Download PDF-Version of this File (940 kB)

A NEW PASSIVATION METHOD FOR EDGE SHUNTS

OF SILICON SOLAR CELLS

M. Hejjo Al-Rifai, J. Carstensen, and H. Föll

Department of Materials Science, Faculty of Engineering, Christian-Albrechts-University of Kiel

Kaiserstr. 2, D–24143 Kiel, Germany

ABSTRACT: The efficiency of silicon solar cells (SC's) can be dramatically degraded by shunts which are usually localized at the edge area. These shunts are introduced during the different production steps of SC's. A new and simple passivation method has been developed to remove edge shunts by an optimized etch procedure using a KOH solution without a significant decrease of the area and thus the short circuit current I_sc. The method leads to an improvement of the relative efficiency h of 10-30% for afflicted SC's while good cells do not suffer any degradation. An improvement of up to 100% can be reached in situations where the SC's have a high density of edge shunts. The magnitude of possible improvements can be estimated by analyzing the temperature dependence of the leakage current. SC's which show improvement after passivation show a (dark) leakage current decreasing with temperature which indicates that the edge shunts are essentially ohmic resistors. In shunt-free SC's the leakage current increases with temperature as would be expected for a p-n-junction. The new method is cheap, efficient, does not degrade good SC's and can be easily added to the production lines at low costs.

Keywords: Shunts - 1: Passivation - 2: Solar Cell Efficiencies – 3

1. INTRODUCTION

Many Silicon SC's show a decrease in efficiency due to localized defects in the pn-junction [1,2] which are usually introduced by the production process, e.g. after the metallization, or the formation of the back and front side contacts, or after the doping of the emitter. These defects usually act as shunts and may occur in the volume as well as at the edge of SC's. Shunts always decrease the efficiency of the SC by increasing the leakage current and decreasing the maximal power point P_m, the open-circuit voltage V_OC, and the fill factor FF [1,2,3,4]. Shunts can be classified in two kinds (Fig. 1). "Volume shunts", the first kind, are very difficult to passivate without destroying a big area of the SC. "Edge shunts", the second kind, are the main subject in this paper. A non-representative statistic from our lab, obtained from the characterization of many commercial available SC's by the newly developed CELLO-techni que [5,6,7], indicates that about 80% of shunts in silicon SC's are edge shunts while just about 20% are volume shunts.

Passivation, i.e. effective removal of edge shunts is therefore of considerable interest to SC manufacturing. Two passivation methods are known from the literature. In the first one (called "coin staking"), the SC's are stacked

and the edges of the "stake" are subsequently plasma etched [8,9]. This method is effective but also quite expensive. With the second method, the edges of a SC are simply cut off with a diamond saw [2]. While this method is relatively cheap, it reduces the area of the SC and thus the short circuit current I_SC.

In this paper we will present a new passivation method for edge shunts which is efficient, cheap, and does not reduce I_SC. It will also be shown that monitoring the (temperature dependent) leakage current before and after a passivation provides a sensitive tool for estimating the possible benefits of a passivation and the gain in performance after the passivation.

2. PASSIVATION METHOD

The principle of the new passivation method employs local chemical etching of the edges of SC's with preferential attack of the defects causing shunts. It is well known that a hot alkaline solution (like KOH) allows to preferentially etch defects in silicon wafers and in SC's [5,10,11,12,13], so it is tempting to use such an etchant for removing and thus passivating shunts. However, just dipping the edge of a silicon SC into hot KOH, always results in a considerable loss of area and thus short circuit current I_SC, because parts of the p-n-junction will be removed in a region of about one mm around the edges of the SC. An additional loss of area could occur due to splashes of hot KOH; moreover, the KOH vapor tends to destroy the AR coating, too. Restricting etching to the edges, avoiding splashes, and removing any vapor, while not impossible, is quite difficult to achieve and needs costly equipment and procedures.

To avoid these problems, a new and simple KOH passivation technique has been developed. First, a defined quantity of cold (i.e. room temperature) KOH is deposited on the edge area of the SC. For this one edge is wetted locally with a cold solution of KOH (10-30%), e.g. by simply touching a sponge saturated with KOH (Fig. 2a).

In the second step, the wetted edge is heated up locally by using e.g. a heating plate kept at 80-120°C (Fig. 2b). The other edges can be passivated sequentially, or, with slightly more sophisticated handling equipment, in parallel. Optionally a cleaning step with distilled water may follow, but the results obtained so far indicate that this might not be necessary.

For the characterization of the effects of this passivation procedure, many SC’s have been characterized before and after the passivation of the edge shunts by the newly developed CELLO-technique [5,6,7] which allows to localize shunts and other defects in a SC. In addition, the IV-curves of the SC have been measured before and after the passivation at 25°C under constant illumination in order to monitor relative changes in I_SC, V_OC, FF, and h . Furthermore, to investigate the kind of shunts [14], and the possible improvement of the SC parameter after the passivation, the temperature dependence of the IV-curves in the dark has been measured, simulated and fitted using two different kinds of shunts in the equivalent circuit.

The new passivation method has been tested and applied to several kinds of good (without shunts) and bad (with strong edge shunts) commercially available silicon SC’s (mono-, multi-crystalline and ribbon silicon SC’s) from several sources.

3. RESULTS AND DISCUSSION

3.1 Passivation

The following example shows results for a mono-crystalline silicon SC (10x10 cm²). The CELLO shunt map before the passivation (Fig. 3a) indicates strong edge shunts (black areas). After the passivation almost all edge shunts have been removed (large reduction of the black areas) (Fig. 3b).

The strong improvement of the main SC parameter after the passivation can also be clearly observed in the IV-curves (c.f. Fig. 4). The fill factor increased from 0.68 to 0.78, the relative efficiency increase is 12%. At the same time a slight reduction in the short circuit current I_sc, i.e. an effective area loss of the SC, has been measured. This reduction is well overcompensated by the shunt reduction and might be reduced by optimizing the process.

The leakage current in the dark is the most sensitive parameter for shunts in the SC. In Fig. 5 the IV-curves in the dark (reverse bias) of the SC before and after the passivation are shown. They demonstrate the large improvement in the leakage current. Generally, improve-
ment in the leakage current are often observed even in cases where only a small or no measurable improvement of the direct SC parameters, was obtained.

3.2 Leakage current temperature dependence

A particular pronounced example for the temperature dependence of leakage currents in the dark before and after the passivation is shown in Fig 6.. Despite of a strong reduction of the leakage current after the passivation, the IV-curves before (Fig. 6a) and after (Fig. 6b) the passivation show the same inverse behavior of leakage currents with increasing temperature T. Obviously the majority of edge shunts has been removed after the passivation because the influence of temperature after the passivation has become weaker. This peculiar inverse temperature dependence of the dark leakage current has only been observed for SC's which are susceptible for passivation. Good SC's having no edge shunts, or bad SC's which do not improve upon passivation, never showed this inverse temperature dependence.

If this would be true, one must expect that the SC of Fig. 6 still contains shunts amenable to passivation and that a second passivation should have a beneficial effect. This proved indeed to be the case; applying the passivation procedure a second time yielded a further improvement of the SC.

4. DISCUSSION

To understand and describe the peculiar behavior of the leakage current with the temperature for SC's amenable to edge passivation, a new equivalent circuit diagram will be suggested (Fig. 7). Two different kinds of shunts are introduced. The first kind, symbolized with R_shFdoes not depend strongly on the temperature (no large positive temperature coefficient, but possibly some voltage dependence); it will in all experience represent process induced volume shunts. The second kind, symbolized with R_shT is a purely ohmic shunt with a strong temperature dependence (temperature coefficient comparable to metals) which is responsible for the peculiar temperature behavior of the leakage current.

The measured IV-curves of leakage currents in the dark at different temperatures have been fitted to this equivalent circuit, representative results are shown in Fig. 8. The results before (Fig. 8a) and after the passivation (Fig. 8b) show a very good fit to the measurements if only R_shuntis increased (to infinity for perfect passivation). The resulting values for R_shuntchanged from

before the passivation to

after the passivation.

The ratio between the linear and the constant part of the ohmic resistor before and after the passivation is nearly constant:

Ratio_before = 0.00205 =Ratio_after = 0.00168

This indicates that the character of the shunts in the SC before and after the passivation did not change, but only the density of the shunts. Note that the temperature coefficient for the change of the resistivity for metals is around 0,4%/^oC or 0,004 if expressed as a ratio as above.

The edge shunts in Si SC's may result from several process steps during the fabrication of SC's. A practical passivation method of edge shunts thus must be able to remove the majority of shunts, must not be harmful to the SC's without shunts, and it must be cheap, simple, and applicable at different steps of the production, i.e. after the emitter diffusion, or before the AR coating.

The new passivation method meets all the required properties. The new passivation method might be applied for all SC's produced since it does not degrade good SC's, or just for tested SC's which might have shunts. A classification of typical shunt characteristics encountered in production is possible by monitoring the dependence of the leakage current on temperature for representative SC's. The passivation of edge shunts not only can improve the efficiency of individual SC's; additionally it can improve the reliability of SC modules since fewer 'hot spots' exist.

5. CONCLUSION

A new passivation method for the edge shunts in the silicon SC’s has been developed. This method can be easily added to production lines; it is cheap, efficient, and does not degrade good SC’s. The method is especially effective in removing the ohmic edge shunts. The temperature behavior of the leakage current may be used to obtain information of the kinds of shunts encountered and on the general amenability of a SC to passivation.

REFERENCES

[1] K. R. McIntosh, C. B. Honsberg, Proceedings of the 16^th Photovoltaic Solar Energy Conference, Glasgow Ed (2000).

[2] P. P. Altermatt, G. Heiser, M. A. Green, Progress in Photovoltaics: Research and Applications, 4 (1996) 335.

[3] J. Pearton, W. Corbett, M. Stavola, Hydrogen in Crystalline Semiconductors, Springer Ed (1992) 28.

[4] R. Hull in: H. Hieslmair, S. A. McHugo, A. A: Istratov, E. R. Weber, Properties of Crystalline Silicon, INSPEC Ed (1998) 777.

[5] J. Carstensen, G. Popkirov, J. Bahr, H. Föll, Proceedings of the 16^th Photovoltaic Solar Energy Conference, Glasgow Ed (2000).

[6] M. H. Al-Rifai, J. Carstensen, H. Föll, Proceedings of the EUROMAT Conference, 13, München VCH Ed (1999) 273.

[7] M. H. Al-Rifai, J. Carstensen, H. Föll, Proceedings of the E-MRS, Strasburg (2001) (to be published).

[8] H. J. Hovel, Semiconductors and Semimetals, 11, New York Ed (1975) 61.

[9] L. D. Patrin, Solar Cells and their Applications, Wiley, Ed (1995) 60.

[10] H. Löwe, P. Keppel, D. Zach, Halbleiterätzverfahren, Akademie-Verlag, Ed (1990) 67.

[11] Q. Vu, D. A. Stricker, P. M. Zavracky, J. Electrochem. Society 143 (1996) 4.

[12] W. Menz, J. Mohr, Mikrosystemtechnik für Ingenieure, Wiley-VCH Ed (1997) 198.

[13] M. Köhler, Ätzverfahren für Mikrotechnik, Wiley Ed (1998) 321.

[14] S. Langkamp, O. Breitenstein, Proceedings of the E-MRS, Strasburg (2001) (to be published).