materials

Article
Fracture Assessment of PEEK under Static Loading
by Means of the Local Strain Energy Density
Mirco Peron *, Seyed Mohammad Javad Razavi ID
, Jan Torgersen and Filippo Berto
Department of Industrial and Mechanical Engineering, Norwegian University of Science and Technology,
Richard Birkelands vei 2b, 7491 Trondheim, Norway; javad.razavi@ntnu.no (S.M.J.R.);
jan.torgersen@ntnu.no (J.T.); filippo.berto@ntnu.no (F.B.)
* Correspondence: mirco.peron@ntnu.no; Tel.: +47-73593850

Received: 12 October 2017; Accepted: 11 December 2017; Published: 13 December 2017

Abstract: Polyetheretherketone (PEEK) has gained interest in many industrial applications

due to its high strength-to-weight ratio, excellent heat tolerance and high corrosion resistance.
Stress concentrators such as notches and geometrical discontinuities are present in many such
components necessitating the reliable assessment of notch sensitivity of PEEK in monotonic
tension. Here we evaluate the applicability of the strain energy density (SED) approach for
the assessment of the fracture strength of experimentally tested notched geometries subject to
corrosion. The fracture behavior of neat, circumferentially razor-grooved dog-bone specimens and
circumferentially U-notched specimens with different notch radii can be predicted with a discrepancy
lower than ±10%. Reliable predictions are shown on two previously published datasets employing
both computed and published mechanical properties as inputs for the SED calculations. This report
presents the first successful application of SED for PEEK as well as the successful prediction of tensile
behavior in corrosive environments. This opens the road towards future applications of PEEK in
fields its compliant use is of growing popularity.

Keywords: tensile behavior; PEEK; strain energy density; SED; notch

1. Introduction
For applications where high specific mechanical properties and high corrosion resistance are
needed, advanced polymers are attractive [1]. Polyetheretherketone (PEEK) emerged as a successful
engineering polymer due to its high strength-to-weight ratio, excellent tribocorrosion and thermal
properties (its glass transition temperature is of 143 ◦ C [2]). It is a reliable substitute for metallic
materials in many industrial applications and a key material in food processing, for impeller wheels
in regenerative pumps, in high pressure and temperature pipes and hose couplings as well as
gearwheels [3,4].
Numerous authors have reported on the strain rate dependency of this material. El-Qoubaa and
Othman [5] assessed the tensile mechanical behavior of PEEK at room temperature at a range of strain
rates (from 0.001 to 1000 s−1 ). The yield stress increased with strain rate, with the strain rate sensitivity
reported higher at higher rates. Albérola et al. [6] reported an increase of Young’s modulus and yield
stress by increasing the strain rate from 10 to 200 s−1 . Similarly, strain rate sensitivity was weaker for
strain rates 0.00001 to 10 s−1 . Most industrial components contain geometrical discontinuities (notches)
negatively affecting fracture and fatigue strength [7–9]. Despite their importance for dimensioning
against failure, a lack of knowledge exists for assessing the notch sensitivity of PEEK. To the best
of authors’ knowledge, only Sobieraj et al. [10] and Chen et al. [11] assessed notch effects on tensile
strength of this semi-crystalline polymer. Sobieraj investigated the static behavior of un-notched and
notched PEEK in corrosive environment at two different strain rates (0.1 and 0.5 s−1 ), whereas Chen

Materials 2017, 10, 1423; doi:10.3390/ma10121423 www.mdpi.com/journal/materials

Materials 2017, 10, 1423 2 of 10
Materials 2017, 10, 1423 2 of 10

whereasthe
explored Chen explored
fracture the fracture
behavior of fourbehavior
differentofaxisymmetric
four different notched
axisymmetric notched
specimens. Bothspecimens.
pioneering Both
pioneering studies state that PEEK is notch-sensitive; yet a fundamental
studies state that PEEK is notch-sensitive; yet a fundamental and reliable failure criterion for PEEK’s and reliable failure criterion
for PEEK’s
tensile behavior tensile
needsbehavior
to be found.needs toThis be found.
criterionThis criterion
should not should not only
only predict the predict the tensile
tensile behavior
behavior independent from type of PEEK base material and geometry,
independent from type of PEEK base material and geometry, but should also account for degradation but should also account for
degradation
effects effects due toimpacts.
due to environmental environmental impacts.
For the first time, Forwe aim the to
first time, such
provide we aim to provide
a criterion such a
employing
criterion employing an energy-based approach, i.e., the strain
an energy-based approach, i.e., the strain energy density (SED) criterion. As years go by, from the energy density (SED) criterion.
As years goofby,
introduction thefrom
Absorbedthe introduction
Specific Fractureof theEnergy
Absorbed (ASFE) Specific
for theFracture Energyof(ASFE)
determination for the
the fracture
determination
toughness and the ofJ-integral
the fracture of lowtoughness
and medium and strength
the J-integral of low
structural and medium
material strength
[12], several structural
energy-based
material [12], several energy-based approaches were developed.
approaches were developed. Among them, the strain energy density (SED) criterion has revealed Among them, the strain energy
density (SED) criterion has revealed to be robust for predicting tensile
to be robust for predicting tensile and fatigue behavior of various metals weakened by geometrical and fatigue behavior of various
metals weakened
discontinuities by geometrical
[8,13–16]. discontinuities
In addition, the authors [8,13–16].
recentlyInprovided
addition, indications
the authors recently
that suggest provided
its
indications that suggest its applicability for polymeric materials [17].
applicability for polymeric materials [17]. Polyurethane foams were tested under different loading Polyurethane foams were tested
under different
conditions providing loading conditionsprediction
a satisfactory providingof a satisfactory
experimental prediction of experimental
tensile strength (+5% to − tensile strength
10% relative
(+5% to −10% relative errors) [18]. Furthermore, polymethylmethacrylate’s
errors) [18]. Furthermore, polymethylmethacrylate’s (PMMA’s) fracture behavior under different (PMMA’s) fracture behavior
under conditions
loading different loading conditions were
and temperatures and temperatures
assessed usingwere assessedThe
SED [19,20]. using SED tested
authors [19,20]. The authors
specimens at
tested specimens at room temperature
◦ and at −60 °C. For both temperature
room temperature and at −60 C. For both temperature conditions, the SED criterion provided reliable conditions, the SED criterion
provided reliable
predictions. The entire predictions.
spectrumThe entire
of 70 spectrum of
experimental 70 experimental
tensile strength values tensile
fellstrength
within avalues fell within
scatterband of
a scatterband of ±15%. Most importantly, the majority of values
±15%. Most importantly, the majority of values fell inside the range of ±10% scatter band, whichfell inside the range of ±10% scatter
is
band, which is considered a reasonable prediction in the engineering
considered a reasonable prediction in the engineering field according to a recent review dealing withfield according to a recent review
dealing
brittle andwith brittle and
quasi-brittle quasi-brittle
fracture of various fracture of various
materials [7]. Here, materials
we show [7].the
Here, we prediction
reliable show the reliable
of the
prediction of the fracture strength of PEEK comparing theoretical
fracture strength of PEEK comparing theoretical SED results to the experimental data published by SED results to the experimental
data published
Sobieraj et al. [10]byand Sobieraj
Chenet etal.
al.[10]
[11].and
In Chen
what et al. [11].the
follows, In what
reader follows,
will bethe reader will
presented with beapresented
reliable
with a reliable estimation of the characteristic strain rate sensitivity
estimation of the characteristic strain rate sensitivity that is independent from specimen geometry, that is independent from
specimen
from geometry,
the reference from
dataset andthe reference
that accountsdataset
for the and
impact thatofaccounts
the degradingfor the impact of the
environment degrading
surrounding
environment
the specimen. surrounding the specimen.

2.2.Experimental
ExperimentalReference
ReferenceData
Data
Sobierajetetal.al.[10]
Sobieraj [10]examined
examinedthethestress-strain
stress-strainbehavior
behaviorofofneat
neatand
andnotched
notchedPEEK
PEEKunder
underuniaxial
uniaxial
loads.Their
loads. Their notched
notched specimens
specimens are circumferentially
are circumferentially grooved
grooved round
round bars bars
with an 8with an 8 mm
mm outer outer
diameter
diameter by
weakened weakened
three differentby three different
types types These
of notches. of notches. These are circumferentially
are circumferentially U-notched
U-notched geometries
geometries
with a 6 mm with
inneradiameter
6 mm inner
withdiameter
moderatewith(0.9moderate
mm) and (0.9
deepmm)(0.45and
mm)deep (0.45
notch mm)Innotch
radii. radii.
addition,
a In addition, a circumferentially
circumferentially razor groovedrazor grooved
dog-bone was dog-bone was(Figure
investigated investigated
1a). (Figure 1a).

(a) (b)
Figure
Figure 1. 1. Schematic
Schematic view
view ofof specimen
specimen geometries:(a)(a)
geometries: Geometriestested
Geometries testedininRef.
Ref. [10];
[10]; (b)(b) Geometries
Geometries
tested
tested inin Ref.
Ref. [11].
[11].
Materials 2017, 10, 1423 3 of 10

Materials 2017, 10, 1423 3 of 10

The specimens, made by OPTIMA LT1™ (Invibio, Inc., West Conshohocken, PA, USA), were tested
The specimens, made by OPTIMA LT1™ (Invibio, Inc., West Conshohocken, PA, USA), were
in a corrosive environment with two different strain rates at 0.1 and 0.5 s−1 , respectively. The maximum
tested in a corrosive environment with two different strain rates at 0.1 and 0.5 s−1, respectively.
axial true stresses together with the experimental scatter are listed in Table 1. The strain rate dependence
The maximum axial true stresses together with the experimental scatter are listed in Table 1.
of Young’s modulus and yield stress was found to be weak (Figure 2), confirming the results obtained
The strain rate dependence of Young’s modulus and−yield stress was found to be weak (Figure 2),
in reference [6] in which strain rates from 0.00001 s 1 to 10 s−1 were reported to negligibly affect
confirming the results obtained in reference [6] in which strain rates from 0.00001 s−1 to 10 s−1 were
these properties.
reported to negligibly affect these properties.
Table 1. Tensile properties of notched and un-notched polyetheretherketone (PEEK) specimens under
Table 1. Tensile
different properties of notched and un-notched polyetheretherketone (PEEK) specimens under
strain rates.
different strain rates.
Condition Un-Notched Moderate Deep Razor
Condition Un-Notched Moderate Deep Razor
−1 )
Strain rate (s−1 0.1 0.50.5 0.1 0.1 0.5 0.5 0.1 0.5 0.5 0.10.1 0.5
) 0.1 0.1 0.5
Max axial true
Max axial true 211 ±±8.16
211 8.16 ± 5.35
225225 ± 5.35 132 ±132 ± 1.11 135 ± 135
1.11 127 ±127
0.42± 0.42 2.33± 2.33129 ± 1.36
129 ± 4.92
± 1.36 119119 123±±4.33
± 4.92 123 4.33
stress(MPa)
stress (MPa)

(a) (b) (c)

Figure
Figure2.2.Control
Controlvolume
volumefor
forsharp
sharpV-notch
V-notch(a),
(a), crack
crack case
case (b)
(b) and
and U
U notch
notch (c)
(c) under
under mode
mode II loading.
loading.

Chen et al. [11] assessed the fracture behavior of PEEK 450G (Röchling Group, Mannheim, Germany)
Chen et al. [11] assessed the fracture behavior of PEEK 450G (Röchling Group, Mannheim,
testing axisymmetric specimens machined with notched radii of 0.5, 1, 2 and 4 mm (Figure 1b),
Germany) testing axisymmetric specimens machined with notched radii of 0.5, 1, 2 and 4 mm
respectively. The specimens were manufactured with an outer diameter of 8 mm and a net section of
(Figure 1b), respectively. The specimens were manufactured with an outer diameter of 8 mm and a
φ = 3 mm. The tensile tests were carried out at room temperature under a strain rate of 0.1 s−1 and the
net section of φ = 3 mm. The tensile tests were carried out at room temperature under a strain rate
true stresses are listed in Table 2 (no experimental scatter were reported in the reference text).
of 0.1 s−1 and the true stresses are listed in Table 2 (no experimental scatter were reported in the
reference text). Table 2. Tensile properties of notched axisymmetric specimens.

r (mm)properties of notched4axisymmetric
Table 2. Tensile 2 1
specimens. 0.5
Maximum axial true stress (MPa) 160 145 150 177
r(mm) 4 2 1 0.5
3. Brief Introduction Maximum
of the SED Approach
axial true stress (MPa) 160 145 150 177

The SED criterion states that the failure of a component occurs when the total strain energy, W,
3. Brief Introduction of the SED Approach
averaged in a circular control volume of radius Rc surrounding a crack or notch tip, reaches its critical
−
valueThe
Wc [21].
SEDThis is material
criterion statesdependent [7]. SED
that the failure of awas applied with
component excellent
occurs when theresults
totaltostrain
the assessment
energy, W,
of the tensile
averaged in aand fatigue
circular behavior
control of different
volume of radius materials weakened
Rc surrounding by several
a crack notch
or notch geometries
tip, reaches its[22–24].
critical
Critical parameters can be analytically derived with only few material properties [21]: The
value Wc [21]. This is material dependent [7]. SED was applied with excellent results to the assessment ultimate
tensile strength
of the tensile andoffatigue
the un-notched
behavior ofmaterial
differentσmaterials
t, the Young’s modulus E and the fracture toughness,
weakened by several notch geometries [22–24].
labeled as KIC in accordance with the convention adopted in [25]. In the case of ductile material, the
Critical parameters can be analytically derived with only few material properties [21]: The ultimate
ultimate tensile strength should be replaced with the maximum true stress or a fictitious stress
tensile strength of the un-notched material σt , the Young’s modulus E and the fracture toughness,
determined using the so-called Equivalent Material Concept [22]. In agreement with Beltrami [26],
the critical value of the total strain energy can be determined by the following:
Materials 2017, 10, 1423 4 of 10

labeled as KIC in accordance with the convention adopted in [25]. In the case of ductile material,
the ultimate tensile strength should be replaced with the maximum true stress or a fictitious stress
determined using the so-called Equivalent Material Concept [22]. In agreement with Beltrami [26],
the critical value of the total strain energy can be determined by the following:

σt2
Wc = (1)
2E
In plane problems the control volume becomes a circular sector or a circle, for V-notches or cracks
respectively (Figure 2a,b). The critical radius Rc is defined as follow [21]:

(1 + v)(5 − 8v) K Ic 2
 
Rc = (2)
4π σt

where ν is the Poisson’s ratio of the material. For a blunt V-notch or a U-notch (Figure 2c), the volume
is assumed to be of a crescent shape, where Rc is the depth measured along the bisector line. The outer
radius of the crescent shape is equal to Rc + r0 with r0 being the distance between the notch tip and the
origin of the local coordinate system (Figure 2c). Such a distance depends on the notch-opening angle
2α and the notch root radius ρ, according to the expression:

(π − 2α)
r0 = ρ (3)
(2π − 2α)

The accuracy of this analytically obtained critical radius has been validated in previous works by
Lazzarin and co-authors [7,8].
However, the fracture toughness is not always available due to the difficulties and time consuming
calculations. Yet, an estimate of the critical radius can be obtained as the radius at which the critical
SED values for two different specimen geometries are identical [27]. The computation of this SED
value can be performed utilizing finite element (FE) modeling. In this work, we employ the Ansys®
(Canonsburg, PA, USA) drastically reducing the effort that otherwise would have been spent using the
complex theoretical derivation obtained in reference [21].

4. Finite Element Model

In order to obtain the SED value, axisymmetric linear elastic 2D analyses were performed on the
notched models. Due to the double symmetry of the geometry, only one quarter of the specimens
were modeled. The 8-nodes axisymmetric element plane 83 was selected for these analyses. A mesh
convergence study was undertaken to ensure that a proper number of elements was used in FE
modelling, with elements size at the crack tip ranging from about 10−3 mm to 10−1 mm. The results
are independent from the mesh being the difference is only 0.11% between the SED value for a coarse
mesh and that for a fine mesh (Figure 3), and thus a coarse mesh was adopted for the analyses.
The mesh-insensitivity of the SED approach was previously reported also by Berto and Lazzarin [23]
for cracked and notched specimens. This represents one of the main advantages of this approach,
together with the capability of assessing the tensile behavior of different materials regardless of
the geometry. In fact, FE codes could also be used for determining the tensile strength of different
components with the stress-strain curve of the un-notched material as the only input. In this case,
however, the results are extremely dependent on the mesh, thus requiring high computational efforts
and time for achieving reliable results. We hence aim to provide a solution to this challenge that is
independent from the mesh and easy to compute.
Figure 4 illustrates the mesh pattern and the boundary conditions used for FE analyses.
Symmetric boundary conditions were used for vertical and horizontal symmetry lines of the models;
however, the top side of the model was able to move along the loading axis.
Materials 2017, 10, 1423 5 of 10
Materials 2017, 10, 1423 5 of 10
Materials 2017, 10, 1423 5 of 10

(a) (b)
(a) (b)
Figure 3. Evaluation of the strain energy density (SED) mesh sensitivity: (a) Number of elements = 2049,
Figure 3. Evaluation
Figure of the
3. Evaluation of strain energy
the strain density
energy (SED)
density mesh
(SED) sensitivity:
mesh (a)(a)Number
sensitivity: Numberofofelements
elements== 2049,
W− = 2.8431 MJ/m3; (b) Number of elements = 23, W =−2.8398 MJ/m3.
W = 2.8431
W = 2.8431 MJ/m 3 ; (b); (b)
MJ/m 3 Number
Number of elements
of elements = 23, WW
= 23, = 2.8398
= 2.8398 MJ/m3 ..
MJ/m 3

Figure 4. Typical mesh pattern of the finite element model near the notch tip and schematics of
Figure
Figure 4. Typical
4. Typical mesh mesh pattern
pattern of the
of the finite
finite element
element model
model nearthe
near thenotch
notchtip
tipand
andschematics
schematics of
of
boundary conditions.
boundary
boundary conditions.
conditions.
Materials 2017, 10, 1423 6 of 10
Materials 2017, 10, 1423 6 of 10

5.5.Results
Resultsand
andDiscussion
Discussion

5.1. Prediction with Known Critical SED

5.1. Prediction with Known Critical SED
Sobieraj et al. tested un-notched and notched PEEK specimens under different strain rates
Sobieraj et al. tested un-notched and notched PEEK specimens under different strain rates and
and in a corrosive environment, and in this section their results were analyzed in terms of SED.
in a corrosive environment, and in this section their results were analyzed in terms of SED.
The application of the SED approach requires the critical value of the radius R of the control volume
The application of the SED approach requires the critical value of the radius Rccof the control volume
and that of the strain energy density W . The critical SED value can be simply evaluated using
and that of the strain energy density Wcc. The critical SED value can be simply evaluated using
3 under a strain rate of 0.5 and
Equation (1), leading to a critical SED value of 7.278 and 6.38 MJ/m
Equation (1), leading to a critical SED value of 7.278 and 6.38 MJ/m under a strain rate of 0.5 and 0.1 s−1,
3
0.1 s−1 , respectively. Concerning the control volume; in reference [10], the fracture toughness has not
respectively. Concerning the control volume; in reference [10], the fracture toughness has not been
been reported. Determining the stress distribution at the crack tip by means of numerical simulations,
reported. Determining the stress distribution at the crack tip by means of numerical simulations, the
the fracture toughness can be estimated via the Notch Stress Intensity Factors (NSIFs) formulation
fracture toughness can be estimated via the Notch Stress Intensity Factors (NSIFs) formulation
derived by Gross and Mendelson [28]:
derived by Gross and Mendelson [28]:
√
KKI I= 22πlimlimrr(1(1− Iλ)I ) θθ (( r,
[σ )]
r ,θ = 00) (4)(4)
r r→00

where
where rr andand θ the radial
θ is the radialand
andangular
angularcoordinate
coordinateofofa apolar
polarcoordinate
coordinate system
system centered
centered at the
at the notch
notch tip,
tip,
σθθ σisθθthe
is the stress
stress component
component according
according to the
to the coordinate
coordinate system
system andand
λI isλthe
I is the William’s
William’s eigenvalue
eigenvalue [29].
[29]. Applying
Applying the tensile
the tensile strength
strength obtained
obtained experimentally
experimentally from from the razor
the razor specimen,specimen, for which
for which λI is
λI is equal
√
equal
to 0.5,tothe0.5, the fracture
fracture toughness
toughness was estimated
was estimated as 4.99as
and4.99 and
5.16 MPa5.16 m
MPa√m for a rate
for a strain strain
of rate of 0.1
0.1 and s−1 ,
0.5and
0.5 s , respectively
−1
respectively (Figure (Figure
5). The 5).increment
The increment
of theoffracture
the fracture toughness
toughness increasing
increasing the strain
the strain rate rate
waswasalso
also observed in reference
observed in reference [30]. [30].

(a) (b)

Figure 5.
Figure 5. Estimation
Estimation of
of the
the fracture
fracture toughness
toughness for
for aa strain
strain rate
rate of
of 0.1
0.1 ss−−11(a)
(a)and
and0.5
0.5ss−(b).
−1 1
(b).

In the FE code, the material was assumed isotropic and linear elastic with the Young’s modulus
In the FE code, the material was assumed isotropic and linear elastic with the Young’s modulus
E = 3500 MPa and the Poisson’s ratio ν = 0.36 as in [10]. The 8-nodes axisymmetric element plane 83
E = 3500 MPa and the Poisson’s ratio ν = 0.36 as in [10]. The 8-nodes axisymmetric element plane 83 was
was selected for these analyses with elements size at the crack tip of about 10−4 mm. Using these
selected for these analyses with elements size at the crack tip of about 10−4 mm. Using these values in
values in Equation (2), the critical radius results to be 0.128 and 0.12 mm for specimens tested at 0.1
Equation (2), the critical radius results to be 0.128 and 0.12 mm for specimens tested at 0.1 and 0.5 s−1 ,
and 0.5 s−1, respectively. The small difference in radius is due to the material’s limited strain rates
respectively. The small−1difference in radius is due to the material’s limited strain rates sensitivity below
sensitivity below 10 s [6]. We predicted the tensile strength by means of FE modeling. In the
10 s−1 [6]. We predicted the tensile strength by means of FE modeling. In the predictions, the critical
predictions, the critical failure load is taken as the one rendering an SED value of the three different
failure load is taken as the one rendering an SED value of the three different specimen geometries
specimen geometries under study (moderate, deep and razor), which is equal to the critical one
under study (moderate, deep and razor), which is equal to the critical one obtained with Equation (1).
obtained with Equation (1). The SED value is proportional to the square of the applied stress. A unit
The SED value is proportional to the square of the applied stress. A unit load was applied in the
load was applied in the simulations. The predicted tensile strength was then computed as the square
simulations. The predicted tensile strength was then computed as the square root of the ratio between
root of the ratio between the critical strain energy density and the SED value. The SED prediction of
the critical strain energy density and the SED value. The SED prediction of the tensile failure for both
the tensile failure for both the strain rates are reported in Table 3.
the strain rates are reported in Table 3.
Materials 2017, 10, 1423 7 of 10

Table 3. Prediction of tensile failure of moderate, deep and razor specimens using the SED approach.

Materials 2017,
Specimen 10, 1423
Geometry Strain Rate (s−1 ) Experimental Data (MPa) SED Prediction (Mpa) 7 of 10
Deviation
0.1 127 118 −6.8%
Deep
Table 3. Prediction of tensile failure of moderate, deep and razor specimens using the SED approach.
0.5 129 124 −4%
Specimen Geometry 0.1Rate (s−1) Experimental
Strain 132 Data (MPa) SED Prediction
132 (Mpa) Deviation
−0.1%
Moderate
0.50.1 135
127 139
118 +3.3%
−6.8%
Deep
0.10.5 129
119 124
118 −4%
−0.8%
Razor
Moderate 0.50.1 132
123 132
124 −0.1%
+0.8%
0.5 135 139 +3.3%
0.1 119 118 −0.8%
Razor
For the geometries and test0.5conditions reported 123in [10], the presented124
approach provides
+0.8% suitable
prediction of tensile failure, where deviations (the difference between the experimental and predicted
Forgenerally
values) are the geometries
lowerandthan
test conditions reported in [10],
4%, for predictions the presented
of moderate andapproach provides merely
razor grooved suitable even
prediction of tensile failure, where deviations (the difference between the experimental
below 1%. Impacts of the corrosive test environment were successfully accounted for. Yet PEEK and predicted
values) are generally lower than 4%, for predictions of moderate and razor grooved merely even
is available in different grades and formats depending on the application and processing method.
below 1%. Impacts of the corrosive test environment were successfully accounted for. Yet PEEK is
The question remains, if the presented approach still holds for different PEEK grades and geometries,
available in different grades and formats depending on the application and processing method. The
especially when
question the critical
remains, SED value
if the presented cannot still
approach be analytically obtained
holds for different PEEKbut the fracture
grades toughness is
and geometries,
known [31–33]. This will be the topic of the next section.
especially when the critical SED value cannot be analytically obtained but the fracture toughness is
known [31–33]. This will be the topic of the next section.
5.2. Predictions with Known Fracture Toughness
5.2. Predictions with Known Fracture Toughness
Chen et al. [11] assessed the tensile behavior of PEEK specimens weakened by four different
Chen et al. [11]
notched geometries. Theassessed theformulation
analytical tensile behavior of PEEK specimens
for determining the SEDweakened by four
critical value different
is not applicable
notched geometries. The analytical formulation for determining the SED critical
as the authors did not test un-notched reference specimens. As previously described, it is possible value is not
applicable as the authors did not test un-notched reference specimens. As previously
to evaluate the critical radius and SED value utilizing FE code only. We may change the radius described, it is of
possible to evaluate the critical radius and SED value utilizing FE code only. We may change the
the control volume of the specimens with two different control radii and iteratively compute the SED
radius of the control volume of the specimens with two different control radii and iteratively compute
value until a satisfying convergence is reached. This approach was shown in the previous section,
the SED value until a satisfying convergence is reached. This approach was shown in the previous
however, for the type of material studied here, the fracture toughness can be taken from literature [34].
section, however, for the type of material studied here, the fracture toughness can be taken from
In this 3
literatureEquation
way, (2)way,
[34]. In this givesEquation
us 0.37 (2)
mm critical
gives radius
us 0.37 mm leading to a SED
critical radius value
leading to aofSED2.84 MJ/m
value of for
axisymmetric ®
2.84 MJ/m3specimens with 4specimens
for axisymmetric mm notch radius
with 4 mm(obtained with
notch radius Ansys with
(obtained , Figure
Ansys6).®, Figure 6).

(a) (b)

Figure 6. Control volume (yellow) in the axisymmetric specimen machined with a notch radius of
Figure 6. Control volume (yellow) in the axisymmetric specimen machined with a notch radius of
4 mm (a) and contour plot of the SED (in MJ/mm3) distribution within the control volume in the
4 mm (a) and contour plot of the SED (in MJ/mm3 ) distribution within the control volume in the
specimen weakened by a notch radius of 4 mm (b).
specimen weakened by a notch radius of 4 mm (b).
Assuming this material is isotropic, linear elastic FE analyses with an elastic modulus of 4000 MPa
and Poisson’s
Assuming thisratio of 0.38iswere
material conducted
isotropic, linear[11]. TheFE
elastic difference
analysesinwith
the elastic properties
an elastic modulusfrom
ofthose
4000 MPa
reported in the previous section is due to different grade of the material provided by two
and Poisson’s ratio of 0.38 were conducted [11]. The difference in the elastic properties from those different
Materials 2017, 10, 1423 8 of 10

reported in the previous section is due to different grade of the material provided by two different
material suppliers.
Materials Different grades correspond to different degree of crystallinity and to8 ofdifferent
2017, 10, 1423 10

length and conformations of macromolecular chains influencing the polymer’s mechanical properties.
material suppliers. Different grades correspond to different degree of crystallinity and to different
In fact, Starkweather et al. [35] reported lower mechanical properties (elastic modulus, yield stress and
length and conformations of macromolecular chains influencing the polymer’s mechanical
tensile strength) with a low degree of crystallinity. In addition, Perkins et al. [36] found an increase in
properties. In fact, Starkweather et al. [35] reported lower mechanical properties (elastic modulus,
the strain to break
yield stress andwith higher
tensile molecular
strength) with aweight. Finally,
low degree the differences
of crystallinity. in gradePerkins
In addition, of the material
et al. [36]under
studyfound an increase in the strain to break with higher molecular weight. Finally, the differences inobtained
are reflected on critical radius and critical SED value, which are different from those grade in
the previous section
of the material (data
under from
study are[10]). Grade
reflected relatedradius
on critical property discrepancies
and critical SED value,also existare
which fordifferent
PMMA [21].
The results
from thoseof the tensileinstrength
obtained prediction
the previous sectionfor thefrom
(data specimens weakened
[10]). Grade relatedbyproperty
a notchdiscrepancies
radii of 0.5, 1 and
2 mm,also exist for PMMA
respectively, [21]. The
are listed results4.of the tensile strength prediction for the specimens weakened
in Table
by a notch radii of 0.5, 1 and 2 mm, respectively, are listed in Table 4.
Table 4. Prediction of tensile failure of axisymmetric specimens machined with notched radii of 0.5, 1
and Table
2 mm 4. Prediction
using the SEDof tensile failure of axisymmetric specimens machined with notched radii of 0.5,
approach.
1 and 2 mm using the SED approach.
SpecimenGeometry
Specimen Geometry Experimental
Experimental Data
Data (MPa)
(MPa) SED
SEDPrediction
Prediction(MPa)
(MPa) Deviation
Deviation
r =r =0.5
0.5 177
177 165
165 −−6.8%
6.8%
r r= =1 1 150
150 155
155 +3.3%
+3.3%
r=2 145 156 +7.5%
r=2 145 156 +7.5%

The The
deviation between
deviation betweenthe thepredicted
predicted values
values and andthe theexperimental
experimental datadata
is inisa in a compliant
compliant scatterscatter
range withwith
range reference to to
reference previous
previouswell-known
well-known studies
studies in inthe
theengineering
engineering field.
field.
To benchmark
To benchmark thethe choiceofofthe
choice the considered
considered controlcontrolvolume,
volume, weweplotted
plotted the the
square root of
square root of
dimensionless strain energy density (W/W ) 0.50.5
as a function of the notch root radius (Figure 7).
dimensionless strain energy density (W/Wc ) as a function of the notch root radius (Figure
c 7).
It appears
It appears proportional
proportional to the
to the fracture
fracture loads
loads of the
of the tested
tested notched
notched samples;the
samples; thecontrol
controlvolume
volumeisishence
hence appropriate with respect to various notch tip radii effects, environmental conditions and strain
appropriate with respect to various notch tip radii effects, environmental conditions and strain rates.
rates. The majority of fracture load predictions are well inside the scatter band of ±10% with a
The majority of fracture load predictions are well inside the scatter band of ±10% with a considerable
considerable number of predicted data within ±5%. This is well in the performance range with many
number of predicted
studies data within
of a SED database that ± 5%. Thisinisawell
appeared recentin review
the performance
dealing withrange
brittlewith
andmany studies of a
quasi-brittle
SED fracture
databaseofthat appeared in a recent review dealing with brittle and quasi-brittle
various materials [7]. We hypothesize the SED’s routine is applicabile for mechanical fracture of various
materials [7]. We hypothesize the SED’s routine is applicabile for mechanical
properties prediction of PEEK in various material grades and geometrical configurations with properties prediction of
PEEK in various
excellent material
predictive grades and geometrical configurations with excellent predictive power.
power.

Figure
Figure 7. Synthesis
7. Synthesis of fracture
of fracture datadata in terms
in terms of normalized
of normalized SED.
SED. TheThe
datadata from
from reference
reference [10]
[10] areare
labeled
labeled in the legend reporting the geometry followed by the strain rate, while the data from
in the legend reporting the geometry followed by the strain rate, while the data from reference [11] reference
[11] report the notch radius in the legend.
report the notch radius in the legend.
Materials 2017, 10, 1423 9 of 10

6. Conclusions
The strain energy density (SED) approach was previously developed for the assessment of
tensile and fatigue behavior of metals weakened by different notch geometries. For the first time,
we successfully employed the SED to predict the tensile strength of PEEK in corrosive environments.
Various notch geometries under different strain rates and environmental conditions were simulated
and show well agreement with experimental results. Simulations both capture the strain- and
notch-sensitivity of the material independent from its grade. Predictive performance of the tensile
strength of PEEK are characterized by a discrepancy to the experimental data in the range of ±10%,
the performance range considered acceptable in many studies of a SED database [7]. This was done
independently of the type of PEEK material.

Author Contributions: Mirco Peron wrote the manuscript and analyzed the data; Seyed Mohammad
Javad Razavi wrote partially the text; Jan Torgersen and Filippo Berto provided scientific guidance, proof reading
and suggestions.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Seymour, R.B.; Carraher, C.E. Selection of Polymers for Special Applications; Springer: New York, NY, USA,
1984; pp. 205–216.
2. Sobieraj, M.C.; Murphy, J.E.; Brinkman, J.G.; Kurtz, S.M.; Rimnac, C.M. Notched fatigue behavior of PEEK.
Biomaterials 2010, 31, 9156–9162. [CrossRef] [PubMed]
3. Platt, D.K. Engineering and High Performance Plastics Market Report: A Rapra Market Report; Smithers Rapra:
Shawbury, UK, 2003.
4. Hoskins, T.J.; Dearn, K.D.; Chen, Y.K.; Kukureka, S.N. The wear of PEEK in rolling–sliding
contact—Simulation of polymer gear applications. Wear 2014, 309, 35–42. [CrossRef]
5. El-Qoubaa, Z.; Othman, R. Tensile Behavior of Polyetheretherketone over a Wide Range of Strain Rates. Int. J.
Polym. Sci. 2015, 2015, 275937. [CrossRef]
6. Albérola, N.D.; Mélé, P.; Bas, C. Tensile mechanical properties of PEEK films over a wide range of strain
rates. II. J. Appl. Polym. Sci. 1997, 64, 1053–1059. [CrossRef]
7. Berto, F.; Lazzarin, P. Recent developments in brittle and quasi-brittle failure assessment of engineering
materials by means of local approaches. Mater. Sci. Eng. R 2014, 75, 1–48. [CrossRef]
8. Berto, F.; Lazzarin, P. The volume-based Strain Energy Density approach applied to static and fatigue
strength assessments of notched and welded structures. Procedia Eng. 2009, 1, 155–158. [CrossRef]
9. Maragoni, L.; Carraro, P.A.; Peron, M.; Quaresimin, M. Fatigue behaviour of glass/epoxy laminates in the
presence of voids. Int. J. Fatigue 2017, 95, 18–28. [CrossRef]
10. Sobieraj, M.C.; Kurtz, S.M.; Rimnac, C.M. Notch sensitivity of PEEK in monotonic tension. Biomaterials 2009,
30, 6485–9644. [CrossRef] [PubMed]
11. Chen, F.; Gatea, S.; Ou, H.; Lu, B.; Long, H. Fracture characteristics of PEEK at various stress triaxialities.
J. Mech. Behav. Biomed. Mater. 2016, 64, 173–186. [CrossRef] [PubMed]
12. Gillemot, F.; Czoboly, E.; Havas, I. Fracture mechanics applications of absorbed specific fracture energy:
Notch and unnotched specimens. Theor. Appl. Fract. Mech. 1985, 5, 39–45. [CrossRef]
13. Razavi, S.M.J.; Peron, M.; Torgersen, J.; Berto, F.; Mutignani, F. Effect of hot dip galvanization on the fatigue
strength of steel bolted connections. Fract. Int. Struct. 2017, 41, 432–439.
14. Razavi, S.M.J.; Peron, M.; Torgersen, J.; Berto, F.; Welo, T. 40CrMoV13.9 notched specimens under multiaxial
fatigue: An overview of recent results. Fract. Int. Struct. 2017, 41, 440–446.
15. Gallo, P.; Razavi, S.M.J.; Peron, M.; Torgersen, J.; Berto, F. Creep behavior of V-notched components.
Fract. Int. Struct. 2017, 41, 456–463.
16. Razavi, S.M.J.; Peron, M.; Torgersen, J.; Berto, F. Static Multiaxial Fracture Behavior of Graphite Components:
A Review of Recent Results. Key Eng. Mater. 2017, 754, 35–38. [CrossRef]
17. Gómez, F.J.; Elices, M.; Berto, F.; Lazzarin, P. Local strain energy to assess the static failure of U-notches in
plates under mixed mode loading. Int. J. Fract. 2007, 145, 29–45. [CrossRef]
Materials 2017, 10, 1423 10 of 10

18. Berto, F.; Marsavina, L.; Razavi, S.M.J.; Ayatollahi, M.R. On the fracture behavior of polyurethane
notched components. Procedia Struct. Integr. 2017, 3, 144–152. [CrossRef]
19. Berto, F.; Cendon, D.A.; Lazzarin, P.; Elices, M. Fracture behaviour of notched round bars made of PMMA
subjected to torsion at −60 ◦ C. Eng. Fract. Mech. 2013, 102, 271–287. [CrossRef]
20. Berto, F.; Elices, M.; Lazzarin, P.; Zappalorto, M. Fracture behaviour of notched round bars made of PMMA
subjected to torsion at room temperature. Eng. Fract. Mech. 2012, 90, 143–160. [CrossRef]
21. Lazzarin, P.; Zambardi, R. A finite-volume-energy based approach to predict the static and fatigue behavior
of components with sharp V-shaped notches. Int. J. Fract. 2001, 112, 275–298. [CrossRef]
22. Torabi, A.R.; Alaei, M. Application of the equivalent material concept to ductile failure prediction of blunt
V-notches encountering moderate-scale yielding. Int. J. Damage Mech. 2016, 25, 853–877. [CrossRef]
23. Berto, F.; Lazzarin, P. A review of the volume-based strain energy density approach applied to V-notches
and welded structures. Theor. Appl. Fract. Mech. 2009, 52, 183–194. [CrossRef]
24. Berto, F.; Campagnolo, A.; Gallo, P. Brittle Failure of Graphite Weakened by V-Notches: A Review of Some
Recent Results Under Different Loading Modes. Strength Mater. 2015, 47, 488–506. [CrossRef]
25. Wallin, K. Fracture Toughness of Engineering Materials: Estimation and Application; EMAS Publishing:
Warrington, UK, 2011.
26. Beltrami, E. Sulle condizioni di resistenza dei corpi elastici. Nuovo Cimento 1885, 18, 145–155. [CrossRef]
27. Razavi, S.M.J.; Ferro, P.; Berto, F.; Torgersen, J. Fatigue strength of blunt V-notched specimens produced by
selective laser melting of Ti-6Al-4V. Theor. Appl. Fract. Mech. 2017, in press. [CrossRef]
28. Gross, B.; Mendelson, A. Plane elastostatic analysis of V-notched plates. Int. J. Fract. Mech. 1972, 8, 267–276.
[CrossRef]
29. Williams, M.L. Stress singularities resulting from various boundary conditions in angular corners on plates
in extension. J. Appl. Mech. 1952, 19, 526–528.
30. Wu, G.-M.; Schultz, J.M. Fracture behavior of oriented poly(ether-ether-ketone) (PEEK). Polym. Eng. Sci.
1989, 29, 405–414. [CrossRef]
31. Peron, M.; Razavi, S.M.J.; Berto, F.; Torgersen, J.; Marsavina, L. Local strain energy density for the fracture
assessment of polyurethane specimens weakened by notches of different shape. Fract. Int. Struct. 2017, 42,
214–222.
32. Peron, M.; Razavi, S.M.J.; Berto, F.; Torgersen, J.; Colussi, M. Fracture assessment of magnetostrictive
materials. Fract. Int. Struct. 2017, 42, 223–230.
33. Peron, M.; Razavi, S.M.J.; Berto, F.; Torgersen, J.; Mutignani, F. Local strain energy density for the fatigue
assessment of hot dip galvanized welded joints: Some recent outcomes. Fract. Int. Struct. 2017, 42, 205–213.
34. Gensler, R.; Béguelin, P.; Plummer, C.J.G.; Kausch, H.-H.; Miinstedt, H. Tensile behaviour and fracture
toughness of poly(ether ether ketone)/poly(ether imide) blends. Polym. Bull. 1996, 37, 111–118. [CrossRef]
35. Starkweather, H.W.; Moore, G.E.; Hansen, J.E.; Roder, T.M.; Brooks, R.E. Effect of crystallinity on the
properties of nylons. J. Polym. Sci. 1956, 21, 189–204. [CrossRef]
36. Perkins, W.G.; Capiati, N.J.; Porter, R.S. The effect of molecular weight on the physical and mechanical
properties of ultra-drawn high density polyethylene. Polym. Eng. Sci. 1976, 16, 200–203. [CrossRef]

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).