Materials scientists often face the problem of understanding why the atoms in their products arrange themselves in the way they do in crystalline solids. Phase homologies offer the occasional clue allowing rules to be established to predict where the next atom might be placed knowing where they sit in closely related compounds. Unfortunately, half a century of rule making has not had a big impact on synthesizing designer materials.

Researchers at Iowa State University, Ames, USA, and their colleagues have turned to information entropy to allow them to data mine bonding parameters in a database of inorganic compounds. They have then coupled the statistics thus extracted together with first-principles calculations to develop a new tool for predicting crystal structures more convincingly than was possible previously.

• Information-Theoretic Approach for the Discovery of Design Rules for Crystal Chemistry,
  Chang Sun Kong, Wei Luo, Sergiu Arapan, Pierre Villars, Shuichi Iwata, Rajeev Ahuja, Krishna Rajan,
  J. Chem. Inf. Model. 2012.
  DOI: 10.1021/ci200628z