Ultrasonics Sonochemistry 6 (1999) 7583

Emulsication by ultrasound: drop size distribution and stability

B. Abisma l, J.P. Canselier*, A.M. Wilhelm, H. Delmas, C. Gourdon
Laboratoire de Ge nie Chimique (UMR CNRS 5503 INPT/UPS), Ecole Nationale Supe rieure dInge nieurs de Ge nie Chimique, 18 chemin de la loge, F31078 Toulouse, France

Abstract The aim of this work is to compare the oil-in-water emulsions produced by mechanical agitation ( Ultra-Turrax, 10 000 rpm, P=170 W ) or power ultrasound (ultrasound horn, 20 kHz, 130 W ) using the same model system : water/ kerosene/polyethoxylated (20 EO) sorbitan monostearate. The following parameters were varied : emulsication time, surfactant concentration, consumed power and volume fraction of oil. With ultrasound, the drop size (Sauter diameter, d ) is much smaller than that given by 32 mechanical agitation under the same conditions, which makes insonated emulsions more stable. For a given drop size (d ), less 32 surfactant is required. 1999 Elsevier Science B.V. All rights reserved.
Keywords: Emulsication; Emulsion stabililty; Power ultrasound

1. Introduction According to P. Becher, an emulsion is a heterogeneous system, consisting of at least one immiscible liquid intimately dispersed in another in the form of droplets, whose diameters, in general, exceed 0.1 mm. Such systems possess a minimal stability, which may be accentuated by such additives as surface-active agents, nely divided solids, etc. [1]. In this respect, surfactants may be present as monolayers or liquid crystals; polymers are another class of emulsiers. Those thermodynamically unstable (i.e., only kinetically stable) liquid liquid dispersions possess a high interfacial area. The type of simple emulsion (water-in-oil or oil-in-water, commonly abbreviated as w/o or o/w) is decided mainly by the volume ratio of the two liquids, their order of addition and the nature of the emulsier [2], characterized, for instance, along the HLB scale [3]. Emulsions are common in a large range of technologies [4] and their preparation should be carefully but easily controlled when scaled up to commercial production. Except in special cases where spontaneous emulsication can occur [1,5], energy must be supplied to produce such metastable mixtures. Thermodynamically speaking,
* Corresponding author. Tel.: +5-62-25-23-31; fax: +5-61-25-23-18.

surfactants reduce the surface free energy required to increase any interfacial area (DG=cDA ), by lowering the interfacial tension, c, and allow nely dispersed media to be created easily. In fact, a much higher amount of energy, with respect to this thermodynamic part, is necessary, since comminution of large droplets into smaller ones involves additional shear forces, so that the viscous resistance during agitation absorbs most of the energy [2,6 ]. The excess energy is dissipated as heat. Energy may be provided through various means, namely mechanical agitation (stirrer, colloid mill, mixer, valve homogenizer, ) or ultrasound generation. Highpower ultrasound must be used, low-intensity acoustic waves leaving the medium unchanged. High-frequency vibrations applied to a diphasic liquid system provide a dierent means of breaking and dispersing a bulk phase: large drops (ca. 80 mm), produced by the instability of interfacial waves, are broken into smaller ones by acoustic cavitation [7,8]. Ultrasound emulsication was reported for the rst time by Wood and Loomis [9].The rst patent was granted in 1944 in Switzerland [10]. Since then, many scientists and industrialists have used dierent types of ultrasound devices (whistle, horn) to make emulsions [1,11]. Beal and Skauen have studied the inuence of exposure time and sample geometry on the quality of emulsion [12]. Higgins et al. have observed that, in general, for short exposure times, higher power

1350-4177/99/$ see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S1 3 5 0 -4 1 7 7 ( 9 8 ) 0 0 02 7 - 3

76

B. Abisma l et al. / Ultrasonics Sonochemistry 6 (1999) 7583

Fig. 1. Destabilization phenomena in emulsions.

setting produces higher specic surface area, but an optimum amount of energy seems to exist in order to produce a maximum interfacial area [13]. Above this optimum, degradation of the surfactant may occur. In fact, it has been observed by Alegria et al. [14] that the surfactant accumulates at the interface of cavitation bubbles, where it can be chemically degraded by H and OH radicals produced by the thermal decomposition of water [15,16 ]. The ultrasound device has been found to produce o/w emulsion with very small particle sizes, more stable than those prepared, for instance, with the Eppenback Homomixer or other mechanical devices [17]. The best emulsions were prepared at optimum surfactant levels, optimum HLB [3] and high power. In the emulsions studied, the eects of HLB and surfactant parameters appear to be more important than those of the power parameters [13]. Besides, Reddy and Fogler [18] claim to have prepared very stable acoustic o/w emulsions (however, at very low volume fraction of oil ) without any surfactant: they invoke electrostatic stabilization due to preferential adsorption of OH ions at the oil/water interface. Emulsions may remain practically unchanged to the naked eye for several months, but will eventually return to their stable state, that is, a phase-separated system.

In emulsion breaking, two kinds of phenomena can be discriminated (Fig. 1): those, generally reversible, involving particle aggregation and migration and those, irreversible, related to particle size modication. On the one hand, reversible occulation of droplets can be followed by creaming or sedimentation, according to the respective densities of dispersed and continuous phases. The migration rate of the particles of the dispersed phase is given by Stokes law. On the other hand, irreversible changes, through Ostwald ripening and coalescence, lead to formation of larger drops, therefore to less and less stable emulsions and eventually to phase separation. Phase inversion can occur with temperature or composition change. To summarize, emulsion stability depends on: $ droplet size, $ density dierence between dispersed and continuous phases, $ viscosity of the contiuous phase, and $ above all, electrostatic and/or steric repulsion between droplets (for which the surfactant plays a major role). Ultrasound emulsication is thus an already known phenomenon [19], but in this work our goal was to undertake a systematic phenomenological study comparing the performances of ultrasound energy with those of classical mechanical agitation, all other things being

B. Abisma l et al. / Ultrasonics Sonochemistry 6 (1999) 7583

77

equal. Our main interests lie in the average size of droplets as well as their size distribution, and in emulsion stability.

Samples of 6 mL of each emulsion were observed in the Turbiscan (back-scattering mode) at 35C for only 25 min.

2. Experimental In the present work, parallel studies on batch emulsication processes were conducted through mechanical agitation and power ultrasound. An Ultra-Turrax (10 000 rpm, 170 W ) and an ultrasound horn (Misonix Sonicator XL 2020, 20 kHz, 130 W ) were used, respectively. Two series of experiments were then carried out with a model o/w emulsion system. The oil was kerosene (CAS Registry No. [8008-20-6 ], required HLB = 12) [19] and the surfactant a polyethoxylated (20 EO) sorbitan monostearate (CAS [9005-67-8], as Tween 60 or Montanox 60, kindly supplied by ICI and SEPPIC, respectively). Its cmc in pure water (8.105 M at 25C, with c =35 mN/m), was determined by surface tencmc sion measurements. The measured micelle size was ca. 4 nm (MALVERN Zetasizer 3000). Surfactant HLB is 14.9 [19,20] and cloud point 69C [21], according to NFT 73-403 annex [22]. Although Tween 60 is a nonionic surfactant, the f potential of the droplets amounts to 18 mV and the conductivity of the aqueous phase is not zero, either of which may be due to residual electrolyte coming from ethoxylation catalyst. Above surfactant cmc, the interfacial tension of the Tween 60H O/ kerosene system is 9.5 mN/m. Since the behavi2 our of metastable systems depends on the conditions of preparation of the nal mixture, the procedure must be rigorously the same for all experiments. A xed volume of emulsion (V=80 mL) was prepared at room temperature, the surfactant being rst dissolved in distilled water; oil was then added and energy supplied over a controlled period of time. The average drop size, expressed as the Sauter diameter (d =S n d 3 /S n d 2, representing a surface average 32 i i i i value), and the drop size distribution, were both obtained by means of a laser diractometer (Malvern MasterSizer S) according to the Mie theory. Emulsion stability was studied by visual inspection and with the help of the Turbiscan MA 1000 (Formulaction) [23]. This is a recently designed automated optical device, working in the near-infrared region with a double detection mode: transmission and backscattering (turbidity range from 0 to 50 000 NTU ). It carries out step-by-step vertical scans of a tube (atbottomed cylindrical cell ) containing a sample of a concentrated dispersion (such as emulsion, suspension or foam) and converts the macroscopic aspect of the mixture into graphics. Designed to work in the kinetic mode, it allows very early visualization of occulation, sedimentation, creaming and coalescence phenomena.

3. Results and discussion The parameters studied were the time of emulsication, the surfactant concentration, the power supplied and the volume fraction of oil. It is important to note that there was no sonochemical degradation of sodium dodecylsulfate, a model surfactant, under our working conditions. Therefore, we conclude that over the very short time of exposure to the power source, the surfactant used in our experiments was not decomposed either. 3.1. Variations of d versus exposure time 32 With a volume fraction of oil (w) of 0.25 and a surfactant concentration (c ) of 10 g /L, the variations of the Sauter diameter, d , as a function of the exposure 32 time are presented in Fig. 2. In both cases, d decreases 32 with the exposure time, but the most severe inuence is observed during the rst 15 s. After 30 s, the slopes approach zero. Anyway, the ultrasound technique always gives smaller droplet sizes and, after 30 s, the diameter is three times smaller with ultrasound than with mechanical agitation. In the following, the emulsication time was xed to 30 s. 3.2. Variations of d versus surfactant concentration 32 The next experiment describes the variation of d vs. 32 surfactant concentration, all other parameters being the same as previously (Fig. 3). The Sauter diameter decreases while the surfactant concentration increases up to ca.10 g /L. This follows a rather general observation, which makes sense, since the amount of surfactant available to form an adsorbed monolayer onto the emulsion droplets controls the interfacial surface area produced. At low surfactant concentration, the temporary interface is not completely covered, therefore drops coalesce and give rise to higher values of d . In fact, 32 knowing the area per polar head of the amphiphilic species from the slope of the Gibbs adsorption isotherm 2) should allow a check on the conditions of (A~115 A total coverage of the interface by surfactant molecules with the monolayer hypothesis. Two further comments may be added. Firstly, the cmc value does not seem to play a role in the variation of d vs. c: it has already been observed that, with non32 ionic surfactants, droplet size is further reduced above the cmc, perhaps because of the solubility of the nonionic in both phases [2]. Secondly, Fig. 3 shows that much less surfactant is required with ultrasound for a desired average drop size.

78

B. Abisma l et al. / Ultrasonics Sonochemistry 6 (1999) 7583

Fig. 2. Variations of d as a function of the exposure time (MA: mechanical agitation; US: power ultrasound ) (c=10 g/L, w=0.25, 32 P =170 W, P =130 W ). MA US

Fig. 3. Variations of d as a function of the surfactant concentration (MA: mechanical agitation, US: power ultrasound ) (t=30 s, w=0.25, 32 P =170 W, P =130 W ). MA US

3.3. Drop size distributions Let us now consider the possible relation between the surfactant concentration, the emulsication mode and the volume drop size distribution. The results are presented in Figs. 4 and 5. Two types of comparisons may

be made. Firstly, if the eect of surfactant concentration is considered, in both cases (ultrasound and agitation) volume drop size distributions atten with higher surfactant concentrations. In the case of ultrasound, we observe: a peaked distribution at low surfactant concentrations (c=1 g /L, standard deviation s =0.737) but a v

B. Abisma l et al. / Ultrasonics Sonochemistry 6 (1999) 7583

79

Fig. 4. Volume drop size distribution of o/w emulsions at various surfactant concentrations: (a) mechanical agitation, (b) power ultrasound.

Fig. 5. Volume drop size distribution: (a) low surfactant concentration, (b) high surfactant concentration.

attened distribution at higher surfactant concentrations (c=15 g /L, s =1.16). The same tendency is observed v in the case of mechanical agitation, with respective s v values of 4.802 and 10.535. The number drop size distribution would show the reverse trend with respect to the concentration scale (s small for high surfactant n concentration). Secondly, if the two techniques are compared, the standard deviation values, s and s , as v n well as the d values, are always much lower with the 32 ultrasound technique. 3.4. Variations of d versus power supplied 32 For an oil volume fraction of 0.25 and with 10 g /L surfactant, mechanical agitation always gives complete emulsication in 30 s, whatever the power involved. On the contrary, with the ultrasound device, a minimum power of 90 W is necessary to emulsify the mixture completely. In fact, in the particularly unfavourable cases of mercury-in-water and analogous liquid metalin-water emulsion systems (very high density dierences), no emulsication occurs below the threshold of cavitation [24]. Now, cavitation is produced above only 2 W/cm2. This means that, in our experimental conditions (geometrical features of the equipment, rather large volume of emulsion and volume fraction of oil ),

the sole ultrasound energy does not allow ecient breaking of the bulk oil phase into primary large drops. This is likely to be due to insucient acoustic streaming under the horn. Gentle stirring in batch operation or continuous ow process should easily improve this rst step. Anyway, between 90 and 130 W, the Sauter diameter values are about four times smaller with ultrasound ( Fig. 6). The reported power P for mechanical agitation is in fact the dierence of power consumption between the loaded and unloaded states at various rotation speeds. Now, let us try to determine the quantity of heat (Q ) absorbed by water per unit of time at the starting of the process. The well-known calorimetric formula reads: Q=m C (dT/dt ) , p t=0 where m is the mass of the system, C its heat capacity, p T the absolute temperature and t the time. The measurement supposes that all ultrasound energy that propagates in the liquid is converted into heat: the loss of energy by thermal exchange at the liquidliquid interface and at the walls of the vessel is therefore neglected.This is veried on the initial straight part of T vs. t curves, whose slopes yield values of Q to be compared with the corresponding electrical inlet power values. According to this approach through very simple

80

B. Abisma l et al. / Ultrasonics Sonochemistry 6 (1999) 7583

Fig. 6. Variations of d as a function of the power (t=30 s, c=10 g /L, w=0.25 ). 32

experiments involving pure water, a much lower dissipated energy has been evidenced in the case of ultrasound generation ( Table 1). Therefore, ultrasound, although providing less energy, aords better emulsication performances. 3.5. Variations of d versus volume fraction of oil 32 Finally, let us consider the eect of w on d . 32 Surfactant concentration is now constant with respect to aqueous phase and not to the total sample volume As expected, we then get the reverse trend: in both cases, the Sauter diameter increases with w ( Fig. 7). Since emulsication is not always complete at high w values, the other purpose of this experiment is to determine the dispersive power of the two techniques towards kerosene. Here again, ultrasound appears as more ecient than agitation, all other things being equal. Besides, phase inversion, water then becoming the dispersed phase, occurs around equal volumes of both phases. The observed phenomenon can be illustrated by a simplied scheme (Fig. 8). On the whole, ultrasound yields more complete and ner dispersions.
Table 1 Electrical power (P ) and evolved heat ux (Q ) values P (W) Mechanical agitation Ultrasound 170 130 Q (W) 120 53

3.6. Compared stability of emulsions Simple visual inspection shows longer stability of ultrasound-made emulsions. For instance, the same extent of creaming appears 6 days after insonation vs. 1 day after mechanical emulsication. The Turbiscan MA 1000 [23] shortens the period of observation required to draw a conclusion, somewhat predicting the future behaviour of an apparently stable emulsion. A parallel study, carried out with this apparatus, shows the following types of back-scattering intensity variations (DI ): $ a drop of DI at the bottom of all samples, due to clarication of the mixture, $ an increase of DI at the top of the less-concentrated samples, with a typical isosbestic point, associated to creaming (particle migration), $ a drop of DI at the top of the more-concentrated samples, without isosbestic point, coming from clarication (drop size increase followed by oil phase separation). Reporting back-scattering intensity data for agitated and insonated emulsions at 35C for 25 min, Fig. 9 shows increasing creaming as a function of time, but the main dierence lies in the larger extent of clarication at the bottom of the tube for an Ultra-Turraxmade emulsion, compared with the insonated sample. Since the continuous and dispersed phases are immiscible, Ostwald ripening can be neglected and destabilization originates only in coalescence. Fig. 10 summarizes

B. Abisma l et al. / Ultrasonics Sonochemistry 6 (1999) 7583

81

Fig. 7. Variation of d as a function of oil fraction (t=30 s, c=10 g/L, P =170 W, P =130 W ). 32 MA US

Fig. 8. General behaviour of agitated (AG) and utrasound-made ( US ) o/w emulsions as a function of oil volume fraction (w).

Fig. 9. Back-scattering variations in the Turbiscan sample tube for agitated ( left) and insonated (right) kerosene-in-water emulsions.

82

B. Abisma l et al. / Ultrasonics Sonochemistry 6 (1999) 7583

Fig. 10. Destabilization of agitated and insonated kerosene-in-water emulsions observed at the top of the samples as a function of oil volume percent.

the behaviour of kerosene emulsions as a function of oil volume fraction. The bars represent the initial destabilization rates, as dierences in back-scattered intensities per hour at the top of the samples. Below 25% oil, both emulsions are rather stable, with some creaming. Above this value, agitated and insonated emulsions behave dierently: clarication of the medium, reecting phase separation, is much faster after mechanical agitation and increases with higher oil volume fraction. It appears that, under those conditions, mechanical and ultrasound emulsication are ecient up to about 25% and 37.5% oil phase, respectively. Consistently with the droplet size dierences, the rst results regarding the stability of emulsions obtained by both techniques therefore indicate a much higher stability of ultrasound-made emulsions. Moreover, at constant overall surfactant concentration, a higher interfacial area implies a higher amount of adsorbed surfactant, therefore a lower amount of micellized surfactant. Then, the depletion interaction (attraction between droplets associated with the presence of micelles) could be weakened, which would enhance emulsion stability.

aords several interesting results in favour of the ultrasound technique: $ smaller average drop sizes d (down to 0.3 mm) can 32 be obtained, $ for a given desired diameter, the surfactant amount required is reduced, $ energy consumption (through heat loss) is lower, $ ultrasound-made emulsions are less polydispersed and more stable. Hopefully, those rst results will be further explained thanks to a better knowledge of cavitation phenomena. An extension of this work would require the direct measurement of acoustic bubble size and their distribution as a function of the working parameters.

Acknowledgements The authors thank Formulaction S.A., and especially K. Puech, for performing the stability study with their specic equipment ( Turbiscan MA1000).

References 4. Conclusion With our simple, three-component, model system, the comparison of two types of emulsication processes, the rst one using mechanical agitation and the second one involving power ultrasound at low frequency (20 kHz),
[1] P. Becher, Emulsions, Theory and Practice, second ed. Reinhold, New York, 1965. [2] S.E. Friberg, S. Jones, in: J.I. Kroschwitz ( Ed.), KirkOthmer Encyclopedia of Chemical Technology, fourth ed. vol. 9 , Wiley, New York, 1994, pp. 393413. [3] P. Becher, J. Disp. Sci. Technol. 5 (1985) 8196.

B. Abisma l et al. / Ultrasonics Sonochemistry 6 (1999) 7583 [4] J.C. Johnson, Emulsiers and Emulsifying Techniques, Noyes Data, Park Ridge, NJ, 1979. [5] N. Shahidzadeh, D. Bonn, O. Aguerre-Chariol, J. Meunier, Proceedings, 2e Congre ` s Mondial de lEmulsion, Bordeaux, 2326 Sept. 1997, vol.1, 1-2-212/0106. . [6 ] P. Walstra, Chem. Eng. Sci. 48 (2) (1993) 333349. [7] M.K. Li, H.S. Fogler, J. Fluid Mech. 88 (3) (1978) 499511. [8] M.K. Li, H.S. Fogler, J. Fluid Mech. 88 (3) (1978) 513528. [9] R.W. Wood, A.L. Loomis, Phil. Mag. 4 (1927) 417. [10] Swiss Patent 394.390, 1944. [11] B. Brown, J.E. Goodman, High-Intensity Ultrasonics, Ilie Books, London, 1965. [12] H.M. Beal, D.M. Skauen, J. Am Pharm. Ass., Sci. Ed. 44 (1955) 490. [13] D.M. Higgins, D.M. Skauen, J. Pharm. Sci. 61 (10) (1972) 1567. [14] A.E. Alegria, Y. Lion, T. Kondo, P. Riesz, J. Phys. Chem. 93 (1989) 49084913. [15] K. Makino, M.M. Mossoba, P. Riesz, J. Am. Chem. Soc. 104 (1982) 8537.

83

[16 ] K. Makino, M.M. Mossoba, P. Riesz, J. Phys. Chem. 87 (1983) 1369. [17] I. Aroutiounov, B. Toumanian, V. Melikian, D. Kovalenko, V. Zubov, Proceedings, 2e Congre ` s Mondial de lEmulsion, Bordeaux, 2326 Sept., vol.1, 1-2-106/0105, 1997. [18] S.R. Reddy, H.S. Fogler, J. Phys. Chem. 84 (1980) 15701575. [19] W.C. Grin, in: M. Grayson (Ed.), KirkOthmer Encyclopedia of Chemical Technology, 3rd ed. vol. 8 , Wiley, New York, 1979, pp. 900930. [20] W.C. Grin, J. Soc. Cosmet. Chem. 5 (1954) 249256. [21] SEPPIC, technical literature. [22] Recueil de normes franc tergents, aises: Agents de surface, de savons, AFNOR, Paris-la-De fense, 1986. [23] G. Meunier, I. Cayre, K. Puech, Proceedings, 2e Congre ` s Mondial de lEmulsion, Bordeaux, 2326 Sept., vol. 2, 2-1-213/0106, 1997. [24] A.E. Crawford, Techniques des Ultrasons. Applications Basse et Haute Puissance, Dunod, Paris, 1959.