Energy-Band Discontinuities at Lattice-Matched Semiconductor Heterojunctions

The thesis is organized as follows: Oh. 1 is a brief introduction to the field of the physics of semiconductor heterostructures: it contains a survey of the materials studied in this thesis, and describes the various types of heterojunctions, with brief comments about their structural characterization and growth techniques. Oh. 2 is specifically devoted to the band-offset problem. After some preliminary definitions, I present the most widely used methods for energy-level measurements, and discuss the trends resulting from the experimental data. A theoretical approach for a systematic study of the band-offset problem, aimed at understanding the nature of the physical phenomena occurring at the interface, is necessary. Oh. 3 highlights the general features and states a crucial problem common to any theoretical approach, even if hidden under several different formulations: i.e. the possible existence of an absolute energy scale to which the band structures of the various materials may be referred. The most significant "model" theories are then critically reviewed in this perspective. The results of some new calculations are also reported in this Chapter; they are attempt of reformulating one of the models in order to make it applicable to the more general case of lattice-matched heterojunctions. The Chapter ends with a general discussion about this kind of model theories which underlines their merits and limits, and stresses the necessity of a more accurate and systematic ab-initio theoretical approach, to which the remaining part of the work is devoted. Ch. 4 explains the basic features of the self-consistent density-functional abinitio pseudopotential method, together with some technical details for its practical implementation to the specific case of heterojunctions; the Chapter also contains a short review of some recent self-consistent calculations which are comparable to the new ones presented in this work. The original results are mainly collected in the remaining Chapters 5 and 6. Ch. 5 reports the results of the self-consistent supercell calculations, with a description of the technical details in the first part. I study some typical heterojunctions, and more exactly one representative example for each of the following classes of heterostructures, in order of increasing complexity: a) isovalent common-anion: I choose GaAs/ AlAs from the III-V group, since there exists a large number of experimental data and theoretical predictions. b) isovalent no-common-ion: I choose as a prototype among the III-V group InP /Ga1-xlnxAs, which has a great technological importance, it is latticematched at the composition x = 0.53, and has been extensively studied from an experimental point of view but not theoretically; c) heterovalent: I study Ge/ GaAs which, among the group IV - group III-V heterojunctions, is the simplest and most studied. Calculations are performed for the thre~ main crystallographic orientations, i.e. (001), (110), and (111). For InP/Gao.41lno.s3As I also study separately the anionic and cationic contribution to the VBO. This system also introduces a new problem, i.e. the treatment of the alloy: the calculations are first performed using the virtual-crystal approximation (VCA). These results are then compared with the ones obtained by considering the "true" Ga and In atoms separately in the galliumindium arsenide region, necessarily in some ordered configuration. In particular the case InP /(GaAs)i(InAsh along the (001) direction is considered, since it is now possible to grow it epitaxially. For this system, I present only some preliminar study; here the presence of internal strains which are no longer negligible as in the treatment of the alloy using the VCA, makes the calculations more difficult and the theoretical study actually performed can just give an idea of the magnitude of the effects of order and internal lattice relaxation. The self-consistent results allow to draw partial conclusions, namely that the VBO is the same in the three main crystallographic orientations for isovalent heterojunctions, and the additivity of anionic and cationic effects in InP /Gao.47Ino.ssAs. However, whether the energy-band discontinuities are a bulk or an interface property cannot be ascertained at this point. Finally, in Ch. 6 I describe and discuss the linear response method. The investigation with this method of the same systems studied in Ch. 5 sheds light on several questions: it appears that for isovalent heterojunctions the energy-band discontinuities are indeed a bulk property within the limit of validity of linear response theory, whereas this is not true for the heterovalent systems in which the band offset can be divided into a "bulk" and an "interface-specific" contribution, the latter being strongly dependent on the structural details of the interface. The validity of the linear response method is also analyzed in this Chapter, by discussing the role of quadratic and higher-order terms. The thesis is concluded with a general discussion about the contribution given by the present work to the investigation of the band-offset problem and with a description of open problems for future work.