Solar Photovoltaic (PV) is one of the most promising solutions for electrical energy production in a sustainable way. Currently the PV market is dominated by crystalline silicon solar cells, and the dissemination of this technology in a global scale has been strongly limited by the high cost of the devices. Si Ribbon technologies have the advantage of dramatically increase the material yield over the traditional Si growth techniques by eliminating kerf losses. However, poor electrical performance due to the high defect density resulting from both impurity content and dislocations generated by the heterogeneous temperature field during crystal growth have been in the past a limitation for this type of technologies. The Silicon over Dust Substrate (SDS) process has the potential to reduce material consumption and achieve higher purity substrates than conventional ribbon processes. However it is still limited by the high defect density generated by thermal stresses that occur during crystallization by Zone Melting Crystallization (ZMC).

The aim of this work was firstly to develop a new ZMC furnace concept that permitted improving the performance and size of the Si ribbon grown via ZMC, and secondly to develop a numerical model to understand the thermal stresses and dislocation dynamics during the growth for the conditions obtained with the developed furnace concept. The furnace was also envisioned to crystallize SDS pre-ribbons produced via CVD for a pilot industrial scale.

To develop the new ZMC apparatus a numerical-experimental approach was pursued, in which Computational Fluid Dynamics Finite Element Analysis (CFD-FEA) was performed in a 3D Computational Aided Design (CAD) environment. The final ZMC furnace CAD model was then subjected to CFD-FEA under different thermal conditions, and a study was conveyed to estimate the ribbon thermal stresses and dislocation density during crystal growth by ZMC. Furthermore, the Hassen-Alexander-Sumino (HAS) model was used to compute the final dislocation density profile of the ribbon.

The final ZMC concept includes as main innovations the possibility to control the temperature gradients in the heating and cooling regions, the use of diode laser bars to produce the floating molten zone with width >50mm, and a translation system to transport the sample that allows continuous operation at high temperatures with low impurity sources. With this furnace the ribbon crystallization area was effectively upsized, reaching processed samples with areas of about 55x95cm2, with no cracking behavior, and no buckling. However, the characterization of the recrystallized test samples revealed that the dislocation density is still in the range of 10e5 #/cm2, which is explained by the numerical model results. The model estimates that stresses in the region adjacent to the molten zone of the ribbon are still considerably higher than the yield stress, where dislocations multiply and reach their maximum very rapidly, requiring further smoothing of the temperature profile. An agreement was found between both modelled and measured results in what concerns the dislocation densities profiles, since the same trend in their distribution is verified. Additionally, both measured and modelled dislocation densities are in the range of 10e5 #/cm2, with a slight offset that can be traced back to the model approximations, or the experienced heterogeneous laser power distribution not contemplated in the CFD-FEA.