Polymer Testing 66 (2018) 137–145

Contents lists available at ScienceDirect

Polymer Testing
journal homepage: www.elsevier.com/locate/polytest

Test Method

A new double-slit rheometrical die for in-process characterization and T

extrusion of thermo-mechanically sensitive polymer systems
P.F. Teixeira, L.L. Ferrás, L. Hilliou∗, J.A. Covas∗∗
IPC/I3N, University of Minho, 4800-058 Guimarães, Portugal

A R T I C L E I N F O A B S T R A C T

Keywords: The twin-slit rheological die can be operated in such a way that the total pressure drop is maintained constant
Extrusion while the shear rate in the measuring slit is changed. The device is particularly useful when coupled to an
Slit die extruder and used to characterize materials that are sensitive to thermo-mechanical conditions. In the present
In-process rheometer work, the twin-slit die design was modified to a double-slit die where the measurement and extrusion channels
Polymer blend
have different geometries, and viscoelastic and morphological characterization can be performed. The flow
Nanocomposite
analysis developed for twin channels was generalized to accommodate for different channel lengths and cross
sections. The new set of equations was used to design the double-slit flow channel. Comparison of off-line with
in-process data validated both the double-slit die and the experimental methodology. The practical utility of the
new die was then demonstrated with the in-process rheological characterization of various polymer systems.

1. Introduction
As more sophisticated polymer materials and products need to be
developed within shorter time-to-market, fast response characterization
tools using small amounts of sample and capable of conveying data
relating rheological response, process-induced morphology and en-
gineering properties become increasingly necessary. A common ap-
proach is to use small-scale processing equipment coupled to on-line or
in-line measuring sensors/devices [1,2].
Coupling instrumented capillary or slit dies to extruders for in-line
measurements seems particularly promising. Not only shear viscosity
[3,4] and normal-stress differences become accessible [5–8], but mea-
surements are done at relatively higher shear rates (in excess of 10 s−1),
i.e., in the region of practical processing. Optical methods are non-in-
vasive and recognized as powerful tools for morphology analysis. Thus,
in-line light scattering studies of polymer blend morphology using a slit
die with an optical window coupled to an extruder started over a
decade ago [9–11]. For example, the authors coupled an extruder to a
modular rheo-optical slit die with rheometrical (viscosity and normal
stress differences) and optical capabilities (small angle light scattering
(SALS) or polarized optical microscopy (POM)); together with the re-
levant downstream equipment, they also produce extrudates for sub-
sequent appraisal of the engineering properties [5,12].
Coupling an extruder to a rheological die and measuring both the
resulting throughput and pressure drop will yield a single point of the
flow curve (viscosity vs. shear rate). To generate a range of shear rates,
either the screw speed (single screw extruder) or the feed rate (twin
screw extruder) must be changed. This action will modify the thermo-
mechanical history of the material in the extruder (residence time,
temperature, hydrodynamic stresses) and so may lead to changes in
homogeneity, thermal stability and/or morphology. At each shear rate,
a possibly distinct material could be characterized [13]. Rauwendaal
and Fernandez [4] observed that their in-process measurements con-
sistently yielded lower viscosity values compared to the results from
off-line capillary and cone-and-plate rheometry. The differences were
more pronounced at higher shear rates (i.e., higher screw speeds) and
attributed to the shorter residence times between extruder and slit die
allowing for less relaxation of the screw-induced effects. According to
Vergnes et al. [13], changes in upstream flow conditions could explain
the anomalous results (negative values of the power law index) ob-
tained by some authors for in-line food rheology [14]. Other authors
confirmed that the methodology had a small effect on the flow curves of
low-density polyethylene melts, but yielded a considerable error in the
case of foods [15,16] and other complex and structured fluids [17,18].
To circumvent this problem, various authors proposed slit die de-
signs enabling the acquisition of various points of the flow curve in a
single experiment. Pabedinskas et al. [19] used a slit die with constant
width and a linear vertical taper, so that the strain rate increased gra-
dually downstream (within a tenfold range). Assuming the flow of a
power-law fluid, from the measurement of pressures at three axial

∗
Corresponding author.
∗∗
Corresponding author.
E-mail addresses: loic@dep.uminho.pt (L. Hilliou), jcovas@dep.uminho.pt (J.A. Covas).

https://doi.org/10.1016/j.polymertesting.2018.01.013
Received 15 November 2017; Received in revised form 16 January 2018; Accepted 17 January 2018
Available online 03 February 2018
0142-9418/ © 2018 Elsevier Ltd. All rights reserved.
P.F. Teixeira et al.
locations of different heights, n and k could be extracted. Agreement
with data obtained from off-line capillary rheometry was very good for
low viscosity materials, but for higher viscosity samples the in-line
values were consistently lower. Because of the convergence of the
channel, the flow is not truly viscometric, and pressure differences will
reflect both viscous and elastic responses [20]. Thus, the usefulness of
this approach may be limited to low viscosity or little elastic fluids.
Horvat et al. [21] developed a multi-step slit die in which the flow
channel is divided into zones of different constant heights and uniform
width. Each zone contains two flush mounted pressure transducers at
distinct downstream locations. Experiments were reported for a three-
step configuration, generating three shear rates spanning one order of
magnitude. Due to the need to mount 6 pressure transducers, the total
length of the slit was quite considerable (370 mm). However, with only
two measurements being done per channel segment, the eventual effect
of pressure or viscous dissipation on viscosity is not readily identifiable.
Moreover, the width/height ratio of the channels was smaller than 10,
which is often considered as the threshold to guarantee the develop-
ment of simple 1D viscometric flow. Coates et al. [22] attached a flow
regulating valve to the exit of a slit. Although various configurations of
this accessory were tried, it was not evident that a viscometric flow was
attained. In practice, only low shear rates were tested (approximately 1
s−1), and agreement with off-line data was only moderate. Kalyon et al.
[23,24] developed an adjustable gap in-line rheometer consisting of a
slit die with a movable plate for continuous height adjustment. In order
to keep the flow rate constant, the device must be coupled to extruders
operated in starve-fed mode. Hochstein et al. [25] modified a multiple-
step slit die by mounting the lower half of the slit on a wedge that can
slide horizontally, and thus vary the gap. Also, a flow valve positioned
near to the die exit was used to control the total pressure drop. Apart
from the limitations already mentioned above, it is worth noting that
variations in the total pressure drop created by these solutions to gen-
erate different shear rates will modify the flow conditions upstream of
the die.
The design proposed by Springer et al. [26] aimed at inducing
variations of shear rate while maintaining constant the total pressure
drop. The entry flow channel is divided into two parallel identical slits,
one being used for the measurements (i.e., with at least two pressure
transducers). Flow distribution in the two slits is controlled by
138
Polymer Testing 66 (2018) 137–145
Fig. 1. Schematic representation of the twin channel
slit following the approach and nomenclature of
Vergnes et al. [13] (a). Left: top view, flow goes from
left to right; valves are shaded in grey; QT is the total
flow rate, Q1 and Q2 are the flow rates in channels 1
and 2, respectively; pe is the entrance pressure. Right:
side view of the two channels; valves move vertically;
(b) flow regions considered to derive the relationship
between QT and pe.
vertically moving valves located at the entrance of each channel. To
change the shear rate at the measuring slit, the corresponding valve is
moved simultaneously with the other one, but in opposite directions, in
such a way that the pressure at the entry channel remains constant.
Shear rates were varied over two orders of magnitude, but no direct
comparison with off-line data was apparently performed. Later, Vergnes
et al. [13] re-visited the concept and developed an analytical model for
a power law fluid relating the aperture of the two valves that forces
constant flow rate and constant entrance pressure. Della Valle et al.
[27,28] and Lach [29] successfully used this type of rheological die,
whose major drawback seems to be the waste of material during mea-
surements. Li et al. [15] used a modified design with perpendicular,
rather than parallel channels. Recently, Robin et al. [30] fitted a par-
allel slit die with adjustable slit heights (by means of the slide solution
developed by Hochstein et al. [25]). The valves were located close to
die exit, allegedly to avoid generating a high amount of shear and en-
ergy dissipation upstream of the measurement zone, as well as to avoid
premature gas bubble nucleation in the starchy melt. However, locating
valves near to discharging end of the die may jeopardize the visco-
metric nature of the flow, as seen above.
Due to its considerable advantages, namely maintaining the total
pressure drop while the shear rate in the measuring slit is changed, the
concept of the twin-slit rheological die is adopted here. However, both
the geometry and the construction were modified in order to improve
the practical utility and the range of measurements of the device. The
measuring slit was made suitable for rheo-optical characterization
[5,12], enabling the determination of both viscoelastic and morpholo-
gical characteristics. The second slit was used for conventional extru-
sion, but its width was made larger, so that standard tensile specimens
could be cut from the extrudate. Thus, the length and width of the two
slits will be distinct. In order to support the new design, the analysis of
Vergnes et al. [13] will be extended to non-identical double-slits.
The outline of the paper is as follows. First, Vergnes's analysis is
extended to non-identical double-slits. Then, the geometry of the flow
channels, construction of the die and experimental set-up are presented.
Next, the theoretical predictions are compared with the experimental
data. Finally, the new die is used to measure the viscoelastic behavior of
several polymer systems.
P.F. Teixeira et al. Polymer Testing 66 (2018) 137–145

2. Designing a new double-slit die Eq. (6) reduces to Eq. (5) for identical slits. To obtain a relationship
between h1 and h2 (maintaining QT and pe constant), the condition
2.1. Flow analysis when valve 1 is fully open (h1 = hc1) and valve 2 is closed (h2 = 0) is
used. In this event, Eq. (6) reduces to:
Fig. 1 shows a schematic (top and lateral views) of the twin channel 1/ n
w1 n ⎛ pe ⎞
slit analyzed by Vergnes et al. [13]. Flow progresses from left to right. QT = Q1 = ⎜ ⎟ hc1/1 n + 2.
At the entrance of each channel, a valve can be moved vertically to 2(2n + 1) ⎝ 2kL1 ⎠ (7)
create a given gap. Relevant geometrical parameters include the length Equalizing Eqs. (6) and (7) creates a general relationship between h1
of the valve, Lv, the length of the channel between the valve and the and h2:
channel exit, Lc (Lv + Lc = L), the height of the channel hc, and the
1/ n 1/ n
vertical gap created by the valve, h. Subscript 1 or 2 refers to each ⎛ ⎡ ⎤ ⎡ ⎤ ⎞
individual channel. The channel has constant width, w. The total flow
pe 1/ n
⎜w1 ⎢ hc11+ 2n ⎥ hc12+ 2n
⎛ ⎞ 1 + 2n
+ w2 ⎢ ⎥ ⎟
⎝ 2k ⎠ ⎢ ⎥ ⎢ hc 2 1 + 2n ⎥
rate, QT (sum of the flow rate in channels 1, Q1 and 2, Q2) and the
entrance pressure, pe, remain constant. The aim is to determine the
⎜⎜
⎝ ⎣
L
⎢ c1 +
h
h1
c1
( ) L v1 ⎥
⎦ ⎢
⎣
Lc2 + h2 ( )
L v2 ⎥
⎦
⎟⎟
⎠
1/ n
value of h2 when h1 is changed in order to induce a specific shear rate. ⎡ ⎤
The simplifying assumptions are: one dimensional (i.e., the ratio hc/w is p 1/ n L1
= w1 ⎛ e ⎞ hc1/1 n + 2 ⇔ ⎢
⎜ ⎟
1 + 2n
⎥
sufficiently small) steady isothermal flow obeying to the power law ⎝ 2kL1 ⎠ ⎢ ⎥
(η = kγ̇ n − 1 where k is the consistency, n is the power law index and γ̇ is
⎢
⎣
Lc1 + ( ) hc1
h1
L v1 ⎥
⎦
1/ n
the shear rate); pressure drop between the channel entry and the piston 1/ n + 2 ⎡ ⎤
valve is identical for the two channels; local flow perturbations (re- w h L1
+ 2 ⎛ c2 ⎞⎜ ⎟
⎢
1 + 2n
⎥ = 1.
w 1 ⎝ h c1 ⎠ ⎢ ⎥
circulation) in the vicinity of the valves are neglected.
Considering the two flow regions in the channel identified in Fig. 1
⎢
⎣
Lc 2 + ( )
hc2
h2
L v2 ⎥
⎦ (8)
(b) (flow in the gap – region a, flow in the slit – region b), the flow rate
can be obtained from the Navier-Stokes equations (together with the 2.2. Die geometry and construction
viscosity following a power-law):
1 + 2n −1/ n Eq. (8) can be used to evaluate the effect of having distinct in-
w1 n ⎛ pe − pf ⎞1/ n 1/ n + 2 ⎡ ⎛ hc ⎞
Q1 = hc ⎢ Lv ⎤
⎜
⎥ , ⎟
dividual channel dimensions on the relationship between h1 and h2 for
2(2n + 1) ⎝ 2k ⎠ ⎣ ⎝ h1 ⎠ ⎦ (1) different n (power law index) and Lv/L (length of the valve relative to
the total channel length). For the twin slit, Vergnes et al. [13] observed
for region a, and
that, when the ratio Lv/L = 0.1, the relative motion of the two valves is
1/ n not very sensitive to n when this is larger than 0.4. In turn, when
w1 n ⎛ pf − 0 ⎞
Q1 = ⎜ ⎟ hc1/ n + 2 [Lc ]−1/ n , n = 0.4 and Lv/L ≈ 0.2, a decrease of h1 is balanced by a quasi-pro-
2(2n + 1) ⎝ 2k ⎠ (2)
portional increase of h2. A linear relationship between the decrease of
for region b, where pf is the pressure at the exit of the valve. Since the h1 and the increase of h2, as well as independence of the pseudoplastic
flow rate is the same in both regions, equalizing these equations yields a nature of the melt, are desirable as they would obviously simplify the
relationship between pressure pe and pf : practical operation of the device. Fig. 2 correlates the required relative
Lc aperture of the two valves for different values of n (Fig. 2 (a)) and Lv/L
pf = pe . (Fig. 2 (b)) for channels with distinct cross-sections (hc1 = 0.8 mm,
Lc + (hc / h1)1 + 2nL v (3)
w1 = 10 mm; hc2 = 1.2 mm, w2 = 12 mm). Channels with equal
Inserting Eq. (3) into Eq. (2) produces the relationship between the (Lc1 = Lc2) and distinct (Lc1 > Lc2) length were also considered. Fig. 2
flow rate, Q1, and the entrance pressure, pe: (a) shows an almost linear relationship between the relative opening of
1 + 2n −1/ n the two valves, except when h1 approaches its fully open position (h1/
w1 n p 1/ n h
Q1 = ⎛ e ⎞ hc1/ n + 2 ⎡Lc + ⎛ c ⎞ Lv ⎤ ⎜ ⎟ . hc1 = 1). The slope of this correlation increases with the value of n. The
2(2n + 1) ⎝ 2k ⎠ ⎢ h ⎥
⎣ ⎝ 1⎠ ⎦ (4) higher the slope, the easier the regulation of the valves, as for each h1
the need of a precise regulation of h2 is minimized. Changing Lc1 re-
The flow rate in channel 2 is quantified by a similar expression
lative to Lc2 has little effect when n ≥ 0.4. Assuming n = 0.4, Fig. 2 (b)
(replace subscript 1 by 2 in Eq. (4)). The total flow rate is then
displays the effect of the Lv/L ratio on the relative aperture of the two
QT = Q1 + Q2 ,
valves. Lv/L was made to change either maintaining Lv1/L1 = Lv2/L2 or
w1 n p 1/ n ⎡ h
1 + 2n −1/ n
making Lv1/L1 < Lv2/L2. In the first case, the range of the approximately
QT = ⎛ e ⎞ hc1/ n + 2 ⎢ ⎡Lc + ⎛ c ⎞ L v⎤ ⎜ ⎟
linear response narrows with increasing Lv/L, but the slope remains
2(2n + 1) ⎝ 2k ⎠ ⎢ h ⎥
⎣⎣ ⎝ 1 ⎠ ⎦ approximately the same. Imposing different Lv/L for each channel did
−1/ n
1 + 2n
⎤ not trigger significant differences. Thus, low Lv/L ratios seem inter-
+⎡ ⎛ hc ⎞ L v⎤
⎢Lc + h2 ⎥, esting in practice.
⎜ ⎟
⎥
⎣ ⎝ ⎠ ⎦ ⎦ (5) Taking into consideration the relationships depicted in Fig. 2, the
as found by Vergnes et al. [13]. If the two slits have distinct geometries configuration and dimensions of the channels were defined and are
(Lc1 ≠ Lc2, w1 ≠ w2 and hc1 ≠ hc2) using the same reasoning as above presented in Fig. 3 (a). The measurement channel has a length
yields the following expression for QT: Lc1 = 75 mm, preceded by a valve with Lv1 = 12 mm (thus
L1 = 87 mm), and a width w1 = 10 mm. The height of this channel is
1/ n
⎛ ⎡ ⎤ hc1 = 0.8 mm (thus w1/hc1 > 10), with a maximum h1 = 0.8 mm. For a
n p 1/ n hc11+ 2n
QT = ⎛ e ⎞ ⎜w1 ⎢ 1 + 2n
⎥ given QT and pe, the generation of average shear rates within two orders
2(2n + 1) ⎝ 2k ⎠ ⎜ ⎢ L + hc1 ⎥
⎜ ⎢ c1
⎝ ⎣
h1
L v1 ⎥
⎦ ( ) of magnitude can be expected. The extrusion channel has a length
Lc2 = 40 mm, a valve also with Lv2 = 12 mm (thus L2 = 52 mm),
1/ n
⎡ ⎤ ⎞ hc2 = 1.2 mm (which matches the maximum value of h2) and
hc12+ 2n w2 = 12 mm.
+ w2 ⎢ 1 + 2n
⎥ ⎟,
⎢ ⎥
⎢
⎣
h
Lc 2 + hc2
2 ( ) L v2 ⎥
⎦
⎟⎟
⎠ (6)
The resulting modular die assembly is represented in Fig. 3 (b). The
main body is coupled to the extruder. The inlet circular channel is

139
P.F. Teixeira et al.
Fig. 2. Relationship between the aperture of the two valves h1 and h2 (relative to the
respective channel height, hc). The channels have different cross-sections (hc1 = 0.8 mm,
w1 = 10 mm; hc2 = 1.2 mm, w2 = 12 mm; Lv1 = Lv2); (a) effect of power law index, n; (b)
effect of Lv/L.
progressively converted into a slit, which is then divided into perpen-
dicular extrusion and measurement channels for easier manipulation of
the extrudates. The body contains the rectangular valves positioned at
the entrance of these channels. They travel vertically by means of
screws rotating in the appropriate direction (vertical displacement of
0.5 mm/rev). The readings from a pressure transducer (Pe) flush
mounted in the flow channel close to the die entrance are used to
guarantee that, on operating the valves, pe remains constant. Mea-
surement and extrusion modules are fixed to this main body. The
former was designed to perform rheological and/or rheo-optical ex-
periments and consists of two halves bolted together. The upper part
can accommodate two flush mounted pressure transducers located
sufficiently far from the entrance and exit, respectively. The lower part
can accept a pressure transducer mounted recessed directly opposite to
the above mentioned pressure transducer near to the exit. Thus, the
hole-pressure method can be used to access the normal stress differ-
ences [5,31]. Top and bottom optical sapphire windows (8 mm dia-
meter and 1 mm thickness) enable optical measurements. In-line rheo-
optical experiments were performed and validated elsewhere [12],
hence they will not be pursued here. The channel width (w1) and height
(hc1) can be changed inserting spacing slides between the two halves.
Spacing slides can also be used in the extrusion module to change the
thickness of the extrudate.
140
Polymer Testing 66 (2018) 137–145
Fig. 3. New double-slit rheological die. (a) Flow channel configuration and dimensions;
the arrows denote the range of vertical aperture of the valves; (b) Die construction: A -
main body, B - rheo-optical module with a replaceable lower lip to enable the use of
different hole geometries in the hole-pressure method. C - extrusion module. P1, P2, P3
and Pe are pressure transducers; V1 and V2 are screws controlling the vertical position of
the valves.
3. Experimental
3.1. Set-up
The new modular slit die can be coupled to prototype modular mini
single or co-rotating twin screw extruders (for more details, see Refs.
[32,33]). Both machines yield outputs ranging from a few tens to a few
hundred grams per hour, depending on screw speed or feed rate. A
mini-gear pump (MS0525-C-230-CCW-S from Xaloy Europe GmbH,
Germany) was inserted between extruder and die in order to minimize
pressure fluctuations arising from inherent extrusion instabilities. A
water cooling tank, haul-off and winder were available in case the ex-
trudate needs to be collected.
The die incorporates four melt pressure transducers (Fig. 3 (b)). One
Dynisco PT422A (0–3000 PSI) is used to monitor the entrance pressure
drop, pe. Two Dynisco PT422A (0–3000 PSI and 0–750 PSI) are flush
mounted to measure the pressure drop along the slit (P1-P2). A Dynisco
PT 435A (0–750 PSI) was mounted recessed for hole-pressure mea-
surements (P3). All transducers have a sensitivity of ± 0.5% and are
sensitive to variations in temperature of ± 0.005 MPa per ± 1 °C.
Each transducer was connected to a Dynisco 1390 strain gage indicator,
with analog retransmission output accuracy span of ± 0.2%. In turn,
these were connected to a data acquisition system DAQPad-6020E from
National Instruments and driven by custom-written LabVIEW ™ rou-
tines.
3.2. Materials
A medium density Polyethylene (mPE M3583 UV, from Total
Petrochemicals, France) with a melt flow index of 8 g/10min (190 °C/
2.16 kg), and a polystyrene (PS RXP 3002 Natural, from Resinex,
Belgium) with a melt flow index of 11 g/10min (200 °C/5.0 kg) were
P.F. Teixeira et al.
used for validating the new die. Several polymer systems were subse-
quently characterized: i) polycarbonate (PC RXP5001 from Resinex,
Belgium) with a melt flow index of 15 g/10min (300 °C/1.2 kg), ii)
physical blends of mPE and PS (mPE/PS 20/80, 50/50 and 80/20 wt
%/wt%), iii) a mPE/MWCNT nanocomposite (with 4 wt% of multiwall
carbon nanotubes NC 7000 from Nanocyl, Belgium). Blends and na-
nocomposite were pre-mixed prior to feeding the extruder.
3.3. Experimental procedure
The determination of a flow curve (viscosity vs. shear rate) is per-
formed at constant screw speed (single screw extruder) or feed rate
(twin screw extruder), each point being obtained with a given combi-
nation of h1,h2 (set by the valves). Each experiment starts with the
extrusion channel shut (h2 = 0 mm) and with the valve upstream of the
measurement channel fully open (h1 = hc1 = 0.8 mm). Thus, the first
point is obtained at the highest shear rate. Both QT and pe are recorded
after reaching steady state. Each of the successive points of the flow
curve is obtained for another combination of h1,h2 after regulating the
valves to decrease h1 and increase h2 so that pe remains constant. The
methodology to compute the shear viscosity and the first normal stress
difference from readings of P1, P2 and P3 has been detailed elsewhere,
together with the calibration procedure of all transducers [5,12]. In a
few cases, flow curves were also obtained conventionally, i.e., gen-
erating the various shear rates with different screw speeds (or different
feed rates). In this case, the minimum flow rate ( ± 45 g/h) needed to
feed the mini-gear pump dictates the minimum throughput, and the
flow rate in the measurement channel can be further reduced by
opening the valve in the extrusion channel.
4. Results and discussion
4.1. Assessment of the flow analysis
The validity of the flow analysis used for designing the new double-
slit die was tested using mPE. The material was processed in the single
screw extruder at 210 °C with a constant screw speed of 50 rpm, giving
a throughput of 205 g/h. The resulting flow curve displayed in the inset
of Fig. 4 was fitted with the power law. The values of the index n and
consistency k were inserted in Eq. (8) to compute the theoretical values
Fig. 4. Comparison between experimental relative openings of the valves (solid circles)
assessed with mPE and theoretical relative openings of the valves (lines). Theoretical
values were computed with Eq. (8), using k and n parameters returned by a power law fit
to the experimental flow curve displayed in the inset and measured for mPE processed
under constant conditions. Theoretical data were calculated for the following valves
lengths (Lv1 = Lv2): 12 mm (squares); 19 mm (triangles), 32 mm (diamonds) and 56 mm
(circles).
141
Polymer Testing 66 (2018) 137–145
of h1/hc1 and h2/hc2 for a set of valves lengths. Fig. 4 shows that these
are at odds with the experimental points (particularly for the curve for
Lv1 = Lv2 = 12 mm): the range of possible relative openings is narrow
and the values set for h1/hc1 are systematically larger than the theore-
tical ones. This mismatch between the two sets of data might be asso-
ciated with the assumptions made for the theoretical derivation. To
start with, the validity of the power law to model the flow curve of mPE
is questionable. As seen in the inset of Fig. 4, a transition regime be-
tween a Newtonian plateau and a power law behavior shows up at
lower shear rates. As such, larger departure from the theoretical pre-
diction is expected for small values of h1/hc1 corresponding to smaller
shear rates in the measuring channel. Also, the assumption of fully
developed flow along the entire flow channel may be problematic,
particularly when the valve defines a narrow gap. Indeed, increasing
the length of the valve (to Lv1 = Lv2 = 19 mm, 32 mm or 75 mm) in Eq.
(8) to compensate for end effects leads to a much better approximation
to the experimental curve. Testing other constitutive equations for
modelling the fluid rheology together with the three-dimensional si-
mulation of the flow in the two channels would probably clarify this
point, but this lies beyond the scope of the present study.
4.2. Validation of the new die
As demonstrated in Fig. 5, despite the limitations of the flow ana-
lysis, flow balancing of the two valves (at constant total flow rate of
305 g/h) was able to generate a range of shear rates of two orders of
magnitude in the measurement slit. Also, within this range, the flow
distribution between extrusion and measurement channels changed
linearly, which simplifies the practical operation of the valves. In par-
ticular, data in Fig. 5 show that equality in volumetric flow rates of both
channels can be achieved for a specific valves tuning, in spite of the
double-slit geometry of the die and respective valve locations. As-
suming a generalized Newtonian relationship between stress and shear
rate for the fluid flowing in both channels, a larger pressure is expected
to be generated in the channel showing the largest flow rate. Given the
good flow balancing, flow curves generated with the new die were di-
rectly compared with the corresponding off-line data acquired with
conventional rheometers. In the case of mPE (Fig. 6), data was gathered
under the extrusion conditions presented in Table 1. Several measure-
ments were duplicated in order to estimate the experimental reprodu-
cibility – differences are less than 0.5%, i.e., within the range of sen-
sitivity of the pressure transducers (Fig. 6 (a)). In addition, as expected
for a typical medium-density polyethylene (a material with good
thermal stability), there is no evidence of effects of extrusion conditions
Fig. 5. Flow distribution between extrusion and measurement channels and corre-
sponding average shear rates at the latter.
P.F. Teixeira et al. Polymer Testing 66 (2018) 137–145

Fig. 7. In-process and off-line measurements of shear viscosity (complex and steady) and
normal stress differences with a PS, at 210 °C. The empirical relationship proposed by
Laun [34] relating the first normal-stress coefficient measured in steady shear experi-
ments to the dynamic shear moduli obtained from oscillatory data is also shown as dashed
line.

and complex shear viscosity η* measurements were performed using an

ARG2 rotational rheometer (TA instruments) equipped with a parallel-
plate geometry (25 mm diameter and 1 mm gap). Dynamic oscillatory
experiments were performed in the linear regime (1% strain), over an
angular frequency (ω) of 0.0628–628.3 rad s−1. Steady viscosity read-
ings (less than 3% variation within 10 s) were obtained while sweeping
the steady shear rates from 0.01 s−1 up to sample fracture. Generally,
reasonable agreement between the curves is observed, with good
overlap of the complex viscosity and viscosity measured with the new
die over a common range of shear rates.
Fig. 7 compares in-process and off-line measurements of shear
viscosity and normal stress differences for PS (same experimental pro-
tocols as for mPE measurements were adopted, albeit with a 0.5% strain
for dynamic experiments). Both N1 and N1 - N2 could not be measured
Fig. 6. Flow curves measured with a mPE: (a) duplicated runs produced at different
for mPE, as the small elasticity of the material hampered the use of the
temperatures with a flow rate of 100 g/h, and produced at 210 °C with different flow rates
achieved by setting different screw speeds; (b) comparison between steady and complex hole-pressure method (pressure differences between P2 and P3 were
shear viscosity measurements obtained by off-line rotational rheometry, and in-process smaller than the sensitivity of the sensors). PS was processed using the
shear viscosity data obtained using the double-slit die. single screw extruder at 210 °C with a constant flow rate of 175 g/h
(corresponding to a screw speed of 25 rpm). Fig. 7 shows excellent
agreement between off-line and in-process measurements of viscosity.
Table 1
Extrusion conditions used to obtain flow curves for mPE. The first normal stress difference, N1, was computed directly from P2 −
P3 readings (see Ref. [5] for details), using a transversal slot
Temperature (°C) Screw Speed (rpm) Flow rate (g/h) (width = 0.5 mm). N1 - N2 was measured with off-line rotational
rheometry, but since N2 is known to be much smaller than N1 for
190 25 100
200 25 100 polymer melts, N1 - N2 should compare well with in-process N1 data.
210 25 100 Unfortunately, there is a lack of overlapping shear rates between the
210 35 140 two methods. Thus, to overcome this experimental inconvenience, the
210 42 170
empirical relationship proposed by Laun [34] relating the first normal-
210 51 205
stress coefficient measured in steady shear experiments to the dynamic
shear moduli obtained from oscillatory data was adopted. The Laun
on the rheological data, since the curves measured at 210 °C with dif- equation reads as
ferent throughputs (100, 140, 170, 205 g/h) overlap. The largest shear a
rate achieved in experiments reported in Fig. 6 (a) is of the order of 60 N1 G′ G′ ⎞2⎤
= 2 2 ⎡1 + ⎛ for ω = γ˙
s−1, which is significantly below the largest values encountered in off- ω ⎢ ⎝ G′′ ⎠ ⎥
γ˙ 2
⎣ ⎦ (9)
line capillary rheometry. Evidently, the shear rate range spanned in
Fig. 6 (a) stems from the mini-extrusion line which delivers outputs in where a is an adjustable power law index (here taken as 1), and G' and
the range 30–300 g/h. Thus, if the proposed double-slit die concept is G″ are the storage and loss moduli, respectively.
transposed to conventional extrusion lines, significantly larger shear The dashed line in Fig. 7 suggests a fair agreement between oscil-
rates are expected to be achieved. Fig. 6 (b) compares the viscosity latory and pressure hole rheological measurements. However, N1 data
curves measured in-process with the steady and complex shear viscos- measured with the double-slit die are quite insensitive to the shear rate.
ities (assuming the validity of the empirical Cox-Merz rule) obtained by This indicates the ability of the hole pressure method to get an order of
standard off-line rheological characterization. Steady shear viscosity η magnitude of the material elastic properties at large shear rates (i.e.
N1), but not to obtain an accurate description of material elasticity.

142
P.F. Teixeira et al.
Fig. 8. Flow curves of mPE/PS immiscible polymer blends. (a) effect of composition on
blends processed at 210 °C in a co-rotating twin-screw extruder operated with a screw
speed of 40 rpm and a feed rate of 180 g/h (solid symbols). Inset: comparison between
steady shear viscosity η of mPE/PS blends measured at 10 s−1 and 40 s−1 (solid circles)
and the viscosity values computed from the log-additivity rule equation (dashed lines):
ln(η) = x ln(ηmPE ) + (1 − x )ln(ηPS ) , where ηPS and ηmPE are the shear viscosity of PS and
mPE, respectively, measured at the corresponding shear rate, and x is the mPE weight
fraction in the blend. (b) mPE/PS 80/20 w%/w% blend processed with a twin screw
(solid symbol) or a single screw (open symbols) extruder under different conditions;
crosses represent the data obtained with the single screw extruder and shear rates are
varied by changing the screw speed.
4.3. Application to various polymer systems
Comparison of off-line with in-process data obtained for mPE and PS
assessed positively the proposed new slit design and experimental
methodology. The device is particularly useful to study materials that
are sensitive to thermo-mechanical conditions. Thus, in this section,
polymer systems possessing a microstructure that depends on flow
history, namely a polymer blend and a nanocomposite, were studied.
The flow curves were measured under constant extrusion conditions
and data will be compared with flow curves measured conventionally
by changing the screw speed or feed rate.
The flow curves of a physical immiscible mPE/PS blend are pre-
sented in Fig. 8. Fig. 8 (a) shows the effect of composition on viscosity,
as measured during extrusion at 210 °C with the co-rotating twin screw
extruder at 40 rpm and a flow rate of 180 g/h. Each individual polymer
was also extruded and characterized under identical conditions. As can
be seen, the viscosity of the blend increases with PS content. The range
of shear rates attained increases with the addition of polystyrene as
higher pressures at lower flow rates are measured. The inset in Fig. 8 (a)
143
Polymer Testing 66 (2018) 137–145
displays a quantitative comparison between the experimental viscosity
values measured at 10 s−1 and 40 s−1, and the values computed with
the log-additivity rule valid for miscible blends [35]. At lower shear
rates, the expected negative deviation from the log-additive rule is
evident, whereas almost no deviation is seen at larger shear rates.
Understanding this effect would require a detailed characterization of
the blend morphology, which is beyond the scope of this study. To
study the effect of processing conditions, a mPE/PS 80/20 w%/w%
blend was produced using the twin screw extruder, processed at 180 g/
h with 40 rpm, and the single screw extruder operated with two screw
speeds: 40 rpm (flow rate 180 g/h) and 61 rpm (flow rate 300 g/h). The
corresponding flow curves plotted in Fig. 8 (b) show differences at
lower shear rates. In particular, the viscosity data obtained using the
single screw extruder with a higher flow rate (300 g/h) are slightly
smaller. This is probably due to a lower dispersion of PS droplets re-
sulting in less contributions from interfacial stresses to the viscoelasti-
city. A flow curve obtained using the conventional testing methodology
whereby the screw speed is varied to generate different shear rates in
the measuring slit is also reported in Fig. 8 (b). This flow curve sits on
top of all the other curves. At each screw speed (and thus for each data
point), a specific blend morphology is probably generated and any
explanation for the corresponding larger viscosity is, therefore, hard to
provide without the support of a morphological characterization.
A nanocomposite consisting of mPE reinforced with 4 wt%/CNT was
manufactured using the twin screw extruder set to 210 °C, with a con-
stant flow rate of 145 g/h and the screws rotating either at 30 or
100 rpm, with the aim of obtaining materials with final distinct filler
dispersion levels. Fig. 9 presents the corresponding viscosity and first
normal stress difference curves obtained while keeping pe constant. A
faster screw speed triggered a smaller viscosity, a small reduction of the
power law index (from 0.63 to 0.60) and enhanced elasticity (see N1
data). This rheological pattern is often taken as evidence of improved
CNT dispersion, since the breakdown of CNT aggregates into smaller
anisotropic particles results in enhanced effective solid volume fraction
and polymer confinement, thereby leading to larger elasticity, en-
hanced shear thinning and smaller apparent viscosity at large shear
rates [36–38].
The same nanocomposite was also prepared by single screw extru-
sion, as this machine develops quite distinct flow patterns, residence
time distribution and thermomechanical environments from those in
the twin screw extruder. The viscosity data could not be obtained by
balancing the flow using the valves, since the pressure at the die inlet
was unsteady (thus, N1 data was also inaccessible). This was probably
due to the limited mixing capacity of the screw extruder, which
Fig. 9. Flow curves of mPE/CNT 96/4 wt%/wt% nanocomposite measured by varying the
shear rate in the measuring channel while keeping constant processing conditions (filled
symbols) with twin screw or single screw extruders, or by varying the processing con-
ditions (symbols).
P.F. Teixeira et al.
Fig. 10. Rheological characterization of a polycarbonate: (a) temperature and total re-
sidence time effects on the mechanical spectra measured with rotational rheometry, on
the flow curves measured with capillary rheometry and on in-process measurements with
the double-slit die coupled to a single screw extruder; (b) degradation kinetics measured
with rotational rheometry at 250 °C, 260 °C and 270 °C (the dash lines indicate the times
(min) at which the complex viscosity decreases 5% with respect to the initial value, for
each temperature).
produced an uneven morphology generating die pressure fluctuations
which could not be alleviated by the mini gear-pump. Thus, the flow
curve was determined varying the screw speed from 12 to 54 rpm (and
consequently the output between 43 and 165 g/h). The resulting set of
data plotted in Fig. 9 is qualitatively and quantitatively different from
that originated from the twin screw extruder, thus confirming that
equipment and/or processing conditions upstream of the rheological
die may create different materials, even if the material recipe at the
entrance of the extruder is the same.
Temperature, residence time and mechanical effects during pro-
cessing of a polymer can also be assessed with the proposed slit die.
This is illustrated for a commercial polycarbonate processed by single
screw extrusion at 250 °C, 260 °C and 270 °C. Fig. 10 (a) shows the
viscosity curves measured off-line from high to low shear rates, using a
Rosand Capillary rheometer RH10 (dies with diameter of 1 mm, and
different length namely 16 mm and 8 mm, for Bagley correction) and
after melting PC during 3 min in the rheometer's reservoir. Mechanical
spectra measured from high to low frequencies with a ARG2 rotational
rheometer (25 mm parallel-plate geometry with 1 mm gap, loaded with
PC disks pre-molded during 2 min at 230 °C, and equilibrated in the
rheometer at the corresponding temperature during 1 min before ap-
plying a 1% strain to ensure both linear regime of viscoelasticity and
satisfactory torque signals), are also plotted in Fig. 10 (a). The
144
Polymer Testing 66 (2018) 137–145
occurrence of a viscosity plateau at lower frequencies in the mechanical
spectra confirms the absence of any significant degradation of the
material during 3.9 min of oscillatory testing. Thus, these spectra can be
used as reference curves if one extends the Cox-Merz equivalence to all
frequencies and shear rates. Flow curves measured with the capillary
rheometer indicate that PC degraded after approximately 6 min inside
the device (see the values of residence times next to the corresponding
curves in Fig. 10 (a)), for all temperatures, as the steady shear viscosity
begins to deviate from the complex viscosity. Thus additional off-line
rheological characterization was also performed to probe effects of
thermal degradation. Fig. 10 (b) shows the degradation kinetics mea-
sured with rotational rheometry at 250 °C, 260 °C and 270 °C, right after
loading of pre-molded disks and thermal equilibration. The thermo-
degradability of PC is inferred from the decrease of the complex visc-
osity. The degradation kinetics seems to be similar at 250 °C and 260 °C,
as a 5% decrease in the complex shear viscosity is measured after 8.9
and 8.6 min, respectively (see the dash lines in Fig. 10 (b)). At 270 °C
the degradation kinetics is faster with a 5% decrease in the complex
shear viscosity after only 6.2 min. Thus, temperature and residence time
will affect the in-process measurements. Double-slit die measurements
were limited by excessive pe at larger flow rates and lower tempera-
tures. However, a flow curve at 250 °C was obtained for a flow rate of
66 g/h (17 rpm) resulting in a residence time of 10 min for the largest
shear rate. Data in Fig. 10 (a) confirm that PC degradation did occur
under such conditions since the in-process steady shear viscosity is
significantly smaller than the complex viscosity. Obviously, the situa-
tion is even worse for lower shear rates as longer residence times are
needed. At 260 °C, two processing conditions were studied: flow rate of
97 and 72 g/h (screw speeds of 26 and 17 rpm respectively), resulting in
minimum residence times of 6.5 and 10.3 min, respectively. For re-
sidence times in excess of 6.5 min, PC thermal degradation is evident as
the steady shear viscosity is apparently shear thickening, which is at
odds with the Newtonian behavior of the complex viscosity. Data ob-
tained after processing at smaller screw speed evidently suffer more
from the PC degradation. Comparing the flow curve obtained con-
ventionally by varying the flow rate from 57 to 97 g/h (16–26 rpm
respectively) to generate the corresponding range of shear rates with
the previous curves at 260 °C (using the valves system), allows separ-
ating the thermo-mechanical effects from those related to residence
time. Two steady shear viscosities in Fig. 10 (a) were obtained with a
similar residence time (9.3 min) but show very different values, due to
differences in mechanical histories. Larger screw speeds (26 rpm) result
in stronger viscous heating, and thus larger PC degradation which leads
to depressed viscosity. At 270 °C, the flow curve was obtained for a flow
rate of 130 g/h (36 rpm), resulting in a residence time of 5.1 min, see
Fig. 10 (a). As the temperature is higher, the viscosity is lower, allowing
higher flow rates and, consequently, shorter residence times. After
5.1 min, the material is not significantly degraded, and shear viscosity
values measured with the double-slit rheometrical die compare well
with data obtained from rotational rheometer.
5. Conclusions
The concept of twin-slit die rheometry operating at constant en-
trance pressure was modified to double-slit die rheometry with mea-
surement and extrusion channels having different designs.
The double-slit concept was first studied by generalizing the flow
analysis initially developed for twin channels to accommodate different
channel lengths and cross sections. The resulting set of equations was
then used to support the design of the double-slit die, assuring the best
compromise between a large range of attainable shear rates and a
simpler practical operation of the two valves. The construction of the
die enables the in-situ measurement of structural information and
coupling to an existing mini extrusion line. The assessment of the new
die with a medium density polyethylene and a polystyrene demon-
strated that the rheological data measured in-process match those
P.F. Teixeira et al.
achieved through off-line rheometry. The study also pointed out the
limitations of the flow analysis, namely in terms of some of its initial
assumptions.
In-process measurements carried out for a polymer blend and a
polymer nanocomposite showed that the double-slit die is adequate for
characterizing fluids with strong viscoelastic sensitivity to thermal and
mechanical histories. In particular, one set of data suggested that a
specific commercial polycarbonate was very sensitive to thermo-
mechanical degradation.
The possibility to assess the shear-induced structure through optical
characterization while delivering simultaneously shear viscosity and
normal stress difference data for a range of deformation rates (at con-
stant extrusion conditions upstream), opens a promising experimental
route to unravel rheology-structure relationships in complex fluids. This
is the case of immiscible polymer blends and (nano)composites, as flow
conditions that mimic industrial processing can be reproduced with the
proposed experimental set-up.
Acknowledgements
The authors acknowledge the financial support by National Funds
through FCT - Portuguese Foundation for Science and Technology,
Reference UID/CTM/50025/2013 and FEDER funds through the
COMPETE 2020 Programme under the project number POCI-01-0145-
FEDER-007688. L.L. Ferrás would like to thank the funding by FCT
through scholarship SFRH/BPD/100353/2014.
References
[1] L. Saerens, C. Vervaet, J.P. Remon, T. De Beer, Process monitoring and visualization
solutions for hot-melt extrusion: a review, J. Pharm. Pharmacol. 66 (2014)
180–203.
[2] I. Alig, B. Steinhoff, D. Lellinger, Monitoring of polymer melt processing, Meas. Sci.
Technol. 21 (2010).
[3] C.D. Han, Measurement of the rheological properties of polymer melts with slit
rheometer. I. Homopolymer systems, J. Appl. Polym. Sci. 15 (1971) 2567–2577.
[4] C. Rauwendaal, F. Fernandez, Experimental study and analysis of a slit die visc-
ometer, Polym. Eng. Sci. 25 (1985) 765–771.
[5] P.F. Teixeira, L. Hilliou, J.A. Covas, J.M. Maia, Assessing the practical utility of the
hole-pressure method for the in-line rheological characterization of polymer melts,
Rheol. Acta 52 (2013) 661–672.
[6] D.G. Baird, First normal stress difference measurements for polymer melts at high
shear rates in a slit-die using hole and exit pressure data, J. Non-Newtonian Fluid
148 (2008) 13–23.
[7] M. Padmanabhan, M. Bhattacharya, In-line measurement of rheological properties
of polymer melts, Rheol. Acta 33 (1994) 71–87.
[8] K. Seethamraju, M. Bhattacharya, Effect of ingredients on the rheological properties
of extruded corn meal, J. Rheol. 38 (1994) 1029–1044.
[9] S. Li, K.B. Migler, E.K. Hobbie, H. Kramer, C.C. Han, E.J. Amis, Light-scattering
photometer with optical microscope for the in-line study of polymer extrusion, J.
Polym. Sci., Polym. Phys. Ed. 35 (1997) 2935–2943.
[10] E.K. Hobbie, K.B. Migler, C.C. Han, E.J. Amis, Light scattering and optical micro-
scopy as in-line probes of polymer blend extrusion, Adv. Polym. Technol. 17 (1998)
307–316.
[11] K.B. Migler, E.K. Hobbie, F. Qiao, In line study of droplet deformation in polymer
blends in channel flow, Polym. Eng. Sci. 39 (1999) 2282–2291.
[12] P.F. Teixeira, J.M. Maia, J.A. Covas, L. Hilliou, In-line particle size assessment of
polymer suspensions during processing, Polym. Test. 37 (2014) 68–77.
145
Polymer Testing 66 (2018) 137–145
[13] B. Vergnes, G. Della Valle, J. Tayeb, A specific slit die rheometer for extruded
starchy products. Design, validation and application to maize starch, Rheol. Acta 32
(1993) 465–476.
[14] M. Padmanabhan, M. Bhattacharya, Flow behavior and exit pressures of corn meal
under high-shear–high-temperature extrusion conditions using a slit diea, J. Rheol.
35 (1991) 315–343.
[15] P. Li, O. Campanella, A. Hardacre, Using an in-line slit-die viscometer to study the
effects of extrusion parameters on corn melt rheology, Cereal Chem. 81 (2004)
70–76.
[16] M. Padmanabhan, M. Bhattacharya, Effect of extrusion processing history on the
rheology of corn meal, J. Food Eng. 18 (1993) 335–349.
[17] J.M. Barbas, A.V. Machado, J.A. Covas, Processing conditions effect on dispersion
evolution in a twin-screw extruder: polypropylene-clay nanocomposites, Chem.
Eng. Technol. 37 (2014) 257–266.
[18] V. Yquel, A. Machado, J. Covas, J. Flat, Contribution of the melting stage to the
evolution of the morphology and chemical conversion of immiscible polyamide/
polyethylene blends in twin-screw extruders, J. Appl. Polym. Sci. 114 (2009)
1768–1776.
[19] A. Pabedinskas, W. Cluett, S. Balke, Development of an in-line rheometer suitable
for reactive extrusion processes, Polym. Eng. Sci. 31 (1991) 365–375.
[20] J.M. Dealy, T.O. Broadhead, Process rheometers for molten plastics: a survey of
existing technology, Polym. Eng. Sci. 33 (1993) 1513–1523.
[21] M. Horvat, M.A. Emin, B. Hochstein, N. Willenbacher, H.P. Schuchmann, A mul-
tiple-step slit die rheometer for rheological characterization of extruded starch
melts, J. Food Eng. 116 (2013) 398–403.
[22] P. Coates, R. Rose, M. Barghash, A'bleed' on line rheometer for polymer melts,
ANTEC 2001 Conference Proceedings, 2001, p. 5.
[23] D. Kalyon, H. Gokturk, I. Boz, An adjustable gap in-line rheometer, Technical papers
of the annual technical conference - society of Plastics Engineers Incorporated, Cités
(1997) 2283–2288.
[24] D.M. Kalyon, H.S. Gokturk, Adjustable Gap Rheometer, Google Patents (1994).
[25] B. Hochstein, Z. Kizilbay, M. Horvat, H.P. Schuchmann, N. Willenbacher,
Innovatives Inline-Rheometer zur Bestimmung des Stärkeabbaus im Extruder,
Chem. Ing. Tech. 87 (2015) 90–94.
[26] P.W. Springer, R.S. Brodkey, R.E. Lynn, Development of an extrusion rheometer
suitable for on-line rheological measurements, Polym. Eng. Sci. 15 (1975) 583–587.
[27] G. Della Valle, P. Colonna, A. Patria, B. Vergnes, Influence of amylose content on
the viscous behavior of low hydrated molten starches, J. Rheol. 40 (1996) 347–362.
[28] G. Della Valle, B. Vergnes, P. Colonna, A. Patria, Relations between rheological
properties of molten starches and their expansion behaviour in extrusion, J. Food
Eng. 31 (1997) 277–295.
[29] L. Lach, Modelling Vapour Expansion of Extruded Cereals, Ecole Polytechnique
Fédérale, Lausanne, Switzerland, 2006.
[30] F. Robin, J. Engmann, D. Tomasi, O. Breton, R. Parker, H.P. Schuchmann, S. Palzer,
Adjustable twin-slit rheometer for shear viscosity measurement of extruded com-
plex starchy melts, Chem. Eng. Technol. 33 (2010) 1672–1678.
[31] D.G. Baird, Possible method for determining normal stress differences from hole
pressure error data, Trans. Soc. Rheol. 19 (1975) 147–151.
[32] R.M. Novais, J.A. Covas, M.C. Paiva, The effect of flow type and chemical func-
tionalization on the dispersion of carbon nanofiber agglomerates in polypropylene,
Compos. Part a-Appl S 43 (2012) 833–841.
[33] P.F. Teixeira, J.M. Maia, J.A. Covas, L. Hilliou, A small-scale experimental extrusion
set-up for exploring relationships between process-induced structures and char-
acteristics of multiphase polymer systems, Macromol. Mater. Eng. 300 (2015)
1278–1289.
[34] H.M. Laun, Prediction of elastic strains of polymer melts in shear and elongation, J.
Rheol. 30 (1986) 459–501.
[35] L.A. Utracki, C.A. Wilkie, Polymer Blends Handbook, Springer, 2002.
[36] R. Larson, The Structure and Rheology of Complex Fluids, vol. 546, Oxford
University Press, New York, 1999.
[37] J. Jancar, J. Douglas, F.W. Starr, S. Kumar, P. Cassagnau, A. Lesser, S.S. Sternstein,
M. Buehler, Current issues in research on structure–property relationships in
polymer nanocomposites, Polymer 51 (2010) 3321–3343.
[38] M. Nobile, Rheology of polymer-carbon nanotube composite melts, Polymer-carbon
nanotube composites, in: T. McNally, P. Pötschke (Eds.), Preparation, Properties
and Application, Woodhead Publishing Limited, Cambridge (UK), 2011, pp.
428–481.