Czochralski crystal growth system

Figure 6. Schematic of driving forces for flows in Czochralski crystal growth system, which shows the regions where the driving forces will produce the strongest motions. The shape functions describing the unknown interface shapes are listed also.

Meniscus-Defined Crystal Growth Systems. In most conventional meniscus-defined growth systems, a seed crystal is dipped into a pool of melt, and the thermal environment is varied so that a crystal grows from the seed as it is pulled slowly out of the pool. Two examples of meniscus-defined growth are shown in Figure 1. The Czochralski (CZ) method (Figure lb) and the closely related liquid-encapsulated Czochralski (LEC) method are batchwise processes in which the crystal is pulled from a crucible with... 

Czochralski Crystal Growth Case Study of Meniscus-Defined Growth System... 

Oliver, D. W., Brower, G. D., and Horn, F, H. 1972. Cold metal crucible system for synthesis, zone refining, and Czochralski crystal growth of refractory metals and semiconductors. J. Cryst. Growth 12 125-131. 

The next most importtmt parameters in Czochralski growth of crystals are the heat flow and heat losses in the system. Actually, aU of the parameters (with the possible exception of 2 and 9) are strongly ciffected by the heat flow within the crystal-pulling system. A tj pical heat-flow pattern in a Czochralski sjretem involves both the crucible and the melt. The pattern of heat-flow is important but we will not expemd upon this topic here. Let it suffice to point out that heat-flow is set up in the melt by the direction of rotation of the cr5rstal being pulled. It is also ctffected by the upper surface of the melt and how well it is thermally insulated from its surroundings. The circular heat flow pattern causes the surface to radiate heat. The crystal also absorbs heat and re-radiates it... 

Figure 1. Schematic diagrams of several commonly used systems for melt crystal growth of electronic materials (a) vertical Bridgman, (b) Czochralski, and (c) small-scale floating-zone systems.

Figure 3. Streamlines (on right) and isotherms (on left) for growth of Si in a prototype Czochralski system. The volume of the melt, at the bottom in each drawing, changes among the calculations, affecting the qualitative form of the convection cell and the shape of the crystal interface. From Theory of Transport Processes in Single Crystal Growth from the Melt, by R. A. Brown, AIChE Journal, Vol. 34, No. 6, pp. 881 -911, 1988 [29]. Reproduced by permission of the American Institute of Chemical Engineers copyright 1988 AIChE.

Figure 24. Streamlines and isotherms for the growth of silicon in a prototype Czochralski system with self-consistent calculation of interface and crystal shapes by using the quasi steady-state thermal-capillary model and the condition that the crystal radius remains constant. Calculations are for decreasing melt volume. The Grashof number (scaled with the maximum temperature difference in the melt) varies between 1.0 X 107 and 2.0 X 107 with decreasing...

The next most important parameters in Czochralski growth of crystals are the heat flow and heat losses in the system. Actually, all of the parameters (with the possible exception of 2 and 9) are strongly affected by the heat flow within the crystal-pulling stem. The next section addresses this factor. 

In the Czochralski method, the material is heated in a crucible just above its melting point. A seed crystal is immersed in the melt and it is slowly pulled out (5mm/min) from the melt vertically while being slowly twisted (2-20 rpm), resulting in continuous growth at the interface. The system is within a chamber with argon to avoid contamination. 

A numerical model for simulation of the global heat transfer and the melt flow in the Czochralski growth of large silicon crystals is presented. The key model features are an extended 3D domain for the 2D/3D computations and a hybrid LES/RANS approach to turbulence modeling. It is shown that use of parallel computations on affordable multiprocessor systems assembled from the COTS hardware could reduce the turn-around time of simulation by an order of magnitude. The model validation using the experimental data on the growth of 100-mm and 300-mm silicon crystals in the industrial pullers Ekz-1300 and Ekz-2405 has proved its predictive power. 

---