NASA NTRS2001
Time dependent convective flows during crystal growth of doped semiconductors lead to fluctuations of the composition, so called dopant striations. In general, it is difficult to decide which is the main mechanism for the generation of these striations, it might be either the fluctuation of the concentration field in the melt and the extent of the solute boundary layer ahead of the solid-liquid interface or a variation of the growth velocity. Direct access to the concentration field is rather complicated to achieve, especially considering the high process temperature and the chemical activity of liquid silicon. The contribution of growth rate fluctuations to the formation of compositional fluctuations can be determined by measuring microscopic growth rates. The classical method of current pulses requires electrical feed-throughs and good electrical contacts, both are critical issues for the growth of high purity silicon crystals. Using a radiation based heating system, the heating power can be modulated very fast and effectively. We added to the normal heater power a sinusoidal off-set in the frequency range of 1 to 10 Hz, generating a narrow spaced weak rippling in the grown crystals which are superposed to the dopant striations caused by natural and by thermocapillary convection. The pulling speed was varied between 1 and 4mm/min. The microscope images of etched crystals slices have been analyzed by peak-search algorithms (measuring the spacing between each artificially induced marker) and by FFT. Performing growth experiments under a time-dependent flow regime, fluctuations of the microscopic growth velocity of Delta(v)/v(sub average) up to 50% have been measured. Damping the time-dependent convection by the use of an axial, static magnetic field of 500mT, the microscopic growth rate became constant within the resolution limit of this method. The results will be discussed using analytical methods for the calculation of microscopic growth velocity and by comparing them with measurements of temperature fluctuations in the melt during growth experiments itself.