This invention relates to multi-phase reactions and in particular to such reactions wherein one phase is liquid. Such multi-phase reactions, i.e. gas/liquid, liquid/liquid, liquid/solid and gas/liquid/solid, are often encountered in fine chemicals manufacture. Such reactions are usually initially developed on a small scale in the laboratory and are then scaled up to give the necessary output. For convenience, the full scale manufacture, which may be several orders of magnitude greater than the laboratory scale on a volume basis, is often effected in a stirred tank. Scaling up from the laboratory scale to full scale often presents unforeseen problems, particularly heat transfer and inter-phase mixing problems.

We have realised that such problems may largely be overcome using by effecting the reaction in a reactor having a plurality of reaction zones operating in parallel with each reaction zone mimicking laboratory scale operation.

Accordingly the present invention provides a process for effecting a multi-phase reaction, wherein at least one of the phases is liquid, comprising effecting the reaction by flowing the reactants through at least one reaction zone comprising at least one channel having a minimum dimension in the range 0.25 to 20 mm.

Preferably the process is effected in a plurality of reaction zones in parallel. References to

"channels" hereinafter should be interpreted to mean "channel or channels" unless the context dictates that either the singular or plural is intended. The reaction zones are preferably in the form of a plurality of channels disposed side by side, through which the reactants flow. The channels typically have a length in the range 10 to 1000 cm and may be of any suitable cross sectional shape. Typically the channels may have a circular or regular polygonal cross section. The minimum cross section dimension is in the range 0.25 to 20 mm, preferably 0.5 - 10 mm and particularly 0.5 - 5 mm. The minimum dimension of the reaction zone will depend on the nature of the phases and the desired flow characteristics. The maximum cross section dimension in the reaction zone of the channel(s) is preferably ≤20 mm, more preferably ≤10 mm and particularly < 5 mm. In particular embodiments, for example where the reaction zone comprises a capillary or a monolith, the maximum dimension of the or each channel may be less than 2mm. When each channel has a dimension in the aforesaid range, transport distances between the different phases are relatively small and may be similar to those employed in an initial laboratory development of the process. This greatly facilitates scaling up a process from initial laboratory work and enables such a process to be scaled to the desired output by increasing the number of channels provided. Also the shear forces between the reactants and the channel walls can be employed to give good mixing between the phases and prevent

coalescence of the dispersed phase or phases. We have found that the heat transfer is greatly improved in such a reactor compared with conventional reactors, for example.

The channels are preferably un-packed, i.e. they preferably do not contain inert or catalytic packing components such as glass, metal or ceramic spheres, mesh, foam or other packing elements. The use of un-packed channels allows a greater throughput of the liquid reactants and minimises pressure drop. The channels may be straight for at least 50%, preferably at least 80%, of their length. The channels are preferably generally linear or may comprise a series of linear portions which are not coaxial. The channels may alternatively comprise one or more curved portions. It is preferred that the channels are regularly shaped. It is preferred that the channels are of generally uniform cross section along their length, preferably for at least 50% of their total length. The or each channel may have a smooth inner surface, which improves the ease with which they may be cleaned. The inner surface of the channels may include protrusions, ridges or partial obstructions to the flow along a channel to encourage the reactants to flow along a tortuous path for improving mixing, heat transfer etc. Alternatively the walls of the channel(s) may be rough.

The channels preferably are disposed such that they extend vertically or nearly vertically downwards. The channels are provided with means to feed at least one fluid reactant to a feed portion of the channel. The liquid may be introduced into the channel by any suitable means, for example by spraying. A gas reactant, if present, may be introduced into the liquid phase before the liquid enters the channel(s) or as the liquid enters the channel(s). The gas feed means are preferably arranged to mix the gas and liquid phases as they enter the channel(s). In the process of the invention it is preferred that the reactants flow downwardly through the channel(s). The feed means and reactor channel(s) are therefore preferably arranged so that the liquid phase, or at least one of the liquid phases in the case of a multi-liquid phase process, flows downwards through the channels. In a preferred embodiment of the invention the process comprises a gas phase and a liquid phase and the gas and liquid phases flow downward co- currently.

The separate vertical channels may be adjacent to one another, for example in the form of an inert monolith or honeycomb. This arrangement leads to substantially adiabatic operation and therefore, in the case of an exothermic reaction, such as hydrogenation, an increasing temperature in the direction of flow. This may adversely affect selectivity in some reactions and so in a preferred embodiment of the invention, the reactor channels are provided with a heat transfer means. This allows control of the temperature along the reaction flow path and the imposition of isothermal conditions in the reactor channel or an alternative temperature profile.

The reactor channel(s) may be provided with heating and/or cooling means together with temperature monitoring and control means, as required for a particular reaction, to regulate the temperature within the channel. The temperature in the reactor channel may vary along the length of the channel. If heat transfer is sufficiently intense, reduced solvent concentrations may be used in reactions where solvent is used to mitigate the adiabatic temperature rise associated with certain reactions.

At its simplest the reactor may take the form of one or more narrow bore tubes similar to a well- known type of shell and tube heat exchanger, in which, when more than one tube is provided, the tubes act as parallel reaction channels and a suitable heat transfer medium is passed on the shell side. Alternatively the reactor may comprise a plurality of channels arranged in a monolith configuration where some channels are used to pass a heat transfer medium to regulate the temperature of the monolith.

One particularly suitable form of reactor is a development of a heat exchanger assembled from a plurality of plates each having a plurality of slots therethrough and having passages for the flow of a heat exchange fluid between adjacent slots and extending for part of the depth of each plate. Such an arrangement is shown in GB2333350. On assembly of a stack of such plates, the through slots are aligned and form the reaction zones: to obtain good thermal control of the reaction, heat exchange fluids may be passed through the aforesaid passages. By providing that the passages of one or more of the plates communicate with the slots, it is possible to arrange that a fluid, e.g. liquid or gas, may be introduced into the reactants passing through the aligned slots at one or more locations within each reaction zone. An alternative arrangement is shown in US 6968892 which shows a particular "pin-fin" type heat exchanger which is a suitable reactor for the process of the invention.

Alternatively the reactor may comprise a structure having a plurality of channels extending therethrough, each channel opening at each end at a surface of the structure. Such structures are known as monoliths and are common in the field of catalysis. The monolith may be formed from metal and/or a ceramic material in which the channels have a minimum cross section dimension in the range 0.25 to 10 mm, more usually 0.5 to 2 mm.

The process may be operated on a continuous basis or on a batch basis. In the latter case, it is preferred that the plurality of reaction zones, e.g. channels, are connected to a reservoir containing one or more of the reactant phases: one or more of the reactants are fed from the reservoir through the reaction zones where the reaction is effected and then returned to the reservoir. The reactants may be flowed through the reactor channel(s) more than once, optionally together with products and other components of the reaction mixture which may have

been present or formed during an earlier pass through the reactor channel. To obtain the desired throughput, a plurality, e.g. 10 or more, of reaction zones are preferably employed in parallel.

The process is preferably catalytic. In a preferred embodiment of the process of the invention, the material of construction of the walls of the reaction zones has essentially no catalytic activity for the reaction. In this way a reactor can be used for a variety of different reactions as required. In this embodiment, the catalyst may be heterogeneous and dispersed or homogeneous and dissolved in a liquid phase. Preferably the catalyst is in the form of solid catalyst particles having an average particle diameter in the range from 1 micron to 50 microns. Thus the reaction may simply be between the liquid and a dispersed solid catalyst phase; alternatively the reaction may be between a gas, or a liquid, and separate liquid phase in which the catalyst is dissolved or dispersed.

The reaction is of particular utility for gas/liquid or gas/liquid/solid reactions. A particularly preferred process of the invention is the reaction of a liquid phase reactant with a gas phase reactant in the presence of a dispersed particulate solid phase catalyst which is dispersed in and flows with the liquid phase, i.e. as a slurry.

The process of the invention is suitable for a range of reactions, e.g. the hydrogenation or oxidation of a liquid in which a catalyst is dispersed or dissolved, using a suitable hydrogenating or oxidising gas, especially hydrogen, oxygen or air as appropriate. One example of such an oxidation reaction using the process of the invention is the oxidation of glycerol using a slurry bubble column reactor. One embodiment of the invention therefore comprises a process for the oxidation of glycerol , wherein a catalyst in the form of solid particles is dispersed in a solution of glycerol and the resulting slurry is caused to flow through at least one reaction zone comprising an un-packed channel having a minimum dimension in the range 0.5 to 20 mm in contact with an oxygen-containing gas stream, which is preferably air. The reaction is preferably carried out at a temperature between ambient (about 20 ⁰ C) and 100 ⁰ C, more preferably between 50 and 70 ⁰ C and at pressures between about 2 bara and 50 bara, more preferably from 2 to 20 bara. The oxygen (partial) pressure is preferably between about 0.25 bara and 3 bara. The catalyst preferably comprises a compound of gold on a solid powdered support which is preferably carbon, e.g. graphite. The amount of catalyst used in the reaction may be varied to provide the optimum amount. Typically the mole ratio of glycerol to gold is in the range from 50,000 - 100 : 1 , more preferably from 20,000 - 500 : 1 , particularly 10,000 - 500 : 1. If the catalyst is of sufficiently high activity then it may be necessary to limit the amount present in the reaction at certain pressures of oxygen to avoid over-oxidation leading to the formation of unwanted products such as oxalic acid or tartronic acid for example. The skilled person would optimise the

reaction conditions, including the amount of catalyst used, oxygen ratios etc to provide the required selectivity for the reaction. Furthermore, the amount of catalyst used should not be sufficient to block the reaction zone channels. Therefore the amount used may also be dependent upon the reactor geometry selected. The catalyst may alternatively be present on the walls of the at least one channel as a coating or impregnated thereon. In this form the reactor preferably comprises a catalytic monolith having a plurality of channels, preferably parallel, extending through a monolith block.

The invention is illustrated by reference to the accompanying drawings in which Figure l is a diagrammatic elevation of a reaction system in accordance with the invention for effecting a gas/liquid or gas/liquid/solid reaction;

Figure 2 is a plan view of one of the plates from which the reactor employed in Figure 1 is assembled;

Figure 3 is a section of the plate shown in Figure 2; Figure 4 is a diagrammatic section of the gas/liquid dispersion generator employed in the reactor system of Figure 1 ;

Figure 5 is a diagrammatic representation of the reaction apparatus used in the accompanying

Examples 1 , 2 & 5

Figure 6 is a diagrammatic representation of the reaction apparatus used in the accompanying Examples 6 & 7.

Figure 7 is a graph showing the conversion and selectivity in the process of Example 6.

Referring to the drawings, the apparatus comprises a reactor 10 having at its upper end a gas dispersion generator 11 to which a slurry of a particulate catalyst in a reactant liquid is supplied via line 12 from a holding tank 13 and gas supplied via line 14. Beneath the gas/liquid dispersion generator 11 is the reactor assembled from a plurality of plates 15 each having a plurality of through slots 16 having a width corresponding to the reaction zone minimum dimension, i.e. in the range 0.5 to 20 mm. Between adjacent slots 16 are heat exchange fluid channels 16 extending for part of the thickness of the plates. Each plate is provided with heat exchange fluid inlet and outlet ports 18, 19 and these are coupled together, by means not shown, to heat exchange fluid inlet and outlet conduits 20, 21. The plates may be welded or otherwise fastened together with their through slots aligned.

Beneath the reactor plates is a gas disengagement and reactant off-take zone 22 from wherein the gas is disengaged from the slurry which is returned to a holding tank 13 via line 23.

The gas dispersion generator 14 comprises a chamber 24 to which the slurry of catalyst in the reactant liquid is supplied via line 12 and the gas supplied via line 14. A constriction 25 is provided at the lower end of the chamber 24. Constriction 25 is dimensioned such that the liquid

flow velocity through the constriction is sufficient to entrain the gas as fine bubbles. Typically the liquid flow rate is of the order of 2 m/s or more.

The slurry, with the gas entrained therein as fine bubbles, passes down through the slots 16 in the reactor plates and the desired reaction effected. On leaving the slots, the slurry is discharged into the gas disengagement zone 22 which is dimensioned such that the downward velocity of the slurry is less than the gas bubble rise velocity. The disengaged gas bubbles rise up through the slots 16, countercurrent to the dispersion flowing downwards therethrough, into the region just beneath the constriction 25.

It has been found that it is often not necessary to supply gas continuously to the gas dispersion generator 11 via line 14: thus the liquid flowing through the constriction can cause the gas bubbles that have risen up through the slots 16 to re-disperse as the desired fine bubbles. Hence the gas need only be supplied via line 14 as necessary to compensate for that consumed by the reaction.

If it is desired to operate the process continuously, a slurry off-take line may be taken from holding tank 13 or line 23 and fresh slurry supplied, e.g. to line 12. As is well known in the art, the catalyst will normally be separated from the product slurry and recycled. The heat transfer fluid supplied via line 20 enables the reaction temperature to be controlled.

The reaction is particularly useful in the hydrogenation of organic compounds such as esters, nitriles, aldehydes, ketones, unsaturated hydrocarbons. In particular the reactor has been found to be advantageous in the hydrogenation of the aromatic unsaturation in aromatic compounds, e.g. in the hydrogenation of resorcinol to 1 ,3-cyclohexanedione. 1 ,3-cyclohexanedione is used as an intermediate in the synthesis of a number of important industrial materials which are used for applications including cosmetics, agrochemicals, polymer additives and pharmaceuticals.

Example 1 The catalytic hydrogenation of resorcinol to 1 ,3-cyclohexanedione in a single channel reactor was studied The reaction occurs in a basic aqueous solution and proceeds via three steps. The first step is the acid-base reaction between resorcinol and NaOH to protect the two C=C bonds of the aryl group during the hydrogenation reaction. The second step is the catalytic hydrogenation of the sodium salt of resorcinol. Finally, the product obtained in the second step is reacted with HCI and the final product is cyclohexanedione that is obtained after the rearrangement of the double bonds.

The reactor used for these studies is shown schematically in Figure 5. The liquid in the receiver 51 , contained the catalyst particles in suspension. The liquid suspension was circulated through the reactor 52 by a pump 58. The reactor was a tube of 500 mm length, 6.35 mm outside diameter and internal diameter 3.86 mm, placed in a vertical orientation. Consequently, the total volume of the tube-reactor was approx 6 cm ³ .

A thermocouple was placed at the top of the reactor and connected with a PID control unit. This controller was linked with the heating tape 54 placed on tube 53 before the reactor to pre-heat the liquid feed to the reaction temperature. A second thermocouple was used to measure the outlet reactor temperature and it was connected to a second PID controller. This second controller was linked with a second heating tape 55, which was used to heat the reactor itself. Both controllers were set to the same temperature during the reaction to ensure accurate control and monitoring was maintained.

The hydrogen gas was injected into the flowing liquid at the top of the reactor through a capillary 1.59 mm O. D. After passing through the reactor the mixture of solid, liquid and gas was decompressed and then the liquid and gas were re-circulated via the receiver 51. The unreacted hydrogen was separated and vented. Depressurisation and degassing was carried out as close to the reactor as possible to minimise end effects. The backpressure regulator 56 had to deal with three phases and consequently there were minor oscillations in the reactor pressure.

Therefore, the reaction pressures quoted are the average pressure over the time period of the experiment.

Sodium hydroxide (11.75g) was dissolved in distilled water (53ml). Resorcinol (28.4 g) was added to the solution and stirred for ca.15 min in the cooling bath. The solution (75 ml) was introduced to receiver 51 with the catalyst and the reaction was started. The catalyst used was 5 wt% Rh/γ-AI ₂ O ₃ supplied by Johnson Matthey, with the following particle size distribution: < 25 microns=15.9 % and <45 microns = 52.5 % and BET surface area of 151 m ² /g. Samples (1 ml each) were taken periodically from sampling point 57 with a syringe, neutralised with 37 % HCI (0.45ml), and then mixed with acetone (2ml). The mixture of resorcinol and cyclohexanedione was extracted with the acetone and analysed by gas chromatography (Varian 3800) equipped with a DB-Wax capillary column and flame ionisation detector.

The reaction time was calculated from the fact that only a volume of 3 ml of liquid is present at any time in the reactor because the ratio gas : liquid used was 1 : 1 so that inside the reactor there is 3 ml of liquid and 3 ml of gas. The reaction time is obtained by multiplying the reaction time by 3 (volume of liquid in the reactor) and divided by 75 (which is the total volume of the solution). The remaining liquid is in the circuit at lower temperature, without stirring and either at

atmospheric pressure, or in the absence of hydrogen. Consequently, we assume no reaction takes place outside the reactor.

Conversion of resorcinol was linear over time up to 99% conversion. At a temperature of 100 ⁰ C and a pressure of 8.6 bar, the reaction rate was 183 mmol/g _cat /hr.

Example 2

Example 1 was repeated using a catalyst consisting of 5% Pd on carbon. The reaction rate was

353 mmol/g _cat /hr.

Example 3 (comparative)

For comparative purposes, a resorcinol hydrogenation was carried out in a Buchi 100 ml stirred autoclave. The 5%Rh/AI ₂ O ₃ catalyst charge in the autoclave was 1g in 75 ml of neutralised resorcinol solution. The stirrer speed was 1400 rpm; temperature was maintained at 100 ⁰ C and the operating pressure was 8.6 bar. Product analysis was performed using the same procedure detailed in Example 1. The reaction rate was 52 mmol/g _ca t/hr.

Example 4 (comparative)

Comparative Example 3 was repeated using the 5% Pd on carbon catalyst. The reaction rate was the same as found in Example 3, indicating that under these reaction conditions, the reaction rate is transport limited.

Example 5

The reaction described in Example 1 was repeated using reactors of different dimensions as shown in Table 1. All reaction rates are reported at a temperature of 100 ⁰ C, using 1 g of catalyst (5% RIVAI ₂ O ₃ ) using a liquid flow rate of 20 ml/min and a gas flow rate of 200 mlN/min.

The results show that although the total flow area of the two parallel reactor tubes is nearly equal to the flow area of the larger, single tube, the reaction rate is significantly higher under similar reaction conditions. Table 1

Example 6 Oxidation of glycerol

The oxidation of glycerol was carried out using air and a 1 % gold on carbon catalyst made by refluxing an aqueous suspension of graphite with chloroauric acid (2.38g in 70 ml water) and then reducing the gold compound with formaldehyde. The catalyst was washed free of chloride ions and dried.

The reactor vessel was a 25 mm diameter by 300 mm tube containing a cordierite monolith having 62 channels per cm ² . Glycerol (0.6 mol I ^"1 ) in sodium hydroxide was used as the liquid phase (glycerol : sodium hydroxide 1 :1) at a reaction temperature of 60 ⁰ C and a liquid/gas (air) ratio 1 :1. The reactor was arranged in a loop configuration for circulating the reactants as described in Enache, D.I., Landon, P., Lok, CM., Pollington, S.D., Stitt, E. H. "Direct Comparison of a Trickle Bed and a Monolith for Hydrogenation of Pyrolysis Gasoline" Industrial & Engineering Chemistry Research 44, 9431-9439 (2005) and reproduced as Fig 6. The reactor (61) was designed for co-current downflow operation, liquid recirculation, and continuous gas feed. The liquid reactant (300 cm ³ ) was introduced into a receiver vessel (62). The liquid reactant was pumped around a closed loop by the use of a metering pump 63 (0-80 L h ^"1 ; discharge pressure 40 barg) or pumped to the reactor via a non-return valve 64. The liquid was introduced to the reactor using a spray nozzle 65 to ensure a uniform flow distribution over the monolith channels. Air was introduced to the reactor using a Brooks 5850S mass flow controller (0-2 L min ^"1 ) 66 where it could be mixed with the liquid spray before the catalyst bed. The lines delivering liquid and gas reactants could be heated if appropriate. A valve (67) was incorporated to allow depressurisation of the reactor exit flow for gas/liquid separation and sampling at atmospheric pressure. The liquid was recycled to the receiver vessel and the air vented. The temperature was monitored by the use of thermocouples (68).

The catalyst flowed through the reactor dispersed as a powder in a gas/liquid flow, as a meso- structured co-current downflow slurry bubble column. The analysis of products was performed by HPLC. The results at different catalyst ratios and oxygen pressures are shown in Table 2. Rate data are presented in terms of both glycerol conversion and oxygen usage (calculated using the stoichiometry to obtain the observed products), based on both reactor volume and mass of gold.

Glycerol + 0.5 O ₂ >■ 1 ,3-dihydroxypropan-2-one

Glycerol + O ₂ > Glyceric acid + H ₂ O

Glycerol + 3 O ₂ > Oxalic acid + 3H ₂ O + CO ₂

Fig 7a shows the conversion and selectivity as the reaction progressed at a glycerol/metal mol ratio of 4730:1 , temperature 60 ⁰ C, pθ ₂ of 2.8 bara, liquid superficial velocity 5.6 mm s ^"1 and gas/liquid ratio of 1 :1. Total conversion of glycerol was observed after 70 minutes. Fig 7b shows the same data at pθ ₂ of 0.5 bara.

Example 7 (comparative)

A cordierite monolith having 62 channels per cm ² was soaked in a mixture based on a phenol formaldehyde resin diluted by the addition of a suitable solvent to reduce the viscosity. After drying at 100 ⁰ C, the monolith was fired under argon at 500 ⁰ C to yield a cordierite monolith coated with a layer of carbon. The carbon was present at a level of 8% on the total weight of the monolith. Gold was added as described for the carbon powder catalyst to achieve 1 wt% Au/C monolith catalyst. The prepared catalysed monolith was then loaded into the reactor tube and used for the oxidation of glycerol as described in Example 6 with the omission of the dispersion of a powdered catalyst in the liquid/gas flow. The results are shown in Table 2.

Significantly lower rates are observed for the monolithic catalyst of Example 7 compared with the meso-structured slurry reaction of Example 6. Hydrodynamic conditions for the monolith and structured reactor are ostensibly the same and, could reasonably be expected to result in the same gas-liquid mass transfer coefficient. The monolith runs would appear therefore to be kinetically limited, inferring that the monolith catalyst is inherently less active than the slurry catalyst. The monolith shows very high selectivity to glyceric acid whereas the structured reaction of the invention yields approximately equal quantities of dihydroxyacetone and glyceric acid. The formation of secondary oxidation products as seen in the structured reactor has also been observed at the highest partial oxygen values in a similar reaction carried out as a batch in an autoclave (Carrettin, S., McMorn, P., Johnston, P., Griffith, K., Hutchings, GJ. , Chemical Communications, (2002) 696-697). This is indicative of higher oxygen availability in the structured reactor used in Example 6 compared to the autoclave and monolith reactor of Example 7 when all are operated at equivalent partial oxygen values. It is apparent therefore that the structured reactor facilitates faster oxygen transport.

Table 2: Oxidation of glycerol using 1 %Au/C catalysts at 60 ⁰ C

Notes: GA: glyceric acid, DHA: dihydroxy acetone, TA: tartronic acid, OA: oxalic acid. * Glycerol conversion of approx 27%; "Glycerol conversion of approx 25%