UCLA

Cleave Engineered Layer Transfer (CELT) is a technique which enables transferring large scale high quality epitaxial layers and devices onto alternative substrates through mechanically engineered substrates. The objective of this dissertation was to formulate a methodology to understand whether the formation of a porous III-V layer, with focus here on indium phosphide, can be feasibly employed to transfer III-V device layers with preserved crystallinity from the host substrate. The porous structures were prepared on (001) oriented InP substrates by electrochemical etching in diluted hydrochloride acid (HCl). The porosity of the layers was reproducibly controlled by etching parameters, such as current density and electrolyte concentration, and a numerical relationship between layer porosity, p and its Young's modulus, E as E = 87(1-p)⁴ was determined. This relationship was important for relating the etching parameters to the mechanical strength of the buried layer. Annealing (temperature range 450 °C-650 °C and time range 10 min- 8 hours) modified the porous layer morphology and was found to be beneficial for subsequent epitaxial growth and layer fracture. The morphology of the porous layers was monitored by scanning electron microscopy (SEM) and the epitaxial layers were studied by high resolution X-ray diffraction (XRD) and transmission electron microscopy (TEM). Nano-indentation was used to measure the Young's modulus of the porous InP layers. In some cases, dual porous layers were produced by changing electrolyte concentration and current density during the porous formation process. The top layer is less porous than the buried layer, with a surface pore fraction of less than 30%. The deeper, more porous buried layer was shown to facilitate fracture in the deeper layer and not through the denser, near surface layer, thus controlling the layer fracture formation through the buried layer only.

Epitaxial deposition of InP on the dual layer porous structure was achieved. The overlying epitaxial InP layer was found to possess a low threading dislocation density and demonstrated complete coalescence of the InP lattice. Specifically, TEM analysis showed the epitaxial layers were single crystalline and lattice registered to the porous layer, and neither grain boundaries nor threading locations were detected over a 35 μm² sampling areas, indicating a threading dislocation density lower than 3�10⁶ cm^-2. In addition, the porous structure was sufficiently weakened after growth and /or annealing to promote facile transfer of the top layer to alternative substrates. Over 1 cm² InP layers (the entire porous region formed) were successfully transferred onto both flexible PDMS substrates and glass substrates. Finally, fractured porous materials on both sides was shown to be readily removed by chemical mechanical polishing, indicating a pathway for employing the CELT process in a reusable process to mitigate the high costs of InP and other III-V substrates.