Most of the large-scale integrated circuits (LSIs), indispensable at present day, are made from Si wafers grown by Czochralski (CZ) method. The size of LSIs continues to grow with the integration of more devices. The diameter of Si crystal has also been increasing to enable the production of large-size LSIs with practical productivity and cost. Generally, a large-diameter crucible is used to produce large single-crystal silicon with practical productivity. However, it is becoming difficult to control the convection of large amounts of silicon melt in large crucibles using the conventional CZ method. Magnetic field-applied CZ (MCZ) is one solution to control the convection of Si melt. There are several kinds of MCZ categorized by magnetic-field direction: horizontal (HMCZ), vertical (VMCZ), Cusp MCZ; and several types of magnets categorized by the magnetic-field generation method (electroconductive and superconductive). In this paper, out-lines of CZ production for single-crystal silicon, MCZ equipment and MCZ production are reviewed.
We fabricated a large Bi2223 hollow cylinder as a superconducting magnetic shield through plasma spraying in order to obtain a very low and calm magnetic-field environment where we can measure the weak pulsed magnetic fields caused by human brain activity. As a substrate, we used a pure-nickel hollow cylinder (323.2mm in O. D., 320mm in I. D. and 660mm in length), the outside of which was coated with a Ag buffer layer, as a diffusion barrier, and then a Bi-system oxide layer about 730μm thick, applied through plasma spraying. However the as-sprayed Bi-system oxide layer showed unclear broad X-ray diffraction patterns of amorphous-like crystal structure and no superconducting properties at 4.2K. Therefore, the sprayed Bi-system oxide hollow cylinder was heat treated at 838°C×100h to form a superconducting phase (Bi2223 phase) through solid-state diffusion reaction. The superconducting Bi2223 hollow cylinder showed a shielding factor of 103 at Z/D=0.98 at 79K in a dc field of 4×10-4T, where Z and D are the distance from the edge of the cylinder to the measuring position at the center line of the cylinder, and the inner diameter of the Bi2223 cylinder, respectively. The same shielding factor, of about 103, was obtained in ac fields of 3×10-5T with frequencies below 10Hz.
For the purpose of the stability analysis of a pool boiling superconductor, many studies on the heat transfer of LHe have been conducted. There are a number of variables that affect heat transfer. It is well-known that to variables are surface orientation and surface treatment. Surface orientation of a superconductor is varied by magnet winding. The change is associated with the variation of gravitational force on the surface, thus causing heat transfer characteristics to change. Usually, the surface of a superconductor is treated to improve heat-transfer characteristics; for example, oxidation. During the winding process, the winding machine may strip off the treatment at some locations. The resulting damage may change the heat-transfer characteristics and degrade the stability of the superconducting magnet. In this study, the heat transfer of polished Cu, oxidized Cu and partially oxidized surfaces were measured as a function of orientation. The critical and minimum heat fluxes depend on the area fraction of oxidation. The calculation method for the heat flux of a partially oxidized surface was established. Heat-transfer measurements of a prototype superconductor with polished surfaces were also conducted to change the surface orientation. The heat-transfer characteristics with the prototype superconductor were degraded as compared with those of the Cu surface. It became clear that heat transfer for a stability analysis must be measured using a prototype superconductor. Next, the recovery current for a large-sized pool boiling superconductor, which is a helical coil superconductor, was calculated according to the experimental results of heat transfer. The dependence of the recovery current on surface orientation and the area fraction of surface treatment was estimated. The calculated results were compared with the measured recovery currents of the short sample. The calculated recovery current agreed well with the experimental result.