Pd-catalysed coupling reactions in supercritical carbon dioxide and under solventless conditions

Abstract

The homocoupling of iodoarenes catalysed by Pd(OCOCF₃)₂/P(2-furyl)₃ occurs best in scCO₂ and under solventless reaction conditions, and provides an attractive alternative to previously reported procedures. A comparison of the results in these cases suggests interesting preferential solvation effects are occurring in scCO₂. Each of the methods investigated has advantages and disadvantages which need to be carefully considered before the method of choice is determined.

---

Green Context

The choice of solvent can play a critical role in the development of a clean process. Here, a comparison of a range of solvents is carried out, with emphasis on scCO₂ and solvent-free conditions. Interestingly, there are many similarities between these two media, the reasons for which are discussed using the coupling reaction of iodoaromatics.

DJM

Introduction

New methods of carbon–carbon bond formation such as those catalysed by transition metals are of great importance to synthetic chemists owing to their versatility and power in assembling complex and often synthetically challenging carbon frameworks. However, they are generally carried out in conventional solvents such as N,N-dimethylformamide which can limit their attractiveness from the environmental point of view.¹ Notable exceptions to this are the use of environmentally acceptable alternative solvents such as water (although this has limitations resulting from the relative insolubility of organic compounds in such a reaction medium)² and ionic liquids.^3,4 Our initial interest in this area was to develop methods to allow such reactions to be carried out in supercritical CO₂ (scCO₂)⁵ which also has promise as an environmentally benign reaction medium, however in the course of these studies we have also found it necessary to investigate reactions in the absence of solvent, which is another alternative green technology rapidly emerging in importance.⁶ Our results reported in this paper demonstrate striking similarities between these two methods, and allow a comparison of the techniques which is of increasing importance when deciding on which particular methodology has the most green potential.⁷

We have recently reported methods which enable Heck, Suzuki and Stille reactions to be efficiently carried out in scCO₂ using simple commercially available reagents.⁸ This relied on the known enhanced solubility of fluorinated palladium sources (e.g. Pd(OCOCF₃)₂ and Pd(F₆-acac)₂) over those more conventionally used (e.g. Pd(OAc)₂), and the use of more lipophilic phosphine ligands (e.g. PBu₃, PCy₃, P(2-furyl)₃ (TFP)) which also have enhanced solubility in scCO₂ compared to the more conventional PPh₃. In the course of this work, we encountered an important and mechanistically interesting Pd-catalysed biaryl formation as a side reaction which would appear to be particularly facile in scCO₂. This forms the basis of the results described in this paper.

Results

In the course of our studies on Pd-catalysed reactions in scCO₂, we decided to investigate the use of acrolein as the alkene component in the Heck coupling reaction with iodobenzene, as it is known to be particularly poor, and we hoped that switching to an unusual reaction medium such as scCO₂ might improve the efficiency of such a reaction. We were disappointed to find that the proposed Heck reaction was unsuccessful, but instead were intrigued to find that the major product was biphenyl, arising from the oxidative coupling of iodobenzene. It was possible to optimise this reaction further by carrying it out in the absence of acrolein, to give biphenyl in 95% isolated yield (Table 1, entry 1).

Table 1 Homocoupling of iodobenzene under various reaction conditions

Entry^a	Catalyst	Ligand	Additive	Solvent	Conversion^b (%)	Yield^c (%)
a Reactions were carried out at 75 °C using Pd(OCOCF3)2 or Pd(OAc)2 (2 mol%) and DIPEA (1.6 equiv.), with TFP (4 mol%) and/or TBAB (50 mol%) as required.b Determined by ^1H NMR.c n.d. = Not determined.
1	Pd(OCOCF3)2	TFP	—	ScCO2	>95	95
2	Pd(OCOCF3)2	TFP	—	Toluene	12	n.d.
3	Pd(OCOCF3)2	—	TBAB	Toluene	90	66
4	Pd(OCOCH3)2	—	TBAB	Toluene	18	n.d.
5	Pd(OCOCH3)2	TFP	TBAB	Toluene	0	n.d.
6	Pd(OCOCF3)2	TFP	—	Solventless	76	n.d.

---

A survey of the literature revealed that similar reactions had been reported before, although they generally required the presence of additives such as tetra-n-butylammonium bromide (TBAB),⁹ Zn/H₂O with Pd/C,¹⁰ the use of palladacycle catalysts¹¹ or Pd(OAc)₂ often at temperatures much higher than the ones we were using (140 cf. 75 °C) requiring polar (and generally toxic) solvents such as DMF and DMA.^12,13 This latter observation is intriguing in that it is this reaction which bears most similarity to our work, yet is reported not to proceed in non-polar solvents such as p-xylene, which would be expected to be most similar to scCO₂ in terms of its solvent properties, particularly when compared to DMF. Indeed, on repeating our reaction in toluene, which is often considered to be a good model solvent for scCO₂, we observed only traces of the coupling product (Table 1, entry 2). This poses some important questions regarding the role solvent, particularly scCO₂, plays in these processes, and general assumptions often made regarding such reactions. We therefore decided to investigate the reaction in more detail to compare its efficiency in scCO₂, in toluene as a representative non-polar solvent, with and without the presence of TBAB for comparison with earlier work,⁹ and under solventless conditions.¹⁴ Initial results compared the coupling reaction of iodobenzene, and are summarised in Table 1.

It can be seen that there is a dramatic difference between reactions carried out in scCO₂ and toluene as already discussed. In the presence of TBAB the reaction in toluene is now significantly better (Table 1, entry 3), but note that Pd(OCOCF₃)₂ is significantly better than Pd(OAc)₂ (Table 1, entry 4), a factor not previously noted. Isolated yields tend to be lower in these reactions due to problems removing tetrabutylammonium residues (Table 1, entry 3).⁹ It is also interesting to note that use of both TFP and TBAB in toluene in the same reaction results in suppressing the reaction totally. Note also that the solventless reaction proceeds well without addition of TBAB, although not quite as well as for scCO₂. In all cases where significant reaction is observed, a red colouration indicative of oxidation of the diisopropylethylamine (DIPEA) is observed which is consistent with the proposed mechanism in previous reports.^9,12

We then investigated the coupling of functionalised iodoarenes to determine the effect of substitutents on the efficiency of the reaction (Table 2). These results show that the reaction is generally favoured by electron donating groups (CH₃, OCH₃), but in the presence of strong electron withdrawing groups (NO₂, CO₂CH₃), yields of the biaryl products are lower due to reductive deiodination which is now a serious side reaction, as is also often the case with alternative procedures.^9–13 In general, observations are similar to those for iodobenzene in that reactions are most efficient in scCO₂, but also work well in the absence of solvent. In toluene, conversions are low, but can be significantly improved by addition of TBAB. In the case of 4-iodonitrobenzene, reactions in scCO₂ and toluene proceed to completion even in the absence of TBAB, however the extent of reduction is significantly lower in toluene if TBAB is present. Importantly, for the solventless reaction, a very low conversion is obtained in this case presumably due to the high melting point of 4-iodonitrobenzene (175–177 °C),^15a and its low solubility in the neat reaction mixture (Table 2, entry 12). This contrasts sharply with the high conversion observed in scCO₂ (Table 2, entry 9) in which the 4-iodonitrobenzene would be expected to have at least some solubility. Attempted coupling of 1-iodonaphthalene was poor under all conditions but proceeded best in toluene in the presence of TBAB (Table 2, entry 16). The poor conversion of the reaction in scCO₂ can be readily explained, as a definite liquid layer was visible in the bottom of the reaction vessel under typical reaction conditions, consistent with the known low solubility of polycyclic aromatics in this medium.¹⁶

Table 2 Homocoupling of substituted iodobenzenes under various reaction conditions

Entry^a	X	Ligand	Additive	Solvent	Conversion^b (%)	Yield^c (%)
a Reactions were carried out at 75 °C using Pd(OCOCF3)2 (2 mol%), DIPEA (1.6 equiv.), with TFP (4 mol%) and/or TBAB (50 mol%) as required.b Determined by ^1H NMR.c n.d. = Not determined; values in parenthesis are ratio of biaryl formation to reductive deiodination.
1	CH3	TFP	—	ScCO2	>95	87
2	CH3	TFP	—	Toluene	11	n.d.
3	CH3	—	TBAB	Toluene	36	n.d.
4	CH3	TFP	—	Solventless	72	n.d.
5	OCH3	TFP	—	ScCO2	>95	82
6	OCH3	TFP	—	Toluene	16	n.d.
7	OCH3	—	TBAB	Toluene	27	n.d.
8	OCH3	TFP	—	Solventless	83	80
9	NO2	TFP	—	ScCO2	>95	44 (1∶1)
10	NO2	TFP	—	Toluene	>95	24 (1∶2)
11	NO2	—	TBAB	Toluene	95	n.d. (7∶3)
12	NO2	TFP	—	Solventless	<5	n.d.
13	CO2CH3	TFP	—	ScCO2	>95	46 (1∶1)
14	1-Iodonaphthalene	TFP	—	ScCO2	15	n.d.(1∶1)
15	1-Iodonaphthalene	TFP	—	Toluene	11	n.d.(1∶4)
16	1-Iodonaphthalene	—	TBAB	Toluene	39	n.d.(1∶3)
17	1-Iodonaphthalene	TFP	—	Solventless	28	n.d.

---

Discussion

The results show that the homocoupling of iodoarenes to form biphenyl derivatives can be an efficient process under a variety of conditions.¹⁷ One of the most striking aspects of these results is the similarity between solventless reaction conditions and those obtained in scCO₂, which contrast sharply with those obtained in a non-polar solvent such as toluene. Previous results from our group have demonstrated related effects, where double bond migration following an intramolecular Heck reaction is significantly suppressed in scCO₂ and under solventless conditions when compared to reactions in conventional solvents such as toluene and acetonitrile.¹⁴ Concern that reactions in scCO₂ could be occurring neat in the bottom of the reactor were addressed by carrying out studies in high pressure view cells. Small traces of liquid can be seen in some cases around the side of the vessels, and solid products precipitate out as they are formed, but the vast majority of reagents in the initial period of the reaction are in a homogeneous phase with the CO₂ unless otherwise stated. The difference in results observed in some cases between neat and scCO₂ reactions (cf. entries 1 and 4, 9 and 12, 14 and 17, Table 2), and analogy with our previously reported homogeneous Pd-catalysed coupling reactions,^8,14 also strongly suggest the reactions are occurring in the supercritical phase. The possibility that these reactions may be occurring neat cannot be totally discounted, but if this were to be the case, the reactions reported here would represent further examples of carbon dioxide accelerated solventless reactions, which are fascinating themselves.¹⁸ However a more likely explanation to account for the similarity of the results obtained in scCO₂ and under solventless can be obtained by considering the nature of reactions in scCO₂ in more detail.

A common problem when using scCO₂ as a reaction medium is that because of its generally weak solvent power, we are often working at the borderline of solubility. Whilst many of the catalyst systems we have devised have high solubility in scCO₂,^8,14 the Pd(OCOCF₃)₂/P(2-furyl)₃ system is somewhat borderline, with small traces of solid being visible. The base (DIPEA) is totally miscible with the scCO₂.¹⁹ However some of the reagents we are using have a finite solubility in pure scCO₂, significantly less than the typical concentrations of our reactions. For example, iodobenzene has a solubility of approximately 1.8 mg ml⁻¹ in scCO₂ at 80 °C, 107 bar¹⁸ and other more polar arenes (e.g. 4-iodonitrobenzene) are likely to be significantly less soluble. We are typically working at a concentration of 18 mg ml⁻¹ iodobenzene with 25 mg ml⁻¹ of DIPEA. The Pd(OCOCF₃)₂ catalyst is essentially insoluble in scCO₂ on its own, but rapidly dissolves in the presence of phosphine and base. It is thus the solubility of the combination of reagents which make up the reaction mixture which is important in determining whether a reaction is homogeneous and likely to proceed effectively. In order to improve the solvent power of scCO₂, co-solvents are often added such as MeOH, MeCN and toluene, and these can dramatically enhance the solvent power of the medium. However, reagents will also act as cosolvents, and in the case of this reaction, the amine, the phosphine and aromatic substrate will all help each other to dissolve in the scCO₂ medium. As a result, it would be expected that there would be preferential interaction between these reagents than with the CO₂ solvent. Indeed, such effects are well established for supercritical fluids²⁰ where local concentrations of cosolvents around substrates can be as high as 10 times the bulk values, particularly near the critical point. Although we are working at temperatures significantly higher than the critical point of CO₂ (31.1 °C), it is important to realise that for a given pressure, solvent power generally decreases with increasing temperature as the solvent density decreases. Hence factors which can enhance the solubility of reagents are still very important under the conditions we are working.

In light of these considerations, it is possible to suggest why these reactions in scCO₂ resemble those of solventless conditions more closely than those in non-polar solvents. If preferential solvation is occurring, then the reaction is most likely to be taking place in areas of exaggerated reagent concentration. This effectively forms a region similar in composition to a solventless reaction mixture, resulting in a similar product distribution. In a conventional non-polar solvent such as toluene, such an effect would not be expected as the individual organic reagents are all independently miscible with the solvent. It would also be expected that the solventless reaction mixture would be significantly more polar than toluene, particularly as organic salts are produced during the reaction. This may account for the similarity with reactions conducted in more polar solvents.^12,13

The relative efficiencies of the different reaction media are worthy of comment. In general, reactions in scCO₂ are cleanest and easiest to purify, but require high-pressure apparatus. Reactions in toluene require the addition of TBAB for best conversions, which results in more complex purification procedures, lower isolated yields and significant waste. The solventless procedures are attractive, particularly on a small scale, although limitations with high melting reagents such as 4-iodonitrobenzene can be problematic, as may be efficient mixing and control of exotherms for larger scale processes. Reaction work-up and purification may require the use of solvents, as is also the case for reactions in scCO₂.²¹ As with many aspects of green chemistry, the final choice of conditions will be determined by a variety of factors depending on the particular reaction. Comparative studies help identify potential advantages and disadvantages of the competing technologies, and will be crucial in assessing their relative impact in the green chemistry arena.

In summary, we have shown that the coupling reaction between iodoarenes occurs best in scCO₂ and under solventless reaction conditions, and provides an attractive alternative to previously reported procedures. The similarity between the results in these cases suggest interesting preferential solvation effects are occurring in scCO₂. Each of the methods investigated has advantages and disadvantages which need to be carefully considered before the method of choice is determined.

Experimental

Typical procedure

CAUTION: As with all reactions under high pressure, appropriate safety precautions must be taken. Iodobenzene (0.200 ml, 1.78 mmol), diisopropylethylamine (0.500 ml, 2.87 mmol), palladium trifluoroacetate (11.9 mg, 0.0356 mmol) and tris-2-furylphosphine (16.6 mg, 0.0712 mmol) were added in the order described to a 20.0 ml high-pressure reaction vessel, which was sealed and connected to a CO₂ supply (an Isco 260D controllable syringe pump with a cooled head).^5,8 The vessel was charged with CO₂ until a pressure of 48 bar was obtained and heated at 80 °C to a stabilised pressure of 71 bar. More CO₂ was added to obtain a pressure of 109 bar (11.0 MPa) and the reactants left to stir under these conditions for 15 h. The stirrer and temperature controls were then switched off, CO₂ gradually released into the solvent trap (40.0 ml), the vessel disconnected and allowed to cool. The apparatus was cleaned with dichloromethane (4 × 1.00 ml) and the washings directed into the solvent trap. The contents of the solvent trap were combined with the dark red solids obtained from the reactor vessel, and the resulting organic phase washed with aqueous saturated ammonium chloride (3 × 20.0 ml), dried with MgSO₄, concentrated and purified [flash silica (40 g); hexane eluent] to give biphenyl (204 mg, 95%) as colourless needles, mp 69–71 °C (from light petroleum bp 40–60 °C, lit.^15b 69–72 °C).

Acknowledgements

We are very grateful to the following members of the Leeds Cleaner Synthesis Group and their respective companies for funding and useful discussions: Drs Andrew Bridge, Patrick Ducouret, and Antonio Guerreiro, Aventis; Drs Mike Loft, Ken Veal and John Strachan, GlaxoSmithKline; Ms Julie MacRae, Drs Jo Negri, and Laurence Harris, Pfizer Central Research; and Mr Bill Sanderson, consultant to Solvay Interox. We also thank the EPSRC and the University of Leeds for funding.

Notes and references

1. R. F. Heck, Comprehensive Organic Synthesis, ed. B.M. Trost, Pergamon Press, Oxfor, vol. 4, ch. 4.3; S. Brase and A. de Meijere, Metal Catalysed Cross Coupling Reactions, ed. F. Diederich and P.J. Stang, J. Wiley-VCH, 1998, ch. 3.
2. For a review, see: J. P. Genet and M. Savignac, J. Organomet. Chem., 1999, 576, 305.
3. C. J. Mathews, P. J. Smith and T. Welton, Chem. Commun., 2000, 1249; for a related biaryl coupling in an ionic liquid, see: J. Howarth, P. James and J. Dai, Tetrahedron Lett., 2000, 41, 10319.
4. T. Welton, Chem. Rev., 1999, 99, 2071.
5. For recent reviews see: R. S. Oakes, A. A. Clifford and C. M. Rayner, J. Chem. Soc., Perkin Trans. 1, 2001, 917; P. G. Jessop, T. Ikariya and R. Noyori, Chem. Rev., 1999, 99, 475.
6. G. W. V. Cave and C. L. Raston, Chem. Comm., 2000, 2199 ; it is known that Heck reactions can be carried out using a tertiary amine base as solvent, see: B. A. Patel, C. B. Ziegler, N. A. Cortese, J. E. Plevyak, T. C. Zebovitz, M. Terpko and R. F. Heck, J. Org. Chem., 1977, 42, 3903 ; see also ref. 1.
7. D. J. C. Constable, A. D. Curzons, L. M. Freitas dos Santos, G. R. Geen, J. Kitteringham, P. Smith, R. E. Hannah, M. A. McGuire, R. L. Webb, M. Yu, J. D. Hayler and J. E. Richardson, Green Chem., 2001, 3, 7.
8. N. Shezad, R. S. Oakes, A. A. Clifford and C. M. Rayner, Tetrahedron Lett., 1999, 40, 2221 ; for related research see: M. A. Carroll and A. B. Holmes, Chem. Commun., 1998, 1395; D. K. Morita, D. R. Pesiri, S. A. David, W. H. Glaze and W. Tumas, Chem. Commun., 1998, 1397; B. M. Bhanage, Y. Ikushima, M. Shirai and M. Arai, Tetrahedron Lett., 1999, 40, 6427; S. Cacchi, G. Fabrizi, F. Gasparrini and C. Villani, Synlett, 1999, 345.
9. J. Hassan, C. Gozzi and M. Lemaire, C. R. Acad. Sci., Ser. IIc, Chim., 2000, 3, 517; J. Hassan, C. Hathroubi, C. Gozzi and M. Lemaire, Tetrahedron Lett., 2000, 41, 8791.
10. S. Venkatraman and C.-J. Li, Org. Lett., 1999, 1, 1133; S. Mukhopadhyay, G. Rothenberg, D. Gitis and Y. Sasson, Org. Lett., 2000, 2, 211; S. Mukhopadhyay, G. Rothenberg, H. Wiener and Y. Sasson, Tetrahedron, 1999, 55, 14763.
11. F.-T. Luo, A. Jeevanandam and M. K. Basu, Tetrahedron Lett., 1998, 39, 7939.
12. M. Brenda, A. Knebelkamp, A. Greiner and W. Heitz, Synlett, 1991, 809.
13. D. D. Hennings, T. Iwama and V. H. Rawal, Org. Lett., 1999, 1, 1205.
14. N. Shezad, A. A. Clifford and C. M. Rayner, Tetrahedron Lett., 2001, 42, 323.
15. Aldrich Handbook of Fine Chemicals and Laboratory Equipment 2000–2001, Sigma Aldrich Co. Ltd., Gillingham, Dorset, UK; (a) p. 997; (b) p. 191.
16. For a discussion of factors affecting solubility in supercritical fluids, see: F. P. Lucien and N. R. Foster, in Chemical Synthesis in Supercritical Fluids, ed. P.G. Jessop and W. Leitner, Wiley-VCH, Weinheim, 1999.
17. For alternative methods of biaryl formation, see: G. Bringmann, R. Walter and R. Weirich, Angew. Chem., Int. Ed. Engl., 1990, 29, 977 ; see also refs. 9–13 and references cited therein.
18. P. Jessop, D. C. Wynne, S. DeHaai and D. Nakawatase, Chem. Commun., 2000, 693.
19. Determined using the method described in: J. M. Dobbs, J. M. Wong and K. P. Johnson, J. Chem. Eng. Data, 1986, 31, 303 ; in related studies, iodobenzene is reported to have a solubility of 1.2 mg ml⁻¹ in scCO₂ at 100 °C, 133 bar, and triethylamine is totally miscible, see: S. Cacchi, G. Fabrizi, F. Gasparrini and C. Villani, Synlett, 1999, 345.
20. For an excellent review in this area, see: J. F. Brennecke and J. E. Chateauneuf, Chem. Rev., 1999, 99, 433.
21. In principle, scCO₂ could be used as a mobile phase for preparative chromatographic purification of such product mixtures which would greatly reduce the amount of solvent waste produced. For a recent review, see: T. L. Chester and D. J. Pinkston, Anal. Chem., 2000, 72(12), 129R.

---