Available online at www.sciencedirect.

com

ScienceDirect
Energy Procedia 92 (2016) 236 – 241

6th International Conference on Silicon Photovoltaics, SiliconPV 2016

Influence of silver-aluminium alloy at solar cell rear side on series

resistance and open circuit voltage
Tobias Urbana, Ansgar Metteb and Johannes Heitmanna
a
Technische Universität Bergakademie Freiberg, Institute of Applied Physics, Leipziger Straße 23, 09599 Freiberg
b
Hanwha Q Cells GmbH, Sonnenallee 17-21, 06766 Bitterfeld-Wolfen

Abstract

The rear side metallization of industrial PERC solar cells was investigated regarding the implication on the silver-aluminium-
alloy formation, in particular the effect of the high resistivity alloy formation and the etching of surface passivation layer. By
changing the dimensions of the printed silver-pads the size of the silver-aluminium-overlap region at the pad-aluminium interface
were varied. The influence on fill factor and open circuit voltage are evaluated. For a three busbar cell design with six pads per
busbar, the fill factor decreases up to 2.2 %abs, if the overlap length between aluminium and silver pad is reduced from 12.5 mm
to 1 mm, whereas the open circuit voltage increases by 2.5 mV. These results are additionally confirmed by a 3D simulation
which validates the measurement results and shows perspectives for reducing the series resistance effect of the silver-aluminium
overlap region up to 40% by increasing the busbar count from 3 to 5.
© 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
© 2016 The Authors. Published by Elsevier Ltd.
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer review by the scientific conference committee of SiliconPV 2016 under responsibility of PSE AG.
Peer review by the scientific conference committee of SiliconPV 2016 under responsibility of PSE AG.
Keywords: AgAl alloy; overlap resistance; AgAl spiking; COMSOL simulation, PERC

1. Introduction

In solar cell manufacturing a change from BSF type solar cells to PERC like structures is visible. This change of
cell concept has a significant impact on the rear side metallization. One point is the formation of a silver-aluminum
(AgAl) alloy at the interface between rear side metallization and soldering pad. In contrast to BSF cells, all the
current flows via the rear side metallization to the pad and soldering interconnection [1]. The resistivity of the silver-
aluminum-alloy is up to 20 times higher than the resistivity of pure aluminum [2] and therefore relevant in regards
to the series resistance (rs) of the solar cell. Additionally, the formation of silver-aluminum spikes is well known [3]
and therefore this interface was investigated concerning the influence on open circuit voltage (Voc) due to
passivation layer damaging.

1876-6102 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer review by the scientific conference committee of SiliconPV 2016 under responsibility of PSE AG.
doi:10.1016/j.egypro.2016.07.065
 Tobias Urban et al. / Energy Procedia 92 (2016) 236 – 241 237

2. Methods

The experimental setup contains 5 different batches were as only the rear side metallization was changed.
Therefore 3 busbar cz-Si-PERC solar cells with overall 18 soldering pads were used. The PERC process includes the
local contact opening via laser ablation and screen printing of silver (Ag) and aluminum (Al) metallization. The rear
side layout of these Ag pads was changed in 4 steps from 1 mm overlap length (OL) to 12.5 mm, therefore the only
contact between Ag and Al metallization was changed in respect to the cross section of the AgAl alloy, see Fig. 1.
The pads themselves (pad length) stays constant. As reference a 5th batch with full Al rear side metallization was
used. After printing the rear side metallization the wafers were randomized to reduce the influence of the front side
screen printing and were fired in an inline fast firing belt furnace which leads to the formation of the AgAl alloy at
the overlap between Ag and Al, see Fig. 1.

Overlap length Aluminum

Silver-Pad
AgAl Alloy

Pad length
(a) (b) (c)

Fig. 1. a) Screen printed Ag metallization pad (1st print), b) after 2nd print of aluminum metallization the overlap of both metal layers at the
long side of the pad and c) after sintering in the fast firing furnace the AgAl alloy gets visible.

The IV measurement was performed with 2 different, calibrated systems. The first one was a Berger inline flasher
with spring loaded contact pins at front and rear side of the solar cell. The number of pins contacting the soldering
spots were 12 and 6 per busbar at front and rear side respectively. This setups enables a measurement which is very
equal to the effective wiring within the module where only soldering leads to a reliable electrical contact [1]. The
second setup was a sun simulator with spring loaded contacts at the front (equal to the inline flasher) and a
conductive chuck for a homogenous, full area contact at the rear side.

Al
Ag
Passivation
Si

(a) (b)

Fig. 2. a) Plan view of symmetry element with schematic current flow to the ground contact at the silver pad and b) cross section of symmetry
element with homogenous current generation within the silicon.

Additionally the sheet resistance of Ag (Rsh-Ag) and Al (Rsh-Al) were determined with classical four-point probe
measurement [4] and the AgAl resistance with special IV setup. Therefore an IV prober was used to execute a
238 Tobias Urban et al. / Energy Procedia 92 (2016) 236 – 241

resistance scan perpendicular to the overlap. The derivation of the resistance vs. distance leads to the sheet
resistance which was averaged over the AgAl alloy area to get the effective sheet resistance (Rsh-AgAl).
Electrical resistance simulation was performed with COMSOL Multiphysics [5] with the measured sheet
resistance and the geometry of the used printing screens. For simplification of simulation the solar cell was reduced
to their smallest symmetry unit, which contains a half pad with a surrounding aluminum rectangle of 2.6 x 2.6 cm
edge length. The light injection and therefore the current flow is assumed to be homogeneous over the cell and
perpendicular to the aluminum metallization which is conducted via the silver pad to ground, see Fig, 2 a).
Therefore the setup is equal to the contacting mechanism of the IV measurement with contact pins only at the silver
pads (inline flasher), Fig. 2 b).

3. Results

3.1 Experimental results

The impact of the AgAl alloy formation on series resistance as well as on open circuit voltage (Voc) was
experimentally determined by varying the rear side metallization layout, as described before. The IV measurement
reveals great differences depending on the rear side contacting scheme. The measurement at the suns simulator
shows very equal fill factor (FF), within the range of the 95% confidence interval. In contrast the pointwise
connection at the flasher measurement shows a significant trend in FF depending on the overlap length, see Fig. 3.
Between 12.5 mm and 1 mm overlap length the FF is decreased by 2.2%abs. Whereas below 10 mm overlap length a
decrease in FF become visible, by comparing with the full aluminum rear side contacted at solder spots (red circles,
Fig. 3.).

79

78
FF [%]

77 rear contacting:
full contact
only at solder pads
76
Simulation
0 2 4 6 8 10 12 14
Overlap Length [mm]
Fig. 3 IV measurement results performed with full area rear contact (sun simulator) and pointwise contacts at soldering spot (inline flasher) for
all different overlap layouts compared with full alumni rear side

On the other hand the AgAl alloy produces spikes which penetrates the rear side passivation and therefore has an
impact on open circuit voltage (Voc). The spiking leads to inverted pyramids, known from contacting Boron emitters
with aluminum containing silver pastes [3]. The imprints of these pyramids are well visible after chemical removing
of the metallization. Edge length of up to 20 μm thus lead to • 10 μm deep spikes. Therefore a reduced overlap area
 Tobias Urban et al. / Energy Procedia 92 (2016) 236 – 241 239

is improving Voc. By changing the OL from 12.5 mm to 1 mm the Voc was reduced up to 2.5 mV. In comparison with
the reference, the pure silver of the pad reduces the Voc by approx. 1 mV, which can be deduced from Fig. 4.
These contrary trends in Voc and FF lead to a tradeoff situation where the influence of the FF is stronger and
therefore the efficiency optimum under the assumption of identical short current density lies in the region of around
10 mm overlap length with a very smooth plateau for overlap length > 10 mm.

659 full aluminium rear side

658
impact of pure silver 20 μm
Voc [mV]

657 R² = 0.83
m = 657.67 mV
n = - 0.2 mV/mmOL
656

655 Experiment
linear fit
0 2 4 6 8 10 12 14
overlap length [mm]

(a) (b)

Fig. 4. a) IV-measurement results of cells with different overlap length, Voc vs. OL with clear visible effect of spiking at the AgAl interface
and b) plan view of AgAl spiking imprints in overlap region after etching of the metallization layer.

3.2. Simulation

Rsh-Al = 15 m:/ Rsh-AgAl = 130 m:/

unit cell: 2.6 x 2.6 cm²
0.4 2.1
rs-AgAl high
rs-Al
calc. FF loss [abs%]

0.3 1.6
rs-loss [:cm²]

0.2 1.0
I [a.u.]

0.1 0.5

0.0 0.0
1 2 3 4 5 6 7 8 9 10 11 12 13 low
overlap length [mm]

(a) (b)

Fig. 5. a) Simulation results for the variation of the overlap length, analogous to the experiment and b) the current distribution for 1 mm (up)
and 12.5 mm (bottom) overlap length with corresponding schematic model.
240 Tobias Urban et al. / Energy Procedia 92 (2016) 236 – 241

The 3D FEM Simulation Software COMSOL Multiphysics was used to reproduce the experimental results. The
corresponding simulation model predicts an identical FF trend which enables a detailed investigation of the root
cause of the resistance loss, see Fig. 3. The change of the overlap length influences the cross section of the AgAl
alloy and additionally the current density in the surrounding aluminium layer, see Fig. 5 b). Accordingly the change
in rs is due to these two effects. The increase of the current density produces a higher loss in the Al layer, following
the classical Ohm’s law where the dissipated power P increases by the square of the current I (P = R/I²). The high
current density in combination with the high sheet resistance of the AgAl alloy results in the second rs contribution.
According to these results there are different possible optimization roots. One is the adjustment of the soldering
spot count, which is not very consistent, thereby the area of the unit cell and therefore the current density per overlap
is reduced. Although this is not the ideal solution because the Ag consumption increases without any further benefits
regarding the other solar cell resistance contributions. Therefore an increase of the busbar count is much favourable,
thereby the whole solar cell benefits from the reduced front side metallization losses [6] in addition to the improved
rear side. The simulation of 4 BB and 5 BB cells with the same busbar design (pad count and dimensions) shows a
decreasing rs-influence with overlap dimension, see Fig. 6. a). By changing the layout from 3 to 5 busbars, the rs
effect is reduced by 40%, in which the effect of the Al-layer is with -45% slightly higher then the effect of the
overlap (-35%), see Fig. 6. b).

jsc = 36.2 mA/cm², overlap length = 6 mm

0.3
3BB 3BB
0.03
4BB 5BB
5BB
- 45 %
'FF [abs%]

0.2
'rs [:cm²]

0.02 -35 %

0.1
0.01

0.0 0.00
0 2 4 6 8 10 12 14 Alu AgAl
overlap length [mm]
(a) (b)

Fig. 6. a) Comparison of 3 to 5 BB cell design with respect to the overlap length influence and b) detailed look for 6 mm overlap length, break
down of rs components.

4. Conclusion

The AgAl alloy which occurs at the overlap between Silver and Aluminum of common PERC rear side
metallization was investigated in regard to series resistance and open circuit voltage. The overlap was varied
between 1 mm up to 12.5 mm for a 3 busbar cell with 6 soldering pads each. The series effect gets visible only if the
solar cells are measured with a realistic setup, were only the pads are electrical connected. A strong decrease of FF
with overlap length is measureable. An overlap length of 1 mm reduces the FF up to 2.2%abs compared to 12.5 mm
or full aluminum rear side. A negative influence of overlap region gets visible if the overlap length drops below 10
mm, compared to full aluminum rear side. In contrast to the series resistance effect the open circuit voltage increases
with decreasing overlap length due to reduced AgAl spiking as effect of reduced area. However, the FF effect is
much stronger than the Voc effect and therefore the efficiency optimum lies in the range of 10 mm overlap length.
 Tobias Urban et al. / Energy Procedia 92 (2016) 236 – 241 241

COMSOL Multiphysics was used to simulate the metallization rear side which predicts the trends in series
resistance effect very well and therefore enables the path to optimize the metallization. Thereby the increasing of the
busbar count seems to be most efficient because additionally to the rear side the contribution of the front side
metallization resistance is improved. The increase of busbar count from 3 to 5 reduces the rs of the rear side
metallization by 40% due to reduction of current density per pad.

Acknowledgements

This work was funded by the German Federal Ministry for Economic Affairs under the acronym “AdmMo” with
identification code 0325775F.

References

[1] T. Urban, M. Heimann, A. Schmid, A. Mette. Analysis of Ohmic Losses Due to Solder and Pressure Interconnection and Related Interface
Resistances for Solar Cells. Energy Procedia 77; 2015. p. 420-427.
[2] J.J. Van den Broek, A.G. Dirks, P.E. Wierenga. The composition dependence of internal stress, ultramicrohardness and electrical resistivity of
binary alloy films containing silver, aluminium, gold, cobalt, copper, iron or nickel. Thin Solid Films 130; 1985. p. 95-101.
[3] S. Fritz, K. Markus, S. Riegel, A. Herguth, H. Matthias. Formation of Ag/Al Screen-Printing Contacts on B Emitters. IEEE J. Photovoltaics 5;
2015. p. 145-151.
[4] L. B. Valdes. Resistivity Measurements on Germanium for Transistors. Proc. I.R.E. Vol 40; 1954. p. 420-427.
[5] Comsol Multiphysics® 5.1 version 5.1.0.180. Copyright 1998-2015.
[6] S. Braun, et. al. The multi-busbar design: an overview. Energy Procedia 43; 2013. p. 86-92.