Manufacturing

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Composites Science and Technology 186 (2020) 107920

Contents lists available at ScienceDirect

Composites Science and Technology


journal homepage: http://www.elsevier.com/locate/compscitech

A proposal to modify the Kelly-Tyson equation to calculate the interfacial


shear strength (IFSS) of composites with low aspect ratio fibers
Laura Aliotta, Andrea Lazzeri *
University of Pisa, Department of Civil and Industrial Engineering, Via Diotisalvi 2, 56122, Pisa, Italy

A R T I C L E I N F O A B S T R A C T

Keywords: The mechanical behavior of short-fiber composites generally differs from long-fiber composites. The mechanical
Adhesion response is influenced by the particular distribution of tensions and by the mechanism of load transfer from the
Mechanical Properties matrix to the fibers. The transfer load from the matrix to the fibers occurs through shear stresses at the surfaces of
Analytical modelling
the fibers (edges or ends effects). The end effects in long fiber composites, involving a small portion of fiber, are
Fiber/matrix bond
Fibers
negligible; nevertheless, for short and very short fibers composites these end effects cannot be neglected.
Interfacial adhesion between the reinforcement and the matrix plays an important role, in fact it influences
both physical and mechanical properties of the composites, but the experimental determination is often laborious
and analytical models are frequently used to evaluate the interfacial shear strength (IFSS).
In this paper a modification of the Kelly-Tyson model for the calculation of the interfacial stress for short
(aspect ratio < 20) and ultra-short fibers composites (aspect ratio < 10) has been proposed to take into account
the end effects that in the original model were not considered. Successively, the Bader and Bowyer model (that
derives from the Kelly-Tyson model) for the evaluation of the IFSS was also modified. A few examples of cal­
culations of the IFSS, using this modified Bader and Bowyer model, have been provided using published liter­
ature data. Furthermore, a mechanical characterization of flax fibers has been carried out and their adhesion to
poly(lactic) acid (PLA) matrix were evaluated for composites containing ultrashort fibers (aspect ratio < 10). The
IFSS value obtained was compared with that obtained from the single fiber fragmentation test (SFFT). It was
found that a very good estimation of IFSS can be done by using this analytical model that can be easily applied
with a limited number of experimental tests.

1. Introduction energy of the filler [12–17].


In the case of continuous fiber composites several experimental
The increasing demand for low cost polymeric composites has led to techniques were developed to characterize the interfacial properties
investigate new possibilities for developing composites materials with such as: single fiber pull-out test, single fiber fragmentation test,
short discontinuous fibers [1]. Short-fiber reinforced polymer (SFRP) microdebond test and fiber push-out test [18–22].
composites are attractive because of their ease of fabrication, economy, In the case of filled particles or very short fiber reinforced compos­
and superior mechanical properties compared to the unreinforced ma­ ites, the experimental calculation of the interfacial shear strength is
trix [2–4]. more complicated and there is not an easy extension of the above
The interface plays an important role in the physical and mechanical mentioned techniques. Different analytical expressions were obtained to
properties of the resultant composites. In fact, usually, a surface fiber derive, not only the IFSS values but also the Krenchel fiber orientation
treatment is often needed to improve the compatibility with the matrix factor (η0) [23,24]. In general, these analytical expressions are based on
[5–11]. In particular, the ability to transfer the stresses across the modifications of the Kelly-Tyson equation [19,25].
interface is often discussed in terms of ‘adhesion’ that in general is Bader and Bowyer (B–B) [26,27] studied the deformation behavior
related to the interfacial shear strength (IFSS or τ). This parameter is of short fiber reinforced composites in which a wide range of
correlated to a combination of different factors such as the interface non-oriented fiber lengths was present; they were able to determine
thickness, the interphase layer, the adhesion strength and the surface fiber/matrix bond strength in a glass/nylon 6.6 and glass polypropylene

* Corresponding author.
E-mail address: andrea.lazzeri@unipi.it (A. Lazzeri).

https://doi.org/10.1016/j.compscitech.2019.107920
Received 7 March 2018; Received in revised form 9 November 2019; Accepted 16 November 2019
Available online 20 November 2019
0266-3538/© 2019 Elsevier Ltd. All rights reserved.
L. Aliotta and A. Lazzeri Composites Science and Technology 186 (2020) 107920

Fig. 1. Fiber stress (a) and interfacial shear stress (b) varying the fiber length according to the classical Kelly-Tyson model and according to the “modified” version
(c) and (d) respectively.

composites. The model proposed by Bader and Bowyer, is an extension have strong effects on the mechanical properties of composites, such as:
of the original Kelly-Tyson equation; it derives the value of the inter­ the tensile strength of matrix and fibers, the aspect ratio, the fibers
facial shear strength, τ, from a combination of the stress-strain curve and orientation factor and the IFSS.
fiber length distribution. In short fiber composites, the load applied to the material is trans­
Thomason [5] improved these methods and illustrated the applica­ ferred to the fibers by the matrix; the load transfer takes place not only
bility to an injection molded glass-fiber-reinforced thermoplastic com­ through the cylindrical surface of the fiber but also through the ends of
posites. He extended the analysis, inserting the composite breaking the fiber itself. When the length of the fiber is much greater than the
stress and strain into the originally Kelly-Tyson equation and obtained length over which the stress transfer takes place, the end effects can be
the average fiber stress at composite failure (σuf). neglected; instead, in very short fiber composites the end effects influ­
Puka�nszky [28] developed a semi empirical equation to predict the ence the behavior and the mechanical response of the material [34].
ultimate tensile properties of different particulate filled and short fiber In most analytical models for the calculation of elastic stresses and
thermoplastic composites. The semi-empirical equation proposed by strains in short fibers composites, the shear lag theory, proposed for the
Puka �nszky describes the effect of filler volume fraction, ϕf, and interface first time by Cox, was often applied [35–38]. According to this theory no
interactions on yield stress and tensile strength of particulate-filled stress is transferred across the fiber ends. This assumption has been
polymers. In this equation, an interaction parameter, B, was intro­ shown not realistic since the fiber stress does not fall to zero at the ends
duced to consider the stress transfer capacity of the composite. The [39]. In particular, the predictions of composite modulus and strength,
Puka �nszky’s equation was successfully applied to anisotropic fillers, for calculated by Cox model, are not sufficiently accurate when the fiber
example in layered silicate nanoparticles, wood fibers, carbon nano­ aspect ratio is small [34,40]. The Cox model provides an underestima­
tubes, calcium carbonate, graphite fibers, calcium sulfate [29–33]. tion of the strength due to the neglect of stress transfer across the fiber
Recently Lazzeri and Phuong [12] established a direct link between ends [34,41–43].
Puka �nszky’s parameter B with other parameters which are known to The original Kelly-Tyson model, instead, considered continuous and

2
L. Aliotta and A. Lazzeri Composites Science and Technology 186 (2020) 107920

long discontinuous fibers (seen as a cylinders), embedded in a ductile


matrix, undergoing multiple breakages while still bearing substantial
fractions of the applied stress until the critical length is attained [19].
In this paper the attention is focused on short (aspect ratio < 20) and
ultra-short (aspect ratio < 10) fibers composites. The Kelly-Tyson
model, widely used to calculate the composite interfacial shear
strength (IFSS), was analyzed and modified taking into account the end
effects. Adding the contribution of fiber stress (not only at the middle of
the fibers but also at their ends) were introduced: a new expression of
critical length and a new formulation of the Kelly-Tyson model. Suc­
cessively, the Bader and Bowyer model (B–B model ) was applied with
the new equations found in order to obtain a simple IFSS evaluation. A
few examples of calculation of the IFSS, τ, with this modified model have
been provided and discussed using published literature data.
However, an evaluation of the IFSS by experimental tests is funda­
mental. The characterization of the interfacial region is very complex
and can be performed at different scales [44]. Among the multiple test
methods (such as: fragmentation, pull-out and microbond), with their
relative advantages and disadvantages largely discussed in the literature
[21,45,46], the single fiber fragmentation test (SFFT) is very popular
and simple to use [21,47]. Consequently, in order to validate this new
model version, a comparison between the IFSS calculated by SFFT and
the modified B–B model was investigated. The experimental validation
was made on a biocomposite because for this class of materials often
Fig. 2. Schematic representation of the stress at the fiber edges. a) Classical
ultra-short fibers are used. These types of fibers derive from industrial
Kelly-Tyson model in which the fiber tensile stress is maximum in the middle of
and/or agriculture wastes and they can be valorized. For this type of
the fiber and the edge effect are negligible b) “Modified” version, for very short
composites, the experimental evaluation of interfacial shear strength is fibers composites, in which the edge effects are not negligible.
complicate and laborious. Moreover, in the case of ultra-short fiber
composites, if the fibers come from wastes, for example, it is difficult to
have fibers sufficiently long to perform these tests and analytical models
are frequently used. This why it is important to have an analytical model
that is able to estimate with good accuracy the IFSS without making
excessive experimental tests.
Very short flax fibers were used and chopped to obtain a very short
aspect ratio (<10), before being added to a PLA matrix in different
amounts. A comparison between the IFSS calculated by single frag­
mentation test (SFFT) and the modified Bader and Bowyer model has
been made.

2. Theorethical analysis

2.1. Critical length review

The main hypotheses of the Kelly-Tyson model are: perfectly plastic


behavior of the matrix and the complete adhesion between the matrix
and the fiber; it follows the assumption of a constant shear stress along
the entire fiber length.
In the original model, the fiber tensile stress is maximum at the
middle of the fiber and at the fiber ends the stresses are negligible
because the fiber is supposed detached from the matrix following fiber Fig. 3. Fiber force balance.
breakage. In the case of composites having very short fibers, the length
of the fibers is just above the critical length or even below this threshold
Maintaining the classical assumptions of the Kelly-Tyson model, in
value. Therefore, when fiber breakage occurs the remaining fragments
this work a particular case of well bonded fibers having relatively low
of the fiber are not capable of bearing substantial fractions of the applied
aspect ratio and a matrix with plastic behavior were considered. In this
stress.
case, it was assumed that debonding does not occur at the fiber ends.
According to the classical Kelly-Tyson model the profile of the fiber
Moreover, since the fibers are very short, not only the fiber ends effects
tensile stress and interfacial shear stress, as a function of the fiber length,
are not negligible, but these effects give a constant stress contribution
is reported in Fig. 1. The equation for the critical length is reported
(σ0) that affects the entire fiber strength distribution. According to these
below:
assumptions, the maximum value that σ0 can reach is equal to the yield
σ f ;b D stress of the matrix (exceeded this value in fact, the matrix and the fibers
Lc ¼ (1)
2τ undergo to detachment); hence, in first approximation, σ0 ¼ σ y,m.
A schematization of the ends fiber stress effects, compared to the
where D is the fiber diameter and σ f,b is the fiber breakage stress. For a classical Kelly-Tyson theory is represented in Fig. 2. Consequently, the
fiber length equals to the critical length, the maximum stress in the profile of the fiber tensile stress was modified, introducing the
middle of the fiber can reach the fiber breakage stress. σ 0 contribution (Fig. 1 (c)).

3
L. Aliotta and A. Lazzeri Composites Science and Technology 186 (2020) 107920

Fig. 4. Schematic representation of theoretical composite strength with volume fraction at different fibers aspect ratio. Comparison between the classical Kelly-Tyson
model and the modified version.

Applying the classical fiber force balance (Fig. 3), the relationship � �
Lc
between the interfacial shear strength and the fiber tensile stress [48] σ f ðL > Lc Þ ¼ σf ;b ⋅ 1 (6)
was obtained: 2L
For L < Lc (Fig. 1 (c)):
π D2 � π D2
σf þ dσf ¼ σ f þ πτDdz (2) Z
4 4 1 τL
σ f ðL < Lc Þ ¼ σ f dL ¼ þ σ0 (7)
L D
where D is the fiber diameter, τ is the interfacial shear strength and σf is
the fiber tensile stress. In the Kelly-Tyson model, the boundary condi­ Also in this case, when the end stresses are negligible, Eq. (7) returns
tions to solve the differential equation were: σf ¼ 0 for z ¼ � L corre­ to the Kelly and Tyson solution.
sponding to a zero stress at the fiber ends; and σf ¼ σ f,max for z ¼ L/2
corresponding to a maximum fiber stress in its middle (coincident to the
fiber breakage stress, σ f,b). With the new assumption, the second 2.3. Composite tensile strength
boundary condition does not change while the first condition includes
the σ0 contribution and becomes: σ f ¼ σ 0 for z ¼ � L. Integrating and For a composite containing more than a minimum volume fraction of
solving Eq. (2) with the new boundary condition, a new definition of fibers, Vmin, the strength of a composite, according to Kelly-Tyson model
critical length was found (Eq. (4)). [12,49], is expressed by the following equation:

Z σf ;max Z
4τ Lc=2 σ c ¼ η0 Vf σ f ðL > Lc Þ þ σ’m 1 Vf (8)
dσ f ¼ dz (3)
D 0
where η0 is the fiber orientation factor; σ0 m is the stress borne by the
σ0

� matrix when the strain of the composite is such that the fibers are
σ f ;max σ0 D
Lc ¼ (4) strained to their ultimate tensile strain, εc. This value can be obtained

experimentally through the stress-strain curves or it can be assumed
approximately equal to Em*εc. The value of σf is the fiber tensile stress
2.2. Fiber tensile stress for L > Lc and L < Lc
before defined.
Refocusing the attention on Fig. 1(c), a new expression of the fiber On the other hand, for a composite containing a volume fraction
average tensile stress was obtained for the cases in which the fiber length below Vmin, the equation is:
is less or greater than the critical length. The new expressions obtained �
σ c ¼ η0 Vf σ f ðL < Lc Þ þ σm 1 Vf (9)
are reported below.
For the case L > Lc (Fig. 1 (c)): where σm is the stress at break of the matrix.
1
Z �
σf ;b σ0 Lc
� Introducing in Eq. (8) and Eq. (9) the new expressions of fiber tensile
σ f ðL > Lc Þ ¼ σf dL ¼ σ f ;b ⋅ 1 ⋅ (5) strength for L > LC and L < LC (Eq. (5) and Eq. (7) respectively) the
L σf ;b 2L
modified expressions of the Kelly-Tyson model have been obtained.
where: σf,max � σ f,b. It can be observed that this new expression for the Fixed the values of σ0 m, σ m, σ0, τ and η0, varying the fiber volume
fiber tensile stress, when σ 0 is negligible (as in the case of long fibers) fraction and the fiber aspect ratio, the composite tensile strength trends
returns to the classical Kelly-Tyson solution (Eq. (6)): have been plotted in Fig. 4. In particular, a comparison between the
classical Kelly-Tyson model and the new version was made.
First of all, it can be observed in Fig. 4 that the intersection point
between the equation referred to L > Lc and L < Lc provides the value of

4
L. Aliotta and A. Lazzeri Composites Science and Technology 186 (2020) 107920

Table 1
Values recalculated from original experimental data from literature.
Material Aspect ratio τaccording to B–B model (MPa) Τexperimental – Pull-out Test (MPa) τaccording to modified B–B model (MPa) Reference

PP/Hemp treated 17 9 6 7 [53]


PP/Hemp untreated 17 7 3 5 [53]
PP/Hemp þ 5 wt% NaOH 24 14 – 8 [54]
PP/Hemp þ 7.5 wt% NaOH 28 15 – 8 [54]
PP/Hemp þ 10 wt% NaOH 30 15 – 11 [54]
PEEK/Glass Fiber 12 80 – 66 [57]
PEEK/Carbon Fiber 14 202 – 179 [57]

Vmin that is the minimum filler volume fraction for which the fibers, (EfεcD/4Lτ)). The final composite stress at any level of strain is reported
having a determined aspect ratio, will have a reinforcing effect on the below:
composite [49]. For a composite having Vf < Vmin the composite will �X� � X� � �� �
τLi Vi Ef ε c D �
undergo fracture without any fiber fragmentation. σ c ¼ η0 þ Ef εc Vj 1 þ Em εc 1 Vf (12)
D 4Lj τ
Increasing the fibers aspect ratio, the reinforcing effect will be i j

greater; it is known in fact, that long fibers reinforce better than short
fibers. Furthermore, it can be observed that the trend differences be­ where i and j indexes are referred to those fibers having L< Lε and L> Lε
tween the classical model and the new version are very marked when the respectively.
aspect ratio is very low (<10). When the aspect ratio increases, the two According to Bader and Bowyer, τ and eventually also η0 can be
models tend to overlap confirming the correct negligibility of fiber end obtained using, at two strain values (ε1 and ε2), the composite stress (σ 1
stresses when the fibers become long. and σ2). The matrix contribution Z at these two strains can be calculated
Also the difference of the intersection point (Vmin) tends to be equal from a tensile test on the pure matrix material and it is used to evaluate
for fiber aspect ratio above 10; below 10 the Vmin difference is very the ratio R of the fiber contributions at the two strains:
marked and classical model requires more fibers to reach the Vmin value. σ1 Z1 ’ X1 þ Y1
This is due to the fact that the starting assumption of the classical model R¼ ; R ¼ (13)
σ2 Z2 X2 þ Y2
and the new version are different. In fact, in the new version, the fibers
Assuming a value of τ, Eq. (12) is used to calculate ratio R0 . The value
having a fiber length under the critical length are not completely de­
of τ is adjusted until R’¼R [5].
tached from the matrix and they are able to bear a part of the load
Substituting in the Eq. (12) the new expressions of the average fiber
applied.
stress (Eq. (5) and Eq. (7)) and using the new expression of critical length
When the fiber aspect ratio is very low (in Fig. 4 is reported the case
(Eq. (4)), a new analytical expression can be obtained:
in which ar ¼ 3) the fibers are too short, all of them are under the critical
length and there is not an intersection point (Vmin). In this case the �X� �
τLi
�� X� �
Ef ε c σ 0 D
�2 �� �
composite stress trends for L > Lc have no significance and only the cases σ c ¼ η0 Vi þ σ0 þ Ef εc Vj 1
D 4Ef εc Lj τ
for L < Lc are considered. It can be observed that the new model pro­ i

j

vides better results; it diverges only for high filler volume fraction that þ Em εc 1 Vf (14)
are impossible to reach in practice. The classical model diverges
Eq. (14) is still iterative, according to the procedure mentioned
immediately giving results that are not consistent with what is found in
above, and when R’¼R the value of IFSS is obtained.
literature. For composites with very low aspect ratio, the decrement/
This new modified model should provide a better IFSS evaluation in
increment of tensile composite stress is not so marked as the classical
the case of short and/or ultra-short fibers composites. It is also expected
model predicts. The results of the modified model seem to be better in
that, in the case of ultra-short fiber composites (ar < 10), the lengths
predicting decrement/increment of the tensile composite stress that
involved are so small that all the fibers are under the critical length; in
appears more realistic [30,50–52].
this case the model is simplified (the super critical fiber contribution
vanishes), leading to a direct and very simple expression of the IFSS (no
more iterative):
2.4. Reformulation of the Bader and Bowyer model
� � �
1 σc σ’m 1 Vf
At this point, with the new stress expressions along the fiber length τðL < Lc Þ ¼ σ0 (15)
ar η 0 Vf
and with the new critical length formulation, the analytical model of
Bader and Bowyer (B–B) [26] can also be modified. Bader and Bowyer, 3. Comparison with literature data
presented an iterative approach for deriving the interfacial shear stress
and the fiber orientation factor η0 (if is not experimentally known) [26, In the present section, the IFSS data for fiber composites reported in
27]. literature are compared to the results obtained with these new modifi­
The Kelly-Tyson model [19] can be simplified to: cations of the Kelly-Tyson and B–B models.
σ c ¼ η0 ðX þ YÞ þ Z (10) Li et al. [53] determined the interfacial shear strength of short,
treated and untreated, hemp fibers composites. Industrial hemp
where X is the sub-critical fiber contribution, Y is the super-critical fiber (Cannabis sativa L.) from Hamilton, New Zealand was used in this
contribution and Z is the matrix contribution. Bader and Bowyer investigation. The matrix used was polypropylene (PP) (Icorene® PP
assumed that at any value of composite strain εc there is a critical fiber CO14RM by Aldrich Chemical) and maleated polypropylene (MAPP)
length Lε that is defined as: (A-C 950P high molecular weight MAPP supplied by Honeywell Inter­
national Inc.). Single Fiber Pull-Out test and Bader and Bowyer model
Ef ε c D
Lε ¼ (11) were performed to determine IFSS. Starting from data reported by Li

et al. [53], the IFSS was calculated for PP composites, according to Eq.
Fibers that are shorter than Lε will carry an average stress, given by (14). The attention was focused on composites with PP matrix with a
the Kelly-Tyson expression, equal to τL/D; similarly, for the fibers that volume fraction of fiber equal to 28.8% because all the necessary data
are above the critical length the average fiber stress is equal to Efεc(1-

5
L. Aliotta and A. Lazzeri Composites Science and Technology 186 (2020) 107920

Table 2
Experimental data from literature used for the IFSS calculation according to the new approach.
Material PP/Hemp PP/Hemp PP/Hemp þ 5 wt% PP/Hemp þ7.5 wt% PP/Hemp þ10 wt% PEEK/Glass PEEK/Carbon
treated untreated NaOH NaOH NaOH Fiber Fiber

Reference [53] [53] [54] [54] [54] [57] [57]


Ef (GPa) 22 24 25.7 26.3 27.6 69 210
Df (μm) 29 30 23.7 24.5 24.6 15 7
Lf (μm) 512 515 569 689 719 175 100
Vf 0.288 0.288 0.301 0.301 0.301 0.179 0.231
ε1 (%) 1 1 0.8 0.8 0.8 0.5 0.5
ε2 (%) 2 2 1.6 1.6 1.6 1 1
σ1,c (MPa) 28.9 25.5 22.1 25.2 27.2 46 70
σ2,c (MPa) 44.5 34.7 37.6 41.9 44.7 82 140
σ0 (MPa) 22 22 18.8 18.8 18.8 91.4 91.4
η0 0.46 0.44 0.29 0.29 0.29 0.4 0.21

Fig. 5. IFSS difference and fibers aspect ratio for the literature cases examined.

for the IFSS calculation are provided in Ref. [53]. The yield stress of the The polymeric matrix used, with and without coupling agent (a
matrix (corresponding to σ0) is not reported and it has been taken from maleic-grafted polypropylene (MAPP)), was PP ISPLEN 090 G2M
the technical datasheet of the supplier whereas the matrix contribution, (Ressol-YPF, Tarragona, Spain). A treatment with NaOH (BASF, Tarra­
at the two strain values chosen, was calculated as Em * εc because the gona, Spain) to eliminate soluble lignin and extractives from the surface
stress-strain curve of the matrix was not provided. The IFSS values ob­ of the fibers was proposed. A prediction of IFSS, using B–B calculation
tained using Eq. (14) and according to literature [53], for treated and was made and it was found that the values were very near to the von
untreated hemp fibers are reported in Table 1; whereas, in Table 2 the Mises criteria, that is considered an upper bound for the theoretical
main data, extrapolated from literature, used for the IFSS calculation are value that τ can assume. Based on the data available (Table 2), PP/hemp
listed. It can be noticed that the IFSS values obtained with this new core composites with different weight percentages of NaOH were
approach, compared to the classic Bader and Bowyer method, are closer analyzed. The results, reported in Table 1 show, also in this case, that the
to the experimental pull-out test. IFSS values obtained with the new model are lower than the classic
Other sets of data suitable for this study were made available in the Bader and Bowyer model. However, these new data appear more real­
work made by del Rey et al. [54]. They studied the mechanical and istic because they are not too close to the maximum possible value given
interface properties of biocomposites with hemp core by-product fiber. by the Von Mises criterion and also taking into account the well-known

Table 3
Percentages of fibers having length shorter and longer of the critical length and critical fiber length value. Comparison between the classical model and the modified
version.
Material Lc according to the classic model Lc according to the new model L < Lc old L < Lc new L > Lc old L > Lc new Reference
(μm) (μm) (%) (%) (%) (%)

PP/Hemp treated 1063.3 1321.6 86.8 91.4 13.2 8.6 [53]


PP/Hemp untreated 1542.9 2094 88.4 93 11.6 7 [53]
PP/Hemp þ 5 wt% 399.5 671.3 88.8 97 11.2 3 [54]
NaOH
PP/Hemp þ 7.5 wt% 447.5 810.3 88.6 97 11.4 3 [54]
NaOH
PP/Hemp þ 10 wt% 478.9 632 94 97 6 3 [54]
NaOH
PEEK/Glass Fiber 155.3 177.8 39.6 47.9 60.4 52.1 [57]
PEEK/Carbon Fiber 89.4 100.9 15.9 26.3 84.1 73.7 [57]

6
L. Aliotta and A. Lazzeri Composites Science and Technology 186 (2020) 107920

(MFI): 6g/10min (210 � C, 2.16 kg), nominal average molar mass:


200,000 g/mol, density: 1.24 g/cm3, content of D-lactic acid units
about 3–6%].
� Reinforcement: raw flax fibers (kindly provided by Professor Luigi
Torre of Perugia University) were used in the form of long heckled
fibers connected together in a roving bundle. The appearance of the
fibers can be seen in Fig. 6 The long fibers were used essentially for
their mechanical characterization in order to obtain the necessary
parameters (Young modulus and stress at break); whereas, for the
composites production, the fibers were chopped with a cutter in
order to obtain fibers having aspect ratio around 8. The fiber density
was taken equal to 1.5 g/cm3 according to literature data [58].

4.1. Fiber tensile tests

Fig. 6. Bundle of raw flax fibers. Axial tensile tests on single fiber filaments, randomly chosen and
extracted from the bundle, were performed at room temperature ac­
cording to ASTM D3379-89. An Instron universal testing machine 5500R
bad adhesion that natural fibers have with the polymeric matrices [7,55,
equipped with a 10 N load cell and interfaced with a computer running
56]. In any case, the interfacial shear strength increases with the NaOH
MERLIN software (INSTRON version 4.42 S/N – 014733H) was used.
treatment demonstrating an improvement of the interface quality with
The support for mounting the fiber filament in the tensile testing ma­
the addition of the NaOH content.
chine was made of thin cardboard and a slot of length equal to the
Short glass (GF) and carbon fibers (CF) reinforced poly-ether-ketone
clamps distance, was cut out in the middle of this support. A single
(PEEK) composites were studied by Sarasua et al. [57]. In this work
filament was pasted at both the ends using a suitable adhesive.
commercial 30 wt% of short glass and carbon fiber reinforced PEEK
The axial Young’s modulus, the ultimate axial tensile strength of
composites were used (trade names: Victrex PEEK 450 GL30 and Victrex
single fiber and all related system parameters (system compliance Cs,
PEEK 450 CA 30). The reinforcing effect of fibers, evaluated in terms of
true compliance C) were calculated according to Ref. [59]. Along the
interfacial shear strength, using the Bader and Bowyer theory was
useful length of the fiber filament, different diameter measurements per
examined. On the basis of the data obtained from literature and reported
single fiber filament were carried out, using an optical microscope, and a
in Table 2, it was found, also in this case, that the new model reduces
mean value was taken for the calculation of the cross sectional area. For
significantly the τ values. However, the IFSS values confirm the good
simplicity it was made the assumption that the fibers have cylindrical
interface adhesion; in particular carbon fibers have a much higher
shape and therefore a circular section was considered.
adhesion to PEEK matrix than the glass fibers.
Natural fibers, due to defects and diameter variation along the entire
The IFSS difference is more pronounced when the fiber aspect ratio is
length, have a very wide scatter of mechanical properties that cannot be
low. In Fig. 5 the IFSS difference and the fiber aspect ratio for the
determined on a limited number of samples. Consequently, a statistical
literature cases examined are reported. It is confirmed that for fiber
study was carried out at 3 different clamp distances (10, 20 and 30 mm)
aspect ratio above 20 the difference with the classical Bader and Bowyer
in which at least 50 fiber filaments per distance were tested.
model (and consequently with the classical Kelly-Tyson model) are not
The Weibull analysis was proved to be suitable for several fibers and
relevant and thus not necessary. In fact, the new modified “model” is
also for flax fibers [60,61]. Owing to the varying severity of flaws along
tailored for short (aspect ratio < 20) and ultra-short (aspect ratio < 10)
the fibers, it was found that the fiber strength is statistically distributed.
fibers in which significant differences with the classical Bader and
The experimental fiber strength distribution, P(σf), under tension (the
Bowyer model can be encountered.
fibers exhibit little diameter variation in the gauge lengths examined)
In Table 3 the differences obtained, using literature values, for the
can be generally described by the standard Weibull distribution [62]:
critical length value and for the percentages of fiber having length
� � �� � �
shorter and longer of the critical length are reported. The new model, � l σf α
P σf ¼ 1 exp (16)
increasing the critical fiber length, decrease the percentage of fibers l0 β
having L > Lc; it can be observed that this decrement is significant, as
expected, for composites with ar < 20. Shorter are the fibers, more the where l is the fiber length, l0 is the reference length (chosen equal to 1
critical length increases and it will provide a greater decrease of the mm for mathematical convenience), α is the shape parameter and β is
percentages of fibers having L > Lc. the scale parameter. A modified Weibull distribution was proposed by
It will be expected that for composite having ar < 10, the composite Watson and Smith [63]; they introduced an empirical exponential
will have all the fibers below the critical length and Eq. (15) can be used parameter, γ, for improving the fitting with experimental data. How­
without the need of iterative calculations. ever, in literature it was found that the two-parameter Weibull distri­
bution provided good agreement with flax fiber strength data [64].
4. Validation with new experimental data Consequently, in this study, Eq. (16) was applied.
The cumulative probability of the fiber failure, P, was estimated as
In order to validate the model, experimental tests were performed in median ranks assigned to the measured strength values at each gauge
which a composite system made of PLA matrix and flax fibers was length, σ f,i, using the following expression:
analyzed. The results of the classical B–B model and the “modified” � i 0:3
version were compared with single fiber fragmentation test (SFTT) and P σf ;i ¼ (17)
n þ 0:4
also the final critical length was compared to the experimental data.
The materials used were: where n is the number of data points (in this case correspond to the
strength measurements performed at a given gauge distance) and i is the
� Matrix: poly(lactic acid), PLA2003D purchased from NatureWorks rank of the i-th number in the ascending ordered strength data point
(grade for thermoforming and extrusion processes) [melt flow index [65].

7
L. Aliotta and A. Lazzeri Composites Science and Technology 186 (2020) 107920

Table 4
Weibull distribution parameters of flax fibers strength (l0 ¼ 1 mm).
Loading Gauge Shape Scale Predicted Measureda
rate length parameter parameter β Strength Strength
(mm/ (mm) α (MPa) (MPa) (MPa)
min)

0.6 10 1.07 6989.17 794.47 734.79


0.6 20 1.03 9921 531.27 544.02
0.6 30 1.91 2782 414.54 410.75
a
average value of 40 results.

Fig. 8. Dependence of the flax fibers average strength on the test length.

Fig. 9. Flax fibers modulus as a function of fiber diameter.

The average strength was evaluated by the following equation:


� � α1 � �
l 1
< σf > ​ ¼ ​ β Γ 1þ (18)
l0 α

where Γ is the Gamma function evaluated at different gauge lengths.


Differently from the fiber stress at break, the elastic modulus (eval­
uated according to Ref. [59]) is independent from the fiber length.
Consequently, all data obtained for each fiber (tested at different gauge
distances) were combined in order to obtain an elastic modulus esti­
mation according to the following equation:
PN
Ei
Ef ¼ i (19)
N

where Ef is the average flax fibers elastic modulus, Ei is the value of the
elastic modulus of each single flax fiber and N is the total number of
tested fibers.
The values of Weibull distribution parameters and the Weibull plots
for tensile strength at different gauge lengths are reported in Table 4 and
Fig. 7. The comparison of the average strength values predicted by Eq.
(18) (at 10, 20 and 30 mm gauge lengths) are very close to the experi­
mentally average values confirming the validity of scaling model based
on standard Weibull analysis.
Plotting the values of the predicted average strength as function of
the gauge length into a log-log plot, a linear trend can be observed.
Interpolating these values with a straight line, it is possible to obtain the
“fibers working line” that predicts the strength at break of the flax fibers at
different lengths (Fig. 8). The average tensile strength is sensitive to
gauge lengths with a decrease in fracture strength with increasing gauge
length. This behavior is very common and it was reported in literature
for brittle fibers. This trend was attributed to the fact that increasing the
test length, there will be more flaws within the fibers and, consequently,
the probability of fiber failure will become larger [65,66].
Fig. 7. Weibull plots for tensile strength of flax fibers at different gauge The value of elastic modulus obtained from Eq. (19) is 36.94 GPa and
lengths: a) 10 mm; b) 20 mm; c) 30 mm. it falls within the range found in the literature for flax fibers [55,67].

8
L. Aliotta and A. Lazzeri Composites Science and Technology 186 (2020) 107920

Fig. 10. Tensile tester apparatus for single fiber fragmentation test (SFFT) (left side) and test set-up under microscope (right side).

Plotting the fiber Young’s modulus as a function of fiber diameter Therefore, the critical fragment length was obtained applying the
(Fig. 9) it seems that there is an inverse hyperbolic correlation between following equation [22]:
the elastic modulus and the fiber diameter; this trend was already found
4
for other natural fibers [68–70]. The thin fibers are more homogeneous lc ¼ l (20)
3
in fact the microfibrils, being more packed, contain fewer defects and
voids. As a consequence, the thinner fibers have a more resistant cross The average shear strength at the interface it was then estimated as:
section at which is associated a higher elastic modulus. Another reason
σf lcÞ D
for the inverse correlation could be in the random distribution of the τ¼ (21)
2lc
greater number of microfibrils. The probability is that in the thicker fi­
bers (having more microfibrils), the rupture will start earlier and this where σ f(lc) is the fiber strength at the critical length, D is the mean fiber
will results in a lower stiffness [68]. diameter.
The specimens for fragmentation tests were prepared from granules
4.2. Single fiber composite tensile tests of pure PLA2003D kept in a circulating air oven at 60 � C for at least 24 h
to avoid water uptake. Sheets of PLA matrix were produced placing
In order to investigate the adhesion between the fiber and the matrix, about 5 g of PLA 2003D between two Teflon sheets and then compres­
single fiber fragmentation test (SFFT) was carried out. In this test, a sion molded by two steel plates in a NOSELAB ATS manual laboratory
tensile load is applied on each single fiber embedded in a matrix spec­ heat press at 195 � C and 2.5–3 tons pressure for 6–8 min. To control the
imen. The specimens, having a dump-bell shape, were elongated in a thickness of the resulting sheets, a thin steel plate was placed between
specific custom made tensile tester (Fig. 10). This tester is able to pull the Teflon sheets. The thickness of the single PLA sheets produced was
slowly both the specimen ends applying a uniform tensile stress distri­ about 0.8 mm. Subsequently, a single polymer sheet was placed again on
bution in the specimen gauge section. the Teflon sheet and a single flax fiber (manually selected from the fibers
The experiment was carried out under a light microscope in order to bundle) was attached to the polymer sheet with an adhesive tape. At this
observe and to measure in situ the fragmentation process. The fiber point, a second polymer sheet was positioned on the top of the first one
inside the matrix starts to break into increasingly smaller fragments until and the two resulting sheets with the fiber in the middle, were pressed
a critical length is reached. For the analysis of the fragments lengths, again in the laboratory press (180 � C, 3 tons, 1 min). The straight fibers
only samples where the crack went completely through the fiber were embedded in the polymer matrix were cut after cooling, using an Elas­
considered. Several specimens were broken in order to have at least 200 tocon cutting die, into dump-bell shaped tensile specimens (ISO 527–2
fragments. A fragments lengths distribution was evaluated and the type A).
weighted average fiber fragment length, l, was calculated.

Fig. 11. Fragments length distribution.

9
L. Aliotta and A. Lazzeri Composites Science and Technology 186 (2020) 107920

Fig. 12. Typical fragments of flax fibers in PLA matrix during fragmentation test.

Table 5 Table 6
Single fiber fragmentation test results. Tensile properties of PLA – Flax composites.
Weighted Mean Critical Flax fibers Interfacial shear Young Modulus Tensile Strength Elongation at break
average fiber fibers length average strength (IFSS), PLA/Flax (wt. (GPa) (MPa) (%)
fragment diameter according to strength at τ, according to %)
length (l) (mm) Eq. (11) critical Eq. (12) (MPa)
100/0 3.29 � 0.1 59.76 � 1.8 3.92 � 0.4
(mm) (mm) length (MPa)
90/10 4.69 � 0.5 63.56 � 0.7 2.8 � 0.2
1.84 0.073 2.46 950 14.10 80/20 4.75 � 0.3 66.6 � 0.3 3.04 � 0.2
60/40 6.30 � 0.2 59.37 � 1.3 1.82 � 0.3

A fragment length distribution, obtained according to the procedure


explained before, is reported if Fig. 11 whereas, some pictures of fiber [72].
fragments recorded during the test are shown in Fig. 12. After preparation and before thermal and mechanical tests, the
The average strenth of flax fibers at critical length was evaluated specimens were stored in a desiccator at room temperature, to avoid
extrapolating from Fig. 8. The SFFT results and all related data are listed water uptake.
in Table 5. Tensile tests were carried out at room temperature and at a crosshead
Generally good fiber/matrix adhesion leads to short fiber fragments speed of 10 mm/min on an Instron universal testing machine 5500R
lengths; on the other hand, poor fiber/matrix adhesion leads to longer equipped with a 10 kN load cell. The machine is interfaced with a
critical fragmentation lengths. computer running MERLIN software (version 4.42S/N–014733H). The
On the basis of the results obtained it can be observed a high value of samples were tested not before 24 h from specimen extrusion. At least
critical fiber length (2.46 mm) at which corresponds a low IFSS value five specimens for each blend were tested and the average values were
(14.10 MPa) indicating that a weak but not null adhesion between the reported.
fibers and the matrix is encountered. This value is in good accordance The fibers were extracted from the composites blends in order to
with other experimental data present in literature in which a critical evaluate: the fiber length distributions and the mean aspect ratio and to
length (measured by pull-out test) of 1.95 mm was calculated with an verify the absence of fiber twisting phenomena due to the extrusion
IFSS value for PLA-flax system of 28.3 MPa [71]. It is necessary to take which may influence the composite IFSS calculation [73]. At least 8–10
into account that the IFSS values, measured by the pull-out test, are in g of each specimen’s blends were dissolved in chloroform and were
generally higher than those measured with fragmentation test as re­ centrifuged, then the extracted fibers were dried at 80 � C for at least 24
ported by several authors in literature. The single-fiber fragmentation h. Subsequently, the fibers were analyzed by SEM (JEOL JSM-5600LV).
test produces values of interfacial shear strength that are in good At least 200 extracted fibers for each composite blends have been
agreement with the effective mechanical properties of the resulting examined and measured with Image J software.
composites [11,46,55]; this is the reason why this test was chosen for the In Table 6 are summarized the results of tensile tests for the different
model validation. fibers content. All specimens showed brittle failure with a stiffness in­
crease with the flax fibers amount. This behavior is very common and it
was encountered in literature for other similar systems [58,72,74]. Also
4.3. Composite tensile tests the tensile strength increases with the fiber content. A maximum is
reached at 20 wt% of fibers and a further fibers increment causes a drop
All materials were dried in a circulating air oven at 60 � C for at least down of tensile strength. Increasing the fiber content in fact, the matrix
24 h. Different composites were produced containing increasing amount is not sufficient enough to wet the fibers and this produces a reduction of
of flax fibers (10, 20 and 40 wt%). Blends were processed with a Thermo the tensile strength [58]. The stiffness increment, moreover, provokes a
Scientific MiniLab Haake conical twin screw extruder at a screw rate of decrease of the elongation at break with fiber content.
100 rpm/min and a cycle time of 60 s; after the extrusion the molten Before to apply the analytical models, it is useful to pay attention on
materials were transferred, through a preheated cylinder (set at the the composites fiber length distributions showed in Fig. 13. It can be
same temperature of the extruder), to a Thermo Scientific Haake MiniJet noticed that for all composites, the fibers lengths involved are all below
II mini injection moulder (mould temperature set at 60 � C) for the the critical length value (calculated by SFFT); the maximum value of
production of Haake type III dog-bone tensile bars (size: 25 x 5 � 1.5 length registered is 1100 μm.
mm). Extrusion and injection molding were carried out at 180 � C, at this An average weighted fiber length and a mean aspect ratio were
temperature PLA is at the molten state and contemporary the flax fibers calculated for the composite blends (the results are listed in Table 7). It
do not degrade. In fact, the thermal degradation of flax fibers takes place can be observed that for all composites systems the aspect ratio is around
at around 200 � C and it is not really significant in the first few minutes

10
L. Aliotta and A. Lazzeri Composites Science and Technology 186 (2020) 107920

Fig. 13. Fiber length distributions: a) PLA þ10 wt% Flax; b) PLA þ20 WT.% Flax; c) PLA þ40 wt% Flax.

Table 8
Table 7
IFSS evaluation for PLA – Flax composites.
Mean (weighted) fiber lengths and mean aspect ratio for PLA – Flax composites.
PLA/Flax IFSS according to Bader and IFSS according to “modified”
PLA/Flax (wt. %) Mean fiber length (μm) Aspect ratio
(wt. %) Bowyer model (MPa) model (MPa)
90/10 530 7.3
90/10 42.82 33.90
80/20 620 8.5
80/20 20.65 13.03
60/40 593 8.2
60/40 19.91 11.94

orientation factor [75].


Being all fibers below the critical length, for all the analytical models,
the only contribution related to the fibers having L < Lc must be
considered. Consequently, for the classic Bader and Bowyer model and
for the modified version, the expressions (Eq. (12) and Eq. (14)) become
very simple and a direct calculation of the IFSS is possible not consid­
ering the contribution of L > Lc. The IFSS trends obtained from these
models are illustrated in Fig. 14.
The IFSS trends of the composites are in agreement with literature
data in which the IFSS was observed to slight decrease with increasing
fiber content [73,76]. The B–B model, extrapolating the data from the
experimental stress-strain curves (that change with the fibers content)
Fig. 14. IFFS trend for PLA – Flax composites.
gives values of the IFSS that varies with the volumetric fibers amount.
However, IFSS is an interface property that is independent from the
8. Consequently, these composite systems belong to the class of ultra- volumetric filler fraction, as a consequence a mean value, among the
short fiber composites (aspect ratio < 10). values obtained at different volumetric percentages, is taken [52,73,77].
It can be observed that with the “modified” version of Bader and
4.4. Model application and validation Bowyer model, the IFSS values (reported in Table 8) are lower and more
realistic. In particular, with the classic model at 10 wt% of fiber content,
Before the application of the analytical models some considerations τ overcomes the Von Mises threshold (35 MPa). This threshold, used to
have to be done. The value of σ0 (fiber end stress) was taken, from the discuss the fibre/matrix interfacial shear strength, is kept as conserva­
tensile tests on pure PLA, as the maximum stress value that matrix can tive value. In fact, the maximum value that the shear strength can reach,
reach (equal to 65 MPa). Due to their very short aspect ratio, a randomly (that cannot be measured directly) can be estimated using the von Mises
fiber disposition was considered, consequently 3/8 was taken as fibers relationship (τ ¼ σm/√3) [78].

11
L. Aliotta and A. Lazzeri Composites Science and Technology 186 (2020) 107920

Table 9 Sci. Manuf. 31 (2000) 1117–1125, https://doi.org/10.1016/S1359-835X(00)


Comparison of different critical lengths and IFSS values. 00068-3.
[3] T. Yu, N. Jiang, Y. Li, Study on short ramie fiber/poly(lactic acid) composites
IFSS value IFSS value Lc according to Lc calculated Lc compatibilized by maleic anhydride, Compos. Part A Appl. Sci. Manuf. 64 (2014)
according to according to the classical according the calculated 139–146, https://doi.org/10.1016/j.compositesa.2014.05.008.
Bader and “modified” Kelly-Tyson “modified” with SFTT [4] Y.U. Tao, L.I. Yan, R.E.N. Jie, Preparation and properties of short natural fiber
Bowyer model (MPa) model for version (mm) (mm) reinforced poly (lactic acid) composites, Trans. Nonferrous Metals Soc. China 19
model (MPa) classic Bader (2009) s651–s655.
and Bowyer [5] J.L. Thomason, Interfacial strength in thermoplastic composites - at last an industry
friendly measurement method? Compos. Part A Appl. Sci. Manuf. 33 (2002)
model (mm)
1283–1288, https://doi.org/10.1016/S1359-835X(02)00150-1.
20.65 13.03 1.67 2.47 2.46 [6] P.J. Herrera-Franco, A. Valadez-Gonz??lez, A study of the mechanical properties of
short natural-fiber reinforced composites, Compos. B Eng. 36 (2005) 597–608,
https://doi.org/10.1016/j.compositesb.2005.04.001.
[7] D. Cho, J.M. Seo, H.S. Lee, C.W. Cho, S.O. Han, W.H. Park, Property improvement
Generally, when the Bader and Bowyer approach is used, a mean of natural fiber-reinforced green composites by water treatment, Adv. Compos.
value of IFSS is considered. It is interesting to note that the classic Bader Mater. Off. J. Japan Soc. Compos. Mater. 16 (2007) 299–314, https://doi.org/
and Bowyer model gives a mean IFSS of 20.65 MPa that it is still over­ 10.1163/156855107782325249.
[8] A. Arbelaiz, G. Cantero, B. Fern� andez, I. Mondragon, P. Ga~ n�
an, J.M. Kenny, Flax
estimated respect to the SFFT value (14.10 MPa).
fiber surface modifications: effects on fiber physico mechanical and flax/
On the other hand, the “modified” Bader and Bowyer model provides polypropylene interface properties, Polym, Compos 26 (2005) 324–332, https://
an IFSS mean value (13.03 MPa) that is more realistic and very close to doi.org/10.1002/pc.20097.
the experimental result confirming that this modified model, tailored for [9] D.A. Jesson, J.F. Watts, The interface and interphase in polymer matrix composites:
effect on mechanical properties and methods for identification, Polym. Rev. 52
ultra-short fibers composites, fits better the experimental data providing (2012) 321–354, https://doi.org/10.1080/15583724.2012.710288.
more realistic results. [10] S. Wong, R.A. Shanks, A. Hodzic, Effect of additives on the interfacial strength of
To confirm further the validity of this new model version, consid­ poly(l-lactic acid) and poly(3-hydroxy butyric acid)-flax fibre composites, Compos.
Sci. Technol. 67 (2007) 2478–2484, https://doi.org/10.1016/j.
ering the IFSS value obtained, the critical length was evaluated. The compscitech.2006.12.016.
standard Kelly-Tyson equation (Eq. (1)) was used for the classic B–B [11] A. Valadez-Gonzalez, J.M. Cervantes-Uc, R. Olayo, P.J. Herrera-Franco, Effect of
model, whereas for the modified version the new critical length equation fiber surface treatment on the fiber-matrix bond strength of natural fiber reinforced
composites, Compos. B Eng. 30 (1999) 309–320, https://doi.org/10.1016/S1359-
(Eq. (4)) was applied. 8368(98)00054-7.
The IFSS values and the critical lengths values are summarized in [12] A. Lazzeri, V.T. Phuong, Dependence of the Puk� anszky’s interaction parameter B
Table 9. It is important to observe that the modified Bader and Bowyer on the interface shear strength (IFSS) of nanofiller- and short fiber-reinforced
polymer composites, Compos. Sci. Technol. 93 (2014) 106–113, https://doi.org/
model provides a critical length value that is very close to the experi­ 10.1016/j.compscitech.2014.01.002.
mental result. On the other hand, the classic model differs strongly and [13] B. Lauke, Determination of adhesion strength between a coated particle and
provides an underestimation of the critical length value. polymer matrix, Compos. Sci. Technol. 66 (2006) 3153–3160, https://doi.org/
10.1016/j.compscitech.2005.01.018.
[14] M. Zappalorto, M. Salviato, M. Quaresimin, Influence of the interphase zone on the
5. Conclusions nanoparticle debonding stress, Compos. Sci. Technol. 72 (2011) 49–55, https://
doi.org/10.1016/j.compscitech.2011.09.016.
[15] I. Sevostianov, M. Kachanov, Effect of interphase layers on the overall elastic and
In this paper a modification of the Kelly-Tyson equation for short and conductive properties of matrix composites. Applications to nanosize inclusion, Int.
ultra-short fiber composites has been proposed. The starting hypothesis J. Solids Struct. 44 (2007) 1304–1315, https://doi.org/10.1016/j.
is the non-negligibility of the fiber end stresses in the case of short and ijsolstr.2006.06.020.
[16] G.M. Odegard, T.C. Clancy, T.S. Gates, Modeling of the mechanical properties of
ultra-short fibers composites. A new equation for the IFSS calculation nanoparticle/polymer composites, Polymer 46 (2005) 553–562, https://doi.org/
using the Bader and Bowyer approach was obtained. 10.1016/j.polymer.2004.11.022.
A few examples of IFSS calculations, τ, have been provided using [17] R.K. Patel, B. Bhattacharya, S. Basu, Effect of interphase properties on the damping
response of polymer nano-composites, Mech. Res. Commun. 35 (2008) 115–125,
published literature data. All results show a lowering of the IFSS values https://doi.org/10.1016/j.mechrescom.2007.08.005.
obtained compared to the classic Bader and Bowyer calculation. The [18] J.K. Kim, C. Baillie, Y.W. Mai, Interfacial debonding and fibre pull-out stresses -
new values appear more coherent and closer to experimental results. Part I Critical comparison of existing theories with experiments, J. Mater. Sci. 27
(1992) 3143–3154, https://doi.org/10.1007/BF01116004.
It has been confirmed that the Bader and Bowyer model becomes
[19] A. Kelly, W.R. Tyson, Tensile properties of fibre-reinforced metals: copper/
increasingly less accurate as the aspect ratio decreases and therefore, for tungsten and copper/molybdenum, J. Mech. Phys. Solids 13 (1965) 329–350,
very low aspect ratio, it needs to be modified. https://doi.org/10.1016/0022-5096(65)90035-9.
[20] A. Kelly, The strengthening of metals by dispersed particles, Proc. R. Soc. A Math.
A validation with experimental data also confirmed the validity of
Phys. Eng. Sci. 282 (1964) 63–79, https://doi.org/10.1098/rspa.1964.0214.
this new approach. The single fiber fragmentation test was used not only [21] L. Sun, Y. Jia, F. Ma, S. Sun, J. Zhao, C.C. Han, Analysis of interfacial adhesion
to evaluate the IFSS value but also to calculate the critical fiber length. behaviors by single-fiber composite tensile tests and surface wettability tests,
The results obtained are very interesting because not only an improve­ Polym. Compos. 31 (2010) 1457–1464, https://doi.org/10.1002/pc.20932.
[22] D. Tripathi, F.R. Jones, Review Single fibre fragmentation test for assessing
ment in the IFSS estimation was reached with this new approach, but adhesion in fibre reinforced composites, J. Mater. Sci. 33 (1998) 1–16, https://doi.
also the new critical length expression gives results that are very close to org/10.1023/A:1004351606897.
the experimental test. [23] H. Krenchel, Fibre Reinforcement : Theoretical and Practical Investigations of the
Elasticity and Strength of Fibre-Reinforced Materials, Akademisk Forlag,
Copenhagen, 1964.
[24] B.F. Blumentritt, B.T. Vu, S.L. Cooper, Mechanical properties of discontinuous fiber
Declaration of competing interest reinforced thermoplastics. II. Random-in-plane fiber orientation, Polym. Eng. Sci.
15 (1975) 428–436, https://doi.org/10.1002/pen.760150606.
[25] M.R. Piggott, Short fibre polymer composites: a fracture-based theory of fibre
The authors declare that they have no known competing financial reinforcement, J. Compos. Mater. 28 (1994), https://doi.org/10.1177/
interests or personal relationships that could have appeared to influence 002199839402800701.
the work reported in this paper. [26] W.H. Bowyer, M.G. Bader, On the re-inforcement of thermoplastics by imperfectly
aligned discontinuous fibres, J. Mater. Sci. 7 (1972) 1315–1321, https://doi.org/
10.1007/BF00550698.
References [27] M.G. Bader, W.H. Bowyer, An improved method of production for high strength
fibre-reinforced thermoplastics, Composites 4 (1973) 150–156, https://doi.org/
10.1016/0010-4361(73)90105-5.
[1] F. Rezaei, R. Yunus, N.A. Ibrahim, E.S. Mahdi, Development of short-carbon-fiber-
[28] B. Puk� anszky, Influence of interface interaction on the ultimate tensile properties
reinforced polypropylene composite for car bonnet, Polym. Plast. Technol. Eng. 47
of polymer composites, Composites 21 (1990) 255–262, https://doi.org/10.1016/
(2008) 351–357, https://doi.org/10.1080/03602550801897323.
0010-4361(90)90240-W.
[2] S.Y. Fu, B. Lauke, E. M€
ader, C.Y. Yue, X. Hu, Tensile properties of short-glass-fiber-
and short-carbon-fiber-reinforced polypropylene composites, Compos. Part A Appl.

12
L. Aliotta and A. Lazzeri Composites Science and Technology 186 (2020) 107920

[29] E. Bilotti, R. Zhang, H. Deng, F. Quero, H.R. Fischer, T. Peijs, Sepiolite needle-like [55] N. Graupner, J. R€ oßler, G. Ziegmann, J. Müssig, Fibre/matrix adhesion of cellulose
clay for PA6 nanocomposites: an alternative to layered silicates? Compos. Sci. fibres in PLA, PP and MAPP: a critical review of pull-out test, microbond test and
Technol. 69 (2009) 2587–2595, https://doi.org/10.1016/j. single fibre fragmentation test results, Compos. Part A Appl. Sci. Manuf. 63 (2014)
compscitech.2009.07.016. 133–148, https://doi.org/10.1016/j.compositesa.2014.04.011.
[30] K. Renner, C. Keny� o, J. M�ocz�
o, B. Puk�
anszky, Micromechanical deformation [56] H. Anuar, A. Zuraida, B. Morlin, J.G. Kov� acs, Micromechanical property
processes in PP/wood composites: particle characteristics, adhesion, mechanisms, investigations of poly(lactic acid)-kenaf fiber biocomposites, J. Nat. Fibers 8
Compos. Part A Appl. Sci. Manuf. 41 (2010) 1653–1661, https://doi.org/10.1016/ (2011) 14–26, https://doi.org/10.1080/15440478.2011.550765.
j.compositesa.2010.08.001. [57] P.M. Remiro, K. Fakultatea, E.H. Unibertsitatea, J. Pouyet, L.D.M. Physique, J.
[31] B. Cioni, A. Lazzeri, The role of interfacial interactions in the toughening of R. Sarasua, The mechanical behaviour of PEEK short fibre composites, J. Mater.
precipitated calcium carbonate–polypropylene nanocomposites, Compos. Interfac. Sci. 30 (1995) 3501–3508, https://doi.org/10.1007/BF00349901.
17 (2010) 533–549, https://doi.org/10.1163/092764410X513486. [58] K. Oksman, M. Skrifvars, J.F. Selin, Natural fibres as reinforcement in polylactic
[32] V. Mittal, Optimization of Polymer Nanocomposite Properties, John Wiley & Sons, acid (PLA) composites, Compos. Sci. Technol. 63 (2003) 1317–1324, https://doi.
2009. org/10.1016/S0266-3538(03)00103-9.
[33] B. Imre, G. Keledi, K. Renner, J. M� ocz�
o, M. Murariu, P. Dubois, B. Puk� anszky, [59] P.K. Ilankeeran, P.M. Mohite, S. Kamle, Axial tensile testing of single fibres, Mod.
Adhesion and micromechanical deformation processes in PLA/CaSO4 composites, Mech. Eng. 02 (2012) 151–156, https://doi.org/10.4236/mme.2012.24020.
Carbohydr. Polym. 89 (2012) 759–767, https://doi.org/10.1016/j. [60] C. Romhany, J. Karger-Kocsis, T. Czigany, Tensile fracture and failure behavior of
carbpol.2012.04.005. technical, J. Appl. Polym. Sci. 90 (2003) 3638–3645.
[34] H.G. Kim, L.K. Kwac, Evaluation of elastic modulus for unidirectionally aligned [61] Y. Zhang, X. Wang, N. Pan, R. Postle, Weibull analysis of the tensile behavior of
short fiber composites, J. Mech. Sci. Technol. 23 (2009) 54–63, https://doi.org/ fibers with geometrical irregularities, J. Mater. Sci. 37 (2002) 1401–1406, https://
10.1007/s12206-008-0810-1. doi.org/10.1023/A:1014580814803.
[35] H.L. Cox, The elasticity and strength of paper and other fibrous materials, Br. J. [62] W. Weibull, A statistical distribution function of wide applicability, J. Appl. Mech.
Appl. Phys. 3 (1952) 72–79, https://doi.org/10.1088/0508-3443/3/3/302. 18 (1951) 293–297, doi:citeulike-article-id:8491543.
[36] P. Zhao, S. Ji, Refinements of shear-lag model and its applications, Tectonophysics [63] A.S. Watson, R.L. Smith, An examination of statistical theories for fibrous materials
279 (1997) 37–53, https://doi.org/10.1016/S0040-1951(97)00129-7. in the light of experimental data, J. Mater. Sci. 20 (1985) 3260–3270, https://doi.
[37] S.J. Eichhorn, R.J. Young, Deformation micromechanics of natural cellulose fibre org/10.1007/BF00545193.
networks and composites, Compos. Sci. Technol. 63 (2003) 1225–1230, https:// [64] R. Joffe, J.A. Andersons, L. Wallstr€ om, Strength and adhesion characteristics of
doi.org/10.1016/S0266-3538(03)00091-5. elementary flax fibres with different surface treatments, Compos. Part A Appl. Sci.
[38] M.J. Starink, S. Syngellakis, Shear lag models for discontinuous composites: fibre Manuf. 34 (2003) 603–612, https://doi.org/10.1016/S1359-835X(03)00099-X.
end stresses and weak interface layers, Mater. Sci. Eng. A 270 (1999) 270–277, [65] F. Wang, J. Shao, Modified Weibull distribution for analyzing the tensile strength
https://doi.org/10.1016/S0921-5093(99)00277-4. of bamboo fibers, Polymers 6 (2014) 3005–3018, https://doi.org/10.3390/
[39] C. Galiotis, R. Young, The study of model polydiacetylene/epoxy composites, polym6123005.
J. Mater. Sci. 19 (1984) 3640–3648, https://doi.org/10.1007/BF02396936. [66] L. Pardini, G.B. Manhani, Influence of the testing gage length on the strength,
[40] J.M. Papazian, P.N. Adler, Tensile properties of short fiber-reinforced SiC/Ai Young’s modulus and Weibull modulus of carbon fibres and glass fibres, Mater.
composites: Part I. effects of matrix precipitates, Metall. Trans. A. 21 (1990) Res. 5 (2002) 411–420, https://doi.org/10.1590/S1516-14392002000400004.
401–410, https://doi.org/10.1007/BF02782420. [67] A. Bledzki, J. Gassan, Composites reinforced with cellulose based fibres, Prog.
[41] T.W. Clyne, A simple development of the shear lag theory appropriate for Polym. Sci. 24 (1999) 221–274.
composites with a relatively small modulus mismatch, Mater. Sci. Eng. A 122 [68] M.C.A. Teles, G.R. Alto� e, P. Amoy Netto, H. Colorado, F.M. Margem, S.N. Monteiro,
(1989) 183–192, https://doi.org/10.1016/0921-5093(89)90629-1. Fique fiber tensile elastic modulus dependence with diameter using the Weibull
[42] H. Kim, Investigation of a stress field evaluated by elasticplastic analysis in statistical analysis, Mater. Res. 18 (2015) 193–199, https://doi.org/10.1590/1516-
discontinuous composites, Int. J. Automot. Technol. 8 (2007) 483–491. 1439.364514.
[43] M. Taya, R.J. Arsenault, A comparison between a shear lag type model and an [69] L. Peponi, J. Biagiotti, L. Torre, J.M. Kenny, I. Mondrag� on, Statistical analysis of
eshelby type model in predicting the mechanical properties of a short fiber the mechanical properties of natural fibers and their composite materials. I.
composite, Scr. Metall. 21 (1987) 349–354, https://doi.org/10.1016/0036-9748 Natural fibers, Polym. Compos. 29 (2008) 313–320, https://doi.org/10.1002/
(87)90227-4. pc.20408.
[44] A. Le Duigou, P. Davies, C. Baley, Interfacial bonding of Flax fibre/Poly(l-lactide) [70] I.M. De Rosa, J.M. Kenny, D. Puglia, C. Santulli, F. Sarasini, Morphological,
bio-composites, Compos. Sci. Technol. 70 (2010) 231–239, https://doi.org/ thermal and mechanical characterization of okra (Abelmoschus esculentus) fibres
10.1016/j.compscitech.2009.10.009. as potential reinforcement in polymer composites, Compos. Sci. Technol. 70 (2010)
[45] P.J.J. Herrera-Franco, L.T. Drzal, Comparison of methods for the measurement of 116–122, https://doi.org/10.1016/j.compscitech.2009.09.013.
fibre/matrix adhesion in composites, Composites 23 (1992) 2–27, https://doi.org/ [71] S.S. Ray, K. Yamada, M. Okamoto, K. Ueda, Polylactide-layered silicate
10.1016/0010-4361(92)90282-Y. nanocomposite: a novel biodegradable material, Nano Lett. 2 (2002) 1093–1096,
[46] V. Rao, P. Herrera-Franco, A.D. Ozzello, L.T. Drzal*, A direct comparison of the https://doi.org/10.1021/nl0202152.
fragmentation test and the microbond pull-out test for determining the interfacial [72] H.L. Bos, J. Müssig, M.J.A. van den Oever, Mechanical properties of short-flax-fibre
shear strength, J. Adhes. 34 (1991) 65–77, https://doi.org/10.1080/ reinforced compounds, Compos. Part A Appl. Sci. Manuf. 37 (2006) 1591–1604,
00218469108026506. https://doi.org/10.1016/j.compositesa.2005.10.011.
[47] A. Pegoretti, L. Fambri, C. Migliaresi, Interfacial stress transfer in nylon-6/E-glass [73] V. Gigante, L. Aliotta, V.T. Phuong, M.B. Coltelli, P. Cinelli, A. Lazzeri, Effects of
microcomposites: effect of temperature and strain rate, Polym, Compos 21 (2000) waviness on fiber-length distribution and interfacial shear strength of natural fibers
466–475, https://doi.org/10.1002/pc.10202. reinforced composites, Compos. Sci. Technol. 152 (2017) 129–138, https://doi.
[48] T.H. Courtney, Mechanical Behavior of Materials, Waveland Press, 2005. org/10.1016/j.compscitech.2017.09.008.
[49] A. Kelly, G.J. Davies, The principles of the fibre reinforcement of metals, Metall. [74] A. Shakoor, R. Muhammad, N.L. Thomas, V.V. Silberschmidt, Mechanical and
Rev. 10 (77) (1965) 1–77, https://doi.org/10.1179/095066065790138357. thermal characterisation of poly (l-lactide) composites reinforced with hemp fibres,
[50] L. Aliotta, V. Gigante, M.B. Coltelli, P. Cinelli, A. Lazzeri, L. Aliotta, V. Gigante, M. J. Phys. Conf. Ser. 451 (2013), https://doi.org/10.1088/1742-6596/451/1/
B. Coltelli, P. Cinelli, A. Lazzeri, Evaluation of mechanical and interfacial 012010.
properties of bio-composites based on poly(lactic acid) with natural cellulose [75] A.R. Sanadi, R.A. Young, C. Clemons, R.M. Rowell, Recycled newspaper fibers as
fibers, Int. J. Mol. Sci. 20 (2019) 960, https://doi.org/10.3390/IJMS20040960, 20 reinforcing fillers in thermoplastics: Part I-analysis of tensile and impact properties
(2019) 960. in polypropylene, J. Reinf. Plast. Compos. 13 (1994) 54–67, https://doi.org/
[51] M.C. Righetti, P. Cinelli, N. Mallegni, C.A. Massa, L. Aliotta, A. Lazzeri, Thermal, 10.1177/073168449401300104.
mechanical, viscoelastic and morphological properties of poly ( lactic acid ) based [76] J.L. Thomason, The influence of fibre length and concentration on the properties of
biocomposites with potato pulp powder treated with waxes, Materials 12 (2019), glass fibre reinforced polypropylene: 7. Interface strength and fibre strain in
https://doi.org/10.3390/ma12060990. injection moulded long fibre PP at high fibre content, Compos. Part A Appl. Sci.
[52] A. Serrano, F.X. Espinach, F. Julian, R. Del Rey, J.A. Mendez, P. Mutje, Estimation Manuf. 38 (2007) 210–216, https://doi.org/10.1016/j.compositesa.2006.01.007.
of the interfacial shears strength, orientation factor and mean equivalent intrinsic [77] M.E. Vallejos, F.X. Espinach, F. Juli� an, L. Torres, F. Vilaseca, P. Mutj�e,
tensile strength in old newspaper fiber/polypropylene composites, Compos. B Eng. Micromechanics of hemp strands in polypropylene composites, Compos. Sci.
50 (2013) 232–238, https://doi.org/10.1016/j.compositesb.2013.02.018. Technol. 72 (2012) 1209–1213, https://doi.org/10.1016/j.
[53] Y. Li, K.L. Pickering, R.L. Farrell, Determination of interfacial shear strength of compscitech.2012.04.005.
white rot fungi treated hemp fibre reinforced polypropylene, Compos. Sci. Technol. [78] A. Pegoretti, C. Della Volpe, M. Detassis, C. Migliaresi, H.D. Wagner,
69 (2009) 1165–1171, https://doi.org/10.1016/j.compscitech.2009.02.018. Thermomechanical behaviour of interfacial region in carbon fibre/epoxy
[54] R. del Rey, R. Serrat, J. Alba, I. Perez, P. Mutje, F.X. Espinach, Effect of sodium composites, Compos. Part A Appl. Sci. Manuf. 27 (1996) 1067–1074, https://doi.
hydroxide treatments on the tensile strength and the interphase quality of hemp org/10.1016/1359-835X(96)00065-6.
core fiber-reinforced polypropylene composites, Polymers 9 (2017) 6–8, https://
doi.org/10.3390/polym9080377.

13

You might also like