pubs.acs.org/Langmuir © 2009 American Chemical Society
From Spherical to Polymorphous Dispersed Phase Transition in Water/Oil Emulsions )
:: M. Schmitt-Rozieres,† J. Kragel,‡ D. O. Grigoriev,‡ L. Liggieri,§ R. Miller,‡ S. Vincent-Bonnieu, and M. Antoni*,† †
)
Aix-Marseille Universit e - Universit e Paul C ezanne UMR-CNRS 6263 ISM2, Centre St. J er^ ome, BP. 451, :: Marseille 13397 Marseille, Cedex 20, France, ‡MPI of Colloids and Interfaces, Am Muhlenberg 1, 14476 Potsdam-Golm, Germany, §CNR-IENI, Via de Marini 6, 16149 Genova, Italy and ESA, Physical Science Unit, Keplerlaan 1, 2200 AG Noordwijk, The Netherlands Received December 22, 2008. Revised Manuscript Received February 5, 2009
Optical scanning tomography is used to characterize bulk properties of transparent water-in-paraffin oil emulsions stabilized with hexadecyl-trimethylammonium bromide (CTAB) and silica nanoparticles. A flow of 500 hundred images is used to analyze each scanning shot with a precision of about 1 μm. The role of silica particles in the shape of the water droplets is investigated. Depending on the concentration of CTAB and silica nanoparticles, a transition occurs in their geometry that changes from spherical to polymorphous. This transition is controlled by the ratio R = [CTAB]/[SiO2] and is described using an identification procedure of the topology of the gray level contours of the tomographic images. The transition occurs for Rcrit ≈ 3 10-2 and is shown to correspond to a pH of the dispersed phase of 8.5.
Introduction It is well established that the properties of liquid/liquid interfaces play a key role in the evolution of emulsions. One common technique used to study this evolution consists of measuring the droplet size distribution with spectroscopic methods and Mie theory. This technique is nonintrusive, but it supplies only averages and does not allow, for example, studies of the dynamics of droplets or the description of their shape. It relies moreover on the assumption of spherically shaped droplets. We show in this work that, in specific conditions, droplets can become highly deformed even for diluted emulsions and that optical scanning tomography can provide important insights into their geometry. Many aspects of emulsion properties1,3 are ruled by the phenomena arising at liquid/liquid interfaces.4,6 For dense emulsions, static models resulting from DLVO theory7 taking into account drainage of interdroplet films8 have been developed and compared with experimental studies. In the more specific case of diluted emulsions, flocculation models were developed to describe the aggregation of droplets under Brownian motion.2 For such flocculated emulsions and for small droplets, there are still doubts about the possibility to (1) Danov, K. D.; Ivanov, I. B.; Gurkov, T. D.; Borwankar, R. P. J. Colloid Interface Sci. 1994, 167, 8–17. :: (2) Dukhin, S. S.; Sjoblom, J. Kinetics of Brownian and gravitational coagulation in dilute emulsions. In Emulsions and emulsion stability; Sjoblom, J., Ed.; Dekker: 1996; Chapter 2. (3) Sanfeld, A.; Steinchen, A. Adv. Colloid Interface Sci. 2008, 140, 1–65. (4) Derjaguin, B. V.; Landau, L. Acta Physicochim. 1941, 14, 633. (5) Verwey, E. J.; Overbeek, J. T. G. Theory of the Stability of Lyophobic Colloids; Elsevier: Amsterdam, 1948. (6) Gurkov, T. D.; Basheva, E. S. In Hydrodynamic Behavior and Stability of Approaching Deformable Drops; Hubbard, A. T., Ed.; Marcel Dekker: New York, 2002. (7) Radoev, B. P.; Scheludko, A. D.; Manev, E. D. J. Colloid Interface Sci. 1983, 95, 254–265. (8) von Smoluchowski, M. Phys. Z. 1916, 17(517), xxx–xxx.
4266
DOI: 10.1021/la804214m
deliver experimental verifications of the droplet flattening phenomenon in the contact region.3 Recent work has shown that this flattening can significantly increase the coupling between deformed droplets.3 This important point is one motivation for the experimental study of model emulsions by optical tomography. In the following, we will however limit our focus on the description of the topology of droplets and on the role of silica nanoparticles. Pickering emulsions are stabilized by solid particles of micro- or nanometer size.9,12 Properties of emulsions involving in particular silica particles attracted attention again in the past decade and have been discussed in the literature with an increasing number of publications.13,14 Due to the presence of nanoparticles at the liquid/liquid15 or liquid/gas16 interfaces, such emulsions raise fundamental problems in the interpretation of the measured interfacial tension as well as in the characterization of their rheological properties.17,18 In particle stabilized emulsions, the interfaces do not involve two phases but three because the particles are solid. The wetting properties of the latter control their adsorption at liquid/liquid interfaces and can be tuned by the addition of opposite charged surfactants. In the case of the silica (9) Binks, B.; Horozov, T. In Colloidal Particles at Liquid Interfaces; Binks, B., Horozov, T. Eds.; Cambridge University Press: 2006. (10) Binks, B.; Fletcher, P. Langmuir 2001, 17, 4708–4710. :: (11) Romain, F. L.; Schmidt, M.; Lowen, H. Phys. Rev. E 2000, 61(5), 5445–5451. (12) Pickering, S. U. J. Chem. Soc. 1907, 91, 2001–2021. :: (13) Horvolgyi, Z.; Mate, M.; Daniel, A.; Szalma, J. Colloids Surf., A 1999, 156, 501–508. :: (14) Kondo, M.; Shinozaki, K.; Bergstrom, L.; Mizutani, N. Langmuir 1995, 11, 394–397. (15) Santini, E.; Liggieri, L.; Sacca, L.; Clausse, D.; Ravera, F. Colloids Surf., A 2007, 309, 270–279. (16) Morehouse, D. S.; Tetrault, R. J.U.S. Patent 3615972, 1976. (17) Hassender, H.; Johansson, B.; Tornell, B. Colloids Surf., A 1989, 40, 93. (18) Lan, Q.; Yang, F.; Zhang, S.; Liu, S.; Xu, J.; Sun, D. Colloids Surf., A 2007, 302, 126–135.
Published on Web 3/12/2009
Langmuir 2009, 25(8), 4266–4270
Letter
nanoparticles that we propose to focus on, surface charge is negative and will be tuned by the addition of cationic surfactant CTAB. Recent studies focused on the interfacial properties of mixed silica particles/CTAB mixtures.19,20 At a critical silica particle/CTAB ratio, an irreversible attachment of particles on the interfaces is evidenced. It is followed by the formation of a solid-like layer.
Materials and Methods Emulsions are formulated in two steps. First, a sample of silica nanoparticle dispersion with suitable particle concentration is obtained, diluting a commercial colloidal silica dispersion (Levasil 200/30%, kindly supplied by the producer Stark GmbH/ Germany) by solutions of the cationic surfactant hexadecyltrimethylammonium bromide (CTAB, Fluka ultra grade 52365) in water (HPLC grade) plus sodium chloride. In the reported study, variable silica and CTAB concentrations were utilized, while the NaCl concentration was always fixed to 1 mM. The size of silica particles in the original Levasil dispersion is of about 10 nm for an actual surface area of about 200 m2/g. The dispersion shows an excellent stability over several months due to the large hydrophilicity of the particles which present a large negative surface charge, with a zeta potential of about -40 eV. As a consequence, the original dispersion is highly alkalic with a stable pH of about 9.2. The diluted dispersion samples with CTAB show instead a decreasing kinetics of the pH, starting from the above value and relaxing to a final value (pHfinal) after typically 4-5 h. Such kinetics originates in the neutralization of the particle surface charge under the effect of the adsorption of the CTAB cations at the particle surface. Once pHfinal is reached, the diluted dispersion is emulsified in paraffin oil (Fluka 76235) which has been used without further purification. The volume ratio dispersion/paraffin oil in all investigated emulsions was fixed to 0.01, and all experiments were performed at 20 °C. The emulsification of the aqueous silica nanoparticle/CTAB mixtures in paraffin oil emulsion has been done under a controlled protocol. After introduction of the dispersion sample in paraffin oil, both fluids are degassed for 10 min under vacuum conditions. Emulsification is then achieved by magnetic stirring over 10 min at 800 rpm. These conditions allow good reproducibility of the experiments in spite of the supplementary constraints in the protocols of emulsification. The resulting emulsions are analyzed with optical tomographic microscopy. This technique combines a classical microscope in transmitted light mode with the acquisition capabilities of CCD camera and a step by step controlled moving objective. It is nonintrusive and allows a direct in situ observation of emulsion features far from the emulsion container walls. The basic ideas of the analysis rely on a precise gray level contouring of each scanned image and finally on the recognition of droplet relevant contours.21 Results presented below are obtained for 1 s tomographic shots of an emulsion volume of 1 mm3 and 5 mm inside the samples where wall effects are negligible. The number of gray level contours used for image treatment is set to 30.
Results and Discussion Figure 1 illustrates the problem in which we are interested to investigate for two different compositions of CTAB and silica nanoparticles. Both images show a very different result in emulsification. While for a higher ratio of R = [CTAB]/[SiO2] spherical drops have been observed, for a lower R emulsions (19) Ravera, F.; Santini, E.; Loglio, G.; Ferrari, M.; Liggieri, L. J. Phys. Chem. B 2006, 110, 19543–19551. (20) Ravera, F.; Ferrari, M.; Liggieri, L.; Loglio, G.; Santini, E.; Zanobini, A. Colloids Surf., A 2008, 323, 99–108. :: (21) Antoni, M.; Kragel, J.; Liggieri, L.; Miller, R.; Sanfeld, A.; Sylvain, J. D. Colloids Surf., A 2007, 309, 280–285.
Langmuir 2009, 25(8), 4266–4270
Figure 1. Snapshots of dispersed phase structure: [CTAB] = 2.91 g/L and [SiO2] = 6.06 g/L (left image); [CTAB] = 3.58 10-1 g/L and [SiO2] = 27.05 g/L (right image). are mainly composed of deformed drops without clear symmetry. This observation demonstrates the role of emulsion stabilization by a classical surfactant and the important modification that can show up when nanoparticles are utilized. The strong deformation of the dispersed phase due to a higher concentration of silica particles is a phenomenon which cannot be explained completely so far.19,20 In the case of bubble systems, non spherical geometries have been discussed in the literature, for example by Subramanian et al.22 The observed deformed droplets are constituted by polymorphic and stiff objects for which no deformation is observable on the time scales on which we worked, typically 1 s, up to several tens of seconds. To characterize emulsions of this type, we introduce the parameter R = [CTAB]/[SiO2] giving the ratio of the concentration of CTAB on that of silica nanoparticles. Regarding to the stable value of the pH of the dispersion samples, we assume that there is no further chemical reaction which modifies the dispersed phase composition after the emulsions have been created. This is confirmed by the realization of further emulsion analysis yielding results similar to the ones shown in Figure 1. The difference in shape of the drops is clearly linked to the composition of the dispersed phase that modifies the interfacial properties. However, the existence of complex silica particle organization inside the drops might also significantly contribute to this effect. To qualitatively analyze the transition from emulsions with spherical droplets to emulsions with polymorphous droplets, we characterize emulsions with different mixing ratios R using optical tomography. Note, however, that for deformed objects as the ones represented in Figure 1 it is impossible to use the simplifications associated to a spherical geometry of droplets. Therefore, the analysis of drop shape has been limited to find the critical dispersion sample composition(s) where the transition from spherical to polymorphic droplets occurs. For this purpose, the capacity of image treatment software to identify every outline for all gray levels and for all images of each tomographic scanning shot has been used. An immediate possibility for the image treatment consists of the analysis of the deformation of the gray level contours by using the standard deviation σ of each contour with respect to a circular reference contour having exactly the same perimeter. With this respect, small values of σ correspond to almost circular contours while large values correspond to deformed situations. Figure 2 illustrates some possible contour topologies for different values of σ. The output analysis of tomographic sequences strongly depends on the quality of the contrasts and of the noise level (22) Subramaniam, A. B.; Mejean, C.; Abkarian, M.; Stone, H. A. Langmuir 2006, 22, 5986.
DOI: 10.1021/la804214m 4267
Letter
Figure 2. Typical topology of gray level contour resulting from image treatment for various values of σ. of each image. In the following, both parameters are adjusted in ways to have 106-107 gray level contours for each tomographic sequence. This is sufficient to perform a reliable statistical analysis on the contours’ topology for each emulsion. Figure 3 illustrates this image treatment procedure for a given image and for two different emulsion compositions. Both emulsions show different emulsification characteristics. Figure 3a shows small droplets and corresponds to a dispersion composition providing spherical drops after emulsification. Conversely, Figure 3b shows larger deformed objects due to the larger amount of nanoparticles in the dispersion. Contour plots of both images show qualitatively that the relative proportion of circular contours (assumed here to be such that σ e 0.05) is larger for the spherical than for the nonspherical dispersed phase. It is the relative number of circular contours and noncircular ones that will allow for the transition location on a quantitative basis. We define to this end the distribution F(σ, GL) which describes the probability to obtain, among all the contours, a contour having a given gray level value (GL) and standard deviation (σ). The resulting distributions for the emulsions presented in Figure 3 are plotted in Figure 4 but now for two full tomographic sequences of 500 images. The shape of F(σ, GL) is clearly different for the two emulsions. The emulsion with R = 1.20 10-2 (respectively R = 0.1) has nonspherical (spherical) drops and shows a flattened (peaked) shape. For R = 0.1, the peak shows up in the range of small and intermediate values of gray levels (GL ≈ 128) as illustrated in Figure 4a. This maximum demonstrates the fact that contours in the surrounding of each drop, that are in the tones of gray for the image corresponding to Figure 4a, have a strong probability to be circular. Figure 4b shows however that the proportion of circular contours in the gray tones is less important because the maximum of the distribution F(σ, GL) is flattened. A stronger broadening of the distribution is also observed as a general rule for polymorphic emulsions. Several criteria allow defining a correct order parameter to discriminate between the structures of emulsions. The roughest criterion has been chosen here in defining simply a fixed constant threshold value of standard deviation σ. This threshold (σthresh) is illustrated in Figure 5 for σthresh = 0.5. Ncirc is then defined as the number of gray level contours such that σ e σthresh, whereas Nnoncirc is the number of contours such 4268
DOI: 10.1021/la804214m
Figure 3. Image treatment for two emulsions with R = 0.1 (a) and
R = 1.20 10-2 (b). Resulting contour plots are represented on the right. Black lines correspond to circular contours (σ e 0.05). Distorted contours are plotted in gray.
Figure 4. Distribution F(σ, GL) (right images) as a function of standard deviation σ and gray level (GL) for the same emulsion than in Figure 3: R = 0.1 (a) and R = 1.20 10-2 (b). Images are reproduced from Figure 3. Distributions F(σ, GL) result from the analysis of 500 images grabbed in 1 s and correspond to a 1 mm3 emulsion volume.
that σ > σthresh. Both Ncirc and Nnoncirc depend on the value of σthresh. Practically, the introduction of this threshold allows a distinction less restrictive than the one used in Figure 3. Indeed, contours still remain close to circles even when σ e 0.5. It is hence possible to interpret here Ncirc (respectively Nnoncirc) as the number of contours “rather” circular (noncircular). From these two definitions, it is possible to introduce the ratio ξ defined as ξðRÞ ¼
Ncirc Nnoncirc
where R was defined previously. When ξ is close to 1, the number of circular contours is larger and the dispersed phase will be spherical. When ξ < 1, noncircular contours dominate and the dispersed phase will be polymorphic. The value of ξ as a function of the dispersion composition is represented in Figure 6. For R < 3 10-2, we get ξ < 0.65, Langmuir 2009, 25(8), 4266–4270
Letter
Figure 5. Illustration of the definition of σthresh for the calculation of ξ(R) in the case of the distribution of Figure 4a. showing that noncircular contours prevail. This is the signature of a polymorphous dispersed phase emulsion. For R > 3 10-2, ξ is typically larger than unity, indicating that circular and noncircular contours have the same statistical weight and that the corresponding dispersed phase has spherical geometry. Figure 6 shows a clear jump in the values of ξ in the transition region. The height of this jump depends on the chosen threshold value σthresh. For σthresh = 0.5, this gap has a value of about 0.4 and is significant regarding the fluctuations in the value of ξ. It hence allows the discrimination of the considered emulsion type and therefore constitutes a reliable quantitative parameter for the characterization of the structure of the dispersed phase. The critical value Rcrit for the dispersion composition in CTAB and silica nanoparticles can be obtained from Figure 6 and gives Rcrit = 3 10-2 ( 1 10-2. It is important to note here that as it is the rate R that determines the transition, the initial dispersion composition nearby the transition is not unique. It is indeed the relative balance of CTAB and nanoparticle concentration in the dispersion composition that is the key parameter here. This indicates that the drops’ deformation results from a complex interplay between both of these components and again suggests that a complex organization of the nanoparticles inside them cannot be excluded to explain their polymorphous geometries. Figure 7 shows the final pHfinal of the dispersion for different ratios R. The measurements of pHfinal have been performed 5 h after the mixture of the components of the dispersed phase (water + CTAB + NaCl + silica nanoparticles). Changes in pHfinal upon increase of R can be understood as a chemical exchange between silanol groups on the nanoparticle surface and surface-active CTA+ cations according to the reaction: SiO - H þ þ CTA þ T SiOCTA þ H þ Three different families of points are observed in Figure 7. The first has been obtained for dispersions which cannot be emulsified in paraffin oil due to a too low concentration of CTAB. The second family corresponds to dispersions that can be emulsified and form spherical droplets. The last family gives rise to emulsions with polymorphic drops. The critical value of Rcrit previously obtained from image treatment is also presented in Figure 7 and corresponds to the critical value of pHfinal of the dispersion that turns out to be close to 8.5. The higher the R value, the more CTA+ cations are bound at the nanoparticles’ surface and the more protons are liberated and Langmuir 2009, 25(8), 4266–4270
Figure 6. Value of ξ(R) as a function of R = [CTAB]/[SiO2] for
σthresh = 0.5. Vertical line indicates the position of the transition region.
Figure 7. Final pHfinal of the dispersion (water + CTAB + NaCl + silica nanoparticles) as a function of R = [CTAB]/[SiO2]. The vertical line corresponds to Rcrit = 3 10-2. Circles (respectively squares) correspond to emulsions with spherical (polymorphous) droplets. Triangles correspond to compositions where emulsification is not possible.
decrease the pH value of the mixture. Because the original dispersion of silica nanoparticles is alkalic (pH = 9.2), the liberation of protons by the action of cations CTA+ results in a gradual neutralization of the initially basic dispersion. Figure 7 represents in fact a typical base-with-acid titration curve. For R > Rcrit until approximately R = 0.1, the steepest decrease of pHfinal is observed with the inflection point around 7, corresponding to the equivalence point of the base-acid titration. It is worth comparing the results of Figures 6 and 7 with zeta potential measurements. Figure 8 shows the zeta potential of a 1 wt % dispersion of SiO2 versus R = [CTAB]/[SiO2].23 As one can see, the value Rcrit in Figure 8 corresponds to a R region still characterized by a large negative surface charge and hence by a stable dispersion. The range of the steepest decrease in pHfinal corresponds to the region of R where the steepest increase of the zeta potential is observed. This correlation additionally supports the electrostatically driven mechanism of adsorption of CTAB on the silica nanoparticles (23) Miller, R. Microgravity Sci. Technol. 2006, 18, 104–107.
DOI: 10.1021/la804214m 4269
Letter
range of R, all negatively charged sites on the surfaces of nanoparticles are screened and additional adsorption of CTA+ cations via hydrophobic interaction approaches almost its equilibrium value. Therefore, in the discussed R range, the adsorbed CTAB should form an almost complete bilayer at the silica surface. This finding agrees well with the data of Wang et al.24 who suggested that, at R = 0.146, CTA+ cations adsorbed at the silica surface form bilayers with a molecular area of 32 nm2.
Conclusions
Figure 8. Zeta potential of a 1 wt % silica dispersion in CTAB aqueous solutions containing 1 mM NaCl. Vertical line corresponds to Rcrit = 3 10-2. at least at low values of R. The isoelectric point is attained at R ≈ 0.07 which almost exactly coincides with the equivalence point in the pHfinal versus R dependence displayed in Figure 7. The transition of zeta potential to positive values simultaneously with the further decrease of pHfinal at R > 0.07 could be interpreted in the framework of the coexistence of two competitive adsorption mechanisms. On one hand, the zeta potential increase to positive values indicates further adsorption of CTA+ cations due to hydrophobic interactions either between already adsorbed and new CTA+ (onset of the CTA+ bilayer formation) or between hydrophobic sites on the silica nanoparticle surface and CH-chains of CTA+. On the other hand, the continuous decrease of pHfinal in the same R range gives evidence for the further liberation of protons, that is, for the continued adsorption by electrostatic interaction. Only at R ≈ 0.15, the pHfinal becomes stable and the zeta potential attains the plateau value of about +26 mV. In this (24) Wang, W. G. B.; Liang, L.; Hamilton, W. A. J. Phys. Chem. B 2004, 108(45), 17477–17483.
4270
DOI: 10.1021/la804214m
Colloidal silica dispersions have been investigated with different amounts of CTAB to tune the hydrophobicity of particles with the aim to understand their effect in diluted water-in-paraffin oil emulsions. Optical tomography studies performed with different CTAB/silica particle mixtures provide information of an irreversible organization of the particles at a critical CTAB/silica particle ratio. Due to this phenomenon, a solid-like behavior appears for the interfaces which influences the morphology of the droplets. To characterize the transition from emulsions with spherical droplets to emulsions with polymorphous droplets, we analyze them with different mixing ratios R using the above-described image treatment routine. Below a critical ratio of Rcrit = 3 10-2, irregular drop shapes have been observed, while above this critical ratio drops are spherical. Due to the definition of sharp criteria, a determination of the transition from spherical to polymorphic droplets is possible. This transition is compared with zeta potential measurements and CTAB layer formation at the nanoparticles interface. Good agreements are found for the critical composition of the dispersed phase. Acknowledgment. This work was partially supported by the European Science Foundation COST Actions P21 and D43, the DFG SPP 1273 Mi 418/16-1, ESA MAP-FASES, CNES, and GdR-CNRS MFA and Mousses. We also thank Annie Steinchen and Albert Sanfeld for important stimulating discussions and help.
Langmuir 2009, 25(8), 4266–4270
