MASTER'S THESIS
Characterisation of Solubility and
Aggregation of Alkaline Extracted Plant Cell Wall Biopolymers
Elizabeth Hagbjer
Master of Science in Engineering Technology Chemical Engineering Design
Luleå University of Technology
Department of Civil, Environmental and Natural Resources Engineering
Characterisation of Solubility and Aggregation of Alkaline Extracted Plant Cell Wall Biopolymers
Elizabeth Hagbjer
Master of Science Chemical Engineering
B7002K: Biochemical and Chemical Process Technology
Supervisors: Dr. David Hodge & Ulrika Rova
August 2012
Abstract
Up to 30% by mass of plant cell walls are comprised of hemicelluloses. The remainder is comprised of cellulose, lignin and extractives. Potential economic uses of hemicellulose include hydrogels, fibre additives in pulp mill paper-making and as a substrate for fermentation processes.
Development of a fermentation process with sugars from hemicellulose has become of increasing interest due to their potential as a feedstock for fermentation-based liquid fuels and other bio-based chemicals. These can be incorporated into existing processes, in particular alkaline chemical pulping mills, where up to 50% of the hemicelluloses are today degraded and eventually combusted.
The main objective of this project is to examine the solubility and aggregation properties of xylans (the predominant hemicellulose), as this will hopefully lead to better solubility-based separations for their recovery. This was done at Michigan State University by alkaline extraction at 85°C on milled birch wood, and at 130 and 170°C (both time-dependent) on birch chips, with 50 g/L sodium hydroxide. This was then followed by precipitation/aggregation experiments with ethanol, polyDADMAC (a polycationic flocculant) and by acidification. Characterisation was done by performing dynamic light scattering (DLS) and size exclusion chromatography (SEC) analysis on resolubilised recovered material from the different extraction conditions. From these, size distributions, molecular weights and degrees of polymerization (DP) could be estimated. The DP values for the extracted polymers were higher than the expected values for hardwood xylans, owing to the incoherent SEC chromatograms. This may be due to aggregate formation with other polymers or re-solubilisation issues of the hemicellulose precipitates. The estimated size range for model xylan was between 100 to 300 nm and the ethanol precipitates seemed to also lie around this region, as detected by DLS. One of the major factors contributing to the difficulty of analysing the results was the issue of re-solubilisation of the hemicellulose precipitates and flocculates.
Preface
The work for this thesis was conducted in the spring of 2011 at Michigan State University (MSU), East Lansing, Michigan, USA.
I'd like to thank first and foremost my supervisors at MSU and Luleå University of Technology (LTU). Thanks to Dr. David Hodge for your guidance and help with this work and for organising my coming to Michigan. Thanks to Ulrika Rova for your help and support throughout this process.
I'd like to thank the friendly staff and students at MSU for all their help in the lab and for making me feel welcome.
Last but not least, I'd like to thank all my family and friends for their love and support throughout this work.
Contents
1 Introduction ... 5
1.1 Background: Forest Biorefineries and Extraction Techniques ... 5
1.2 Hemicelluloses ... 6
2 Materials and Methods ... 7
2.1 Material ... 7
2.2 Dynamic Light Scattering ... 7
2.3 Characterisation and Composition of Biopolymers... 9
2.4 Alkaline Extraction ... 10
2.5 Ethanol Precipitation Tests ... 10
2.6 pH Adjustments ... 11
2.7 Polydiallyldimethylammonium Chloride (PolyDADMAC) Flocculation ... 11
3 Results ... 12
3.1 Extraction and Recovery ... 12
3.1.1 Composition Analysis ... 12
3.1.2 Precipitation Yields ... 13
3.2 Biopolymer Characterisation ... 14
3.3 DLS ... 16
4 Conclusions and Discussion ... 17
5 Future Work & Recommendations ... 19 References
Appendix
5
1 Introduction
1.1 Background: Forest Biorefineries and Extraction Techniques
A biorefinery utilises processes involving plant lignocellulose conversion in order to produce different end-products, all incorporated in the same plant [1]. The future goal being to phase out fossil fuel based chemicals and replace them with those produced from renewable and sustainable feedstocks. Finding a way of incorporating them into pulp and paper facilities has also been of interest in order to minimise capital costs and raise revenue of existing mills. Being able to integrate into existing processes also gives these so-called forest based biorefineries an advantage over agricultural based ones.
A suggested implementation of a biorefinery into an existing chemical plant is by adding an additional pre-treatment to the wood chips prior to kraft pulping [2], thereby extracting the hemicellulose that makes up part of the plant cell wall. This has the added benefit of separating out the extracted xylans (for further hydrolysis/depolymerisation) from the high concentration of toxic by-products present during chemical pulping. Hemicellulose has been of recent interest due to their source of xylose and other sugars [3] that make up the hemicellulose xylan polymer. These sugars in turn could be utilised as fermentation feedstock for fermentation-based liquid fuels and other value-added chemicals. The fact that hemicellulose is underutilised during a chemical pulping process also makes it a good candidate as a renewable resource [3]. One method of extracting hemicellulose is by an alkaline pre-treatment of the biomass (Fig. 1). The extracted xylan can then be precipitated out of the liquor by addition of alcohol, or other solvents, and recovered by e.g.
filtration. It is the isolation and separation of the extracted hemicellulose that has been the main focus of this study.
Figure 1: Possible process pathway for fermentation-based fuel and chemical production.
Aggregation and solubility experiments have been performed in order to develop improved solubility-based recovery. Characterisation was done using size exclusion chromatography (SEC), dynamic light scattering (DLS) and by calculating precipitation yields. Thus obtaining estimates of number average degree of polymerisation (DP), hydrodynamic radius (approximate size ranges) and approximate extracted amounts of hemicellulose respectively. Hemicellulose from silver birch (Betula pendula) was isolated under different extraction conditions. Extractions were done on milled wood chips in a water bath at 85ºC and in a lab-scale digester at 130ºC and 170ºC. Silver birch is a common commercial hardwood from the northern part of Sweden and wood chips for this study were provided from Smurfit Kappa in Piteå, Sweden. Different methods were used in order to
6
recover the extracted hemicellulose, including: precipitation with ethanol (EtOH), addition of polydiallyldimethylammonium chloride (polyDADMAC, a polycationic flocculant) and acidification with H2SO4.
1.2 Hemicelluloses
Hardwoods consist mainly of cellulose (approximately 45%), hemicellulose (approximately 30%) [3], lignin (approximately 25%) and extractives (around 5%). Hardwood hemicellulose (birchwood, in particular) in turn consists mainly of O-acetyl 4-O-methyl glucuronoxylan polymers with 1 → 4 linked ß-D-xylopyranose backbones. Acetylation of the 2 and 3 carbons can occur, and contain either 4-O-methyl glucuronic acid or only a glucuronic acid substitution in an α-1→ 6 linkage (Fig.
2). On average, the mole ratio between glucuronic acid to xylan is 1:10 [3]. These side chains also give rise to the polymer acting as a weakly charged electrolyte. The amorphous structure of hemicellulose makes it less resistant to hydrolysis. Despite this, it can still undergo hydrogen bonding between chains, but not to the same extent as cellulose. Published works state that they have a degree of polymerisation that ranges from 185 to over 230 [4]. The acetyl groups of the xylan chains contribute to their insolubility. It is also believed they prevent aggregation by averting hydrogen bonding between chains. Aggregate formation may also be due to hydrophobic stacking of the polymers. Decreasing aggregate formation should thus lead to increased solubility of the polymer chains.
A large portion of the hemicelluloses are degraded during alkaline chemical pulping [5]. Dislodging the xylan from the cell wall may however be impeded by the presence of lignin-carbohydrate complexes [6] and their ester and ether linkages. It is believed that hydrogen bonding between the polysaccharides also contributes to difficulty in removing the xylan components in hemicellulose, as well as causing aggregation between extracted polymers. However, by implementing a pre- treatment step, it is possible to retrieve a significant portion of the xylans. These xylans can then be further depolymerised (by hydrolysis) to xylose, a potential feedstock for fermentation. Most extractions for recovering xylans from biomass occur in one or two-step procedures, the two-step involving a delignification step, followed by an oxidative/alkaline step. Therefore, if there is no delignification step, some of the recovered xylan may contain significant amounts of lignin aggregates and other undesired extractives. Here a one-step extraction has been studied and applied, using an alkaline sodium hydroxide solution (50 g/L) at different temperature conditions. Using alkaline solvents has the added benefit of impregnating cooking alkali into the wood chips [7], before chemical pulping.
Different methods [8, 9] of recovering the xylans have been carried out, for instance, precipitation by lowering the pH and by addition of ethanol (up to 66% (v/v)) and by the addition of a flocculating agent (polyDADMAC). In the case of ethanol precipitation, the recovered, dried pellet is very hard to resolubilise. This is thought to be in part due to strong inter- and intramolecular hydrogen bonds that form during the drying process (pellet dried in oven overnight at 105ºC), facilitated also by the presence of larger unsubstituted sections of the xylan chains (causing interchain aggregation through strong hydrogen bonding [10]). Another theory is that covalent bonding between polysaccharides and lignin/other undesirable extractives through ether, ester and Figure 2: Structure of hardwood hemicellulose, O-acetyl 4-O-methyl glucuronoxylan.
7
carbon-carbon linkages [11] also contribute to dissolution problems. This difficulty in isolation of the xylans makes it challenging to properly evaluate their solubility and aggregation properties.
Easy recovery of the extracted xylans is also important from a processing point of view, in order to efficiently remove the xylans for further degradation into monomer sugars. An alternative to using alkaline solvents is steaming or hot water treatments of wood chips. This method solubilises xylans of lower DP and MW (further depolymerisation), whilst preserving some acetylation [10]. Acidic treatments can also be used and also seem to preserving some acetylation, whilst enhancing degradation of the xylan [12]. Reported molecular weights of hardwood lignin-glucuronoxylan complexes have been in the range ~15.7·103 g/mol [8].
2 Materials and Methods 2.1 Material
An abundant hardwood that is extensive throughout Europe is silver birch (Betula pendula). A deciduous tree, classed amongst the dicotyledons, it reaches around 15 – 25 metres in height, with stems approximately 40 cm in diameter. The crown has drooping branches with wind-pollinated catkins and the bark is typically white with dark patches, especially around the base. In the northern part of Sweden it is one of the main components in forestry and paper-making plants. For instance, at Smufit Kappa in Piteå, Sweden where birch-wood is pulped and processed into containerboard, for the production of corrugated cardboard. The birch-wood used throughout this study was provided by Smurfit Kappa, Piteå (12/2/2010) with an average moisture content of 4.8-wt% (after air drying). These were then milled in a Wiley MiniMill to <2 mm for the 85ºC extraction, but maintained as chips for the other two conditions (130ºC and 170ºC).
2.2 Dynamic Light Scattering
Dynamic light scattering (DLS) tests were performed on a 90 Plus/BI-MAS from Brookhaven Instuments. The device hardware consisted of a 660 nm laser with a Perkin Elmer ADP detector.
Received data from the detector was then processed using MAS OPTION particle size analysing software program.
Light, when impinged on matter, gives rise to an electric field that induces an oscillating polarisation of the electrons in the molecules of the substance. This created secondary source of light from the molecules then radiates (scatters) the light [13]. Light that scatters in all directions, having translational and rotational degrees of freedom (thus giving rise to frequency shifts) is referred to as Rayleigh scattering and gives rise to an observed fluctuation in intensity which is time-dependant. The cause of this fluctuation is due to the fact that molecules in solution are in constant motion, so called Brownian motion, a mathematical model used to describe the random movement of particles suspended throughout a fluid. The frequency shifts, angular distributions and intensity of the molecules can then be determined by the size, shape and molecular interactions (e.g.
diffusion properties) in the scattered light. In a typical set up, light from the laser is directed towards a sample, scattered and then received by a detector, placed at a certain angle relative to the direction of the incident beam. The detector (in this case using the quasi-elastic laser light scattering method) registers the scattered waves as an average intensity with superimposed fluctuations. The data [14]
is then sent to an autocorrelator where the dynamic information is derived from an autocorrelation
8 of the intensity trace according to:
g2(q; τ)=[ I (t) I (t+τ)]/[ I (t)]2 (1)
which gives a second order autocorrelation curve. The function describes the autocorrelation at a particular wave vector, q, and delay time, τ as a function of intensity, I. The longer the time delay, the more the correlation decays exponentially, and this in turn can be related to the motion of the particles. By assuming certain distributions, numerical methods can be applied to fit the exponential decay. In order to relate the second order autocorrelation function to the first order autocorrelation function, the Siegert equation is used:
g2(q; τ)=1+β[ g1(q ; τ)]2 (2)
Where β is a correction factor approximately equal to the inverse of the number of speckle from the light collected, and relates to the geometry and alignment of the laser beam. Data generated from the autocorrelation function can then be analysed to determine for example, a size distribution.
By assuming a monodisperse solution, the first order autocorrelation function can be treated as a single exponential decay:
g1(q; τ)=e(−Γ τ) (3)
where Г is the decay rate. This then leads to the determination of the translational diffusion coefficient from:
Γ=q2Dt (4)
where q is:
q=[(4 π n0)/λ]sin (θ/2) (5)
λ is the wavelength of the incident beam of the laser and n0 is the refractive index of the sample.
The angle θ, is the angle between the detector and the sample cell. Depending on the wave vector q, Dt can be determined from a single angle, θ or from multiple angles.
The Stokes-Einstein equation, along with the calculated Dt is used to calculate the hydrodynamic radius of a sphere (i.e. a polymer, or polymer aggregates) according to:
Dt=(kBT )/ π η r (6)
where kB is Boltzmann's constant, T is the absolute temperature, η is the viscosity of the solution and r is the radius of a sphere. The hydrodynamic is the “apparent” or effective size of the molecule, based on the diffusion properties of the particles in the solution, assumed to be hard spherical particles in constant motion. This is why different determined size distributions can be observed for the same solution sample. Electrostatic forces between ions and collisions between particles interfere with scattering analysis and can be minimised by dilution of the solution and by the additions of salts (in this case NaNO3).
The samples for DLS analysis were prepared by adding 0.01M NaOH and 0.1M NaNO3 solutions to the recovered material and then heating and soniccating for 1 hour. After cooling to room temperature, the samples were then diluted and filtered through 0.22 µm syringe filters. 3 mL of solution were then placed in a sample cell and analysed at 660 nm and detected at 90º with respect to the sample cell. The viscosity of the solution was assumed to be the same as water (due to high
9
dilution factor). The resulting tests thus produced size distribution graphs, autocorrelation curves and summary data of these.
2.3 Characterisation and Composition of Biopolymers
The composition of residues and precipitates was analysed according to the National Renewable Energy Laboratory (NREL) procedure “Determination of Structural Carbohydrates and Lignin in Biomass” [15]. A weighed amount of sample was transferred to a pressure tube and 72 wt-% of sulphuric acid was added. This was then set in a 30ºC water bath and stirred every 5-10 minutes, for 1 hour. Deionised water was then poured into the tubes before they, along with a set of sugar recovery standards, were sealed and autoclaved for 1 hour at 121ºC. The sugar recovery standards consisted of arabinose, glucose and xylose prepared at 0.1, 0.5 and 1 g/L. After autoclaving, the samples were filtered, the filtrate saved for monosaccharide detection HPLC analysis on a HPX- 87H Aminex column from Bio Rad, and the residue dried and weighed to determine the acid insoluble lignin content.
Size exclusion chromatography (SEC) was performed on a Waters Ultrahydrogel 250 column with RI and UV detection at 280 nm. SEC separates molecules in solution according to their size, as opposed to other common modes of chromatographic separation, for instance adsorption or ion exchange [16]. Porous particles make up the interior of the column, thus molecules smaller than the pore size of the particles will be trapped inside the pores and take longer to pass through. The eluate, the solution filtered through the column, is collected at the end of the column in constant volumes at different times and is referred to as fractions. These fractions are then presented as UV/RI vs. time chromatograms, giving a size distribution resembling Gaussian peaks. A baseline is then drawn below these peaks in order to estimate DP.
For the presented results, the mobile phase (eluate) consisted of 0.1 M NaNO3 and 0.01 M NaOH.
Precipitate samples from different conditions were taken and dissolved by heating and sonication in mobile phase, filtered and then diluted into 1 mL glass vials. SEC analysis of molecular weight distributions rely on the assumption that size effects are the only contribution to separation, and that there is no interaction between polymers. It is also assumed that the polymers will generate a signal that is proportional to their mass, and that the dextran calibration standards translate to xylan polymer sizes. From this it was then possible to estimate degree of polymerisation (DP) and molecular weight data of the extracted, precipitated and flocculated material. This was done first by indexing the initial data (from the received signal) in order to compress the amount of data. The retention times were then converted to retention volumes by multiplying with the flowrate (0.6 mL/min). The height from the baseline to the curve points were then calculated by taking the respective response values and subtracting the most negative, outlying response value from them. A DP was then estimated by the logarithmic linearization of the dextran standards and the conversion factor of 132 g/monomer subunit for a xylan polymer. Height was then multiplied by DP, and DP squared and summarized. These summations were then used to estimate DP based on number and weight.
10
2.4 Alkaline Extraction
All extractions were performed on silver birch wood chips supplied by Smurfit Kappa (Piteå, Sweden), with a moisture content of 4 – 5 wt-%. This was determined by drying weighed amounts of biomass at 105ºC for 24 h in tarred, aluminium trays and taking the mass loss as the moisture fraction of the sample. Alkaline extractions were carried out with a sodium hydroxide (NaOH) concentration of 50 g/L at 85, 130 and 170ºC. All conditions had a liquor to wood ratio of 20:1, the 85 and 130ºC being held at their respective extraction conditions for 1 hour, pH being recorded both before and after extraction.
For the 85ºC experiments, a milled sample (<2 mm fraction) of the silver birch wood chips was added, according to the desired loading, and taking into account moisture content. The samples were deposited into empty 50 mL Falcon tubes. Deionised water and NaOH solution were pipetted into different tubes and pH measured before being added to the biomass. The tubes were subsequently placed in a water bath at 85ºC for 1 hour. Tubes were then placed in an ice bath for 10 minutes before being centrifuged at 4000 rpm for 10 minutes. After the supernatant was decanted and collected, the procedure was repeated with deionised water until the pH of the wash water was around neutral. The biomass residue was then transferred to dried, weighed aluminium trays and dried overnight at 105ºC (see Fig. 4).
For the 130 ºC and 170ºC extractions, a sample of wood chips was weighed out and placed into a lab-scale digester along with corresponding amounts of NaOH solution and deionised water. For the 130ºC extraction, the heat was ramped from 25ºC to 130ºC at 1.5ºC/min for 70 minutes. The temperature was then held at 130ºC for 1 hour, whereby the extraction liquor was collected and the chips washed in a bucket of deionised water until pH was neutral. The wood chips where then placed in a 105ºC oven until dry. The ramp rate for the 170ºC extraction was 0.75ºC/min until 100ºC (100 minutes), and then 0.5ºC/min until 170ºC (140 minutes) was reached. The cook then ended with no more heat being applied, and let to cool to room temperature. One litre of the extraction liquor was then collected and the residue wood chips washed and handled similarly to the 130ºC extraction. The collected supernatants from all extractions were used to determine the recovery of the solubilised, extracted hemicellulose and were refrigerated until used. It should be noted however, that the total amount solubilised material for the 170ºC extraction was based on a different cook, the only difference between the two trials being that the second cook was held at 170ºC for 1 hour.
2.5 Ethanol Precipitation Tests
Ethanol (EtOH) precipitation is one of the most effective ways to determine how much can be extracted from an alkaline pre-treatment of biomass. The tests were carried out with 200 proof ethanol, the resulting precipitates being then used for DLS, composition analysis and weight-based yield calculations. A measured volume of extraction liquor was poured into marked Falcon tubes and a corresponding amount of ethanol was added according to the desired volume percentage (33- vol% to 66-vol%). The sample was then stored at -4°C for 2 hours, then decanted and washed and centrifuged 5 times with ethanol at 4000 rpm. The precipitated pellet (which is the xylan-rich fraction) was collected and dried in an incubator (40°C) overnight, then stored in a desiccator before weighing.
11
2.6 pH Adjustments
Another way to precipitate hemicellulose out from an alkaline solution is by lowering the pH. This is in reality a form of lignin recovery where the solubility of lignin is decreased, thus increasing aggregation. Here it was done by addition of 72-wt% sulphuric acid solution to the extraction liquor.
The extraction liquor was poured into Falcon tubes and pH was measured. Sulphuric acid (H2SO4) was then added until the desired pH was reached. The resulting precipitate was then centrifuged and washed first with deionised water and then with ethanol. The pellet was collected and left to dry overnight in an incubator at 40°C before being weighed. The recovered material was then used for DLS, composition analysis and weight-based yield estimates.
2.7 Polydiallyldimethylammonium Chloride (PolyDADMAC) Flocculation
Cationic polyelectrolytes are used widely throughout the pulp and paper industry for retention and de-watering purposes. Polydiallyldimethylammonium chloride (polyDADMAC) is a high-charged cationic homopolymer, that is, a repeating polymer chain consisting of the same fundamental monomer (Fig. 3). This makes it a good candidate for flocculation of xylans, due to the weak polyelectrolyte behaviour [17] of their glucuronic substitutions. Molecular complexes are formed between the cationic polymers and the wood-based anionic polymers [13], and could thus be used to separate them out from the extraction solution. A 20-wt% solution of polyDADMAC from Sigma Aldrich was used with low molecular weight (MW average between 100 000 and 200 000). The solution was then diluted to 0.5% polyDADMAC, this being within the recommended range for application according to the manufacturer.
A recorded amount of extraction liquor was poured into an empty Falcon tube. The pH was measured and 0.5-vol% of polyDADMAC solution was added according to the desired concentration (around 0.0025mL polyDADMAC/mL extraction solution). There being no significant difference between yields for higher concentrations than lower concentrations of polyDADMAC, a 33-vol% of 0.5-vol% polyDADMAC concentration was ultimately deemed adequate. The pH was then adjusted with 72-wt% sulphuric acid, until pH was around 5 – 6. The sample was then centrifuged at 5000 rpm for 30 minutes and washed with ethanol, before being poured into aluminium trays and dried overnight in an incubator (40°C). The resulting aggregates were then subjected to DLS, SEC and weight-based yield analysis.
Figure 3: Basic polyDADMAC monomer.
12
3 Results
3.1 Extraction and Recovery 3.1.1 Composition Analysis
Composition analysis of the 130ºC ethanol precipitates (50g/L NaOH) show that around 8 – 13 wt-
% were composed of xylose (Table 1). This gave an average xylan content of 10.8 wt-% for the precipitates and an overall xylan content of 3.8 wt-% of total solubilised material. No other sugars (e.g. arabinose or glucose) were detected.
UV spectrometry also confirmed the previously stated fraction of lignin found in hardwoods (Table 2). Around 20% of lignin was detected in the biomass residue.
Table 1: Xylose fraction for ethanol precipitated material (130°C).
Yields from different added amounts of ethanol.
Sample Xylose fraction (wt-%)
EtOH 33% (v/v) 13.6
EtOH 40% (v/v) 11.1
EtOH 57% (v/v) 8.8
EtOH 67% (v/v) 8.0
Average xylose content of
above precipitates 10.4
Xylose content of total
material solubilised 3.8
Table 2: Lignin content of 85C biomass residue (after extraction).
Sample
%Acid souble lignin (%ASL)
%Acid insoluble lignin
(%AIL) %Total lignin content
85°C biomass residue 0.99 18.17 20.14
130°C biomass residue 0.97 17.28 18.25
13
3.1.2 Precipitation Yields
As can been seen from Figure 4, the amount of solubilised polymers (xylan and lignin) that can be recovered, decreases with increasing temperature. The total solubilised amount however, seems to increase with increasing temperature.
Figure 4: Recovery yields for all conditions and recovery methods, compared to total amount solubilised material.
More extracted material seems to be able to be recovered from the 85ºC, than for the 130ºC and 170ºC extraction liquors. This is may be due to the xylan being subject to more degradation at higher temperatures and the longer residence time in the reactor due to heat-up. Also, some of what is being recovered is thought to be organics (Na+ and SO42-
), rather than polymers.
It seems that the greatest amount of material is recovered by pH adjustment, except for the 170ºC condition. Here polyDADMAC seemed to recover the most material, whereas ethanol and polyDADMAC seemed to recover similar amounts for the other two conditions.
14
3.2 Biopolymer Characterisation
All the recovered material gave strange chromatograms (Fig. 6 – 7), which resulted in strange DP and molecular weights for both the ethanol precipitated and the polyDADMAC flocculated samples.
This could be due to solubilisation issues of the samples, as the recovered material was difficult to resolubilise. The chromatograms therefore only depict the xylan fractions that were soluble in solution. The molecular weights for all conditions seemed to be higher than those exhibited by previously published materials [10, 8]. Ranging from 21 000 g/mol (for 170ºC EtOH precipitates and polyDADMAC flocculates) to 70 000 to 100 000 g/mol (for 85 and 130ºC precipitates and flocculates), according to dextran calibration. This could be due in part to the presence of supramolecular structure with high molecular weight components within the material being collected. It could also be due to the presence of contaminants affecting the overall assessed weight.
The same problem occurred for the subsequent DP values, and have therefore been omitted due to their unreliability. Figure 5 showed a single, clear peak for the model birch xylan, within the calibration range (10.5 – 15 minutes). The data from the SEC plots was difficult to analyse, due to incomplete separation within the calibration range observed for most of the samples, as can be seen in the figures below (Fig. 6 and 7). Figure 8 is the calibration standard used for the molecular weight.
Figure 5: SEC plot of model birch xylan (normalized data).
15
Figure 6: SEC plot of ethanol precipitates (normalized data).
Figure 7: SEC plot of polyDADMAC flocculated material (normalized data).
Figure 8: SEC standard for dextran, for determination of molecular weight.
16
3.3 DLS
The dynamic light scattering results showed a broad size range distribution for the various samples.
Most samples exhibited two clear peaks, one believed to represent the “single chain polymers” and the other the aggregated xylans (in accordance with [17]). The polyDADMAC flocculated materials seemed to exhibit the largest overall size averages, especially for the aggregate peak. This could be due to polyDADMAC being a fairly large polymer in itself and thereby increasing the overall average, as well as any other contaminating extractives also adding to the overall size. Figure 9 shows a comparison between model birch xylan and the 130ºC recovered material. The other conditions have been omitted as there were different soluble yields for each condition, the 130ºC results giving a good indication of the overall trend. Due to the great variation in results, DLS was deemed to be more a method of general approximation than a reliable source of quantification.
Figure 9: DLS comparison of model xylan, ethanol precipitated xylan and polyDADMAC flocculated xylan (both from 130°C extraction).
17
4 Conclusions and Discussion
The SEC distribution from the model xylan exhibits a distinct peak corresponding to around 70 000 g/mol (Fig. 5), according to the dextran standard. This high molecular weight is attributed to aggregation formation of the model xylan during dissolution. This peak therefore probably includes both single polymer chains as well as aggregates of xylan. SEC analysis gave molecular weights ranging from 21 000 g/mol (for 170ºC EtOH precipitates and polyDADMAC flocculates) to 70 000 to 100 000 g/mol (for 85ºC and 130ºC precipitates and flocculates), according to dextran calibration.
All of these are above previously published values (~15 700 g/mol) for hemicellulose [18]. This may be due to other extractives and contaminants (such as inorganics) being recovered along with the xylan, along with insufficient washing of the samples. But more probably due to the fact that only a certain fraction of the precipitates were able to be re-dissolved. So what is given is an approximation of the soluble fraction of the recovered material. Many of the precipitates exhibited long elution times, outside of the calibration range, which also seems indicative of dissolution problems with the precipitates. The extracted hemicellulose undergoes crystallisation when dried, which makes it extremely difficult to dissolve. The samples were heated and sonicated for 1 hour in order to try and overcome the re-solubilisation issue. However, more testing is needed in order to resolve this more efficiently. Had the recovered materials resolubilised more effectively, they would likely have given rise to two distribution peaks, representing perhaps single chain polymers and aggregates of xylan respectively.
The DLS results (Fig. 9) also seemed to correspond well to previously documented observations of single chain polymers and aggregates. As size ranges for xylans have not been very well documented by this method, the DLS was mainly used to see distributions (i.e. single chain polymer peak and aggregate peak), and the results seemed to show that these did indeed exist. Both the ethanol recovered and the model xylans exhibited similar distributions, and gave a size approximation of individual xylan chains as well as formed aggregates. The polyDADMAC flocculates from the 130ºC on the other hand, seemed to exhibit higher size ranges. It is therefore likely that only the larger polymer chains or aggregates were able to resolubilise from the flocculates. Also as polyDADMAC is in itself a fairly large polymer [13], this may also contribute to an overall increase in apparent size. The polyDADMAC from the 85ºC and 170ºC extractions displayed size ranges similar to the conditions presented in Figure 9, and have therefore been omitted for clarity. Here it would seem therefore that method of extraction from biomass seems to have an impact on the relative size of xylan opposite to what has previously been stated (i.e. higher degradation leading to smaller polymers at higher extraction conditions). Again, the measurable fraction of the samples is only the polymers that are able to resolubilise after recovery. This, along with the greatly varying results, deemed DLS more a tool for general approximation than for documentable analysis. This variation is probably due to the time-dependant nature of the analysis (see section 2.2).
The composition analysis of the birch residue (for 85ºC and 130ºC extractions) showed that the residue comprised about 20% lignin (after extraction). As hardwood contains 25% lignin, most of the lignin has remained in the biomass, and has therefore probably not contributed much towards aggregation and solubility of the extracted xylan. No other sugars were detected besides xylose.
This seems to confirm that the main component extracted from the biomass was xylan. However, as the ethanol precipitates were composed of only 8 – 13 wt-% of xylose, it would seem that xylan is either not efficiently being recovered from the extraction liquor, or is still retained in the biomass after extraction.
18
Ethanol and polyDADMAC appear to have similar recovery amounts for all conditions except 170ºC (Fig. 4). This is most likely an error in calculation or washing of the flocculate, and is believed to follow the same trend as the other conditions. Following this trend, the precipitation yields for the pH adjustments seemed to recover the most material. This is probably due to other extractives, organic materials and polymers being recovered from the solution when lowering the pH, thus affecting the overall apparent weight of the recovered pellet. So in terms of recoverability, it would seem that lowering the pH is the best method, based on amount material recovered per gram dry wood. Although, most of the material recovered is thought to be inorganics (Na+ and SO42-
). Purity requirements of these recovered materials would also be an issue [19] when upscaling, depending on the accepted toxicity levels of the material (e.g. based on what type of organism is being used for further fermentation processing etc.).
As expected, the 170ºC extracted the most soluble amount of material, owing to higher degradation and longer residence time in the reactor. This condition also had the least amount of recoverable material, based on per gram dry wood. Again, probably owing to the further degradation of the polymers at this temperature as well as alkalinity.
It cannot be concluded whether any method of recovery is superior to the others. PolyDADMAC has the advantage of requiring only very small additions (around 2.5µL polyDADMAC/mL solution) in order to achieve the same recovery as ethanol. The main issue was in handling the recovered flocculates, as it was easily resolubilised back into solution. This was overcome with washing with ethanol.
The main concern throughout this study has been the resolubilisation of the collected material.
Samples were heated and sonicated in order to overcome this, but it was apparent according to the characterisation analyses that only a fraction of the material was able to be resolubilised. Further investigation is therefore needed in order to resolve this issue.
19
5 Future Work & Recommendations
Although the DLS for the extracted hemicellulose seemed to correspond well to the model xylan, the data was often erratic and would give different distributions for the same sample. The DLS was therefore deemed to be an unreliable form of analysis.
More testing is also recommended in order to solve re-solubilisation issues with hemicellulose precipitates and overcome the above-mentioned problems with crystallisation. More extensive testing will be done on different types of biomass under the supervision of Dr. David Hodge at Michigan State University, especially for conditions of extraction at 170 ºC.
20
References
1. Helmerius J, Integration of a Hemicellulose Extraction Step into a Forest Biorefinery for Production of Green Chemicals, Licenciate thesis, Luleå University of Technology, Luleå, Sweden, 2010.
2. Al-Dajani W.W & Tschirner U.W, Pre-extraction of Hemicelluloses and Subsequent Kraft Pulping Part I: Alkaline Extraction, Tappi Journal 7, 2008, 6, pp. 3 - 8.
3. Ebringerova A, Structural Diversity and Application Potential of Hemicelluloses, Macromol. Symp., 2006, 232, pp. 1 – 12.
4. Goring D.A.I. & Timell T.E, Molecular properties of Six 4-O-Methylglucuronoxylans, J Phy. Chem, 1960, 64, pp. 1426 – 1430.
5. Bikova T. & Treimanis A, Solubility and Molecular Weight of Hemicelluloses from Alnus Incana and Alnus Glutinosa. Effect of Tree Age, Elsevier, Plant Physiol. Biochem, 2002, 40, pp. 347 – 353.
6. Saake B, Kruse Th. & Puls J, Investigation on Molar Mass, Solubility and Enzymatic Fragmentation of Xylans by Multi-detected SEC Chromatography, Elsevier, Bioresource Technology, 2001, 80, pp. 195 – 204.
7. Helmerius J, Vinblad von Walter J, Rova U, Berglund K.A. & Hodge D.B, Impact of Hemicellulose Pre-extraction for Bioconversion on Birch Kraft Pulp Properties, Elsevier, Bioresource Technology, 2010, 101, pp. 5996 – 6005.
8. Roubroeks J.P, Saake B, Glasser W.G. & Gatenholm P, Contribution of the Molecular Architecture of 4-O-Methyl Glucuronoxylan to its Aggregation Behaviour in Solution, Hemicelluloses: Science and Technology, ACS Symposium Series, American Chemical Society, 2003, pp. 167 – 183.
9. Chaunliaud E, Saulnier L. & Thibault J.-F, Alkaline Extraction and Characterization of Heteroxylans from Maize Bran, Journal of Cereal Science, 1995, 21, pp. 195 – 203.
10. Ebringerova A, Xylan and Xylan Derivatives – Biopolymers with Valuable Properties 1, Macromol. Rapid Commun., 2000, 21 (9), pp.542 – 556.
11. Koshijima T. & Wananabe T, Association Between Lignin and Carbohydrates in Wood and Other Plant Tissues, Berlin, Germany, Springer, 2003.
12. Kabel M.A, Bos G, Zeevalking J, Voragen A.G.J. & Schols H.A, Effect of Pretreatment Severity on Xylan Breakdown of the Remaining Cellulose from Wheat Straw, Elsevier, Bioresource Technology, 2007, 98, pp. 2034 – 2042.
13. Lappan R.E, Pelton R, McLennan I, Patry J. & Hrymak A.N, Kraft Lignin-Poly(DADMAC) Precipitate Formation, Ind. Eng. Chem. Res, 1997, 36, pp. 1171 – 1175.
14. Bruce J.B. & Pecora R, Dynamic Light Scattering: with applications to chemistry, biology and physics, Toronto, Canada, Dover, 2000.
21
15. Sluiter A, Hames B, Ruiz R, Scarlata C, Sluiter J, Templeton D. & Crocker D, Laboratory analytical procedure, “Determination of Structural Carbohydrates and Lignin in Biomass”, NREL, 2011 ( http://www.nrel.gov/biomass/pdfs/42618.pdf ).
16. Mori S & Barth H.G, Size Exclusion Chromatography, Berlin, Germany, Springer, 1999.
17. Chanliaud E, Roger P, Saulnier L. & Thibault J.F, Static and Dynamic Light Scattering Studies of Heteroxylans from Maize Bran in Aqueous Solution, Elsevier, Carbohydrate Polymers, 1996, 31, pp. 41 – 46.
18. Chanliaud E, Saulnier L. & Thibault J.F, Heteroxylans from Maize Bran in Aqueous Solution. Part II: Studies of Polyelectrolyte Behaviour, Elsevier, Carbohydrate Polymers, 1997, 32, pp. 315 – 320.
19. Norman A.G, The Hemicelluloses, III. Extraction and Preparation, Biochemistry Section, Rothamsted Experimental Station, Harpenden, Herts, 1937, pp. 1579 – 1585.
22
Appendix
Standard curves used for SEC analysis of sugars:
