You are on page 1of 17

Evaporation of an emulsion trapped in a yield

stress uid
G. Guéna∗ J. Cordes S. Fouilloux J.-B. d'Espinose
F. Lequeux & L. Talini.
(PPMD-ESPCI ; Paris ; France.)

The present work deals with emulsions of volatile alkanes in an
aqueous clay suspension, Laponite, which forms a yield stress uid.
For large enough yield stress (i.e. Laponite concentration), the oil
droplets are prevented from creaming and the emulsions are thus me-
chanically stabilized. We have studied the kinetics of evaporation of
the oil phase of those emulsions in contact with the atmosphere. We
show that the evaporation process is characterized by the formation
of a sharp front separating the emulsion from a droplet-free Laponite
phase, and that the displacement of the front vs. time follows a dif-
fusion law. Experimental data are confronted to a diusion controlled
model, in the case where the limiting step is the diusion of the dis-
solved oil through the aqueous phase. The nature of the alkane, as
well as its volume fraction in the emulsion, has been varied. Quanti-
tative agreement with the model is achieved without any adjustable
parameter and we describe the mechanism leading to the formation of
a front.

1 Introduction
Emulsions provide an ecient way to transport and/or store small quan-
tities of a chemical compound in a large volume of an immiscible liquid. Due
to a high surface to volume ratio, and thus a high energetical cost associated

e-mail :


5 % w/w . we address the specic question of the evaporation of the oil contained in a o/w emulsion whose continuous phase is a colloidal clay suspension that forms an aging yield stress uid. kinetic stabilization of emulsions is crucial and ecient stabilization processes are sought. it has been emphasized that the volatility of either the emulsifying or the emulsied phase can govern the long time evolution of an emulsion [2. these media are not a stable state of matter.g. Here. Emulsied phase : The emulsied phase is a light linear alkane whose vapor pressure is comparable but smaller than the atmospheric pressure : 1It is unclear that the presence of a third compound may aect or not the relaxation process e. demixing of the two phases occurs through a hydrodynamical process of creaming and coalescence of the droplets which leads to the development of a surnaging phase. creaming and diusion controlled re- laxation. there is no need for the addition of any surfactant in 1 the system . recover its liquid state by application of a strong shear [6] . Over longer time scales. The dispersion is stirred for 72 hours to ensure full dispersion and exfoliation of the clay platelets. 5] . [1] .with their formation. It is further let at rest for 24 hours. diusion controlled destabilization such as Ostwald ripening can be observed [1] . the eects due to evaporation are not hindered or modied by hydrody- namic ones. The advantage of such emulsions is twofold. Second. There- fore. In this paper. 2 .e. we show that the kinetics of evaporation is controlled in an unconventional manner by the concentration prole of alkane. Recently. 3] . 2 Experimental set-up Emulsifying phase : The emulsifying phase is an aqueous suspension of Laponite RD (Rockwood) of concentration φm = 2. the time scale of the yield stress increase is short compared to the destabilization processes of the emulsion : coalescence. we observe (and explain why we can observe) the formation of a front separating the emulsion from a droplet-free aqueous phase. At short times (typically less than 24 hours). As a consequence. The pH is set to 10 by adding hydroxide sodium in order to avoid chemical dissolution of the clay particles [4. First : as the droplets are prevented from creaming during the whole duration of the experiment. The uid aging induces a growing yield stress with the time elapsed after emulsication. It can then be rejuvenated. i. the uid then evolves from a low viscosity liquid to a paste with a large apparent yield stress within a few hours. The originality of our experiment is that during the evapora- tion process.

The volume fraction of the emulsion is varied between 1 % and 10 % . heptane ∗ The solubility of those alkanes x in water is very low [7. Error on the volume fraction is more important for the more volatile compound but it is less than 10 % for pentane whereas it is less than 1% for heptane. x∗ ≡ solubility in water from [7] . the typical size of the emulsied droplets is ao ' 20 µm and varianceσa ' 10 µm . any adsorption of the alkane on the surface of the platelets would lead to a large concentration in alkane even after the evaporation of the oil from the emulsion.05 ± 0. we did not measure any signicant change in the solubility of alkane in a laponite dispersion compared to the one in water. may increase the apparent solubility of the oil [10] . 8] .4 10−7 99 _ Table 1: Physical constants of light linear hydrocarbones.0 10−5 35 1. In the following. 10 −7 mol/g of water ). Emulsication process : Emulsions are prepared using a ultrasonic emul- sier without adding any surfactant and they are stable for several weeks. we assume that the solubility of alkane in a Laponite dispersion is the same as in water. During the emulsication process. values for the solubility found in the literature are given in table 1 . However.41 C7 5. three alkanes have been used for this study : .06 ± 0. Due to the large surface to volume ratio of the clay platelets (it is of the order of ' 10 m /m for a mass fraction 8 2 3 φm ' 1 %). Tb ≡ Boiling Temperature . Specic interaction : A specic interaction of the alkane molecules with the clay platelets. 3 . one would nd cads ' 10 −3 mol/g of dispersion which is signicantly larger than the solubility of the alkane in the aqueous phase (from table 1. assuming close packing adsorption of alkane. using Chemical Oxygen Demand test.1 10−6 69 1. It is determined a priori by weighting the dierent compound and it has been checked a posteriori with optical technics by measuring the amount of alkane contained in the droplets in a given volume of the emulsion squeezed between two slabs. Dm ≡ Mass Diusion coecient in water from [9] .12 C6 2. Emulsions are polydisperse. leading for instance to the adsorption of the alkane on the surface of the platelets. x∗ Tb Dm ◦ −9 mol/mol C 10 m2 /s C5 1. pentane hexane and . For instance. the laponite dispersion is rejuvenated by the strong shear applied but becomes pasty again when at rest. the solubility of pentane ∗ is c = 6.

18.001 % . 4 Evaporation process Evaporation of the emulsions takes place in open air. A question that naturally arises concerning the stability of the emulsions is whether the Laponite platelets play a role in it. the oil droplets remain well dispersed within the aqueous phase over large time scales (a few months) . 19] . Some emulsions  called Pickering emulsions  are known to be stabilized by the trapping of nanopar- ticules at the oil/water interfaces. 12] . Moreover. Optical observation of the 4 .1 P a . which means that the atmosphere in contact with the emulsion is never saturated in alkane (we will see a justication of that latter). 17. which is far smaller than the critical mass fraction we have found. if Pickering stabilization was signicant in our case. Pickering adsorption has been invoked to explain the stability of some emulsions in Laponite dispersions [16. it should prevent the droplets from coalescence for much smaller con- centration in Laponite.3 Stabilization mechanisms Qualitative observations : The stability of the emulsions has been tested by varying the mass fraction of Laponite in the emulsifying phase while the volume fraction of the oil phase has been kept constant atφv = 1 % . The stress values and time scales at stake are consistent with values found in the literature [14. We ob- serve that for concentration in Laponite smaller than φm ' 2 % w/w . Note that in any case the Laponite platelets are in large excess. Indeed. this takes place by a diusive process through an hydrodynamical boundary layer which is less than 1 cm and then by advection due to air draft. when the Laponite is concentrated enough. The yield stress needed to prevent a 10 µm droplet from creaming is Fb /S = 34 ∆ρ g R ' 0. the concentration needed for the total coverage of the oil/water interfaces being φm ∼ 0. we clearly observe that the emulsions are stable only in presence of a yield stress. 15] . The critical concentration in Laponite corresponds to the one for which Laponite disper- sions exhibit a yield stress that is large enough to counterbalance buoyancy stresses [5. Oil drop- lets therefore remain well dispersed in the Laponite provided the yield stress reaches such a value within time scales short compared to the characteristic time of creaming. the oil droplets cream and further separate into an oil phase . However. 11. Emulsions are mechanically stable when the yield stress σY is comparable to this stress Fb /S [13] . on the contrary. It is well known [Ref Béatrice] that the transport of the volatile alkane in the atmosphere.

Thermodynamics : The alkane contained in the droplets of the emulsion is pressurized by the eect of capillarity and the equilibrium concentration in alkane at the interface of the droplets is shifted from equilibrium values in the bulk (this mechanism is known to be responsible for Ostwald ripening in emulsions [1]). Since we focus on the evaporation of the alkane phase only. that has the same properties regarding the transport of alkane as an interface emul- sion/air. The striking feature that appears on these pictures is that a sharp front forms in the vicinity of the interfaces with the atmosphere. 1.(a). the variation with time of the distance between the front and the interface. Water evaporation would indeed modify the evaporation process of the emulsion as emphasized in previous works [2. The front sep- arates a droplet-free phase. Typical micrographs are shown in gures 1. from the emulsion and moves away from the interface as time elapses. it is favorable for the oil phase to empty out into the atmosphere. evaporation of the alkane is the same in the vicinity of an interface emulsion/silicon oil as of an interface emulsion/air . We have observed a similar behavior whatever the nature of the alkane. For the observation purpose. On the other hand. and as we have experimentally checked.(c) . L(t) . which is in contact with the atmosphere. a small volume of silicon oil is inserted between the emul- sion and the atmosphere. Quanti- tative measurements of the front displacement have been performed in the 1D geometry of a capillary tube. 5 . Therefore. as shown in the following. The atmosphere with which the emulsion is in contact is not saturated in alkane either at the beginning of. The silicon oil volume therefore acts as a cork that prevents evaporation for water but not for the alkane. As in the example shown in gure 2.evaporation of the emulsions has been performed using microscopic technics. The evaporation kinetics of the oil phase therefore presents a diusive behavior in each case. in the following. the emulsion is quenched between two glass slabs 100 µm apart. follows a power law with an exponent 1/2 . the chemical potential of the alkane being larger within the droplets than in the atmosphere. 3] .(b) & 1. the interface therefore refers to the emulsion/silicon oil. Silicon oil is miscible with alkane in all proportion whereas it is not miscible with water. nor during the evaporation process. An extra overpressure may also result from the stress exerted by the Laponite on the droplet but this eect will not be considered here.

It means that as long as alkane droplets exist. the molar D ∗ volume of the alkane. However. we nd ac ' 4 nm corresponding to a dierence of the order of 0. the average distance between the droplets is d ≡ ao / 3 φv . Equating the chemical potential of the alkane in the droplets to the chem- ical potential of the alkane dissolved in the aqueous phase. the concentration in alkane in their vicinity is x ' x . Two boundary conditions have to be added to equation (3). Let us focus on equation (1) . In the following. we always have xV < x∗ and in practice. As this time is short compared to the time scale of the front displacement (of the order of τ ∼ 10 s). First. we can always assume x  V x∗ . the interfacial tension with water is γ12 = 48.5 Diusion controlled evaporation model We denote as x the molar fraction in alkane dissolved in the aqueous phase. using values given in the litterature . Second. 100 µm and therefore. x = x given by equation (2).44 mN/m for pentane and γ12 = 49. as we explained later. D √ In the emulsion. Let us now consider the equation of transport for the alkane within the laponite dispersion. one nds the D concentration at the corresponding interface x [8] : xD ac = exp (1) x∗ a Cn Cn wherea is the radius of the droplet and ac = 2γvm /RT with vm . at the interface between the droplets and the laponite. the local concentration in alkane 3 2 From [20]. the concentration in alkane D V is x = x given by equation (1). The diusion of the alkane in the aqueous phase simply writes : ∂x = Dm ∆x (3) ∂t where Dm is the diusion coecient of the alkane in the laponite dispersion. the dierence in 2 those values is small since. Necessarily : x > x . we will assume that the transport is controlled by the dierence D between the concentration in the emulsion at the droplets interface x and V at the vapor interface x . The concentration in alkane at the interface with the atmosphere is given by Henry's law : xV Pv = (2) x∗ Psat Here.46 ± 0. at the interface. the characteristic time associated with diusion is τ ∼ 10 s (assuming a typical value of the mass diusion coecient Dm ' 10−9 m2 /s).41 ± 0.1% .50 mN/m for hexane @20◦ C 6 .

our model predicts that the front displacement follows a diusive power law √ L= 2Dt with F x∗ D= Dm . the concentration in alkane follows a one-dimensionnal V diusion equation similar to equation : ∂t x = Dm ∂zz x with x = x and D x = x at z = 0 and z = L . It is crucial to note that the front between the emulsion and the droplet- free Laponite is very sharp and does not broaden with time. The mass ux towards the interface then reads ( see gure 3) : δx jx = Dm (5) L δx = xD i − xi . xi ' 0 . V Integrating the rst order equation resulting from equations (4) and (5). As pointed out in what precedes. the droplets being V with rather large. equations (1) and (2) give δx ' x∗ . In the absence of droplet. owing to surface tension eects. we simply have x = x .is the stationary one given by thermodynamics following equation (1). the size of the droplets decreases. In that case. the displacement of the front is related to the mass ux through the conservation of mass : φv d L jx = (4) F dt  where F ≡ vm Cn H2 O /vm is a numerical prefactor that ensures conversion from volume fraction to molar fraction. As schematized in gure 3. Therefore. Thus. there are thus two dierent length scales in the prob- lem. Thus. The concentration in droplets as a function of the distance to the interface can then be reasonnably described by a step function. Its stability results from the fact that. the eective diusion coecient D is proportional to the ∗ ∗ solubility of the oil x . Note that because F x /φv  1 . Within our model. the smaller is a droplet. As the atmosphere is under saturated in alkane. z is the distance to the interface with the atmosphere. x D ' x∗ . The problem can therefore be described as follow: in the region where D the concentration in droplets is large enough. when the front moves. the eective diusion coecient D is very small compared to Dm . which leads to an increase of the capillary pressure inside them and accelerates the leaking of the alkane away from the droplets. (6) φv As expected. the faster the alkane it contains dissolves into the aqueous phase. Here. the assumption of a 7 . the concentration prole can be assumed to be quasi- stationary with xed values of the concentration at the front and at the interface.

leads to the formation of a shock wave that is analogous to the one observed for instance in sedimentation of suspensions of monodisperses non-Brownian solid particles. Second. The kinetics is therefore controlled by the diusion of the dissolved oil through the aqueous phase. Dm . The alkane mass ux through the emulsifying phase is jlap = MCn /vmH2 O · (Dm δx) /L . we can neglect the alkane transport kinetics in the gas phase (explicite calculation is given in appendix). the eective diusion coecient of the front ∗ D should only depend on the solubility of the emulsied oil x and on the volume fraction of the emulsion φv . this 5 means that as soon as fresh atmosphere is convected with boundary layers smaller than 1m. the increase in the front velocity with the oil volume fraction. the droplets play the role of a reservoir that feeds the ux of oil towards the interface. In gure 4(a). in the gaz phase the concentration step is : δx = Psat /Po ∼ 0. In the latter case. we see that the same data collapse onto on a single curve when L is plotted as a function of 2Dt with D given by equation (6). Furthermore. the mass diusion coecient of alkanes is much larger in the gas phase [21] : for pentane G . the slower the front moves away from the interface with the atmosphere. 6 Experimental results If we refer to equation (6).  As- suming diusion controlled transport yields : jatm = MCn /vm · Dm δx G G V /Λ where Λ is the typical length over which the concentration gradient estab- lishes in the atmosphere. the settling velocity increases for decreasing particle concentration. the larger this reservoir and therefore. it can however be simply understood : during the evaporation process. the mass ux is far larger in the atmosphere as long as Λ  10 L . which is a poorly known quantity. Let us compare that ux with the one in the atmosphere. which results in a sharp front separating the suspension from the particle-free uid that is very similar to the one we observe during the evaporation of emulsions. In other words. D in particular depends on the mass diusion coecient of the alkane. On gure 4(b).quasi-stationary concentration prole is consistent with our description.5 mol/mol . Dm = 3. The larger the volume fraction φv . this nding is similar to the evaporation mechanism proposed in creamed emulsions [3] . Finally. In practice. We have used 8 . The D ∝ φv−1 dependence is not intuitive. Therefore. we can now consider the assumption made that x V ' 0 . calculation V 5 shows that the mass uxes are equal if Λ = 10 L .7 10 −6 m2 /s . we report measurements when those two parameters are varied. First. in the case where the interface would be at saturation.

(b) : present study 7 Conclusion We present herein a work dealing with emulsions that are mechanically stabilized. The collapse of the data onto a master curve in the representation of gure 4(b) is remarkable given the large spread of the initial data cf gure 4(a). 22] for diusion of alkanes in water. limited by the diusion of the dissolved alkane through the emulsifying phase.05 ± 0.the value of the mass diusion coecient of light alkanes in water currently −9 found in the literature [9] : Dm ' 10 m2 /s . that can thus be extrapolated. Ta- ble 2 compares the values of the mass coecient Dm of alkanes in a laponite dispersion to data available in the literature [9. √ The somewhat larger values found experimentally as compared to 2Dt . however Dm should depend on the nature of the alkane. It validates in particular our description of the evaporation process. The D ∝ φv dependency is well observed. the emulsifying phase presenting a yield stress that prevents the oil droplets from creaming. (a) (b) Dm Dm C5 1. Owing to its weak concentration (particle volume fraction close to 1%). especially in the cases of pentane and hexane. The dierences for the three alkanes can only be attributed to dierent values of Dm . This point clearly appears on table (2) where the eective diusion co- ecient has been normalized by a quantity taking into account the alkane −1 solubility and molar volume and is shown as a function of φv . However the large dier- ences observed are intriguing and more experimental data in both water and Laponite is needed to draw conclusions.06 ± 0.3 ± 0.5 C7 _ 1. The displacement of the front formed during the 9 . the presence of Laponite in the emulsifying phase is not likely to induce a signicant variation of the mass diusion coecient as compared to the one in water . doubtlessly result from the approximate value of the mass diusion coecient we have used.25 Table 2: Mass diusion coecient Dm (10−9 m2 /s) (a) : from [9] . the evaluation of the mass diusion coecient is not direct as in a NMR measurement. We have studied the evaporation of the oil phase from such an emulsion.0 ± 0.12 6±1 C6 1. in our experiment.41 2. Indeed.

Rheol. J. 20:2069. Ostwald ripening in emulsions (1) : Direct observations of ostwald ripening in emulsions. Clint. Liquides. 2002. Col. Mongondry. H. 118:590. and E. B. [8] B. J. 2003. J. 10 . Chem. Pertzov. The variation of the eective diusion coecient of the oil with the oil volume fraction in the emulsion is in particular well captured by the model we develop. S. Int. K. Tanase. [2] I. I. 2005. 2004. 1987. B. Bonn. 104:5892. Revised state diagram of laponite dispersions. We show that the experimental data is well described assuming that the oil transport is controlled by the diusion of the dissolved oil through the aqueous phase. Tassin. A. Langmuir. Mc Aulie. J. Phys. [4] S. Tanaka. which have been poorly studied in the past. Aranberri. Dynamical mechanical proper- ties of gelling colloidal disks. S. and P. Beverley. 163:478. J. Binks. Fletcher. P. Fletcher. B. Shchukin. Sciences. How do emulsions evaporates? Langmuir. Sci.. D. References [1] A. Price and O. 1969. [7] C. Let.evaporation process has been measured as a function of time for a large set of parameters. Kabalnov. 44:585. Clint. [3] I. and T. Int. We also propose a mechanism explaining the formation of a sharp front between the emulsion and the droplet-free aqueous phase.. Tassin. Cocard. V. Belin. and T. A. Aranberri.. J. J. [9] W. Col. Solubility in water of normal c9 and c10 alkane hydrocar- bons. Phys. and J. 2002. 2000. 89:015701. These results shed new light on the evaporation process of emulsions. [6] D.-F. Evapora- tion rate of water from concentrated oil-in-water emulsions. J. 2000. Sci. Abou. Söderman. Jacques Meunier. and P. Nicolai. Cabane and S. [5] P. Self-diusion coecients of some hydrocar- bons in water : Measurements and scaling relations. Nicolai. Laponite: Aging and shear rejuvenation of a colloidal glass.. 18:3471. 283:397. Rev. Binks. H. D. Hénon.-F.

L. Wilhelm. and S. Selective adsorption of hydrocarbone gases on clays and organic matter. P. 2007. Flow induced by a sphere settling in an aging yield-stress uid.L. Peysson. 1999. 277:257. Ashby and B. Lyonnard. 49:672. Chowdhry. Binks. [14] C. and J. A. M. Euro- phys. Torres. Chem. Mourchid. 306:285. Yang. 1984. Thieme. J. Huang. Eect of dispersion ph on the formation and stability of pickering emulsions stabilised by layered double hydrox- ides particules. A. The mathematical theory of non uni- form gases. Levitz. Frigaard. Local rheological probes for complex uids: Application to laponite suspen- sions. 2007.. Q. Leharne. and P. 302:439. [21] S.. N. Chapman and T. [16] J. A. Herzhaft. Chem. Niu. Mourchid. 66:021502. B. [20] Y. A. [17] N. Colloid Polym. Propagation and stopping of air bubbles in carbopol solutions. Organic Geochemistry. [19] F. [18] L. Cheng and W. A. Sci. structure. Surf. 2000. 29:407. 2006. Delville. and D. J. Phys. Dubash and I. Lecolier. and C. Liquid-solid transition of laponite suspensions at very low ionic strength: Long-range electrostatic stabilisation of anisotropic colloids. Y. [15] B. Elias. Klyomlyat. Col. and rheology of laponite suspensions. 35:413. Browaeys. Int. S. J. Abend. Sun. Rev. E.Lett. Snowden. Phase diagram of colloidal dispersions of anisotropic charged particles: Equilib- rium properties. 2002. Chem.-C. 2007.[10] A. Lecolier. 2000. Phys. 1995. Bacri. J. 142:123. and M. Phys. 2004. G. E. Morl. Iturbe. Phys. E. 1970. Prepa- ration of o/w emulsions stabilised by solid particles and their character- isation by oscillatory rheology. Aggregation in pickering emulsions. Levitz. Delville. Pickering emulsions stabilised by laponite clay particles. J. Sci. B. [11] A. Third edition. P. [13] N. Surface and interfacial tensions and their combined properties in seven binary.. Data. [12] P. and S. Eng. Tsul. Allain. Fluids. 11:1942. Lagaly. Talini. Gueslin. 11 . Non-Newtonian Fluid Mech. Cowling. Ponton. Lang- muir. A. 2:5640. and G. Cambridge University Press. F. Col. L.. immiscible liquid-liquid- vapor systems. Lambard. 18:103101. R..

Stilbs. Phys. Molecular dynamics and nmr study of methane-water systems. 1991.. Mol. 12 . Laaksonen and P. 74:747.[22] A.

Figure 1: (a) (b) & (c) : Photographs taken in 45 minutes interval for an emulsion of pentane at room temperature. The emulsion appears on the upper part of the photograph. A sharp front separating the emulsion from the droplet-free Laponite dispersion develops from the interface with the atmosphere. 13 .

06 1 10 100 1000 t (h) Figure 2: Distance between the front and the interface. 6 1 L (mm) 1/2 0. The experimental data (squares) is well described by a diusive law −12 (full line) with an eective diusion coecient D = 3. as a function of time. L. Emulsion of hexane @ φv = 3% . 10 m2 /s .1 0. 14 . The experiment lasts one mounth.

Figure 3: Spatial variations of the droplet volume fraction and of the mo- lar fraction of the dissolved alkane as considered in the description of the transport of alkane to the atmosphere. 15 .

04 0.1 0.01 0. The dashed line represents 2Dt . (b) Rescaled lengths by assuming Dm = 1.1 1 10 2 2Dt mm ( ) 0.1 0.1 1 10 2 2Dt (mm ) (b) Rescaled lengths Figure 4: (a) Distance of the front from the interface as a function of time.04 0. 4 pentane hexane heptane 1 L (mm) 0.04 0.001 0.001 0. 10 √ −9 m2 /s . 16 .1 0.1 1 10 100 1 000 10 000 3 t (10 s) (a) Raw measurements 4 pentane hexane heptane 1 L (mm) 4 1 L (mm) 0.01 0. The time variations of L are shown for three dierent alkanes (dierent colors) and when varying the volume fraction in oil of the emulsions φv (dierent symbols).

As the eective diusion coecient spreads over several orders of magnitude depending on the alkane. Straight lines stand for Cn H O D = Dm /φv .Figure 5: Eect of the volume fraction of the emulsion φv . it has been rescaled by the solubility and by the molar volume ∗ of each alkane D → D/F x where F = vm /vm2 . 17 .