Roald Hoffmann at Cornell University, US,  tells us why boron carbide is non-stoichiometric and on the new method he and his colleagues are developing to study such solids. 

Please explain, for a non-specialist, the significance of your article.. 

Boron carbide has always presented a real puzzle in its composition and properties. Although it is usually identified as B4C, it is always short of carbon to a variable degree. Chemists have an antipathy to non-stoichiometry, so generally this is imagined to be due to some inefficiency in the synthesis of the material. We find that the non-stoichiometry is essential; it follows from boron carbide's need to expel some electrons from its frontier orbital region so as to attain higher stability. We found a phenomenal strengthening of pi-bonding in the structure at the expense of sigma-bonding to meet the spatial constraints enforced on the C-B-C chains. We also found that classical chemical constructs such as steric interactions between bulky substituents (in this case B12) and exo-polyhedral multiple bonding are useful in understanding the bonding, structure and properties of boron carbide.  

What has motivated you to conduct this work?

In different ways, myself and my colleagues has been fascinated with polyhedral boron hydrides; they were the subject of my Ph.D. thesis at Harvard a mere 45 years ago, and figured importantly in Pancharatna and Balakrishnarajan's Ph.D. research at the University of Hyderabad in India years later. Recently, all three of us have looked upon the analogy between 2-D aromatic organic compounds and 3-D aromatic boranes and extended the concept of exocyclic double bonds to clusters. A search for its presence in icosahedral borides led us to boron carbide, which is infinitely attractive due to the plethora of mysteries around its structure and bonding, coupled with its technological importance. 

Where do you see this work developing in the future? 

Although cluster bonding may seem complicated because it is delocalized, it is reasonably well understood and has predictive value in molecular domains. But when it comes to extended structures, the wealth of knowledge gained on molecules remains to be effectively used. We think that a molecular orbital approach, coupling knowledge of discrete cluster chemistry and the methods of solid state physics for extended structures, can lead to the design of new materials.

Are there any particular challenges facing future research in this area? 

Compared to computational molecular chemistry, computational materials science is only at its infancy. We have many challenges, both with respect to prediction and analysis. The two go hand-in-hand; simple calculation/simulation does not by itself provide understanding and concepts. Our approach, which associates molecules, or molecular fragments, with materials and builds up the latter from molecular building blocks, provides a comprehensive way to look at bonding and properties in extended materials.