Oxidation kinetics of metal films and diffusion in NiO for data storage

Abstract

The growth of pore-free oxide layers on metals is determined by formation and migration of point defects and thus is an issue of fundamental importance. So far, mainly oxidation kinetics of thick metal films, crystals, and bulk samples were investigated in the literature. The oxidation of thin metal films on insulating substrates can easily be monitored by measuring the conductivity of the remaining metal. This dissertation focuses on understanding the processes that determine the oxide growth rate (e.g., surface reaction or chemical diffusion) in thin films, and also attempts to increase the reaction rate constant for potential applications in irreversible data storage systems. In this work, the resistance changes upon oxidation were measured by electrical impedance spectroscopy during the oxidation of metal films on Al2O3 substrates, and the oxide thicknesses were calculated. The oxide growth follows the parabolic rate law of oxidation for Cr, Al, Ti, V, Zn, Ni and Co films with a thickness typically ranging from 10 to 150 nm. Thus, in spite of their small thicknesses, the rate determining process of the oxidation of these metal films is found to be diffusion through the oxide layer according to the Wagner theory. Ni and Co exhibit a higher oxide growth rate than Cr, Al, Ti, V, and Zn. The oxidation rate constant of Ni is not significantly changed by applying different conditions such as different pO2, UV illumination, ozone exposure, and varying metal film thickness. This confirms the validity of the parabolic rate law of oxidation for the samples with 10-150 nm thickness in the temperature range of 250-500 °C. A comparison of Ni tracer diffusion coefficients between single- and polycrystalline NiO from literature studies and the present work with polycrystalline films (grain size: 10-30 nm) shows that a decreased grain size increases the effective diffusion coefficient by orders of magnitude pointing towards fast Ni diffusion along the grain boundaries in NiO. NiO was chosen as an example for a more detailed investigation of oxidation kinetics and diffusion. Ni diffusion in NiO is mainly controlled by Ni vacancies, and the vacancies are compensated by electron holes in undoped NiO. Thus, donor doping of NiO (e.g., with Al3+, Cr3+, etc.) is expected to increase the reaction rate constant by increasing the Ni vacancy concentration. However, the oxidation rate constant of Cr-doped Ni films with different thicknesses and different Cr concentrations (0.1 and 1 %) showed no significant change compared to undoped NiO. Depth profiles of Cr concentration in Cr-doped Ni films before and after oxidation were obtained by X-ray photoelectron spectroscopy (XPS). The results indicated that Cr was not homogeneously distributed in the films. While there is a Cr-rich surface layer in the metal form and almost homogenous distribution of Cr over the rest of the metal film, in the oxide form of the sample most of the Cr remained at the interface of the oxide and the substrate. Therefore the inhomogeneous Cr distribution in the growing oxide film prevented a faster oxidation. Since it is difficult to precisely control the grain size and to obtain a homogenous dopant distribution in the growing oxide films, the transport properties of undoped and donor-doped NiO ceramic samples were studied by conductivity relaxation measurements. NiO powders were synthesized by nitrate-glycine synthesis method and compacted to dense ceramic samples by spark plasma sintering (SPS). The grain size of the ceramic samples was controlled by additional annealing. Conductivity (σ) and chemical diffusion coefficient (Dδ) values were obtained as a function of oxygen partial pressure and temperature. These values were found to depend in a nontrivial way on Cr content (0.1, 0.3 and 1 %), grain size and thermal history of the samples. The comparison of σ and Dδ values of as-prepared (SPS, 5 min at 1000 °C) and annealed (8 h at 1500 °C) samples showed that the Cr dopants seem to be electrochemically inactive for the as-prepared ceramics. For annealed samples, 0.1 % Cr-doped NiO showed a significant decrease in σ and increase in Dδ compared to undoped NiO, indicating that the dopants are activated. However, a further increase in Cr content caused a decrease in chemical diffusion coefficient for annealed samples. The reason for this unexpected behavior is an inhomogeneous Cr distribution in the samples. The inhomogeneous Cr distribution and formation of undesired NiCr2O4 spinel is detected by transmission electron microscopy/energy dispersive X-ray spectroscopy (TEM/EDX) at grain boundaries for almost all samples, and for high Cr concentrations even in the grains. The obtained conductivities and chemical diffusion coefficients indicate a much lower solubility limit for Cr than reported in the literature. The maximum achievable increase of the chemical diffusion coefficient by donor doping is more than one order of magnitude at 700 °C. In-situ oxidation kinetics measurements of metals revealed that Co films show the highest reaction rate constants within the studied metals; extrapolation yields a millisecond oxidation time at a temperature of 540 °C. The thin films in this study also showed much higher oxidation rate constant than the literature values because of their very small grain size. However, the oxidation rate constant is still not in the desired range for a routine long-term data storage application, which requires microsecond oxidation time. Nonetheless, for special purposes, Co thin films could be candidates for a long-time data archiving system owing to the very low oxidation rate constants at room temperature.