Such a method is described in WO-A-93/23164, which patent publication is concerned with separating catalyst particles in a screen bottomed reactor from a mixture which contains catalyst particles and another solid substance, by giving the catalyst particles such a size relative to the other solid substance that there is a clear difference in the range of particle sizes between the catalyst and the other solid substance present in the suspension. In the said patent publication no details are given of the technological execution of the experiments, which in any case were only carried out on a small scale (litre scale).

It has been found, however, that when, as seems logical, the agitator and the direction of rotation of the agitator are chosen in such a way that the suspension is pumped downwards in the centre of the agitator, only a limited flow through the screen bottom can be achieved with good separation between the first solid substance and the remainder of the suspension still taking place; and that particularly

vigorous stirring is necessary, as a result of which the useful life of the screen is particularly adversely affected and the solid substances, for example the catalyst in the known method, can also be damaged.

Furthermore, these adverse effects were found to be much more severe on a large scale (1-10 m3 scale) than on a small scale, to such an extent even that such a process was found to be impractical on a large scale.

In addition, it was found that poorer results were also obtained when an agitator producing an essentially radial flow was used, for example a turbine agitator.

The invention has the object of providing a method in which a good separation between the first solid substance and the other solid substance is achieved at a high flow rate, without the screen and the solid substances being severely damaged.

According to the invention, this is achieved by ensuring that the type of agitator and the direction of rotation of the agitator are chosen in such a way that the suspension is pumped upwards it the centre of the agitator.

Suitable agitators which can be used in the method according to the invention are, for example, pitched-blade agitators, propeller agitators, MIG agitators, InterMIG agitators, Lightnin agitators (e.g.

type A310, A320, C100 and C110), Chemineer agitators (HE3), Sigma agitators, Prochem massflow D agitators, Isojet agitators, Interprop agitators and Mixell TTP agitators. In practice, the axis of the agitator will usually be located vertically and centrally in the reactor, with the shaft at which the agitator is level fitted preferably being chosen such that the clearance, defined as the ratio of the distance between the bottom

in the centre of the reactor vessel (lowest point) and the lowest element of the agitator to the agitator diameter, is less than 0.3, in particular less than 0.2.

The agitator is installed and operated in the screen bottomed reactor in such a way that the suspension is pumped upwards in the centre of the agitator, that is to say that a movement starting from the filter is produced and maintained in the suspension by the rotary movement of the agitator in the centre of the agitator. The optimum stirring speed can be simply determined by the specialist and is in practice usually an optimization of, on the one hand, maximization of the flow through the screen and, on the other hand, the limitation of wear to this screen.

Suitable types of screens which can be used in the method according to the invention are for example woven screens and slotted screens. The pore size of the screen which is chosen is in practice usually between the average particle size (in terms of mass) of the first solid substance and that of the other solid substance. The optimum pore size of the screen can be simply determined by the specialist on the basis of the desired efficiency of separation. The available screen surface is chosen to be as large as possible, taking into account the desired mechanical strength of the screen.

In the context of this invention the first solid substance and the other solid substance can each comprise one compound or a mixture of compounds.

The slurry obtained after the first solid substance has been separated off can then be separated if desired, for example by filtration or decantation, after which the remaining clear stream can if desired

be used again to dilute the contents of the reactor, with the aim of improving the separation between the first solid substance and the other solid substance.

The invention is particularly suitable for use in the separation of solid catalyst particles, in particular catalysts on carriers, immobilized enzymes or whole cells from a suspension which still contains another solid substance; in this process the catalyst particles can be bigger on average than the other solid substance or slightly smaller. Because of the good separation and the small chance of damage, the catalyst can usually be re-used many times - which has a major positive effect on the economics of the processes.

When enzymes are used as the catalyst, they can for example be used as immobilized enzymes or in the form of (immobilized) whole cells or cell homogenates. Examples of immobilized enzymes which are much used in practice are acylases, amidases, hydantoinases, decarbamoylases and esterases. For immobilized enzymes the average particle size (in terms of mass) is in practice usually between 10 and 1000 ym, in particular greater than 50 ym, more particularly greater than 100 gm.

An important application of the method according to the invention consists, for example, in the production of -lactam derivatives as antibiotics, with a P-lactam nucleus being subjected to an enzymatic acylation reaction with the aid of an immobilized enzyme and a suitable acylation agent. Such processes are found to be carried out with advantage at high concentrations, with one or more output substances and/or formed substances being present in the reaction

mixture, partially in solid form, at the chosen end of the reaction.

Another important application consists, for example, in the further processing of the reaction mixture remaining after the enzymatic acylation reaction described above and after the enzyme and the solid -lactam derivative obtained have been separated off. In order to recover valuable components which are still dissolved in the reaction mixture remaining, the mixture can be subjected to an enzymatic hydrolysis reaction with the aid of an immobilized enzyme, in which the -lactam derivative is converted into the -lactam nucleus and the acylation agent remaining is converted into the corresponding acid, after which the -lactam nucleus and the acid are also present as solid substances in the reaction mixture, as well as the enzyme. After the enzyme has been separated off according to the invention, the valuable hydrolysis products can then be recovered in a simple manner.

An immobilized enzyme is used with advantage in the stated applications. A suitable immobilization technology is, for example, described in EP-A-222462.

Another suitable technology consists of immobilizing Penicillin G acylase on a carrier which contains a gelling agent, for example gelatin, and a polymer with free amino groups, for example alginate amine, chitosan or polyethyleneimine. In addition, enzymes can also be used in crystalline form (Clecs).

Of the immobilized enzymes which are commercially available, those which have been found to be particularly suitable are for example the Escherichia coli enzyme from Boehringer Mannheim GmbH which is commercially available under the name Enzygel,

the immobilized Penicillin G acylase from Recordati and the immobilized Penicillin G acylase from Pharma Biotechnology, Hannover.

Suitable enzymes which can be used in the stated enzymatic acylation reaction and the enzymatic hydrolysis reaction are for example amidases or acylases, in particular penicillin amidases or acylases. Such enzymes are for example described in J.G. Shewale et al., Process Biochemistry, August 1989, pp. 146-154, and in J.G. Shewale et al., Process Biochemistry International, June 1990, pp. 97-103.

Examples of suitable enzymes are enzymes derived from Acetobacter, in particular Acetobacter pasteurianum, Aeromonas, Alcaligenes, in particular Alcaligenes faecalis, Aphanocladium, Bacillus sr., in particular Bacillus megaterium, Cephalosporium, Escherichia, in particular Escherichia coli, Flavobacterium, Fusarium, in particular Fusarium oxysporum and Fusarium solani, Kluyvera, Mycoplana, Protaminobacter, Proteus, in particular Proteus rettqeri, Pseudomonas and Xanthomonas, in particular Xanthomonas citrii.

The method according to the invention can be suitably applied in the production of p-lactam antibiotics, for example cephalexin, amoxicillin, ampicillin, cefaclor, cefradin, cefadroxil, cefotaxime and cefazolin.

Any -lactam nucleus can in principle be used; in particular a -lactam nucleus with the general formula (1)

in which R0 stands for H or an alkoxy group with 1-3 C atoms; Y represents CH2, O, S or an oxidized form of sulphur; and represents in which Rl for example represents H, OH, halogen, an alkoxy group with 1-5 C atoms, an alkyl group with 1-5 C atoms, a cycloalkyl group with 4-8 C atoms, an aryl or a heteroaryl group with 6-10 C atoms, the groups being able to be unsubstituted or substituted by an alkyl, aryl or an alkoxy group with 1-8 C atoms, for example.

Suitable examples of -lactam nuclei which can be used in the method according to the invention are penicillin derivatives, for example 6- aminopenicillanic acid (6-APA), and cephalosporin derivatives, for example 7-aminocephalosporanic acid with or without a substituent at position 3 (7-ACA), for example 7-aminodesacetoxycephalosporanic acid (7- ADCA), 7-amino-3-chloro-ceph-3-em-4-carboxylic acid (7- ACCA) and 7-amino-3-chloro-8-oxo-l- azabicyclo[4.2.0]oct-2-ene-2-carboxylic acid.

In the (enzymatic) acylation reaction a phenylglycine in activated form, preferably a (primary, secondary or tertiary) amide or salt thereof, or a lower alkyl (1-4C) ester, for example a methyl ester,

can be used as acylation agent; phenylglycines which are either substituted or unsubstituted, in particular phenylglycine, p-hydroxyphenylglycine and dihydro- phenylglycine, are for example eligible as phenylglycines.

The temperature at which the enzymatic acylation reaction is carried out is usually lower than 40"C, preferably between 0 and 350C. The pH at which the enzymatic acylation reaction is carried out is usually between 6 and 10, preferably between 6.5 and 9.

The (enzymatic) acylation reaction and the further processing of the reaction mixture are in practice usually carried out in water. If desired, the reaction mixture can also contain an organic solvent or a mixture of organic solvents, preferably less than 30 volt. Examples of organic solvents which can be used are alcohols with 1-7 C atoms, for example a mono- alcohol, in particular methanol or ethanol; a diol, in particular ethylene glycol; or a triol, in particular glycerol.

The invention is now explained by means of the examples, without, however, being limited by them.

Abbreviations CEX = cephalexin monohydrate 6-APA = 6-aminopenicillanic acid 7-ADCA = 7-aminodesacetoxycephalosporanic acid 7-ACCA = 7-amino-3-chloro-ceph-3-em-4-carboxylic acid PGA = D-phenylglycine amide PG = D-phenylglycine HPGA = D-p-hydroxyphenylglycine amide

HPG = D-p-hydroxyphenylglycine HPGM = D-p-hydroxyphenylglycine methyl ester AssemblaseTM is an immobilized Escherichia coli penicillin acylase from E. coli ATCC 11105, as described in WO-A-97/04086. The immobilization is carried out as described in EP-A-222462, with gelatin and chitosan being used as gelling agents and glutaraldehyde as crosslinker.

The final activity of the Escherichia coli penicillin acylase is determined by the amount of enzyme added to the activated globules and was 3 ASU/g dry weight, with 1 ASU (Amoxicillin Synthesis Unit) being defined as the amount of enzyme which generates 1 g of amoxicillin.3H2O per hour from 6-APA and HPGM (at 200C; 6.5% of 6-APA and 6.5% of HPGM).

Separase-G is an immobilized Alcaligenes faecalis acylase from ATCC 19018, as described in EP-A-453047.

The immobilization is carried out as described in EP-A-222462, with gelatin and chitosan being used as gelling agents and glutaraldehyde as crosslinker. The final activity of the Alcaligenes faecalis acylase is determined by the amount of enzyme added to the activated globules and was 1600 PAU/g dry weight, with 1 PAU being defined as the amount of enzyme which hydrolyses 1 ymol penicillin G per minute under standard conditions (100 g/l penicillin G potassium salt, 0.05 M potassium phosphate buffer, pH = 8.0, 280C).

Example I: Amoxicillin I.a. Amoxicillin trihydrate was prepared in a vessel fitted with an anchor agitator, cooling jacket, pH

measuring device and temperature measuring device. The reactor was loaded with 23 litres of water at 10~C, 8.1 kg of 6-APA and 9.25 keg of HPGM, while stirring. 90 litres of enzyme suspension in water were then added.

This suspension contained 50.0 kg of enzyme net wet AssemblaseTM. 8.1 kg of 6-APA and 9.25 kg of HPGM were then added again, and the reaction volume was made up to 150 litres with water. The mixture was mixed and thermostatted at 100C. The starting pH was 6.2. After a few hours the pH rose to about 6.8, then fell again to 6.2. After 10 hours, with the pH at 6.2, the reaction had finished, and a viscous suspension had been produced. This mixture was diluted with 70 kg of demineralized water.

I.b. 190 litres of the mixture from Example I.a were transferred to a screen bottomed reactor (diameter 600 mm), fitted with a cooling jacket, temperature measuring device, agitator, 2 baffles and a bottom stop valve below the screen. The diameter of the screen placed horizontally in the centre of the bottom was 370 mm. The woven screen had a 100 gm mesh and an open surface of 40%. The pitched-blade agitator consisted of two elements with two blades. The diameter of the bottom blade was 380 mm. The distance between the underside of the agitator and the screen bottom was 30 mm.

The mixture was thermostatted at 100C. The stirring speed was set at 160 rpm. The direction of rotation of the agitator was chosen in such a way that the suspension was pumped upwards in the centre of the agitator. The draw-off pipe downstream of the bottom stop valve was connected to a vessel with a manometer

which was at an absolute pressure of 0.9 bar (the receiving tank).

After the transfer of the suspension into the screen bottomed reactor, the bottom stop valve was opened and a flow rate of 400 litres/hour to the receiving tank was set. The crystal concentration in the effluent was approximately the same as the crystal concentration in the reactor. The absolute pressure in the receiving tank was 0.9 bar.

The volume in the screen bottomed reactor was kept at 190 1 with the remaining 30 litres of the mixture. After addition of these 30 litres the volume in the screen bottomed reactor was reduced to 100 1.

The bottom stop valve was closed when a volume of 100 litres was attained in the screen bottomed reactor.

Filtration of the suspension into the receiving tank (about 120 litres) via a chamber filter press (Dieffenbach type, filter area 4.35 m2, chamber volume 60 litres, filter cloth nycot 2794) was commenced.

The subsequent continuous process was then carried out by means of the set-up illustrated in Fig.

1, in which R1 represents the conversion reactor, R2 the screen bottomed reactor, V1 a receiving tank, F1 a chamber filter press and V2 a liquid tank. The bottom stop valve of the screen bottomed reactor was opened and a flow rate of 400 litres/hour to the receiving tank (V1) was set. At the same time filtrate was fed from the chamber filter press to the screen bottomed reactor, so that the volume in the screen bottomed reactor remained about 100 litres. In this way 180 litres of filtrate were circulated through the screen bottomed reactor. The filtrate was then conveyed from the chamber filter press into the filtrate tank (V2). The screen bottomed reactor was emptied to a

volume of 90 litres. The receiving tank was completely emptied. The following product streams were obtained: [1] a suspension of immobilized enzyme in about 75 litres of liquid (total 90 litres), containing 0.6 kg of crystalline amoxicillin and a small amount of crystalline HPG, [2] a filter cake containing 30 litres of the conversion liquid and 29 kg of crystalline amoxicillin, and [3] 70 litres of filtrate.

I.c. The procedure of Example I.a. was repeated, with the difference that the enzyme suspension was then this time derived from product stream [1] of Example I.b.

The procedure of Example I.b. was then repeated. The following product streams were obtained: [1] a suspension of immobilized enzyme in about 75 litres of the conversion liquid (total 90 litres), containing 0.6 kg of crystalline amoxicillin and a small amount of crystalline HPG, [2] a filter cake containing 30 litres of the conversion liquid and 29 kg of crystalline amoxicillin and 2 kg of crystalline HPG, and [3] 70 litres of filtrate.

I.d. The procedure of Example I.a. was repeated, with the difference that the enzyme suspension was now derived from product stream [1] of Example I.c. The procedure of Example I.b. was then repeated up to the point where the entire conversion mixture had been transferred to the screen bottomed reactor. This reactor contained 190 litres of suspension with the immobilized enzyme. The stirring speed in the screen bottomed reactor was then increased from 160 to 200 rpm. The flow rate through the screen bottom was increased from 400 litres/hour to the following values:

Suspension Screening Calculated Calculated difference time I flow rate flow volume (seconds) (litres/ through in screen hour) screen bottomed (litres/m2/m reactor in) (litres) 190 - 180 56 643 100 180 -+ 170 45 800 124 170 -, 160 55 655 102 160 -, 150 52 692 107 150 -, 140 52 692 107 When a volume of 100 litres was attained in the screen bottomed reactor, the bottom stop valve was closed. Filtration of the suspension into the receiving tank (about 120 litres) via a decanter (Flottweg Z23-3 type, drum diameter 230 mm, diameter of phase separation disc 138 mm) at a rate of 500 itres/hour was commenced. The drum speed is 6000 rpm. The differential speed between the drum and propeller is 60 rpm.

The subsequent continuous process was then carried out in the setup illustrated in Fig. 1, in which F1 represents the decanter. The bottom stop valve of the screen bottomed reactor was opened and a flow rate of 500 litres/hour to the receiving tank was set.

At the same time filtrate from the decanter was added to the screen bottomed reactor, so that the volume in the screen bottomed reactor remained about 100 litres.

In this way 180 litres of filtrate were circulated through the screen bottomed reactor. The filtrate from the decanter was then conveyed into the filtrate tank (V2). The screen bottomed reactor was emptied to a

volume of 90 litres. The receiving tank was completely emptied. The following product streams were obtained: [1] a suspension of immobilized enzyme in about 75 litres of the liquid (total 90 litres), containing 0.6 kg of crystalline amoxicillin and a small amount of crystalline HPG, [2] a filter cake containing 30 litres of the conversion liquid and 30 kg of crystalline amoxicillin and 2 kg of crystalline HPG, and [3] 70 litres of filtrate.

I.e. A sample of about 4 1 was taken from the reaction mixture obtained in Example I.c, after dilution with water and before transfer to the screen bottomed reactor. This sample was transferred to a second, non- baffled screen bottomed reactor with a diameter of 160 mm, a horizontal woven screen with a 100 ym mesh and an open surface of 40%. The screen bottomed reactor was equipped with a pitched-blade agitator (one element with two blades; diameter 112 m). The distance from the underside of the agitator to the screen bottom was 3 mm. In addition to the immobilized enzyme, the sample contained 14 wt% of amoxicillin crystals (needles with a length < 20 ym) and 1 wt% of precipitated HPG. The effluent was returned to the second screen bottomed reactor. The flow rate was increased in stepwise fashion. The stirring speed which was set was 612 rpm.

The bottom stop valve was opened and suspension was pumped out of the reactor by means of a peristaltic pump.

The direction of rotation of the agitator was chosen in such a way that the suspension was pumped upwards in the centre of the agitator. The screen- bottom stream was adjusted with the pump in such a way

that the crystals were removed from the reactor contents, in such a way that the crystal concentration in the effluent was approximately the same as the crystal concentration in the reactor. When separation was good, the pressure in the suction pipe was about 0.9 bara (bar absolute). The flow rate could be set in such a way that the flow through the screen was 110 litres/m2/min with good separation.

Comparative Experiment A: Example I.e was repeated, but now the direction of stirring was reversed so that the suspension was pumped downwards in the centre of the agitator. The maximum flow which could be set was only 60 litres/m2/min.

Example II: Ampicillin II.a. A PGA solution was prepared by suspending 301.6 g of PGA (purity 99.5 wt%) in 650 g of demineralized water at 50C. 102.1 g of 96 wt% sulphuric acid were gradually added over a period of about 1 hour, while stirring and cooling. A clear PGA.SH2SO4 solution was obtained, with a pH of 3.8 at 220C.

II.b. Ampicillin trihydrate was prepared in a screen bottomed reactor fitted with an agitator, 4 baffles, cooling jacket, pH meter, titration facility, thermometer and bottom stop valve below the screen. The horizontal screen bottom of the reactor had a diameter of 90 mm. The woven screen had a 175 ym mesh and an open surface of 36%. The pitched-blade agitator (one element with four blades) had a diameter of 50 mm. The

distance between the underside of the agitator and the screen bottom was 5 mm. The draw-off below the screen with the bottom stop valve was connected to a vacuum receiving flask. The reactor was loaded with a suspension of AssemblaseTM in water. After 1 minute of draining the 1net wet" weight was 300 g.

131.6 g of 6-APA (purity 98.6 wt%), 30.2 g of PGA (purity 99.5 wt%) and 400 g of demineralized water at 10~C were supplied in succession to a production vat.

This suspension, hereinafter called the production mixture, was stirred for 15 minutes at a temperature of 100C.

The reaction was then started by transferring the production mixture with 100 g of demineralized water at 100C to the screen bottomed reactor. The suspension was stirred and thermostatted at 100C.

423.7 g of PGA.SH2SO4 solution were then metered in over a period of 283 minutes at a constant rate. During this entire process the temperature and pH of the suspension remained 100C and 6.3, respectively. Starting from 328 minutes after the start of the reaction the pH was kept constant at 6.3 by titration with a 6N (aqueous) H2SO4 solution at a temperature of 100C. After 540 minutes from the start of the reaction the pH was reduced from 6.3 to 5.6 by adding 6N H2S04 solution.

II.c. The 1410 g of slurry obtained in Example II.b contained 16 wt% of ampicillin trihydrate crystals, 3.6 wtk of PG crystals, 300 g net wet weight of immobilized enzyme and conversion liquid. The stirring speed which was set was 500 rpm. An absolute pressure of 0.9 bar was set in the vacuum receiving flask. The

bottom stop valve was opened, after which a suspension was drawn from the reactor.

The direction of rotation of the agitator was chosen in such a way that the suspension was pumped upwards in the centre of the agitator. The sieve-bottom flow was adjusted with the pump in such a way that the crystals were removed from the reactor contents, in such a way that the crystal concentration in the effluent, on the basis of a visual inspection, was the same as the crystal concentration in the reactor. When separation was good, the absolute pressure in the suction pipe was about 0.9 bar.

The flow rate could be adjusted in such a way that the flow during the good separation of the first 300 g of suspension by the screen was 30 litres/m2/min on average.

II.d. The suspension obtained from Example II.c was filtered through a G3 filter. This yielded about 150 g of filtrate. This filtrate was returned to the screen bottomed reactor. A further 300 g of suspension were then drained off through the screen and filtered through the same filter with the filter cake already there. This yielded about 200 g of filtrate, which were again returned to the screen bottomed reactor. This procedure was repeated until a volume of 6 litres of filtrate was returned. During this process the flow through the screen bottom, with good separation, increased from 30 to >500 litres/m2/min.

The filter cake on the G3 filter contained more than 98% of the solid ampicillin trihydrate and PG initially present in the screen bottomed reactor. After the separation the screen bottomed reactor contained

more than 99.5 wt of the immobilized enzyme initially present.

Comparative Experiment B: Example II.c was repeated, but now the direction of stirring was reversed so that the suspension was pumped downwards in the centre of the agitator. The maximum flow which could be set was now less than 5 litres/m2/min.

Example III: Cefaclor: III.a. Cefaclor: -naphthol:water (2:1:6 moles) complex (Cefaclor- -naphthol complex) was prepared in a screen bottomed reactor fitted with an agitator, 4 baffles, cooling jacket, pH meter, thermometer and bottom stop valve below the screen. The horizontal screen bottom of the reactor had a diameter of 90 mm. The woven screen had a 175 Hm mesh and an open surface of 36%. The pitched-blade agitator (one element with four blades) had a diameter of 50 mm. The distance between the underside of the agitator and the screen bottom was 5 mm. The draw-off below the screen with the bottom stop valve was connected to a vacuum receiving flask. The reactor was loaded with a suspension of AssemblaseTM in water. After 1 minute of draining the "net wet" weight was 200 g.

95.1 g of 7-ACCA (purity 97 wit), 83.4 g of PGA (purity 99.5 wit), 4.0 g of sodium bisulphite and 568.7 g of demineralized water at 10~C were supplied in succession to a production vat. The suspension was then mixed and 28.8 g of -naphthol added. The pH was then brought back from 7.7 to 7.0 with 4N H2S04 solution, at

a temperature of 10"C. The suspension, hereinafter called the production mixture, was stirred for 15 minutes at a temperature of 10~C.

The reaction was then started by transferring the production mixture with 171.8 g of demineralized water at 100C to the screen bottomed reactor. The suspension of immobilized enzyme, 7-ACCA and liquid was stirred and thermostatted at 100C. Five minutes after the start of the reaction 130 mg of Cefaclor- -naphthol complex were metered in as seed crystals. During the reaction the pH was kept at 7.0 by titration with a 4N H2SO4 solution. The total consumption of 4N H2SO4 solution for pH adjustment and titration was 116.2 g.

The reaction was considered to have finished 176 minutes after the start.

III.b. The 1288 g of slurry obtained in Example III.a contained 14 wt% of Cefaclor- -naphthol complex crystals with an average diameter of 5 ym, 0.1 wt% of naphthol crystals with an average diameter of 5 gm and 200 g net wet immobilized enzyme with an average diameter of 420 ijm. The stirring speed which was set was 500 rpm. A pressure of 0.9 bara was set in the vacuum receiving flask. The bottom stop valve was opened, as a result of which a suspension was drawn from the reactor.

The direction of rotation of the agitator was chosen in such a way that the suspension was pumped upwards in the centre of the agitator. The sieve-bottom flow was adjusted with the pump in such a way that the crystals were removed from the reactor contents, in such a way that the crystal concentration in the effluent, on the basis of a visual inspection, was the

same as the crystal concentration in the reactor. When separation was good, the absolute pressure in the suction pipe was about 0.9 bar.

The flow rate could be adjusted in such a way that the flow during the good separation of 950 g of suspension (about 75% of the reactor contents) by the screen was 25 litres/m2/min on average.

Comparative Experiment C: Example III.b was repeated, but now the direction of stirring was reversed so that the suspension was pumped downwards in the centre of the agitator. The maximum flow which could be set was less than 5 litres/m2/min.

III.c. The suspension obtained from Example III.b was filtered through a G3 filter. This yielded about 600 g of filtrate. This filtrate was returned to the screen bottomed reactor and the suspension was stirred for 4 minutes. About 1050 g of suspension from the reactor were then separated with an average flow of 110 litres/m2/min and filtered through the same filter with the filter cake already there. The filtrate obtained was again returned to the screen bottomed reactor, and the screening and filtration procedure was repeated once.

The filter cake on the G3 filter contained more than 95% of the Cefaclor- -naphthol crystals initially present in the screen bottomed reactor. After the separation the screen bottomed reactor contained more than 99.5 wt% of the immobilized enzyme initially present.

Example IV: Cefalexin: IV.a. Processing of a reaction mixture, obtained after enzymatic acylation, via enzymatic hydrolysis in a screen bottomed reactor fitted with an agitator, 2 baffles, meter, thermometer and bottom stop valve below the screen. The conical screen bottom of the reactor had a diameter of 200 mm projected on the horizontal plane. The angle between the screen and the horizon was 15". The woven screen had a 175 ym mesh and an open surface of 36%. The agitator was of the InterMIG type and had a diameter of 100 mm. The distance between the underside of the agitator and the centre of the screen bottom was 14 mm. The distance from the tip of the agitator to the screen bottom was 2 mm. The reactor was loaded with an immobilized enzyme suspension (SeparaseTM type) in water. After 1 minute of draining the "net wet" weight was 525 g.

3120 g of demineralized water, 14.0 g of sodium bisulphite, 10.5 g of PG, 140.0 g of 7-ADCA, 70.0 g of CEX, 56.0 g of PGA and 87.5 g of ammonium sulphate were supplied in succession to a production vat thermostatted at 200C. The pH was adjusted to 7.5 with about 40 g of 25 wt% ammonia solution. The mixture was stirred for 15 minutes. The almost clear mixture was then filtered through a G3 filter. The filtrate was transferred into the production vat and thermostatted at 200C.

The contents of the production vat were added to the screen bottomed reactor over a period of 8 minutes. After about 1 minute, stirring was commenced in the screen bottomed reactor. About 2 minutes later a white suspension of PG crystals was obtained. During

the reaction the pH rose from 7.5 to 8.0. The reaction was considered to have finished 30 minutes after the start of supply from the production vat.

IV.b. The 3540 g of suspension obtained contained 1.3 wt% of PG crystals with an average diameter of 50 ijm, 525 g "net wet" immobilized enzyme with an average diameter of 450 ym, and conversion liquid.

The draw-off below the screen with the bottom stop valve was connected to a circuit consisting of a manometer, a circulation pump, a flowmeter, a heat exchanger and a dipped return pipe to the screen bottomed reactor. The stirring speed which was set was 150 rpm. The direction of rotation of the agitator assembly was chosen in such a way that the suspension was pumped upwards in the centre of the agitator. The bottom stop valve below the screen was opened and a flow was set at such a level that for at least 5 minutes the crystal concentration in the circulation pipe, on the basis of a visual inspection, was the same as the crystal concentration in the reactor. The maximum flow corresponded to a flow of 63 litres/m2/min through the screen.

The accuracy of the visual inspection was confirmed by transferring about 100 ml of the reactor contents into a measuring cylinder during the screening process. The measuring cylinder was then set aside for 2 minutes, in an absolutely stationary state. In this period a brown-coloured layer of solid material with a volume of about 15 ml was formed by precipitation. This layer consisted of the immobilized enzyme. A white layer of solid material with a volume of about 5 ml was formed on top of this layer by precipitation. This layer consisted of PG crystals. The same test was

repeated with about 100 ml of suspension from the circulation run. After 2 minutes of precipitation time the brown-coloured layer was found to be absent and a white layer of solid material with a volume of about 5 ml was formed by precipitation. This layer consisted of PG crystals.

The flow rate maximization was repeated at stirring speeds of 200, 250, 300 and 350 rpm. Results of measurement are given in Table 1.

The pressure in the suction pipe of the circulation pump was approximately 0.9 bar during all the screening processes.

Comparative Experiment D: Example IV.b was repeated, but the direction of stirring was reversed this time, so that the suspension was pumped downwards in the centre of the agitator, the maximum flow was 31 litres/m2/min.

The flow rate maximization was repeated at stirring speeds of 200, 250, 300 and 350 rpm. Results of measurement are given in Table 1.

Table 1 Number of Direction of Flow through revolutions pumping in screen ( rpm) centre of (litres/m2/mi agitator n) 150 up 66 down 31 200 up 110 down 70 250 up 171

down 96 300 up >171 down 113 350 up >171 down 131 I IV.c. Processing of a reaction mixture, obtained after enzymatic acylation, via enzymatic hydrolysis in a screen bottomed reactor fitted with an agitator, 2 baffles, pH meter, thermometer, demineralized water spraying system and bottom stop valve below the screen.

The conical screen bottom of the reactor had a diameter of 1400 mm projected on the horizontal plane. The angle between the screen and the horizon was 15". The woven screen had a 175 ym mesh and an open surface of 36%.

The agitator consisted of one element with two blades, with the characteristic that the directions of pumping in the centre and at the outer ends of the agitator elements were opposite, and had a diameter of 700 mm.

The distance between the underside of the agitator and the middle of the screen bottom was 144 mm. The distance from the tip of the agitator to the screen bottom was 57 mm. The reactor was loaded with a suspension of SeparaseTM in water. After 1 minute of draining the "net wet" weight was 150 kg.

A production vat contained 924 litres of clear liquid with a temperature of 21"C and a pH of 7.6.

This mixture contained 0.74 wt% of PG, 2.77 wt% of 7-ADCA, 1.94 wt% of CEX and 0.59 wt% of PGA (determined by HPLC).

The contents of the production vat were added to the screen bottomed reactor over a period of 8

minutes. After about 2 minutes, stirring was commenced in the screen bottomed reactor. About 2 minutes later a white suspension of PG crystals was obtained. During the reaction the pH rose from 7.5 to 8.0. The reaction was considered to have finished 30 minutes after the start of supply from the production vat.

IV.d. The approximately 1050 litres of suspension contained 1.2 wt% of PG crystals with an average diameter of 50 ym and 150 kg net wet immobilized enzyme with an average diameter of 450 ym.

The draw-off below the screen with the bottom stop valve was connected to a circuit consisting of a manometer, a circulation pump, a flowmeter and a dipped return pipe to the screen bottomed reactor. The stirring speed which was set was 32 rpm. The direction of rotation of the agitator was chosen in such a way that the suspension was pumped upwards in the centre of the agitator. The bottom stop valve below the screen was opened and a flow was set at such a level that for at least 5 minutes the crystal concentration in the circulation pipe, on the basis of a visual inspection, was the same as the crystal concentration in the reactor. The maximum flow corresponded to a flow of 38 litres/m2/min through the screen.

The accuracy of the visual inspection was confirmed by transferring about 100 ml of the reactor contents into a measuring cylinder during the screening process. The measuring cylinder was then set aside for 2 minutes, in a state of absolute rest. In this period a brown-coloured layer of solid material with a volume of about 15 ml was formed by precipitation. This layer consisted of the immobilized enzyme. A white layer of solid material with a volume of about 5 ml was formed

on top of this layer by precipitation. This layer consisted of PG crystals. The same test was repeated with about 100 ml of suspension from the circulation run. After 2 minutes of precipitation time the brown-coloured layer was found to be absent and a white layer of solid material with a volume of about 5 ml was formed by precipitation. This layer consisted of PG crystals.

The flow rate maximization was repeated at stirring speeds of 46, 60 and 200 rpm. Results of measurement are given in Table 2.

The pressure in the suction pipe of the circulation pump was approximately 0.9 bar during all the screening processes.

Comparative Experiment E: Example IV.d was repeated, but the direction of stirring was reversed, so that the suspension was pumped downwards in the centre of the agitator; the maximum flow was c5 litres/m2/min this time.

The flow rate maximization was repeated at stirring speeds of 46, 60 and 200 rpm. Results of measurement are given in Table 2.

Table 2 Number of Direction of Flow through revolutions pumping in screen ( rpm) centre of (litres/m2/mi agitator n) 32 up 38 down <5 46 up 56

down <5 60 up 63 down <5 200 up not measured down <5 IV.e. The bottom stop valve below the screen in the set-up described in Example IV.d was closed and the suspension in the circuit was returned to the reactor.

The suspension present in the reactor had a white colour. The draw-off below the screen was then connected in succession to a manometer, a pump, a flowmeter, a chamber filter press (filter area 3.2 m2, chamber volume 60 litres, filter sheets T-1000) and a bifurcation to (1) a dipped return pipe to the screen bottomed reactor and (2) a discharge pipe to a receiving tank. The stirring speed which was set was 60 rpm.

The direction of rotation of the agitator was chosen in such a way that the suspension was pumped upwards in the centre of the agitator. The dipped return pipe was opened and the discharge pipe to the receiving tank was closed. The bottom stop valve below the screen was opened and a flow of 70 litres/minute was set. The crystal concentration in the suction pipe of the circulation pump, on the basis of a visual inspection, was the same as the crystal concentration in the reactor. The flow corresponded to a flow of 50 litres/m2/min through the screen.

The accuracy of the visual inspection was confirmed in the same way as described in Example IV.d.

For a period of 112 minutes, suspension was in this way removed from the reactor contents and filtered, and filtrate was returned to the reactor contents. As a result of an increase in the cake resistance in the chamber filter press, the circulation flow rate decreased from 70 litres/min to 30 litres/min. The total volume of filtrate returned was about 4500 litres. The suspension left in the reactor had a brown colour.

The discharge pipe to the receiving tank was then opened and the dipped return pipe to the screen bottomed reactor was closed. A suspension with few crystals was removed from the reactor at a rate of 30 litres/min via the screen. The agitator was stopped as soon as the elements of the agitator became visible.

In this way an enzyme bed was formed on the screen. As soon as air was drawn through the enzyme bed by the pump below the screen bottom, 230 litres of demineralized water were metered in at a rate of 20 litres/min to the screen bottomed reactor via the sprayers. The enzyme was then drained for 2 minutes and the pump was stopped. Nitrogen was then blown through the pipes and the chamber filter press for 15 minutes in order to collect liquid and dry the filter cake.

The chamber filter press was evacuated. The filter cake removed weighed 44.2 kg and contained 14.1 kg of PG. More than 95% of the PG crystals present in the reactor were isolated by this method.