Finding the reconstructions of semiconductor surfaces via a genetic algorithm

In this article we show that the reconstructions of semiconductor surfaces can be determined using a genetic procedure. Coupled with highly optimized interatomic potentials, the present approach represents an efficient tool for finding and sorting good structural candidates for further electronic structure calculations and comparison with scanning tunneling microscope (STM) images. We illustrate the method for the case of Si(1 0 5), and build a database of structures that includes the previously found low-energy models, as well as a number of novel configurations.