www.elsevier.com/locate/solener
Received 31 March 2004; received in revised form 23 September 2004; accepted 24 September 2004
Available online 30 October 2004
Abstract
Our recent R&D activities of IIIV compound multi-junction (MJ) solar cells are presented. Conversion eciency of
InGaP/InGaAs/Ge has been improved up to 3132% (AM1.5) as a result of technologies development such as double
hetero-wide band-gap tunnel junction, InGaPGe hetero-face structure bottom cell, and precise lattice-matching of
InGaAs middle cell to Ge substrate by adding indium into the conventional GaAs layer. For concentrator applications,
grid structure has been designed in order to reduce the energy loss due to series resistance, and world-record eciency
InGaP/InGaAs/Ge 3-junction concentrator solar cell with an eciency of 37.4% (AM1.5G, 200-suns) has been fabri-
cated. In addition, we have also demonstrated high-eciency and large-area (7000 cm2) concentrator InGaP/InGaAs/
Ge 3-junction solar cell modules of an outdoor eciency of 27% as a result of developing high-eciency InGaP/InG-
aAs/Ge 3-junction cells, low optical loss Fresnel lens and homogenizers, and designing high thermal conductivity
modules.
Future prospects are also presented. We have proposed concentrator IIIV compound MJ solar cells as the 3rd gen-
eration solar cells in addition to 1st generation crystalline Si solar cells and 2nd generation thin-lm solar cells. We are
now developing low-cost and high output power concentrator MJ solar cell modules with an output power of 400 W/m2
for terrestrial applications.
2004 Elsevier Ltd. All rights reserved.
1. Introduction
0038-092X/$ - see front matter 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.solener.2004.09.018
M. Yamaguchi et al. / Solar Energy 79 (2005) 7885 79
tandem cells based on their computer analysis. Although space applications because of the possibility of high con-
AlGaAs/GaAs tandem cells, including tunnel junctions version eciency of over 30%. In fact, the commercial sa-
and metal interconnectors, were developed in the early tellite (HS 601HP) with 2-junction GaInP/GaAs-on Ge
years, a high eciency close to 20% was not obtained solar arrays was launched in 1997 (Brown et al., 1997).
(Hutchby et al., 1985). This is because of diculties in We have proposed AlInPInGaP double hetero-
making high performance and stable tunnel junctions, (DH) structure top cell, wide-bandgap InGaP DH struc-
and the defects related to the oxygen in the AlGaAs ture tunnel junction for sub-cell interconnection, and
materials (Ando et al., 1987). lattice-matched InGaAs middle cell. As a result of the
One of the authors has conducted his research on Al- above technology developments, the mechanically
GaAs/GaAs 2-junction solar cells since 1982. His group stacked InGaP/GaAs//InGaAs 3-junction cells (1 cm2)
has demonstrated 20.2% eciency (Amano et al., 1987) have reached the highest (1-sun world-record) eciency
by proposing double hetero-structure tunnel junction as of 33.3% at 1-sun AM1.5G following joint work by Ja-
a sub-cell interconnection in 1987 (Yamaguchi et al., pan Energy Co., Sumitomo Electric Co. and Toyota
1987). A double hetero- (DH) structure tunnel junction Tech. Inst. (Takamoto et al., 1997b). Since FY2001, this
was found to be useful for preventing diusion from R&D project has been shifted to the project for Con-
the tunnel junction and improving the tunnel junction centrator MJ Solar Cells and Modules.
performance by the authors (Sugiura et al., 1988). In this paper, more recent results for high-eciency
In Japan, based on such an activity, R&D project for IIIV compound multi-junction (MJ) solar cells, con-
Super-high Eciency MJ Solar Cells has been con- centrator MJ solar cells and modules conducted under
ducted under support by NEDO since scal year (FY) the New Sunshine Project in Japan are presented. Future
1990 (Yamaguchi and Wakamatsu, 1996) as a long-term prospects for super high-eciency and low-cost MJ and
target to the early 21st century, in which challenges and concentrator cells and modules are also presented.
eorts are made in the development of super-high e-
ciency solar cell technology, aiming at a dramatic in-
crease in conversion eciency of over 40% and 2. Key issues for obtaining high-eciency multi-junction
developing innovativel technologies. (MJ) solar cells
An InGaP material for the top cell was proposed by
Olson et al. (1990). As a result of performance improve- Conversion eciency of InGaP/GaAs based MJ solar
ments in tunnel junction and top cell, over 30% eciency cells has been improved by the following technologies. A
has been obtained with InGaP/GaAs tandem cells by the schematic illustration of the InGaP/(In)GaAs/Ge triple
authors (Takamoto et al., 1997a). Recently, InGaP/GaAs junction solar cell and key technologies for improving
2-junction solar cells have drawn increased attention for conversion eciency is shown in Fig. 1.
Fig. 1. Schematic illustration of a triple-junction cell and approaches for improving eciency of the cell.
80 M. Yamaguchi et al. / Solar Energy 79 (2005) 7885
2.1. Wide band-gap tunnel junction hetero-growth layer. In the case of GaAs hetero-growth
layer, junction depth was measured to be around 1 lm.
A wide band-gap tunnel junction which consists of On the other hand, thickness of n-type layer produced
double hetero-structure p-AlInGaP/p-AlGaAs/n-(Al)In- by phosphor from the InGaP layer was 0.1 lm. An in-
GaP/n-AlInGaP increases incident light into the (In)- crease in Ge quantum eciency was conrmed to be
GaAs middle cell and produces eective potential due to a reduction in junction depth.
barriers for both minority-carriers generated in the top It was found that the absorption edge of the InGaP
and middle cells. The wide band-gap tunnel junction with- top cell shifted to the longer wavelength region, by using
out absorption and recombination losses improves both the InGaP rst hetero-layer. Band-gap of the InGaP top
Voc and Isc of the cells. It is dicult to obtain high tunnel- cell reduced from 1.86 to 1.81 eV by changing the het-
ing peak current with wide gap tunnel junction, so thin- ero-growth layer from GaAs to InGaP. The fact that
ning depletion layer width by formation of highly doped the band-gap increased with the growth temperature
junction is quite necessary. Since impurity diusion oc- indicated that this phenomenon was due to the ordering
curs during growth of the top cell (Sugiura et al., 1988), eect in the InGaP material (Gomyo et al., 1987). Since
carbon and silicon, which have low diusion coecient, the band-gap narrowing of the top cell decreases Voc of
are used for p-type AlGaAs and n-type (Al)InGaP, the triple-junction cell, an approach for growth of less-
respectively. Furthermore, the double hetero-structure is ordering InGaP should be necessary. As a matter of a
supposed to suppress impurity diusion from the highly fact, conversion eciency has been improved up to
doped tunnel junction (Takamoto et al., 1999). The sec- 30% (AM0) by increasing the top cell band-gap up to
ond tunnel junction between middle and bottom cells con- 1.89 eV (King et al., 2002).
sists of p-InGaP/p-(In)GaAs/n-(In)GaAs/n-InGaP which
have wider band-gap than middle cell materials. 2.3. Precise lattice-matching to Ge substrate
2.2. InGaP/Ge hetero-face structure bottom cell Although 0.08% lattice-mismatch between GaAs and
Ge was thought to be negligibly small, mist-dislocations
InGaP/GaAs cell layers are grown on a p-type Ge were generated in thick GaAs layers and deteriorated cell
substrate. PN junction is formed automatically during performance. By adding about 1% indium into the In-
MOCVD growth by diusion of V-group atom from GaP/GaAs cell layers, all cell layers are lattice-matched
the rst layer grown on the Ge substrate. So, the mate- precisely to the Ge substrate. As a result, cross-hatch pat-
rial of the rst hetero-layer is important for the perfor- tern caused by mist-dislocations due to lattice-mismatch
mance of Ge bottom cell. An InGaP layer is thought was disappeared in the surface morphology of the cell
to be suitable material for the rst hetero- layer, because with 1% indium, as shown in Fig. 3. The mist-disloca-
phosphor has lower diusion coecient in Ge than ar- tions were found to inuence not to Isc but to Voc of
senic and indium has lower solubility in Ge than gal- the cell. Voc was improved by eliminating mist-disloca-
lium. Fig. 2 shows the change in spectral response of tions for the cell with 1% indium. In addition, wavelength
the triple-junction cell by changing the rst hetero- of the absorption edge became longer and Isc of both top
growth layer on Ge from GaAs to InGaP. Quantum e- and middle cells increased, by adding 1% indium.
ciency of the Ge bottom cell was improved by the InGaP
2.4. Widening of top cell band-gap by AlInGaP
Fig. 3. Surface morphology of InGaAs with various indium composition grown on Ge.
Fig. 6. 7000 cm2 and 400 concentrator module with 36 receivers connected in series and dome-shaped Fresnel lens made by injection
mold.
M. Yamaguchi et al. / Solar Energy 79 (2005) 7885 83
A 400 and 7000 cm2 concentrator module was fab- (3) A new encapsulating polymer that survives exposure
ricated with 36 pieces of randomly-selected solar cells of high concentration UV and heat cycles.
connected in series and the same number of the newly- (4) Beam-shaping technology that illuminates the
developed dome-shape Fresnel lenses as shown in Fig. square aperture of the solar cells from round con-
6. The module without heat sinks and external cooling centration spot.
utilities were used. (5) Homogenizer technologies giving uniform ux and
The eciency in a hot summer day (35 C of ambient prevent from conversion loss stemmed from chro-
temperature) was 27.0% as shown in Fig. 7 (Araki et al., matic aberrations and surface voltage variation.
in press). This value was well above on the record e- (6) Allowing as large as 1.75 mm assemble tolerance.
ciency of 22.7% (Bett et al., 2003) established by Fraun- There is no need of special optical alignment. Even
hofer Institute Germany in October 2002 with one-order local mechanical industries can assemble the main
less area (768 cm2). This achievement resulted from sev- body.
eral new technologies: (7) Shaped Fresnel lens designed by non-imaging optics
theory and made by low-cost injection mold that
(1) High-pressure and vacuum-free lamination of the gives tight variation of optical eciency as narrow
solar cell that suppress temperature rise to only as 0.5%.
8C under 400 geometrical concentration illumina-
tion of sunbeam. The technological roadmap toward 31% ecient
(2) Direct and voids-free soldering technologies of fat module is shown in Table 2. Since we have identied de-
metal ribbon to solar cells that suppresses the hot tailed technological problems and how to solve them, we
spots and resistance by 300 times and give a higher expect that the target will be achieved well in advance.
output current than normal non-concentration
operation.
5. Future prospects
Table 2
Roadmap table of more than 31% eciency module
Technologies (%)
November 2002 August 2003 March 2004 March 2005 March 2006
A Cell eciency at 1-sun 30.1 30.3 32
B Cell eciency concentration 34.4 35.3 37 40
C Lens eciency 72.4 85.4 85.8 91
D Homogenizer eciency 94.4 96.3 97.5 97.5 97.5
E Ohmic loss in circuit 0.1 0.1 0.1 0.1 0.1
F Spectrum mismatching loss 5.3 5.1 3
G Current mismatching loss 3.7 3.7 2 2 2
H Loss by temperature rise 1.2 1.3 1.3 1
I Total eciency 21.7 27.0 29 >29 >31
84 M. Yamaguchi et al. / Solar Energy 79 (2005) 7885
45
40
35 10
30
25
20
1 2 3 4
Number of junction
1
1995 2000 2005 2010 2015 2020
Fig. 8. Theoretical and realistically expected conversion e- Year
ciencies of single-junction and multi-junction solar cells in
comparison with experimentally realized eciencies under 1- Fig. 10. Scenario of electricity cost reduction by developing
sun and concentration conditions. concentrator solar cells.
is lattice-matched to Ge substrates and has a theoretical reduction for widespread PV applications as shown in
1-sun AM0 eciency of about 42%. This system has Fig. 10 (Yamaguchi, 2002).
also potential of over 45% under 500-suns AM1.5
condition.
We are now challenging ourselves to develop low- 6. Summary
cost and high output power concentrator MJ solar cell
modules with an output power of 400 W/m2 for terres- Conversion eciency of InGaP/InGaAs/Ge has been
trial applications as shown in Fig. 9. improved up to 3132% (AM1.5) and 2930% (AM0) as
Concentrator operation of the MJ cells is essential a result of technologies development such as double het-
for their terrestrial applications. Since the concentrator ero-wide band-gap tunnel junction, InGaPGe hetero-
PV systems have potential of cost reduction, R&D on face structure bottom cell, and precise lattice-matching
concentrator technologies including MJ cells is started of InGaAs middle cell to Ge substrate by adding indium
in Japan. Therefore, concentrator MJ and crystalline into the conventional GaAs layer. For concentrator
Si solar cells are expected to contribute to electricity cost applications, grid structure has been designed in order
to reduce the energy loss due to series resistance, and
world-record eciency of 37.4% (AM1.5G, 200 suns)
has been demonstrated for InGaP/InGaAs/Ge 3-junc-
tion concentrator solar cells. In addition, we have real-
ized high-eciency and large-area (7000 cm2)
concentrator InGaP/InGaAs/Ge 3-junction solar cell
modules of an outdoor eciency of 27% as a result of
developing high-eciency InGaP/InGaAs/Ge 3-junction
cells, designing grid contact with low series resistance,
developing non-imaging Fresnel lens and 2nd optics
(homogenizers) with low optical loss, and designing
modules with low series resistance and high thermal
conductivity.
Future prospects are also presented. We have pro-
posed concentrator IIIV compound MJ solar cells as
the 3rd generation solar cells in addition to 1st genera-
Fig. 9. Projection of concentrator cell module output in tion crystalline Si solar cells and 2nd generation thin-
comparison with output projection of crystalline Si and thin lm solar cells. We are now challenging ourselves to
lm solar cell modules. develop low-cost and high output power concentrator
M. Yamaguchi et al. / Solar Energy 79 (2005) 7885 85
MJ solar cell modules with an output power of 400 W/ Hutchby, J.A., Markunas, R.J., Bedair, S.M., 1985. A review of
m2 for terrestrial applications. multijunction concentrator solar cells. In: Proceedings of the
18th IEEE Photovoltaic Specialists Conference, Las Vegas,
USA. IEEE, New York, p. 20.
Acknowledgments King, R., Fetzer, C., Colter, P., Edmondson, K., Ermer, J.,
Cotal, H., Yoon, H., Stavrides, A., Kinsey, G., Kurzt, D.,
This work is partially supported by the New Energy Karam, N., 2002. High-eciency space and terrestrial
and Industrial Technology Development Organization multijunction solar cells through bandgap control in cell
structures. In: Proceedings of the 29th IEEE Photovoltaic
as part of the New Sunshine Program under the Minis-
Specialists Conference, New Orleans, USA. IEEE, New
try of Economy, Trade and Industry, Japan. This work
York, p. 776.
is also partially supported by the Japan Ministry of Edu- Kurtz, S.R., Myers, D., Olson, J.M., 1997. Projected perfor-
cation, Culture, Sports, Science and technology as a part mance of three- and four-junction devices using GaAs and
of the Private University Academic Frontier Center Pro- GaInP. In: Proceedings of the 26th IEEE Photovoltaic
gram Super-High Eciency Photovoltaic Research Specialists Conference, Anaheim, USA. IEEE, New York,
Center. p. 875.
Olson, J.M, Kurtz, S.R., Kibbler, K.E., 1990. A 27.3% ecient
Ga0.5In0.5P/GaAs tandem solar cell. Appl. Phys. Lett. 56,
References 623.
Sugiura, H., Amano, C., Yamamoto, A., Yamaguchi, M., 1988.
Amano, C., Sugiura, H., Yamamoto, A., Yamaguchi, M., 1987. Double hetero-structure GaAs tunnel junction for an
20.2% Eciency AlGaAs/GaAs tandem solar cells grown by AlGaAs/GaAs tandem solar cell. Jpn. J. Appl. Phys. 27,
MBE. Appl. Phys. Lett. 51, 1998. 269.
Ando, H., Amano, C., Sugiura, H., Yamaguchi, M., Salates, Takamoto, T., Ikeda, E., Kurita, H., Ohmori, M., Yamaguchi,
A., 1987. Nonradiative eh recombination of mid-gap trap M., 1997a. Two-terminal monolithic In0.5Ga0.5P/GaAs tan-
in AlGaAs by MBE. Jpn. J. Appl. Phys. 26, L266. dem solar cells with a high conversion eciency of over
Araki, K., Kondo, M., Uozumi, H., Yamaguchi, M., 2003. 30%. Jpn. J. Appl. Phys. 36, 6215.
Development of a robust and high eciency concentrator Takamoto, T., Ikeda, E., Agui, T., Kurita, H., Tanabe, T.,
receiver. In: Proceedings of the 3rd World Conference on Tanaka, S., Matsubara, H., Mine, Y., Takagishi, S.,
Photovoltaic Energy Conversion, Osaka, Japan. p. 630. Yamaguchi, M., 1997b. InGaP/GaAs and InGaAs mechan-
Araki, K., Kondo, M., Uozumi, H., Ekins-Daukes, N.J., ically-stacked triple-junction solar cells. In: Proceedings of
Yamaguchi, M., in press. M., Packaging IIIV tandem the 26th IEEE Photovoltaic Specialists Conference, Ana-
solar cells for practical terrestrial applications achievable to heim, USA. IEEE, New York, p. 1031.
28% of module eciency by conventional machine assemble Takamoto, T., Yamaguchi, M., Ikeda, E., Agui, T., Kurita, H.,
technology. Solar Energy Mater. Solar Cells. Al-Jassim, M., 1999. Mechanism of Zn and Si diusion
Bett, A.W., Baur, C., Dimroth, F., Lange, G., Meusel, M., van from a highly doped tunnel junction for InGaP/GaAs
Riesen, S., Siefer, G., Andreev, V.M., Rumyantsev, V.D., tandem solar cells. J. Appl. Phys. 85, 1481.
Sadchikov, N.A., 2003. FLATCONTM-modules: technol- Takamoto, T., Agui, T., Kamimura, K., Kaneiwa, M., Imaiz-
ogy and characterisation. In: Proceedings of the 3rd World umi, M., Matsuda, S., Yamaguchi, M., 2003. Multijunction
Conference on Photovoltaic Energy Conversion, Osaka, solar cell technologies-high eciency, radiation resistance,
Japan. p. 634. and concentrator applications. In: Proceedings of the 3rd
Brown, M.R., Goldhammer, L.J., Goodelle, G.S., Lortz, C.U., World Conference on Photovoltaic Energy Conversion,
Perron, J.N., Powe, J.S., Schwartz, J.A., Cavicchi, B.T., Osaka, Japan. p. 581.
Gillanders, M.S., Krut, D.D., 1997. Characterization testing Wolf, M., 1960. Limitations and possibilities for improvement
of dual GaInP2/GaAs/Ge solar cell assemblies. In: Proceed- of photovoltaic solar energy converters. Proc. Inst. Radio
ings of the 26th IEEE Photovoltaic Specialists Conference, Engineers 48, 1246.
Anaheim, USA. IEEE, New York, p. 805. Yamaguchi, M., 2002. Present status of R&D for super-high-
Dimroth, F, Baur, C., Meusel, M., van Riesen, S., Bett, A.W., eciency IIIV compound solar cells in Japan. In: Pro-
2003. 5-Junction IIIV solar cells for space applications. In: ceedings of the 17th European Photovoltaic Solar Energy
Proceedings of the 3rd World Conference on Photovoltaic Conference, Munich, Germany. WIP, Munich, Germany, p.
Energy Conversion, Osaka, Japan. p. 616. 2144.
Fan, J.C.C., Tsaur, B-Y., Palm, B.J., 1982. Optimal design of Yamaguchi, M., Wakamatsu, S., 1996. Super-high eciency
high-eciency tandem cells. In: Proceedings of the 16th solar cell R&D program in Japan. In: Proceedings of the
IEEE Photovoltaic Specialists Conference, San Diego, 25th IEEE Photovoltaic Specialists Conference, Washing-
USA. IEEE, New York, p. 692. ton DC, USA. IEEE, New York, p. 9.
Gomyo, A., Suzuki, T., Kobayashi, K., Kawata, S., Hino, I., Yamaguchi, M., Amano, C., Sugiura, H., Yamamoto, A., 1987.
Yuasa, T., 1987. Evidence for the existence of an order state High eciency AlGaAsGaAs tandem solar cell with tunnel
in Ga0.5In0.5P grown by metalorganic vapor-phase epitaxy junction. In: Proceedings of the 19th IEEE Photovoltaic
and its relation to band-gap energy. Appl. Phys. Lett. 50, Specialists Conference, New Orleans, USA. IEEE, New
673. York, p. 1484.