Structure prediction and targeted synthesis: A new $NanN2$ diazenide crystalline structure

Significant progress in theoretical and computational techniques for predicting stable crystal structures has recently begun to stimulate targeted synthesis of such predicted structures. Using a global space-group optimization (GSGO) approach that locates ground-state structures and stable stoichiometries from first-principles energy functionals by objectively starting from randomly selected lattice vectors and random atomic positions, we predict the first alkali diazenide compound $NanN2$⁠, manifesting homopolar N–N bonds. The previously predicted $Na3N$ structure manifests only heteropolar Na–N bonds and has positive formation enthalpy. It was calculated based on local Hartree–Fock relaxation of a fixed-structure type (⁠$Li3P$-type) found by searching an electrostatic point-ion model. Synthesis attempts of this positive $ΔH$ compound using activated nitrogen yielded another structure (anti-$ReO3$-type). The currently predicted (negative formation enthalpy) diazenide $Na2N2$ completes the series of previously known $BaN2$ and $SrN2$ diazenides where the metal sublattice transfers charge into the empty $N2$ $Πg$ orbital. This points to a new class of alkali nitrides with fundamentally different bonding, i.e., homopolar rather than heteropolar bonds and, at the same time, illustrates some of the crucial subtleties and pitfalls involved in structure predictions versus planned synthesis. Attempts at synthesis of the stable $Na2N2$ predicted here will be interesting.