Summary of the 20th International Conference on Defects in Semiconductors (ICDS-20) in Berkeley, CA, July 26

Summary of the 20th International Conference on Defects in Semiconductors (ICDS-20) in Berkeley, CA, July 26 - 30, 1999.

This conference has been regularly held since 1959, and attracted nearly 300 participants. It covers the basic properties of defects in all semiconductors, from point defects and impurities to extended defects. Here we shall only briefly cover the part of the program that was concerned with III-nitrides.

The electronic properties of dislocations in GaN were discussed in several contributions. There was a common agreement that dislocations generally are severe recombination centers. This is true both for grown-in dislocations, believed to be decorated with point defects and impurities, and fresh clean dislocations. In the latter case it was shown by a real time video sequence that dislocation segments can move by glide under the influence of electron irradiation at room temperature. This was interpreted as evidence of recombination enhanced defect motion ( i e not a heating effect).

The basic properties of Mg acceptors were discussed. Optically detected magnetic resonance studies confirm a binding energy of 260 meV, and a presence of compensating donors of unknown origin. For effective mass like acceptors the g factor is expected to be completely anisotropic, while experimentally only an anisotropy of a few % is found. It was found that this anisotropy was slightly dependent on the doping level, decreasing at high doping. A model of an axially distorted wavefunction was suggested, where the bound hole would be shared in different proportions between the bond aligned with the c-axis and the 3 bonds forming an angle with the c-axis, which could explain the observations. Other magnetic resonance data reported a triplet signal in bulk Mg-doped GaN, taken as evidence of the presence of neutral Mg-O complexes.

Several contributions discussed the possible defects responsible for the common problem of reduced acceptor activation in Mg-doped GaN, but there is so far no identification of these defects. Activation energies obtain from electrical measurements disagree strongly with optical data, which is still not explained. Theoretically a number of H related defect complexes are expected to occur in annealed material, which could be part of the complexity. The accuracy of theoretical calculations is limited by the assumptions of bulk thermal equilibrium for calculations of defect formation enthalpies, not valid at the usual growth conditions.

In another report it was shown that Mg may show up in a second ordered phase in some GaN materials, in particular for N polarity growth. Two different structural models were discussed.

A well structured PL band related to silver in GaN occurs at about 1.6 eV. It was shown that this band is related to one Ag atom on a Ga site.

An update was given on the studies of vacancy type defects by positron annihilation spectroscopy. It was shown that in Mg-doped GaN the Mg acceptors are the main source of donor compensation, as the amount of Ga vacancies decreases. This type of experiment can quantitatively give the concentration of Ga vacancy type defects, but there is no clear discrimination between different negative charge states of these vacancies, neither is it possible to exclude that complexes of these vacancies with impurities are observed. The data further suggest that the Ga vacancy is essentially undistorted, but gives rise to a slight outward relaxation of the adjacent atoms. Interestingly, the concentration of these vacancies is found to be high (of the order 10¹⁸ cm^-3) in n-type O-rich GaN, but very low in n-type Si-doped GaN. This observation raises the question whether the stable defects observed might be V_Ga-O complexes. Unfortunately this technique is not sensitive to nitrogen vacancies, as these are expected to be positively charged.

Tentative results were reported for the primary defects in irradiated GaN. A major annealing stage for the recovery of electrical properties at about 500 C was confirmed. However, residual defects are stable up to much higher temperatures. It was suggested that one of the primary defects is mobile at room temperature. Much more work is needed to identify these defects, however (no conclusive magnetic resonance data as yet).

A correlated result was demonstrated for Schottky barrier formation with the aid of various deposition techniques. Clearly sputtering is creating a large amount of defects in GaN, very detrimental to the reverse voltage characteristic. Electron beam deposition of the metal can lead to excellent results for the reverse leakage current if the electron beam is carefully screened from the GaN surface during the evaporation.

The annealing of implanted GaN was discussed. Cd is an interesting dopant to study, since in that case perturbed angular correlation (PAC) spectroscopy can be used. For annealing above 1100 C about 60 % active Cd acceptors at Ga sites were recorded. An interesting technique for surface protection during GaN anneal was to include metallic Al in the ampoule, which gave a protective AlN layer on the surface. It was claimed that this simple procedure was effective for annealing temperatures up to 1600 C. Data from sample characterization were preliminary.

A report was given on Si donors in GaNAs with a N content of about 1.5 %. Two donor states are observed in electrical measurements at high pressure, one of these shows metastable properties similar to the DX donors in AlGaAs.

Si doping of GaN is known to have a strong effect of the strain status of heteroepitaxial layers. It was demonstrated that this effect is correlated with a surprisingly strong variation of the lattice parameters in the layers. The origin of this effect requires further study.