A Microfluidic Device for Conducting Gas-Liquid-Solid Hydrogenation Reactions

See allHide authors and affiliations

Science  28 May 2004:
Vol. 304, Issue 5675, pp. 1305-1308
DOI: 10.1126/science.1096956


We have developed an efficient system for triphase reactions using a microchannel reactor. Using this system, we conducted hydrogenation reactions that proceeded smoothly to afford the desired products quantitatively within 2 minutes for a variety of substrates. The system could also be applied to deprotection reactions. We could achieve an effective interaction between hydrogen, substrates, and a palladium catalyst using extremely large interfacial areas and the short path required for molecular diffusion in the very narrow channel space. This concept could be extended to other multiphase reactions that use gas-phase reagents such as oxygen and carbon dioxide.

Multiphase catalytic reactions play important roles not only in the research laboratory but also in the chemical and pharmaceutical industries (1). They are classified according to the phases involved, such as gas-liquid, gas-liquid-liquid, or gas-liquid-solid reactions. Although numerous multiphase catalytic reactions are known and many are used in industry, these reactions are still difficult to conduct when compared to homogeneous reactions, because the efficiency of interaction and mass transfer between different phases is extremely low, and thus in most cases the reaction rates are slow. In general, to accelerate multiphase catalytic reactions, some treatment producing high interfacial area between the two or three reacting phases, such as vigorous stirring or additional equipment, is needed, and the development of more effective, simple devices that can produce such a high interfacial area between different phases is a much-sought-after goal.

To achieve efficient multiphase catalytic reactions, we focused on a new device, which has a very small channel (nanometer- to micrometer-sized in width and depth and centimeter- to meter-sized in length) in a glass plate (210). A similar device, the so-called “microchannel reactor,” is used mainly in the field of analytical chemistry (11). The device has a very large specific interfacial area per unit of volume. In concrete figures, this area rises to 10,000 ∼ 50,000 m2/m3, as opposed to only 100 m2/m3 for conventional reactors used in chemical processes (12). Our idea is to immobilize a solid catalyst on the wall of the microchannel and then to flow liquid and gas materials into the channel. Provided that the flow is well controlled, it should be possible to pass the gas through the center of the channel and the liquid along the inner surface of the channel at a particular gas pressure (Fig. 1) (13, 14). In this system, efficient gas-liquid-solid reactions might occur, because effective interaction of the three phases is expected because of the extremely large interfacial areas and the short path required for molecular diffusion in the very narrow channel space.

Fig. 1.

Ideal device for multiphase reactions.

To put this idea into practice, we chose hydrogenation catalyzed by palladium (Pd) as a model gas-liquid-solid reaction. Pd catalysts are often used in organic synthesis, and versatile transformations using Pd catalysts have been developed (1518). Hydrogenation using Pd catalysts is one of the most important and widely used reactions in synthetic organic chemistry (19, 20).

Our first task was to immobilize an active Pd catalyst on the wall of the glass channel. Although there have been several reports concerning the immobilization of metals on a glass wall (21, 22), lowered reactivity and leaching of the metals during reactions are sometimes serious problems. We have developed a new method for this immobilization.

We selected a microchannel reactor having a channel 200 μm in width, 100 μm in depth, and 45 cm in length (Fig. 2). First, amine groups were introduced onto the surface of the glass channel to form 3. Microencapsulated (MC) Pd (2), prepared from Pd(PPh3)4 and copolymer 1 in dichloromethane-t-amyl alcohol (2330), was used as the Pd source. A colloidal solution of the MC Pd 2 was passed through the microchannel to form 4, which was heated at 150°C for 5 hours. During the heating, cross-linking of the polymer occurred, and the desired Pd-immobilized microchannel was successfully prepared (Fig. 3A).

Fig. 2.

Experimental hydrogenation system using a microreactor and immobilization of the Pd catalyst.

Fig. 3.

The appearance of the reaction system in the microchannel. (A) View of the microchannel without catalyst (left) and with the Pd catalyst (right). (B) Triphase reaction system.

We then conducted hydrogenation of benzalacetone using this Pd-immobilized microchannel (Fig. 2). A teflon tube (200 μm × 10 cm) was connected to the end of the microchannel reactor. The reaction was conducted under continuous flow conditions at ambient temperature by introducing a tetrahydrofuran (THF) solution of a substrate (0.1 M) through one inlet and introducing H2 through the other inlet via a mass-flow controller. Yields were determined by collecting a measured volume of the product from the outlet of the microchannel reactor and quantifying it by 1H nuclear magnetic resonance (NMR). When the flow rate of H2 was relatively slow, alternate slugs of the liquid and gas were observed, and the yield was insufficient [63% yield with a flow rate of the liquid substrate of 0.8 ml/hour, and a flow rate of H2 of 0.15 ml/min]. We then tried the reaction with an increased flow rate of hydrogen (1.0 ml/min) and a decreased flow rate of the substrate (0.1 ml/hour). In this case, as we expected, the desired flow system shown in Fig. 1 was realized; the liquid flowed close to the channel surface where the catalyst existed, and the gas flowed through the center (Fig. 3B). The reaction proceeded smoothly to afford the product quantitatively. The mean residence time (which equals the reaction time) of the starting materials was only 2 min. In this case, the interfacial area between the liquid, the gas, and the solid catalyst was dramatically increased, providing a very efficient environment for hydrogenation.

Next, we examined hydrogenation with this Pd-immobilized microchannel reactor (Table 1) using several other substrates. The flow rate of the liquid phase and of H2 was kept constant for each case. Double bonds, including tri-substituted olefins, as well as triple bonds were reduced efficiently, and deprotection of a benzyl ether and of a carbamate group also proceeded smoothly. In all cases, the reductions went to completion in 2 min to afford the desired products quantitatively. Furthermore, chemoselective reduction of 8 was successfully conducted, and the triple-bond moiety was reduced without deprotection of the benzyl ether moiety.

Table 1.

Hydrogenation using the Pd-immobilized microchannel. Unless otherwise noted, the reaction conditions were 1 ml/min H2 in THF at room temperature for 2 min, and yields were quantitative (side products are noted) as determined by 1H NMR.

In addition to the high reactivity it provides, this reactor has several practical advantages. In nearly all cases, products were isolated almost pure simply by passing the starting material through the reactor and then removing the solvent. No Pd was detected in the product solutions in most cases [checked by inductively coupled plasma (ICP) analysis], which meant that no leaching of the Pd and no Pd contamination of the products occurred. The microchannel reactors were reused several times without loss of activity. Although one reactor produces only a small amount of the product, it would be easy to scale up reactions by using a number of chips in parallel (12).

The high reactivity and efficiency of this microchannel reactor have been proved by several control experiments (31). The amount of Pd catalyst immobilized in the microchannel was determined by ICP analysis, and the time course for the amount of substrate present in the microchannel reactor during the reaction was calculated on the basis of the flow rate and the actual residence time. In addition, the pressure of H2 in the microchannel reactor system was measured and found to be almost 1 atm. On the basis of these data, control experiments on reduction of olefin (6) and deprotection of benzyl ether (7) were conducted in batch systems. The yields were 11% and <1%, respectively, which are much lower than those obtained using the microchannel reactor. We also calculated a space-time yield (32). When referred to the reaction channel volume and the amount of the catalyst, the space-time yield was 140,000 times higher than those produced by ordinary laboratory flasks.

Our approach should lead to high efficiency in other multiphase reactions; for example, those using O2, CO, CO2, and other gas-phase molecules or using other metals as catalysts. Because it is easy to scale up the reaction by using a number of chips in parallel with shared flow, this system can provide the desired products easily and quickly in the required volumes, as well as in pure form at the point of use without the need for any treatment such as isolation or purification. Such an approach lessens reagent consumption and the time and space needed for synthesis.

Supporting Online Material

Materials and Methods


References and Notes

Stay Connected to Science

Navigate This Article