Contents

Objective of this study 1

Introduction 2 Wood 2 Kraft pulping 4 Paper strength 6 The action of xylan 10

Experimental 13

Results and Discussion 16 Xylan dissolution, degradation, and redeposition 16 Xylan precipitation onto cellulose fibres 18 Effects of xylan on tensile strength properties 21 Xylan substituent reaction kinetics 27 Cooking temperature and tensile strength 36 Method development for uronic acid quantification 39

Conclusions 43

Future Work 45

Acknowledgements 47

References 48 Objective of this study

The increase in tensile properties of a paper sheet due to xylan enrichment has been observed several times but is still not fully understood. This thesis deals with the mechanisms underlying this strength-increasing effect of xylan in kraft cooking and discusses the most important parameters governing the strength increase and the nature of xylan. The use of birch xylan as a dry-strength agent in the softwood kraft cook was investigated with the aim of developing a system with a high yield of very easy-to-beat pulp. This can result in a paper product possessing high tensile strength and stiffness as well as high bulk and bending stiffness.

1 Introduction

Wood Wood is a very heterogeneous material and its composition and structure differ depending on the species, individual, and location in the tree. The wood used as a raw material in pulp production originates usually from either softwood or hardwood trees. Although the diversity of wood structure is high, it always consists mainly of cellulose, hemicellulose and lignin together with small amounts of extractives, pectin and proteins.

Cellulose is a linear polysaccharide consisting of β-1-4 linked glucose units, with a degree of polymerisation of approximately 10 0001. Cellulose molecules are naturally organised into β-sheets, which in turn are organised in bundles of sheets called fibrils. The fibrils also contain other cellulose with the less-ordered, so-called paracrystalline structure. The cellulose fibrils form aggregates that build up the wood cell wall.

Hemicelluloses are found in the matrix between cellulose fibrils in the cell wall, and they have been shown to be closely associated with both cellulose and lignin. Xylan comprises a group of hemicelluloses that will be focused on throughout this thesis. Softwood xylans differ slightly in structure from those found in hardwoods, although both have a backbone of β-(1Æ4)-D-xylopyranosyl residues. Figure 1 presents a representative formula for O-acetyl-(4-O-methylglucurono)xylan, which is the main hemicellulose in birch, representing 27% of the dry mass of birch wood. Hemicelluloses are unlike cellulose branched molecules, and the degree of polymerisation of birch xylan is between 101 and 122 units2. There is on average one 4-O-methyl-α-D-glucuronic acid (MeGlcA) side group per 8−20 xylopyranosyl residues, introducing charges to the polymer.

2

Figure 1. A representative formula of glucuronoxylan.

Figure 2 presents a model of a wood cell with its different cell wall layers. The thin primary wall (P) and the middle lamella (ML) have high lignin contents but also consist of cellulose, hemicellulose and small amounts of extractives, pectin and proteins. The middle lamella is located outside the primary wall and holds the wood cells together. The fibril aggregates in the primary wall build up a network extending in random directions, as depicted in the figure. The secondary wall consists of three sub layers, in each of which the fibrils run in a specific direction. In the thin outer layer, S1, the fibrils run in a directly almost perpendicular to the fibre direction. The S2 layer is the most important one for the physical properties of the wood and thus for its papermaking properties. The fibrils are oriented at a small angle to the fibre direction, and this angle is different for different species and is very important for the mechanical properties of the fibres. This layer is very thick compared to the other layers, and more than 50% of the wood material is found in S2. The third sub layer in the secondary cell wall, S3, resembles S1 in terms of fibril direction and is the thinnest secondary sub layer. The wood cell shown in the figure is, in technical texts, frequently referred to simply as “wood fibre”, a usage that will be adopted in this thesis.

3

Figure 2. The wood cell according to Côté 3. Fibril aggregates are shown as dark lines.

Kraft pulping The aim of kraft cooking is to remove most of the lignin so that the fibres can be separated from each other and to prepare the pulp for subsequent processes. The degradation and dissolution of lignin is carried out in steel digesters at temperatures in the 140−170°C range and at pressures of 6−9 bar. The degradation of lignin into smaller soluble units is performed through chemical reactions involving the active species, the hydroxide ion and the hydrogen sulphide ion. The pulp is often subsequently bleached to a desired brightness. However, in this thesis, unbleached fibres have been studied exclusively.

The delignification of wood in kraft cooking is accompanied by the simultaneous dissolution of carbohydrates. Different carbohydrates are dissolved to different extents, and it is the balance between lignin and carbohydrate dissolution that determines the selectivity of the process. The rate of delignification in kraft cooking is controlled by three distinct reaction kinetics4-6. The observed delignification is the sum of the three kinetics, and at any given time during the kraft cook one of them is dominant. In the initial phase the rate of delignification is high, and almost 20% of the lignin is dissolved via initial phase kinetics. In the bulk phase the delignification rate is lower, and by the time this phase ends and the residual phase begins, approximately 90% of the lignin has been removed. The delignification rate in the residual phase is very slow and may even completely stop7. The kinetics of

4 carbohydrate dissolution are somewhat different. There are only two observed kinetic phases of carbohydrate dissolution, with a very high rate of dissolution in the first and a lower rate in the second. The transition point between the two phases comes very early in the cook, at least for spruce8. After this point the carbohydrate dissolution proceeds as the bulk delignification does, and the selectivity of the cook becomes relatively high. In terms of selectivity, it is important never to reach the point at which residual delignification starts to dominate. The energy of activation of the bulk phase delignification9,10 is lower than that of the second-phase carbohydrate dissolution8,11, which gives an increased selectivity when the cooking temperature is lowered.

In the 1950s it had already been shown that some of the dissolved carbohydrates, especially xylan, are relatively stable in the dissolved phase as whereas other hemicelluloses, such as mannans, are degraded into smaller units under kraft cooking conditions12,13. At that time, it was also shown that the dissolved xylans can be redeposited onto the fibre surface later on in the cook14,15. There is still today, a significant amount of xylan lost as polysaccharide and the possibility of using this dissolved part in the pulp, instead of as today; degrade it and burn it in the recovery boiler, is investigated in this thesis.

Kraft pulping is the dominant chemical pulp production process around the world and has been so for the past 50 years. In the 1970s and 1980s several important improvements were made to the process to increase its selectivity, meaning a higher delignification rate and reduced carbohydrate degradation16-18. In conventional kraft cooks, the chips and liquor were loaded into the digester at the same time, and heated under pressure for a certain length of time until the desired degree of delignification was achieved. This cooking method involves high alkalinity early in the cook, leading to severe carbohydrate degradation and low alkalinity by the end of the cook; this results in a low overall delignification rate and may even cause precipitation of lignin, as the solubility boundary of lignin in terms of pH, temperature, and ionic strength is very close to the conditions encountered in the kraft cook19,20. The findings made in the 1970s and 1980s have resulted in several

5 new industrial processes21. All of them involve the exchange of cooking liquor at certain points in the cooking process, in order to control the chemical concentration, ionic strength, and amounts of dissolved wood components in the liquor. These process changes made it possible to delignify the wood to a greater extent without affecting pulp strength or yield. However, in the fibre processing line, the cooking is still the least selective unit operation; it is now common to interrupt the cook at rather high kappa numbers and continue processing using the more selective oxygen delignification process for 1 or 2 reactors, followed by multistage bleaching.

Paper strength The strength of a paper is a multifaceted concept, as paper can be pulled, torn, folded back and forth, and compressed until it breaks − and all this in different directions. Depending on the manufacturing operations and the use of the end product, a particular set of quality requirements has to be fulfilled. One common quality requirement of many products is tensile strength, and many theories have been advanced regarding this property. Paper is a network of fibres, and its tensile strength depends upon the strength of the fibres themselves, fibre shape in the load direction, the contact area between fibres, and the fibre−fibre joint strength22 (i.e., the force it takes to pull apart two fibres with a given contact area). Some studies demonstrate the existence of intact fibres in the rupture zone23 while others show that a large proportion of the fibres are broken24. Whether the limiting factor of paper strength is related to the fibres themselves or rather to the fibre−fibre interactions is determined by the degree of bonding in the sheet.

Fibre strength is primarily determined by the nature of the raw material. Latewood fibres have a higher tensile strength than do earlywood fibres25-27, and spruce fibres are stronger than those of pine28. This difference can be attributed to the S2 fibril angle in the fibres and to the cell wall thickness. The greater the angle, the greater the tensile strength and tensile stiffness of the fibres28; moreover, if the cell wall is thicker, there is more material to bear the load. The stretch at break may be shorter

6 for fibres with a high fibril angle. However, fibre strength does not depend solely on the origin of the fibres, but also on the cooking and bleaching process used. The chemical composition greatly affects fibre strength, and increased hemicellulose content may give an increased fibre strength29.

When fibre−fibre joints are being formed during the consolidation of a paper sheet through pressing and drying, the flexibility of the fibre surfaces is of great importance. The fibres are swollen in water and shrink considerably when water is removed, especially in terms of thickness. At this stage, a flexible fibre surface will be able to connect and adapt its shape to the neighbouring fibres, giving the possibility of a high degree of contact. Dry strength agents are developed from this view to increase either the contact area between fibres or the fibre−fibre joint strength by modifying the fibre surface. The knowledge of today on paper strength and paper strength agents is well summarised in a recent publicated review 30.

The theory of paper tensile strength presented by Page22 relates fibre strength and fibre−fibre joint strength to the tensile sheet strength through theoretical relationships, as expressed in Equation 1.

1 9 12g = + (1) T 8Z bLα where T is the tensile strength of the sheet, Z is the fibre strength measured as zero span, g is the acceleration due to gravity, b is shear bond strength per unit bonded area, L is the fibre length, and α is the bonded area per gram of fibrous material.

The developed theory describes the relationships well, but suffers from some shortcomings when it comes to practical application. There are five main drawbacks, as follows. The first concerns sheet formation: the theory is valid only for sheets of perfect formation, and these are never encountered in practice. The formation of a sheet affects its tensile properties tremendously, as good formation will minimise the weak points of the sheet. A test strip always breaks at its weakest

7 point, which explains the strong effect of formation on tensile strength. Second, Page’s theory does not take account of kinks or other weak areas in the fibre itself, as fibre strength is assumed to be the same along the whole fibre length. The third drawback concerns the distribution of contact zones, a factor that is not included in the model. Two fibres connected by many small contact zones at different locations along the fibres may require a higher force to be pulled apart compared to two fibres sharing only one large contact zone of the same total contact area. Fibre shape is another important parameter not taken account of by the theory. Curled fibres do not actually bear any load until they have been straightened out, but Page assumes that all fibres are completely straight. These first four drawbacks can be seen as instances of idealisation, whereas the last drawback is even more serious. The relative bonded area between fibres, a parameter used in this theory, causes problems as it is determined spectroscopically. The wavelength of the light used in this method is far too long for measuring actual molecular contacts. This problem is depicted in Figure 3 where two fibres in contact are shown. The degree of contact estimated using this method will thus be overestimated, the error being presumably different for different fibre materials. However, this drawback has more to do with the analytical method than with the theory as such.

Partly joined area Area in molecular contact

Area not available for light scattering, d < λ/2

Figure 3. The relative bonded area is strongly overestimated by light scattering measurements. Reprinted with permission from Lindström et al.30

8 It might be possible to measure the contact area by means of irradiation with very small wavelengths, for example, by neutron scattering. However, no such method has yet been reported to be successful, mainly because of difficulties interpreting the generated spectras. Still, despite these problems, Page’s theory is, for engineering purposes, the most accepted one due to its simplicity and rationality. However, considerable research is being done in the area, and in the future, sophisticated network models may be further developed to the point of practical applicability.

To increase the tensile strength of the paper sheet, the pulp is beaten prior to the papermaking. Fibres in water experience a subtle balance between swelling forces, mainly originating in charged groups in the hemicellulose and lignin, and restraining forces located in the supramolecular structures of the fibrils. During beating, delamination of the fibre wall takes place31; this breaks down the internal structure of the fibrils, allowing the fibres to swell more. Internal and external fibrillation is also caused by beating, and these phenomena help create large contact areas between fibres. Formation of fines is another process taking place during beating, which enhances the fibre−fibre joint strength. Beating has also some negative effects on the pulp: it increases the time required for pulp dewatering, and may reduce the production rate on a pulp drier or a papermaking machine. The density of the formed sheet is increased by beating; this decreases the bending stiffness, an effect that has severe consequences for many paper products. The interaction of the formed sheet with water and moisture is increased by beating, reducing the dimensional stability of the product when it is subject to moisture. This property is of great importance for both packaging materials and printing papers. Hence, process changes that allow pulp to be produced with less beating, i.e., increased beatability, achieve great improvements in both the production process and product properties.

9 The action of xylan It has been known for many years that the tensile properties of pulp and paper can be enhanced by increasing the amount of hemicelluloses in the pulp32. Studies carried out to demonstrate this effect include; addition of hemicelluloses33-35, removal of hemicelluloses35-37, and the use of different cooking conditions to manufacture pulps with different amounts of hemicelluloses29,38,39. As xylan is relatively stable under kraft cooking conditions and has been identified as the hemicellulose affecting the fibre strength the most25,36, it is of great interest to increase the xylan content of fibres. The mechanism underlying this strength- increasing effect has yet to be fully understood. Leopold and McIntosh25 reported a decrease in individual fibre strength with decreased xylan content. In contrast, Molin and Teder29 observed no change in fibre strength when pulp was cooked under different cooking conditions to achieve different hemicellulose contents; they observed no increase in bond strength either, measured as the z-direction tensile index, although the tensile index of the sheet was increased. However, Schönberg et al.35 reported a clear trend in the Scott bond value with changing xylan content. The results from these studies are hard to compare, since the xylan content was varied in totally different ways. They can still be used to illustrate the general picture of xylan acting as a strength-increasing agent in some cases while producing no effect in others and our understanding behind this is lacking. The theory suggested early on was that xylan can act as a glue between fibres as well as increasing their swelling capacity38; both these effects have been used to explain the observed strength increase. Centula and Borruso40 related the increase in strength with increased hemicellulose content to the structure of the fibre surface rather than to the chemical composition. It has also been shown that exchanging cooking liquors, i.e., removing liquors at one point in the process and feeding them back at another, can have positive effects on pulp properties, such as decreased alkali demand in the cook, increased yield, and increased beatabilty41,42. It was recently shown that the strength-increasing effect of xylan is completely associated with the xylan located on the surface of the fibres, no correlation being found between the

10 xylan content of the inner part of the fibres and the sheet strength43. This idea was discussed early on15,32, but could not be confirmed experimentally until 2004.

The native form of birch xylan depicted in Figure 1 is modified during kraft cooking. The three degradation reactions − endwise peeling, alkaline hydrolysis, and secondary peeling − all decrease the molecular weight of all polysaccharides to different extents. In the alkaline media, the acetate groups are immediately hydrolysed. Furthermore, the MeGlcA side group on the xylan backbone is partly removed from the backbone. In the early 1960s methanol was detected as a degradation product of MeGlcA, and a reaction mechanism for the formation of hexenuronic acid was suggested44. The existence of this mechanism was later supported by reactions with model compounds45. It was however not until 1995 that the structure 4-deoxy-β-L-threo-hex-4-enopyranosyluronic acid (HexA) was identified in kraft pulps and black liquor46, a finding that stimulated great interest in HexA in pulping research. Such interest has remained strong, because of HexA’s contribution to the kappa number47, its reactivity towards bleaching chemicals48, and its strong involvement in the brightness reversion of pulp and paper49. Due to its negative effects, it is crucial to minimise the amount of hexenuronic acid created in the kraft cook. Some studies have considered the kinetics of hexenuronic acid formation and degradation, together with the degradation of 4-O-Me-glucuronic acid, resulting in the reaction scheme presented in Figure 4. The suggested conditions for minimising the HexA content of pulp often involve high temperature and high alkali levels50,51. Such kraft cooking conditions, however, also minimise the xylan content of the produced pulp. As already discussed, high xylan content is very important, both in terms of its contribution to pulp yield,42,52 and its ability to improve pulp strength25,29,35,38. What needs to be studied and optimised is the degree of substitution of the side-group hexenuronic acid on the xylan backbone, to produce a pulp with high yield, and high strength properties and low hexenuronic acid content. Furthermore, the amounts of carboxylic acid groups have been correlated with increased tensile strength and fibre−fibre joint strength53,

11 It is possible that highly charged xylan increases strength more efficiently than less- charged xylan does.

O k1 O O O O O - CH3OH HO O HO O

OH OH O O OH OH OCH HOOC 3 HOOC

k2 k3 O O

OH OH O O O O O OH OH OCH3 HO HOOC OH HOOC

MeGlcA Unsubstituted HexA xylose unit

Figure 4. The formation of hexenuronic acid from 4-O-methylglucuronic acid and the degradation of the two substituents of xylan.

12 Experimental

A more detailed description of the experiments is given in the related papers; only a general introduction and overview of the methods used is provided here.

The kraft cooking experiments were carried out using two different sets of apparatus. One consisted of autoclaves that were charged with chips and liquor and rotated in a temperature-controlled polyethylene glycol bath. It is possible to remove one autoclave from the bath and cool it while the cook is prolonged in the others. This system was used whenever many different cooking times were being compared, for example, in kinetic studies of black liquor xylan precipitation onto cotton fibres and reactions involving xylan substituents. The disadvantage of using such a set-up is the limited amount of chips that can be treated at a time (approximately 300 g OD). The other set-up consisted of a forced circulation digester, where the cooking liquor is circulated between the digester vessel and a steam-heat exchanger. A drawback of this system is that it yields just one batch of pulp in every experiment; on the other hand, the maximum chip capacity is over 2 kg and it is possible to exchange cooking liquors in this system − a central focus of this thesis.

A high liquor-to-wood ratio of 75:1 was used when studying the kinetics of the kraft cooking reactions in the autoclave system and in validation cooks in the circulation digester. By cooking at high liquor-to-wood ratios, the concentrations of cooking chemicals, dissolved wood components, and ionic strength can be held reasonably constant. When cotton fibres were cooked with birch black liquor, the liquor-to-fibre ratio was 1:10. For all other cooks, the liquor-to-wood was 4:1, which is similar to those used in industrial processing.

The softwood pulping experiments were performed using a cooking design including liquor exchange and a liquor-to-wood ratio of 4:1. The exchange of liquor was carried out when 1 hour of delignification time remained. The replacement

13 liquor was generally black liquor from a birch kraft cook, withdrawn at different points in the birch cook. Reference pulps were produced using identical cooking parameters, but using only white liquor as the replacement liquor. All pulps had a final kappa number between 17 and 19.

After cooking, all pulps were washed for 12 h in deionised water and defibrated in a water-jet NAF defibrator (Nordiska Armatur Fabriken). Kappa number (SCAN C:100), pulp viscosity (SCAN-CM 15:99), residual hydroxide ion concentration (SCAN N 33:94), and hydrogen sulphide ion concentration (SCAN N 34:96) were determined.

Xylan was isolated from birch black liquors according to the method set forth in Axelsson et al.54, which involves the neutralisation of the black liquor and the precipitation of xylan through the addition of ethanol. The amounts of Klason lignin in the wood, pulps, and precipitated black liquor xylan were determined using the TAPPI method, T222 om-83, while the carbohydrate composition was determined after acid hydrolysis, according to the method of Theander and Westerlund55. The hexenuronic acid contents of pulp and precipitated xylan were determined using the method described in Gellerstedt and Li56, which involves hydrolysis with mercury acetate, followed by oxidation, and finally condensation with barbituric acid, giving a coloured product suitable for HPLC separation and UV detection. The total uronic acid contents of the pulps and precipitated xylan were determined using the colorimetric method involving carbazole addition, as described in Filisetti-Cozzi and Carpita57. The light absorbance detection was carried out from 400 to 700 nm, and the absorbance at 525 nm was used for calibration with the D-galacturonic acid standard and then for quantifying the samples. After systematic study of the rationale of the carbazole method, the total absorbance at 525 nm was found to include the contributions of HexA and cellulose. Therefore, the total uronic acid content was obtained by direct reading at 525 nm for the xylan samples. For pulp samples, the sample was always charged to contain 200–300 nmol total uronic acids, so the absorbance of the cellulose could be omitted. The MeGlcA contents were subsequently calculated by subtracting the

14 HexA content from the total uronic acid contents. The amount of MeGlcA obtained was later determined through methanolysis, followed by GC detection58 to evaluate the accuracy of the different methods. The molecular weight of dissolved xylan was determined using a gel permeation chromatography (GPC) method, as described by Jacobs and Dahlman2.

Both the surface and total charges were determined for all unbeaten pulps. The total charge was determined using conductometric titration according to Katz et al.59, while the surface charge was determined using polyelectrolyte titration. Poly- diallyldimethylammonium chloride (pDADMAC) was adsorbed onto fibres in a solution as described by Winter et al.60 at low electrolyte concentration as recommended by Horvath et al. 61 (10–5 mol/l NaHCO3). The fibres were separated by means of filtration, and the excess pDADMAC was titrated using potassium polyvinyl sulphate62 in the presence of orthotoluidine blue indicator.

Some pulps were beaten in a PFI mill, made into paper sheets, and strength tested according to standard procedures.

15 Results and Discussion

Xylan dissolution, degradation, and redeposition Birch wood was kraft cooked at 165°C for different lengths of time. The amounts of dissolved xylan and pulp xylan were determined, and the results are presented in Figure 5. The cooking times include the heating-up time, but have been re- calculated to be valid for the final cooking temperature using the activation energy of 117 kJ/mol according to Lindgren and Lindström10.

70 60 50 40 Pulp 30 BL 20 wood xylan (%) Fraction of birch 10 0 0123 Cooking time (hours)

Figure 5. The amount of xylan in pulp and black liquor (BL) after different cooking times at a final cooking temperature of 165°C.

As can be seen in Figure 5, 20% of the birch xylan was lost after even a very brief cooking time. The rate of xylan dissolution was initially high, and the amount of xylan in the cooking liquor increased at first; however, after only a brief cooking time, the rate of xylan redeposition and degradation came to exceed the rate of dissolution, and the amount of xylan in the liquor decreased. Dissolution and redeposition onto the fibres seemed to be important processes in the first hour of cooking, whereas later on degradation reactions were dominant. It should be kept in mind that the method used to purify the black liquor xylan requires a certain molecular weight (smaller molecules will not precipitate, but will stay in solution during purification and so be classified as degraded in this study).

16 There are different ways of determining the molecular weight of hemicelluloses and to do that from wood without modifying them during purifications is a challenge.

However, Jacobs et al.2 determined the Mw of native birch 4-O- methylglucuronoxylan to 13.7–16.5 kg/mol, in a GPC system coupled with MALDI-TOF MS. The same system was used for the isolated black liquor xylan samples and as seen in Table 1, the molecular weight decreased throughout the entire cook, ending up at a Mw of 5.9 kg/mol.

Table 1. Pulps after different cooking times and the corresponding molecular weights of dissolved xylan.

Sample Time1 H-factors Kappa no. Total yield [OH–]res Mw (kg/mol) (min) (%) mol/l Dis. Xylan 1 0.2 2 - 82.5 0.651 12.2 2 10.2 100 - 65.6 0.227 11.7 3 70.2 660 18 51.0 0.258 8.8 4 180 1700 13 50.6 0.166 5.9 1 At final temperature.

The loss in molecular weight can be related to the amount of degraded xylan if one assumes that peeling reactions are alone responsible for degradation. The number of lost monomers was used to calculate the mass of degraded xylan, as can be seen in the right column in Table 2. A more direct method of estimating the amount of degraded xylan was also used. By simply subtracting the amount of xylan found in the pulp and liquor from the amount in the wood, the amount of degraded xylan could be determined; the values calculated in this way are shown in the fourth column in Table 2. As seen in the table, the results of the two methods correlated very well for the first two samples. For samples 3 and 4, the decrease in molecular weight could not be used to estimate the amount of degraded xylan, which means that not only peeling took place but probably also alkaline hydrolysis. Alkaline hydrolysis dramatically affects the molecular weight but does not degrade carbohydrates into iso-saccharinic acid, so the material is still found in solution.

17 Table 2. The fraction of the birch wood xylan found in black liquor and pulp and the calculated amounts of degraded xylan, for samples 1–4 described in Table 1; figures in (%). Degraded, assuming only Black Degraded peeling degrades xylan

Sample Liquor Pulp (100% − BL − Pulp) (loss in Mw) 1 16.6 61.9 21.6 20.1 2 29.3 45.7 25 24.1 3 19.5 48.3 32.2 48.5 4 8.1 48.5 43.4 73.1

Xylan precipitation onto cellulose fibres The mechanism underlying the process by which xylan goes from being in solution to being attached to the fibres during kraft cooking remains unclear. Whether it is via an adsorption or precipitation/crystallisation process has been discussed, and many different notations describing the phenomenon are cited in the literature. The concept of precipitation involves a loss of solubility, and this is likely to be the case when the hydroxide ion concentration is lowered. Linder et al.63 suggest a mechanism for the assembly of xylan on cellulose surfaces in which the hydrophobic interactions between the lignin structures associated with xylan molecules play an important role in the formation of aggregates. Another possibility is that aggregates are formed via hydrophobic interactions between low- charged segments of the xylan molecules. These processes are examples of precipitation. However, it is also possible that the low solubility of xylan when pH is decreased is not related to any hydrophobic interactions at all. Examining xylan−cellulose interactions using NMR measurements revealed that there was a change in the conformation of xylan, and it was suggested that the xylan adapted to the three-dimensional structure of cellulose when attached64,65. This implies that there are molecular attractions between the molecules that make them come together, inducing a change of the xylan conformation. How exactly this process proceeds is under discussion: there might well be a crystallisation process involving

18 the xylan on the cellulose fibrils, or it is possible that pure adsorption takes place, followed by a re-conformation of xylan on the cellulose surface.

The influence of temperature on xylan deposition was investigated under kraft cooking conditions. Cotton fibres were heated together with birch black liquor corresponding to sample 4 in Table 1. The degradation of the cellulosic material was determined and compensated for by cooking cotton fibres together with white liquor under the applied conditions. In all cotton cooks, the hydrogen sulphide ion concentration was 0.2 mol/l and the sodium ion concentration 2.0 mol/l. Figure 7 indicates only a slight increase in the amount of deposited material with increased temperature. This deposition was, however, found to be facilitated by a low hydroxide ion concentration (as seen in Figure 6), which is in accordance with earlier findings41,66,67. One common explaination is that the partly deprotonised

hydroxyl groups on xylan, whose pKa is close to 14, will be protonised when the hydroxide ion concentration drops. This would off course strongly influence the solubility.

5 5

4 4

3 3

2 2 T = 140°C [OH¯] = 0.2 mol/l 1 1 T = 160°C Weight increase (%) [OH¯] = 0.6 mol/l Weight increase (%) 0 0 012345 012345 Cooking time (hours) Cooking time (hours)

Figure 6. The precipitation of xylan onto Figure 7. The same experiment carried cotton fibres at different hydroxide ion out at different temperatures. The hydroxide concentrations. The temperature was ion concentration was 0.4 mol/l. 150°C.

From the temperature dependency of the initial rate of adsorption and from Arrhenius’ law of kinetics, the energy of activation for xylan deposition onto

19 cellulose fibres was determined to be 8.4 kJ/mol. The energies of activation reported earlier were somewhat higher: 4 and 6 kcal/mol, i.e., 17 and 25 kJ/mol68,69. This can be explained by the different environments of the adsorption process examined and the different kinds of xylan used in these two earlier studies. All these values are extremely low, suggesting that diffusion exerts a rate- controlling influence.

The solubility of xylan is clearly related to the uronic acid content33. Uronic acids affect the properties of xylan and may increase the strength of the formed sheet. The amount of carboxylic acids on the fibre surface was measured as the surface charge of the cotton fibres. The reference fibres were also measured and their contribution to the surface charge was subtracted from the measured value of the samples. The charges on the reference fibres are due to both the small portion of hemicellulose in the cotton and to degradation products. In Figure 8 and 9, the number of surface-charged groups is normalised to the amount of xylan in the fibres. This allows the cleavage of the glycosidic bonds connecting the uronic acids to the xylan backbone, together with the degradation of the xylan itself, to be estimated.

2 2 [OH¯] = 0.2 mol/l T = 140°C [OH¯] = 0.6 mol/l 1.5 1.5 T = 160°C 1 1

Xylan charges 0.5 0.5 (ekv/100 xylose)

0 0 01234501234 Cooking time (hours) Cooking time (hours)

Figure 8. The degree of substitution of charged Figure 9. The degree of substitution of groups of the precipitated xylan at different charged groups of the precipitated xylan at hydroxide ion concentrations. The temperature was different cooking temperatures The 150°C. hydroxide ion concentration was 0.4 mol/l.

20 As seen in Figure 8, a high hydroxide ion concentration gives lower degrees of substitution of the xylan precipitated onto the cotton fibres. Hydroxide ion concentration is important for xylan stability, and it also affects the number of uronic acid groups in xylan, as seen in the figure. It can be hypothesised that both the rate of cleavage of the glycosidic bonds connecting the uronic acids to the xylan backbone and the rate of degradation reactions of xylan itself are increased, but to different extents, by an increase in hydroxide ion concentration. The effect of temperature is also evident, as seen in Figure 9. The temperature dependency can be used yet again in estimating the energy of activation for the cleavage reaction, which was determined to be 89 kJ/mol. However, it is important to remember that this value for the energy of activation of the cleavage of charged groups on the xylan backbone was seen for this typical study. Only the precipitated xylans were included in the study and later in this thesis a more thorough kinetic study of the reactions of the side groups in xylan is presented.

Effects of xylan on tensile strength properties What could be observed at both high and low temperatures and both high and low hydroxide ion concentrations was that the adsorption rate decreased within the first hour. Most of the adsorbable material was attached to the fibres within 1 h, after which adsorption was slow. The potential for xylan precipitation onto the fibres later in the cook, thereby contributing to yield and strength, is an interesting matter. Since very little xylan is present in softwood pulping, using waste black liquor from hardwood kraft cooking in softwood pulping is an attractive option. This is possible in mills having both a softwood and hardwood fibre production line.

Two xylan-rich black liquors from birch cooking were added to the softwood kraft cook for the last hour of cooking time. The types of xylan in the two added liquors were quite different: the black liquor corresponding to sample 2 in Table 1 was used, as was that corresponding to sample 4. Figure 10 shows the tensile index for the softwood pulps at different degrees of beating, shown as sheet density.

21

120

115

110 HMw x (kNm/kg) 105 LMw 100 Ref

95 Tensile inde

90 700 720 740 760 780 800 820 Sheet density (kg/m3)

Figure 10. Tensile index of different softwood kraft pulps. HMw indicates that birch

black liquor containing xylan with a Mw of 11.7 kg/mol was added to the cook, while

LMw indicates that birch black liquor containing xylan with a Mw of 5.9 kg/mol was added. The mean confidence interval of 95% is shown as error bars. The variation in sheet density was obtained by different degrees of beating.

All the studied pulps were beaten to 500, 1000, 2000, and 4000 PFI revolutions, resulting in different sheet densities; as seen in the graph, xylan addition causes a slight densification of the sheet. The tensile stiffness for the same pulps is shown in Figure 11. The mean confidence interval of 95% is indicated in both Figures 10 and 11 by error bars. The deviation in thickness direction is very small due to the large number of measured points.

22 11.5 x ess inde n 11

(kNm/kg) HMw LMw Tensile stiff Ref 10.5 700 750 800 Sheet density (kg/m3)

Figure 11. The tensile stiffness index for the same pulps as presented in Figure 10. The mean confidence interval of 95% is shown as error bars.

The highest tensile strength and tensile stiffness were achieved when the replacement liquor had been extracted early in the birch kraft cook. The black liquor corresponding to sample 2 as described in Table 1 contained xylan of a high molecular weight (Mw = 11.7 kg/mol) and is denoted HMw in Figures 10 and 11. The pulp produced with liquor from cook 4 (as described in Table 1) present during the last hour is denoted LMw, since the Mw of the xylan in this case was only 5.9 kg/mol. The reference pulp was produced using only white liquor as the replacement liquor. The data shown in the HMw series correspond to the pulp produced by adding diluted HMw black liquor to give the same xylan concentration as was given by the LMw black liquor (7.85 g/l). One kraft cook was also carried out in which HMw black liquor was added so as to give a xylan concentration of 20.4 g/l, but no significant further increase in tensile strength was observed. The pulp produced in the last cook will be referred to as “HMw at high conc.”

As shown in Figure 10, xylan addition can produce less-dense sheets of the same tensile index if less beating is used. For example, at a tensile index of 105 kNm/kg, the density of sheets made of the HMw pulp was 740 kg/m3, whereas it was 760

23 kg/m3 for sheets made of the reference pulp. This means that in making products for which bulk is vital, material can be saved. If the thickness must be at least 110 µm, for example, the amount of material can be reduced from 84 to 81 g/m2, which means 3.3% less pulp is needed to produce the same amount of product. For a mill with a pulp production of 500.000 tons/year, such a process change would mean an additional profit of € 7.3 million per annum if the pulp costs € 440/ton (this does not include the savings related to the use of less energy for beating or the additional profit related to an increase in pulp yield).

The tensile stiffness was increased by the addition of black liquor containing low- molecular-weight xylan (LMw); however, the increase was doubled when high- molecular-weight xylan dissolved in black liquor (HMw) was used. During kraft cooking, both the molecular weight and degree of substitution of the xylan molecules are affected. For the xylan in the LMw liquor, the degree of substitution was determined from the sum of hexenuronic acid and 4-O-methylglucuronic acid content, and was found to account for a third of the xylan found in the HMw liquor. This results in different charge distributions of the different manufactured softwood fibres, as shown in Table 3.

Table 3. Charge distributions for unbeaten pulps. Xylan Surface Total Fraction of Difference from Surface Pulps Content charge charge Total charge reference (%) selectivity (%) (µekv/g) (µekv/g) on surface (%) Surface Total (surf/tot) Reference 7.5 4.2 46.1 9 LMw 8.0 4.8 52.6 9 13 14 1 HMw 8.2 5.6 50.6 11 31 10 3 HMw at high conc. 9.0 8.9 53.5 17 110 16 7

The HMw xylan introduces more charges into the pulp fibres than the LMw xylan does. Moreover, the LMw xylan does not seem to be more attached to the surface than to the bulk of the fibres. The surface selectivity, here defined as the change in surface charges divided by the change in total charges, is 1 for the LMw xylan

24 addition. When the high-molecular-weight xylan is added, the surface charges increase three times as much as the bulk charges do, and at the same time the strength of the sheet increases further. When adding the same type of xylan at a higher concentration, more is adsorbed onto the fibres, introducing more charges, especially on the surface of the fibres. Despite this high increase in surface charges, no further increase was evident in either the tensile strength or tensile stiffness of the sheet. This indicates that it is no longer the surface charges that limit the strength. The limiting factor still seems to be related to the joint strength or the contact area rather than to the fibre strength, since the strength is still enhanced by beating. This supposition was supported by zero-span tensile testing, which found no significant difference between the pulps presented in Table 3. The water retention value was also determined for the unbeaten pulps and no significant change was observed.

Many studies have found tensile strength to be increased by a higher xylan content of the pulp38,43,70. Mechanisms possibly underlying this strength increase have been proposed, but no clear correlation between the molecular weight of the hemicellulose and the magnitude of the strength increase has yet been reported. The influence of the molecular weight and/or degree of substitution on tensile strength, as seen in Figure 10, suggests an adsorption process determined by either the molecular weight or the degree of substitution. The two different types of xylan might be attached to the fibres in different ways. It has been reported that a high degree of polymerisation favours xylan adsorption onto cotton fibres33. It is seen in Table 3 that the xylan content is not very different, although it is clear that the HMw pulp contains more xylan than the LMw pulp does. It is unlikely that this difference explains the difference in strength between the two pulps, since the HMw at high conc. pulp contains a considerably higher amount of xylan but does not display correspondingly higher strength. The surface charge density of fibres has been correlated with increased fibre–fibre joint strength71. As discussed above, the effect of increased surface charge density was observed to be positive for the sheet tensile strength at low values but levelled out at surface charge densities of 5.6 µekv/g or greater.

25 As shown in Figure 10, the addition of low-molecular-weight xylan slightly improves the tensile strength of sheets produced with less beating. With more revolutions of beating, the strength-enhancing effect is as great as is achieved by the addition of high-molecular-weight xylan. One possible interpretation is that part of the xylan enters the fibre wall; by beating the fibres long enough, this xylan becomes accessible to neighbouring fibres, making it possible to increase either the fibre–fibre joint strength or the contact area between fibres, or to improve the formation of the sheet by decreasing the friction between fibres. The difference in surface selectivity between HMw and LMw, presented in Table 3, supports the theory that low-molecular-weight xylan enters the fibre wall to a greater extent than high-molecular-weight xylan does. That is to say, the low-molecular-weight xylan molecules seem to be small enough to enter most pores in the cell wall. Moreover, it has been shown that xylan can form aggregates when adsorbing onto cellulose surfaces72. According to Axelsson et al.73, 70% of the xylan in unbleached birch pulp is located in so-called lignin–carbohydrate complexes, and this fact affects how the xylan is attached to the fibres. The amount of xylan located in lignin– carbohydrate complexes in the different liquors is as yet unknown. As discussed earlier, xylan can form aggregates and the size of these will affect how much of the xylan that can reach the inner part of the fibre and how much is staying close to the fibre surface. It is possible that the high-molecular-weight xylan has a greater tendency to form large aggregates than low-molecular-weight xylan does.

The results concerning strength properties suggest that high-molecular-weight xylan more effectively increases the fibre–fibre joint strength, contact area, or number of joints between fibres than the low-molecular-weight xylan does. Part of this effect can be explained by the increased surface charge density, but there are still other factors that improve the final sheet strength. It has earlier been shown that addition of xyloglucan has a positive effect on both sheet formation and tensile strength74. Sheet formation is likely improved by the xylan enrichment of the fibre surfaces through a possible decrease in friction between the wet fibres.

26 Xylan substituent reaction kinetics The degree of substitution of the xylan is of great importance, not only due to its influence on the solubility of xylan and strength of the pulp as discussed, but also in terms of bleachability and brightness stability. As stated in the introduction, hexenuronic acid forms during kraft cooking. The kinetics of the reaction shown in Figure 4 were investigated under kraft cooking conditions at a high liquor-to-wood ratio.

The amounts of glucuronic acid in pulp xylan and dissolved xylan are shown in Figure 12 for two different cooking temperatures at different degrees of delignification of birch wood.

13

lan 140°C dissolved xylan

) 12

xy 160°C dissolved xylan d lose

y 140°C pulp xylan 11 160°C pulp xylan 100 x / s

p 10 Grou ( 9 MeGlcA in dissolve

8 10 12 14 16 18 20 22 kappa*

Figure 12 The methylglucuronic acid contents of dissolved and of pulp xylan presented as the number of groups per 100 xylose units at different degrees of delignification.

The degree of delignification is here indicated by kappa*, which indicates the corrected kappa number − corrected to exclude the contribution of HexA. The contribution of HexA to the kappa number was estimated according to Li and Gellerstedt47. As seen in Figure 12, the degree of substitution of MeGlcA on xylan is higher for the dissolved than for the pulp xylan. It is also seen that the influence of cooking temperature is great in the dissolved phase, whereas the amount of

27 MeGlcA in pulp xylan is not markedly affected by increased temperature. To be able to calculate the rate of MeGlcA degradation, it was assumed that the change with time of the glucuronic acid content can be expressed as follows: d[]MeGlcA = k ⋅[MeGlcA] (2) dt MeGlcA

where kMeGlcA is the rate constant of all glucuronic acid−consuming reactions. The solution to Equation 2 is shown in Equation 3, where [MeGlcA]0 is the MeGlcA content of wood.

−kMeGlcA⋅t []MeGlcA = []MeGlcA 0 ⋅e (3)

The rate constant, kMeGlcA, was estimated from the least square fit of Equation 3 to the amounts of MeGlcA shown in Figure 12 for different cooking times. The energy of activation for the consumption of MeGlcA in both pulp and dissolved xylan can be estimated from the temperature dependency of the rate constant kMeGlcA, using Arrhenius’ law of kinetics as shown in Equation 4.

−Ea k = A ⋅ e R⋅T (4)

Here, k is the rate constant, A is Arrhenius’ pre-exponential factor, Ea the energy of activation, R the general gas constant, and T the absolute temperature. The same type of equation was used to describe the rate of delignification at the different studied temperatures, as shown in Equation 5. d(kappa*) = k ⋅kappa * (5) dt delignification

28

Table 4. Observed k values. Temperature kMeGlcA(pulp) kMeGlcA(solution) kdelignification

(°C) (min–1) (min–1) (min–1) 140 0.0013 0.0003 0.0042 150 0.0031 0.0013 0.0090 160 0.0050 0.0022 0.0159

Ea (kJ/mol) 120 180 118

The energy of activation values were found to be 120 kJ/mol and 180 kJ/mol in the pulp and black liquor, respectively. This should be compared to the energy of activation for the bulk delignification of birch, which was previously reported to be 117 kJ/mol10 and was here found to be 118 kJ/mol. The decrease in the glucuronic acid content of pulp xylan displays a temperature dependency very close to that of the delignification, which implies that it is difficult to control the glucuronic acid content of pulp xylan by means of temperature. For dissolved xylan, the situation is different, and the temperature dependency of the methylglucuronic acid−consuming reactions is higher than the comparable value for delignification. It seems as though part of the methylglucuronic acid content of the pulp xylan is stable during the kraft cook independent of temperature, while another part dissolves together with the xylan, degrading to various extents depending on the temperature.

The hexenuronic acid content of the same set of pulp and black liquor xylan samples as shown in Figure 12 is shown in Figure 13.

29 4 n a 140°C dissolved xylan xyl n 3 160°C dissolved xylan 140°C pulp xylan 160°C pulp xylan HexA i

(Groups/100 xylose) 2 10 12 14 16 18 20 22

kappa* kappa* Figure 13. The hexenuronic acid content of the dissolved and the pulp xylan presented as the number of groups per 100 xylose units at different degrees of delignification.

Figure 13 shows the parameter one should try to minimise, namely, the amount of hexenuronic acid in pulp xylan. Although the cooking temperature was varied within the range of values found in industrial processing, the amount of hexenuronic acid in pulp xylan was only slightly lower at the higher cooking temperature. Also in this case, the dissolved xylan displays greater substitution, especially at the lower temperature. As seen in the figure, minimising the hexenuronic acid content of pulp by increasing the temperature does not significantly affect the actual hexenuronic acid content of xylan.

Examining the degree of side group substitution in the xylan that goes into solution and to some extent also readsorbs onto the fibres is troublesome. The change in MeGlcA content with cooking time might well be related to either the selective dissolution of high-substituted xylans and/or to the readsorption of low- substituted xylans. To avoid this problem, the total amounts of MeGlcA and HexA (in both dissolved and pulp xylan) were determined for different cooking times and temperatures. To obtain addable data for pulp and dissolved xylan, the amounts of hexenuronic acid and 4-O-methylglucuronic acid will from now on be expressed in terms of µmol/g of wood cooked. This means that when xylan is degraded, a

30 decrease in the MeGlcA content, and to some extent in the HexA content, is observed.

As seen in Figure 4, the change in the hexenuronic acid content of xylan is determined by both the methylglucuronic and hexenuronic acid contents, as expressed in Equation 6. d[]HexA = k ⋅[]MeGlcA − k ⋅[HexA] (6) dt 1 3

The initial condition is [HexA] = 0 at t = 0. This differential equation can be solved using Equation 3, and the solution is shown in Equation 7.

k ⋅[]MeGlcA −k ⋅t −(k +k )⋅t []HexA = 1 0 (e 3 − e 1 2 ) (7) k1 + k2 − k3

In Table 4, kMeGlcA is shown; it can be seen in Figure 4, that the reactions consuming MeGlcA are controlled by k1 and k2 which give Equation 8.

kMeGlcA = k1 + k2 (8)

Using Equation 7, together with obtained data, the constants can be estimated as follows. For each temperature the equation was written as a ratio between two points

− ⋅ − + ⋅ []HexA (t ) e k3 t 1 − e (k1 k2 ) t 1 1 = (9) −k3 ⋅t 2 −(k1 +k2 )⋅t 2 []HexA (t 2 ) e − e

where t1 is one cooking time and t2 is another. The ratio can be created for any two points in a series; in this study the first and last points were used.

31 By using Equation 8, the only unknown left in Equation 9 is k3, leading to a unique solution. When k3 has been determined, Equations 7 and 8 can be used again.

k ⋅[]MeGlcA − ⋅ − ⋅ []HexA (t ) = 1 0 (e k3 t − e kGlcA t ) (10) kMeGlcA − k3

Here, the only unknown is k1, so it would be possible to calculate k1 for all values of [HexA] at different cooking times. However, these values should all produce the same results, so the average value was used in this study.

Once k1 has been determined, k2 can also be determined from Equation 8. The values of the rate constants are shown in Table 3, together with the energies of activation that were calculated in the same way as above.

Table 5. Rate constants at different cooking temperatures. Temperature k1 k2 k3

(°C) (min–1) (min–1) (min–1) 140 0.0006 0.0001 0.0011 150 0.0016 0.0004 0.0028 160 0.0036 0.0007 0.0071 Ea (kJ/mol) 129 143 141

It can be seen in the table that the degrees of temperature dependency of the degradation reactions of MeGlcA and HexA are very similar. These degradation reactions are not only due to the cleavage of the side groups but also to the degradation of xylan itself. The energy of activation for HexA formation is slightly lower; but as the energy of activation for delignification is lower still, it is possible to control these reactions by adjusting the cooking temperature. An increase in temperature will affect the rate of these reactions more than it will affect the delignification rate. This means that cooking at low temperatures produces a xylan richer in MeGlcA than when xylan is cooked at a higher temperature. As seen in Figure 12, this highly charged xylan stays in solution and the MeGlcA content of

32 pulp xylan is only slightly affected by the change in cooking temperature. The situation is different for the hexenuronic acid, since the reaction rates of both formation and degradation are more temperature sensitive than the delignification rate is. The effect of temperature will not be as pronounced as in the case of MeGlcA. However, the HexA content is higher for xylan cooked at lower rather than higher temperatures, since the energy of activation is higher for the degradation than for the formation of HexA. All HexA and MeGlcA contents are shown in Figure 14: the solid lines represent the models expressed in Equations 3 and 7 with the k values shown in Table 5. The MeGlcA content of wood was estimated from the best fit to the measured points (134 µmol/g wood) and not surprisingly, it was lower than literature data on the MeGlcA content in birch wood (213 µmol/g wood)2.

mol/g wood µ

Cooking time (minutes)

Figure 14. The upper curves show the methylglucuronic acid content and the lower curves show the hexenuronic acid content of the sum of the pulp xylan and the dissolved xylan manufactured at different temperatures; (o) indicates 140°C, (*) 150°C, and (+) 160°C.

The energies of activation shown in Table 5 are all considerably higher than was recently reported by Simao et al. for Eucalyptus globulus50. Their approach was somewhat different and no analyses were made of the dissolved xylan. As well, they divided the methylglucuronic acid content into different portions in terms of their

33 reactivity; moreover, they did not realise that the results of the colorimetric method using carbazole give the total amount of uronic acids including HexA. These differences in method make it very difficult to compare the different studies. However, both these studies of hardwood pulps found that the methylglucuronic acid is not totally degraded early in the cook, as has been reported in the case of softwood xylan48,75. Approximately half of the methylglucuronic acid content of pulp xylan seems to remain at the end of the cook, while even more remains in the dissolved xylan.

The ionic strength in the cooking liquor is typically higher in the industrial process than in laboratory cooks. This is due to dead-load caused by imperfections in the chemical recovery and black liquor being recirculated and added to the early parts of the cook, alternatively used as impregnation liquor. The effect of ionic strength, measured as sodium ion concentration, on the xylan substituents reactions was investigated. Two cooks were carried out at different sodium ion concentrations and the degree of substitution of pulp xylan is shown in Figure 15.

12.0 d 11.0

10.0 s/100 p 9.0

8.0

7.0

6.0 xylan (Grou MeGlcA in dissolve 5.0 10 12 14 16 18 20 22 kappa *

Figure 15. The amount of methylglucuronic acid is higher in dissolved xylan cooked at low ionic strength, compared at the same corrected kappa number. Circles (o) indicate that the sodium ion concentration [Na+] = 1.0 mol/l and crosses (x) that [Na+] = 2.0 mol/l in the cook.

34 The lower degree of substitution in the dissolved xylan cooked at a high ionic strength can be related to the longer cooking times needed in order to reach the same degree of delignification. A higher ionic strength thus greatly decreases the rate of delignification. For pulp xylan there was also a higher degree of substitution in the xylan cooked at a low sodium ion concentration, as seen in Figure 16, although here the difference is smaller, as was the case with the different cooking temperatures.

10.0

100 9.5 / s

p 9.0

8.5 Grou ( 8.0 lan MeGlcA in pulp y

x 7.5 10 12 14 16 18 20 22 kappa* Figure 16. The amount of MeGlcA in pulp xylan for pulps manufactured at cooks of different ionic strengths. Circles (o) indicate that the sodium ion concentration [Na+] = 1.0 mol/l and crosses (x) that [Na+] = 2.0 mol/l in the cook.

Hexenuronic acid was also determined and the amounts found in both pulp and dissolved xylan were found to be very similar. The effect of ionic strength on the cook was also very small. The curves are shown in Figure 17.

35

hydroxide ionactivity. the negativeinfluence forces the sorptionandthisismostlikelyrelate of thesereactions.Itisalsowe a changeinionicstrengthaffectstherate hexenuronic andmethylglucuronic selectivity, delig concentration, hasanegativeimpactonthekraftcookingperformanceintermsof It iswellestablishe validate thisconclusion andc temperature isverysmallintermsoftheam The kineticstudyofthesidegroupsin Cooking temperatureandtensilestrength circulation digesterusing a strength, twopulpswere manufacturedat

64 delignification. indicating low low sodiumionconcent Figure 17. 17. Figure

. HexA in xylan (Groups/100 xylose) 3. 4. 0. 1. 2. 0 0 0 0 0 nification rateandpulpbrig 10 Hexenuronic

sodiumionconcentrationan d thathighionicstrength,measuredintermsofsodiumion ofsodium Thehydroxideio 12 r atio high liquor-to-woodratio. Th a c n andboxeshi id conte ll knownt o ntribute acidreacti ionsca 14 n t ofpulpxyla 36 d totheshieldingofrepulsiveelectrostatic h totheunderst at of delignif gh) anddissolvedxylan(unfilledcircles 140°Cand160°C,respectively, inthe n concentrationisalsoimportantf d xylan demonstratedthattheeffectof n beattributedtoaloweringofthe ht kappa anincreaseinionicstrengthenhances cro ount ofsubs on kinetics,thoughitseemsas 16 ness sses high)at n (filledmark * 10, 7 6 i cation morethanthekinetics ,7 7 18 . I e pulpswerethenanalysed tituents inpulpxylan.To t hasbeensuggestedthat a different ers, nding ofpapershee ci rcles i 20 degreesof n dicating 22

o r t and strength tested. The strength of the two pulps at different degrees of beating is shown in Figure 18. The 95% confidence interval is indicated by error bars.

130

125

120

x (kNm/kg) 115 T = 140°C 110 T = 160°C 105 Tensile inde 100 0 1000 2000 3000 4000 PFI revolutions

Figure 18. The tensile index at different degrees of refining for two birch kraft pulps cooked at different temperatures. The mean 95% confidence interval is indicated by error bars.

As the temperature is increased, the rates of the kraft cooking reactions that take place simultaneously with the delignification process increase to different extents, resulting in a change in pulp properties. The pulp manufactured at 140°C has a higher beatability than the pulp manufactured at 160°C does. As discussed above, the degree of substitution of pulp xylan is only slightly affected by a change in temperature. The sheet strength is often controlled by the strength of the fibre−fibre joints, the number of these joints, and the molecular contact area between the fibres. Forsström et al.78 were able to correlate the fibre−fibre joint strength to tensile strength for a number of different pulps and as mentioned earlier, the increase in the strength of the fibre−fibre joint has been correlated with the surface charge of the fibres71. For the two pulps shown here, there was no significant difference in surface charge, as seen in Table 6. The chemical composition was also very similar in the two different pulps. The yield was of course higher for the pulp manufactured at the lower temperature, and it can also

37 be seen that the hexenuronic acid content was higher than for the pulp manufactured at the higher temperature, in accordance with the discussion above. To find the difference between the two pulps that corresponds to the difference in beatability, the surface material was enzymatically peeled off and analysed according to Dahlman and Sjöberg79; these results are also presented in Table 6.

Table 6. Pulp properties Cooking temperature 140°C 160°C Kappa number 16.4 16.0 Kappa* 12.2 12.5 Pulp yield (%) 48.3 47.4 Pulp viscosity (dm3/kg) 1410 1310 Xylose content (%) 21.3 20.9 Glucose content (%) 77.9 78.3 MeGlcA (subst/100 Xyl) 10.3 10.7 HexA (subst/100 Xyl) 3.2 2.7 Surface charge (µekv/g) 3.78 3.71 Surface material (5% of total weight) Xylan content (%) 32.1 28.6 Glucose (%) 64.6 69.0 MeGlcA (subst/100 Xyl)* 1.4 1.9 HexA (subst/100 Xyl) 3.0 3.4 * Uncertain figures due to weak detection signal.

The surface properties of fibres largely determine the strength properties of the formed sheet, as discussed in the introduction. It can be seen that the chemical composition of the surface of the fibres differs from that of the entire fibres. Alkaline hydrolysis and endwise peeling harmed the two pulps to different extents, as can be seen from the difference in viscosity. Carbohydrate degradation decreases the molecular weight of the carbohydrates. For xylan to precipitate, the molecular weight is vital, as smaller molecules will stay in solution and not contribute to paper strength. It seems as though the amount of hemicelluloses on the fibre surface is the parameter controlling the strength in this experimental set-up.

38 This study demonstrates that the improved beatability of pulps manufactured at lower cooking temperatures cannot be attributed to differences in surface charge or to the degree of substitution of either hexenuronic acid or methylglucuronic acid. The chemical composition of the fibre surface, i.e., the xylan content, is what affects the tensile strength of the final sheet.

Method development for uronic acid quantification The same two pulps were used in a thorough study of different available methods for the determination of uronic acid residues in xylan. The aim of this study was to determine the reliability of the results of the carbazole method and to obtain further evidence that the difference between the uronic acid content of the pulps in Table 6 is negligible.

Generally speaking, quantifying uronic acids is challenging and there are several conventional methods for uronic acid quantification. The two methods most commonly used in chemical laboratories are colorimetric carbazole−uronic acid assays57, which have been used so far in this thesis, and the methanolysis−gas chromatography method44,58. Both methods are strictly dependent on the operating conditions. In addition, chemical changes to the native uronic acid structures arising during different chemical processing or storage treatments usually escape effective detection and quantification by these methods.

The carbazole method used in this thesis was comprehensively studied on the model substances depicted in Figure 19 together with microcrystalline cellulose, commercial xylan, xylose, and D-Galacturonic acid.

39 OHC O OHC O CH2OH

Furfural Hydroxymethylfurfural (HMF)

OHC O COOH HOOC O

2-furancarboxylic 5-formyl-2-furancarboxylic acid (FA) acid (FFA)

Figure 19. Four furan structures that commonly result from the acid hydrolysis and dehydration of lignocellulosic materials. Furfural and HMF are formed from pentoses and hexoses, respectively. FA and FFA are formed from HexA and other uronic acids.

The substances depicted in Figure 19 are generated during analysis from the different sugars present in the pulp, as indicated in the caption. It was demonstrated that they all contribute to varying extents to light absorption at the detected wavelength of 525 nm. It was earlier stated that hexoses cause much stronger interference than pentoses do, which in fact have negligible interference effects57. For this reason, it is important to consider the light absorption of the structures originating from non uronic acid sugars. In the case of pulp samples one must consider the cellulose content, as this could result in a higher light absorption value at the detected wavelength of 525 nm than would be found in the case of pure D-galacturonic acid or xylan analysis. However, the influence of cellulose is found to vanish when a sample contains close to or more than 300 nmol uronic acids. In the kinetic study, therefore, the sample always contained 200−300 nmol of uronic acids. This way, the interference with cellulose was minimised.

On the other hand, the methanolysis−GC method as described in the literature is not correctly calibrated by applying an isolated MeGlcA. The hydrolysis and

40 detection of glycosidic MeGlcA in xylan should differ from those in model monomeric MeGlcA. To avoid these problems, a combination of carbazole quantification of the MeGlcA content of commercial xylan and applying commercial xylan to calibrate methanolysis−GC was suggested, as discussed in paper III. By using the carbazole method to quantify the MeGlcA content of commercial xylan, in which there are no interfering hexoses, the accuracy of these methods can be significantly improved. It was found that the modified methanolysis method gave an accurate standard measurement of commercial xylan. Mass spectra data also revealed that after methanolysis, there were small peaks representing the other uronic acids that were separated. It seems as if these originate from the native MeGlcA as its isomers and analogues, presumably as 4-O- methyl-idopyranosyluronic acid and 3-O-methyl-idonic acid, which have been earlier identified in both alkaline-treated model structures of MeGlcA-xylose45 and in kraft pulps80. Galacturonic acid is the end group in birch xylan, but due to degradation reactions during kraft cooking, these groups are not detected in pulp xylan.

The verification results for the two methods used on the two birch pulps mentioned in Table 6 are listed in Table 7. The carbazole method values shown comprise the detected values from which the contribution of cellulose has been subtracted. It can be seen that the methanolysis−GC method produces a figure for the total content of MeGlcA and of its isomers or analogues. The difference between the amount detected using the carbazole method and that detected using methanolysis−GC is very close to the amount of hexenuronic acid as determined using mercury acetate−HPLC.

41 Table 7. Comparison and integration of the two methods. Uronic acids Method Description 140°C 160°C detected Total uronic acids − Direct determination 176 170 Carbazole H2SO4 (µmol/g) MeGlcA (µmol/g) 64.2 60.4 Calibration by commercial Isomers and Methanolysis−GC xylan analogues of 58.2 55.7 MeGlcA (µmol/g) The difference between Integration carbazole and HexA (µmol/g) 54 54 methanolysis HgAc2–HPLC HexA (µmol/g) 49.2 40.7 Verification by − MeGlcA and its other methods Carbazole H2SO4 after 125 126 HgAc2 isomers (µmol/g)

42 Conclusions

The research presented demonstrates that it is possible, using simple means, to modify the kraft cooking process in such a way that the tensile properties of the produced paper are significantly improved. Dissolved xylan can work as a strength- increasing additive during the kraft cook and also directly increase the pulp yield. This was done through adding birch black liquors to a softwood kraft cook which is possible for kraft pulp mills having one hardwood fiberline and one softwood fiberline. The increased strength was observed as a high beatability, i.e. a higher tensile index at a given PFI revolution. If a tensile index of 105 kNm/kg is required, sheet density can be reduced from 765 to 740 kg/m3 and properties associated with bulk, such as bending stiffness, can be increased; alternatively, 3.3% less material can be used to achieve the same tensile strength and bulk.

It was observed that in increasing the beatability, the molecular weight of the xylan added is crucial. The high-molecular-weight xylan dissolved early on in the cook increases the strength to a greater extent than does the low-molecular-weight xylan present in the black liquor towards the end of the cook. This effect was observed in terms of both tensile strength and tensile stiffness. An increase in surface charge of the fibres was correlated with the increase in tensile strength and tensile stiffness up to a certain point, after which both properties become independent of the fibre surface charge.

Kraft cooking modifies the wood’s xylan structure by converting part of the MeGlcA into HexA and into several isomers and analogues of MeGlcA. The carbazole−H2SO4 colorimetric method reveals the total content of all these uronic acids, while the methanolysis−gas chromatography method can quantify MeGlcA and its isomers and analogues but not HexA. By integrating both methods the comprehensive uronic acid composition can be quantified, together with a more accurate determination of the xylose content and thus of the ratio between various uronic acid structures and the xylose unit. The energies of activation for the

43 formation of hexenuronic acid, the degradation of methylglucuronic acid, and the degradation of hexenuronic acid in the kraft cooking of birch were found to be 129, 143, and 141 kJ/mol, respectively. The energy of activation for the delignification of birch was found to be 118 kJ/mol. The differences in activation energies imply that xylan cooked at lower temperatures contains more acidic charged groups than does xylan cooked at higher temperatures. It was demonstrated that this is indeed the case for the dissolved xylan, whereas pulp xylan is not markedly affected by a change in cooking temperature. For all cooks, dissolved xylans were richer in 4-O-methylglucuronic acid and hexenuronic acid than were the xylans located in the pulp. This strongly suggests that the number of acidic groups greatly affects the solubility of xylan in kraft cooking. The hexenuronic acid content of pulp xylan was only slightly affected by a change in cooking temperature, indicating that high cooking temperature is not a suitable way of minimising hexenuronic acid, due to other negative effects such as decreased carbohydrate yield resulting in a lower overall pulp yield.

44 Future Work

This research contributes to completing the scientific puzzle concerning how kraft pulping affects final paper properties, though it raises more questions than it answers. There are many loose ends to grab hold of, some of which are presented here.

The amount of xylan deposited onto the fibres has been shown to depend on the hydroxide ion concentration in the cook. In the study carried out here, this concentration was held constant at 0.4 mol/l; it would be very interesting to study the strength of the produced pulp when the high-molecular-weight xylan is added to the softwood cook at lower hydroxide ion concentrations. If no further increase is seen, we have most likely reached the maximum effect of dissolved xylan in terms of strength properties.

It has been shown in earlier studies that carcoxymethyl cellulose, CMC can also significantly increase the tensile strength when added to the kraft cook. This suggests that many different polysaccharides may be suitable for use as strength- increasing agents and for evaluating various additives; identifying the ones with the largest potential would be of great interest.

Exchanging liquors within softwood pulping might also be an attractive alternative for mills with only one fibre production line or with only softwood as a raw material. It is likely that the effects on strength demonstrated in this thesis may also appear with the use of softwood xylan, although the available amounts of xylan are limited.

The conversion of the native MeGlcA into its isomers and analogues should be studied further, concerning their exact structures and possible alterations on the present kinetic and paper strength studies.

45 So how does xylan increase the strength of the formed paper? This question still cannot be answered satisfactorily. Ideas concerning flexible surfaces have been proposed, but it is difficult to find ways to prove the different hypothesises involved. Work with adhesion measurements of model substances, together with thorough studies concerning the mechanism of xylan assembly, can contribute to further understanding.

46 Acknowledgements

First of all I would like to express my gratitude to my boss and supervisor, Associate Professor Mikael Lindström: thank you for all your support and valuable guidance, and for making the Pulping Technology Group such a stimulating and enjoyable place in which to work. My colleagues in the Group are also acknowledged, as they have contributed to the pleasant working atmosphere and good humour at work. Our neighbours, the wood chemists and fibre technologists, are almost as nice as we are and are very much appreciated. All my colleagues in other divisions at the Department of Fibre and Polymer Technology are acknowledged for being helpful and for tough floor ball games. The colleagues at STFI are also acknowledged for their kind help and scientific discussions.

Special thanks are due to Koki Kisara for his good co-operation and skilful laboratory work and for his perfect forehand. Dr. Jiebing Li is thanked for sharing his considerable knowledge and for generously offering his help. Dr. Anna Jacobs is thanked for always being helpful and for having a deep knowledge in GPC methods and enzymatic peeling of fibres. Professor emeritus Ants Teder, Professor Göran Gellerstedt, Associate Professor Gunnar Henriksson and Elisabet Brännvall are all heartily thanked for their constructive criticism and discussion.

To Marie − thank you for your enormous support and love.

The Biofibre Materials Centre (BiMaC) is gratefully acknowledged for generously financing the project.

47 References

1. Sjöström, E., Wood Chemistry: Fundamentals and Applications, 2nd Edition. 1993. 293 pp.

2. Jacobs, A. and O. Dahlman, Characterization of the molar masses of hemicelluloses from wood and pulps employing size exclusion chromatography and matrix-assisted laser desorption ionization time-of-flight mass spectrometry. Biomacromolecules, 2001. 2(3): p. 894-905.

3. Côté, W.A., Wood Ultrastructure. 1967: University of Washington Press, Seattle.

4. Wilder, H.D. and E.J. Daleski, Jr., Kraft pulping kinetics. II. Delignification rate studies. Tappi, 1965. 48(5): p. 293-7.

5. Kleinert, T.N., Mechanisms of alkaline delignification. I. The overall reaction pattern. Tappi, 1966. 49(2): p. 53-7.

6. Lémon, S. and A. Teder, Kinetics of the delignification in kraft pulping. I. Bulk delignification of pine. Svensk Papperstidning, 1973. 76(11): p. 407-14.

7. Axelsson, P., Aspects on birch kraft pulping and the relation between cooking conditions and pulp bleachability, in Department of Fibre and Polymer Technology. 2004, Royal Institute of Technology: Stockholm.

8. Lindgren, C.T., Kraft Pulping Kinetics and Modeling, in Department of Pulp and Paper Chemistry and Technology. 1997, KTH, Royal Institute of Technology: Stockholm. p. 65.

9. Vroom, K.E., The \"H\" factor: a means of expressing cooking times and temperatures as a single variable. Pulp Paper Mag. Can., 1957. 38(No. 2): p. 228-31.

10. Lindgren, C.T. and M.E. Lindström, Kinetics of the bulk and residual delignification in kraft pulping of birch and factors affecting the amount of residual phase lignin. Nordic Pulp & Paper Research Journal, 1997. 12(2): p. 124-127,134.

11. Li, Z., J. Li, and G.J. Kubes, Kinetics of delignification and cellulose degradation during kraft pulping with polysulphide and anthraquinone. Journal of Pulp and Paper Science, 2002. 28(7): p. 234-239.

12. Saarnio, J. and C. Gustafsson, The dissolving and destruction of carbohydrates during the sulfate cook. Paper and Timber (Finland), 1953(No. 3, Teollisuuden Keskuslab. Tiedonantoja No. 115): p. 65-6,78.

13. Yllner, S. and B. Enström, The adsorption of xylan on cellulose fibers during the sulfate cook. II. Svensk Papperstidning, 1957. 60: p. 549-54.

48 14. Yllner, S. and B. Enström, Adsorption of xylan on cellulose fibers during the sulfate cook. I. Svensk Papperstidn., 1956. 59: p. 229-32.

15. Clayton, D.W. and J.E. Stone, The redeposition of hemicellulose during pulping. I. The use of tritium-labeled xylan. Pulp & Paper Magazine of Canada, 1963. 64(11): p. T459-T468.

16. Hartler, N., Extended delignification in kraft cooking - a new concept. Svensk Papperstidning, 1978. 81(15): p. 483-4.

17. Norden, S. and A. Teder, Modified kraft processes for softwood bleached-grade pulp. Tappi, 1979. 62(7): p. 49-51.

18. Teder, A. and L. Olm, Extended delignification by combination of modified kraft pulping and oxygen bleaching. Paperi ja Puu, 1981. 63(4a): p. 315-18, 321-2, 325-6.

19. Surewicz, I.W., Sorption of organic components from cooking liquor by cellulose fibers; its relation to the \"dangerous cooking crest\" in alkaline pulping. Tappi, 1962. 45: p. 570-8.

20. Norgren, M., H. Edlund, L. Wågberg, B. Lindström, and G. Annergren, Aggregation of kraft lignin derivatives under conditions relevant to the process, Part I. Phase behavior. Colloids and Surfaces, A: Physicochemical and Engineering Aspects, 2001. 194(1-3): p. 85-96.

21. Johansson, B., J. Mjöberg, P. Sandström, and A. Teder, Modified continuous kraft pulping - now a reality. Svensk Papperstidning, 1984. 87(10): p. 30-5.

22. Page, D.H., Theory for the tensile strength of paper. Tappi, 1969. 52(4): p. 674-81.

23. Davison, R.W., Weak link in paper dry strength. Tappi, 1972. 55(4): p. 567-73.

24. Van den Akker, J.A., A.L. Lathrop, M.H. Voelker, and L.R. Dearth, Importance of fiber strength to sheet strength. Tappi, 1958. 41: p. 416-25.

25. Leopold, B. and D.C. McIntosh, Chemical composition and physical properties of wood fibers. III. Tensile strength of individual fibers from alkali-extracted loblolly pine holocellulose. Tappi, 1961. 44: p. 235-40.

26. McIntosh, D.C., Tensile and bonding strengths of loblolly pine kraft fibers cooked to different yields. Tappi, 1963. 46: p. 273-7.

27. Leopold, B. and J.L. Thorpe, Effect of pulping on strength properties of dry and wet pulp fibers from Norway spruce. Tappi, 1968. 51(7): p. 304-8.

28. Leopold, B., Effect of pulp processing on individual fiber strength. Tappi, 1966. 49(7): p. 315-18.

29. Molin, U. and A. Teder, Importance of cellulose/hemicellulose-ratio for pulp strength. Nordic Pulp & Paper Research Journal, 2002. 17(1): p. 14-19,28.

49 30. Lindström, T., L. Wågberg, and T. Larsson. On the nature of joint strength in paper -A review of dry and wet strength resins used in paper manufacturing. 13th Fundamental Research Symposium. 2005. Cambridge: The Pulp and Paper Fundamental Research Society. p. 457-562

31. Page, D.H. and J.H. De Grace, Delamination of fiber walls by beating and refining. Tappi, 1967. 50(10): p. 489-95.

32. Rydholm, S.A., Pulping processes. Vol. 1. 1965, New York: John Wiley & Sons, Ltd. p1159.

33. Walker, E.F., Effects of the uronic acid carboxyls on the sorption of 4-O- methylglucuronoarabinoxylans and their influence on papermaking properties of cellulose fibers. Tappi, 1965. 48(5): p. 298-303.

34. Bhaduri, S.K., I.N. Ghosh, and N.L. Deb Sarkar, Ramie hemicellulose as beater additive in papermaking from jute-stick kraft pulp. Industrial Crops and Products, 1995. 4(2): p. 79-84.

35. Schönberg, C., T. Oksanen, A. Suurnakki, H. Kettunen, and J. Buchert, The importance of xylan for the strength properties of spruce kraft pulp fibres. Holzforschung, 2001. 55(6): p. 639-644.

36. Spiegelberg, H.L., The effect of hemicelluloses on the mechanical properties of individual pulp fibers. Tappi, 1966. 49(9): p. 388-96.

37. Fiserova, M., E. Opalena, and J. Farkas, Effect of hemicelluloses on the papermaking properties of pulps prepared from poplar wood. Cellulose Chemistry and Technology, 1987. 21(4): p. 419-30.

38. Pettersson, S.E. and S.A. Rydholm, Hemicelluloses and paper properties of birch pulps. III. Svensk Papperstidning, 1961. 64: p. 4-17.

39. Kettunen, J., J.E. Laine, I. Yrjala, and N.E. Virkola, Strength development in fibers produced by different pulping methods. Paperi ja Puu, 1982. 64(4): p. 205-11.

40. Centola, G. and D. Borruso, Influence of hemicelluloses on the beatability of pulps. Tappi, 1967. 50(7): p. 344-7.

41. Aurell, R., Increasing kraft pulp yield by redeposition of hemicelluloses. Tappi, 1965. 48(2): p. 80-4.

42. Dillen, S. and S. Noreus, Kraft pulping with recirculation of cooking liquor containing hemicellulose. Svensk Papperstidning, 1968. 71(15): p. 509-14.

43. Sjöberg, J., M. Kleen, O. Dahlman, R. Agnemo, and H. Sundvall, Fiber surface composition and its relations to papermaking properties of soda-anthraquinone and kraft pulps. Nordic Pulp & Paper Research Journal, 2004. 19(3): p. 392-396.

50 44. Clayton, D.W., The alkaline degradation of some hardwood 4-O-methyl-D- glucuronoxylans. Svensk Papperstidning, 1963. 66: p. 115-24.

45. Johansson, M.H. and O. Samuelson, Epimerization and degradation of 2-O-(4-O- methyl-a-D-glucopyranosyluronic acid)-D-xylitol in alkaline medium. Carbohydrate Research, 1977. 54(2): p. 295-9.

46. Teleman, A., V. Harjunpaa, M. Tenkanen, J. Buchert, T. Hausalo, T. Drakenberg, and T. Vuorinen, Characterization of 4-deoxy-β-L-threo-hex-4- enopyranosyluronic acid attached to xylan in pine kraft pulp and pulping liquor by carbon-13 and proton NMR spectrometry. Carbohydrate Research, 1995. 272(1): p. 55-71.

47. Li, J. and G. Gellerstedt, The contribution to kappa number from hexeneuronic acid groups in pulp xylan. Carbohydrate Research, 1997. 302(3-4): p. 213-218.

48. Buchert, J., A. Teleman, V. Harjunpaa, M. Tenkanen, L. Viikari, and T. Vuorinen, Effect of cooking and bleaching on the structure of xylan in conventional pine kraft pulp. Tappi Journal, 1995. 78(11): p. 125-30.

49. Granström, A., T. Eriksson, G. Gellerstedt, C. Rööst, and P. Larsson, Variables affecting the thermal yellowing of TCF-bleached birch kraft pulps. Nordic Pulp & Paper Research Journal, 2001. 16(1): p. 18-23.

50. Simao, J.P.F., A.P.V. Egas, C.M.S.G. Baptista, and M.G. Carvalho, Heterogeneous Kinetic Model for the Methylglucuronic and Hexenuronic Acids Reactions during Kraft Pulping of Eucalyptus globulus. Industrial & Engineering Chemistry Research, 2005. 44(9): p. 2997-3002.

51. Shatalov, A.A. and H. Pereira, Uronic (hexenuronic) acid profile of ethanol-alkali delignification of giant reed Arundo donax L. Cellulose (Dordrecht, Netherlands), 2004. 11(1): p. 109-117.

52. Sjöblom, K., Extended delignification in kraft cooking through improved selectivity. Part 3. The effect of dissolved xylan on pulp yield. Nordic Pulp & Paper Research Journal, 1988. 3(1): p. 34-7.

53. Barzyk, D., D.H. Page, and A. Ragauskas, Acidic group topochemistry and fiber- to-fiber specific bond strength. Journal of Pulp and Paper Science, 1997. 23(2): p. J59- J61.

54. Axelsson, S., I. Croon, and B. Enström, Solution of hemicelluloses during sulfate pulping. I. Isolation of hemicelluloses from the cooking liquor at different stages of a birch soda cook. Svensk Papperstidning, 1962. 65: p. 693-7.

55. Theander, O. and E.A. Westerlund, Studies on Dietary Fiber. 3. Improved Procedures for Analysis of Dietary Fiber. Journal of Agricultural and Food Chemistry, 1986. 34: p. 330-336.

51 56. Gellerstedt, G. and J. Li, An HPLC method for the quantitative determination of hexeneuronic acid groups in chemical pulps. Carbohydrate research, 1996. 294: p. 41- 51.

57. Filisetti-Cozzi, T.M.C.C. and N.C. Carpita, Measurement of uronic acids without interference from neutral sugars. Analytical Biochemistry, 1991. 197(1): p. 157-62.

58. Bertaud, F., A. Sundberg, and B. Holmbom, Evaluation of acid methanolysis for analysis of wood hemicelluloses and pectins. Carbohydrate Polymers, 2002. 48(3): p. 319-324.

59. Katz, S., R.P. Beatson, and A.M. Scallan, The determination of strong and weak acidic groups in sulfite pulps. Svensk Papperstidning, 1984. 87(6): p. R48-R53.

60. Winter, L., L. Wågberg, L. Ödberg, and T. Lindström, Polyelectrolytes adsorbed on the surface of cellulosic materials. Journal of Colloid and Interface Science, 1986. 111(2): p. 537-43.

61. Horvath, A.E., T. Lindstroem, and J. Laine, On the Indirect Polyelectrolyte Titration of Cellulosic Fibers. Conditions for Charge Stoichiometry and Comparison with ESCA. Langmuir, 2005: p. ACS ASAP.

62. Terayama, H., Method of colloid titration (a new titration between polymer ions). Journal of Polymer Science, 1952. 8(2): p. 243-253.

63. Linder, A.P., R. Bergman, A. Bodin, and P. Gatenholm, Mechanism of assembly of xylan onto cellulose surfaces. Langmuir, 2003. 19(12): p. 5072-5077.

64. Mitikka, M., R. Teeaar, M. Tenkanen, J. Laine, and T. Vuorinen, Sorption of xylans on cellulose fibers. International Symposium on Wood and Pulping Chemistry, 8th, Helsinki, June 6-9, 1995, 1995. 3: p. 231-236.

65. Paananen, A., P.T. Larsson, P. Biely, G. Gilli, and L. Kroon-Batenburg, FAIR-CT96-1624 The Strength of Wood Fibres: Association Between Hemicellulose and Cellulose at the Molecular Level Final Report, in FAIR Programme Area 1.3 - Forestry- Wood Chain, I. SEOANE, Editor. 2000, EU.

66. Aurell, R., The effect of lowered pH at the end of birch kraft cooks. Svensk Papperstidning, 1963. 66(11): p. 437-42.

67. Hansson, J.A. and N. Hartler, Sorption of hemicelluloses on cellulose fibers. I. Sorption of xylans. Svensk Papperstidning, 1969. 72(17): p. 521-30.

68. Hansson, J.A., Sorption of hemicelluloses on cellulose fibers. 3. Temperature dependence on sorption of birch xylan and pine glucomannan at kraft pulping conditions. Svensk Papperstidning, 1970. 73(3): p. 49-53.

69. Clayton, D.W. and G.R. Phelps, Sorption of glucomannan and xylan on a-cellulose wood fibers. Journal of Polymer Science, Part C: Polymer Symposia, 1965. No. 11: p. 197-220.

52 70. Sihtola, H. and L. Blomberg, Hemicelluloses precipitated from steeping liquor in the viscose process as additives in papermaking. Cellulose Chemistry and Technology, 1975. 9(5): p. 555-60.

71. Torgnysdotter, A. and L. Wågberg, Study of the joint strength between regenerated cellulose fibers and its influence on the sheet strength. Nordic Pulp & Paper Research Journal, 2003. 18(4): p. 455-459.

72. Mora, F., K. Ruel, J. Comtat, and J.P. Joseleau, Aspect of native and redeposited xylans at the surface of cellulose microfibrils. Holzforschung, 1986. 40(2): p. 85-91.

73. Axelsson, P., R. Berggren, F. Berthold, and M.E. Lindström, Molecular Mass Distributions of Lignin and Lignin-Carbohydrate Complexes in Birch Kraft Pulps - Changes Caused by the conditions in the cook and their Relation to Unbleached Pulp Brightness and Bleachability. Journal of Pulp and Paper Science, 2005. 31(1): p. 1-20.

74. Christiernin, M., G. Henriksson, M.E. Lindström, H. Brumer, T.T. Teeri, T. Lindström, and J. Laine, The effects of xyloglucan on the properties of paper made from bleached kraft pulp. Nordic Pulp & Paper Research Journal, 2003. 18(2): p. 182-187.

75. Gustavsson, C.A.S. and W.W. Al-Dajani, The influence of cooking conditions on the degradation of hexenuronic acid, xylan, glucomannan, and cellulose during kraft pulping of softwood. Nordic Pulp & Paper Research Journal, 2000. 15(2): p. 160-167.

76. Axelsson, P. and M.E. Lindström, Influence of the conditions during birch kraft cooking on unbleached brightness, and on ECF- and TCF-bleachability. Nordic Pulp & Paper Research Journal, 2004. 19(3): p. 309-317.

77. Sjödahl, R.G., M. Ek, and M.E. Lindström, The effect of sodium ion concentration and dissolved wood components on the kraft pulping of softwood. Nordic Pulp & Paper Research Journal, 2004. 19(3): p. 325-329.

78. Forsström, J., A. Torgnysdotter, and L. Wågberg, Influence of fibre/fibre joint strength and fibre flexibility on the strength of papers from unbleached kraft fibres. Nordic Pulp & Paper Research Journal, 2005. 20(2): p. 186-191.

79. Dahlman, O. and J. Sjöberg. 7th European Workshop on Lignocellulosics and Pulp,. 2002. Åbo, Finland. p. 111-114.

80. Nelson, E.R., P.F. Nelson, and O. Samuelson, A new methylated uronic acid from paper pulp hydrolyzates. Acta Chemica Scandinavica (1947-1973), 1968. 22(2): p. 691-3.

53