Langmuir 2007, 23, 12827-12834
12827
Microstructure and Stability of a Lamellar Liquid Crystalline and Gel Phase Formed by a Polyglycerol Ester Mixture in Dilute Aqueous Solution N. Duerr-Auster,* J. Kohlbrecher,† T. Zuercher, R. Gunde, P. Fischer, and E. Windhab Institute of Food Science and Nutrition, ETH Zurich, 8092 Zurich, Switzerland and Laboratory of Neutron Scattering, ETH Zurich & Paul Scherrer Institute, 5232 Villigen PSI, Switzerland ReceiVed July 24, 2007. In Final Form: September 21, 2007 The self-assembly behavior of a commercial mixture of polyglycerol fatty acid esters (PGE) and water is investigated as a function of temperature and surfactant content. The phase diagram of this pseudo-binary mixture was characterized using a combination of cross-polarized light and freeze-fracture electron microscopy (cryo-SEM), X-ray diffraction (XRD), small-angle neutron scattering (SANS), and differential scanning calorimetry (DSC). Our experiments show that the morphology of the supramolecular aggregates is lamellar and present in the form of a continuous or dispersed phase (multilamellar vesicles) depending on the water content of the system. Under the effect of temperature, the shortand long-range order of the bimolecular layers successively changes from a biphasic surfactant dispersion to a lamellar liquid-crystalline (LR) and a stable lamellar gel phase (Lβ) upon cooling; this transition is found to be irreversible. Formation of the lamellar aggregates can be related to the average molecular structure and shape factor of PGE. The stability of the resulting gel phase (Lβ) appears to be due to the presence of small amounts of unreacted ionic cosurfactant, namely, fatty acid soaps, in this per se nonionic commercial mixture.
Introduction Characterization of self-assembled supramolecular aggregates formed by surfactants in water has been of considerable interest for both fundamental and applied research over the past few decades.1-3 To date, many aggregate structures with various morphologies have been identified, among which lyotropic liquid crystalline phases represent an important group of nanostructured materials. Such liquid crystalline phases and dispersions thereof find a multitude of applications in food, cosmetic, and pharmaceutical products4-7 and have recently attracted a lot of attention due to the growing need to create biocompatible material templates for encapsulation purposes.8,9 For such applications there is a rising interest to use commercially available surfactants with well-defined physicochemical properties. However, many commercial surfactant products are complex mixtures, and it is a well-established fact that surfactant mixtures can exhibit substantially different properties than single surfactants.10 Furthermore, these mixtures may contain not only different kinds * To whom correspondence should be addressed. Phone: +41 (0)44 632 75 94. Fax: +41 (0)44 632 11 55. E-mail:
[email protected]. ethz.ch. † Laboratory of Neutron Scattering (PSI), Switzerland. (1) Small, D. M. Handbook of Lipid Reserach: The Physical Chemistry of Lipids; Plenum Press: New York, 1986. (2) Laughlin, R. G. The Aqueous Phase BehaVior of Surfactants; Academic Press: London, 1994. (3) Holmberg, K. B.; Jo¨nsson, B.; Kronberg B. Lindman B. Surfactants and Polymers in Aqueous Solution; J. Wiley & Sons Ltd.: England 2002. (4) Larsson, K. In Lipids-Molecular Organization, Physical Function and Technical Applications; The Oily Press: Dundee, U.K., 1994. (5) Krog, N. J. In Food Emulsions; Friberg, S. E., Larsson, K., Eds.; Marcel Dekker Inc.: New York 1997. (6) Heertje, I.; Roijers, E. C.; Hendrickx, H. A. Lebensm.-Wiss. Technol. 1998, 31, 387. (7) Kumar, T. N.; Sastry, Y. S. R.; Lakshminarayana, G. J. Am. Oil. Chem. Soc. 1989, 66, 153. (8) Uchegbu, I. F.; Vyas, S. P.; Ijeoma, F.; Suresh, P. Int. J. Pharm. 1998, 172, 33. (9) Walde, P. In Encyclopedia of Nanoscience and Nanotechnology; Nalwa, H. S., Ed.; American Scientific Publishers: California, 2004. (10) Abe, M.; Scamehorn, J. F. Mixed Surfactant Systems; Marcel Dekker: New York, 2005.
of surface-active ingredients but also a number of unreacted components, which are generally considered as impurities and may have a considerable impact on the phase behavior. Hence, it is necessary to carry out comprehensive studies not only about the composition and functionality of the main components but also about the role of minor ingredients in the phase formation and stability of surfactant mixtures. This context has made lipid-derived surfactants, especially monoacylglycerols, a focus of attention over the past few years.6,11-15 The lyotropic phase behavior and phase stability of different monoglycerol fatty acid esters in aqueous solution is very well-known today. Another representative class of lipidderived surfactants, which have received less attention with respect to fundamental studies, is polyglycerol fatty acid esters (PGE). Similar to polyoxyethylene alkyl ethers, PGE consist of a polymeric hydrophilic head group of n glycerol units and a hydrophobic tail group of alkyl chains. The process of synthesis of commercial products of PGE yields unrefined mixtures including molecules with a varying degrees of polymerization and esterification16,17 and trace amounts of unreacted glycerol and fatty acid soaps.5 At present, the phase behavior of some commercial tetraglycerol,5 pentaglycerol,18 and decaglycerol fatty acid esters19,20 and the surfactant properties of some purified di(11) Krog, N.; Larsson, K. Chem. Phys. Lipids 1968, 2, 129. (12) Vauthey, S.; Milo, C.; Frossard, P.; Garti, N.; Leser, M. E.; Watzke, H. J. J. Agric. Food Chem. 2000, 48, 4808. (13) Mezzenga, R.; Meyer, C.; Servais, C.; Romoscanu, A. I.; Sagalowicz, L.; Hayard, R. C. Langmuir 2005, 21, 3322. (14) Pitzalis, P.; Monduzzi, M.; Krog, N.; Larsson, H.; Ljusberg-Wahlen, H.; Nylander, T. Langmuir 2000, 16, 6358. (15) de Campo, L.; Yaghmur, A.; Sagalowicz, L.; Leser, M. E.; Watzke, H.; Glatter, O. Langmuir 2004, 20, 5254. (16) Andersen, T.; Holm, A.; Skuland, I. L.; Trones, R.; Greibrokk, T. J. Sep. Sci. 2003, 26, 1133. (17) Meulenaer, B. D.; Van Royen, G.; Vanhouette, B.; Huyghebaert, A. J. Chromatogr. A 2000, 896, 239. (18) Izquierdo, P.; Acharya, D. P.; Hirayama, K.; Asaoka, H.; Ihara, K.; Tsunehiro, T.; Shimada, Y.; Asano, Y.; Kokubo, S.; Kunieda, H. J. Dispersion Sci. Technol. 2006, 27, 99. (19) Ai, S.; Ishitobi, M J. Colloid Interface Sci. 2006, 296, 685. (20) Ishitobi, M.; Kunieda, H. Colloid Polym. Sci. 2000, 278, 899.
10.1021/la702242v CCC: $37.00 © 2007 American Chemical Society Published on Web 11/22/2007
12828 Langmuir, Vol. 23, No. 26, 2007
Duerr-Auster et al. meter with a glass electrode. From the inflection point of the titration curve the concentration of fatty acid soap, C, contained in the dry surfactant mixture was calculated in mass percent as C ) 100(VNM)/m
Figure 1. Schematic molecular structure of a triglycerol diester of hexadecanoic and octadecanoic acid.
to pentaglycerol fatty acid esters have been reported.21 However, in those systematic approaches the phase behavior of a commercial triglycerol fatty acid ester and the influence of impurities, especially of small amounts of charged components such as fatty acid soaps, has not yet been discussed in detail. The objective of the present work was to characterize the phase behavior of a specific unrefined commercial PGE with a predominant amount of a triglycerol ester of hexadecanoic and octadecanoic acid in water considering a concentration and temperature range typical for industrial application. In particular, we focus on the influence of small amounts of unreacted fatty acid soap, namely, hexadecanoate and octadecanoate, on the phase formation and stabilization in the dilute regime by combining differential scanning calorimetry (DSC), polarized light microscopy, and freeze-fracture scanning electron microscopy (cryo-SEM). The characteristic parameters of the phases are analyzed by X-ray diffraction (XRD) and small-angle neutron scattering (SANS). The investigation aims at discussing the molecular mechanisms leading to the phase formation of an unrefined commercial PGE mixture and describing the important role played by two minor components in the stabilization of this phase. Experimental Section Materials. PGE 55 was a generous gift from Danisco (Brabrand, Denmark) and used as received. This commercial-grade polyglycerol ester of fatty acids (PGE) is unrefined; thus, a surfactant mixture contains traces of unreacted glycerol and unreacted fatty acid soaps. The hydrocarbon chains consist predominantly of octadecanoic (54 wt %) and hexadecanoic acid (45 wt %). The hydrophilic head groups have a degree of polymerization of two (32 wt %), three (54 wt %), and four (14 wt %). An example of one possible molecular structure is depicted in Figure 1. The water used was purified with a Milli-Q Biocel (Millipore) pure water system. For the preparation of samples for neutron scattering experiments, pure deuterium oxide (Sigma Aldrich, Switzerland) instead of water was used to increase contrast. For the titration experiments a 4 M solution of hydrochloric acid (Fluka, Switzerland) was used and diluted to the appropriate concentrations of use; for precipitation experiments calcium chloride (Fluka, Switzerland) was employed. Sample Preparation. Samples of different surfactant composition between 0.6 and 12.5 wt % were prepared by weighing appropriate amounts of surfactant and water into clean and dry glass containers. These were then gently mixed by heating in a water bath to a temperature of 80 °C for least 10 min after the appearance of turbidity. The samples were then cooled to room temperature and equilibrated for 24 h prior to analysis. Titration Experiments and pH Measurements. The titration curve of dilute aqueous solutions was obtained by first adding a defined amount of 1 M hydrochloric acid solution, second letting the samples equilibrate for 24 h at room temperature, and third measuring the pH using a previously calibrated Metrohm 744 pH (21) Kato, T.; Nakamura, T.; Yamashita, M.; Kawaguchi, M.; Itoh, T. J. Surfactants Deterg. 2003, 6, 331.
(1)
where V is the volume of hydrochloric acid at the inflection point, N the molarity of the titer, M the average molar mass of octadecanoate and hexadecanoate, and m the absolute mass of PGE in solution. Differential Scanning Calorimetry (DSC). DSC measurements were performed to quantify the temperature-dependent phase behavior. A Perkin-Elmer (model DSC 7) calorimeter was used in the temperature range of 10-80 °C. Aluminum crucibles were filled with 20 µL of solution, and an empty crucible was used as a reference. Heating-cooling-heating cycles were performed at a heating rate of 10 °C/min. Polarized Light Microscopy. The optical properties (birefringent, nonbirefrigent) of the material were identified by polarized light microscopy. The microscope used was a Leica (model Leica DM IRB), and the images were acquired using a digital camera (model DFC 320). Cryo Scanning Electron Microscopy (cryo-SEM). Freezefracture SEM was performed for selected dilutions to visualize the sample morphology. For fracturing, solution was filled into an aluminum tube of 3.1 mm diameter and immediately plunged into liquid ethane before transferring into a liquid nitrogen bath where it was fractured. After fracturing, the sample was etched and coated with a 6 nm thick film of platinum powder using a high-vacuum sputter coater unit (model MED 010). For imaging, such a freezefractured sample was transferred into a Hitachi SEM (model S-900 FEG) using a gattan holder, and the images were recorded at magnifications up to 20 000 times. Wide-Angle X-ray Powder Diffractometry (XRD). The shortrange order of PGE in solution was investigated by X-ray powder diffractometry. Mixtures of various concentrations of PGE were prepared as described above and filled into mark tubes with a diameter of 0.7 mm (Hilgenberg, Germany). The diffraction patterns collected at room temperature were recorded using a STOE (model STAPI P) diffractometer in the 2θ range 10-40° for Co KR1 radiation. Temperature-controlled measurements were performed with an Oxford single-crystal diffractometer (model Xcalibur) with a cryojet in the same 2θ range using Mo KR radiation. Raw data were collected on an area detector and integrated using the Fit2D software22 to convert to intensity versus 2θ. For the quantitative analysis of the unit cell dimensions, eqs 2 (with Θ/2 ) θ) and 3 were used. [We chose to use the common notation in X-ray diffractometry for XRD experiment, where 2θ is the angle comprised between the incident and scattered beam; for SANS experiments this angle is denoted as Θ.] Small Angle Neutron Scattering (SANS). SANS experiments were performed to confirm the phase behavior and quantify the lattice constants of the individual phases. The SANS I beamline at the Paul Scherer Institute (PSI Villigen, Switzerland) was used for all experiments.23 Samples were filled into custom-built circular glass cells (1 mm thickness; 500 µL) and thermostated using a circulating water bath regulated with a precision of 0.1 °C. Data was collected for temperatures between 25 and 59 °C. The neutron wavelength (λ) was fixed at 0.8 nm, and the time-averaged data were collected on a two-dimensional 3He detector at distances of 2, 6, and 18 m resulting in an effective q range from 0.03 to 2.5 nm-1, which corresponds to a resolvable linear real space from 2.5 to 250 nm. All measurements were carried out with maximum collimation (18 m) to avoid any broadening of the Bragg peaks of the lamellar structures due to beam divergence. The momentum transfer, q, is defined as (22) Hammersley, A. P. ESRF Internal Report, ESRF98HA01T; ESRF: Grenoble, France, 1986. (23) Kohlbrecher, J.; Wagner, W. J. Appl. Crystallogr. 2000, 33, 804.
Polyglycerol Fatty Acid Ester Phase Formation q ) (4π/λ) sin Θ/2
Langmuir, Vol. 23, No. 26, 2007 12829 (2)
where Θ is the scattering angle between the incident and scattered neutron beam. The two-dimensional raw data were corrected for background, transmission, and detector efficiency according to standard procedures24 and radially averaged to yield the onedimensional intensity distribution, I(q), using the BerSANS software package.25 The periodic repeat distance, d, was calculated from the maximum scattering intensity of the first-order peak at, qmax, using eq 3 d ) 2π/qmax
(3)
Results and Discussion Macroscopic and Microscopic Appearance. During the standard sample preparation procedure, it was observed that the macroscopic appearance of all solutions changed irreversibly when the samples were heated to 80 °C and subsequently cooled to room temperature; this phenomenon is shown photographically in Figure 2. When dry PGE was added to water at room temperature, the surfactant powder remained in suspension and two separate phases could be discerned by the naked eye. With time, the volume occupied by the surfactant phase increased; however, over the period of several weeks, the solution remained inhomogeneous as depicted in Figure 2.I. When subjected to heat treatment, all solutions became homogeneously turbid above a temperature of 58 °C, and this opacity remained after cooling (Figure 2.II). For samples with a PGE content exceeding 2 wt %, no phase separation could be detected by visual means during an observation time of 4 weeks when stored at room temperature (Figure 2.III). For highly dilute solutions, slight sedimentation was noted after 1 week but the overall aspect of the samples remained turbid. The morphology of the surfactant aggregates causing the appearance of turbidity were probed by different microscopic techniques, and representative micrographs thereof are depicted in Figure 3. Figure 3a shows a cryo-SEM picture of a 5 wt % PGE dispersion prior to any heat treatment. One can clearly discern randomly distributed surfactant aggregates (light gray) mostly sized above 20 µm which are separated by water (dark gray). The arrow in Figure 3a indicates that these dispersed aggregates are characterized by a layered microstructure. By analogy to the macroscopic observations, the same bulk solution was subjected to heat treatment and imaged again. Figure 3b depicts a cryo-SEM micrograph of a 5 wt % PGE solution after heat treatment in which one can again clearly distinguish two different materials, namely, water (dark gray) and lipid (light gray). As opposed to the non-preheated solution, both water and lipid are now homogeneously distributed and form a continuous, ordered matrix consisting of alternating layers with a distance varying between 200 and 800 nm. In some areas the lipid layers form closed, spherical to ellipsoidal shells with a diameter of approximately 5 µm. With increasing surfactant concentration a similar but much denser microstructure was observed. Highly dilute solutions, i.e., solution with a surfactant content below 5 wt %, were imaged at room temperature using a light microscope with cross-polarized light. The resulting micrographs typically featured strongly birefringent spherical particles of approximate diameter ranging between 1 and 5 µm. Furthermore, the particles displayed an interference pattern consisting of two intersecting black bars that form a Maltese cross-like pattern which, in surfactant dispersions, are indicative for multilamellar vesicles. The macroscopic and microscopic aspect of PGE-water mixtures (24) Strunz, P. Saroun, J., Keiderling, U., Wiedenmann, A.; Przenioslo, R. J. Appl. Crystallogr. 2000, 33, 829-833. (25) Keiderling, U. Appl. Phys. A 2002, 74, 1455.
Figure 2. Macroscopic appearance of (I) a non-preheated dispersion of PGE, (II) a heated and cooled solution of PGE, and (III) a heated, cooled, and aged solution at room temperature.
both indicated that the surfactant is not water soluble at room temperature but forms aggregates, which are more or less homogeneously distributed depending on the temperature history. The dimensions of all aggregates, whether present as dispersed or continuous structure, seem to exceed the wavelength of white light and thus cause the appearance turbidity. Temperature-Dependent Behavior. The behavior of the surfactant solution was further analyzed as a function of temperature using differential scanning calorimetry (DSC) and temperature-controlled X-ray diffraction (XRD). Typical thermograms resulting from calorimetric measurements in the heating direction are depicted in Figure 4. When the non-preheated surfactant dispersion was heated from 20 to 80 °C (Figure 4a, first upscan) a fairly sharp endothermic transition appeared at a temperature of 56 °C (the peak temperature will be used to describe the transition unless otherwise specified) with a corresponding enthalpy change (∆Ηh) of 3.8 J/g as calculated from the peak area. This transition temperature matched with the appearance of macroscopic turbidity. During the second and subsequent heating scans performed immediately after cooling, this main transition was shifted to a slightly higher and significantly broader endothermic peak at 58 °C (Tc2) with a smaller ∆Ηh of 2.82 J/g. Furthermore, during the second heating scan an additional, minor endothermic transition appeared at a temperature of 44 °C (Figure 4a, second upscan (Tc1)) characterized by a corresponding enthalpy change of 0.33 J/g. When the sample was cooled, both transition temperatures were shifted to about 10 °C lower values due to supercooling. Under the experimental conditions, the data collected from the second and subsequent heating-cooling cycles showed very good reproducibility. In order to probe the stability of the preheated PGE solutions as a function of time, filled crucibles were stored for 4 weeks at room temperature and the DSC experiments were repeated under identical conditions. A representative example of the resulting thermograms in the heating direction is depicted in Figure 4a (third upscan), wherein one can clearly observe a broad main transition at 58 °C matching with the second upscan where the peak area (∆Ηh) amounts to 3.2 J/g. A striking feature of the third upscan is that, in contrast to the second upscan, the minor pretransition previously located at 44 °C has almost vanished and the thermogram only shows evidence of an extremely small peak. This peak has a corresponding ∆Ηh of 0.07 J/g. Instead, the enthalpy change of the main transition of the aged sample (third upscan) corresponds to the sum of enthalpy changes measured for the pre- and main transition of the preheated, fresh sample (second upscan). In order to interpret the calorimetric data as measured by DSC with respect to structural changes of the surfactant aggregates associated with the heat treatment, static X-ray diffraction (XRD) experiments were carried out for solutions of similar concentration
12830 Langmuir, Vol. 23, No. 26, 2007
Duerr-Auster et al.
Figure 4. (a) DSC heating traces of the first, second, and third heating step of a 10 wt % PGE solution. The first and second heating cycles were successively performed on the same day. The third heating cycle was performed after 4 weeks of storage at room temperature. The inset shows the visual appearance corresponding to DSC measurements of (I) non-heated sample, (II) heated sample, and (III) heated sample after 4 weeks storage at room temperature. (b) XRD profile of a heated and not-heated (crystalline dispersion) PGE solution. The difference in the peak area results from a concentration difference of the samples, i.e., the not preheated sample had higher concentration. The inset shows the XRD profile of a 10 wt % PGE solution at 25 and 60 °C. For the sake of clarity, the scattering profiles have been multiplied by an arbitrary factor.
Figure 3. (a) Cryo-scanning electron micrograph of a non-preheated 5 wt % PGE solution. The scale bar corresponds to 5 µm. (b) Cryoscanning electron micrograph of a 5 wt % PGE solution after one heating-cooling cycle. The solution temperature prior to freezing was 65 °C. The scale bar corresponds to 5 µm. (c) Polarized light micrograph of a dilute a 2.5 wt % PGE solution at room temperature featuring characteristic Maltese crosses. The scale bar corresponds to 20 µm.
and at different temperatures ranging between 25 and 65 °C. The 2θ range used for XRD experiments allows one to resolve crystalline structures in the order of 2-10 Å and thus gives insight into the degree and type of crystalline orientation adopted by the hydrophobic alkyl chains of the surfactant lattice (shortrange order). Typical examples of the observed diffraction profiles are depicted in Figure 4b. The main figure shows the profiles recorded at room temperature for a non-preheated (PGE suspension) and a preheated (PGE solution) sample, which
corresponds to the samples used for the first and second upscan in the DSC experiments, respectively. Both XRD curves show a single, sharp wide-angle reflection corresponding to a repeat spacing of 4.2 Å, which for lipid systems is indicative of nonspecific crystalline hexagonal chain packing.1 It should be noted that the difference in the peak area of the XRD profiles results from the concentration difference of the measured samples. The constancy in position of the wide-angle reflection detected before and after heating seems to point out that PGE has a characteristic short-range order and that heating and subsequent cooling does not affect the lattice structure adopted by the alkyl chains; in other words, PGE is not polymorphic. However, when the environmental temperature of a preheated solution is increased above the main transition temperature of 58 °C (Tc2), the sharp wide-angle reflection fades and gives way to a broad, diffuse band (inset Figure 4b) characteristic for a disordered, fluid state of the alkyl chains. The main endothermic transition observed in DSC and the fading peak seen in XRD thus can be interpreted as the melting of the hydrophobic alkyl chains. Between 25 and 58 °C, no significant shifts in the diffraction profile could be detected. SANS Experiments. In order to analyze the degree of longrange order and determine the morphology of the supramolecular surfactant aggregates as a function of concentration and tem-
Polyglycerol Fatty Acid Ester Phase Formation
Figure 5. (a) SANS data obtained for 1, 5, 10, and 12.5 wt % PGE in D2O at 25 °C. Increasing concentration supports formation of lamellae with a higher degree of order. The best fit using a lamellar model with long-range order (solid lines) shows good agreement with the data in the middle and higher q range and is representative for a multilamellar phase. (b) SANS data for 12.5 wt % PGE solution measured at 25, 46, and 59 °C. The inset shows the Kratky plot of the same data (46 and 59 °C) emphasizing that the order is least in the LR phase. To better visualize the data and fits to the data, profiles shown in plots a and b have been multiplied by arbitrary rescaling factors.
perature, small-angle neutron scattering experiments were carried out at PSI (Villigen, Switzerland) for solutions with concentrations between 1 and 12.5 wt % and in a temperature range between 25 and 59 °C. With respect to concentration, two types of SANS profiles representative for the different morphologies were observed, and examples thereof are depicted in Figure 5a. Below a surfactant concentration of 5 wt % PGE, the profiles were characterized by the absence of Bragg scattering, possibly due to large interlayer spacings. Moreover, in the low-q region, the intensity exhibits a q-2 dependence, indicative of large-sized lamellar structures,26 which cannot be resolved by SANS; such micrometer-sized multilamellar aggregates have indeed been observed in light microscopy (Figure 3b). Above a critical concentration of 5 wt % PGE, Bragg peaks up to third order and in equidistant position occurred at all temperatures, and these profiles are characteristic for lamellar phases27 which are made (26) Nieh, M.-P.; Harroun, T. A.; Raghunathan, V. A.; Glinka, C. J.; Katsaras, J. Biophys. J. 2004, 86, 2615.
Langmuir, Vol. 23, No. 26, 2007 12831
of an ordered array of “open-ended” bimolecular lipid layers with intercalated water layers. As the concentration was increased above this critical value, the first-order reflection became sharper and there was a greater total number of detectable reflections, indicating a more ordered structure with better defined interlayer spacings. Between the two highest concentrations measured, i.e., 10 and 12.5 wt % PGE, there exists no significant difference in the scattering profile but the intensity maxima (Imax) are progressively shifted to lower q values (corresponding to higher layer distances), which seems to point out that the lamellae experience the same kind of long-range order despite a significant difference in the interlayer spacing. We turn now to the effect of temperature on the degree of order of the surfactant aggregates as probed by SANS. Figure 5b shows the scattering profiles measured for a 12.5 wt % PGE solution at 25, 46, and 59 °C. At this concentration, Bragg scattering could be observed in all profiles, pointing out that there exists long-range order at all investigated temperatures. However, one can observe in Figure 5b that the number of peaks and their sharpness decrease above the main transition temperature (Tc2 )58 °C), and moreover, the profile is progressively shifted to lower q values, which is indicative for less well-defined and larger bilayer spacings, respectively. Both results imply that the degree of order decreases as the temperature is increased. The SANS profiles of samples showing Bragg scattering were analyzed quantitatively to determine the periodic repeat spacing (d) as well as the dimensions of the lipid bilayer (dl) and the water layer (dw) as a function of concentration and temperature. A schematic representation of the relation between these parameters (d, dl, and dw) for a lamellar phase is given by the schematic drawing in Figure 6a. For the interpretation of the following results, it should be noted that the value of d, as extracted from the scattering data, represents the sum of dw and dl. The periodic repeat distance, d, was calculated from the maximum scattering intensity of the first-order peak, qmax, using eq 3. Information about dl was obtained by fitting the scattering profile with a lamellar model with long-range order. The model consists of finite-sized lamellae (dl) with hydrophobic cores and hydrophilic shells along with a thermal disorder parameter for the distribution of the interlamellar spacings (d), which strongly affects the intensity and width of the peaks.28-30 Figure 6b shows the evolution of d as a function of the surfactant volume fractions at 25, 46, and 59 °C. For the calculation of the lipid volume fraction from the concentration in weight percent, a density of 0.84 g/cm3 corresponding to the density of long chain liquid fatty acids1 was assumed. At temperatures below 58 °C and between a concentration of 7.5 and 12.5 wt %, d increases linearly from 38 to 62 nm, thus showing evidence of onedimensional swelling.31-33 With further decrease of the surfactant concentration (5 wt %), the periodic repeat distance increases to about 70 nm and deviates from the linear trend indicating a diminished swelling ability. Raising the temperature to 59 °C causes a net increase of all periodic repeat distances, and this increase, as indicated by the arrows in Figure 6b, is highest at (27) Chapman, D. In The Structure of Lipids by Spectroscopic and X-ray Techniques; Methuen and Co Ltd.: London, 1965. (28) Fru¨hwirth, T.; Fritz, G.; Freiberger, N.; Glatter, O. J. Appl. Crystallogr. 2004. 37, 703. (29) Pabst, G.; Koschuch, R.; Pozo-Navas, B.; Rappolt, M.; Lohner, K.; Laggner, P. J. Appl. Crystallogr. 2003, 36, 1378. (30) Pabst, G.; Rappolt, M.; Amenitsch, H.; Laggner, P. Phys. ReV. E 2000, 62 (3), 4000. (31) Luzzati, V. Biol. Membr. Phys. Facts Funct. 1968, 1, 71. (32) Kunieda, H.; Akahane, A.; Feng, J.; Ishitobi, M. J. Colloid Interface Sci. 1999, 218, 88. (33) Evans, D. F.; Wennerstro¨m, H. In The Colloidal Domain: Where Physics, Chemistry, Biology, and Technology Meet; VCH: New York, 1994.
12832 Langmuir, Vol. 23, No. 26, 2007
Duerr-Auster et al.
Figure 7. Pseudo-binary phase diagram of PGE in water (LR) Lamellar liquid crystal, Lβ ) Lamellar gel, MLV ) Multilamellar vesicles, 1Ψ ) single phase, 2Ψ ) two-phase region).
Figure 6. (a) Schematic representation of the lamellar phase with characteristic long-range order parameters (d, dw, and dl) below and above the main transition temperature (Tc2 ) 58 °C). (b) Variation of the repeat distance, d, as measured by SANS as a function of the inverse lipid volume fraction of the surfactant, 1/φ, at 25, 46, and 59 °C. The corresponding concentration in weight percent is indicated. The solid line represents the linear fit of eq 4 to the experimental data. The slope indicates the resulting bilayer thickness, dl, in nanometers. (1ψ) Single-phase region; (2ψ) two-phase region. (c) Lipid bilayer thickness (dl) as a function of temperature obtained by fully analyzing the SANS scattering curve.
the lowest surfactant concentration and vice versa. At the same time, the lipid bilayer thickness, dl, decreases from an initial value of 4.9 nm at 25 °C to 3.9 nm at 59 °C (Figure 6c) corresponding to a conformational change from the fully extended form of octadecanoic acid34 to its highly disordered, fluid state31 as depicted schematically in Figure 6c. It can be noted that the quantitative information, i.e., the values of the repeat spacings extracted from SANS, differ significantly from the dimensions observed in cryo-SEM. However, the qualitative structural information, i.e., kind of structure, is the same in both cases, and differences in the quantitative values can probably be attributed (34) Tanford, C. In The Hydrophobic Effect: formation of micelles and biological membranes; Wiley: New York, 1973.
to artifacts resulting from the sample preparation procedure of SEM samples. Lyotropic Phase Behavior. By means of microscopy, DSC, XRD, and SANS, and from the visual appearance, a pseudobinary phase diagram of PGE in water as a function of temperature is generated. The diagram presented in Figure 7 only takes into account the structure of the samples subjected to at least one heating-cooling cycle, which appeared to be necessary to provide homogeneous solutions and reversible (equilibrium) experimental conditions; the irreversible transition observed during the first heating cycle shall be discussed in a later section. In Figure 7 one can clearly distinguish four different areas, whichswith respect to concentrationsgroup into two main blocks, i.e., a one-phase (1Ψ) region at higher surfactant concentration and a two-phase (2Ψ) region at lower PGE concentration. As probed by SANS, all phases below the critical concentration of 5 wt % PGE are two-phase regions (2Ψ) and thus always contain a mixture of water and dispersed self-assembled surfactant structure. These aggregates are multilamellar vesicles (MLV) and consist of several alternating bimolecular lipid and water layers, which are arranged in a concentric way to form a soft spherical particle. Examples of such structures could be observed indirectly in polarized light microscopy and directly in cryo-SEM micrographs. Above the critical surfactant concentration of 5 wt %, the solvent (water) can be completely withheld in the PGE matrix and two lamellar, single-phase regions (1Ψ), namely, LR and Lβ, appear. Lamellar phases are made of an ordered array of “open-ended” bimolecular lipid layers with intercalated water layers; examples thereof are depicted in Figure 3b. By raising the temperature at any surfactant concentration above the main transition temperature of 58 °C the lipid bilayer of the lamellar and multilamellar vesicle phase undergoes a phase transition from gel (solid) to liquidcrystalline (fluid) and two different bilayer states, indicated by the index β (solid) or R (fluid), are successively encountered. This event is commonly referred to as “chain melting transition” and was probed by DSC and XRD. If solutions are not equilibrated for a sufficient amount of time (nonequilibrium), an apparent pretransition can be detected by calorimetric measurements, which probably originates from the premelting of partially crystallized alkyl chains. Equilibrated solutions only show evidence of one endothermic peak upon heating; for that reason the equilibrium phase diagram contains information only about the main transition. Within the investigated concentration range, with increasing temperature, the morphology, in other words the kind of longrange order of the self-assembled surfactant aggregate, thus remains similar but only the physical state of the bilayer, indicated by the short-range order, undergoes drastic changes. A similar
Polyglycerol Fatty Acid Ester Phase Formation
Langmuir, Vol. 23, No. 26, 2007 12833
Table 1. Molecular Dimensions of a Diglycerol (DG) and Triglycerol (TG) Fatty Acid Ester of Hexadecanoic and Octadecaoic Acid with One (C16/18) or Two (2C16/18) Alkyl Chains and Resulting Packing Parameter (F) a0 (nm2)a lc (nm)b a0lc (nm3) V0 (nm3)c F ) V0/a0lc (-) DG C16/18 TG C16/18 DG 2C16/18 TG 2C16/18 a
0.2 0.4 0.4 0.4
1.95 1.95 1.95 1.95
0.39 0.78 0.39 0.78
0.485 0.485 0.97 0.97
1.2 0.6 2.5 1.2
From ref 42. b Extracted from SANS. c From ref 38.
behavior has been described by Holstborg et al.35 for a series of diglycerol monoesters for which the supramolecular aggregate structure remains lamellar at all temperatures and over the entire concentration range. Origin of the Phase Behavior. The self-driven assembly of surfactants in solution into aggregated structures generally depends on the balance of intermolecular and interaggregate forces.36 In polar solvents such as water, surfactants with a pronounced hydrophobic part and relatively small head groups have tendency to assemble into liquid crystalline rather than micellar structures.37 Above the chain melting temperature, the ability of a pure surfactant to form a liquid crystalline aggregate with a specific geometry can generally be predicted by the packing parameter F (F ) V/a0lc), which considers the molecular geometry of the amphiphile. For surfactant mixtures, as in our case, the packing parameter represents an average calculated from the weighted F of all molecules present in the mixture.38 For molecules to assemble into flexible to planar bilayers, the average value of the shape factor must be between 0.5 and 1 or close to unity, respectively. The presently investigated nonionic surfactant mixture of PGE contains a predominant amount of triglycerol (54 wt %) and diglycerol (32 wt %) ester of hexadecanoic and octadecanoic acid with an unknown degree of esterification. Values for the head group area a0 occupied by triglycerol and diglycerol fatty acid esters of long-chained lipids have been reported in the literature and correspond to 0.4 and 0.2 nm2, respectively.39 In the liquid state, the lipid bilayer thickness has been determined by SANS and corresponds to 3.9 nm, and onehalf of this value (1.95 nm) can be used as an average approximation for the alkyl chain length, lc, of hexadecanoic and octadecanoic acid. For a triglycerol and diglycerol monoester of fatty acids, the product of a0 and lc corresponds to a value of 0.78 and 0.39 nm3, respectively. The average molecular volume of a saturated hydrocarbon chain with 16 and 18 carbon atoms can be calculated using the equation by Tanford34 and corresponds to 0.485 nm3, and the ratio of V and a0lc thus amounts to 0.6 and 1.2 for a diglycerol and triglycerol, respectively. As summarized in Table 1, if the diglycerol and triglycerol head group is esterified with two alkyl chains, the packing parameter amounts to 2.5 and 1.2, respectively. It follows that, at temperatures above the main chain melting temperature, the PGE mixture containing a predominant amount of molecules with a higher degree of polymerization, i.e., n ) 3, as in our case, and having one or two long hydrocarbon chains (16 or 18 carbon atoms), the packing parameter F is between 0.5 and 1 or close to unity, thus favoring formation of flexible to planar bilayers above the critical (35) Holstborg, J.; Pedersen, B. V.; Krog, N.; Olesen, S. K. Colloids Surf. B: Biointerfaces 1999, 12, 383. (36) Israelachvili, J. In Intermolecular and Surface Forces; Academic Press Ltd.: London, 1991. (37) Tiddy, G. J. Phys. Rep. 1980, 57, 1. (38) Kratzat, K.; Stubenrauch C.; Finkelmann H. Colloid Polym. Sci. 1995, 273, 257. (39) Kawaguchi, M.; Yamamoto, M.; Nakamura, T.; Yamashita, M.; Kato, T. Langmuir 2001, 17, 4677.
aggregation concentration. It seems thus very likely that the reason for the morphology of the phase formed by PGE in water can be explained by its molecular structure. To further verify this hypothesis, future work aims at quantifying the exact head group of the PGE mixture. For PGE mixtures with a much higher degree of esterification and thus a higher degree of hydrophobicity, the packing parameter F is greater than unity, and thus, inverted hexagonal structures, consisting of rod-shaped micelles of indefinite length packed in a hexagonal array, are likely to form at sufficiently high surfactant concentration. Another important point, which requires consideration when discussing the origin of the phase behavior, is the capacity of the lamellae to interact with water despite the nonsolubility of PGE. Similar to mono- and diglycerides, a significant part of the ability to incorporate water in the interlamellar space (swelling) certainly owes to the presence of a large number of unesterified hydroxyl groups present in the polymeric glycerol head group. However, most nonionic lipids, which undergo lyotropic mesomorphism, have a finite capacity to swell, and in the case of pure monoglycerides the equilibrium interlayer spacing, resulting from the force balance between attractive long-range van der Waals forces and osmotic repulsion forces, amounts to approximately 20 Å.1 When considering the large interlayer distances of several hundred Ångstroms reported in the SANS measurements, it seems that additional hydrophilicity and thus interlayer hydration is provided by the presence of electrostatic forces, that is by the presence of small amounts of charged molecules which, in the case of a nonrefined surfactant mixture, are most probably unreacted fatty acid soaps, namely, hexadecaoate and octadecanoate. Indeed, for other surfactants systems3,40 it is a wellknown fact that such large interlayer distances are associated with the presence of ionized fatty acids in the bilayer leaflet. It thus seems that besides the average molecular geometry of the surfactant mixture, another reason for the existence of the lamellar liquid crystalline and gel phase at such high degrees of dilution is the presence of small amounts of ionized molecules in the bilayer leaflet which enable a significantly higher and more homogeneous interlayer hydration, i.e., long-range order. To determine the functionality of these small amounts of charged co-surfactant, we first quantified the amount of ionized fatty acid present in the PGE mixture and then investigated by visual inspection the effects of charge on the formation and stability of the lamellar phases. Figure 8a shows a typical potentiometric titration curve obtained from a 1 wt % PGE (previously submitted to one heating-cooling cycle) and a 0.5 mM Na-Palmitate solution. The initial pH values measured for both solutions amount to 8.7 and 9.5, respectively, and imply the presence of alkaline soaps in solution which, considering the fatty acid composition of the surfactant, must have a chain length of 16 and 18 carbon atoms. The nature and shape of both curves shows a similar s-shaped trend with an inflection point at a corresponding pH value of 5.5, which is in good agreement with the pKa reported for octadecanoate.41 From the inflection point of the curve, the weight fraction of fatty acid soap contained in dry PGE was calculated according to eq 1 and corresponded to 1.4 wt % (or 0.5 mM for a 1 wt % PGE solution). Due to the high initial pH of PGE solutions, it is very likely that under the effect of heat the amount of soap encountered in the solution increases by hydrolytic cleavage of the ester bonds. This reaction could account for the irreversible transition from a heterogeneous dispersion to a homogeneous solution observed during the first heating cycle. (40) Krog, N.; Borip, A. P. J. Food Agric. Sci. 1973, 24, 691. (41) Rajam, S.; Heywood, B. R.; Walker, J. B.; Mann, S. J. Chem. Soc., Faraday Trans. 1991, 87, 727.
12834 Langmuir, Vol. 23, No. 26, 2007
Duerr-Auster et al.
hydration must be related to an increased hydrophilic hydration force, probably involving stronger electrostatic interactions. The experiment described above clearly shows that complexation of the fatty acid soaps suppresses the irreversible transition. Thus, in the dilute domain, trace amounts of ionized fatty acids seem to have two functions: first, they participate in a chemical reaction, which generates enough charged molecules to incorporate excess water in the interlamellar space; second, they stabilize the interlamellar water content at room temperature.
Conclusions
Figure 8. (a) Titration curve of a 1 wt % PGE solution in comparison to a 0.5 mM Na Palmitate at room temperature. (b) Visual appearance of a 1 wt % PGE solution prepared with 0, 0.015, and 1 wt % CaCl2 heated to 80 °C and cooled back to room temperature.
The phase formation and stability with and without the presence of charge has been investigated by adding a defined amount of calcium chloride to the PGE solution prior to and after heating, and a photograph of the visual appearance of such samples at room temperature is shown in Figure 8b. It can be clearly seen that adding an amount of divalent salt equivalent or higher to the measured concentration of fatty acid soap prior to heating leads to precipitation of the surfactant aggregate. A similar phase separation can be observed if salt or acid is added to the solution at room temperature after heating. The fact that the equilibrium of the lamellar gel phase can be easily disturbed by addition of small amounts of counterions seems to confirm the hypothesis that for the intrinsically nonionic PGE the main reason for the stability of the lamellar gel phase in excess water at room temperature is the presence of small amounts charges. In addition, it seems that charged fatty acid soaps not only stabilize the gel phase but also modify the pH of the solution such that, at higher temperatures, conditions for hydrolysis are given, and the total amount of charged “co-surfactant” in the lamellae possibly increases during heat treatment. Most likely, this reaction accounts for the irreversible transition observed during the first heatingcooling cycle depicted in Figure 2I and 2II, during which a poorly hydrated, crystalline suspension, commonly referred to as ‘coagel’,42,43 transforms into a fully hydrated, homogeneously opaque gel phase. Obviously, the difference in the degree of (42) Palma, S.; Manzo, R. H.; Allemandi, D.; Fratoni, L.; Lo Nostro, P. Langmuir 2002, 18, 9219. (43) Muga, A.; Casal, H. L. Langmuir 1990, 94, 7265.
We investigated the microstructure and stability of the lyotropic liquid crystalline phases formed by a mixture of a polyglycerol ester of palmitic and stearic acid as a function of temperature in dilute aqueous solution using a combination of DSC, polarized light, and cryo-freeze-fracture electron microscopy as well as different scattering techniques such as XRD and SANS. The combination of these techniques has allowed a detailed description of the morphology of the supramolecular surfactant aggregate and characterization of the dimensions associated with the characteristic long- and short-range order as well as changes thereof under the effect of heating or cooling. Microscopy and small-angle neutron scattering data showed that in the entire investigated concentration range the aggregate structure presents a lamellar morphology which can exist in the form of a single continuous phase or as dispersed multilamellar vesicles in solution. Considerations of the average molecular packing parameter confirmed that the PGE mixture is predisposed to form a flexible to planar bilayer in polar solvents. Using DSC we have shown that the lamellar aggregates can be present in three different physical states, namely, coagel, liquid crystalline (LR), or gel (Lβ), depending on the thermal history of the sample. When PGE is dispersed in water at room temperature it remains as a biphasic surfactant dispersion (coagel) which, when heated and cooled above the main transition temperature, successively forms a homogeneous lamellar liquid crystalline and a lamellar gel phase. Contrary to monoglyceride esters of similar chain length, this presently investigated gel phase appears to be stable with time. Two factors were identified to be of importance in the preservation of its structural order: (i) PGE was shown not to be polymorphic and there is consequently no alteration of the hydrophobic chain packing over time and (ii) the per se nonionic surfactant mixture was shown to contain a small amount of charged co-surfactant, namely, hexadecanoate and octadecanoate, which significantly influences and probably preserves the longrange order at various temperatures. Acknowledgment. ETH Zurich is acknowledged for financial support. The authors thank Bruno de Meulenaer (University Ghent) for the chemical characterization of PGE head groups. We thank Stephan Handschin for his help with light microscopy. We thank Martin Colussi (ETH Zurich) for his support with DSC measurements and Christian Ba¨rlocher (ETH Zurich) for his help with XRD measurements. Critical discussions with Peter Walde (ETH Zurich) are highly appreciated. The SANS data are based on experiments performed at the Swiss spallation neutron source SINQ, Paul Scherrer Institute, Villingen, Switzerland. LA702242V
