Efficient defect passivation by hot-wire hydrogenation

Article abstract of DOI:10.1063/1.119371

Atomic hydrogen, produced at a hot wire, passivates bulk defects in polycrystalline silicon without damaging surface regions. Solar cells from such polycrystalline silicon respond much more favorably to hot-wire hydrogenation than to low-energy ion implantation or a direct-current plasma treatment. Hot-wire passivation yields a hydrogen concentration close to the surface of 8 × 10¹⁹ cm^-3 and improves the minority carrier diffusion length of solar cells by up to 100%. Implantation as well as conventional plasma treatment result in lower hydrogen concentration and, consequently, in much smaller improvements of diffusion lengths.

Full text of DOI:10.1063/1.119371

Efficient defect passivation by hot-wire hydrogenation
R. Plieninger, H. N. Wanka, J. Kühnle, and J. H. Werner
Citation: Applied Physics Letters 71, 2169 (1997); doi: 10.1063/1.119371
View online: http://dx.doi.org/10.1063/1.119371
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/71/15?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Damp and dry heat degradation of thermal oxide passivation of p + silicon
J. Appl. Phys. 115, 114505 (2014); 10.1063/1.4869057
Hydrogen engineering via plasma immersion ion implantation and flash lamp annealing in silicon-based solar cell
substrates
J. Appl. Phys. 115, 064505 (2014); 10.1063/1.4865737
Interdigitated back contact silicon heterojunction solar cell and the effect of front surface passivation
Appl. Phys. Lett. 91, 063507 (2007); 10.1063/1.2768635
Defect passivation in multicrystalline silicon for solar cells
Appl. Phys. Lett. 85, 4346 (2004); 10.1063/1.1815380
Hydrogen plasma passivation of GaAs on Si substrates for solar cell fabrication
J. Appl. Phys. 87, 2285 (2000); 10.1063/1.372174
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:
131.193.242.21 On: Thu, 27 Nov 2014 22:43:28

Efficient defect passivation by hot-wire hydrogenation
R. Plieningera)
¨
¨
Max-Planck-Institut fur Festkorperforschung, Heisenbergstrasse 1, D-70569 Stuttgart, Germany
H. N. Wanka
¨
¨
Universitat Stuttgart, Institut fur Physikalische Elektronik, Pfaffenwaldring 47, D-70569 Stuttgart, Germany
¨
J. Kuhnle
¨
¨
Max-Planck-Institut fur Festkorperforschung, Heisenbergstrasse 1, D-70569 Stuttgart, Germany
J. H. Werner
¨
¨
Universitat Stuttgart, Institut fur Physikalische Elektronik, Pfaffenwaldring 47, D-70569 Stuttgart, Germany
͑Received 28 May 1997; accepted for publication 13 August 1997͒
Atomic hydrogen, produced at a hot wire, passivates bulk defects in polycrystalline silicon without
damaging surface regions. Solar cells from such polycrystalline silicon respond much more
favorably to hot-wire hydrogenation than to low-energy ion implantation or a direct-current plasma
treatment. Hot-wire passivation yields a hydrogen concentration close to the surface of 8ϫ1019
cmϪ3 and improves the minority carrier diffusion length of solar cells by up to 100%. Implantation
as well as conventional plasma treatment result in lower hydrogen concentration and, consequently,
in much smaller improvements of diffusion lengths. © 1997 American Institute of Physics.
͓S0003-6951͑97͒04741-4͔
The high cost of monocrystalline silicon wafers forces
many laboratories to investigate solar cells from polycrystal-
line silicon.1,2 Polycrystalline silicon in the form of thin lay-
ers promises strong cost reduction.3,4 Unfortunately, crystal
defects and impurities reduce diffusion lengths in polycrys-
talline silicon. For efficient solar cells, diffusion lengths ex-
ceeding the cell thickness are required. Several authors
showed an increase of the diffusion length in polycrystalline
silicon after different hydrogen treatments.5–7 However, not
in all cases was efficient defect passivation obtained. Novel
defect passivation methods are of utmost importance in solar
cell technology.
This letter reports on passivation of polycrystalline sili-
con layers by hydrogen radicals produced at a hot wire. In
this method, hydrogen molecules are cracked at a thin tung-
sten wire that is resistively heated to temperatures around
1900 °C. Deposition of amorphous and microcrystalline8
silicon with the help of a hot wire is already well known.
Only in the case of microcrystalline9 ͑grain size Ͻ1 ␮m͒ and
single-crystalline silicon10 solar cells were improvements by
a hot-wire hydrogen treatment reported. However, there is no
study on the use of hot wires for hydrogen passivation of
large grained polycrystalline silicon.
͑HWHP͒ with two well established methods of hydrogen
treatment: low energy ion implantation5,6,13 and remote
plasma treatment.7,14 Hydrogen implantation is carried out in
a commercial system using a Kauffmann-type ion gun. Dur-
ing the 30 min treatment with an ion energy of 1.2 keV and
an ion current of 0.75 mA/cm2, the sample reaches a tem-
perature of about 200 °C. The rather high energy is neces-
sary to reach reasonable sample temperatures as the setup
does not offer a direct sample heating. The ion implantation
conditions had been optimized previously for highest solar
cell efficiency using the same raw material. The direct cur-
rent ͑dc͒ remote plasma system consists of a 1 kV dc glow
discharge operated at 150 Pa pure hydrogen gas pressure. A
300 V bias voltage accelerates the hydrogen ions from the
plasma towards the sample, which is heated to 350 °C during
the 3 h treatment. One could regard this method as a low-
energy ion implantation with heated sample stage as well.
The hot-wire reactor uses a resistively heated tungsten wire
͑temperature 1900 °C͒ which cracks the molecules of hydro-
gen partly into radicals. Without further acceleration, the hy-
drogen atoms reach the sample, which is in a distance of 2
cm from the wire and is heated to 350 °C.
We determine the hydrogen concentration at and close to
15
Polycrystalline silicon used in this study is provided to
us by Mitsubishi Electric Corp., Japan. These silicon layers
are produced by atmospheric pressure chemical vapor depo-
sition from SiH2Cl2 at 1100 °C on zone-melting recrystal-
lized silicon seeds.11 The layer thickness is 100 ␮m the bo-
the surface by elastic recoil detection analysis ͑ERDA͒
with an 4 MeV Heϩ primary ion beam from the Pelletron
¨
accelerator at the Max-Planck-Institut fur Metallforschung,
Stuttgart. Hall effect measurements of the majority carrier
concentration p and mobility ␮ quantify the reduction of
active doping concentration by the hydrogen treatment.16 Fi-
nally, the analysis of spectral internal quantum efficiency
measurements17 of test solar cells with a diffused, highly
n-type doped emitter yields the minority carrier diffusion
length in the bulk silicon prior to and after the hydrogen
incorporation.
ron and doping concentration is 2.15ϫ1016 cmϪ3
.
Dislocations, twins, and stacking faults are the main crystal
defects in our polycrystalline silicon with grain sizes ranging
from millimeters to several centimeters.12 Similar films are
used for the preparation of solar cells with efficiencies up to
16.4% after hydrogen passivation.1
Our study compares hot-wire hydrogen passivation
Figure 1 shows the ERDA profiles taken from the hydro-
gen exposed side of three samples treated with either dc re-
mote plasma, or ion implantation, or hot-wire hydrogen. We
a͒
Electronic mail: plie@quasix.mpi-stuttgart.mpg.de
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:
Appl. Phys. Lett. 71 (15), 13 October 1997 0003-6951/97/71(15)/2169/3/$10.00 © 1997 American Institute of Physics 2169
131.193.242.21 On: Thu, 27 Nov 2014 22:43:28

TABLE II. Majority carrier ͑hole͒ concentration and mobility before ͑p, ␮)
and after ͑pH , ␮H) hydrogenation.
Before treatment
After treatment
␮
pH
␮
H
Treatment
p ͓cmϪ3
͔
͓cm2/V s͔
͓cmϪ3
͔
͓cm2/V s͔ ͑pϪpH)/p
dc plasma
2.16ϫ1016
2.14ϫ1016
268
271
265
2.04ϫ1016
2.07ϫ1016
1.67ϫ1016
270
273
275
6%
3%
23%
Hϩ implant
hot wire H 2.17ϫ1016
concentration of Nbulkϭ8ϫ1019 cmϪ3. Ion implantation
could perhaps yield similar hydrogen concentrations in the
bulk, if the samples were heated to 350 °C.
Table II compares the Hall carrier concentrations and
mobilities before and after hydrogenation. Among the three
passivation methods, hot-wire treatment shows the strongest
reduction of the carrier concentration, which decreases by
23%, while the dc plasma treatment yields a reduction of
only 6%. This number averages over the complete sample
thickness, therefore, the acceptor passivation or compensa-
tion close to the surface may be even higher. None of the
three methods affects the majority carrier mobility ␮Hϭ270
cm2/V s, a value close to the one of monocrystalline silicon
with the same doping.
Table III compiles the results for the test solar cells with
a structure shown in the inset of Fig. 2. The carrier diffusion
length LB is the most appropriate quantity to characterize the
minority carriers. Prior to hydrogenation, all test cells show
initial electron diffusion lengths LB between LBϭ30 ␮m and
LBϭ33 ␮m. The dc plasma treatment from the back side
does not significantly change this value. However, after ion
implantation as well as after hot-wire treatment from the
back side the diffusion length after hydrogenation, LBH in-
creases to approximately LBHϭ50 ␮m. Front side treatment
with the remote dc plasma yields LBHϭ40 ␮m. Unfortu-
nately, ion implantation at 1.2 keV from the front side causes
considerable emitter damage and is not studied in further
detail here. The best result is obtained by hot-wire treatment
from the front side, yielding a diffusion length of LBHϭ66
␮m. After annealing for 30 min at 350 °C in a pure argon
atmosphere, majority carrier concentration as well as minor-
ity carrier diffusion length of ion implanted samples return
exactly to the values before the hydrogen implantation.
The results in Table III indicate the efficiency of hot-
wire passivation for defects in the p-type base of our boron
doped solar cells. We found similar results in all solar cells
investigated so far. However, technological advantages in
FIG. 1. ERDA analysis of polycrystalline silicon after different hydrogena-
tion treatments. The peak in the hydrogen count rate stems from the hydro-
gen at the surface; the signal at lower energies ͑channel numbers͒ originates
from hydrogen within the bulk. The depth scale is calculated according to
the experimental conditions.
detect only the hydrogen atoms recoiled from the sample.
The peak at high energies ͑channel numbers͒ corresponds to
the surface, lower energy hydrogen ions originate from in-
side the sample. Table I summarizes the hydrogen concen-
trations Nsurf at the surface and Nbulk in the bulk close to the
surface ͑less than 1 ␮m depth͒ deduced from these measure-
ments. Treatments by hot-wire as well as by dc plasma yield
hydrogen surface concentrations of Nsurfϭ3.5ϫ1015 cmϪ2
.
This number almost equals the areal surface density of sili-
con atoms and suggests therefore a completely hydrogen ter-
minated surface. After ion implantation, we find an even
higher surface concentration of hydrogen of Nsurfϭ1ϫ1016
cmϪ2. This enormous areal concentration is either due to
surface damage caused by the impinging hydrogen ions and
the resulting increased surface area, or this number corre-
sponds to a thin layer with very high volume concentration
directly beneath the surface. Due to the low sample tempera-
ture of only 200 °C during ion implantation the diffusion of
hydrogen into the bulk is expected to be slower than at the
higher temperature of 350 °C used for the other treatments
where direct sample heating was possible. In the case of the
dc plasma treatment and hydrogen implantation, the small
count rates for lower energy hydrogen atoms correspond to a
hydrogen concentration below Nbulkϭ2ϫ1019 cmϪ3. Low
energy ions in the primary beam, which produce low energy
recoils from surface hydrogen atoms, as well as multiple
scattering events are responsible for this background deter-
mining the detection limit.18 However, the hydrogen signal
from the near surface region of the hot-wire treated sample is
clearly higher than the background and indicates a hydrogen
TABLE III. Minority carrier diffusion length LB in the base of test solar
cells before and after hydrogenation from the back or front side.
Before
LB ͓␮m͔
After
LBH ͓␮m͔
Improvement
͑LBHϪLB)/LB
Treatment
method
TABLE I. Hydrogen concentration Nsurf at the surface and Nbulk in the bulk
close to the surface after hydrogenation as determined by ERDA measure-
ments.
dc plasma ͑back͒
dc plasma ͑front͒
Hϩ implant ͑back͒
thermal anneal
hot wire ͑back͒
hot wire ͑front͒
34
34
32
34
40
49
32
46
66
0%
18%
50%
Nsurf ͓cmϪ2
͔
Nbulk ͓cmϪ3
͔
Treatment
dc plasma
Hϩ implant
hot wire H
͑3.5 Ϯ0.3)ϫ1015
͑1Ϯ0.05͒ϫ1016
͑3.5Ϯ0.3͒ϫ1015
Ͻ2ϫ1019
Ͻ2ϫ1019
34
33
38%
100%
͑8Ϯ2͒ϫ1019
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:
2170 Appl. Phys. Lett., Vol. 71, No. 15, 13 October 1997 Plieninger et al.
131.193.242.21 On: Thu, 27 Nov 2014 22:43:28

technique with respect to achieving a high hydrogen concen-
tration in the bulk. Compared to ion implantation, at least
three times more hydrogen is incorporated into bulk silicon.
Finally, hot-wire passivation has the additional advantage of
simple setup and operation.
The authors thank Mitsubishi Electric Corp. for provid-
ing polycrystalline silicon material, and in particular H.
Morikawa for performing the implantations. We are grateful
to N. Pazarkas for ERDA measurements, B. Winter and B.
Fischer for sample processing, M. Wolf for help with the
quantum efficiency measurements, J. Weber for granting ac-
cess to the dc plasma system and useful discussions, and H.
J. Queisser, K. Namba, T. Ishihara, H.-D. Carstanjen and M.
Schubert for their continuous support.
FIG. 2. Internal quantum efficiency of a test solar cell before and after hot
wire treatment from the front side. The short wavelength regime indicates
no degradation due to surface damage, the long wavelength quantum effi-
ciency increases drastically after the hydrogenation. The inset shows the
sample geometry used for the experiments.
1 J. Nijs, S. Sivoththaman, J. Szlufcik, J. Poortmans, and R. Mertens, Solid
State Phenomena 51–52, 461 ͑1996͒.
2 A. Rohatgi, S. Narasimha, and C. P. Khattak, IEEE Electron Device Lett.
17, 401 ͑1996͒.
3
¨
J. H. Werner, R. Bergmann, and R. Brendel, in Festkorperprobleme/
Advances in Solid State Physics, edited by R. Helbig ͑Vieweg,
Braunschweig/Wiesbaden, 1994͒, Vol. 34, p. 115.
solar cell processing are only achieved if the defect hydro-
genation of the cell volume does not simultaneously damage
the nϩ emitter at the surface. We investigate the surface
damage by the analysis of internal quantum efficiency mea-
surements. Figure 2 presents results for a hot-wire passivated
cell. The short wavelength regime around 450 nm reveals no
deleterious influence of the front surface hot-wire treatment,
indicating a negligible emitter damage. This lack of damage
is a direct consequence of the low energy ͑Ͻ200 meV͒ of the
hydrogen radicals. In the case of 1.2 keV ion implantation,
we find a decrease of quantum efficiency of about 50% rela-
tive in the short wavelength range. This effect of surface
damage by the impinging ions might also be reduced by
reducing the implantation energy to 400 eV.19 However, in
this case the sample has to be heated in order to guarantee a
diffusion of hydrogen from the surface into the bulk.
In conclusion, hot-wire hydrogenation represents a new
tool for the passivation of electronic defects in polycrystal-
line silicon. Hot-wire passivation from the front side in-
creases the bulk diffusion length of test solar cells by 100%
without degrading the quantum efficiency in the short wave-
length regime by surface damage of the emitter. In contrast,
ion implantation from the front side degrades the emitter
response. Back side hydrogenation with hot-wire as well as
ion implantation only yields a 50% increase in bulk diffusion
length. Hydrogenation by dc remote plasma is even less ef-
ficient. The hot-wire treatment is also the most effective
4 T. Matsuyama, M. Tanak, S. Tsuda, S. Nakano, and Y. Kuwano, Jpn. J.
Appl. Phys., Part 1 32, 3720 ͑1993͒.
5 E. Katz, M. Koltun, and L. Polyak, Solid State Phenomena 51–52, 479
͑1996͒.
6 J. C. Muller, Y. Abadou, A. Barhdadi, E. Courcelle, S. Unamuno, D.
Salles, P. Siffert, and J. Fally, Sol. Cells 17, 201 ͑1986͒.
7 T. Vermeulen, J. Poortmans, K. Said, O. Evrard, W. Laureys, M. Caymax,
J. Nijs, and R. Mertens, in Proceedings of the 25th IEEE Photovoltaics
Specialist Conference ͑IEEE, Piscataway, NJ, 1996͒, p. 653.
8 M. Heintze, R. Zedlitz, H. N. Wanka, and M. B. Schubert, J. Appl. Phys.
79, 2699 ͑1996͒.
9 A. R. Middya, J. Guillet, J. Perrin, and J. E. Bouree, Mater. Res. Soc.
Symp. Proc. 420, 289 ͑1996͒.
10 M. W. M. van Cleef, J. K. Rath, F. A. Rubinelli, C. H. M. van der Werf,
and R. E. I. Schropp, in Proceedings of the 25th IEEE Photovoltaics
Specialist Conference ͑IEEE, Piscataway, NJ, 1996͒, p. 429.
11 T. Ishihara, S. Arimoto, H. Kumabe, and T. Murotani, Prog. Photovolt. 3,
105 ͑1995͒.
12 A. Takami, S. Arimoto, H. Naomoto, S. Hamamoto, T. Ishihara, H. Kum-
abe, and T. Murotani, in Proceedings of the 1st World Conference on
Photovoltaic Energy Conversion ͑IEEE, Piscataway, NJ, 1994͒, p. 1394.
13 T. Ishihara, S. Arimoto, H. Morikawa, H. Kumabe, and T. Murotani,
Appl. Phys. Lett. 63, 3604 ͑1993͒.
14 M. Ghannam, G. Palmers, H. E. Elgamel, J. Nijs, R. Mertens, R. Peruzzi,
and D. Margadonna, Appl. Phys. Lett. 62, 1280 ͑1993͒.
15 F. H. P. M. Habraken, Nucl. Instrum. Methods Phys. Res. B 29, 181
͑1992͒.
16 S. J. Pearton, J. W. Corbett, and T. S. Shi, Appl. Phys. A: Solids Surf. 43,
153 ͑1987͒.
17 R. Brendel, M. Hirsch, R. Plieninger, and J. H. Werner, IEEE Trans.
Electron Devices 43, 1104 ͑1996͒.
18
¨
O. Kruse, Ph.D. thesis, Universitat Stuttgart, 1995.
19 H. Morikawa ͑private communication͒.
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:
Appl. Phys. Lett., Vol. 71, No. 15, 13 October 1997 Plieninger et al. 2171
131.193.242.21 On: Thu, 27 Nov 2014 22:43:28