J . Phys. Chem. 1990, 94, 5308-5312
5308
Conclusions In summary, we propose that ethylene decomposition on Pt(210) follows the reaction mechanism shown in Figure 11. When ethylene adsorbs onto a 110 K Pt(210) sample a a-bound complex is formed. Some of the a-bound ethylene desorbs upon heating to 250 K, and some ethane is produced around the same temperature. However, most of the a-bound ethylene is stable to room temperature. Further heating results in conversion of this *-bound ethylene to form a mixture of ethylylidyne, methyl species, surface carbon, and hydrogen. The ethylylidyne species decomposes to form a mixture of ethynyl, methylidyne, hydrogen, and adsorbed carbon. The methyl groups decompose to form methylene and (?5) Yagasaki, E.; Masel, R. 1. J . A m . Chem. SOC.,in press.
methylidyne. Further decomposition of these species results in adsorbed carbon and C2 species. The r-bound ethylene remains stable to higher temperatures on Pt(210) than on any other face of platinum. Thus, it appears that the Pt(210) surface has unique properties for ethylene decomposition. Acknowledgment. This work was supported by the National Science Foundation under grant CBT 86- 13258 and by Amoco Oil Co and Shell USA. Sample preparation was done using the facilities of the University of Illinois Center for Microanalysis of Materials which is supported as a national facility, under National Science Foundation Grant DMR 86-12860. Equipment was provided by NSF grants CPE 83-51648 and CBT 87-04667. The experimental assistance of Eriko Yagasaki is greatly appreciated. Registry No. Pt, 7440-06-4; C2H,, 74-85-1.
Structure of Cubic Mesomorphic Phases Determined by Low-Temperature Transmission Electron Microscopy and Small-Angle X-ray Scattering Janet L. Burns, The Procter and Gamble Company, Miami Valley Laboratories, P.O. Box 398707, Cincinnati, Ohio 45239-8707
Yachin Cohen, and Yeshayahu Talmon* Department of Chemical Engineering, Technion-Israel Institute of Technology, Haija 32000, Israel (Received: August 2, 1989; In Final Form: January 8 , 1990)
Cubic phases exist in many amphiphile/water systems at high solute concentrations between the lamellar and hexagonal phases, and at lower amphiphile concentrations between micellar and hexagonal phases. The structure of the latter is still controversial. Structural models that have been suggested to satisfy the experimental data all lacked direct evidence to prove their validity. The structural model we report here is based on a combination of direct and indirect experimental techniques. Direct imaging by low-temperature (cryo) transmission electron microscopy (TEM) of vitrified specimens provides the geometry and size of the "building blocks" of the phase, in this case globular micelles, and shows directly the existence of a cubic phase. The indirect method, small-angle X-ray scattering (SAXS), provides the symmetry group of these cubic phases, the size of the unit cells, and the positions of the 24 globular micelles in these unit cells. Beyond solving the microstructure of model cubic phases of the anionic surfactant HCO-60/water system and providing, for the first time, direct images of this cubic phase and of lysolecithin/water, we demonstrate a general approach for combining direct and indirect methods to solve microstructures of complex fluids.
Introduction Cubic mesomorphic phases have been identified in many phase diagrams of amphiphile/water Cubic phases that exist between the lamellar and hexagonal phases are by now rather well understood in terms of a bicontinuous topology of three-dimensional networks of identical rodlike or lamellar elements linked t ~ g e t h e r . I * However, ~*~ the structure of cubic phases at lower amphiphile concentrations, situated between the micellar and hexagonal phases, is still controversial. Balmbra et aL6 were first to identify this "second" cubic phase in dod_ecyltrimethylammonium chloride/water as belonging to the Pm3n space group. Tardieu and Luzzati2 identified this structure in several other systems and suggested a model based on a three-dimensional network of finite-length rods (cylindrical micelles) forming the bars of a system of cages, each enclosing a spherical micelle. NMR self-diffusion measurements on these failed to support the continuous hydrocarbon phase suggested by the model and suggested the existence of finite, closed micelles. A model by Fox,'* based on a unit cell containing 48 spherical micelles, was subsequently rejected" due to overlap of those micelles. An alternative structure1I*I2 has eight short rodlike or prolate micelles *To whom correspondence should be addressed.
0022-3654/90/2094-5308$02.50/0
located at the special positions a(2) and d(6) of the Pm3n unit cell. That model has been supported by NMR Some of the systems that exhibit the cubic P m h phase involve poly(ethy1ene oxide)-based nonionic surfactant^.^*^^-" For none ( I ) Luuati, V.; Tardieu, A,; Gulik-Krzywicki, T.; Rivas, E.; Reis-Husson, F. Nature 1968, 220, 485. (2) Tardieu, A.; Luzzati, V. Biochim. Biophys. Acta 1970, 219, 11. (3) Fontell, K. Mol. Cryst. Liq. Crysf. 1981, 63, 59. (4) Charvolin, J. J . Phys. (Puris), C3 1985, 46, 173. ( 5 ) Lindblom, G.;Rilfors, L. Biochim. Biophys. Acto, in press. (6) Balmbra, R. R.; Clunie, J. S.; Goodman, J. F. Suture 1969, 222, 1159. (7) Bull, T.; Lindman, B. Mol. Crysr. Liq. Cryst. 1974, 28, 155. (8) Lindblom, G.;Wennerstrom, H. Biophys. Chem. 1977, 6, 167. (9) Eriksson, P. 0.;Khan, A,; Lindblom, G. J . Pfiys. Chem. 1982.86, 387. (10) Fox, K. K. Mol. Cryst. Liq. Cryst. (Lett.) 1983, 92, 135. (11) Fontell, K.; Fox, K. K.; Hansson, E. Mol. Cryst. Liq. Crysr., Lett. Secr. 1985, I , 9. ( 1 2 ) Eriksson, P. 0.;Lindblom, G . ;Arvidson, G. J . Pfiys. Cfiem. 1985,89, 1050. ( 1 3) Eriksson, P. 0.;Lindblom, G.; Arvidson, G. J. Phys. Cfiem. 1987, 91,
846. (14) Soderman, 0.;Walderhaug, H.; Henriksson, U.;Stilbs, P. J . Pfiys. Chem. 1985,89, 3693. ( 1 5 ) Ekwal, P.; Mandell, L.; Fontell, K. Mol. Cryst. Liq. Cryst. 196?3,8,
157. (16) Mitchell, D. J.; Tiddy, G.J. T.; Waring, L.; Bostock, T.; McDonald, M . P..I. Chem. Soc., Faraday Trans. I 1983. 79. 975.
0 1990 American Chemical Society
The Journal of Physical Chemistry. Vol. 94, No. 13. 1990 5309
Structure of Cubic Mesomorphic Phases 0 II
-
CHn 0 - C (CH& ~
I : :I::
-
(0- CH, CH,). OH I CH (CH& CHI ~
- -
-
(0- CH2 - CH& I
- OH
-
CH - 0 - C (CH2)m CH - (CHt)r CH, ~
(0 C K - Cb). ~
- OH
I
CHa - 0 - C - CHI)^ - CH - (Cb)s - CHa Fimre I. The molecular structure of HCO-60: x + y + z = 60.
of these systems has the structure been firmly established by X-ray diffraction, but NMR indicates a discontinuous hydrocarbon phase with possibly anisometric micelles."J* A recent TEM study of freezefracture replicas of oleoyldecalds(e1hyleneoxide)I9revealed what seemed to be a cubic lattice of nearly spherical micelles. The main uncertainty regarding the structure of the Pm3n cubic phase is the nature of the basic micellar elements. i.e., whether they are spherical, cylindrical, or distorted globular aggregates. Only transmission electron microscopyz0 can provide direct high-resolution images of these "building blocks", and two-dimensional projections of the phases they make up. However, this technique has stringent requirements on specimen preparation, and thus, complementary indirect methods are needed. X-ray scattering from the bulk phase involves no special pretreatment. However, the interpretation of X-ray diffraction data requires a model of the proposed structure. The optimal combination that we demonstrate here is to use electron microscopy to suggest the nature. size, and packing symmetry of the structural building blocks. to determine the exact symmetry by X-ray scattering, and lo establish the parameters of the unit cell by adjusting the model to satisfy scattering intensities.
Materials and Methods For our study we chose as a model system the nonionic surfactant HCO-60 (Nikko Chemicals Co., Tokyo), because it forms large micelles that can be relatively easily visualized directly. HCO-60 is a derivative of castor oil (92% triolein) with a chain of approximately 20 ethylene oxide groups attached to carbon number 9 of the fatty acid chain (Figure I). This surfactant is rather well-known in the pharmaceutical industry. where it is used to solubilize a variety of drugs?' Unfortunately, no detailed information on the phase diagram of this surfactant is available. HCO-60 forms micellar solutions with water up to moderately high concentrations (approximately 20 w i %). Linoleic acid (LA) can be solubilized in these micelles; it causes size increase and improves cosolubilizaton of other materials?' Direct imaging by electron microscopy was also done on micellar solutions and cubic phases of L-a-monopalmitoyllecithin(Sigma, USA). Direct images of the investigated systems were obtained by the technique described by Bellare et aLZ2 The technique is based on vitrifying thin films of the liquid system by ultrafast cooling in liquid ethane at its freezing point, and examination of the vitrified specimen direcfly in the transmission electron micraswpe (TEM). Specimen preparation and handling were done under controlled conditions (using the amtrolled environment vitrification system (CEVS)) to avoid microstructural changes in the system. The vitrified specimens were examined either in a JEOL l2OCX or a Hitachi H-500 microscopes, using Gatan cold-stage systems. Aceleration voltage was 100 kV. Micrographs were recorded on Kodak SO-163 film developed to maximum speed in full-strength Kodak D-19developer. (17) Nilrsan, P. G.: Wenncntrh. H.; Lindmsn. 8.1. Phys. Chem. 1983. 87. 1311. (18) Nilsson. P. G.; Wennentr7" H.: Lindman. B. Chem.Scr. 1985.25, 67
(19) Jousma, H.: Bouwstra. J. A.; Spies. F.; Junginger. H. E. Colloid Polym. So'. 1987. 265. 830. (20) Talmon. Y. ColloidsSurf. 1986. 19, 237. (21) Takada. K.; Yorhimura. H.: Yorhikawa. H.; Muranishi. S.;Yarumura. T.;Takahim. 0. Phorm. Re$. 1986, 3. 48. (22) Bellarc. J. R.: Davis. H. T.; Scrivcn. L. E.;Talman. Y . J. Elecrron Mirmrc. Tmh. 1988. IO. 87.
Figure 2. Cry-TEM micrograph of a vitrified spcimen (-170 "C) of 2 w t W HCO-60 and 0.6% linoleic acid aqueous solution, showing individual globular micella. Specimen was quenched from 22 OC. Bar = 100 nm.
We also used freeze-fracture replication electron microscopy as described by Burns and Talmon?' In this method a layer of the liquid is vitrified under fontrolled conditions. The vitrified specimen is then fractured, and a platinum-carbon replica of the fractured surface is prepared. We used a Bakers 400T f r e e z e fracture apparatus operated a t -160 "C. The replicas were examined a t room temwrature in a Hitachi I2A transmission electron microscope. SAXS oatterns were recordedwith a modified Rizaku-Denki small-angie X-ray camera and a homebuilt linear picon-sensitive detector of the delay-line type?' Cu-Ka radiation from a fine-focus sealed-tube generator was used. Monochromatization was provided with a nickel filter and one Franks mirror, and the beam was collimated to a height of 4 mm and a width of about 350 pm in the plane of the sample. The samples, placed in q u a m capillaries, were mounted parallel to the height of the beam. Sample-tdetector distance was 460 mm and e x p u r e times were about IO h. Background scattering due to short-range density fluctuations within the water phase, as well as parasitic scattering, were removed from the experimentally measured intensity by subtracting the intensity scattered from a water-filled capillary, scaled to fit the scattering from the samples at the wide-angle tail of the scattering pattern. The measured intensities were corrected for the finite height of the incident beam by using the method of Lake?l
Results and Disflrssion Figures 2 and 3 show direct images of vitrified specimens of micellar solutions of the HCO-60/LA/water system. Individual micelles are seen in Figure 2; they seem to be globular, not spherical. The mesomorphic phase that is formed a t higher concentration and a t room temperature is too viscous to allow specimen preparation for direct imaging. To overcome this problem we took advantage of the phenomenon of suspended aggregates being locally concentrated during spaimen preparation, due IO flow patterns that develop in the specimen during its thinning.20~zzThus, under properly chosen conditions we can produce the high-viscosity phase on the electron microscope grid itselF; this is an example of "on-the-grid processing", a concept we have used elsewhere to prepare specimens of phases that cannot be prepared directly as thin fluid films?6~z8 Figure 3 shows the (23) Burns. J. L.;Talmon. Y . 1.Eltcrmn Micmc. Tech. 1988, IO. 113. (24) SAXS measurements were performed at the Low Angle X-Ray Scatterinn L a b " . the Polvmn Research Deoanmenl. Weizmann In( 2 5 ) Lake. J. A. Aria Crysrollog?. 1967. 23. 191.
(26) Talmon.
Y.: Burns, 1. L.; Chestnut. M.H.:SicgeI, D. P. J. E l m m n
Micrmc. Tech. 1990. 14. 6. 6. ( 2 1 ) lnrernarional Tables for X-ray. Crysrollopphy: Kynoeh . .
Birmingham. U.K..1952; Vol: Vol. 1. ((28) 2 8 ) Sicgel. D. P.;Burns. J. L.:Chestnut, 1989. 56. 161.
Ress:
M.H.: Talmon. Y . Blophys. J.
5310 The Journal of Physical Chemistry, Vol. 94. No. 13. 1990
viscous phase that is made of globular micelles packed into arrays. The twodifferent projections (Figure 3 , a and b) show how the apparent symmetry of the various "grains" changes with specimen tilt relative to the beam. The patterns of rows and local areas of hexagonal symmetry indicate a cubic structure. The specimen
Figure 4. Freezefracturereplica of a thermally fixed sample of 24 micellea in the fracture face of the cubic phase. Bar = 100 nm.
wt %
Burns et al. shown in Figum 2 and 3 were prepared by vitrifying liquid sample originally at 22 OC. Similar micrographs were obtained from samples "quenched" from 0 'C. Freezefracture replicas can be made of bulk specimens of the viscous phase, with no n d for on-the-grid processing. They too show the cubic packing of globular micelles as is demonstrated in Figure 4. To quantitatively evaluate the structural parameters of this particular cubic phase, we used small-angle X-ray scattering (SAXS). The scattering patterns are shown in Figure 5 (curves a, b. and d), in which four diffraction peaks can be seen. The positions of the observed peaks, as well as the ratios of the peak positions relative to that of the first peak, are given in Table 1. These peak positions are in accord with the ratios expected for the Pmjn or P43n space groups. The dimension of the unit cell can be calculated from the position of the first peak (1.1.0): 18.0 and 19.0 nm for the 32% HCO-60 at 24 and 0 'C, respectively. and 25.5 nm for 24% HCO-60/7.5% LA at 0 OC. At 24 "C the HCOdO/LA/water system exhibited only one clear interference peak (curve c). A structural model of the cubic phase, with predictions of the shape. size, and positions of the structural elements, cannot be obtained by a single experimental technique. Refinement of a structural model based on scattering measurements exhibiting only four peaks is not possible. However, TEM yields direct evidence that the basic structural elements of this cubic phase are globular (but not perfectly spherical) micelles with diameters ofca. 6 nm for HCO-60 alone, and 8 nm with solubilized linoleic acid. The positions of the diffraction peaks in the SAXS patterns indicate the space group and the unit cell dimensions. Further structural analysis concerns determining the number and positions of the spherical micelles within the unit cell, as given by the equivalent positions of one of the general or special cases of space groups PmSn or P 4 h . The special positions k of P m h , with 24 micelles per unit cell, are chosen because a reasonable volume fraction (about 30-4070) is predicted by using the sphere diameters and cube dimensions mentioned above. No further extinctions are imposed on the observed reflections. and the coordinates of the special positions have only two adjustable parameters (the special positions k of h 3 n are equivalent to the general positions of P43n with one of the three coordinate parameters equal to zero). Our suggested model is shown schematically in Figure 6. Twenty-four micelles are located at the special positions k of P m h 4 micelles are centered on each face of the cubic unit cell, while
HCO-60.7.5% linoleic acid in water, originally at 0 OC,showing globular
Structure of Cubic Mesomorphic Phases
The Journal of Physical Chemistry, Vol. 94, No. 13, 1990 5311
TABLE I: Positions of the Scattering Peaks (S = 2 sin @/A [A-']) and Their Ratio Relative to the Position of the First Peak 24 wt % HCO-601 7.5 wt % LA 0oc
32 wt % HCO-60
0 "C
24 'C
1.41 1.58 1.73
0.01 1 40 0.01300 0.01430
1.43 1.63 1.79
0.01030 0.01 170 0.01300
1.36 1.55 1.72
0.007 55 0.008 95 0.009 70
1.36 1.61 1.75
TABLE I 1 Comparison of the Measured Intensities with Those Calculated bv Using the Model Structure 32% HCO-60 24 OC 0 OC (h,k,l) measd calcd measd calcd l,l,O 2,0,0 2,1,0 2,1,1
280 64 53 51
400 131 125 164
290 62 47 49
408 133 131 147
+
24% HCO-60 7.5% LA 0 OC measd calcd 618 600 55 57 48 35 30 59
TABLE I11 Fitted Parameters for the Model Structure of the Cubic Phase D,Aa,A y z a,A2 32 wt % HCO-60, 24 OC 32 wt % HCO-60, 0 OC 24 wt % HCO-60 7.5 wt % LA, 0 "C
+
00
0
001
s ik' 1
60 60 80
180 0.170 0.265 10000 190 0.170 0.267 4000 255 0.165 0.255 20000
scattered by the model structure to the (h,k,l) peaks were calculated as follows:
Figure 5. SAXS patterns: 32 wt % HCO-60 at 24 OC (a) and 0 OC (b); 24% HCO-60, 7.5% linoleic acid at 24 OC (c) and 0 OC (d).
Ic(h,k,l) = m(h,k,l)J2(S)lF(h,k,I)l2 exp(-aS2)/S2
(1)
where m(h,k,l) is the multiplicity of the (h,k,l) peak and S is the scattering vector
+ k 2 + 12)'/*
S = 2 sin O/X = (l/a)(h2
(2)
where X is the wavelength of radiation and 20 is the scattering angle. f(S) is the shape factor of the micelle. As a first approximation we assume the micelles to be perfect spheres of uniform electron density. This assumption is reasonable for the nonionic ethylene oxide-based surfactants, as compared with ionic surfactants for which a core-shell distribution of electron density is more appropriate. A
n A
lolo
where i l
4 y.a L Figure 6. The model cubic P m h structure showing location of micelles on two external faces, and on a midplane (A-A section) of the cube.
12 micelles are located within the cube, so that 4 are centered on each midplane perpendicular to a principal axis. The coordinates of the special positions are given in ref 27, structure no. 223. The intensities of the scattering peaks (measured as the peak height) were compared to calculations based on the suggested model. Five parameters are used in this calculation: D , average micelle diameter; a, cubic unit cell dimension; y and z, parameters of special positions k of PmJn for location of the micelles (given as fractons of a); a,a crystal distortion parameter. The intensities
f(S) = (sin
+ - fi cos
(3)
+ = TDS. The interference factor F(h,k,l) is given by F(h,k,l) = Cexp[-2ri(hXn n
+ kYn + lz,)]
(4)
in which (X,Y,Z),are the fractional coordinates of the 24 micelles within the unit cell. They are given in terms of the y and z parameters by the special positions k of P m h 2 ' The factor ( l/S2) is a Lorentz correction factor. Calculated and measured intensities of the scattering peaks are compared in Table 11. The calculated intensities were normalized so that their sum equals that of the four reflections. The values of the parameters used in the calculation are given in Table 111. Micelle diameters are based on TEM results, and the unit cell dimension was taken as determined from the peak positions. The fit was performed by considering increasing values of the distortion parameter cy and varying the coordinate parameters y and z . The relative intensities of the four peaks, in particular the dominance of the first peak, are very sensitive to small variations in t h e y and z parameters. As a state of higher symmetry is approached, such as the case where y = z = 0.25 (discussed below) the interference factor of the (l,l,O)peak rapidly diminishes. T h e y
5312 The Journal of Physicar Chemisfry. Vol. 94. No. 13. 1990
Z L-a-monopalmitoyllecithinin water. originally at 0 OC. showing individual globular micdlcs. Bar = 100 nm. F l y r e 7. Vitrified specimen of 10 wt
and L parameters. which were obtained from the fit, satisfy the requirement that the micelles do not overlap. That the values of these parameters for the three cases are almost identical is intriguing. At present we do not have sufficient evidence to link this to the nature of interactions between the micelles. The distortion parameters naxssary to fit the data are large. Indeed the disorder in the systems under study is significant. as evident qualitatively from observations by TEM. Several types of disorder exist in the system and contribute to the decrease in intensity of the higher order peaks. All disorder effects are taken into a a u n t in this analysis, as a first approximation, by the single parameter u in the functional form used in eq I . It is beyond the scope of this analysis to account in greater detail for the disorder effects. The main result of this analysis is in demonstration that a cubic structure of spherical micelles, of the size observed by TEM. is compatible with the measurements of the scattered intensities. It has been predicted that spherical micelles of molecules with long ethylene oxide chains retain their spherical shape at high wncentrations.'+" Moreover, Mitchell et aLt6even suggested that those micelles would aggregate and form primitive cubic rather than close-packed structures. In our system they and z parameters are 0.17 and 0.26. If these were 0.25 and 0.25, the PmJn structure would degenerate into eight identical c u b s each containing three micelles (an 'A-B-centered cube"), Le.. an intermediate between a bcdyentered cube wntaining two spheres. and a faceentered cube having four spheres per cube. Mitchell et aLt6 suggested for dodecyldodecakis(ethy1ene oxide) a cubic phase that is a mixture of bce and fa structures. A single structure as given by our model may account for such observations. HCO-60 served as a very useful model system because of the large micelles it forms. However, the system is not isomerically pure. Thus we have started to study much better defined cubic phases. One such system is L-u-monopalmitoyllecithinin water, for which a cubic Pm3n phase has been identified.2 Figure 7 is a micrograph showing globular micelles of L-a-monopalmitoyllecithin in water, and Figure 8 shows that the spherical micelles pack a t higher wncentrations into a cubic phase. This indicates that the model of a cubic Pm3n phase formed of spherical micelles is not restricted to the HCO-60 systems. Further studies on such cubic phases are currently in progress.
Burns et al.
F l y r e 8 . Micrograph of a vitrified aqueous solution of 25 wt % L-amonopalmitoyllecithin (originally at 0 "C). Note the 'grains" of the cubic phase and the "latticc fringes" from the vitrified liquid-crystalline structures. Bar = 100 nm.
Conclusions We have demonstrated here the application of a wmbination of direct, cryo-TEM, and indirect, SAXS, techniques for the determination of the microstructure of a model cubic phase. Two electron microscopy techniques were used. Direct imaging of vitrified specimens enabled us to wllect several projections from the same volume of a thin film (this is related to three-dimensional data), whereas freeze-fracture-replication provided views of a fractured surface of a thermally fixed 'bulk" phase. The combination of the two electron microscopy techniques gave a clear indication of the nature of the system under investigation. and of the "building blocks" from which the system is composed. Without that information the interpretation of the quantitative data from indirect methods is not unique and can lead to erroneous models such as the one by Tardieu and Luzzati? With the information from electron micrographs wmbined with the quantitative data from SAXS we were able lo construct a reliable model for the mesophase we studied. HCO-60 based systems were useful to demonstrate our approach, because they are built of relatively large micelles that can be rather easily visualized by electron microscopy. We have shown here that our approach can be used with more wmmon systems of smaller building elements. In fact, these chemically and isomerically cleaner systems, such as the lysolecithins that we are currently studying. show much more order. as demonstrated by the sample micrograph of Figure 8. Our approach of combining direct and indirect microstructural analysis methods is quite general and wuld be taken whenever microstructured systems of synthetic or biological origin are studied. Acknowledgment. We thank Dr. Ellen Wachtel of the Weizmann Institute of Science for h a invaluable help with the SAXS work and for her wmments to the manuscript, and Mr. Alan E. Bruns of the Procter & Gamble Co. for technical help with the HCO-60 system. This project was supported by grants to one of us (Y.T.)from the United States-Israel Binational Science Foundation (BSF). Jerusalem, from the Fund for the Promotion of Research at the Technion, and from the Technion V.P.R. Fund-Albert Einstein Research Fund.
