You are on page 1of 4

ELIMINATING SHUNTS FROM INDUSTRIAL SILICON SOLAR CELLS BY SPATIALLY RESOLVED

ANALYSIS

S. A. Correiaa, J. Lossena, M. Bährb


a
ErSol Solar Energy AG®,
Wilhelm-Wolff-Str. 23, D-99099 Erfurt, Germany
b
Solar Zentrum Erfurt – CiS Institut für Mikrosensorik GmbH,
Konrad-Zuse-Straße 14, 99099 Erfurt, Germany

ABSTRACT: In this work a easy-to-use system was built and used for fast shunt analysis of Industrial silicon solar
cells using liquid crystal thermal foils. With this method, shunt detection is possible trough the visualization of the
local power dissipation under an external bias. Six main shunt types where identified. After a detailed
characterization their causes could be partially overcome. The shunt analysis enabled the statistic study of production
shunts and the improvement of the production process, especially in case of edge isolation.
Keywords: Shunts, Defects, Edge Isolation.

1 INTRODUCTION same shunt types as in lock-in thermography at reverse


bias at typical applied voltages between -2V and -10V in
This work focuses on the study of shunts in industrial very short times [7]. At the present time the foils
crystalline silicon solar cells. Shunts can heavily reduce available in the market have a temperature resolution
the solar cell’s conversion efficiency and decrease the around 1K [8].
module performance.
The efficiency of a cell is affected because shunts 2 METHODOLOGY
reduce the Fill Factor (FF) and the Open Circuit Voltage
(Voc). This effect becomes more dominant under low In this work, a device using liquid crystal foils was
light conditions [1]. built and used for shunt analysis in an industrial
A low shunt resistance can lead to hot-spots in environment. The analysis was performed at voltages
reverse biased cells, especially when the power between -0.5 and -20V. In addition optical microscopy,
dissipation occurs in a small area [2]. Scanning Electron Microscopy (SEM), Energy
Dispersive X-ray Spectroscopy (EDX) and Light Beam
1.1 Shunts in Crystalline Silicon Solar Cells Induced Current (LBIC) Analysis were used as
A shunt is a local increase in the dark forward current complementary analysis techniques.
of a cell. This increase can be caused by material defects A routinely analysis of industrial, off specification,
or it can be process induced. cells with a dark reverse current bigger than 3A at -10V
Material induced shunts can occur due to a high was performed. As a rule each 10th cell from a group of
density of dislocations, voids or impurities as well as 100 cells lot was analysed.
metal-decorated small angle grain boundaries, grow-in Further, shunt analysis was used as an auxiliary
macroscopic SixNy inclusions and inversion layers technique for the optimization of the laser edge isolation
crossing the wafer [3][4]. process.
Shunts can be created during processing by residues
of the emitter at the cell edge, by cracks and holes, by 2.1 Construction of the Device
scratches and by aluminium particles at the cell surface. The device consists of a temperature stable metal
Schottky type shunts can also occur below grid lines [4]. plate built as a vacuum chuck for cells, a DC power
Nine different types of shunts have been already source, a vacuum pump and a liquid crystal thermal foil.
found and identified in silicon screen printed crystalline The patterns in the LC foil can be registered and
solar cells by Lock-in Thermography [4]. They differ by catalogued with a digital camera. Fig. 1 shows a picture
their I-V characteristic (linear/ohmic or non-linear) and of the finalized device.
physical origin [4].

1.2 Shunt Analysis with Polymer Dispersed Liquid


Crystal Foils
Lock-in thermography is a well accredit technique for
shunt analysis. With a sensitivity limit well below 1 mK,
the technique enables the study of weak shunts at the
normal operating point of 0.5V forward bias or below
[5]. However, its analysis time is long when considering
a in-line utilization.
An alternative can be the use of temperature sensitive
polymer dispersed liquid crystal foils [6] which have a
temperature dependent reflectivity. For the foils used in Fig. 1: Picture of the device.
this work the blue light is preferentially reflected at high
temperatures and the red light at lower temperatures. The 3 RESULTS AND DISCUSSION
technique permits the immediate detection of almost the

1
Shunts due to the metallization
Fig. 2 shows the shunt distribution in relative The sintering parameters are set for high cell
numbers for a multi crystalline and a mono crystalline performance. However front metallization shunts become
material from off specification cells. The cells have a difficult to avoid when low contact resistance values need
dark reverse current bigger than 3A at -10V bias. The to be achieved. A thermal image of a shunt occurring at
sum of all frequencies is bigger than one because more the busbars can be seen in Fig. 4.
than one shunt type can occur simultaneously in one cell.
It is seen that multi crystalline cells are more likely to
be affected by material shunts than mono crystalline
cells. Besides that it can also be seen than shunts at the
front metallization busbars occur more often in multi
crystalline cells than in mono crystalline cells. The values
related to shunts due to scratches, cracks, besides paste
contaminations are used to screen handling issues during
processing.

Fig. 4: Thermal image of shunts at the front metallization


busbars.

Extreme firing conditions can cause the formation of


shunts on the back surface of the cell. Fig. 5 (Left) shows
the thermal image of a cell with shunts on the back side
metallization and Fig. 5 (Right) shows a SEM cross
section from the point marked in Fig. 5 (Left). It can be
seen that the Al metallization is fired to deep. The same
figure shows a big material void inside the formed Al/Si
crystallite.

Fig. 2: Shunt distribution for mono crystalline and multi


crystalline silicon cells.

3.1 Shunt Characterization


The shunt types that could be identified in 156x156
mm mono and multi crystalline cells, at the Ersol Solar
Energy AG® cell production line, are described bellow:

Edge shunts
A cell with edge shunts, is visualized in Fig. 3. The Fig. 5: (Left)Thermal image of a cell with shunts caused
effect of an edge shunt in the cell output will depend on by the back surface metallization. (Right) Cross section
the amount of emitter remaining on the cell edges and the SEM image of a crystallite grown at the back side of the
affected area. cell.

Process induced contaminations


The thermal image from Fig. 6 shows a cell with
point sized shunts at the front busbars and a shunt due to
an Al contamination occurred during screen printing.

Fig. 3: Thermal image of a linear edge shunt in a


156mmx156mm poly-crystalline cell with Rsh=0.32Ω at
-1V bias.

The increase of the relative frequency of this type of


shunt can be an immediate alert to a plasma failure or bad
Fig. 6: Cell with shunts due to the front metallization and
cell staking as main failure causes for plasma edge
Al contamination during screen printing. The red square
isolation. laser edge isolation can be ineffective if the cell
highlights the point contaminated with aluminium.
geometry is outside specified tolerances or if the wrong
laser parameters are being used.
Fig. 7 shows the cross sectional SEM picture of the
cell at the contamination shunt location. It can be seen

2
that one of the crystallites goes trough the emitter region
inside the cell. A darker region extends until Cracks and holes were reduced by applying more
approximately 500nm below the interface between the Al strict handling norms and by reducing manual handling.
crystallites and the Silicon. An EDX scan revealed that
this region is composed of Al and Si in equal Material induced shunts
concentrations. Material shunt detection is generally only possible by
this method at voltages above -15V to -20V and when
other shunt types don’t dominate. However cells with a
high density of crystallographic defects or material
contamination can be easily visualized at lower biases,
depending on how sever the shunt is.
The thermal map of Fig. 10 corresponds to a cell with
a material contamination induced during crystal growth.
The cell belongs to a group of cells from the same ingot
with the same shunt thermal image pattern. It can be seen
in Fig. 11 that the shunt correlates with a region with
Fig. 7: (Right) Cross sectional view of the shunt extremely poor LBIC response. Fig.12 shows the SEM
highlighted in Fig. 6 by the red box. An Al particle pictures marked by the white arrow. At the grain
causes the shunt. boundaries a formation of crystallites is observed. These
could be identified by EDX as SixCy crystallites.
Shunts due to scratches
Scratches can damage the cell junction when they go
deeper than the SixNy anti-reflective coating. The shape
of the scratch, which is not clear in the thermal image
from Fig.8, can be recognized at its minimal visualization
voltage.

Fig. 10: Cell with a material induced shunt.

Fig. 8: Thermal image of shunt caused by a scratch.

Scratches visible on the top of the anti-reflective


coating are an indication of handling problems during
screen printing. Process correlation will be more difficult
for scratches performed prior to the anti-reflective
coating. This case can only be possible if, for example,
the shape of a shunt matches the shape of a handling part.

Cracks and holes


Cracks and holes compromise automatically the
performance of a cell. Holes created at metallization
positions during printing allow the metal paste to pass Fig. 11: LBIC Scan of the cell from Fig. 10. The red
trough the wafer. Very deep linear shunts are created this arrow points to a region analysed by optical microscopy.
way. An example of a cell with a shunt caused by a crack
is presented Fig. 9.

Fig. 12: SEM image of the region pointed in Fig. 11. The
white structures are SixCy crystallites.
Fig. 9: Thermal image of shunt due to a crack.
3
The causes of the shunts in a production line where
4 OPTIMIZATION OF EDGE ISOLATION identified. They could be reduced with improvements in
cell handling, reduction of contaminations and process
Plasma edge isolation and laser edge isolation are optimization.
two well accepted edge isolation processes in industry. Shunt control and prevention resulted in a cell
The laser edge isolation process has two main efficiency increase.
advantages in relation to plasma isolation: (1) it permits
the in-line processing and this way avoids handling 6 AKNOWLEDGEMENTS
problems like bad stacking. Bad stacking causes
simultaneously insufficient and excessive emitter removal Thanks to Dr. Stefan Dauwe and Kristin
in cells from the same stack. This conduces to a bigger Neckermann, SZE Erfurt, for their help with the
repeatability, which means that the distribution of shunt preparation and interpretation of the SEM samples.
resistance values, and therefore output of the cell, will
have a small distribution. (2) It avoids the usage of non- 7 REFERENCES
environmentally friendly gases like CF4. Output
differences will always depend on process parameters [1] P. Grunow, S. Lust, D. Sauter, V. Hoffman, C.
such as process energy and the size of the machine, just Beneking, B. Litzenburger, L. Podlowski,
to quote. 19thEuropean Photovoltaic Solar Energy Conference,
The effectiveness of the laser ablation can be mainly 2004, Paris ,pp. 2190-2193.
controlled by pulse energy, scribing distance and on
which side the isolation is performed. Fig. 13 shows the [2] M. C. Alonso Garcia, W. Herrmann, W. Bömer, B.
shunt resistance values for laser isolated cells taking into Proisy, Prog. Photovoltaics: Research and
account distance from the edge and side variation. Appications, Vol. 11, 2003, pp. 293–307.

[3] M. Langenkamp, O. Breitenstein, Solar Energy


Materials & Solar Cells, Vol. 72, 2002, pp. 433-440.

[4] O. Breitenstein, J. P. Rakotoniaina, M.H. Al Rifai,


M. Werner, Progress in Photovoltaics: Research and
Applications, Vol. 12, 2004, pp. 529-538.

[5] O. Breitenstein, M. Langenkamp, O. Lang, A.


Schirrmacher, Solar Energy Materials & Solar Cells,
Vol. 65, 2001, pp. 55-62.

[6] Jan Schmidt, Ingo Dierking, Progress in


Photovoltaics: Research and Applications, Vol. 9,
2001, pp. 263-271.
Fig. 13: Typical parameters for laser edge isolation.
[7] C. Ballif, S. Peters, J. Isenberg, S.Riepe, D. Borchert,
Fig. 14 illustrates a limitation found for laser edge Procedings of the 29th IEEE PVSC (2002); Paper No
isolation. Shunts can occur at the corners of the cell when 4P3.10.
a deep ablation depth (z>25µm) is cancelled by a very
high material re-deposition due to a lower isolation speed [8] www.edmundoptics.com
at the corners of the cell. This problem was eliminated by
adjusting ablation parameters.

Fig. 14: (Left) Thermal image of a shunt occurred during


Laser edge isolation. (Right) SEM picture of the shunted
region at the cell’s corner.

5 CONCLUSION

Shunt analysis using liquid crystal foils enabled the


effective detection of both process and material induced
shunt.

You might also like