Firing and Gettering Dependence of Effective Defect Density in Material Exhibiting LeTID

Daniel Skorkaa), Annika Zuschlagb) and Giso Hahnc)

University of Konstanz, Department of Physics, 78457 Konstanz, Germany

a)Corresponding author: daniel.skorka@uni-konstanz.de b)annika.zuschlag@uni-konstanz.de

b)giso.hahn@uni-konstanz.de

Abstract. The investigation of LeTID effects in mc-Si material requires repeated measurements of the minority charge carrier lifetime eff. Due to the inhomogeneous nature of mc-Si, measurements of eff should be carried out in a spatially resolved way. In this work, we show results for the impact of firing and gettering on LeTID in high quality mc-Si material and analyse them in terms of the effective defect density Neff while retaining the spatial information. We show that especially for low peak firing temperature degradation is not homogeneous and that firing and gettering influence the degree of spatial inhomogeneity.

INTRODUCTION

Multicrystalline (mc) Si material with p-type doping exhibits a recently discovered degradation effect called Light and elevated Temperature Induced Degradation (LeTID), which has received a lot of attention because it causes efficiency losses in PERC(Passivated Emitter and Rear Cell)-type solar cells of up to 10%rel [1, 2]. It could be shown that gettering [3, 4] and firing temperature [5] as well as the cool-down temperature ramp after the peak firing temperature [6] influence the extent of degradation.

Lifetime samples offer more flexibility with regard to processing steps and direct access to effective minority charge carrier lifetime eff and are therefore often used for LeTID investigations. Typically, data is presented as some weighted mean of eff versus treatment time, where the resulting average value is formed from a ring-shaped area in the sample above the coil containing areas of high and low eff [7]. This has the disadvantage that spatial information is lost; in addition, while eff is a commonly used metric, it is influenced by a multitude of factors, with LeTID being only one. In this work, measurements of eff are spatially and temporally resolved to study and understand the behaviour of the defect in more detail. To this end, we employ Time-Resolved PhotoLuminescence Imaging (TR-PLI) [8] which is one method of several available to obtain spatially resolved values of eff. Its specific advantage in application is that it is a fast, calibration-free imaging method. A disadvantage of this method, shared with most PL imaging methods (but also the mode of operation of a solar cell in the field), is that this image of eff is obtained at a constant illumination level, and hence not at a constant level of excess charge carriers n, To better assess the changes under the assumption of a single defect, we use the effective defect density

(1)

which is defined as the change in recombination rate with respect to a chosen initial state.

SiliconPV 2018, The 8th International Conference on Crystalline Silicon PhotovoltaicsAIP Conf. Proc. 1999, 130015-1–130015-5; https://doi.org/10.1063/1.5049334

Published by AIP Publishing. 978-0-7354-1715-1/$30.00

130015-1

Konstanzer Online-Publikations-System (KOPS) URL: http://nbn-resolving.de/urn:nbn:de:bsz:352-2-18okq8fmiekc55

Erschienen in: SiliconPV, the 8th International Conference on Crystalline Silicon Photovoltaics / Ballif, Christopher et al. (Hrsg.). - Melville, NY : AIP Publishing, 2018. - (AIP Conference Proceedings ; 1999,1). - 130015. - ISBN 978-0-7354-1715-1

https://dx.doi.org/10.1063/1.5049334

EXPERIMENTAL

Sample Preparation

Six lifetime samples were prepared from ~1 cm B-doped mc-Si material of high quality in the following way: Sister wafers of 50x50 mm² square size were etched in hot KOH and HF/HNO3/CH3COOH to remove surface damage due to wafer cutting. After surface cleaning using a hot H2SO4/H2O2 solution and HF dip, samples B1-B3 received a POCl3 diffusion in a quartz tube furnace (55 �) and the phosphorus silicate glass (PSG) and emitter was removed afterwards in diluted HF. The emitter was then removed using HF/HNO3/CH3COOH. Samples A1-A3 received no diffusion. Afterwards, all samples were again cleaned in a hot H2SO4/H2O2 solution with a following HF dip. Finally, SiNx:H was deposited on both sides using a commercial direct plasma PECVD tool (see Tab. 1).

TABLE 1. Process flow of B-doped mc-Si samples under investigation. Set A Set B

Etch & cleanPhosphorous diffusion, 55 �

PSG & emitter etch Clean

Both sided SiNx:H deposition, 75 nmFiring, 670°C / 730°C / 810°C measured wafer peak T

Degradation

Samples were fired with a measured peak sample temperature of either T1 = 670°C, T2 = 730°C or T3 = 810°C in a fast firing belt furnace. Directly after firing, the spatially resolved eff of each sample was measured using TR-PLI with a spatially homogenous, but temporally modulated, illumination equivalent to about 0.7 suns. The samples were then subjected to degradation treatments at 75°C on a hotplate under 1 sun illumination. Periodically, the samples were removed from the hotplate for eff measurement using the same conditions as applied directly after firing (at room temperature). The samples were then placed back onto the hotplate.

Results

The degradation of eff can easily be seen from maps of eff before the degradation treatment (“initial”) and at complete degradation (“degraded”, at lowest value of the harmonic average of eff of the complete sample after 150h to 300h) as given in Fig. 1 (ungettered samples A) and 2 (gettered samples B), top and middle row. However, since the starting value of eff varies both within one sample due to material inhomogeneities caused by the underlying defect structure and between samples due peak firing temperature influencing surface passivation and varying application of gettering, no clear statement can be made about whether the degradation is the same in all samples and all areas.

Within the presented experiment we only measure eff, and cannot directly determine bulk, nor can we separate the LeTID effect from other effects that might occur in the bulk at the same time. Here, we assume that neither surface passivation nor other defects show appreciable changes during the investigated time period.

The maximum defect density at each point, Neff(t)|max (which is not reached at the same time for every point), can be seen in Fig. 1 and 2, bottom row, to vary strongly within samples A1 and B1, leading to an inhomogeneous distribution. Within this experiment, n during TR-PLI measurement ranges from 5·1013 cm-3 to 5·1015 cm-3. This affects the evaluation of Neff(t)|max by giving a larger difference than would be obtained at constant n in all areas of the sample. However, due to the shallow injection dependence of the LeTID effect [9], we estimate that the relative error between areas within one sample is considerably lower than the observed variation.

Compared to the map of initial eff (top rows), we observe that the highest defect densities correlate with regions of low starting eff (e.g., area II). In contrast, samples fired at 730°C show less and those fired at 810°C the least difference between grains, leading to a more homogeneous distribution. There is a general trend, where degradation becomes more homogeneous, but also stronger, with increasing firing temperature. Gettering acts to reduce Neff in areas like area II, while leaving Neff in areas of high starting eff like area I virtually unaffected, compare for example samples A2 and B2, or A3 and B3. Note the same scaling in Fig. 1 and 2.

130015-2

A1: ungettered, 670°C A2: ungettered, 730°C A3: ungettered, 810°C

initi

al

log 1

0(ef

f / μ

s)

degr

aded

log 1

0(ef

f / μ

s)

defe

ct d

ensi

ty

log 1

0(N

eff(t

)| max

/ s-1

)

FIGURE 1. Map of eff in initial state (top row) and at maximum degradation (of the harmonic average) (middle row) as well as map of maximum effective defect concentration of each individual point (bottom row) for ungettered samples A1-A3. Note the

logarithmic scaling of the colour axis. Area II in the samples shows some internal structure. A similar pattern is visible in eff before degradation and after

subsequent regeneration (not shown) for all samples, see Fig. 3. However, the pattern appears washed out in Neff for higher firing temperatures and/or in gettered samples.

Area I

Area II

130015-3

B1: gettered, 670°C B2: gettered, 730°C B3: gettered, 810°C

initi

al

log 1

0(ef

f / μ

s)

degr

aded

log 1

0(ef

f / μ

s)

defe

ct d

ensi

ty

log 1

0(N

eff(t

)| max

/ s-1

)

FIGURE 2. Map of eff in initial state (top row) and at maximum degradation (of the harmonic average) (middle row) as well as map of maximum effective defect concentration of each individual point (bottom row) for gettered samples B1-B3. Sample B1

shows an additional vertical structure due to a crack. Note the logarithmic scaling of the colour axis.

log 1

0(ef

f / μ

s)

log 1

0(N

eff(t

)| max

/ s-1

)

eff initial Neff(t)|max

FIGURE 3. Maps of initial eff and Neff(t)|max in areas I and II for all samples. Note the internal structure of area II, visible more strongly at low firing T and without gettering. The crack in sample B1 was excluded. Note the logarithmic scaling of the colour

axis.

A1

A2

A3

B1

B2

B3

I II

I II

Area I

Area II

130015-4

DISCUSSION

We show quantitatively that the strength of LeTID can be very different in different parts of a mc-Si sample and is in fact not homogeneous at all, especially at low firing temperatures.

Higher firing temperatures lead to a stronger and more homogeneous distribution of the effect. This could be explained by a distributed impurity, possibly existing in the form of precipitates before firing, as proposed in [5].

It is already known that a gettering step can reduce the degree of LeTID [3,4]. In our work, we present evidence that gettering reduces Neff only in areas like area II (area of lower starting eff). However, this behaviour would be surprising for a distributed (metal) impurity, since gettering should be more efficient in good grains like area I [10] due to the absence of competing internal gettering sites. It is also suspected that gettering can reduce the dislocation density [11].

It has also been proposed that hydrogen might be involved in the LeTID effect [3, 6, 12, 13]. Under the assumption that hydrogen in some form contributes to the effect either directly or indirectly (instead of alleviating it, as proposed by [12]), these observations might be explained by the fact that at high firing temperatures, more hydrogen could be released from the dielectric layers into the Si bulk. The structures in Neff for area II become less pronounced at the same time, suggesting that the maximum defect concentration might be related to the hydrogen concentration retained in the bulk after firing, and the hydrogen concentration in turn related to the dislocation density.

CONCLUSIONS

At lower firing temperatures, LeTID related degradation occurs to a smaller extent and is less homogeneous within one mc-Si sample. Gettering reduces the extent of degradation only in areas with initially low lifetime. These observations point to an involvement of hydrogen released from the SiNx:H layers into the silicon bulk during the fast firing step in the formation of the LeTID effect.

ACKNOWLEDGEMENTS

Part of this work was supported by the German BMWi under contract numbers 0324001, 0324204B and 0325763B. The content is the responsibility of the authors.

REFERENCES

1. K. Ramspeck, S. Zimmermann, H. Nagel, A. Metz, Y. Gassenbauer, B. Birkmann, A. Seidl, "Light induceddegradation of rear passivated mc-si solar cells," in Proc. 27th EUPVSEC, Frankfurt am Main, Germany, 2012,pp. 861-865.

2. F. Kersten, P. Engelhart, H.-C. Ploigt, A. Stekolnikov, T. Lindner, F. Stenzel, M. Bartzsch, A. Szpeth, K. Petter,J. Heitmann, J.W. Müller, Sol. Energ. Mat. Sol. Cells 142, 83-86 (2015).

3. A. Zuschlag, D. Skorka, G. Hahn, Progr. Photovolt. Res. Appl. 25(7), 545-552 (2017).4. S. Chakraborty; Y. Huang; M. Wilson; A. Aberle; J. Li, Phys. Status Solidi A 44, 1800160 (2018).5. D. Bredemeier, D. Walter, S. Herlufsen, J. Schmidt, AIP Advances 6(3), 35119-035119-8 (2016).6. R. Eberle, W. Kwapil, F. Schindler, M. Schubert, S. Glunz, Phys. Status Solidi RRL 12(10), 861-865 (2016)7. D. Kiliani, G. Micard, B. Steuer, B. Raabe, A. Herguth, G. Hahn, J. Appl. Phys. 110, 054508 (2011).8. D. Kiliani, A. Herguth, G. Micard, J. Ebser, G. Hahn, Sol. Energ. Mat. Sol. Cells 106, 55-59 (2012)9. D. Bredemeier, D. Walter, S. Herlufsen, J. Schmidt, En. Procedia 92, 773-778 (2016)S. Gindner, P. Karzel,

B. Herzog, G. Hahn, IEEE J. Photovolt. 4(4), 1063-1070 (2014).10. S. Gindner, P. Karzel, B. Herzog, G. Hahn, IEEE J. Photovolt. 4(4), 1063-1070 (2014).11. D. Fenning, A. Zuschlag, J. Hofstetter, A. Frey, M. Bertoni, B. Lai, G. Hahn, T. Buonassisi, IEEE J. Photovolt.

4(3), 866-873 (2014)12. K. Nakayashiki, J. Hofstetter, A. Morishige, A. Tsu-Tsung, D. Needleman, M Jensen, T. Buonassisi, IEEE J.

Photovolt. 6(4), 860-868 (2016)13. T.H. Fung; M. Kim; D. Chen; C. Chan; B. Hallam; R. Chen; D. Payne; A. Ciesla, S. Wenham; M. Abbott, Sol.

Energ. Mat. Sol. Cells 184, 48-56 (2018)

130015-5