You are on page 1of 6

Rheol Acta

DOI 10.1007/s00397-006-0089-z ORIGINA L CONTRIBUTION

Julien Mougel
Oscar Alvarez
Aging of an unstable w/o gel emulsion
Christophe Baravian
François Caton
with a nonionic surfactant
Philippe Marchal
Marie-José Stébé
Lionel Choplin

Received: 29 June 2005 O. Alvarez . P. Marchal . L. Choplin


Sci 112:427–437, 1986) model does
Accepted: 6 February 2006 Génie chimique des milieux not apply. A dimensional analysis
# Springer-Verlag 2006 rhéologiquement complexes BP 451, based on the hypothesis of dominant
1 rue Grandville, van der Waals forces is proposed for
Nancy, Cedex 54001, France nonionic surfactants, which is in good
M.-J. Stébé agreement with experimental data. We
Groupe Physico-Chimie des colloïdes also show that the measured average
CNRS UMR 7565, droplet volume increases linearly with
Université Henri Poincaré BP 239, time and that the coalescence rate
Vandoeuvre-lès-Nancy,
Cedex 54506, France strongly depends on the volume frac-
tion in relation with different topolo-
Abstract We study an unstable gical conformations of droplets.
This work was presented at the congress highly concentrated emulsion of water
AERC 2005 in Grenoble, France. droplets in oil with a nonionic sur- Keywords Concentrated emulsion .
factant. A technique of light diffusion Coalescence rate . Nonionic
J. Mougel (*) . C. Baravian . F. Caton surfactant . Elastic modulus .
Laboratoire d’énergétique et de coupled to a rheometer allows simul-
mécanique théorique et appliquée taneous measurement of average Water-in-oil emulsion
CNRS UMR 7563, droplet radius R and emulsion shear
2 avenue de la Forêt de Haye BP 160, elastic modulus G0 during time. Over
Vandoeuvre-lès-Nancy,
Cedex 54501, France the studied range of volume fraction
e-mail: julien.mougel@ensem.inpl-nancy.fr (from 71 to 95%), we show that
Tel.: +33-3-83595710 Princen and Kiss’ (J Colloid Interface

Introduction Another coalescence model is Ostwald ripening, which


occurs because of the polydispersity of the system. Small
Long time stability of concentrated emulsion is a widely droplets disappear for the profit of bigger ones because of
studied and complex field of investigation (e.g., Princen the Laplace pressure difference and due to the dispersed
and Kiss 1986; Bibette et al. 1999; Langevin 1998; Taylor fluid diffusion through the continuous phase (Langevin
1995). Indeed, three main mechanisms were shown to lead 1998). The speed of this mechanism depends on the
to droplet coalescence. Firstly, the Schmoluchowski model diffusion coefficient of the dispersed phase in the other and
(see, for example, Langevin 1998) is built on the hypoth- the interfacial tension γ (see Taylor 1995; Walstra 1996)
esis that two droplets in contact coalesce. The main also leading to an evolution of R with t1=3 :
consequence is that the number N of droplets in volume For emulsions with a concentration that is higher than
unit decreases during time leading to a complete phase the close packing fraction, droplets are highly deformed
separation. This evolution is given by dN =dt / N 2 and and separated by a thin film of continuous phase (like in
implies that the droplet radius follows R  t 1=3 : foams). The coalescence happens when this film breaks up.
This complex mechanism involves, for emulsions, the (99% Purum) and the surfactant, dispersed in oil, is a
stability of the interface between oily and aqueous phases sorbitan monooleate [Span80, hydrophile-lipophile bal-
(see Bibette et al. 1999; Langevin 1998). Thermal agitation ance (HLB)=4.3] purchased from Aldrich. The refractive
can create fluctuations of the film thickness and amplitude indices of water and oil phase are 1.33 and 1.443,
increases when droplets get closer. The fluctuations of respectively. The composition of the samples in mass
concentration in the layer of surfactant are another source fraction is between 74 and 96% for the aqueous phase and
of instability. the surfactant/oil weight ratio is 0.81 to keep the same
In parallel to the modeling complexity, experimental surface tension γ ¼ 2:8 mN/m for all samples. Indeed, the
studies are difficult due to the nonstability of the system surfactant concentration is, under this condition, greater
and its opacity to visible light. Indeed, because of the than the critical micellar concentration for all the samples
turbidity of concentrated suspensions, direct microscopy or where the volume fractions in dispersed phase are 71, 73,
small angle light scattering techniques cannot be used 75, 78, 85, 90, 92, 93.5, and 95%. The total quantity of
easily. Moreover, in many cases, the samples are unstable each sample prepared is 84.1 g.
with respect to dilution or can simply not be diluted in The emulsions are prepared at 27 C by incorporation of
water [e.g., water-in-oil (w/o) emulsions or foams]. In the aqueous phase into the oil phase containing the
contrast, multiple light scattering systems, such as steady surfactant (semi-batch system). The emulsification process
light transport (Baravian et al. 2004) or dynamic light was divided in three steps: (1) dispersion of Span80 in oil
scattering (Brown 1993), are well suited for concentrated and homogenization (10 min at 500 rpm), (2) aqueous
and opaque suspensions. phase incorporation at a flow rate of 4 g/min and an
Most experimental studies on emulsion stability uses agitation speed of 500 rpm, and (3) homogenization of the
ionic surfactants, making electrostatic repulsion forces emulsion during 5 min at the same agitation speed. The
dominant. In this case, the heuristic relation proposed by agitation system used was a Turbo test 33/750 P (Rayneri
Princen and Kiss (1986) is verified (e.g., Princen and Kiss Groupe VMI) equipped with a speed control to ensure a
1986; Ravey et al. 1998; Taylor 1996) or slightly modified constant stirring rate independent on the evolution of the
(Mason 1999; Bressy et al. 2003). To our knowledge, only emulsion viscosity. Our samples are prepared with a 45
Pal (1999) performed droplet size and elastic modulus G0 pitched-blade turbine of 50 mm diameter in a cylindrical
measurements on a nonionic emulsion.1 On his figure 9, we glass tank (70 mm diameter). The turbine is positioned on
see that the variation of experimental G0 R=γ with the the free surface of the oil phase before aqueous phase
volume fraction is not well described by Princen and Kiss’ incorporation (see, e.g., Shervin et al. 1991).
model. In the case of a nonionic surfactant, because there is The measurement device is a steady light transport
no electrostatic charge on the surface of droplets, the apparatus (see Baravian et al. 2004, 2005) placed on an
dominant behavior of the system could be due to van der AR1000 rheometer from TA Instrument. This apparatus
Waals forces (see Bibette et al. 1999). A natural question allows simultaneous light transport length l  and
arising from these observations is whether the nonionic or rheological measurements. Each experiment begins
ionic characteristic of the surfactant does fundamentally roughly 2 h after the end of the preparation process and
change the properties of the emulsion. lasts for about 64 h.
To explore this possibility, we study a gel emulsion with The transport length l  is related to Mie theory through
a nonionic surfactant different from that of Pal (1999). For φl ¼ ð43 πR3 Þ=½Csca ðx; mÞð1  gðx; mÞÞ where x ¼ kR

various concentrations in dispersed phase, we follow the with k as the wave number and m as the ratio of the
temporal evolution of droplet radius and sample shear particle to medium refractive indices. Csca and g are the
elastic modulus. In the first part of this work, we describe scattering cross-section and the anisotropy factor, respec-
the chemical composition of the samples, the preparation tively. Knowing the optical properties of the emulsion
process, and briefly the measurement method. The second phases and the volume fraction φ of droplets, Mie theory
part deals with the coalescence mechanism and finally, a directly relates the transport length to the average radius
new dependence of the elastic modulus with the average of droplets. Under our experimental conditions
droplet size and the volume fraction is proposed on the (m ’ 0:92; 100 < φl  < 700 μ m), the average droplet
hypothesis of dominant van der Waals forces. radius R (in μ m) shows a linear dependence with φl 
(in μ m).2 This dependence is well described by R ¼ α
φl  β where α ¼ 0:038 and β ¼ 0:766 μ m are the best-
Materials and method fit parameters using a standard least square method.
Water was deionized and milli-Q filtrated with a Millipore
apparatus. The oil is a Dodecane obtained from Labosi 2 Although our samples are highly concentrated, dependent diffusion

is not involved because the average distance between droplets center


1 The following occurrences of “nonionic (ionic) emulsion” will be is always big enough compared to the laser wavelength (Tsang et al.
denoted by an emulsion in which the surfactant is nonionic (ionic). 2000).
Rheological test consists of the application of oscillation 100
0.01 0.1 1

stress of 1 Pa at a frequency of 1 Hz to the sample, which is 1−φ


30
much smaller than the droplet coalescence rate. As shown
on Fig. 1, G0 decreases with time and increases with re- 10

R0 (µm)
95%
spect to the volume fraction. We follow the temporal
evolution of two properties of the emulsion: the elastic 1

R (µm)
20
modulus and the average radius of droplets. 93.5%

92%

Coalescence 10
93.5% (stabilized)
We present on Fig. 2 the results obtained with the steady
light transport apparatus. First, we observe that the average 71%
size of droplets increases during time. Two tests were done 0
at a volume fraction of 95%, showing an excellent 1 10 100
reproducibility (see Fig. 2). To verify that the measured t-t0 (hours)
size evolution is significant, a stabilized emulsion at φ ¼
93:5% was prepared by adding NaCl (about 2.5% weight Fig. 2 Time evolution of average particle radius R for various
of the aqueous phase). Compared to the nonstabilized one, volume fractions.  Experimental points, solid lines are the best fits of
Eq. 1. þ Stabilized emulsion at 93.5% in volume fraction. Insert: 
the evolution of the stabilized emulsion is negligible. We initial radius as a function of volume fraction , solid line
also notice that whatever the volume fraction, the same 0:795=ð1  Þ. For clarity reason, not all the data are plotted
agitation rate is used during the preparation process so that
the first measured radius R0 is different for each sample.
The graphic in the insert of Fig. 2 shows that the variation
of R0 with the volume fraction φ of water droplets is where ω is the coalescence rate, R0 and t0 correspond to
proportional to 1=ð1  φÞ with a factor of 0:795μ m in the the first measurement. A standard least square adjustment
whole range of volume fraction. We verified that every- of Eq. 1 to experimental curves (see Fig. 2) allows the
thing else being equal, a different R0 (obtained by determination of ω for each sample. This coalescence rate
changing the agitation rate) does not significantly change is known, for a given volume fraction, to be strongly
the kinetics. dependent on the HLB (Aronson and Petko 1993). Figure 3
For all samples (from φ ¼ 0:71 to 0.95), we verify that shows that ω increases with the volume fraction until
the average radius follows a t 1=3 evolution law: φ1 ’ 0:75, then remains constant until φ2 ’ 0:90 , and
finally increases again. φ1 and φ2 could be related to the
RðtÞ ¼ ½R30 þ ω  ðt  t0 Þ1=3; (1) clustering of droplets from the close maximum packing
fraction without significant deformation to a polyhedral
1000
shape when φ2 is reached.
95% 0.1
800 93.5%

92%
600
85%
G’ (Pa)

0.01
ω (µm3/s)

78%
400

0.001
200

0
1 10 100 0.0001
t-t0 (hours) 0.7 0.8 0.9 1
Volume fraction φ
Fig. 1 Evolution of the samples elastic modulus G0 at various
volume fractions. For clarity reason, not all the data are plotted Fig. 3 Variation of the coalescence rate ! with the volume fraction
Fig. 4 G0 vs R. Except for the 7
sample at 73% in volume
fraction, the basic plot of the Slope -2
data shows that, in our case,
G0 is correlated with 1=R2 6.5 92%

85%

Ln G’ (Pa)
6
75% 95%

5.5
90%
73% 93.5%

5
78%
Slope -1
4.5
1 1.5 2 2.5 3 3.5
Ln R (µm)

Elastic modulus than 75%, corresponding to the first transition regime in


Fig. 3. Another experimental observation is the divergence
Figure 4 shows the elastic modulus (G0 ) dependence with of G0 R2 when φ approaches 1 (Fig. 5). This divergence is,
the average droplet radius R for different volume fractions. again, not taken into account in Princen’s model.
This figure is obtained from Figs. 1 and 2. It is clear that the From dimensional considerations, we can express the
well-known Princen and Kiss (1986) model does not apply elasticity as follows
to our experimental data because G0 does not vary like 1=R
(the corresponding slope 1 is plotted on Fig. 4). Indeed, W
G0  (2)
we find that G0 is better correlated to 1=R2 (see slope 2 V
on Figs. 4 and 5), especially for volume fractions greater
where W has the dimension of an energy and V the
100
dimension of a volume. To take into account the divergence
Present work of G0, we must consider V as the continuous phase volume
Pal’s experiments by unit droplet Vf :
φ /(φc − φ )
1φ
· Princen for the present work Vf / R3 : (3)
2
Princen for Pal φ
G' R
10
2π AD0γ The energy W therefore has to be proportional to R to
describe the correlation of G0 with 1=R2.
In a general point of view, elasticity of concentrated
dispersions was closely related to the pair interaction
potential energy (e.g., Quemada and Berli 2002). On the
other hand, in the case of emulsions concentrated above the
1
0.7 0.8 0.9 1
critical random packing volume fraction, the elasticity is
related to the interfacial energy of deformed droplets
Volume fraction φ (Princen and Kiss 1986). Nevertheless, this local elasticity
Fig. 5 G0 R2 =ð2AD0 Þ vs . We compare our results () and Pal’s is transmitted at the sample scale through the interaction
experimental data () to the model of Eq. 7 (solid line). The value potential between droplets (electrostatic repulsion in all
of the parameters are A ¼ 3:66 and c ¼ 0:96. For Pal’s work, we cases where Princen model is verified). In the following,
take R ¼ 1:675 m (R½4; 3 from the size distribution) and  ¼ we propose to explore the possibility that the dominant
0:77 mN/m. The dashed-dotted and dashed lines represent
Princen’s model for the present work at t ¼ 2 h and for Pal, interaction energy W is due to van der Waals forces in the
respectively case of nonionic surfactant.
Amplitude of van der Waals energy between two points of Fig. 4 are drawn. The best fit to our
identical droplets in a different surrounding medium can experimental data provides the following values: A ¼ 3:
be expressed as (Israelachvili 1991): 66 and φc ’ 0:96 , taking D0 ¼ 30 nm (Dukhin and
Goetz 2004) as the micelle diameter.
AH R The square symbols on the Fig. 5 shows the experi-
W ¼ ; (4)
12D mental data of Pal (1999). With the same φc , the same
value of A and D0 ¼ 6 nm (Molina-Bolivar et al. 2002),
where D is the distance between the surfaces of the two Pal’s data superimposes on the present work results without
droplets and AH ¼ 24πγD20 is the Hamaker constant with any adjustment parameters, although a different nonionic
D0 the intermolecular distance. In our case, as the droplets surfactant (Triton X-100, HLB=13.5, γ ¼ 0:77 mN/m)
are in contact, we assume that D does not depend on the was used in an oil-in-water (o/w) emulsion.
volume fraction and is of the order of the intermolecular
distance, i.e., D  D0 . Therefore we write
Conclusion
W ¼ 2πγD0 R: (5)
The steady light transport technique allows the simulta-
neous measurement of the temporal evolution of the
Equations 2, 3, and 5 lead to the following expression sample shear elastic modulus G0 and droplet average radius
for G0: R for samples at various volume fractions φ.
We find that the coalescence law R  t 1=3 applies in our
2πγD0 φ system for the whole range of volume fractions. The
G0 / : (6)
R2 ð1  φÞ coalescence rate shows different domains in relation with
the clustering of droplets.
We see that this approach leads to a divergence at φ ¼ 1. The relation that we found between G0, R, and φ (Eq. 7)
However, a better empirical relation in comparison with differs from Princen and Kiss’ (1986) model, probably due
experiments is to the use of a nonionic surfactant (Span80) in the
preparation of our samples. Based on dominant van der
2πγAD0 φ Waals interactions, a tentative model is built up. This model
G0 ¼ ; (7) also describes other results obtained with another nonionic
R2 ðφc  φÞ
surfactant for an o/w emulsion (Pal 1999).
Further investigations with other nonionic surfactants
where A is the proportionality coefficient. The critical have to be performed to verify if this behavior is only valid
volume fraction φc might account for the droplet surface in some particular cases or if it is a more general one.
curvature and very short-range interactions. This model
is valid for a wide range of volume fractions and droplet Acknowledgement We thank Daniel Quemada for useful remarks
sizes, as shown on Fig. 5 on which all experimental and comments.

References
Aronson MP, Petko MF (1993) Highly Bibette J, Leal-Calderon F, Poulin P (1999) Langevin D (1998) La coalescence. Bulletin
concentrated water-in-oil emulsions: Emulsions: basic principles. Rep de la SFPS 115:9–13
influence of electrolyte on their prop- Prog Phys 62:969–1033 Mason TG (1999) New fundamental con-
erties and stability. J Colloid Interface Bressy L, Hebraud P, Schmitt V, Bibette J cepts in emulsion rheology. Curr Opin
Sci 159:134–149 (2003) Rheology of emulsions stabi- Colloid Interface Sci 4:231–238
Baravian C, Caton F, Dillet J (2004) Steady lized by solid interfaces. Langmuir Molina-Bolivar JA, Aguiar J, Carnero Ruiz
light diffusion application to rheology: 19:598–604 C (2002) Growth and hydratation of
a new tool for the characterization of Brown W (1993) Dynamic light scattering: triton x-100 micelles in monovalent
concentrated suspensions. Rheol Acta the method and some applications. alkali salts: a light scattering study.
43:427–432 Claredon, Oxford J Phys Chem B 106:870–877
Baravian C, Caton F, Dillet J, Mougel J Dukhin AS, Goetz PJ (2004) Ionic proper- Pal R (1999) Yield stress and viscoelastic
(2005) Steady light transport under ties of so called non-ionic surfactants properties of high internal phase ratio
flow: characterization of evolving in non-polar liquids. http://www. emulsions. Colloid Polym Sci
dense random media. Phys Rev E 71 dispersion.com/pages/newsletter/ 277:583–588
(066603) articles/Newsletter15a. Cited Nov
2005
Israelachvili J (1991) Intermolecular and
surface forces, 2nd edn. Academic.
ISBN 0-12-375181-0 (Chapter 11)
Princen HM, Kiss AD (1986) Rheology of Ravey JC, Stébé MJ, Sauvage S (1998) Tsang L, Kong JA, Ding KH, Ao CO (2000)
foams and highly concentrated emul- Water in fluorocarbon gel emulsion: Scattering of electromagnetic waves:
sions. J Colloid Interface Sci 112: structure and rheology. Colloids Surf numerical simulations, vol. 2. Wiley,
427–437 14:237–257 New York. ISBN 0-471-38800-9
Quemada D, Berli C (2002) Energy of Shervin CR, Raughley DA, Romaszewski (Chapter 8)
interaction in colloids and its implica- RA (1991) Flow visualization scale up Walstra P (1996) Emulsion stability. In:
tions in rheological modeling. Adv studies for the mixing of viscoelastic Becher P (ed) Encyclopedia of emul-
Colloid Interface Sci 98:51–85 fluids. Chem Eng Sci 46:2867–2873 sion technology, vol. 4. Dekker, New
Taylor P (1995) Ostwald ripening in emul- York, pp 1–62
sions. Colloids Surf A 99:175–185
Taylor P (1996) The effect of an anionic
surfactant on the rheology and stability
of high volume fraction o/w emulsion
stabilized by pva. Colloid Polym Sci
274:1061–1071

You might also like