Colloids and Surfaces B: Biointerfaces 128 (2015) 127–131
Contents lists available at ScienceDirect
Colloids and Surfaces B: Biointerfaces journal homepage: www.elsevier.com/locate/colsurfb
Formulation and optimization by experimental design of eco-friendly emulsions based on d-limonene Luis M. Pérez-Mosqueda, Luis A. Trujillo-Cayado, Francisco Carrillo, ˜ Pablo Ramírez ∗ , José Munoz Departamento de Ingeniería Química, Facultad de Química, Universidad de Sevilla, P. García González 1, 41012 Sevilla, Spain
a r t i c l e
i n f o
Article history: Received 11 December 2014 Received in revised form 28 January 2015 Accepted 13 February 2015 Available online 21 February 2015 Keywords: d-Limonene Emulsion Ostwald ripening Pluronic Response surface methodology
a b s t r a c t d-Limonene is a natural occurring solvent that can replace more pollutant chemicals in agrochemical formulations. In the present work, a comprehensive study of the influence of dispersed phase mass fraction, , and of the surfactant/oil ratio, R, on the emulsion stability and droplet size distribution of d-limonene-in-water emulsions stabilized by a non-ionic triblock copolymer surfactant has been carried out. An experimental full factorial design 32 was conducted in order to optimize the emulsion formulation. The independent variables, and R were studied in the range 10–50 wt% and 0.02–0.1, respectively. The emulsions studied were mainly destabilized by both creaming and Ostwald ripening. Therefore, initial droplet size and an overall destabilization parameter, the so-called turbiscan stability index, were used as dependent variables. The optimal formulation, comprising minimum droplet size and maximum stability was achieved at = 50 wt%; R = 0.062. Furthermore, the surface response methodology allowed us to obtain the formulation yielding submicron emulsions by using a single step rotor/stator homogenizer process instead of most commonly used two-step emulsification methods. In addition, the optimal formulation was further improved against Ostwald ripening by adding silicone oil to the dispersed phase. The combination of these experimental findings allowed us to gain a deeper insight into the stability of these emulsions, which can be applied to the rational development of new formulations with potential application in agrochemical formulations. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Many traditional industrial products are being gradually replaced by environmental friendly alternatives. Organic solvents are the main source of waste from academic and industrial installations as reaction medium, in separation procedures, and as diluters. The role of green solvents in the chemical and pharmaceutical industries is becoming more and more important [1]. This work deals with the development of emulsions containing a green solvent, which may find applications as matrices for incorporation of agrochemical active ingredients. The role of solvents in agrochemical industry is becoming increasingly more important [2]. More than 25% of all pesticides contain high concentrations of organic solvents, which represent a fire hazard, may be toxic and contribute to atmospheric volatile compound (VOC) emissions [3].
∗ Corresponding author. Tel.: +34 954557179. E-mail address: [email protected] (P. Ramírez). http://dx.doi.org/10.1016/j.colsurfb.2015.02.030 0927-7765/© 2015 Elsevier B.V. All rights reserved.
d-Limonene is a naturally occurring a cyclic monoterpene, which is commonly found in the rinds of citrus fruits such as grapefruit, lemon, lime, and in particular, oranges. Limonene exhibits good biodegradability; hence it may be proposed as an interesting alternative to organic solvents, meeting the ever-increasing safety and environmental demands of the 21st Century [4,5]. Furthermore, d-limonene also finds applications in many other fields such as food and pharmaceutical industries [6]. Given that d-limonene is not soluble in water, emulsification is an interesting alternative to solubilize and protect d-limonene from oxidative degradation [7]. In order to emulsify d-limonene, non-ionic polymeric surfactants, Pluronics, have been used. Their chemical structure is denoted by PEOx–PPOy–PEOx, where PEO and PPO stands for poly(ethylene) oxide and poly(propylene) oxide, respectively. They have been used as foam and emulsions stabilizers in industrial applications [8,9]. Recently, the interfacial properties of Pluronic PE9400 adsorbed at the limonene–water interface has been reported by [10], showing that PE9400 is irreversible adsorbed
128
L.M. Pérez-Mosqueda et al. / Colloids and Surfaces B: Biointerfaces 128 (2015) 127–131
at this interface, thereby being a potential candidate to prevent emulsion coalescence. Emulsions are thermodynamically unstable colloidal dispersions, which are destabilized by several mechanisms such as flocculation, coalescence, creaming and Ostwald ripening. It has been shown that orange oil suffers from creaming and Ostwald ripening [11]. It is known that both destabilization processes, creaming and Ostwald ripening, depend on dispersed phase mass fraction, , and surfactant concentration [12,13]. Therefore, a right selection of composition and physicochemical formulation is essential to improve the emulsification process [14]. Multivariate statistical techniques such as response surface methodology (RSM) have shown to be an interesting tool for the optimization of emulsions and nanoemulsions [15]. With this approach, a mathematical model connecting the response variables to the process ones is obtained with a minimum number of experiments. Hence, the development of new formulations becomes more effective [16]. The most common emulsification methods are based on mechanical energy input to the system by an external source. As a rule, emulsions are prepared in two steps; the aim of the first (primary homogenization) is to create droplets of dispersed phase such that a coarse emulsion is formed. The goal of the second step is to reduce the size of preexisting droplets, which usually involves the use of a different homogenizer, such as ultrasonic probes, highpressure valve homogenizers or microfluidizers [17]. Nevertheless, in the present study a single-step rotor/stator homogenizer was used in order to obtain monomodal submicron emulsions with dispersed phase mass fraction in the range from dilute to slightly concentrate emulsions ( = 10–50 wt%) and with a low ratio of surfactant concentration over solvent concentration (R < 0.1). An experimental full factorial design 32 was conducted in order to optimize the emulsion formulation.
2.3. Droplet size measurements Droplet size distribution (DSD) was determined by using laser diffraction measurements performed with a Mastersizer X (Malvern) particle size analyzer. The mean droplet diameter was expressed as the Sauter diameter, d3,2 : d3,2 =
N D3 i i i2 i
Ni Di
(1)
where Ni is the number of droplets with a diameter, Di . Measurements were done in triplicate. 2.4. Physical stability Multiple light scattering measurements with a Turbiscan Expert LabMeasuring were used in order to study the physical stability of the emulsions. This instrument is capable to detect destabilization by creaming before it is visible by naked eye [20,21]. In order to quantify simultaneously the destabilization processes in our emulsions the Turbiscan Stability Index (TSI) was used. This index is a statistical factor and its value is calculated as the sum of all the destabilization processes in the measuring cell and it is given by [22]: TSI =
|scanref (hj ) − scani (hj )|
(2)
j
where scanref and scani are the initial backscattering value and the backscattering value at a given time, respectively, hj is a given height in the measuring cell and TSI is the sum of all the scan differences from the bottom to the top of the vial. 2.5. Design of experiments
2. Materials and methods 2.1. Materials d-Limonene (4-isopropenyl-1-methylciclohexane) was used as dispersed phase in emulsion formulation. d-Limonene 97% was purchased from Sigma-Aldrich® and purified with Florisil® resins (Fluka, 60–10 mesh) prior to use by following the procedure used elsewhere [18,19]. Namely, a mixture of oil and Florisil® in proportion 2:1 (w/w) was shaken gently for 3 h and then separated by decanting. The triblock copolymer Pluronic PE9400 (PEO21 –PPO50 –PEO21 , Mw = 4600 g mol−1 and HLB = 12–18) was kindly provided by BASF and used as received. Ultrapure water cleaned using a Milli-Q water purification system was used. All glassware was washed with 10% Micro-90 cleaning solution and exhaustively rinsed with tap water, isopropanol, deionized water, and ultrapure water in this sequence. Silicone Oil E-200 viscosity 200 cP was purchased by Dow Corning and also was used as received.
2.2. Emulsion preparation In the preliminaries studies emulsions containing 30 wt% dlimonene as dispersed phase and 1.8 wt% Pluronic PE9400 as emulsifier were prepared. These O/W emulsions were carried out using a rotor–stator homogenizer (Ultra Turrax T-25/KV11, IKA Instruments, Germany) at different process conditions. When focusing on formulation, homogenization rate was fixed at 17,500 rpm during 60 s in the studied emulsions. The final emulsion weight was 50 g in all cases.
In order to rationally develop an optimal formulation of d-limonene-in-water emulsions prepared with a rotor/stator homogenizer, a second-order experimental design and mathematical model was required. In the present work, the two process variables selected were: surfactant/oil ratio (R) and dispersed phase mass fraction (). With only two factors, a full factorial design 32 with three replications of the center point was selected. The quadratic model is given below: Y = ˇ0 + ˇ1 R + ˇ2 + ˇ11 R2 + ˇ22 2 + ˇ12 R
(3)
Each factor is measured at three levels which were coded to take the value −1 when the factor is at its low level, 0 when at its medium level and +1 when at its high level. The effect of the two independent variables on emulsion mean droplet size (d3,2 ) and on destabilization processes using the Turbiscan Stability Index (TSI) of d-limonene-in-water emulsion was studied. Experiments were randomly carried out in order to minimize the effects of random error in the observed responses. The center point in the design was done in triplicate to calculate the repeatability of the method [16] and to check the fitting quality of the mathematical model [23]. For model construction, terms with p > 0.05 were removed and the analysis was recalculated without this terms. The suitability of the models was determined by using coefficient of determination (R2 ) and the lack of fit test (Flof ). Joklegar and May [24] proposed that R2 should be higher than 0.80 to obtain a good fitting. Flof is the ratio between mean squares due to lack of fit and mean squares for pure error. This Flof value must be compared with a table value of Fcrit for ˛ level of significance and the degrees of freedom of the mean squares employed. It was assumed that the proposed model did not exhibit lack of fit at ˛ level of significance when Fcrit exceeds Flof .
L.M. Pérez-Mosqueda et al. / Colloids and Surfaces B: Biointerfaces 128 (2015) 127–131
129
3. Results and discussion 3.1. Processing conditions It has been reported that emulsion droplet size is a function of the processing conditions; homogenization rate and emulsification time [25] in addition to the formulation used. First the influence of the homogenization rate in the range 6500–24,000 rpm during 60 s was studied. As seen in Fig. 1, a minimum mean droplet size value was obtained at 17,500 rpm. The increase of droplet size above 17,500 rpm may be related to a recoalescence phenomenon induced by an excess of mechanical energy input during the emulsification process. Recoalescence phenomenon is due to the fact that emulsion droplets are subjected to excessive kinetic energy as a result of high-intensity turbulence in emulsification systems, which in turn yields the partial rupture of the interface of some droplets [14]. Second, the influence of the emulsification time in the range 30–180 s at 17,500 rpm was studied. The minimum Sauter diameter value was reached at 60 s (data not shown). Therefore, the optimal processing conditions were found to be 17,500 rpm and 60 s of homogenization rate and emulsification time. 3.2. Focusing on formulation The lowest value for the surfactant/oil ratio (R) was chosen such that the minimum size of stable droplets that can be theoretically produced was 1 m, according to the following equation [26]: dmin =
6sat Cs
(4)
where Cs is the surfactant concentration in the emulsion and sat is the excess surface concentration of the surfactant at saturation. A value of sat = 3 mg m−2 was used [10]. The highest limit was set as 5 times the lowest one, so R varied in the range 0.02–0.1. The dispersed phase fraction () was studied in the range 10–50 wt%, from dilute to slightly concentrated emulsion. Fig. 2 shows the Sauter mean diameter (d3,2 ) values obtained for all the experiments. For a given dispersed phase mass fraction, increasing R yielded a decrease of the droplet size until reaching a minimum value. This d3,2 variation was expected since more emulsifier was available to fully cover the O/W interface of smaller droplets as R is increased. However, once a certain ratio was achieved a further increase of surfactant did not result in smaller droplets. The limiting factor may be (a) the actual emulsification process, which was not capable to generate enough energy to further reduce the droplet size and/or (b) the fact that the emulsifier
Fig. 1. Sauter mean diameter as a function of homogenization rate for d-limonenein-water emulsion with R = 0.06 and = 30%. Symbols are the mean of three replicates and the error bars show the standard deviation of the measurements.
Fig. 2. Sauter mean diameter as a function of surfactant/oil ratio and dispersed mass fraction for emulsions at t = 0. Symbols are the mean of three replicates and the error bars show the standard deviation of the measurements.
did not adsorb fast enough to form a protective layer for preventing recoalescence phenomena [26]. On the other hand, the influence of on the droplet size was certainly more striking. The lack of major differences in d3,2 with was to be expected. However, a clear tendency to decrease d3,2 with increasing was observed. For a given surfactant/oil ratio (R) if the amount of oil (dispersed phase) is increased, the surfactant concentration in the continuous phase is also increased, enhancing the viscosity of the continuous phase, c . It has been reported that, the viscosity of the continuous phase, influences the droplet size diameter obtained in rotor/stator homogenizers. It is stated that an increase in c will lead to higher turbulent eddies which can be higher than the smaller droplets and therefore emulsification will proceed via a turbulent viscous regime producing smaller droplets [27]. In order to study the physical stability of the emulsions, laser diffraction and multiple light scattering measurements as a function of aging time were carried out. Fig. 3 shows the droplet size distribution for emulsions with R = 0.06 and = 10 wt% and 50 wt% for t = 0 and t = 12 h by way of example. The shift toward higher sizes of the droplet size distribution is usually associated with the occurrence of a destabilization process by coalescence and/or Ostwald ripening. On the one hand, if coalescence is the leading destabilization process a linear plot of 1/r2 vs. t [28] will be shown; on the other hand, a linear plot of r3 vs. t [29] will be obtained if Ostwald ripening is the main destabilization process. In all the systems studied a linear variation of r3 with time was observed confirming the occurrence of Ostwald ripening (data not shown). A backscattering
Fig. 3. Droplet size distribution for emulsions with R = 0.06 and = 10 wt% (triangles) and = 50 wt% (squares). t = 0 (solid lines and closed symbols) and t = 12 h (dashed lines and open symbols).
130
L.M. Pérez-Mosqueda et al. / Colloids and Surfaces B: Biointerfaces 128 (2015) 127–131
Fig. 4. Turbiscan Stability Index as a function of surfactant/oil ratio and dispersed mass fraction for emulsions at t = 0. Symbols are the mean of three replicates and the error bars show the standard deviation of the measurements.
decrease in the lower zone of the measuring cell was observed for all samples (data not shown). The drop in backscattering observed at the bottom of the measuring cell clearly indicated the occurrence of a destabilization mechanism by creaming, in that the dispersed phase possessed a lower density than the continuous phase. An overall analysis of the laser diffraction and multiple light scattering results allows us to conclude that all emulsions were affected by creaming and increase of Ostwald ripening. It has been shown that d-limonene emulsions are prone to break down due to creaming and/or Ostwald ripening [31]. Depending on two different behaviors were detected: for = 10 wt% and 30 wt% both destabilization processes, creaming and increase of the droplet size were readily noticeable. Nevertheless, for emulsion with = 50 wt% the creaming was not very perceptible whereas the increase in the droplet size was very marked. Eventually, this emulsion was also affected by creaming probably due to the increase of the droplet size. Taking into account that the emulsions were destabilized simultaneously by both processes, creaming and Ostwald ripening, the influence of R and in the global destabilization parameter, TSI, was studied. Fig. 4 shows the TSI values for all the emulsions. It should be noted that the most stable emulsions (lowest values of TSI) are the formulations with 50% . 3.3. Optimum formulation Response surface methodology is an affective statistical technique particularly appropriate for optimization of formulations or
processing [30]. Fig. 5 illustrates the three-dimension response surface curves and the model equations of d3,2 and TSI for the studied variables. These 3D plots and their respective contour plots provide a visual interpretation of the interaction between two factors. The variation of each response variable was initially assessed as a quadratic function of surfactant/oil ratio and dispersed phase mass fraction. The lack of fit shown by the fitting of TSI was solved adding a new cubic interaction term in these models: ˇ122 R * 2 . The results indicated that the models employed were adequate, showing no significant lack of fit (Fcrit > Flof with ˛ = 0.05) and very satisfactory values of R2 for all responses. The R2 values for droplet size and turbiscan stability index were found to be 0.959 and 0.991, respectively. From full factorial design we deduced that the most significant factors affecting droplet size and the main destabilization phenomenon in our system are the linear and quadratic terms of surfactant/oil ratio and the linear term of dispersed phase mass fraction. The equation for the droplet size of the fitted model was: d3,2 = 0.94 − 0.35 ∗ R − 0.12 ∗ + 0.25 ∗ R2
(5)
In addition, the model equation for the TSI behavior was: TSI = 5.8 − 4.6 ∗ + 4.0 ∗ R + 3.1 ∗ R2 + 2.5 ∗ R ∗ − 3.6 ∗ R ∗ 2 (6) Fig. 5 also shows the dependence of TSI as a function of both, R and . It is worth noting the existence of an area with TSI values close to zero, for R values in the range 0.03–0.05 and for the highest value (50%). An optimum process condition can be set for emulsion of submicron size with minimum destabilization. As noted above, the lowest TSI values are found for the emulsions containing 50 wt% of dispersed phase. Fig. 6 shows the normalized values of TSI and droplet size for the emulsions containing 50% of dispersed phase as function of surfactant/oil ratio. The intersection of both curves gave an optimal value with minimum destabilization and droplet size for R = 0.062. 3.4. Formulation improvement Ostwald ripening can be prevented by addition of a small amount of an oil that has very low solubility in the continuous phase [32]. A solute has been added to the dispersed phase of the optimal formulation in order to inhibit Oswald ripening [33]. Silicone oil has been used as a second component of the dispersed phase so that it possesses a large molecular size, is insoluble in water and increases the viscosity of the dispersed phase. Several papers
Fig. 5. Response surface 3D plots and model equations of (A) Sauter mean diameter and (B) Turbiscan Stability Index as a function of surfactant/oil ratio and dispersed mass fraction for emulsions at t = 0.
L.M. Pérez-Mosqueda et al. / Colloids and Surfaces B: Biointerfaces 128 (2015) 127–131
131
of the variables, surfactant/oil ratio and dispersed phase mass fraction a surface response methodology has been implemented.dLimonene-in-water emulsions exhibited destabilization by both, creaming and Ostwald ripening growth. The former was more marked for dilute emulsions, whereas the latter was faster at higher dispersed phase mass fraction values. In order to take into account the influence of both destabilization processes a global destabilization parameter has been used, turbiscan stability index (TSI). An optimal formulation, with minimum TSI and minimum droplet size values, was obtained for R = 0.062 and = 50 wt%. An improvement of the optimal formulation was achieved by the addition of silicone oil in the dispersed phase, which inhibited the droplet growth by Ostwald ripening.
Fig. 6. Normalized Sauter mean diameter (dashed line) and Turbiscan Stability Index (solid line) as a function of surfactant/oil ratio for emulsions with = 50 wt%. The intersection of both curves indicates the optimum formulation.
Fig. 7. Droplet size distributions for the emulsion without silicone oil and for the emulsion containing 2 wt% of silicone oil in the dispersed phase a function of aging time.
have studied the addition of silicone oils in the dispersed phase to prevent Ostwald ripening [34,35]. Fig. 7 shows the droplet size distributions for the emulsion without silicone oil and for the emulsion containing 2 wt% of silicone oil in the dispersed phase as a function of aging time. It should be noted that the Sauter diameter for the emulsion containing 2 wt% silicone oil is slightly lower than the emulsion without the additive (see the inset in Fig. 5). Furthermore, the stability of the emulsions was greatly enhanced since the droplet distributions remained constant after 260 h of aging time. This is probably caused by the fact that, on the one hand as the small droplets volume decreases, their silicone oil concentration increases because it cannot leave the droplet. On the other hand, as the large droplets grow their silicone oil concentration decreases. This sets up a chemical potential difference opposite to the Laplace effect inhibiting Ostwald ripening [32]. 4. Conclusions Monomodal, submicron d-limonene-in-water emulsions were obtained by rotor/stator homogenizer by using pluronic PE9400 as emulsifier. In order to gain a deeper knowledge of the droplet size distribution and stability of the emulsions as function
References [1] F. Kerton, Alternative Solvents for Green Chemistry, Royal Society of Chemistry, Cambridge, 2009. [2] T.F. Tadros, Colloids in Agrochemicals, Wiley, 2009. [3] R. Höfer, J. Bigorra, Green Chem. 9 (2007) 203–212. [4] W.S. Brud, Industrial uses of essential oils, in: K. Hüsnü Can Baser, G. Buchbauer (Eds.), Handbook of Essential Oils, CRC Press, 2010. ˜ [5] L.M. Pérez-Mosqueda, P. Ramírez, L.A. Trujillo-Cayado, J. Santos, J. Munoz, Colloids Surf. B: Biointerfaces 123 (2014) 797–802. [6] M. Klein, A. Aserin, I. Svitov, N. Garti, Colloids Surf. B: Biointerfaces 77 (2010) 75–81. [7] A. Soottitantawat, H. Yoshii, T. Furuta, M. Ohkawara, P. Linko, J. Food Sci. 68 (2003) 2256–2262. [8] A. Torcello-Gómez, M. Wulff-Pérez, M.J. Gálvez-Ruiz, A. Martín-Rodríguez, M. Cabrerizo-Vílchez, J. Maldonado-Valderrama, Adv. Colloid Interface Sci. 206 (2014) 414–427. [9] E.V. Batrakova, A.V. Kabanov, J. Control. Release 130 (2008) 98–106. [10] L.M. Pérez-Mosqueda, J. Maldonado-Valderrama, P. Ramírez, M.A. Cabrerizo˜ Colloids Surf. B: Biointerfaces 111 (2013) 171–178. Vílchez, J. Munoz, [11] S.S. Lim, M.Y. Baik, E.A. Decker, L. Henson, L. Michael Popplewell, D.J. McClements, S.J. Choi, Food Chem. 128 (2011) 1023–1028. [12] R. Chanamai, D.J. McClements, Colloids Surf. A: Physicochem. Eng. Aspects 172 (2000) 79–86. [13] J. Weiss, C. Canceliere, D.J. McClements, Langmuir 16 (2000) 6833–6838. ˜ [14] J. Santos, L.A. Trujillo-Cayado, N. Calero, J. Munoz, AIChE J. 60 (2014) 2644. [15] S.H. Musa, M. Basri, H.R.F. Masoumi, R.A. Karjiban, E.A. Malek, H. Basri, A.F. Shamsuddin, Colloids Surf. B: Biointerfaces 112 (2013) 113–119. [16] D.C. Montgomery, Design and Analysis of Experiments, Wiley, New York, 2001. [17] K. Urban, G. Wagner, D. Schaffner, D. Röglin, J. Ulrich, Chem. Eng. Technol. 29 (2006) 24–31. [18] J. Maldonado-Valderrama, V.B. Fainerman, M.J. Gálvez-Ruiz, A. MartínRodriguez, M.A. Cabrerizo-Vílchez, R. Miller, J. Phys. Chem. B 109 (2005) 17608–17616. [19] J. Maldonado-Valderrama, R. Miller, V.B. Fainerman, P.J. Wilde, V.J. Morris, Langmuir 26 (2010) 15901–15908. ˜ [20] J. Santos, L.A. Trujillo-Cayado, N. Calero, M.C. Alfaro, J. Munoz, Chem. Eng. Technol. 36 (2013) 1883–1890. ˜ [21] L.A. Trujillo-Cayado, P. Ramírez, M.C. Alfaro, M. Ruíz, J. Munoz, Colloids Surf. B: Biointerfaces 122 (2014) 623–629. [22] C. Lesaint, W.R. Glomm, L.E. Lundgaard, J. Sjöblom, Colloids Surf. A: Physicochem. Eng. Aspects 352 (2009) 63–69. [23] A.I. Khuri, J.A. Cornell, Response Surfaces: Designs and Analyses, Marcel Dekker Inc., New York, 1996. [24] A.M. Joklegar, A.T. May, Cereal Foods World 32 (1987) 857–868. [25] J.M. Franco, A. Raymundo, I. Sousa, C. Gallegos, J. Agric. Food Chem. 46 (1998) 3109–3115. [26] D.J. McClements, Food Emulsions: Principles, Practices and Techniques, CRC Press, Boca Raton, 2005. [27] N. Vankova, S. Tcholakova, N.D. Denkov, I.B. Ivanov, V. Vulchev, T. Danner, J. Colloid Interface Sci. 312 (2007) 363–380. [28] A. Kabalnov, Coalescence in emulsions, in: Binks (Ed.), Modern Aspects of Emulsions Science, Royal Society of Chemistry, Cambridge, 1998. [29] J.G. Weers, Ostwald ripening in emulsions, in: Binks (Ed.), Modern Aspects of Emulsions Science, Royal Society of Chemistry, Cambridge, 1998. [30] A. Raymundo, J.M. Franco, J. Empis, I. Sousa, J. Am. Oil Chem. Soc. 79 (2002) 783–790. [31] D.J. McClements, L. Henson, L.M. Popplewell, E.A. Decker, S. Jun Choi, J. Food Sci. 77 (2012) 33–38. [32] A. Imhof, D.J. Pine, J. Colloid Interface Sci. 192 (2) (1997) 368–374. [33] V. Schmitt, C. Cattelet, F. Leal-Calderon, Langmuir 20 (2004) 46–52. [34] S. Graves, K. Meleson, J. Wilking, M.Y. Lin, T.G. Mason, J. Chem. Phys. 122 (13) (2005) 134703. [35] J.M. Gutiérrez, C. González, A. Maestro, I. Sole, C.M. Pey, J. Nolla, Curr. Opin. Colloid Interface Sci. 13 (4) (2008) 245–251.
Contents lists available at ScienceDirect
Colloids and Surfaces B: Biointerfaces journal homepage: www.elsevier.com/locate/colsurfb
Formulation and optimization by experimental design of eco-friendly emulsions based on d-limonene Luis M. Pérez-Mosqueda, Luis A. Trujillo-Cayado, Francisco Carrillo, ˜ Pablo Ramírez ∗ , José Munoz Departamento de Ingeniería Química, Facultad de Química, Universidad de Sevilla, P. García González 1, 41012 Sevilla, Spain
a r t i c l e
i n f o
Article history: Received 11 December 2014 Received in revised form 28 January 2015 Accepted 13 February 2015 Available online 21 February 2015 Keywords: d-Limonene Emulsion Ostwald ripening Pluronic Response surface methodology
a b s t r a c t d-Limonene is a natural occurring solvent that can replace more pollutant chemicals in agrochemical formulations. In the present work, a comprehensive study of the influence of dispersed phase mass fraction, , and of the surfactant/oil ratio, R, on the emulsion stability and droplet size distribution of d-limonene-in-water emulsions stabilized by a non-ionic triblock copolymer surfactant has been carried out. An experimental full factorial design 32 was conducted in order to optimize the emulsion formulation. The independent variables, and R were studied in the range 10–50 wt% and 0.02–0.1, respectively. The emulsions studied were mainly destabilized by both creaming and Ostwald ripening. Therefore, initial droplet size and an overall destabilization parameter, the so-called turbiscan stability index, were used as dependent variables. The optimal formulation, comprising minimum droplet size and maximum stability was achieved at = 50 wt%; R = 0.062. Furthermore, the surface response methodology allowed us to obtain the formulation yielding submicron emulsions by using a single step rotor/stator homogenizer process instead of most commonly used two-step emulsification methods. In addition, the optimal formulation was further improved against Ostwald ripening by adding silicone oil to the dispersed phase. The combination of these experimental findings allowed us to gain a deeper insight into the stability of these emulsions, which can be applied to the rational development of new formulations with potential application in agrochemical formulations. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Many traditional industrial products are being gradually replaced by environmental friendly alternatives. Organic solvents are the main source of waste from academic and industrial installations as reaction medium, in separation procedures, and as diluters. The role of green solvents in the chemical and pharmaceutical industries is becoming more and more important [1]. This work deals with the development of emulsions containing a green solvent, which may find applications as matrices for incorporation of agrochemical active ingredients. The role of solvents in agrochemical industry is becoming increasingly more important [2]. More than 25% of all pesticides contain high concentrations of organic solvents, which represent a fire hazard, may be toxic and contribute to atmospheric volatile compound (VOC) emissions [3].
∗ Corresponding author. Tel.: +34 954557179. E-mail address: [email protected] (P. Ramírez). http://dx.doi.org/10.1016/j.colsurfb.2015.02.030 0927-7765/© 2015 Elsevier B.V. All rights reserved.
d-Limonene is a naturally occurring a cyclic monoterpene, which is commonly found in the rinds of citrus fruits such as grapefruit, lemon, lime, and in particular, oranges. Limonene exhibits good biodegradability; hence it may be proposed as an interesting alternative to organic solvents, meeting the ever-increasing safety and environmental demands of the 21st Century [4,5]. Furthermore, d-limonene also finds applications in many other fields such as food and pharmaceutical industries [6]. Given that d-limonene is not soluble in water, emulsification is an interesting alternative to solubilize and protect d-limonene from oxidative degradation [7]. In order to emulsify d-limonene, non-ionic polymeric surfactants, Pluronics, have been used. Their chemical structure is denoted by PEOx–PPOy–PEOx, where PEO and PPO stands for poly(ethylene) oxide and poly(propylene) oxide, respectively. They have been used as foam and emulsions stabilizers in industrial applications [8,9]. Recently, the interfacial properties of Pluronic PE9400 adsorbed at the limonene–water interface has been reported by [10], showing that PE9400 is irreversible adsorbed
128
L.M. Pérez-Mosqueda et al. / Colloids and Surfaces B: Biointerfaces 128 (2015) 127–131
at this interface, thereby being a potential candidate to prevent emulsion coalescence. Emulsions are thermodynamically unstable colloidal dispersions, which are destabilized by several mechanisms such as flocculation, coalescence, creaming and Ostwald ripening. It has been shown that orange oil suffers from creaming and Ostwald ripening [11]. It is known that both destabilization processes, creaming and Ostwald ripening, depend on dispersed phase mass fraction, , and surfactant concentration [12,13]. Therefore, a right selection of composition and physicochemical formulation is essential to improve the emulsification process [14]. Multivariate statistical techniques such as response surface methodology (RSM) have shown to be an interesting tool for the optimization of emulsions and nanoemulsions [15]. With this approach, a mathematical model connecting the response variables to the process ones is obtained with a minimum number of experiments. Hence, the development of new formulations becomes more effective [16]. The most common emulsification methods are based on mechanical energy input to the system by an external source. As a rule, emulsions are prepared in two steps; the aim of the first (primary homogenization) is to create droplets of dispersed phase such that a coarse emulsion is formed. The goal of the second step is to reduce the size of preexisting droplets, which usually involves the use of a different homogenizer, such as ultrasonic probes, highpressure valve homogenizers or microfluidizers [17]. Nevertheless, in the present study a single-step rotor/stator homogenizer was used in order to obtain monomodal submicron emulsions with dispersed phase mass fraction in the range from dilute to slightly concentrate emulsions ( = 10–50 wt%) and with a low ratio of surfactant concentration over solvent concentration (R < 0.1). An experimental full factorial design 32 was conducted in order to optimize the emulsion formulation.
2.3. Droplet size measurements Droplet size distribution (DSD) was determined by using laser diffraction measurements performed with a Mastersizer X (Malvern) particle size analyzer. The mean droplet diameter was expressed as the Sauter diameter, d3,2 : d3,2 =
N D3 i i i2 i
Ni Di
(1)
where Ni is the number of droplets with a diameter, Di . Measurements were done in triplicate. 2.4. Physical stability Multiple light scattering measurements with a Turbiscan Expert LabMeasuring were used in order to study the physical stability of the emulsions. This instrument is capable to detect destabilization by creaming before it is visible by naked eye [20,21]. In order to quantify simultaneously the destabilization processes in our emulsions the Turbiscan Stability Index (TSI) was used. This index is a statistical factor and its value is calculated as the sum of all the destabilization processes in the measuring cell and it is given by [22]: TSI =
|scanref (hj ) − scani (hj )|
(2)
j
where scanref and scani are the initial backscattering value and the backscattering value at a given time, respectively, hj is a given height in the measuring cell and TSI is the sum of all the scan differences from the bottom to the top of the vial. 2.5. Design of experiments
2. Materials and methods 2.1. Materials d-Limonene (4-isopropenyl-1-methylciclohexane) was used as dispersed phase in emulsion formulation. d-Limonene 97% was purchased from Sigma-Aldrich® and purified with Florisil® resins (Fluka, 60–10 mesh) prior to use by following the procedure used elsewhere [18,19]. Namely, a mixture of oil and Florisil® in proportion 2:1 (w/w) was shaken gently for 3 h and then separated by decanting. The triblock copolymer Pluronic PE9400 (PEO21 –PPO50 –PEO21 , Mw = 4600 g mol−1 and HLB = 12–18) was kindly provided by BASF and used as received. Ultrapure water cleaned using a Milli-Q water purification system was used. All glassware was washed with 10% Micro-90 cleaning solution and exhaustively rinsed with tap water, isopropanol, deionized water, and ultrapure water in this sequence. Silicone Oil E-200 viscosity 200 cP was purchased by Dow Corning and also was used as received.
2.2. Emulsion preparation In the preliminaries studies emulsions containing 30 wt% dlimonene as dispersed phase and 1.8 wt% Pluronic PE9400 as emulsifier were prepared. These O/W emulsions were carried out using a rotor–stator homogenizer (Ultra Turrax T-25/KV11, IKA Instruments, Germany) at different process conditions. When focusing on formulation, homogenization rate was fixed at 17,500 rpm during 60 s in the studied emulsions. The final emulsion weight was 50 g in all cases.
In order to rationally develop an optimal formulation of d-limonene-in-water emulsions prepared with a rotor/stator homogenizer, a second-order experimental design and mathematical model was required. In the present work, the two process variables selected were: surfactant/oil ratio (R) and dispersed phase mass fraction (). With only two factors, a full factorial design 32 with three replications of the center point was selected. The quadratic model is given below: Y = ˇ0 + ˇ1 R + ˇ2 + ˇ11 R2 + ˇ22 2 + ˇ12 R
(3)
Each factor is measured at three levels which were coded to take the value −1 when the factor is at its low level, 0 when at its medium level and +1 when at its high level. The effect of the two independent variables on emulsion mean droplet size (d3,2 ) and on destabilization processes using the Turbiscan Stability Index (TSI) of d-limonene-in-water emulsion was studied. Experiments were randomly carried out in order to minimize the effects of random error in the observed responses. The center point in the design was done in triplicate to calculate the repeatability of the method [16] and to check the fitting quality of the mathematical model [23]. For model construction, terms with p > 0.05 were removed and the analysis was recalculated without this terms. The suitability of the models was determined by using coefficient of determination (R2 ) and the lack of fit test (Flof ). Joklegar and May [24] proposed that R2 should be higher than 0.80 to obtain a good fitting. Flof is the ratio between mean squares due to lack of fit and mean squares for pure error. This Flof value must be compared with a table value of Fcrit for ˛ level of significance and the degrees of freedom of the mean squares employed. It was assumed that the proposed model did not exhibit lack of fit at ˛ level of significance when Fcrit exceeds Flof .
L.M. Pérez-Mosqueda et al. / Colloids and Surfaces B: Biointerfaces 128 (2015) 127–131
129
3. Results and discussion 3.1. Processing conditions It has been reported that emulsion droplet size is a function of the processing conditions; homogenization rate and emulsification time [25] in addition to the formulation used. First the influence of the homogenization rate in the range 6500–24,000 rpm during 60 s was studied. As seen in Fig. 1, a minimum mean droplet size value was obtained at 17,500 rpm. The increase of droplet size above 17,500 rpm may be related to a recoalescence phenomenon induced by an excess of mechanical energy input during the emulsification process. Recoalescence phenomenon is due to the fact that emulsion droplets are subjected to excessive kinetic energy as a result of high-intensity turbulence in emulsification systems, which in turn yields the partial rupture of the interface of some droplets [14]. Second, the influence of the emulsification time in the range 30–180 s at 17,500 rpm was studied. The minimum Sauter diameter value was reached at 60 s (data not shown). Therefore, the optimal processing conditions were found to be 17,500 rpm and 60 s of homogenization rate and emulsification time. 3.2. Focusing on formulation The lowest value for the surfactant/oil ratio (R) was chosen such that the minimum size of stable droplets that can be theoretically produced was 1 m, according to the following equation [26]: dmin =
6sat Cs
(4)
where Cs is the surfactant concentration in the emulsion and sat is the excess surface concentration of the surfactant at saturation. A value of sat = 3 mg m−2 was used [10]. The highest limit was set as 5 times the lowest one, so R varied in the range 0.02–0.1. The dispersed phase fraction () was studied in the range 10–50 wt%, from dilute to slightly concentrated emulsion. Fig. 2 shows the Sauter mean diameter (d3,2 ) values obtained for all the experiments. For a given dispersed phase mass fraction, increasing R yielded a decrease of the droplet size until reaching a minimum value. This d3,2 variation was expected since more emulsifier was available to fully cover the O/W interface of smaller droplets as R is increased. However, once a certain ratio was achieved a further increase of surfactant did not result in smaller droplets. The limiting factor may be (a) the actual emulsification process, which was not capable to generate enough energy to further reduce the droplet size and/or (b) the fact that the emulsifier
Fig. 1. Sauter mean diameter as a function of homogenization rate for d-limonenein-water emulsion with R = 0.06 and = 30%. Symbols are the mean of three replicates and the error bars show the standard deviation of the measurements.
Fig. 2. Sauter mean diameter as a function of surfactant/oil ratio and dispersed mass fraction for emulsions at t = 0. Symbols are the mean of three replicates and the error bars show the standard deviation of the measurements.
did not adsorb fast enough to form a protective layer for preventing recoalescence phenomena [26]. On the other hand, the influence of on the droplet size was certainly more striking. The lack of major differences in d3,2 with was to be expected. However, a clear tendency to decrease d3,2 with increasing was observed. For a given surfactant/oil ratio (R) if the amount of oil (dispersed phase) is increased, the surfactant concentration in the continuous phase is also increased, enhancing the viscosity of the continuous phase, c . It has been reported that, the viscosity of the continuous phase, influences the droplet size diameter obtained in rotor/stator homogenizers. It is stated that an increase in c will lead to higher turbulent eddies which can be higher than the smaller droplets and therefore emulsification will proceed via a turbulent viscous regime producing smaller droplets [27]. In order to study the physical stability of the emulsions, laser diffraction and multiple light scattering measurements as a function of aging time were carried out. Fig. 3 shows the droplet size distribution for emulsions with R = 0.06 and = 10 wt% and 50 wt% for t = 0 and t = 12 h by way of example. The shift toward higher sizes of the droplet size distribution is usually associated with the occurrence of a destabilization process by coalescence and/or Ostwald ripening. On the one hand, if coalescence is the leading destabilization process a linear plot of 1/r2 vs. t [28] will be shown; on the other hand, a linear plot of r3 vs. t [29] will be obtained if Ostwald ripening is the main destabilization process. In all the systems studied a linear variation of r3 with time was observed confirming the occurrence of Ostwald ripening (data not shown). A backscattering
Fig. 3. Droplet size distribution for emulsions with R = 0.06 and = 10 wt% (triangles) and = 50 wt% (squares). t = 0 (solid lines and closed symbols) and t = 12 h (dashed lines and open symbols).
130
L.M. Pérez-Mosqueda et al. / Colloids and Surfaces B: Biointerfaces 128 (2015) 127–131
Fig. 4. Turbiscan Stability Index as a function of surfactant/oil ratio and dispersed mass fraction for emulsions at t = 0. Symbols are the mean of three replicates and the error bars show the standard deviation of the measurements.
decrease in the lower zone of the measuring cell was observed for all samples (data not shown). The drop in backscattering observed at the bottom of the measuring cell clearly indicated the occurrence of a destabilization mechanism by creaming, in that the dispersed phase possessed a lower density than the continuous phase. An overall analysis of the laser diffraction and multiple light scattering results allows us to conclude that all emulsions were affected by creaming and increase of Ostwald ripening. It has been shown that d-limonene emulsions are prone to break down due to creaming and/or Ostwald ripening [31]. Depending on two different behaviors were detected: for = 10 wt% and 30 wt% both destabilization processes, creaming and increase of the droplet size were readily noticeable. Nevertheless, for emulsion with = 50 wt% the creaming was not very perceptible whereas the increase in the droplet size was very marked. Eventually, this emulsion was also affected by creaming probably due to the increase of the droplet size. Taking into account that the emulsions were destabilized simultaneously by both processes, creaming and Ostwald ripening, the influence of R and in the global destabilization parameter, TSI, was studied. Fig. 4 shows the TSI values for all the emulsions. It should be noted that the most stable emulsions (lowest values of TSI) are the formulations with 50% . 3.3. Optimum formulation Response surface methodology is an affective statistical technique particularly appropriate for optimization of formulations or
processing [30]. Fig. 5 illustrates the three-dimension response surface curves and the model equations of d3,2 and TSI for the studied variables. These 3D plots and their respective contour plots provide a visual interpretation of the interaction between two factors. The variation of each response variable was initially assessed as a quadratic function of surfactant/oil ratio and dispersed phase mass fraction. The lack of fit shown by the fitting of TSI was solved adding a new cubic interaction term in these models: ˇ122 R * 2 . The results indicated that the models employed were adequate, showing no significant lack of fit (Fcrit > Flof with ˛ = 0.05) and very satisfactory values of R2 for all responses. The R2 values for droplet size and turbiscan stability index were found to be 0.959 and 0.991, respectively. From full factorial design we deduced that the most significant factors affecting droplet size and the main destabilization phenomenon in our system are the linear and quadratic terms of surfactant/oil ratio and the linear term of dispersed phase mass fraction. The equation for the droplet size of the fitted model was: d3,2 = 0.94 − 0.35 ∗ R − 0.12 ∗ + 0.25 ∗ R2
(5)
In addition, the model equation for the TSI behavior was: TSI = 5.8 − 4.6 ∗ + 4.0 ∗ R + 3.1 ∗ R2 + 2.5 ∗ R ∗ − 3.6 ∗ R ∗ 2 (6) Fig. 5 also shows the dependence of TSI as a function of both, R and . It is worth noting the existence of an area with TSI values close to zero, for R values in the range 0.03–0.05 and for the highest value (50%). An optimum process condition can be set for emulsion of submicron size with minimum destabilization. As noted above, the lowest TSI values are found for the emulsions containing 50 wt% of dispersed phase. Fig. 6 shows the normalized values of TSI and droplet size for the emulsions containing 50% of dispersed phase as function of surfactant/oil ratio. The intersection of both curves gave an optimal value with minimum destabilization and droplet size for R = 0.062. 3.4. Formulation improvement Ostwald ripening can be prevented by addition of a small amount of an oil that has very low solubility in the continuous phase [32]. A solute has been added to the dispersed phase of the optimal formulation in order to inhibit Oswald ripening [33]. Silicone oil has been used as a second component of the dispersed phase so that it possesses a large molecular size, is insoluble in water and increases the viscosity of the dispersed phase. Several papers
Fig. 5. Response surface 3D plots and model equations of (A) Sauter mean diameter and (B) Turbiscan Stability Index as a function of surfactant/oil ratio and dispersed mass fraction for emulsions at t = 0.
L.M. Pérez-Mosqueda et al. / Colloids and Surfaces B: Biointerfaces 128 (2015) 127–131
131
of the variables, surfactant/oil ratio and dispersed phase mass fraction a surface response methodology has been implemented.dLimonene-in-water emulsions exhibited destabilization by both, creaming and Ostwald ripening growth. The former was more marked for dilute emulsions, whereas the latter was faster at higher dispersed phase mass fraction values. In order to take into account the influence of both destabilization processes a global destabilization parameter has been used, turbiscan stability index (TSI). An optimal formulation, with minimum TSI and minimum droplet size values, was obtained for R = 0.062 and = 50 wt%. An improvement of the optimal formulation was achieved by the addition of silicone oil in the dispersed phase, which inhibited the droplet growth by Ostwald ripening.
Fig. 6. Normalized Sauter mean diameter (dashed line) and Turbiscan Stability Index (solid line) as a function of surfactant/oil ratio for emulsions with = 50 wt%. The intersection of both curves indicates the optimum formulation.
Fig. 7. Droplet size distributions for the emulsion without silicone oil and for the emulsion containing 2 wt% of silicone oil in the dispersed phase a function of aging time.
have studied the addition of silicone oils in the dispersed phase to prevent Ostwald ripening [34,35]. Fig. 7 shows the droplet size distributions for the emulsion without silicone oil and for the emulsion containing 2 wt% of silicone oil in the dispersed phase as a function of aging time. It should be noted that the Sauter diameter for the emulsion containing 2 wt% silicone oil is slightly lower than the emulsion without the additive (see the inset in Fig. 5). Furthermore, the stability of the emulsions was greatly enhanced since the droplet distributions remained constant after 260 h of aging time. This is probably caused by the fact that, on the one hand as the small droplets volume decreases, their silicone oil concentration increases because it cannot leave the droplet. On the other hand, as the large droplets grow their silicone oil concentration decreases. This sets up a chemical potential difference opposite to the Laplace effect inhibiting Ostwald ripening [32]. 4. Conclusions Monomodal, submicron d-limonene-in-water emulsions were obtained by rotor/stator homogenizer by using pluronic PE9400 as emulsifier. In order to gain a deeper knowledge of the droplet size distribution and stability of the emulsions as function
References [1] F. Kerton, Alternative Solvents for Green Chemistry, Royal Society of Chemistry, Cambridge, 2009. [2] T.F. Tadros, Colloids in Agrochemicals, Wiley, 2009. [3] R. Höfer, J. Bigorra, Green Chem. 9 (2007) 203–212. [4] W.S. Brud, Industrial uses of essential oils, in: K. Hüsnü Can Baser, G. Buchbauer (Eds.), Handbook of Essential Oils, CRC Press, 2010. ˜ [5] L.M. Pérez-Mosqueda, P. Ramírez, L.A. Trujillo-Cayado, J. Santos, J. Munoz, Colloids Surf. B: Biointerfaces 123 (2014) 797–802. [6] M. Klein, A. Aserin, I. Svitov, N. Garti, Colloids Surf. B: Biointerfaces 77 (2010) 75–81. [7] A. Soottitantawat, H. Yoshii, T. Furuta, M. Ohkawara, P. Linko, J. Food Sci. 68 (2003) 2256–2262. [8] A. Torcello-Gómez, M. Wulff-Pérez, M.J. Gálvez-Ruiz, A. Martín-Rodríguez, M. Cabrerizo-Vílchez, J. Maldonado-Valderrama, Adv. Colloid Interface Sci. 206 (2014) 414–427. [9] E.V. Batrakova, A.V. Kabanov, J. Control. Release 130 (2008) 98–106. [10] L.M. Pérez-Mosqueda, J. Maldonado-Valderrama, P. Ramírez, M.A. Cabrerizo˜ Colloids Surf. B: Biointerfaces 111 (2013) 171–178. Vílchez, J. Munoz, [11] S.S. Lim, M.Y. Baik, E.A. Decker, L. Henson, L. Michael Popplewell, D.J. McClements, S.J. Choi, Food Chem. 128 (2011) 1023–1028. [12] R. Chanamai, D.J. McClements, Colloids Surf. A: Physicochem. Eng. Aspects 172 (2000) 79–86. [13] J. Weiss, C. Canceliere, D.J. McClements, Langmuir 16 (2000) 6833–6838. ˜ [14] J. Santos, L.A. Trujillo-Cayado, N. Calero, J. Munoz, AIChE J. 60 (2014) 2644. [15] S.H. Musa, M. Basri, H.R.F. Masoumi, R.A. Karjiban, E.A. Malek, H. Basri, A.F. Shamsuddin, Colloids Surf. B: Biointerfaces 112 (2013) 113–119. [16] D.C. Montgomery, Design and Analysis of Experiments, Wiley, New York, 2001. [17] K. Urban, G. Wagner, D. Schaffner, D. Röglin, J. Ulrich, Chem. Eng. Technol. 29 (2006) 24–31. [18] J. Maldonado-Valderrama, V.B. Fainerman, M.J. Gálvez-Ruiz, A. MartínRodriguez, M.A. Cabrerizo-Vílchez, R. Miller, J. Phys. Chem. B 109 (2005) 17608–17616. [19] J. Maldonado-Valderrama, R. Miller, V.B. Fainerman, P.J. Wilde, V.J. Morris, Langmuir 26 (2010) 15901–15908. ˜ [20] J. Santos, L.A. Trujillo-Cayado, N. Calero, M.C. Alfaro, J. Munoz, Chem. Eng. Technol. 36 (2013) 1883–1890. ˜ [21] L.A. Trujillo-Cayado, P. Ramírez, M.C. Alfaro, M. Ruíz, J. Munoz, Colloids Surf. B: Biointerfaces 122 (2014) 623–629. [22] C. Lesaint, W.R. Glomm, L.E. Lundgaard, J. Sjöblom, Colloids Surf. A: Physicochem. Eng. Aspects 352 (2009) 63–69. [23] A.I. Khuri, J.A. Cornell, Response Surfaces: Designs and Analyses, Marcel Dekker Inc., New York, 1996. [24] A.M. Joklegar, A.T. May, Cereal Foods World 32 (1987) 857–868. [25] J.M. Franco, A. Raymundo, I. Sousa, C. Gallegos, J. Agric. Food Chem. 46 (1998) 3109–3115. [26] D.J. McClements, Food Emulsions: Principles, Practices and Techniques, CRC Press, Boca Raton, 2005. [27] N. Vankova, S. Tcholakova, N.D. Denkov, I.B. Ivanov, V. Vulchev, T. Danner, J. Colloid Interface Sci. 312 (2007) 363–380. [28] A. Kabalnov, Coalescence in emulsions, in: Binks (Ed.), Modern Aspects of Emulsions Science, Royal Society of Chemistry, Cambridge, 1998. [29] J.G. Weers, Ostwald ripening in emulsions, in: Binks (Ed.), Modern Aspects of Emulsions Science, Royal Society of Chemistry, Cambridge, 1998. [30] A. Raymundo, J.M. Franco, J. Empis, I. Sousa, J. Am. Oil Chem. Soc. 79 (2002) 783–790. [31] D.J. McClements, L. Henson, L.M. Popplewell, E.A. Decker, S. Jun Choi, J. Food Sci. 77 (2012) 33–38. [32] A. Imhof, D.J. Pine, J. Colloid Interface Sci. 192 (2) (1997) 368–374. [33] V. Schmitt, C. Cattelet, F. Leal-Calderon, Langmuir 20 (2004) 46–52. [34] S. Graves, K. Meleson, J. Wilking, M.Y. Lin, T.G. Mason, J. Chem. Phys. 122 (13) (2005) 134703. [35] J.M. Gutiérrez, C. González, A. Maestro, I. Sole, C.M. Pey, J. Nolla, Curr. Opin. Colloid Interface Sci. 13 (4) (2008) 245–251.
