Abstract: Reducing the carbon concentration in Czochralski (CZ) silicon crystals is crucial in order to improve the properties of high-power devices, such as on-resistance and carrier lifetime. To achieve carbon concentration reduction, it is necessary to reduce carbon monoxide (CO) contamination from the CZ furnace graphite components and to remove the carbon impurities originating from the starting material. In this study, suppressing the chemical reaction between silicon monoxide (SiO) and the graphite heater effectively reduced the CO contamination rate. Furthermore, we attempted to promote CO evaporation during the CZ process in order to remove carbon impurities from the melt. Increasing the Ar gas flow velocity above the melt surface was found to be effective in increasing the CO evaporation rate during both the melting and growth processes. The CO evaporation rate during the melting process of 8-inch CZ silicon was calculated as being of the order of 10-2 μg/s. Owing to the effects of the CO evaporation, 8-inch CZ silicon crystals with carbon concentrations lower than 2.0 × 1014 atoms/cm3 at a solidified fraction of 0.85 were grown.
Abstract: To determine the electrically inactive fraction of As or P in heavily doped as-grown Czochralski Si 4-point resistivity and SIMS measurements were carried out. No clear trend for the electrical inactive fraction was found with an increasing dopant concentration, though a mean electrical inactive fraction of 11.5% for As doping could be determined.Experimental results on a dopant-vacancy complex in as-grown Si are scarce, hence temperature-dependent positron annihilation lifetime spectroscopy (PALS) was carried out on several heavily As and P doped as-grown Si samples. The measured average positron annihilation lifetime τav is between 218 ps and 220 ps. No temperature dependent effect on τav could be observed. Therefore, it can be concluded that in the studied doping range the dopant-vacancy complexes do not exist. The reason for the inactivation of the dopant has to be found elsewhere. A possible explanation can be the formation of dopant precipitates.
Abstract: In an attempt to understand how and where dislocations are introduced into Si ingots by temperature gradients, bulk dislocation-free FZ crystals are exposed to temperature gradients similar to those in Bridgman Si crystal growth. This heat treatment introduces dislocations, which were analyzed using X-ray topography (XRT) and Scanning InfraRed Polariscopy (SIRP). Hereby, the orientation dependency is taken into account and ingots in (001) and (111) growth orientation are evaluated in this work. It can be found that the dislocation generation takes place at similar regions of the crystal and is independent of orientation, however, their propagation and multiplication differs. This leads to an overall different shape of the dislocation network. Especially intriguing are the long slip lines in the (111)-crystal, which cannot be found in the (001)-crystal. This suggests a different magnitude of slip propagation depending on the sample orientation. This effect should be explained by a different activation of slip systems and is discussed in the paper.
Abstract: In recent years, silicon solar cells continue to remain the main stream in photovoltaic (PV) industry, particularly of made from multi-crystalline silicon (mc-Si). The progress of crystal growth technology for mc-Si ingot using directional solidification (DS) is particularly significant. With the breakthrough of the so-called high-performance (HP) mc-Si technology in 2011, the mc-Si solar cell efficiency had increased from 16.6% in 2011 to 18 % or beyond in 2013. Nowadays, HP mc-Si, solar cells from a normal screen-printing aluminum back surface field (Al-BSF) production line could easily reach 18.3%. With the passivated emitter and rear cell (PERC) structure using PECVDalumina passivation, an average efficiency of over 19.2% could also be obtained. The emerging of HP mc-Si almost blocked the development of mono-like technology in 2012, and pushed p-type mono-Si cells to higher efficiency by using advanced technology. Unlike the conventional way of having large grains and electrically-inactive twin boundaries, the growth of HP mc-Si is from small and uniform grains having more random GBs. The grains developed from such grain structures significantly relaxes the thermal stress and suppresses the massive generation and propagation of dislocation clusters. Currently, most of commercial mc-Si ingots are grown by this concept, which could be implemented by seeded with small silicon particles or using nucleation agent coatings. The seeded growth has been well adopted in industry. However, the melting control of the seed layer and the thick red zone induced remain key issues in mass production. Several methods have been considered to resolve these issues with some success. The use of nucleation agent layers is a simpler approach, but the control of initial grain structures remains challenging.
Abstract: We have proposed single seed cast Si growth and developed a furnace for 50 cm square ingots. By optimizing growth parameters, improving gas condition, coating, the quality of mono Si ingot has improved. Namely, dislocation density, the concentrations of substitutional carbon and interstitial oxygen have been significantly reduced. The conversion efficiency of cast Si solar cells has become comparable with those of CZ Si wafers.
Abstract: The grain evolution of multicrystalline Si was studied using the ingot grown from microcrystalline template. The grain shape evolution and width increase are not monotonic but may have 3 stages. On the other hand, the grain boundary (GB) analysis suggests that there exit 2 reactions, namely random GB annihilation at the initial stage and Σ3 generation and annihilation at the steady state.
Abstract: Rice husk, an agricultural waste product obtained in large quantities in many countries including Nigeria, is very rich in siliceous materials. It has been known for several decades that, with careful processing, rice husk can be a source of metallurgical grade silicon . The question remains as to whether rice husk ash (RHA) can be purified by a cost-effective, low technology route to produce solar-grade silicon suitable for use in photovoltaic devices. In Nigeria this would have the benefit of transforming large volumes (> 600,000 tonnes per annum) of agricultural waste into a partial solution to that country's issue with energy distribution.In this work, high purity silica has been prepared from RHA (ashed at 700°C for 5 and 12 hours) using a hydrometallurgical process. We report on the effect of natural variations in the rice husk composition on the effectiveness of the hydrometallurgical purification; the effectiveness of each stage of the hydrometallurgical process in removing impurity elements. While the hydrometallurgical purification of RHA is effective in removing impurities such as Ti and Fe to levels below the limits of detection of X-ray fluorescence (XRF), B and P levels need to be reduced to < 1017cm-3 well below the detection limits of XRF. The resultant silica has been subsequently reduced to metallurgical-grade silicon (MG-Si) by direct reduction using Mg powder.
Abstract: Surface states have hindered and degraded many semiconductor devices since the Bardeen era. Surface states originate from dangling bonds on the surface. This paper discusses a generic solution to surface states, i.e. valence-mending passivation. For the Si (100) surface, a single atomic layer of valence-mending sulfur, selenium or tellurium can terminate ~99% of the dangling bonds, while group VII fluorine or chlorine can terminate the remaining 1%. Valence-mending passivation of Si (100) has been demonstrated using CVD, MBE and solution passivation. The keys to valence-mending passivation include an atomically-clean Si (100) surface for passivation and precisely one monolayer of valence-mending atoms on the surface. The passivated surface exhibits unprecedented properties. Electronically the Schottky barrier height between various metals and valence-mended Si (100) now follows more closely the Mott-Schottky theory. With metals of extreme workfunctions, new records for low and high Schottky barriers are created on Si (100). The highest barrier so far is 1.14 eV, i.e. a larger-than-bandgap barrier, and the lowest barrier is below 0.08 eV and potentially negative. Chemically silicidation between metal and valence-mended Si (100) is suppressed up to 500 °C, and the thermally-stable record Schottky barriers enable their applications in nanoelectronic, optoelectronic and photovoltaic devices. Another application is transition metal dichalcogenides. Valence-mended Si (100) is an ideal starting surface for growth of dichalcogenides, as it provides only van der Waals bonding to the dichalcogenide.
Abstract: Deep Level Transient Spectroscopy (DLTS) has been applied to Metal-Insulator-Semiconductor (MIS) capacitors, consisting of a p+ or n+ a-Si:H gate on an intrinsic i-a-Si:H passivation layer deposited on crystalline silicon n-or p-type substrates. It is shown that the type of gate has a pronounced impact on the obtained spectra, whereby both the kind of defects (dangling bonds at the a-Si:H/(100) c-Si interface (Pb0 defects) or in the amorphous silicon layer (D defects) and their relative importance (peak amplitude) may be varied. The highest trap densities have been found for the p+ a-Si:H gate capacitors on an n-type Si substrate. In addition, the spectra may exhibit unexpected negative peaks, suggesting minority carrier capture. These features are tentatively associated with interface states at the p+ or n+ a-Si:H/i-a-Si:H interface. Their absence in Al-gate capacitors is in support of this hypothesis.