Benchtop factory for cross-coupling reactions by circulatory catalyst flow using low-viscosity ionic liquid as reaction medium and catalyst support

1 Introduction

In the past two decades, new reaction media, such as ionic liquids [1–7] and fluorous solvents [8–11], have emerged as a greener substitute for conventional organic solvents. These solvents are less volatile than organic solvents, and biphasic treatment with organic solvents provides a useful means for the separation of these solvents and the products [12]. Generally, separation and reuse of homogeneous catalysts are not easy; however, the immobilization of the catalyst in these solvents allows for the facile separation of the products and the catalyst. Whereas the limited lipophilic nature of these solvents merits for the separation of the products by biphasic treatment with organic solvents, the solubility issue of the substrates and the reagents often plagued the efficiency of the reaction. It is conceivable that effective mixing is a clue for efficient reaction whenever these new reaction media are used as solvents.

Recently, evolving microreaction technology has opened up new avenues for organic synthesis and chemical production [13–22]. Flow microreactors have several advantages over conventional batch reactors, which include efficient mixing, rapid heat and mass transfer, and operational safety inherent to a tiny reaction space. Work in our laboratory over a decade has focused on the use of microreactors for practical organic synthesis [22], which includes catalytic reactions [23–26], radical reactions [27–29], carbonylation reactions [30–32], and photo-irradiation reactions [33–37].

As for the catalysis, much of the early work on the use of microreactors focused on heterogeneous catalysis because it can utilize the high volume-to-surface ratio ensured by microchannels (Scheme 1). The potential of this new flow technology, however, is not necessarily restricted to heterogeneous catalysis. The application of the microreaction technology to reactions using a homogeneous catalyst in conjunction with a new reaction medium should have merit for effective reaction, as the effective mixing of biphasic phases would be ensured by the high-performance micromixer. Even at the stage of extractive workup, the use of a micromixer instead of a separation funnel would contribute to smooth biphasic separation. Furthermore, the continuous flow reaction/workup system would be sequenced by the continuous separation of the product from the reaction medium containing catalyst, which can be recycled to create a circulatory flow catalytic system. In such a system, the minimum use of somewhat precious designer solvents would be investigated. In this article, we discuss the usefulness of continuous flow reactions in cross-coupling reactions using a low-viscosity ionic liquid as an efficient recycling medium and a catalyst support. Overall, such a circulatory catalyst flow system can mimic flow reaction using a heterogeneous catalyst.

Scheme 1

Heterogeneous and homogeneous catalytic reactions in microflow system.

2 Cross-coupling reactions using low-viscosity ionic liquid, [bmim]NTf₂

Ionic liquids are becoming increasingly popular as a reaction medium for synthetic organic reactions, which involve a wide variety of catalytic reactions [38]. Large-scale synthesis of the fine chemicals in ionic liquids in a batch reactor can often be impeded by its lipophobic nature and relatively high viscosity compared to that of conventional organic solvents. We found that the combination of such a reaction system with a micromixer would be beneficial; however, we soon realized that because of the large pressure drop, high-viscosity ionic liquids are not necessarily suitable for flow reactions using microreactors. Thus, we need low-viscosity ionic liquids for continuous-flow reactions. In Scheme 2, the viscosity and solubility data of two types of imidazolinium ionic liquids, [bmim]PF₆ and [bmim]NTf₂, are summarized [38–40]. Compared to [bmim]NTf₂, [bmim]PF₆ has a higher viscosity; however, a problem associated with the use of [bmim]NTf₂ is encountered at the stage of extractive workup. As [bmim]NTf₂ has higher solubility in organic solvents, biphasic separation is not perfect. For example, ether, a typical extractive solvent, is not a good solvent for the biphasic separation with [bmim]NTf₂ because a significant amount of the ionic liquid goes into the ether layer. After a survey of several organic solvents, however, we found that cyclohexane is a good countersolvent for [bmim]NTf₂ (Scheme 2).

Scheme 2

Two types of imidazolinium ionic liquids. Viscosity and solubility in organic solvent.

In 2002, we reported that with [bmim]PF₆ as a solvent, the Sonogashira reaction does proceed without the use of copper co-catalyst [23]. In this report, we also described the first microflow Sonogashira reaction using Pd-NHC complex A as a catalyst, which was interfaced with an IMM micromixer having a 40-μm channel width. Some of the updated data of flow Cu-free Sonogashira reaction are shown in Scheme 3. In 2004, we first demonstrated that a low-viscosity ionic liquid, [bmim]NTf₂, is an excellent reaction medium for cross-coupling reactions, such as the Mizoroki-Heck reaction, Suzuki-Miyaura coupling reaction, and Stille coupling reaction [39]. For example, using Pd-NHC complex A as a catalyst and [bmim]NTf₂ as a reaction medium, the Mizoroki-Heck reaction of iodobenzene and butyl acrylate to give butyl cinnamate can be repeatedly carried out with catalyst and solvent recycling (Scheme 4). We also reported Pd-accelerated atom transfer carbonylation [41], which was conducted in ionic liquids such as [bmim]PF₆ and [bmim]NTf₂ (Scheme 5) [40]. After biphasic treatment using an organic solvent such as cyclohexane, the products can be separated from the catalyst solution of the ionic liquid. The ionic liquid containing the Pd catalyst can be recycled for the second-run reaction.

Scheme 3

Copper-free Sonogashira coupling reaction using a flow microreactor.

Scheme 4

Mizoroki-Heck reaction with catalyst/solvent recycling.

Scheme 5

Metal-accelerated atom transfer carbonylation using ionic liquids and Pd-NHC complex.

3 Circulatory microflow Mizoroki-Heck reaction using ionic liquid

As mentioned above, a low-viscosity ionic liquid, [bmim]NTf₂, is an excellent reaction medium for Pd-catalyzed cross-coupling reactions [39]. The use of the low-viscosity ionic liquid for flow chemistry would have two advantages: (i) rapid diffusion permitting rapid reaction and (ii) small pressure drop ensuring a smooth flow. Using the Mizoroki-Heck reaction of iodobenzene with butyl acrylate in [bmim]NTf₂ as a model (Scheme 4), we attempted to create a continuous microflow system with automated catalyst and solvent recycling (Scheme 6) [24].

Scheme 6

Continuous-flow Mizoroki-Heck reaction and triphasic workup.

For the continuous reaction system, we used an automated microflow apparatus, CPC CYTOS Lab System (Cellular Process Chemistry GmbH, Mainz, Germany), which is equipped with two pumps, a micromixer (channel width=100 μm, inner volume=2 ml), and a residence time unit (inner volume=15 ml), with an intelligent control unit. Using this automated microflow system, we introduced a mixture containing iodobenzene, butyl acrylate, and tripropylamine from one inlet of the micromixer (0.5 ml/min), and a [bmim]NTf₂ solution containing Pd-NHC complex A was introduced from the other inlet (0.5 ml/min) (Scheme 7). The catalytic reaction took place after the two solutions were mixed and was brought to completion in the residence time unit (residence time, 17 min, at a total flow rate of 1.0 ml/min), where the temperature was controlled at 130°C. From the resulting mixture, the coupling product, butyl cinnamate, was obtained nearly quantitatively by conventional extraction with hexane. The by-product, an ammonium salt, was removed from the resulting ionic liquid layer by washing with a copious amount of water. The recovered ionic liquid could be used again in the next run without a significant drop in product yield. We also tested a high-viscosity ionic liquid, [bmim]PF₆, but, contrary to expectation, this did not flow smoothly, overburdened the pump units, and resulted in inferior yields.

Scheme 7

Mizoroki-Heck reaction using a continuous microflow system.

This successful result with [bmim]NTf₂ led us to explore a totally automated flow system with continuous catalyst/solvent recycling. We designed a dual flow workup system based on the use of two serially connected T-shaped micromixers, whose setup is schematically shown in Scheme 8. The product could be extracted by mixing with an organic solvent, such as hexane, and ammonium salt was removed by mixing with H₂O. The resulting mixture separated into three phases: a hexane layer containing the product and excess amount of tripropylamine, an aqueous layer containing the inorganic salt, and the ionic liquid layer containing the Pd catalyst. The ionic liquid layer was continuously pumped back to the container to be introduced in the reaction system.

Scheme 8

Workup system based on dual flow microextraction system.

In Scheme 9, a totally automated, circulatory continuous-flow system connected with microextraction units is illustrated. We used two serially connected residence time units (total volume=30 ml). The reaction mixture exiting from the CYTOS Lab System was introduced into a T-shaped micromixer (300 μm), where hexane was mixed to extract the coupling product. The mixture was then mixed with 0.5 M NaOH aqueous solution in another T-shaped micromixer (300 μm) to remove ammonium salt by-product. We used a Y-shaped glass flask as a reservoir for the phase separation. On standing in the Y-shaped glass flask, the resulting mixture separated into three phases: a hexane layer containing the product, an aqueous layer, and the ionic liquid layer containing the Pd catalyst. Overflowed organic and aqueous phases were collected in a separate container. The ionic liquid in the bottom layer was pumped back to the catalyst solution container for recycling. After running the whole system for 11.5 h, where 144.8 g (0.71 mol) of iodobenzene together with the corresponding amount of acrylate and amine were consumed (total volume=408 ml), 115.3 g of trans-butyl cinnamate was obtained in an 80% (10 g/h) yield after purification by silica gel chromatography. This corresponds to a performance in which the ionic liquid with the Pd catalyst (90 ml) was recycled about five times during this overall catalytic reaction.

Scheme 9

Production of butyl cinnamate using circulatory catalyst flow system.

4 Microflow Sonogashira reaction

We then investigated the synthesis of a key intermediate for a matrix metalloproteinase (MMP) inhibitor 1 [42], using a continuous-flow Sonogashira coupling reaction of 2-bromothiophene 2 with p-tolylacetylene as the key step (Scheme 10). Consequently, we succeeded in the 100 g order synthesis of the MMP inhibitor in a continuous-flow system using PdCl₂(PPh₃)₂ and CuBr as the catalyst and dimethylformamide (DMF) as the solvent (Scheme 10) [25]. We then improved the reaction conditions to be Cu-free (Scheme 11), for which Pd-NHC catalyst A worked well as the catalyst. For a quick search of optimized flow reaction conditions, we used an automated microreactor system for conditions search, MiChS System X-1 (MiChS Co. Ltd., Yao, Osaka, Japan) [43], which proved to be very useful. After testing 15 conditions, we found that the reaction conditions of 5 min of the residence time at 100°C gave a satisfactory result.

Scheme 10

Continuous-flow synthesis of MMP inhibitor using conventional Sonogashira reaction.

Scheme 11

Quick optimization for Cu-free Sonogashira reaction in DMF using an automated microreactor system, MiChS System X-1.

Then, we tested [emim]NTf₂ instead of DMF to create a catalyst recycling system, whose viscosity value of 34 mPa·s at 20°C is even smaller than that of [bmim]NTf₂ [38]. The reaction using [emim]NTf₂ worked quite well but often encountered solubility issues with the substrate 2 to this ionic liquid and 1 to extracting solvent such as cyclohexane. To circumvent such limitations, we reconsidered another synthetic route suitable for the microflow system (Scheme 12). We planned to use sulfonyl ester derivatives 3 as a coupling partner of p-tolylacetylene and expected that the coupling product 4 would react with methyl ester of tryptophan 5 to give the methyl ester 1'.

Scheme 12

Modified synthetic route to the MMP inhibitor.

We were delighted to find that the envisioned route indeed worked well in the batch reaction (Scheme 13). Thus, when the reaction of pentafluorophenyl bromothiophenylsulfonate 3 with tolylacetylene in [emim]NTf₂ was carried out in the presence of ⁱPr₂NEt as a base and Pd-NHC A as a catalyst, a homogeneous solution resulted at room temperature. The mixture remained homogeneous during and after the reaction (80°C to room temperature). The coupling product 4 could be extracted by 30% ether in cyclohexane. After silica gel chromatography, 4 was obtained in 70% yield. The obtained coupling product 4 reacted with 5 to give 1' in good yield.

Scheme 13

Reaction of pentafluorophenyl ester 3 with p-tolylacetylene in a batch reactor system.

We then investigated the same reaction in a microflow system using a T-shaped mixer (1000 μm i.d.) and a stainless steel tube (1000 μm i.d.×50 cm). The reaction at 95°C with 1.75 min residence time gave the coupling product 4 with complete conversion (Scheme 14).

Scheme 14

Reaction of pentafluorophenyl ester 3 with p-tolylacetylene in a flow reactor system.

5 Copper-free Sonogashira reaction accompanied by circulatory catalyst flow

We then applied the synthesis of the intermediate of MMP inhibitor to the circulatory microflow system consisting of continuous flow reaction, flow extraction and separation, and catalyst/solvent recycling [23]. Because the Sonogashira reaction is fast, we used a residence time unit having a larger diameter of 2000 μm to obtain better throughput. Another reason we modified the system is the introduction of water to the residence time unit. This is because the reaction mixture obtained in a flow reaction had a higher viscosity due to the formation of an ammonium salt by-product. Consequently, we employed a mixed-flow two-layer system, using a 2000 μm i.d. tube (1 m length) as a residence time unit, in which 1.5 M potassium carbonate aqueous solution was mixed with the reaction mixture to remove the ammonium salt quickly from the ionic liquid layer. This worked quite well. Thus, a homogeneous mixture of 3, p-tolylacetylene, and ⁱPr₂NEt was mixed with [emim]NTf₂ containing a Pd catalyst using a T-shaped mixer (600 μm i.d.). The mixture was then treated with 1.5 M potassium carbonate aqueous solution, and the mixture was introduced into a 2000 μm i.d. tube that was heated at 95°C for 1.75 min (residence time). Then, a blended cyclohexane with diethyl ether (30%) was mixed using another T-shaped mixer (400 μm i.d.) and passed through another tube (2000 μm i.d.×25 cm, 40°C) for extraction. The exiting reaction mixture was collected in the Y-shaped tube. The bottom ionic liquid layer was continuously fed into the reaction system using another pump. The process was operated for 5.5 h, where 269 mmol of the starting material 3 was fed into the flow system. After purification of the crude product by silica gel chromatography, 103 g of the key precursor of MMP inhibitor was obtained (86% yield) (Scheme 15).

Scheme 15

One hundred gram scale production of the key intermediate of MMP inhibitor using circulatory catalyst flow system.

6 Conclusion

In this article we have demonstrated a flow benchtop factory based on a circulatory catalyst flow system integrated with ionic liquid. The circulatory flow system with efficient catalyst recycling was constructed by using a low-viscosity ionic liquid, such as [bmim]NTf₂ and [emim]NTf₂, in conjunction with an efficient self-designed microextraction/catalyst recycling system. Thus, running the circulatory catalyst flow system for Pd-catalyzed Mizoroki-Heck reaction gave 115 g of the coupling product, butyl cinnamate, after 11.5 h. After a similar circulatory flow system was run for 5.5 h, 103 g of the key intermediate of the MMP inhibitor was synthesized by Cu-free Sonogashira reaction of bromothiophenyl compound 3 with p-tolylacetylene. Coupled with the continuous-flow system, the Pd catalyst immobilized in the ionic liquid phase circulates around the system, analogous to a heterogeneous catalyst.