6.6.3 Simulation of Bulk Silicon Crystallisation
As an example of the temperature simulation of silicon ingot crystallisation, the SOPLIN
casting technique is selected. To simulate this process, a finite element mesh of about
230 000 elements was built to describe the furnace geometry. This mesh includes the
silicon ingot, the mould, all insulation materials and active heating and cooling facilities,
248
BULK CRYSTAL GROWTH AND WAFERING FOR PV
Pivotable
top of the
furnace
>
1600
>
1460
>
1320
>
1180
>
1040
>
900
>
760
>
620
>
480
>
340
>
200
[°
C
]
Passive
insulation
Mould
support
Silicon
Liquid
Solid
Mould
Figure 6.29
Finite element geometry of an ingot casting furnace and simulated temperature distribution
during a reference process. The liquid – solid interface is marked by the black line
as shown in Figure 6.29. Because of confidentiality agreements with the industry, the
heating and cooling systems are not shown in detail in this figure. All heat conductance
and capacity effects as well as the non-stationary release of latent heat are taken into
account. All material contact regions between silicon, mould or insulation materials are
modelled by heat flow–resistance parameters. To describe the heat flux by radiation inside
the furnace, a view-factor model is included in the software. All material data are treated
in their temperature dependency and all the internal control systems of the furnace are
added to the simulation software.
To start one simulation run, only the cooling water temperature and the time-
dependent process control information are necessary as input data, as they are entered in
the crystallisation furnace. Output from one calculation is the three-dimensional tem-
perature history in the furnace, beginning after pouring the melt and ending with a
homogeneous temperature of about 300
◦
C inside the ingot. This calculation needs less
than 6 h on a common one-processor workstation.
In Figure 6.29, an example of the temperature distribution during a reference pro-
cess is shown in the middle, cut through the furnace. The liquid–solid interface is marked
by the melting temperature isotherm. The solidification front is mostly flat, and a slight
non-symmetry is caused by the specific construction of the heating system. These simu-
lation results are verified in an experimental crystallisation furnace with good agreement
to the measurement in the ingot volume during crystallisation.
In general, the shape of the solidification front is controlled by the lateral heat flux,
while the vertical heating and cooling conditions control the solidification velocity. To
investigate these general reflections for the described furnace, variations of the process
control were simulated. In Figure 6.30, two variations are presented. By a 30% raise of
heating power at the side walls of the ingot, the shape of the solidification front becomes
more convex. Otherwise, a reduction of heating power by about 20% turns the solidifica-
tion front to a more concave shape. Additionally, to this more or less predictable effect,
simulation results show an increase in the solidification time for the convex crystallisation
of 44% and a 30% reduced processing time for the concave solidification. Both effects
are due to the total varying power input in the furnace. These simulation results enable
NUMERICAL SIMULATIONS OF CRYSTAL GROWTH TECHNIQUES
249
>
1600
>
1460
>
1320
>
1180
>
1040
>
900
>
760
>
620
>
480
>
340
>
200
[°
C
]
Liquid
Solid
Liquid
Solid
Convex
Concave
Do'stlaringiz bilan baham: |