Diberyllium or Be2 is is not supposed to exist yet pursued by scientists for over 70 years with over 100 published papers. Recently Merrit et al. (DOI) have recorded new experimental evidence for the molecule but prefer to call it the beryllium dimer.
A simple MO diagram explains why Be2 must be indefinitely unstable. When the two beryllium 2s orbitals interact they split in two energy levels - one bonding (lower energy) and one antibonding (higher energy) - and when the four electrons of both Beryllium atoms are divided over these two new molecular orbitals both levels get fully occupied and there is no net energy gain and therefore no bond. Many inprobable diatomic molecules spend their lives in obscurity. For some time in Wikipedia dilithium was called dilithium (real) because the actual dilithium page was devoted to dilithium: the fictional Star Trek chemical element.
The physics and experimental setup as always in this type of research is fuzzy, the Meritt work is no exception. Creating a beryllium gas in the first place is difficult due to the high melting point of the metal, the ease of oxidation and the toxicity of beryllium oxides. It is created by pulsed laser vaporation - must be something like pulsed laser deposition but without the deposition - and cooled by so-called pulsed supersonic expansion of the helium carrier gas into a vacuum. Two more lasers (one called the pump the other one the dump) are then pointed at this gas and resonance fluorescence decay from excited beryllium dimers is somehow measured at right angles of the laser beam.
After a lot of data processing the end result is a Morse potential graph describing the dimer as it snugly sits in an energy well with the two beryllium atoms separated by 2.5 angstrom, hence a true chemical bond. The precise shape of this graph of course defies all theory and is attributed to various stages of orbital hybridization and will no doubt attract the attention of theoretical chemists for years to come.