Full Paper

Renewable Resources as Reinforcement of

Polymeric Matrices: Composites Based on
Phenolic Thermosets and Chemically
Modified Sisal Fibers

Jackson D. MegiattoJr., Franciéli B. Oliveira, Derval S. Rosa,

Christian Gardrat, Alain Castellan,* Elisabete Frollini*

Lignocellulosic materials can significantly contribute to the development of composites, since

it is possible to chemically and/or physically modify their main components, cellulose, hemi-
celluloses and lignin. This may result in materials more stable and with more uniform properties.
It has previously been shown that chemically modified sisal fibers by ClO2 oxidation and reaction
with FA and PFA presented a thin coating layer of PFA on their surface. FA and PFA were chosen as
reagents because these alcohols can be obtained from renewable sources. In the present work, the
effects of the polymeric coating layer as coupling agent in phenolic/sisal fibers composites were
studied. For a more detailed characterization of the fibers, IGC was used to evaluate the changes
that occurred at the sisal fibers surface after the chemical modifications. The dispersive and
acid-base properties of untreated and treated sisal fibers surfaces were determined. Biodegrada-
tion experiments were also carried out. In a complementary study, another PFA modification was
made on sisal fibers, using K2Cr2O7 as oxidizing agent. In this case the oxidation effects involve
mainly the cellulose polymer instead of lignin, as observed when the oxidation was carried out
with ClO2. The SEM images showed that the oxidation of sisal fibers followed by reaction with FA
or PFA favored the fiber/phenolic matrix interaction
at the interface. However, because the fibers were
partially degraded by the chemical treatment, the
impact strength of the sisal-reinforced composites
decreased. By contrast, the chemical modification of
fibers led to an increase of the water diffusion coeffi-
cient and to a decrease of the water absorption of the
composites reinforced with modified fibers. The latter
property is very important for certain applications,
such as in the automotive industry.

J. D. Megiatto Jr., F. B. Oliveira, E. Frollini

Instituto de Quı́mica de São Carlos, Universidade de São Paulo,
USP, C.P. 780, CEP 13560-970 São Carlos, SP, Brazil
Fax: þ 551633739952; E-mail: elisabete@iqsc.usp.br
J. D. Megiatto Jr., C. Gardrat, A. Castellan
Université Bordeaux 1, US2B, UMR 5103 CNRS-INRA-UBx1, F-33405
Talence Cedex, France
Fax þ 33540006439; E-mail: a.castellan@us2b.u-bordeaux1.fr
D. S. Rosa
Universidade de São Francisco, Programa de Pós-Graduação em
Engenharia e Ciência dos Materiais, Laboratório de Polı́meros
Biodegradáveis e Soluções Ambientais - CEP 13251-900 Itatiba,
SP, Brazil

Macromol. Biosci. 2007, 7, 1121–1131

ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim DOI: 10.1002/mabi.200700083 1121

Introduction
In recent years, the incorporation of lignocellulosic
materials as reinforcing agents or as fillers in polymer
composites has received an increased attention.[1–6] The
main advantage of lignocellulosic fibers upon their
mineral counterpart is their environmental friendliness,
due, for instance, to the CO2 neutral life cycle and the
possibility of biodegradation or incineration for energy
recovery after disposal.[7] The use of lignocellulosic
materials as replacements for synthetic fibers in the
production of composites has gained significant impor-
tance in technical applications, such as in the automotive,
as well as in the packaging industries.[8–10]
Unfortunately, several disadvantages of natural fibers
such as mechanical and thermal degradation during
processing can make them undesirable for certain applica-
tions. Thermal degradation of the fibers occurs above
200 8C, resulting in gaseous products, which can cause high
porosity and lower mechanical properties of the compo-
sites.[8] This problem is considerably diminished in the
present work because, contrarily to many thermoplastics
matrices, the processing of phenolic thermosetting resins
takes place at temperatures below 200 8C, as it will be
detailed later.
However, the main disadvantage of these natural fiber/
polymer composites seems to be the incompatibility be-
tween hydrophilic lignocellulosic fibers and the hydro-
phobic polymeric matrices. The resulting composites
present low mechanical properties, due to a lack of adhe-
sion between the fibers and the resin at the interface. This
problem is minimized when phenolic matrices are used, as
in the present work, due to polar hydroxyl groups in their
structure. To produce composites with good mechanical
properties, it is necessary to promote improvements in the
interface area, for example by increasing the adhesion
between the fiber and the polymeric matrix through a
chemical modification of the fiber surface and/or by
decreasing the thickness of the fiber bundles. There are
several methods in the literature for promoting interfacial
adhesion in systems where lignocellulosics are used as
reinforcement. However, in a majority of these methods,
the modifier agents are obtained from petroleum.
In a series of papers,[9–12] we have applied a new way of
chemically modifying the surface of lignocellulosic fibers.
This method involves a selective modification of the lignin
polymer partly preserving the cellulose. The oxidation of
the fibers concerned a selective oxidation by sodium
periodate and/or chlorine dioxide of guaiacyl and syringyl
units of lignin, generating ortho and para-quinones and
muconic derivatives able to react with furfuryl alcohol
(FA). This latter substance can be easily obtained by
reduction of furfural, which is isolated from renewable
sources.[13] In a first attempt, this method was applied to
Macromol. Biosci. 2007, 7, 1121–1131
1122 ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
J. D. Megiatto Jr., F. B. Oliveira, D. S. Rosa, C. Gardrat, A. Castellan, E. Frollini
sugarcane and curaua fibers. The results showed that the
FA-modified sugarcane and curaua fibers had a thin
coating layer of poly(furfuryl alcohol) (PFA), indicating that
the FA polymerization reaction had occurred.[10] The PFA
coating layer formed after the treatments favored fiber/
phenolic matrix interaction at the interface, which
increased the adhesion between fibers and matrix polymer
in composites materials.[10] However, the chemical treat-
ments also caused some fibers degradation that affected
the mechanical properties and decreased the impact
strength of the corresponding composites, indicating that
the chemical process conditions were too harsh for those
natural fibers.[11]
In our continuous interest to elaborate composites
based on phenolic resins and various natural modified
fibers,[14–16] we here present results obtained with chemi-
cally modified sisal fibers. The aim of the study was to
submit the sisal fibers to milder chemical process condi-
tions than those used in the sugarcane and curaua fibers,
aiming to improve both fiber/matrix interactions at the
interface and the composites properties. In a companion
paper,[17] the sisal fibers were oxidized with ClO2 and
treated with FA and PFA (previously prepared). The chemi-
cal modification was based on a selective oxidation of the
guaiacyl and syringyl phenolic groups of lignin into ortho-
and para-quinones and muconic derivatives. In order to
assure that only the chemically grafted chains were
retained at the surface, after both reactions (with FA
and PFA) the chemically modified fibers were submitted to
Soxhlet extraction, using in each case the adequate sol-
vent. The modified sisal fibers were characterized by
spectroscopy techniques and thermal analysis. The scan-
ning electron microscopy (SEM) images suggested that the
both modified sisal fibers presented a thin coating layer of
PFA on their surfaces.[17] In the present work, other kinds of
fibers grafted with PFA were prepared, for comparison
with those previously modified.[17]
In this work, in order to investigate the effects of this
polymeric coating layer around the fibers as coupling
agent between sisal fibers and a phenolic matrix, com-
posites were prepared and characterized. SEM images were
used to evaluate fiber/phenolic matrix interaction at the
interface. The mechanical behavior of the composites and
fibers were analyzed by impact and tensile tests,
respectively. Water absorption capacity also was studied.
For more detailed characterization of the fibers, Inverse
Gas Chromatography (IGC) was used to evaluate the
changes that occurred at the sisal fibers surface after
the chemical modifications. Biodegradation experiments
(simulated soil and cellulase tests) of the unmodified and
chemically modified fibers were also carried out. To the
best of our knowledge, this is the first study related to the
chemical modification of sisal fibers using FA or PFA as
reagents, as well as for the first time results obtained from
DOI: 10.1002/mabi.200700083
Renewable Resources as Reinforcement of Polymeric Matrices: . . .

IGC, tensile strength and biodegradation experiments,
carried out for unmodified and modified sisal fibers, are
correlated to each other and with the properties of
composites reinforced with those fibers.
TA Instruments operating at 25 8C and 1 N/min up to 18 N. A
minimum of thirty samples of each material was tested and
average values are reported in the next section.
Biodegradation experiments were carried out using the
following conditions:

Immersion in a liquid medium with cellulase: each sample was

Experimental Part
General
Sisal fibers (kindly donated by Incomar SP, Brazil) were Soxhlet
extracted with a mixture of cyclohexane/ethanol (1:1, v:v) for 48 h,
and subsequently with water for 24 h. The fibers were dried in an
air-circulating oven (100 8C) until constant weight. The ‘‘extra-
ctives-free’’ sisal fibers were then used to make composites.
Alkane series (pentane to decane) were obtained from Aldrich,
tetrahydrofuran from J. T. Baker and chloroform from Merck.
placed in a vial filled with 10 ml of acetate buffer, pH 6.0, contain-
ing 0.25 g of cellulase, and then incubated in a thermostated oven
at 35 8C. The buffer/enzyme system was changed every 24 h
during the evaluation period in order to maintain the original
level of enzymatic activity. Every 48 h, the samples were removed
from the incubation medium, washed with distilled water, wiped
dry, weighed, and examined by light microscopy before being put
back in the incubation medium. The control samples were incu-
bated in buffer without enzyme.

Exposure in simulated soil: the samples were weighed and buried

in the simulated soil containing 23% loamy silt, 23% organic
Sisal Fibers Treatments matter (cow manure), 23% sand and 31% distilled water (w/w).
The biodegradation was monitored every 15 days by measuring
The preparation of PFA and the chemical modification of sisal
the mass retention.
fibers (ClO2 þ FA and ClO2 þ PFA) were described in full detail in
the companion paper.[17] In general lines, sisal fibers were oxidized
IGC was used to evaluate the dispersive and acid-base pro-
by an aqueous ClO2 solution, at 55 8C. The oxidized fibers were
perties of the untreated and treated fibers. For this purpose, the
washed with water until neutrality and, after drying impregnated
fibers (with lengths close to 1 mm) were packed into a column and
with FA and heated at 55 8C for 45 min. The excess of FA was
analyzed by a Shimadzu GC-17A gas chromatograph equipped
removed by Soxhlet extraction using ethanol for 16 h. The
with a flame ionization detector (FID). The columns were made
experiment with PFA was conducted under the same experi-
from stainless steel (316) with an internal diameter of 5.0 mm and
mental conditions, except the final extraction, which was made
a length of 2.0 m. The injector and detector temperatures were
with acetone.
150 8C, and the column temperature was 35 8C. The flow rate of
In the present work, the grafting reaction was carried out
the carrier gas (N2) was 30 mL  min1, the injected volume of the
through the generation of radicals produced by a K2Cr2O7 action
vapor probes was 10 mL and the injection was carried out with a
on carbohydrates according to the procedure of Nguyen.[18]
gas-tight Hamilton micro-syringe (very small amounts of the
K2Cr2O7 (5 mmol) was dissolved in distilled water (100 mL,
probes were injected in order to work at infinite dilution condi-
pH
2) upon stirring, followed by addition of PFA (2.5 g). The
tions). The non-polar probes were a series of n-alkanes from pentane
mixture was heated at 90 8C under stirring until complete
to decane, with methane as the marker, and the polar probes were
dissolution of PFA. At this time, sisal fibers (1 g, length: 30 mm)
tetrahydrofuran and chloroform. Each probe was tested in triplicate
were added and the stirring was continued for 20 min. The fibers
and the reliability of the three trials was very good for all
were soxhlet extracted with acetone for 16 h, and then dried for
measurements.
24 h at 55 8C. The weight gain was about 13%.
The dispersive component of the surface free energy was
calculated according to the Schultz approach[19] and was corrected
for asymmetric peaks.[20,21] The time each n-alkane probe was
Fiber Characterization
retained on the GC column was recorded and from this retention
Contents of humidity, ash, Klason lignin, hemicelluloses, time, tR, and that of the marker, to, the net retention volume, Vn, was
a-cellulose (pure cellulose), crystallinity index of sisal fibers were obtained according to:
determined in a companion paper.[17] Averages were calculated
Vn ¼ JDðtR  to Þ (1)
based on a minimum of three samples for all the mentioned
analyses (see Table 1).
The tensile strength of the fiber bundles (length, 15 mm; where J is the flow-rate of the carrier gas, and D is a correction
diameter, near 0.5 mm) was analyzed on a DMA model 2980, from factor of the gas compressibility.[20]

Table 1. Contents of unmodified sisal fiber in %.

Component Cellulosea) Hemicelluloseb) Klason Lignin Ash

Content 65 W 1 20 W 1 12 W 1 1 W 0.1

a)
The crystallinity index is 66%; b)Determined as difference between holocellulose (cellulose R hemicellulose) and cellulose content.

Macromol. Biosci. 2007, 7, 1121–1131

ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mbs-journal.de 1123

Since only an extremely small volume of each probe was
injected, the adsorption-desorption process could be considered as
the dominant process in the GC, and diffusion and absorption
phenomena could be neglected as a first approximation.[20,22]
Thus, the following thermodynamic relation was used to deter-
mine the London component of the surface free energy of the
cellulose (g DS ):

1=2
1=2
RT ln Vn ¼ 2N g DS a g DL
Here N is Avogadro’s number, a the surface area of the probe
molecule, g DL the dispersive component of the surface free
energy of the probe in the liquid state, R the gas constant
(8.314 J K1 mol1) and T the absolute temperature of the column.
When n-alkane series are used as probes, their interactions with
the substrates under investigation are due only to the dispersion
forces. A plot of the observed value of RT ln Vn against a (g DL )1/2
would thus give rise to a straight line of slope (g DS )1/2, known as the
reference line (figures not shown).[22,23]
The acid-base properties of fiber surfaces can be obtained from
Equation (3).
 
ln Vn
DGSP
A ¼N a WASP ¼ RT
ln Vnref
where DGSP
A is the free energy of adsorption, Vn is the retention
volume of the polar probe under consideration and Vnref is
obtained from the n-alkane reference line, at the appropriate
(g DL )1/2 value. Several approaches are mentioned in the literature to
evaluate the acid-base properties of solid surfaces using IGC, and
among these, probably one of the simplest is related to the
investigation of specific interactions between the surface under
analysis and two polar molecules, i.e. donor and acceptor
probes.[20,23] The DGA value is determined for each probe, and
the obtained values are considered to describe the acid (ANs) and
basic (DNs) properties of the surface [Equation (4) and (5),
respectively].[24]
sp
DGA donor  ANs
sp
DGA acceptor  DNs
The value of the ratio ANs/DNs can be taken as an indication of
the predominance of acidic (ANs/DNs  1.1) or basic (ANs/
DNs  0.9) sites at the interface under investigation. In this
arbitrary scale, amphoteric and neutral (non-polar) surfaces are
related to ANs/DNs values in the range of 0.9–1.1 and near 0,
respectively.[24]
A detailed description of the theory involved in the determina-
tion of the dispersive and polar characteristics of the fiber surfaces
by IGC is beyond the scope of this paper, but elaborated
explanations on this subject can be found elsewhere.[20,24,25]
Composite Preparations[11,26]
A phenolic pre-polymer was synthesized by mixing phenol,
formaldehyde (37%) and potassium hydroxide (1.38%:1.00%:
Macromol. Biosci. 2007, 7, 1121–1131
1124 ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
J. D. Megiatto Jr., F. B. Oliveira, D. S. Rosa, C. Gardrat, A. Castellan, E. Frollini
0.06%) under mechanical stirring, at 70 8C, for 1 h. Subsequently,
the solution was cooled to room temperature, water was elimi-
nated under reduced pressure and the mixture was neutralized
with concentrated hydrochloric acid.
Composites reinforced with sisal fibers (either chemically
modified or unmodified) were obtained by adding the fibers (40 g)
to the pre-polymers (229 g), and submitting the mixture to
mechanical stirring for 30 min at 50 8C (JVJ mixer, Pardinho, SP,
(2) Brazil). This was done in order to obtain an optimum impregna-
tion of the lignocellulosic materials by the pre-polymer. The
compression molding was carried out in a mold (300  140 
5 mm) under pressure. The molding cure cycle (75 8C/30 min/
5.0 ton; 75 8C/30 min/7.5 ton; 85 8C/30 min/10.0 ton; 85 8C/
30 min/12.5 ton; 85 8C/30 min/15.0 ton; 95 8C/30 min/17.5 ton;
105 8C/30 min/20.0 ton; 115 8C/1 h/20.0 ton; 125 8C/1.5 h/20.0 ton)
was previously determined by DSC measurements.[27] Composites
were prepared with randomly oriented fibers (near 30 mm
in length).
Composite Characterizations[11]
For the Izod impact tests, ten unnotched samples were cut from
each plate (63.5  12.7  4.0 mm) and shaped according to the
(3)
ASTM D256 standard. The impact strength was assessed using an
Izod impact tester (CEAST Resil 25) and the tests were carried out
at room temperature with an impact speed of 4 m  s1 and an
incident energy of 5.5 J. In each experiment, at least ten
measurements were performed to give an average calculation.
SEM was carried out in a Zeiss-Leica apparatus model 440,
electron acceleration 20 kV. Fractured samples were covered with
a thin layer of gold in a sputter coating system prior to analysis.
Water absorption studies were performed according to the
ASTM D 570-98 standard. The dimensions of the specimens for the
water uptake experiments were 76.2  25.4  3.2 mm. Three
specimens of each sample were submerged in distilled water at
three temperatures, i.e. 23, 55 and 72 8C.[28] The specimens were
removed from the water after certain periods of time, wiped with
a cloth, and weighed in a high precision balance before they were
(4) submersed again in water. The content of water was calculated by
the weight difference.
(5)
Results and Discussion
Chemical Composition of Sisal Fibers
Table 1[17] presents the chemical composition of the sisal
fibers. The high content of cellulose and the high value of
crystallinity index make these fibers very appropriate
reinforcement material to be applied in polymer compo-
sites, since normally those properties lead to good
mechanical resistance.
Dispersive and Acid-Base Properties of
Untreated and Treated Sisal Fibers
Dispersive and acid-base properties of unmodified and
chemically modified sisal fibers were obtained through
DOI: 10.1002/mabi.200700083
Renewable Resources as Reinforcement of Polymeric Matrices: . . .
Scheme 1. PFA chain grafted on sisal fiber surface.
IGC analyses. As it was already mentioned, PFA chains
were introduced at fibers surfaces using three different
processes: ClO2 þ AF, ClO2 þ PFA (described in a companion
paper, as previously mentioned)[17] and K2Cr2O7 þ PFA. The
PFA chain structure that was grafted onto the sisal fiber
surfaces is exemplified in Scheme 1.
Values of the dispersive component of the surface free
energy (g DS ) and the acid-base properties of the unmodified
and modified sisal fiber surfaces could be obtained from
IGC data and Equation (2), (4) and (5) (Table 2).
The data of Table 2 show that all types of chemical
treatment considerably increased the dispersive compo-
nent of the free energy surface of the fibers (g DS ). The
stronger the interaction between the sample and the
non-polar volatile probes, the greater were the slope of
the reference line, (g DS )1/2, the net retention volume Vn
[Equation (2)] and the column retention time tR [Equation
(1)]. Thus, a treatment leading to high values of g DS is an
indication of decreasing polar characteristics of the fiber
surface.
The previous studies on sugarcane bagasse and curaua
fibers[9–11] have shown that lignin is the main component
involved in the reaction with ClO2/FA and ClO2/PFA. In the
companion paper,[17] similar experiments were performed
on sisal fibers. In order to get a better understanding of the
chemical modification involved, the reaction was per-
formed on lignin extracted from sisal fibers. Hydroxyl
Table 2. Dispersive component of surface free energy and acid-
base properties of sisal fibers at 30 8C.
Sisal fiber g DS ANs DNs
mJ  mS2
Untreated 21.0 4 676 2 669
K2Cr2O7 R PFA 70.0 3 617 2 665
ClO2 R AF 63.0 4 369 3 036
ClO2 R PFA 65.0 3 481 2 003
Macromol. Biosci. 2007, 7, 1121–1131
ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
groups, and more specifically free phenolic hydroxyls,
define the reactivity of lignin during the oxidation process
and the treatments with FA and PFA. To quantify the
different hydroxyl groups present in sisal lignin samples,
31
P NMR spectra of the phosphitylated polymers were
recorded. The results showed that the reaction induces a
reduction of both aliphatic and phenolic hydroxyl groups
(11% and 31%, respectively). Concerning the fibers, the
results obtained from 13C NMR CP-MAS (cross polarization
magic angle spin nuclear magnetic resonance)[17] indi-
cated that, besides lignin, the carbohydrate matrix was
also slightly affected. Then, the lower availability of the
polar hydroxyl groups on the fiber surfaces led to a
decreasing of the polar characteristics of the fiber surface,
and then to higher g DS values of the treated fibers when
compared to the untreated ones. Furthermore, the poly-
meric chains introduced in the fiber surface had non-polar
moieties, which could also increase the g DS value (Table 2).
The reaction of fibers with K2Cr2O7 þ PFA led to an even
higher g DS value when compared to those found for the
fibers reacted with ClO2/FA or ClO2/PFA (Table 2). This
result can be taken as an indication of the higher
effectiveness of the K2Cr2O7 oxidation reaction. This
increase in the non-polar character of the fiber surfaces,
due to the chemical treatment, agreed with the results
obtained from water absorption measurements (discussed
later), among others.
The Table 2 shows the data related to the acid-base
character of the fiber surfaces. All fibers (both treated and
untreated) presented more sites with an acidic character
than basic ones [Table 2, AN/DN  1.1, see Exp. Part,
Equation (4), (5)]. However, the inspection of the individual
ANs and DNs values can give more detailed information.
When compared to the untreated fibers, those treated
with K2Cr2O7 and then with PFA showed lower ANs and
similar DNs values. It can be considered that the acid
properties (ANs) of the fiber surfaces are mainly due to the
hydroxyl groups that are naturally present in the fibers, as
well as to hydroxyl groups linked at the end of the PFA
chains (Scheme 1) introduced at the fiber surface by the
chemical modification. It can be proposed that an amount
of the hydroxyl groups naturally present on the fiber
surface had reacted, thus decreasing the inherent surface
acid properties. This could be compensated by hydroxyl
groups present at the end of the introduced PFA chains
(Scheme 1). However, the unreacted hydroxyl groups
ANs/DNs present at the fiber surface were probably less accessible to
the probes after reaction, due to the steric hindrance
caused by the presence of PFA, leading to a lower value of
ANs. Concerning the basic sites, the grafting of PFA chains
1.8
was able to hinder the interactions between the basic sites
1.4 present at the fiber surface and the probe, but also to
1.4 introduce several new basic sites, leading to a DNs value
1.7 near that of the untreated fiber.
www.mbs-journal.de 1125

The fiber treated with ClO2/PFA presented an acidic
value (ANs, Table 2) similar to that of the fibers treated
with K2Cr2O7/PFA, and the same previous discussion can
be considered. However, the DNs value was somewhat
lower than that obtained for the untreated fibers. As was
mentioned in the previous discussion, the oxidation
reaction seemed to have occurred more efficiently
with K2Cr2O7 than with ClO2, which in turn probably
led to a more efficient reaction with PFA. Thus, the fibers
having reacted with ClO2/PFA probably had a lower
content of PFA chains on their surface, and consequently a
lower content of basic sites were introduced. Therefore, the
difficulty of accessibility to the basic sites that are
naturally present on the fiber surface was not compen-
sated, as it probably was for the fibers treated
with K2Cr2O7/PFA, as previously mentioned.
The fibers treated with ClO2 and then with FA had a
value of ANs closer to that of the untreated fibers than to
the other treated fibers (Table 2). Considering that the PFA
chains were introduced on the fiber surface via an in situ
polymerization, probably the chains were shorter than
those introduced when the PFA was prepared prior to their
introduction on fiber surface. Thus, both hydroxyl groups
(present on the fiber surface and at the end of PFA chains,
Scheme 1) were more accessible, giving rise to a value of
ANs closer to that of untreated fibers. The DNs value for the
fiber treated with ClO2/FA was higher than that of the
untreated fibers (Table 2). Again, the PFA short chains,
which were probably introduced through in situ poly-
merization, could at a first glance explain this result. The
access to both basic sites inherent of fiber surface and
those present at short PFA chains was easier, as compared
to the fibers having reacted with PFA prepared prior to
their introduction on fiber surface, leading to a DNs value
even higher than that of the untreated fibers.
In the present work, alongside the intrinsic interest in
verifying the influence of the chemical treatments on both
the acid-base properties and the dispersive component of
surface free energy, it was also an objective to search for a
possible correlation between these parameters and the
fiber/matrix interactions at the interface of the composites
reinforced with these fibers, as it will be discussed later.
Tensile Strength
The tensile strength of the fibers is shown in Table 3. All
types of chemical modification led to decreases in the
tensile strength and in the percentage of elongation of the
sisal fibers. However, the results obtained for the fibers
oxidized with K2Cr2O7 showed that this reagent probably
affected the polysaccharides matrix quite severely. The
degradation of the cellulose chain was possibly the main
factor responsible for the lowering of the mechanical
Macromol. Biosci. 2007, 7, 1121–1131
1126 ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
J. D. Megiatto Jr., F. B. Oliveira, D. S. Rosa, C. Gardrat, A. Castellan, E. Frollini
Table 3. Tensile strength of unmodified and chemically modified
sisal fibers.
Sisal fibers Tensile Strength Elongation
MPa %
Unmodified 188 W 20 1.4 W 0.3
ClO2 oxidized 149 W 33 1.4 W 0.4
ClO2 oxidized and 147 W 25 0.9 W 0.5
modified with FA
ClO2 oxidized and 157 W 18 0.6 W 0.1
modified with PFA
K2Cr2O7 oxidized and 42 W 13 0.5 W 0.1
modified with PFA
properties observed for the fibers after oxidization
with K2Cr2O7 (Table 3).
Biodegradation
Figure 1 shows the results obtained from biodegradation
analysis of the fibers. Hemicellulose can be considered as
the fiber component most vulnerable to biodegradation,
because it is located in the non-crystalline domains. Next
in line are non-crystalline and crystalline cellulose,
whereas protolignin is less prone to biodegradation, due
to its crosslinked and aromatic character.
Results from soil simulation tests showed that, early on
in the test, fibers oxidized with ClO2 and reacted with FA
were less resistant to biodegradation than those modified
with ClO2 and PFA. As already indicated by the IGC results,
probably the polymeric chains grafted on the fiber surface
during the modification with FA were shorter than those
grafted with PFA chains. Moreover, this shorter chain
length was believed to facilitate the contact between the
fibers and the microorganisms.
In the companion paper,[17] we showed that the chemi-
cal modification of sisal fibers with chlorine dioxide and FA
should mainly affect the lignin component, but the CP-
MAS 13C NMR analyses indicated that the carbohydrate
matrix was also affected, as already mentioned in the
present paper. The peaks due to carbohydrates were less
intense, probably due to partial oxidation of both hemi-
celluloses and amorphous cellulose.[17]
The biodegradation results showed that the reaction
with K2Cr2O7 reached various fiber domains as compared
to the fibers oxidized with ClO2. The fibers oxidized
with K2Cr2O7 and further reacted with PFA seemed to be
protected against the microorganism degradation during
the first 35 d. The previously discussed properties of the
fibers oxidized with K2Cr2O7 showed that this reagent
DOI: 10.1002/mabi.200700083
Renewable Resources as Reinforcement of Polymeric Matrices: . . .
Figure 1. Biodegradation results for unmodified and modified
fibers: a) soil simulation and b) cellulase tests.
probably also touched the polysaccharide matrix to a
reasonable extent, and thus between 35 and 45 d, the
microorganisms were able to totally degrade these fibers.
This partial degradation of the fibers treated with
K2Cr2O7 and ClO2 was believed to have facilitated the
access of the microorganisms to all the domains of the
fibers, since the treated fibers became totally biodegraded
during the considered time interval. On the other hand, the
untreated fibers only displayed a mass loss of 40% of mass
(i.e. near 60% retained mass) during the same time interval
(Figure 1). It is probable that the mass loss for the
untreated fiber was mainly related to the biodegradation
of hemicelluloses (near 19%, Table 1) and non-crystalline
cellulose (total content of cellulose: near 65%, Table 1).
Regarding the biodegradation experiments with cellu-
lase, both untreated and fibers treated with ClO2 (þFA and
Macromol. Biosci. 2007, 7, 1121–1131
ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
PFA) behaved similarly up to 350 h. The presence of PFA
(not sensible to cellulase) led to higher contents of retained
mass at the final steps of biodegradation, i.e. >350 h, as
compared to the untreated fibers. In the considered time
interval, the fibers treated with K2Cr2O7 þ PFA were more
resistant to the cellulase tests. These results suggest that
the layer of PFA was thick enough to impede the diffusion
of the enzyme to the polysaccharide matrix. The variation
in mass observed for this fiber could be related to water
adsorption-desorption processes.
The untreated and treated fibers were then used to
reinforce thermoset phenolic matrix composites.
Characterization of Phenolic Thermoset Composites
Impact Strength
The mechanical properties of a composite, as for instance
impact strength, are influenced by the properties of both
the matrix and the fibers, in addition to interfacial
properties between fiber and matrix. Numerous micro-
failure mechanisms can occur during impact loading,[12]
but in the present work, the loss of mechanical properties
of the fibers due to their chemical treatments (Table 3)
seems to be an important factor affecting the impact
strength of the corresponding composites (Table 4). During
the impact, the matrix seemed to transfer the load to a
poorer reinforcing agent when modified fibers were used.
The action of lignocellulosic fibers in composites is
strongly influenced by the preservation of the cellulose
matrix, when the fibers have previously been submitted to
treatments. The impact strength results of composites
confirmed that the reaction with K2Cr2O7 affected a con-
siderable fraction of the cellulose domain, as previously
mentioned. The impact strength value of the composite
diminished near tenfold, after the fibers had been oxidized
Table 4. Izod impact strength of unmodified and chemically
modified sisal fibers reinforced phenolic matrices composites
(unnotched samples).
Sisal fibers reinforcing Impact
composites Strength
J  mS1
Unmodified 443 W 20
Oxidized with ClO2 and 150 W 20
modified with FA
Oxidized with ClO2 and 155 W 12
modified with PFA
Oxidized with K2Cr2O7 and 42 W 5
modified with PFA
www.mbs-journal.de 1127
 J. D. Megiatto Jr., F. B. Oliveira, D. S. Rosa, C. Gardrat, A. Castellan, E. Frollini

Figure 2. SEM micrograph of fractured surfaces of composites reinforced with chemically modified sisal fibers. (a) Composite 15% (w/w)
sisal þ ClO2 þ AF (2 500) The arrows indicate the excellent adhesion fiber/matrix at interface and the partial coating of fibers for PFA
(visualized mainly as granules) that was grafted into fiber surface trough FA in situ polymerization. (b) Composite 15% (w/w) sisal þ
ClO2 þ AF (1 000) The arrows indicate the diffusion of resin in part of the internal fiber ducts. (c) Composite 15% (w/w) sisal þ ClO2 þ PFA
(1 000) The arrows indicate the excellent adhesion fiber/matrix at interface. (d) Composite 15% (w/w) sisal þ ClO2 þ PFA (600) The
arrows indicate the homogeneous coating of fiber by PFA (polymerized previously to be grafted on fiber surface). (e) Composite 15% (w/w)
sisal þ K2Cr2O7 þ PFA (1 300) The arrows indicate that the resin practically had diffused for all internal fiber ducts, evidencing that the
chemical modification lead also to more separated fiber bundles. (f) Composite 15% (w/w) sisal þ K2Cr2O7 þ PFA (800) The arrows indicate
the excellent adhesion fiber/matrix at interface.

Macromol. Biosci. 2007, 7, 1121–1131

1128 ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim DOI: 10.1002/mabi.200700083
Renewable Resources as Reinforcement of Polymeric Matrices: . . .

with this reagent and treated with PFA (Table 4). The
composites reinforced with fibers that had been submitted
to other chemical treatments also showed lower impact
strengths than those reinforced with unmodified fibers.
However, the reduction was almost three times less than
the one noted after K2Cr2O7 oxidation. Nonetheless,
considering that the impact strength of the phenolic
thermoset is near 12 J  m1,[15] it can be inferred that for all
composites the presence of unmodified or modified sisal
fibers increased the impact strength, that is, the necessary
energy to break the samples was higher for the composites
than for the thermoset, confirming that the lignocellulosic
materials act as a reinforcement of phenolic matrices.

SEM
Figure 2 shows scanning electron micrographs of impact-
fractured samples of composites reinforced with treated
fibers. As can be seen in that figure, some fibers were
pulled out due to the impact (see the arrows, Figure 2)
indicating that there were regions with poor adhesion at
the interface. Although the chemical modifications of the
fibers deteriorated their mechanical properties (Table 3)
resulting in the decrease of the impact strength of the
composites (Table 4), the interactions between fiber and
matrix at the interface were considerably improved as a
consequence of those treatments (Figure 2). During the
steps of impregnation of the fiber with the pre-polymer
and the initial cure, there occurred an interdiffusion of the
pre-polymer (a viscous liquid) on the fiber bundles. This
interdiffusion depends on the correlation between the
cohesive energy of the pre-polymer and the surface energy
of the fiber, among other factors. The increase in the
non-polar character of the fiber surfaces after the chemical
treatment, as indicated by IGC data, probably favored the
interactions with the least polar moieties of the pre-
polymers, and later in the matrix, at the interface of the
obtained composites. Furthermore, the introduction of
hydroxyl groups on the fiber surface and at the end of the
PFA chains as well as on the basic sites of the furan rings
probably intensified the hydrogen bond interactions
between the phenolic polymer and the sisal fibers. The
comparison of Figure (2c) and (2d) with (2a) and (2b)
showed that the fiber was more homogeneously coated
with a polymeric layer when PFA had been pre-poly-
merized and grafted to the fiber surface than when it was
polymerized in situ
Figure (2a)–(2f) proves that the chemical treatment
of fibers facilitated the diffusion of the pre-polymer in
the lignocellulosic fiber network, thus intensifying fiber/
matrix interactions in the composite. This effect was more Figure 3. Water absorption as a function of time for composites
reinforced with sisal fibers (at different temperatures): —&—
significant for fibers previously oxidized with K2Cr2O7
unmodified, —~— ClO2 oxidized þAF, —$— ClO2 oxidized
(Figure (2e), (2f)) than with ClO2 (Figure (2c), (2d)) indicat- þPFA and —^— K2Cr2O7 oxidized þPFA.
ing once again that K2Cr2O7 was able to access different
domains of fibers. Yet, the correlation between the polar

Macromol. Biosci. 2007, 7, 1121–1131

ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.mbs-journal.de 1129

Table 5. Water diffusion coefficient D for composites reinforced with unmodified and chemically modified sisal fibers.
T D
-C
unmodified ClO2 R FA modified
fibers fibers
25 2.8 3.0
55 3.5 4.9
72 5.9 12.0
and non-polar characteristics of both polymer and fiber, as
well as the effectiveness of the resin in accessing the inner
parts of the fiber bundles in addition to the fiber surface,
was also influenced by the roughness of the surface and by
the separation of the fiber bundles. This separation could
in turn be improved by the action of solvents and reagents
used during the chemical modification considered in the
present work.
Water Absorption
The composites were also characterized with respect to
water absorption (Figure 3) at three temperatures. It
should be pointed out that the test conditions were very
aggressive (see the Experimental Part) and that they did
not reproduce the end use conditions of this kind of
material. The aggressive conditions were, however,
deliberately chosen in order to subject the material to
the worst possible conditions. The diffusion of water in the
bulk of the composite is strongly influenced by the speed
at which hydrogen bonds are broken and re-established,
and examples of such bonds included H2O/H2O, H2O/
polymeric matrix, H2O/fibers. Moreover, the water molec-
ules were considered to diffuse through the matrix, the
fibers and the interface.
The experimental values were well fitted using Equa-
tion (6).[28,29]
log Mt =M1 ¼ logðkÞ  n logðtÞ
where Mt and M1 correspond to the water content at time
‘‘t’’ and at equilibrium, respectively. The linear part of the
curve that was obtained when log Mt/M1 was plotted as a
function of time enabled the determination of the linear
and angular coefficients ‘‘k’’ and ‘‘n’’, respectively.
For all experiments the value of ‘‘n’’ was found to be
equal to about 0.5. At such conditions, it was considered
that the diffusion of water through the composites
followed Fick’s law.[28,29]
Macromol. Biosci. 2007, 7, 1121–1131
1130 ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
J. D. Megiatto Jr., F. B. Oliveira, D. S. Rosa, C. Gardrat, A. Castellan, E. Frollini
103  mm2  sS1
ClO2 R PFA modified K2Cr2O7 R PFA modified
fibers fibers
2.5 2.3
4.5 4.0
15.6 14.6
The value of ‘‘k’’ gives information on the affinity
between the material and water molecules. The higher the
value of ‘‘k’’, the stronger is the affinity. The ‘‘k’’ values for
the composites reinforced with unmodified fibers were 0.7
for all considered temperatures, whereas those for the
composites reinforced with modified fibers were in the
range 0.1–0.3, regardless of temperature and fiber treat-
ment. These ‘‘k’’ values indicated that the chemical
modifications of fibers gave rise to less hygroscopic
composites. The fiber modifications conduced to a larger
diffusion of resin inside the fiber bundles, resulting in
fibers that were partially filled by the phenolic polymer in
the composites and also in more intense fiber/matrix
interactions at the interface. This, in turn, rendered the
polar groups of the fibers and the thermoset polymer less
available for interactions with water. Moreover, as was
shown by IGC data (Table 2), the non-polar character of the
fibers increased with the chemical treatment, thus
contributing to a decrease in the affinity for water of
the whole material, as compared to for the composite
made with unmodified fibers.
It is possible to calculate the diffusion coefficient (D) of
water in the composites using Equation (7).
 
Mt 4 D 0;5 0;5
¼ t (7)
M1 L p
(6)
Here L corresponds to the thickness of the composite.
Table 5 shows the values obtained for the water diffusion
coefficient of composites reinforced with modified and
unmodified sisal fibers.
As expected, the D increased with temperature (Table 5).
It was also found to be higher for composites reinforced
with chemically modified sisal fibers when the tem-
perature increased (55 and 72 8C). However, for all studied
temperatures, they absorbed smaller quantities of water
than those reinforced with unmodified fibers (Figure 3).
The amount of water absorbed by the composites also
DOI: 10.1002/mabi.200700083
Renewable Resources as Reinforcement of Polymeric Matrices: . . .
depended on the quantity of microcavities present, which
in turn were affected by interfacial fiber/matrix interac-
tions. The higher the adhesion at the interface, the lower
was the quantity of microcavities available for storing
water molecules. Furthermore, if the internal ducts of the
fibers were more filled with resin, the possibility of storing
water molecules decreased even further. As was already
mentioned, the chemical modifications of the fibers
improved the interactions between fiber and matrix at
the interface and also increased the accessibility of the
resin to the inner parts of fibers, thus reducing the water
absorbing capacity of the composite. However, these same
factors could render the diffusion of water through the
interface and internal ducts of the fibers more difficult
leading to the water probably diffusing mainly through
the bulk matrix. The phenolic polymer had fewer groups
that could interact with water via hydrogen bonds than
the fibers, and this could likely explain the increased dif-
fusion coefficient mainly observed for higher temperatures
(Table 5), since the development of such interactions
decelerated the diffusion of water through the material.
Conclusion
The chemical modification intensified the interaction
between the fiber and the phenolic matrix at the interface
for sisal-reinforced composites, but also caused some fiber
degradation that affected the mechanical properties and
decreased the impact strength of the corresponding
composites. In contrast, the chemical modification of
fibers considerably decreased the water absorption in
composites reinforced with modified fibers. This is an
important property for certain composite applications.
Moreover, new important data concerning composites
reinforced with sisal fibers were obtained in relation to the
detailed characterizations of the unmodified and chemi-
cally modified fibers, shown in the companion paper.[17]
This data will certainly contribute to reaching the target
that is under constant consideration in this research area
i.e. submitting fibers to modifications that improve both
the fiber/matrix interface and the composites properties.
Acknowledgements: The authors gratefully acknowledge CAPES/
COFECUB (Project 422/03/05) for enabling traveling missions
between Brazil and France and CAPES for the doctoral fellowship
of F.B.O. E. F. is grateful to CNPq (National Research Council, Brazil)
for research productivity fellowship and financial support and to
FAPESP (The State of São Paulo Research Foundation, Brazil) for the
doctoral fellowship for J.D.M. as well as for financial support.
Received: March 23, 2007; Revised: June 1, 2007; Accepted: June 4,
2007; DOI: 10.1002/mabi.200700083
Macromol. Biosci. 2007, 7, 1121–1131
ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Keywords: composite water absorption; sisal fiber, biodegrada-
tion; sisal fiber, chemical modifications; sisal fiber, inverse gas
chromatography; sisal fiber-reinforced composites
[1] V. A. Alvarez, A. N. Fraga, A. Vasquez, J. Appl. Polym. Sci. 2004,
4007.
[2] M. Brahmakaumar, C. Pavithran, R. M. Pillai, Composites Sci.
Technol. 2005, 65, 563.
[3] A. S. Luyt, M. E. Malunka, Thermochim. Acta 2005, 426, 101.
[4] A. P. Mathew, K. Oksman, M. J. Sain, Appl. Polym. Sci. 2005,
97, 2014.
[5] S. Mishra, A. K. Mohanty, L. T. Drzal, M. Misra, G. A.
Hinrichsen, Macromol. Mater. Eng. 2004, 289, 955.
[6] S. M. Sapuan, A. Leenie, M. Harimi, Y. K. Beng, Mater. Design
2006, 27, 689.
[7] C. Canevali, M. Orlandi, L. Zoia, R. Scotti, E. Tolppa, J. Sipila,
F. Agnoli, F. Morazzoni, Biomacromolecules 2005, 6, 1592.
[8] S. T. Georgopoulos, P. A. Tarantili, E. Avgerinos, A. G.
Andreopoulos, E. G. Koukios, Polym. Degrad. Stab. 2005, 90,
303.
[9] W. Hoareau, W. G. Trindade, B. Siegmund, A. Castellan, E.
Frollini, Polym. Degrad. Stab. 2004, 86, 567.
[10] W. G. Trindade, W. Hoareau, I. A. T. Razera, R. Ruggiero, E.
Frollini, A. Castellan, Macromol. Mater. Eng. 2004, 289, 728.
[11] W. G. Trindade, W. Hoareau, J. D. Megiatto, I. A. T. Razera, A.
Castellan, E. Frollini, Biomacromolecules 2005, 6, 2485.
[12] W. Hoareau, F. B. Oliveira, S. Grelier, B. Siegmund, E. Frollini,
A. Castellan, Macromol. Mater. Eng. 2006, 291, 829.
[13] D. Fengel, G. Wegener, ‘‘Wood: Chemistry Ultrastructure
Reactions’’, Walter de Gruyter, Berlin 1989, p. 613.
[14] E. Frollini, J. M. F. Paiva, W. G. Trindade, I. A. T. Razera, S. P.
Tita, ‘‘Natural Fibers, Polymers and Composites - Recent
Advances’’, F. Wallenberger, N. Weston, Eds., Kluwer Aca-
demic Publishers, Norwell USA 2004, p. 193 ff.
[15] I. A. T. Razera, E. Frollini, J. Appl. Polym. Sci. 2004, 91, 1077.
[16] J. M. F. Paiva, W. G. Trindade, E. Frollini, L. C. Pardini, Polym.
Plast. Technol. Eng. 2004, 43, 1179.
[17] J. D. Megiatto, Jr., W. Hoareau, C. Gardrat, A. Castellan, E.
Frollini, submitted.
[18] C. Nguyen, E. Zavarin, J. Appl. Polym. Sci. 1992, 46, 1103.
[19] J. Schultz, L. Lavielle, ACS Symp. Ser. 1989, 391, 185.
[20] M. N. Belgacem, A. Gandini, ‘‘Interfacial Phenomena in
Chromatography’’, Marcel Dekker, New York 1999, p. 59.
[21] M. A. Tshabalala, J. Appl. Polym. Sci. 1997, 65, 1013.
[22] M. N. Belgacem, G. Czeremuszkin, S. Sapieha, Cellulose 1995,
2, 145.
[23] A. Van Asten, N. Van Veenendaal, S. J. Koster, J. Chro-
matogr. A 2000, 888, 175.
[24] F. M. Fowkes, J. Phys. Chem. 1962, 66, 382.
[25] M. C. Gutierrez, F. Rubio, J. L. Oteo, J. Chromatogr. A 1999,
845, 53.
[26] J. M. F. Paiva, E. Frollini, Macromol. Mater. Eng. 2006, 291,
405.
[27] J. M. F. Paiva, E. Frollini, J. Appl. Polym. Sci. 2002, 83,
880.
[28] A. Espert, F. Vilaplana, S. Karlsson, Composites, Part A. 2004,
35, 1267.
[29] L. A. Pothan, S. Thomas, G. Groeninckx, Composites, Part A
2006, 37, 1260.
www.mbs-journal.de 1131