Unintentional doping in GaN
Dept. of Materials Science and Metallurgy, University of Cambridge, Pembroke Street, Cambridge, CB2 3QZ. Physical Chemistry Chemical Physics
(Impact Factor: 4.49).
06/2012; 14(27):9558-73. DOI: 10.1039/c2cp40998d
The optimisation of GaN-based electronic and optoelectronic devices requires control over the doping of the material. However, device performance, particular for lateral transport electronic devices, is degraded by the presence of unintentional doping, which for heteroepitaxial GaN layers grown in the polar (0001) orientation is mainly confined to a layer adjacent to the GaN/substrate interface. The use of scanning capacitance microscopy (SCM) has demonstrated that this layer forms due to the high rate of incorporation of gas phase impurities, primarily oxygen, during the early stages of growth, when N-rich semi-polar facets are often present. The presence of such facets leads to additional unintentional doping when defect density reduction strategies involving a three-dimensional growth phase (such as epitaxial lateral overgrowth) are employed. Many semi-polar epitaxial layers, on the other hand, exhibit significant unintentional doping throughout their thickness, except when a three-dimensional growth phase is introduced to aid in defect density reduction resulting in the presence of (0001) and non-polar facets which incorporate less dopant. Non-polar epitaxial samples exhibit behaviour more similar to (0001)-oriented material, but oxygen diffusion from the sapphire substrate along prismatic stacking faults also locally affects the extent of the unintentional doping in this case.
Available from: Tongtong Zhu
- "Cross-sectional samples were prepared by scribing a small notch on the sample surface with a diamond scribe and then cleaving the sample along the notch. SCM reveals dopant profiles  and has previously been applied to the analysis of polar  and nonpolar  GaN structures grown on sapphire. Plane-view and cross-sectional SEM-CL measurements were performed on selected areas identified by optical microscopy. "
[Show abstract] [Hide abstract]
ABSTRACT: We investigated the properties of a GaN epilayer grown by metalorganic
vapour phase epitaxy on a c-plane bulk GaN substrate obtained by
ammonothermal growth. X-ray diffraction measurements showed that the
epilayer and substrate were fully relaxed, had a miscut angle of
0.3±0.05° towards m and had omega rocking curve width values
of 20-30 arcsec, limited by the instrumental broadening. Scanning
capacitance microscopy data of the sample in cross-section indicated
that the substrate had n-type conductivity with a carrier concentration
of at least 1019 cm-3. Combined optical Nomarski
microscopy, atomic-force microscopy and scanning electron
microscope-cathodoluminescence studies showed the presence of large
hexagonal pyramids on the surface, each associated with one or two
dislocations with a screw-component threading from the substrate. This
observation leads us to calculate a lower limit of the threading
dislocation density of 3×102 cm-2. We
predict that the formation of such hexagonal hillocks during epitaxy can
be avoided with a slightly larger miscut angle of 0.4° or 0.5°.
Another type of defect observed were ridge-like surface structures with
narrow arrays of edge-type threading defects with a local density of
109 cm-2. However, the absence of threading
defects below the regrowth interface at a ridge suggested that this type
of structure is linked to (polishing) damage to the substrate surface
and is therefore rated as an avoidable problem.
Journal of Crystal Growth 11/2013; 383:12-18. DOI:10.1016/j.jcrysgro.2013.07.035 · 1.70 Impact Factor
[Show abstract] [Hide abstract]
ABSTRACT: Using first principles total energy calculations we have investigated the initial stages of the adsorption of Sc and ScN thin films on GaN(0 0 0 1) surfaces under both N and Ga rich conditions. In an ideally GaN(0 0 0 1) bulk terminated surface, and when the Sc atom is constrained to remain on top of the surface, the T4 site configuration is the most favorable. However a structure in which the Sc atom replaces a Ga atom of the first monolayer and the displaced Ga atom occupies a T4 site (forming bonds with Ga atoms only) has the lowest energy. Results are similar for Ga rich conditions: if the Sc atom is constrained on top of the surface, it occupies the T4 site. However, if it is allowed, it will occupy sites in the third (from top) Ga layer and it will form ScN. For a full monolayer of Sc atoms, three different configurations are possible, in all of them there is formation of scandium nitride: a ScN bilayer terminated configuration for N rich conditions, a ScN bilayer underneath a Ga bilayer for Ga rich conditions, and a ScN bilayer under a Ga layer for intermediate configurations. In all three geometries, the ScN are in wurtzite like configurations with distorted bond angles and the surfaces are metallic.
Applied Surface Science 03/2013; 268:16–21. DOI:10.1016/j.apsusc.2012.11.094 · 2.71 Impact Factor
[Show abstract] [Hide abstract]
ABSTRACT: For the characterisation of group III nitrides by X-ray diffraction
there are several challenges. In particular: a) reliable reference data
on cell parameters of the end members and b) a simple and reliable
method to obtain the alloy content of a thin film or multi-quantum layer
structure that is fully strained to a template, the latter being fully
relaxed or still showing some residual strain. A broad range of
reference values for both cell parameters and elastic constants have
been reported. In this work, we investigate the cell parameters and
strain of (0001) GaN templates grown on sapphire and a bulk ammono GaN
sample to derive a coherent set of cell parameters and elastic
constants. The values are found to be applicable to a variety of
samples, with different crystalline quality (threading dislocation
density) or doping level. Subsequently different methods for determining
the composition of group III nitride alloys on such strained GaN layers
are compared. For simpler analysis, the template is often assumed fully
relaxed. We investigate here this approximation and find that standard
``relative'' method fortuitously provides very good results.
Japanese Journal of Applied Physics 08/2013; 52(8). DOI:10.7567/JJAP.52.08JB29 · 1.13 Impact Factor
Data provided are for informational purposes only. Although carefully collected, accuracy cannot be guaranteed. The impact factor represents a rough estimation of the journal's impact factor and does not reflect the actual current impact factor. Publisher conditions are provided by RoMEO. Differing provisions from the publisher's actual policy or licence agreement may be applicable.