Вы находитесь на странице: 1из 12

A New Proposal for Polymer Dynamics in Steady

Shearing Flows
VIJAY R. MHETAR, L. A. ARCHER
Department of Chemical Engineering, Texas A & M University, College Station, Texas 77843

Received 22 March 1999; revised 20 September 1999; accepted 4 October 1999

ABSTRACT: Beginning with a recently proposed expression for the drag force on a single
macromolecule pulled with constant velocity through a fluid of long-entangled molecules (V. R. Mhetar and L. A. Archer, Macromolecules 1998, 31, 6639), we investigate
the effect of entanglement loss on polymer dynamics in steady shearing flows. At
steady-state, a balance between the elastic restoring force and viscous drag acting on
entangled polymer segments reveals a critical molecular strain g m,c beyond which the
drag force exerted on polymer molecules by their neighbors is insufficient to support
arbitrarily small orientation angles. Specifically, we find that in fast steady shear flows
t 21
, g , t 21
d
Rouse, polymer orientation in the shear plane approaches a limiting angle
x c ' atau(1/(1 1 g m,c )) beyond which flow becomes incapable of producing further
molecular alignment. Shear flow experiments using a series of concentrated polystyrene/diethyl phthalate solutions with fixed entanglement spacing, but variable polymer
molecular weight 0.94 3 10 6 # M w # 5.48 3 10 6 , reveal a limiting steady-state
orientation angle between 6 and 9 over a range of shear rates; confirming the
theoretical result. Orientation angle undershoots observed during start-up of fast
steady shearing flows are also explained in terms of a transient imbalance of elastic
restoring force and viscous drag on oriented polymer molecules. Our findings suggest
that the DoiEdwards affine orientation tensor (Q) is not universal, but rather depends
on deformation type and deformation history through a balance of elastic force and
viscous drag on polymer molecules. 2000 John Wiley & Sons, Inc. J Polym Sci B: Polym Phys
38: 222233, 2000

Keywords: polymer dynamics; steady shear flow; orientation angle; DoiEdwards


theory; entanglement loss; partial retraction; convected constraint release; viscous drag

INTRODUCTION
The DoiEdwards (D-E) theory contends that in
melts and solutions of entangled macromolecules
translational motion of individual molecules is
confined to tube-like regions surrounding the molecular contour. When an entangled polymer is
subjected to sudden straining, the tube and chain
deform affinely. Stresses induced by the deformation relax on two time-scales. At short times of
Correspondence to: L. A. Archer (E-mail: laa@unix.
tamu.edu)
Journal of Polymer Science: Part B: Polymer Physics, Vol. 38, 222233 (2000)
2000 John Wiley & Sons, Inc.

222

order the longest Rouse relaxation time of the free


polymer, t Rouse, stretched chain segments are assumed to retract along their respective tubes to
restore the equilibrium contour length. Following
equilibration of contour length, affine orientation
of chain segments relax by reptation. For t
. t Rouse the instantaneous stress in the material
is then,

s 5 ~15/4!G NQ~E! c ~t!,

(1)

where G N is the elastic modulus of the entangled


polymer network; E the deformation gradient
tensor; Q(E) the second moment tensor of seg-

POLYMER DYNAMICS IN STEADY SHEARING FLOWS

ment orientation vectors u; Q 5 (1/^uE z uu&)^(E


z u)(E z u)/uE z uu&; and c (t) a reptation relaxation function, c (t) 5 p;odd 8/p 2 p 2 exp(2p 2 t/ t d ),
where t d0 is the terminal orientation relaxation
or reptation time.1
Equation 1 predicts that for t . t Rouse stress is
factorizable into time-dependent, G N c (t), and
strain-dependent, Q(E), parts. For a shear strain
of magnitude g, the strain-dependent shear stress
contribution may be approximated as Q xy ( g ) '
4/15g 2
. Experimental results using weak to
~1 1 g2 /5!
moderately entangled solutions of narrow molecular weight distribution poly(styrene) provide
support for the latter result.3,4 This functional
form of Q xy ( g ) has nonetheless been the source of
a long-standing paradox in the D-E theory. Specifically, the prediction for Q xy ( g ) implies that for
shear strains g . =5 a larger imposed strain
yields a smaller shear stress, s xy ( g ) ; 1/ g at high
strains. This unusual result follows directly from
the assumption that polymer chain segments orient affinely in response to an imposed strain,
regardless of the size of that strain. Monotonic
decrease in the second moment tensor Q xy ( g ) at
high shear strains reflects enhanced segment
alignment parallel to the direction of shear, x,
compared with alignment parallel to the direction
of the shear gradient, y.
The multivalued shear stress result observed
in step shear flows carries over to D-E stress
predictions in all time-dependent shearing flows.
In these flows, the shear stress at any instant in
time is taken to be the affine stress accumulated
over the history of the deformation, weighted by
the materials memory ( c (t 2 t9)) of that stress,
t
s xy ( g ) 5 15/4 G (0)
N * 2` Q xy (E(t, t9)) c (t 2 t9)/t9
1
dt9. Thus, a continuous shearing deformation
imposed at a rate g , t 21
Rouse is assumed to orient
polymer chain segments in proportion to the effective strain accumulated by molecular segments
at steady-state. The orientation of polymer segment vectors is therefore anticipated to monotonically approach the direction of shear as shearing
becomes progressively faster. The consequent decrease in shear stress with increasing rate yields
the unusual prediction that at high rates apparent steady-state viscosity h 5 ( s xy ( g )/ g ) scales
with shear rate as h ; g 23/2. Thus, the very same
DoiEdwards orientation tensor Q that provides
favorable predictions in step-shear deformations
yields unrealistic levels of shear thinning in
steady shear flow. Any attempt to fix Q to improve predictions in steady shearing has the un-

223

desirable effect of reducing the quality of the D-E


step-strain predictions.
Several proposals have been advanced in the
literature to correct the D-E steady shear predictions without changing Q. All but one,5,6 achieve
this by including more realistic descriptions of
shear stress contributions due to flow-induced
changes in polymer contour length. Reviews outlining the successes and shortcomings of each of
these proposals are available,5,7,8 and will not be
discussed here. Instead, a different approach for
evaluating polymer orientation dynamics in
steady flows is addressed. In this approach the
orientation of entangled polymer chain segments
in a deformation field is contended to be determined by a balance between the elastic restoring
force due to molecular alignment parallel to the
direction of shear and viscous drag exerted on
chain segments by their neighbors, which retards
restoration of the equilibrium random distribution of molecular segment vectors. Specifically, we
propose that arbitrarily small orientation angles
x between entangled polymer segments and a
macroscopic flow are possible only if the viscous
drag force on segments is sufficiently large to
balance the entropic, elastic restoring force that
continually attempts to return segments to their
equilibrium orientation distribution.
This approach is different from that taken in
the DoiEdwards theory, where orientation of
polymer segments in any flow field is completely
described by the affine orientation tensor, Q, for
that flow field. A consequence of the new proposal
is that at moderate shear rates, in a steady shear
flow, the orientation of polymer chain segments
approaches a limiting angle beyond which the
flow becomes incapable of inducing further molecular alignment. Our result appears to resolve the
long-standing discrepancy between good agreement between DoiEdwards step-strain predictions and experimental results, and the unusually
high levels of shear thinning predicted in steady
shearing flows.1,5 Experimental support for the
proposal is discussed in the Results and Discussion section of the article.

Theory
Consider a homogeneous melt of flexible, entangled macromolecules each with degree of polymerization N and entanglement spacing N e0 . At equilibrium, each of the N-mer molecules will on average entangle with (N/N e0 2 1) surrounding

224

MHETAR AND ARCHER

Figure 1. A polymer melt subjected to a simple shear


flow. At low shear rates, g ! 1/t d0 (N), the primitive
(N-mer) chain is essentially a Gaussian coil. The polymer chain moves with an average curvilinear velocity,
V T , which is related to its center of mass velocity V c by,
V C 5 V T R/L c (N).

molecules. If a uniform shear flow is applied to


the system (N-mer chain and surrounding mole21
cules at equilibrium), such that g ! t d0
, the
frictional drag force experienced by each N-mer is
identical to the force that would arise if the molecule is pulled by its center of mass with a directed velocity V c 5 R g through a sea of chemically identical, but stationary molecules (P-mers),
(see Fig. 1).
Recently, Mhetar and Archer,9 showed that a
pulled macromolecule (N-mer) will not instinctively follow a curvilinear path around entanglement points. These authors contended, as did
Bueche almost half a century ago,10 that the Nmer will instead continually attempt to drag surrounding molecules entangled with it (P-mers) as
it translates through the entangled polymer network. Because P-mers are themselves entangled
and the N-mer is here free at both its ends the
pulled N-mer cannot drag P-mers indefinitely as
argued by Bueche,10 rather the N-mer will on
average only induce P-mer translation over a
short distance, of order the mean entanglement
spacing a 5 =N e0 b. Beyond that point, the large
frictional resistance offered by polymer molecules
entangled with the P-mer arrests its motion and,
thereby, compels the pulled N-mer to slide
through its entanglement with the P-mer to maintain the applied velocity. This sliding motion
traces out a curvilinear (tube-like) path around
the N-mer contour with diameter equal to the
mesh size of the entanglement network =N e0 b.
The frictional resistance the N-mer experiences
per entanglement site is therefore P z m , the drag
coefficient per N-mer molecule caused by the combined dragging and sliding motions is then,

zN 5

N
N z m,
N e0

(2)

where z m is the monomeric friction coefficient of


the N-mer, and we have assumed P 5 N.
The predicted halting curvilinear motion of the
N-mer molecule progresses at an average speed
V T that is related to the induced center of mass
velocity V c by L c (N)/V T 5 R/V c , where L c (N) and
R are the N-mers mean contour length and radius of gyration, respectively. The total drag force
offered by the surrounding molecules to the Nmers motion is therefore not the usual Rouse
drag assumed for polymer chains sliding intuitively through monomer-filled tubes: fdrag
5 V T z Rouse 5 V T N z m , rather f drag 5 V T z N .
Before proceeding, it is useful to derive the
limiting Newtonian viscosity h0 and terminal molecular relaxation time t d0 predicted by eq 2 for
an entangled melt of equal-sized molecules. In a
slow steady shear flow, the rate of viscous dissipation per unit volume is h0g 2. The dissipation
due to a single N-mer is z C V 2C 5 z C R 2 g 2 , where z C
is the center of mass friction coefficient. The rate
of dissipation per unit volume is therefore
z C R 2 g 2 /Nb 3 , yielding h 0 5 z C /b where b is the
monomer size. When the N-mer moves by L c (N)
along its curvilinear path, its center of mass
translates a distance R. Equating dissipation
rates yields,

zC 5 zN

S D S D
VT
VC

5 zN

L c~N!
R

<

N3
2 zm
N e0

(3)

where V T is the curvilinear velocity of the N-mer.


Thus, for well entangled polymers N/N e0 @ 1, the
limiting Newtonian shear viscosity is predicted to
2
2
be h 0 5 5 ( z m /b) N 3 /N e0
5 h m N 3 /N e0
, which is
similar to the result obtained by integrating the
relaxation modulus predicted by the DoiEdwards theory.1
Thermal fluctuations in the N-mer contour
length modify L c (N) in eq 3 by its fluctuating
equivalent, L c (N) 5 L c (N)(1 2 b ( =N e0 /N)). 11
The limiting shear viscosity then becomes, h 0
2
5 h m N 3 /N e0
(1 2 b ( =N e0 /N)) 2 ; where b is a
constant of order unity. The shear modulus of the
network is just G N ' 3 kT/(N e0 b 3 ). These two
results define a terminal molecular relaxation
time, td0 [ (h0/GN) ' tm(N3/Ne0(1 2 b(=Ne0/N))2).
Here h m ' ( z m /b) and t m 5 h m /(3 kT/b 3 ) are the
monomeric viscosity and segment jump time, respectively.
The frictional drag force acting on the N-mer
21
molecule at low shear rates, ( g ! t d0
), is

POLYMER DYNAMICS IN STEADY SHEARING FLOWS

f drag 5 z NV T <

S D
S D
N2
z V
Ne m T
5

N2
N3
z m L cg < 3/2 z mb g
Ne
Ne

(4)

where, for simplicity, we have ignored the fluctuation correction to L c (N). Substituting z m
5 3 t m kT/b 2 yields,
f drag 5

225

3kT

N eb

S D
tm

N3
3kT
t g .
g <
Ne
N eb d0

(5)

At steady-state, this frictional drag force is balanced by an entropic elastic restoring force,
f elastic. The elastic force results from flow-induced
orientation of polymer chain segments and may
be estimated as follows. Consider a N-mer molecule under the influence of a shear strain g m . The
strain is taken to be sufficiently small that deviations from Gaussian coil statistics are minimal.
The elastic force acting on the molecule is then,
f elastic 5

3 kT
3kT
g
2 L c~N! g m <
Nb
N e0b m

Figure 2. (a) Primitive N-mer segment of unit length


at equilibrium, average orientation angle x 5 45, (b)
oriented primitive N-mer segment subjected to a shear
strain g m . Here, L x and L y are the projections of the
primitive chain segment length in the shear ( x) and
shear gradient ( y) directions, respectively.

must first find its average orientation at that rate.


Again, let g m be the effective molecular shear
strain at steady-state. Also, let l x and l y be the
projection of a N-mer segment of unit length in
the shear ( x) and shear-gradient ( y) directions,
respectively (see Fig. 2). At steady-state, the average orientation of the N-mer segment in the
shear plane can then be estimated in terms of l x ,
l y , and g m as

(6)

Equating eqs 5 and 6 yields g m 5 t d0 g , which


simply states that at steady-state the effective
shear strain experienced by polymer molecules is
proportional to the product of their terminal relaxation time and the imposed deformation rate; a
result that is supported by experiment. Thus, unlike the imposed strain g 5 g t, g m is a timeindependent molecular strain that reflects the
balance between flow-induced deformation and
molecular relaxation at steady-state.
The preceding slow-flow analysis can be ex21
tended to higher shear rates, g . t d0
. For the
sake of simplicity, the analysis will be restricted
to shear rates less than the reciprocal longest
Rouse relaxation time of the N-mer (g , t21
Rouse). It
is further assumed that complete retraction of the
N-mer contour length is achieved in a time tRouse
following sudden imposition of strain. Though the
second restriction is not expected to hold strictly,9
it allows us to argue that the curvilinear length of
the N-mer chain is independent of deformation
21
rate in the range ( t d0
, g , t 21
Rouse). Shear rates
in this range are, however, large enough that the
equilibrium configuration distribution of polymer
segment vectors is significantly perturbed by
shearing. Thus, to estimate the viscous drag acting on the N-mer molecule at a shear rate g , we

L x 5 L Cl x < L C

1 1 gm

1 1 ~1 1 g m! 2

L y 5 L Cl y < L C

1 1 ~1 1 g m! 2

The mean center of mass velocity of the polymer molecule is then V C 5 L y g . Since V T /L C
5 V C /L x the curvilinear, tube velocity is V T
5 L C g (L y /L x ). The convective constraint release
(CCR) renewal time introduced by Ianniruberto
and Marrucci8 is then t CCR 5 L C /V T 5 L x /(L y g ).
21
Thus, at low shear rates ( g ! t d0
), where L x
' L y , the CCR process yields a tube renewal time
tCCR 5 g 21, which is eclipsed by the faster renewal process due to simple reptation. At high
shear rates where CCR is anticipated to dominate, we find tCCR 5 g m / g . Thus, in the absence of
some other process for accelerating reptation at
higher shear rates, tube renewal by convective
constraint release should therefore occur on a
characteristic timescale ( t CCR ' t d0 ), at best.
In steady shear flow at rates g , t21
Rouse, retraction of the primitive N-mer molecule to maintain
its equilibrium contour length reduces the number of entanglements the N-mer makes with surrounding molecules (see Fig. 3). In the simplest
case, the surrounding temporary network could
be assumed to deform affinely and the N-mer to

226

MHETAR AND ARCHER

constraint release) should lower the level of molecular orientation possible in a steady-shear flow
compared to the D-E prediction.
To quantify this, we first estimate the viscous
drag on a N-mer molecule at steady-state,

Figure 3. (a) N-mer molecule trapped in a temporary


network of chemically identical molecules at equilibrium. Dots represent network constraints. (b) Retracted N-mer molecule in a deformed network. The
number of network constraints exerted on the N-mer is
seen to be lower than at equilibrium.

recover its equilibrium contour length in a time


tRouse. The steady-state entanglement density can
therefore be computed from, N/N e ( g m ) 5 N/
N e0 =2/(1 1 (1 1 g m ) 2 ). For well-entangled
polymer liquids, tRouse is much shorter that any
other timescale of rheological significance. Entanglement loss should therefore begin essentially
upon start-up of steady shearing and persist at
steady-state, where its fundamental effect is to
reduce frictional drag between entangled macromolecules and their neighbors.
Material properties, such as network modulus,
viscosity, and orientation relaxation time, that
depend either directly on entanglement density,
or indirectly through their dependence on the molecular drag coefficient, will clearly be altered by
entanglement loss. Mead et al., for example, have
recently shown that the effect of entanglement
loss on polymer orientation relaxation can be captured phenomenologically in the DoiEdwards
constitutive model if a strain-dependent rate of
entanglement loss k is introduced.12 Mhetar and
Archer have proposed that the combination of
partial chain retraction and entanglement loss in
flow yields a strain-dependent stretch relaxation
time, t s ' t d0 /^uE z uu 1/ 2 &, in entangled polymer
21
liquids.8 At shear rates t d0
, g , t 21
Rouse, both
approaches remove the downturn in shear stress
predicted by the DoiEdwards constitutive equation. Viewed in terms of the required balance
between viscous drag and elastic restoring force
at steady-state, extra orientation relaxation produced by lost entanglements can be thought of as
a means for lowering the elastic restoring force
per molecule, so that at steady-state this force
remains in balance with the lower drag force
caused by entanglement loss. Thus, either consequence of entanglement loss (reduction of drag
force, or acceleration of molecular relaxation by

f drag~ g ! 5 z NV T <
<

S D

Ly
N2
z mL cg
Ne
Lx

2
1 1 ~1 1 g m! 2

1/2

g m 3 kT
L,
1 1 g m Nb 2 c

(7)

where in the last equality we have used g m / t d0


5 g . Equation 7 indicates that at high shear
rates, or equivalently at large g m , the viscous
drag on the primitive chain, f drag, decreases with
increasing shear rate roughly as g 21. At shear
rates in the range of interest, the elastic restoring
force on the primitive N-mer chain is here taken
to be determined by its orientation only. The force
is directed along the chain contour and can be
estimated from,
f elastice T 5

3 kT
L $@l ~ g ! 2 l x~0!#e x
Nb 2 c x m
1 @l y~ g m! 2 l y~0!#e y%.

(8)

At large g m , (l y /l x ) 3 0 and the component of the


restoring force parallel to the direction of shear is
more important,
f elastic <

3 kT
L @l ~ g ! 2 l x~0!#
Nb 2 c x m
5

3 kT
L
Nb 2 c

1 1 gm

1 1 ~1 1 g m!

(9)

Equation 9 indicates that the elastic restoring


force increases in proportion to the accumulated
molecular shear strain at low strains, but eventually saturates at large shear strain.
In a steady shear flow, this restoring force
must be balanced by the drag force component
parallel to the direction of shear if the molecular
orientation causing the restoring force is to be
maintained at steady-state. Herein we believe lies
the source of the aforementioned step-shear/
steady-shear contradiction in the D-E theory. In
both deformation types, the theory contends that
(l y /l x ) 3 0 monotonically with increasing strain
or strain rate. This prediction is reasonable for
step shear because in a suddenly imposed defor-

POLYMER DYNAMICS IN STEADY SHEARING FLOWS

mation the instantaneous orientation of a primitive N-mer molecule is entirely determined by its
affine deformation. The macroscopic stress developed by an ensemble of such molecules is therefore unsurprisingly elastic. The D-E prediction is
unrealistic for steady shear flow, however, because there the viscous drag required to maintain
the high levels of molecular orientation predicted
at steady-state decrease with increasing shear
strain [see Fig. 4(a)].
Equations 7 and 9 in fact indicate that in
21
steady-shearing flows at rates t d0
, g , t 21
Rouse,
shear-induced entanglement loss ultimately
causes the frictional drag force on polymer molecules to become insufficient to sustain arbitrarily
low orientation angles at steady-state. Thus, if
polymer chains are long enough to entangle, the
steady-state molecular orientation in a steady
shear flow is predicted to approach and become
locked to a limiting/plateau angle x c '
atan(1/(1 1 g m,c )), regardless of the strain rate
and molecular details of the polymer. At higher
shear rates g . t21
Rouse, retraction becomes impossible and shear-induced entanglement loss
ceases. The molecular orientation angle should
then resume its descent to 0 at shear rates exceeding t21
Rouse. A similar result is predicted by the
theory of Mead et al. (see Figure 14 of ref. 12),
except there the entanglement densities considered are too small to allow a clear distinction
between the plateau orientation angle regime and
the high shear regime, g . t21
Rouse, where molecules eventually align parallel to the direction of
flow.
Strain-dependent drag, f drag and restoring
forces f el , estimated from eqs 7 and 9 are presented in Figure 3(a). A critical molecular strain
g m,c ' 2.75 (strain corresponding to minimum
allowed steady-state orientation angle), can be
identified from the figure. The corresponding limiting orientation angle is x c ' 16.6. If the assumption that molecules completely recover their
equilibrium contour length after tRouse is replaced
by L c u t5 t Rouse 5 L c ^uE z uu& 1/ 2 which supports only
partial contour length recovery,8 a much smaller
number of entanglements are lost under shear,
N/N e ( g m ) ' N/N e0 (2/1 1 (1 1 g m ) 2 ) 1/4 . In that
case, comparison of elastic and drag force components parallel to the direction of shear [Fig. 4(a)],
yields g m,c ' 13.2 and x c ' 4.5.
Before concluding this section, it is useful to
consider the critical strain and limiting orientation angle estimates provided by variants of eqs 7
and 9, modified to produce a smooth crossover

227

Figure 4. (a) Components of the viscous drag force


f drag and elastic restoring force ( f el ) parallel to the
direction of shear. f drag is computed under the assumption that molecules recover their equilibrium contour
length in a time tRouse following imposition of a step
strain. f drag,PSE is computed under the assumption that
only partial recovery of contour length occurs in tRouse,
21
L c u t5 t Rouse 5 L c ^uE z uu& 1/ 2 . At low shear rates ( g , t d0
)
both f drag and f el increase linearly with molecular
21
strain g m 5 g t d0 . However, at high shear rates ( t d0
21
, g , t Rouse), the viscous drag force decreases with
molecular strain f drag ; g 21
m , while the elastic force
becomes independent of molecular f el ; g 0m . (b) Components of the viscous drag force f drag and elastic restoring force ( f el ) parallel to the direction of shear
predicted by eqs 10 and 11. Except for differences in the
limiting orientation angles predicted, f drag and f el show
the same qualitative behavior at low and high shear
rates as those predicted by eqs 7 and 9.

between f drag and f elastic estimates at low and high


flow rates. Again assuming complete retraction,
the amended drag and elastic restoring force estimates are,

228

MHETAR AND ARCHER

Table I. Structural and Linear Viscoelastic Data for Poly(styrene)/Diethyl Phthalate Solutions
Used in the Study
Polymer
Sample #

#
M
w
3 10 26

PI

h0 [Pa z s]

J 0e [Pa21]

t d0 [s]

l d0 [s]

G N [Pa]

PS-P93M
PS-1P8M
PS-2P9M
PS-3P8M

0.935
1.810
2.890
3.840

1.01
1.03
1.09
1.04

7.05 3 102
7.46 3 103
5.41 3 104
1.38 3 105

6.8 3 1024
7.6 3 1024
7.9 3 1024
7.7 3 1024

0.5
5.7
42.8
106.3

2.5
11.7
99.5
198.4

4.8 3 103
4.6 3 103
4.6 3 103
4.7 3 103

1/2

gm
1 1 Qgm

fdrag~gm ! 5

3 kT
2
2 Lc
Nb
1 1 ~1 1 gm !2

f elastice T 5

3 kT
L $@l ~ g ! 2 l x~0!#e x
Q 2Nb 2 c x m
1 @l y~ g m! 2 l y~0!#e y%,

(10)

(11)

where Q ' 1/2. f drag,x and f el, x determined from


these equations are compared in Figure 4(b).
Again a critical molecular strain g m,c is observed,
beyond which further molecular orientation cannot be sustained by viscous drag at steady-state.
The corresponding limiting orientation angle x c
5 atan(1/(1 1 Q g m,c )) is estimated to be x c ' 33
(assuming complete retraction) and x c ' 15.6
(assuming only partial retraction). Thus, the refined estimates of f el and f drag support the prediction that a nonzero, limiting steady-state orientation angle exists in steady shear flow, but suggest
that this angle could be reached at even lower
shear rates than predicted by eqs 7 and 9.

EXPERIMENTAL
Materials and Methods
Several high molecular weight, narrow molar
mass distribution polystyrenes (PS) were purchased from Tosoh Corp., Japan and from Aldrich
Chemicals. Weight-averaged molecular weights,
polydispersity indices, and abbreviated names
are provided in Table I for all polymers used in
the study. Solutions containing 18%, by weight,
poly(styrene) in diethyl phthalate (Aldrich) were
made up using the following procedure. First, the
required amounts of polymer and diethyl phthalate (DEP) were weighed and placed into a labelled glass container, the weight of the container
(including its contents) was recorded. A large excess of dichloromethane [methylene chloride] (Al-

drich) cosolvent was added to the PS/DEP mixture to accelerate dissolution. The container containing the PS/DEP/methylene chloride mixture
was then sealed and agitated for two days using a
room-temperature (24.5C) laboratory stirrer/
shaker device. For all polymers used in the study
this time was sufficient to produce transparent,
well-mixed solutions. Subsequent exposure of the
PS/DEP/methylene chloride mixture to air, at
room temperature, resulted in complete evaporation of the cosolvent within a week.
Linear viscoelastic properties of the PS/DEP
solutions were studied using a Paar Physica
UDS-200 mechanical rheometer. Frequency-dependent polymer behavior in low-amplitude oscillatory shear and polymer relaxation following
small amplitude step strain measurements g
5 0.2 were investigated using the rheometer. The
rheometer was equipped with 25-mm diameter)
6-mm thick, bead-blasted, stainless steel cone
(cone angle, 4) and plate fixtures. Using a Stylus
profilometer (Surfanalyzer 5000), the root-mean
square surface roughness of the bead-blasted fixtures was determined to be 3.1 mm. In highly
entangled polymers, clean surfaces with micrometer-size roughness have been shown to reduce
apparent slip at low shear stresses and to delay
the transition from weak (rheometrically insignificant), to strong slip at high stresses.13
Frequency-dependent storage G9( v ) and loss
G0( v ) moduli for the PS/DEP solutions used in
the study are presented in Figure 5(a, b). Complex shear viscosities uh*u(v) deduced from the
dynamic moduli are presented in Figure 5(c). Except for sample PS-3P8M, all measurements were
performed at a fixed temperature of 25.5C; a
recirculating oil bath facilitated temperature control. Data presented for PS-3P8M was obtained
by superimposing G9( v ) 2 G0( v ) data at two
temperatures (25.5 and 65.5C) to 25.5C by horizontal shifting alone. It is noteworthy that the

POLYMER DYNAMICS IN STEADY SHEARING FLOWS

Figure 5. (a c) Dynamic storage moduli G9( v ), loss


moduli G0( v ), and complex viscosities uh*u(v) for the 18
wt % poly(styrene)/diethyl phthalate solutions used in
the study. Low-amplitude oscillatory shear flow measurements on the lower molecular weight poly(styrene)
(PSP93M, PS1P8M, and PS2P9M) were performed at
25.5C using bead-blasted, stainless steel cone-andplate rheometer fixtures. Data presented for the high

229

G9( v ), and uh*u(v) curves for all solutions converge at high frequencies. This behavior is expected for narrow molecular weight distribution polymers of differing molecular weights,
but with the same entanglement spacing. Material properties deduced from the frequencydependent moduli are presented in Table I. The
following definitions were used in extracting this
2
data: h0 5 uh*u(v)uv30; J 0e 5 h 22
0 (G9( v )/ v u v 30 );
0
t d0 5 h 0 J e ; G N 5 G9( v m ), where v m is the
frequency at which G0( v ) has a local minimum.
Values of G N and J 0e deduced in this way are
related as follows, G N ' 3.5 6 0.2 ( J 0e ) 21 . Terminal relaxation times l d0 determined from the
long-time slope of semilogarithmic plots of shear
stress versus time following linear step strains
are also included in the table. Except for the lowest molecular weight polymer studied PS-P93M,
relaxation times obtained from step strain measurements are seen to be related to those from
oscillatory shear by, l d0 ' (2.0 6 0.2) t d0 .
Steady shear flow rheological measurements
were also performed on the PS/DEP solutions using the UDS-200 rheometer. These measurements were conducted at 25.5C with the same
bead-blasted, stainless steel cone-and-plate rheometer fixtures employed in the oscillatory shear
experiments. In addition to the possibility of slip
violations at the cone/plate surface, edge fracture
presents a well-known challenge to cone-andplate, steady shear rheology measurements on
highly entangled polymers. In this work, a combination of procedures proved successful in delaying edge fracture. First, prior to each test the cone
and plate were thoroughly cleaned using methylene chloride and acetone, and the cone edge lubricated using a fluoropolymer lubricant. During
steady shear measurements, the cone was submersed to a depth of about 4 mm in a small
puddle of the polymer solution under study. Second, at the highest shear rates studied, experiments were actively monitored and stopped
shortly after steady-state was achieved. This last
procedure was motivated by the observation that
edge fracture requires a shear-rate dependent
critical strain for its onset; interrupting experiments before this strain was achieved proved suc-

est molecular weight polymer, PS-3P8M was obtained


by superimposing G9( v ) 2 G0( v ) data at two temperatures (25.5 and 65.5C) to 25.5C by horizontal
shifting.

230

MHETAR AND ARCHER

cessful in delaying the edge instability.14 Using


this combination of procedures steady shear measurements were possible up to shear rates of 3 s21
without any visible appearance of edge lacerations or transient shear stress decay normally
associated with edge fracture and wall slip.
For each material studied, time-dependent
shear stress, s xy (t) and first normal stress differences N 1 (t) were collected over three decades of
shear rate. In every case, steady-state stresses
were deduced from this information by taking the
average of twenty instantaneous stress values at
long times where s xy (t) and N 1 (t) were both independent of time. The averaging procedure minimizes errors caused by inevitable transient
stress fluctuations about the mean, steady-state
value. It is significant that the steady-state
stresses reported here were determined directly
from the respective time-dependent stress data
measured at constant shear rate, rather from
shear rate sweeps at long times. The former approach facilitates steady-state stress measurements under conditions where transient artifacts
due to slip at the walls, rheometer compliance,
and edge fracture can be safely ruled out. Averaging over several repeat experiments also permitted confidence intervals (error bars on plots) to
be defined, run-to-run repeatability of steadystate stresses is therefore guaranteed within the
defined intervals.

RESULTS AND DISCUSSION


Plots of steady-state shear stress s ss and first
normal stress differences N 1,ss versus dimensionless shear rate are presented in Figure 6(a). The
dimensionless shear rate or Weissenberg number
is defined by, Wi 5 g t d0 . The similarity of the s ss
versus Wi and N 1,ss versus Wi plots for PS/DEP
solutions with varying PS molecular weight is
remarkable, but nevertheless well known.1,2 The
near-perfect overlap of the steady-state data at
Wi , 10 suggests that for the studied materials,
the terminal molecular relaxation time is more
accurately determined from zero shear viscosity
and steady-state creep compliance, than from
longtime stress relaxation data. It is significant
21
that at high rates g . t d0
, the experimental
steady-state shear stress and normal stress differences exhibit a virtual lack of dependence on
shear rate over a fairly broad range of rates. As
discussed in the previous section of the paper,

steady-state shear and normal stress differences


were determined directly from time-dependent
data measured at fixed shear rate. No evidence of
unstable flow or unusual transient stress behavior was observed in the range of experimental
shear rates and times for which data are reported,
indicating that the observed dependencies are
real. Plateaus in steady-state shear stress and
first normal stress differences have also been reported by Bercea et al.15 from cone-and-plate
shearing experiments using solutions of high molecular weight poly(methylmethacrylate) M w 5
23.8 3 10 6 in toluene (N/N e0 5 42 and 125). As
in the present study, these authors observed that
the range of shear rates over which the plateau is
seen in N 1,ss is shorter than the range of rates at
which a plateau in shear stress is observed. Cates
et al. have suggested that plateaus in steadystate shear stress and first normal stress differences are expected in entangled polymers when
the finite lateral size of the tube is taken into
account in stress calculations.16 Specifically,
these authors contend that because of its finite
cross section, slight differences in flow velocity
are experienced by different lateral sections of the
tube. This effect yields a contribution to shear
stress that increases linearly with rate and balances the down-turn in shear stress that would be
21
predicted at g . t d0
, if the only source of stress
was molecular orientation.
To investigate the steady shear behavior of
entangled polymers further, it is advantageous to
recast the two sets of data presented in Figure
6(a) in terms of steady-state orientation angles
x ss [ 0.5 3 atan(2 s ss /N 1,ss ) [see Fig. 6(b)]. For
all solutions investigated, a defined dependence
on Wi is clearly apparent from the data at Wi
, 10, indicating that the steady-state orientation
angle is set by g and t d0 only, in this range of
shear rates. At higher Wi, x ss is observed to become nearly independent of Wi over a range of
values that becomes larger with increasing polymer molecular weight. A limiting orientation angle x c may be deduced from the experimental
results. The limiting angle is found to lie in the
range of 6 and 9 for a nearly four-fold increase
in polymer molecular weight, confirming that x c
is indeed insensitive to polymer molecular
weight.
Steady-state orientation angles for entangled
poly(butadiene)/Flexon391 solutions, high molecular weight poly(methylmethacrylate)/toluene solutions, a poly(styrene)/tricrecyl phosphate solution, and PS-1P8M are provided in Figure 6(c) for

POLYMER DYNAMICS IN STEADY SHEARING FLOWS

231

Figure 6. (a) Steady-state shear stress s ss (open symbols) and first normal stress
difference N 1,ss (filled symbols) versus dimensionless shear rate Wi 5 g t d0 . Error bars
define confidence intervals in which run-to-run repeatability of the steady-state
stresses are guaranteed. For all the polymers studied, both sets of data ( s ss and N 1,ss )
are observed to overlap over a rather wide range of dimensionless shear rates, indicating a dependence of steady-state polymer material properties on Wi only. (b) Steadystate orientation angle x 5 0.5 atan((2 s ss )/N 1,ss ) versus dimensionless shear rate for
various 18 wt % PS/DEP solutions. (c) Steady-state orientation angle x 5 0.5
atan((2 s ss )/N 1,ss ) versus dimensionless shear rate for various poly(butadiene)/
Flexon391 (PBA-PBD) [from Menezes and Graessley18]; ultra-high molecular weight
poly(methylmethacrylate)/toluene (HPMMA1 and HPMMA2) [from Bercea et al.15];
poly(styrene)/tricrecyl phosphate (PS/TCP) [Hua et al.17]; and poly(styrene)/diethylphthalate (PS-1P8M) solutions. (d) Plot of orientation angle x versus shear strain g t
at various Wi during start-up of steady shearing. The data are for an 18 wt % solution
of PS-2P9M in DEP.

variable Wi. The poly(butadiene) orientation angles were recovered from time-dependent shear
and normal stress measurements by Menezes and
Graessley obtained using a stiffened Weissenberg
Rheogoniometer.17 Narrow distribution polybutadienes M w /M n , 1.05) of variable molar mass
(PBA-Mw 5 200,000, N/Ne ' 7; PBB-Mw 5 350,000,

N/N e ' 12; PBC-M w 5 517,000, N/N e ' 18 and


PBD-M w 5 813,000, N/N e ' 29), but fixed polymer volume fraction (f 5 0.0755) were used in the
study. It is noteworthy that while s ss (Wi) results
for the two most highly entangled poly(butadiene)/Flexon solutions (PBC and PBD) manifest
a regime of very weak to no shear stress depen-

232

MHETAR AND ARCHER

dence on Wi at steady-state, the corresponding


plateau in N 1,ss is not observed in these solutions.
Orientation angle data for the high molecular
weight poly(methylmethacrylate)/toluene samples (HPMMA1, f ' 0.023 and HPMMA2, f
' 0.07) studied by Bercea et al., was determined
from s ss vs. g and N 1,ss vs. g results reported by
the authors. Data for the PS/TCP solution was
obtained from orientation angle results reported
by Hua et al.18 for solutions of a high molecular
weight poly(styrene), M w 5 1.9 3 10 6 , in tricrecyl phosphate ( f 5 0.13; N/N e ' 13), with the
exception that a terminal time t d0 ' 7.5 s (exactly one-half the value reported), was used to
compute Wi. Orientation angle data for PS-1P8M
were obtained from the present study.
Several features of the orientation angle results in Figure 6(c) are noteworthy. First, for Wi
, 10 the Wi-dependent orientation angles for the
PB/Flexon391, PS/tricrecyl phosphate solutions,
are qualitatively similar to those found for the
PS/DEP solutions investigated in the present
study. Furthermore, as was the case for the PS/
DEP solutions, orientation angles for the PB and
PS/TCP solutions are observed to superimpose
well when plotted as a function of Wi; confirming
that in this shear rate range g and t d0 are the
dominant dynamic parameters that set the level
of molecular orientation at steady-state. Second,
for Wi . 10, the PB/Flexon391 and PS/TCP solutions both manifest a regime of Wi values
where x ss shows a rather weak dependence on
Wi. As pointed out already, a similar flattening of
the Wi dependence of x is observed for PS/DEP in
the same Wi range. Remarkably, the limiting orientation angle projected for the poly(butadiene)/
Flexon 391 and poly(styrene)/diethylphthalate solutions, x c ' 7 is in the same range as the
plateau orientation angles observed for the PS/
DEP solutions.
These results are, however, quite different
from orientation angles observed for the two high
molecular weight PMMA/toluene solutions. In
these solutions two orientation angle plateau regions are apparent. Beginning at Wi value
around 2, a first plateau between 35 and 30
is observed for HPMMA1 and HPMMA2. In
HPMMA2, this first plateau region is followed by
a high-Wi regime where the orientation angle
decreases with Wi in a manner strikingly similar
to the initial decrease observed for the PS/DEP,
PS/TCP, and PB/Flexon solutions. The finding of
two orientation angle plateaus, rather than one,
in the PMMA/toluene solutions could be rational-

ized using the approach presented in the theory


section of the paper, if the transition from complete contour length retraction to partial retraction is taken into account.8 Flow birefringence
experiments using polymer solutions of comparable entanglement density to the materials studied
by Bercea et al. could help resolve the source of
the orientation angle differences between PMMA/
toluene and the other polymer solutions.
Additional support for the proposal is provided
by transient orientation angle measurements following start-up of steady shearing [Figure 6(d)].
Specifically, at high shear rates, the orientation
angle is observed to undergo an undershoot before
ultimately approaching x c . This behavior was
seen in all the PS/DEP solutions investigated in
this study, but is not predicted by the DoiEdwards theory. Time-dependent birefringence
measurements during start-up of circular Couette
shear of a 3% solution of PS-8P42M in tricrecyl
phosphate also reveal similar transient undershoots of orientation angle.19 A limiting steadystate orientation angle between 10 and 14 is
suggested by the birefringence data. An undershoot in orientation angle during start-up of
steady shearing is not predicted by the recent
theory of Mead et al.,12 but is expected if the
required balance between entropic elastic restoring force and viscous drag at steady-state is enforced. Specifically, at Wi $ 1, the network and
N-mer strands trapped in it initially orient affinely. The elastic restoring force created by the
high levels of affine molecular orientation is initially larger than the viscous drag force exerted
on molecules by their neighbors. The force imbalance causes polymer segments to relax the excess
orientation until the elastic and viscous drag
forces balance at steady-state.

CONCLUSION
In this article a simple, new proposal has been
presented for polymer dynamics in fast steady
shearing flows. Specifically, we contend that in
steady shear flow a balance between the viscous
drag on polymer segments and the entropic elastic restoring force due to primitive chain orientation and stretch must be met at steady-state.
Thus, the level of molecular orientation achievable at steady-state is determined both by the
direct effect of flow on molecular orientation and
by the indirect effect of flow-induced orientation

POLYMER DYNAMICS IN STEADY SHEARING FLOWS

and chain stretch on the drag force exerted on


molecules by their neighbors. For entangled polymers, the second effect is dominated by entanglement loss due to contour length retraction. Its
consequence is quantified by a new expression for
entanglement friction between molecules under
action of a directed external force.9
In a simplified analysis enforcing the proposed
balance of forces, we find that at high shear rates
21
t d0
, g , t 21
Rouse in a steady shear flow the
steady-state orientation of polymer molecules approach a limiting value x c ' atan(1/(1 1 g m,c ))
beyond which flow becomes incapable of producing further molecular orientation. If molecules
are assumed to retain their equilibrium contour
length at shear rates g , t21
Rouse, the limiting orientation angle is found to be x c ' 33. This angle
changes to x c ' 15.6 when shear-induced
changes in polymer contour length at rates below
t21
Rouse are taken into account. The latter prediction is higher than limiting orientation angles
determined from steady shear rheological measurements using poly(styrene)/diethyl phthalate,
poly(butadiene)/Flexon391, poly(methylmethacrylate)/toluene, and poly(styrene)/tricresylphosphate
solutions. These solutions show limiting orientation angles in the range of 6 to 9 over a wide
range of polymer molecular weight and monomer
chemistries. Orientation angle undershoots deduced from flow birefringence measurements during start-up of steady shearing are confirmed by
our PS/DEP experiments. Both sets of results are
consistent with relaxation of extra molecular orientation induced by a transient imbalance of viscous drag, and elastic restoring force acting on
polymer molecules during start-up of steady
shearing.
The authors are grateful to the National Science Foundation Grants number DMR9816105 and CTS9624254,

233

and to the Texas Higher Education Coordinating


Board, Advanced Technology Program for supporting
this study.

REFERENCES AND NOTES


1. Doi, M.; Edwards, S. F. The Theory of Polymer
Dynamics; Oxford Science Publications: New York,
1986.
2. Larson, R. G. Constitutive Equations for Polymer
Melts and Solutions, Butterworths: London, 1988.
3. Osaki, K.; Nishizawa, K.; Kurata, M. Macromolecules 1982, 15, 1068.
4. Osaki, K. Rheol Acta 1993, 32, 429.
5. Marrucci, G. J Non-Newtonian Fluid Mech 1996,
62, 279.
6. Ianniruberto, G.; Marrucci, G. J Non-Newtonian
Fluid Mech 1996, 65, 241.
7. Archer, L. A.; Mhetar, V. R. Rheol Acta 1998, 37,
170.
8. Mhetar, V. R.; Archer, L. A. J Non-Newtonian
Fluid Mech 1999, 81, 71.
9. Mhetar, V. R.; Archer, L. A. Macromolecules 1998,
31, 6639.
10. Bueche, F. J Chem Phys 1952, 20, 1959.
11. Doi, M. J Polym Sci Polym Phys Ed 1983, 21, 667.
12. Mead, D. W.; Larson, R. G.; Doi, M. Macromolecules 1998, 31, 7895.
13. Mhetar, V. R.; Archer, L. A. Macromolecules 1998,
31, 8617.
14. Crawley, R. L.; Graessley, W. W. J Rheol 1977, 21,
19.
15. Bercea, M.; Peiti, C.; Simionescu, B.; Navard, P.
Macromolecules 1993, 26, 7075.
16. Cates, M. E.; McLeish, T. C. B.; Marrucci, G. Europhys Lett 1993, 21, 451.
17. Menezes, E. V.; Graessley, W. W. J Polym Sci
Polym Phys 1982, 20, 1817.
18. Hua, C. C.; Schieber, J. D.; Venerus, D. C. J Rheol
1999, 43, 701.
19. Oberhauser, J. P.; Leal, G. L.; Mead, D. W. J Polym
Sci Polym Phys Ed 1998, 35, 265.

Вам также может понравиться