The control of defects, particularly impurities, to tune the concentrations of electrons and holes is of utmost importance in the use of semiconductor materials. To estimate the amount of dopant that can be added to a semiconductor without precipitating secondary phases, a detailed phase diagram is needed. The ability of ab initio computational methods to predict defect stability can greatly accelerate the discovery of new semiconductors by calculating phase diagrams when time-consuming experimental ones are not available. DFT defect energy calculations are particularly successful in identifying doping strategies by determining the energy of multiple defect charge states in large band gap semiconductors and insulators. In metals, detailed phase diagrams can be determined from such calculations but only one, uncharged defect is needed. In this work, we have calculated dopant solubilities of Br and Na in the thermoelectric material PbSe by mapping its solvus boundaries in different regions of the respective ternary phase diagrams using DFT defect energy calculations. The narrow gap PbSe provides an example where defects with nominal charge state (based on valence counting) have properly-localized charge states. However, defects with unexpected charge states produce delocalized electrons, which are then, in effect, defects with the expected charge state. Simply applying the methods for calculating multiple defect charge states in PbSe and treating them as separate defects fails to predict properties measured by experiments. Performing thermodynamic calculations using only the expected charge states, excluding others, enables accurate prediction of experimentally measured doping efficiencies and phase diagrams. Identifying which defect charge states to include in thermodynamic calculations will expedite the use of such calculations for other semiconductors in understanding phase diagrams and devising effective doping strategies.
ASJC Scopus subject areas
- Materials Chemistry