9558
J. Phys. Chem. 1995,99, 9558-9566
Structures of Organogels Based upon Cholesteryl 4-(2-Anthryloxy)butanoate, a Highly Efficient Luminescing Gelator: Neutron and X-ray Small-Angle Scattering Investigations P. Terech,*st.$I. Furman$J and R. G . Weiss’ CEA-Dkpartement de Recherche Fondamentale sur la Matikre Condenske, SESAM-PCM, I7 Rue des Martyrs, 38054 Grenoble Ckdex 09, France, and Department of Chemistry, Georgetown University, Washington, DC 20057 Received: February 20, 1995@
Organogels of cholesteryl 4-(2-anthryloxy)butanoate (CAB) in decane and butanol have been studied by scattering techniques. The neutron and X-ray scattering curves of these gels consist of a mix of the form factor of the aggregates and the structure factor of their organized interacting domains. The results demonstrate that the aggregates are long and rigid fibers, the diameter (d) of which is slightly sensitive to the solvent type: d = 160 8, in decane and d = 192 A in butanol (assuming a homogeneous circular cross-sectional symmetry hypothesis). The fibers are interconnected by “junction zones” in a random three-dimensional network. In decane, the structures of the aggregates in the junction zones are lyotropic organizations obtained through a transformation from the hexagonal packing of the crystalline state. The fibrils with a diameter of about 75.6 A, corresponding to approximately twice the molecular length, are “swollen” to ca. 102 A. In alcohols, the structural organization is closer to that of the solid state and exhibits sharp interfaces with the solvent. The results obtained in the present studies are discussed in the context of previous electron microscopy and fluorescence studies of CAB gels and compared to other structurally related gelators.
1. Introduction Some dilute binary solutions of low-molecular weight compounds can form thermally-reversible viscoelastic materials in appropriate organic solvents. The sample obtained through a liquid-to-solid-like phase transition, often on cooling of an isotropic solution, has mechanical properties which can be that of a gel. Besides qualitative and somewhat subjective recognition methods, such as visual tests of the nonflowing character in a test-tube-tilting method, this class of physical gels is well identified by basic rheology experiments.’ The elastic and viscous components mixed in such a viscoelastic material and the elasticity measured by the storage modulus are much larger than the dissipated energy. Frequently, these systems are characterized by a yield value (Le. minimum stress which has to be applied before a significant flow can occur). The microscopic meaning of this characteristic mechanical behavior lies in an aggregation process leading to a three-dimensional (3D) network of interacting particles immersed in a fluid phase. The characteristics of the related nodes, or, more realistically, of the extended junction zones of the aggregates, determine most of the mechanical and dynamical properties of the gels. The largest aggregate units, colloidal particles, interact with moderate strength and exhibit a thermally reversible sol-to-gel phase transition. The morphology of the aggregates from binary systems is usually one-directional. Beyond the gelation threshold (e*),where the finite aggregates are connected, an infinite 3D network is formed. Phase separation is avoided by the cohesion of the “nodes” of the porous structure. This paper describes some microscopic structural aspects of gels containing small amounts of cholesteryl 4-(2-anthryloxy)butanoate (CAB), a member of a family of recently discovered
* Author to whom correspondence should be addressed.
’ CEA-DCpartement de Recherche Fondamentale sur la Matibre Cond-
ensCe. = Member of CNRS. 5 Georgetown University. Present address: INSERM U298 CHRU, 49133 Angers, France. @Abstractpublished in Advance ACS Abstracts, May 15, 1995.
“gelators” (ALS molecules which consist of an aromatic A and a steroidal S group coupled through a linking group L).2 These molecules form gels with a wide variety of organic liquids. The gels are very luminescent, allowing changes in fluorescence intensity to be used as a method to measure Tg,the gelation temperatures. On cooling of appropriate isotropic solutions, ALS molecules self-associate spontaneously. As described by statistical mechanic^,^.^ aggregation in organic liquids occurs for concentrations above a lower limit or range, Ctim,when there is a difference in the cohesive energies between the molecules in the aggregated and the dispersed (“monomeric”) states. Frequently, for the organogelators, Cli, is hardly distinguished from C*. Intra-aggregate forces, such as directional polar interactions and molecular packing considerations, determine the structure of the aggregates. One-dimensional aggregates (Le. multimolecular structures with very high aspect ratios) are formed at Ctim, and the resulting density distribution of molecules in aggregates is highly polydisperse. The mean aggregation number (also proportional to the mean statistical length of the aggregate) results from thermal equilibration and is very sensitive to gelator concentration and to the gelatorgelator attraction energy. Although some work has been devoted to determining the molecular and aggregate packing arrangements of small aqueous gelators, such as the bile salts? the structures of gels with organic media have been largely unexplored. During the past several years, numerous new organogelators have been discovered, and more systematic chemical ~ t u d i e are s ~ now ~ ~ ~being ~ performed. Recognition of the molecular chemical requirements which are responsible for the gelation ability of a particular compound or class of compounds is stimulating exploratory programs of chemical syntheses of different families of molecules. One example is the aforementioned class of small steroid derivatives ( A L S molecules).2 The gelation behavior of one member of the family, CAB, has been examined in great detail.* Its gels are extremely sensitive to the nature of the liquid component.
0022-365419512099-9558$09.00/0 0 1995 American Chemical Society
Structures of Organogels Based upon CAB
J. Phys. Chem., Vol. 99, No. 23, 1995 9559
CAB is of special interest to us because it is composed of a steroid and aromatic part connected by a flexible functional link,allowing a direct comparison between the structure of its
and SAXS complementary methods. Thus, it provides a coherent description of the aggregates in the gel phases. The results are compared to previous electron microscopy investigations of CAB gels and discussed in the framework of the DDOA and STNH gelator systems.
2. Experimental Section CAB
m:-DDOA
&“ STNH
gels and those formed by either of two molecules, 2,3-bis(ndecy1oxy)anthracene (DDOA6) and an androstanyl derivative (STNP), which are close analogues of the individual anthryl and cholesteryl parts of CAB. STNH is known to gel only selected hydrocarbons, while DDOA is an excellent gelator of a variety of organic liquids of different polarity, ranging from alkanes to alcohols and nitriles. In general, understanding the structural factors which control gel formation would allow a priori design of new gelators with specific properties of their gels (such as strength and stability, chirality, electronic conjugation, solvent compatibility, conformations, etc.). The study of the mechanism of gelation of organic liquids by very small amounts of low-molecular weight compounds is facilitated by the use of nondestructive techniques such as scattering by high-flux neutron sources and synchrotron brilliant X-ray beams. Small-angle scattering (SAS) of the two radiation sources provides complementary information on long correlation distances in the range typical of colloidal systems (1000 A d 20 A) and can take advantage of some contrast variation opportunities. The intensity distribution of long-wavelength neutrons scattered within small angles (small-angle neutron scattering technique (SANS)) provides a scattering curve typical of the material and consists of an interference pattern whose reverse Fourier transform provides real-space structural information. The long-range fluctuations depend upon the contrast ot the particles in the medium which, for neutron radiation, has a direct dependence on the isotopic composition of the two-phase system. Consequently, any change in the constitution of the system (solvent or surfactant composition) can lead to a variation in the scattering level in a way that is related to the morphology, internal homogeneity, and composition of the particles.I0 Neutron data can then be compared to small-angle X-ray scattering (SAXS) data for which the contrast is a function of the atomic number Z of the elements. Despite the fact that the structural models from analyses of SAS curves cannot be proved to be uniquely correct, (due to the lack of phase information and spatial resolution), SAS has proved to be very efficient in characterizing colloidal aggregates, especially when it is complemented by another technique. The present paper details a structural study of the gels formed by CAB in both polar (butanol, octanol) and nonpolar (decane) solvents using S A N S
CAB was available material prepared by Dr. Yih-chyuan Lim2 Due to the light sensitivity of its anthryl group, CAB was kept in the dark until fresh gels were prepared by cooling of a vigorously heated mixture with a selected organic solvent. The gel samples were stable at ambient temperatures over a 3-4day period before solid-liquid phase separation was visually evident; the alcohol gels were the more stable. Quartz cells with 1- or 2-mm gaps and 1-mm-path length cells with capton windows were used for neutron and X-ray experiments, respectively. Fully deuterated 1-butanol (Aldrich) (designated D-butanol) and decane (Cambridge) (designated D-decane) were used for the neutron scattering experiments, while natural isotropic abundance decane and 1-octanol (Aldrich) were used for the X-ray study. These solvents are known to give rather stable gels,2 and their high boiling points minimized solvent evaporation. When the heated solutions were cooled in a test tube, a solid-like material was obtained which did not exhibit flow when the sample holder was inverted. New gels with low concentrations of CAB were almost transparent with a slight violet color (Tyndall effect). Turbidity was noticed in more concentrated or older gel samples. Only fresh samples were used for the scattering experiments. The S A N S study was performed using the 8- and 30-m spectrometers of the 20-Mw reactor at NIST (National Institute of Standards and Technology, Gaithersburg, MD). A common wavelength of 10 A was chosen for the two spectrometers (mean triangular wavelength distribution of 0.25 and 0.31, respectively). The bidimensional detectors were 64 x 64 cm2 3He position-sensitiveproportional counters giving 128 x 128 arrays of raw data. Absolute intensities (cm-I) were obtained by calibration with the standard scatterings of a silica gel and a polystyrene sample for the 8- and 30-m spectrometers, respectively. For the 8-m spectrometer, the detector was pivoted by 7.5 degrees around the sample to extend the angular range and a multibeam converging collimation system was used. The S A N S data reduction and usual corrections for the detector response, substractions of the various transmission corrected backgrounds, masking steps of the detector and absolute calibration were performed with the NIST standard programs. The momentum transfer Q (.$-I) was defined as (4n/A)sin 8, as usual for pure elastic scattering, where 8 is half the scattering angle and A the neutron wavelength. The Q-range investigated was 0.003-0.2 A-I. Due to the limited stability of the samples, a contrast variation study was not performed. SAXS data were obtained at the DCI synchrotron source “Laboratoire pour 1’Utilisation du Rayonnement Electromagnttique” (LURE, Orsay, France) for two sets of experiments on the D22 instrument at A = 1.458 A ( E = 8500 eV) and two distances (1.75 and 0.714 m). A gas-filled (Xe-C02) detector with a 0.217 “/channel spatial resolution was used (over 512 channels). Intensities (au) were transmission corrected for the empty beam signal and solvent scattering according to I = Isam/ (I0,samTsamtsam) - Iback/(IO,backTbackfback), where the subscripts Sam and back refer to the sample and background, respectively, t is the corresponding counting time, IO is the incident X-ray intensity, and T is the transmission. The Q-range investigated was 0.005-0.3 A-I. All spectra were recorded at room temperature.
Terech et al.
9560 J. Phys. Chem., Vol. 99, No. 23, 1995
u
.
-1
0.01
0:1
0.01
Q
Figure 1. SANS spectra of CAB organogels. Logarithmic plot of absolute intensities I (cm-I) normalized by the CAB concentration ( C ) vs Q &I): (1,O) D-decane, C = 0.0054 g - ~ m - ~ (2,O) ; D-butanol, C = 0.00476 ~ c m - ~ . c3
1o1
10-1
10”
1o
.~ 0.01
0.1
0.
Q
Figure 2. SAXS spec$a of CAB organogels. Logarithmic plot of intensities I (au) vs Q (A-I): (1, 0 ) decane, C = 0.0192 g ~ m - ~(2,; 0) l-octanol, C = 0.01058 g ~ m - ~Intensities . for decane are multiplied by 10 for the sake of comparison.
3. Results All the neutron spectra of the organogels are isotropic regardless of the cell thickness used (between 1 and 2 mm). Figure 1 shows the neutron scattering curves of CAB gels in deuterated decane and l-butanol. The neutron scattering profiles are clearly different in the two solvents. Additionally, a striking feature of these scattering curves is the presence of a very strong intensity oscillation, or peak, at Q 0.0683 for the decane gel which is absent in the alcoholic gel. The X-ray scattering curves in undeuterated decane and l-octanol are also very different, as shown in Figure 2. In decane, three intensity oscillations, or broad peaks, are clearly seen at Q 0.071, 0.136, and 0.195 A-I. In l-octanol, the intensity decrease is monotonous except at Q % 0.195 A-I, where a weak peak is observed. The diffraction pattern of the CAB solid powder is shown in Figure 3 within the same Q-range as that for the undeuterated gels (synchrotron source). The Bragg peaks are shown with arrows at Q % 0.094, 0.163, 0.188, 0.248, and 0.281 k’.
.
0.1
Q
Figure 3. X-ray diffraction pattern of CAB powder recorded under the same experimental scattering conditions as those used for the organogels (identical logarithmic representation). The peaks are indicated by arrows (see text).
Nat.is the number of atoms in a CAB molecule (C45H6003) and NA the Avogadro number. b, is the neutron scattering length for the atom i, es is the neutron density of the solvent ( c m ~ m - ~ ) , and bCAB is the specific neutron density of the surfactant bCAB = NAC21bilM, where the summation is extended over the neutron scattering length values (bi) of the 108 atoms of CAB. M is the CAB molecular weight, and VCAB, the specific volume of CAB in the aggregate. The neutron scattering length densities of the solvents @sare 78.28 x lo9 cm*g-’ for D-decane and 71.03 x lo9 cmqg-l for D-butanol. Using eq 1, the specific contrasts of CAB aggregates in the two solvents are IhbCAB/decanel = 57.43 x 109 cmmg-1 and IbCABlbutanoll = 56.78 x lo9 cm-g-I, if, in the first approach, the CAB aggregates of both gel types are considered to exist in a packing close to that of the solid state (VCAB 1 cm3*g-’). Since theneutron contrast values are rather comparable ((iiCABlbutanol/AbCAB,decane)2 % 0.98), any major scattering difference between the two gel samples can be attributed to the specific influences of the solvent type upon the morphology of the CAB aggregates. At this stage, considering the scattering profiles, it can be observed that the different solvent types induce significant structural changes in the aggregates which are analyzed below. In most examples of rather dilute organogels ( C I l%), the aggregates have an unidirectional shape (wormlike chains, rods, or rigid fibers) which accounts for the overlap of their spherical reorientational volume at rather low concentration. It is thus reasonable to undertake the analysis in this For a system of monodisperse, noninteracting particles with a rodlike local structure, the scattering can be expressed as
where F(& is the form factor of the rodlike particle at a given orientation relative to the momentum transfer Q, while ( ) Q stands for an average over all such orientations. The form factor of a homogeneous rodlike particle of length L = 21 and radius ro is
4. Analysis
To analyze the neutron scattering properties of CAB in D-butanol and D-decane, the neutron specific contrast for a CAB aggregate (cm-g-I) was calculated using eq 1. The indices
CAB and S refer to the solute CAB and the solvent, respectively.
where V is the volume of the scatterer, a is the angle between Q and the rod axis, eo is the scattering amplitude density relative to the solvent, and J I is the first-order Bessel function of the first kind. Because the scattering of the CAB gels is isotropic in the experimental Q-range, the related CAB aggregates are uncor-
J. Phys. Chem., Vol. 99, No. 23, 1995 9561
Structures of Organogels Based upon CAB related in position and orientation, and, as a consequence, the calculation of the scattered intensity is simplified in the socalled conditions of particle scattering. The intensity then reduces to a sum of the spherically averaged scattered intensity of each particle. The averaging of eq 3 for long rods (I5 >> 2ro) leads to the simplified eq 4of the scattered intensity, where
C is the surfactant concentration ( g ~ m - ~ )Equation . 4 is the product of both (1) an axial term proportional to l/Q and the molecular weight per unit length ML of the linear aggregate multiplied by the square of the contrast and (2) a cross-sectional term involving the Bessel function J I . In appropriate concentration and Q-range conditions, the scattering curve of gels is that of particle scattering in which the contribution of the nodes of the network is minimized a n d or rejected in a lower-Q domain. The average distance between the particles defining the porosity (4)of the network is rather large when compared to the particle diameter (2ro) and prevents the orientation effects. This situation is different in the vicinity of the junction zones, where the contribution to the scattering of the nodes can be estimated (see for instance ref 20) in an idealized, ordered, and concentrated network. From small to large angles, various typical real-space distances of the unidirectional aggregates can be probed within appropriate Q-range conditions, as defined by 2Jt/Qmi, d 2n/Qmax. The following structural parameters can be estimated: the statistical hydrodynamic radius of gyration of the whole aggregate correlated to the contour length and the persistence length, the rodlike rigidity, the cross-sectionalradius of gyration, the homogeneity, the shape and polydispersity of the aggregate's cross section, the type of its interface with the solvent, and the intemal molecular organization. For CAB gels, the low-angle part of the scattering is a monotonic decrease from relatively high intensity values (Figures 1 and 2), indicating that the overall length of the aggregates is not being probed. Expansions of eq 4 in appropriate angular domains are used to characterize the aggregates. In the low-Q range, the crosssectional term (Bessel function) reduces to unity and the asymptotic behavior is a Q-l dependence observed in the range 1/15 or l/(lp)> ca. 1.9 x 21Qmi, 800 A exist in both solvents. The scattering curves in the QZ vs Q representation show a strong intensity decay after the typical plateau. Such an intensity decrease is consistent with eq 4,which reduces to eq 5 in the (5)
low-Q limit. It appears that the intensity decay is due to the finite size of the cross section, while the infinite axial dimension leads to the low-Q Q-' intensity divergence. In the corresponding Q-range (l/(lp) < Q < l/Rc), the cross-sectional intensity decay is Gaussian (eq 5). In eq 5, R, is the cross-sectional radius of gyration defined as the second moment of the distribution of the scattering density (eq 6). Rc depends on the geometry, homogeneity, and dispersity
Terech et al.
9562 J. Phys. Chem., Vol. 99, No. 23, 1995
0
0.0004
0.0008
@
Figure 5. “Guinier plots” In (QZ)lis Q2 for SANS (A,B) and SAXS (a,b) experiments of CAB organogels; (A, A) D-decane; (B, A) D-butanol; (a, 0 )decane; (b, 0) 1-octanol. Concentrations and results are collected in Tables 1 and 2.
TABLE 1: Guinier Analysis of the SANS Data for CAB Gels in D-Decane and D-Butanol CAB/solvent
CABID-decane
C ABD-butanol
concn ( g ~ m - ~ ) mass fraction (%)
0.0054 0.636 59
0.00476 0.514 61
Rc (A)
TABLE 2: Guinier Analysis of the SAXS Data for CAB Gels in Decane and 1-Butanol CAB/solvent concn ( g ~ m - ~ ) mass fraction ( 7 ~ ) Rc (A)
CAB/decane
CAB/octanol
0.0192 2.564 54
0.0 1058 1.263 69
of the cross section. Figure 5 shows the so-called “Guinier plots” for CAB in alkanes and alcohols as measured by S A N S and SAXS experiments. The data can be fitted in the Q-range by a linear dependence which justifies the use of the specialized formalism of eq 5. For QR, < 1, R, values are deduced and collected in Table 1 for deuterated solvents and Table 2 for undeuterated solvents. The determination has been made only for Q < (RJ-I (ca. 0.015 A-I) even if the straight lines of Figure 5 can be extended over broader domains. The estimation of accuracy of the determination of R, is approximately 2 A. For CAB gels in alcohols, the slopes are comparable for the two scattering techniques, suggesting that the cross-sectional contrast with respect to the solvent can be considered as homogeneous within the range of the instrumental resolution. Equation 6 relates the contrast dependence of the radius of gyration of a three-dimensional particle, where RG is the radius of gyration of the whole scatterer and V is its volume. An equation similar to (6) can describe the two-dimensional situation of cross sections in fibrillar aggregates. When the scattering density is not uniform, the center of gravity of the contrast profile is not the same for neutron and X-ray scattering and thus leads to a variation of the radius of gyration estimation (see ref 10 and references cited therein). This situation may be found, for instance, in structures where a solid-type packing is mixed with solvation areas. Tables 1 and 2 show that there is a slight variation in the R, values (59 and 54 A) from the S A N S andSAXS data for CAB/decane samples which suggests a heterogeneous contrast profile for the related aggregates. Except for CAB in D-decane, the experimental points in the lowest Q-range remain aligned (Figure 5 ) , indicating that the unidirectional character of the aggregates is maintained over long distances corresponding to about 2z/Q,i, 2000 A. In a first approach, the cross section is considered as being homogeneous in terms of neutron or X-ray scattering power without any prejudgment of the internal molecular organization. Within
0.01
0.1
Q
Figure 6. SAXS scattering curves of CAB organogels. Points are ~; 1-octanol, C experimental: (1,O) decane, C = 0.0192 g ~ m - (2,O) = 1.263 g ~ m - ~Full . lines are best fits according to eq 4: (1) Ro = 82 A, E = 0.24; (2) Ro = 85 A, E = 0.24. The dotted line is the same fit but includes a Gaussian cross-sectional polydispersity ( E = 0.1) to emphasize the noncoincidence between the theoretical and experimental secondary maxima. this framework, the mean values for the radii of gyration are 68 8, for the CAB fibers in alcohols and 56.5 8, in alkanes. If a circular cross section is assumed, the geometrical radii (Ro) are related to the radius of gyration by Ro = Rc2It2. Consequently, the diameters of the fibers are slightly larger in alcohols (192 A) than in alkanes (160 A). At still larger Q-values, after the asymptotic Q-’ plateau (in a QZ vs Q representation, Figure 4) and the intensity decay due to the finite size of the cross section, three rather intense oscillations (mainly the first two ones) are clearly seen for CAB gels in undeuterated decane (only one is seen in D-decane due to the experimental Q-range cutoff, Figure 4A) in contrast to CAB gels in alcohols. Figure 6 shows a best fit of eq 4 to experimental data, assuming a homogeneous circular cross section for the scattering particles. For decane or alcohol gels, the agreement in the lowangle region is very satisfactory. In the wide-angle region, in the case of the decane gels, a clear discrepancy exists between the form-factor oscillations generated by the Bessel function and the experimental data. The origin of these large-angle scattering features can be assigned either to cross-sectional formfactor oscillations of a specific geometry, homogeneity, and dispersity or to the structure factor of some crystalline heterogeneities within the gel network. In the following analysis, the scattering curve is interpreted as resulting from the mixing of the form factor of the CAB fibers (low-angle part) and the structure factor (wide-angle part) of pseudocrystalline organizations in the interaction zones of the gel network. A direct consequence of this mixing is that the exact cross-sectional profile of the CAB aggregates in decane gels cannot be characterized by the SAS technique. To minimize the number of variational parameters in the calculations, it is not necessary at this stage to add further parameters such as the internal heterogeneity and/or cross-sectional anisotropy. In alcohols, the scattering curve does not show the diffraction peaks observed in decane except at large Q (0.195 k’), where a single weak peak is seen (Figure 4). The form factor of the connecting fibers is consequently observed and fitted in a much larger Q-range (Figure 6). A satisfactory agreement to eq 4 is found which supports the proposed structural model. The deduced structure (d = 170 A) can be confirmed by using the so-called Porod analysis. Equations 7 relate the scattering to the surface area per unit volume (2)in a transfer momentum range corresponding to the sharp interface of the scatterers. This law is independent of the geometry and topology of the sample.
Structures of Organogels Based upon CAB
Inv. = JwQ2Z(Q) dQ = 23244 1 - q5)(Ae)2
J. Phys. Chem., Vol. 99, No. 23, 1995 9563
and
In eqs 7, Ag is the volumic contrast of the particles and q5 is the aggregate volume fraction. For cylinders of radius R, the relation = 2qYR is obtained when the contribution from the extremities is neglected. For CAB in D-butanol (4 = 0.005; see Table l), a @Z vs Q plot (Figure 7) shows an extended bump characterizing the form factor of the aggregates. The invariant, obtained by inte ration up to Q values lower than the Bragg peak at Q % 0.2 - I , is found to be ca. 0.00016 A-2 cm-I, while l i m p = 1.26 x A-4 cm-’ and then R 81 A. This result is in good agreement with the previous Guinier treatment and low-angle fit of the scattering curve (see Tables 1 and 2 and Figure 5). Figure 7 shows the Porod plot together with the above-mentioned low-angle best fit obtained according to eq 4. The CAB aggregates in alcohols are fibers with a diameter d of approximately 162 8, which form sharp interfaces with the solvent. In 1-octanol, a similar data treatment gives Inv. = 1.06 x lop6 A-2 cm-I, l i m p - (@I) = 9.0 x k4cm-’ and R % 75 A. Due to the mixing of the formfactor intensity and the Bragg peaks in the intermediary Q-range, such data treatment cannot be applied in the case of decane gels. Because neutron beams can be calibrated with respect to a standard sample, absolute intensities can be used to deduce some quantitative information. The uncertainty of the method is rather high considering both the intrinsic error (about 20%) and that specific to the CAB samples (concentration uncertainty with respect to some crystallization reactions, sample homogeneity, contrast evaluation, etc.). To circumvent these drawbacks, a “correction factor” is calculated from the data at large angles with the Porod limit assuming R = 81 8, (eq 7b). Introducing this factor into the low-angle Guinier determination of the extrapolated intensity at Q = 0, an estimation of the number of CAB molecules per unit length of fiber can be made using the combination of eqs 1,4, and 5. Doing so, nL = MJM is found to be about 18 mo1ecules.A-I. This value can be compared with the density of a CAB aggregate in a solid-state packing. For a cylinder of a radius of 81 A and length of 1 A, the number, n ~ of , CAB molecules is n(82)**1*1*Na/Mc~~ % 19.1 m0l.A-I (assuming VCAB % 1 cm3*g-l). The above two values of n~ are comparable and support the consistency of the low and largeangle analysis. The SAXS spectra of the decane gels show secondary maxima (see Figure 6) in the intermediary and large Q-range at positions different from those of the solid powder. The reflections of the crystalline powder (0.096, 0.163, 0.188, 0.248, and 0.281 A-I) are almost exactly in the ratio l:&:&:&:h, suggesting hexagonal ordering of the solid. It is known that a twodimensional hexagonal array of scatterers (space group p6m) generates reflections whose spacing is defined by eq 8,16where Qhk is the reciprocal spacing of the reflection of the plane with indices (h,k) and a* is the reciprocal cell parameter.
x
(PI)
Qhk
= 2na*(h2
+ k2 - hk)”*
(8)
The Corresponding intercolumnar distance is DO = (2/d3). (2n/Q(ll,) = 75.57 A. To account for the maxima found in the decane gels, the displacement of the Bragg peaks in swollen
0
0.04
0.08
Q
Figure 7. Porod analysis. SANS experiments for CAB gel in D-butanol: lim I@ = 1.22 x Inv. = 0.00015, Ro = 78.3 A. The full line is a fit according to eq 4 with Ro = 82 A and using a Gaussian cross-sectional polydispersity, E % 0.24.
hexagonal arrangements is considered. It has been shown for alkaline-earth soapsI7that the swelling is parallel to the variation of the intercolumnar distance, D, according to eq 9, where DO is the intercolumnar distance of the pure surfactant, 42: is the weight fraction of the solvent of the lyotropic diffracting unit, and Vsudacmt and Vsolvent are the specific volumes of the surfactant and solvent, respectively.
For the gel in undeuterated decane, vsolvent = 1.37 cm3.g-’, 1 cm3.g-’, and, assuming that the intensity apex in the diffraction pattern of the gel at Q % 0.071 A-1 is the (1 1) reflection of a swollen hexagonal microphase, the solvent fraction 42: in the mesophase can be deduced. A value of 42: = 0.425 is calculated using eq 9, suggesting that the gel at 4 b d k - 0.974 (Table 2) is a heterogeneous network of fibers connected in the junction zones by swollen hexagonal microdomains (intercolumnar distance = 102.2 A). As expected, a much higher surfactant fraction, 4 0.57, than that in the bulk binary system is found in the lyotropic domains. An identical calculation made for the gel sample in D-decane with a “peak” at Q % 0.0683 A-1 gives 42; % 0.5 for a bulk concentration &’:y” = 0.994 (Table 1). The concentration dependence of the peak position, its specific shape, curved at the apex, and its large width (hwmh = 0.02 A-l, CAB/D-decane) typical of a diffuse ring, are scattering features consistent with a lyotropic organization in the gel network. In passing, it can be observed that the surfactant concentration within the supposed swollen hexagonal arrangements is in the usual range found for some aqueous binary systems of soaps (see, for instance, ref 17). The spacings of the peak positions observed in the gel state are in the ratio of 1:1.92:2.75 and have to be indexed according to the most probable symmetry of the system. Various structural transformations of periodic arrangements in two dimensions can be considered to describe the probable symmetries in the orthogonal plane to the bundle of fibers within the junction zones of a gel network. These modifications have already been studied in different surfactant-made lyotropic ternary systems. For instance, a progression from hexagonal to rectangular to lamellar phases is observed by gradual addition of decanol to aqueous solutions of the ionic surfactant sodium decyl sulfate.Is For a hexagonal symmetry, the spacings 1:d 3 :d 4 :d 7 :d9 of the typical reflections and the corresponding diffracting families of planes can be deduced from eq 8. The origin of the spacing at 1.92 can be due to the 4 3 spacing having a very weak intensity resulting from a specific three-dimensional organization Vsufiafacmt
Terech et al.
9564 J. Phys. Chem., Vol. 99, No. 23, 1995
8 10
10
0.1
0
0.2
Q
0.01
0‘3
0.1
Q
Figure 9. CAB organogels: comparison between SAXS and electron microscopy results. Points are experimental (SAXS): (1, 0 )decane, C = 0.0192 g ~ m - (2, ~ ; 0) 1-octanol, C = 0.01058 g ~ m - ~Lines . are theoretical scattering curves following previous E.M. measurements (see text and ref 2c). Equation 10 is used: (1) decane, b = 52 A, k = 0.5; (2) 1-octanol, b = 41, k = 0.31.
0.12
0.08
0.16
Q O.* Figure 8. (A) Comparison of the X-ray scattering patterns (linear ) the scales) of a CAB/decane organogel (0,C = 0.0192 g ~ m - ~and crystalline powder (+). (B) Enlargement of the second peak of the CAB/decane gel diffraction pattern (0).Expected diffracting positions of the expected 4 3 and 4 4 theoretical reflections of a swollen hexagonal ordering (see text) resulting from the transformation of the solid-state ordering (+). of the domains or to the overlap of the two expected 4 3 and 4 4 diffuse reflections. Knowing the displacement ratio of the (11) reflection (0.76), the two diffuse rings are expected at Q x 0.163 x 0.76 0.124 8,-] and Q 0.188 x 0.76 0.143 (mean value 0.134 8,-’), while the measured position is at Q x 0.136 A-1. Figure 8B shows an enlargement of the scattering curves which demonstrates that the observed single broad peak could reasonably come from the fusion of the two marked reflections. The s acing 2.75 in the gel can be assigned to the expected one at 7 (2.65) or to the fusion of 4 7 and 4 9 reflections in a mean reflection at 2.8. Figures 3 and 8 illustrate the recovery process of the diffuse reflections. Now, if a rectangular symmetry (sides a,b) is considered, reflections with spacings defined by Q h k = 2 ~ c ( h ~ ( a *f) ~ k2(b*)2)”2are generated with cmm and p g g as possible space groups. Using the same argument as detailed above, there is no significant improvement of the indexation whatever the combination of a and b values. For instance, in a cmm symmetry, the expected sequence of spacings would be l:(l/ 2a)[a2 b2]”2:2.0:bla, which does not improve the fit to the data. If a transformation to a lamellar organization is considered, spacings in the sequence 1:2:3:4 are expected. The lamellar organization is excluded since the reflection observed at a spacing 2.75 cannot be assimilated to that expected at 3.0. In the decane gel spectra, taking into account the contrast variation between SANS and SAXS as measured by the crosssectional radii of gyration values and that no traces of the diffraction peaks typical of the dry solid state can be observed, it is assumed that the CAB aggregates are microdomains of swollen mesophases (probably hexagonally packed). In butanol, the CAB aggregates are in a packing arrangement closer to that of a solid state (see above) but in a structure which
J
+
can be different from the original hexagonal ordering. This situation is in contrast to that found for decane gels for which the diffraction data suggest a participation of the solvent in the structure. A single peak is definitely insufficient to propose a crystallographic space group for the aggregates in alcohols, but it is interesting to note that its position is similar to that of the fourth reflection (Q(sI,)in the corresponding swollen hexagonal decane microphase. A detailed crystallographic study is needed to investigate the structural organizations and correlations between the crystalline solid state and the gels in the two solvent systems. 5. Discussion
The structural deductions of the present scattering experiments on the native gels (with their solvent) can be compared with some previous electron microscopy results.2c Using a freezeetching procedure, transmission electron micrographs have shown that the CAB gels are composed of long and rigid fibers, the cross-section shape of which is slightly rectangular (Le. ribbonlike fibers). The cross-sectional areas have been shown to be 209 x 104 = 21736 A2 in decane and 263 x 82 = 21566 A2 in 1-octanol (where a twisted structure with a mean pitch of 1200 8, was also observed). The corresponding theoretical scattering curves, calculated from the parameters deduced from the electron microscopy measurements, are reported together with the actual SAXS experimental data (Figure 9) and using the theoretical eq lO.I9
2c --(Ab)2MLJ Q
2n
sin(Qu cos q ) sin(Qka cos q ) Qacosg, Qkucosq
[
In eq 10, a is half the long side of the rectangular homogeneous cross section, b is half the thickness, and k is its related anisometric ratio ( k = bh). As far as the low-angle part of the scattering curves is concerned, a moderate agreement can be found between the SAXS experimental data and the fit based on the values obtained from the electron microscopy experiments. The unidirectionality of the aggregates and the average radii of gyration of their cross sections are confirmed, while the slight low-angle departures are induced by the moderate cross-sectional anisometry introduced in the model. For the decane samples, the comparison of the curves (Figure 9) is made only after the Gaussian intensity decay due to the presence of mixing of the structure factor to the form factor. For the CAB/alcohol samples, the two curves
Structures of Organogels Based upon CAB can be compared further, and significant differences are seen which cannot be explained simply by introducing an assumed significant cross-sectional polydispersity. Despite the fact that the cross-sectional anisometry cannot be confirmed by the present scattering experiments, it is interesting to note that the hexagonal lattice with a 6-fold symmetry and one aggregate per unit cell are consistent with a centered rectangular phase (two aggregates per unit cell). When the ratio of the wider side to the narrower one of the rectangle is no longer 4 3 , as it is in the hexagonal symmetry, the hexagonal phase can transform into a rectangular one (see ref 16). The cross-sectional anisometry, k, varies from 2.0 to 3.2 from decane to 1-octanol gels as measured by electron microscopy. Taking into account intrinsic errors that may be involved in electron microscopy measurements, the experimental ratio k 2 in decane is not very different from the theoretical one of 4 3 for hexagonal arrangements. In alcohols, a drastic transformation of the hexagonal ordering to a more compact arrangement has to be invoked. The hypothesis of a helical structure is not considered in the modeling of the scattering data because the supposed amplitude of its related pitch (1200 A), which is large compared to fiber diameter, would not induce very significant effects in the low-Q range (simulations according to eq 16 of ref 9; not shown). The actual molecular length of CAB (ca. 38 A) can be compared to the different characteristic sizes determined in the present study (75.7 8, in the solid; ca. 160 8, for the diameter of the fibers; ca. 102.2 8, for the intercolumnar spacing in the decane swollen junction zones). It appears that bimolecular pairs can be involved in the columns of the solid state while in gels; depending upon the mechanisms of molecular interaction, different multiples may be proposed according to various headto-tail, head-to-head, or interdigited configurations for the associated molecules in the aggregates. When crystalline ordering is involved in the junction zones of the gel network, solidlike gels exhibiting high yield stress values20are obtained. This behavior is observed for CAB, the anthracenyl gelator DDOA,2’and the steroid STNH derivative.22 As far as scattering is concerned, the increased molecular size of CAB compared to that of STNH accounts for the truncated nature of the form factors of the aggregates when long-range ordering is involved (N.B., CAB in decane). In decane gels, a swollen hexagonal arrangement, reminiscent of the hexagonal ordering of the solid state, is observed. Other examples are known of such correlations of the gel structure with the solid state or liquid-crystalline state of the native Still, the gelation in decane cannot be seen as a continuous swelling process since gel samples are prepared by a complete solubilization of the CAB at high temperature. The solvent (decane) participation in the structures remains unusual and differs from what is commonly observed with some other systems?O where the junction zones are “dry” microcrystallites whose internal symmetry is lower than that of the crystalline solid. The more compact structures in alcohols may be due to Coulombic interactions between the solvent and the aggregates. However, there is evidence that steroidal groups and 1-alkanols can associate favorably under other conditions. For instance, Lawrence has found that 2: 1 molar ratios of 1-alkanols (containing 12 to 18 carbon atoms) and cholesterol form mesophases even though the individual components are not m e s o m o r p h i ~ . ~ ~ Presumably, hydrogen bonding plays an important role in stabilizing the mesophase aggregates. A similar interaction between CAB and 1-butanol or 1-octanol in the gelator strands may be inhibited by esterification of the steroidal hydroxy group. The relatively high T, of CAB/l-octanol gels (in comparison
J. Phys. Chem., Vol. 99, No. 23, 1995 9565 to the T . of CABln-alkane gels; vide infra), ca. 63 OC with 0.8% CAB, is further evidence for the exclusion of alcohol molecules from gelator strands. Since spectroscopic studies support a packing model in which the aromatic parts of CAB molecules are stacked,2 it is reasonable to assume in CAB/decane gels that the decane molecules penetrating the gelator strands are located in regions rich in aliphatic (steroidal) groups. Furthermore, strands must be swelled in ways that depend upon the general shapes of the alkanes: the T, of 0.8% CABln-alkane gels are constant (ca. 42 “C) for n-alkanes from heptane to hexadecane; Tg of 0.8% CAB/methylcyclohexane gels are about 5 degrees lower, and the gels have significantly shorter lifetimes; these values are much lower than those of the corresponding 0.8% CAB/alcohol gels.2c As mentioned previously, absorption, fluorescence, and circular dichroism spectra of CAB gels with alkanes and alcohols have led to the conclusion that there is significant overlap of anthryl groups within gelator strands.2 The similarity of the diffraction patterns of solid CAB and DDOA2’ support this contention. Furthermore, the X-ray diffraction pattern (in the same experimental Q-window) of an anthraquinone derivative of CAB (CAQ) is that of an amorphous material without significant scattering features, while CAQ gels in decane and alcohols exhibit diffuse peaks, similar to those found in decane gels of CAB and which are typical of a lyotropic organization.26 In CAQ, the steroid and aromatic parts are more rigidly linked than in CAB, suggesting that molecular packing, solvent swelling properties, as well as the stability of the samples and the polarities of the aggregates and the solvent, are correlated in a complex manner. Regardless of the details, associations among aromatic groups play an important role in determining gel structure, stability, and aggregate type. An understanding of the factors responsible for gel stability, strand chirality, lyotropism, photosensitivity, etc., when gelators are composed of both a steroid and an aromatic part has been an objective of this work. Although the results presented help to unravel some of the intricacies associated with A L S gelators and their gels, much work remains to be done. Future studies will address the outstanding issues. Acknowledgment. The National Institute of Standards and Technology (NIST) is acknowledged for providing the neutron and X-ray beams and all technical and financial supports. NATO and the Polymers Division at NIST are thanked for support to P.T. We also thank Dr. G. B. McKenna (NIST) for discussions during the course of the work. The “Laboratoire pour 1’Utilisationdu Rayonnement Electromagnktique” (LURE, Orsay, France) is thanked for providing the synchrotron beam used for some of the diffraction and scattering experiments. The National Science Foundation (USA) is acknowledged for its support to R.G.W. References and Notes (1) Almdal, K.; Dyre, J.; Hvidt, S.; Kramer, 0. Polym. Gels Networks 1993,I , 5. (2) (a) Lin, Y. Ph.D. thesis, Georgetown University, Washington, DC, 1987. (b) Lin, Y.; Weiss, R. G. Liq. Cryst. 1989,4 , 367. (c) Lin, Y.; Kachar, B.; Weiss, R. G. J. Am. Chem. Soc. 1989, 111, 5542. (d) Mukkamala, R.; Weiss, R. G. J. Chem. Soc., Chem. Commun. 1995,375. (3) Ruckenstein, E.; Nagarajan, R. J . Phys. Chem. 1980,84, 1349. (4) Israelachvili, J. N. Intermoleculur and surface forces, 3rd ed.; Academic Press: London, 1992; p 341. ( 5 ) Sugihara, G.; Ueda, T.; Kaneshina, S.; Tanaka, M. Bull. Chem. Soc. Jpn. 1977,50, 604. (6) Brotin, T.; Utermohlen, R.; Fages, F.; Bouas-Laurent, H.; Desvergne, J. P.J . Chem. Soc., Chem. Commun. 1991, 416.
Terech et al.
9566 J. Phys. Chem., Vol. 99, No. 23, 1995 (7) Murata, K.; Aoki, M.; Susuki, T.; Harada, T.; Kawabata, H.; Komori, T.; Ohseto, F.; Ueda, K.; Shinkai, S. J . Am. Chem. SOC.1994, 116, 6664. (8) Furman, I.; Weiss, R. G. Langmuir 1994, 9, 2084. (9) Terech, P.; Ramasseul, R.; Volino, F. J . Phys. France 1985, 46, 895. (10) Jacrot, B. Rep. Prog. Phys. 1976, 39, 911. (1 1) Glatter, 0.; Kratky, 0. Small Angle X-ray Scattering; Academic Press: London, 1982. (12) Guinier, A.; Foumet, G. Small Angle Scattering ofX-rays; Wiley: New York, 1955. (13) Cabane, C. In Surfactant Solutions; Zana, R., Ed.; Surfactant Science Series; Dekker: New York, 1987; Vol. 22, p 57. (14) Burchard, W.; Kajiwara, K. Proc. R. SOC.London A 1970, 316, 185. (15) Hjelm, R. P. J . Appl. Cryst. 1985, 18, 452. (16) Ladd, M. F. C.; Palmer, R. A. Structure Determination by X-ray Crystallography; Plenum Press: New York, 1985.
(17) Luzzati, V.; Mustacchi, H.; Skoulios, A,; Husson, F. Acra Cvstallogr. 1960, 13, 660.
(18) Hendrikx, Y.; Charvolin, J. Liq. Cryst. 1992, 5, 677. (19) Mittelbach, P.; Porod, G. Acta Phys. Austriaca 1961, 14, 185. (20) Terech, P.; Rodriguez, V.; Barnes, J. D.; McKenna, G. B. Langmuir 1994, IO, 3406. (21) Terech, P.; Desvergnes, J. P.; Bouas-Laurent, H. J . Colloid Interface Sci., in press. (22) Terech, P. AIP Con5 Proc. 1991, 226, 518. (23) Terech, P.; Chachaty, C.; Gaillard, J.; Giroud-Godquin, A. M. J . Phys. (Paris) 1987, 48, 663. (24) Terech, P. 1I Nuevo Cimento, in press. (25) Lawrence, A. S. C. In Liquid Crystals and Ordered Fluids; Johnson, J. F., Porter, R. S.,Eds.; Plenum Press: New York, 1970; p 289. (26) Terech, P.; Weiss, R. G. To be published. JF'950486J
