6956
Langmuir 2004, 20, 6956-6963
Amphiphilic Polysaccharides: Useful Tools for the Preparation of Nanoparticles with Controlled Surface Characteristics A. Durand,* E. Marie, E. Rotureau, M. Leonard, and E. Dellacherie Laboratoire de Chimie Physique Macromole´ culaire, UMR CNRS-INPL 7568, Groupe ENSIC, BP 451, 54001 Nancy Cedex, France Received April 16, 2004. In Final Form: June 1, 2004 Polymeric surfactants obtained by hydrophobic modification of dextran are used as stabilizers for oilin-water emulsions. The kinetics of interfacial tension decrease is studied as a function of polymer structural characteristics (degree of hydrophobic substitution) and at various polymer concentrations. Several hydrocarbon oils, either aliphatic (octane, decane, dodecane, and hexadecane) or aromatic (styrene), are tested. Kinetics exhibits the same general trends no matter which oil or polymer is considered. The emulsifying properties of the polymeric surfactants are illustrated by the preparation of oil-in-water emulsions. The droplet size at the preparation is correlated to the amount of oil and to the polymer concentration in the aqueous phase. For low polymer/oil ratios, it is shown that the droplet size is limited by the initial amount of polymer. On the contrary, for high polymer/oil ratios, the droplet size seems to level down, indicating that other parameters become predominant. Emulsion aging occurs by Ostwald ripening, and it is demonstrated that the theoretical equation of Lifshitz, Slyozov, and Wagner (LSW) correctly describes the experimental results. The nature of the oil has important effects on emulsion aging, as described by the LSW equation. The aging of emulsions containing oil mixtures is quantitatively described on the basis of the results with pure oils. The influence of polymer chemical structure can be conveniently correlated to interfacial tension results through the LSW equation. On the contrary, the influence of oil volume fraction seems to be overestimated by the usual correction factor, k(φ). The effect of temperature on emulsion aging is finally examined. Miniemulsions stabilized with dextran derivatives are used for the radical polymerization of styrene. Following this procedure, polysaccharide-covered polystyrene nanoparticles are prepared and characterized (size and surface coverage). The size of the particles is directly correlated to that of the initial droplets for styrene volume fractions around 10%. On the contrary, for initial styrene volume fractions around 20%, particles exhibit a larger size than the initial droplets, indicating that coalescence processes take place during polymerization. The amount of dextran at the surface of the particles is determined and compared to the adsorbed amounts resulting from emulsion preparation.
Introduction The potential use of polymeric nanoparticles as drug carriers has led to the development of many different drug delivery vehicles. The surface characteristics of such nanoparticles are of fundamental importance for their interactions with the living medium. Therefore, polysaccharide-coated nanoparticles and nanodroplets are very attractive candidates for biomedical applications, since polysaccharides are natural polymers exhibiting biocompatibility. Our laboratory has been developing the hydrophobic modification of polysaccharides for ∼15 years. The amphiphilic polymers obtained have been used for the preparation of polysaccharide-covered nanoparticles. Several processes were followed such as direct adsorption of amphiphilic polysaccharides onto preformed polymeric nanoparticles and emulsion preparation followed by solvent evaporation using amphiphilic polysaccharides as polymeric emulsifiers.1-4 The surfactant properties of amphiphilic derivatives of dextran (a neutral and slightly branched bacterial polysac* To whom correspondence should be addressed. Phone: 33 (0)3 83 17 52 92. Fax: 33 (0)3 83 37 99 77. E-mail: alain.durand@ ensic.inpl-nancy.fr. (1) Fournier, C.; Le´onard, M.; Le Coq-Le´onard, I.; Dellacherie, E. Langmuir 1995, 11, 2344. (2) Fournier, C.; Le´onard, M.; Dellacherie, E.; Chikhi, M.; Hommel, H.; Legrand, A. P. J. Colloid Interface Sci. 1998, 198, 27. (3) Rouzes, C.; Gref, R.; Le´onard, M.; De Sousa-Delgado, A.; Dellacherie, E. J. Biomed. Mater. Res. 2000, 50, 557. (4) Rouzes, C.; Le´onard, M.; Durand, A.; Dellacherie, E. Colloids Surf., B 2003, 32, 125.
charide consisting of R-1,6-glucopyranose units) have been particularly studied in our group. Thanks to an optimized modification process, a series of dextran derivatives with varying molecular characteristics were prepared with various types and amounts of hydrophobic groups (aromatic rings or aliphatic hydrocarbon chains) and the introduction of controlled amounts of anionic groups. The surface active properties of such polymers were studied and related to their structural parameters.5,6 The use of amphiphilic dextran derivatives as emulsifiers was also examined. The initial droplet size (well below 1 µm) as well as the stability of the dodecane-in-water emulsions prepared were correlated to the polymer characteristics. In particular, it was demonstrated that Ostwald ripening was the major aging process.7-9 This paper describes a new process for the preparation of polysaccharide-covered nanoparticles involving the miniemulsion radical polymerization of styrene stabilized by amphiphilic derivatives of dextran. Aqueous miniemulsions are obtained via intense shearing of a mixture of monomer, water, a stabilizer, and a highly water (5) Rouzes, C.; Durand, A.; Le´onard, M.; Dellacherie, E. J. Colloid Interface Sci. 2002, 253, 217. (6) Rotureau, E.; Leonard, M.; Dellacherie, E.; Durand, A. Submitted for publication, 2004. (7) Rotureau, E.; Leonard, M.; Dellacherie, E.; Durand, A. Phys. Chem. Chem. Phys. 2004, 6, 1430. (8) Sadtler, V.; Imbert, P.; Dellacherie, E. J. Colloid Interface Sci. 2002, 254, 355. (9) Imbert, P.; Sadtler, V.; Dellacherie, E. Colloids Surf., A 2002, 211, 157.
10.1021/la0490341 CCC: $27.50 © 2004 American Chemical Society Published on Web 07/10/2004
Amphiphilic Polysaccharides
insoluble compound, the so-called hydrophobe. This compound suppresses the mass exchange between the oil droplets and the Ostwald ripening, and the stabilizer avoids droplet coalescence.10-12 Polymerization of these stable nanodroplets leads to particles which ideally keep their size.13 For the formulation of miniemulsions, a wide range of ionic14-16 and nonionic17-20 surfactants have been used, leading to differently charged and stabilized latexes. In this paper, we intend to provide a complete description of the whole process, from the emulsifying step to the polymerization ones. The adsorption of hydrophobic dextran derivatives at the oil/water interface is first studied, especially its kinetics which is indeed known to be very different from that of molecular surfactants. The emulsifying properties of the polymeric surfactants are studied through the preparation of oil-in-water emulsions. The initial droplet size as well as the amount of adsorbed polymer are analyzed. Emulsion aging is then described with a particular attempt to establish quantitative correlations to predict the size variation with time as a function of the polymer properties, nature of the oil, oil volume fraction, and aging temperature. Then, the polymerization of styrene is carried out using a standard miniemulsion procedure21 with hydrophobic derivatives of dextran as a stabilizer, producing monodisperse polysaccharide-covered polystyrene nanoparticles. The characteristics of the latex particles produced (size and amount of covering of polysaccharide) are determined and related to the polymerization conditions as well as to the emulsion stability. Experimental Section Materials. The amphiphilic derivatives of dextran were prepared from dextran T40 obtained from Pharmacia (Uppsala, Sweden). The other chemicals were purchased from Aldrich (St Quentin Fallavier, France). Styrene was distilled under reduced pressure prior to utilization. The other chemicals were used as received. MilliQ water was used for all the experiments. Polymer Synthesis. The amphiphilic derivatives of dextran (Scheme 1) were prepared by chemical modification of a commercial dextran denoted Dex. This dextran has been characterh w ) 40 000 g/mol, ized by SEC-MALLS: M h n ) 26 000 g/mol, M and Ip ) 1.6. Hydrophobic dextran derivatives were prepared by reacting dextran with phenyl glycidyl ether in an aqueous solution of sodium hydroxide as previously described.1,5 The degree of phenoxy substitution, τ (%), is defined by τ ) 100[y/(x + y)] (see Scheme 1). The polymer is then named DexPτ. The value of τ was determined by 1H NMR in deuterated dimethyl sulfoxide. The dextran derivatives used in that work possess degrees of substitution between 5 and 25%. Emulsion Preparation. Oil-in-water emulsions were prepared by sonication (pulsed mode, 10 W, two sonication steps of 1 min separated by a rest of 2 min) using a Vibracell model 600W instrument (Sonics & Materials Inc., Danbury, CT). Octane, (10) Bechthold, N.; Tiarks, F.; Willert, M.; Landfester, K.; Antonietti, M. Macromol. Symp. 2000, 151, 549. (11) Landfester, K. Macromol. Rapid Commun. 2001, 22, 896. (12) Schork, F. J.; Poehlein, G. W.; Wang, S.; Reimers, J.; Rodrigues, J.; Samer, C. Colloids Surf., A 1999, 153, 39. (13) Landfester, K.; Bechtold, N.; Fo¨rster, S.; Antonietti, M. Macromol. Rapid Commun. 1999, 20, 81. (14) Paunov, V. N.; Sandler, S. I.; Kaler, E. W. Langmuir 2001, 17, 4126. (15) Landfester, K.; Bechtold, N.; Tiarks, F.; Antonietti, M. Macromolecules 1999, 32, 2679. (16) Marie, E.; Landfester, K.; Antonietti, M. Biomacromolecules 2002, 3, 475. (17) Chern, C. S.; Liou, Y. C. Polymer 1999, 40, 373. (18) Kim, N.; Sudol, E. D.; Dimonie, V. L.; El-Aasser, M. S. Macromolecules 2003, 36, 5573. (19) Landfester, K. Surfactant Sci. Ser. 2003, 115, 225. (20) Lim, M. S.; Chen, H. J. Polym. Sci., Part A: Polym. Chem. 2000, 38, 1818. (21) Antonietti, M.; Landfester, K. ChemPhysChem 2001, 2, 207.
Langmuir, Vol. 20, No. 16, 2004 6957 Scheme 1
a
a Note that this scheme does not imply that the same hydroxyl group is substituted within all the glucopyranose units. A detailed study of the position modified by vinyl glycidyl ether has not been performed yet.
decane, dodecane, and hexadecane were used without further purification. The volumes of the oil and the aqueous phase were kept equal to 1 and 10 mL, respectively, for all the emulsions prepared. The polymer was previously dissolved in the aqueous phase during a period of 20 h. Polymerization. The amphiphilic derivatives of dextrane (DexP5, DexP13, and DexP18) were dissolved in MilliQ water. The organic phase is composed of styrene, hexadecane (5 vol % related to styrene), and AIBN (1.5 wt % related to styrene). The volume fraction of the organic phase was varied from 5 to 20 vol %. After stirring for 1 h, emulsification was achieved via sonication (pulsed mode, 10 W, 3 min) using a Vibracell model 600W instrument (Sonics & Materials Inc., Danbury, CT). To avoid polymerization due to heating, the mixture was ice-cooled during sonication. Polymerization was then performed at 75 °C for 24 h. Dynamic Surface/Interfacial Tension. We used a dynamic tensiometer from Interfacial Technology Concept (Longessaigne, France). For the measurement of dynamic surface tension, an air bubble was formed in the aqueous polymer solution. The shape of the bubble was followed by a CCD camera, and the surface tension was deduced from the mathematical analysis of this shape. The polymer solution was maintained at 25 ( 0.5 °C by a circulating water bath. The volume of the bubble was 5 mm3 for all measurements. This volume was kept constant by an automatic regulation. The time t ) 0 was taken immediately after the formation of the bubble. It has been demonstrated that the volume of the bubble can have a significant influence on the value of the surface tension: the higher the volume, the more accurate the calculation is.22 We obtained the following values for pure water at 25 °C: 71.5 mN/m for 12 mm3 and 70 mN/m for 5 mm3. These values are similar to those given in the detailed study of Lin et al. who demonstrated that the volume effect was especially important below 5 mm3.22 Size Measurement of Emulsions and Latex Particles. Droplet and particle sizes were measured by dynamic light scattering at low concentration using an HPPS-ET instrument from Malvern. Although this apparatus is able to measure relatively concentrated samples, the emulsions and latexes were diluted in pure water. Indeed, dilution with oil, saturated water, or the polymer solution gave the same results. Furthermore, the presence of additional small droplets of oil and of aggregates of polymers increases the uncertainty of the measurements. Viscometry. Viscometric measurements with aqueous DexP solutions were carried out using an Ostwald-type capillary viscometer (0.46 mm diameter). The temperature was regulated by a circulating bath. Prior to measurements, the aqueous solutions were filtered through 0.2 mm filters. Polymer concentration was checked by weighing the dry extracts obtained after drying the solutions for 24 h in an oven at 110 °C. The found values were always ∼90% of the calculated ones. No kinetic corrections were required, since we verified that the flow time was proportional to the cinematic viscosity.
Results and Discussion Kinetics of Interfacial Tension Decrease by Dextran Derivatives. Since this point has been detailed (22) Lin, S.-Y.; Wang, W.-J.; Lin, L.-W.; Chen, L.-J. Colloids Surf., A 1996, 114, 31.
6958
Langmuir, Vol. 20, No. 16, 2004
Durand et al.
Figure 1. Dodecane/water interfacial tension as a function of time with DexPτ polymers in the aqueous phase at 25 °C: DexP7, 10 g/L (O); DexP16, 10 g/L (0); DexP23, 1 g/L (b), 10 g/L ([), and 20 g/L (2). The lines are curve fittings made by using eq 1. Inset: characteristic time, t*, of the various curves as a function of CPh-0.5.
Figure 2. Initial droplet size as a function of the mass ratio [DexP23]/[oil] for dodecane-in-water emulsions varying either the DexP23 concentration with φ ) 9.1% (b) or the oil volume fraction with C ) 2 g/L (O). The line represents the result of a mass balance assuming a surface coverage equal to 1 mg/m2.7
Table 1. Characteristic Parameters Obtained from the Kinetics of Dodecane/Water Interfacial Tension Decrease (For Details, See the Text) at T ) 25 °C
time, γie. Varying the degree of substitution and varying the polymer concentration in the bulk are not equivalent for γie. From these results, we can extract general trends of the kinetics of interfacial tension decrease. First, it appears to be mainly controlled by the concentration of hydrophobic units in the bulk. As a result, we may infer that it is a diffusion-limited phenomenon. The limiting step is probably the diffusion in the layer of already adsorbed macromolecules rather than the diffusion in the bulk. The latter was shown to occur only for polymer concentrations lower than 0.01 g/L.6 As for the interfacial tension at infinite time, it is mainly influenced by the ability of the polymer to adsorb at the interface. This fact is closely related to the degree of substitution, as reported previously.5 Consequently, the concentration of hydrophobic tails in the bulk is no longer a relevant parameter. On the contrary, increasing the degree of substitution increasing the number of anchoring groups for each macromolecule at the oil/water interface. This leads to a lower equilibrium value of the interfacial tension. Emulsifying Properties. Emulsion Preparation. Oilin-water emulsions were prepared by sonication in the presence of dextran derivatives previously dissolved in the aqueous phase. For a given polymer, the polymer concentration in the aqueous phase, C (g/L), as well as the oil volume fraction, φ, were varied. It appears that plotting the initial droplet diameter as a function of the initial weight ratio R ) [polymer]/[oil] allows one to gather all the experiments onto a single curve (Figure 2). Up to R ) 0.05, the variation of the droplet diameter with R is close to that predicted by a mass balance25 assuming a constant surface coverage equal to 1 mg/m2. In other words, the droplet diameter is controlled by the amount of polymer available for covering the oil/water interface. For the emulsions prepared in conditions such that R > 0.05, the droplet diameter remains around 180 nm for dodecane and 160 nm for styrene. As a result, the amount of polymer is no longer the limiting parameter and the droplet diameter is fixed by the mechanical energy input, oil physical properties, rate of polymer adsorption onto the newly created oil/water interface, polymer chains stiffness, etc. Such leveling off of the droplet diameter has already been reported but for miniemulsions prepared by a
polymer
C (g/L)
t* a (s)
γieb (mN/m)
DexP7 DexP16 DexP23
10 10 1 10 20
20.6 11.9 28.9 9.5 7.3
20.0 7.1 7.5 6.9 6.2
a
Characteristic time, t*. b Interfacial tension at infinite time,
γ
ie.
elsewhere,6 we will restrict its discussion here to the polymer concentration range (1-20 g/L) used in the part devoted to emulsion preparation. The variation of dodecane/water interfacial tension in the presence of DexP23 in the aqueous phase was examined at various concentrations (Figure 1). In another series of experiments, several DexPτ polymers with degrees of substitution varying between 7 and 23% were used at a given concentration of 10 g/L (Figure 1). In the explored concentration range, the variation of the interfacial tension can be conveniently depicted (Figure 1) by a semiempirical equation similar to that proposed by Filippov:23,24
γi0 - γi(t) γi(t) - γie
)
xt*t
(1)
In eq 1, γi0, γi, and γie are the oil/water interfacial tensions at t ) 0, at t, and at infinite time, respectively, and t* is a semiempirical parameter having the units of time and corresponding to the time at which γi ) (γi0 + γie)/2. This parameter allows a quantitative comparison of the different experiments. It appears that eq 1 fits the experimental curves rather satisfactorily for t > 2000 s in the concentration range considered. The values obtained for γie and t* are given in Table 1. When the characteristic time, t*, is plotted as a function of the concentration of hydrophobic tails in the bulk, CPh (mol of phenoxy groups/ L), a single curve is obtained by either varying the polymer concentration or varying the degree of substitution of the polymer. Indeed, t* is found to be inversely proportional to xCPh (inset of Figure 1). Such a unified curve is not obtained when plotting the interfacial tension at infinite (23) Filippov, L. K. J. Colloid Interface Sci. 1994, 163, 49. (24) Filippov, L. K. J. Colloid Interface Sci. 1994, 164, 471.
(25) Canselier, J. P.; Delmas, H.; Wilhelm, A. M.; Abismaı¨l, B. J. Dispersion Sci. Technol. 2002, 23, 333.
Amphiphilic Polysaccharides
Langmuir, Vol. 20, No. 16, 2004 6959 Table 2. Rate of Ostwald Ripening Obtained with Dodecane-in-Water Emulsions Stabilized with Various Dextran Derivativesa polymeric surfactant DexP7 DexP11 DexP23
Cb (g/L)
ωexp (m3/s)
ωcalcc (m3/s)
ωexp/ωcalc
10 10 5 10 20
3.1 × 10-26 2.3 × 10-26 1.6 × 10-26 1.5 × 10-26 1.3 × 10-26
8.8 × 10-27 4.3 × 10-27 3.0 × 10-27 3.0 × 10-27 2.7 × 10-27
3.5 5.3 5.6 5.0 4.8
a The dodecane volume fraction was 9.1% in all cases. b Initial polymer concentration in the aqueous phase before emulsion preparation. c Values calculated from eq 3 and the measured interfacial tensions. For dodecane, the values given in ref 49 were used, i.e., D ) 5.4 × 10-10 m2/s and C∞ ) 2.4 × 10-5 mol/m3.
Figure 3. Surface coverage of oil droplets by DexP23 immediately after emulsion preparation as a function of the polymer concentration remaining in the aqueous phase. Dodecane-in-water emulsions stabilized by DexP23, φ ) 9.1%. The line is a curve fitting made by using the Scatchard equation.29
microfluidizer.26 In the case of miniemulsions obtained by sonication, no leveling off was reported in the case of molecular surfactants.27 We could formulate two assumptions: either the polymer chain stiffness prevents the adsorption onto small droplets because of surface curvature or polymer diffusion onto newly created droplets is too slow to prevent their growth by collisions up to a certain diameter. Nevertheless, theoretical data are still lacking to propose a deeper interpretation. The amount of nonadsorbed polymer remaining in the aqueous phase immediately after sonication was determined using the anthrone method.28 The data confirm that, with conditions such that R < 0.05, the surface coverage is of the order of 1 mg/m2 and almost all the polymer initially added is adsorbed. A minimum surface coverage is required to prevent coalescence. As a result, all droplets with a lower surface coverage are not stable enough and they coalesce. Consequently, in the R domain where the amount of polymer is limiting, the droplets formed have a similar surface coverage around 1 mg/m2. The polymer adsorption is controlled by the droplet stability. In the domain 0.05 < R < 0.25, the droplet diameter remains between 210 and 190 nm, while the surface coverage increases. With those conditions, more polymer is adsorbed when R increases but it has a far more limited effect on the droplet diameter than that in the domain where R < 0.05. When plotting the droplet surface coverage, Γ (mg/m2), as a function of the concentration of nonadsorbed polymer remaining in the aqueous phase, Cexcess (g/L), for the experimental points corresponding to R > 0.05, we get a curve having the shape of an adsorption isotherm (Figure 3). Furthermore, the curve can be linearized using the Scatchard method.29,30 In that domain, the polymer adsorption equilibrium controls the amount of polymer adsorbed onto the droplets. With the Scatchard linearization method, we get a maximum surface coverage equal to 4.3 mg/m2, which agrees with the experimental results (Figure 3). Even if our picture of emulsion preparation is certainly oversimplified, it is helpful as a first approach. Several polymers were used for emulsion preparation, varying the degree of substitution between 7 and 23%. (26) Chern, C. S.; Chen, T. J. Colloids Surf., A 1998, 138, 65. (27) Landfester, K. Prog. Colloid Polym. Sci. 2001, 117, 101. (28) Scott, T. A.; Melvin, E. H. Anal. Chem. 1953, 25, 1656. (29) Scatchard, G. Ann. N.Y. Acad. Sci. 1949, 51, 660. (30) Klotz, I. M.; Hunston, D. Biochemistry 1971, 10, 3065.
Nevertheless, no significant influence on the droplet diameter could be evidenced. It is probable that, in our sonication conditions, the energy input is so important that all the polymers are able to stabilize droplets around 200 nm within the emulsification time. Thus, the surface active properties do not vary enough within the DexPτ family considered (5% < τ < 23%) to become a limiting parameter in emulsion preparation. Nevertheless, the various polymers are not equivalent for emulsion aging, as will be illustrated in the following. Emulsion Aging. It has been already demonstrated that submicronic emulsions stabilized with DexPτ polymers undergo aging mainly because of Ostwald ripening.7,8 Indeed, the main characteristics of emulsion aging are the linear variation of the cube of the average droplet radius with time, the overall displacement of the droplet size distribution, and the direct dependence of the aging rate on the oil solubility. A theoretical equation of Ostwald ripening has been derived by Lifshitz and Slyozov31 and independently by Wagner:32
R h 3(t) ) R h 3(0) + ωt
(2)
In eq 2, R h is the average radius of the droplets (m), t is the time (s) evolved since emulsion preparation, and ω (m3/s) is the rate of Ostwald ripening. This latter parameter is expressed as a function of the physicochemical properties of the oil and the interface:
8γiDV2mC∞ 9RT
ω ) k(φ)
(3)
In eq 3, γi is the interfacial tension (N/m), D is the diffusion coefficient of the oil in water (m2/s), Vm is the molar volume of the oil (m3/mol), C∞ is the solubility of the oil in pure water (mol/m3), R is the gas constant (8.314 J/mol K), T is the absolute temperature (K), and k(φ) is a correction factor that takes into account the influence of the oil volume fraction. The correction factor, k(φ), was added in more recent approaches, since the initial treatment corresponded to the limit of zero volume fraction. Starting from these data and in view of the use of DexPτ polymers as stabilizers in miniemulsion polymerization, several parameters were examined so as to quantify their effects on emulsion aging. Although the degree of hydrophobic modification of dextran has no significant influence on the initial droplet diameter, it modifies the rate of Ostwald ripening (Table 2). This effect is related to the higher interfacial activity (31) Lifshitz, I. M.; Slyozov, V. V. J. Phys. Chem. Solids 1961, 19, 35. (32) Wagner, C. Z. Elektrochem. 1961, 35, 581.
6960
Langmuir, Vol. 20, No. 16, 2004
Figure 4. Variation of the Ostwald ripening rate with the oil volume fraction for a dodecane-in-water emulsion stabilized with DexP23 (C ) 20 g/L). The line is the prediction given by theoretical equations.33,35
observed when τ increases. A detailed comparison between the experimental and calculated values of the slope, ω, has been reported previously.7 Moreover, with initial polymer concentrations between 5 and 20 g/L, the rate of Ostwald ripening varies only slightly (Table 2). This result can be related to the previous data about the surface coverage of the droplets. In that concentration range, we have R > 0.05 so that the surface coverage does not vary significantly, leading to only a small variation of ω. A 20% dodecane-in-water emulsion was prepared and diluted in pure water. Emulsion aging was monitored for the initial emulsion as well as for the diluted ones so as to check the variation of the Ostwald ripening rate with the oil volume fraction (Figure 4). It has been demonstrated that it is equivalent to perform the dilution either in pure water or in the initial polymer solution.7 Several theoretical derivations of the dependence of the Ostwald ripening rate on the oil volume fraction have been proposed,33-36 leading to several expressions of the function k(φ). Although ω appears to vary with the oil volume fraction, its variation is much less pronounced than that given by theoretical treatments. Taylor37 reports similar discrepancy for decane-in-water emulsions stabilized with sodium dodecyl sulfate (SDS) and up to φ ) 0.3. As a result, it seems that, in the case of oil-in-water emulsions, the variation of ω with φ is lower that predicted by theoretical descriptions of Ostwald ripening. These theoretical approaches never deal with emulsions. In all cases, spherical particles are considered in a continuous medium. The particles are supposed to interact through their neighboring diffusion fields and also by soft collisions. This last type of interaction is sometimes said to be predominant34 but could be questionable for oil droplets stabilized by a surfactant layer. The influence of the nature of the oil has been described quantitatively in a previous paper.7 Nevertheless, the use of oil mixtures was not treated. The rates of Ostwald ripening observed with decane/dodecane mixtures appear to decrease as the amount of dodecane is increased (Figure 5). Furthermore, using the equation proposed by (33) Brailsford, A. D.; Wynblatt, P. Acta Metall. 1979, 27, 489. (34) Voorhees, P. W. J. Stat. Phys. 1985, 38, 231. (35) Enomoto, Y.; Tokuyama, M.; Kawasaki, K. Acta Metall. 1986, 34, 2119. (36) Tokuyama, M.; Kawasaki, K. Physica A 1984, 123, 386. (37) Taylor, P. Colloids Surf., A 1995, 99, 175.
Durand et al.
Figure 5. Ostwald ripening rate of oil-in-water emulsions containing decane/dodecane mixtures as a function of the dodecane volume fraction in the oil phase. DexP23, 5 g/L; φ ) 9.1%. The line is a curve fitting made by using the equation of Kabal’nov.38
Figure 6. Variation of the Ostwald ripening rate of oil-inwater emulsions as a function of aging temperature: Decane (b); dodecane ([); hexadecane (2). DexP23, 5 g/L; φ ) 9.1%. The lines are the predictions made by using eq 3 and thermodynamic data from Abraham41 (- - -) or Taylor42 (---).
Kabal’nov38 (eq 4), the experimental points are fairly well described.
ωmixture )
(
)
φ1 φ2 + ω1 ω2
-1
(4)
In eq 4, ωmixture, ω1, and ω2 are the rates of Ostwald ripening of the emulsion containing the oil mixture, oil 1 alone, and oil 2 alone, respectively. φ1 and φ2 are the oil volume fractions of oils 1 and 2 in the oil mixture, respectively. The aging temperature was varied between 25 and 75 °C for emulsions stabilized with DexP23 and by using three different oils: decane, dodecane, and hexadecane. We observed that two phenomena occur when increasing the aging temperature. First, the rate of Ostwald ripening was increased following an Arrhenius-like curve (Figure 6). Second, a partial droplet aggregation started after ∼20 h at 75 °C, leading to macroscopic white aggregates which gave rise to an important creaming in the tubes and made the size measurements quite difficult. These two phenomena can be assumed to be independent, since aggregation was observed to occur similarly with all three (38) Kabal’nov, A. S.; Pertsov, A. V.; Aprosin, Y. D.; Shchukin, E. D. Colloid J. 1985, 47, 898.
Amphiphilic Polysaccharides
Langmuir, Vol. 20, No. 16, 2004 6961
Table 3. Thermodynamic Data of a Solution at 298 K from the Literature and Apparent Activation Energies of Ostwald Ripening Determined Experimentally for the Three Different Oils Used 0 ∆Gs(lfaq) (kJ/mol)
0 ∆Hs(lfaq) (kJ/mol)
0 ∆Ss(lfaq) (J/mol‚K)
oil
ref 41
ref 42
ref 41
ref 42
ref 41
ref 42
Eapp (kJ/mol)
decane dodecane hexadecane
47.1 54.5 69.4
47.3 53.9
11.0 16.2 26.6
7.5 21.3
-121 -128 -143
-133 -109
47.5 57.1 72.5
oils, whereas the rate of Ostwald ripening differs strongly from decane to hexadecane. The increase of the Ostwald ripening rate with temperature can be treated using eq 3. The molar volume of the oil, Vm, can be calculated thanks to a correlation giving its density at each temperature.39 The diffusion coefficient of the oil in water is calculated with the correlation proposed by Wilke and Chang.40 The solubility of the oil in water is estimated by two means: on one hand, by the thermodynamic data given by Abraham41 and, on the other hand, by the thermodynamic data obtained by Taylor with oil-in-water microemulsions diluted in water.42 In both references, the data are rather similar for decane and dodecane (Table 3). The correlation given by Abraham was used for hexadecane, but the authors are aware that its validity for this compound is questionable. The variation of the interfacial tension with temperature is neglected. A correction factor between the experimental and theoretical values of ω was used as determined previously (Table 2). The predicted variation of ω with temperature is rather close to the experimentally observed variation for the three oils no matter what thermodynamic data are used (Figure 6). Nevertheless, the experimental variation is always sharper than the predicted variation. This fact could probably be explained by a contribution of aggregate formation as described below. The experimental variation of ω with T can be depicted by an Arrhenius-like equation of the type
ω)
() (
)
Eapp a exp T RT
(5)
In eq 5, a (m3‚K/s) is a pre-exponential factor, Eapp (J/ mol) is the apparent activation energy of the molecular diffusion process, and R (J/mol‚K) is the gas constant. The pre-exponential term inversely proportional to T was introduced because ω is by itself inversely proportional to T (see eq 3). The found values of Eapp for the three oils are given in Table 3. The activation energy, Eapp, increases linearly with the number of carbon atoms in the oil, n, following the equation Eapp ) 6700 + 4100n (in J/mol). This variation is related to the thermodynamic properties of the oil solution. The formation of aggregates at high temperature seems to be independent of the increase of the Ostwald ripening rate. It is largely known that cellulose ether aqueous solutions undergo phase separation upon heating.43 Nevertheless, no such behavior was detected for the dextran derivatives used in this study. A similar absence of cloud point has been reported by Zhang et al. for nonionic saccharide surfactants with amide linkages.44 The behavior of aqueous solutions of DexP15 has been investigated upon heating by viscometric measurements. It appears (39) Perry, R. H. Perry’s Chemical Engineers’ Handbook; McGrawHill: New York, 2002. (40) Wilke, C. R.; Chang, P. AIChE J. 1955, 1, 264. (41) Abraham, M. H. J. Chem. Soc., Faraday Trans. 1 1984, 80, 153. (42) Taylor, P. Adv. Colloid Interface Sci. 2003, 106, 261. (43) Klug, E. D. J. Polym. Sci. 1971, 36, 491. (44) Zhang, T.; Marchant, R. E. J. Colloid Interface Sci. 1996, 177, 419.
Figure 7. Intrinsic viscosity of DexP16 in water at various temperatures (b) and comparison with the data of Guner45 for dextran T40 (O).
that the intrinsic viscosity of DexP15 in water decreases slightly when the temperature goes from 25 to 55 °C (Figure 7). Nevertheless, the variation is much less pronounced than that reported by Gu¨ner45 for an unmodified dextran with a similar molecular weight (Figure 7). This could be attributed to the aggregated state of the partly hydrophobized macromolecules. As a result, the droplet aggregation observed around 75 °C cannot be attributed to a significant dehydration of DexP macromolecules. This point should be clarified by future experiments. Miniemulsion Polymerization of Styrene. From Styrene Droplets to Latex Particles. Styrene-in-water emulsions were prepared by sonication in the presence of three dextran derivatives: DexP5, DexP13, and DexP18. For a given polymer, the polymer concentration in the aqueous phase, C (g/L), as well as the styrene volume fraction, φ, were varied. It appears that plotting the initial droplet diameter as a function of the initial weight ratio R ) [polymer]/[oil] allows one to gather all the experiments onto a single curve, as already described in the case of dodecane-in-water emulsions. The droplet size was around 160 nm at maximal surface coverage for the three amphiphilic copolymers (Figure 8). However, drastic differences were observed during the polymerization step for the three derivatives. The best results were obtained in the case of DexP18 for a styrene volume fraction, φ, of 10%. In this case, the particles size was close to the initial droplet size and the coagulate amount was below 5 wt %. Polystyrene particles covered by a hydrophilic shell composed of amphiphilic polysaccharide could thus be synthesized by this method. However, an increase of φ led to an increase of the particles size and the amount of coagulum. Furthermore, the particle size and amount of coagulum strongly increased by decreasing τ. In the case of DexP5, no latex was obtained. This has to be correlated to the instability observed at higher temperatures for the dodecane-in-water emulsions. (45) Gu¨ner, A. J. Appl. Polym. Sci. 1999, 72, 871.
6962
Langmuir, Vol. 20, No. 16, 2004
Figure 8. Droplet diameter of the initial styrene emulsion (O) and particle diameter (b) of the final latex for miniemulsion polymerizations stabilized by DexP18 (for reaction conditions, see the text). The oil volume fraction is 10%.
Figure 9. Styrene conversion in miniemulsion polymerization at 75 °C (for reaction conditions, see the text) stabilized by DexP18. The line is the prediction made by using classical kinetic equations and rate constants.46
On the basis of former results, the Ostwald aging rate of styrene-in-water emulsions can be calculated and we get ω ) 3.7 × 10-22 m3/s; however, the addition of 5% hexadecane in the aqueous phase reduces it by a factor of 100 000 (ω ) 2.4 × 10-27 m3/s)! Furthermore, no variation of the droplet size of the styrene-in-water emulsions was observed within a few hours. Finally, the kinetics of styrene polymerization is fast, since the maximum monomer conversion is reached within 2 h (Figure 9). As a result, Ostwald ripening cannot be the destabilization process of our miniemulsions, which is probably due to aggregation of the droplets, as already observed in the case of oil-in-water emulsions. Polymerization Kinetics. Concerning the kinetics of styrene polymerization, the shape of the experimental curve is correctly described by the usual theoretical equation giving monomer conversion as a function of time and taking literature values of the rate constants for styrene and AIBN46 (Figure 9). Nevertheless, the theoretical equation does not take into account the fact that the droplet core becomes more and more viscous with polymer formation. As a result, the experimental curve is below the predicted one and does not reach a complete conversion even after 6 h (it is limited to ∼90%). Similar (46) Odian, G. Principles of Polymerization, 3rd ed.; Wiley: New York, 1994.
Durand et al.
Figure 10. Surface coverage of the initial styrene droplets by DexP18 immediately after emulsion preparation and surface coverage of final latex particles as a function of the polymer concentration remaining in the aqueous phase (for reaction conditions, see the text). The line is a curve fitting made by the Scatchard equation.29
limiting conversions have been reported for styrene miniemulsion polymerization.47,48 Surface Characteristics of the Obtained Particles. The amount of nonadsorbed polymer remaining in the aqueous phase immediately after sonication was determined using the anthrone method. When plotting the droplet surface coverage, Γ (mg/m2), as a function of the concentration of nonadsorbed polymer remaining in the aqueous phase, Cexcess (g/L), we get a curve having the shape of an adsorption isotherm (Figure 10). Furthermore, the curve can be linearized using the Scatchard method29,30 and we get a maximum surface coverage equal to 5.1 mg/m2, which agrees with the experimental results (Figure 10). The results are in total accordance with the experimental data obtained in the case of emulsions. This proves that no desorption of the dextrane derivatives occurred during the polymerization step. Conclusion In the foregoing, the preparation of polysaccharidecovered polystyrene nanoparticles was explored carefully, starting from the earliest stage of the process, which is the initial surface tension decrease observed for the amphiphilic polysaccharides used in this study. Its kinetics was studied as a function of polymer structural characteristics (degree of hydrophobic substitution) and at various polymer concentrations. Several hydrocarbon oils, either aliphatic (octane, decane, dodecane, and hexadecane) or aromatic (styrene), were tested. It was shown that the interfacial tension at equilibrium is only attained after a very long time, maybe because of the slow diffusion of the polymer to the surface, followed by a rearrangement of the chains at the interface. The emulsifying properties of the polymeric surfactants were illustrated by the preparation of oil-in-water emulsions. For low polymer/oil ratios, the droplet size is controlled by the initial amount of polymer. On the contrary, for high polymer/oil ratios, the droplet size seems to level down, indicating that other parameters (such as the diffusion coefficient of the polymeric surfactant) become predominant. Emulsion aging occurs by Ostwald (47) Chern, C. S.; Liou, Y. C. J. Polym. Sci., Part A: Polym. Chem. 1999, 37, 2537. (48) Wang, C. C.; Yu, N. S.; Chen, C. Y.; Kuo, J. F. Polymer 1996, 37, 2509. (49) Sakai, T.; Kamogawa, K.; Nishiyama, K.; Sakai, H.; Abe, M. Langmuir 2002, 18, 1985.
Amphiphilic Polysaccharides
ripening, and it was demonstrated that the theoretical equation of Lifshitz, Slyozov, and Wagner (LSW) correctly describes all the experimental results. Especially the influence of polymer chemical structure can be conveniently correlated to interfacial tension results through the LSW equation. Miniemulsions stabilized with dextran derivatives were used for the radical polymerization of styrene. Following this procedure, polysaccharide-covered polystyrene nanoparticles were prepared and characterized (size and surface coverage). The size of the particles is directly correlated to that of the initial droplets for styrene volume fractions around 10% and a degree of substitution of the
Langmuir, Vol. 20, No. 16, 2004 6963
polysaccharide of 18%. On the contrary, for initial styrene volume fractions around 20%, particles exhibit a larger size than the initial droplets, indicating that coalescence processes take place during polymerization. Furthermore, at a lower degree of substitution, this coalescence becomes predominant and no latex is obtained. However, amphiphilic derivatives of dextran with a high degree of modification are well-suited surfactants for the miniemulsion polymerization of styrene and polysaccharide-covered nanoparticles have been successfully synthesized in this way. LA0490341
