Bottom cell growth aspects
for triple junction InGaP(In)GaAsGe1043_a
三接面 InGaP / (In)GaAs / Ge底部型太陽能電池成長
指導教授: 李洋憲
Reporter:自控三甲 69712007尤振猛
Bottom cell growth aspects for triple junction
InGaP/(In)GaAs/Ge solar cells
三接面 InGaP / (In)GaAs / Ge底部型太陽能電池成長
G. Timò*, C. Flores, and R. Campesato CESI S.p.A. Via Rubattino No. 54 , 20134
Milano, Italy
Received 13 December 2004, revised 15 March 2005, accepted 6 April 2005
Published online 15 September 2005
Key words MOVPE, growth, solar cells, multijunction, GaAs/Ge, AlGaAs/Ge,
InGaP/Ge, Ge bottom cell. PACS 81.10.-h
The paper discusses the problems of nucleation layer and
substrate specification selection for a bottom Ge cell
performance. GaAs/Ge, AlGaAs/Ge and InGaP/Ge
heterojuctions have been compared showing how lattice
matching and dopant interdiffusion control are key aspects
for improving the Ge bottom cell photovoltaic response. The
influence of substrates orientation, polarity and resistivity
on the electrical performances of the bottom cells are
響的問題。GaAs / Ge、AlGaAs / Ge和InGaP / Ge的異質接面已
1 Introduction
Multi-junctions (MJ) solar cells made from elements of the groups III-V and IV
of the periodic table have been considered by the photovoltaic community for
their quasi-ideal energy gap combination which allows the achievement of
high solar conversion efficiency [1]. Whereas U.S companies have already
showed excellent results producing solar cells reaching AMO efficiency of
29% [2], Europeans suffer of a delay in the development of such a devices and
are trying to shorten the technology gap in the frame of European Space
Agency programs [3,4].
CESI participating these programs, has been working in the realisation of
dual and triple junction (TJ) devices for some years by using AIXTRON
Planetary rectors and VEECO vertical reactor technologies (see figure 1),
facing the related problematic growth aspects and potentiality of the
MOCVD systems. We found, for example, that the control of temperature
and thickness uniformity has to be carefully taken into account to grow
high effciency TJs and a such control is indeed possible even on these
production MOCVD systems, in spite of the big size of their reactor cambers.
A typical N on P TJ structure, shown in figure 2, is made of several layers,
which are deposited to form three junctions (“bottom”, “middle” and the “top”
cell) interconnected by two tunnel diodes. The choice of the solar cell
materials is the result of a compromise between the demand of optimising
light collection and the need of using materials with similar lattice constants
and thermal expansion coefficients. A good compromise is accomplished by
selecting Ge, (In) GaAs and InGaP semiconductor material for the bottom,
middle and top cell, respectively.
互相連接三接面 (底部、中間和頂部電池)。太陽能電池的材料選擇要在最佳
Ge bottom cell, however, has to be optimised by selecting a proper
“nucleation layer”, that is a suitable buffer material to be grown between the
substrate and the active region of the device. The nucleation layer plays a
fundamental role in the formation of the bottom cell junction and it has the
function of passivating its surface [5]. In this paper we focalise our attention
on the growth GaAs, AlGaAs and InGaP semiconductors, grown by different
reactors, as nucleation layer candidates for bottom Ge cells. The influence of
substrate orientation, polarity and resistivity on the electrical performances
of the bottom cells will be showed as well.
2 Experiment
Growth of MJ structures has been carried out by using a Planetary reactor
AIX2400 and a vertical VEECO 450 gold reactor. TMGa, TMIn, TMAl, DMZn,
AsH3, PH3 and Si2H6 have been used as precursors. Growth pressure in the
AIXTRON and VEECO system was changed between 50 and 150 torr. Growth
temperature was investigated in the range 650-750°C. Growth rate of GaAs and
AlGaAs have been varied between 2 and 10μm/h, while for InGaP, it was
maintained fixed around 1.8 μm/h. The solar cell samples were grown on Ge
substrates 150 μm thick, with disorientation of 6°, 9° off from [100] to the
nearest [111]. The surfacemorphology of the samples was characterised by
optical microscopy while in-depth doping profiles were measured by a WEP
Electrochemical CV-profiler CVP 21. Photovoltaic performance was measured
by using a single source WACOM solar simulator.
藉由使用AIX2400與VEECO 450成長多接面結構。TMGa、TMIn、TMAl、
VEECO系統都是在50~150 torr。成長溫度介於650~750度。GaAs與AlGaAs的
長晶速率在2~10 um/hr之間,InGaP的長晶速率維持在1.8 um/hr。此太陽能電
池樣本是成長在6度、9度且150 um厚的鍺基板上。樣本的表面型態利用光學顯
微鏡做分析。摻雜分佈利用WEP electrochemical CV-profiler CVP21量測。光
電特性使用single source WACOM solar simulator做量測。
Fig. 1 (a) View of the AIXTRON 2400
growth chamber. (b) View of the
VEECOGOLD 450 Reactor.
圖一 AIXTRON2400長晶機台外觀
圖二 Veeco450機台外觀
Fig. 2 Triple junction
InGaP/(In)GaAs/Ge solar cell structure
with N on P polarity.
圖二 三接面InGaP/(In)GaAs/Ge N
on P太陽能電池結構
3 Results
The Ge junction can be obtained by atoms solid state diffusion from the nucleation layer to
the Ge substrate.Depending on the polarity of the TJ device different solutions can be
considered. We have analysed GaAs/Ge, AlGaAs/Ge and InGaP/Ge heterojunctions (HJs),
introducing Zn, P, and As as dopants in order to compare the resulting growth morphologies
and the electrical performances of the Ge bottom cells. In figure 3 the main results obtained
on surface morpologies for the different HJs are shown. GaAs/Ge HJ presents misfit
dislocation (MDs) due to the different lattice and thermal expansion coefficient of the two
semiconductors, while the growth of lattice matched ternary materials on Ge can be
complicated by roughness problem related to the substrate polarity and preparation. In the
case of AlGaAs/Ge and InGaP/Ge HJs we found in fact that the amount of arsine introduced
in the reactor before growing and the exposure time of the substrate to this atmosphere is
crucial for preparing the Ge surface and then improving the morphology. Further the
substrate preparation condition are dependent on the polarity of the substrate.
In the case of GaAs/Ge HJ, we found in the past that the open circuit
voltage of the cells is related to the MDs density [see again 5]. In this
paper, we report how the GaAs(N)/Ge(N) HJ photovoltaic behaviour is
also a function of the temperature profile over the wafer. In the AIX2400
reactor, by regulating the current supply of the infrared lamps installed
underneath the susceptor, we could select two different temperature
profiles over the wafer and check the photovoltaic behaviour of a GaAs
single junction solar cell structure on GaAs/Ge HJ. In particular we
measured the open circuit voltage (Voc) distribution, sectioning out of
each wafer eigh (2x4 cm2) solar cells. The experiment was performed
selecting the “temperature growth window” over which the photovoltage
produced by the GaAs/Ge HJ is stronger [6]. The results are reported in
figure 4.
Fig. 3 Optical microscope morphology of GaAs/Ge,
AlGaAs/Ge and InGaP/Ge HJs. The improvement of
surface morphology owing to different substrate
preparation is showed. Growths were performed on
VEECO 450 reactor. See text for details.
圖三 GaAs/Ge, AlGaAs/Ge and InGaP/Ge 異質接面的光學顯微
Fig. 4 Effect of the growth temperature and temperature profile over the wafer on the open circuit voltage
value of GaAs single junction solar cell structure grown on GaAs/Ge HJ. a) Average Voc measured from
the 8 cells out of each wafers in function of the growth temperature; b) Temperature distribution over the
wafer obtained with a different infrared lamps setting; c) and d) Voc distribution obtained with the different
temperature profile; e) solar cell structure.
下的平均開路電壓 (每個晶圓取八個點) (b)不同紅外線燈管設定對應晶圓的溫度分佈
(c)(d)不同溫度分佈下的開路電壓分佈 (e)太陽能電池結構
The different Voc values obtained by using the two temperature profiles agree with the
thermal distribution over the wafer: higher voltages are found with the temperature profile
No.8 which presents higher temperature values all over the wafer. It is possible to conclude
that for GaAs (N)/Ge (N) HJs, both temperature and MDs distribution over the wafer
influence the uniformity of the photovoltaic performances. Similar results have been
obtained by growing the GaAs/Ge HJs with the Veeco reactor, at different growth pressures.
We have just to point out that the temperature window over which the Ge cell is more active
can be shifted to lower or higher temperatures depending on the reactor type. Since, even
with uniform temperature profile the presence of uniformly distributed MDs would limit the
GaAs/Ge HJ photovoltage, it would be useful, to consider lattice matched material on Ge
substrates, in order to reduce the recombination velocity at the Ge surface. In the case of
MJ with P on N polarity we have proposed a new AlGaAs/Ge lattice matched HJ [7], while for
N on P devices the InGaP/Ge HJ seems the preferred one.
料是有益的,可減少在鍺基板界面的再結合速度。我們提出在P on N多接面上新
的AlGaAs/Ge晶格匹配的異質接面,同時N on P的InGaP/Ge異質接面元件似乎
Infact, even if both AlGaAs and InGaP can be grown lattice matched on Ge,
AlGaAs presents the advantage with respect to InGaP to allow a better
composition control and it is less sensitive to temperature variation. Thus, for
the P on N polarity, AlGaAs should be used, while for the N on P polarity, an
analysis of the atoms interdiffusion confirms that InGaP is a more suitable
ternary material than AlGaAs. We have carried out an easy study to show the
strong differences in the diffusion profile among Zn, As, and P which are Ptype and N-type dopants in Ge [8].
更佳的組成控制與對溫度變化較不敏感。因此對於P on N應該使用AlGaAs,同
時對於N on P,原子相互擴散分析證實InGa更適合當第三材料。我們已經完成
The simulation of the diffusion profile has been performed by utilising the
“thin-film, constant-surface concentration solution” of the Fick’s second law
[9], utilising as concentration value at the Ge surface, the solid solubility
values of As and P, and in the case of Zn diffusion, the doping level
introduced in the ternary layer grown on Ge. The diffusion coefficient was
collected from S.M.Sze [10], whereas the solid solubility values were taken
from F.A.Trumbore [11]. The calculated diffusion profiles at the growth
temperature of 700 °C are reported in figure 5.
圖五 Zn、P、As在Ge中的計算擴
圖六 異質接面結構的極性子分佈
Since the bottom Ge cell efficiency decreases as the thickness of the emitter
increases, mainly owing to the lowering of the short circuit current [12], we expect
to find a stronger degradation of the bottom cell performances when faster
diffuser are utilised, that is, when arsenic is used instead of phosphorous for N on
P polarity. Since Zn is the slowest dopant among those considered, we can also
conclude that the P on N polarity HJ should be more stable than N on P one. The
Ge diffusion into the nucleation layer has to be considered as well. If, for example,
InGaP(P)/Ge(N,P) or AlGaAs(P)/Ge(P,N) HJs are grown, high Zn doping levels have
to be used in the ternary layers in order to avoid any degradation of the device
efficiency. The reason of worsening of the HJ electrical performances is the
polarity inversion which can occur in the ternary layer, owing to Ge autodoping
from the substrate (see figure 6). The influence of substrate orientation and
resistivity is showed for the selected InGaP (N)/Ge(P) and AlGaAs (P)/Ge (N) HJs
in table 1.
快速的擴散被利用時,我們預期發現更強的底部電池性能劣化,當N on P的P被
As取代時。因為Zn是考慮的摻雜物中最慢的,我們也可以推斷P on N應該比N
on P異質接面更穩定。鍺擴散進入成核層同時也被考慮。例如InGaP(P)/Ge(N,P)
表格一 不同基板位向與電阻率下未鍍膜異質接面電性總整理
The higher short circuit current values found in the InGaP/Ge HJs with respect to
the AlGaAs/Ge HJ are probably due to the better interface between the P-base
ternary material and Ge; however, these values are reduced when the remaining
materials of the TJ structure are grown, because they have lower energy gap and
produce strong light absorption. The following figures has to be considered:
InGaP/Ge Isc is 3% reduced when the AlGaAs/GaAs tunnel diode is subsequently
grown, and decreased by 19% when another InGaAs layer 1 μm thick is added.
When high resistivity substrates are used, the built-in voltage of the Ge junction is
lower, consequently the Voc is reduced. High resistivity substrates give rise better
results in the case of P on N polarity with respect to the N on P because of the
higher value of the lifetime in N-type material. Finally higher substrate orientation
improves the surface morphology, the quality of the InGaP/Ge interface, lowering
the surface recombination velocity of the Ge sub-cell, and in turn, raising the Voc
和產生較強的光吸收。之後的圖必須被考慮:InGaP/Ge Isc是3%,這個數值減
電阻率的基板在P on N會比N or P上具有更佳的效果,因為N型材料具有較高的
3 Conclusion
Lattice matching and dopant interdiffusion control along with substrate resistivity
and polarity are key aspects in selecting a proper HJ for improving the Ge bottom
cell photovoltaic response. Theoretical analisis of dopants diffussion shows that
P on N HJs should degrades less than N on P ones when subjected to the thermal
load produced by the subsequent MJ cell structure growth. However since asgrown InGaP (N)/Ge (P) HJs shows better short circuit currents than AlGaAs
(P)/Ge (N) ones the two selected HJs should shows similar photovoltaic
performances after the entire growth cycle of the triple junction cell.
的理論分析顯示P on N異質接面應該比N on P劣化更慢。然而InGaP(N)/Ge(P)
[1] S. R. Kurtz, P. Faine, and J. M. Olson, J. Appl. Phys. 68, 1890 (1990).
[2] M. A. Stan, D. J. Aiken, P. R. Sharps, N. S. Fatemi, F. A. Spadafora, and J. Hills, Twentyninth IEEE Photovoltaic Spec. Conf. , May 19-24, 2002, pp 816-819.
[3] W. Bensch, J. Hilgarth, J. Kunze, G. La Roche, K. D. Bogus, and C. Signorini, 16th
European Photovoltaic Solar Energy Conf. 1-5 May 2000, pp 935-938.
[4] G. Strobl, J. Hilgarth, M. Nell, R. Dietrich, R. Kern. W. Kostler, A. W. Bett, F. Dimroth, U.
Schubert, C. Flores, R. Campesato, G. Timò, G. La Roche, C. Signorini, and K. Bogus, 6th
Eurpean Space Power Conf. 6-10 May 2002 pp 539-543.
[5] G. Timò, C. Flores, R. Campesato, D. Passoni, and B. Bollani, Mat. Sci. Forum 203, 97
[6] J. C. Chen, M. Ladle Ristow, J. I. Cubbage, and J. G. Werthen, Appl. Phys. Lett. 58, 2282
[7] G. Timò, C. Flores, R. Campesato, G. Smekens, and J. Vanbegin, PV in Europe - 711/10/02 Rome.
[8] R. A. Smith, Semiconductors, Cambridge University Press 1978, p 425.
[9] Brian Tuck, Atomic diffusion in III-V semiconductors, Adam Hilger, Bristol and
Philadelphia, 1988, p. 18.
[10] S. M. Sze, Physics of semiconductor devices, p. 68.
[11] F. A. Trumbore, The Bell system technical journal, January 1960, pp 205-230.
[12] Flavio Iurato, Thesis: “Ottimizzazione di celle solari al GaAs per missioni sulla
superficie di Marte”, Anno accademico 2003-2004- Politecnico di Milano, Facoltà di
Ingegneria industriale, Corso di laurea in Ingegneria Aerospaziale.

similar documents