You are on page 1of 8

Solar Energy Materials & Solar Cells 74 (2002) 365–372

A-Si:H buffer in a-SiGe:H solar cells


D. Lundszien*, F. Finger, H. Wagner
.
Institute of Photovoltaics, Forschungszentrum Julich .
GmbH, D-52425 Julich, Germany

Abstract

Profiled a-SiGe:H-buffer layers between the doped and the absorption layers of amorphous
silicon germanium (a-SiGe:H) solar cells are routinely used to avoid bandgap discontinuities
and high-defect densities at the p/i- and i/n interface. Here, we present a much simpler
approach replacing the profiled a-SiGe:H-buffer layers at both interfaces by a-Si:H-buffer
layers. It is demonstrated that for a-SiGe:H solar cells (thickness of the EG ¼ 1:5 eV part is
54 nm) these structures yield similar open circuit voltage VOC and fill factor (FF) compared to
the bandgap profiled layer at the same short circuit current density jSC : The influence of
thickness, optical bandgap and position of the buffer layers on the solar cell performance is
investigated. r 2002 Elsevier Science B.V. All rights reserved.

Keywords: Thin film solar cells; a-SiGe:H; a-Si:H; Bandgap profiling; Buffer

1. Introduction

For amorphous silicon (a-Si:H) based stacked solar cells the classical red absorber
is amorphous silicon germanium (a-SiGe:H), which has a lower optical bandgap
(depending on the Ge concentration) as compared with a-Si:H. These solar cells in
p–i–n (or n–i–p) configuration usually contain a-Si:H p- and n layers. Consequently,
there is a bandgap discontinuity at the p/i- and i/n interface. The distribution of these
discontinuities between conduction and valence band is still a matter of debate.
However, since the defect density of a-SiGe:H increases with decreasing optical
bandgap (i.e. increasing Ge content), one has to expect a high-defect density at the
p/i- and i/n interfaces which will adversely affect the internal electric field and the
carrier collection and which results in poor open circuit voltages VOC and FFs.

*Corresponding author. Tel.: +49-2461-61-3932; fax: +49-2461-61-3735.


E-mail address: d.lundszien@fz-juelich.de (D. Lundszien).

0927-0248/02/$ - see front matter r 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 7 - 0 2 4 8 ( 0 2 ) 0 0 0 9 6 - X
366 D. Lundszien et al. / Solar Energy Materials & Solar Cells 74 (2002) 365–372

Therefore, considerable effort has been made to counteract these effects by smooth
bandgap grading at the p/i- and i/n interface and in fact even throughout the entire
a-SiGe:H absorber layer [1–5]. Most literature reports propose that such a bandgap
grading of the entire absorber layer is a prerequisite for high-efficiency solar cells. In
the present study, we will present an alternative approach where only the region
within 15 nm to the p/i- and i/n interface is modified while the intrinsic a-SiGe:H
absorber layer is kept at a constant bandgap (1.5 eV). The influence of thickness,
optical bandgap and position of the buffer layers on the solar cell performance is
described. It will be shown that the profiled layers at the p/i- and i/n interface can be
replaced by a-Si:H-buffer layers without any loss in FF and VOC and that the
bandgap profiling of the entire absorber layer is not necessary to achieve high solar
cell performance.

2. Experiment

All cells were deposited in a multichamber UHV glow discharge system with diode
type electrode configuration and the substrate located at the unpowered electrode
(100 cm2; 2 cm electrode spacing). Si2H6, GeH4 and H2 are used as process gases. The
deposition conditions were: pressure 0.7 Torr, power density 35 mW/cm2, substrate
temperature 2001C. All cells were deposited on textured SnO2 (ASAHI, type ‘‘U’’)
and had Ag back reflectors. Cell area is 1 cm2. The a-SiGe:H i-layer has a constant
bandgap of EG ¼ 1:5 eV. This cell structure was not optimized to deliver high-
current densities because we used only a thin a-SiGe:H absorber layer of
54 nm thickness. Profiled and a-Si:H buffer with different thicknesses were applied.
The general solar cell structure for all cells of this study is sketched in Fig. 1. The
details of the buffer shapes are shown in the following diagrams together with the
results.
The J-V parameters of the cells were measured under red light using a
590 nm cut-on filter to simulate the light exposure of the bottom cell in a tandem
stack.

interface regions
(0-15 nm)
Glass/SnO2

p n Ag
AM1.5 a-SiGe:H
+ og590 EG = 1.5 eV

Fig. 1. General solar cell structure for all cells used of this study. More detailed information about the
buffer shape can be found in Figs. 2–5.
D. Lundszien et al. / Solar Energy Materials & Solar Cells 74 (2002) 365–372 367

3. Results

3.1. Comparison between profiled and a-Si:H buffer

In a first experiment the profiled a-SiGe:H buffer were replaced by a-Si:H-buffer


layers at the p/i interface, at the i/n interface and finally at both interfaces. The buffer
layer thickness is 9 nm. The thickness of the a-SiGe:H absorber layer was kept
constant at 54 nm except for cell no. 5 with slighly thicker a-SiGe:H layer to obtain
the same short circuit current density. The results are summarized in Table 1.
It is seen that the profiled a-SiGe:H buffer can be replaced by a simple a-Si:H
buffer on both interfaces with improved FF and VOC (compare cell no. 1 with 4). On
the other hand, solar cell no. 1 exhibits the highest jSC due to the increased
incorporation of germanium into the profiled a-SiGe:H buffer which has to be
compensated by an increase of the i-layer absorber thickness (for cell no. 5).
In a second experiment, we compare cell structures with (a) profiled buffer (like
cell no. 1) and (b) two a-Si:H buffers (like cell no. 4) using various thicknesses for the
a-Si:H-buffer layers. In both cases the thickness of the two buffer layers was
simultaneously increased from 0 to 12 nm yielding a symmetric buffer shape. To
achieve similar current densities jSC for cases (a) and (b), the i-layer thickness had to
be adjusted to take care of the increased absorption in the profiled a-SiGe:H buffers
which contain a-SiGe:H.
The results from J2V measurements are shown in Fig. 2. We see that the
adjustment of the i-layer thickness between cases (a) and (b) was successful: very
similar currents densities are obtained for all buffer layer thicknesses. The VOC shows
no difference between profiled and nonprofiled buffer shape upon changing the
buffer thickness. VOC increases from 625 mV (without buffer) to 690 mV using a
buffer thickness of about 12 nm.
For the FF we observe a pronounced difference. While the FF increases for
a-Si:H-buffer layer thickness up to p/i=i/n=10 nm, for the profiled a-SiGe:H-buffer
shape the FF first remains at a low level up to 3 nm buffer thickness. Between 3 and
10 nm the FF Ist nearly 2% (absolute) higher for the a-Si:H buffer compared to the
profiled a-SiGe:H buffer. Above 10 nm the FF for the cell structure with a-Si:H

Table 1
A-SiGe:H solar cells with different buffer shapes

No. Shape FF(%) VOC (mV) JSC (mA/cm2)

1 p/54 nm/p 69.0 685 6.4


2 b/54 nm/p 68.0 680 6.2
3 p/54 nm/b 70.0 680 6.1
4 b/54 nm/b 70.5 690 5.9
5 b/57 nm/b 70.0 685 6.1

All cells are measured with AM 1.5 and a 590 nm cut-on filter. b=a-Si:H buffer; p=a-SiGe:H profiling; xx
nm=thickness of Egmin ¼ 1:5 eV.
368 D. Lundszien et al. / Solar Energy Materials & Solar Cells 74 (2002) 365–372

72
3.0

Pmax [mW/cm ]
70

2
p
„normal“
n

FF [%]
a-SiGe:H 68
1.5eV 2.5
66
no buffer
„a-Si:H“ 2.0 64
p p/i i/n n 700
a-SiGe:H
6.5
1.5eV
680

jsc [mA/cm ]
2
Voc [mV]

6.0
660

p
„inverse“
n 640 5.5
p/i a-SiGe:H i/n
1.5eV
620 5.0
0 2 4 6 8 10 12 0 2 4 6 8 10 12
d(p/i = i/n) [nm] d(p/i = i/n) [nm]

Fig. 2. Comparison between a-SiGe:H solar cells with a-Si:H-buffer layers and profiled a-SiGe:H buffer at
the p/i- and i/n interface. The values for FF and VOC are measured as a function of the thickness d of the
buffer layer. d(p/i)=d(i/n). (&) Cells with a-Si:H buffer, (D) cells with profiled a-SiGe:H buffer, (J) cell
with inverse profiled a-SiGe:H buffer.

buffer finally decreases because of the thicker 1.5 eV a-SiGe:H i-layer necessary to
obtain the same jSC :
The results show that profiled a-SiGe:H buffer can be replaced by a-Si:H-buffer
layers without any losses in FF and VOC : An buffer layer of a certain thickness
(10 nm) is needed to obtain high VOC and FF, but apparently a profiled a-SiGe:H
buffer is not necessary.

3.2. Inverse profiled a-SiGe:H buffer

The performance of an ‘‘inverse’’ profiled structure is also presented in Fig. 2. In


this structure the ‘‘worst case’’ is realized by applying two bandgap steps, a small
bandgap (1.5 eV) and an enhanced defect density at both interfaces. Surprisingly, it
turns out that the bandgap profiling at the p/i- and i/n interface can be even inverted
without any loss in FF and VOC : In the following the expressions ‘‘normal’’ and
‘‘inverse’’ bandgap profiled a-SiGe:H buffer are used to characterize the two profiled
structures (see Fig. 2).

3.3. Thickness of the a-Si:H buffer

In the next experiment a variation of buffer layer thickness on one side keeping the
buffer layer on the other side at constant thickness is carried out.
The values for FF and VOC as a function of the thickness of the buffer layer are
plotted in Fig. 3. For the VOC nearly the same behaviour upon increasing the a-Si:H-
buffer thickness is observed for both interfaces. Up to a buffer layer thickness of
D. Lundszien et al. / Solar Energy Materials & Solar Cells 74 (2002) 365–372 369

720 72
p p/i i/n n
a-SiGe:H, 1.5 eV 700
70

Voc (mV)

FF (%)
680
68
660
p p/i i/n n 66
640
a-SiGe:H, 1.5 eV no buffer
620 64
0 5 10 15 20 25 0 5 10 15 20 25
d(buffer) [nm] d(buffer) [nm]

Fig. 3. Comparison between a-Si:H-buffer layer at the p/i- and i/n interface. The values for FF and VOC
are measured as a function of the thickness d of the buffer layer. (&) p/i interface, (J) i/n interface. The
dashed line indicates the values for a cell without buffer layers.

5 nm VOC increases strongly and saturates above 15 nm. To obtain a high VOC a
buffer at each interface is necessary—a single buffer (d ¼ 0 nm) shows only small
increase as compared to the VOC level for a cell with no buffer layer (dashed line).
The FF exhibits different trends upon varying the buffer thickness. The FF remains
nearly unaffected at high values upon changes of buffer thickness at the p/i interface.
On the other hand, the FF is very sensitive to changes of the a-Si:H-buffer thickness
at the i/n interface. Starting at low FF values (no buffer=dashed line) the FF
increases strongly with buffer thickness and saturates above 10 nm.

3.4. Role of bandgap of the buffer

To investigate the influence of the bandgap of the buffer layer on the solar cell
performance, the bandgap of the buffer layer at the p/i–i/n interface was varied
keeping the bandgap of the buffer at the i/n–p/i interface constant at EG ¼ 1:8 eV
(a-Si:H). The thickness of the buffers at both interfaces is 9 nm. The values for FF
and VOC as a function of the bandgap of the buffer layer are plotted in Fig. 4. For
VOC the same trend upon decreasing the optical bandgap of the buffer at the p/i- and
i/n interface is observed. A small decrease in VOC occurs for EG > 1:5 eV for the
buffer followed by a strong decease for EG o1:5 eV.
The FF remains unchanged down to EG values of 1.5 eV (1.6 eV) for the buffer at
the p/i–i/n interface, respectively. Below these optical bandgaps the FF decreases.
For changes at the i/n interface the decrease in FF is more pronounced and occurs
already at higher bandgaps. So the FF is more sensitive upon changes of the optical
bandgap at the i/n interface than at the p/i interface.

3.5. Role of a-Si:H-buffer-position

This experiment is motivated by the results of the inverse profiled a-SiGe:H


buffers. The idea that the position of the a-Si:H at the beginning (or the end) of the
graded a-SiGe:H buffer plays an important role is examined by a very thin a-Si:H
370 D. Lundszien et al. / Solar Energy Materials & Solar Cells 74 (2002) 365–372

700
70
p n
i/n
a-SiGe:H, 1.5 eV 650 65

VOC [mV]

FF [%]
600
no buffer
60

p 550 55
n
p/i
a-SiGe:H, 1.5 eV
500 50
1.4 1.5 1.6 1.7 1.8 1.4 1.5 1.6 1.7 1.8
Eg (buffer) [eV] Eg (buffer) [eV]

Fig. 4. Comparison between buffer layers (9 nm) of different optical bandgap EG at the p/i- and I/n buffer.
The values for FF and VOC are measured as a function of the optical bandgap of the buffer layer. (&) p/i
interface, (J) i/n interface, (,) cell with EG ¼ 1:65 eV for both buffers. The dashed line indicates the
values for a cell without buffer layers.

700

72
680
p i/n n
Voc [mV]

a-SiGe:H

FF [%]
68
660

64
640
buffer on one side
p p/i no buffer
n buffer on one side
620 60
a-SiGe:H 0 5 10 15 20 0 5 10 15 20
d(p/n-buffer) [nm] d(p/n-buffer) [nm]

Fig. 5. Comparison between a-Si:H-buffer layer at the p/i- and i/n interface. The values for FF and VOC
are measured as a function of the position d of a 3 nm a-Si:H layer. (&) p/i interface, (J) i/n interface.
The dashed line indicates the values for a cell without buffer layers.

buffer (of only 3 nm) which is introduced at various positions. The distance d
between the thin a-Si:H buffer (3 nm) and the doped layer at the p/i–i/n interface was
varied, keeping the a-Si:H-buffer thickness at the i/n–p/i interface constant at 9 nm.
Fig. 5 shows FF and VOC as a function of the distance d between the doped layer and
the a-Si:H layer. Again (compare Fig. 3) the VOC behaves very similar upon
variation of the buffer layer—this time the distance d between doped layer and the
thin a-Si:H buffer on either interface side. The VOC remains unchanged upon a shift
of the a-Si:H buffer away from the doped layer up to a distance of d ¼ 9 nm. Above
this distance the VOC decreases. Already with this thin a-Si:H buffer the VOC is
considerably enhanced.
For the FF instead, remarkable differences are found for the variation of the
distance d at the p/i- and i/n interface, respectively. A high FF is obtained using no
buffer layer at the p/i interface. Upon increasing the distance d between the p-layer
and the a-Si:H buffer the FF decreases. At the i/n interface using no or only this thin
D. Lundszien et al. / Solar Energy Materials & Solar Cells 74 (2002) 365–372 371

a-Si:H buffer the FF is on a low level (compare Fig. 3). Surprisingly, the FF
increases if the thin a-Si:H buffer is shifted away from the n-layer! For 3 nm
odo6 nm the FF reaches a maximum and decreases for d > 6 nm.

4. Discussion

The results show that different shapes (profiled, inverse profiled and nonprofiled)
of the buffer yield nearly the same performance once the buffer layer thickness
exceeds a certain value. In particular, the VOC reaches the same high level for any of
the applied buffer layers, provided a critical buffer layer thickness and buffer layer
bandgap are used on both sides. Apparently bandgap profiling to avoid bandgap
steps near the p/i- and i/n interface like proposed in previous literature reports [3,5,6]
are not necessary.
The fact that the same solar cell performance is reached for inverse and normal
profiling a-SiGe:H buffers raises the question what both shapes have in common that
leads to the high FF values. One possibility is that the beneficial effect of a profiled
a-SiGe:H buffer is only caused by the increased average bandgap in the profiled a-
SiGe:H buffer irrespective of the shape of these buffer layer. If this is true, constant
buffer bandgap layers of, say, 1.65 eV, should have the same effect as a profiled a-
SiGe:H buffer with 1.8 eVoEG o1.5 eV [1.65=(1.8+1.5)/2]. An average optical
bandgap of 1.65 eV at both interfaces is sufficient for a high FF as shown in Fig. 4.
However, this shape leads to small values for VOC compared to the structures with
profiled a-SiGe:H buffers (see Fig. 2). So the normal and inverse profiled a-SiGe:H
buffers cannot be easily replaced by buffer layers with the same average bandgap.
The FF for the inverse profiling is as high as for the a-Si:H-buffer structure. As
there is no beneficial effect of the bandgap profiling itself, it is possible that the
a-Si:H in the inverse profiled a-SiGe:H buffer at the i/n interface is responsible for
the high FF values. This is supported by the results presented in Fig. 5 for the thin a-
Si:H buffer. In this context it should be noted that for the inverse profiled a-SiGe:H
buffer with a small region of highly defective low bandgap a-SiGe:H material located
close to the doped layers, the observed high values for VOC and FF were not
expected.

5. Summary

In this study, the different buffer shapes of thin a-SiGe:H (1.5 eV, 54 nm) solar
cells were investigated. Profiled a-SiGe:H buffers are compared with a-Si:H buffers
as buffer layer in a-SiGe:H solar cells. The profiled buffers can be replaced by simple
a-Si:H-buffer layers without any drawbacks. It is shown that different shapes
(profiled, inverse profiled and nonprofiled) of the buffer layer yield nearly the same
performance once the buffer layer thickness exceeds a certain value.
The VOC shows the same dependence upon manipulations of the buffer shape at
both the p/i- and the i/n interface. On one hand, the FF is not affected by an a-Si:H
372 D. Lundszien et al. / Solar Energy Materials & Solar Cells 74 (2002) 365–372

buffer at the p/i interface, while on the other hand the FF shows a very complex
behaviour upon changes near the i/n interface. In particular, the FF shows a
maximum for a thin a-Si:H buffer located 5 nm away from the n-layer.
The bandgap grading of the entire absorber layer is apparently no prerequisite for
high-efficiency solar cells as proposed in the literature. The reason for this behaviour
is not known at this point and should be challenge for device simulations.

Acknowledgements

This work was supported by the ASE GmbH, Division Phototronics. We thank A.
.
Lambertz, F. Birmans and S. Michel for technical assistance and J. Muller and B.
Rech for helpful discussions.

References

[1] S. Guha, J. Jang, A. Pawlikiewitz, T. Glatfelder, R. Ross, S.R. Ovshinsky, Band-gap profiling for
improving the efficiency of amorphous silicon alloy solar cells, Appl. Phys. Lett. 54 (23) (1989)
2330–2332.
[2] E. Maruyama, K. Sayama, A. Terakawa, K Ninomiya, S. Tsuda, H. Tarui, S. Nakano, Y. Kuwano,
Improvement for high-efficiency, stable multi-junction cells, Proceedings of the 23rd IEEE PV
Specialists Conference, Louisville, KY, 1993, pp. 827–832.
.
[3] J. Folsch, H. Stiebig, F. Finger, B. Rech, D. Lundszien, A. Lambertz, H. Wagner, Role of bandgap
grading for the performance of a-SiGe:H solar cells, Proceedings of the 25th IEEE PV Specialists
Conference, Washington, DC, 1996, pp. 1133–1136.
[4] J. Zimmer, H. Stiebig, H. Wagner, A-SiGe:H based solar cells with graded absorption layer, J. Appl.
Phys. 84 (1) (1998) 611–617.
[5] R.A.C.M.M. Swaaij van, M. Zeman, S. Arnoult, J.W. Metselaar, Performance dependance on grading
width of a-SiGe:H component solar cells, Proceedings of the 28th IEEE PV Specialists Conference,
Anchorage, 2000, pp. 869–872.
[6] F. Smole, M. Topic, J. Furlan, Buffer layers versus band grading in a-SiGe:H solar cells, Proceedings
of the 14th European Photovoltaic Solar Energy Conference, Barcelona, Spain, 1997, pp. 632–635.

You might also like