Date

2018

Document Type

Dissertation

Degree

Doctor of Philosophy

Department

Materials Science and Engineering

First Adviser

Stranwitz, Nicholas C.

Abstract

Oxide/semiconductor interfaces are ubiquitous in modern electronics. If these interfaces are of poor quality (e.g., large amount of electronic traps) or possess traits that are undesirable (e.g. fixed charge in metal oxide semiconductor field effect transistors, MOSFETs), the device will not function properly or at all. Atomic layer deposition (ALD) has been shown to produce films with a wide variety of properties, due in part to the variability in processing parameters, such that thin films can be grown for a wide variety of applications.One such variance in process parameters is the chemistry utilized to grow ALD films. Typically, Al2O3 films are grown with trimethylaluminum (TMA) and water. Non-hydrolytic ALD (NHALD) utilizes less oxidative precursors than water, which can affect the film and interface properties. I have shown that using isopropyl alcohol (IPA) instead of water as the oxygen precursor in the ALD growth of Al2O3 results in films with a larger fixed charge (NF), but also a higher concentration of carbon impurities that lead to a higher amount of electronic defects at the interface (Dit) and even in the oxide bulk. Preliminary work was done to make hybrid traditional/NHALD films by which the volume density of electronic traps can be measured and the overall film properties can be modified.Another parameter by which film properties can be modified is the deposition temperature during ALD. I have deposited films using temperatures ranging from 50°C to 300°C on both hydrogen-terminated silicon and silicon containing a native oxide. Fixed charge and interface trap state density was measured on metal oxide capacitor (MOSCap) structures. We have shown that fixed charge magnitude is low at high deposition temperatures and at 50°C, with a maximum at ~100°C. We have also shown that oxides deposited at 50°C contain a defect state that corresponds to silicon dangling bonds at the surface. The presence of dangling bonds implies that insufficient hydrogen is present to passivate them, and the low NF implies that less excess oxygen is present to contribute to NF. This is surprising as films deposited at low temperatures contain more –OH groups trapped in the film. However, Al2O3 deposited at lower temperatures is less dense and therefore allows more –OH groups to effuse from the film during annealing, thus explaining the apparent hydrogen and oxygen deficiency.In addition to MOSCap structures, I have also made and measured MOS diodes, which have the same structure as MOSCaps but contain an oxide layer that is less than 2.5 nm, thus allowing direct tunneling of carriers between the metal and semiconductor. These structures are useful in low-resistance metal/semiconductor contacts and as rectifying junctions in metal-insulator-semiconductor photovoltaics (MISPVs). Using numerical simulations and analytical expressions, the Schottky barrier height (ϕB) of the MOS diodes should rely both on the fixed charge and oxide thickness. By making devices with high and low NF magnitudes (annealed and as-deposited, respectively) and with different oxide thicknesses, the ability of the fixed charge to modify the ϕB was shown to not exist. Our data instead imply that electronic dipoles at the Al2O3/SiO2 interface control the barrier height. The barrier height trends are explained by dipoles that depend on the thickness and chemical character of the SiO2 interface layer. I have carried out some preliminary investigations of the Al2O3/SiO2 dipole for oxides using water, IPA, and O3 as oxygen precursors as well as just begin an experiment to measure the presence (or absence) of a dipole at the Al2O3/TiO2 interface.

Share

COinS