Development of Methodology to Determine O(H) Content in GaN
Associate Professor Ewa M. Goldys
April 2001
Project description
1. Aims and Significance
The general aim of our project is to develop a suite of spectroscopic assessment methods of oxygen and hydrogen (O(H)) content in gallium nitride (GaN). These methods will have important mapping capabilities. Further, we aim to apply these methods to investigate the influence of these important inadvertent impurities in practical growth and post-growth processing techniques. We aim to solve a range of specific problems in GaN such as doping, compensation, impurity correlation and ohmic contact stability which require such analysis tools. Advanced materials such as GaN represent the frontier for microelectronic and optoelectronic device applications. The outcomes will help establish the optimum conditions for the manufacture and processing technologies of high quality GaN films for applications in the microelectronics industry. The proposed investigations are aiming to make further distinctive advances in engineering of materials properties for improved device technologies.
1.1 Expected outcomes:
The results will be of both scientific and industrial significance in that they will contribute new knowledge of doping and compensation mechanisms in GaN, address limitations of existing technologies for p-type contact formation using hydrogen-drawing Zr layers and develop a unique set of tools for fingerprinting and mapping of O and H in GaN. All these outcomes are of special interest from an engineering point of view and necessary for the improvement of electronic and optoelectronic GaN devices.
1.2 Issues of significance in GaN technology
In the the last five years GaN and related compounds have emerged as the key materials for applications in light-emitting devices such as blue diodes, short wavelength lasers, UV-sensitive light detectors, and in high temperature high power electronics. Applications in high-density optical data storage (one of the fastest growing sectors in this multibillion dollar industry), medical imaging and diagnosis, and high temperature chemical detection/monitoring help motivate the program. There has been a substantial worldwide interest in the development of GaN-based low-noise circuit elements, particularly transistors with superior characteristics.The fundamental material properties of GaN such as its high direct bandgap and excellent thermal characteristics ensure its dominant role in future generation optoelectronic and electronics devices. However the capabilities of these devices presently fall short of the requirements for long lifetime, high electron mobility and low and stable p-type contact resistance. It is well known that GaN films generally suffer from poor lattice matching to the substrate, typically sapphire. Notable advances have been made in the past several years in the development of low temperature buffers which facilitate growth on such lattice mismatched substrates by acting as fairly efficient barriers for dislocation propagation. Thanks to these developments, industrial GaN growth processes by metalorganic chemical vapour deposition (MOCVD), Molecular Beam Epitaxy (MBE) and Hydride Vapour Phase Epitaxy (HVPE) are now available. However, as we have recently found, structural defects such as dislocations are still present, accompanied by a range of secondary effects such as accumulation of impurities in the vicinity of structural defects. These phenomena, are facilitated by high growth temperature, in MOCVD and HVPE. Some of the impurities, such as hydrogen are inadvertently present in the MOCVD and HVPE growth environment, others, such as oxygen, may be present only in residual quantities but seem to be preferentially incorporated. As a result the quality of these materials, particularly the in-plane and in-depth uniformity required by device technologies, and the effectiveness of the processing methods, for example for p-type contact formation, critically depend on the way the inadvertent impurities are incorporated. Methods of defect assessment and control are thus believed to be one of the remaining critical issues in GaN technology.
1.3 How the present project will address these issues of significance
In the last two years we have completed independent studies on the role of defects in GaN films, particularly of oxygen, hydrogen, and native defects and their interaction with structural defects. We have achieved results on the existence of the dislocation networks on the surface of commercial GaN films very similar to those reported elsewhere. We have also reported novel results relating to the use of cathodoluminescence imaging and depth profiling as a tool to assess uniformity of the films. In the case of oxygen- and hydrogen-related defects, we have led the way internationally including demonstrations of the role of these impurities in the important yellow emission in GaN. We have identified a similar red emission in cubic GaN grown by MBE, and demonstrated its link to similar defects as those responsible for the yellow emission.1.3.1 Motivation to Study oxygen in GaN
The full significance of oxygen in GaN has just begun to emerge. Below we present the key issues that we plan to resolve within this project once sensitive methods to quantify the content of electrically active oxygen and hydrogen become available.a) Oxygen doping of GaN: Oxygen is known to be an n-type dopant that can be incorporated at very high densities, leading to metallic conductivity in bulk GaN grown under high pressure. Its potential as an industrial n-type dopant superior to the presently used Si can only be realised if oxygen can be introduced in a controlled manner and monitored. The role of O in nominally undoped GaN has not been fully explored as well. Undoped GaN is generally strongly n-type with room temperature free electron concentration in excess of 1017 cm-3. The autodoping process has been attributed to nitrogen vacancies as well as to O and other impurities , but it has not yet been fully clarified. We will be in a position to address the issue of autodoping, if we develop methods of oxygen detection that are both quantitative and sensitive to electrically active oxygen to a sufficient degree.
b) Nonuniformities of resistivity due to oxygen: We have reasons to believe that the present MOCVD and HVPE GaN films may have spatially nonuniform electrical properties because of preferential oxygen incorporation in the region of enhanced density of structural defects. In this project we wish to test the latter hypothesis. Detrimental consequences of planar and in-depth nonuniformities are anticipated for device technologies, particularly for high power devices. Our preliminary studies of MOCVD GaN including the world first demonstration of oxygen involvement in the yellow emission point to the link between oxygen incorporation and structural defects. Previously only indirect evidence was available, (see for example a theoretical report by Eisner et al) of preferential incorporation of oxygen into the core of certain dislocations, but, on the other hand, numerous earlier reports relate the yellow emission to structural defects thus bridging the link. In a separate work we have observed a granular structure in the edge emission in GaN due to grouping of dislocations. Thus, if oxygen indeed accumulates at these dislocations, it may locally increase the electron concentration and lead to nonuniformities of electrical properties. By contrast, we have already extensively characterised HVPE GaN grown on sapphire and identified distinctive regions of a greatly enhanced electron concentration in distinctive areas in the near-interface region where dislocations would accumulate. Importantly, in a recent work we have shown that this effect is absent in HVPE GaN grown on a MOCVD GaN template, thus revealing the link between an enhanced electron concentration and dislocations. Oxygen is a likely candidate to be the cause of this enhanced electron concentration, as it yields shallow donor states. However an assessment technique is required to properly judge the quantity of oxygen in these highly conductive regions.
c) Oxygen as a compensator in p-type doping Oxygen is also theoretically postulated to be an important contributor to compensation in p-type Mg-doped GaN. It is anticipated that magnesium is preferentially incorporated in the immediate proximity of oxygen, if the latter is present. The resulting Mg-O pairs are electrically inactive. Experimental identification of such pairs would contribute to future better control of p-type doping in GaN.On the basis of the above it is clear that the an extensive evaluation of the process of oxygen incorporation must be undertaken for a proper control of oxygen in growing GaN films to be of industrial significance. Such evaluation is only possible if methods to quantify the amount of oxygen in various lattice locations can be developed.
The first major aim of this project is thus to complete comprehensive experimental investigations of the signature characteristics of oxygen in GaN films. We anticipate to develop procedures based on various optical spectroscopies and mapping that will enable us to gauge the concentration of residual electrically active oxygen in the GaN films, as well as its spatial distribution in the films. We intend to pursue this as a spectroscopic sensing strategy for applications in industry. These studies will be carried out in conjunction with microstructural characterisations. We will gain important insights on the link between oxygen and various structural defects, and their respective emission fingerprints (yellow luminescence, donor-acceptor pair and bound exciton emissions). As oxygen is a common contaminant, we will aim to fully elucidate the mechanism involved in auto-doping and self-compensation in n-GaN as well as compensation in p-GaN.
1.3.2 Motivation to study hydrogen in GaN
The synthesis of post-growth activated p-type GaN in 1989 and the realisation of elevated hole concentrations in the early nineties has stimulated a rapid progress of GaN device technologies. However further increases in the doping level in the p-type material appeared to be more difficult. Reasons for this are not presently known, but it is widely suspected that hydrogen may be responsible. At the same time, hole mobilities are extremely low, suggesting very high levels of compensation. Thus one of the key problems which we plan to address is:the role of hydrogen in p-type doping. Mg-doped GaN is generally grown in the
presence of hydrogen with an attendant advantage of minimising acceptor
compensation by native donor defects. The Mg acceptors after the growth are not
electrically active and require activation by either annealing in a nitrogen
atmosphere or by low energy electron beam irradiation. The activation process
has been attributed to the dissociation of Mg-N-H complexes and subsequent
removal of H [18]. However, there are major inconsistencies on the reported
effects of H on the photoluminescence spectra of Mg-doped GaN [19]. Recent
theoretical studies suggest that the H behaviour is more complex than the mere
formation and dissociation of Mg-N-H complexes. This is supported by the
research on stability of contacts to p-type material, which appear to
deteriorate with time. Thus the second problem that we plan to investigate is:
The second major aim is thus to identify the optimum conditions for activation of p-type dopants through the studies of hydrogen content in GaN and the role of hydrogen in dopant deactivation processess. These investigations will include development of the p-type ohmic contact technologies meeting appropriate requirements of contact stability using hydrogen-drawing zirconium layers. The relevant outcome will have direct implications for device technologies, and also will offer a unique testing ground to understand the incorporation of hydrogen and its interaction with other defects. In parallel to a similar aim with respect to oxygen, this project aims to devise a spectroscopically useful method of fingerprinting of hydrogen in the GaN films.