J, Rubb. Res. Imt.

Malaya, 22(2), 191-200 (1969)

Some Factors Affecting the Service Life of

Natural Rubber Articles
C.J. DERHAM, G.J. LAKE and A.G. THOMAS
Natural Rubber Producers' Research Association, Welwyn Garden City, Herts., U. K.

The paper is mainly concerned with failure under repeated stressing (fatigue) and creep
behaviour.
Fatigue failure is a crack growth process from small naturally occurring flaws. A critical
minimum severity of strain is necessary to produce any mechanical crack growth at all and this
governs the 'fatigue limit', the strain below which the fatigue life is very great. Factors which
influence this minimum severity include the molecular structure of the vulcanisate, the presence
of oxygen and anti-oxidants and the minimum strain during the strain cycle. Ozone attack can
be very important in fatigue at low strains.
Creep and stress relaxation are due to a combination of physical and chemical effects. The
former is affected by humidity, pre-stressing and the nature of the vulcanising system, whereas
the latter depends on the chemical stability of the vulcanisate.

Many factors can influence the useful life of a in engineering applications as it may be rapid
rubber article in service. It may, for example, and catastrophic.
become excessively hard or soft or become As the failure is essentially a crack growth
mechanically weak due to chemical changes. process from small flaws, the crack growth be-
These effects are countered by suitable com- haviour of the rubber and the size of the initial
pounding to improve the chemical stability. flaw are the essential factors determining fa-
The present paper is mainly concerned with two tigue life.The crack growth characteristics of vul-
factors which are of particular concern in en- canised rubbers have now been quite extensive-
gineering applications: mechanical failure due ly studied (THOMAS, 1958; LAKE AND LINDLEY,
to repeated stressing (fatigue), and the creep (or 1964 and 1965). The form in which they have
stress relaxation) which occurs under a constant been expressed has been derived from earlier
load (or deformation). In both cases, chemical work on tear behaviour (RIVLIN AND THOMAS,
as well as physical processes are important. 1953; THOMAS, 1960). This approach considered
the energetics of crack propagation. If a crack
FATIGUE FAILURE of length c in a strained test piece grows by a
Under repeated stressing, rubber components small amount thus producing an area dA of
may eventually fail. If the stressing is vigorous new surface, the amount of energy (U) in the
and the component is large, failure may be due strained test piece is reduced by oU. This lost
to excessive heat build-up, as may happen in energy can be considered as the energy available
large tyres at high speed. In this paper, however, for driving the crack through the rubber, and a
cases where the temperature rise itself is a major quantity 3 UfoA can be denned which can loosely
factor are not being considered. Failure at nor- be described as the driving force trying to make
mal temperatures under repeated stressing is the crack grow. This quantity is known as the
due to the growth of small cracks across the tearing energy and is denoted by T. It has been
specimen until finally rupture occurs. This is a shown that T governs the magnitude of the
mode of failure which must be guarded against stresses around the tip of the crack (THOMAS,
COMMUNICATION 512

191
 Journal of the Rubber Research Institute of Malaya, Volume 22, Part 2, 1969

1955; GREENSMITH et al, 1963), and also deter-

mines the amount or rate of crack growth that
occurs in the strained test piece (THOMAS, 1958).
There are several test pieces for which the rela-
tion between T and the applied forces or strains
can be calculated., so that, using these, the tear
behaviour can be described in terms of the
tearing energy. The relation between, say, rate
of crack propagation and T is then an intrinsic
property of the material and independent of the
particular type of test piece used to obtain it.
The test piece used for most of the cut growth
work is shown in Figure J, and for this the rela-
tion for T is
T~=2kcW ...(1)
where c is the crack length, W the strain energy
density in the bulk of the test piece (calculable Figure 1. Strip test piece containing edge cut
from the applied stress and the stress-strain of length c.
curve) and k a slowly varying function of the
strain, but which is approximately 2 for most strain energy density when the test piece is ex-
cases (RivLiN AND THOMAS, 1953; GREEN- tended to its maximum.
SMITH, 1963). Equation (3) has been found to represent
For natural rubber vulcanisates, provided quite well the dependence of fatigue life on
the value of T for catastrophic tearing Te is not strain over the expected range of applicability
exceeded, there is a negligible amount of purely of Equation (2). From parallel fatigue and cut
time dependent cut growth under a constant growth measurements the size of the initial
load. If the test piece is repeatedly stressed and flaw can be deduced to be about 2.5 x 10~3 cm.
completely relaxed, however, cut growth occurs The size of these flaws is clearly of great im-
during each loading cycle. The amount of crack portance in determining the life. Usually, al-
growth depends on the maximum value of T though not invariably, the worst flaws are at
attained each cycle. Thus the crack growth the surface and are associated with the prepara-
behaviour can be described by the relation tion of the sample by cutting or moulding.
between the cut growth per cycle, dcjdn, where However, particles of dirt are frequently about
n is the number of cycles, and this value of T. the above size and have been observed to ini-
It has been found for natural rubber that the tiate failure from the body of the material. It
seems probable that poorly dispersed com-
relation (THOMAS, 1958) pounding ingredients can also act in a similar
...(2) way.
G an = Natural rubber shows virtually no purely
adequately describes the cut growth behaviour time dependent crack growth under a constant
for T values between about 105 and 108 erg/ load, as mentioned above. This behaviour
cm2. Using this relation and Equation (1) it is appears to. be associated with strain-induced
easily shown that the fatigue life JV of a test crystallisation which also gives it its characteris-
piece cycled in simple extension and initially tic high tensile strength. Non-crystallising rub-
containing a flaw of size c0 is given by GENT bers such as SBR show time dependent crack
et al (1964): growth so that under even a constant load
failure may eventually occur.
Equation (2) is not valid at low tearing ener-
(2kW)*c0 "- gies. Figure 2 shows a typical crack growth re-
where, as for the test piece of Figure 7, Wis the lation for T values up to about 105 erg/cm2.

192

COPYRIGHT © MALAYSIAN RUBBER BOARD

 C. J. DERHAM et al.: Some Factors Affecting the Service Life of Natural Rubber Articles

Also, networks containing polysulphide cross-

links have generally higher T0 values than those
containing monosulphide or carbon-carbon
cross-links (LAKE AND THOMAS, 1967).
A test piece cycled below its fatigue limit in
air may in practice have a finite life due to
attack by ozone, which also produces crack
growth. Ozone crack growth behaviour is
shown schematically in terms of T in Figure 3.
Provided a very small T value (Tt) is exceeded
the rate of growth is independent of T but pro-
portional to the ozone concentration (AN-
DREWS et al, 1963). Presence of chemical anti-
ozonants increases Tz and reduces the growth
Tearing energy, I(P erg/em? rate above Tg. Unfortunately under repeated
stressing the enhancement of Tz does not occur,
Figure 2. Relation between tearing energy and due probably to mechanical rupture of a pro-
crack growth rate dc/dn for a natural rubber
tective surface bloom, but the rate reduction
vulcanisate at low tearing energies. The intercept remains. Ozone crack growth is additional to
at To gives the minimum tearing energy for any
crack growth to occur. that due directly to mechanical stressing and is,
of course, very important at low T values. The
crack growth behaviour under the combined
action of ozone and mechanical stressing can
be represented by
There is an intercept on the T axis, and below
this T value (T0) there is virtually no crack dc K. x (ozone concentration)
,..(4)
growth due to mechanical stressing (LAKE AND dn~ (frequency of cycling)
LINDLEY, 1965). Corresponding to TQ, there is a where 9 (T) gives the mechanical crack growth
fatigue limit for tensile stressing below which and the second term the ozone crack growth
the life is virtually infinite. For the normal flaw (LAKE AND LINDLEY, 1966). The parameter K
sizes this fatigue limit is usually between 50 and
100% tensile strain. Its magnitude is of consi-
derable practical significance as most rubber
components subjected to repeated stressing will
be designed to work below it. As soon as it is
exceeded the fatigue life drops to a few hundred
thousand cycles, which is often inadequate.
The magnitude of T0 is governed by the mo-
lecular structure of the vulcanisate and the
strengths of the chemical bonds, and can be
calculated approximately from a fundamental
molecular approach (LAKE AND THOMAS, 1967).
ID 2 I04 I0e
It is not significantly influenced by mechanical
Tearing energy, erg/cm2
hysteresis or crystallisation in contrast with
tensile and tear strength. T0 increases by about Figure 3. Schematic dependence of crack
a factor of two in vacuo, and some of this in- growth rate due to ozone on tearing energy. The
crease can be obtained in air by incorporating dashed curve shows the effect of adding a chemi-
suitable anti-oxidants such as PEN or certain cal anti-ozonant. The rate of crack growth above
paraphenylenediamines. Thus oxygen facili- the critical tearing energy is proportional to the
tates the rupture of the molecules under stress. ozone concentration.
193
 Journal of the Rubber Research Institute of Malaya, Volume 22, Part 2, 1969

depends on the particular vulcanisate and the growing crack which does not disappear on
form of the strain cycle. By using Equation (4), partial relaxation and thus produces an effec-
which can be determined by crack growth mea- tive strengthening of the material. The effective
surements, instead of Equation (2) the fatigue value of To increases and the rate of crack
life in the presence of ozone can be predicted growth at T values above it decreases. The
and compared with experiment, as shown in effect of incomplete relaxation on fatigue is
Figure 4. The agreement, using the flaw size much less in a non-crystallising rubber such as
deduced from high strain measurements of SBR as Table 1 indicates (LAKE AND LINDLEY,
2.5 x 10~3 cm, is satisfactory. 1966).
In practice, most spring applications do not
allow the rubber to be completely relaxed. This
can greatly enhance the fatigue life. This en- TABLE 1. EFFECT OF STRAIN
hancement appears to be due to the presence of CYCLE ON FATIGUE LIFE
strain-induced crystallisation at the tip of the
Fatigue life (kilocycles)
Strain cycle
NR SBR

0-250 13
50-240 13 000

Although the above work has been carried

out in tension the results are qualitatively
similar in other types of deformation such as
shear or compression of a bonded block. Fai-
lure in these cases is associated with tensile
components of the stress which tend to make a
crack open and thus encourage its propagation.
There are several factors of importance in
ensuring a long life for a component under re-
peated stressing. If complete relaxation occurs
each cycle, as for example in a tyre sidewall,
the deformation should not give T values above
To- The life is then likely to be governed by
ozone crack growth or production of large
flaws by accidental mechanical damage. If
possible, the component should be designed so
that incomplete relaxation occurs, and then
relatively large alternating deformations can be
sustained. Again, ozone attack is likely to be of
importance in determining the ultimate life.
Incorporation of anti-ozonants is beneficial
under static or alternating loads in reducing
Figure 4. Dependence of fatigue life on strain the ozone crack growth rate. If the component
in the presence of ozone. Curves are theoretical. has only intermittent dynamic use an anti-
Upper curve for ozone concentration of 0.3 ozonant may completely suppress growth dur-
p.p.h.m, lower (dashed) curve for concentration ing the static period.
of 0,8p.p.h.m.Points are experimental values: For components used in compression ozone
• 0.3 p.p.h.m., O0.8p.p.h.m. attack can be much less significant than mea-

194
 C. J. DERHAM et al.: Some Factors Affecting the Service Life of Natural Rubber Articles

surements in tension may suggest. A rubber straightforward. This question will be consi-
disc 2 cm diameter and 3 mm thick bonded to dered later.
metal on its major faces was compressed by Some of the factors affecting physical creep
16% and exposed to ozone at 600 parts per mil- and stress relaxation will now be considered.
lion for 25 hours. This is equivalent to about
fifty years at normal atmospheric concentra- (i) Humidity
tions. Although cracks grew, they penetrated It was apparent from stress relaxation mea-
only about 1 mm into the body of the rubber, surements carried out at room temperature that
equivalent to a rate of growth only about 450 the atmospheric humidity was influencing the
of that observed in tension. The reason is pro- results appreciably. A systematic study was
bably that the very small opening of the cracks therefore made. The sample, about 1 mm thick,
under this strain makes access of ozone difficult. was immersed in a suitable aqueous solution
This effect is fortunate as many rubber com- for at least 20 minutes, rapidly extended while
ponents for long term applications are used in still in the solution and the subsequent stress
compression. decay followed. The apparatus was thermo-
CREEP AND STRESS RELAXATION stated at 25°C.
Under constant load the deformation of a rub- Saturated aqueous solutions of various in-
ber sample increases steadily with time, the organic salts were used giving a range of relative
phenomenon of creep. At constant deformation humidities from 15% upwards. Check mea-
relaxation of stress occurs. The two effects are surements with air at various humidities showed
aspects of the same molecular processes and the that the salts themselves did not influence the
relation between them is governed by the force- relaxation and presumably did not enter the
deflection curve. If this is known, one can be rubber. Results for vulcanisates A (a conven-
calculated from the other (GENT, 1962a). tional CBS-accelerated sulphur mix given in the
Broadly speaking therefore, compounding and Appendix) and B (a peroxide vulcam'sate) are
other changes which improve one will improve shown in Figure 5 for 100 % extension. Varying
the other. the humidity has two effects:(a)the overall rela-
In springing and load-bearing applications xation rate increases with humidity; and (b) the
generally, creep is probably the main disadvan- shape of the curves varies systematically, those
tage of rubber as compared with steel, and in at high humidity curving upwards. It has been
severe conditions, such as at elevated tempera- reported that accelerated sulphur vulcanisates
tures, may limit the useful life. Understanding similar to this one give linear log(tkne) relations
of the phenomenon, prediction of the amount for physical relaxation (GENT, 1962a). It ap-
of creep over long periods and improvements pears that this is somewhat fortuitous and arises
in materials are thus of importance. because laboratory humidities are usually in
There are two components of creep: (i) a the range 30 -60% where this relation holds
physical effect which reflects the internal visco- approximately.
sity, and (ii) chemical effects which are due to The substantial magnitude of the effect of
cross-link and main chain scission. The physical humidity shows the desirability of control of
component contributes a term approximately this variable during testing and also the possible
proportional to the logarithm of the time after important effect of atmospheric humidity on
loading (GENT, 1962a) and thus gets progres- service performance.
sively less important at longer times. The chemi-
cal component is approximately proportional (ii) Vulcanising System
to the time, and at very long times becomes the Relaxation rates depend on the degree of
dominating effect (GENT, 1962b). The two com- cross-linking so that this must be held constant
ponents have different variations of rate with when other compounding variables are consi-
temperature so that extrapolation of creep rates dered (GENT AND MASON, 1963). The moduli of
to service temperatures is not completely the two rubbers shown in Figure 5 are similar,
195
 5-O
*N
92%

4-0 /•/ 79%

jj > S/C BS

///
J/ / 66%

47%

3-0 _,/ • 32%

s« // *X
//-/
15%
* ft* /
;• 79%

//// / Peroxide
2-0

10 100
Time, minutes

Figure 5. Effect of relative humidity on physical stress relaxation for natural rubber SjCBS and
peroxide vulcanisates A and B respectively. The figures by each curve give the relative humidity.
 C. J. DERHAM et al.: Some Factors Affecting the Service Life of Natural Rubber Articles

and it can be seen that the overall rate and sen-

sitivity to humidity are lower for the peroxide
4-0
vulcanisate. Because the shapes of the curves •—5
are somewhat different, quantitative compari- no pre-strain
son is difficult, but arbitrarily taking the relaxa- 100%
tion between 1 and 100 minutes the peroxide 3-5
pre-strain
vulcanisate has a rate about 60 % of that of the
accelerated sulphur vulcanisate. This effect is
present over a range of cross-link densities. 3-0 S/CBS + 40 parts
A correct choice of vulcanising system can lampblacK
therefore help in designing a compound with a
ic
low physical relaxation rate. Usually a com- 2-5
promise between different properties has to be
made, for example, the superior behaviour of
a peroxide vulcanisate in creep may have to 60 80 100 120 140
be balanced against its generally somewhat % extension
inferior strength properties. Figure 6. Creep rate as a function of strain for
(Hi) Effect of Pre-stressing vulcanisate C, containing 40 p.p.h.r. lampblack,
It is known that the presence of filler can in- for virgin test pieces and ones pre-strained to
crease the relaxation rate of natural rubber 100%.
vulcanisates by a factor of two or more, de-
pending on the nature and loading of the filler. This effect is not solely due to the filler. A
The stiffness of filled vulcanisates can be sub- corresponding gum rubber was pre-stressed to
stantially reduced by pre-stressing and there is a similar stress to that applied to the black-
some limited published data suggesting that filled rubber and the above experiments were
pre-stressing can also influence relaxation rates repeated with the results shown in Figure 7.
(GENT, 1962c).
The creep behaviour of vulcanisate C, a
CBS-accelerated sulphur mix containing 40 3-0
parts per hundred rubber (p.p.h.r.) lampblack,
was measured over a range of extensions using
fresh samples for each point and taking care 300%
not to over-strain them on applying the load. 2-5 pre-strain
Under the test conditions quite accurately
linear log (time) plots were obtained so that the
creep rate could be expressed as a percentage 8-
increase in deformation per decade of time, the 2-0 - Sulphur CBS gum
deformation at one minute being taken as the
initial value.
A number of test pieces were then strained to 100 200 300 400
100% for 1 hour and allowed to recover un- % extension
stressed for 24 hours so that subsequent re- Figure 7. Creep rate as a function of strain for
covery was negligible. They were then each an accelerated sulphur gum vulcanisate similar
tested at a strain in the range 60 to 150%. The in composition to the base material of the black-
results of these two series of tests are shown in filled vulcanisate C. Results for virgin and per-
Figure 6. It is apparent that there is a very sub- strained test pieces. The 300% pre-strain en-
stantial reduction in creep rate in the region of tailed a stress similar to that required to pre-
the pre-strain value. strain vulcanisate C to 100%.

197
 Journal of the Rubber Research Institute of Malaya, Volume 22, Part 2, 1969

The pre-strain in this case was 300%. The re- AC, •}

logt+3-46xlO" J (t-l)
sults are qualitatively remarkably similar to ea
those for the black-filled rubber suggesting that
part at least of the effect is due to changes in the
gum matrix. A similar suggestion has been
made for the stress-softening observed with
filled rubbers (HARWOOD et al., 1967).
The effects described above emphasise the
difficulties of making valid assessments of creep
behaviour. They also suggest that a funda-
mental molecular interpretation of the relaxa-
tion of natural rubber vulcanisates of the types
discussed is likely to be difficult,
Chemical relaxation processes, which are
relatively important at long times, will be
I 10 100 1000
superimposed on the physical component. To
Time, mlnules
estimate the long-term relaxation at service
temperatures it is necessary to separate the two Figure 8. Creep at 70°C of an accelerated
components and to examine their temperature sulphur vulcanisate. Full curve is a relation of the
dependence individually so that a reasonable form Ae/eo=[Alog(t/to)+B(t—to)] where A
extrapolation may be made. and B represent physical and chemical creep
For a vulcanisate whose physical creep is components respectively. Their values have been
proportional to log (time) the total fractional chosen to give the best fit to the experimental
creep £\eje0 is well described by the equation points.
- +B(t-t0) ..-(5)
where to is the time after loading of the first with vulcanising system and anti-oxidant, and
reading, when the strain is ea and t\e the in- the benefits of incomplete relaxation of stress
crease in strain at time /. The agreement with during cycling are all important. Attack by
experiment is shown in Figure 8. The para- ozone is particularly important in tension, but
meters A and B give respectively the physical for compressive stresses may be much less so.
and chemical components. Many engineering components, such as en-
The relative importance of physical and gine and anti-vibration mountings and bridge
chemical relaxation cannot be readily expressed bearings, are used primarily in compression
in terms of their amounts as the amount of and have relatively small surface to volume
physical creep depends strongly on the time of ratios. In these cases not only is ozone attack of
the first measurement after stretching. It is less concern, but also the attack on the rubber
therefore better to compare the two rates of by mineral oils may be unimportant as diffusion
relaxation, and on this basis it appears that for may limit the penetration very substantially
a typical vulcanisate the rates become compa- (SOUTHERN, 1967).
rable after one to two years at 10°C.
There are also indications that chemical
CONCLUSIONS creep due to oxidation, which is often the most
It is usually important to make the chance of important chemical creep process, is reduced in
an engineering component failing by fatigue bulky components (CUTHBERT, 1967), probably
remote, as such a failure may be rapid and again due to diffusion effects.
catastrophic. The work described above enu- This work forms part of the research pro-
merates some of the relevant factors. The gramme of the Natural Rubber Producers'
magnitude of the fatigue limit, its variation Research Association.

198
 C. J. DERHAM et al.: Some Factors Affecting the Service Life of Natural Rubber Articles

REFERENCES LAKE, G.J. AND LINDLEY, P.B. (1964b) Ozone cracking,

flex cracking and fatigue of rubber. Part two. Tech-
ANDREWS, E.H., BARNARD, D., BRADEN, M. AND GENT, nological aspects. Rubb. J., 146(11), 30.
A.N. (1963) Ozone attack on rubbers. The Chemistry
and Physics of Rubber-like Substances (Bateman, L., LAKE, G.J. AND LINDLEY, P.B. (1965) The mechanical
ed.), 329. London: Maclaren and Sons Ltd. fatigue limit for rubber. J. appl. Polym. Sci., 9(4),
1233.
CUTHBERT, C. (1967) Some properties of large rubber LAKE, G.J. AND LINDLEY, P.B. (1966)Fatigue of rubber.
units. Use of Rubber in Engineering (Alien, P.W.,
Lindley, P. B. and Payne, A.R., ed.), 92. London: Proc. Conf. phys. Basis Yield Fracture Oxford
Maclaren and Sons Ltd. 1966, 176.
GENT, A.N. (1962a) Relaxation processes in vulcanised LAKE, G.J. AND THOMAS, A.G. (1967) The strength of
rubber. I. Relation among stress relaxation, creep, highly elastic materials. Proc. R. Soc. A, 300, 108.
recovery, and hysteresis. /. appl. Polym. Sci., 6(22), RIVLIN, R.S. AND THOMAS, A.G. (1953) Rupture of
433. rubber. I. Characteristic energy for tearing. /.
GENT, A.N. (1962b) Relaxation processes in vulcanised Polym. Sd., 10(3), 291.
rubber. II. Secondary relaxation due to network SOUTHERN, E, (1967) The swelling of rubber by oils and
breakdown. J. appl. Polym. Sci., 6(22), 442. other liquids. Use of Rubber in Engineering (Alien,
GENT, A.N. (I962c) Relaxation processes in vulcanised P.W., Lindley, P.B. and Payne, A.R., ed,}, 49.
rubber. III. Relaxation at large strains and the
London: Maclaren and Sons Ltd.
effect of fillers. Proc. Rubb. Technot. Conf. London THOMAS, A.G. (1955) Rupture of rubber. II. The strain
1962, 348. concentration at an incision. J. Polym. Sci., 18(88),
177.
GENT, A.N. AND MASON, P. (1963) Viscoelastic be-
haviour. The Chemistry and Physics of Rubber-like THOMAS, A.G. (1958) Rupture of rubber. V. Cut growth
Substances (Bateman, L.,ed.), 187. London: Macla- in natural rubber vulcanisates. /. Polym. Sci.,
ren and Sons Ltd. 31(123), 467.
GENT, A.N., LINDLEY, P.B. AND THOMAS, A.G. (1964) THOMAS, A.G. (1960) Rupture of rubber. VI. Further
Cut growth and fatigue of rubbers. I .The relation- experiments on the tear criterion. /. appl. Polym.
ship between cut growth and fatigue. J. appl. Polym. Sci., 3(8), 168.
Sci., 8(1), 455.
GREENSMITH, H.W. (1963) Rupture of rubber. X. The APPENDIX
change in stored energy on making a small cut in a
test piece held in simple extension. J. appl. Polym. Item Vulcanisal e
Sci., 7(3), 993. A B C
GREENSMITH, H.W. AND THOMAS, A.G. (1955) Rupture RSS1 (SMR5) 100 100 100
of rubber. III. Determination of tear properties. ZnO 5 - 5
J. Polym. Sci., 18(88), 189. Stearic acid 2 _ _
GREENSMITH, H.W., MULUNS, L. AND THOMAS, A.G. Lauric acid _ — 1
(1963) Strength of rubbers. The"Chemistry and Sulphur 2.5 - 2.7
Physics of Rubber-like Substances (Bateman. L., CBS 0.6 _ 0.54
ed.), 249. London: Maclaren and Sons Ltd. Nonox ZA 1.5 _ _
Flectol H _ 1.5 1.5
HARWOOD, J.A.C., MULLINS, L. AND PAYNE, A.R. (1967) Lampblack - - 40
Stress softening in rubbers—a review. /, Instn Dicumyl peroxide - 3.5 -
Rubb. fnd., 1(1), 17.
LAKE, G.J. AND LINDLEY, P.B. (1964a) Ozone cracking, Vulcanisation 40' at 10' at 30' at
flex cracking and fatigue of rubber. Part one. Cut conditions 140°C 100°C 140°C
growth mechanisms and how they result in fatigue + 60' at
failure. Rubb. J., 146(10), 24. 150CC

DISCUSSION
Chairman: Dr. A. A. Watson
Dr. P.Thirion asked about the effect of humidity on stress relaxation and the time of recovery of the free strain.
Mr. Thomas replied that die humidity effect was due probablyto the non-rubbers rather than the polyisoprene
itself, because the diffusion of water into rubber was controlled by the non-rubbers present. Dr. C.E. Scott en-
quired if similar humidity effects were shown by synthetic cw-polyispprene whose non-rubber constituents were
greatly reduced. Mr. Thomas replied that humidity affected synthetic cw-polyisoprene as well as natural rubber,
but comparative data—especially on synthetic—were lacking. Mr. S.W. Shi asked if the effect of humidity

199
 Journal of the Rubber Research Institute of Malaya, Volume, 22, Part 2, 1969

could be due to water acting as a physical plastitiser. Mr. Thomas said that plasticisation often reduced stress
relaxation rates; perhaps water softened some components and imparted a stiffness dependent on time, which
would contribute to the creep. Mr. Thomas said that work was in progress on EV vulcanisation systems, which
Mr. J. Payne suggested would show less creep than the conventional use of sulphur; Dr, J.I. Cunneen added
that EV compounds should be better than conventional sulphur vulcanisates in long-term creep where chemical
effects were important, but the behaviour of EV systems and conventional vulcanisates were very similar in
short-term creep,
Mr. R.A. Billett said that lampblack—mentioned in the paper—was virtually unobtainable outside Britain.
Mr. Thomas said lampblack was the material used in the experiments, but any soft non-reinforcing black could
be used. Dr. P.M. Sorgo stated that carbon blacks were very specific in their applications and should not be
used indiscriminately. However, a black with the desired properties—such as large particles and non-rein-
forcing—could always be obtained. Dr. A.A. Watson considered that medium thermal black would provide a
good replacement if lampblack was not available. He asked if it was known what gave rise to the extra creep.
Mr. J. O'Connell commented that lampblack increased stiffness with minimum loading. Mr. Thomas informed
Mr. I.K. Yeo that creep behaviour in rubber under compression was under investigation.

200