Abstract: Oxygen precipitates are among the most detrimental oxygen-related silicon bulk defects formed during solar cell manufacturing. These defects are formed only during high-temperature processes, impeding an identification of prone materials during incoming inspection. Moreover, the prediction of oxygen-precipitate-related bulk charge carrier recombination currently requires advanced numerical simulation. This work presents an easily implementable model to predict the bulk carrier lifetime limit, using the temperature-time profile of a high-temperature process as well as the material properties as the input data. In addition to published analytical descriptions of oxygen precipitation, an empirical description of the retarded growth of small precipitates is included. Furthermore, the time-lag in nucleation is explicitly considered, which is, to our knowledge, not implemented in oxygen precipitation modeling so far. The calibration of the two free parameters of the model is achieved using the experimental data of 19 different thermal process combinations performed using a single material. This results in a good agreement not only for the material used for calibration but also for other silicon materials. A validation based on passivated emitter and rear cells as well as on test structures confirms the ability of the model to predict bulk carrier lifetimes after solar cell processing

It was fOlllld as long ago as 1954 that heating oxygen rich silicon to around 450°C produced electrical active defects - the so called thermal donors. The inference was that the donors were created by some defect produced by the aggregation of oxygen. Since then, there has been an enor mous amount of work carried out to elucidate the detailed mechanism by which they, and other defects, are generated. This task has been made all the more relevant as silicon is one of the most important technological ma terials in everyday use and oxygen is its most common impurity. However, even after forty years, the details of the processes by which the donors and other defects are generated are still obscure. The difficulty of the problem is made more apparent when it is realised that there is only one oxygen atom in about ten thousand silicon atoms and so it is difficult to devise experiments to 'see' what happens during the early stages of oxygen precipitation when complexes of two, three or four 0xygen atoms are formed. However, new important new findings have emerged from experiments such as the careful monitoring of the changes in the infra red lattice absorption spectra over long durations, the observation of the growth of new bands which are correlated with electronic infra-red data, and high resolution ENDOR studies. In addition, progress has been made in the improved control of samples containing oxygen, carbon, nitrogen and hydrogen.

The vast majority of modern microelectronic devices are fabricated on single-crystal silicon wafers, which are produced predominantly by the Czochralski (CZ) melt-growth process. Important metrics that ultimately influence the quality of the silicon wafers include the concentration of impurities and the distribution of lattice defects (collectively known as microdefects). This thesis provides a multiscale quantitative modeling framework for describing physics of microdefects formation in silicon crystals, with particular emphasis on oxide precipitates.Among the most prevalent microdefects found in silicon crystals are nanoscale voids and oxide precipitates. Oxide precipitates, in particular, are critically important because they provide gettering sites for highly detrimental metallic atoms introduced during wafer processing and also enhance the mechanical strength of large-diameter wafers during high-temperature annealing. On the other hand, like any other crystalline defect species, they are undesirable in the surface region of the wafer where microelectronic devices are fabricated. Although much progress has been made with regards to oxide precipitate prediction and optimization, it has been surprisingly difficult to generate a robust, quantitative model that can accurately predict the distribution and density of precipitates over a wide range of crystal growth and wafer annealing conditions.In the first part of this thesis, a process scale model for oxide precipitation is presented. The model combines continuum mass transport balances, continuum thermodynamic and mechanical principles, and information from detailed atomic-scale simulations to describe the complex physics of coupled vacancy aggregation and oxide precipitation in silicon crystals. Results for various processing situations are shown and comparisons are made to experimental data demonstrating the predictive capability of the model.In the second part of this thesis, atomistic simulations are performed to study the stress field and strain energy of oblate spheroidal precipitates in silicon crystals as a function of precipitate shape and size. Although the stress field of a precipitate in silicon crystals may be studied within a continuum mechanics framework, atomic scale modeling does not require the idealized mechanical properties (and precipitate shapes) assumed in continuum models and therefore provides additional valuable insight. The atomistic simulations are based on a Tersoff empirical potential framework for silicon, germanium and oxygen. Stress distributions and stress energies are computed for coherent germanium precipitates and for incoherent, amorphous silicon dioxide precipitates in a crystalline silicon matrix. The impacts of precipitate size and shape are considered in detail, and for the case of oxide precipitates, special emphasis is placed on the role of interfacial relaxation. Whenever possible, the atomistic simulation results are compared with analytical solutions.