Crystal interfaces

Abstract

Interfaces to other semiconductors, producing a heterojunction, or to conductors, acting as contacts or barriers, are important parts of almost every semiconductor device. Their basic interface properties are a decisive element of the device operation and its performance. Layers of different semiconductors with a common interface may be coherently strained up to a to a critical layer thickness, which is roughly inverse to the mismatch of their in-plane lattice parameters. In thicker layers strain is at least partially relaxed by misfit dislocations. The electronic properties of semiconductor heterojunctions and metal-semiconductor contacts are governed by the alignment of their electronic bands. Early models describe band offsets and barrier heights as the difference of two bulk properties. While related chemical trends are found for certain conditions, the band lineup cannot be predicted with sufficient accuracy by a single universal model. Interdiffusion on an atomic scale, defects located at the interface, and leaking out of eigenfunctions from one into the other material create interface dipoles, modify alignments guessed from simple properties of the bulk materials. Additional shifts originate from strain. Various models exist, each describing certain groups of materials forming interfaces. Linear models define reference levels within each material such as charge neutrality levels or branch-point energies within the bandgap, or average interstitial potentials, or they use localized states of impurities lying deep in the bandgap. Nonlinear models account for the formation of interface dipoles; such a charge accumulation can be induced by band states near the interface in one semiconductor lying in the bandgap of the other or by disorder at the interface inducing gap states. More recent first-principle approaches model heterointerfaces explicitly or align bands with respect to the vacuum level by including surfaces.