Viscoelastic Behavior of Polymers PDF
Viscoelastic Behavior of Polymers PDF
Viscoelastic Behavior of Polymers PDF
10.1 STRESS RELAXATION AND CREEP In a stress relaxation experiment the sample is rapidly stretched to the required length, and the stress is recorded as a function of time. The length of the sample remains constant, so there is no macroscopic movement of the body during the experiment. Usually the temperature remains constant also (1).
Introduction to Physical Polymer Science, by L.H. Sperling ISBN 0-471-70606-X Copyright 2006 by John Wiley & Sons, Inc.
507
508
Creep experiments are conducted in the inverse manner. A constant stress is applied to a sample, and the dimensions are recorded as a function of time. Of course, these experiments can be generalized to include shear motions, compression, and so on. Section 8.1.6 dened the modulus of a material as a measure of its stiffness, and compliance as a measure of its softness. Under conditions far from a transition, E @ 1/J. Frequently stress relaxation experiments are reported as the time-dependent modulus, E(t), whereas creep experiments are reported as the time-dependent compliance, J(t). 10.1.1 Molecular Bases of Stress Relaxation and Creep While the exact molecular causes of stress relaxation and creep are varied, they can be grouped into ve general categories (2): 1. Chain scission. Oxidative degradation and hydrolysis are the primary causes. The reduction in modulus caused by chain scission during stress relaxation can be illustrated by a model where three chains are bearing a load and one is cut:
Oxidative scission
(10.1)
s 2s 3
Inversely, this causes an increase in elongation during creep. 2. Bond interchange. While this is not a degradation in the sense that the molecular weight is decreased, chain portions changing partners cause a release of stress. Examples of stress relaxation by bond interchange include polyesters and polysiloxanes. Equation (10.2) provides a simple example (3). Bond interchange is going on constantly in polysiloxanes, with or without stress. In the presence of a stress, however, the statistical rearrangements tend to reform the chains so that the stress is reduced. 3. Viscous ow. Caused by linear chains slipping past one another, this mechanism is responsible for viscous ow in pipes and elongational ow under stress. An example is the pulling out of Silly Putty. 4. Thirion relaxation (4).This is a reversible relaxation of the physical crosslinks or trapped entanglements in elastomeric networks. Figure 10.1 illustrates the motions involved. Usually an elastomeric network will relax about 5% by this mechanism, most of it in a few seconds. It must be emphasized that the chains are in constant motion of the reptation type (5) (see Section 5.4).
10.1
509
Figure 10.1 An illustration of reversible motion in a trapped entanglement under stress. The marked point of reference moves with time. When the stress is released, entropic forces return the chains to near their original positions.
CH3 Si O
CH3 Si O
CH3 Si O
s = constant
CH3
CH3
CH3 H3C O Si O Si
CH3 H3C
(10.2) CH3
CH3 Si O
Si
O
s=0
CH3
5. Molecular relaxation, especially near Tg. This topic was the major subject of Chapter 8, where it was pointed out that near Tg the chains relax at about the same rate as the time frame of the experiment. If the chains are under stress during the experiment, the motions will tend to relieve the stress. It must be emphasized that more than one of the relaxation modes above may be operative during any real experiment.
510
10.1.2 Models for Analyzing Stress Relaxation and Creep To permit a mathematical analysis of the creep and relaxation phenomenon, spring and dashpot elements are frequently used (see Figure 10.2). A spring behaves exactly like a metal spring, stretching instantly under stress and holding that stress indenitely.A dashpot is full of a purely viscous uid. Under stress, the plunger moves through the uid at a rate proportional to the stress. On removing the stress, there is no recovery. Both elements may be deformed indenitely.
10.1.2.1 The Maxwell and Kelvin Elements The springs and dashpots can be put together to develop mathematically amenable models of viscoelastic behavior. Figure 10.3 illustrates the simplest two such arrangements, the Maxwell and the Kelvin (sometimes called the Voigt) elements. While the spring and the dashpot are in series in the Maxwell element, they are in parallel in the Kelvin element. In such arrangements it is convenient to assign moduli E to the various springs, and viscosities h to the dashpots. In the Maxwell element both the spring and the dashpot are subjected to the same stress but are permitted independent strains. The inverse is true for the Kelvin element, which is equivalent to saying that the horizontal connecting portions on the right-hand side of Figure 10.3 are constrained to remain parallel. As examples of the behavior of combinations of springs and dashpots, the Maxwell and Kelvin elements will be subjected to creep experiments. In such an experiment a stress, s, is applied to the ends of the elements, and the strain, e, is recorded as a function of time. The results are illustrated in Figure 10.4. On application of the stress to the Maxwell element, the spring instantly responds, as illustrated by the vertical line in Figure 10.4. The height of the line is given by e = s/E. The spring term remains extended, as the dashpot gradually pulls out, yielding the slanted upward line. This model illustrates elasticity plus ow. The spring and the dashpot of the Kelvin element undergo concerted motions, since the top and bottom bars (see Figure 10.3) are constrained to remain parallel. The dashpot responds slowly to the stress, bearing all of it initially and gradually transferring it to the spring as the latter becomes extended.
Figure 10.2 Springs and dashpots are the basic elements in modeling stress relaxation and creep phenomena.
10.1
511
Figure 10.3 The Maxwell and Kelvin (or Voigt) elements, representing simple series and parallel arrays of springs and dashpots.
Figure 10.4 Creep behavior of the Maxwell and Kelvin elements. The Maxwell element exhibits viscous ow throughout the time of deformation, whereas the Kelvin element reaches and asymptotic limit to deformation.
When the spring bears all the stress, both the spring and the dashpot stop deforming together, and creep stops. Thus, at long times, the Kelvin element exhibits the asymptotic behavior illustrated in Figure 10.4. In more complex arrangements of springs and dashpots, the Kelvin element contributes a retarded elastic effect.
10.1.2.2 The Four-Element Model While a few problems in viscoelasticity can be solved with the Maxwell or Kelvin elements alone, more often they are used together or in other combinations. Figure 10.5 illustrates the combination of the Maxwell element and the Kelvin element in series, known as the four-element model. It is the simplest model that exhibits all the essential features of viscoelasticity.
512
Figure 10.5 (a) The four-element model. (b) Creep behavior as predicted by this model. At t1, the stress is relaxed, and the model makes a partial recovery.
On the application of a stress, s, the model (Figure 10.5a) undergoes an elastic deformation, followed by creep (Figure 10.5b). The deformation due to h3, true ow, is nonrecoverable. Thus, on removal of the stress, the model undergoes a partial recovery. The four-element model exhibits some familiar behavior patterns. Consider the effects of stretching a rubber band around a book. Initially E1 stretching takes place. As time passes, E2 + h2 + h3 relaxations take place. On removing the rubber band at a later time, the remaining E1 recovers. Usually the rubber band circle is larger than it was initially. This permanent stretch is due to h3. Although less obvious, the Kelvin element motions can also be observed by measuring the rubber band dimensions immediately after removal and again at a later time. The quantities E and h of the models shown above are not, of course, simple values of modulus and viscosity. However, as shown below, they can be used in numerous calculations to provide excellent predictions or understanding of viscoelastic creep and stress relaxation. It must be emphasized that E and h themselves can be governed by theoretical equations. For example, if the polymer is above Tg, the theory of rubber elasticity can be used. Likewise the WLF equation can be used to represent that portion of the deformation due to viscous ow, or for the viscous portion of the Kelvin element. More complex arrangements of elements are often used, especially if multiple relaxations are involved or if accurate representations of engineering data are required. The MaxwellWeichert model consists of a very large (or innite) number of Maxwell elements in parallel (2). The generalized VoigtKelvin model places a number of Kelvin elements in series. In each of these models, a spring or a dashpot may be placed alone, indicating elastic or viscous contributions.
10.1
513
10.1.2.3 The Takayanagi Models Most polymer blends, blocks, grafts, and interpenetrating polymer networks are phase-separated. Frequently one phase is elastomeric, and the other is plastic. The mechanical behavior of such a system can be represented by the Takayanagi models (6). Instead of the arrays of springs and dashpots, arrays of rubbery (R) and plastic (P) phases are indicated (see Figure 10.6) (7). The quantities l and j or their indicated multiplications indicate volume fractions of the materials. As with springs and dashpots, the Takayanagi models may also be expressed analytically. For parallel model Figure 10.6a, the horizontal bars connecting the two elements must remain parallel and horizontal, yielding an isostrain condition (eP = eR). Then
s = sR +sP s i = eEi , i = P, R E = (1 - l )EP + lER (10.4) (10.5) (10.6)
Figure 10.6a represents an upper bound model, meaning that the modulus predicted is the highest achievable for a two-phased mixture.
Figure 10.6 The Takayanagi models for two-phase systems: (a) an isostrain model; (b) isostress model; (c, d ) combinations. The area of each diagram is proportional to a volume fraction of the phase.
514
For example, take a 50/50 blend of a plastic and an elastomer. Typical Youngs moduli are EP = 3 109 Pa, and ER = 2 106 Pa. The quantity l equals 0.5, and equation (10.6) yields 1.5 109 Pa. To an excellent approximation, this is half the modulus of the plastic. For the series model Figure 10.6b, an isostress strain condition exists. The strains are additive, e = eP + eR yielding a Youngs modulus of E= 1-j j + ER EP
-1
(10.7)
(10.8)
This is a lower bound modulus, meaning that equation (10.8) yields the lowest modulus any blend material may have. If the example above is used, with j = 0.5, equation (10.8) yields E = 4 106 Pa, or approximately twice the elastomer value. Surprisingly, it is possible to make two different 50/50 blends of plastics and elastomers and obtain very nearly both of the results predicted by the above examples. This depends on the morphology, and particularly on phase continuity; see Figure 4.3. When the plastic phase is continuous in space and the elastomer dispersed (Figure 10.6a), the material will be stiff, and substantially exhibit upper bound behavior. On the other hand, when the elastomer is continuous and the plastic discontinuous (Figure 10.6b), a much softer material results and a lower bound situation prevails. The rubber phase is much more obviously discontinuous in the models represented in Figures 10.6c and 10.6d. While the examples above used Youngs modulus, many other parameters may be substituted. These include other moduli, rheological functions such as creep, stress relaxation, melt viscosity, and rubber elasticity. Each element may itself be expressed by temperature-, time-, or frequency-dependent quantities. It must also be noted that these models nd application in composite problems as well. For example, a composite of continuous bers in a plastic matrix can be described by Figure 10.6a if deformed in the direction of the bers, and by Figure 10.6b if deformed in the transverse direction. Figures 10.6c and 10.6d represent a higher level of sophistication, allowing for more precise calculations involving continuous and discontinuous phases. For Figures 10.6c and 10.6d, respectively, Youngs moduli are given by 1-j j E= + EP lER + (1 - l )EP 1-j j E=l + ER EP
-1 -1
(10.9) (10.10)
+ (1 - l )EP
10.2
515
For these latter models, the volume fraction of each phase is given by a function of l multiplied by a function of j. The volume fraction of the upper righthand portion of Figure 10.6d is given by l (1 - j). Equations (10.9) and (10.10) simplify for numerical analysis if j is taken equal to l.
10.2 RELAXATION AND RETARDATION TIMES The various models were invented explicitly to provide a method of mathematical analysis of polymeric viscoelastic behavior. The Maxwell element expresses a combination of Hookes and Newtons laws. For the spring, s = Ee the time dependence of the strain may be expressed as de 1 d s = dt E dt The time dependence of the strain on the dashpot is given by de s = dt h (10.13) (10.12) (10.11)
Since the Maxwell model has a spring and dashpot in series, the strain on the model is the sum of the strains of its components: de 1 d s s = + dt E dt h 10.2.1 The Relaxation Time For a Maxwell element the relaxation time is dened by t1 = h E (10.15) (10.14)
The viscosity, h, has the units of Pa s, and the modulus, E, has the units of Pa, so t1 has the units of time. Thus t1 relates modulus and viscosity. On a molecular scale the relaxation time of a polymer indicates the order of magnitude of time required for a certain proportion of the polymer chains to relaxthat is, to respond to the external stress by thermal motion. It should be noted that the chains are in constant thermal motion whether there is an external stress or not. The stress tends to be relieved, however, when the chains happen to move in the right direction, degrade, and so on.