The self-assembled formation of ordered, vertically stacked rocksalt/wurtzite MgxZn1−xO heterostructures by planar phase separation is shown. These heterostructures form quasi “natural” two-dimensional hetero-interfaces between the different phases upon annealing of MgO-oversaturated wurtzite MgxZn1−xO layers grown by plasma-assisted molecular beam epitaxy on c-plane sapphire substrates. The optical absorption spectra show a red shift simultaneous with the appearance of a cubic phase upon annealing at temperatures between 900 °C and 1000 °C. Transmission electron microscopy reveals that these effects are caused by phase separation leading to the formation of a vertically ordered rock salt/wurtzite heterostructures. To explain these observations, we suggest a phase separation epitaxy model that considers this process being initiated by the formation of a cubic (Mg,Zn)Al2O4 spinel layer at the interface to the sapphire substrate, acting as a planar seed for the epitaxial precipitation of rock salt MgxZn1−xO. The equilibrium fraction x of magnesium in the resulting wurtzite (rock salt) layers is approximately 0.15 (0.85), independent of the MgO content of the as-grown layer and determined by the annealing temperature. This model is confirmed by photoluminescence analysis of the resulting layer systems after different annealing temperatures. In addition, we show that the thermal annealing process results in a significant reduction in the density of edge- and screw-type dislocations, providing the possibility to fabricate high quality templates for quasi-homoepitaxial growth.
References
Due to the low deposition temperature and the high MgO-concentration, a broad absorption peak was observed for samples with x0 ≥ 0.20. Hence, determination of the absorption edge by consideration of exciton-phonon complexes as demonstrated in Ref. 25 was not possible for all samples. Furthermore, lateral fluctuations of the MgO-content of up to ± 0.02 were observed. The Mg-fractions in the as-deposited reference samples, determined according to Refs. 6 and 8 were 0.054, 0.11, 0.22, 0.29, and 0.38. As different parts of the same sample were used for the annealing experiments, we restrict the precision in the MgO-content to 0.05, 0.10, 0.20, 0.30, and 0.40 following the arguments above.
In addition to lateral inhomogeneities of the phase separation process also rs-MgO precipitates with diameters > 100 nm and different orientation ((001), (111)) with respect to the sapphire substrates were found in some cases. We attribute the presence of such inclusions to inhomogeneities in the MgO-concentration of the as-grown sample.
We assign the presence of (002) rs-MgxZn1−xO diffraction peak in the HRXRD analysis to MgO precipitates that have been observed.
