Lattice-mismatched epitaxy in the III-V compound semiconductor system based on III-AsP and related alloys are of enormous importance, and considerable research interest, for many years. The reason is straightforward if one considers the limitations placed on available materials properties for devices dictated by lattice matching to the dominant substrate technologies - Si, GaAs (and/or Ge) and InP. For III-V epitaxy, the lattice constants of these substrates have defined a generation or more of device advances since growth of heterostructures possessing the same equilibrium lattice constants as the substrate yields essentially defect-free (specifically extended defect-free) materials and devices. Removing the restriction of lattice matching to current substrate technology opens a rich spectrum of bandgaps, bandgap combinations, conduction and valence band offsets, etc., that are desirable and exploitable for advancing device technologies for new functionality and greater performance. However successful exploitation of these properties requires mitigation of a variety of extended defects that result from the lattice mismatch between substrate and epitaxial heterostructures.
A well known method to achieve this solution is through the use of compositionally (lattice constant-graded) buffer interlayers, in which the equilibrium lattice constants of interlayers are slowly altered by controlled changes in layer composition so that the mismatch strain between the initial substrate and the final device layers is spread across a thickness of buffer. The research accomplished has yielded success for both lattice constant ranges Si – GaAs and GaAs - InP. For the Si – GaAs system, a three step GaP nucleation process on Si has been developed and demonstrated, which maintains total avoidance of creating coalescence-related defects such as antiphase domains and stacking faults resulting from the initial III-V/IV interfaces while reducing overall threading dislocation density by ~10x, to a range of 1x107 cm-2, compared to current state of the art. This reduction can now enable future III-V/Si solar cells based on GaAsP metamorphic buffers in which the underlying Si substrate can participate as an active sub-cell, and such buffers have been demonstrated in this research. Second, in this same lattice constant range, novel GaP/SiGe interfaces on Si were grown and demonstrated to eliminate the small, but not negligible lattice misfit between GaP and Si, and provides a second pathway for future III-V/Si solar cell integration through subsequent metamorphic buffer growth. For the GaAs-InP range of lattice constants, multiple metamorphic buffer strategies, including those based on anion-specific quaternary GaInAsP, combinations of step and linearly-graded buffers, and buffers with multiple ternary alloys were all investigated. Micro-scale phase separation within quaternary anion-graded GaInAsP was identified as a mechanism to significantly inhibit proper lattice misfit strain relaxation, which was explained by thermodynamic arguments consistent with theoretical phase separation. This led to the creation of hybrid step and linearly graded InGaAs/InGaP metamorphic buffers through which phase separation was totally eliminated by avoiding specific compositions that were identified as sources for phase separation. These findings have enabled a realistic path for accessing the full range of bandgaps needed for future high efficiency III-V solar cells through optimized metamorphic III-V grading strategies.