Investigation on the Long-Term Stability of AlO x /SiN y :H and SiN y :H Passivation Layers during Illuminated Annealing at Elevated Temperatures

. Most crystalline Si based solar cells, e.g. passivated emitter and rear cells, rely on SiN y :H and AlO x /SiN y :H passivation layers. In this work, the long-term behavior of minority charge carrier lifetime in such symmetrically passivated samples during illuminated annealing at elevated temperatures is investigated by means of photoconductance decay based lifetime measurements, corona charging and capacitance voltage measurements. Thereby, AlO x layers, which are known to reduce H in-diffusion due to their barrier properties, deposited by atmospheric pressure chemical vapor deposition as well as by atomic layer deposition were considered enabling a comparison of different deposition techniques. The frequently published behavior of the bulk related degradation could be confirmed and the qualitative correlation between maximum defect density and the changing total amount of H in the Si bulk due to the barrier properties of the individual layers dielectric layers could be shown. Furthermore, for the subsequently observed degradation accelerated by a treatment at higher temperatures, literature indicates degradation to be caused by surface related degradation. Investigations on field effect passivation during degradation by means of corona charging and CV measurements showed a large drop in fixed negative charges in the passivation layer stacks.


Introduction
Bulk related degradation (BRD) phenomena, such as light-and elevated temperature-induced degradation (LeTID), can reduce bulk minority charge carrier lifetime and thereby affect the efficiency of Si solar cells.However, the effective minority charge carrier lifetime eff may suffer from surface related degradation (SRD) as well [1], [2].Literature indicates that SRD can occur in samples passivated by SiNy:H and AlOx/SiNy:H [2] both passivation layer systems found in passivated emitter and rear cells (PERC).For both, LeTID and SRD, it is known that properties of the passivation layer can influence the kinetics of LeTID [3] and SRD [4].In this work, short-and long-term stability of eff in samples passivated by SiNy:H and AlOx/SiNy:H during illuminated annealing at elevated temperatures is investigated.AlOx layers from two different deposition tools, atmospheric pressure chemical vapor deposition (APCVD) and atomic layer deposition (ALD), are considered to check whether different AlOx layer microstructures have an impact on degradation kinetics as already indicated in a previous study [5].

Experimental
For the degradation experiments ~1 Ωcm B-doped FZ-Si wafers serve as base material.After Preprint an HF etching step, 5-18°nm AlOx is deposited on both sides by two different deposition techniques, APCVD at 590°C and ALD at 300°C, except for reference samples.All samples are then coated on both sides with 75 nm SiNy:H via plasma-enhanced chemical vapor deposition and subsequently fired at a sample peak firing temperature of Tsample,peak = 800°C.LeTID related degradation and regeneration is carried out at 80°C and 0.9(1) suns.After reaching a maximum in eff, indicating that SRD starts to become dominant, treatment temperature is increased to 200°C to accelerate this second degradation.eff is determined at room temperature using photoconductance decay (PCD) evaluated at an excess charge carrier density Δn = 0.1 p0 ≈ 1.5•10 15 cm -3 , with p0 being the base doping.From PCD measurements, the lifetime equivalent maximum defect density (Δn  0) ΔN*leq, is calculated as described in [6], [7].Additionally, the surface saturation current density j0 is determined.The evaluation of j0 during both, BRD and SRD was performed by the j0 difference analysis according to [7].
For the investigation of the passivation mechanism during SRD, the alternating PCD measurement and applying of corona charging (CC) are performed at two specific times, directly at the beginning of the temperature treatment at higher temperature (t2,initial) and at the eff minimum of SRD (t2,min).In addition, the fixed charge density Qf in the SiNy:H and AlOx/SiNy:H layers is determined by capacitance voltage (CV) measurements directly after firing and at the eff minimum of SRD.For CV measurements, Al was evaporated on one side of the samples followed by production of laser fired contacts (LFC).The evaluation of the CV data is performed according to [8] via the determination of the slope of the 1/C 2 curve.The H content in the Si bulk is analyzed via resistivity measurements to determine the amount of acceptor-hydrogen pairs according to [9].Compared to ALD AlOx, APCVD AlOx shows barrier properties which are not as strong.After switching the treatment temperature to 200°C, the differently passivated samples suffer from a second degradation, whereby the samples with AlOx/SiNy:H stacks show a significantly delayed degradation compared to the SiNy:H reference.The steady increase in j0 of the SiNy:H passivated sample suggests that this second degradation is related to surface passivation (SRD).A slight increase in j0 is also observable for the AlOx/SiNy:H stacks but especially for the thinner AlOx samples only to a very limited extent.Considering ΔN*leq, especially for the sample group with thinner APCVD AlOx, it can be observed that something seems to change in the bulk for >20 h.This can also be confirmed in Fig. 3 (right).The injection-resolved lifetime curves show a reverse behavior (lower lifetimes at low injection) for t >20 h, which can undoubtedly be assigned to a BRD.So far, it is not clear what triggers this phenomenon.Focusing on the surface, the behavior during CC of the samples is investigated, presented in Fig. 4, where the measured eff is shown depending on charging time.At the beginning of SRD, the respective eff minimum of the AlOx/SiNy:H stacks is almost at the same level.Compared to the SiNy:H sample, the samples with AlOx/SiNy:H stacks show a higher minimum.However, a different behavior can be observed at the minimum in eff of the following treatment at 200°C (Fig. 4, right).The minima of the AlOx/SiNy:H stacks are on a similar or even slightly higher level than at t2,initial, but the CC time has decreased tremendously.Conversely, it can be observed for the SiNy:H sample that the eff minimum is slightly decreased and the corona charging time is very similar compared to t2,initial.Note that a rather uncharged surface also leads to the fact that lifetime may not be described by an injection-independent j0 value.This may explain the drop in j0 during the second degradation (see Fig. 1, right) for the thinner AlOx layer stacks.For a comparison of the CC measurements, the CC duration until reaching the eff minimum is shown for the differently passivated samples in Fig. 5 (left).In addition, CV measurements are applied to measure the absolute fixed charge density which is shown in Fig. 5 (right).Samples with AlOx/SiNy:H layers show a decline of negative fixed charges during SRD measured by CC as well as by CV.However, the quantitative amount measured by CV cannot be directly compared with the corona charging time, especially for APCVD AlOx interlayers, which may be due to the differences in the layer microstructure.While the CC time of the SiNy:H passivated samples does not change significantly during SRD, the CV measurements show that there is a slight increase in positive fixed charge density.There is thus a net decline of negative charge density or net increase of positive charge density during SRD independent of the passivation type.However, which of these is the case cannot be clearly assigned by the CV measurement.This allows the hypothesis to be made that the net fixed charge is involved in the SRD and opens the further question whether the positively charged H species plays a key role in SRD.

Conclusion
It could be shown that AlOx layers from ALD and APCVD both serve as a barrier layer for H during firing.Since the ΔN*leq and -Δp values of the ALD AlOx/SiNy:H passivated samples are lower than those of the APCVD AlOx/SiNy:H passivated samples, it is concluded that the ALD AlOx layer has a stronger barrier effect and/or is a less effective source of H.This also confirms a correlation between ΔN*leq and H amount for LeTID.The second degradation for the SiNy:H passivated samples could be attributed to SRD by the steady increase of j0 and by the injection resolved lifetime.On the other hand, the AlOx/SiNy:H passivated samples with ALD and thin APCVD AlOx layers show only a slight increase in Δj0 and in addition the thin APCVD AlOx layer sample group showed a reverse dependence in the injection-resolved lifetimes for >10 h at 200°C, indicating another and yet unknown BRD effect.From the CC and CV measurements, a decline of negative fixed charges in the AlOx/SiNy:H layers could be observed during SRD while in the SiNy:H layer there is a slight increase in positive fixed charge density.

Figure 1 .
Figure 1.Measured eff (top) and resulting ΔN*leq (middle) and change in j0 (bottom) compared to initial values versus accumulated time for SiNy:H and AlOx/SiNy:H passivated samples.To trigger degradation and regeneration related to LeTID, samples are first illuminated at 80°C (left) and subsequently at 200°C for accelerated study of SRD (right).

Figure 2 .
Figure 2. Change in hole concentration -Δp versus accumulated time during dark anneal at 220°C for the differently passivated samples (fired).

Figure 3 .
Figure 3. Injection resolved τeff over accumulated time of the sample with SiNy:H (left) and of a sample with APCVD AlOx/SiNy:H (right) during degradation at 200°C (right).

Figure 4 .
Figure 4. Corona charging time at the eff minimum of the different samples with SiNy:H layer or AlOx/SiNy:H layer stacks at the beginning (left) and at the minimum in eff of the second degradation at 200°C (right).Negative charging for SiNy:H, positive charging for AlOx/SiNy:H.Thus, SRD in the SiNy:H passivated sample could be explained by the decrease in chemical surface passivation quality, whereas the second degradation, between ~1-10 h at 200°C, in AlOx/SiNy:H passivated samples, seems to be related to a decrease in field effect passivation.It is remarkable that despite the strongly different extent of initial BRD, both, samples with ALD and APCVD AlOx layers, show a very similar course during the second degradation.

Figure 5 .
Figure 5. (Left) Corona charging time required to reach the τeff minimum (in Fig. 4) of the different samples with SiNy:H layer or AlOx/SiNy:H layer stacks at the beginning (filled) and at the minimum of the second degradation (unfilled).(Right) Absolute fixed charge density |Qf| from CV measurements of the different samples with SiNy:H layer or AlOx/SiNy:H layer stacks directly after firing and after reaching the minimum of the second degradation.Blue colours: ALD AlOx, orange colours: APCVD AlOx.