Adjustment of heater power distribution during VGF
growth of GaAs crystals
Fig. 1.
Schematic view of the VGF setup for growing 6 inch diameter GaAs crystals
|
The Bridgman technique is widely applied to grow
high quality III-V semiconductors such as GaAs, InP, etc. The vertical gradient
freeze method (VGF) is a modification of the conventional Bridgman method without
moving the ampoule or heaters. In this case, the crystallization process is provided
by changing the heat supply in the electrodynamic gradient multi-zone furnace.
Optimization of heat transfer in the furnace, and the melt and gas flows is
the key point in growing quality crystals.
The following physical phenomena affect crystallization in VGF growth:
- heat transfer by radiation and diffusion in a particular hot zone;
- melt convection and convective heat exchange under the
crystallization front, including free surface phenomena and magnetic field effects;
- gas convection and heat exchange at the gas/crystal interface
(encapsulant flow during liquid encapsulated crystal growth);
- latent heat release on the crystallization front.
Thermal management and recipe optimization
Optimization of process recipe is a powerful approach to increase yield and to improve
crystal quality. In the presented example, optimization of the process has been performed
without any hardware modifications. Optimized recipe allowed us to achieve more
convex melt/crystal interface shape in the beginning of cylindrical part of
the crystal and significantly reduce probability of multi-crystalline material growth.
Lower values of temperature gradients reduce dislocation density in the crystal and reduce
probability of crystal cracking in the end of the process caused by residual stresses.
Original operating regime of the heater and computed
temperature distribution in the reactor are shown in Figure 2.
(a) |
(b) |
Fig. 2. Heater power
evolution during the growth process (a) and computed temperature field in the reactor (b).
|
For each of the three heaters, we suggested individual dependence of the power as a function of time.
Optimized heater power evolution and the resulting temperature distribution
in the reactor are shown in Figure 3.
(a) |
(b) |
Fig. 3. Heater power
evolution during the growth process (a) and computed temperature field in the reactor (b).
|
Effect of the heater power optimization is illustrated in Figure 4.
One can see a noticeable effect on the shape of the crystallization front,
as well as significant drop in temperature gradients at the melt-crystal interface,
which allows us to hope for a considerable improvement of the crystal quality.
|
Fig. 4.
Effect of the heater power optimization. Evolution of the crystallization
front before (black) and after the optimization (red). The correlation
between the crystallization front shape and the temperature
gradients along the front.
|
Heater power optimization is not the only way to tackle this problem.
Other ways to decrease the temperature gradients and dislocation
density during VGF growth include:
- heat transfer by radiation and diffusion in a particular hot zone;
- special design of heaters and increasing the number
of independent heaters;
- heat shields around the “seed” part of the crystal;
- temperature engineering of the bottom part of the ampoule;
- design of the side and upper insulations.
Modeling of dopant distribution
GaAs crystals are doped with different elements to achieve the required electric conductivity
for semi-conducting or semi-insulating wafers. Control of segregation of doping
materials is important to increase a high yield of GaAs wafers with electric conductivity
values within the specification limits. Modeling of melt flow coupled with segregation
helps to obtain distribution of dopant, for example Si or Zn in the crystal.
Figure 5 Si doping distribution
|
Figure 6 Zn doping distribution
|
