J. B. McClain, D. E. Betts, D. A. Canelas, E. T. Samulski, J. M. DeSimone, ^* J. D. Londono, H. D. Cochran, G. D. Wignall, ^* D. Chillura-Martino, R. Triolo

Interfacially active block copolymer amphiphiles have been synthesized and their self-assembly into micelles in supercritical carbon dioxide (CO₂) has been demonstrated with small-angle neutron scattering (SANS). These materials establish the design criteria for molecularly engineered surfactants that can stabilize and disperse otherwise insoluble matter into a CO₂ continuous phase. Polystyrene-b-poly(1,1-dihydroperfluorooctyl acrylate) copolymers self-assembled into polydisperse core-shell-type micelles as a result of the disparate solubility characteristics of the different block segments in CO₂. These nonionic surfactants for CO₂ were shown by SANS to be capable of emulsifying up to 20 percent by weight of a CO₂-insoluble hydrocarbon into CO₂. This result demonstrates the efficacy of surfactant-modified CO₂ in reducing the large volumes of organic and halogenated solvent waste streams released into our environment by solvent-intensive manufacturing and process industries.

J. B. McClain, D. E. Betts, D. A. Canelas, E. T. Samulski, J. M. DeSimone, Department of Chemistry, University of North Carolina, CB 3290, Venable and Kenan Laboratories, Chapel Hill, NC 27599, USA.
J. D. Londono, H. D. Cochran, G. D. Wignall, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA.
D. Chillura-Martino and R. Triolo, Departimento di Chimica Fisica, University of Palermo, 90123 Palermo, Italy.
^*   To whom correspondence should be addressed.

Block and graft copolymers composed of chain segments with dissimilar solubility characteristics self-assemble into well-defined structures when placed in a medium that is a good solvent for one of the segments (the lyophilic segment) and a poor solvent for the other segment (the lyophobic segment) (6, 7, 8, 9). These nonionic surfactants are typified by micelles or more complex aggregates in which the lyophobic segments form a core surrounded by a shell of the highly solvated lyophilic segments that extend into the continuous phase (6). The core regions of such micelles are technologically useful, as they are capable of emulsifying otherwise insoluble materials into a microphase-separated environment within a preferred continuous solvent phase (9). Here, we report the direct structural characterization by SANS of well-defined, spherical micelles resulting from a series of molecularly engineered block copolymer surfactants in scCO₂. Our results also demonstrate the efficacy of these surfactants in emulsifying insoluble solutes into CO₂. This development may ultimately enable surfactant-modified CO₂ to be used as a replacement for conventional solvent systems currently used in manufacturing and service industries, such as precision cleaning (metal finishing, microelectronics, optics, or electroplating), medical device fabrication, and dry (garment) cleaning, as well as in the chemical manufacturing and coating industries.

Several laboratories have demonstrated progress in the micellization of aqueous and polar materials in many dense sc fluids including alkanes, chlorofluorocarbons, hydrofluorocarbons, and CO₂ (10, 11, 12, 13, 14, 15). However, success in designing generic surfactants for CO₂ has been hindered by the challenge of identifying lyophilic segments for CO₂. A promising lead for the design of highly effective surfactants for CO₂ arose out of two related discoveries: (i) that CO₂ is thermodynamically a good solvent for fluorinated acrylate polymers [positive second virial coefficient (13, 16)], and (ii) that these same polymers could be synthesized under homogeneous conditions in CO₂ (17). Therefore, the fluorinated acrylate segment can be considered as a vehicle to bring insoluble components into CO₂. Recent evidence has shown that fluoroacrylate polymers could be used to bring hydrated, water-soluble polyethylene glycol segments into CO₂ (10).

The block copolymers that we have molecularly engineered to be interfacially active in CO₂ are composed of a CO₂-insoluble polystyrene (PS) segment and a CO₂-soluble poly(1,1-dihydroperfluorooctyl acrylate) (PFOA) segment. Because of these solubility differences in CO₂, such block copolymers are predisposed to self-assemble spontaneously into micellular structures in CO₂. We synthesized PS-b-PFOA diblock copolymers (Fig. 1) with the "iniferter" technique developed by Otsu (18, 19, 20). We first synthesized the lyophobic PS segment of the surfactant by using tetraethylthiuram disulfide as the initiator. We determined the functionality of the telechelic PS segment by using ultraviolet-visible spectroscopy, and the number-average molecular weight (M_n) was determined from gel permeation chromatography. This functionalized PS segment was then used as a macroinitiator in the photopolymerization of the second monomer, FOA, to form the second lyophilic segment of the diblock copolymer (for simplicity, we will henceforth refer to the block copolymers by listing M_n for each block, M_nstyrene-b-M_nFOA).

During the past two decades, SANS has emerged as a powerful technique for studying the self-assembly of amphiphiles in aqueous media (21). Experiments to determine the characteristics of PS-b-PFOA micelles in scCO₂ were performed on the W. C. Koehler 30-m SANS facility at the Oak Ridge National Laboratory (22). The neutron wavelength  was 4.75 Å (/  5%), and the sample detector distance ranged from 6.3 to 10 m, resulting in a range of momentum transfer 0.006 < Q = 4¹ sin  < 0.1 Å¹, where 2 is the angle of scatter. Instrumental procedures and calibration as well as the procedure for SANS measurements from scCO₂ systems have been described (23, 24, 25, 26, 27).

Neutron scattering curves (Fig. 2) from PS-b-PFOA in CO₂ were fitted to a spherical "core plus shell" model based on a solid PS core and a uniform-thickness shell containing PFOA chains swollen by CO₂ solvent molecules. Parameters of the model include the aggregation number N_agg (the number of block copolymer units forming each micelle), the number of CO₂ molecules N_CO2 (that swell the PFOA corona per FOA unit), and the polydispersity parameter Z (related to the variance of the Schultz distribution of the particle sizes: increasing polydispersity with decreasing Z). The radius of the core (R_core) and of the total particle (R_total) can be easily derived from these parameters. If we assume no orientational correlations, the coherent differential cross section, d(Q)/d, for such a model is given by

(1)

(28), where N_p is the number density of particles, S(Q) is the structure function arising from interparticle scattering, B is the background, and F(Q) is the scattering amplitude of the spherical particle.

A series of PS-b-PFOA surfactants having differing lyophilic PFOA segment lengths and lyophobic PS segment lengths were dissolved in CO₂ at 65°C and 340 bar (density  = 0.842 g cm³) and characterized by SANS (Table 1). If we rely on calibrated SANS curves, the fit of the scattering results to the core-shell model indicates that the block copolymers self-assemble into spherical core-shell structures (see Fig. 2). A systematic increase of the PFOA block length (the first three entries in Table 1) results in a variation of only the "shell-dependent" properties: an increase in R_total and a decrease in the swelling of the corona, evident in N_CO2. The "core-dependent" properties, R_core and N_agg, remain essentially constant for surfactants of constant PS length. Control of the core-dependent properties is evident in the smooth increases in R_core with increasing PS segment length. The behavior of PS-b-PFOA surfactants in CO₂ follows the expected scaling laws of micelles composed of block copolymers with variable segment lengths in traditional solvents (29, 30).

Table 1. Characterization of PS-b-PFOA micelles in CO2 (65°C, 340 bar,  = 0.842 g cm3) as a function of block lengths. Surfactant Mnstyrene-b-MnFOA Nagg Rcore (Å) Rtotal (Å) NCO2 Zcore 3.7k-b-16.6k 7 27 85 40 10 3.7k-b-39.8k 7 27 89 16 8 3.7k-b-61.2k 6 26 100 17 8 4.5k-b-24.5k 8 30 75 16 3 6.6k-b-34.9k 6 34 77 11 3

The solvent strength of scCO₂ is easily tunable with changes in the system density (directly related to temperature and pressure) (2, 13, 16, 31). Thus, the association of amphiphiles in CO₂ should be strongly dependent on the CO₂ density because of changes in polymer solubility as a function of solvent quality. As the density of CO₂ is increased (32), the solvation of both segments becomes greater, which decreases aggregation and creates more dynamic micelles. Thus, we anticipate the existence of a critical micelle density, analogous to a critical micelle concentration, to describe the phenomenon of unimer-to-aggregate transitions for amphiphilic materials in sc fluids with changing solvent (quality) density. Table 2 shows the effect of two experimental conditions--65°C, 340 bar,  = 0.842 g cm³ and 40°C, 340 bar,  = 0.934 g cm³ (31)--on the association of PS-b-PFOA surfactants. With increasing solvent strength (density), smaller and more polydisperse micelles are evident, especially for surfactants with a higher PS/PFOA ratio. When the PFOA segment length is increased to 135 repeat units, the micelles are essentially unaffected by CO₂ density over this range of conditions.

Table 2. The effect of CO2 density on the self-assembly of PS-b-PFOA surfactants of varying block lengths. Surfactant* CO2 density (g cm3) Nagg Rcore (Å) Rtotal (Å) NCO2 Zcore 3.7k-b-39.8k 0.934 3.3 21 68 16 4.5 3.7k-b-39.8k 0.842 7 27 89 16 8 4.5k-b-24.5k 0.934 5 26 67 21 3 4.5k-b-24.5k 0.842 8 30 75 16 3 6.6k-b-34.9k 0.934 16 43 128 28 9 6.6k-b-34.9k 0.842 22 49 142 22 11 3.7k-b-61.2k 0.934 6 25 97 18 9 3.7k-b-61.2k 0.842 6 26 100 17 8 * Surfactant concentration = 4.0% (w/v). Value for pure CO2 at the temperature and pressure of the experiment (32).

We demonstrated the ability of a PS-b-PFOA micellar solution in CO₂ to emulsify CO₂-insoluble materials by adding model CO₂-insoluble hydrocarbon oligomers (500 g per mole of oligomers of both deuterated and hydrogenated PS, referred to as D-oligomer and H-oligomer, respectively) to the system and monitoring by SANS (33). We used both H- and D-oligomer samples to take advantage of different neutron scattering contrasts and to quantify the fraction of oligomer inside the micelle core versus oligomer free in solution or stabilized by free surfactant (25). SANS characterization of micelles of PS-b-PFOA surfactants in CO₂ (65°C, 340 bar,  = 0.842 g cm³) with added H-oligomer and D-oligomer display stabilization of >99% of the added oligomer into the core of the micelle. The micellar core volume increases with added H-oligomer and D-oligomer as a function of oligomer concentration for the 3.7^k-b-39.8^k surfactant (20) (Fig. 3). By a comparison of the unswollen surfactant micelles with the swollen system containing added oligomer in Table 3, we see an approximately eightfold increase in the volume of the micelle core with the addition of up to 20% (w/w) oligomer.

3

Table 3. Effect of CO2 density on the swelling of PS-b-PFOA surfactant micelles with a PS oligomer. Surfactant* [Oligomer] CO2 density (g cm3) Nagg Rcore (Å) Rtotal (Å) NCO2 Zcore 3.7k-b-39.8k 13 0.934 10 44 124 13 11 3.7k-b-39.8k 13 0.842 12 47 120 7 11 3.7k-b-39.8k 16.6 0.934 11 48 126 9 12 3.7k-b-39.8k 16.6 0.842 13 50 119 5 11 4.5k-b-24.5k 20 0.934 40 70 157 10 10 4.5k-b-24.5k 20 0.842 39 69 152 7 8 6.6k-b-34.9k 20 0.934 20 64 153 15 9 6.6k-b-34.9k 20 0.842 22 66 162 11 10 * Surfactant concentration = 4.0% (w/v). Oligomer concentration = percent (w/w) of surfactant; all experiments were performed with D-oligomer. Value for pure CO2 at the temperature and pressure of the experiment (32).

The density of CO₂, over the same range studied for unswollen aggregates, seems to have little effect on the structure of oligomer-swollen micelles (Table 3). Highly swollen systems were observed for three different surfactants at CO₂ densities ranging from 0.842 to 0.934 g cm³ with less than 5% change in the dimensions or characteristics of the micelles. This is very different from the association behavior of similar unswollen surfactant micelles. In the oligomer-swollen system, the surfactant molecules are forced to an interface between insoluble PS oligomer and CO₂. This effect results in diminished changes in the overall structural dimensions with solvent strength so long as CO₂ remains a good solvent for the PFOA shell segments. Earlier work has shown that PFOA is in a thermodynamically favorable solvent at both densities (16) and therefore retains its propensity to stabilize a larger CO₂-insoluble core.

REFERENCES AND NOTES

1. K. A. Walsh, Chem. Week 158, 38 (1996).
2. M. A. McHugh and V. J. Krukonis, Supercritical Fluids Extraction: Principles and Practice (Butterworth-Heinman, Stoneham, MA, ed. 2, 1993).
3. K. A. Shaffer and J. M. DeSimone, Trends Polym. Sci. 3, 146 (1995).
4. J. A. Hyatt, J. Org. Chem. 49, 5097 (1984) .
5. V. S. Smith, P. O. Campbell, V. Vandana, A. S. Teja, Int. J. Thermophys. 1, 23 (1996).
6. Z. Tuzar and P. Kratochiv, in Surface and Colloid Science, E. Matijevic, Ed. (Plenum, New York, 1993), vol. 15, pp. 1-83.
7. B. Chu, Langmuir 11, 414 (1995).
8. J. C. Ravey, M. Buzier, C. Picot, J. Colloid Interface Sci. 97, 9 (1984) .
9. K. A. Cogan and A. P. Gast, Macromolecules 23, 745 (1990) .
10. J. L. Fulton et al., Langmuir 11, 4241 (1995).
11. K. P. Johnston et al., Science 271, 624 (1996) .
12. K. A. Bartscherer, M. Minier, H. Renon, Fluid Phase Equilibria 107, 93 (1995).
13. D. Chillura-Martin et al., J. Mol. Struct., in press.
14. J. Eastoe, Z. Bayazit, S. Martel, Langmuir 12, 1423 (1996).
15. T. A. Hoefling, R. M. Enick, E. J. Beckman, J. Phys. Chem. 95, 7127 (1991) ; E. J. Beckman, Science 271, 613 (1996) .
16. J. B. McClain et al., J. Am. Chem. Soc. 118, 917 (1996) .
17. J. M. DeSimone, Z. Guan and C. S. Elsbernd, Science 257, 945 (1992) .
18. T. Otsu and M. Yoshida, Makromol. Chem. Rapid Commun. 3, 127 (1982).
19. Z. Guan and J. M. DeSimone, Macromolecules 27, 5527 (1994).
20. Summary of data from the synthesis of PS-b-PFOA diblock copolymers: Telechelic PS: M_n = 3.7 kg mol¹, polydispersity index (PDI) = 1.6, functionality = 1.6; 3.7^k-b-16.6^k diblock copolymer: M_n = 20.4 kg mol¹, 49.4 mol % styrene, and 50.6 mol % FOA; 3.7^k-b-39.8^k diblock copolymer: M_n = 43.5 kg mol¹, 28.9 mol % styrene, and 71.1 mol % FOA; 3.7^k-b-61.2^k diblock copolymer: M_n = 64.9 kg mol¹, 20.9 mol % styrene, and 79.1 mol % FOA. Telechelic PS: M_n = 4.5 kg mol¹, PDI = 1.7, functionality = 1.6; 4.5^k-b-24.5^k diblock copolymer: M_n = 28.9 kg mol¹, 44.5 mol % styrene, and 55.5 mol % FOA. Telechelic PS: M_n = 6.6 kg mol¹, PDI = 1.8, functionality = 2.3; 4.5^k-b-34.9^k diblock copolymer: M_n = 41.5 kg mol¹, 45.1 mol % styrene, and 54.9 mol % FOA.
21. L. J. Magid, Colloid Surf. Sci. 19, 129 (1986).
22. W. C. Koehler, Physica (Utrecht) B 137, 320 (1986).
23. R. J. Nelmes et al., Acta Crystallogr. A49, 427 (1993) .
24. S. Janssen, D. Schwahn, K. Mortensen, T. Springer, Macromolecules 26, 5587 (1993) .
25. G. D. Wignall, in The Physical Properties of Polymers, J. E. Mark, Ed. (American Chemical Society, Washington, DC, 1993), pp. 313-378.
26. ___, F. S. Bates, J. Appl. Crystallogr. 20, 28 (1986).
27. Experiments were performed under equilibrium conditions, justified by an approximately 1-hour delay between the mixing of the system and the taking of data and by the observation that there was no change in the scattering curve during 3 to 8 hours of data acquisition.
28. E. Caponetti and R. Triolo, Adv. Colloid Interface Sci. 32, 235 (1990) ; see also R. Triolo and E. Caponetti, in Industrial and Technical Applications of Neutrons, F. Rustichelli, M. Fontana, R. Coppola, Eds. (North-Holland, Amsterdam, 1992), pp. 403-424.
29. A. Halperin, M. Tirrell, T. P. Lodge, Adv. Polym. Sci. 100, 31 (1992).
30. T. M. Birshtein and E. B. Zhulina, Polymer 30, 171 (1989).
31. Y. Hsiao et al., Macromolecules 28, 8159 (1995) .
32. J. F. Ely, CO₂PAC: A Computer Program to Calculate Physical Properties of Pure CO₂ (National Bureau of Standards, Boulder, CO, 1986).
33. The model CO₂-insoluble oligomers were PS (500 g mol¹) made by living anionic polymerization. Also, to simplify the experimental system and to ensure minimal free oligomer dissolved in CO₂ under the experimental conditions, we extracted oligomer samples in advance with CO₂ at 50°C and 400 bar.
34. The research at the University of North Carolina was supported by the Environmentally Benign Chemical Synthesis and Processing Initiative of NSF and the U.S. Environmental Protection Agency; the Presidential Faculty Fellowship Program (J.M.D.); and the Consortium for the Synthesis and Processing of Polymeric Materials at the University of North Carolina (sponsored by Air Products and Chemicals, Bayer, B. F. Goodrich, DuPont, Eastman Chemical, General Electric, Hoechst-Celanese, and Xerox). The research at Oak Ridge National Laboratory was supported by Laboratory Directed Research and Development Program and the Division of Material Science, U.S. Department of Energy, under contract DE-AC05-96OR22464 with Lockheed Martin Energy Research Corp.

12 June 1996; accepted 25 September 1996