Heat Setting of Oriented Poly (ethylene

Terephthalate) : Effects of Deformation Mode,

Strain Level, and Heating Medium

C. P. BUCKLEY, Department of Mechanical Engineering, UMIST,

Manchester M60 lQD, United Kingdom, and D. R. SALEM, TRIIPrinceton,
601 Prospect Avenue, PO Box 625, Princeton, New Jersey 08542

Synopsis
We continue a previous study of the a and a' mechanical relaxation processes in oriented
poly( ethylene terephthalate) by means of heat-setting experiments on drawn filaments between
room temperature and the melting region. Both processes were exhibited in both torsion and
bending deformations, but heat setting was found consistently to be greater in bending than in
torsion: this was accounted for by the known anisotropy of the a relaxation. In both modes of
deformation, increasing strain level gave reduced recovery, and results suggested this arose from
high stresses activating at room temperature the stress relaxation mechanisms, which at low stresses
occurred only at elevated temperatures ( a if as-drawn, a' if preset). Results were found to be
independent of whether blown air or stirred oil was used as heat transfer medium, in contrast to
some previous reports in the literature; the reason lay in differences in the precise sequences of
deformation and temperature employed.

INTRODUCTION

Heat-setting experiments are of importance in two respects. First, they pro-

vide a method of obtaining meaningful measurements on mechanical relaxation
processes in temperature regions where a polymer is structurally unstable. The
thermal history associated with each data point is well defined. Second, they
provide a model in the laboratory of industrial heat-setting processes. These
are a particular feature of the manufacture of products from oriented
poly (ethylene terephthalate) (PET) fibers or film.
In a previous article we exploited heat-setting experiments in investigating
the high-temperature viscoelasticity of oriented PET in the temperature regime
where crystallization and other structural evolution take place during mechan-
ical tests. Drawn PET filaments were deformed in small strain torsion, heat
set, cooled, and released. Measurements of fractional recovery (the fraction of
strain induced that was recovered on release) revealed the viscoelasticity of
PET above the glass transition to be governed by another time-dependent re-
laxation process, distinct from the glass transition ( a ) process; and it was
labeled a'. The characteristic features were highly sensitive to the microstruc-
tural state of the polymer. Specifically, the center of the a' process moved to
longer relaxation times and higher temperatures with increasing crystallization.
It was suggested that the origin of the process is the slipping of stressed en-
tanglements by molecular diffusion, in a polymer network that is incompletely
crosslinked by crystals.

Journal of Applied Polymer Science, Vol. 41, 1707-1721 (1990)

0 1990 John Wiley & Sons, Inc. CCC 0021-8995/90/7-81707-15$04.00
1708 BUCKLEY AND SALEM

The purpose of this work is to report further details of the a and a' processes
in P E T as revealed by heat-setting experiments. The additional parameters
considered are the mode of deformation (torsion vs. bending), strain level, and
heating medium.
Previous studies of heat setting of PET in extension, 2,3 torsion, 4,5 and
bending4 have been reported in the literature, but it is now necessary to reex-
amine the phenomenon in light of the discovery of the a' relaxation.

EXPERIMENTAL DETAILS

The oriented PET monofilament used in this work had been prepared at
UMIST by melt extrusion, followed by drawing X5 at 80°C. It had a density
of 1381 kg/m3 and a diameter of 51 pm. A new specimen was cut from the
filament for each experiment. Further details of the material were given in the
previous article, * including microstructural details.
A schematic diagram of the main elements of the heat-setting experiment
is given in Figure 1. In the present work the following conditions were kept
constant: T o = 20°C, ta = 60 s, t c - t B = 60 s, to - tc = 300 s. The mode of
deformation was either torsion of the filament about its own axis or bending
of the filament around a cylinder. All results shown here refer to a heat-setting
time t, = 120 s. Other experimental variables were the temperature of heat
setting T,,the maximum (skin) shear strain of the twisted filament (in torsion)
yo, or the maximum tensile/compressive strain at the outer edge of the filament
(in bending) to, and the temperature of any thermal pretreatment of the filament

te mperature

I - - .
o tA ts te tc tD time
Fig. 1. Schematic diagram of a heat-setting experiment in torsion: the specimen was ( a ) a t
zero time, twisted to maximum shear strain yo; ( b ) at time t a , heated to temperature T.; ( c ) at
time ts, cooled to temperature To; ( d ) at time tc, released. Recovery strain yr was measured a t
time t D .Fractional recovery was defined f = y,/yo. For heat setting in bending a similar sequence
was followed, with bending deformation replacing twisting. Strain level was specified through the
maximum tensile/compressive strain a t the outer edge of the filament: eo for strain induced, e, for
strain recovered in this case f = c,/co.
 HEAT SETTING OF ORIENTED PET 1709

T,,. (In this paper, shear strain in a twisted filament is expressed as the tangent
of the shear angle.)
The experimental procedure for heat-setting filaments in torsion was de-
scribed before.' A summary of the experimental sequence for filaments heat
set in bending is given in the following paragraphs.

Deformation
For strains to < 0.1, the filament was wound one turn around a Tufnol cylinder
of the appropriate diameter and held in a screw clamp. For to > 0.1, the filament
was bent around a glass rod and secured in a clamp. Details of the apparatus
used are given elsewhere!

Heat
The filament, clamped at constant length on the cylinder or rod was immersed
in a bath of stirred, thermostatically controlled silicone oil at the required
temperature T, for the required time t,. T, was held constant to k O . 1 K. Alter-
natively (for heat setting in air), the clamped filament was heated in a forced-
air furnace (details follow).

Cool
The filament was removed from the oil and quenched in stirred silicone oil
a t 20°C. Alternatively, filaments heated in the forced-air furnace were brought
to room temperature by forcing cold air through the furnace.

Release
For to < 0.1, the filament was released from the Tufnol cylinder and placed
in a petri dish containing a thin layer of silicone oil, which ensured minimal
frictional resistance to recovery. For €0 > 0.1, the filament was released from
the glass rod and placed on a microscope slide with a thin layer of silicone oil.

Measurement
For to < 0.1, a camera with extension tube was mounted above the petri dish
and focused on the circularly deformed filament. Beneath the dish was placed
a measuring grid. From a magnified photographic negative of the filament, the
recovered radius of curvature of the filament was measured. For to > 0.1, a
photomicrograph was taken of the deformed filament. Since the recovered fil-
ament usually formed less than half a circle, it was not possible to take a
diameter measurement directly from the photomicrograph negative. Therefore,
with the aid of a digitizing table, the coordinates of three points along the
filament curve were obtained from magnified negatives, and the equation of
the circle was solved to obtain the recovered radius of curvature.
Fractional recovery f was determined as the fraction of applied curvature
(reciprocal radius of curvature) that was recovered on release (see Fig. 1) . The
random measuring error in f , as determined from repeat experiments, was found
to be 0.01 in torsion and in low-strain bending tests and 0.03 in high-strain
bending tests.
1710 BUCKLEY AND SALEM

Some specimens were thermally preset prior to heat setting. They were held
undeformed a t constant length and immersed in thermostatically controlled
silicone oil at the required presetting temperature T p ,for a time t p . Following
quenching in silicone oil at 20"C, they were subjected to heat-setting sequences
in the usual way, as described here and in Ref. 1. In the experiments reported
here, the presetting time was held constant at tp = 2100 s.

Forced-air Furnace
A forced-air furnace was designed for the rapid heating and cooling of fibers
in air. To heat the specimen, compressed air was forced through a coil of copper
pipe submerged in hot, stirred silicone oil, from where it was directed into a
chamber containing the specimen (see Fig. 2 ) .
The specimen chamber (Fig. 3 ) consisted of a stainless steel outer cylinder
(diameter 96 mm, height 72 mm) and an inner cylinder (diameter 65 mm,
height 56 mm) that, to minimize thermal mass and maximize heating unifor-
mity, consisted of aluminum foil, wrapped around a light stainless steel frame.
The gap between inner and outer walls was filled with insulating material. A
small hole through the wall of the chamber accommodated the thermocouple
of a digital thermometer. Hot air passed through the 15-mm pipe at the bottom
of the chamber and was exhausted through a funnel of equal diameter in the
lid. In order to spread the air uniformly throughout the chamber a baffle of
aluminum foil (Fig. 3 ) was fixed just above the hot-air inlet pipe. This also
created turbulence, to enhance the rate of heat transfer to the specimen.
To cool the specimen, compressed air entered from the side of the chamber,
directed via a U-shaped cooling tube submerged in a vacuum flask of crushed
ice and water.
The heating and cooling cycles were activated by three remote-controlled
solenoid valves. For example, the direction of air flow through the valves (Fig.
2 ) and the flow sequence for a typical setting experiment was as follows.

Cooling
Unit

Oil - bath

Fig. 2. Diagram showing main elements of the hot-air furnace used for heat-setting specimen
in air.
 HEAT SETTING OF ORIENTED PET 1711

Hot air out

t

t
Hot air in

Fig. 3. Section through the specimen chamber of the hot-air furnace outlined in Figure 2.

( a ) Valves 1 and 2 open, valve 3 closed cold air entered the chamber via
valve 1to maintain the specimen at a constant initial temperature; hot
air passing through the heating coil went to exhaust via valve 2.
( b ) Valves 1 and 2 closed, valve 3 open: cold air was shut off from the
chamber at valve 1; hot air was prevented from going to exhaust by

Fig. 4. Fractional recovery from low-strain heat setting in torsion ( y o = 0.006) and bending
(to= 0.005). Specimens had been preset at 200°C. In both modes of deformation a and a’relaxation
processes are visible as maxima in the gradient of f versus T,.
1712 BUCKLEY AND SALEM

0 50 100 150 200 2 50

Fig. 5. Fractional recovery from heat setting in torsion (yo = 0.06) and bending ( co = 0.06)
for specimens that were as-drawn. Only one relaxation process ( a )is visible.

valve 2 and instead entered the chamber via valve 3 and exhausted
through the funnel in the lid.
( c ) Valves 1 and 2 open, valve 3 closed as ( a ) .
Adjustment of the hand-operated valve 4 allowed the cold air flow rate during
phases ( a ) and ( b ) to be varied.
This system was designed to achieve high rates of change of the air tem-
perature so that step changes in temperature required by the ideal heat-setting
experiment (Fig. 1) could be closely approximated. Thus when heating and
cooling between ambient and temperatures in the range 125-25OoC, the rate
of heating of the air in the chamber was greater than 40 K/s, and the rate of
cooling was greater than 73 K/s.
RESULTS AND DISCUSSION
Deformation Mode
Results presented in Figures 4-6 compare the variations in fractional recovery
f with T,for bending and torsional modes of deformation. In each case the
 I I I I I I I I
1
torsion
f

I 1 I I I I I I

0 50 100 150 200 0 50 100 150 200

Ts ("C)

Fig. 6. Effects of varying presetting temperature Tpon the fractional recovery ffollowing heat
setting at temperature T,in torsion (yo= 0.006) and bending (c,, = 0.005).

I I I I

1.0

Tp = 200°C
0.8

0.6

04

0.2

a
-0.1
50 100 150 200
Ts ("C)

Fig. 7. Effects of varying strain level yoon fractional recovery following heat setting in torsion.
Specimens had been preset at Tp = 200°C.
1714 BUCKLEY AND SALEM

50 100 150 200

T, ("C 1
Fig. 8. Effects of varying strain level to on fractional recovery following heat setting in bending.
Specimens had been preset at Tp = 200°C.

pattern is clearly the same. The a and a' relaxations are evident as maxima in
the gradients of f versus T, plots. In as-drawn specimens only the glass transition
( a ) mechanism is apparent (Fig. 5 ) , but in preset specimens both a and the
higher temperature a' mechanisms are clearly visible. Increasing presetting
temperature Tpshifts the center of a' to higher T,(Fig. 6 ) . As reported pre-
viously, this phenomenon was explored in some detail for low-strain heat set-
ting in torsion and was found to reflect an increase in relaxation times of the
time-dependent a' process, presumed to be due to the crystallization occurring
during presetting. The clear message of these results is that both a and a'
processes are operative in both torsional and bending deformations, and that
their time/temperature behavior is the same, irrespective of the mode of de-
formation. It is clear from Figures 4-6, however, that heat setting in the two
modes does not produce identical results. In all cases it was found that the
magnitude of f was greater for heat setting in torsion than in bending: that is,
a greater degree of set was achieved in bending than in torsion.
For the as-drawn specimens deformed in torsion, the slight negative recovery
(overtwist) at T > 125°C has been observed before4s5and is believed to arise
from a small contribution from the anisotropy of thermal expansion combined
with the helical anisotropy of the filaments after t ~ i s t i n g . Such
~ . ~ an effect
 HEAT SETTING OF ORIENTED PET 1715

I 1 I I 1 I I

1.0

0.8

0.6

0.4

0.2

- 0.1 I I I I I I I
50 100 150 200
TS ("C)
Fig. 9. Effects of varying strain level y o on fractional recovery following heat setting in torsion
for specimens that were as-drawn.

would not be expected to the same degree in bending, and indeed it was not
observed (Fig. 5).
The present work did not include heat setting in pure extension, but close
examination of the data obtained in such a study by Lunn et a1.2 suggests that
a' relaxation was evident in their data for oriented, annealed PET but was
overlooked. Since bending deformation of a filament occurs by axial extension
and compression strains, the occurrence of the a' relaxation in extension would
be expected in view of the present results for heat setting in bending.
To interpret the results of heat-setting experiments it is helpful to employ
the methods of nonisothermal viscoelasticity. Fractional recovery f is then seen
to correspond approximately to the stress relaxation modulus applying to time
t, and temperature T, of setting, normalized with respect to the modulus at
To.' It follows that if f i s lower for heat setting in bending than in torsion, the
fall in axial tensile modulus with rise in temperature in the region considered
must be greater than that in axial shear modulus.
Evidence that this is indeed so can be found in the dynamic mechanical
study by Davies and Ward of anisotropy of the a relaxation in annealed, oriented
PET.' Tan 6 ( a t 250-300 Hz) was measured with tensile stress at angles O",
45", and 90" to the orientation direction ( O D ) . In both materials studied-
1716 BUCKLEY AND SALEM

0.41

torsion, p r e - s e t

I \ pre-set \

O torsion, not p r e - s td

0.02 0.06 0.10 0.14

6, ( t o r s i o n ) or €,(bending1
Fig. 10. Comparison of the strain dependence of fractional recovery in torsion and bending
for specimens preset at T, = 150°C and then heat set at T,= 200°C. The lower curve is for as-
drawn specimens deformed in torsion.

hydrostatically extruded rod and film drawn at constant width-it was found
that

tan 6 o > tan 645

throughout the a region of temperature. The ratio of peak heights (tan do :tan
dd5) was 1 : 0.73 and 1 : 0.79 in the two cases, respectively. Since tensile defor-
mation in these materials at 45' to OD would be dominated by axial shear, the
results are consistent with the anisotropy of heat setting observed in the pres-
ent work.
The explanation for the anisotropy lies in the anisotropy of mechanical cou-
pling of the external stress to the amorphous fraction, the relaxing element of
the microstructure in the glass transition ( a ) process. The structure of an
annealed oriented semicrystalline polymer, such as P E T under tensile stress
parallel to OD, provides approximately series coupling to the amorphous frac-
tion, as first shown quantitatively by Buckley et al.' Series coupling produces
maximum strain energy density in the low modulus (amorphous) fraction and
hence yields maximum tan 6 and fractional drop in modulus at the glass tran-
sition.
 HEAT SETTING OF ORIENTED PET 1717

I 1 I I

”
0 0.05 0.10 0.15 0.20
YO
Fig. 11. Recovered shear strain yr versus induced shear strain yo for torsional heat setting of
preset specimens at setting temperatures T,is shown. Presetting temperature was T, = 200°C.
Nonlinearity is clearly apparent.

The practical situation where heat setting of oriented PET filaments in

torsion and bending is of vital importance is the false-twist texturing process.
As the yarn is first twist set, individual filaments within it are heat set rapidly
into a helical shape, involving both bending and torsional deformations. The
yarn is then de-twisted, causing each filament to adopt a buckled or “crimped”
shape. Research on the kinetics of this process has shown clearly that the rate
at which it can be carried out is limited by the time-dependent setting mech-
anism within the filaments of the yarn.” To examine the effects of varying
degree of set in torsion and bending, an approximate analytical model of the
crimp shape was solved as part of the present study.6 The level of crimp was
defined (somewhat arbitrarily, but comparably to methods used in practice)
as the force to hold the textured yarn at 15% contraction from its twist-set
length. The effects of varying f i n bending and torsion were examined by cal-
culation. It was found that the level of crimp achieved is dominated by the
degree of set in bending, while the effect of degree of set in torsion is almost
negligible.

Strain Level

For preset specimens deformed in either torsion (Fig. 7 ) or bending (Fig.

8),the effect of increasing the applied strain level was to reduce f (i.e., to
increase the degree of set) a t all values of T,. The same was true for as-drawn
specimens at temperatures below 125”C, but at higher temperatures (in the
region of negative f ) increasing yo led to lower overtwist, as predicted by theory5
(Fig. 9 ) .
Figure 10 compares the dependence of f on strain level for preset fibers with
Tp = 200°C, set at T,= 150°C (i.e., in the plateau region) in bending or tor-
sion. It is clear that fibers deformed in bending display lower recovery f a t
1718 BUCKLEY AND SALEM

f I I I I I 1 1

1.0
( a ) torsion

C
0.8 - 9

0.6 - Air
___ Oil.

0.4

0.2

0
50 100 150 200 T, (*CI
Fig. 12. Fractional recovery f versus setting temperature T,for specimens heat set in air in
torsion or in bending. Dashed lines show corresponding curves for specimens heat set in silicone
oil. The effect of varying the heating medium is shown to be insignificant. Labels on curves indicate
( A ) as-drawn specimens; ( B ) preset a t T, = 150°C; ( C ) preset a t Tp = 200°C. Strain levels were
yo = 0.006, eo = 0.005.

all strain levels explored. From the torsional data in Figure 10, it is also
apparent that the effect on f of varying yo becomes measurable at a strain
level in the region of 0.02. This is a manifestation of nonlinearity in thermo-
viscoelastic behavior and is represented differently in Figure 11, where
the recovered strain y r is plotted versus yo for heat setting in torsion at
two temperatures. The point of measurable departure from linearity can
be seen to occur at an applied strain of approximately yo = 0.016 in
both cases.
It is a well-known feature of the nonlinear viscoelastic deformation of poly-
mers that with increase in applied strain level the proportion of unrecovered
strain increases on any given time or temperature scale. This was apparent,
for example, in the torsional heat-setting data of Buckley et al.5 and the ex-
tensional data of Lunn et a1.2The present results, however, reveal an interesting
difference between specimens that have or have not been preset. In preset
specimens the reduction in f with increase in strain appears independent of T,
up to the plateau region separating a and a’ regimes (Figs. 7 and 8 ) , showing
 HEAT SETTING OF ORIENTED P E T 1719

f I I I I

1-0
( b ) bending

0.8

- Air.
_ _ - Oil.
0.6

0.4

0.2

0
50 100 150 200 T,Kl
Fig. 12. (Continuedfrom the previous page.)

that it is associated with stress relaxation by the a' process. The suggestion
made previously was that the a' mechanism consists of entanglement slippage.'
If this is so, the new results provide evidence for entanglement slip even during
room temperature deformation, when strains are sufficiently large. In the case
of as-drawn specimens, however, the reduction in f with increase in strain
appears to be associated with the a process (Fig. 9),although in these specimens
this may be overlapped by a'.

Heating Medium

Figure 12 demonstrates that fractional recovery from heat setting does not
depend on whether the heating medium is oil or air. For both torsion and
bending, fractional recovery from heat setting in the hot-air furnace was found
to be the same, to within experimental error, as that obtained in silicone oil,
other conditions being the same.
This point is of interest in view of the scattered reports in the literature of
major differences in thermal shrinkage and microstructure in annealed oriented
PET, caused by use of different heat transfer media; see, for example, Wilson."
1720 BUCKLEY AND SALEM

The fact that even inert fluids such as air and silicone oil can give large differ-
ences in behavior is readily explained by the heat transfer coefficient being the
critical parameter, through its control of heating rate. Wilson’s work showed
that the origin of such effects is competition between the kinetics of thermal
shrinkage on the one hand and crystallization on the other. When the latter
takes place in a drawn filament, it tends to stabilize the oriented state and
hence to reduce subsequent shrinkage. Faster heating allows more shrinkage
to occur in advance of crystallization, and hence gives rise to greater shrinkage
and amorphous disorientation at the temperatures above ca. 100°C where
crystallization occurs in PET. This explanation of events is consistent
with the present results. All heat setting was carried out at constant specimen
geometry, merely by means of stress relaxation at the setting tempera-
ture. There was therefore no simultaneous deformation and crystalliza-
tion involved, and the results would be expected to be independent of
the precise value of heating rate and hence of the heat transfer medium
employed.

CONCLUSIONS

This work has shown the pattern of heat-setting behavior discovered by us

previously, in small-strain torsional setting of oriented PET filaments, to extend
to bending deformations and to larger strains in the nonlinear viscoelastic
regime. In as-drawn specimens there is a single relaxation process in the glass
transition ( aprocess) region. Preset specimens, however, exhibit both the (Y
and at relaxation processes.
In all cases explored, magnitudes of fractional recovery were found to be
lower for bending than for torsional deformation, for given presetting condi-
tions, setting time and temperature, and maximum strain level. This result
is of practical importance in that yarn crimp developed in the false-twist
texturing process is more sensitive to the degree of set in bending than in
torsion.
Increasing the strain level leads to decreasing fractional recovery for both
torsion and bending, and in both as-drawn and preset specimens. In terms of
the suggested mechanism of at relaxation, the results for both modes of defor-
mation suggest that this effect is due in preset specimens to entanglement
slippage occurring in room temperature deformation, when strains are suffi-
ciently high.
Comparison of results obtained with blown air or stirred silicone oil as heat
transfer media showed no significant differences, in contrast to some previous
reports in the literature. The reason is that the present experiments were de-
signed to avoid concurrent deformation and crystallization, so results were not
so sensitive to competition between deformation and crystallization kinetics
and hence to the precise heating rates employed.

The work described in this article was carried out at the Department of Textiles, UMIST. The
authors are grateful to Professor J. W. S. Hearle for advice and encouragement and to ICI Fibres
and to the Science and Engineering Research Council for financial support.
 HEAT SETTING OF ORIENTED PET 1721

References
1. C. P. Buckley and D. R. Salem, Polymer, 2 8 , 6 9 ( 1987).
2. A. C. Lunn, B. L. Lee, and I. V. Yannas, Polym. Engng. Sci., 1 4 , 6 1 0 (1974).
3. R. Shishoo and K. M. Bergh, Text. Res. J., 4 7 , 5 6 (1977).
4. S. Arghyros and S. Backer, Text. Res. J., 5 2 , 295 (1982).
5. C. P. Buckley, J. W. S. Hearle, and R. Mandal, J. Text. Inst., 7 6 , 264 (1985).
6. D. R. Salem, Ph.D. Thesis, University of Manchester, UK, 1982.
7. C. P. Buckley, Int. J. Mech. Sci., 23, 503 (1981).
8. G . R. Davies and I. M. Ward, J. Polym. Sci. A-2, 1 0 , 1153 (1972).
9. C. P. Buckley, R. W. Gray, and N. G. McCrum, J . Polym. Sci. B, 7 , 8 3 5 ( 1969).
10. D. S. Barnes and W. J. Morris, J. Text. Inst., 7 1 , 291 (1980).
11. M. P. W. Wilson, Polymer, 1 5 , 2 7 7 (1974).

Received May 18, 1989

Accepted November 20, 1989