Vorticity banding in rodlike virus suspensions
Kyongok Kang,1M. P. Lettinga,1Z. Dogic,2and Jan K. G. Dhont1
1Institute für Festkörper Forschung (IFF), Weiche Materie, Forschungszentrum Jülich, D52425 Jülich, Germany
2Rowland Institute at Harvard, 100 Edwin H. Land Boulevard, Cambridge, Massachusetts 02142, USA
?Received 16 February 2006; published 28 August 2006?
Vorticity banding under steady shear flow is observed in a suspension of semiflexible colloidal rods ?fd virus
particles? within a part of the paranematic-nematic biphasic region. Banding occurs uniformly throughout the
cell gap within a shear-rate interval ?? ˙−,? ˙+?, which depends on the fd concentration. For shear rates below the
lower-border shear rate ? ˙−only shear elongation of inhomogeneities, which are formed due to paranematic-
nematic phase separation, is observed. Within a small region just above the upper-border shear rate ? ˙+, banding
occurs heterogeneously. An essential difference in the kinetics of vorticity banding is observed, depending on
the morphology of inhomogeneities formed during the initial stages of the paranematic-nematic phase separa-
tion. Particle tracking and polarization experiments indicate that the vorticity bands are in a weak rolling flow,
superimposed on the applied shear flow. We propose a mechanism for the origin of the banding instability and
the transient stability of the banded states. This mechanism is related to the normal stresses generated by
inhomogeneities formed due to the underlying paranematic-nematic phase transition.
DOI: 10.1103/PhysRevE.74.026307PACS number?s?: 47.20.Ft, 47.55.?t, 47.15.Fe, 82.70.?y
At equilibrium, complex fluids undergo a variety of order-
ing transitions that are driven by purely thermodynamic
forces ?see, for example, Refs. ?1,2??. External fields can
greatly affect the phase behavior of these systems. For non-
conservative fields such as a shear flow no thermodynamic
analog is yet known, where the equality of scalar quantities
in coexisting phases ?the analogs to pressure and chemical
potential? suffices to predict the location of phase transition
lines as a function of the strength of the driving force. The
search for such a formalism remains at the forefront of re-
search in soft condensed matter and nonequilibrium physics
?3,4?. In addition to nonequilibrium phases, systems driven
by a shear flow frequently exhibit hydrodynamically driven
pattern formation which have no equilibrium analogs. Some
representative examples include Taylor-Couette flow at high
shear rates and shear banding at much lower shear rates
A particularly important example of a complex fluid
whose phase behavior is greatly affected by a shear flow is a
system of rodlike colloids. At equilibrium rods undergo a
thermodynamically driven phase transition from an isotropic
to a liquid crystalline nematic phase ?2?. Shear flow strongly
aligns rods and therefore affects the location of the isotropic
?paranematic-? nematic phase transition ?8–10?. An isotropic
state under shear flow is referred to as a “paranematic” state
to indicate that flow partially aligns otherwise isotropic rods.
The paranematic-nematic binodal is defined as the locus of
points that separates the one-phase region from the region
where phase coexistence occurs. The spinodal under shear
flow is defined as the locus of points where the system be-
comes unstable against infinitesimally small perturbations of
the orientational order parameter. The spinodal and binodal
referred to here and hereafter in the present paper are con-
nected to the paranematic-nematic phase transition, that is,
the shear-affected isotropic-nematic phase transition that also
occurs in the absence of flow. Whenever a spinodal or bin-
odal is mentioned in this paper, it refers to the shear-affected
isotropic-nematic phase transition and not to the banding
transition, except when explicitly mentioned otherwise.
Besides shifts of the equilibrium binodals and spinodals,
shear flow can also lead to the formation of banded structures
in a number of complex fluids including rodlike colloids
studied here. In general two types of banding transitions can
be distinguished: gradient banding and vorticity banding. In
the case of gradient banding, coexisting regions ?“bands”?
extend along the gradient direction. The shear rate is essen-
tially constant within these bands ?see Refs. ?11–16??. The
gradient banding transition is relatively well understood and
it occurs when the shear stress decreases with increasing
shear rate ?17–21?. In case of vorticity banding, regions of
different internal structure are alternately stacked along the
vorticity direction ?9,10,22–24?. The origin and the mecha-
nism of the vorticity banding transition are not yet known. It
was suggested in Ref. ?17? that vorticity banding can occur
when the shear stress is a multivalued function of the shear
rate. As far as we know, there are no systematic experimental
studies of vorticity banding concerning shear-band formation
kinetics, the characteristic features of vorticity-banded struc-
tures, the internal structure of individual bands, and the pos-
sible connection to the underlying nonequilibrium phase be-
In the present paper we systematically study pattern for-
mation ?vorticity-banding transition? under steady shear flow
conditions of a suspension of rodlike colloids in shear flow,
where bands of different orientational order are stacked
along the vorticity direction. We quantify the relationship
between pattern formation and the underlying nonequilib-
rium, shear-affected paranematic-nematic phase transition.
Vorticity banding is observed within a part of the biphasic
isotropic-nematic region, under both controlled shear-rate
and shear-stress conditions. We propose a possible mecha-
nism that describes the vorticity-banding instability and also
explains the temporary stability of the quasistationary
banded states. The proposed mechanism implies that there is
no genuine stationary vorticity-banded state, and that its tran-
PHYSICAL REVIEW E 74, 026307 ?2006?
©2006 The American Physical Society026307-1
sient stability relies on the presence of inhomogeneities
formed due to paranematic-nematic phase separation. The
lifetime of the vorticity-banded state is thus set by the life-
time of the inhomogeneities. As soon as the inhomogeneities
disappear, for example due to sedimentation, the vorticity-
banded state also disappears. This process takes a few days
as compared to the formation of bands within an hour. Since
paranematic-nematic region, the inhomogeneities that form
after a shear-rate quench due to paranematic-nematic phase
separation seem to play a crucial role in rendering the system
unstable against vorticity banding. The kinetics of the band-
ing transition is found to be fundamentally different depend-
ing on whether these inhomogeneities are isolated or form an
interconnected structure. Furthermore, above the region
where bands are formed homogeneously, there is a small
shear-rate range where heterogenous banding is observed.
As an experimental model system we use the monodis-
perse rodlike virus fd. Extensive experiments have shown
that the equilibrium isotropic-nematic phase transition of fd
virus is quantitatively described by Onsager’s theory ?25?
when it is extended to take into account fd flexibility and its
surface charge ?26–29?. Besides fd, numerous other systems
of rodlike particles exhibit an isotropic-nematic phase tran-
sition, with important examples including DNA?30?, tobacco
mosaic virus ?31? and synthetic Boehmite rods ?32?. The
phase behavior of fd has been studied in the presence of an
external magnetic ?conservative? field in Ref. ?33?. The in-
duced shift of isotropic-nematic binodals and spinodals for
such a conservative external field can be defined thermody-
namically, simply by adding the corresponding potential en-
ergy to the Hamiltonian. The analogous procedure is not al-
lowed for the shear flow due to its nonconservative nature.
Besides previous experiments on fd ?9,10?, the only other
experimental study of the paranematic-nematic phase transi-
tion of colloidal rods in shear flow, that we are aware of, is in
dispersions of hydroxypropylcellulose ?15?. However, there
is significant work done on shear banding and phase transi-
tions of related systems such as wormlike micelles
?11,12,23,34? and thermotropic liquid crystals ?13,14?. A
complication of wormlike micelles systems, when compared
to suspensions of rods, is that the worm-length distribution
and the scission and recombination kinetics depend on the
The main body of this paper is organized as follows. In
Sec. II, details of the fd virus suspensions, the experimental
setup, and the data analysis are given. The topology of the
shear-induced nonequilibrium phase diagram is given in Sec.
III. The kinetics of band formation is described in Sec. IV. In
Sec. V we describe experiments which indicate that the
bands are in rolling flow. Finally, we propose a mechanism
for the vorticity-banding instability in Sec. VI. This mecha-
nism is reminiscent of the well-known elastic instability of
polymers ?35–39?, where nonuniform elastic deformation of
the polymers leads to a rolling flow. In the present case,
elastic deformations of inhomogeneities formed due to
paranematic-nematic phase separation lead to the vorticity-
banding transition and the associated rolling flow.
onlyinside the biphasic
II. EXPERIMENTAL DETAILS
In this section we discus the colloidal system and the
experimental setup together with the data analysis.
A. The colloidal system
We use monodisperse rodlike fd viruses which are a good
model system for studies of liquid crystalline phase behavior
?26–29?. The bacteriophage fd is a semiflexible filamentous
molecule with a contour length L=880 nm and a diameter
D=6.6 nm. The persistence length is P=2200 nm, which is
more than twice its contour length. Fd virus is thus relatively
stiff. The molecular weight of native fd is 1.64?107g/mol.
Bacteriophage fd was grown and purified following standard
biological protocols, using the Xl1-blue strain of Escherichia
coli as the host bacteria ?40?. The standard yield was ap-
proximatively 15 mg of fd per liter of infected bacterial cul-
ture. The virus particles were purified by repeated centrifu-
gation ?105g for 5–6 h?, and redispersed in high-ionic-
strength buffer to screen electrostatic interactions ?20 mM
tris-HCl, pH 8.15, 100 mM NaCl?. Dextran ?507 kD, radius
of gyration 16 nm, Sigma-Aldrich? was mixed with the fd
dispersion in order to widen the biphasic region and enhance
the phase-separation kinetics ?29?.
Two different dispersions are used in the experiments
with two different dextran concentrations. To study the non-
equilibrium kinetics and phase diagram we used a low dex-
tran concentration ?10.6 mg/ml dextran, 21.7 mg/ml fd?. To
study the internal structure of bands we used a high dextran
concentration ?14.5 mg/ml dextran, 21.7 mg/ml fd?. Sus-
pensions on which experiments were performed have been
prepared as follows. The homogeneous mixtures were al-
lowed to phase separate for a few hours, after which full
phase separation was achieved by gentle centrifugation ?104g
overnight?. The binodal concentrations with added low con-
centration of dextran are determined to be 17.5±0.5 and
29.0±1.5 mg/ml. The width of the phase-coexistence region
is considerably wider when compared to a pure fd suspen-
sion, where binodal points are at 21 and 23 mg/ml, respec-
tively. A volume Visofrom the isotropic phase is mixed with
a volume Vnemof the coexisting nematic phase. The concen-
tration of fd virus particles in such a mixture is expressed
in terms of the quantity ?nem?Vnem/?Vnem+Viso?, which
varies from0 forthelower
?17.5±0.5 mg/ml? to 1 for the upper binodal concentration
?29.0±1.5 mg/ml? in the absence of flow. Homogenized
mixtures within the biphasic region at various concentrations
?nemare used for the vorticity-banding experiments. Such
mixtures are prepared to be at constant dextran chemical po-
tential, independent of the fd concentration. The samples
with higher dextran concentration form thicker and more
regular vorticity bands, and are thus more suitable to study
the internal structure of the bands.
The variable ?nemdefined above is not introduced here as
an “order parameter,” but rather as a convenient measure for
the concentration of a sample relative to the two binodal
concentrations. In all our experiments the actual fd concen-
tration is always between the two binodal concentrations, so
that ?nemalways lies between 0 and 1.
KANG et al.
PHYSICAL REVIEW E 74, 026307 ?2006?
B. Experimental setup and data analysis
Transparent couette shear cells are used with gap widths
ranging from 1.0 to 2.0 mm. The rotating inner cylinder has
a radius of 24 mm while the outer cylinder is fixed. The
experiments are performed under controlled-rate conditions.
For one concentration, the determination of the border shear
rates is repeated under controlled-stress conditions, using a
Bohlin rheometer ?CVO?. The optical couette cell is placed
between a polarizer and an analyzer and is illuminated from
one side with a white light source, as sketched in Fig. 1. The
polarizer P is oriented along the flow direction. The optimal
contrast between vorticity bands is obtained when the polar-
izer and analyzer A are not exactly crossed. Typically, the
angle between the analyzer and the flow direction is 80°. The
reason for this is related to the weak rolling flow within the
bands and is discussed in Sec. V. The transmitted light inten-
sity is monitored with a 12-bit charge-coupled device CCD
camera ?RS Princeton Instruments? equipped with a telecen-
tric lens ?Computar 5.5 mm?. One pixel corresponds to
8.8 ?m in real space, which sets the spatial resolution in the
vorticity direction of our setup. The depth of the focus is
about 1 mm. Therefore, an image is smeared along the gra-
dient direction over a large part of the gap. Since the orien-
tational order differs in the two types of vorticity bands, they
appear in transmission as alternating bright and dark stripes,
stacked along the vorticity direction ?see the inset in Fig. 1?.
For lower dextran concentrations, the extent of a single
dark or bright band is between 50 and 120 ?m. A single
image at each time covers a region that includes about 50
vorticity bands divided over 582 pixels. The intensity at a
particular height is the average over 80–100 adjacent pixels
along the flow direction. Typically, ten such cuts from a
single image are analyzed as described below and averaged.
To probe the evolution of banded structures right after a
shear-rate quench, optical morphologies are recorded every
10–60 s, depending on the rate of band formation, which is
a function of the cell gap width, fd concentration, and shear
rate. The total recording time varies from 1 to 3 h, which is
the time to reach the quasistationary banded state.
The intensity profiles are analyzed as follows. An inten-
sity profile I?z,t? at a particular time t obtained from images
as described above is first represented by a Fourier series as
I?z,t? = I0?t? +?
??n?t?sin?2?z/L? + ?n?t?cos?2?z/L??,
where I0is the average intensity, L is the total height of the
image, and 0?z?L is the height variable along the vorticity
direction. In order to avoid high-frequency peaks due to
noise being identified as a vorticity band, we averaged three
adjacent intensities, corresponding to 26 ?m. The maximum
number of Fourier modes Nmaxis therefore equal to one-third
of the number of image pixels. A band thus encompasses at
least three pixels. The average number n ¯ of bright and dark
bands is now obtained from
n ¯?t? = ?
where Nminis chosen equal to 3 in order to eliminate spuri-
ous long-wavelength variations in light intensity which are
much longer than a typical bandwidth, while
2?t? + ?n
2?t? + ?n
is the normalized probability for a Fourier mode of order n.
The average width at a given time of a vorticity band is then
H?t? = L/2n ¯?t?.
As mentioned above, the choices for Nmaxand Nminare such
that high-frequency contributions ?due to noise? are elimi-
nated and low-frequency variations ?due to nonuniform illu-
mination? do not lead to erroneous results. We confirmed by
counting the number of bands by hand for a number of ex-
periments that the procedure described above gives the cor-
rect number of bands.
III. THE NONEQUILIBRIUM PHASE DIAGRAM
The paranematic-nematic and vorticity-banding phase
diagram in the shear rate versus fd concentration plane is
presented in Fig. 2 ?similar phase diagrams can be found in
Refs. ?9,10??. For fd virus suspensions, vorticity banding is
observed within a part of the two-phase paranematic-nematic
region which is bounded by the binodal. Dextran is added to
fd rods, which induces depletion attractions. This in turn
leads to a widening of the biphasic region ?29? and an en-
FIG. 1. Schematic of the experimental setup. The optical shear
cell is placed between two polarizers, and spatial-temporal images
of the banded structure are recorded with a CCD camera equipped
with a telecentric lens. Additionally we show an image of a typical
banded structure. The polarizers are not exactly crossed for the
reasons discussed in Sec. V. The unit vector z ˆ indicates the vorticity
VORTICITY BANDING IN RODLIKE VIRUS SUSPENSIONSPHYSICAL REVIEW E 74, 026307 ?2006?
hancement of both phase-separation and vorticity-banding
kinetics. It is still uncertain whether banding occurs in fd
suspensions without any dextran.
We first turn our attention to determining the location of
the paranematic-nematic binodal under flow, which was ac-
complished using time-resolved shear-stress measurements
after a shear-rate quench ?10?. Starting at a high shear rate,
where the homogeneous state is stable, the shear rate is
quenched to a lower value and the shear stress is measured as
a function of time. If the system crosses the binodal, inho-
mogeneities in concentration and orientation order develop
with time, giving rise to an increase of the shear stress. On
quenching from a high to a lower shear rate, isotropic inho-
mogeneities are formed in a nematic background. Since these
isotropic inclusions have a higher viscosity as compared to
the nematic background, their growth is accompanied by an
increase in the shear viscosity. Such time-resolved stress
measurements allow the determination of the paranematic-
nematic binodal in the fd concentration versus shear rate
plane and lead to the upper solid curve in Fig. 2. The binodal
determined in this way marks the concentrations of phase
coexistence after completion of phase separation, since the
homogeneous phases that coexist become metastable or un-
stable on increasing ?for the lower binodal? or decreasing
?for the upper binodal? the fd concentration. Note that this
binodal is not connected to a shear-induced phase transition,
contrary to many wormlike micellar systems. Here, the bin-
odal is merely shear affected, and also occurs in the absence
The above-described method is not suitable for determin-
ing the paranematic-nematic binodal for the case of either
high dextran concentration or high fd concentration. With
increasing dextran concentrations the binodal is located at
increasingly higher shear rates. As a result, the difference in
the orientational order across the binodal is not as pro-
nounced as for samples with lower dextran concentration.
This leads to a decrease of the stress response after a shear-
rate quench and renders the experimental determination of
the binodal difficult. For this reason we have not shown the
binodal in the inset of Fig. 2. At high fd concentration
??nem?0.7? close to the homogeneous nematic phase, states
with periodic tumbling or wagging of rods in shear flow are
observed ?10,41?. This is evidenced by an oscillating shear-
stress response subsequent to a shear-rate quench. For homo-
geneous nematic phases, theory predicts such behavior for
certain rod concentrations and shear rates ?37,42–46?. Due to
the oscillatory response, the location of the binodal as well as
the extent of the vorticity-banding transition cannot be deter-
mined for ?nem?0.7. Therefore, the measurements of the
binodal and vorticity banding are limited to lower fd concen-
trations. More details about the experimental determination
of the binodal are available in Ref. ?10?.
The vorticity-banding region is determined from profiles
taken at various shear rates similar to the ones shown in Fig.
3. The profiles in Fig. 3 persist for more than a week and are
taken for a fixed fd concentration ??nem=0.23?. The shear
rate ? ˙−is the lower-border shear rate of the vorticity-banding
region ?the lower bound of the shaded region in the phase
diagram in Fig. 2?, and ? ˙+is the upper-border shear rate. As
can be seen in Fig. 3, for shear rates below ? ˙−, no clear
bright and dark bands are formed. Inhomogeneities that are
formed due to phase separation are stretched to a certain
extent ?leftmost profile in Fig. 3?, but do not give rise to the
formation of vorticity bands. For shear rates ?? ˙−?? ˙ ?? ˙+?
within the vorticity-banding region, clear bright and dark
bands are formed ?two middle profiles in Fig. 3?. These pro-
files exhibit relatively large intensity variations and a longer-
wavelength structure when compared to those outside the
banding region. For shear rates slightly above ? ˙+, heteroge-
neous vorticity banding is observed ?rightmost profile in Fig.
3?. The shear-rate range ?? ˙+?? ˙ ?? ˙het? where heterogeneous
banding occurs is about 5–10 % of the homogenous
vorticity-banding shear-rate range ?? ˙−?? ˙ ?? ˙+?. Experi-
ments of this kind are repeated for different fd concentrations
FIG. 2. The nonequilibrium phase diagram in the shear rate
versus concentration plane for an overall fd concentration of
21.0 mg/ml and an overall dextran concentration of 10.6 mg/ml.
The fd concentration is expressed in terms of ?nemas defined in
Sec. II A. The upper solid curve is the paranematic-nematic binodal
and the shaded area is the region where vorticity banding is ob-
served. The ? are experimental points where banding occurs for the
first time on increasing the shear rate while the ? are the experi-
mental data where banding ceases to occur. The inset shows part of
the vorticity-banding region for an overall dextran concentration of
14.5 mg/ml. Here, the binodal is not shown. The lower figure
?which is taken from Ref. ?9?? shows a banded state for the same
dextran concentration of 14.5 mg/ml. The bandwidth is about
2 mm. The two enlargements on the right show the inhomogeneities
that are present within the bands. The thickness of these inhomoge-
neities is of the order of 10–20 ?m.
KANG et al.
PHYSICAL REVIEW E 74, 026307 ?2006?
to determine the entire vorticity-banding region in the fd
concentration vs shear rate plane. For each fd concentration
the final shear rate in the two-phase region is systematically
varied in steps of 0.02–0.04 s−1. The prequench shear rate is
always 10 s−1, which is far above the binodal.
Due to sedimentation, the system will eventually phase
separate into a coexisting paranematic and a sheared nematic
bulk phase. This can be seen in Fig. 4 in Ref. ?9?, which
shows a partly demixed system where sedimentation of
denser nematic inhomogeneities occurred to some extent.
The upper phase in this figure is a homogeneous paranematic
phase that does not contain any inhomogeneities. The ab-
sence of a banded structure in the upper phase demonstrates
that the lower binodal in Fig. 2 coincides with the lower
bound of the vorticity-banding region.
The inset in Fig. 2 shows the vorticity-banding region ?up
to ?nem=0.4? for the higher dextran concentration of
14.5 mg/ml. In this case the vorticity-banding region ex-
tends to much higher shear rates since the attractions be-
tween the rods are increased. Still, this region is contained
within the biphasic region as bounded by the binodal. The
vorticity bands at higher overall dextran concentrations are
larger and more regular. Such bands are used for experiments
on the internal flow and orientational order within the bands
as described in Sec. V. The typical quasistationary banded
structures that is observed for these higher dextran concen-
trations are shown in the lower part of Fig. 2. On the right-
hand side there are two microscopy images that show the
inhomogeneous microstructure found within each of the
bands. These inhomogeneities are due to paranematic-
nematic phase separation and are formed right after the
shear-rate quench. In the proposed mechanism that underlies
vorticity banding as presented in Sec. VI, these inhomogene-
ities play an essential role in the stabilization of the banded
IV. KINETICS OF VORTICITY BANDING
The experiments discussed in the present section are done
on fd-dextran suspensions with the lower overall dextran
concentrationof 10.6 mg/ml.
dependent banded intensity profiles subsequent to a quench
from an initial shear rate of 10 s−1to four final shear rates
located in the biphasic paranematic-nematic region. The pro-
files in Fig. 4?a? are for a final shear rate 0.15 s−1just below
the lower-border shear rate ? ˙−=0.16 s−1. No banding is ob-
served. Here, the striped pattern is due to inhomogeneities
that are formed due to paranematic-nematic phase separation
and are elongated in shear flow. For Fig. 4?b?, the final shear
rate 0.17 s−1is just above ? ˙−. Here, the growth of the vortic-
ity bands is clearly visible as the appearance and coarsening
of bright and dark bands. Similarly, for a shear rate 0.45 s−1
just below the upper-border shear rate ? ˙+=0.46 s−1, the
banding transition is clearly observed in Fig. 4?c?. In a small
Figure4 shows time-
FIG. 3. Quasistationary intensity profiles taken a few hours after
the shear-rate quench into the vorticity-banding region. The gap
width is 2.0 mm and the fd concentration is ?nem=0.23. The overall
dextran concentration is 10.6 mg/ml, which complies with the
phase diagram in Fig. 2. The border shear rates ? ˙−and ? ˙+are the
lower and higher shear rates that bound the vorticity-banding region
in the phase diagram. The leftmost profile shows an image for a
shear rate just below ? ˙−. The two middle profiles are for shear rates
within the banding region, while the rightmost profile is taken for
the shear rate just above ? ˙+, where heterogeneous banding occurs.
The vertical axis is given in camera-pixel number and a scale bar is
added on the left side. The intensity scales are the same for all four
FIG. 4. Temporal evolution of the vorticity banding morpholo-
gies at various shear rates for an fd concentration of ?nem=0.23 and
a gap width of 2.0 mm. The overall dextran concentration is
10.6 mg/ml. The numbers above the intensity profile images refer
to the time after the shear-rate quench in minutes. ?a?, ?b? Shear
rates 0.15 and 0.17 s−1just below and above the lower-border shear
rate ? ˙−=0.16 s−1, respectively. ?c?, ?d? Shear rates 0.45 and ? ˙+
=0.47 s−1just below and above the upper-border shear rate ? ˙+
=0.46 s−1, respectively. Heterogeneous banding is observed in ?d?
just above the upper-border shear rate.
VORTICITY BANDING IN RODLIKE VIRUS SUSPENSIONSPHYSICAL REVIEW E 74, 026307 ?2006?
shear-rate region just above the upper-border shear rate ? ˙+,
heterogeneous banding occurs, as can be seen in Fig. 4?d?.
In order to quantify the growth kinetics of vorticity bands,
the time-dependent bandwidth is obtained from intensity pro-
files by the Fourier-mode analysis described in Sec. II B. A
typical result of such an analysis is shown in Fig. 5?b?. There
is a well-defined time t0, beyond which vorticity banding
occurs. Afterward the bandwidth increases continuously until
it saturates at long times ?the times indicated as td–thare in
this regime?. The time t0corresponds to the profile marked
with tc. From the corresponding image in Fig. 5?a? it can be
seen that, indeed, at the time t0clear bright and dark stripes
begin to appear. On average the bright and dark regions have
an equal width during their growth.
It is much harder to ascertain the behavior of the system
prior to the onset of the shear-banding transition. Since the
depth of the camera focus is 1 mm, the acquired image is a
superposition of spatial variations of inhomogeneities along
the gradient direction. Therefore, the apparent bandwidth
found in this region using image analysis is highly suscep-
tible to artifacts. The best way to study the behavior of in-
homogeneities formed in this region is with small-angle light
scattering ?SALS?. SALS data showing the evolution of co-
existing droplets ?inhomogeneities? upon a shear-rate quench
into a biphasic region are shown in Ref. ?9?. The conclusion
drawn from these experiments is that right after the shear-
rate quench the inhomogeneities formed due to paranematic-
nematic phase separation are shear stretched along the flow
direction. This process is usually complete in about 10 min.
After this initial stage we do not see any significant change
in the SALS pattern. This indicates that any further coarsen-
ing process of coexisting phases is suppressed by the shear
flow. Even after vorticity bands are fully developed it is pos-
sible to observe the presence of inhomogeneities in both
bands ?see lower panel of Fig. 2?.
The relevant parameters obtained from growth curves like
the one in Fig. 5?b? are the growth time and final bandwidth
in the stationary state. These parameters are extracted as fol-
lows. We write
H?t? = H0+ H??t?
for t ? t0,
where t0is the time at which banding sets in, H0is the
bandwidth at time t0, and H? describes the growth of the
bands. The time dependence of the vorticity-bandwidth, to
within experimental error, can be described with a single
H??t? = A?1 − exp??t0− t?/????t ? t0?,
where A is the total increase of the bandwidth at long times.
This will be referred to as the growth amplitude while ? is
the band-growth time. The final bandwidth is equal to H0
+A ?see Fig. 5?. The solid line in Fig. 5?b? is a best fit to Eqs.
?4? and ?6? for t?t0. The parameters A and ? characterize the
growth kinetics of the vorticity bands.
An exponential growth of the bandwidth is observed to
within experimental error. The growth time ? is therefore the
relevant experimental measure for the growth rate of bands.
There is as yet no theory concerning band-growth kinetics
that explains exponential growth or possible deviations from
exponential growth at longer times.
The growth time ? and the growth amplitude A are mea-
sured as functions of shear rate and fd concentration ?within
the shaded region in Fig. 2?. To compare such measurements
for different fd concentrations, the shear rate is normalized in
dimensionless units as
? ˙ − ? ˙−
? ˙+− ? ˙−
Here, ? ˙−and ? ˙+are the lower-border and upper-border shear
Systematic measurements of kinetic parameters are done
for two different fd concentrations ?nem=0.23 and 0.35. The
final bandwidth H0+A and the growth time ? are given in
Figs. 6?a? and 6?b? as a function of the normalized shear rate.
FIG. 5. ?a? Temporal development of vorticity profiles right after
a shear-rate quench into the banding region. ?b? The bandwidth H as
a function of time right after a shear quench, as measured from
profiles as given in ?a?. The solid line is an exponential fit to the
data according to Eqs. ?4? and ?6?. Here, the times ta–tcin both
figures are related to shear stretching of inhomogeneities at several
times after the quench, while td–threlate to growth of vorticity
bands. The fd concentration is ?nem=0.23, the shear-cell gap width
is 2.0 mm, and the shear rate is 0.25 s−1. The overall dextran con-
centration is 10.6 mg/ml. The shear rate is located in the middle of
the vorticity-banding region.
KANG et al.
PHYSICAL REVIEW E 74, 026307 ?2006?
The relatively small shear-rate range beyond ? ˙+where het-
erogeneous banding occurs is also indicated in this figure.
Since a Fourier-mode analysis in this region is not possible,
the dashed lines are simple estimates from profiles such as
the one given in Fig. 4?d?.
We find an important difference in the kinetic behavior
depending on the fd concentration. For both cases the band-
width at the instant of time where banding sets in is H0
=60±5 ?m. The final bandwidth is always larger than the
bandwidth H0when banding starts. The final bandwidth for
lower fd concentration ?nem=0.23 varies by about 10%
throughout the banding region. The band-growth time ? di-
verges both in the vicinity of the lower-border shear rate ? ˙−
and below the shear rate ? ˙hetwhere heterogeneous banding
ceases to occur. Hence, for low fd concentration, banding
ceases to occur for shear rates lower than ? ˙−or higher than
? ˙hetdue to a vanishing growth rate 1/?. In contrast, for high
fd concentration ??nem=0.35? band-growth rates are finite
throughout the banding region ?lower curve in Fig. 6?b??.
The reason that banding ceases to occur in this case, for
shear rates below ? ˙−and above ? ˙het, is that the growth am-
plitude A vanishes ?lower curve in Fig. 6?a??.
Additionally we note that the time t0is constant through-
out the region where homogeneous banding occurs, to within
experimental error. For ?nem=0.23, t0=11±2 min, while for
?nem=0.35, t0=8±2 min.
Assuming that the inhomogeneities formed due to phase
separation are at the origin of the vorticity-banding instabil-
ity, the different banding kinetics for the two fd concentra-
tions are probably related to the different morphology and
mechanical properties of these inhomogeneities. Figure 7
shows the morphologies for three different fd concentrations
during the early stages of phase separation, obtained by con-
focal microscopy. These images are taken after a shear-rate
quench to a zero shear rate. For the lowest concentration, an
interconnected morphology is observed during the initial
stages of phase separation. This is reminiscent of spinodal
decomposition. After about 10 min, the connectedness is
lost, and a blurry morphology of inhomogeneities is formed.
At the highest fd concentration, nucleation of isotropic tac-
toids in a nematic background is observed. A somewhat or-
dered, noninterconnected structure exists after about 10 min.
For the middle concentration, which is close to the spinodal,
interconnectedness still exists to some extent after 10 min.
Slow phase separation in the vicinity of the spinodal leads to
a still interconnected structure after a relatively long time.
The fact that spinodal decomposition is observed at low fd
concentration and nucleation and growth at high concentra-
tion is due to the residual alignment of the fd rods subse-
quent to a shear quench ?47,48?. A quantitative analysis of
this type of phase-separation kinetics is given in Ref. ?48?,
where the relevant spinodal concentration is found to be
around an fd concentration of ?nem=0.6. The morphologies
shown in Fig. 7 will be strongly deformed under shear flow,
but will probably still be very different with changing fd
concentrations. Different mechanical properties of these
shear-deformed inhomogeneities might be at the origin of the
observed difference in the banding kinetics for the two fd
concentrations. A more systematic microscopic investigation
is necessary to quantify the relation between the shear-
deformed morphology of inhomogeneities and the observed
As mentioned before, the spinodal and binodal referred to
here are connected to the paranematic-nematic phase transi-
tion, that is, the shear-affected isotropic-nematic phase tran-
sition which also occurs in the absence of flow. It refers to
the shear-affected isotropic-nematic phase transition and not
to the banding transition.
FIG. 6. ?a? The final bandwidth as a function of the normalized
shear rate as defined in Eq. ?7? for two fd concentrations ?nem
=0.23 and 0.35, as indicated in the figure. In both cases the overall
dextran concentration is 10.6 mg/ml. ?b? The band-growth time ?
as a function of shear rate for the same fd concentrations. Above the
upper-border shear rate ? ˙+, heterogeneous banding occurs for both
concentrations. The dotted lines are estimates of bandwidths and
growth rates from intensity profiles like in Fig. 4?d?.
FIG. 7. Confocal images taken in reflection mode of the mor-
phology of phase-separating suspensions, 90, 300, and 600 s after a
shear-rate quench to zero shear rate, for three different fd concen-
trations, ?nem=0.23, 0.52, and 0.70, as indicated in the figure. The
bright regions are nematic, the dark regions are isotropic. The field
of view is 180 ?m.
VORTICITY BANDING IN RODLIKE VIRUS SUSPENSIONSPHYSICAL REVIEW E 74, 026307 ?2006?
As will be discussed in Sec. VI, gradients in shear rate
determine the normal stress that is generated by the inhomo-
geneities, which in turn stabilizes the quasistationary state.
Since the morphology of inhomogeneities at the time that
banding sets in depends on the overall fd concentration, as
discussed above, the final bandwidth, therefore, might be
concentration dependent. As can be seen from Fig. 8?a?, the
final bandwidth is indeed seen to be depending on the fd
concentration. The final bandwidth decreases with increasing
fd concentration. There should also be a gap-width depen-
dence of the final bandwidth due to the fact that the gradients
in shear rate increase with increasing gap width. Such a gap-
width dependence is indeed found, as shown in Fig. 8?b?. As
can be seen, the bandwidth increases as the gradients in shear
rate become larger.
V. STRUCTURE OF QUASISTATIONARY VORTICITY-
Two experiments provide strong indication that the vor-
ticity banding is due to a weak rolling flow superimposed
onto the applied shear flow. For both of these experiments
we have used a somewhat higher dextran concentration of
12.3 mg/ml to produce large and regularly stacked vorticity
bands. The width of the quasistationary bands under these
conditions is about 1 mm. In a first experiment we use po-
larization optics to probe the orientational order of fd rods
within adjacent bands. The couette cell is placed between an
exactly crossed polarizer and analyzer ?see Fig. 1?. A ?/2
plate is placed between the couette cell and the analyzer,
with a variable angle ? with respect to the polarization di-
rection of the analyzer. When the optical axis of the ?/2
plate is perpendicular ?or parallel? to A there is no visible
contrast between the two bands ?middle image in Fig. 9?a??.
This implies that the orientational order averaged along the
gradient direction is the same in both bands. However, all the
rods along the gradient direction are not necessarily aligned
in the same direction. In other words the rod orientation can
form a left- or right-handed helix. This can be shown by
changing the angle of the ?/2 plate. The polarized light is
rotated equally by the two bands but in exactly opposite
directions. This is nicely demonstrated by the leftmost and
rightmost images in Fig. 9?a?, where the ?/2 plate is at
angles 80° and 100° with respect to A. The brightness of a
band in the left image is equal to the brightness of adjacent
bands in the right image. The transmitted intensity in the two
bands as a function of the angle ? is plotted in Fig. 9?b?.
Such behavior is consistent with banded structures that are in
a rolling motion ?as depicted in Fig. 10?b??. Since the rolling
motion is opposite in direction for two adjacent bands, the
change of the polarization direction of light is equal but op-
posite in sign for the two bands. This optical phenomenon is
similar to propagation of light in cholesteric liquid crystals
In the second experiment focused on determining the
structure of vorticity bands we have tracked tracer particles
within a vorticity band. Here, a couette cell is used with two
counter-rotating cylinders. The height of the cell is adjusted
to keep track of the particle along the vorticity direction. The
position of the tracer particle in the gradient direction is
probed by changing the relative angular velocity of the two
cylinders so as to shift the plane of zero velocity in order to
keep the particle within the field of view. The relative angu-
lar velocities are adjusted in such a way that the average
shear rate remains unchanged. As can be seen from Fig.
10?a?, the height z of the tracer sphere, which is the compo-
nent of its position along the vorticity direction, oscillates in
time with an amplitude that is roughly equal to the band-
width. There is drift of the z position due to sedimentation of
the large tracer particle. The experimental measure for the
location of the tracer sphere in the gradient direction is the
so-called speed ratio S. S is defined as
FIG. 8. ?a? The concentration dependence of the final bandwidth
H for a shear rate equal to 0.34 s−1and for the lower overall con-
centration of 10.6 mg/ml. The gap width of the shear cell is
2.0 mm. ?b? The gap-width dependence of the final bandwidth for
?nem=0.23. The data points are averages over shear rates within the
banding region, and the error bars relate to the spread in bandwidth
on variation of the shear rate ?see the upper curve in Fig. 6?a??.
FIG. 9. ?a? Images of shear bands illuminated with white light
with a polarization direction parallel to the flow and the crossed
analyzer directed along the vorticity direction. Between the ana-
lyzer and the couette cell there is a ?1/2?? platelet with its optical
axis at an angle of 80°, 90°, and 100° with respect to the analyzer
?from left to right?. The height of the images is 6.5 mm. ?b? The
transmitted intensity of two adjacent bands ?the bands marked as 1
and 2 as a function of the angle ?. ? is the angle between the
analyzer and the ?1/2?? platelet. The solid lines are best fits to a
linear combination of a sine and cosine with the same period and
the same offset phase. The overall dextran concentration is
12.3 mg/ml, the shear rate is 0.26 s−1, and ?nem=0.17.
KANG et al.
PHYSICAL REVIEW E 74, 026307 ?2006?
where ?iand ?oare the ?absolute values of? the rotational
velocities of the counter-rotating inner and outer cylinders,
respectively. The rolling velocity is simply superimposed
onto the velocity v0in the perpendicular direction that one
would have for a Newtonian fluid in a couette cell. Indeed,
heterodyne dynamic light scattering experiments within the
banded state show an essentially linear velocity profile as for
a Newtonian fluid. The fluid flow velocity v0is given as a
function of the radial distance ? from the centerline of rota-
2− ??o+ ?i?Ro
where Riand Roare the radii of the inner and outer cylinders,
respectively. The location ?0of the plane of zero velocity is
thus equal to
2+ ?1 − S?Ri
where the definition of the experimental parameter S in Eq.
?8? has been substituted. This relation is used to construct the
plot in Fig. 10?a? for the radial position of the tracer particle.
Although there is a drift of the particle toward the outer
cylinder due to gradients in the shear rate, the radial position
seems to exhibit an oscillatory behavior. Since the tracer par-
ticle will migrate to the outer cylinder due to gradients in
shear rate and its initial radial position is in the middle of the
gap of the shear cell, these oscillations are not very pro-
nounced. Therefore, the data for ?0are too noisy to unam-
biguously correlate the height and the radial position in order
to prove that there is a rolling flow. The data in Fig. 10,
however, strongly indicate the existence of a rolling flow, as
depicted in Fig. 10?b?.
Typically, 1–5 h are needed to establish a banded state
that remains unchanged for at least a week. There is a differ-
ence between density of nematic and paranematic phase and
therefore the inhomogeneities will slowly sediment over a
long period of time. This will eventually lead to a state where
two homogeneous bulk phases coexist: a paranematic and a
sheared nematic phase ?see Fig. 4 in Ref. ?9?.?. Microscopy
images and small-angle light scattering experiments indicate
that inhomogeneities are present within the bands ?9,10?.
These inhomogeneities are necessary to maintain a normal
stress along the gradient direction that stabilizes the vorticity
bands. As soon as these inhomogeneities coalesce, for ex-
ample due to sedimentation, the stabilizing normal stress
ceases to exist, and bands will disappear. The banded struc-
ture is therefore a long-lived, transient state. In this sense the
banded structure is referred to as quasistationary, since the
inhomogeneities will not persist for ever.
VI. POSSIBLE MECHANISM FOR VORTICITY BANDING
We have measured the flow curve of fd-dextran mixtures
for the possible rheological signs of a gradient-banding tran-
sition. In the case of gradient banding, the flow curve of the
homogeneous system, before banding occurs, exhibits a van
der Waals looplike dependence on the shear rate. That is,
there is a region of shear rates where the shear stress of the
homogeneously sheared suspension decreases with increas-
ing shear rate ?see Fig. 11?a??. A mechanism similar to gra-
dient banding would play a role in the present system when
the shear stress of the suspension just before banding sets in
FIG. 10. ?a? On the right axis, the position z in the vorticity
direction of a tracer sphere with a diameter of 50 ?m is shown as a
function of time in the quasistationary banded state as measured by
video microscopy with a counter-rotating couette cell, where the
radii of the inner and outer cylinders are 18.5 and 20.0 mm, respec-
tively. The time t=0 is the time after which the optical trapping of
the tracer sphere is released. The radial position ?0of the tracer
sphere is shown on the left axis. The overall dextran concentration
is 12.3 mg/ml, the fd concentration is ?nem=0.17, and the shear
rate is 0.88 s−1. The bandwidth is about 1 mm. ?b? A sketch of the
rolling flow that complies with the observed oscillatory behavior of
the position coordinate of the tracer particle.
FIG. 11. ?a? A typical van der Waals looplike shear-rate depen-
dence of the shear stress for a homogeneously sheared system that
will exhibit gradient banding. The viscosity ? and the shear stress
?=? ˙ ? as a function of shear rate are shown in ?b? and ?c?, respec-
tively. Vorticity banding is observed within the shaded region. The
vertical bounds of the shaded area correspond to the lower- and
upper-border shear rates. The data points are measured 5 min after
a shear-rate quench, just before vorticity banding occurs. The fd
concentration is ?nem=0.35. ?d? The shear stress for higher shear
rates, for ?nem=0.33. The overall dextran concentration is
10.6 mg/ml. The vertical line indicates the location of the binodal.
The gap width is 1.0 mm.
VORTICITY BANDING IN RODLIKE VIRUS SUSPENSIONSPHYSICAL REVIEW E 74, 026307 ?2006?
would exhibit such a decrease of the shear stress with in-
creasing shear rate. When gradient bands are allowed to fully
develop before a stress measurement is done, a plateau in the
shear stress as a function of shear rate is observed under
controlled shear-rate conditions ?17–21?. This plateau can be
tilted when the stress couples strongly to variables such as
concentration ?17?. There is no such signature in the flow
curve for the fd suspensions within the vorticity-banding re-
gion as seen in Figs. 11?b? and 11?c?. Here, the shaded area
indicates the vorticity-banding region, that is, the shear-rate
range ?? ˙−,? ˙+? in the phase diagram in Fig. 2 where vorticity
banding is observed. The stress and viscosity data in Figs.
11?b? and 11?c? are obtained 5 min after a shear-rate quench.
Vorticity banding sets in after 10 min. These data essentially
do not change when measured within a time window of
about 5–10 min after the quench. In the case of gradient
banding such curves would exhibit a van der Waals looplike
behavior as depicted schematically in Fig. 11?a?. The system
is only weakly shear thinning ?see Fig. 11?b?? and does not
show any sign of a van der Waals loop nor a stress plateau.
In addition, we have repeated the determination of the border
shear rates for one concentration under controlled stress con-
ditions. No differences were found with the controlled rate
experiments: both border shear rates are the same. The
mechanism for vorticity banding is thus clearly different
from that of gradient banding. Finally we confirmed that the
shear stress is also well behaved throughout the entire bipha-
sic region, as shown in Fig. 11?d?. Here, the vertical line
indicates the location of the binodal.
In view of the rolling flow within the bands ?see Sec. V?,
the vorticity-banding instability might have a similar origin
as the elastic instability studied in polymeric systems
?35–39?. The origin of this well-known instability is as fol-
lows. When there is a gradient in shear rate, as in a couette
cell, the shear-induced stretching of polymer chains leads to
normal stresses along the gradient direction. Chains which
are not perfectly aligned along the streamlines are stretched
in a nonuniform way due to gradients in the shear rate ?Fig.
12?a??. On average, such nonuniform stretching of chains
leads to normal stresses that pull a volume element toward
the rotating inner cylinder ?also depicted in Fig. 12?a??.
These “hoop stresses”’ set the fluid in motion toward the
inner cylinder. This leads to a rolling flow ?as sketched in
Fig. 10?b??, since at the inner cylinder the flow velocity must
change to the vorticity direction. In case of a free surface, the
fluid may climb the inner cylinder, in which case the increase
in hydrostatic pressure compensates the normal stress in the
gradient direction. This is known as the Weissenberg or rod-
We propose that, in a similar way, nonuniform shear-
induced deformation of the inhomogeneities ?instead of poly-
mer chains? may be at the origin of the vorticity-banding
instability, as depicted in Fig. 12?b?. Nonuniform deforma-
tion of inhomogeneities ?formed during the initial stages of
paranematic-nematic phase separation? due to gradients in
shear rate are thus responsible for the vorticity-banding in-
stability and the stabilization of the banded structures.
There are, however, additional forces that might play a
role here. Since inhomogeneities are very much extended
along streamlines, bending elasticity might give rise to sig-
nificant normal stresses in the opposite direction, away from
the center of the couette cell ?see Fig. 12?c??. Bending elas-
ticity ?49? might counteract the stretching forces in generat-
ing hoop stresses.
The following experimental observations can be intu-
itively understood on the basis of the above-proposed
?i? At a given overall shear rate, gradients in the shear rate
in a couette cell increase with increasing gap width of the
couette cell. Therefore the driving force for rolling flow will
increase with increasing gap width. This will probably lead
to an increase of the bandwidth. Indeed an increase of the
bandwidth with increasing gap width is observed ?see Fig.
?ii? Shear gradients in a couette cell are large when the
overall shear rate is large. This might explain why vorticity
banding occurs only at sufficiently high shear rates, that is,
the lower-border shear rate ? ˙−is larger than zero. Gradients
in shear rate are not sufficiently pronounced for shear rates
below ? ˙−to render the normal stresses strong enough to in-
duce vorticity banding.
?iii? At larger shear rates, close to the upper-border shear
rate, inhomogeneities are relatively thin due to shear stretch-
ing. This diminishes the nonuniform stretching within the
inhomogeneity ?see Fig. 12?b?, where now the inhomogene-
ity is very thin?. In addition, the bending forces will be
smaller, since a thinner inhomogeneity is more easily bent as
compared to a thick inhomogeneity. This might explain why
banding ceases to occur above the upper-border shear rate
? ˙+, which is well within the two-phase region.
?iv? Contrary to gradient banding, the region where vor-
ticity banding occurs and the final bandwidth are indepen-
FIG. 12. ?a? The origin of the well-known elastic instability of
polymer systems. The dots on the polymer chain are used to indi-
cate the degree of stretching. Without stretching or for uniform
stretching, the dots would be equally spaced. When a chain is not
aligned along streamlines, stretching is nonuniform and normal
stresses are generated. The resulting forces pull a volume element
toward the rotating inner cylinder, as indicated by the arrow. ?b?
Similar stretching of the inhomogeneities that are formed due to
nonequilibrium paranematic-nematic phase separation. The dots are
used to visualize nonuniform stretching. ?c? In comparison to flex-
ible polymer chains, bending elasticity may play a role as well. In
such cases normal stresses act in the opposite direction, as depicted
by the arrows in the figure.
KANG et al.
PHYSICAL REVIEW E 74, 026307 ?2006?
dent of whether the shear rate or shear stress is controlled.
This agrees with the mechanism proposed above, since nor-
mal stresses are important instead of shear-gradient stresses.
Although the mechanism proposed above is in accord
with a number of experimental observations, more theoreti-
cal and experimental work is needed to validate these sug-
gestions. Measurements of normal stresses along the gradient
direction during the initial stages of banding could be per-
formed. This will not be straightforward since, in weak roll-
ing motion, these stresses are probably quite small. In addi-
tion, a detailed experimental study of the dependence of the
flow pattern within the bands on varying gap width would be
Stationary vorticity-banded states are also found in aggre-
gated nanotube suspensions ?50?. Instead of the inhomoge-
neities formed due to phase separation as in our fd virus
suspensions, here, similar deformation of the nanorod aggre-
gates may be at the origin of the vorticity banding. Normal
stress measurements reported in Ref. ?50? indeed indicate
that such stresses play a role in vorticity banding. Stationary
vorticity banding has also been observed in wormlike micel-
lar systems ?23?. There are two possibilities here: either in-
paranematic-nematic phase transition, or the wormlike mi-
celles themselves are nonlinearly stretched ?like polymer
chains in the elastic instability of polymer systems?. In those
paranematic-nematic phase transition occurs and no vorticity
banding is observed, the mechanical properties related to
stretching of the inhomogeneities are probably such that nor-
mal stresses are not large enough to give rise to rolling flow.
In the weakly flocculating systems in Ref. ?22?, the nonuni-
form stretching of flocs of colloidal particles is probably at
the origin of the observed vorticity banding.
We have performed experiments under steady shear-flow
conditions on the structure and kinetics of vorticity banding
in sheared suspensions of rodlike fd virus. We determined
the vorticity- banding region which is entirely enclosed by
the paranematic-nematic binodal. Under both controlled
shear-rate and shear-stress conditions banding occurs be-
tween the lower- and upper-border shear rates ? ˙−and ? ˙+,
respectively, where ? ˙−is larger than zero. After a shear-rate
quench from a high shear rate into this region, inhomogene-
ities are formed due to phase separation. These inhomogene-
ities are shear elongated up to a well-defined time after
which vorticity banding occurs. The growth of the vorticity-
bandwidth can be described, to within experimental error, by
a single-exponential function of time. There are two impor-
tant parameters that characterize the kinetics of band forma-
tion: ?i? the band-growth time ? in the exponential, the in-
verse of which measures the growth rate of band formation,
and ?ii? the amplitude A of the time exponential, which is
related to the total growth of the bandwidth as compared to
the initial apparent bandwidth of shear-stretched inhomoge-
neities. The growth kinetics depends on the morphology and
the mechanical properties of the inhomogeneities formed due
to the phase separation. Two different scenarios have been
found. For small fd concentration, ? diverges at the border
shear rates, while the amplitude A remains finite. The growth
rate of bands thus vanishes. For larger fd concentration, the
amplitude A vanishes at the border shear rates, while ? re-
Experiments indicate that there is a weak rolling flow
within the bands. A possible mechanism that is at the origin
of the banding instability as well as the stabilization of qua-
sistationary banded states is proposed, where the mechanical
properties of the inhomogeneities are essential. The proposed
mechanism is similar to the elastic instability for polymer
systems, where the inhomogeneities play the role of the
The proposed mechanism explains, on an intuitive level, a
number of the observed phenomena, like the gap-width de-
pendence of the vorticity-bandwidth, the large bandwidth
with increasing dextran concentration, and the fact that band-
ing occurs only beyond a certain nonzero shear rate and
ceases to occur above another shear rate. Moreover, the pro-
posed mechanism explains why the border shear rates are
independent of whether controlled shear rates or controlled
shear stresses are applied. Just above the upper-border shear
rate there is a finite probability to have a localized assembly
of neighboring inhomogeneities which are still thick enough
to give rise to banding. This leads to the observed heteroge-
neous band formation.
More experiments are necessary to unambiguously vali-
date the proposed mechanism. In addition, theory should be
developed to confirm such a scenario. In particular, the ex-
pression for the stress tensor as obtained in Ref. ?51?, which
is valid for highly inhomogeneous systems of stiff, long, and
thin rods, could serve as a starting point for the theoretical
validation of the proposed mechanism. Simulations also
might provide a better understanding of the behavior of col-
loidal rods under shear flow ?52,53?.
This work was performed within the framework of the
Transregio SFB TR6 “Physics of Colloidal Dispersions in
External fields,” and is also part of the research being per-
formed within the European Network of Excellence “Soft
Matter Composites” ?SoftComp?. Z.D. is supported by the
Rowland Institute at Harvard. The confocal microscopy ex-
periments have been done in cooperation with D. Derks, A.
Imhof, and A. van Blaaderen, from the Utrecht University in
The Netherlands. The tracking experiment has been done in
cooperation with M. Lenoble and B. Pouligny from the Cen-
tre Recherche Paul Pascal, CNRS, Pesac in France.
VORTICITY BANDING IN RODLIKE VIRUS SUSPENSIONSPHYSICAL REVIEW E 74, 026307 ?2006?
?1? W. C. K. Poon and P. N. Pusey, in Observation, Prediction, Download full-text
and Simulation of Phase Transitions in Complex Fluids, edited
by M. Baus, L. F. Rull, and J. P. Ryckaert ?Kluwer Academic
Publishers, Dordrecht, 1995?.
?2? G. J. Vroege and H. N. W. Lekkerkerker, Rep. Prog. Phys. 8,
?3? A. Baranyai and P. T. Cummings, Phys. Rev. E 60, 5522
?4? S. Butler and P. Harrowell, Nature ?London? 415, 1008 ?2002?.
?5? M. C. Cross and P. C. Hohenberg, Rev. Mod. Phys. 65, 851
?6? J. F. Berret, Langmuir 13, 2277 ?1997?.
?7? P. Boltenhagen, Y. Hu, E. F. Matthys, and D. J. Pine, Phys.
Rev. Lett. 79, 2359 ?1997?.
?8? T. A. J. Lenstra, Z. Dogic, and J. K. G. Dhont, J. Chem. Phys.
114, 10151 ?2001?.
?9? J. K. G. Dhont, M. P. Lettinga, Z. Dogic, T. A. J. Lenstra, H.
Wang, S. Rathgeber, P. Carletto, L. Willner, H. Frielinghaus,
and P. Lindner, Faraday Discuss. 123, 157 ?2003?.
?10? M. P. Lettinga and J. K. G. Dhont, J. Phys.: Condens. Matter
16, S3929 ?2004?.
?11? J.-F. Berret, Langmuir 13, 2227 ?1997?.
?12? E. Fischer and P. T. Callaghan, Phys. Rev. E 64, 011501
?13? C. Pujolle-Robic and L. Noirez, Nature ?London? 409, 167
?14? P. T. Mather, A. Romo-Uribe, C. D. Han, and S. S. Kim, Mac-
romolecules 30, 7977 ?1997?.
?15? N. Grizzuti and P. L. Maffettone, J. Chem. Phys. 118, 5195
?16? S. Lerouge, J. P. Decruppe, and C. Humbert, Phys. Rev. Lett.
81, 5457 ?1998?.
?17? P. D. Olmsted and C.-Y. David Lu, Phys. Rev. E 60, 4397
?18? C.-Y. David Lu, P. D. Olmsted, and R. C. Ball, Phys. Rev. Lett.
84, 642 ?2000?.
?19? P. D. Olmsted, O. Radulescu, and C.-Y. Lu, J. Rheol. 44, 257
?20? J. L. Goveas and P. D. Olmsted, Eur. Phys. J. E 6, 79 ?2001?.
?21? J. K. G. Dhont, Phys. Rev. E 60, 4534 ?1999?.
?22? J. Vermant, L. Raynaud, J. Mewis, B. Ernst, and G. G. Fuller,
J. Colloid Interface Sci. 211, 221 ?1999?.
?23? D. Bonn, J. Meunier, O. Greffier, A. Al-Kahwaji, and H.
Kellay, Phys. Rev. E 58, 2115 ?1998?.
?24? P. Fischer, E. K. Wheeler, and G. G. Fuller, Rheol. Acta 41, 35
?25? L. Onsager, Ann. N.Y. Acad. Sci. 51, 62 ?1949?.
?26? Z. Dogic and S. Fraden, in Soft Condensed Matter, edited by
G. Gompper and M. Schick ?Viley-VCH, Weinheim, 2006?.
?27? J. Tang and S. Fraden, Liq. Cryst. 19, 459 ?1995?.
?28? E. Grelet and S. Fraden, Phys. Rev. Lett. 90, 198302 ?2003?.
?29? Z. Dogic, K. R. Purdy, E. Grelet, M. Adams, and S. Fraden,
Phys. Rev. E 69, 051702 ?2004?.
?30? F. Livolant, Physica A 176, 117 ?1991?.
?31? S. Fraden, G. Maret, D. L. D. Caspar, and R. B. Meyer, Phys.
Rev. Lett. 63, 2068 ?1989?.
?32? M. P. B. van Bruggen, J. K. G. Dhont, and H. N. W. Lek-
kerkerker, Macromolecules 32, 7037 ?1999?.
?33? J. Tang and S. Fraden, Phys. Rev. Lett. 71, 3509 ?1993?.
?34? J.-F. Berret, D. C. Roux, and G. Porte, J. Phys. II 4, 1261
?35? K. Weissenberg, Nature ?London? 159, 310 ?1947?.
?36? C. W. Macosko, Rheology, Principles, Measurements and Ap-
plications ?VCH Publishers, Cambridge, U.K., 1994?.
?37? R. G. Larson, S. J. Muller, and E. S. G. Shaqfeh, J. Fluid
Mech. 218, 573 ?1990?.
?38? P. Pakdel and G. H. McKinley, Phys. Rev. Lett. 77, 2459
?39? A. Groisman and V. Steinberg, Phys. Fluids 10, 2451 ?1998?.
?40? T. Maniatis, J. Sambrook, and E. F. Fritsch, Molecular Clon-
ing: A Laboratory Manual, 2nd ed. ?Cold Spring Harbor Labo-
ratory Press, Plainview, NY, 1989?.
?41? M. P. Lettinga, Z. Dogic, H. Wang, and J. Vermant, Langmuir
21, 8048 ?2005?.
?42? S. Hess, Z. Naturforsch. A 31, 1034 ?1976?.
?43? G. Rienäcker, M. Kröger, and S. Hess, Phys. Rev. E 66,
?44? S. Hess and M. Kröger, J. Phys.: Condens. Matter 16, S3835
?45? M. G. Forest, Q. Wang, and R. Zhou, Rheol. Acta 44, 80
?46? M. G. Forest, X. Zheng, R. Zhou, Q. Wang, and R. Lipton,
Adv. Funct. Mater. 15, 2029 ?2005?.
?47? J. K. G. Dhont and W. J. Briels, Phys. Rev. E 72, 031404
?48? M. P. Lettinga, K. Kang, A. Imhof, D. Derks, and J. K. G.
Dhont, J. Phys.: Condens. Matter 17, S3609 ?2005?.
?49? P. G. de Gennes and J. Prost, The Physics of Liquid Crystals,
2nd ed. ?Clarendon Press, Oxford, 1999?.
?50? S. Lin-Gibson, J. A. Pathak, E. A. Grulke, H. Wang, and E. K.
Hobbie, Phys. Rev. Lett. 92, 048302 ?2004?.
?51? J. K. G. Dhont and W. J. Briels, J. Chem. Phys. 117, 3992
?2002?; 118, 1466 ?2003?.
?52? R. Zhou, M. G. Forest, and Q. Wang, Multiscale Model. Simul.
3, 853 ?2005?.
?53? G. Germano and F. Schmid, J. Chem. Phys. 123, 214703
KANG et al.
PHYSICAL REVIEW E 74, 026307 ?2006?