A Microfluidic System for the Continuous Recycling of Unmodified Homogeneous Palladium Catalysts through Liquid/Liquid Phase Separation The MIT Faculty has made this article openly available. Please share how this access benefits you. Your story matters. Citation Li, Pengfei, Jason S. Moore , and Klavs F. Jensen. “A Microfluidic System for the Continuous Recycling of Unmodified Homogeneous Palladium Catalysts through Liquid/Liquid Phase Separation.” ChemCatChem 5, no. 7 (May 16, 2013): 1729–1733. As Published http://dx.doi.org/10.1002/cctc.201300054 Publisher Wiley Blackwell Version Author's final manuscript Citable link http://hdl.handle.net/1721.1/92778 Terms of Use Creative Commons Attribution-Noncommercial-Share Alike Detailed Terms http://creativecommons.org/licenses/by-nc-sa/4.0/ Continuous Recycling DOI: A Microfluidic System for Continuous Recycling of Unmodified Homogeneous Palladium Catalysts via Liquid-Liquid Phase Separation** Pengfei Li,* Jason S. Moore, Klavs F. Jensen* With their highly tunable nature and the resulting high activity and and received considerable attention.[16,17] Ryu and coworkers selectivity in promoting transformations, homogeneous catalysts reported a continuous-flow process for recycling of an ionic liquid- have dramatically shaped modern organic chemistry.[1] Industrial supported palladium catalyst in Mizoroki-Heck reaction.[12a] The applications are limited, however, by the difficulties associated with ligand and the ionic liquid used in this reaction shared the same catalyst recovery and recycling.[1b,c] As a result, many strategies structural framework. Theberge and Huck designed an have been devised to overcome those challenges,[2] including aqueous/fluorous biphasic microflow setup and used it for immobilizing the catalysts onto solid or liquid supports by structural continuous recycling of a fluorous guanidine-ligated palladium modifications,[3-7] enlarging the catalysts for ultra- or catalyst for Suzuki-Miyaura coupling.[12b] In both examples, static nanofiltration,[8] and using controlled solubility variations.[9,10] vials were used for passive liquid-liquid phase separation of the Although leaching of catalyst has remained a general challenge, pooled multiphasic mixtures. Furthermore, a “release-trap” strategy some of these methods have proved successful. However, the using coordinative polymers has been reported for continuously approaches can fall short for demanding reactions, such as reuse of catalysts.[12c,d] However, there have been no precedents palladium-catalyzed cross-coupling reactions often involving strong about continuous recycling of unmodified palladium catalysts based bases and high temperatures[11,12] and requiring deliberately on advanced ligands.[5a,18] In this communication, we describe a designed ligands. Therefore, it would be preferable to recycle the microfluidic system for continuous catalyst recycling in palladium- unmodified (“native”) catalyst directly under the best performing catalyzed hydroxylation of aryl bromides. A highly-active bulky conditions. phosphine was used as the supporting ligand and a microfluidic There are both chemical and technical hurdles associated with device for liquid-liquid phase separation. catalyst recycling of unmodified systems. First, highly active Palladium-catalyzed selective hydroxylation of aryl halides catalysts could deactivate when out of the reaction environment using biaryl phosphine ligands, first reported by Buchwald in and/or during the recycling processes. Second, convenient and 2006,[19] is a general method to prepare various phenols from readily efficient catalyst isolation is non-trivial and sometimes impossible. available aryl halides.[20] Later, Beller developed N-aryl imidazole- Third, manual operations of sensitive catalysts must be carefully based phosphine ligands for this reaction[20b,21] and interesting implemented, and the process can be very time-consuming. cationic ligands for catalyst recycling.[11a] This represented a rare Our labs have been developing continuous microflow processes example of recycling palladium catalysts in complex coupling of palladium-catalyzed cross-coupling reactions.[13] Owing to the reactions. However, the moisture-sensitive reaction conditions and high surface-volume ratio of the liquids in microchannels, heat- and using product-saturated 1,4-dioxane as the solvent for product mass-transfer are very efficient, enabling easy and precise control of separation limited the practicality. We felt that if an immiscible process parameters.[14] Along with other merits, such as producing organic/water mixture could work for this reaction, the water- less waste, reducing footprint, and ready scale-up, continuous-flow soluble phenolate could be efficiently separated from the catalyst- reactions have been regarded as one of the key green processes[15] containing organic phase. Moreover, we have previously developed liquid-liquid biphasic systems for palladium-catalyzed reactions under continuous-flow conditions.[13] Consequently, we commenced [] Prof. Dr. P. Li Center for Organic Chemistry our study by searching for an efficient biphasic catalytic system Frontier Institute of Science and Technology using 3-bromoanisole as a model substrate. Previously, two Xi’an Jiaotong University phosphines L1 and L2 were shown effective as supporting ligands 99 Yanxiang Road, Xi’an, Shaanxi, 710054 (China) for this reaction in 1,4-dioxane, with L2 performing better in most E-mail: [email protected] cases.[19a] With these ligands, we found that in a mixture of toluene and water, the reaction also proceeded well, but only when a phase- J. S. Moore, Prof. Dr. K. F. Jensen transfer catalyst (tetrabutylammonium bromide, TBAB) was added Department of Chemical Engineering as a promoter. Further optimizations under batch conditions were Massachusetts Institute of Technology 77 Massachusetts Avenue, Cambridge, MA 02139 (USA) conducted and selected results are shown in Table 1. E-mail: [email protected] Under the otherwise same conditions (Pd2(dba)3 0.5 mol%, ligand 2 mol%, 100 oC, 120 min), the reaction was complete using J. S. Moore ligand L2, while L1 led to 60% conversion (entry 2 vs. entry 1). For Dow Chemical Company an efficient flow reaction, a faster reaction would be desirable. 2301 N. Brazosport Blvd., Freeport, TX 77541 (USA) When the reaction time was decreased to 10 minutes, however, even [] P. L., J. S. M., and K. F. J. thank the Novartis International 1 mol% Pd2(dba)3 and 4 mol% of L2 gave an incomplete conversion AG for funding. P. L. thanks Prof. Stephen L. Buchwald for (92%, entry 3). At this point, a more electron-rich ligand L3[22] was his support. tested and proved more effective. With only 0.5 mol% of Pd2(dba)3 Supporting information for this article is available on the and 2 mol% of L3, the product was quantitatively formed (entry 4). WWW under or from the author. Very recently, L3 has been successfully utilized in several challenging coupling reactions;[23] here we extend its usefulness to 1 o selective hydroxylation of aryl halides. Because Pd2(dba)3 is only soluble standard sodium tosylate were heated at 100 C for 5 slightly soluble in toluene at room temperature, which may not be minutes. After the mixture was cooled down to room temperature, suitable for studying under flow conditions, allylpalladium chloride the aqueous phase containing the phenolate was carefully removed dimer was selected as a readily soluble palladium source and proved using a syringe. The yield was based on UHPLC analysis of the similarly effective (entry 5). However, we observed that palladium aqueous phase taking sodium tosylate as the internal standard. 3- black was formed immediately after reaching full conversion, which Bromoanisole (100 mol%) and the aqueous solution (same as the was in contrast to the case of Pd2(dba)3, where no palladium black first run) were added to the reaction tube via syringes, and the was observed. Further decrease in reaction time led to incomplete resulting mixture was again heated at 100 oC for 5 minutes, followed conversion and the reaction using allylpalladium chloride dimer was by cooling down and phase separation. This procedure was repeated slightly faster than the one using Pd2(dba)3 (entry 6 and 7), and both up to a total five runs. The results in Scheme 1 did show that the reactions did not form significant amounts of palladium black. catalyst was recyclable under CIC conditions. However, the yields fluctuated significantly, apparently due to the difficulties associated Table 1. Optimizations in batch.[a] with precise control of the reaction times and temperatures and possible errors during manual operations. [Pd] x mol%, Br OMe L 2x mol% HO OMe [allylPdCl]2 0.5mol%, toluene, 2M KOH, Br OMe L3 2 mol%, toluene, HO OMe TBAB 5 mol%, 100 oC 2M KOH, TBAB 5 mol%, o Me 100 C, 5 min OMe Me Me Run Yield (%) P(t-Bu)2 MeO P(t-Bu) Me P(t-Bu) 2 i-Pr i-Pr 2 i-Pr i-Pr 1 68 i-Pr i-Pr 2 70 3 58 ArBr, [Pd], L, 4 63 biphenyl in Tol. i-Pr i-Pr i-Pr 5 55 aq. ArOK, KOH, TBAB, NaOTs L1 L2 L3 Scheme 1. Catalyst recycling in batch conditions. Entr Ligand Pd source t (min) Conversion/ y Yield (%)[b] In order to obtain reaction profiles in a flow setting, we assembled a continuous-flow setup using syringe pumps, PFA 1 L1 (2 mol%) Pd2(dba)3 120 60/58 (0.5 mol%) tubing, tee and cross junctions, and a packed-bed reactor. (see SI). Tubular reactors, filled with stainless steel spheres (60-120 m 2 L2 (2 mol%) Pd2(dba)3 120 100/95 (0.5 mol%) diameter), have been used in continuous-flow reactions to efficiently enhance biphasic mass- and heat-transfer processes.[13b,e,g] Initially, a 3 L2 (4 mol%) Pd2(dba)3 10 92/85 (1 mol%) stock solution of allylpalladium chloride dimer in 1,4-dioxane was used as one of the streams and was pumped into the reaction. We 4 L3 (2 mol%) Pd2(dba)3 10 100/98 (0.5 mol%) found that the reaction results were not reproducible, probably because the catalyst (Pd0L) formation process was not effective.