Low-temperature growth of polycrystalline Si and Ge films by redox reactions of Si2H6 and GeF4

Article abstract of DOI:10.1016/S0022-328X(00)00400-9

Low-temperature growth of polycrystalline Si and Ge films has been investigated by low-pressure chemical vapor deposition (LP-CVD) featuring a redox reaction of disilane (Si₂H₆) and Germanium tetrafluoride (GeF₄). The film growth is established in a wide range of 0.4-100 torr and in low-temperature range of 300-450°C. The Si content significantly depends on the reaction modes in which either Si₂H₆ or GeF₄ is activated preferentially, and thus the film composition is varied from > 90 atm% Si to > 90 atm% Ge. It is found that the isolated nuclei are formed directly on the substrate surface and provides a good basis of high crystallinity in a thin film irrespective of the film contents. The mechanism of film growth and the origin of low-temperature crystal growth are discussed from a chemical point of view according to the experimental results.

Full text of DOI:10.1016/S0022-328X(00)00400-9

www.elsevier.nl/locate/jorganchem
Journal of Organometallic Chemistry 611 (2000) 531–536
Low-temperature growth of polycrystalline Si and Ge films by
redox reactions of Si2H6 and GeF4
Jun-ichi Hanna *, Kousaku Shimizu
Imaging Science and Engineering Laboratory, Tokyo Institute of Technology, Nagatsuta Midori-ku, Yokohama 226-8503, Japan
Received 3 March 2000; accepted 26 March 2000
Abstract
Low-temperature growth of polycrystalline Si and Ge films has been investigated by low-pressure chemical vapor deposition
(LP-CVD) featuring a redox reaction of disilane (Si2H6) and Germanium tetrafluoride (GeF4). The film growth is established in
a wide range of 0.4–100 torr and in low-temperature range of 300–450°C. The Si content significantly depends on the reaction
modes in which either Si2H6 or GeF4 is activated preferentially, and thus the film composition is varied from \90 atm% Si to
\90 atm% Ge. It is found that the isolated nuclei are formed directly on the substrate surface and provides a good basis of high
crystallinity in a thin film irrespective of the film contents. The mechanism of film growth and the origin of low-temperature
crystal growth are discussed from a chemical point of view according to the experimental results. © 2000 Elsevier Science S.A. All
rights reserved.
Keywords: Germanium tetrafluoride; Disilane; Thermal CVD; Polysilicon; Crystal growth
1. Introduction
tive doping efficiency, and a large absorption coefficient
in infra-red region compared with the a-Si:H thin films.
Several preparation techniques for low-temperature
growth of poly-Si have been proposed, which include
plasma CVD [3], Hot-wire CVD [4], and photo-CVD
[5]. Within these techniques, however, remain some
problems to be solved: the crystallinity is poor at an
early stage of film growth, resulting in inhomogeneous
crystallinity and a low crystal fraction; high crystallinity
and high rate growth hardly go together. The former is
attributed to delayed nuclei formation during film
growth and their competitive growth to the grains
[6a,b] and the latter to the condensation of reactive
species, ie., film precursors generated in the gas phase,
resulting in powder formation. The latter problem is
not so easy for us to solve in these conventional CVDs,
because it is associated with a principle of their low-
temperature film growth: high deposition rate requires
high concentration of the precursors, resulting in their
gas phase condensation.
Nowadays, the hydrogenated amorphous silicon (a-
Si:H) thin film prepared by rf-grow discharge of silane
(SiH4) is established for large-area electronic devices
such as thin film transistors (TFTs) for liquid crystal
displays, solar cells, and xerographic photoreceptors for
copiers and laser printers. The a-Si:H films, however,
remains a serious problem of photo-induced degrada-
tion of electrical properties [1], and basically can not
provide us with high mobility over 10 cm2 V−1 s−1 due
to the amorphous nature [2]. Thus, a new material is
required for advanced large-area electronic devices such
as TFT arrays with driving circuits for low-cost and
high performance display devices and stacked thin film
solar cells for power generation.
Polycrystalline Si (poly-Si) films prepared at a tem-
perature where glass substrates are readily employed
are a promising candidate for such advanced large-area
electronic devices, because of a good compatibility with
conventional Si device fabrication processes and of
superior electronic properties, e.g., high mobility, effec-
To solve this dilemma, we have proposed a new
principle for low-temperature growth of films without
decomposing the source materials in the gas phase, i.e.,
reactive thermal CVD [7].
* Corresponding author. Fax: +81-45-9211492.
E-mail address: hanna@isl.titech.ac.jp (J.-i. Hanna).
0022-328X/00/$ - see front matter © 2000 Elsevier Science S.A. All rights reserved.
PII: S0022-328X(00)00400-9

532
J.-i. Hanna, K. Shimizu / Journal of Organometallic Chemistry 611 (2000) 531–536
For poly-Si film growth, we selected disilane (Si2H6)
silane molecules [10]; in hot-wire CVD, the heated
W-filaments promote the decomposition of silane
molecules [4]. In all these CVD processes, the reactive
Si-related radicals are responsible for film growth at
lower temperatures where the parent molecule, e.g.,
SiH4 never decomposes so as to deposit the films.
This principle, however, has a dilemma about the
low-temperature film growth and high rate film growth
for large-area applications, because the high concentra-
tion of the reactive species in the gas phase leads to
powder formation. One of the solutions to solve this
dilemma is to confine the reactive zone where the
source materials are excited to decompose like the
thermal CVD, in which the source materials are ther-
mally activated to decompose in the vicinity of the
heated substrate.
and germanium tetrafluoride (GeF4) as a set of the
reactive source materials. Interestingly, we found that
the crystal growth takes place at 450°C or less with
compositional variety of Ge-rich to Si-rich and that
isolated nuclei are formed directly on the SiO2 substrate
and the selective growth can be achieved by choosing
growth parameters [8].
In this report, we will discuss the mechanism of film
growth and the origin of low-temperature crystal
growth from a chemical point of view on the basis of
the experimental results.
2. Results and discussion
In a new principle, the redox reactions in a set of the
source materials is featured to promote their decompo-
sition leading to film growth as shown in Fig. 1. In fact,
it was demonstrated in SiH4–F2 system that chemical
reactions between SiH4 and F2 at the reduced pressure
are effective to generate the reactive species and deposit
Si films at 250°C [11]. This chemical process, however,
can not escape from the above dilemma because of the
spontaneous gas phase reactions of SiH4 and F2 at the
ambient temperature. Thus, a new redox pair of source
materials was chosen, i.e., Si2H6 and GeF4, which can
react with each other only when they are activated
thermally or catalytically on the heated surface. In fact,
the film deposition took place at heated substrate while
no decomposition of these source gases is observed at
the ambient temperature, which is confirmed by in-situ
IR spectroscopy. [12]
For the metalization process in semiconductor tech-
nologies, W-CVD and its related processes [13] has
been investigated, in which SiH4 and WF6 are used for
the source materials. The basic concept of this film
growth is very similar to the present principle, but it
was not recognized well in such a manner described
here.
2.1. A new principal for low-temperature film growth
In order to reduce the growth temperature for film
deposition in CVD processes, there are two approaches:
one is to find out a new source material exhibiting a
lower pyrolytic temperature: the other is to assist the
decomposition of source materials with external energy.
As far as the Si-related compounds are concerned, there
are few materials with pyrolytic temperatures lower
than those of SiH4 (\500°C) and Si2H6 (\350°C)
readily available. Therefore, the major effort to estab-
lish a low-temperature CVD process has been made in
the latter approach.
In fact, in the low-temperature deposition of Si films,
the source materials are decomposed at a certain region
away from the heated substrate where the external
energy is provided to excite the source materials for
promoting their decomposition: in the plasma CVDs,
for example, SiH4 is decomposed in the plasma where
accelerated electrons give kinetic energy to silane
molecules, resulting in the dissociation of electronically
excited silane molecules [9]; in the Hg-sensitized photo-
CVD, photo-excited Hg atoms transfer their energy to
2.2. Growth of Ge-rich films
In the combination of Si2H6 and GeF4, there is a
chance for Ge to be deposited when Si2H6 works as a
reducing agent and is oxidized into stable final products
such as SiF4 and SiHF3 without forming Si-related
reactive intermediates on the surface and/or in its vicin-
ity. In fact, it was found that Ge-rich (\90 atm%)
films could be deposited very easily [14]. Interestingly,
there was a peak at around 375°C in the deposition rate
as a function of temperature as shown in Fig. 2. This
indicates that there exist two film growth modes con-
cerning film deposition. In the low temperature region
of B400°C, epitaxial growth of Ge-rich film took place
when Si (100) wafer was used even at the temperature
Fig. 1. Concept of reactive thermal CVD.

J.-i. Hanna, K. Shimizu / Journal of Organometallic Chemistry 611 (2000) 531–536
533
200°C) is formed on the Si wafer to which GeF4 is
exposed [15] and polymeric deposits of GeF2, ie.,
(GeF2)n is formed on the wall [12]. Therefore, it is
plausible that GeF4 is activated on the heated Si sub-
strate when absorbed and results in being reduced
easily into Ge in the presence of the reducing agent,
Si2H6, even at the low temperature of 300°C where
none of GeF4 and Si2H6 are decomposed.
Interestingly, selective growth in which the film
growth occurs only on bare Si surface and not on the
SiO2 [16] when SiO2-patterned Si wafer was applied as
a substrate as shown in Fig. 3. This fact provides us
with a confirmation of the present idea, because no
activation of GeF4 results in no film deposition even at
375°C.
At a high temperature region over 400°C, non-selec-
tive growth took place, in which the film deposition
occurred irrespective of substrate materials. In this
regime, the continuous nucleation took place on the
substrate surface at the initial stage of film growth and
it enhanced when the reaction pressure was increased.
This suggests that nucleation took place homoge-
neously, i.e., in the gas phase near the substrate surface.
The final conclusion can not be drawn until what
happens in the vicinity of the surface is clarified, but we
speculate that thermally activated Si2H6 and GeF4
could react each other in the gas phase near the surface,
resulting in Ge cluster formation which leads to Ge
nuclei on the surface.
Fig. 2. Ge growth rate as a function of substrate temperature. GeF4
flow rates were 1.5 (open circles) and 2 sccm (closed circles), while
Si2H6 and He of dilutent gas flow rates were fixed at 30 and 300 sccm,
respectively, and total pressure was at 1 Torr.
As discussed above, it is concluded that two chemical
mechanisms are concerned with the low-temperature
growth of Ge-rich films: one is catalytic activation of
GeF4 on the surface, which makes it possible to selec-
tive and epitaxial growth of Ge-rich films; the other is
thermal acceleration of reaction rate between GeF4 and
Si2H6, which leads to non-selective polycrystalline
growth. In this film growth, the reaction pressure could
be increased up to 100 Torr. This is one of good
advantages over the plasma processes that can not be
applied in such a high pressure, where it helps the
substrate temperature keep uniform and reduces con-
tamination from residual gas.
Interestingly, in polycrystalline growth of Ge, iso-
lated nuclei were formed on the SiO2 substrate without
accompanying any amorphous tissue. Thus, with the
combination of this nucleation and selective growth of
resulting nuclei, the high crystallinity SiGe was pre-
pared [17].
Fig. 3. SEM micrograph of Ge films selectively grown on Si surface
on SiO2-patterned Si substrate.
as low as 300–320°C. It should be noted that neither
GeF4 nor Si2H6 can be decomposed thermally in this
temperature range. The crystallinity of the epitaxial
films was degraded as the growth temperature was
increased over 400°C. It turned out that this degrada-
tion is due to the nucleation of Ge crystallites while the
grain growth takes place simultaneously. In a typical
epitaxial growth of Si, the higher the growth tempera-
ture is, the better the crystallinity is established, because
the crystal is the most stable form in thermal equi-
librium condition.
2.3. Growth of Si-rich films
It is well known that GeF4 exhibits etching activity
for Si. This fact indicates that GeF4 is activated to be
reduced probably into GeF2 when absorbed on the Si
surface. In fact, it was revealed from ESCA and in-situ
IR measurements that very thin deposit of Ge (100–
As described in the previous section, Ge-rich film is
easily deposited at the lower temperature, but it is
difficult to increase Si content in the film when the
substrate temperature was set below 450°C. The Si
content was 20 atm% at most in spite of a large excess

534
J.-i. Hanna, K. Shimizu / Journal of Organometallic Chemistry 611 (2000) 531–536
gas flow ratio of S2H6 against GeF4, e.g. 7–8. It was
found that Si content can be increased up to \90
atm% when the substrate temperature was increased
over 450°C and a higher gas flow ratios over 10 were
applied [17]. As is clear in Fig. 4, the Si content of films
exceeds 95 atm%, which is estimated on the basis of
Vegard’s law from the X-ray diffraction peak of Si
(220) in the film. As for the deposition rate, it was
increased with an increase in the GeF4 flow rate, but
the normalized deposition rate with a given GeF4 flow
rate was almost constant in spite of the film contents.
This fact indicates that GeF4 plays the major role for
film deposition even in the growth condition where the
film is rich in Si. In this figure, the Ge content is
increased very steeply when the GeF4 gas flow rate was
increased over 1.3 sccm. This characteristic behavior of
this materials system should be related to the catalytic
effect of Ge surface on the activation of source materi-
als, i.e. GeF4.
On the basis of the fact that the Si incorporation into
the film significantly depends on the substrate tempera-
ture, we can point out that thermal activation of Si2H6
is an essential requirement to achieve effective Si incor-
poration into the films, so that the GeF4 may help
promoting not only the film growth but also structural
relaxation of Si-network into the crystal growth. This
will be discussed in more detail later on.
2.4. Low-temperature crystal growth
The crystal growth of Ge is achieved at 350°C by the
thermal CVD using GeH4 as a source material [19]. In
this process, the growth temperature of films is deter-
mined primarily by the pyrolytic temperature of GeH4.
It is well known that the bulk crystallization tempera-
ture of Ge is 400°C, so that it is natural that the crystal
growth occurs at that temperature. However, it is not
natural that the crystal growth of Si-rich film whose Ge
content is less than 5 atm% occurs at the temperatures
lower than 500°C, because incorporated Ge in 5 atm%
or less does not change much in the crystallization
temperature of films. In fact, the thermal CVD process
for crystalline SiGe alloy requires high growth tempera-
tures over 550°C when GeH4 and Si2H6 were employed
as source materials [20]. In addition, it should be noted
that thermal decomposition of Si2H6 always afford
amorphous films when the substrate temperature is
lower than 500°C. Therefore, some specific mechanism
of crystallization has to be associated with the present
crystal growth of Si-rich films at 450°C.
The low-temperature crystal growth has been re-
ported so far in various CVD processes including
SiH4–H2 system with plasma [21] or hot W filaments,
SiF4–H2 plasma systems [4], SiH2Cl2–H2 plasma sys-
tems [22], and SiH4–F2 and Si2H6–GeF4 reactive sys-
tems [11,17]. There are several mechanisms proposed
for explaining the low-temperature growth of crys-
talline Si in each CVD process: Veprek et al. [23]
discussed the plasma condition where the crystal
growth takes place and proposed a chemical transport
of Si in the equilibrium condition in the presence of
atomic hydrogen; Tsai et al. [21] proposed preferential
etching of amorphous tissue by atomic hydrogen for
the crystal growth; Matsuda pointed out the impor-
tance of hydrogen coverage when the crystal growth
takes place and explained that the hydrogen covered
surface helps migrating the film precursors to find out
the sites energetically favorable which leads to the
crystal growth [24]. According to the fact that the film
surface is covered with chlorine in the growth condition
where the crystal growth takes place in SiH2Cl2–H2
plasma [22], Matsuda et al. discussed this in terms of
the effective migration of the precursors on the film
surface.
In this process, immediate and isolated nucleation
took place on the SiO2 surface as described in the Ge
film growth. Furthermore, the selective growth was also
realized by choosing the growth condition. Thus, the
combination of this step by step process, the nuclei
formation on the substrate and its selective growth to
grains provided us with high crystallinity poly-Si-rich
SiGe films at 450°C [18].
In most of the CVD processes cited above, hydrogen
plasma is associated with the film growth. Indeed all
these explanation have a good basis for low-tempera-
ture crystal growth, but SiH4–F2 system is out of this
discussion because of no atomic hydrogen incorpora-
tion in the crystal growth. Furthermore, the hydrogen
coverage of the surface can not be expected in Si2H6–
GeF4 system because of high growth temperature over
400°C where thermal dissociation of HꢀSi bonds in the
surface are activated.
Fig. 4. Si-molar fraction and deposition rates of films as a function of
GeF4 flow rate. Si2H6 and He flow rates were fixed at 20 and 300
sccm, respectively. Total pressure was kept constant at 0.45 Torr.

J.-i. Hanna, K. Shimizu / Journal of Organometallic Chemistry 611 (2000) 531–536
535
Therefore, one can recognize some contribution of
these reactive species to the relaxation of Si-network
during the film growth.
In the presence of chemical species that exhibit etch-
ing activity, the SiꢀSi bond formation and its kinetic
dissociation occur simultaneously on the top surface
during the film growth. Thus, the Si-network near the
surface propagates in a kind of chemical equilibrium
resulting from the contribution of the reactive species as
schematically illustrated in Fig. 5. This idea can explain
how the relaxation of Si-network takes place in all the
CVD processes described above: in the processes where
atomic hydrogen are involved, atomic hydrogen medi-
ates this chemical equilibrium and promotes the rear-
rangement of SiꢀSi bonds, leading to thermody-
namically stable Si-network, i.e., nucleation; in SiH4–
F2 and Si2H6–GeF4 systems, F2 and GeF4 mediate it
and give the same result as in the case with atomic
hydrogen, respectively. This is a speculation based on
the common chemical reactivity of the reactive species
involved in these CVD processes.
Fig. 5. Structural relaxation vs chemical equilibrium of Si-network.
This illustrates an idea explaining the structural relaxation of Si-net-
work into crystalline form via chemical equilibrium with the contribu-
tion of reactive species in the gas phase.
Therefore, further studies including in-situ analysis
and chemical characterization of the film surface and
gas phase products in these CVD processes, will give
insight into the chemical equilibrium assumed. In fact,
the growing surface is always covered with hydrogen
when the crystal growth of Si takes place in silane
plasma process as Matsuda pointed out. As the surface
is covered with hydrogen, so it was done with chlorine
in SiH2Cl2 plasma process when the crystal growth
takes place. This can be a good sign for this idea
because such a chemically equilibrated surface should
be covered with the active species involved, i.e., H and
Cl.
Fig. 6. Schematic diagram of LP-CVD apparatus and typical growth
condition.
3. Experimental
In a point of view that the crystal growth is a kind of
the continuous propagation of relaxed Si-network
where Si atoms are arranged in sp3 configuration, the
relaxation of Si-network at the initial stage of the film
growth is very important for following network propa-
gation. Once the Si-network is formed, it requires high
temperatures above 600°C to transform it into the
crystalline form. Therefore, in the low-temperature
crystal growth there must be some specific dynamic
chemical processes that rearrange Si-network into more
stable sp3-like configuration during the film growth.
As discussed in the proposed mechanisms for the
low-temperature crystal growth, most of the above
mentioned CVD processes involve the intensive contri-
bution of atomic hydrogen, which exhibits etching ac-
tivity for Si films, in the crystal growth. This is true in
SiH4–F2 and Si2H6–GeF4 systems as well in terms of
the contribution of chemical species exhibiting etching
activity, which are atomic fluorine, fluorine, or GeF4.
The experimental apparatus used for film deposition
is a typical cold-wall type of low-pressure CVD reactor
as shown in Fig. 6. It consists of a main chamber
equipped with a gas nozzle, a substrate platform, a
loading chamber, and two vacuum systems for baking
and deposition. The substrates, a piece of SiO2/c-Si
wafer, a quartz plate, and a Corning 7059 glass plate
were set on the substrate platform equipped with a
heater unit and their temperatures were monitored by a
thermocouple and controlled at 450°C.
In a typical experiment, flow rates of source materi-
als, i.e., Si2H6, GeF4, and He as a dilutent were 2–20,
0.1–2.7, and 300–500 sccm, respectively, which were
regulated by mass flow controllers. The reaction pres-
sure was kept at 0.4–100 Torr by using an automatic
pressure controller.
The film thickness was determined by scanning elec-
tron microscope (SEM). The film composition was

536
J.-i. Hanna, K. Shimizu / Journal of Organometallic Chemistry 611 (2000) 531–536
[10] H. Niki, D.J. Mains, J. Chem. Phys. 68 (1964) 304.
[11] T. Komiya, A. Kamo, H. Kujirai, H.D. Nguyen, I. Shimizu, J.
Hanna, Mat. Res. Soc. Proc. 194 (1991) 63.
[12] Y. Matsu, J. Hanna, unpublished data.
[13] S.W. Wang, S. Frukawa, Tungsten and Other Advanced Metals
for VLSI/ULSI Applications, Materials Research Society, Pitts-
burgh, 1990.
estimated by Vegard’s law from X-ray diffraction peak
of Si (220) face. Structural properties of the films were
determined by Raman and X-ray diffraction spectro-
scopies.
[14] M. Yamamato, M. Miyauchi, J. Hanna, Appl. Phys. Lett. 63
(1993) 2508.
References
[15] M. Yamamato, M. Miyauchi, J. Hanna, unpublished data.
[16] M. Yamamato, Y. Takada, J. Hanna, Appl. Phys. Lett. 64
(1994) 3467.
[17] K. Shiota, D. Inoue, K. Minami, M. Yamamoto, J. Hanna, Jpn.
J. Appl. Phys. 36 (1997) L989.
[18] J. Hanna, K. Shiota, M. Yamamoto, Mat. Res. Soc. Proc. 420
(1998) 945.
[19] J.A. Comfort, R. Reif, Electochem. Soc. Proc. CVD-X, (1987)
265.
[20] P.J. Wang, B.S. Mayerson, F.F. Fang, J. Nocera, B. Parker,
Appl. Phys. Lett. 55 (1989) 2333.
[21] C.C. Tsai, G.B. Anderson, R. Thompson, B. Wacker, J. Non-
cryst. Solids 114 (1989) 151.
[22] L. Guo, Y. Toyoshima, M. Kondo, A. Matsuda, Appl. Phys.
Lett. 75 (1999) 3515.
[1] D.L. Steabler, C.R. Wronski, Appl. Phys. Lett. 31 (1977) 292.
[2] J.I. Pankov, Semiconductor and Semimetals, vol. 21, Academic
Press, New York, 1984.
[3] T. Hamasaki, H. Kurata, M. Hirose, Y. Osaka, Appl. Phys.
Lett. 37 (1980) 1084.
[4] H. Matsumura, Jpn. J. Appl. Phys. 30 (1991) L1522.
[5] S. Nishida, H. Tasaki, M. Konagai, K. Takahashi, J. Appl.
Phys. 58 (1985) 1427.
[6] (a) T. Nagahara, K. Fujimoto, N. Kohno, Y. Kashiwagi, H.
Kakinoki, Jpn. J. Appl. Phys. 31 (1992) 4555. (b) K. Endo, M.
Bunyo, I. Shimizu, J. Hanna, Mat. Res. Soc. Symp. Proc. 283
(1993) 641.
[7] J. Hanna, Proc. Mat. Res. Soc. 37 (1995) 105.
[8] J. Hanna, K. Shiota, M. Yamamoto, Mat. Res. Soc. Proc. 507
(1998) 945.
[23] S. Veprek, M. Maracek, Solid State Electron. 11 (1968) 683.
[24] A. Matsuda, J. Non-cryst. Solids 59/60 (1993) 767.
[9] A. Matsuda, T. Kaga, H. Tanaka, L. Malhotra, K. Tanka, Jpn.
J. Appl. Phys. 22 (1983) L115.
.