Remember the good old days when carbon atoms could bond together by not more than a bond order of three as in acetylene?. Sason Shaik et al. in this month's Nature chemistry argue that for dicarbon it can be four as well giving C2 a quadruple bond (DOI).
In the conventional scheme of things each of the carbon atoms in the dicarbon MO diagram is bringing a total of 6 electrons to the table. The 1s and 2s electrons combine and fill out both bonding and antibonding MO's and therefore do not contribute to bonding. The 2p electrons combine and fill out two degenerate pi MO's making the bond order 2. That is an unusual case of 2 pi bonds without any sigma bonds but with two sigma lone pairs. In an alternative picture C2 is a diradical with bond order 3.
In the new calculations Shaik arrives at a bond order of 4 with an energy of the fourth bond of between 11 and 14 kcal/mole compared to 100 kcal/mole for the sigma bond and 95 kcal/mole for the pi bonds. Valence bond theory and MO theory support each other on this one. This bond is called an inverted bond because just as in tetrahedrane the pi-shaped lobes are actually pointing away from each other.
Realistic? Just how the electrons in this bond are supposed to interact is unclear. The article hints at an experimental confirmation for the energy involved from spectroscopy but a review is cited and not a research article. As far as future experimental evidence Shaik suggests an investigation into the radical properties.
C2 by the way is just one in a whole sick list of troublesome molecules. In the accompanying editorial Jörg Grunenberg explains why. Compared to their colleagues of twenty years ago, these days computational chemists have all the computing power they need.