Particleboards From Peanut-Shell Flour: Leonel Batalla, Adria N J. Nun Ez, Norma E. Marcovich
Particleboards From Peanut-Shell Flour: Leonel Batalla, Adria N J. Nun Ez, Norma E. Marcovich
Particleboards From Peanut-Shell Flour: Leonel Batalla, Adria N J. Nun Ez, Norma E. Marcovich
ABSTRACT: Peanuts have been cultivated worldwide for mined and used in an attempt to predict the mechanical
hundreds, if not thousands, of years. However, most pea- response of the panels. The stiffness of the particleboard
nuts are sold without the shell, and so large quantities of could be greatly improved by a reduction of the porosity,
peanut shells remain as byproducts in the field, not being but even a low void content had a critical effect on the
used properly. In this work, the feasibility of making parti- strength of the composites. The stiffness of the composites
cleboard from milled peanut shells was studied. To obtain could only be reasonably represented with simple theoreti-
medium-density panels, a low compaction pressure and a cal models when the effect of the porosity was incorporated.
high fiber content were used. The physical and mechanical © 2005 Wiley Periodicals, Inc. J Appl Polym Sci 97: 916 –923, 2005
properties of the panels were similar to those reported for
wood-based particleboard when 80 wt % filler was used. Key words: composites; fillers; mechanical properties; ther-
The void content of the panels was experimentally deter- mosets
ing the curing of the polymer can lead to voids in the matrix is proposed. Under this assumption, when a
final material. The most liquid and castable resol con- uniform stress is applied at the boundary, the elastic
tains methylene and ether bonds, which have reactive modulus of the composite is given by
methyl end groups and thus can be cured via acid or
the application of heat. In practical applications, the
most frequent method for conversion is heat-activated
curing. At higher temperatures, the ether bonds be-
Ec ⫽ Em 冋 1 ⫹ 共E f /E m ⫺ 1兲V f 2/3
1 ⫹ 共E f /E m ⫺ 1兲共V f 2/3 ⫺ V f 冊 (3)
冋 册
relatively short branched connections via methylene
bonds in phenolic resins result in brittle mechanical Vf
Ec ⫽ Em 1 ⫹ (4)
properties that can be slightly improved by the addi- 共E f /E m 兲/共E f /E m ⫺ 1兲 ⫺ V f 1/3
tion of fillers.11
Peanuts have been cultivated worldwide for hun- Another interesting approach is the simple model pro-
dreds, if not thousands, of years. However, most pea- posed by Counto,14 which assumes perfect bonding
nuts are sold without the shell, and so large quantities between the particles and the matrix. In this case, the
of peanut shells remain as byproducts in the field, not modulus of the composite is given by
being used properly.
The aim of this work is to study the feasibility of 1 1 ⫺ V f 1/2 1
making experimental particleboards from milled pea- ⫽ ⫹ (5)
Ec Em 共1 ⫺ V f 1/2 兲/V f 1/2 E m ⫹ E f
nut shells. To obtain medium-density panels, a low
compaction pressure and high fiber content were Even though these models have proved to be a good
used. The particleboards were physically and mechan- approximation to composite properties, all of them
ically characterized. The void content of the panels assume that the composite porosity content is equal to
was experimentally determined and used in an at- zero.14 Besides the effect of lowering the composite
tempt to predict the mechanical response of the pan- load-bearing volume, the porosity also affects the
els. composite mechanical properties by introducing stress
concentrations into the material.14 Therefore, as these
effect were not included in the derivation of eqs. (1)–
REINFORCEMENT THEORIES FOR
PARTICULATE POLYMER COMPOSITES (5), they will not be able to represent the stiffness of a
particleboard.
In conventional unidirectional and continuous com- One alternative to including the effect of porosity in
posites, the traditional way of predicting composite calculating the stiffness of a composite material con-
stiffness is to apply the simple rule of mixtures.14 For sists of attributing all the voids to the matrix phase
a random distribution of the individual phases (par- through the calculation of an effective matrix Young’s
ticulate composite), the model is also useful, simplify- modulus and then using it instead of the true matrix
ing the arrangement to a two-phase model in which modulus in conventional models. This approach was
average stresses and strains are considered to exist in successfully applied previously.16
each of the phases.15 The upper and lower bounds for Alternatively, Madsen and Lilholt14 proposed that
this model are given by eqs. (1) and (2), respectively: the effect of porosity on material stiffness can be ap-
proximated by
E c ⫽ V f E f ⫹ 共1 ⫺ V f 兲E m (1)
E v ⫽ E d 共1 ⫺ V v 兲 2 (6)
E fE m
Ec ⫽ (2)
共1 ⫺ V f 兲E f ⫹ V f E m where the subscripts d and v denote the fully dense
material and the porous material, respectively.
where E is the elastic modulus and V is the volumetric On the other hand, the effect of porosity on the
fraction. The subscripts c, f, and m indicate the com- mechanical properties has been extensively studied
posite, fiber, and matrix, respectively. for ceramic materials. Ceramics have inherently ran-
Improved bounds for the modulus of two-phase dom microstructures, and the pores in these materials
media were obtained by Paul and by Ishai and Co- have random shapes, sizes, and distributions;17 this is
hen.15 In the approximate solution obtained by Paul, the same behavior expected in particleboards. A rela-
the constituents are assumed to be in a state of mac- tion of the form
roscopically homogeneous stress. Perfect adhesion at
the interface of a cubic inclusion embedded in a cubic E v ⫽ E d 共1 ⫺ V v 兲 n (7)
918 BATALLA, NUÑEZ, AND MARCOVICH
c ⫽ m 冉 1 ⫺ Vf
a ⫹ bV f ⫹ cV f 2 冊 (9)
following thermal schedule: 30 min at 90°C, 60 min at
140°C, and 10 min at 160°C.
High-density particleboards were prepared for
comparison. In this case, resol and peanut shells were
EXPERIMENTAL mixed in a Brabender-type mixer (volume capacity
⫽ 250 cm3) for about 15 min. The paste was filled in
Materials
the metal mold, and the reaction was carried out un-
A commercial alcoholic solution, liquid resol (R472,
Atanor, Buenos Aires, Argentina), with 60.6 wt %
solids, a viscosity of 230 cp at 19°C, a pH of 8.0 – 8.6 at
25°C, and a water concentration of 12–20 wt %, was
used as the matrix. The resol was produced by the
condensation reaction of phenol with formaldehyde in
a basic medium. The first step of the reaction was the
formation of addition compounds known as methylol
derivatives (the reaction taking place at the ortho or
para position). In the presence of alkaline catalysts and
with more formaldehyde, the methylol phenols con-
densed either through methylene linkages or through
ether linkages. In the latter case, the subsequent loss of
formaldehyde could occur with methylene bridge for-
mation. This type of product was soluble and fusible
but contained alcohol groups. When the reactions
leading to their formation were carried out further,
large numbers of phenolic nuclei could condense to
form networks.
The mechanical and physical properties of the pure
resol, shown in Table I, were obtained from the liter- Figure 1 Particle size distribution of peanut-shell flour.
PARTICLEBOARDS FROM PEANUT-SHELL FLOUR 919
Methods
The theoretical value for the volume fraction of the
filler, Vft, was calculated as follows:
V ft ⫽ 共W f / f 兲关共W f / f 兲 ⫹ 共W m / m 兲兴 (10)
TABLE II
Thermal Degradation Data of Peanut-Shell Flour and
Composites at 10°C/min in Air
Peanut Desorbed
shell (%) water (%) Td (°C) Tb (°C) Char (%)
TABLE III
Physical Properties of Peanut-Shell Composites
Fiber content Void content Water absorption Thickness swelling
(wt %) Density (kg/m3) (vol %) (wt %) (%)
TABLE IV
Mechanical Properties of Peanut-Shell Medium-Density Particleboards
Fiber content Tensile modulus Tensile maximum Flexural modulus Flexural maximum
(wt %) (GPa) stress (MPa) (GPa) stress (MPa)
ple, or in other words, water diffusion is more difficult exhibit much higher moduli than the corresponding
as the density of the particleboard increases.6 The medium-density ones, and this indicates that the stiff-
measured values are comparable to those of other ness of the composites could be greatly improved by
particleboards made from vegetable reinforcements.5 the reduction of the void content.
On the other hand, the thickness swelling is very low,
compared with moisture sorption, because it is more
affected by the bond quality25 and adhesive proper-
Modeling the particleboard mechanical behavior
ties, and as discussed previously, the compatibility
between the filler and the matrix is excellent in this Because of a lack of information about peanut-shell
case. stiffness, the range of 4.9 –14 GPa, suggested by But-
Table IV shows the mechanical properties of par- trey26 for the modulus of saw dust, was adopted.
ticleboards determined in tensile and flexural tests. Therefore, in the following calculations, a value of 10
Both the moduli and strengths decrease as the filler GPa was taken as Young’s modulus of the peanut-
content increases because the void content also in- shell flour. However, the effect of the filler modulus
creases. The porosity reduces the load-bearing volume on the composite predictions will be discussed later.
of the sample and introduces stress concentrations, Equations (1)–(5) were used to model the experi-
which make the material less stiff and resistant. How- mental modulus as a function of the filler volume
ever, even the tensile strength of the 80 wt % particle- fraction. First, all the selected theoretical models pre-
board is in the range of those found in wood particle- dict a continuous increase in the composite modulus
boards (ⱖ0.35 MPa, ref. 7). with the filler concentration if Young’s modulus of the
Selected properties of high-density composites from fiber is higher than that of the matrix, yet the opposite
resol and peanut-shell flour prepared for comparison behavior was experimentally observed. Thus, to re-
are reported in Table V. First, it was not possible to verse this tendency, two different approaches were
obtain a composite with no voids because the evapo- used.
ration of water and solvents from the resol during Figure 4 shows the experimental tensile and flexural
curing led to porous systems, although pressure was moduli together with the fitting curves obtained with
applied. Both the modulus and strength increase with the effective matrix modulus as a function of the ac-
the filler content because the porosity of these com- tual fiber volume fraction. In all cases, eqs. (1)–(5)
posites is relatively low, and thus the reduction of the were applied two times: first to calculate an effective
load-bearing volume is not as important as it is for matrix modulus with the void volume fraction as the
medium-density particleboards. Moreover, the tensile filler concentration with the modulus equal to zero
strength of the high-density composites is in the same and second to calculate the composite modulus with
range as that of the medium-density one, and this the effective matrix modulus as the modulus of the
indicates that even a low void concentration has a matrix. An exception was made with eq. (2), for which
critical effect on the strength of the composites be- the effective matrix modulus was calculated with the
cause of the stress concentrations introduced by po- upper bond of the rule of mixtures [eq. (1)]. The use of
rosity. On the other hand, high-density particleboards a matrix modulus corrected by porosity is not suffi-
TABLE V
Physical and Mechanical Properties of Peanut-Shell High-Density Particleboards
Fiber Density (kg/m3)
content Tensile modulus Tensile maximum Void content
(wt %) (GPa) stress (MPa)) Experimental Theoretical (vol %)
Figure 4 Experimental (F) tensile and (■) flexural moduli Figure 6 Fitting curves with the lower bound corrected by
and fitting curves (lines) obtained with the effective matrix eq. (7) as a function of Ef: (– – –) 14, (- - -) 10, and (—) 4.9 GPa
modulus approach as a function of the actual fiber volume and (F) experimental tensile modulus.
fraction (Vf): (– 䡠 –) upper bound, (– – –) lower bound, (- 䡠 -)
Ishai–Cohen, ( 䡠 䡠 䡠 ) Paul, and (–) Counto models.
successfully represent them. The lower bond of the
rule of mixtures is, as expected, the model that gives
cient to obtain good agreement, especially at high filler
the lowest values and thus the best fit.
(high void) contents.
The inability of both approaches to accurately
In the second method, the correction derived by
model the experimental data, especially at higher filler
MacKenzie27 and successfully applied by Madsen and
contents, can be attributed to different facts that were
Lilholt14 in a recent publication was used. Figure 5
not considered in the development of the equations. In
shows the predictions of eqs. (1)–(5), corrected by eq.
practice, the particle distribution may be less homo-
(6), as well as the experimental values as a function of
geneous as the particles may not be completely sepa-
the actual fiber volume fraction. The fittings get closer
rated from one another (e.g., aggregates of smaller
to the experimental data, although none of the models
particles may be present). Thus, the applied stress will
then not be distributed evenly between the particles
and the aggregates, and the assumption of either
isostress or isostrain will not be valid.28 In addition,
the amount of the binder would be not enough to
bound the filler particles together, and hence only part
of the filler would act as a reinforcement.
The correction proposed in eq. (7), which was
proved to be useful for calculating the modulus of
porous ceramic materials, was also used. The solid
line in Figure 5 represents the fitting calculated with
the lower bond of the rule of mixture corrected by eq.
(7). The constant n must be equal to 4 to obtain the best
agreement with the experimental data. This value was
in the range of those found for porous ceramics17 and
represents the stress concentration factor due to the
presence of pores in the material. A factor higher than
2 is necessary to obtain a reasonable agreement with
experimental results.
Finally, Figure 6 shows the predictions obtained
Figure 5 Experimental (F) tensile and (■) flexural moduli with eq. (2) corrected by eq. (7) when different values
and fitting curves (lines) obtained with the void correction are used as the peanut-shell Young’s modulus. Al-
approach as a function of the actual fiber volume fraction though the upper limit (14 GPa) represents properly
(Vf). The following models were corrected with eq. (6): (– 䡠 –)
upper bound, (– – –) lower bound, (- 䡠 -) Ishai–Cohen, the data at lower filler volume fractions, the lower
(䡠 䡠 䡠) Paul, and (–) Counto. (—) The lower-bound model was limit (4.9 GPa) prediction is closer to the experimental
corrected with eq. (7). points for higher filler concentrations. However, the
PARTICLEBOARDS FROM PEANUT-SHELL FLOUR 923
References