Computational Alchemy: The Search for New Superhard Materials

Publisher Website
Google Scholar
Abstract
A central challenge to modern materials science is the rational design and synthesis of new materials possessing exceptional properties. Recent advances in first-principles modeling methods and the availability of increasingly powerful computational resources make this goal increasingly achievable. The strength of these modeling methods lies in their predictive ability. They are able to reproduce the crystal structures and elastic properties of a large class of materials to within 2–3% of experimental values and have predicted a number of phase transitions that have been verified experimentally.Despite the power of these methods, the process of designing materials from first principles is not usually a straight-forward or simple one. It requires overcoming a number of obstacles, some of them quite formidable. First a calculable figure of merit that correlates well with the desired property must be identified. While this may be straightforward in some cases, in others—such as predicting the ability of a material to isolate radionuclides over million-year time scales—the process of reducing complex properties to a few calculable variables can be rather difficult. Next a promising chemical system and a realistic set of crystal structures must be selected. This is not trivial because predicting the structures that can crystallize in a given system can be exceedingly challenging. However a wide variety of methods are available to aid in the generation of promising structures — comparative crystallography, algorithms based upon the concepts of crystalline nets and close packing, modern alloy theory methods, and simulated annealing strategies being some examples.