The co-existence of multi-component liquid and solid intermediate phases before the hetero-LPE of III–V solid solutions

Liquid phase epitaxial growth of III–V solid solutions invariably involves contact between the multi-component saturated or undersaturated liquid and solid phases which are not in thermodynamic equilibrium because either the number of components is different or the composition of the layer to be grown differs from the composition of the underlying layer. The non-equilibrium system must relax to the final equilibrium state through some intermediate ones.

The main point of the present review is to show that all non-equilibrium systems encounted in hetero-LPE come through the following stages of relaxation:

• 1.

  1. Partial dissolution of the solid with simultaneous formation of a thin diffusive dividing layer (0.5–3 nm thick) (DDL) at the solid/liquid interface (in the subsurface region of the solid). The layer contains all the components of the given system and in some cases the quasi-equilibrium between the saturaed multi-component liquid and the solid diffusive dividing layer can be observed experimentally. If the DDL is mismatched to the substrate the former must be strained. So, the Gibbs potential of the solid increased additionally and the liquid must become supersaturated by additionally dissolving the substrate.

• 2.

  2. The nucleation and the growth of centres of a new phase at some points on the solid/liquid interface with the simultaneous dissolution of the solid at other areas of the same interface (mechanism of “etch-back and regrowth”).

• 3.

  3. The formation of a continuous epitaxial dividing layer (EDL), separating a bulk solid from a multi-component liquid. At this stage the system stratifies into a three-layer configuration: multi-component liquid/EDL/substrate, further relaxation is limited by a solid state diffusion.

It is only during dedicated experiments on selected systems that stages 1 to 3 can be observed separately in their more or less pure form. Usually, the first and the second stages almost escape experimental detection due to an enormous variability in time scales and an inherent randomness of events during the “fast” stage 2. So, in practice the varied combination of relaxation modes is observed. It is, however, true that, among many parameters which influence the mode of non-equilibrium solid/liquid interface relaxation, the most critical are the temperature of isothermal contact between liquid and solid phases and the lattice mismatch between a substrate and a new solid phase. This concerns not only the magnitude of the mismatch but its sign as well.