Recycling of a homogeneous catalyst using switchable water

Abstract

Aqueous/organic biphasic catalysis allows easy separation of a homogeneous catalyst from product, but is often inefficient when hydrophobic substrates are used. A system based on switchable water is monophasic in the absence of CO₂ and biphasic in its presence. Catalysis can be performed in the monophasic solvent, and then switched to a biphasic system, separating catalyst from product. Removal of CO₂ allows for easy recycling of the catalyst. Hydroformylations have been achieved using this solvent system. The catalyst was recycled several times with minimal loss of activity.

---

Homogeneous catalysts are often more active and selective than heterogeneous catalysts but are far more difficult to separate from the product. The development of new means to separate, recover, and recycle homogeneous catalysts is an important area of research for both industry and academia alike.¹ One approach, already industrialized in a few cases, is catalysis in an aqueous/organic biphasic mixture.² Such reactions involve the dissolution of a catalyst into the aqueous phase and the dissolution of the reagents into the organic solvent. The small partitioning of the reagent into the catalyst-bearing aqueous phase allows the reaction to proceed. Aqueous/organic biphasic systems are currently used for the industrial hydroformylation of short alkenes such as propene.^2a These systems utilize transition metals ligated by sulfonated phosphines to increase the water solubility of the catalyst, causing it to reside in the aqueous phase. After the reaction is completed, the product (organic) phase is decanted and the aqueous catalyst-bearing phase is used again.

Even though these aqueous/organic biphasic catalytic systems have the advantage of facile separation of catalyst from product, they can suffer from slow reaction rates. These result from the catalyst and the reactants being in two different phases, especially when the reactant (such as 1-octene or styrene) is so hydrophobic that its concentration in the aqueous phase is extremely low. One solution is to manipulate the hydrophilicity of the catalyst by protonation/deprotonation of the ligands. This allows a reaction to be run in a monophasic organic solvent and the catalyst is later drawn into an aqueous phase, away from the products, using a pH change. This pH change is often enacted by introducing CO₂ into the aqueous phase.³ Similarly, catalysts which can be drawn into an aqueous phase based on temperature changes have also been explored.⁴ Another solution to this problem is to design a trigger to make the aqueous and organic phases merge into one phase during the catalysis and then to separate into two phases again after the reaction is complete. Demonstrations of this concept using thermomorphic liquids⁵ and analogously with fluorous compounds⁶ have been performed. CO₂ based-organic/aqueous tunable solvents (OATS)⁷ have also been previously published. However, the OATS method required high pressures. Herein, we describe a method for achieving monophasic catalysis and biphasic separation in a similar vein to those methods. However through our new method using switchable water, these outcomes can be realized using mild reaction conditions, common reagents, with the reversible separations being performed at only 1 bar of CO₂.

We propose that such switching between monophasic and biphasic can be achieved at ambient pressure if the aqueous phase is “switchable water”. Switchable water is an aqueous solution of an amine that reversibly forms the corresponding bicarbonate salt upon exposure to CO₂;⁸ The solution thus reversibly switches from low ionic strength to high ionic strength via generation of trialkylammonium bicarbonate salts. A combination of switchable water and a water-miscible organic solvent such as tert-butanol is a monophasic liquid mixture in the absence of CO₂, but a biphasic mixture in its presence. The presence of an organic co-solvent (which is later salted out) would allow for higher loading of hydrophobic substrates than with water alone. The proposed catalytic process would involve homogeneous catalysis (such as hydroformylation of an alkene) in the monophasic mixture, after which CO₂ would be added (Fig. 1). The CO₂ would react with the amine, cause a rise in the ionic strength of the solution and thereby trigger the salting out of both the organic solvent and the product from the aqueous phase. If a suitably hydrophilic catalyst were selected, the catalyst would then remain in the aqueous phase, isolated from the products of reaction. After decantation of the product phase, the removal of CO₂ from the water would then regenerate a low ionic strength aqueous phase that could now accept fresh reagents and organic solvent and the reaction could be repeated.

Fig. 1 Monophasic hydroformylation and biphasic separation in a liquid mixture of switchable water and tert-butanol.

The hydroformylation of styrene is an attractive model system because its aldehyde products have potential use in pharmaceutical and fine chemical production and its water solubility is low⁹ which would make a traditional aqueous/organic biphasic reaction inefficient. Our past studies of salting-out using switchable water showed that THF could easily be forced out aqueous solution upon the addition of CO₂.⁸ However, initial studies into hydroformylation of styrene found that this weakly coordinating organic solvent hinders the hydroformylation reaction. We therefore searched for other water-miscible organic solvents that could solubilize styrene and still later be salted out by our trialkylammonium bicarbonate salts. Preliminary tests showed that tert-butanol was an appropriate organic solvent choice. We ultimately developed a solvent system comprised of tert-butanol, water and 0.8 molar N,N,N′,N′-tetramethyl-1,4-diaminobutane (TMDAB, Me₂NC₄H₈NMe₂) that performs hydroformylations efficiently and in a recyclable fashion. Styrene, was reacted in the solvent system with synthesis gas in the presence of a rhodium catalyst ([Rh(COD)Cl]₂) and the sodium salt of sulfonated triphenylphosphine (TPPTS, P(C₆H₄-m-SO₃Na)₃) at 100 °C for 3 h (Scheme 1).

Scheme 1 Hydroformylation of styrene using a Rh pre-catalyst and water soluble phosphine, TPPTS.

After the reaction, CO₂ was bubbled through the solution, increasing the ionic strength and generating a biphasic system where the catalyst resided in the aqueous phase and the product resided in the tert-butanol phase. The organic phase could then be removed and analyzed. The low ionic strength form of the aqueous phase was then regenerated by expelling CO₂. This was achieved by heating to 65 °C and bubbling an inert gas through the solution. A fresh supply of organic solvent and substrate was then added and the reaction was repeated (Table 1). Aldehyde products were obtained with good conversions and selectivities over multiple cycles.

Table 1 Hydroformylation of styrene in a monophasic water/tert-butanol mixture followed by separation in a biphasic water/tert-butanol mixture, where the switch between monophasic and biphasic is achieved using the CO₂-induced phase separation of switchable water^a

Entry	Cycle 1			Cycle 2			Cycle 3			Cycle 4			Cycle 5
	% Conv.	% Aldehyde	B:L	% Conv.	% Aldehyde	B:L	% Conv.	% Aldehyde	B:L	% Conv.	% Aldehyde	B:L	% Conv.	% Aldehyde	B:L
a Standard conditions: 6 mL tert-butanol, 4 mL H2O, 0.80 molar N,N,N′,N′-tetramethyl-1,4-diaminobutane (TMDAB), relative to water, 0.15 mL styrene (1.3 mmol). 0.25 mol% [Rh(COD)Cl]2 (1.5 mg), 7 : 1 P/Rh (P=TPPTS), 100 °C, 10 bar (1 : 1 CO/H2), 3 h. b Same conditions as [a] except 1.4 molar N,N-dimethylethanolamine (DMEA) used instead of 0.80 molar TMDAB. c 12 : 1 P/Rh initially, 0.10 mL styrene (0.9 mmol), 1.5 mg [Rh(COD)Cl]2, 1.5 h. d Same conditions as [c] except 5 bar syngas used. e Same conditions as [c] except 0.30 mL 1-octene (1.9 mmol) reacted instead of 0.10 mL styrene, 3 h.
1	98	98	10.2	96	98	5.7	63	99	4.7
2^b	95	99	8.8	96	99	6.5	74	98	5.7
3^c	99	97	7.5	99	98	5.5	99	99	5.2	97	98	5.1	73	99	5.2
4^d	84	99	3.0	77	99	3.4	90	98	3.8	83	99	3.8	52	95	2.9
5^e	99	98	0.5	99	99	0.3	99	99	0.3

---

Our initial system suffered from decreasing conversions over the three cycles of the first reaction (Table 1, Entry 1). We speculated that deactivation of the catalyst was due to oxidation of the TPPTS (as observed in ³¹P NMR spectra, see Fig. S3†) as well as the possible competitive coordination of the bidentate diamine TMDAB to the metal. Leaching of TPPTS into the organic phase was deemed unlikely, as was later confirmed by ³¹P NMR. To test the diamine coordination hypothesis, the hydroformylation reaction was performed using 1.4 molar of a monoamine, N,N-dimethylethanolamine (DMEA, Me₂NCH₂CH₂OH) as the switchable water additive. Although conversions were less affected during the cycles (Table 1, Entry 2), a significant loss of aqueous volume was observed (likely due to DMEA partitioning into the organic phase to a larger extent than TMDAB, see Table S1†). A discoloration of the product containing organic phase was also observed suggesting significant leaching of the catalyst over the cycles when DMEA is used (Fig. S2 & Table S2†).

We therefore returned to the TMDAB system but increased the initial phosphine loading to 12 : 1 P/Rh so that TPPTS could better compete with the diamine for positions on the metal. This proved to remedy the issue of decreasing aqueous volume while maintaining high conversions during the cycles. This system was able to be recycled for four runs with minimal loss of conversion and selectivity to the aldehyde products with only a slight loss of activity on the fifth cycle (Table 1, Entry 3). For this initial phosphine loading, the amount of styrene loaded into the system had to be decreased slightly so that it would remain soluble in the aqueous solvent of greater initial ionic strength. The reaction time was also decreased accordingly to 1.5 h. Although high conversions were often observed for this system, some unreacted substrate was always observed in the gas chromatograms; therefore catalyst activity was consistent until the final run and the substrate/catalyst ratio was appropriate. Despite the presence of high concentrations of amine in the reaction mixture, we did not observe any aldol by-products for these reactions. A control experiment with no TPPTS in solution afforded poor conversion (<6%) with a B:L ratio of roughly 0.8.

To further confirm the stability of the catalytic species and the solvent system, we monitored the gas uptake for the hydroformylation of styrene over five cycles (Fig. S1†). These reactions (Table 1, Entry 4) were performed at 5 bar syngas as opposed to 10 bar. The rate of gas uptake was greater than or equal to that of the first cycle for the first four cycles. The fifth cycle showed slower gas uptake, suggesting catalyst deactivation which is in agreement with the lower conversion observed on the fifth cycle compared to first four cycles in entries 3 and 4 of Table 1.

Additionally, we performed the hydroformylation of 1-octene which showed good conversion and selectivity over three cycles (Table 1, Entry 5). The selectivity to the linear isomer over the branched isomer is consistent with past literature. Our pressures are lower than typically used in the literature but we used higher catalyst loadings; as a result our turnover numbers are generally lower than previous examples (i.e. ∼300 compared to ∼1000).^3c–e Our system affords an average turnover frequency of ∼100 h⁻¹ while recent work on traditional biphasic hydroformylations of 1-octene using Rh/TPPTS afforded ∼20 h⁻¹.¹⁰ Overall, the mild reaction conditions, high selectivity, low ligand loading, and recyclability make this system an attractive alternative to previous described systems.

A notable decrease in the branched/linear product ratio (B/L) was observed during the second cycles for all the hydroformylations of styrene with the exception of entry 4; however the regioselectivity remained consistent during the later cycles (Table 1, Entry 3). For entry 4, where the reactions were performed at lower syngas pressures, the B:L ratios were lower than all other runs and were more consistent over the five cycles. The dependence of regioselectivity on syngas pressure is well established.¹¹ As a profound colour change of the solutions occur during the CO₂-induced phase separations (Fig. 2), we suspect that newly formed bicarbonate anions (which first form in solution during the initial introduction of CO₂ after the first hydroformylation) are complexing the rhodium, displacing some phosphine. Coordination of bicarbonate to rhodium is well known.¹² When CO₂ is later removed from the system (to regenerate the monophasic system), it is possible that some bicarbonate remains complexed to the rhodium (as the original bright yellow solution is not completely regenerated). As the ligand system has already then incorporated bicarbonate after the first introduction of CO₂, a plateau of regioselectivity for later cycles is plausible. Introduction of CO₂ into the system before the first hydroformylation (to induce the phase separation) did not cause this colour change. It appears the post-reaction metal species are susceptible to bicarbonate coordination but not the pre-catalyst.

Fig. 2 Hydroformylation of styrene using switchable water. (Left) All reagents mixed before reaction or CO₂ treatment. (Centre) After completion of the hydroformylation reaction. (Right) After CO₂ treatment for 45 min. From Top to Bottom: Cycles 1 to 3 using the same catalyst and aqueous phase.

It is also possible that pH effects may be the cause of the diminished regioselectivity between the first and second cycles and the plateau after the second cycle. pH effects on the regioselectivity of biphasic hydroformylations have been noted previously.¹³ When removing CO₂ to regenerate the biphasic system, it is likely that not all the CO₂ is removed. This would then prohibit the solvent from completely returning to the initial pH. In fact, the first cycle is performed at pH = 12.1, while subsequent cycles are performed with pH around 9.6. As the pH swing is relatively consistent for cycles two to five, this could have lead to the more consistent regioselectivity observed.

Additionally, it was hypothesized that the regioselectivity is diminished due to the decreasing amount of TPPTS in the system through the cycles as TPPTS is oxidized (as was observed by ³¹P NMR). In such a case, we would then expect the regioselectivity to continually drop throughout the cycles as more and more TPPTS degrades. However we did not observe such an effect. In a control experiment we found that adding fresh TPPTS during the recycle stages (i.e. between cycles 1 and 2, 2 and 3, etc.) still did not maintain the high regioselectivity observed in the first cycle despite replenishment of the ligand. It is more likely that the presence of complexed bicarbonate and/or pH effects are the greater contributors to the initial drop of regioselectivity between cycles one and two and then the more consistent regioselectivity from cycles two to five.

Finally, the leaching of rhodium metal into the organic product phase was monitored using ICP-MS through three cycles (Table 1, entry 3). With an initial rhodium concentration of approximately 170 mg L⁻¹ in solution, the TMDAB system showed low leaching of rhodium into the product-containing organic phase (1.04 ± 0.01, 1.13 ± 0.02, and 1.52 ± 0.04 mg L⁻¹ in the first three cycles), further demonstrating the potential utility of this system for future catalysis. Of course, any leaching of precious metal catalysts is undesirable for industrial purposes. To this end, future research will delve into minimizing catalyst leaching into the organic product phase as well as the development of non-precious metal catalysts.

In summary, we have described the first example of the use of switchable water additives to allow homogeneous catalysis to take place in a monophasic solvent mixture and yet allow the subsequent catalyst/product separation to take place in a biphasic solvent mixture. This method does not suffer from the traditional mass transfer issues that accompany biphasic reactions because the system is monophasic during the catalysis. Furthermore it can also tolerate alkenes of lower water solubility than traditional biphasic aqueous/organic catalysis. The hydroformylation reactions performed in switchable water are run at low syngas pressures, on short timescales, and with facile separation of catalyst from product because of CO₂-induced phase separation. Recycling of the catalyst solution is performed with ease by removing CO₂ from the solution by sparging with an inert gas or air and moderate heating. The Rh/TPPTS catalyst maintains high conversion, product selectivity, and good regioselectivity through recycling. This method for solving the inherent rate limitations of conventional biphasic catalysis does not require high pressure CO₂ or expensive fluorous or ionic liquid solvents. Further studies into additional reactions, substrates, and the implementation of flow reactions are currently underway in our laboratory.

The authors thank the National Science and Engineering Research Council of Canada, the Canada Research Chairs program and the Walter C. Sumner Foundation programs for funding and the Killam Trusts for teaching release. The authors also thank Dr Diane Beauchemin and Dr Alemayehu Asfaw of Queen's University for assistance with rhodium analysis by ICP-MS.

Notes and references

1. (a) Catalyst Separation, Recovery and Recycling: Chemistry and Process Design, ed. D. J. Cole-Hamilton and R. P. Tooze, Springer, Dordrecht, 2006; (b) M. J. Muldoon, Dalton Trans., 2010, 39, 337.
2. (a) B. Cornils and E. G. Kuntz, in Aqueous-Phase Organometallic Catalysis, ed. B. Cornils and W. G. Hermann, Wiley-VCH, Weinheim, 2nd edn, 2004; (b) U. Hintermair, W. Leitner and P. G. Jessop, in Supercritical Solvents, ed. W. Leitner and P. G. Jessop Wiley-VCH, Weinheim, 2010.
3. (a) A. Andreetta, G. Barberis and G. Gregorio, Chim. Ind., 1978, 60, 891; (b) A. Buhling, P. C. J. Kamer, P. W. N. M. van Leeuwen, J. W. Elgersma, K. Goubitz and J. Fraanje, Organometallics, 1997, 16, 3027; (c) A. Buhling, P. C. J. Kamer, P. W. N. M. van Leeuwen and J. W. Elgersma, J. Mol. Catal. A: Chem., 1997, 116, 297; (d) S. L. Desset and D. J. Cole-Hamilton, Angew. Chem., Int. Ed., 2009, 48, 1472; (e) M. Mokhadinyana, S. L. Desset, D. B. G. Williams and D. J. Cole-Hamilton, Angew. Chem., Int. Ed., 2012, 51, 1648.
4. (a) Z. Jin, X. Zheng and B. Fell, J. Mol. Catal. A: Chem., 1997, 116, 55; (b) Z. Jin, Y. Wang and X. Zheng, in Aqueous-Phase Organometallic Catalysis, ed. B. Cornils and W. G. Hermann, Wiley-VCH, Weinheim, 2nd edn, 2004; (c) A. Behr, G. Henze and R. Schomäcker, Adv. Synth. Catal., 2006, 348, 1485.
5. (a) D. E. Bergbreiter, P. L. Osburn, A. Wilson and E. M. Sink, J. Am. Chem. Soc., 2000, 122, 9058; (b) D. E. Bergbreiter, P. L. Osburn and J. D. Frels, J. Am. Chem. Soc., 2001, 123, 11105; (c) D. E. Bergbreiter, J. Tian and C. Hongfa, Chem. Rev., 2009, 109, 530.
6. J. A. Gladysz and R. Corrêa da Costa, in Handbook of Fluorous Chemistry, ed. J. A. Gladysz, D. P. Curran and I. T. Horváth, Wiley-VCH, Weinheim, 2004.
7. (a) J. Lu, M. L. Lazzaroni, J. P. Hallet, A. S. Bommarius, C. L. Liotta and C. A. Eckert, Ind. Eng. Chem. Res., 2004, 43, 1586; (b) J. P. Hallet, J. W. Ford, R. S. Jones, P. Pollett, C. A. Thomas, C. L. Liotta and C. A. Eckert, Ind. Eng. Chem. Res., 2008, 47, 2585; (c) P. Pollett, R. J. Hart, C. A. Eckert and C. L. Liotta, Acc. Chem. Res., 2010, 43, 1237.
8. (a) S. M. Mercer and P. G. Jessop, ChemSusChem, 2010, 3, 467; (b) S. M. Mercer, T. Robert, D. V. Dixon, C.-S. Chen, Z. Ghoshouni, J. R. Harjani, S. Jahangiri, G. H. Peslherbe and P. G. Jessop, Green Chem., 2012, 14, 832.
9. CRC Handbook of Chemistry and Physics, ed. D. R. Lide, CRC Press, Boca Raton, 79th edn, 1998.
10. Z. Ma, X. Liu, G. Yang and C. Liu, Fuel Process. Technol., 2009, 90, 1241.
11. A. L. Watkins and C. R. Landis, J. Am. Chem. Soc., 2010, 132, 10306.
12. T. Yoshida, D. L. Thorn, T. Okano, J. A. Ibers and S. Otsuka, J. Am. Chem. Soc., 1979, 101, 4212.
13. R. M. Deshpande, Purwanto, H. Delmas and R. V. Chaudhari, J. Mol. Cat. A: Chem., 1997, 126, 133.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c2cy20095c

---