The Influence of Micro Cracks in Multi-Crystalline Silicon ... - Q-Cells

THE INFLUENCE OF MICRO CRACKS IN
MULTI-CRYSTALLINE SILICON SOLAR
CELLS ON THE RELIABILITY OF PV
MODULES
P. Grunow 1 , P. Clemens 1 , V. Hoffmann 1 , B. Litzenburger 2 , L. Podlowski 2
1
Q-Cells AG, Guardianstr. 16, D-06766 Thalheim, Germany, phone: +49 3494 6686 -0, fax: -10, email: p.grunow@q-cells.com
2
Solon PV GmbH, Ederstr. 16, D-12059 Berlin, Germany, phone: +49 30 818 79 -100, fax: -110, email: l.podlowski@solon-pv.com; b.litzenburger@solon-pv.com
Motivation
The need to produce more Wp per kg silicon in a period
of poly silicon shortage has lead to rigid thickness
reductions for mono- and multicrystalline silicon wafers.
Micro cracks in wafers and cells are therefore
investigated with special effort [1][2], because they
limit the production yields and they are suspected to
reduce the module reliability.
Wafers with different thickness were laminated on
module cover glass for bending tests and the
propagation of micro cracks was detected by near
infrared scattering (NIRS). In difference to conventional
approaches [3], NIRS detects the reflection at the micro
crack induced internal interfaces in the wafer material.
Experimental: module bending tester and NIRS
Fig. 1:. Module bending tester with adjustable elongation
between 0 and 12 cm. The load was increased in 1 cm
steps and the laminated wafers were monitored in respect
to micro crack occurance by NIRS, i.e. scattered light in
the near infrared range transmitting the wafer.
A conventional digital camera with switched off IR filtering
was used together with a halogene bulb as light source.
Fig. 2: NIR Scattering image of a wafer laminated on glass
showing micro cracks after mechanical loading in the module
bending test. In difference to the conventional transmission
technique, the cracks appear dark in the NIRS image.
The cross in the center was prepared by laser scribing with a
defined groove length (20mm) and depth (120µm) before the
bending test as a defined reference damage.
Mechanical bending tests
Fig 3: The elongation from the module bending tester was used
to calculate the curvature radius using the deflection curve of a
simple beam. The critical radius defined by the first occurence
of micro cracks as propagating from the laser scribed grooves is
compared to the value reached in the mechanical load tests
according to IEC 61215. The thermal cycling tests were
performed in parallel and turned out to be less critical
regarding tensile stress on cells in the considered module
Electrical Measurements
#1 #2
Fig. 6: Q6L multicrystalline single-cell-modules, that were
prepared with different crack positions induced by
scratching the module back sheet with a metal stick
Type crack position
#3
?
#4 #5 #6
#1 parallel on both sides ot both bus bars
#2 parallel to one bus bar outside
#3 corner
#4 position unknown, natural crack from bending
#5 parallel and centered between the bus bars
#6 centered but perpendicular to the bus bars
Table 1: Different crack position as sketched in Fig. 5
change
critical curvature radius/cm
600
500
400
300
200
100
0
220 µm
270 µm
IEC 61215 heavy load test 540 kg/m²
0 200 400 600 800 1000
groove length/ µm
Fig. 4: Module bending tests for multicrystalline wafers
with 280 and 220 µm thickness laminated on glass
with one EVA layer and prepared with laser grooves with
different length but constant depth of 120 µm. The
dashed line visualises the minimum radius in the IEC
61215 heavy load test for a module with
4 mm x 943 mm x 1593 mm glass cover.
0%
-10%
-20%
-30%
-40%
-50%
# 1 # 2 # 3 # 4 # 5 # 6
crack position
Fig. 7: Change of the detailed electrical parameters of the singlecell-modules
before and after applying the cracks as described in
Figure 6 and Table 1
rel. Module Pmax
101%
100%
99%
98%
97%
96%
IEC 61215 heavy load test 540 kg/m²
> 1000 640 430 320 260 210 180 160
module centre curvature radius /cm
Figure 8: The relative power Pmax for 60 cell modules is
decreasing to lower curvature radii (i.e. to higher maximum
elongation in the module center). Repeated cycling tests at a
constant load did not further decrease the module power.
Isc
Voc
FF
standard#1
standard#2
crackle#1
crackle#2
crackle#3
crackle#4
crackle#5
crackle#6
critical curvature radius/cm
700
600
500
400
300
200
100
0
1 2 3 4 5
No. of EVA layers
130 µm
280 µm
IEC 61215 heavy load test 540 kg/m²
Fig. 5: Module bending test for 280 and 130 µm thick
wafers laminated on glass with increasing No. of 0,4mm
thick EVA layers. For the 280µm wafers no micro cracks
could be detected up to the maximum elongation limit of 12
cm for > 2 EVA layers
CONCLUSION
• Micro cracks were sucessfully detected by near
infrared scattering (NIRS)
• The groove length and depth of the micro cracks are
shown to be critical parameters for the reliability of
PV modules according to IEC 61215. For 270µm
wafers the critical length is 1000 µm and for 220µm
it is 300µm as shown for laser scribed grooves with a
depth of 120µm
• The level of reliability of modules with 270µm cells
can be maintained for 130µm thin cells, when using
at least 5 layers of EVA instead of 1 layer.
• The power loss in modules with crackling cells after
mechanical load test are in the order of 1% at loads
according to IEC 61215. The power loss increases to
higher loads but does not depend on the number of
the load cycles.
References
1. A. Schneider, E. Rueland, A. Kraenzl, P. Fath, 19th EPVSEC, 2004, Paris,
France, p.496
2. J. P. Rakotoniaina1, O.Breitenstein1, M. Hejjo Al Rifai1, D. Franke, A.
Schneider, EPVSEC, 2004, Paris, France, p. 640
3. E. Rueland, S. Recht, S. Wansleben, H. Feist, P. Fath, 19th EPVSEC,
2004, Paris, France, p.810

INFLUENCE OF MICRO CRACKS IN MULTI-CRYSTALLINE SILICON SOLAR CELLS ON THE
RELIABILITY OF PV MODULES
P. Grunow 1 , P. Clemens 1 , V. Hoffmann 1 , B. Litzenburger 2 , L. Podlowski 2
1 Q-Cells AG, Guardianstr. 16, D-06766 Thalheim, Germany,
email: p.grunow@q-cells.com
2 Solon PV GmbH, Ederstr. 16, D-12059 Berlin, Germany, email: l.podlowski@solon-pv.com
ABSTRACT: The mechanical strength of multi- and monocrystalline silicon wafers and cells is strongly dependent
on the length and the position of micro cracks in the silicon wafer material. Micro cracks are increasing the breakage
risk over the whole value chain from the wafer to the finished module, because the wafer or cell is exposed to tensile
stress during handling and processing. The actual need to use larger and thinner crystalline silicon wafers increases
the risk of yield losses. Besides the breakage risk in the production, the cells have to withstand the tensile stresses
under outdoor operation in the finished modules. Here the tensile stress is induced by temperature changes and
mechanical loads from wind and snow. Currently, commercial PV modules have 25 years warranty and their
reliability over this time period is simply considered as proven, when a module type was successfully qualified
according to IEC 61215. In industrial practice, there is currently no parameter for the mechanical strength of cells
and wafers available, which guarantees the reliability of an individual module.
Wafers of different thickness were laminated into modules, which were then exposed to mechanical load tests similar
to IEC 61215. The micro crack propagation due to the load test was monitored by scattered light on the micro crack
planes in the near infrared range. Finally, the power drops resulting from mechanical load tests were compared for
standard modules using cells with and without micro cracks.
Keywords: Micro cracks, Multi-Crystalline Silicon, Reliability
1 INTRODUCTION
The need to produce more Wp per kg of silicon in a
period of poly silicon shortage has lead to rigid thickness
reductions for mono- and multicrystalline silicon wafers.
Micro cracks in wafers and cells are investigated with
increased effort [1][2][3] because they limit the
production yields and are suspected to reduce the module
reliability. Due to the laws of fracture mechanics this is
becoming even more critical for thinner wafers [4].
Wafers of different thickness were laminated on
module cover glass for bending tests, and the propagation
of eventual micro cracks was detected by near infrared
scattering (NIRS) after the bending test. In contrast to
conventional approaches [5], NIRS uses the scattering of
near infrared light at the micro cracks planes inside the
wafer.
In addition, the electrical power loss of a single cell
will depend on the position of the crack in the cell as
investigated for single-cell-modules and large standard
modules.
test laminates were bended by a line of force applied to
the centre of the module. The displacement was increased
in 0.5 cm steps from 3 cm to 12 cm with the apparatus
shown in Fig. 1. The distance between the supports
holding the glass was 123 cm. The wafers are exposed to
tensile stress because the wafer is glued to the extreme
axis of the 4 mm glass. The maximum elongation of the
glass was increased in steps of 0.5 cm followed by a
subsequent check for micro crack formation, which was
done by near infrared scattering NIRS.
2.2 Micro crack detection with NIRS
NIRS is using NIR light transmitted through the
wafer and reflected at the internal surfaces due to the
micro cracks extending into the wafer volume. A
conventional digital camera, Sony Digital 8 Handy cam
DCR-TRV140E PAL, with switched-off IR filter was
used together with a halogen bulb as the light source. In
contrast to the conventional through-light transmission
techniques, the cracks appear dark in the NIRS image,
see Fig. 2.
2 MECHANICAL LOAD TESTS
2.1 Experimental
Multicrystalline as cut wafers 156 mm x 156 mm
270 µm were laminated on 4 mm x 823 mm x 1653 mm
glass between 2 sheets of 0.4 mm Ethylene Vinyl Acetate
(EVA) without using a back sheet on the back to allow
optical transmission of NIR light for micro crack
detection. For the same reason, only wafers were used
instead of complete cells, because the full area aluminum
print on the back side blocks the NIR light. The
mechanical stability of as cut wafers can differ to fired
cells [1], but this difference was balanced out by using a
reference damage, which has the same critical effect on
the wafer stability as a spontaneous micro crack. During
lamination the membrane of the laminator was protected
from the EVA with a Teflon-coated fiber glass foil. These
wafers laminated on glass
Figure 1: Module bending tester with adjustable
elongations between 0 and 12 cm. The elongation was
increased in steps of 0.5 cm. The wafers are laminated to
the lower side of the glass without a Tedlar back sheet.

The minimum detectable crack width, or the
minimum detectable distance of two opposing internal
micro crack surfaces, is given by the wavelength of the
used NIR light (1.1 µm). The reflection gets distorted for
distances smaller than the wavelength. The minimum
detectable area of the micro crack planes depends on the
resolution of the digital camera. In this work, a length
down to 100 µm and a depth of 130 µm could be
detected by NIRS on laser scribed grooves.
2.3 Critical crack length
To determine the critical crack length of micro cracks
in regards of the reliability of cells inside a module, laser
grooves of defined depth and length were scribed in the
wafer centre with a pulsed Nd:YAG Laser at 532 nm.
According to expression (1), the critical tensile stress,
which leads to wafer breakage, is inverse proportional to
the square root of the crack length:
1 2
K Ic
= σ ⋅ a ⋅π
= 0,9 MPa m for silicon (1)
with K Ic
= critical stress intensity factor, σ = tensile
stress perpendicular to the crack, a = critical length
of the micro crack.
These laser grooved wafers were used as a model to
study the critical parameter of natural micro cracks, e.g.
cracks which were induced in the wafer cutting and were
not removed in the saw damage etch. The length and
depth of the laser grooves can be used as parameters to
investigate their influence on the mechanical stability of
the wafer. The width, which is much larger for the laser
grooves than for natural micro cracks, is not a critical
parameter for the mechanical stability. The influence of
the laser scribing process itself was not investigated here.
Fig. 2 shows the NIRS image of a 270 µm test wafer
with two crossing laser grooves of 20 mm length and a
depth of 130 µm scribed before the bending test. The
bending radius was set to 350 cm and equals the radius
for a framed module with a 4 mm x 943 mm x 1593 mm
glass cover (P200 from Solon PV GmbH) exposed to the
heavy load test (540 kg/m²) in IEC 61215 [6]. The same
type of module was used in the electrical tests.
between -40°C and +85°C on the same sample showed
no further micro crack propagation either from the
vertical or horizontal laser grooves, although the thermal
expansion of the cover glass will induce tensile stress in
both orientations. I.e., that the tensile stress resulting
from the mechanical heavy load tests were higher than
those resulting from the thermal cycling test required by
IEC 61215.
The two bifurcations in the micro crack propagation
in the upper and lower part of Fig. 2 represent a typical
breakage pattern observed on the samples. A possible
explanation for these deviations from the straight line
propagation could be that local points of higher stability
in the multicrystalline grain structure block the
propagation parallel to the lines of maximum tensile
stress.
2.3 Critical curvature radius
The maximum elongation w
max
of the 4 mm glass
module in the bending test was used to calculate the
curvature radius R using the deflection curve of a simple
beam, see Fig. 3, for the case of small elongations:
2
l
R = −
(2)
12⋅
w
max
with l as the distance between the supports in the
bending tester.
The critical curvature radius was then defined by the
first occurrence of micro cracks while increasing the
maximum elongation in steps of 0.5 cm. For comparison
those critical curvature radii are shown together with the
minimum curvature radius calculated for the heavy load
test of the IEC 61215 on a framed P200 module from
Solon PV GmbH (see the dashed horizontal line in the
following figures). Obviously, its value depends on the
module design, i.e. dimension, framing and the
encapsulation material used. Because the latter
intermediates the tensile stress applied by the cover glass
to the wafer in the bending test, the encapsulant plays a
critical role in the reliability of modules with thinner
cells.
20mm
Figure 2: NIR scattering image of a wafer laminated on
glass showing micro cracks after mechanical loading in
the module bending test. The cross in the center was
prepared before the bending test by laser scribing with a
defined groove length (20mm) and depth (130µm).
The micro crack in Fig. 2 starts from the vertical laser
groove and propagates than parallel to the applied force
of the bending tester. A subsequent thermal cycling test
Figure 3: Deflection curve of a simple beam as used
for the calculation of the module bending in the tester
2.4 Influence of the wafer thickness
The critical crack length was investigated for two
groups of 15 wafers of different thickness, 220 µm and
270 µm. As described above, they were prepared with
laser grooves set to a fixed depth of 130 µm and a groove

length ranging from 100 µm to 1 mm. The critical
curvature radius was calculated from the maximum
elongation and the centre position of the specific wafer in
respect to the line of maximum curvature in the module
centre. Fig 4 shows the critical curvature radius at wafer
breakage vs. different groove length.
critical curvature radius/cm
Figure 4: Module bending tests for multicrystalline
wafers with 280 and 220 µm thickness laminated on glass
with one layer of EVA. The dashed line visualizes the
minimum radius in the IEC 61215 heavy load test for a
framed module using 4 mm x 943 mm x 1593 mm glass
cover.
In summarising Figure 4, the critical crack length for
the considered module design is 1000 µm for the 270 µm
thick wafers and 300 µm for the 220 µm thick wafers.
Micro cracks exceeding these limits will cause cell
breakage during the heavy load test conditions of the
IEC 61215. Although thinner wafers are more flexible,
they break at lower mechanical and thermal loads than
thicker wafers and thereby decrease the reliability of the
module. In consequence the module design should be
reconsidered, when using thinner cells.
2.5 Influence of the encapsulation
Very thin multicrystalline wafers of 130 µm thickness
and 156 mm edge length were processed successfully in
one of the production lines at Q-Cells. These wafers were
compared to 280 µm thick wafers in another bending test
on 4 mm glass using different thicknesses of EVA
between the cover glass and the cells. Only one layer of
0.4 mm EVA and no Tedlar was used on the opposite
side. The occurrence of micro cracks was monitored by
NIRS as a function of the curvature radius, see Fig. 5.
critical curvature radius/cm
600
500
400
300
200
220 µm
100
270 µm
IEC 61215 heavy load test 540 kg/m²
0
0 200 400 600 800 1000
700
600
500
400
300
200
100
0
groove length/ µm
1 2 3 4 5
No. of EVA layers
130 µm
280 µm
IEC 61215 heavy load test 540 kg/m²
Figure 5: Module bending test for 280 and 130 µm thick
wafers laminated on glass with increasing No. of 0.4mm
thick EVA layers. For the 280µm wafers no micro cracks
could be detected up to the maximum elongation limit of
12 cm for > 2 EVA layers.
Although these wafers were prepared with laser
grooves using a reduced groove depth of 60µm, the
breakage was dominated by natural micro cracks as was
concluded from the fact that the observed breakage
patterns in the NIRS images did not originate from the
laser scribed grooves.
According to Fig. 5 the module design for 130 µm
cells needs at least 3 layers of EVA to withstand the
heavy load test from IEC 61215 for the P200 module. At
least 5 layers of EVA are necessary to achieve level of
mechanical strength as for 280 µm wafers. The increased
thickness of the EVA layer adjacent to the cover glass
reduces the tensile stress on the wafers effectively. An
alternative to the increased number of EVA layers would
be the use of encapsulation material with adjusted elastic
properties.
3 ELECTRICAL TESTS
3.1 Experimental
To investigate the influence of the position of the
cracks on the electrical parameters of the individual cells,
6 different crack patterns were applied to single-cellmodules
by simply cracking the cell through the back
sheet with a metal pen, see Fig. 6 and table 1.
#1 #2
#3
?
#4 #5 #6
Figure 6: Q6L 156mm multicrystalline single-cellmodules
prepared with different crack positions induced
by scratching the module back sheet with a metal stick
Type crack position
#1 parallel on both sides of the both bus bars
#2 parallel to one bus bar outside
#3 corner crack
#4 position unknown, natural crack from bending
#5 parallel and centred between the bus bars
#6 centred but perpendicular to the bus bars
Table 1: Different crack position as sketched in Fig. 6
Only sample #4 was treated differently. This crack
was induced by bending the 3mm x 200 mm x 200 mm
cover glass to a maximum elongation of 2 mm, i.e. the
precise position of the crack is not known, but will be
concluded from the results of the electrical
measurements. The change in the detailed parameters for
each pattern is reported in Fig. 7.
3.2 Results
Sample #1 with an overall power drop of 60% is
giving the expected worst case situation, while the

eakage along the center lines for the samples #5 and #6
result in nearly no power drop.
change
Figure 7: Change of the electrical parameters of the
single-cell-modules due to cracks of different position
and extension as described in Figure 6 and Table 1
From the relatively low power drop 1000 640 430 320 260 210 180 160
module centre curvature radius /cm
standard#1
standard#2
crackle#1
crackle#2
crackle#3
crackle#4
crackle#5
crackle#6
Figure 8: The relative power Pmax for 60 cell modules
decreasing to lower curvature radii (i.e. to higher
maximum elongation in the module centre). Repeated
cycling tests at a constant load did not further decrease
the module power.
In Fig. 8 the approach of sample #4 was extended to
un-framed standard P200 laminates with sixty 270µm
cells connected in series. In addition, a second group of
P200 modules was tested using only cells with micro
cracks, which were detected by the means of a simple
acoustic test called a “crackling” or “twist” test.
The modules with standard cells show better stability
than modules using cells with micro cracks for small
loads or large curvature radii, see Fig. 8. The cracked cell
Isc
Voc
FF
modules decreased by 1% in power for mechanical loads
equal to the heavy load test in IEC 61215. Both groups
were the same for maximum loads (12 cm displacement)
and has a power drop of around 2%. Repeating the load
test at a fixed maximum elongation does not lead to
further decrease of the power, i.e. it is independent of the
number of cycles. This gives good confidence in the
reliability even for modules with cracked cells. In
conclusion, it is reasonable to perform heavy load tests as
final test in module production to exclude spontaneous
power drops of the module. For the module type
considered here the mechanical load test was found to be
more critical regarding the tensile stress than the thermal
cycling test. This should not be generalised and needs to
be tested again for each individual module design.
4 CONCLUSIONS
Micro cracks were successfully detected by near infrared
scattering (NIRS).
The groove length and depth of the micro cracks were
shown to be critical parameters for cell breakage in PV
modules under mechanical load. For 270 µm wafers the
critical crack length is 1000 µm and for 220 µm it is
300 µm, as shown for laser scribed grooves with a depth
of 130 µm
The level of reliability of modules with 270 µm cells can
be maintained for 130 µm thin cells by using at least 5
layers of EVA instead of 1 layer.
The power loss after mechanical load test on modules
with cracked cells is in the order of 1% for loads in
accordance with the criteria of IEC 61215. The power
loss increases at higher loads but does not depend on the
number of the mechanical load cycles.
4 ACKNOWLEDGEMENTS
We would like to thank Toralf Patzlaff from Q-Cells AG
for the laser groove scribing, Nick Sommer from Solon-
PV GmbH for the module preparation and Thomas Spiess
from Q-Cells AG for the special treatment of the singlecell-modules.
5 REFERENCES
[1] A. Schneider, Proc. of the Conf. on PV in Europe
from Technology to Energy Solution, Rome, I (2002)
p. 347.
[2] P. Rakotoniaina1, O. Breitenstein, M. Hejjo Al
Rifai1, D. Franke, A. Schneider, EPVSEC (2004)
Paris, France, p. 640.
[3] D. Franke, Proceedings of the 17th EC PVSEC,
Munich, D, (2001) p. 1830-1833
[4] Dissertation D. Kray, University of Konstanz,
Germany (2004).
[5] E. Rueland, S. Recht, S. Wansleben, H. Feist, P.
Fath, 19th EPVSEC (2004) Paris, France, p.810.
[6] IEC 61215 Ed.2: Crystalline silicon terrestrial
photovoltaic (PV) modules – Design qualification
and type approval, 10.16.3