DOI 10.1515/hf-2012-0135 Holzforschung 2013; aop
Dinesh Fernando * , Dmitri Gorski , Marc Sabourin and Geoffrey Daniel
Characterization of fiber development in high- and low-consistency refining of primary mechanical pulp Abstract: Primary refined softwood was subjected to high-
consistency (HC) or low-consistency (LC) secondary refin-
ing, and the nature of the development of the internal
and external fiber microstructure and ultrastructure has
been compared. The primary refining of mixed softwood
as a raw material was performed in pilot scale by the
advanced thermomechanical pulp process. The study was
aiming at the comparative characterization of LC and HC
pulps at the fiber level when produced with similar and
well-characterized handsheet properties. The formerly
described Simons ’ staining method was applied. A signifi-
cant degree of fiber wall delamination/internal fibrilla-
tion (D/IF) was observed during both LC and HC refining.
Both the energy input and the refining consistency had a
significant impact on elevating the degree of fiber wall D/IF.
The statistical evaluation of internal fiber development
indicated that the fiber populations in LC- and HC-refined
pulps had a similar degree of fiber wall D/IF despite hav-
ing a large difference in refining energy input (420 kW h
odt -1 ), confirming that D/IF was promoted more energy-
efficiently in LC than in HC refining. The characteristic of
the external fiber development from HC and LC refining
was very different. Secondary LC refining promoted fiber
surfaces with ribbons of thin hairlike threads arising from
the inner secondary S2 layer that occasionally developed
along the whole fiber length. Broad sheet- and lamellae-
type external fibrillation from the S2 was typical for HC
refining, and these characteristics were rarely observed in
the LC pulps. The mechanisms for LC and HC fiber devel-
opment are proposed. The cell wall characteristics (inter-
nal and external) of the pulp fibers appear to govern most
of the physical and optical properties in handsheets.
Keywords: ATMP, delamination/internal fibrillation
(D/IF), fiber characterization, fiber development, surface
ultrastructure of fibers, HC refining, LC refining, SEM,
Simons ’ staining
*Corresponding author: Dinesh Fernando , Department of Forest
Products/Wood Science, Wood Ultrastructure Research Centre
(WURC), Swedish University of Agricultural Sciences, SE-75007
Uppsala, Sweden, e-mail: [email protected]
Dmitri Gorski: Norske Skog Saugbrugs , N-1756 Halden, Norway; and
Department of Mechanical Engineering , Pulp and Paper Centre, The
University of British Columbia, 2385 East Mall, Vancouver, British
Columbia, Canada V6T 1Z4
Marc Sabourin: Andritz , Inc., 3200 Upper Valley Pike, Springfield,
OH 45504, USA
Geoffrey Daniel: Department of Forest Products/Wood Science ,
Wood Ultrastructure Research Centre (WURC), Swedish University of
Agricultural Sciences, SE-75007 Uppsala, Sweden
Introduction Mechanical pulping is expected to produce pulp of suit-
able quality that meets the customer requirements at the
lowest possible energy expenditure. The advanced thermo-
mechanical pulp (ATMP) process gives rise to pulps with
similar properties to those produced by a traditional TMP
process, but the energy input is significantly lower (Gorski
et al. 2011a, 2012c ; Johansson et al. 2011 ). The low-consist-
ency (LC) refining of primary ATMP further reduces the total
energy requirement: ~ 300 kW h odt -1 less energy is needed to
achieve the target pulp and paper properties by secondary
LC refining compared with the secondary high-consistency
(HC) refining of primary ATMP (Gorski et al., 2012a).
The cross-sectional dimensions of fibers decrease with
HC refining (Kure 1999 ) but remain unchanged or even
increase with LC refining perhaps due to internal delami-
nation and swelling (Gorski et al., 2012a). The distribution
of fiber fractions from the two processes also differs con-
siderably. During LC refining, the long fiber fraction (R30)
is lowered and the middle fraction (P30/R100) is elevated
(Gorski et al. 2012b ). LC-refined fibers are more flexible
but have a smaller external surface area compared with
HC-refined fibers (Gorski et al., 2012a). The interpretation
is that LC refining creates less external fibrillation (EF)
and more internal delamination compared with HC refin-
ing, although the characteristic of the fibrillation has not
been studied. The mechanisms leading to these observa-
tions are unknown.
The quality of the final paper product correlates
strongly with the development of fiber properties such as
the amount of split fibers and decreased cross-sectional
Brought to you by | Sveriges LantbruksuniversitetAuthenticated | [email protected] author's copy
Download Date | 3/14/13 10:40 AM
2 D. Fernando et al.: Fiber development in refining
dimensions (Reme et al. 1998 ; Kure 1999 ; Gorski and Hill
2012 ), fiber flexibility (Corson 1989 ; Fernando et al. 2011 ,
2012; Gorski et al. 2011b,c , 2012a ), fiber wall thickness
(Braaten 2000 ; Ferluc et al. 2010 ), fiber bonding (Skowron-
ski 1990 ; Reme et al. 1998 ), collapsibility of fibers (Reme et
al. 1998 ; Kure 1999 ), development of the long fiber fraction
(Corson et al. 2003 ), and EF/internal fibrillation (EF/IF)
(Fernando et al. 2011 , 2012; Gorski et al., 2012a ). EF/IF can
presumably be tailored by the parameters of the refining
process (Miles and Karnis 1991 ; Kang and Paulapuro 2006 ;
Fernando et al. 2011 ). It is known that IF (Claudio -da-Silva
1983 ; Abitz and Luner 1989 ; Mohlin 1989 ; Paavilainen
1993 ; Fernando et al. 2011 , 2012) and EF (Mohlin 1989 ;
Kang and Paulapuro 2006 ; Fernando 2007 ) determine, to
a large extent, both the strength and the optical properties
of paper.
The focus of this study was to investigate the mecha-
nisms of fiber development in secondary HC and LC refin-
ing of primary ATMP. Therefore, the characterization of
fiber development was performed, aiming to improve the
understanding of the interrelation between HC- and LC-
refined fiber profiles and the optical and strength properties
of paper sheets. The intention was to contribute to a better
understanding of the fundamental mechanisms gover-
ning this fiber property development in LC- and HC-refined
pulps at the microstructural and ultrastructural levels of
cell walls. Most of all, the influence of refining processes
on the structural changes to the fiber walls and thereby the
final paper quality should be better understood.
Materials and methods
Raw materials and refi ning The mixed soft wood chips were obtained from a Canadian TMP
mill. Composition: ~ 80 % lodgepole pine and 20 % Sitka spruce
and western balsam fi r from the interior of British Columbia. The
ATMP process (Hill et al. 2010 ; Johansson et al. 2011 ) for HC refi n-
ing was conducted at the Andritz pilot plant (Springfi eld, OH, USA).
The chips were preheated and defi brated by means of a mechanical
pretreatment consisting of a modular screw device (Impressafi ner;
Andritz, 40 kW h odt -1 energy input) and a Fiberizer (Andritz; 220
kW h odt -1 energy input). The fi berized material was fi rst stage re-
fi ned under conditions of RTS (low retention time, high tempera-
ture, and high refi ner speed) according to Sabourin et al. (1997) ,
with an energy input of 530 kW h odt -1 ; 3.1 % bisulfi te was added
through the dilution water.
A part of the fi rst-stage pulp was further refi ned by an atmos-
pheric HC double-disc refi ner with four diff erent energy inputs
(610 – 1000 kW h odt -1 ). LC refi ning of the fi rst-stage ATMP was
conducted in two optimized stages at the pilot plant in the Pulp
and Paper Centre, The University of British Columbia (Vancouver,
British Columbia, Canada). The pulp was disintegrated in 60 ° C
water for 4 h before refi ning at LC (3 – 3.4 % ) using a 14-inch LC pilot
refi ner (Aikawa Fiber Technologies, Sherbrooke, Quebec, Canada)
equipped with a 16-inch overhung segments and driven by a 110
kW variable frequency motor. The refi ning conditions were varied
by changing the gap of the refi ner while the throughput was kept
constant. This gave specifi c energy inputs of 70 – 180 kW h odt -1 in
the fi rst LC stage and 70 – 240 kW h odt -1 in the second LC stage.
The FineBar (Aikawa Fiber Technologies) segments were used (bar
edge length 5.59 km, groove depth 4.8 mm, groove width 2.4 mm,
bar width 1 mm, and bar angle 15 ° ). The pulps and handsheets
formed from the pulps (Table 1 ) were subjected to the Technical
Association of the Pulp and Paper Industry standards testing as
described previously (Gorski et al. 2012b ).
Pulps for fiber characterization Five pulps were chosen for the characterization of fi ber properties
(Table 1) based on specifi c energy consumption (SEC) and the properties
of handsheets made from the refi ned pulps for representing extreme
levels and/or similar properties (Figure 1 ; cf. Table 1). This is due to the
fact that paper with similar physical properties can be produced from
LC-refi ned pulp with signifi cantly lower energy input compared with
HC-refi ned pulp. Therefore, in this study, an emphasis was given to
the comparisons of pulps with a similar level of handsheet properties.
Simons ’ staining of pulp fibers The development of fi ber wall delamination/IF (D/IF) was evaluated
according to Fernando and Daniel ’ s (2010) method of Simons ’ stain-
ing (SS). Staining was performed on ~ 1 g of never-dried pulp from
each of the fi ve samples and then immediately examined and ana-
lyzed by light microscopy as described in the quoted literature.
Table 1 Abbreviations and essential data of the pulps observed.
Code Position (stage) SEC (kW h odt -1 ) Basic properties
TI (N m g -1 ) Density (kg m -3 ) TEA (J m -2 ) CSF (ml) LS (m 2 kg -1 )
HC1 First HC stage 800 21.6 298 16.4 487 50.1
HC2a Second HC stage 1410 37.8 371 37.8 170 57.0
HC2b Second HC stage 1570 43.6 401 50.0 117 60.3
LC2 Second LC stage 980 29.7 351 29.2 238 55.3
LC3 Third LC stage 1150 39.4 398 32.2 110 59.2
TI, tensile index; CSF, Canadian standard freeness; LS, light scattering coefficient.
Brought to you by | Sveriges LantbruksuniversitetAuthenticated | [email protected] author's copy
Download Date | 3/14/13 10:40 AM
D. Fernando et al.: Fiber development in refining 3
20
25
30
35
40
45
50
55
600 800 1000 1200 1400 1600 1800 2000
Tens
ile in
dex
(Nm
g-1
)
SEC (kW h odt-1)
HC Primary
HC Secondary
LC Secondary
HC2a
HC2b
HC1
LC2
LC3
Figure 1 Tensile index of handsheets formed from pulps character-
ized in this study (labeled inside squares; see Table 1).
Data collection and statistical analysis Five subfi ber populations (SFPs; see Fernando and Daniel 2010 )
that represent the diff erent categories of fi bers possessing varying
degrees of D/IF in a given pulp were identifi ed following SS. Each
category refl ects varying levels of internal fi ber development in a
given pulp. The SFPs from stained samples were obtained manually
by light microscopy (200 fi bers from each pulp were evaluated) and
analyzed as described (Fernando and Daniel 2010 ). Raw data were
fi rst summarized graphically for the ease and direct observation of D/
IF development in the cell walls. Thereaft er, the data were evaluated
and statistically analysed with SAS soft ware (SAS/STAT version 9.3
for Windows XP-Pro platform; SAS Institute, Cary, NC, USA). Ordinal
logistic regression tests were performed for the assessment of the sig-
nifi cance of the diff erences concerning the degree of D/IF of the pulp
fi ber cell walls.
Scanning electron microscopy The characterization of surface ultrastructure and EF of pulp fi bers
was performed by scanning electron microscopy (SEM). Approxi-
mately 2 g of wet samples were dehydrated separately in a series of
increasing ethanol and then acetone concentrations (Fernando and
Daniel 2008 ). The samples were then dried by means of an Agar
E3000 critical point dryer (Agar Scientifi c, Stansted, UK) with CO 2 as
a drying agent and coated with gold with an Emitech K550X sputter
device (Quorum Technologies Ltd, Ashford, Kent, UK). SEM instru-
ment: Philips XL 30 ESEM (FEI Company, Eindhoven, Netherlands)
operated at 10 kV.
Results and discussion The handsheets with a tensile index of 40 Nm g -1 and a
light scattering of 59 m 2 kg -1 were produced from both HC-
and LC-refined ATMP. The HC refining required ~ 1450 kW h
odt -1 SEC, whereas ~ 300 kW h odt -1 less SEC was consumed
in LC refining (HC2a vs. LC3; Figure 1). The apparent
density of the handsheets developed similarly for both
HC and LC refining compared at equal tensile index level,
whereas the tear index, length weighted average fiber
length, stretch, and tensile energy absorption (TEA) were
lower for the LC-refined pulp (Table 1). LC refining signifi-
cantly reduced the size of R30 fiber fraction and consider-
ably increased the middle fraction (Gorski et al. 2012a,b ).
Internal fiber development
The response of individual fibers to SS differed; fibers
stained in varying color intensities from blue to yellow/
orange with increasing severity of the fiber wall D/IF.
The observation is in line with that of previous works
(Blanchette et al. 1992 ; Fernando and Daniel 2010 ). This
indicates a diverse distribution of the individual fiber
development within a given pulp. This can be expected
because the heterogeneous wood fibers are treated dif-
ferently depending on SEC, temperature, and intensity,
and these parameters change both the chemical and the
morphologic structures of the native fiber wall as illus-
trated in the literature (Claudio -da-Silva 1983 ; Fernando
2007 ; Daniel et al. 2009 ). For example, an increase in SEC
or refining intensity enhances the proportion of treated
fibers with high degree of D/IF, thereby improving the
overall internal fiber development of a pulp (Fernando
et al. 2012 ).
An overview of the results from the SS study is shown
in Figure 2 and outlines an initial assessment of fiber
development. The five SFPs were simplified into three
major groups: “ non-D/IF ” , “ low D/IF ” , and “ high D/IF ” ,
as described in the method for ease of understanding
(Figure 2a). As expected, increasing the SEC significantly
reduced the percentage of untreated fiber population
resulting in enhanced fiber development (Figure 2a, red
line). The difference was especially pronounced when the
initial primary-stage pulp (HC1) was compared with any
of the secondary pulps irrespective of refining consist-
ency (e.g., HC1 vs. LC2). The primary pulp with the lowest
refining energy input (HC1; 800 kW h odt -1 ) consisted of
~ 63 % untreated stiff fibers and only 16 % fibers with a high
degree of D/IF. In contrast, pulp HC2b with the highest
SEC (1570 kW h odt -1 ) was the most developed at the fiber
wall level. It was dominated by a treated fiber population
( ~ 65 % ), with the majority of fibers from the high D/IF
group ( ~ 38 % ) that represents the most flexible fiber frac-
tion. The preliminary analysis also indicated that pulps
LC3 and HC2b appeared to have more or less similar fiber
populations concerning the degree of D/IF (61 % and 65 %
Brought to you by | Sveriges LantbruksuniversitetAuthenticated | [email protected] author's copy
Download Date | 3/14/13 10:40 AM
4 D. Fernando et al.: Fiber development in refining
treated fiber population, respectively). However, LC3 con-
sumed 420 kW h odt -1 less SEC than HC2b (Figure 1). This
finding confirms that LC refining induces wall D/IF more
energy-efficiently, thereby generating more flexible fibers
than during HC refining. These fiber development values
are in good agreement with mill-scale pulp data, where
primary double-disc refined pulp had ~ 40 % untreated
and 35 % well-treated fibers at 1700 – 1900 kW h odt -1 and
tensile index of 45 Nm g -1 (Fernando et al. 2011 ).
The percentage of almost untreated fibers (i.e.,
SFP light blue) was lower in the secondary HC- and LC-
refined pulps compared with the primary pulp (Figure 2b).
However, the secondary LC pulps also contained more of
the almost untreated fibers ( ~ 25 % light blue compared
with ~ 15 % in the HC-refined pulps). If it is assumed that
a fiber always develops some degree of D/IF on receiving
an impact (mechanical action) inside a refiner, a con-
clusion can then be drawn from this finding; less fibers
receive any impact during LC compared with HC refining
at a given pulp quality level (LC3 vs. HC2a; Figure 2b). The
proportion of well-treated fibers increased with energy
input for both HC and LC. However, the energy increment
required was significantly smaller in LC refining ( ~ 350
vs. 750 kW h odt -1 SEC for LC3 and HC2b, respectively;
Figure 2a) to achieve a similar level of internal fiber devel-
opment. Accordingly, each impact on fibers in a LC refiner
HC1 LC2 LC3 HC2a HC2b
(800) (980) (1150) (1410) (1570)SEC
Pulp name
Per
cent
age
of fi
bres
(%)
0
10
20
30
40
LB DB Green Or/Yl < 1/2 Or/Yl > 1/2
100
50
0
HighD/IF
NonD/IF
LowD/IF
a
b
Figure 2 Percentage of fibers stained in different color intensities
following SS.
(a) Fractions of the three major groups of fibers representing differ-
ent levels in the degree of wall D/IF. (b) Raw data for each of the five
SFPs in the pulps investigated.
promotes D/IF of the cell wall more efficiently than HC
refining. This implies that, within the LC system, more
energy was transferred effectively onto fibers (i.e., interac-
tion effect of SEC and LC system) as discussed below.
Statistical analysis for the degree of D/IF
Detailed statistical analyses by means of the ordinal logis-
tic regression test (Table 2 ) shows (a) the overall signifi-
cance in the degree of D/IF among all pulps, (b) the effect
of the two refining conditions (i.e., the refining energy
input and the refining consistency), and (c) the impact of
the different refining conditions on morphologic changes
(i.e., D/IF) in the different pulps by analyzing two at a
time.
The greatest difference was found in the degree of
wall D/IF among the five pulps at 0.01 % significance
level (P < 0.0001; Table 2). This resulted from both increas-
ing SEC, which is in line with previous results (Fernando
et al. 2011 , 2012), and changing the refining conditions
(HC/LC). The statistical evidence for this was provided
when analyzing the two variables separately. The results
indicated that both the energy input (P < 0.0001) and the
LC/HC refining conditions (P < 0.0001) have highly signifi-
cant influence on enhancing fiber wall D/IF (i.e., making
stiff wood fibers flexible).
The statistical analysis performed pairwise is also
informative and provides further evidence on the novel
information regarding the internal fiber development
mechanisms during the two processes. There is a sig-
nificant difference in the development of fiber wall D/
IF between primary HC and secondary LC pulps (HC1
vs. LC2; P = 0.0472). The difference in SEC between the
Table 2 Logistic regression statistics for type 3 analysis of ordinal
logistic regression test for significant differences between HC- and
LC-refined pulps on the degree of fiber wall D/IF.
Source DF χ 2 Pr > χ 2
Pulps a 4 53.18 < 0.0001
Energy b 1 41.39 < 0.0001
HC/LC system b 1 15.66 < 0.0001
HC1 vs. LC2 c 1 3.94 0.0472
LC2 vs. LC3 c 1 8.43 0.0037
LC3 vs. HC2a c 1 0.04 0.8461
HC2a vs. HC2b c 1 2.36 0.1241
LC3 vs. HC2b c 1 1.50 0.2213
a Overall significance.
b Two refining conditions (energy and HC/LC system effect).
c Comparing two pulps for differences in overall degree of D/IF of
their fibers.
Brought to you by | Sveriges LantbruksuniversitetAuthenticated | [email protected] author's copy
Download Date | 3/14/13 10:40 AM
D. Fernando et al.: Fiber development in refining 5
two pulps was 180 kW h odt -1 . The internal fiber develop-
ment between the two LC pulps is also significantly dif-
ferent with an increase in SEC of 170 kW h odt -1 (LC2 vs.
LC3; P = 0.0037). However, the degree of D/IF developed
between the two HC pulps was not significantly different,
although the SEC difference of the two pulps was similar
to that of the two LC pulps (HC2a vs. HC2b; P = 0.1241). The
results thus indicate an interaction effect of energy and LC
system that accelerates internal fiber development during
LC refining. This is reflected in LC3 pulps that contained
a greater percentage of treated fibers compared with LC2
(Figure 2a). An evidence for the interaction effect is also
given by the raw data itself (Figure 2b) as discussed above.
However, it should be noted that the interaction effect
could not be statistically estimated because all the pos-
sible combinations of energy and consistency were not
available in the data. For example, the combination high
energy input and LC was missing due to the practical limi-
tations of the refining system.
Furthermore, statistical analysis indicated that LC
refining with low SEC produced pulps that was similar
to high-energy HC pulps regarding the degree of internal
fiber development (LC3 vs. HC2b; P = 0.2213). This provides
a statistical evidence for the preliminary analysis and con-
firmed the improved energy efficiency of LC refining.
S1
S1
S1
S1
S1
S1
S2
CML
CM
LS
2
a b
dc
S2
S2
S1
S2
S2
S2
S2
S1S2 S1
S2
S2
S2 S2 S2
Figure 3 SEM micrographs of HC- and LC-refined TMP fibers with their surface morphologic characteristics (see Table 1).
(a) HC1 fibers with the secondary S1 layer of the cell wall as the outer layer. The retention of the native geometrical shape of the fibers indi-
cates inherent fiber stiffness. (b) Almost all the HC2b fibers possess the S2 as the outer layer with ribbon-type fibrillation or clear surfaces
presumably due to complete peeling of fibrils. Collapsed never-dried fibers lost their square-like form. (c) Most fibers of LC2 show S1 as the
outer layer. (d) The majority of LC3 fibers have the S2 layer exposed most often with hairlike fibrils protruding from the surface. Bars, 40 μ m
(a – d).
Characterization of the fiber surface ultra-structure by SEM
There were differences in the fiber surfaces especially
between the primary HC1 pulp (Figure 3 a) and the second-
ary HC2b pulp produced with high energy input (Figure 3b).
The majority of fibers in HC1 (SEC of 800 kW h odt -1 )
had the S1 secondary wall as their outer surface layer
(Figures 3a and 4a). It was most often fibrillated into
typical flake-like fibrils (Figure 4 a, arrows), sometimes
with remaining parts of the compound middle lamella
(CML; Figures 3a and 4a). In addition, most of the HC1
fibers retained their native geometrical shape of wood
fibers, which is indicative of intrinsic stiffness. On the
contrary, almost all the fibers in the HC2b pulp (SEC of
1560 kW h odt -1 ) displayed an exposed cellulose-rich S2
layer with a typical ribbon-type fibrillation and the fibers
exhibited greater collapsibility/conformability (Figure 3b).
The collapse of the fibers was most likely attributable to
their thinner walls, which resulted from the removal of the
major part of fiber wall materials as shown in Figure 3b.
Most of the fibers from the LC2 and LC3 pulps also
exhibited an exposed S2 layer, often with typical ribbon-
like fibrils projecting from the fiber surface (Figure 3c
and d). Both LC-refined pulps appeared morphologically
Brought to you by | Sveriges LantbruksuniversitetAuthenticated | [email protected] author's copy
Download Date | 3/14/13 10:40 AM
6 D. Fernando et al.: Fiber development in refining
similar, although the LC2 with lower energy input had
more fibers covered with outer S1 wall material, indicat-
ing an inferior peeling effect compared with the LC3 pulp
from higher SEC.
A detailed SEM observation revealed important infor-
mation related to external fiber development mecha-
nisms during HC and LC refining. Severe fiber splitting
was already initiated during the ATMP refining process
in the first-stage pulp (Figure 4b, arrows). This is consist-
ent with earlier observations where a split fiber index of
a
f g h
i j k
l
m
d
cb
e
S1
S2
CML
S2
S2
S2 S2
S2
S2
Figure 4 SEM micrographs detailing surface ultrastructure and EF of HC- and LC-refined fibers.
(a) HC1 fiber has a rugged appearance due to bricks-like structures (arrows) from the fibrillated S1 surface layer. Parts of the remaining CML
and retention of the square-like native fiber form are visible. (b) Opening of some HC1 fibers exposing the cell lumen due to fiber splitting
(arrows). (c) HC2b fibers with typical ribbon-type S2 fibrillation reflecting the native hierarchical structure of the fiber wall. (d) Thin thread/
string-like ribbons most likely representing the smallest macrofibrils (arrows). (e) Wide band-like ribbons formed by aggregates of smaller
numbers of macrofibrils (arrowheads in d and e). (f) Broad sheet-like ribbons (arrows). (g and h) Broad carpet forming lamellae sheet
peeling from the S2 (curved arrow). (i) Hairlike structures of ribbons from the S2 surface layer developed along the fibers (pulp LC3) during
LC refining. (j and k) Hairlike ribbons were either thin thread and string-like fibrils presumably representing single or small macrofibrils
(arrowheads) or slightly wide band-like ribbon fibrils (arrows) from the S2. (l and m) LC3 fibers illustrating distinct S2 EF along the fiber so
that they appear as “ hairs ” projecting from the surface. Bars, 20 μ m (a, d – f, and j), 40 μ m (b and c), 10 μ m (g, h, and k – m), and 100 μ m (i).
~ 0.30 was determined for first-stage pulp (Gorski et al.,
2012a ). Although this implied a strong refining action on
the fibers already in the first stage, an extensive peeling of
the fiber wall material into the inner secondary wall layers
was observed only at high SEC.
Strong dissimilarities in fiber surface morphology were
found between HC and LC pulps. HC refining with high
SEC generated various types of characteristic S2 fibrilla-
tion (Figure 4c; e.g., ribbon-type fibrils) ranging from thin
thread-like fibrils (Figure 4d, arrows) to broad sheet-like
Brought to you by | Sveriges LantbruksuniversitetAuthenticated | [email protected] author's copy
Download Date | 3/14/13 10:40 AM
D. Fernando et al.: Fiber development in refining 7
fibrils (Figure 4e and f) and lamellar sheets (Figure 4g
and h) peeling from the fiber surfaces. Particularly, the
broad sheet-like fibrils and lamellar sheets were common
on fibers in HC2b, although they were also observed occa-
sionally in the HC2a pulp with lower energy input.
LC refining promoted a mechanical interaction that
led to a distinct fiber development mechanism, where fiber
surfaces most often showed thin hairlike threads, some-
times along the whole fiber length (Figure 4i, l, and m).
Broad sheet-like and lamellae types of EF were rarely
observed in the LC pulps and the major hairlike surface
wall structures were thread-like fibrils (i.e., similar to
single macrofibrils; Figure 4j and k, arrowheads) or very
thin ribbon-like fibrils originating from the S2 layer (Figure
4j and k, arrows). In mechanical pulping refining, micro-
cracks are known to develop along naturally occurring
weak zones in the lamellae of already exposed S2 layer
(i.e., between single macrofibrils or aggregates of various
sizes of macrofibrils within lamellae; Fernando and Daniel
2004 ). In this trial, LC refining enhanced the development
of microcracks predominantly between smaller aggregates
of macrofibrils (Figure 5 a). The resulting ribbons of thin
threads were thereby easily peeled off from the surface by
the action of shearing forces inside the refiner (Figure 5b
and c). The actual events taking place in the fiber outer
wall during LC refining are illustrated in Figure 5.
A possible cause for the differences in the fiber surface
ultrastructure may be the different refining environments
in HC and LC refiners, which could influence the soften-
ing and surface properties of refined fibers. For example,
Kerekes (2011) describes the difference in the coefficient of
friction measured in HC (0.5 – 0.8) and LC (0.1 – 0.15) refin-
ers. The motor effect and typical energy inputs in HC and
LC refiners are also reported to follow the same relation-
ship and are approximately 5 – 10 times less in the case of
LC. The lower friction inside the LC refiners suggests a dif-
ferent type of mechanical interaction, which may explain
the distinct fibrillation type observed.
Hypothesis on the mechanisms of HC and LC refining
Based on the results, it can be hypothesized that shear
forces may dominate in HC refining and influence the
observed fibrillation types (e.g., broader sheets/lamellae
fibrils), which have a greater relative bonded area (RBA).
In the case of LC refining, on the contrary, compressive
forces are dominant. They are generated from fiber-to-
fiber interaction by the hitting of less concentrated fibers/
fiber bundles against each other, which is less probable in
HC refining. The refiner bars play also a more prominent
role (plate-to-fiber action) for cyclic compression in LC
refiners. Therefore, it is most probable that these mechan-
ical interactions (impact on fibers) in LC refiners impose
high internal stresses/strains on fibers that should trigger
a c d
b
e
Figure 5 SEM micrographs of LC3 fibers at high magnifications highlighting the phenomena occurring in the fiber wall during LC refining.
(a) Development of many microcracks (arrows) along the inherent weak zones within lamellae. These zones lie between individual mac-
rofibrils (red circle; popping up of a single macrofibril) or small aggregates of macrofibrils, which form band-like ribbons (arrowheads). (b)
Microcrack on the wall (arrowhead) that developed into a large opening stimulating peeling of the respective band-like narrow fibril (curved
arrow) and strings of small macrofibrils peeling and protruding from the surface contributing to the hairlike appearance. (c) Remnants on
the wall of a fiber providing evidence for stripping of narrow band-like ribbons (yellow lines indicate the width of the individual ribbon
peeled out). (d and e) Latter stage of hairlike narrow S2 fibrillation by peeling (curved arrows). Bars, 1.0 μ m (a and b), 2.0 μ m (c), and 8.0 μ m
(d and e).
Brought to you by | Sveriges LantbruksuniversitetAuthenticated | [email protected] author's copy
Download Date | 3/14/13 10:40 AM
8 D. Fernando et al.: Fiber development in refining
fiber wall D/IF efficiently and generate more microcracks
in the wall leading to hairlike fibrillation. The previous
studies on the action of shear and compressive forces con-
clude that the former contributes predominantly to EF via
peeling mechanisms, whereas IF is the primary effect of
cyclical repetitive compression (Hartman 1984 ; Kang and
Paulapuro 2006 ; Kerekes and Senger 2006 ). Accordingly,
the hypothesis proposed on refining mechanisms with dif-
ferent and selective refining actions between HC and LC
refiners is the principle reason for the different fibrillation
mechanisms and fiber responses observed.
Several studies report on the importance of plate
clearance in terms of fiber properties ( H ä rk ö nen et al.
2003 ; Murton and Duffy 2005 ; Kerekes and Senger 2006 ;
Mohlin 2006 ; Eriksen and Hammar 2007 ). Compres-
sive forces are known to dominate when the plate gap
is small, and shearing forces are prominent when the
gap is large. H ä rk ö nen et al. (2003) showed how consist-
ency controls fiber residence time and plate clearance.
The smallest values for the latter two parameters are
attained with the lowest consistency. The plate clear-
ance in LC refiners is very small compared with HC
refiners, where it is several times larger (Eriksen and
Hammar 2007 ). In addition, a very intensive turbulent
motion of the pulps and the extension of the turbulence
toward the outer radius of the refiner plate are obtained
when the pulp consistency is lowest ( H ä rk ö nen et al.
2003 ). These conditions most likely prevailed in the LC
refiners of the present trial (at ~ 3 % consistency), which
promoted strong fiber-to-fiber as well as fiber-to-bar
impacts (i.e., compressive forces). In contrast, at higher
consistencies, repetitive compression was less influen-
tial during HC refining. However, more fundamental
work is needed to clarify the mechanisms proposed for
HC and LC refining.
Characteristic of fiber development and final product quality
The relationship of the three major D/IF groups to the
tensile strength is established in earlier work (Fernando et
al. 2011 ). The present study also showed very similar rela-
tionships (Figure 6 a). A very strong positive correlation
(r 2 = 0.96) to tensile index of handsheets was exhibited by
pulp fibers with high D/IF. Untreated non-D/IF fibers had
an inverse correlation with the same strength (r 2 = 0.99).
Fibers with low D/IF showed a positive trend with lower
strength gain. Furthermore, the three groups of fibers
were correlated to the density of the handsheets with
exactly the same strength and direction as they did with
tensile strength (Figure 6b). The results thus emphasize
the direct influence of the relative proportions of the two
groups of high and low D/IF, which contain desirable flex-
ible fibers, in a given pulp for the development of strength
and density.
Accordingly, the significant differences in the degree
of D/IF among the five pulps are the fundamental bases for
varying tensile strengths, freenesses, and densities of the
pulps developed (Table 1). The elevated D/IF in LC3 and
HC2b pulps made them highly flexible and thereby improved
the collapsibility and conformability of fibers. This accounts
for the enhanced fiber-to-fiber bonding and better densifica-
tion of handsheets (Stone and Scallan 1965 ; Abitz and Luner
1989 ; Heikkurinen et al. 1991 ; Paavilainen 1993 ).
In addition, the surface ultrastructure of each pulp
also contributed substantially to most of the physical and
optical properties of their handsheets. For example, the
exposed S2 layer with enhanced EF (long ribbon-like cel-
lulose fibrils), particularly the broad sheet-like and lamel-
lae fibrils, exhibited by pulp HC2b were key characters
for good bonding ability of the pulp (Kang and Paulapuro
R2=0.9907
R2=0.7277
R2=0.9599
0
10
20
30
40
50
60
70
Fibr
es s
tain
ed w
ith d
iffer
ent c
olou
rs (%
)
Tensile index (Nm g-1)
Non D/IF
Low D/IF
High D/IF
R2=0.9358
R2=0.6618
R2=0.9171
0
10
20
30
40
50
60
70
20 25 30 35 40 45 280 300 320 340 360 380 400 420Density (kg m-3)
Non D/IF
Low D/IF
High D/IF
a b
Figure 6 Linear correlation of the three major groups of SFPs in a pulp to the tensile index (a) and apparent density (b) of their handsheets.
The non D/IF and high D/IF groups had very strong but opposite correlation. The figure illustrates the influence of D/IF on the properties of
the sheets produces from the fibers.
Brought to you by | Sveriges LantbruksuniversitetAuthenticated | [email protected] author's copy
Download Date | 3/14/13 10:40 AM
D. Fernando et al.: Fiber development in refining 9
2006 ; Li et al. 2010 ; Fernando et al. 2012 ). Ribbon-type
fibrils acquire higher bonding potential, particularly the
broad and long fibrils possessing greater RBA (Braaten
2000 ) and promote sheet density, tensile index, and Scott
bond strength (Kang and Paulapuro 2006 ). The higher
tensile index and density of HC2b and LC3 pulps were
thus explained by enhanced fiber development (i.e., inter-
nal and external cell wall characteristics) of their fibers.
In contrast, the inferior EF reflected by the lignin-rich S1
surface layer with flake-like short S1 fibrils of poor bonding
potential is partly responsible for the lowest density and
strength properties in HC1 and LC2 (Table 1). A consider-
ably larger proportion of their untreated stiff fibers (63 %
for HC1 and 54 % for LC2; Figure 2a) also collectively con-
tributed to the observed lower quality of the pulp.
Light scattering was on the same level for LC- and HC-
refined pulps compared at equal tensile indices (HC2b vs.
LC3; Table 1). Gorski et al. (2012a) observed lower EF for
LC pulps than for HC pulps. However, this study demon-
strates that LC refining has a different mechanism of EF,
which promoted hairlike long ribbons. Braaten (2000)
describes RBA of external fibrils as the determining factor
for both strength and optical properties of a paper sheet
and the important role of thread-like and narrow ribbon-
shaped fibrils for higher light scattering properties. Con-
sequently, the morphologic characteristic of EF found in
LC3 pulps was presumably the principle reason for its
equally high light scattering ability. According to Braaten
(1997, 2000) , the RBA of the hairlike narrow long ribbons
of LC3 fibers is very low and thus significantly influences
the light scattering coefficient of the pulp.
A significantly lower stretch and TEA of handsheets
was shown by the LC compared with the HC-refined pulp
(Table 1). This means that the fiber network of sheets
from the LC pulps was not able to stretch as much as
the network of HC-refined fibers. This may be caused
by several factors. LC-refined fibers are straighter and
shorter compared with HC-refined fibers (Gorski et al.,
2012a), which would lead to less stretch in the individual
LC fibers, when a handsheet is subjected to tension. The
morphologic characteristic of EF, which was deficient
in carpet forming broad sheets and lamellae (which are
excellent binders between fibers; Braaten 1997, 2000 )
compared with that of HC fibers, may have also collec-
tively contributed to the inferior stretchability of the LC
fiber network. A higher degree of D/IF normally causes
higher flexibility and stretchability of individual fibers
(Fernando et al. 2012 ), which did not seem to compensate
here for the higher straightness of the LC-refined fibers. A
possible explanation is that the highly flexible LC fibers
formed bonds with a larger bonded area, which could
stretch out the fiber segments between the bonds leading
to less network stretchability.
Conclusions HC and LC refining of primary ATMP along with the increas-
ing SEC significantly elevated the degree of fiber wall D/
IF. The internal fiber development during secondary HC
and LC refining was similar at a similar level of handsheet
tensile index. However, the refining energy input for the LC
refining was 420 kW h odt -1 less. LC promoted the fiber wall
D/IF more energy-efficiently than HC refining. The external
fiber development during HC and LC refining was very dif-
ferent as revealed by the ultrastructural characterization of
the pulp fiber surfaces. LC refining is suggested to induce
a distinct action that promotes fiber surfaces rich in thin
hairlike threads of S2 ribbons, sometimes along the whole
fiber length in addition to efficient D/IF development. In
the course of HC refining, broad sheet and lamellar types
of EF of the S2 layer occur, leading to high bonding poten-
tial of the fibers. These characteristics were rarely observed
in LC-refined pulps. The different mechanisms of fiber
development at the cell wall level were explored in HC and
LC refining that governs most of the physical and optical
properties of the pulps investigated.
Acknowledgements: Our work was carried out within the
framework of the “ Branschforskningsprogram f ö r skogs-
och tr ä industrin ” and “ Process and product developments
through unique knowledge of wood fiber ultrastructure ”
(2007 – 03230). The program is financed by VINNOVA, six
forest-based industries (Eka Chemicals, Holmen, SCA,
Smurfit Kappa Kraftliner, StoraEnso and S ö dra Cell), and
involves collaborative work among SLU, Innventia, KTH,
and MSU. The authors would like to express their grati-
tude to the staff of The University of British Columbia Pulp
and Paper Centre and the Andritz pilot plant in Spring-
field, OH, especially Dr. Antti Luukkonen. Dr. Jens Heymer
(Aikawa Fiber Technologies) and Prof. James Olson (The
University of British Columbia Pulp and Paper Centre) are
acknowledged for their advice and support during the LC
refining trials. Jan Hill (QualTech AB) is acknowledged for
valuable comments on the article. Dmitri Gorski would
also like to acknowledge the financial support of G å l ö stif-
telsen, Hans Werth é n Fonden, and Jansons Legat who all
contributed to his postdoctoral at The University of British
Columbia.
Received August 13, 2012; accepted February 11, 2013; previously
published online xx
Brought to you by | Sveriges LantbruksuniversitetAuthenticated | [email protected] author's copy
Download Date | 3/14/13 10:40 AM
10 D. Fernando et al.: Fiber development in refining
References Abitz, P., Luner, P. (1989) The effect of refining on wet fibre flexibility
and its relationship to sheet properties. In: Fundamentals of
Papermaking, Transactions of the 9th Fundamental Research
Symposium, Cambridge, UK. Vol. 1, pp. 67 – 86.
Blanchette, R.A., Akhtar, M., Attridge, M.C. (1992) Using Simons
stain to evaluate fiber characteristics of biomechanical pulps.
Tappi J. 75:121 – 124.
Braaten, K.R. (1997) The impact of fibre geometry, fibre splitting and
fibrillation on the light scattering. In: International Mechanical
Pulping Conference. SPCI, Stockholm, Sweden, June 9 – 13. pp.
349 – 353.
Braaten, K.R. (2000) Fibre and fibril properties versus light
scattering and surface smoothness for mechanical pulps. Pulp
Pap. Can. 101:122.
Claudio-da-Silva, E. (1983) The flexibility of pulp fibres – a structural
approach. In: Tappi International Paper Physics Conference,
Harwichport, MA. pp. 13 – 25.
Corson, S.R. (1989) Aspects of mechanical pulp fibre separation and
development in a disc refiner. Pap Puu-Pap Tim. 7:801 – 814.
Corson, S.R., Flowers, A.G., Morgan, D.G., Richardson, J.D. (2003)
Manipulation of paper structure and printability by control of
the fibrous elements. Int. Mech. Pulp. Conf., Quebec, Canada.
pp. 33.
Daniel, G., Bardage, S., Fernando, D., Hafr é n, J., Ander, P.
(2009) Energy consumption in refining of Scots pine and
Norway spruce TMP is governed by fibre morphology
and ultrastructure. In: International Mechanical Pulping
Conference, Sundsvall, Sweden. pp. 82 – 86.
Eriksen, O., Hammar, L.- Å . (2007) Refining mechanisms and
development of TMP properties in a low-consistency refiner.
Int. Mech. Pulp. Conf., Minneapolis, MN, USA.
Ferluc, A., Lanouette, R., Bousquet, J.-P., Bussiere, S. (2010)
Optimum refining of TMP pulp by fractionation after the first
refining stage. Appita J. 63:308.
Fernando, D. (2007) Ultrastructural characterization (morphological
and topochemical) of wood pulp fibres; effects of mechanical
and kraft processes. Doctoral thesis No. 2007:39. Swedish
University of Agricultural Sciences (SLU), Department of Forest
Products/Wood Science, Uppsala, Sweden.
Fernando, D., Daniel, G. (2004) Micro-morphological observations
on spruce TMP fibre fractions with emphasis on fibre cell wall
fibrillation and splitting. Nord. Pulp Pap. Res. J. 19:278 – 285.
Fernando, D., Daniel, G. (2008) Exploring Scots pine fibre
development mechanisms during TMP processing; impact of
cell wall ultrastructure (morphological and topochemical) on
negative behaviour. Holzforschung 62:597 – 607.
Fernando, D., Daniel, G. (2010) Characterization of spruce thermo-
mechanical pulps (TMP) at the fibre cell wall level: a method
for quantitative assessment of pulp fibre development using
Simons ’ stain. Tappi J. 9:47 – 55.
Fernando, D., Muhi ć , D., Engstrand, P., Daniel, G. (2011)
Fundamental understanding of pulp property development
during different TMP refining conditions using the application
of new Simons ’ stain method and fibre surface ultrastructure.
Holzforschung 65:777 – 786.
Fernando, D., Muhi ć , D., Engstrand, P., Daniel, G. (2012)
Surface and internal micro/ultrastructure of TMP fibres
produced during high-intensity refining elucidate the
development of pulp and paper properties. Holzforschung
66:467 – 475.
Gorski, D., Hill, J. (2012) Improved quality of SC magazine paper
through enhanced fibre development using the ATMP process.
Nord. Pulp Pap. Res. J. 27:35 – 41.
Gorski, D., M ö rseburg, K., Axelsson, P., Engstrand, P. (2011a)
Peroxide-based ATMP refining of spruce: influence of chemical
conditions on energy efficiency, fibre properties and pulp
quality. Nord. Pulp Pap. Res. J. 26:1 – 47.
Gorski, D., M ö rseburg, K., Axelsson, P., Engstrand, P. (2011b)
Peroxide-based ATMP refining of spruce: energy efficiency,
fibre properties and pulp quality. Nord. Pulp Pap. Res. J.
26:47 – 63.
Gorski, D., M ö rseburg, K., Johansson, L. (2011c) Influence of
process conditions and addition of chemicals on energy
efficiency and development of spruce fibers in peroxide-based
ATMP refining. Nord. Pulp Pap. Res. J. 26:232 – 247.
Gorski, D., M ö rseburg, K., Olson, J., Luukkonen, A. (2012a) Fibre
and fines quality development in pilot scale high and low
consistency refining of ATMP. Nord. Pulp Pap. Res. J.
27:872 – 881.
Gorski, D., Olson, J., Luukkonen, A. (2012b) Two-stage
low-consistency refining of mechanical pulp. Appita J.
65:244 – 249.
Gorski, D., Johansson, L., Engstrand, P. (2012c) Refiner bleaching in
a peroxide-based ATMP process compared with conventional
bleaching. Holzforschung 66:275 – 281.
Hartman, R.R. (1984) Mechanical treatment of pulp fibres for
property development. Doctoral thesis. Institute of Paper
Chemistry, Appleton, WI.
Heikkurinen, A., Levlin, J.-E., Paulapuro, H. (1991) Principles and
methods in pulp characterization- basic fibre properties. Pap.
Puu-Paper Timb. 73:411 – 417.
Hill, J., Sabourin, M., Aichinger, J., Johansson, L., Gorski, D.,
Axelsson, P. (2010) Combining selective bleaching chemistries
and ATMP technology for low energy mechanical pulping at
higher brightness. 7th Int. Sem. Fund. Mech. Pulp. Nanjing,
China. pp. 164.
H ä rk ö nen, E., Kortelainen, J., Virtanen, J., Vuorio, P. (2003) Fiber
development in TMP main line. International Mechanical
Pulping Conference, Quebec, Canada, pp. 171 – 178.
Johansson, L., Hill, J., Gorski, D., Axelsson, P. (2011) Improvement of
energy efficiency in TMP process by selective wood disinte-
gration and targeted application of chemicals. Nord. Pulp Pap.
Res. J. 26:31 – 46.
Kang, T., Paulapuro, H. (2006) New mechanical treatment for
chemical pulp. J. Process Mech. Eng. 220:161 – 166.
Kerekes, R.J. (2011) Force-based characterization of refining
intensity. Nord. Pulp Pap. Res. J. 26:14 – 20.
Kerekes, R.J., Senger, J.J. (2006) Characterizing refining action in
low-consistency refiners by forces on fibres. J. Pulp Pap. Sci.
32:1 – 8.
Kure, K.-A. (1999) On the relationship between process input
variables and fibre characteristics in thermomechanical
pulping. Doctoral thesis. NTNU, Trondheim, Norway.
Li, K., Lei, X., Lu, L., Camm, C. (2010) Surface characterization and
surface modification of mechanical pulp fibres. Pulp Pap.
Canada 111:28 – 33.
Brought to you by | Sveriges LantbruksuniversitetAuthenticated | [email protected] author's copy
Download Date | 3/14/13 10:40 AM
D. Fernando et al.: Fiber development in refining 11
Miles, K.B., Karnis, A. (1991) The response of mechanical and
chemical pulps to refining. Tappi J. 74:157 – 164.
Mohlin, U.-B. (1989) Fiber bonding ability a key pulp quality
parameter for mechanical pulps to be used in printing papers.
In: International Mechanical Pulping Conference, Helsinki,
Finland. Vol. 1. pp. 49 – 57.
Mohlin, U.-B. (2006) Refining intensity and gap clearance. 9th PIRA
Int. Refining Conf., Vienna, Austria, February 22 – 23.
Murton, K., Duffy, G. (2005) The influence of pulp residence
time on refining intensity and pulp quality. Appita J. 58:
64 – 71.
Paavilainen, L. (1993) Conformability – flexibility and collapsibility
– of sulphate pulp fibers. Pap. Puu-Paper Timb. 75:689 – 702.
Reme, P.A., Helle, T., Johnsen, P.O. (1998) Fibre characteristics of
some mechanical pulp grades. Nord. Pulp Pap. Res. J. 13:263.
Sabourin, M., Xu, E., Cort, B., Boileau, I., Waller, A. (1997) Optimizing
residence time, temperature and speed to improve TMP pulp
properties and reduce energy. Pulp Pap. Can. 98:111 – 118.
Skowronski, J. (1990) Surface roughening of pre-calendered
basesheets during coating. J. Pulp Pap. Sci. 16:102.
Stone, J.E., Scallan, A.M. (1965) A study of cell wall structure by
nitrogen adsorption. Pulp Pap. Mag. Can. 66:T407 – T414.
Brought to you by | Sveriges LantbruksuniversitetAuthenticated | [email protected] author's copy
Download Date | 3/14/13 10:40 AM