60th Annual Report on Research 2015

We have been working to identify catalyst modifications that will accelerate the cycloreversion step with less-strained olefin substrates.  In this regard, we have identified several aspects of catalyst design that we believe may offer increased rates (Figure 2).  Among these, we believe that angle strain is potentially the most powerful structural variable.  The general idea is to design catalysts that with larger C–N–N bond angles that will favor the all sp²-hybridized cycloreversion products (with preferred angles of ~120¼) while destabilizing the all sp³-hybridized cycloadducts (with preferred angles of ~109¼).

With this in mind, we took note that our original catalyst, [2.2.1]-bicyclic hydrazine 9, had a C–N–N angle of 104¼, clearly the wrong direction for favoring sp²-hybridization.  Notably, we prepared [2.2.2]-bicyclic hydrazine catalyst 10 with a calculated C–N–N angle of 111¼ and indeed observed a significant enhancement in rate for the cycloreversion step with a norbornene-derived cycloadduct.  Encouraged by this result, we have been pursuing new hydrazine catalysts with significantly expanded angles.  Two such designs include the tricycles 13 and 16 (Figures 3 and 4), which have calculated angles of 116¼ and 117¼ respectively.  Our targeted synthesis of 13 relies on the intermediacy of the known diketone 11, which can be prepared on large scale from 1,3-acetone dicarboxylate and oxalate esters.  Subsequent reductive amination provides ready access to diamine 12, and we are now working to identify conditions that will forge the key N–N bond. A similar strategy is expected to furnish the tricycle 16 from the known diketone 14 (Figure 4). 

Another catalyst system we are pursuing also possesses large C–N–N bond angle and, we speculate, may benefit from aromatic stabilization for the cycloreversion step.  Specifically, we are investigating benzocinnoline derivatives as catalysts for ring-opening metathesis reactions.  The general reaction design is shown in Figure 5, in which benzocinnoline (17) is alkylated by an alkyl halide 18 to produce, after deprotonation, the diazonium ylide 19.  (This ylide is identical to azomethine imine 20).  Importantly, the diaza-containing central ring is aromatic in intermediate 19 but not in the cycloadduct 22.  Because of this loss of aromaticity and large C–N–N angle, we speculate that intermediate 22 will be especially prone to undergo cycloreversion to the alternative diaza ylide 24 and the metathesized olefin 23. Subsequent protonation and Von Braun dealkylation would then release the new alkyl halide 25 and return benzocinnoline to the catalytic cycle. Our preliminary efforts have demonstrated the feasibility of the alkylation equilibrium as well as ylide formation, and we are currently working to identify conditions for efficient cycloaddition with intermediates 19. 

In addition to the above efforts, we have also worked on another approach to carbonyl-olefin metathesis using transition metal catalysis. To date, a transition metal-catalyzed carbonyl-olefin metathesis has proven elusive, mainly owing to the challenges associated with conversion of a metal oxo to the requisite metallooxetane species.

A proposed Chauvin-type catalytic cycle for a carbonyl-olefin metathesis is shown in Figure 6.  Thus a metal oxo species 26 engages an olefin substrate to form a metallooxetane 27. This intermediate undergoes cycloreversion to generate an intermediate 28 bearing a carbonyl and a metal alkylidene. The latter moiety proceeds to olefinate another carbonyl substrate via metallooxetane formation and cycloreversion, thereby producing the new olefin 29 and returning the metal oxo to the catalytic cycle.  Although the envisioned process is conceptually straightforward, and many examples of carbonyl olefination with metal alkylidenes are known, the real challenge lays in step A: the conversion of the metal oxo species to the metallooxetane. Nevertheless, evidence exists that certain metal oxo species can engage olefins to form metallooxetane intermediates. Particularly intriguing for our purposes was a recent report from Chen, who found that certain electron-deficient, four-coordinate rhenium oxo complexes are capable of catalyzing the ROMP of norbornene. These results clearly implicate the generation of a rhenium alkylidene species, which represents the first half of the catalytic cycle shown in Figure 6 (the generation of aldehyde functionality was not noted but is presumed). Since it is established that rhenium alkylidenes olefinate aldehydes, a plausible scenario exists in which rhenium oxo compounds might serve as viable carbonyl-olefin metathesis catalysts.

To test this hypothesis, we treated a mixture of norbornene and nitrobenzaldehyde with 10 mol% MeReO₃ and (C₆F₅)₃B and observed 20% of the ring-opening carbonyl-olefin metathesis product 30 (eq 1).  While this conversion is admittedly modest, the crucial point is that this result represents turnover of the metathesis catalyst – a feat that has not been reported in this area. We subsequently found increasingly electron-rich aldehydes to be significantly less reactive, and NMR studies reveal that this diminished reactivity is due to coordination of the Lewis acid by the aldehyde, leaving the Lewis acid unavailable to activate MeReO₃.  Presumably, if the rhenium catalyst could be rendered sufficiently activated, then Lewis acid activation of the catalyst would not be required, and the capacity of the rhenium catalyst to engage the olefin substrate would be dictated simply by the equilibrium of aldehyde coordination by the rhenium. Toward this end, we propose to synthesize and evaluate a series of electron-deficient aryl trioxorhenium complexes. Our studies will initially focus on compounds related to pentafluorophenyl trioxorhenium, which Chen showed to be an optimal catalyst for norbornene ROMP.