You are on page 1of 7

Dynamic Properties

MEASUREMENT OF DYNAMIC
PROPERTIES OF RUBBER
ROBERT S. MARVIN
National Bureau of Standards, Washington 25, D. C.

T HE dynamic mechani-
cal properties of ma-
terials are generally under-
T h e relation of the dynamic modulus to other rheologi-
cal measurements on rubberlike materials is discussed.
The experimental methods used in determining dynamic
limiting slope greatly exag-
gerated. The recovery after
removal of the load is also
stood t o be those that govern properties of rubber are classified into three groups de- shown. As indicated by
the response of matter to pending on the relative values of the sample dimensions broken lines, this creep
stresses alternating with and the wave length of sound in the sample. Examples of curve may be broken up into
time. They are usually ex- each group, with a short discussion of advantages and three additive parts: an in-
pressed in a form which as- disadvantages of each, are given. An appendix includes stantaneouselastic response,
mmes that both stress and derivations, based on an infinite, linear differential equa- a delayed elastic response,
strain have a sinusoidal time tion, for forced and free vibration, and relates some of the and a viscous flow. Of these
variation but can be defined more commonly used methods of expressing the results only the, first two are re-
under other conditions, as of dynamic measurements. coverable. This type of be-
will be discussed below. havior has been predicted
We shall assume through- for linear amorphous poly-
out that at a fixed frequency the strain is directly proportional to mers generally, and the fact that many materials actually do follow
the strese amplitude; this is equivalent to assuming that stress such curves, with the recovery curve representing the mirror
and strain can be related by a linear differential equation. ill- image of the instantaneous elastic and delayed elastic portions
though the total stress-strain curve for rubber is decidedly non- of the deformation curve, has been confirmed experimentally
linear, the approximation introduced here is reasonably accurate (9, 24, 27, 28). Parenthetically, it is often very difficult experi-
for the small strains that are encountered in dynamic testing mentally to differentiate between a true viscous flow and a de-
(usually under 1% and almost never greater than 10%). layed elastic response with a very long relaxation time (9, SO).
Since rubber is a material of the type generally termed visco- To the extent that the creep function, E ( ~ ) / (where
u u repre-
elastic, the dynamic modulus must express both the effective sents stress and E strain), is independent of the magnitude of the
“spring constant,” and the damping properties or energy loss stress, the behavior of the material is linear. This is equivalent
during cyclic deformation. These properties are of great prac- to saying that the superposition principle applies-that is, the
tical importance in the design of those rubber articles whose use +
strain due to the combined stresses u1 u2 is equal to the sum of
involves alternating rather than static deformations, and are also the strain due to u1 alone plus that due to U Z alone. It is this same
of considerable theoretical interest in that they must provide ulti- linearity of behavior, of course, that permits breaking of the
mately a measure of the forces existing between the tangled creep function into three components as in Figure 1.
chains that make up the rubber. Another specification of the properties of such a material
The detailed theoretical analysis which would predict the employs a general linear differential equation, known as the
rheological behavior of a rubber sample in terms of these forces operator equation ( 3 ) :
is Bot yet available, hence a phenomenological treatment in which
the general strese-strain relations are given in terms of certain
macroscopic material constants will be used. n = O m = O

TYPES OF RHEOLOGICAL SPECIFICATIONS where


There are several ways in which such a general relation (a D“ = an/a tn
rheological equation of state) may be expressed. The con-
stants occurring in a general differential equation relating stress The coefficients CY,, and pmspecify the rheological properties of a
and strain may be specified, or the response of the material to material, just as does the creep function, and may be derived
some specified stress or strain may be given. I n principle, all from it, at least in principle.
such specifications are equivalent, and methods for converting Equation 1 (and the creep function also) is perfectly general
from one type to any of the others have been developed (9,3, 8, and if solved subject to the appropriate boundary conditions
18, 27, 48, 68). These conversions involve mathematical dif- would give the response of the material to any arbitrary stress
ficulties and necessitate a knowledge of the functions over a com- or strain. Such solutions, however, may be rather difficult t o
plete range of time or frequency. Since such information is obtain, and in any event it is often convenient to have a “model”
generally not available, in practice various approximation meth- of this equation from Fhich a good deal of qualitative informa-
ods are required (2,3, IS, 49, 6463,67). tion may be deduced without attempting a formal solution.
Perhaps the simplest of these rheological relations to discuss One type of model which can represent Equation 1 is shown in
is the creep function, which gives the response of the material to a Figure 2. Such models are made up of combinations of “springs,”
constant stress a9 B function of time. Figure 1 shows such a across which the force is proportional to the strain, and “dash-
curve for a material like an unvulcanized rubber exhibiting ir- pots,” across which the force is proportional to the rate of strain.
reversible flow, with the initial instantaneous response and the When arranged in the form of Figure 2, a spring-dashpot pair is
696 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 44, No. 4
-ELASTOMERS-Dynamic Properties-
known as a Voigt element, and the whole array as a Voigt Model real and imaginary parts of the ratio of stress to strain when the
(66). .time dependence of each is expressed as exp (id), where i =
It should perhaps be stressed that such models need not imply 2/--i. They may be combined into a complex modulus de-
any physical assumptions regarding the actual molecular proc- fined by G* = G' + iG",which is the ratio of complex stress to
esses occurring in the material. They can be regarded as merely complex strain (the actual stress and strain are the real parts of
an alternate rheological specification of material properties. the complex expressions). The ratio G"/G' is the tangent of the
Here, the direct experimental functions are considered as the angle, 8, between stress and strain, called the loss tangent.
fundamental specifications and the others as derived. Alternately, the reciprocal of the loss tangent, designated Q, is
used by some authors. The whole system of complex notation
is exactly like that used in alternating current theory.
A set of definitions which hold for any type of periodic stress
and strain, and reduce to those given here for sinusoids, can be
made in terms of energy concepts. If D is defined as the maxi-
mum energy stored in the sample per unit volume during a cycle,
w as the energy dissipated per cycle per unit volume, and 60 as
the amplitude of the strain, the components of the complex
modulus may be defined as:
G' = 2v/eoa (2)
G' = w/regZ (3)
I
I ' - and the complex modulus, G*, as before. Although the quantities
I - \
w and 0 would be rather difficult t o measure directly, they are
t0 tl concepts which can at least be visualized for a generalized periodic
TIME -t
deformation. Several other methods' of expressing the results of
Figure 1. Creep Function and Its Components
dynamic experiments have been used by various workers, and a
Instantaneous elastic response - ---
Stress imposed at h: released at ti
Delayed elastic response * - - -
summary of some of these with equations relating them for the
case of sinusoidal stress and strain is given in the appendix.
viscous flow =
Total response -
- - --
-
0-
b
The creep function, already considered, expresses directly cer-
tain experimental results. If it-could be determined over the
complete range of times from zero t o infinity, it would provide a
complete rheological epecification of a material. However, ex-
perimental limitations deny a direct measurement of this func-
tion a t very short times, say less than 1 second. Speaking in
terms of the model of Figure 2, the elements whpse relaxation
times, 7, are less than about 1 second will appear during a creep
experiment as pure springs, and their delayed elastic character
can only be evidenced by experiments conducted on a time scale
of the order of magnitude of their relaxation times.
Dynamic experiments provide, of course, the obvious means of
obtaining such information, since the time scale of a steady-state
sinusoidal vibration is equal to the reciprocal of the frequency.
Hence, the definition of the dynamic modulus is discussed and
then the various experimental methods which have been used
in its measurement.

DYNAMIC MODULUS FUNCTION 0-


Figure 2. Voigt Model
If a stress having a sinusoidal time variation is applied to a
linear viscoelantic material, the steady-state strain will also have ti
sinusoidal time dependence but will in general be out of phase with The dynamic modulus function is another specification of the
the stress as shown in Figure 3A or, since both quantities are rheological properties of a material, and is of course equivalent
sinusoids with the same period, by the rotating vector scheme of in principle to all the others. It can be determined directly in the
Figure 3B. range of short times but ia rather difficult to evaluate experi-
o is the radian frequency, equal to 2r times the frequency,
mentally a t long times or very low frequencies.
Y , and t is the time in seconds.
It happens that the transformation from a Voigt model of the
Figure 3 shows that two quantities are required to specify the type of Figure 2 to a dynamic modulus function is fairly simple,
response to a sinusoidal stress. These may be taken as the ratio and it is interesting to make such a transformation for a model
of the amplitudes of stress and strain, and the phase angle be- which would yield a creep function qualitatively like that of
tween them, or an two equivalent quantities. The two used here Figure 1. The results, for an arbitrary choice of parameters, are
PS basic quantities for this report can be defined as follows:
shown in Figure 4, where the real and imaginary parts of the
Component of stress in phase with strain modulus are shown logarithmically and the loss tangent semilog-
G' = arithmically.
Strain
Component of stress 90' out of phase with strain At very low frequencies both G' and G" go t o zero, since the
G,, model selected exhibits irreversible flow; they approach zero
Strain
in such a way that their ratio, tan 6, goes to infinity. As the
These quantities are chosen because they are, respectively, the frequency is increased, the response of the purely viscous ele-
April 1952 INDUSTRIAL AND ENGINEERING CHEMISTRY 697
-ELASTOMERS-Dynamic Propertie-
ment decreases (since velocity is proportional to frequency) and note that increasing frequency at constant temperature corre
that of the delayed elastic element predominates. I n this region sponds to decreasing temperature at constant frequency.
G’ remains nearly constant, while G” and tan 6 increase with
increasing frequency until the response of the pure spring is TYPES OF MEASUREMENTS
appreciable compared to that of the damped spring. As this
frequency is approached tan 6 goes through a rather sharp No attempt will be made here to give a complete account of
maximum corresponding to a maximum energy loss, G” also all the methods which have been developed for measuring dy-
rises to a maximum, and G’ approaches asymptotically the value namic properties of rubber. Rather a classification of the various
of the undamped spring, GI. At very high frequencies, the veloc- types of methods will be set up and certain specific examples will
ity of motion is so large that the dashpots represent essentially be noted for each type of measurement.
zero compliances, and the whole motion is associated with the Some recent papers review this subject from various points of
undamped spring. view: A discussion of many of these methods, particularly those
used as test methods by the rubber industry, was given by Dillon
and Gehman (10). Ferry, Sawyer, and Ashworth ( 1 4 )mention a
few methods and classify and discuss the various possibilities,
and Nolle (58) in a report of his outstanding developmental work
in this field of instrumentation has given discussions of five dif-
ferent methods and has included derivations of the operating
equations for each.
Leaderman ($9) has presented a discussion of the response of
viscoelastic systems to various types of strain and proposed a
systematic nomenclature.
It must be evident that in order to measure the whole range of
frequency of interest in dynamic measurements, extending
roughly from 1 to lo7 cycles per second, several widely different
types of measurement will be necessary. One way to clamify
such methods is based on the relative values of the sample dimen-
sions and the wave length of sound in the material a t the fre-
Figure 3. Sinusoidal Stress and.Strain quency under consideration (14). If the dimension of the sample
in the direction of the imposed strain is much smaller than the
wave length of sound in the sample a t the test frequency the
The model used here is idealized chiefly in that its dynamic inertia of the sample may be neglected, and the analysis proceeds
modulus function shows much more abrupt changes with fre- as though the sample were a single damped spring with fre-
quency than do actual rubberlike materials; this explains why so quency-dependent spring and damping constants. If this di-
many of the earlier workers in this field found, for a narrow fre- mension is much larger than the wave length, the strain will be
quency range, dynamic moduli independent of frequency. damped out, and it becomes possible to measure the attenuation
Two groups of workers have fairly recently investigated a wide of waves in the material. If the two quantities are of the same
enough frequency range to cover the sharp rise in G’ as it ap- order of magnitude standing waves can be set up.
proaches its high-frequency limiting value: Guth and his co- Methods corresponding to all three of these cases are being used
workers ($3) for Butyl and natural rubber and Nolle (39) for experimentally for dynamic measurements on rubber. By far
natural, Buna-N, Butyl, neoprene, and GR-S. the most common, those applicable at low frequencies (up to a
Results now available from a large scale cooperative program few kilocycles per second) are the first type where wave motion
initiated by the National Bureau of Standards in which over 20 in the sample can be neglected. Usually the rubber sample is
laboratories throughout the world have made measurements of used to provide the restoring force in a mass-spring system
mamples from a single lot of polyisobutylene covering the f r e which is set into either forced or free vibration. In the case of
quency range of 0.01 to 7 x 105 cycles per second, with stress forced vibration the “spring constant” or G’ is equal to a geomet-
relaxation measurements extending the equivalent frequency ric constant depending on the dimensions of the sample, times
range to about 10-8 cycle per second, demonstrate essentially the mass, times the square of the radian frequency of resonance.
all of the behavior shown in Figure 4. As expected, the func- Strictly speaking the frequency a t which the velocity of motion
tions change much more gradually with frequency than indi- of the mass (for a constant force) is a maximum should be used,
cated in Figure 4, the final rise in the G‘ curve extending over a t but the use of an amplitude maximum rather than a velocity
least six decades. maximum leads to only a small error (which can be corrected for
if desired) if the damping is not extreme. The methods used by
EffECT Of TEMPERATURE Kosten and Zwikker ($6),Gehman (16), n’aunton and Waring
(37), Moyal and Fletcher (%), Oberto and Palandri (41), Geh-
It has been recognized for a long time that decreasing the man, Woodford, and Stambaugh (17, 51), and Dillon, Pretty-
temperature has an effect on the dynamic modulus similar to man, and Hall (11)are examples of this type.
that due to increasing the frequency. This has been explained In general such devices, or a t least the more recent ones,
qualitatively by the common assumption that the effects of in- have employed electromagnetic drives which are convenient to
dividual viscous mechanisms are proportional to exp (AI?“),A a use and permit the design of a relatively compact apparatus.
constant. It has also been generally assumed, based on the A slightly different type of resonance vibrator, which still
kinetic theory of rubberlike elasticity, that individual spring must be considered under this classification, however, has been
mechanisms have moduli directly proportional t o the absolute developed recently by Rorden and Grieco (44) and described by
temperature. On these conceptions, Ferry (12)has recently based Hopkins ( l d ) . They measure the change in resonance frequency
a reduction scheme by which dynamic measurements made at and breadth of the resonance curve of an electrically driven
various temperatures and frequencies may be compared, and tuning fork caused by the introduction of rubber samples strained
has shown experimentally that such measurements do form single in shear by the motion of the fork, covering a frequency range
curves when so reduced. of from 100 to 14,000 cycles per second.
For the purposes of this discussion, it will be sufficient to Nonresonance forced vibrators have been applied to rubberlike
,698 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 44, No. 4
-ELASTOMERS-Dynamic Propertie-
materials somewhat less frequently. I n such methods the real in a sample is measured are the longitudinal wave technique
and imaginary parts of the dynamic modulus can be measured (4, 20, 38),which has been applied more extensively to textile
directly in terms of the definitions given here by a measurement fibers, and the compressional wave technique (6'8,40, 46). An-
of the force and velocity (or Bmplitude) and the phase angle be- other method which has been used at high frequencies (35) in-
tween them. Electromagnetic drives are usually employed, and volves measuring the loading introduced on an electromechanical
electromagnetic pickups can conveniently be used to measure the transducer by the presence of a sample. This type of measure-
velocity (SI),although condenser pickups or strain gages could be ment has also been used by Smith, Ferry, and Schremp (60) at
employed. low frequencies for measurements on polymer solutions.
The frequencies covered by nonresonance methods may be
varied continuously, and very small strains (10-6 or less) can be COMPARISON OF nqETHODS
employed, thus practically ensuring that the material will re- The selection of the best of these several methods for any
spond in a linear manner. They are thus considerably more flex- particular application or the comparison of the results obtained
ible than rfisonance vibrators. by these various methods is somewhat complicated. The first
Blizard (6) has described measurements with a device of this factor is a consideration of the frequency range to be covered,
type, using a special generator, over the frequency range of 0.0125 as indicated in the previous discussion. Changes in the design
t o 750 cycles per second. A nonresonance method employing a of the apparatus or size of the sample can allow a certain exten-
mechanical drive and optical recording, but in which only the real sion of the range of applicability, but even so there is a limita-
part of the modulus was measured, was used by Aleksandrov and tion of the order of three decades of frequency on the appli-
Lazurkin (1). cability of any one type of measurement.
A somewhat different nonresonance method was developed by
Roelig (43)and has been used fairly extensively. I n this method
the force applied to the mass in the mass-spring system is not 1 ' " " ' ' 1 " 1 " ' ' " '
measured, but rather the force transmitted through the sample.
Roelig uses an optical method to trace out a stress-strain ellipse
which can be analyzed to yield G' and G". The advantage of this
method over those above is that the inertia of the apparatus does
not enter into the equations for calculating the modulus, at least
u p to frequencies of the order of 100 cycles per second. The
machine as designed by Roelig uses rotating eccentric weights to
apply the driving force. This type of driving system, plus the fact
that the machine is designed to accept large test pieces, leads to
a rather large piece of apparatus.
Most of the very low frequency measurements (10 cycles per
second and below) have been made utilizing free vibration de-
vices (7, 19, 26, 55, 56). The directly measured quantities are
the period of oscillation and the logarithmic decrement. The
conversion from these quantities to the dynamic modulus as
used here is given in the appendix. Since the strain ik not
sinusoidal, but a damped sinusoid, there is, strictly speaking,
a n approximation involved in calculating an equivalent dynamic
modulus from these results, but the few measurements available
t o date by which various methods can be compared indicate
that any error involved is actually quite small.
Probably the most familiar example of this class of measure-
ments is the Yerzley oscillograph in which a rubber sample
strained in compression (modifications have been used employ-
ing shear) supports an oscillating beam carrying adjustable
weights and a pen which makes a record of the decaying oscil-
lations oa a rotating drum.
Many other devices employing free vibrations in compression,
shear, and torsion have been developed, and a great deal of in- Figure 4. Dynamic Response of Four-Parameter Model
formation has been obtained with them. They share with the
resonance vibrators the disadvantage of requiring that a change
in either the sample size or the inertia member must be made to Most of the measurements made a t low frequencies use either
change the frequency of measurement. free vibration or forced resonance methods. Certain approxi-
The most common example of a standing wave method for use mations are inherent in the analysis of such results as usually
with solid rubbers is the vibrating reed technique. In this method carried out. In the free vibration method the amplitude is con-
a thin reed of material is vibrated either as a cantilever beam (38) stantly decreasing during the course of a measurement, so the
or longitudinally (47). The amplitude of vibration of the free dependence of modulus on amplitude cannot be checked too
end of the beam may be observed directly with the aid of a low easily. In the resonance method, if resonance is detected by
power microscope, or the modulated intensity of a light beam changing the frequency as is quite common, the assumption must
partially interrupted by the vibrating reed may be measured with be made that the components of the modulus are frequency-
a photocell. The frequency and band width of the amplitude independent over a limited frequency range.
resonance may be measured, the amplitude of free vibrations The experimental comparisons so far available between reso-
may be followed as they decay, or the amplitude-force ellipse nance and nonresonance vibrations seem to indicate that these
may be displayed on an oscilloscope screen and a quantity pro- approximations are valid within the precision of most of the
portional to G" calculated from the area of this ellipse divided methods now in use. This leaves, then, as the primary objections
by the square of the strain amplitude. to the forced resonance method, the fact that it is relatively in-
Methods in which the decay with distance of the wave motion convenient to make measurements a t a large number of fre-
April 1952 INDUSTRIAL AND ENGINEERING CHEMISTRY 699
-ELASTOMERS-Dynamic Properties-
quencies within the range of the apparatus and also that this tion, since the modulus which governs such waves is given by
total range itself is quite limited; in practice most such devices +
K* 4/3 G*, where K* is a complex bulk modulus. The cal-
are operated in the range 30 to 150 cycles per second, and many culation of the components of the shear modulus from such
are used a t 60 cycles per second only. measurements, even assuming that the imaginary part of K* is
The free damped vibration measurements are quite conven- zero, requires some independent determination of K', the real
ient to set up and operate, but the analysis of their results is not part of K*.
too precise with the instrumentation ordinarily used. They are Nolle ( 4 0 ) and Guth (25) have made the assumption that the
used most widely (Yerzley oscillograph) as a fairly routine method bulk modulus of their polymers could be taken as equal to that
for comparing various stocks, but the instrumentation can be of water. While they obtained values which fitted a reasonable
refined, and this method & the most convenient for measure- extrapolation of lower frequency measurements of the shear
ments a t frequencies below the range of ordinary electronic modulus by the use of this estimate in connection with data taken
equipment. a t about room temperature, there seems to be no real basis for
Forced, nonresonance methods have the advantage, aside expecting this assumption to hold true in general.
from yielding a direct measurement of the complex modulus, of At lower frequencies, up to a t least several kilocycles, there is
being extremely flexible in the frequency and amplitude of strain evidence based on comparison of measured shear and Young's
employed. With electromagnetic driving and pickup units the moduli (Young's moduli measured by extension methods) that
sensitivity can be made very high. Poisson's ratio is very close to one half as it is in the static case
Such instruments do, of course, have certain inherent draw- (31).
backs. Although their total frequency range can be extended to
perhaps 1500 or 2000 cycles per second, they require extraordi- PRECISION OF MEASUREMENTS
narily careful design and construction in order to avoid extrane- The precision and accuracy of all these methods is a subject
ous resonances which can lead to false results a t frequencies as on which the information now available is not too satisfactory.
low as 200 cycles per second unless care is taken. This work, The variability between duplicate results on the same sample
and the assembly of the associated electronic equipment, require seems to be of the order of *5oJ, for the best resonance and
a good deal of time, and to ensure reliable results experienced nonresonance forced vibration measurements and of the order
operators are required. I n view of these factors and the rather of f 1 0 to 15% for some of the other methods which involve
high costs of such equipment, they have been used so far only as somewhat less precise observations. The variability between re-
research instruments. sults obtained on duplicate samples ranges from about &5% to
The vibrating reed technique, in its simpler form where the *20%.
amplitude is measured with the aid of a low powered microscope, This is not a particularly satisfactory state of affairs, since the
requires relatively simple and inexpensive equipment and appears differences between many types of stocks can often fall within
to be a rather convenient experimental technique. With a given this range. A number of experimenters are now working on the
sample, measurements can be obtained a t only one frequency for problem of improving such measurements, and the outlook for
each temperature, and the determination of absolute modulus attaining a precision of 1 or 2% seems, in some cases, rather good.
values is difficult because of the uncertainties in measuring the One factor which is extremely important and which has received
dimensions of the reed accurately. far too little attention in most of the work to date is that of
The measurements employed at higher frequencies all require close temperature control. Based on the polyisobutylene results
the use of somewhat involved electronic circuits and require from the National Bureau of Standards' cooperative program,
personnel with some electronic experience for their successful it appears that for this material a t 25' C. and 60 cycles per sec-
operation. Such equipment has been principally in the class of ond, the temperature coefficient of G' is about 4% per ' C.,
specially built research equipment up to the present time. whereas a t 25' C. and 4000 cycles the coefficient is about five
times. It should be noted, that for many rubbers (natural for
TYPES OF STRAIN AND AMPLITUDE EPPECTS instance) the variation of G' at room temperature and 60 cycles is
much less, since in this case the measurements are on a relatively
Most of the low frequency test methods can be designed to Rat part of the curve.
employ any of the ordinary types of strain: shear (including As a general rule, then, it seems safe to say that in certain
torsion), compression' or, for a vulcanized material, extension. frequency and temperature regions, control to 0.1 or 0.2" C.
Of these, shear or extension appear8 to be the most satisfactory should be sought. Also, temperatures within the samples should
if an absolute modulus value is desired, as the effects of sample be checked by means of thermocouples, although this should
shape and size seem to have a great effect on the results of com- probably be done in a dummy experiment the results of which
pression measurements. The use of a shape factor defined by the are not used, since the presence of a thermocouple would very
ratio area of one bonded surface/area of free surface ( 4 5 ) has likely affect the stress concentration in some samples.
been used with considerable success in static measurements but
has not been tested as completely for dynamic measurements.
NONSINUSOIDAL STRESS
Dynamic measurements in compression seem to yield values
that depend not only on the shape factor but on the amplitude So far the discussion has been limited to methods in which
of strain and the static strain superposed during the test. It has steady-state sinusoidal stresses were employed and from which
been assumed throughout this discussion that dynamic moduli both the real and imaginary parts of a modulus could be calcu-
are independent of amplitude, but this does not hold true for lated. There have been used, and are used today, a number of
strains greater than 1%, and there seems to be an effect present methods which do not fulfill one or both of these conditions,
even below 1% strain. For both shear and compression the but which should still be classified as dynamic methods.
dependence is such that G' and G" decrease with increasing am- Some of the earliest measurements of "dynamic" properties
plitude, and i t is apparently not a temperature effect ( 1 6 , 32, 4 1 ) . were made with various types of flexometers (IO). These devices,
The effect is very pronounced with tread stocks but can be ob- in general, apply alternating stress or strain to a rubber sample
served with gum stocks and even with pure polymer. No really but measure only the temperature rise as a result of the vibra-
satisfactory explanation seems to exist as yet, and a good deal tion. Although such a measurement can in principle be made the
more careful experimental work on this effect is needed. basis for a determination of G", since temperature rise is an in-
The high frequency measurements that involve the propaga- dication of energy loss, the complications involved io calculating
tion of compressional waves present an additional complica- the actual energy loss are such that in practice this is not done,
700 I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY VoI. 44, No. 4
-ELASTOMERS-Dynamic Propertie-
and they are used only to compare various stocks. The various For the forced vibration case, express the time dependence as
devices are made to operate a t either constant amplitude or con- exp ( i d )and there results
stant force, and, as Dillon and Gehman and many others have P(iw) u = Q(iw)e
pointed out, care must be used in the interpretation of such re-
sults since actual apfilications of rubber articles may involve or
neither constant amplitude or force but some rather complicated U/€ = &(iw)/P(iw) (5)
intermediate case. Such devices seem definitely unsuited to any
which is, by definition, the complex modulus. The ratio Q / P
fundamental investigation. in Equation 5 contains both real and imaginary parts which are
Mooney and Black (34) have developed an instrument in functions of the frequency and of the material constants, an,
which the energy loss per cycle is measured for samples under- &,,. Therefore,
going approximately sinusoidal, forced vibration.
G’ = Re [ Q ( i w ) / P ( i w ) ]
The various pendulum and rebound methods, in which the
ratio of recovered to initial energy is measured by observing (es- G” = Im [ & ( i w ) / P ( i w ) ] (6)
sentially) the height of fall and recovery of a weight impinging For sinusoidal stress and strain only, the use of a simple dif-
on a sample, have been used also for many years with appar- ferential e uation containing frequency-dependent “constants”
ently rather satisfactory results in the comparison of different in place of%quation 4 can be justified. If 7’ = G ” / w , Equation
stocks. Such methods yield results from which the loss tangent 6 shows that, for any given frequency, Equation 4 may be
can be calculated directly, but the frequency with which each written
measurement should be associated is somewhat difficult to deter-
mine accurately, and this frequency will of course depend on the
u = G‘ e + 7’ De (7)
modulus of the material under test. It is interesting to note, which is the starting point for the analysis of dynamic experi-
ments generally used.
however, that the results of Powell and Aston using the Dunlop For free vibration experiments, the stress is due to an inertial
Pendulum and the Tripsometer on the “standard” polyisobutyl- force and may be expressed as - ( M / a ) D % ,and the strain must
ene already referred to, agree surprisingly well with other more be written €0 exp ( i w - X ) t , the factor, exp ( - A t ) , expressing the
direct measurements. This result is termed surprising not only decay of strain with time. Substituting these quantities into
Equation 4 there results
because of the difficulties in deciding on the frequency to associ-
ate with the measurements but also because such measurements - ( M / u ) (iw - X)2P ( i w - X)e = Q(iw - X)e (8)
approximate only a half cycle of sinusoidal vibration. -4t first
Our aim here is to find some way of calculating G‘ and G”
glance one might be tempted t o assume that a steady,state as defined in Equation 6 from the experimental results of a free
condition would not be set up in such an experiment, and a total vibration experiment in which W , or the period, and X, the at-
lack of agreement with steady-state measurements would hardly tenuation of the vibration with time, are measured, To do this
have been surprising. a first ap roximation may be made by assuming that X is so
much smaser than w that Equation 8 may be written
Finally, some rather recent methods which involved steady-
state, nonsinusoidal vibrations should be mentioned. Breazeale + 2iwX) = & ( i w ) / P ( i w )
( M / a ) (w2 (9)
The right-hand side of Equation 9 is simply G’ + iG”from Equa-
(6) has developed an apparatus for measurements on textile
fibers (which can also be used with some success on rubbers) tion 6. Thus
employing triangular or saw-tooth longitudinal vibrations.
Tolstoy and Feofilov (64) have discussed the development of a G’ = w2M/a
method employing square-wave pulses and observing the shape G” = 2wXM/a (10)
of the resulting strain. While such methods have not been fully
evaluated (so far as this writer is aware) they offer interesting In Equations $, 9, and 10 a is a constant depending on the sam-
ple dimensions such that: force/displacement = a stress/strain.
possibilities for a t least qualitative evaluations of dynamic prop- A number of different methods of expressing experimental re-
erties. Since a square wave can be thought of, thrpugh a Four- sults have been used by workers in this field, all of which can be
ier analysis, as representing the simultaneous application of related (exactly or t o a close approximation) if sinusoidal stress
sinusoidal waves consisting of the fundamental and all its har- and strain are assumed,
For forced vibrations, the energy loss per cycle, per unit
monics, there seems a t least a speculative possibility here of ob- volume, is
taining some sort of modulus function which would be an inte-
grated result of the dynamic moduli over the whole frequency w = S c d r = r e o 2 G” (11)
range on the basis of a single measurement. No doubt such re- It should be noted that in evaluating this integral the actual
sults would be difficult t o compare with moduli based on sinu- stress and strain-that is, the real parts of the complex expres-
soidal stress measurements, but the possibilities are intriguing. sions-must be used.
The specific damping capacity is defined as the ratio of the
energy loss, w, to the maximum energy stored In the sample dur-
APPENDIX ing a cycle, Y
$ = w / v = 2 a tan 6 c (12)
RELATION OF DYNAMIC MODULUS TO OPERATOR
EQUATION AND TO EXPERIMENTAL RESULTS Roelig defines a specific damping capacity in a slightly dif-
ferent manner, as the area of his stress-strain ellipse (w) divided
There follows a derivation relating the fundamental ( f r e by ‘/e the product of maximum stress and maximum strain
quency-independent) parameters in the linear differential equa- (which he also measures directly from his ellipse). This gives a
tion for a viscoelastic material to the quantities measured in specific d p p i n g capacity equal to:
free and forced vibration (where wave motion in the sample can
be neglected) (49, and a summary of some of the other methods
that have been used to express dynamic results,
In his vibrating reed measurements, Sack uses the quantity
The operator equation (1) may be written as [SI W / E O ~asa measure of energy loss
P ( D ) u = Q ( D )t ($1 W/E$ = wG” (14)

where P = 2
n - 0
anDn, Q =
m

m = O
&Dm
The more common analysis of such measurements, wherein the
frequency of amplitude resonance, YO, and the half-power width
of the resonance curve, AV (the difference between frequencies a t
April 1952 INDUSTRIAL A N D ENGINEERING CHEMISTRY 701
ELASTOMERS-Dynamic Properties-
which the amplitude is 1/.\/2 of the maximum), are measured, (21) Halt, W. L., and McPherson, A. T., J . Research Natl. Bur.
yields Standards, 17, 657-78 (1936) [Rubber Chem. and Technol., 10,
412-31 (1937)l.
A v / v ~= t a n 6 (15) (22) Hopkins, I., Trans. Am. SOC.Mech. Ennrs., 73, 195-204 (1951).
The results of free damped vibrations are often expressed (23) Ivey, D. G., Mrcmca, B. A., and Guth, E., J . Applied Phys., 20,
in terms of the period of vibration, T = 2 K / W , and the logarithmic 486-92 (1949) [Rubber Chem. and Technol., 23,172-84 (1950)l.
decrement, which is defined as the logarithm of the ratio of the (24) Kohlrausch, F., A n n . physik. Chem., 158,337-75 (1876).
amplitudes of two successive vibrations. (25) Kosten, C. W., and Zwikker, C., Physica, 4, 221-34 (1937); see
also Kosten, C. W., “Proceedings Rubber Technology Confer-
ence,” pp. 987-1002, Cambridge, W.Heffer and Sons, Ltd.,
1938 [Rubber Chem. and Technol., 12,381-93 (1939)l.
(26) Kuhn, W., and Kunzle, O., Helu. Chim. Acta, 30, 839-58 (1947).
Dynamic resilience is defined as the ratio of energy stored (27) Leaderman, H., ”Elastic and Creep Properties of Filamentous
in two successive cycles of free vibrations. Since this energy is Materials and Other High Polymers,” Washington, D. C.,
proportional t o the displacement squared, the dynamic resilience The Textile Foundation, 1943.
1s (28) Leaderman, H., IND. ENC.CHEM.,35, 374-8 (1943).
(29) Leaderman, H., J . Colloid Sci., 4, 193-210 (1949).
(30) Lethersich, W., Brit. J . Applied Phys., 1, 294-301 (1950).
(31) Marvin, R. S.,Fitsgerald, E. R., and Ferry, J. D., J . Applied
PhyS., 21,197-203 (1950).
For pendulum or rebound tests only half a cycle of vibration (32) Marvin, R. S., Weiss, F., and Mandel, J., National Bureau of
is involved, so the resilience as measured in such tests is the Standards, unpublished experiments.
square root of t h a t defined by Equation 17 f33) Mason, W. P., Baker, TV. O., McSkimin, H. J., and Heiss,
J. H., Phys. Rev., 75, 936-46 (1949); McSkimin, H. J., J.
R, = exp ( - R tan 6) (18) Acoustical SOC.Am., 23, 429-34 (1951).
R, is also generally used to express the results obtained with the (34) Mooney, M., and Black, S.A,, Can. J . Research, 28F, 83-100
(1950) [Rubber Chem. and Technol., 23, 744-59 (1950)l.
Yerzley oscillograph. (35) Mooney, M.. and Gerke, R. H., R&ber Chem. and Technol., 14.
3 5 4 4 (1941); 15, 158 (1942).
LITERATURE CITED (36) Moyal, J. E., and Fletcher, W. P., J. Sci. Instruments, 22, 167-70
(1945) [Rubber Chem. and Technol., 19,163-9 (1946)l.
(1) Aleksandrov, A. P., and Lazurkin, Yu. S., J . Tech. Phys. (37) Naunton, W. J. S.,and Waring, J. R. S., “Proceedings Rubber
( U S S R ) ,9,1249-60,1261-66, 1267-79 (1939). Technology Conference,” pp. 805-20, Cambridge, W. Heffer
(2) Alfrey, Jr., T., “Mechanical Behavior of High Polymers,” and Sons, Ltd., 1938 [Rubber Chem. and Technol., 12, 332-43
New York, Interscience Publishing Co., 1948. (1939)1.
(3) Alfrey, Jr., T., and Doty, Paul M., J . Applied Phys., 16, 70S13 (38) Nolle, A. W., J. Applied Phys., 19, 763-74 (1948).
(1945). (39) Nolle, A. W., J . Polymer Sci., 5, 1-54 (1950).
(4) Ballou, J. W., and Smith, J. C., Ibid., 20, 493-502 (1949). (40) Nolle, A. W., and Mowry, S. C., J . Acousfical SOC.Ant., 20.
(5) Blizard, R. B., Ibid., 22, 730-5 (1951); Ph.D. thesis, Massa- 432-9 (1948).
chusetts Institute of Technology, 1949. (41) Oberto, S., and Palandri, G., Rubber A g e ( N . Y . ) , 63, 725-34
(6) Breazeale, F., and Whisnant, J., J . Applied Phys., 20, 621-6 (1948) [Rubber Chem. and Technol., 22, 1009-27 (1949)].
(1949). (42) Rivlin, R. S., Trans. Inst. Rubber Ind., 26, 78-85 (1950).
(7)Cassie, A. B. D., Jones, M., and Naunton, W. J. S.,Trans. (43) Roelig, H., “Proceedings Rubber Technology Conference,”
Inst. Rubber Ind., 12,49-84 (1936). pp. 821-9, Cambridge, W. Heffer and Sons, Ltd., 1938
( 8 ) Cini, M., J . Applied Phys., 21, 8-10 (1950). [Rubber Chem. and Technol., 12, 394-400 (193911.
(9) Dahlquist, C. A., Hendricks, J. O., and Taylor, N. W., IND. (44) Rorden, H. C., and Grieco, A,, J . Applied Phys., 22, 842-5
ENQ.CHEM.,43, 1404-10 (1951). (1951).
(10) Dillon, J. H., and Gehman, J. D., India Rubber World, 115, (45) “Rubber in Engineering,” pp. 266-7, London, His Majesty’s
61-8, 76, 217-22 (1946) [Rubber Chem. and Technol., 20, Stationery Office, 1946.
827-58 (1947)l. (46) Sack, H. S., and Aldrich, R. W., Phys. Rea., 75,1285 (1949).
(11) Dillon, J. H., Prettyman, I. B., and Hall, G. L., J . Applied (47) Sack, H. S., Motz, J., Raub, H. L., and Work, R. N., J . Applied
Phys., 15,309-23 (1944) [Rubber Chem. and Technol., 17,597- Phus.. 18. 450-6 (1947).
616 (1944)l. (48) Simha, R., j.Applied Phys., 13,201-8 (1942).
(12) Ferry, J. D., J . Am. C h a . Soc., 72,3746-52 (1950). (49) Sips, R . , J . Polymer Sci., 5, 69-89 (1960).
(13) Ferry, J. D., Fitzgerald, E. R., Johnson, M. F., and Grandine, (50) Smith, T. L., Ferry, J. D., and Schremp, F. W., J . Applied Phys.,
Jr., L. D., J . Applied Phys., 22,717-22 (1951). 20, 144-53 (1949).
(14) Ferry, J. D., Sawyer, W. M., and Ashworth, J. N., J . Polymer (51) Stambaugh, R. B., IND.ENQ.CHEM.,34,1358-65 (1942) [Rubber
Sci., 2, 593-611 (1947). Chem. and Technol., 16,400-18 (1943)l.
(15) Fletcher, W. P., and Gent, A. N., Trans. Inst. Rubber Ind., (52) ter Haar, D., Physica, 16, 719-52, 83S50 (1950); J . Polymer
26,45-63 (1950). S C ~ 6,
. , 247-50 (1951).
(16) Gehman, S . D., J . Applied Phys., 13, 402-13 (1942) [Rubber (53) Tobolsky, A. V., Dunnell, B. A , , and Andrews, R. D., Teztile
Chem. and Technol., 15,860-73 (1942)l. Research J . , 21,404-11 (1951).
(17) Gehman, S. D., Woodford, D. E., and Stambaugh, R. H., (54) Tolstoy, N. A., and Feofilov, P. P., J . Experimenlal and Theo-
IND.ENG.CHEM., 33, 1032-8 (1941) [Rubber Chem. and ret. Phys. (U.S.S.R.), 19,421-30 (1949).
Technol., 14,842-57 (194111. (55) Voigt, W., “Lehrbuch der Krystallphysik,” Leipsig, 1928.
(18) Gross, B., J. Applied Phys., 18, 212-21 (1948); 19, 257-64 (56) Yerzley, F. L., IND. ENG. CHEM.,ANAL. ED., 9, 392-4 (1937)
(1949). [Rubber Chem. and Technol., 13,149-58 (1939)l.
(19) Harris, C. O., J . Applied Mech., 9, A129-Al35 (1942) [Rubber (57) Zener, C., “Elasticity and Anelasticity of Metals,” Chicago,
Chem. and Technol.. 16. 136-54 (1943)l. The University of Chicago Press, 1948.
(20) Hillier, K. W., and Kolsky, H . , ‘ P T o c . - P ~ ~SOC.
s. (London),
62B, 111-21 (1949). RECEIVED
for review October 2, 1951. ACCEPTE~D
January 31, 1952.

At
-

702 INDUSTRIAL AND ENGINEERING CHEMISTRY Vel. 44, No. 4

You might also like