Worm-like Micelles of CTAB and Sodium Salicylate under Turbulent

Langmuir 2008, 24, 13875-13879

13875

Worm-like Micelles of CTAB and Sodium Salicylate under Turbulent Flow Roberta K. Rodrigues, Marcelo A. da Silva, and Edvaldo Sabadini* Department of Physical-Chemistry, Institute of Chemistry, UniVersity of Campinas - UNICAMP, P.O. Box 6154, 13084-862, Campinas, SP, Brazil ReceiVed September 3, 2008. ReVised Manuscript ReceiVed October 6, 2008 Polymers with high molecular weight and worm-like micelles are drag-reducing agents under turbulent flow. However, in contrast to the polymeric systems, the worm-like micelles do not undergo mechanical degradation due to the turbulence, because their macromolecular structure can be spontaneously restored. This very favorable property, together with their drag-reduction capability, offer the possibility to use such worm-like micelles in heating and cooling systems to recirculate water while expending less energy. The formation, growth, and stability of worm-like micelles formed by cetyltrimethylammonium bromide (CTAB) and sodium salicylate (NaSal) were investigated using the self-fluorescence of salicylate ions and the ability of the giant micelles to promote hydrodynamic drag reduction under turbulent flow. The turbulence in solutions of CTAB-Sal was produced within the double-gap cell of a rotational rheometer. Detailed diagrams were obtained for different ratios of Sal and CTAB, which revealed transitions associated with the thermal stability of giant micelles under turbulent flow.

1. Introduction Surfactants in solution will spontaneously form supramolecular aggregates (when above the critical micellar concentration, cmc), and the micellar morphology covers a large range of shapes and sizes.1 In solutions with low surfactant concentration and absence of other cosolutes, the aggregates will form roughly spherical structures. However, changes in the surfactant concentration, ionic strength, or addition of binding cosolutes can induce a uniaxial growth leading to the formation of long and flexible cylindrical micelles, usually denominated worm-like or threadlike micelles.2 These micellar systems display a rheological behavior analogous to that of flexible polymers solutions. However, the worm-like micelles structure is maintained by a delicate balance of transient molecular interactions in a dynamic system where the micelles are continuously breaking and reforming within a finite time scale.3 This particular property has attracted the attention of researchers interested in the dragreduction phenomenon, also named the Toms Effect. B. A. Toms in 1948 found that a very dilute high-molecular weight polymer solution under turbulent flow required a lower pipe flow pressure gradient than did the pure solvent to produce the same flow rate.4,5 Drag-reduction levels may reach 80% under laboratory conditions.2 Consequently, the phenomenon has become of considerable engineering interest, mainly in pumping processes.6-10 * Corresponding author. E-mail: [email protected]. (1) Israelachvili, J. N.; Mitchell, D. J.; Ninham, B. W. J. Chem. Soc., Faraday Trans. 2 1976, 72, 1525–1568. (2) Ezrahi, S.; Tuval, E.; Aserin, A. AdV. Colloid Interface Sci. 2006, 128-130, 77–102. (3) Hoffmann, H.; Ebert, G. Angew. Chem., Int. Ed. Engl. 1988, 27, 902–912. (4) Toms, B. A. Proceedings of the 1st Congr. on Rheology; North-Holland Publishing Co.: Amsterdam, 1948; pp 135-141. (5) Virk, P. S.; Merrill, E. W.; Mickley, H. S.; Smith, K. A.; Mollo-Christensen, E. L. J. Fluid Mech. 1967, 30, 305–328. (6) Alkschbirs, M. I.; Bizotto, V. C.; Oliveira, M. G.; Sabadini, E. Langmuir 2004, 20, 11315–11320. (7) McCormick, C. L.; Hester, R. D.; Morgan, S. E.; Safieddine, A. M. Macromolecules 1990, 23, 2132–2139. (8) Bailey, F. E.; Koleske, V. J. Poly(ethylene oxide); Academic Press: New York, 1976. (9) Figueredo, R. C. R.; Sabadini, E. Colloids Surf., A 2003, 215, 77–86.

According to Tabor and de Gennes,11 the Toms Effect can be explained by the interaction of the polymer chain with the small vortices created within the turbulent flow. The process of stretching-contraction of the polymer chain affects the evolution of the vortices cascade (which dissipates the kinetic energy of the fluid) by storing some of the turbulence energy in the chain. Therefore, drag-reduction additives have essentially long and flexible molecular structures, for example, high molecular weightpolymers or worm-like micelles.12 Worm-like micelles have an additional advantage when used to reduce pumping energy costs in recirculating systems. Polymeric additives are irreversibly mechanical degraded during the turbulent flow, whereas micellar systems can recover their macromolecular structure and maintain their ability to promote drag reduction.13 One of the most studied worm-like micelle systems is that based on cetyltrimethylammonium (CTA) and salicylate (Sal) salts. The micelles grow due to the spontaneous insertion of Sal between CTA. The majority of the studies involving CTA-Sal used concentrations higher than 50 mM, for which the Toms Effect is very expressive.14 However, few articles are dedicated to the formation, growth, and stability of such a system in the diluted regime. The main purpose of this study is to investigate the formation and growth of the CTA-Sal worm-like micelle in the diluted regime, as close as possible to the CTA cmc. In this concentration regime, the formation and thermal stability of the delicate supramolecular structures were investigated using the selffluorescence of the salicylate ions and the drag-reduction capability of the giant micelles. The turbulent flow and the consequent hydrodynamic drag effects were studied in a double(10) Kulicke, W. M.; Gra¨gem, H.; Ko¨tter, M. Drag Reduction Phenomenon with Special Emphasis on Homogenous Polymer Solutions - Polymer Characterization/Polymer Solutions; Springer-Verlag: Berlin, 1989. (11) Tabor, M.; de Gennes, P. G. Europhys. Lett. 1986, 2, 519–522. (12) Zakin, J. L.; Lu, B.; Bewersdorff, H. W. ReV. Chem. Eng. 1998, 14, 253–320. (13) Smith, B. C.; et al. In Structure and Flow in Surfactant Solutions; Herb, C. A., Prud’Homme, R. K., Eds.; American Chemical Society: Washington, DC, 1974; Chapter 26. (14) Lu, B.; Zheng, Y.; Davis, H. T.; Talmon, Y.; Zakin, J. L. Rheol. Acta 1998, 37, 528–548.

10.1021/la802890x CCC: $40.75  2008 American Chemical Society Published on Web 11/18/2008

13876 Langmuir, Vol. 24, No. 24, 2008

Rodrigues et al.

gap cell of a rotational rheometer, which allowed the determination of the stability of the worm-like micelles under flow.

2. Material and Methods 2.1. Materials. Cetyltrimethylammonium bromide (CTAB) and sodium salicylate (NaSal) were obtained from Sigma and used without any further treatment. The samples were prepared by dilution of previously prepared stock solutions. The stock solutions were prepared by weighting (within (0.0001 g) adequate quantities of CTAB or NaSal as required. Sodium hydroxide was added to the samples to keep the pH ) 10. Ultrapure water (18.2 MΩ cm) was used throughout. 2.2. Methods. Rheological Measurements. All rheological experiments were conduct with a Haake RheoStress 1 rheometer equipped with a double-gap cell. The cup internal and external diameters were 17.75 and 21.70 mm, respectively, and the rotor internal and external diameters diameter was 18.35 and 20.99 mm, respectively. The rotor length was 55.00 mm. Temperature was controlled by an external water-bath system with a precision better than 0.1 °C. Flow curves were obtained over an angular velocity range of 0-920 rpm. Special care was taken to avoid foam interference. The temperature sweep experiments were measured over a temperature range of 25-75 °C in 3900 s at a fixed angular velocity of 612 or 921 rpm. Fluorescence Measurements. The fluorescence experiments with the solutions containing CTAB and sodium salicylate in different ratios were carried out with a Perkin-Elmer luminescence spectrometer, model LS 55. The excitation wavelength used was 348 nm, and the spectra were obtained between 360 and 540 nm.

3. Results and Discussion The Toms Effect in a worm-like system is necessarily associated with the viscoelasticity of the solution, and the range of concentrations and proportions in which the CTAB-NaSal solution behaves as a viscoelastic fluid was determined using a large number of samples (ca. 200). In these experiments, a very simple procedure was carried out that consists of swirling vials containing the solutions and observing the movement of the small air bubbles trapped in the solution.15 If the solution is nonviscoelastic, the bubbles stop as soon as the circular movement of the vial is interrupted. However, when the solutions are viscoelastic, the bubbles move in an opposite trajectory (backward) when the swirling is interrupted. This simple test allows the determination of the boundary between viscoelastic and nonviscoelastic solutions for a large number of samples with different concentrations and temperatures, as shown in the diagram of Figure 1. The Toms Effect was studied as close as possible to the critical micellar concentration of CTAB (0.8 mmol L-1 at 25 °C).16 According to the data in Figure 1, the nonviscoelastic region extends to lower CTAB concentrations at a constant Sal concentration as the temperature is raised. The boundary is roughly linear with a constant ratio between NaSal/CTAB (defined as ξ) at each temperature. At 25 °C, ξ = 0.4, and when the temperature is increased, higher amounts of NaSal are needed to maintain the viscoelastic behavior, resulting in an increase in the values of ξ. Αt 45 °C, for example, the boundary is observed at ξ ) 0.6, and at 55 °C, only the nonviscoelastic behavior is observed. Rod-to-spheres transitions are observed when the temperature of solutions of CTAB-Sal is increased.17,18 (15) Nash, T. J. Appl. Chem. 1956, 6, 539–546. (16) Gamboa, C.; Rios, H.; Sepulveda, L. J. Phys. Chem. 1989, 93, 5540– 5543. (17) Yoneda, M.; Endoh, K.; Suga, H.; Hirata, H. Thermochim. Acta 1996, 289, 1–7. (18) Faetibol, E.; Michels, B.; Waton, G. J. Phys. Chem. 1996, 100, 20063– 20067.

Figure 1. Phase diagram for CTAB-NaSal solutions at several temperatures. The regions of the diagrams are classified in terms of the rheological (viscoelastic and nonviscoelastic) behavior of the studied samples. The values for the ratios ξ ) [NaSal]/[CTAB] obtained from the fluorescence study at the transition (at 25 °C) are also indicated.

Because the Toms Effect depends on the presence of larges structures, the diagram was useful to delineate the ratios between the two components and the temperature range in which dragreduction phenomenon can be effective. The progressive incorporation of salicylate molecules into the micelles and their consequent growth were studied using the self-fluorescence of this cosolute. In principle, a fluorescent molecule that has its movement restricted can be more fluorescent, as the nonradioactive processes such as diffusion and collision decrease. The fluorescence spectra of fixed concentrations of sodium salicylate (0.5, 0.7, 0.9, and 1.2 mmol L-1) were obtained (not shown) in the presence of CTAB in the range of concentration between 0 and 4.5 mmol L-1 at 25 °C. The cmc for CTAB-NaSal in this NaSal concentration range is below 0.05 mmol L-1. Figure 2 shows the variation of the fluorescence intensity of salicylate at 406 nm, as a function of CTAB concentration. As expected, the intensity of the fluorescence (comparing the four solutions of NaSal) is higher for more concentrated solutions of the cosolute. However, for each fixed concentration of NaSal, the intensity of the fluorescence increases as CTAB is added. For example, the fluorescence of the solutions of NaSal in the presence of 4.5 mmol L-1 of CTAB is practically twice as high as that for solutions without CTAB. The curves can be analyzed in two CTAB concentration ranges. At low CTAB concentrations (in which ξ is high), the intensity of the fluorescence of salicylate increases sharply with the CTAB concentration, and at high concentrations (in which ξ is low), the intensity increases much more slowly with the concentrations of CTAB. As indicated in the plots, the transition between the two regimes is observed at a critical value of ξ; ξc = 0.3, 0.4, 0.5, and 0.7, respectively, for the concentrations 0.5, 0.7, 0.9, and 1.2 mmol L-1 of sodium salicylate. These four sodium salicylate concentrations are indicated by the dot arrows in the diagram of Figure 1. The tip of the arrow corresponds to ξc (determined from the fluorescent study) for each solution. The transitions in the fluorescence measurements are smooth with ξc in the range between 0.4 and 1.0, which is close to the ξ corresponding to the boundary for the viscoelastic-nonviscoelastic transition (Figure 1). Therefore, the transition observed by fluorescence may also be associated with changes in the size of

Worm-like Micelles of CTAB and Sodium Salicylate

Langmuir, Vol. 24, No. 24, 2008 13877

Figure 2. Intensity (in arbitrary units) of fluorescence for salicylate solutions (0.5, 0.7, 0.9, and 1.2 mol L-1) as a function of CTAB concentration. The values for the ratios ξ ) [NaSal]/CTAB] at the transition are also indicated. ξ decreases from left to right.

the worm-like micelles. For ξ > ξc, the association of Sal and the surfactant molecules can induce the formation of longer micelles, which is responsible for the viscoelastic behavior of the solution (Figure 1). For ξ < ξc, the balance between the CTAB and sodium salicylate concentrations results in a transition of the worm-like micelles to smaller structures, because the incorporation of salicylate molecules is reduced. The solutions of CTAB-NaSal were submitted to turbulent flow, to verify correlations between ξ and the Toms Effect. The turbulent flow was created in a rheometer containing a doublegap cell. Experiments in such system are very suitable for the study of the Toms Effect, mainly due to its high accuracy ((3%).19,20 A detailed description of the technique can be obtained elsewhere.20 Basically, a flow curve is obtained by shearing a solution in a range of angular velocity (Ω) and measuring the correspondent shear stress (τ). The apparent viscosity (ηapp) can be defined as the slope of the flow curve as a function of Ω, which is a direct measurement of the hydrodynamic drag. For a Newtonian fluid, after the laminar flow regime (characterized by a plateau, in which ηapp is the shear viscosity), ηapp increases gradually as the angular velocity increases. In the experimental setup used, the laminar flow for water was observed at up to ca. 138 rpm (see Figure 3A). The increase in the apparent viscosity for water is due to the dissipative flow structures. Beyond 138 rpm, the sharp increase in the apparent viscosity is due to the Taylor vortices (which consist of two counter-rotating pairs of vortices overlapped with the Couette flow).21,22 On increasing the angular velocity, the Taylor vortices become wavy and the flow field eventually becomes chaotic and turbulent if the velocity is increased further.19 The pattern of the flow curve in the turbulent regime is changed if a drag-reducing agent (such as some ppm of a polymer with very high molecular weight) is present in the solution.20 The longer macromolecular chains absorb the energy of the vortices and inhibit the development of the turbulent cascade, which results in a lower torque (directly associated with ηapp) relative to that applied in the pure solvent (see, for instance, ref 20). (19) Nakken, T.; Tande, M.; Elgsaeter, A. J. Non-Newtonian Fluid Mech. 2001, 97, 1–12. (20) Bizotto, V. C.; Sabadini, E. J. Appl. Polym. Sci. 2008, 110, 1844–1850. (21) Groisman, A.; Steinberg, V. Phys. ReV. Lett. 1996, 77, 1480–1483. (22) Taylor, G. I. Proc. R. Soc. London 1936, A157, 537–546.

Figure 3. Flow behavior of CTAB-NaSal solutions at 25 °C with different NaSal/CTAB ratios (ξ) and CTAB concentrations of (a) 2.0 mmol L-1, (b) 3.0 mmol L-1, and (c) 5.0 mmol L-1. The standard deviation for ηapp is (3%.

For CTAB-NaSal, the flow curve was obtained as a function of the concentrations of both components (Figure 3). In these experiments, the CTAB concentration was kept fixed at 2.0, 3.0, and 5.0 mmol L-1, and the concentrations of NaSal were changed to obtain ξ ) 0.2, 0.6, 1.0, and 2.5. For the solutions of CTAB containing NaSal with ξ > 0.6, the pattern of the curves changed drastically in comparison to the curve for pure water. In the range of angular velocities >300 rpm, the apparent viscosity is practically constant, showing the high performance of the worm-like micelles to promote drag

13878 Langmuir, Vol. 24, No. 24, 2008

reduction. In comparison, for a polymeric drag-reducing agent, the apparent viscosity is lower than that for the pure solvent, but it still grows smoothly as the angular velocity increases.20 The difference in the flow curve for polymers and worm-like micelles can be rationalized by considering the size and concentration of both drag-reduction agents. For polymers with very large molecular weight, the critical concentration necessary to observe the drag-reduction effect is very low (lower than 100 ppm), which results in very low shear viscosity.23 The persistence length of worm-like micelles is in the range from 100 to 400 Å, and charged micelles can even reach micrometric lengths, which is longer than the persistence length for polymers.24 However, to form worm-like micelles, higher concentrations of surfactant are necessary. Therefore, in this case, the large structures and the relatively high concentrations yield solutions in which the viscoelastic behavior is very pronounced. This means that solutions of worm-like micelles are not only elastic, but also viscous (high shear viscosity), resulting in flow curves different from those for polymers in drag-reduction conditions. In the laminar flow, the apparent viscosity for the worm-like micelles is 4-5 times higher than that for pure water, and the onset point for the Taylor vortices cannot be determined. Considering the differences in the viscosity, the Reynolds number should be used, to compare the different systems. However, it is difficult to determine, as the structure of the micelles changes depending on their composition and rotation rates. The micelles are very effective in reducing the hydrodynamic drag. For example, for 2.0 mmol L-1 CTAB with ξ ) 0.6, the apparent viscosity for Ω > 750 rpm is even lower than that for pure water, despite the relative high shear viscosity of the solutions (shown in the inset of Figure 3A). The experiments cannot be extended to angular velocity higher than 920 rpm, where effects of larger magnitude are expected, due to the mechanical instability of the flow. Beyond this angular velocity, the solution is centrifuged out of cup, and foam formation is observed. For ξ > 0.6, the apparent viscosity is practically independent of the ratios and concentrations of CTAB and NaSal. We can infer that this is associated with the continuous process of breaking (due the turbulent flow) and self-reforming of the worm-like micelles. For ξ ) 0.2 and 0.4, the flow curves for solutions with 2.0, 3.0, and 5.0 mmol L-1 of CTAB are practically superimposed on the flow curve for water, indicating no evidence of drag reduction. Based on the Toms Effect of the CTAB-NaSal solutions, the transition from a noneffective to effective capability to promote drag reduction is observed when ξ ) 0.6. The result is quite interesting because, at this proportion, it can be suggested that the flow induces the formation of worm-like micelles, characterized by the pronounced increase in ηapp, as Ω is increased. The onset point for this transition is 99, 69, and 46 rpm for 2, 3, and 5 mol L-1 of CTAB, respectively. In shear flow under appropriated conditions, nonlinear flow phenomenon can be observed and is usually attributed to the so-called shear-induced structures (SIS).25 The nature of SIS is not yet fully understood,26 but has (23) Gyr, A.; Bewersdorff, H. W. Drag Reduction of Turbulent Flows by AdditiVes (Fluid Mechanics and Its Applications); Springer: Berlin, 1995. (24) Dreiss, C. A. Soft Matter 2007, 3, 956–970. (25) Richtering, W. Curr. Opin. Colloid Interface Sci. 2001, 446–450. (26) Miller, E.; Rhonstein, J. P. J. Non-Newtonian Fluid Mech. 2001, 143, 22–37.

Rodrigues et al.

Figure 4. Apparent viscosity of 2 mmol L-1 CTAB solutions containing NaSal with ξ ) 0.6, 0.8, and 1.2 as a function of the temperature. The flow curve for pure water is also included as a reference. The angular velocity (Ω) was fixed at 612 rpm.

been attributed to micellar growth,27-29 flow instabilities,30 phase transition,31 or flow-induced gelation.32 The flow curves for CTAB and NaSal were obtained at different temperatures, to investigate the thermal stability of the wormlike micelles under drag-reduction conditions.33,34 Figure 4 shows the flow behavior of 2.0 mmol L-1 CTAB with ξ ) 0.6, 0.8, 1.2, and 2.5, when Ω was fixed at 612 rpm and the temperature is increased from 25 to 70 °C (at ca. 0.8 °C/min). The flow curve for water was included in the figure as a guide to interpret the results. Two transitions (T1 and T2) can be clearly observed for the four CTAB-NaSal solutions. At lower temperatures, the worm-like micelles are intact, and the transitions at T1 correspond to the temperature in which the viscosity is compensated by the drag-reducing effect promoted by the large molecular structures. In other words, the apparent viscosity in this turbulent flow is the result of antagonist effects. On one hand, the shear viscosity contributes positively to ηapp, and, on the other, the drag-reduction effect contributes negatively. For ξ ) 0.6 and temperature above 34 °C, ηapp is even lower than that for pure water, because the shear viscosity decreased and the drag-reduction effect is practically not affected as the temperature was enhanced. However, at 40 °C, a second transition (T2) is observed, corresponding to an abrupt increasing in ηapp, which is associated with the thermal destruction of the worm-like micelles and the consequent loss in their ability to promote drag reduction. For other solutions with ξ > 0.6, T1 and T2 are higher, which indicate that the thermal stability increases as the concentration of NaSal is increased. Table 1 shows the values for T1 and T2 for solutions at CTAB with 2.0 and 3.0 mmol L-1 at different values of ξ. It can be observed that both temperatures are slightly higher for the CTAB solutions at 3.0 mmol L-1. (27) Wang, S. Q. Macromolecules 1991, 24, 3004–3009. (28) Hu, Y.; Rajaram, C. V.; Wang, S. Q.; Jamieson, A. M. Langmuir 1994, 10, 80–85. (29) Ruckstein, E.; Brunn, P. O.; Holweg, J. Langmuir 1988, 4, 350–354. (30) Tsenoglouw, C.; Voyiatzis, E. J. Non-Newtonian Fluid Mech. 2008, 151, 119–128. (31) Lee, J.; Fuller, G.; Hudson, N.; Yuan, X. J. Rheol. 2005, 49, 537–550. (32) Turner, M. S.; Cates, M. E. J. Phys.: Condens. Matter 1992, 4, 3719– 3741. (33) Zhang, Y.; Schmidt, J.; Talmon, Y.; Zakin, J. L. J. Colloid Interface Sci. 2005, 286, 696–709. (34) Zhang, Y.; Qi, Y.; Zakin, J. L. Rheol. Acta 2005, 45, 42–58.

Worm-like Micelles of CTAB and Sodium Salicylate

Langmuir, Vol. 24, No. 24, 2008 13879

Table 1. Values for the Transition Temperatures, T1 and T2, for CTAB (2.0 and 3.0 mmol L-1) and NaSal Solutions with Different Values of ξa NaSal (mmol L-1)

CTAB (mmol L-1)

ξ

T1 (°C)

T2 (°C)

0.8 1.2 1.6 2.0 5.0 1.8 2.4 3.0 7.5

2.0 2.0 2.0 2.0 2.0 3.0 3.0 3.0 3.0

0.4 0.6 0.8 1.0 2.5 0.6 0.8 1.0 2.5

N.O. 30 34 39 42 29 36 41 48

N.O. 42 52 56 63 44 55 57 67

a The precision of the transition temperature is approximately (2 °C. N.O. ) not observed.

T1 and T2 are practically constant when the flow curves for CTAB-NaSal are obtained at higher angular velocity. Figure 5 shows the flow curve for a 2.0 mmol L-1 solution of CTAB with ξ ) 0.6 obtained at Ω of 612 and 921 rpm. As expected, the turbulence at 921 rpm is much higher than that at 612 rpm, and, consequently, the magnitude of the variation in ηapp at T2 is higher, when the worm-like micelles are destroyed.

4. Conclusions The incorporation of salicylate ions into CTA micelles and consequent growth of worm-like micelles was studied using the self-fluorescence of the salicylate ion and the drag-reducing effect promoted by the giant micelles. The drag-reduction effect was studied when the solutions were submitted to a turbulent flow within the double-gap cell of a rotational rheometer. The ability of the worm-like micelles to reduce the hydrodynamic drag is associated with their interaction with the dissipative vortices formed within the turbulence. A critical ratio between NaSal and CTAB concentrations (ξc) was observed in which the worm-like micelles are formed, and the values for ξc can be clearly determined from the fluorescence and flow curves experiments. The thermal stability of the worm-like CTAB-NaSal micelles was studied on the basis of the flow curves obtained by keeping

Figure 5. Apparent viscosity of 2.0 mmol L-1 CTAB solutions containing NaSal with ξ ) 0.6, as a function of the temperature, when Ω was fixed at 612 and 921 rpm.

the double-gap cell rotating at a fixed angular velocity and changing the temperature. When ξ g ξc, two transitions can be observed at specifics temperatures designed as T1 and T2. At T1, the drag-reduction effect promoted by the worm-like micelles overcomes the shear viscosity, resulting in lower apparent viscosities. At T2, the worm-like micelles are destroyed, which results in the abrupt increase in the apparent viscosity of the solution due to the loss of the drag-reduction mechanism. Acknowledgment. We thank the Fundac¸a˜o de Amparo a` Pesquisa do Estado de Sa˜o Paulo (FAPESP, Sa˜o Paulo, Brazil) and the Conselho Nacional de Desenvolvimento Cientı´fico e Tecnolo´gico (CNPq, Brası´lia, Brazil) for financial support and fellowships. We also thank Professor Fred Y. Fujiwara for useful comments. LA802890X