Investigation of the threshold voltage shift effect of lanthanum(III) oxide on tin/hafnium dioxide/lanthanum oxide/silicon dioxide/silicon stacks

The semiconductor industry continues to scale (shrink) transistor dimensions to both increase the number of transistors per integrated circuit and their speed. One important aspect of scaling is the need to decrease the equivalent oxide thickness of the transistor gate dielectric while minimizing leakage current. Traditional thin layer SiO2 or SiOxNy films have been replaced by higher dielectric constant film stacks Here we study one example, the HfO2/La2O3/SiO 2 stack. This dissertation describes an investigation of the use of La2O3 to reduce the threshold voltage of TiN/HfO 2/SiO2/Si stacks (high-kappa/metal gate stacks). A significant aspect of this study is the determination of band alignment for a series of high-kappa/metal gate stacks that explore the effect of placement and thickness of the Lanthanum oxide layer. In order to achieve this goal, a number of film stack properties were determined including film thicknesses, band gap of the high-kappa oxides, the flat band voltages, Si band bending, and the valence band and conduction band offsets.;The first part of this work was measurement of individual layer thickness in the multi-layer film stacks using spectroscopic ellipsometry (SE) and other complementary techniques. In order to more completely understand the SE measurements, complementary techniques were used. These techniques include angle resolved X-ray photoelectron spectroscopy (ARXPS), X-ray reflectivity (XRR), transmission electron microscopy (TEM), and Rutherford backscattering spectroscopy (RBS). In this dissertation, we show that SE can simultaneously measure HfO 2 and SiO2 thicknesses in HfO2/SiO2/Si stacks. We discuss the difficulties in simultaneous measurement of all films in the La oxide Hf oxide film stack.;The second part of this dissertation is the measurement of the band gap of high-kappa films. The band gap of a high-kappa film is an important parameter because it affects the conduction band offset (CBO) between high-kappa and Si substrate. The CBO affects the gate leakage current of the transistor. The band gap of high-kappa films was determined from the complex refractive index using several different methods. Comparisons of plots of the extinction coefficient (k), absorption coefficient (alpha), and optical models for imaginary part of the dielectric function (epsilon2) show that each method gives slightly different values for the band gap. The Cody Lorentz model for the dielectric function provides a useful model for the defect induced sub-band gap absorption. We show the impact of the subband gap absorption on band gap extrapolation. Because the existence of sub band gap states is well documented in the literature, we use the Cody Lorentz model to determine the band gap.;The next step was to determine band alignment of the valence and conduction bands the layers in the film stack. X-Ray photoelectron spectroscopy (XPS) measurements were used to determine the valence band offset (VBO) and silicon band bending. The conduction band levels were determined from the valence band energy levels and the band gap. The CBO we measured (1.77eV) is well above the specified minimum CBO for Hf oxide (1.0eV). We developed a band alignment model to account for the trends that we observed. Our data is consistent with the presence of a dipole at the high-kappa/SiO2 interface. According to this model, the change in VBO is a direct measure of the change in the interface dipole moment.;Because the combination of capacitance -- voltage (C-V) and XPS to measure the flat band voltage and Si band bending, respectively, has rarely been used, relationship between these two quantities has not been discussed in the literature. The agreement between an empirical, theoretical relationship between flat band voltage vs. Si band and our data suggests that XPS can be a useful tool for examining VT shift layers in high-kappa gate stacks.;We also investigated the effect of the SiO2 thickness and growth method on the flat band voltage of TiN/HfO2/La2O 3/SiO2/Si stacks. We observed no change in flat band voltage for stacks with nominally 12, 20, and 30 A thermally grown SiO 2 layers. The stack with the nominally 8 A chemically grown SiO 2 layer, however, showed a flat band voltage that was significantly more negative. This difference in flat band voltage may be due to an increase in sub-oxide concentration in the 8 A chemically grown SiO2 layer which could affect the chemistry at the La2O3/SiO 2 interface and thus the interface dipole.