Effect of different methods of oxidation on SiSiO2 interface state properties

Abstract

Three different methods of oxidation - thermal, TCE, and anodic — were applied to n-type (111) silicon 10 ohm cm resistivity samples. MOS test samples were fabricated and their interface state properties were characterized by C-V and AC field effect techniques. From C-V measurements the interface state density at mid gap was found to be less in TCE (2 × 10¹⁰ cm⁻² eV⁻¹) and anodic (1 × 10¹⁰ cm⁻² eV⁻¹) samples than in dry (5 × 10¹⁰ cm⁻² eV⁻¹) oxidized samples. The mobile charges were also less in TCE (2 × 10¹⁰ cm⁻²) and anodic (5 × 10¹⁰ cm⁻²) samples. Using the AC field effect technique, the frequency (2–100 kHz) and temperature dependence of field effect mobility, μ_FE, were studied. By applying Garrett's theory of frequency dependence of μ_FE, the relaxation times of interface states were found to vary from 30 to 3 μs in dry, 8 to 4 μs in TCE, and 4.5 to 1.5 μs in anodic samples in the temperature range 230 to 370 K. Using Rupprecht's theory of temperature dependence of relaxation times, thedominant energy levels, E_c - E_T, were found to be shallower in TCE (0.04 eV) and anodic (0.06 eV) than in dry (0.1 eV) oxidized samples. The capture cross-section of these samples was found to be small, in the range 10⁻²⁰ to 10⁻²¹ cm². In TCE and anodic samples the shallow interface state levels indicate stronger interactions between silicon and oxygen atoms at the interface. The observed low densities of interface states and mobile charges in these samples also show improved passivation of silicon.