GAP CZOCHRALSKI PROCESS

Abstract

This study, at Agilent Technologies, Singapore, looks into the feasibility of incorporating In-grower Synthesis (IGS) into the existing GaP Liquid-Encapsulated Czochralski (LEC) Process. IGS is an in-situ method of producing GaP, from Ga and P, prior to pulling a single crystal. It is found that IGS is possible but has disadvantages of being hazardous and difficult to control. It is demonstrated that upgrading IGS from a 1.9”- to a 2.9”- process is feasible, though improvements must be made to the equipment. It is found that using the Dash Technique with an early seed lift reduces the tendency for twinning. Visibility can be improved by using a tinted, polarising filter. In addition, reducing the vaporisation of P through careful heating, cleaning of P deposits, a minimum pressure of 32 atm and packing boric oxide on top of GaP helps improve visibility. Convection currents in Ar worsen visibility and cause thermal instabilities that can lead to creation of defects in the crystal. This is exacerbated by the forced (due to stirring) and natural convection currents in the melt. It is recommended that the Ar pressure be maintained at apprroximately 70 atm during LEC. The presence of instabilities also limits the single crystal yield to approximately 70%. Through experiments, it is found that the maximum pulling rate may be approximately 0.6 in./h. However, in view of the adverse effects on dislocation density and electrical characteristics, the maximum pull rate for a 2.9”-boule should be approximately 0.4 in./h. It is also verified that twinning occurs on the (111) planes and that a faster pull speed will result in increased dopant content, as predicted in the Buton-Prim-Slichter equation. Finally, it is found that for the current charge weight of 1460 g, pulling at 0.385 in./h for 7.2 hours gives an immediate improvement in productivity. It leads to a 17% reduction in cycle time, from 16.1 to 13.3 h. Furthermore, this will increase the throughput by 20% to 110 g/h. Increasing, the charge weight to give a longer crystal is limited to 1700 g by the crucible dimensions. Moreover, increasing the charge weight means that some time has to be spent in optimising the crucible height for future LEC.