No abstract

No abstract

The density of molten silicon doped with 0.1 at% boron or 0.1 at% gallium was measured over a temperature range from the melting point to 1650° C by using an improved Archimedean method with modified dipping procedure to study the influence of impurity doping on density anomaly. Density anomaly with the thermal volume expansion coefficient of about 8.0×10-4 K-1 has been observed from 1420° C to 1435° C for the pure molten silicon, together with the drasric decrease in density, regarded as the stage prior to solidification, from the melring temperature to 1420° C. This anomaly was also observed in 0.1 at% boron-doped silicon melt, but completely suppressed in 0.1 at% gallium-doped melt. No change was observed, however, in the thermal volume expansion coefficient, which remained about 0.6×10-4 K-1, regardless of the addition of such impurities over 1435° C. Concentration dependence of density anomalies was also investigated using the molten silicon doped with 0.1-1.0 at% gallium, showing no sign of the anomalous temperature coefficient.

The oxygen incorporation mechanism at the growth interface during Czochralski (CZ) silicon single crystal growth has been studied using melt quenching technique developed by the double-layered Czochralski (DLCZ) process. A slow crystal rotation rate of 0.5 rpm was chosen to investigate the influence of melt convection in relation to the source of oxygen on the oxygen inhomogeneity in the grown crystal. Micro-Fourier transform infrared spectroscopy (micro-FTIR) measurement and preferential etching revealed that the oxygen variation in the crystal was determined by the balance of the bulk melt with a high oxygen content and the oxygen-depleted melt from the free surface. The major cause of the periodic intake of the oxygen-depleted melt flow during one crystal rotation was probably the inhomogeneous radial temperature gradient. This suggests that thermal asymmetry in the melt is a fatal factor in oxygen inhomogeneity in the crystal. We have postulated that the equilibrium oxygen segregation coefficient is not much smaller than unity, but is rather close to it.

No abstract