See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/265903728

Passivated Busbars from Screen-printed Low-temperature Copper Paste

Article in Energy Procedia · September 2014

DOI: 10.1016/j.egypro.2014.08.052

CITATIONS READS

25 1,574

16 authors, including:

Don Wood I. Kuzma

Hemlock Semiconductor imec
19 PUBLICATIONS 219 CITATIONS 50 PUBLICATIONS 624 CITATIONS

SEE PROFILE SEE PROFILE

Richard Russell Filip Duerinckx

imec imec
48 PUBLICATIONS 614 CITATIONS 91 PUBLICATIONS 1,411 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by N.E. Powell on 30 September 2014.

The user has requested enhancement of the downloaded file.

 Available online at www.sciencedirect.com

ScienceDirect
Energy Procedia 55 (2014) 724 – 732

4th International Conference on Silicon Photovoltaics, SiliconPV 2014

Passivated busbars from screen-printed low-temperature

copper paste
Don Wooda, Izabela Kuzma-Filipekb, Richard Russellb, Filip Duerinckxb, Nick Powellc,
Adriana Zambovac, Brian Chisleac, Pierre Chevaliera, Caroline Boulorda,
Alexandre Beuchera, Nicolas Zeghersa, WeiWei Dengd, Zhiqiang Fengd,
Pierre Verlindend, Jozef Szlufcikb, Guy Beaucarnea
a
Dow Corning Europe SA, Seneffe 7180, Belgium.
b
Imec, Kapeldreef 75, Leuven 3001, Belgium
c
Dow Corning, 2200 West Salzburg Road, Midland, MI 48686, USA
d
State Key Laboratory of PV Science and Technology, Trina Solar, 2 Trina Road, Trina PV Industrial Park, Jiangsu Province, China 213031

Abstract

A screen-printable copper paste has been developed by Dow Corning to replace the standard screen-printable silver paste for use
in front busbars for solar cells. Solar cells produced with these ‘passivated copper busbars’ have shown increased conversion
efficiency due to an improved device operating voltage and current while maintaining a similar fill factor when compared to cells
with standard ‘fired-through’ silver busbars. In addition to the improved cell efficiency, the use of copper paste may provide
major cost savings compared to the use of silver, potentially giving a very significant reduction in cost per Watt. 60-cell modules
have been produced at Trina Solar with passivated copper busbars, showing similar performance to the reference modules with
silver busbars. Module reliability has been shown to be well in excess of the IEC 61215 standard requirements.
© 2014 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
© 2014 The Authors. Published by Elsevier Ltd.
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Peer-review underresponsibility
Peer-review under responsibilityof of
thethe scientific
scientific committee
committee of theofSiliconPV
the SiliconPV 2014 conference.
2014 conference

Keywords: Photovoltaics; metallization; copper; screen-printing;

1876-6102 © 2014 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Peer-review under responsibility of the scientific committee of the SiliconPV 2014 conference
doi:10.1016/j.egypro.2014.08.052
 Don Wood et al. / Energy Procedia 55 (2014) 724 – 732 725

1. Introduction

1.1. Motivation

Screen printing of silver-based fired-through front contacts remains the dominant metallization process for solar
cells due to its simplicity, low cost and high throughput. While module prices have seen a long-term 20% reduction
in cost per doubling of manufactured volume, the cost of the silver metallization process per Watt peak is currently
5% higher than it was in 2008 [1]; this process now constitutes the greatest part of the cost of the module, after the
silicon. Bulk copper has a similar conductivity to silver (1.7 µΩ-cm for copper, 1.6 µΩ-cm for silver), but costs
~100 times less (~5 Euro per kg), making it an excellent potential replacement.

Dow Corning has developed a screen-printable paste [2] that uses copper as the main conductive component in
place of silver. This paste is not conductive enough to form the entire front grid on a solar cell; the conductivity
requirement for the fingers is far higher than the requirement for the busbars. With a typical 3-busbar solar cell
design, the fingers must carry the current for up to 2.5 cm to reach a busbar. Although the current density in the
busbar is then higher, the busbar only needs to carry the current through it’s thickness, typically ~20 microns, which
is almost 1,000 times less distance than in the fingers, before being collected by the thick copper tabbing (assuming
continuous soldering onto the busbar). The main requirement for the busbars then, is to have enough conductivity to
carry the current over a short distance without adding significantly to the overall cell series resistance, and to
provide a solder-able surface onto which the tabbing can be soldered.

Therefore, we have used this paste to print ‘passivated copper busbars’ on top of standard screen-printed and
fired-through silver fingers [3] with aluminium ‘back surface field’(Al-BSF), using a dual print process. In addition
to the low cost, this scheme avoids direct metal-semiconductor contact, reducing charge carrier recombination under
the busbars. This has been shown to increase the cell operating voltage in most cases, giving improved cell
efficiency.

1.2. Low temperature screen-printable copper paste

The copper paste consists primarily of copper and solder particles held in a polymer matrix (Fig. 1), as detailed in
[2]. The paste has been engineered to be highly thixotropic, having a high viscosity at rest, but low viscosity when
under stress during the screen printing process. Immediately after printing, the paste has a brown-red colour due to
the presence of the copper. However, after ‘curing’ at ~250 °C in an oxygen-free atmosphere, the colour of the
paste changes to grey due to the fluxing of the copper and subsequent coating by the solder. Our concept of a low
temperature screen-printable copper paste with the copper encapsulated by low melting point alloys has also
subsequently been demonstrated by other groups [4], as has the use of a low temperature paste to form busbars in a
dual print process [3, 5].

 

 

 
 
  



     

Fig. 1. Dow Corning's screen printable copper paste. After printing (left) and after curing (right).
726 Don Wood et al. / Energy Procedia 55 (2014) 724 – 732

2. Experimental

Initial ‘proof of concept’ cells were produced following the general process outlined in Fig. 2. This was a dual
print process [6], with the copper busbars printed and cured after the printing and firing of the silver fingers and
aluminium BSF. 20 cells were produced using Dow Corning’s copper paste for the busbars, and 15 controls used
standard fired-through silver busbars.

Following the successful proof of concept trial, a larger number of more advanced industrial cells were
processed. These cells were given a selectively doped emitter, so that the doping under the metalized regions was
higher than that between the fingers. Approximately 300 cells were subjected to the ‘dual print’ process with fired
through silver fingers and copper paste busbars, and approximately 200 were processed through the standard single
print process with full silver front side metallization. After printing of the copper busbars, curing was performed in
a belt furnace with the oxygen content below 10 ppm. Soldering of the cells into strings for module fabrication was
performed using copper ribbon coated with Sn60Pb40 solder. The ribbon was 1.5 mm wide and 0.25 mm thick. All
printed busbars, both copper and silver, were nominally 1.5 mm wide.

Fig. 2. Cell process flows and resulting structures. Above left: standard silver fired-through process. Above right: modified process for
passivated copper busbars. Bottom: Increased recombination is expected at the metal-semiconductor interface (indicated in red). This
recombination active area is reduced in the case of the ‘passivated busbar’ design (right), which is expected to give improved cell performance.

3. Results

3.1 Busbar resistivity

In preliminary trials, the copper paste formulation was intentionally adjusted to provide a resistivity from ~30
µOhm-cm (corresponding to our standard paste) up to ~150 µOhm-cm. These copper pastes were used to form the
busbars on standard ‘fired-through’ Al-BSF solar cells with silver fingers. Before tabbing, no reduction in cell fill
factor could be observed up to a resistivity of 60 µOhm-cm, demonstrating the ‘over engineering’ of the current
 Don Wood et al. / Energy Procedia 55 (2014) 724 – 732 727

front grid, which uses a fired-through silver busbar with resistivity of ~3 µOhm-cm. After soldering standard
copper tabs onto the copper busbars, no difference in cell fill factor could be observed, even with a busbar resistivity
up to 150 µOhm-cm. Both before and after tabbing, the IV measurement was made using probes spaced ~1 cm
along the busbar.

3.2 Proof of concept: Al BSF

The IV results of the ‘proof of concept’ Al-BSF cells produced at Imec are summarized in Fig. 3 and Fig. 4. The
most significant feature of these results is the 2.5 mV improvement in median open circuit voltage (Voc) for the
cells with the passivated copper busbars. This is believed to be related to the reduced recombination under the
passivated copper busbars compared to that at metal-semiconductor interface under the standard silver fired-through
busbars. This reduced front surface recombination is also believed to be the cause of the small (~1%) but
statistically significant improvement in the device current. No statistically significant difference in the cell series
resistance or fill factor could be detected. These device parameters lead to the 0.2% absolute improvement in cell
efficiency that we observed. Following these promising results, a similar comparison on high-efficiency PERC-type
cells will be published at a later time.

626 36.0
Jsc (mA/cm2)
624.3 35.5
624
Voc (mV)

35.0 35.0
622 621.8 34.7
34.5

620
34.0
Ag Cu Ag Cu

0.8
Rs (Ohm-cm2)

79.5
79.2
0.65
FF (%)

79.0 79.1
0.6 0.60

78.5
0.4
78.0
Ag Cu Ag Cu
Busbar Busbar

Fig. 3. Proof of concept cells, comparing cells with silver and copper busbars. All cells are screen printed with Al-BSF. The box plots show the
upper and lower quartiles; each symbol represents an individual cell; the values given are the medians for each group.
728 Don Wood et al. / Energy Procedia 55 (2014) 724 – 732

18.0

Efficiency (%)
17.5
17.3
17.1
17.0

16.5
Ag Cu
Busbar

Fig. 4. Cell efficiency for the Imec proof of concept cells, comparing the silver and copper busbars. The box plots show the upper and lower
quartiles; each symbol represents an individual cell; the values given are the medians for each group. A t-test demonstrated that the difference
between the two groups is significant at the 5% level.

3.3 Cells and modules produced in an advanced industrial line

A higher volume trial was conducted with the two aims of assessing the cell efficiency with more statistical
confidence and fabricating full sized modules. The bulk resistivity of the copper busbar was measured by 4-point
probe to be 29 µΩ-cm, which was consistent with previously measured values for this material. For the silver fired-
through busbars, the bulk resistivity was measured to be 3.5 µΩ-cm. The cell level IV performance is summarized
in Table 1 and Fig. 5. For cells with copper busbars, the cell Jsc and shunt resistance are all improved relative to the
cells with silver busbars. However, the series resistance was higher (i.e. poorer) at cell level, leading to lower fill
factor and cell efficiency. This reduction in fill factor is essentially a measurement issue – for measurement systems
with a wider probe spacing, the lateral conduction of the busbar contributes more to the overall cell series resistance;
at module level, after tabbing, the lateral resistivity of the busbar does not impact the fill factor, since lateral current
flow is through the highly conductive soldered tabs. For the current trial, the module level series resistance is the
same for the silver and copper busbar (Table 2). The improvement in Voc was modest, at just 0.3 mV, although it
was statistically significant, with a t-test giving the 95% confidence interval ranging from an improvement of 0.1 to
0.5 mV. However, measurements of Joe,metal, which will be reported elsewhere, indicated an improvement of 2 to 3
mV could be expected. T-tests revealed that the differences in Jsc, FF and Rs were also statistically significant at
the 5% level, although no significant difference in shunt resistance was detected.

After tabbing, there is only a minor difference in module performance between the two modules with copper
busbars and the two with silver busbars. The average maximum power is 270.8 and 271.1 Watts respectively. This
is equivalent to a difference in cell efficiency of just 0.02% absolute. Although, as discussed above, the series
resistance is similar at module level in both cases, the slope of the IV curve at low current indicates a shunt
resistance for the module with copper busbars of about half of that for the reference case, leading to a slightly lower
fill factor (0.4% absolute). This may be related to sub-optimal current matching between cells, due to the limited
number of cells available from which to select for the two modules. This problem would obviously be eliminated at
high volume manufacturing. As a qualitative measure of the reproducibility of the module performance, Fig. 6
shows that differences between the IV curves for the four modules are not easily resolvable.
 Don Wood et al. / Energy Procedia 55 (2014) 724 – 732 729

Table 1. Cell parameters from the advanced industrial process, before tabbing.

Isc Voc FF Rs Rsh Efficiency

(A) (mV) (%) (mOhm) (Ohm) (%)
Ave 8.89 640.4 78.9 3.1 852 18.80
Cu BB
Max 8.92 641.0 79.1 3.5 643 18.92
(297 cells)
Stdev 0.02 0.8 0.2 0.1 312 0.03

Ave 8.87 640.1 79.4 2.6 834 18.87

Ag BB
Max 8.87 641.0 79.6 2.9 1007 18.95
(189 cells)
Stdev 0.01 0.5 0.1 0.1 296 0.02

Cell Performance
Before tabbing
19.0
Efficiency (%)

0.6425
18.87
Voc (Volts)
18.8 18.80 0.6403
N=189 N=287 0.6400 0.6402
18.6

18.4 0.6375

18.2 0.6350
37.4 80
Jsc (mA/cm2)

Fill Factor (%)

79.4
37.2 37.2 79 78.9
37.1
37.0
78
36.8
77
Ag cu Ag cu
Busbar Busbar

Fig. 5. Cell parameters from the advanced industrial process.

Table 2. Average module level performance for the advanced industrial process.









!
"#
"(#
"#
"#
" #



 
 
 





 



 

730 Don Wood et al. / Energy Procedia 55 (2014) 724 – 732

    



 

 


 
 

 

 
 


 
 
 
 


 
 

Fig. 6. IV curves from the four modules of the advanced industrial process. The maximum power points are shown with a large symbol, close to
the ‘knee’ of the curves. Differences between the four curves are not easily resolvable.

3.4 Thermal stress testing and damp heat exposure of Al-BSF cells

Unencapsulated Al-BSF cells from the ‘proof of concept’ trial have been subjected to ‘stress testing’ at 300 °C in
a vacuum oven for up to 600 hours. The cell pseudo fill factor (pFF) was determined from Suns-Voc after various
time intervals. The cell pFF gives a measure of the junction quality – the contribution of carrier recombination
processes to the fill factor, and is not affected by changes in series resistance. Therefore, any diffusion of copper
through the silicon nitride and into the silicon would be immediately detected by a drop in the cell pFF. No
difference in change of pFF was observed between the silver and copper busbars (Fig. 7), and none of the cells
declined by more than 1% relative.

Other cells were used to fabricate single-cell encapsulates before undergoing ‘damp-heat’ testing, according to
the IEC 61215 standard, which involves holding the encapsulated cells at 85 °C with 85 % relative humidity. The
cells were tested on an IV ‘flasher’ with electrical connection at the tab ends with the backsheet masked to prevent
rear reflection. As shown in Fig. 7, after 1,000 hours of damp heat (i.e. 1 x IEC), no degradation could be observed.
After 3,000 hours (3 x IEC), limited degradation was observed, with the cells with passivated copper busbars
performing better than those with standard fired-through silver busbars. However, in either case, the degradation
was well within the 5% allowed by the IEC 61215 standard.

Fig. 7. Al-BSF cells, with copper busbars or standard full silver metallization. Left: no significant change in pseudo fill factor was observed after
exposure to 300 °C in a vacuum oven. Right: Minor degradation after 3,000 hours of damp heat for single cell encapsulates.
 Don Wood et al. / Energy Procedia 55 (2014) 724 – 732 731

3.5 Reliability results from the advanced industrial process

The full-sized industrial modules have been subjected to thermal cycling (‘TC’) and to damp heat (‘DH’), in an
attempt to verify their expected longevity in the field, with the tests conducted according to the International
Standard IEC 61215 [7]. For the damp heat test, the module is simply kept at 85 degrees centigrade at 85% relative
humidity for a fixed number of hours. The IEC specification is a degradation of less than 5% in maximum power
for a module subjected to 200 thermal cycles or 1000 hours of damp heat. The modules were subjected to light
induced degradation before either the damp heat or the thermal cycling tests.

The maximum power of the module subjected to thermal cycling degraded by only 0.5% (Table 3), which is well
within the IEC requirement. This degradation was related to a reduced fill factor due to an increased series
resistance. A small improvement was observed for both Jsc and Voc after thermal cycling, although the same trend
was observed for the silver busbar reference cells.

After 1000 hours under damp heat, the module with copper busbars degraded by 1.2%, compared to 1.8% for the
module with silver busbars (Table 3). This degradation was despite a significant improvement in the module fill
factor of 1.3% relative for the copper module, related to a similar reduction in the module series resistance. This
improvement is also accompanied by improved cell-to-cell uniformity of the electroluminescence. However, the
overall degradation is related to a reduction of ~3% relative in the device current. The mechanism of that
degradation is not known. However, given that the series resistance didn’t increase, it seems unlikely to be related
to the copper busbars. Changes to the anti-reflection coating or yellowing of the encapsulant are possible causes.

Table 3. Relative changes in cell performance due to damp heat and thermal cycling. The performance after damp heat is that after 1000 hours at
85 °C and 85% humidity. ‘TC 200’ refers to 200 cycles between -40 °C and +85 °C.

(














 








 






 


%$$






 

4. Conclusions

We have demonstrated the use of a screen-printable copper paste as a feasible, low cost alternative to silver paste
for use in the front busbars of solar cells. The bulk resistivity of the paste is ~30 µOhm-cm, which is at least two
times better than required. For the Al-BSF cells discussed in the current paper, we have achieved an average
improvement in Voc of ~3 mV due to reduced recombination activity under the ‘passivated’ busbars. We believe
that these passivated copper busbars will give even greater benefit in combination with higher efficiency cell designs
with, for example, low front doping and plated contacts. Results for such designs will be reported elsewhere.

60-cell modules have been produced at Trina Solar with passivated copper busbars, showing similar
performance to the reference modules with silver busbars, and confirming that the cell series resistance for copper
and silver busbars are identical after tabbing, regardless of the ~10x increase in bulk resistivity of the low
temperature copper paste. Cells and encapsulated cells with printed Cu BB have shown good reliability under damp
heat and thermal cycling. Module reliability has been shown to be well in excess of the IEC 61215 standard
requirements for TC and DH, and a temperature stress test on cells with Cu paste busbar at 300 °C has shown that
there is no degradation of the solar cells through diffusion of Cu through the SiN ARC and into the Si.
732 Don Wood et al. / Energy Procedia 55 (2014) 724 – 732

References

[1] Verlinden P, Cost analysis of current production and strategy for future silicon PV modules, Presented at the 28th EUPVSEC, Paris (2013).
[2] Caroline Boulord, Pierre Chevalier, Nick Powell, Adriana Zambova, Composition and conductor formed therefrom, WO2013/090570A1.
[3] Guy Beaucarne, Jorg Horzel, Nick Powell, Loic Tous, Don Wood, Adriana Zambova, Photovoltaic cell and method of forming the same,
WO2013/090562A3.
[4] Manabu Yoshida, Hideo Tokuhisa, Uichi Itoh, Toshihide Kamata, Isao Sumita, Shigenobu Sekine, Novel low-temperature-sintering type Cu-
alloy pastes for silicon solar cells, 3rd Workshop on Metallization for Crystalline Silicon Solar Cells, 25 – 26 October 2011, Charleroi,
Belgium
[5] Shuhei Yoshiba, Marwan Dhamrin, Manabu Yoshida, Abdullah Uzum, Uichi Itoh, Hideo Tokuhisa, Shigenobu Sekine, Tadashi Saitoh and
Koichi Kamisako, Cost effective front contact metallization by copper paste for screen-printed crystalline silicon solar cells, 27th EUPVSEC,
Hamburg, 25-28 September 2012.
[6] Jaap Hoonstra, Benoit Heurtault, Stencil print applications and progress for crystalline silicon solar cells, 24th EUPVSEC, 21-25 September
2009.
[7] IEC61215, edition 2.0, published 2005-04. Geneva, Switzerland, www.iec.ch

View publication stats