Selective deposition of polycrystalline silicon thin films at low temperature by hot-wire chemical vapor deposition

Article abstract of DOI:10.1063/1.116280

Polycrystalline silicon thin films have been selectively deposited at a substrate temperature of 300 °C on molybdenum or silicon over silicon dioxide, silicon nitride or Corning 7059 glass substrates in a continuous hot-wire chemical vapor deposition (HWCVD) process involving hydrogen and disilane. Excellent selectivity is achieved on features as small as 1 μm spaced molybdenum lines. The deposition rate on molybdenum is 60-100 A/min which is higher than that obtained by a pulsed gas plasma-enhanced chemical vapor deposition (PECVD) or a very high-frequency PECVD (VHF-PECVD). The selective deposition rate obtained in our system is attributed to the high concentration of atomic hydrogen generated when molecular hydrogen passes over a heated tungsten filament.

Full text of DOI:10.1063/1.116280

Selective deposition of polycrystalline silicon thin films at low temperature by hotwire
chemical vapor deposition
S. Yu, E. Gulari, and J. Kanicki
Citation: Applied Physics Letters 68, 2681 (1996); doi: 10.1063/1.116280
View online: http://dx.doi.org/10.1063/1.116280
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/68/19?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Ion–surface interactions in low temperature silicon epitaxy by remote plasma enhanced chemical–vapor
deposition
J. Vac. Sci. Technol. A 14, 3024 (1996); 10.1116/1.580166
Stability of hotwire deposited amorphoussilicon thinfilm transistors
Appl. Phys. Lett. 69, 1062 (1996); 10.1063/1.116931
Effects of hydrogen surface pretreatment of silicon dioxide on the nucleation and surface roughness of
polycrystalline silicon films prepared by rapid thermal chemical vapor deposition
Appl. Phys. Lett. 69, 485 (1996); 10.1063/1.118148
Amorphous and microcrystalline silicon films deposited by hotwire chemical vapor deposition at filament
temperatures between 1500 and 1900°C
J. Appl. Phys. 79, 8748 (1996); 10.1063/1.362501
Fluorinated amorphous carbon thin films grown by helicon plasma enhanced chemical vapor deposition for low
dielectric constant interlayer dielectrics
Appl. Phys. Lett. 68, 2864 (1996); 10.1063/1.116350
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:
69.166.47.144 On: Tue, 25 Nov 2014 05:00:28

Selective deposition of polycrystalline silicon thin films at low temperature
by hot-wire chemical vapor deposition
S. Yu,a) E. Gulari,a) and J. Kanickib)
Center for Display Technology and Manufacturing, University of Michigan, Ann Arbor, Michigan 48109
͑
Received 19 December 1995; accepted for publication 1 March 1996͒
Polycrystalline silicon thin films have been selectively deposited at a substrate temperature of
00 °C on molybdenum or silicon over silicon dioxide, silicon nitride or Corning 7059 glass
3
substrates in a continuous hot-wire chemical vapor deposition ͑HWCVD͒ process involving
hydrogen and disilane. Excellent selectivity is achieved on features as small as 1 ␮m spaced
molybdenum lines. The deposition rate on molybdenum is 60–100 Å/min which is higher than that
obtained by a pulsed gas plasma-enhanced chemical vapor deposition ͑PECVD͒ or a very
high-frequency PECVD ͑VHF-PECVD͒. The selective deposition rate obtained in our system is
attributed to the high concentration of atomic hydrogen generated when molecular hydrogen passes
over a heated tungsten filament. © 1996 American Institute of Physics.
͓
S0003-6951͑96͒01619-1͔
Selective deposition of polycrystalline silicon ͑poly-Si͒
thin films has important applications in fabrication of hydro-
genated amorphous silicon ͑a-Si:H͒ thin-film transistors
clamped to a stainless-steel susceptor which is heated by two
inserted cartridge heaters. The substrate temperature is moni-
tored with a thermocouple clamped to the susceptor surface
close to the substrate. The distance between the filament and
the substrate is 4 cm. We pass hydrogen gas over the surface
of the heated filament where it is catalytically disassociated
into atomic hydrogen. Unlike the HWCVD process used by
other groups,4 we flow source gas disilane through a dis-
persal ring bypassing the filament and let it react with atomic
hydrogen downstream of the filament to form film forming
species. In this way, we avoid both the generation of atomic
Si which has a high sticking coefficient undesirable for good
quality films and the deterioration of the tungsten filament
when silicon atoms react with tungsten to form tungsten sil-
icides. The formation of tungsten silicides will also reduce
the efficiency of the tungsten wire to produce atomic hydro-
gen needed for a high rate selective deposition process. Us-
ing the process described above, we have deposited poly-Si
films at 300 °C on glass substrates with certain hydrogen to
disilane ratio.7
In this study, poly-Si thin films are deposited at a sub-
strate temperature of 300 °C, hydrogen flow rate of 48–97
sccm, disilane flow rate of 0.53–1.6 sccm, pressure of 40–65
mTorr, and filament temperature of 1800 °C. Silicon content
in the gas phase is varied by adjusting the flow rate of hy-
drogen or disilane. Molybdenum is deposited by sputtering
and then patterned by photolithography or by using a me-
chanical mask to cover part of the substrate during sputter-
ing. Under these conditions, poly-Si thin films have been
selectively deposited on molybdenum, amorphous, polycrys-
talline, and single-crystal silicon over thermal or PECVD
silicon dioxide, silicon nitride and Corning 7059 glass sub-
strates. Selective deposition on indium tin oxide ͑ITO͒, how-
ever, is difficult to achieve in our system, since reduction of
ITO by the atomic hydrogen present in the process occurs
before film growth stars. The film thickness and crystallinity
are examined by cross-section transmission electron micros-
copy ͑TEM͒. The deposition rate is calculated by dividing
the film thickness by the total process time. The selectively
deposited films on patterned substrates are examined by
scanning electron microscopy ͑SEM͒.
͑TFTs͒ for liquid crystal displays. For example, in top- or
bottom-gate TFT configuration, by selective depositing
source and drain contact layers the a-Si:H TFT array fabri-
cation process can be simplified. This simplification will pro-
vide a path to low cost TFT array manufacturing and will
extend present a-Si:H technology to a very high resolution,
large area displays. Selective deposition of microcrystalline
silicon thin films has been achieved by a pulsed gas PECVD1
or a very high-frequency plasma-enhanced chemical vapor
,8
2
deposition ͑VHF-PECVD͒ process using SiH and H as
4
2
reactants or a conventional rf-plasma CVD using SiH and
4
3
SiF mixtures. However, the growth rates of the selectively
4
deposited films by these methods are low and/or a substrate
temperature higher than 300 °C is required. In this letter, we
have demonstrated that selective deposition of polycrystal-
line silicon can be achieved by hot-wire chemical vapor
deposition ͑HWCVD͒ at a growth rate of 100 Å/min and a
substrate temperature of 300 °C with excellent selectivity
over metal lines spaced 1 ␮m apart.
HWCVD employs a heated tungsten filament instead of
a plasma to activate reactive gases. Its gas phase chemistry is
simpler than that of PECVD, there are no film damages
caused by bombardment of energetic species, and it is cost
effective for large area applications. A variety of good qual-
ity materials such as amorphous silicon and silicon nitride
have been deposited at high growth rates by this method.4
The HWCVD reactor employed in this study has been de-
scribed in some detail in previous papers.6 In brief, the
reactor consists of a cold wall 6 in. six-way stainless-steel
chamber. The reactor has a quartz viewport through which
the filament temperature is monitored with an optical pyrom-
eter. The filament unit consists of a tungsten filament of
about 1.0 m in length and 0.25 mm in diameter wrapped
around two quartz rods. Substrates of 2 in. in diameter are
,5
,7
a͒
Also with the Department of Chemical Engineering.
Also with the Department of Electrical Engineering and Computer Sci-
b͒
ence.
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. 68 (19), 6 May 1996 0003-6951/96/68(19)/2681/3/$10.00 © 1996 American Institute of Physics 2681
9.166.47.144 On: Tue, 25 Nov 2014 05:00:28
6

FIG. 1. Dependence of deposition rate and selectivity on Si content in the
gas phase.
FIG. 2. TEM cross-section micrograph with selected area diffraction pattern
The dependence of deposition rate and selectivity on Mo
of a film selectively deposited on Mo over SiO at 1.1% Si showing that the
2
and c-Si over SiO at a substrate temperature of 300 °C on Si
film is nontextured polysilicon.
2
content in the gas phase is shown in Fig. 1. Selective depo-
sition is obtained at Si content below 2.2% in the gas phase.
At higher Si content in the gas phase, selectively is lost and
served that the poly-Si film is also deposited on the side
walls of the Mo lines. The experimental conditions for this
deposition are the same as the conditions used for the film
whose TEM micrograph is shown in Fig. 2. Therefore, the
film is polycrystalline. In addition, by comparing the thick-
ness of Mo lines before and after deposition from cross-
section SEM, the thickness of the film is ϳ1200 Å for a 20
min deposition process. This gives a deposition rate of 60
Å/min, which is very close to that measured by TEM of the
films are deposited on both Mo and SiO with the net depo-
2
sition rate on Mo higher than on SiO . The selective deposi-
2
tion rate on Mo or c-Si over SiO shown in Fig. 1 is higher
2
1
than the rate obtained by the pulsed gas PECVD or
VHF-PECVD or the rf-plasma CVD using SiH and SiF
2
4
4
3
mixtures. Under the selective deposition conditions, we
have observed that a-Si:H is etched at a rate at least one
order of magnitude faster than the etching rate observed in a
VHF-PECVD system. This indicates that the concentration
of atomic hydrogen in our system is much higher than that in
a VHF-PECVD system, which implies that catalytic disasso-
ciation of molecular hydrogen by a heated tungsten filament
is a much more efficient way than a plasma in generating
atomic hydrogen.
The generation rate of atomic hydrogen increases with
filament temperature and hydrogen pressure. At higher fila-
ment temperatures contamination by tungsten will become a
problem. However, at a filament temperature of below
2
100 °C, tungsten level in a HWCVD silicon film is at the
9
detection limit of secondary ion mass spectroscopy. There-
fore, at a filament temperature of 1800 °C used in this study,
tungsten contamination should not be a concern.
The TEM cross-section micrograph with the selected
area electron diffraction pattern of a film selectively depos-
ited on Mo over SiO at a Si content of 1.1% in the gas phase
2
is shown in Fig. 2. The TEM micrograph as well as the
selected area diffraction pattern shows that the film is non-
textured poly-Si with randomly oriented crystallites. When
they are examined by cross-section TEM, the poly-Si films
nonselectively deposited on both Mo and SiO at higher Si
2
content in the gas phase exhibit similar microstructure and
crystallinity.
The SEM micrographs of patterned Mo lines of 1720 Å
thick with a spacing of 1 ␮m on SiO before and after selec-
2
tive deposition are shown in Fig. 3. It can be seen that the
poly-Si film is deposited on 1 ␮m spaced Mo lines but not on
FIG. 3. SEM micrographs of 1 ␮m spaced Mo lines over SiO2 before
deposition ͑a͒, and after deposition ͑b͒.
SiO . By cross-section SEM of this sample, we have ob-
2
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:
682 Appl. Phys. Lett., Vol. 68, No. 19, 6 May 1996 Yu, Gulari, and Kanicki
9.166.47.144 On: Tue, 25 Nov 2014 05:00:28
2
6

film selectively deposited under the same condition on Mo
does not provide an explanation for the difference of nucle-
ation on different substrates. In our opinion, the microscopic
mechanism of substrate selectivity still remains unclear and
further investigation is needed.
over SiO of a much larger area.
2
It has been found that atomic hydrogen plays a key role
in substrate selective deposition.2 The net growth rate of
the selective deposition is a result of the difference of a sili-
con deposition rate and etching rate of strained Si–Si bonds
by atomic hydrogen. The substrate selectivity is attributed to
the difference of nucleation of Si–Si bond strength on dif-
ferent substrates. However, the origin of the difference of
nucleation on different substrates is not pinpointed. Recently,
Roca i Cabarrocas et al. have suggested a-Si:H nucleation
versus homogeneous growth mechanisms to explain the sub-
,10
1
G. N. Parsons, Appl. Phys. Lett. 59, 2546 ͑1991͒.
W. Westlake and M. Heintze, J. Appl. Phys. 77, 879 ͑1995͒.
K. Baert, P. Deschepper, J. Nijs, and R. Mertens, Mater. Res. Soc. Symp.
2
3
Proc. 164, 395 ͑1989͒.
A. H. Mahan, J. Carapella, B. P. Nelson, and R. S. Crandall, J. Appl. Phys.
4
69, 6728 ͑1991͒.
5
S. V. Deshpande, J. L. Dupuie, and E. Gulari, Appl. Phys. Lett. 61, 1420
͑1992͒.
J. L. Dupuie and E. Gulari, J. Vac. Sci. Technol. A 10, 18 ͑1992͒.
S. Yu, S. V. Deshpande, E. Gulari, and J. Kanicki, Mater. Res. Soc. Symp.
Proc. 377, 69 ͑1995͒.
H. Matsumura, Y. Tashiro, K. Sasaki, and S. Furukawa, Jpn. J. Appl.
Phys. 33, L1209 ͑1994͒.
C. Horbach, W. Beyer, and H. Wagner, J. Non-Cryst. Solids 137&138,
6
7
1
1
strate selectivity. However, contradictory to the homoge-
neous growth on a-Si:H observed by Roca i Cabarrocas
8
9
0
1
1
et al., we have deposited poly-Si selectively on a-Si:H sub-
strate that has been etched during the selective deposition
process. Etching of a-Si:H substrate is in agreement with the
mechanism that etching by atomic hydrogen takes place dur-
ing the initial nucleation process. However, this mechanism
661 ͑1991͒.
1
J. J. Boland and G. N. Parsons, Science 256, 1304 ͑1992͒.
P. Roca i Cabarrocas, N. Layadi, T. Heitz, and B. Drevillon, Appl. Phys.
Lett. 66, 3609 ͑1995͒.
11
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. 68, No. 19, 6 May 1996 Yu, Gulari, and Kanicki 2683
69.166.47.144 On: Tue, 25 Nov 2014 05:00:28