773Aldol reactions between N-silylated or 2,3-disubstituted J>sultams and uti> nesor aldehydes resultin the C-4aldolderivatives 2 and 6. Desilylation of 2 with TBAFyieldstheN-unsubstituted compounds 3. Mixtures of diastereomersobtained from reactions withaldehydes are separated by ce, and their structures areelucidated by spectroscopic data.Carbapenems, as thienamycin, differ from penicillins and cephalosporins by their characteristic a-hydroxyethyl side chain, which is believed to be responsible for their great antibacterial activity and their stability against~-lacta mases!', As 1,2-thiazetidine l,l-dioxides (~-sultams) are highly reactive sulfon analogues of !3-1actams, it is of special interest to introduce the hydroxyalkyl side chain into position 4 of the~sultam ring. Here, we wish to report about our results using N-substituted~-sultams as model compounds for the introduction of the a-hydroxyalkyl function by aldol reactions with ketones and aldehydes.Refering to a procedure from Dursr-), the N-substituted -sultams la,b were allowed to react at -78°C with the secondfold amount of LOA, and acetone or benzophenone. We isolated the aldols 2a 3 ), 2b, and 2c as stable, colorless, crystalline compounds. The highest yields were obtained, when the freshly prepared LOA reacts not longer than 30 sec with the~-sultam, and the overall reaction time does not exceed 5 min.The structure of 2 is confumedby their IR-and tH_NMR spectra,which showa strongabsorption of the hydroxy grouparound 3500em", the two absorptions of the S~group at 1290-1275 and 1140em", and the typical ABXpatternof 4-monosubstituted J>sultams.Desilylation of 2b,c was easily done by treatment with TBAF on silica gel in ethanol at room temp. yielding the desilylated products 3b and 3c with yields >80%, which can be stored in vacuofor some weeks without decomposition.The reaction with aromatic aldehydes was first studied with benzaldehyde, which was reacted with equimolar amounts of lb and BuLL After work-up, we found a compound, C17H2sN204S2Si, mlz = 416 (M+-), m.p, 128°C (methanol), which obviously was not the expected aldol. From the analytical and spectral data the structure 4 results, which might be formed by dimerisationt' of the anion of Ib and reaction of the N,N'-bis-silylated intermediate with benzaldehyde followed by protonation during work-up. To avoid this reaction. we used 2 equivalents of LOA and a slight excess of the aromatic aldehyde. By this procedure, the aldols 2d-g were obtained from lb. All aldols were isolated as diastereomeric mixtures, which were separated by CC (silica gel). The (oR·, 4S·)-isomersS) (A) were first eluated with dichloromethane, the (~, 4S·)-isomers (B) we obtained by subsequent extraction with ethyl acetate. The ratio of isomers was always around 1:1. The rield of the aldol 2g from acetaldehyde was maximal 15%6. When the a,~-unsaturated mesityl oxide was used as an electrophil, only the 1,2-addition products 2h were isolated. We did not find any 1,4-addition product, which is in agreement with other experiments...