Etched Back Aluminium Emitter Passivation

The passivation quality of aluminium oxide/silicon nitride and silicon nitride on differently deep etched back asymmetric aluminium emitter samples with a homogeneously etched back FSF of 100 Ω/sq is analysed. The assessment is done using the so called Implied-V_oc.

First, the emitter profile is given for a sample, that has not been etched back at all. Second, the analysis of the emitter passivation is carried out.

5.2.2.1. Sample Preparation and Evaluation Sequence

Figure 5.11.: Process sequence The samples were fabricated in a similar manner as the

FALCON solar cells that are presented in section 5.4.

Mainly, three exceptions can be given. The FSF was fully etched back to 100 Ω/sq and therefore the samples feature only a homogeneous FSF. Since the samples do not need a metallization, the rear passivation was not opened locally by a laser and no pastes were screen-printed on the passivation.

Starting from 6^” textured 11 Ωcm wafers, samples simi-lar to the previously in section 5.2.1 presented structure featuring a homogeneous FSF and SiO2/CT-SiNx front side passivation, were fabricated.One exception during processing was the single side etching, which was car-ried out on the rear in order to remove the highly doped phosphorus layer. After the front side passivation was finished, the emitter was alloyed on the rear side by full area screen-printing of aluminium paste (Al Q) and sub-sequent co-firing at a set peak temperature of 840‰and a belt speed of 4400 mm/min with the screen-printed side facing the top. The aluminium paste was etched off in a 37 % HCl solution for approx. 30 minutes. Sub-sequently, separation of the samples was carried out by using a Powerline 100D laser from Rofin, which in-cluded marking of the wafers on the rear side.

The alloyed emitter, which had a junction depth of ap-prox. 5µm was etched back in an approx. 20 % NaOH solution at 80‰ to various emitter depths. The etch back was monitored by weight differences prior and af-ter the etch back. Furthermore, resulting sheet

resis-tances were measured using a 4-point probe measurement assembly. This was followed by the first lifetime measurement using a Sinton Lifetime tester WCT-120, from which the saturation current density j0 and implied-Voc was calculated. This was followed by a further HCl/HF and Piraniha cleaning step. Two different rear side passivations were applied. One half of the samples were coated with aluminium oxide in the earlier described rpa-ALD of 15 nm thickness, whereas the other half were directly passivated by means of remote PECVD (SiNA-SiNx), which was also used for the first group as capping layer (in both cases about 150 nm thick). Then, the resulting implied-Voc and j0e values were measured again with the help of the Sinton Lifetime tester. All samples were then fired at a set peak temperature of 850‰ and a belt speed of

5.2. Aluminium Emitter Formation and Passivation

5400 mm/min. This was done to reproduce real solar cell manufacturing, in which the silicon nitride passivation is activated and thereby the incorporated hydrogen freed. Finally, a third PCD measurement was carried out.

5.2.2.2. Emitter Formation and Passivation Results

The emitter formation is analysed by mapping of sheet resistance with the help of a 4-point probe and ECV measurements. ECV measurements are based on successively etching back a semiconductor surface e.g. silicon. By exposing the surface to an electrolyte, which under reverse bias conditions is acting as a Schottky-like contact (for n-type silicon), it is possible to determine the electrically active dopant concentration. Detailed explanation on this measure-ment technique can be found in [52] or [51].

Figure 5.12.: Emitter profile obtained by ECV measurements for one non-etched back sample and indicated minimum etch back depth.

The surface peak of the non-etched sample exceeds 3×10¹⁹ atoms cm^-3 due to aluminium-rich residuals on the surface (eutectic and lamellas). The junction depth is approx. 5.5µm.

The emitter passivation quality was analysed by comparing measured implied-Voc values of the differently deep etched back samples prior to the deposition of the passivation layer/stack, after it and past the firing step. The measured sheet resistances of etched back samples were not usable, since they vary strongly over the whole etched back surface. Strong variations of the obtained j₀ values for similarly deep etched back emitter samples were found. This is due to the fact, that j0 is not calculated directly, but from the slope of a fit of the Auger corrected inverse lifetime data as a function of the minority charge carrier density [51].

Therefore, using the implied-V_oc as the emitter passivation quality criterion leads to much lower spread of the data, since it is directly calculated from the excess charge carrier density as it is shown in equation 3.13.

5.2. Aluminium Emitter Formation and Passivation

In figure 5.13 the measured implied-Voc values are given for non-passivated differently deep etched back emitter samples as well as for both rear side passivations.

Figure 5.13.: Measured implied-V_ocvalues for differently deep etched back samples prior deposi-tion of the rear passivadeposi-tion and after firing. Al₂O₃/SiNA-SiN_xand SiNA-SiN_xrear passivated samples are shown. Analysed at an injection level of ∆n = 5×10¹⁵cm^-3; Note: The dashed lines are a guide to the eye.

Starting from a non-passivated rear emitter, the measured implied-Voc values are increasingly reduced the closer the surface gets to the actual pn-junction and stay almost constant after the emitter is fully removed. This is in good agreement with the simulation shown in section 4.3.2 (for the respective SRV). Furthermore, that clearly shows the effect of the field-effect passivation due to the emitter.

After firing at a set peak temperature of 850‰ with a belt speed of 5400 mm/min, the pas-sivation quality of samples passivated by means of Al2O3/SiNA-SiNx is showing implied-Voc

values of up to 650 mV to 660 mV over the whole etch back spectrum from about 2 µm to 8 µm. In contrary to the expectations, no increase of the implied-V_oc was found compared to the non-etched unpassivated rear side. Therefore, the passivation of the etched back emitter is not leading to a reduced effective SRV. An improvement was expected, since Bock [72, 75]

showed enhanced rear side performance for etched back and passivated emitters.

Samples, passivated on the rear by SiNA-SiNxshow a similar result after firing of the passivation.

While for slightly etched back emitters, the implied-Voc results in higher values than for the Al₂O₃/SiNA-SiN_x passivation, the result changes with increased etch back depth and thereby lower field-effect passivation by the emitter. Then, the passivation quality is reduced for SiNA-SiNx passivated samples, resulting in lower implied-Voc values.

5.2. Aluminium Emitter Formation and Passivation

5.2.2.3. Renewed Passivation for Selected Samples

The passivation of selected samples was renewed, since the measured passivation quality was not improved compared to the unpassivated and non-etched case. Therefore the best applicable passivation was carried out in order to evaluate, if an improvement compared to the unpassivated rear can be nevertheless found.

First, the front side of selected samples from the last experiment was fully covered by protective wax using a Schmid Inkjet DoD 300. Then, the passivation on the rear side was removed in a 10 % HF solution. This was followed by a stripping step similar to the one described in section 4.1.2 in order to remove the protective wax on the front side. Before a renewed aluminium oxide deposit was possible, a HCl/HF cleaning sequence was carried out. The aluminium oxide was deposited in a FlexAl-reactor (rpa-ALD) at 300‰ for approx. 30 minutes, leading to a 30 nm thick layer on the rear side. After a 30 min anneal at 420‰ in a nitrogen atmosphere, the samples were again analysed using a Sinton lifetime tester at an injection level of ∆n = 5×10¹⁵ cm^-3 in QSS mode.

Figure 5.14.: Measured implied-Voc values for differently deep etched back samples prior to the first deposition of the rear passivation and after the second passivation, which further was annealed at 420‰ in a nitrogen atmosphere for 30 minutes. The samples were analysed at an injection level of ∆n = 5×10¹⁵ cm^-3; Note: The dashed lines are a guide to the eye

The data indicates, that even under a high quality passivation of annealed aluminium oxide, which usually leads to well passivated p⁺⁺ surfaces [70], the passivation quality of the etched back emitter could not be improved, since implied-V_oc values are reduced compared to non-etched unpassivated case. For shallow emitters, the passivation is performing better than the etched back unpassivated case. Further the bulk is well passivated using aluminium oxide, too.

5.2. Aluminium Emitter Formation and Passivation