Professional Documents
Culture Documents
1007/s003399900062
Applied Physics A
Materials
Science & Processing
Springer-Verlag 1999
Abstract. Research activities in the field of III-V solar cells shown efficiencies beyond 30% and are expected to reach the
are reviewed. III-V compound semiconductors are used for 40% barrier in the future.
space solar cells, concentrator solar cells, and in thermopho- This paper reports on the progress and current research
tovoltaic generators. The epitaxial growth of ternary and areas of III-V materials for photovoltaics. It is divided into
quaternary material by MOVPE and LPE allows us to re- three sections. Section 1 refers to important research aims,
alize various band gaps. Multi-junction solar cells with dif- applications and markets for III-V solar cells. Section 2 de-
ferent band gaps are necessary to obtain efficiencies larger scribes the present status of the metalorganic vapor phase
than 30%. Recent results of the III-V solar cell research at epitaxy (MOVPE) growth of III-V solar cells at Fraunhofer
the Fraunhofer ISE are presented. A mechanically stacked ISE. Results of GaAs and GaInP solar cells are presented.
GaAs/GaSb tandem concentrator solar cell achieved an ef- The third section deals with the technology of high-efficiency
ficiency of 31.1% under 100 × AM1.5d. An efficiency of mechanically stacked tandem concentrator solar cells de-
23% for a two-terminal concentrator module (486 cm2 ) with veloped at Fraunhofer ISE. Low-cost technologies were used
Fresnel lenses has been measured under realistic outdoor to produce the p-n structures. Two systems are of current in-
conditions. terest: GaAs/GaSb and Al0.26 Ga0.74 As/Si. The top cells are
produced by liquid phase epitaxy (LPE) processes. The p-n
PACS: 71.55.Eq; 81.05.Ea; 85.30.De junction of the GaSb cell is fabricated by a Zn vapor phase
diffusion technique. The efficiency of the GaAs/GaSb tan-
dem cell is 31.1%, which is the highest value for cells pro-
III-V compounds are the basic materials for modern opto- duced without metalorganic vapor phase epitaxy (MOVPE)
electronic devices. Apart from this, they often show superior technology. Special emphasis is given to the growth technolo-
properties compared to Si in the field of microelectronics, for gies and the demands for concentrator application.
example high-speed transistors.
In the field of solar cells, GaAs has higher theoretical and
practical efficiencies than Si. According to well-accepted effi-
ciency tables, GaAs has already shown an efficiency of 25.1%
compared to 24.0% in the case of Si [1]. In addition, the
radiation hardness of III-V solar cells is better. As a conse-
quence, solar cells based on III-V compound semiconductors
are becoming increasingly dominant for space applications.
On the terrestrial solar cell market the less expensive silicon
is still the most widely used material, but the application of
high-concentration systems opens new perspectives for III-V
materials in this field as well. The required area of semicon-
ductor material in these systems can be reduced by up to three
orders of magnitude compared to conventional flat modules,
so that high-efficiency III-V solar cells become cost-efficient.
In addition, the broad spectrum of ternary and quater-
nary III-V compounds, that can be grown lattice-matched on
GaAs or Ge substrates (see Fig. 1), offers a wide range of
possibilities for advanced device structures such as mono-
lithic multi-junction cells. These novel concepts have already Fig. 1. Band gap versus lattice constant for different III-V materials
120
1 Research activities in III-V compound materials – high dislocation densities due to a large lattice-mismatch
of 4%,
1.1 GaAs-based space cells – wafer bowing and cracks due to thermal mismatch.
There was a permanent progress in improving the GaAs ma-
The highest reported efficiency for a single-junction GaAs terial quality by introducing special growth techniques, such
solar cell is 25.7% (AM1.5g, 0.25 cm2 ) [2]. High efficiency as thermocyclic growth, thermocyclic annealing [15, 16] or
is one reason why markets such as solar cells for satellites strained superlattices [17, 18]. Solar cells made of MOVPE-
show an ongoing increase in the use of GaAs. The broad- grown GaAs on Si substrates achieved an efficiency of 18.3%
ening of the commercial satellite market is caused by new (AM0) [19]. Recently monolithic tandem cells of AlGaAs on
demands for telecommunication. According to estimations, active Si could achieve efficiencies of 21.2% (AM0) on active
50%–70% of the satellites under construction are equipped area [20].
with III-V solar cells [3]. In the case of space solar cells the
high radiation resistance is another major advantage of III-
V materials. In this context, two definitions of efficiencies 1.3 InP-based space cells
are commonly used in the space community. The beginning-
of-life (BOL) efficiency is measured directly after cell pro- InP is another III-V material under investigation for space
duction. The end-of-life (EOL) efficiency is defined as the applications. InP has an even higher radiation hardness com-
efficiency after the end of the satellite mission in space (5 pared to GaAs [5], however it is more expensive. Due to
to 15 years [4]). The BOL efficiency is lowered in space the higher cost, InP solar cells play a negligible role in the
due to the exposure to high-energy proton and electron flu- present space market and are not commercially available.
ences. The magnitude of the fluences depends on the alti- There might be a market for InP solar cells in the future
tude and inclination of the satellites. Irradiation experiments because the next generation of communication satellites is
using 1 MeV electrons and fluences between 3 × 1014 e/cm2 expected to be launched in the medium earth orbit (2000 to
and 1 × 1015 e/cm2 show an efficiency decrease to 75%–65% 10 000 km height) [21]. These orbits are located in the midst
for Si and 85%–76% for GaAs/Ge [5]. A high EOL effi- of the Van Allen radiation belts characterized by the highest
ciency of space cells is important for diminishing the overall radiation intensities.
system cost and weight of satellites. The weight is an issue The highest reported InP space solar cell efficiency is
because the launch cost of satellites in the low-earth orbit 19.9% (AM0) [4]. Worldwide about a dozen research groups
(up to 2000 km) is 11 000 $/kg, and up to 66 000 $/kg in the are working on the development of InP solar cells. The main
geosynchronous earth orbit [4]. topics of investigation are heteroepitaxy, radiation stability
The GaAs space solar cells are grown by metalorganic va- testing, hydrogen passivation, and optimization of the emitter
por phase epitaxy (MOVPE) on Ge substrates. Ge is nearly structure. The comparably high price of InP bulk material in-
lattice-matched to GaAs (see Fig. 1). Therefore, an epitax- hibits a homoepitaxial approach. Therefore, some heteroepi-
ial growth of GaAs with sufficient quality is possible. Ge taxial thin-film approaches are pursued to combine low-cost
substrates are less expensive compared to GaAs and exhibit substrates with the high radiation hardness of InP. Three dif-
more mechanical strength. The substrate can be thinned to ferent substrates are under investigation: GaAs, Ge, and Si.
80 µm saving weight for the main load of the satellites. The lattice-mismatch between InP and Si is 8% and 4% in
According to estimations of a Ge substrate supplier, the respect to GaAs or Ge (see Fig. 1). The most interesting con-
world-wide consumption of Ge substrates for solar cells is cept is InP on Si due to the low substrate costs. With the
more than 6000 m2 [6]. The next generation of satellites will help of GaAs or graded Inx Ga1−x As intermediate layers the
be equipped with GaInP/GaAs/Ge dual- and triple-junction number of dislocations originating from the lattice-mismatch
cells [7, 8]. The first satellite equipped with such cells was can be greatly reduced. The trap concentration due to lattice-
launched in 1997 [9]. Details of multi-junction cells are given mismatch can also be reduced from about 1014 cm−3 to about
below in Secs. 1.4 and 3. 1012 cm−3 with a hydrogen treatment [22]. The highly dislo-
Another interesting approach to increase both efficiency cated material is somewhat predamaged, so that the radiation
and radiation resistance is the use of space concentrator cells. damage is rather small. The highest efficiency of a InP/Si so-
An additional advantage of the concentrator concept is the re- lar cell is 13.0% (AM0) [23].
duction of expensive semiconductor area by a factor similar A drawback of InP is that no lattice-matched window
to the concentration ratio. Concentrator space solar cells and layer for surface passivation is available. Therefore, the emit-
modules are under development [10, 11]. ter structure is of high significance. Several structures are
investigated with very thin p+ emitters (30 nm) or p+ p emit-
ters with a total thickness of 250 nm [24], or n+ emitters
1.2 GaAs on Si substrate (25 nm) [25].
The weight and cost issue is the driving force for the research 1.4 Multi-junction cells
of GaAs growth on Si substrates. Especially in the late 80s
a lot of research work was performed in this field [12–14]. So far, only single-junction cells were discussed. In order
The main problems arising from the heteroepitaxial growth to enhance solar cell efficiencies to values higher than 26%,
are: new concepts have to be considered. The III-V semiconduc-
– antiphase domains due to a polar on a non-polar material tors crystallizing in the zincblende structure have an import-
growth, ant impact for increasing solar cell efficiencies beyond 30%:
121
the band gap of the material can be designed by growing the concentrator module technology is so far somewhat im-
ternary and quaternary alloys lattice-matched to a substrate mature. Problems related to accurate tracking [39], soil-
(see Fig. 1). ing [40], cheap and enhanced optics [41, 42], and spectrum
This allows us to develop solar cells having multiple p-n adapted multi-junction solar cells [43–45] are subjects of cur-
junctions in different semiconductor materials. In this case rent research. Prototype concentrator systems with medium
the solar spectrum can be used more efficiently. The opti- level concentration (up to 250 suns) and 20 kWp using Si
mized band gaps for triple-junction or quattro-junction solar solar cells [35] are already available. BP Solar is currently
cells have been calculated theoretically for different spec- constructing a concentrator solar power plant with 480 kWp
tra [26, 27]. For example, the efficiency limit calculated by on the Spanish island of Tenerife. A prototype module for this
detailed balance theory is 53.6% for a AM1.5d spectrum and plant was developed within the framework of the European
a four-junction device. The necessary materials should have project EUCLIDES [46]. Parabolic mirrors as concentrating
band gaps of 0.71 eV, 1.13 eV, 1.55 eV, and 2.13 eV [26]. The optics with one-axis tracking were used. This demonstrates
calculations show that a band gap of 1.0–1.1 eV is essential that concentrator technology is of large interest and may be-
for tandem- as well as for triple- and quattro-junction solar come an important technology for the future solar power
cells. From Fig. 1 one can see that there is no commonly used market.
material available which is lattice-matched to Ge or GaAs. In this respect, III-V multi-junction solar cells as de-
However, recently it was shown that the quaternary material scribed in Sect. 1.4 are attractive for high concentration lev-
GaInAsN having a band gap energy in the range of 1.0 eV can els (> 500 suns). Figure 2 shows our modeling results of
be grown lattice-matched to a GaAs substrate [28, 29]. The efficiency limits versus band gap of the top and bottom
quality of this promising material is still not sufficient and has cell. The calculations were performed assuming a 500 sun
to be improved for solar cell applications. concentration and the AM1.5d spectrum. The cell tempera-
Multiple-junction cells can be realized in two prin- ture is 315 K. Only radiative recombination is considered.
cipal ways: mechanically stacked or monolithically inte- A maximum efficiency of > 40% is expected. Practical effi-
grated. The highest reported efficiency of a monolithical ciency of 35% should be achievable in the very near future.
one-sun GaInP/GaAs tandem cell is 30.3% (AM1.5g) [30] The highest reported efficiency for monolithic GaInP/GaAs
and 26.9% (AM0) [31]. A two-terminal InP/InGaAs tan- is 30.2% (AM1.5d, 115–260 suns) [47]. InP/GaInAs three-
dem cell has achieved an efficiency of 22.2% (AM0) [32]. terminal concentrator cells achieved an efficiency of 31.8%
Monolithical GaInP/GaAs/Ge triple cells achieved 25.8%
(AM0) [8]. Another triple cell consisting of a monolithical
tandem GaInP/GaAs cell stacked on a InGaAs cell grown
on InP achieved an efficiency in the range of 33%–34%
(AM1.5g) [33]. The given efficiencies can only be some ex-
amples of the broad spectrum of the multi-junction approach.
(AM1.5d, 50 suns) [48]. Fraas et al. reported an efficiency groups [57] use different theoretical models to prove this ef-
of > 35% (AM1.5d, 100 suns) for a GaAs/GaSb mechanic- fect. Apart from the expected efficiency enhancement, the
ally stacked tandem [49]. This efficiency was corrected due current matching in monolithical multi-junction solar cells
to Sandia recalibration some years later to 32.6% (AM1.5d, can be achieved by the application of MQWs [58]. The
100 suns) [1]. change of the absorption behavior due to the introduction of
In Sect. 2 of this paper we will report on the status of con- the pseudomorphically strained thin layers into the bulk ma-
centrator tandem solar cell research at Fraunhofer ISE. We terial gives a new degree of freedom for the design of such
are investigating the mechanically stacked and monolithically cells.
integrated approach.
N (B"TCBTFOYDN
ON #4'PS#SBHHSFGMQYDN
ON (B"TCVGGFSOYDN
N (B"TTVCTUSBUFOYDN
Fig. 3. Scheme of two different TPV approaches. a Selective emitter as
radiator in combination with adapted solar cells. b Greybody radiator in
combination with filter and low-band-gap cells Fig. 4. Typical GaAs solar cell structure grown by MOVPE technology
123
ON (B"TDBQOYDN cell. Therefore, the thickness of the optimized ARCs must be
slightly changed. However, the cells shown in Fig. 6 have the
ON "M*O1QYDN same ARC leading to the differences in the EQEs.
ON (B*O1FNJUUFSOYDN The exchange of Al0.85 Ga0.15 As to GaInP as the window
ON (B*O1VOEPQFE material is clearly visible in the EQE (Fig. 6). The thick-
ness of the GaInP layer was 30 nm. The higher absorption
ON (B*O1CBTFQYDN in the direct GaInP material lowers the EQE in the range
from 450 nm to 650 nm. However, the efficiency world record
ON (B*O1#4'QYDN of 25.7% AM1.5g for a GaAs solar cell (see Table 1) was
(B"TCVGGFSQYDN achieved with a GaInP window layer for a 0.25-cm2 n-p
ON
cell [2]. In comparison, 25.1% was achieved on a 4-cm2 de-
N (B"TTVCTUSBUFQYDN vice using an Al0.85 Ga0.15 As window layer [1].
The EQE of the cells shown in Fig. 6 has a high value of
> 94% in the spectral range from 600 nm to 900 nm. Con-
Fig. 5. Typical GaInP solar cell structure grown on GaAs substrate by sidering the losses due to grid shadowing (roughly 4%) and
MOVPE technology reflectivity (< 1%), one can conclude that the internal collec-
tion efficiency is close to 100%.
The structure of the back surface field was also varied.
GaAs and GaInP structures were processed to solar cells. A highly doped layer or a Bragg reflector was used. The
Photolithography was used to define the front grid structure of Bragg reflector consists of a five Al0.3 Ga0.7 As/GaAs-stack in
the solar cells. TiPdAg (5 nm/10 nm/100 nm) contacts were order to enhance the light reflectivity near the band edge of
evaporated sequentially and annealed at 380 ◦ C for 2 min. GaAs. It was found that this Bragg reflector works very effi-
The rear contact is made of AuGe (12% Ge). The contacts ciently. Therefore, it was possible to design GaAs solar cells
are electroplated with Au. The GaAs cap layer is removed having a solar active layer (emitter and base) thickness of
on the active cell area by means of selective etching (citric < 2 µm. Such cells are very useful for space applications be-
acid + H2 O(1 : 1wt)/H2 O2 = 3 : 1). An optimized double- cause they are expected to exhibit a higher radiation hardness.
layer antireflection coating (ARC) of MgF2 /TiO2 is used to Andreev et al. have shown the benefits of a Bragg reflector
minimize the reflection losses. for a 2.5-µm-thin solar cell [59]. Table 1 gives an overview of
obtained efficiencies for different GaAs solar cell structures.
Figure 7 shows the measured EQE curves of a GaInP
2.2 Results and discussion of MOVPE-grown structures solar cell. We have introduced a nominally undoped layer
as suggested by Tsai [60]. The intrinsic layer is slightly n-
Various GaAs solar cell structures with different doped in the order of ≈ 1 × 1015 cm−3 . This layer between
Al0.85 Ga0.15 As and GaInP window layer thicknesses were the n-doped emitter and p-doped base helps to enhance the
investigated for cells with p-n and n-p polarity. Each polar- collection efficiency of minority carriers due to the extended
ity has its specific advantages and disadvantages. Figure 6 electrical field. In addition, the intrinsic layer reduces the in-
shows the measured external quantum efficiencies (EQE) of fluence of the Zn diffusion from the base into the n-doped
p-n GaAs solar cells with different window layers. The de- emitter during the growth of the subsequent layers. Figure 7
pendence on the Al0.85 Ga0.15 As thickness is clearly visible shows calculations of what part of a AM1.5g spectrum is
in the short-wavelength part of the spectrum. The thickness absorbed in which layer. It is obvious that a large part of
of the thin window layer influences the reflectivity of the the incoming light is absorbed in the 500-nm-thick intrinsic
layer. Table 1 summarizes the efficiency values for GaInP so-
Table 1. Efficiencies of GaAs and GaInP solar cells grown by MOVPE tech-
nology at Fraunhofer ISE. The used material of the window layer is given
in brackets. For comparison, the world-record efficiencies are shown
lar cells. The maximum efficiency reported in literature was 3.1.2 LPE-EBR GaAs solar cell. Figure 2 shows that
18.5% at AM1.5g [61]. The Fraunhofer ISE has achieved a GaAs/GaSb tandem concentrator cell is one of the most
14.7% (AM1.5g) so far. This efficiency is limited by the fill promising approaches to achieve maximum efficiencies. Both
factor. Further improvement of fill factor and efficiency will cells can be fabricated by relatively simple technologies. The
be achieved with different layer structures and a suitable front heteroface Al0.8 Ga0.2 As/GaAs top cell structure is grown
grid design. using the isothermal etchback regrowth LPE (LPE-EBR)
method. A conventional H2 -purified LPE system was used.
A Ga-based melt contains Al (2.4 at. %) and Zn (0.7 at. %).
3 Mechanically stacked tandem concentrator solar cells The melt is saturated with respect to arsenic at a temperature
of 850 ◦ C. After homogenisation the melt is in contact with
3.1 Experimental a GaAs substrate for 40 min at 865 ◦ C. Three main processes
take place: first, the melt etches the GaAs substrate back due
3.1.1 Al0.26 Ga0.74 As solar cell. The Al0.26 Ga0.74 As solar cell to the undersaturation of the melt. Second, Zn diffuses into
structure was produced by a LPE process. A special two-melt the n-doped GaAs substrate material forming the p-n junc-
multi-substrate crucible was developed. The temperature ver- tion, and third, a thin AlGaAs layer is isothermally grown
sus time diagram of the process is shown in Fig. 8. At first, due to the difference in the chemical potentials between sub-
a Ga-based melt with Sn (8.34 at. %), Al (0.29 at. %) and As strate and melt. More details of this process were published
(1.37 at. %) is homogenised at 820 ◦ C and saturated at a tem- elsewhere [63, 64]. In order to improve the throughput of this
perature of 810 ◦ C. The 30–50-µm-thick n-doped base layer method, special multi-wafer crucibles were developed [65].
of the solar cell structure is grown by cooling technique. In The simple structure of the solar cell is shown in Fig. 9.
order to get a better wetting, the melt is 5 ◦ C supersaturated It is noteworthy that no back surface field and a relatively
before contacting the substrate. After 15 min of an isother- thick emitter of 2 µm is used. An efficiency of 23.3% AM1.5g
mal contact a temperature ramp of 0.8 ◦ C/min is initiated. and 25.8% 100 × AM1.5d [66] was achieved with this type of
After removing the first melt at 735 ◦ C, a second Ga-based solar cell. In addition, the thick emitter is helpful in the case
with Al (1.0 at. %), Zn (0.62 at. %), As (0.21 at. %) is used to of concentrator application. The sheet resistance is reduced
Fig. 9. Scheme of the LPE-EBR heteroface Alx Ga11−x As/GaAs solar cell
Fig. 11. Efficiency versus concentration for LPE-EBR GaAs solar cells
and plays a minor role. The maximum efficiency of 25.8% at in dependence of the contact resistivity. Symbols represent measurements,
100-sun concentration is limited by the contact resistance. lines show modeling results
resistance different approaches were investigated. A simple choosing a conductive transparent ITO back contact, applying
Zn post-diffusion process was developed [71]. This process a metal grid structure on the rear side, or by decreasing the
leads to a 10 times higher doping concentration at the sur- cell diameter.
face reducing effectively the contact resistance for a given So far, no mechanically stacked Al0.26 Ga0.74 As/Si con-
metal alloy. Moreover, different metal alloys were investi- centrator cell was fabricated. However, a 60-µm-thick
gated. Figure 12 shows the specific contact resistivity of dif- Al0.26 Ga0.74 As layer was grown by LPE. The GaAs substrate
ferent alloys versus the annealing temperature. The annealing was removed and the front- and rear side were covered with
time was always 1 min. The contact resistivity was measured an optimized ARC. We used this layer as filter for measur-
using the transmission line technique as described in [72]. ing the efficiency of Si concentrator cells. A monocrystalline
The open symbols represent contacts showing Schottky be- 0.2-cm2 local back surface field (LBSF) concentrator solar
havior whereas ohmic contacts are shown by solid symbols. cell was used [73]. Under the Al0.26 Ga0.74 As filter a 7.9%
Ni, Cr, and Ti used as adhesion materials in combination with (8.9% at C = 20) efficiency was measured for the Si cell.
Au lead to contact resistivities in the range of 1 × 10−4 Ω cm2
(see Fig. 12a). The AuZn contact exhibits the best perform- 3.2.2 GaAs/GaSb tandem cell. Figure 14 shows the scheme
ance concerning the contact resisitivity but the adhesion was of a mechanically stacked GaAs/GaSb tandem concentra-
fairly bad. It turns out that CrZnAu exhibits an acceptable tor cell. An important aspect of the mechanical stacking
contact resistivity (see Fig. 12b) combined with good adhe- is the optical behavior of the top cell. Figure 15 shows
sion properties. the measured transmittance and reflectance for two types of
ARC approaches. One GaAs cell uses double-layer ARCs of
3.2 Results and discussion MgF2 /TiOx (front side: 115/65 nm; back side: 225/135 nm),
the other one uses a prismatic cover (d ≈ 300 µm, n = 1.48)
3.2.1 AlGaAs/Si tandem cell. Figure 13 shows measured and TiOx (65 nm) at the front side and MgF2 /TiOx on the
EQE curves of AlGaAs solar cells with different emitter back side. The prismatic cover reduces effectively the shad-
thicknesses. If the emitter thickness is too large (> 1 µm), the owing losses of the front grid [74]. Unfortunately, it causes
diffusion length in the p-doped emitter is not high enough to also some optical losses due to the increase of reflectance
provide good carrier collection. In this case the EQE is lower. and decrease of transmittance (see Fig. 15). However, the
If the emitter thickness becomes too thin, the light with en- prismatic cover has an additional advantage. It allows us to
ergy close to the band gap is partly absorbed in the base of the
cell structure. The minority carriers generated in the base can
diffuse towards the p-n junction or towards the substrate. The
latter carriers are lost through recombination because no back Fresnel lens
surface field is used. A 1-µm-thick emitter was determined Concentrated
as optimum for this structure. The diameter of the concen- Prismatic cover
sunlight
trator cell is 4 mm. An efficiency of 16.3% AM1.5g (Voc = + Single AR-coating
1297 mV, Isc = 14.3 mA/cm2, FF = 0.88) was obtained for Double AR coatings
a one-sun application. The maximum efficiency was 17.1% Eg = 1.42 eV GaAs
Cu
at C = 30 AM1.5g. After etching of the GaAs substrate the
Prismatic cover
maximum efficiency was 15.2% obtained at a low concentra-
+ Single AR-coating
tion level of seven suns. In this case the high resistance in the Eg = 0.72 eV Cu GaSb
30–50-µm-thick base is the main reason for power losses. No
metal contact was evaporated on the rear side of the 0.13-cm2
active solar cell area in order to achieve a high transmis- Fig. 14. Scheme of the mechanically stacked GaAs/GaSb tandem concen-
sion. The problem of resistance losses may be overcome by trator cell
Transmittance, Reflectance
Ext. Quantum Efficiency (%)
80 0,8
use a more dense grid structure which reduces the electrical out, that for a 300-µm-thick wafer and a 4-mm-diameter cell
losses. a n-doping level of 2 × 1017 cm−3 is a good compromise.
Another effect which has to be considered is the reduc- The best GaAs/GaSb tandem cell achieved an efficiency
tion of transmittance due to sub-band-gap free-carrier absorp- of 31.1% at 100 suns. The cell was measured under labora-
tion. Figure 16 shows experimental results of measured short- tory standard conditions: AM1.5d normalized at 1000 W/m2 ,
circuit currents of a GaSb cell under different GaAs substrates 25 ◦ C. This is the highest efficiency ever achieved for cells
with optimized ARCs on both sides. A strong free-carrier ab- fabricated without MOVPE technology.
sorption occurs in p-doped material. Therefore, a n-on-p solar While standard conditions are very valuable for compar-
cell is not favorable for mechanical stacking. The substrate ing cell efficiencies, it is obvious that realistic conditions are
with a n-doping of 2 × 1017 cm−3 shows only little more ab- quite different. At medium latitudes of 40–50◦ such as in
sorption than a semi-insulating GaAs material. Freiburg (Germany) the direct solar spectrum has normally
The optical losses have to be regarded in combination less than 1000 W/m2 energy flux. The ambient temperature
with the electrical behavior of the solar cell. One example and especially the cell temperature is generally higher than
is power loss due to high resistance in the base. Figure 17 25 ◦ C. Effects of wind speed are of practical interest. There-
shows calculations of losses caused by absorption and electri- fore, we decided to measure module efficiencies under realis-
cal losses in the base of a typical LPE-EBR GaAs concentra- tic outdoor conditions. We want to stress this point because it
tor cell. The calculations assume a concentration of 200 suns is important for the evaluation of the given results.
and a 300-µm-thick wafer. The base doping level and the In practice, modules with a two-terminal connection are
diameter of the solar cell was varied. The losses are given desirable. The mechanically stacked GaAs/GaSb cell has the
in relative percentage to a tandem efficiency of 31%. It turns main advantage that the cells can be connected without cur-
rent matching. On the other hand, in a tandem cell design
using monolithically integrated cells such as GaInP/GaAs
a current matching of the two cells is absolutely necessary.
15 The lower current of the two cells limits the current of the
whole tandem cell. In this case spectral changes of illumina-
tion, for example caused by weather conditions or different
J sc, GaSb (mA/cm²)
ce
18 -3
1 x 10 cm
0 Acknowledgements. The authors wish to thank the “III-V group” at Fraun-
0 5 10 15 20 hofer ISE for the excellent work, support and discussion. We gratefully
acknowledge helpful discussions with Prof. W. Wettling and the continuous
Diameter of Solar Cell (mm) interest and support of our work by Prof. Luther. The Cusanuswerk Bonn,
Germany gave financial support to one of us (G. Stollwerck). The Ger-
Fig. 17. Optical and electrical losses in the base of 200× LPE-ER GaAs man Ministry for Education, Science, Research and Technology (BMBF)
concentrator cell versus diameter of the cell. The losses are given in relative supported the work under contract number 0328554D. The authors are re-
percentage to a 31% tandem solar cell sponsible for the content of the publication.
128
References 30. T. Takamoto, E. Ikeda, H. Kurita, M. Ohmori: Appl. Phys. Lett. 70,
381 (1997)
31. T. Takamoto, M. Yamaguchi, S.J. Taylor, E. Ikeka, T. Agui, H. Kurita:
1. M.A. Green, K. Emery, K. Bücher, D.L. King, S. Igari: Progr. Photo- In Proc. 26th IEEE Photovoltaic Specialists Conference (IEEE, New
volt. 6, 35 (1998) York 1997) p. 1451
2. S.R. Kurtz, J.M. Olson, A. Kibbler: In Proc. 21st IEEE Photovoltaic 32. W.M. Wanlass, J.S. Ward, K. Emery, A. Duda, T.J. Coutts: In Proc. 1st
Specialists Conference (IEEE, New York 1990) p. 138 World Conference on Photovoltaic Energy Conversion, (IEEE, New
3. M. Meyer, R.A. Metzger: Compd. Semicond. 2, 22 (1996) York 1994) p. 1717
4. S.G. Bailey, D.J. Flood: Progr. Photovolt. 6, 1 (1998) 33. T. Takamoto, E. Ikeda, T. Agui, H. Kurita, T. Tanabe, S. Tanaka,
5. M. Yamaguchi: J. Appl. Phys. 78, 1476 (1995) H. Matsubara, Y. Mine, S. Takagishi, M. Yamaguchi: In Proc. 26th
6. M.D. Hont, P. Mijlemans, I. Moerman, P. Deemester: Compd. Semi- IEEE Photovoltaic Specialists Conference (IEEE, New York 1997)
cond. 4, 30 (1998) p. 1451
7. R.A. Metzger: Compd. Semicond. 2, 25 (1996) 34. P.E. Gregory, P.G. Borden, M.J. Ludowise, R.J. Owen, N. Kaminar,
8. N.H. Karam, J.H. Ermer, R.R. King, M. Haddad, L. Cai, D.E. Joslin, R.A. Larue, R.J. Boettcher: Solar Cells 6, 103 (1982)
D.D. Krut, M. Takahashi, J.W. Eldrege, W. Nishikawa, B.T. Cavicchi, 35. S. Yoon, V. Garboushian: In Proc. 1st World Conference on Photo-
D.R. Lillington: In Proc. 2nd World Conference on Photovoltaic So- voltaic Energy Conversion (IEEE, New York 1994) p. 1500
lar Energy Conversion, ed. by J. Schmid, H.A. Ossenbrink, P. Helm, 36. C. Algora del Valle, V.D. Luque: In Proc. 14th European Photo-
H. Ehmann, E.D. Dunlop (Joint Research Centre, European Commis- voltaic Solar Energy Conference, ed. by H.A. Ossenbrink, P. Helm,
sion, Ispra, Italy 1998) p. 3534 H. Ehmann (H.S. Stephens, Bedford, UK 1997) p. 1724
9. B.T. Cavicchi, J.H. Ermer, D.D. Krut, D.E. Joslin, M.S. Gillanders, 37. V.M. Andreev, V.P. Khvostikov, V.R. Larionov, V.D. Rumyantsev,
D.K. Zemmrich: In Proc. 2nd World Conference on Photovoltaic So- E.V. Paleeva, M.Z. Shvarts, C. Algora: In Proc. 2nd World Conference
lar Energy Conversion, ed. by J. Schmid, H.A. Ossenbrink, P. Helm, on Photovoltaic Solar Energy Conversion, ed. by J. Schmid, H.A. Os-
H. Ehmann, E.D. Dunlop (Joint Research Centre, European Commis- senbrink, P. Helm, H. Ehmann, E.D. Dunlop (Joint Research Centre,
sion, Ispra, Italy 1998) p. 3515 European Commission, Ispra, Italy 1998) p. 3719
10. V.M. Andreev, A.B. Kazantsev, V.P. Khvostikov, E.V. Paleeva, 38. M.S. Kuryla, M.L. Ristow, L.D. Partain, J.E. Bigger: In Proc. 22nd
V.D. Rumyantsev, M.Z. Shvarts: In Proc. 1st World Conference on IEEE Photovoltaic Specialists Conference (IEEE, New York 1991) p.
Photovoltaic Energy Conversion (IEEE, New York 1994) p. 2096 506
11. D.M. Murphy, M.I. Eskenazi: In Proc. 26th IEEE Photovoltaic Special- 39. J.C. Arboiro, G. Sala, J.I. Molina, L. Hernando, E. Camblor: In Proc.
ists Conference (IEEE, New York 1997) p. 861 2nd World Conference on Photovoltaic Solar Energy Conversion,
12. J.C.C. Fan, J.M. Poate: Heteroepitaxy on Silicon (Material Research ed. by J. Schmid, H.A. Ossenbrink, P. Helm, H. Ehmann, E.D. Dun-
Society, Pittsburgh, PA 1986) lop (Joint Research Centre, European Commission, Ispra, Italy 1998)
13. H.K. Choi, R. Hull, H. Ishiwara, R.J. Nemanich: Heteroepitaxy on Sil- p. 2229
icon, Fundamentals, Structure, and Devices (Material Research Soci- 40. R. Hammond, D. Srinivasan, A. Harris, K. Whitfield: In Proc. 14th
ety, Pittsburgh, PA 1988) European Photovoltaic Solar Energy Conference, ed. by H.A. Ossen-
14. S.F. Fang, K. Adomi, S. Iyer, H. Morkoc, H. Zabel, C. Choi, N. Otsuka: brink, P. Helm, H. Ehmann (H.S. Stephens, Bedford, UK 1997) p. 301
J. Appl. Phys. 68, R31 (1990) 41. P. Benitez, J.L. Diaz: In Proc. 2nd World Conference on Photovoltaic
15. M. Yamaguchi, A. Yamamoto, M. Tachikawa, Y. Itho, M. Sugo: Appl. Solar Energy Conversion, ed. by J. Schmid, H.A. Ossenbrink, P. Helm,
Phys. Lett. 53, 2293 (1988) H. Ehmann, E.D. Dunlop (Joint Research Centre, European Commis-
16. M. Yamaguchi, M. Tachikawa, Y. Itho, M. Sugo, S. Kondo: J. Appl. sion, Ispra, Italy 1998) p. 2237
Phys. 68, 4518 (1990) 42. J.L. Alvarez, M. Hernandez, P. Benitez, J.C. Minano: In Proc. 2nd
17. M. Yamaguchi, T. Nishioka, M. Sugo: Appl. Phys. Lett. 54, 24 (1989) World Conference on Photovoltaic Solar Energy Conversion, ed. by
18. K. Zieger, G. Hahn, T. Forner, J.S. Im, A. Hangleiter, D. Haase, J. Schmid, H.A. Ossenbrink, P. Helm, H. Ehmann, E.D. Dunlop (Joint
A. Dörnen, F. Phillipp, F. Scholz, V. Frese, J. Hilgarth, M. Braun: In Research Centre, European Commission, Ispra, Italy 1998) p. 2233
Proc. 13th European Photovoltaic Solar Energy Conference, ed. by 43. P. Faine, S.R. Kurtz, C. Riordan, J.M. Olson: Solar Cells 31, 259
W. Freiersleben, W. Palz, H.A. Ossenbrink, P. Helm (H.S. Stephens, (1991)
Bedford, UK 1995) p. 81 44. S.R. Kurtz, D.J. Friedman, J.M. Olson: In Proc. 1st World Conference
19. Y. Ohmachi, T. Ohara, Y. Kadota: In Proc. 21st IEEE Photovoltaic Spe- on Photovoltaic Energy Conversion (IEEE, New York 1994) p. 2402
cialists Conference (IEEE, New York 1990) p. 89 45. L.W. James: In Proc. 1st World Conference on Photovoltaic Energy
20. M. Umeno, T. Soga, K. Baskar, T. Jimbo: Solar Energy Mater. Solar Conversion (IEEE, New York 1994) p. 2402
Cells 50, 203 (1998) 46. A. Luque, G. Sala, J.C. Arboiro, T. Bruton, D. Cunningham, N. Mason:
21. R.A. Metzger: Compd. Semicond. 2, 25 (1996) Progr. Photovolt. 5 195 (1997)
22. B. Chatterjee, S.A. Ringel, R. Hoffman: Prog. Photovolt. Res. Appl. 4, 47. D.J. Friedman, S.R. Kurtz, K.A. Bertness, A.E. Kibbler, C. Kramer,
91 (1996) J.M. Olson, D.L. King, B.R. Hansen, J.K. Snyder: Progr. Photovolt. 3,
23. S.J. Wojtczuk, N.H. Karam, P. Gouker, P. Colter, S.M. Vernon, 47 (1995)
G.P. Summers, R.L. Walters, R. Statler: In Proc. 1st World Conference 48. M.W. Wanlass, T.J. Coutts, J.S. Ward, K.A. Emery, T.A. Gessert,
on Photovoltaic Energy Conversion (IEEE, New York 1994) p. 1705 C.R. Osterwald: In Proc. 22nd IEEE Photovoltaic Specialists Confer-
24. R. Hoffman, N.S. Fatemi, P.P. Jenkins, V.G. Weizer, M.A. Stan, ence (IEEE, New York 1991) p. 38
S.A. Ringel, D.A. Scheiman, D.M. Wilt, D.J. Brinker, R.J. Walters, 49. L.M. Fraas, J.E. Avery, V.S. Sundaram, V.T. Dinh, T.M. Davenport,
S.R. Messenger: In Proc. 26th IEEE Photovoltaic Specialists Confer- J.W. Yerkes, J.M. Gee, K.A. Emery: In Proc. 21st IEEE Photovoltaic
ence (IEEE, New York 1997) p. 815 Specialists Conference (IEEE, New York 1990) p. 190
25. R.J. Walters, M.J. Romero-Florez, D. Araujo, R. Garcia, S.R. Messen- 50. T.J. Coutts, J.P. Benner: The First NREL Conference on Thermophoto-
ger, G.P. Summers: In Proc. 2nd World Conference on Photovoltaic voltaic Generation of Electricity (AIP321, New York 1994)
Solar Energy Conversion, ed. by J. Schmid, H.A. Ossenbrink, P. Helm, 51. T.J. Coutts, J.P. Benner: The Second NREL Conference on Thermopho-
H. Ehmann, E.D. Dunlop (Joint Research Centre, European Commis- tovoltaic Generation of Electricity (AIP358, New York 1995)
sion, Ispra, Italy 1998) p. 3550 52. T.J. Coutts, C.S. Allman, J.P. Benner: Thermophotovoltaic Generation
26. A. Marti, G. L. Araújo: Solar Energy Mater. Solar Cells 43, 203 (1996) of Electricity, Third NREL Conference (AIP401, Woodbury New York
27. S. Kurtz: In Proc. 26th IEEE Photovoltaic Specialists Conference 1997)
(IEEE, New York 1997) p. 875 53. T.J. Coutts, J.P. Benner: Thermophotovoltaic Generation of Electricity,
28. M. Kondow, T. Kitatani, S. Nakatsuka, M.C. Larson, K. Nakahara, 4th NREL Conference (AIP460, Woodbury New York, 1998)
Y. Yazawa, M. Okai, K. Uomi: IEEE J. Sel. Top. Quantum Electron. 54. K.W.J. Barnham, G. Duggan: J. Appl. Phys. 67, 3490 (1990)
3, 719 (1997) 55. A. L. Araújo, A. Marti, F.W. Ragay, J.H. Wolter: In Proc. 12th Euro-
29. D.J. Friedman, J.F. Geisz, S.R. Kurtz, J.M. Olson: In Proc. 2nd pean Photovoltaic Solar Energy Conference, ed. by R. Hill, W. Palz,
World Conference on Photovoltaic Solar Energy Conversion, ed. by P. Helm, (H.S. Stephens, Bedford, UK 1994) p. 1429
J. Schmid, H.A. Ossenbrink, P. Helm, H. Ehmann, E.D. Dunlop (Joint 56. F.W. Ragay, A. Marti, G.L. Araujo, J.H. Wolter: Solar Energy Mater.
Research Centre, European Commission, Ispra, Italy 1998) p. 3 Solar Cells 40, 5 (1996)
129
57. S. Kettenmann, J.F. Guillemoldes: In Proc. 13th European Photovoltaic 67. A.W. Bett, S. Keser, O.V. Sulima: J. Cryst. Growth 181, 181 (1997)
Solar Energy Conference, ed. by W. Freiersleben, W. Palz, H.A. Os- 68. A.W. Bett, S. Keser, G. Stollwerck, O.V. Sulima: In Proc. Thermopho-
senbrink, P. Helm (H.S. Stephens, Bedford, UK 1995) p. 119 tovoltaic Generation of Electricity, Third NREL Conference, ed. by
58. A. Freundlich, I. Serdiukova: In Proc. 2nd World Conference on Pho- T.J. Coutts, C.S. Allman, J.P. Benner (AIP, Woodbury New York 1997)
tovoltaic Solar Energy Conversion, ed. by J. Schmid, H.A. Ossenbrink, p. 41
P. Helm, H. Ehmann, E.D. Dunlop (Joint Research Centre, European 69. U. Blieske, A. Baldus, A. Bett, F. Lutz, T. Nguyen, C. Schetter,
Commission, Ispra, Italy 1998) p. 3707 K. Schitterer, O.V. Sulima, W. Wettling: In Proc. 23rd IEEE Photo-
59. V.M. Andreev, O.I. Chosta, V.P. Khvostikov, V.M. Lantratov, voltaic Specialists Conference (IEEE, New York 1993) p. 1490
E.V. Paleeva, M.Z. Shvarts: In Proc. 14th European Photovoltaic So- 70. U. Blieske, A. Bett, T. Duong, C. Schetter, O.V. Sulima: In Proc. 12th
lar Energy Conference, ed. by H.A. Ossenbrink, P. Helm, H. Ehmann, European Photovoltaic Solar Energy Conference, ed. by R. Hill,
(H.S. Stephens, Bedford, UK 1997) p. 1744 W. Palz, P. Helm (H.S. Stephens, Bedford, UK 1994) p. 1409
60. C.J. Tsai, W.H. Bloss, K. Zieger, F. Scholz, V. Frese, U. Blieske: In 71. A. Blug, A. Baldus, A. Bett, U. Blieske, G. Stollwerck, O. Sulima,
Proc. 13th European Photovoltaic Solar Energy Conference, ed. by W. Wettling: In Proc. 13th European Photovoltaic Solar Energy Con-
W. Freiersleben, W. Palz, H.A. Ossenbrink, P. Helm (H.S. Stephens, ference, ed. by W. Freiersleben, W. Palz, H.A. Ossenbrink, P. Helm
Bedford, UK 1995) p. 918 (H.S. Stephens, Bedford, UK 1995) p. 910
61. T. Takamoto, E. Ikeda, H. Kurita, M. Ohmori: In Proc. 1st World Con- 72. G.K. Reeves, H.B. Harrison: IEEE Electron. Dev. Lett. EDL-3, 111
ference on Photovoltaic Energy Conversion (IEEE, New York 1994) (1982)
p. 2402 73. S. Sterk, A. Schönecker, K. Schitterer, W. Wettling: In Proc. 12th Eu-
62. F. Dimroth, A.W. Bett, W. Wettling: J. Cryst. Growth 179, 41 (1997) ropean Photovoltaic Solar Energy Conference, ed. by R. Hill, W. Palz,
63. J.M. Woodall, H.J. Hovel: Appl. Phys. Lett. 30, 492 (1977) P. Helm, (H.S. Stephens, Bedford, UK 1994) p. 527
64. A. Baldus, A.W. Bett, U. Blieske, O.V. Sulima, W. Wettling: J. Cryst. 74. M. J. O’Neill: In Proc. 19th IEEE Photovoltaic Specialists Conference
Growth 146, 299 (1995) (IEEE, New York 1987) p. 1402
65. A. Baldus, A. Bett, U. Blieske, O.V. Sulima, W. Wettling: In Proc. 75. A.W. Bett, G. Stollwerck, O.V. Sulima, W. Wettling: In Proc. 2nd
12th European Photovoltaic Solar Energy Conference, ed. by R. Hill, World Conference on Photovoltaic Solar Energy Conversion, ed. by
W. Palz, P. Helm (H.S. Stephens, Bedford, UK 1994) p. 1485 J. Schmid, H.A. Ossenbrink, P. Helm, H. Ehmann, E.D. Dunlop (Joint
66. A.W. Bett, S. Keser, G. Stollwerck, O.V. Sulima, W. Wettling: In Proc. Research Centre, European Commission, Ispra, Italy 1998) p. 268
26th IEEE Photovoltaic Specialists Conference (IEEE, New York
1997) p. 1451