Presented at the 22nd European Photovoltaic Solar Energy Conference and Exhibition, 3.-7.
September 2007, Milano
SILICON SOLAR CELLS WITH SCREEN-PRINTED FRONT SIDE METALLIZATION EXCEEDING
19% EFFICIENCY
O. Schultz1, A. Mette1,2, R. Preu1, S.W. Glunz1
1
Fraunhofer ISE, Heidenhofstrasse 2, 79110 Freiburg, Germany,
Phone ++761-4588-5355; Fax ++761-4588-9250, email:
[email protected] 2
Now with Q-Cells AG, Germany
ABSTRACT: Thermal oxides are commonly used for the surface passivation of high-efficiency crystalline silicon
solar cells and have led to the highest conversion efficiencies of mono- and multicrystalline silicon reported so far. In
contrast to the laboratory fabrication of solar cell structures with lithographically defined and evaporated front
contacts, in this study we combine several technologies ready for industrial implementation. On the front side,
surface passivation of the lowly doped emitters up to 90 Ω/sq is achieved with sputtered SiNx:H, the contact grid is
created with screen-printing of a hotmelt silver paste, the electrical properties of the grid are improved by light-
induced silver plating. Since contact formation on the front requires a short high-temperature firing process, the rear
surface passivation layer needs to be thermally stable. A wet thermal oxidation is applied on the rear and the surface
recombination velocity of the fired oxide is experimentally determined to be below S≤ 38 cm/s at the end of the
process sequence. This includes evaporation of aluminium on the rear, the formation of laser-fired contacts and a
short annealing process under air ambience. Monocrystalline solar cells were produced and 19.3% efficiency were
obtained as best value on 4 cm2 cell area.
Keywords: High-Efficiency, Silicon Solar Cells, Screen-Printing
1 INTRODUCTION
In order to improve the cost-effectiveness of solar
cells from mono- and multicrystalline silicon, the
efficiency has to be increased and wafer thickness has to
be reduced. One very important component of the cell
influencing the performance especially for thin wafers is
the rear surface. The increasing demands for optical
quality require a higher internal reflection and also the
electrical quality can be improved applying a more Alloyed Aluminium
sophisticated surface passivation than a full-area
aluminium back-surface field (Al-BSF). Therefore, all
designs for high-efficiency silicon solar cells today
feature a dielectric passivation layer on the rear. Provided
that the refractive index of this layer is low enough, this
results in a very good optical mirror with high internal
reflection. This leads to a good light-trapping that can be
further improved by evaporation of a thin layer of
aluminium. To extract the carriers from the cell, a point-
contact pattern has proven sufficient for good carrier
collection [1]. Both cell structures are sketched in
Figure 1. A very good surface passivation is required in Evaporated Point Contacts
order to realise an efficiency gain of the passivated rear Aluminium Oxid
in comparison to a full-area contacted screen-printed Al-
BSF. Fig. 1: Solar cell structures with screen-printed front
A common technology meeting the requirements of metallization and full area Al-BSF (top) and passivated
passivation quality and optical properties is a thick rear with point contacts (bottom)
thermal oxidation (≥100 nm) at high temperatures. This
has led to the highest conversion efficiencies on
monocrystalline silicon so far [2]. A technological process and therefore a viable technology for industrial
modification of the standard oxides grown at 1050°C in silicon solar cell processing.
dry ambience is the use of pyrogenic steam [3]. This Following the development of high-efficiency cells
increases the growth rate and/or reduces the oxidation using photolithography technology, in the present study
temperature. The latter is especially beneficial for high- cells are manufactured employing an industrial front
temperature sensitive material and enabled the first cells structure with screen-printed contacts. The front surface
in excess of 20% efficiency on multicrystalline silicon consists of a phosphorus doped emitter passivated by a
using a process at 800°C [4]. This wet oxidation is about single layer antireflection coating of silicon nitride
one order of magnitude faster than the dry oxidation (SiNx:H). The contact grid is screen-printed hotmelt
silver paste [5], which needs to be fired through the SiNx-
Presented at the 22nd European Photovoltaic Solar Energy Conference and Exhibition, 3.-7. September 2007, Milano
layer. This short high-temperature process of 800°C - a contact pitch of 750 µm, this results in effective surface
900°C can potentially degrade passivation layers. recombination velocities of Seff = 112 cm/s and 143 cm/s,
Therefore, a good process design needs to take into respectively [8].
account the passivation quality at the end of the process Although the recombination velocities are
sequence, i.e. the rear passivation layer has to be significantly higher than the values derived for the wet
temperature stable. oxide without the firing process [9], this is low enough to
The question to be answered is: Can a wet thermal provide excellent surface passivation and allows for
oxide yield low rear surface recombination velocities and highly efficient solar cells. Therefore, the oxide was
withstand the firing process? implemented in the process sequence as described in the
following section.
2 EXPERIMENTS
2.2 Process sequence
2.1 Thermal stability of thermal oxide for rear surface
passivation In order to test the cell structure, high-quality boron-
doped FZ silicon of 0.5 Ω cm base resistivity and 250 µm
Bright-etched boron-doped FZ silicon wafers of thickness was used as a substrate for seven cells of 4 cm2
1 Ω cm base resistivity and of 330 µm thickness were aperture area on each wafer. Both, cells with a standard
cleaned and oxidised in steam ambience. The wafers industrial Al-BSF and cells with an oxide passivation
were fired in an inline belt furnace at the same settings were processed in order to experimentally determine the
like they were used in solar cell fabrication (see next benefit of the point-contacted oxide-passivated rear
section). Then aluminium of 2 µm thickness was compared to the full-area Al-BSF. Recently developed
evaporated on both sides and the wafers were annealed at technologies for crystalline silicon solar cell production
350°C for 15 min before the metallic layers were etched like wet oxidation [4], sputtering of silicon nitride [10],
off in hydrochloric acid. The effective carrier lifetime of screen-printing of hotmelt silver paste [5], laser-fired
the samples was measured with the QssPC technique [6]. contacts [11] and light-induced silver plating [12] were
The average lifetime was 335 µs. Evaluation according to implemented in an industrial process sequence shown in
equation Figure 2 resulting in the cell structures shown in
1 1 2S Figure 1. The thermal oxide was applied at the very
≅ + beginning of the process sequence and grown thick
τ eff τ bulk W
enough to withstand the random pyramid etching, to
with the assumption of τbulk = 1500 µs results in an upper mask the local POCl3-diffusion and to resist the short
limit of S ≤ 38 cm/s. This can be converted to 54 cm/s for etching of the phosphorus glass.
0.5 Ω cm material [7]. Applying laser-fired contacts with
4’’ FZ wafer (250µm thick, 0.5 Ω cm, p-doped, bright etched)
Wet chemical cleaning
Dry oxidation Wet oxidation
Selective removal of oxide from front side (2 x 2 cm² sized areas)
Random pyramids texture of front
POCl3 planar diffusion (40, 60, 90 Ω/sq)
PSG etching, cleaning
Removal of oxide from rear
Sputtered SiNx:H antireflection coating
Screen-printing and drying of rear contacts (conv. Al paste)
Screen-printing and drying of front contacts (Ag hotmelt paste)
Contact formation (fast firing belt furnace)
Evaporation of aluminium on rear surface
Laser fired contacts (LFC)
Annealing of rear surface
Light-induced plating of front contacts
Annealing in air ambient
Fig. 2: Process flow diagram of manufactured solar cells featuring screen-printed front contacts thickened by light-induced
plating. The left flow shows the Al-BSF process, on the right-hand side the rear side is passivated by wet thermal oxide.
Presented at the 22nd European Photovoltaic Solar Energy Conference and Exhibition, 3.-7. September 2007, Milano
This has the advantage that the rear surface is covered In the short wavelength region, and therefore linked to
during the whole process sequence and no emitter etch- the properties of the front surface of the cells, the
back nor sophisticated cleaning processes have to be difference in the quantum efficiency is due to the
applied except at the very beginning. different emitter doping profiles and thus the two types of
cell structure show nearly equal performance. In the long
2.3 Results wavelength region, representing the rear surface
properties, the Al-BSF and the oxide passivated cells
The results of the IV-measurements of the best cells exhibit distinct differences. The superior rear surface
under standard testing conditions are shown in Table 1. passivation and the high internal reflectance lead to a
significant gain in the short-circuit current density and
Table 1: IV-parameters of the best solar cells with Al-
open-circuit voltage compared to the Al-BSF cells.
BSF or oxide passivated rear.
Additionally, differences in the reflectance between
*Calibrated measurement at Fraunhofer ISE Calibration
1150 nm and 1200 nm can be observed for the three
Laboratory
different emitters. This can be attributed to free-carrier
absorption of the long-wavelength photons in the highly
Rsh rear Voc jsc FF η
doped emitter layers.leading to slightly higher quantum
[Ω/sq] side [mV] [mA/cm2] [%] [%]
efficiencies for the 60 Ω/sq and 90 Ω/sq emitter also in
Al-BSF 632 35.2 81.2 18.1 the long-wavelength region.
40
Oxide 639 36.7 80.0 18.8 The fill factor FF of the point-contacted cell is
Al-BSF 632 35.6 81.2 18.3 slightly lower than the fill factor of the full-area
60
Oxide 647 37.1 79.6 19.1 contacted Al-BSF. This can be explained by increased
Al-BSF 640 36.5 79.8 18.7 series resistance on the rear due to the point-contact
90
Oxide 655 38.2 76.9 19.3* structure and increased spreading-resistance [13, 14]. The
differences in FF are small for the 40 Ω/sq and the
For all emitter diffusions the Voc and jsc values are 60 Ω/sq emitter, for the 90 Ω/sq the difference cannot be
significantly increased for the oxide passivated cells. For explained by the cell design but must be due to a
further analysis the quantum efficiency and reflectance technological problem. However, the high gain in jsc and
measurements are shown in Figure 3. Voc still allow a gain of 0.6% absolute. An electrical
conversion efficiency of 19.3% was measured in a
1.0 calibrated measurement at Fraunhofer ISE CalLab. This
0.9 efficiency is, to the knowledge of the authors, the highest
0.8 value reported for a screen-printed front metallization
pattern to date [15].
0.7
0.6 90 Ω/sq oxide passivated rear 2.4 Simulation
0.5 60 Ω/sq oxide passivated rear
IQE
40 Ω/sq oxide passivated rear
0.4 90 Ω/sq Al-BSF In order to assess the potential of the cells structure
60 Ω/sq Al-BSF and the process, 1-dimensional modelling using PC1D
0.3 40 Ω/sq Al-BSF
0.2
[16] was performed for the cells with 60 Ω/sq emitter.
The doping profile was measured by SIMS and used for
0.1
the Al-BSF and the oxide passivated rear. First, the
0.0 reflectance was simulated by adjusting the internal
300 400 500 600 700 800 900 1000 1100 1200
Wavelength [nm] reflectance values in PC1D in a way that a good fit was
obtained. This resulted in Rback = 66% (diffuse) on the
rear for the cell with Al-BSF and Rback = 92% (specular)
1.0 for the oxide passivated one.
0.9
90 Ω /sq oxide passivated rear 1.0
0.8 60 Ω /sq oxide passivated rear 0.9
0.7 40 Ω /sq oxide passivated rear
90 Ω /sq Al-BSF 0.8
0.6 60 Ω /sq Al-BSF
Reflectance
40 Ω /sq Al-BSF
0.7
0.5
0.6 60 Ω/sq oxide passivated rear
0.4 60 Ω/sq Al-BSF
0.5
IQE
Simulation
0.3
0.4
0.2 0.3
0.1 0.2
0.0 0.1
300 400 500 600 700 800 900 1000 1100 1200
Wavelength [nm] 0.0
300 400 500 600 700 800 900 1000 1100 1200
Wavelength [nm]
Fig. 3: Measurement of internal quantum efficiency IQE
and reflectance for the Al-BSF and for the oxide
passivated rear. Fig. 4: Simulated internal quantum efficiency and
reflectance (lines) fitted to the measurements (dots).
Presented at the 22nd European Photovoltaic Solar Energy Conference and Exhibition, 3.-7. September 2007, Milano
Then the rear surface recombination velocity Seff of surface recombination velocities of Seff =143 cm/s on
the Al-BSF cell was adjusted. In order to obtain a good 0.5Ω cm base resistivity were calculated for a contact
fit of the measured quantum efficiency a value of pitch of 750 µm. Solar cells of 4 cm2 aperture area were
Seff =900 cm/s was chosen. For the oxide passivated cell fabricated and efficiencies in excess of 19% were
the value of Seff = 143 cm/s calculated in section 2.1 was reached for emitter sheet resistances of 60 Ω/sq and
used and allowed a good description of the measured 90 Ω/sq.
data. In both cases a bulk lifetime of τbulk = 500 µs was
assumed. The simulation results are displayed in Fig. 4, ACKNOWLEDGEMENT
the main fit parameters for the 60 Ω/sq. cells are
summarized in Table 2. The authors would like to thank all members of the
Fraunhofer ISE Solar Cell Department for their
Table 2: Internal reflectivities (Rback) and effective rear contributions to this work.
surface recombination velocities at the back (Seff) as
extracted from solar cell modelling ((d) = diffuse and (s) REFERENCES
specular reflection). In addition the jsc and Voc simulation
results are presented. The simulation was performed on [1] A.W. Blakers and M.A. Green, Applied Physics
cells with 60 Ω/sq emitter. Letters 48 (1986) 215
Seff [2] J. Zhao, A. Wang and M.A. Green, Progress in
Rback Voc jsc
Rear side Photovoltaics: Research and Applications 7 (1999)
[%] [cm/s] [mV] [mA/cm2]
471
Al-BSF 66 (d) 900 637 35.5 [3] M. Stocks and A. Cuevas, Proceedings of the 2nd
World Conference on Photovoltaic Energy
oxide 92 (s) 143 645 37.1 Conversion, Vienna, Austria (1998) 1623
[4] O. Schultz, S.W. Glunz and G.P. Willeke, Progress
The relative gain by using the oxide passivation instead in Photovoltaics: Research and Applications 12
of a full-area Al-BSF is shown in Figure 5. According to (2004) 553
the simulation, the improved rear surface passivation and [5] A. Mette, D. Erath, R. Ruiz, G. Emanuel, E. Kasper
higher internal reflectance increase Voc and jsc by 1.3% and R. Preu, Proceedings of the 20th European
and 4.5%, respectively. The difference in FF was Photovoltaic Solar Energy Conference, Barcelona,
experimentally determined to be 2% relative. Overall, a Spain (2005) 873
relative gain in efficiency of 3.8% for a standard 60 Ω/sq [6] R.A. Sinton and A. Cuevas, Applied Physics Letters
emitter is calculated, this gain will increase for even 69 (1996) 2510
lower doped emitters. [7] O. Schultz Dissertation, Fakultät für Physik,
Universität Konstanz, Konstanz 2005
[8] D. Kray and S.W. Glunz, Progress in Photovoltaics:
10 Research and Applications 14 (2006); 195
9
8
[9] O. Schultz, S.W. Glunz, D. Kray, M. Dhamrin, I.
Voc jsc FF η
7 Yamaga, T. Saitoh and G.P. Willeke, Proceedings of
the 20th European Photovoltaic Solar Energy
Relative gain [%]
6
5 Conference, Barcelona, Spain (2005) 741
4
3
[10] W. Wolke, J. Catoir, R. Preu, G. Emanuel, J. Liu, M.
2 Ruske, Proceedings of the 20th European
1 Photovoltaic Solar Energy Conference, Barcelona,
0 Spain (2005) 733
-1
-2
[11] E. Schneiderlöchner, R. Preu, R. Lüdemann and
-3 S.W. Glunz, Progress in Photovoltaics: Research and
Applications 10 (2002) 29
[12] A. Mette, C. Schetter, D. Wissen, S. Lust, S.W.
Fig.5: Simulation results illustrating the relative gain of Glunz and G. Willeke, Proceedings of the 4th World
solar cell parameters by using the oxide passivation Conference on Photovoltaic Energy Conversion,
instead of a full area Al-BSF. Modelling refers to a Waikoloa, Hawaii, USA (2006) 1056
240 µm thick solar cell with a base doping of 0.5 Ω cm [13] K.R. Catchpole and A.W. Blakers, Proceedings of
and an emitter sheet resistance of 60 Ω/sq. the 16th European Photovoltaic Solar Energy
Conference, Glasgow, UK (2000) 1719
[14] B. Fischer, Dissertation, Fakultät für Physik,
3 SUMMARY Universität Konstanz, Konstanz 2003
[15] M.M. Hilali, K. Nakayashiki, A. Ebong and A.
A wet thermal oxide was implemented in the Rohatgi, Progress in Photovoltaics: Research and
beginning of an industrial-like process sequence. It Applications 14 (2006); 135
served as a masking layer against random pyramid [16] D.A. Clugston and P.A. Basore, Proceedings of the
texturing and POCl3 emitter diffusion. The front contacts 26th IEEE Photovoltaic Specialists Conference,
were formed with screen-printed and fired silver paste. Anaheim, California, USA (1997) 207
After aluminium annealing of the rear, excellent surface
recombination velocities of S≤ 38 cm/s were measured.
In combination with laser-fired contacts effective rear