Strain-Induced Crystallization Behavior of Poly(Ether
Ether Ketone) (PEEK)
MING C. CHIEN' and R. A. WEISS'
Polymer S c i e n c e Program and D e p t . of C h e m i c a l Engineering
U n i v e r s i t y of Connecticut
Storrs, Connecticut 06268
Strain-induced crystallization of poly(etherether ketone)
(PEEK)was studied by the use of a parallel plate rheometer.
The experimental variables included preheating time,
crystallization temperature, and shear rate. The crystallization kinetics were characterized by means of a n induction time, defined a s the time elapsed from the start of
shearing to the onset of crystallization. The experimental
results showed that the induction time for strain-induced
crystallization of PEEK was significantly shorter than that
for crystallization under quiescent condition, and that
strain-induced crystallization was much less temperature
dependent than quiescent crystallization. The activation
energy for strain-induced crystallization was found to be
0.035 kcal/mole. which was considerably smaller than the
reported activation energy for quiescent crystallization.
Photomicrographs of the sheared specimens indicated that
PEEK crystallites orient along the flow direction.
INTRODUCTION
train-induced crystallization of polymers is
a commonly observed phenomenon associated with polymer processing operations. The
understanding of this subject is therefore of
practical importance, in order to achieve the
optimum conditions for processing a n d to determine processing-structure-property relationships. Unlike crystallization under quiescent
conditions, strain-induced crystallization of
polymers is not very well understood. Nevertheless, research on this subject h a s been actively
pursued. Experiments can be generally categorized into four basic areas: stretching amorphous polymers in the glassy state, stretching
cross-linked polymers in the rubbery state, stirring dilute polymer solutions, and shearing linear polymer melts (1).
In general, the crystallization kinetics a n d the
crystalline textures formed are different when
the material is deformed during crystallization
than when crystallized under quiescent conditions. Chain extension and molecular alignment that may occur during deformation can
increase the nucleation rate of crystallites a n d
affect the crystallite morphology as well. This,
initurn, may have a significant impact on the
final polymer properties.
' C urrcnr addrrss John Brown E Rr C lrlc I 7 Amelia Place P 0 Box 1432
Poly(ether ether ketone), PEEK,
S
Stdmlord C7 06904
To \lilioni c o r r e s p n n d r n r t should he addressrd
6
a relatively new high-performance thermoplastic, is being evaluated for a variety of potential
applications. Because of its high melting point
and excellent solvent resistance, PEEK is very
attractive to the various industries that require
thermoplastics for use a t high temperature or
in corrosive environments.
In the past few years, the crystallization behavior of PEEK has been of interest because of
the importance of crystallization with respect
to material properties. Various experiments on
the crystallization of PEEK have been reported
(2-7). Basically, these experiments were all carried out under quiescent conditions; that is,
with no application of shear or extensional
forces during crystallization.
Rlundell and Osborn (2)studied the crystallization and melting behavior of PEEK and found
that it was closely analogous to poly(ethy1ene
terephthalate), PET, in its crystallization behavior exceDt that the main transitions occurred abouf 75°C higher. Based on differential
scanning calorimetry (DSC) isothermal experi-
POLYMER ENGINEERING AND SCIENCE, MID-JANUARY, 7988, VOI. 28, NO. 7
Strain-InducedCrystallization Behavior of PoZy[Ether Ether Ketone) [PEEK)
ments, they reported a maximum rate of crystallization of PEEK a t around 230°C. This observation was later confirmed by Shukla and
Sichina (3)by studies of the crystallization of
PEEK in PEEK/graphite fiber composites.
Cebe and Hong (7)studied the crystallization
behavior of PEEK under isothermal a n d nonisothermal conditions. Their experimental results during isothermal crystallization from the
melt and from the rubbery amorphous state
were fit with the Avrami equation. The Avrami
exponent was calculated as about 3, which corresponds to heterogeneous nucleation and
three-dimensional spherulitic growth. Based on
nonisothermal experiments, the activation energy was found to be 68 kcal/mole for crystallization from the melt and 5 2 kcal/mole from
the rubbery amorphous state. These values
were considerably smaller t h a n the 322 kcal/
mole reported by Kemmish a n d Hay (8)for isothermal crystallization. Cebe and Hong claimed
that their values of activation energies for PEEK
were more reasonable since they were comparable to those of PET for both isothermal and
nonisothermal crystallizations.
The objective of the study reported herein was
to determine if and how strain would perturb
the crystallization behavior a n d the morphology
of PEEK.
EXPERIMENTAL
Experimental Material
Extruded PEEK film, 0.25-mm thick, was
supplied by the Advanced Systems Division of
Northrop Corporation. It was used as received
for both quiescent and strain-induced crystallization studies.
Isothermal Crystallization
Isothermal crystallization experiments were
run to determine the experimental conditions
for the strain-induced crystallization experiments. All the thermal measurements were
made with a Perkin Elmer differential scanning
calorimeter (DSC), Model 2, equipped with a
mechanical cooler and a Model 3600 data station. During the experiments, the DSC cell was
purged with dry nitrogen gas.
Isothermal crystallization studies of PEEK
were made between 157°C a n d 167°C a n d between 297°C and 327°C. Samples of approximately 1 0 mg were cut from the as-received
extruded PEEK film a n d were crimped inside
aluminum sample pans. For the studies in the
lower temperature range, the sample was
heated a t about 320"C/min from just below T g .
145"C, to the desired crystallization temperature, and the differential heat flux referenced
to a n empty pan was monitored as a function
of time. For the studies in the higher temperature range, the sample was first heated to about
377°C and held there for 5 min to allow the
PEEK to melt completely. The sample was then
quickly cooled, with a cooling rate of about
2OO"C/min, to the crystallization temperature
and the rate of crystallization was followed as
before. The time a t which the crystallization
exotherm reached its maximum value was
taken by convention as the crystallization halftime, &.
Strain-induced Crystallization
Strain-induced crystallization was measured
with Rheometrics System 4 Mechanical Spectrometer using a parallel plate geometry. A 1 to
2-mm thick sample was placed between a pair
of 25-mm diameter machined aluminum plates
in the convection oven chamber. The sample
was first heated to 375°C for a period of time
sufficient to achieve a homogeneous melt. The
oven was then programmed to cool rapidly in
two steps to the test temperature that was chosen to be below the melting point of PEEK, but
high enough so that normal isothermal crystallization would not occur in the time frame of
the experiment. First, the temperature was lowered to 340°C near the melting point of PEEK.
Once the instrument equilibrated here, the temperature was further cooled to the test temperature, which took about 5 min. The temperature
of the sample was monitored by a thermocouple
in contact with the upper plate of the fixture.
Once the sample was stabilized a t the desired
temperature, the upper plate was set to rotate
a t a fixed speed. The shear stress and the normal stress of the sample were monitored by
2000 g-cm transducers mounted on the lower
plate. These stresses, which are very sensitive
to the presence of small amounts of crystallites.
were used to determine the onset of crystallization. The crystallization induction time was
defined as the time elapsed from the start of
shearing to the onset of crystallization as manifested by a n increase in either the shear stress
or the normal stress, as shown schematically in
Fig. 1.
The three experimental variables studied in
the strain-induced crystallization experiments
Induction T i m e
Start of
Shaaring
Time
Fig. 1 . Schematic f o r calculating induction timef o r crystallization under strain.
POLYMER €NG/N€ER/NG AND SC/€NC€, MID-JANUARY, 1988, Yo/. 28, No. 7
7
Ming C. Chien and R. A. Weiss
were: ( 1 ) preheating time in the melt, (2) the
crystallization temperature, a n d (3) the shear
rate.
A number of sheared specimens were microtomed into thin slices of about 50 pm. The
morphology was observed with a Nikon Biophot
polarizing optical microscope, and photomicrographs were taken with a Nikon Microflex AFX
camera.
Thermal analysis of shear-crystallized PEEK
samples was conducted with the DSC using a
heating rate of 40"Clmin. Two r u n s were done
for each sample. After the first heating, each
sample was cooled at a rate of 200"C/min back
to room temperature and heated a second time.
The second heating scan should have the characteristics of a quiescently crystallized sample.
Crystallinity and onset temperature of melting
were measured for the two scans to evaluate
the effect of strain.
RESULTS AND DISCUSSION
Isothermal Crystallization
The as-received PEEK was amorphous as confirmed by DSC. The DSC thermogram obtained
on the initial heating of the sample showed a Tg
at 145°C followed by a crystallization exotherm
at 179°C and finally a melting transition a t
339°C (Fig. 2 ) .After the first heating, the sample was cooled back to room temperature a t
200"C/min for the second heating. The second
heating thermogram showed a broad diffuse T g
near 147°C and no crystallization exotherm. A
melting transition was again observed near
339°C indicating that the sample was semicrystalline. This result was expected and simply
serves to demonstrate that if the amorphous
film is heated sufficiently above its T g ,crystallization will occur.
A set of isothermal crystallization experiments were made to determine the rate of crystallization at different temperatures and to provide a reference for the strain-induced crystallization studies. The measured crystallization
half -times are plotted vs. crystallization ternperature in Fig. 3. The results are consistent
with those reported by other researchers (2, 3)
and indicate a maximum crystallization rate in
the vicinity of about 230°C. It should be noted
that crystallization occurs very slowly above
300°C.
Strain-inducedCrystallization
Strain-induced crystallization was studied a t
three temperatures: 330"C, 332°C. and 335°C.
Based on the isothermal crystallization experimental results, quiescent crystallization was
not expected to occur a t these temperatures in
the time frame of the strain-induced crystallization experiments. The strain-induced crystallization experiment was also attempted a t
340°C which is just above the melting point,
but no crystallization was observed.
Figure 4 shows some representative data collected for the strain-induced crystallization experiments. At a constant shear rate, both the
shear stress a n d normal stress are functions of
time. The initial transient peaks in the data a t
short times are due to the rheological response
of the polymer melt a n d are unrelated to crystallization. If no crystallization took place, the
stress would eventually reach a steady state
value. However, because crystallization occurs,
the stress begins to increase a t a certain point.
For both stresses, the onsets of stress increase
coincide. Therefore, either shear stress or nor-
-L
"
100
,,,/
10.
t
-I"
c
I'
D
Y
01
001
I
I
0 001'
'
I30
I70
210
250
lsorhernol C - y s t a l I ~ z o r ~ o Tne m p a r o t u r e
290
3.
(C)
Fig. 3. Crystallization hay-time us. temperature.
0.2 Ilssc 8 330.C
Fig. 4 . Shear stress and normal stress at 330°C us. time
for a shear rate of 0.2 s - I .
8
POLYMER ENGINEERING AND SCIENCE, MID-JANUARY, 7988, Vol. 28, NO. I
Stra in-1nduced Crystal1izat ion Behavior of Pol y(E t her Ether Ketone) (PEEK)
ma1 stress could be used to determine the onset
of crystallization of the polymer.
To prove that the stress increases discussed
above were indeed due to the strain-induced
crystallization of the polymer, one PEEK specimen was tested using oscillatory shear with a
small strain amplitude (a minimum strain was
used only to keep the torque at a detectable
level). Figure 5 shows the storage modulus, G',
and loss modulus, G", measured with a strain
of 0.5%and a frequency of 1 Hz a t 330°C. The
moduli did not change during a n experimental
time of 20 min, which indicated that no crystallization occurred. This result confirms that
the stress increases observed in the steady
shear experiments are truly related to the crystallization of the polymer. It also proves that the
crystallization observed in the steady shear experiments is due to strain a n d not due to any
instrumental characteristics.
For all the steady shear experiments, the
PEEK sample was heated to 375°C about 36°C
higher than the melting point of PEEK, to destroy any existing nuclei in the sample. The
preheating time in the melt was varied from 1 5
to 35 min to determine any possible effect of
this variable. No obvious differences were observed in the results with respect to the onset
of crystallization. Thus, within the preheating
time range studied, either no nuclei persist or if
they do, they have no influence on the crystallization behavior during shearing. A preheating
time of 30 min was then chosen for all subsequent steady shear experiments.
The effect of shear rate is shown in Fig. 6. At
a constant temperature, increasing the shear
rate decreased the crystallization induction
time. For example, a t 330°C the induction time
for a shear rate of 0.5 s-' was about 300 s,
while for a shear rate of 0.2 s-' it was about
500 s. In addition, the rate of stress increase
associated with the higher shear rate appeared
greater, implying a faster rate of crystallization.
Figure 7 shows that the experimental temperature also affects the crystallization kinetics. The induction time was shorter for the
higher degree of supercooling. For example, this
was about 500 s a t 300°C vs. about 800 s at
335°C for a constant shear rate of 0.2 s - I .
To achieve good confidence in the data, it was
necessary to make several runs for each shear
rate since the dimensional changes of the specimens due to crystallization introduce some uncertainty in the data. Qualitatively, the reproducibility of the experiments appeared reasonably good; see, for example, Fig. 8,which shows
I .
0s-o.
TIMELSEC. >
O.OE-00
I.X.0.
Fig. 6. Effect of s h e a r rate on t h e s h e a r s t r e s s a t 330°C
after coolingfrom t h e melt.
1
1
I 330-c
0.2 I/.=
,.oE*o.j-0. O E I O O
,
;
I
TIYELSEC.)
7---A
Fig. 7. Effect of temperature O R the s h e a r s t r e s s after
coolingfrom t h e melt. The s h e a r r a t e is 0.2 s-'.
I'
oSHEAR RATE OF 0. 2
..1 . . . . . . . . . . . . . . . *. .. .. .. . ,
.
.
.
.
.
.
.
.
.
.
*
.
A
I
I-+
T----A
0. O E I M
9.w-0.
l/S I 33%
i
z. OL.0,
Fig. 5. Dynamic moduli of PEEK a t 330°C as afunction
of time after coolingfrom the melt.
Y T - - - - - - T I
I
0.w-00
TIYELYIN.)
T I M E LSEC.
>
a. OItDl
Fig. 8. Four experimental r u n s of s h e a r s t r e s s us. time
a t 335°C a n d a s h e a r r a t e of 0.2 s-'.
POLYMER ENGINEERING AND SCIENCE, MIDJANUARV, 1988, Vol. 28, No. 1
9
Ming C.Chien and R. A. Weiss
four different experimental runs using a shear
rate of 0.2 s-' a t 335°C. To assess the quantitative reproducibility, a curve smoothing technique was used as an objective procedure for
determining the induction time. To define the
error range of the data, all the experimental
results were treated by a statistical analysis.
Tables 1 , 2, and 3 show the variation of three
sets of data based on statistical 95%confidence
limits.
Figure 9 gives the induction time vs. shear
rate for the three experimental temperatures.
Three experimental curves were used to generate each data point. A s a general observation,
the induction time decreases as shear rate inTable 1. Induction Time at 33OOC with 95% Confidence Limits.
ShearRate
Induction Time (s)
~
(VS)
Shear Stress
Normal Stress
0.05
0.08
0.2
0.5
0.8
974 (84)
792 (122)
455 (68)
311 (127)
203 (32)
1033 (139)
830 (68)
533 (174)
225 (68)
188 (90)
Table 2. Induction Time at 332OC with 95% Confidence Limits.
~~~~~~~~
Shear Rate
Ills)
0.05
0.08
0.2
0.5
0.8
~
induction Time (s)
- - -- --- - ---
--- --
Shear Stress
Normal Stress
1545 (315)
1226 (840)
783 (121)
420
1513 (266)
1173 (410)
720 (185)
355
214 (78)
Table 3. Induction Time at 335OC with 95% Confidence Limits.
Shear Rate
(11s)
0.05
0.08
0.2
0.5
0.8
1.o
.01
Induction Time (s)
~~ _ _ _ _ ~ - - ~
Normal Stress
Shear Stress
21 19 (552)
1422 (483)
898 (109)
595 (145)
1821 (586)
1422 (352)
883 (167)
422 (83)
268 (30)
203 (24)
creases a n d as temperature decreases. I t is important to recognize the significance of strain
with respect to the crystallization kinetics as
shown in this plot. Based on the previous isothermal crystallization experiments, under
quiescent conditions a t a temperature of 330°C
it would take on the order of several thousand
seconds for crystallization to occur. However,
under the effect of strain, the induction time
for crystallization was as small as 200 s for 0.8
s-' and about 1000 s for 0.08 s - I . At 3 3 5 ° C .
where the normal isothermal crystallization
would take days, the induction time for straininduced crystallization was still on a relatively
convenient time scale for these experiments:
300 s for 0.8 s-' a n d 2000 s for 0.08 s-'.
On the whole, the induction time decreased
as the shear rate increased. Shearing a polymer
melt causes the molecules to become elongated
and thus enhances the formation of crystals
aligned in the flow direction. These crystals
may serve as additional nucleating sites for
other crystallites. A s the shear rate increases,
the rate of formation of these primary nuclei
also increases. The number of primary nuclei
due to this shearing effect is much higher than
in a quiescent melt. A s a result, strain-induced
crystallization shows a much shorter induction
time than does quiescent crystallization.
Another important phenomenon observed in
Fig. 9 is the decrease in temperature dependence of crystallization as the shear rate increases. To show this in a clearer way, the data
are replotted in Fig. 10 as induction time vs.
temperature for the different shear rates. At the
higher shear rate, 0.8 s-', the induction time is
almost independent of temperature. Apparently, the effect of strain is the primary cause
for crystallization as previously discussed since
the quiescent isothermal crystallization at the
temperatures studied here would take a n extremely long time. The effect of supercooling is
diminished when the shear rate is increased.
Yeh, et al. (9) proposed that the induction
time for crystallization, ti,can be related to the
38
. I
Sheor R o t e (l/sec)
Fig. 9. Induction time us. shear rate a t three temperatures.
10
Fig. 10. Induction time us. temperature for three shear
rates.
POLYMER ENGINEERING AND SCIENCE, MID-JANUARY, 1988, Yo/. 28, NO. 7
Strain-Induced Crystallization Behavior of Poly(Ether Ether Ketone) (PEEK)
crystallization temperature, T,by the following
empirical equation:
l / t i = A exp [ E I R T )
where A is a pre-exponential factor, R is the gas
constant, and E is the activation energy for
crystallization. It was suggested that l / t , is a
measure of the nucleation rate since the induction time can be viewed as the time required to
generate a critical number of nuclei per unit
volume, causing a n observable increase in the
stress. According to this equation, the activation energy could be calculated through a plot
of In (to vs. 1/T in which the slope is (- E/R).
From this analysis, the activation energy for
strain-induced crystallization of PEEK was
found to be relatively independent of strain rate
over the range studied and was calculated to be
about 0.035 kcal/mole based on the average of
the three slopes, Fig. 1 1 . This value is significantly smaller than those reported for the crystallization of PEEK under quiescent conditions
by Cebe and Hong (7)and by Kemmish and Hay
(8).This result, however, was not unexpected
because the application of strain to a polymer
network would greatly facilitate the initiation
of nuclei.
A number of samples taken from the center
of each strain-crystallized specimen were analyzed by DSC with a heating rate of 40"C/min.
The crystallinity of PEEK was calculated using
the reported heat of fusion of 130 J/g (10).
It
was expected that the crystallinity of the samples would increase with increasing shearing
time and that the melting temperatures for the
first heating scans would be higher. However,
no significant difference was observed in the
data. The crystallinity in all samples was about
25% to 28%, and the melting temperatures for
both heating scans were practically the same.
Hong reports that a method to restrain the melt
from internal relaxation during melting is considered necessary to observe the elevation of
the melting temperature that corresponds to the
shearing effect [ 1).
Figure 12 shows photomicrographs of the
i
51
i
4
L-
1. 6 4
1.644
6' 6 1-. 6 4 8
1.652
1.656
lOOO/TEHPERATURE (HI
Fig. 1 1 , Log induction time us. reciprocal temperaturefor
three shear rates.
F i g . 12. Photomicrographs of sheared samples: (a)alignment ofcrystallites with shearing direction: @) disturbed
f l o w near the edge of the parallel plates; and (c) microvoids due to changes in the sample dimensions resulting
from crystallization and deformation.
sheared samples taken with a n optical microscope. The morphology of the specimens is a
mixture of crystalline and amorphous materials. It is difficult to determine the actual size
of the crystallites since they are too fine for the
resolving power of the microscope. However, it
appears that the crystallites are aligned with
the shearing direction [the horizontal axis in
these photos). Because of the relatively high
shear rate near the edge of the parallel plates,
the flow there appears more disturbed (Fig.
1%). Some microvoids are observed as a result
of the change in sample dimension due to crystallization (Fig.12c).
POLYMER ENGINEERING AND SCIENCE, MIDJANUARY, 1988, Vol. 28, No. 1
11
Ming C . Chien and R. A. Weiss
CONCLUSIONS
The induction time for crystallization of
PEEK is significantly decreased by the application of strain during isothermal crystallization. Shearing increas& the rate of crystal formation. A s the shear rate increases, the induction time for crystallization decreases.
Compared with quiescent crystallization,
strain-induced crystallization is not as temperature-dependent since the onset of crystallization is primarily due to the deformation of the
melt. Shearing is the predominant cause for the
initiation of nucleation at temperatures very
close to the melting point of PEEK.
The morphology in the sheared samples is a
mixture of crystalline and amorphous materials. Crystallites are stretched and aligned
along the shearing direction.
The activation energy for strain-induced crystallization is considerably smaller than that for
quiescent isothermal crystallization. Shearing
a PEEK sample lowers the energy barrier for
the initiation of nucleation.
12
ACKNOWLEDGMENT
We gratefully acknowledge the support by the
Advanced Systems Division of Northrop Corporation.
REFERENCES
1. K. 2. Hong, Ph.D. Thesis, University of Michigan
( 1981).
2. D. J. Blundell and B. N. Osborn, Polymer. 24, 453
(1 983).
3. J. G. Shukla and W. J . Sichina. SPE ANTEC Tech.
Papers, 30, 265 (1984).
4. C. N. Velisaris and J. C. Seferis, SPE ANTEC Tech.
Papers, 31,401 (1985).
5. Y . C. Lee and R. S. Porter, Polym. Eng. Sci., 26,633
(1986).
6. W.J.Sichina and P. J. Gill. SPE ANTEC Tech. Papers.
31,293 (1985).
7. P. Cebe and S. D. Hong, “Crystallization Behavior of
Poly(ether ether ketone),”Polymer (in press, 1986).
8. D. J. Kemmish and J. N. Hay. Polymer. 26,905(1985).
9. G . S. Y. Yeh. K. Z. Hong. and D. L. Krueger. Polyrn.
Eng. Sci.,19,6 (1979).
10. J. N. Hay, D. J. Kemmish, J. I. Langford. and A. I. M.
Rae, Polym. Commun., 25,175 (1984).
POLYMER ENGINEERING AND SCIENCE, MID-JANUARY, 7988, Yo/. 28, No. 7