Shapes and shaping of biopolymer drops in a hyperbolic flow
-
Upload
lars-hamberg -
Category
Documents
-
view
216 -
download
3
Transcript of Shapes and shaping of biopolymer drops in a hyperbolic flow
![Page 1: Shapes and shaping of biopolymer drops in a hyperbolic flow](https://reader036.fdocuments.us/reader036/viewer/2022081214/5750755b1a28abdd2e992282/html5/thumbnails/1.jpg)
Shapes and shaping of biopolymer drops in a hyperbolic flow
Lars Hamberga, Mathias Wohlwendb, Pernilla Walkenstroma, Anne-Marie Hermanssona,*
aSIK, The Swedish Institute for Food and Biotechnology, P.O. Box 5401, SE-402 29 Goteborg, SwedenbETH, Laboratory of Food Engineering, Swiss Federal Institute of Technology, 8092 Zurich, Switzerland
Received 11 July 2002; revised 4 October 2002; accepted 18 December 2002
Abstract
The shaping of drops in a model system based on k-carrageenan-emulsion drops in the millimetre range in silicon oil has been studied. The
drops were shaped by exposing them to drag forces in a hyperbolic flow, while their shape was fixed simultaneously by introducing gel
formation of the biopolymer in the drop.
The shape and the shaping process were studied and evaluated with image analysis of macrograph sequences of the shaping. The effect of
process conditions, flow speed and cooling temperature on the final shape and shape progress was investigated as well as the effect of
different k-carrageenan drop characteristics, such as drop viscosity and gel strength. Drop viscosity was altered by addition of locust bean
gum, LBG, and the gel strength was altered by addition of ions. The k-carrageenan solutions in the drop were characterised by rheological
investigations.
With the same type of flow, different shapes could be achieved with small process changes and with high reproducibility. The fixation of
the characteristic drop features, perimeter, area, Feret’s X and Y ; does not occur at the same time and position. For the different process
parameters investigated, a change in speed affected the process in a similar way to a change in the viscosity ratio. This applies if the viscosity
ratio is changed at a constant temperature, but if the change in the viscosity ratio is temperature-induced, the effect is different. The final
shape of the produced drops could be graded into three classes, correlated to the position in the flow field where the drops were fixed. A shape
map of the different drop shapes obtained was presented.
q 2003 Elsevier Science Ltd. All rights reserved.
Keywords: Shape; Shaping; Elongation; Deformation; Drop; k-carrageenan
1. Introduction
Emulsion droplet deformation during flow processing
could lead to emulsion-based products with new properties
if the shape of the deformed drop is conserved. Small
deformation of drops induced by flow has been studied for a
long time (Rallison, 1984; Stone, 1994; Walther, 2001).
However, large deformation of a drop into a specific drop
shape is an area where, until recently, only a limited number
of studies have been reported (Hamberg, Walkenstrom, &
Hermansson, 2002; Walther, Walkenstrom, Hermansson,
Fischer, & Windhab, 2002; Wolf, Scirocco, Frith, & Norton,
2000). To increase the functionality of a fluid drop, it is not
enough to simply deform and shape it. It is also necessary to
fix and make the drop resistant to further changes when a
desirable shape has been accomplished. One way to achieve
fixation is by gel formation of the biopolymers in the drop.
The fixation and deformation process often interact, affect
one another and could therefore be treated as one process,
shaping (Hamberg et al., 2002).
When drops or particles are shaped, their contribution to
the functionality of the suspension is believed to be different
from the contribution for a spherical drop or particle. Little
is investigated for biopolymers (Stokes, Wolf, & Frith,
2001; Wolf, Frith, Singeleton, Tassieri, & Norton, 2001),
but in the nearby fibre-suspension area, interesting parallels
could be found. The shape effect of a hard symmetric fibre
suspension on rheology is known from previous studies and
reviewed both for theory (Petrie, 1999) and for practice
(Djalili-Moghaddam, 2001). On addition of fibres and
spheres of the same number and individual volume, the
viscosity increases more markedly with addition of fibre.
Even small alterations in the shape of the fibres make a
difference. When the shape of a fibre is a changed from the
normal symmetry to become somewhat curved, even a small
0268-005X/03/$ - see front matter q 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0268-005X(03)00009-2
Food Hydrocolloids 17 (2003) 641–652
www.elsevier.com/locate/foodhyd
* Corresponding author. Tel.: þ46-31-33-55-600; fax: þ46-31-83-
37-82.
E-mail address: [email protected] (A.-M. Hermansson).
![Page 2: Shapes and shaping of biopolymer drops in a hyperbolic flow](https://reader036.fdocuments.us/reader036/viewer/2022081214/5750755b1a28abdd2e992282/html5/thumbnails/2.jpg)
curvature ,5% can increase the solution viscosity by an
order of 10 to 15% compared with a solution with straight
fibres (Joung, Phan-Thien, & Fan, 2002). The curvature
does not have to be fixed; flexible fibres have been
theoretically treated, and it was found that a sufficient
viscosity increase could be expected (Joung, Phan-Thien, &
Fan, 2001). If parallels could be found and explored in the
case of biopolymer drops, shaping would be an area of great
interest.
The drop-shaping method used in this work has recently
been presented by Walther et al. (2002). In their work,
shaping was done with millimetre-sized gelatine drops in
hyperbolic flow, and both elongated drops and drops of
complex form were formed. They found that the region
where the drop was fixed was directly related to the created
shape. The impact of the process conditions on the final
shape of the drop was clear and it was possible to control
process with high reproducibility. Hamberg et al. (2002)
also used the same type of hyperbolic flow to shape small
ellipsoidal drops. They investigated the impact of different
gel formation kinetics by comparing sub-millimetre drops
of gelatine and k-carrageenan under the same conditions.
Wolf et al. (2000) have presented a method using pure shear
flow instead of hyperbolic flow. It resulted in stretched and
ellipsoid drops with large differences in the length to width
ratio.
Although shear flow has long been used to deform drops,
the focus has shifted more and more to other flow types,
such as hyperbolic flow. Hyperbolic flow can easily be
generated in the centre of a four-roll mill 4-RM (Taylor,
1934). By following the centre flow-line from the centre and
out between one pair of rollers, one obtains first an
elongation flow, then a contraction flow. Elongation flow
has the advantage of being more effective in terms of
creating interfacial area than shear flow (Erwin, 1991).
However, elongation flow is limited in residence-time and
space for practical reasons. Therefore, processes with
elongation flow are limited to a time scale of seconds.
This places high demands on analysing tools as well as on
the time scale of properties of the flow, the drop and the
biopolymer.
In this work, the shaping of drops in a model system
based on k-carrageenan-emulsion drops in the millimetre
range in silicon oil has been studied. k-carrageenan is cold-
set biopolymer, i.e. a water/k-carrageenan system in the
drop is in solution at temperatures over Tgel; and fixes when
forming a gel at temperatures below Tgel: The viscosity of
the biopolymer solution above Tgel is strongly temperature-
dependent and rises drastically just before Tgel: The gel
formation properties of carrageenan are strongly dependent
on the type of carrageenan and the counter-ions present in
the solution, (Hermansson, Eriksson, & Jordansson, 1991;
Rochas, & Rinaudo, 1980, 1984; Rochas, Rinaudo, &
Landry, 1989). The ion-dependence makes it possible to
choose a suitable Tgel by introducing the proper ion-
compound and ion-strength of the solution. In this food
emulsion model system, the silicon oil used is neither food
grade nor is the size of the drops at the same size as the
drops found in food products. The oil is present in the
system primary as ‘a shaping agent’ and is not mixed with
the product, i.e. the drops. The oil is chosen because the high
viscous silicon oil made it possible to study the process in
the 4-RM at lower speeds than low viscous oil. The use of
other oils will probably not change the mechanism in other
means than scaling the force from the flow and changing
interfacial properties.
The aim of this study was to investigate the final shape
and the shape development of millimetre k-carrageenan
drops during a combination of cold-set gel formation and
deformation by elongation flow. The shape and the process
have been evaluated and studied in terms of macrograph
sequences and image analysis to be able to investigate not
only the final shape but also the changes in the shape during
the process. The effect of process conditions, flow speed and
cooling temperature, on the final shape and shape progress
was investigated as well as the effect of different k-
carrageenan drop characteristics, drop viscosity and gel
strength. The k-carrageenan solutions were characterised by
rheological investigations. The reason for exploring differ-
ent shapes is not only to study the shape as a final product,
but also to pave the way for future studies of the special
functional properties that could be related to the final drop
shape in future emulsions with smaller shaped drops.
2. Materials and methods
2.1. Biopolymer
The k-carrageenan used was a product of Danisco Cultor
(8220 Brabrand, Denmark). It is a commercial potassium-k-
carrageenan. The ion content was 8.5% K, 0.18% Ca and
1.5% Na. The locust bean gum, LBG, was also a
commercial product of Danisco Cultor (8220 Brabrand,
Denmark), called Grindstede LBG 246.
2.2. Oil
The oil used as the continuous phase was silicone oil,
polydimethylsiloxane, PDMS, named Wacker Silicone
Fluid AK5 000 from Wacker-Chemie GmbH, Burghausen,
Germany. The oil had a measured viscosity of 6.4 Pas at
10 8C and 5.2 Pas at 20 8C.
2.3. Additive
To increase the quality of the macrographs and reduce
the image measurement error, aniline blue was used as a
contrast agent. Aniline blue was obtained from Riedel de
Haen AG, Seelze-Hannover, Germany. Addition of standard
NaCl was used to alter the Naþ concentration.
L. Hamberg et al. / Food Hydrocolloids 17 (2003) 641–652642
![Page 3: Shapes and shaping of biopolymer drops in a hyperbolic flow](https://reader036.fdocuments.us/reader036/viewer/2022081214/5750755b1a28abdd2e992282/html5/thumbnails/3.jpg)
2.4. Solution preparation
2.4.1. Preparation of the k-carrageenan solution
Three different types of 1% w/w k-carrageenan water
solutions with different additions of sodium chloride were
prepared. The three solutions had a final sodium concen-
tration, including sodium from the k-carrageenan, of 17,
109 and 212 mM, respectively. The amount of Kþ ion
present from the k-carrageenan powder resulted in a final
concentration of Kþ of 22 mM. Although only the Naþ ion
concentration is adjusted, presence of Kþ ions is necessary
for the gel formation and vital for the final gel strength
(Hermansson et al., 1991). Aniline blue was added to the
solutions at a concentration of 0.5% w/w. To dissolve all the
ingredients, the solution was heated to and then held at
95 8C for 15 min while stirring.
2.4.2. Preparation of k-carrageenan blended with LBG
For the solution of 1% k-carrageenan blended with LBG,
0.25% w/w LBG was added and the sodium chloride
addition was adjusted to obtain the same final properties
as for the k-carrageenan solution with 109 mM Naþion
concentration. In order to dissolve the k-carrageenan/LBG
mixture, the solution was first pre-stirred for 30 min at room
temperature before the solution was held at 95 8C for 15 min
during stirring.
2.5. Shaping
2.5.1. Flow
The hyperbolic flow was generated in a four-roll mill,
4-RM (Taylor, 1934). The 4-RM was built and adapted to
fit a microscope by the workshop at the Institute of Food
Science/Food Process Engineering at ETH, Zurich,
Switzerland and has been described in more detail earlier,
(Hamberg, Walkenstrom, Stading, & Hermansson, 2001).
By letting one pair of the rolls rotate in one direction and
the other pair in the opposite direction, the hyperbolic
flow-field will be generated in the centre of the chamber
(Feng & Leal, 1997). In the investigations, the speed of
the rolls was 5, 10, 15, 20 and 25 rpm. The corresponding
flow along the centre line for different roll speeds
constantly increases through the inlet area at an elongation
rate of around 0.5 s21 for the lowest roller speed and
around 2.5 s21 for the highest speed. The flow speed of
the continuous phase reaches its maximum in the
narrowest gap, with values around 12 and 48 mm/s for
the lowest and highest speed, respectively. In the same
area, the elongation rate drops from positive to negative
with the same magnitude as for the inflow. The flow along
the centre line slows down and approaches immobility at
a distance approximately 70 mm from the stagnation
point. The elongation rate along the centre line drops to
zero from the peak values at the beginning of the outlet
from around 20.5 and 22.5 s21.
2.5.2. Process
The shaping process started with injection of a 10 ml
large drop of the biopolymer solution into the oil in the 4-
RM. The rolls were started in advance so that a fully
developed hyperbolic flow field was generated. The
biopolymer solution was heated to 90 8C and the
surrounding oil to 10 or 20 8C. The drop was injected at
the centre line, 4 mm from the stagnation point in the
centre. First, the drop was accelerated from the injection
point to the narrowest gap between the rollers and then,
after passing the narrowest gap, the drop was exposed to
retardation in the outflow from the rollers. The tempera-
ture of the biopolymer phase was chosen in order to have
a large gap to the gel formation temperature. This ensures
that the drops is deformable until the relaxation flow in
the outlet and enable the possibility to form complex-
shaped drops. The oil temperatures were chosen to be
sufficiently under gel formation temperature to ensure a
rapid cooling of the drop around gel formation
temperature.
2.5.3. Macroscopy
The shaping was recorded with a macroscopic technique,
in which a sequence of macrographs was taken to observe
the shaping progress. The macrographs were taken with a
Hamamatsu C6157, 3 CCD Colour Video Camera, Hama-
matsu Photonics K.K., Hamamatsu-city, Japan, connected
to a frame-grabber in a workstation equipped with the image
analysis program MicroGOP 2000/S, Contextvision AB,
Linkoping, Sweden. The camera was tilted at an angle of 58
from the normal of the oil surface in the 4-RM to obtain
good picture quality. To avoid motion-induced blurring, the
electronic shutter of the camera was set to 1/10,000 s. The
time resolution was higher than 0.1 s, corresponding to a
measured frame rate of around 12 frames/s.
2.5.4. Image analysis
The macrographs were converted into binary images and
the drop was isolated before drop parameters were
measured. The parameters used were defined as follows:
Area The projection area of a drop on an x–y plane.
Perimeter The projection perimeter of a drop on an x–y
plane.
Feret’s X The projection length of a drop along the centre
line.
Feret’s Y The projection length of a drop perpendicular to
the centre line.
Position x-coordinate of the centre of gravity of a drop.
Graphic illustrations of the response parameters are
given in Fig. 1.
L. Hamberg et al. / Food Hydrocolloids 17 (2003) 641–652 643
![Page 4: Shapes and shaping of biopolymer drops in a hyperbolic flow](https://reader036.fdocuments.us/reader036/viewer/2022081214/5750755b1a28abdd2e992282/html5/thumbnails/4.jpg)
2.6. Rheological characterisation
2.6.1. Viscosity
To characterise the properties of the solutions before gel
formation, viscosity measurements were made. The instru-
ment used was a Bohlin VOR Rheometer, Bohlin Rheology,
Chichester, UK, equipped with Millennium VOR software,
Reologen i Lund, Oved, Sweden, with a double gap
geometry measuring system. The system had an inner and
outer rotating wall with diameter 21.7/27.4 mm and a fixed
bob with a diameter of 23.9/24.9 mm. A volume of 10 ml
from the different solutions was used and shear sweeps were
performed at 90, 60, 50, 40 and 35 8C.
2.6.2. Gel formation temperature
Oscillatory measurements were also made while the
solution was cooled from 50 to 10 8C to measure the gel
formation temperature, Tgel: A stress-controlled StressTech
HR Rheometer, Rheologica Instruments AB, Lund, Swe-
den, with Peltier cooling was used with a cone and plate
geometry, angle 48, diameter 30 mm. 0.5 ml of the warm
solution was placed on the plate and measurement was made
with a cooling rate of 5 8C/min, frequency of 1 Hz, and with
a strain of 0.001. Tgel was determined as the temperature
where the storage modulus, G0; crosses over the loss
modulus, G00:
2.6.3. Gel strength
To measure the bulk gel strength of the drops, gel tubes
were prepared from solutions with the same properties as
the different drops by pouring a hot solution, heated to
95 8C, in a metal cylinder. The metal tube was put in an ice-
bath to cool the solution and to build a gel. Then the gel
tubes were taken out of the cylinders and cut into 5 mm
thick slices. The measurements were made with a Bohlin
VOR Rheometer, as described earlier. The measurement
system had riffled plate-plate geometry, with a diameter of
30 mm. The measurements were carried out at a tempera-
ture of 10 and 20 8C with a strain of 0.025, amplitude of 5%
and a frequency of 1 Hz. The gel slices were applied to
the measurement system and compressed 5% before the
start of the measurements.
3. Results and discussion
This part is divided into four sections of which Section
3.1 describes typical shaping of a drop. Section 3.2 covers
the rheological characterisation of the k-carrageenan
solution. Section 3.3 focuses on the end shape of a
processed drop, and Section 3.4 concerns measurements
and image analysis of the shaping process.
3.1. Shaping
Typical shaping of a drop is shown in the 10 macrograph-
long sequence in Fig. 2. The macrographs cover an area
from the centre of the 4-RM, on the left-hand side of the
macrographs, where the injection is situated, along
the centre line, to the outflow region on the right-hand
side. The process ends approximately 70 mm from the
centre of the 4-RM. Just to the left of the middle of the
macrograph the tip of two of the rollers is visible, indicating
where the narrowest gap is located.
The macrographs have been taken from a sequence with
the experimental conditions 1% k-carrageenan in 109 mM
Naþ, 15 rpm roller speed and 10 8C cooling temperature.
The time between each macrograph is approximately 0.8 s.
The first macrograph was taken when the injection needle
was positioned. The second macrograph was taken during
the injection. Sequences 3–5 show the actual elongation
process and the contraction is shown in sequences 5–7.
From macrograph 7 to the end in macrograph 10, no large
Fig. 1. Definitions of the different responses measured: (a) area, (b)
perimeter, (c) Feret’s X and (d) Feret’s Y :
Fig. 2. Sequence showing the progress of shape formation. The
macrographs were taken with approximately 0.8 s delays starting at the
top. Conditions of 15 rpm, 10 8C and 109 mM Naþ.
L. Hamberg et al. / Food Hydrocolloids 17 (2003) 641–652644
![Page 5: Shapes and shaping of biopolymer drops in a hyperbolic flow](https://reader036.fdocuments.us/reader036/viewer/2022081214/5750755b1a28abdd2e992282/html5/thumbnails/5.jpg)
change in shape is visible, but transportation along the
centre line is noted.
Fig. 2 also shows the development of the different
characteristic parts of the drop: the body, the limbs and the
blob. The characteristic parts are illustrated in Fig. 3. The
first limb to be developed is the bow-limb, macrographs 3–
5. The limb is stretched out from the bow as the bow
formation accelerates into the gap. When the bow of the
drop has passed the narrowest gap, a blob is developed at the
bow limb, macrographs 4–7. The forming of the aft limb
starts later than the bow limb, as can be seen in the
macrographs 4–6, and seems to form when the body of the
drop is accelerated around the narrowest gap. The body of
the drop is already present from the beginning. First, it is
located in the centre, later when the bow is accelerated the
body is closer to the aft. In the end, at the outflow, the body
contracts along the centreline and expands in the perpen-
dicular direction. The body seems to be the last part of the
drop to be fixed.
3.2. Rheological characterisation of k-carrageenan
3.2.1. Viscosity
The result of the viscosity measurements of pure k-
carrageenan solution and k-carrageenan solution blended
with LBG at temperatures between 90 and 35 8C is
presented in Table 1. The measurements were made with
solutions at the 109 mM Naþ ion concentration. In the table,
only one viscosity value from the different measured flow
curves is presented. For the pure k-carrageenan solution, no
shear dependence was found for shear rates between 5 and
100 s21 at temperatures between 90 and 40 8C. At 35 8C a
very weak shear dependence was detectable along the
lowest shear rates, and therefore the value in the table is
presented for a shear rate at 50 s21. Limited shear-
dependence was also found for the k-carrageenan solution
blended with LBG, and hence all the values in Table 1 for k-
carrageenan/LBG are presented for 50 s21. The k-carragee-
nan solution tended to be more shear thinning at lower
temperatures than at higher temperatures.
The viscosities for both the pure k-carrageenan solution
and the solution blended with LBG were highly tempera-
ture-dependent, and increased rapidly when the temperature
came close to Tgel: The viscosities of the pure k-carrageenan
solution were in general more than six times lower than for
the k-carrageenan/LBG solution. It was consequently
concluded that differences in drop viscosity due to the
cooling temperature, 10 and 20 8C, of the oil, were much
less than those due to addition of LBG for all temperatures
away from Tgel surroundings. The viscosity ratio between
the pure k-carrageenan solution in the drop and the oil
changes during the process varied from 0.0006 at 90 8C to
0.003 at 35 8C. The rapid increase in viscosity near Tgel
increases the ratio further. Corresponding values for the
blended drops were 0.004 at 90 8C and 0.03 at 35 8C. All
viscosity ratios refer to the oil temperature of 10 8C.
3.2.2. Gel-formation temperature
The aim was to have as little variation as possible in Tgel:
Tgel measurements are presented in Table 1. All gel-
formation temperatures are around 30 8C and therefore well
above the cooling temperature for the experiments. The
variation is less than 10% of the total temperature drop. The
measurements showed that Tgel is also slightly dependent on
Naþ concentration, but the effect of LBG addition on Tgel is
minor.
3.2.3. Gel strength
The foremost reason for investigating the gel strength
was to control the parameter for determining the influence
on the final drop shape. The ion-dependence of the shape
was expected to be minor since the ions did not affect the
viscosity but only the gel strength, and thereby should only
be crucial in the very last stages of the process. Another
reason for measuring and varying the gel strength was to
pave the way for future investigations not only of drops of
different shapes, but also of the rheological effects in a
dispersion of shaped drops of different flexibility.
The results of the gel strength measurements for the
different cooling temperatures and the three different Naþ
salt concentrations are presented in Table 2. The values of
the storage modulus, G0; was shown to be highly dependent
on the Naþ salt concentrations, with values from below
3 kPa for the lowest concentration to values around 9 kPa
for the highest concentration. Differences in G0 due to
Fig. 3. Different characteristic parts of a drop: (a) the body, (b) the aft limb,
(c) the bow limb and (d) the blob.
Table 1
Viscosity and gel transition temperature and surface tension
k-carrageenan k-carrageenan þ LBG
h 90 8C, 109 mM NaCl 3.7 (mPas) 23 (mPas)
h 60 8C, 109 mM NaCl 6.7 (mPas) 51 (mPas)
h 50 8C, 109 mM NaCl 10 (mPas) 74 (mPas)
h 40 8C, 109 mM NaCl 16 (mPas) 110 (mPas)
h 35 8C, 109 mM NaCla 21 (mPas) 200 (mPas)
Tgel 17 mM NaCl 29 (8C) –
Tgel 109 mM NaCl 32 (8C) 31 (8C)
Tgel 212 mM NaCl 35 (8C) –
a Large deviation occurs between different measurements.
L. Hamberg et al. / Food Hydrocolloids 17 (2003) 641–652 645
![Page 6: Shapes and shaping of biopolymer drops in a hyperbolic flow](https://reader036.fdocuments.us/reader036/viewer/2022081214/5750755b1a28abdd2e992282/html5/thumbnails/6.jpg)
cooling temperature 10 and 20 8C are less than the gel
strength differences due to addition of Naþ ions. The loss
modules, G00; were shown to be more temperature-
dependent than ion-dependent, with lower values for the
high temperature.
3.3. Final drop shape
To find out the diversity of possible shapes that could be
produced under the limited process conditions, the process
and drop parameters: speed, cooling temperature and the
amount of added Naþ salt, were altered. The reason for
exploring different shapes is not only to study the shape as a
final product but also to open up the possibility for future
studies of the special functional properties that could be
related to the final drop shape.
3.3.1. Shape map
Fig. 4 show the final shape of the drop cooled in the
continuous phase with a temperature of 10 8C. The Naþ ion
concentrations in the drop are 17, 109 and 212 mM,
respectively, and the speed of the rollers varies from 5 to
25 rpm. The macrographs of the individual drops in the
figure are picked from a number of trials and are chosen to
be representative and to have good contrast. Fig. 5 shows
drops from the same salt and speed conditions but cooled in
a continuous phase instead with a temperature of 20 8C.
The total number of drops investigated for the figures is 63.
Common to all drops in Figs. 4 and 5 is that the shape of the
drops could be divided into three classes:
† Class I—a solid body with no limb.
† Class II—a solid body with one bow limb to the right.
† Class III—a solid body with one bow limb to the right
and one aft limb to the left.
The first class, with no limb, consists of all the drops
formed with the speed condition of 5 rpm and the drop
formed with the speed of 10 rpm, 17 mM Naþ ion
concentration, and the cooling temperature of 20 8C. In
general, these drops have an ellipsoidal shape with the
longest axis oriented perpendicularly to the centre line.
However, two of the drops in the class differ substantially
from the others. The first drop that differs has the strongest
gel strength, Table 2, and is formed under the conditions
5 rpm, 212 mM Naþ, and 10 8C. This drop has the
ellipsoidal shape, but its longest axis is oriented along
the centre line instead of being perpendicular to it. The
second drop that differs is the triangular drop formed under
the conditions 10 rpm, 17 mM Naþ and 20 8C. This drop
has the weakest gel strength within the class (Table 2).
Within class I the impact of different Naþ concentrations
and cooling temperature is small. The only notable fact is
that the differences are larger for the drops cooled at the
low temperature compared with the ones cooled at the high
temperature.
The second class consists of the three drops with one
limb and all are formed with the speed condition 10 rpm. In
Fig. 4, with a drop cooling temperature of 10 8C, only the
drop with 17 mM Naþ ion concentration, and in Fig. 5,
20 8C, the drops with the two highest Naþ ion concen-
trations belong to the class. All the drops in class II have the
limb at the bow of the drop and the limb ends with a distinct
blob. Although the class has a limited number of objects,
some trends could be found. Within the class, an increase inFig. 4. Shape map of end shapes of the drops regarding speed and Naþ
concentration at 10 8C.
Fig. 5. Shape map of end shapes of the drops regarding speed and Naþ
concentration at 20 8C.
Table 2
Gel strength
Naþ conc.(mM) Temperature (8C) G0 (Pa) G00 (Pa)
17 20 2878 54
17 10 3674 465
109 20 7367 224
109 10 7561 986
212 20 8856 309
212 10 9460 901
L. Hamberg et al. / Food Hydrocolloids 17 (2003) 641–652646
![Page 7: Shapes and shaping of biopolymer drops in a hyperbolic flow](https://reader036.fdocuments.us/reader036/viewer/2022081214/5750755b1a28abdd2e992282/html5/thumbnails/7.jpg)
the Naþ salt concentration tends to favour a higher Feret’s Y
at the expense of Feret’s X. The same trend is found for a
decrease in temperature, so the effects are hard to separate,
since temperature and Naþ ion concentration both affect the
gel strength of the drop (Table 2).
The last and biggest class involves all the drops with two
limbs, and they are formed at the roller speeds 15, 20 and
25 rpm. In addition, the two drops at 10 rpm, with 109 and
212 mM Naþ ion concentration and 10 8C cooling tempera-
ture belong to this class. The limb at the bow of the drop
ends with a distinct blob but the aft does not, or has a very
indistinct one. The bow limb is also longer than the aft limb.
In general, the drops within the class have an increasing
projection perpendicular to the centre line, Feret’s Y ; with
higher roller speed and for increasing temperature. How-
ever, an increase in Naþ ion concentrations both projections
seem to increase. The drops with a low Naþ ion
concentration have a more narrow distribution in Feret’s X
and Y than the drops with higher Naþ ion concentrations.
Furthermore, drops with higher Naþ ion concentration have
thinner limbs.
3.3.2. Fixation time and position
Previous investigations have shown that the final drop
shape is related to the drop fixation position (Hamberg et al.,
2002; Walther et al., 2002). Therefore, Fig. 6 shows a
diagram where the position of fixation and the time to
fixation are presented for the different roller speeds and
cooling temperatures.
The general trend is that the fixation position is shifted
towards the end with increasing speed. The differences in
position found between the lower speed-rates are larger than
those between the higher rates. The fixation positions seem
to reflect the shape classes found in Figs. 4 and 5. The
fixation for class I positioned first around 46 mm and class
III later around 60 mm. The similarities between position
and shape for classes III and I are in agreement with
previous results. The same agreement could not be
concluded for class II, because the two bars representing
10 rpm could not be interpreted in terms of classes as
10 rpm includes drops from all three classes.
Inverse dependence for different roller speeds is found
for the time to fixation. The plain bars show that even if the
drops in class III are fixed at later positions, they still
become fixed earlier than the drops from class I. A higher
roller speed moves the drop more quickly to the end of the
outflow area with the low forces. Furthermore, it is possible
that more elongated drops at the higher speeds could cool
faster due to a larger surface area.
There is no obvious cooling temperature effect found in
the fixation time data. Either the temperature difference is
too small, or competing phenomena could be affecting the
fixation time. A lower cooling temperature results in a
slightly higher cooling rate, which should mean that the
temperature drops sooner to below Tgel for drops cooled at
Fig. 6. Time to fixation, plain, and position of fixation, netted, for different speeds and cooling temperatures: (grey) 10 8C and (white) 20 8C.
L. Hamberg et al. / Food Hydrocolloids 17 (2003) 641–652 647
![Page 8: Shapes and shaping of biopolymer drops in a hyperbolic flow](https://reader036.fdocuments.us/reader036/viewer/2022081214/5750755b1a28abdd2e992282/html5/thumbnails/8.jpg)
10 8C. Nevertheless, the cooling rate is also known to alter
the measured Tgel; (Hermansson et al., 1991) and shift it
towards lower values, resulting in a later decrease to
temperature below Tgel: On time scales corresponding to the
ones in the experiments, and for the complex system that the
small moving drop represents, proper measurements to
establish the superior phenomena are difficult.
3.4. Shaping-process measurements
To explain and conclude in what way the process
parameters results in the different shapes in Figs. 4 and 5, it
is necessary to study more than the pre-set parameters and
the final result. Therefore, measurements of the shaping
process were made in terms of image analysis of the
macrographs from the process with the different parameters.
This makes it possible to interpret the mechanisms involved
in shaping, by following the development of the shape of the
drop systematically, by plotting characteristic features
(Fig. 1).
3.4.1. Position
The shaping shown in Fig. 2 could also be represented as
in Fig. 7, where some characteristic parameters of the drop
are plotted against the position. The parameters shown in the
graph are: area, perimeter, Feret’s X and Y with the
definitions according to Fig. 1. Note that perimeter is scaled
down by a factor of two, and all units are in millimetre
except that the unit for area is mm2. The position refers to
the distance along the centre line, between the centre of the
4-RM and the centre of gravity of the drop.
The perimeter and the Feret’s X curve have the same
characteristics, and they start with a steep increase in the
beginning until around the narrowest gap located at
27.5 mm. In the divergent outflow, the two parameters
decrease until they both flatten out. The two curves level off
at different positions—first, the perimeter around 45 mm,
and later Feret’s X; at approximately 55 mm. There could be
a dual reason for this. First, as the drop is cooled from the
surrounding fluid, a skin could be developed around
the drop, fixing the perimeter. The second and probably
the main reason is revealed in combination with Fig. 2, in
macrograph 6 and onwards. There the bow limb appears to
be flexible, allowing Feret’s X to continue to change after
fixation. Therefore, the levelling of the perimeter indicates
the fixation of the drop shape. However, the levelling of
Feret’s X not only indicates that the fixation of a shape has
passed, but also that rotating movements and flexibility have
ended.
The curve for Feret’s Y ; the projection perpendicular to
the centre line, shows no significant changes before the
narrowest gap. In the divergent outflow, the curve rises as
the drop stretches in the Y-direction, until it flattens just
before the position of 50 mm.
The measured projected area is also shown in Fig. 7. The
changes in this graph are a result of movements
in the z-direction. A larger area results in a ‘thinner’ object.
The increase in area is present up to a stage later than
the maximum length, and the final decrease seems to be at the
same time and position as for the development of the solid
body (Fig. 2). Therefore, the body should have the largest
extension in the z-direction.
3.4.2. Time
The data in Fig. 8 come from the same drop as in Figs. 2
and 7. The graph shows how the shape develops with time
instead of position, and it presents the same characteristic
parameters as above. By plotting the data against time, it is
Fig. 7. Shape curve showing the progress of (-·-) Area, (—) Perimeter/2,
(–-) Feret’s X and (· · ·) Feret’s Y as a function of position. Conditions of
15 rpm, 10 8C and 109 mM Naþ.
Fig. 8. Shape curve showing the progress of (-·-) Area, (—) Perimeter/2,
(---) Feret’s X and (· · ·) Feret’s Y as a function of time. Conditions of
15 rpm, 10 8C and 109 mM Naþ.
L. Hamberg et al. / Food Hydrocolloids 17 (2003) 641–652648
![Page 9: Shapes and shaping of biopolymer drops in a hyperbolic flow](https://reader036.fdocuments.us/reader036/viewer/2022081214/5750755b1a28abdd2e992282/html5/thumbnails/9.jpg)
not only easier to relate the graphs to the macrograph
sequence in Fig. 2, but it also reveals the time scales for the
different parts of the process better. The time scales
involved in the development of the different characteristics
can be hard to decide from Fig. 7 as the drop moves at
different speeds at different positions.
The shaping in Fig. 8 was divided into four zones and
was interpreted in terms of external and internal forces. The
external forces are forces from the flow field acting on and
deforming the drop. The internal forces are the surface
forces that keep the drop spherical and the forces from the
internal plastic and elastic response in terms of drop
viscosity and gel strength.
In the first zone, from time 0 to 2.2 s, the external forces
are superior to the internal forces. The accelerating increase
in drop length is expected, since the acceleration at the
centre line in a hyperbolic flow should be constant in this
zone.
In the second zone, from 2.2 to 3.6, where the drop has its
highest elongation, an unexpected symmetry exists for
perimeter and Feret’s X: This symmetry is surprising,
although an ideal hyperbolic flow field should generate
symmetric external forces. The internal properties of the
drop are neither symmetrical with time, nor should one
expect the result from the surface forces to be symmetrical.
As the drop in the process is chilled down from 90 to 10 8C,
the drop viscosity changes, and the change in the viscosity
ratio should therefore result in unsymmetrical forces. The
surface forces should counteract with the elongation of the
drop before the narrowest gap. However, in the contraction
behind the gap, the surface forces should contribute to, and
instead increase the contraction of the drop and thereby
result in an unsymmetrical graph. From this, it is possible to
conclude that flow has a greater effect on the change of
shape than internal forces.
In the third zone, from 3.6 to 4.5 s, the superior force
goes from being external in the beginning to internal at the
end. This is where the fixation occurs. The total hold-up
time is ,1 s, and the forces that are involved are gel
formation, viscosity increase, and lowering of the external
forces. The first two involve the temperature and the heat
transfer out from the drop. One could only speculate about
the temperature, since the temperature gradient is high as it
decreases with an overall gradient higher than 10 8C/s, and
is therefore hard to measure. The fact that the perimeter
reached its final value half a second earlier could point to an
uneven temperature profile in the drop. Hence, the earlier
fixation of the perimeter could be interpreted as a skin
formation. The equal time to a steady value for area and
Feret’s Y strengthen the assumption that the drop is fixed at
4.5 s, although the graph for Feret’s X has not reached its
final value.
In the final zone, from 4.5 s to the end of the
measurement, all graphs but the one for Feret’s X are
stable. Feret’s X indicates that there is still some flexibility
left in the drop until at least 5.6 s. In the end, the external
forces acting on the drop could be so small that they do not
even show or indicate the flexibility of the drop.
3.4.3. Speed
Fig. 9 presents Feret’s X; the projection length of the drop
along the centre line versus its centre position, for different
roller speed rates from 5 to 25 rpm. The curves are taken
from experiments with a drop Naþ salt concentration of
109 mM and a cooling temperature of 10 8C. They therefore
correspond to the final drop shapes in the middle row in
Fig. 4. For all speed rates, the curves show that the drop
elongates until a position fore the centre of the drop around
29 mm. Hence, the centre has already passed the narrowest
gap at 27.5 mm when the drop has its maximum elongation.
In the outflow, all drops contract and the curves drop and
flatten out in the end indicating a fixation of the drop.
The trends seen in Fig. 9 are expected; an increase in
speed results in a higher maximum length during the first
stages of the process. In addition, the slope in the increasing
part follows the trend of higher values at higher speed; this
is also found in the descending part. The slopes and the
outline of the graphs again seem rather symmetrical around
the narrowest gap. See discussion earlier.
The interpretations of the later stages of the curves are
hazardous because the drop could start to rotate and some of
the limbs are flexible after their fixation. However, the
tendency for the higher speed-rates is that, although the
differences are large at the narrowest gap, the differences
subside and are small at the end of the process. This leads to
the conclusion that high deformation does not alter the
cooling substantially.
3.4.4. Temperature
Temperature is an important factor in the shaping process
when using a cold set biopolymer for preserving the shape.
It is not only by the direct affect of determining when and
Fig. 9. Effect of different roller speeds on Feret’s X: 109 mM Naþ and
10 8C. (þþ ) 5 rpm, (· · ·) 10 rpm, (-·-) 15 rpm, (---) 20 rpm, (—) 25 rpm.
L. Hamberg et al. / Food Hydrocolloids 17 (2003) 641–652 649
![Page 10: Shapes and shaping of biopolymer drops in a hyperbolic flow](https://reader036.fdocuments.us/reader036/viewer/2022081214/5750755b1a28abdd2e992282/html5/thumbnails/10.jpg)
where the gel formation in the drop should occur in the
process, but also by changing other important parameters
like viscosity/viscosity ratio and surface tension. Therefore,
a close monitoring of the drop-temperature distribution
inside the drop during the shaping would be of great
relevance and help when trying to control and analyse the
process. Unfortunately in this investigation it was not
possible to follow the temperature development in the drop
by measurements as the process was on a timescale of
seconds.
The different cooling temperatures influence the system
in different ways. First, a lower temperature increases the
cooling rate of the drop, and therefore the drop cooled at
10 8C should be fixed earlier due to an earlier passing of Tgel
than for a drop cooled at 20 8C. Second, the faster cooling of
the drop should also lead to a lower temperature of the k-
carrageenan-solution inside the drop, at a certain position. A
lower temperature corresponds to a higher drop viscosity,
Table 2, and therefore a different viscosity ratio. Finally, the
different cooling temperatures could also affect the viscosity
of the continuous phase.
Effects of temperature on Feret’s X and Y for the process
are shown in Fig. 10, where the solid and dotted graphs
represent a cooling temperature of 10 and 20 8C, respect-
ively. The graphs for Feret’s X start together, but during the
increase in the inlet between the rollers, the graph with a
cooling temperature of 10 8C increases more and reach a
higher maximum value. The graphs keep approximately this
difference towards the end during the decrease and finish at
a higher value than the graph with a cooling temperature of
20 8C. For Feret’s Y ; the curves have approximately the
same value and profile until the final stages, where the lower
cooling temperature curve flattens out earlier at a position
below 50 mm and the other does not flatten until around
55 mm.
The temperature effect on drop and oil viscosity oppose
each other in terms of shaping, and it is clear from the
graphs of Feret’s X that the last effect is more dominant in
the beginning. The higher viscosity in the continuous phase,
and thereby lower viscosity ratio, elongates the drop more at
the beginning of the process. It is also reasonable to believe
that the drop viscosity changes due to different temperatures
are less at higher temperatures far from Tgel; than for
temperatures in Tgel surroundings. This increase in tem-
perature dependence near Tgel could explain the behaviour
of the 10 and 20 8C Feret’s X-curves shown in Fig. 10. In the
figure, the graphs do not end at the same level, as the
viscosity-ratio-change increases at the end of the process.
The effect of the viscosity increase of the continuous
phase should also be notable for Feret’s Y and lead to a
similar difference between these curves. No such effect is
visible but instead the first effect, earlier passing of Tgel;
results in a shorter drop in the Y-direction. Hence, a
decrease in cooling temperature leads to an increase in the
X-direction but a decrease in Y-direction.
3.4.5. Viscosity
LBG was added to the solutions for some samples to
investigate the effect of the viscosity ratio, as the change in
drop viscosity was shown to have a complex effect when the
viscosity change was temperature-induced. Tgel for k-
carrageenan blended with LBG is known not to differ
much from Tgel for pure k-carrageenan (Stading &
Hermansson, 1993). The effective difference in viscosity
was found to be ,7 at 60 8C and ,10 times higher at 35 8C
than the original pure k-carrageenan solution for the
solution with LBG (Table 1). The rheological characteris-
ation also showed that Tgel for the blended solution was also
only slightly affected (Table 1).
The different curves shown in Fig. 11 are the progress of
Feret’s X and Y plotted as a function of the centre position.
Fig. 10. Effect of different cooling temperatures: (---) Feret’s X-10 8C, (—)
Feret’s X-20 8C, (· · ·) Feret’s Y-10 8C, (-·-) Feret’s Y-20 8C. Conditions of
109 mM Naþ and 15 rpm.
Fig. 11. Effect of different viscosities on Feret’s X and Y : 109 mM Naþ and
15 rpm: (—) Pure k-carrageenan and (· · ·) k-carrageenan mixed with LBG.
L. Hamberg et al. / Food Hydrocolloids 17 (2003) 641–652650
![Page 11: Shapes and shaping of biopolymer drops in a hyperbolic flow](https://reader036.fdocuments.us/reader036/viewer/2022081214/5750755b1a28abdd2e992282/html5/thumbnails/11.jpg)
All curves are from drop conditions of 109 mM Naþ salt
concentration, 10 8C as cooling temperature and a roller
speed of 15 rpm. The curves are two replicas for pure k-
carrageenan and k-carrageenan with LBG, respectively.
The two Feret’s X curves with pure k-carrageenan are
superior to the k-carrageenan/LBG curves for positions
around the narrowest gap in the 4-RM. Close to the start and
at the end, there are no large differences. All curves in Fig.
11 have a similar pattern except for the altitude of the
maximum and show similarity to the difference among the
speed graphs in Fig. 9. Hence, it could be reasonable to
believe that the viscosity ratio and speed effects could be
exchangeable at the same temperature.
This type of viscosity change shows a different pattern
from the temperature-induced change. The differences are
found in the later stages of the process, where the
differences between the curves decrease in Fig. 11 instead
of being nearly constant in Fig. 10. The most likely reason is
that the differences in the viscosity ratio are similar both at
the start and at the end in Fig. 11.
Perpendicular to the centre line, Feret’s Y shows no
changes until the outflow, after the narrowest gap around
30 mm. There, a steady increase become visible and
continues until the curves reach a stable plateau at the
end, just before the 50 mm position. Again, as for Feret’s X;
the curves with the lowest viscosity reach the highest values
for Feret’s Y :
The multiple Feret’s X and Y curves also show the
differences within a group of curves created under the same
conditions. The differences shown for pure k-carrageenan
are among the highest ones found in the measurements,
while the differences for the blended solution are among the
smallest. The individual differences within samples with the
same conditions also tend to be larger for a cooling
temperature of 10 8C than for 20 8C and for samples with
higher Naþ concentrations. This could be explained that a
more rapid cooling, in combination with the thin limbs
created from drops with higher gel strength, makes the
individual drop more sensitive to the differences that could
be found in k-carrageenan solutions.
The shape reproducibility in Figs. 4 and 5 was
surprisingly high, and even under the most shaped
conditions the differences due to changes in the process
parameters were larger than the differences within the
replicates.
3.4.6. Drop size
The influence of different drop sizes is an important issue
for the shaping that has been omitted in this study. A change
in size will not only affect the deformation but also fixation
of the drop, since the cooling time is influenced by the drop
size. In this investigation, the drops have an initial diameter
of 2.7 mm, and therefore too large for many applications.
Nevertheless, the questions addressed concerning shaping in
the small sub-millimetre range have proved to be hard to
answer due to limited observability (Hamberg et al., 2002).
The limitations in observability on the time scales noted
here are even bigger on the smaller scale. One could expect
that some of the observations and conclusions could be
applied to the downscaled systems as well.
Although the drop sizes are in the millimetre range,
surface phenomena are expected to play a role in the general
shaping mechanisms. Here, as the effect of differences in
surface tension was believed to be minor between all
different investigation set-ups and since the surface tension
was difficult to measure on the time scales involved in this
process, surface measurements were omitted in this
investigation. Characterisation will be difficult, as the
surface measurements would have featured problems
when the biopolymer solution in the drops changes from a
viscous solution to a visco-elastic gel through gel formation.
Another unsatisfactory circumstance is that no steady state
is believed to occur at the surface on the short time scale of
the shaping process. Nevertheless, the surface properties
and function must be investigated to fully master the
shaping. Therefore, the role of surfactants in shaping and
surface tension will be studied further as a challenging task
for the future.
4. Conclusions
The measurements of the shaping process, in terms of
image analysis of the different parameters on the
macrographs from the process, was found to be a good
new tool for investigating how the process parameters
determined the different shapes. By plotting the charac-
teristic features of the drop with position and time, it was
possible to interpret the mechanisms and time scales
involved in shaping. The fixation occurs on a time scale
of ,1 s, and the mechanisms involved are gel formation,
viscosity increase, and lowering of the external forces.
With the same hyperbolic flow, very different shapes
could be achieved with small process changes and with
high reproducibility. The final shape of all the drops
produced could be graded into three classes. The first
class contained all drops with just a solid body. The
second class contained the drops with a body and one
bow limb. In the last shape-class, the drops had a solid
body with one bow limb and one aft limb. The class to
which the drop belonged seemed to be correlated to the
position in the flow where the drop was fixed. Earlier
position gives the simpler shape. Time to fixation and
fixation position had an inverse behaviour. However, the
differences between the characteristic features inside a
class did not correlate with position or time.
The fixation of the characteristic features, perimeter,
area, Feret’s X and Y ; does not occur at the same time
and position. The first feature to be fixed is the
perimeter, and this is believed to be due to skin
formation of the drop. Then comes the synchronous
fixation of area and Feret’s Y : This fixation point has
L. Hamberg et al. / Food Hydrocolloids 17 (2003) 641–652 651
![Page 12: Shapes and shaping of biopolymer drops in a hyperbolic flow](https://reader036.fdocuments.us/reader036/viewer/2022081214/5750755b1a28abdd2e992282/html5/thumbnails/12.jpg)
been used as the end of the shaping. Under some
conditions, the fixation of Feret’s X is separated from the
others due to flexibility of the drop. Flexibility has not
been included as a part of the shaping.
For the different process-parameters investigated, a
change in speed and viscosity ratio affect the process in
the same straight way as if the viscosity ratio is changed at a
constant temperature. If the change in viscosity ratio is
temperature-induced, the effect is complex. Various
additions of Naþ ions to alter the drop flexibility do not
change the process or the shapes much.
The set-up and technique have proven to be a good new
tool for studying and producing differently shaped drops.
This tool could be applied in the future to study the
rheological impact of drop shape and drop flexibility on
smaller drops suitable for real food products.
Acknowledgements
This work has been carried out with financial support
from the Swedish LiFT program (Future Technologies for
Food Production) and the EU project Structure engineering
of emulsions by micro-machined elongation flow proces-
sing (QLK-CT-2000-01543) within the RTD programme
Quality of Life and Management of Living Resources.
Special thanks to Bernhard Walther, SIK.
References
Djalili-Moghaddam, M. (2001). Rheological measurements on fibre
suspensions. Goteborg: Chalmers University of Technology.
Erwin, L. (1991). Principles of laminar fluid/fluid mixing. In C.
Rauwendaal (Ed.), Mixing in polymer processing (pp. 1–16). New
York: Marcel Dekker.
Feng, J., & Leal, L. G. (1997). Numerical simulations of the flow of dilute
polymer solutions in a four-roll mill. Journal of Non-Newtonian Fluid
Mechanics, 72, 187–218.
Hamberg, L., Walkenstrom, P., & Hermansson, A.-M. (2002). Shaping of
gelling biopolymer drops in an elongation flow. Journal of Colloid and
Interface Science, 252, 297–308.
Hamberg, L., Walkenstrom, P., Stading, M., & Hermansson, A.-M.
(2001). Aggregation, viscosity measurements and direct observation of
protein coated latex particles under shear. Food Hydrocolloids, 15,
139–151.
Hermansson, A.-M., Eriksson, E., & Jordansson, E. (1991). Effects of
potassium, sodium and calcium on the microstructure and rheological
behaviour of k-carrageenan gels. Carbohydrate Polymers, 16,
297–320.
Joung, C. G., Phan-Thien, N., & Fan, X. J. (2001). Direct simulation of
flexible fibers. Journal of Non-Newtonian Fluid Mechanics, 99, 1–36.
Joung, C. G., Phan-Thien, N., & Fan, X. J. (2002). Viscosity of curved
fibres in suspension. Journal of Non-Newtonian Fluid Mechanics, 102,
1–17.
Petrie, C. J. S. (1999). The rheology of fibre suspensions. Journal of Non-
Newtonian Fluid Mechanics, 87, 369–402.
Rallison, J. M. (1984). The deformation of small viscous drops and bubbles
in shear flows. Annual Review of Fluid Mechanics, 16, 45–66.
Rochas, C., & Rinaudo, M. (1980). Activity coefficients of counterions and
conformation in k-carrageenan systems. Biopolymers, 19, 1675–1687.
Rochas, C., & Rinaudo, M. (1984). Mechanism of gel formation in k-
carrageenan. Biopolymers, 23, 735–745.
Rochas, C., Rinaudo, M., & Landry, S. (1989). Relation between the
molecular structure and mechanical properties of carrageenan gels.
Carbohydrate Polymers, 10, 115–127.
Stading, M., & Hermansson, A.-M. (1993). Rheological behaviour of
mixed gels of k-carrageenan-locust bean gum. Carbohydrate Polymers,
22, 49–56.
Stokes, J. R., Wolf, B., & Frith, W. J. (2001). Phase-separated biopolymer
mixture rheology: prediction using a viscoelastic emulsion model.
Journal of Rheology, 45(5), 1173–1191.
Stone, H. (1994). Dynamics of drop deformation and breakup in viscous
fluids. Annual Review of Fluid Mechanics, 26, 65–102.
Taylor, G. I. (1934). The formation of emulsions in definable fields of flow.
Proceedings of the Royal Society of London, A146, 501–523.
Walther, B. (2001). Deformation and break-up of drops and filaments in
laminar flow—A literature review. SIK-report 2001 No. 682, Goteborg:
SIK.
Walther, B., Walkenstrom, P., Hermansson, A.-M., Fischer, P., & Windhab,
E. (2002). Flow processing and gel formation—a promising combi-
nation for the design of the shape of gelatine drops. Food
Hydrocolloids, 16, 633–643.
Wolf, B., Frith, W. J., Singeleton, S., Tassieri, M., & Norton, I. T. (2001).
Shear behaviour of biopolymer suspensions with spheroidal and
cylindrical particles. Rheologica Acta, 40, 238–247.
Wolf, B., Scirocco, R., Frith, W. J., & Norton, I. T. (2000). Shear-induced
anisotropic microstructure in phase-separated biopolymer mixtures.
Food Hydrocolloids, 14, 217–225.
L. Hamberg et al. / Food Hydrocolloids 17 (2003) 641–652652