M. A. Moram

X-ray diffraction of III-nitrides

The III-nitrides include the semiconductors AlN, GaN and InN, which have band gaps spanning the entire UV and visible ranges. Thin films of III-nitrides are used to make UV, violet, blue and green light-emitting diodes and lasers, as well as solar cells, high-electron mobility transistors (HEMTs) and other devices. However, the film growth process gives rise to unusually high strain and high defect densities, which can affect the device performance. X-ray diffraction is a popular, non-destructive technique used to characterize films and device structures, allowing improvements in device efficiencies to be made. It provides information on crystalline lattice parameters (from which strain and composition are determined), misorientation (from which defect types and densities may be deduced), crystallite size and microstrain, wafer bowing, residual stress, alloy ordering, phase separation (if present) along with film thicknesses and superlattice (quantum well) thicknesses, compositions and non-uniformities. These topics are reviewed, along with the basic principles of x-ray diffraction of thin films and areas of special current interest, such as analysis of non-polar, semipolar and cubic III-nitrides. A summary of useful values needed in calculations, including elastic constants and lattice parameters, is also given. Such topics are also likely to be relevant to other highly lattice-mismatched wurtzite-structure materials such as heteroepitaxial ZnO and ZnSe.

Understanding x-ray diffraction of nonpolar gallium nitride films

X-ray diffraction (XRD) is widely used for the rapid evaluation of the structural quality of thin films. In order to determine how defect densities relate to XRD data, we investigated a series of heteroepitaxial nonpolar a-plane GaN films with different densities of dislocations and basal plane stacking faults (determined by transmission electron microscopy). Factors influencing XRD data include surface roughness effects, limited lateral coherence lengths, lateral microstrain, mosaic tilt, and wafer curvature, in addition to the defects present. No direct correlation between defect densities and any measured XRD parameter was found. However, the structural imperfections dominating XRD data can be identified by specific analysis of each individual broadening factor. This reductive approach permits full explanation of the in-plane rotational anisotropy of symmetric ω-scan widths for both a-plane and m-plane films: in these samples, mosaic tilt is the dominant factor.

On the origin of threading dislocations in GaN films

A series of GaN films were grown by metalorganic vapor phase epitaxy on nitrided sapphire using an initial annealed low-temperature nucleation layer (LT-NL), without employing any conventional threading dislocation (TD) reduction methods. Film thicknesses ranging from the LT-NL to 500 nm were used. The island network morphology was investigated at each growth stage using atomic force microscopy. Data from cathodoluminescence studies showed initially uniform luminescence, followed by the gradual development of bright (low TD) regions which had lateral sizes different from the island sizes at all times and which continued to increase in size after coalescence. The formation of low-energy arrays of a-type TDs also continued after island coalescence. X-ray diffraction, transmission electron microscopy (TEM) and AFM data indicated that the highest (a+c)-type TD densities were found in the LT-NL, but subsequently decreased due to TD loop formation (promoted by island facets) and reaction to produce a-type TDs. ...

Accurate experimental determination of the Poisson’s ratio of GaN using high-resolution x-ray diffraction

An accurate Poisson’s ratio value of 0.183±0.003 for a typical c-axis-oriented GaN film grown by metal-organic vapor-phase epitaxy deposition has been determined using a wafer bending apparatus combined with high-resolution x-ray diffraction lattice parameter measurements. The precision of this method has improved ten fold over typical methods used for thin film samples, enabling future study of the effects of doping, compositional changes, or structural defects on the Poisson’s ratio of GaN. The obtained Poisson’s ratio value is lower than most calculated values, which is attributed to the presence of strain-relieving edge dislocations in the GaN sample. Unstrained film lattice parameters can also be found using this method, and are shown to be a=3.1884±0.0002A and c=5.1850±0.0002A (assuming an unstrained c∕a ratio of 1.6262). A brief review of Poisson’s ratio and unstrained lattice parameter values for GaN in the literature is also given.

Dislocation reduction in gallium nitride films using scandium nitride interlayers

We describe a method of reducing threading dislocation densities in 0001-oriented GaN from (5.0±0.5)×109cm−2to(3.1±0.4)×107cm−2 (for coalesced films) or to below 5×106cm−2 (for partially coalesced films) in a single step, without lithography. Lattice-matched, dislocation-blocking scandium nitride interlayers are deposited on a 500nm GaN-on-sapphire template. Dislocation-free GaN islands grown on the ScN interlayer nucleate both on the interlayer and on tiny areas of the GaN template exposed through openings in the interlayer. However, some dislocations are generated above the interlayer during subsequent island coalescence.

Defect reduction in (112¯2) semipolar GaN grown on m-plane sapphire using ScN interlayers

The effect of ScN interlayer thickness on the defect density of (112¯2) semipolar GaN grown on m-plane sapphire was studied by transmission electron microscopy. The interlayers comprised Sc metal deposited on a GaN seedlayer that was nitrided before GaN overgrowth by metal-organic vapor-phase epitaxy. Both interlayer thicknesses reduced the dislocation density by a factor of 100 to low-108 cm−2. The 8.5 nm interlayer produced regions that were free from basal plane stacking faults (BSF) and dislocations. The overall BSF density here was reduced by a factor of 5, to (6.49±0.07)×104 cm−1, without the need for an ex situ mask patterning step.

The effects of Si doping on dislocation movement and tensile stress in GaN films

Dislocations in undoped GaN move in response to the in-plane tensile stress present during film growth. Dislocation movement during growth relieves tensile stress, produces arrays of a-type dislocations and reduces the overall dislocation density, with preferential reduction of (a+c)-type dislocations. However, Si-doping limits dislocation movement, limiting the relief of the tensile stress that develops during growth and limiting dislocation reduction, probably due to the formation of Si impurity atmospheres at dislocations. Consequently, Si-doped films are under relatively greater tensile stress compared to undoped GaN films grown under similar conditions. Alternative dopants could be chosen to reduce tensile stress development, such as Ge.

Dislocation movement in GaN films

We demonstrate that significant dislocation movement occurs below the surface of heteroepitaxial c-plane GaN films during their growth by metalorganic vapor phase epitaxy. Dislocations move primarily by vacancy-assisted climb, which appears to be driven by the high in-plane biaxial stresses present during growth. Annealing low dislocation density (4.3×108 cm−2) GaN films promotes dislocation climb and thus reduces both dislocation densities and in-plane stresses (at high temperatures), independent of epilayer growth conditions.

Piezoelectric coefficients and spontaneous polarization of ScAlN

We present a computational study of spontaneous polarization and piezoelectricity in Sc(x)Al(1-x)N alloys in the compositional range from x = 0 to x = 0.5, obtained in the context of density functional theory and the Berry-phase theory of electric polarization using large periodic supercells. We report composition-dependent values of piezoelectric coefficients e(ij), piezoelectric moduli d(ij) and elastic constants C(ij). The theoretical findings are complemented with experimental measurement of e33 for a series of sputtered ScAlN films carried out with a piezoelectric resonator. The rapid increase with Sc content of the piezoelectric response reported in previous studies is confirmed for the available data. A detailed description of the full methodology required to calculate the piezoelectric properties of ScAlN, with application to other complex alloys, is presented. In particular, we find that the large amount of internal strain present in ScAlN and its intricate relation with electric polarization make configurational sampling and the use of large supercells at different compositions necessary in order to accurately derive the piezoelectric response of the material.

Elastic constants and critical thicknesses of ScGaN and ScAlN

Elastic constants of hexagonal ScxGa1−xN and ScxAl1−xN up to x = 0.375 were calculated using a stress-strain approach. C11, C33, C44, and C66 decreased while C12 and C13 increased slightly with increasing x. The biaxial [0001] Poisson ratios increased from 0.21 for GaN to 0.38 for Sc0.375Ga0.625 N and from 0.22 for AlN to 0.40 for Sc0.375Al0.625N, due to greater u values, in-plane bond lengths and bond ionicities. Subsequently, critical thicknesses for stress relaxation were calculated for ScxAl1−xN/AlN, ScxGa1−xN/GaN, and ScxAl1−xN/GaN heterostructures using an energy balance model. These range from 2 nm for Sc0.375Al0.625N/AlN and Sc0.375Ga0.625N/GaN to infinity for lattice-matched Sc0.18Al0.82N/GaN.