We calculated the optical excitation and thermal ionization energies of oxygen vacancies in mHfO2 using atomic basis sets, a non-local density functional and periodic supercell. The thermal ionization energies of negatively charged V − and V 2− centres are consistent with values obtained by the electrical measurements. The results suggest that negative oxygen vacancies are the likely candidates for intrinsic electron traps in the hafnum-based gate stack devices.Hafnium based oxides are currently considered as a practical solution satisfying stringent criteria for integration of high-k materials in the devices in future technology nodes. However, high-k transistor performance is often affected by high and unstable threshold potential 1 , V t , and low carrier mobility 2 . These effects are usually attributed to a high concentration of charge traps and scattering centers in the bulk of the dielectric and/or at its interface with the silicon channel. Although the reported trap densities vary greatly with fabrication techniques, the majority of data point to existence of a specific intrinsic shallow electron centre common to all HfO 2 based stacks while some extrinsic defects, such as Zr substitution in HfO 2 , have also been considered 3 .Oxygen vacancies are dominating intrinsic defects in the bulk of many transition metal oxides including HfO 2 and ZrO 2 , and are thought to be also present in high concentrations in thin films. However, in spite of numerous experimental studies, evidence relating oxygen vacancies to measured characteristics of interface traps in high-k stacks is still mostly circumstantial. Therefore, accurate theoretical characterization of these defects is highly desirable.Previous theoretical calculations of oxygen vacancy in HfO 2 and ZrO 2 reported the ground state properties obtained within local or semilocal approximations to density functional theory (DFT) methods (see 4,5 for a review). This approach, however, significantly underestimates band gaps, which hampers determining energies of defect levels with respect to the band edges and precludes identifying shallow defect states 5,6,7 . As a result, most of the early local DFT calculations (except, perhaps, ref. 8 ) failed to predict unambiguosly negative charge states of oxygen vacancy in HfO 2 . Significant improvement was achieved by Robertson et al. 4,9,10 who used screened exchange approximation to predict vacancy energy levels including V − charge state. However, these calculations were performed using a small periodic supercell and therefore corresponded to extremely high vacancy concentrations. The quality of the functional used is also largely unknown and needs independent verification.In this work we used much bigger supercells and a nonlocal functional to calculate optical excitation and thermal ionization energies of oxygen vacancies in five charge states. To relate these energies to experimental data we distinguish optical absorption/reflection type measurements involving Frank-Condon type (vertical) excitations, and electric...

The earliest ideas of the polaron recognized that the coupling of an electron to ionic vibrations would affect its apparent mass and could effectively immobilize the carrier (self-trapping). We discuss how these basic ideas have been generalized to recognize new materials and new phenomena. First, there is an interplay between self-trapping and trapping associated with defects or with fluctuations in an amorphous solid. In high dielectric constant oxides, like HfO 2 , this leads to oxygen vacancies having as many as five charge states. In colossal magnetoresistance manganites, this interplay makes possible the scanning tunnelling microscopy (STM) observation of polarons. Second, excitons can self-trap and, by doing so, localize energy in ways that can modify the material properties. Third, new materials introduce new features, with polaron-related ideas emerging for uranium dioxide, gate dielectric oxides, Jahn-Teller systems, semiconducting polymers and biological systems. The phonon modes that initiate self-trapping can be quite different from the longitudinal optic modes usually assumed to dominate. Fourth, there are new phenomena, like possible magnetism in simple oxides, or with the evolution of short-lived polarons, like muons or excitons. The central idea remains that of a particle whose properties are modified by polarizing or deforming its host solid, sometimes profoundly. However, some of the simpler standard assumptions can give a limited, indeed misleading, description of real systems, with qualitative inconsistencies. We discuss representative cases for which theory and experiment can be compared in detail.

No abstract