Structure factor and pair correlation: Energy-dispersive x-ray diffraction showed reduced intensities and structure factor of Si as well as a temperature-dependent change in the nearest interatomic distance and coordination number. Small-angle scattering showed no evidence of cluster formation in a Si melt near to the melting temperature.
Binding of impurity gallium: Modified extended X-ray absorption fine-structure spectroscopy showed that impurity-level gallium atoms are tightly bound by three Si atoms in a Si melt. This bonding may be the key to understanding why a density anomaly arises near to the melting point and why Ga addition destroys it.
Crystal-melt relaxation and anomalous effects: The Si melt density after melting clearly showed a relaxation tendency. The equilibration time was 3 hours, the well-known aging period in crystal growth. After relaxation an anomalous change took place up to 15 degrees higher than the melting point. In the anomalous region, the thermal expansion was calculated to be about an order of magnitude larger than in the higher temperature region. A property anomaly in the same temperature range was also found for the surface tension and viscosity. The addition of 0.1% boron did not show any effect on the density anomaly while that of 0.1% gallium or antimony wiped out the anomaly.
Effects of antimony doping: Highly antimony-doped Si crystals are known to often become oxygen deficient, while the controlled addition of oxygen is essential. The oxygen saturation solubility was found not to decrease, but to increase upon antimony addition, while a volatile species of Sb2O formed in proportion to the square of the Sb concentration, and then evaporated.
Oxygen in a Si melt: The results of conventional simulation techniques quantitatively agreed with that by x-ray fluoroscope observations of Si melts in a crucible. Oxygen dissolved from the contact part of the Si melt with the crucible. Using the determined evaporation rate, the most eroded part of the crucible was indicated to be the boundary between the inside wall and the bottom surface, and experimentally confirmed. Also, the radial oxygen distribution in the crystal agreed well with the calculated result. Thus, the simulation technique is useful and reliable as long as proper boundary conditions are employed.