New light on an old problem: defects and nonstoichiometry

That compounds have definite compositions is taken as a matter of faith and yet there are several inorganic solids which exhibit a wide range of compositions or show no simple correspondence between the composition and the detailed structure (or chemical identity). It has been known since the 1920s that stoichiometric FeO1.00 does not fall in the stability range of iron (II) oxide (FeO1.05–FeO1.15). Point defects in crystals such as vacancies and interstitials first described by Schottky, Frenkel and Wagner account for the transport properties of ionic solids, but there are serious difficulties in applying the point-defect formalism to solids possessing a wide stoichiometric range or to those solids exhibiting ordering of defects or extended defects (such as crystallographic shear planes). Although there is no clear-cut transition between the point-defect regime and the regime of highly ordered structural imperfections in nonstoichiometric solids, we can certainly say that the point-defect model is really valid only when the defect concentration (or the deviation from stoichiometry) is extremely small. Only in such dilute point-defect systems can one satisfactorily relate the electronic properties and the nonstoichiometry to the concentration of point defects.