EFG silicon material

By applying the standard photolithography based process with an additional short front side oxidation step, Käs et al. achieved the to date highest solar cell efficiency on EFG material [116]. Spatially resolved characterisation via LBIC (Figure 7-3, left side) reveals the very homogeneous grain quality, without larger regions of high recombination activity. Therefore, the 18.2% efficiency of this solar cell is not necessarily due to material limitations but also possibly limited by the applied solar cell process. Hence the development of a more advanced solar cell process which is described in the previous chapters was necessary. The first adaption is the implementation of a plasma texture on the front side, which is unfortunately not compatible with the front side oxidation step (see chapter 4.4). Table 7-3 shows the beneficial influence of the plasma texture on EFG solar cells without front surface oxidation. Although the Voc of plasma textured EFG solar cells is reduced, the increased jsc in total leads to higher efficiencies.

Table 7-3: Comparison of IV parameters for solar cells from EFG wafers with and without plasma texture.

Solar cells with Al-BSF and SiO2/SiNx stack on the Rear Side (RS)* are from neighbouring wafers, respectively. All solar cells feature a SARC. Al-BSF and SiO2/SiNx passivated solar cells are not comparable as they do not originate from neighbouring wafers.

RS Texture FF [%] j_sc [mA/cm²] V_oc [mV] η [%]

Mean values for EFG (3-4 solar cells per process)

Al-BSF none 78.3 ± 0.3 30.9 ± 1.4 585 ± 6.8 14.1 ± 0.7 Al-BSF PT 78.1 ± 0.3 32.2 ± 0.4 577 ± 4.3 14.5 ± 0.3 SiO₂/SiN_x none 75.5 ± 0.9 33.0 ± 0.1 621 ± 4 15.5 ± 0.3 SiO₂/SiN_x PT* 76.8 ± 0.7 35.2 ± 0.3 618 ± 7 16.7 ± 0.3

* No degradation due to the plasma textured surface occurs, as the rear side oxidation step is carried out before the plasma texturing step.

Next, the full are Al-BSF is replaced by a dielectrically passivated, locally contacted rear side. This leads to a better surface passivation and improved optical quality of the rear side. Thereby, Voc and jsc are improved.

As dielectric rear side passivation layers a SiO2/SiNx stack and an Al2O3/SiNx stack are investigated. Very encouraging results are obtained for the SiO2/SiNx stack on EFG material of very high quality (Table 7-3, lower half). Due to the high thermal budget of the oxidation, material of lower quality is significantly degraded by the “bleeding” of gettering sites and fine (re)dispersion of impurities in the EFG wafer (see chapter 6.1).

High quality EFG material usually exhibits a bulk resistivity of 2-3 cm, which is optimized for the industrial screen printing process. This yields high series resistance contributions from the rear side and thus significantly reduces the FF, if a reasonable LFC

pitch⁴¹ is chosen (see Table 7-3, lower half). 1 cm EFG material would be better suited for local contacting, but in most cases the material quality is lower and thus 1 cm EFG material can suffer from the high thermal budget of the oxidation step.

By replacing the SiO2 dielectric by an Al2O3 layer the thermal budget in the advanced solar cell process is significantly reduced due to the lower deposition temperature of the Al2O3 (see chapter 6.2). This allows 1 cm EFG material to be successfully processed.

Due to the lower bulk resistivity higher FFs than on the 2-3 cm EFG material passivated by the SiO2/SiNx stack are obtained. The mean efficiencies are comparable, which might be explained by the large bulk quality variation, indicated by the high variation of Voc. The highest efficiencies obtained for the respective materials show larger discrepancies mainly due to the higher FFs obtained on the 1 cm EFG material.

Table 7-4: Comparison of IV parameters for solar cells from EFG wafers with SiO2/SiNx stack or Al2O3/SiNx stack as rear side passivation. Mean values are from the batch that includes the best solar cell, respectively. All solar cells feature a plasma texture and a SARC.

RS R_bulk [cm] FF [%] j_sc [mA/cm²] V_oc [mV] η [%]

Mean values for 7 (not neighbouring) solar cells with SARC

SiO₂/SiN_x 2-3 75.7 ± 1.1 34.7 ± 0.6 608 ± 10 16.0 ± 0.7 Al₂O₃/SiN_x ~1 77.4 ± 0.6 34.1 ± 1.5 619 ± 14 16.3 ± 1.1

IV parameters of the best solar cells with SARC

SiO2/SiNx 2-3 76.8 35.5 623 17.0

Al2O3/SiNx ~1 78.2 35.7 628 17.5

The right side of Figure 7-3 depicts an LBIC measurement of a 1 cm EFG solar cell which was processed according to the advanced process described in chapter 3.2.2. The solar cell exhibits a similar efficiency (18.1%) as the Al-BSF solar cell fabricated on EFG material with 2-3 cm by Käs et al. [116] on the left side (18.2%).

Although the overall solar cell efficiencies are comparable, the LBIC scan reveals that the 1 cm EFG solar cell on the right side of Figure 7-3 shows a significant amount of efficiency limiting, defect-rich areas. This leads to the conclusion that significantly higher conversion efficiencies should be possible on EFG material exhibiting a similarly homogeneous material quality like the EFG solar cell depicted on the left side of Figure 7-3.

41 The pitch should be significantly higher than the diameter of a single LFC point, as otherwise the areal portion of the rather highly recombination active contact area on the rear side becomes too large and the average Srear drops below the Srear of a common Al-BSF.

Process evolution for the different materials under investigation

Figure 7-3: LBIC scan of the best Al-BSF solar cell on 2-3 cm EFG material (left) and the best dielectrically passivated solar cell from 1 cm EFG material (right) [182], both with DARC. Although the overall cell performance is comparable, the solar cell on the right shows large areas with low crystal quality indicated by a low IQE. The maximum possible efficiency for dielectrically passivated 1 cm EFG material therefore lies well above 18% if only areas of high crystal quality are considered.

To address the special elongated grain structure of EFG wafers, asymmetric LFC patterns with higher contact density perpendicular to the grain structure than along the grains are investigated. This should reduce series resistance losses, as charge carriers in average have to cross a smaller number or even no grain boundaries when travelling to a single LFC to be extracted. Figure 7-4 depicts two asymmetric LFC patterns on the elongated grain structure of an EFG solar cell rear side. Experiments on neighbouring EFG solar cells which only differ in the rear side LFC pattern show, that the material variation over 2-3 cm of the crystal structure has much more influence on the series resistance than the LFC pattern. Therefore, experiments on a large number of neighbouring solar cells are necessary to validate the theory described above on a statistical basis.

Figure 7-4: Asymmetric LFC pitch layout for EFG solar cells. The EFG wafer with the vertically oriented, elongated grain structure is depicted in light gray; grain boundaries as black lines. The LFCs are depicted as darker gray points. The left side shows a pattern that minimizes the current transport through grain boundaries. Rotation of 90° yields the pattern on the right side. Some grains are not directly contacted, which forces more charge carriers to travel through grain boundaries before extraction.

The investigation of Light Induced Degradation (LID) on EFG solar cells shows that EFG solar cells are not affected by this phenomenon. This is in accordance with the expectations, as the oxygen content of EFG wafers is very low [78] compared to block cast mc or Cz Si wafers. The fast LID due to a high amount of transition metals (Fe and/or Cu) described in chapter 5.4 is also not observed in EFG solar cells.