Anda di halaman 1dari 20

Thesis -- Title Page 1 of 1

TRANSIENT SHEAR AND EXTENSIONAL


RHEOLOGY OF DILUTE AND SEMI-
DILUTE POLYMER SOLUTIONS
A Thesis

Presented in Partial Fulfillment of the Requirements for


the Degree Masters of Science in the
Graduate School of The Ohio State University

By

David Scott Shackleford, B. S.

*****

The Ohio State University


1996

Master's Examination Committee:

Dr. Kurt W. Koelling, Adviser


Dr. Jacques L. Zakin
Department of Chemical Engineering

Table of Contents

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/TITLE.HTM 4/14/2002
Thesis -- Contents Page 1 of 1

TABLE OF CONTENTS

Abstract

Chapters:

1. Introduction
1.1 Definition of Extensional Flow
1.2 Practical Examples of Extensional Flow
1.3 Rheology of Polymers
1.4 Motivation for this Study

2. Literature Review
2.1 Dilute and Semi-Dilute Concentration Regimes
2.2 Intrinsic Viscosity and Polymer Entanglement
2.3 Experimental Estimation of Critical Overlap Concentration
2.4 Prior Methods of Extensional Viscosity Measurement
2.4.1 Contraction Flows
2.4.2 Stagnation Point Flows
2.4.3 Fiber Spinning
2.4.4 The Rotating Clamp Technique
2.5 The Filament Stretching Extensional Rheometer
2.6 Mathematical Modeling of Extensional Flow
2.7 Test Fluids Used in Extensional Flow Studies

Bibliography

More to be added later

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/contents.htm 4/14/2002
Thesis -- 1. Introduction Page 1 of 5

CHAPTER 1

INTRODUCTION

1.1 Definition of Extensional Flow

Extensional flow, or extensional deformation, can be defined as deformation in which the velocity of the
flow varies with the direction of flow. Examples include squeezing or stretching flows. For solids,
extensional deformation is described by Hook's law, where the strain of the solid is proportional to the
normal stress acting on it. The proportionality constant is Young's modulus, which is a property of the
type of solid. Extensional deformation is shown in Figure 1.1 (a).

Figure 1.1 Comparison of Extensional and Shear Deformation for a Solid. (a) Extensional Deformation,
(b) Shear Deformation.

Shear deformation is distinguished from extensional deformation because it involves a velocity field
which varies in the direction perpendicular to the direction of flow. For solids, a shear modulus can be
defined as the ratio of the shear stress to the shear strain. For an ideal Hookian solid, the Young's
modulus is three times the shear modulus. Shear deformation is shown in Figure 1.1 (b).

Liquids can be deformed both in shear and extension, as well. Deformation of a liquid is typically called
flow, and examples of extensional and shear flow are shown in Figure 1.2:

Figure 1.2 Comparison of Extensional and Shear Flow in a Liquid. (a) Extensional Flow, (b) Shear
Flow.

For liquids, a property analogous to the modulus can be defined -- the viscosity. The shear viscosity is
the ratio of the shear stress to the shear strain rate. The shear strain rate is the derivative of the strain
with respect to time. Since fluid mechanics is usually more concerned with velocity than strain, shear
strain rate is rewritten as the derivative of the velocity with respect to the direction perpendicular to the
direction of flow:

(1.1)

Thus the shear viscosity, η, can be defined for a liquid to be:

(1.2)

where τyx is the shear stress, which is equal to the shear force acting on the fluid in the x-direction
divided by the surface area of a surface normal to the y-direction.

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/I_INTRO.HTM 4/14/2002
Thesis -- 1. Introduction Page 2 of 5

For liquids in extensional flow, another property, the extensional viscosity, ηE, can be defined as the
ratio of the first normal stress difference to the extensional strain rate, or the extension rate, :

(1.3)

where τ11 and τ22 are the normal stresses in the directions parallel to and perpendicular to the direction
of stretching respectively. For uniaxial extensional flow, extension rate, , is defined as:

(1.4)

where L is defined as the length of a fluid element in the direction of stretching and t is time.
Extensional viscosity has the same units as shear viscosity and has an analogous meaning for
extensional flow. Extensional viscosity for a Newtonian fluid is three times the shear viscosity.

Newton's Law of viscosity describes the behavior of ideal viscous liquids just as Hook's law of elasticity
describes ideal elastic solids. Real materials often do not behave as either purely solid or purely liquid,
but rather have some characteristics in between the two. This can be seen if the modulus seems to
depend on strain rate as well as strain or if the viscosity depends not only on strain rate but also the total
strain. Rheology, defined as the study of deformation and flow of matter, is the field of science which
seeks to characterize materials showing these non-ideal properties. Thus far, many materials have been
studied in shear and extension for mostly solid behavior, and in shear flow for mostly liquid behavior.
Extensional flow of liquids has proven to be very difficult to study until recently, and thus there is little
valuable data available. One of the purposes of this investigation is to generate extensional flow data by
measuring the extensional viscosity of fluids.

1.2 Practical Examples of Extensional Flow

There are many industrial processes that involve extensional flow. Manufacturers of artificial fibers
draw out thin filaments of fluid and stretch them as they solidify in order to increase their strength and
uniformity. Thin plastic sheets are produced by stretching out thicker sheets to many times there original
length and width. Extensional flow in polymer processing is both valuable in creating products with
molecular orientation and essential in shaping polymer products. On the other hand, extensional flow
has been attributed as the cause of flow-induced molecular degradation of polymers, which reduces the
quality of the product (Tadmor, 1979).

Other industries are also affected by extensional flow. Most chemical processes involve flow of liquids
through pipes which often have contractions or expansions. If a fluid has a very high extensional
viscosity, these parts of the flow process can have exceedingly high resistance. Pumping fluids through
fiber mats or filters can be extremely difficult if the fluid has a large extensional viscosity. If a fluid has
suspended solid particles, there is an extensional flow around these particles which may account for
difficulty in handling solid-liquid mixtures. Aerosol droplets tend to resist breaking up into smaller
droplets because of the extensional flow that must occur. Liquid-gas foams tend to be more stable if the
fluid resists the extensional flow required to collapse liquid bridges. Turbulent drag reduction observed
in dilute surfactant or polymer solutions may be related to the extensional flow properties of these fluids.

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/I_INTRO.HTM 4/14/2002
Thesis -- 1. Introduction Page 3 of 5

Some of these examples of extensional flow are illustrated in Figure 1.3.

Figure 1.3: Examples of Extensional Flow. (a) Fiber Spinning Process; (b) Polymer Sheet or Film
Stretching Process; (c) Flow Through a Pipe Contraction.

Figure 1.3 (continued): Examples of Extensional Flow. (d) Flow through a bed of cylindrical fibers; (e)
flow through a column packed with porous media; (f) break-up of aerosol droplets.

1.3 Rheology of Polymers

Extensional Flow is often an important characteristic of the processes used to handle polymers. The
rheology of polymers also tends to defy simplistic classification. Newton's law of viscosity usually fails
to describe the behavior of polymeric liquids, and Hook's law of elasticity fails to describe the behavior
of their solids. Quite often, however, it is the complexity of the rheology of these materials which makes
them ideal for use commercially. The remarkable elasticity of rubber, the tendency of plastics to bend
rather than break, and their wide range of strength to weight ratios, are some of their most important
qualities. Thus it is particularly interesting to study viscoelastic flow properties of polymers.
There are two basic types of polymeric liquids -- melts and solutions. Polymer melts are pure or nearly
pure polymer liquids and are generally formed by raising solid polymers to temperatures above their
melting temperatures which are usually around 200 o C. Since these materials are often solids at room
temperature, they must be handled at high temperatures in order to study their flow behavior. In
addition, polymer processing operations are usually performed at high temperatures, making high
temperature rheology studies more valuable. High temperature operation complicates laboratory
experiments which must be run in enclosed, temperature-controlled ovens (although many commercial
shear rheometers are equipped for this purpose). Polymers also tend to chemically degrade when held at
high temperatures for a prolonged period of time. Polymer solutions, on the other hand, are formed by
dissolving some polymer in a low molecular weight solvent which is usually a liquid at room
temperature. The solution does not need to be heated to high temperatures and thus handling and testing
is much simpler. Also, if the solution is stable, then the same fluid preparation can be used repeatedly
over a long period of time in a variety of ways without fear of degradation or a change in properties.
Both solutions and melts are used and handled in the polymer industry and understanding the rheology
of both is potentially useful.

Much work has already been done, and a general picture of polymer rheology has developed. Polymer
melts tend to be shear thinning -- that is, as the shear rate increases the viscosity drops. At very low and
very high shear rates, viscosity tends to level off and Newtonian behavior is observed. Many models,
called Generalized Newtonian Fluid (GNF) constitutive equations (Macosko, 1994), are available to fit
this shear rate dependence of viscosity. One example is the Cross Model:

(1.5)

Where is the viscosity plateau at high shear rates, ηo is the viscosity plateau at low shear rates, and n and
Kc are parameters describing the shear thinning region.

Melts also tend to be viscoelastic -- that is the viscosity is dependent on the total strain imposed as well
as the strain rate. This is often described by saying the fluid has a memory of all past strains. The Linear

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/I_INTRO.HTM 4/14/2002
Thesis -- 1. Introduction Page 4 of 5

Viscoelastic (LVE) constitutive equation (Macosko, 1994) integrates the strain rate imposed on the fluid
over time from negative infinity to the present.

(1.6)

Where G is the modulus of the material. This model is already quite complex (So much so that most
problems of real interest can only be solved numerically), and yet it is only a first approximation to the
behavior of viscoelastic fluids.

In shear flow, viscoelasticity makes the shear stress increase as time proceeds until a certain time is
reached at which the shear stress levels off and reaches a steady-state. This steady-state value is the
viscosity usually quoted for viscoelastic fluids. The time required to reach steady state is one measure of
the fluids relaxation time.

Viscoelasticity affects the extensional flow behavior of polymers significantly. Extensional viscosity of
polymers depends strongly on the total strain imposed. In other words the past history of strain in the
fluid changes the fluids resistance to further strain. The strain dependence may be more important than
the strain rate dependence in extensional flow. This supposition, however, is premature considering the
lack of experimental measurements which truly test a wide range of either strain or strain rates.

Polymer solutions, made by dissolving solid polymer in Newtonian liquids, are quite common and have
been studied deeply. The rheology of a solution tends to be a combination of the properties of the
polymer and the solvent. This is quite valuable for rheologists who want to select some optimal behavior
for their fluids. The viscosity of a solution is usually between that of the pure polymer and the pure
solvent, at the same shear rate. This can be deceptive, however, since viscosity is a strong function of
temperature and often the polymer and the solvent are not both liquids at any temperature (the solvent
may vaporize at a temperature lower than the polymer's melting point).

The elastic modulus of the polymer may be quite large while for the solvent it is usually zero. The
solution will have an elasticity between that of the pure polymer and the pure solvent. This means that
polymer solutions can be made which exhibit a significant amount of viscoelasticity.

An interesting type of solution, the Boger Fluid, can have almost no shear-thinning behavior at all, but
can have viscoelastic properties (Boger, 1977/78). This is achieved by dissolving a small amount of high
molecular weight polymer in a Newtonian solvent which already has a large shear viscosity. The
polymer contributes little to the viscosity but does have a large elastic contribution. The extensional
viscosity of these fluids appears to be many times the shear viscosity.

1.4 Motivation for this Study

Constitutive equations may describe extensional flow quite well, however, this can only be verified with
highly accurate experimental studies of extensional flow. Until recently, extensional flow theories were
difficult to test. Many variables of importance in extensional flow have not been measured
experimentally, and those that have been tested in the past have often fallen victim to misleading
influences and ambiguous results. Tests that cannot maintain a constant extension rate or that cannot
exposed all fluid elements to the same strain, cannot separate the influences of these two properties of
fluid.

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/I_INTRO.HTM 4/14/2002
Thesis -- 1. Introduction Page 5 of 5

The current study is intended to measure fluid properties in extensional flow as accurately as possible
using a novel experimental apparatus -- the filament stretching rheometer. These results can serve a
variety of purposes including: improving constitutive modeling, giving insight into the deformation
process at the molecular level, and understanding polymer processing design problems.

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/I_INTRO.HTM 4/14/2002
Thesis -- 2. Literature Review Page 1 of 13

CHAPTER 2

LITERATURE REVIEW

This research project involves experimental studies which build upon previous work conducted in
several areas. The critical overlap concentration of a Boger fluid composed of polyisobutylene,
polybutene, and tetradecane was measured using intrinsic viscosity and molecular arguments. The
extensional and shear rheology of these polymer solutions were studied, and the filament stretching
extensional rheometer was built and characterized. These results were then compared to molecular
theories and constitutive equations. The work done previously in these areas and the theoretical basis for
this work, are summarized in this chapter.

2.1 Dilute and Semi-Dilute Concentration Regimes

The concentration of the polymer in a solution, even when very low, has a large impact on the behavior
of the solution. It has been observed that the properties of a material are affected by the degree to which
polymer molecules entangle with one another (Graessley, 1974). In extremely low concentrations, this
entanglement is almost nonexistent and the rheology of the fluid is affected only by the nature of the
individual polymer chains. As the concentration is increased, the polymer entanglement adds a new
effect to the properties of the fluid. The point at which entanglement becomes important, known as the
critical overlap concentration, C*, has been a subject of much interest in rheology.

The concentration at which polymers begin to overlap could be calculated making, some assumptions
about the shape and distribution of the polymers. If one assumes that polymer chains are freely, and
randomly oriented so that their shape is a spherical gaussian distribution, then a measure of their size is
given by their radius of gyration, the average distance between any link in the chain and the center of
mass. It can be shown (Bird, Armstrong, Hassager, 1987) that this random state gives a radius of
gyration:

(2.1)

where l is the length of one bond on the backbone chain (equal to 0.154 nm for a carbon-carbon bond),
and n is the number of bonds:

(2.2)

where M0 is the molecular weight of the repeat unit and MW is the molecular weight of the polymer.

The critical overlap concentration, C*, is defined to be the concentration at which the volume
encompassed by polymers will be equal to the volume of the entire system (Fujita, 1990). This point can
be calculated quite specifically without any assumption about the polymer packing distribution in the
solution. The critical value can be defined as the concentration at which the volume swept out by the
radii of gyration of the polymer molecules present is equal to the total volume of the system:

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/ii_lit.htm 4/14/2002
Thesis -- 2. Literature Review Page 2 of 13

(2.3)

Solving for C* we arrive at:

(2.4)

One problem with this definition is that, in a real solution, each molecule will have a different size and
shape, each will move, and, as a group, they will be distributed randomly not uniformly. Thus, there is
no non-zero concentration at which no entanglement occurs, only a gradual progression of increasing
entanglement as the polymer density increases. The critical concentration does not represent any
fundamental break point in entanglement.

The assumption that the polymers are randomly coiled gaussian chains is not applicable to real
solutions, in general, because of two effects: (1) polymer-solvent affinity, and (2) the excluded volume
effect (Doi, Edwards, 1986). The molecular affinity of the functional groups of the polymer and solvent
influence the tendency of the coil to spread out or to contract in solution. This means that a good solvent
will cause the radius of gyration to be larger than a poor solvent. The excluded volume effect accounts
for the fact that two atoms cannot occupy the same space. Each atom must have a minimum finite
volume. This means the coil will be slightly larger than the gaussian distribution would predict. Systems
where these two effects exactly offset each other are said to be in the theta condition or Flory condition
(Flory, 1953). Since the solution thermodynamics depend on temperature, a solution will actually be in
the theta condition only at one particular temperature, known as the theta temperature, Θ. In the theta
condition, the radius of gyration is exactly equal to that defined by the gaussian distribution used above.
For other conditions a correction factor, , (Polymer Handbook, 1975) is defined by:

(2.5)

This factor, called the characteristic ratio or expansion ratio, was shown by Flory to be related to
temperature and Molecular Weight (Flory, 1953):

(2.6)

Where ψ1 is a factor relating to the entropy of dilution and CM is a function which very weakly depends
on molecular weight and concentration but can often be considered constant. This equation indicates that
larger molecules (with higher molecular weights) tend to expand more. Also it predicts that solvents
have more affinity at higher temperatures, and as the theta temperature is approached, reduces to
unity. Below the theta temperature, negative values result -- an impossible state, indicating that the
polymer should no longer remain in solution. This theory, however, may not apply to low molecular
weight polymers which may remain in solution even below the theta temperature.

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/ii_lit.htm 4/14/2002
Thesis -- 2. Literature Review Page 3 of 13

Using the correction factor, , it is possible to calculate the critical overlap concentration of a
polymer solution by the following equation:

(2.7)

Below this concentration, polymer entanglement is rare enough that we might expect it not to effect
physical properties of the fluid. Somewhere near this point we expect entanglement to have increasing
importance in physical behavior.

2.2 Intrinsic Viscosity and Polymer Entanglement

In dilute solution, a relationship between molecular weight and solution viscosity has been shown. In
order to interpret this relationship systematically, a quantity known as intrinsic viscosity, [ η], is defined:

(2.8)

The Mark-Houwink relation for intrinsic viscosity (Flory, 1953) and molecular weight is:

(2.9)

where K, or K' and a are adjustable parameters and are considered to be constant for the polymer-solvent
system. M v is the viscosity-average molecular weight, which usually falls between the number-average
and weight-average molecular weights for polydisperse systems. This relationship is often used to
determine the molecular weight of a polymer from viscosity measurements.

Some have proposed that the critical Simha parameter, the product of the intrinsic viscosity and the
critical overlap concentration, [ η] C*, should be a constant (Simha, Zakin, 1960). Quite a large amount
of experimental data from intrinsic viscosity and light scattering measurements is available to show that
the parameter [ η] C* is approximately equal to unity (Brown, Mortensen, 1988; Brown, Zhou, 1991;
Matsumoto, Mishioka, Fujita, 1972; Nakamura, Akasaka, Kataama, Norisuye, Teramoto, 1992; Fox,
Flory, 1951). Experimental measurements of this quantity vary over about half an order of magnitude
but unity is a good estimate. Making an assumption about the critical Simha parameter, it is then
possible to calculate the critical overlap concentration from measurements of intrinsic viscosity.

The relationship between viscosity and polymer size has been studied in terms of bead-spring models
(Graessley, 1974). These models assume that a polymer can be represented as a chain of beads
connected by individually stretching or compressing springs. The Rouse-Bueche model predicts that a
bead-spring chain will have a series of internal modes of motion which, when deformation occurs, are
manifested in a series of relaxation times:

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/ii_lit.htm 4/14/2002
Thesis -- 2. Literature Review Page 4 of 13

i = 1, 2, 3 ... (2.10)

where ζ0 is a frictional coefficient per segment, k is boltzmann's constant and T is temperature. The
viscosity of a dilute solution with no entanglement effects depends on the relaxation times as follows:

(2.11)

where ν is the number of chains per unit volume. Thus it can be shown for the Rouse-Bueche model
that:

(2.12)

Substituting for radius of gyration in theta conditions, the intrinsic viscosity is predicted to be exactly
proportional to molecular weight. This result in not realistic because the Rouse-Bueche model does not
include intramolecular interaction. Zimm's model is more general in this respect and predicts:

(2.13)

In theta conditions, Zimm's model predicts the Mark Houwink coefficient a = 1/2 which is consistent

with experimental data. In a very good solvent (where ) the value of a is 4/5 (Doi,
Edwards, 1986). Thus the polymer-solvent interaction and molecular weight can be used to predict a
dilute solution's viscosity. This also allows molecular weight measurement and polymer-solvent
interaction to be predicted from viscosity measurements. Zimm's model assumes that the solution is
dilute; if the concentration is high enough that entanglements begin, then a deviation from Zimm's
model is expected.

2.3 Experimental Estimation of Critical Overlap Concentration

Many Researchers have made an effort to determine the effect of entanglement on the properties of
polymer solutions by estimating the critical overlap concentration. Several different methods have been
used to experimentally determine the value, especially static light scattering (SLS) (Kunst, 1950),
dynamic light scattering (DLS -- also known as quasi-elastic light scattering or QELS) (Brown,
Mortensen, 1988), pulsed-field-gradient nuclear-magnet -resonance spectroscopy (PFG NMR) (Brown,
Zhou, 1991), small-angle neutron scattering (SANS) (Brown, Mortensen, 1988), other diffusion
coefficient measurements, and intrinsic viscosity measurements (Matsumoto, Nishioka, Fujita, 1972). Of
these methods, dynamic light scattering and intrinsic viscosity, are by far the most prevalent
measurements due to their simplicity, their accuracy and their correlation with theory.

Static light scattering can give a very accurate measure of the radius of gyration of a polymer molecule
in dilute solution. Once the actual radius of gyration is known, it is then possible to calculate the critical

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/ii_lit.htm 4/14/2002
Thesis -- 2. Literature Review Page 5 of 13

overlap concentration. It is still necessary, however, to measure some macroscopic property, such as
viscosity, to determine if there is any effect from polymer entanglement. Static light scattering requires
that the fluid sample be completely free of contaminating particles. If there are contaminants larger in
size than the polymer molecules, the large particles will scatter far more light than the polymer --
rendering the measurement meaningless. Dynamic light scattering can distinguish between elastically
scattering solid particles and inelastic dissolved polymer chains, and yields better results. Still, the
sample must be filtered to as high a degree of purity as possible, to maximize the effect of the polymer
on the overall scattering intensity.

E. D. Kunst (1950) used static light scattering to measure the radius of gyration of polyisobutylene
dissolved in n-heptane. Fox and Flory (1951) compared these measurements with the intrinsic
viscosities of the same fluids and concluded that a relationship between intrinsic viscosity and radius of
gyration existed:

(2.14)

where is nearly constant for all polymer-solvent systems and is approximately 2.5x10 23 (When
consistent units are used is 2.5x10 23 and is dimensionless, however, the value is often given when RG is
given in cm and [η] is given in dl/g such that the value of is 2.5x1021 , see Fox, Flory, 1951). Theoretical
analysis of the hydrodynamics predicts a value of 3.6x10 23 , however the value 2.5x10 23 agrees with
light scattering and intrinsic viscosity measurements (Fox, Flory, 1951).

Light Scattering has also been performed on several other polyisobutylene solutions and similar values
for have been determined. (Fujita, 1990) Other methods have been used to measure radius of gyration
and have been shown to yield values approximately the same as that measured by light scattering
(Brown, Zhou 1991) (Nakamura, et. al. 1992) (Matsumoto, Nishioka, Fujita, 1972).

Simha and Zakin (1960) used this relation developed by Fox and Flory and assumed an hexagonal
packing distribution to determine that the product of the critical entanglement concentration and the
intrinsic viscosity should be a constant, C*[ η] = 1.08. Graessley (1980) applied an statistical average
distribution to determine a value of C*[η] = 0.77. Fujita (1990) arrived at a different result by avoiding
an assumption of packing distribution, and determined a value of C*[ η] = 1.46. The exact number
determined depends on which assumptions are made, however, a value of approximately C*[η] = 1
seems reasonable.

2.4 Prior Methods of Extensional Viscosity Measurement

Measurements in extensional flow have been made in a variety of ways each with some distinct
advantages and drawbacks. Extensional flow is difficult to study because rarely can it be isolated
experimentally from shear flow. It is always at least two -dimensional and usually three-dimensional as
well as time dependent, and thus the equation of motion is usually a partial differential equation. This
makes isolating specific variables difficult to do from anything but very simple experiments. Thus, some
researchers have made efforts to perform complex flow experiments which do not yield actual viscosity
measurement but may establish trends and relationships between variables. Some methods used to study
extensional flow are illustrated in Figure 2.1.

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/ii_lit.htm 4/14/2002
Thesis -- 2. Literature Review Page 6 of 13

Figure 2.1: Methods Used to Experimentally Study Extensional Flow. (a) Contraction Flow; (b)
Opposed Nozzles; (c) Crossed Channel Flow; (d) The Four-Roll Mill; (e) The Spinline Rheometer; (f)
The Rotating Clamp Technique.

2.4.1 Contraction Flows

Since contraction flows are so common in industry, there has been a significant amount of study of the
extensional flow through contractions. This process is illustrated in Figure 2.1(a). Modification of
streamlines due to the extensional flow has been observed quite effectively and some measure of
extensional flow resistance can be made based on the difference in pressure drop due to extensional flow
(James, Saringer, 1982).

Planar contraction flows have been used in both experiments and constitutive model simulations
(Quinzani, Armstrong, Brown, 1995). Laser Doppler Velocimetry and flow birefringence were used to
measure the strain rate in the contraction flow. The simulations were successful in qualitatively
reproducing the flow field.

2.4.2 Stagnation Point Flows

Extension occurs around stagnation points and several devices have been designed to use this method to
generate extensional flow. The most prominent device using this method is the Opposed-Nozzles
Apparatus, shown in Figure 2.1(b). (Fuller, Cathey, Hubbard, Zebrowski, 1987) This device has two
cylindrical tubes submerged in a fluid which are oriented to oppose each other across a small gap. Fluid
is sucked in through both nozzles, inducing a stagnation point in the center of the gap. Extensional flow
occurs around the stagnation point as the fluid comes into the gap radially and exits axially. By
measuring the force acting to pull the tubes together, and the flow rate entering the tubes, an
approximate extensional viscosity can be estimated. This is the most common method used to measure
extensional viscosity, and it has the advantage that it produces similar strains to flow through
contractions or orifices.

One major problem with this method, however, is its failure to determine extensional viscosity as a
function of extension rate. For many fluids of interest, extensional viscosity rises strongly with
extension rate, but the opposed-nozzles' results show an increase at low extension rates and a decrease at
high extension rates. This may be due to the change in residence time of fluid in the gap. As flow rate is
increased (in order to produce higher extension rates), the residence time drops resulting in less total
strain on the fluid (Cathey, Fuller, 1990). In addition, this method does not isolate extensional flow from
shear flow and thus does not yield exact measures for extensional viscosity.

Stagnation point flows have been generated in Crossed-Channel devices, Figure 2.1(c), as well (Farrell,
Keller, Miles, Pope, 1980). These devices have four square channels which meet in the center of the
device. Fluid flows in from two opposite sides and flows out the other two, causing extensional flow at
the junction point. These devices are good for birefringence studies because of the rectangular geometry
and flat outer surfaces which minimize refractive distortion. They have also been used to estimate
extensional viscosity by measuring the pressure drop through the system. These measurements appear to
be highly approximate and difficult to interpret.

Four-Roll Mills, shown in Figure 2.1(d), have been used to study the birefringence occurring as a result
of extensional flow and shear flow (Fuller, Leal, 1980). The four different rollers can be rotated

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/ii_lit.htm 4/14/2002
Thesis -- 2. Literature Review Page 7 of 13

independently at different speeds to increase or decrease the amount of extension occurring. This system
has been useful for understanding some aspects of extensional flow but cannot measure extensional
viscosity.

2.4.3 Fiber Spinning

A more successful method of measuring extensional viscosity involves fiber spinning (Acierno,
Titomanlino, Greco, 1974; Ramanan, Gauri, Koelling, Bechtel, Forest; Gauri, 1996). A spinline
rheometer, shown in Figure 2.1(e), pumps fluid out of a small nozzle and then winds up the fluid on a
roller, placed some distance away. By measuring the force acting on the nozzle and the diameter of the
fluid fiber, it is possible to calculate an average extensional viscosity. Several problems develop using
this method as well. The extension rate is not constant but varies along the length of the fiber. This
complicates the attempt to relate extensional viscosity to extension rate. The fluid has some shear history
because of the flow through a long tube before being stretched. Also, the fluid shows some die swell,
after leaving the tube.

Because of these complexities, an involved mathematical treatment is required. Still, this device is very
similar to the industrial fiber spinning process and may be very important in applying extensional flow
theory to practical problems.

2.4.4 The Rotating Clamp Technique

A technique designed for polymer melts and very high viscosity materials involves stretching thin sheets
or cylinders between two rotating clamps or between one stationary clamp and a rotating clamp
(Meissner, Raible, Stephenson, 1981; Padmanabhan, Kasehagen, Macosko, 1996). This process is
shown in Figure 2.1(f). The clamp rotation rate is increased exponentially with time such that the
extension rate remains constant. The extension rate can be determined from video camera monitoring of
the sample diameter. Often, however, polymer melts require high temperatures which can only be
attained in closed ovens, making it difficult to monitor the sample with a camera. In such cases, the
extension rate can be determined from the wind-up rate and from geometric considerations.

The chief advantage of this geometry is that the flow field approximates a simple start-up of steady
extensional flow, which can be easily modeled. This advantage has made the rotating clamp technique
the best method for testing the extensional viscosity of high viscosity fluids such as polymer melts.

2.5 Filament Stretching Extensional Rheometer

For polymer solutions, the rotating clamp technique is often impractical which has motivated some to
use a new technique called filament stretching (Tirtaatmadja, Sridhar, 1993). Similar to the spinline
rheometer, a thin fluid fiber is stretched, the axial force on the fluid is measured and the diameter of the
fiber is measured. Unlike the spinline however, the extension rate is held constant throughout the
experiment and the fluid is initially at rest, so that minimal strain history is present in the fluid.

The specific design considerations are discussed in more detail in section 3.4, but the general procedure
is as follows. Two circular plates are mounted parallel to one another and the fluid is loaded between
them. The surface tension of the fluid holds it to the plates such that it forms a liquid bridge. An attempt
is made to add just enough fluid so that the shape of the fluid sample is cylindrical. After the sample is
loaded and allowed to relax, the plates are pulled apart from one another; slowly at first, but with a

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/ii_lit.htm 4/14/2002
Thesis -- 2. Literature Review Page 8 of 13

velocity growing exponentially with time. In order to maintain a constant extension rate, the length of a
cylindrical fluid sample must increase according to:

(2.15)

where L is the length the sample, L 0 is the initial length, t is time and is the extension rate which is
intentionally held constant. The diameter of a constant-volume cylindrical sample is related to the length
by Vol = (1/4) π D2 L. The diameter can be measured either by a video camera or more precisely by a
plane laser. Thus the constant stretching rate restricts the change in diameter to:

(2.16)

Where D0 is the initial diameter. A real fluid sample does not deform as a perfect cylinder but forms a
narrow hour-glass shape. Most of the fluid sample has a diameter very close to that measured at the
centerline. Thus, if the stretching can be controlled such that the diameter at the center changes
according to the expression above, then most of the fluid will experience the same extension rate.

The force required to stretch the fluid can be measured with a force transducer, and the ratio of the
measured force to the cross-sectional area of the filament at the centerline is related to the first normal
stress difference by a force balance. Dividing the stress by the extension rate gives an extensional
viscosity for the fluid. If the fluid were Newtonian, the extensional viscosity would be three times the
shear viscosity and would be constant throughout the process, after the filament attains a uniform
diameter. Viscoelastic fluids, however, have shown very strong increases in stress as a function of strain.

The first such device was developed by Tirtaatmadja and Sridhar (1991) and was oriented vertically.
This design makes some experiments simpler but adds the uneven influence of gravity. Initially, the
extension rates were not held constant, but over time they were able to develop an algorithm to maintain
a constant extension rate (1993). A later design was oriented horizontally in order to minimize the
importance of gravity. The maximum length that the filament could attain was 1000 mm and the largest
Hencky stains were approximately 7.2. They tested three Boger fluids (labeled fluids A, B, and M1) and
one shear-thinning fluid (A1) at a variety of extension rates ranging from 0.6 to 9.0 s -1 (but with data
predominantly between 1.0 and 5.0 s -1 ). The results show a rapid increase in extensional viscosity as the
strain grows but show little difference for different extension rates when plotted as extensional viscosity
vs. Hencky strain. A steady-state extensional viscosity appears to be attained for some runs of the
experiment at strains ranging from 5.0 to 7.0. This steady state value was reached at a Trouton ratio of
approximately 1000 to 3000 for fluids A and B and at approximately 8000 for fluid M1. No steady state
was reached for fluid A1 but Trouton ratios did reach 1000.

The same research group went on to test other fluids such as the standard test fluid S1 (Ooi, Sridhar,
1994a). Extensional viscosities measured were similar in magnitude to those measured for fluid A1,
however, a steady-state value was reached with S1 for extensional viscosities between 600 and 20000
Pa-s. They attempted to determine a trend in steady-state extensional viscosity as a function of extension
rate, and found that the values rise at low extension rates, reach a maximum at about 2.0 s -1 and begin to
fall at higher shear rates. They suggested that the decreasing extensional viscosity could be attributed to
the shear-thinning nature of the fluid.

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/ii_lit.htm 4/14/2002
Thesis -- 2. Literature Review Page 9 of 13

Ooi and Sridhar (1994b) continued their investigations by testing a lyotropic liquid crystal polymer
solution of (hydroxypropyl) cellulose in glacial acetic acid. Because of the liquid crystal structure of the
fluid, it was possible to consider two theories to predict the extensional stress growth. The Leslie-
Ericksen continuum theory and the Doi molecular theory were compared to experimental results. The
extensional viscosity measured by filament stretching, approached steady-state although most of the data
did not actually reach a constant value before the end of the run. The results show some qualitative
agreement between the models and experiments but both theories predict much more rapid approaches
to steady state than the experimental evidence supports. The theories did demonstrate the proper trends
relating stress and extension rate as the extensional viscosities approached steady-state.

Speigelberg, Ables and McKinley (1996) built a similar filament stretching rheometer, using a video
camera to study the nonuniform changes in the surface shape of the fluid sample. This device was much
smaller, reducing the range of Hencky strains accessible. The experiments, however, were intended to
study the variation in shape at low strains as the fluid sample initially begins to stretch. They predicted
the surface shape evolution using a modified lubrication model. They concluded that the aspect ratio --
or the ratio of filament's initial length to radius -- does have a pronounced effect on the extensional
viscosity measured immediately after the experiment begins. The effect decreases as the experiment
continues, however, it may have some impact on the results even at large times. They also compare the
results of a Boger fluid and a Newtonian fluid, showing that both fluids behave the same way at low
strains, and that the viscoelastic character of the Boger fluid develops over time as the strain grows.

Solomon and Muller (1996c) tested three polystyrene-based Boger fluids which varied in molecular
weight, concentration, and solvent. They compared the extensional viscosity of fluids with good and
poor solvents in order to determine the relationships between solvent quality, polymer extensibility and
rheological properties. Results showed that the stress response of a polymer solution with a good
solvent, and thus reduced extensibility, was weaker than that of the same polymer in a poor solvent.

Berg, Kroger, and Rath (1994) developed a self-contained rheometer that can be placed in a drop tower
and allowed to free fall for as long as 4.7 seconds. This free fall eliminates the effects of gravity and
allows for extensional viscosity measurement in near weightlessness. The design was further improved
by adjustable-diameter sample holders which reduce the necking phenomena observed in other
experiments. These new sample holders force the fluid to maintain a cylindrical shape so that the fluid
motion is very close to uniaxial extension. The maximum length for the fluid sample was 20.0 cm which
limited their experiment to Hencky strains of less than 1.4. The mechanical constraints of the system
limited the extension rates to between 0.4 and 1.0 s -1 .

This device was used to determine the extensional viscosity of a technical acrylamide copolymer
(Stockhausen PAA 2530), with a weight-average molecular weight of 5.6x10 6, dissolved in water at
various concentrations. These fluids were strongly shear thinning and show zero-shear viscosities
between 0.2 Pa-s for a 0.0125 weight percent solution up to 30 Pa-s for a 0.2 weight percent solution.
The results show that the extensional viscosity increased with increasing stretching time and all data at
different extension rates collapsed quite closely when plotted as extensional viscosity versus strain. The
extensional viscosities found were much larger for the higher concentration fluids and extensional
viscosity at individual strain values appears to increase linearly with concentration.

Kroger and Rath (1995) visualized the velocity field in an elongating liquid bridge at small strains after
the initiation of steady extension rate in the center of the bridge. Video imaging was used to follow
tracer particles in the liquid bridge to determine the local velocity, strain and rate of deformation. Their
concern was the non-uniformity of the extension rate to which the fluid was exposed. For viscoelastic
fluids, it was found that the contours of the stretching bridge could be correlated with a modified

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/ii_lit.htm 4/14/2002
Thesis -- 2. Literature Review Page 10 of 13

capillary number:

(2.17)

which compares the extensional viscous force to the surface tension force. They determined that only
with capillary numbers of one or greater would the fluid retain a mostly cylindrical shape.

2.6 Mathematical Modeling of Extensional Flow

In order to understand fluid flow quantitatively and to verify theoretical ideas with experiments, it is
necessary to develop constitutive equations relating stress to strain, strain rate, strain history, and fluid
properties. These models are often based on a combination of molecular level interactions and on
macroscopic force balances. They must be constructed in mathematically rigorous fashion, while
representing as many experimental observations as possible. These requirements often lead to such
complex equations that using them becomes impractical. Furthermore, complex models are highly
sensitive to experimental error when curve-fitting is performed. The multiple parameters used in
complex models require the availability of a wide range of high quality experimental data. Simplistic
models, however, fail to predict some of the fundamental behavioral characteristics of many common
fluids. It is thus important to strike a balance between complexity and practicality when developing and
using constitutive models.

In extensional flow, the difficulties of modeling are accentuated by poor experimental data, and a lack of
molecular level understanding of deformation phenomena. Recent advances in extensional flow
measurement methods should help to rectify the first problem. It is now possible to test fluids at nearly
constant extension rates and to measure their extensional viscosities as a function of strain and strain
history. These experimental techniques, however, must be used to glean more information about
molecular level deformation, before constitutive models can accurately predict the behavior of fluids in
extension.

Specifically for polymer systems, experimenters must learn how molecules deform and relax in
extensional flow and how these processes differ in shear flow; how molecular interactions affect
deformation; how polymers and solvents contribute to the behavior of solutions and mixtures; and how
well bead-spring approximations represent the behavior of real molecules. Once these concepts are well
established, constitutive models should prove to be effective in describing actual behavior.

Models currently available for describing extensional flow, were developed without the benefit of many
extensional flow experiments and so were based largely on viscoelastic behavior in shear flow. Often
these models produce some semblance of extensional flow behavior, but quantitative predictions fall
orders of magnitude away from actual data. This failure leaves these models almost useless for
engineering work.

One answer to this problem is to increase the number of parameters of the models. It then becomes
possible to reproduce the laboratory results quite quantitatively. The physical significance of the many
parameters is questionable, however. Often, it is also impossible to solve these complex equations
without numerical simulations, with the accompanying numerical error.

Although these problems persist, comparison of currently available models with experimental data is

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/ii_lit.htm 4/14/2002
Thesis -- 2. Literature Review Page 11 of 13

necessary in order to develop better models in the future. The simplest model, the linear viscoelastic
model (LVE), or Maxwell model (Bird, Armstrong, Hassager, 1987), is a two parameter model. The
steady, zero-shear-rate viscosity and a relaxation time, are used to characterize the fluid. This model
does predict extensional viscosity growth but predicts no steady state in extensional viscosity, and fails
to describe extension rate dependence. In its tensor form, the Maxwell Model is:

(2.18)

One major limitation of this model is its dependence on the reference frame of the coordinate system
used. In order to generalize the results for all coordinate systems, a convected time derivative can be
used in place of the partial derivative. This convected derivative is defined as (Bird, Armstrong,
Hassager, 1987):

(2.19)

where the capitalized differential operator is the substantial time derivative:

(2.20)

Another improvement can be achieved by considering the contribution of a polymer in solution


separately from the contribution of the solvent:

(2.21)

(2.22)

(2.23)

This model is known as the convected Jeffreys model or the Oldroyd-B model. It is better able to predict
the transient growth of extensional viscosity in real fluids but still predicts no steady state.

A further improvement in the model can be achieved by incorporating a term that is quadratic in stress.
If this method is chosen and the stress contributions of the solvent and polymer are considered
separately, the result is the Giesekus Model (Giesekus, 1982):

(2.24)

(2.25)

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/ii_lit.htm 4/14/2002
Thesis -- 2. Literature Review Page 12 of 13

(2.26)

where τs, τp , ηs, and ηp are the solvent and polymer contributions to the stress tensor and the viscosity,
respectively. The parameter α is the "mobility factor" and is added in order to determine the magnitude
of the quadratic term. The Giesekus model has four parameters and can be used to effectively describe
the behavior of extensional flow, under both transient and steady state conditions. It fails to describe the
exact slope of the transient behavior, but this is to be expected in a model with so few parameters.

Other non-linear models are available, but their performance is generally no better than the Giesekus
Model for extensional flow studies. The available alternatives include the White-Metzner Model, the
Bird-DeAguiar Model, the FENE-P Model (Bird, Armstrong, Hassager, 1987), and the Chilcott-Rallison
Model (Chilcott, Rallison, 1988).

In order to improve the curve-fitting ability of these models, more than one relaxation mode can be
included (Bird, Armstrong, Hassager, 1987). It is assumed that each mode contributes to the stress
tensor independently such that the total stress tensor is simply the sum of the contributions from each
mode:

(2.27)

where τi is the contribution of the i-th mode, and m is the total number of modes used. Each mode
involves a separate set of parameters, which are used in the model to determine each modes stress
contribution.

2.7 Test Fluids Used in Extensional Flow Studies

It has proven to be difficult to measure the extensional viscosity of polymer solutions with the
techniques described above. The filament stretching rheometer is a new device and appears to be
promising, however, the other devices have often failed to produce meaningful experimental values.
Because the measurements from different techniques were often very different, several standardized test
fluids were developed in order to compare measurement devices. Fluids M1, S1 and A1 (named for
Monash University, University of Strathclyde, and the University College of Wales, Aberystwyth) were
three of these fluids. Each fluid was prepared in one large batch and was shipped to various laboratories
where they could be tested. The A1 project results were summarized by Hudson and Jones (1993). The
fluid was tested in shear using rotational viscometers and in extension using spinline rheometers,
pendant drop tests, opposed jets, and converging channel devices. Comparisons of shear rheology were
quite good, but extensional viscosity measurements differed by several orders of magnitude, and often
did not show the same trends. The results were so different that even the trend of extensional viscosity
with extension rate was not the same for all experiments.

Although the filament stretching rheometer appears to yield good measurements for extensional
viscosity, more work is needed in order to demonstrate that the results are meaningful, useful, and
reproducible. The present study was developed to build an extensional rheometer based on the design of
Tirtaatmadja and Sridhar (1993) and test its ability to perform measurements with which previous

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/ii_lit.htm 4/14/2002
Thesis -- 2. Literature Review Page 13 of 13

methods have failed.

http://kcgl1.eng.ohio-state.edu/~shackleford/thesis/ii_lit.htm 4/14/2002

Anda mungkin juga menyukai