Molecules take very intriguing forms. You might know cubane, the literal molecular cube. Now, scientists recently reported the synthesis and characterization of 1-azahomocubane, one of the first of hopefully many more cubane analogues. Similar to weird species like Dewar benzene, these molecules can offer chemists new insights into effects and limits of ring strain.
Today we’ll start with a truly elegant oldschool synthesis of simple cubane, and then dive into azahomocubane’s modern synthesis and properties. Doing so, we also learn: 1) how people make the absolutely mad octanitrocubane; 2) how chemists can use the unstable antiaromatic cyclobutadiene as a reaction partner; 3) why azide groups are very useful for rearrangement reactions and last, 4) why you should never throw away your old products.
Hard to believe, but cubane was first synthesized already in 1964. Philip Eaton, part of the first cubane synthesis, continued to be a driving force of cubane research. Almost 4 decades years later, he also led the first synthesis of octanitro-cubane. As you can imagine, this required some unpleasant reaction conditions – and of course, this thing goes boom. The introduction of nitro-groups worked via so-called interfacial nitration of a cubyl anion at the melting interface of frozen THF and N2O4. I’ll gladly pass on this. Also, the last step requires you to think: “mhm ah yes, let’s do an ozonolysis of a nitrosyl-heptanitro-cubane”.
Octanitrocubane was hypothesized to a potentially best-in-class non-nuclear explosive based on theory – but its experimental density was shown to lower. There are no public records of larger-scale synthesis and testing, so it’s just researched from a computational point and remains elusive.
Back to normale cubane – mindblowingly, just after 2 years, an incredibly efficient synthesis was published. As promised, it uses cyclobutadiene as a starting material, which seems crazy but also very logical given cubane is full of squares. You will know that this is an anti-aromatic compound and very unstable, so it is not possible to store it in a bottle. However, oxidizing this iron complex, which can be handled better, leads to release of cyclobutadiene within the reaction vessel – in this case, allowing for a [4+2] cycloaddition with this diketone which bears two bromo groups and another olefin – both essential to this efficient synthesis.
After the endo cycloaddition, the two double bonds are well positioned to engage in a [2+2] photoaddition, obviously one of the key reactions available for construction of cyclobutane rings. Now we have this basket-like intermediate, which can undergo a base-mediated double quasi-Favorskii reaction, leading to ring contraction of the cyclopentane rings. This creates the cubane system, and now you can simply remove the acid groups in some step via decarboxylation. The yields are suspiciously consistent at 80% but even if they would be lower, it would not take away from the nice sequence and use of cyclobutadiene.
Very timely, there was a super recent publication on another, metal-free way to liberate cyclobutadiene – quite nice work and much better than creating some toxic iron compounds. This works via a retro [2+2] addition to release nitrogen and create reactive cyclobutadiene. [J. Am. Chem. Soc. 2023, 145, 10, 5631]
So the little cubane boy is done – but how do we create the much more complicated azahomocubane? Well, a logical intermediate would be a simple cubane-amine. Because it’s not 1964 anymore, we don’t have to create cubanes bottom-up, rather, we can simply buy them because there are crazy folks performing cubane synthesis on kilo scale.
This di-ester-cubane is not cheap and although it comes with the cubane, it also has two esters. Because we need the mono-substituted amino-cubane, we need to get rid of one of them. This is achieved via hydrolysis with just one equivalent of hydroxide and subsequent Barton-decarboxylation of the activated acid.
So, how do we get the amine from the ester? After hydrolysis, the acid was converted to the azyl azide. This prepares the Curtius-rearrangement which is facilitated by the strong energy gain of N2 release. The intermediate isocyanate can be trapped with various nucleophiles, in this case, with tert-butanol to give the N-Boc carbamate product.
Simple acidic exposure releases the cubyl amine. What now? It’s azide time again! Exposure to triflyl azide, a terrifyingly reactive azide transfer reaction, creates cubyl azide via substitution. Both of these azides come with a hefty explosion warning, so they are best handled only in solution and behind a blast shield. Adding some acid now, you guessed it, leads to another Curtius-like rearrangement with migration of the alkyl rest to displace N2. The resulting carbocation is then trapped by the acetic acid to give this product. After all of this, we now have a single nitrogen in the correct oxidation state.
Because we need the nitrogen in a corner position, there is more to be done. Once again, let’s do a sweet rearrangement. N-chlorination creates the opportunity for a new C-N bond formation, initiated by the fragmentation of the acetate. This is reminiscent of the Favorskii ring contraction we saw in normal cubane’s synthesis, just this time, the halogen leaving group is bound to nitrogen. The intermediary amide can be hydrolyzed and voila, we finally positioned nitrogen in a cyclobutane. Now, we just need to create the final linkage of the extra carbon and nitrogen. We have the ester carbon to work with, so the final steps used the methyl ester. Now it’s just a simple orgo freshman sequence of ester reduction, chlorination and intramolecular SN2 to get to aza-homocubane. As this final product is volatile, they prepared the salt form for easier handling.
So does azahomocubane go boom like octanitrocubane? No, clearly not – just because it looks strained, does not it can release nitrogen or CO2 as part of an explosion. It’s decomposition is much less exciting actually, exposure to acid or simple storage in the refrigerator led to ring opening of the cubane. This step was irreversible, a testament to the high instability of azahomocubane. It also makes you think that the last SN2 with the primary leaving group is probably one of the only reasonable ways to create the system in the first place.
In terms of geometry, the team obviously expected the product to be different than your generic tertiary amine. DFT calculations indicated that this was definitely less than ideal sp3 geometry, and the crystal structure – funnily found via serendipitous discovery – was consistent with this. You can see that the five-membered ring distorts the picture quite a bit, so homocubanes look more like baskets or houses, instead of cubes – if that makes sense.
What is the impact on basicity? The 1-azahomocubane nitrogen is not happy with strain energy being an order of magnitude higher – but this means that basicity is more than 10-fold lower! You might think that because there is more strain, the nitrogen might be happier to be present in some other configuration. However, looking nitrogen NMR chemical shift analysis showed that the nitrogen in azahomocubane is less electron rich compared to the other frameworks, which is aligned with the basicity trend. Notably, the hypothetical azacubane has 45 kcal/mol more strain energy than azahomocubane, which makes you wonder how long it will take to synthesize it in the lab.
Finally, they looked into hypothetical atom exchange and hypohomodesmotic reaction calculations. What is that? It’s basically a nerdy theory-crafting method of computing bond separation and formation energies, and the name is due to different sets of reaction conditions qualifying for different reaction types. For example, the hypohomodesmotic reactions shown here can be used to compute strain in cycloalkanes. Basically, you’re taking heats of formation of the individual molecules – which are known values – and figure out what is the enthalpy value Q needed to balance out the hypothetical reaction. You can see that while cyclohexane has a Q value of almost zero – because it is not strained at all – Q is much higher for cyclopentane, cyclobutane and cyclopropane as strain increases.
Applying this methodology to our question, we see that azahomocubane is significantly more stable than all-carbon homocubane. But why is that?
To shed some light on this, they modelled nitrogen lone-pair sigma star interactions with the carbon framework. This revealed that there is substantial hyperconjugation from the nitrogen with both adjacent cubic and basket handle C-C bonds, somewhat stabilizing the system compared to a normal homocubane. Both hyperconjugative effects and general orbital re-configuration are at play. For example, Bent’s rule describes orbital re-orientation once you add in electronegative substituents like fluorine or nitrogen. In such settings, orbitals with s character are pointed towards more electropositive substituents like hydrogen or carbon, so this leads to some changes in bond geometry and lengths.
This work doesn’t answer all our questions, but it shows that aza-variants of strained systems will be interesting playground for chemists. Catch you in the next one!
- Azahomocubane | Chem. Sci., 2023,14, 2821 https://pubs.rsc.org/en/content/articlelanding/2023/SC/D3SC00001J
- Cubanes in Medicinal Chemistry: Synthesis of Functionalized Building Blocks https://pubs.acs.org/doi/10.1021/ol501750k | Org. Lett. 2014, 16, 16, 4094
- Hepta- and Octanitrocubanes: https://doi.org/10.1002/(SICI)1521-3773(20000117)39:2%3C401::AID-ANIE401%3E3.0.CO;2-P | Angew. Chem. Int. Ed. 2000, 39, 401