New procedure for recycling homogeneous catalyst: propylene carbonate synthesis under supercritical CO₂ conditions

Abstract

Polyfluoroalkyl phosphonium iodides, Rf₃RPI (Rf = C₄F₉C₂H₄, C₆F₁₃C₂H₄, C₈F₁₇C₂H₄; R = Me, Rf), catalyzed propylene carbonate synthesis from propylene oxide and carbon dioxide under supercritical CO₂ conditions, where propylene carbonate was spontaneously separated out of the supercritical CO₂ phase. The Rf₃RPI catalyst could be recycled with maintaining a high CO₂ pressure and temperature by separating the propylene carbonate from the bottom of the reactor followed by supplying propylene oxide and CO₂ to the upper supercritical CO₂ phase in which the Rf₃RPI remained.

---

Green Context

Cyclic carbonates are useful synthetic intermediates which can be advantageously synthesised in supercritical CO₂. Here, it is shown that a fluorous phosphonium catalyst can be used in scCO₂ as a homogeneous catalyst. This has the advantage of allowing the synthesis of e.g. propylene carbonate to take place under supercritical conditions, with the direct and spontaneous separation of the carbonate. The supercritical phase retains the catalyst allowing for continuous reaction.

DJM

Due to its environmentally friendly nature, supercritical carbon dioxide has attracted much interest as a substitute for organic solvents in various fields including homogeneous catalysis.¹ Supercritical CO₂ is also very effective for accelerating the transformation of CO₂ which is usually very inert.² In addition, by utilizing supercritical CO₂ as a new dense phase that separates organics and water, we can simplify the separation process. Especially, if the product spontaneously comes out of the supercritical CO₂ phase, an efficient separation without losing high pressure and temperature can be realized.

Five-membered cyclic carbonates such as ethylene carbonate and propylene carbonate are synthesized by the cycloaddition of CO₂ to epoxides in the presence of homogeneous or heterogeneous catalysts.³ In our recent studies on propylene carbonate synthesis using supercritical CO₂, we found that propylene oxide and supercritical CO₂ initially formed a uniform phase while the produced propylene carbonate spontaneously separated out of the supercritical CO₂ phase by forming a lower phase in the reactor, as illustrated in Fig. 1.⁴ This finding suggested to us that the product could be recovered from the bottom of the reactor, while maintaining a high CO₂ pressure and temperature inside the reactor. In addition, if one can obtain a catalyst which is selectively soluble in supercritical CO₂, the reaction can be repeated without catalyst separation.⁵

Fig. 1 Schematic diagram of the reaction behavior for the propylene carbonate synthesis from propylene oxide and supercritical CO₂ at 100 °C and 14 MPa. PO; propylene oxide, scCO₂; supercritical CO₂, PC; propylene carbonate.

We synthesized novel polyfluoroalkyl phosphonium iodides, Rf₃RPI (Rf = C₆F₁₃C₂H₄, R = Me (1); Rf = C₈F₁₇C₂H₄, R = Me (2); Rf = R = C₄F₉C₂H₄ (3); Rf = R = C₆F₁₃C₂H₄ (4)), by reacting tri(polyfluoroalkyl)phosphines with either methyl iodide or the corresponding polyfluoroalkyl iodides.⁶ The catalytic performances of Rf₃RPI for the cycloaddition reaction of CO₂ to propylene oxide were first evaluated using a conventional batch reactor (20 cm³ inner volume) in the same manner as previously described.^4a,7Table 1 summarizes the yield and selectivity of propylene carbonate at 100 °C and 14 MPa. All the phosphonium catalysts (1–4) exhibited high yields and selectivities comparable to those of a conventional catalyst, Bu₄PI,⁸ although the yield obtained by the catalyst with the shortest fluoroalkyl chain (3) was slightly lower than those obtained by the others. The CO₂ pressure dependence of the reaction using 1 clearly demonstrated the preferential effect of the supercritical conditions for promoting the reactivity of CO₂ as seen in Fig. 2. The yield and selectivity increased with the increasing CO₂ pressure, and a high CO₂ pressure of 10 MPa or above was notably effective for achieving high yields.

Table 1 Propylene carbonate synthesis from propylene oxide and carbon dioxide catalyzed by polyfluoroalkyl phosphonium iodides^a

Entry	Catalyst	Yield (%)	Selectivity (%)
a Reactions were carried out using a conventional batch reactor (20 cm^3 inner volume). Reaction conditions: catalyst (0.572 mmol, 1 mol%), propylene oxide (57.2 mmol), CO2 (14 MPa), 100 °C, 24 h.
1	(C6F13C2H4)3MePI (1)	93	99
2	(C8F17C2H4)3MePI (2)	92	97
3	(C4F9C2H4)4PI (3)	83	97
4	(C6F13C2H4)4PI (4)	89	99
5	Bu4PI	90	99

---

Fig. 2 CO₂ pressure dependence of the yield (●) and selectivity (○) of propylene carbonate for 1. Reaction conditions: 1 (0.572 mmol, 1 mol%), propylene oxide (57.2 mmol), 100 °C, 24 h.

We next investigated the product separation and catalyst recycling employing a reactor (20 cm³ inner volume) equipped with a mechanical stirrer, sapphire windows, and a valve at the bottom of the reactor for recovering the product. A typical procedure is as follows. The reaction of propylene oxide (57.2 mmol) and CO₂ was first run at 100 °C and 14 MPa in the presence of Rf₃RPI (0.572 mmol, 1 mol%) and biphenyl (200 mg, internal standard for GC analysis). Visual observation through sapphire windows revealed that all the components were miscible and formed a uniform phase at the beginning of the reaction, confirming the homogeneous catalysis of Rf₃RPI. As the reaction proceeded, the product solution separated from the uniform phase gradually accumulated to form a lower phase in the reactor as previously observed,^4a and finally the volume ratio of the upper supercritical phase to the lower phase became approximately three. After 24 h, the lower phase was taken out of the reactor by slowly opening the valve. During this process the pressure decreased from 14 to 11 MPa. The reaction was then repeated by supplying propylene oxide (57.2 mmol) containing biphenyl (200 mg, internal standard for GC analysis) to the reactor at 11 MPa followed by readjusting the pressure to 14 MPa upon the introduction of CO₂. These results are summarized in Fig. 3, where the yield was calculated based on the amount of propylene carbonate in the separated lower phase and the amount of supplied propylene oxide. When using polyfluorinated phosphonium salts (1 and 3), propylene carbonate was produced in the second run with almost the same yield as the first run showing that Rf₃RPI remains in the upper supercritical phase as we expected. In a separate experiment, deposition of the catalyst was observed inside the reactor as a white solid when CO₂ was released after the separation of the lower phase, indicating that the catalyst had been dissolved in the supercritical phase.

Fig. 3 Yield of propylene carbonate during the repeated reaction using (a) 1, (b) 3, and (c) Bu₄PI catalysts. Reaction conditions: catalyst (0.572 mmol, 1 mol%), propylene oxide (57.2 mmol), CO₂ (14 MPa), 100 °C, 24 h. The reaction was repeated by removing the product solution at 100 °C and 14 MPa followed by supplying propylene oxide (57.2 mmol) and CO₂ (14 MPa). The yield of Bu₄PI in the second run was determined by releasing CO₂; see text.

In contrast, Bu₄PI is preferentially dissolved in the lower-phase carbonate solution. Hence, once the lower phase was removed, no carbonate phase appeared in the second run. Note that the yield in the second run determined by releasing CO₂ was only 3% (Fig. 3). On the other hand, the Rf₃RPI catalysts gave a high yield even in the third run. Thus, the fundamental idea of the new catalyst recycling during homogeneous catalysis has been demonstrated by using supercritical CO₂ and a CO₂-philic catalyst. The merit of the present procedure is easy separation of the product, catalyst, and supercritical CO₂ without losing the high pressure and temperature of the supercritical phase.

In conclusion, we have demonstrated that propylene carbonate can be repeatedly synthesized from propylene oxide and CO₂ by using supercritical CO₂- and CO₂-philic polyfluoroalkyl phosphonium iodides with maintaining a high CO₂ pressure and temperature. The utilization of supercritical CO₂ is also advantageous in terms of the reactivity enhancement and solvent-free processes. In order to improve the efficiency of the catalyst recycling, modification of the catalyst structure, optimization of the separation conditions, and addition of a third component will be future subjects.

Acknowledgement

L.-N. H. acknowledges the New Energy and Industry Technology Development Organization (NEDO) of Japan for a postdoctoral fellowship.

References

1. (a) Chemical Synthesis Using Supercritical Fluids, ed. P. G. Jessop and W. Leitner, Wiley-VCH, Weinheim, 1999; (b) P. G. Jessop, T. Ikariya and R. Noyori, Chem. Rev., 1999, 99, 475–493; (c) R. T. Baker and W. Tumas, Science, 1999, 284, 1477–1479; (d) M. Poliakoff and P. King, Nature, 2001, 412, 125; (e) S. L. Wells and J. DeSimone, Angew. Chem., Int. Ed., 2001, 40, 518–527.
2. (a) W. Leitner, Angew. Chem., Int. Ed. Engl., 1995, 34, 2207–2221; (b) P. G. Jessop, T. Ikariya and R. Noyori, Chem. Rev., 1995, 95, 259–272.
3. (a) D. J. Darensbourg and M. W. Holtcamp, Coord. Chem. Rev., 1996, 153, 155–174 , and references therein. For recent reports on cyclic carbonate synthesis, see; (b) T. Yano, H. Matsui, T. Koike, H. Ishiguro, H. Fujihara, M. Yoshihara and T. Maeshima, Chem. Commun., 1997, 1129–1130; (c) K. Yamaguchi, K. Ebitani, T. Yoshida, H. Yoshida and K. Kaneda, J. Am. Chem. Soc., 1999, 121, 4526–4527; (d) T. Zhao, Y. Han and Y. Sun, Phys. Chem. Chem. Phys., 1999, 1, 3047–3051; (e) H. S. Kim, J. J. Kim, B. G. Lee, O. S. Jung, H. G. Jang and S. O. Kang, Angew. Chem., Int. Ed., 2000, 39, 4096–4098; (f) T. Iwasaki, N. Kihara and T. Endo, Bull. Chem. Soc. Jpn., 2000, 73, 713–719; (g) D. Ji, X. Lu and R. He, Appl. Catal. A, 2000, 203, 329–333; (h) H. Kawanami and Y. Ikushima, Chem. Commun., 2000, 2089–2090; (i) M. Tu and R. J. Davis, J. Catal., 2001, 199, 85–91; (j) B. M. Bhanage, S. Fujita, Y. Ikushima and M. Arai, Appl. Catal. A, 2001, 219, 259–266; (k) R. L. Paddock and S. T. Nguyen, J. Am. Chem. Soc., 2001, 123, 11498–11499; (l) J. Peng and Y. Deng, New J. Chem., 2001, 25, 639–641; (m) M. Aresta and A. Dibenedetto, J. Mol. Catal. A, 2002, 182–183, 399–409.
4. (a) H. Yasuda, L.-N. He and T. Sakakura, J. Catal., 2002, 209, 547–550 . Very recently, the similar phase separation has been reported for the reaction of ethylene oxide and scCO₂, see; (b) X.-B. Lu, R. He and C.-X. Bai, J. Mol. Catal. A, 2002, 186, 1–11.
5. Attempts to immobilize palladium and rhodium catalysts in scCO₂ by using fluorinated phosphine ligands have been reported, see: (a) D. Hâncu and E. J. Beckman, Green Chem., 2001, 3, 80–86; (b) M. McCarthy, H. Stemmer and W. Leitner, Green Chem., 2002, 4, 501–504.
6. Experimental procedure for the synthesis of (C₆F₁₃C₂H₄)₃MePI (1): (C₆F₁₃C₂H₄)₃P was prepared according to the literature: P. Bhattacharyya, D. Gudmunsen, E. G. Hope, R. D. W. Kemmitt, D. R. Paige and A. M. Stuart, J. Chem. Soc., Perkin Trans. 1, 1997, 3609–3612 . A mixture of (C₆F₁₃C₂H₄)₃P (8.58 g, 8.00 mmol) and methyl iodide (0.50 mL, 8.00 mmol) in acetone (20 mL) was refluxed for 3 days under an argon atmosphere. After evaporation of the acetone under reduced pressure, the crude salt was washed several times with dehydrated ether and dried in a vacuum (yield 5.83 g, 60%). ¹H NMR (400 MHz in acetone-d₆ at 25 °C): δ 1.88 (d, 3H, J = 14.4 Hz, CH₃P), 2.21 (m, 6H, CH₂P), 2.54 (m, 6H, CH₂CF₂). ¹³C{¹H} NMR (100.4 MHz in acetone-d₆ at 25 °C): δ 4.98 (d, J = 50.4 Hz, CH₃P), 13.73 (d, J = 53.7 Hz, CH₂P), 24.58 (t, J = 22.3 Hz, CH₂CF₂), 107.8–119.7 (m, C₆F₁₃). ³¹P{¹H} NMR (161.7 MHz in acetone-d₆ at 25 °C): δ 38.47. Anal. Calc. for C₂₅H₁₅F₃₉IP: C, 24.73; H, 1.25. Found: C, 24.70; H, 1.20%. (C₈F₁₇C₂H₄)₃MePI (2), (C₄F₉C₂H₄)₄PI (3), and (C₆F₁₃C₂H₄)₄PI (4) were analogously synthesized by reacting the tri(polyfluoroalkyl)phosphines with either methyl iodide or the corresponding polyfluoroalkyl iodides.
7. After the reaction at 100 °C and 14 MPa for 24 h, the autoclave was cooled to room temperature and CO₂ was slowly released by bubbling into DMF at −20 °C for trapping any unreacted propylene oxide. The liquid products and the DMF solution were mixed and analyzed by means of GC and GC-MS.
8. T. Nishikubo, A. Kameyama, J. Yamashita, T. Fukumitsu, C. Maejima and M. Tomoi, J. Polym. Sci., Polym. Chem., 1995, 33, 1011–1017.

---