Professional Documents
Culture Documents
a r t i c l e in fo
abstract
Article history:
Received 24 March 2008
Received in revised form
25 July 2008
Accepted 5 August 2008
Available online 21 September 2008
Using a simple process, high-efciency silicon concentrator solar cells have proved to achieve up to 21%
efciency at 100 . The purpose of this work is to prove the feasibility of their industrialisation by
setting up a pilot line and manufacturing a signicant number of cells for a 100 concentrator system.
The process has been successfully veried by modifying the antireection coating, the annealing
process and the back contact. This yielded an average efciency of 18.5% at 100 with 70% of cells
having an efciency 418% and costs ranging from 0.31 to 0.41 h/W. A fast learning curve is shown which
suggests optimistic results indeed for further industrialisation.
& 2008 Elsevier B.V. All rights reserved.
Keywords:
Silicon solar cells
Concentrator
Industrialization
1. Introduction
Photovoltaic concentrator systems (CPV) have been available
for almost 30 years in laboratories and research elds to provide
green electricity to the market. As in the development of at
plate photovoltaic systems (at plate PV or at plates) for
terrestrial purposes, R+D into CPV systems started soon after the
oil crisis in 1973. Unlike at plate systems which found a niche
application for remote systems (emergency and telecommunications, for example) and in developing countries, concentrator
systems were developed for large power plant applications where
they competed with well-established thermal power stations. As
oil prices came down again, the need for a commercial CPV system
disappeared together with a high percentage of funding [1].
Meanwhile, a number of interesting CPV systems with very
different approaches and degrees of success have been eld
tested, but none of them has achieved the mass production level
of at plates [2]. These systems include different optics (mirrors or
lensesFresnel or anidolic), spot sizes and geometries (point or
linear focus, from just a few to several hundred suns), tracking
strategies (static, single or double axis), refrigeration systems
(active or passive) and cells (Si or IIIV compound semiconductors, whether single or multijunction).
Though most of the demonstration systems have been
manufactured using silicon solar cells, the current, main research
effort into concentrator solar cells focuses on multijunction
ARTICLE IN PRESS
1698
M. Castro et al. / Solar Energy Materials & Solar Cells 92 (2008) 16971705
ARTICLE IN PRESS
M. Castro et al. / Solar Energy Materials & Solar Cells 92 (2008) 16971705
1699
Table 1
IES-UPM cell characteristics and its performance at 100 concentration, 25 1C
simulated in PC1D
Material
1.3 3 cm2
1 3 cm2
0.2470.01 O cm
28973 mm
50 O/sq
5.65%
9.6 mO cm2
14.4 mO cm2
0.9 (6%) mO cm2
1.0 (7%) mO cm2
2.5 (17%) mO cm2
4.8 (33%) mO cm2
5.3 (36%) mO cm2
3.47 A/cm2
776 mV
81.2%
21.83%
reducing its lifetime. To make sure this did not occur, lifetime
measurements using Photo Conductance Decay (PCD) and Quasi
Steady State Photo Conductance (QSSPD) methods were taken in
wafers after all thermal processes.
In order to process these cells, a new pilot line has been set up
at the IES-UPM facilities. This pilot line has been put into
operation using some of the previous laboratory installations
and adding some new equipment. Chemical and furnace steps are
carried out in batches of 20 wafers, evaporation steps in batches of
14 wafers and batches of 8 wafers were used for silver electrolysis.
Photolithography and lift-off steps are carried out wafer by wafer.
In an industrial environment, batch size could be uniform for
every process step, thereby permitting an improved manufacturing sequence and higher throughput.
After the devices are cut from the wafer into individual cells,
they are characterised under concentration. Measurements are
taken using a solar simulator with ash illumination and a data
acquisition system [15], while the cell is connected using a
specially-designed jig permitting rapid cell positioning and
introducing low resistance to the measurement prior to any
encapsulation or soldering process, (Fig. 3). The light intensity is
10 W/cm2 and the real concentration level is checked using a
previously-calibrated silicon concentrator cell and assuming
current linearity. The fast ash system allows the cell to be
measured at 25 1C.
ARTICLE IN PRESS
1700
M. Castro et al. / Solar Energy Materials & Solar Cells 92 (2008) 16971705
Fig. 4. Efciency vs. concentration for several sample cells in the rst batch,
showing high series resistance, very low efciency and large dispersion in values.
Table 2
Electrical performance at 100 suns for several sample cells taken from the rst batch processed
Cell
Fill factor FF
Efciency Z (%)
Series resistance Rs
(mO)
B9A
B9D
B9E
C3A
C3B
C3I
Average
Deviation (%)
3.1
3.0
3.1
3.7
3.0
1.8
3.0
21
752
743
759
753
736
748
749
1
0.59
0.48
0.75
0.41
0.38
0.35
0.49
31
13.9
10.7
17.8
10.5
7.6
4.5
10.8
43
10
17
N/A
21
18
14
16
26
ARTICLE IN PRESS
M. Castro et al. / Solar Energy Materials & Solar Cells 92 (2008) 16971705
1701
Table 3
Rear side resistive layer removal processes
Process
Etched materials
Time (min)
Temperature (1C)
Protective coating
All
SiO2
Al+SiO2
Al2O3/Al+SiO2
5
10
40+10
13+5
N/A
Ambient
3740+Ambient
3540+Ambient
No
Yes
Yes
Yes
Table 4
Characteristics at 100 concentration of the average and best cells of each process
Process
Average Rs (mO)
Average Z (%)
Initial
Mech. polish
BHF
HCl+BHF
NH3+BHF
1873
673
471
472
2775
872
1872
1772
1972
672
21
2
2
N/A
26
10.5
21.0
20.0
21.3
8.2
74877
76673
75673
76472
75073
2.4
1.1
2.1
0.3
Fig. 5. Efciency vs. concentration for several back resistive layer removal
processes, (A). average and (B). best cell. The best cell for HCl etch was measured
only at 100 suns.
ARTICLE IN PRESS
1702
M. Castro et al. / Solar Energy Materials & Solar Cells 92 (2008) 16971705
Fig. 8. Efciency vs. concentration measurements of one of the best cells produced
which maintained almost 20% efciency up to 250 .
In these new runs, the problems arising during the rst run
were avoided. Short-circuit current densities higher than 35 mA/cm2
were measured, as a result of decreasing the shadowing factor and
optimal ARC thickness. Cells had a solderable rear surface and no
back contact resistance was found. The series resistance was of
the order of 45 mO, as per design. Some of the cells were
characterised up to 250 and gave efciencies as high as 19.8% at
250 , as can be seen in Fig. 8, which shows the expected
performance of the cells simulated with PC1D as well. The main
difference between the measured and the simulated performance
was due to a higher shadowing factor in the real cell. The last
curve in Fig. 8 shows the simulation in PC1D of the designed cell
with an increased shadowing factor, in agreement with the
measured performance.
Fig. 9a shows the histogram of the measurements at 100
concentration level and the corresponding normal distribution
ARTICLE IN PRESS
M. Castro et al. / Solar Energy Materials & Solar Cells 92 (2008) 16971705
1703
Fig. 10. Short circuit current of main production according to runs and its
corresponding t to a normal distribution (lines).
Fig. 11. Learning curve: breakages and performance for each batch. Batches D, E
and F belong to Run 1, the remainder to Run 2. Black dashed line is the total
yield loss; the black solid line indicates those measured cells with Z416%.
ARTICLE IN PRESS
1704
M. Castro et al. / Solar Energy Materials & Solar Cells 92 (2008) 16971705
Table 5
Key gures used in cost calculations
70% Yield, 18% efciency
Key equipment
Wet benches
Diffusion furnaces
Aligner
E-beam evaporators
Grinders
Dicing saws
4
1 Stack with 4 tubes
1
8
2
2
4
1 Stack with 4 tubes
1
6
2
2
Labour
Direct labour
Engineering
Administration and Directors
19
5
8
14
5
8
Persons/shift
Persons
Persons
Materials
Silicon wafers
Silicon usage
Electronic grade chemicals
High purity gasses
Metals
Dopants
126,986
0.74
100 103
5.0 103
48
2.5
91,170
0.53
70 103
3.6 103
35
1.8
Wafers/year
g/W
l/year
m3/year
Kg/year
l/year
Units
Fig. 12. Cost distribution of the process for the two scenarios considered.
Though in this work we have not gone into the whole system
costs and characteristics, it is very reasonable to admit that with
these advantages, the whole 100 concentration system (140
geometric) can achieve prices comparable to the current market
prices at a mass production level, not being limited by the cell
production or price.
7. Further improvements
The feasibility of manufacturing 100 concentration cells
using a simple process is demonstrated in this paper. However,
the process was additionally complicated during production
development by adding a highly resistive rear side layer removal
process. This layer appears during the growth of the silicon oxide
ARC during drive-in. For large mass production, other ARCs that do
not form an oxidised back surface should be considered. These
could be TiO2, SiN grown by chemical vapour deposition (CVD) or
plasma-enhanced chemical vapour deposition (PECVD), lowtemperature silicon oxide. PECVD silicon oxide could also be
easily used for this purpose, depending on the required refractive
index and passivation. All of these layers are deposited at low
temperature, preventing further lifetime degradation. SiN has the
ARTICLE IN PRESS
M. Castro et al. / Solar Energy Materials & Solar Cells 92 (2008) 16971705
advantage of having good bulk and surface passivation characteristics when grown in an excess of hydrogen [20].
Regarding yield, the weakest point is the ultrasonic lift-off
process. Further improvements in this specic step are needed
before tackling mass production. A better approach would be to use
metal etch instead of the lift-off process to dene the front metal
grid. This would avoid the ultrasonic step and probably produce a
better yield. This process would change the sequence to: front metal
evaporation/photolithography/metal etch, instead of the photolithography/front metal evaporation/lift-off sequence used here. The
main disadvantage of this sequence is the need to open the contacts
in the passivation and the antireection oxide before metal
evaporation processes, something that would require another
photolithography step and extremely precise alignment with the
metal etch mask after metal deposition. For mass production, the
trade-off between an increase in the number of photolithography
steps and the breakages in the ultrasonic lift-off process should be
evaluated in order to choose from these two alternatives. It was
denitely not worth it in the existing facilities at IES-UPM.
Last, but not least, the performance of the cells under
concentration should be considered. Although good performances
were achieved, dispersion of the results was quite high. More
automatic, continuous large volume processing would improve
the average efciency in addition to reducing the dispersion of
results. This is due to more controlled and stable processes, the
opposite of what happens under less repetitive conditions typical
of a research laboratory such as the facilities at IES-UPM. For
example, dispersions found in the emitter due to the variable
conditions of the diffusion furnaces during production could have
been easily avoided.
8. Conclusions
In this work, the feasibility of fabricating 100 concentration
cells using a simple process is proven. However, some unexpected
problems such as the highly resistive back contact layer arose
which required some major changes to the initial processing
sequence. These changes include annealing before back contact
evaporation, the need to add a polishing step on the back surface
in order to remove the highly resistive layer formed during the
drive-in and an ARC growth step.
Using this new process, more than 1000 good performance
cells were processed. This is one of the largest productions
reported for this simple manufacture process of high concentration (100 ) silicon solar cell of conventional structure, and just
two photolithography masking steps, showing very good performance, with an average efciency of 18.5% and 70% of the cells
produced having efciency in excess of 18% at 10 W/cm2. Cells of
up to 21.7% efciency at 100 were also measured.
The weakest points of the production line were the dicing of
the cells and the manual lift-off process. These results could be
easily improved by automating both of these manual processes. In
addition to which, the use of different ARC layers would prevent
the back resistive layer, resulting in a more simplied process.
The cost associated to this process has been evaluated and
ranges from 0.31 to 0.41 h/W. This indeed is perhaps the most
signicant purpose of this work: to demonstrate that the IES-UPM
base line and the technologies used for the space silicon solar cells
is capable of producing high-efciency concentration silicon solar
cells in mass production when manipulated in a production
environment, and cost-effectively.
1705
Acknowledgements
eiro and J.C.
The authors would like to thank J. Montes, J. Pin
Zamorano for the assistance in the setting up of the pilot and
izo for the help in calculating
during the cell processing; C. Can
costs and J. C. Jimeno and his team at TiM in Bilbao for their
photolithography support.
This work has been supported by The Spanish Ministry of
Education and Science under its Consolider Ingenio 2010 Programme, through the GENESIS-FV(CSD2006-0004) Project.
References
[1] R.M. Swanson, The promise of concentrators, Prog. Photovoltaics Res. Appl. 8
(2000) 93111.
[2] G. Sala, D. Pachon, I, Anton, Test, Rating and Specication of PV Concentrator
Components and Systems (C-Rating Project). Book 1. Classication of PV
Concentrators, Universidad Politecnica de Madrid, Spain, 2002. /http://
www.ies-def.upm.es/ies/CRATING/Documents.htmS.
[3] R.R. King, D.C. Law, K.M. Edmondson, C.M. Fetzer, G.S. Kinsey, H. Yoon, D.D.
Krut, J.H. Ermer, R.A. Sherif, N.H. Karam, Metamorphic Concentrator Solar
Cells with Over 40% Conversion Efciency, in: Proceedings of the Fourth
International Conference on Solar Concentrators for the Generation of
Electricity or Hydrogen, El Escorial, Spain, 2007, pp. 58.
[4] G. Sala, I. Anton, J. Monedero, P. Valera, M.P. Friend, M. Cendagorta, F. Perez, E.
Mera, E. Camblor, The Euclides-thermie Concentrator Power Plant in
Continuous Operation, in: Proceedings of the 17th EUPVSEC, Munich,
Germany, 2001, pp. 488491.
[5] R.M. Swanson, S.K. Beckwith, R.A. Crane, W.D. Eades, Y.H. Kwark, R.A. Sinton,
S.E. Swirhun, Point-contact silicon solar cells, IEEE Trans. Electron Devices ED31 (5) (1984) 661664.
[6] V. Garboushian, S. Yoon, G. Turner, A. Gunn, D. Fair, A Novel High
Concentration PV Technology For Cost Competitive Utility Bulk Power
Generation, in: Proceedings of the rst WCPEC, Hawaii, USA, 1994,
pp. 10601063.
[7] P. J. Verlinden, A. Lewandowski, C. Bingham, G. S. Kinsey, R. A. Sherif, and J. B.
Lasich, Performance and Reliability of Multijunction IIIV Modules for
Concentrator Dish and Central Receiver Applications, in: Proceedings of the
fourth WCPEC, Hawaii, USA, 2006, pp. 592597.
[8] W.P. Mulligan, D.H. Rose, M.J. Cudzinovic, D.M. De Ceuster, K.R. McIntosh, D.D.
Smith, R.M. Swanson, Manufacture of Solar Cells with 21% Efciency, in:
Proceedings of the 19th EUPVSEC, Paris, France, 2004, pp. 387390.
[9] A. Slade, R. Gordon, D. Dutra, V. Garboushian, Long Term Reliability of MassProduced High-Efciency Silicon Point-Contact Solar Cells Under 250
Concentration, in: Proceedings of the 29th IEEE PVSC, New Orleans, USA,
2002, pp. 10151018.
[10] G.S. Kinsey, P. Pien, P. Hebert, R.A. Sherif, Multijunction Solar Cells
for CPV, in: CPV today. Concentrated Photovoltaic Summit 08, Madrid, Spain,
2008.
[11] J. Coello, M. Castro, I. Anton, G. Sala, M.A. Vazquez, Conversion of commercial
Si solar cells to keep their efcient performance at 15 Suns, Prog.
Photovoltaics: Res. Appl. 12 (2004) 323331.
[12] M.J. Terron, Desarrollo de Celulas Solares de Facil industrializacion Para su
Uso en Concentracion, Ph.D. Thesis, Universidad Politecnica de Madrid, Spain,
1997.
izo, R. Lago, I. Pou, A. El Moussaoui, A. Luque, F. Schomann, W.
[13] C. del Can
Schmidt, Pilot Line Production of Silicon Concentration Solar Cells, in:
Proceedings of the 16th EUPVSEC, Glasgow, UK, 2000, pp. 16631666.
[14] A. El Moussaoui, Fabricacion de Celulas Solares en Silicio de Bajo Coste, Ph.D.
Thesis, Universidad Politecnica de Madrid, Spain, 1993.
[15] I. Anton, Metodos y Equipos para la Caracterizacion de Sistemas Fotovoltaicos
de Concentracion, Ph.D. Thesis, Universidad Politecnica de Madrid, Spain,
2004.
[16] D.K. Schroder, D.L. Meier, Solar cell contact resistancea review, IEEE Trans.
Electron Devices ED-31 (5) (1984) 637653.
[17] R. Lago Aurrekoetxea, Recombinacion Supercial y de Volumen en Celulas
Solares con Tecnologa Fosforo-Aluminio sobre Silicio, Ph.D. Thesis, Universidad Politecnica de Madrid, Spain, 2002.
[18] R.A. Crane, P.J. Verlinden, R.M. Swanson, Building a Cost-effective, Fabrication
Facility for Silicon Solar Cell R&D and Production, in: Proceedings of the 25th
IEEE PVSC, Washington D.C., USA, 1996, pp. 529532.
izo, G. del Coso, W. Sinke, Crystalline Silicon Solar Module
[19] C. del Can
Technology: Towards the 1 Euro per Watt-peak Goal, Submitted to Progress
in Photovoltaics: Research and Applications, (2008).
[20] A. Luque, S. Hegedus (Eds.), Handbook of Photovoltaic Science and Engineering, Wiley, New York, 2003, pp. 283285.