You are on page 1of 10

Mechanics of Materials 35 (2003) 1139–1148

www.elsevier.com/locate/mechmat

Modeling of deformation behavior and strain-induced


crystallization in poly(ethylene terephthalate)
above the glass transition temperature
a,b,* a,c a,c a,d
S. Ahzi , A. Makradi , R.V. Gregory , D.D. Edie
a
Center for Advanced Engineering Fibers and Films, Clemson University, Clemson, SC 29634, USA
b
Institute Professionnel des Sciences et Technologies, Universit
e Louis Pasteur, 15-17 Rue du Mar
echal Lef
ebvre,
IMFS-UMR7507, Strasbourg 67100, France
c
School of Materials Science and Engineering, Clemson University, Clemson, SC 29634, USA
d
Department of Chemical Engineering, Clemson University, Clemson, SC 29634, USA
Received 24 July 2002

Abstract

A constitutive model for large deformation stress–strain behavior and strain-induced crystallization in poly(ethylene
terephthalate), at temperatures above the glass transition temperature, is proposed. In this model, the intermolecular
resistance is treated in a composite framework where the crystalline and amorphous phases are considered as two
separate resistances coupled through two different analog representations leading to the upper and the lower bound
approaches. The crystallization rate is expressed following a non-isothermal phenomenological expression based on the
modified Avrami equation. Our predicted results are compared to existing experimental results and good agreement is
found.
Ó 2003 Elsevier Ltd. All rights reserved.

Keywords: Micromechanical modeling; Polymer hot drawing; Strain-induced crystallization; PET

1. Introduction peratures can be described in terms of stretching


an elastic–viscoplastic analog system formed with
In previous research works, it is proposed that intermolecular resistance acting in parallel with a
the large deformation behavior of poly(ethylene network resistance (Buckley and Jones, 1995;
terephthalate) (PET) fibers and films at high tem- Boyce et al., 2000). Above the glass transition
temperature the stress–strain curves (Boyce et al.,
2000; Llana and Boyce, 1999) show four basic
*
features. Under deformation, the material exhibits
Corresponding author. Address: Institute Professionnel des a relatively stiff initial response followed by a rol-
Sciences et Technologies, Universite Louis Pasteur, 15-17 Rue
du Marechal Lefebvre, IMFS-UMR7507, Strasbourg 67100,
lover to yield. After yielding, the stress increases
France. Tel.: +33-3-90244951; fax: +33-3-90244972. steadily with the strain then increases dramati-
E-mail address: said.ahzi@ipst-ulp.u-strasbg.fr (S. Ahzi). cally at large strains. The intermolecular resistance

0167-6636/$ - see front matter Ó 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0167-6636(03)00004-8
1140 S. Ahzi et al. / Mechanics of Materials 35 (2003) 1139–1148

results in the initial stiff response, the increase in


the flow stress due to strain hardening and to
crystallization, and the rate and temperature de-
pendence of the yield. The network resistance
produces the anisotropic strain hardening (orien-
tational hardening) behavior resulting from mo-
lecular alignment. Other relevant experimental and
modeling works are provided in Chandran and
Jabarin (1993), Buckley et al. (1996), Adams et al.
(1994), Matthews et al. (1997), GÕSell et al. (1999,
2000), Vigny et al. (1999) and Gorlier et al. (2001),
among many others. The present work follows
closely the above described scheme, particularly
the work of Boyce et al. (2000) where the network
resistance is modeled as a network orientation
process (Arruda and Boyce, 1993) together with a
molecular relaxation process (Bergstrom and
Boyce, 1998). However, unlike in the work of
Fig. 1. Schematic representation of the breakdown of the total
Boyce et al. (2000) and the work of Buckley and resistance into an intermolecular resistance (A) acting in par-
Jones (1995), we explicitly account for strain- allel with network resistance (B). (a) Upper bound analog
induced crystallization. In this case, the inter- model representation of the intermolecular resistance where the
molecular resistance is treated in a composite amorphous stiffness and flow are in parallel with those of the
framework where the crystalline and amorphous crystalline phase and (b) lower bound analog model represen-
tation of the intermolecular resistance where the amorphous
phases are considered as two separate resistances stiffness and flow are in series with those of the crystalline
coupled through two different analog representa- phase.
tions leading to the upper bound and the lower
bound micro-mechanical approaches (see Fig. 1).
As pointed out by Salem and co-workers (Clauss havior of PET above the glass transition temper-
and Salem, 1992; Salem, 1992, 1994) the occur- ature. The analog representation is shown in Fig.
rence of strain-induced crystallization is controlled 1, considering intermolecular resistance (resistance
by the test parameters such as the temperature and A) acting in parallel with a network resistance
the strain rate. The mathematical modeling of (resistance B). The total imposed deformation
crystallization for such processes requires the gradient F is identical to both the intermolecular
knowledge of crystallization rate as a function of and network deformation gradients FA and FB ,
temperature and molecular orientation. To ac- respectively.
count for the growth of the crystalline phase we
expressed the rate of crystallization following a F ¼ FA ¼ FB : ð1Þ
non-isothermal phenomenological expression de- The total Cauchy stress tensor T is the sum of the
veloped by McHugh and co-workers (Dairanieh intermolecular stress TA and the network stress TB .
et al., 1998; Doufas et al., 1999, 2000) using a
T ¼ TA þ TB : ð2Þ
modified Avrami equation.
In the present work, we propose to study in some
details the contribution of the intermolecular re-
2. Model formulation sistance, including strain-induced crystallization
effects, to the total mechanical response. We will
The mathematical representation is based on briefly address the contribution of the network
the constitutive model presented by Boyce et al. resistance for which the detailed modeling is given
(2000) for the finite deformation stress–strain be- in the work of Boyce et al. (2000).
S. Ahzi et al. / Mechanics of Materials 35 (2003) 1139–1148 1141

2.1. Network stress and flow model can be used to get the pressure term (Arr-
uda and Boyce, 1993; Bergstrom and Boyce, 1998;
The deformation acting on the network re- Anand, 1996).
sistance (see Fig. 1) represents the network orien-
tation process (Arruda and Boyce, 1993) and 2.2. Intermolecular stress and flow
the molecular relaxation process (Bergstrom and
Boyce, 1998). The deformation gradient FB can be Above the glass transition and at a certain level
decomposed into a network orientation part FBN of strain, the deformation of PET can result in
and a flow part FBf . The multiplicative decompo- strain-induced crystallization, which, in turn, in-
sition is therefore given as creases the intermolecular (resistance) barriers to
deformation. The co-existence of the crystalline
FB ¼ FBN FBf : ð3Þ
and the amorphous phases and their contribution
The network stress is determined by the following to the intermolecular response can be treated fol-
stress–stretch relation, which is described by the lowing two different composite models: the upper
Arruda and BoyceÕs eight chains model (1993). bound and lower bound approaches. We assume
pffiffiffiffi " # that the rate of viscous flow in both the crystalline
1 N 1 kN h N i
TB ¼ CR  ‘ pffiffiffiffi B  ðkN Þ2 I ; ð4Þ and the amorphous phases follows the Arhenius
JB kN N type expression proposed by Argon (1973) and
modified by Boyce et al. (2000) for the amorphous
where JB is the network volume change, N is the
flow.
number of rigid links between entanglements and  
CR is the rubbery modulus and ‘1 is the in- DGi ð1  si =si Þ
c_ pi ¼ c_ 0;i exp  ð7Þ
verse Langevin pffiffiffiffifunction given by ‘ðbÞ ¼ cothðbÞ kh
ð1=bÞ‘1 ½kN = N .
The index ‘‘i’’ represents one of the phases, crys-
The stretch on each chain in the network, kN , is
talline (c) or morphous (a). The parameters c_ 0;i ,
the root mean square of the distortional applied
N 1=2 DGi , k and h are, respectively, the reference shear
network stretches kN ¼ ½ð1=3ÞtrðB Þ , where
N N N N 1=3 N rate of phase i, the activation energy of phase i, the
B ¼ F B F B T and F B ¼ ðJB Þ FB . Boltzman constant and the temperature. The re-
The relaxation process is prescribed by Berg- sistances to shearing in the crystalline and amor-
strom and BoyceÕs model (1998) via the stress- phous phases, si , evolve with strain hardening
assisted chain reptation based on the Doi and according to:
Edwards (1986) theory. The rate of relaxation is
taken to be s_ i ¼ hi c_ i ; ð8Þ
  where hi is the hardening rate. The individual
1
c_ FB ¼ C sB ; ð5Þ constituents are assumed to behave as power law
kF  1
hardening type material. The hardening rates can
T 1=2
where kF ¼ ½ð1=3ÞðtrðFBF FBF ÞÞ and C is a para- therefore be expressed in the following form
meter that expresses the temperature dependence. (Stringfellow et al., 1992):
This shear rate is used in the flow rule to get the  n
si si i
flow strain rate Df , which is in turn used to com- hi ¼ ; ð9Þ
pute FBf . ni s i

c_ f where ni is material constants, and si is the effective


Df ¼ pffiffiffi TB0 : ð6Þ shear strength levels (initial critical shear stresses)
2s B
of each pure phase. The effective shear strength
The effective shear stress, sB , is given by sB ¼ level for the crystalline phase, sc , is assumed to be
½ð1=2ÞðTB0 TB0 Þ1=2 . Here, TB0 is the deviatoric part of proportional to the amorphous phase one
TB . We note that relation (4) gives a deviatoric part
of TB but a compressible version of the eight-chain sc ¼ jc sa : ð10Þ
1142 S. Ahzi et al. / Mechanics of Materials 35 (2003) 1139–1148

The effective shear stresses of the two phases, sc 1 c 1 a


and sa , are related to their corresponding devia- TAc ¼ C ½ln FAc e ; and TAa ¼ C ½ln FAa e :
0 0
Jc Ja
toric Cauchy stresses, TAc and TAa , by ð16Þ
 1=2  1=2
1 c0 c0 1 a0 a0
sc ¼ T T ; sa ¼ T T : ð11Þ Here, Jc ¼ det FAc e and Ja ¼ det FAa e are the phase
2 A A 2 A A volume changes. C c and C a are the elastic stiffness
Here, and in what follows, the exponents ÔcÕ and ÔaÕ tensors of the crystalline and amorphous phases,
designate crystalline and amorphous phases. respectively. Since we assume isotropic elasticity,
The phase strain rates Dp c and Dp a are given the components of the elastic stiffness tensors, C c
by the following flow rule: and C a , are determined by the phase bulk moduli,
Kc and Ka , and the phase shear moduli, Gc and Ga .
c_ 0 c_ 0
Dp c ¼ pffifficffi TAc ; Dp a
¼ pffiffiaffi TAa : ð12Þ The plastic stretching of each phase is described
2s c 2s a by the shear rates c_ c and c_ a , which are given by Eq.
(7). These are used in (12) to obtain the plastic
2.2.1. Upper bound model strain rates of the two phases, which in turn are
The analog representation is shown in Fig. 1a in used to compute the plastic part of the corre-
which the intermolecular resistance is composed of sponding deformation gradients.
amorphous phase resistance acting in parallel with
a crystalline phase resistance. The intermolecular 2.2.2. Lower bound model
deformation gradient, FA , is therefore the same for The amorphous and crystalline phase resis-
both phases: tances are taken in series to define the total inter-
molecular resistance as shown in Fig. 1(b). For this
FA ¼ FAc ¼ FAa ; ð13Þ model, the total intermolecular stress (Cauchy
where FAc and FAa are the deformation gradient in stress), TA , is equal to both the stress of the
the crystalline and the amorphous phases, respec- amorphous and that of the crystalline phases:
tively. 1 e
The overall stress–strain response is initially stiff TA ¼ TAc ¼ TAa ¼ Ccom ½lnðF e Þ: ð17Þ
J
due to the intermolecular barriers to deformation e
The effective composite elastic stiffness tensor Ccom
followed by intermolecular flow. This behavior can
is defined by the composite bulk modulus, Kcom ,
be represented by an elastic spring in series with a
and the composite shear modulus, Gcom .
viscous element in each phase. The deformation
gradient can therefore be decomposed into an 1 1 1
¼x þ ð1  xÞ ;
elastic part, FAc e (or FAa e ), and a plastic part, FAc p Kcom Kc Ka
(or FAa p ), via the multiplicative decomposition: 1 1 1
¼ x þ ð1  xÞ : ð18Þ
Gcom Gc Ga
FAc ¼ FAc e FAc p ; FAa ¼ FAa e FAap : ð14Þ
The equivalent viscous element contribution to
Here the exponent c and a designate the crystalline
deformation gives the intermolecular rate of plas-
and amorphous phases; ÔeÕ and ÔpÕ designate the
tic deformation DpA , as function of the phase
elastic and plastic parts, respectively.
plastic strain rates, by the following rule of mix-
The total intermolecular stress is given by the
ture:
rule of mixture:
DpA ¼ xDpA c þ ð1  xÞDpA a : ð19Þ
TA ¼ xTAc þ ð1  xÞTAa ; ð15Þ
Here again, x represents the volume fraction of the
where x is the volume fraction of the crystalline crystalline phase.
phase. The phase stresses TAc and TAa are related to Finally, for the two models described above,
the elastic deformation of the corresponding phase the evolution of the individual constituent shear
by the following constitutive relation: strength levels, si , and the percent crystallinity, x,
S. Ahzi et al. / Mechanics of Materials 35 (2003) 1139–1148 1143

are calculated at each time step using the implicit mental work of Salem (1992, 1994) and of Clauss
Euler approximation: and Salem (1992), a shift in the crystallinity evo-
ðtþDtÞ ðtÞ lution is observed for PET deformed under dif-
si ¼ si þ s_ Dt; ð20Þ
ferent constant strain rates. This is not accounted
for in the expression given above for the rate of
y ðtþDtÞ ¼ y ðtÞ þ y_ Dt; ð21Þ
crystallinity. In order to account for the strain rate
where s_ i is calculated from Eq. (8) and y_ is the rate effect, we phenomenologicaly modified equation
of the degree of transformation described below. (23) to capture a low rate of transformation at low
strain rate and high rate of transformation at high
strain rate.
3. Rate of crystallization e_ ðm1Þ=m
y_ ¼ mKav ðhÞ½ lnð1  yÞ ð1  yÞ
e_ref
If we denote by /1 the maximum degree of  
crystallinity, the total percent of crystallinity x is tr T
 exp n ð25Þ
given by G
x ¼ /1 y; ð22Þ Here, the reference strain rate, e_ref , is taken as the
maximum strain rate for which experimental re-
where y is the total degree of transformation. The
sults are available for the calibration of the model
rate of transformation is expressed as function of
parameters. This expression can be generalized by
the developed stress T and the shear moduli G,
choosing an appropriate function of strain rate
following a phenomenological expression derived
and temperature instead of the strain rate ratio
by Doufas and co-workers (Doufas et al., 2000;
used in (25).
Dairanieh et al., 1998; Doufas et al., 1999), which
is based on the non-isothermal version of the
Avrani equation.
4. Results and comparison with experiments
dy
y_ ¼
dt   In the first part of this section, we show results
ðm1Þ=m tr T from our parametric study. In the second part, we
¼ mKav ðhÞ½ lnð1  yÞ ð1  yÞ exp n ;
G compare our predicted results to those of the ex-
ð23Þ perimental uniaxial drawing tests of Salem (1992,
1994) and of Clauss and Salem (1992). Both the
where m is the Avrami exponent, n is a dimen- stress–strain curves and the evolution of the degree
sionless model parameter, h is the test temperature of crystallinity are presented. We also compare our
and Kav ðhÞ is the transformation rate function, predicted stress–strain response to the experimen-
which is defined in case of poly(ethylene tere- tal uniaxial compression results of Arruda and
phthalate) as follows: Boyce (1993). These uniaxial tension or compres-
 1=3 "  2 #
h  141 sion results are conducted under constant strain
3 4pNu
Kav ¼ 1:47  10 exp  rate. The considered strain rates are 0.42 and 0.01
3/1 47:33
s1 in the case of uniaxial drawing, 0.05 and 0.5 s1
ðs1 ; h in °CÞ: ð24Þ in the case of uniaxial compression, and the ref-
erence strain rate, e_ref , used in Eq. (25) is taken as
Here, Nu is the number density of nuclei initially
0.42 s1 .
present within the amorphous phase.
The PET under deformation shows a low rate
of crystallization which can be simulated using 4.1. Parametric studies
Eq. (22) together with low n values. In this case
the exponential term becomes negligible and the In what follows, we study the effect of different
model loses the effect of triaxiality. In the experi- parameters such as the triaxiallity coefficient n,
1144 S. Ahzi et al. / Mechanics of Materials 35 (2003) 1139–1148

Table 1
Parameter Equation Value
Model and material parameters
/1 (22) and (24) 0.3
Nu (24) 108
m (23) and (25) 3
n (15) 0.0
c_ 0;a ¼ c_0;c (s1 ) (7) 1.6  109
sa (MPa) (9) 4.5
jc (10) 6
nc (9) 0.1
na (9) 5.5
Kf (MPa s)1 (5) 0.6  103
DGa ¼ DGc (J) (7) 1.6  1019
Processing conditions
Temperature, h (°C) 90
Applied strain rate (s1 ) 0. 42
Physical properties of PET
Young modulus, E (MPa) – 18
Poison ratio, t – 0.49
Rubber modulus, CR (MPa) (4) 0.7
Number of rigid chain links, N (4) 9

strain rate and temperature. The model parame- 4.1.2. Strain rate e_
ters as well as the PET physical properties and the Fig. 3(a) and (b) show the effect of strain rate
processing conditions, used to fit the hot drawing on the mechanical response and the evolution
experiments, are shown in Table 1. These studies crystallinity as function of the true strain under
are conducted under uniaxial tension. axisymmetric tension. The stress–strain and the
crystallinity curves are fitted to experimental data
4.1.1. Parameter n of Salem (1992) at a temperature h ¼ 90 °C and a
The effect of triaxiality on the mechanical re- strain rate e_ ¼ 0:42 s1 . This strain rate is used as
sponse and the evolution of the percent crystal- reference, e_ref , together with the same model pa-
linity is controlled by the parameter n as illustrated rameters to produce the stress–strain and the
in Fig. 2(a) and (b) for axisymmetric tension under crystallinity curves at the lower strain rate e_ ¼
a constant strain rate of 0.42 s1 and for a test 0:01 s1 .
temperature h ¼ 90 °C. As expected, the increase
of n enhances the crystallization rate (Fig. 2(b)), 4.1.3. Temperature h
which in turn, together with the molecular orien- In Fig. 4(a) and (b) we present the effect of
tation effect increases the hardening observed on temperature on the mechanical response and the
the stress–strain curves (Fig. 2(a)). At low values percent crystallinity of PET under axisymmetric
of n this effect becomes negligible. This is due to tension. At the temperature h ¼ 90 °C and the
the fact that the exponential term in Eq. (23) or strain rate e_ ¼ 0:1 s1 , the stress–strain and the
(25) becomes close to unity. Since the suggested crystallinity curves are fitted to the experimental
value of n for PET is very small (Doufas et al., data of Salem then the same parameters are used
2000), we therefore set n ¼ 0:0 in what follows. for the response at h ¼ 96 °C. The figures show
This renders the crystallinity evolution indepen- that the experimentally measured effect of temper-
dent of the stress and thus independent of the se- ature is captured by the proposed model. How-
lected composite model (upper or lower bound). ever, the starting points of the crystallinity for
S. Ahzi et al. / Mechanics of Materials 35 (2003) 1139–1148 1145

40 0.35
lower ξ=0.01
lower ξ=0.01
upper ξ=0.01 0.30
upper ξ=0.01
lower ξ=0.1
30 lower ξ=0.1
upper ξ=0.1 0.25 upper ξ=0.1
True Stress (MPa)

lower ξ=1
lower ξ=1

Crystallinity
upper ξ=1
0.20 upper ξ=1
20
0.15

0.10
10

0.05

0 0.00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2 .0

(a) True Strain (b) True Strain

Fig. 2. (a) Effect of the parameter n on the stress–strain behavior of PET at temperature h ¼ 90 °C and strain rate e_ ¼ 0:42 s1 under
uniaxial tension and (b) effect of the parameter n on the crystallinity profile of PET at temperature h ¼ 90 °C and strain rate e_ ¼ 0:42
s1 under uniaxial tension.

25 0.35

-1
lower 0.01 s 0.30 0.01 s-1
20 upper 0.01 s-1 0.42 s-1
lower 0.42 s-1 0.25
True Stress (MPa)

upper 0.42 s-1


Crystallinity

15
0.20

0.15
10

0.10

5
0.05

0 0.00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
(a) True Strain (b) True Strain

Fig. 3. (a) Effect of strain rate, e_ on the stress–strain behavior of PET at temperature h ¼ 90 °C and n ¼ 0:0 under uniaxial tension and
(b) effect of strain rate, e_ on the crystallinity profile of PET at temperature h ¼ 90 °C and n ¼ 0:0 under uniaxial tension.

0.30

50
lower θ=96oC 0.25 θ=96oC
upper θ=96oC θ=90oC
40
lower θ=90oC
True Stress (MPa)

0.20
upper θ=90oC
Crystallinity

30
0.15

20
0.10

10 0.05

0 0.00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
(a) True Strain (b) True Strain

Fig. 4. (a) Effect of temperature on the stress–strain behavior of PET at strain rate e_ ¼ 0:1 s1 and n ¼ 0:0 under uniaxial tension and
(b) effect of temperature on the crystallinity profile of PET at strain rate e_ ¼ 0:1 s1 and n ¼ 0:0 under uniaxial tension.
1146 S. Ahzi et al. / Mechanics of Materials 35 (2003) 1139–1148

both temperatures are chosen separately to fit the which decrease the stress. Beyond these deforma-
experimental results. This shows the limitation of tion values, the macromolecular network reaches
our crystallization model to correctly account for the preferential orientation which results in strain-
the shift due to temperature. induced crystallization and the disappearance of
the molecular relaxation. At very large deforma-
4.2. Comparisons with the experimental results tion a dramatic hardening takes place due to the
increase of the elastic barrier to deformation with
In this section, we compare our predicted re- the increase of both the molecular orientation and
sults for hot drawing of PET to the experimental the crystalline volume fraction.
results in the literature. We first consider the uni- The proposed models (bounds) predict different
axial tension and compare the predicted stress values for the stress–strain response as shown in
strain curves and the evolution of crystallinity to Fig. 5(a). The upper bound stress–strain curve is
the experimental results of Salem (1992, 1994). higher relative to that predicted by the lower
Then we consider uniaxial compression and com- bound model. This is expected since the upper
pare the predicted stress–strain curves to the ex- bound model is stiffer. For the comparison with
perimental results of Llana and Boyce (1999). The the experimental results, Fig. 5 shows that: (1) a
experimental results for the evolution of crystal- good agreement is obtained before the crystalli-
linity are not available for the compression test. zation; (2) the experimental results are closer to the
lower bound model at the beginning of crystalli-
4.2.1. Uniaxial tension at a constant strain rate zation and tend to join the upper bound prediction
Here, as in the work of Salem (1992) we con- at large strains (higher crystallinity). In Fig. 5(b),
sider two strain rates e_ ¼ 0:42 and 0.01 s1 at we show the evolution of crystallinity with strain
temperature h ¼ 90 °C. The model an material for both strain rates. This comparison shows a fair
parameters are listed in Table 1. The experimental agreement between our crystallinity model pre-
results show that up to a total equivalent strain dictions and the experimental results.
eeq ¼ 0:6 and 0.9 for the mechanical response at
strain rates e_ ¼ 0:42 and 0.01 s1 , respectively, the 4.2.2. Uniaxial compression at a constant strain rate
PET samples are fully amorphous. The steady in- To compare to the experimental results of
crease of stress with strain is due to an antagonist Llana and Boyce (1999) we consider the axisym-
effect between the network orientation, which in- metric compression tests at two different strain
creases the stress, and the molecular relaxation, rates e_ ¼ 0:5 and 0.05 s1 , and at a temperature

50 0.35
expirement 0.42s-1
lower 0.42-1 0.30 experiment 0.01 s-1
40 upper 0.42s-1 calculated 0.01 s-1
expirement 0.01s-1 0.25 experiment 0.42 s-1
True Stress (MPa)

lower 0.01s-1
Crystallinity

calculated 0.42 s-1


30 upper 0.01s-1
0.20

0.15
20

0.10

10
0.05

0 0.00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
(a) True Strain (b) True Strain

Fig. 5. (a) Comparison of model predictions and experimental data for the stress–strain behavior of PET at temperature h ¼ 90 °C and
strain rate e_ ¼ 0:42 s1 and e_ ¼ 0:01 s1 under uniaxial tension and (b) comparison of model predictions and experimental data for the
crystallinity profile of PET at temperature h ¼ 90°C and strain rate e_ ¼ 0:42 s1 and e_ ¼ 0:01 s1 under uniaxial tension.
S. Ahzi et al. / Mechanics of Materials 35 (2003) 1139–1148 1147

30 0.35

experiment 0.05 s-1


0.30
25 lower 0.05 s-1
0.5 s-1
upper 0.05 s-1
0.05 s-1
True Stress (MPa)

20 experiment 0.5 s-1 0.25

lower 0.5 s-1

Crystallinity
upper 0.5 s-1 0.20
15
0.15

10
0.10

5
0.05

0 0.00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8
(a) True Strain (b) True Strain

Fig. 6. (a) Comparison of model predictions and experimental data for the stress–strain behavior of PET at temperature h ¼ 90 °C and
strain rate e_ ¼ 0:5 s1 and e_ ¼ 0:05 s1 under uniaxial compression and (b) model predictions for the crystallinity profile of PET at
temperature h ¼ 90 °C and strain rate e_ ¼ 0:5 s1 and e_ ¼ 0:05 s1 under uniaxial compression.

h ¼ 90 °C. Since the samples used by these authors account for the strain rate effect on the transfor-
are different from those used in the work of Salem, mation rate. The results of our simulation for the
the values of some material properties are changed mechanical response and the evolution of the
from those given in Table 1. These are E ¼ 28 percent crystallinity under tension are compared to
MPa, CR ¼ 2:1 MPa and sa ¼ 1:02 MPa. Fig. 6a the experimental observations and fair agreement
shows the simulated stress–strain curves in com- is found.
parison with the experimental results. Again here, Modification of the rate of transformation to
for both strain rates, one can see that a good account for both the strain rate and temperature
agreement is found before crystallization starts, effects on crystallization is needed. In addition, the
then the experimental results lie between the upper effects of orientation of the crystalline phase are
and lower results. Fig. 6(b) shows the predicted neglected in the present work. These effects must
evolution of crystallinity for both tests. Experi- be included to generalize the proposed modeling
mental results for crystallinity are not available for approach.
these tests.

Acknowledgements
5. Conclusion
This work was supported primarily by the ERC
We have successfully developed and imple- Program of the National Science Foundation
mented a constitutive elastic–viscoplastic model under Award Number EEC-9731680 with Clem-
for the finite deformation stress–strain behavior of son University.
poly(ethylene terephthalate) at temperature above
the glass transition. The contribution to the in-
termolecular barrier to deformation is accounted References
for by assuming the crystalline phase and the
amorphous phase arranged in two different analog Adams, A.M., Buckley, C.P., Jones, D.P., 1994. Biaxial, hot
models: upper bound model and lower bound drawing of amorphous poly(ethylene terephthalate). In:
model. The crystallization formulation, which is Proceedings of the Ninth International Conference in
Deformation, Yield and Fracture of Polymers, Cambridge,
based on the non-isothermal version of Avrami UK.
equation, is adopted to describe the rate of crys- Argon, A.S., 1973. A theory for the low-temperature plastic
tallization. A simple modification is proposed to deformation of glassy polymers. Philos. Mag. 28, 839–865.
1148 S. Ahzi et al. / Mechanics of Materials 35 (2003) 1139–1148

Arruda, E.M., Boyce, M.C., 1993. A three-dimensional con- Model development and predictions. J. Non-Newtonian
stitutive model for the large stretch behavior of rubber– Fluid Mech. 92, 27–66.
elastic materials. J. Mech. Phys. Solids 41, 398–412. Gorlier, E., Haudin, J.M., Billon, N., 2001. Strain-induced
Anand, L., 1996. A constitutive model for compressible crystallization in bulk amorphous PET under uni-axial
elastomeric solids. Computat. Mech. 18, 339. loading. Polymer 42, 9541–9549.
Bergstrom, J.S., Boyce, M.C., 1998. Constitutive modeling GÕSell, C., Cabot, C., Hiver, J.M., Vigny, M., Aubert, A., 1999.
of the large strain time dependent behavior of elastomers. Influence of induced crystallization on strain hardening
J. Mech. Phys. Solids 46, 931–957. when streching poly(ethylene terephthalate). In: Proceedings
Boyce, M.C., Socrate, S., Llana, P.G., 2000. Constitutive model of the International Conference of the Polymer Processing
for the finite deformation stress–strain behavior of poly(eth- Society, Bangkok, Thailand, 1–3 December.
ylene terephthalate) above the glass transition. Polymer 41, GÕSell, C., Hiver, J.M., Elkoun, S., Vigny, M., Cabot, C.,
2183–2201. Aubert, A., Dahoun, A., 2000. In: Proceedings of the 11th
Buckley, C.P., Jones, D.C., 1995. Glass–rubber constitutive International Conference On Deformation Yield and Frac-
model for amorphous polymers near the glass transition. ture Of Polymers, Cambridge, UK, pp. 375–378.
Polymer 36, 3301–3312. Llana, P.G., Boyce, M.C., 1999. Finite strain behavior of PET
Buckley, P.C., Jones, D.C., Jones, D.P., 1996. Hot-drawing of above the glass transition temperature. Polymer 40, 6729–
poly(ethylene terephthalate) under biaxial stress: applica- 6751.
tion of a three-dimensional glass–rubber constitutive model. Matthews, R.G., Duckett, R.A., Word, I.M., Jones, P.D., 1997.
Polymer 37, 2403–2414. The biaxial drawing behavior of poly(ethylene terephtha-
Chandran, P., Jabarin, S., 1993. Biaxial orientation of poly(eth- late). Polymer 38, 4795–4802.
ylene terephthalate), part I: nature of the stress–strain Salem, D.R., 1992. Development of crystalline order during
curves. Adv. Polym. Technol. 12, 119–132. hot-drawing of PET film: influence of strain rate. Polymer
Clauss, B., Salem, D.R., 1992. Characterization of the non- 33, 3182–3188.
crystalline phase of oriented poly(ethylene terephthalate) by Salem, D.R., 1994. Crystallization during hot-drawing of
chain-intrinsic fluorescence. Polymer 33, 3193–3202. poly(ethylene terephthalate) film: influence of temperature
Dairanieh, I.S., McHugh, A.J., Doufas, A.K., 1998. A phe- on strain-rate/draw-time superposition. Polymer 35, 771–
nomenological model for flow-induced crystallization. Proc. 776.
SPE/ANTEC Conf. 56, 212–216. Stringfellow, R.G., Parks, D.M., Olson, G.B., 1992. A constit-
Doi, M., Edwards, M.F., 1986. The Theory of Polymer utive model for transformation plasticity accompanying
Dynamics. Oxford University Press, Oxford. strain-induced martensitic transformations in metastable
Doufas, A.K., Dairanieh, I.S., McHugh, A.J., 1999. A contin- austenitic steels. Acta. Metall. Mater. 40, 1703–1716.
uum model for flow-induced crystallization of polymer Vigny, M., Aubert, A., Hiver, J.M., Aboulfaraj, M., GÕSell, C.,
melts. J. Rheol. 43, 85–109. 1999. Constitutive viscoplastic behavior of amorphous PET
Doufas, A.K., McHugh, A.J., Miller, C., 2000. Simulation of during plane-strain tensile stretching. Polym. Eng. Sci. 39,
melt spinning including flow-induced crystallization Part I. 2366–2376.

You might also like