fixed charge [33]. The negative fixed charges seem to be originating from Al vacancies, which are a result of a preferred orientation of Al bondings close to the interfacial layer [33]. In figure 2.13C, the band diagram including a thin interfacial oxide layer is shown. Due to the repelling of majority carriers at the surface, a depletion region is formed close to the interfacial SiOxlayer.
This results in band bending. Furthermore, the fixed charges are shielded by the accumulation of the minority carriers, which leads to inversion.
The effect of field-effect passivation on the underlying substrate depends strongly on the doping type and level. For n-type surfaces, field-effect passivation repels the majority carriers (elec-trons) from the surface, which leads for large densities of negative fixed charges to inversion, since more holes exist close to the surface than electrons. At p-type surfaces, high-quality aluminium oxide passivation leads to an accumulation of the majority carriers, resulting in a strong increase of the ratio of majority to minority charge carrier concentrations. The amount of fixed charges is crucial for field-effect passivation, since Hoex et al. [33] showed, that Seff scales with Q12
f
starting at a fixed charge of|Qf|= 3×1011cm-2. In Figure 2.13Ban important effect is shown on moderately doped n-type silicon. While p-type silicon exhibits no increase of the effective SRV (Seff) at around 1011 cm2 negative fixed charges in the aluminium oxide layer, negatively doped silicon shows an increase in SRV. This is due to the fact, that electrons on the surface of n-type silicon are pushed away and hence the ratio of minority to majority carrier gets equalized, when the effective SRV is peaking. For higher negative fixed charges of the aluminium oxide, the electrons, therefore the majority carrier concentration gets even smaller at the surface than the minority charge carrier concentration, leading to the in figure 2.13 Cvisible effect of a band inversion. Then, the Seff decreases with Qf2.
2.7. Alloying of Aluminium in Silicon
Within this work, aluminium is alloyed with n-type silicon at the rear of the wafer to form the emitter. This takes place while the Al-Si interface gets molten, which starts at the melting point of Al at 660 , when the co-firing process takes place. Huster [25] describes the formation of the back surface field (BSF) within a p-type doped silicon wafer. Since the background doping has no influence on the alloying process, the emitter formation is explained using the model proposed by Huster. In figure 2.14, the binary phase diagram of aluminium and silicon is given. The phase diagram shows the thermodynamic behaviour for different concentrations of aluminium-silicon alloys and therefore is key for understanding the formation of the emitter.
The first step starts with bringing the aluminium accurately defined on the wafer, which is done by screen-printing an aluminium containing paste on the rear surface of the wafer. Besides aluminium in particles in sizes from 1-10 µm, which are encapsulated with a thin aluminium oxide layer, the paste also contains glass frites for improved sintering, organic binders and solvents for improved printing properties such as viscosity and stability on fresh air.
In order to have defined edges and structures, which are covered with aluminium paste, a screen is used as a mask, which is made out of small braided stainless steel wires, that are covered with a resist. Typically, about 5 to 7 mg/cm2 is screen-printed on the rear, leading to a thickness of approx. 30-40 µm [25].
The alloying process takes place within an infra-red belt firing furnace, in which the wafer is placed on a chain and is transported through different temperature zones. The zone’s temper-ature and the chain speed can be varied, which was done within this work.
While the screen-printed wafer passes through several temperature zones within the furnace, temperature changes lead to different conditions of the Al-Si interface, moving the alloy along
2.7. Alloying of Aluminium in Silicon
Figure 2.14.: Part of the Al-Si binary phase diagram after Murray [36] with different interesting points highlighted
the liquidus curve, which is explained below and schematically shown in figure 2.15
Figure 2.15.: Schematic representation of the aluminium alloying in silicon with A the drying of the paste, B heating up, C the melting, D the reached peak temperature, E the actual alloying process andF the solidification
A - Paste drying A short anneal for about one minute leads to the evaporation of incorporated solvents, creating a porous matrix of aluminium filled oxide shells, with total residual paste contents of 50-70 % aluminium [25].
B - Heating up After drying of the paste, increasing temperature in a bigger, second chain firing furnace leads to a burning out of the remaining organic binders, while the wafer is
2.7. Alloying of Aluminium in Silicon
heated up to a few hundred degrees. When a temperature of 660degrees is reached, the aluminium in the oxide shells begins to melt. The oxide layers locally break, leading to direct contact between the aluminium and silicon. The melting can be seen in the measured temperature profile of the wafer, as a small plateau occurs as a result of the latent heat. Silicon dissolves in the aluminium melt, following the liquidus curve of the silicon and aluminium alloy. The oxide shells, which are natively about 1 nm thick [25], get thicker and stabilize the residual paste matrix.
C - Melting With increasing temperature, more and more silicon dissolves in the increasing amount of molten aluminium. The locally broken oxide shells keep on growing in thickness and stabilize the porous matrix even further. Aluminium and silicon both experience strong diffusion in opposite directions. While the silicon dissolves in the molten inner part of the shells travelling through the openings of the oxide shells caused by sintering of the glass frittes, the aluminium needs to flow in direction of the wafer surface.
D - Peak At the maximum temperature of the co-firing process, about 30 atomic-% silicon has dissolved in the liquid aluminium. A liquid Al-Si lake should, under ideal situations, cover the whole rear of the solar cell.
E - Alloying During cooling down, the silicon transport is reversed, as the system moves back along the liquidus curve in the binary phase diagram to lower solubility for lower temper-atures. While more and more silicon is rejected from the melt, it recrystallizes epitaxially at the rear surface of the wafer, incorporating aluminium atoms according to their solid solubility at a given temperature. This leads to the formation of a positively doped region with a surface concentration of up to 3·1019 atoms per cm2 (see figure 2.16).
F - Solidification Reaching 577 , the remaining aluminium lake solidifies, leading to the for-mation of the eutectic layer with 12.2 atomic-percent of silicon remaining.
Figure 2.16.: Typical emitter profile measured by ECV within this work
Typical aluminium-alloyed emitter have a profile similar to the one shown in figure 2.16.
In general, alloying depth, surface concentra-tion and trend of the curve are depending on the used aluminium paste. Ordinarily, the ex-act composition is unknown. The used pastes are described by acronyms, that are used here at the University of Konstanz. A typical value within this work for the surface concentra-tion of positively doped surfaces was around 2-3·1019 atoms cm-3, which leads to an over-all sheet resistance of roughly 10 Ω/sq for an emitter depth of 5µm. This high surface con-centration is mainly attributed to boron that is added to the aluminium paste.