NO160210B - N-TRIMETHYLSILYLOXYCARBONYL-6-AMINOPENICILLIN INTERMEDIATE FOR THE PREPARATION OF A PENICILLIN. - Google Patents

Description

The present invention relates to an N-trimethylsilyloxy-carbonyl-6-amino-penicillin intermediate for the production of a penicillin with the formula

where R-CO is oc-aminoarylacetyl or phenoxyacetyl, where aryl means phenyl, which is optionally substituted with an OH^ group.

More specifically, the invention relates to intermediates for use in the production of antibacterial agents of the class usually called semisynthetic penicillins, and preferably of the subclass characterized by an α-amino group and the acyl side chain in the 6-position, such as in ampicillin and amoxicillin.

The invention further relates to a method for producing a compound as described above.

The first commercial penicillin with an α-amino group on the 6-acylamidoside chain was ampicillin, which is 6-(D-α-amino-α-phenylacetamido)penicillanic acid (see US-PS 2,985,648).

Amoxicillin is an antibacterial agent used in human therapy and is marketed as a trihydrate of the free acid (ie the zwitterion). The connection is e.g. described in GB-PS 978 178, in "J. Chem. Soc." (London), pp. 1920-1922

(1971) and in "Antimicrobial Agents and Chemotherapy", 1970, pp. 407-430 (1971). The chemical name is 6-[D-α-amino-α-(p-hydroxyphenyl)acetamido]penicillanic acid.

The use of amino acid chloride-hydrochloride for the production of such penicillins was discussed in the patent literature, e.g. in GB-PS 938 321 and 959 853 under anhydrous conditions (the latter used protection during the acylation of the carboxy group of 6-aminopenicillanic acid with a silyl group, as also discussed in GB-PS 1 008 468 and US-PS 3 249 622) as well as in GB- PS 962 719 in cold aqueous acetone. These penicillins are amphoteric amino acids and therefore certain aliphatic unsymmetrical branched secondary amines (often called liquid amine resins), which had previously been used for the isolation of 6-aminopenicillanic acid which is also an amphoteric amino acid (see US-PS 3,008,956). Improved methods for the isolation and purification of such penicillins are discussed e.g. in US-PS 3,180,862 via P-naphthalene sulfonates and in US-PS 3,198,804 via intermediate isolation and subsequent light hydrolysis of hetacillin.

The use of a silyl group to protect the carboxy group of a natural penicillin during chemical cleavage to 6-aminopenicillanic acid is disclosed in US-PS 3,499,909. The use of silylated 6-aminopenicillanic acid during anhydrous acylation with amino acid chloride hydrochlorides is disclosed in numerous patents , e.g. in US-PS 3,479,018, 3,595,855, 3,654,266, 3,479,338 and 3,487,073. Some of these patents also deal with the use of liquid amine resins; see also US-PS 3,912,719, 3,980,637 and 4,128,547.

GB-PS 1 339 605 contains various specific and detailed examples of the preparation of amoxicillin by reacting a silylated derivative of 6-aminopenicillanic acid with a reactive derivative (including chloride, hydrochloride) of D-( - )-cx-amino-p-hydroxy enylacetic acid, where the amino group is protected, then removal of the silyl group or groups by hydrolysis or alcoholysis and then, if possible, recovery of the amoxicillin, usually as the crystalline trihydrate. Thus, crystalline amoxicillin was obtained in example 1 by isoelectric precipitation from an aqueous solution, e.g. at a pH value of 4.7. Purification was probably achieved in this example by dissolving the crude product (before isoelectric precipitation) in water at an acidic pH value such as 1.0 (e.g. in aqueous hydrochloric acid) in the presence of a water-immiscible organic solvent, such as methyl -isobutyl ketone (4-methylpentan-2-one). Essentially the same method was used in US-PS 3,674,776.

From "Synthesis", May 1970, pages 259-260, a method is known for the preparation of N-silyloxycarbonyl amino acid derivatives by adding carbon dioxide. However, the article in question contains nothing that could draw attention to using such a reaction and a penicillin compound to obtain a carbamate intermediate that can withstand acylation during the formation of desired penicillins, as it is well known that the 3-lactam ring in penicillin is very sensitive.

The intermediate product as described above is characterized by the fact that the compound is dissolved in an anhydrous, organic solvent and that it has the formula

where B means a cleavable ester protecting trimethylsilyl group.

As mentioned, the invention also includes a method for producing a compound with the formula where B has the above meaning, and this method is characterized by adding dry carbon dioxide to a reading of an N-trimethylsilylpenicillin compound with the formula

where B has the above meaning, in an anhydrous, inert organic solvent at a temperature within the range 0-100°C, until the carbonylation reaction is complete.

The intermediates according to the invention provide the following surprising properties and advantages: a) the acylation reaction can be carried out at a much higher temperature than usual without loss of yield; b) the acylation mixture is very stable and is kept for 24 hours at 5°C without yield loss. This is in stark contrast to other anhydrous acylation processes which show extensive degradation over this time period at this temperature; c) very high yields of 85# or more are always obtained; d) the method can be used for the production of amoxicillin of very high quality and which does not contain toxic bases such as N,N-dimethylaniline or N-methylmorpholine; e) the carboxylated mixture containing the new intermediate has a practically infinite stability. Nothing

loss of yield may be noted after storage at

room temperature for several weeks;

f) no exothermic reaction occurs during the acylation, which would require cooling in the same way as in other

acylation processes.

In a preferred embodiment for the preparation of the compound of the formula

dry carbon dioxide as gas is added to a solution of trimethylsilyl-6-trimethylsilylaminopenicillanate in an anhydrous, organic, inert solvent, preferably methylene chloride, at room temperature, or at a temperature in the range 0-100°C until the reaction is complete.

A compound with the formula is further produced as an embodiment of the present invention

where B means an easily cleavable ester protecting group or more particularly an easily cleavable ester protecting group selected from the group consisting of trimethylsilyl, benzhydryl, benzyl, p-nitrobenzyl, p-methoxybenzyl, trichloroethyl, phenacyl, acetonyl, methoxymethyl, 5-lndanyl, 3 -phthalidyl, l-[(ethoxycarbonyl)oxy]ethyl, pivaloyloxymethyl and acetoxymethyl. A compound with the formula is further produced as a preferred embodiment

This compound is referred to here by various trivial names such as bis-silylated carbamate of 6-APA, SCA, 6-trimethyl-silyloxycarbonylpenicillanic acid TMS-ester and TMS02C, APA,

TMS.

The very existence of this compound is surprising in view of the well-known fact that reaction of 6-APA with carbon dioxide destroys 6-APA and gives 8-hydroxy-penicillanic acid as discussed in TJS-PS 3 225 033.

One means of obtaining quantitative yields of 6-trimethylsilyloxycarbonylaminopenicillanic acid trimethylsilyl ester (TMSOi.APA.TMS) lies in completely preparing the 6-APA bis-TMS precursor in the first instance. This was achieved by reacting 6-APA with hexamethylenedisilazane (HMDS) as in the following reaction scheme:

The completion of the bis-trimethylsilylation reaction can be easily followed using NMR. The 3-trimethylsilyloxycarbonyl group shows a methylsilyl singlet at 0.31 ppm (tetramethylsilane = 0), while the 6-trimethylsilylamino group shows a methylsilyl singlet at 0.09 ppm.

The 6-APA trimethylsilylation reaction has so far only been carried out in methylene chloride. Other solvents can, however, be used such as e.g. acetonitrile, dimethylformamide or even HMDS.

Conversion of the trimethylsilylamino group to the trimethylsilyloxycarbonylamino group is easily achieved by bubbling dry CO2 through the reaction solution. The transformation is easily followed by NMR because the trimethylsilylamino singlet at 0.09 ppm declines as a new singlet appears for the trimethylsilyloxycarbonylamino group at 0.27 ppm.

When the invention is used in connection with the manufacture of ampicillin, ampicillin anhydrate, ampicillin trihydrate, amoxicillin and amoxicillin trihydrate, the end products are isolated and purified according to conventional methods well known in the art, as set forth in US Patent Nos. 3,912,719, 3,980,637 and 4 128 547, as well as other patents and publications specified therein.

The acid chlorides used in the following examples can be replaced by a wide range of other acid chlorides for the production of conventional penicillins.

Thus, the acyl halide can be selected for the introduction of an arbitrary desired acyl group in the 6-amino position, as is well known in this art, described e.g. in US-PS 3,741,959.

The invention shall be illustrated by means of figure 1 and figure 2 which show 80 MHz NMR spectra in the range 0-2.0 ppm.

Thus shows:

Figure 1 80 MHz NMR spectra of the 6-APA silylation mixture in the range 0-2 ppm. Prepared integration data show 100# silylation of the 3CO2H and 6-amino groups, integrated against GEM-dimethyl I 6-APA; Figure 2 shows 80 MHz NMR spectra of the carboxylation mixture after gas treatment of the previous silylation mixture (Figure 1) with CO2. This spectrum shows total conversion of 6-APA-bis-TMS to the carbamate (no 6-amino-TMS can be detected).

Acid chlorides are normally prepared under vigorous conditions such as by treating the acid under reflux with thionyl chloride, but they can, when sensitive groups are present, including sensitive blocking groups, be prepared under practically neutral conditions by reacting a salt of the acid with oxalyl chloride.

The method according to the invention will be explained in more detail in the following examples.

Example 1

To a mixture of 6-aminopenicillanic acid (6-APA) and 10 ml of CD 2 Cl 2 and 1.13 ml of trimethylchlorosilane at a temperature of 26-27' C, 1.23 ml of triethylamine was added dropwise over the course of 30 minutes. Stirring was continued for another 2 hours. Dry carbon dioxide gas was then bubbled through the mixture for approx. 3 hours. At the end of this period, NMR spectra showed the presence of 6096 silylated carboxy-6-APA (SCA) with the structure

The mixture was kept in a refrigerator overnight. The next morning, 0.77 ml of N,N-dimethylaniline was added and the mixture was cooled to -8°C. Then 1.2 g of D-(-)-p-hydroxy-2-phenylglycyl chloride hydrochloride with a purity of 79# was added in portions as follows:

At the end of the 310 minute reaction period, thin layer chromatography (TLC) performed on a sample of the reaction mixture using a solvent system of 0% ethyl acetate, 20% acetic acid and 20% water shows the presence of amoxicillin.

To a cold 2 ml sample of the final reaction mixture was added 1.0 ml of D2O. After separation by centrifugation, the aqueous phase was shown by NMR to contain 78% amoxicillin and approx. 20% 6-APA. The presence of amoxicillin was also confirmed by TLC.

Example 2

A mixture of 5.4 g or 0.025 mol of 6-aminopenicillanic acid and 6.2 ml of 93% hexamethylenedisilazane (HMDS; 0.0275 mol), as well as 0.07 g or approx. 0.001 mol of imidazole in 40 ml of CH2Cl2 was refluxed under a nitrogen purge for approx. 17.5 hours. At the end of this period, 0.13 ml or approx. 0.001 mol trimethylchlorosilane (TMCS). The resolution became unclear. Refluxing continued for an additional 7 hours and deposits of NH 4 Cl were observed in the condenser. AT this point, NMR spectra showed approx. 100% silylation of both the amino and carboxyl groups in 6-APA. 0.2 ml or 0.00125 mol (approx. 5 mol %) of HMDS and 0.06 ml or approx. 0.0005 mol TMCS, and the reflux under nitrogen purging was continued for another 17 hours. At this point, the NMR spectrum was the same as before, but with the addition of small amounts of HMDS and TMCS. Dry carbon dioxide was then bubbled through the reaction mixture at room temperature for 175 minutes, after which NMR spectra showed no HMDS and more than 92% silylated carboxy-6-APA (SCA). 4.45 ml or 0.035 mol of N,N-dimethylaniline (DMA) was then added and the mixture was cooled to -3°C. 5.65 g or 0.026 mol of D-(-)-2-phenylglycyl chloride with a purity of 95% was then added in portions as follows:

The reaction was followed by NMR spectra which showed very little change approx. 5 hours after the start of the reaction. The temperature was then 3°C. The reaction mixture was kept packed in ice for the next 16 hours. It was then removed from refrigeration and stirred for 3.5 hours at room temperature (about 20-24°C). A large amount of solids was still present. The reaction mixture was then stirred at room temperature, 22-24°C, for approx. 63 hours. At the end of this period there was only a slight blur. Upon D20 extraction of a sample, NMR spectra showed ampicillin and 6-APA.

The reaction mixture was cooled to approx. 0°C and stirred for 5 minutes in the cold after adding 35 ml of ice water. After smooth filtration, the mixture was washed with cold water and CH 2 Cl 2 . The aqueous phase showed, after separation by TLC, a large zone which was slower than ampicillin and 6-APA, and which represented a new intermediate X.

The aqueous phase was adjusted to a pH value of 3.0 with NH 4 OH and was inoculated with ampicillin. Methyl isobutyl ketone (MIBK) was added in an amount of 35 ml and the mixture was stirred, adjusted to a pH of 5.2 with more NH 4 OH, stirred at 20°C for 1 hour, stirred in an ice bath for another 1 hour and cooled overnight. The precipitate of ampicillin was collected by filtration, first washed with 25 ml of cold water and then with 40 ml of MIBK and finally with 40 ml of a mixture of 85 parts of isopropyl alcohol and 15 parts of water, dried at 50°C and found to weigh 4 .5 g as the identity as ampicillin was confirmed by TLC.

Example 3

A mixture of 5.4 g 6-APA, 6.2 ml 93% wt HMDS and 0.06 g imidazole in 50 ml CH 2 Cl 2 was refluxed and purged with nitrogen for 18 hours. Then 0.1 ml of TMCS was added which caused cloudiness. Refluxing for a further 2 hours gave a clear solution of NH4CI in the condenser. An additional 0.1 ml of TMCS was then added which left only a very slight cloudiness. The reflux was continued under a nitrogen purge for the next 65 hours. The mixture was then cooled to approx. 22°C and the addition of dry carbon dioxide was started. After 75 minutes, NMR spectra showed formation of more than 90% bis-silylated carbamate (SCA). 4.45 ml of DMA were then added and then 5.6 g of D-(-)-2-phenylglycyl chloride hydrochloride with 97% purity in proportions as follows:

After this mixture was stirred for a further 17 hours, TLC was performed on samples of the reaction mixture and on a diluted reaction mixture (1 ml of the mixture diluted with 2 ml of CH2Cl2) and found in each a small zone of ampicillin and a large zone of a new intermediate X.

The reaction mixture was then cooled to 0°C, 40 ml of ice water was added and the mixture was stirred for 5 minutes, smooth filtered and washed with water and with CH 2 Cl 2 . The aqueous phase was separated, 10% removed for testing and the remainder adjusted to a pH of 3.0 with NH 4 OH, inoculated with ampicillin and stirred. After addition of another 40 ml of MIBK, the mixture was stirred and the pH was adjusted to 5.2 with NH 4 OH and stirred at room temperature for 1 hour and then in an ice bath for another 1 hour. Crystals were precipitated. After cooling overnight, the crystalline product was collected by filtration, washed successively with MIBK, water and MIBK and then with 40 ml of isopropanol-water in a ratio of 85:15 and dried at 45°C to obtain 6.25 g of ampicillin (6 .8 g corrected for sampling, corresponding to a yield of 68$).

Example 4

To a mixture of 1.0 6-APA and 1.13 ml TMCS in 10 ml CD 2 Cl 2 1.23 ml TEA 1 was added dropwise over 30 minutes, and the mixture was stirred for a further 2 hours. Dry carbon dioxide was bubbled through for 4 hours. At this point, NMR showed approx. B5- b0% carboxysilylation. The mixture was kept in the refrigerator overnight. In the morning, 0.77 ml of DMA was added, the mixture was stirred, cooled to -8°C and 1.2 g of D-(-)-p-hydroxy-2-phenylglycyl chloride hydrochloride was added in portions as follows:

At the end of the 310-minute period, NMR showed approx. 18% amoxicillin and approx. 20$ 6-APA.

Example 5

10.0 g or 46.24 mmol corresponding to 1.0 equivalent of dry 6-aminopenicillanic acid was suspended in 175 ml of anhydrous methylene chloride with stirring at 25°C. 10.76 g or 106.23 mmol corresponding to 2.30 equivalents of triethylamine was added

at 25°C followed by the addition of 11.70 g or 107.75 mmol corresponding to 2.33 equivalents of trimethylchlorosilane over 10-15 minutes, the temperature being maintained below about 30°C by the rate of addition of trimethylchlorosilane. After stirring for 20-30 minutes, the mixture containing precipitated triethylamine hydrochloride was analyzed for complete silylation by 80 MHz NMR. The mixture was then treated with carbon dioxide at 20°C for about 2 hours and analyzed for complete carboxylation by 20 MHz NMR. Further. gas addition was necessary during this time. The volume of the carboxylation mixture was, if necessary, adjusted to about 175 mL with dry methylene chloride. After complete carboxylation, the slurry was treated with 2.95 g or 3.56 mL or 50.87 mmol corresponding to 1, 1 equivalent of propylene oxide and cooled to 0-5"C. D-(-)-2-(p-hydroxyphenyl)-glycyl chloride hydrochloride hemidioxane solvate was added in five portions to 2.71 g at approx. 2' C (total 13.54 g or 50.87 mmol corresponding to 1.1 equivalent). Each portion of acid chloride was allowed to dissolve before the next portion was added. (Stirring was stopped and the mixture examined for any solids at the bottom of the flask. The slurry must not be heated above 5°C for this test, as the results may be erroneous). This required approx. 20 minutes per portion. This portionwise addition was very important. The final acylation mixture was examined for any undissolved acid chloride hydrochloride. The mixture was kept at 0-5°C for 30 minutes and treated with cold (0-5<*>C) deionized water (DI) in an amount of 100 ml under very rapid stirring for 10 minutes. The mixture was allowed to separate and the lower methylene chloride phase was removed. The aqueous phase was smooth filtered (very little solids) through a thin ("Dicalite") layer of diatomaceous earth and the cake was washed with cold (0-5°C) DI water in an amount of 15 ml. Any lower phase of organic layer was removed before crystallization. The clear, pale yellow, aqueous solution with a pH value of 2-2.5 was adjusted to pH 3.5 at 0-5°C and if necessary inoculated. The slurry was kept at 0-5°C for 40 minutes and the pH- the value was adjusted to 4.8-5.0 with 6N ammonium hydroxide and crystallized for 2 hours. The slurry was filtered and the solid amoxicillin thus collected was washed with a cold (0-5°C) mixture of 1:1 isopropanol:water and the cake was washed with methylene chloride in an amount of 30 ml, yielding about 13.5 g or about 70% snow white amoxicillin trihydrate.

Example 6

108 g or 0.5 mol of 6-aminopenicillanic acid, 1.0 g or 0.017 mL of imidazole, 800 mL of dry methylene chloride, and 120 mL or 0.56 mol of 98% wt HMDS were stirred and heated to reflux for 3.3 hours. The reaction mixture was flushed with dry nitrogen gas during reflux to remove the NH 3 formed during the reaction. Then 2.0 ml or 0.016 mol of trimethylchlorosilane (TMCS) was added. Refluxing was continued with N 2 purging for an additional 19 hours, and then in the condenser sublimed NH 4 Cl was removed and 2.6 mL or 0.0306 mol of TMCS was added to the reaction. Refluxing with N2 purging continued for a further 34 hours. The volume of the reaction mixture was brought to 1000 ml with dry methylene chloride. NMR then showed 100% silylation of the amino and carboxyl group on the 6-aminopenicillanic acid. The solution was covered with N2 gas and kept like that for 9 days. NMR spectra confirmed the above and the stability. The solution was stirred and C02 bubbled through for approx. 90 minutes. The temperature was 20-22°C. NMR showed 100% conversion of bis-trimethylsilyl-6-aminopenicillanic acid to bis-trimethylsilylcarboxy-6-aminopenicillanic acid (SCA).

This base mixture was used for the acylation experiments described below. The chemical compound in this solution had the formula

NMR showed that the bis-trimethylsilylcarboxy-6-amlnopenicillanic acid was still stable after 9 days.

100 ml of the base mixture (SCA equivalent to 10.8 g of α-amino-pencilillanic acid, 0.05 mol) was stirred at 22°C and 8.0 g or 0.058 mol of TEA.HCl and 4.2 ml or 0 .06 mole of propylene oxide (see US-PS 3,741,959). Some TEA.HCl was precipitated. The mixture was stirred and cooled to +3°C. Then 15.5 g, or 0.55 mol of D-(-)-p-hydroxyphenylglycyl chloride-hydrochlorold-hemidioxane solvate with a purity of 75% was added in portions as follows:

After a further 70 minutes, approx. 50 ml of dry methylene chloride to the reaction mixture to reduce viscosity.

After another 160 minutes, a 2 ml sample was removed and added to 1.0 ml D2O. After centrifugation, NMR analysis of the aqueous phase showed approx. 6# non-acylated 6-amlnopenicillanic acid.

10 minutes later, the reaction mixture was transferred to a 600 ml beaker and the transfer was terminated by washing with 50 ml of methylene chloride. While stirring in an ice bath, 60 ml of cold deionized ice water was added to produce a solution of two phases without solids content and with a pH value of 1.0. 15.0 ml of "LA-1" type liquid anion exchange resin was added to the two-phase system under stirring and inoculation at a pH value of 2.0. A crystallization began. An additional 10.0 ml of "LA-1" was slowly added over 5 minutes. The pH value was 3.0. Then 0.15 g of NaBH4 was added. Then 5.0 ml of "LA-1" was added and the pH value was 4.5. Stirring was continued and 1.0 g of NaHSO, sodium bisulphite, in 4.0 ml of water was added dropwise. Then 10.0 ml of "LA-1" was added and the pH continued to rise. A total of 40 ml of "LA-1" was added and the final pH value was 5.6. Then 5 ml of acetone was added. At this point, 1.5 g of NaHSO 3 dissolved in 6.0 ml of water was added over 30 minutes. Stirring in an ice bath was continued. The precipitated product was collected by filtration and the cake was washed successively with 50 ml of methylene chloride, 40 ml of water, 100 ml of isopropyl alcohol-water in a ratio of 80:20 and 100 ml of methylene chloride. The cake was then dried at atmospheric pressure and 45°C to obtain 18.2 g of amoxicillin trihydrate, which represented a yield of 87% calculated for 6-aminopenicillanic acid, corrected for 1% for testing the overall yield was approx. 88%. "LA-1" liquid anion exchange resin is a mixture of secondary amines, each secondary amine having the formula wherein each R<1>, R<2> and R<3> is an aliphatic hydrocarbon group and where R<1>, R<2 > and R<3> in the aggregate contains from 11-14 carbon atoms, this special mixture of secondary amines which is sometimes referred to as "Liquid Amine Mixture No. I", is a clear amber liquid with the following physical properties: viscosity at 25°C equals 70 cPs; specific gravity at 20°C equal to 0.845; refractive index at 45°C equal to 1.467; distillation interval at 10 nm: up to 160°C - 4#, 160-210°C - 5*, 210-220°C - 755É, above 220°C - 17*.

Example 7

A methylene chloride solution (5.0 mL) of 0.54 g or 2.497 mmol of trimethylsilyl-6-trimethylsilyloxycarbonylaminopenicillinate was treated with 0.20 g or 1.45 mmol of triethylamine hydrochloride followed by 0.162 g or 2.75 mmol of propylene oxide at 25 "C. The mixture was stirred at 25"C for 20 minutes to facilitate dissolution of most of the triethylamine hydrochloride. 0.43 g or 2.75 mmol of phenoxyacetyl chloride was added dropwise at 25°C and the mixture was stirred at 25°C for 30 minutes. A sample was removed and analyzed by CMR analysis at 20.0 MHz. CMR (carbon-13 nuclear magnetic resonance) showed complete disappearance of the phenoxyacetyl chloride as well as the APA carbamate and the presence of penicillin V-trlmethylslylester. The presence of penicillin-V-trimethylsilyl ester was demonstrated by spectral comparison with an identical sample prepared by silylation of the free penicillin-V acid with triethylamine and trimethylchlorosilane. The yield was estimated from the CMR spectrum to be 85-9056.

In the same way, using the same molar amounts of reagents and the appropriate acid chloride, cloxacillin, dicloxacillin, staphcillin and nafcillin can be prepared. CMR data on these acylation mixtures suggest an extremely pure acylation mixture with yields estimated to be at least 85*.

Claims (2)

1. N-trimethylsilyloxycarbonyl-6-amino-penicillin intermediate for the preparation of a penicillin of the formula where R-CO is α-aminoarylacetyl or phenoxyacetyl, where aryl means phenyl, which is optionally substituted with an OH group, characterized in that the compound is dissolved in an anhydrous, organic solvent and that it has the formula where B means a cleavable ester protecting trimethylsilyl group. 2. Process for the preparation of an N-trimethylsilyloxycarbonyl-6-aminopenicillin intermediate according to claim 1 with the formula where B has the above meaning, characterized by adding dry carbon dioxide to a solution of an N-trimethylsilylpenicillin compound with the formula where B has the above meaning, in an anhydrous, inert organic solvent at a temperature within the range 0-100°C, until the carbonylation reaction is complete.