It has been a long-standing goal to epitaxially integrate III-V alloys with Si substrates which can enable low-cost microelectronic and optoelectronic systems. Among the III-V alloys, gallium phosphide (GaP) is a strong candidate, especially for solar cells applications. Gallium phosphide with small lattice mismatch (~0.4%) to Si enables coherent/pseudomorphic epitaxial growth with little crystalline defect creation. The band offset between Si and GaP suggests that GaP can function as an electron-selective contact, and it has been theoretically shown that GaP/Si integrated solar cells have the potential to overcome the limitations of common a-Si based heterojunction (SHJ) solar cells.

Despite the promising potential of GaP/Si heterojunction solar cells, there are two main obstacles to realize high performance photovoltaic devices from this structure. First, the growth of the polar material (GaP) on the non-polar material (Si) is a challenge in how to suppress the formation of structural defects, such as anti-phase domains (APD). Further, it is widely observed that the minority-carrier lifetime of the Si substrates is significantly decreased during epitaxially growth of GaP on Si.

In this dissertation, two different GaP growth methods were compared and analyzed, including migration-enhanced epitaxy (MEE) and traditional molecular beam epitaxy (MBE). High quality GaP can be realized on precisely oriented (001) Si substrates by MBE growth, and the investigation of structural defect creation in the GaP/Si epitaxial structures was conducted using high resolution X-ray diffraction (HRXRD) and high resolution transmission electron microscopy (HRTEM).

The mechanisms responsible for lifetime degradation were further investigated, and it was found that external fast diffusors are the origin for the degradation. Two practical approaches including the use of both a SiNx diffusion barrier layer and P-diffused layers, to suppress the Si minority-carrier lifetime degradation during GaP epitaxial growth on Si by MBE were proposed. To achieve high performance of GaP/Si solar cells, different GaP/Si structures were designed, fabricated and compared, including GaP as a hetero-emitter, GaP as a heterojunction on the rear side, inserting passivation membrane layers at the GaP/Si interface, and GaP/wet-oxide functioning as a passivation contact. A designed of a-Si free carrier-selective contact MoOx/Si/GaP solar cells demonstrated 14.1% power conversion efficiency.

GaAs-based solar cells have attracted much interest because of their high conversion efficiencies of ~28% under one sun illumination. The main carrier recombination mechanisms in the GaAs-based solar cells are surface recombination, radiative recombination and non-radiative recombination. Photon recycling reduces the effect of radiative recombination and is an approach to obtain the device performance described by detailed balance theory. The photon recycling model has been developed and was applied to investigate the loss mechanisms in the state-of-the-art GaAs-based solar cell structures using PC1D software. A standard fabrication process of the GaAs-based solar cells is as follows: wafer preparation, individual cell isolation by mesa, n- and p-type metallization, rapid thermal annealing (RTA), cap layer etching, and anti-reflection coating (ARC). The growth rate for GaAs-based materials is one of critical factors to determine the cost for the growth of GaAs-based solar cells. The cost for fabricating GaAs-based solar cells can be reduced if the growth rate is increased without degrading the crystalline quality. The solar cell wafers grown at different growth rates of 14 μm/hour and 55 μm/hour were discussed in this work. The structural properties of the wafers were characterized by X-ray diffraction (XRD) to identify the crystalline quality, and then the as-grown wafers were fabricated into solar cell devices under the same process conditions. The optical and electrical properties such as surface reflection, external quantum efficiency (EQE), dark I-V, Suns-Voc, and illuminated I-V under one sun using a solar simulator were measured to compare the performances of the solar cells with different growth rates. Some simulations in PC1D have been demonstrated to investigate the reasons of the different device performances between fast growth and slow growth structures. A further analysis of the minority carrier lifetime is needed to investigate into the difference in device performances.