W double emulsion by polysaccharides as weak gels

Available online at www.sciencedirect.com

Colloids and Surfaces A: Physicochem. Eng. Aspects 316 (2008) 46–54

Stabilisation of W/O/W double emulsion by polysaccharides as weak gels M. Benna-Zayani a,∗ , N. Kbir-Ariguib a , M. Trabelsi-Ayadi a , J.-L. Grossiord b a

Laboratoire d’applications de la chimie aux substances et ressources naturelles et a` l’environnement, Facult´e des Sciences de Bizerte, 7021 Zarzouna, Bizerte, Tunisia b Laboratoire de Physique Pharmaceutique, Facult´ e de Pharmacie, Universit´e Paris-Sud 5, rue Jean-Baptiste Cl´ement, 92296 Chˆatenay-Malabry Cedex, France Received 2 February 2007; received in revised form 16 August 2007; accepted 17 August 2007 Available online 23 August 2007

Abstract Aqueous solutions of scleroglucan, carageenan and mixture of xanthan and locust bean gum polysaccharides have been tested as the external aqueous phase in water in oil in water (W/O/W) double emulsion formulations. This has been done in order to stabilize the double emulsions in time by preventing the creaming and eventually the phenomenon of coalescence. For this purpose, rheological properties of polysaccharides aqueous solutions at low concentrations were studied. These so-called weak gels exhibit a yield stress (τ c ). The existence of this threshold was validated by two methods. The measure of shear strain (γ) induced by an increasing steady shear stress (τ) shows a sharp increase at τ ≈ τ c . The oscillatory viscoelastic analysis by imposing an increasing sinusoidal shear stress shows that the viscous modulus (G”) passes through a defined maximum at the critical value of stress. The obtained values of yield stresses are, as required, relatively low and the viscosities of the pseudo gels are sufficiently low which will allow preventing the breakdown of the double droplets during preparation. Moreover, the times necessary for the rebuilding of the tested weak gel after shearing have been estimated to few minutes indicating that they go back rapidly to their initial pseudo three-dimensional network. The W/O/W double emulsions prepared are relatively stable towards creaming especially those prepared with scleroglucan and mixture of xanthan and locust bean gum. © 2007 Elsevier B.V. All rights reserved. Keywords: Multiple emulsion; Stability; Polysaccharides; Creaming; Weak gels

1. Introduction Emulsified systems and particularly double emulsions water in oil in water emulsion (W/O/W) are known to be thermodynamically instable. And one of the major problems is creaming which could result at the end on coalescence and separation. It is well-known that many studies had attempted to slow down creaming of emulsified systems by adding a thickener. However, a high viscosity of the external continuous phase causes a crucial problem on fabrication of the double emulsion because it can cause the breakdown of the oily globules especially because of the relatively high diameter of oily globules. Other authors suggested the use of a colloidal solution or suspension which forms a three-dimensional network in the external continuous phase

∗

Corresponding author. Tel.: +216 98 696 223; fax: +216 72 590 566. E-mail addresses: [email protected], [email protected] (M. Benna-Zayani), [email protected] (J.-L. Grossiord). 0927-7757/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.colsurfa.2007.08.019

of the simple emulsion. But this gelation automatically induces an increase of the external phase viscosity. The ideal for a multiple emulsion will be a material which exhibits, in solution, a three-dimensional network without any higher value of viscosity. This can be achieved using the so-called weak gels which are formed by solution of polysaccharides in water in presence or absence of salt (depending on the nature of the biopolymer). In the last decade, some authors have been interested in the rheological behaviour of the so-called weak gels and their eventual application in food industry such as reconstituted milk or in cosmetic O/W simple emulsions [1–9] in order to stabilize aqueous suspensions or O/W emulsions by forming an extended network which induces high viscosity of the continuous phase at low shear and thus slowing down the droplet motion. All these previous works are relative to simple emulsions and used polysaccharide solutions to thicken the continuous phase of an emulsion. And to the best of our knowledge, no study on slow down creaming using only weak gels in multiple emulsions (W/O/W) was carried out.

M. Benna-Zayani et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 316 (2008) 46–54

Moreover, according to the Taylor theory, the phenomenon of fragmentation depends on the droplets size and the viscosity of the external phase [10]. Thus for a simple emulsion characterised by a diameter which is between 2 and 5 ␮m, a relatively high viscosity can be required to slow creaming without causing fragmentation during manufacturing. In contrast, for multiple emulsions and particularly those intended for pharmaceutical and cosmetic applications oily globules are large (typically in the order of 20–50 ␮m [11–14]) and the viscosity of the external phase should be taken relatively low to prevent fragmentation. So the approaches are different and the employment of polysaccharide solution in double emulsion formulation can have two advantages: their low plastic viscosity at low concentrations which, from a practical point of view, prevents the breakdown of multiple droplets during the double emulsion manufacturing and at the same time the value of their yield stress, sufficient to maintain large droplets suspended in the external phase. Thus, this work aims to prevent creaming in W/O/W double emulsions that can be used in numerous applications by using polysaccharides weak gels in the external aqueous phase. For this purpose, W/O/W double emulsions with relatively large oily globules are studied knowing that if one managed to prevent the creaming of large droplets, it will obviously be able to apply the results obtained to multiple globules of lower size. 2. Theoretical background In order to estimate the minimal value of the yield stress (τ 0 ) of the weak gel (used as the external phase of the double emulsion) required to stabilize droplets and prevent creaming in the emulsion, let us at first consider the case of an isolated sphere suspended in a liquid. At very low velocity, the sphere creates stresses (τ) in its immediate vicinity that, in a first approximation, can be expressed as: τ=

4 Fb = ρgR A 3

(1)

where Fb is the buoyancy force exerted on the sphere, A = πR2 is the frontal area of the sphere, ρ the difference in density between the sphere and the suspending liquid, g the gravity acceleration and R the radius of the sphere. Consequently, to have quasi-static conditions in which the sphere remains motionless (corresponding to a zero velocity), the suspending liquid has to have a yield stress value (τ 0 ) higher or equal to τ: τ0 ≥

4 ρgR 3

(2)

In a more general way, instead of 4/3 coefficient, some authors [15–17] have defined a dimensionless number describing the yield stress effects to gravity effects Y, such as: τ0 ≥ YρgR

(3)

Experimental studies concerning also an isolated sphere suspended in a liquid have shown that Y is ranging between 6 × 10−2

47

and 7 × 10−2 which is relatively different from the simplified 4/3 coefficient. If we calculate the minimal value of τ 0 of a double emulsion like those studied in this work, characterised by a radius R ≈ 50 ␮m, a difference of ρ ≈ 30 kg m−3 , the obtained value, considering a proportional constant Y of 7 × 10−2 and gravity acceleration g = 10 m s−2 , is equal to 0.001 Pa. Moreover, even if we calculate this minimal value of τ 0 for the same double emulsion considering the value of 4/3 for the proportional constant, the obtained value is only equal to 0.02 Pa. However, there are many interacting droplets in an emulsion and the interdroplet interactions must be taken into account for concentrated emulsions. But as the double emulsions studied in this article are diluted (20%, w/w of the simple emulsion), we can assume that the minimal yield stress value required settling droplets will be nearly of τ 0 (relations (2) and (3)). 3. Materials and methods 3.1. Materials The polysaccharides tested are scleroglucan (named Amigel and has an average molecular weight of 4 × 106 Da), carageenan (named Stiagel USC 150 and has an average molecular weight of 6 × 105 Da) and mixture of xanthan (named Satiaxane CX 2 QD and has an average molecular weight of 2 × 106 Da) and locust bean gum (named Viscogum BCR 13 and has an average molecular weight of 5 × 105 Da). These biopolymers are all pure and are kindly given by Degussa texturant systems (Currently called Cargill Texturizing solutions). They have been used as received without any treatment. The oil phase used is coco oil named Miglyol 810 Neutraloel and is provided by Sasol Germany GmbH. The sulphate magnesium (MgSO4 ·7H2 O), sodium chloride and glucose used as electrolytes were purchased from Sigma in pure analytical form. The surfactants used are cethyl dimethicone copolyol with an average chain of 14000 Da (Abil EM90) as the lipophillic one and a copolymer of polyethylene and polypropylene oxides with an average chain of 12000 Da (Synperonic) as the hydrophilic one. Both surfactants were bought from Goldschmidt France. The preserver used for the double emulsions is the sodium methylparahydroxybenzoate. 3.2. Preparation 3.2.1. Polysaccharide solutions Scleroglucan gels have been prepared in demineralised water at room temperature with a vigorous stirring using an Ultraturax (PT 3100 polytron) at a rotation speed of 8000 rpm. Three percentages (w/w) of this polysaccharide have been used: 0.4, 0.6 and 0.8%. The preparation of carageenan gels in demineralised water has been done at 90 ◦ C and under magnetic stirring during 30 min. Different percentages were tested in presence and absence of NaCl. They are summarized in Table 1. Gels prepared from mixtures of xanthan and locust bean have been prepared by dispersing at room temperature the two

48

M. Benna-Zayani et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 316 (2008) 46–54

Table 1 Composition of the tested carrageen and mixtures of xanthan/carob aqueous suspensions Carrageenan aqueous suspensions %(w/w) of carrageenan % (w/w) of NaCl

1 0

1 0.75

1 0.25

1 0.15

1 0.1

1 0.01

0.15 0.75

Global % (w/w)

Ratio of xanthan/locust bean

Xanthan/carob aqueous suspensions 0.02 0.08

50/50, 60/40, 80/20 20/80, 50/50, 80/20, 90/10

biopolymers in demineralised water under magnetic stirring during 10 min then during 30 min with heating the mixture until it reached 90 ◦ C. After that heating was stopped but stirring continued for 15 min. Table 1 summarizes the different proportions and global percentages tested. 3.2.2. Double emulsion preparation All the double emulsions studied in this work have been prepared by the two steps method [18]. At the first step, a simple W/O emulsion is prepared (Table 2) which will serve for all double emulsion formulations. Its preparation is done at 70 ◦ C with a homogeniser type Rayneri Turbo Test at a rotation speed of 2000 rpm during 45 min. In the second step 20% (w/w) of the simple emulsion (W/O) is dispersed in the external aqueous phase which is a suspension of the weak gel in presence of the hydrophilic surfactant and the preserver. One of the external phases is composed of demineralised water instead of the weak gel and has served as reference to study the stability against creaming. The percentage of 20% of the simple emulsion has been chosen because, as it is low, it favours the creaming. This will allow the studying of the effect of polysaccharide in the external phase on inhibiting creaming. Table 2 summarizes the composition of the double emulsions. All the prepared multiple emulsions were observed by an optic microscope in order to ensure itself of the formation of double emulsion. For instance, Fig. 1 shows the microscopic confirmation of the multiple emulsion formation.

0.15 0.5

0.2 0.5

a range of measurement of 0.04–2000 ␮m and equipped with 132 detectors. The primary W/O emulsion was diluted in the oily phase (refractive index equal to 1.44) and taking into account the size of the droplets (<10 ␮m), the Mie’s model was used to characterize the globule size distribution. Each double emulsion was diluted in a 1% (w/w) glucose aqueous solution before any measurement. Three measurements were carried out for each sample and the given value of diameter represents the mean value. The analysis model used to charac-

3.2.3. Droplet size measurements Droplet size distribution is determined by laser diffraction (λ = 750 nm) using a Coulter LS 230 (Coultronics, France) with Table 2 Composition of the prepared multiple emulsions % (w/w) Primary w/o emulsion (PE) Miglyol 810 Abil EM90 H2 O MgSO4 ·7H2 O

27 4 68.3 0.7

Multiple W/O/W emulsion (ME) Primary emulsion Synperonic Sodium methylparahydroxybenzoate Polysaccharide solution

20 0.5 0.05 79.45

Fig. 1. Examples of optical micrographs showing some of the obtained W/O/W double emulsions.

M. Benna-Zayani et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 316 (2008) 46–54

terize the outer globule size distribution was that of Fraunhofer because of the relatively high value of the droplets’ diameter (>20 ␮m) and because it would be very difficult and even impossible to introduce an exact value of the refractive index of the primary emulsion if we had wanted to use the Mie’s model. Distributions were given as a function of the volume and the diameter of the globules. The parameter used in the present study was the mean droplet size d53 which is defined as [19,20]:  ni d 5 d53 = i i3 i n i di where ni is the particle number with a diameter equal to di . This parameter has been chosen because it is a relevant parameter for the polydispersity and because it is highly influenced by the large sizes which govern the possible phenomenon of creaming. However, it is important to note that the very high values obtained for d53 are not very accurate because they may be within the limit of the apparatus. 3.2.4. Density In order to evaluate the minimum value of yield stress which inhibits the creaming, the density values of polysaccharides solutions and those of the simple emulsion, the demineralised water and the miglyol oil have been measured. Unlike the density of the simple emulsion that has been measured by picnometre weighting, all the other densities have been measured with a capillary densimeter Beer analyser 2 from Anton Paar with a precision less than 10−5 (Courtaboeuf, France).

49

The rheological characterisation of the prepared gels has been done mainly by adopting three different tests. For the first test, the existence (or the absence) of a yield stress is checked by measuring shear strain (γ) induced by an increasing shear stress (τ). The shear stress was varied by step steady values and the shear strain was measured 30 s after each applied shear stress step. The graph of Fig. 2a shows a typical ideal curve obtained by a yield stress fluid. Indeed, the strain (γ) can be expressed as: log γ = log τ + log [A(t)] For shear stresses (τ) below the yield value (τ 0 ), the A(t) function can be expressed, regarding the generalized KelvinVoigt model, as: A(t) = J0 +

i=n 

Ji (1 − e−t/θi )

(τ < τc )

i=1

where J0 is the instantaneous elasticity compliance, Ji the delay elastic compliance, θ i the delay time (θ i = ηi Ji ) and n the KelvinVoigt unit number.

3.2.5. Osmolarity A Bioblock Scientific Roebling Micro-osmometer has been used to measure the osmolarity of the external and internal aqueous phases. This osmometer is a cryometer. The calibration was done with demineralised water (0 osmol L−1 ) and an etalon solution of 300 × 10−3 osmol L−1 . The measurement was obtained after direct reading with a precision of 10−3 osmol L−1 . 3.2.6. Visual assessment of creaming Prepared double emulsions were poured into a 100 mL cylindrical glass bottles immediately after preparation. These bottles were then kept at rest and any movement of creaming boundaries has been followed visually with time. It was determined by means of volume fraction of the serum after creaming. For this, the ratio of the height of the serum after creaming by the height of the fresh-prepared emulsion in the bottle was calculated. 3.2.7. Rheological measurements All the rheological measurements have been carried out on a shear-stress controlled rheometer: the Haake Rheostress 600 (Thermo Electron Corporation) equipped with a Haake Universal Temperature Controller. The geometry used is a cone plate (C60) with a diameter of 60 mm and an angle of 1◦ . A solvent trap has been used to minimize the drying out effects (water evaporation) and dust settlement. During the experiments, the temperature has been maintained at 20 ◦ C.

Fig. 2. Ideal curves for yield stress determination for a plastic fluid. (a) Steady shear stress by step and (b) Viscoelastic analysis.

50

M. Benna-Zayani et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 316 (2008) 46–54

For values of shear stress higher than τ 0 , A(t) is expressed as:

A(t) = J0 +

i=n  i=1

Ji (1 − e−t/θi ) +

t η0

(τ > τc )

And thus, theoretically, before and after the yield value (τ 0 ), providing that the time t is independent of τ, we obtain: d log γ =1 d log τ Consequently, beyond and below the yield value no change in the slope of the log γ–log τ curve has to be remarked. But at τ = τ 0 , we expect a change in the rheological behaviour which passes from that of an elastic solid to a viscous liquid. So, the strain undergone by the sample for the shear stresses higher than the threshold is largely higher than that before the flow. Thus, at τ = τ 0 , a sharp increase of the level of the strain appears, which can be indicative of the yield stress value. As the measured and required values of yield stress are relatively low, the detection of the presence of the yield value is very delicate. For this reason we decided to confirm the existence of the yield stress in the studied weak gels. For this purpose, an oscillatory viscoelastic analysis has been carried out at a frequency of 1 Hz. This second test consists in imposing an increasing sinusoidal shear stress and to measure the storage modulus (G’), the loss modulus (G”) and the loss angle (δ). For fluids exhibiting yield stress results expect for τ < τ 0 a linear domain of the storage and loss modula in which the amplitude of the applied shear stress is not enough to affect the structure of the tested material. Viscoelastic quantities are constant and the value of the loss angle δ is very low and indicative of an elastic state (Fig. 2b). At the critical value of stress, the storage modulus decreases drastically and the loss modulus passes through a maximum and the more pronounced the maximum, the higher the value of the yield stress of the tested gel. For viscoelastic fluids without yield stress, G” does not show any maximum. Finally, as a gel structure is characterized by a threedimensional network established by interparticular bonds, when a shear stress higher than the yield value is applied on the gel, the interparticular bonds are broken and the three-dimensional structure breaks down. When the shear stress ceases being applied, the material can return to its initial state of gel by again establishing the broken interparticular bonds. The time necessary to again find the threedimensional starting structure (characterized by a given value of G’) is called rebuilding time. To estimate the rebuilding time of the weak gels, we have proceeded to an oscillatory viscoelastic analysis (at a constant applied shear stress smaller than the yield value and at a constant frequency of 1 Hz) after shearing the sample with an increasing shear rate from 0 to 1000 s−1 . G’ and δ were measured in time and the rebuilding time is obtained when the values of G’ and δ were stabilized (G’ must reach its value in the linear regime).

Fig. 3. Detection of yield stress: 0.08% (w/w) mixture of X/L (90/10).

4. Results and discussion 4.1. Rheological characterisation of polysaccharides in water The results obtained by the tested polysaccharides solutions show that they can exhibit a critic value of shear stress (τ c ) even at low concentrations. The critical value of shear stress (τ c ) indicates a transition between two regimes of strain: the linear regime and the non-linear regime. Depending on the material, τ c can be taken as the yield stress value. And this can be done if the viscoelastic analysis shows a maximum of G” [21,22]. The graph given in Fig. 3 is an example of stress–strain obtained curves by the polysaccharides tested. All the obtained values of τ c are grouped in Table 3. This table also contains an order of magnitude of viscosities measured just after τ c . These values have allowed us to choose, after characterisation, the gels which will serve to double emulsion formulation. Results of the oscillatory viscoelastic analysis show that the loss modulus (G”) passes through a defined maximum for scleroglucan and xanthan/locust bean aqueous solutions, whereas no maximum of the viscous modulus was observed for carageenan solutions. Fig. 4 shows an example of the obtained curves for each of the tested polysaccharides. Results found by varying the steady shear stress by step and by the oscillatory viscoelastic analysis allow, on the one hand, to confirm the presence of yield stress in the aqueous solutions of scleroglucan and mixtures of xanthan and locust bean gum and, on the other hand, to show that carageenan solutions do not present any yield stress even if the strain–shear stress curves show a critical shear stress. The latter could be attributed to the end of the linear regime or to a drastic change in rheological behaviour but not to the presence of a threshold. In addition, the deviation observed for these fluid gels (scleroglucan and mixtures of xanthan and locust bean) with regard to the ideal case of a yield stress fluid, could be explained by

M. Benna-Zayani et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 316 (2008) 46–54

51

Table 3 Values of yield stress, viscosity (measured just after the yield value) and rebuilding time of the tested polysaccharide aqueous solutions Scleroglucan Composition (w/w) τ c (Pa) Viscosity (Pa s) Rebuilding time (s)

0.4% 0.60 1.3 60

0.6% 0.85 6.4 90

0.8% 2.8 13 60

Carageenan %Carrageenan + %NaCl (w/w) τ c (Pa) Viscosity (Pa. s) Rebuilding time (s)

1+0 0 – –

1 + 0.75 29 40 –

1 + 0.25 15 88 × 10−3 –

1 + 0.15 2.4 1.6 × 10−2 60

1 + 0.1 1.3 – 120

1 + 0.01 0 – –

0.15 + 0.75 1.0 58 × 10−3 200

0.15 + 0.5 0.45 – –

0.2 + 0.5 2.6 38 × 10−3 300

Xantan/locust bean (X/L) Global (%, w/w) (ratio of X/L) τ c (Pa) Viscosity (Pa s) Rebuilding time (s)

0.02 (50/50) 0.11 3.5 × 10−3

0.02 0.20 2.2 × 10−3

0.02 (80/20) 0.30 10 × 10−3

Fig. 4. Oscillatory viscoelastic analysis (a) 0.6% (w/w) scleroglucan, (b) 0.08% (w/w) mixture of x/c (80/20) and (c) 1% (w/w) carageenan + 0.15% (w/w) NaCl.

0.08 (20/80) 0.19 38

0.08 (50/50) 0.67 23

0.08 (80/20) 0.70 14 × 10−3 125

0.08 (90/10) 0.87 54 × 10−3 90

Fig. 5. Determination of rebuilding times (a) 0.4% (w/w) scleroglucan, (b) 0.08% (w/w) mixture of x/c (90/10) and (c) 1% (w/w) carageenan + 0.1% (w/w) NaCl.

52

M. Benna-Zayani et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 316 (2008) 46–54 Table 5 The mean droplet size d53 of the studied multiple emulsions Emulsion MEref ME01 ME02

Just after preparation 32 ± 1.2 29 ± 0.6 20 × 101 ± 5.1

ME05

44 ± 1.4

ME06

100 ± 1.5

d53 (␮m)

21 days after: 30 ± 1.8 22 days after: 21 × 101 ± 5 13 days after: 2.5 × 102 ± 3.8 47 days after: 76 ± 1.9

2 years after: 229 ± 35

2 years after: 104 ± 3

Fig. 6. The droplet size distribution of the primary W/O emulsion.

the probable formation of pseudo gels formed by aggregates of particles (micro gels) rather than a gelified structure with a three-dimensional network [23]. Finally, the times necessary to allow the rebuilding of the pseudo gels after destruction have been estimated to few minutes (Table 3 and Fig. 5). This indicates that after shearing, the gel or pseudo gel structures go back rapidly to their initial pseudo three-dimensional networks. 4.2. Characterisation of prepared double emulsions The droplet size distribution of the primary W/O emulsion is shown in Fig. 6 and the average droplet diameter d43 is equal to 4.00 ± 0.3 ␮m. In the following, only findings related to the double droplets will be reported. 4.2.1. External phases composition in double emulsion The polysaccharide solutions used in double emulsion formulation were chosen considering the yield stress and the viscosity values exhibited by the tested gels. For gels prepared with scleroglucan solutions, that at 0.4% (w/w) is the only one which has been tested in double emulsion formulation because the two others have too high viscosities values. For xanthan/locust bean mixtures and carageenan, only two compositions have been kept (Table 4). The emulsions prepared with scleroglucan and xanthan/locust bean mixtures have been formulated under two osmolar conditions: (i) iso-osmolarity between the external and internal aqueous phases by adding glucose in the external phase Table 4 Composition and osmolarity of external phases of the formulated emulsions Emulsion

External phase composition(% w/w)

Osmolarity of the external phase (mosmol L−1 )

MEref ME01 ME02 ME03 ME04 ME05 ME06

100% H2 O + 0.94% glucose 0.4% Scleroglucan + 0.94% glucose 0.4% Scleroglucan 0.15% Carageenan + 0.75% NaCl 0.2% Carageenan + 0.5% NaCl 0.08% Xantan/locust bean (90/10) 0.08% Xantan/locust bean (80/20) + 0.94% glucose

58 58 9 243 159 6 58

(ME01 and ME06). This aims fundamentally to get a constant diameter with time and (ii) hypo-osmolarity with an osmolarity of the external phase lower than that of the internal one in order to have an idea on the kinetic of swelling for an eventual application of drug delivery by swelling breakdown mechanism (ME02 and ME05) [24,25]. With carageenan solutions, it was not possible to prepare emulsion under the previous conditions because of the presence of NaCl (ME03 and ME04). 4.2.2. Stability of emulsions and discussion The stability against creaming of the prepared multiple emulsions has been improved compared with that of the emulsion taken as reference (MEref). Indeed, the latter begins to cream after 30 min of preparation in spite of the relatively low value of oily globules size (Table 5) whereas the incorporation of polysaccharides in the external phase has allowed reaching appreciably higher stability’s times (Tables 6 and 7): Multiple emulsions prepared using carageenan solutions as the external phase are less stable. The creaming happens after a week for the ME04 and after 25 days for the ME03. After 21 months, the volume fraction of the serum was of 0.83 for ME04 and 0.67 for ME03. These results were expected since the rheological characterisation of carageenan solutions revealed the non-existence of a yield stress. Thus, the noted stability’s times higher than that of MEref are due to the delay of creaming caused by the increase of the external phase viscosity. The emulsions prepared with the mixtures of xanthan and locust bean gum are more stable than the previous ones. ME05 had a moderate oily droplet size just after preparation and as expected, swelling was observed after 13 days (Table 5). It creamed very slightly after 1 month and after 21 months, the volume fraction of serum was 0.33. The fresh prepared ME06 showed a relatively high droplets diameter for an emulsion prepared under iso-osmolar condition. But after 47 days, the mean diameter decreases (Table 5). After 2 years, the average diameter practically returns to its initial value. This emulsion creamed very slightly only after 2 months and after 2 years the volume fraction of serum was 0.067. The initial decrease of the diameter and the observed slight creaming could be attributed to a breakdown of some of the largest oily globules. Nevertheless, the assumption of an experimental error allotted to the initial decrease observed for the droplets diameter remains possible.

M. Benna-Zayani et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 316 (2008) 46–54

53

Table 6 Calculated values of minima yield stress required to inhibit the creaming (τ 0min ) and the measured values of yield stress of the weak gels used as external phases (τ 0 ) Emulsion

Measured d53 /2 (␮m)

ρ = ρ − ρPE (kg. m−3 )

Calculated τ 0min (mPa) (Eq. (2))

Measured τ 0 (mPa)

MEref ME01 ME02 ME05 ME06

16 14.5 100 22 50

25.188 26.611 26.611 25.469 25.469

5.4 5.1 (41 after 2 years) 35 (37 after 22days) 7.5 (42 after 13days) 17 (18 after 2 years)

– 600 600 870 1.70 × 103

Table 7 Serum volume fraction (Φv ) Values after creaming Emulsion

Φv (v/v)

MEref ME01 ME02 ME03 ME04 ME05 ME06

After 30 min: 0.10 After 28 months: 0.0 After 28 months: 0.0 After 25 days: 0.017 After 1 week: 0.017 After 1 month: 0.017 After 2 months: 0.017

After 21 months: 0.67 After 21 months: 0.83 After 21 months: 0.33 After 24 months: 0.067

The more stable emulsions against creaming are those prepared with scleroglucan. Indeed, ME01 is the more stable emulsion and does not show any creaming even after two years. It has also the smallest droplets diameter (Table 5). But after two years, the oily globules swell in spite of the iso-osmolar condition of preparation. ME02 which is prepared with the same concentration of scleroglucan as ME01 but with a hypo-osmolarity does not show any creaming even after 20 months despite the very high value of droplets diameter (Table 5). But this emulsion has a particular texture compared to the other emulsions. It has the aspect of foam. ME01 and ME06 present two iso-osmolar formulations which normally should not exhibit any swelling with time. But results show that ME01 swell after 2 years. This result is possibly due to a coalescence of oily globules in ME01. The stability of emulsions prepared with scleroglucan and with mixtures of xanthan/locust bean solutions could be related to the three-dimensional network structure exhibited by polysaccharides in these weak gels. Indeed, if we refer to the assumption of a structure composed by micro-gels, as already mentioned in the rheological section, the gelation of polysaccharide could lead to jointed micro-gels. This would depend primarily on the nature of polysaccharide as well as arrangement of its chains in solution. Moreover, in this case, the stability of the emulsion would also depend on the size of oily globules compared to that of the meshes of micro-gels. Taking into account the difference in stability observed between the emulsions prepared with the scleroglucan and those with xanthan/locust bean mixtures, one could think that this assumption is rather valid for the xanthane/caroub mixtures than scleroglucan. Thus, according to the most probable assumption one could attribute a structure of well-developed three-dimensional gel through the whole of the sample for scleroglucan solutions and a structure of connected micro-gels for the xanthane/caroub mixtures.

Finally, it is interesting to point out the very low values of times necessary to allow the rebuilding of the pseudo gels after destruction compared to the stability’s time of the emulsion free of polysaccharide and taken as a reference. And this means that if an emulsion is shaken momentarily inducing the destruction of the structure, a rapid rebuilding will happen without observing any creaming. 5. Conclusion The main goal of the paper was to stabilize multiple emulsions (W/O/W) against creaming by using polysaccharides aqueous solutions as the external phase during preparation. With the exception of carageenan aqueous solutions which do not exhibit a yield stress, the other tested weak gels seem to be appropriate to prevent creaming in multiple emulsions. Scleroglucan and mixtures of xanthan/locust bean solutions exhibit relatively low values of yield stress and plastic viscosity. They go back rapidly to their initial structure after shearing. These rheological characteristics allow the easy preparation of multiple emulsions and the suspending of oily droplets in the external phase by preventing creaming. Acknowledgement The authors would like to thank Mr. Desprairies and Mrs. Le Franc¸ois from Degussa (Cargill) company for the interest they showed for this work. The authors are also grateful and thank Mr. Azib Abdelwahed for his invaluable assistance for the correction of the English of the manuscript. References [1] M. Grassi, R. Lapasin, S. Pricl, A study of the rheological behaviour weak gel systems of scleroglucan, Carbohydr. Polym. 29 (1996) 169–181. [2] A. Paraskevopoulou, D. Boskou, V. Kiosseoglou, Stabilization of olive oil – lemon juice emulsion with polysaccharides, Food Chem. 90 (2005) 627–634. [3] A.B. Rodd, C.R. Davis, D.E. Dunstan, B.A. Forrest, D.V. Boger, Rheological characterisation of ‘weak gel’ carrageenan stabilised milks, Food Hydrocolloids 14 (2000) 445–454. [4] C. Michon, G. Cuvelier, E. Aubr´ee, B. Launay, Caract´erisation rh´eologique de gels fluides aux petites d´eformations et a` la rupture, Les Cahiers de Rh´eologie 16 (1999) 46–53. [5] C. Michon, C. Chapuis, V. Langendorff, P. Boulenguer, G. Cuvelier, Strainhardening properties of physical weak gels of biopolymers, Food Hydrocolloids 18 (2004) 999–1005. [6] M. Renaud, M.N. Belgacem, M. Rinaudo, Rheological behaviour of polysaccharide aqueous solutions, Polymer 46 (2005) 12348–12358.

54

M. Benna-Zayani et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 316 (2008) 46–54

[7] C. Valenta, K. Schultz, Influence of carrageenan on the rheology and skin permeation of microemulsion formulations, J. Control. Release 95 (2004) 257–265. [8] D. Bais, A. Trevisan, R. Lapasin, P. Partal, C. Gallegos, Rheological characterization of polysaccharide–surfactant matrices for cosmetic O/W emulsions, J. Colloid Interf. Sci. 290 (2005) 546–556. [9] M. Akhtar, B.S. Murray, E. Dickinson, Perception of creaminess of model oil-in-water dairy emulsions: Influence of the shear-thinning nature of viscosity-controlling hydrocolloid, Food hydrocolloids 20 (2006) 839– 847. [10] V. Muguet, M. Seiller, G. Barratt, D. Clausse, J.P. Marty, J.L. Grossiord, W/O/W multiple emulsions submitted to a linear shear flow: Correlation between fragmentation and release, J. Colloids Interf. Sci. 218 (1999) 335–337. [11] N. Garti, Double emulsions – scope, limitations and new achievements, Colloids Surf. A 123/124 (1997) 233–246. [12] H. Okochi, M. Nakano, Preparation and evaluation of W/O/W type emulsions containing vancomycin, Adv. Drug Deliv. Rev. 45 (2000) 5–26. [13] G.T. Vladisavljevic, H. Schubert, Influence of process parameters on droplet size distribution in SPG membrane emulsification and stability of prepared emulsion droplets, J. Membr. Sci. 225 (2003) 15–23. [14] M.J. Devani, M. Ashford, D.Q.M. Craig, The development and characterisation of triglyceride-based ‘spontaneous’ multiple emulsions, Int. J. Pharm. 300 (2005) 76–88. [15] R.P. Chhabra, Bubbles Drops and Particles in Non-Newtonian Fluids, CRC Press, Boca Raton, FL, 1993.

[16] L. Jossic, A. Magnin, Drag stability of objects in a yield stress fluid, AIChE J. 42–12 (2001) 2666–2672. [17] L. Jossic, A. Magnin, Structuring of gelled suspensions flowing through a sudden three-dimensional expansion, J. Non-Newtonian Fluid Mech. 127 (2005) 201–212. [18] J.L. Grossiord, M. Seiller, Multiple emulsions: Structure, properties and applications, Edition de Sant´e, Paris, 1998. [19] M.J. Hey, F. Al-Sagheer, Interphase transfer rates in emulsions studied by NMR spectroscopy, Langmuir 10 (1994) 1370–1376. [20] P. Walstra, Physical Chemistry of Foods, Marcel Dekker, New York, 1998. [21] P. Coussot, J.L. Grossiord, Comprendre la rh´eologie. De la circulation du sang a` la prise du b´eton, EDF Sciences, GFR, Casablanca, 2001. [22] S. Laribi, J.M. Fleureau, J.L. Grossiord, N. Kbir-Ariguib, Comparative yield stress determination for pure and interstratified smectite clays, Rheol. Acta 44 (2005) 262–269. [23] R.J. Ketz, R.K. Prud’homme, W.W. Graessley, Rheology of concentrated micro-gel solutions, Rheol. Acta 27 (1988) 531–539. [24] S. Geiger, S. Tokgoz, A. Fructus, N. Jager-Lezer, M. Seiller, C. Lacombe, J.L. Grossiord, Kinetics of swelling–breakdown of a W/O/W multiple emulsion: possible mechanisms for the lipophilic surfactant effect, J. Control. Release 52 (1998) 99–107. [25] J.L. Grossiord, M. Seiller, W/O/W multiple emulsions: a review of the release mechanisms by break-up of the oily membrane, S.T.P. Pharma Sci. 11-5 (2001) 331–339.