60th Annual Report on Research 2015

The first line of inquiry consisted of a convenient synthesis of a phosphatriptycene derivative suitable for elaboration at the aryl positions flanking the phosphorus. Here we have borrowed a key trick from the work of Sawamura and coworkers on 9-sila-10-phosphatriptycenes: the use of tris(2-lithiophenyl)phosphine, generated from tris(2-bromophenyl)phosphine, as the key intermediate. Note that we could not simply start with Sawamura's architecture, as we need the bridgehead atom in the 9-position to be smaller than phosphorus, to make the intended pocket around the phosphorus large enough to accept a metal atom. Therefore we closed the triptycene framework using the phosgene equivalent phenyl chloroformate. After protic workup, 9-hydroxy-10-phosphatriptycene is obtained in 57% crystallized yield, on multigram scale. For comparison, the elegant initial preparation of the parent phosphatriptycene by Bickelhaupt and coworkers in 1974 requires five steps from commercially available precursors, the last of which proceeds in 35% yield. The hydroxyl group, present for synthetic convenience, presents a synthetic challenge, in that it could direct metalation in positions adjacent to it, rather than to the phosphorus. We therefore protected this group as a triethylsilyl ether in 90% yield.

Before proceeding toward the desired supramolecular architecture, we prepared a copper(I) complex of the 9-siloxy-10-phosphatriptycene to measure key metrics by X-ray crystallography. The copper(I) chloride complex of this ligand crystallizes from a solution in dichloromethane / hexanes as a pentamer with bridging chlorides.

The constrained geometry of this ligand relative to triphenylphosphine is evident, for (Ph₃P)CuCl crystallizes as a chloride-bridged tetramer. Geometric calculations based on this structure confirm that a pocket about the phosphorus, created by replacement of flanking C–H bonds with C–O–aryl moieties, should accommodate first-row transition metals readily.

The elaboration of this ligand using directed metalation has proved frustrating. We knew of no example in which a phosphine itself acts as a directing group toward bases such as lithium amides or magnesium amides, but hoped that reagents based on zinc or copper, both of which have a greater affinity for phosphorus, would bind the phosphorus and deprotonate the nearest C–H bond. Bases such as (2,2,6,6-tetramethylpiperidyl)zinc chloride, lithium di(tert-butyl)(2,2,6,6-tetramethylpiperidyl) zincate, and (2,2,6,6-tetramethylpiperidyl)copper(I), however, showed no reaction with the 9-siloxy-10-phosphatriptycene. The metalation of phosphine oxides by very strong bases such as tert-butyllithium is known, and we were able to oxidize the phosphatriptycene to the corresponding phosphatriptycene oxide cleanly, but our substrate gave rise to a complex mixture of ring-opened products under these conditions. Less reactive magnesium amide bases, disappointingly, gave no reaction at all. In view of the affinity of zinc and copper for sulfur, we prepared the phosphatriptycene-P-sulfide and treated it with the zinc and copper amide reagents, but again observed no reaction.

We are currently investigating a modified strategy, in which the positions adjacent to phosphorus are already functionalized before the triptycene framework is assembled. The directed lithiation of 3-bromo-1-(N,N-diethylcarbamoyl)phenol by lithium diisopropylamide has been reported. We are attempting to use this aryllithium to prepare the corresponding triarylphosphine. At present, we have succeeded in placing two of these aryl groups on phosphorus, as judged by ³¹P NMR spectroscopy and mass spectrometry; the third P–C bond formation remains difficult, possibly due to steric encumbrance. We cannot heat the reaction to drive it to completion, as the aryllithium undergoes lithium bromide elimination to form a reactive benzyne. We remain confident that conditions can be found to assemble this new triarylphosphine. With it in hand, we could readily assemble a phosphatriptycene with flanking hydroxy groups, the arylation of which should be feasible.

Finally, we sought to synthesize 9-aza-10-phosphatriptycenes, believing that directed metalation of triarylamines could likewise offer a convenient route to an already-elaborated architecture, one that could be incorporated before the triptycene was assembled. We chose the methoxy substituent for a first pass. Tris(3-methoxyphenyl)amine is a known compound, readily prepared in high yield in a single step from commercially available materials by Buchwald-Hartwig coupling. A methoxy group is one of the best known directing groups for arene lithiation, and we knew that directed lithiation normally proceeds between two directing groups if they are present in a meta-arrangement. We did not know, however, whether the poorly basic triarylamine nitrogen could serve as the second directing group, to achieve the desired geometry. We learned that it does so quite effectively for a single metalation, but that attempted triple metalation gives rise to loss of selectivity.

Building on our findings in directed lithiation, however, we have achieved the synthesis of an acridone, not previously known, from a diarylamine. The use of a carbamate anion as a removable directing group was known for the single lithiation of N-alkylanilines, but we did not know whether it would work cleanly for a double lithiation. Using this route, we have developed a multigram-scale synthesis of a previously unknown acridone, and have gone on to elaborate this framework into what we hope will be a useful ligand. Although very different than the phosphatriptycenes we set out to make — this is a nitrogen-based s-donor with negligible p-accepting characteristics — this diarylated acridone offers an array of new directions as a ligand in its own right.