Вы находитесь на странице: 1из 5

Swelling of a rubber cylinder

in torsion: Part 2. Experimental


K. M. Loke, M. Dickinson and L. R. G. Treloar
Department of Polymer and Fibre Science, University of Manchester Institute of
Science and Technology, Sackville Street, Manchester M60 1QD, UK
(Received 14 September 1971)

Experiments have been carried out on the effect of torsion on the swelling of a
rubber cylinder in a low molecular-weight liquid, the object being to test the
theory given by Treloar. Both natural rubber and a synthetic poly(cis-isoprene)
were used, with toluene as the swelling liquid. Application of torsion resulted
in a reduction in the swelling by an amount approximately proportional to the
square of the torsion, as predicted theoretically. The values of this reduction,
however, were less than the predicted values by amounts varying from 12% to
23%, depending to some extent on the type of rubber and on the degree of
crosslinking. These discrepancies are believed to arise mainly from experimental
difficulties.

INTRODUCTION o! a dumbbell having substantial conical ends terminated


by square blocks for the application of the torsional stress
In the previous paper z the theory of the effect of torsion (Figure 1). The junction between the central cylindrical
on the equilibrium swelling of a crosslinked rubber was portion and the conical end was rounded off to avoid a
developed. It was shown that the expected effect is a stress concentration in this region. The internal dimen-
reduction of the degree of swelling by an amount which is sions of the clamp (Figure 2) were larger than the cor-
approximately proportional to the square of the torsion. responding dimensions of the unswollen test-piece, but
The present paper is concerned with the experimental smaller than its freely swollen dimensions; this, together
examination of this problem in relation to the theory
developed.
Since the expected effect depends on the square of the
torsional strain, it is important for the purpose of obtain-
ing an accurate result that this torsional strain shall be as
large as possible. This involves considerable practical
difficulty, since under conditions of prolonged loading in
the highly swollen state the strength of the rubber is
greatly diminished. Also, in order to avoid buckling in the
twisted state it is necessary to apply some extension in the
axial direction, and this further increases the total stress
to be borne. In practice the maximum torsion which could
be applied before the specimen ruptured corresponded to a
value of the parameter ~a (surface shear strain) of about 112mm
0-8, which yielded a change of swelling due to twist of
about 8 ~ . This, although not as large as it had been
hoped to obtain, was sufficient to provide a fair basis for 96mm
comparison with the theory.
The experiments were carried out on both natural 71mm
rubber (smoked sheet grade I) and 100~o poly(cis-l,4-
isoprene), swollen in toluene (Analar grade). The vul-
canizates were compounded according to the recipes given
in Table 1 using cold mixing on an open mill.
b L 5.Omm diam
The design of the test-piece and of the clamps took into
account the necessity to avoid high stress concentrations
in the region of the clamps. The test-piece was in the form Figure 1 Test-piece dimensions: (a) perspective; (b) elevation

POLYMER, 1972, Vol 13, May 283


Swelling of a rubber cylinder in torsion (2): K. M. Loke et al.

with a rounding-off of the edge of the clamp from which enabled the required amount of torsion and axial exten-
the rubber emerged, was intended to reduce the stresses at sion to be applied while the specimen was immersed in
the edge of the clamp, while retaining a positive pressure the swelling liquid. Rotation of the clamp A was effected
on the conical region of the sample. by means of the worm-gear mechanism C, and changes in
The degree of swelling of the cylinder in torsion was axial length by the bevel gear D, which enabled the car-
obtained from measurements of axial length and diameter riage B to be shifted horizontally (by rotation of the
obtained by means of a Pye 2-way travelling microscope threaded shaft F). The assembly was mounted in a metal
reading to 0.005 mm. The accuracy of the swelling mea- trough filled with the swelling liquid up to a level at which
surement was limited by the accuracy with which the the clamps were immersed. The trough was covered with a
diameter of the specimen could be measured. For the sheet of glass to reduce evaporation, and stood in an outer
highest accuracy it is desirable that this diameter should water bath thermostatically controlled at 25 _+0.05 °C.
be as large as possible; however, the greater the diameter The measurements of axial length were obtained from
the longer is the time required to attain equilibrium reference marks 3 to 4 cm apart inscribed on the surface
swelling and the greater, therefore, the probability of of the specimen as shown in Figure 1. A rubber embossing
fracture. The diameter chosen (Smm) represented a ink was used for the natural rubber/sulphur vulcanizates,
practical compromise, and gave a time to reach equili- whilst for the dicumyl peroxide cured samples of both
brium of between 10 and 20 hours. natural rubber and Natsyn an ordinary 'biro' ink proved
The moulding presented a problem. The mould was satisfactory.
split in the diametral plane of the cylinder. Closure of the The specimen was allowed to swell for 2 to 3 days, with
two halves caused the trapping of a variable amount of periodic adjustments of axial length, in order to achieve
rubber in the 'bite'; this resulted in a non-circular section swelling equilibrium without 'buckling' of the cylinder.
of the final test-piece. To overcome this difficulty mould- This procedure was adopted after it has been found that
ing was carried out by a transfer moulding process which stretching of the unswollen cylinder to the final length
involved injection of the rubber into the closed mould followed by immersion in liquid resulted in early rupture.
through a small hole in the end-plate, excess 'flash' being A considerable tension was required to achieve the
extruded through a similar hole at the other end. Even desired extension of the dry rubber, and in the presence of
with the precaution, however, significant deviations from the liquid the strength was so much reduced that rupture
circularity of the cross-section were not entirely eliminated occurred before this tension could be sufficiently relieved
and the experimental procedure had to be designed to by swelling. In the method used here elongation proceeded
minimize their effects (see below). concurrently with swelling and high stresses were avoided.
(A similar principle was adopted by Flory and Rehner3 to
overcome this difficulty.)
METHOD OF MEASUREMENT After a final measurement of dimensions the torsion
was then applied, keeping the length between clamps
The specimen was mounted between clamps A and B constant. Because of the slight irregularity in shape of the
attached to the apparatus shown in Figure 2. This cross-section of the sample already referred to, the

I_.
-
377mm
-]
..j

C ' A B E ,_

j
r--'-.. !
_ ::~]~1~111/111/11,1111111111111111~11111~111~111 "
F
I(o
I
~11~1~1/11111111111/111/111111111/111/1/111~111~11111111111~J
T
76 mm

a
E l

c•C A B D

1--J . ~ . ~ . . - ~IDlli111111tI111lllIl,l,]llIIl~IIi~11tlll I |~I111111111111lI11111111111111111111lI111111lI1111~I111111~II

E
b

Figure 2 Apparatus for combined torsion and extension: (a) plan; (b) elevation

204 POLYMER, 1972, V o l 13, M a y


Swelling of a rubber cylinder in torsion (2) : K. M. Loke et al.

Table 1 Rubber compounding ingredients (parts by weight)

Sample A B C D E F G

Smoked sheet, Grade 1 100 100 100 100


Natsyn 100 100 100
Dicumyl peroxide 3-2 3.5 3-0 3"0 3"0 3"0
M.B.T.S. Vulcafor (accelerator) 0'5
Zinc oxide 0.5 2 0.5 0.5
Stearic acid 0"5
Carbon black 0.25
N-phenyl-2-naphthylamine (antioxidant) 0.5
Sulphur 6
Cure time (min) at platen temp. of 140°C 35 30 33 55 30 32 30
Mc (swelling) 8891 8184 12 380 8313 11 294 11 312 14 566

swollen volume corresponding to the untwisted state was The successive twisted states for any one sample did not
obtained from measurements of the dimensions of the correspond to exactly the same axial length, or axial
swollen cylinder immediately after application of the extension ratio flz, between reference marks. Since the
torsion. In the time required to carry out this measure- objective is to derive the effect of torsion on equilibrium
ment ( ~ l min) any change of swelling would have been swelling, at constant axial length, it was necessary to apply
very small. This was confirmed by checks against the a correction to the measured swollen volume to obtain the
volume before twisting; the differences (volume after twist- equivalent degree of swelling at a fixed value of/33, for
ing minus volume before twisting) ranged from +0.039 which the mean over the complete range of torsion values
to -0.011 cm a, with an arithmetic mean of +0.0141 or was taken. The required correction was obtained by dif-
+0.3%, i.e. in the opposite sense to the subsequent ferentiating the following equation (equivalent to equation
change of swelling. The important consideration was that (i 1) of Part 1x) with respect to v2:
by focusing always on the same diametral section of the
sample (midway between the reference marks) the effects ln(1 - vz) + v2 + XVZ~= - p V l / Mcfl8 (2)
of any irregularity of section were avoided. Measure-
and writing ~ V/V= - 8v2/v2. This gives:
ments were made periodically until a limiting value of
swelling was reached, after which the twist was removed. .... pVl/Me 3o (3)
The sample was then allowed to swell overnight to v / v - - [2X-Z i i ( l - vz)]v~fl~ p3
equilibrium in the untwisted state (at constant axial
length). A new twist was then applied and the above This formula is strictly valid only for zero torsion, but
procedure repeated. The torsional strain was obtained was assumed to be sufficiently accurate for the present
from measurement of the axial length of one or more purpose, the corrections involved being quite small
complete turns of twist, using the mould closure line as (~V/VxO.6% at most).
reference mark.
A small irreversible increase in swelling occurred over
the entire period of experimentation, lasting from 10 to TREATMENT OF EXPERIMENTAL DATA
14 days. However, this slight irreversibility, in which suc-
cessive equilibrium volumes in the untwisted state dif- The experiments were carried out on 4 natural rubber
fered by less than 0.5 %, had no significant effect on the vulcanizates and 3 Natsyn vulcanizates, each group
value of relative volume change obtained over a single run comprising a range of chain molecular weights Me
occupying from 8 to 13 hours. (Table 1). Detailed consideration of the data, and of the
methods of calculation employed, will be presented for
one sample only--the Natsyn 'A' sample. Corresponding
CHARACTERIZATION OF SAMPLES results for the remaining samples will be limited to the
comparison of theoretical and experimental volume
The mean chain molecular weight Me may be determined changes on twisting.
from measurements of either modulus or swelling
equilibrium. Since the relevant problem is concerned with
swelling, it was more appropriate to use the swelling Derivation of Me
equilibrium for the purpose. The cylindrical portion of the For the Natsyn 'A' vulcanizate the measured free swell-
rubber sample was cut out and free-swollen in toluene to ing equilibrium at 25°C corresponded to a volume frac-
equilibrium (2-3 days), after which it was removed, tion (vz) of rubber of 0.1893. The measured density of the
surface-dried, and weighed. Weighings were then made at rubber was 0-9016gcm -3, and of the solvent 0-862. The
5 h intervals during drying, which required 3 days. These molecular weight of toluene being 92.13, this gives a
weighings, together with the measured densities of the dry molar volume (V1) of 106.9. Taking X=0.40 (as for natural
rubber and of the toluene, enabled the volume fraction of rubber in toluene") substitution of these data in equation
rubber (v2) to be obtained (assuming additivity of (I) yields the result Me=8891.
volumes). The value of Me was then calculated from the
equation:
Attainment of equilibrium
i n ( l -- V2)+ v2+XV~ = - ( p V 1 / M c ) v l/3 (I)
Figure 3 shows the course of the volume changes sub-
Values calculated in this way are given in Table 1. sequent to the introduction of a particular value of twist,

POLYMER, 1972, Vol 13, May 205


Swelling of a rubber cylinder in torsion (2): K. M. Loke et al.

Since there is the danger here of a circular argument,


4-26 1 this procedure needs some justification. We note first that
the value of Me is determined from free swelling, hence

~" 4.18
! T the calculation of v2 for the swollen stress-free state is
simply a conversion back to an original swelling measure-
ment. The only truly theoretical part of the calculation is
therefore the effect of the axial extension on the swelling.
~ 4-10 For the case considered the value of v2 in free swelling
was 0.1893, corresponding to a change of linear dimen-
sions in the ratio 1-742 : 1. The relatively small exten-
4"02 sion to fla = 1.938 gives a calculated change of v2 to 0"l 813,
I I I I I
i.e. a reduction in v2 of 4.4%. It is only this 4.4% which
0 2 4 6 8 IO 12
relies on the applicability of the theory, and on the basis
Time (h } of previous experiments on simple extension 4 it would
seem reasonable to assume that this will not be in error by
Figure 3 Change of volume with time after application of torsion. more than, say, 5 % of itself, or 0.2 % of the whole swollen
Natsyn sample A, ~bao=0. 4383 volume. This would produce an error of the same amount
(0.2 %) in ao2, which is practically negligible.

1.00
COMPARISON WITH THEORY

~: o.9a The calculated values of the change of volume due to


torsion (A V/Vo) were based on the theory discussed in the

0"96

a: 0.94

0.92
0 0
Time (h)
4 ~ JD,~ v

Figure 4 Relative change of volume with time for Natsyn A, for


following values of ~bao:A, 0.1806; B, 0.2706; C, 0.3609; D, 0.4383;
E, 0.5157; F, 0.5670 0

represented by 4~a0=0.4383. The volumes shown have # o


been corrected to take account of the small difference of
axial extension ratio fla, which in this case was 1.934, 4
from the mean, 1.938, for all experiments on this sample;
this correction varied from +0.0036 to +0.0038 cm a dur- ? o
ing the course of equilibrium. The total reduction of
volume due to torsion amounted to 0.2207cm a on an
4
original volume of 4.2467 cm a, giving AV/Vo = - 5.20 %,
for this particular twist.
The complete set of curves for the six values of twist 0
employed is shown in Figure 4. These have been plotted,
for ease o f comparison, in terms of relative volume, 4
referred to the untwisted state.
0
Derivation of ao
4
For comparison with the theory, which gives the change
of swelling most conveniently in terms of the dimension- I I I
less parameter ~ba0, it is necessary to determine the un- 0 0-05 0"10 0"15 0"20 0"2.5 0"30
swollen radius a0. Because of the non-circularity of the
cross-section this was obtained not by direct measure- ~2Oo2
ment of the dry rubber, but from the measured diameter in
the swollen untwisted state, at the extension ratio fla, Figure 5 Relative reduction of volume due to torsion for rubbers
together with the calculated value of v~ for this state. listed in Table 1. ~ , Calculated;---, observed

206 POLYMER, 1972, Vol 13, May


Swelling of a rubber cylinder in torsion (2): K. M. Loke et al.

previous paper 1. For sample A the modification referred would have the effect of reducing the true torsional strain
to in Part 1was used, in which the numerical integration and hence the change of volume.
was performed with respect to r 2, and the limit of the More generally, it is important to bear in mind that the
axial region (rl) coincided with the step length 8(r2). For effect of torsion on the swelling equilibrium, in contrast to
all the other samples the integration was performed with the effect of extension, is essentially a second-order effect,
respect to r, using a value of 8r of 10-3 and a value of rl proportional to the square of the strain. As a result it is
sufficient to give a difference between/31 and/3z of 10-6 likely to be affected to a disproportionate extent by any
(rl ~ 10-z). A check was made, however, to ensure that reduction of the true elastic strain, associated with stress
these two methods of calculation gave substantially relaxation or irreversible structural changes, or indeed by
identical results. The results are shown in Figure 5 in the any deviation of the material from the postulated
form of plots of - A 1I/Vo against ~2ao2. idealized elastic structure. Any such effects would be in
The experimental results show the expected form of the direction of a reduction of the magnitude of the change
dependence of swelling on torsional strain, though there is in swelling, in line with the experimental deviation.
a consistent quantitative discrepancy from the theoretical In view of these experimental difficulties the present
curves. Based on the maximum values of ~a0, this dis- observations cannot be regarded as providing conclusive
crepancy ranges from - l l . 7 % (sample A) to - 2 2 . 9 % evidence of a genuine difference between the theoretical
(sample F), with a mean of - 17.2 %. The discrepancy was and actual effects of torsion on the equilibrium swelling.
somewhat lower for the Natsyn samples ( - 14.1%) than
for the natural rubber samples ( - 1 9 . 6 % ) . There was a
slight indication of a correlation with Me, the smaller dis- ACKNOWLEDGEMENT
crepancies in either group being obtained with the samples
having lower Mc values. The authors acknowledge with thanks the provision of
These discrepancies are well outside the experimental maintenance grants by the Malayan Rubber Fund Board
variations. There may, however, be systematic errors. One (for K.M.L.) and by the Science Research Council (for
possible source of error might be the failure to attain the M.D.). They are also indebted to Dr M. D. Heaton and
true swelling equilibrium after twisting. A compromise is Dr R. F. T. Stepto for advice and assistance relating to
necessary owing to the gradual upward drift in swelling the computational aspects of the work.
referred to previously; after prolonged equilibration the
initial reduction is followed by a slow increase in volume.
(There is a slight indication of this in the top curve of REFERENCES
Figure 4.) The total volume change may on this account
1 Treloar, L. R. G. Polymer 1972, 13, 195
be slightly underestimated. Another error might arise from 2 Blanchard, A. F. and Wootton, P. M. J. Polym. Sci. 1959, 34, 627
the very slight residual buckling in the specimen at high 3 Flory, P. J'. and Rehner, J. J. Chem. Phys. 1944, 12, 412
twists, which could not be entirely eliminated. This also 4 Treloar, L. R. G. Trans. Faraday Soc. 1950, 46, 783

POLYMER, 1972, Vol 13, May 21}7

Вам также может понравиться