Because superconductors can conduct electricity without energy loss, the potential technological benefits of materials such as YBCO are enormous. However the mechanism responsible for high-temperature superconductivity is not yet fully understood. Such an understanding is crucial if we are ever to create new materials with improved properties (e.g. superconducting transitions nearer room temperature). Some theories have linked lattice distortions to the phenomenon, but these are not always easy to observe experimentally.
Using neutron total scattering and reverse Monte Carlo (RMC) analysis, we have investigated the structure of YBa2Cu3O6.93 (YBCO), the canonical high-temperature superconductor. Of interest is a specific copper–oxygen bond length, thought to be involved in charge transfer between Cu/O chains and planes. Historically there has been some disagreement between average structure methods and local structure techniques such as EXAFS: the average structure shows only a single Cu–O bond length while the local structure suggests two distinct bond lengths. The large box models produced by RMC refinements are fitted to both local structure data and average structure data simultaneously, so are consistent with both data sets. Our results indicate that previous local structure studies were correct in identifying two bond lengths, but that the atomic displacements are correlated such that the atomic sites are unimodal — hence the apparent single Cu–O bond length in the average structure.
Access to large atomic-scale models of YBCO allows us to look for correlations beyond individual bond distortions. An initial exploration of spatial correlations between short bonds revealed the slightest hint at a preference for clustering into pairs. We have much more work to do to determine the validity of this picture, but the work will hopefully be worth the wait: if accompanied by charge localisation then the clustering we see could be strong experimental evidence for a bipolaron model.