J O U R N A L OF M A T E R I A L S S C I E N C E 4 (1969) 1088-1096

The Tensile Behaviour of Polyethylene

Terephthalate
J. M. S T E A R N E * , I. M. W A R D t
I C I Ltd, Petrochemical and Polymer Laboratory, Runcorn, Cheshire, U K

Received 17 June 1969

Measurements of the load/extension curves of polyethylene terephthalate (PET) over a

wide range of temperatures showed four regions of behaviour. These were brittle fractu re,
ductile, cold-drawing, and uniform extension. A particular study was made of the transitions
between brittle fracture and ductile failure, and between ductile and cold-drawing, since
these define the limits of the three regimes of failure observed at temperatures below the
glass transition and softening range.
The effects of molecular weight and crystallinity were examined. The brittle strengths
measured in tension at low temperatures showed a very large scatter. There was evidence,
in spite of this scatter, that the brittle strength falls with decreasing molecular weight. The
yield behaviour was not affected, so that the brittle/ductile transition moves to higher
temperatures.
Crystallinity affects both brittle strength and yield behaviour. The brittle strengthfalls with
increasing crystallinity, whereas the yield stress rises. Both effects combine to raise the
temperature of the brittle/ductile and the ductile/cold-drawing transitions.
Stress/temperature curves were also constructed for notched specimens. Notching raises
the effective yield stress and reduces the brittle strength so that the brittle/ductile transition
is moved to higher temperatures. The observed effects are in qualitative agreement with
theoretical predictions of the plastic constraint at the tip of a notch, and thus the latter
gives a satisfactory qualitative explanation of notch sensitivity. Notching leads to brittle
failure at room temperature, and in notched specimens the brittle strength rises as the
temperature is decreased. The brittle strength of the lowest molecular weight sample was
again significantly less than that of the higher molecular weight samples.

1. Introduction this laboratory, with reference to yield stress and

There is an increasing interest in the failure impact strength [2].
phenomena in polymers, whether this occurs by
brittle fracture or ductile yield. It is evident that 2. Experimental
in some respects polymers behave similarly to 2.1. Preparation of Samples
metals. For example, brittle/ductile transitions The tensile samples of P E T were prepared by
play an important r61e in their behaviour [1 ]. injection moulding, Two samples sizes were
The present investigation examines the failure of chosen; for their dimensions see fig. 1. It was
polyethylene terephthalate over a wide range of appreciated that such samples are not randomly
temperatures at various strain rates. oriented. A separate experiment was therefore
Particular points of interest are the notch undertaken to investigate this effect and this is
sensitivity and the effects of molecular weight and discussed in section 3 below. Samples with differ-
crystallinity. The latter aspects have been dis- ent degrees of crystallinity were obtained either
cussed previously in an earlier publication from by varying the time in the mould and the mould
*Now at ICIANZ Ltd, Ascot Vale, Melbourne, Australia.
~-Joint appointment between ICI Ltd and Bristol University.
1088 9 1969 Chapman and Hall Ltd.
TENSILE BEHAVIOUR OF PET
temperature or, as in the case of the small dum-
bell specimens, by using a specially heated jig to
prevent distortion of the samples. This jig was
mounted in a heating block oven at temperatures
ranging from 150 to 180 ~ C and for times varying
from 7 to 25 min.
GATE 0"5in.RADIUS
, -~-
A
o17s o-i7
!
l~ 4"5in.
THICKNESS=O.125in.
GATE 0-5 in.RADIUS
,-L 3in
THICKNESS = O - O 6 3 i n ,
Figure I Dumb-bell dimensions for tensile experiments.
l in. = 2.54 cm.
2.2 Tensile Measurements
The tensile experiments were carried out on a
Hounsfield E-type Tensometer, equipped with a
variable temperature control cabinet, capable of
producing temperatures in the range - - 9 0 to
-F 200 ~ C. The brittle strengths at - - 1 9 6 ~
were obtained using a specially designed tensile
rig mounted in a thermos flask containing liquid
nitrogen. Temperatures between - - 1 9 6 and
-- 90 ~C were obtained by passing N2 gas through
a germanium-silver coil heat exchanger immersed
in a thermos vessel containing liquid nitrogen,
see fig. 2.
2.3. Measurement of Crystallinity in PET
Samples
The degree of crystallinity, We, was calculated
from the density of the samples, assuming the
amorphous and crystalline phases to be additive
(see [2] for details). A calcium nitrate density
gradient column was used to measure density.
2.4. Measurement of Molecular Weight
Values for the number average molecular weight
Mn were estimated from solution viscosity
measurements using the Mark-Houwinck rela-
TENSOMETER
tu
SPECIMEN~ ~ ~ ~'~====~~ ~ N2
LIQUIDN;t
I
Figure 2 L o w temperature for tensile experiments.
tionship [,/] = K~rn" where [7] is the intrinsic
viscosity; K and a are constants assumed, on the
basis of previous work [3], to be 1.7 x 10-4 and
0.83 respectively.
3. Results and Discussion ; Te ns ile
Fracture Experiments
3.1. Amorphous PET
Diagrammatic representations of the different
types of failure observed in amorphous P E T a r e
shown in fig. 3. In a recent article Vincent [4]
distinguishes four of the five modes of failure
shown in this figure. These are brittle, necking
rupture, cold-drawing and uniform extension. In
P E T we observe the behaviour termed "ductile"
at temperatures immediately above the brittle
range, but at this stage we do not see the forma-
tion of a neck as in the necking rupture case
described by Vincent.
The stress-temperature plots, shown for
example in figs. 4 and 5 reveal three regions of
behaviour: a brittle region, a yield region (which
embraces both ductile and necking rupture) and
a cold-drawing region. Necking rupture is only
seen in a temperature range just below the cold-
drawing region. For this reason in our work we
have distinguished a brittle region and two
regions of yield behaviour.
It was feared that injection moulding might
introduce a degree of molecular orientation
which would invalidate the results. The effects of
annealing the samples for 2 h at 80 ~ C were
therefore examined, and the data are included in
fig. 4. No significant differences were observed
between annealed and unannealed samples, either
in respect of yield stress or brittle fracture stress.
Annealing produced a reduction in the low
1089
 J. M . S T E A R N E , I. M. W A R D

degrees of birefringence observed when the

sample was viewed between crossed polarisers,

/,,: I BRITTLE I DUCTILE

and small amounts of shrinkage (,~ 109/oo
maximum). These small amounts of shrinkage in
amorphous P E T are consistent with the very low
birefringence [5]. Previous studies of the influ-
ence of molecular orientation on yield stress [6]
i i
I
I
I are consistent with this small degree of bire-
i
EXTENSION EXTENSION fringence introducing a negligible increase in the
yield stress above that for an isotropic sample.
The effect of molecular weight on the stress/
temperature behaviour is shown in fig. 4. It can
be seen that the yield stress is independent of
molecular weight (over a range of molecular
NECKING weight from 11 000 to 27 000). This appears to
COLD DRAWING
be true for both regions of yield behaviour,
although it must be emphasised that very small
o changes in the extent of the lower region of yield
i.e. the ductile region, would not be detectable.
These results show that the brittle stress is not
F really affected by molecular weight from 16 500
EXTENSION EXTENSION to 27 000 but that the brittle stress of the 11 000
sample is very significantly lower. This result
suggests that there may be a threshold in the
molecular weight required for ductile behaviour
at room temperature. This is an important result
UNIFORM EXTENSION and will be investigated further.
The effect of strain rate on the behaviour is
shown in figs. 5 and 6. At crosshead speeds from
0.5 to 15 in. min -1. no change in the brittle stress
is observed, whereas the yield stress in both
EXTENSION regions increases uniformly. Fig. 5 shows that the
brittle stress is temperature and strain rate
Figure 3 Diagrammatic representation of the different independent to a very good approximation. The
types of failure. brittle/ductile transition and the ductile/cold-
Mn
EXTENSION RATE 5 i n rain -1 e I.V.----0'39 11000

24 ~ I,V.=0'54 16500
_~ 13 []
22 ............. ,~. . . . . . . & I.V.=O.58 18000
20 e A^~. x I.V.=0.60 18500

18 x '~ | -"~'- ~ - , ,e~ ~' I . V . = 0 . 8 4 27000

.E 16
_a | e

|
I--
~o
z 8 ~G e | |

o6 A x ee
x
~' 4
0
2

r I I I 1 q I t r I
210 l I
-- O 0 --180 --160 --140 --120 --100 --80 --60 --40 --20 0 40 60
TEMPERATURE ~
Figure 4 E f f e c t o f molecular weight on amorphous PET. 1 in rain, - 1 = 2.54 c m rain - 1
*1 in. m i n . - z = 2.45 cm rnin -1.
TENSILE gEHAVIOUR OF PET

I,V. 0'58 IM 0"54 I.V. 0"84

9 EXTENSION RATE Sin,rain"1 x EXTENSION RATE 0'5 in.rain "t e EXTENSION RATE 0'5 in.rain4

&EXTENSION RATE15in.m|n-1 vEXTENSION RATE Sin.rain 4

24
" X X
. . . . ~ . . . . . . . .

2O

"7-, 16

z
x
'0
~ 4
m t~

F i i i i ~ J I ~ i i i F

-200 -t80 -t60 -140 -120 -100 -80 -60 -40 -20 0 20 40 6O
TEMPERATURE OC

Figure 5 Effect of strain rate on amorphous P E T . 1 in. min -1 = 2.54 cm min - 1 .

22 ~ AMORPHOUS PET ([.v.=0.54}

| "~'--~ e EXTENSION RATE 10 in.rain-1

20 "~t--.... y, EXTENSION RATE 40in.rain-

18

16

14

,E
~12

t~

mtO
d
<
z

0 g
z
~•
0
~ 6

i i i 1 i F I I

-120 -100 -80 -60 -40 -20 0 20 40 60

TEMPERATURE~
Figure 6 Effect of high strain rates. 1 in. min - j = 2.54 cm rain -1, 1091
5

drawing transition both move to higher tempera-
tures with increasing strain rate.
At the higher rates of extension the stress
values had to be measured from an oscilloscope
response because of the lag on the recorder pen.
Three regions of failure were evident, but at the
highest rate of strain the ductile/cold-drawing
transition was not very clearly defined. At the
highest crosshead speed of 40 in. rain -1 the yield
stress values fell below those for 10 in. min -1
(see fig. 6), which suggests that adiabatic heating
occurs.
The experimental results show that there are
two well defined regions of yield behaviour. As
well as differing in the nature of strain hardening
i.e. in post-yield behaviour, the regions differ in
the nature of the yield process, as revealed by the
strain-rate sensitivity. A possible explanation for
this is as follows. At high temperatures, cold-
drawing occurs and this has been described [6] as
the stretching of a molecular network. In this
region it was shown by the effects of cross-linking
that the yield process involves freeing the
molecular chains between entanglement points.
At lower temperatures, the polymer ruptures i.e.
the strain hardening associated with the orienta-
tion of the molecular network does not occur. It
is therefore proposed that at these temperatures
the yield process involves the pulling out of
entanglements. This yield process would be
24
2O
16!
w
~12
Z
o
z8
I I I I
-200 -t80 -160 -140 -120 -100
Figure 7 Effect of crystallinity, 1 in. min - I = 2,54 cm min - 1 .
*1 p s i = 6894.76 N m -~.
1092
J. M. S T E A R N E , I. M. W A R D
expected to have a higher activation energy than
the high temperature yield process. The data
tend to confirm this, although the range of strain
rates is not sufficiently great to allow a detailed
interpretation.
The change in behaviour at ~ -- 60 to -- 40~
could be associated with the ),-relaxation
transition in this polymer [7].
3.2. Crystalline P E T
The effect of crystallinity on the stress/tempera-
ture behaviour is shown in fig. 7. It can be seen
that up to 35 % crystallinity there is little influ-
ence on the brittle stress, but the yield stress is
much increased. The brittle/ductile and ductile/
cold-drawing transitions therefore move to
higher temperatures. Increasing the crystallinity
from 35 to 44% (see fig. 8) has a dramatic effect
on the brittle stress, reducing it from about
20000 to 15000 psi.* This could be due either to
annealing faults or to a change in the morpho-
logical structure. The transition points are still
evident, and there is a further increase in the yield
stress in both the ductile and the cold-drawing
regions. These results show that both yield
mechanisms are affected by crystallinity. This is
an acceptable result, in terms of our tentative
proposals for these yield mechanisms. The
presence of crystallites is likely to make it more
difficult either to disrupt the structure entirely, or
PET I.V.--.=0 ' 5 4
EXTENSION RATE 5"0 in.rain -1
| 35% CRYSTALLINE
x. AMORPHOUS
I / I I ! I / I I
-80 -60 -40 -20 0 20 40 60
TEMPERATURE ~
TENSILE BEHAVIOUR OF PET

x ~ ~ x | IiV,= 0"45~ 43~ EXTENSION RATE S'O in.rnin-1

14 • xx ~ ' ~ , ~ . X I.V.= 0'54; 43% CRYSTALLINITY:, EXTENSION RATE S'O in.rnfn-I
| | x|215174|
| | | |
12
x

'clO e ~
x •

x•

.1
~
Z

I I i I I I I I
- O -60 -40 -20 O 20 40 60 80
TEMPERATURE ~
Figure8 Effect of molecular weight on crystalline PET. 1 in. rain -1 ~ 2.54 cm rain -1

to allow chain sliding which is the essence of the
high temperature yield process.
It is interesting to note a direct effect of
molecular weight. Fig. 8 shows that decreasing
the molecular weight from 16 500 to 11 000
moves the brittle/ductile transition from - - 3 0
to -- 10 ~ C. This appears to be due entirely to a
reduction in the brittle stress. The ductile/cold-
drawing transition does not move significantly.
This confirms that crystallinity is more important
than molecular weight in determining the yield
behaviour, a result to be deduced from the data
of figs. 4 and 7. The data for the crystalline
material is summarised in table I.
3.3. E f f e c t o f N o t c h i n g
Notched samples were examined for stress/strain
behaviour at various temperatures.
3.3.1. Preparation of Samples
A mould was constructed for producing samples
that could be effectively notched and the
dimensions of this mould are shown in fig. la. A
Charpy 0.010 notching tool was mounted in a
specially constructed holder and cutting of the
notches was carried out on a milling machine.
Two notch depths were used, 0.075 in. and 0.110
in.
3.3.2. Amorphous P E r
Experimental results for P E T show that notching
can drastically alter the pattern of the character-
istic behaviour for a given polymer. In fig. 9 the
notched and unnotched stress/temperature curves
and Notch Sensitivity for amorphous P E T at a cross-head speed of 0.1
in. rain -1 are shown. The notched sample has
two notches diammetrically opposed, cut to a

TABLE I Summary of results from the stress-strain temperature exp3rimants carried out on crystalhne PET.

Cross-head speed Mol wt % crystallinity Brittle/ductile Ductile/cold- Brittle

in./min transition drawing strength
temperature ~ transition psi
temperature ~ (-- 80~C)
5.0 16 500 35 -- 65 -- 30 21 000
5.0 16 500 43 -- 30 50 15 500
5.0 11 000 44 -- 10 50 13 500
lin min-1 = 2.54 cm rain-l; lpsi = 6894.76 N m -2.
1093
 J. M . S T E A R N E , I. M . W A R D

LV.=O.54. x NOTCHED (DOUBLE)

EXTENSION RATE O.1 in,rnin -1
2 4 .,,& & & & | UNNOTCHED

EXTENSION RATE O.Sin.min "1 & UNNOTCHED

2O
A
'c

t6

w
rr
I-
mt2
...1
z

O 8
z
x
?
o
4

-200 -180 -160 -140 -120 -1OO -80 -60 -40 -20 O 20 40 60
TEMPERATURE~

Figure 9 Notched and unnotched amorphous P E T . 1 in. m i n - 1 = 2.54 c m m i n - 1 .

depth of 0.110 in. In the cold-drawing region the temperature of the brittle/ductile transition very
average yield stress has been raised by notching appreciably so that notched PET shows brittle
by a constraint factor, but not to the extent that behaviour at room temperature. The brittle
the calculations for such notch dimensions strength is temperature dependent, rising with
predict (see appendix). decreasing temperature to a value close to the
Notching also affects the brittle stress, reduc-
unnotched brittle strength. The fracture surfaces
ing this so that when the stresses are calculated
on the basis of the remaining cross-sectional area indicated brittle behaviour over this temperature
the notched brittle stress curve undercuts that for range.
the unnotched sample. The notched brittle strengths are molecular
These effects taken together increase the weight dependent, as shown in fig. 10.

24f
22
EXTENSION RATE O'1 in.rain-1 |

•
I.V,=O,84

I,V.= 0 ' 5 4
2O A I.V, = 0 " 4 0
T *
.~ 18

m 1 6 -~, •
if)
uJ
w 14

j12 A | X X

0
z 8

O" i I I I I -- 810 I I I I I I
-200 -180 --160 -140 --120 -IOO --60 --40 -20 0 20 40 60
TEMPERATURE ~

Figure 10 Effect of molecular weight on double notched amorphous P E T . 1 in. m i n - 1 = 2.54 c m m i n - t .

1094
TENSILE BEHAVIOUR OF P E T

20
• UNNOTCHED

| NOTCHED

u)
u.112

M
Z 8

O
Z

)
I
-100 -80 - 6 0' - 4 0' - 2 0' 0' 2o 4o 6o 80
' l e' o
TEM PERATU RE ~

Figure 11 Notched and unnotched crystalline.

3,3.3. Crystalline P E T 4. Conclusion

The increase in yield stress on notching (in the
cold-drawing region) appears to be greater in the
case of the crystalline samples, and the transition
shifts to a higher temperature region, in keeping
with the result observed with the unnotched
samples. In fig. 11 the notched and unnotched
stress/temperature curves for a 43 % crystalline
P E T sample are shown. The undercutting of the
ductile region of the unnotched curve due to
notching is quite substantial, but again the brittle
stress as measured at -- 196 ~ C is not lowered
very greatly (see table II). Increasing crystallinity,
as in the unnotched case, causes a shift in the
brittle/ductile transition to higher temperatures
as can be seen in fig. 12.
This survey of fracture in P E T provides a num-
ber of guide lines for understanding the failure
of this polymer.
(i) There is some similarity between brittle/
ductile transitions in polymers and those in
metals (as proposed by Vincent [1 ]). The brittle
stress is almost temperature and strain rate
independent, whereas the yield behaviour is
markedly sensitive to these parameters. Notch
sensitivity, i.e. embrittlement due to notching,
relates to the effective raising of the yield stress
due to the plastic constraint at the tip of the
notch.
(ii) The yield stress is not sensitive to molecular
weight. It is, however, increased by increasing

24

9 NOTCHED PET 450/0 CRYSTALLINE

20 X NOTCHED PET 30~

| NOTCHED PET 60/0 CRYSTALLINE

.E
.o
~16
u3
u3
Ld
I--
m12

Z ~ ~ O _ _ 9 ".% 9

Z
x
b
4

/ I I I I I I I I
-100 -80 -60 -40 -20 O 20 40 60 80
TEMPERATURE ~
Figure 12 Effect of crystallinity on notched P E T .
1095
 J. M. STEARNE, I. M. WARD

T A B L E II Summary of notched tensile date.

Mol wt ~ Crystallinity Ductile/cold-drawing Brittle strength, Brittle strength,

transition unnotched, notched,
temperature ~ -- 196 ~ C -- 196 ~ C
psi psi
16 500 0 25 23 000 22 000
16 500 43 64 18 000 17 250

*1 psi = 6894.76 N m -=.

crystallinity and this can lead to brittle behaviour (i) For a 0.075 in. notch and r = 0.010 in. the
if this yield stress exceeds the fracture stress, constraint factor is 2.09.
which is reduced by crystallisation. (ii) If r is 0.005 in. then the constraint factor is
(iii) The brittle stress is comparatively insensitive 2.12.
to molecular weight at high molecular weights (iii) For a 0.110 notch and r = 0.010 in. the
but falls rapidly at low molecular weights. The constraint factor is 2.04.
brittle/ductile transition therefore moves to
higher temperatures as the molecular weight of Acknowledgement
the polymer falls and can even lead to brittle The authors would like to thank Mr T. A.
behaviour at room temperature for a very low Roberts who carried out much of the experi-
molecular weight. mental work and Mr P. I. Vincent for helpful
comments on the draft manuscript.
Appendix
Constraint Factor for Deep Wedge-Shaped References
Notches with Circular Roots [8] 1. P. I. VINCENT, Polymer 1 (1960) 425.
From the longitudinal tensile stress distribution 2. J. B. J A C K S O N a n d E. R. D I X O N , J. Materials Sci. 3
across the minimum section of a long notched (1968) 464.
bar, use of Hencky's theorem leads to a yield 3. D. A. S. RAVENS and I. M. WARD, Trans. Faraday
Soc. 57 (1961) 150.
point load (L) per unit thickness of,
4. P . I . VINCENT, Encyclopedia of Polymer Science and

L=4ka
[ (1 + a ) - - a ( er~ - - i --a)
] Technology Vol. 7 (1967) p. 292.
5. P. R. PINNOCK and I. M. WARD, Trans. FaradaySoc.
62 (1966) 1308.
if a/r /> e ~ -- 1. Here r is the radius of the circu- 6. s. w . ALLISON and I. M. WARD, Brit. J. Appl. Phys.
lar root, of angular span 2a. 2a is the width of the 18 (1967) 1151.
minimum section of the bar and 2k is the tensile 7. A. B. THOMPSON and D. W. WOODS, Trans. Faraday
yield stress. Soc. 52 (1956) 1383.
The constraint factor is the ratio of the yield 8. R. HILt, "Plasticity" (Clarendon Press, Oxford,
point load to the yield point load for tensile yield 1950) p. 250.
of the same thickness of width 2a and equals
L/4ka .

~ ' ~ ~ / / 2 0=45~
ct=t35 ~
\
\ / 2(a "38in
\ /
\ /

v
A1

1096