Microstructural dependence of electron and hole transport in low-temperature-grown polycrystalline-silicon thin-film solar cells

Article abstract of DOI:10.1063/1.1527979

The carrier transport properties of undoped polycrystalline silicon (poly-Si) thin films were investigated. It was shown that the ac-conductivity measurement within n-i-n and p-i-p sample structures can be used to determine individual electron and hole co

Full text of DOI:10.1063/1.1527979

Microstructural dependence of electron and hole transport in low-temperature-grown
polycrystalline-silicon thin-film solar cells
Takuya Matsui, Riza Muhida, Tomohiro Kawamura, Toshihiko Toyama, Hiroaki Okamoto, Tsutomu Yamazaki,
Shinya Honda, Hideyuki Takakura, and Yoshihiro Hamakawa
Citation: Applied Physics Letters 81, 4751 (2002); doi: 10.1063/1.1527979
View online: http://dx.doi.org/10.1063/1.1527979
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/81/25?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Impact of the n+ emitter layer on the structural and electrical properties of p-type polycrystalline silicon thin-film
solar cells
J. Appl. Phys. 114, 134505 (2013); 10.1063/1.4823811
Chemical speciation at buried interfaces in high-temperature processed polycrystalline silicon thin-film solar cells
on ZnO:Al
J. Appl. Phys. 113, 044519 (2013); 10.1063/1.4789599
N/I buffer layer for substrate microcrystalline thin film silicon solar cell
J. Appl. Phys. 104, 104505 (2008); 10.1063/1.3021053
Defect density and recombination lifetime in microcrystalline silicon absorbers of highly efficient thin-film solar
cells determined by numerical device simulations
J. Appl. Phys. 94, 1035 (2003); 10.1063/1.1577813
Local characterization of electronic transport in microcrystalline silicon thin films with submicron resolution
Appl. Phys. Lett. 74, 1475 (1999); 10.1063/1.123585
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.42.202.150 On: Sat, 22 Nov 2014 00:42:01

APPLIED PHYSICS LETTERS
VOLUME 81, NUMBER 25
16 DECEMBER 2002
Microstructural dependence of electron and hole transport in low-
temperature-grown polycrystalline-silicon thin-film solar cells
Takuya Matsui,a) Riza Muhida, Tomohiro Kawamura, Toshihiko Toyama,
and Hiroaki Okamoto
Department of Physical Science, Graduate School of Engineering Science, Osaka University, Toyonaka,
Osaka, 560-8531, Japan
Tsutomu Yamazaki, Shinya Honda, Hideyuki Takakura, and Yoshihiro Hamakawa
Department of Photonics, Ritsumeikan University, Kusatsu, Shiga, 525-8577, Japan
͑
Received 5 April 2002; accepted 16 October 2002͒
Carrier transport properties of undoped polycrystalline silicon ͑poly-Si͒ thin films prepared by
SiH –H plasma at low temperature have been investigated. The ac-conductivity measurement
4
2
technique has been applied to poly-Si i layers with an n-i-n junction structure in order to
characterize the electron conductivity along the growth direction. Furthermore, the hole
conductivity has been measured with p-i-p junction structures. The temperature dependence of ac
conductivity reveals that poly-Si films with relatively low crystalline volume fraction (X ϳ50%)
c
exhibit intrinsic character, while the poly-Si films with high X (Ͼ50%) exhibit n-type character
c
with activation energies less than 0.15 eV. Based on these results, the relationship among
microstructure, carrier transport, and photovoltaic performance of poly-Si solar cells is
discussed. © 2002 American Institute of Physics. ͓DOI: 10.1063/1.1527979͔
Low-temperature-grown polycrystalline silicon ͑poly-Si͒
thin films produced by plasma-enhanced chemical vapor
deposition ͑PECVD͒ are currently attracting considerable at-
tention as a low-cost and stable photovoltaic material.1 Re-
cently, several groups have reported that poly-Si films pre-
pared under various conditions in the PECVD process show
a wide variety of microstructures from a crystalline-
amorphous mixed state to an almost perfect crystalline state
with different crystallographic orientations.2 These struc-
tural changes directly affect the carrier transport property
which plays a dominant role in determining the photovoltaic
performance. Due to such complicated microstructures, how-
ever, it is not straightforward to explore the carrier transport
properties of poly-Si films. Moreover, little effort has been
made to identify the effect of carrier transport on the photo-
voltaic performance.6
Si/metal, however, one does not observe the simple linear
Ohm’s law in dc current–voltage characteristics due to the
presence of carrier depletion acting as a contact barrier at
metal/poly-Si interfaces. The ac-conductivity measurement
technique has been proposed as a powerful tool for measur-
ing the material conductivity with a sandwich electrode
–5
8
geometry. In this letter, we report on the results of electrical
conductivities due to electrons and holes in the poly-Si films
using the ac-conductivity technique. We show that the Fermi
energy level and photosensitivity in the poly-Si layer are
quite sensitive to the film microstructure. Based on these
results, we discuss the relationship between electrical prop-
erties and photovoltaic performance of poly-Si films.
,3,5
The solar cells prepared in this work have a structure of
5
Ag-grid/ZnO/p-i-n/ZnO/Ag/glass. Highly conductive mi-
,7
crocrystalline silicon p and n layers that were deposited by a
rf–PECVD system. The poly-Si i layers were deposited at
For the case of the columnar grains oriented perpendicu-
larly to substrate, which is often encountered in ͑220͒ pref-
erentially orientated poly-Si films, the carrier transport can
be separated into two components; one is along the film ͑par-
allel͒ and another is along the growth direction ͑perpendicu-
lar͒. In this case, the parallel conductivity should be strongly
affected by the grain boundaries, whereas the perpendicular
conductivity is affected very little. For the PECVD produced
poly-Si films, such an anisotropic electrical transport has
1
80 °C by another PECVD system with a very high fre-
quency ͑VHF͒ excitation source ͑100 MHz͒. A mixture of
SiH and H with a H concentration of 4.5% was used for
4
2
4
the growth of the poly-Si i layer. The film crystallinity was
changed by deposition pressure in VHF–PECVD process.5
The crystalline volume fraction, X , and crystallographic
c
orientation were determined by Raman and x-ray diffraction
͑
XRD͒ measurements, respectively.5
8
been reported by Ko cˇ ka et al. Concerning practical p-i-n
The ac-conductivity measurement was performed on
poly-Si solar cells, such carrier transport parameters as elec-
trical conductivity of poly-Si films should be characterized in
the form of sandwich electrode geometry, as in solar cell
structure, because the perpendicular carrier transport clearly
dominates the device operation, rather than the parallel trans-
port. With the sandwich electrode geometry, e.g., metal/poly-
2-␮m-thick poly-Si i layers with different sample structures
consisting of ZnO/n-i-n/SnO2 /glass and ZnO/p-i-p/
SnO /glass. The flat ZnO and SnO films were employed as
2
2
front and rear side transparent electrodes, respectively. The
dark ac conductivity was measured in the frequency range of
5
0 Hz–1 MHz under an applied ac voltage of 10 mVrms .
a͒Present address: National Institute of Advanced Industrial Science and
Technology, Japan; electronic mail: t-matsui@aist.go.jp
Since the ac-conductivity spectra involve the effects of car-
rier depletion at contacts and electrode resistance, the true
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
003-6951/2002/81(25)/4751/3/$19.00 4751 © 2002 American Institute of Physics
28.42.202.150 On: Sat, 22 Nov 2014 00:42:01
0
1

4752
Appl. Phys. Lett., Vol. 81, No. 25, 16 December 2002
Matsui et al.
FIG. 1. Short circuit current density (J ) and open circuit voltage (V ) of
sc
oc
poly-Si solar cells with different i-layer crystalline volume fractions of Xc
ϳ50% ͑closed͒ and ϳ80% ͑open͒ as a function of i-layer thickness.
Dashed and solid lines represent theoretical plots calculated with S
FIG. 2. Temperature dependence of dark conductivity of poly-Si i layers
GB
1
3
Ϫ3
ϭ10 cm/s, NGBϭ10 cm
͑model A͒ and SGBϭ20 cm/s, NGB
with different crystalline volume fractions of X ϳ50% ͑circles͒ and ϳ80%
c
1
6
Ϫ3
ϭ10 cm ͑model B͒, respectively.
͑squares͒ measured with n-i-n ͑closed͒ and p-i-p ͑open͒ sample structures.
film conductivity of poly-Si i layer can be determined at
noted that poly-Si i layers in the n-i-n and p-i-p samples
are generally expected to be different due to the difference in
the nucleation behavior on n and p layers. Nevertheless, in
this series, no significant structural difference is found be-
tween poly-Si i-layer deposited on n layer and that on p
layer from the Raman and XRD measurements. In Fig. 2, a
remarkable difference in the dark conductivity behaviors is
observed for these two poly-Si i layers. For the poly-Si film
frequencies where the relative permittivity, ␧ , approaches
eff
8
a proper value of ␧ (Si)ϭ(Cd/␧ S)Ϸ12. Here, C, d, ␧ ,
eff
0
0
and S denote the capacitance corresponding to the imaginary
part of measured admittance spectra, the sample thickness,
the vacuum permittivity, and the sample area, respectively.
The photoconductivity was also measured under air mass
2
1
.5, 100 mW/cm illumination through the glass substrate.
Figure 1 shows the short circuit current density, J , and
with X ϳ50%, the activation energies of the dark conduc-
sc
c
open circuit voltage, V , of poly-Si solar cells as a function
tivity, ␧ , which is deduced from the Arrhenius plot in the
oc
a
of i-layer thickness in the range of 1.0–5.6 ␮m. In Fig. 1, the
higher temperature regime ͑300–400 K͒, are almost identical
around 0.55 eV for both n-i-n and p-i-p samples. For the
poly-Si film with X ϳ80%, on the other hand, ␧ for n-i-n
results on poly-Si with X ϳ50% are compared to those on
c
poly-Si with X ϳ80%. Both series of poly-Si layers exhibit
c
c
a
5
͑
220͒ preferential orientation. The solid and dashed lines in
is much smaller than that for p-i-p. These electrical proper-
ties depending on sample structures can be interpreted as
follows. For the n-i-n sample, the n/i interfaces should act
as barriers for the hole extraction, and only electrons can
contribute to the current flow if the applied voltage is much
smaller than the barrier height impeding hole extraction. On
the contrary, the p-i-p sample permits only holes for the
current flow. Since the activation energy of the dark conduc-
tivity is a good measure of the Fermi energy from the band
edge for the carrier transport, the result on the poly-Si film
with X ϳ50%, i.e., ␧ ϳ0.55 eV being independent of
Fig. 1 represent the theoretical fits which will be discussed
later. Regardless of X , J linearly increases with increasing
c
sc
thickness up to the i layer thickness of 3 ␮m. No significant
difference is found for the improvement in Jsc between solar
cells with X of ϳ50% and ϳ80%. However, J rapidly
c
sc
starts to decrease at the thickness above 3 ␮m for Xc
ϳ80%, whereas Jsc gradually increases at the thickness
above 5 ␮m for X ϳ50%. On the other hand, V signifi-
c
oc
cantly decreases with an increase in the i-layer thickness. In
particular, independent of i-layer thicknesses, Voc for Xc
c
a
ϳ80% is always lower than that for X ϳ50%. It is evident
sample structure suggests that the Fermi level locates at the
center of the band gap. Thus, this material is almost intrinsic.
c
that the poly-Si layer with X ϳ50% is advantageous as the
c
photovoltaic material in comparison with that with Xc
For the poly-Si film with X ϳ80%, on the other hand, the
c
ϳ80%. In fact, using the poly-Si i layer with X ϳ50%, a
small ␧a for electron transport and the large ␧a for hole
transport provide strong evidence that the Fermi level is
close to the conduction band edge. Thus, this material is n
c
maximum conversion efficiency of 9.18% (Jsc
2
ϭ25.1 mA/cm , V ϭ0.527 V, fill factorϭ0.7) has been
oc
achieved with the i-layer thickness of 2.5 ␮m, by employing
type. Table I summarizes room temperature ␴ taken from
the Arrhenius plot shown in Fig. 2, together with the results
d
an appropriate light trapping structure.9
In order to elucidate the relationship between poly-Si
microstructure and photovoltaic performance, electrical char-
acterization was performed on poly-Si films by the ac-
conductivity technique. In Fig. 2, the temperature depen-
of photoconductivity (␴ ) and photosensitivity (␴ /␴ ).
ph
ph
d
In Table I, ␴ph is less dependent on poly-Si crystallinity com-
pared to ␴ , implying that ␴ is determined by the number
d
ph
of photogenerated carriers under light illumination, rather
than by the Fermi level at thermal equilibrium. Accordingly,
dence of dark conductivity (␴ ) measured for poly-Si i
d
layers with different crystalline volume fractions (X ϳ50%
␴ph /␴d is predominantly determined by ␴ . In particular,
c
d
and ϳ80%) is shown. Figure 2 also demonstrates ␴ mea-
for the poly-Si films with X ϳ80%, ␴ markedly varies
d
c
d
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
sured for the n-i-n and the p-i-p samples. It should be
with sample structure. As a result, ␴ /␴ for p-i-p is found
ph d
128.42.202.150 On: Sat, 22 Nov 2014 00:42:01

Appl. Phys. Lett., Vol. 81, No. 25, 16 December 2002
Matsui et al.
4753
TABLE I. Dark conductivity (␴ ), photoconductivity (␴ph), and photosen-
d
sitivity (␴ph /␴ ) of poly-Si i layers with different crystalline volume frac-
d
tions (X ) measured with n-i-n and p-i-p sample structures.
c
Xc
n-i-n
␴d
␴ph
͑%͒
(p-i-p)
͑S/cm͒
͑S/cm͒
␴ph /␴d
Ϫ7
Ϫ5
3.4ϫ102
1.8ϫ102
2.7
ϳ50
n-i-n
p-i-p)
n-i-n
1.7ϫ10
8.3ϫ10
4.1ϫ10
5.8ϫ10
1.5ϫ10
1.1ϫ10
5.4ϫ10
Ϫ9
Ϫ5
Ϫ6
Ϫ4
Ϫ6
(
(
ϳ80
Ϫ10
1.4ϫ104
p-i-p)
3.8ϫ10
to be much higher than that for n-i-n, where ␴ph /␴ differs
d
FIG. 3. Calculated carrier recombination rate profiles along the grain bound-
ary position, at short circuit condition (Vϭ0 V) and at forward biased con-
dition (Vϭ0.43 V). Dashed and solid lines represent the calculation results
using S ϭ10 cm/s, N ϭ1013 cm ͑model A͒ and SGBϭ20 cm/s, NGB
by four orders magnitude. This result provides additional
support that photogenerated holes dominate minority carrier
transport in the highly crystallized poly-Si films.
It has been reported that the origin of earlier-mentioned
n-type character observed in highly crystallized poly-Si films
Ϫ3
GB
GB
ϭ1016 cm ͑model B͒, respectively.
Ϫ3
is attributed either to intrinsic nature or to the electrically
forward applied voltage is nearly close to open circuit con-
dition. In Fig. 3, model B reveals that the intensive carrier
recombination takes place near the p/i interface when V
ϭ0.43 V. Due to its n-type character, the built-in electric
field locally concentrates at p/i interface. Therefore, carrier
density at this depletion region is dramatically increased by
the forward bias application. Since the recombination rate is
determined by the product of majority carrier density and
minority carrier density, holes recombine with ‘‘majority’’
electrons at the p/i interface via the defective grain bound-
aries. This effect would account for the finding that Voc for
X ϳ80% is always lower than that for X ϳ50%.
active oxygen-related complex.2
,10–12
In general, it is widely
accepted that the oxygen-related donors concentrate at grain
boundaries in the poly-Si films,11 and oxygen contaminated
grain boundaries give rise to similar defect energy distribu-
1
2
tion as Si/SiO interface. This idea of such an electrically
2
activated grain boundaries with high donor concentration in
the poly-Si films allows us to give clear explanations for the
results of the thickness dependence of photovoltaic perfor-
mance shown in Fig. 1. The lines shown in Fig. 1 represent
calculated results based on the two-dimensional grain bound-
1
3
ary model in which uniform columnar grains in the i layer
with lateral grain size of 0.01 ␮m are assumed. Here, the
dashed lines ͑model A: intrinsic͒ indicate simulation results
calculated with the surface recombination velocity at grain
boundary of SGBϭ10 cm/s and the effective donor concen-
tration at grain boundary of NGBϭ1013 cm . For the solid
c
c
In conclusion, we demonstrate that the ac-conductivity
measurement with n-i-n and p-i-p sample structures can be
applied to determine the individual electron and hole con-
ductivities along the poly-Si growth direction. The poly-Si
films with relatively low crystalline volume fraction Xc
ϳ50% exhibit intrinsic character, leading to the successful
high efficiency solar cell fabrication. In contrast, poly-Si
films with high crystalline volume fraction induce higher
film conductivity with n-type character. This effect is pro-
posed to be responsible for the pronounced reduction in Voc .
Ϫ3
line ͑model B: n type͒ SGBϭ20 cm/s and N ϭ1016 cm
Ϫ3
GB
are assumed. Preliminary studies1
3,14
have suggested that
SGB should be lower than 100 cm/s in order to reproduce
recent experimental results of poly-Si solar cells. NGB values
given in models A and B are set as their macroscopic dark
conductivities to be identical with the measured dark con-
ductivities for X ϳ50% and ϳ80%, respectively. Figure 3
c
1
shows the calculated carrier recombination rate along the
grain boundary position for both cases of models A and B, at
short circuit condition (Vϭ0 V) and at forward biased con-
dition (Vϭ0.43 V). The i-layer thickness is set at 5.0 ␮m.
Under short circuit condition, carrier recombination rate in
model B is much higher than that in model A particularly at
an i-layer position adjacent to i/n interface. The considerable
recombination loss of photogenerated carriers found in
model B is ascribed to the higher SGB and to the lower
built-in electric field in the i layer due to n-type character.
This calculation result is consistent with the measured spec-
K. Yamamoto, IEEE Trans. Electron Devices 46, 2041 ͑1999͒.
H. Keppner, J. Meier, P. Torres, D. Fischer, and A. Shah, Appl. Phys. A:
Mater. Sci. Process. 69, 169 ͑1999͒.
O. Vetterl, R. Carius, L. Houben, C. Scholten, M. Luysberg, A. Lambertz,
F. Finger, and H. Wagner, Mater. Res. Soc. Symp. Proc. 609, A15 ͑2000͒.
Y. Nasuno, M. Kondo, and A. Matsuda, Jpn. J. Appl. Phys., Part 2 40,
L303 ͑2001͒.
T. Matsui, M. Tsukiji, H. Saika, T. Toyama, and H. Okamoto, Jpn. J. Appl.
Phys., Part 1 41, 20 ͑2002͒.
M. Goerlitzer, P. Torres, N. Beck, N. Wyrsch, and A. Shah, J. Non-Cryst.
Solids 227–230, 996 ͑1998͒.
O. Vetterl, A. Groß, T. Jana, S. Ray, A. Lambertz, R. Carius, and F. Finger,
J. Non-Cryst. Solids 299–302, 772 ͑2002͒.
J. Ko cˇ ka, A. Fejfar, V. Vorl ´ı cˇ ek, H. Stuchl ´ı kov a´ , and J. Stuchl ´ı k, Mater.
Res. Soc. Symp. Proc. 557, 490 ͑1999͒.
T. Matsui, M. Tsukiji, H. Saika, T. Toyama, and H. Okamoto, J. Non-
Cryst. Solids 299–302, 1152 ͑2002͒.
S. Vep ˇr ek, Z. Iqbal, R. O. K u¨ hne, P. Capezzuto, F. A. Sarott, and J. K.
Gimzewski, Solid State Phys. 16, 6241 ͑1983͒.
Y. Nasuno, M. Kondo, and A. Matsuda, Appl. Phys. Lett. 78, 2330 ͑2001͒.
J. H. Werner, Inst. Phys. Conf. Ser. 104, 63 ͑1989͒.
T. Matsui, T. Yamazaki, A. Nagatani, K. Kino, H. Takakura, and Y. Ha-
makawa, Sol. Energy Mater. Sol. Cells 65, 87 ͑2001͒.
J. H. Werner, Technical Digest of 13th Sunshine Workshop on Thin Film
2
3
4
5
6
7
8
9
0
tral response. In the solar cells with X ϳ80%, the rapid
c
decrease in Jsc at the thickness Ͼ3 ␮m arises from the sig-
nificant decrease in the quantum efficiencies at longer wave-
lengths (Ͼ600 nm). Since the long wavelength response is
mostly determined by the diffusion length of minority carri-
ers generated at rear side of the solar cell, it can be inter-
preted as that the photogenerated holes are no longer col-
lected effectively if the cell thickness exceeds 3 ␮m.
1
11
1
1
2
3
14
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
The situation is quite different at Vϭ0.43 V at which
Solar Cells ͑NEDO, Tokyo, 2000͒, p. 41.
128.42.202.150 On: Sat, 22 Nov 2014 00:42:01