It was recalled that the conductivity of the Sn-doped material depended critically upon the amount of Sn-doping and the O partial pressure during preparation and annealing, and that Frank and Köstlin had rationalized the observed carrier concentration dependence by proposing the formation of 2 types of neutral defect clusters at medium Sn-doping levels. One was termed reducible, and the other non-reducible, to reflect their ability to create carriers under reduction. Both clusters were composed of one O interstitial and two Sn atoms which substituted for In and were positioned in non-nearest and nearest coordination, respectively. The present work sought to distinguish the reducible and non-reducible clusters by using an atomistic model. It revealed only a weak correlation of O interstitial binding energies with the relative positioning of Sn dopants. The number of Sn dopants in the vicinity of the interstitial had a much larger effect upon how strongly it was bound; a simple consequence of Coulomb interactions. It was proposed that O interstitials became non-reducible when clustered with 3 or more SnIn. This occurred at higher doping levels, as reducible clusters aggregated and shared Sn atoms. A simple probabilistic model could estimate the average number of clusters which so aggregated, and yielded a qualitatively correct description of the carrier density in reduced samples as a function of the Sn-doping level.

Defect Cluster Aggregation and Nonreducibility in Tin-Doped Indium Oxide. O.Warschkow, D.E.Ellis, G.B.González, T.O.Mason: Journal of the American Ceramic Society, 2003, 86[10], 1707-11