الفهرس 
Chapter 1: IntroductionOnly 14 pages are availabe for public view
 |  | from | 113 |  |  |
| | | from | 113 | | |
Abstract
The properties of silicon starting materials and its contamination by metal impurities are considered to be the major concern in silicon processing. Even extremely small concentrations of impurities, when present in a harmful form in the device area, can have a detrimental effect on the device yield. In particular, Fe is incorporated as a highly mobile and soluble species during device processing. To improve device performance, we need to understand the properties of Fe-related defects in silicon and to design efficient processing methods to remove iron from active device region. One of the most important techniques used to determine the electrical characteristics of a defect is known as Deep Level Transient Spectroscopy (DLTS). DLTS is therefore extensively used to study the concentrations and electrical properties of iron-related defects and mechanisms for their formation as well as its depth distribution. This study has revealed some exciting new aspects of the well-studied Fe impurity in silicon. As a result of the present studies, our knowledge on aluminum gettering has become substantially more systematic. The main results obtained from this thesis and conclusions are summarized below. This thesis is divided into four parts, firstly a description of CV characteristics at different stages of treatments. This is followed by the characterization of FeB pair in iron indiffused samples and subsequently the effects of different protocols of wet chemical etching. The interaction between hydrogen released from the wet chemical etching and iron, and its partial passivation of boron are discussed. To summarize, wet chemical etching of Fe contaminated p-type Si leads to decrease of electrically active FeB detected by DLTS. The most probable reason for this decrease is the direct dissociation of FeB pairs by hydrogen, which associated with boron passivation of the surface of the sample. Unfortunately, a rough estimate of hydrogen concentration calculated from the carrier concentration profiles of p-type samples is not unique. This gives guidelines for future work. For example, the hydrogen concentration with such etchant dilutions is usually substantially limited by temperature and therefore, it is necessary to measure the pH value before starting the wet chemical etching at the appropriate temperature. These experimental observations indicate that it is crucial to know the hydrogen activity for each mixture of the etchant solution; a process needs new technology other than the conventional analytical chemistry ways. DLTS is used to study the segregation of Fe from crystalline Si to an Al:Si liquid at its surface, which is the basic mechanism of aluminum gettering used in silicon photovoltaics. The second part contains the detailed results obtained from aluminum gettering of iron in p-type FZ silicon annealed at temperature range from 950-1100°C. Gettering of iron is examined during different treatments of annealing. Simulated and experimental studies are performed to obtain a reliable segregation coefficient for iron. Some interesting properties of segregation coefficient are also discussed in regard to the phase diagram. The measured segregation coefficient is smaller than estimates from the binary Fe:Si and Al:Fe phase diagrams. This apparent discrepancy originates from the ternary character of the system where the solubility of Fe in Si in equilibrium with the Al-doped αFeSi2 has to be taken as a reference. Our data suggest that this solubility exceeds that in the binary Fe:Si system by two orders of magnitude. However, it would be beneficial for a future study to confirm this with complementary studies, such as DLTS and Neutron Activation Analysis. Since quantitative data on this effect are not available, experimentally measured values of segregation coefficients are required if predictive simulations of Al gettering shall yield reliable results. In the third part, we briefly discuss a generalized belief that aluminum gettering injects vacancy. In this work, the quantitative marker methods, platinum and gold diffusion, are used to profile the concentration of vacancies in the bulk formed by Al alloying. Undoubtedly, the experimental data obtained from platinum and gold indiffusion was able to proof that aluminum gettering injects vacancies with homogenous depth profile. Depth profiles of Au in p-type Si have been studied by the DLTS after diffusion at 850ºC for 2h prior to and subsequent to AlG. The donor levels attributed to Pts and Aus are detected in much higher concentrations compared with that measured in samples without AlG. The increase in Pts or Aus concentration in the silicon bulk provides the most direct evidence so far that AlG with 400nm Al layer, for 50 min at 1100°C injects vacancies with non-equilibrium concentrations (1015 cm-3). The obtained results show not only the increase of vacancy concentration as a result of used AlG procedure, but also its homogenous distribution. The interface between aluminum and silicon and the formation of silicon dioxide and aluminum oxide after annealing has been investigated by HRTEM and EDX. The underlying mechanism of vacancy injection during AlG is not known so far and needs further investigations. Longer-term future research including the combination of different techniques such as DLTS and HRTEM, which has already been started, should be continued to answer the question whether the reaction of an oxide layer with the AlSi alloy might also contribute to the generation of vacancies at the AlSi/Si interface. In addition, future work will focus on understanding what forms vacancies exist after AlG and the dependence of their profile on the cooling rate or the gettering duration. For example, we will attempt to answer whether vacancies exist in the form of some complex like VO2 or as agglomerations and what concentration produced by different thickness of Al layer. This work will be accomplished using techniques such as positron annihilation spectroscopy. Based on the experimental results, AlG was able to inject vacancies into Si-wafers; consequently it can strongly influence the process of impurity gettering by Al-Si alloy and should be taking into account during computer simulation of AlG. Finally, the last part of this thesis deals with the light-induced defects after aluminum gettering of iron doped samples. The influence of experimental parameters such as iron concentration, anneal temperature, illumination time and possible association/dissociation of FeB are discussed. Particular attention has been given to the interaction of iron and vacancies released from aluminum gettering leading to a novel deep level produced under illumination. One of the most interesting results obtained was the discovery of a new Fe-related defects which has been proofed to be a metastable complex of iron-vacancy pair. It is a donor level with energy position Ev+0.33 eV, named FeD. This FeD defect is stable up to about 175°C where it dissociates reversibly in case of small iron concentrations and irreversibly for high iron concentrations. Owing to the enhanced vacancy concentration observed after AlG, we have described a tentative scenario of light-induced FeD formation and thermal dissociation involving the reaction of Fei and V to the metastable FeiV pair described theoretically in the literature. In order to be consistent with these calculations we have proposed that FeiV is a mobile complex which is a hypothesis that needs to be verified or disproved by further experiments or theoretical investigations. According to our knowledge, detection of FeD by DLTS was firstly described in this study and currently a long discussion has been started about FeD components and importance. Future work will be conducted to determine the roll of FeD on the minority carrier lifetime and diffusion length to optimize the efficiency of solar cell. A copy of published papers containing the discussed results and conclusions, have been included at the end of the thesis.