CPMAS 13C NMR Analysis of Fully Bleached Eucalypt Pulp Samples: Links to Handsheet Hygroexpansivity and Strength Properties
The objective of this study was to examine the crystallinity of the cellulose of seven fully bleached pulp samples by using solid-state cross-polarization/magic angle spinning (CP/MAS) 13C Nuclear Magnetic Resonance (NMR) spectra in combination with line-fitting analysis and how it affected the hygroexpansional properties of handsheets. Differential scanning calorimetry was used to measure the pore size distribution of the samples and to characterize the water inside the fiber wall. Combined with the CP/MAS 13C NMR analysis, mechanisms behind the moisture absorbing capacities of various hardwoods were studied. Based on the results, the hygroexpansional behavior of the handsheets and the Crystallinity Index of the cellulose in the samples were not inter-correlated. The fiber wall thickness measured with a FiberLab® fiber analyzer had a moderate correlation with the amount of crystalline cellulose and xylan of the samples, implying that the crystalline and the amorphous regions were evenly distributed in the fiber wall.
Solid state CP/MAS 13C NMR spectroscopy is a versatile tool for
morphology studies of cellulosic materials (Atalla and Hart,
1999; Hult et al., 2001, 2003).
The supra-molecular structures of cellulose were investigated by using CP/MAS
13C NMR spectra along with non-linear spectral fitting, which analyzes
the crystalline allomorph and disordered domains (Larsson
et al., 1995, 1997; Wickholm
et al., 1998; Hult et al., 2003).
The line-shape spectral fitting analysis of solid state spectra allows a very
detailed comparison and characterization of cellulose supra-molecular structures
and has been extensively used to investigate the structural characteristics
of cellulose and its derivatives. The relative amounts of different cellulose
forms in complex cellulosic materials can be determined by using spectral fitting
analysis of the C4-region of CP/MAS 13C NMR spectra (Fig.
1). The ratio of crystalline and amorphous cellulose can be characterized
with a crystallinity index (CrI) (Newman and Hemmingson,
1990; Liitiä et al., 2003).
The microfibrils in bleached chemical pulp fibers consist of two regions of
cellulose: crystalline and amorphous (Bertran and Dale,
1986). There is also a region between these two forms of cellulose, normally
referred as para-crystalline cellulose (Larson et al., 1997). The highly
ordered cellulose fibrils are barely accessible to water molecules. The amorphous
regions consisting of less-ordered cellulose molecules and hemicelluloses absorb
water. In cell walls, only 60-70 % of cellulose chains are in crystalline form
(Alince, 2002). The rest are considered disordered (amorphous)
and fully accessible to water molecules. A trace amount of adsorbed water on
the surface of crystalline region is almost negligible, the amount of absorbed
water in fibers, also called the bound water, almost totally relies on the accessible
sorption sites of amorphous regions.
||Spectral fitting for the C-4 region of the CP/MAS 13C
NMR spectrum of a birch kraft pulp sample
Determination of the size of amorphous region of fibers can be conducted by
Differential Scanning Calorimetry (DSC), which is commonly used for determination
of bound water in fibers (Maloney et al., 1998;
Salmén and Berthold, 1997). In this study, DSC
was used to characterize the pore structure for different hardwood fibers. Focus
was also placed on the effect of cellulose crystallinity on the pore size distribution.
Hygroexpansion is a paper property of great interest in many applications,
for instance in printing operations, but also in relationship to mechano-sorptive
creep. Paper expands or shrinks as the relative humidity in the local environment
changes. When hygroexpansivity is plotted against relative humidity, paper generally
shows complex hygroexpansion behavior, dependent on its humidity history before
and during testing (Uesaka, 2002; Salmén,
1993). Sheet dimensional changes occur in concert with the changes in fiber
dimensions. At the hydrogen-bonded fiber crossing areas, the more stable direction
of a fiber is able to restrict change in the less stable direction of another,
so that intra-planar dimensional changes are far smaller than those in the z-direction.
The hygroexpansivity is highly influenced by restrictions affecting the orientation
of both the cellulose and the hemicellulose chains (Salmén
et al., 1987a, b; Uesaka,
2002). The dried-in stresses in hemicelluloses can be released by increasing
the relative humidity to about 50-60%, thus increasing the hygroexpansivity
of paper if it has been dried under restraint to very low humidities. As hemicelluloses
and the amorphous cellulose are responsible for the water uptake of pulp fibers
(Salmén and Berthold, 1997), NMR studies provide
information about the influence of the crystallinity degree and the xylan content
in the fiber wall on hygroexpansivity and strength properties.
The purpose of this investigation was to examine how various hardwood samples differ in terms of their degree of cellulose crystallinity and their pore size. With the information obtained, the characteristics of the hygroexpansional behavior of handsheets made of bleached hardwood kraft pulp samples in relation to cellulose crystallinity can be studied. Statistical analysis was used to find links between the NMR data and the strength and structural properties of the handsheets.
MATERIALS AND METHODS
Pulp samples and carbohydrate testing: Bleached industrial hardwood
kraft pulps (Eucalyptus grandis E. dunnii, Uruguay; Betula
pendula/B. pubescens, Finland; Acacia magnium, Indonesia; E. globulus,
Portugal; E. urograndis/E. grandis, South Africa; E. grandis/E. saligna,
Brazil; E.urograndis, Brazil) were obtained from pulp mills as dry sheets.
Laboratory handsheets with a grammage of approximately 60 g m-2 were
prepared in a conventional sheet former according to ISO5269-1:1998. Anisotropic
sheets were made with a dynamic sheet former at KCL Research Center. The wire
speed was set to 1000 m min-1. The jet speed was set to 920 m min-1
(1.6 bar). The feed pulp concentration was 4 g L-1. The grammage
of anisotropic sheets was 60 g m-2. Tensile properties were measured
according to EN ISO 1924 and tensile stiffness was measured according to SCAN-P
67:93. The handsheets were prepared and tested at KCL research center (Espoo,
The carbohydrate composition of the pulps was determined using the following method: utilizing sulfuric acid hydrolysis at 120°C to degrade the carbohydrates into monosugars, based on Tappi standard T222-om0. The monosugars were then analyzed by ion chromatography (Metrohm 817 Bioscan system). The carbohydrate measurements were conducted in Botnia mill at Kemi.
The FiberLab® measurement apparatus consists of an analyzer and a s ample
unit. Fiber images are captured by two CCD cameras. The direct results are the
fiber length, fiber width and fiber wall thickness. The calculated values are
the curl index, coarseness, cross-sectional area and volume index. Procedures
of fiber and image processing to obtain fiber properties have been described
in detail elsewhere (Anonymous, 2006). Kajaani FiberLab®
data for the hardwoods was recorded at the Botnia mill in Kemi. This study was
conducted in Espoo, Finland, from 2nd of January to 5th of June 2009.
NMR measurements: The solid-state CP/MAS 13C NMR experiments
(Atalla and Hart, 1984) were performed on a Bruker Avance-400
spectrometer operating at frequencies of 100.59 MHz for 13C. All
the experiments were carried out at ambient temperature using a Bruker 4-mm
MAS probe. The pulp samples (~45% moisture) were packed in a 4 mm ZrO rotor.
CP/MAS 13C NMR data was acquired with 8000 scans accumulated per
sample. The samples were analyzed at Georgia Institute of Technology, Atlanta,
Pore measurements: The thermoporosimetry measurements were carried out
using a Mettler DSC 821. In the technique used in this study, the energy that
is absorbed when water in frozen pulp fibers is melted in a stepwise manner
approaching 0°C is measured. The melting energy is assumed to be directly
proportional to the amount of melted water. This technique is based on the fact
that water contained within pores sufficiently small or inside a swollen hemicellulose
gel is at an elevated pressure and therefore has a depressed melting temperature.
The details of the measurements have been described by Maloney
and Paulapuro (1999). The pore size measurements were conducted at Helsinki
University of Technology.
Measurement of hygroexpansion: Hygroexpansion was measured on test pieces
subjected to a relative humidity of 66% RH prior to testing. A non-standardized
apparatus based on laser measurement of the dimensional change in the sample
length was used to measure the hygroexpansion of paper strips. The measurements
were performed at STFI Packforsk, with an instrument similar to a previous version
described by Salmén (1993). It consists of 30
rigid clamps and 30 freely movable clamps, with 100-mm gaps between the clamps,
in which 30 paper strips are placed independently in a horizontal position.
A weight is placed on the strips to eliminate any effects ofbuckling on length
registration. These clamps were placed in a chamber with regulated humidity.
The relative humidity was changed from 50±2%RH to 22±3%RH to 33±2%RH
to 66±2%RH and the length changes between 33±2%RH and 66±2%RH
The hygroexpansion coefficient (β) was determined from the gradient of
the linear part of the curves in the low-moisture range and is expressed as
the hygroexpansional strain divided by the moisture change of the sample (Nanri
and Uesaka, 1993).
Data analysis of the results: Table 1 shows the results
of the correlation analysis (Milton and Arnold, 1995)
performed between the measured and calculated pulp fiber properties, the handsheet
properties of un-refined fibers and the NMR data. Since, the number of samples
was only seven, the critical value for making the results statistically significant
for a two-tailed test (<5%) was 0.75. Table 2 shows the
correlations between the NMR data and the hygroexpansivity measurements.
||Correlations of NMR data, handsheet strength properties and
fiber dimensions (2-tailed test, statistically significant (p<0.05).
Level of significance > 0.74 (7 samples)
|A correlation analysis on the crystallinity index and xylan
content (relative amount in percent), strength properties of dynamic handsheets
in machine and cross direction and fibre characteristics (fibre length,
widgth, wall thickness and fibre curl). 1: Crl%, 2: Xylan% (NMR), 3: Tensile
stiffness index (kNm g-1) (MD), 4: Tensile stiffness index (kNm
g-1) (CD), 5: Elastic modulus (Mpa). 6: Tensile index (Nm g-1)
(MD), 7: Tensile index (Nm g-1), (CD), 8: Fibre wall thickness
(m), 9: Fibre width (m), 10: Fibre length (m), 11: Curl (%)
||Correlations of NMR data and the hygroexpansivity measurements
(2-tailed test, statistically significant (p<0.05). Level of significance>0.74
|A correlation analysis on the crystallinity index and xylan
(measured with NMR in relative amount in percent), the hygroexpansional
properties of isotropic and dynamic handsheets and the mean and standard
deviation of fiber width and fiber wall thickness. The hygroexpansitivities
obtained from Pulkkinen et al. (2009b)
||The line-fitting results of the C-4 region of CP/MAS 13C
NMR spectrum of pulp sample (values are relative intensity as a percentage,
|Values taken from Pulkkinen et al.(2009b)
The crystallinity index (CrI) has been considered an important factor for monitoring
the structural features of the cellulosic substrate. The calculated crystallinity
index (CrI) values are presented in Table 3 along with the
rest of the results of the spectral fitting. CrI was calculated as a percentage
of the sum of crystalline and para-crystalline cellulose fraction in the C-4
signal cluster (Lennholm et al., 1994). The correlation
between the values determined by ion chromatography and those determined by
CP/MAS 13C NMR was good (Fig. 2), implying that
there is a certain fraction of xylan that is intimately associated with the
cellulose, as has been suggested for bleached birch kraft pulp (Liitiä
et al., 2003). The correlation between the results of carbohydrate
analysis and the relative amount of xylan observed from the spectral analysis
also gives us confidence that xylan amounts between samples of this study are
comparable. Recent evidence of the association between xylan and cellulose based
on dynamic FT-IR has also been presented by Dammström
et al. (2009).
Based on Table 1 and 2, the crystallinity
index (CrI) and xylan content of the samples had good correlations with tensile
strength properties in the cross direction and moderate correlations with the
fiber wall thickness index and fiber width. With an increasing cellulose crystallinity,
tensile strength in the cross-direction decreased, probably due to the lower
amount of xylan in the fiber wall (Table 1). With higher xylan
contents the strength of the handsheets increased. This kind of behavior with
regard to strength properties has been confirmed earlier (Annergren
et al., 1963; Kettunen et al., 1982;
Kibblewhite and Bawden, 1989), as samples with a higher
amount of crystalline cellulose produce sheets with lower tensile strength.
The only significant correlation observed for the hygroexpansional data was
between the hygroexpansivity of the handsheets and the xylan content of the
fiber wall. The negative correlation can be at least partly explained by increasing
fiber wall thickness value as the xylan content increases. More coarse fibers
decrease the hygroexpansivity coefficient of the handsheets (Pulkkinen
et al., 2009a).
There seems to be no apparent correlation between the measured hygroexpansion
coefficients and the cellulose crystallinity index.
||The correlation between xylan measured chromatographically
and xylan measured with NMR
Hygroexpansivity of isotropic handsheets (diamonds) and anisotropic
(dynamic mean value) handsheets; in CD (triangles), in MD (crosses) and
the geometric mean (squares) for laser measurement apparatus (Pulkkinen
et al., 2009b
Only a weak correlation
was observed for conventional handsheets (Table 2). The observed values for hygroexpansion of conventional and dynamic handsheets
were extracted from our previous publication (Pulkkinen et
al., 2009b) and are shown here as a function of crystallinity index
in Fig. 3.
The water absorption capacity of pulp fibers due to the amount of amorphous
regions of the fibers wasindirectly detected with the Differential Scanning
Calorimetry (DSC). The results are shown in Table 4 forthe
pulp samples studied. The amount of non-freezing water and bound water obtained
with the DSC should be quite similar for never-dried and dried samples (Salmén
and Berthold, 1997; Maloney and Paulapuro, 1999).
Therefore, the differences in drying of the pulp samples were not assumed to
cause any differences in the measurements with the DSC (Salmén,
1993; Maloney and Paulapuro, 1999).
Since, the pulps were obtained from different sources, the fact that the amount
of lignin and hemicelluloses varies between samples has to be taken into account.
This will affect the amount of bound water measured with the DSC (Salmén
and Berthold, 1997) as pore sizes develop into larger ones when the removal
of lignin and hemicelluloses progresses further. As shown in Fig.
4a and b, the mean pore size measured with the DSC is a function of both
the crystallinity index and the amount of xylan measured with NMR. Therefore,
it seems that fibers with low xylan content had a slightly lower mean pore size
independent of the amount of water contained in the amorphous regions of the
fiber wall (Table 4). This is consistent with the observations
by Duchesne et al. (2001). They observed that
the fibril aggregates (macrofibrils) with high hemicellulose content were well
organized with a porous structure. Fibril aggregates with low hemicellulose
content had a smoother, flatter and more compact surface structure.
No correlation between the non-freezing water content and the crystallinity of the cellulose was observed. The amount of non-freezing water was approximately constant for all samples (~0.3 g/g). The fiber wall thickness measured with a FiberLab® analyzer and the crystallinity of the cellulose had a moderate correlation with the crystallinity index (Fig. 5).
The interpretation of the xylan from the spectral data is not straightforward
and should be handled with caution (Liitiä et al.,
2003). Therefore, the observed weak correlation between the fiber wall thickness
and the xylan content (Table 1) may be due to the mild affinity
between the cellulose and the hemicelluloses.
As a fiber analyzer can only detect the birefringent part of the fibers, i.e.,
cellulose (Piirainen, 1985; Olsson
et al., 1995), the fiber wall thickness value may be partly influenced
by the fiber detection procedure (Jordan and ONeill,
1994). Keeping in mind the principles of fiber wall detection with an analyzer,
the most important observation is the correlation between the fiber wall thickness
index and the crystallinity of the cellulose.
The increased hydrogen bonding, where the fibrils can form tighter aggregates
when the amount of xylan decreases, seems to affect the strength properties
of thefiber network. If the effect of the transverse dimensions of the fibers
on network strength is based on the open micropore structure present in un-dried
fibers, after drying the transverse dimensions lose their impact on the strength
properties unless refining is applied.
The hygroexpansion coefficient only covers the low range of moistures of the
samples (Nanri and Uesaka, 1993). It has been previously
observed by Pulkkinen et al. (2009a) that samples
that indicate a low hygroexpansivity coefficient often undergo major changes
in sheet dimensions across the whole humidity range.
The mean pore size as a function of the (a) Crystallinity
Index (CrI)(right) and as a function of (b) xylan measured with NMR (left)
||Results of the DSC analysis. NFW = non-freezing bound water
|FBW: Free bound warter (mL g-1), TBW: Total bound
water (mL g-1)
||Fiber wall thickness as a function of the crystallinity index
||Hygroexpansivity of isotropic handsheets as a function of
mean pore size measured with a DSC
Low hygroexpansivity is believed to be associated with a high fiber wall thickness
index (Pulkkinen et al., 2009a; Nanko
and Wu, 1995).
The low amount of total water absorbed by the sample of A. magnium was probably affected by the low amount of xylan in the fiber wall. The amount of bound water in the sample was by far the lowest among those studied (Table 4). Also, the small average pore size probably affected the moisture absorbance of the fibers, enabling them to absorb water at the fastest rate, resulting in a high hygroexpansion coefficient. Further evidence is given by Fig. 6, which shows the hygroexpansivity of an isotropic handsheet plotted against the average pore size.
The findings of this study give us reason to speculate that the differences
seen between the strength properties of the unrefined samples are dependent
on the crystallinity degree of the cellulose and on the transverse dimensions
of the fibers. The hygroexpansivity of isotropic handsheets were seen to be
dependent on theaverage pore size of the fibers. The amount of amorphous regions
and xylan was also a determining factor in isotropic hygroexpansion.
The algorithm used in this study was not able to include Lorentzian lines for
modeling the signals originating from the highly crystalline domain in C(4)
region (Larsson and Westlund, 2005) and only Gaussian
lineshapes were used. In this study, the Iβ form of crystalline
cellulose (Atalla and Hart, 1984; Larsson
and Westlund, 2005) was not observed for the hardwood samples measured.
However, it is possible that this was a feature of the measurement accuracy
of the spectral analysis applied. Because of the previous facts it is possible
that the values for the Crystallinity Index or the relative amount of xylan
were not as accurate as those obtained with combined use of Lorenzian and Gaussian
lines. However, all studied spectra were treated in the same manner, so obtained
results are comparable within this study.
The moderate correlation between fiber wall thickness, fiber width and CrI
is an interesting and new finding. Based on this study, the crystallinity of
the fiber wall decreases (and the xylan content increases), as the fiber wall
thickness increases. One probable reason is the birefringence, i.e., double
refraction, of particles containing cellulose (Piirainen,
1985; Olsson et al., 1995). This would mean
that the hemicelluloses and amorphous regions of fibers would be undetectable
in a fiber analyzer. This may also hinder the observation of the linkage between
xylan content and hygroexpansivity. This contradicts with the positive correlation
between xylan and fiber wall thickness index observed in this study. This can
be partly due to the higher swelling ability of the outer surface of the fiber
due to the higher xylan content (Dahlman et al.,
2003). This can promote more easily detectable outer surface for a scanned
fiber flowing through a measurement capillary, as cellulose fibrils and xylan
are in close association with each other.
The detection of cellulose-xylan-matrix with the optical analyzers needs to
be studied in more detail to shed some light over the influence of the location
of the xylan on the detection process. Dahlman et al.
(2003) have observed that the outer layer (S1) of hardwood pulp fibers have
a higher relative content of xylan than the corresponding inner layer (S2+S3).
However, regarding the samples of this study, it is safe to assume that the
S1 layer has been removed during the pulp manufacturing process. The correlation
between the xylan contents and the hygroexpansivity of conventional handsheets
and the absence of correlation between the xylan content and oriented sheet
seem to imply that obtaining similar orientation for all the dynamic sheets
was unsuccessful. However, in our previous study (Pulkkinen
et al., 2009a), the effect of anisotropy, measured as the MD/CD ratio
of tensile strengths, was observed to be similar for allsamples. Due to the
fact that fibrils are always laid at an angle to the fiber axis, the orientation
of the hemicellulose chains may have more influence on the cross-fiber dimensions,
thus affecting the hygroexpansivity of the paper.
The objective of this study was to investigate how hardwood fibers that differ
in their ultrastructure and dimensions perform mechanically and what the effect
of ultrastructure is on the hygroexpansivity of paper. Based on these results,
the hygroexpansional behavior of the handsheets and the crystallinity index
of the cellulose in the samples were not inter-correlated. The fiber wall thickness
had a moderate correlation with the amount of crystalline cellulose in the samples,
implying that the crystalline and the amorphous cellulose were evenly distributed
in the fiber wall. The observed correlation between fiber wall thickness and
xylan content could be due to the mild affinity between the cellulose and hemicelluloses.
The water-absorbing capability of the fibers measured with a DSC did not give
indications of the hygroexpansional behavior of the fiber network.
The findings of this study led us to speculate that the differences seen between the strength properties of the unrefined samples used in this study were mainly dependent on the crystallinity degree of the cellulose and the transverse dimensions of the fibers. The dimensional stability of the dynamic sheet samples was not affected by the crystallinity of the cellulose. The distinction between the effects of xylan in fibers and fiber surfaces and fiber wall thickness could be investigated in a separate study using fibers with similar dimensions and varying amounts of xylan.
Sirkka-Liisa Maunu for her valuable comments during the preparation of this manuscript. Professor Arthur Ragauskas and Dr. Yunqiao Pu for providing the results of NMR spectroscopy.
1: Alince, B., 2002. Porosity of swollen pulp fibers revisited. Nordic Pulp Pap. Res. J., 17: 71-73.
2: Annergren, G., S. Rydholm and S. Vardheim, 1963. Influence of raw material and pulping process on the chemical composition and physical properties of paper pulps. Svensk Papperstidning, 66: 196-210.
3: Atalla, R.H. and D.L.V. Hart, 1999. The role of solid state 13 C NMR spectroscopy in studies of the nature of native celluloses. Solid State Nucl. Magnetic Resonance, 15: 1-19.
4: Bertran, M.S. and B.E. Dale, 1986. Determination of cellulose accessibility by differential scanning calorimetry. J. Appl. Polym. Sci., 32: 4241-4253.
5: Dahlman, O., A. Jacobs and J. Sjoberg, 2003. Molecular properties of hemicelluloses located in the surface and inner layers of hardwood and softwood pulps. Cellulose, 36: 325-325.
Direct Link |
6: Dammstrom, S., L. Salmen and P. Gatenholm, 2009. On the interaction between cellulose and xylan a biomimetic simulation of the hardwood cell wall. BioResources, 4: 3-14.
7: Duchesne, I., E.L. Hult, U. Molin, G. Daniel, T. Iversen and H. Lennholm, 2001. The influence of hemicellulose on fibril aggregation of kraft pulp fibres as revealed by FE-SEM and CP/MAS 13C NMR. Cellulose, 8: 103-111.
8: Hult, E.L., P.T. Larsson and T. Iversen, 2001. Cellulose fibril aggregation an inherent property of kraft pulps. Polymer, 42: 3309-3314.
9: Hult, E.L., T. Iversen and J. Sugiyama, 2003. Characterization of the supermolecular structure of cellulose in wood pulp fibers. Cellulose, 10: 103-110.
10: Jordan, B.D. and M.A.O. Neill, 1994. The birefringence of softwood mechanical pulp fines. J. Pulp Paper Sci., 20: 172-174.
11: Kettunen, J., J.E. Laine, I. Yrjäla and N.E. Virkola, 1982. Aspects of strength development in fibres produced by different pulping methods. Paperi Ja Puu, 64: 205-211.
12: Kibblewhite, P. and D. Bawden, 1989. Structural organizations and papermaking qualities of kraft, soda AQ neutral sulfite AQ polysulfide and polysulfide-AQ pulps. Appita J., 42: 275-281.
13: Larsson, P.T., U. Westermark and T. Iversen, 1995. Determination of the cellulose Iα allomorph content in a tunicate cellulose by CP/MAS 13 C-NMR spectroscopy. Carbohydr. Res., 278: 339-343.
14: Larsson, P.T., K. Wickholm and T. Iversen, 1997. A CP/MAS13 C NMR investigation of molecular ordering in celluloses. Carbohydr. Res., 302: 19-25.
15: Larsson, P.T. and P.O. Westlund, 2005. Line shapes in CP/MAS 13 C NMR spectra of cellulose I. Spectrochim. Acta A Mol. Biomol. Spectrosc., 62: 539-546.
16: Lennholm, H., T. Larsson and T. Iversen, 1994. Determination of cellulose Iα and Iβ in lignocellulosic materials. Carbohydr. Res., 261: 119-131.
17: Liitiä, T., S.L. Maunu, B. Hortling, T. Tamminen, O. Pekkala and A. Varhimo, 2003. Cellulose crystallinity and ordering of hemicelluloses in pine and birch pulps as revealed by solid-state NMR spectroscopic methods. Cellulose, 10: 307-316.
Direct Link |
18: Maloney, T.C., P. Stenius and H. Paulapuro, 1998. Hydration and swelling of pulp fibers measured with differential scanning calorimetry. Nord. Pulp Paper Res. J., 13: 31-36.
19: Maloney, T.C. and H. Paulapuro, 1999. The formation of pores in the cell wall. J. Pulp Paper Sci., 25: 430-436.
20: Milton, J.S. and J.C. Arnold, 1995. Introduction to Probability and Statistics. McGraw Hill, New York, pp: 811
21: Nanko, H. and J. Wu, 1995. Mechanisms of paper shrinkage during drying. Proceedings of the 1995 International Paper Physics Conference, (IPPC‘1995), TAPPI Press, Atlanta, GA., pp: 159-171
22: Nanri, T. and T. Uesaka, 1993. Dimensional stability of mechanical pulps-drying shrinkage and hygroexpansivity. Tappi J., 76: 62-66.
23: Newman, R.H. and J.A. Hemmingson, 1990. Determination of the degree of cellulose crystallinity in wood by carbon-13 nuclear magnetic resonance spectroscopy. Holzforschung, 44: 351-355.
24: Olsson, J.A., A.G. Robertson, T.D. Finnigan and R.R.H. Turner, 1995. An analyzer for fiber shape and length. J. Pulp Paper Sci., 21: 367-373.
25: Piirainen, R., 1985. Optical method provides quick and accurate analysis of fiber length. Pulp Paper, 59: 69-71.
26: Pulkkinen, I., J. Fiskari and V. Alopaeus, 2009. The effect of hardwood fiber morphology on the hygroexpansivity of paper. BioResources, 4: 126-141.
27: Pulkkinen, I., J. Fiskari and V. Alopaeus, 2009. The effect of sample size and shape on the hygroexpansion coefficient-a study made with advanced methods for hygroexpansion measurement. Tappsa, March, pp: 26-33.
28: Salmen, L., C. Fellers and M. Htun, 1987. The development and release of dried-in stresses in paper. Nordic Pulp Pap. Res. J., 2: 44-48.
29: Salmen, L., R. Boman, C. Fellers and M. Htun, 1987. The implication of fiber and sheet structure for the hygroexpansivity of paper. Nordic Pulp Paper Res. J., 2: 127-131.
30: Salmen, L., 1993. Responses of Paper Properties to Changes in Moisture Content and Temperature. In: Products of Papermaking, Baker, C.F. (Ed.). Pira International, Cambridge, UK., pp: 369-430
31: Salmen, L. and J. Berthold, 1997. The Swelling Ability of Pulp Fibres. In: The Fundamentals of Papermaking Material, Baker, C.F. (Ed.). Pira International, UK., pp: 683-701
32: Uesaka, T., 2002. Dimensional Stability and Environmental Effects on Paper Properties. In: Handbook of Physical Testing of Paper, Mark, R.E. (Ed.). Dekker, pp: 115-171
33: Wickholm, K., P.T. Larsson and T. Iversen, 1998. Assignment of non-crystalline forms in cellulose I by CP/MAS 13 C NMR spectroscopy. Carbohydr. Res., 312: 123-129.
34: Anonymous, 2006. KajaaniFiberLab™. Käyttäjän käsikirja K02642 V1.0 FI Metso Automation 3/2006.
35: Atalla, R.H. and D.L. Vanderhart, 1984. Native cellulose. Composite of two distinct crystalline form. Science, 223: 283-285.