ABSTRACT
CONTENTS 9.1 Introduction ........................................................................................................................ 283 9.2 Ionic Oxides and Defect Densities................................................................................... 284 9.3 Bulk Band Structures......................................................................................................... 284 9.4 Defect Types and Their Formation Energies ................................................................. 285 9.5 Defect Energy Levels......................................................................................................... 286
9.5.1 Oxygen Vacancy ..................................................................................................... 286 9.5.2 Oxygen Interstitial .................................................................................................. 290 9.5.3 Defects in Hf Silicate .............................................................................................. 291
9.6 Discussion ........................................................................................................................... 294 9.7 Interface Defects................................................................................................................. 297 9.8 Defect Passivation.............................................................................................................. 298 9.9 Summary............................................................................................................................. 301 References.................................................................................................................................... 301
The scaling of complementary metal oxide semiconductor (CMOS) transistors is leading to the replacement of SiO2 gate oxide by oxides of higher dielectric constant (K) such as HfO2 to avoid excessive gate leakage currents. The leading candidates are presently HfO2, Hf silicate, and their nitrogenated alloys [1-5]. They will be introduced at the 45 nm node. Their use has required a wide-ranging optimization of these materials. There are three main problems with devices using high-K gate oxides. First, there is a
large fast transient charge trapping, which causes a shift in the transistor threshold voltages, instability, and ultimately a loss of reliability [5-9]. The problem is a band of charge traps in the bulk oxide that lie at an energy near the Si conduction band edge [6,8]. The second problem is a degradation of the Si carrier mobility, which is lower than in the equivalent SiO2 gate oxide devices, particularly for negative MOS (NMOS) [5,10-13]. This is partly due to remote phonon scattering by soft phonons [11], and partly due to a remote Coulombic scattering by trapped charge in the oxide. The third problem is controlling the
gate This complex effect may be due to dipoles and due to charged vacancies [15]. This indicates a critical need to identify the nature of point defects present in the oxides and to find their energy levels. The oxides of interest are oxides of early transition metals or rare earths, with fully
ionized d shells, so that they have a large gap of order 6 eV. They were chosen to be thermodynamically stable in contact with Si and to have a large conduction band offset with Si [3,4]. In general, we consider HfO2, ZrO2, La2O3, Y2O3, and Gd2O3. The most common defect in such oxides is the oxygen vacancy. This class of oxides, especially CeO2 and ZrO2, is also of interest as fast ion conductors and as catalyst supports, such as in the three-way catalyst in car exhausts, where oxygen deficiency plays an important role. Thus, we focus particularly on the oxygen vacancy. This chapter first reviews calculations of the electronic structure and energy levels of the
defects. It then considers experimental data on the energy levels from optical and electrical studies, which support the importance of the oxygen vacancy. The interesting fact is that the traps are electron traps, meaning that an electron becomes trapped at an O vacancy. This is counterintuitive, as the O vacancy is expected to be positively charged, in compounds of such electropositive metals. The chapter then describes the effects of the oxygen vacancy in trapping, mobility, and the interaction with gate electrodes. One question is the concentration of defects, and why such large concentrations can occur for defects with a large formation energy.
