2016 年 43 巻 4 号 p. 222-232
The III/V alloy semiconductors have become of commercial importance for light emitting diode, solar cell, high speed electronics, and many other applications. These materials can be grown by several epitaxial processes; however, commercial production is most often done using organometallic vapor phase epitaxy (OMVPE or equivalently MOVPE or MOCVD). Even after decades of effort to understand the fundamental aspects of both the growth processes and the materials properties, our understanding of both is still incomplete. However, thermodynamics gives underlying, vital information that leads to an understanding of both the growth processes and the materials properties. In particular, thermodynamics is decisive in determining the optimal growth process. Only so called “non-equilibrium” processes such as OMVPE and molecular beam epitaxy (MBE) can be used for the growth of alloys containing both Al and In, such as AlGaInP and AlGaInN, two materials from which the highest performance LEDs and solar cells are fabricated. The thermodynamics of mixing of these materials can be understood using the delta-lattice-parameter (DLP) model. It allows the prediction of the parameters necessary for the growth of specific alloys for particular device applications. It also gives the key information allowing us to understand the microstructures of the resultant materials. When constituent lattice constants differ widely, the alloys exhibit miscibility gaps. This means that many materials cannot be grown using equilibrium growth techniques. Nevertheless, metastable materials can be grown by OMVPE and MBE using kinetic factors to prevent the establishment of thermodynamic equilibrium during epitaxial growth. However, these materials often exhibit compositional fluctuations driven by thermodynamic forces. The resulting microstructures have enormous consequences for the materials properties and device performance. This review will focus on some of these materials, notably the metastable alloys AlGaInN, GaAsN, and GaAsBi.