Article

pubs.acs.org/Macromolecules

Viscoelasticity, Nonlinear Shear Start-up, and Relaxation of

Entangled Star Polymers
Frank Snijkers,† Kedar Ratkanthwar,‡,§ Dimitris Vlassopoulos,*,†,⊥ and Nikos Hadjichristidis*,§,∥
†
Institute of Electronic Structure and Laser, FORTH, Heraklion 71110, Crete, Greece
‡
School of Chemical Sciences, Swami Ramanand Teerth Marathwada University, Vishnupuri, Nanded-431606, India
§
Department of Chemistry, University of Athens, Athens 15771, Greece
⊥
Department of Materials Science & Technology, University of Crete, Heraklion 71003, Crete, Greece
∥
Division of Physical Sciences & Engineering, KAUST Catalysis Center, Polymer Synthesis Laboratory, King Abdullah University of
Science and Technology (KAUST), Thuwal, Kingdom of Saudi Arabia

ABSTRACT: We report on a detailed rheological investigation of

well-defined symmetric entangled polymer stars of low functionality
with varying number of arms, molar mass of the arms, and solvent
content. Emphasis is placed on the response of the stars in simple
shear, during start-up, and for relaxation upon flow cessation. To
reduce experimental artifacts associated with edge fracture (primarily)
and wall slip, we employ a homemade cone-partitioned plate fixture
which was successfully implemented in recent studies. Reliable data for
these highly entangled stars could be obtained for Weissenberg
numbers below 300. The appearance of a stress overshoot during start-
up with a corresponding strain approaching a value of 2 suggests that in
the investigated shear regime the stars orient but do not stretch. This is
corroborated by the fact that the empirical Cox−Merx rule appears to
be validated, within experimental error. On the other hand, the (shear) rate dependent steady shear viscosity data exhibit a slope
smaller than the convective constraint release slope of −1 (for linear polymers) for the investigated range of rates. The broadness
of the stress overshoot reflects the broad linear relaxation spectrum of the stars. The initial stress relaxation rate, reflecting the
initial loss of entanglements due to the action of convective constraint release in steady shear flow, increases with Weissenberg
number. More importantly, when compared against the relevant rates for comb polymers with relatively short arms, the latter are
slower at larger Weissenberg numbers. At long times, the relaxation data are consistent with the linear viscoelastic data on these
systems.

I. INTRODUCTION characterized polymers is on the other hand far less explored

Entangled branched polymers represent a topic of great and understood. In particular, with the exception of a few data
scientific and technological significance. In particular, under- sets only, which are discussed below,14 systematic studies of
standing their rheology is central to linking processing shear rheology are essentially lacking. In order to have a reliable
performance to consumer products and to designing new and versatile predictive toolbox for any entangled branched
materials with improved properties. Thanks to the substantial polymer, it is important to have reliable data for various well-
advances of the last two decades, the rheology of commercial defined architectures. It turns out that this remains an open
branched polymers can be successfully predicted at molecular challenge for at least three reasons. (i) The anionic high-
level, starting from their polymerization.1 Nevertheless, it is vacuum synthesis methods for the production of high-quality,
important to elucidate the role of exact molecular features well-characterized complex polymers are very involved15,16 and
(number, molar mass and distribution of branches) on the the resulting sample amounts very small, which limits the
rheology of branched polymers. To this end, the role of well- applicable techniques to study their behavior (for example, the
characterized complex polymers is essential.2 In fact, the linear use of capillary rheometry is impossible). (ii) Moreover, while
viscoelastic behavior of well-characterized polymer architectures the linear behavior can be measured with ease in commercial
is relatively well understood and the current versions of the rotational rheometers, the nonlinear flow behavior is much
tube models describe the relaxation behavior of star more difficult to study due to flow instabilities of different
polymers,3,4 H polymers,5,6 combs,7−9 starcombs,10 and Cayley
trees11−13 nearly quantitatively. The success is largely due to Received: April 2, 2013
the synergy of anionic synthesis, characterization methods, Revised: June 25, 2013
rheology and modeling. The nonlinear flow behavior of well- Published: July 8, 2013

© 2013 American Chemical Society 5702 dx.doi.org/10.1021/ma400662b | Macromolecules 2013, 46, 5702−5713
Macromolecules Article

Scheme 1. General Reactions for the Synthesis of 4-Arm and 8-Arm Star Polyisoprenesa

a
8-SiCl: {Si[CH2CH2SiCl2(CH3)]4}. Red (outer part) represents polyisoprene chains and blue (inner part) polybutadiene chains.

origin, especially in shear flow.17−19 (iii) Even if anionic that nonlinear stress relaxation measurements are convenient as
synthesis is the best means to obtain exact architectures, there they are somewhat less prone to instabilities and other artifacts,
are always shortcomings and the issue of architectural dispersity presumably due to the delay in the built-up of normal forces
is emerging as an important one.6,20,21 during short-time deformation at relatively small strains,
In recent work, we have addressed the nonlinear shear although they are not without challenges such as slip.29−31
response of well-characterized entangled polymers with comb The results for the damping function for symmetric stars were
architecture.22 A main finding of that work is that, in fast flows, found to be in agreement with the Doi−Edwards prediction for
the combs with relatively short arms behave qualitatively much linear polymers. Using rheo-optical methods, Tezel et al.32,33
like linear polymers since their arms relax very fast. The same examined the validity of the convective constraint release
conclusion was drawn by Ianniruberto and Marrucci for combs (CCR) mechanism (proposed for entangled linear polymers)34
in elongational flows.23 It is therefore interesting to explore the for the steady-state shear viscosity of star solutions. They
role of the arms in more detail, and hence a systematic concluded that, given some uncertainties, in general CCR
investigation of the simplest possible branched polymer works in roughly the same way for stars as for linear polymers.
architecture, i.e., monodisperse star polymers, is in order. We Menezes and Graessley35 reported on the start-up shear
study symmetric entangled star polymers of low functionality, behavior of a well-defined entangled star solution. Because of
where each arm has an identical molar mass and in which the the occurrence of instabilities, the nonlinear region they
number of arms per star is small; hence, there is no effect of the accessed was very limited, but they were able to capture the
core. We address points i and ii above: i.e., we investigate the first normal stress reliably. For completion, we note that Ye and
nonlinear shear rheology of model stars in the melt and in Sridhar36 studied the steady extensional viscosity of entangled
solution using appropriate instrumentation to handle small star solutions using a filament stretching rheometer (FSR). The
sample quantities and to avoid flow instabilities. Specifically for steady extensional viscosity initially decreased with strain rate,
stars, the linear relaxation behavior was found to be dominated and at higher strain rates a region of extension thickening was
by contour-length fluctuations of the arms, constraint release observed. The results are qualitatively similar to those obtained
and the associated dynamic dilution,14,24,25 in sharp contrast to with solutions of linear polymers. However, for star polymers
the response of linear monodisperse polymers, where reptation the region of extensional thinning was much wider compared to
is the key mechanism. The relaxation of low-functionality stars the region for linear polymers, presumably due to the broad
is furthermore independent of the number of arms. Although relaxation spectrum of the stars, reflecting a large difference
not without outstanding issues, the star dynamics is rather well between the longest relaxation time for renewal of chain
understood by now2,24 and the current versions of the popular conformations and the Rouse time of the entangled arms.
generic tube models, such as the branch-on-branch model Here, we explore the nonlinear shear flow behavior of
(BOB model), 4 capture the relaxation behavior near entangled symmetric stars using a specially designed homemade
quantitatively. Turning now to the state-of-the-art of the cone partitioned-plate geometry that delays edge fracture
nonlinear rheological behavior, there have been several studies problems and allows for more accurate nonlinear shear
on the stress relaxation of (mildly) entangled star polymer measurements.37 Possible wall-slip issues are reduced in this
solutions.26−28 From a practical perspective, it can be stated type of geometry due to the large normal stresses in the middle
5703 dx.doi.org/10.1021/ma400662b | Macromolecules 2013, 46, 5702−5713
Macromolecules Article

Figure 1. Size exclusion chromatography traces of PI 4a-56k before (a) and after (b) fractionation.

region of the geometry where the viscoelastic properties are determined after passing PI through the SEC instrument which was
measured.38 The focus here is on start-up shear flow during a calibrated with polystyrene (PS) standards. It relates to the fact that PS
sudden imposition of a constant shear rate and then subsequent and PI have different hydrodynamic volumes (the key parameter in
relaxation upon cessation of steady-state flow. The star SEC).40 Indeed, our (unpublished) SEC experimental results from
many anionically prepared PS and PI (of high 1,4 content) models
polymers used and experimental details are described in section revealed a constant ratio of the molar masses of 1.6, in a plot of molar
II. Then, the results are presented in section III, starting from mass versus elution volume. This is consistent with the well-known
linear viscoelasticity and continuing with nonlinear start-up and Mark−Houwink equation for the intrinsic viscosity of polymers as a
relaxation of shear stress. Section IV discusses the experimental function of molar mass M, [η] = KMα, with K and α being constants
results and compares against those available in the literature. depending on the polymer−solvent system. For all samples, the
Finally the main conclusions are summarized in section V. polydispersity index was lower than 1.1. NMR spectra, generated with
a Bruker 400-MHz instrument with CDCl3 at 25 °C, revealed that all
PIs have a high 1,4-content (93 to 95%). Molar masses of the arms
II. MATERIALS AND METHODS (Ma) and other characteristics are given in Table 1. The coding
II.1. Synthesis and Characterization of 4-Arm and 8-Arm
Star Polyisoprenes. The synthesis of the 4- and 8-arm polyisoprenes Table 1. Molecular Characteristics of the Investigated
(PI) was achieved by using anionic polymerization high vacuum Polymersa
techniques and chlorosilane chemistry (Scheme 1). All polymer-
izations and linking reactions were carried out in evacuated, n-BuLi- Ma MTOT Me
washed, and solvent-rinsed glass reactors. Reagents were introduced sample code [kg/mol] [kg/mol] [kg/mol] Za f
via break-seals and aliquots for characterization were removed by heat- PI 4a-56k 56 224 5.0 11.2 4
sealing of constrictions. Full details of the high vacuum techniques are PI 8a-56k 56 448 5.0 11.2 8
given elsewhere.39 First, a narrow molecular weight linear living
PI 4a-103k 103 412 5.0 20.6 4
polyisoprene was prepared, in benzene at 25 °C, with sec-BuLi as the
PI 4a-103k 86% 103 412 6.1 16.9 4
initiator. A small fraction was removed and terminated with degassed
MeOH (arm of the star). The living polyisoprenyllithium was capped PI 4a-103k 62% 103 412 9.5 10.8 4
a
with a few units of butadiene (Bd) prior to reaction with the See text for definitions of the quantities.
multifunctional chlorosilane compound. The 4-arm star was prepared
by linking of the arms with tetrachlorosilane (SiCl4) and the 8-arm star
was obtained with tetra(methyl dichlorosilylethyl)silane {Si- indicates the chemistry (PI) followed by the number of arms (e.g., 4a)
[CH2CH2SiCl2(CH3)]4} (8-SiCl). and finally the weight-averaged molar mass of a single arm (e.g., 56 kg/
About 10% (4-arm PI) to 50% (8-arm PI) excess of the living PI, mol). The first three samples in the table are melts, the two following
end-capped with Bd, was used in order to force the linking reaction to samples are solutions prepared from sample PI 4a-103k. The solutions
completion. The linking reaction was monitored by comparison of the were prepared in the good solvent squalene (Sigma-Aldrich) using the
high molecular weight “star” peak and the low molecular weight “arm” cosolvent tetrahydrofuran (THF) (Sigma-Aldrich). After dissolution of
peak of size exclusion chromatography (SEC) traces. The excess living the polymer, the volatile cosolvent THF was removed by evaporation
polymer was terminated with degassed methanol. The polymers were under vacuum at room temperature for several days. The solutions also
extensively fractionated (solvent/non solvent: toluene/methanol) in contained a tiny amount of antioxidant to improve their stability. The
order to remove the excess arm material. An example (4-arm PI, Mw = molar mass between entanglements Me (in kg/mol) in Table 1 is
56k) is given in Figure 1. calculated from the plateau moduli of the melts or solutions as14
All intermediates and final products were analyzed by SEC and Me(ϕ) = (ϕρR(T + 273.2))/G0N(ϕ), with G0N(ϕ) the plateau modulus
nuclear magnetic resonance (NMR). SEC-experiments were per- of the melt or solution (see discussion below), ϕ the volume fraction
formed at 25 °C with a Waters model 510 pump, a Waters model 410 of polymer in the solution, ρ the mass density of the polymer (in kg/
differential refractometer, and three Styragel columns having a porosity m3; at 20 °C, ρ = 910 kg/m3),41 R the universal gas constant, and T
range from 103 to 106 Å. The carrier solvent was a mixture of the temperature (in °C). In addition to the weight-averaged molar
chloroform/triethylamine (95/5, v/v) at a flow rate of 1.0 mL/min. mass of an arm, Ma, Table 1 lists the weight-averaged total molar mass
Polystyrene standards were used for calibration, the Mn was obtained of the star, MTOT, the number of entanglements per arm Za = Ma/Me
after applying a correction coefficient of 1.6 for PI. The latter was and the number of arms f = MTOT/Ma. Note that MTOT was

5704 dx.doi.org/10.1021/ma400662b | Macromolecules 2013, 46, 5702−5713

Macromolecules Article

Figure 2. (a) Master curves of the dynamic moduli G′ and G″ as a function of angular frequency ω for the melts at 20 °C; PI 4a-56k (G′, ○; G″, ▽),
PI 8a-56k (G′, blue ◇; G″, blue □), and PI 4a-103k (G′, red ○; G″, red ▽ for oscillatory data; and G′, red −; G″, red -- for creep converted data).
(b) Master curves for the solutions at 20 °C; PI 4a-103k (G′, red ○; G″, red ▽ for oscillatory data; and G′, −; G″, -- for creep converted data), PI 4a-
103k 86% (G′, green ○; G″, green ▽), and PI 4a-103k 62% (G′, gray ○; G″, gray ▽). (c) Horizontal aT and vertical bT time−temperature shift
factors as a function of temperature T at a reference temperature of 20 °C used to obtain the master curves; PI 4a-56k (aT ○; bT, ●), PI 8a-56k (aT,
blue △; bT, blue ▲), PI 4a-103k (aT, red □; bT, red ■), PI 4a-103k 86% (aT, green ◇; bT, green ◆), and PI 4a-103k 62% (aT, gray ▽; bT, gray ▼).
The three full lines are fits with the WLF equation of the horizontal shift factors of the melts (black), the PI 4a-103k 86% (green), and the PI 4a-
103k 62% (gray).

determined either from the linear prepolymer molar mass and the star ±0.01 °C was achieved with a Peltier system, under nitrogen
functionality, or by direct measurement using static light scattering. atmosphere. However, the actual overall temperature control in the
The results were identical within 2%. sample was not as good (closer to ±0.1 °C, which was still acceptable)
II.2. Methods. II.2.1. Rheological Measurements. All linear due to air flow variations in the top part of the Peltier. Linearity of the
oscillatory and nonlinear rheological measurements have been creep measurements was ensured by performing tests at different
performed on a strain-controlled ARES-rheometer (TA Instruments, stresses. Converting creep compliance versus time curves to the
USA) equipped with a force rebalance transducer (2KFRTN1). The moduli as a function of frequency is well-known to be problematic as it
linear measurements were always performed in a parallel plate is a mathematically ill-conditioned problem. There are several ways to
geometry of 8 mm diameter with a temperature control of ±0.1 °C, do this, each with their own peculiarities,42−44 and here we have
and in a nitrogen environment to reduce the risk of degradation followed the approach of Pasquino et al.44
(achieved via an air/nitrogen convection oven and a liquid nitrogen II.2.2. The Cone Partitioned-Plate (CPP). The study of nonlinear
Dewar). A temperature range of −60 to +80 °C was used, ensuring flow properties of polymer melts with a rotational rheometer is
that the sample remained stable, and subsequently the data were complicated by the occurrence of flow instabilities and other artifacts.
shifted via time−temperature superposition. Details of the procedures Of particular concern for melts (with non-negligible second normal
for the linear measurements and the treatment of the data to obtain force) is edge fracture, which is shown to be directly associated with
master curves can be found in previous publications.8,10,20,37 The shear banding, i.e., nonlinear velocity profile in the melt.45,46 Following
nonlinear step-rate measurements were performed in a special cone the idea of Meissner et al.47 and the subsequent extensive
partitioned-plate geometry, briefly motivated and discussed in the investigations of Schweizer et al.45,48−50 and Li and Wang,18 we
following section with reference to an earlier paper.37 Also the constructed a so-called cone partitioned-plate (CPP) with temperature
nonlinear measurements were performed in a controlled environment control for the strain-controlled ARES rheometer.37 The CPP allows
(under nitrogen) with a temperature control of ±0.1 °C. Finally, long- us to perform nonlinear flow measurements unaffected by edge
time creep measurements were performed for the melt with the fracture up to higher strains and rates as compared to a regular cone
highest molar mass of the arms (PI 4a-103k) to extend the accessed and plate geometry (eventually, also for the CPP setup, edge fracture
frequency range to lower values without the need to use high takes over, the effects are simply delayed). The technical realization
temperatures at which the polymer degrades. The creep measurements and performance of our simple CPP on linear, monodisperse
have been performed with the Physica MCR 501 (Anton Paar, entangled polymers (PI and PS) was reported recently37 and the
Austria) stress-controlled rheometer using a homemade parallel plate results for the start-up viscosity η+ were found to be identical to the
geometry with diameter 8 mm as top plate. Temperature control of results of Schweizer et al.50 on a similar sample. The CPP has also

5705 dx.doi.org/10.1021/ma400662b | Macromolecules 2013, 46, 5702−5713

Macromolecules Article

Table 2. Viscoelastic Properties of the Investigated Polymers at TREF = 20 °Ca

sample code G0N × 10−5 [Pa] η0 × 10−5 [Pa·s] JS0 × 105 [Pa−1] τ0 [s] τCross [s] S.W. τR × 103 [s]
PI 4a-56k 4.4 11.5 ± 0.1 2.0 ± 0.1 23 ± 1 8.3 8.8 2.5
PI 8a-56k 4.4 12.0 ± 0.1 2.4 ± 0.1 30 ± 2 8.6 11 2.5
PI 4a-103k 4.4 2360 ± 50 10000b 4000 8.5
PI 4a-103k 86% 3.1 110 ± 9 4.9 ± 0.6 540 ± 80 200 15 1.5
PI 4a-103k 62% 1.4 2.0 ± 0.2 7±1 14 ± 3 4.6 10 0.21
a
See text for definitions of the quantities. bOrder of magnitude estimation of the terminal relaxation time for sample PI 4a-103k based on the roughly
constant ratio of order 3 between τCross and τ0 as found for the other samples.

been used to study the nonlinear shear flow behavior of model combs mass of the arms. Both observations are in agreement with the
with relatively short arms.22 In the present work we use the CPP setup literature.2,14,24,25
to investigate the model stars in nonlinear shear deformations. Table 2 lists an overview of the viscoelastic properties of the
Specifically, the CPP was composed of a standard 25 mm stainless stars as obtained from the linear data in Figure 2. The values for
steel cone with an angle of 0.1 rad, an Invar parallel plate of radius
the plateau moduli G0N(ϕ) are taken as value of the elastic
Rplate = 3 mm and an appropriately sized partitioned-plate part made of
stainless steel, forming a gap of 0.15 mm. For the measurements, the modulus G′ at the minimum of the phase angle tan(δ) = G″/
sample overfilled the inner plate by about 80% in diameter and the G′. Note that the volume fraction of the polymer is nearly
radius used for measuring the torque was Rplate. This level of overfill identical to its mass fraction here, due to the similarity in
did not substantially affect the linear viscoelastic data, conforming to density of polyisoprene and squalene. The relative error on the
earlier findings.37 concentration introduced in this way is in fact smaller than 2%.
The data for the plateau modulus for the solutions obeys the
III. EXPERIMENTAL RESULTS usual scaling for good solvents14,52 G0N(ϕ) = G0N(1)ϕ2.3. The
zero-shear viscosity η0 = limω→0(G″/ω) and the zero-shear
III.1. Linear Viscoelastic Properties. The master curves recoverable compliance JS0 = (1/η02) limω→0(G′/ω2) are
for the dynamic moduli G′ and G″ as a function of angular calculated from the linear oscillatory data. Their values are
frequency ω for the stars are shown in Figure 2, parts a and b, obtained as averages and standard deviations from three
at a reference temperature of 20 °C. Figure 2a shows the master different estimations. First, (G″/ω) and (G′/ω2) are calculated,
curves for the three melts and Figure 2b shows those for the and then the zero-shear limits of these two functions are
two solutions and their parent melt. For the star with the approximated by three different methods: simply taking the
longest arms (PI 4a-103k) and hence the slowest relaxation, we values at the lowest reliable frequency, fits with a Carreau-type
could not reach the terminal relaxation regime at the highest model, X/XCarreau,0 = [1 + (τCarreau,0ω)2](nCarreau−1)/2 and a Cross-
temperature of 80 °C, and hence we choose to perform long- type model X/XCross,0 = 1/(1 + (τCrossω)nCross) with X the
time creep measurements in the linear regime and then convert function of interest (being (G″/ω) or (G′/ω2)), XCross,0 and
the obtained creep compliance to dynamic moduli as explained X Carreau,0 the fitted zero-shear limits for the function,
before. The moduli as obtained via the conversion of the creep τCross,τCarreau, nCross, and nCarreau, further fitting parameters, and
compliance are shown as full light gray lines in Figures 2a and ω the angular frequency.53 As the terminal regions were
2b. reached in most cases, all three estimates were rather close and
The horizontal and vertical shift factors used to obtain the error was limited. Only for the PI 4a-103k, we have not been
master curves are shown in Figure 1c. The vertical shift factors able to get a reliable value for the recoverable compliance. The
are determined from the change of thermal tension and density longest relaxation time τ0 listed in Table 2 was calculated as14 τ0
with temperature bT = (ρ(TREF)(TREF + 273.2))/(ρ(T)(T + = JS0η0. It is used further on to define a characteristic
273.2)), with T the temperature (in °C) and with the Weissenberg number Wi0 = τ0γ̇, with γ̇ the shear rate, to
temperature dependence of the density ρ (in g/cm3):41 ρ(T) enable comparisons between different polymers and quantify
= 0.918−5.34 × 10−4(T + 273.2) − 4.701 × 10−8(T + 273.2)2. deviations from linear behavior. The inverse of the terminal
The vertical shift factors are assumed to be identical for the crossover frequency τCross is shown as well. It is less prone to
melts and solutions, although for the solutions the use of the experimental error but its physical interpretation is less clear.
density shift is more obscure. The horizontal shift factors result The ratio between the two relaxation times is roughly constant
from the least-squares fit as done by the Orchestrator software. over the different samples (3.0 ± 0.4). Since we have not been
The horizontal shift factors can be fitted with the WLF- able to get a reliable value for the recoverable compliance for
equation:51 log(aT) = (−C1(T − TREF))/(C2 + T − TREF), and sample PI 4a-103k, we provide an order of magnitude
with TREF = 20 °C the following parameters result for the melts: estimation for the terminal relaxation time for sample PI 4a-
C1 = 5.3 and C2 = 138.1 °C. The respective values at the glass 103k based on the mentioned ratio between τCross and τ0. Next
transition temperature of PI (Tg = −65 °C) are C1g = 6.7 and column in Table 2 lists the spectral width (S.W.) defined as14
C2g = 53.1 °C. For the PI 4a-103k 86% solution, we obtain C1 = J0SG0N. It is a measure of the breadth of the relaxation spectrum.
4.6 and C2 = 134.8 °C, and for the PI 4a-103k 62% solution, we And finally, the last column lists the Rouse time τR for the stars
obtain C1 = 4.3 and C2 = 151.7 °C. using the relaxation time of an entanglement τE from Table 3
By comparing the behavior of stars PI 4a-56k and PI 8a-56k (see further) together with the number of entanglements of an
in Figure 2a, we confirm that the linear relaxation of the stars is arm Za from Table 1 in the usual way14 τR = τE(Za)2. The values
independent of the number of arms (the data are for τR are in all cases below 10−2 s, i.e., orders of magnitude
indistinguishable). On the other hand, comparison of the below the terminal relaxation times τ0.
response of stars PI 4a-56k and PI 4a-103k in Figure 2a shows III.2. Modeling of the Master Curves with the Branch-
that the relaxation is clearly strongly dependent on the molar on-Branch Model (BOB). A further critical test of the sample
5706 dx.doi.org/10.1021/ma400662b | Macromolecules 2013, 46, 5702−5713
Macromolecules Article

Table 3. Parameters Used in the Simulations with the BOB certain factor as indicated in the caption. The black lines in
Model Figure 3 are the predictions of the BOB model.1,4,54 The
parameters used in the BOB model (at a reference temperature
sample code τE × 105 [s] Ma [kg/mol] ϕ
of 20 °C) are a value of 0.43 MPa for the plateau modulus GN0
PI 4a-56k 2.0 56 1 and 4.1 kg/mol for the molar mass between entanglements Me.
PI 8a-56k 2.0 56 1 For the BOB model the molar mass between entanglements
PI 4a-103k 2.0 100 1 follows from the plateau modulus according to Me(ϕ) = (4/
PI 4a-103k 86% 0.54 100 0.86 5)(ϕρRT/G0N(ϕ)), i.e., including the 4/5 prefactor, unlike for
PI 4a-103k 62% 0.18 100 0.62 the values in Table 1. For the p2-parameter we used a value of
1/40 and for the dynamic dilution parameter α we used 1
quality and accuracy of the measurements can be obtained by (these are the commonly values in the BOB model).4 All
comparing the experimental results with predictions of one of samples are modeled as being perfectly monodisperse. For the
the well-developed tube models. Figure 3 depicts the solutions, the solvent is considered by artificially introducing
very small monodisperse linear polymers with molar mass of
410 g/mol, i.e., the molar mass of the solvent squalene. The
remaining parameters in the model are taken as shown in Table
3, with τE the relaxation time of an entanglement, Ma the molar
mass of an arm, and ϕ the volume fraction of polymer. Note
that the value of 20 μs for τE for the melts is identical to the
value used by Auhl et al.55 for linear polyisoprenes when
comparing at the same reference temperature. We consider the
obtained predictions very satisfactory, in agreement with the
quality of the predictions in the literature for stars.4 The
plateaus of the predictions are somewhat below the plateaus for
the experimental data in all cases. This is a common problem of
the tube-based modeling.4,56 Finally, note that for the polymer
with the largest molar mass of the arms of 103 kg/mol, the
value used in the model is slightly lower (100 kg/mol) and still
Figure 3. Master curves of the dynamic moduli G′ and G″ as a
function of angular frequency ω for the melts at 20 °C compared with
the model predictions display a somewhat slower relaxation
predictions of the BOB model. PI 4a-56k, vertically shifted ×100 than the experimental data, and this for the melt as well as for
(experimental data, ○; BOB prediction, full black line). PI 8a-56k, the two solutions. Moreover, it could also be interesting to
vertically shifted ×100 (experimental data, blue △; BOB prediction, examine the possibility of some physics missing in the BoB
full black line). PI 4a-103k, vertically shifted ×10 (experimental data, model as the source of the slight disagreement between
red □ for oscillatory data and red line for creep converted data; BOB predictions and experimental data. Overall, however, the
prediction, dashed black line). PI 4a-103k 86% (experimental data, agreement is satisfactory and such subtle issues could be
green ◇; BOB prediction, dashed-dot black line). PI 4a-103k 62% resolved with more detailed characterization of the stars and
(experimental data, gray ▽; BOB prediction, dotted black line). further analysis, which are beyond the scope of the present
contribution.
experimental master curves of the dynamic moduli G′ and G″ III.3. Nonlinear Start-up Shear Flow. In Figure 4, two
as a function of angular frequency ω for all the investigated examples are shown of start-up simple shear flow for samples PI
samples at 20 °C as in Figures 2a and 2b. To ensure a clear 4a-56k and PI 4a-103k. The transient start-up viscosity η+ is
presentation, some of the curves are shifted vertically with a depicted as a function of time t for different shear rates γ̇ . The

Figure 4. Transient start-up shear viscosity η+ as a function of time t for the PI 4a-56k (a) and PI 4a-103k (b) (Table 1). The linear viscoelastic lines
are shown as thick dark gray lines and are calculated from the linear data (Figure 1a). The shear rates are indicated in s−1 next to the lines as
“imposed rate at measurement temperature ... shear rate shifted to reference temperature of 20°C”. The results for the melt PI 4a-56k were obtained
at 20 °C directly, the results for the PI 4a-103k were obtained at 80 °C and shifted to 20 °C using the shift factors from the linear data (Figure 1c).

5707 dx.doi.org/10.1021/ma400662b | Macromolecules 2013, 46, 5702−5713

Macromolecules Article

Figure 5. Steady viscosity scaled with zero-shear viscosity ηSTEADY/η0 (a), maximum peak-viscosity scaled with steady viscosity ηMAX/ηSTEADY (b),
strain where the maximum occurs γMAX (c), and strain of the peak-broadness γPEAK BROADNESS (d) as a function of the characteristic Weissenberg
number Wi0 for all samples. The black ○, blue △, red □, green ◇, and gray ▽ symbols are for the nonlinear data of the PI 4a-56k, PI 8a-56k, PI 4a-
103k, PI 4a-103k 86%, and PI 4a-103k 62%, respectively. In part a, the full black, blue, red, and dashed green and dark gray lines represent the
respective complex viscosities (scaled with zero-shear viscosity) as a function of frequency (multiplied by τ0), as calculated from the linear data for
the PI 4a-56k, PI 8a-56k, PI 4a-103k, PI 4a-103k 86%, and PI 4a-103k 62%, respectively. Error bars are present but often obscured by the symbols.

linear viscoelastic line is shown as well (dark gray line). This responsible, then it is not unexpected that star arms do not
line is obtained by direct transformation of the master curves contribute to this undershoot in this range of Wi0. Apart from
(from Figure 1) by applying the Cox−Merz rule57 η(γ̇) = the undershoot, however, the generic findings of Figure 4
η*(ω)|ω=γ̇ in conjunction with the Gleissle rule58 η+(t) = discussed here are qualitatively similar to the behavior of
η(γ̇)|γ̇=1/t. The start-up data should follow the LVE-line for low entangled linear polymer melts and solutions,35,37,59,60 and
rates and strains. The data in Figure 4a for star PI 4a-56k was entangled comb polymer melts and solutions.22 However the
obtained at 20 °C directly. The data for star PI 4a-103k (Figure scaling analysis of the data reveals specific features that signal
4b) was obtained at 80 °C and then shifted to 20 °C using the the star response.
shift factors from the linear data (Figure 2c) as indicated by the To be able to compare the different data sets, several
axes in Figure 4b. Start-up results for the other investigated parameters are extracted from the start-up transient viscosity
samples are not shown here as transient viscosity versus time as data: the steady-state shear viscosity ηSTEADY is calculated as
there are no apparent qualitative differences. The key average over the steady-state portion of the start-up curves. The
observation from Figure 4 is the progressive deviation of the standard deviation was always checked to make sure that
transient viscosity data from the LVE envelope as the shear rate reasonable values were obtained and the right portion of the
increases; this deviation is marked by a steady-state viscosity curve was used. The standard deviation on the steady-state
value below the LVE one. As the shear rate increases the time- shear viscosity ηSTEADY calculated in this way was found to be
dependent viscosity exhibits an overshoot and reaches a negligible compared to other sources of error. As second
maximum before eventually dropping to its steady-state value. quantity of interest, the maximum peak-viscosity ηMAX is
Unlike our previous works on combs22 and linear polymers,37 extracted by simply taking the maximum of the transient start-
here we did not observe an undershoot between the maximum up curve. As third quantity, we extract the position (as strain)
and steady portions of the start-up curves at any of the where the maximum peak-viscosity occurs γMAX, and finally, we
experimental investigated rates. We note that whereas under- quantify the broadness or width of the peaks γPEAK BROADNESS
shoots before steady-state have also been observed by Auhl et (as a strain) at half height, hence at a viscosity value of (ηMAX +
al.55 for linear monodisperse polymers and predicted by tube- ηSTEADY)/2. The latter two quantities are generally found to be
based theory for linear polymers,59,60 their physical origin more scattered. In many cases, it is more prone to error at low
remains elusive. If recoil (following stretch) is primarily rates due to the torque resolution, while at high rates, the start-
5708 dx.doi.org/10.1021/ma400662b | Macromolecules 2013, 46, 5702−5713
Macromolecules Article

Figure 6. Transient shut-down stress σ− divided by steady stress σSTEADY as a function of time t divided by terminal relaxation time τ0 for different
shear rates for the PI 4a-56k (a) and PI 4a-103k (b) stars at 20 °C. The shear rates are indicated in s−1 next to the lines as “imposed rate...
temperature shifted shear rate” as in Figure 4b. The results for the melt PI 4a-56k were obtained at 20 °C directly, the results for the PI 4a-103k were
obtained at 80 °C and shifted to 20 °C using the shift factors from the linear data. Note that the horizontal axes scales are not identical in the two
figures.

Figure 7. Initial relaxation rates 1/τ taken as average over the dimensionless time t/τ0 from 0.02 to 0.03 (a) and long-time relaxation rates 1/τ taken
as average over the dimensionless time t/τ0 from 0.9 to 1.1 (b), as a function of characteristic Weissenberg number Wi0. The black ○, blue △, red □,
green ◇, and gray ▽ symbols are for the nonlinear data of the PI 4a-56k, PI 8a-56k, PI 4a-103k, PI 4a-103k 86%, and PI 4a-103k 62%, respectively.
The error bars are the standard deviations of the least-squares fits to the data in the mentioned time range. The ● symbols in (a) are the respective
data obtained in ref 22 for entangled combs. The horizontal black line in (b) is the average value of 1.2 for the long-time relaxation rate.

up time of the motor, combined with the resolution of the master curve. This will be further discussed below in
standard data acquisition of the ARES rheometer increase conjunction with literature data.
experimental error. We define the error on the strains here Parts b−d of Figure 5 show the peak viscosity ηMAX
artificially by simply multiplying the shear rate with the normalized with steady viscosity ηSTEADY, the strain where the
maximum time resolution of the standard data acquisition of peak occurs γMAX, and the strain of the peak-broadness
the ARES (0.01 s). In this paper, all reported data on stars is γPEAK BROADNESS as a function of the characteristic Weissenberg
obtained at rates below 10 s−1. number Wi0. Also for these three quantities, the differences
The steady-state shear viscosity ηSTEADY scaled with zero- between the different samples are very small, especially when
shear viscosity η0 (Table 2) is shown as a function of the considering experimental error and the very small scale on the
characteristic Weissenberg number Wi0 for all the samples in y-axes of the three figures. If anything, in Figure 5, parts c and d,
Figure 5a. The symbols represent the data obtained from the one could discern a tiny systematic dependence on the number
nonlinear step-rates, the lines are obtained from the linear of entanglements of an arm for γMAX and γPEAK BROADNESS, with
oscillatory data by dividing the complex viscosity η+ by η0 and the overshoot becoming somewhat less broad and occurring at
plotting this quantity as a function of frequency ω multiplied by smaller strains as the number of entanglements of an arm
τ0, following the empirical Cox−Merz rule:57 η(γ̇) = η*(ω)|ω=γ̇. increases. On the other hand, the more entangled the arms, the
The Cox−Merz rule seems to be followed rather well in all broader their relaxation spectrum14,24 (see also section III.1
cases, with the nonlinear data being slightly below the linear above). However, small differences among few data sets and the
data at the higher Weissenberg numbers in all cases. Within experimental errors involved in the measurements preclude a
experimental error and for the rates considered here, all data for more detailed analysis. For linear monodisperse polymers, Doi
the two solutions and the three melts scale together on a single and Edwards61 predicted that ηMAX, when arising solely from
5709 dx.doi.org/10.1021/ma400662b | Macromolecules 2013, 46, 5702−5713
Macromolecules Article

orientation of the chains (without stretch), should occur at a shear viscosity data (normalized to the zero-shear viscosity),
strain of γMAX = 2.3. In experiments, one generally finds a value confirming the action of CCR which is the same for both
between 2 and 2.3 for γMAX.35,37,50 Here, for the stars, the values architectures. However, as Wi0 increases above 10, the rate of
for γMAX as reported in Figure 5c are around the expected value the combs remains clearly slower, in contrast to the much faster
for orientation and slightly below, for all samples. Indirectly, increase of the inverse shear viscosity, and hence rate and
this suggests that for all investigated shear rates and for all inverse viscosity do not coincide anymore neither for stars nor
samples, the stars are orienting without stretching at this range for combs. Apparently, in this regime there is a complex
of Weissenberg numbers. Note that, for stars the terminal interplay of dilution, friction due to branches and CCR, whose
relaxation time and the stretch time are widely separated (see elucidation needs more work in the future. Assuming that no
section IV for further discussion). significant stretching takes place in this regime (see Figure 5c
III.4. Relaxation upon Cessation of Steady-State Shear and section IV below), a tentative interpretation calls for a
Flow. The relaxation of stress from steady-state σ− at shear rate different action of CCR in the two architectures. Related to
γ was monitored as a function of time t since the flow was this, we expect that the orientation of the comb backbone
stopped (γ = 0 s−1 for t ≥ 0 s). Figure 6 shows typical data for (taking place at higher Wi0) has a slower relaxation (due to the
the transient shut-down stress σ− (divided by its steady-state lack of free ends), hence driving the comb initial relaxation
value σSTEADY at time t = 0) as a function of time t for different rates to lower values.
shear rates for the two stars PI 4a-56k and PI 4a-103k. We note We also looked at the long-time relaxation. This is more
that, qualitatively, the observations are identical to those of difficult to analyze since the relaxation spectra for the stars are
Menezes and Graessley62 for solutions of linear polymers in the very broad and determination of a truly constant relaxation rate
entangled regime and to those of Snijkers et al.22 for entangled is often ambiguous, mainly due to the torque resolution. We
comb polymer melts and solutions. First, the relaxation managed to perform the analysis for only two samples and
transients decrease monotonically with time for all rates. chose a dimensionless time region t/τ0 between 0.9 and 1.1 for
Second, the relaxation rate in the initial stages accelerates as the samples following the analysis for the combs.22 The results for
shear rate increases. And third, for long enough times after flow the long-time relaxation rates as a function of Wi0 are shown in
cessation, the relaxation rate becomes roughly independent of Figure 7b and suggest that, within error the relaxation rates are
shear rate. shear rate independent. An average value of 1.2 ± 0.8 is found
In order to compare the different samples, the initial for the long-time relaxation rates, hence only slightly above
relaxation rates for the stress decay function were calculated terminal, exactly as for the combs.22 In this respect, it is
by assuming a single exponential relaxation over a certain range encouraging that the relaxation data are consistent with the
of dimensionless times t/τ0 as σ−(t/τ0) = C exp(−t/τ0τ), with dynamic frequency sweep data of Figure 2 (section III.1).
C an arbitrary constant and τ the relaxation time over the time
region of interest. Therefore, the initial relaxation rate 1/τ IV. FURTHER DISCUSSION AND COMPARISON WITH
could be obtained by fitting a straight line to the plot ln(σ−/ THE LITERATURE
σSTEADY) vs t/τ0, with the only challenge the choice of the Key observations in the presented data set is the near identity
relevant dimensionless time range for the fit. To this end, some between the scaled quantities (in Figure 5) for the set of
considerations were crucial. Because of instrumental constraints polymer star samples with systematically varying number of
discussed earlier, fitting at times below 0.1 s was excluded. We arms, molar mass of the arms, and solvent content. Especially
chose a dimensionless time region t/τ0 between 0.02 and 0.03 the observation of a roughly constant strain at peak viscosity
for all samples, identical to the dimensionless time range we γMAX with a value of about 2 (Figure 5c) is interesting as it
analyzed for the combs.22 Figure 7a depicts the results for these indicates that, in the investigated range of rates, we are not
“initial” relaxation rates as a function of Wi0 for the investigated stretching the polymers, but just orienting them, even though
star samples with the error bars coming directly from the least- we reach high characteristic Weissenberg numbers. As shown in
squares fits. We observe that the initial relaxation rate increases the last column of Table 2, the values for the Rouse time τR for
with the characteristic Weissenberg number Wi0. All data fall the stars are in all cases below 10−2 s, i.e., orders of magnitude
onto a single master curve within experimental error, which below the terminal relaxation times τ0 (Table 2). In terms of
suggests that regardless of the star functionality and molar characteristic Weissenberg number, this leads to an order of
masses of the arms, the orientation of the arm relaxes with a magnitude estimation for the range of values needed to stretch
rate proportional to the inverse terminal relaxation time τ0 and the stars in all cases (Wi0 > 104). In the performed experiments
hence it randomizes its orientation mainly via the usual linear edge fracture sets in at Wi0 at least an order of magnitude
relaxation mechanisms (arm retraction and constraint release), lower, hence we could not reach the required rates for stretch.
qualitatively identical to the observations for the combs.22 It Note that this naivë analysis is in full agreement with the work
appears that the initial relaxation rate increases with Wi0 of Ye and Sridhar36 on extensional flow of star solutions
because the initial steady-stress state of the entangled polymer discussed in the Introduction.
at the relaxation experiment contains less entanglements A comparison with the data of Menezes and Graessley,35
compared to that at the beginning of the start-up experiment which is to the best of our knowledge the only other data set
at constant shear rate, due to the action of CCR.22 At higher available with start-up simple shear data of star polymers, shows
Wi0 the action of CCR is stronger and consequently the mainly the severe problems with edge fracture, which limited
relaxation rate becomes faster. We also compare this behavior their data to very low rates (Wi0 < 3). The data obtained with
with that of combs22 in Figure 7a. Note that, in that case, the symmetric polybutadiene 4-arm stars having arm molar mass of
comb response is dominated by the backbone due to short (i.e., 49.5 kg/mol, in solution at a concentration of 26.6% (i.e.,
very fast retracting) arms.22 Clearly, for Wi0 < 10 the fast between 4 and 5 entanglements per arm), is in excellent
relaxation rate of both architectures is nearly identical. agreement with the data presented here (Figure 5). As stated in
Moreover, in this regime, the data are identical to the inverse the Introduction, the focus of Menezes and Graessley was on
5710 dx.doi.org/10.1021/ma400662b | Macromolecules 2013, 46, 5702−5713
Macromolecules Article

obtaining reliable normal stress data through the use of the for these entangled stars, as they deviate from the −1 viscosity-
specially stiffened rheometer.62 shear rate slope.34 CCR is proven to work for entangled linear
In Figure 8, a comparison is made between the steady-state polymers,34,37,63 but for stars there are subtle issues such as
shear data of Figure 5a with those of Tezel et al.33 Tezel et dynamic dilution, and hence deviations are not a surprise.
In Figure 9, a comparison is made between the nonlinear
data of the stars as presented in Figure 5, and the nonlinear
start-up shear data of two linear polymers (black filled circles)
measured earlier.37 The two linear polymers are monodisperse
polymers of polyisoprene and polystyrene with molar masses of
60 kg/mol (Z ∼ 12) and 182 kg/mol (Z ∼ 11), respectively. As
reported in ref 37, we found a master curve scaling for the
nonlinear quantities reported in Figure 9 for the two linear
polymers due to their similarity in the number of entangle-
ments and hence we do not use distinct symbols for the two
polymers in Figure 9. In Figure 9, the quantities are plotted as a
function of the characteristic Weissenberg number Wi0 based
on the longest relaxation time τ0. The values for τ0 for the linear
polymers, calculated in the same way as for the stars described
earlier, are 0.16 s at 10 °C for the PI sample and 0.33s at 180
Figure 8. Nonlinear steady shear data obtained on the stars and shown
previously in Figure 5a compared to the steady shear data from Tezel °C for the PS. Note that the number of entanglements of the
et al.32 as a steady viscosity scaled with zero-shear viscosity ηSTEADY/η0 linear polymers Z is roughly identical to the number of
as a function of the characteristic Weissenberg number Wi0. The ○ entanglements of an arm Za as reported in Table 1 for the PI
symbols are a collection of all the nonlinear shear data from Figure 5a. 4a-56k, PI 8a-56k, and PI 4a-103k 62%. Figure 9a shows that,
The red ▲ and blue ■ symbols are for two of the samples from ref 33, despite the common onset of thinning at Wi0 ∼ 1, the linear
coded star 2 and star 6, respectively. The full red and full blue lines are polymers (black symbols) are stronger thinning and that their
the predictions of the adjusted MLD model for star 2 and star 6 transition to a (seemingly) terminal thinning slope is more
respectively; the dashed red and dashed blue lines are the predictions sudden, i.e., more narrow than for the stars, reflecting a broader
of the adjusted GLaMM model for star 2 and star 6, respectively. The
relaxation spectrum for the stars. Concerning the overshoot,
full black line has a slope of −1 (CCR prediction) and serves to guide
the eye. when looking at a fixed, relatively high Weissenberg number
(say Wi0 = 10), the linear polymers have a stronger overshoot
(Figure 9b), with a peak that is shifted toward larger strains
al.32,33 obtained the experimental data using two-color (Figure 9c) and which are somewhat broader (Figure 9d). This
birefringence in a Couette device and modeled the data with might reflect the fact that, in contrast to stars, for linear
an adjusted GLaMM59 or MLD63 models. The adjustments of polymers the stretch time τR is closer to the terminal relaxation
the two models relate to the fact that they turned the reptation time τ0. Actually, in terms of order of magnitude, the linear
mechanism off. Tezel et al.32,33 measured polybutadiene 4-arm polymers are around a stretch Weissenberg number (i.e., a
stars dissolved in low molar mass linear polybutadiene. The Weissenberg number based on the Rouse time) of 1 at the
data shown in Figure 8 is for the samples with coding Star 2 (in highest investigated rates,37 whereas for the stars considered
dark gray; Ma = 140 kg/mol, 13 wt %, yielding Za = 10) and here, we are in all cases still almost two decades too low to
Star 6 (in light gray; Ma = 130 kg/mol, 12 wt %, yielding Za = reach a stretch Weissenberg number of 1.36 Another
8.6).33 The data shown in Figure 8 are the two most extreme comparison concerns the occurrence of a local minimum, i.e.,
data sets from Tezel et al.,33 as in a plot of ηSTEADY/η0 versus undershoot, in the start-up viscosity curves between the
characteristic Weissenberg number Wi0 the other data sets fall maximum and the steady-state portions. As already mentioned,
in between these two. This comparison is relative as the for the linear polymers we observed a very mild undershoot at
experimental data and both models are in all cases set to start at the highest few rates,37 which is absent for the stars in all cases.
viscosity ratio ηSTEADY/η0 = 1 (by tuning the η0-values Possibly, also the occurrence of the undershoot relates to
individually, which have a rather large spread).33 However, stretch: The linear polymers are mildly stretching, while the
such a representation helps performing a relative comparison of stars are still far from stretching.
different stars. Furthermore, the longest relaxation times are the
ones reported in the article, which are calculated differently
V. CONCLUSIONS
from those in Table 2.33 Nevertheless, several observations can
be made: (i) Optical measurements (based on the validity of In this work, we have presented a detailed experimental
the stress-optical rule) appear to be less accurate and lead to a investigation of the linear and nonlinear rheology of well-
significantly larger error on the experimental data as compared defined symmetric entangled polymer stars of low functionality
to mechanical data obtained with the CPP (data are much more with varying number of arms, molar mass of the arms and
scattered in Figure.8). (ii) A master curve scaling was not solvent content. We have focused in particular on the response
obtained, neither for the data, nor for the two models. (iii) The of the stars in simple shear, during start-up and relaxation upon
optical data and the models appear to exhibit somewhat flow cessation. To reduce experimental artifacts associated with
stronger thinning; i.e., they are less broad in their spectrum edge fracture (primarily) and wall slip, which are unavoidable in
than the data from the CPP. Slightly larger polydispersity (for nonlinear shear rheology, we have utilized a homemade cone-
star 2 of ref 33) and the approximate way in which CCR was partitioned plate fixture which was successfully implemented in
modeled are the most likely reasons for this.32,33 (iv) The recent studies with entangled linear and comb polymers. The
experimental data from CPP suggests that CCR may not hold following conclusions can be drawn:
5711 dx.doi.org/10.1021/ma400662b | Macromolecules 2013, 46, 5702−5713
Macromolecules Article

Figure 9. Steady viscosity scaled with zero-shear viscosity ηSTEADY/η0 (a), maximum peak-viscosity scaled with steady viscosity ηMAX/ηSTEADY (b),
strain where the maximum occurs γMAX (c), and strain of the peak-broadness γPEAK BROADNESS (d) as a function of the characteristic Weissenberg
number Wi0 for all the star samples and the linear, monodisperse polymers from ref 37. The ● symbols are for the linear, monodisperse polymers
from ref 37. (PI 60k and PS 182k), all other symbols are for the star samples (legend as in Figure 5). The black lines in part a are drawn to guide the
eye. Error bars are present but often obscured by the symbols.

(1) For the star polymers considered here with up to about orientation compared to that of arms, apparently due to
20 entanglements per arm, reliable nonlinear data could the lack of free ends.
be obtained with the CPP for characteristic Weissenberg (6) The experimental data set presented here, i.e., linear
numbers (based on the terminal time) below 300. viscoelasticity, stress start-up, and relaxation is self-
(2) During start-up simple shear, the stress overshoot consistent and may be useful toward refining models for
occurred nearly always at a strain approaching a value the nonlinear response of entangled star polymers.

■
of 2, suggesting that in the investigated shear regime the
star arms orient but do not stretch. AUTHOR INFORMATION
(3) The validity of the empirical Cox−Merx rule was
confirmed, within experimental error. On the other Corresponding Author
hand, the steady shear viscosity data exhibit a slope *E-mail: (D.V.) [email protected]. (N.H.) nikolaos.
smaller than the CCR slope of −1 in the investigated [email protected].
range of rates. Notes
(4) The broadness of the stress overshoot increased with The authors declare no competing financial interest.
Weissenberg number. A relation with the broad linear
viscoelastic relaxation spectrum of the stars is tempting,
albeit not yet quantified.
■ ACKNOWLEDGMENTS
We thank R. Pasquino for help with the creep analysis and
(5) The initial stress relaxation rate upon cessation of steady helpful discussions, as well as the reviewers of this work for
shear flow, reflecting the initial loss of entanglements due their very contructive comments. Partial support from the EU
to the action of CCR in steady shear flow prior to (FP7 ITN DYNACOP Grant 214627) is gratefully acknowl-
relaxation, increased with Weissenberg number. More- edged.
over, when compared against the relevant rate for comb
polymers, the latter was clearly slower at Weissenberg
numbers exceeding a value of about 10. This could be
■ REFERENCES
(1) Read, D. J.; Auhl, D.; Das, C.; den Doelder, J.; Kapnistos, M.;
suggestive of a different action of CCR in stars and Vittorias, I.; McLeish, T. C. B. Science 2011, 333, 1871−1874.
combs and a related slower relaxation of comb backbone (2) McLeish, T. C. B. Adv. Phys. 2002, 51, 1379−1527.

5712 dx.doi.org/10.1021/ma400662b | Macromolecules 2013, 46, 5702−5713

Macromolecules Article

(3) Frischknecht, A. L.; Milner, S. T.; Pryke, A.; Young, R. N.; (39) Hadjichristidis, N.; Iatrou, H.; Pispas, S.; Pitsikalis, M. J. Polym.
Hawkins, R.; McLeish, T. C. B. Macromolecules 2002, 35, 4801−4820. Sci., Part A 2000, 38, 3211−3234.
(4) Das, C.; Inkson, N. J.; Read, D. J.; Kelmanson, M. A.; McLeish, T. (40) Mori, S.; Barth, H. G. Size exclusion chromatography; Springer
C. B. J. Rheol. 2006, 50, 207−234. Verlag: Berlin, 1999.
(5) McLeish, T. C. B.; Allgaier, J.; Bick, D. K.; Bishko, G.; Biswas, P.; (41) Zoller, P.; Walsh, D. Standard Pressure-Volume-Temperature Data
Blackwell, R.; Blottiere, B.; Clarke, N.; Gibbs, B.; Groves, D. J.; Hakiki, for Polymers; Technomic Publishing Co.: New York, 1995.
A.; Heenan, R. K.; Johnson, J. M.; Kant, R.; Read, D. J.; Young, R. N. (42) Schwarzl, F. R. Rheol. Acta 1969, 8, 6−17.
Macromolecules 1999, 32, 6734−6758. (43) Evans, R. M. L.; Tassieri, M.; Auhl, D.; Waigh, T. A. Phys. Rev. E
(6) Chen, X.; Shahinur Rahman, M.; Lee, H.; Mays, J.; Chang, T.; 2009, 80, 012501−1.
Larson, R. G. Macromolecules 2011, 44, 7799−7809. (44) Pasquino, R.; Zhang, B.; Sigel, R.; Yu, H.; Ottiger, M.; Bertran,
(7) Daniels, D. R.; McLeish, T. C. B.; Crosby, B. J.; Young, R. N.; O.; Aleman, C.; Schlüter, A. D.; Vlassopoulos, D. Macromolecules 2012,
Fernyhough, C. M. Macromolecules 2001, 34, 7025−7033. 45, 8813−8823.
(8) Kapnistos, M.; Vlassopoulos, D.; Roovers, J.; Leal, L. G. (45) Schweizer, T.; Stöckli, M. J. Rheol. 2008, 52, 713−727.
Macromolecules 2005, 38, 7852−7862. (46) Skorski, S.; Olmsted, P. D. J. Rheol. 2011, 55, 1219−1246.
(9) Kirkwood, K. M.; Leal, L. G.; Vlassopoulos, D.; Driva, P.; (47) Meissner, J.; Garbella, R. W.; Hostettler, J. J. Rheol. 1981, 33,
843−864.
Hadjichristidis, N. Macromolecules 2009, 42, 9592−9608.
(10) Kapnistos, M.; Koutalas, G.; Hadjichristidis, N.; Roovers, J.; (48) Schweizer, T. Rheol. Acta 2002, 41, 337−344.
(49) Schweizer, T. J. Rheol. 2003, 47, 1071−1085.
Lohse, D.; Vlassopoulos, D. Rheol. Acta 2006, 46, 273−286.
(50) Schweizer, T.; van Meerveld, J.; Ö ttinger, H. C. J. Rheol. 2004,
(11) Blackwell, R. J.; Harlen, O. G.; McLeish, T. C. B. Macromolecules
48, 1345−1363.
2001, 34, 2579−2596. (51) Ferry, J. D. Viscoelastic Properties of Polymers, 3rd ed.; Wiley:
(12) van Ruymbeke, E.; Orfanou, K.; Kapnistos, M.; Iatrou, H.;
New York, 1980.
Pitsikalis, M.; Hadjichristidis, N.; Lohse, D. J.; Vlassopoulos, D. (52) Colby, R. H.; Rubinstein, M. Macromolecules 1990, 23, 2753−
Macromolecules 2007, 40, 5941−5952. 2757.
(13) van Ruymbeke, E.; Muliawan, E. B.; Hatzikiriakos, S. G.; (53) Macosko, C. W., Rheology; VCH: New York, 1994.
Watanabe, T.; Hirao, A.; Vlassopoulos, D. J. Rheol. 2010, 54, 643−662. (54) den Doelder, J.; Das, C.; Read, D. J. Rheol. Acta 2011, 50, 469−
(14) Graessley, W. W. Polymeric Liquids & Networks: Structure & 484.
Rheology; Garland Science: New York, 2008. (55) Auhl, D.; Ramirez, J.; Likhtman, A. E.; Chambon, P.;
(15) Roovers, J. Polymer 1979, 20, 843−849. Fernyhough, C. J. Rheol. 2008, 52, 801−835.
(16) Lee, J. H.; Driva, P.; Hadjichristidis, N.; Wright, P. J.; Rucker, S. (56) Likhtman, A. E.; McLeish, T. C. B. Macromolecules 2002, 35,
P.; Lohse, D. J. Macromolecules 2009, 42, 1392−1399. 6332−6343.
(17) Larson, R. G. Rheol. Acta 1992, 31, 213−263. (57) Cox, W. P.; Merz, E. H. J. Polym. Sci. 1958, 28, 619−622.
(18) Li, X.; Wang, S.-Q. Rheol. Acta 2010, 49, 985−991. (58) Gleissle, W. Rheology; Astarita, G., Marrucci, G., Nicolais, L.,
(19) Sui, C.; McKenna, G. B. Rheol. Acta 2007, 46, 877−888. Eds; Proceedings of the 8th International Congress on Rheology;
(20) Snijkers, F.; van Ruymbeke, E.; Kim, P.; Lee, H.; Nikopoulou, Plenum: New York, 1980; Vol. 2, Fluids.
A.; Chang, T.; Hadjichristidis, N.; Pathak, J.; Vlassopoulos, D. (59) Graham, R. S.; Likhtman, A. E.; McLeish, T. C. B.; Milner, S. T.
Macromolecules 2011, 44, 8631−8643. J. Rheol. 2003, 47, 1171−1200.
(21) Li, S. W.; Park, H. E.; Dealy, J. M. J. Rheol. 2011, 55, 1341− (60) Pearson, D.; Herbolzheimer, E.; Grizzuti, N.; Marrucci, G. J.
1373. Polym. Sci., Part B 1991, 29, 1589−1597.
(22) Snijkers, F.; Vlassopoulos, D.; Lee, H.; Yang, J.; Chang, T.; (61) Doi, M.; Edwards, S. F. J. Chem. Soc. Faraday Trans. II 1979, 75,
Driva, P.; Hadjichristidis, N. J. Rheol. 2013, 57, 1079−1100. 38−54.
(23) Ianniruberto, G.; Marrucci, G. Macromolecules 2013, 46, 267− (62) Menezes, E. V.; Graessley, W. W. Rheol. Acta 1980, 19, 38−50.
275. (63) Mead, D. W.; Larson, R. G.; Doi, M. Macromolecules 1998, 31,
(24) Milner, S. T.; McLeish, T. C. B. Macromolecules 1997, 30, 7895−7914.
2159−2166.
(25) Pakula, T.; Geyler, S.; Edling, T.; Boese, D. Rheol. Acta 1996, 35,
631−644.
(26) Vrentas, C. M.; Graessley, W. W. J. Rheol. 1982, 26, 359−371.
(27) Osaki, K.; Takatori, E.; Kurata, M.; Watanabe, H.; Yoshida, H.;
Kotakat, T. Macromolecules 1990, 23, 4392−4396.
(28) Vega, D. A.; Milner, S. T. J. Polym. Sci., Part B 2007, 45, 3117−
3136.
(29) Osaki, K. Rheol. Acta 1993, 32, 429−437.
(30) Kapnistos, M.; Kirkwood, K. M.; Ramirez, J.; Vlassopoulos, D.;
Leal, L. G. J. Rheol. 2009, 53, 1133−1153.
(31) Wang, S.-Q.; Ravindranath, S.; Boukany, P. E. Macromolecules
2011, 44, 183−190.
(32) Tezel, A. K.; Leal, L. G.; McLeish, T. C. B. Macromolecules 2005,
38, 1451−1455.
(33) Tezel, A. K.; Oberhauser, J. P.; Graham, R. S.; Jagannathan, K.;
McLeish, T. C. B.; Leal, L. G. J. Rheol. 2009, 53, 1193−1214.
(34) Marrucci, G. J. Non-Newtonian Fluid Mech. 1996, 62, 279−289.
(35) Menezes, E. V.; Graessley, W. W. J. Polym. Sci., Polym. Phys. Ed.
1982, 20, 1817−1833.
(36) Ye, X.; Sridhar, T. Macromolecules 2001, 34, 8270−8277.
(37) Snijkers, F.; Vlassopoulos, D. J. Rheol. 2011, 55, 1167−1186.
(38) Schweizer, T. How reliable is N2 of a polymer melt determined with
a cone-partitioned plate rheometer? Presented at the 6th Annual
European Rheology Conference, Göteborg, Sweden, April 2010

5713 dx.doi.org/10.1021/ma400662b | Macromolecules 2013, 46, 5702−5713