JOURNAL OF CHEMICAL PHYSICS VOLUME 110, NUMBER 6 8 FEBRUARY 1999 Periodic surfaces and cubic phases in mixtures of oil, water, and surfactant Alina Ciach and Robert Hołysta) Institute of Physical Chemistry PAS and College of Science, Dept. III, Kasprzaka 44/52, 01224 Warsaw, Poland ~Received 9 December 1997; accepted 4 November 1998! We study a ternary mixture of oil, water, and surfactant in the case of equal volume fractions of oil and water using the Landau–Ginzburg model derived from a lattice model of this ternary mixture. We concentrate on a phase region close to a coexistence line between microemulsion and cubic phases. In our model the bicontinuous cubic phases exist in a narrow window of the volume fraction of surfactant r s '0.6. The sequence of phase transitions is L→G→D→ P→C as we increase r s along the cubic-microemulsion bifurcation line. Here L stands for the lamellar phase and C for the cubic micellar phase. The gyroid G, primitive P, and diamond D phases are all bicontinuous. The transitions weakly depend on the temperature. The increase of r s is accompanied by the increase of the surface area per unit volume. In the case of fluctuating monolayers the interface is diffused and the average area of the monolayer per unit volume is larger than the ‘‘projected area,’’ i.e., the area of the surface describing the average position of the monolayer, per unit volume. The effect is the strongest in the L and the weakest in the C structure. © 1999 American Institute of Physics. @S0021-9606~99!51306-8# I. INTRODUCTION bic cell can yield partial information about a topology of the structure. One can expect a small unit cell and the small surface area per side of the unit cell in structures of the simple topology8 and the large values for the structures of complex topology. Typically the surface area per side of the cubic cell is between 2 and 4 for simple topology surfaces such as P,D,G,I-W P 6 , O-CTO 6 etc., but it can be larger than 7 for the complex topology surfaces.9–11 In binary mixtures of the water, the surfactants or the lipids the most common structure is the gyroid one, G, existing usually on the phase diagram between the hexagonal and lamellar mesophases. This structure has been observed in a very large number of systems1,12–15 and in the computer simulations of the systems.16 The G phase is found at rather high surfactant concentrations, usually much above 50% by weight. Other cubic phases even at very low concentration of surfactants can be found in ternary mixtures ~with water and oil!.8,17,18 For the didodecyldimethylammonium bromide, water, and styrene system, the periodic surfaces are found over the huge range of water fraction from 11% to over 80%. Most of the studies concentrated on the cases with large amounts of water and surfactant and the influence of added oil on the phase transitions. It was shown that the emerging phases depend on the properties of the oil, i.e., whether it penetrates the surfactant or swells the bilayer. In the highly asymmetric case of a large amount of the water and the surfactant and a small amount of the oil it has been found that with increasing the volume fraction of water one finds the following progression of cubic phases: G→D→ P. These studies are time consuming due to the very long equilibration times of weeks or even months. Nevertheless the experimental data are very rich in this case.19–23 Here we concentrate our theoretical studies concerning A periodic surface is a surface that moves onto itself under a unit translation in one, two, or three coordinate directions similarly as in a periodic arrangement of atoms in regular crystals. Triply periodic minimal surfaces are periodic in all three directions and are in addition characterized by the zero mean curvature at each point of the surface. The first triply periodic minimal surface ~primitive, P, surface! was discovered by Herman Schwarz in 1865. The interest in periodic surfaces in this century was due to the experimental observation ~Luzzati et al.1! that bilayers of lipids ~or surfactants! in water solutions form at suitable thermodynamic conditions ordered bicontinuous structures. To describe these mesophases the ideas of minimal surfaces and related hyperbolic structures were next used by Larsson, and Hyde et al.2–5 Also, new triply periodic minimal surfaces ~among them the gyroid, G, surface! were discovered by a mathematician Alan Schoen.6 If we draw a surface through the middle of the triply periodic lipid ~or surfactant! bilayer, it divides the volume into two disjoint subvolumes each continuous in the whole volume of the system.7 Therefore, the name bicontinuous has been given for such structures. The x-ray scattering provides information about the symmetry of the structure and, therefore, discerns between different periodic surfaces. Here we concentrate on the cubic symmetries. The set of the reflections for the P cubic structure (Im3m symmetry! index to &: A4: A6: A10: A12: A14: A16, that of the D structure ( Pn3m symmetry! index to &:): A4: A6: A8: A9: A10, and that of the G structure (Ia3d symmetry! index to A6: A8: A14: A16: A20: A22: A24. Finally, the measurements of the surface area inside the cua! Electronic mail: [email protected] 0021-9606/99/110(6)/3207/8/$15.00 3207 © 1999 American Institute of Physics Downloaded 23 Feb 2006 to 213.135.39.26. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp 3208 J. Chem. Phys., Vol. 110, No. 6, 8 February 1999 ordered cubic phases on the largely unexplored region of phase diagram of the ternary mixture of water, oil, and surfactant, where the volume fractions of oil and water are comparable. It is well known that bicontinuous microemulsions are formed in such systems. In bicontinuous microemulsions the surface covered with the surfactant divides volume into the water rich and the oil rich subvolumes. The diffusion measurements provide direct information about the bicontinuity of the structure. The pulsed gradient nuclear magnetic resonance ~NMR! self-diffusion technique gives the selfdiffusion rates of all the components in the structure,24 providing a direct check on the continuity and the extension of a region occupied by the components. If any of the components of the system is closed in a finite ~small! volume, then its effective diffusion constant ~measured as a mean-squared displacement divided by time! goes to zero, whereas in a continuous structure the effective diffusion coefficient is nonzero. The range of order in microemulsions is comparable to the typical length of the structure ~domain size!. Topological properties of the surfactant monolayers resemble those of periodic surfaces. It is still not clear, however, whether the transition between the bicontinuous microemulsion and the ordered bicontinuous cubic phases occurs in nature. When the volume fractions of the oil and the water are equal, one finds the cubic phases in a narrow window of the surfactant concentration around 0.5 weight fraction.19,20 However, it is not known whether these phases are bicontinuous. No experimental evidence has shown that there exist bicontinuous cubic phases with the ordered surfactant monolayer, rather than bilayer, forming the periodic surface. We study the possibility of such ordering transition between microemulsion ~random isotropic monolayer! and ordered bicontinuous cubic phases ~ordered periodic monolayer! within Landau– Ginzburg approach in this work. For the comparable volume fractions of oil and water, bicontinuous microemulsions and the other phases with vanishing spontaneous curvature are stable. The lack of an oil– water symmetry is for this part of phase diagram not of primary importance, since the surfactant monolayer is not biased towards either oil or water occupied region. Effectively oil and water play symmetrical roles with respect to the surfactant in this limited part of the phase diagram. Hence for an equal oil and water volume fractions one can describe the system by an oil–water symmetric model. The Landau–Ginzburg model used here25 has been derived from the lattice model of the ternary mixture.26 A volume occupied by a single molecule is fixed in the lattice model. Due to the oil–water symmetry in the model the spontaneous curvature is zero. In the model the important field is the local oil–water volume fraction difference and the fields describing local assembling and orientational ordering of amphiphiles. The model differs strongly from the theoretical description introduced in terms of the geometrical properties of the surfactant molecules and the surface covered by them, where the various phase transitions are interpreted in terms of the geometry of the system and change of the curvature or the surface area per head of surfactants.27–29 Nevertheless we show a connection between the results of the A. Ciach and R. Hołyst Landau–Ginzburg model and the geometry of the surface dividing the volume into oil-rich and water-rich regions. In our model we compute the sequence of phase transitions between G, D, and P phases as a function of r s . The paper is organized as follows. In Sec. II we present the equations describing different phases and discuss their geometrical properties. In Sec. III we discuss the model and present its simplified form near the coexistence line between the microemulsion and cubic phases. In Sec. IV we show the results and discuss the calculated sequence of phase transitions. The conclusions are also contained in Sec. IV. II. THE L, P, D, G, AND C STRUCTURES STUDIED IN THE MODEL Let f~r! denote the scalar field representing the difference between oil and water volume fraction at the point r. The structures studied in this paper can be represented by the following Fourier series: f ~ r! 5 ( A ~ k ! cos@ 2 p k•r2 a ~ k !# , k ~2.1! where k describes the reciprocal lattice vectors for a given lattice, a (k) is a phase shift, and A(k) is an amplitude associated with a modulus, k, of a given k-vector. Close to the bifurcation line ~a line of continuous phase transition and/or divergence of the static structure factor! the amplitudes are very small and it is justified to use only the first term in the Fourier series. The lamellar structure, L, is given by f ~ r! 5A L cos~ Z ! . ~2.2! The P structure is described by f ~ r! 5A P ~ cos~ X ! 1cos~ Y ! 1cos~ Z !! , ~2.3! the D structure is given by f ~ r! 5A D ~ cos~ X ! cos~ Y ! cos~ Z ! 2sin~ X ! sin~ Y ! sin~ Z !! , ~2.4! while the gyroid G structure by f ~ r! 5A G ~ sin~ X ! cos~ Y ! 1sin~ Y ! cos~ Z ! 1cos~ X ! sin~ Z !! . ~2.5! Finally the cubic micellar structure, C, is represented by f ~ r! 5A C cos~ X ! cos~ Y ! cos~ Z ! . ~2.6! Here X52 p x/d, Y 52 p y/d, Z52 p z/d, A i , ~i is L, P, D, G, or C! is an amplitude and d is the size of the unit cell. Due to the symmetry between oil and water the dividing surface where the surfactant is located is given by f ~ r! 50. ~2.7! The surfaces defined by Eqs. ~2.3!–~2.5! are called the nodal surfaces.30–34 Although the nodal surfaces are neither minimal nor constant mean curvature they can be used as an ansatz for such surfaces.35 For example, the topology of the minimal surface and its symmetry are exactly the same as the ones of the corresponding nodal surfaces. The geometrical properties such as the surface area per side of the unit cell (Sd) ~here S is the surface area per unit volume and d is the size of the unit cell! are very well represented by the corre- Downloaded 23 Feb 2006 to 213.135.39.26. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp J. Chem. Phys., Vol. 110, No. 6, 8 February 1999 A. Ciach and R. Hołyst 3209 FIG. 1. The nodal P primitive surface given by Eqs. ~3.3! and ~3.7! inside the unit cell of the primitive bicontinuous cubic structure ~a view along a diagonal!. FIG. 3. The nodal G gyroid surface given by Eqs. ~3.5! and ~3.7! inside the unit cell of the gyroid bicontinuous cubic structure ~a view along a diagonal!. sponding properties of these nodal surfaces. The quantity mentioned above has the value very close to the one for the minimal surface ~differences about 0.5%!. We find using Eq. ~2.7! together with Eqs. ~2.3!–~2.5! the following values for these nodal surfaces: for P, Sd52.353, for the G surface Sd53.092 and for the D surface we have Sd53.839. In Figs. 1–4 we show the nodal P, D, and G surfaces and our cubic micellar C surface in the unit cell of the given structure. In the actual calculations we will use the Fourier amplitudes of the structures.25 ever, cubic phases, present in mixtures with strong surfactants, are not stable in this model.9–11 To study such phases one has to consider more complicated LG models, in which additional order–parameter, describing orientational ordering of surfactant particles, is present. In case of binary mixtures such model was introduced in Ref. 37 whereas for ternary mixtures a model in which cubic phases are stable was introduced and studied in Ref. 25. Important advantage of the latter model is the fact that all the coupling constants are expressed in terms of the surfactant volume fraction r s , temperature, and a single phenomenological parameter g describing the amphiphilicity of the surfactant through the strength of interparticle interactions. Small g corresponds to weak, and large g to strong surfactants. No parameters in this model are fitted, and all the calculated quantities are expressed in terms of directly measured quantities and a single phenomenological parameter specifying the material properties of the system. No additional assumptions concerning for- III. THE LANDAU–GINZBURG „LG… MODEL NEAR THE BIFURCATION LINE Many properties of mixtures with weak surfactants are well described by the elegant Gompper–Schick ~GS! model,36 in which the free energy is a functional of the local difference between volume fractions of oil and water. How- FIG. 2. The nodal D diamond surface given by Eqs. ~3.4! and ~3.7! inside the unit cell of the double diamond bicontinuous cubic structure ~a view along a diagonal!. FIG. 4. The nodal C surface given by Eqs. ~3.6! and ~3.7! inside the unit cell of the micellar cubic structure ~a view along a diagonal!. Downloaded 23 Feb 2006 to 213.135.39.26. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp 3210 J. Chem. Phys., Vol. 110, No. 6, 8 February 1999 A. Ciach and R. Hołyst mation of the monolayer or its properties are made. Within this LG model one can calculate the positions of surfactant surfaces and their properties for different r s , and draw conclusions about such physical properties, as the sequence of cubic phases or the effect of diffused ~or delocalized! oil– water interfaces on the average surface area. Please note that S is a projected area per unit volume and in general is not equal to the average area per unit volume S/V of the surfactant-occupied surface. The projected area is the area of the surface that describes the average position of the interface between water rich and oil rich domains @given by Eq. ~2.7!#. Instead of introducing the LG model on symmetry grounds, in Ref. 25~a! the free—energy functional is derived from the lattice microscopic model26 in which only two parameters characterize the interactions in the case of the oil– water symmetry: b is the strength of the water–water ~oil– oil! interaction, and c describes interaction between the water ~oil! and the amphiphiles ~surfactants!. Eventually only a combination of b and c enters as a single interaction parameter g 5(2c/b) 2 . Interaction between the amphiphiles and ordinary particles is proportional to a scalar product between the orientation of the amphiphile and the distance between the particles. In this way the amphiphilicity is explicitly taken into account. The interactions between amphiphiles in the simplest version of the model are neglected. In the model of Ref. 26 the microscopic density distributions in the case of close packing of the molecules are given by r̂ a (r) where a denotes oil, water, or surfactant in different orientations and r denotes a lattice site. r̂ a (r) 51(0) if the site r is ~is not! occupied by the component a. This microscopic model can be investigated in the meanfield ~MF! approximation, in which microscopic configurations r̂ a (r) occur with a probability ;exp@2bHMF# , where for any microscopic state the MF—Hamiltonian H MF is given by H @ r̂ a ~ r!# 5 MF (r (a c a~ r!~ r̂ a~ r! 2 1 2 r a ~ r!! . ~3.1! r a (r) is the average density ~introduced here to compensate for double counting of pairs of molecules! and c a (r) is the mean field felt by the particle of the kind a at r, resulting from interactions with the remaining molecules. In other words, H MF@ r̂ a (r) # is equal to the total energy of the microscopic state @ r̂ a (r) # in a hypothetic system in which every particle experiences only an external field c a (r), which is a functional of the average density.26 The strength of this hypothetic external field at a given point is equal to the field provided by the rest of the system, as if it were in the hypothetic state corresponding to the minimum of the grand thermodynamical potential.26~b! In the Weiss-type formulation of the mean field, the average density is given by r a 0 ~ r0 ! 5const ( @ r̂ a ~ r!# r̂ a 0 ~ r0 ! exp~ 2 b H A particular microscopic state in such ordered phases appears with a probability ;exp(2bHMF@ r̂ a (r) # ) which strongly depends on the average densities @through c a (r)#. For the structures with sharply localized average states the microscopic configurations different than the average solution appear with negligible probability. However, if r a (r) are slowly varying functions, as is the case close to secondorder transitions,26 then many microscopic configurations occur with comparable probability @see Eq. ~3.2!#. In an equivalent formulation of the MF, in order to find the equilibrium state, i.e., the average density, with the probability distribution ;exp@2bHMF# , one minimizes the MF approximation for the grand thermodynamical potential.26 Close to the continuous transition one can further simplify the analysis, by using the continuous approximation for the lattice model discussed above. The LG functional can be defined as a continuous approximation for the MF thermodynamical potential. In Ref. 25~a! the LG functional corresponding to the microscopic model described above is derived in a standard way and is found to be a functional of three-order parameter ~OP! fields: f~r!, the concentration difference between oil and water, r~r!, the deviation of the average of the local density of surfactant from the global value, and a vector field u~r! describing the orientational ordering of the amphiphiles. The resulting functional assumes the form V eff5 ~ V 2 1V int! b, with V 25 dr@ 21 a 2 f 2 1 21 ~ ¹ f ! 2 1 21 a 2 r 2 1 12 ~ ¹ r ! 2 ~3.3b! and V int5 E dr@ 3!1 ~a3r31b3f2r1c3uuu 2 r ! 1 4!1 ~a4f41a4r41b4f2r21c4r2uuu 2 1A 4 u uu 4 !# . ~3.3c! All the coupling constants can be expressed in terms of the average surfactant volume fraction r s , temperature t 5kT/b, and a parameter describing the amphiphilicity g 5(2c/b) 2 . The grand-thermodynamical potential and the temperature are calculated in units of b. Macroscopically b is related to the critical temperature of oil–water separation by kT c 5d(12 r s )b. The explicit expressions for the coupling constants which are used here for bifurcation analysis are a 2 52 S F a 2 52 @ r̂ a ~ r!# ! , which is to be solved self-consistently. For ordered states the average densities r a (r) are not constant, but rather have some kind of oscillatory behavior. E 1 12 @ u uu 2 1 ~ ¹•u! 2 1 ~ ¹3u! 2 # 2Ju•¹ f # MF ~3.2! ~3.3a! J5 D t 2d , 12 r s G t 2d , r s ~ 12 r s ! S D b 35 ~3.4a! 2 r sg 3t ~3.4b! 1/2 , ~3.4c! 6& t , ~ 12 r s ! 2 ~3.4d! Downloaded 23 Feb 2006 to 213.135.39.26. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp J. Chem. Phys., Vol. 110, No. 6, 8 February 1999 c 3 52 a 45 3& , rs 8t ~ 12 r s ! 3 A 45 A. Ciach and R. Hołyst 3211 ~3.4e! ~3.4f! , 18 . 5trs ~3.4g! The other coupling constants which are not needed in the present paper are given in Ref. 25~a!. The length unit in Eq. ~3.3! is by construction of the LG model equal to the lattice constant a[1 of the original model. In the lattice model it was assumed that a lattice cell is occupied by a single particle ~in the case of water by a cluster of particles!. Hence the lattice constant is identified with the linear size of the amphiphile, and thus the length unit used here is ;25 Å. In our mesoscopic description the volume is measured in units of the volume occupied by a surfactant molecule, and the area is measured in units of the area occupied by an amphiphile. In other words, in our model the area of the monolayer is the dimensionless quantity equal to the number of amphiphiles residing on the monolayer. Hence, it should be identified with the area rescaled by the surfactant parameter of the corresponding structure ~see the discussion in Ref. 38!, when comparing our results with those of the microscopic, geometrical models.38 The rescaled area of the monolayer per unit volume S is equal to the surfactant volume fraction if no surfactant molecules occur outside the monolayer.38 An example of a phase diagram, calculated in Ref. 25~a!, is shown in Fig. 5~A! for g 550 ~strong surfactant!. For low surfactant-volume fraction there is a continuous transition between the homogeneous fluid and coexisting homogeneous oil- and water-rich phases, given by a 2 50. Next there is a Lifshitz point, at r s 5 r Ls , with r Ls '0.1 for g 550, at which the Lifshitz line meets the bifurcation line. The Lifshitz line separates the uniform phase into a structureless region ~above! and microemulsion ~below!. Here by microemulsion we mean macroscopically uniform phase in which water– water structure factor assumes maximum for the wavenumber kÞ0. For r s . r Ls the instability of the uniform phase for the wave-number k5k b Þ0 occurs at temperatures higher than the instability with respect to oil–water separation (a 2 50 and k50) and is given by25 e 2 [ ~ 11a 2 2J 2 12 Aa 2 ! /450, and k b 5a 1/4 2 . ~3.5! The bifurcation line with k5k b Þ0 defines continuous transition between microemulsion and different ordered phases. In Ref. 25~a! it was found that the continuous transition to the ordered phases terminates at the tricritical point indicated by the last cross on the right in Fig. 5~A!. For r s larger than at the tricritical point the transition between microemulsion and ordered phases becomes first order within the Landau model. The different ordered phases coexist below the bifurcation line. The first-order transition lines meet the continuous transition to the microemulsion at points indicated by the crosses in Fig. 5. Fig. 5~B! shows a portion of the bifurcation line corresponding to the stability region of the cubic phases. FIG. 5. ~A! The bifurcation ~solid! line between a homogeneous fluid and cubic phases. t is the dimensionless temperature @see below Eq. ~3.3!# and r s is the surfactant volume fraction ~the volume fraction of oil and water are equal!. The dashed line is the Lifshitz line below which the water–water structure factor assumes maximum for a wave-number kÞ0, as in microemulsions. o/w denotes the coexisting oil and water-rich phases. lc denotes liquid–crystalline phases, separated by crosses. The last cross is the tricritical point separating the continuous and the first-order phase transitions. ~B! The part of the bifurcation line @shown in full in ~A!# between the microemulsion and the cubic phases. t is the dimensionless temperature @see below Eq. ~3.3!# and r s is the surfactant-volume fraction. L, G, D, P, and C denote the lamellar, gyroid, diamond, primitive, and cubic micellar structures, respectively. Along the bifurcation line the static structure factor diverges for the wave vector k b . The locations of the coexistence regions for lower temperatures are not indicated. We limit ourselves to the vicinity of the continuous transition, where analytical results can be obtained. Much below the continuous transition extensive numerical calculations are necessary. The full MF analysis of phase diagrams for weak as well as for strong surfactants has been performed in Ref. 26~c! for the simplified version of the present model. The interactions in the model of Ref. 26~c! are the same as in the present case, but the locations of the particles are restricted to sites of a simple-cubic lattice and, unlike in the present model, the orientations of amphiphiles are restricted to six directions compatible with the lattice structure. Because of this restriction the ordered phases other than the D phase are not stable, as they would correspond to ‘‘forbidden’’ orientations of amphiphiles ~for example, in the case of the lamellar phase Downloaded 23 Feb 2006 to 213.135.39.26. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp 3212 J. Chem. Phys., Vol. 110, No. 6, 8 February 1999 A. Ciach and R. Hołyst the direction of oscillations would correspond to the diagonal of the lattice cell!.26~c! For the uniform phases both models give the same results, and the general features of the phase diagrams of the full and the simplified model should be similar. For strong surfactants @Fig. 5~b! of Ref. 26~c!# it was found that for low temperatures there are large three- and two-phase regions; for example,26~c! for temperatures t ,1, the oil-, water-, and D-phase coexistence occurs for 0, r s ,0.7. For higher temperatures ( t .1.5), however, a substantial part of the phase space is occupied by the ordered D phase, as determined by the bifurcation analysis in perfect agreement with the full MF study.26~c! The results obtained in the simplified model indicate that also in the full model substantial regions of stabilities of the ordered phases are to be expected below the bifurcation line, and that only for significantly lower temperatures the large two- and three-phase coexistence regions occur. This feature of the phase diagrams indicates that the bifurcation analysis is not only limited to the immediate vicinity of the bifurcation line, but should apply to a rather extended temperature region. We should also stress that in this model the pure surfactant system does not undergo ordering phase transition, because in the model introduced in Ref. 25~a!, the interactions between amphiphiles are neglected. For a very high surfactant concentration the results obtained in this model are not expected to agree with experiments. The amphiphiles do not order by themselves, thus if there is too much surfactant, the ordering effect of interactions between ordinary particles and amphiphiles becomes too weak and the system becomes disordered, whereas in reality the interactions between the amphiphiles support formation of a lamellar phase for very high surfactant volume fraction. We extend here the bifurcation analysis of Ref. 25~a!, and consider in addition to the already studied structures, the gyroid phase. From now on we concentrate on r s . r Ls , for which the instability of the disordered phase corresponds to k5k b .0. Close and below the bifurcation line t 5 t b ( r s ), given explicitly by Eq. ~3.5!, V has a much simpler form, if the contributions of order O( e 6 ), are neglected.25 In equilibrium V assumes minimum. This condition allows to express r and u in terms of f (k)5O( e ) corresponding to bifurcation, and V5 2k 2b (k 11k b 2 2 2 e u f̃ ~ k ! u 1 1 8 d Kr 4! k1 ,k2 ,k3 ,k4 ( S( D 4 i ki 4 3@ A ~ $ k̂i % ! 2G ~ $ ki % !! ] )i f̃ ~ ki ! 1O ~ e 6 ! . ~3.6! In the above k b 5a 1/4 2 is the wave number corresponding to bifurcation,25~a! and A ~ $ k̂i % ! 5a 4 1A 4 a 2 ~ k̂1 •k̂2 !~ k̂3 •k̂4 ! , G ~ $ ki % ! 5 ~ b 3 2c 3 k1 •k2 !~ b 3 2c 3 k3 2k4 ! . 3 ~ a 2 1 u k1 1k2 u 2 ! ~3.7a! ~3.7b! In Eq. ~3.6! ( 8 is a summation over vectors such that u ki u 5k b . The stable phase is the one giving the lowest value of Eq. ~3.6! for the given value of r s and for u ku 5k b given by Eq. ~3.5!. The Fourier amplitudes of the structures defined in Eqs. ~2.2!–~2.6! just below the bifurcation can be written in a form F m m j j G f̃ ~ k! 5F w ( d Kr ~ k2k b p̂ j ! 1w * ( d Kr ~ k1k b p̂ j ! . ~3.8! Here w * denotes the complex conjugate to w. For the structures L, P, and C w51, for D w511i and for G w5i. The number and orientations of the unit vectors p̂ j are different for different structures. For the L phase there is m51 vector in the direction of oscillations. For the D and C phases m 54 and p̂ j form tetrahedron. For the G phase m56 and each unit vector p̂ j is parallel to one of the 6 diagonals of the three adjacent sides of the cube. The size of the unit cell is d 52 p /k b for the L and P phases; d52 p )/k b for the D and C phases and d52 p &/k b for the G phase. In Ref. 25~a! the hexagonal phase was considered in addition to the cubic phases listed in the Sec. II. It was shown that this phase is not stable in this part of the phase diagram, that is close to the stability region of microemulsion. Hence, we do not consider this phase in the present work. Just below the bifurcation f~r! is a slowly varying function of the position. For example, in the simplest, lamellar phase with the modulations in a ẑ direction and for the surfactant surface located at z50, f (z); e sin(2pz/kb). The density of surfactant is given by ; e cos(2pz/kb), hence it deviates from zero in an extended region. The width of the surfactant-occupied interface is very large. IV. RESULTS, THEIR GEOMETRICAL INTERPRETATION AND DISCUSSION We have calculated V given by Eq. ~3.6! for the structures described in the Sec. II and the portion of the phase diagram corresponding to stable cubic phases is shown in Fig. 5~b! for strong surfactants ( g 550). For weaker surfactants ~but still strong, i.e., with water–amphiphile interactions stronger than the water–water interactions! the sequence of phases is the same, L→G→D→ P→C. For sufficiently strong surfactants, for which cubic phases are stable, the only dependence of the phase diagram on the strength of surfactant is the extent of the region of stability of the lamellar phase. It grows with the strength of the surfactant. For surfactants weaker than in the case shown in Fig. 5 the cubic phases occur for somewhat lower surfactantvolume fraction, for example, for g 516 cubic phases appear for r s '0.45. It is worth noting that the surfactant-volume fraction in cubic phases found in this model agrees very well with the surfactant-volume fraction for which unidentified cubic phases are stable in ternary mixtures with equal oil and water volume fractions for C i E j with (i, j)5(8,4) or ~10,5!.19 It is also important to note that the bicontinuous cubic phases exist in a very narrow range of surfactantvolume fractions in our model, and that they transform into the cubic micellar phase when r s is increased. Downloaded 23 Feb 2006 to 213.135.39.26. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp J. Chem. Phys., Vol. 110, No. 6, 8 February 1999 FIG. 6. The projected rescaled surface area @i.e., the area multiplied by the surfactant parameter ~Ref. 38!# per unit volume S divided by the surfactant volume fraction for different structures along the bifurcation line as a function of surfactant volume fraction r s . Please note that due to the geometrical constraints this quantity cannot exceed the length of the surfactant l . Here we set l 51 for convenience. In a structure with all surfactant molecules located at monolayers, the volume fraction of surfactant should be equal to the average surface area times the surfactant parameter, times the width of the monolayer divided by the volume, i.e., r s 5S l /V, where S is the surface area rescaled by the surfactant parameter. This is true for a single surface with no intersections. In real systems the microscopic configurations of the monolayers @which are taken into account in the derivation of the Landau–Ginzburg model in Eq. ~3.2!# can differ from the average configuration and the rescaled average surface area per unit volume S/V is larger than the rescaled ‘‘projected’’ surface area per unit volume S. The surfactant density averaged over these configurations is not a sharply localized function, as for a single configuration of a monolayer. It is a smooth function, nonvanishing in a region of width comparable to the standard deviation from the average position of the monolayer. The projected surface corresponds to the average location of the interface between oil-rich and water-rich domains and in our case is given by Eq. ~2.7! @ f (r)50 # and Eqs. ~2.2!–~2.6!. Hence we expect S/ r s < l 21 . The ratio S/ r s , calculated for different phases below the bifurcation is shown in Fig. 6. Recall that S is the rescaled surface area per unit volume, i.e., the area multiplied by the surfactant parameter.38 The rescaled surface area per volume is an increasing function of the surfactant-volume fraction and it determines the sequence of phases. Moreover, we have found that the deviations of the value S/ r s from unity ~for convenience we set l 51 in our model! are different in different phases. In the model considered here within the MF approximation the surfactant-occupied interface is very diffuse close to the second-order transition. As discussed in Sec. III, this indicates that different microscopic configurations appear with comparable, nonvanishing probability @Eqs. ~3.1! and ~3.2!#. Such microscopic configurations, when averaged with the Boltzmann probability distribution ~3.2!, lead to the diffusive interface. The relevant ~with nonvanishing probability! microscopic configurations in the A. Ciach and R. Hołyst 3213 FIG. 7. The size d of the unit cell for the structures stable just below the bifurcation line. The unit of length is the thickness of the surfactant monolayer. r s is the surfactant-volume fraction. model studied here play a similar role as the undulations of the sharp monolayer—they broaden the oil–water interface. We have found that the effect of broadening of the interface on the value S/ r s in different phases is different, and we have a quantitative measure of its strength through the projected area. As expected, the effect is the stronger, the more independent the surfactant-occupied surfaces are. Indeed, in the lamellar phase the difference between the average and the projected surface area is the largest, ~about 20% of the relative difference!. In the C phase the structure is quite stiff due to intersections of the surface, and there are fewer microscopic configurations deviating from the MF solution coming from the minimization of Eq. ~3.5!. The effect of broadening of the interface is weaker in the C phase than in the other structures and consequently S/ r s is the largest as shown in Fig. 6. In the special case of the C phase it is clear from Eq. ~2.6! that the surface would intersect, therefore, in the computation of S/ r s we have subtracted the volume occupied along the lines of intersection, since otherwise it would be counted twice. The value of Sd for the considered structures is known ~Sec. II!, and d is determined by the wave number corresponding to bifurcation, k b , so that for the L and P phases d52 p /k b , for D and C phases d52 p )/k b and for G phase d52 p &/k b ~see Fig. 7!. In our case the length unit is l and should be identified with the length of the surfactant ~;25 Å!. From this estimate we find that d ranges from ;100 Å to ;200 Å. The broadening of the surfactant profiles, due to the microscopic configurations39 taken into account via Eqs. ~3.1! and ~3.2!, occur on the small length scale.39 It is set by the size of the calculated f~r! profile. It gives at most 20% ‘‘absorption’’ of the surface area. On the other hand, the area absorption in microemulsions in the typical system can be as large as 80% due to the long wavelength undulations studied by Helfrich40 for the lamellar phase. These fluctuations diverge logarithmically with the system size. In fact they can be included in the model if we go beyond the mean-field description of the Landau–Ginzburg model @Eq. ~3.3!#. Downloaded 23 Feb 2006 to 213.135.39.26. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp 3214 J. Chem. Phys., Vol. 110, No. 6, 8 February 1999 In concluding remarks we should mention limitations of our approach. First, we only considered a selection of structures, and it is possible that some other, possibly more complicated structures stabilize as well and the phase diagram is even more complicated and rich. The second limitation is the bifurcation analysis, which can only be applied close to the continuous transition between microemulsion and the cubic phases. The full phase diagram of the model can only be determined numerically in the future work. Finally, we use the LG, mean-fieldlike approximation, in which fluctuations are neglected. It is possible that the fluctuation-induced weakly first-order phase transition can occur instead of the continuous transition, as found in some models of such systems.41–43 The effect of fluctuations in the simplified version26 of the present model has been partially taken into account in Ref. 43 and indeed, the fluctuation-induced firstorder transition was found. The D phase, found previously within the bifurcation analysis, however, remained stable. It is hence quite probable that in the present model the bicontinuous cubic phases remain stable beyond the LG approximation in the narrow window of surfactant volume fractions. ACKNOWLEDGMENTS We would like to thank Dr. Góźdź for many helpful discussions. This work was partially supported by the KBN Grant No. 2P03B12516 and the Maria Skłodowska Curie Joint Fund II. V. Luzzati and P. A. Spegt, Nature ~London! 215, 701 ~1967!; V. Luzzati, A. Tardieu, and T. Gulik-Krzywicki, ibid. 217, 1028 ~1968!; V. Luzzati, T. Gulik-Krzywicki, and A. Tardieu, ibid. 218, 103 ~1968!; V. Luzzati, A. Tardieu, T. Gulik-Krzywicki, E. Rivas, and F. Reiss-Husson, ibid. 220, 485 ~1968!. 2 K. Larsson, Z. Phys. Chem. ~Munich! 56, 173 ~1973!. 3 S. T. Hyde, S. Anderssonn, B. Ericsson, and K. Larsson, Z. Kristallogr. 168, 213 ~1984!. 4 S. Andersson, S. T. Hyde, K. Larssonn, and S. Lidin, Chem. Rev. 88, 221 ~1988!. 5 S. T. HYde, J. Phys. Chem. 93, 1458 ~1989!. 6 A. H. Schoen, Infinite periodic minimal surfaces without selfintersections, NASA Technical Note No. D-5541 ~1970!. 7 J. Charvolin and J. F. Sadoc, J. Phys.~Paris! 48, 1559 ~1987!. 8 P. Barois, D. Eidam, and S. T. Hyde, J. Phys. ~Paris!, Colloq. 51, C7-25 ~1990!. 9 W. Góźdź, and R. Hołyst, Macromol. Theory Simul. 5, 321 ~1996!. 10 W. Góźdź and R. Hołyst, Phys. Rev. Lett. 76, 2726 ~1996!. 1 A. Ciach and R. Hołyst W. Góźdź and R. Hołyst, Phys. Rev. E 54, 5012 ~1996!. G. Arvidson, I. Brentel, A. Khan, G. Lindblom, and K. Fontell, Eur. J. Biochem. 152, 753 ~1985!. 13 K. Larsson, J. Phys. Chem. 93, 7304 ~1989!. 14 M. Clerck, A. M. Levelut, and J. F. Sadoc, J. Phys. ~Paris!, Colloq. 51, C7 ~1990!. 15 D. C. Turner, Z-G Wang, S. M. Gruner, D. A. Mannock, and R. N. McElhaney, J. Phys. II France 2, 2039 ~1992!. 16 R. G. Larson, J. Phys. II France 6, 1441 ~1996!. 17 D. M. Anderson and H. Wennerström, J. Phys. Chem. 94, 8683 ~1990!. 18 T. Landh, J. Phys. Chem. 98, 8453 ~1994!. 19 M. Kahlweit, R. Strey, and P. Firman, J. Phys. Chem. 90, 671 ~1986!. 20 P. Ström and D. M. Anderson, Langmuir 8, 691 ~1992!. 21 P. Ström, J. Colloid Interface Sci. 154, 184 ~1992!. 22 P. Puvvada, S. B. Qadri, and B. R. Ratna, Langmuir 10, 2972 ~1994!. 23 P. J. Maddaford and C. Toprakcioglu, Langmuir 9, 2868 ~1993!. 24 B. Lindman, K. Shinoda, U. Olsson, D. Anderson, D. Karlström, and H. Wenneström, Colloids Surf. 38, 205 ~1989!. 25 ~a! A. Ciach, J. Chem. Phys. 104, 2376 ~1996!; ~b! A. Ciach, Phys. Rev. E 56, 1954 ~1997!. 26 ~a! A. Ciach, J. S. Ho” ye, and G. Stell, J. Phys. A 21, L777 ~1988!; ~b! A. Ciach and J. S. Ho” ye, J. Chem. Phys. 90, 1222 ~1989!; ~c! A. Ciach, J. Chem. Phys. 96, 1399 ~1992!. 27 J. N. Israelachvili, S. Marcelja, and R. Horn, Rev. Biophys. 13, 121 ~1980!. 28 J. N. Israelachvili, Intermolecular Interactions And Surface Forces ~Academic, New York, 1985!. 29 S. T. Hyde, J. Phys. ~Paris!, Colloq. 51, C7 ~1990!. 30 A. L. Mackay, Proc. R. Soc. London, Ser. A 442, 47 ~1993!. 31 A. L. Mackay, Chem. Phys. Lett. 221, 317 ~1994!. 32 A. L. Mackay, Curr. Sci. 69, 151 ~1995!. 33 H. G. von Schnering and R. Nesper, Z. Phys. B 83, 407 ~1991!. 34 J. Klinowski, A. L. Mackay, and H. Terrones, Philos. Trans. R. Soc. London, Ser. A 354, 1975 ~1996!. 35 C. A. Lambert, R. H. Radzilowski, and L. E. Thomas, Philos. Trans. R. Soc. London, Ser. A 354, 2009 ~1996!. 36 G. Gompper and M. Schick, Phys. Rev. Lett. 65, 1116 ~1990!; G. Gompper and M. Schick, Phase Transitions and Critical Phenomena, 1st ed. ~Academic, New York, 1994!, Vol. 16. 37 G. Gompper and S. Klein, J. Phys. II France 2, 1725 ~1992!; G. Gompper and U. Schwarz, Z. Phys. B 97, 233 ~1995!. 38 J. N. Israelachvili, D. J. Mitchell, and B. W. Ninham, J. Chem. Soc., Faraday Trans. 2 72, 1525 ~1976!; S. T. Hyde, J. Phys. Chem. 93, 1458 ~1989!. 39 S. Rowlinson and B. Widom, Molecular Theory of Capillarity ~Clarendon, Oxford, 1982!, p. 64. 40 W. Helfrich, Liquids at Interfaces, Les Houches, session XLVIII, edited by J. Charvolin, J. F. Joanny, and J. Zinn-Justin ~1988!. 41 S. A. Brazovskii, Sov. Phys. JETP 41, 85 ~1975!. 42 R. Hołyst and W. T. Góźdź, J. Chem. Phys. 106, 4773 ~1997!. 43 C. Buzano, L R. Evangelista, and A. Palizzola, Phys. Rev. E 56, 770 ~1997!. 11 12 Downloaded 23 Feb 2006 to 213.135.39.26. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp
© Copyright 2026 Paperzz