Dissertations / Theses on the topic 'Czochralski silicon'

Disertační práce se zabývá studiem defektů v monokrystalech Czochralskiho křemíku legovaných bórem. Práce studuje vznik kruhových obrazců vrstevných chyb pozorovaných na povrchu křemíkových desek po oxidaci. Hlavním cílem práce je objasnit mechanismy vzniku pozorovaného rozložení vrstevných chyb na studovaných deskách a vyvinout metody pro řízení tohoto jevu. Na základě experimentálních analýz a rozborů obecných mechanismů vzniku defektů jsou objasňovány vazby mezi vznikem defektů různého typu. Tyto jsou pak diskutovány v souvislosti s parametry krystalu i procesu jeho růstu. Takto sestavený model je využit pro vývoj procesu růstu krystalů, kterým je potlačen nadměrný vznik defektů ve studovaných deskách. Za účelem studia defektů jsou zaváděny a vyvíjeny nové analytické metody. Disertační práce byla vytvořena za podpory ON Semiconductor Czech Republic, Rožnov pod Radhoštěm.

For the purpose of this thesis work, four n-type CZ silicon ingots with different crown tapered angle and shouldering area were characterized in order to understand the oxygen behavior and related defects on ingot top cuts and its influence on material lifetime. A p-type CZ silicon ingot was also characterized in order to have a reference material for comparison. Differences in lifetime between the crowns were observed and a strong correlation between the crown tapered angle and oxygen concentration and distribution was established. The crown with the higher tapered angle has the highest lifetime. In contrast, the crown with the lower crown tapered angle has the lowest lifetime. The crystal body quality can be influenced by the top ingot quality in what concerns the interstitial oxygen concentration and distribution. Thus, analyzing the early body of each ingot, it might be possible to predict the crystal body quality.

Low voltage power electronics are made from dislocation free silicon heavily doped with arsenic or antimony to provide low electrical resistivity. Attempts to grow crystals with decreased resistivity have led to a higher probability of twinning during growth, so that the crystal no longer possesses the required crystallographic orientation for device fabrication. The source of the twins must be identified so that crystal growth process conditions can be designed to eliminate this defect mechanism, allowing lower resistivity crystals to be grown reliably. In lightly doped crystals, twinning was ascribed to presence of carbon impurity or a low probability atomic stacking accident, neither of which should be affected by increased concentration of arsenic or antimony. Crystals that twinned during growth were characterized by resistivity, Laue back-reflection x-ray diffraction, optical and scanning electron microscopy, energy dispersive x-ray spectroscopy, spreading resistance, x-ray computed tomography and electron backscatter diffraction. The twin nucleation site of silicon crystals that were grown heavily doped with arsenic or antimony were compared to lightly doped crystals which twinned, and crystals that exhibited other defects. The initial twinning in the <100> orientation heavily doped crystals occurred from small gas bubbles bursting at a {111} facet at the three phase boundary, and forming a twin orientation domain on that facet. The gas bubbles likely consist of argon, the process gas used during solidification to remove silicon monoxide gas from the growth system. The higher levels of arsenic or antimony dopant may have changed the silicon surface tension, or provided additional impurities into the liquid silicon. Either effect may have changed the number or size of argon bubbles in the liquid silicon, leading to a higher incidence of gas bubbles near the {111} facet during solidification. Similar but smaller crater features were observed on two lightly boron-doped silicon crystals that twinned. Two other lightly doped crystals formed twins from carbon inclusions, consistent with carbon as a cause. Some heavily-doped twinned samples also show high concentrations of metals at the twin nucleation site, which could affect surface energy. Measurement of the geometry of crystal surface-to-facet radius eliminated a recently-proposed twin nucleation theory from consideration. Constitutional supercooling was demonstrated to not be a major contributing factor to twin nucleation. It was shown that deliberately introducing additional arsenic dopant during solidification would nucleate twins, but twins did not occur if only elemental carbon was introduced.

This thesis presents an investigation of the phenomenon of efficient, room temperature luminescence from dislocation-engineered (DE) silicon. Previous work had demonstrated that the introduction of near-surface dislocation loops to a silicon substrate by boron ion implantation and high temperature annealing produced efficient electroluminescence at room temperature. However, the mechanism by which high efficiency luminescence is produced was not understood. A wide matrix of specimens containing dislocations was fabricated by a variety of methods, including ion implantation, and their luminescence efficiencies were correlated to their physical properties. Transmission electron microscopy was used to characterise the defect structures created by ion implantation. In the majority of specimens a band of dislocation loops in close proximity to the surface was observed. The dislocation loops were shown to be consistent with a mixture of Frank and perfect dislocation loops, the relative proportions of which were dependent upon processing conditions. The thermal evolution of the dislocation loop size distribution was investigated. For the first time, a size distribution displaying a double peak was observed. The size distribution was shown to be consistent with the Gaussian distribution of two defect populations of different mean diameter. The thermal evolution of the size distribution was investigated in silicon implanted samples. A flux of self-interstitials from Frank dislocation loops to perfect dislocation loops was deduced. The evolution of the dislocation loop sizes was found to be consistent with Ostwald ripening. Cathodoluminescence (CL) was used to investigate the luminescent properties of silicon at room temperature for the first time. A new CL system was installed for this work, initially the CL system was characterised and a routine to ensure a high degree of reproducibility was formed. The luminescence mechanism of DE-silicon was shown to be the same as in unprocessed silicon wafers; TO phonon-assisted recombination. The mechanism of enhanced luminescence from DE-silicon was unambiguously shown to be due to the gettering of electrically active impurities from the specimen bulk. A reduction in the bulk transition metal impurity concentration of up to 35 times was inferred. In samples which were implanted with boron the degree of gettering was found to show a logarithmic dependence on the dislocation density. Using a crosssectional mapping technique, implanted samples were shown to contain a lower concentration of transition metal impurities throughout the entire wafer in comparison to as-received, unprocessed specimens. Furthermore, the impurity concentration was found to be lowest in close proximity to the band of dislocation loops. The dislocation loops were found to act as non-radiative recombination centres and their strength was strongly influenced by the local carrier concentration. The high doping levels of samples implanted with boron were found to minimise the non-radiative recombination action of the dislocations. Low temperature annealing was used to improve the luminescence efficiency of DE-silicon further.

Cast multi-crystalline silicon substrates are used in more than 50% of the photovoltaic modules produced today. The random grain orientations of multi-crystalline silicon wafers inhibit the formation of an effective surface texturing using conventional techniques. The other main substrate used is single crystalline Czochralski wafers (~30% of the market share). Czochralski silicon material is known to suffer from the formation of a metastable defect under carrier injection, sometimes referred to as light induced degradation (LID). Light induced degradation in low-cost photovoltaic grade silicon is studied. Trap formation is shown to occur at temperatures above 200oC. Efficiency degradation reduced from 0.75% to 0.24% when the cell thickness was reduced from 378 to 157m. The presence of light induced degradation in ribbon silicon solar cells is documented for the first time in this thesis. Trap generation and annihilation are observed in high lifetime regions of multi-crystalline silicon samples. No degradation was observed over a 24-hour period at 25oC in high efficiency ribbon solar cells (>16%), but at an elevated temperature of ~75oC, appreciable efficiency degradation was observed. Czochralski silicon solar cells showed full degradation within 24 hours at 25o C. Part two of this thesis involves the development of a surface texturing suitable for all crystalline silicon. Only 6 to 10 seconds in a 200:1 HF to HNO3 solution at room temperature allows for the formation of an effective porous silicon anti-reflection coating. This resulted in a porous silicon anti-reflection coated solar cell efficiency of 15.3% on a float zone Si sample with an excellent fill factor (78.7%). The typical process used in the literature involves porous silicon etching as the final step in the solar cell fabrication sequence. The major problem associated with this process sequence is fill factor degradation. This problem was overcome in this research by porous silicon etching prior to cell processing. It is shown that incorporating an acid texture prior to porous silicon etching can improve the surface reflectance for cast multi-crystalline and Czochralski silicon samples. Solar cell efficiencies of 14.8% for Cz Si and 13.6% for cast mc-Si were achieved.

Combinations of analytical and experimental results indicate that deep level doping of Czochralski grown silicon wafers is capable of providing very high resistivity wafers suitable for silicon-on-insulator (SOI), integrated passive devices (IPD) and 3D integration configurations. Deep level doping involves adding trace elements to silicon that compensate for background free carriers introduced by impurities in the silicon and pin the Fermi level near the mid bandgap intrinsic level. Starting from n-type Czochralski-silicon wafers with a nominal resistivity of 50 Ωcm, gold ion implantation and subsequent annealing were used to increase the resistivity of silicon wafers by up to 3 orders of magnitude, to values as high as 93 kΩcm. Hall measurements performed over a large temperature range show that the increase in resistivity is solely due to a decrease in carrier concentration and not a decrease in mobility. The carrier concentration is only one order of magnitude larger than that of intrinsic silicon over a temperature range of 200-360 K. Hall results also show that the resistivity of the compensated material remains up to two orders of magnitude larger than that of the uncompensated material at near operating temperatures. High frequency attenuation measurements in the 1-67 GHz range for coplanar waveguides show attenuation reductions of up to 76% from 0.76 dB/mm to 0.18 dB/mm at 10 GHz for those fabricated on uncompensated and compensated silicon respectively. Spiral inductors fabricated on both compensated and uncompensated silicon show up to a factor of 10 increase in the maximum quality factor from 0.3 to 3.1 for inductors on uncompensated and compensated silicon respectively. A 70% increase in maximum quality factor from 9 to 15.2 is exhibited by inductors commercially fabricated on compensated silicon when compared to those on float-zone silicon. The coplanar waveguide and spiral inductor results provide clear evidence that deep level dopant compensation is effective in improving the performance of passive devices in the GHz frequency range.

Neste trabalho é estudado o efeito do carbono na formação de defeitos em silício Czochralski crescido na direção em amostras submetidas a tratamentos térmicos variados. Medidas de espalhamento difuso de raios-X, espectroscopia de infravermelho, medidas de resistividade, topografia de raios-X e microscopia eletrônica de transmissão mostraram que os defeitos nas amostras \"como crescidas\" podem ser relacionados com os microdefeitos tipo B. Tratamento térmico a 450ºC mostrou a presença de vacâncias nas amostras com baixa concentração de carbono enquanto que nas amostras com alta concentração de carbono ocorre a inibição da formação dos doadores térmicos (\"Thermal Donors - TD\"). Os resultados confirmam os modelos de Newman e Mathiot para a geração dos TD. Para tratamento térmico a 650ºC o carbono promove a formação de Novos Doadores (\"New Donors - ND\"). Os resultados mostram que estes defeitos são de natureza predominante de vacância e concordam com os modelos de geração que envolvem átomos de oxigênio substitucional. Os doadores observados a 550ºC puderam ser relacionados aos Novos Doadores Térmicos (\"New Thermal Donors - NTD\") observados por Kamiura et al..
Effect of carbon concentration upon defect formation in oxygen rich Czochralski grown silicon has been investigated by combining various furnace thermal anneals. Diffuse X-ray scattering, infrared spectroscopy, resistivity, x-ray topography, and transmission electron microscopy have shown that defects in as-grown samples could be related to the B swirls. 450ºC anneals have shown the presence of vacancies in low carbon samples while high carbon concentration inhibited Thermal Donor (TD) formation. Our results confirm models by Newman and Mathiot for thermal donors generation. For 650ºC anneals carbon promotes New Donors (ND) formation. Our results show that these defects are mainly vacancy in nature and agrees with the substitutional oxygen models proposed for these donors. Donor formation was observed at 550ºC which could be related to New Thermal Donors (NTD) proposed by Kamiura et al..

Les cellules photovoltaïques à base de silicium cristallin représentent plus de 90% du marché photovoltaïque mondial. Des architectures de cellules à haut rendement de conversion sont actuellement développées. Pour atteindre leurs performances maximales, ces architectures nécessitent néanmoins une amélioration des propriétés électriques des substrats de silicium cristallin. Les objectifs de cette thèse sont d’identifier les défauts limitant les propriétés électriques de ces substrats, de comprendre les mécanismes menant à leur formation et de proposer des moyens permettant leur neutralisation. Les matériaux étudiés sont des plaquettes de silicium Czochralski de type n, généralement utilisé pour les applications à haut rendement. Le four de tirage Czochralski a d’abord été modélisé afin de comprendre comment le passé thermique subi par le lingot de silicium lors de la cristallisation affecte la génération des défauts. Ces travaux ont été confirmés via des confrontations avec des données expérimentales, en utilisant une méthode originale développée dans le cadre de ce travail. Nous avons ensuite étudié l’influence du budget thermique lié aux procédés de fabrication des cellules sur la population de défauts. Nous avons ainsi pu montrer que la nature des défauts limitant les propriétés électriques du silicium était grandement modifiée selon le procédé de fabrication de cellules utilisé. Nous avons en outre mis en évidence une dégradation inattendue des propriétés électriques du silicium Czochralski de type n sous illumination, liée à la formation d’un défaut volumique inconnu. Les conditions de formation et de suppression de ce défaut ont été étudiées en profondeur. Enfin, les principaux défauts limitant les propriétés électriques du silicium ayant été identifiés et les mécanismes menant à leur formation compris, nous proposons dans un dernier chapitre des nouvelles techniques de caractérisation permettant de détecter les plaquettes défectueuses en début de ligne de production de cellules photovoltaïques, et ce à une cadence industrielle
Photovoltaic solar cells based on crystalline silicon represent more than 90% of the worldwide photovoltaic market. High efficiency solar cell architectures are currently being developed. In order to allow their maximal performances to be reached, the electronic properties of their crystalline silicon substrate must however be enhanced. The goals of the present work are to identify the defects limiting the electronic properties of the substrate, to understand the mechanisms leading to their formation and to propose routes for their neutralization. The studied materials are n-type Czochralski silicon wafers, usually used as substrates for high efficiency photovoltaic applications. The Czochralski puller was first modeled in order to understand how the thermal history experienced by the silicon ingot during crystallization affects the defects generation. This study were validated through the comparison with experimental data using an original method developed in the frame of this work. We then studied the influence of the thermal budget associated to solar cell fabrication processes on the defects population. We thus showed that the nature of lifetime-limiting defects was completely changed depending on the solar cell fabrication process. Besides, we evidenced an unexpected degradation of the electronic properties of n-type Czochralski silicon under illumination, related to the formation of an unknown bulk defect. The formation and deactivation features of this defect were extensively studied. Finally, the main limiting defects being identified and the mechanisms resulting in their formation understood, we propose in a last chapter new characterization techniques for the detection of defective wafers at the beginning of production lines at an industrial throughput

Visando compreender os fluxos na fase líquida do crescimento de silício pelo método Czochralski, é feita a Simulação Numérica do silício fundido, resolvendo-se as equações que governam o fenômeno da convecção forçada no fluido: Balanço de Quantidade de Movimento e Balanço de Massa. A técnica numérica escolhida é a de Elementos Finitos, onde é utilizada a formulação de Galerkin, com aproximações quadráticas nas componentes da velocidade e linear na pressão. A partir de várias combinações de rotações cadinho-cristal, os perfis de velocidade obtidos são analisados com relação aos efeitos de incorporação de impurezas e/ou dopantes no cristal em crescimento.
In order to visualise the flow conditions during crystal growth of Silicon by the Czochralski technique, a numerical simulation is done. It is used the Finite Element Method with the Galerkin Formulation , and with quadratic approximations on the components of the velocity and linear approximations on the pressure. Many combinations of crystal and crucible rotations are analised and discussed considering optimal growth conditions.

A novel dislocation locking technique is used to study the behaviour of nitrogen and oxygen in silicon. Specimens containing well-defined arrays of dislocation half-loops are subjected to isothermal anneals of controlled duration, during which nitrogen or oxygen diffuses to the dislocations. The stress required to move the dislocations away from the impurities is then measured. Measurement of this unlocking stress as a function of annealing time and temperature allows information on the transport of nitrogen and oxygen to be deduced. Despite being present in a concentration of just 3E14cm-3 in some specimens, nitrogen is found to provide substantial benefits to the mechanical properties of float-zone silicon (FZ-Si). The segregation of nitrogen at dislocations is stable to at least 1200 degrees centigrade and the unlocking stress measured at 550 degrees centigrade is of similar magnitude to that found previously for oxygen in Czochralski silicon (Cz-Si). The unlocking stress initially rises linearly with annealing time, before it takes a constant value. The rate of the initial rise is dependent on temperature and the 1.5eV activation energy found agrees with that found previously. The rate of the initial rise also depends on nitrogen concentration. In the 500 to 700 degrees centigrade temperature range, the unlocking stress is found to decrease linearly as the temperature at which the unlocking process takes place increases. The results of a pre-annealing experiment confirm that oxygen monomers and dimers in Cz-Si exist in thermodynamic equilibrium at 550 degrees centigrade. Numerical simulation of oxygen diffusion to dislocations allows values of the effective diffusivity of oxygen in Cz-Si with four different oxygen concentrations to be deduced. At 500 degrees centigrade, the effective diffusivity depends upon oxygen concentration in a way which is consistent with oxygen dimers being responsible for transport. The transport of oxygen in Cz-Si at 550 to 600 degrees centigrade is found to be unaffected by nitrogen doping at a level of 2.1E15cm-3. The dislocation locking technique has also been used to study the effect of high concentrations of shallow dopants on oxygen transport in Cz-Si in the 350 to 550 degrees centigrade temperature range. Oxygen transport has been found to be unaffected by a high antimony concentration ~3E18cm-3, but is found to be enhanced by, on average, a factor of approximately 44 in Cz-Si with a high boron concentration ~5E18cm-3. Furthermore, deep-level transient spectroscopy (DLTS) and high-resolution DLTS (HR-DLTS) are used to study the electrical activity of defects in silicon. A deep-level with an enthalpy of 0.50eV and a concentration of order 10E11cm-3 is found in n-type nitrogen-doped FZ-Si and n-type nitrogen-doped neutron transmutation doped FZ-Si. No additional deep-levels are found in either material, for which the detection limit is 6E10cm-3. No deep-levels are found in p-type nitrogen-doped Cz-Si, for which the detection limit is approximately 10E12cm-3. DLTS and HR-DLTS are also used to investigate the electrical activity of oxygen-decorated dislocations in Cz-Si and states associated with oxygen at dislocation cores have been identified.

A determinação do estado de deformação de monocristais planos através da técnica pseudo-Kossel, combinada com o estudo do espalhamento difuso de raios X próximo a uma reflexão de Bragg, permite uma caracterização dos defeitos quanto a sua natureza, tamanho e simetria se houver poucas variedades de defeitos diferentes ou quando ocorrer a predominância de um deles. Estudamos monocristais de silício Czochralski dopados com antimônio ou com boro, e que também continham oxigênio e carbono devido ao processo de crescimento. Estas amostras apresentam aglomerados de defeitos pontuais com dimensões características em torno de 102 - 103 nm. Medidas de espalhamentos interplanares mostram que estas são mais fortemente influenciadas pela direção de crescimento que pela concentração de dopantes, e que as maiores alterações ocorrem à medida que 0 angulo entre a direção de crescimento e os planos diminuem. A partir das medidas dos espalhamentos interplanares podemos levantar 0 estado de deformação dos cristais através do tensor de \"strain\" e foi possível analisar a configuração \"stressstrain\". Desta análise obtivemos, além da variação do parâmetro de rede, 0 tensor de \"stress\" e dele 0 tensor momento de dipolo elástico P. 0 tensor P pode nos dar informações acerca da simetria do defeito &nos permitiu calcular 0 espalhamento difuso de raios X próximo de uma reflexão de Bragg. Vimos que algumas amostras estão deformadas numa configuração muito próxima daquela de cisalhamento simples e que a simetria do tensor P indica que em praticamente todas as amostras tem uma variedade grande de defeitos diferentes. Uma das amostras, Si com Sb (CSb= 8. 1018 dopantes/ Cm3)apresenta um tensor P pr6ximo daquele gerado por defeito que temsimetria ortorrombica e, para ela foram construidas curvas de isso intensidades em torno do ponto (004) da rede recíproca, calculados a partir das componentes de P, e que foram comparadas com as curvas de \"rocking\" experimentais.
The accurate determination of strain distributions in single crystals through the pseudo-Kossel technique together with the analysis of the diffuse X-ray scattering near a Bragg reflection permits the characterization (nature, size and shape) of predominant defects. Czochralski silicon single crystals doped by diffusion during growth with antimony and boron were studied. These samples presented clusters of 102 - 103 nm. Interplanar spacings determined by the pseudo- Kossel method showed a significant dependence on growth direction,planes parallel to growth direction were the most affected. From and the dipole tensor P determined. This dipole tensor was employed in the calculation of the diffuse X-ray scattering. For one of the samples Si:Sb P presents an orthorrombic simetry. The results for this sample were compared with experimental rocking.

Ce travail a pour but de comprendre les effets de deux principaux défauts liés à l’oxygène, les complexes bore-oxygène et les donneurs thermiques, sur les propriétés électriques et photovoltaïques du silicium. Plus particulièrement, les interactions des impuretés isovalentes, connues pour modifier la distribution spatiale de l’oxygène, avec ces défauts ont été étudiées. Deux protocoles expérimentaux ont d’abord été développés pour évaluer la dégradation de la durée de vie des porteurs de charge sous éclairement dans le silicium riche en fer. Ensuite, il a été mis en évidence que l’introduction de germanium et d’étain en très grande quantité dans le silicium n’influence pas de façon significative le rendement de conversion des cellules. Cependant, contrairement à ce qui a été récemment avancé dans la littérature, aucune limitation due au co-dopage au germanium ou à l’étain de la dégradation sous éclairement des performances photovoltaïques n’a été observée. Par contre, il a été montré que le carbone entraîne un ralentissement de la dégradation due aux complexes bore-oxygène. Egalement, contrairement à l’étain qui n’influence pas la génération des donneurs thermiques, le germanium conduit à un ralentissement de la formation de ces défauts. Une expression empirique a été proposée pour rendre compte de cet effet, et ce pour une large gamme de teneurs en germanium. Enfin, dans le silicium très dopé et très compensé, la génération des donneurs thermiques est identique au cas du silicium standard. Ceci constitue un résultat marquant puisqu’il permet de valider par l’expérience le fait que la formation des donneurs thermiques est limitée par la teneur en électrons
This study aims at understanding the effects of two main defects related to oxygen, the boron-oxygen complexes (responsible for light-induced degradation of the carrier lifetime) and the thermal donors (among other things, responsible for variations of the conductivity), on the electric and photovoltaic properties of silicon. More precisely, the interactions of isovalent impurities, known for modifying the oxygen spatial distribution, with these defects were studied. Two experimental protocols were first developed to evaluate the light-induced degradation of the carrier lifetime in iron-rich silicon. Then, the introduction in silicon of germanium and tin in high quantity were shown not to significantly influence the conversion efficiency of the cells. However, contrary to recent studies from the literature, no reduction due to germanium co-doping or to tin co-doping of the light-induced degradation of the photovoltaic performances was observed. However carbon was shown to lead to a slowdown of the degradation due to boron-oxygen complexes. Moreover contrary to tin which has no influence on the thermal donor generation, germanium slows down their formation. An empirical expression has been proposed to take into account this effect for a large range of germanium concentrations. Eventually in highly doped and compensated silicon, the thermal donor generation is identical as in conventional silicon, which experimentally confirms that the thermal donor formation is limited by the electron density

Thesis (Ph.D.)--Massachusetts Institute of Technology, Dept. of Chemical Engineering, 2000.
Includes bibliographical references (p. 367-387).
Microdefect formation in crystalline material is strongly correlated to the processing conditions for growth of crystals important in microelectronic processing. The geometry and operation conditions for crystal growth systems affect the temperature profile in the crystal and melt, which influences microdefect formation. The objectives of this thesis are to formulate the computational framework to establish the linkage between microdefect formation in crystal and proce5sing conditions of crystal growth system. The research focuses on two industrially important crystal growth problems: Czochralski (CZ) growth of single-crystal silicon and growth of bismuth germanium oxide (Bi4Ge30 12:BGO) by the vertical Bridgman method. A sequential, two-step approach is taken for linking mathematical modeling between processing conditions and microdefect formation in crystals. An accurate model of heat transfer in CZ growth of silicon is developed by including all the components in the system. Microdefoct formation in the crystal is then modEled by imposing the temperature profile obtained by the global heat transfer simulation. The integrated hydrodynamics thermal-capillary model (IHTCM) of CZ crystal growth includes radiative and conductive heat transfer between all components of the system. An important component of this simulation is the incorporation of a model of turbulence in the melt. A low Reynolds number k-c model is incorporated into the IHTCM for CZ system. The coupled k-c/IHTCM is applicable to any CZ system geometry and operating conditions because of the self-consistency of the model. Also a robust numerical solution method is developed to solve numerically unstable k-c equations by a finite-element approximation. The comparison between simulations and experiments for CZ growth of an 8" diameter crystal shows semi-quantitative agreement in melt/crystal interface, oxygen concentration in the crystal, and the location of a neutral zone, where the concentrations of two intrinsic point defects balance, in the crystal. Microdefect formation in CZ silicon is modeled with intrinsic point defects (vacancies and self-interstitials) and their agglomerates. The model is two-dimensional in space and predicts the radial profiles of point defects, which are determined near the melt/crystal interface, and the axial development of size distribution of voids and self-interstitial agglomerates, which is a function of point defect supersaturation and the temperature profile. The model provides quantitative links between operating conditions and microdefect distribution in the entire crystal. An effective numerical method with parallel processing is developed using a mixed local discontinuous Galerkin method. The predicted agglomeration temperatures and densities for vacancy and self-interstitial clusters are within the ranges of experimental data. The predictions also include the location for ring-like oxidation-induced stacking fault (OSF) formation, assuming the OSF-ring is formed at the radial location with the peak in residual vacancy concentration after the onset of vacancy agglomeration. The simulations clearly reproduce the radial distribution of microdefects observed by experiments. Starting from the crystal center and moving to the edge, the simulations predict a void region, the OSF-ring as a region of locally high vacancy concentration, a defect free region, a region dominated by self-interstitial clusters, and finally a defect free region near the crystal edge. The defect free region at the crystal edge results from the radial diffusion of point defects caused by reactions at the crystal surface. The heat transfer model in the vertical Bridgman system for BGO crystal growth incorporates internal radiation in the semi-transparent BGO crystal and conduction and radiation for all components of the heat transfer system. A band approximation is used to model internal radiation in the crystal. The global heat transfer model provides quantitative understanding of the heat transfer within the semi-transparent BGO crystal as well as in the entire system. Comparison of the temperature profile at the crucible wall between simulations and experiments for the large 11 cm diameter BGO crystal growth shows good agreement. The detailed analysis of heat transfer near the solidification interface gives insight for the control of bubble defects in BGO crystal formed by constitutional supercooling. The framework for numerical simulations developed in this thesis quantitatively demonstrates the linkage between processing conditions and microdefect formation in crystalline material. The linkage is established by the coupling of self-consistent modeling of global heat transfer in the crystal growth systems and microdefect formation in crystals.
by Tatsuo Mori.
Ph.D.

The thesis deals with the study and analysis of crystallographic defects on the surface of silicon wafers produced by Czochralski method. It focuses primarily on growth defects and oxygen precipitates, which play an important role in the development of appropriate nucleation centers for growth of stacking faults. The growth of stacking faults near the surface of silicon wafers is supported by their oxidation and selective etching. Such a highlighted stacking faults are known as the OISF (Oxidation Induced Stacking Fault). Spatial distribution of OISF on the wafer gives feedback to the process of pulling silicon single crystal and wafers surface quality. Moreover the work describes the device for automatic detection and analysis of OISF, which was developed for ON Semiconductor company in Rožnov Radhoštěm.

Electrically detected magnetic resonance (EDMR) is a sensitive spectroscopic technique, which can be used to readout few to single electron spins in semiconductor and carbon nanodevices for applications in solid state quantum information processing (QIP). Since only electrically active defects contribute to the EDMR signal, this technique can be used further to investigate defects and impurities in photovoltaic devices, in which they limit the sunlight-to-energy conversion efficiency significantly. Here, I employ X-band EDMR for semiconductor defect analysis and identify the most important recombination centres in Czochralski silicon with oxide precipitates, which can be intentionally grown to confine detrimental metallic impurities to inactive regions of the wafer in order to serve as a defect-free substrate for modern silicon photovoltaic devices. Those experiments show that oxide precipitation is accompanied by the formation of silicon dangling bonds. Furthermore, I describe a very promising route towards the fabrication and readout of few to single electron spins in carbon nanotube devices, which can be characterised structurally via transmission electron microscopy in order to relate their electrical and spin properties with their structure. Finally, I employ EDMR to read out electron spin states in donor-doped silicon field-effect transistors as a prerequisite for their application in QIP. I report on a novel cryogenic probe head for EDMR experiments in resonant microwave cavities operating at 0.35 T (9.7 GHz, X-band) and 3.34 T (94 GHz, W-band). This approach overcomes the inherent limitations of conventional X-band EDMR and permits the investigation of paramagnetic states with a higher spectroscopic resolution and signal intensity. Both advantages are demonstrated and discussed. I further report on a novel mechanism giving rise to the EDMR effect in donor-doped silicon field-effect transistors, which is capable of explaining why the EDMR signal intensities of the conduction electrons are enhanced by a factor of ∼100, while the donor resonance signals increase by a factor of ∼20 from X- to W-band only. The spin-relaxation and dephasing times are extracted from a series of pulsed-EDMR measurements and confirm this model. The author gratefully acknowledges funding from Trinity College Oxford, Department of Materials, EPSRC DTA, and Konrad-Adenauer-Stiftung e.V. (Begabtenförderung).

Made available in DSpace on 2014-10-09T12:35:01Z (GMT). No. of bitstreams: 0
Made available in DSpace on 2014-10-09T14:00:05Z (GMT). No. of bitstreams: 0
Tese (Doutoramento)
IPEN/T
Instituto de Pesquisas Energeticas e Nucleares - IPEN-CNEN/SP

La partie principale de cette these est consacree a l'etude du compose a fermions lourds uir#2si#2. L'elaboration de monocristaux de grande purete chimique en a ete une etape importante. L'etude et la mise au point d'un nouveau four de tirage czochralski a chauffage tri-arcs sont decrites. Cet appareillage, entierement concu en technique ultravide, a permis la synthese de monocristaux dont les caracteristiques sont meilleures que celles publiees, jusqu'a ce jour. La recherche de nouveaux intermetalliques ternaires proches, dans le diagramme d'equilibre, des composes a fermions lourds de formule generale 1-2-2 est egalement abordee. Trois nouvelles familles de composes ont ete ainsi mises en evidence: m#2t#3x, m#6t#1#6x#7 et m#4t#1#3x#9 (avec m = terre rare ou uranium, t = element de transition et x = silicium ou germanium). Leurs proprietes cristallographiques et physiques ont ete etudiees. Les composes des deux premieres familles sont paramagnetiques de type pauli. Ceux de la troisieme sont, en revanche, tous ordonnes, avec des comportements magnetiques varies. En particulier, nos mesures revelent que u#4ir#1#3si#9 et u#4ir#1#3ge#9 sont deux nouveaux fermions lourds presentant trois transitions magnetiques. Les etudes realisees au cours de ce travail sur uir#2si#2 confirment toute la complexite de ce compose a fermions lourds. Les caracterisations structurales de nos monocristaux ont mis en evidence, pour la premiere fois, l'existence de substitutions d'iridium, par du silicium, sur les deux sites cristallographiques de l'iridium. Les sites de silicium ne sont pas affectes par ce melange, ce qui n'avait jamais ete soupconne. L'influence du recuit et de la stoechiometrie sur les proprietes de uir#2si#2 est egalement presentee. Les etudes par diffraction des neutrons ont montre que uir#2si#2 est antiferromagnetique de type i, que la densite d'aimantation est isotrope et presque uniquement portee par l'uranium et qu'il existe une contribution positive supplementaire attribuable aux electrons de conduction

[EN] Crystalline silicon solar cell performance and durability is highly influenced by substrate structure and composition. In this sense, the presence of impurities and lattice imperfections degrades the electrical behavior of the substrates limiting the power conversion efficiency potential of photovoltaic devices. The influence of material electrical behavior on device performance, is further increased in next generation silicon solar cells demanding high quality substrates to reach high efficiency. A industrial-like Czochralski (Cz) p-type, ingot was grown and sawed in wafers with resistivity ranging from 0.7 Ohm.cm to 2.1 Ohm.cm and thicknesses of 200 µm. The characterization of the radial and longitudinal distribution of resistivity as well as carbon, oxygen and iron content lead to a broad characterization ot the impurity distribution at the ingot level. The low carbon content ([C]<5E15 cm-3) contrasts with remarkable high interstitial oxygen (Oi). On the other hand, while effective carrier lifetime is not significativelly affected by the high Oi content, low Fei concentrations in the ingot tail ([Fei]>1E11 cm-3) and surface ([Fei]>1E12 cm-3) led to significant lifetime degradation. In addition, lifetime mapping of the tail of the ingot showed high density of crystal dislocations. The influence of the boron and phosphorus diffusion on the electrical performance of silicon substrates was studied. Phosphorus diffusion from a POCl3 source showed a beneficial effect on lifetime that was associated with the reduction of Fei content (external gettering effect). On the other hand, boron diffusion from a BCl3 source revealed a detrimental effect that could be related to a substantial increase of [Fei] in the range of one order of magnitude. A process sequence including consecutive boron and phosphorus diffusions showed a substantial degradation of lifetime at the ingot top that could be explained by Oi precipitation rings. The influence of the crystal defects, found at the ingot tail, on lifetime degradation after the diffusion processes was also confirmed. Al-BSF cells were manufactured from a batch of wafers extracted along the ingot with efficiencies ranging between 17.8% and 18.2% and showing low dependence of solar cell performance on the solidified fraction of the ingot. Internal Quantum Efficiency (IQE) and minority carrier diffusion length (Ld) mappings further confirmed the external gettering effect developed by the phosphorus diffusion. In a further stage, Passivated Emiteer Rear Totally-diffused (PERT) cells were manufactured with efficiencies in the range of 18.8% and 19.6%. As opposed to Al-BSF cells, electrical parameters of PERT cells were influenced by solidified fraction with a reduction of cell performance near the top and tail of the ingot. In a later stage, an additional batch of solar cells of three different architectures, including Al-BSF, PERT and Passivated Emitter Rear Contact (PERC), was successfully manufactured from the same region of a Cz boron-doped ingot. Real time monitoring of the Voc enabled the study of the influence of the cell architecture on the Light Induced Degradation (LID) and Regeneration (LIR) kinetics and amplitudes. Strong correlation between the cell architectures and the degradation amplitude were observed with higher efficiency losses for PERC cells mainly due to a higher sensibility to bulk carrier lifetime degradation. In addition, the potential of LIR processes to permanently suppress the LID effects on the studied solar cell architectures was confirmed. Main causes driving the influence of the architecture in LID amplitude and LIR kinetics were further investigated.
[ES] Las características de los substratos sobre los que son fabricadas las células fotovoltaicas cristalinas constituyen un elemento fundamental en su comportamiento y durabilidad. En este sentido, la presencia de impurezas e imperfecciones en la red cristalina reduce el comportamiento eléctrico de los substratos limitando el potencial de generación eléctrica de las células. Esta limitación se acrecienta en las células de nueva generación, en las que su potencial para aumentar la eficiencia de conversión fotovoltaica solo puede hacerse efectivo al utilizar substratos con altas prestaciones eléctricas. En la presente investigación se ha procedido a la cristalización mediante la técnica Czochralski (Cz) de un lingote monocristalino, tipo p con resistividad en el rango de 0.7-2.1 Ohm.cm y tamaño industrial. Su posterior corte en obleas de 200 µm de espesor nominal ha permitido disponer de obleas distribuidas a lo largo del lingote en una fracción solidificada entre el 2.0% y el 99.4%. La caracterización del lingote ha incluido la medida de la resistividad así como la reconstrucción longitudinal y radial de su concentración en carbono, oxígeno y hierro. El reducido contenido en carbono obtenido ([C]<5E15 cm-3) contrasta con una elevada concentración en oxígeno intersticial (Oi) sin efectos apreciables en las propiedades eléctricas del lingote. Contrariamente, la reducción del tiempo de vida efectivo en la periferia y cola del lingote, es atribuido al aumento de la concentración en hierro intersticial [Fei] en dichas regiones. El estudio de la influencia de las difusiones en la evolución de las propiedades eléctricas del material incluyó difusiones de boro y fósforo mediante BCl3 y POCl3. El estudio ha constatado el efecto benéfico de la difusión de fósforo en la periferia del lingote asociado a la reducción del contenido en Fei. Por el contrario, se ha constatado el efecto perjudicial de la difusión de boro explicado en buena medida por un aumento superior a un orden de magnitud en [Fei]. La aplicación sucesiva de difusiones de boro y fósforo ha revelado una degradación apreciable en la cabeza del lingote tras la aparición de patrones circulares en la medida del tiempo de vida característicos de la precipitación de Oi. Se ha podido constatar la influencia de la elevada densidad de dislocaciones, observada en la cola del lingote, en la respuesta del substrato ante las difusiones. La fabricación de células de tecnología Al-BSF, con eficiencias comprendidas entre 17.8% y 18.2%, ha evidenciado una reducida influencia de las variaciones de las propiedades eléctricas del substrato en el comportamiento de las células finales. A su vez, la cartografía de la eficiencia quántica interna (IQE) y de la longitud de difusión (Ld) ha confirmado el efecto benéfico de la difusión de boro en la periferia y cola del lingote. La fabricación de células Passivated Emiteer Rear Totally-difused (PERT), ha permitido la obtención de dispositivos con eficiencias entre 18.8% y 19.6%. Las eficiencias muestran valores máximos en el centro del lingote y una reducción apreciable en la cabeza y cola.En una última etapa, se ha procedido a la fabricación de arquitecturas de células fotovoltaicas Al-BSF, PERT y Passivated Emitter Rear Contact (PERC), con el objetivo de estudiar los procesos de degradación y regeneración asociados a la formación de pares boro-oxígeno (B-O). El estudio de la evolución de Voc con el tiempo ha constatado una dependencia de la amplitud de degradación y cinética de regeneración con la arquitectura de célula. En este sentido se ha observado una degradación significativamente superior en las células PERC principalmente atribuible a una mayor sensibilidad a degradaciones de la calidad del substrato. A su vez se ha confirmado el potencial de los procesos de regeneración inducidos por la luz para suprimir de manera permanente los efectos de la degradación inducida por la for
[CAT] Les característiques dels substrats sobre els quals són fabricades les cèl·lules fotovoltaiques cristal·lines constitueixen un element fonamental en el seu comportament i durabilitat. En aquest sentit, la presència d'impureses i imperfeccions en la xarxa cristal·lina redueix el comportament elèctric dels substrats limitant el potencial de generació elèctrica de les cèl·lules. Aquesta limitació es veu augmentada en les cèl·lules de nova generació, en què el seu potencial per augmentar l'eficiència de conversió fotovoltaica només pot fer-se efectiu mitjançant la utilització de substrats amb altes prestacions elèctriques. En la present investigació s'ha procedit a la cristal·lització mitjançant la tècnica Czochralski (Cz) d'un lingot monocristal·lí, tipus p amb resistivitat en l'interval de 0.7-2.1 Ohm.cm i mida industrial. El seu posterior tall en oblees de 200 micres de gruix nominal ha permès disposar d'oblees distribuïdes al llarg del lingot en una fracció solidificada entre el 2.0% i el 99.4%. La caracterització exhaustiva del lingot ha inclòs la mesura de la resistivitat així com la reconstrucció longitudinal i radial de la concentració en carboni, oxigen i ferro. El reduït contingut en carboni obtingut ([C]<5E15 cm-3) contrasta amb una elevada concentració d'oxigen intersticial (Oi) sense efectes apreciables en les propietats elèctriques del lingot. Contràriament, la reducció del temps de vida efectiu dels portadors de càrrega a la perifèria i cua del lingot, és atribuïda a l'augment de la concentració en ferro intersticial [Fei] en aquestes regions. L'estudi de la influència de les difusions en l'evolució de les propietats elèctriques del material va incloure difusions de bor i fòsfor mitjançant BCl3 i POCl3. L'estudi ha constatat l'efecte benèfic de la difusió de fòsfor en la perifèria del lingot associat a la reducció del contingut en Fei. Per contra, s'ha constatat l'efecte perjudicial de la difusió de bor explicat en bona mesura per un augment superior a un ordre de magnitud en [Fei]. L'aplicació successiva de difusions de bor i fòsfor ha revelat una degradació apreciable en el cap del lingot després de l'aparició de patrons circulars en la mesura del temps de vida efectiu característics de la precipitació d'Oi. Igualment s'ha pogut constatar la influència d'una elevada densitat de dislocacions, observada a la cua del lingot, en la resposta del substrat davant les difusions. La fabricació de cèl·lules de tecnologia A-BSF, amb eficiències compreses entre 17.8% i 18.2%, va permetre evidenciar una reduïda influència de les variacions de les propietats elèctriques del substrat en el comportament de les cèl·lules finals. Al mateix temps, la cartografia de l'eficiència quàntica interna (IQE) i de la longitud de difusió (Ld) ha confirmat l'efecte benèfic de la difusió de bor en la perifèria i cua del lingot. La fabricació de cèl·lules Passivated Emiteer Rear Totally-difused (PERT), ha permès l'obtenció de dispositius amb eficiències entre 18.8% i 19.6%. Les eficiències mostren valors màxims al centre del lingot i una reducció apreciable en el cap i cua. En una última etapa, s'ha procedit a la fabricació d'arquitectures de cèl·lules fotovoltaiques Al-BSF, PERT i Passivated Emitter Rear Contact(PERC), amb l'objectiu d'estudiar els processos de degradació i regeneració associats a la formació de parells bor-oxigen (B-O). L'estudi de l'evolució de Voc amb el temps s'ha constatat una dependència de l'amplitud de degradació i cinètica de regeneració amb l'arquitectura de cèl·lula. En aquest sentit es va observar una degradació significativament superior en les cèl·lules PERC principalment atribuïble a una major sensibilitat a degradacions de la qualitat del substrat. Al mateix temps es va confirmar el potencial dels processos de regeneració induïts per la llum, per suprimir de manera permanent els efectes de la degradaci
Cascant López, M. (2016). INFLUENCIA DE LA COMPOSICIÓN Y ESTRUCTURA DE LOS SUBSTRATOS DE SILICIO MONOCRISTALINO EN EL COMPORTAMIENTO ELÉCTRICO DE LAS CÉLULAS FOTOVOLTAICAS CRISTALINAS DE ALTA EFICIENCIA [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/62356
TESIS

碩士
國立臺灣大學
化學工程學研究所
94
The solar cell industry need low cost monocrystalline wafer to enlarge it’s application due to it’s remarkable high material cost. Reducing manufacturing cost is the most important task in the photovoltaics (PV) industry without down grading the quality of the solar ingot. In this paper, the hot-zone of the Kayex CG 6000 puller is well systematic designed through hot-zone simulation and experiment verification to raise throughput (in term of production per hours (PPH) and reduce production cost. The new generation hot-zone design including thermal shield cone, inside insulation and bottom insulation were developed through the software simulation and experiment verification. 　　　　Without sacrificing the quality of solar ingot, the power consumption was significantly reduced from the original 84KW to 68KW. Also PPH is greatly promoted from original 1.34 kg/hr to 2.21 kg/hr.

碩士
國立虎尾科技大學
光電與材料科技研究所
98
This study investigates the influence of temperature, seed pulling rate, diameter control, and rising speeds and rotating rate of crucible on the solid - liquid interface quality during the growth of single crystal by Czochralski method.。 In this study, regarding the heat flow field in the furnace body and the diameter control of ingot, increasing the rotation speed of pillar crystal will lead to more intensive isotherm lines below the pillar crystal region. In this region the temperature changes very violently, where the cover bowl symptoms occur when ingot rod is formed. Meanwhile the melt re-flowing field in the crucible will gradually decrease, which will affect the dopant distribution becoming not uniform. When the rotation speed of crucible increases, the internal center temperature distribution of the melt inside the crucible will become non- uniform and form a cold region then resulting in crystal rods stick out during crystal pulling process. This will damage the ingot rod formation. In the flow field the generation of forced convection results in a non-uniformity for dopant distribution which will also affect the quality of pillar crystals. When the crucible has a relative rotation with ingot rod, if the ingot rod and crucible rotation speed decrease, the vortex at the bottom of crucible symmetry axis will become smaller, and the dopant distribution non-uniform situation will reduce. The results verify the equations of silicon ingot solidified rate (dm/dt) and temperature gradient (dt/dx). And growing 8” silicon ingot succeeded in 16 hours by AST silicon ingot puller. Optimized growth parameters can also control the silicon ingot diameter and quality accurately.

碩士
國立中央大學
機械工程學系
102
Nowadays the Czochralski (Cz) technique has become a main method to grow large single silicon crystal. In order to enhance the quality of crystal, the level of oxygen concentration in the ingot as well as its electric properties has to be controlled. The optimal resistivity for the silicon epitaxial wafers can be obtained by adding directly some common dopants (boron, arsenic, phosphorus, antimony …) into the liquid silicon during the growth process. In this study, the effect of doping arsenic on the oxygen concentration along the freezing interface is numerically investigated by finite volume method (FVM). In order to compare with the experimental resistivity provided by SAS Company, the conversion of crystal resistivity from arsenic concentration is made by using the standard transformation formulation. It is clear that the simulation predictions have similar tendency with the experimental ones in crystal resistivity. The computational results show the mechanism of oxygen content reduction in heavily arsenic-doped Cz silicon melt as compared with non-doped melt. This is because arsenic doping decreases the thermodynamic activity coefficient of oxygen dissolved into the bulk melt from silica. Arsenic content also increases along the length of crystal due to its small segregation coefficient (k0=0.3). The arsenic atoms concentrated in the ingot center are much more than their concentration in the region of crystal edge. Furthermore, the increase in doping level causes a decrease of oxygen content in the growth direction while this increases the radial segregation of arsenic. There is an inverse relationship between dopant concentration and crystal resistivity. Last but not least, the effect of pulling rate and rotation rate on the resistivity is also predicted numerically. The results indicate that the radial resistivity variation of ingot increases with increasing the growth rate as well as crucible rotation rate while this trend is reversed as the crystal rotation rate is accelerated.

碩士
國立中興大學
材料工程學研究所
87
The main purpose of this research is to study the influence of the silicon interstitials generated during wet oxidation on the growth of microdefects in a two-step anneals. The low-high two-step anneal scheme using 650℃ as nucleation temperature and 1150℃ as growth temperature is the base-line process. The results observed in this research is quiet different from what have been observed using 1000℃ as growth temperature. With a short nucleation anneal, the cluster precipitates is the dominant type of defect when growth temperature is 1000℃. However, when growth temperature is 1150℃, cluster precipitates is never found in these samples. At 1150℃ anneal, the stacking faults length is very long , (10μm∼100μm)，but density is low when nucleation anneal is short. When nucleation annealing time increase to 128h, the stacking faults become short, but uniform in length (4μm∼5μm), and density become high. A dissolution test is conducted to study the evolution of stacking faults at high temperature anneal. The most striking result is that the short wet oxidation enhances the oxygen precipitation rate at 1150℃. This result is against conventional understanding about the role of silicon interstitials generated during oxidation.

碩士
國立中興大學
材料工程學研究所
84
In this study, we report a new set of experimentalresults on oxygen precipitation carried out usingCzochralski silicon wafers : a low-high two stepheat-treatment in N2 ambient was employed : thefirst step heat-treatment low temperature ( at 450℃ for 0 - 1024 hours ) is for SiO2 precipitatenucleation and the high temperature step ( at 1000℃ for 0 - 40 hours ) is for precipitate growth. Theoxygen precipitation rate is monitored by measuring the interstitial oxygen concentration in the siliconwafer. Two precipitation retardation phenomena wereobserved on wafers that received prolonged nucleationheart- treatment. First retardation peak whick is morepronounced occurred on the wafers which received l6h - 64 h at 450 ℃ with the microdefect features observed in this study , we believed that thisretardation phenomenon reported by Tan and Kung on 750 ℃ - 1050 ℃ two-step heat-treatment testSecond retardation peak occurred on the wafersheat-treated at 450 ℃for 512 h or longer. Thisretardation peak is relatively weak. And we still notquiet understand the mechanism of this retardationphenomenon. A set of idential wafers were preheated at 1200 ℃ / 1 h in dry N2, prior to the two-step heat-treatment described above. The short hightemperature pre-heat-treated wafers enhanced theoxygen precipitation rate and also revealed thesimiliar retardation phenomena described above, wereport the microdefect features and FT/IR spectra onthe final stage of two-step heat-treatments. Aprofound model is still needed to cover all thephenomena observed.

博士
國立臺灣大學
化學工程學研究所
92
Several hot-zone designs are presented for Czocharlski silicon growth for photovoltaic applications. Without sacrificing the crystal quality, a significant reduction of power and argon consumption was achieved, while the pulling rate was significantly increased. More importantly, the oxygen content was greatly reduced leading to longer minority lifetime of the wafers. According to the results of experiments and simulations, the variation of the axial oxygen distribution could be improved by gradually increasing the crucible rotation speed during the growth.. The major works of the thesis were hot-zone design and computer simulation. The experiments to prove the results of the designs were carried out by SAS. The design reported here included the radiation shield (molybdenum, graphite with different coatings, and composite cone) additional side and bottom insulations (graphite and graphite-felt), and upper side insulation. This thesis made simulations on the power consumption, interface concavity, the crucible and the heater temperature. Then, the growth experiments with the new hot-zone design were carried by SAS. Good agreement was found in the power consumption and a reference temperature near the heater between computer modeling and experimental measurements. The best hot zone design so far has let to a power consumption reduced from 59.1 kWh/kg to 17.4 kWh/kg, the pulling rate was increased from 0.8 mm/min to 1.32 mm/min, the average oxygen content was decreased from 17 ppma to 6.3 ppma, the consumption of argon was also reduced from 93.3 c.f../kg to 27.1 c.f./kg, the degradation rate of the graphite elements was greatly reduced, while, the interface concavity was still remained the same. According to the comparison of productivity, cost, and quality, our design has overtaken that by Siemens Solar Inustries(SSI).

碩士
國立中興大學
精密工程研究所
90
The P- epitaxy on heavy boron doped wafer has been well employed in power MOS device for reducing vertical on-resistance and also provide effective latch-up hardening for advanced CMOS technology. However, the micro-defect generated from the substrate greatly affects the quality of epitaxial layer. To control the density of bulk micro-defect as well as to create a comfortable denuded zone is an important research subject of applying heavy boron doped substrate for IC manufacture. In this resear- ch, a systematical two-step (low — high) and three-step (high-low- high) anneals was carried out to investigate the bulk-micro -defect (BMD) for B+ Cz silicon. Some important information described as follows were obtained. The increase of boron concentration enhanced the oxygen precipitation. The high density of stacking faults was observed with much short nucleation annealing. The feature of etched stacking faults in heavy boron doped silicon is much plump (more less like elliptic shape) than that in light boron doped silicon. The feature of the bulk defects in the heavy doped silicon is different from that of the light doped silicon, in upon with the annealing time and annealing temperature. The denuded zone generated after two-step anneal is not easy to be observed, A long high temperature annealing prior to the two-step anneal is needed. 1150 ℃ annealing up to 4 hours can generate a denuded zone size of about 20 um. The wet ambient can provide a much wide denuded zone than dry N2 anneal. The poly-silicon film deposition process reduce the denuded zone size drastically for the medium to high oxygen concentration silicon used in this research. Therefore, it is not appropriated to use the poly film as back side gettering for high oxygen concentration wafers.

碩士
國立中央大學
能源工程研究所
106
Abstract The single crystalline-silicon (sc-si) is a major material for semiconductor. Recently, the demand for low resistivity wafer is increasing. In order to produce low resistivity wafer, the dopant concentration should be enhanced during the growth of Czochralski silicon. As the dopant concentration increased, the probability of losing structure increased as well. The phenomenon of losing structure is also known as constitutional supercooling. Increasing the temperature gradient in crystal is one method to prevent constitutional supercooling. The Cz furnace model is numerically investigated by CGSim (Crystal Growth Simulator). In this study, the numerical simulation has been performed to analyze how to enhanced the temperature gradient in crystal. First, compare the effect of adding cooled jacket in the furnace of 75kg 6 inch silicon. As the length of cooled jacket increased, the temperature gradient enhanced. Increasing the emissivity of cooled jacket could enhance the temperature gradient as well. Based on this hot zone design, comparing the effect of different crystal and crucible rotation rates. In the case of 75kg 6 inch silicon, the result showed that using the original hot zone with optimized growth parameter can increase the temperature gradient by 13.6% (from 4.6 to 5.2 K/mm). While using the cooled jacket design with optimized growth parameter could increase the temperature gradient 21.9% (from 4.6 to 5.6 K/mm). In addition, this study discusses the larger size production. If we using the optimized hot zone design with optimized growth parameter, the temperature gradient could be increased 39% (from 3.5 to 4.9 K/mm).

碩士
國立中央大學
機械工程學系
102
A three-dimensional numerical simulation has been performed to understand the motion of the melt flow, thermal field and oxygen distributions during the Czochralski silicon single crystal growth process under the influence of a transverse magnetic field. With the application of a transverse magnetic, the velocity, temperature and oxygen concentration fields in the melt become three-dimensional and asymmetric. There were two different flow patterns on the plane parallel and crossing transverse magnetic field, separately. Therefore, the presence of a transverse magnetic field decreases the oxygen concentration level along the melt-crystal interface. The uniformity of oxygen concentration at the melt-crystal interface is also improved when the magnetic field is applied. However, the two flow motion will cause the different temperature distributions form distorted in the whole melt. It is hard to simulation and crystal growth. In this study, the numerical simulation has been performed to clear the mechanism of oxygen transportation, such as the distribution of oxygen concentration in the melt is related to the crystal rotation rate and crucible rate. The lower temperature at the crucible wall and the free surface velocity decrease as the crucible rotation rate decrease. When the crucible rotation rate reaches below 1 rpm, the oxygen concentration value along the melt-crystal interface decrease enlarges. The uniformity of oxygen concentration is better for higher crystal diameters. The crystal rotation rate has negligible influence on the oxygen concentration. But the radial distribution of oxygen uniformity is improved at higher crystal rotation rates. In the case of transverse field, the crucible rotation rate is a key parameter in the control of oxygen concentration in the crystal. The quantity of the oxygen transportation and silica concentration on the free surface can be increased by increasing the gas flow rate. Because the argon gas velocity affect the radial velocity and interfere the free surface flow motion. However, the crystal oxygen concentration was increased with an increase in the flow velocity of argon gas in the TMCZ. This thesis analysis silicon crystal growth process under magnetic Czochralski method, this trend is in consistence with the experimental one. The variation of the axial oxygen concentration with the growth length of the silicon crystal is related to the melt depth of the crucible, the flow structure inside the melt, the crucible temperature, and the argon flow speed along the free surface. In order to improve the axial non-uniform of oxygen concentration, the heater position and crucible rates are adjusted. The axial non-uniform of oxygen concentration can be improved approximately 24.7% and 6.6% by revising the crucible rates and modifying the heater position.

碩士
國立中興大學
電機工程學系
93
Abstract Defect in silicon wafer is detrimental to integrated circuits, to study the defect formation mechanism in silicon is very important to improve the device yield. In this study, we randomly chose hundreds of silicon wafer and employed high temperature oxidation process to reveal the defects on wafer surface. We have observed few defects patterns which relate to crystal growth; 1) OISF ring pattern. 2) Swirl pattern 3) Center core defect pattern 4) Dark ring pattern and 5) the mixed of the patterns. We try to correlate these patterns to crystal growth condition.

碩士
國立中央大學
機械工程學系
107
Czochralski (Cz) method is widely used for the production of high quality silicon single crystal. Under high temperature condition of growth process, the undesirable impurities, such as oxygen and carbon, enter the silicon melt and their content strongly affects the resistivity and the dislocation density of the silicon wafer. A precise control of these impurities at a low concentration and uniform distribution, therefore, has played an important role for improving the quality of silicon crystals, especially large-sized crystals. To our best knowledge, there are few publications showing the effects of the operation parameters of CZ growth process on carbon concentration. In this study, a 2D axisymmetric numerical model is used to study the heat and carbon transport during the growth of a 6 inch-diameter silicon ingot. Different rotation speed and direction of the seed and crucible are considered to investigate their effects on the variation of heat, flow, and carbon characteristics. The numerical simulations show that the carbon concentration gets lowest when the counter rotation rates of seed and crucible are 10 rpm and -3rpm, respectively. The behavior of carbon movement is related to the melt convection and the velocity under crystal-melt interface. While the temperature on free melt surface and oxygen in the melt will affect the quantity of carbon. The flow structure is included three main vortices: Taylor-Proudman cell (1), under the crystal-melt interface, buoyancy driven cell near the crucible wall (3), and the secondary cell (2) between (1) and (3). It was found that the carbon atoms are carried by cell (3) into the silicon melt. The carbon atoms are got out of the melt from the free melt surface. The larger effective evaporation area may reduce the carbon content in the melt due to the larger evaporation rate of carbon. Moreover, the secondary vortex (2) also affects the carbon transportation. Appropriate cell (2) may keep the carbon atoms stay longer in the melt and buoyancy cell (3) is easier to bring them to the free melt surface.

博士
國立中央大學
能源工程研究所
106
The Czochralski process nowadays becomes a main pulling method for the production of the commercial silicon wafers due to its relatively high growth rate and possible weight control. This technique, however, is facing to the big challenges how one can improve the growth system and operation conditions to produce silicon crystals with a good quality. From this study, it is expected to find the better growth process for the effective control of the heat, flow, oxygen transport, and defect formation. 2D global numerical simulations of heat and mass transfer under the influence of magnetic field in an industrial Cz-Si growth were conducted. The flow characteristics, the distributions of temperature, oxygen concentration, and the formation of point defect under different crystal-crucible rotation conditions without and with a cusp magnetic field were analyzed. The simulation results showed that the oxygen content along the crystal-melt interface is determined by the competition between the diffusion and convection mechanisms. At a low difference between the crystal and crucible iso-rotation rates, the effect of the diffusion process on oxygen transport in the melt becomes stronger than the effect of convection. A lower concentration and uniform radial distribution of oxygen can be obtained under the iso-rotation condition. It was found that a flow transition occurred in the silicon melt when the crystal and crucible have the same iso-rotation rate. Since that, the ratio between crystal and crucible rotational Reynolds numbers, Res/Rec, is 0.5842. Under the application of a cusp magnetic field, the electric force and the magnetic force changes the magnitude and direction of velocity in the silicon melt. The magnetic field has the stronger effect on the oxygen concentration in counter-rotation cases in comparison with iso-rotation ones. Reducing or enhancing the oxygen content by a cusp magnetic field depends on the differences between the crystal and crucible rotation rates. Using an unbalanced cusp reduces the radial uniformity of oxygen content, especially in counter-rotation cases, as compared to using a balanced one. Rotating the crystal and the crucible in the same direction also produces a flatter defect transition and a lower concentration of point defects. Producing a convex crystal-melt interface in iso-rotation cases is good for growing the defect-free crystals due to the outward diffusion of point defects from the central region to the edge of ingot. The axial temperature gradient is enhanced in iso-rotation cases. This may allow the faster pulling of crystal from the silicon melt and prevent the super-cooling during the growth.