Errors and Pitfalls

Abstract

When a crystal structure has been successfully solved and refined by the methods described here, the resulting information is generally of a high level of accuracy which can only very exceptionally be obtained by spectroscopic methods. The indirect nature of the structure determination, depending as it does on the acceptance of a structure model, can, however, sometimes lead to serious errors. Some very nasty traps lie in wait for those who never look carefully at photographic or area detector data, and whose unit cells and space groups are determined by automatic diffractometers and are not critically considered! It is, of course virtually unthinkable that a completely nonsensical structure model could actually yield calculated F _c-values for several thousand reflections which agreed with the observed F _o-values within a few percent. On the other hand, there certainly are situations where the structure model largely reflects reality, and gives a credible R-factor although some small, but possibly important, parts of the structure are wrong. These can arise, for example, when the refinement converges not to the global minimum but to a local false minimum. In such a case, a solution has been obtained which contains errors which prevent the true structure being attained by further refinement in the normal way. There are other reasons for this to occur than a wrong choice of unit cell of space group, and specific cases of these will be examined in the following sections.