Pulp delignification and refining: impact on the supramolecular structure of softwood fibers

The effect on softwood fiber wall nanostructure of kraft cooking, oxygen delignification and refining was evaluated by X-ray scattering. A recently developed simulation method for modelling small angle X-ray scattering (SAXS) data was used to estimate the apparent average sizes of solids (AAPS) and interstitial spaces in the fiber wall (AACS). Fiber saturation point and wide angle X-ray scattering were also used to calculate the pore volume in the fiber wall and the crystallite size of the fibril, respectively. The experimental modelled SAXS data was able to give consistent values for each kraft-cooked and oxygen-delignified pulp. Kraft delignification seems to have the major influence on the fiber nanostructure modification, while oxygen delignification has little or no significant impact even for different kappa numbers. The particle sizes values were more stable than the cavities sizes and no significant differences were seen between different delignification processes, refining or delignification degree. Pulps evaluated after PFI-refining, showed an increase in the fiber wall porosity evaluated by FSP and an increase in the interstitial spaces in the fiber wall, while the crystallite size and the particle sizes were very little or not affected at all.


Introduction
Cellulose I, the dominating allomorph in cellulosic fibers, is made of β-(1,4) D-glucan polymers, which are arranged in bundles, herein called fibrils.Fibrils are considered to be the smallest fiber wall structures and are estimated to be between 2 and 4 nm in diameter (Fengel and Wegener 1984).It is proposed by Fengel (1971) that cellulose fibrils are surrounded by loosely packed hemiceluloses surrounded by an layer of lignin.During delignification, lignin and hemicellulose removal tends to cause some aggregation of the cellulose fibrils due to the increased surface contact (Hult et al. 2001).The larger structures formed are called fibril aggregates (Bardage et al. 2004;Hult et al. 2001;Wickholm et al. 2001).The diameter of fibril aggregates have been reported to range from 10 to 30 nm according to Fengel and Wegener (1984).Apart from fibril aggregation, removal of lignin and hemicellulose can also give rise to cavities in the fiber wall, increasing therefore their porosity (Stone and Scallan 1965).The cellulose fibrils and fibril aggregates is part of the cellulose supramolecular structure of the fiber wall together with the interstitial spaces between the fibrils and fibril aggregates (Fig. 1).
It is expected that different delignification processes will give pulps with different chemical composition and component distribution, which will probably cause different supramolecular arrangements within the fiber wall.
Kraft cooking is the main delignification process used worldwide for chemical pulping, due to its versatility and good pulp properties (Sjöström 1993).Kraft cooking aims to remove lignin while preserving the carbohydrates in the fiber matrix as much as possible.It is performed under alkaline conditions with hydroxide (OH − ) and hydrosulfide (HS − ) anions as the active chemical species, together with high temperature (150-170 °C) during a certain cooking time (Sjöström 1993).From this delignification process, pulps within a wide range of lignin content (also presented as kappa number) can be obtained.The delignification degree will depend on the final purpose and properties desired for the pulps.
Oxygen delignification is a complementary process to continue to remove lignin from the fiber Fig. 1 Simplified and idealized representation of wood matrix structure (represented by the early and late wood), fibers (represented by the different layers: (M) middle lamella, (P) primary layer and secondary layers (S1, S2 and S3), fibrils, and fibril aggregates.The elementary fibrils are composed by the D-glucose chains subunits from cellulose linked by β-1-4 glycosidic bonds.The dimensions are relative to the lateral dimension of the structures matrix after kraft cooking.An alkaline environment is used together with oxygen at lower temperatures compared to kraft cooking (90-110 °C).The degree of delignification is usually lower than in cooking, as the major part of the lignin is removed by kraft cooking.However, extended oxygen delignification used for pulps with high lignin content after kraft cooking can hold great potential as means to improve pulp properties, such as higher swelling and mechanical strength (Esteves et al. 2020(Esteves et al. , 2021b)).
Besides the chemical delignification, the interior of the fiber wall and the fiber surface structures are also affected by mechanical action, such as the refining process, which is an important treatment in papermaking to increase sheet strength.The structure of the fiber wall is affected by mechanical forces through internal and external fibrillation, increasing fiber flexibility and conformability (Page 1985(Page , 1989)).Internal fibrillation is the most important effect obtained through beating, leading to the breakage of inner joints between cellulose fibrils and promoting an increased abundance of pores inside the fiber wall (Maloney and Paulapuro 1999).External fibrillation is the peeling off of fibrils from the fiber surface, which will give rise to better interfiber bonding (Clark 1969;Page 1989).
Pulp properties, such as chemical composition, distribution of components within the fiber wall, water absorption capacity, refinability as well as strength of the fiber network, are influenced by the delignification processes and chemical conditions used.It can be hypothesized that differences observed on a macroscale caused by the processes also affect the fiber wall on a nanoscale.Understanding the relationship between fiber nanostructure and pulp properties and how these processes can affect the fiber nanostructure opens the possibility to modify process conditions to obtain tailored pulp properties.Few methods have the resolution to analyze the fiber wall on a nanoscale, but X-ray scattering techniques, such as SAXS and WAXS, offer a powerful tool to examine fiber wall structures down to a fraction of a nanometer and therefore they were used in this study.The aim is to investigate changes in the fiber wall nanostructure caused by chemical and mechanical processes.

Materials
Pine or a mixture of pine and spruce wood chips with different delignification degrees, obtained by kraft cooking and oxygen delignification, were used in this study.The procedure is schematically illustrated in Fig. 2.

Kraft cooking
The kraft cook to obtain pulps denominated K91 and K52 were carried out with an effective alkali (EA) of 23.4% and 47% sulfidity in a recirculated digester.The pulps denominated K28 and K31 were obtained by kraft cooking with an EA of 22% and 30% sulfidity in autoclaves.The liquor/wood ratio was 4.5 l/kg and the cooking temperature 160 °C in all cooks.The trials were stopped at different H-factors (cooking times) depending on the desired final kappa number.
After the cooking step, the spent liquor was drained off and the delignified wood chips were washed in deionized water for 10 h in self-emptying metal cylinders and subsequently defibrated and screened in a NAF water jet defibrator (Nordiska Armaturfabriken).

Oxygen delignification
The oxygen delignification was carried out by mixing kraft cooked pulp with sodium hydroxide (NaOH), magnesium sulfate (MgSO 4 ) and water at a final consistency of 12% in polyethylene bags.The bags were closed by heat-welding, kneaded manually and in a shaker for greater homogenization of the chemicals.After the mixing, the pulps were removed from the bags and placed in pressurized steel autoclaves coated with Teflon.The autoclaves were closed, pressurized with 0.7 mPa O 2 and placed in an electrical-heated glycol bath at 100 °C, with rotation and a slight inclination.After the oxygen delignification, the pulps were washed with deionized water, filtrated, centrifuged and defibrated.

PFI-refining
The pulps were refined in a PFI-mill according to ISO 5264-2 standard.Two of the kraft delignified pulps were refined at three different levels (4000, 6000, 10,000 PFI-revolutions), and oxygen-delignified pulps were beaten 4000 PFI-revolutions.

Analysis
The pulp analysis was performed on never-dried pulps.

Fiber saturation point
Fibre saturation point (FSP) gives the mass of liquid in the pore system per unit of solid mass.This is a technique for liquid swollen porous samples that uses the solute exclusion method to measure the water inside the fiber wall that is inaccessible to a dextran solution.FSP was measured with a similar approach to that of Stone and Scallan (1967) but using only one high molecular mass dextran (Dextran 1500-2800, CAS No. 9004-54-0, about 2 MDa).A wet pulp sample with a known solids content was immersed with a 1% (dextran mass/solution mass) dextran solution (approximately one (1) mass unit of wet sample mass being mixed with three (3) mass units of dextran solution).After mixing, the sample was stored in a sealed vessel at room temperature for at least three (3) days to equilibrate.A calibration curve was measured based on three different dextran concentrations (0.5, 1.0, and 1.5%), using the optical rotation of polarized light using a Polartronic M100 Touch polarimeter (Schmidt + Haensch GmbH & Co., Berlin, Germany), operating at 586 nm with a resolution of 0.001° (angular degree) and precision of ± 0.005° at 589 nm.The liquid from the immersed samples was subsequently filtered through a Puradisc syringe filter (Whatman, Maidstone, UK) equipped with a 0.45 µm polytetrafluoroethylene membrane in a polypropylene housing (VWR International AB, Stockholm, Sweden) and measured in the polarimeter.The concentration of dextran in the sample was determined using the previously established calibration curve for the optical rotation of polarized light measured.
Dynamic light scattering was used to determine the hydrodynamic diameter of the dextran molecules at high dilution in deionized water (Zetasizer ZEN3600; Malvern Instruments Ltd., Malvern, UK), using a He-Ne 4.0 mW, 633 nm laser and a detector angle of 178°.The hydrodynamic diameter of the dextran molecules was found to be 97 ± 2 nm with a polydispersity index of 0.2, measured at a dextran The length of the arrows shows the degree of delignification.Pulps are denominated KX_OY, where X is the kappa number of the cooked pulp and Y is the kappa number of the oxygen delignified pulp.The trial with double "O" means double oxygen delignification stage without washing in between the stages.Single oxygen trials were performed at 100 °C and concentration of 0.15 g per L solution.Based on the determined size of the dextran, the results obtained for the FSP were interpreted as representing liquid contained in pores smaller than approximately 97 nm in diameter.The FSP value is expressed as the unitless ratio of the mass of pore water to the mass of dry solids (g/g).For each sample, the FSP measurement was done in triplicate.

Small angle and wide angle X-ray scattering (SAXS and WAXS)
Measurements were performed on an Anton Paar SAXSpoint 2.0 system (Anton Paar, Graz, Austria) equipped with a Microsource X-ray source (Cu Kα radiation, wavelength 0.15418 nm) and a Dectris 2D CMOS Eiger R 1 M detector with 75 by 75 µm pixel size.All measurements were performed with a beam size of about 500 µm in diameter, at a sample stage temperature of 25 °C with a beam path pressure at 1-2 mBar.For SAXS measurements, the sample to detector distance (SDD) was 561.9 mm.For WAXS measurements, the sample to detector distance (SDD) was 109.1 mm.Samples were mounted on a Multi-paste Holder mounted on a Heated Sampler and a VarioStage (Anton Paar, Graz, Austria) and kept between Kapton foils in hermetically sealed compartments and were not exposed to vacuum during measurement.10 frames of 6 min duration were recorded, giving a total measurement time of 1 h per sample, the transmittance was determined and used for scaling the scattering intensities.Scattering data recorded on deionized water with the same experimental setup was used for background subtraction, an arbitrary scaling of the water signal was used.Binning was used to generate graphs with between 500 and 510 data points, which were used for graphs and modelling.The software used for instrument control was SAXSdrive version 2.01.224 (Anton Paar, Graz, Austria), and post-acquisition data processing was performed using the software SAXSanalysis version 3.00.042(Anton Paar, Graz, Austria).
The average crystallite size was determined using the Scherrer equation (Wikipedia 2022) based on the full width at half maximum (FWHM) of the cellulose I lattice (200) signal and a shape factor (K parameter) of 0.9.The (200) FWHM was obtained by fitting (non-linear least-squares) the WAXS diffractogram, after background subtraction, using Gaussian functions, subsequently calculating the FWHM for the Gaussian function used model the (200) signal.
SAXS results were evaluated by modelling using a method, developed by Larsson et al. (2022).The SAXS modelling method was used to obtain quantitative information on the apparent average particle and cavity sizes as reported by the modelling method.For each pulp, one SAXS measurement was modelled.For water swollen pulps, a SAXS measurement on pure water was used for background subtraction prior to modelling.Scaling factor for subtraction of SAXS data recorded on water, were calculated from the fiber saturation point measurement of the pulp.

Total fiber charge
The total fiber charge was determined by conductometric titration according to the method described by Katz et al. (1984), where the pulp in the H + protonform was dispersed in deionized water with NaCl and HCl and titrated with 0.1 M of NaOH.The titration was carried out in a microprocessor controlled titrator (Metrohm-Titrino 702SM) and the data was treated in the Tiamo 2.3 software.The measurements were done in duplicate.

Kappa number
The kappa number of the pulps was measured according to the standard ISO 302:2004 and the measurements were done in duplicate.

Carbohydrate composition
The carbohydrate composition was determined according to SCAN-CM 71:09 and the measurements were done in duplicate.The xylan, glucomannan and cellulose contents were calculated from the monosaccharides (Table S1) according to Janson (1974), using a uronic acid content of 4.25 and a glucose:mannose ratio of 4.17.

Carbohydrate composition
Carbohydrate composition and total fiber charge were determined-Table 1.The oxygen delignified pulps Vol:.( 1234567890) presented a higher total fiber charge when compared to the kraft cooked pulps, which was expected.The pulps subjected to extended oxygen delignification (K91_O53 and K91_OO37) obtained the highest total fiber charge.

Fiber saturation point
The fiber saturation point, FSP, gives the mass of liquid in the fiber wall voids in a water swollen fiber as g water/g solid material.Figure 3 shows FSP for unrefined pulps delignified with either kraft cooking or kraft cooking followed by oxygen delignification.
Kraft cooked pulps with kappa numbers ranging from 91 down to 28 showed a small variation in FSP, from 1.3 to 1.6 g/g.No correlation between FSP and degree of delignification was observed, in fact the pulps with the greatest difference in delignification degree; K91 and K28, had the same FSP (1.7 g/g).
These results are in line with the study by Andreasson et al. (2003) where water retention value (WRV) measurements showed small changes (1.6-1.8 g/g) for pulps with kappa numbers in the range 110-60.A maximum was observed at kappa number 35 (2.0 g/g), and continued delignification to kappa number 15 gave a WRV of 1.8.Since lignin provides rigidity to the fiber wall, it is likely that at some point, voids created by the depletion of lignin could collapse and thereby result in a decrease in FSP and WRV.
When kraft cooked pulps were further delignified by oxygen, the fiber saturation point was unaffected in two cases (broken black arrows in Fig. 3) which is in accordance with previous studies (Esteves et al. 2021a).However, when the kraft pulp with kappa number 91 was oxygen delignified to a kappa number of 37 (K91_OO37), an increase in FSP was obtained (red broken arrow in Fig. 3).One possible explanation can be the higher total fiber charges obtained after oxygen delignification, which can overcome the shrinking forces inside the fiber wall and help the pores maintain or increase their volume.The reduction in kappa number was achieved by performing the oxygen delignification in two stages.
Table 1 Chemical composition and fiber charge of the kraft cooked and oxygen delignified pulps (% in pulp) *The xylan, glucomannan and cellulose contents were calculated from the monosaccharides according to Janson (1974).The uronic acid content and the ratio between glucose and mannose are calculated depending on wood specie (for Pine is 4.6, for the mixture is 4.25).Uronic Acid, HexA, Acetic Acid are not analyzed.The galactoglucomannan ratio was 4.1 for pine and 4.17 for the mixture

WAXS-crystallite size
The crystallite size was obtained from WAXS data (Figures S3 and S4) by the use of the Scherrer equation.The crystallite size determined from the (200) signal was interpreted as representing the width of the crystalline inner part of the cellulose fibril, which typically correlates with the average lateral fibril dimensions that can be obtained from CP/MAS 13 C-NMR measurements.As seen in Fig. 4, the crystallite size was about 5 nm in the unrefined pulps regardless of the kappa number which suggests that the crystallite size is not affected when the kappa number is decreasing from 91 to 28.The crystallite size obtained in the present study is in concordance with previous studies.The crystallite size in birch pulp fibers delignified to kappa numbers below 20 was reported to be between 4-5 nm (Borrega et al. 2018).Hult et al. (2003) also showed a crystallite size of 4 nm in spruce kraft delignified pulps.Crystallite sizes of 2.5 to 3.5 nm were reported in wood chips subjected to either acidic or alkaline pretreatment which removed only a minor part of the lignin (Brännvall et al. 2021).A clear correlation between increased crystallite size and increased cellulose content in bagasse pulp fibers was reported by Driemeier et al. (2011).The crystallite size was around 3 nm in the starting material and as the removal of lignin and hemicellulose increased, the crystallite size increased as well, ending up at 5 nm (Driemeier et al. 2011).
It was suggested that crystallite growth would occur as non-crystalline cellulose crystallizes on crystallite surfaces.

SAXS-modelling
A method, developed by Larsson et al. (2022), was used to model experimental SAXS data (Figures S1  and S2) to obtain quantitative information on the nanostructure of the fiber wall.The modeling method describes the recorded SAXS profile as a superposition of components, each carrying information about the apparent average cavity size (AACS) and the apparent average particle size (AAPS) for each component.Further, the relative weight of each model component can also give information on changes occurring in the fiber wall nanostructure as a result of altered processing conditions.For all pulps examined in this study, three intensity components (I1, I2 and I3) were used to model the experimental SAXS intensity (Figure S5).The ranges obtained by modelling the apparent average cavity size (AACS) and apparent average particle size (AAPS) for each component are summarized in Table 2.
Table 2 shows that for all pulps in this study, the SAXS modelling could distinguish three well-separated size ranges for both the apparent average cavity and particle sizes.This was interpreted as being consistent with the well-known hierarchical structure of the fiber wall.Further, the non-overlapping size ranges for AACS and AAPS, respectively, simplify comparisons between samples based on the relative weights of the modelled components.
The pulps used in this study contained different amounts of cellulose, lignin and hemicellulose, why modelling SAXS results as a two-component system (solid particles and water-filled cavities) is an  approximation.Because of the chemical composition of the pulps, all three components, cellulose, lignin and hemicelluloses, can influence the AAPS and the AACS obtained by modelling.
Although the size ranges for AACS and AAPS averages associated with the modelled intensity components were not overlapping, their interpretation in terms of structural element types may be overlapping due to the inherent variability in structures such as cellulose fibril aggregates and fibril aggregate composites of cellulose, lignin and hemicellulose (Fig. 5).For these reasons, the interpretation of the size-ranges associated with the modelled intensity components (I1, I2, I3) was made as follows: • The I1 component was interpreted to model composite structure describing larger-scale structure, such as cellulose fibril aggregates and fibril aggregate composites of cellulose and lignin (I1 AAPS) and the interstitial cavities between such particles (I1 AACS).In pulps with high kappa numbers, possibly also lignin agglomerates.• The I2 component was interpreted to describe the width of cellulose fibrils and fibrils associated to lignin and/or hemicellulose (I2 AAPS) and smaller interstitial cavities between fibril aggregates/aggregate composites (I2 ACCS).
• The I3 components was interpreted to describe cellulose fibrils of smaller width, not yet coalesced into their large counterpart due to the presence of lignin and hemicellulose (I3 AAPS) and smaller-scale cavities in fibril aggregate/aggregate composites (I3 AACS).

Pulps-intensity component weights
From Fig. 6, it is seen that the relative weights of the model components I1 to I3 were similar for all the kraft pulps, apart from the K91 and the subsequent oxygen delignification was not associated with any major changes in the relative weights.The K91 pulp showed a significantly higher weight for component I3, representing a relatively higher abundance of smaller structure.The reason for this is not clear but could be related to the higher lignin content in the K91 pulp.For the component I2 the greater difference was seen for K52 and K52_O42.As for K91 the reason for that is not fully understood once is higher than the following samples with higher delignification, however, the difference is maintained after oxygen delignification which makes it consistent.
The relative weights of the model components I1 to I3 remained unchanged for the oxygen-delignified pulp, compared with the pulps prior to oxygen delignification (indicated by the horizontal arrows in Fig. 6).The interpretation made from this was that the main structural changes in the fiber wall occurred during the delignification induced by the kraft cooking, even though the pulps present significant differences in the chemical composition (seen in Table 1).

Pulps-apparent average of particle size (AAPS)
In Fig. 7, the apparent average particle sizes (AAPS) are shown.The largest solid structures in kraftcooked pulps, represented by I1, show no clear correlation with the degree of delignification.The smallest AAPS value (12.5 nm) was obtained for the pulp with the highest kappa number (K91, highest lignin content), reaching a maximum at 18 nm for the pulp with kappa number 31, and decreasing to about 15 nm for the K28 pulp.Measurements and modelling were repeated for the K28 pulp (K28 rep) with similar results.The size of fibril aggregates in kraft pulp is normally in the range of 17-30 nm (Brännvall et al. 2021;Hult et al. 2003;Larsson et al. 2017) so it is likely that the particle size obtained for the I1 component relates to the fibril aggregates.The values  2022) for the I1 component for solid structures (AAPS) in softwood and hardwood pulps in a water-swollen state were much larger, 78 and 48 nm, respectively.However, all pulps in Larsson et al. (2022) had higher cellulose purity than the pulps used in the present study and had been dried and re-swollen prior to measurements and modelling, whereas the pulps in the present study were never-dried.Drying of cellulosic materials is well-known to cause hornification associated with an increase in the degree of cellulose by aggregation and partial fiber wall pore collapse, hornification becomes most severe in cellulose-rich pulps.
The I2 component for AAPS in kraft-cooked pulps was approximately 4 nm, which correlates well with the crystallite size of 5.0 nm obtained from Fig. 4. It is also similar to the lateral fibril dimensions measured by CP/MAS 13 C-NMR (Brännvall et al. 2021).
The smallest solid structure, obtained from the I3 component, was 2 nm, which is significantly smaller than the lateral fibril dimensions.This is similar to findings for fibril diameters estimated by SAXS measurements made on wood tissue (2.5 nm) and could represent remnants of a state of cellulose order present prior to isolation and drying (Jakob et al. 1994).
AAPS for the three modelled intensity components remained essentially unaffected by oxygen delignification (shown by the arrows) in all studied cases, despite the decrease in lignin content (Table 1).Fibril aggregation does not seem to be occurring during oxygen delignification, not even for extended delignification degree, such as for pulp K91_OO37.

Pulps-apparent average cavity size (AACS)
The apparent average cavity size (AACS) describes the water-filled interstitial spaces between solid particles.The intensity component I1, representing the largest structures, was in the size range from 29 to 40 nm.The intensity component I2 ranged from 8 to 11 nm and I3 from 4 to 5.5 nm-Fig.8.
The size range of the intensity component I1 agrees with the reported values in the literature for larger pore sizes.At the very initial stages of pulping the pores in never-dried softwood fibers were 40-65 nm in diameter when measured by FSP and CP/MAS 13 C-NMR (Brännvall et al. 2021).Forsström et al. (2005) reported pore sizes of 30-35 nm diameter in pulps with kappa numbers ranging from 110 to 16. Kimura et al. (2014) used BET analysis on softwood fibers and reported a wide size range of 10-20 nm in diameter and also mesopore diameters in the range of 4-5 nm.These two size ranges agree with the size range of the intensity components I2 and I3, respectively.Lovikka et al. (2016) gave an average pore width of 8 nm for softwood, but also report pores of 3-4 nm.
The AACS were unaffected when the K91 pulp was oxygen delignified to kappa 53 (K91_O53) and when the K52 pulp was oxygen delignified to kappa number 42 (K52_O42).However, when oxygen delignification was performed on the K91 pulp in two stages down to kappa number 37 (K91_OO37), the cavity size increased.The same behavior was seen for FSP, Fig. 3.A higher FSP was obtained after oxygen delignification only in the case of K91_OO37.

Refining effect
Refining is well known process to induce macroscopic effects on the fibers, such as fiber cutting, internal and external fibrillation (Page 1985(Page , 1989) ) leading to higher porosity and therefore higher fiber swellability (Clark 1985).However, the refining effect on the supramolecular structure of fibers is still unclear.

Refined pulps-fiber saturation point
Figure 9 shows the effect of PFI-refining on FSP for kraft-cooked and oxygen-delignified pulps.As expected, after 4000 PFI-revolutions the FSP increased for all pulps, regardless of the delignification process.Continuing refining to 6000 and 10 000 PFI-refining for kraft-cooked pulps (K91 and K52) had no significant change in FSP, suggesting that extensive refining does not affect the internal fiber wall structure further-Fig.9a.Oxygen delignified pulps seemed to have a more significant increase in FSP after 4000 PFI-revolutions than the cooked pulps-Fig.9b.While for kraft-cooked pulps the values seem to stabilize near 2 g/g, for oxygen-delignified pulps the FSP value after 4000 PFI-revolutions is already higher; 2.5 g/g.
Internal fibrillation is the main refining effect on FSP values, as refining will open up fiber pores leading to the FSP increase (Maloney and Paulapuro 1999;Stone et al. 1968).The increase in FSP is due to the random breaking of the crosslinks in the interfibrillar matrix during refining (Page 1989).The levelling-off of FSP seen for kraft-cooked pulps with a higher refining degree, is due to the internal fibrillation stabilization, also reported before (Kang and Paulapuro 2006).

Refined pulps-crystallite size
The crystallite size does not change significantly due to refining, the values are in the range of 4.7 -5.6-Table 3.

Refined pulps-Intensity component weights
Relative weights of modelling components I1, I2 and I3 are shown in Fig. 10 for kraft-cooked pulps with kappa number 52 and 91.Refining led to systematic changes in the weight of the I2 and I3 components.Refining increased the weight of both the largest (I1) and smallest (I3) structures, while the intermediate (I2) structures decreased.This suggests that refining can break apart the aggregated particles into smaller structures.Pulp K91 seems to have higher weight of smaller structures than K52 (I3-component).This could be related to the lower delignification degree when compared to the K52, resulting in lower degree of fibril aggregation.
Looking at oxygen-delignified pulps, Fig. 11, the same behavior was seen as for kraft-cooked pulps; refining increased the abundance of the larger and smaller structures (I1 and I3 components), while decreasing the abundance of the intermediate sizes (I2-component) sizes.
A significant decrease in medium structures is seen for K52_O42 due to the higher weight from the kraft cooked pulp K52.If we consider that pulp K52 had higher fibril aggregation than K91, due to longer delignification, it is suggested that refining is breaking apart a higher amount of aggregates.All the pulps seems to end up with similar values after refining.

Refined pulps-apparent average of particle size (AAPS)
Figure 12 shows the AAPS before and after refining for kraft-cooked pulps.Refining resulted in only small changes in the AAPS for all three intensity components.The effects on AAPS were not systematic with respect to degree of refining.
Neither was the particle size of oxygen-delignified pulps significantly affected by refining-Fig.13.

Refined pulps-apparent average of cavities size (AACS)
Figures 14 and 15 show the AACS after PFI-refining of kraft-cooked and oxygen-delignified pulps.Unlike AAPS, AACS values are significantly affected by refining.Clearly, the interstitial spaces between solid particles increased after refining to   all the pulp samples at all three size levels scales described by the modelled intensity components I1, I2 and I3.
For the kraft-cooked pulps, large cavities (I1) increased from 30 to 46 nm after refining.The I2 increased from 8 to a range of 11.4-12.6nm and I3, from 3.8 to approximately 6 nm.
In the oxygen delignified pulp the difference in AACS was also significant-Fig.15.For the oxygen delignified pulps, the large cavities (I1) increased from 30 to 48 nm, I2 increased from 8 to a range of 14 nm and I3, from 3.8 to approximately 7 nm.
The results are expected since, as reported previously, refining tends to increase the interstitial spaces due to the random fracture of the interfibrillar matrix (Maloney and Paulapuro 1999;Page 1989;Stone et al. 1968).
Figure 16 illustrates how the refining can affect the structures, by increasing the cavities sizes and breaking apart the medium structures.

Conclusions
The effect of delignification and refining on the fiber wall nanostructure was studied using X-ray scattering techniques.
Results from SAXS data modelling suggests that refining can break apart aggregated particles into smaller particles, as the abundance of smaller structures increased while the abundance of the intermediate sizes decreased.Refining had a pronounced impact on cavity sizes, increasing the size of small, medium and large cavities.
Delignification had a smaller impact on the fiber wall nanostructure.WAXS results showed that neither kraft cooking nor combined kraft cooking and oxygen delignification affected crystallite size.It remained at 5 nm whether the kappa number was 91 or 28.Modelling of SAXS data showed that the apparent average particle size, AAPS, corresponding to the medium size structures in the fiber wall was approximately 4 nm for all pulps, which correlates well with the crystallite size.The main change in AAPS occurred for the largest particles, reaching a maximum of 18 nm at kappa number 31, and being lower both at higher and lower kappa number.The AAPS was same after oxygen delignification as it had been in the kraft cooked pulp, indicating that fibril aggregation does not occur during oxygen delignification, even at high degrees of delignification.SAXS data modeling also revealed a solid structure of 2 nm, which is significantly smaller than the lateral fibril dimensions.It is hypothesized that this could represent remnants of cellulose structures present prior to isolation and drying.

Fig. 2
Fig. 2 Schematic representation of the kraft cook trials (black arrows), followed by oxygen delignification (white arrows).The length of the arrows shows the degree of delignification.Pulps are denominated KX_OY, where X is the kappa number of the cooked pulp and Y is the kappa number of the oxygen delignified pulp.The trial with double "O" means double

Fig. 3
Fig. 3 Fiber saturation point as a function of kappa number for unrefined pulps subjected to a kraft cook and oxygen delignification; the dashed arrows connect the original kraft-cooked pulp to the final oxygen-delignified pulps

Fig. 4
Fig. 4 Crystallite size as a function of kappa number for unrefined pulps subjected to kraft cook and oxygen delignification, determined by WAXS.The standard deviation errors are represented in the figure but are not visible

Fig. 5
Fig. 5 Illustration of how the fibrils, cellulose and lignin are organized before and after delignification.The figure intends to illustrate the meaning of the different modelled components obtained from the SAXS analyses

Fig. 6
Fig. 6 Relative weights the three modelled intensity components (I1, I2 and I3) used to model the recorded SAXS data for unrefined a kraft cooked pulps and b oxygen delignified

Fig. 8
Fig. 8 AACS for the three modelled intensity components (I1, I2 and I3) used to model the recorded SAXS data for unrefined a kraft cooked pulps and b oxygen delignified pulps.The com-

Fig. 9
Fig. 9 Fiber saturation point for unrefined and refined pulps subjected to a a kraft cook and b oxygen delignification

Fig. 10
Fig. 10 Relative weights for the three modelled intensity components (I1, I2 and I3) used to model the recorded SAXS data for kraft cooked PFI-refined pulps a K52 and b K91.The com-

Fig. 14
Fig. 14 Apparent average cavity size for kraft-cooked pulps unrefined and PFI-refined

Table 2
The AACS and AAPS ranges obtained by modelling the SAXS data recorded on the pulps