic Acid Content (Determined by FDAM-Labeling) in the Course of Solvent Peeling. As the changes in MMD and in
the ratio between high- and low-molecular-mass fractions for cotton linters, flax, and hemp suggest, the solution process is very likely to start from the better accessible regions, where the major part of the hemicelluloses that are more rich in uronic acids is contained. This observation can be further sustained by determination of the uronic acid content in the solvent peeling fractions. EWKP and flax, two pulps with high carboxyl group content, show a clear decrease in uronic acid content that correlates well with the changes in MMD (Figure 9). High uronic acid concentrations are correlated with high hemicellulose contents, that is, xylan having 4-O-methylglucuronic acid residues. The decrease in uronic acid groups over time allows us to monitor the dissolution of the xylan portion in the pulp. Hemicelluloses are located in less-ordered regions and carry the main part of the uronic acids, as can be seen in Figure 10. The EWKP has the highest xylan content because of the xylan redeposition during pulping explained above. The major part of this hemicellulose fraction dissolves within the first hour. The hemicelluloses in flax pulp are also preferentially dissolved within 876
dx.doi.org/10.1021/bm101555q |Biomacromolecules 2011, 12, 871–879
Biomacromolecules
ARTICLE
Figure 10. Comparison between the development of elution profiles and uronic acid groups in the course of dissolution of flax pulp. Left: RI signal. Right: concentration of uronic acid groups.
Figure 11. Optical micrographs of fibres from different pulps during dissolution. Left: Fibers from Whatman filter paper shows intense ballooning. Middle: first balloon in flax fibres observed after 120 min. Right: BWSP dissolves instantly. Bar represents 50 μm.
the first 2 h, but the speed of dissolution is lower and comparable to that for establishing a constant MWD (cf. Figure 2). Because the crystallinity index of flax is higher than the one of EWKP, it takes longer to dissolve the sample and to access all uronic acids in the flax sample. Bagasse pulp contains fewer uronic acid groups in its low-molecular-mass region than the other two pulps. It dissolves very fast and quickly establishes constant uronic acid content, already after 1 h. Fiber Morphology. Rather than the original fibers surface, the swelling behavior in DMAc/LiCl (9% v/w) and in particular the tendency of the fiber to form so-called “balloons” in the dissolving process is more instructive for evaluating the dissolution. In general, ballooning can be observed in all plant fibres. It originates from the difference in solubility between the S-layers that are rich in cellulose and easier to dissolve and the P-layer with its more complex composition. The ballooning depends predominantly on the processing of the fiber: Only when most of the P-layer of the fiber still exists and only minor cracks are present to allow the restricted penetration of the swelling agent or solvent can ballooning be observed. Its occurrence is an indicator of the dissolution process starting from the inner layers (the S2-layer), that is, from the lumen side, whereas the P layer still withstands the dissolution process and bulges.16 In Figure 11, snapshots of three pulp samples have been selected to depict their differing swelling and dissolution behavior.
Cotton fibers from Whatman filter paper showed the most pronounced ballooning tendency, whereas for hemp no ballooning was observed at all. In bagasse and flax pulp, some ballooning could be observed. The ballooning tendency correlates with the development of the MMD in the course of solvent peeling and allows a rough preview of the dissolution behavior. In the case of fast dissolution by immediate disintegration into fiber fragments, the final MMD that characterizes the bulk sample is obtained immediately (c.f. Figure 4). One example for this is BWSP, where the P layer was already destroyed in the acidic pulping process. In fibers that exhibit ballooning, such as flax, the readily accessible regions containing both hemicellulose and cellulose are dissolved first and are only later followed by the more structured cellulose regions (c.f. Figure 2, left). This can be nicely seen in the SEM pictures of flax pulp upon dissolution in Figure 12, which shows the remaining, undissolved parts of the samples after different dissolution times (2, 5, and 10 min). The primary observation is that fibrils present at the fiber surface are dissolved during the first minutes (Figure 12, left), a closer look at the fiber surface shows separated fibrils (Figure 12, middle). These changes of the fiber surface result from the dissolution and removal of easily accessible and reactive regions. In later dissolution stages, the solvent enters the cracks of the P/S1 layer and starts to attack the S2 layer. This was made visible by stopping the dissolution process after 10-15 min (Figure 12, right): the P layer was burst 877
dx.doi.org/10.1021/bm101555q |Biomacromolecules 2011, 12, 871–879
Biomacromolecules
ARTICLE
Figure 12. SEM pictures of flax pulp during progressing dissolution. Left: starting flax pulp; fibers show some signs of defibrillation. Middle: fiber surface of flax pulp after 2 min of dissolution; fibrils from the fiber surface are completely dissolved, and separated fibrils become immanent. Right: the burst P layer allows a view on the S1 layer that has been swollen (ballooning) and now had partially collapsed after solvent removal.
open, and the wrinkled structure of a previous balloon of the S1 layer becomes visible. This correlates well with the observed MMD that changed accordingly in the course of dissolution. Therefore, two dissolution mechanisms can be distinguished: areas with intact P layer dissolve by ballooning, whereas in areas without intact P layer, dissolution of outer layers and swelling of S layers result in longitudinal cracks (Figure 12, middle) and hence a better penetration of the solvent into the fiber structure.
’ CONCLUSIONS Two different cellulose sources were analyzed in this study: pulps from annual plants (cotton linters, flax, hemp, and bagasse) and wood pulps (softwood mixtures, BWSP, and EWKP). In the course of dissolution, changes in the shape of the MMD were observed for some pulps investigated. Annual plant pulps tend to change their MMDs over dissolution time, whereas celluloses derived from wood fibers maintain the initial shape of the MMD throughout the dissolution process. One exception is bagasse pulp, which showed a behavior similar to the woodderived pulps. The change of MMD in pulps from annual plants indicates that initially lower-molecular-mass fractions (cellulose and hemicellulose) are preferably dissolved, which are followed later by higher molecular fractions mainly of cellulose. It can be concluded that dissolution begins in the less-ordered regions of the cellulose fibers, and only later will the crystalline regions of the fiber that are rich in cellulose and contain longer cellulose chains be affected. The path of dissolution was further elucidated by labeling uronic acids that are specific markers of xylans. Xylans are preferentially dissolved in the course of dissolution in hardwood pulps and annual plant celluloses. Navard and Cuissinat assumed that the dissolution behavior was controlled by the physical and chemical organization of the cellulose fiber itself17,18 and did not link the different dissolution behavior to the length of the cellulose chain or to the nature of the solvent. Our study shows that in general the fiber source and fiber crystallinity correlates with the time needed to reach a constant state of dissolution: annual fiber celluloses with high crystallinity need more time to reach a constant MMD, weightaverage molecular mass (Mw), recovered mass, and uronic acid content than wood plant celluloses with low crystallinity. The conformation plots additionally underline the clear differences between the solution behavior of wood pulp and annual plant
pulp samples. All annual plant pulp samples tend to have a higher ν value, indicating higher stiffness of the molecules in solution. For most pulps, the dissolution time can be significantly shortened. This applies to wood pulps obtained after different pulping procedures, which require ∼2 h but also to the less wellsoluble annual plant celluloses, which require about 2-6 h. Sulfite pulps dissolve even faster. More than 1440 min (24 h) of dissolution time did not improve the results of the GPC analysis in any of the pulps investigated. For the analysis of a high number of samples, as required in biorefinery analytics, a quick screening, making use of dissolution times no longer than necessary, could be helpful to reduce significantly analysis time and sample throughput for GPC analysis of cellulosics.
’ ASSOCIATED CONTENT
bS
Supporting Information. Characterization of different pulps (data partially provided by the producers), crystallinity of the pulps as determined by cross-polarization magic-angle spinning NMR (CPMAS NMR),12-14 and ν-values derived from the conformation plots (GPC-MALLS) for wood and annual plant cellulose samples. This material is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected]. Tel.: þ43-1-47654-6071. Fax: þ43-1-47654-6059.
’ ACKNOWLEDGMENT We thank Sandra Schlader, Gerhard Zuckerst€adter, and Gabriele Schild (Lenzing AG) for taking SEM pictures and for the determination of crystallinity, Susanne M€oderl (Lenzing AG) for compilation and analysis of pulp properties, and Sonja Schiehser (BOKU Vienna) for skillful lab work (GPC analysis) and data evaluation. ’ REFERENCES (1) McCormick, C. L.; Callais, P. A. Polymer 1987, 28, 2317–2323. (2) McCormick, C. L.; Callais, P. A.; Hutchinson, B. H. R. Macromolecules 1985, 18, 2394–2401. (3) Jerosch, H.; Lavedrine, B.; Cherton, J.-C. J. Chromatogr., A 2001, 927, 31–38. 878
dx.doi.org/10.1021/bm101555q |Biomacromolecules 2011, 12, 871–879
Biomacromolecules
ARTICLE
(4) Potthast, A.; Rosenau, T.; Sixta, H.; Kosma, P. Tetrahedron Lett. 2002, 43, 7757–7759. (5) Ishii, D.; Tatsumi, D.; Matsumoto, T. Biomacromolecules 2003, 4, 1238–1243. (6) Schelosky, N.; R€oder, T.; Baldinger, T. Papier (Bingen, Ger.) 1999, 12, 728–739. (7) Dupont, A.-L. Polymer 2003, 44, 4117–4126. (8) Dupont, A.-L.; Mortha, G. J. Chromatogr., A 2004, 1026, 129–141. (9) Sundholm, F.; Tahvanainen, M. J. Chromatogr., A 2003, 1008, 129–134. (10) Engstrom, A.-C.; Ek, M.; Henriksson, G. Biomacromolecules 2006, 7, 2027–2031. (11) Bohrn, R.; Potthast, A.; Schiehser, S.; Rosenau, T.; Sixta, H.; Kosma, P. Biomacromolecules 2006, 7, 1743–1750. (12) Larsson, P. T.; Wickholm, K.; Iversen, T. Carbohydr. Res. 1997, 302, 19–25. (13) Newman, R. H. Holzforschung 1999, 53, 335–340. (14) Wickholm, K.; Larsson, P. T.; Iversen, T. Carbohydr. Res. 1998, 312, 123–129. (15) Schulz, L.; Seger, B.; Burchard, W. Macromol. Chem. Phys. 2000, 201, 2008–2022. (16) Nicolas Le, M.; Emilie, M.; Catherine, P.; Herman, H.; Patrick, N. Macromol. Symp. 2008, 262, 65–71. (17) Cuissinat, C.; Navard, P. Cellulose 2008, 15, 67–74. (18) Cuissinat, C.; Navard, P.; Heinze, T. Carbohydr. Polym. 2008, 72, 590–596 .
879
dx.doi.org/10.1021/bm101555q |Biomacromolecules 2011, 12, 871–879
