N-type Czochralski silicon solidification: Oxygen- and copper-related defects formation
Abstract
This PhD work was focused on the formation of oxygen- and copper-related defects in ntype
Czochralski silicon for solar cells. These defects are a very important source of
material and solar cell performance degradation, and a better understanding of the
mechanism behind their formation is still needed.
The first scope of this thesis was to understand the impact of process parameters on the
quality of n-type silicon ingots. We have seen that the oxygen incorporation into the
crystal is very sensitive to the evaporation rate at the free melt surface and a proper
control of the transition from neck to body (crown plus shoulder) is of extreme
importance to the material, as large diameter changes are needed. In industry, the crown
pull speed is set to a specific value; the crown shape may anyway change from ingot to
ingot. This shape can be indirectly correlated to the average melt temperature after
necking, where lower average temperature leads to a higher growth rate of the crown.
Longer time to grow the crown results in higher oxygen evaporation at the free melt
surface and consequently a lower oxygen concentration is measured at the top of the
ingot body. Thermal donor annealing heat treatment results in an improvement of the
average lifetime and reduction of its inhomogeneities. However, only small
improvements are found on the early body region with higher oxygen concentration due
to the formation of larger oxygen-related defects.
The following work on the optimization of the process was based on changes of the
crown pulling speed (from 0.6 to 0.8 mm/min). It was seen that higher pull speed results
in a more elongated crown and consequently lower oxygen incorporation into the crystal
body. The formation of oxygen precipitates is very high at regions below the P-band and
the lifetime degradation is correlated to the high precipitate density. The precipitation at
the first part of the body is reduced for ingots grown with higher crown pull speeds,
although a strict control of the process parameters (e.g. heater power) is still needed.
Lower solid-liquid interface deflections are found for the top of the ingot, grown with
higher crown pull speed. As the interface is normally concave towards the melt during
further body growth, small interface shape changes take place in this ingot, with lower
probability of structures loss.
After optimization of the growth of the Cz silicon ingots, the focus has been set on the
impact of impurities on the quality of n-type Cz silicon ingots. In order to study the
impact of metallic impurities, phosphorus and oxygen on the defects generation, four
ingots were grown with different impurity contents. The main difference between top and
middle of the ingots is the oxygen concentration due to its higher concentration compared
to other impurities. Metallic impurities are still a limiting factor for lifetime at the ingot
middle, and can have a different impact depending on feedstock quality. The reduction of
the oxygen incorporation is effective on the overall improvement of the electrical
properties, mainly after high temperature processes.
In this work, ring-pattern defects were only visualized after thermal oxidation, but still
without significant fluctuations of the interstitial oxygen distribution. Striations are
correlated to the oxygen precipitate density, except for ingots grown with lower oxygen
incorporation. However, the precipitates density is not enough to correlate with the lifetime fluctuations, as their nature may differ from one striation to another. After heat
treatment, a large reduction of the average lifetime is obtained at the top of the ingots due
to the oxygen precipitation and formation of stacking faults. The degradation of the solar
cells performance, made from these materials, is mainly attributed to the precipitation of
oxygen during their fabrication. The formation of ring-pattern defects in as-grown
samples is correlated to the interaction between vacancies and oxygen during crystal
cooling. The local differences can be minor through the as-grown samples radial
direction, but may be enhanced during high temperature processes, as the ones during
solar cells fabrication.
At last, a study of lifetime degradation close to the ingot end was carried out. We have
initially measured a large lifetime degradation of the material quality on a block, obtained
from the end of an n-type ingot at room temperature. The source of its degradation was
unknown until we measured a high average Cu concentration in this block with GDMS.
Cu depth profiles were measured and a very high concentration was found close to the
sample surface, with a decrease towards the bulk. Lifetime degradation in regions of
moderate Cu contamination can thus be associated to the continuous Cu precipitation
close to the samples surface. It is a result of the high diffusivity and low solubility of Cu
in silicon, even at room temperature. The ingot end may thus be a source of
contamination for the rest of the ingot and need to be quickly removed. The Cu
concentration and precipitation close to the surface depends on the surface area, where a
larger area enhances the precipitation close to the surface and reduction of the bulk
contamination.