Nature Journal Club

Lynne B. McCusker

Laboratory of Crystallography, ETH Zurich, Switzerland

A crystallographer celebrates a method with niche applications.

In 2004, Oszlányi and Sü introduced a new way to determine crystal structures from diffraction data. To many crystallographers, including myself, this was a remarkable development. Although most of us had assumed that the trend of incremental but significant improvements to existing methods would continue, we had not expected a completely different approach to be discovered. The algorithm is an elegant one, based on a very simple perturbation (called charge flipping) of electron-density maps that are generated during the structure solution process.

Initially, the algorithm was viewed as a curiosity. After all, existing methods for solving structures work very well about 95% of the time, so a new technique was not really needed. However, the algorithm caught the attention of some inquisitive crystallographers, who tested it on their favourite problem cases. The result is that, just 4 years after its development, the approach has found niches in areas in which traditional methods flounder (Acta Cryst. A64, 123–134; 2008).

Scientists studying aperiodic materials (modulated structures and quasicrystals, whose structures are best described in more than three dimensions) were among the first to recognize the possibilities offered by the algorithm, because it could be easily adapted to work in higher dimensions. Charge flipping has enjoyed great success with such structures, and is now considered the method of choice by this community.

The algorithm has also proved effective in solving the structures of polycrystalline materials, mainly because complementary information from other sources (such as chemical analysis and electron microscopy) can be easily included. Now small protein structures and neutron- and electron-diffraction data are being explored — no doubt further niches will be found.


Comments are closed.