Sixteen OCH 2 CH 2 N 3 P(i-BuNCH 2 CH 2 ) 3 N substituents (A) containing the highly basic bicyclic azido-phosphine moiety shown, have been incorporated into the dendrimer [CH 2 Keywords: catalyst; dendrimers; organic catalysis; phosphine imine; proazaphosphatrane Dendrimer synthesis has fascinated chemists for more than a decade.[1] Phosphorus-containing dendrimers make up an active subarea [2] which have materials, biological and catalysis applications.[3] To our knowledge, however, no reports of catalytic reactions promoted by phosphorus-containing dendrimers have appeared in which the phosphorus-containing moiety itself functions as the active site. We disclose here the first such example and show that it is an efficient catalyst for useful organic transformations such as Michael addition, aldol condensation and isocyanate trimerization. Acyclic phosphine imines [RN ¼ P(NR 2 ) 3 ] constitute a class of strong nonionic bases [4] wherein C 6 H 5 N ¼ P(NMe 2 ) 3 , for example, displays a pK a of 20.9 in acetonitrile.[5] Such phosphorus imines have been found to be very useful in amino acid synthesis. [6,7] We have found that bicyclic 1a is a powerful catalyst for acylation, [8] Michael addition [9] and isocyanate trimerization.[10] We were thus attracted to the possibility of grafting such a bicyclic base to a dendrimeric framework. The synthetic approach we report here is based upon a previous study [11] in which we found that 1a is stable in the azide form [11a] whereas the analogous reaction with phenyl azide gives the iminophosphorane 1b.[11b]The strategy we chose for assembling the dendrimeric framework is shown in Scheme 1. In order to realize a substantial number of basic sites in the molecule, we decided to employ the diazido acid 6 to effect branching. Methyl 3,5-dihydroxybenzoate 3 on treatment with dibromoethane in the presence of 18-crown-6 and K 2 CO 3 gave the dibromoester 4 in 87% yield.Although a few procedures have been reported [12] for the conversion of a meta-dihydroxybenzene (namely, 1-benzyloxy-3,5-dihydroxybenzene) to the corresponding 3,5-bis(2-bromoethyloxy) derivative, none of them afforded the 3,5-bis(2-bromoethyloxy) derivative 4 as a clean product. We therefore developed a modified procedure in which dibromoethane is used as the solvent in the presence of finely powdered K 2 CO 3 to give 4, which was pure by 1 H and 13 C NMR spectroscopy. Conversion of 4 to 5 in 93% yield, followed by saponification and acidification gave diazido acid 6 in 90% yield, which was then condensed with DAB-Am-8 dendrimer to afford 7 in 35% yield. The hexadeca-azido dendrimer 7 was found to be pure by NMR spectroscopy, although its polydispersity value (M w /M n ¼ 1.159 versus a polystyrene standard) obtained by GPC analysis was not ideal (1.00), presumably due to aggregation that occurred in the carrier solvent (THF).[13] Molecular weights estimated by GPC relative to a polystyrene standard depend on the shape and flexibility of the molecules and were considerably lower than calculated values,...