Low-temperature growth of crystalline silicon on a chlorine-terminated surface

Article abstract of DOI:10.1063/1.125373

We report on the role of surface termination during growth of crystalline silicon at low temperatures. Macrocrystalline silicon was fabricated using plasma-enhanced chemical-vapor deposition with a hydrogen-diluted dichlorosilane (SiH₂Cl₂)/monosilane (SiH₄) mixture to study the role of hydrogen and chlorine in crystal formation. When varying the fraction of SiH₂Cl₂, x = [SiH₂Cl₂]/([SiH₄]+[SiH₂Cl ₂]), good crystallinity was obtained for x = 0 and 1, whereas the crystallinity was markedly deteriorated for intermediate x. Optical emission spectroscopy of the plasma suggests that film precursors different from SiH_x fragments and atomic chlorine are generated for x ≠ 0 and that atomic hydrogen is generated in all cases. Infrared reflection absorption spectroscopy indicates that the surface coverage is hydrogen for x = 0, chlorine for x = 1, and a hydrogen-chlorine mixture for intermediate x. These results imply that low-temperature growth of crystalline silicon is facilitated on a chlorine-terminated surface as well as on a hydrogen-terminated surface under the presence of atomic hydrogen. The cooperative roles of chlorine and hydrogen are proposed in the crystal growth of Si.

Full text of DOI:10.1063/1.125373

Low-temperature growth of crystalline silicon on a chlorine-terminated surface
Lihui Guo, Yasutake Toyoshima, Michio Kondo, and Akihisa Matsuda
Citation: Applied Physics Letters 75, 3515 (1999); doi: 10.1063/1.125373
View online: http://dx.doi.org/10.1063/1.125373
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/75/22?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Low-temperature synthesis of homogeneous nanocrystalline cubic silicon carbide films
J. Appl. Phys. 102, 056101 (2007); 10.1063/1.2776155
Microstructure and initial growth characteristics of the low temperature microcrystalline silicon films on silicon
nitride surface
J. Appl. Phys. 90, 217 (2001); 10.1063/1.1378334
Particle generation and thin film surface morphology in the tetraethylorthosilicate/oxygen plasma enhanced
chemical vapor deposition process
J. Appl. Phys. 88, 3047 (2000); 10.1063/1.1287783
In situ Fourier transform infrared spectroscopy and stochastic modeling of surface chemistry of amorphous
silicon growth
J. Appl. Phys. 83, 6096 (1998); 10.1063/1.367503
Aluminum mediated low temperature growth of crystalline silicon by plasma-enhanced chemical vapor and
sputter deposition
Appl. Phys. Lett. 72, 1510 (1998); 10.1063/1.121042
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:
128.82.252.58 On: Tue, 23 Dec 2014 16:47:19

APPLIED PHYSICS LETTERS
VOLUME 75, NUMBER 22
29 NOVEMBER 1999
Low-temperature growth of crystalline silicon
on a chlorine-terminated surface
Lihui Guo,a) Yasutake Toyoshima, Michio Kondo,b) and Akihisa Matsuda
Thin Film Si Solar Cells Super Laboratory, Electrotechnical Laboratory, Tsukuba, Ibaraki 305, Japan
͑
Received 13 July 1999; accepted for publication 7 October 1999͒
We report on the role of surface termination during growth of crystalline silicon at low
temperatures. Microcrystalline silicon was fabricated using plasma-enhanced chemical-vapor
deposition with a hydrogen-diluted dichlorosilane (SiH Cl )/monosilane (SiH ) mixture to study
2
2
4
the role of hydrogen and chlorine in crystal formation. When varying the fraction of SiH Cl ,
2
2
xϭ͓SiH Cl ͔/͓͑SiH ͔ϩ͓SiH Cl ͔͒, good crystallinity was obtained for xϭ0 and 1, whereas the
2
2
4
2
2
crystallinity was markedly deteriorated for intermediate x. Optical emission spectroscopy of the
plasma suggests that film precursors different from SiH fragments and atomic chlorine are
x
generated for xÞ0 and that atomic hydrogen is generated in all cases. Infrared reflection absorption
spectroscopy indicates that the surface coverage is hydrogen for xϭ0, chlorine for xϭ1, and a
hydrogen–chlorine mixture for intermediate x. These results imply that low-temperature growth of
crystalline silicon is facilitated on a chlorine-terminated surface as well as on a hydrogen-terminated
surface under the presence of atomic hydrogen. The cooperative roles of chlorine and hydrogen are
proposed in the crystal growth of Si. © 1999 American Institute of Physics.
͓
S0003-6951͑99͒01648-4͔
Low-temperature growth (TϽ450 °C) of crystalline sili-
con is of intense interest from the viewpoint of applications
to low-cost devices on glass substrates, such as solar cells
and thin-film transistors. Plasma-enhanced chemical-vapor
deposition ͑PECVD͒ has been widely used for the growth of
crystalline silicon with a SiH /H gas mixture at 13.56 MHz.
plasma was performed to investigate generated radicals in
the plasma using a highly sensitive charge-coupled device
camera ͑Princeton Research i-max 500͒. Surface termination
was observed using infrared reflection absorption spectros-
copy.
4
2
Figure 1 shows the deposition rate and crystallinity as a
function of the SiH Cl2 fraction, xϭ͓SiH Cl ͔/͓͑SiH ͔
Although the mechanism of crystal formation at low tem-
peratures is still controversial,1 hydrogen is considered to
play an important role in the formation mechanism. On the
other hand, as an alternative method, dichlorosilane
2
2
2
4
–4
ϩ͓SiH Cl ͔͒,
maintaining
the
flow
rates
of
2
2
͓
SiH ͔ϩ͓SiH Cl ͔ and ͓H ͔ constant at 10 and 390 sccm,
4 2 2 2
respectively. The growth temperature was 350 °C. As re-
9
(
SiH Cl ) and tetrafluorosilane (SiF ) source gases have
2 2 4
ported elsewhere, a high deposition pressure of 4 Torr was
5
–7
been examined with the addition of hydrogen. A growing
surface when using SiH /H has been studied by infrared
used in this study, and thereby a deposition rate of 8 Å/s was
obtained for xϭ0. With increasing x, the growth rate is
slightly increased and the crystallinity deteriorates abruptly.
4
2
8
reflection absorption spectroscopy ͑IRRAS͒, whereas less
knowledge has been obtained upon growth in halogen cases.
In this letter, we report a study of crystalline silicon
growth at temperatures of 200–350 °C using rf-PECVD with
a SiH /SiH Cl /H gas mixture to clarify the role of halo-
4
2
2
2
gens and hydrogen in crystal growth. It was found that either
a SiH /H or SiH Cl /H source gas mixture results in crys-
4
2
2
2
2
tal formation and that a mixture of SiH and SiH Cl dete-
4
2
2
riorates the crystallinity. The surface termination was identi-
fied to be halogen in the SiH Cl /H case using IRRAS
2
2
2
observation. Crystal formation on a halogen-terminated sur-
face is discussed in relation to the hydrogenated surface.
Crystalline Si films have been deposited on glass sub-
strates from dichlorosilane (SiH Cl )/monosilane (SiH )
2
2
4
mixtures diluted by H2 using a conventional rf-PECVD
method. The crystallinity of the samples was evaluated from
a Raman spectrum by taking the ratio of the signal height for
Ϫ1
the crystalline peak at 520.5 cm to the amorphous peak at
4
FIG. 1. Growth rate and Raman crystallinity as a function of dichlorosilane
Ϫ1
80 cm . Optical emission spectroscopy ͑OES͒ of the
fraction, xϭ͓SiH2Cl2͔/͓͑SiH2Cl2͔ϩ͓SiH4͔). The Raman crystallinity was
Ϫ1
obtained from the ratio of the intensities at 520 cm (c-Si phase͒ to the
Ϫ1
value at 480 cm (a-Si phase͒, I /I . The total flow rate of SiH and
c
a
4
a͒
NEDO Fellow.
Electronic mail: mkondo@etl.go.jp
SiH Cl was 10 sccm, hydrogen flow rate 390 sccm, total pressure 4 Torr,
2 2
b͒
and rf power 100 W applied to a cathode 10 cm in diameter.
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:
003-6951/99/75(22)/3515/3/$15.00 3515 © 1999 American Institute of Physics
28.82.252.58 On: Tue, 23 Dec 2014 16:47:19
0
1

3516
Appl. Phys. Lett., Vol. 75, No. 22, 29 November 1999
Guo et al.
FIG. 3. Optical emission intensity from the plasma as a function of the
SiH Cl fraction for the Cl* ͑755.4 nm͒ and H ͑655 nm͒ lines. The flow
2
2
␣
rate of SiH Cl was constant at 10 sccm, and only the hydrogen flow rate
2
2
was varied. Total pressure was 0.4 Torr, and rf power 20 W.
was used to normalize all the previous spectra. As a result,
Ϫ1
FIG. 2. Optical emission intensity from the plasma as a function of dichlo-
Si–D bands at around 1500 cm are seen as upward peaks.
x
rosilane fraction, xϭ͓SiH Cl ͔/͓͑SiH Cl ͔ϩ͓SiH ͔) in percent for H ͑655
2
2
2
2
4
␣
For xϭ0, the surface is mostly terminated by hydrogen as
reported before. For xϭ1, we could not observe surface
hydrogen at all, but did observe surface Si–Cl around 600
cm . These surface diagnosis results and their relevance to
the crystallinity shown in Fig. 1 suggest that a growing sur-
face covered by chlorine facilitates crystal growth as well as
a hydrogen covering surface. We also confirmed that coex-
isting hydrogen and chlorine termination disrupt the crystal
growth. Since the growth temperature in this work is much
lower than the thermal desorption threshold of hydrogen–
halogen-related species from the crystalline silicon surface,
the surface dangling bonds are expected to be mostly termi-
nated by chlorine for xϭ1.
The surface termination under the coexistence of atomic
hydrogen and chlorine flux was studied using molecular
beams generated by pyrolisis of H2 and Cl2 gases,
nm͒, Si* ͑SiCl͒ ͑288 nm͒, and SiH* ͑414 nm͒ lines. Experimental condi-
tions for the plasma were the same as those for Fig. 1.
8
Ϫ1
For intermediate fractions, xϭ0.2 and 0.4, no crystalline
phase appeared. Increasing x greater than 0.5 results in a
reduction in the growth rate, which may be attributed to
higher dissociation energy of SiH Cl than SiH , and a crys-
10
2
2
4
talline phase appears again. For xϭ1, the crystallinity is
higher than that for xϭ0.
OES measurements were carried out for various dichlo-
rosilane fractions x, and we monitored several lines related to
11
Si* ͑or SiCl*) ͑288 nm͒, SiH* ͑414 nm͒, H ͑656 nm͒, and
␣
Cl* ͑755 nm͒. Since Si* and SiCl* lines have almost iden-
tical photon energies, the line at 288 nm may contain the
contribution of both the Si* and SiCl* lines. Figure 2 shows
the SiH Cl fraction dependence of Si*, SiH*, and H emis-
12
2
2
␣
respectively, and it was reported that the surface termina-
sion intensities I(Si*, SiH*, H ). The spatial distribution of
␣
tion at room temperature is mostly chlorine, which is consis-
tent with the IRRAS results in this work, although a differ-
ence in the flux density of hydrogen and chlorine and in
substrate temperature should be taken into account.
emission intensity is integrated over the electrode spacing to
obtain I(Si*, SiH*, H ). With increasing x, the I(H ), and
␣
␣
I(Si*) are almost constant, and only the I(SiH*) decreases
linearly. For xϭ1.0, the growth rate was proportional to
I(Si*ϩSiCl*) emission intensity, but not to the I(SiH*).
This is different from the pure silane (xϭ0) case.10 When
varying the hydrogen dilution ratio of dichlorosilane, the
I(Cl*) increases with increasing SiH Cl fraction, while the
2
2
I(H ) decreases, as shown in Fig. 3. These OES results sug-
␣
gest the coexistence of atomic chlorine and hydrogen and
their competing reaction on the growing surface depending
on x.
Surface diagnosis using IRRAS has been carried out to
identify the surface-termination species in a different appa-
ratus under identical conditions to those for these deposi-
tions. First, a SiH Cl /H mixture ͑2/100 sccm, 350 mTorr,
2
2
2
2
40 °C͒ plasma was operated to grow the film. IRRAS spec-
tra of a chlorinated surface were observed during the growth
Fig. 4͑a͔͒, as well as after the growth ͓Fig. 4͑b͔͒. Then,
͓
FIG. 4. Infrared absorption spectra observed by an IRRAS technique for ͑a͒
the growing surface during deposition with dichlorosilane, ͑b͒ the surface
after the growth of ͑a͒, and ͑c͒ the hydrogenated surface after deposition
with silane. All the spectra were normalized by the spectrum for the deuter-
several monolayers were deposited at xϭ0 ͓Fig. 4͑c͔͒ to
achieve a totally hydrogenated surface. Finally, deuterium
plasma exposure was performed to completely exchange the
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:
surface hydrogen with deuterium, so the last IR spectrum
ated surface ͑not shown͒.
1
28.82.252.58 On: Tue, 23 Dec 2014 16:47:19

Appl. Phys. Lett., Vol. 75, No. 22, 29 November 1999
Guo et al.
3517
A correlation between crystal formation and surface hy-
One of the authors ͑L.G.͒ would like to thank the New
Energy and Industrial Technology Development Organiza-
tion ͑NEDO͒ for their support of this work through a fellow-
ship.
drogen coverage has been pointed out in the silane case (x
2
,10
ϭ0).
On a surface partly terminated by hydrogen at high
1
0
temperatures, the crystallinity is deteriorated. In addition,
in molecular beam epitaxy of silicon, the crystallinity dete-
riorates with surface hydrogen coverage.13 These results
have a common aspect with the results in the present study:
Crystal growth is facilitated on a ‘‘homogeneous’’ surface
but disrupted on an ‘‘inhomogeneous’’ surface. In crystal
growth, adsorbates on the surface should find an energeti-
cally stable site to form a crystalline network through surface
diffusion. If the surface is inhomogeneous, adsorbates will
suffer from having different energy barriers during diffusion,
resulting in a reduced diffusion length, i.e., a kind of local-
ization takes place. Since hydrogen dilution is necessary
even in the dichlorosilane process, we suggest a cooperative
effect of atomic hydrogen flux onto the surface and chlorine
termination in crystal formation at low temperatures.
1
S. Veprek, Z. Iqbal, and F. A. Sarrot, Philos. Mag. B 45, 137 ͑1982͒.
A. Matsuda, J. Non-Cryst. Solids 59&60, 767 ͑1983͒.
C. C. Tsai, G. B. Anderson, and R. Thompson, Mater. Res. Soc. Symp.
2
3
Proc. 192, 475 ͑1990͒.
T. Akasaka and I. Shimizu, Appl. Phys. Lett. 66, 3441 ͑1995͒.
T. Arai, T. Nakamura, and H. Shirai, Jpn. J. Appl. Phys., Part 1 36, 4907
4
5
͑
1997͒.
6
K. Nakahata, A. Miida, T. Kamiya, Y. Maeda, C. M. Fortmann, and I.
Shimizu, Jpn. J. Appl. Phys., Part 2 37, L1026 ͑1998͒.
R. Platz and S. Wagner, Appl. Phys. Lett. 73, 1236 ͑1998͒.
Y. Toyoshima, K. Arai, A. Matsuda, and K. Tanaka, Appl. Phys. Lett. 56,
7
8
1540 ͑1990͒.
9
L. Guo, M. Kondo, M. Fukawa, K. Saitoh, and A. Matsuda, Jpn. J. Appl.
Phys., Part 2 37, L1116 ͑1998͒.
A. Matsuda and T. Goto, Mater. Res. Soc. Symp. Proc. 164, 3 ͑1989͒.
P. A. Coon, P. Gupta, M. L. Wise, and S. M. George, J. Vac. Sci. Technol.
A 10, 324 ͑1992͒.
M.-C. Chuang and J. W. Coburn, J. Vac. Sci. Technol. A 8, 1969 ͑1990͒.
D. J. Eaglesham, F. C. Unterwald, H. Luftman, D. P. Adams, and S. M.
Yalisove, J. Appl. Phys. 74, 6615 ͑1993͒.
In summary, we have provided evidence of the enhanced
growth of crystalline silicon on a chlorine-terminated surface
at low temperatures. The mechanism of the chlorine termi-
nation and its beneficial role are discussed in terms of the
surface chemistry, and a similarity to the growth on a hydro-
genated surface is also discussed.
1
1
0
1
1
2
13
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:
28.82.252.58 On: Tue, 23 Dec 2014 16:47:19
1