Catalytic hydrogenations are among the standard methods in organic chemistry. These methods are also employed on an industrial scale for the synthesis of optionally enantiomerically enriched products. These can be used-as in the case of L-dopa-as important bioactive substances.

For the hydrogenation of dissolved, unsaturated, organic compounds in a homogeneous or heterogeneous manner, the hydrogen needed has to be fed into the reaction in some way. Various processes are available for this purpose to. the person skilled in the art.

A distinction is made between on the one hand the use of reducing agents that contain the hydrogen as a reduction equivalent, and in which a by-product is formed from the reducing agent. In this case, the separation of the by- product and of the hydrogen donor, which has to be used in excess, represents a disadvantage. In the case of alcohols, e. g. 2-propanol, or of formate, the selection of the solvent is also limited. This is due to the fact that, for a favourable shift of the equilibrium, the hydrogen donors have to be used in a greater than stoichiometric quantity and, in the case of alcohols, usually even as solvents. The limiting solubility of salt-like substances in hydrophobic solvents is a disadvantage of the use of formate.

The above disadvantages are removed if hydrogen is used directly. It can be used in a large excess, since it can readily be removed as a substance that is gaseous under normal conditions. On the other hand, hydrogenation with

molecular hydrogen is atom economical in the sense of a complete transfer to the target product without the formation of other by-products.

However, the use of molecular hydrogen has various disadvantages: For hydrogenation with gaseous molecular hydrogen, it is unavoidable that the hydrogen has to dissolve in the liquid medium being gassed through a gas/liquid phase interface so that the reaction takes place at a measurable rate. It can happen that this transition rate is limiting for the entire reaction. To counter this, various methods of improving the phase transition are commonly used. This can be done by increasing the hydrogen gradient as a propelling force for the phase transition as a result of applying increased pressure in the reaction autoclave or by increasing the limiting interface by means of suitable stirrers and agitators. By stirring the liquid medium being gassed, the interface is increased, which increases the rate of dissolution of hydrogen in the liquid medium and thus contributes to the more rapid supply of hydrogen to the liquid medium, which in turn can positively affect the rate of hydrogenation.

Common to these approaches is the fact that they relate to a selected geometry, and when the scale is increased, these geometries must necessarily change. The use of higher pressures in small volumes, however, also creates safety aspects that cannot be ignored. In the further handling of the liquid phase, it must also be borne in mind that this is under the applied pressure.

Owing to the high stirring speeds and the maximisation of the phase surface, those reactors containing more than a 30% gas volume are usually used for hydrogenation. This volume is not advantageous for achieving high space-time yields. Furthermore, the technical production of gas

pressure vessels is associated with high costs, since the high compressibility of gas compared with that of liquids can lead to explosive expansion. Particular attention has to be paid to the gas space of a pressure vessel.

In addition, the high stirrer speeds that are set lead to a high energy input, which has to be compensated by increased cooling. At the same time, the energy required to supply a reaction in a liquid volume increases disproportionately with the volume and reaches discrete limits, depending on the reactor and stirrer geometry, which cannot be overcome.

It is known from DE 19528871 to supply cell cultures in an aqueous medium with oxygen. For this purpose, the oxygen is metered into these media through an oxygen-permeable membrane.

In continuation of the efforts to apply gaseous hydrogen to a liquid medium, the object of the present invention was to make available a process and a suitable device, allowing an optimised supply of the liquid medium with hydrogen. The process and the device should manage without the addition of other hydrogen-producing substances and yet should be suitable for use on an industrial scale, in terms of apparatus, in an economically and ecologically advantageous way.

The object is achieved in accordance with the claims. Claim 1 relates to a process for the gassing of liquid media with hydrogen. Claims 2 to 6 are preferred embodiments of the process according to the invention. Claim 7 relates to a special device for gassing with hydrogen, whereas claims 8 to 10 reflect preferred versions of the device.

By metering molecular hydrogen into the liquid medium through a hydrogen-permeable membrane in a process for gassing liquid media with molecular hydrogen, the object set is achieved in a very simple, but no less advantageous,

manner. In this process, the gaseous hydrogen diffuses from the gas space through the membrane to the side of the membrane in contact with the liquid medium, where the hydrogen is directly dissolved. As a result, the phenomena of disadvantageous coalescence described in reaction autoclaves are avoided from the outset.

The process according to the invention is preferably used for the hydrogenation of unsaturated organic compounds in the presence of a hydrogenation catalyst in the liquid medium in such a way that the hydrogen is metered into the liquid medium through a hydrogen-permeable membrane.

All catalysts that can be considered for this purpose by the person skilled in the art, whether chiral or non- chiral, homogeneously soluble or heterogeneous or molecular-weight-enlarged with polymers, can be used as hydrogenation catalysts. In particular the catalysts known from Ojima, Catalytic Asymmetric Synthesis, Wiley-VCH, 1993, pp. 1 to 33; DE 10002976; DE 10002975; DE 10002973; DE 10100971; J. M. Brown, in Comprehensive asymmetric catalysis, Vol. 1 (Eds.: E. N. Jacobson, A. Pfaltz, H.

Yamamoto), Springer Verlag, Berlin, Heidelberg, New York, 1999, p. 121; M. J. Burk, M. F. Gross, T. G. P. Harper, C.

S. Kalberg, J. R. Lee, J. P. Martinez, PureApplChem 1996, 68, 37; H. B. Kagan, in Comprehensive Organometallic Chemistry, Vol. 8 (Eds.: G. Wilkinson, F. G. A. Stone, E.

W. Abel), Pergamon, London, 1982, p. 463; W. S. Knowles, AccChemRes 1983, 16, 106; R. Noyori, Asymmetric Catalysis in Organic Synthesis, John Wiley & Sons, Tokyo, 1994 ; G. X.

Zhu, A. L. Casalnuovo, X. M. Zhang, Journal of Organic Chemistry 1998, 63, 8100 can be mentioned as suitable catalysts. The disclosures of these documents are considered to be included herein.

The liquid medium to be used depends on the solubilities of the products, educts or catalysts that are present during the reaction or the possibly positive effect of some

solvents on the hydrogenation itself, and can otherwise be freely selected by the person skilled in the art.

As unsaturated organic compounds, all those with which the person skilled in the art is familiar for this purpose can also be used for the process according to the invention.

Individually, organic compounds having a C=C-, C=C-, C=O-, C=N-, C-N-, C-P-, C=P-, N=N-, N=O-, N=0-, N=P-or NP bond can be used as educts for the present invention.

Particular attention should be paid in the present invention to the selection of a suitable membrane material.

In principle, the person skilled in the art is free in this selection, but the following requirement should be observed. On the one hand, hydrogen must be able to diffuse through the membrane sufficiently well without damaging the membrane material by reaction.

Preferably, a membrane is selected which consists of a ceramic material (as mentioned in M. Schmidt et al. Chez : Ing. Techn. 1999, 71, 199-206 and M. Cheryan Ultrafiltration and microfiltration, Technomic Publ. & Co., Lancaster, 1998). Ceramic materials have the advantage that they are very resistant towards reaction media and cope well with the requirements of a robust industrial process.

Alternatively, organic polymer materials can also be used as membrane materials. These have the advantage of being pliable and thus capable of being flexibly adapted to the reactor geometry. In addition, extremely small pore sizes can be achieved with these materials, which lead to hydrogen being given off in relatively small portions at the side of the membrane facing the liquid medium, which contributes to a more rapid dissolution of the hydrogen owing to the surface/volume ratio, which is thereby increased. The use of membranes made of perfluoro-and fluoropolymers (PTFE), polyimides or silicones is advantageous in this context. These are mentioned e. g. in

DE 19528871. These disclosures are considered to be included herein.

The present process is characterised in that, as far as possible, no hydrogen bubbles are produced in the liquid medium. For this reason, the hydrogen pressure in the gas space should be adjusted such that work is performed below the bubble point (formation of bubbles visible to the eye) of the relevant membrane material. A pressure difference of 100 and 0.1 bar, preferably between 50 and 1 bar, particularly preferably between 25 and 1 bar, is preferably established between the gas phase and the liquid medium.

The present process can be carried out in a so-called batch process and also continuously. By means of so-called molecular-weight-enlarged catalysts (see above for lit.) the process can be carried out continuously in a membrane reactor. In the context of the invention, the repetitive batch method (batch UF) is also meant by continuous reaction. Here, the reactor is emptied apart from the polymer and the liquid medium is pressed through the reactor membrane. The catalyst remains in the reactor and reacts with the newly added substrate and the catalysis cycle starts from the beginning.

The continuous method can be carried out in the so-called cross-flow filtration mode (Fig. 2) or as dead-end filtration (Fig. 1).

In dead-end operation, catalyst and liquid medium are placed in the reactor and the dissolved substrate is then metered in, with the hydrogen feed according to the invention being guaranteed at the same time. The substrate is reduced by means of the catalyst, optionally enantio-or regioselectively, and then discharged from the membrane reactor together with the liquid medium through the filtration membrane.

In the cross-flow method, the liquid medium, containing substrate, product and catalyst together with hydrogen fed

in according to the invention, are passed over a membrane at which a pressure difference exists.

For both cases, the metering in of the dissolved substrate takes place at a rate such that the permeated liquid medium predominantly contains optionally enantio-or regio- selectively hydrogenated product. Both process variants are described in the prior art (Engineering Processes for Bioseparations, Ed.: L. R. Weatherley, Heinemann, 1994, 135-165).

In the context of the invention, membrane reactor means any reaction vessel in which the molecular-weight-enlarged catalyst is enclosed in a reactor, while low molecular- weight substances can be fed into or discharged from the reactor. The reactor membrane can be integrated directly into the reaction chamber or can be incorporated externally in a separate filtration module, in which the reaction solution flows through the filtration module continuously or intermittently and the retentate is fed back into the reactor. Suitable embodiments are described e. g. in W098/22415 and in Wandrey et al. in Jahrbuch 1998, Verfahrenstechnik und Chemieingenieurwesen, VDI p. 151ff. ; Wandrey et al. in Applied Homogeneous Catalysis with Organometallic Compounds, Vol. 2, VCH 1996, p. 832 ff. ; Kragl et al. , Angew. Chem. 1996,6, 684f.; U. Kragl et al., Trends Biotechnology 2001,19, 442 ff; S. Laue et al., Adv.

Cat. Synth. 2001,343, 711 ff.

In its next form, the invention concerns a device for the gassing of liquids with molecular hydrogen, exhibiting an open or closed hydrogenation vessel (21), which is filled with an optionally stirred liquid medium (22), which is separated by a hydrogen-permeable membrane (23) from a gas space (24) containing hydrogen, and in which

- hydrogen is metered into the liquid phase through the hydrogen-permeable membrane (Fig. 3).

By means of such a device, which can be designed in a similar way to that in US 5,601, 757, the gassing according to the invention is possible with the advantages described in the process aspect.

The embodiment in which the device according to the invention possesses a hydrogen consumption measuring device (28), by means of which the hydrogen pressure in the gas space can be kept constant, is especially advantageous.

In respect of the selection of the membrane material, reference is made to the statements relating to the process. Here too, membranes made of ceramics or an organic polymer material are advantageously used.

A membrane made of perfluoro-and fluoropolymers (PTFE), polyimides or silicones is quite advantageously suitable for this purpose.

With regard to the batch or continuous method, the remarks made above for the process apply correspondingly here.

With the device according to the invention, it is possible, both in batch and in continuous operation, to achieve a supply of the reaction in the liquid medium with molecular- weight-enlarged catalysts, which is advantageously suitable for avoiding transfer hydrogenation conditions and the associated disadvantages and yet the difficulties described for hydrogenation with gaseous hydrogen do not have to be accepted.

It should be noted that the hydrogen-permeable membrane in the process according to the invention and the device according to the invention can be designed in accordance with ideas familiar to the person skilled in the art. In particular, it can be a closed tube or a closed hose.

Similarly, however, modules can also be used, as suggested

in DE 19528871. These are open on both sides with a hydrogen feed and a hydrogen discharge, between which the hydrogen-permeable membrane is in contact with the liquid medium.

The documents cited are considered to be included in the description.

Description of the figures: Fig. 1 shows a membrane reactor with dead-end filtration.

The substrate 1 is transferred via a pump 2 into the reactor chamber 3, which has a membrane 5. In the stirrer- driven reactor chamber, in addition to the solvent, are the catalyst 4, the product 6 and unreacted substrate 1.

Through the membrane 5, mainly low molecular-weight 6 is filtered off. The hydrogen is fed into the reactor via the pipe 7.

Fig. 2 shows a membrane reactor with cross-flow filtration.

Here, the substrate 7 is transferred via the pump 8 into the stirred reactor chamber, in which solvent, catalyst 9 and product 14 are also present. Via the pump 16 a flow of liquid medium is established, which leads through an optionally present heat exchanger 12 into the cross-flow filtration cell 15. Here, the low molecular-weight product 14 is separated off by the membrane 13. High molecular- weight catalyst 9 is then passed with the solvent flow, optionally via a heat exchanger 12 again, optionally via the valve 11, back into the reactor 10. The hydrogen is fed into the reactor via the pipe 17.

Fig. 3: the liquid medium 22 metered into the reaction vessel by the pump 27 contains substrate and catalyst. The liquid medium 22 is optionally set in motion by the stirrer 26, while hydrogen is fed into it by the hydrogen feed 25, which opens out into the gas space 24, through the hydrogen-permeable membrane 23. The hydrogen pressure is kept as constant as possible by means of a measuring device 28, which is connected to the pump 29. Through the outlet 28, after hydrogenation is complete, the liquid medium 22 can be drawn off with catalyst and product.

Fig. 4: Conversion-time curve of the volume-gassed reaction Circles: conversion according to quantity of hydrogen, squares conversion according to quantity of substance (see also example 1) Y axis: conversion (no unit) X axis: time in hours Fig. 5: Conversion-time curve Squares: conversion according to quantity of substance (see also example 2) Y axis: conversion (no unit) X axis: time in hours Fig. 6: Membrane reactor conversion-time curve (example 3) Squares: conversion according to quantity of substance (capillary electrophoresis) Y axis: conversion (no unit) X axis: time in hours Examples: Example 1

12 mM of methyl-N-acetylcinnamate (1), 1.2 mM of BPPM (3) and 1.1 mM of rhodium biscyclooctadiene tetra- flourosulfonate in 250 ml of 2-propanol are initially placed in a 500 ml glass vessel with a Teflon-coated magnetic stirring rod under an argon atmosphere. The solution is degassed. A polytetraflouroethylene (PTFE) hose with a length of 5 m (internal diameter 0.8 mm; external diameter 1.6 mm) is introduced into the solution in such a way that it is completely covered. Hydrogen is introduced into the hose at 0.8 MPa and the mass flow of the for maintaining the pressure is monitored and recorded. Samples are taken from the unpressurised vessel for reaction monitoring purposes and are investigated by gas chromatography with respect to yield and selectivity as described in H. Frank, G. J. Nicholson, E. Bayer, Journal of Chromatography 1978, 146, 197 or T. Dwars, J. Haberland, I. Grassert, G. Oehme, U. Kragl, Journal of Molecular Catalysis A: Chemical 2001, 168, 81. The curve of the conversions according to hydrogen transition and gas chromatography is shown in Fig. 4. It is shown that the conversion is limited by the hydrogen transition and the hydrogen feed takes place through the membrane.

Quantitative conversion is achieved with an enantiomer ratio of er=19 (enantiomeric excess ee=90%).

Example 2:

Degassed methanol with 250 mM of N-acetylcinnamic acid (1 for Me=H) and 0.6 mM of pyrphos-rhodium cyclooctadiene tetraflouroborate (4-RhCODBF4) is introduced into a closed recirculation device with a membrane unit (10 m PTFE hose, internal diameter 1.0 mm; external diameter 1.6 mm wound on to a solid block) and circulating pump. The total volume of the liquid phase is 0.15 1. A pressure of 8 bar is applied to the liquid phase and initially a pressure of 11 bar H2 to the inside of the hose. After 2.5 h (vertical line in Fig. 5) the internal pressure on the gas side is raised to 12 bar.

Quantitative conversion is achieved, with an enantiomer ratio er=32 (ee=94%). For this example too, the rate of reaction is limited by the hydrogen uptake.

Capillary electrophoresis conditions: U=30 kV; pH=10.2 ; c (phosphate buffer) = 125 mM; c (DIME-beta-cyclodextrin) = 25 mM, T = 16°C, Beckman Pace/MDQ

Example 3: Methanol and 250 mM of N-acetylcinnamic acid (1 for Me=H) and 60 mg of the polymer-enlarged catalyst (0.003082 mmol/polymer, 10% functionalised, produced in accordance with the examples of DE10002975) are introduced into the reactor in the apparatus from example 2, supplemented by a device for nanofiltration with a ceramic membrane (prototype from the Hermstorfer Institut). (Conditions as in example 2). 25 ml/h of the substrate solution are then metered in with a metering pump (corresponding to a residence time of 10 h; V=0.25 1). The conversion and the selectivity are determined by capillary electrophoresis.