Influence of Thermomechanical Pulp Fiber Compositions

on Internal Curing of Cementitious Materials

Andrea Mezencevova1; Victor Garas2; Hiroki Nanko3; and Kimberly E. Kurtis4
Downloaded from ascelibrary.org by Sri Krishna College of Engineering & Technology on 01/31/17. Copyright ASCE. For personal use only; all rights reserved.

Abstract: This research examines the efficiency of thermomechanical pulp (TMP) fibers, both untreated and after chemical treatment to
produce holocellulose and α-cellulose, for internal curing. In addition, the effect of TMP fibers on early hydration behavior was investigated.
The results show that TMP and α-cellulose fibers had a negligible effect on cement hydration. Holocellulose, conversely, significantly re-
duced the rate of hydration, noticeably delaying setting time. The addition of the fibers in dosages to provide an additional amount of
entrained water of 0.05 g per gram of cement resulted in a decrease of early autogenous shrinkage of cement paste. Holocellulose, reducing
autogenous shrinkage of cement paste by 93%, was shown to be the most effective for internal curing; however, its adverse effect on cement
hydration may require acceleratory admixtures when used in concrete. Thermomechanical pulp and α-cellulose showed a similar efficiency in
mitigating autogenous shrinkage; these fibers reduced shrinkage of cement paste by approximately 51 and 45%, respectively. DOI: 10.1061/
(ASCE)MT.1943-5533.0000446. © 2012 American Society of Civil Engineers.
CE Database subject headings: Cement; Composite materials; Curing; Fibers; Shrinkage.
Author keywords: Autogenous deformation; Cement; Composite; Curing; Wood fiber.

Introduction
Wood pulp fibers have been increasingly used to reinforce cemen-
titious materials as they are nontoxic, renewable, and readily avail-
able at relatively low cost compared with other commercial fibers
(Savastano et al. 2009; Tonoli et al. 2007; Mohr et al. 2006; Naik
et al. 2004; MacVicar et al. 1999; Balaguru 1994). More recently,
wood-derived fibers and powders have been shown to be suitable
internal curing agents in cement-based materials (Mohr et al. 2005;
Kawashima and Shah 2011), providing mitigation against autog-
enous shrinkage, which is most prevalent at lower water-to-cement
ratios, i.e., less than 0.40 but especially less than 0.35 (Bentz and
Jensen 2004; Lura et al. 2007). Variability in the performance of
different types of wood-derived fibers and powders suggests that
their performance as internal curing agents may be related to their
structure and/or chemical composition.
An improved understanding of the mechanism(s) by which
wood-derived materials function this way would provide the basis
for their optimization for this application. Being hygroscopic,
wood-derived fibers contain both free and bound water, with the
1
Postdoctoral Fellow, School of Civil and Environmental Engineering,
Georgia Institute of Technology, 790 Atlantic Dr. NW, Atlanta, GA 30332.
E-mail: andrea.mezen@gatech.edu
2
ExxonMobil Upstream Research Company, Houston, TX 77252;
formerly, Graduate Student Research Assistant, School of Civil and
Environmental Engineering, Georgia Institute of Technology, Atlanta, GA.
E-mail: victor.y.garas-yanni@exxonmobil.com
3
President, Insight Technology International, LLC, 301 Barrington Hall
Dr., Suite 205, Macon, GA 31220. E-mail: hnanko@bellsouth.net
4 Experimental Study
Professor, School of Civil and Environmental Engineering, Georgia
Institute of Technology, 790 Atlantic Dr. NW, Atlanta, GA 30332-0355
(corresponding author). E-mail: kkurtis@ce.gatech.edu
Note. This manuscript was submitted on May 4, 2011; approved on
December 12, 2011; published online on December 14, 2011. Discussion
period open until January 1, 2013; separate discussions must be submitted
for individual papers. This paper is part of the Journal of Materials in Civil
Engineering, Vol. 24, No. 8, August 1, 2012. ©ASCE, ISSN 0899-1561/
2012/8-970–975/$25.00.
free water held in the lumen and between the fibers, and the bound
water found predominantly associated with the cell wall. For an
individual fiber or powder, the amount of free water present
may be related to the size of the lumen and its structure, e.g., col-
lapsed versus open and confined by cell wall versus unconfined in a
broken fiber, which are in turn related to the wood from which the
materials are derived and to the processing, e.g., pulping, drying,
and refining, methods. Hydrogen bonding is responsible for the
binding of polar water molecules to free hydroxyl groups present
on the primary chemical components of wood, cellulose, hemicel-
lulose, and lignin. The composition of the wood can be altered by
chemical and, to some degree, by mechanical processing, affecting
the relative capacity of the material to bind moisture. Currently, the
relative contributions of free and bound water to internal curing are
not fully understood.
In this research, the performance of a series of thermomechan-
ical pulp (TMP) fibers for internal curing of cementitious compo-
sites was assessed; various treatments were applied to the TMP
fibers to alter their composition to allow for the examination of
the relationship between pulp fiber chemical composition and in-
ternal curing. In this paper, ordinary TMP containing cellulose,
hemicelluloses, and lignin was compared with holocellulose
TMP containing primarily cellulose and hemicelluloses and
α-cellulose TMP containing primarily cellulose. On the basis of
data obtained by Mohr (2005), the capacity of the fibers to bind
water was expected to decrease from ordinary TMP to holocellu-
lose TMP to α-cellulose TMP.
Materials
A commercial TMP softwood fiber and the products resulting from
its chemical treatment, holocellulose and α-cellulose, were selected
for this study. Thermochemical pulp fibers are produced from
high-temperature refining processes; they appear to be attractive
for this application in terms of their high yields (80–90%) and

970 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / AUGUST 2012

J. Mater. Civ. Eng., 2012, 24(8): 970-975

Downloaded from ascelibrary.org by Sri Krishna College of Engineering & Technology on 01/31/17. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. SEM micrographs of (a) TMP fiber and the products of its chemical treatment; (b) holocellulose; (c) α-cellulose; LEO 1530 thermally assisted
field emission scanning electron microscope

low costs. The TMP fiber consisted of uncollapsed rodlike fibers used in the pastes. This amount of superplasticizer was used in
(open lumen) with an average fiber length of 1–2 mm, as shown in each of the samples to eliminate any potential effects of variations
scanning electron microscope (SEM) micrographs (Fig. 1) obtained in superplasticizer use. All tools and mixture components were
using a thermally assisted field emission scanning electron micro- kept at 25°C
3°C for 24 h before mixing.
scope (LEO 1530). The TMP fiber, with high lignin and hemicel-
lulose content, was chemically treated with hydrogen peroxide
and acetic acid to produce holocellulose, the carbohydrate Table 1. Oxide Analysis and Bogue Potential Composition for ASTM
portion of the fiber, with very low lignin content. Holocellulose Type I Portland Cement
was then treated with 18% sodium hydroxide to extract hemicellu- Component Percentage
loses from the fiber and yield α-cellulose (Rowell et al. 2005).
The fiber-cement pastes were prepared with a fixed water-to- SiO2 21.08
cement ratio (w∕c) of 0.30 on understanding that autogenous Al2 O3 5.03
shrinkage is increasingly significant at w∕c values below 0.35 Fe2 O3 2.97
(Lura et al. 2007). These were prepared with ASTM Type I CaO 62.85
Portland cement (Lafarge, North America) and deionized water MgO 3.15
(resistivity of 18:2 MΩ·m). Oxide analysis and Bogue potential Na2 O 0.09
composition of the cement is given in Table 1. For isothermal
K2 O 0.56
calorimetry experiments, fibers were added at a constant
Total alkalies as Na2 Oeq 0.46
dry fiber to cement ratio (f ∕c) by mass of 0.01, i.e., 1% fiber.
TiO2 0.26
For autogenous deformation measurements, differing fiber mass
fractions were used, depending on the absorption capacities of Mn2 O3 0.14
the fibers, to entrain an equivalent amount of entrained water P2 O5 0.08
(we ) of 0.05 g per 1 g of cement (we ∕c ¼ 0:05). This we ∕c SrO 0.04
ratio was chosen on data in (Jensen and Hansen 2001; Bentz et al. BaO 0.03
2005). SO3 2.8
The fiber-cement pastes were prepared by mixing the fibers in Loss on ignition 0.93
their as-received moisture state and approximately half of the Insoluble residue —
proportioned water using a single planetary mixer for 1 min to C3 S 49.65
ensure separation of the fibers. The cement and the remaining water C3 A 8.29
were subsequently added to the mixture, and the paste was mixed
C2 S 22.97
for 4 min. To aid workability, 2 μL of a carboxylated polyether
C4 AF 9.04
superplasticizer (ADVA100, WR Grace) per 1 g of cement was

JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / AUGUST 2012 / 971

J. Mater. Civ. Eng., 2012, 24(8): 970-975

 Absorption Capacity of Fibers in the pastes. Early heat evolution data of fiber-cement pastes
obtained from isothermal calorimetry, described previously, were
The capacity of fibers for “entraining” water or providing internally
compared with the control pastes to determine the “effective
stored curing water in cement-based materials was evaluated using
w∕c” (w∕ce ), which was then used to determine how to adjust
isothermal calorimetry. Absorption capacity (k) of the fibers was
(Δk) the assumed fiber absorption capacity (k A ) to calculate the
determined relying on heat evolution with time from data generated
actual absorption capacity (k) of the fibers [Eqs. (4) and (5)]. In
during isothermal hydration. In this method, which was described
Eq. (4), the numerator represents the water entrained by the fiber,
in detail by Johansen et al (2009), it is assumed that, depending on
and the denominator represents the relative dry-fiber mass. The cal-
their moisture content (MC) and assumed absorption capacity (k A ),
culated value of k was then used for additional executions of the
the fibers will either absorb, when MC < k, or release, when
method until the actual absorption capacity k was obtained. Hydra-
Downloaded from ascelibrary.org by Sri Krishna College of Engineering & Technology on 01/31/17. Copyright ASCE. For personal use only; all rights reserved.

MC > k, water from the cement mixture to equilibrate at their

tion curves of fiber-cement pastes behaved similarly to that of the
absorption capacity. The water stored in the fibers’ structure will
control mixture with no fibers at the fixed w∕c of 0.30.
be subsequently available for internal curing.
First, the moisture content MC (Table 2) of the as-received fiber w∕c  w∕ce
was calculated from Eq. (1), where M AR is the weight of the Δk ¼ ð4Þ
f ∕c
as-received fiber and M OD is its weight after drying in an oven
at 100°C for approximately 24 h:
M AR  M OD k ¼ kA þ Δk ð5Þ
MC ð%Þ ¼ × 100 ð1Þ
M OD Absorption capacities (k) of the studied fibers were determined
as follows: k ¼ 3:6 for TMP; k ¼ 3:3 for holocellulose; and
The assumed absorption capacity (k A ) of the studied fibers was k ¼ 5:0 for α-cellulose. These were then used for determining
chosen to be 3.3 on the basis of data obtained by Mohr (2005). the mixture proportions for the fiber-cement pastes for the
Fibers were added in the as-received condition (moist) at a dry-fiber hydration kinetics and autogenous deformation experiments.
mass-to-cement ratio f ∕c of 0.01. Moisture adjustments in the
design were on the basis of the difference between the MC of Moisture Desorption Rate of Fibers
the as-received fibers and their assumed absorption capacity
(kA ). The mass of mixing water (M W ) to maintain the w∕c of The tested fibers in their “saturated surface dry” (SSD) moisture
the mixture was found according to Eq. (2), with M CEM being condition were placed in an environmental chamber at 23°C,
the mass of cement proportioned for the mixture, and the second 95% relative humidity (RH), and allowed to equilibrate. The RH
term in Eq. (2) being the fraction of water added or removed from of 95% was selected to simulate moderate drying conditions within
the mixture by the fiber: the hydrated cement paste matrix. It has been shown that the inter-
nal relative humidity of a low water-to-cement ratio paste under
M W ¼ M CEM × fw∕c  ½ðMC∕100  k A Þ × f ∕cg ð2Þ sealed curing conditions can drop to 85% at early ages because
of self-desiccation (Henkensiefken et al. 2009; Jiang 2006). There-
The mass of the as-received fiber (M F ) was then determined as fore, moisture desorption of the fibers was also examined at this
follows, completing the design for the internally cured mixture: lower relative humidity. The fibers were brought to the SSD state
from the as-received condition by adjusting their water content so
M F ¼ M CEM × f ∕c × ð1 þ MC∕100Þ ð3Þ that the MC was equal to their absorption capacity k. First, the
moisture content of the fibers was determined and then a calculated
In addition to the fiber-cement pastes, ordinary cement pastes amount of water was mixed with the fibers to increase their MC to
were prepared at the w∕c ¼ 0:30 and at w∕c ratios above and below the level that corresponded to their absorption capacity k.
that w∕c (0.26 and 0.34) that served as bounding curves. To aid
workability, 2 μL of superplasticizer per 1 g of cement was used Isothermal Calorimetry
Ordinary Portland cement paste and cement pastes with addition of
Table 2. Composition of Cement Paste and Fiber-Cement Pastes for
1% dry fibers (f ∕c ¼ 0:01) were prepared. A total of 5–7 g of each
Autogenous Deformation Measurement sample was placed in pretared, temperature-equilibrated, polyethyl-
ene ampoules, which were then sealed and weighed. The time spent
Fiber-cement mixture loading the ampoule with the paste and placing into the calorimeter
Control
Parameter mixture TMP Holocellulose α-cellulose was minimized to 2 min to avoid further uncontrolled effects of
preparation on the hydration of the cement. The samples were
MC (%) — 44 222 244
maintained in an isothermal calorimeter (TAM Air, Thermometric
k — 3.6 3.3 5.0
AB, Sweden) at 25:0
1°C for 24 h. Measurements were recorded
Cement (g) 100 100 100 100 automatically every 60 s. The resulting heat of hydration data were
Water (g) 30 34.39 31.64 32.56 normalized per gram of cement and averaged over three replicates
Fiber, dry — 1.39 1.52 1.00 for each sample type.
mass (g)
w∕c 0.30 0.30 0.30 0.30 Autogenous Deformation
we ∕ca — 0.05 0.05 0.05 Autogenous deformations were measured for cement pastes con-
SP (μL∕g 2 2 2 2 taining fibers (TMP, holocellulose, and α-cellulose) and compared
of cement) with the cement paste with no fibers. The fibers were added at mass
Note: MC = fiber’s moisture content; k = fiber’s absorption capacity; fractions that would correspond to a constant entrained water-to-
SP = superplasticizer. cement ratio we ∕c ¼ 0:05. This we ∕c ratio was chosen on data from
a
Entrained w∕c, assuming that all water in fibers given by the k value will be Jensen and Hansen (2001) and Bentz et al. (2005) showing that at a
available for internal curing. basic w∕c of 0.30, additional entrained water approximately equal

972 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / AUGUST 2012

J. Mater. Civ. Eng., 2012, 24(8): 970-975

 to 0:05 g∕g cement should mitigate autogenous shrinkage of the
cement paste by providing enough water to prevent self-
desiccation. Depending on the absorption capacity k, previously
evaluated by isothermal calorimetry, the fibers were added at dry-
fiber to cement ratios of 0.014, 0.015, and 0.010 for TMP, holo-
cellulose, and α-cellulose, respectively (see Table 2).
Autogenous deformation of cement pastes during 28 days was
measured by the technique described by Jensen and Hansen (1995).
After mixing, the pastes were cast in corrugated rigid polyethylene
tubes, which were then sealed to prevent moisture exchange be-
Downloaded from ascelibrary.org by Sri Krishna College of Engineering & Technology on 01/31/17. Copyright ASCE. For personal use only; all rights reserved.
desorption rate was examined on fibers exposed to air; however,
in real cementitious systems, additional effects, e.g., cement pore
solution composition, may influence the rate and mass of water
released by fibers.
Early Hydration Kinetics of Fiber-Cement Pastes
The effect of the addition of TMP fiber before and after chemical
treatment (holocellulose, and α-cellulose) on the rate of hydration
of cement paste was assessed using isothermal calorimetry. Early

hydration curves, shown in Figs. 3 and 4, indicate that differences

tween the paste and environment and stored in an environmental between the control mixture and the mixtures containing TMP and
chamber at 23°C for the whole testing period. The initial linear de- α-cellulose are likely negligible in practice. These two fibers
formation measurement was taken at final set determined by Vicat showed little effect on the rate of hydration. It can be observed
needle penetration test (ASTM C 191) to exclude plastic deforma- in the hydration curves of these samples that the period of
tion. Subsequent length measurements were then taken on the first dormancy corresponding to a relatively low chemical reactivity
day at 1, 2, 4, and 6 h, then daily for a week, and then weekly for the was completed after approximately 2 h from the beginning of
rest of the testing period. the test, and was followed by an exothermic shoulder peak at ap-
proximately 9 h; then the heat flux decreased until the end of the
test. The cumulative heat evolved after 24 h was similar for TMP
Results and Discussion and α-cellulose. In addition, overall hydration, i.e., heat evolved,
for the two fibers was slightly higher (2.5–6%) as compared
Moisture Desorption Rate of Fibers with the control mixture (Fig. 4). These fibers, when incorporated
within the cement matrix, do not seem to present any notable
Holocellulose, TMP, and α-cellulose were dried at RH of 95% and
23°C to assess the rate of moisture desorption under conditions sim-
ulating moderate drying conditions within the hydrated cement
paste matrix. The moisture desorption curves for all the fibers
showed a very similar trend, as seen in Fig. 2. Among the fibers
examined, the TMP fiber exhibited the fastest desorption rate, los-
ing 60% of its moisture content during the first 2 days. Holocellu-
lose and α-cellulose had somewhat slower desorption rates; during
the first 2 days they released 45% of their initial moisture content.
Moisture desorption of the fibers was largely complete after 6 days
with all the fibers losing similar amount of water between 91
and 93%.
To evaluate the amount of moisture released from the fibers
under stronger drying, the fibers were equilibrated at 85% RH
and 23°C. The decrease in RH from 95 to 85% caused an additional
loss of moisture content in the fibers of approximately 3–5% still
without significant variation among the different fibers. On the ba-
sis of these results, it may be assumed that approximately 95% of Fig. 3. Heat evolution curves of cement pastes (w∕c ¼ 0:3) containing
water initially present in fibers, either because of presoaking or ab- internal curing materials (TMP, holocellulose, and α-cellulose) at the
sorbed during mixing, will be lost when the internal relative humid- dry-fiber dosage of 0.01 compared with the control cement paste
ity in cement-based materials decreases to 85%. The water (w∕c ¼ 0:3)

Fig. 2. Change in normalized moisture content of the fibers during Fig. 4. Cumulative heat evolved for cement pastes (w∕c ¼ 0:3) con-
moisture desorption at 95 and 85% RH and 23°C; one sample of taining internal curing materials (TMP, holocellulose, and α-cellulose)
each fiber was tested; moisture content of a fiber at its SSD is at the dry-fiber dosage of 0.01 compared with the control cement paste
considered 100% (w∕c ¼ 0:3)

JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / AUGUST 2012 / 973

J. Mater. Civ. Eng., 2012, 24(8): 970-975

 incompatibilities that would prohibit their use. However, testing on
concretes should be further performed.
Conversely the early hydration behavior of the cement mixture
containing holocellulose was different. On its hydration curve, the
period of dormancy lasted approximately 10 h and was followed by
a definite peak at 17 h indicating a delay in hydration of the fiber
mixture by approximately 8 h in comparison with the control mix-
ture. The delayed hydration of holocellulose-cement paste can be
observed in Fig. 4, in which the raise in cumulative heat, indicating
Downloaded from ascelibrary.org by Sri Krishna College of Engineering & Technology on 01/31/17. Copyright ASCE. For personal use only; all rights reserved.
the end of the dormancy period, appeared several hours later in
comparison with the control cement paste. This could be explained
by the fact that holocellulose contains a high amount of hemicel-
lulose, which is known to be very unstable in an alkaline medium,
being degraded into simple sugars that inhibit cement hydration
(Defo et al. 2004). It has been proposed that sugars and sugar acids
adsorb onto the Ca2þ ions that form the positively charged surface
of hydrating C3 S particles, thus poisoning the CSH and CaðOHÞ2
nucleation sites and inhibiting the hydration (Milestone 1979;
Thomas and Birchall 1983). A significant retardation of cement
hydration caused by hemicellulose was observed by Bilba et al.
(2003) who studied the influence of the botanical components,
i.e., water extractives, hemicellulose, cellulose, and lignin, on
the setting properties of bagasse/cement composites. In their tests,
a composite with hemicellulose addition did not set for 30 h.
However, as Fig. 3 shows, TMP fiber that is high in both lignin
and hemicelluloses content did not affect hydration of cement.
Similar results were obtained by Patil (1992) who found the cement
hydration to be retarded with higher holocellulose content but not
with the lignin. It is known that the hemicelluloses are soluble in
alkali, and can therefore be readily separated from the cellulose
component by alkali extraction. However, this can be done only
when the fiber has first been delignified (Roberts 1996). The hemi-
celluloses are likely linked to lignin through covalent ester link-
ages, which were found to be resistant to alkaline cleavage.
These bonds need to be cleaved before the hemicelluloses
can be solubilized (Obst 1982). This may probably be a possible
Fig. 5. Autogenous deformation for cement pastes containing wood
fibers compared with control nonfibrous mixture
Table 3. Results of Autogenous Deformation Measurements of the Cement- and Fiber-Cement Pastes
Parameter Control
Dry-fiber dosage (%) Not applicable
Time to final set (h) 6.40
Autogenous shrinkage (με) 642:3
54:4
Reduction in shrinkage (%) Not applicable
974 / JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / AUGUST 2012
J. Mater. Civ. Eng., 2012, 24(8): 970-975
explanation why TMP fiber, despite its high content of hemi-
cellulose, did not retard cement hydration. Lignin present in the
unbleached TMP fiber stiffens the fiber cell wall, and may act
as a physical chemical barrier between the hemicellulose and
alkaline cement pore solution.
Autogenous Deformation
To assess the efficiency of wood-derived fibers for internal curing
of cementitious composites, autogenous deformations were
measured for cement pastes containing TMP, holocellulose, and
α-cellulose fibers. As previously discussed, the fibers were added
at differing dry mass fractions, depending on their absorption
capacities, to entrain an equivalent additional amount of water
of 0.05 g per 1 g of cement (we ∕c ¼ 0:05). The dry-fiber dosages
were as follows: 1.39% TMP, 1.52% holocellulose, and 1.00%
α-cellulose (Tables 2 and 3). Length measurements began at final
set. Good workability was achieved at the tested fiber mass frac-
tions, with the TMP fiber being the most easily incorporated into
the fresh cement mixture. This effect may primarily result from the
shorter fiber length and, in the case of the TMP fibers, also from the
presence of lignin stiffening the fiber cell wall.
Results for autogenous deformation of the pastes are shown in
Fig. 5 and Table 3. Incorporating both untreated and chemically
treated TMP fibers resulted in a decrease of early autogenous
shrinkage. After 28 days, as seen in Fig. 5, the fiber composites
exhibited autogenous shrinkage of 315:3
29:2 με for the
TMP, 42:8
23:0 με for holocellulose, and 353:1
22:1 με
for α-cellulose, which, compared with the control nonfibrous mix-
ture, corresponded to approximately a 51, 93, and 45% decrease in
autogenous shrinkage of the cement paste, respectively. This effect
is likely because of the additional water provided by the fibers,
which counteracts self-desiccation in the mixture. Despite the ex-
pectation the fibers provide the same amount of entrained water in
each case, their effect on the early autogenous shrinkage of the
fiber-cement pastes was different. The most significant reduction
in autogenous shrinkage (93%) was observed for holocellulose.
During the first 2 days, composites incorporating this fiber exhib-
ited length expansion that was then followed by minimal shrinkage.
This suggests that the water contained in the holocellulose fiber was
being slowly released to the self-desiccating cement matrix, provid-
ing the moisture needed for continued internal curing. The
TMP fiber composite also showed some early expansion, although
the amount was less significant and occurred over a shorter duration
than that in holocellulose-cement paste. In addition, with the TMP
composites the expansion was followed by a greater shrinkage
(45% after 28 days). Among the tested fibers, α-cellulose was
shown to be the least effective in mitigating autogenous shrinkage
of cement pastes. After 28 days, the fiber decreased shrinkage of
the cement composite by approximately 51%. Similar autogenous
shrinkage behavior of fiber-cement composites was observed by
Mohr et al. (2005) when kraft fiber and cellulose powder compo-
sites exhibited greater autogenous shrinkage in comparison with
composites containing TMP fibers.
TMP Holocellulose α-cellulose
1.39 1.52 1.00
8.30 23.40 6.70
315:3
29:2 42:8
23:0 353:1
22:1
Approximately 51 Approximately 93 Approximately 45
 Results also show that autogenous shrinkage decreased on in-
creasing the dry-fiber content despite the fiber type (Table 3), which
may indicate a potential mechanical effect of the fibers because of
the internal restraint. Further tests should be performed with cement
pastes containing equivalent fiber dosage rates for all tested
fibers to elucidate the influence of mechanical effect of fibers
on autogenous deformation.
Fig. 5 shows that lowering the lignin content, i.e., TMP versus
holocellulose, resulted in an 86% decrease in autogenous shrinkage
while lowering both the lignin and the hemicellulose contents,
Downloaded from ascelibrary.org by Sri Krishna College of Engineering & Technology on 01/31/17. Copyright ASCE. For personal use only; all rights reserved.
i.e., TMP versus α-cellulose, did not result in a significant change
in autogenous deformation. This result suggests that additional
factors, likely microstructural but related to the cell wall composi-
tion and likely the rate of moisture diffusivity through it, play a
role in internal curing efficacy and require further investigation.
Overall, the results indicate that among the natural fibers
studied, holocellulose seems to be the most effective for internal
curing of concrete, reducing autogenous shrinkage of cement paste
by 93%. Conversely, addition of holocellulose significantly
retarded the early hydration of the fiber-cement pastes.
Conclusions
The objective of this work was to investigate wood-derived fibers
for their application as internal curing materials for cement-based
composites. Thermomechanical pulp fibers before and after chemi-
cal treatment (holocellulose and α-cellulose) were evaluated for
their capability to mitigate autogenous shrinkage. The effect of
the fibers on early cement hydration was also examined. On the
basis of the results obtained from the testing conducted, the
following conclusions can be drawn:
• Holocellulose significantly affected the rate of hydration, no-
ticeably delaying setting time. Conversely, TMP and α-cellulose
fibers showed little effect on cement hydration and did not seem
to present any notable incompatibilities that would prohibit their
use in cementitious composites.
• Incorporating the tested wood-derived fibers in dosages corre-
sponding to an equivalent additional entrained water dosage of
0.05 resulted in decreasing of early autogenous shrinkage of ce-
ment paste. Holocellulose, reducing autogenous shrinkage of
cement paste by 93%, was shown to be the most effective
for internal curing. However, its adverse effect on cement hy-
dration should be addressed, potentially with accelerating
chemical admixtures.
• Thermomechanical pulp and α-cellulose showed a similar effi-
ciency in mitigating autogenous shrinkage; these fibers reduced
shrinkage of cement paste by approximately 51 and 45%, res-
pectively. Higher fiber rates should be used to evaluate a
possible increase in their internal curing effect, and testing on
concretes should be performed. Also, further tests should
be performed with cement pastes containing equivalent fiber
dosage rates for all tested fibers to elucidate the influence of their
mechanical effect (internal restraint) on autogenous shrinkage. In
addition, further assessment to investigate the roles of additional
factors, i.e., microstructure and composition of fiber cell wall and
fiber distribution, in internal curing efficacy is needed.
References
ASTM. (2008). “Standard test methods for time of setting of hydraulic
cement by Vicat needle.” C191-08, West Conshohocken, PA.
Balaguru, P. (1994). “Contribution of fibers to crack reduction of
cement composites during the initial and final setting period.”
Mater. J., 91(3), 280–288.
JOURNAL OF MATERIALS IN CIVIL ENGINEERING © ASCE / AUGUST 2012 / 975
J. Mater. Civ. Eng., 2012, 24(8): 970-975
Bentz, D. P., and Jensen, O. M. (2004). “Mitigation strategies for autog-
enous shrinkage cracking.” Cem. Concr. Compos., 26(6), 677–685.
Bentz, D. P., Lura, P., and Roberts, J. W. (2005). “Mixture proportioning for
internal curing.” Concr. Int., 27(2), 35–40.
Bilba, K., Arsene, M.-A., and Ouensanga, A. (2003). “Sugar cane bagasse
fibre reinforced cement composites. Part, I. Influence of the botanical
components of bagasse on the setting of bagasse/cement composite.”
Cem. Concr. Compos., 25(1), 91–96.
Defo, M., Cloutier, A., and Riedl, B. (2004). “Wood-cement compatibility
of some eastern Canadian woods by isothermal calorimetry.” For. Prod.
J., 54(10), 49–56.
Henkensiefken, R., Bentz, D., Nantung, T., and Weiss, J. (2009). “Volume
change and cracking in internally cured mixtures made with saturated
lightweight aggregate under sealed and unsealed conditions.” Cem.
Concr. Compos., 31(7), 427–437.
Jensen, O. M., and Hansen, P. F. (1995). “A dilatometer for measuring
autogenous deformation in hardening cement paste.” Mater. Struct.,
28(181), 406–409.
Jensen, O. M., and Hansen, P. F. (2001). “Water entrained cement-based
materials I. Principles and theoretical background.” Cem. Concr.
Res., 31(4), 647–654.
Jiang, Z., Sun, Z., and Wang, P. (2006). “Internal relative humidity distri-
bution in high-performance cement paste due to moisture diffusion and
self-desiccation.” Cem. Concr. Res., 36(2), 320–325.
Johansen, N. A., Millard, M. J., Mezencevova, A., Garas, V. Y., and Kurtis,
K. E. (2009). “New method for determination of absorption capacity of
internal curing agents.” Cem. Concr. Res., 39(1), 65–68.
Kawashima, S., and Shah, S. P. (2011). “Early-age autogenous and drying
shrinkage behavior of cellulose fiber-reinforced cementitious material.”
Cem. Concr. Compos., 33(2), 201–208.
Lura, P., Jensen, O. M., and Igarashi, S.-I. (2007). “Experimental observation
of internal water curing of concrete.” Mater. Struct., 40(2), 211–220.
MacVicar, R., Matuana, L. M., and Balatinecz, J. J. (1999). “Aging mech-
anisms in cellulose fiber reinforced cement composites.” Cem. Concr.
Compos., 21(3), 189–196.
Milestone, N. B. (1979). “Hydration of tricalcium silicate in the presence of
lignosulfonates, glucose and sodium gluconates.” J. Am. Ceram. Soc.,
62(7–8), 321–324.
Mohr, B. J. (2005). “Durability of pulp fiber-cement composites.” Ph.D.
thesis, Georgia Institute of Technology, Atlanta, GA.
Mohr, B. J., Nanko, H., and Kurtis, K. E. (2006). “Aligned kraft pulp
fiber sheets for reinforcing mortar.” Cem. Concr. Compos., 28(2),
161–172.
Mohr, B. J., Premenko, L., Nanko, H., and Kurtis, K. E. (2005). “Exami-
nation of wood-derived powders and fibers for internal curing of
cement-based materials.” Proc., 4th Int. Seminar on Self-Desiccation
and Its Importance in Concrete Technology, B. Persson, D. Bentz,
and L. O. Nilsson, eds., NIST, Gaithersburg, MD, 229–244.
Naik, T. R., Friberg, T. S., and Chun, Y. M. (2004). “Use of pulp and paper
mill residual solids in production of cellucrete.” Cem. Concr. Res.,
34(7), 1229–1234.
Obst, J. R. (1982). “Frequency and alkaline resistance of lignin-
carbohydrate bonds in wood.” Tappi J., 65(4), 109–112.
Patil, S. G. (1992). “Study on the utilization of some waste materials of
forests for wood cement composites and pulping.” Ph.D.. thesis, Dr.
Yashwant Singh Parmar Univ. of Horticulture and Forestry, Dept. of
Forest Products, Solan, India.
Roberts, J. C. (1996). The chemistry of paper, Royal Society of Chemistry,
Cambridge, UK, 190.
Rowell, R. M., Pettersen, R., Han, J. S., Rowell, J. S., and Tshabalala, M. A.
(2005). “Cell wall chemistry.” Handbook of wood chemistry and wood
composites, R. M. Rowell, ed., CRC, Boca Raton, FL, 63–64.
Savastano, H., Santos, S. F., Radonjic, M., and Soboyejo, W. O. (2009).
“Fracture and fatigue of natural fiber reinforced cementitious compo-
sites.” Cem. Concr. Compos., 31(4), 232–243.
Thomas, N. L., and Birchall, J. D. (1983). “The retarding action on cement
hydration.” Cem. Concr. Res., 13(6), 830–842.
Tonoli, G. H. D., Joaquim, A. P., Arsene, M. A., Bilba, K., and Savastano,
H. (2007). “Performance and durability of cement-based composites
reinforced with refined sisal pulp.” Mater. Manuf., 22(2), 149–156.