Preparation of ultra-high modulus linear

polyethylenes; effect of molecular weight and

molecular weight distribution on drawing
behaviour and mechanical properties
G. Capaccio and I. M. Ward
Department of Physics, University of Leeds, Leeds LS2 9JT, UK
(Received 27 March 1973; revised 30 May 1973)

A systematic investigation of the effect of molecular weight and molecular weight distri-
bution on the cold drawing behaviour of linear polyethylene has been undertaken. In the
molecular weight range studied, the natural draw ratio was very sensitive to the morphology
of the initial material; spectacular effects on the natural draw ratio were observed provided
that an optimum initial morphology was achieved. These effects can be related to both
molecular weight and molecular weight distribution.
The extensional modulus and melting behaviour of the drawn material was also examined.
To a first approximation the extensional modulus related to the natural draw ratio, and at
very high draw ratios (~30) extremely high extensional moduli (~700kbar) were obtained.
The structure and properties of the drawn material did, however, also depend on the
molecular weight and molecular weight distribution. In particular, when certain molecular
weight requirements were satisfied, the oriented samples showed the presence of extended
chain material. It does, however, appear that although differences in molecular weight and
molecular weight distribution give rise to differences in extensional moduli, the presence
of extended chain crystallization per se is not a necessary requirement for the production
of high modulus material.

INTRODUCTION hence on the draw ratio and final modulus; and (b) the
formation of extended chain crystals in drawn samples
A number of values have been proposed for the Young's
as a function of molecular weight and draw ratio.
modulus of fully aligned linear polyethylene (LPE) on Furthermore we present a new material, in the form of
the basis of theoretical calculations and various experi- a highly oriented film, which exhibits the highest uni-
mental determinations of the crystalline stiffness 1, 2. All
directional modulus ever obtained at room temperature
these estimates, although far from being in good agree-
for LPE, and almost in the range predicted for fully
ment amongst themselves, indicate that a Young's
aligned chains.*
modulus of at least 2 Mbar should be expected for a
completely oriented sample. This value is several orders
of magnitude higher than is usually achieved3, 4 by cold EXPERIMENTAL
drawing conventional LPE to a typical draw ratio (A)
of about 9-10. There is therefore considerable interest The materials used were four commercial grades of
in attempting to obtain higher draw ratios and hence Rigidex (BP Chemicals) LPE homopolymer and the
higher moduli, and also in exploring the structural gel permeation chromatographic (g.p.c.) molecular
factors which prevent the theoretical moduli from being weight data in Table 1 were kindly supplied by Mr
achieved. In a previous publication 5 the effect of mole- J. M. Squire, BP Research & Development Dept.,
cular weight has been studied using a number of samples Grangemouth Division, Grangemouth, UK.
having different number-average molecular weights Sheets 0-05-0.07cm thick were obtained by com-
(/Qn) as well as different weight-average molecular pression moulding pellets of polymer at 160°C between
weights (/Qw). The results, due to the particular selection copper plates. Some sheets were immediately quenched
of samples available, could not be considered conclusive; in cold water while others were slowly cooled, at a
nevertheless they indicated a dependence of draw ratio rate of 7-9°C/min, to a temperature T~ (measured
on molecular weight and suggested that there was a either inside the sheet or on the surface of the copper
good correlation between the Young's modulus and plates) and then quenched in cold water.
the draw ratio achieved. The density of the isotropic sheets was measured at
The present paper discusses: (a) the effect of molecular * G. Capaccio and I. M. Ward, Br. Pat. Appln. 10746/73 (filed
weight on the morphology of the initial material and 6 March 1973).

POLYMER, 1974, Vol 15, April 233

Ultra-high modulus finear polyethylenes: G. Capaccio and I. M. Ward
Table 1 G.p.c. molecular weight data of LPE samples used I f a few samples are now taken along this cooling
path and quenched in cold water the morphology of the
Sample No. Rigidex grade Batch No. /~n Mw quenched material will be directly related to the morph-
1 9 BL 9022 6 060 126 600 ology at the final temperature on the slow cooling path
2 50 9099 6 180 101 450 and will therefore be a function of the temperature T~
3 25 6050/1 12 950 98 800 at which the slow cooling process has been suddenly
4 140-60 2H31/A/278 13 350 67 800 stopped. As a consequence we would expect to obtain
significant variations in am for different values of Tq. Figure
2 and Table 2 show very well the extent of these variations.
23 °C in a density gradient column using either isopropyl
All four samples when quenched at Tq= 160°C show
alcohol and diethyleneglycol or isopropyl alcohol and
a low am and this value should be quite typical for any
water as solvents. The mass fraction of the crystalline
sample quenched at a temperature higher than the
phase (~ra) as a function of the sample density (p),
observed crystallization temperature. For Tq=II0°C
the density of the crystalline phase (pc) and the density
a dramatic increase in O~mis observed; the crystallization
of the amorphous phase (pa) was calculated from the
process has, at this stage, involved most of the molecules
equation:
available, and the differences between different samples
~,. = [(p - p=)/(pc- p~)]P¢
P
assuming a value of 1.000g/cm 3 for pc and a value in 150
the range 0.855-0-870g/cm 3 for p=6-s.
Dumbbell samples of dimensions 2 cm x 0.2 cm were
cut and drawn on an Instron Tensile Testing machine
at 75°C at a draw speed of 10cm/min for 60-90see. ~ 13C
Only sample 2 had to be drawn at a speed of 1 cm/min
to attain ,~=30. The draw ratio was measured from
marks on the surface of the undrawn samples spaced
at intervals of 0.2 or 0.1 cm. In general at very high
draw ratio (>25) only 30-50 % of the drawn segment
~ 90
showed a homogeneous draw.
The melting curves were measured on a Du Pont
Thermal Analyzer with a Du Pont DSC cell and operated
7G
in a differential scanning calorimetry (d.s.c.) heat mode.
I I I
Untreated aluminium pans were used. The temperature
was measured by chromel/alumel thermocouples and 0 5 IO 15
Time (mini
corrected in accordance with the Du Pont manual. The
experiment was carried out in an atmosphere of nitrogen Figure I Cooling curve of an isotropic film of sample 4 com-
at normal pressure and the gas flow was (1-3)x 103 pression moulded at 160°C
cma/min. The samples were heated up at a rate of
5 °C/min from a starting temperature of 90°C.
The room temperature Young's modulus was measured O-970!
by a dead-loading creep experimenta, taking the 10 sec
response at stresses (~) corresponding to strains (E)
below 0.01.
O.96~

RESULTS AND DISCUSSION E

Pre-treatment o.
0.962
The cooling pattern of the four samples, under the
same conditions, is very similar and the cooling curves c

do not show any appreciable differences; as an example,

the curve for sample 4 is plotted in Figure 1. O.
0.958
All the samples show a phase transition in the region E
120-123°C which is associated with the main part of
the primary crystallization of the polymer. The tem-
0.954
perature range in which this occurs is in good agreement
with predictions obtained by comparing the molecular
weight distribution of the samples and their rate of r'm

cooling with the crystallization rate data published for 0-950

LPE fractions isothermally crystallized at high super- o

cooling10, zz. One therefore expects to obtain a polymer I I I I I I I 0 I

of low crystalline fraction e~ra at temperatures higher 160 120 80

than 123 °C whereas at lower temperatures the crystalline Quenching temperature, T9(°C}
fraction should be much higher and keep increasing Figure 2 Density p of the isotropic samples as a function of the
slowly with decreasing temperature owing to further quenching temperature Tq. r--I, Sample 1 ; I , Sample 2; O, sample
crystallization. 3; 0 , sample 4

234 POLYMER, 1974, V o l 15, A p r i l

 Ultra-high modulus linear polyethylenes: G. Capaccio and I. M. Ward
Table £ Crystalline mass fraction c=m as Drawing behaviour
a function of Tq for different samples.
Values calculated for pa=0"855g/cm3 (bot- The drawing behaviour of these samples supports
tom) and pa=0"870g/cm 8 (top) this interpretation of the initial morphology and a
close analysis of the variation of the yield stress (~y)
Sample No. and draw ratio with Tq provides more information on
Tq (°C) 1 2 3 4 the possible morphological features of some of the
samples.
160 0"67 0"66 0"65 0-65
0"71 0"70 0"69 0"69
The yield stress data do not show any dramatic
110 0"74 0"76 0"75 0"76 dependence on molecular weight and molecular weight
0"77 0"79 0"78 0"79 distribution. The oy values for all four samples are
20 0"79 0"78 0"77 0"78 fairly close and are only a function of Tq. As an example
0"82 0"81 0"80 0"81 of the trends and fluctuations the values of the yield
stress for samples 3 and 4 are shown in Figure 3. An
can then be related, in a qualitative manner, to the initial increase of over 50 % is observed in the region
differences in molecular weight. Tq=130-110°C after which the yield stress levels off
Previous work has shown the effect of molecular and no temperature dependence is detectable below
weight on both the crystallization rate 10, 11 and the ll0°C.
equilibrium crystallinitylL The crystallization rate vs. The initially low value of the yield stress at high
molecular weight curve for isothermally crystallized quenching temperatures may be related to the more
LPE fractions at high supercooling, for example 121 °C, compliant behaviour of samples of low crystallinity. The
goes through a maximum 11 at a molecular weight of development of further crystallization, at and just
about 2 × 104 and is asymmetric, being very sharp on below the observed crystallization temperature, is
the very low molecular weight side and fairly smooth likely to produce a net increase in the stiffness of the
on the high molecular weight side. On the other hand, material which can account for the observed increase
the equilibrium crystallinity is constant at low molecular of the yield stress. This reaches a maximum value at
weights, but fails off monotonically~2 at molecular weights T~=ll0°C when presumably an optimum morphology
greater than about 104. Although these results were has been achieved. In fact, the yield stress might be
obtained for fractionated polymers, there are indications expected to reflect the presence of two conflicting
that in unfractionated polymer fractionation occurs effects; the increase in crystallinity which results in a
during crystallization13, and so it would be expected stiffening of the structure and a segregation process, by
that they would still be of relevance to the present in- which non-crystallized low molecular weight molecules
vestigation. are more and more segregated at the spherulite bound-
In our experiments the effect of crystallization rate aries. In principle, therefore, the ultimate stage should
is probably predominant. However, the results can be be represented by a very weak structure in which large
interpreted either in terms of kinetic effects alone or spherulites are well separated by highly compliant low
of a combination of kinetic effects and equilibrium molecular weight material. According to this picture
crystallinity. Thus, sample 1 with its long very high the plot of cru vs. Tq should go through a maximum
molecular weight tail as well as very low molecular which is not observed in our results but which has been
weight tail in the molecular weight distribution curve found in other polymers 14.
has the lowest am. Sample 2, on the other hand, with It does, however, appear that more complex effects
the same /fftn but a narrower distribution exhibits a are seen in the strain hardening behaviour, as revealed
relatively higher am. by the pattern of draw ratio with quenching temperature.
Sample 3, with a higher 3~tn but with 3~w fairly close In particular, the draw ratio shows a maximum value
to that of sample 2, has a lower value of am than the (~max) as a function of quenching temperature at
latter because of its molecular weight distribution which Ta = 110°C for samples 2, 3 and 4, whereas for sample 1
is shifted as a whole towards higher molecular weight. the draw ratio remains constant for Tq < 110°C.
Sample 4 if compared with sample 3, of almost identical In Figure 4 the average values of several determinations
3~rn, shows the expected increase of am for a narrower
molecular weight distribution and reaches the value
190
measured for sample 2. The equivalence of samples 2
and 4 can therefore be attributed to the combination
of effects acting on two opposite directions; first, the "C
narrowing of the molecular weight distribution, which 2 iso
reduces the proportions of both very low and very high
molecular weight material and secondly, the overall
shift of the molecular weight distribution towards
higher Mn.
When slowly cooled down to room temperature
(Tq =20°C), a very small increase in am is observed in
samples 2, 3 and 4, whilst sample 1 reaches a value
higher than the others. This unusual behaviour can be
P llO

70
J O

160 ' 12'O ' 8'0 ' 4'0 '

attributed to the continuous crystallization, even at
very low temperatures, of very low molecular weight Quenching tzmpcroture 7"9(°C)
molecules 7 which are present in substantial quantity Figure3 Yield stress ay as a function of quenching temperature
in this sample of low 3~tn and very broad molecular Tq for samples 3 (O) and 4 (Q) drawn at 75°C at a draw speed
weight distribution. of 10cm/min

POLYMER, 1974, V o l 15, A p r i l 235

 Ultra-high modulus linear polyethylenes: G. Capaccio and I. M. Ward
40 Thermal analysis
A preliminary investigation of the texture of the
drawn samples has been carried out by examining their
melting behaviour (Figures 5 and 6). Samples 1 and 2
32

I AT=IOC

o" 24
4,J
2 0

E 0

"10
X=8
p_,=
0
z
16
i
i I10 0
o J X = 14
,Sa mpie

13 III
8 []

I I I I I , I
X=I3
160 140 120 I00
Quenching temperature Tq(OC} X =26 ,Sample ;
Figure 4 Variation of draw ratio A with quenching temperature
Tq for samples drawn at 75°C at a draw speed of 10cm/min.
FI, Sample 1 ; II, sample 2; O, sample 3; O, sample 4 X =29
I I I I I
of the draw ratio are shown for the four samples as a II0 130 150 170
function of the quenching temperature in the temperature Temperature (oc)
range 160-110°C. The results for Tq<ll0°C, which Figure 5 D.s.c. traces for samples 1 and 2 at different draw
are not shown in the Figure, showed much greater ratios A
scatter, but except for sample 1, they consistently
indicated that the draw ratio reached a value which is
intermediate between Amax and the draw ratio for
Tq = 160°C. AT= Ioc
These results also clearly show for the first time the
effect of molecular weight and molecular weight distri-
bution on the draw ratio. A sharper maximum in the
draw ratio as a function of quenching temperature
appears to be characteristic of the samples with higher
Mn, i.e. samples 3 and 4. At the same time, by comparing
samples 1 and 2 with samples 3 and 4, we can see that %
there is a spectacular effect o f / i w . At a given value
o f / i n , a substantial increase in draw ratio is always - - X=ll[Sample3
obtained by narrowing the molecular weight distribution
{
---- x=2oj
i.e. by reducing/iw. U
At high draw ratios there is an excellent correlation o
between draw ratio and the crystallinity of the undrawn
material, the highest draw ratios having the highest
crystallinity. This suggests that both are, to some
extent, dependent on the same parameter--the concen-
tration of very high molecular weight material. This
view is supported by the behaviour of sample 1, which X=13
has the highest /iw, and would therefore be expected
to possess many interlamellar tie molecules TM 16. This
sample shows no change in draw ratio for quenching X=2C ~Sample 4
temperatures below 110°C.
However, very high molecular weight material can
be only partly responsible for these effects. The presence
of a maximum in the draw ratio for the other three ~ x-34
samples (2, 3 and 4) suggests that other factors, such 1 l l I I I

as the extent of segregation and the nature of the material i iO i30 150 17O
segregated during the slow cooling of the isotropie Temperature (oc)
sheet, also play a significant role in the plastic deforma- Figure 6 D.s.c. traces for samples 3 and 4 at different draw
tion process. ratios A

236 POLYMER, 1974, V o l 15, A p r i l

 Ultra-high modulus linear polyethylenes: G. Capaccio and I. M. Ward
show an increase in the melting temperature Tm
(Tm~133°C for the isotropic material) with increasing
draw ratio. At its highest attainable draw ratio sample 2 570
melted at ~137°C. Much more remarkable behaviour
is shown by samples 3 and 4 even at lower draw ratio.
Sample 4 always shows a melting peak at 138.5°C
¢,.
whereas sample 3 gives a single peak at 137.5°C at O
..Q
A= 11 which splits into a double peak at A= 20 with a
substantial fraction melting at 138-5°C. This shows ~ 450

that the plastic deformation of samples 3 and 4 has

produced crystalline material with a crystallite thickness 23

greater than ~ 1500A 17, is, conventionally termed ex- o

E
tended chain crystals.
The fact that only samples 3 and 4 exhibit this peculiar "5 330
c
behaviour suggests that the formation of extended chain 0
¢-
material is related to a critical molecular weight range,
i.e. a critical value o f / f i n . For very narrow molecular
weight distributions (sample 4) the result becomes
completely independent of the initial morphology and 210
/
draw ratio. With a broader molecular weight distri-
bution, as obtained with sample 3, the phenomenon
occurs only partially and only at high draw ratio. It
appears therefore that extended chain samples of LPE
can be easily prepared by cold drawing provided that 90
polymers of suitable molecular weight and molecular
weight distribution are chosen. The present results do
I I I I I I I
not rule out the possibility that even at values of 3~n II I9 27 35
which are lower or higher than ~ 1 3 0 0 0 the same Draw ratio, X

Figure8 Young's modulus, calculated from the 10sec isochronal

stress-strain curves, as a function of draw ratio ,~, at various
strains, for samples 3 (O) and 4 (0). A, E=0.001 ; B, ~=0.002;
69C
C, ~=0"005

effect may be obtained with an appropriately matched

molecular weight distribution.
We are aware that this particular aspect of the problem
57(
requires further detailed discussion and will present a
t-
O
more extensive study in a future paper.
r~

~o Mechanical properties
The room temperature creep behaviour was studied
2 45( C
for each sample. From the creep response, the 10sec
O
E isochronal stress-strain curves were constructed, so
that values of the Young's modulus could be obtained
c
0
for various strains. The results are shown in Figures 7
c and 8 as a function of sample draw ratio.
33( A complete comparison cannot be obtained from the
i,i results shown in Figure 7 because sample 1 could be
drawn only to a relatively low draw ratio. However,
taking the results shown in Figures 7 and 8 overall, it
is clear that the modulus is, to a first approximation,
210 related only to draw ratio, at least for ,~<30. This
result is consistent with that obtained previously at a
lower range of draw ratios and modulP. It is possible
that results shown in Figure 8 may be indicative of a
small effect due to molecular weight at draw ratio 20.
90 The results show also that the achievement of high
modulus does not necessarily require the presence of
extended chain crystals. This result is consistent with
19 ' 2'7 35 the view that the non-crystalline phase is likely to give
Draw ratio~ X the major contribution to the overall compliance.
It is therefore reasonable to expect that the macro-
Figure 7 Young's modulus, calculated from the 10sec isochronal scopic deformation would not be affected by any
stress-strain curves, as a function of draw ratio ,~, at various
strains, for samples 1 (R) and 2 (11). A, E=0.001; B, c=0.002; change in the crystalline structure, as distinct from the
C, ~=0"005 fraction of crystalline material or the degree of molecular

POLYMER, 1974, Vol 15, A p r i l 2,37

Ultra-high modulus linear polyethylenes: G. Capaccio and I. M. Ward
orientation. On this view the concentration of fully extended A highly oriented material of very high stiffness
chains in the non-crystalline regions is mainly responsible (Young's modulus>600kbar) has been prepared. It
for the stiffness of the polymer, i.e. oriented tie molecules. appears that the presence of extended chain crystals,
It is also possible that extended chain crystals could formed during the cold drawing process, is not directly
play a similar role to oriented extended chain tie mole- relevant to the value of the extensional modulus achieved.
cules in contributing to the overall stiffness.
It would appear that there are also some molecular
weight effects on the modulus. Sample 2 at a draw ACKNOWLEDGEMENTS
ratio of 30 has a higher modulus than sample 4 at the We wish to thank Mr J. D. Mellor for his assistance in
same draw ratio. This suggests that the presence of the preparation of the samples, and Dr F. Jones, Depart-
high molecular weight molecules is important, because ment of Colour Chemistry, Leeds University for gener-
although sample 4 has a higher 2fin value than sample 2, ously undertaking the thermograms. We also wish
it has a narrower molecular weight distribution. The to acknowledge that one of us (G. C.) was supported
flattening of the modulus/draw ratio plot for sample 4 by the Science Research Council.
supports the view that the amount of high molecular
weight material may in the limit determine the highest
modulus attainable. REFERENCES
Orientation studies by wide angle X-ray and broad
line nuclear magnetic resonance methods are presently 1 Frank, F. C. Proc. R. Soc. (A) 1970, 319, 127
2 HoUiday, L. and White, J. W. Pure AppL Chem. 1971, 26,
being carried out in an attempt to resolve some of 545
these problems and will be reported very soon. 3 Meinel, G. and Peterlin, A. J. Polym. Sci. (A-2) 1968, 6, 587
4 Gupta, V. B. and Ward, I. M. 3". MacromoL Sci. 1968, B2, 89
5 Andrews, J. M. and Ward, I. M.J. Mat. Sci. 1970, 5, 411
CONCLUSIONS 6 Glenz, W., Morosoff, J. and Peterlin, A. J. Polym. Sci. (B)
1971, 9, 211
The present study has dearly shown the effect of initial 7 Chiang, R. and Flory, P. J. J. Am. Chem. Soc. 1961, 83, 2857
morphology and molecular weight on the cold drawing 80kada, T. and Mandelkern, L. J. Polym. Sci. (A-2) 1967,
behaviour and mechanical properties of LPE. 5, 239
The results presented indicate that, under the experi- 9 Gupta, V. B. and Ward, I. M. J. Macromol. Sci. 1967, B1,
373
mental conditions adopted, an optimum morphology 10 Mandelkern, L., Fatou, J. M. G. and Ohno, K. J. Polym.
must be obtained to achieve very high deformation Sci. (B) 1968, 6, 615
ratios. 11 Fatou, J. M. G. and Barrales Rienda, J. M. J. Polym. Sci.
We suggest that the extent of nucleation in the crystal- (A-2) 1969, 7, 1755
12 Fatou, J. G. and Mandelkern, L. J. Phys. Chem. 1965, 69, 2,
lization process, crystallite growth and segregation of 417
low molecular weight material are all likely to influence 13 Anderson, F. R. J. Polym. Sci. (C) 1963, 3, 123
the critical morphology. All these parameters are 14 Way, J. L. and Atkinson, J. R. Plastics and Polymers in press
molecular weight dependent. In this investigation the 15 Mandelkem, L. J. Polym. Sci. (C) 1967, 18, 51
16 Steidl, J. and Pelzbauer, Z. J. Polym. Sci. (C) 1972, 38
effects of AStn and the ratio -,Qw/.t~n have been clearly 17 Meinel, G., Morosoff, M. and Peterlin, A. J. Polym. Sci.
demonstrated with respect to their influence on draw (A-2) 1970, 8, 1723
ratio and final modulus. 18 Bassett, D. C. personal communication

238 POLYMER, 1974, Vol 15, April