Herstellung und Charakterisierung von metamorphen Pufferschichten für Ga(AsP)-Tandem-Solarzellen auf Si

The aim of this work was to develop a 1.7eV Ga(AsP)-single-junction solar cell on a Si-substrate with the help of a metamorphic buffer-layer. This device serves as a proof-of-concept for a later tandem solar cell with a Si-bottom-cell, which could reach up to 43% efficiency. This task can be divided into three parts: (i) The polar III-V-nucleation layer on the non-polar Si-substrate. (ii) The optimisation of the metamorphic buffer which shifts the lattice constant to a larger value while simultaneously minimizing the threading dislocation density (TDD) in the following layers. (III) The growth and characterization of the Ga(AsP)-solar cell. The samples were grown by metal organic vapour phase epitaxy and were characterized with the help of X-ray-diffraction, transmission electron microscopy (TEM), atomic force microscopy (AFM) and photoluminescence-spectroscopy (PL). The first part of this work focused on expanding the knowledge of the GaP-nucleation, which was optimized for exact Si-substrates, to off-oriented Wafers. This was expected to lower Anti-phase-domain (APD) densities. By using the information gained from AFM and X-ray reflectometry measurements it was possible to adjust the partial pressure of the TEGa in a low temperature flow modulated epitaxy to assure the optimal layer thickness and no metal-droplets, which can lead to the formation of defects. Step-bunching can occur in the second, high temperature growth step on off-oriented surfaces. Lowering the temperature and the V/III-ratio reduce this effect. The decrease in temperature leads to bigger APDs because the anti-phase-boundaries lie on {111}-planes instead of {112}-planes for higher temperatures. To optimize the metamorphic buffer the relaxation of Ga(AsP)-layers with low lattice mismatch (<1.5%) have been investigated. A "two-step-model" has been developed to describe the process. After a few nanometer of pseudomorphic growth, the layer starts to relax by the nucleation of half-loops at the surface, which glide to the interface between the strained layers and form 60°-dislocations. The relaxation starts, when the product of the thickness and the lattice mismatch reaches 80 nm %. The residual strain is decreasing linearly with increasing layer thickness. During the first step the relaxation in [110]-direction is promoted compared to the [1-10]-direction due to different glide velocities of the dislocations caused by the polarity of the crystal. When a second, higher strained layer, is grown on top of the first layer, the strain of the later is drastically increased and the second relaxation process sets in. Due to the high, already existing defect density dislocation blocking happens and leads to defect multiplication processes, by which the first layer is then fully relaxed. With the „two-step-model“ it is possible to estimate the strain state of the last layer by knowing it’s thickness. This allows the growth of a following unstrained layer to assure no further relaxation within the solar cell. A tilt of the (001)-crystal-planes around the <110>-axis perpendicular to the off-cut-direction is caused by an inhomogeneous amount the 60°-dislocations with different Burgers-vectors. The negative tilt is proportional to the degree of off-cut and the lattice mismatch. The effective strain on the dislocations due to the projection on different glide planes causes this inhomogeneity . The positive tilt can be correlated to a surface effect which stabilizes steps and thus promotes the nucleation of one kind of half-loop dislocations. It is influenced by the V/III-ratio, the growth temperature and the crystal composition, with a maximum of tilt at 50%-As-concentration. The two tilts superimpose each other, thus the positive tilt is only visible for a low lattice mismatch. The measured high tilts are caused by the two-step relaxation process: The inhomogeneity arises due to the half-loop nucleation at the surface during the first step. During the second step the number of dislocations is increased by multiplication, however this is not influencing the ratio of the different dislocations. This causes an increase of tilt during the second step. If the layer relaxes further during the first step, the dislocation-network interacts with the newly formed half-loops and promotes the dislocations with different burgers-vectors, thus reducing the tilt. The Ga(AsP)-single-junction solar cell reached an efficiency of 9.8%, matching the highest reported values for this concept. The performance is limited by a TDD in the range of 1E8 1/cm². The EQE and IV-curve show the influence of the high defect density. The reduced values of the VOC and the other solar cell parameters can quantitatively described when the reduced carrier-diffusion-length is included.